CONF-820544

CONF-820544—

DE83 006404 International Conference on

WELDING TECHNOLOGY for
ENERGY APPLICATIONS

Gatlinbnrg, Tennessee
May 16-19,1982

Compiled by
S. A. David
G. M. Slaughter

Date Published: September 1982

Sponsored by

American Welding Society

Northeast Tennessee Section
Oak Ridge National Laboratory
American Society for
Metals Joining Division

prepared by OAK RIDGE NATIONAL LABORATORY Oak Ridge. Tennessee 3 7 8 3 0

operated by UNION CARBIDE CORPORATION foi the DEPARTMENT OF ENERGY
Contract No. W-7405-eng-26

DISTRIBUTION OF THIS DOCUMENT IS

 FOREWORD
In view of the critical role of welding in the fabrication technology for economical and reliable
energy systems, a three-day International Conference on Welding Technology for Energy Applications
was held in Gatlinburg, Tennessee on May 16-19, 1982. Jointly sponsored by the American Welding
Society Northeast Tennessee Section, the Oak Ridge National Laboratory, and the American Society
for Metals Joining Division, the Conference brought together outstanding welding researchers,
engineers, and administrators from around the world.
The program covered a variety of joining research and development subjects related to such
important energy sources as fossil and nuclear, with special emphasis on pressure vessels, heat
exchangers, automation, and materials considerations. Segments of the Conference addressed
advanced materials development, welding process improvements, and metallurgical aspects of joining
technology such as dissimilar metal welds.
A special feature of the Conference was the assessment of the status of welding throughout the
world by renowned executives from Europe and Asia as well as the United States.
Most of the speakers prepared written papers detailing the subject of their oral presentation.
These papers have been compiled in this Proceedings of the Conference, which will document for
history the excellent work discussed.
We wish to express our appreciation to the Organizing Committee for its overall help and support.
Also, we wish to express thanks to Vivian Jacobs of Oak Ridge National Laboratory who provided
editorial assistance for the Conference Abstracts and Proceedings. Finally, our sincere thanks are given
to the Session Chairmen and authors for their contributions in making this Conference a major success.
We sincerely hope that the attendees enjoyed their stay in beautiful Gatlinburg and the visits many
made to the 1982 World's Fair, held simultaneously in nearby Knoxville.

S. A. David, Conference General Chairman

G. M. Slaughter, Program Chairman

 CONTENTS

Foreword iii
Session A. International Assessment

Welding Technology for Energy Conversion

A. A. Wells 3

Materials and Welding Process Development for Nuclear Application in the

Federal Republic of Germany
K. Kussmaul, W. Stoppler, and R. Sinz 17

Welding Technologies and Materials for Energy Applications in Japan

/. Masumoto, K. Masubuchi, and 7. Fujimura 71

Session B. Materials Considerations

Critical Materials Availability—Opportunities in Conservation and

Substitution Technology
Allen G. Gray 91

Session C. Fossil Applications I

Welding Practices in Fossil-Fired Units

G. G. Stephenson 105

Processes, Quality Assurance and Examination of Weld Metal Overlays

John J. Meyer 119

Hardfacing Materials and Processes for Valve Applications

S. J. Mathews and P. Crook 127

Session D. Gas Shielded Processes

Weld Penetration Variation in GTA Welding of Some 304L Stainless Steels

J. L Fihey and R. Simoneau 139

Evaluation of Welding Techniques and Weldments for High Temperature

Components Used in th9 Gasification of Coal
D. K. Schuhmacher et al 155
 Process Parameter Effects on Arc Physics and Heat Flow in GTAW
J. F. Key et al 179

Materials and Welding of Small Bore Pipework for Nuclear Fuel Reprocessing Plant "-•
L. Carrick and A. Paton 201

Session I . Fossil Applications II

High Temperature Properties of the Seam Weld Portion in 1 % Cr-0.5% Mo

ERW Boiler Tubing After Long-term Aging
T. Nakazawa et al 217

Substructure of a 16-8-2 GTA Weld as Observed Through TEM

J R. Foulds and J. Moteff 231

Creep Deformation and Crack Growth in a Low Alloy Steel Welded Pressure
Vessel Containing Defects
M. C. Coleman 255

Session F. Advanced Processes

Improvement of Fusion Welding Through Modeling, Measurement and

Real-Time Control
Koichi Masubuchi et al 281

Robotics in Welding
John R. Dwyer 301

Laser Welding of Structural Alloys

£. A. Metzbower, D. W. Moon and F. W. Fraser 313

Electron Beam Welding for Heavy Section Steel Plates

J. Tanaka. S. Koauge, and H. Furukawa 331

Session G. Nuclear Applications

Welding Metallurgy of SA508 Cl II Heat Affected Zones

P. J. Alberry and J. A. Lambert 347

Effect of Weld Residual Stresses on Intergranular Stress Corrosion Cracking

of Type 304 Stainless Steel Boiling Water Reactor Piping
Joseph C. Danko 379

Remote Automatic Plasma Arc Closure Welding of a Dry Storage Canister

for Spent Nuclear Fuel and High-Level Radioactive Waste
R. P. Sprecace and IV. P. Blankenship 395

Session H. High Deposition Rate Processes

Performance of the CaF2 - CaO - SiO2 System as a

Submerged Arc Welding Flux for a Niobium Based HSLA Steel
S. Liu K. C, C. B. Dallam, and D. L. Olson 445

vi
Toughness of Submerged Arc Weld Metals of Controlled Rolled Nb
Bearing Steel
Tadamasa Yamaguchi et al 467

Creep Characteristics of Butt Welded Joints for Thick 2V* Cr-1 Mo

Pipework and Pressure Vessels
G. Cumino and F. Rinaldi t 483

Influence of Flux Composition on the Chemistry and Cleanliness of

2% Cr-1 Mo Steel Electroslag Weld Metal
J. J. Odani, R. H. Frost, and G. R. Edwards 509

Session J . Heat Ixchangar Applications

Effect of Joint Design and Fabrication Limitations on

Tube-to-Tubesheet Welding
Samuel D. Reynolds, Jr. 533

Welding Equipment and Process Development for Inaccessible

Tube-to-Tubesheet Welding on a Sodium-Heated Steam Generator
J. H. Kiefer 573

Session K. Transition Joints

Creep Rupture and Creep Crack Growth Behavior of Transition Joints

a Nath 597

Damage Accumulation and Failure Criteria in 'Creep Brittle' Ferritic

Weldment Structures
B. J. Cane 623

Dissimilar Weld Failure Analysis and Development Program

K. H. Holko and C. C. Li 641

Some Observations on the Physical Metallurgy of Nickel Alloy Weld Metals

C. G. Skillern and A. C. Lingenfelter 673

vai
 WELDING TECHNOLOGY FOR ENERGY CONVERSION

A. A. Wells

The Welding Institute

Abington Hall
Cambridge, U.K.

ABSTRACT

Energy conversions from the principal fuels such as coal,

oil, gas, nuclear, and to a lesser extent the renewable sources
assert demands upon materials and joining to the limits of
human ingenuity. The quest for efficiency leads to extremes of
operating temperature both high and low, at which for reasons
of safety there must be adequate strengths and ductilities of
materials under transient as well as steady conditions. The
demand for large unit plant sizes arises from the need to
achieve economies of scale, originally in terms of minimising
first cost but also increasingly because of space limitations
at site. It has the effect of progressively increasing the
section sizes over which such material properties must be sus-
tained, and also leads to a preference for welded joints which
approach the homogeneity of parent materials. To add to the
challenge, the economic pressure to use impure sources of fuels
creates additional problems of erosion and corrosion of many
different types, which call increasingly for techniques of
metallic cladding and surfacing.
The approach to materials and joining must be interdis-
ciplinary under these circumstances, and must also have regard
at least to the broad classification of components into such
as pressure and storage vessels, piping, heat and mass transfer
equipments, and dynamic parts for pumps, turbines and electric
generators. The field is vast, and the paper is concerned more
narrowly in this context with the present development and
future prospects of the newer welding processes such as electron
beam, friction and diffusion bonding, in comparison with the
range of arc fusion processes, together with the joint properties
that are obtained. It also makes reference to particular
problems such as site construction, repair, remote welding and
heat treatment.
 INTRODUCTION

Being preoccupied with other technological matters, such as a world

revolution in communications and control through the application of micro-
electronics, we are inclined to forget as citizens the large extent to
which we now rely for future security upon a copious availability of
energy to support human life. We are committed to sustain it in the
developed countries and provide for its increase elsewhere, yet we are
also obliged to remember that the rate of consumption of energy has
reached current levels from a base near zero entirely over the past two
centuries, which is a mere instant in geological time. Whereas most
members of society would voluntarily accept that liberal supplies of
energy should continue, it is not clear that all would will the means to
ensure it. Part of the difficulty lies with understanding the magnitude
of the challenge; for instance, the approximate annual consumption of
thermal energy in the UK as a small country with a moderate per capita
consumption can be expressed in various ways such as, 10 Joule, average
load 300 gigawatt, 300 million tonnes of coal equivalent, 0.3 cubic kilo-
metre of coal equivalent. Only the last of these figures begins to acquire
meaning, and it might have appealed to Sir Winston Churchill who complained
about "those damned noughts", probably when considering a wartime logistic
problem.
One of the first lessons that emerges from comprehension of the scale
of energy demand is that supply must make use of a mix in which no com-
ponent is neglected, comprising oil, gas, coal, nuclear and renewable
sources, wherever they are to be found. The next lesson relates to
manageable plant sizes. If a national energy demand is of the order as
above of 1 million tonnes of coal equivalent a day, and present-day con-
version plant sizes are of the order of 1000 tonnes/day, then there will
obviously be an aggregate of about 1 thousand plants in service at one
time. Even if each lasts for 25 years, the replacement rate will have to
be about 1 per week, and if each takes 5 years to build there will always
be 200 under construction at any one time. It follows that there is a
constant need to increase unit plant sizes so that the task (and the site
space) can be accommodated.
Some apology may be needed for making such statements of the obvious
in a simplistic manner, but there is a degree of urgency in the message,
and it relates to the fate of energy planning at a time of recession and
high interest rates worldwide. The active pursuit of microelectronics is
readily defended under such conditions, since the short timescale befits
the high interest rates so that profitability is or can be retained. The
short timescale is not only related to the small physical sizes of solid-
state electronic components so that construction times are small; the
feasible rates of innovation and obsolescence are also comparatively
fast. The hardware of energy conversion represents an antithesis to this
state of affairs. It is massive and takes much longer times to construct.
What is more important is *"nat such hardware takes much longer periods of
time to develop, since experience from service is not quickly acquired,
so that decisions need to be taken in the near future with regard to plant
which will come into operation near the end of the century. Nowhere is
this more relevant than with regard to the materials of construction and
methods of fabrication including welding and joining that will be needed
for energy conversion plants to be built in large sizes and numbers in
one or two decades from now. There is a need to diminish construction
times, and this is likely to be met in part by improvements in organisa-
tion; there will also be a need for improved qualities of construction,
defined by such as fitness-for-purpose methods, and substantial efforts
will be needed to ensure that the pursuit of quality is made to be com-
patible with shorter construction times.
However, it seems to this observer that the greatest benefits would
accrue if more vigorous attention were to be given to several new welding
and joining methods which are appropriate because of their potential
productive capacity, to be fostered as long-term developments. Long-term
developments are unfashionable at a time of universally high interest
rates, but means should be found to accommodate them in a field which has
so much importance relative to the future needs of society.
Each of these joining methods is well beyond the laboratory stage and
some have been demonstrated at a large industrial scale, but years of work
are required in each case before they can be applied as a matter of
routine, in relation to the mechanical properties of the joints associated
with them, firstly in relation to field service and secondly in relation
to suitable accelerated laboratory tests related to quality assurance.
In order to consider materials and joining methods for energy con-
versions it is first necessary to introduce separate classifications, of
the physical and chemical processes employed, and the associated plant
components, as follows,
Power plants for electricity generation - nuclear, fossil fuel,
hydro, tidal, wind and wave.
Coal, oil and gas winning.
Coal/gas and coal/oil conversion, purification and cracking.

Types of equipment - drill, transmission and plant piping,

pressure and storage vessels, boilers,
heat and mass transfer equipments,
dynamic parts of pumps, turbines and generators.

The welding and joining processes which will mainly be discussed are
as follows,
Electron beam, friction, diffusion, cladding and surfacing.

ELECTRON BEAM WELDING

Whereas the centenary of the invention in Russia by Benardos of

carbon arc fusion welding occurred in 1981, the first applications of
electron beam welding were made about 25 years ago. Fully developed
equipments capable of making single pass butt welds with characteristic-
ally narrow fused zones in metals more than 100mm thick have been avail-
able for a decade, but it is only recently that the process has been
successfully applied for production of heavy pressure vessels, '" There
has been much to learn concerning the control of liquid metal flow and
outgassing of the weld in vacuum. Thus, just as submerged-arc welding
 is mainly confined to the flat and horizontal-vertical positions and
electroslag welding to the vertical position (with dams), heavy section
electron beam welding is best performed in the horizontal-vertical or
vertical up/down positions. These limitations in each case arise from
the need to retain the position of the melt with respect to gravity
effects. Since the electron beam process is usually autogenous, using
neither filler metal nor flux, there are limitations on parent metals
with respect to avoiding weld defects which resemble those for high
quality fusion welding, but are not identical. Indeed, the differences
may have been more important than the similarities.
One of the elusive defects in heavy section electron beam welding
has taken the form of deep worm holes which can also be more or less
elongated, superficially like a form of porosity although the cause has
been traced elsewhere. This defect is characteristically due to
momentary loss of the beam through a flashover discharge in the electron
gun caused by ingress of metal vapour. The metal vapour is created in
large quantities at the melt and is drawn to the separately evacuated
gun which is normally at a higher vacuum than the work chamber. This
effect can be neutralised by the use of suppressors which cause the
collimated electron beam to be deflected through a chicane, and such
devices can be very effective. Care should also be taken to eliminate
flashover discharges within the high voltage power supply and connecting
cable, and these are similarly controlled by design considerations.
Centre plane weakness defects can also occur from metallurgical
causes and resemble those sometimes seen in electroslag or even submerged-
arc welds. They erise from segregation of low melting constituents just
as in solidifying ingots, and are exacerbated at higher welding speeds.
The explanation of the latter effect is associated with the transient
heat flow, which is well understood. At low welding speeds the melted
zone approaches the cylindrical but at higher speeds, characterised by
the product of speed and weld width, the fused zone develops a long tail
so that much of the solidification front is essentially parallel to weld
length. Favourable conditions are then created for segregations to con-
centrate at the mid-plane. Such effects are best countered by improving
the cleanliness of the parent metal, and avoiding high welding speeds,
although the latter are relative; the optimal values are always much
higher than for comparable welding processes because of the narrowness
of the zone fused by the collimated electron beam. Width/thickness
aspect ratios are of the order 50 even for steel butt welds of 150mm
thickness.
In comparison with these now well understood and avoidable defect
occurrences the advantages of electron beam welding are substantial. For
instance, hydrogen cracking problems are minimal because of the vacuum
operation. The process is exceptionally productive and single pass welds
can be made in large thicknesses at welding speeds in the range 0.1-1
metre/minute, or nearly 2 orders of magnitude faster than electroslag
welding. Circumferential welds can be made with impeccable quality at
slope-out where the finished weld returns to the starting position.
Coarse grain heat-affected zones are eliminated by the combination of
high welding speed and small weld width. Residual stress and distortion
effects are also minimised by the narrow heated zone; indeed, it might
be claimed that they are eliminated for practical purposes. It is
 debatable whether heavy section electron beam welds need post-weld heat
treatment. The minimisation of heat degradation effects can be illus-
trated in a striking manner in terms of overal heat input; it is possible
to electron beam weld for an hour on a heavy component, which is only
hand warm when it emerges from the vacuum chamber. (Considerable use of
this property at a smaller scale has been made in electron beam closing
welds for the cases of vehicle automatic transmissions containing
precision machined and finished gear components. It is claimed that the
grease applied to these components at the time of assembly is not melted.)
The enclosure requirements for heavy section electron beam welding
are more easily satisfied than might be thought, and comprise operator
protection from x-radiation provided by 5-10mm of lead, and vacuum of
10~2 to 10~4 Torr (since the electron gun may be separately pumped to
higher vacuum using a small integrally mounted equipment). The principal
requirement for a vacuum chamber to enclose large workpieces is for it to
be pumpad down in a short time so that the equipment as a whole is pro-
ductive. For instance, the lOOm^ chamber at The Welding Institute is
pumped down to 10" 3 Torr in about 40 minutes. Once this facility is pro-
vided the sealing of leaks ceases to be a problem. Again, the TWI
facility makes use of a 4 metre square door, for which it proved to be
unnecessary to machine the mating and sealing faces. The significance of
this experience is that it would be fully practicable to employ
temporarily constructed vacuum spaces under shop or site conditions for
the joining of large assemblies, and the comparative success with which
various forms of sliding seal have been employed, notably in France and
Japan, bear out this assertion. Figure 1 shows an elaboration of the
system first developed and used experimentally by Sciaky in France,
which could be employed for the construction on site of large, vertical
axis cylindrical high pressure reaction vessels for energy conversions
if it became necessary to do so. Successive ring sections would be
added at the bottom, the vessel being progressively lifted within the
framework which would support it during service. Each ring would first
be formed from petals by vertical-up EB welding, and the practicability
of this would be enhanced by the freedom from distortion exhibited by the
welding method.
Although EB welding has a similar high potential for producing the
quality welds required in nuclear reactor vessels of the light water type
the process has appeared rather late on the scene to offer economic
advantage, bearing in mind the large investment required in weld proced-
ure qualifications. Nevertheless, it should be considered as a serious
candidate for welding the closely spaced set-through primary circuit
nozzles in PWR vessels, because of its excellent characteristics with
regard to low distortion and residual stresses, and the facility with
which the process is controlled. This would also be a suitable initial
application to LWR construction because of the compactness of the local
vacuum chamber. Post-weld heat treatment would follow with such con-
structions but it is unlikely that preheat would be required.
Expressed in the most general terms, these examples afford an
excellent example of the forethought required with the development of the
larger scale welding processes. It would be unlikely for either to come
to maturity in less than 5 to 10 years, although the rewards to be had in
terms of scope of achievement as well as improved productivity are
significantly large.

A — Electron gun with z, d motions B, C — Self supporting conical and dome vacuum closures
D - Hydraulic supports

Fig. 1. Schematic arrangement for EB welding of large vessels in situ.

 FRICTION WELDING

It is not surprising that there is such a multiplicity of different

welding processes (with 10 or more classes and many more variants) when
the whole range of industrial joining problems is considered. Some of
these are candidates in the heavy pressure vessel field for energy con-
versions, but a different range should be considered for all types of
pipe and pipeline. Although arc fusion welding processes take precedence
at the present time, with a burgeoning use of automatic methods, the
situation will continue to be dynamic. An important feature of pipe
welding is that it will often be performed on site so that light-weight
equipment is preferred, and that the chosen processes must be flexible
with respect to access and working position. The long-standing prefer-
ence for stovepipe manual welding of pipelines is thus adequately
explained, and if. was not until pipelines were laid offshore from
expensive-to-run lay barges that automatic fusion welding methods came
into their own through economic pressures.
Any alternative candidate process for pipe joining must offer
improvement in productivity, consistency of quality and facility for non-
destructive examination, preferably in full combination. One of the
candidate processes is friction welding, now widely recognised as well-
suited to joining end attachments to drill pipe under shop conditions.
Flash welding is another process which has been patiently developed in
the USSR for the field joining of larger diameter pipes. One of the dif-
ficulties that has plagued pipe flash welding for many years is that of
locally poor fusion (arcing flat spots) arising from uneven pressures and
related conditions around the periphery, and these areas are difficult to
detect by non-destructive examination. It is probable that these diffi-
culties have been adequately mastered in the USSR during the 25 years
that the development has occupied, but the evidence remains to be pub-
lished. Although friction welding has yet to be successfully applied to
the field joining of large-diameter pipes, the process has two potential
advantages. Firstly, energy conversion for field joining will be
mainly through internal combustion engines, and the weight of the equip-
ment is much increase^1 if there has then to be conversion to electricity,
and if transformers also have to be used. In principle, this step would
be removed with a mechanical process sucii as friction welding. Secondly,
friction welds, being typical of solid-phase joints, usually exhibit a
high consistency with respect to destructive tests irrespective of dis-
tribution around the periphery, so that non-destructive examination is
of less importance than establishment of proof that the welding cycle has
been performed within limits which themselves exhibit a fairly wide
tolerance.
Friction welds are also considered to be subject to a number of dis-
advantages for pipes. There are internal and external flashes, which may
have to be removed to facilitate internal fluid flow and external coating.
Conventional friction welding in which one end of the pipe is rotated
relatively to the remainder is unsuited to other than straight pipes.
Indeed, there is a greater disadvantage that the rotation occurs at slow
speeds with very large torques and this, coupled with the necessity to
apply large axial forging loads through the bearing system, has hitherto
 10

totally precluded the operation for large diameters and field joining.
However, the weight and cost of the equipment would not necessarily be so
serious for lay barge mounting, so that the process in this application
should stand comparison with flash welding.
Radial friction welding, introduced by The Welding Institute, is a
variant in which there is no relative rotation between the pipes being
joined. Instead of this a somewhat larger diameter narrow coupling ring
is introduced over the butt joint between the pipes and this ring is
rotated and forged down by the continuous squeezing action of the mount-
ing chuck as the friction process proceeds. The pipe ends are also
clamped to form rigid abutments, so that no axial forge is required.
Furthermore, a plug temporarily inserted inside the pipe both supports
the latter and suppresses the formation of an internal flash. This form
of friction welding has been sufficiently developed to justify it as a
promising candidate for the offshore assembly of flow lines of the
smaller diameters.
These experiences collectively suggest that friction welding con-
tinues to have development potential in spite of its accumulated history
of 30 years of development. Another variant which has been demonstrated
is orbital welding in which both components are rotated together with an
axial offset while being subjected to axial pressure. This has proved as
expected to be effective for solid bars, in the sense that the friction
velocity is constant over the whole cross-section. Studies of the micro-
structural deformation near the friction surfaces even show superiority
in comparison with conventional friction welding, and better mechanical
properties. It is considered that this group of processes should receive
substantial future attention in terms of pipe joining in the field.

DIFFUSION BONDING

This solid-phase welding process dates from antiquity as smith weld-

ing although it is well-known in a modern context in terms of roll bond-
ing, and is recognised for its capacity to produce sound joints between
dissimilar metals. There is much still to be learned about the micro-
mechanisms of growth of crystallites across well-fitting interfaces,
although the relative influences of elevated temperature, time, inter-
facial pressure and surface roughness have recently been well system-
atised for several metals by Derby and Wallach. As a result of refined
but small-scale studies conducted in vacuo by Bucklow at The Welding
Institute he has additionally emphasized the favourable influence of
quite small tangential relative motions at the interface under otherwise
suitable bonding conditions and these, of course, are familiar as the
unwanted seizures of sliding parts in mechanical engineering. Hence,
diffusion bonding is not only the oldest of the welding processes for
metals but it also promises considerable future development. It has the
ultimate potential for joining without microstructural discontinuity or
residual stress. It is both isothermal and autcgenous and therefore
offers full joint efficiencies for all forms of loading. It also poses
the ultimate challenge for non-destructive examination, although the
mechanical performances of roll-clad plates for chemical vessels have
 11

been consistently good and place that particular need into an appropriate
perspective.
Diffusion bonding is in many cases a natural alternative candidate
process compared with bra2ing which would extend the mechanical perform-
ances of joined components to higher temperatures. It deserves considera-
tion for the construction of dynamically loaded or rotating parts such as
are associated with turbines and pumps, and is relevant to energy conver-
sion in relation to which a proportion of these components habitually
operate at high temperatures. Like electron beam welding, with which it
might be considered to compete*, it is facilitated by operation in vacuum,
but the requirements may not eventually prove to be onerous. It is often
considered at the present time that diffusion bonding has to make use of
substantial external loading in order to provide the necessary local
plastic deformation at interfaces to counteract the local asperity con-
tact that arises even with superior surface finishes. However, external
loading should be replaceable in many instances by interference fits
such as are produced by wedges and tapered pins, since these also have
the capacity to provide tangential motions at the interface. Considera-
tion should be given to circumstances in which such components are seated
by a light blow when the appropriate bonding temperature has been
reached, as for instance with the detonation of small explosive charges.
The best features of brazing and explos've welding might be combined in
this way. There is much scope for innc.sration with this process, which
has been too little explored.

CLADDING AND SURFACING

It is clear that there will be a future need in coal conversion

equipments for improved methods of applying protective cladding, in view
of the erosion and corrosion effects which will be experienced, arising
from abrasion and impurities such as sulphur compounds in the fuels
which will be required to be used. It is likely that the most suitable
surfacing materials will not be those optimised for the membrane strengths
and ductilities required of the containing vessels and piping.
There are already several weld cladding processes in addition to
roll bonding and the most favoured of these make use of submerged arcs
and metallic fillers in strip form. Recent developments in Japan have
demonstrated that strips up to 0.3m wide can be deposited uniformly in
thick layers on the electroslag welding principle, by making use of
exterral magnetic fields which both improve the uniformity of heat input
across the width and also assert a damming action on the melted metal.

*It may be noted however that, whereas diffusion bonding is a well-

controlled process for joining dissimilar metals, there are sometimes
difficulties with EB welding in that context, arising from the electro-
magnetic disturbance of the beam due to large circulating currents of
thermoelectric origin.
 12

Attention should also be given to the electrostatic capability of

electron flows, made familiar in sputtering and ion bombardment, which
can transfer materials in the vapour phase to conducting surfaces. Such
processes should be amenable to the uniform coating of large areas in a
controlled manner, with considerable scope as to the electrical conduc-
tance of the transferred materials. Such processes to be conducted on
the large scale have reached little more than a gestation stage, but
deserve considerable attention now if they are to be available when
needed on a large scale in a decade or more from now. It has already
been seen in this brief review how little progress can be made with large-
scale fabrication process developments in periods as brief as 5 years,
simply because the processes of trial by service followed by modification
cannot easily be accelerated, and large investments in fabrication plant
cannot economically be written off before they have been suitably pro-
ductive.

THE CONTROL OF WELDING ARCS

In the period about 20 years ago there was a leap forward in the
understanding of metal droplet transfer across welding arcs from consum-
able wire electrodes with inert gas shielding. There is a natural trans-
fer rate dependent upon current and the electromagnetic detachment
forces which are generated by it, which is slow with large globules at
low currents and spray with small globules at large currents. The spray
transfer also has a pronounced scouring effect in the weld pool, which
produces a finger-like penetration. The first use of this understanding
was to select suitable welding currents to facilitate metal transfer.
However, these welding currents are not necessarily at the values which
would be chosen to obtain the optimal transient heating and melting of
parent metal to effect a joint. The next step was to assert control on
droplet detachment rate in order to fill the joint at the best rate from
more general considerations, and this can be done by pulsing, using
solid-state electronic power supplies. Pulse control is also effective
in non-consumable tungsten arc welding where there are positional varia-
tions with respect to gravity. Both the pulse concentrations and their
rates can be controlled electronically. The Welding Institute has made
a significant contribution to this development, and great improvements
can be made to the joint quality of arc welds using the so-called
synergic pulsing. The nuclear, aerospace and engine industries have
taken up the development with considerable determination and thrust, and
have been able to make considerable improvements thereby in joint per-
formances .
Although much of industry would wish to persist with conventional
arc welding power sources, using transformers, rectifiers and DC genera-
tors, which have the merit of being relatively cheap and rugged after
many decades of development, it has become evident from the experience
of developing synergic pulse arc welding that much more benefit could
accrue from the use of solid-state welding power sources, and the micro-
processors which are so readily associated with them. The drive in this
direction is inexorable, and may be compared with developments which have
 13

taken place in the machine tool field.

However, not all the technical problems have been solved, and a
typical difficulty that is experienced relates to fluctuating electrical
supplies, or interference from adjacent electrical equipments; it has
the effect of reducing the repeatability with which the desired settings
for welding can be transferred from one location to another. Such dif-
ficulties are typical of those which require sustained effort to over-
come, which The Welding Institute is dedicated to provide, but they are
not of such magnitude as to cancel the advantages of the developments in
themselves. They serve more to explain the requirement for new develop-
ments in welding technology to be nurtured with patience for times which
are generally ionger than would be expected at the outset.
The necessity for full visual control of welding arcs is usually
accepted without question, and the welder's mask with dark glass is
commonplace. Much was learned a generation ago about arc behaviour,
using high-speed cine photography, and the techniques are now highly
developed. Work has continued since on the use of video recording, which
has the additional great value of application in real time, leading to
the possibility of remote viewing and control. Video techniques have
also reached a high state of development, including the accurate render-
ing of colour, and considerable magnification is possible. The stage
has certainly been reached in UK work where the video camera can displace
the welder's mask, so that the welder can operate remotely in the use of
automatic and semi-automatic open arc welding processes. The techniques
of pattern recognition have also been developed in association with tele-
vision monitoring, and offer substantial promise in the extension of
viewing to automatic control of position. It must surely be argued in
relation to welding on large costly items of capital plant, where the
operator has at present to concentrate attention for long periods of
time, and lapse of concentration may lead to expensive mistakes, that
the provision of automatic control should receive the same priority as
it would in the control of aircraft by the pilot.

PROBLEMS OF REPAIR

Considerable strides have recently been taken in the conduct of

repairs by remote welding on nuclear reactor components worn by erosion
during long service, although these have hitherto mainly been concerned
with tubes of relatively small diameters. Considerable use has been
made of closed circuit television in conjunction with robotic welding
equipment, in confined spaces with elevated temperatures and in some
cases in the presence of back pressures. The existences of minor back-
ground irradiations under these circumstances have not been of undue sig-
nificance. The present status of this type of work is one of steadily
growing experience and a moderate increase of confidence. It is also
considered to be likely that cladding repairs in LWR components have
proved to be feasible and have been successfully carried through,
although there is no open knowledge of circumstances. All these situa-
tions have in common that no isothermal post-weld heat treatments would
be required, whether during first manufacture or repair.
 14

It can be conjectured that the need will sometime arise for weld
repairs to be conducted on heavy-walled pressure components in reactor
service, under circumstances which would normally call for post-weld
heat treatment. A special difficulty is that the circumstances could
vary so widely that no general strategy can be foreseen at the present
time. On the one hand there are many circumstances where large welded
repairs have been successfully and repeatedly conducted on heavy compon-
ents such as rolling mill housings, made of steels of doubtful weldability,
and without post-weld heat treatments. On the other hand there would
seem to be no justification at the present time for relaxing such a
requirement for repair welds on nuclear pressure components where this
would be mandatory for new constructions. An answer to this dilemma is
best likely to be found by making a systematic study of the performances
of weld repair techniques, and there is ample justification for doing
this now, while the issue is unclouded by pressing and particular needs.
There is the advantage that a full and effective range of remote welding
techniques is already available. It is the need for PWHT which is in
doubt.

DISCUSSION AND CONCLUSIONS

A paper such as this loses significance if it is phrased in sensa-

tional or overtly political terms but it is entitled to have a theme.
The chosen theme relates to the vital necessity for sustenance of energy
supplies into the future a generation from now, that the investment be
maintained in research and development on fabrication, welding and join-
ing for energy production and conversion plant, because of the long
timescale in work of this kind. The message is not only intended for
industrialists, and governments where appropriate, but also for society
at large. There should be adequate public debate to ensure that the
alternatives are clearly understood. It is one thing for the public to
choose freely to revert to a state of lesser dependence on abundant
energy; it would be another for that condition to arise by default. It
is a legitimate responsibility for engineers and technologists to ensure
that the issues are known and understood.

REFERENCES

1. A. N. Kornienko, "One Hundred Years of Arc Welding", Metal

Construction, 14(3): 144-147 (March 1982).

2. M. Endo et al., "Electron Beam Welding for Pressure Hull of Deep

Submergence Research Vehicle made of 10Ni-8Co Ultra High Steel",
Trans. Japan Welding Society, 12(2): 24-33 (October 1981).

3. K. Yokota et al., "2000m Deep Submergence Research Vehicle 'SHINKAI

2000' ", Mitsubishi Heavy Industries Technical Review, 18(3): 236-
247 (October 1981).
 15

U. B. Derby and E. R. Wallach, "Theoretical Model for Diffusion Bond-

ing", Metal Science, 16: 49-56 (January 1982).
 MATERIALS AND HELPING PROCESS DEVELOPMENT
FOR NUCLEAR APPLICATION
IN THE FEDERAL REPUBLIC OF GERMANY

K. Kussmaul, W. Stoppler and R. Sinz

Staatliche Material^ -lifungsanstalt (MPA)
University of Stuttgart
Pfaffenwaldring 32
D-7000 Stuttgart 80

ABSTRACT

The concept of Basis Safety was developed and gradually implemented

in the Federal Republic of Germany (FRG) in order to achieve an utmost
availability and EXCLUSION OF CATASTROPHIC FAILURE of pressure boundary
components in Light Water Reactors (LWRs).
This concept provides, amongst others, redundant safeties both
for long-term service as well as for an accident by improving toughness
properties of the non-influenced base material (BM), heat affected zone
(HAZ) and deposited metal (DM) of components with relevance to safety.
The toughness properties and the susceptibility to cracking can de-
cisively be influenced by a well balanced chemical composition at very low
sulphur contents, melting, forming, and heat treatment technique.
With the help of the welding simulation technique it was possible to
systematically investigate and to prove, e.g. in fine-grained structural
steels, the toughness properties in the HAZ as a function of the stress-
relief temperature. Submerged-arc welding (SAW) on heavy section components
were intensively investigated during and after welding. This helped to shed
light on the microstructural development of the HAZ and beads as a function
of the particular parameters of the welding process. Furthermore, this
helped to delineate boundary conditions for an appropriate computer pro-
gramme with the aid of which the design and optimization of a joint and its
HAZ can be reconstructed and simulated by computation. The advantage of the
narrow gap technique is the extremely low amount of sensitive coarse grained
zones.
Processing tough filler materials with optimum weld parameters will
provide excellent material properties at an elevated strength level in
shape-welded heavy section components.
The martensitic steel X 20 CrMoV 12 has on the one hand an excellent
weldability of .ne BM (HAZ without problems), but on the other hand it has
not been possible up to now, to achieve equal properties for the deposited
material. This is also the case for martensitic welding (preheat-temperature
at about 250 °C) which only enables large wall thicknesses to be welded.
As far as austenitic steels are concerned the sulphur content had also
to be lowered. Sufficient toughness had to be proved also in the case of im-
pact loading for austenitic welded joints subsequent to thermal aging.
Finally, the reason for the susceptibility to corrosion of individual aus-
tenitic materials and material conditions was investigated and remedies
elaborated.
 18

1. SAFETY STRATEGY
All research and development activities in the FRG regarding the
safety of ferritic components of LWRs have the feature of a two-fold
safety strategy /I/.
On the one hand, the available failure criteria are to be examined
as to their validity for boundary or worst case conditions, and on the
other hand, the basis for a deterministic EXCLUSION of CATASTROPHIC FAIL IRE
has to be established and to be proven. This can be achieved by optimizir. ]
the chemical composition of the material and all further processing steps
as well as the design of components and the operating conditions.
This strategy is illustrated in Fig. 1, where the safety margin is
given by the ratio of the applied load (K) and the material resistance (R).
If the condition K< R is infringed, failure will occur. In general,
it can be caused by
- extensive local or general deformation including instability
- crack initiation and (stable)crack growth
- unstable crack growth (ductile and brittle fracture)
The research and development work is, therefore, concentrated on in-
vestigation of factors which may lead to catastrophic failure and the
measures which can be derived from basic knowledge with respect to failure
prevention. Regarding the reactor pressure vessel, the exclusion of (cata-
strophic) rupture is tantamount to the requirement of preventing crack in-
itiation right from the beginning or, in the c^se of initiation, at least
to assure crack arrest. The condition K < R can only be fulfilled, if all
factors which influence loading, and load bearing capacity are completely
and quantitatively taken into consideration in production and during
service.
From this the above mentioned strategy can be derived / 2 , 3/. On the
one hand, the remaining safety margin can be determined on the basis of
R m i n and K m a x (lower bound concept) and, on the other hand, a high R is
desired in connection with a low K (upper bound concept). A similar method
was chosen, e.g. for 12 % Cr-steels and their welded joints. These steels
are used for high temperature reactors and fast breeders. The same concept
can also be applied to austenitic materials that have been extensively ana-
lysed in view of the Bethe-Tait-energy-excursion.
Because a sophisticated welding technology is highly affected by
materials and design, the authors attempt to present an overall view.
1.1. Lower Bound Concept
1.1.1. Realization of boundary conditions
The load bearing capacity of a component is determined by the tough-
ness and flaw conditions of base materials and welded joints. All quality
degrading factors, Table 1, that may occur during manufacturing and in
service have to be considered, if a realistic lower limit of load bearing
capacity of materials has to be quantified. Embrittlement, and ultimately
cracking can already be caused during manufacturing, if there is an un-
favourable chemical composition in connection with non-optimized processing
of base material and welded joints. On the other hand^ embrittled areas
will promote cracking in service, in which environmental factors will have
an aggravating effect.
 19

It was possible to delineate these boundary conditions in the re-

search programmes in the production of base materials as well as welded
joints by means of systematically selected parameters which exceeded these
boundaries. The base material of various toughness levels with and without
natural cracks, and welded joints with original micro and macrocracks that
are found in the HAZ and in the weld material define the basis of the load
bearing capacity diagram represented in the spatial system of coordinates,
Fig. 2. The respective load bearing capacity of the material condition may
T5e quantified with the help of a fracture concept in which all possible
combinations of flaw sizes as well as toughness levels are included. All
these combinations also incorporate time-depending changes in the real com-
ponent up to the end of life (EOL) state. The materials and welded joints
mentioned have to be loaded under the most stringent conditions ( K m a x ) , in
order to quantify the smallest safety margin; they have to be in accordance
with the lower bound concept. Finally, this means that representative
specimens have to be tested in dimensions comparable to the components
in which a conservative stress condition (high multiaxial ratio) is
achieved by notches or cracks and residual stresses. Complex stress con-
ditions can be seen, e.g. in the LWR pressure vessel in connection with
thermal loading (loss of coolant accident, small leak). In the case of
the fast breeder, a sudden energy excursion of 370 MWs that can only be
contained by complete plastification of components has to be considered.
Exceptional testing facilities for large scale specimens and vessels
are required to perform tests :nder such conditions. The test specimens
and test vessels are, in special cases, subjected to combined loading by
internal pressure, and thermal shock, including residual stresses.

1.1.2. Quantification of the safety margin and verification

of safety concepts
On a conventional basis, the following necessary restrictions are
required:
- limitation of general loading under static load
- limitation of local strain in high stress regions under static
and cyclic load.
A sufficient capability of deformation may be guaranteed by the necessary
minimum of elongation at fracture and reduction of area in the tensile
test, and charpy impact energy. Additionally, in the case of ferritic
material, fracture below yield stress has to be excluded by the indroduc-
tion of brittle fracture transition temperature, and the definition of
allowable temperature areas (e.g. Pellini, and Porse diagrams).
The failure of a precracked component is described by fracture mech-
anics concept:
- linear-elastic fracture mechanics (LEFM), ASME III, App. G and
ASME XI, App. A
- Elastic-plastic fracture mechanics (EPFM), COD and J-Integral
concept each with the corresponding design curve
- Plastic limit load calculation.
It is possible to use the failure assessment diagram / 4 / and the failure
analysis diagram /5/ universally for all ranges of toughness.
The validity of these safety concepts may be demonstrated with the
help of material and loading combinations described above under special
consideration of lower bound conditions.
 20

1.2. Optimization of Material Manufacture and Design for LWRs

In addition to the quantification of the minimum safety margin, the
optimization of material manufacture and processing (Rmax) as we11 as
the design of components and operational process (Km-jn) have been persued.
Quality degrading factors from manufacture and operation, Table 1,
initiated development work in the field of base materials and welded joints
as described in the following. An examination had to be made as to whether
the fracture mechanics parameters can be correlated with conventional
strength and toughness parameters that have been determined as part of
the quality assurance tests. The following new requirements on material
toughness completed within a revised version of the chemical composition
for RPVs - 22 NiMoCr 37 (similar to A 508 Cl. 2 ) , 20 MnMoNi 55 (similar
to A 508 Cl. 3) - and containment vessel - 15 MnNi 63 - steels, Table 2,
are intended to guarantee deterministically the structural integrity even
with local defects and cracks, Fig. 3, /6, 7, 8/:
upper shelf energy ^100 J ) for base material,
notch impact energy at 0 5 41 J ) HAZ, weld metal,
notch impact energy at + 33 K 2: 68 J ) (transverse
lateral expansion at + 33 K £0,9 mm ) direction)
reduction of area at fracture 2: 45 % all directions
notch impact energy of
HAZ-simulated base material >
at service conditions — 68 J
An additional part of the basis safety concept are design rules/9/, such as
- conservative limitation of loads which arise
- large water gap in the RPV to reduce neutron exposure
- reinforcement of nozzle areas in the vessel
- reduction of welds, in future no weldments in the core region
- no welds at highly loaded positions
- access for repeated nondestructive testing.
The inherent safety resulting from these requirements is adquate to ex-
clude catastrophic failure. Quality control in manufacture and surveillance
and testing in service provide additional redundance under these conditions
for the integrity of components. From this point of view, the most fre-
quently required weld repair at small discontinuities becomes more ques-
tionable, because they are without significance in the case of basic
safety.
2. MATERIAL DEVELOPMENT
2.1. Optimization of Factors Influencing Material Toughness and
Weldability
The material condition is influenced by chemical composition, melting
technology, forging and rolling, respectively, as well as heat treatment,
Fig. 4. In addition, the chemical composition also has an effect on the
weldability of the base material.
 21

2.1.1. Chemical composition

Since the influence of chemical composition encroaches on the compo-
nent integrity, it is ot the greatest importance to optimize the same.
Materials become increasingly susceptible to processing because of un-
favourable composition and, consequently, flaws, which may lead to re-
jection of the component, cannot be prevented.
Findings from investigations conducted or, the coarse-grained heat
affected zone (HAZ) in actual welds and under simulated conditions were
utilized for compiling new advanced analyses, because the thermal-mechanical
process in the HAZ brings about extremely stringed marginal conditions
that are not possible otherwise during fabrication history. In view of the
temperature-time-function, the technique of HAZ-simulation is based en
extensive temperature field measurements performed on component welds, cf.
paragraph 3.2. This technique consists of simulation of overheating and
stress-relieving /10 - 13/ with subsequent material testing. The
temperature-time-function is controlled in such a manner that the struc-
tural formation and precipitation conditions of the specimen ( d m a x = 5 5 mm)
are to a great extent identical with those of the HAZ of original weld-
ments. The annealing effect in multi-layer welds can be taken into account
by multiple overheating. The homogeneous formation of the overheating
structure over the specimen cross-section allows the simulated material
condition to be quantified conservatively without difficulty. Short-
time creep tests serve c.s a first step to reconstruct creep and relax-
ation processes during stress-relieving. Different strain levels up to
max. 2 % (without fracture) are generated in another testing sequence by
applying different load values at a constant testing time. Material prop-
erties can be attributed to those strain values in mechanical and techno-
logical post-examinations. The influence of chemical composition of ma-
terial on toughness and weldability can be assessed from the rupture curve
of the short-time creep test and the toughness in C v and tensile tests
as a function of increasing creep strain. Further realistic conclusions
are provided by the "heating relaxation test" mainly conducted if the
values required for the residual toughness are not totally achieved /13/.
The subsequent analysis of a great number of simulation tests made
it possible to assess the tendency to embrittlement /14/ of some alloy
types, Fig. 5 and, subsequently, to select those elements which have a
decisive influence on toughness reduction and relaxation crack formation
/15, 16/. Apart from micro-alloying carbide forming elements, minor and
trace elements proved to be especially critical. Stress-relief crack forma-
tion in the HAZ of welded joints has to be expected if the given threshold
values in at least two elements are exceeded. Results obtainedfrom testing
irradiated specimens from surveillance and R and D programmes of a variety
of materials, show favourable behaviour which can be traced back to the
beneficial influence of restriction of the same elements.
The qualification of materials was conducted in close cooperation
with industry and research institutes. Consequently, material types were
improved or rejected, when their safety proved not to be up to standard.
2.1.2. Melting technology
New advanced chemical compositions /17 - 19/ and mainly the reduced
content of minor and tract elements made it necessary to acquire either
 22

improved or even new melting technologies. Steels can be fairly well de-
oxidized by vacuum carbon deoxidization (VCD), and the hydrogen content
in the melt can be reduced considerably by stream degassing. Thus, flake
'".racking can be avoided. The sulphur content can be decreased to values
as low as 0.008 % or more to reduce directionality of toughness and to
avoid hot cracking during welding. This was achieved with the low sulphur
process or electro slag remelting /21 , 22/. Among the low sulphur processes,
the calcium argcn (CA) treatment can be successfully employed without much
expenditure. With the help of this method, it is feasible to fix or trans-
fer a great deal of the sulphur into the slag, the remaining sulphur chang-
ing to a globular structure. Furthermore, ingot segregation, and conse-
quently, inhomogeneous distribution of alloying elements can be drasti-
cally reduced by multiple casting with decreasing concentration of
alloying elements (multipouring, afterpouring), as well as electro slag
remelting. In the case of larger ingots, the control of cooling leads to
circulation and equilibrium in the residual melt that contributes to reduce
segregation and segregation cracking.

2.1.3. Forging and rolling

All subsequently mentioned factors and requirements have to be
considered in the forging procedure sheet during the forging of ingots to
semi-finished products. The degree of deformation perpendicular to
possibly existing discontinuities has to be sufficient to remove them.
The combination of tool shape and geometrical arrangement of dies and
fogging has to ensure that no tensile stresses occur in the ingot core
because of deformation. In the final transformation passes, fine-grained
structural development can be achieved by appropriate temperature control
during forging. However, the main aim of forming is to work the material
thoroughly and uniformly / 8 , 23/. This ensures that the transverse tough-
ness is enhanced as tar as possible, and means that the directionality
is reduced. Fig. 6. The same considerations are leading to a combined
forging-rolling process i i the fabrication of plates /24/. Forging (up-
setting) prior to rolling has brought about excellent values of the re-
duction of area after fracture as a result of a well deformed core region
in the ingot /17/.

2.1.4. Heat treatment

Highly tough, fine grained materials will be tne result of an opti-
mum adjustment of temperature, holding time and cooling rate during
quenching and tempering. Therefore, the stability of structure, es-
specially, the precipitates, can be guaranteed up to the end of life of
the component under service temperature. There is a possibility of en-
hancing the toughness of materials having non-optimized analyses by
annealing with partial austenitizing. This will cause a new micro-
structural constituent in the area of former austenite boundaries /25/
due to short-term heating sequence of this process may improve further
the toughness of the material, Fig. 7. However, there are still not
enough results available concerning proof of the long-term stability of
such structures.
 23

2.2. Toughness and Weldability as a Function of Development of

the Base Material
In materials for reactor pressure vessels and steel containment
vessels, the above mentioned improvements were subsequently employed. This
caused the change of micro-alloyed material types with relatively high
content of minor elements to steels without micro-alloying elements with
extremely low sulphur content 20 MnMoNi 55 and 15 MnNi 63. The influence
of material toughness, as a result from this development is for RPV
materials very well demonstrated by a large number of substantial melts
with an overall weight of 1000 t in the Research Programme "Integrity
of Components (FKS). The selection of melts in the F KS ranges from re-
jected semi-finished products with "historical" chemical composition and
production to 20 MnMoNi 55 melts having been produced under consideration
of all optimization measures. Investigation are conducted to delineate
the lower boundaries of specifications for chemical composition and tough-
ness with material qualities falling significantly below this boundary.
Therefore, they are adequate for demonstration of the effects of forging
and heat treatment.

2.2.1. Revised specification for chemical composition

In Fig. 8 the former boundaries of the chemical composition of
22 NiMoCr 37 and 20 MnMoNi 55 are compared with newly established values.
The range of the chemical composition of materials investigated in the FKS
programme can also be found in Fig. 8. KS 01 serves as an example for the
chemical composition of "historical" 22 NiMoCr 37 melts (rejected RPV
shell course), KS 04 (steam generator tube sheet) for the optimized
22 NiMoCr 37 melts, and KS 17 (RPV shell course) for 20 MnMoNi 55 of the
latest generation. The chemical compositions of materials representative
of RPVs investigated in the FKS and of steel containment vessels (non-
optimized WSt E 51 and 19 Mn6, optimized 15 MnNi 63) are listed in Tables
3 and 4.
2.2.2. Influence of optimized chemical composition and
production on material toughness
The upper shelf of the notch impact energy of materials ranges from
the lowest (40 J) to the highest (>200 J) values, Figs. 9 and 10. The
reduction of area at fracture of 20 % and the upper shelf of the notch
impact energy of 45 J of materials which do not attain the lower bounds
are to be compared with 70 % and 200 J of an optimized 20 MnMoNi 55 melt,
Fig. 11. The NDT-temperatures range from -30 °C to +30 °C.
The Cv-temperature curves (upper shelf energy and temperature tran-
sition) of FKS melts in Fig. 11 represent a dependency on the chemical
composition and manufacturing condition. They/cover, in a wide range, all
requirements of the basis safety concept. A toughness in the transverse
direction, comparable with the end of life condition, caused by neutron
embrittlement in the core region has been found in the test melt KS 07A
(Basis 22 NiMoCr 37). This is due to exceeding the lower chemical com-
position boundary and by significantly changed melting, forging and heat
treatment tech-^logy. The "historical" melt KS 01 (22 NiMoCr 37) lies
 24

just below the new 100 J-minimum requirement of the basis safety concept
for upper shelf energy. This concept is supported by numerous tests con-
ducted on this material in the field of conventional material testing as
well as representative large scale specimen testing.
Both of the 20 MnMoNi 55 melts matching the optimized chemical com-
position and production technique form the upper limit of the toughness
spectrum with upper shelf energies of 160 J and 200 J, respectively.
Critical fracture toughness values close to the boundary curve of ASMEXI
occurred in the vanadium alloyed RPV steel 15 MnMoNiV 53, Fig. 12, for-
merly used.
The influence of chemical composition and manufacturing technology
can also be noted in parameters of post-yield fracture mechanics. Ex-
tremely low crack resistances (T-modulus, T = 8) are linked with low
crack initiation values (J| = 50 N/mm) for KS 07, Fig. 11. The non-
optimized material for containment steel vessels 19 Mn 6 has a similar
jR-curve and a comparable upper shelf energy in the notch impact bend
test (approx. 50 J ) . The crack_initiation and T-modulus are already in-
creased to Ji = 170 N/mm, and T - 300, Fig. 11, for a 20 MnMoNi 55 melt
(KS 14), and the corresponding weld metal with a medium upper shelf energy
ranging from 160 J to 200 J in the notch impact bend test. The crack tip
opening-resistance-curves in Fig. 11 show a similar course.
Steels having low purity and high sulphur content exhibit a decrease
in transverse Cv-toughness with increasing degree of deformation during
forging (max 10 %) /26/. This was also determined by another author in a
highly tough 20 MnMoNi 55 material, Fig. 13, /27/. The F KS melts KS 07 A
(22 NiMoCr 37 modified) and KS 15 (20 MnMoNi 55 modified) have the same
tendency up to an extremely high degree (30) of deformation. The simul-
taneously increasing values in the main forging direction results in ani-
sotropy with a longitudinal: transverse ratio of 4 : 1, Fig. 13. The
upper shelf of the notch impact energy in the transverse direction re-
mains constant when increasing the degree of deformation from 6 to 10 in
an 22 NiMoCr 37 melt KS 02 with .006 % sulphur content, whereas the C v -
energy in main forging direction increases by 20 %. An enhanced upper
shelf toughness in the transverse direction was required for forging an
ingot of KS 05 (22 NiMoCr 37 with .011 % S ) . An improvement of about 30 %
from 95 J to 125 J could be achieved by reversing the forging direction,
i.e. deformation perpendicular to the previous main forging direction. The
influence of upsetting on the transverse toughness is especially obvious
in the lower boundary melt KS 15. Forging with upsetting resulted in 100 J
upper shelf energy, i.e. 100 % increase, whereas forging without upsetting
only changed the low upper shelf notch impact energy in transverse direc-
tion of 50 J insignificantly. The low initial upper shelf energy as well
as the high transition temperature of a forged plate of 22 NifiOCr 37 test
melt containing .014 % sulphur could be improved by adequate reforming,
Fig. 14. In spite of the non-optimized chemical composition, a material
toughness of 150 J upper shelf and 68 J at 0 °C could be attained by
additional tempering with partial austenitizing, and was comparable with
that of considerable purer materials.

2.2.3. Improving weldability

The course of creep rupture curves resulting from constant load
 25

tests on HAZ-simulated base material, Fig. 15, puts emphasis on in-

creased toughness after overheating due to optimized chemical compo-
sition. While the "historical" 22 NiMoCr 37 melt has the minimum of
elongation at fracture at 0.5 %, it is increased up to approx. 1.5 %
for KS 04 (22 NiMoCr 37 optimized) and to 6 % for a 20 MnMoNi 55 melt
comparable with KS 17. The lower boundary of the scatter band for
22 NiMoCr 37 sheds light on the susceptibility of the non-optimized
22 NiMoCr 37 melts to strain embrittlement and stress-relief cracking.
However, as for 20 MnMoNi 55 material, already at the minimum value of
£Br = 1-2 % any embrittlement, or crack formation during stress-
relieving is not to be expected. It can be assumed that creep strain up
to approx. 1.0 % is likely to occur in the original HAZ during stress-
relieving. Therefore, it is possible to have the KS 04 classified as
completely non-susceptible to stress-relief cracking, because the notch
impact energy exceeds 150 J at this creep strain, Fig. 16. No decrease
in toughness can be observed for the optimized 20 MnMoNi 55 melt even at
1.5 % creep strain, though microcracks are already present. On the other
hand, considerable differences can be found in the Cv-toughness of
materials for steel containment vessels after previous loading in the
short time constant load test. The micro-alloyed steel W ST E 51 shows
a strong decrease in impact energy caused by temperature induced em-
brittlement, whereas the optimized steel 15 MnNi 63 retains its ex-
cellent notch impact energy of more than 160 J without any crack forma-
tion even at a creep strain of more than 2 %, Fig. 16.

2.2.4. Improving stabilized austenitic stainless steels

Sufficient toughness, especially also in the deposited metal and in the
HAZ had to be proved after annealing for austenitic steels.
Embrittlement and sensitizing of austenitic steels against intergranular
corrosion (IC) have to be investigated and guarded for overheated areas in
the HAZ. A simulation technique was carried out in accordance with the procedure
for fine-grained structural steels, since the HAZ of austenitic welded joints
is not available in a size required for removing specimens, because the compo-
nents (plates, pipes) are not very thick. The temperature-time-function of
overheating is orientated on the course measured on original joints taking
especially into account the peak temperature which is decisive for structural
development in stabilized qualities.
The conditions for hot cracking, reduction in toughness due to hot crack-
ing and precipitation of secondary carbides as a consequence of heat treatment
were investigated in extensive parameter studies on overheated specimens.
Another object of these tests was the IC-susceptibility as a function of the
annealing temperature and time.
Serious changes in the precipitation condition are the consequence of
dissolving niobium carbon-nitrides in stabilized steels. In this case, after
overheating only approx. 50 % of the previously precipitated niobium content
are present in form of an eutectic solidified phase, the residue remains in
solution. If annealing takes place in the relevant temperature ranges, second-
ary precipitation of chromium carbides and niobium carbon-nitrides, or Z-phase
(Fe, Mo, Cr) 4 Nb 3 N 3 will be caused. Hot cracking, embrittlement, notch sen-
 26

sitivity at elevated temperatures and corrosion sensitizing will be the con-

sequence. No embrittlement phenomena could be proved during short-time creep
tests, which can be compared with strain embrittlement and stress-relief
cracking in the HAZ of fine-grained structural steels.
The primary ferritic solidification of the structure in cooling down
from peak temperature is in the first place used as a remedy. It dis-
tinctly reduces the risk of hot cracking in the HAZ because of lower heat ex-
pansion in the high-temperature range and higher solubility for impurities
in ferrite. If the carbon-nitrides of the stabilizing element are as much as
possible coagulated, complete dissolution and reprecipitation on grain
boundaries can be prevented to a great degree. A low carbon content in con-
junction with stabilizing and adequate forging and rolling as well as heat
treatment technique can help to avoid the damage mechanisms mentioned.

3. OPTIMIZATION OF FACTORS WHICH INFLUENCE WELDED JOINTS

The most preferred welding methods employed for the nuclear reactor
field are submerged-arc and manual metal arc. For these methods a gradual
build-up of the seam by a great number cf beads is required. When depositing
sucha bead a time dependant temperature field with hiqh gradients and peaks up
to melting temperature will be developed in the adjacent material. In cer-
tain areas of such a heat affected zone, structural changes take place,
building up together with the cooling down process a residual stress field
of different intensity. The existing structure and residual stress con-
dition of this area are altered by further temperature transients.
It has been found that apart from hot and hydrogen induced cold cracks
an additional inter-crystalline type of crack in the micro and macro range
can occur in the coarse grained zone of the HAZ of such a welded joint
during stress-relief heat treatment. This is due to relaxation processes
during stress-relieving because the ductility of the material is not suf-
ficient in that structural condition. This phenomenon is known as stress-
relief cracking /28, 29, 30/.
Cold crack formation can in the main, be prevented by welding with
preheated electrodes and flux (at 200 °C to 300 °C) and by means of a
hydrogen anneal at approximately 280 °C for several hours immediately
following the welding process. The risk of hot cracking is generally
avoided by the above mentioned optimization of base material, in this
case, by reducing impurities (especially, sulphur).
The weldability of the martensitic steel X 20 CrMoV 12 1 is basic-
ally influenced by preheating temperature. In Cthe case of wall thick-
ness > 40 mm, a preheating temperature of 450 C is usually required
/31, 32/. This means that, the austenitic struct"-^ o f the beads is main-
tained during the complete welding process. The necessary intermediate
cooling below the martensitic point prior to stress-relieving has to be
performed for reasons of toughness. In this case, the notch impact energy
obtained in the deposited metal is not of the best giving 34 0 at 300 °C.
Better results are achieved with 64 J at 300 °C by "martensitic welding"
at a preheat temperature of approximately 250 °C. Only in this way it
is possible to fabricate joints without macrocracks in components of wall
thickness >100 mm. Due to the multiple annealing, the stress during
structural transformation is confined to the individual layers. However,
 27

the effect gained may again be lost if long-time annealing at high tem-
rature takes place. Fig. 17.
Apart from improving the properties of the base material, it was
necessary to optimize the HAZ by other measures. Fig. 18.

3.1. Optimizing Stress-relieving Temperature

Welded constructions of a wall thickness of more than 30 mm and
38 mm, resp. have to be processed by stress-relief annealing in the FRG.
Prior to final annealing, it is possible that intermediate stress-relieving
procedures are conducted once or several times. In this connection, either
the whole component or a larger area adjacent to the weld are heated to
a temperature depending on the material between 500 °C and 620 °C. An-
nealing is generally conducted in a furnace, and in the case of individual
joint or joints at the construction site mainly by local induction heating.
As shown in Table 5 stress-relief annealing for RPV anc! steam gener-
ators are mostly conducted in stages with one or several intermediate an-
neals at temperatures ranging from 530 °C to 550 °C, and a final anneal at
about 600 °C. The holding period lasted from 5 hrs to 25 hrs (as a rule
2 min/mm wall thickness). Heating and cooling rates range from 15 °C/h
to 40 °C/h. Intermediate annealing is considered necessary to restrict
distortion and to avoid shrinkage cracking. If welding repairs are still
necessary after final annealing, it has to be repeated.
With the help of the welding simulation technique already described,
it is possible to state an optimal stress-relieving temperature. The in-
vestigation described in /16/ confirmed that it is not opportune to choose
a temperature of 550 °C.
Therefore, based on results obtained from the investigation carried
out up to now, a general increase in the annealing temperature for non-
optimized materials is recommended, provided that the mechanical prop-
erties will not be too much reduced /31/. The suggested annealing tempera-
tures concerning the most important materials are listed in Table 5. In
this connection, it has to be mentioned that the temperature interval
between 500 °C and 550 °C should generally be traversed as fast as possible.
Therefore, in the case of thick walled components, it is proposed that a
two-stage final anneal with a longer holding period at 475 C is conducted
to obtain an improved temperature equalization. But it has to be born in
mind that it is not necessary in all cases to anneal basic safe materials
with high toughness for the purpose of stress-relieving.

3.2. Optimization of Welding Parameters

In spite of all the progress achieved in increasing the toughness and
ductility by optimizing the chemical composition of the base material and
of the stress-relief anneal, it has to be noted that, from an under-
standing of the fact that in the overheated zone, in which the grain
structure has not been refined (coarse-grained zoen), high tensile residual
stresses meet with reduced deformability, there arises the demand for a
welding seam that is free of coarse-grain structure as far as possible. In
order to meet this demand, the marginal conditions leading to the forma-
tion of a coarse-grain structure should be known. Therefore, temperature
 28

measurements were carried out in the heat affected zone for determining
the local and time-dependant temperature field in the course of numerous
submerged-arc test welds with varying joint shapes in plates and forgings
of 27 mm to 270 mm thickness, fabricated from St E 51, 15 MnNi 63,
22 NiMoCr 37, 15 MnMoNiV 53, and X 20 CrMoV 12 1 steels /31, 33/. The heat
input ranged from 16 KJ/cm to 32 kJ/cm. The chemical composition of these
materials are listed in Table 6 and the welding data in Table 7.
Temperature-Time-Curves, which are shown by way of illustration for
one measuring point, Fig. 19, in a 250 mm thick seam in Fig. 20, were ana-
lised in respect to their rise time to peak temperature T m a x , Fig. 21, as
regards the duration of tjj for which the temperature was exceeded, Fig. 22,
and in respect to the peak temperature T m a x as a function of the distance
from the fusion line,Fig. 23.
Moreover, it was found that the achieved peak temperature is inde-
pendent of
1.. the heat input in the range of 16 kJ/cm to 32 kJ/cm,
2. the wall thickness in the range above 27 mm,
3. the thermal conductivity of the material at 20 °C in
the range of 29 J/m s K to 44 J/m s K and,
4. the actual height within the partially completed jointduring welding.
On the other hand, the peak temperature is directly r'opendant on the
preheating temperature.
In the investigated range, the analysis of the cooling time & t (800 C C
to 500 °C) gave times between 9 sand 21 s, Fig. 24. The cooling time A t
(1300 °C to 1000 °C), Fig. 24, and the duration time tjj above 800 °C,
Fig. 25, as a function of the heat input were taken as a measure for degree
of austenitization of the heat affected zone and coarse grain formation.
Thus, the region of the HAZ is subjected to temperatures above 800 °C for
4 to 15s. This time is obviously sufficient for completely austenitizing
the structural area. If the cooling lines that occurinreal welds are plotted
on a TTT-diagram (1300 °C austenitizing temperature), specially produced for
welding simulation, 45 % to 70 % of the structure will be bainitic, Fig. 26.
It could be shown that the theoretically determined areas of microstructures
apDroximately agree with those found in the HAZ of real joints.
From the knowledge of the achieved peak temperature as a function of
the distance from the fusion line, the distances of isothermal lines
around the individual welding layer of any arbitrary number of submerged-
arc weld beads can be furnished. Thereby, it is possible to determine the
temperature history for each point in the HAZ.
It was found that, the decisive temperature for the grain size, is
that which finally lies above 800 °C, Fig. 27.
Therefore, the temperature transient on one side of the joint is of
decisive influence on the coarse grained portion of the heat affected zone.
This is a consequence of the positioning of beads and bead sequence, which
again is dependent on the shape of the welded joint.
The welding process can be optimized experimentally with regard to
the peak temperature sequence in the HAZ either by means of numerous test
welds or by theoretical modelling of a welded joint.
A computer programme "SNB" was established giving due regard to geo-
metric marginal conditions, Fig. 28. This programme is capable of comput-
ing a multi-pass weld under realistic conditions.
 29

It is possible to evaluate nearly all the usual shapes of welded

joints. First of all, the programme calculates the number of beads to be
produced from the area of joint, the area of the bead, and the degree of
overlap1'', and with the help of the shape of the weld the provisional
coordinates of the centre of gravity of the beads are determined. The
pass is approximated by elliptical and circular components. The beads
are positioned by joining the provisional coordinates within the joint.
This ensures that, on average, the degree of overlap is obtained for the
complete joint.
The individual geometric marginal conditions will be met by shifting
and rotating the bead in a marginal area. The coarse-grained zones and
their modification are calculated with the help of the function T m a x = F ( A )
by subsequent beads in the area of the side of the joint.
The result for a welded joint calculated by the programme "SNB" is
represented in Fig. 29. The basis was the shape of welded joint and data
of the real joint shown in Fig. 20.
With the help of this programme it could be confirmed that it is
possible to achieve with the narrow gap welding technique already em-
ployed in practice /35/, a considerably reduced coarse grained area in
the HAZ. However, the angle of the gap should be 0° as shown in Figs. 30
and 31. Originally, this process was developed more than twenty years
ago and used successfully by Brown Boverie and Cie for fabricating welded
turbine shafts /36, 37/. Narrow gap welds, Fig. 32, for a German 1300 MW
PWR were first manufactured in USA. Recently, by reducing the angle of the
gap from 3° to 0°, a 650 mm thick joint for a RPV of a 600 MW HWR power
plant in which the coarse grained portion was less than 5 % was-produced
in Germany, Fig. 33.
To ensure a high quality of the joint, the KTA Guide Lines in Germany
require the most important welding data such as preheat temperature,
current, voltage, rate of deposit, etc. to be continuously recorded during
fabrication /38/.
The successful optimization of welding process, heat control and
use of tough filler material, and consideration of the safety requirement
as well, leads ultimately to the production of shape welded pressure
vessels or turbine rotors consisting only of weld metal. In this con-
nection, components up to 73 t weight (in future up to about 400 t weight),
fabricated with the submerged-arc welding method are being employed on
backing-up equipment /39-41/.
Shape welded components, especially those of large wall thickness,
have the advantage of homogeneous and isotropic material properties. For
instance, the yield point in the shape welded material 10 MnMoNi 55 is at
least 10 to 20 % higher both at ambient and at service temperature, and the
ultimate tensile strength is 10 % higher than in the MnMoNi 55 for plates
and forged materials. These favourable strength properties are complemented
additionally by excellent material toughness with a NDT-temperature of
-50 °C (compared to -30 °C), and an upper shelf toughness of 175 to 200 J
at temperatures higher than + 0 °C. The Cv-temperature curves of the shape
welded materials are in the upper scatter band of material 22 NiMOCr 37,
Fig. 9. Moreover, the fracture mechanics characteristics of deposited
metal are in line with those of optimum 20 MnMoNi 55 melts, Figs. 11 and
12. Apart from these advantages, it is possible that materials of poor
weldability can be processed by shape welding. Furthermore, areas in a
 30

component can be adjusted to the stress state and environmental conditions

by using various filler materials.
The application of shape welding can consequently be brought into
practice for heavy section large scale components with the benefit of both
safety and economy.
 QUALITY DEGRADING TOUGHNESS DEGRADATION CRACK FORMATION
INFLUENCES

PRODUCTION - UNFAVOURABLE CHEM. - SEGREGATION CRACKING

COMPOSITION - LAMELLAR TEARING
- NON-OPTIMIZED - FORGING AND ROLLING
HEAT TREATMENT DEFECTS
- SEGREGATION - HOT CRACKING
- HARDENING - COLD CRACKING
- TEMPERATURE INDUCED - STRESS-RELIEF CRACKING (SRC)
EMBRITTLEMENT (TIE)
- STRAIN INDUCED
EMBRITTLEMENT (SIE)

SERVICE - TEMPERATURE INDUCED - INAPPROPRIATE LOADING

EMBRITTLEMENT - FATIGUE
- STRAIN INDUCED - CORROSION
EMBRITTLEMENT
- CORROSION FATIGUE
- NEUTRON IRRADIATION
- CREEP
- HYDROGEN

Table 1: Quality degrading influence on toughness and crack formation

 wt %

C Si Mn P S Cr Mo Ni Cu

22 NiMoCr 3 7 -.20 .20 + .85* -.008 -.008 f.40 ^.55 1.20+ £.10

.15 .10 1.15 .40 .45

20 MnMoNi 5 5 -.012 £.008 -.20
.25 .35 1.55 .55 .85 (<-'.10)

.12 .15 1.20 .50

15 MnNi 6 3 £.015 -.005 -M5 -.05 -.06
.18 .35 1.65 .85

wt %

continuing Al V Sn N As Sb Ta Co

.010
22 NiMoCr 3 7 *.01 -.011 -.013 -.015 £.005 -.030 ^.030
.040

.010
20 MnMoNi 5 5 -.02 -.011 -.013 ^.025 - -.030 ^.030
-040

.02
15 MnNi 6 3 ^.02 ^.01 *.O15 -.015
.05

guiding value ( ) belt I ine region

Table 2: Requirements for chemical composition (RPV and Steel Containment

Vessels) according to the "Basic Safety Concept"
 22 N i M o C r 3 7

LADLE: ANALYSIS wt %

C Si Mn P S Cr Mo Ni AT Cu V Sn Co As Sb Ta

KS 01 .23 .20 .73 .015 .014 .42 .74 .95 .032 .12 .01 .01 .016 _ _ .01

KS 02 .19 .20 .93 .008 .006 .50 .56 1.29 .016 .10 - .007 .014 .028 - .01

KS 04 .21 .17 .66 .007 .007 .41 .57 1.16 .01 .09 .01 .008 .015 - - .01

KS 05 .24 .20 .69 .010 .011 .48 .75 .71 .008 .19 .01 .011 .016 .021 .005 -

KS 07 A/B .030 .29 .63 .021 .021 .46 .99 .74 .004 .26 .05 .011 .017 .027 .004 -

20 MnMoNi 5 5

KS 14 .21 .30 1.27 .006 .007 .13 .52 .74 .026 .08 .01 .006 .013 .01

KS 15 .24 .32 1.40 .014 .014 .15 .51 .59 .016 .15 .015 .007 .010 .016 .002

KS 17 .19 .25 1.38 .COS .006 .09 .50 .76 .026 .05 .01 - .011 - - .005

1)Modified Chemical Composition

Table 3: Chemical composition of the F KS heats

 Q
—i
<C
to
—<
t/i

3
+•>

•a:
~*-
..._,.,

LO
CO
<—
LO

z
c
o

<:
c_>
<t

^
-a

CM
CD
15 HnNi 6 3 min. 0.12 0.15 0.50 0.02

KO
CD
CD
in

CD
(VdTOV 1040) max.

CD
CD
0.35

CD
0.18 0.015

U")
C\J
0.15 0.05 0.85 0.05 0.06 0.015 0.02 0.004 0.015

O
CM
W ST E 51

*—
CD
CD
min. 0.10 0.50

CD
CD
(VdTOV 358)

CD
CD
max.

r»s
0.20 0.50

ro
CM

CD
CM
0.035 0.95 0.020
34

CD
ro
19 Hn 6 V min. 0.35

CD

CD
CD
CD
^r
(VdTOV 369) max

r*-.
0.23 0.60 0.040

&
o
E
C

01
o

03
a.
t/1
a>
STRESS RELIEVING
MATERIAL ANNEALING TEMPERATURE °C, (f) HEATING- AND (i)COOLING RATE °C/H APPLICATION
UP-TO-NOW OPT. MATERIALS PROPOSAL (NOT OPT. MATERIALS)

INTERMEDIATE ANNEALING (I) TWO-STEP Fir- AL ANNEALING F 2

WITH ONE- STEP FINAL ANNEALING (F
1
22 NIMOCR 37 (I) 530 - 550 f20 - 40 • 20 (F2) 4/5 • 20 REACTOR PRESSURE
VESSEL, PIPINGS
20 PINMONI 55 (F2) 595- 620 • 20 - 40 » 20 630 - 540 f 30 <30
STE 51 (FG51W)1} 530 - 530 * 60 * 60
620 - 640 f 60 H O CONTAINMENT
(BH51) 560 - 600 4 60 • 00
15 MNNI 63 530 - 580 4 60 * 50 580 - 590 f CO 160
15 NICUMONB 5 1 5 550 - 600 4 60 • 60 500 - 620 4 60 * 60 PIPINGS

1}
NOT USED IN NEW EQUIPMENT

Tab!e 5: Employed and proposed s t r e s s - r e l i e v i n g heat treatment f o r

low alloyed materials
 36

to
J j en

o ,, 1
O>
3
:- O

Ol

• ;

• - :

>
C
o

: i
o
6 6 Q-
-; ) 0
O
U
D
J : j 0 O

J 0

OJ
0
- •
r
_

f-

•T: tr. QJ
:
3 ?
1 £ M >

6
>

E:
o
0
c 'c
i>
. - 1
p
 i-.,,-.u: • • • • • - •

i ' - : • • :

:.': .;rHo- U- 0 •+ C-
- • • •

-CM'. -
., it. - -•.•- c
j r j-i.v:

'•'-C-'O •<
Ars J:I :-
.. R:;-:

Lii : : i A ""'.'Ox ; G O G tyo 0. ' • .?

'/
:.;.• in B >: , r 0 Ea'.-kiinr o=o 0. -'
Sur.
«; I'/ '.WOx'.OOO .X

kioI.XQ-K •y.rJ-lo.N'iV il.OOxlOOO Tulit- Cr'iOTT 0," ; •"Cv ; o o

Kt." £:.'jpei

•G

L* iiLNPjO-H 3O-,5x7?5 one-sided :-?o 29 7.0

X250 Tulip
cr WP77 W 111 5" i000x1000 2/ 5-* 25/"0 Oi>-^ 1 5 OPilTT 5>-o 70 ^ '• 0,MJ
-o X27
UP5H 'i5HhHi65 •i 600x400 X 25/50 3?-Ni570 lT;a^ 'TT 600 PF o, ev 2r-. ' ?o,.": o
--]
X30, " 0 0 i?OCrNoV "i 000x1000 cne-sidod LWJ20 5'.0 m 26 0.59 ••f-.o
'.2-1 xlOO Tu 1 ip
;:2o/.?7o 1500x1000
x?70

cut oat from :"orf-ed section

1abie 7: Data of submerged - arc test weldings

 Qualitatsmindernde
Einflusse
OPTIMIZED - Zahigkeitsminderung
MATERIAL - Rissbildung
QUALITY DEGRADING
INFLUENCES
- TOUGHNESS DEGRADATION
T> -CRACK FORMATION

SAFETY MARGIN

Beanspruchungs-
erhohende Faktoren
- Eigenspannungen
-Storfalle
STRESS RISING INFLUENCES
OPTIMIZED
DESIGN AND
SERVICE
R
-RESIDUAL STRESSES
- U P S E T CONDITIONS

K
Beanspruchung Belastbarkeit
LOADING LOADABILITY

Figure 1: Safety margin of components and systems

 39

Belastbarkeit B
LOADABILfTY B=B(Z.F)

Zdhigkeitsminderung Z
TOUGHNESS DEGRADATION

Fehlergrbfle F
FLAW SIZE

Fi g u r e 2 : Boundaries of loadability
 40

CO
<l§ HIGH UPPER SHELF CO

OPTIMIZATION j
LU
LL
<
CO
LOW UPPER SHELF

TEMPERATURE
CO
SE
KMA! 4UFACT

CO
UJ
IPERT

cc CD
Q_ CO
UJ UJ
CO
<I UJ

OPTIMIZATION
 41

C H E M I C A L C O M P O S I T I O N

SELECTION OF
ORE AND SCRAP

MELTING

FORGING AND
ROLLING

HEAT TREATMENT

TOUGHNESS AND FLAW CONDI- WELDABILITY OF

TION OF BASE MATERIAL BASE MATERIAL

WELDING PROCESS

TEMPERATURE CONTROL

ADDED MATERIAL

TOUGHNESS AND FLAW CONDI-

TION OF WELDED JOINT

INTEGRITY OF COMPONENT

F i gure 4: Influences on the integrity of c o m p o n e n t s

 S u s c e p t i b i1 i ty to Type of
G roup Steel
S1E or SRC TIE All oy

I CMn StE 36

x E MnNi 15 MnNi 63
II X(XX) XXIXXX) MaNi - V St E 43-51

X XXX(XX) MnNi - Ti St E 43-^-7

1 XI MnMo - , N i 20 MnMoNi 55

X XX MnMo-Nilfb 13 MnNiMoIfb 54

III XXX XXX MnMo-NiV 15 MnMoNiV 53

XX XX MnMo-NiV 15 MnNiMoV 53

XXX XXX MnMo-V 17 MnMoV 64

IV XXX IX(XX) X j NiMo-Cr 22 NiMoCr 3?

1
V -oooonxxxi —000 &1SI! Ni-GuMoNt) 15 NiCuMoIfb 5

SIE Strain-induced ombrittIement no susceptibility detected

SRC Stress-relief cracking x not very susceptible
TIE Temperature-i nduced embr i 11Iement xx moderately susceptible
mater i a I opt im i zed xxx very susceptible
material modified

Figure 5: S u s c e p t i b i l i t y o fl o w a l l o y e d s t e e l s d u r i n g p o s t w e l d h e a t treatment
 43

C,-SPECIMENS, REMOVAL : T / t

</>
FORGING WITH
UPSETTING
UJ
z
X
o
3
o
I—
FORGING WITHOUT T-L
UPSETTING

FORGING RATIO
Figure 6: Dependence of the material toughness on forging
technique and forging ratio

t FORGING WITH UPSETTING

AND PARTIAL ^ T
AUSTENtTIZATIONx / f 1
If)

Hi
FORGING WITH / / / F O R G I N G WITHOUT
X iiDcnTTiur.
UPSETTING. / . / / UPSETTING
o
ID
O /u
TEMPERATURE
Figure 7: Toughness as a function of forging
and heat treatment
 44

22NiMoCr37 %
15 0,15
UJ
1,3

Mn Mo
iT
Ni KS O r HISTORICAL" MELT
= KS04 OPTIMIZEO MELT
n FKS
0,10
n V0TUV365
UJ R BASIS SAFETY CONCEPT
o * GUIDING VALUE
2 0.5- V
0,05
o Cu s Al
a:
UJ 7s
0 7, 0

I
0/
/o 20 MnMoNi 55 %
Mn

i
X 0,15
O = KS17 OPTIMIZEO MELT
UJ
n FKS
m VDTUV TENT. OCT. 76
10 F] BASIS SAFETY CONCEPT 0,10
LLJ Ni
CD
Mo
0,5
o
UJ
a. Sn
0 0
Figure 8: Range of chemical composition of RPV materials
22 NiMoCr 37 and 20 MnMoNi 55
 FORGING SUPPLY ^22NiMoCr37(SCATTERBAN0
H Q 22 NiMoCr 37 RPV HEATS
ISO-V TRANSVERSE rzz20MnMoNi55 «

0
-100 0 100 200 300 °C
TEST TEMPERATURE
Fig. 9: Scatter band of the Charpy-V-energy of RPV steels
 46

v-SPECIMENSJRANSVERSE

-—15MnNi63
19Mn6
-40 0 40 80 °C 120
TEMPERATURE
Figure 10: Scatter band of Charpy-V-energy of Steel Containment
Vessels
 47

TOUGHNESS VALUES
SYMBOL MATERIAL MOT-T
KS17 20MnMoNi5< - 3 0 °C
KSU 20MnMoNi55 -20°C
KS01 22NiMoCr37 * 5°C
/ / / • KS07 22NiMoCr37 + 30°C
M00IFIE0
-50 50 150 °C 250
TEMPERATURE
1000 TENSILE-TEST 800
N —SPECIMENS CT-50-
SPECIMENS
=20% mm
15MnNi63

0 10 20 % 30 0 0^ 0,8 mm 1,2
STRAIN e CRACK EXTENSION Aa
0,8
mm

0 2 £mm 6 0 2 J>mm6
CRACK EXTENSION Aa CRACK EXTENSION Aa

Figure 11: Toughness values of materials for RPV and Steel

Containment Vessels
 48

6000
• KSO1 NOTT =
• KS04 NONSEGREGATED
N/mmw D KS04 SEGREGATED
O KSO7 A/B
A KS12 (ESR)
5000 O KS12 (conv.)
« KS13
A KS15
• KS21
<>15MnMoNi V 53
UI00 ® BUILT-UP WELDING
10MnMoNi55
CO
U-J

3000

2000

1000

-KJO HOTT K 100

TEMPERATURE RELATIVE TO NDT T

Figure 12: Fracture toughness, specimens up to 10 T-CT

 49

ORIENTATION MELT
o • 20 MnMoNi 55, JSW
• • Mn = 837oiP=.0257o,S=.0237o(n.C0UPETTE)
A A 22NiMoCr37(KS02)JFKS
® * 22NiMoCr37modified(K$07A/B),FKS
300 - H a 20 MnMoNi 55 modified (KS15), FKS -
C t - SfECIMEBS, KMOWl ;T/».

J
o
LLJ

LLJ
200
I—
O

100

0 2 t 6 8 10 20 30
FORGING RATIO
Figure 13: Dependence of material toughness on the forging ratio
 50

AC3

TIME
200
J KS16, 22NiMoCr 37, modified

100

MAIN FORGING
DIRECTION
0 I

-200 0 200 °C 400

TEMPERATURE
Figure 14: Influence of forging and heat treatment on the
material toughness
 51

COARSE GRAIN CREEP RUPTURE CURVES

CYCLE 13 22NiMoCr37
CONSTANT LOAD 22 NiMoCr37 OPTIMIZED
TEST AT 610 °C 20MnMoNi55
SIMILAR TO KS17
T=610°C A T=580°

22NiMoCr37
NOT OPTIMIZED

0,1 100 h

Figure 15: Creep rupture curves - scatter band of RPV steels

 52

COARSE GRAIN OVERHEATING CONSTANT LOAD TEST

22NiMoCr37
22 NiMoCr37 OPTIMIZED 610°C/6h
20MnMoNi55
15MnNi 63 580°C/2h
Cv -TESTS AT 320°C and 200 °C
15MnNi63 (200
2 160
LLJ SIMILAR TO
,KS17
^ 120
O

80
X
o
22NiMoCr37
NOT OPTIMIZED
7c 15 2,5
CREEP STRAIN £ r
Fi gure 16 Post damage tests
 53

100
DEPOSIT MATERIAL
X 2 0 CrMoV 121
OPTIMIZED
CHEMICAL COMPOSITION

75
MARTENSITIC"
WELDING
Hi
LU
50
O
LONG TIME
ANNEALING
I 550°C/10000h
>•

o
25

0 25 50 75 100 °C 125
TEMPERATURE
Figure 17: C -T curve for welded joint
made of X20 CrMoV 12 1 steel
 OPTIMIZATION OF THE HEAT AFFECTED ZONE

REDUCTION OF THE REDUCTION OF

INCREASE OF
COARSE GRAINED RESIDUAL
TOUGHNESS
PORTION STRESSES

OPTIMIZATION OF THE OPTIMIZATION OF OPTIMIZATION OF THE

TEMPER-BEAD
PEAK TEMPERATURE STRESS RELIEVING CHEM.COMPOSITION
TECHNIQUE
SEQUENCE OF BASE MATERIAL

WELDING PROCESS TEMPERATURE

SHAPE OF WELD
HEAT RATE

POSITIONING
OF THE LAYER COOLING
RATE
BEAD SEQUENCE

Figure 18: Measurements for optimizing the mechanical properties of

the heat affected zone
 1750
°C MEASURING POINT 1

1500

1250
LU
(Z
Z)
1000
<
on
LU U1

Q_ 750
LU

500

250
TEMPERATURE DURATION A B O v K COOLING TIME
RISING TIME TEMPERATURE ^ At800/500°C
0
0,5 1,0 5 10 50 s 100
TIME
Figure 19: Temperature "history" for measuring point 1 during welding
 56

BEREICH MIT
THERMOELEMENTEN

REGION WITH
THERMOCOUPLES

Figure 20: Building-up of test weld S 2 RN

 LEVEL A ABOVE
CENTER OF GRAVITY OF LAYER S
AREA OF
LEVEL B BELOW WELO
CENTER OF GRAVITY OF LAYERS LEVEL A
Tmax
LEVEL B

SUBMERGED-ARC TEST WELDING

LONG. WELD III A III B
WALL THICKNESS mm 150 150
HEAT INPUT kJ/cm 319 27,9
MATERIAL 22NiMoCr37
PREHEAT TEMPERATURE 180 °C

0,5 10 2 5 10
RISING TIME OF TEMPERATURE
Figure 21: Rising time to peak t e m p e r a t u r e (test w e l d i n g LN III A , B )
 1500
°C
ATK
1250 1000
900
800
1000 700
1
01
600
LU
Q_ 750 500
LU CD

300
< 500 TIME
LU SUBMERGED-ARC TEST WELDING 200
Q_
LONG.WELD IIIB TltU \ HEAT INPUT 100
LONG. WELD III A 31,9 j kJ/cm
250 MATERIAL 22NiMoCr37
PREHEATING TEMPERATURE 180 °C
0 j I
1 2 5 10 20 50 s 100
DURATION t•uH ABOVE TEMPERATURE
Figure 2 2 : D u r a t i o n o f t e m p e r a t u r e i n t e r v a l A T i n c l u d i n g peak t e m p e r a t u r e
 S U B M E R G E D - A R C TEST WELDING
A V E R A G E CURVE
HEAT INPUT 1 5 - 3 2 kJ/cm
PREHEAT TEMPERATURE 150 - /»OO °C
WALL THICKNESS 11 ~ 250 mm
- 15D0*Ty / 1500-Ty \ • u /, _A_

0,5 1 2 5 10 20 50 m m 100
DISTANCE FROM FUSION LINE A
Figure 23: Peak temperature as a function of the distance from fusion line
and preheating temperature for submerged - arc test weldings
 40
kJ/cm ^800/500

30
LLJ

20
SUBMERGED-ARCTEST WELDING
C7»
o
22NiMoCr37 15MnMoNiV53
<
S1RN A LN III A a HLN 60-M
LLJ 10 D SUN A LN IIIB o HLN 60-D
x O WP38 15MnNi 63 © S2RN-W • LN IY a HLN250-M
/ s
• WP27 WStE51 9 II SS'n PREHEAT TEMP 150-200o0
O BHW3513MnNiMoNb5<*© S2RN-D i i

0 5 10 15 20 25 30 35
COOLING TIME A t
Figure 24: H e a t i n p u t a sa f u n c t i o n o ft h e c o o l i n g t i m e b e t w e e n 8 0 0 ° C a n d
5 0 0 ° C , a s w e l l a s 1 3 0 0 ° C a n d )000°C
 61

25 PEAK
s PREHEATING TEMPERATURE 150 -200 °C TEMPERATURE

1500
o °C
o U00
CO
LJJ 1300
o 15 1200
CD 1100
<
1000
O
10 900
5
Q

SUBMERGED-ARC TEST WELDING

-S1.S2.HLN6Q, HLN250
Hfflff
A-LNDIA
A-X20/100, X20/270

10 20 30 kJ/cm
HEAT INPUT E

Figure 2 5 : Duration above 800°C as a function of heat input and

peak t e m p e r a t u r e
 CHEMICAL COMPOSITION % 22NiMoCr37

AUSTENITISATION TEMPERATURE
1300°C
HEAT INPUT
LLJ E=32kJ/cri

C o o l i n g r e g i o n for s u b m e r g e d - arc t e s t w e l d i n g s in t h e
t e m p e r a t u r e - time - t r a n s f o r m a t i o n ( T T T ) - d i a g r a m
 63

1600
© S 2 RN-W A LONG. WELD
°c © S 2 RN-D ' ONLY ONE
1500 • S1 R U N HEATING CVCLE^
MATERIAL 22 NiMoCr 37

U00
E
LJJ
1300
3

01
LJJ 1200
GL
LU
1100

LLJ
1000

900

800
0 2,5 5,0 7,5 10
GRAIN S I Z E ( A S T M )
Figure 27: Grain size in the heat affected zone as a function of
the peek temperature
 CORNER

ASIDE
OVERLAPPING

MAX. DEGREE OF \
PENETRATION OVERLAPPING AREA
OF THE CORNER OF LAYER
Figure 28: Marginal conditions in a m u l t i - l a y e r welding
 65

Figure 29: Multi-layer welding computed with code "SNB"

 66

1100 °C
800 °C

Grobkornzone
COARSE GRAINED
ZONE (~307o)

Figure 30: computed with "SNB", angle of groove 1°

 67

1100°C

800 °C

Grobkornzone
COARSE GRAINED
ZONE H O 7.)

Figure 31; S e a m c o m p u t e d w i t h " S N B " , a n g l e o f g r o o v e 0'

 CD
C

Schweifigut W a r m e e i n f I uflzone
DEPOSIT MATERIAL HEAT AFFECTED ZONE
SchmalspaltschweiOnaht
NARROW GAP WELDING
Figure 32: N a r r o w g a p w e l d i n g f o r a G e r m a n 1 3 0 0 MW PWR,
3 6 0 mm w a l l t h i c k n e s s
 Kraftwerk Union

7cm 13,5 1cm 2 1 1mm 201

Narrow Gap Weld of 20 MnMoNi 55

Thickness of Weld: 650 mm
Figure 3 3 : 650 mm thick narrow gap welding for a 6 0 0 MW HWR
 WELDIUG TECHNOLOGIES AND MATERIALS
FOR ENERGY APPLICATIONS IN JAPAN

I. Masumoto

Nagoya University
Nagoya, Aichi-ken, Japan

K. Masubuchi

Massachusetts Institute of Technology

Cambridge, Massachusetts 02139

T. Fujimura

Technical Research Association

for Integrity of Structures
at Elevated Service Temperatures
Tokyo, Japan

INTRODUCTION

Japan with her little energy resources, will face a grave energy
situation in the 1990s. Japan had brought an economic prosperity de-
pendent on oil for energy since the 1960s. Then she was forced to change
her energy policy due to an oil crisis in 1978.
In Japan, before 1940 the increase rate of primary energy demand was
approximately 4.5%/year, but after 1950 it has become 9.9%/year, and
after the first oil crisis it has dropped to 4%/year, as shown in Fig.l.
The most important change in Japan's energy policy was the diversi-
fication of energy resources. An important energy resource in place of
oil is, of course, nuclear energy. Nuclear power stations of light water
reactors have been constructed, and their energy production will account
for 5% of the energy consumption and 15% of the electric power generation
in 1980. However, imported oil still account for 60% of the energy demand.
Japan's energy policy aims at reducing this figure to 45% by maintaining
the present level of oil consumption of 300 million kilo-liters per year
in 1990. For this purpose, the belated use of natural gas must be

71
 72

encouraged. However, long distance transportation and the development

of storage facilities for imported liquid gas pose great problems for
this island country.

c
UJ (1 S

ro 0,2

^ p r —i o i

Fig. 1. Primary Energy Status in Japan

On the other hand, although a speedy development of fast breeder

reactors is urged in the nuclear energy development plan, the plan has
been lagging behind by as much as five years, and the operation of the
first commercial plant is not expected before early 1990. The wider
utilization of coal is the only way to fill this energy gap. In Japan,
coal energy accounts for only 16% of the total energy consumption and
20% of the energy for electric power generation in 1980. These low
figures of coal consumption will have to be doubled by 1990. The large-
scale utilization of coal in this densely-populated, narrow and mountain-
ous land gives rise to great environmental problems such as the disposal
of ash and long distance transportation for import. Therefore, the
utilization of refining coal by coal conversion is called for. Although
at present the utilization of a coal-oil mixture is being studied as a
short term development measure, the import of synthetic oil will be
necessary in the long run. The development of this process jointly with
the United States and Australia will become an important issue for Japan.
The development of coal gasification is also necessary. In 1990, with
the increase in gas production utilities, utilization of coal in a gas
state will be 10 times that of the present. The development of a coal
gasification process, therefore, is one of the most important projects
together with the nuclear energy development program. The construction
 73

of coal gasification plants must bo planned in connection with the de-

velopment of high temperature gas-cooled nuclear reactors.
Research and development efforts related to energy applications
cover a wide range of subjects including construction of facilities for
transportation and storage of energy-related materials such as petroleum,
liquefied natural gas and so on, offshore oil drilling rigs, dee; sub-
mersibles for searching oil and gas under the deer sea, coal conversion
plants, just to name a few. Japan uses extensively up-to-date technolo-
gies on materials and fabrication.
This paper first presents an overview of research and develo; r^nt
efforts in Japan on welding technologies and materials for unt-roy a; pli-
cations.

IMPROVEMENT OF MATERIALS PROPERTIES

High Quality Steels for Nuclear Power Plant

The electric furnace steel making process h=is been completely re-
placed by the LD convertor process for the manufacture of higher nualir
and larger-scale materials. Steels made by the convertor include n:;ly
a small amount of impurities as shown in Fig. 2.

-IS i
- ir

"i 1
| J i j I
SSSs s 8

fcn fe-.ili: Irc^c

• - . " : . <

, ! !
J. LLI i I

Crygen (ffM!

Fig. 2. Cleanness of Heavy Section Steels Made in Japan

Fig. 3 shows the dimension limits of large-scale steel products.

Especially in recent years forging technology has made great progress
and, as an example, a high integrity pressure vessel, using large-sctle
ring forgings with no longitudinal welding joints, is being made, as
shown in Fig. 4.
 74

U S Hcjvy S e c !.on P l a t e

*•'•&

He'ding Seat

(b) R:ng forging

Conventional fldvjccd
Design Design
by Pistes by Ring
fc'.sr E.tsfir Finning Forging

Fig. 3 Dimension Limits of Fig. 4 Design and Fabrication

Large-scale Steel Products of of Large-scale Pressure Vessel in
Japanese Manufacturers Japan

2-r-Cr-l Mo Steel for Petroleum Industries

2-i-Cr-l Mo steel is the material used for large vessels such as

resolvers of commercial coal conversion plants. Its low strength is a
problem in structural design. An effective way to improve its strength,
especially its yield strength, is to change the heat treatment from
normalizing and tempering to quenching and tempering. It is extremely
difficult, in the steel manufacturing process or the fabrication process,
to maintain the quality of products by uniform heat treatment on large-
scale materials. Fig. 5 shows the characteristics of Japanese quenched
and tempered 2-A-Cr-l Mo steel (ASTM A 542 steel). It also shows the
excellent properties of welded parts.

Hastelloy for the Structural Components

at Extra-high Temperature Service

The materials for gas turbines of aircraft are used at high temper-
atures however they are small in size and their service time does not
exceed several thousand hours. There have been little experience in
manufacturing and using materials for plant components that are large-
 75

50
3 25
s 72
(b! Hjct;li'y vc S-.rrngth
j-20

|BO
15 Specified Value
fttt _a 3 3
!
Sp:c:(iti Vilue for Clrrngth

P?I •
Specified Value

|70J

.65
s tz.1%] \;-\
I eTiTied Value

Suffice Quarter Half Ouir'.er T.'l 1/4 ,L_L U

Thickness Thicknexx 50 E0 70 BO SO 100 110

Politico Petition

Roll Direction Trinsverj Pepotlt He»t-Rffected Tensile Strength

Direction Ketil Zone

Bi.-t Melil IfclrfcJ Joint

Fig. 5 Properties of 2 -j- Cr-lMo Steel Plate (Thickness: lOOitun)

scale and used for several ten thousand hours and over. Therefore, a
research project was organized in Japan to evaluate if Hastelloy-X,
selected from among many high temperature materials, could be adapted
for long service at higher temperature service.

Corrosion-Resistant Double Tube^

Recently Tight Fit Tubing (T.F.T.) has been developed to produce

the corrosion-resistant pipe for corrosive flow lines by a thermo-
hydraulic fit (T.H.F.) method.
T.F.T. meet particularly the serious requirements for the oil and
gas drilling wells.
The principle and the procedure of T.H.F. method is illustrated in
Fig. 6. In short, the accuracy of fitting is attained by the plastic
expansion of liner and the desired fit is achieved by the thermal
shrinkage of the tubing. Specific combination of heating temperature of
tubing and expanding pressure of liner can be computed when dimensions
and material properties of both pipes are given.

PROGRESS OF WELDING PROCESSES AND TECHNOLOGY

Application of Electron Beam (EB) Welding 3

 76

(1) Establishment of Qualification

Test for EB Welding Procedure by JWES
fined
A committee on Specification of
Electron Beam Welding to Pressure
Vessels (called SEP Committee) was
organized in the Japan Welding Engi-
neering Society (JWES) three years
ago. As a result of the committee's
activity, WES standard 1601(1981)
"Method of Qualification Test for
Electron Beam Welding Procedure" was
diometsr established as JWES's specification
initial gap - fir-in gap on July 1st, 1981.
This specification describes
Fig. 6 Principle of Thermo-
the requirements for EB Welding equip-
Hydraulic Fit Method
ments as follows: a) change of ac-
celerating voltage < + 2% of max., b) change of beam current < + 2% of
max., c) change of beam focus current < + 1% of setting value, d) change
of osillation coil current < + 5% of setting value, e) precision of weld-
ing jig; change of travelling speed < + 2% of max., and deviation of jig
< + 0.2mm, and f) capacity of vacuum system 7.5 x 10 ° torr.
Types of welding equipment are classified into three, i.e. full,
local and non-vacuum, and vacuum and non-vacuum are defined as
< 7.5 x 10~ 2 torr and 1 to 760 torr, respectively. Welding parameters
should be maintained within following limits: a) a change in the beam
current < + 5%, b) a change in the accelarating voltage < + 2%, c) a
change in the welding speed < + 4%, d) a change in the beam focus current
< + 3%, e) a change in the gun-to-work distance < + 5% and f) any change
of oscillation such as from using to no use.

(2) Application for Large Components

EB welding are successfully applied in such industries as automobile,

aircraft, heavy electrical power plant fabricator etc. Particularily
in the last one various large components are fabricated by high power
EB welding.
Fig. 7 illustrates the diaphram of steam turbines of the 150 -
1,000 m class, in which many nozzle are mounted in hollow disk, and high-
precision EB welding is applied to mount them. As EB welding can be per-
formed without preheating, higher precision is maintained.
For the critical plasma testing equipment JT-60 for developing the
nuclear fusion energy, now being constructed in Japan Atomic Energy
Research Institute, several thousands tons of high manganese non-magnetic
steel (18Mn-5Cr and 14Mn-2Cr-2Ni) is used for reinforcing frame of troidal
field coil.
For welded structures, high accuracy is demanded, and also stress
relief annealing is not applicable because of temper brittleness. Accord-
ingly, the EB welding was adopted because it features reduced welding
deformation and small residual stress.
 77

Since high manganese steel con-

Typical dimensions
tained oxygen and nitrogen in large P 20 -80mm
quantities, many voids were produced 0 800 -1700mm
PT 20 -50mm
when EB welding was made by pene- H 40- -300mm
T
trating the steel plate having a 5 ~20mm
thickness of 50mm-150mm.
Accordingly a steel plate for
EB welding was developed by re-
ducing oxygen to about 50ppm and
nitrogen to about 200ppm, and good
welding performances were obtained.
Table 1 shows mechanical proper-
ties of EB welded joint.
Fig. 8 indicates the rein- 5 0 - 1 0 0 Passes

forcing frame for troidal field coil (At Submerged arc welding
of JT-60. 4
A vessel, which is an accumu- Fig. 7 External View of a
lator for nitrogen gas service, Model of Diaphragm

having a diameter of 2.3m

and a plate thickness of 96mm
was made by EB welding of its
equator after setting two semi-
spherical shaped head plates. With
36 KW beam output and welding speed
lOOmm/min., piercing welding was
made by one pass of horizontal beam.
It took about 70 minutes to complete
this welding. Fig. 9 shows shape of
the vessel.
Electron beam welding procedures
have been established for joining
Fig. 8 Reinforcing Frame of 10Ni-8Co ultra-high strength steel
Toroidal Field Coil of JT-60 plates to be used for a pressure bulk
of a deep submergible capable of
operating at waters 6,000 meters deep. 5

Table 1. Mechanical properties of EB welded joint

for non magnetic reinforcing frame
of toroidal field coil of JT-60
| 2vEo of
'Tfield strsn jth'Te nsile slrsglh ' Elongation R.A. Permeability
! weld metal

• \

18Mn-5Cr 59 99 ; 38 41 ;<i.oi
I [
" " """i
t
14Mn-2Cr-2Ni | 38 1 84 j 47 53 ! 2 0 !<i.oi
 78

Shape of Ordinary maX

pr«ssur«. 20.6MPa
Work
Design pressure 21.1MPa
Design tamp. -26-+53C

Fig. 9 Pressure Vessels Fabricated by EB Welding Process

Main benefit of EB welding is

reduction of welding distortion in
comparison with TIG welding to fabri-
cate a real spherical shell. Full-
scale model is shown in Fig. 10.
The mechanical properties of EB
weldment of the mode are summarized
in Fig. 11 and 12.

Narrow Gap Welding

Japan Pressure Vessel Research

Council (JPVRC) Fabrication Division
has tentatively given a definition Fig. 10 Appearance of Full
to narrow gap welding (NGW) as Scale Model of a Pressure Bulk
follows: "The arc welding process ° P e r a t i n 9 a t Waters 6,000m Deep
of a narrow groove joint whose gap
is considerably small compared with the thickness of the plate to be
welded. Specifically the gap shall be not more than 20mm for plate thick-
ness 200mm and under, and not more than 30mm for plate thickness over
200mm. The welding process is limited to mechanized or automatic arc
welding process."
 79

f tail! =-.„

_
1 1
25

:o
(i
r
JlTl

fin r rt
s

| J.C T=
i I ] j
|
! ; !

I3O>
!
•o i I

0 T
w W Bond j HAZ l.[
loo; ,,;r
(
£t~ *: T ~J "' TIC
t 0 5 0
I Eq.jior j o . 1 1 i t n) i GO (8C 11 [_ (OOll | ' ;

T i

Fig. 11 Tensile Properties of Fig. 12 Charpy Impact Proper-

Weld Metal of Full Scale Model in ties of Full Scale Model in Fig. 10
Fig. 10

In general the NGW process was well known to have two major bottle-
necks, that is, lack of penetration for both sidewalls of the groove and
unstable quality of welded joints in case of heavy duty welding oper-
ation .
A number of Japanese fabricators have been successfully overriding
these formidable tasks and putting NGW into practical use for every
field of steel constructions. NGW processes practically used in Japan
are summarized as shown in Fig. 13 and classified basically into N-GMAW
(Narrow Gap Gas Metal Arc Welding), N-GTAW (Narrow Gap Gas Tungsten Arc
Welding) and N-SAW (Narrow Gap Submerged Arc Welding).
The development of N-GMAW process has been related with the im-
provement of special devices to give sound penetration on both side
walls of the groove. Development of N-GTAW process has been counting
on the improvement of welding equipment of higher precision for narrow
groove. On the other hand, development of N-SAW has been undergone in
terms or flux to give easily removable welding slags.
In Fig. 14 progress of development in Japan is summarized for
various kinds of NGW processes. It should be noted that various kinds
of N-GMAW processes have been developed successively since 1975 and
widely put into practical operation, and that N-SAW processes have been
realized recently and some of them are entering into the stage of prac-
tical operation. The applications in various fields and the kind of
processes of narrow gap welding are summarized as shown in Fig. 15.
It is worth noting from Fig. 15 that boilers and pressure vessels
ranks first (52.5%), and followed industrial machinery construction
(25%), ships & offshore structures (12.5%) and penstocks (10%). As far
as welding process is concerned, the most extensively used welding
process is GMAW-process (MIG and MAG), taking 75%, and followed by SAW
(20%) and GTAW(5%).
 80

[Cj r r o c c s s ~ i i t h i s c l l U l t r i g ] .l.COPSAP.
Arc , n of p.rveil e l t u r o d a u l r e I l P r o c e s « '
j •.- •: :

! l"ll|_':|^iI5^ t lL e I '?•»"" '

! 'KAT::./.!'

Fig. 13 Classification of Narrow Gap Welding Process in Japan

From the application of NOfl to

pressure vessels made of higher grade fn-JJiV-i n
•[;.'>::•!

steels; which are operated at high •:;a

pressure and/or elevated tempera-

ture, it is obvious that welded \?\
joints by N-GMAW and N-GTAW can be !
characterized to be excellent in !J I L
mechanical properties, particularly
in notch toughness, compared with
conventional welding processes such
as SMAW, SAW and ESW.
As usual practice, N-GMiftW is I N--.B
T
adopting a shielding gas of 8 0 ~ 9 0 %
Ar and CO2 from the view point of
assuring mechanical properties and
workability. Its heat input is as
low as 15~25KJ/cm.
It can be counted on N-GMAW P
that fine HAZ austenitic grain size
along fusion boundary is obtained
through the effect of low heat input
and arc oscillating process. >
One example of mechanical
Flg
properties of welded joint by N-GTAW 1 . : 1 4 Cumulated Number of
is shown in Table 2. Evidently, Applications and Publications of the
Develo
impact toughness of weld metal is P e d N G * Processes
remarkably high. The good toughness
as well as less hydrogen in weld metal allows N-GTAW for welding of high
strength steels of yield strength up to 90 kgf/mm^.
Temper embrittlement susceptibility is examined for welded joint of
2 -A— Cr-1 Mo steel by conducting GE type step-cooling test process.
 81

\ i.:-..-|

\L \

Fig. 15 (a) NGW Applications (b) Kind of NGW Processes Put in

in Various Industrial Fields, Col- Applications from the Inquiries
lected from the Inquiries

Table 2 Mechanical Properties of N-GTA Weldment

of ASTM A514 Steel
Tiv.r; 1 Oil
vE-50
T.S 2 El . , /j
R. A. I.uc:.'l tiou of k S £-ra(ff-lbs) kgf- u-lbs)
n-n £r;ic turf.'i (F.L.)
83 . ! ( 1.18) It 2 VI (W.M.) 23(166) J 16(116)

As noted from it, the shift in 40ft-lb transition temperature after

step cooling is less than 25 °C (45°F) for both welded metal and fusion
line. Embrittlement index (vTr40 + 1.5 AvTr40) is less than -38°C
(-36.4°F). Evidently it can be concluded that welded joint by NGW has
extremely high toughness against temper embrittlement.
Furthermore, impact toughness for 2-^— Cr-1 Mo steel is examined in
terms of heat input and post-weld-heat-treatment conditions, whose
results are shown in Fig. 16.

Laser Welding7

Laser surface treatments such as alloying and transformation harden-

ing are superior to other thermal surface treatments in being able to
treat a given localized area with the minimum amount of energy. A new
method for laser surface treatment using a uniform intensity rectangular
beam (UIKB) is shown here. Fig. 17 (a) and (b) show an optical system
for shaping the UIRB and a hardened part in a cross-section. From
Fig. 17(a),it is seen th&t a parallel beam with a nearly Gaussian mode
is focussed and converted into a rectangular beam with a uniform intensi-
ty (power density) in the direction along the width, by means of three
simple Cu mirrors.
 82

*C *C
•^ - J

(43. )kj/ln) PLTIT

-•-o—
JiVj - Ju - ruut+s.c.
-<9
tat. kj/ln)
~o-~ TMZ
*O'*J lOUj/lnJ PWHT*S.C.
-58 -IB 71 -it.5 - •+—
ciai
cue P»tifl

50

-120 -100 -80 -60 ~'i0 -20 0

Tes tinj» teinpijr.iture, "C

Fig. 16 Effect of Weld Heat Input on HAZ Toughness of N-GMA Weld-

ments of ASTM A387 Gr.22 C1.2 Steel (V-notch located at the middle of
HAZ)

Pig. 17 Cross Sectional Shapes of Iron Surface Layers Carburized

with Uniform Intensity Rectangular Beam (UIRB) and Defocused Circular
Beam (DCB)

Diffusion Bonding^

Based on the basic research, a full shrouded impeller with Cr-Mo

steel and 13Cr steel (150mm dia.) has been successfully fabricated by
diffusion bonding. Two segments, the hub and the shroud, were prepared
by machining and subsequently joined by diffusion bonding into a full
shrouded impeller. The impeller was evaluated by spin testing and
 83

metallographic examination. It worked successfully up to 58000 r.p.m.

and the bond line of the impeller was not observed at all in the metal-
lographic examination after the spin test as shown in Fig. 18.
, The established bonding tech-
nology is applicable to other
machine parts and metals although it
is necessary to carefully select
bonding temperature, pressure and
time to suit the characteristics of
the metal in question.

Diffusion Bonded Impeller

(Cr-Mo Steel) Submerged Arc Welding
of 2 J Cr-lMo Thick Plates
for Pressure Vessels without
Intermediate Stress Relief Annealing

In recent years, thicker materi-

als with higher strengths have been
Fig. 18 Microstructure of required to meet the needs of large-
Diffusion Bonded Joint (x530) size, efficient equipment in heavy
industry. Examples are high-pressure
vessels used in the chemical industry and nuclear reactor pressure
vessels having wall thicknesses in the range of 200 to 300 mm. These
vessels are usually fabricated by the submerged arc welding process.
When thick materials are joined by the submerged arc welding process,
cracking transverse to the weld direction sometimes occurs in the weld
metal, and in some cases, cracks propagate through total thickness.
To avoid this kind of cracking, an intermediate stress relief annealing
(ISR) in a furnace has been utilized as the usual process of construc-
tion. However, the heavier the structure, the more the intermediate
stress relief annealing is required, and in consequence deterioration of
the toughness and the strength of the material may be induced; there is
also the great disadvantage of the cost of the construction. For these
reasons, a postweld heat treatment under a temperature of about 200 to
300°C, by which the diffusible hydrogen concentrations which are con-
sidered to be closely related to the crackings may be decreased, was
recently recommended instead of the conventional ISR.
Judging from comprehensive experimental results, it is clear that
the transverse crack is a cold cracking induced in the weld metal both
by high longitudinal residual stress and by high concentration of di-
ffusible hydrogen. Moreover, taking into account the fact that the
highest longitudinal residual stress is nearly constant regardless of
he plate thickness of the butt welded joint, it is easily seen that
prevention of the cracks can be made possible only by decreasing the con-
centration of diffusible hydrogen.
Fig. 19 shows the distribution of hydrogen concentration across the
thickness when Low Temperature Post-Weld Heat Treatment(LTPWHT) is employed
Preheat and interpass temperatures during welding were 200°C (392°F) and
the temperature of LTPWHT was changed as shown in Fig. 19; the time was
 84

about 2 h. The results indicate

that LTPWHT is a very useful method
for decreasing the concentration of
diffusible hydrogen, and its effect
increases with an increase in tem-
perature. Consequently, this treat-
ment is considered a promising
method for avoiding the kind of
crackings superseding the conven-
tional intermediate stress relief
annealing.
Unlike the conventional ISR
that is expected to lower the two
causes of the cracks — namely, re-
sidual stress and hydrogen concen-
tration — LTPWHT is a method for
preventing the cracks by decreasing
only hydrogen concentration. Fig. 19 Distributions of Hy-
The LTPWHT condition necessary drogen Concentrations across the
for the prevention of transverse Thickness when LTPWHT is Employed
cracks where a conventional fused
type of flux was used was obtained as shown in Fig. 20 by applying the
aforementioned procedure. It is seen that its condition changes exten-
sively depending upon the dimensions of the weld zone and the welding
conditions.
(It is obvious that if a lower-hydrogen type of flux is used, the
residual hydrogen concentration just after welding will decrease, so that
the LTPWHT condition required to prevent the cracks will be mitigated.)

Fig. 20 Preventing Transverse Fig. 21 Gas Burnnrs for LTPWHT

Cracks by LTPWHT where a Convention-
al Fused Type of Flux is Used

The validity of this approach was confirmed by a prototype of the

forged rings of 3000 mm inner diameter and 250 mm thickness. Girth
welding to join these shell courses was performed by the tandem submerged
arc welding process with a combination of conventional flux and wire.
 85

The welded region was sectioned for magnetic particle examination. No

cracks were observed.
Also, a number of thick-wall pressure vessels have been made by
utilizing LTPWHT instead of ISR — for example, hydrocracker reactors
of 3900 mm inner diameter, 222 mm wall thickness, and a weight of 510
tons. All excess weld reinforcement was removed by grinding after
LTPWHT; and UT inspection, magnetic particle examination, and radio-
graphic inspections specified by the ASME Code were performed to de-
termine if weld defects were present. None were found.
Figure 21 shows gas burners for LTPWHT.

CONSTRUCTION OF LNG STORAGE TANK

Liquefied natural gas (LNG) is receiving increasing attention as

the most important source of energy in the interim period before coal
and nuclear energy can take 'ieir turn in large-scale energy production.
Among the key elements to constitute an LNG supply system are the LNG
tankers to transport the fuel from the gas liquefaction stations over-
seas and the LNG tanks for storing the imported fuel in reception cen-
ters at home.
In respect of LNG tanks, there have been or are being constructed
in various reception centers in Japan, while many more are being planned,
as shown in Table 3. Large LNG storage tanks constructed on land may
be classified generally into above-ground tanks and underground tanks,
of which the former comprises by far the greater number. The most
general above-ground tank design is the flat bottom cylindrical tank of
double wall construction provided wit', an insulating layer. The ma-
terials used today for the inner tank shells directly exposed to cryo-
genic temperatures are 5083-0 aluminium alloy and 9% nickel steel, both
of which display excellent cryogenic properties, weldability and machi-
nability. The maximum thickness of their lowest strakes of above-ground
tanks are 69 mm in 5083-0 and 32 mm in 9% nickel steel. On the other
hand, underground tanks are constructed of a flat bottom cylindrical re-
inforced concrete outer w^ll forming the structural member, inside which
the leakproof inner wall of austenitic stainless steel membrane struc-
ture is supported through a layer of thermal insulating material. The
material used for the inner membrane wall is type 304 stainless steel of
2 mm thick, which possesses superior low-temperature toughness. The
suitable shape of corrugations are provided on the shell membrane in
lattice pattern and on the bottom membrane in concentric circles in
order to absorb the thermal construction.
The number of LNG storage tanks have been constructed and are
currently under construction, roughly to say, about 30% have been made
of aluminium alloy, 40% of 9% nickel steel and 30% of stainless steel
membrane. These tanks have also undergone a tremendous growth in capaci-
ty enlargement, from about 10,000 kl initially to about 130,000 kl today,
and which have involved no problems either in the course of construction
or in later service.
 86

Table 3 Examples of Mechanical Test Results

of Butt and Lapped Welded Joint
of Al-Mg Alloy 5083-0.9% Ni-steel and 304 Stainless Steel

V-notch Charpy Impact

Tensile Bend
Welding Welding Joint Thick- Test"(VE.lSJ) ( k f . )
Matonal ness, strength t e s t "
position process type mni Weld
k{/«» : R=31/3 Bond HAZ
metal
Good
Flat High current MlG Butl so 30.7,30.2
Good
— — —
Al-Mg alloy 50B3-0
Automatic MlG wilh Good
Vertical Butt 50 30.8.30.5 — — —
(Wlcr w,re:5IB3) oscillation Good
Good
Horizonlal Automatic MlG Butt 50 31.0.30.6 — — —
Good
Good
Flat Submerged-arc Butt 24 76.5.76.2 10 8 11.0 16.4
9 5&Ni-5leel(filler w.rj- Good

70Ni-Mo-W type) Automatic TIG wilh Good

Vertical Butl 24 76 0.75.8 11.2 15.2 11.2
oscillation Good
Good
Horizontal Submerged-arc BL.lt 24 74 6.74.7 11.4 9.8 11.6
Good
Automatic tow
Flat Lapped 60 5.59.6 — — — —
304 stainless pulsed TIG 2
sleel
{filler wife, Automatic low
Vertical Lapped o —
pulsed TIG 57.2.56.1 — —
type 308L)
Automatic low
Horizontal Lapped 53 9,56.3 — — —
pulsed TIG
5pcc>l>ed value of tensile strength of welded
A ! M S alloy 5083-0 -2tk(/m>(JIS)
9 <•; N,.stool 68 8kj/--'(ASM£)
G r i d tost Al-Mfl alloy 50B3 S.rio bond test
9 T* Ni-stopl Lmi',.Lid.'ial root
•Spcc.lifd value ol ab-.orbed e r v e y (VE-!96) for 9 ;N''•stool veldmonl •• 3 5»t"(ASME)

The capacity increases of these tanks as well as their construction

in large numbers in recent years have largely been made possible by
improving the weldability of materials, by the adoption of better designs,
and in particular by rationalising construction methods and by automation
of welding operations. Figure 22 shows the construction of above-ground
tank and underground tank in progress at a LNG storage base.
Comparing the three types of
1WG tanks, the above-ground tanks of
Al-alloy and 9% nickel steel have
the advantage of comparatively brief
construction works and relative
freedom from limitations of ground
conditions, on account of their
being constructed above-ground; such
structures, on the other hand, de-
mand uneconomical utilization of
land. The advantages of the membrane
type underground tank are their
effective utilization of land, their
limited impairment of the environ-
mental scene, in addition to the Fig. 22 Above-ground and under-
favorable advantage in respect of ground Tanks at a Large LNG Storage
safety against earthquakes and fire, Yard
while they are more susceptible to
limitations of ground conditions.
 87

Which among the three types of tank should be adopted for a specific
installation is a choice that will be decided in consideration of the
site conditions, ground characteristics, construction costs, and the
policy of the owner. At any rate, there is no doubt that the number of
LNG storage tanks to be constructed will increase in time, in keeping
with the rising demand for this form of fuel.
Whatever the type of tank, the development and application of auto-
matic welding process suited for each inner tank material is a most
vital factor for the establishment of the facilities. Concerted research
is currently being carried out to expand the range of automatic welding
and also to introduce more highly developed non-destructive inspection
techniques as a means of constructing storage tanks of the highest
quality for cryogenic service.

CONCLUDING REMARKS

This report is a short overview of some topics in present Japanese

industries for energy applications.
Energy shortage is a most serious problem not only in Japan but in
whole world. Further developments of structural materials and fabri-
cation processes including welding technology are required for new
energy creation, conversion, storage and transportation. For this
purpose international mutual cooperation and exchange of experiences
should be an important short cut.

ACKNOWLEDGEMENT

The authors are much obliged to Dr. I Sejima et al of Hitach Ltd.,

Dr. E. Takahashi et al of Kobe Steel Ltd., Dr. T. Shimoyama et al of
Mitsubishi Heavy Industries Ltd. and Mr. H. Nagaoka of Ishikawajima
Harima Heavy Industries Co., Ltd. for their cooperation in the prepa-
ration of this report. Particularily the authors would like to extend
their appreciation very much to the above mentioned companies for the
coutesy of valuable data and photos.

REFERENCES

1. T. Kobayashi and T. Fujimura, Research Approach on Materials for

Coal Conversion Plants, 1981 International Gas Research Conference.

2. T. Yoshida, S. Matsuda, S. Matsui, T. Atsuta, S. Toma and K. Itoga,

The Development of Corrosion-Resistant Tubing, 13th Annual Offshore
Technology Conference (Houston), May 4-7 (1981).

3. WES 1601-1981, Method of Qualification Test for Electron Beam

 88

Welding Procedure, OWES. July 1, (1981).

4. H. Kita and I. Sejima, Application of High Power Electron Beam

Melding, Advanced Welding Technology in 1981 - Japan-UK-USA Joint
Seminar, Tokyo April 17 (1981).

5. M. Endoh, M. Hirose, T. Toyohara, H. Morihana, K. Fuchigami,

T. Shimoyama and G. Takano, Study on Electron Beam Procedure for
Pressure Hull of Deep Submergence Research Vehicle made of lONi-
8Co Ultra High Strength Steel with High Toughness, Yosetsu-Gakkaishi
(Journal of the Japan Welding Society), Vol. 50 (1981) No. 8, 820-
827.

6. Fabrication Division of the Japan Pressure Vessel Research Council,

September (1981).

7. 1st International Laser Processing Conference, Nov. 1981, Anaheim

U.S.A., Laser Institute of America.

8. A. Kohno, S. Saikawa, H. Honma and K. Koide, Development of Dif-

fusion Bonded Impeller, IIW, i-GT-dif 01-79.

9. E. Takahashi and K. Iwai, Omission of Intermediate Postweld Heat

Treatment (PWHTJ by Utilization Low-Temperature PWHT for Welds in
Pressure Vessels, Symposium on Application of 2-i-Cr-lMo Steel for
Thick Wall Pressure Vessels, Denvar, May 18-23 (1980).

10. Y. Kuriyama, K. Sasaki, T. Kuriyama, N. Nagaoka and N. Katayama,

Automatic Welding of Cryogenic Storage Tanks, Metal Construction,
Vol. 9 (1977), No. 7.
•©•
 CRITICAL MATERIALS AVAILABILITY-OPPORTUNITIES
IN CONSERVATION AND SUBSTITUTION TECHNOLOGY
Allen G. Gray *
American; Society for Metals
Metals Park, 011 44073

ABSTRACT

The United States is heavily dependent on foreign sources for sup-

plies of key materials that are essential to the nation's defense and
for the operation of its vital industries. For example, this country
imports almost 100 percent of the strategic metals, chromium, cobalt,
columbium, tantalum and manganese.

Over-all the U. S. is dependent on foreign sources for 22 of the 27

metals considered vital to this country's economy. The problem affects
I!. S. industry over-all as much as potential defense needs. However,
until recently there has been all too little awareness of our enormous
dependence on foreign sources of critical nonfuel minerals.

The essential nature of critical and strategic metals requires that

all options to provide supplies or alternatives be pursued. On the
supply side, effective steps are being taken to strengthen the strategic
stockpile and develop domestic resources. The development of alterna-
tives requires that viable materials technology options derived from
substitution and conservation be available at reasonable costs.

It should be noted that the need for high performance alloy steels,
stainless steels and superalloys requiring the strategic alloying ele-
ments is increasing as demands have grown for higher durability, plus
higher performance fuel efficient aircraft turbine engines. Likewise,
it should be recognized that a transfer of this high performance mate-
rials and processing technology will be a major asset in the develop-
ment of this country's alternative energy programs.

Chromium continues to appear more critical than the other strategic

metalc. Unless we develop alternative materials technologies, a cutoff
of oil supply and chromium supply would attack this country on two
fronts.

The theme of this presentation is that technology is among the most

viable options for reducing this country's vulnerability to a cutoff of
supply of critical metals. Also an important point of emphasis, is that
now is the time to start to develop these technologies for substitution
and conservation.

Technical Director, American Society for Metals

91
 92

As a spokesman for "substitution preparedness," in Congressional

testimony, the proposal was made that the United States should embark on
an organized effort to foster and support research and development pro-
grams that will advance the practical appi ieation of substitution and
conservation technology to reduce the impact of supply interruptions
in critical metals, particularly chromium and cobaJt. Likewise, it's
evident that a plan should be developed to document known substitution
technologies and "stockpile" this information.

As R & D programs are developed to plug g^Ps aUcl create new options
for substituting for critical metals, this technological information
would be added to the information stockpile.

Substitution programs should be an important part of strategic-

materials planning for individual manufacturing firms. The objectives
of a national initiative on substitution can be applied to company sub-
stitution R & D programs.

It's not news to those of you who are participating in this Inter-
national Conference that the United States, as well as Kurope and Japan,
are heavily dependent on foreign sources for supplies of many key metals,
that are essential to the nation's defense and necessary for the opera-
tion of this country's vital industries. For example, the U. S. imports
almost 100 percent of the strategic metals, cobalt, chromium, niobium,
tantalum, manganese, and the platinum metals.

The issue of critical and strategic minerals continue:-; to be urgent.

It is gaining more attention as the complexities of this country's mate-
rials vulnerability become known.

Never before have strategic materials and minerals received as much

attention from the President, the Congress, and the many federal agencies
with materials responsibilities. The next few years will be particularly
significant to the U. S. in the development of national materials policy.
Substitutes, conservation, reclamation and innovative new materials can
all help reduce the country's materials dependency.

As previously noted, this country is almost 100 percent dependent

on foreign sources for chromium, cobalt, niobium, tantalum, manganese
and the platinum metals, and it should be noted that over-all, the U. S.
depends oxi foreign sources for 22 of the 27 metals considered vital to
this country's economy.

The problem affects U. S. industry over-all as much as potential

defense needs. However, until recently there has been all too little
awareness of our enormous dependence on foreign sources of critical non-
fuel minerals.
 93

Aside from the r act tl.at without chromium we cannot make an effi-
cient durable gas turbine engine, the metal has many important indus-
trial applications: oil refineries, petro-chemicals, conventional and
nuclear power plants, tanker trucks, industrial machinery and all
stainless steals.

Aii Options Should be Pursued

The essential nature of critical and strategic metals requires that

all options; to provide supplies or alternatives must be pursued. On the
supply side, effective steps must be taken to strengthen the strategic
stockpiLe and develop domestic resources. The development of alterna-
tivts Lequires that viable materials technology options derived from
substitution, conservation, coating systems, and reclamation be a v a i l -
able at reasonable costs.

1L should be noted that the need For high performance alloy steels,
stainless steels and superalloys requiring the strategic metals is in-
creasing as demands have grown for higher durability, plus higher
periormanco, fuel efficient aircraft: turbine engines. An example is the
prospect for increased requirements for tantalum in single crystal tur-
1) ine S 1 ados .

Likewise, the transfer of high temperature technology will be a

major asset in the development of this country's alternative energy p r o -
grams, including coal gasification and liquefaction, even deep well e x -
ploitation of domestic oil resources. It's significant that extensive
tests on alloys for these uses conducted by the Metal Properties Council
iiave shown that a minimum chromium content of about 2 5 % is required for
long term service.

This brings us to the point that materials availability either

short term during critical situations or long term due to resource deple-
tion or overconsumption, is of vital concern to industry and to this
country's security. Without proper materials many companies would be
forced to close unless Lhey made alternative plans well in advance.

In fact, supplies of critical materials and their rising cost, and

1 emphasize the latter, present what may be the most serious challenge
facing the United States in the remainder of the twentieth century.

Progress • /ard a National Materials Policy

The pursuit of our most important national goals dictates that we

be concerned about materials, and that we develop an enlightened n a t i o n -
al materials policy, and we are making progress in this area.

The critical materials issue has been recognized in legislative

circles, largely because of experience with the oil situation. I had
the privilege of appearing in the House of Representatives in mid-1979
and in the U. S. Senate in July 1980, to present testimony on critical
 94

materials issues.

Concern in Congressional circles van evidenced in the closing days

of the 96th Congress, when the "National Materials and Minerals Policy,
Research & Development Act of 1980" was passed by the Li. S. Senate and
signed into Public Law 96-479. The original bill was introduced in the
House of Representatives by Congressman Don Fuqua in 1979.

The act represents the first major statement on materials policy

developed in over tiiree decades and calls on the President to take a
number of steps in addressing materials problems. This Includes the
requirement to provide Congress a prospertive plan within a year after-
passage of the law, which would have been in October 1981. Various
government agencies contributed information for this report to Congress.

A new bill (11R 4281), known as the "Critical Materials Act of 193L,"
also introduced by Congressman Fuqua in July 1981, is intended to focus
on "organization concerns" in inplementing the original legislation.
This bill calls for the creation of a "ComiciJ on Critical Materials,"
a small coordinating council within the White House Executive Olfice, lu
oversee the multitude of federal materials programs. In introducing
the bill, Congressman i'uqua noted that in materials research alone there
are over 14 agencies and departments involved with over .*? 1 billion in
research programs.

The Council, as visualized by Congressman l-'uqua, will act as the

President's focal point to analyze and evaluate materials policy as it
would concern our energy, minerals, defense, and foreign policies, among
others. It will also provide important information and data concerning
materials needs, problems, and projections, particularly as they relate
to materials research and development.

Evidence of further action on the governmental front is provided hv

a bill introduced by Representative .lames (Jim) Santini known as the
National Minerals Security Act (HR 3364).

The bill proposes to establish a Council on Minerals and Materials

to provide an ongoing minerals voice within the Executive Branch. It is
set up along mandates similar to the lines of the U. S. Council of
Environmental Quality except it, of course, focuses on minerals. An-
other part of the bill, which is the brunt of most controversies, per-
tains to policy decisions on U. S. public lands of which 68% has been
restricted or closed to domestic mineral exploration.

It would be unwise to view these legislative actions as the total

solution to our materials problems. The developments in materials
policy legislation is only a start in dealing with many complex problems.
 95

Key Aspects of Substitution and Conservation

Use of technology for substitution and conservation can provide

industry and defense with flexibility to deal with shortages and crises
ia availability.

1 have been a strong proponent of the substitution initiative, par-

ticularly the development of an information stockpile on substitution
technology, as a national resource. This viewpoint was included in
testimony in hearings for the National Materials Policy Bill in the U. S.
Senate in July last year and in the U. S. House of Representatives
shortly after the bill was introduced.

I proposed that this country should embark on an organized effort

to foster and support research and development programs that will ad-
vance the practical application of substitution and conservation tech-
nology to reduce the impact of supply interruptions in critical metals.

Likewise, it's evident that a plan should be developed to document

known substitution capabilities and to "stockpile" this information just
as we stockpile strategic metals.

As R & D programs are developed to plug gaps and create new options
for substituting for critical metals, this technological information
would be added to the information stockpile.

As a spokesman for "substitution preparedness," which is how I re-

ferred to my recommendations in Congressional testimony, I emphasized
that it is essential that appraisal and recording of substitution pro-
cedures in specific applications be as complete as available information
allows.

Needs for additional information should be identified and defined

so that research and development projects could be assigned to gather
data required to expand capabilities in those areas which appear to be
sound and achievable with available means.

It should be noted that leadtime is very short in an emergency

situation. Thus a stockpile of information on substitution technology
is also a stockpile of time, and a valuable yet relatively low cost
supplement to the commodity stockpile.

Substitution programs should be an important part of strategic

materials management for individual manufacturing firms. The objectives
of a national initiative on substitution can be applied in internal (com-
pany) substitution programs to develop alternatives for critical metals
or options among the family of critical materials.
 Guidelines on Substitution

My recommendations for materials substitution and conservation

initiatives, include the following points:

1. Find out just how much we know about substitutes for the most crit-
ical strategic metals, particularly chromium, cobalt, manganese,
columbium, tantalum and platinum group metals.

2. Appraise the reliability of the information that is available on

substitutes for these critical metals and determine what research
is needed to round out available data so that the substitute
materials can be used with full confidence.

J. Document known substitution technologies and stockpiJe this infor-

mation in a form that can be utilized immediately by industry and
defense in an emergency. This is "Substitution Preparedness."

4. Identify basic research programs needed to create new options ior

substituting for these most critical, metals and as this information
is developed, add it to the substitution information stockpile.
Avoid waiting for a crisis.

5. Structure research to have substitution as its only goal. Research

priority would be determined on basis of probability of need, and
of sioccess.

6. Develop ways to improve awareness, interactions, and communications

in substitution technology in government, industry, and universities.

7. Emphasize research to utilize substitutes with wide distribution

and with domestic and near domestic sources such as vanadium,
molybdenum, tungsten, and nickel. Expand research to develop
manganese-aluminum alloy systems to substitute for chromium stain-
less steels on the basis of long term availability of manganese
through ocean mining.

8. Recognize that the government has an important stake in replacements

and that the nation as a whole and its security depend upon imports
of critical metals. Realize that industry is involved in solving
near term problems and has increasingly turned to application type
work and that if government leaves the task of substitution research
totally to industry it probably will not get done.

Regarding the final point, although these are good reasons for
business firms in the private sector to underwrite work on substitutes
for metals crucial to their existence, I would like to comment on the
rationale for government support.

An important consideration is to realize that the tendency is for

industry not to address a critical supply situation until a material is
 97

unavailable and a crisis has developed. Then it's likely that before
anything can be done the situation would be beyond the critical stage.

For example, chromium is very effective in hardenability applica-

tions, so as long as it is available, and is relatively economical, as
it is today, there's no incentive to substitute for it in heat treatable
steels, as well as in many other uses. But since substitution appears
possible in the area of heat treatable steels it makes good sense to
conduct research and "stockpile" information now on the production and
heat treatment of chromiumfree steels for production of important
engineering components. The data could be drawn on immediately in a
chromium emergency.

Chromium is Most Critical

On the broader picture, chromium is now recognized as the most im-

portant critical and strategic element. This is due to its versatility
in widespread industrial and defense applications, its techrical quali-
ties, the geographical concentration of the world's supply in iolitical
sensitive areas, and the lack of economic incentive to develop alterna-
tives.

In every national emergency beginning with World War I, chromium

supply has been a national concern, which wad made more serious by lack
of information on conservation and substitution technology.

Chromium uses include stainless steels, tool steels, structural and

heat treatable alloy steels, heat resistant and corrosion resistant
alloys, high performance superalloys, alloy cast iron, pigments, metal
plating and finishing, leather tanning, chemicals, and refractories for
metallurgical furnaces.

Domestic resources of chromium are low grade and under the most
critical conditions, domestic production could supply only a small por-
tion of U. S. requirements. There is no expectation that either the
private sector, or the private and public sectors working jointly, will
be able to increase proven reserves of chromium in politically stable
areas of the world.

Hence, one of the most feasible options open to industry and govern-
ment is to have available for immediate use, information on technologies
for substitution and conservation which could be employed to reduce this
country's dependence on chromium as well as that of other OCED countries.

It should be noted that new tools are available today for studying
substitution and interchangeability of alloying elements. Computer sys-
tems provide new approaches to the design of steels and alloys with
specific properties for substitution applications.

While substitution is one important approach to meeting the chal-

lenges of strategic critical metals, there is, of course, an underlying
 98

need to understand and develop other options to survive any threatened

metals cartel, including such things as processing technology for high
performance, near net shapes, utilization of domestice resources, sur-
face modified metals and reclamation.

Substitution Within the Critical Metals Group

An aspect of substitution technology that should not be overlooked

is the opportunity to utilize substitution within the family of critical
metals.

An example of the use of niobium to alleviate the critical supply

situation and increased cost of cobalt is provided by changes in turbine
disk materials, a major application for high temperature alloys.

Several nickel-base alloys containing cobalt have been tradition-

ally used in this heavy application, including Waspaloy (13% Co), Rene'
41 (11% C o ) , and Astroloy (17% Co). Engine designers are gradually
changing the disk alloy to an established cobaltfree, nickel base alloy,
Inconel 718. In terms of substitutabiiity relative to critical elements,
it should be noted that Inconel 718 contains 5% niobium in addition to
19% chromium and 3% molybdenum.

Another example is that the critical element niobium is contribu-

ting to metallurgy to reduce future dependence on foreign sources of
critical chromium through tlie development of the modified 9Cr-l Mo
steel for steam generation and other high temperature applications in
place of higher chromium austenitic stainless steels. Work is being
conducted at Oak Ridge National Laboratory and at Nippon-Kokan K..K. in
Japan on these new steels.

Opportunities in Processing

Processing innovations will play a major role in advancing both

substitution and conservation technology. For instance, high speed
steels without cobalt utilizing increased quantities of molybdenum and
tungsten are being produced by hot isostatic pressing (HIP) of powders.

It's expected that hot Isostatic pressing will make increasingly

important contributions to conservation by reducing the amount of
strategic raw materials required to produce a component.

Rapid solidification technology (RST) holds promise for the devel-

opment of materials with higher performance as well as conservation of
critical elements. Some of the potentials that relate to critical mate-
rial, if this new technology can be moved forward, are exciting:

1. Cobaltfree, high creep strength, nickel base superalloys for tur-

bine airfoils.
 99

2. An appreciable increase in incipient melting point of nickel base

superalloys.

3. Oxidation/corrosion resistant iron-aluminum alloys and perhaps

manganese-aluminum alloys as substitutes for chromium stainless
steels.

4. Laser Glazing to produce layers with different alloy compositions.

It also should be noted that RST promises to improve aluminum and

titanium alloys as well as superalloys.

It's logical that research efforts should be aimed at developing

materials technologies to utilize to the fullest extent domestic and
near domestic metal resources. (Incidentally, it appears that the
Soviets are giving much consideration to this.) These more available
resources include molybdenum, nickel, vanadium and tungsten.

Manganese is of interest in the long term as a resource in alloy

systems utilizing aluminum to substitute for chromium in stainless
steels. In addition to the fact that manganese is more widely distrib-
uted than chromium the long term potential for abundant supplies of
manganese from ocean mining of nodules suggests research support for
alloy designers in this area. The replacement of nickel by manganese
and chromium by aluminum yields a Fe-Mn-Al alloy that is useful for
cryogenic applications.

Potential for Metallurgical Coating Systems

Application of existing and developing technology for metallurgical

coating systems offers a great potential to save critical metals. For
example, hard facing as a method of effective wear and corrosion con-
troJ is receiving increasing attention to impart essential surface pro-
perties with minimum use of chromium and cobalt. A tremendous potential
exists for design of hard facing into fabrication of original equipment
in a conscientious effort to conserve critical metals and make compon-
ents last longer.

Likewise, designing with available clad steels and other clad

materials can economize on critical metals compared to use of solid high
alloy products. Clad materials properly applied can provide all the
advantages of a solid critical alloy composition.

Surface modified metals offer many possible substitutes in appli-

cations requiring solid alloys containing strategic metals like chromium
and cobalt. Several techniques are available or are in the development
stage including ion implantation, laser and electron beam high rate
fusion, spraying and surface coating with ceramics.

The concept behind surface modification is basically to put the

properties of the critical metals where they are required rather than
 100

making the entire component from a critical metal alloy. High rate sur-
face fusion provides rapid solidification rates which give valuable sur-
face properties. Ion implantation with chromium or cobalt provides the
basis metal with greater corrosion and wear resistance. Design goals
can be achieved with implanted ions.

Strategic Materials Management

I would like to emphasize the need for the business sector to take
a more active role in dealing with the realities of critical metals.
This includes the establishment of vigorous materials management policies
and programs that cover availability, conservation, substitution, recla-
mation and the minimal use of critical strategic materials.

Critical material requirements should be considered in new product

and process developments and in long term planning. Specific technical
responses should include research in cooperation with customers and
suppliers which focuses on systems based on available metals and alloys
with only small additions of critical materials.

Also, work should be pushed to enhance the mechanical properties of

less critical alloys as well as to explore the use of advanced materials
and coating systems which will displace or reduce the over-all need tor
critical metals.

It is most important that this type of information on critical

materials issues be discussed openly at all levels of the company and
especially at corporate levels where the discipline of assignments can
be established.

Finally, I would like to leave a few suggestions with you on what

individual firms can do to avoid, or at least reduce the detrimental
impact of sharp changes in materials availability and price.

First, let's accept the premise that the problem is real. That
America truly faces in the decades ahead, a growing "resource war," a
worldwide battle not necessarily involving military forces, for the
world's mineral resources.

Also, that at this point the government is not the complete answer,
although there are favorable recent developments which may open up
opportunities for cooperative action.

In seizing the initiative for dealing realistically with supply

uncertainies, I recommend the development and implementation for an
active effort of strategic materials planning including an aggressive
program on materials substitution and conservation technology.

Broadly, strategic materials planning is a formal process by which

supply network requirements can be integrated with a firm's specific
manufacturing and business objectives. The planning process addresses
 101

major supply issues that vitally affect the allocation of the firm's
resources and future performance. It seeks to answer questions such as
the following:

1. Will supply disruptions detrimentally affect manufacturing effi-

ciency? Product quality? Product cost? New product introduc-
tions? Profitability?

2. What actions are needed to secure supply networks? What are the
financial requirements? Risks? Management implications?

3. Should a Materials Resources Division or Company be established

within the organization?

4. How much is known about substitutes? What percentage of R&D should

be allocated to this on basis of probability of need and probabil-
ity of success?

Such questions deal with some of the important characteristics of

effective strategic materials planning.

Keep in mind that: first, the perspective of effective strategic

materials planning is broader than the traditional limits of the pur-
chasing function.

Second, effective strategic materials plans are the result of a

disciplined, analytical process that formally indentifies, prioritizes,
and evaluates critical materials supply issues. It must utilize a high
degree of technological input and receive periodic updating and board
room review. The planning process becomes a decision mechanism for
allocating efforts among promising options including R&D on substitution
and other supply-defensive options. Reassessment of requirements is
essential as market conditions and business priorities change.

As I indicated previously, technology can provide answers (some say

the best answers) for reducing this country's vulnerability to the
potential of cutoff of supplies of critical metals.

An important reality, however, is that technology cannot provide

quick fixes in most cases. It takes time and considerable investment to
develop, quantify and establish designer confidence in a new material or
even a modified alloy.

However, an overriding reality in my opinion is that technological

solutions will probably come much faster than the political and social
solutions necessary for the peaceful international sharing of the finite
resources of our planet.

One final note. I hope I have left you with an understanding that
the severity of the critical materials problem is such that we must
attack it on all fronts. This means: decrease demand to the extent we
 102

can through substitution and conservation, and increase the supply to

the extent we can through exploration, research and the astute practice
of international politics.
 Session C.

FOSSIL APPLICATIONS I
Co-Chairmen
R.A.Bradley S. J. Dapkuna*
Oak Ridge National Laboratory Dtpcrimartf of fiwrgy
Oak Ridge, Tennessee Werftoatan, D.C.
 WELDING PRACTICES IN
FOSSIL-FIRED UNITS

G.G. Stephenson

Technical Superintendent - Metallurgy

Thermal Generation Division
Ontario Hydro
Toronto, Canada
M5G 1X6

ABSTRACT

Welding plays a vital role in the design, construction, and

repair of fossil-fired equipment. Welds are exposed to static and
cyclic loads in severe environments, so unless welds of the highest
quality are made, failures will occur and even then premature failures
can happen due to unforeseen conditions. This paper concentrates on
maintenance repairs which are made to original welds or as a result of
material deteriorating during service. Maintenance welding often
involves compromise between theory and practice. It is not always
possible to use shop procedures, and this paper describes how modified
welding procedures have been developed for field applications.

Qualification of welding procedures and welders are discussed.

Shielded metallic arc and gas tungsten arc welding processes are
mainly used to join a wide range of materials, from carbon to low
alloy creep resisting steels to stainless steels.

Examples are given of repair techniques used for various

components. Pulverizer rolls are refurbished using a hardsurfacing
build-up. Repairs to boiler drums and headers, using a modified
temper bead technique, are also discussed. Ontario Hydro's current
practice in evaluating and replacing transition welds between low
alloy chrome-molybdenum and austenitic stainless steels using Inconel
filler materials in the shop are detailed.
 106

Strategy involved in repairing cracks detected in

chrome-molybdenum-vanadium steel turbine valve chests and turbine
diaphragm solid particle erosion damage is discussed. Rebrazing
turbine blade lacing wire is another common occurrence. Precautions
to avoid hard spots developing during brazing are described.

INTRODUCTION

Ontario Hydro has 32 operational fossil-fueled generating

stations with a total capacity of 7764 MWs. They range in size from
66 MW up to 500 MW, generally operating at a temperature of 540°C
(1000° F) and 17 MPa (2350 psi) pressure. Each unit contains
thousands of pressure vessel welds, so the performance of these welds
and repair welds is very critical. In addition to high temperature
and pressure operating conditions, the majority of our fossil units
are two shifted, ie, operating for two shifts and off for one shift.
This fatigue cycling has caused numerous problems, especially in
welded joints which are highly restrained.

Welding problems in generating stations have been reported in

large volumes in the literature.(lj(2)(3)(4)(5) I t i s n o t t h e
intention of this paper to duplicate these, but to briefly describe
some of the welding areas that have been prominent in Ontario Hydro
equipment.

Due to the severe operating conditions and high costs of a forced

outage, the weld quality has to be uniform and of a very high
standard. Repair welds are more difficult to make due to poor access,
environmental conditions, welding on deteriorated material, etc.
Procedures used by Ontario Hydro and examples of maintenance weld
techniques and practices under these conditions are described.

WELDING QUALIFICATIONS

Welding procedures and welders are qualified according to

Section IX of the ASME Boiler and Pressure Vessel Code. All pressure
welding must meet the requirements of the Boiler and Pressure Vessel
Act of the Province of Ontario as administered by the Boiler
Inspection Branch of the Ontario Ministry of Consumer and Commercial
Relations. The Act stipulates that pressure vessels and piping shall
be designed and built to the requirements of Canadian Standards
Association, CSA-B51 "Code for Construction and Inspection of Boilers
and Pressure Vessels."
 107

Ontario Hydro has a welding manual which, besides containing over

100 welding procedures, also has a lot of information on general
welding background. This is invaluable tc the stations as they can
get details from it or. welding terminology, safety, processes,
electrodes, heat treatment, and inspection.

Table I - Typical Welder Performance

Qualification Tickets

Test Pipe Sizes

No. Classification Material Process OD Thickness

1 F6-F4-A4 Up to 6 percent GTA and >1" 1/16-0.872"

alloy steel SMA

2 F4-A4 Up to 6 percent SMA >1" 1/16-0.872"

alloy steel

3 F6-F4-A4 Up to 6 percent GTA and >2 7/8" 3/16-maximum

alloy steel SMA

4 F6-F4-A5 9 percent Cr- GTA and >1" 1/16-0.872"

1 percent Mo SMA
steel

5 F6-F5-A8 Austenitic GTA and >1" 1/16-0.872"

Stainless Steel SMA

6 F5-A8 Austenitic SMA >1" 1/16-0.872"

Stainless Steel

7 F43 Inconel GTA and >1" 1/16-0.872"

SMA

F - Electrode Classification Number

A - Weld Metal Analysis
GTA - Gas Tungsten Arc Welding
SMA - Shielded Metallic Arc Welding

In repair welding, gas tungsten arc and shielded metallic arc

welding processes are used to join a wide range of materials from
carbon to low alloy creep resisting to stainless steels. Welders are
tested annually on pipework in the 6G position for the above processes
and materials to qualify them for pipe sizes required. Typical
tickets obtained are shown in Table I. If different filler materials,
welding processes or combinations, or thicknesses are to be used,
additional tests will be required. Alternatively, tests may be
omitted if the combination of variables is not to be used in the year.
 108

REPAIR PHILOSOPHY

The type of repair done on a component depends on factors such as

access, possible distortion, effect on adjacent parts, and time
constraints. If the unit comes down on a forced outage, then
generally the defect will be ground out and rewelded using a matching
filler material. In this way, the unit is returned to service in the
shortest time. If the unit is available for a longer time, then
original or superior materials may be inserted or parts built up with
material more resistant to the environment. Preventative maintenance
is also an important field where nondestructive examination can detect
future problem areas. In this way, defective areas can be found
before failure occurs and repaired on a planned outage.

It is not always possible to repair components using the original

shop welding procedure. Due to size, access, possible distortion,
etc, postweld heat treatment, although required, may not be
practical. In these cases, the welding procedure requires modifying,
such as by using the temper bead technique'^) or a more ductile weld
filler material.^ ' Repair may not always be possible and a new
component must be used; however, a spare part may not always be
available so a temporary repair will be necessary.

Ontario Hydro has a large maintenance staff at each station which

includes qualified welders. Therefore, we have the capability of
doing most weld repairs ourselves. However, some major repairs are
contracted out, generally to the original manufacturer or a
specialized welding shop. In these cases, the welding and inspection
procedures are first approved by us and the repair monitored.

Maintenance welding practices are never cut and dried like most
production shop welds. Each field repair has to be considered
separately and the technique evaluated considering other factors
besides welding parameters. It is difficult to make definite rules
and practices to follow. Therefore, it is best to give some brief
examples which will cover the above points in a wide variety of
components.

REPAIR PRACTICES FOR VARIOUS COMPONENTS

Pulverizer Rolls

Alloyed martensitic white cast iron (ASTM A532, Class 1, Type B)

rolls used in coal pulverizing mills have been rebuilt using an iron
based high carbon, high chromium hard surfacing alloy (see Figure 1 ) .
It has been found that restoring the rolls to original size and shape
has improved the life by about 50 percent at a third of the cost of
new rolls. Excellent abrasion resistance is achieved from large
chromium carbides supported in a tough alloy-steel matrix, giving a
Rockwell hardness of approximately 60 Re.
 109

Figure 1 - Pulverizer Rolls

After and Before Weld Repair

The rolls are built up using submerged arc or gas metal arc
welding with a preheat temperature of 200°C. Upon completion of
welding, the roll is wrapped in an insulating blanket and slow
cooled. The rolls are used in the as-welded condition which will
show stress-relief cracks throughout. Rolls are generally
reconditioned after 25 mm of wear per side.

Boilers

Ontario Hydro has a boiler tube failure (BTF) reporting system.

A form is filled out for every BTF indicating tube location, failure
cause, repair procedure, etc. This information is stored on a
computer and is analyzed so that any trends can be evaluated and
recommendations given to improve the overall performance of the
boiler.<8)

BTPs account for the major percentage of unreliability in our

units causing a large number of outages and generation loss (see
Table I I ) . Of these, weld failures are responsible for about
10 percent of the total (see Table III).
 110

Table II - Number of Outages Caused by

Boiler Tube Failures in Ontario Hydro System

Number of Percent Maintenance Generation

Year Forced Outages Total Outages Loss GW.d

1978 59 26 32 38.1

1979 51 20 24 33.1

1980 86 34 30 64.5

Table III - Causes of Boiler Tube

Failures in Ontario Hydro System

Percent Failures
Year DD DW DM IC EB MS

1978 30 .7 9 .4 1.4 41.1 8.0 9.4

1979 23 .1 13 .4 0 45.3 8.1 10. 1

1980 63 .8 9 .9 1.2 13.9 6. 5 4.7

Legend:

DD Design Defects IC Internal Corrosion

DW Welding Defects ED External Deterioration
DM Material Defects MS Miscellaneous

Tube Repairs

The majority of boiler tube leaks are repaired by grinding out

the defects and welding back up with a matching filler material. An
exception to this is transition joints between ASME SA 213 T22 (2-1/4
Cr-1 Mo) and SA 213 TP 321H (stabilized austenitic stainless steel).
Failures in these welds have been widely investigated.(9)(10)(ll)
It is Ontario Hydro's practice to do this weld in the shop under ideal
conditions, and the two like joints are welded in the field. The
standard procedure uses a gas tungsten arc root pass with ERNiCr-3
filler material and SMAW filler passes with an ENiCrFe-3 electrode. A
preheat temperature of 230°C is applied using gas torches on the
chrome-molybdenum side of the joint. No postheat is required as the
tubes are less than 100 mm 0D and also less than 12.5 mm thick. All
welds are radiographed.
 Ill

Boiler Steam Drums and Headers

Corrosion fatigue cracks have been detected in the lower water

wall headers and main steam drums of some boilers. These have
occurred on the inside of the drum at the toe of reinforcing fillet
welds between nozzles and the shell. (Figure 2 ) . The circumferential
cracks are removed by grinding and provided sufficient thickness is
remaining, they are blended (width to depth 3:1) and left.
(Figure 3 ) . This is determined from design calculations according to
ASME Boiler and Pressure Vessel Code, Section 1.

OUTSIDE

Figure 2 - Crack Locations at Toe

of Fillet Welds in Boiler Header
 112

OUTSIDE

Figure 3 - Details of Boiler Header

Showing Crack Grindout Profiles

A. modified temper bead technique has been developed for weld

repairs. Temper bead welding involves a controlled technique for
deposition of successive weld passes so that the heat affected zone is
tempered by the heat of welding. In this way, heavy sections in
carboti steels can be repaired without the mandatory high temperature
postweld heat treatment. We have shown that a reduction in preheat
temperature to 9 5°C (from required 180°C) and eliminating a grinding
step between the first and second weld layers will produce a sound
weld with acceptable metallurgical structure and hardness.
 113

Table IV - Modified Temper Bead Welding Procedure

for Lower Waterwall Headers and Steam Drums

Material: Carbon Steel

Weld Preparation: Grind out crack (or carbon arc-air gouge

using preheat). 60 degree included
angle, cavity top 1.3 times depth.

Inspection: Magnetic particle inspection (MPI)

following grinding.

Preheat: 95°C using electrical resistance

heating.

Welding Process: Shielded metallic arc using stringer

beads.

Weld Sequence: I. Butter whole surface using 2.4 mm

diameter E7018 electrodes.

2. MPI. Half the first layer is not

removed by grinding.

3. Deposit second layer using 3.2 mm

diameter E7018 rods. This layer
softens and toughens HAZ created by
first layer.

4. Deposit remaining passes as per

normal procedure.

Postheat: Two hours at 230°C to 280°C immediately

following welding using electrical
resistance heating.

Inspection: MPI before and after postheat and after

weld has been at ambient temperature for
48 hours.

The details of the procedure are shown in Table IV. It is

essential that the electrodes are specially heat treated to achieve
maximum dryness and then held at elevated temperature until used.
This will prevent hydrogen being dissolved in the weld causing
cracking.
 114

Turbine Steam Chests

Thermal fatigue type cracking at sharp radii on the inside of

main and reheat steam chests were in the past ground out and weld
repaired. As the chests are manufactured from low alloy creep
resisting chrome-molybdenum-vanadium steel, preheat and postheat are
required. This led to a very long, difficult access and costly repair
as the whole chest required heating.

Current practice is to leave the cracks and monitor them with

NDE, provided the chest integrity is not affected. Factors considered
in making this judgement are previous experience of crack propagation
rates, crack appearance, location, chest thickness, radiography
inspection, and possible exploratory grinding.

The procedure used for weld repairs involves preheating to 230°C

using electrical resistance heating, gouging out the cracks, magnetic
particle examination to ensure that all the cracks are removed and
shielded metallic arc welding with an E9018-B3 rod. Postheating
immediately following welding involves holding for eight hours at
700°C with a maximum heating and cooling rate of 100°C/hour or 6250/t
°C/hour, where t is the thickness of the material in mm, whichever is
smaller. A stringer bead technique is required and each pass must
overlap the previous bead by 50 percent to refine the grain size and
minimize reheat cracking. As the material is notch sensitive, the
finished weld must be hot ground to remove all notch effects prior to
postweld heat treatment.

Turbine Blade Lacing Wire Brazing

Any brazing done on 410 martensitic stainless steel turbine

blades is a critical operation requiring much care and skill.
Incorrect procedures can result in catastrophic blade failures. To
avoid this, particular attention is paid to:

(1) Joint preparation giving a clean surface for capillary flow of

the braze to ensure that a sound joint is made.

(2) Torch heating to avoid blade overheating during flame

application. This prevents local blade hardening at the critical
sharp radii of the lacing wire holes which could initiate blade
cracks.

(3) Removal of the brazing flux so that no corrosion type failures

occur.
 115

Old cracked brazes can be removed by heating or preferably by

dropping hot (55°C to 60°C) nitric acid onto it. Brazing is done with
the blade in the horizontal position, with the concave side of the
blade uppermost. An oxy-propane gas mixture is used for torch heating
with a flux conforming to AWS Brazing Flux 3A and ASME Group F101,
Classification BAg-1 brazing filler metal.

Turbine Diaphragm Solid Particle Erosion (SPE)

Ontario Hdro has experienced extensive SPE damage to the IP

eighth stage and most recently to the LP fourteenth stage diaphragms.
The extent of damage and repair options has been discussed
previously.' '

Weld repair of SPE damage, (Figure A ) , on the eighth stage

diaphragm has been done using two different welding procedures. The
martensitic stainless steel vanes have been gas tungsten arc welded
using a matching filler material with preheat and postheat or ERNiCr-3
filler using preheat only. The latter is generally used for minor
repairs or where time is a factor as no postheat is applied. Both
repairs have given comparable service performance. A weld repaired
diaphragm prior to grinding to original contour is shown in Figure 5.

Figure 4 - Solid Particle Erosion

Damage on Turbine Diaphragm
 116

Figure 5 - Weld Repair of Turbine

Diaphragn Solid Particle Erosion Damage

SUMMARY

A brief review of maintenance weld repair procedures and

techniques has been presented. It has been shown that maintenance
weld repairs can differ significantly from routine shop production
fabrication.

Examples have been presented covering many different types of

equipment and materials. Weld repairs involved using the same
original procedure or modifying it (i) to meet the latest
developments, (ii) by using a different filler material or (iii) an
alternate procedure such as the temper bead technique, have been
given.

ACKNOWLEDGEMENTS

The author acknowledges the permission of Ontario Hydro to

publish the paper. The views expressed are the personal ones of the
author and are not meant to indicate Ontario Hydro philosophy.
 117

REFERENCES

1. Burgess, N.T., Some Weld Failures In Power Stations, Welding and

Metal Fabrication, March 1962.

2. Central Electricity Generating Board (CEGB), Welding Problems in

Power Station Operation, Welding and Metal Fabrication, May 1976.

3. Stephenson, G.G., and A. Frantzem, Boiler Tube Weld Failures in

Fossil-Fueled Generating Stations, 1975-1977," Ontario Hydro
report CTS-31090-7, July 1978.

4. Lawson, W.H.S., and D. Mills, Working Within Metallurgical

Limitations in the Successful Repair Welding of Power-Plant
Components, CIM Bulletin, Volume 74, No. 831, July 1981.

5. Coleraan, M.C., The Structure of Weldments and its Relevance to

High Temperature Failure, Central Electricity Generating Board
(CEGB) Digest, July 1979.

6. Lawson, W.H.S., Development of Temper Bead Welding Methods for

Repairs of Carbon Steel Components, Ontario Hydro Research Report
80-308-K, July 19 80.

7. Gooch, T.G., Repair Welding with Austenitic Stainless Steel MMA

Electrodes, Metal Construction, November 1980.

8. Dooley, R.B., et al, Boiler Tube Failures in Fossil-Fueled

Generating Stations, 19 80, Ontario Hydro CTS Report 31090-12,
April 1981.

9. Dooley, R.B., et al, Ontario Hydro's Experience with Dissimilar

Metal Welds in Boiler Tubing, American Welding Journal, in press.

10. King, J.F., Behaviour and Properties of Welded Transition Joints

Between Austenitic Stainless Steels and Ferritic Steels—A
Literature Review, ORNL-TM-5163, November 1975.

11. IIW Documents IX-1200-81 and IXH-91-81, Welded Joints Between

Austenitic Stainless and Low Alloy Steels; Current Position,
19 81.

12. Dooley, R.B, et al, Ontario Hydro's Experience with Solid

Particle Erosion of Steam Turbines, presented at ASME/ASTM/MPC
Task Group and EPRI Workshop on Solid Particle Erosion, Atlanta,
May 6, 1980.
 PROCESSES QUALITY ASSURANCE AND EXAMINATION
OF WELD METAL OVERLAYS

John J. Meyer
Nooter Corporation
St. Louis, Missouri

The use of corrosion resistant weld metal to overlay

carbon and low alloy steels in pressure vessels is of
increasing importance to the industry with literally
millions of pounds being deposited annually worldwide.
These overlays are being done with a wide variety of
processes and overlay materials to rigid standards and high
levels of quality. The purpose of this paper is to give an
overview of the processes currently being employed, the
quality level being specified, and the examination of the
product.

Several processes are generally required to overlay

an average pressure vessel. The large areas such as heads
and shells would be done with one of several high deposition
rate processes. The nozzles and other appurtenances would
be overlaid with manual or semiautomatic processes.

Any welding process which achieves an overlay that is

fused to the base metal will result in some dilution of the
base metal into the deposit. The amount of and the control
of the dilution is very important because it will have a
direct influence on the chemistry of the deposit. It will
also impact the cost of the overlaid product.
Shielded metal arc welding (SMAW), manual gas metal
arc welding (GMAW) and gas tungsten inert welding (GTAW)
generally are done with a multilayer procedure of two,
three, or wore layers to achieve the desired chemistry.
The first or barrier layer may be of a higher alloy than
is required in the final chemistry. This will allow for

119
 120

diluting steel or low alloy base metal into the barrier

layer deposit. An example would be the use of 309
stainless steel barrier followed by 308 stainless steel to
achieve a 304 stainless steel deposit. If an extra-low-
carbon (ELC) deposit is required, a 309 stainless steel
barrier followed by two layers of 308 ELC would be used.
The manual SMAW, GMAW and GTAW processes are low deposition
rate methods; however, they may be required to clad areas
which are inaccessible to larger equipment capable of
higher deposition rates. There are also times where the
desired chemistry and corrosion resistance to certain media
can be achieved by no other processes.

Economically speaking, weld metal overlays are competing

with roll bonded clad plates or explosively bonded clad
plates. There is a thickness over which weld metal
overlays begin to be more economical then clad plates.
Three things affect the cost of the overlay: the
deposition rate; the cost of the consumables involved; and
the thickness of the deposit. When deposition rates of
2 3 kg per hour are reached, the cost of the consumables
such as wire, flux, strip and gas becomes the major cost
factor to determine the cost of the overlay.

In an effort to increase the deposition rate, many

processes have been used over the years. Automatic
submerged arc (SAW) can be used with one wire depositing
approximately 7 kg per hour. Two-wire setups, twin and
tandem, are used depositing 14 kg per hour. The dilution
rates of the steel backer into the weld overlay would, of
course, be fairly high with these methods. Two-wire setups
are used in the series arc method where each of the wires
goes through a contact tip connected to one side of an AC
constant current power supply. Deposition rates are about
14 kg per hour and could be raised to 23 kg per hour by
adding a third cold wire. Dilution rates with this method
. <n vary from 10 to 20%.

Metal powders can be added to submerged arc systems

to obtain chemistry and increase the deposition rate. Most
of these applications are for hard facing and hard metal
applications. It is a method of modifying the chemistry
of the deposit by adding small metered amounts of metal
powder to the flux or directly to the base metal to be
welded through.

Automatic GMAW with oscillation has been used but tends

to have fairly high dilution thus requiring multiple layers
to achieve chemistry. It is frequently used to overlay the
inside of a small nozzle from 4" I.D. and up to a length
of approximately 6 feet. Two or three layers may be
 121

required to achieve proper chemistry. Shingling and

overlapping of beads may be used to reduce dilution.
As mentioned earlier, the cost of the overlay is
dependent on the cost of the consumables once a deposition
rate of 20 to 30 kg per hour is achieved. In an effort to
get deposition rates that high, a number of important
processes have been developed. These would include
submerged arc strip, electroslag strip, "Multi-Weave"
multiwire using self-shielded stainless steel and
submerged arc, plasma mig, and plasma hot-wire.

Welding with strip is certainly not new. It has been

done on pressure vessels for twenty years or more. Recent
improvements in fluxes used with the process along with
equipment improvements have enhanced the quality of the
deposits greatly.
The strip will vary in thickness from 0.4mm to 0.75mm.
The width of the strip used can be 25mm wide to 300mm wide.
A typical width used in production would be 60mm to 75mm.

Most of the strip overlay work has been done by the

submerged arc process. By modifying the quantity of
fluorides in the flux, it is possible to electroslag weld
rather than submerged arc weld.

When using wider strips, there does seem to be an

advantage to electroslag welding. We know that with
submerged arc welding the arc will travel along the width
of the strip and that this travel is not always uniform
and thus the penetration is not uniform. In the slag mode,
the penetration does seem to be more uniform because
melting of the strip takes place across the full width.
With wide strip, magnetic forces can be a problem because
they distort the molten slag, molten metal, and can cause
ragged undercuts at the tie-ins. This problem can be
controlled by generating a Lorentz force by placement of
electrical coils adjacent to the strip.

The materials which are deposited by strip would

typically be 309L, 309L barrier layer followed by 308L,
347 or 316L, if the lowest carbon content is required.
308L, 347 and 316L in a single layer may be used but the
chemistry will be marginal. Strip of nickel-chrome-iron
has been used to produce Inconel type overlays.

The "Multi-Weave", multiwire system has used as many

as six wires being fed into a common weld puddle. The
early multiwire systems were done by submerged arc.
Later, self-shielding, open-arc fabrication wire was
 122

developed. This material is about 85% efficient with lo'Z

of its weight being the self shielding flux. It produced
its own gas shield around the arc and a slag cover over the
weld deposit.
The wires available for this method are 308L, 3161,,
34? and 309L. It is obviously a high deposition rate
process. In recent years, the six-wire system of open arc
welding has given way to two-and three-wire heads which are
not quite as difficult to operate.

The plasma arc method of overlay welding can be

separated into three different methods of accomplishing
a weld overlay. The plasma overlaying discussed here does
not include methods which used metal powder and are designed
to overlay small areas such as valve seats.

There is a single wire plasma mig which feeds a wire

through the body of the plasma torch. This equipment is
designed primarily for butt welding; however, it is possible
to do high quality weld overlays with the equipment.
The gas metal plasma arc process uses two wires fed
through a contact tip from the opposite sides of the tcrch
to meet under the plasma torch. One power source is
connected to and feeds current through an arc running
between the work piece and the plasma torch. The filler
wires are fed into the plasma just below the torch nozzle.
The second power source is connected to the torch and to a
sliding contact on each of the two wires. This power
source drives current through the wires and through the
plasma torch. The wires melt continuously and the molten
metal is transferred to the weld pool in a stream of
droplets. The torch and the contact tips are covered by
an inert gas shield and the assembly is mounted on an
oscillator. This assembly oscillates perpendicularly to
the weld direction to produce a wide weld bead on the
surface of the base metal.

The plasma hot-wire method of weld overlay uses, as

its primary source of heat, a DC power supply and a
transferred plasma arc torch. A second source of heat
comes from the heating of the two filler wires by I 2 R
resistance. This is done by an AC power supply. The wire
is continuously fed through two separate contact tips and
meet within the molten pool where the electrical circuit
is completed. Because these wires stick out some 75mm
from the contact tip to the molten pool, they become
heated to incandescence by I 2 R resistance. This heat,
along with the plasma torch, results in melting wire and
base metal in a controlled manner to produce the desired
 123

overlay. The plasma torch and the hot-wire torches are

both mounted on a trailing shield assembly which provides
an inert gas protection. The entire assembly of trailing
shield, plasma torch, and hot-wire torch is mounted to and
carried by an oscillator. The plasma overlay system is
capable of deposition rates in the 25 to 30 Kg per hour
range and is of high quality with minimal base metal
penetration. Most austenitic stainless steel plate
chemistries can be achieved in a single layer without a
barrier layer using the plasma process.

In addition to the austenitic stainless steels, Table

I also includes other materials which have been deposited
bv the plasma hot-wire method.

Table I Filler Material Deposited

by
Plasma Hot-Wire Overlay Process

"ER308 ER316 ER309Cb

ER308L ER316L
ER309 ER317L
ER310 ' ER347
Other alloys include:

ERNi-1 (Nickel 61) ERNiCr-3 (Inconel 12)

ERNiCrMo-4 (Hastelloy C-276)
76) ERNiCu-7 (Monel 60)
ERNiMo-7 (Hastelloy B-2 ) ERNiCrMo-3 (Inconel 625)
ERNiFeCr-1 (Incoloy 65) ERCu (Copper)
ERCuNi (Monel 67) 25Cr 22Ni 2Mo 5Mn
ERCuAlA-2 (Al. Bronze) 19-15H
Haynes Alloy 25 904L
Nitronic 50 (22Cr 13Ni 5Mn)

The plasma method of overlay has great versatility

in that most corrosion resistant alloys which can be drawn
into a wire can be successfully overlaid on the interior
surfaces of pressure vessels. The plasma process is not
limited by the availability of a suitable flux for the
high alloys as would be required by strip or submerged arc
welding.

As mentioned earlier, when the deposition rate reaches

25 to 30 Kg per hour, the major cost per square foot is the
cost of the filler metal. In most cases, wire will have
an economic advantage over the cost of strip and flux.
 124

Regardless of the process chosen for weld metal

overlay, the quality of the final product must be assured.
Careful weld test plate studies must be done as a first
step toward quality products. ASME provides minimum rules
to determine the quality as outlined in QW214 of Section
IX. QW214 says that the procedure test plate shall be
liquid penetrant examined to look for surface defects; the
plate shall be cut up and four guided side bends shall be
made; and a chemical analysis shall be taken of the deposit.

More tests can be done and perhaps should be done to

assure quality. In the case where austenitic stainless
steel is the overlay material, the percentage of ferrite
present is very important. If the ferrite is too low,
microfissures could become a problem and, if too high,
sigma phase could develop in elevated temperature service.
A range of 4 to 10% ferrite is desirable except in some
critical corrosion service where very low ferrite is
required. The ferrite can be checked with a Ferritescope,
by chemical analysis and a constitution diagram, or by
a magnetic gauge. The Ferritescope can be used to check
each bead as it is deposited and thus can spot procedure
or chemistry problems immediately.
"Crack blocks" of the overlay have proven valuable in
some applications by finding microfissures at various
locations below the surface. "Crack blocks" are done by
making a series of cuts starting at the surface and
progressing through the overlay. At each level, the surface
is polished, etched, and examined with a 30-power microscope
to look at the amounts of porosity, slag inclusions, and
fissures. Chemical analysis at various locations should
be checked to assure homogeneity of the deposit.

Inspection and quality control of the overlay during

fabrication are also important. As mentioned in the case
of austenitic stainless steel, frequent checks for ferrite
should be done. Detailed chemical analysis of the overlay
should be carried out and spot-checking of the as-supplied
wire should be done before overlay starts as well as close
examination of the mill test reports.

In-service problems with weld metal overlays generally

do not occur if proper research and development work and
quality control have been done.

Corrosion can be avoided if proper material evaluations

and testing have been done.
 125

It has been reported that pressure vessels which have

been overlaid with austenitic stainless steel and are in
high pressure hydrogen service such as oil refinery and
coal conversion have suffered from disbonding of the
overlay. This could, of course, lead to serious in-service
problems with the vessel.

This problem is being studied by various groups such

as the Welding Institute, where specimens of most of the
overlaid processes discussed in this paper will be
autoclaved at high temperature and pressure in a hydrogen
atmosphere. They will then be carefully examined by
metallography for disbonding.

Disbonding does not appear to be widespread and may be

related to the manner in which the equipment is operated
and shutdown.
Weld metal overlay will continue to be an important
fabrication method of providing an economical protection
against corrosion, erosion, and hydrogen embrittlement of
pressure vessels for the chemical and petroleum industries.
Careful selection of the materials, the processes, and the
methods of examination will ensure long life in service.
 126

REFERENCES

1. "Maglay The New Surfacing Process with Electro-slag

Welding Using Wide Strip Electrodes" Kawasaki Steel
Corporation.

2. "Mechanical and Physical Properties of Stainless

Steel Deposits Produced by Stoody Fabricated Electrodes
and Wires" Stoody Company.

3. "Gas-Metal-Plasma-Arc Welding a New Method for Welding

Cladding" by Dr. Eric Smars AGA AB and Mr. Bachstrom
Uddcomb Sweden AB. The Welding Institute Reprint
Exploiting Welding in Production Technology London
22-24 April 1975.

4. "New Advances in Plasma - Mig Welding" Philips Welding,

The Netherlands.

5. "Plasma Hot-Wire Overlay Welding" Linde Division

Union Carbide.

6. "Hydrogen Induced Disbonding of Weld Overlay in Pressure

Vessels and its Prevention" Presented at the I18th
Chemical Plant Welding Research Committee of the Japan
Welding Engineering Society in Tokyo, August 7, 1981,
by T. Sakai, K. Asami, M. Katsumata, H. Takada and
0. Tanaka Kobe Steel Ltd.
 HARDFACING MATERIALS AND PROCESSES FOR VALVE APPLICATIONS

by
S. J. Matthews
and
P. Crook

Cabot Corporation
1020 West Park Avenue
Kokomo, IN 46901

INTRODUCTION

Hardfacing is a term describing the application of a hard material

to the surface of a component by weld overlay surfacing for the main
purpose of reducing wear. Wear can be defined as the loss of material
by mechanical impact, abrasion, erosion (solid particle/liquid drop-
let/cavitation), or metal-to-metal sliding. It should be recognized
that wear is a surface phenomenon. As the surface of a component
wears away, it usually becomes less efficient or even disabled,
usually because of loss of dimensional clearance or configuration.
The hardfacing of valve components, such as gate valve seats, has long
been recognized as a method of improving the operating efficiency,
performance and life of fluid flow control devices. The purpose of
this paper is to review some of the hardfacing materials and procecses
associated with valve applications.

HARDFACING MATERIALS

Hardfacing alloys for valve service are expected to possess a wide

variety of properties including high compressive strength, low coef-
ficient of friction, resistance to metal-to-metal wear, resistance to
cavitation erosion, impact resistance, corrosion resistance, and good
hot hardness, when elevated temperature service is involved.

Of particular concern to the valve industry is the phenomenon of

galling. Galling is the term used to describe the interfacial damage
which commonly occurs when two metallic surfaces are brought together
under high load and subjected to relative motion. Localized damage on
valve seat faces may lead to leakage and excessive galling may even
result in valve seizure.

The galling phenomenon has been the subject of much work in recent
years. Two schools of thought have developed. One theory proposed is
the atomic bonding of surface asperities with shear failure occurring
in the weaker of the two mating faces. Bond strength is related to
real contact area, which in turn, is influenced by the deformation

127
 128

characteristics of the two surfaces. The other theory is that surface

degradation is caused by subsurface crack nucleation and growth, in
response to surface stress waves.

To protect the seat faces of many industrial valves, the technique

of hardfacing is utilized and the primary choice of material is
usually a high performance cobalt-chromium wear alloy. Laboratory
test data suggest that few other alloys resist galling to the same
extent.

The nominal chemical compositions of the well known cobalt-

chromium hardfacing alloys used in valve applications are given in
Table I.

Table I
Nominal Chemical Composition of
Traditional Cobalt-Base Hardfacing Alloys

Co Cr W Mo C
STELLITE* alloy No. 1 Bal 31 13 - 2.45
STELLITE alloy No. 6 Bal 28 4.5 - 1.0
STELLITE alloy No. 12 Bal 30 8.5 - 1.5
STELLITE alloy No. 21 Bal 27 - 5.5 0.25

Like other wear-resistant materials, the cobalt-chromium alloys

are comprised of a hard phase dispersed throughout a metallic matrix.
The hard phase, which is formed during solidification, and thus morph-
ologically dependent upon hardfacing parameters, is a chromium-rich
M7C3 carbide. The matrix is essentially the corrosion and oxida-
tion resistant binary cobalt - 20 wt% chromium, strengthened by solute
atoms of tungsten or molybdenum.

In addition to outstanding self-mated anti-galling properties, the

alloys are moderately resistant to abrasion, have good resistance to
chemically aggressive media, possess outstanding resistance to both
cavitation and liquid droplet erosion, and retain their strength even
at high temperatures.

The difference between the individual cobalt-chromium alloys is

chiefly one of carbide volume fraction and morphology, hence resist-
ance to abrasion. The toughness of these alloys (and also resist-
ance to cooling stresses during welding) generally decreases with in-
creasing hard phase volume fraction.

The outstanding anti-galling properties and cavitation/liquid

droplet erosion behavior of the cobalt-chromium wear alloys are attri-
buted predominantly to their matrix deformation and fracture

*STELLITE is a registered trademark of Cabot Corporation

 129

characteristics, which are in turn related to stacking fault

energy2'-* and the tendency to transform (from a face centered cubic
to hexagonal close packed structure) or twin under stress.^»*
Contrary to popular belief, both resistance to galling and cavitation
erosion have been found to be largely independent of carbide volume
fraction, and hence bulk hardness, for this class of alloy. Those
alloys, therefore, with a low carbon level and the benefit of appre-
ciable toughness (i.e., STELLITE alloy Nos. 6 and 21) find much use in
the hardfacing of valve seats.

In the wake of the 1978 cobalt price increase, a number of new

hardfacing materials, based largely on the systems nickel-chromium,
nickel-iron-chromium, and cobalt-iron-chromium were developed. Only
those based on cobalt-iron-chromium, however, have demonstrated good
self-mated anti-galling properties.

Examples of two relatively new hardfacing alloys are STELLITE

alloy No. 2006 and Developmental alloy No. 2008. The latter has
proved to have exceptional anti-galling properties in room temperature
air, not only under self-mated conditions, but also against the 300
Type Stainless Steels and the corrosion-resistant nickel alloy
HASTELLOY alloy C-276, features which the traditional cobalt-chromium
alloys do not possess. The nominal compositions of alloy Nos. 2006
and 2008 are given in Table II.

Table II
Nominal Chemical Composition of
Two Relatively New Hardfacing Alloys

Co Ni Fe Cr W Mo_ C Cb
STELLITE alloy No. 2006 33 8 18.5 3 1 - 8 1.2
Developmental alloy No. 2008 32 8 25.5 26.5 2.5 - 0.4 5

The choice of approximately 30 wt% cobalt and of an iron balance

for these two alloys was influenced by experience with diluted
cobalt-chromium deposits and previous work concerning alloys of
reduced cobalt content.6 While STELLITE alloy No. 2006 contains
relatively high levels of chromium and molybdenum to enhance solid
solution strengthening, Developmental alloy No. 2008 relies for its
matrix strength upon the release of chromium from its usual carbide
forming role. Columbium, an active MC carbide former, is thus an
essential ingredient of the alloy.

Resistance of these alloys to cavitation erosion, which is caused

by the formation and collapse of bubbles on surfaces in high pressure
fluid systems (the liquid jets resulting from collapse being responsi-
ble for surface stress waves), was determined according to ASTM
 130

G32-77 test procedures. Each test involved attaching a sample, in the

form of a stub, to an ultrasonic horn and vibration of the horn in
deionized water (maintained at 27°C) at a frequency of 20 KHz over a
100-hour period. Weight loss of each sample was measured at 25 hour
intervals, and mean depth of erosion calculated through density consid-
erations. The experimental data showed that alloy Nos. 2006 and 2008
have cavitation erosion resistance similar to that of the traditional
cobalt-chromium hardfacing alloys.

To determine the resistance to galling of these alloys, a pin-

on-block test similar to that described in reference 7 was utilized.
Essentially, the test involved the twisting under load of a cylin-
drical pin against a stationary block by means of a wrench. Threshold
stress is defined as the lowest at which galling was observed on
either pin or block after one revolution of the pin. The maximum pos-
sible stress using such a test, and pins of diameter 9.5 mm,
was found to be 501 MPa. Stress increments of 63 MPa were used during
test. The results, summarized in Table III, describe the galling re-
sistance of the new alloys in comparison to traditional cobalt-
chromium alloys used in valve service.

Table III
Galling Test Data Given in Terms of
Threshold Galling Stress (MPa)

Counterface Block Materials

Alloy (Pin) 1020 Steel 316 Steel C-276 STELLITE 6 Self-
Wrought Wrought Wrought Oxy Deposit Mated
STELLITE alloy No.
6 (2 layer oxy
deposit) <_ 251 125 63 > 501

STELLITE alloy No.

21 (2 layer
GTA deposit) <_ 63 £63 125 > 501 > 501

STELLITE alloy No.

2006 (2 layer
GTA deposit) > 501 < 63 251 > 501 > 501

Developmental alloy No.

2008 (2 layer
GTA deposit) 251 438 > 501 > 501 > 501

HARDFACING PROCESSES

Hardfacing alloys, especially STELLITE alloy No. 6, are available

in a wide variety of different product forms for depositing weld over-
lays (bare cast rod, coated electrodes, tubular wires, solid wires,
and powder). Most welding processes are readily adaptable to hard-
facing providing proper techniques are employed to prevent "stress"
 131

cracking and to minimize base metal dilution. Stress cracking

generally occurs as a result of thermally induced stresses, the
occurrence of which can be minimized by liberal use of preheat, high
interpass temperature, and very slow cooling. The hardfacing of
transformation hardenable steels (such as type 410 Stainless) can com-
pound the stresses operating on the hardfacing deposit during cooling
and require special precautions to minimize cracking. Specifically,
preheat and interpass temperatures should be maintained above the M s
temperature of the steel. Depending on the size and mass of the part,
a post weld heat treatment immediately after hardfacing may also be
required to minimize cracking.

Hardfacing techniques should always attempt to minimize base metal

dilution for the obvious reason that excessive base metal dilution
will compromise the metallurgical effectiveness of the hardfacing mat-
erial. This technique is the opposite of welding and joining, where
often increased base metal penetration is desirable. For economic
reasons, the deposition rate for hardfacing processes should ideally
be as high as possible.

Three processes which have been used extensively for hardfacing

valve components are oxyacetylene, gas tungsten arc and plasma trans-
ferred arc. The oxyacetylene process is characterized by the lowest
achievable base metal dilution factor (< 5%) but unfortunately the
process is relatively slow and time consuming (about 1 Kg/hr). Fur-
thermore, the very low carbon Co-Cr alloys such as alloy No. 21
intrinsically do not have good oxyacetylene weldabililty. Alloy No. 6
can be deposited by oxyacetylene. However, proper melting practice
and chemistry control are required during manufacture of the cast rod
in order to produce consumables which do not "boil" during
oxyacetylene deposition.

Oxyacetylene methods have given way in many cases to the gas tung-
sten arc processes, especially when hardfacing austenitic stainless
steels that potentially sensitize if exposed to the carburizing "feat-
her" of an oxyacetylene flame. Since gas tungsten arc is a more in-
tense heat source, more base metal dilution (~ 202) can be
expected. However, overall dilution can usually be minimized by using
two or more layers of hardfacing deposit. Hot cracking can be a
potential problem in gas tungsten arc hardfacing. Hot cracking may be
caused by high levels of deleterious elements (i.e., sulfur).
Attempts to hardface a "free machining steel", such as type 303 or 303
Se, may result in hot cracking since presence of harmful elements will
be introduced to the deposit via base metal dilution.

The gas tungsten arc process can be readily automated if the hard-
facing filler metal is available either as long lengths of bare cast
rod (usually 3.2mm diameter or larger) or as a continuous spool of
wire (usually 1.6mm diameter). Long lengths of bare cast rod are pro-
duced either by continuous casting or by butt welding shorter lengths
together. Continuous spools of STELLITE alloy No. 6 solid diameter
 132

wire are produced by a proprietary process. The selection of con-

sumable size and form of course depends on the size of the valve com-
ponent part to be hardfaced. Small diameter valve seats (i.e.,
12.5mm) would warrant the use of small diameter solid wire, while lar-
ger valve seats (i.e., about 220mm in diameter) would best utilize a
larger bare cast rod. Tubular wires are also available for gas
tungsten arc hardfacing. However, if selecting a tubular wire, the
utilization of a fully compacted product is necessary. If a loosely
compacted tube wire is fed into the arc, "mini" explosions are
possible and expulsion of filler metal tends to foul the tungsten
electrode producing an erratic arc.

The plasma transferred arc process is characterized by the ideal

combination of relatively low base metal dilution (~ 10%) plus a
relatively high deposition rate (up to 5 kg/hr). A further advantage
of the plasma transferred arc process is that powder is used as the
consumable. Hence, almost any hardfacing alloy composition can be
melted, atomized, and screened into this product form overcoming any
obstacles set forth by the traditional methods of casting rod,
wroughting wire, or compacting tube wires.

Electrode

Plasma Gas

Shielding
Gas
Powder and
Powder Gas

Deposit

V/77/77/Z77/77//71
Figure 1. Schematic illustration of the
plasma transferred arc hardfacing process
 133

The plasma transferred arc process is schematically illustrated in

Figure 1 and can be described briefly as follows: a tungsten elec-
trode, recessed into a torch body, generates a transferred arc to the
workpiece. Plasma gas (usually argon) is ionized within the torch and
exists through a constricted orifice. At this location, hardfacing
"filler" material is introduced in powder form through powder injec-
tion ports assisted by a carrier gas. The carrier gas is usually
argon, although sometimes a secondary gas is used such as helium or an
argon-hydrogen mixture. The powder particles (previously screened to
approximately a 100/325 mesh size) completely melt and resolidify as a
fusion welded overlay. Only a small portion of the powder (referred
to as "overspray") may escape this action. Hence, the process is
about 95% efficient. Microstructural features of PTA deposits are
very similar to those achieved by conventional gas tungsten arc using
a bare cast rod.
Figure 2 compares the microstructure of STELLITE alloy No. 6 depo-
sited by oxyacetylene, gas tungsten arc, and plasma transferred arc
processes.

-xv. >„

a) oxyacetylene b) gas tungsten arc c) plasma transferred

arc
Figure 2. Microstructure of STELLITE
alloy No. 6 deposited by three different
processes | |
2 0 pm

The plasma transferred arc process was introduced about 20 years

ago, but has received increased attention over the last 5 years as an
efficient method of depositing wear resistance weld overlays on valve
components. It has been estimated that some 200 PTA machines are cur-
rently operating in the U.S. and over half of these are being used for
valve applications. Users have selected PTA because of its ability to
deposit a sound fusion welded overlay with remarkable smoothness,
thereby minimizing final machining operations. Figure 3 shows a valve
component before and after hardfacing, illustrating this feature.
 134

Figure 3. Valve component before and after

hardfacing with plasma transferred arc
(courtesy of Masoneilan International, Inc.)

Perhaps the primary reason for renewed interest in PTA is that the
process is naturally conducive for mechanization and automation.
There has been a distinct trend in this country over the p3st ten
years toward increased automation and mechanization of hardfacing pro-
cesses, especially in the valve industry. Mechanization serves not
only to increase productivity but also to eliminate the problems of
manual fatigue thereby improving product quality, consistency, and
reliability.

Efforts to mechanize the PTA process have taken full advantage of

recent instrumentation technology utilizing programable sequence con-
trollers, microprocessors, and other computerized methods."

Innovative advances have also been made in terms of novel torch

designs that can reach difficult places such as deeply seated areas of
integral valve bodies and inside diameters of small bores.

SUMMARY

The subject of hardfacing, perhaps thought by some as a "black

art", is on the contrary a very high technology effort especially in
 135

the valve industry. The technology is manifested by the need for so-
phisticated high performance hardfacing alloys required to resist the
demanding environments of fluid flow control valve service. High
technology is also found in the automated methods currently being used
to efficiently deposit high quality hardfacing overlays.

REFERENCES

1. Silence, W. L., "Effect of Structure on Wear Resistance of Co-,Fe-,

and Ni-Base Alloys", Journal of Lubrication Technology, 1978, 100
(3), 428-435.

2. Bhansali, K. J. and Miller, A. E., "Role of Stacking Fault Energy

on the Galling and Wear Behavior of a Cobalt-Base Alloy",
Proceedings of the Third International Conference on Wear of
Materials, ASME, 1981, 179-185.

3. Woodford, D. A., "Cavitation-Erosion-Induced Phase Transformations

in Alloys", Metallurgical Transactions, 1972, 3, 1137-1145.

4. Remy, L. and Pineau, A., "Twinning and Strain-Induced FCC to HCP

Transformation on the Mechanical Properties of Co-Ni-Cr-Mo
Alloys", Materials Science and Engineering, 1976, 26, 123-132.

5. Antony, K. C. and Silence, W. L., "The Effect of Composition and

Microstructure on Cavitation Erosion Resistance", Proceedings of
the Fifth International Conference on Erosion by Solid and Liquid
Impact, 1979, 671-677.

6. Cobalt Monograph, Centre d'information du Cobalt, 1960, 362-384.

7. Schumacher, W. J., "Wear and Galling can Knock Out Equipment",

Chemical Engineering, 1977, 84 (10), 155-160.

8. Zuchowski, R. S. and Culbertson, R. P., "Plasma Arc Weld

Surfacing", Welding Journal, June 1962, p. 548.

9. Kapus, B. G., "Computer Control Takes the Variations Out of

Hardfacing", Welding Design and Fabrication, June 1980.
 Session D.

GAS SHIELDED PROCESSES

Co-Chairmen
H. L. Helmbrecht T. M. Mustafeski
Babcock and Wlcox union Catbk* QxparaHoft Ntic*** DMion
Barbwton, ONo O<k FSdo*. T«
 WELD PENETRATION VARIATION IN GTA WELDING
OF SOME 304 L STAINLESS STEELS

J.L. Fihey and R. Simoneau

Recherche et Essais, Materiaux

IREQ, Institut de recherche
d'Hydro-Quebec, Varennes
Quebec, JOL 2P0

ABSTRACT

A weldability problem has been experienced with the

automatic GTA welding of stainless steel 304 L used for
instrumentation tubings in a nuclear power plant. Eventhough
the welding parameters were the same, the weld profile was
observed to change markedly for some heats. Low sulphur and
oxygen heats have a larger width to depth ratio (W/D) than
usual. The problem is particularly severe when a high and a
low W/D heat are butt welded since the weld deviates toward
the high W/D heat resulting in a lack of root fusion. Experi-
mental measurements of the arc force show that it has a
negligible effect at such low current (20 to 40 A ) . Cinemato-
graphic observations with and without laser beam illumination
show that the penetration pattern is linked to the anodic spot
geometry. A diffused anodic spot produces a poorer penetra-
tion than a constricted anodic spot. A model based on our
experimental results and results given in the literature is
proposed. The anodic spot mode depends on the presence of
minor elements. Elements having a high electron affinity
(0,S, halogens) within a given concentration range tend to
oroduce a constricted spot whereas a very low level of those
elements combined with the presence of easily ionized elements
(Cs, K, Al, Ca, Cr, Mn...) tend to produce a stable diffused
anodic spot visualized by a blue ionized metallic vapor at the
surface of the weld pool. The arc voltage is lower in the
diffused anodic spot mode which explains its tendency to
spread over the high W/D heat in a butt weld. The model
explains all the experimental results given in the literature
to date, including the contradictory observations about the
effect of oxygen. Addition of 0.1% oxygen to the argon
shielding gas is shown to produce a constricted anodic spot on
all heats, thus minimizing the weld deviation. An excess of
oxygen (l%) restores a poor penetration pattern (diffused
anodic spot) independent of heat composition.
 140

INTRODUCTION

The automatic GTA welding of 304 L tubes is commonplace in CANDU

nuclear power plants. However in the past few years, a number of lack
of fusion defects have been observed with this process^* > , as illus-
trated by the typical assymetrical geometry shown in fig. 1. Similar
weldability problems have been reported in the literature for 3 0 4 ^ ^ ,
316'^^ and 304 L'^' stainless steels. The weld deviation phenomena is
linked to the long known problem of variation of penetration from heat
to heat. Surprisingly when two heats of different weldability .are join-
ad the weld deviates toward the poor weldability heat. No satisfactory
explanation had yet been given to this phenomena.

20X
Fig. I. Typical cross-section of a defective weld.

In this study we analyze the effect of sulphur and oxygen content

in the base metal and oxygen content of the shielding gas upon 304 L
stainless steel weldability. The mechanism of weld deviation is brought
to light through metallographic examination of the welds cross-section
and visual and cinephotographic observation of the weld pool. It is
suggested that the anodic spot size and its position are responsible for
the change in penetration profile.

EXPERIMENTAL DETAILS

Chemical analyses of seven different heats of 304 L stainless

steels used in this study are given in table 1. The materials comprised
 141

two 6.4 mm diameter tubes (Tl and T2) with a wall thickness of 1.3 mm,
four 9.5 mm diameter tubes (.3 to T6) with the same wall thickness and
one plate 5 mm thick (Pi). The tubes were mainly used for weld devia-
tion testing and the plate for depth of penetration testing. The
sulphur content of the tubes varies from 20 to 220 ppm weight and the
oxygen content from 57 to 136 ppm weight. The percentage of the other
elements are quite comparable from heat to heat except for phosphorus
(60 to 300 ppm weight).

Table 1. Chemical analyses of the seven heats used in this study

Ma t e r i a 1
dent i ficat ion Shape C Si Ni Cr Mn P S N 0

Tl Tube .021 .65 10.3 18.5 1.31 .022 .002 .053 .0073

T2 Tube .020 .58 10.3 18.4 1.25 .008 .010 .053 .0136

Tl Tube .025 .60 9.4 18.4 1.70 .026 .022 .064 .0076

n Tube .020 .53 10.4 18.5 1.27 .006 .010 .034 .0108

V) Tube .021 .63 9.8 18.7 1.29 .025 .003 .064 .0057

Th Tube .011 .63 10.2 18.4 1 39 .019 .013 .066 .0063

l'l Plate .022 .67 10.1 18.3 1.77 .030 .003 .036 .0058

Pairs of tubes were tacked and the joint rigorously cleaned (acid
stripping + ultrasonic cleaning in acetone). Welding was carried out in
a chamber. The tubes rotated at a constant speed and the current was
preprogrammed using a microcomputer for power source control. Argon
backing gas at a constant pressure of 76 mm of water was flowing inside
the tube. The electrode diameter was 2.4 mm with a 30° vertex angle.
Four shielding gas compositions were uset\ pure argon (99.999), argon -
0.1% O2, argon - 1% 0 2 and argon - 1% H 2 . The flowrate was 12 1/min.
The welding current was continuously decreased to compensate for the
heating of the tubes. In order to optimize the procedure, we measured
the temperature close to the weld pool with an infrared pyrometer. This
temperature had to be kept constant for uniform penetration. Figure 2
shows the recording of current and temperature for a butt weld between
two 9.5 mm diameter tubes. The welding speed was 2.4 mm/s. The temper-
ature lies between 1220°C and 1360°C during 15 s or 2.5 s more than
needed for a complete rotation thus allowing for an overlap of the weld
bead at the end of the cycle. A variation of 140°C in the vicinity of
the weld pool is quite low considering the extremely large thermal
gradients (2000 to 5000°C/cm in this area. In our experiments the arc was
abruptly extinguished before one complete revolution in order to keep
track of the original joint and to have a complete image of the weld pool
surface. The joints were then cut in half for metallographic examination
of the weld cross-section.
 142

T(°C)

Argon flowrate 12J/mm

Trovel speed 0,24cm/s

10
Time (s)
Fig. 2. Welding current and temperature close to the fusion
line during automatic GTA tube to tube welding.

RESULTS

A - Effect of material composition

The three combinations possible with tubes Tl and T2 were welded.

Pure argon shielding gas was used. The tubes T2 have a better weldability
or lower width to depth ratio (W/D) of the fused zone and they have a
higher sulphur and oxygen content. Any attempt to weld those heats
together results in weld deviation to the side of the poorly weldable tube
T2 (Fig. 3 ) . Visual and cinematographic observation, with and without
laser beam illumination, of the arc and weld pool show two different
behaviours. On tube Tl the pool is stable (constant solidification rate)
producing a smooth weld surface without any solidification ripples. A
blue vapor covers all the surface of the weld pool. According to spec-
rroscopic analyses^") this vapor is mainly constituted of ionized manga-
nese, chromium and iron. The weld pool on tube T2 is unsteady i.e. the
liquid-solid interface progresses irregularly producing solidification
ripples. The blue vapor covers only a portion of the pool generally in
the center and the size of this blue region varies very rapidly. When the
ionized vapor extends in one direction so does the pool suggesting that
energy is transferred through this vapor. We used two watts argon laser
beam illumination (wavelength = 488 mm) in order to see through the arc.
The camera was looking across the pool through a narrow band filter. The
film was run at 250 frames/s. The blue vapor is no longer visible but in
the case of the unsteady pool there is some activity at the surface which
delineates the position of the blue zone. The very high speed of the
observed contrast variation suggests that this is probably due to metal
vaporization. However, in the case of the stable weld pool even the bulk
of the liquid metal is steady. One can detect only two rather slow con-
 143

a 10X 10x
Fig. 3. Weld penetration profile a) T2-T2 butt joint, b) T1-T2
butt joint. Same welding parameters, cross-section 180°
from weld initiation.

HIGH SUIPHUR HEAT LOW SULPHUR HEAT

Fig. 4. Fusion pass from tube T2 to tube Tl. Note the difference
in noise level and weld bead appearance.
 144

vective movements on both side of t\ic arc. In a particular experiment: we

made a fusion path starting cm tube T2 and ending on tube Tl and we moni-
tored simultaneously the arc voltage and weld pool radiations using the
infrared pyrometer.The welding current was 25A. The results are summed up
in Fig. 4. The unsteady weld pool gives very noisy voltage and radiation
recordings. Many ripples are seen on the bead surface. The voltage and
radiation noise level decreases dramatically when the pool stabilizes on
tube Tl. In this experiment the mean value of arc voltage was higher on
tube T2 (high penetration, low W / D ) . This is a general trend with a volt-
age difference of 0.4 t-j 0.6 volts. When tubes Tl and T2 are butt welded
the tail of the pool deviates towards Tl as shown in figure 5. The blue
zone is located on the side of Tl extending from the root of the arc to
the rear of the pool.

A second series of experiments was carried out with the 9.5 mm dia-
meter tubes. The initial current was 46 A and it was decreased at a rate
of 1.3 A/s (optimized welding program). Three different combinations were
tested, the results are summarized in table 2. With combination T3-T4 the
blue region is constricted and slightly displaced towards tube T4. With
the two other combinations the pool moves crabwise as already described

Table 2. Effect of material composition on weld geometry parameters

1

Ms Ui Ws/Wi a b Ws/b Off centre Direct ion l.ac

Combinat ion (mm) (mm) (mm) (mm) (mm) of of
deviat ion f US 1

T3 - T4 4.40 3.50 1.27 5.51 4.32 1.03 0.35 T4 No

T4 - T5 4.29 2.90 1.48 5.31 3.73 1.15 O.55 T"> No

T4 - T6 4.83 4.25 1.14 6.58 3.76 1.28 0.75 T6 No

,. Torch
• Blue ionized
vapor
Asymmetricol ripples

v
Arc
Low sulfur heat Weld pool \A
-Wil
Joint High sulfur heal
Fig. 5. Schematic view of the migration of the weld pool towards
the low sulphur, low oxygen heat.
 145

(fig. 5 ) . Generally the pool deviates towards the low sulphur heat, but
in the case of combination T4-T6 the sulphur level being almost the same
the pool deviates towards the lower oxygen heat.

B - Effect of shielding gas composition

Four different shielding gas were tested with combination T4-T5. The
same welding parameters as above were used. The results are given in
table 3. The addition of 1% H2 to pure argon widens and lengthens the
weld pool. The crabwise movement of the pool is accentuated (Ws/b in-
creases) and the deviation becomes extremely large. Oxygen on the other
hand reduces the deviation as shown on Fig. 6. Addition of only 0.1% O2
changes drastically the weld geometry. The axis of the weld pool is
parallel to the travelling direction. The blue region is constricted and
slightly deviated towards tube T5. With 1% oxygen the deviation is very
small but the energy input is inadequate resulting in a lack of root
fus iop.

Table 3 Effect of shielding gas composition on weld geometry parameters

Shi eldinR Ws Wi Ws/Wi a b Ws/b Off centre Di rect ion Lack

Cumbi on gaa ( nun) (mm) (mi 1 ) (mm) (mm) of of
COIlip osi t ion dev i at ion fusion

IV. - r> Pure argon .29 2.90 1.48 5 31 3.73 1.15 0.55 T5 No
(99 .999)

T4 - T5 Argon -IX H2 5 .60 4.90 1. 14 8 36 4.46 1.26 1.60 T5 No

T4 - Tb Argon - 0 . 1 1 0^ 4 .40 3.70 1.19 7 04 4.19 1.05 0.50 T5 No

T4 - T5 Argon -1% (Vj 4 .20 1.3 3.23 * — 0.35 T4 Yes

* Those values could not be measured because it was not possible

to localize the weld pool surface contour.

10x b)
Fig. 6. Combination T4-T5 welded with the highest initial current,
a) pure argon - b) Ar - 0.1% O2.
 146

C - Effect of heat input

It was recognized early during the experimental work that increasing

the heat input to the weldment would reduce the likeLihood of lack of
fusion defects related to weld pool deviation. In order to test this
observation the combination T4-T5 was welded with two different heat
input. The heat input was varied by changing only the initial current
while keeping the same decreasing rate (1.3 A / s ) . Pure argon shielding
gas was used. The results for each combination are presented in table 4.
An increase in the heat input decreases the deviation of the weld pool
probably because of a better homogenization. Moreover a high heat input
is more forgiving as far as lack of fusion defects are concerned because
the inner bead width is larger. One drawback of using high heat input is
the excessive inside bead convexity. It can be controlled by a finer
adjustment of the backing gas pressure. For our experiments the pressure
of 76 mm of water was not always sufficient to keep a flat inner bead.

Table 4 Effect of heat input on weld geometry parameters

Ws Wi Ws/Wi Off contre D i r e c t ion Lack

Combinat ion Init i a l (mm) (mm) (mm) of of
current deviat ion fusion

T4 - T5 42 3.95 2.46 1.61 1.02 T5 Yes

T4 - T5 46 4.29 2.90 1.48 0.55 T5 No

D - Arc force measurement

The plasma arc exerts a pressure on the weld pool which could be
responsible for the observed weld deviation. In order to check this
assumption the arc force for currents of up to 200 A was measured. The
experimental details are reported elsewhere^''. The arc force was mea-
sured continuously from 10 A to 200 A. The force varied as the square of
the current intensity reaching 1.1 10~ N at 200 A (Fig. 7 ) . These re-
sults are in agreement with measurements reported by Eagar and Burleigh
' ° ) . Within the experimental error the force was independant of the com-
position of the workpiece (stainless steel, carbon steel or copper).

-.-2 N
In the range of 20 to 50 A the arc force was less than 0.05 10~'
which is of the order of the force exerted by the argon gas flow. This
force is very small and cannot play a major role in the weld pool devia-
tion phenomena.
 147

i—i—i—r T " n—i—i—i—i—i i—r

L Arc force on 304 stainless sleel

Arc length 2 mm
10-2 Electrode Tungsten-2% Thorium
2.4mm diameter
20° vertex angle
Shielding gas: pure Argon, 12£/min

100
Current (A)
Fig. 7. Arc force on 304 stainless steel versus welding current.

DISCUSSION

A - Literature review

Within the past 25 years numerous authors have observed heat to heat
penetration profile variations. These were generally attributed to the
presence of minor elements. Two recent review articles by Metcalfe and
Quigley'"-' and Glickstein and Y e n i s c a v i c h ' ^ ' summarized the observations.
Depending on the explanations given to the observed phenomena two main
trends can be identified:

1 - Penetration controlled by the anodic spot configuration

The electrons emitted at the tungsten cathode enter the weld pool
through some fraction of its surface called the anodic spot. This spot
can either be diffused covering most of the weld pool surface or restrict-
ed to a smaller region. The anodic spot can be identified as the heat
source. Ludwig'3) stated that the presence of low ionization energy
impurities in the weld metal ( C s s K, C a ) widens the anodic spot and
accordingly reduces the penetration. Ten years before he had observed
that some impurities of chlorine in zircalloy would increase the anode
fall and the weld penetration^1*'. He suggested that this was due to the
presence of negative chlorine ions in the anodic space. Those ideas were
revived with some variations by Savage et a l . ^ 2 ) and ^^
 148

2 - Penetration controlled by convection in the liquid pool

The weld penetration profile may be controlled by convective heat

transfer in the weld pool. A first group of authors^^ > '•^ > '•"' proposed
that Lorentz forces created by the diverging current lines in the weld
pool play a dominant role in liquid metal motion and thus convective heat
transfer. Andrews and Craine'^'' have mathematically demonstrated that
with a concentrated anodic spot the Lorentz forces induce an axial motion
from the surface to the bottom of the pool, whereas a diffused anodic spot
induces a radial motion from the center to the edge. For a second group
of authors'- ^ > *9) surface tension is the driving force for liquid metal
motion. Heiple and Roper^'°^ identified experimentally sulphur, selenium
and tellurium as elements increasing the depth of penetration and aluminum
and cerium as elements decreasing the depth of penetration. These authors
suggested that the first group of elements were surface active elements
which promoted a direct dependance of surface tension with respect to tem-
perature. In this case the liquid metal flows inward toward the center of
the pool and then down thus increasing penetration. Whereas in the pre-
sence of aluminum or cerium and when the elements of the first group are
at a very low level there is an inverse surface tension-temperature rela-
tionship and the liquid metal flows radially outward from the center of
the pool.

B - Fundamental role of the anodic spot

The experimental facts we have collected to date do not allow us to

settle the matter definitively as to which one of the proposed mechanisms
is dominant. However, there are some signs indicating that the anodic
spot size variation is the controlling factor. The blue region observed
in our experiments was already mentioned by Metcalfe and Quigley^') and
Mills'^'. It can be used as a tracer of the anodic spot since it indi-
cates a high temperature region. The weld pool deviation is related to
the position of the blue vapor or the anodic spot over the weld pool.
Moreover the weld deviation is a first order effect which affects the
whole of the weld, thus it should be related to a major heat generation
mechanism that is the electron potential energy (workfunction) or the
anode fall'20)_ Experimental measurements on small GTA weld pools showed
very small temperature gradients in the liquid in the range 0-8°C/mm'21y
and our experiments with relatively large (15 mm diameter) stainless steel
weld pools show a maximum superheat of 80 to 90°C^22,)_ The excess heat
content of a stainless steel weld pool with such a superheat is about 16.2
cal/g whereas the heat of fusion is a factor of 4 higher (67 cal/g) .
Accordingly heat transfer by convection should have a second order effect
on the penetration profile of small weld pools.
All the elements which have been shown to enhance penetration belong
to groups VI B (S,Se,Te) and VII B (halogens) of the periodic table of the
elements. They have a high electron affinity and thus are susceptible to
form negative ions. On the other hand the elements which decrease the
penetration have a low ionization energy (Cs, Ca, K, Al) and they form
positive ions. Oxygen is the first element of group VI B, thus it should
behave as sulphur and form negative ions. Yet contradictory results are
 149

reported in the literature. There is a patent which claims that oxydation

of the workpiece surface prior to welding improves the penetration'23)
whereas oxygen addition to the argon shielding gas has been shown to
decrease the depth of penet rat i o n ^ ^ ) _ in order to test the last observa-
tion, we made one GTA fusion pabs on a low sulphur - low oxygen 304 L
stainless steel plate (PI, see chemical analyses in table 1 ) . The welding
current was 100 A, the travelling speed 1 mrn/s and the arc length 2mm.
Pure argon shielding gas was used at the beginning of the pass, the arc
voltage was 8.8 volts, a diffused anodic spot (blue region) covered almost
completely the weld pool and the penetration was poor (W/D = 6.2). The
shielding gas was then switched to the mixture argon - 0.1% O2 without
extinction 0? the arc. Dramatic changes in the arc and weld pool behavior
occurred. The anodic spot became constricted with a small jet of blue
vapor emerging from a depression visible in the center of the pool. The
penetration increased markedly (w/D = 1.8) and the arc voltage increased
to 10.3 volts. The shielding gas was then again switched to the mixture
argon - 1% O2 resulting in the complete disappearance of the blue vapor, a
widening of the pool and the poor penetration pattern resumed (W/D = 7.1).
However the arc voltage was still higher, 10.7 volts.
Figure 8 shows the cross-section of the weld bead corresponding to
the three shielding gas together with a schematic view of the anodic space
charge distribution. All our observations to date, including spectroscopic
measurements, suggest that in the absence of elements having a high elec-
tron affinity a diffused anodic spot is formed and the elements having the
Lowest ionization energy (Cr and Mn in this case) are positively ionized.
With some oxygen in the shielding gas stable negative ions are formed
creating a potential barrier to the incoming electrons, the anodic spot is
constricted in one or a few small regions from which negative ions are
L spelled.
This constriction of the anodic spot results in an increase of the
anodic current density and anode fall as measured by Ludwig'^1' which must
account for the arc voltage increase. Above some oxygen level the forma-
tion of positive ions region is no longer possible and the anodic spot
becomes diffused again, the electrons have to cross the negative ions layer
which explains the high value for the arc voltage. Those observations are
consistent with the results of Savage and a l . ^ 2 ) showing good penetration
in a specimen of commercial grade inconel alloy, with 240 ppm O2 in argon
and poor penetration with 600 ppm O2 . The critical oxygen level is
certainly dependent on the residual elements content of the alloy.
The Rosenthal model^4) f or a po int heat source and heat transfer by
conduction alone predicts that the ratio W/D takes the value of two. Thus
the value 1.8 obtained with the mixture Ar - 0.1% O2 is a normal value for
a constricted anodic spot. With a diffused heat source W/D is expected to
increase as observed and the melting efficiency may decrease because the
surface losses are likely to be higher.

C - Proposed mechanism of weld pool deviation

According to our experimental results the weld deviates constantly

towards the low sulphur or low oxygen heat. Phosphorus does not seem to
play a role since the weld deviates alternatively towards the low or high
 150

Diffused anode spot

e e ee

10 x
Pure Argon(99.999)

Constricted anode spot

t> e

10 x
Argon- 0.1% O2

Diffused anode spot

e e e e e

10x
Argon - 1% 0 2
Fig. 8. Effect of oxygen content of the argon shielding gas upon
weld penetration and anodic space charge distribution.
 151

phosphorus heat. Tubes T2 and T5 have a very low sulphur and oxygen
content and a diffused anodic spot with respect to standard GTA welding
procedure using pure argon. The other tubes are of the constricted anodic
spot type. The anodic spot may be more or less constricted depending on
the particular chemical composition and its size may vary rapidly on the
same heat leading to weld pool instability (Fig. 4 ) .
Butt welding of two tubes of different anodic spot behavior results
in catastrophic weld deviation. The anodic spot has a tendency to spread
over the low sulphur - low oxygen tube thus reducing the arc voltage.
This behavior produces the observed crabwise motion of the fused zone.
Using an argon-oxygen mixture restores a concentrated anodic spot and
minimizes the deviation. Too much oxygen decreases the penetration and
may cause lack of fusion defects. With two tubes of the constricted anodic
spot type there may be still some slight deviation towards the low sulphur
or low oxygen tube but the weld pool keeps a straight motion and the
penetration is uniform. In this case, lack of fusion defects are unlikely
unless the welding parameters are not well adjusted.

CONCLUSION

It has been shown that, all other parameters being constant, two
factors influence GTA weld penetration profile on 304 L stainless steel
materials:
- Sulphur and oxygen content of the stainless steel.
- Oxygen content of the argon shielding gas.
The lower the sulphur and oxygen content of the stainless steel and
the oxygen content of the shielding gas, the higher the ratio W/D (poor
weldability).
A mechanism based on the anodic spot geometry has been proposed. It
is suggested that a lack of elements having a high electron affinity
either in the metal or in the shielding gas promotes vaporization and ion-
ization of metallic elements (blue region). A stable diffused anodic spot
is formed resulting in high W/D ratio. The presence of easily ionized
elements (Ca, Cs, Al, Cr, Mn...) amplifies this trend. On the other hand
if elements having a high electron affinity are available (0,S, halogens)
they form negative ions which constrict the anodic spot in a small posi-
tive ions rich region. The resulting concentrated heat source produces a
normal penetration profile (W/D = 2 ) . Increasing the concentration of
those elements increases the arc voltage and decreases the anodic spot
size up to a point where the formation of a positive ions rich region is
no longer possible and a diffused anodic spot resumes.
This model explains all the results given in the literature to date
and particularly the contradictory observations upon the effect of oxygen.
Weld pool instability may occur at the transition between two anodic spot
mode. Catastrophic weld deviation happens when two tubes having a differ-
ent anodic spot behavior are GTA butt welded. The weld deviates toward
the diffused anodic spot type material (lower arc voltage). Adding some
oxygen ( 0.1%) to the argon shielding gas restores a constricted anodic
spot on both materials and reduces dramatically the weld deviation.
 is:

Stainless steel manufacturers are constantly improving the

cleanliness of their product which arc more and more of the diffused
anodic spot type. Thus the incidence of weld deviation on weld defects
should diminish but at the expense of poor penetration. Tt mav be i good
practice to add some oxygen to the argon shielding pas in order to keep :\
constricted anodic spot and good penetration.

ACKNOWLEDGEMENTS

The authors are grateful to M. Drouet for assistance in the arc forc<
measurement and R. Roberge for reading the manuscript and valuable com-
ments. They wish to thank M. Fontaine for the cinematographic work and
P. Kieffer for performing the spectroscopic analyses and operating: the
laser. They finally thank A. Di Vincenzo, .1. Larouche and A. Jolv for
assistance in the experimental work.

REFERENCES

1. I. Grant and R.G. Ford, "Variation in Penetration Profile in

Automatic Gas Tungsten Arc Welding of Instrument Tube". AEGL
Report, M-3783, USI-2960610 (1981).

2. J.L. Fihey and R. Simoneau, "Expertise de soudures GTAW

defectueuses dans des tubulures en 304 L", IREQ Report (1982).

3. H.C. Ludwig, "Current Density and Anode Spot Size in the Gas
Tungsten Arc", Welding Journal, 47_, 234s (1968).

4. A.F. Taylor "Control of Welding Processes", Developments in

Pressure Vessel. Technology - 3 , Ed. R.W. Nichols, Applied Science
Publishers Ltd, London, 267 (1980).

5. T. Moisio and J. Leinonen, "The Influence of Minor Elements on th°

Welding Arc and Penetration in Austenitic Stainless Steel Welding",
Int. Conf. P r o c , The Welding Institute, 285 (1979>.

6. P. Kieffer, "Spectroscopie de 1'arc de soudure", IREQ Report 2513

(1982).

7. J.L. Fihey and M. Drouet to be submitted for publication.

8. T.W. Eagar and T.D. Burleigh, "The Effect of Arc Force on Weld Bead
Shape", presented at the 1981 AWS Convention, Cleveland, Ohio.

9. J.C. Metcalfe and M.B.C. Quigley, "Arc and Pool Instability in GTA
Welding", Welding Journal, 56, 133s (1977).
 153

10. S.S. Glirkstein and W. Yeniscavich, "A Review of Minor Elements

Effects on the Welding Arc and Weld Penetration", W.R.C. Bulletin,
226 (1977).

11. H.C. Ludwig, "Arc Welding of Vacuum and Inert Atmosphere Melted
Zircalloy-2", Welding Journal, J36_, 335s (1957).

12. W.F. Savage, C D . Lundin and G. Goodwin, "An effect of Shielding

Gas on Penetration in Inconel Weldments", Welding Journal, 4_7_, 313s
(1968).

13. B.M. Patchett, "MIG - Welding of Aluminum with an Argon-Chlorine

Mixture", Metal Construction, _H3, 484 (1978).

14. W.H.S. Lawson and H.W. Kerr, "Fluid Motion in GTA Weld Pools, Part
1", Welding Research International, ^, 1 (1976).

15. W.it.S. Lawson and H.W. Kerr, "Fluid Motion in GTA Weld Pools, Part
2", Welding Research International, jj, 63 (1976).

16. G.S. Mi U s , "Arc/Weld Pool Interactions and their Effects on GTA

Weld Penetration", Int. Conf. Proc., The Welding Institute, London,
May 1979.

17. J,G. Andrews and R. Craine, "Fluid Flow in a Hemisphere Induced by

a Distributed Flow of Current", J. Fluid. Mech., 94_, 281 (1978).

18. K. Ishizaki, "A new Approach to the Mechanism of Penetration",

Int. Conf. P r o c , The Welding Institute, London, April 1980.

19. C.R. Heiple and J.R. Roper, "Effects of Minor Elements on GTAW
Fusion Zone Shape", Rockwell Int., Preprint RFP 3224 (1981).

20. M.B.C. Quigley and al. "Heat Flow to the Workpiece from a TIG
Welding Arc", J. Phys. D. Appl. Phys. _6, 2250 (1973).

21. E.J. Morgan-Waren, "Weld Metal Solidification and its Control by

Stirring", Int. Conf. Proc., The Welding Institute, London, May
1979.

22. J.L. Fihey, "Regulation en temps reel du soudage automatique a

l'arc electrique - Rapport interimaire n° 1", IREQ Report 2452 C
(1981).

23. J.C. Majetic and R.B.G. Yeo" Method of Welding Stainless Steel",
U.S. Patent 3, 584, 187, June 8 (1971).

24. D. Rosenthal, Trans. Amer. Soc. Mech. Eng., 68, 849 (1946).
 Evaluation of welding techniques and weldments for high temperature
components used in the gasification of coal

D.K, Schuhmacher, W. Schulze Frielinghaus and J. Ptitz

Rheini Braunkohlenwerke AG, Cologne, FRG

R. van Gaever and F. Eichhorn

Welding Institute, Techn. University, Aachen, FRG

INTRODUCTION

The situation on the energy markets for the present and the future
will require optimal utilization as well as substitution of scarce energy
supplies such as crude oil and natural gas. Liquid and gaseous products
derived from coal and lignite are required for the chemical industry, for
fuel production and for part of our heating requirements. Therefore
research and development on advanced coal gasification technologies are
in progress. Effort is concentrated on high pressure - high temperature
systems and on the application of the fluidized coal bed technique. In
the early 70's Rheinbraun started investigations on the development of
the high temperature Winkler process (HTW) for the production of
synthesis gas from brown coal and on hydrogasification of coal for
generating a methane rich gas which can be used as substitute natural gas
(SNG). The feasibility of both processes is being tested in pilot plants.
A demonstration plant with an output of 37 000 m-yh synthesis gas is in
the construction phase now and is scheduled to go on stream in 1984-
Five years of experience in the successful operation of bench scale
installations and pilot plants provide a good basis to evaluate the
requirements for advanced high temperature alloys and their fabrication
for the advanced 1922 Winkler Process and especially for the new
Hydrogasification process.
This report presents some results of investigations of welding
techniques for high temperature components used in the hydrogasification
of lignite. The results obtained from mechanical tests can be transfered
to the requirements in the Winkler gasification process as well as to
other high temperature processes in the petrochemical industry.
Hydrogasification is a process for converting coal into methane with
hydrogen in a fluidized bed at pressures of about 80 - 100 bar and at
temperatures of 850 - 950 °C. The hydrogen required can be provided by
two different methods, as shown in Fig. 1.
- The residual char obtained by hydrogasification is gasified with steam
and oxygen in a separate gasification reactor, e.g. the
high-temperature-Winkler gasifier,
- Part of the methane produced is converted into hydrogen by combining
hydrogasification with a high temperature nuclear reactor (HTR). In
this process methane is converted into hydrogen in a nuclear-heated
steam reformer. This combined process involves a cut in coal
consumption by up to 40 percent and in addition has other economic
advantages.
 156

This gasification system sets up severe requirements on the behavior of

high temperature materials and their weldments in corrosive environments.
There are two different aspects from which these requirements have to be
considered: On the one hand, metals in the gasification reactor operate
as conventional high temperature plant components in conditions similar
to the complex carburizing/oxidizing atmospheres in chemical industry. On
the other hand, nuclear gasification means a challenge to the absolute
integrity of high temperature alloys as a barrier to the primary circuit
in the nuclear heated steam reformer in the special steam reformer
atmosphere.

TEST PROGRAM

In an extensive materials evaluation program, funded by the German

government and carried out in cooperation with the German nuclear
industry, the coal companies, research centers and university institutes,
various alloys under consideration for gasifier internals and for steam
reforming tubes are being tested under simulated service conditions. Test
materials are solid solution hardening austenitic steels and nickel base
alloys which are available as wrought and cast reformer tubes and as
plates (Table I).
As the usability of such high temperature alloys will depend in
large part on their ability to be joined satisfactorily, their general
weldability was evaluated prior to extensive ageing and mechanical
property evaluations. Emphasis was given to the optimization of
mechanized welding processes. Welding was done with mechanized gas
tungsten arc (GTA), gas metal arc (GMA) and plasma arc (PA) welding
processes (Table II). The welding of joints with GTA and PA was always
carried out with filler metal. Parent like fillers as well as dissimilar
nickel base filler wires were used as shown in Table II. Weldments were
evaluated by
- tensile and impact tests prior to and after ageing at different
temperatures
- long term creep rupture tests in a simulated steam reformer atmosphere
- corrosion tests in process gas with oxidizing and carburizing gas
components
- exposure tests of some alloys as components in the high temperature
zone of the hydrogasification pilot plant.

WELD PREPARATION

In selecting a suitable weld preparation no differentiation was made

between tubes and plates. The thickness of the specimens varied between
10 to 20 mm. Preliminary investigations showed, that a simple vee groove
preparation, as is recommended by most manufacturers for base metal
thickness of the same dimensions, did not lead to reproducible root
passes with complete penetration. A complete root penetration is to be
 157

expected if a root opening of about 1 to 2 mm is maintained during

welding. However this is impossible even by using a large number of tack
welds. The shrinkage of the alloys during welding is so great, that a
reduction of the root opening and even a root closure is produced. Best
weldments were made with weld preparation as shown in Pig. 2. This
involves more machining, but has the advantage, that the base metals can
be welded without root opening and an even, reproducable fusion at the
root is guaranteed. A groove angle of 40° is sufficient to assure a
secure side wall fusion. Larger angles lead to a greater weld volume,
which in turn causes a larger angular missalignment. Before welding, the
turned or milled weld preparations must be carefully cleaned.

MECHANIZED GTA WELDING

The mechanized GTAW process was mainly used to weld the root passes.
When welding plates in flat position the torch was driven over the base
metal. Tubes were horizontally fixed to a turntable and welded under a
stationary torch. In both cases purified argon was needed for shielding
and backing. For plates uncooled copper backing with argon shielding was
used. For the GTAW tests a high power source was used in connection with
an automatic torch elevation adjustment for the raaintainance of a
constant arc length, along with control of the time slope of the welding
current, the transport of filler metal and the rotation of the turntable.
This device provided the basis for an automatic welding process for
circumferential root passes with reproducible welding parameters and
guaranteed defect free weldments. (Fig. ]5)
Plates to be welded were in contact with start on and run off plates
and welded with the automatic arc length control. This way of root
welding was successful with all base metals exept the nickel base
alloy 617 containing up to 12 % of cobalt. The viscosity of the 617 melt
was so high, that even with extreme variations of welding parameters no
sufficient root penetration could be achieved. In this case the pulsed
gas metal arc welding process was applied.
Filler passes were welded with GTA. When filling up the last passes
of the grove and the cover, it is necessary to oscillate the torch and
the filler wire. To avoid contaminations and glassy scales on the weld
bead the surface of each pass was carefully cleaned by wire brushing.
Weld surfaces were ground every 4-5 passes. Fig. 4 shows a GTA welded
joint c:i an Incoloy 800 H reformer tube specimen with one root pass, ten
filler passes and one cover pass.

MECHANIZED GAS METAL ARC WELDING

In order to increase the yield when welding thick specimens GMA

welding with spray transfer was applied for filler passes. Though GMA
welding is not recommended for nickel base alloys by the manufacturers,
good weldments were produced, when special manipulations were applied to
 158

avoid cold lapping and lack of fusion. In multipass welding beads with
high crowns or pronounced convexity increase the tendency for cold
lapping to occur. In addition to this hot cracking appeared on the base
metal Incoloy 800 H. Cracks were found in high numbers as grain boundary
cracks in the base metal close to the fusion zone. Unfavourable cooling
rates in the large melt together with a restraint of shrinkage caused
these defects. Using the same heat input during GTA welding hot cracking
did not occur.

MECHANIZED PULSED GMA WELDING

The mechanized pulsed GMA welding process proved to be i.vorable for

welding Inconel 617. Root penetration with the GTA process was not
possible. As the pulsed arc technology combines low heat input with a
static-free and constant welding process free of short circuits and
spatter, it was used for both filler passes and cover passes. Fig. 5
shows a defect-free welded joint of Incoael 617 and the welding
parameters. An effective current of 220 Amperes, an effective voltage
of 25 Volts and a travel speed of 35 cm/min result in a heat input of
9,5 kJ/cm, which is as low as welding by GTA. But unlike GTA one pass was
welded on top of the other with an oscillating torch. Each pass shows
good fusion to the side walls and a suitable surface for the subsequent
pass.

MECHANIZED PLASMA ARC WELDING

Mechanized PA welding tests with matching filler metals were made in

Inconel 617 plates with a thickness of 10 mm by the single pass key hole
technique. The plates were butt welded with a root face thickness of 8 mm
and a groove angle of 60° and with backing. Fig. 6 shows a joint, welded
with 325 Amperes, 27 Volts and a travel speed of 25 cm/min. There was
only a little addition of filler metal of about 10 g/min. For all tests a
three port orifice with an inner diameter of 4 mm was used. PA weldments
were not only defect free as related to pores and hot cracks, but they
also showed the most uniform and even surfaces of welds and reverse sides
of welds. However PA welding imposes severe requirements on weld
preparation and machining. In contrary to the optimal conditions in a
welding laboratory with firmly fixed plates to be welded and a perfect
control of welding parameters the fabrication of thick components by PA
welding in practise must take into account the missalignment of plate
edges and of root opening.

MECHANICAL PROPERTIES OF WELDING JOINTS

Mechanical tests of welded specimens and the parent metals were

 159

carried out both at room temperature and at high temperature in the range
of 800 °C to 950 °C.
In performing tensile tests on welded high temperature alloys, the
following aspects and characteristics which will influence the test
results have to be considered:
- The microstructure of the weld has a cast like structure and is
normally more coarse-grained than the parent metal.
- The structure is directionally solidified depending on the process and
the welding parameters.
- The joint design as V-groove or U-groove produces a non symmetric weld
area within the gauge length.
This results in mechanical properties, especially strain and toughnes.°,
differing from the base metal. Therefore it is necessary to use specimens
with long gauge lengths in order to allow for the total strain of both
the weld and the parent metal.

Typical strain and fracture characteristics of tensile specimens are

shown in Pig.7
Tensile tests were performed with strain control on standardized
high temperature tensile specimens with gauge di-meters of 10 mm and
gauge lengths of 50 mm on cast alloys and 100 mm on wrought alloys.
Fig. 8 compares the strength of the different parent alloys tested at
elevated temperature. The alloy with the best strength is Inconel 617
followed by Nimonic 86 and Hastelloy X. The centricast reformer tube
materials Manaurite 36X and IN 519 show comparable strengths. The results
for weldments tested at room temperature and at elevated temperature are
shown in Fig. 9«
The ultimata tensile strength of wrought alloys such as Incoloy 800 H,
Incoloy 802 und Inconel 617 as well as the cast alloy IN 519 is lower
than that of the parent alloys, even when using different filler metals.
Comparing the strengths of welded specimens to the parent metal, there is
a 10 - 20 % reduction in strength relative to the unwelded alloy. This
behavior can not be attributed to the weakness or the strength of the
different filler metals, as - with some exceptions - failure of the
specimens has always been in the base metal or the heat affected zone. It
can only be concluded that the different strain behavior of weld and
parent metal affects the ultimate tensile strength more significantly
than the strength of the filler metal itself. Fig. 10 shows an analysis
of the average strain behavior of some weldments compared to the base
metal. The differences in ductility between weld joint and parent metal
which are small at room temperature, increase with higher temperatures.
Welds show on the average a reduction of elongation of about 50 %.

MECHANICAL PROPERTIES AFTER AGEING

Tensile tests on welds and base alloys in the mill condition are
supplemented by investigations on the changes in tensile strength and
ductility following ageing. Fig. 11 plots for the alloy Manaurite 36X the
 160

trends found for changes in the ultimate tensile strength and the
elongation at room temperature after ageing up to 10 000 h. Parent metal
and weld loose their strength with increasing time. Elongation drops to
very low values due to embrittlement by precipitations. Strength and
ductility diminish with increasing ageing temperature. Stress and
ductility behavior tested at elevated temperature show a similar
tendency.
High toughness of structural alloys for high temperature service and
for high pressure is regarded as necessary for resistance to crack
propagation and sudden failure. Charpy V-notch impact teats were
performed to evaluate the toughness of the welded joints prior to and
after ageing. There are outstanding differences in changes of toughness
depending on the type of the weldment and the parent metal. Welds from
wrought alloys such as Alloy 800 H show remarkably lower impact values
than the parent alloys while welded specimens from cast alloys increase
their toughness. (Fig. 12). The conclusion is simple considering that the
fracture has a tendency to run from the notch into the most embrittled
area. That is in the case of a wrought alloy, the weld and in the case of
a cast parent metal, the base alloy or the heat affected zone.
An analysis of the fracture behavior of weldments from numerous impact
tests on the cast alloy Manaurite 36X in the as received condition tested
both at room temperature and at elevated temperature shows on a plot
three different toughness zones according to the fracture appearance
(Fig. 13). High toughness data are found when the fracture occured within
the weld zone. Fractures at the weld in the heat affected zone show
rather low impact values of about 50 J/m . When the cleavage runs into
the base metal, the low toughness of the cast parent alloy is obtained.

CREEP RUPTURE PROPERTIES

The design of high temperature components is based on an estimation

of their lifetime. Various codes have been developed to standardize
procedures, for example the American Boiler and Pressure Vessel Code.
Guidelines are given for the design of components to operate at
temperatures where creep effects are significant. Creep and creep rupture
data are the main elements for design calculations. In the past years the
main interest turns upon the influence of corrosive environments on the
creep rupture properties of high temperatur alloys (2,3). A number of
mechanisms exist both for strenthening and for weakening of alloys by
corrosion under creep load.
Creep rupture testing of welds and parent metals has been carried
out in a simulated steam reformer atmosphere. According to operational
results of a helium heated methane-steam reformer pilot plant operated by
the nuclear research centre in Jiilich, FRG, a standard process gas
composition has been defined:
50 % H 2 0 , 35 % H 2 > 5 % CH 4 , 5 % CO, 5 % C0 2 ,Fig. 14. This gas
composition with oxidizing and carburizing components corresponds to the
thermodynamic equilibrium at 40 bar and 850 "C for a HpO/CH. mol
fraction of the feed gas of 4 : 1. Oxygen and carbon activities do not
 161

change significantly with pressure between 1 and 40 bar and high pressure
conditions can be simulated at atmospheric pressure.
Some results to date of creep rupture tests on Alloy 800 H, 802 and
Manaurite 36X are shown in Fig. 15-17 as log stress - log rupture time
curves. Data points in the diagrams refer to results of creep rupture in
process gas, the continous curves refer to the average of the scatterband
from tests being carried out within the project in air and in simulated
High-Temperature Reactor helium environment.
At 850 °C the rupture strength both of parent metals and of welds
show very good consistency with the center line of the scatterband. At
950 °C rupture points of welds are situated below the fitted curves while
the parent metals of the same heats show a higher strength. This tendency
already known from the behavior of welds in HTR-helirm atmosphere can
distinctly be seen from the creep rupture data of the cast alloy
Manaurite 36X with its matching filler metal (Fig. 11). It's remarkable
that nevertheless most of the ruptures occured not in the weld but in the
base metal or in the heat affected zone. This behavior is independant
from the use of a matching filler metal or a dissimilar filler metal like
Inconel 625. Finally there were nc apparent differences due to process
gas test atmosphere.

CORROSION TESTS

Components in the high temperature regime have to satisfy high

strength and ductility requirements superposed by corrosion attack. Long
term corrosion tests without mechanical load up to 20 000 h together with
thermogravimetric examinations has been conducted in the above mentioned
atmosphere to evaluate the corrosion resistance of welds and parent
metals. Due to the high oxygen potential of the process gas in the
reformer atmosphere and the high chromium content of the alloys and
filler metals tested, only oxidation phenomena have been found. The depth
of penetration of the internal and selective oxidation determined from
metallographic cross sections increases with temperature and exposure
time. Exposure with intermediate cooling down to room temperature after
every 500 h contributes to a further increase of penetration.
Corrosion of weldments for two alloys (IN 802 and IN 617) which have
widely different behavior is shown in Fig. 18 and 19« Fig. 18 compares
the cross sections of the upper surface of the weld and the base metal
Incoloy 802 after exposure times of 1500 h and 5000 h. Filler metal was
Inconel 625. After 5000 h of exposure internal oxidation penetrates up to
200 am into the base metal, but only 60 \m in the weld. Oxides and
precipitates are present in the grain boundaries and consist mainly of
Al^O . Points of selective oxidation penetrating deeply into the bulk
were analized as TiOp.
Weldments of Inconel 617 (Fig. 19) with matching filler metal show
essentially better corrosion resistance. The penetration in the weld and
the base metal is equally deep and approaches 70 um. The surface is
covered and shielded by an oxide scale of 20 urn. A carbon depleated zone
exeeds the front of internal oxidation. These results are in agreement
 162

with long time corrosion tests of different high temperature alloys for
exposure times of 20 000 h.

CORROSION AND CREEP

Regarding the correlation between corrosion and creep in the process

gas atmosphere we can confirm that grain boundary oxidation initiates the
formation of creep cracks (4). On the other hand, it can be assumed that
creep deformation will accelerate the oxidation of grain boundaries.
Creep cracks on the surface as well as in the center of the specimen are
severely oxidized. Spalling of the oxide scale can rarely be observed.
Even the strained surface of a creep specimen is usually covered by a
thin oxide film which seems to heal quickly in the atmosphere after
cracking.
Considering to the different elongation of the weld zone and the
base metal we find a varying behaviour of the oxide scale. Fig. 20
presents scanning microscope pictures of a creep rupture specimen of
Inconel 617 welded with filler metal 617. Rupture occured in the base
metal, but the weld was strained and constricted, too. The dense oxide
layer on the weld and the parent metal consists of CrpO,. Areas of
the base metal, strongly elongated, show a network of oxide scale cracks,
extending deeply into the bulk metal. As a parent like filler metal was
used, there is no distinct transition of scale morphology for the parent
metal to the weld. Here only some small and superficial cracks in the
oxide layer can be found.
Though creep deformation, grain boundary oxidation, and oxide scale
formation in process gas will influence each other, an effect of the
corrosive atmosphere on the properties of high temperature alloys and
their weldments has not been found to date.

WELD BEHAVIOR OF PLANT COMPONENTS

(AN EXAMPLE)

The closest available simulation to service conditions anticipated

in commercial gasification plants is materials evaluation and exposure in
pilot plants together with the analysis of component failures. In the
high temperature zone of a gasifier shrouds, gas baffles, thermowells and
cooling pipes are fabricated and welded from high temperature alloys.
They are dismantled after fixed operational periods or after failure and
analyzed. Fig. 21 shows a liner of the gasifier welded to a gas baffle
and corroded along the weld seam. The alloy is Incoloy 800 H. Corrosion
attack could be analyzed as alkaline corrosion. Alkaline containing ashes
with a high amount of sodium salts reached the space between the wall of
the liner and the gasifier refractory wall and formed with the aluminium
oxide of the refractory a low melting slag. This compound dissolved step
by step the protective oxide layer of the alloy.
 163

Very interesting is the fact, that the weld resisted to the

corrosion attack. The component was TIG-welded with filler metal 112,
containing up to 23 % Cr and 12 % Mo. Cross sections through the corroded
area next to the weld reveal that the corrosion front penetrated into the
base metal and stopped at the weld bead. Due to its higher Cr-content the
weld surface was able to form adherent Cr20-j scales that were not
attacked. As a corrective measure the design of the liner and the gas
baffle was changed to prevent slag deposit.

SUMMARY

The report present some investigations of welding techniques for

high temperature components used in the Hydrogasification of lignite.
Weldments in various alloys under consideration for gasifier internals
and for steam reforming tubes required for the production of hydrogen
were produced using mechanized processes such as Gas Tungsten Arc
welding, Gas Metal Arc welding and Plasma Arc welding. Base metals were
the nickel-chromium-iron alloys 800 H and 802, the nickel alloys
Hastelloy X and 617 as plates and tubes and the centri-cast alloys 519
and Manaurite 36X. Matching filler metals as well as the recommended
nickel base filler wires were used. For welding high temperature alloys
the following conclusions can be made:
- The mechanized GTAW process is well suited for welding root passes,
filler passes are defect-free, but demand much operating time due to
the little deposit efficiency.
- The mechanized GNAW process is well suited for filler passes, but hot
cracking in Incoloy 800 H cannot be excluded.
- Good welding results are obtained on Inconel 617 with the pulsed GMAW
process.
- Using the PAW process with key holing, austenitic stainless steels and
nickel-base alloys can be welded in the single pass technique with
filler metal.

Weldments were evaluated by:

- tensile tests prior to and after ageing
- corrosion tests in synthetic process gas atmosphere
- Creep reputure tests in process gas atmosphere
- exposure tests as components in the high temperature zone of the
hydrogasification pilot plant.

The ultimate tensile strength of all weldments tested, is lower than

that of the parent metal, even when using different filler metals. The
reduction of strength related to the unwelded alloy amounts to 10 - 20 5?.
After ageing parent metal and weld loose their strength and ductility
with increasing exposure time.
Creep rupture, strength of weldments tested at 850 °C in process gas can
be compared to the properties of the parent metal. But at 950 °C creep
rupture strength is lower than the average of the scatterband of air and
helium data on unwelded specimens.
 164

The principal corrosion effect in the process gas atmosphere is

subsurface oxidation of Aluminum at the grain boundaries and within the
grains.
Test results are supplemented by the failure analysis of a high
temperature component in the pilot plant gasifier.

ACKNOWLEDGEMENT

The work discussed here has been performed under the terms of the
co-operative agreement between
Bergbau-Forschung GmbH
Interatom GmbH
Hochtemperatur-Reaktorbau GmbH
Kernforschungsanlage Jiilich GmbH
Rheinische Braunkohlenwerke AG
dealing with the development of processes for the conversion of solid
fossil fuels using heat from high temperature reactors. Materials
research and development is sponsored by the Ministry of Economics, Sma"! 1
Business and Transport of the state North Rhine/Westfalia.
 165

References

1. Welding of WIGGIN-Nickel Alloys

Henry Wiggin + Co., Publ. Nr. 3339, Sept. 73

2. R.H. Cook,
"Rupture Strengths of PNP Alloy Weldments",
HTMP Technical Memorandum No. 8010/1, 01.10.80

3. P.W. Smith
"The Behavior of Weldments of Wrought Alloys in a
PNP Environment"
HTMP-Report No. 59, April 81

4. M. Schiitze, A. Rahmel
Observations on the Cracking Behaviour of Oxide Scales
During Creep of the Base Metal and on Creep Enhanced
Internal Corrosion
Corrosion in Coal Conversion Systems
Conf. 5th - 7th. May 1982, London
 166

alloy C Ni Cr Mo Co Al Ti Fe Si

INC. 800 H 0,07 33,83 19,72 0,23 0,04 0,47 0,42 Rest 0,06
INC. 802 0,30 31,64 21,18 0,19 0,03 0,43 0,87 Rest 0,32
HAST. X 0,07 Rest 21,98 8,86 2,05 0,24 0,01 18,7 0,30
NIMONIC 86 0,07 Rest 24,60 10,10 0,81 0,05 O,O2 - 0,23
INCONEL 617 0,07 Rest 21,97 8,65 12,12 1,12 0,26 - 0,16
IN 519 0336 24,17 24,55 - - - - Rest 0,84
Manaurite 0,41 32,9 25,18 0,14 - - - Rest 1,42
36 X

Chemical composition of high temperature

RHE1NBRMJN
alloys tested and welded

Tab. I

Base metal Filler metal

Incone 82
Incotoy 800 H
Thermanit 21/33

Inconel 625
Incoloy 802
Incone 82

Manaurite 36 X Manaurite 36 X

IN 519 Inconel 82

Inconel 6 1 7 Inconel 617

Nimontc 86 Nimonic 86

Hastelloy X Hastelloy X

Base metals and filler metals

ISF used for weldments RHE1NBMUN

Tab. II
 167

CH4 i iigas treatment -. SNG

-CO;

HTR

-it? , .
RSO

HKV

OKA

HKV hydrogasification
HTW high-temperature-Wink'ef gasification HTW
OKA steam generation plant
HTR high• temperature nuclear reactor
RSO steam reformer

FJCJ 1 Combination of processes for gasification

of brown coahHKV RSO or HKV/HTW RHEMBRAUN

Weld preparation used for all weldments

ISF
Fig. 2
 168

Current 180-190 A
Voltage 13 V
Trevel sp&ed 14 cm/min
S h i e l d i n g gas 8 l.'min Argon

JSF M e c h a n i z e d GTA rootpasswelding of IN 5 1 9 RHEINBRAUN

Fig. 3

5 mm

t mm

ISF Mechanized GTA welding of Incoloy 800 H RHEINBRAUN

Fig. 4
 169

Fig. 5 Pulsed GMA welding of Inconel 617 RHEINBRMIN

5 mm

TT
/ ^
C.Mrrt>nt
V<itt,)(ic
:i?Fi ft
?7 V
; I Tr.ivf'l ^pn^H ?S cm/min
_J t f i l l e r met^l deposit 10 g/min

PA welding of a 10mm thick Inconel 617

plate in single pass technique

Fig. 6
 170

Typical strain and fracture charakteristics of

RMEMBfMUN
welded high temperature alloys

Fig. 7

• Incoloy 800H Atitmonic QG

A Incoloy 802 ^ Mcooel 6 IT

Haslelloy X

Tensile strength of high temperature alloys

Fig. 8
 171

i c o l o , BO? I

600 900 950

ipmppmtorp (' C)

HOC 9O0

Ultimate lonsilc strength of high temperature alloys and the weldments RUBNBRAUN

Fig. 9

BOO BOO 950

( o m D e f a l u r u ( • CJ

Elongation of high temperature alloys and their weldments

Fig. 10
ro

o a
c
c o a
3
\ I
)n
a 3
2.
o
L II
j :::|t:- r—>
-4
- - . ' •

-pi-)H
I E

.' 1
-*4H--
1 1

--r y
i i_^
i1 I . •

i i
UTT
 173

| T80-
(iller metal: Inconel 82
*u> 1 6 0 "
0)
—ra 140 -
weld fracture
a 100-
E
80-
60- -2 HAZ fracture
40-

-t- -I H
800'C 900'C 950"C

test lemperatuin

Toughness and fracture analysis of Manaurite 36X

weldments tested at ditferent lemperatures

Fig. 13

gas
components vol%
H2 35
H2O 50
CH4 5
CO 5
CO 2 5
standard gas composition
RHEINBMUN
of steamreformer atmosphere

Fig. 14
 174

rupture time h

tiller nuitat 82

Fig
jQ
" '
"J5 Creep Rupture Properties of Alloy 8 0 0 H
Weldments in Process Gas Atmosphere
RH&NBMUN

rupture tirne h
• weidment, filler rtiGlal625

o patent meial

FJQ. 1 6 Creep Rupture Properties of Alloy 602

Weldments in Process Gas Atmosphere ftMHNSIMUN
 175

Creep Rupture Properties of Manaurite 36X

Weldments in Process Gas Atmosphere RHBHBMUH

Fig. 17

e*POEt-f*? tfne 15"0 fi parent

welc e»po5ufe time 5000 h oarent

Corrosion of weldments in process gas atmosphere

Incoloy 802, filler metal Inconei 625

Fig. 18
 176

weld -exposure time 7500 h- parent melaJ

E
a,
o
s
I
iveld -exposure time 5000 fi- parent metal

Corrosion of weldments in process gas atmosphere RHEINBRAUN

Inconel 617. filler metal Inconel 617

Fig. 19

base metal surface 'eld surface

Effect of strain on the oxide scale of RHEINBPAUN

Inconel 617 weldments

Fig. 20
 177

Corroded liner with gas baffle RHEINBRAUN

Fig. 21
 PROCESS PARAMETER EFFECTS ON ARC PHYSICS
AND HEAL' FLOW IN GTAW

J. F. Key, 11. i3. Smart t, J. W. Chan and M. E. Me II wain*

EGt«G I d a n o , I n c .
P. 0 . Box 162 3
Idaho F a l l s , ID 83413

ABSTRACT

Tile need for a more basic understanding of welding processes has

engendered efforts to provide an experimental data base for use in weld
prjrt'.s.s modeling. U'ork to determine interrelationships of neat source,
molten pool, and weld bead geometry i .s described here, and the sensitivity
ol tnese interrelationships to welding variables is discussed. file
efiects on bead morpnology of cuangiug current, welding speed, and iiller
wire speed, witn practLC.il boundaries, are presented as a mult ivariaole
plot. A model of the relationship between process parameters and weld
bead morpnology is presented; the model involves dynamic heat and mass
balances in a control volume containing the moLten pool.

INTRODUCTION

A twofold need for more basic understanding of welding processes nas

emerged. Mechanized (semiautomatic) equipment has developed to the point
that an engineer must select an optimum combination of as many as 20 or
more process v a n a D l e s for a procedure and the welder must often make a
rapid correction during a weld. This is a formidable task for both. The
engineer generally must resort to expensive trial-and-error solutions,
especially if available experience in mechanized welding is limited. The
welder may be forced to stop in the middle of a weld if misalignment,
mismatch, alloy variations, or other problems are encountered. These
problems are caused by (I) a lack of information on process variable
interrelationships witli attending effects on weld structure and properties
and (2) the lack, of closed-loop automatic welding systems that can make
instantaneous process control decisions.

*Geo-Centers, Inc., Newton Upper Falls, MA

179
 180

A program designed to meet these technological needs has been

initiated. The studies are directed toward (1) a detailed understanding
of heat source/molten pool/weld bead geometry relationships and the manner
in which these relationships are influenced by welding variables, and (2)
development of a process model. Results to date are discussed in this
context.

APPROACH

The need for a process model was recognized from the beginning since
this would be the optimum means of communicating quantitative
characterization of parameter interrelationships coupled to weld effects,
and would also serve as a basis for automatic process control programs.
To model a process with a large number of interrelated control
parameters and complex heat transfer, fluid mechanics, and metallurgical
effects on weldments is an exceedingly difficult task. A manageable
approach was selected that divided the process into segments, related to
(a) the heat source, (b) the molten pool, and (c) the weld bead geometry.
Experiments of a correlative nature were conducted to establish dominant
variables or mechanisms. Hie three areas were studied both separately and
in combination, as the appropriate variables dictated. The list of
variables includes current, voltage, shielding gas composition, cathode
tip geometry, weld groove geometry, filler wire speed, and welding speed.
The results will be integrated into an overall process model.

EXPERIMENTAL TECHNIQUES

Weld ing

Gas tungsten arc welding (GTAW) was chosen for initial process
modeling due to its widespread use in semiautomatic welding of complex
alloys in critical structures and its relative simplicity from a physics
viewpoint. The basic process consists of a steady-state dc electric arc
heating the parts to be welded. The welding methods used and the
techniques employed for the associated studies have been described
previously .*-~^
For the weld bead geometry studies discussed in this paper, the
actual parameters used appear to be significant to the results and are
therefore summarized here: Partial-penetration, single-pass,
bead-on-plate welds were made on 12.7-mm (1/2-in.) thick. AISI 304
stainless steel plate using straight polarity direct current. Welds were
made using a commercially available automatic GTAW machine typical of
those used for orbital welding of large diameter pipe in the nuclear
industry. A 2% thoriated tungsten electrode with a 30° vertex angle was
used, with argon shielding gas. Welding currents of 100, 150, and 200 A
were used at respective arc voltages of 9.0, 9.6, and 10.5 V. Welding
speeds were calculated to give gross arc energy values per unit length of
weld bead of approximately 500, 750, and 1000 J/mm. Individual welds were
made using filler wire speeds starting at 0 mm/s (0 ipm) and increasing in
increments of 2.12 mm/s (5 ipm) to some maximum value.
 181

Arc Measurements

The arc characterization portion of this work required sophisticated

measuring techniques to determine both temperature and physical behavior.
Arc temperatures were measured by time-resolved emission spectroscopy,
using a two-line technique.3
Arc characterization near the heat source/molten pool interface uses
high-speed holographic interferometry for both arc temperature
measurements and plasma jet detection. Although che method is unproved
for these applications, optimism is warranted since spectral emission from
the pool should not interfere with measurements. Gas density is
calculated from fringe spacing (Fig. 1) and is related to pressure and
temperature.

Fig. 1. Characteristic holographic iaterferogram of arc,

DISCUSSION OF RESULTS

Process Parameter Influence on Bead Shape

A number of process variables influence the effectiveness of the arc

as a heat source and hence weld pool behavior and weld bead morphology.
Process variables of initial interest were gas composition, tungsten
electrode tip geometry, and weld groove geometry. These variables were
chosen for two reasons: (1) their interrelationships were not well
understood but effects on bead shape were known to be important, and (2)
process developments in mechanized welding were oriented toward use of
narrow-groove joint geometries.
 182

Results ot" this investigation were reported by Key.*- The most

important findings of this study indicated that penetration for welds in
12.7-mm Type 304 stainless steel plate increased remarkably as tne cathode
vertex angle was increased or as helium was added to the argon shielding
gas (Fig. 2 ) . These results were valid for spot or bead-on-plate weld,s
made at low speeds. However, effects of cathode geometry and., gas
composition were not additive. Furthermore, as the welding speed
increased, the benefits of a blunt cathode tip or a high helium content
shielding gas were somewhat diminished.
The previously referenced work also revealed that weld bead
geometries obtained on flat plate cannot be used to predict results of
welds made in grooves of various geometries. Its study of the effect of
groove geometry (75° V-groove, 40° U-groove, and 10° narrow-groove) also
showed that, in general, a more sharply pointed tungsten electrode, e.g.,
30° vertex angle, gave optimum performance due to better arc stability.
This is especially true for 75° V-groove and 10° narrow-groove joint
geometries. The reason becomes clear when one imagines an electrode viitn
a blunt tip shape (e.g., 90° vertex) in either of these joint geometries.
The shortest path to ground is between the electrode shoulder and the '
groove wall rather than between the electrode tip and the groove bottom.
However, the effects of helium additions to the shielding gas were similar
to those for welding on flat plate, i.e., deeper penetration and a larger
bead cross-sectional area.
Many theories have been proposed to explain certain results mentioned
above. Since the purpose of the arc is to provide a source of heat to
melt the base and filler materials (if any), the temperature distribution
of the arc should be an important factor in controlling weld bead shape.
Unfortunately, several other factors complicate analysis of bead shape.
The arc has an associated pressure distribution which may cause a surface
depression and influence fluid motion in the weld pool at high currents.
An electromagnetic field exists about the arc and may be coupled with the
molten pool, also producing fluid motion. Heiple and Roper^ have
suggested that temperature gradients within the pool cause differential
surface tension forces (Marangoni forces) which in turn cause fluid
motion. Whichever the controlling mechanism for fluid flow and heat
transfer in the pool may be, bead shape is fundamentally related to
properties of the arc. The next phase of the program, therefore, studied
effects of welding variables on the temperature and pressure distributions
in the arc as a preparatory step to developing an arc model. (The work on
pressure distributions is in an early stage and will be reported later.)

Welding Variable Effects on Arc Temperature Distributions

An investigation of welding variable influence on arc temperature

distribution as measured by time-resolved emission spectroscopy was
reported by Key et al.2,3 ^he latter reference contains a short review
of prior work in the field. The most important finding of this
investigation was that as helium is added to the shielding gas, the arc
becomes more nearly isothermal, with a much larger diameter (Fig. 3 ) .
High helium compositions exhibit a bi-modal distribution (Fig. 3b) with a
relatively cool center. Although the exajt mechanism for this phenomenon
is uncertain at present, the following possibilities are proposed: (1) a
Fig. 2. Fusion zone profile as a function of electrode tip geometry and shielding gas composition
(150 A, 2.0 s spot-on-plate weld). (a) Effect of electrode geometry in 100% Ar.
Fig. 2b. Interrelationship of electrode geometry and gas composition.
 185

15000

—<—1—1
100ZAr-
X 30* VERTEX .
^ V 1.00mm
1 1—1—1—1—1—1—1—1—1—1—

10000 .. 150A
J\TT

UJ
Q_ 5000
UJ

in (T) CM (T)
i i i

ARC POSITION [mm]

I) Pure argon shield.

15000
—1—1—1—1—1—1—1—1—1—1—1—1—I—1—

30* VERTEX .
1. 00mm
10000 .. 150A
*-*
UJ

UJ 5000 ..
UJ

-*••-«—-H—*—• •+—1 1- -*—»—*•—t—*—*—•- -t—t *—H 1—1—1 11 1—

in •* m N
m
i i i I

ARC POSITION [mm]

(b) 10 Ar/90 He shield.

Fig. 3. Radial temperature distribution at arc midpoint.

 186

plasma dynamic pumping of cool gases into the center of the arc from the
periphery or (2) a high frequency arc rotation that increases the
effective arc diameter seem to be reasonable explanations. Peak
temperature is generally unaffected by electrode vertex angle. Tungsten
electrode vertex angle has only a moderate effect on plasma diameter.
Figure 4 shows semiquantitatively how gas composition and tungsten
electrode vertex angle affect plasma diameter and weld bead geometry.
These schematic representations are taken from data similar to those
presented in Figs. 2 and 3. A 30° vertex angle operating in pure argon
shielding gas at 150 A welding current produces a plasma diameter that is
much less than the diameter of the weld spot (Fig. 4 a ) . Plasma diameter
is defined here as that portion of the arc having reliably detectable
Ar II lines. If the shielding gas composition is changed to 10 vol.%
argon/90 vol.% helium (10 Ar/90 He) with other parameters kept constant, a
more than twofold increase in plasma diameter occurs (Fig. 4(b)), making
the plasma diameter, now nearly equivalent to the spot diameter.
Figure 4(c) shows that a 90° vertex angle in argon causes a constriction
in the plasma diameter near the tungsten electrode and a divergence toward
the workpiece. A weld bead shape midway between those shown in Figs. 4a
and 4b is produced. Lffects of gas composition and electrode vertex
angle, as stated previously, are not additive (Fig. 4 d ) . The plasma
diameters shown in Figs. 4b and 4d are nearly equal near the workpiece and
the bead cross sections are similar. The 90° electrode does cause a
significant constriction in the center of the arc.
Addition of hydrogen to the argon shielding gas (95 Ar/5 H2) has
little effect on peak temperature but does cause an expansion of the
plasma diameter. When using this shielding gas composition, plasma
temperature or shape are insensitive to elecurode geometry or current
changes. Effects of this gas composition on weld bead geometry are
related to recombination of atomic hydrogen at the workpiece surface.
Although welding current has a potent influence on bead shape, it has
only a moderate effect on arc temperature distribution. For example, an
increase in current from 150 to 300 A causes a 10 to 20 percent increase
in peak temperature and a considerable increase in plasma diameter that
depends somewhat on composition, i.e., more in argon less in helium.
These results are valid for stationary spot welds, pulsed-current
welds that produce overlapping spots, and constant-current bead welds at
relatively slow travel speeds. Constant-current bead welds at faster
speeds have little effect on plasma diameter but do alter the geometry of
heat flow into the base metal, which in turn affects bead shape.
Although arc temperature distributions measured by spectroscopy
techniques give insight into effects of the arc on the pool, their
usefulness for modeling purposes is limited by the inability to measure
temperatures at the arc/molten pool interface. Other methods are being
developed to make this measurement.

Arc Modeling

A true understanding of how process variables influence arc

properties requires an electric field model of the arc that is capable of
describing it as an energy source with associated temperature and pressure
distributions and magnetohydrodynamic effects. The detailed temperature
 187

30-degree vertex angle

10% Ar/90% He shield
100% Ar shield 150 A
150 A
Visible
arc

Molten
pool

8ase
material

IUEL 2 0J77 J

a. b.

Tungsten Tungsten
electrode electrode

90-degree vertex angle

90-degree vertex angle 10% Ar)90% He shield
100% Ar shield 150 A
150 A

INEL 2 0276 J

INEL 2 0276 J

C. d.
Fig. 4. Process parameter effects on plasma boundary and molten pool
geometry.
 188

measurements that were performed allow progress on the electric field

model without undue simplifying assumptions. Development of an eLectric
field model of the arc as a function of process variables is in progress.

Weld Bead Geometry Controlling Factors

Another phase of the program reported by Smartt, et al.,^ conducted

concurrently with arc characterization, involved a detaiLed investigation
of factors controlling weld bead geometry.
The GTAW process has been studied extensively in recent years.
Effects of electrode shape 6-11 shieLding gas composition,2,11 and
nominal process parameters^ otl autogenous (i.e., without added filler
metal) bead geometry have been examined. Temperature distribution in the
base metal,13-15 a n d temperature and energy distributions in the
arc ,16-21 have also been investigated. Such studies have led to models
of heat transfer from the arc to the base metal,22-24 anc j prediction of
temperature distribution in the base metal by analytical and numerical
means.12,14,15,25-51 Early analytical work calculated the temperature
distribution in the base metal for the general case of a moving point or
line heat source of a strength proportional ro the total arc energy.
Later work calculated the molten weld pool di.. ^nsions, witli recognition
that the welding arc is of finite size. Yet, as far as is known, there
has been no published study of th.j effects of filler wire addition on GTAW
bead geometry, nor has there been any attempt to predict bead shape as a
function of filler wire addition rate.
This experimental study has been designed to obtain an understanding
of the process-parameter/weld-bead-geometry relationships for GTAW, with
emphasis on the effects of filler wire addition. To ensure that the
information gained is of general applicability, the work has been limited
to straight-line welding using constant values cf current, arc voltage,
welding speed, and filler wire speed for partial-penetration bead-on-plate
welding.
In order to help clarify the relationship of process parameters to
weld bead geometry, penetration is plotted in Fig. 5 as a function of bead
width. The straight line extending from the origin indicates the ratio of
penetration to bead width predicted by a classical point-heat-source
modell^>15j25} neglecting convection in the molten weld pool. To
explain the data, the liberty has been taken to schematically indicate the
effects of changing current, welding speed, and filler wire speed. The
data are grouped into three sets by current, with each set enclosed by a
three-sided region. The boundary lines are labeled for the 200 A current
level. The "autogenous" line forms the upper left (solid) boundary.
Decreasing welding speeds moves one along this line to the upper right.
The extrapolated limit thus obtained for zero welding speed is indicated
by the solid circle. The addition of filler wire results in one moving
downward from the "autogenous" line. The addition of excessive amounts of
filler wire determines the "excessive reinforcement" line which forms the
lower right (dashed) boundary of the region. The right-hand (dashed)
boundary of the region, labeled "stationary, nonautogenous," is for zero
welding speed and parallels the path for filler wire addition. Each
current level has a zero-speed solid circle, the locus of which, as
current :.s varied, determines the "spot-on-plate" line, the upper boundary
 189

BEAD WIDTH (mm) "

F i g . 5. P e n e t r a t i o n as a f u n c t i o n of bead w i d t h .

below which a L L pent: t r a L ion/b ead-w id tli combinations must l i e . In r e c e n t

mod i.'Ling work, Cl i c k s t u in and Fr iedinaii5 2. aave p r e d i c t e d
pene t r a t ion/bcad-wid t\\ r a t i o s c o r r e s p o n d i n g t o t h a t p o r t i o n of the
" s p o t - o n - p L a t e " l i n e which i s p a r a l l e L t o t h e p o i n t - h e a t - s o u r c e l i n e .
Thus far no modeLing e f f o r t known has p r e d i c t e d t h a t p o r t i o n of the
s p o t - o n - p l a t e l i n e which d e v i a t e s from t h e p o i n t - h e a t - s o u r c e l i n e .
Bead width i s e s s e n t i a l l y independent of f i l L e r wire a d d i t i o n . In
c o n t r a s t , p e n e t r a t i o n d e c r e a s e s with i n c r e a s i n g f i l l e r wire s p e e d .
C e n t e r ! i n e p e n e t r a t i o n and reinforcement h e i g h t a r e p l o t t e d a s a f u n c t i o n
of f i l l e r wire speed in F i g . 6. Trie dashed l i n e s in F i g . 6 denote t h e
maximum penetral-.ji.on a t 200 A c u r r e n t . The r e s u l t i n g e f f e c t on t h e shape
of t u e bead c r o s s .section a t 200 A i s i l L u s t r a t e d i n F i g . 7, where t h e
c e n t e r l i n e p e n e t r a t i o n i s n o t i c e a b l y l e s s than t n e maximum p e n e t r a t i o n .
Tne d a t a shovn in F i g . 6 provide i n s i g h t i n t o t h e e f f e c t s of f i l l e r
wire a d d i t i o n on weLd bead geometry. Consider, f o r example, the 100 A
c u r r e n t d a t a in P i g . 6. As t h e f i L l e r wire speed i n c r e a s e s , t h e
reinforcement i n c r e a s e s and the p e n e t r a t i o n d e c r e a s e s ; i n e f f e c t t h e weld
bead aioves up on top of the p l a t e . This g e n e r a l t r e n d i s e x h i b i t e d by aLL
of t h e d a t a in F i g . 6. For 200 A c u r r e n t t h i s e f f e c t i s enhanced a t the
bead c e n t e r l i n e a s mentioned; for t h e lower c u r r e n t l e v e l s t h e e f f e c t i s
distributed.
Reinforcement area increases linearly witli increasing f i l l e r wire
speed as shown in Fig. 8, where reinforcement area is plotted for
nonautogeneous weld samples as a function of the r a t i o of f i l l e r wire
speed to welding speed. The failure of the data to extrapolate to zero
for zero wire speed i s due to a small amount of d i s t o r t i o n in the p l a t e s .
 190

z
111
LJ
U
OL
O
LL

C/L

750 J/tnn

-3 . * 100 A
+ 150 A
0 200 A
-4
in SI in

WIRE SPEED Cmm/m

Fig. 6. Reinforcement and penetration as a function of filler wire speed

for a gross heat input of 750 J/mm.

Weld Bead Geometry Modeling

In discujsmg these results it is worthwhile to briefly review a

simple model of the relationship between process parameters and weld Dead
morpnology. This may be accomplished by placing a control volume around
the molten weld pool, and performing a heat/mass balance on the control
volume. The dominant factor to consider is the heat input, tho source of
this heat being the arc plasma. In the absence of filler additio", and
assuming the weld pool is stationary, pool shape is governed by heat
transfer from the arc through the pool into tne base metal. Thus, the
heat input distriDution from the arc, neat conduction and connection in
the molten pool, and the base metal geometry and composition are
controlling factors.
Motion of the weld pool results in mass flow through the control
volume, this mass being that portion of the base metal through whicn the
control vol.;.-"? passes. The weld pool shape (i.e., the liquid/solid
interface p^. .lion) changes relative to that of a stationary pool, to
accommodate the effect of the enthalpy change (i.e., sensible, latent, and
superheat) of this mass as it is brought up to the pool temperature. When
filler metal is added, the shape of the arc/molten pool interface is
changed by the mess addition, the filler metal is preheated by the arc
prior to entering the molten pool, and the mass addition is localized.
It is possible to obtain quantitative information from this model by
making the following three assumptions: First, bead width (e.g., the
molten pool width) is governed by the distributed heat source and
convection in the weld pool. Second, tne enthalpy change of mass passing
through the fusion zone is equal to heat input from the arc. Third, the
 191

Reinforcement
height

Centerltne
penetration

Maximum
penetration

Fi-g. 7. Typicai weld transverse cross section^ showing measured

quantities. A is reinforcement area; 13 is base metal area,

transverse cross-sectional siiape of the molten pool is elliptical.. It is

also necessary to make a slight change in the control volume.
Specifically, it is assumed tnat the front half of the weld pool is
hemispherical in shape. In order to maintain the liquid/solid interface,
between the weld pool and the uase metal, at tne melting temperature
(Tni) tliere must be a solid region ahead of this interface in which the
temperature is elevated above the nominal base metal temperature ( T o ) .
For simplicity it is assumed that the solid region is a portion of a
spherical shell. The control volume surrounding the weld is extended to
include this solid region (Fig. 9a). Applying the first law of
thermodynamics to the control volume, the enthalpy change of the mass (m)
passing through the control volume is equated to the heat input (Qi n )
from the arc to the base metal. It is assumed that all of the heat input
passes through the molten pool, and that the molten pool may be considered
to be at some average temperature (Ti). Consider the plane A-A,
Fig. 9b; the mass flow through the control volume may be divided into two
parts, that passing through the molten pool (1) and that passing only
through the solid shell (s). Thus the total enthalpy change (per unit
 192

m •* in ID c-N CD CD El
WIRE SPEED/WELD SPEED
Fig. 8. Reinforcement transverse cross-sectional area versus filLer wire
speed/weld speed ratio.

time) of the mass flowing through the control volume is the sum of the
enthalpy changes of the liquid and solid zones. The details of this
analysis are given in Reference 4.
The enthalpy change (per unit time) of mass passing through the
molten pool is plotted as a function of total arc power in Fig. 10. The
related mass flow rates and current levels are also indicated. The
melting efficiency, defined by Niles and Jackson53 a s that portion of
the total arc energy required to melt the weld bead, is listed on Fig.. 10
for the autogenous welding conditions studied. Two trends may be seen.
First, the melting efficiency increases with increasing current; second,
it also increases with increasing welding speed. These trends and the
magnitudes of melting efficiencies are in agreement with the work of Niles
and Jackson.
In the control volume presented above, the weld bead is assumed to
have a semicircular cross section. Assuming the shape to be
semielliptical, for a given cross-sectional area (A) and observed bead
width (W), the penetration (P) may be readily calculaced from the
relationship p = A/TTW . The actual and calculated penetration values are
plotted as a function of current for autogenous welding in Fig. 11. The
good agreement between actual and calculated values indicates that the
bead cross-sectional area for autogenous welds is controlled by heat
input. Thus the decrease in penetration occurring at 200 A current is
caused by a larger bead width than would be expected from a
point-heat-source model.
The penetration of nonautogenous welds may be estimated in a similar
manner. In this case it is necessary to add the enthalpy change of the
filler wire entering the control volume to that of the base metal passing
through the control volume. However, the filler metal was observed to
enter the weld pool in the liquid phase. Thus only the enthalpy change in
 193

rfi

a.
Control volume

INEL-S-35 930

Fig. 9. Geometric model of weld pool showing control volume used for
calculations. (a) Side view. (b) Front view.
 194

ARC POWER CW)

900 14 30 ,"!i00
60 .,

50 + *-500 J/mm
+ -750 J <• m m
a 0-1000 J .• mm
ro 4 0
+ * 4 08 ^
a
E
p
u 30 4.
+3C6 V
<
ft. u
20
5
i
j | ?.04 a.

o x
J i BP. j=
•'.NO

0i 1 0
in
CURRENT (COL,!,

Fig. 10. Rate of metal melting and associated enthalpy change rate as a
function of current, showing melting efficiency for autogenous welds.

4. AUTOGENOUS
c-ACTUAL
E o-CALCULATED

3
z •
o
- ^ ^ ^ ^ c ^ ^ - ^ ^ C ^ X ^ 1000
^O"''' ^^750 J/mm
12 a

HI
jj

1. • •

B
in in Q in
r
CURRENT <.K>

Fig. 11. Actual and calculated values of penetration as a function of

current for autogenecus welds.
 195

going from T m to T[ wad considered. Since tiie width of the pool is

essentially independent of the addition of filler m e t a l , the model was
then solved in tne same manner as for autogenous welding, including use of
the same arc efficiency values. Representative results are shown in
Fig. 12, where penetration is plotted as a function of filler wire speed
for 150 A current at a gross heat input of 73U J/ram, using an arc
efficiency of 3 5 % . The agreement between actual and calculated values is
reasonably good.
The model presented predicts the general trends observed
experimentally and is thus believed to represent the physical phenomena
involved in the welding process. The model is not capable of predicting
bead w i d t h , as it contains no information concerning the controlling
phenomena. These are believed to be the effective heat transfer radius of
Ltu1 a r c ^ and convtciive heat transfer in the molten pool.

150 A 7 5 0 J/mm
1
a--ACTUAL
o - C A L C U L A T E D Cr^

(\j m -* m a) CD O)

WIRE SPEED (mm/«>

Fig. 12. Actual and calculated values of penetration as a function of

filler wire speed for 150 A current, 750 J/mm gross heat input, and
'5 5Z arc efficiency.
CONCLUSIONS

The foregoing results and discassion of an investigation of the

influence of primary welding variables on radial plasma temperature
distribution and weld bead geometry in the GTAW process lead to the
following conclusions:

1. Shielding gas composition has a significant effect on plasma

diameter and radial temperature distribution. It has an
insignificant effect on peak temperature. Further work is
required to accurately model heat flow in an arc shielded by gas
mixtures.
 2. E l e c t r o d i' t e a t n o d e ) v e r t e x a n g l e l i a s a n e g l i g i ci 1 e e t i e c t o n p e a k
temperature. The9 0 ° vertex angle ca u s e s .1 s l i g h t , c o n s t r i c L l o n
in argon a n d a significant, coastri.cti. o n i n 1 0 Ai'/'K) l i e m i d w a v
between electrodes. Thedivergence1 t o w a r d tile w o r k p i e c e m a y
inlluence bead shape.
j. A L a r g e i n c r e a s e i n a r cc u r r e n t g e n e r a l l y e . u s e s a
s i n a l I - t o - m o d e r a t e i n c r e a s e i n p e ii< L e u p e r . i t. a r e a n d p l a s m a
dlamet er .
4. A m o l t e n a n o d e h a s n i )e l l e e l ^ i n a r c t e m p e r a t u r e s i n t i e 1 r e g i o n s
measured. A n y c h a n g e s d u e t o v a p o r i s a t i o n a r ec o n f i n e d t o a
r e g i o n L e s s m a n 0 . 3 0 i;i;a a h o v e t l i e s u n a c e - - a r e g i o n t h a t i s
d i t t i c u i t t om e a s u r e b ye m i s s i o n spec t rose opy .
3. A r cL e n g t u lias n oa p p r e c i a b l e n i t l u r n . - o o n t e m p e r a t u r e
measurements i n tliese e x p e r i m e n t s .
o. F u rc u r i ' e n t s n o te x c e e d i n g i ' 3 0 A , u n d e r t n e e x p e r i m e n t a l
c o n d i t i o n s u s e d i n t i n s w o r k , t h er a t i o o l p e n e t r a t i o n L u Ne.id
w i d t n t o r a u t o g e n o u s w e i J m g a p p r o a c n e s tti.it l o r a c l a s s i c . l l
p o m t — h e a t - s o u r c e m o d e l , 1 ' 4 t o r / . e f ow e l d i n g s p e e d . As w e l u m g
speed i s increased, M\A a s t i l l e r wire i s added t o t h e molten
p o o l , t h ew e l i l b e a d p e n e t r a t i o n d e c r e a s e s s i g ni t L C j u t L y t r o i i i
t h a t a s s o c i a t e d w i t l i a c l a s s i c a l p o i nt - h e a t - s o u r e e m o d e l .
7. F o r i n c r e . i s e e . w e l d i n g . s p e e d o r f i l l e r w' r e s p e e d , t n e v o l u m e o l
m e t a l m e l t e d p e ru n i t t i m e , a n d t h u s t h e D e a d transverse
cross-sectional area, i s r e l a t e d t o t h eh e a t i n p u t f r o m t n e a t e
t o t h eb a s e m e t a l . Specifically, t h ee n t h a l p y c h a n g e o l m a s s
p a s s i n g t h r o u g h t h ef u s i o n /.one m a y b e e q u a t e d d i r e c t l y t o h e a t
i n p u t f r o m t h ea r c t o t h eb * s e m e t a l .
5. Assuming an e l l i p t i c a l t r a n s v e r s e shape for the weld pool,
p e n e t r a t i o n may be c a l c u l a t e d , given an observed bead width.
The above enthalpy model for area and p e n e t r a t i o n may be
extended t o nonautogenous welding by c o n s i d e r i n g the entnalpy
change of f i l l e r wire e n t e r i n g the molten pool.
9. For c u r r e n t l e v e l s above 130 A and up t o 200 A i n c l u s i v e , the
weld-bead-penetration/bead-width r a t i o d e v i a t e s s i g n i f i c a n t l y
from that predicted by a c l a s s i c a l p o i n t - h e a t - s o u r c e model.
S p e c i f i c a l l y , the p e n e t r a t i o n i s c o n s i d e r a b l y lower than would
be expected. However, the volume of metal melted per u n i t time
i s s t i l l r e l a t e d to heat input from tne arc to the base m e t a l .
10. The decreased p e n e t r a t i o n o c c u r r i n g a t high c u r r e n t l e v e l s i s
a s s o c i a t e d with a n o n c l a s s i c a l i n c r e a s e in weld pool width.
This i n c r e a s e in width i s probably a s s o c i a t e d with an i n c r e a s e
i n the arc r a d i u s as c u r r e n t i n c r e a s e s , and changes in
h o r i z o n t a l convection of heat i n the weld p o o l .

ACKNOWLEDGEMENT

The a u t h o r s t h a n k L. D. R e y n o l d s , T . A. Hauk, a n d U. S . W a l l a c e f o r
technical a s s i s t a n c e . Appreciation i s also expressed to
Dr. Nils Christensen for his helpfi,;. discussions, and to Mr. Harry Brown
for technical e d i t i n g .
This work was supported by the U.S. Department of Energy, Office of
Energy Research, Office of Basic Energy Sciences, under DOE Contract
No. DE-AC0 7-7 6IDO15 70.
 197

REFERENCES

1. J . F. Key, "Anode/Cathode Geometry and S h i e l d i n g Gas

I n t e r r e l a t i o n s h i p s in GTAW," Weld. J . , 5 9 ( 1 2 ) : 365-s—370-s ( 1 9 8 0 ) .
2. J . F. Key, M. E. Mcllwain, and L. I s a a c s o n , "A Plasma D i a g n o s t i c s
Approach t o Welding Heat Source Molten Pool I n t e r a c t i o n , " S i x t h
I n t e r n a t i o n a l Conference on Gas Discharges and t h e i r A p p l i c a t i o n s -
P a r t 2, H e r i o t - W a t t U n i v e r s i t y , Edinburgh, UK, September 8 - 1 1 , 1980",
Conference P u b l i c a t i o n Number 189, I n s t i t u t e of E l e c t r i c a l E n g i n e e r s ,
p . 2 35 e t seq.
3. J . F. Key, J . W. Cnan, and M. E. Mcllwain, " S h i e l d i n g Gas Composition
and E l e c t r o d e Geometry Influence on Arc P r o p e r t i e s , " p r e s e n t e d a t AWS
Annual Meeting, C l e v e l a n d , A p r i l 6-10, 1981 and submitted t o the
Welding J o u r n a l .
4. H. B. Smartt and J . F. Key, "An I n v e s t i g a t i o n of F a c t o r s C o n t r o l l i n g
GTA Weld Bead Geometry," Proceedings of ASM Conference on Trends i n
Welding Research in the United S t a t e s , New O r l e a n s , L o u i s i a n a ,
November 16-18, 1981.
5. C. R. Heiple and J . R. Roper, "Minor Element E f f e c t s on GTA Fusion
Zone Geometry," Proceedings of ASM Conference on Trends i n Welding
Research in the United S t a t e s , New O r l e a n s , L o u i s i a n a ,
November 16-18, 1981.
6. W. f. Savage, S. S. Struuck, and Y. Ishikawa, "The E f f e c t of
E l e c t r o d e Geometry in Gas Tungsten-Arc Welding," Weld. J . , 4 4 ( 1 1 / :
489-s--4%-s (1965).
I. R. A. C h i n o s k i , "The E f f e c t s of Varying E l e c t r o d e Shape on Arc,
O p e r a t i o n s , and Quality of Welds in 2014-T6 Aluminum," We1d. J . ,
47(5): 21O-S--222-S (1968).
8. R. A. C h i h o s k i , "The Rationing of Power Between t h e Gas Tungsten Arc
and E l e c t r o d e , " We 1d. J . , 4 9 ( 2 ) : 69-s—82-s (1970).
9. T. N. P e t r i e and E. Pfender, "The Influence of t h e Cathode Tip on
Temperature and Velocity F i e l d s in a Gas-Tungsten A r c , " Weld. J . ,
49(12): 588-s--596-s (1970).
10. A. A. Erokhin, V. A. Bukarov, and Yu. S. Ischenko, " I n f l u e n c e of
Tungsten Cathode Geometry on Some Welding Arc C h a r a c t e r i s t i c s and
Metal P e n e t r a t i o n , " Svar. P r o i z . , 12: 17-19 ( 1 9 7 1 ) .
II. J . F. Key, "Anode/Cathode Geometry and S h i e l d i n g Gas
I n t e r r e l a t i o n s h i p s in GTAW," Weld. J . , 5 9 ( 1 2 ) : 3 6 4 - s - - 3 6 9 - s ( 1 9 8 0 ) .
12. E. Friedman and S. S. G l i c k s t e i n , "An I n v e s t i g a t i o n of the Thermal
Response of S t a t i o n a r y Gas Tungsten Arc Welds," Weld. J . , 5 4 ( 1 2 ) :
408-s—420-s (1976).

13. W. Spraragen and G. E. Claussen, "Temperature D i s t r i b u t i o n During

Welding: A Review of the L i t e r a t u r e t o January 1, 1937," Weld. J . ,
16(9): 4-s--10-s (1937).
14. D. Ros&nthal, "Mathematical Theory of Heat D i s t r i b u t i o n During
Welding and C u t t i n g , " Weld. J . , 2 0 ( 5 ) : 220-s—234-s (1941).
15. N. C h r i s t e n s e n , V. de L. Davies, and K. Gjermundsen, " D i s t r i b u t i o n of
Temperatures in Arc Welding," B r i t i s h Weld. J . , 12(2): 54-75 (1965).
16. J . F. L a n c a s t e r , "Energy D i s t r i b u t i o n in Argon-Shielded Welaing
A r c s , " B r i t i s h Weld. J . , 1 ( 9 ) : 412-426 (1954).
17. H. N. Olsen, "Thermal and E l e c t r i c a l P r o p e r t i e s of an Argon Plasma,"
The Physics of F l u i d s , 2 ( 6 ) : 614-623 (1959).
 198

18. 0 . H. Nestor, "Heat I n t e n s i t y and Current Density D i s t r i b u t i o n s a t

the Anode of High Current I n e r t Gas A r c s , " J . App. P h y s i c s , 3 3 :
1638-1648(1962).
19. H. C. Ludwig, "Current Density and Anode Spot S i z e i n t h e Gas
Tungsten A r c , " Weld. J . , 4 7 ( 5 ) : 234-s—240-s (1968).
20. A. E. F. Gick, M. B. C. Q u i g l e y , and P. H. R i c h a r d s , "The Use of
E l e c t r o s t a t i c Probes t o Measure the Temperature P r o f i l e s of Welding
A r c s , " J . phys. D: Appl. P h y s . , 6: 1941-1949 ( 1 9 7 3 ) .
21. S. S. G l i c k s t e i n , "Temperature Measurements i n a Free Burning A r c , "
Weld. J . , 5 5 ( 8 ) : 2 2 2 - s - - 2 2 9 - s ( 1 9 7 6 ) .
22. E. R. G. Eckert and E. Pfender, "PLasma Energy T r a n s f e r t o a Surface
With and Without E l e c t r i c C u r r e n t , " Weld. J . , 4 6 ( 1 0 ) : 471-s—480-s
(1967).
23. M. B. C. Quigley e t a l . , "Heat Flow t o the Workpiece from a TIG
Welding A r c , " J . Phys. D: Appl. Phys. , 6: 2250-2258 ( 1 9 7 3 ) .
24. E. Pfender, "Energy T r a n s p o r t in Thermal P l a s m a s , " Pure and Appl.
Chem., 5 2 ( 7 ) : 1773-1800 (1980).
25. D. Rosenthal, "The Theory of Moving Sources of Heat and I t s
A p p l i c a t i o n t o Metal T r e a t m e n t s , " T r a n s . ASME, 6 8 ( 1 1 ) : 849-866
(1946).
26. E. F. Nippes, L. L. M e r r i l l , and W. F. Savage, "Cooling Rates in Arc
Welds in 1 / 2 - i n . P l a t e , " Weld. J . , 2 8 ( 1 1 ) : 556-s—564-s ( 1 9 4 9 ) .
27. A. A. Wells, "Heat Flow i n Welding," Weld. J . , 3 1 ( 5 ) : 263-s—2o7-s
(1952).
28. D. K. Roberts and A. A. Wells, "Fusion Welding of Aluminum A l l o y s , "
B r i t i s h Weld. J . , 1 ( 1 2 ) : 553-560 (1954).
29. C. M. Adams, J r . , "Cooling Rates and Peak Temperatures in Fusion
Welding," Weld. J . , 3 7 ( 5 ) : 210-s—215-s ( 1 9 5 8 ) .
30. P. J h a v e r i , W. G. M o f f a t t , and C. M. Adams, J r . , "Tne E f f e c t of P l a t e
Thickness and R a d i a t i o n on Heat Flow i n Welding and C u t t i n g , " Weld.
J^, 4 1 ( 1 ) : 1 2 - s - - 1 6 - s ( 1 9 6 2 ) .
31. J . M. Barry, Z. P a l e y , and C. M. Adams, J r . , "Heat Conduction from
Moving Arcs i n Welding," Weld. J . , 4 1 ( 3 ) : 975-s—1004-s ( 1 9 6 3 ) .
32. Z. P a l e y , J . N. Lynch, and C. M. Adams, J r . , "Heat Flow i n Welding
Heavy S t e e l P l a t e , " Weld. J . , 4 3 ( 2 ) : 71-s—79-s ( 1 9 6 4 ) .
33. N. N. Rykalin and A. I . Beketor, " C a l c u l a t i n g the Thermal Cycle in
the Heat Affected Zone from the Two Dimensional O u t l i n e of the Molten
P o o l , " Svar. P r o i z . , 9: 22-25 ( 1 9 6 7 ) .
34. K. E. Dorschu, "Control of Cooling Rates i n S t e e l Weld M e t a l , " Weld.
J^, 4 7 ( 2 ) : 4 9 - s - - 6 2 - s ( 1 9 6 8 ) .
35. A. A. Erokhin, V. A. Bukarov, and Yu. S. Ishchenko, " C a l c u l a t i o n of
the P r i n c i p a l Parameters of the Pool i n the Welding of P l a t e s , " Svar.
P r o i z , 12: 1-3 ( 1 9 7 0 ) .
36. D. E. S c h i l l i n g e r , I . G. Betz, and H. Markus, " S i m p l i f i e d
Determination of Thermal Experience i n Fusion Welding," Weld. J . ,
49(9): 410-s--418-s (1970).
37. A. A. Erokhin, V. A. Burkarov, and Yu. S. Ishchenko, " C a l c u l a t i n g the
Automatic Welding Conditions for Butt J o i n t s with S p e c i f i e d Underside
Reinforcement," Svar. P r o i z . , 2: 22-25 ( 1 9 7 1 ) .
38. E. G. S i g n e s , "A S i m p l i f i e d Method for C a l c u l a t i n g Cooling Rates in
Mild and Low Alloy S t e e l Weld M e t a l s , " Weld. J . , 5 1 ( 1 0 ) :
4 3 7 - s — 4 8 4 - s (1972).
 199

39. B. M. Krantz and V. D. Coppolecchia, "The Effects of Covered

Electrode Welding Variables on Weld Metal Cooling R a t e s , " Weld. J . ,
51(3): 117-s—-121-s (1972).
40. L- Ts. Kalev, A. Zhelev, and Tsv. Tr. Stoinov, "Certain Special
Features of Calculating the Shape of the S o l i d i f i c a t i o n Front i n Weld
Metal," Avt. Svarka, 2: 8-11 (1973).
41. V. I- Oboturov and S. M. Chashin, "Calculation of the Principal
parameters of the Pool in the Automatic Welding of Aluminum A l l o y s , "
Svar. Proiz, 10: 14-15 (1975).
42. E. Friedman, "Thermomechanical Analysis of the Welding Process Using
the F i n i t e Element Method," Trans. ASME, 97(3): J . Pressure Vessel
Technology, 206-213 (1975).
43. Z. Paley and P. D. Hibbert, "Computation of Temperatures in Actual
Weld Designs," Weld. J . , 54(11): 385-s--392-s (1975).
44. A. Ya. Nedoseka, V. A. Sanchenko, and G. A. Vorona, " D i s t r i b u t i o n of
Temperature When a Concentrated Source of Heat Acts on the Surface of
a P l a t e , " Avt. Svarka, 6: 1-4 (1977).
45. S. C. Hsu, S. Chakravorty, and R. Mehrabian, "Rapid Melting and
S o l i d i f i c a t i o n of a Surface Layer," Met. Trans. B, 98: 221-229
(1978).
46. H. W. Ghent e t a l . , "Heat Conduction Modeling of Two-Dimensional TIG
Weld P o o l s , " Conference on Arc Physics and Weld pool Behavior, The
Welding I n s t i t u t e , London, May 8-10, 1979.
47. V. Pavelic and K. C. Tsao, "Weld Puddle Shape and Size Correlation in
a Metal P l a t e Welded by the Gas-Tungsten-Arc (GTA) P r o c e s s , "
Conference on Arc Physics and Weld Pool Behavior, The Welding
i n s t i t u t e , London, May 8-10, 1979.
48. S. C. Hsu, S. Kou, and R. Mehrabian, "Rapid Melting and
S o l i d i f i c a t i o n of a Surface Due to a Stationary Heat Flux," Met.
Trans. B, i l B : 29-38 (1980).
49. L. A. Bertram and F. J . Zanner, " I n t e r a c t i o n Between Computational
Modeling and Experiments for Vacuum Consumable Arc Remelting,"
Conference on Modeling of Casting and Welding Processes, Engineering
Foundation, Rindge. NH, August 3-8, 1980^
50. S. Kou, "3-Dimensional Heat Flow During Fusion Welding," Conference
on Modeling of Casting and Welding Processes, Engineering Foundation,
"Rindge, NH, August 3-8,198Ck
51. S. S. G l i c k s t e i n and E. Friedman, "Effect of Weld Pool Configuration
on Heat-Affected Zone Shape," Weld. J . , 60(6): 110-s—112-s (1981).
52. S. S. G l i c k s t e i n and E. Friedman, Temperature and D i s t o r t i o n
Transient in Gas Tungsten-Arc Weldments, WAPD-TM-1428 (October 1979).
53. R. W. Niles and C. E. Jackson, "Weld Thermal Efficiency of the GTAW
P r o c e s s , " Weld, J . , 54(1): 25-s—32-s (1975).
 MATERIALS AND WELDING OF SMALL BORE PIPEWORK
FOR NUCLEAR FUEL REPROCESSING PLANT

L Carrick and A Paton

British Nuclear Fuels Limited

Risley Warrington, England

ABSTRACT

British Nuclear Fuels Limited (BNFL) is one of the world's largest

organisations providing an integrated nuclear fuel service. For 25
years the nuclear fuel reprocessing plant of BNFL has been constructed
from two grades of stainless steel; ICCr - !3Ni - INb and lfiCr - 8Ni - ITi.
18Cr - 13Ni - INb is used for plant handling corrosive liquids at high
temperatures and in areas of high radioactivity behind concrete
biological shielding, where access for maintenance is difficult or
impossible and complete reliability is required. 18Cr - 3Ni - ITi is used
for plant items containing less corrosive liquid or where lower levels
of radioactivity exist and decontamination and repairs can be effected.
Both stainless steels have successfully completed the duty required of
them and continue to provide a safe operating plant. However, there
have been difficulties experienced with certain compositional elements
in the stainless steels mainly with respect to weldability and localised
corrosion mechanisms.
The steel manufacturing industry has progressed considerably in the
past 10 years with the introduction of modern steelmaking techniques to
improve the quality of stainless steels. New stainless steels have been
developed with increased corrosion resistance and improved weldability.
BNFL have evaluated some of these new manufacturing techniques and
improved stainless steeli against nuclear reprocessing plant environments.
The results of this work in relation to the problems which occurred in
the original materials will be discussed in the paper.
Traditionally, manual welding techniques have been employed for the
welding of BNFL's small bore austenitic stainless steel pipes both in the
fabrication shop and on site. Corrosion resistance is of paramount
importance in nuclear fuel reprocessing plant and the welding techniques
employed have to produce velds with a corrosion resistance adequate for
their duty which necessitates the use of special welding procedures.
These special procedures and the problems associated with them will be
discussed in the paper.

201
 202

The aim of BNFL is continually to explore ways to improve the

reliability of welded joints and reduce the weld repair rate during
manufacture. To this end much effort has been given over the last 3
years to an evaluation of automatic orbital TIG welding of small bore
austenitic stainless steel pipes in order to overcome the problems
experienced by manual welding. The paper will discuss this work and
explain how the developments have been successfully extended to the
application of automatic orbital welding to the site construction of
BNFL nuclear processing plant.

INTRODUCTION

The Nuclear Fuel Reprocessing Plant of British Nuclear Fuels

Limited (BNFL) at Sellafield, Cumbria, England has successfully
completed 25 years of reprocessing magnox nuclear fuel. Construction
is now in progress to extend existing reprocessing facilities. Plans
are also well advanced for a new oxide fuel reprocessing plant and a
plant to convert fission product waste to a solid glass like form for
long term storage.
As a result of this expansion it will be necessary to weld
approximately 500,000 stainless steel tube butt welds to a very
demanding acceptance standard. The tubes will range in size from
10 - 200 mm NB; 1.7 - 6.0 mm wall thickness, with the most common
size 40 mm NB; 2.1 mm and 4.1 mm wall thickness.
In the past difficulty has been experienced with manual welding
techniques in striving to achieve the required acceptance standards
whilst retaining a reasonable production rate. In an attempt to
enhance the production of high quality welds with accurate repeatability
changes in materials and fabrication techniques, resulting from three
years of development, have been made.
The majority of this paper is devoted to describing the processes
of modifying material specifications and introducing a new fabrication
technique into site construction procedures in order to resolve both
material and fabrication problems experienced over a number of years.
Although essentially a practical review, an attempt has been made to
provide a limited appreciation of all the technical aspects involved.
This approach has been used so that the complete transition from the
development programmes to practical usage can be covered.
 203

MATERIALS

In choosing materials and fabrication processes for reprocessing

plant the criterion of main concern is corrosion, consideration of such
criteria as strength and fatigue being of secondary importance in most
cases. However the process of selection must include a need for
complete integrity and minimal repair and maintenance. Recently,
increased emphasis has been given to weldability, a trend which is
probably true for most design authorities.
The environment in very general terms ranges from comparatively
dilute nitric acid up to 10 molar at temperatures between ambient and
100°C, organic solvents, and cooling water. Austenitic stainless steels
are therefore adequate in resisting the general corrosion attack.
However there are a number of localised corrosion mechanisms which
need to be given close attention.
For areas of the plant which experience less severe environments
and are maintainable 18Cr - 8Ni - ITi welded with 18Cr - 9Ni - INb
filler has been used successfully. However there have been isolated
problems of knife line attack at welds and localised pitting in the
parent material, described elsewhere by Crook and Shawl The stringer-
type inclusions resulting from the use of titanium as a stabiliser do
tend to give the material a poor surface condition.
To improve the corrosion resistance and remove the need for the
titanium stabiliser a change of material from 18Cr - 8Ni - ITi to 3O4L
has been made.
For highly active conditions, where little or no maintenance is
possible, a fully austenitic 18Cr - 13Ni - INb stainless steel is used
welded by autogenous root fusion and an 18Cr - 9Ni - INb filler. The
fully austenitic weld metal is prone to hot cracking during solidification.
Gross cracks which follow the centre line of the weld are the extreme
cases but micro cracks can occur in the root of the welds.

Stainless Steel Evaluation

Due to the improvements in stainless steel manufacturing technology

over the last 10 years it was decided that a re-appraisal of stainless
steels should be completed before commencing the main expansion programme
at Sellafield. Four different stainless steel suppliers, six different
grades of stainless steel in three product forms, and three different
methods of refining were used to provide a total of 25 variants for
evaluation. The aim was to evaluate the corrosion resistance of the
different grades of stainless steel in three environments, and to compare
the currently used 18Cr - 13Ni - INb with the other grades.
The programme produced the same ranking order of corrosion resistance
for the material groups in all three liquid environments: 310L > 3O4L >
18.13.1. The initial superiority of the 310L variants was found to
diminish on prolonged testing and their corrosion rates eventually
approached those of 3O4L.
 204

Although the 3O4L produced a wider spread of results than any other
material group all results were favourable so further effort is now
underway to establish a compositional specification for a 3O4L stainless
steel that could be used in reprocessing plant for primary containment.
The weldability of 3O4L pipework in various bore sizes and thicknesses
will also be evaluated.
In spite of the inherent welding difficulties associated with
18Cr - 13Ni - INb, BNFL has over 25 years operational experience with
this material, so that any change will not be taken without proper
justification, against the technical and commercial constraints likely
to be imposed in meeting the future programme of planned expansion.

WELDING

The plants at Sellafield have been manually welded using the

tungsten inert gas and manual metal arc processes. For tube welding,
the tungsten inert gas process has been used almost without exception.
The 18Cr - 8Ni - ITi and latterly 3O4L materials have given little
technical and procedural problems, the only difficulty has been
achieving the very demanding acceptance levels.
The 18Cr - 13Ni - INb material however, has suffered from
metallurgical defects as described previously. The main problem
being associated with the autogenous root fusion which is made with
the intention of presenting a homogeneous structure to the highly
active reprocessing liquors. The strict procedural controls necessary
to overcome these metallurgical problems have been aggravated by the
need to meet the standard acceptance levels, particularly the uniform
positive root profile in the bore of the tubes. The welder skill
required to produce acceptable welds in 18Cr - 13Ni - INb is high, this
is coupled with an increased level of welder control and supervision,
which must ensure that the following parameters are followed accurately:

1. Platform type weld preparation with tightly toleranced dimensions.

2. Tacking procedures transverse to the weld seam.

3. Current levels as low as possible but sufficient to give full

penetration.

4. Arc maintained close to the centre line of the weld preparation.

5. Automatic arc decay to be used.

6. Decay the arc of the root run up the side wall.

 205

During a routine radiographic inspection of a butt weld between a

100 mm NB 18Cr - 13Ni - INb tube and a 180° bend an overpenetrated weld
was chosen for further inspection. Metallurgical examination revealed
gross micro cracking, (figure 1)

magnification x 7

Figure 1 Weld section showing microcracks in the root.

An evaluation programme established that the level of phosphorus

in the stainless steel was a major contributor to the problem; restraint
and a high heat input welding procedure were the other factors. This
experience has led to a reassessment of the phosphorus levels in 18Cr -
13Ni - INb stainless steel. The maximum allowable level has now been
reduced from 0.04% to 0.015%.
Against this background a number of alternative fabrication and
welding techniques have been investigated to enhance the production of
high quality welds with accurate repeatability. The technique found
most promising and the one developed and finally introduced was automatic
orbital welding.
 206

ORBITAL WELDING

Equipment

Two different automatic orbital TIG systems have been purchased for
development and use on the Sellafield Site. Both units are of the
horseshoe type, that is the welding head clamps directly onto the tube.
Typical welding heads of the two systems are shown in Figure 2.

Figure 2 Welding heads from two different orbital systems

first system on the left.

With these type of systems a number of welding heads are required

to cover a range of tube sizes. The first system required four welding
heads to cover tubes from 10 - 80 mm NB, the second system required three
welding heads to cover tubes from 10 - 150 mm NB.
All the basic development work was carried out using the first
orbital system which was not equipped with a programmer or an automatic
arc voltage control. Without these functions difficulty was experienced
in achieving welds with no root concavity in tubes over 50 mm NB.
 207

The second orbital system had a built in programmer and automatic

arc voltage control.

Development

As stated in the section on Materials the main criterion for

material choice is corrosion so prior to any commitment being made to
orbital welding a programme of work was carried out to compare the
corrosion resirLance of orbital and manual welds." Results of the
programme showed there was no significant difference between orbital
and manual welds.
Initial welding development in the laboratory was concentrated on
18Cr - 13Ni - INb tubing ranging from 15 to 50 mm NB and 3.3 to 4.5 mm
wall thickness. Tolerant weld procedures were developed using the same
weld preparation as for manual techniques.
Development work continued on 3O4L tubing ranging from 15 to 50 mm
NB and 1.7 to 2.7 mm wall thickness. Weld procedures were developed
using both single run and double run techniques with square edge weld
preparations.
At this stage the equipment was transferred to a BNFL production
environment at Sellafield Site where the weld procedures already
developed were successfully repeated using industrial welders under
site workshop conditions.
After several months of handling by industrial workers the orbital
equipment had a calibration check and was found to be accurate, thus
giving confidence in the reliability of the equipment.
The next important step was to apply orbital welding to the new
construction site. A project was identified and agreements made with
the mechanical contractor to introduce orbital welding with the help of
BNFL's experience.

Project for the introduction of Orbital Welding

Tube in 3O4L up to 80 mm NB was chosen to be welded by automatic

orbital welding. An 'on site' training area was established where the
contractors staff and welders were familiarised with the equipment and
trained in the procedures of orbital welding by BNFL's staff. Single
run procedures were used for simplicity so that the controls on the
equipment needed minimum change during the welding process.
To ensure that weld position, access difficulties and operational
sequences typical of those likely to arise in practice were experienced,
a rig was fabricated with various bends and tube lengths duplicating the
tube assemblies in the plant.
 208

During welding on the rig using the single run procedures variations
in weld penetration were observed on tubes of the same size. An example
of this phenomenon is shown in figure 3.

magnification x 7
Figure 3 Sections of two welds produced by identical welding
parameters but showing variation in bead penetration.

After a thorough evaluation of this problem it was concluded that

variations in weldability from cast to cast were being experienced.
A literature review of cast to cast variations in TIG welding has
been published by Lucas and Eardley, where similar problems are discussed.

CAST TO CAST WELDABILITY

Having identified the problem the more tolerant two run welding
procedures were used but, although there was some improvement, the
problem was not entirely resolved. It was at this stage that the second
orbital system was purchased for another project using 3O4L tubing. The
two projects together had 34 different casts of 3O4L in various tube
sizes.
To quantify the cast to cast problem a simple test was developed.
A sample of tube was placed under a TIG torch at a set distance and using
standard welding conditions the number of pulses to produce penetration
were counted. Penetration was determined by a light sensitive device in
the bore of the tube. With argon as a shielding gas the casts of 304L
available required between 5 and 30 pulses to achieve penetration. Using
this test each cast of 3O4L was given a number relating to it's weldability.
It was also possible to define the limits of the welding procedures, for
example a procedure for 50 mm NB 304L tube could weld casts with pulse
counts of between 10 and 24. New welding procedures were developed on the
first orbital system to provide the flexibility needed to weld a group of
 209

casts in a particular tube siae. This approach was successful in enabling

most of the tubes of a given size to be welded with one procedure.
With the second orbital system, the problem of cast to cast weld-
ability was resolved by changing the shielding gas. Various gases and
gas mixtures were evaluated, the most successful being Argon/5% Hydrogen.
Using the test described above, the pulses to penetration were reduced
from 5 to 30 with Argon to 4 to 8 with Argon/5% Hydrogen. This resulted
in all the casts in one size of tube being welded by one procedure.
To achieve success with the Argon/5% Hydrogen shielding gas, automatic
arc voltage control is necessary to maintain a short arc, without which
excessive porosity occurs.
Once the problem of cast to cast weldability variation was resolved
welding on the rig was finalised and production welding started.

NEGATIVE PURGE

The second project required the welding of tubes up to 160 mm NB.

With the larger diameter tubes difficulty was experienced in achieving
welds without root concavity. To overcome this problem a negative purge
technique was introduced with tubes over 50 mm NB.
To operate this technique a negative pressure of 25 mm of water was
applied to the tube during the welding of the root run, resulting in a
more positively penetrated root bead. This can be seen in figure 4 which
compares two weld profiles on a 150 mm NB tube, one with and one without
negative purge.

magnification x 7

Figure 4 Two weld profiles showing the effect of negative

purge - left with negative purge, right without
negative purge.
 210

ORBITAL WELDING IN PRODUCTION

With the technical problems resolved both orbital systems started

production welding. It is still in the enrly stages but the welders are
enthusiastic, production rates are higher than manual techniques and
rejection rates are less than 10%. Access to difficult areas is easier
with orbital welding heads and the completed welds do not require any
grinding before radiography.
It must be emphasised that effort has been made to give the orbital
welding system every chance of success, especially in creating a good
weld joint. To help with this two auxiliary equipments were developed.

AUXILIARY EQUIPMENT FOR ORBITAL WELDING

Past experience has shown that for successful operation any welding
system, automatic or manual, requires accurate assembly of the weld joint.
Weld preparation, alignment of the tubes and weld gap must all be set to
tight tolerances. This is especially so when welding on site and to
achieve this two sets of equipment have been developed and used effectively
in achieving their objectives.

Tube end forming tool

The eccentricity tolerance of the 'as received1 tubes is t 10% of the

mean wall thickness. On tubes up to 50 mm NB, these tolerances do not
create undue problems with the weld joint set up, but above this size the
eccentricity and thickness tolerances can give difficulties both in machin-
ing accurate weld preparations and producing accurate bore alignment in
the weld joint. With tubes 80 mm NB and above, benefits have been gained
by rounding up the tubes before and after machining the weld preparations.
The rounding up operation is carried out using a portable hydraulic form-
ing tool (figure 5) which is 165 mm diameter, 275 mm long and weighs
23 Kgms. The tool uses a range of collets and dies to cover the range of
tube sizes; changing these items to accommodate the different tube sizes
is a simple and quick operation. Several tubes with damaged ends have
been rounded successfully.
 211

Figure 5 Tube end forming tool

Tube alignment jig

It is common practice when fitting small bore tubes together for

welding on site to use an angle bar and two G clamps to assemble the
joint. If tubes will not come together accurately force is often used
to fit the tubes into the angle bar before restraining with the clamps
rather than re-setting the pipe assembly to give an unrestrained weld
joint. Highly restrained weld joint assemblies are not recommended and
in severe cases can lead to weld defects or subsequent problems with the
weld in operation due to the residual stresses in the pipe work.
To overcome these problems a development programme was initiated to
produce a tube alignment jig which could be used with either manual or
orbital welding systems. The jig produced (figure 6) has the following
specification:-

1. Length 525 mm, radial clearance 100 mm.

2. Weight 10 Kgms.

3. Load applied to the tubes sufficient to overcome the weight of a

6 metre length.
 212

4. Maximum tube bore mismatch of 0.4 mm max.

Figure 6 Tube alignment jig.

The jig operates by placing the tubes on the clamps and activating
three pneumatic cylinders which bring the tubes together to form a joint
assembly ready for welding.

SUMMARY

The material used for the areas of the reprocessing plant which
experience less severe environments and are maintainable has been changed
from 18Cr - 8Ni - ITi to 3O4L.
A stainless steel evaluation programme has been completed to compare
other grades of stainless steel to 18Cr - 13Ni - INb for the highly active
areas of the plant. The results of this evaluation programme give a rank-
ing order of 310L> 304L > 18.13.1. Other important factors will now be
considered before a decision on the material choice is finally made.
Orbital welding has been introduced to site conditions. Using a tube
end rounding tool and a jig alignment tool to set the weld joint assembly
an orbital system has achieved weld pass rates of over 90%.
This is very much in the early stages but sifficient work has been
carried out to clearly demonstrate that orbital welding of small bore
austenitic tubes to a high standard on site can be successful.
 213

Cast to cast variations in weldability were identified with 3O4L

but by using various gas mixtures and grouping the casts using a special
test the problem was resolved. A longer term solution to the problem
is well advanced.
To give positive tube bore profile around the complete tube
circumference with tubes over 50 mm NB a negative purge technique
has been used with satisfactory results.

REFERENCES

1. C E Crook and R D Shaw,

Materials Applications in the Windscale Nuclear Fuel
Reprocessing Plant - Pipework, BNES, London, 1979.

2. B Poulsan and R Robinson,

Orbital Welding Development for BNFL (corrosion test)
NEI Clark Chapman Power Engineering Limited, Industrial
Technology Unit, February 1979.

3. W Lucas and J A Eardley,

Effect of Cast to Cast Material Variations in TIG
Welding - Literature Review, Welding Institute Research
Report, December 1981.
 mm 8R
(MA mm m mi

--•;.»V""
' - " . '.'• ,<'-•

" • ' ^
• • \ ^ ~

: • > - 'U'' 1
"-•'•••' ,-/"
v^;:' ~ - " ••'".. :,

if

Sliilf

/-
Ig-p
 HIGH TEMPERATURE PROPERTIES OF THE SEAM WELD PORTION
IN 1% C r - 0 . 5 % Mo ERW BOILER TUBING
AFTER LONG-TERM AGING

T. Nakazawa, Y. Hosoi, H. Shimada, H. Komatsu,

T. F u j i i * , and H. Nogata**

I l l - F u n d a m e n t a l Research Laboratory
(Corrosion and Heat R e s i s t a n t M a t e r i a l s )

* : I-Fundamental Research Laboratory

( S t r e n g t h and Toughness)
** : Kimitsu Works
Nippon Steel Corporation
1618 Ida, Nakaharaku, Kawasaki, 211 Japan

ABSTRACT

This study is aimed at the evaluation of reliability of

the seam weld portion of 1% Cr-0.5% Mo ERW (Electric Resistance
Welding) boiler tubings for high temperature use. High temper-
ature properties of materials aged in the temperature between
500 and 600°C for 5000 h were evaluated by a slow extension
rate test (SERT) and a low cycle fatigue test at high tempera-
tures. Microscopic observation was also made with a scanning
electron microscope and a transmission electron microscope.
Proof stress and tensile strength obtained by SERT tend
to decrease with an increase in aging temperature. It is con-
firmed that the strength of the seam weld portion is higher
than that of the base metal. The fatigue crack propagation
rates studied by test pieces with a side-notch either at the
base matal or at the seam weld portion were found to be lower
at the seam weld portion compared with that at the base metal.
Dislocations of high density and fine precipitates of carbides
were observed with electron microscopy in the seam weld portion.
It is also found that the size of dislocation cells formed
during fatigue is smaller in the seam weld portion than in the
base metal. It is concluded that higher strength of the seam
weld portion is closely related to the above mentioned micro-
structure originated in the dissolution of carbides during
ERW process.

217
 218

INTRODUCTION

There are certain advantage for ERW (Electric Resistance Welding)

tubings over seamless tubings: ovallity, uniform thickness, and excel-
lent mechanical properties, since ERW tubings are made of hot-rolled
plates. On the other hand, there still seems to exist unknown factors
regarding the extent of weld defects only because of the limit of re-
solution power of presently available non-destructive inspection methods.
Recently, however, Haga et al^-' classified the welding phenomena
in ERW into three types according to (1) the distance and shape between
the weld point and the V-convergency, and (2) periodicity of the weld-
ing process. They found that the first and the second type phenomena
would be essentially free from any defect if sufficient metal is melted
and squeezed out from the groove upon welding. Furthermore, Ando et
al2) have devised an ingenious method for shielding completely the
welding zone with inert gas. Based on these findings and the device,
Nippon Steel Corporation has developed high quality small diameter ERW
tubings with defect free seam weld.
The present study is aimed at the evaluation of high temperature
properties of the seam weld portion produced through above mentioned
process especially from the metallurgical view point.

MATERIALS AND PROCEDURES

PRODUCTION OF ERW TUBINGS

Table I. Chemical Composition of the Materials Tested (%)

C Si Mn P S Cr Mo

0.13 0.23 0.52 0.017 0.006 0.90 0.51

Chemical composition of the ERW tubings tested is shown in Table I.

A 250 ton heat was refined by a basic-oxygen process and a vacuum
degassing process to obtain a clean steel, which is subsequently cast
into 24 ton ingots. The ingots were slabbed, hot-rolled, and cut into
plates with the prescribed width. The plates were tube-formed and
seam-welded by a high-frequency induction method. (Capacity: 600 kW,
frequency: 180 kHz.) After welding, both internal and external flahses
were automatically removed, and the entire length of the seam weld por-
tion was continuously inspected by ultrasonic testing for mill control.
The tubings were then normalized at 920°C for 10 min followed by temper-
ing at 670 c C for 1 h and inspected automatically by eddy current test-
ing and ultrasonic testing.
 219

The materials employed for the present study were chosen randomly
out of the tubings. Size of the tubings were 63.5 mm in outer diameter
and 7.5 mm in thickness. The width of the fused zone was approximately
0.1 mm in the trangential direction.

PREPARATION AND TESTING

ERW tubings were cut into semicylindrical shells and flattened to
plates 180 mm in length and 70 mm or more in width. The plates were
subjected to the same heat treatments as those of tubings described
above and, then, aged for 100 to 5000 h at temperatures in the range
of 500 to 600°C.
From the plates, following test pieces were prepared in the trans-
verse (circumferential) direction: (1) Charpy V-notch impact test pieces
7 mm in thickness, (2) slow extension rate test pieces 6 mm in diameter
and 30 mm in gauge length, and (3) low cycle fatigue test pieces 5 mm
in thickness, 10 mm in width and 20 mm in gauge length. Some of these
test pieces with the seam weld portion were notched within the regions
of 0.5 mm from the fused zone after the plates were etched: 1 mm in
notch depth, 45° in notch angle, and 0.1 mm in root radius.
A slow extension rate test^) was carried out at 550°C under a
strain rate of 5.5 x 10~ 6 /S. The push-pull strain controlled fatigue
tests were performed in a servo-hydraulic fatigue machine using an
axial extensometer to control longitudinal strain in the total strain
range of 0.3 to 0.6%.

METALLOGRAPHY:
Microstructure of the above described test pieces was examined
using a high resolution scanning electron microscope of field emission
type (SEM) and a lMeV transmission electron microscope (TEM).

RESULTS AND DISCUSSION

IMPACT PROPERTIES
The results of the transverse Charpy V-notch impact test are given
in Table II. The impact properties of the base metal and seam weld
portion were not significantly influenced by aging temperature and time
except for the overaged weld portion at 600°C for 5000 h. Absorbed
energy tested at 40°C is comparable between the seam weld portion and
the base metal although the crystallinity of fracture appearance in the
seam weld portion is higher. Absorbed energy tested at 80°C is higher
in the seam weld portion than that in the base metal, which is supposed-
ly due to finer microstructure and higher strength of the seam weld
portion.
The results of the longitudinal Charpy V-notch impact test reveal
that 50% FATT is about -20°C, and the shelf energy is higher than or
equal to 160J both in the base metal and in the seam weld portion.
 220

Table II. Comparison of Transverse Charpy V-notch

Impact Properties of Long-term-aged Materials
with those of Normalized-and-tempered
Materials (N-T)

Minimum Value of Absorbed Energy(J)(*)

Posi- and Crystalinity in parenthese
tion
of Tested at 40°C Tested at 80°C
Notch Aging time (h) Aging time (h)
N-T 100 1000 5000 N-T 100 1000 5000

Base 51 48 41 43 58 64 62 63
Metal (30%) (25%)(30%) (30%) (0%) (0%) (0%) (5%)
500
Aging Temperature (°C)

Seam 34 64 44 47 83 90 83 79
Weld (40%) (45%)(45%) (40%) :io%)(25%) (15%) (15%)

Base 55 55 56 68 72 76
Metal (20%)(35%) (20%) (0%) (0%) (5%)
550 dittos dit-
Seam 56 73 55 tos 88 98 95
Weld (40%)(35%)(45%) (10%) (5%) (15%)

Base 55 38 52 72 63 74
Metal (25%)(45%)(35%) (0%)(10%) (5%)
600 dittos dit-
Seam 65 50 22 tos 93 83 62
Weld (40%) (55%)(60%) (15%)(25%)(25%)

(*) Minimum value of three test pieces 7 mm in thickness.

HIGH TEMPERATURE PROPERTIES

Before describing the present results, typical tensile properties
and creep rupture strength of the ERW tubings are shown in Table III.
The results of a slow extension rate test at 550°C are given in
Fig. 1. All the test pieces with the seam weld portion located at the
center were ruptured at the position 5 to 7mm apart from the center. It
is recognized that 0.2% offset proof stress values are higher than those
of the test pieces without the seam weld portion but for a test piece aged
at 600°C. These results suggest that higher strength of the seam weld
portion raises the effective strength of the neighboring base metal.
Compared with Table III, it is noticed that the proof stress of
normalized-and-tempered materials in Fig. 1 is higher than that evalu-
ated under a higher strain rate at 550°C and shown in Table III. It
is suggested that rearrangement of dislocations will occur to some ex-
tent during SERT as the stage I creep phenomenon. It is further noticed
 221

Table III. Tensile Properties and Internal Pressure Creep

Rupture Strength for 5000h Life of l%Cr -0.5%Mo
ERW Tubings (MPa)

Testing Temperature (°C)

Ambient 500 550 600

Tensile Strength 441 432 392 297

Yield Strength 259 184 193 184

Creep Rupture (*)

- (279) 129 64
Strength

(*) Estimated value by extrapolation from data at 550°C and

600°C.

400
B.Mi Base MetaL
S.W: Seam Weld
350 • </>

_ 300
a
0.
5 250
to
200
CO
150

CM
d IOO I
50

0 I B.M.ISW. 3.M.ISW.
Normalized 500 *C 550 C
and
Tempered Aging
#
I
3.M.I&W.
6O0*C
Temperature

Fig. 1 Effects of Long-Term Aging Treatments on Slow

Extension Rate Tensile Properties at 550°C
under Strain Rate of 5.5 xlO" b /S.
 222

that the proof stress does not significantly change after long-term
aging at 500°C indicating the stage II creep henomenon while the proof
stress of the base metal is reduced so much by long-term aging at 550°C
giving the creep rupture strength. Thus, the combination of long-term
aging treatments with SERT provides a good means for evaluating the
creep rupture strength.

10

Fig. 2 Comparison of Low Cycle Fatigue Life at 500°C between

Test Pieces with and without the Seam Weld Portion
of Materials aged at 500°C for 5000h.
Tha test was made in an axial strain control mode,
E>/ery test pieces was fractured in the base metal
even of those with the seam weld portion.

Nf
Fig. 3 Effect of Long-Term Aging Treatment at 600°C on Low
Cycle Fatigue Life at 600°C. The modes of test and
fracture were the same as those of Fig. 2.
"N-T" denotes "nonnalized-and-tempered".

The results of a low cycle fatigue test of smooth test peices are
given in Fig. 2 and Fig. 3. Most of fatigue cracks were initiated and
propagated in the base metal even with the seam weld portion at the
center of test pieces. Some cracks initiated in the seam weld portion
propagated away from the fused zone, the others being left as sub-cracks.
 223

Figure 2 shows that there is no substantial difference in the fatigue

life between test pieces with and without the seam weld portion as pre-
dicted by the fracture location, namely ai the base metal. Figure 3
further shows that long-term aging at high temperature does not affect
the fatigue life substantially in contrast to the results of SERT shown
in Fig. 1. It is supposedly due to the microstructural change specific-
ally introduced in strain cycling, as wiil be described later.

10-

8.0/tmA:ycle)
o

(da/dN-IO.2/im/cycle)

O 200 AOO 600 BOO I0O0 1200

N

Fig. 4 Comparison of Load Drops due to the Fatigue Crack

Propagation between Test Pieces with and without
the Seam Weld Portion of Materials aged at 550°C
for 5000h. The mode of test was the same as that
of Fig. 2, except that the test pieces were
notched on one sides of them. (Strain range: 0^0.4%)

The results of a fatigue test of side-notched test pieces are given

in Fig. 4 where the load decreases with number of fatigue cycles. It is
to be noted that the decrease rate is lower for the test piece with the
seam weld portion. The propagation rates of the fatigue cracks estimat-
ed from Fig. 4 are 8.0 ym/cycle for the seam weld portion and 10.2 ym/
cycle for the base metal respectively. The obtained values for the
propagation rate are considered to be reasonable based on our previous
study^) with an austenitic stainless steel, where actually measured
values for the propagation rate agree well with the estimated values.
It should be noted that cracks initiated in the seam weld portion tend
to propagate away from the fused zone into the neighboring zone where
the preexisting banded microstructure is disarrayed by upsetting during
ERW process. Thus, finer microstructure of the fused zone and dis-
arrayed microstructure of the neighboring zone are considered to be the
reasons for the higher resistance to crack initiation and propagation
in the seam weld portion.
 224

MICROSTRUCTURE

Base Metal Seam Weld

N-T

500'C

Ageing

(soooh)
550 'C

600 C

40p
Fig. 5 Optical Micrographs of the Base Metal and the Seam
Weld Portion, showing the coagulation of pearlite
lamella after long-term aging treatment. "N-T" de-
notes "normalized-and-tempered".

550t 5000h 6 0 0 C 5OOOh

Fig. 6 SEM Micrographs of the Seam Weld Portion as

Normalized-and-Tempered, and Those a f t e r Long-Term
Aging. N-T denotes "normalized-and-tempered".
 225

The micrographs of the materials tested are given in Fig. 5 through

Fig. 10. Figure 5 shows that no substantial difference is discerned
both in size and distribution of ferrite and pearlite between the base
metal and the seam weld portion. It is clearly shown even in Pig. 5
that the carbides in pearlite lamella are coagulated by long-term aging
at 600°C. As shown in Fig. 6, observation of the microstructure by a
SEM reveals that long-term aging does not affect the morphology of
carbides essentially at 500°C but promotes the coagulation considerably
at 550°C or 600°C.

(a) Base Metal (b) Seam Weld

Fig. 7 TEM Micrographs of the Base Metal and the Seam Weld
Portion as Normalized-and-tempered.

J»" '. V".

(a) Base Metal (b) Seam Weld

Fig. 8 TEM Micrographs of the Base Metal and the Seam Weld
Portion after Aging at 55O°C for 5000h.
 226

Figure 7 shows that dislocation density is higher in the seam weld

portion than in the base metal of the normalized-and-tempered material,
and that more unidentified precipitates are observed in the former.
It is resonable to speculate that carbon, chromium, and molybdenum are
fully in solid solution in the fused zone and in the neighboring zone
after ERW. Higher hardenability due to the above mentioned solid solu-
tion is considered to be the cause of the higher dislocation density in
the seam weld portion as normalized. It is further reasonable to consider
that chromium and molybdenum in solid solution retard the climbing motion
of dislocations^) and also tend to form finely-dispersed carbides in the
zone during tempering.
Figure 8 shows that needle-like precipitates and unidentified
plate-like precipitates are observed both in the base metal and in the
seam weld portion of the material aged at 550°C for 5000h. The former
precipitates are identified as M 0 2 C It is recognized that the preci-
pitates are rather finer and more densely-dispersed in the same weld
portion. There are alternative explanations for the morphology of
precipitates shown in Fig. 8 : (1) Dislocations of high density in the
seam weld portion shown in Fig. 7 (b) provide more nucleation sites for the
heterogeneous precipitation during long-term aging, (2) finely-dispersed
precipitates shown in Fig. 7(b) grow into the observed precipitates
during long-term aging. It is evident from comparison between Fig. 7
and Fig. 8 that the recovery and the coarsening of precipitates are
responsible for the decrease in the proof stress after long-term aging
at 550°C which is shown in Fig. 1. The coherency strain of precipitates
M02C in the <100> direction of ferrite matrix for a steel bearing l%Cr&)
is estimated to be 0.03, which is much larger than the elastic strain of
the order of 0.001 or less estimated for the creep rupture strength in
Table III. Hence, it is likely that the coarsening of precipitates is
not substantially affected by the applied stress of the creep rupture
test. Thus, it is confirmed from the metallurgical view point that the
combination of long-term aging treatments with SERT provides a good
means for evaluating the creep rupture strength of the present materials.
Figure 9 shows the change in sub-structure of the normalized-and-
tempered material in the course of low-cycle fatigue at 550°C. Compared
with Fig. 7, it is recognized that distinct cellular structure of dis-
locations is developed by strain cycling both in the base metal and in
the seam weld portion. There are alternative explanations for the
difference of Fig. 9(b) from Fig. 9(a) : (1) Dislocations of higher
density in Fig. 7(b) are aggregated by strain cycling to form the finer
cells, (2) Individual dislocations are limited their movement by the
precipitation of chromium and molydenum of larger amount from super-
saturated solid solution to form the finer cells in Fig. 9(b).
Figure 10 shows the sub-structure of a test piece which was
fractured in a low-cycle fatigue test at 550°C after a long-term aging
treatment at 550°C. Compared with Fig. 8, it is recognized that coarse
needle-like precipitates formed by long-term aging are dissolved by the
subsequent strain cycling in the range of 0 to 0.4%. Micrographs shown
in Fig. 7 through Fig. 10 and the fatigue test results shown in Fig, 3
suggest that the dissolution of the precipitates takes place in so un-
expectedly early stage of strain cycling and thereby the long-term aging
treatment does not affect the fatigue life. It is recognized that plate-
 227

like precipitates are more densely distributed in the seam weld portion
and considered to be the cause of higher resistance to crack initiation
and propagation.

Base Metal (b) Seam Weld

Fig. 9 TEM Micrographs of the Base Metal and the Seam Weld
Portion after Low-Cycle-Fatigue at 550°C suceeding
to Normalizing-and-Ternpering. Cyclic strain of
0 to 0.4% was applied to the test pieces for about
half the life. The mode of test was the same as
that of Fig. 2.

(a) Base Metal (b) Seam Weld

Fig. 10 TEM Micrographs of the Base Metal and the Seam Weld
Portion after Low-Cycle-Fatigue Fracture at 550°C
suceeding to Aging at 550°C for 5000h. (Strain
range: C v O . 4 % , fatigue life: 2450 cycle).
 228

SUMMARY

High temperature properties of the seam weld portion of l%Cr-0.5%Mo

ERW tubings were evaluated by a slow extension rate test (SERT) and a
low-cycle fatigue test, in comparison with those of the base metal.
Normalized-and-tempered materials were aged at temperatures in the range
of 500 to 600°C for 5000h, in order to assess the effect of long-term
aging on high temperature properties.
Test results by SERT are summarized as follows:
(1) 0.2% offset proof stress of test pieces with the seam weld portion
is higher higher than of those with the base metal except for the
test pieces aged at 600°C for 5000h.
(2) The proof stress of normalized-and-tempered materials at 550°C is
higher when tested under lower strain rate, whereas the proof stress
of the base metal at 550°C is reduced close to the creep rupture
strength by aging at 550°C for corresponding time to the rupture
life.
The low-cycle fatigue test results are summarized as folows:
(1) Less chance of crack initiation and lower crack propagations rate
in the seam weld portion lead to no substantial difference in the
fatigue life between test pieces with and without the seam weld
portion,
(2) Long-term aging at high temperature does not affect the fatigue
life essentially in contrast to the results by SERT.
Microstructural aspects of the above-described properties may be
summarized as follows:
(1) Banded structure of ferrite and pearlite developed in the base
metal is disarrayed in the seam weld portion by upsetting during ERW
process.
(2) Higher dislocation density and fine precipitates are observed in
the seam weld portion than in the base metal of the normalized-and-
tempered material.
(3) Dislocation cells of smaller size are developed in the seam weld
portion by strain cycling in the fatigue test for the normalized-
and-tempered material.
(4) At the seam weld portion, more densely-dispersed needle-like pre-
cipitates identified as M02C, together with unidentified plate-like
precipitates are observed when aged at 550°C for 5000h.
(5) The needle-like precipitates, Mo 2 C, are dissolved by strain cycling
in the range of 0 to 0.4% at 550°C, supposedly in the early stage of
strain cycling.
Thus, the seam weld portion of l%Cr-0.5%Mo ERW tubings has proved
to be superior to the base metal both in high temperature tensile pro-
perties and in resistance to low cycle fatigue cracking at high tempe-
ratures if the tubings are produced with care so as to avoid defects
during ERW process. In addition, the ERW tubings has also proved to
possess high toughness in the transverse direction at above 40°C and in
the logitudinal direction at above -20°C,
 229

REFERENCES

1. H. Haga, K. Aoki, and T. S a t o , "Welding Phenomena and Welding

Mechanisms i n High Frequency E l e c t r i c R e s i s t a n c e Welding - 1 s t
R e p o r t " , Weld. J . , 59 (7) : 208-S216-S ( 1 9 8 0 ) .
d i t t o s , "The Mechanisms of Formation of Weld Defects i n High-
Frequency E l e c t r i c R e s i s t a n c e Welding", i b i d , 6 0 ( 7 ) : 104-S-109-S
(1981).

2 . N. Ando, T. Fukuda, H. Nogata, Y. Kimiya, Tetsu t o Hagane

( J a p a n e s e ) , 6 8 ( 5 ) : S339 ( 1 9 8 2 ) .

3 . Y. H o s o i , M. S a k a k i b a r a , T. Nakazawa, and H. Shimada, Tetsu t o

Hagane ( J a p a n e s e ) , 6 6 ( 1 1 ) : S1190 (1980).

4. A. Uehara, Y. Hosoi, and H. Shimada, unpublished work.

5 . H. Morikawa, and T. S a t o , " I n t e r a c t i o n of S o l u t e Atoms i n a - I r o n

with P o i n t Defects induced by I r r a d i a t i o n i n a lMeV E l e c t r o n
Microscope", P r o c . 5th I n t . Conf. on High Voltage E l e c t r o n
Microscopy, Kyoto, 531-534 (1977).

6. H. Morikawa, H. Komatsu, and M. Tanino, " E f f e c t of Chromium upon

Coherency between M2C P r e c i p i t a t e s and a - I r o n M a t r i x i n 0.1C-
10Ni-8Co-lMo-Cr S t e e l s " , J . E l e c t r o n Microscopy, 2 2 ( 1 ) : 90-101
(1973).
JtZO
 SUBSTRUCTURE OF A 16-8-2 GTA WELD AS
OBSERVED THROUGH TEM

J. R. Foulds and J. Moteff

Department of Materials Science

and Metallurgical Engineering
University of Cincinnati
Cincinnati, Ohio 45221

ABSTRACT

A weldment section from 'formed and welded' large diam-

eter (0.91m OD) thin wall (12.7mm) 316 stainless steel pipe
for structural use in the hot-leg of the liquid-metal fast
breeder reactor is characterized microstructurally through
TEM. The GTA weldment (welded with 16-8-2 filler) examined
is sectioned from pipe in the 'as-received' (formed + welded
+ annealed + straightened) and the 'reannealed* (formed +
welded + annealed + straightened + reannealed) condition.
The optically observed weld metal dendritic substructure
appears as a dislocation substructure with low-angle (<1.0°
misorientation) boundaries demonstrated to be stable on re-
annealing and built up on interdendritic planes of mismatch.
Using microhardness as a room-temperature strength probe be-
fore and after the reanneal, it is shown that the weld and
parent metal exhibit distinctly different elevated tempera-
ture recovery characteristics. The contribution of stable
low-angle dislocation boundaries (subboundaries) to recovery
at high temperature explains the consistent differences be-
tween weld and parent metal behavior.
The substructure size and subboundary energy or misori-
entation are shown to be effected by variations in the weld
heat-input rate. Subboundary analyses indicate the build-up
or polygonization occurs on interdendritic 'growth' disloca-
tions accommodating the mismatch between adjacently growing
dendrites.

The authors gratefully acknowledge the receipt of welded specimens

for this study from Dr. J. W. McEnerney and Dr. V. K. Sikka of the Oak
Ridge National Laboratory, Tennessee. This work was supported, in part,
by the Department of Energy through the Reactor Research and Development
Division under Contract AT(11-1)-2107.

INTRODUCTION

Much of the research in the area of austenitic stainless steel

welding has been understandably concerned with the establishing of a
'macroscopic' compatibility between weld and parent metal through vari-

231
 232

ations in the welding process and parameters. 'Macroscopic1 here refers

to the resulting chemistry, optically-observed grain structures, and re-
sults of conventional non-destructive inspection techniques as applied
to weldments. This 'macroscopic' weld/parent metal compatibility, how-
ever, does not explain consistent differences between weld and parent
metal elevated temperature behavior. 1
Most recent high temperature mechanical property data on automatic
gas tungsten-arc girth welded (with 16-8-2 filler) large diameter 316
stainless steel pipe weldments show a higher yield strength, lower ulti-
mate tensile strength, lower stress-rupture life and stress-minimum creep
rate curves for the weld metal as compared with the base metal. The ob-
servations, in fact, represent a general comparison of weld and parent
metal elevated temperature mechanical behavior seen for a variety of
austenitic stainless steel weldments. In each case proferred explana-
tions for these based on chemistry and grain size have been unable to
explain all the property differences; moreover, they have been found
wanting with regard to possible generalizations to other weldments.
Beyond the 'macroscopic' compatibility mentioned above, the subtle
but consistent weld/parent metal behavior differences must be explain-
able by consequences of the very nature of the weld pool solidification
process. As a first step in understanding weld metal mechanical behav-
ior based on its characteristic dendritic structure, a 16-8-2 GTA seam-
welded 316 stainless steel pipe weldment section is substructurally char-
acterized using the Transmission Electron Microscope (TEM). Although the
primary application of the material investigated here is that of a struc-
tural piping component in the hot-leg of the liquid-metal fast breeder
reactor, the results of this characterization may be applicable not only
to all welds, but also to all rapidly-solidified crystallographic systems.

EXPERIMENTAL PROCEDURE

Material

The large diameter (0.91m OD) thin wall (12.7mm) 316 stainless steel
pipe was manufactured in 1.83m lengths by a commercial vendor through a
forming and longitudinal seam-welding process as per ASME Code specifi-
cations. Subsequent to the post-weld solution-anneal (1060°C, 1/2 h in
Argon + water-quench) the pipe was subjected to a 'sizing and straighten-
ing1 to eliminate the distortion from residual stress recovery during
annealing. This introduces some cold-work into the pipe section. The
'formed + welded + solution-annealed + straightened1 condition is hence-
forth referred to as an 'as-received' one. A transverse weldment sec-
tion from pipe in this condition was used for this study. The section
was then re-solution-annealed at 1065°C for a half hour in Argon, air-
cooled, and re-examined. Characterization was thus conducted on the
weldment in both 'as-received' and 'reannealed1 conditions.
Figure 1 is a macroetched view of the transverse weldment section
studied with the welding pass sequence. Table I is a spectrographic
chemical analysis of the plate (base metal away from weld) and weld
metal (from center of weld with minimum dilution) and Table II details
 233

the actual welding parameters employed.

1.0 cm

5. Oscillated hot-wire
4. Oscillated hot-wire
3. I'ot-wire GTAW
2. Cold-wire GTAW - Root
1. Oscillated cold-wire

Figure 1. GTA weldment section with beads and pass sequence.

Table I
Chemical Analysis of Formed and Welded Pipe

Element C Mn P S Si Nl Cr Mo
Plate (wt. %) 0.066 1.63 0.032 0.009 0.58 11.29 16.09 2.85
Weld (wt. Z) 0.053 1.57 0.014 0.014 0.49 9.42 16.18 1.95

Element V Nb Ti Co Cu Al B N
Plate (wt. %) 0.06 <0.01 0.01 0.17 0.22 0.01 0.001 0.039
Meld (wt. %) 0.04 <0.01 0.01 0.06 0.13 <0.01 0.001 0.041
 234

Table II
Welding Parameters for GTA Seam-Weld

# Passes 5

Welding Position Flat

Filler Metal 3/32" dia. 16-8-2 Cr-Ni-Mo

Electrode 2% Thoriated Tungsten

Electrode dia. 5/32"

Inert gas 20%Ar, 80%He @ 50 cft./hr.

Current dcsp

Amperage, A 400A, 11-14V on inside pass

& 250-275A, 11-14V on root pass
Voltage, V up to 400A, 11-14V on rem. passes

Feed Speed 6"/tnin. on inside pass

(FS) 8"/min. on root pass
9"/min. on rem. passes

Heat-input rate 4.4-5.6 x 10 on inside pass

2.1-2.9 x 10^ on the root pass
A.V.
(joules/in) <3.7 x 10^ on rem. passes
FS
Preheat 60°F min. on all passes
300°F max. on inside pass

Table III
Average Grain Intercept (ym)

Region

Weld HAZ Base

Outside Root
Outside Center Inside Outside Center Inside
pass pass
Size
128 87 121 106 113 132 73 114
(}Jm)
 235

Microscopy

(a) Optical: Optical mi-

croscopy on the base metal, heat-
affected zone (HAZ) and weld met-
al regions enabled a characteri-
zation of the grain structure and
the absence of any effect of re-
annealing on it. Figure 2 shows
the typical weld metal and base
metal grain structure. The quan-
titative information on grain
sizes estimated by a random line-
intercept technique indicates in
Table III that the weld and base
metal grain sizes are comparable.
Also, no significant grain coars-
ening was visible in the HAZ
300
areas.
Figure 3 is the optically
observed typical weld metal den-
dritic substructure with approx-
imately a 2% by area QTM globular
delta ferrite^ in the 'as-re-
ceived' condition. Reannealing

Figure 2. Comparison of (a)

weld metal and (b)
base metal grain
structures.

Figure 3. Typical weld metal

dendritic structure
with globular ferrite.

300
 236

only very slightly coarsened the substructure with no measurable change

in the size, content and distribution of ferrite. The secondary inter-
dendritic arm spacing was visually estimated at about l-3ym. The loca-
tion of ferrite (inter- or intra-dendritic) is difficult to determine
optically. It must be mentioned that TEM observations indicated ferrite
globules varied in size from 0.7 to 6.0pm with a large fraction between
1.5 to 2.5ym before reannealing. Reannealing appeared to decrease the
maximum and median size. In both the 'as-received1 and 'reannealed' welds
ferrite was generally found (~two-thirds of ferrite observed) not associ-
ated with any dislocation boundaries, which, from the succeeding discussion
on dislocation boundaries, indicates it to be mostly intradendritlc.
(b) TEM: Thin foils about 0.3ym thickness were examined on a JEOL-
Model JEM 200A electron microscope operated at 200 kV. The microscope
was equipped with a 30° tilt-360° azimuth (tilt axis position) specimen
stage enabling defect characterization. To prepare the foils, 2.5mm
square x 0.4mm thin elements were sectioned from the weldment of Figure 1
so that the element thickness was parallel to the pipe wall thickness di-
mension. Approximately twenty elements from each weldment (as-received
and reannealed) representing various locations in the base metal, HAZ and
weld metal were selected. These were next ground to -0.15mm thickness
and electropolished to perforation in a 90:10 Acetic Acid:Perchloric Acid
by volume electrolyte using a twin-jet technique at 40 volts and a cur-
rent density of 1.75 Amps/cm .
A single grain representing the 'average' dislocation structure for
that foil, estimated visually, was selected for characterization. Char-
acterization included, where possible, a measure of:
(i) dislocation density, p (cm~^) — estimated from two-dimensional micro-
graphs taken under a variety of two-beam conditions using Ham's technique,
(ii) dislocation Burgers vector, b, and line direction, t — after deter-
mining the grain orientation with respect to the incident electron beam
under fixed conditions of tilt and azimuth (represented by a stereographic
projection), the Burgers vector was determined using the ^'o — 0 criterion
for invisibility, where ~g is the operating diffracting beam vector (for
edge dislocations of the b = •§• <110> kind, a is the lattice parameter,
g**D x t is too small for a visible image with "g*T> ==0 ) ^ ;
(iii) subgrain or substructure size, A (ym) — obtained from two-dimensional
micrographs in a manner as for grain size estimation;
(iv) subboundary or low-angle dislocation boundary misorientation, 0(°) —
measured by the shift in a Kikuchi line intersection in traversing the
boundary.
Details on TEM methods used may be found in Reference 5.

Microhardness

In order to establish a correlation between mechanical behavior and

microstructure it is necessary to use a test specimen having dimensions
within the range over which microstructural variations would be estimated
to have little or no effect on deformation characteristics. In the case
of weldments, fairly small test specimens would be necessary for this.
As an alternative, 'microhardness' would be an effective strength micro-
probe, providing an estimate of strength over relatively short distances.
 237

In this study room-temperature hardness is used as a room temperature

strength estimate.
Microhardness measurements were made using a Wilson Tukon-Model MO
microhardness tester with a 136° diamond pyramid indenter and a 500 g
load. Indentations were made on the mechanically polished (to 0.3ym
alumina) weldment surface of Figure 1 in both 'as-received' and 'rean-
nealed' conditions. The indentation scheme is as shown in Figure 4.

1 cm
Figure '. Jloom temperature microhardness indentation scheme.

Since microstructural variations within a given weldment zone (weld, HAZ

or base metal) were evident over the pipe thickness dimension, the large
volume of data was reduced to mean values of a given horizontal row for
each zone separately. These values were then plotted as a function of
position or distance from the outside pipe surface (location of row).
Each mean value represents an aveiage from at least ten microhardness
readings. In case of weld metal and HAZ where the number of indenta-
tions per row for each zone was less than ten, the specimen was re-
polished to a depth of at least 1.0mm and the indenting process re-
peated. This enabled a larger sample space for estimation of a mean
hardness. Hardness values are reported in diamond pyramid hardness
(DPH) units of kg/mm 2 .
The method of data reduction employed here assumes an insignificant
(with regard to effect on mechanical behavior) microstructural variation
along a row approximately tangential to the pipe circumference within a
given zone of base metal s HAZ or weld metal. The validity of this assump-
tion is supported where the hardness variance does not significantly ex-
ceed that commonly observed for a microstructurally homogeneous test
specimen under similar test conditions.

RESULTS AND DISCUSSION

General Substructure Characterization

The microstructure as a function of location in the 'as-received'

weldment section is visible in Figure 5. The accompanying microhardness

y
fell i ,- 7:i

Figure 5. Microhardness and dislocation structure variation across

'as-received' GTA weldment section.
 239

numbers are shown as strength estimates obtained at room temperature with

a 300 gram load and 136° diamond pyramid hardness (DPH) indentor. The
obvious difference in weld and base metal microstructure is the existence
of curved low-angle dislocation boundaries (subboundaries) in the weld,
not seen in the base metal. Areas in base metal exhibit a random dis-
location distribution with no segregation into subboundaries. Table IV
summarizes some measured aspects of the dislocation substructure with the
hardness values for locations described in Figure 5.

Jable IV
Dislocation Substructure for Locations
in Weld Metal Zone of As-received GTA. Weldment

Substructure
Average Estimated
Misorientation Substructure
Location DPH (kg/mm ) (degrees) Size (Mm)
14 177 0.34 2.8
15 170 0.38 2.5
23 169 0.46 2.3
33 168 0.37 2.3
34 176 N.M. C N.M.
35 173 N.M. 2.8
Root (" 42 166 No Subboundaries
Pass |_ 43 166 No Subboundaries
53 169 N.M. N.M.
54 176 0.57 1.7

a: XY Location X — Row from outside to inside increasing in number

in Fig. 5 Y — Location examined from left to right along row
b: from two-dimensior.al line-intercept estimate
c: N.M. = Not measured

Within a given zone of weldment (base metal, HAZ or weld) microstruc-

tural variation is more evident across the pipe wall than along the cir-
cumference. Pipe straightening not only accounts for a dislocation density
higher than normally observed for solution-annealed material, but also re-
sults in the variations in microhardness and dislocation density seen a-
cross the pipe wall (non-uniform cold-work across pipe thickness). For
the base metal and HAZ, dislocation densities and hardness values were
generally higher nearer the pipe surfaces and decreased toward the center.
T-he weld metal has a dislocation substructure of size comparable with the
optically-estimated interdendritic arm spacing, and appearing to decrease
somewhat with a smaller bead cross-section (or lower heat-input rate). The
'subgrains', which are regions of crystal separated by subboundaries had
average misorientations between neighbors of the order of 0.5°, also ap-
pearing to decrease with a decreasing heat input rate. In fact, subbound-
aries at the root pass with the lowest heat-input rate, are believed to
 240

have energies or misorientations so low as to be very poorly-defined and

characterized as 'absent'.
Reannealing resulted in a lowering of hardness and random disloca-
tion density everywhere with the persistence of subboundaries in the
weld metal region. Figure 6 and Table V are microstructural descrip-
tions of the 'reannealed' weldment.

136
..r-

152 A\

157*.

/142

V N

Figure 6. Microhardness and dislocation structure variation

across 'reannealed' GTA weldment section.

In the base metal and HAZ the dislocation density and hardness de-
creased such that regions of higher density and hardness before reanneal-
ing became regions of lower density and hardness after the reanneal. In
fact, microhardness and dislocation density in the base metal and HAZ were

* For a subboundary misorientation, 0 < 0.2°, the density, p, of disloca-

tions in a subboundary - that observed in a 'solution-annealed' random
dislocation structure (~10° c m ~ 2 ) ; p = 1/d2, where d = mean dislocation
spacing (cm) in boundary estimable from: d = |b|/0 (rad.)-
 241

Table V
Dislocation Substructure Evaluation for Locations
in Weld Metal Zone of Reannealed GTA Weldment

Substructure
Average Estimated
9 Misorientation Substructure
Location DPH (kg/mm ) (degrees) Size (lim)
13 151 0.69 2.3
14 161 0.68 4.0
23 153 0.43 4.2
33 162 0.40 2.2
34 166 0.37 2.5
35 152 N.M. C 5.0
Root f 42 145 N.M. 1.4
Pass L 43 152 No Subboundaries
44 150 No Subboundaries
53 162 0.75 3.6
54 160 0.62 3.7

a: XY Location X — Row from outside to inside increasing in number

in Fig. 6 Y — Location examined from left to right along row
b: from two-dimensional line-intercept estimate
c: N.M. = Not measured

used to evaluate the strengthening effect of random dislocations, some of

which are as a result of pipe straightening. Figure 7 shows the strength-
ening as a linear function of the square root of dislocation density, y/p ,
representing the forest hardening effect of random dislocations.
Reannealing of the weld metal resulted in a slight coarsening of the
substructure (increase in substructure size) accompanied by an apparent
increase in the subboundary misorientation, the decreasing size and mis-
orientation with decreasing bead width or heat-input rate being maintained.
The persistence of subboundaries in the weld results in a 'residual1 weld
metal strength even after reannealing manifesting itself as a somewhat
higher (than base metal) weld metal hardness. Subboundaries have been
shown to strengthen by reducing the effective grain size in a Hall-Petch
fashion. Reannealing effected some recovery of the cold-work in straight-
ening and no recrystallization in either weld or base metal regions.
Thus, the optically visible weld metal dendritic structure appears as
a dislocation substructure when viewed through the TEM, the 'optical' den-
dritic arm spacing being comparable with the TEM estimated substructure
size. Also, the subboundaries, as will be shown below, have characteris-
tics indicating they occur as a result of solidification and correspond
to the interdendritic etched areas seen optically.
Solidification, and consequently substructure characteristics, are
complex functions of local undercooling in the melt, dependent on a variety
of process and material parameters.6 For this reason, a quantitative cor-
relation between the two substructure characteristics measured (size and
 242

180 180

170

160

ISO
DPH(kg/mm2) = 134 + 5 . 0 6 ( / p cm" )

140

130 — 130
I . I I I I
0 1 2 3 4 5 6 7 8 9

Square root dislocation density, / p , (cm ) x 10

Figure 7. Strengthening effect of random dislocations in base metal.

boundary misorientation) and a single parameter such as weld heat-input

rate is not attempted here. In general, however, factors such as a low
heat-input rate resulting in a faster cooling rate (G.R, G E thermal
gradient in melt, R = growth rate) effect a finer substructure. Factors
contributing to a reduction in G/R or an increased propensity for den-
dritic growth effect higher energy and higher misorientation (tighter,
more dislocation dense) subboundaries.

Subboundary Characterization

A total of fifteen subboundaries or networks from the weld metal

regions of both the 'as-received' and 'reannealed' weldments were ana-
lyzed. Since the subboundaries in both cases had identical character-
istics, these are discussed without specific reference to the weldment
condition.
The manner of subboundary characterization carried out is based on
the various mechanisms for their formation proposed by Lindroos and
Miekk-oja. Curiously, they recognized rapid solidification as a means
of generating dislocation substructures to study their influence on de-
formation. Following are a few definitions of terms used to describe
the boundaries:
forest plane: this is the plane of the network and contains the
dislocations making up the boundary. In cases where the network exhibits
a curvature the forest plane is continuously varying and determined from
analysis over a limited region of boundary;
 243

forest dislocations: these are one or more families of line defects

lying in the forest plane, often sessile and forming the basis for a net-
work;
glide-source dislocations: these are dislocations produced within
the crystal (not at the forest) capable of moving under stress, which on
reaction with the forest dislocations result in a network;
junction dislocations: these are produced as a result of reaction
between a glide-source dislocation and a forest dislocation; since obser-
vation of the network cannot differentiate between a glide-source and a
junction dislocation segment, all segments not of the 'forest' kind are
referred to as junction dislocations;
node: this refers to the joint common to several dislocation seg-
ments in the network, usually 3-fold with three being the number of seg-
ments at the point;
Burgers vector plane: the plane defined by and containing the Burgers
vectors of the segments at a node.
The subboundaries examined were all curved hexagonal networks (see
Fig. 8(a)), the dislocation segments at each node being nearly coplanar.
Of the three dislocation families, the 'forest' one was easily distin-
guishable by the large edge component in its Burgers vector. Junction
dislocations are generally expected to have small edge components if the
glide-source dislocation is of strong screw character. Figure 8(b) de-
tails a typical network examined. Note the large edge character of the
forest dislocation family ' C and the nearly pure screw nature of the
junction dislocation families 'A' and 'B'. Table VI shows results on
the forest planes of subboundaries analyzed. The large edge character
of the 'immobile1, 'non-glide' forest dislocations and the often approx-
imate {110} boundary as forest plane support the theory that "polygoni-
zation' or 'build-up' occurs on dislocations (forest) accommodating the
mismatch between adjacently solidifying dendrites or dendrite arms.
The last to solidify interdendritic regions are the ones accommo-
dating the crystallographic mismatch (misorientation) between the ad-
jacently grown dendrites in the 'growth substructure.' This explanation
of rapid growth first proposed by Rutter and Chalmers is consistent
with actual observations on the substructure , also seen here. The mis-
match regions — interdendritic in this case — comprise a system of dis-
locations at least locally planar and capable of accommodating a rela-
tive 'twist' about an axis perpendicular to this plane (twist boundary),
a 'tilt' about an axis in the plane of the boundary (tilt boundary), or
both a tilt and twist (tilt 'n' twist boundary) about two orthogonal
axes.^O The plane here is called a 'forest plane' and the accommodating
dislocations — 'forest dislocations.' Evidently these dislocations by
the very nature of their creation seldom belong to the slip systems,
primary or conjugate, of either dendrite at the forest plane. The net-
works or subboundaries seen after solidification are often the result
of mobile glide-source dislocations from intradendritic regions inter-

* The approximate {110} mismatch plane with its curvature suggests the
deviation from classical prismatic growth expected with dendritic solidi-
fication; it also supports its existence as an interdendritic mismatch
plane.
 244

acting with the forest dislocations either during cooling or post-weld

stress/temperature treatments. This process of substructure formation
is similar to the 'microscopic polygonization1 observed with recovery
after 'turbulent' flow in FCC crystals, wherein interaction of glide
dislocations with intracrystalline Cottrell barriers is responsible for
the substructure. Deformation substructures like those generated at

»••"•*

t*,,.<.><••% a

1.0 Mm

i Dislocation
C, forest lllO]

A, junction |-[011]
B, junction [111]

Burgers vector plane (ill)

Forest plane ~(110)

Figure 8. (a) Curved low-angle boundary in weld metal, and

(b) example of analyzed network.
 245

elevated temperatures, however, comprise dislocations at the subboundary

belonging to the matrix glide systems and are therefore distinguishable
from the 'growth' substructures seen here.

Table VI
# Subboundaries with a Given Forest Plane

Forest Plane

{111} {110} {112}

2 11 2

The basic 'knitting' mechanism for a hexagonal network described by

Lindroos and Miekk-oja is reproduced here. Figure 9(a) shows a (110)
forest plane with a system of parallel edge forest dislocations, C, with
b = NP = 2"[110] intersecting the (111) plane of a Thompson tetrahedron.
The pure screw glide-source dislocation, A, with b = MM = ytlOl] on the
(111) plane glides toward the forest under a shear stress, T . A and C
align by climb [that dislocations reacting at elevated temperatures al-
ways form attractive junctions through alignment by vacancy transport
along dislocation cores has been demonstrated-*^] and react as shown in
Figure 8(b):

MN + NP •* MP or f[101] + §[110] •+ f [Oil], A + C -*• B

The junction dislocation, B, has a Burgers vector, MP = ^-[011] lying in
the forest plane and not being in its normal glide plane, being held there
by elastic and/or chemical forces. Its line tension pulls dislocation A
onto the same plane. In identical fashion, junction dislocations are
formed at all forest dislocations and the hexagonal network of Figure 9(b)
established. Various other more complex mechanisms based on the same
principle of attractive junction formation by stress-induced climb have
been proposed, resulting in a variety of networks.'J!3>14>15 j t ^ s noted
that the curvature in the network is effected by increasing or decreasing
line segments with corresponding increases and decreases in their Burgers
vector edge components respectively. Also, the very nature of the knit-
ting mechanism and subboundary suggests a possible recovery mechanism
important to elevated temperature weld metal deformation.
As per this recovery model, glide source dislocations of opposite
sign from adjacent cells arriving at the boundary under stress are capable
of becoming part of the boundary as junction dislocations in the manner
described for network knitting.' Being of opposite signs, however, the
two dislocations knit networks in opposing fashions capable of annihilating
each other. The forest or subboundary thus acts as a kind of "collector"
or "sink" where dislocations meet and annihilate each other. The signif-
icant aspect of the model is the proposed path for vacancy transport —
along the dislocation core — effecting the stress-induced climb. The ac-
tivation energy for this core diffusion or vacancy pipe diffusion has been,
in some cases, found to be as low as one-half that for vacancy lattice dif-
 246

tUO|

(Ul)

(b)

Figure 9. The basic knitting mechanism on a symmetric (1102

boundary after^Lindroos and Miekk-oja; MN = 4 I
N*P = |[110] , PM = | [ ]

fusion. ° This relatively rapid pipe diffusion is capable of influencing

even the 'time-dependent' weld deformation behavior at high temperatures.
Weld metal recovery behavior is established through microhardness measure-
ment before and after the reanneal. The results are described below.

Microhardness

Figure 10 demonstrates the room temperature microhardness variation

across the pipe wall for each of the three zones of weld, HAZ and base
metal. The profile for the base metal zone indicates higher hardnesses
 247

Outside Inside
1 pipe surface pipe
180 surface
"" { 1.0 cm

170

160

, Base
150 N

140

170
^y
160 ^—--^
150
S HAZ . ^
•
140

170 V t

160

150 ~ Weld

140 _ — As-received DPH (kg/mm2)

Reannealed

Figure 10. Room temperature microhardness profiles across

GTA weldment pipe wall,
(a) Base metal, (b) HAZ, (c) Weld metal.

nearer the pipe surfaces decreasing toward the center before reannealing.
This is in agreement with the higher dislocation densities observed nearer
the surfaces introduced as a result of the pipe straightening operation.
Reannealing effected a reversal in the hardness profile with the center
being slightly harder than the surface regions. The weld zone does not
exhibit any hardness trends across the pipe wall. However, the general
shape of the profile persists on reannealing, and does not show the re-
versal seen with the base metal zone. The HAZ also shows no trends in
hardness across the wall and exhibits reannealing effects characteristic
of both weld and base metal, thereby representing a zone of transition
between the two, at least in this respect.
 248

Hardness has often been shown related to tensile properties of yield

and ultimate tensile strength. » • » The correlation, however, is
influenced by the tensile stress-strain behavior. Thus, strength compari-
sons made from direct hardness measurements necessarily assume a constant
stress-strain behavior. For this reason, Figure 10 does not seek to com-
pare the strengths of base, HAZ and weld metal, but rather to demonstrate
strength variations on a relatively microscopic scale within a given zone
of weldment assumedly exhibiting a constant stress-strain behavior, and
the effect of reannealing on it.
Since reannealing primarily effects a recovery of internal stresses,
much of which is induced by the pipe-straightening operation, with little
or no recrystallization (TEM shows little substructure coarsening or in-
creases in misorientation across subboundaries on reannealing) or grain
growth (optical microscopy shows little difference in grain size before
and after the reanneal), microhardness decrements at a specific location
on reannealing are used to represent the amount of recovery of the internal
stress.
Recovery occurs by the motion and subsequent annihilation of disloca-
tions under an internal stress, 'a^', the rate of decrease in the internal
stress may be written as:

-c-J (Ref. 21) (1)

where t = time, C = constant, v = climb velocity of dislocations, and
d = distance moved by dislocation for its elimination. The equation re-
duces to:

%t - - i • «>
21
for the proposed mechanisms of annealing by dislocation climb. 'A' is
a constant varying inversely as the temperature and directly as the vacancy
diffusion coefficient; 'm' is an inverse measure of the mean dislocation
travel, 'd 1 , required for its elimination, which is itself related to the
internal stress, 'c^ 1 .
Without detailing the derivations, possible 'm* values based on dis-
location theory are m = 2 , 3, 4. m = 3 , most generally for annealing by
climb of a randomly dispersed dislocation network. In this case, 'd' is
the network size, 'c^' the forest hardening inversely proportional to it
or direccly as the square root of the dislocation density.
m = 2, 4 is satisfied by the presence of piled-up groups at intercrys-
talline (grain boundaries) or intracrystalline (e.g., Cottrell in FCC crys-
tals) barriers respectively. The mechanism, although applicable only for
small values of the internal stress, is a useful basis for differentiating
between recovery mechanisms in base and weld metal. Following is a de-
scription of the method used tc estimate 'm' from the hardness data gen-
erated.
Integrating eq. (2) over the annealing time from t = 0 to t = t gives:
for m # 1, a}-™ = o±~m -At(l-m) (3a)
if o
and for m = 1, a.* = o± exp(-At) (3b)
 249

where subscripts 0, f refer to states at t = 0, t = t respectively. As

'm' approaches 1, for low internal hardening, equation (2) or (3b) ap-
proaches that proposed for a low-temperature anneal by cross-slip closely
satisfied by some high stacking fault energy metals. In this study the
contribution of this form of annealing to recovery is expected to be insig-
nificant. O j - and O-^ are represented by the room temperature hardness
Hf and H o after and before reannealing respectively, so that equation (3)
becomes:
for m * 1, 4 ~ m = H^~m - At(l-m) (4a)
and for m = 1, H f = H Q exp(-At) (4b)
Ten pairs of hardness values (H o , Hf ) from different locations across
the pipe wall were selected from each of the base and weld metal hardness
profiles of Figure 10. Treating 'At' as a constant, the data was then fit
to equations (4a) and (4b) by a least-squares method, minimizing with re-
spect to At for m = 1, 2, 3, 4. This yielded a set of equations with At,
m as known quantities in (4a) and (4b). The statistic, |v / c \ •"
which represents the error between observed hardness A = IZ^^Iti- - Kf J
values H? and values calculated from eq. (4), Hf , y" JQ_;———
was determined for each case (m = 1, 2, 3, 4) and K the value 'm' for which
A is a minimum is assumed to best represent the dominating recovery pro-
cess. Table VII has the optimum values of m and At determined for base
and weld metal zones of the weldment. All observed values of Hf lay with-
in the 95% confidence interval on the prediction equation.

Table VII
Recovery Parameters 3 for Base and Weld Metal

Zone m At £
3 2 1
Base 2 -0.4189 x 10 (kg/mm ) 8.12 kg/mm 2
Weld 4 -0.1975 x 10~ 7 (kg/mm 2 )" 3 1.45 kg/mm 2

a: m, At from eq. (4)

b: error between post-reannealing and calculated (from eq. (4))
hardness

The data of Table VII confirms the different recovery processes in

base and weld metal. While annealing in the base metal is suggested to
occur by climb of dislocations ahead of a piled-up group at a grain bound-
ary, the weld metal 'm' value indicates a considerable decrease in the
dislocation travel for annihilation with the existence of intracrystalline
barriers. Thus, the difference in recovery characteristics of base and
weld metal confirmed by the significantly different 'm' values obtained,
are attributable to the intra-crystalline low-angle dislocation boundaries
(subboundaries) described in detail earlier. Subboundaries are expected
to influence dynamic recovery during elevated temperature deformation in
identical fashion.
 250

Subboundaries and Deformation

From the results described thus far and a consideration of the pro-
posed mechanisms for subboundary contributions to recovery, * an explana-
tion for weld/base metal mechanical behavior differences based on the
existence of stable low-angle solidification dislocation boundaries in
the weld appears very attractive. Subboundaries, by virtue of their pre-
senting barriers to dislocation motion, are material strengtheners. A
reasonable approach to describing their contribution to the yield strength
takes the form of a modified Hall-Petch expression:

where ao is the Hall-Petch 'friction stress 1 , K, k, m are constants, A

is the substructure size, D the grain size, and a v s the yield strength.
At least below 0.6 T m (T m = melting temperature in °K) 'm' has been shown
to be 1. k, which represents the effectiveness of the substructure in
strengthening, is dependent on the subboundary energy or misorientation.
At room temperature, assuming a kA contribution of subboundaries to
strength and a nearly identical subboundary strengthening behavior as ob-
served with deformed 304 and 316 stainless steels,2J-26 f r o m

H - H o = AH = 50A" 1 * (Ref. 24) (6)

where A = subgrain intercept (ym)
AH = substructure strengthening measured in Knoop Hardness
Number (KHN) in kg/mm 2
one obtains, for an average substructure size of ~3 ym observed for the
weld here — AH ~ -i®. x 16.7 kg/mm . For a linear hardness-yield strength
dependency 1 7 ' 1 8 aiTd H o = 150 kg/mm 2 for a y s = 19.7 kg/mm 2 (Ref. 24) the
expected increase in room temperature a v s from base metal with no sub-
structure to weld metal with a 3 ]im substructure is approximately 2 kg/mm 2
(~21 MPa). Actually, tensile tests on base and weld metal indicate this
order of strengthening. The increase in weld metal yield strength as a
result of subboundaries cannot be totally neglected. At least part of
the weld metal residual stresses sometimes observed even after post-weld
stress-relief treatments are effected by the stable dislocation arrange-
ments (subboundaries) resulting from rapid solidification.
Differences in the plastic deformation behavior of weld and base metal,
more significant with increasing temperature, are also explained by the
weld substructure. Challenger and Moteff 2 7 have demonstrated the effec-
tiveness of a dislocation substructure in decreasing 'n'. Work on 304
stainless steel 2 ^ has indicated the establishing of a cellular structure
to approximately coincide with the transient (increasing slope — dlno/dlne
to decreasing one) of the Incr vs. lne curve with increasing strain during
hot (649°) testing. This suggests an effect of 'cells' in decreasing
dlna/dlne or 'n' in the Ludwik power function a = Ken. That the weld
metal usually exhibits a lower strain hardening ability (n) and ultimate
tensile strength is also in agreement with the proposed dynamic recovery
effects of the dislocation substructure.

* Although results of Ref. 27 are on 304 stainless steel, the subboundary

strengthening is expected to follow the same behavior for 316 stainless
steel.
 251

CONCLUSIONS

Following are the conclusions derived from this characterization

study:
(1) The weld metal dendritic structure appears as a dislocation sub-
structure of a size comparable to that of the secondary interdendritic
arm spacing with subboundaries having a misorientation of less than 1.0°.
(2) Within the weld metal zone the substructure size and subboundary mis-
orientation appeared to decrease with bead width or heat-input rate.
(3) The base or parent metal had a random dislocation structure (no sub-
boundaries) shown to effect strengthening in a forest hardening fashion.
(4) Reannealing resulted in an overall decrease in dislocation density
and therefore strength (hardness) with the persistence of weld metal sub-
structure, the effect of a slight increase in substructure size and sub-
boundary misorientation being felt.
(5) The residual weld metal strength after reannealing, manifesting itself
as a weld metal hardness greater than the base metal hardness, may be at-
tributed to the strengthening effect of stable, persistent subboundaries
in the weld, realizing a decrease in effective grain size.
(6) Subboundary analysis indicated a 'build-up' or 'polygonlzation' on
'forest' or 'growth' dislocations accommodating the mismatch between ad-
jacently solidifying dendrites.
(7) The proposed mechanism for this 'polygonization' or 'knitting' appears
valid based on the analysis conducted.
(8) Room temperature microhardness on the weldment before and after rean-
nealing established the significantly different recovery behaviors of weld
and base metal.
(9) A recovery model involving stress-induced climb of dislocations at
subboundaries is applicable to the weld metal exhibiting a stable dislo-
cation substructure.
(10) The apparent contribution to recovery from subboundaries influences
both time-independent and -dependent weld metal deformation behavior in
a manner consistent with actual observation.
It has been the object of this paper to characterize the weld rental
dislocation substructure, demonstrate its existence as a consequence of
rapid solidification and possibly extend its effects to other weld sys-
tems. However, in any extension of this work to other weld systems, it
should be recognized that the effects of grain size, chemistry, etc.
possibly supersede those of the dislocation substructure. Even so, the
importance of this aspect of weld metal microstructure to arriving at an
understanding of weld deformation behavior cannot be neglected. This
represents the first stage in a different but universally applicable ap-
proach to the understanding of weld metal deformation behavior based on
its apparently stable solidification substructure.

REFERENCES

1. Klueh, R. L. and Edmonds, D. P., "Mechanical Properties of 16-8-2

Large-Diameter Pipe Girth Welds," Oak Ridge National Laboratory
Report ORNL-5660 (September 1980).
 252

2. McEnerney, J. W. and Sikka, V. K., "Characterization of Time-Indepen-

dent Properties of Formed and Welded Pipe for Breeder Reactor Applica-
tions," Oak Ridge National Laboratory Report, ORNL/TM-6609 (April
1979).

3. Ham, R. K., "The Determination of Dislocation Densities in Thin Films,"

Phil. Mag., vol. 6, pp. 1183, 1184 (1961).

4. Hirsch, P. B., Howie, A., Nicholson, R. B., Pashley, D. W., and Whelan,
M. J., "Electron Microscopy Thin Crystals." Butterworths, London, pp.
181, 182 (1965).

5. Foulds, J. R., "Substructure of 316 Stainless Steel Pipe Weldments,"

Ph.D. Dissertation, University of Cincinnati, 1981. pp. 217-225.

6. Davies, G. J. and Garland, J. G., "Solidification Structure and Prop-

erties of Fusion Welds," International Metallurgical Reviews, Rev.
#196, vol. 20, pp. 83-106 (1975).

7. Lindroos, V. K. and Miekk-oja, H. M., "The Structure and Formation of

Dislocation Networks in Aluminum-Magnesium Alloys," Phil. Mag., vol.
16., pp. 593-610 (1967).

8. Rutter, J. W. and Chalmers, B., "A Prismatic Substructure Formed

During Solidification of Metals," Canad. J. Physics, vol. 31, pp. 15-
39 (1953).

9. Hulme, K. F., "Note on the Disorientation and Impurity Substructures

in Zinc Single Crystals," Acta Met., vol. 2, 1954, pp. 810-815.

10. Weertman, J. and Weertman, J. R., "Elementary Dislocation Theory,"

McMillan, New York, pp. 187-196 (1964).

11. Friedel, J., Boulanger C. and Crussard, C , "Elastic Modulus and In-
ternal Friction of Polygonized Aluminum land Aluminum Alloys]," Acta
Met., vol. 3, pp. 380-391 (1955).

12. Turunen, M. J, and Lindroos, V. K., "Model for Dislocation Climb by a

Pipe Diffusion Mechanism," Phil. Mag., vol. 29, pp. 701-708, (1974).

13. Miekk-oja, H. M., "A Mechanism for Knitting the Dislocation Network,"
Phil. Mag., vol. 13, pp. 367-370 (1966).

14. Lindroos, V. K. and Miekk-oja, H. M., "Knitting of Dislocation Net-

works by means of Stress-Induced Climb in an Aluminum-Magnesium Alloy,"
Phil. Mag., vol. 17, pp. 119-133 (1968).

15. Miekk-oja, H. M. and Lindroos, V. K., "The Formation of Dislocation

Networks," Surface Science, vol. 31, pp. 422-455 (1972).
 253

16. Hirsch, P. B. and Humphreys, F. J., Proc. Roy. Soc. (London) A318,
p. 45 (1970).

17. Tabor, D., "The Hardness and Strength of Metals," J. Inst. Metals,
vol. 79, pp. 1-18 (1951).

18. Oliver, B. R. and Bowers, J. E., "A Method for the Determination of
Yield Stress from Hardness Measurements in Aluminum and Copper,"
J. Inst. Metals, vol. 94, pp. 223-225 (1966).

19. Cahoon, J. R., Broughton, W. H. and Kutzak, A. R., "The Determination

of Yield Strength from Hardness Measurements," Met. Trans., vol. 2A(7),
pp. 1979-1983 (1971).

20. Cahoon, J. R., "An Improved Equation Relating Hardness to Ultimate

Strength," Met. Trans., Vol. 3A(11), p. 3040 (1972).

21. Friedel, J., "Dislocations," Addison-Wesley, Massachusetts, pp. 277-

279, (1974).

22. Kosik, 0. Abson, D. J., and Jonas, J. J., "Strengthening effect of

hot-work subgrains at room temperature," J. Iron & Steel Inst., Vol.
209(8), pp. 624-629 (1971).

23. Challenger, K. D. and Moteff, J., "Quantitative characterization of

the substructure of AISI 316 Stainless Steel Resulting from Creep,"
Met. Trans., vol. 4A(3), pp. 749-756 (1973).

24. Foulds, J. R., Ermi, A. M. and Moteff, J., "Substructural Development

in the Hot Tensile Testing of Type AISI 304 Stainless Steel," Mat. Sci.
& Eng., Vol. 45, pp. 137-141 (1980).

25. Michel, D. J., Moteff, J. and Lovell, A. J., "Substructure of Type 316
Stainless Steel Deformed in Slow tension at temperatures between 21°C
and 816°C," Acta Met., Vol. 21, pp. 1269-1277 (1973).

26. Sikka, V. K., Nahm, H. and Moteff, J., "Some Aspects of Subboundary
and Mobile Dislocations during High Temperature Creep of AISI 316
and 304 Stainless Steels," Mat. Sci. & Eng., Vol. 20, pp. 55-62
(1975).

27. Challenger, K. D. and Moteff, J., "A Correlation Between Strain Hard-
ening Parameters and Dislocation Substructure in Austenitic Stainless
Steels," Scripta Met., Vol. 6, pp. 155-160 (1972).
 CREEP DEFORMATION AND CRACK GROWTH IN A LOW ALLOY STEEL
WELDED PRESSURE VESSEL CONTAINING DEFECTS

M.C. Coleman

Central Electricity Generating Board

Marchwood Engineering Laboratories
Marchwood, Southampton SO4 4ZB, UK.

ABSTRACT

A full size pressure vessel has been constructed to examine the

effects of welding residual stresses on the creep deformation and crack
growth occurring in heavy section low alloy steel weldments. The vessel
was based on jCr§Mo|V main steam pipe and contained four 2CrMo manual
metal arc welds, two in the as-welded condition and two stress relieved
for 3 h at 700 C. All the welds contained pre-existing defects
machined in the heat affected zones.
Testing was carried out in a pressure vessel testing facility at
two internal steam pressures, 250 and 350 bar, and a temperature of
565 C. Cracked and uncracked areas of the vessel were monitored
continuously using DC potential drop equipment and high temperature
capacitance gauges. During inspection periods, at ambient conditions,
the weldments were non-destructively examined using ultrasonics and an
acetate film replication technique. In addition, residual stress and
strain measurements were carried out.
Results are presented for the continuous creep deformation observed
in both the hoop and axial directions of the welds throughout the
11,400 h of testing, as well as the intermittent strain data obtained
during inspections. In addition, crack growth observations are
described based on the non-destructive examination results. The residual
stresses measured are also given for both the as-welded and stress
relieved weldments.
While the pressure vessel experiment is not complete, the results
obtained are discussed in terms of the effects of welding residual
stress on the hoop and axial deformations observed in the welds.
Similarly, the effects of residual stress on creep crack growth are
considered together with compositional and microstructural implications.

255
 256

INTRODUCTION

The design of welded components for use at high temperatures in

power generating steam plant is based on parent material properties
only. Safety factors are built in which allow for variation in
materials properties and to take account of such factors as thinning at
bends. However, no account is taken of property changes due to the
microstructural and compositional differences introduced by a weldment,
nor of the effects of welding residual stresses.
During the operation of high temperature components, the first
problems usually occur in the weldments. Invariably these problems
are the result of crack-like defects located either in the weld heat
affected zone (HAZ) or the weld metal. Defects may be present
immediately after fabrication as, for example, stress relief cracks
nucleated in susceptible microstructures,particularly if post weld
heat treatment is not properly controlled. Similarly, when the welded
component is in service, creep deformation driven by operational
stresses, as well as any remaining residual stresses, may cause further
crack nucleation and propagation, both in the HAZ and in the weld metal.
Irrespective of how they initiate, it is essential to establish if such
defects significantly affect the continued operation of the welded
component.
Within the CEGB, defect assessments have been based on a correlation
obtained from uniaxial test data, where the crack growth X^te, a, is
related to the linear elastic stress intensity factor, K. ' There is
service experience to suggest that this assessment procedure is too
pessimistic. Furthermore, it has been argued that it is fundamentally
incorrect to describe crack growth in the creep regime using a linear
elastic parameter. As a result, a number of alternative,models have
been proposed using stress ' , crack opening displacement and a strain
rate parameter, C* 7,8^ again based on small scale laboratory data.
However, it is not known whether any of these models provide an adequate
description of the behaviour of defects in large components under
complex stress conditions.
Various research programmes are currently being carried
out in the Pressure Vessel Testing Facility (PVTF) of the CEGB's
Marchwood Engineering Laboratories (MEL), in which deformation,
crack growth and failure are being studied in full size welded
components operating under creep conditions. One of these
programmes examines the effects of residual stress on deformation and
crack growth in weldments which contain pre-existing defects machined
in the HAZs. This paper reports the work carried out in constructing
the pressure vessel and presents the creep deformation and crack growth
results obtained during the first 11,400 h of testing. The
implications of these results are then discussed for residual stresses
in plant and defect assessment.
 257

EXPERIMENTAL PROCEDURE

Pressure Vessel Manufacture

The pressure vessel was constructed using a normalised and tempered

main steam pipe and 2CrMo weld metal whose compositions are
given in Table 1. A welding procedure was required to produce a
consistent defined HAZ microstructural distribution so that defect
depths and crack growth could be related to specific HAZ microstructures.
Also it was required that the HAZ/fusion boundary interface should be
macroscopically normal to the pipe surface, again to facilitate defect
machining and crack growth data analysis.

Table I. Parent pipe and weld metal composition, wt %

Element C Cr Mo V Mn Si S P As Sb Sn

Parent 0.11 0.40 0.50 0.24 0.48 0.20 0.031 0.030 0.02 <0.02 <0.005

Weld 0.05 2.25 0.90 <0.01 1.14 0.50 0.021 0.020 0.015 <0.001 0.004

Welding trials were carried out, based on HAZ modelling

techniques"> , with the objective of producing a high proportion of
coarse grained, creep brittle, microstructure in the HAZ. These showed
that a downhand buttering technique using 6 mm diameter manual metal arc
electrodes for the first layer followed by 2.5 ram and then 3.25 mm
diameter electrodes, to avoid burn through, Figure la, consistently
produced a HAZ containing ^ 85% coarse grained bainite.H This was made
up of a coarse grained region extending for about 13 mm from the outside
surface then five repeating regions through to the bore, where the
refined and coarse grained regions extended for 1 - 2 mm and about 8 mm,
respectively, corresponding to the overlapping of the buttered layer
weld beads, Figure la. The same welding procedure was used to butter
the machined ends of all the parent material sections in the pressure
vessel so that each interface contained nominally this same HAZ micro-
structural distribution. After buttering, the interface layers were
built up using 4 and 5 mm diameter electrodes, to produce the more
usual "J" preparation profile. After lightly machining this profile,
the weld was completed using predominently 4 mm electrodes in a
conventional downhand welding technique, as shown schematically in
Figure lb.
The complete pressure vessel consisted of four welds between three
pipe sections and two forged end caps. The weld separation was
selected such that elastic interaction between the welds would be
negligible. First two welds were completed in one half of the vessel,
which was then heac treated for 3 h at 700°C. The final two welds were
 258

left in the as-welded condition, a filler bar being inserted in the

pressure vessel to reduce the stored explosive energy during testing.
The main dimensions of the pressure vessel are given in Figure lc.

1.HAZ
b)
2. Buttered
layer HAZ defect
3.Backing
layers

60 wall
f
y
Buttered interface

Section X - X
0

_37Q_ 400 400 Dimensions mm

Fig.t a)Buttered interface b} Schematic weldment and c) Pressure vessel

dimensions

Residual stress measurements were made on the outer surface of the

weldments immediately before defects were machined in the HAZs. These
measurements were obtained using the hole drilling technique , which
resulted in holes, lj nn diameter by about l| mm deep, in the vessel.
Parent material, HAZ and weld metal locations were examined at different
positions around the circumference in order to obtain a representative
residual stress distribution for both the as-welded and stress releived
weldments.
A spark erosion technique was employed to produce the defects, using
3 mm thick copper electrodes shaped at the tip to an included angle of
60 . At least two electrodes were used to produce each defect to
compensate for electrode wear that occurred during machining- The
resulting defect tip radius was i> 0.1 mm.
The locations and dimensions of the HAZ pre-existing part circum-
ferential defects are given in Table II and their positions are shown
in Figure 1c. These were selected to cover a range of depths, a, and
surface lenghts 2JJ, and to provide a direct comparison between defect
behaviour in as-welded and stress relieved weldments. Furthermore,
 259

the absolute depths were chosen, taking account of the HAZ micro-
structural distribution previously described, such that the defect tips
would be located either in the first or second region of coarse grained
bainitic microstructure.

Table II. HAZ defect locations and dimensions

Weld HAZ Interface Depth Surface length

Condition a, mm 2% mm

C 18 180
D 16 160
Stress relieved
G 10 100
H 6 60

K 6 60
L 6 120
As-welded
0 10 100
P 16 160

Pressure Vessel Testing Procedure

13 14
The pressure vessel was installed in the PVTF at MEL ' , as
sbown in Figure 2, and the test history is given in Table III.
Testing was carried out at 565°C and 250 bar for the first 2000 h. No
crack growth was detected during this period so the pressure was
increased to 350 bar for the subsequent 9400 h in a deliberate attempt
to promote crack growth. Heating to the test temperature was achieved
using an air circulation bell furnace. At temperature the vessel was
pressurised incrementally with steam up to the test pressure, which in
all cases was achieved within a period of about one hour. A short time
was allowed to elapse after each pressure increment to ensure that
temperature gradients in the vessels were kept to a minimum.
The deformation occurring in the vessel was monitored continuously
using Planer high temperature capacitance gauges developed at the CEGB's
Central Electricity Research Laboratories. These function up to
'v 700 C and have an effective gauge length of "\> 20 mm. Consequently,
both hoop and axial strains were obtained from relatively small areas
of the vessels pipe material and two of the weld centre sections, as
indicated in Figure 2. In general, the gauges have a strain range_of
^ 1% and drift rates at temperature typically are less than 3 x 10
strain per hour. Creep crack growth from the defects was also monitored
continuously throughout the test periods. This was carried out by
the DC potential drop technique, the potential data being converted to
crack depth using a previously established calibration.*^
 260

+ •A

D a Q

Stress relieved * E Hoop and axial

strain gauges

J
K
-SJL +L
Creep pips
As-welded +M
+N

Fig. 2. General appearance of the pressure vessel installation and schematic

representation of the strain monitoring positions.
Table IIT. Pressure vessel test history

Test Period Temperature Pressure Time Total Time

bar hours hours

565 250 2,000 2 ,000

565 350 2,600 4,600
565 350 3,500 8,100
565 350 3,300 11 ,400

At the end of each test period, the vessel was returned to

ambient conditions for examination bv reducing first pessure then temp-
erature. Intermittent strain data were then obtained from micrometer
measurements across creep pips. The creep pips, consisting of
stellite domes welded to small cylinders compatible with the parent
material, were tungsten inert gas welded to the vessel surface, equi-
spaced, at different levels around the circumference of the vessel pipe
sections, either adjacent to the weldments or about the mid - length
 261

position in the pipe sections. This arrangement is shown schematically

in Figure 2. Hoop strains were then calculated based on measurements
made between diametrically opposite pips at the same level and involved
gauge lengths of ^ 400 mm. Axial strains were calculated from measure-
ments taken between adjacent pips at different levels and involved gauge
lengths of ^ 100 mm for the parent material and weldments. The micro-
meters could be*'read to within ± 0.005 mm which, on the above gauge
lengths, represents accuracies of ± 0.001% and ± 0.005% for the hoop
and axial strains, respectively.
To avoid any localised cracking or weld metal interaction effects,
the creep pips were not attached directly to the surface of the weld
metals but to the parent material ^ 5 mm outside the HAZ. Hence in the
hoop direction these intermittent strain values relate to deformation
recorded in the parent material immediately adjacent to the weld HAZ,
while in the axial direction, data include a combination of weld metal,
HAZ and pipe deformation.
During the periods at ambient temperature, the weldments and
defects were non-destructively examined using ultrasonic and magnetic
particle techniques, and the defect edge openings were measured using
slip gauges. In addition, acetate film replicas were taken for
metallographic examination from polished areas across the ends of the
defects and from defect free areas of the weldments. " Finally,
residual stress measurements were made across the as-welded and stress
relieved weldments.

RESULTS

Capacitance gauge strains

The capacitance gauges showed a linear relationship between the

hoop and axial strains during pressurisation, with the hoop strain
approximately four times greater than the axial strain. In addition,
the strains measured in the parent pipe and weld metals were the same
indicating that a uniform elastic stress distribution was established
during pressurisation.
The creep deformations in the parent pipe, stress relieved and as-
welded welds are given in Figures 3, 4 and 5. In the parent pipe, the
hoop strain rate decreased with time during testing at 250 bar,
indicative of a primary creep region. The creep rate increased as a
result of increasing the test pressure to 350 bar, but immediately
began decreasing and continued., to do so throughout the test, finally
reaching a rate of 6 x 10 h In the axial direction a small
amount of primary creep strain was observed, but after about 1000 h the
strain rate became zero and remained so throughout the test periods at
250 and 350 bar.
In the stress relieved weld, the hoop strains and strain rates
were almost identical to those observed in the parent pipe, with the
final strain rate being 6.7 x 10"^ h" 1 . In the axial direction however,
the stress relieved weld showed positive strain accumulation throughout
the test. The general shape of the strain against time plot was
 262

l<tU

120
Pressure
, 100 - 250 bar 350 bar ^ ^ ^ Hoop
o
K 80 -

'g 60 •

4—

40 -

20
Axial
n - i i i • •
2000 4000 6000 8000 10 000 12 000
Time, hours

Fig 3. The variation of hoop and axial strain with time for the
^ i parent pipe at two pressures and 565°C

140

120 Pressure Hoop

100 _ 250 bar I 350 bar

80

2000 4000 6000 8000 10 000 12 000

Time, hours

Fig.4. The variation of hoop and axial strain with time for the stress
relieved 2CrMo weld metal at two pressures and 565°C
 263

160

2000 4000 6000 8000 10 000 32000

Time,hours

Fig.5. The variation of hoop and axial strain with time for the as-welded
2CrMo weld metal at two pressures and 565° C

similar to that for the hoop direction, Figure 4, but the axial strains
were always about 30% less than the hoop strains.
In the as-welded weld metal, the hoop and axial strains were
virtually identical during the 2000 h at 250 bar, Figure 5, and the
total strain accumulated during this time was about twice that
accumulated in the hoop direction in the parent pipe and stress relieved
welds. During the period at 350 bar however, the hoop strain
accumulation, 100 x 10 , was very similar to that in the parent pipe
and stress relieved welds, 95 and 97 x 10~5 respectively, with the final
strain rate approaching 7 x 10~8 h~l. The axial accumulation was again
similar to that in the hoop direction during the first 2000 h at 350 bar,
but thereafter the strain rate decreased more rapidly in the axial
direction than in the hoop direction. At the end of testing the axial
rate was 6.3 x 1 0 ~ 8 h~l.

Creep Pip Strains

The hoop strain pattern observed in the vessel is shown in

Figure 6. After 2000 h of testing at 250 bar, the strains measured
adjacent to the as-welded welds, KL and OP, were similar and about three
times greater than those adjacent to the stress relieved welds, CD and GH.
 264

Measurement time
a 11 ^00 hours
8100
o 46OO
x 2000

AB CD EF GH I J KL MN OP CL
11 11 JJL
Stress relieved As-welded

Fig.6. The variation of average hoop strain, from creep pip measurements,
with time and position in the pressure vessel

This clearly indicated that the higher residual stresses in the as-
welded weldments had produced a faster rate of deformation. Testing
at 350 bar however, produced a rapid and significant change. The outer-
most welds, stress relieved CD and as-welded OP, reached similar total
strains, despite the earlier differences. In contrast, the innermost
welds, stress relieved GH and as-welded KL, continued to maintain the
strain differences established during the first test period.
These hoop strains are based on four diametral measurements at each
of the positions, A to R indicated in Figure 6. At the final
inspection the spread in measurements obtained at the outermost weld
positions, C and P, was about five times greater than that at the two
innermost welds, GH and KL, resulting in a strain spread of + 16 x 10
compared with 3 x 1O~ • Furthermore, the maximum measurements at the
outermost welds were always across the diameter on which the defects
were centred, see Figure 1, whereas no such trend was observed at the
innermost welds.
In addition to the above observations, it was established, as will
be detailed later, that the residual stresses dropped significantly
during the first 4000 h of testing and, furthermore, that no significant
crack growth occurred until after the 8100 h and 11400b. inspections.
Consequently, while the early effects of residual stress were clear
from the strains measured after 2000 h at 250 bar, the large strains
 265

measured at the outermost welds immediately after the first period of

testing at 350 bar are not considered to be the effect of either
residual stress or crack growth. Instead, the deformation at the
outermost welds is considered to be a result of the original deep pre-
existing defect geometry. Analysis of the deformation arising from
this complex defect/weldment interaction will not be considered further
in this paper. The more shallow defects present in the innermost welds
do not however appear to have affected the strain measurements. As a
consequence a valid comparison can be made between the stress relieved
and as-welded welds, GH and KL, respectively.
Hoop and axial strain data for the two innermost welds and the
adjoining parent pipe are presented in Figure 7. The hoop measure-
ments were very consistent, but considerable spread was observed in the
axial measurements, as indicated by the data bars in Figure 7. In the
hoop direction the parent pipe and stress relieved weldment behaved
similarly throughout the complete test programme. The as-welded weld-
ment contained five times more strain after the first test period at
250 bar, but then deformed at the same rate as the parent pipe and
stress relieved weldment. In the axial direction, the as-welded
weldment showed twice as much strain after the first 2000 h as the
stress relieved weldment, 200 x 10~ 5 compared with 100 x 10"^, but
subsequently the deformation rates were similar. In the parent pipe,

jCr^MoiV Parent Hoop Axial

2CrMo As-welded K-L
500 2Crlio Stress relieved G-H

400 Pressure Data spread

250 bar ', 350 bar __
1
300

1200

-100

Fig. 7. The variation of hoop and axial strain with time from creep pip
measurements for the parent pipe stress relieved and
as-welded welds at two pressures and 565°C.
 266

negative axial strains were measured initially, rather than the

expected slightly positive strains, but the strain rate tended
finally towards zero. This initial compressive behaviour is
considered to be a consequence of the high positive strains in the
adjacent welds causing a tipping rotation of the creep pips at the
parent pipe positions I and J, rather than a genuine feature of the
parent material behaviour.
The main difference between thr; creep pip and capacitance gauge
data was the absolute values of strain measured. In particular, the
creep pips showed axial strains greater than the hoop strians while the
capacitance gauges showed the reverse. This was considered to be a
result of the different locations of the measurements, in that the
creep pip data only reflect what is happening at the weld metal centre
by way of measurements either adjacent to the weld or spanning a range
of materials, including the weld metal, while the capacitance gauges
measure what is happening at the weld metal centre itself.

Crack Growth and Displacement

Continuous monitoring,using the DC potential drop technique,

produced no indication of crack growth from any of the pre-existing
defects during either the 250 bar or 350 bar periods of testing.
Intermittent monitoring using non-destructive technqiues also revealed
no extension of the defects during the 2000 h at 250 bar. However,
crack initiation and growth were detected during the periods of testing
at 350 bar.
Crack growth was detected from the 16 mm deep defect in the as-
welded HAZ during the inspection at 4600 h. Ultrasonic inspection
indicated that 3 mm growth had occurred through the wall, while magnetic
particle inspection showed 5 mm circumferential extension from one end
of the defect. The inspection at 8100 h revealed further growth to a
maximum depth of 23 mm with the 5 mm surface circumferential crack
extending to a length of about 10 mm. No further growth was detected
at the last inspection at 11,400 h.
Within the accuracy of the ultrasonic measuring technique,± 2 mm
at best, no significant growth was detected from any of the other pre-
existing defects in either the as-welded or stress relieved HAZs during
the three test periods up to 8100 h. During the final test period up
to 11,400 h however, crack growth was detected from all defects, as
given in Table IV. As can be seen, the range of growth measured in the
as-welded HAZs, 5 - 9 mm, was very similar to that in the stress relieved
HAZs, 3 - 1 0 mm.
The defect edge opening displacement measurements taken during the
inspection periods are given in Table V. These show no significant
change from the original defect width after 2000 h at 250 bar.
However, following the first period at 350 bar, a 50% increase in width
was measured at the two deepest, outermost, defects in the vessel,
10 - 15% increase for the intermediate depth defects, but no change
across the shallowest defects. More importantly however, the dis-
placements measured at 4600 h did not alter significantly throughout
 267

the following test periods.

Table IV. HAZ defect depths

Defect Depth mm
Weld HAZ Interface Original
Condition Depth mm Oh 2000 4600 8100 11400

C 18 18 18 18 18 25
Stress D 16 16 16 16 16 19
Relieved G 10 10 10 10 10 20
H 6 6 6 6 6 10

K 6 6 6 6 6 15
As-welded L 6 6 6 6 6 11
0 10 10 10 10 10 17
P 16 16 16 19 23 23

Table V. HAZ defect edge openings

Edge Openings mm
Weld HAZ Interface Original
Condition Depth mm Oh 2000 4600 8100 11400

C 18 3.03 3.06 4.58 4.65 4.65

Stress D 16 3.01 3.05 3.29 3.36 3.36
Relieved G 10 3.02 3.10 3.58 3.58 3.60
H 6 3.00 2.90 2.90 2.86 2.89

K 6 3.05 3.06 3.09 3.08 3.08

As-welded L 6 3.01 2.96 2.97 3.00 3.00
0 10 3.01 3.06 3.29 3.17 3.36
P 16 3.02 3.08 4.41 4.50 4.57

Acetate Film Replicas

Metallographic examination of the replicas taken from the as-welded

welds revealed that creep cracking had initiated from the 16 mm deep
defect between 2000 and 4600 h on test. This generated a circum-
ferential HAZ macro crack extending 5 mm from one end of the defect
with slight microcracking from the other end. After 8100 h the macro
 268

crack had extended to 10 mm meeting the weld fusion boundary. Micro-

cracking and cavitation were present ahead of the main crack tip,
orientated circumferentially in the HAZ and axial in the weld metal, as
shown in Figure 8.

X: . ->;»,!^--V.v ••-.-^.V -, •- ••.-..-V-1 '• rr. ~*~ :'• ->?V&t,- >• .

Acetate film micrograph x 65

Fig. 8. Creep crack growth extending from the end of the 16mm deep defect
around the circumference of the as-welded weldment.

Apart from isolated cavities extending from the end of the deepest
defect in the stress relieved weld at 11,400 h, no other significant
surface creep damage was observed at the pre-existing defects or within
the weldtnents.

Residual Stresses

The measurements made across the weldments prior to testing and

after the first three test periods are shown in Figure 9. These show
that high hoop stresses existed across the weldment into the parent pipe
in the as-welded condition. In the weld metal the axial stress was
compressive but in the HAZ and parent pipe high tensile stresses were
present. In the stress relieved weldments the hoop and axial stresses
were equi-biaxial in all but the weld metal. Also, apart from the
weld metal axial stress, all stresses were considerably lower than in
the as-welded weldment.
 269

Parent pipe Hoop • Axial a

HAZ
Weld tnefal

4000, „ - ^ 8000
^ Time.hours

Fig. 9. The change in residual stress with time across the outer
surface of the as-welded and stress relieved weldments.

The residual stresses measured after 2000 h at 250 bar were all
less than^those measured prior to testing. The most significant change
occurred in the as-welded weld, where the hoop and axial stresses were
similar to those measured in the stress relieved welds. At 4600 h
little change was measured in the weld metals, but the stresses in the
parent pipe were now the same for both the as-welded and stress
relieved cases. The final measurements at 8100 h showed slight
increases in stress in the stress relieved weldments, but, in general,
very low values of residual stress in the as-welded weldments.

DISCUSSION

The present testing programme is still continuing and it is

intended to generate further crack growth. However, several clear
trends are already apparent from the results obtained. These relate
specifically to the effects of residual stress on deformation and to
HAZ crack growth.
 270

Creep Deformation

It is well established that the steady state creep deformation

occurring in plain pipes is controlled by the von Mises equivalent
stress, a A , where

°*= A
The strain rates in a given direction are then obtained from an
equation of the form

v i (v
where A and n are obtained from the uniaxial creep properties of the
pipe and a,, a«, and a- are the steady state principal creep stresses
fromthe Bailey equations.
Using the above equations, with A and n values of 3.2 x lO"^ and
4, respectively, where the units of stress are MN m and strain rate
h~l,19 anc | a t a n internal pressure of 350 bar, the steady state strain
rates calculated are 4.9 x 10~° h~^ in the hoop direction and, as with
all plain pipes, zero in the axial direction. While the strain rate
measured in the hoop direction by the capacitance gauge still appears
to be decreasing, Figure 3, the minimum rate, measured over the last
1000 h on test, is 6 x 10~8 h~ . In the axial direction a steady zero
rate was measured throughout most of the test. These rates are in very
good agreement with the theoretical predictions and, since the pipe
section is behaving as a plain pipe, indicate that the weldment
separation was sufficient to prevent any interactions. The data from
the pipe section therefore provides a base line against which weld
behaviour can be assessed.
In the stress relieved weld metal, the hoop deformation that
occurred during the 250 bar and 350 bar periods of testing were almost
identical to those observed in the parent pipe, as can be seen in
Figures 3 and 4. This compatibility of hoop strains and strain rates
between the parent pipe and stress relieved weld metal is a direct
consequence of stress off-loading from the weaker weld metal to the
stronger pipe, as suggested by finite element stress analysis and
substantiated by other welded pressure vessel programmes.21 Further-
more, since there is no significant difference between the total strains
observed at any time in the stress relieved weld and parent pipe, the
stress redistribution must have occurred rapidly.
In the axial direction, the strains and strain rates in the stress
relieved weld and the parent pipe were different. The zero axial
strain rate in the pipe, resulting from a surface steady stnte hoop to
axial stress ratio, cr /a , of 2 to 1, is replaced in the weld with
substantial positive strains and a final strain rate, 6.1 x 10~8 h" 1 ,
 271

similar to that in the hoop direction. Just as the surface elastic

hoop stresses redistribute to give a steady state hoop stress
distribution, the surface elastic axial stresses also redistribute to
give a steady state axial stress distribution. In the axial orient-
ation however, redistribution of stress takes place through the wall
thickness and not by off-loading along the pipe. The final results of
these stress redistributions can be assessed by examining the hoop to
axial strain rate ratio in the weld metal. Here, as in previous
work^ljthe hoop to axial strain rate ratio is close to unity, indicating
that the redistributed steady state hoop and axial stresses on the
surface of the stress relieved weld are equal.
The as-welded weld shows two phases of behaviour. Consideration
of Figure 5 shows that during the 2000 h period at 250 bar the hoop
and axial deformations were the same. Comparing these data with
Figure 4 indicates that at 2000 h the hoop and axial st\-ains are two
and three times greater, respectively, than those in the stress
relieved welds. After 2000 h the hoop and axial strain rates began to
decrease and eventually approached values similar to those in the stress
relieved weld. Uniaxial tests have shown that there is no significant
difference between the creep properties of the as-welded and stress
relieved 2CrMo welds. " Hence the reason for the differences arising
in the behaviour of the pressure vessel welds must be the significantly
higher residual stresses that are generally present in the as-welded
weld, Figure 9.
While this suggestion appears qualitatively satisfactory, detailed
examination produces some anomalies. In the hoop direction, the
residual stress in the weld metal is 'v 150 MN m~^, clearly greater than
the 40 MN m~^ measured in the stress relieved weld ;_tal. In contrast
in the axial direction the residual stress in the as-welded weld metal
is compressive and lower than the tensile value measured in the stress
relieved weld metal. These measurements, however, only relate to the
surface of the weldrnent and it is known that the residual stress
distribution varies through the thickness, peaking at a distance about
10% of the wall thickness below the outer surface. ^ In addition to
this, microstructural variations that occur in a weld, as a result of
weld bead overlap and start-stops in (he welding procedure, can lead *"0
considerable scatter in the residual stresses measured, depending on the
precise location of the measuring hole.^3,24 Furthermore, detailed
examination of 2CrMo welds, similar to those present in the pressure
vessel, has shown that while general stress distributions are under-
standable, very large variations are found even for preselected regions
of the weld metal microstructure.^2
This scatter in residual stress data makes it extremely difficult
to quantify the effects of residial stress on deformation. What is
readily apparent however, is that the residual stresses, especially the
weld metal hoop stress, dropped markedly in the as-welded weld during
the first 2000 h on test, Figure 9, and by 4000 h were about equal to
those in the stress relieved weld, At the same time, from 2000 h
onwards the hoop deformation in the as-welded weld metal became similar
to that in the stress relieved weld. Additionally, after about 4000 h,
the axial strain rate in the as-welded weld metal dropped below that in
the hoop direction and at the end of testing matched the axial rate in
 272

the stress relieved welds. These observations are a further

qualitative illustration of the relationship between residual stress and
deformation, showing that the residual stress effects are short term and
that after about 4000 h the pressure stresses alone are mainly
responsible for deformation.
The important consideration for operating plant is whether the
enhanced deformation associated with the reduction of residual stresses,
short term though it may be, results in any microscopic or macroscopic
creep damage. If such damage does arise, then it may extend to
generate cracks that have a significant influence on the performance of
the weldment and on the integrity of the plant. This aspect can be
appraised by considering the creep crack growth data obtained from the
pre-existing defects in the presure vessel.

Creep Crack Growth

During the first 2000 h at 250 bar no creep damage or cracking was
detected in any of the pressure vessel weldments. For this reason and
since the residual stresses had decayed considerably during this period,
the pressure was increased to 250 bar in a deliberate attempt to
generate creep damage and crack growth.
The first signs of creep damage, Figure 8, occurred from the 16 mm
deep defect in the as-welded weld which grew in successive periods
reaching a maximum total depth of 23 mm at 8100 h, after which no
further growth was detected, Table IV. However, during the final test
period, crack growth was detected from all the other pre-existing defects
in the pressure vessel which had previously been dormant. This occurred
irrespective of whether they were in as-welded or stress releived HAZs
and the amount of growth was about the same magnitude in all locations.
These observations show that, while creep crack growth initiated
first at the deepest defect in an as-welded HAZ, in general there is no
clear evidence that the welding residual stresses have affected the
extent of creep crack growth so far observed from the pre-existing
defects.
The metallographic results obtained from examination of the acetate
film replicas showed some creep damage from the ends of the deepest
defect, but no other significant creep damage at any other location in
the weldments. Therefore, it appears that the weldment microstructures
at the surface do not, in general, contain creep damage due to either
the presence of the pre-existing defects or the residual stresses.
At this stage it is important to mention the possible effects of
composition. The numerous examples of cracking in low alloy ferritic
steel weldments, giving rise to circumferential HAZ^5 ancj transverse
weld metal^° cracking, either during heat treatment or early in the
service life of the component, are a clear illustration of the effects
of residual stress on the integrity of weldments. However, these
phonomena have also been closely related to composition. High levels
of the alloying element vanadium^' and impurity elements such as
antimony and tin^°>29 have been shown to increase the susceptibility of
CrMoV materials to circumferential HAZ cracking. Similar observations
 273

have been made with regard to the effects of the alloying elements
manganese and silicon^" and impurities such as phosphorus on
transverse weld metal cracking in 2CrMo weld metal. In the present
work, the jCr|Mo|V parent material contains 0.24% vanadium and is
within specification for trace element contents. The absence of any
microstructural creep damage, even with residual stresses present,is
therefore also consistent with a material composition which is not
susceptible to stress relief cracking. Whether any longer term effects
arise will be established by continued testing of the weldments.
One factor that may explain the creep cracking observed relates to
microstructure. Knowledge of the HAZ microstructural distribution
suggests that the shallowest pre-existing defects have grown only in
the coarse grained HAZ region adjacent to the top weld bead while the
deeper defects have grown to about the limit of the second coarse
grained HAZ region. In addition, the 16 mm deep defect in the as-
welded weld HAZ appears to have arrested at a depth equal to the limit
of this second coarse grained HAZ region. These observations are
consistent with the growth patterns observed in plant components^ ancj
uniaxial tests confirming that the refined microstructural regions of
CrMoV HAZs are more resistant to creep crack growth than the coarse
grained regions. It remains to be established by further testing if
the growth data from all defects will show this growth and arrest
pattern.
Finally, the mechanics of creep crack growth must be considered.
The inability of the DC potential drop technique to provide a continuous
record of crack depth, probably because of oxide bridging-^2 restricts
any appraisal to the intermittent measurements made during the
inspection periods. From the ultrasonic measurements of crack depth
it is clear that, except for the deepest defect in an as-welded HAZ,
growth occurred during the last period only. At this stage, therefore,
crack growth analysis would be premature. However, the data provides
a basis for preliminary comment on existing crack growth models.
Considering the stress intensity factor, K, approach first. The
standard CEGB assessment procedure-^ applied to the deepest defects in
the pressure vessel predicts that th^y would double in size in about
1000 h. This is more than an order of magnitude greater than the
observed growth rates and confirms that the K based approach is too
pessimistic. Furthermore, if creep crack growth were dependent on K,
which in turn is dependent only on the defect size and stress, then
growth should occur from the start of testing. The present data
however clearly show that an incubation period elapses before growth
occurs from any of the pre-existing defects. Finally, the K approach
would predict markedly different growth rates from the different
defect sizes and locations, while no such distinct differences have
arisen in the pressure vessel test.
Considering the edge opening displacements of the pre-existing
defects, Table V, two types of behaviour are apparent. The three
shallow defects, 6 mm deep, have retained their original width through-
out the test programme, including the last test period to 1J400 h when
crack growth occurred, Table IV. The five deepest defects showed a
step change in edge opening after 4600 h, coinciding with the test
pressure being increased but only in the as-welded, 16 mm deep defect
 274

did crack growth occur. The edge openings at these defects remained
constant through to the end of the test, but it was only during the
final test period that crack growth took place and by this time the as-
welded 16 mm deep defect seems to have become dormant. These observ-
ations show no obvious correlation between edge opening displacement
and creep crack growth, except perhaps that they appear to occur
independently of one another. It may be, however, that the crack tip
displacements are so small at this stage that the slip gauge technique
is not sufficiently sensitive to monitor them. While this may be so,
at present the data serve only to indicate the difficulties likely to
be encountered in monitoring displacements in real components,
particularly where brittle materials are concerned, but provides no
basis for supporting any existing relationships between displacement or
displacement rate and creep crack growth.

CONCLUDING REMARKS

The present pressure vessel work has reached an interim stage at

which certain observations regarding creep deformation and crack growth
are apparent.
1. Welding residual stresses significantly increase the hoop and axial
strains accumulated early in the life of as-welded weld metal compared
with that in either parent pipe or stress relieved weld metal.
2. The residual stresses present in as-welded weldments decay to values
equal to those in stress relieved welds in a relatively short time,
about 4000 h under the present test conditions.
3. Coincidental with the relaxation of residual stresses, the creep
strain rates observed in an as-welded weld metal progressively decrease
until they approach rates similar to those observed in the stress
relieved weld metal.
4. There is no clear evidence for this pressure vessel material that
the presence of welding residual stresses has affected the growth of creep
cracks from the pre-existing circumferential defects in the weld HAZ.
5. The creep crack growth data generated during the pressure vessel
test does not support any of the existing stress or strain based models
used to describe creep crack growth data obtained from small scale
uniaxial tests.

ACKNOWLDEGEMENT

This paper is published by permission of the Central Electricity

Generating Board.
 275

REFERENCES

1. L.H. Toft and D.E. Yeldham, "Weld performance in high pressure

steam generating plant in the Midlands Region CEGB".
International Conference on Welding Research related to Power
Plant, CEGB.London (1972).

2. M.J. Siverns and A.T. Price, "Crack growth undsr creep conditions".
Nature, 228, 760-761 (1970).

3. I. Mogford and A.T. Price, "Application of fracture mechanics to

predict weld performance". International Conference on Welding
Research related to Power Plant, CEGB, London (1972)

4. G.J. Neate and M.J. Siverns. "The application of fracture mechanics

to creep crack growth". International Conference on Creep and
Fatigue in Elevated Temperature Applications. Inst. Mech Engrs.,
Philadelphia (1973) (Sheffield, 1974).

5. R.D. Nicholson and C.L. Formby, "The validity of various fracture

mechanics methods at creep temperature". Int. J Fracture, 11,
595 - 604 (1975).

6. J.R. Haigh, "The growth of fatigue cracks at high temperature,

under predominantly elastic loading". Engng. Fracture Mech.,
271-284 (1975).

7. K.M. Nikbin, G.A. Webster and C.E. Turner, "Relevance of non-linear

fracture mechanics to creep cracking". ASTM STP 601, 47 - 62 (1976).

8. M.P. Harper and E.G. Ellison, "The use of the C* parameter in

predicting creep crack propogation rates". J. Strain Analysis,
12 (3) , 167-179 (1977).

9. P.J. Alberry and W.K.C. Jones, "Diagram for the prediction of weld
heat affected zone microstructure". Metals Tech., 4, 360-364 (1977),

10. P.J. Alberry, T. Rowley and D. Yapp, "Control of manual metal arc
weld quality by deposition sequence". International Conference on
Advances in Welding Processes. The Welding Institute, 105-116
(1978).

11. M.C. Coleman, "The structure of weldments and its relevence to

high temperature failure". The Fifth Bolton Landing Conference
on Weldments - Physical Metallurgy and Failure Phenomena, General
Electric Co., 409-420 (1978).

12. E.M. Beaney and E. Proctor, "A critical evaluation of the centre
hole drilling technique for the measurement of residual stresses".
Strain, 10, 7-14 (1974).
 276

13. N.F. Eaton and T. Rowley, "Experimental evaluation of creep

behaviour of welded vessels". International Institute of Welding
Colloquium, Toronto (1972).
14. M.C. Coleman, R.Fidler and J.A. Williams, "Crack growth monitoring
in pressure vessels under creep conditions". In Detection and
Measurement of Cracks. The Welding Institute, Cambridge,
40-44 (1976).
15. B.E. Noltingk, D.F.A. McLachlan, C.K.V. Owen and P.C. O'Neill,
"High stability capacitance strain gauges for use at elevated
temperatures", Proc. Inst. Elec. Eng., 119, 897 (1972).
16. W. Arnswald, R. Blum, B. Neubauer and K.E. Poulsen, "The use of
surface replicas for the examination of creep loaded power station
components". V.G.B. Kraftwerkstechnik, 59 (7), 581 (1979).
17. M.C. Coleman, J.D. Parker, D.J. Walters and J.A. Williams, "The
deformation behaviour of thick walled pipe at elevated temperatures".
Third International Conference on Mechanical Behaviour of
Materials, Cambridge, 2, 193-202 (1979).

18. R.W. Bailey, "The utilisation of creep test data in engineering

design". Proc. Inst. Mech. Engrs., 131, 131 (1935).
19. B.J. Cane and C.J. Middleton, Private communication.

20. D.J. Walters, "The stress analysis of cylindrical butt-welds under

creep conditions". CEGB Note No RD/B/N3716 (1976).

21. M.C, Coleman and J.D. Parker, "Deformation in 2CrMo -

pressure vessel weldments at elevated temperature". International
Conference on Creep and Fracture of Engineering Materials and
Components, Swansea, 629-643 (1981).
22. R. Fidler, Private communcation.

23. R. Fidler and K. Jerram, "Residual stresses in 2CrMo welds".

British Nuclear Energy Society Conference on Ferritic Steels for
Fast Reactor Steam Generators, London, Paper 77 (1977).

24. A.K. Hardy, "Circumferential distribution of residual stress in

a single bead MMA weld on a reheat pipe". CEGB Note R/M/N833 (1975)
25. M.C. Coleman, "Creep crack growth in a CrMoV throttle valve
pressure vessel". Fifth International Conference on Fracture,
Cannes, 3, 1235-1242 (1981).

26. D.A. Wolstenholme, "Intergranular embrittlement of 2CrMo manual

metal arc weld metals". The Institution of Metallurgists
Conference on Grain Boundaries, C7-C12 (1976).
 277

27. J.D. Murray, "Stress relief cracking in carbon and low alloy
steels". Brit. Weld. J., 14, 447-456 (1967).

28. B..L. King, "Some effects of composition and microstructure on the

high temperature ductility of CrMoV steels". CEGB Report
RD/L/R1945 (1976).

29. J. Myers and A.T. Price " Stress relief cracking studies of
0.5CrMoV and 2CrMo steels using HAZ simulation". Metals Tech.
4, 406-410 (1977).

30. D.J. Gooch and B.L. King "High temperature crack propogation in
2|Cr-lMo manual metal arc weld metals". The Fifth Bolton
Landing Conference on Weldments - Physical Metallurgy and Failure
Phenomena. General Electric Co., 393-408 (1978).

31. D.J. Gooch and B.L. King, "Creep crack growth in controlled
microstructure CrMoV heat affected zones". Weld J. 59, 10 -18s,
(1980)

32. B.L. Freeman and G..1. Neate, "Limitations to the DC potential drop
technique of crack length measurement at elevated temperature".
CEGB Report SSD/MID/R24/78, (1978).
 M?sii^i;-A&f^x0^i:

% 1
" ' i

.t
•'me?

*5<V'W
 IMPROVEMENT OF FUSION WELDING THROUGH MODELLING,
MEASUREMENT AND REAL-TIME CONTROL

Koichi Masubuchi, David E. Hardt, Henry M. Paynter,

William Unkel, Jose Converti, and Miriam Zacksenhouse

Massachusetts Institute of Technology

Cambridge, Massachusetts 02139

ABSTRACT

This paper discusses the motivation for and the initital results
of a research effort within the Laboratory for Manufacturing and Pro-
ductivity at MIT to introduce in-process sensing and control to weld-
ing operations. Over the years the cost of constructing critical
structures has increased steadily. A significant portion of this cost
has come from the inspection and repair of welds. One solution to this
problem is to reduce the occurrence of weld defects by developing com-
puter controlled, or "intelligent" welding machines. The intelligent
welding machine is differentiated from a mere mechanized or pre-programmed
welder in that it controls the quality of the weld directly rather than
simply maintaining the welding parameters within specified limits of the
values based upon experience or upon trial welds. This is done by in-
process measurement and real-time control of the weld conditions. Our
basic approach for developing such welding machines is to combine funda-
mental research into process physics with research directed at attempting
to control the process in real-time.

Initial efforts have concentrated on the fundamentals of static and

dynamic characteristics of the welding arc, in particular those aspects
that are important in determining the stability of the arc to disturbances
and the interaction between the arc and the molten pool. The strength of
the plasma jet formed by a gas tungsten arc was measured by a non-intrusive
technique and strength was shown to vary as the square of the arc current.
The strength of the jet was shown to determine the stiffness of the arc in
resisting external distrubances. Hydrodynamic phenomena were also shown
to be important in the response of the arc to rapid changes in current.

281
 282

Our initial measurement and control work has centered on closed loop
control of penetration for root pass welds where complete melting of the
of the root is required. The paper describes efforts to use measurements
of the mechanical impedance of the weld pool to define the weld pool cross
sectional shape, as well as experiments with a rudimentary controller
design for this purpose.

INTRODUCTION

It is reasonable to state that over 50% of all repair welds done in

the world today are unnecessary and some repair welds have even damaged
the structures to some extent, although we have no data to substantiate
this comment (1). We must find some ways to stop the trend of increased
inspection and repairs. There are basically two solutions to this problem:

(1) To develop new rationales for accepting defects which

are not harmful; and
(2) To improve the reliability of welding operations to
reduce the occurrence of weld defects.

Much effort has been conducted to evaluate the significance of defects

on the service behaviors of welded structures. It is hoped that these re-
sults will be used to revise codes and specifications to make them more
rational. There remains a basic difficulty in incorporating scientific
data in codes and specifications. How to find defects is a matter of
applied physics—radiography, ultrasonic waves, magnetism, etc. How to
evaluate effects of defects that have been detected on the service be-
haviors of welded structures is a matter of applied mechanics—brittle
fracture, fatigue, etc. However, what to do with structures containing
some d€:fects involves many technical, financial, and legal considerations.
A ]iitited study was made at MIT on analysis of decision-making for repairing
or not repairing defects in welded structures (5). The concept of utility
of a welded structure containing defects was used in the analysis. Efforts
similar to this study and other efforts on systems analysis of the reli-
ability of welded structures may provide some foundation for developing
rational standards for welded structures.

The other approach is to make welding operations more reliable than

they are now. One possible way is to develop welding machines which have
the capability of in-process sensing and control. These machines may be
called "intelligent" welding machines. One basic problem in welding today
is that most welding operations are done manually, and there is always
room for human errors. Although many "automatic" welding machines have
been developed so far, these machines use welding conditions which have
been determined in previous experiments. Therefore, they should be called
"pre-programmed" or "mechanized" machines. No machine yet has a complete
in-process sensing and control capability in real-time. However, with the
use of modern electronics and computer technologies it should be possible
to develop welding machines equipped with such a closed loop control system.
 283

Research Approach

There are various kinds of weld defects including cracks, slag in-
clusions, porosity, lack of fusion, insufficient penetration, undercut,
etc (2). They are caused by the following:

(1) Less than perfect joint fit-up, including mismatch;

(2) Improper welding operations;

(3) Improper selection of welding conditions; and

(4) Improper selection of materials.

The research described here is concerned primarily with the first two
factors. It is assumed that weld defects can still occur even when mate-
rials are reasonably good and welding conditions employed are basically
adequate. It is therefore assumed that the occurrence of defects can
be greatly reduced if proper controls are carried out while welding is
performed. The present research is aimed at developing welding machines
for girth welding of pipes in which welding conditions must be changed
during welding to compensate for in-process disturbances, such as thick-
ness changes, mismatch and jigging heat transfer variations.

The concept of closed loop weld control is rather new to welding.

Although some efforts are being made to develop welding machines equipped
with some control systems, there is no generic approach, to the best of
our knowledge, on automatic control of welding. Thus one of the major
efforts in our research is to develop such a basic strategy.

This control strategy, in bread terms, is to seek to de-couple,

and then separately regulate all of the determinants of a "good weld".
For example, this would be realized if weld bead width, penetration, cross-
sectional shape, and local temperature distrubution history could all be
independently controlled so as to maintain certain specified strength,
toughness and fatigue characteristics in the joint.

At the present time there is no consistent lexicon to describe arc

welding automation. While the label "automatic" has been attached to
many welding systems, in many cases these are as primitive as mechanized
welders, where the torch travel speed, the wire feed rate and power source
characteristics are pre-adjusted to fixed values based upon experience and/
or experiments for the particular weld being done. Feedback has been used
however, in such tasks as torch guiding and seam tracking in robotic torch
manipulation.

There are presently two major difficulties in the application of

advanced (or even rudimentary) closed loop weld control strategies:

(1) A lack of an adequate understanding o.f arc welding statics

and dynamics; and
 284

(2) A lack of techniques to measure the quality of the weld,

(in particular to measure penetration during welding).

Rational control strategies can only be developed after the above defi-
ciencies are corrected.

Levels of Control in Welding

The concept of closed loop weld control is illustrated by the block

diagram in figure 1. This shows in a schematic fashion the intelligent
welding machine comprising, the following control schemes:

(1) Inner loop (or welder) control

(2) Overall control (or weld) control

The inner loop control includes the sensors and controllers that main-
tain the welder parameters (such as current, wire feed rate, etc.)
within an acceptable tolerance of a specified value. This inner loop
control, by itself, would allow pre-programmed welding: welding where
previous experience of the operator is used to determine how the welder
parameters should vary.

The outer loop represents the intelligent welding concept. As with

any dynamic system proper control requires three essential elements:

(1) A model of the system to be controlled

(2) Measurement (directly or indirectly) of key output quantities

(3) A Controller to use 1) and 2) to meet certain control objectives.

In the case of fusion welding using TIG processes, the objectives of

control are many and they include weld pool size, HAZ size and weldment
thermal history. A complete real-time quality controller would have to
account for all these to insure reliable autonomous operations. Our
eventual goal in this research is to develop the tools to model, measure
and control such a multiple input-multiple output, highly coupled system.
However, at present we are concentrating on the singular objective of
real-time regulation of weld pool size.

To accomplish this control objective, we have pursued modelling of

the arc, to understand how it affects (or could be manipulated to affect)
weld geometry; measurement of weld geometry using pool mechanical im-
pedance; and control of the geometry using direct digital control tech-
niques. This paper presents the current state of these efforts in the
above order.
 285

DESIRED wan
CHARACTERISTICS

HEAT WORK
1 PIFCE
WELD ~ f WF! DFR 1 ^ WORK
CONTROLLER —O—1 I PIECE
i
1
INNER LOOP

OUTER LOOP

Figure 1. SCHEMATIC OF OVERALL CONTROL STRATEGY

FOR SMART WELDER.

BASIC ARC STUDIES - STATICS AND DYNAMICS

Basic studies of the static and dynamic characteristics of the

welding arc have been conducted as a first step in improving the
understanding of welding arc/weld puddle behavior. In the near term
these basic studies allow better control of inner loops and provide
sensors and diagnostic techniques as tools in the control studies. In
the long term, the payoff from the basic studies is to suggest and
support process modifications which decouple control parameters of the
complex process of welding; when effective, this decoupling can improve
the performance of a welding process by extending its range beyond
present limits possible with conventional systems. Further, in cases
where decoupling is not possible, the models developed in the studies
can be used by the controller to establish the Interrelationships
between the inputs and the output (weld quality) thereby allowing more
sophisticated control strategies to be used.

This section summarizes the work to characterize the welding arc

and concentrates on the electromagnetically driven fluid motion
associated with the arc (11). The fluid motion is important not only in
the behavior of the arc itself, but is also an important part of the
interaction of the arc with the puddle. Specifically the plasma jet
formed by the arc is important because it 1) affects the stability of
the arc to external disturbances; 2) is a strong factor in the
mechanical interaction of the arc and the puddle; and 3) affects the
distribution of heat input. The static character istics of the arc are
first discussed, starting with a review of the plasma jet phenomena,
followed by a description of a non-intrusive technique to measure the
strength of the jet and finally by a presentation of some of the
 286

experimental results. The transient response of the arc to rapid

changes in welding current have also been studied as a step toward
understanding the reasons for the success of the current pulsing
strategies empirically developed for GTAW. These studies, which have
lead to the development of models for arcs with current transients, are
summarized and again show the importance of fluid phenomena.

Plasma Jet Momentum

Plasma jets form in welding arcs because of the expansion of the

current path from the imposed constriction at the torch tip to a larger
area. The main (axial) welding current induces a magnetic field
oriented in the azimuthal direction as shown in Figure 2; the axial and
radial current components interace with this self-induced field to cause
motion of the fluid as indicated in Figure 2. The so-called pressure
force results from the interaction of the axial current and the
azimuthal magnetic field; the total force on the plate (or puddle) is
independent of the distribution of
U 2
current and is given as F = — I While this term
pressure 8it

alone can be used to get a rough estimate of the plasma jet

characteristics, there is no net axial force exerted on the fluid by
this effect. The interaction of the axial current with the azimuthal
magnetic field results in an axially directed body force as indicated in
Figure 2. Since the radial current is highest near the tip, i.e. where
the main current expansion occurs, the momentum is imparted to the fluid
in the tip region. The flux of momentum is converted to a force on the
plate, since the flow direction is altered by the plate. For a fixed,
flat plate the force is

2
u
F = -- I (2 An R 2 /Ri)
8TT

where R2 and R2 are the radii at the tip and after expansion,
respectively. As will be shown by the measurements the axial force is
typically 3 times as large as the force from the radial pressure
distribution.
 287

PRESSURE DISTRIBUTION AXIAL FORCE DISTRIBUTION

IN SPREADING ARC IN SPREADING ARC

R
2-

FORCE ON PLATE = MOMENTUM TO FLUID =

±- i n d W R , ) I 2

FIGURE 2. PLASMA JET PHENOMENA

The expected square law dependence is noteworthy because it

indicates that t<e force will increase faster than the heat input, which
increases roughly linearly with current. One effect of current pulsing
at frequencies faster than the thermal response time of the puddle,
would be to increase the force on the puddle at a particular value of
heat input; i.e.

F~ <I 2 >, while Q~ <I>.

This may be one reason for the improved weld quality achieved
with high current pulsing.

Experimental Apparatus

The experimental measurements of the plasma jet momentum were

conducted in a stationary torch setup shown schematically in Figure 3. A
conventional GTA torch was used, typically with argon shielding gas, to
create the arc plasma. For the arc studies the work piece was a
water-cooled copper plate; a water water-cooled steel work piece was
used when a molten puddle was desired. The rig is designed to allow
full optical access to the welding region, and to provide for
 288

HELIWLTZ.
COILS

ARGON FLOW MAGNETIC

FIELD
ELECTRODE

X WATER COOLED)
PROJECTION SYSTEM
AND SINGLE DIODE OR
CCPD LINEAR ARRAY
ELECTRODE DETECTORS

FIGURE 3. SCHEMATIC OF GTAW TORCH AND

ARC DEFLECTION APPARATUS.

easy modification and interfacing with diagnostic and control

instrumentation. The steady-stage plasma jet measurements used a charge
coupled photo diode (CCPD) linear array, manufactured by Reticon, to
measure the light emission intensity. The output of the CCPD was
recorded using an oscilloscope. Transient measurements were made using
a single photo-diode (EG&G model SGD-100A), which allowed fast response
measurements for single location in the arc. Other diagnostic devices
included a monochrometer and a voltage probe which swings through the
arc.

A unique current controller designed and constructed, to our

specifications, by Alexander Kusko, Inc. (6). This transistor based
device places the welding current under tight control and allows
variation of welding current over its full range (0-300 amps) with a
bandwidth greater than 20 kHz. In the basic studies, the device has
been used to impose, on the arc, current changes which are faster than
the arc response time; this allows us to determine the response of the
arc. The current from the device can be controlled directly by a
computer and in the long term this device will be used in the actual
smart welding device.

Momentum Flux Measurements

Previous techniques to determine the force on the puddle have

included direct measurements with scales and indirect measurements using
a traversing pressure tap; both techniques are cumbersome and prone to
high error. In our work, (7,8)the momentum of the plasma jet was
determined by measuring the deflection of the arc caused by the
application of an external magnetic field, as shown schematically in
Figure 4. A tungsten electrode with tip angle is positioned vertically
as shown in Figure 5. The CCPD array is placed a distance, h from the
tip and the radiation intensity profile determined from the diode array.
 289

If the deflection is small and the arc is assumed to be tangent to che

axis of the electrode at the tip, the flux of momentum, M o is determined
from the deflection of the center of the arc, d as
2
Mo = Ih B/d (1)

where B is the magnetic field strength and I is the welding current.

Although a measurement of the deflection at a single combination of
magnetic field strength and separation distance is sufficient to
determine the momentum flux at a value of arc current, in the present
measurements the momentum was determined from measurements over a range
of magnetic field and separation distance. Typically the variation in
values was less than 10%

ELECTRODE

FIGURE 4. GEOMETRY USED IN ARC DEFLECTION

MEASUREMENT OF PLASMA JET MOMENTUM

The variation of momentum flux with current for a torch with a tip
angle of 32° is shown in Figure 5 to follow the expected square law
dependence. For comparison the force from the radial pressure
distribution is shown and can be seen to be approximately a factor of 3
lower, at least for the tip angle of these experiments. The variation
of momentum flux with tip angle has also been measured and shows the
expected trend of increased momentum at sharper angles. The results for
angles of 32 to 60 degrees are consistent with a constant current
density from the cathode and indicate a current density of 100
amp/mm**2. Further details are given in Reference (7).
 290

... 4.0

PLASMA JET MOMENTUM

PRESSURE FORCE

20 40 60 80 100 120 140 160

I(Amps.)

FIGURE 5. VARIATION OF PLASMA JET MOMENTUM

WITH CURRENT FOR AN ARGON ARC.

Previous investigators have correlated the stiffness of the arc to

the arc current (9) without understanding the physics. Equation 1 shows
explicitly that the plasma jet momentum controls the ability of the arc
to resist deflection from external magnetic fields, such as those
developed from nonuniform current passage through the base material.
For a constant current arc the deflection of the arc will vary inversely
with the arc current; the deflection can be decreased by decreasing the
tip angle since this increases the momentum flux at a given current.
Current pulsing can also be used to increase the stiffness for a fixed
average current.

Transient Arc Characteristics

The response of voltage and radiation emission intensity from the

arc to a rapid increase and to a rapid decrease in arc current has also
been measured (7, 10). The transistor current controller was capable of
changing the current from 100 amps to 160 amps in approximately 15
micro-seconds; this allowed us to examine the arc response. The
expected result was that the emission intensity would first increase in
response to the excess Joule dissipation and then, as the fluid was
replaced by fluid which had transitted through the tip region after the
current change, the emission intensity would again change. Since the
jet velocity decreases with radius it was expected that the arrival time
of the second part of the response would increase with distance from the
centerline. At the centerline the propagation speed should be equal to
the centerline velocity which for an arc current of 100 amps should be
about 100 to 150 m/sec. In contrast, the response of the emission
intensity to a rapid increase in current showed that the delay time was
consistent with a speed of about 50 m/sec at current levels of 100 amps
 291

and that the delay was not a function of distance from the arc
centerline. The measured response was similar to the observations of
hydrodynamic starting jets. A simple model was formulated to calculate
the propagation speed of such a vortex and this model gave speeds close
to those measured. While additional measurements and analysis are
necessary, the results to date indicate the need to consider
hydrodynamic effects when modeling the steady and transient behavior of
welding arcs.

Application to Welding Control

The improved understanding of the arc behavior gained in the

initial work will allow us to obtain tighter control of the arc.
Specifically, the momentum of the plasma jet can be used as a direct
indication of the ability of the arc to resist disturbances. The models
developed can be used by inner loop arc controllers if this becomes
necessary. As discussed below the ability to modulate the plasma jet
force independent of the heat input is being used to advantage in the
puddle impedence measurements. The long term benefit of the studies,
however, requires that the precise nature of the interaction of the arc
with the weld pool be understood; the further basic efforts in this
project will address this issue.
 292

WELD GEOMETRY CONTROL

As stated earlier, three major determinants of weld quality are:

1) weld pool geometry, 2) metallurgical property of the weld metal, and
3) heat effect to the base metal. Although a complete in-process control
system would account for all three, our efforts to date have concentrated
on weld geometry since it is the primary determinant of integrity.

In the molten state, weld geometry can be defined by the maximum

width and penetration of the weld puddle at any instant, and it is these
dimensions that we wish to place under active control. By having closed
loop geometry control, the resulting welding device will produce welds
of a consistent geometry that can be varied to suit the welding conditions
and, most importantly, this consistency will be maintained despite varia-
tions in the welding environment such as thickness changes, fit up errors
and jigging changes. Any automated welding device must have these pro-
perties to produce useful welds successfully without supervision.

This is illustrated by the example of autogenous GTA root pass weld

on pipe where complete melting of the root is required. The major pro-
blems are incomplete melting or burn through. Active geometry control
applied here would eliminate these problems by commanding the weld device
to melt only enough material to completely fuse the joint without risking
a burn through, and then to maintain a specified backbead width.

Root Pass Geometry Measurement using Weld Pool Dynamics

The occurrence of full penetration of a weld puddle is marked by

two distinct physical changes: 1) a transition to puddle support forces
arising from surface tension only, and 2) a bulk movement of the puddle
(caused primarily by gravity) to a new equilibrium position. If either
of these changes can be sensed, the control of a full penetration can
be accomplished. We have been addressing the use of both of these pheno-
mena to detect full penetration, the first by looking for the appearance
of a distinct mechanical resonance in the weld puddle, and the second by
trying to measure, using arc length changes, the depression of a puddle
upon full penetration when welding horizontally.

The first technique is based on the observation, shown in Figure 6,

that when full penetration occurs the weight of the puddle is supported
entirely by surface tension; and as the puddle moves vertically, the angle
of these forces changes. This provides a position-dependent vertical
force or, in other words, a spring. The puddle mass combined with these
surface tension springs represents a simple mechanical oscillator with a
natural frequency given by k/m where k is the effective spring constant
and m is the puddle mass.

Based upon this model, a method for sensing full penetration can be
proposed that would involve exciting the puddle with a vertical force
(originating, for example, from the plasma jet momentum) and then looking
for this characteristic frequency of oscillation. Our current experiments
 293

involve performing "frequency response" measurements on fully penetrated

puddles to verify the existence of such oscillations.

PLASMA FORCE

all ali
PLASMA

\ ^ _ ^

pVg

FORCES ACTING ON POOL SPRING MASS MODEL

Figure 6. MODEL OF A COMPLETELY MELTED WELD BEAD.

The procedure for performing the frequency response (i.e., force

input, motion output) experiments on a weld pool was as follows. A
stationary weld pool was formed using the current controlled TIG apparatus
described earlier. The current profile was a D.C. component (usually of
60 amp magnitude) and a sinusoidal component in the range 1 to 100 Hz.
and with amplitudes up to 30 amp. This provided a sinusoidal flux of
momentum change and therefore a sinusoidally varying force input to the
pool. The output, which is the resulting vertical motion of the pool,
was measured by passing a well collimated 1 cm. wide beam of laser light
along the bottom surface of the weldment (see figure 7 ) . The light was
then focused on a photodiode; if any part of this beam is occluded, the
intensity of light received by the diode will decrease. Since the beam
was aimed so as to pass below the weldpool, any movement of the lower
surface of the pool below the level of the weldment will modulate the
output from the photodiode. Thus we can impose a sinusoidal force on
the pool and measure the resulting mechanical response.
 294

A series of experiments were performed to use this technique to look

for the expected "spring-mass" or second crder system response from the
pool. In the first test, the D.C. current was 60 amps, and the superimposed
sinusoid had a 30 amp. amplitude. Using a 0.1" thick steel specimen, input
frequencies of 2, 4, 14, and 20 Hz. were imposed at separate times. The
resulting motion of the pool showed that the output motion (which was
sinusoidal and of the same frequency as the input) increased in amplitude
until 14 Hz. ..and then rapidly decreased, thereby suggesting the existence
of a resonance near 14 Hz.

As a more complete exploration of this pool response a test using

white noise (in this case a signal containing all frequencies from 10 to
100 Hz.) superimposed on the D.C. current was then attempted. With such
an excitation the pool output, when decomposed into frequency components
using a discrete Fourier Transform should immediately show the character-
istics of a second order resonant system it such is present. Several
tests were performed using noise amplitudes of 10 amp. on the D.C. current
of 60 amp. The output frequency spectrum for the pool is shown in figure
8. As can be seen in the figure, an output peak occurs at about 14 Hz.,
and (more importantly) the output "rolls off" at a -40db./ decade slope
beyond 14 Hz. clearly indicating second order response for the pool
dynamic system. Thus the pool resonance concept appears well founded.

\rrr/\ j//^/j
light beam
"photo'-
diode

Figure 7. POOL DISPLACEMENT MEASUREMENT METHOD.

Present experiments are endeavoring to exploit this measurement

method to quantify pool size by correlating it with resonant frequency.
At the same time, methods for "decoding" the pool motion from arc voltage
changes rather than from the light blocking method are being pursued.
With the latter accomplished it should be possible to then close the con-
trol loop to modulate heat imput so as to maintain a particular pool reson-
ant frequency and therefore pool size despite weldment environment
 295

variations.

-60
100
frequency (Hz)
Figure 8. FREQUENCY RESPONSE OF PUDDLE

As a more immediate control signal we have been investigating the

use of the steady state puddle depression that results when full penetra-
tion occurs. By measuring this steady state puddle depression when weld-
ing horizontally, we have been able to effect a rudimentary closed loop
controller for maintaining a full penetration weld. The puddle depression
upon complete melting is measured by looking for an increase in arc voltage
(under conditions of constant current and constant electrode-workpiece
separation). This voltage increase is a result of an overall arc column
length increase caused by the puddle depression. Such a voltage change
is illustrated in figure 9 where a stationary puddle was formed using
a GTAW torch. As the puddle goes from partial to full penetration, the
arc voltage can be seen to increase rapidly by nearly 0.5 volts.
 296

r—Melt-through

Arc 12 .
Voltage
(volts)

10
Time

FIGURE 9. ARC VOLTAGE AT MELT-THROUGH FOR A

STATIONARY WELD POOL.

A penetration control system was devised to exploit this signal,

and is illustrated schematically in Figure 10. This "bang-bang"
controller will regulate to an average full penetration weld by
continuously decreasing or increasing the local heat flux according to
whether or not full penetration is detected. As shown in the figure,
the feedback sensed is the arc voltage, and when a certain threshold is
exceeded, full penetration is assumed. In response to full penetration,
the heat input is reduced until the voltage goes below the threshold.
At that time the heat input is again increased. By always seeking the
full penetration transition, the adaptive qualities of the controller
are realized since any change in the welding conditions, such as
thickness or heat conductivity,will be reflected in a change in the
point of full penetration. This change will be tracked by the
controller thereby maintaining full penetration despite the disturbance.

With this admittedly simple control scheme, the weld depth will be
forced to oscillate about full penetration, but the amplitude of the
oscillation may be minimized by proper selection of controller
parameters.

A preliminary experiment with this control scheme has been

performed and the results follow closely the predicted performance of
Figure 10. In this experiment the torch voltage was sensed by a
microcomputer, filtered to remove unwanted electrical noise, and then
comp ed to a threshold voltage (which was determined automatically,
right after torch start-up). The result of the comparison (i.e. above
or below threshold) caused the computer to command a torch velocity
increase or decrease (which was effectively a heat input decrease or
increase).
 297

HEAT INPUT
(Control Variable)

WELD POOL DEPTH

Complete Penetration

ARC VOLTAGE /-Threshold Level

(Measurement)
Time

FIGURE 10. ROOT PASS PENETRATION CONTROL SCHEME.

A typical result is shown In Figure 11, where the expected changes

in arc voltage are seen driving the torch velocity in an oscillatory
fashion. The resulting weldment showed periodic full to partial
penetration as expected. The oscillation is more severe than desired,
because after full penetration has been obtained, the effective heat
input must be significantly decreased before full penetration is lost.
Nevertheless, some degree of control of penetration has been achieved
without sensors mounted at the rear of the weldment.

In summary, we are exploring two methods of full penetration

sensing, and have demonstrated the use of the steady state puddle
depression measurement in a closed-loop penetration control experiment.
The results verify the utility of this measurement for control but also
point to the need for further refinement in both the measurement
technique (e.g. to provide anticipatory information about full
penetration) and in the control algorithms to be used with this
measurement. We expect these refinements to develop from both basic
studies of arc physics and arc-puddle interaction and from control
modeling and simulation efforts that are now in-progress.
 298

Transition Voltage

FIGURE 11. RESULTS OF FULL PENETRATION CONTROL EXPERIMENT.

REFERENCES

(1) J. ConvertI, et al., "Improvement of Reliability of Welding by

In-Process Sensing and Control (Development of Smart Welding
Machines for Girth Welding of Pipes)," First, Second and Third
Progress Reports for the Department of Energy from M.I.T. under
Control No. DE-ACO2-79ER10474.A0O0 (M.I.T. DSP #88547), November
1979, March 1980, and June 1981, respectively.

(2) K. Masubuchi, "New Directions in Welding Research and Development,"

a special lecture given at the Inter-national Conference on Welding
Research in the 1980's, October 27-29, 1980, Osaka, Japan.

(3) K. Masubuchi, Analysis of Welded Structures, Pergamon Press, 1980.

(4) J.C. Wormeli, "Alyeska - a welding milestone," Welding Journal,

Vol. 56 No. 9, 1977, pp. 31-38.

(5) K. Masubuchi and N. Yirioka, "A Systems Analysis of Significance of

Weld Defects," Significance of Defects in Welded Structures,
Proceedings of the Japan-U.S. Seminar, 1973, Tokyo, University of
Tokyo Press, 1974.
 299

(6) D. Galler, A. Kusko and J. Converti, "A Fast Response Transistor

Current Regulator for Welding Research," IEEE-IAS 1981 Annual
Meeting, Philadelphia, PA.

(7) J. Convert!, "Plasma-Jets in Arc Welding," Ph.D. Thesis, M.I.T.,

1981

(8) J. Convert! and W. Unkel, "Plasma Jet Momentum," 33rd Gaseous

Electronics Conference, Norman, OK., October 1980.

(9) I.A. Bachelis, "Calculation of Deflection of Welding Arc In a

Constant Transverse Magnetic Field," Welding Production, 10 (7),
1963.

(10) J. Converti and W. Unkel, "Hydrodynamic Transient in a Tip-Plane

Arc Configuration," 34th Gaseous Electronics Conference, Boston,
MA., October 1981.

(11) K. Masubuchi, D. flardt, H.M. Paynter and W. Unkel, "Improvement

of Reliability of Welding by In-Process Sensing and Control,"
ASM Conference, New Orleans, LA., November 1981.
 ROBOTICS IN WELDING

John R. Dwyer

Merrick Engineering, Inc.

3070 Sidco Drive
Nashville, Tennessee 37204

301
 302

Merrick Engineering, Inc. has been manufacturing complete welding

systems since 1967. The company has been involved with robots and micro-
processor-based welder controls for the last several years, and numbers
many of the Fortune 500 among their customers.
This paper will try not only to give an overview of robots in welding
in the U.S.A. and Japan, but also discuss the commitments that must be
made by management. There are several lessons to be learned here. One
is that robots will drastically change the structure of labor industries.
For tasks such as welding, painting and even many assembly operations,
reasonably sophisticated robots working two shifts a day can pay for
themselves in less than a year. In this environment, companies relying
on labor will be extremely vulnerable to competition from companies which
have switched much of their manufacturing process to robots.
Second, robots provide recession resistance. Anticipating high vol-
atility in their mature domestic markets and uncertainties in their export
markets, Japanese blue chip companies are trying to build operations that
will make money at anything over 70% capacity. They are finding that
robots, which can work many shifts, are the key to lowering break-even
points. And furthermore, dedicated machines require long production
runs while robots can be re-programmed. While competitors suffer from
operating losses and sleepless nights worrying about layoffs and union
resistance, robot-run plants can simply switch to one-shift operation.
Of course, the pay-back to the investment sets a limit.
Third, robots reduce barriers to entering into high production. One
of the most fascinating aspects of robotics in Japan is that many smaller
manufacturers are taking the lead in installing sophisticated robots.
At the end of 1979, more than half of the numerically controlled robots
in Japan were installed in plants with fewer than 1,000 workers. Sophis-
ticated robots can carry out complex machining, welding, assembly, and
other skilled operations with flawless and tireless accuracy. Unlike
traditional automation, which tended to replace simple manual workers,
robots can replace experienced and skilled workers.
Fourth, robots greatly increase manufacturing flexibility. With
robots, managers can choose plant sites in order to minimize logistics
and transportation costs, attract first-class management talent or do
almost anything that will give the company a strategic edge. Robots give
companies greater freedom to meet different specifications for a given
product line.
Fifth, and probably most important, is that companies can benefit
from robots only if they take great care in preparing for their intro-
duction. Management must make sure, for example, that it doesn't alienate
workers and unions. Even in Japan, worker resistance has stymied robot
introduction in cases where workers felt that robots were taking away the
pleasant and easy jobs, leaving dirty and unpleasant work for people.
Where robots have replaced people in hostile working environments, such
as welding, painting, cutting, and grinding, Japanese companies have
found that workers learn to live with robots and even come to appreciate
their value. The production floor must be ready for robotics.
Several recommendations on how to ensure minimum worker resistance
are:
• Never surprise the labor force with a robot on the floor on
 303

Monday morning. Get people involved early.

• Let your personnel people get involved in what to call the new
automation. Do you want to call it a robot or an automatic mani-
pulator?
• Get a public relations program going for the robot introduction.
Create a positive first impression. Publicize that the increased
efficiency will protect jobs by keeping the company competitive.
• Avoid layoffs. Reduce the number of employees through attrition.
Upgrade workers wherever possible.
The introduction of industrial robots and the competition to develop
more sophisticated robots have expanded to a global scale. In the West,
the motto of the day is "Let's pass robot leader Japan," and efforts are
underway to reorganize production systems through the use of large numbers
of robots. And the Communist Bloc which is suffering from a slowdown in
productivity is showing more enthusiasm than the capitalist countries in
negotiating for robot purchases. (See Tables I and II.)
In Japan, a major reason for acquiring robots is the lack of skilled
workers. Their demand for robots is high because Japan, unlike other
advanced countries, has a chronic labor shortage as a result of a virtual
ban on the entry of foreign workers. According to a survey conducted by
the Japanese Ministry of Labor, as of June, 1980, there was a nationwide
shortage of 34,600 welders and 21,600 painters, and most of the shortage
of skilled labor was concentrated in companies with fewer than 100
employees. This same area is becoming an ever-increasing problem in the
United States.
Estimates published in Business Week predict that U. S. robot
production by existing makers alone will reach only 23,000 units by 1990;
however, if computer makers enter the field, this figure could rise as
high as 200,000 units. In this connection, the attitudes of the world's
largest computer maker, IBM, and largest semi-conductor maker, Texas
Instruments, bear watching, and IBM has already announced a new family
of robots.
In 1980, American industry spent about ninety million dollars on
robots — a 50% increase from the 1979 level. Purchases soared again in
1981, to an estimated one-hundred-thirty million dollars. By 1990, experts
predict, the United States will represent a two billion dollar a year
market for robotics.
In Tennessee, we will shortly see the completion of the Nissan Plant
which will have approximately 270 robots of which some 54 will be used
for MIG welding. An equal number or more will be used for resistance
welding, the balance for spray painting, and pick-and-place.
Today, interest in robots is such that people may disdain talk about
last fall's World Series (after all, it was just the Yankees and Dodgers
again) to hear more about the intriguing devices. That is the "E. F.
Hutton effect," says James A. Baker of General Electric. "People whose
eyes glaze over at announcements of the most far-reaching electronic
breakthroughs or business developments suddenly are gripped with feverish
interest when someone mentions the word 'robot,'" observes Mr. Baker.
One thing they need to learn is that there is more to robotics than
buying a robot. General Electric1s Mr. Baker, who has more experience
as a robot user than as a robotmaker, advises, "Avoid the impulse to
 304

TABLE I : INTERNATIONAL RO3OT POPULATION

Country Type A Type B Type C Type D TypeE Total

Japan 6.899 — 7.347 53 189 67.435

USA 400 2.000 1.500 200 — 4.100
West Germany 290 830 200 100 10.000 11.420
France 120 500 2.000 6.000 30.000 38 620
USSR - — — _ — 3.000
Switzerland 10 40 — — 8.000 8.050
Sweden 250 150 250 50 100 800
Norway 20 50 120 20 50 260
Czechoslovakia 150 50 100 30 200 530
Giear Brilain — — - - — 371
Poland 60 115 15 50 120 360
Denmark 11 25 30 0 110 176
Finland 35 16 43 22 51 167
Belgium 22 20 0 0 82 124
Netherlands 48 3 0 0 30 81
Yugoslavia 2 3 .5 0 15 25

TOTAL 1.418 10.701 4.293 13.819 101.947 135.519

Type A Programmable, servoconlrolled. continuous path

Type B Programmable, serviiconlrolled. point t o p o i n l
Type C Programmable, nonservo robots lor general purpose use
Type D Programmable, nonsurvo robots (or iliecasling and molding machines
Type E Mechanical transfer de nr.es (pick a id placel

TABLE I I : FORECAST OF ROBOT GROWTH

Country 1985 1990

Japan 16,000 29,000

USA 7,715 31.350
West Germany 5.000 12.000
Switzerland 600 5,000
Sweden 2.300 5.000
Norway 1,000 2.000
Great Britain 3,000 21.500
Poland 200 1.200-1.500
Denmark 110 250
Finland 950 3.000
Belgium 150-200 —
Yugoslavia 100-150 300
 305

install a few token robots in the factory and expect them to guard the
doors of your business against the competitive inroads of the Japanese."
Says James S. Albus, head of the robotics research laboratory at the
(US) National Bureau of Standards, "The human race is now poised on the
brink of a new industrial revolution that will at least equal, if not far
exceed, the first Industrial Revolution in its impact on mankind."
Immune to government and union regulations on heat, fumes, noise, radia-
tion, and other safety hazards, robots can tirelessly work around the
clock, three shifts a day with the lights and heat turned down or off.
Upper management sees the robot "as a way of magically substituting
dependable machines for difficult-to-manage personnel," he says. So it
issues orders to manufacturing to get the robots on board. And operations
managers, who may not even know what a robot looks like, respond. They
are beginning their greatest learning experience in recent decades.
Nonusers of industrial robots do not expect labor to accept the
introduction of robots into plants, yet users indicate that labor has ac-
cepted the introduction of robots very well. Users know from experience
that their workers are able to operate robots, because robots require
minimal attention after start-up; non-users do not think their workers
are skilled enough to operate robots.
It is not surprising that non-users believe that there would be a
significantly lower economic benefit from using robots than users do.
The Robot Institute of America, an industrial trade group, offers
a definition of a robot: "A reprogrammable, multifunctional manipulator
designed to move material, parts, tools, or specialized devices, through
variable programmed motions for the performance of a variety of tasks."
U. S. demand for robots began growing in 1977, when industry, es-
pecially the auto industry, increased investments to rationalize produc-
tion operations.
The auto industry was the first to install robots on the assembly
line, but because the workers did not keep them properly maintained, there
were even cases where a separate assembly line staffed by workers was set
up in addition to the robotized line.
Three types of industrial robots may be termed the "pillars" of the
industry at present: spot welders, arc welders, and painting robots.
Ninety percent of the spot welders are for auto industry use. In contrast,
over half of the arc welders and painting robots, sales of which have
rapidly increased over the last year or two, are purchased by smaller
companies.
Those who have become most perplexed and worried over the appearance
of these superior robots are skilled welders and painters. For example,
after the introduction of an arc welding robot at a certain medium-sized
company, the professional welders, who had not been very diligent until
then, changed their attitude and really began to work. They undoubtedly
felt that their jobs were being threatened, for the robots were getting
more work done. In one shop, it was purely accidental: the robot sales-
man demonstrating his product did not know that the weld was supposed
to be three weld passes and he programmed his robot for a single pass;
it did the job. Human welders followed suit.
The right robot must be chosen. Many companies are disappointed be-
cause they install robots that are "overqualified" for their tasks.
 306

On today's production floors, there are still many work stations requiring
rather simple motions. There's no need for a fancy, multi-directional
robot to perform, say, simple loading and unloading operations.
At present, robots cannot do all kinds of work. For example, arrange-
ment of the materials with which the robot is to work, retrieval of the
processed parts, and removal and cleaning up of the remaining scrap mater-
ial are all tasks still performed by humans. The time that a person spends
operating a single machine has been reduced so that while previously it
was not unusual for a person to be responsible for only one machine, now
it is common for a single worker to be in charge of four or five.
Also, the robot is not flawless. If it has a problem, it will
either correct the problem itself, summon its human supervisor with a
bell or whistle, or just quit. Robots are like computers - garbage in -
garbage out. Where parts have poor fit-up, mismatch, and are inaccurate,
proper cleanliness and care have not been taken, the final results will
not be good. Robots are not a cure-all. The same argument can be made1
for welding tooling. It must be made precise. The robot will come back
to the same point within its accuracy rating and, if the part is not held
correctly, it may weld off the seam.
Because the robots do not know when a torch is full of spatter or
the gas flow over the weld is impaired, several types of self-cleaning
arrangements have been devised. One means is by actuating a cleaning
mechanism inside a special MIG torch followed by a blast of gas or air
to blow away the spatter. Another is for the robot to present the torch
to a rotating brush or a blast of cleaner. Some of these seem to work
better than others.
We have seen robots used for plasma welding of catalytic converters
in Japan, and they are doing plasma cutting as well.
Robots weld automotive parts on Chrysler's K-cars assembly line and
spray-paint file drawers for Steelcase Company; they cast ceramic molds
for motor parts for the Evinrude Outboard Motor Company and manufacture
portions of Xerox machines.
They must be cost-justified, correctly used, and properly maintained.
They never do more than they're told to. And no robot has ever been
known to contribute an innovative idea.
A common user error is the tendency to over-specify the robot's control
system. A typical robot task (other than welding) requires only 30 steps
or less. There is no need to specify a robot and control system capable
of spot welding any car body in the world, especially if they are destined
for pick-and-place duty. Such overkill backfires if the equipment breaks
down and no one knows how to fix it.
In the United States, a robot is generally applied to the most
difficult job in the shop. This requires top-of-the-line equipment. It
also imparts a false sense of security to those who think that if the job
should be phased out, the robot can be reassigned to other tasks. Prob-
lems inherent in dealing with this sophistication on a first-time basis
cause frustration on the plant floor which may lead to the suspension
of the robot experiment.
In contrast, the Japanese attempt to apply the simplest robot avail-
able to the simplest job. This way the user builds experience and confi-
dence in the equipment, while giving the work force ample opportunity to
 307

accept the automation.

The welding machine must be carefully interfaced to the robot. An
example of one of these interfaces is shown here (Figure 1). Many of the
welding manufacturers and robot manufacturers have already developed
these interfaces for MIG welding.
In addition, because of multi-programs required by certain welding
tasks, they have developed multi-program wire feed systems, such as the
Miller System Nine (Figure 2) which will accept, store, and execute nine
different and complete weld programs. Merrick Engineering has developed
a microcomputer-based (Figure 3) welding process control for TIG welding
with from one to fourteen time intervals, a 12" CRT monitor for data dis-
play, two channels, eight full programs which may be recorded/loaded on
a UV Prom, and it is shown by itself and in a Merrick TXR-100 (Figure 4)
unit, available in a stand-alone or in a power supply. However, this
and other units not totally integrated require independent movement
memory (robot) and welding parameter memory. OTC robots are integrated
and most of the newer welding robots, such as Hitachi and General Electric
among them, also are fully integrated.
There are developments going on now for non-contact seam followers --
one in this country, and one in Germany, which are beyond the prototype
stage and are in working environments.
Attention must be paid by shop management to keep the worker from
being hurt by the robot. Japan has already had its first death as a result
of a worker bypassing the safety interlocks and entering the envelope
where the robot was working. All manufacturers are designing the necessary
safety features to disable the robot before working on it or entering
its work area.
It is suggested that our present hourly wage for a worker in the
automotive or farm equipment industry is $15 to $20 per hour. The average
robot will work at a cost of $4.15 to $5.00 per hour, and multiple shifts
will decrease this amount even more.
There are 156 manufacturers of robots world-wide with more coming
onstream every day.
Robots come in three basic types of technology:
• Low - non-servo
• Medium - some servo with outside computer
• High - completely computer controlled and servo controlled
The price of a low technology robot can be from $8,000 to $16,000.
This has a high degree of flexibility and runs high batches with a low
run of parts for each batch.
Medium technology robots are $20,000 to $60,000 and approximately
4,000 robots in the United States fall into this category or about 75%
of all robots purchased.
The point to point robots are not suitable for arc welding where con-
tinuous path is required. One can imagine this if welding a wheel-rim.
Newer robots, dubbed servo robots, make the simple, common versions
look stupid by comparison. They have memory and control systems, most
often in the form of microcomputers that allow the robots to be program-
med to carry out a variety of work routines and reprogrammed when
necessary.
Moreover, work is underway to create still smarter robots that can
 308

Motor Control Cord

From Robot Conirolltr

115V/Contactai
Control Co.d

Diagram Courtesy of Miller Electric

Typical Connection Between a Robot

and a MIG Welding System

FIGURE 1
309
310

CO
UJ

o
 311

FIGURE 4
 312

"see" by deciphering what appears on a television camera, to sense what

their "hands" pick up, and to answer commands from the human voice, thus
enabling robots to make "decisions" from a number of programmed options.
One robot expert, for example, predicts that in fifteen years there
will be no human coal miners.
In 1982, five percent of all assembly systems will use robotic tech-
nology. By 1985, twenty percent of the labor in the final assembly of
autos will be replaced by automation. In the same year, "scene analysis"
will provide enough feedback for robots to select parts scrambled in a
bin. By 1987, some fifteen percent of all assembly systems will use robot
technology. By 1989, fifty percent of the labor in small component as-
sembly will be replaced by automation. And by 1990, the development of
sensory techniques will enable robots to approximate human capability
in assembly.
Predicts Edward Fredkin, professor of computer science at MIT, "We
are creating what is going to be an immense new industry, perhaps as big
as the auto industry."
Technological advances in robotics are moving at rates paralleling
those of hand-held calculators — faster than builders can deliver
hardware to the marketplace.
Paired with robots, machine vision is more of a technical curiosity
blown out of proportion by the trade press than commercially common-
place. It is still too expensive for most applications.
We are, like it or not, victims -- or victors — of our own techno-
logical ingenuity.
Merrick sells robots but does not manufacture them. We do the tool-
ing necessary to hold the part, sometimes cool it, and often must design
almost as intricate a control, which presents the parts to the robot, as
the robot itself.
We are on the threshhold of a new industrial revolution, and
Merrick looks forward to being a part of this revolution.
 LASER WELDING OF STRUCTURAL ALLOYS

E. A. Metzbower, D. W. Moon and F. W. Fraser

Material Science and Technology Division

Naval Research Laboratory
Washington, DC 20375

ABSTRACT

Laser beam weldments of structural alloys have been

fabricated in a single pass, both as autogeneous butt welds
and narrow gap Vee welds with filler metal additions. The
autogeneous welds were made in thicknesses up to 16mm (0.625
in), whereas the welds with filler metal additions were made
in thicknesses up to 19mm (0.75 in). The aluminum alloy 5456
of 12mm (0.5 in) plate thickness was normally welded at 7
kilowatts and 28mm per second (65 inches per minute).
Titanium alloys of 12mm (0.5 in) pjate thickness were welded
at 11 kilowatts and 15mm per second (35 inches per minute).
Steels of 12mm (0.5 in) plate thickness were welded at power
levels from 10 to 14 kilowatts and speeds from 13 to 19mm per
second (30 to 45 inches per minute.) A description of the
laser power/welding speed and its variation on plasma
suppression is given. For all of the alloys except aluminum
the mechanical properties corresponded to the base plate
properties. In the case of aluminum the weldments frac-
tured in the fusion zone with a degradation of the plate
properties. The requirements of a bend test were satisfied
by the weldments. Hardness traverses were made across the
base plate, heat affected and fusion zones. Microstructures
were determined in the three zones and correlated with the
mechanical properties and hardnesses. Fusion zone purifi-
cation was observed in certain alloys.
Fracture toughness testing was carried out on the
weldments. The aluminum and titanium alloys satisfied the
present toughness criteria. The HY-130 steel weldments also
satisfied these criteria but the HY-80 laser beam weldments

313
 314

showed evidence of solidification cracking in the autogeneous

welds. Results of recent experimental work of heat treating
these weldments will be discussed. The A36 laser beam
weldments had Charpy V-notch energies similar to those found
in the base plate. When these weldments were stress relieved
an enhancement in the CVN energy was found at -40cC (-40°F).
An analysis is given for this enhancement.

INTRODUCTION

Life cycle cost - the total cost of a system through its manufac-
turing stage, operational lifetime, until its final abandonment as a
viable entity - is a major concern and driving force for most complex
structures. A major component of the life cycle cost of large struc-
tures is the number of man-hours required to fabricate and maintain it.
Any and all improvements in fabrication procedures that reduce cost,
while maintaining structural integrity and reliability, are warmly
welcomed by the fabrication community. In order to reduce the number
of man-hours required for fabrication, automated welding processes and
the development of robotics are currently being studied both in this
country and abroad. Automated welding processes offer the opportunity
to decrease the welding time simply by increasing the welding speed.
However as the process is automated, the integrity and reliability of
the assembly must be maintained.
Laser beam welding is an automated welding process that offers
the potential for cost savings by means of increased welding speeds.
Collateral savings can also be accrued by decreased preheat temperature
and time, by decreased straightening resulting from less distortion,
and by decreased filler metal consumption.
This paper will explain the process of laser beam welding, report
on the mechanical properties and fracture toughnesses of some struc-
tural alloys (steels, aluminum and titanium), compare the microstruc-
tures of the fusion and heat-affected zones to the base plate, and
correlate the properties and structures.

LASER WELDING

Thick section laser beam welding is achieved by focusing a high

power laser beam onto the surface of an alloy. This focused, highly
collimated source of heat evaporates the alloy and creates a "keyhole",
or metal-vapor-filled cavity, through the thickness of the workpiece.
Surrounding this keyhole is the weld pool, a volume of molten metal in
dynamic equilibrium with the metal vapor. As the workpiece is moved,
melting takes place at the leading edge of the weld pool and solidifi-
cation occurs to the rear. The depth of penetration of the keyhole and
the shape of the weld pool are governed by a variety of factors such
as: the amount of energy absorbed and the amount reflected at various
 315

temperatures, laser beam power at the focused spot, the size of the
focused spot, travel speed, the gas used to shield the weld pool, the
method of gas shielding, the thermal properties of the alloy being
welded, etc.
At the top of the keyhole, above the workpiece, the evaporated
alloy forms an ionized gas called plasma. If this plasma is allowed to
remain on top of the keyhole, it will absorb some of the laser beam and
reradiate it in all directions, thus decreasing the depth of penetra-
tion of the weld. The plasma can easily be displaced by a stream of
inert gas across the top of the keyhole. Helium, because of its high
ionization energy, is usually used for plasma control, although a
helium-argon mixture may be used effectively ( 1 ) . Thus, plasma
formation is not considered to be a limiting factor in laser beam
welding.
The relative position of the top of the plate and the focal point
of the laser beam has been shown (2) to affect the penetration cap-
ability for a fixed power and speed. In order to fabricate the deepest
welds with a high depth-to-width ratio, the focal point of the laser is
usually positioned about 2.5mm (0.1 in.) below the top surface of the
plate.
For a given thickness of plate, the combination of laser beam
power and travel speed, that produces a satisfactory weld, must be
determined. In order to maintain a stable keyhole, the laser power and
welding speed must be adjusted. One means of determining these values
is to produce a series of bead-on-plate weldments covering a range of
powers and speeds. At too slow a welding speed severe undercut
results. At too fast a welding speed complete penetration is not
achieved. At too high a power the keyhole falls through, whereas too
low a power results in incomplete penetration.

EXPERIMENTAL METHOD

Laser beam welds have been made in many alloys. Aluminum (3),
titanium (3,4,5), steels (3, 6, 7, 8, 9 ) , and a variety of other alloys
(7) have been used to demonstrate the ability of the laser to weld
thick section plates.
The structural alloys studied in this investigation were steels,
aluminum and titanium. The laser beam welding parameters are given in
Table I. The laser power is measured at the exit of a telescope
external to the laser. The laser beam impinges on four mirrors before
the beam strikes the workpiece. The spot size of the laser beam is
approximately 0.75mm (0.030 in) at the workpiece. The final focusing
mirror has a f-number of about 6.
Mechanical property measurements were made using transverse
weld specimens. Charpy V-notch and dynamic tear testing was done
at different temperatures. Hardness tests were made on polished
and etched cross sections using a microhardness indenter. All test-
ing was performed in accordance with the pertinent ASTM specifi-
cations.
 316

Table I. Laser Beam Welding Parameters

Thickness Power Travel Speed

cm (in) kw mm/sec (iptn)

6 (.25) 6 21 (50)
9 (.38) 10 17 (40)
12 (.50) 13 14 (33)
15 (.63) 13 6 (15)

6 (.25) 8 25 (60)
12 (.5) 12 13 (30)

Al 12 (.50) 8 28 (65)

Ti 12 (.50) 11 15 (35)

The microstructures of the base plate, heat-affected and fusion

zones were observed on polished and etched cross sections using an
optical microscope. A field emission scanning electron microscope
was used to observe the separation processes on the fracture surfaces
of the different specimens. All fractographic observations were made
using stereographic pairs. Variations in alloying elements across the
fusion zone were determined with an electron mlcroprobe. Differences
in alloy concentrations on the fracture surface were determined by
energy dispersive x-ray analysis.

RESULTS AND DISCUSSIONS

In order to provide a coherent discussion, this section will be

arranged according to alloys. Thus the steels will be discussed first,
followed by aluminum, and finally by titanium.

STEELS

The mechanical properties of laser beam weldments of the steels

are given in Table II. These data are those of the base plate since
all of the specimens fractured in the base plate. The weld itself,
therefore, is stronger than the base plate. The energy absorbed by the
Charpy V-notch (CVN) specimen for A36 base plate and laser beam welds
in the as-welded and stress relief condition, as a function of tempera-
ture, is shown in Fig. 1. Dynamic tear (DT) energies of the HY steels
have also been measured as a function of temperature (3,10). In HY-130
a very sharp ductile to brittle transition temperature occured at -11°C
 317

Table II. Mechanical Properties of Laser Beam Weldments

Yield Ultimate Elongation/

Thickness Strength Strength Hardness
Alloy mm (In) MPa (ksi) MPa (ksi) % HRC
A36 6 (.25) 276 (40) 442 (64) 23
9 (.38) 269 (39) 428 (62) 22 10
12 (.50) 276 (40) 448 (65) 24
15 (.63) 269 (39) 435 (63) 26
HY80 6 (.25) 621 (90) 745 (108) 25 40
12 (.50) 627 (91) 752 (109) 24 43
HY100 6 (.25) 697 (101) 759 (110) 20 41
HY130 6 (.25) 966 (140) 1022 (148) 13 40
12 (.50) 962 (138) 987 (143) 19 40

280 1 1 i i I

240 - A( • L-T
/ # LBW
• LBW-SR
200 -

cc
160 -
\^M •
X
o
120
O

>
0. 80 -
<
X
o
40 -

1 I 1
-60 -20 20 60 100
Figure 1 Charpy V-Notch energy as a function of temperature for A36
plate and laser beam weldments.
 318

(12°F). A stress relief heat treatment (tempering at 590°C (1100°F))

restored the impact energy to the upper shelf value when measured at
-18°C (0°F). The HY-80 laser beam weldments exhibited a low DT energy
which was the result of solidification cracking (10). A comparison of
the energy dispersive x-ray spectra, obtained from a region of solidi-
fication cracking and a region of microvoid coalescence, indicates a
definite increase in the amount of silicon in the area of solidification
cracking.
The microstructures of the base plate, heat-affected and fusion
zones of the HY-130 steels have been identified (3, 9, 10). The base
plate has a martensitic structure, th'i heat-affected zone (HAZ) is
primarily martensite with some bainite, whereas the fusion zone is
martensite and bainite. The hardness of the base plate is about 30-31
HRC, increases to almost 40 HRC in the HAZ, and is in the range of
37-39 HRC in the fusion zone. The microstructure of HY-80 and 100
laser beam weldments is similar to that of the HY-130 weldments but the
hardness values in the HAZ and fusion zone are slightly higher.
In the A36 laser beam weldments (1) the base plate is ferrite and
pearlite, the HAZ is refined ferrite and pearlite, whereas the fusion
zone is bainite. These structures are shown in Fig. 2. The hardness
values of these regions are shown in Fig. 3 for both the as-welded and
stress relief conditions.

V
a) BASE PLATE b) BASE PLATE

c) HEAT AFFECTED ZONE d) FUSION ZONE

Figure 2 Microstructures of base plate, heat affected and fusion
zones of laser beam weldments of A36.
 319

1
•LIEF
UJ
Q
UJ
11 5
O QC L
CD
•
• • •
UJ
CO
TRES 1 I
UJ
<O
CO
CO / m CO
<
i OQ
< CO
/ «
•
1
N

-—
2 z
CO O
3 N

\l "
UJ

•1 Q.

u •r
UJ

| 1 1 1
1 i
O O o
CM
CM
00 o

(NAH) SS3NQdVH
Figure 3 Hardness traverses across the base plate, heat affected and
fusion zones of laser beam weldments of A36.
 320

Recent experiments have measured the resistance of laser beam

weldments of HY.-130 steel to stress corrosion cracking (SCC). These
experiments indicate that the average KTc/->f value of the laser beam
weldments is 108 MPaYi" (98 ksiYTa) which compares favorably to the
value of 103 MPaYTn" (94 ksi Yin) reported by Fujii (12) for the same
thickness base plate. Fractographlc examination of the specimens
revealed a considerable difference in the fracture separation mechanism
from one side of a specimen to the other, i.e., from the top to the
bottom of the weld. The top of the weld failed by itdcrovoid coales-
cence, the center by a combination of microvoid coalescence and
cleavage, and the bottom by intergranular with some cleavage. In order
to determine the reasons for this, a series of specimens were tested
identically but not allowed to fail, i.e., the SCC crack was started
but not allowed to propagate the entire length of the weld. Each
specimen was then sectioned as in Fig. 4. These sections were polished
and etched to reveal the microstructure, grain structure and solidifi-
cation structure. The top of the weld solidified in a cellular mode
and the crack propagated through the cells. The center of the weld
solidified in a cellular/cellular dendritic mode and the crack propa-
gated along the cell walls and through the dendrites. The bottom of

METALLOGRAPHIC SECTIONING
OF SCC SPECIMEN

TOP

CENTER

BOTTOM

Figure 4 Sectioning technique for SCC specimens of HY130 laser beam

weldments
 321

the weld solidified in a cellular dendritic mode and the crack propa-
gated along the dendrites. The solidification structure and crack path
through the three distinct areas i& shown in Fig. 5. The terminology
is that of Savage (13). Examination of the grain structure in these
sections and in a transverse section indicates a consistency with the
solidification mode. The amount of constitutional supercooling
increases from the top of the weld to the bottom. Hardness traverses
across these sections indicated that the fusion zone in the bottom
section was 4-5 points harder on the HRC scale than either the center
or top sections.

ALUMINUM 5456

Laser beam welding of aluminum alloys has produced mixed results.

Porosity which can be a problem in welding aluminum by any process is
a critical problem in laser beam welding with its Inherent keyhole
and fast welding speeds. The porosity tends to reduce elongation and
ultimate strength but not to effect yield strength (3, 14). DT tests
indicate that, despite porosity, the DT values of the lasei- beam welds
are greater than those of the base plate but less than those of gas
metal arc welds. A fractographic study of the DT specimens showed the
separation process to be microvoid coalescence. A considerable amount
of porosity was observed on the fracture surfaces.
The keyhole phenomena of laser beam welding in an alloy in which
there is a considerable difference in melting point/boiling point
temperatures of the alloying elements, leads to vaporization of those
elements which have low boiling temperatures. Table III lists the
principal elements found in the Al 5456 alloy and their melting and
boiling point temperatures. Examination of the Table leads one to
conclude that magnesium would be easily evaporated. Electron micro-
probe analyses on the cross sections of the laser beam welds reveal a
decrease in the magnesium content as one proceeds from the base plate
to the center of the weld. This decrease is about ten percent or 0.5
w/o. A decrease of 0.5 w/o Mg in aluminum magnesium alloys has been
shown to decrease the yield strength (15).
The keyhole phenomena also softens the strain hardened aluminum
alloy by fusion zone purify "tion and a refinement and redistribution
of solutes.

Table III. Melting and Vaporization Temperatures

TM °C TB °C
Element (Melting Temp) (Boiling Temp)

Al 660.4 2467
Mg 648.8 1090
Mn 1244.0 1962
Cr 1857.0 2672
 322

TOP

CENTER

BOTTOM

Figure 5 Solidification structures and crack path at the top, center

and bottom of laser beam welds of HY130
 323

The microstructures of the base plate, HAZ and fusion zone are
shown in Fig. 6. The base plate consists of a fine, elongated distri-
bution of the precipitate Mg Al in a matrix of Al. Also present
in the matrix are large insoruble particles of Mg Si and (Fe.Mn)Al .
The fusion zone consists solely of refined, fine grained Al and coarser
precipitates of Mg Al , whereas in the HAZ a redistribution and shrink-
ing of the large insoluble particles has occurred in the matrix of Al
and Mg A l y

TITANIUM

Laser beam welding of a titanium alloy (Ti-6Al-2Cb-lTa-0.8Mo)

has resulted in satisfactory mechanical properties and fracture tough-
ness. The plates were grit blasted and the edges to be welded were
pickeled in a HF-HNO solution, washed with alcohol and dried with
nitrogen. The butt welds had a flow of helium gas as a bottom shield.
An area in front of the laser beam/workpiece interaction zone was
flooded with helium and a long trailing shield was used. The results
of testing the weldments are given in Table IV. The CVN and DT values
were determined at both 0° and 25°C (32° and 77°F). These values
compare favorably to those of the base plate (16).
A precracked cantilever beam specimen was fractured in air prior
to SCC testing. A K value of 90 MPaTS (82 ksiYin) was obtained for
the air break. As seen from Table IV, this value falls within the
K range, indicating nonsusceptibility of the Ti-6211 laser beam
welds to SCC under the test conditions.

Table IV

Mechanical Properties of Ti-6211 Laser Beam Welds

Range

YS 834-883 MPa (121-128 ksi)

UTS 924-945 MPa (134-137 ksi)
% Elong. 10-19
% R.A. 15-43
CVN 42-53 Nm (31-39 ft. lbs.)
DT 245-256 Nm (181-189 ft. lbs.)
KT 82-95 MPaVST (75-86 ksi \irT)
Iscc

The microstructure of the base plate (Fig. 7) consisted of elon-

gated primary alpha and Widmanstatten alpha plus beta. Tha far HAZ
contained partially dissolved, or ghost alpha and Widmanstatten alpha
plus beta. The coarse grained, near HAZ, the betatized region adjacent
to the fusion zone, consisted primarily of a coarse Widmanstatten
structure and globular alpha. The microstructure of the fusion zone is
 324

Base Plate

Heat Affected Zone

Fusion Zone
Figure 6 Microstructures of base plate, heat affected and fusion zones
of laser beam weldments of A15456.
 325

Base Plate

Heat Affected Zone

Fusion Zone
Figure 7 Microstructure of base plate, heat affected and fusion zones
of laser beam weldment of titanium alloy.
 326

predominately martensitic alpha. Variations within the fusion zone

microstructure (Fig. 8) include: A) a coarse acicular branched struc-
ture, B) a fine acicular alpha structure, C) a fine Widmansta'tten
structure, and D) a fine lamellar structure.

COARSE ACICUIAR - BRANCHED PINE ACICULAR

WIDMANSTATTEN LAMELLAR
Figure 8 Various microstructures found in fusion zone of titanium
laser beam weldment.

The martensitic alpha is a result of the rapid cooling of the

molten weld pool. A fully martensitic structure is difficult to
develop in alloys which are lean in beta-stabilizers, since the kinet-
ics of the competing alpha phase nucleation and growth reaction are
more rapid (17). Thus the variations in the alpha phase that are
observed are consistent with other observations.
The fractography of the SCC area is interesting in that inter-
facial separation between favorably oriented adjacent alpha grains is
observed as flutes. Yet the flutes did not degrade the SCC resistance
of the fusion zone.
 327

SUMMARY AND CONCLUSIONS

Laser beam welding of A36 steel produces weldments that have

mechanical properties and impact properties at least equal to or
greater than those of the base plate. A stress relief heat treatment
raises the upper shelf level of CVN curves and also lowers the transi-
tion temperature. The reasons for this are the refinement in the
fusion zone, the bainitic microstructure, the decrease in hardness
values, and evidence for a less defined center line (11) in the stress
relief weldment. This alloy has been welded in a variety of thick-
nesses with good results and minimal distortion.
Quench and temper steels produced an unterapered martensitic
structure in the fusion zone and high hardness values. The result
is that in good welds a high ductile-to-brittle transition occurs.
In alloys which have high concentrations of the trace elements (Si, S,
P, etc.), although the fusion zone is cleaned up by the fusion zone
purification process, solidification cracking occurs.
Aluminum alloys, when laser beam welded, showed the usual aluminum
welding problem - porosity. The keyhole and the fast welding speed
trap gases in the fusion zone which is refined. In strain hardened
alloys, welding disrupts the strengthening process. A refinement and
redistribution of the second phases results in a softening (lower
hardness and ultimate strength) of the fusion zone. Conversely a
a modest increase in fracture toughness results when compared to the
base plate.
Titanium alloys must be protected from atmospheric contamination
when they are laser beam welded. Good mechanical properties and
fracture toughness have been observed in this alloy. The laser beam
welds have properties equal to, or better than, those of the base
plate, gas metal arc and gas tungsten arc weldments. The high energy
density, low heat input process produces weldments in which the oxygen
and nitrogen pickup is low and the toughness and SCC resistance are
satisfactory. The fast cooling process results in a predominantly
martensitic alpha prime raicrostructure which, when fractured, separates
by microvoid coalescence.

REFERENCES

1. F. D. Seaman, "The Role of Shielding Gas in High Power CO. (CW)

Laser Welding," Technical Paper MR77-982, Society of Manufacturing
Engineers, Dearborn, Michigan, 1977.

2. F. D. Seaman, "The Role of Focus in Heavy-Plate Laser Welding,"

Technical Paper MR78-345, Society of Manufacturing Engineers, Dearborn,
Michigan, 1978.

3. E. A. Metzbower and D. W. Moon, "Mechanical Properties, Fracture

Toughnesses, and Microstructures of Laser Welds of High Strength
 328

Alloys," in Applications of Lasers in Materials Processing, E. A.

Metzbower, Editor, American Society for Metals, Metals Park, Ohio,
1979.

4. C. M. Banas, D. M. Royster, D. Rotz and D. Anderson, "Comparison

of Electron Beam, Laser Beam and Plasma Arc Welding of Ti-6A1-4V,"
presented at the 56th Annual Meeting of the American Welding Society,
Cleveland, OH, April 21-25, 1975.

5. F. W. Fraser and E. A. Metzbower, "Laser Welding of a Titanium

Alloy," to be published in Advanced Processing Methods for Titanium,
D. F. Hasson and C. H. Hamilton, Eds., IMS, Warrendale, PA, 1982.

6. E. M. Breinan and C. M. Banas, "Fusion Zone Purification During

Welding with High Power CO Lasers," Proceedings of the Second Inter-
national Symposium of the Japan Welding Society, Osaka, Japan, August
24-28, 1975.

7. E. M. Breinan and C. M. Banas, "Preliminary Evaluation of Laser

Welding of X-80 Artie Pipeline Steel," WRC Bulletin No. 201, December
1974.

8. E. M. Breinan, C. M. Banas, and M. A. Greenfield, "Laser Welding -

the Present State-of-the-Art," AFML-TR-75-149, Nov. 1975 (also in
"Source Book on Electron Beam and Laser Welding," M. M. Schwartz,
Editor, American Society for Metals, Metals Park, Ohio, 1981).

9. J. Stoop and E. A. Metzbower, "A Metallurgical Characterization of

HY-130 Steel Welds," Welding Journal, Vol. 57, November 1978, p. 346S.

10. E. A. Metzbower and D. W. Moon, "Fractography of Laser Welds,"

in Fractography and Materials Science, ASTM STP 733, L. N. Gilbertson
and R. D. Zipp, Eds., ASTM, Philadelphia, PA, 1981.

11. E. A. Metzbower, D. W. Moon and F. W. Fraser, "Mechanical Behavior

of Laser Beam Weldments," to be published in Trends in Welding Research
in the United States, American Society for Metals, Metals Park, OH,
1981.

12. C. T. Fujii, "Effect of Specimen Dimensions on K ^ Determina-

tion by the Cantilever Method," NRL Report 8236, Washington*; D. C., May
31, 1978.

13. Warren F. Savage, "Solidification, Segregation and Weld Defects,"

in the Proceedings of the Fifth Bolton Landing Conference, R. J.
Christoffel, E. F. Nippes and H. D. Solomon, Eds., General Electric
Company, Schenectady, N.Y., 1978.

14. D. W. Moon and E. A. Metzbower, "Laser Welding of Aluminum Alloy

5456," presented at AWS Annual Meeting, Kansas City, MO, April 1982,
and submitted for publication in the Welding Journal.
 329

15. L. F. Mondolfo, Aluminum Alloys, Butterworth Ltd., Washington,

D. C., 1976.

16. Metals Handbook, Ninth Edition, Vol. 3, 1980, American Society

for Metals, Metals Park, Ohio, p. 386.

17. J. C. Chesnutt, C. G. Rhodes, and J. C. Williams, "Relationship

Between Mechanical Properties, Microstructures, and Fracture Topography
in Alpha+Beta Titanium Alloys," Fractography-Microscopic Cracking
Processes, AST STP 600, American Society for Testing Materials, 1976.
330
 ELECTRON BEAM WELDING
FOR HEAVY SECTION STEEL PLATES

J.Tanaka, S.Kosuge and H.Furukawa

T e c h n i c a l Research C e n t e r
Nippon Kokan K.K.
M n a m i - w a t a r i d a , Kawasaki, 210, Japan
ABSTRACT

A s t u d y has been made of t h e s e l e c t i o n of e l e c t r o n

beam ( EB ) welding p r o c e d u r e s t o o b t a i n sound welds and
the EB welded joint properties for heavy section steel plates.
The EB machine used has a capacity of 75 kW( 150kV x 500mA ).
The materials examined are carbon steel(ASTM A516 Gr.70) and
2.25Cr-1Mo steel(ASTM A38? Gr.22 C12) with thickness in the
range 50-200mm.
Porosity and solidification crack were identified as the
two major defects likely to occur in EB welds of thicker steel
plates. These defects are influenced by steel composition and
welding procedures. I t was found that they depended strongly
on the solidification pattern for such steels with low oxygen
and nitrogen content as used in this experiment and could be
prevented by the selection of welding procedures. Beam
oscillation was the effective technique to prevent weld defects.
Horizontal position was found to be preferred for welding steel
plates more than 100mm thick and sound welds could be obtained
in 200mm thick steel plates.
The Charpy impact properties of carbon steel and
2.25Cr-iMo steel welded joints have been examined. Very good
results were obtained in 2.25Cr-1Mo steel for weld metal and
HAZ. As for carbon steels, reduction in Ceq. of base metal
was benificial to improve toughness. The toughness of A516-
Gr. 70 steel containing Q.2V/0 C was very poor in the as-welded
condition, while good toughness could be obtained after PWHT.
HAZ Charpy properties were generally superior to those of weld
metal. The results of other joint proprties such as tension,
bend, hardness, COD, creep rupture and NRL drop weight tests
were also comfirmed to be satisfactory.

331
 332

These r e s u l t s lead t o t h e conclusion t h a t t h e EB welding

process was f u l l y applicable t o welding of heavy section
s t e e l p l a t e s up t o 200mm t h i c k .

1. INTRODUCTION

The application of electron beam welding (hereinafter abbreviated as

EB welding), up t o now, has been r e s t r i c t e d t o t h e i n d u s t r i a l f i e l d of
small-sized mass production, where thinner m a t e r i a l s a r e mostly used.
However, i n recent years t h e tendency t o apply EB welding t o t h e heavy
section s t r u c t u r e s has become stronger since high power EB welding machines
with high r e l i a b i l i t y have been developed, and some examples of p r a c t i c a l
application of EB welding have been already reported.''
The authors have concentrated mainly on t h e s e l e c t i o n of EB welding
conditions t o obtain sound welds and t h e welded j o i n t p r o p e r t i e s f o r thick
carbon s t e e l and low alloy s t e e l p l a t e s by using a high power EB welding
machine with a capacity of 75kW. Through t h i s study i t was confirmed
t h a t sound welds could be obtained with appropriate beam o s c i l l a t i o n and
the joint properties were satisfactory.
This paper describes the effects of process factors and steel
composition on weld defects and the joint properties of A516 Gr.70 carbon
steel and 2.25Cr-1Mo steel.

2 . MATERIALS USED AND EQUIPMENT

Welding t e s t s were carried out on ASTM A516 Gr.70 carbon s t e e l with

thickness i n t h e range 50-150mm. Welded j o i n t p r o p e r t i e s were studied on
100mm thick A516 Gr.70 carbon s t e e l and 200mm thick A?87 Gr.22 C12
2.25Cr-1Mo s t e e l . In order to i n v e s t i g a t e t h e effect of s t e e l composition
on s o l i d i f i c a t i o n crack and toughness of weld metal, additional laboratory
s t e e l s of 50mm i n thickness were also used. Chemical composition of t h e
materials used i s l i s t e d i n Table 1.
The EB welding machine used has a capacity of 75kW(15OkV x 500mA)
with nearly p a r a l l e l beam shape. EB welding was done under a vacuum of
5 x 10-4 Torr.
Table 1 Chemical composition of materials used
(
Thickness •
•\aterials C ci Ml P 3 Cu Ki Or
(mm)

50 0.23 0.22 1.06 0.012 0.004 0.15 0.25 0.'6 tr

A516 Gr.70 100 0.21 0.22 1.07 0.010 0.003 0.17 0.25 O.15 tr
Commercial
150 0.18 0.2} 1.09 0.012 0.002 0.15 0.25 0.18 tr

A387 Gr.22 C12 200 0.15 0.09 0.53 0.005 0.004 0.01 0.03 2.36 1.00

0.10 0.10 0/?5

Laboratory 50 -%•• 0.011 0.007 tr tr tr tr
0T35 0.40 1^60
 333

5. WELDING CONDITIONS FOR CARBON STEELS AND LOW ALLOY STEELS

3.1 Fenetration Depth

The effects of beam current and welding speed on penetration depth

were investigated by keeping accelerating voltage and focal point
constant at 15OkV and just on the surface of base metal respectively.
These results are given in
Fig. 1 (flat position) and
Fig. 2 (horizontal position),
where Ib: beam current (mA),
vb: welding speed (cm/min.),
PF, P H J penetration depth in
flat position and in horizontal
position respectively (mm).
As shown in Fig.1, Pp in flat
position is given by
Pp=2.0 Ib//vb in the range of
50
Ib//vb =45. In the range of
lb//Vb
Ib//vb~>45 indicating high
beam current and low welding Pig. 1 Penetration depth in flat
speed, the penetration depth position
almost leveled off due to beam
cavity closure under hydrostatics
forces of molten metal during
welding.
On the other hand, beam V=15O k v
cavity is kept stable during
welding in horizontal position x**
because most of molten metal
which are drawn behind the
beam cavity by surface tension
solidify at that place without
flowing into beam cavity due to
small molten metal head in
comparison with flat position.
Figure 2 explains that the
penetration increases
progressively as beam current
increases and welding speed
decreases. The depth is given
9
by PH=2.9 Ib°* /vb°*4.
Therefore horizontal Fig. 2 Penetration depth in horizontal
position was found to be position
preferred for welding steel
plates more than 100mm thick.
 334

3.2 hetal-drop-through in Flat Position

In case of through-thickness beam welding of thick steel plate in

flat position, metal-drop-through occurs i f the weight of molten metal
becomes heavier than the surface tension of back weld bead which supports
molten metal. The force supporting molten metal by surface tension of
back bead i s given by T(1/R1 + Vife)* where Ri i s the radius of curvature
in the welding direction, R2 i s one in the transverse direction and T i s
surface tension(1i50dyne/cm at 1500°C). This force can be approximated
to X/R.2 because R-| i s almost infinite. The balance condition between
the weight of molten metal(pgh) and the supporting force by the surface
tension(2JT/B^R) i s given by pgh=2lr/%R.
where
Pi density of metal (g/cm3) 15
h: thickness (cm) - calculated curve
Bfofft(=2R2)s back bead width(cm)
The occurrence of metal-drop- • :drop-throucfi
through i s strongly affected by O:no "
)O -
thickness of base metal and back
bead width, and this correlation •drop-through
for A516 Gr.70 carbon steel i s
shown in Fig. 3« I t can be found
that the calculated values are
in good accordance with the
experimental values. Taking
off-seam defect into consideration,
i t may be necessary to keep So
at least 2.5mm of bead width. h (mm)
Accordingly the allowable
thickness of carbon steel to be Fig. 3 Metal-drop-through in
welded without metal-drop- f l a t position
through keeping 2.5mm of back
bead width i s restricted to about 20mm.
For plate thickness greater than 20mm, a backing bar must be attached
to avoid metal-drop-through.

3.3 Metal-run-out from Front Bead

One of the difficulties when welding thick steel plates in horizontal

or vertical position i s metal-run-out from the front bead, which causes
streaks of molten metal to solidify on the joint surface and to produce
large pores in weld metal. Relationship between metal-run-out and welding
parameters when welding without a fronting bar to support molten metal
and to prevent metal-run-out was investigated. These results in
horizontal position are shown in Fig. 4. In case of non through-thickness
beam welding, i t was found that the welding speed to prevent metal-run-out
 335

should be increased with increase in beam current, and that metal-run-out

could not be avoided at a beam current of above 270mA even though welding
speed was increased.
This indicates that the extent
of excess weld metal on the O non through -thickness beam
front bead must be controlled no meial nxi out with non D through // MX)™1*1
through-t
appropriate to avoid metal-run-
out. Maximum thickness of Solid • metal run out
Open :no *•
carbon steels to be welded < i beam oscibton Airphtude

without metal-run-out was

restricted to approximately
75mm. ]
When welding with through- J
thickness beam, the plate
thickness capable of welding
without metal-run-out becomes
greater. This i s partly because
the reinforcement of front bead
becomes small due to the back Fig. 4 Metal-run-out in horizontal
bead formation, and because a position
large amount of metal vapour
in the beam cavity which has a detrimental influence on front bead
formation can escape out via beam cavity of back side. A 150mm thick
carbon steel could be welded without metal-run-out. If the extent of back
bead formation i s not sufficient, the molten metal drops out from the
front bead, and this extreme case corresponds to that in the welding with
non through-thickness beam.
The results obtained in vertical-up position are given in Fig.5*
The risk of metal-run-out at the front joint in vertical-up position was
observed to be higher than in horizontal position because excessive
reinforcement was liable to be formed at the front face. In order to
prevent metal-run-out i t was found necessary to make bead width small at
the front face in addition to good bead formation at the back bead.
The regions to be welded without metal-run-out in vertical-up position
were narrower in comparison with
those in horizontal position, V=15O><V
and i t was impossible to prevent
lal run
metal-run-out at a beam current "° ™ «•<
O ' 50"""lhKk
of above 150mA b e c a u s e wide O = 75"™ "
molten p o o l was formed a t t h e 5 A : 100"™ "

f r o n t f a c e . Maximum t h i c k n e s s
for carbon steel to be welded I Solid : metal nn out
without metal-run-out i s about % Open: no ^

100mm, but i t s region without ^1

metal-run-out i s very small.
The steel thickness which could
be stably welded was found to be
100
up to 75 It, ( mA)

Fig. 5 Metal-run-out in vertical-up

position
 336

3.4 Weld Defects and Their Prevention

3.4.1 Porosity

An example of correlation
between welding conditions and V=150 kv
weld defect occurrence for 0-.no oscillation
A516 Gr.70 carbon steel in 500- D: oscillation
horizontal position is shown 400- Solid: defects
in Fig. 6. The welds were Open: no „
: 300"
examined by radiography. P' porosity
Porosity was observed at £•200" fc • > VC: vertical crack
almost all welding conditions 100- a HC: horizontal //

when welding without beam -10 15 25 ^5

oscillation, and most of V (cm/min) b

porosities were found in the

vicinity of mid-thickness Fig. 6 Correlation between welding
position. When welding with conditions and weld defects
beam oscillation, porosity
disappeared irrespective of
beam oscillation patterns
such as longitudinal, transverse, 1SO X 2 0 0 kv mA
X 12°""'"*" 1
Thicknsss: lOO ™ 11

and circular deflection. no oscillation X-oscillation Y-oscillation

In order to clarify its effect on solidification line
preventing porosity occurrence,

Hi fff
solidification pattern in the
longitudinal direction was c
i
investigated as shown in Fig. 7.
In case of no beam oscillation, - W.D. W.D. - W.D.
it was observed that molten metal
solidified faster near the top Fig. 7 Effect of beam oscillation
and the bottom side than at mid- on solidification pattern
thickness. Such a solidification
pattern indicates that gas and metal vapour existing around mid-thickness
can not easily escape out of weld and are liable to be trapped around
there, which results in porosity occurrence in the weld metal.
When welds were made using beam oscillation, it was found that
solidification started on the bottom side and ended on the top side
without the delay of solidification around mid-thickness, and that beam
cavity was being kept stable without cavity closure during welding.
In this case, gas and metal vapour can easily escape out and porosity
does not occur.
From these results, it was found that porosity depended strongly on
the solidification pattern for such steel plates with low oxygen content
as used in this study. It was considered necessary to keep beam cavity
stable and to avoid cavity closure during welding so as to prevent
porosity. Beam oscillation was the essential and effective technique
for avoiding porosity in other welding positiore such as flat and vertical
up position.
 33"

This technique was also veiy effective to prevent shrinkage crack due to
solidification instability.
The porosity occurrence i s affected by steel composition as well as
welding parameters. In particular, oxygen content iii the base metal was
found to be the most important element controlling porosity. This i s ,
as described below, because the oxide inclusions in steel can be deoxidized
by carbon and CO gas causing porosity formation can bu easily produced.
As the oxygen potential of AI2O3 i s the lowest among oxide inclusions such
as TiO2, SiO2 and AI2O3, the deoxidizing reaction of AI2O3 was taken up
as an example. I t s reaction .an be exemplified by the following equations,
where the temperature of molten metal impinged by EB and the pressure in
the beam cavity were supposed to be 1800°C and 1 Torr(1.3 x 1Cr3atm)
respectively.
2A1 + 3/202 = AI2O3 AG°2O73 = -248,500 cal (1)
3C + 3/202 = 3C0 AG°2O73 = -210,400 cal (2)
(2)-(1) AI2O5 + 3C = 2A1 + 3C0 AG°2O73 = 38,100 c a l
Pco = 4.5 x 10-2atm
.*. Pco > 1.3 x 10-3 atm ( 1 Torr )
Accordingly i t i s presumed that even AI2O3 can be easily deoxidized by
carbon and CO gas i s produced. Another reaction, i . e . AI2O3 = 2A1 + 3/202,
cannot progress because the dissociation pressure of AI2O5 at 1800°C i s
approximately lO-^atm. I t has been reported that CO gas amount in the
vacuum atmosphere chamber remarkably increase during welding, particularly
when welding rimmed steels.2
In order to confirm the deoxidizing reaction of oxide during welding,
TIO2 powder was added to the weld metal. The oxygen potential of TiO2
( AGO2073=-131kcal) i s higher than that of Al203( AG°2O73=-i66kcal),
whereas i s lower than that of SiO2( AGo2073=l!-117kcal). TiCp powder was
painted as evenly as possible onto the joint surface of 50itan thick
A516 Gr.70 carbon steel(0.23% C) by using methyl alcohol before welding.
TiO2 amounts of the order 0.01-0.05% were added to the weld metal.
The weights of TiO2 powder to be added were calculated by assuming 5.0mm
of weld width at mid-thickness so as to give an additional 0.006-0.03% Ti
and 40-200ppm oxygen in the weld metal. These results are l i s t e d in
Table 2.

Table 2 Ti,0 and C in weld metal in case of Ti0 2 addition

(wt°/o)

Ti in deld Metal 0 in rfeld Ketal * C in .Veld Ketal

TiO2 Addition
Calculated Experimental Calculated Experimental
. 0 tr tr - 18 0.23
0.01 0.006 0.007 40 14 0.23
0.03 0.018 0.016 120 15 0.?3
0.05 0.030 0.028 200 10 0.2?
Base Metal : A516 Gr.70 of 50 mm thick * ppm
 338

They show that Ti contents in the weld metal are nearly equal to the
calculated values but oxygen are not added to the weld metal at a l l , which
indicates that TiO2 cannot remain unchanged as i t i s in the weld metal.
Oxygen i s estimated to escape out as CO gas due to the deoxidation of TiO2
by carbon. There i s a slight decrease in carbon content at 0.05% TiO2.
However, the exact estimation of carbon content i s difficult because
carbon content i s by far more than oxygen content. I t was observed that
more addition of TiO2> i*e. 2.0;6, caused j e t gas to scatter out almost a l l
molten metal as very fine spatters.
Thus, oxygen content and plate thickness are important for avoiding
porosity, and oxygen levels must be lowered as plate thickness increases.
Oxygen content for 100mm thick carbon steel containing 0.21>£ C appears to
be in the range up to 40ppm to obtain porosity free welds.

5.4.2 Solidification Crack

As mentioned above, i t was clarified that shrinkage crack due to

solidification instability could be prevented by use of beam oscillation.
In this study, the influences of welding conditions and steel composition
on the solidification crack were investigated when welding with beam
oscillation. The correlation between weld width at mid-thickness, welding
speed and solidification crack was investigated on O.t^C~O.3/&>i-1.6$fci-
O.OJ$Jb steel with thickness in the range 10-50mm when welding with
through-thickness beam in horizontal position. The incidence of
solidification crack(vertical crack) was plotted against weld width at
mid-thickness (Bwir) and welding speed(v s ) in Fig. 8.

O.1C-O.3Si-l6Mn-o,Q3Nb

O O -—500 •

X3 32mm thick §

$50mm thick Q 10 mm thick

m 20 mm thick

AA : 10mm thick
O9 • 20mm thick open : no crack
6* : 32mm thick solid : crack
O* : 50mm thick
I 1
50 100 150 200
Vs (cm/min)

Fig. 8 Correlation between weld width(BwJ), welding speed(v s )

and solidification crack
 339

Solidification cracking tendency became greater as welding speed and weld

width increased, and crack was more likely to occur in thick plate than
in thin plate. These crack/no crack boundaries are also affected by steel
composition, in particular carbon.
Therefore, the effect of chemical composition on the crsetting
tendency was investigated by using laboratory carbon steels of 50mm thick.
As listed in Table 1, G,Si and Mi content were changed keeping S and P
content constant. Welds were produced changing welding speed in thft range
15-4Ocm/min with through-thickness beam in horizontal position. Good
correlation between solidification crack, weld width(Bw£) and carbon
content [C] x welding speed (v8) was found as shown in Fig. 9. It was
found that the cracking tendency became greater as carbon content, welding
speed and weld width increased respectively. The obvious effect of other
elements, i.e. Si or Mn, was not observed within the present content range.
It was concluded that the appropriate selection of welding procedures, -
i.e. welding speed and weld width, at a given steel composition was very
important for preventing solidification crack. Appropriate welding
conditions without defects for thick steel plates for three welding
positions are
given in
Table 3*
C : 0.10-0.3S V s B i5~4ocm/min
Si :010-040
Mn:0.75~1.«O o:no crack
P: 0.011 • crack
i S : 0.007

E 4 o 8 o

f
I 2

5 10 15
[C] x vs (%-cm/min)

Fig. 9 Relationship between weld width (Bw£), (c) x welding

speed(v8) and solidification crack
Table 3 Appropriate welding conditions for thick steel plates

Accslerating Beam Welding Beam Oscillation

Thickness Voltage
Welding Current Speed Pattern Cycle Amplitude
I-OBltion (kV) (cm/min.)
(ran) (mA) (Hz) (ran)
100 150 200 12.0 Elliptical 100 2.5
Horizontal
200 150 390 9.0 Elliptical 300 2.0
Flat 100 150 300 22.0 Longitudinal 100 2.0
Vertical-up 100 150 1}0 6.5 Transverse 100 2.0
 340

4. WELDED JOINT PROPERTIES

Welded j o i n t property t e s t s were carried out mainly on 100mm thick

A516 Gr.70 carbon s t e e l and 200mm thick 2.25Cr-1lfo s t e e l . Laboratory
carbon s t e e l s were also used for toughness t e s t of weld metal. Welds were
made with welding
conditions in
horizontal position
as l i s t e d in Table 3.
The transverse section
of weld made for 200mm
thick 2.25Cr-1Mo s t e e l
i s shown Photo. 1.
Welded j o i n t s of
A516 Gr.70 carbon
s t e e l and 2.25Cr-1Mo Photo. 1 Transverse section of 200mm thick
s t e e l were subjected 2.25Cr-1Mo s t e e l weld
to PWHT of 600°C x 4hours
and 69O°C x 30 hours respectively.

4.1 Tension and Bend Tests

The r e s u l t s of
tension t e s t s are l i s t e d Table 4 Results of tension t e s t s
i n Table 4» I t could be
found t h a t a l l welded
j o i n t s fractured at base Materials Thickness TS Fractured
metal and t h e i r strength (mm) (kgf/mm?) Position
exceeded the specified
requirements. 53.7 B.M.
All bend t e s t s such A516 Gr.70 100
54.5 B.M.
as face, root and side
bend showed good d u c t i l i t y . 58.2 B.M.
2.25Cr-1lfo 200
58.0 B.M.

4.2 Hardness Distribution

Maximum hardness a f t e r PWHT, which were observed i n the heat affected

zone(HAZ) for both s t e e l s , were Hv1J?5 for A516 Gr.70 carbon s t e e l and
Hv190 for 2.25Cr-1Mo s t e e l . These r e s u l t s indicated t h a t both welded
j o i n t s were f u l l y tempered.
 341

4.3 Charpy Impact Tests

The results of Charpy impact t e s t s are given in Table 5« I t was

observed that good results were obtained on both steels for »n notch
locations, i . e . weld metal, fusion boundary(bond) and HAZ. Particularly
2.25Cr-1Jto s t e e l had such a superior toughness that could not be expected
for arc welds.
Table 5 Results of Charpy impact t e s t s

Thickness Notch vEo vE-20 vE-40 vTs

Materials Location
(mm) (kgf • m) (kgf-m) (kgf-m) (°O
tf.K. 16.7 12.7 7.5 -19
A516 Gr.70 100 Bond 15.5 13.3 10.2 -20

HAZ 25.5 23.2 21.0 -57

W.M. 29.1 28.5 27.8 <r90
2.25Cr-1Mo 200 Bond 29.7 29.0 27.7 <-90
HAZ 25.7 25.0 20.0 -75
Sampling Position : 1/2t

I t was found that cabon

equivalent(Ceq.) was one of 60
the major factors controlling
toughness and low level Ceq. 40
was beneficial. ~ 20
The correlation between u
fracture appearance transition L o
temperature(vTs) and Ceq. for
C and C-Nn steels including £-20
100-150mm thick eommercialy
available steels was *-40
investigated as indicated in -60-
Fig. 10, where Ceq. i s given
-fl
by Ceq.(JIS) = C + Si/24 + 1.25 0.35 0.45
Nta/6 + Ni/40 + Cr/5 + Mo/4 +
V/14 and the data are of Ceq. (96)
after PWHT.
f i g . 10 Correlation between vTs and
Ceq. for commercial C steels
 342

I t was observed that toughness became better as Ceq. decreased.

Moreover, 50mm thick
laboratory carbon s t e e l s were
added to i n v e s t i g a t e the
toughness for weld metal i n As Welded
the as-welded condition. Thickness = 5 0 m m
Figure 11 shows the correlation so
between absorbed energy at 0°C
(vEo) and Ceq..
I t was found that toughness
depended strongly on Ceq..
With respect to the micro-
structures observed i n the
weld metal for carbon s t e e l s ,
the structure consists mostly
of polygonal f e r r i t e i n the
range Ceq.Sabout 0.50%, while
much f e r r i t e - s i d e - p l a t e •3
> 10
structure appears i n the range
Ceq. > about 0.30# .

"0 0.25 O30 0.35 0.40 045 0.50 055

Ceq. ( % )

Fig. 11 Relationship between vEo and

Ceq. for laboratory C s t e e l s

4>4 Fracture Toughness Tests

COD and NHL drop weight t e s t s

were carried out on 2.25Cr-1Mo
s t e e l . Good COD r e s u l t s were Table 6 Results of COD t e s t s
obtained as l i s t e d i n Table 6.
NOT temperature obtained
were -50°C i n the weld metal, Testing COD Values (mm)
-45°C at fusion boundary and Temperature WM Bond HAZ
-40°C i n the HAZ.
0 >2.0 >2.0 >2.0
-30 >2.0 >2.0 >2.0
-60 >2.0 1.3 0.80
-90 0.17 0.59 0.31
 343

4.5 Other Welded Joint Property Tests

Other welded joint property tests, i.e. high temperature tension,

creep rupture and stress relief(SR) embrittlement test, were carried out
on 2.25Cr-1Mo steel* The test results obtained were all good.

5. SUMMARY

A study has been made of the selection of £5 welding procedures to

obtain sound welds and the E5 welded joint properties for heavy section
steel plates by using a high power £5 welding machine (75kW). It was
confirmed that sound welds could be obtained with appropriate beam
oscillation and the joint properties were all satisfactory. These results
lead to the conclusion that the EB welding process was fully applicable
to welding of heavy section steel plates up to 200mm thick.
The major results obtained are summarized as follows ;
(1) Beam oscillation is a very effective technique to avoid porosity and
shrinkage crack. This is mainly because beam cavity can be kept
stable without cavity closure during welding by use of beam
oscillation.
(2) The horizontal welding position is preferred for welding thick steel
plates, and sound welds can be obtained for up to 200mm thick steel
plates.
(3) With respect to weld defects, oxygen content and carbon content
have a str^au. correlation with porosity and solidification crack
respective-^".
(4) With respect to solidification crack, crack/no crack boundaries are
affected by steel composition, welding speed, weld width and plate
thickness.
(5) Welded joint properties of 100mm thick A516 Gr.70 carbon steel and
200mm thick 2.25Cr-1Mo steel were all found to be good.
Weld toughness for carbon steels has a strong correlation with cabon
equivalent(Ceq.) specified by the following equation.
Ceq. = C + Si/24 + Mn/6 + Ni/40 + Cr/5 + Mo/4 + V/14

REFERENCES

1. H.Kit a, I.Karatsu et al : "Application of Electron Beam Welding to

the Heavy Section Structures," IIW Document, No.W-279-80
2. M.J.Bibby : "Gases Evolved from Electron Beam Welds in Plain Carbon
Steels," Welding Journal, December,1972
 Session G.

NUCLEAR APPLICATIONS
CoCHctmeiT
R.E.Smflh
J. A.JonMAf}pltdRMtcn^Connpcfiy
Chortottt, North Carofcw OokDklO^T
 THE WELDING METALLURGY OF SA5O8 Cl II
HEAT AFFECTED ZONES

P.J. Alberry and J.A. Lambert

Central Electricity Generating Board

Marchwood Engineering Laboratories
Marchwood, Southampton, Hampshire,
United Kingdom.

ABSTRACT

A weld thermal simulation technique has been used to

investigate the metallurgical response of SA5O8 class II
material during welding. Dynamic Aci and Ac3 data,
grain growth kinetics and continuous cooling transformation
diagrams have been measured. The heat affected zone
structure, grain size and precipitate distribution are
described in terms of the weld thermal cycle experienced
and compared with a weld heat affected zone. The as-
welded hardness and tempering response of a range of
possible heat affected zone structures has been established.
The tempering effects of various weld thermal cycles are
calculated from isothermal tempering data. The likely
tempering effects during welding are estimated and compared
with the tempering of actual welds during welding and in
subsequent conventional post weld heat treatment.

347
 348

INTRODUCTION

A computer model for the prediction of the distribution of heat

affected zone (HAZ) microstructures has been developed at Marchwood
Engineering Laboratories as part of a more general programme.
The model translates welding conditions and basic metallurgical
information into a graphical display of the overal HAZ structure
distribution for a given joint geometry. The model has been successfully
used to identify tolerant welding procedures designed to produce HAZ
refinement in CrMoV compositions and its use is now routine in research
work where close control of HAZ structure is required. The model has
also been used to assess the likely effects of various weld procedure
options and provide a quantitative assessment of the likely operational
variability in terms of HAZ structure^.
This work is concerned with the development of the basic HAZ
metallurgy required by the model, prior to the prediction of multipass
HAZ structures in SA5O8 cl II material. The information required
includes continuous cooling transformation data for various HAZ regions,
dynamic Aci and AC3 data, prior austenite grain growth kinetics and
detailed hardness and tempering measurements. The work carried out to
obtain these measurements is described. In addition, a detailed
metallographic and electron microscope examination of the various HAZ
structures on SA5O8 cl II material corresponding to multipass weld
deposits is presented. The work has been carried out using weld thermal
simulation techniques in combination with weld HAZ studies.
The work was partly funded by Babcock and Wilcox, Alliance, as part
of an Electric Power Research Institute Contract, RP1236, Repair Welding
of Heavy Section Steel Nozzles3.

EXPERIMENTAL AND RESULTS

Weld simulation specimens of dimensions 100 mm long and 9 mm

diameter, were machined from a block 130 mm x 75 mm x 75 mm of SA5O8
cl II supplied by Babcock and Wilcox, Alliance. The composition was
designed to represent one standard deviation to the alloy rich side of
BWR pressure vessel materials from various manufacturers, which are in
service. The composition is given in Table I.

Continuous Cooling Transformation Data

4
Weld simulation techniques were used to cycle a series of specimens
to a range of weld thermal cycle peak temperatures corresponding to
those experienced by various regions of the HAZ of a tungsten inert gas
(TIG) weld. Thermal cycle peak temperatures of 1300°C, 1200°C, 1100°C
and 1000°C were chosen as representative of coarse and progressively
finer grained regions of the HAZ. High speed dilatometric measurements,
 349

Table I. Material Compositions (Wt%)

Weld Simulation Specimens Tempering Specimens

Element SA508 cl II SA508 cl II

C 0.23 0.24
Mn 0.62 0.61
P 0.010 0.017
S 0.018 0.010
Si 0.17 0.31
Ni 0.95 0.93
Cr 0.28 0.39
Mo 0.70 0.72
V 0.01 0.025
Nb 0.01 0.013
Al 0.017 0.015
Ti 0.01 0.01
B 0.001 N/A
Cu 0.062 0.073
Sn 0.015 0.023
As 0.020 0.040
Sb N/A <0.01

with a sampling frequency of 10/s, were carried out during the heating
and cooling portions of each weld thermal cycle. A series of weld
thermal cycles of constant peak temperatures and representing increasing
arc power at constant travel speed, 3 mm/s, and preheat, 100°C, were
used for each of the peak temperatures selected. This produced a
series of progressively slower cooling rates, corresponding to progress-
ively higher heat inputs, Figure 1. Diametral strain and temperature
during heating and cooling were logged automatically using an HP9825
minicomputer. Data plots of temperature versus strain were produced
for each thermal cycle in the series and for each peak temperature
selected. The phase transformations during heating and cooling were
determined from the deviation from linearity of the temperature:strain
plots. The continuous cooling data from each of the peak temperatures
1300°C, 1200°C, 1100°C and 1000°C are shown in Figures 2-5. Prior
austenite grain sizes were measured, using standard ASTM charts, as
ASTM 2, 3, 5 and < 8 for peak temperatures of 1300°C, 1200°C, 1100°C
and 1000°C, respectively. These correspond to 0.1 mm, 0.8 mm, 0.04 mm
and < 0.015 mm mean grain diameters of regular tetrakaidecahedra.
Provided that the initial carbide distribution present in the base
material is dissolved during each successive weld thermal cycle, it is
known that a single specimen can be used to produce a complete continuous
cooling transformation (CCT) diagram for particular low alloy ferritic
steels. This implies that each thermal cycle successively re-austenitises
 350

Welding conditions
. Arc power 1 -100 k J / s
N. Travel speed 3mm/s
\ . Preheat 100 C

. \

40 80 120 160 200

Time s

Fig. 1. Programmed wsld thermal cycles used to determine the

continuous cooling transformation data.

1000r

80C(-
Upper
Lower bainite
60O bamite

400
Martensite

200

40 80 120 160 200 240

Time s

Fig. 2. Continuous cooling transformation curves for O.I mm

prior austenite grain size.

the structure and that in the absence of carbides, the prior austenite
grain size and cooling rate determine the transformation product in a
given material. This was shown to be the case for those CCT curves
relating to the 1300°C, 12OO°C and 11OO°C peak temperatures. The
results, in Figures 2-4, include CCT data from specimens which received
only a single cycle together with CCT data from a single specimen in
each case, which received multiple cycles. The data from single and
 351

1000
-
800

600-
Upper bainite
. Lower bainite __-
400
Martensite __,^ —
200
-

80 120 160 200 240 280

Time s

Fig. 3. Continuous cooling transformation curves for O.O8 mm

prior austenite grain size.

1000

800

C Upper bainite
— * Lower
.Martensite

200
i i i i < i i i 1 1 i 1
40 80 120 160 200 240
Time s

Fig. 4. Continuous cooling transformation curves for O.O4 mm

prior austenite grain size.

multiple weld thermal cycles did not show any significant differences.
Strictly, for the 11OO°C specimens, the first cycle did not quite
dissolve all the carbides present in the base plate, these only fully
dissolved on successive cycles. However, a sufficient quantity were
dissolved such that the transformation temperatures measured after
single cycles on separate specimens were identical to those measured
after multiple cycles.
The multiple cycle CCT curve, Figure 5, relates to that portion of
an HAZ which experienced an initial thermal cycle which dissolved all
the carbides and which then experienced a second thermal cycle with a
peak temperature of 1000°C. The curve was generated by first cycling
 352

1000r

5 800p
3 600
[ Upper bamite
I 400 Z~ Lower
Martensite
bamite

200

80 120 150 200

Time s

Fig. 5. Continuous cooling transformation curves for < O.O15 mm

prior austenite grain size.

to 1300 C to dissolve all carbides and then repeated cycling to 1000°C

followed by different cooling rates to generate the CCT curve. However,
the outer edges of weld HAZs can receive only low temperature thermal
cycles which promote incomplete carbide dissolution. Consequently a
second CCT curve was generated from 1000°C using a separate series of
determinations which were carried out using several specimens which
received only a single thermal cycle with a peak temperature of 1OOO°C,
Figure 6. These two conditions are common in multipass weldments due
to the influence of overlapping HAZs from successive weld beads.
The dynamic Ac^ and AC3 temperatures on heating were estimated as
808 C and 892°C. These values are based on data taken every 0.1 s at
a heating rate of % 180°C/s. The values given are expected to be
accurate to ± 18°C. The dynamic Ac^^ was confirmed metallographically,
(as shown under HAZ Structures - Double thermal cycles) to be
approximately 800°C.

900

700
Ferrite

500
' Lower bainite
i"300
Martensite
100
80 120 160
Time s

Fig. 6. Continuous cooling transformation curve for < 0.015 mm

prior austenite grain size, with the carbides not fully in solution.
 353

HAZ Structures

Single thermal cycles

Weld HAZs resulting from the deposition of a single weld bead on a

thick section plate consist of a range of structures and prior austenite
grain sizes which reflect the weld thermal cycle experienced and the
basic metallurgy of the material. In the SA508 cl II composition used
in this investigation, these range from coarse grained bainite/marten-
site mixtures close to the fusion boundary to fine grained ferrite/bain-
ite mixtures at the outer edge of the HAZ. These structures have been
reproduced using a range of weld thermal cycles designated TIG 1300,
TIG 1200, TIG 1100, TIG 1000, TIG 900, TIG 800 and TIG 700. The
nomenclature refers to a weld thermal cycle with heating and cooling
rates resulting from a heat input of 1 KJ/mm and a preheat of 100°C with
the appropriate weld thermal cycle peak temperature. Specimens were
polished and etched in 2% nital and the structures are shown in Figure 7
together with the measured transformation range and Vickers diamond
pyramid hardness at a load of 10 Kg (HvlOKg).
Carbon extraction replicas were taken from each structure to
establish the effect of thermal cycle peak temperature on precipitate
morphology and type. Replicas were removed from the polished surfaces
of specimens cycled to a range of peak temperatures. Extraction times
of ^ 3 minutes were achieved using 2% nital. The replicated regions
were examined on a Jeol 100CX transmission electron microscope and the
predominant precipitates were identified using energy dispersive analysis
of X-rays, EDAX,together with selected area diffraction of several
particles. Additional confirmation of precipitate structure was
achieved by dissolving the ferritic matrix in either 10% HC1 in methanol
or 10% nital and examining the residue using an X-ray diffractometer.
A summary of carbide types identified by this method is given in Table
II.

Table II. Analysis of powder extracts using X-ray diffractometry.

Material and Etchant Carbides Identified

Parent Plate 10% HC1 HC + Mo 2 C (Hexagonal)

Parent Plate 10% nital M..C + Cr
23 C 6

900°C thermal cycle 10% HC1 +

M C Mo 2 C

1100°C thermal cycle 10% HC1 M3C

 354

130G-Peak temperature

Mt^i^'^ 420-260aC-Transformation range

^55-Hardness

1100

52?-3OO°C

U9

900

657-S11°C

267

800
685-485°C

238
I 700

100 Mm

Fig. 7. Optical micrograph! of simulated HAZ structures after

single weld thernal cycles.
 355

The predominant carbide type in SA5O8 cl II parent material was

M3C with an orthorhombic cementite structure. Quantitative EDAX analysis
of a range of particle sizes showed that Mo, Cr and Mn were substituted
into the M3C replacing Fe, Table III. The amount of Fe was estimated
as <\> 87%, with a Mn:Cr ratio of 1.5. The significance of the Mn:Cr
ratio will be discussed in a later paper comparing SA5O8 cl II and
SA533B. Fine, needlelike M02C dispersions were also present in the
SA5O8 cl II parent plate together with a few coarse ^ 2 3 ^ particles.

Table III. Analysis of M3C carbides using EDAX

Composition, wt%
Element Parent Plate 900°C Thermal Cycle IGOQOC Thermal Cycle
Standard w Standard ,, Standard
Mean _ . . Mean _ . . Mean _
Deviation Deviation Deviation
Mo 1.5 0.5 1. 9 0. 8 2.0 2.1

Cr 5.7 1.4 6. 5 0. 8 2.0 0.7

Mn 6.2 0.7 5. 5 0. 4 1.5 0.5

Fe 86.6 2.3 86 1. 2 94.6 2.4

The effects of single thermal cycles with various peak temperatures

are shown in Table IV and Figures 8,9. With increasing thermal cycle
peak temperature, the cementite spherodises and rejects the substitut-
ional transition elements Mo, Cr and Mn to the matrix. After a thermal
cycle to a peak temperature of 1000°C, rejection is almost complete,
Table III.
Progressive dissolution occurs at thermal cycle peak temperatures
above 1000°C and dissolution is complete by 1200°C although some
reprecipitation occurs on cooling, Figures 8,9. The M02C needles
become progressively more spherodised above 900°C and some coarsening
takes place up to 11OO°C. Dissolution is complete between 1100 0 C and
1200°C.
Measured lattice spacings of (103), (210), (121), (211), (122) and
(113) planes for the M3C and Fe3C cementites were within + 0.003 A*.
This indicates that substitution of Mo, Cr and Mn does not significantly
distort the orthorhombic structure.

Double thermal cycles

Weld HAZs resulting from the deposition of two weld beads exper-
ience a double thermal cycle. Previous work^, has shown that in low
alloy ferritic material the HAZ produced by successive thermal cycles
 356

ijj 1200 - Peak temperature

420-268 C- Transformation range

471-Hardness

^ l 5
tfti! **>•!.«*

f-.^atr-H*'*

238
700

202
Pttrent

202
10pm

Fig. 8. Carbon extraction replicas from simulated HAZ structures

after single weld thermal cycles.
 357

•„*!?.•>• "--rv*' .' \_' I _ C J Peak temperature

;"'.;•* 420*268 C Transformation range

Hardnes3

1100

572-300°C

1000
449

*<n 545-277 C

900

657-511°C

. 800

238

.V

**••

Parent 202 f
1/im

Pig. 9. Carbon extraction replicas from simulated HAZ structures

after single weld thermal cycles.
 358

Table IV. Summary of the type and size of carbides in SA508 cl II

after different weld simulations.

Single Double
Carbide Type Carbide Type
Thermal Thermal
and Size and Size
Cycles Cycles

1300°C Carbides dissolved. 1300 + 600 Fine Fe^C.

1200°C Carbides dissolved. 1300 + 700 Coarse, Dense

No Mo 2 C.

1100°C Fe^C - partially 1300 + 800 Coarse, Dense

dissolved. Coarse with precipitate
Spherical M02C. free zones along
grain boundaries

iooo°c M3C _ partially 1300 + 900 Fine Fe.,C .

dissolved.
Spherical M02C.

900°C M3C - with Mo, Cr, 1300 + 1000 Patchy Fe3C colonies
Mn. and areas with no
Spherical M02C. carbides dissolved.

800°C Elongated M3C - 1300 + 1100 Carbides dissolved.

with Mo, Cr, Mn,
Fine needlelike
M02C.

700°C Elongated M3C -- with 1300 + 1200 Carbides dissolved.

Mo, Cr, Mn.
Fine needlelike Mo 2 C

Parent Elongated M3C -• containing Mo, Cr and Mn.

Plate Fine needlelike M02C. Few coarse C r
23C6'

is largely representative of the last thermal cycle experienced for

thermal cycles with peak temperatures in excess of ^ 1000°C. During
a weld thermal cycle, carbide dissolution occurs on heating and grain
growth takes place whilst the material is austenitic. The austenite
is grain boundary nucleated and grain growth is duplicated for each
weld thermal cycle as soon as the transformation to austenite on
heating is complete. The presence of a stable precipitate dispersion
 359

in the austenite can inhibit grain growth and some inhibition during
the early stages of carbide dissolution on heating can occur. This
depends on the precipitate size, volume fraction and on the prior
austenite grain size. Although the previous thermal cycle history may
have produced different carbide dispersions prior to the final weld
thermal cycle, carbide dissolution is rapid and the final structure is
largely independent of the prior thermal history.
A range of specimens has been subjected to a series of double
thermal cycles. In all cases, the first thermal cycle corresponds to
TIG 1300 and the second thermal cycle to TIG 1300, TIG 1200, TIG 1100,
TIG 1000, TIG 900, TIG 800, TIG 700 or TIG 600 respectively. This
nomenclature also refers to a weld thermal cycle with heating and cool-
ing rates resulting from a heat input of ^ 1 KJ/mm and a preheat of
100°C, with the appropriate weld thermal cycle peak temperature. The
specimens were polished and etched in 2% nital and the structures are
shown in Figure 10 together with the measured transformation ranges and
Vickers 10 Kg hardness values. Those second thermal cycles with peak
temperatures less than the dynamic Ac^ i.e. TIG 700 and TIG 600, do
not result in any transformation to austenite but have tempered coarse
grained structures which result from the first thermal cycle, TIG 1300.
It is clear from Figure 10 that the TIG 1300, TIG 800 combination
results in a coarse grained structure with the prior austenite grain
boundaries outlined by an ultra-fine grained ferritic transformation
product. This decorated structure has been cycled to a temperature
just above the dynamic Ac^, causing grain boundary nucleation of
austenite as described previously and confirms that the dynamic Acj
temperature is approximately 800°C,
Carbon extraction replicas were taken from each structure and the
precipitate type and morphology established as described previously.
Examples of the extraction replicas and precipitates are shown in
Figures 11, 12. After a single thermal cycle to a peak temperature of
1300°C, all the original carbides in the base plate were dissolved.
The effects of a second thermal cycle to various peak temperatures are
summarised in Table IV. The main features common to all specimens
which have received two thermal cycles is that M02C remains in solution
and the cementite particles do not contain any dissolved transition
metal elements.
The size and distribution of the Fe3C particles is found to be a
function of the thermal cycle experienced, Figures 11, 12. Weld
thermal cycles up to the Acj_, t iper the coarse HAZ structure produced
by the first thermal cycle and produce cementite precipitation; the
amount of tempering being a function of the time:temperature history of
the weld thermal cycle. The most dense precipitation occurs for the
TIG 1300, TIG 700 and TIG 1300, TIG 800 thermal cycles, Figures 11, 12.
In addition, the TIG 1300, TIG 800 specimen shows some retransformation
along the prior austenite grain boundaries producing a continuous band
of ferritic grains containing a low carbide density, Figure 11. Second
thermal cycle peak temperatures in the range 900 - 1000°C produce some
cementite precipitation in the form of isolated colonies during heating
at temperatures up to the Ac^. These colonies are subsequently
partially dissolved at temperatures above the Acj^ during the thermal
 360

Peak
<
' '" ^ ' ' | § j | § 436-284°C Transformation range

Hardness
1300 */< .
800 J
'' J
v
-•'<1gM-'' -;••• / ^ > ? 1300
v *-*~.-.-...^-. -•.••vifc: 1 2 0 0
315
416-268C

mmmmm IVlfl '•..' • • ' • " > w ' " V l - i * ' - V •'•'•••<»

w«£5SSS?«S
276°C
mm*mi*& 1300%
600 ^

mmm

900 202
r*;. r; .: : , ; . . i - ^ . ^

,100

Fig. 10. Optical micrographs of simulated HAZ structures after

double thermal cycles.
 361

!»$• ^ 0 0 ^ea^ temperatures

i/ 416-268°C Transformation range

Hardness

'. K .t ;^.;,, 1300

315 *;

'•:•'••: •••;^:::/"/'.€45i

! 130
' ^' °
; '/ 1000
349

'V ..a*

388

10pm

Fig. 11. Carbon extraction replicas from simulated HAZ structures

after double thermal cycles.
 362

tem
Peratures

416-268°C Transformation range

464 Hardness
1300 *t$*/ : :>-<
800 V & v ' V

418 230 c
~ °

Fig. 12. Carbon extraction replicas from simulated HAZ structures

after double thermal cycles.
 363

cycle when the material is austenitic. Thermal cycle peak temperatures

above 1000°C produced complete cementite dissolution, Figures 11, 12.

Weld HAZ structures

A single bead was deposited on a plate of identical composition to

the weld simulation specimens. Welding was carried out at Babcock and
Wilcox, Alliance, using 200 A, 9 V, 3.6 mm/s, and a wire feed speed of
27.5 mm/s with a 0.9 mm wire diameter and 100°C preheat. The range
of HAZ structures from the fusion boundary out to the base plate are
shown in Figure 13. The range of structures correspond closely with
those observed in the single weld thermal cycle simulations. The
structures are, of course, continuously changing due to the steep
thermal gradients in the weld HAZ and it is not possible to achieve the
same spatial resolution in terms of thermal cycle peak temperature
experienced for the weld HAZ as was possible for the weld simulated HAZs.
The maximum coarse HAZ grain size measured using a mean linear
intercept technique at a distance within 0.4 mm of the fusion boundary
was 0.15 - 0.2 mm. The effective heat input for this weld was estimated
to be 0.5 KJ/mm. This grain size can be compared with the coarse HAZ
grain size in the TIG 1300 specimen of 0.097 mm corresponding to an
effective heat input of 1 KJ/mm. The difference in grain size is due
to the different thermal cycles and thermal cycle peak temperatures
experienced by the weld HAZ and the weld simulation specimen. However,
it is clear from Figures 2 and 3 that an increase in grain size from
0.08 mm to 0.1 mm caused by an increase in thermal cycle peak temper-
ature from 1200°C to 1300°C has only a small effect on the CCT curve.
Consequently, the CCT curve determined by the TIG 1300 weld simulations
can be taken as representative of the maximum coarse HAZ grain size in
a weld HAZ. This was confirmed for peak thermal cycle temperatures
up to 137O°C for that region of the CCT curve relating to those cooling
rates appropriate to TIG welds by using a TIG 1370 weld thermal cycle.
The transformation start and finish temperatures on cooling were
measured as 428°C and 250°C respectively. This can be compared with
42O°C and 26O°C for a TIG 1300 thermal cycle with the same heating and
cooling rate.
The weld HAZ hardness was measured at various positions in the HAZs
shown in Figure 13. An approximate comparison between the weld HAZ
hardness and weld simulated HAZ hardness at estimated equivalent posit-
ions is given in Table V.
Carbon extraction replicas were removed from the TIG weld HAZ,
covering a coarse grained bainitic structure at the top to a parent
plate structure at the bottom, Figure 14. In the coarse grained region
the microstructure was carbide free. Beyond the start of the fine HAZ
at position A, a dense band of partially dissolved Fe3C was present,
Figure 15. The very fine grained regions of the HAZ, position B in
Figure 14, contained M3C with progressively increasing amounts of Mo,
Cr and Mn. Spherical M02C particles are also present in this region,
Figure 15. The intercritical HAZ baseplate structure around position C,
 364

Coarse HAZ

A55 Hardness

Medium HAZ

450 Hardness

'$ Fine HAZ

h
? 350 Hardness

V.fine HAZ •
275 M
™ mmmmmmm
Intercritical HAZ
100 nm
225 Hardness

L
" '•'>•' ' ,• - ^ V : f - ** •""''' ',•

x
_ ••"--.-;r--.K-'- ,t:^i'i^.,.,, ,-l)l B a s e plate

Fig. 13. Optical micrographs from weld HAZ structures.

 365

' 5.

10|im 500

Fig. 14. Weld HAZ microstructures in SA5O8 cl II.

 366

HAZ

HAZ

Base plate

Fig. 15. Weld HAZ microstructures in SA508 cl II, from a refined

region A, intercritical region B and a tempered region C.
 367

Table V. Comparison of weld simulated and weld HAZ hardnesses

from bead on plate welds deposited at 200 A, 9 V, 3.6 mm/s travel speed,
27.5 mm/s wire feed speed with 0.9 mm wire diameter and 100°C preheat.

Weld HAZ Hardness Weld Simulation Hardness Grain Size

HvlOKg HvlOKg in HAZ

455 450 - 471 Coarse

450 400 - 450 Medium
350 401 Fine
275 267 Very Fine
225 238 Intercritical

Figure 14, contains fully substituted M3C and spherical and needlelike
M02C particles, Figure 15, similar to those in the baseplate.

Grain Growth

A knowledge of the weld thermal cycle experienced and the result-

ant grain size produced can be used? to estimate f.he grain growth
constant, A. For small initial grain sizes, A is defined by the
equation:-

t*
(i)
£ exp
^r(t7
.At

\
where t is the time at which transformation to austenite is
s complete on heating.

t* is the time at which transformation from austenite to the

T ferritic transformation product is complete during cooling
and all grain growth ceases.
 368

T(t) is the time dependence of the temperature during the thermal

cycle.

Q is tlie apparent activation energy for the grain growth

pp
process, assumed to be 460,000 J/mole.

R is the gas constant.

At is the time step used to integrate numerically the time

dependence of the temperature.

G is the final grain size.

n is a constant and assumed to be 2.73 over the experimental

range of conditions.

N is a constant determined by the HAZ thermal gradient and

has a value of 1 for shallow thermal gradients and ^ 0 . 5
for thermal gradients appropriate to weld HAZs.
Q
This approach has also been used for a high tensile steel and
should prove generally suitable for a wide range of steels.
A weld simulation specimen was subjected to a weld thermal cycle,
TIG 1300, and the thermal cycle temperatures logged automatically every
0.1 s. The value of the numerical integral in equation (1) was calcu-
lated to be 8.27 x 10""16 s anc j the measured final grain size, G, was
0.097 mm. Hence the value of A was calculated to be 2.1 x 10^2 m m 2 7

Tempering

The tempering curves for a range of HAZ structures were determined

by furnace heat treating various weld simulated as-welded microstruct-
ures corresponding to different grain sizes and transformation temper-
atures. Grain sizes corresponding to coarse, fine and "refined coarse"
with transformation temperatures of 420 - 26O°C, 545 - 277°C and
440 - 26O°C, respectively, were used. These transformation temperat-
ures relate to typical TIG weld thermal cycles at a preheat of 100°C.
In addition, coarse,medium and fine grain sizes with transformation
temperatures of 457 - 297°C, 511 - 332°C and 634 - 489°C, respectively,
were used. These transformation temperatures relate to weld thermal
cycles of heat input ^ 6 . 6 KJ/mm and a 100°C preheat. This heat input
was used simply to provide cooling rates for each grain size condition
slow enough to promote the formation of high temperature transformation
products. Each specimen was tempered in a furnace for 0.5 h at
temperatures of 300°C, 400°C, 500°C, 600°C, 650°C and 700°C. Three
separate Vickers hardness determinations were carried out on each
specimen. The results are shown in Table VI. For each of the
microstructural conditions the tempering results were plotted against
 Table VI. Tempering data for various HAZ microstructures in SA5O8 cl II.

Grain Transformation Transformation Vickers Hardness., Hv30Kg

Size Product Temperature As |h at |h at ^h at in at |h at |h at
Range, C Quenched 300°C 400°C 500°C 600°C 65O°C "00°C

Coarse Martensite 420-260 447 450 427 385 360 332 283

Upper Bainite
Fine Lower Bainite 545-277 451 449 437 389 369 323 in
Martensite
Refined Lower Bainite
440-260 433 452 427 383 356 317 266
Coarse Martensite

Coarse Lower Bainite 457-297 429 502 415 366 358 311 268
Martensite 10

Upper Bainite
Medium Lower Bainite 511-332 434 418 409 375 339 311 272
Martensite

Ferrite 278 242 226

Fine 634-489 293 328 291 261
Upper Bainite
 370

s
o

300C
>1eLded
a
JZ
200
Grain size
o HAZ
0.015-0.1 mm; complete carbide
100 dissolution
0.015mm; incomplete carbide
dissolution
20 000 30000 40000
Larson-Miller Parameter
Fig. 16. Tempering curves in as-welded SA5O8 cl II HAZs.

the Larson-Miller parameter defined as

(1.8T + 492) (20 + log t) (2)

where T is the temperature in C and t is the cime in hours. The

choice of this parameter is discussed later under Discussion. The
tempering curves for these microstructural conditions which produce
combinations of martensite, lower bainite and/or upper bainite with
various prior austenite grain sizes, are almost identical, Figure 16.
The exception is the fine grained material with the highest transform-
ation temperature 634 - 489°C. This has produced a predominantly fine
grained ferrite transformation product. Consequently, the measured
hardnesses are correspondingly lower at all times and temperatures,
 371

although the curve retains the same general shape.

The hardness results for the lower temperature transformation
products have been fitted to a parabolic equation using a regression
technique. The hardness can be calculated from the Larson-Miller
parameter (LMP) over the range 20,000 - 35,000 by the equation:-

(Hv30Kg) = 243 + 0.0228 (LMP) - 0.OO00OO63 (LMP) 2 (3)

DISCUSSION

A detailed discussion of the work presented and a comparison with

other data from the literature will be left for a later report. The
results described have shown that the microstructure and hardness of
weld simulated SA5O8 cl II HAZs are sensitive to the weld thermal cycle
experienced. In the following sections, some explanations for the
effects observed and their implications for weld HAZs are discussed.

Microstructure

Single TIG weld thermal cycles with high peak temperatures produced
coarse grained martensite or coarse grained martensite/lower bainite
mixtures which correspond closely to the microstructures observed at
the fusion boundary of the weld HAZs. Weld thermal cycles with lower
peak temperatures produce finer grained microstructures, the prior
austenite grain size decreasing with peak temperature. Thermal cycles
with peak temperatures in the range ^ 800 - 900°C produce various degrees
of reaustenitisation, resulting in intercritical structures which are
representative of those structures present at the extremities of weld
HAZs at the HAZ/base metal interface.
The decrease in grain size with thermal cycle peak temperature
causes a shift in the CCT curve which promotes the formation of higher
temperature transformation products at a given cooling rate. A given
weld HAZ could contain martensitic transformation products at the
fusion boundary changing progressively to lower bainite, upper bainite
and ferrite as the grain size decreases at various distances away from
the fusion boundary. Even relatively simple single cycle weld HAZs
will contain a range of grain size and transformation structures in
this composition at typical welding cooling rates.
Multipass weld HAZs are more complex. They experience multiple
weld thermal cycles with peak temperatures ranging between the melting
point and the preheat temperature. These thermal cycles can leave
the structure basically unaltered, change the transformation product or
temper the HAZ structure, depending on the peak temperature experienced
and the cooling rate. The two basic mechanisms responsible for these
changes are carbide dissolution or precipitation and grain growth while
the material is austenitic.
Austenite is grain boundary nucleated on heating so that the grain
 372

growth process is duplicated for each weld thermal cycle as soon as

transformation to austenite on heating is complete. Grain growth
usually stops on transformation to a ferritic product on cooling. The
presence of a stable precipitate distribution in the austenite can
inhibit grain growth and some inhibition during the early stages of
carbide dissolution on heating; may occur. However, the carbide diss-
olution is rapid and almost complete by 1100°C, so that the final prior
austenite grain size achieved is largely characteristic of the last
thermal cycle experienced. The effects of previous weld thermal cycles
on the precipitate distribution only have a minor effect on the subse-
quent grain growth kinetics. The transformation product produced after
a given weld thermal cycle is almost independent of any previous thermal
cycles experienced. This is because the last thermal cycle determines
the prior austenite grain size, amount of carbide dissolution and the
cooling rate, which, in turn, largely determine the transformation
product. The transformation temperature on cooling can readily
change from 430°C for the coarse grained condition to 58O°C for the
fine grained condition across a multipass weld HAZ welded with typical
TIG welding conditions, so that the range of transformation products and
their distribution will be complex.

Carbide dissolution ar,i precipitation

The as-received base plate consists of a dense dispersion of M3C

containing 86.6% Fe, 6.2% Mn, 5.7% Cr, 1.5% Mo and M02C. On heating,
during a weld thermal cycle, these particles coarsen and dissolve
depending on the weld thermal cycle peak temperature. The cementite
can reprecipitate during the cooling portion of the weld thermal cycle
but the Fe3C reprecipitated does not contain dissolved transition
elements. It is suggested that during dissolution the carbon moves
rapidly away from the precipitate site followed by the transition
elements. During cooling there is insufficient time for the transition
elements to diffuse back to the cementite nucleus and precipitation of
Fe3C occurs directly.
The transition metals are taken up by the cementite during subse-
quent stress relief heat treatments, the take-up rate and final equil-
ibrium composition depending on base plate composition and stress re-
lieving time and temperature. These results will be discussed in a
future report which relates carbide morphology to the cementite compo-
sition for a number of PWR steels.
A comparison has been made between the carbide morphology across
a weld HAZ and in single cycle simulated HAZs. Close to the fusion
boundary the coarse grained bainitic microstructure of the weld is
carbide free and this structure is closely reproduced using peak cycles
of either 1300°C or 1200°C, the exact peak temperature depending on the
grain size required. In the fine grained regions of the weld, dense,
partially dissolved, Fe3C is present, and there are no M02C carbides.
Simulations using peak temperatures of 1100°C and 1000°C reproduced
the partially dissolved Fe3C as observed at position A, Figure 14, but
there are small differences in carbide densities between weld and sim-
ulated HAZ specimens, which may be due to segregation in the base plate.
 373

The very fine HAZ region, position B, Figure 14, contains coarse
M3C and spherical M02C particles. This dispersion is reproduced by
simulation to peak temperatures in the range 800°C - 900°C.
Generally the carbide morphology, density and type varies across
the weld HAZ as a result of the weld thermal cycle experienced. The
particular carbide morphology, density and type present in a given HAZ
region can be accurately reproduced by careful selection of the approp-
riate weld thermal cycle and weld thermal cycle peak temperature.

Hardness

The hardness produced in the HAZ depends on the separate contrib-

utions from the transformation product, those elements taken into solid
solution, the carbide distribution, grain size and the dislocation
density. It is observed that changes in grain size from 0.1 mm to
0.015 mm, changes in transformation temperature range from 420 - 26O°C
to 545 - 277°C and changes in carbide distribution do not radically
affect the as-welded hardness. For simple single bead weld HAZs this
is expected to be in the range 400 - 470 HvlOKg for those regions which
have experienced temperatures in excess of 1000°C. The outer edge of
the HAZ is softer, typically 225 - 350 HvlOKg depending on the exact
position.
Table V shows a comparison of measured weld simulated and weld HAZ
hardnesses from different regions of the HAZ corresponding to coarse,
medium, fine, very fine and intercritical prior austenite grain sizes.
The values are in reasonable agreement which reflects the accurate weld
simulation of both the transformation product and the precipitate
distribution present in weld HAZs. Unambiguous comparisons of weld
HAZ and simulated HAZ structures are difficult due to the rapid varia-
tion in structure across weld HAZs caused by the steep thermal gradients
present during welding.
Multipass weld HAZs which have experienced peak temperatures in
excess of 1000°C, but which have not experienced any significant temper-
ing effects from successive thermal cycles, are typically in the range
430 - 460 HvlOKg. In those situations where no subsequent tempering
occurs, the HAZ hardness produced is largely representative of the last
thermal cycle experienced. This is because each successive weld ther-
mal cycle renucleates austenite on heating and the extent of grain
growth and precipitate dissolution and reprecipitation determines the
HAZ structure and hardness.

Tempering in multipass weld deposits

The HAZ hardness of a multipass weld deposit is determined by the

particular combination of weld thermal cycles experienced at a given
position within the weld HAZ. These cycles can retransform various
regions, if the peak temperature experienced is above the dynamic Aci,
producing hard, as-welded transformation structures. These structures
can be progressively tempered by the action of various lower peak
 374

temperature weld thermal cycles, provided that the peak temperature is

less than the dynamic Ac^.
HAZ hardnesses in multipass weld deposits may be calculated, in
principle, from a knowledge of the weld thermal cycle experienced, the
appropriate CCT curves and tempering data. These data have been
measured for SA508 cl II in this investigation and an attempt has been
made to apply this data to the prediction of hardness in weld simulated
HAZ structures which have experienced known weld thermal cycles. The
hardness variation as a function of time and temperature for the coarse
grained and intermediate grain sized material was independent of the
transformation temperature and could be represented by equation (3).
The tempering effects during a weld thermal cycle can be estimated from
equation (3) if the Larson-Miller parameter can be obtained for the
weld thermal cycle. This is possible by numerical integration. This
effectively divides the weld thermal cycle into a large number of
discontinuous isothermal steps such that the time interval for each
step is short with respect to the weld thermal cycle. The Larson-
Miller parameter for the arbitrary temperature and time interval is:-

(1.8 T(t) + 492) (20 + log At) (4)

where T(t) represents the time dependence of the weld thermal cycle and
At is the time step for numerical integration, taken as 0.1 s. However,
it is not possible to summate individual Larson-Miller parameters for
each time interval. It is necessary to select an arbitrary normalising
temperature, chosen as 800°C in this case. Each Larson-Miller parameter
is converted into an equivalent time at 800°C and the equivalent times
summated for each isothermal step across the weld thermal cycle. The
equivalent time at 800°C for the weld thermal cycle is given by:-

- 20] (5)

The constant, 1932, arises from the choice of 800°C as the normalising
temperature. The Larson-Miller parameter for the weld thermal cycle
is then obtained from the equation

LMP = (1932) (20 + log £ £ t® 0 0 ) (6)

The tempering effects can then be estimated by substituting the calcu-

lated Larson-Miller value for the weld thermal cycle into equation (3).
Hardness values for weld simulated HAZs which had experienced
TIG 1300, TIG 800; TIG 1300, TIG ^00 and TIG 1300, TIG 600 have been
calculated using the above procedure. The values are 329, 370 and
404 Hv30Kg respectively. The corresponding measured values after weld
simulation are 315 ± 24, 349 ± 18 and 388 ± 8 HvlOKg. These results
show a small systematic difference which is not expected to be due to
the difference in load used to measure the Vickers hardness. It is
thought to be due to the small compositional differences between those
specimens used for the tempering experiments and those used for weld
simulation. The specimens used for tempering should have a slightly
 375

prior austenite grain size, precipitate distribution and transformation

products present in single and multipass weld deposits in SA5O8 cl II
material.
4. The structures and hardness produced in SA5O8 cl II weld HAZs can
be understood in terms of the weld thermal cycles experienced.
5. The effect of time and temperature on as-welded HAZ hardness in
SA5O8 cl II can be calculated for weld thermal cycles and for conven-
tional post weld heat treatment using standard tempering data.
6. It is theoretically possible to achieve as-welded HAZ hardnesses
in SA5O8 cl II which approach the level obtained after post weld heat
treatment. This may have implications for the control of HAZ hardness
during mechanised welding.

REFERENCES

1. P.J.Alberry, R.R.L. Brunnstrom and K.E.Jones. (1981). A computer

model for the prediction of heat affected zone structures in
mechanised tungsten inert gas multipass weld deposits. CEGB Report
RD/M/1158R81.
2. P.J.Alberry and R.D. Nicholson. (1981). A heat affected zone com-
puter comparison of ASME III and CEGB two layer refinement procedures
for BWR and PWR cladding repairs. CEGB Report RD/M/1156R81.
3. P.S. Ayres. (1978). Repair welding of heavy section steel nozzles.
2nd Int. Nuclear Power Conf., Cologne, September 28th.
4. J. Myers. (1972). Weld simulation in a study of stress relief crack-
ing in a CrMoV steel. Welding Institute Seminar on weld thermal
simulators for research and problem solving.
5. P.J.Alberry and W.K.C.Jones. (1977). The structure and hardness of
|CrMoV and 2CrMo simulated heat affected zones. Metals Tech.,
December, p557-566.
6. T.Gladman. (1966). On the theory of the effect of precipitate
particles on grain growth in metals. Proc. Roy. Soc., 249A, p298-3O9.
7. P.J. Alberry, B. Chew and W.K.C.Jones. (1977). Prior austenite
grain growth in the heat affected zone of a |CrMoV steel. Metals
Tech., June, p317-325.
8. H. Ikawa, H. Oshige, S.Noi, H.Date and K. Uchikawa. (1978). Relation
between welding conditions and grain size in weld HAZ. Trans. Jap.
Weld S o c , £ , No.l, p47-51.
9. T.Yada and K. Ishikawa. (1977). Experimental evaluation of half
bead weld repairment for nuclear pressure vessels. Fracture Mechanics
and Technology, Proc. Int.Conf., Hong Kong, March 21-25, p527-537.
 376

higher hardenability as indicated by chemical analysis, Table I. The

specimens used for tempering containing 0.39 wt% Cr and 0.31 wt% Si
compared to 0.28 wt% Cr and 0.17 wt% Si for the weld simulation specimens.
The calculation of HAZ hardness has been incorporated into a weld
HAZ computer model developed at Marchwood Engineering Laboratories and
enables detailed multipass weldments and HAZ hardness traverses to be
calculated at any specified position within a multipass HAZ. The
calculations have recently been extended to cover the effects of post
weld heat treatment using the logic described previously. The model
calculates the tempering effects due to adjacent weld beads in the
first layer and continues to summate the tempering effects of those
thermal cycles which result from the deposition of successive layers.
The reduced as-welded peak HAZ hardness which is achievable for most
practical welding situations is calculated to be in the range 300 -
320 Hv3OKg. The peak HAZ hardness after a conventional post weld heat
treatment can also be estimated from the tempering data via the Larson-
Miller parameter. The calculated peak HAZ hardnesses after 5 and
17 h at 615°C are 308 and 291 Hv30Kg, respectively. This may be com-
pared with an experimentally measured^ peak HAZ hardness value of
272 Hv30Kg for SA5O8 cl II after 17 h at 615°C ± 20°C, whose composition
was reported to be 0.18 C; 0.77 Mn; 0.033 Si; 0.06 P; 0.011 S;
0.8 Ni; 0.36 Cr and 0.64 Mo. The carbon level is 0.05 wt% lower than
the composition used for the SA5O8 cl II tempering data in this work
which may account for the slightly lower measured HAZ hardness after
17 h at 615°C.
These results demonstrate the feasibility of a welding technique
where successive welding passes are used to control the HAZ structure
and hardness. It seems possible that HAZ hardnesses can be achieved
which are equivalent to those achieved during conventional post weld
heat treatment. If the practical viability of these techniques can
be demonstrated then this may provide some confidence that the structure
and properties of the HAZs of non post weld heat treated weldments can
be left in an acceptable condition metallurgically. However, it is
unlikely that such techniques will have a major effect on the as-welded
residual stress distribution and further work is required to assess the
fitness for purpose of weld repair situations which do not involve post
weld heat treatment.

CONCLUSIONS

1. The dominant precipitates in the base plate of SA5O8 cl II are

molybdenum carbide and cementite containing chromium, manganese and
molybdenum. These are rapidly taken into solution during a weld thermal
cycle. Dissolution on heating is complete by ^ 1100°C. Reprecipit-
ation of cementite occurs on cooling. The cementite is present as a
fine dispersion and does not contain any dissolved chromium, manganese
or molybdenum.
2. The continuous cooling transformation products in HAZs of welded
SA5O8 cl II depend strongly on prior austenite grain size and cooling
rate.
3. Weld thermal simulation techniques can accurately reproduce the
 377

ACKNOWLEDGEMENTS

Material and funds for part of this work were supplied by Babcock
and Wilcox, Alliance, Ohio, as part of an EPRI Contract No. 1236.
Mr. H. Fletcher, Marchwood Engineering Laboratories, carried out the
experimental work.
The work was carried out at Marchwood Engineering Laboratories and
the paper is published by permission of the Central Electricity
Generating Board, Babcock and Wilcox, Alliance and EPRI.
 EFFECT OF WELD RESIDUAL STRESSES ON INTERGRANULAR STRESS CORROSION
CRACKING OF TYPE 304 STAINLESS STEEL BOILING WATER REACTOR PIPING

BY

JOSEPH C. DANKO
ELECTRIC POWER RESEARCH INSTITUTE
PALO ALTO, CALIFORNIA 94303

ABSTRACT

The role of weld residual stresses on the intergranular stress

corrosion cracking of Type AISI 304 stainless steel piping in boiling
water reactors was investigated. Pipes ranging in size from 10 cm
(4 inch) to 66 cm (26 inch) in diameter were fabricated using field
welding procedures. X-ray and stress relief methods were used to
measure the surface and throughwall residual stresses. Peak axial
residual stresses on the internal surface were found to be the highest
for the 10-cm pipe, intermediate for the 25-cra, and the lowest for the
66-cm pipe. Throughwall residual stress measurements showed axial
tensile stresses from the Inner surface to approximately 50 to 75% of
the wall thickness for the 10-cm pipe whereas the stresses were strongly
compressive at a depth of approximately 25% of the wall thickness for
the 66-cm pipe. The peak axial residual stresses exceeded the yield
stress of the 304 stainless steel base material over the range of
reactor operating temperatures and were found to be a major contributor
to stress corrosion cracking.

INTRODUCTION

Incidents of intergranular stress corrosion cracking (IGSCC) of

welded Type 304 stainless steel (304 SS) piping systems In boiling water
reactors (BWRs) have been occuring since the introduction of the first
commercial unit. A total of 272 worldwide incidents have been
reported; these are listed by the various pipeline types in Table I.
Most of the incidents have occurred in three pipeline systems: reclr-
culation bypass, core spray, and reactor water cleanup lines.

TABLE I
WORLDWIDE IGSCC INCIDENTS BY PIPE LINE TYPE

Pipe Line Type Total Incidents

Recirculation bypass 50
Core spray 82
Control rod drive return 3
Reactor water clean up 67
Large recirculation _>_ 25 cm (10 inches) 18
Small 2. 7.6 cm (3 inches) 16
Other _36_
TOTAL 272

379
 380

Some of these pipelines are shown in Figure 1, a schematic of a BWR

recirculating piping system. Pipe diameters of the 304 SS lines vary
from 10 cm (4 inches) to 71 cm (28 inches) depending on which of the six
BWR models is being considered. In a typical recirculation piping
system, the number of welded joints may vary from 150 to 200 depending
on the particular BWR model.

INLET
VALVE

RCCIRCULCTION
INLCT

•tCCIRCULATIOH
PUMP

SUCTION VALVE

Figure 1. Schematic of a boiling water reactor recirculation

piping system.

IGSCC incidents have occurred in pipe sizes ranging from 8 cm

(3 inches) to 61 cm (24 inches) in diameter. The frequency of cracking
as a function of pipe diameter is shown in Figure 2. Cracking has been
detected by the presence of small leaks, by ultrasonic inspection, and
dye penetrant examination.

The IGSCC incidents have adversely affected plant reliability,

availability, economics, and the radiation exposure of skilled personnel
involved in the inspection and repair operations. However, pipe crack-
ing is not a safety issue that represents a hazard to the safety of the
public. Of the 272 worldwide incidents, there has never been an occur-
rence of a pipe severance. Small leaks or cracks have been reported.
The field failure data base was used in a Pipe integrity analysis that
resulted in a "leak-before-break" concept.^ ' In this concept, a leak
will occur before the tough ductile austenitic stainless steel will lose
its structural.integrity. A special Nuclear Regulatory Commission Pipe
Study Group ' also adopted this position.
 381

100

90 TOTAL INCIDENTS
WORLD WIDE THRU
1981 - 272
80

70

</i 60

2 50
•i
Z
a
2 30
. 20

s 10 —
0

0
0 2
5
"t
10
6
15
8
20
10
25
li12
31
1')
36
I
If.
k)
I I20
18
M6 51
22
56
24
61
26
66
28
71
CINCH)
CCM)

3OT SS PIPE DIAMETER

Figure 2. Frequency of pipe cracking incidents as a function

of pipe diameter.

As part of the pipe integrity analysis, the design and stresses on

the piping systems were reexamined. Analysis of the piping systems of
operating plants were performed in accordance with ANSI Code for
Pressure Piping, B31.1. For new plants, the ASME Boiler Code,
Section III, "Nuclear Power Plant Components" is used. The analysis
showed that the piping systems did indeed meet the code requirements.
However, the codes did not include the weld residual stresses in the
piping design analysis since they were not considered to play an impor-
tant role in the normal modes of failure because of plastic shakedown.

In response to this important utility problem, the Electric Power

Research Institute initiated in 1975 a number of research and develop-
ment projects. Some of these included investigations on the effect of
weld residual stresses on the IGSCC of the 304 SS pipes; the results of
these projects provide the basis of this paper.

EXPERIMENTAL PROCEDURE

Several laboratories were involved in the investigations. The

experimental procedures used between the laboratories were very similar
and in some cases pipes were exchanged between the organizations.

Pipe Material: The 304 SS pipes were ordered to ASTM specifica-

tions A376 and A358 for the small-diameter seamless and the large-
diameter plate rolled and welded pipe, respectively. A number of heats
of the different size pipes in the mill-annealed condition were used in
the investigations.
 382

Welding Procedure; The welding parameters used for the 304 SS butt
girth welds were based on field welding specifications employed by the
Phillip Getscow Company. An extended land type of weld bevel was used
with Grinnel Type 308 SS consumable inserts. The internal diameters of
all pipe sections were machined counterbored to dimension. Welding was
performed in the 2G position and consisted of: gas-tungsten-arc welding
(GTAW) of the Grinnel insert; GTAW of the second layer using bare filler
wire (ER-308); and shielded-raetal-arc welding (SMAW) of subsequent
layers using coated electrodes of (E308-16). Weld heat input varied
with the layer and pipe diameter.

Heat Sink Welding (HSW): This welding process used a water spray
or flowing water on the inside of the pipe after the root pass and the
first and second layers were deposited. The reason for using this
method was to reduce the level of sensitization of the HAZ in the
304 SS.

Residual Stress Measurements: Two methods for the measurement of

residual stresses were used: X-rays and stress relief. X-ray residual
measurements ' were used to determine the internal surface and
subsurface stresses near the pipe weld. This method is limited to
depths of approximately 0.001 cm (0.0005 inch). The accuracy of the
X-ray method used in the pipe welds was +35 to 50 MPa (+5 to 7 ksi).

The stress relief method measures ' the bulk residual stresses in
the region of the surface of the body or through the wall. This method
involves attaching strain gages to the outside and inside surface of the
pipe weldment and then carefully sectioning the weldment. Location of
the strain gages on a pipe weldment is shown in Figure 3.

To obtain throughwall residual stress the pipe segments were cut

into thin sections, new strain .gages attached and further sectioning
performed. Calculations of residual stresses from the measured strains
were made using isotopic plane-stress, stress-strain relations. During
the sectioning process, there was a redistribution of the stresses and
this was accounted for in the analysis. Accuracy of this method for
throughwall stress measurements was approximately +35 MPa (5 k s i ) .

RESULTS
(8)
The first measurements of residual stresses were performed on
butt welded 304 SS pipes of 10-cm (4-inch), 25-cm (10-inch), and 66-cm
(26-inch) diameter. Strain gage measurements on the internal surface of
the pipes showed considerable azimuthal variations in both the axial and
hoop stresses. These variations were more prominent in the 10-cm and
25-cm pipes and diminished in the 66-cm pipe. This is illustrated in
Figures 4 and 5 for the 25-cm and 66-cm pipe. Axial residual stresses
as a function of the distance from the weld centerline were determined
by the combination of X-ray and strain gage measurements and are shown
in Figure 6 for the three pipe sizes. The peak tensile surface residual
 383

ROSETTE STRAIN GAGE

315'

PIECE NO 1

PIECE NO 2

STRAIN GAGE ARRANGEMENT OF THE INSIDE SURFACE - 2254>gr M AZIMUTH

NOTES 111 THE OUTSIDE GAGES WE RE ON RADIAL PLANES ABOVE GAGES 1. 2. 3.6. 7. AND 8 - 225 dtgrn AZIMUTH
THE OUTSIDE GAGES WERE DISPLACED AXIALLY RELATIVE TO THE INSIDE GAGES SO AS TO REMAIN
THE SAME DISTANCE FROM THE FILLET EDGE

(21 FOR SYMMETRY. 6 GAGES WERE LOCATED AT 315. 45, AND 135 dtgrtts -OUTSIDE AND
INSIDE PAIRS AT POSITIONS 1.3. AND 6
131 33 GAGES TOTAL

Figure 3. Strain gage arrangement on pipe weld.

Figure 4. Variations of axial surface residual stresses on

25.4 (10 inch) 304 SS welded pipe.
 384

Figure 5. Variations of axial surface residual stresses on 66 cm

(26 inch) 304 SS welded pipe.

stresses are pipe size dependent with the highest stresses in the 10-cm
(4-inch), intermediate in the 25-cm (10-inch), and the lowest in the
66-cm (26-inch) diameter pipes. These peak tensile stresses occurred in
the HAZ of the weldment. The axial residual stresses also showed a
steep gradient as a function of distance from the weld centerline and
were generally axisymmetrical. At approximately 0.7 cm (0.3 inch) from
the weld centerline the axial residual stresses changed from tensile to
compressive and continued to become increasingly more compressive,
reaching levels of approximately -500 MPa (-70 ksi) at I cm (0.4 inch)
from the weld centerline for the three different pipe sizes •

In the field fabrication, the weld counterbores were machined, and

in some cases ground. These operations cold work the surface; there-
fore, X-ray residual stress measurements were performed on such surfaces
to determine the state of residual stresses. The results are reported
in Table II for a 10-cm (4-inch) Schedule 80 304 SS pipe.

High surface tensile residual stresses were produced by both the

machining and grinding operations with the higher stresses oriented in
the circumferential direction. The as-received pipe had small compres-
sive residual stresses on the inside surface.
 385

Distance From Weld Centerline (in)

0.6 0.4 0.2 0.2 0.4

Note Total longitudinal residual

MTMses StT*ts as a 80
500 function distance from weld

£ 400 60 _

| 300
40 o
,1 200
10 16 cm diameter
20
100

100
20
_ 20° 0
CO 40 |
il 300

60 |
I
t 500
o 80
u
600
25.4 cm
700 100

800
1.0 0.5 0.5 10 1.5
Distance From Weld Centerline (cm)

Figure 6. Peak axial tensile residual stresses in welded 304 SS

pipes.

An experiment was performed to determine the relaxation of these

surface stresses. Flat coupons of 304 SS were abusively ground and then
heat treated at 300°C (570°F) in a vacuum for times up to 336 hours.
This temperature was selected because it is the nominal BWR steady-state
operating condition- X-ray measurements on these coupons showed essen-
tially no relaxation of the tensile surface residual stresses.

In order to determine the effect of the welding process on the

yield stress of the weldment, miniature tensile specimens were machined
from the 66-cm (26-inch) pipe weld. Specimens were tested at 290 C
(550°F) and the measured yield strength profile across the weldment is
shown in Figure 7. The yield strength in the HAZ was significantly
higher than that of the base material and there was a smooth transition
in the yield strength from the weld metal to the base material.

Other measurements of the weld residual stresses were performed

by another laboratory using pipe weldments prepared from the same heats
of 304 SS and using the same weld procedures. Pipe weldraents of 10-cra
(4-inch) and 66-cm (26-inch) diameter were u«ed. Welds of 10 cm
(4 inch) and 25 cm (10 inch) removed from field failures were also
included.
 386

TABLE II

MAXIMUM SURFACE RESIDUAL STRESS IN 10cm (A INCH) 304 SS PIPE

DUE TO SURFACE TREATMENTS

Maximum Surface Residual Stress

MPa (ksi)

Surface
Sample Preparation Circumferential Axial

Inside Surface Machined 550 (80) T 70 (10) T

Inside Surface Light Grind 550 (80) T 0 (0)

(Hand Grinder)

Inside Surface Heavy Grind 550 (80) T 120 (17) T

(Hand Grinder)

Inside Surface Shot Peened 290 (42) C

Inside Surface As Received 105 (15) C

T « tension; C = compression

Grinding in Circumferential Direction

Strain gage measurements were made on three different butt welds of

10-cm (4-inch) Schedule 80 304 SS pipes. The bulk internal surface
residual stresses showed the same azimuthal variation in axial and hoop
stress as reported previously (Figures 4 and 5) and the magnitude of the
peak stresses was essentially the same. Throughwall distribution of the
residual stresses of the three pipe welds revealed similar residual
stress patterns in that the residual stresses were predominately tensile
through the whole wall near the weld fusion line, a region where
sensitization would be expected to be most severe.

The throughwall distribution of residual stresses of the 66-cm

(26-inch) pipe at a distance of 3.2 mm (0.13 inch) from the weld center-
line is illustrated in Figure 8. Residual stresses varied from moder-
ately tensile on the internal surface to compression at approximately
12 percent of the wall thickness and reached a peak at approximately
25 percent of the wall thickness and remained compressive to 75 percent
of the wall thickness before going tensile again. In comparison to the
throughwall residual stresses of the 25-cm pipes, the magnitude of the
nonlinear stresses and the region of compressive residual stresses were
significantly different.
 387

Distance from Weld Fusion Lmt, mm

0 z_ A __ _s s
1 1 1 1 ! 1 1
weld Weld 1 Bose —
metal HAZ marenal

1 1 1 | 1
0 01 02 03
Distance from Weld Fusion Line, inch

Figure 7. Yield strength profile across the weldment of a

66 cm (26 inch) 304 SS pipe.

STRESS IMPol STRESS IMPol

00-276 -136 C p 138 276 276-138 138 276

/l
1
w I 1 ' 1 ' 1

1/2 - -

VA - -

m I 1 11 h•4-J |
-40-30-20-10 0 10 20 30 10 -40-30-20-10 O 10 20 30 40
STRESS (ksi) STRESS (ksi)

(e)

Figure 8. Throughwall distribution of axial residual stresses on

each side of weld 3 ram (0.13 inch) from the weld fusion
line in 66 cm (26 inch) pipe.
The residual stress measurements on the as-welded 304 SS pipes
between the two laboratories were in good agreement. Peak surface axial
tensile residual stresses for the 10-cm (4-inch) and the 66-cm (26-inch)
pipes were 352 MPa (51 ksi) versus 380 MPa (55 ksi) and 193 MPa (28 ksi)
versus 240 MPa (35 ksi), respectively. The 25-cm (10-inch) pipe
 388

measurements by on** laboratory only showed intermediate values between

the small- and large-diameter pipes of 345 MPa (50 ksi).

Residual stress measurements were made on 25-cm (10-inch) and a

61-cm (24-inch) 304 SS pipe removed from operating plants. The results
showed that the peak surface axial stresses on the 25-cm pipe were
430 MPa (60 ksi). Thts compares very favorably with the values of
345 MPa (50 ksi) for as-welded 25-cm (10-inch) pipe. For the 61-cm pipe
an uncorrected value of the peak axial residual stress was 225 MPa
(32 ksi). Throughwall residual stress distribution of this pipe shown
in Figure 9 closely resembles that of the 66-cm (26-inch) pipe
(Figure 8).

Heat sink welding of a 10-cm (4-inch) Schedule 80 pipe of 304 SS

with flowing water produced compressive residual stresses on the inter-
nal surface of -400 MPa (-58 ksi). This compares to a value of 380 MPa
(55 ksi) axial tensile stresses for a conventional welded pipe. Mea-
surements on 25 cm (10 inch), 41 cm (16 inch), and 61 cm (24 inch) pipes
also showed axial compressive residual stresses on the internal surface
but the magnitude decreased as the pipe diameter increased. '
Figure 10 shows the residual stress patterns for a 25-cm (10~inch) pipe

MP
ion 2oo*° sop ° STRESS
1 I 1 I— »»'

-30 -20 -K) 20 V 40

Figure 9. Throughwall distribution of axial residual stresses

in the pipe 61 cm (24 inch) removed from service,
a. 3 mm (0.13 inch) b. 8 mm (0.31 inch) from weld
fusion line
 389

TOTAL A IAI R ESIDUAL STKSS

to a P.F1CE
DIA HETEIt PIPCS
TYPE 3 04 SS
•i L ! _ 3 0 4 S S
304S_S_
—ik. * * TENSIOI
400

I
50
300
40
1
200 SO
20
100
. 1 , 1 1 , 1 .
oa 04 04 01 (ml
20 10 10 20 lc>! - 10
100
DISTANCE 20
FfOM » C l
2C0 SO
"cOHWENTIOMt"^
JUtlO 40
JO0

- viI
SO
400
-
1 I (0

400 70
•0
600
K
TOO ~

aoo
HEAT
SINK
VELO
*

I
\\ \ :
100
IK)
120
•00

1000 : J
COMPRESSION H P H
a

i
\ - ISO

CONPdtSSION
140

Figure 10. Total axial internal surface residual stresses for heat
sink welded and normal welded 25 cm (10 inch) 304 SS
pipes.

welded by conventional and heat sink methods. This illustrates the

considerable alteration produced by heat sink welding.

DISCUSSION

The measurements of weld residue: stresses in the various diameters

of 304 SS pipe provided important information on the elucidation of
IGSCC in the piping systems in boiling water reactors. These include
the azimuthal variations of residual stresses, the peak stresses as a
function of pipe diameter, the throughwall distribution of residual
stresses, relaxation of residual stresses and the use of special
processes that altered the weld residual stresses.

Azirauthal variations in both the axial and circumferential tensile

residual stresses were measured on the internal surface of pipes ranging
from 10 cm (4 inch) to 66 cm (26 inch) in diameter (Figures 4, 5, and
6 ) . The magnitude of these tensile stresses was more pronounced for the
 390

10 cm (4 inch) and the 25 cm (10 inch) and was much less in the 66-cm
(26-inch) pipe. An attempt was made to correlate these variations with
the field welding procedures used in preparation of the girth welds.
Examination of such factors as weld start and stop positions, the weld-
ing parameters and variations in the pipe wall thickness failed to
produce any correlation. The variations in the axial tensile residual
stresses may explain the cracking patterns observed in some of the field
failures. For the small-diameter 304 SS pipes (10 cm - 4 inch), leaks
were generally very localized and the circumferential cracks caused by
the axial stresses were very intermittent. On the other hand, cracking
in the 61-cm (24-cm) pipe was all circumferential. This behavior may be
related to the magnitude and oscillations of the axial residual stresses
in the 10-cm pipe compared to the large diameter pipe. The residual
stresses in both cases contribute to the driving force for the
initiation and propagation of IGSCC.

The peak axial tensile stresses on the internal surface also showed
a pipe size dependency (Figure 6 ) . Axial surface tensile stresses are
produced by the radial shrinkage that cause throughwall bending near the
weld. The bending moment in the larger-diameter pipe is less because of
the stiffness and lower temperatures during the welding operation com-
pared to the small-diameter pipes. The importance of these axial
residual stresses to the IGSCC problem may be determined by examining
the incidents as a function of pipe size (Figure 2 ) . Most of the
incidents have occurred in the smaller-diameter pipes where the surface
tensile weld residual stresses are the greatest. In the large-diameter
(61-cm) 304 SS pipes, only one reactor after 10 years of operation
experienced cracking and none of those cracks were throughwall. When
the peak weld residual stresses are added to those calculated in the
code design analysis, the yield stress of the weldment is exceeded. The
magnitude of these combined stresses decreases with increasing pipe size
and correlates well with the pattern of field incidents.

Measurements of the throughwall weld residual stress distribution

show a definite relationship to the 304 SS pipe diameter. For the 10-cm
(4-inch) diameter 304 SS pipe the axial weld residual stresses in the
region of the HAZ showed tensile stresses throughout a large fraction of
the wall thickness. In contrast, the 66-cm (26-lnch) pipe showed an
Initial axial tensile stress on the internal surface which became com-
pressive at approximately 15% of the wall and then become very compres-
sive reaching a peak at approximately 50% of the wall thickness before
decreasing and again changing to tensile stresses (Figure 8 ) . There is
a much smoother transition in the throughwall residual stress distri-
bution for the 66-cm (26-inch) compared to the 10-cm (4-inch) pipe.
Thus, the residual axial tensile stresses provide a greater driving
force for crack propagation In the smaller-diameter pipes. This is not
inconsistent with the results of the field incidents which show a
greater propensity for IGSCC in the small-diameter pipes.

The existence of the peak axial residual stresses that were well
ahcve the yield stress of the base material may appear to be an
 391

anomaly. However, the plastic deformation in the HAZ that results from
the welding operation significantly increases the yield stress in this
area. As shown in Figure 7, the peak yield stress in the HAZ is twice
that of the base material at the operating temperature of 282°C
(55O°F). At room temperature, the yield stress will, of course, be much
higher and will be at the levels of the measured residual stresses. The
relaxation of the residual stresses at the reactor operating temperature
is another important factor in the IGSCC behavior of the welded pipes.
Measurements made on coupons at 304 SS with abusively ground surface
before and after heat treatment at 300°C (570°F) showed essentially no
change in the residual stresses after 336 hours. These results showed
no relaxation and residual stresses would be expected to remain at
operating conditions.

Additional data supporting the permanence of residual stresses was

obtained from measurements made on welded pipes removed in the field.
Peak residual stresses from a 25-cm (10-inch) pipe were 430 MPa
(60 ksi); this compares to residual stresses of 345 MPa (50 ksi) for the
as-welded pipe. The surface and throughwall residual stresses of a
61-cm (24-inch) pipe removed from service after 10 years of operation
(Figure 9) compares favorably with that of an as-welded pipe of 66 cm
(26 inch) in diameter (Figure 8 ) . This comparison is justified since
the residual stresses measured in a number of as-welded 304 SS pipes
ranging in diameter from 51 cm (20 inch) to 66 cm (26 inch) show very
similar stress patterns. These results from the field pipe show a
permanence of the weld residual stresses. They do not shakedown at the
reactor operating conditions.

These results clearly demonstrate the contribution of the weld

residual stresses to the IGSCC of the 304 SS pipes in the BWRs. These
high stresses when combined with the operating stresses exceed the yield
stress of the material; this is one of the necessary requisites for
IGSCC. There are two other requisites: weld sensitization and the BWR
environment that contains dissolved oxygen in sufficent levels in the
high purity primary coolant water. A model ( ' ' for IGSCC in welded
304 SS in BWR has been developed and is based on the crexistence of
these three factors in the HAZ.

Finally the effect of weld residual stresses on the IGSCC of 304 SS

piping is further demonstrated by the results of the experiments on heat
sink welded (HSW) pipes. This process was found to significantly alter
the stress distribution in the welded pipe. Internal surface and
throughwall axial residual stresses were highly compressive compared to
conventional welded pipes (Figure 1 0 ) . The magnitude of the change in
residual stresses between normal welded and heat sink welded decreases
with increasing pipe diameter and appears to level off beyond 41 cm
(16 inch) in diameter. ' IGSCC tests were performed on full-size
10-cm (4-inch) 304 SS pipes prepared by normal and heat sink welding
procedures. A statistical test matrix was formulated to quantify
the benefits of heat sink welding. Results of these tests showed the
heat sink welded pipes to have a factor of 15 improvement in the time
 392

for IGSCC failure over the normal welded pipes.' ' This was attributed
primarily to the compressive residual stress patterns resulting from the
process. Heat sink welding also reduces the level of sensitization and
this may also contribute somewhat to the reduced susceptibility of
IGSCC.

In the application of heat sink welding, postweld grinding must be

avoided. Residual stress measurements on 304 SS coupons with ground
surfaces showed very high tensile residual stresses (Table I I ) . These
were considerably above the weld residual stresses resulting from normal
welding. Such high surface residual stresses will contribute to the
initiation of IGSCC.

SUMMARY

Incidents of IGSCC in welded 304 SS pipes in boiling water reactors

are a recurring problem for the utilities. Piping systems are designed
to a code that does not include the weld residual stresses. Measure-
ments were made on the interal surface and throughwall residual stresses
on 304 SS pipes preapred with field welding procedures. Results of
these measurements have been compared to the information from the field
incidents and special laboratory tests to illustrate the effect that
weld residual stresses have on the IGSCC behavior of 304 SS pipes in
BWRs.

In considering the stresses on the 304 SS pipes, the peak axial

tensile surface residual stresses, when added to the code calculated
stresses, exceed the yield stress in the weld heat-affected zone. The
magnitude of these stresses decreases with increasing pipe diameter.
IGSCC incidents also show a pipe size dependency with the highest
frequency of incidents occurring in small pipes (10 - 25 cm) and the
lowest in large pipes (61 cm - 25 inch). The throughwall residual
stress distribution patterns were also a function of pipe diameter.
Small pipes showed axial stresses through most of the wall thickness,
whereas compressive axial stresses existed in the large pipes. In the
only IGSCC Incident in a 61-cm (24-inch) pipe, the crack tip coincided
with the region of compressive residual stresses. For the small-
diameter pipes, the throughwall axial tensile stresses provide a driving
force for crack propagation.

Weld residual stresses in the 304 SS pipes do not appear to

shakedown during operating and cracking incidents would be expected to
continue in the older operating plants. This is consistent with field
experience.

The heat sink welding process results in compressive residual

stresses on the pipe internal surface and partially throughwall. By
altering the residual stresses, the resistance to IGSCC was improved
substantially compared to pipes fabricated by conventional field
procedures.
 393

Postweld grinding of the pipe weldment introduces very high tensile

residual stresses on the internal surface. These stresses contribute
significantly to the initiation of IGSCC

The results of these investigations illustrate the deleterious

effect of tensile weld residual stresses on the IGSCC of 304 SS pipes in
boiling water reactors.

ACKNOWLEDGEMENT

Information reported in this paper was obtained from jointly

sponsored contracts between the Electric Power Research Institute and
the General Electric Company, and from contracts with Argonne National
Laboratory, Battelle Columbus Laboratories, and Southwest Research
Institute. The Boiling Water Reactor Owners Group IGSCC Research
Program provided the support for these projects.

REFERENCES

1. "Pipe Cracking Experience in Light Water Reactors", Technical

Report, NUREG-D679, August 1980, U.S. Nuclear Regulatory
Commission.

2. H. H. Klepfer et al., "Investigation of Cause of Cracking in

Austenitlc Stainless Steel Piping", NEDO-21000 Vol. 1, July 1975,
General Electric Co.

3. Technical Report, "investigation and Evaluation of Cracking in

Austenitlc Stainless Steel Piping of Boiling Water Reactors",
NUREG-75/067, October 1975, U.S. Dept. of Commerce Report
PB-246 645.

4. "Investigation and Evaluation of Stress Corrosion Cracking in

Piping of Light Water Reactor Plants", NUREG-0531, U.S. Nuclear
Regulatory Commission, February 1979.

5. N. Hughes, "Solution Heat Treatment, Corrosion-Resistant Cladding

and Heat Sink Welding of 304 Stainless Steel Pipe Welds", EPRI
WS-79-174, Vol. 1, May 1980.

6. "Residual Stress Measurements by X-Ray Diffraction", SAE J784a,

August 1971, published by Society of Automotive Engineers.

7. R. M. Chrenko, "Residual Stress Measurements on Type 304 Stainless

Steel Welded Pipes", EPRI WS-79-174, Vol. 1, May 1980.

8. "Studies on AISI Type 304 Stainless Steel Piping Weldments for Use
in BWR Applications", NP-944, EPRI Final Report on RP449-2,
December 1979.
 394

9. "Measurement of Residual Stresses in Type 304 Stainless Steel

Piping Butt Weldments", NP-1413, EPRI Phase Report on RP449-1, June
1980.

10. "Controlling Residual Stresses by Heat Sink Welding", NP-2159-LD,

EPRI Final Report on RPT109-1 and 1576-1, December 1981.

11. R. E. Hanneraan, "Models for the Intergranular Stress Corrosion

Cracking of Welded Type 304 Stainless Steel Piping in BWR", EPRI
WS-79-174, Vol. 1, May 1980.

12. R. Sasaki et al., "Mitigation of Inside Surface Residual Stresses

of Type 304 Stainless Steel Pipe Welds by Inside Water Cooling
Method", EPRI WS-79-174, Vol. 1, May 1980.

13. R. Post and J. C. Lemaire, "Statistical Approach to Qualify

Countermeasures", EPRI WS-79-174, Vol. 1, May 1980.

14. N. Hughes, "Qualification of Solution Heat Treatments, Corrosion-

Resistant Cladding and Heat Sink Welding of 304 Stainless Steel
Pipe Welds", EPRI WS-79-174, Vol. 1, May 1980.
 REMOTE AUTOMATIC PLASMA ARC CLOSURE WELDING OF A DRY STORAGE
CANISTER FOR SPENT NUCLEAR FUEL AND HIGH LEVEL RADIOACTIVE WASTE*

R. P. Sprecace and W. P. Blankensbip

Westingbouse Electric Corporation

Advanced Energy Systems Division
P. 0. Box 10864
Pittsburgh, Pennsylvania 15236

ABSTRACT

A carbon steel storage canister has been designed for the dry
encapsulation of spent nuclear fuel assemblies or of "logs" of
v i t r i f i e d high level radioactive waste. The canister design i s in
conformance with the requirements of the ASME Code, Section I I I ,
Division 1 for a Class 3 vessel. The canisters will be loaded and
sealed as part of a completely remote process sequence to be per-
formed in the hot bay of an experimental encapsulation f a c i l i t y at
the Nevada Test S i t e . The final closure to be made i s a full
penetration butt weld between the canister body, a 12.75-in O.D. x
0.25-in wall pipe, and a mating semiellipsoidal closure l i d . Due to
a combination of design, application and f a c i l i t y constraints, the
closure weld must be made in the 2G position (canister v e r t i c a l ) .
A process selection study and process verification testing
program culminated in the selection of plasma arc welding, using the
autogenous keyhole technique, as the welding process to be developed
for making the final closure weld. Specifications were prepared for

*Work sponsored by the U.S. Department of Energy under the

following contracts:

• Experimental Demonstration Package Program, Contract No. CPFF-854

(with Rockwell Hanford Operations, Division of Rockwell
International).

• Commercial Waste and Spent Fuel Packaging Program, Contract

No. DE-AC08-81NV10171.

395
 396

a completely automatic, programmable plasma arc welding system

including an automatic welding head custom designed to interface
with the canister closure configuration. A system conforming to
these specifications was then procured on a competitive basis;
delivery to Westinghouse AESD/Pittsburgh was in April, 1980. A
program of experimental welding was then initiated in which a number
of welding procedure details were investigated and, more signifi-
cantly, the capabilities and shortcomings of the welding system were
assessed and appropriate improvements made and/or defined.
In October, 1980 the system was shipped to the encapsulation
facility at the Nevada Test Site, operated for USDOE by Westinghouse
AESD/Nevada. Here the final phases of equipment system modifica-
tions and welding procedure development were conducted. These
efforts resulted in the definition of a completely automatic welding
procedure for performing the subject closure weld with the loaded
canister and welding head located remotely in the encapsulation hot
bay. The weld is performed using the keyhole technique, and is
autogenous except for a small filler metal addition made at the
point of final keyhole closure to permit meeting the maximum
allowable undercut limit imposed by the controlling specification.
The welding system is described, and the final welding procedure
is described and discussed in detail. Several aspects and results
of the procedure development activity, which are of both specific
and general interest, are highlighted; these include:

• The critical welding torch features which must be

exactly controlled to permit reproducible energy input
to, and gas stream interaction with, the weld puddle.

• A comparison of results using automatic arc voltage

control with those obtained using a mechanically fixed
initial arc gap.

• The optimization of a keyhole initiation procedure.

• A comparison of results using an autogenous keyhole

closure procedure with those obtained using a filler
metal addition.

• The sensitivity of the welding process and procedure to

variations in joint configuration and dimensions and to
variations in base metal chemistry.

Finally, the advantages and disadvantages of the plasma arc

process for this application are summarized from the current view-
point, and the applicability of this process to other similar appli-
cations is briefly indicated.
 397

INTRODUCTION

In support of the Nuclear Waste Terminal Storage Program a car-

bon steel dry storage canister, compatible with currently licensed
truck-mounted shipping casks, was designed for the packaging of
spent light water reactor fuel assemblies to be emplaced in various
envisioned geologic repository experiments and demonstrations. A
later expansion of scope resulted in the design of an alternate (but
externally identical) version for packaging of "logs" of vitrified
high level radioactive waste.
The selection of a joining process to be used for making the
final hermetic structural closure of the canister was an important
aspect of this design effort. Early in the program, a process se-
lection study was performed which resulted in the identification
of three viable process candidates. These were plasma arc welding
(PAW), using the autogenous keyhole technique; pulsed-current gas
metal arc welding (GMAW); and pulsed-current gas tungsten arc weld-
ing (GTAW) with filler wire addition. A verification testing task
was then undertaken in which the feasibility of each of these can-
didates was experimentally assessed and demonstrated. Finally, a
trade study and evaluation was performed which resulted in the
selection and recommendation of PAW as the preferred approach.
The implementation of this recommendation included the procure-
ment of a PAW equipment system, a demonstration of the ability to
perform a fully automatic closure weld using this system, the devel-
opment of a detailed closure welding procedure, and the installation
and qualification of the system and procedure in the field facility
where remote encapsulation operations will be performed. The
details of the implementation sequence constitute the balance of
this paper.

CANISTER DESIGN DESCRIPTION

The carbon steel storage canister design configuration is shown

in its two alternative versions in Fig. 1. Each canister (less
contents) is an assembly of six components: body, lower end cap,
support ring, closure lid, assembly (backing) ring and lifting
pintle. The body is a section of ASTM A106 (or A53) Grade B pipe,
nominally 12.75-in O.D. x 0.25-in wall x 160-in long. The lower end
cap is fabricated from a standard 2:1 ellipsoidal end cap made from
0.38-in thick ASTM A516 (or A515) Grade 70 plate. The support ring
is 0.5-in thick ASTM A36 plate. A canister body subassembly is made
from these three parts by tack welding the support ring to a tapered
seat machined inside the lower end cap, and then joining the lower
end cap to the body with a full penetration structural weld.
 398

LIFTING PINTLE
LIFTING PINTLE SPOTFACE FOR 'SPOTFACE' FOR ALIGNMENT/
ALIGNMENT ATTACHMENT OF WELDER
OF WELDER
CLOSURE LID
CLOSURE LID CLOSURE LID 2:1 ELLIPSOID
2:1 ELLIPSOID ASSEMBLY RING
CLOSURE LID

LIFTING BAIL
HiJf BUMPER
ASSEMBLY RING

CANISTER MAIN BODY

VITRIFIED HIGH
LEVEL WASTE CANISTER
CANISTER PWR FUEL ASSEMBLY
MAIN BODY

ADAPTER

LOWER LOWER SUPPORT RING

SUPPORT RING
LOWER END CAP LOWER END CAP
2:1 ELLIPSOID 2:1 ELLIPSOID
HIGH LEVEL WASTE VERSION SPENT FUEL VERSION

Fig. 1. Storage canister design configuration.

The closure lid is fabricated from an end cap identical to that

used for the lower end cap. A hole bored in the center of the cap,
and a surrounding flat machined surface, are used for aligning the
ASTM A1C8 Grade 1018 lifting pintle which i s then joined to the cap
with a full penetration structural weld. The weld preparation on
the bottom edge of the closure lid is machined parallel to the upper
flat to permit use of the flat surface for mounting and aligning the
automatic closure welding head. A closure lid subassembly i s then
completed by tack welding the assembly ring (made from any of a
number of carbon steel grades) inside the lower edge of the l i d .
Details of the closure lid/canister body weld joint configur-
ation are shown in Fig. 2. Due to a combination of encapsulation
facility constraints and concerns related tb t i l t i n g and rotation of
an unsealed canister containing a spent fuel assembly, the closure
weld is made in the 2G position (canister v e r t i c a l ) . The weld
preparation i s a 10-degree butt joint, angled upward from outside to
inside through tha canister wall. The close-fitting assembly ring
spans the joint interface, and contains a deep circumferential
groove located directly behind the interface. The ring serves three
purposes: i t facilitates assembly and alignment of the closure lid
with the canister body, provides a shield to prevent weld splatter
from impinging on the canister contents, and provides (via the
groove) an escape plenum for the plasma gas column which penetrates
the canister wall during welding. The plasma gas contained in the
groove also provides an inert atmosphere purge which prevents burn-
ing of the weld root surface. This i s somewhat superfluous in the
current application, since the assembly will be backfilled with an
 399

CLOSURE LID

BACKING RING

RELIEF GROOVE

CANISTER
BODY

Fig. 2. Closure weld joint configuration.

inert gas mixture prior to welding, but could be very significant in

other applications where the inert gas backfill would not be
utilized.

CLOSURE WELD REQUIREMENTS

Two key factors influencing the selection of a canister closure

welding process were the requirement for completely remote welding
in the encapsulation facility hot bay, and the requirement that the
canister design meet the intent of the ASME Boiler and Pressure
Vessel Code, Section III, Division 1 for a Class 3 vessel. The
first of these dictated that the process be highly reliable and
amenable to fully automatic equipment system operation and welding
process control.
Consideration of the second factor led to the stipulation that
the design would exactly conform to Code requirements except in
areas where those conflicted with the functional requirements placed
on the canister for acceptable performance in its intended applica-
tion. This affected the closure weld in the areas described below.
 400

Weld Joint Design

The selected joint configuration is consistent with those

allowed by Section I I I , Article ND 3352.

Weld Procedure Qualification

The final detailed welding procedure, with the equipment system

installed in the remote encapsulation facility, will be qualified in
accordance with tbe requirements of Section IX as specified in
Section I I I , Article ND 4300.

Weld Inspection

Inspection i s relevant both to qualification and to tbe accep-

tance of production welds. Techniques applicable to tbe completed
weld joint fall into four categories, as summarized below:

Visual Examination - Tbe requirements for a Class 3 vessel are

stated in Section I I I , Articles ND 4424 and ND 4426.1. With
reference to Fig. 3, these are basically as follows:

CLOSURE LID

UNDERCUT
.031-IN. MAX.

CONCAVITY

ROOT SIDE
FACE (CROWN) SIDE

CANISTER BODY

REINFORCEMENT
.094-IN. MAX.

Fig. 3. Definition of closure weld features.

 401

• Weld bead surfaces must be free from coarse ripples,

grooves, overlaps, porosity and abrupt ridges or
valleys.

• Undercut shall not exceed 1/32 in and shall not

encroach on the required section thickness.

• Reinforcement shall not exceed 3/32 in.

• Concavity on the root side is permitted when the

resulting thickness of the weld is at least equal to
the thickness of the thinner member.

Nondestructive Surface Examination - The applicable techniques

are liquid penetrant and magnetic particle testing. These "hands-
on" procedures are not feasible for inspection of production welds,
since the radiation levels preclude personnel access to the
unshielded canister following closure welding. For procedure
qualification, it was judged that neither technique would provide
significant information beyond that obtained in the other examina-
tions and tests to be utilized. Thus nondestructive surface
examination was not performed on this program.

Nondestructive Volumetric Examination - The applicable

techniques in this category are radiography and ultrasonic testing.
Although not required for weld procedure qualification, welds were
radiographed in accordance with Code procedures to provide addi-
tional verification of weld joint integrity. In production welding,
radiography is not feasible due to the presence of a diffuse, high-
intensity radiation source inside the canister. Elimination of
radiographic inspection is acceptable provided that the weld proce-
dure is qualified and that a reduced weld efficiency (45 percent in
this case, per Section III, Article ND 3352.l(c)) is used in design
calculations. Ultrasonic testing is conceptually feasible for
volumetric examination, but is not a Code requirement. Since
standard ultrasonic testing equipment is not available for the
configuration and encapsulation conditions of this canister, ultra-
sonic examination was not included in the inspection requirements.

Destructive Examination and Testing - The applicable

requirements, as stated in Section IX, Article QW451 consist of
tensile tests, bend tests and certain documentation actions.
The tensile test specimen definition for pipe welds is provided
in Section IX, Article QW452.1(b), together with the requirements
placed on the tensile testing procedure. The acceptance criterion
is that measured tensile strength of the welded specimen must be
greater than the minimum allowable tensile strength of the weaker
base metal (in this case, the canister body).
Section IX, Article QW466 provides the necessary bend test
specimen definitions, a listing of which tests must be performed and
the requirements for the bend testing procedure. The acceptance
 402

criteria for a l l bend tests i s that all "open defects" must be less
than 1/8-in deep after bending, except for cracks initiated at the
edge of the specimen which did not i n i t i a t e at slag inclusions or
internal weld defects. (No limit i s placed on the size of edge
cracks which i n i t i a t e in sound weld metal.)
No elongation measurements, hardness testing or metallographic
examination i s required by the Code for weld procedure qualifica-
tion. However, these were performed on qualification weld specimens
to develop a more complete characterization of the structure and
properties of the canister closure weld.

PLASMA ARC WELDING SYSTEM

System Procurement

Following t h e p r e p a r a t i o n of an a p p r o p r i a t e equipment s p e c i f i -
c a t i o n , based on t h e c a n i s t e r design and on t h e p h y s i c a l and
environmental constraints of the application and the encapsulation
facility, a competitive procurement cycle was initiated. This
culminated in the placement of an order with Hobart Brothers, Inc.
of Troy, OH. The Hobart system included automatic welding heads
designed and fabricated on a subcontract basis by The General Atomic
Company of San Diego, CA. The system was delivered to Westingbouse
AESD/Pittsburgh, PA in April, 1980, where acceptance testing was
successfully performed to complete the procurement activity.

As-Delivered System Description

The principal components of the welding power supply/ control

system are shown in Fig. 4. The features and functions of these
components are briefly described as follows:

Hobart Cyber-Tig CT-300 DC Power Supply - Rated for 300 amps/20

volts at 100 percent duty cycle, with 80 volts maximum at open
c i r c u i t . Dual welding current ranges (3-30 or 3-300 amps). Module
also contains programmer and meters for welding current and power
supply output voltage.

Hobart 800 Series Programmer - Provides basic capability for

programmed automatic control of the closure weld cycle, including
current upslope during weld initiation, current taper during main
weld cycle, and current downslope during weld termination. Also
controls prepurge and high-frequency arc starting. Provides appro-
priate interfaces for cycle features controlled by other system
components.
 403

Fig. 4. Welding power supply/control system. <A) Power supply,

(B) Programmer/meter panel, (C) Plasma control unit, (D) Arc voltage
control unit, (E) Motor speed controller, (F) Four-channel recorder.

Hobart HPW-400 Plasma Control Unit - Regulates and meters the

plasma and shielding gas flows to the welding torch; controls and
displays pilot arc current and voltage; and controls the downslope
of plasma gas flow rate during weld termination.
 404

Cyclomatic Model 266 Arc Voltage Control (AVC) Unit - For PAW
closure welding, this unit establishes and maintains the arc voltage
in the transferred arc welding mode by continuously adjusting,
through a feedback control system, the electrode-to-work separation
necessary to maintain the arc voltage at the preset value. Two
control ranges, 5-30 and 5-50 volts, are provided. A voltmeter
continuously displays the average arc voltage as measured at the
sensing leads on the welding bead. A manual in/out torch-position-
ing setup control is also wrovided.

Hobart Cyber-Tig Motor Speed Controller - Controls the drive

motors mounted ot» the automatic welding head which provide bead
rotation, filler wire feed and vertical torch travel. Contains set
point potentiometers for the drive motors, master weld cycle timer
for programming the duration of the complete weld cycle, programming
controls for initiation and termination of filler wire feed, and
override switches to permit manual operation of all three motors.

Gould 2400 Series Four-Channel Strip Chart Recorder - Records

weld head rotation drive motor voltage, weld current, arc voltage
and filler wire feed drive motor voltage.

Two additional system components are shown in Fig. 5. These

are:

Fig. 5. Welding system components requiring location in

hot bay. (A) Recirculating torch coolant system, (B) High-frequency
arc start booster.

Thermal Dynamics Model HP 100 Water Circulator - Provides the

necessary cooling water to the PAW torch. A recirculating system,
installed in the hot bay near the welding station, i s required due
to r e s t r i c t i o n s (arising from nuclear c r i t i c a l i t y concerns) on flows
of water through the hot bay shielding boundaries.
 405

Hobart Model 1611 Remote Arc Starter - Ibis booster unit for the
high-frequency pilot arc ignition process, interposed between the
power supply/control system and tbe PAW torch, is needed because of
the long (y- 65 ft) electrical cables required for remote hot bay
installation of tbe automatic welding head.

Tbe General Atomic automatic welding head, mounted on a canister

closure mockup assembly, is shown in Fig. 6. The functional compon-
ents of this bead are described as follows:

Fig. 6. Automatic welding bead. (A) Lifting bail, (B) Attachment

hardware, (C) Rotation drive motor, (D) Eddy current proximity
sensor, (E) PAW torch, (F) Trailer shield, (G) Torch tilt angle
adjustment, (H) Torch lead/lag angle adjustment, (I) Manual torch
in/out controls, (J) Cables and hoses supported by overhead boom,
(K) AVC drive unit, (L) Filler wire feed system.

Hobart Model HPW-400 Plasma Arc Welding Torch - Performs the

canister closure weld. Mounted on brackets which permit adjustment
of torch vertical tilt and horizontal lead/lag angles.
 406

Cyclomatic Model 266 Arc Voltage Control Drive Unit - Drives the
plasma torch toward or away from the weld joint as required to
maintain the arc voltage at the preset value, and provides controls
to manually adjust the torch/workpiece gap during setup.

Head Rotation Drive Motor - Drives the torch carousel through

interaction with a stationary ring gear, thus rotating the plasma
torch along the weld joint interface.

Vertical Torch Travel Motor - Drives the torch in the vertical

direction (for setup or emergency adjustments) through a
rack-and-pinion arrangement.

Filler Wire Feed System - Delivers weld f i l l e r wire to the weld

puddle when/if required.

Eddy Current Proximity Sensor - Provides capability for

establishing torch/workpiece separation gap at the preset dimension
prior to initiation of transferred arc welding mode.

Attachment Mechanism - Three-clamp system for attaching the

welding head to the closure lid such that the orbit of torch
rotation is concentric with the lifting pintle axis and the plane of
the orbit is parallel to the plane of the edge of the closure weld
preparation.

Lifting Bail - Provides a means for lifting the welding head for
installation on and removal from the closure lid, either manually or
using remote manipulator-actuated equipment.

System Modifications

During the course of welding system installation, checkout,

calibration and procedure development a c t i v i t i e s , certain highly
desirable or mandatory modifications were identified and imple-
mented. While many of these were considered normal for the process
of making a complex custom-designed system operational, a few were
of fundamental significance in the context of achieving acceptable
system functioning and performance for the intended application.
These latter items are briefly described in the following paragraphs.
The as-delivered system contained a master weld cycle timer for
programming the duration of the complete weld cycle. During the
early procedure development experiments, i t was determined that the
accuracy and repeatability of this timer were unacceptable for
providing the necessary control over the points at which various
weld termination sequences could be initiated. Analysis and vendor
discussions led to the conclusion that no substitute timer having
adequate range could be identified to provide the necessary
capability.
 407

Accordingly, an alternate control scheme based on azirauthal position

sensing was conceived and implemented. Through incorporation of a
microswitch and tripping pin on (respectively) the rotating and
fixed elements of the welding head, plos restricted-duration
precision timers with appropriate control system interconnections,
the following cycle control sequence was established:

• Establish an exact azimuthal starting point reference

by initial tripping of the microswitch prior to weld
initiation.

• Sense exactly 360 degrees of weld travel by again

tripping the microswitcb as head rotation proceeds
during welding.

• Initiate and control the weld termination sequence

using restricted-range precision timers actuated by a
signal from the above event.

A second area in which significant modifications were required

was the control of weld travel speed as determined by welding head
rotation rate. Despite a number of attempts to improve the accuracy
and repeatability of the as-delivered open-loop motor speed control
subsystem, in-cycle speed variations of up to 8 percent from the
mean, and cycle-to-cycle variations of similar magnitude at a fixed
speed control setting, continued to be experienced. Variations of
this magnitude, with their attendant effect on weld heat input and
bead shape, were judged to be intolerable for this application. The
situation was resolved by replacing the entire as-delivered sub-
system with a closed-loop control system incorporating an integrated
gearmotor/tachometer drive unit. The functioning of the closed-loop
system has subsequently been found to be entirely satisfactory.
Modifications were also required to the filler wire feed drive
and control subsystem. The positioning mechanisms provided for the
wire guide system were found to be inadequate for making the pre-
cise, repeatable guide settings necessary to feed wire into the
keyhole closure region during weld termination. This was corrected
by designing, fabricating and installing a replacement positioning
system incorporating a dovetail slide and other precision adjustment
capabilities. In addition, the wire feed control timer circuitry
was reconfigured to permit the initiation of wire feed either beforr.
or after the initiation of final current Jownslope, in order to
provide maximum flexibility in the development of a weld termination
procedure incorporating filler wire addition.
Finally, it was necessary to modify the plasma torch orifice
bodies in order to achieve predictable, repeatable energy input to
the weld joint. This is discussed in detail in the section on
procedure deve1opment.
 408

WELDING PROCEDURE DEVELOPMENT

Materials

To provide specimens for procedure development and related

investigations, mockups of the canister body and closure lid were
designed as shown in Figs. 7 and 8. Body mockups were fabricated
from seamless pipe, with both ASTM A106 Glade B (3 heats) and A53
Type S Grade B (1 heat) materials represented in the specimen
population. Closure lid mockups were fabricated from end caps which
were formed from plate; ASTM A515 Grade 65 (2 heats), A515 Grade 70
(3 heats) and A516 Grade 70 (4 heats) were included. End cap
fabrication practices included cold forming with and without a sub-
sequent stress relief heat treatment, and hot forming.

.250 ±.010

6.0+0.1

Fig. 7. Canister body mockup design.

The specification requirements for chemical composition,

together with the heat analyses for a l l mockups used in the proce-
dure development and qualification work, are given in Table 1. No
basic weldability problems were experienced with any of these
materials, although (as noted later) some significant grade-to-
grade variations in weld bead shape were obtained with given welding
parameters.
Filler wire, used in the keyhole closure region of a number of
developmental welds and in the final procedure, was 0.035-in dia.
AWS Type ER80S-D2. This was used for reasons' of experimental con-
venience; any of a number of lower strength grades could have been
satisfactorily substituted.
 409

1.54 t .03

10° to°3O'

Fig. 8. Closure lid mockup design.

Procedure Description and Terminology

For convenience of discussion, the overall welding procedure

will be divided into five separate parts, defined as follows:

Setup - Welding head is attached to the closure lid and

adjusted; the head/lid assembly is then placed on the canister
body. Final adjustments are made automatically. An inert
atmosphere is established inside the canister assembly, and the
system is ready for initiation of the welding procedure.

Tack Welding - Closure lid is tack welded to canister body.

Weld Initiation - Switch is made from pilot arc to transferred

arc welding mode, keyhole is established and weld travel is
initiated.

Main Weld - Autogenous butt weld is produced as the torch is

rotated around the joint circumference.
 410

TABLE 1 Chemical composition of canister mockup materials

for weld procedure development and qualification

Composition (weight percent)

Mn Si

Canister Body Spec. 0.30 max 0.29/1.06 0.048 max 0.058 max 0.10 min
ASTM A106 Limits
Grade B Heat 0.16 0.83 0.009 0.013 0.19
Analyses 0.17 0.82 0.008 0.012 0.19
0.17 0.82 0.007 0.012 0.19

Canister Body Spec. 0.30 max 1.20 max 0.05 max 0.06 max
ASTM A53 Limits
Type S Heat
Grade B Analysis 0.24 1.00 0.010 0.015

Closure Lid Spec. 0.27 max 0.85/1.20 0.035 max 0.04 max 0.15/0.30
ASTM A516 Limits
Grade 70 Heat 0.23 1.09 0.005 0.011 0.27
Analyses 0.24 1.01 0.006 0.017 0.25
0.25 1.10 0.010 0.016 0.23
0.24 1.06 0.008 0.017 0.25

Closure Lid Spec. 0.28 max 0.90 max 0.035 max 0.04 max 0.15/0.30
ASTM A515 Limits
Grade 65 Heat 0.20 0.79 0.011 0.028
Analyses 0.19 1.13 0.005 0.023 0.27

Closure Lid Spec. 0.31 max 0.90 max 0.035 max 0.04 in.^x 0.15/0.30
ASTM A515 Limit s
Grade 70 Heat 0.22 0.47 0.007 0.022
Analyses 0.27 0.75 0.010 0.023 0.22
0.27 0.77 0.008 0.021 0.22
 411

Weld Termination - Weld travel is terminated; weld current and

plasma gas flow are terminated; keybole is closed; weld bead shape
in keyhole closure region is adjusted through filler wire addition,
if necessary, to meet application requirements.

Certain system or setup features and welding parameters were

fixed prior to the initiation of experimental welding, and main-
tained constant throughout the program. These are listed below:

• Plasma gas composition: argon.

• Shielding gas composition: argon.
• Electrode: turigsten/2 percent ThO2, 0.125-in dia.,
tip ground to a point, 60-degree included angle.
• Torch plasma orifice: 0.125-in dia.
• Torch tilt angle: 10° (up).
• Torch lead angle: 0° (perpendicular to surface).

Setup

To begin the process of making a developmental weld, the closure

lid and canister body mockups are thoroughly cleaned to remove soil,
waxes, oils and other contaminants, with a final joint surface
cleaning using oxylene or tnetbanol solvent being performed just
prior to use. The automatic welding head is attached to the can-
ister lid subassembly, and all necessary checks and torch position-
ing or other adjustments made. (In an actual encapsulation, this
would be the final hands-on activity prior to evacuation of per-
sonnel from the hot bay and initiation of the remote encapsulation
process sequence.) The head/lid assembly is then lifted and placed
on the canister body mockup, which is positioned on a baseplate or
on a "full volume" simulator to be described later. (The actual
remote encapsulation process sequence will be described toward the
end of the paper.)
An inert atmosphere is then established inside the mockup
assembly, either by purging through penetrations in the baseplate or
by evacuation and backfilling of the simulator. Helium was used for
this purpose throughout the development program. (In an actual
encapsulation, either helium or a mixture of helium and another
inert gas such as krypton or neon will be used. The helium will
provide a basis for helium leak testing the canister after closure
welding. The other gas, when used, will provide a means for
assessing continued canister integrity after shipment under
conditions where the detection of helium has ambiguous
implications.)
Finally, the torch is automatically driven in toward the
workpiece until the desired torch/work separation distance (i.e.,
 412

arc gap)* is established through a signal from the eddy current

proximity sensor to the AVC drive control. Welding is then
initiated as quickly as possible to minimize the opportunity for
helium out-leakage from the assembly.

Tack Welding

A set of three tack welds was adopted to prevent the development

of an axial gap at the joint interface in response to weld solidifi-
cation shrinkage forces produced early in the closure welding cycle.
(Gaps as large as 0.06-in were observed in mockups closure welded
without prior tack welding.) The 1- to 2-in long tack welds are
made in sequence at 90°, 270° and 180° from the point where closure
welding would be initiated. A keyhole is not established during
tack welding; instead, the lid and body are melted to a shallow
depth and then solidify to join the two pieces. The tack is
completely remelted during the closure weld cycle, so there is no
effect on the integrity of the final weld.
Two tack welding procedures were investigated. The key
parameters of the first of these are:

• Travel speed: 4 in/min.

• Plasma gas flow rate: 2-3 cfh
• Weld current: 100 amp
• Arc voltage: 28-30 volts
• Torch/work distance: 0.25 in
• AVC control: locked out

This procedure gave acceptable results for A516 Grade 70 closure

lids, but cracking occurred in a few instances with A515 Grade 70
material. Accordingly, a second procedure was developed incorpora-
ting a filler addition to give a M).010-in reinforcement. This
final procedure, performed with the AVC control locked out and using
an initial torch /work distance of 0.25 in, is shown graphically in
Fig. 9. This produced crack-free tack welds in all of the closure
lids used in the balance of the program.

* Throughout this program, torch/work separation distance is

defined as the dimension EW shown in Fig. 12. The actual arc
gap is the dimension FW shown in the same figure.
 413

200

WELD
CURRENT 100
(amp)
I I I I
0

E
PLASMA
FLOW RATE
(cfht

100

FILLER WIRE
FEED RATE
(in /min)

LJ
TRAVEL
SPEED
(in /min) I I
0 10 15 20 25 30 35 40
TIME (sec)
Fig. 9. Final tack welding procedure.

Weld Initiation

Three basic approaches to closure weld initiation were investi-

gated, only one of which yielded entirely acceptable results. The
first of these, shown graphically in Fig. 10(a), consisted of the
following steps:

• With torch stationary, switch from pilot arc to trans-

ferred arc mode at a high weld current level, 190 amps,
to establish a keyhole through the joint interface.

• With torch remaining stationary, after 7 second delay,

increase plasma gas flow rate from pilot level of 2 cfh
to main weld level of 7 cfh, and initiate weld current
downslope.

• Downslope current in 8 seconds from 190 amps to main

weld current level of (e.g.) 150 amps; initiate weld
travel of (e.g.) 4 in/min at the end of this dowoslope.
 414

200

WELD
CURRENT 100
(amp)

0
10
PLASMA
FLOW RATE
(cfh) 0

TRAVEL
SPEED
(ir. /mini r 2 4 6 8 10
(A) FIRST. TRIAL INITIATION PROCEDURE
12 14 16 18 20

200

WELD
CURRENT 100
(amp)

L
10
PLASMA
FLOW RATE
iefh) 0

TRAVEL
SPEED
(in /min) Li 2 4 6 8 10 12 14 16 18 20
(B) SECOND TRIAL INITIATION PROCEDURE

200

WELD
CURRENT 100
(amp)

0
1 1 1 1 1 1

10
PLASMA
FLOW RAT E
(cfh) 1 1 1 1 I I | | | I

TRAVEL
SPEED
(in /min)
5

n r
0
1
2
|
4
1
6
1
8
1
10
TIME (sec)
1
12
I
14
I
16
1
18
|
20

(C) FINAL INITIATION PROCEDURE

Fig. 10. Weld initiation procedures.

Two features of the resultant solidified weld start region were

undesirable, and could not be satisfactorily improved with minor
t r i a l variations in initial current level and/or downslope time.
F i r s t , a large residual tbrougb-bole was left at tbe start, which
bad to be assimilated into tbe weld puddle (in tbe overlap following
tbe f i r s t 360 degrees of weld travel) before tbe weld termination
 415

procedure could begin; and second, substantial sagging of the

starting puddle produced a "lump" of weld reinforcement at the lower
edge of the weld bead face which violated the 3/32-in acceptance
limit for reinforcement.
To improve the appearance of the weld start region and minimize
the weld puddle perturbation during overlap, a second approach to
weld initiation was developed. Shown graphically in Fig. 10(b), the
final version of this approach consisted of the following steps:

• With torch stationary, switch to transferred arc mode

at very low weld current level, sufficient to produce
only shallow melting at the weld joint face.

After 2 second delay, initiate weld travel of 4 in/min

and initiate weld current upslope.

• Upslope current in 4 seconds to main weld current level

of 140 amps; after 3 seconds of upslope, increase
plasma gas flow rate from pilot level of 2 cfh to main
weld level of 5 cfh. Keyhole is established "on the
fly" as the current approaches 140 amps.
Several combinations of initial current, travel start delay time
and upslope time were tried, with best results obtained at the
values noted above. The AVC control was locked out for a period of
7 seconds after transferred arc initiation to allow time for the
plasma column to stabilize. (Without this stabilization period,
large fluctuations occurred during startup, producing a very uneven
surface.) Using this initiation procedure, no residual through-hole
was produced, and the excessive reinforcement resulting from puddle
sagging was reduced somewhat. However, attempts to further reduce
the reinforcement through additional parameter optimization were
unsuccessful. This presented an intolerable handicap to the devel-
opment of an acceptable weld termination procedure, so additional
weld initiation development was pursued.
The third and ultimately successful approach to weld initiation
was based on the concept of a hole premachined through the canister
wall at the joint interface. By aligning the torch with such a hole
prior to initating the automatic closure weld sequence, it was
possible to establish and propogate a keyhole very quickly, with a
minimum of extraneous workpiece heatup and corresponding minimum
initial puddle size, and with elimination of the puddle turbulence
and splashback associated with the establishment of a keyhole by
directly melting through a solid wall with the high velocity plasma
column.
Initial development was done with premachined circular holes of
0.12 to 0.25-in dia. having their axes coincident with the plane of
the joint interface. The final configuration is a 0.22-in dia.
semicircular hole penetrating only the closure lid half of the
joint. This is simpler and less costly to machine, requires no
 416

matchup between closure lid and canister body, and introduces no

discernable penalty in weld initiation procedure complexity or in
acceptability of weld bead shape in the initiation region.
Figure 11 shows this final premachined hole configuration. (The
adjacent interconnected slot is discussed in a subsequent section.)
Figure 10(c) graphically shows the optimized initiation procedure
developed for use with the premachined hole; it consists of the
following steps:

WELD DIRECTION
Fig. 11. Preinachined keyhole starting penetration and
pressure relief slot.

• With torch stationary, switch to transferred arc mode

at main weld current level of 140 amps, simultaneously
increasing plasma gas flow rate from pilot level of
2 cfh to main weld level of 5 cfh.

• After 6 second delay, initiate weld travel of 4 in/min.

Main Weld

The main weld extends from the point at which a stable,

propogating keyhole i s established to the point at which weld cur-
rent downslope i s initiated. The welding parameters are held con-
stant during t h i s ^ 360-degree period (workpiece heatup effects
 417

are not sufficient to require compensatory weld current tapering),

except for arc gap/arc voltage adjustments made by the AVC unit
(when used) to compensate for ovality and/or thermal distortion.
The development of a main weld procedure involved the optimization
of various parameters, resolution of certain equipment system
problems, and choices among alternative procedural strategies, as
discussed below.
Weld Travel Speed - Main weld procedures were investigated at
travel speeds of 4, 6, 8 and 10 in/min. Equivalent weld heat input
levels and bead shapes were obtained using weld currents of approxi-
mately 140, 190, 240 and 270 amps, respectively. At speeds of
6 in/min and above, however, "double arcing" was experienced, with
the frequency of incidence increasing with increasing travel speed.
This is a phenomenon unique to plasma arc welding in which the
transferred arc established between electrode and workpiece suddenly
becomes diverted and split into two separate arcs, one between elec-
trode and torch orifice and the other between orifice and workpiece.
Several resultant effects, all deleterious, make it impossible to
produce an acceptable closure weld while double arcing is
occurring.
Several approaches to the elimination of double arcing are
known; however, those which do not. require modifications to the
plasma arc torch (such as increasing plasma gas flow rate at a given
travel speed and weld current) were tried without success. Accor-
dingly, a travel speed of 4 in/min was adopted for all remaining
work. No significant penalty was associated with this decision, as
the 10-minute weld time at this speed represents a minor fraction of
the time required for the entire encapsulation sequence.
Plasma Torch Characteristics and Performance - Throughout the
early phases of the procedure development activity, problems were
experienced with lack of reproducibility of weld joints made with
nominally identical control system settings, and with the failure of
systematic variations in certain settings to produce interpretable
results. Many of these difficulties were due to deficiencies in the
travel speed control subsystem, the resolution of which was
discussed earlier, and to various other minor electrical and
mechanical defects which were resolved in a straightforward manner.
The problems persisted, however, and a perception developed that the
source of these might lie in the plasma arc torch itself; one (of
several) reason for this was the periodic observation of the
occurrence of greenish coloration of the plasma column accompanied
by spitting sounds and plasma column fluctuations, all suggesting
that torch cooling water might be leaking into the plasma gas
stream.
To investigate the latter possibility, an intensive study of
torch characteristics and performance was undertaken. The pertinent
torch components and dimensional features are shown schematically in
Fig. 12. For study purposes, the workpiece was a water-cooled
copper block, and all tests were performed with the torch station-
ary. A set of 14 orifice bodies, as-received from the manufacturer,
 418

-TORCH BODY SHIELDING GAS CUP

WORK
PIECE

*
t-E W

-ELECTRODE OftJFICE BODY

Fig. 12. PAW torch schematic.

was collected for evaluation. Using fixed weld parameters and

electrode positions within the torch, it was quickly determined that
an unacceptably wide range of arc voltages was obtained over the
orifice bodies in the set. Representative test results illustrating
this conclusion are given in Table 2. Also, indications of torch
coolant leakage into the plasma gas stream were obtained with one
orifice body and suspected with others.
Dimensional and shape measurements were made to assess the
physical similarity of the orifice bodies. The results indicated
large variations it> features which were judged to be critical to the
reproducibiiity of plasma gas flow characteristics and of the
electrode to orifice body relationship. The "As-Received" lumns
of Table 3 indicate these variations for the dimensions AB <. ' AD.
Others included a range of ^0.030 in for the orifice length CD,
several mils of variation in orifice diameter, a noncoplanar
condition between the inner and outer contact surfaces at plane A,
and nonperpendicularity between the contact surfaces A and the
orifice body mounting thread axis.
To bring the set of orifice bodies into a reasonably standard
configuration, a hand forging die was designed and fabricated. Each
body was then forged to make the inner and outer contact surfaces at
A coplanar to within 0.001 in. The effect of this operation on Che
AB and AD dimensions is shown in the "After Forging" columns of
Table 3. Machining operations were then performed to bring the AB
and AD dimensions to the "After Machining" values shown in Table 3,
and to standardize the orifice length CD and the orifice diameter.
 419

TABLE 2 Test results illustrating nonreproducible performance

obtained with as-received torch orifice bodies

Orifice PIa sma Ga s Plasma Gas Pressure (in H2O) Arc

Body Supply Pressure Before I n i t i a t i n g After Initiating Voltage
Number (psig) Transferred Arc Transferred Arc (volts)

01 29.2 24 34 27.0
02 29.2 24 35 27.5
03 29.2 24 35 28.0
04 29.2 24 34 27.5
05 29.2 24 37 28.5
06 29.3 24 36 27.5
07 29.4 24 35 28.0
08 29.3 24 35 27.2
09 29.3 24 36 29.0
10 29.3 24 34 28.0
11* - - - -
12* - - - -
13* - - - -
—
14* — —

*Not tested prior to modification

TABLE 3 Critical plasma torch orifice body dimensions before

and after modification

Orifice Dimension AB* (in) Dimension AD* (in)

Body As- After After As- After After
Number Received Forging Machining Received Forging Machining

01 0.303 0.292 0.312 0.464 0.459 0.456

02 0.320 0.303 0.311 0.500 0.458 0.456
03 0.320 0.306 0.312 0.500 0.467 0.456
04** 0.294 0.285 0.286 0.453 0.435 0.434
05 0.305 0.294 0.310 0.480 0.457 0.456
06 0.303 0.292 0.311 0.480 0.453 0.452
07** 0.300 0.286 0.286 0.480 0.450 0.434
08** 0.321 0.318 0.332 0.508 0.490 0.464
09 0.310 0.306 0.311 0.480 0.472 0.456
10** 0.325 0.316 0.332 0.495 0.465 0.464
11 0.303 0.298 0.311 0.474 0.464 0.457
12** 0.314 0.305 0.311 0.474 0.465 -
13 0.306 0.294 0.311 0.474 0.448 0.449
14 0.305 0.303 0.311 0.453 0.449 0.449

*Dimensions defined in Fig. 12.

**As modified configuration judged unacceptable for further use.
 420

With these modifications completed, i t was possible to proceed with

a systematic study and optimization of electrode to orifice body
relationship. Listed below are the final standardized and/or
optimized* orifice body dimensions and electrode position relation-
ships, with pertinent comments regarding the establishment of each.
As indicated in Table 3, five of the 14 orifice bodies were
disqualified from further use due to excessive variations from these
standardized/optimized values.

• Electrode stickout, dimension AF: Optimized value is

0.300 + 0.001 in. Values from 0.280 to 0.320 in were
investigated.

• Dimension AB: Standardized value i s 0.311 +_ 0.001 in.

• Electrode tip to edge of orifice entry angle, dimension

FB: Standardized value i s 0.010 n'nnn i-n« ^ e
study
showed the following:

Values in the optimum range of 0.010 to 0.020 in

produce stable, predictable arc voltages with no
discernable electrode tip erosion.

At values greater than 0.020 in, rapid tip erosion

occurs.

- If the electrode is advanced so that the tip falls

within the orifice entry angle region BC, unstable
arc voltage behavior i s observed.

• Orifice entry angle, region BC: Standardized value is

118° (included), the as-received value.

• Orifice length, dimension CD: Standardized value i s

0.100 _+ 0.003 in. For a fixed arc gap FW, arc voltage
increased slightly as CD was varied from 0.095 to 0.103
in.

• Orifice diameter: Standardized value i s 0.125 +_ 0.001,

established by reaming each as-received orifice body.

• Dimension AD: Standardized value is 0.457 n'nnc. i-n<

As a result of standardizing dimensions AF and AD, the arc
gap FW can be conveniently established during setup by

*Note that t h i s "optimization" applies only for the complete set of

dimensions arrived at here. Other "standardized" dimensions could
have been established, and corresponding different "optimized"
values developed, that would produce equally satisfactory r e s u l t s .
 421

fixing the externally determinable dimension DW (or, as was done on

this program, the externally determinable dimension EW with the
interrelating dimension ED under close control).
Table 4 gives representative test results indicating the
performance repeatability among the (usable) members of the orifice
body set following modification and standardization. Orifice bodies
conforming to this standardized configuration were used exclusively
through the balance cf the program.

TABLE 4 Test results illustrating reproducible performance

obtained with modified, standardized orifice bodies
and optimized electrode-orifice relationship

Orifice Orifice Plasma Gas Plasma Gas Pressure After Arc

Body Length Supply Pressure Initiating Transferred Arc Voltage
Number (in) (psig) (in H£0) (volts)

01 0.101 29.7 32.5 25.0

02 0.102 29.7 33.0 25.0
03 0.101 29.7 33.0 25.0
05 0.103 29.7 33.0 25.0
06 0.098 29.8 33.0 24.9
09 0.102 29.8 33.0 25.1
11 0.103 29.8 33.5 24.9

13 0.095 29.7 33.0 24.6*

14 0.095 29.7 33.0 24.6*

*The reduced arc voltages for 013 and 014 are a direct consequence
of their shorter orifice lengths, as predicted by earlier test
r e su 11 s.

Assembly Ring Configuration and Joint Venting - To mitigate

concerns regarding internal pressure buildup which might lead :o
weld puddle blowout, considerable attention was devoted to the
geometric configuration of the assembly ring and to venting of the
weld joint. The final assembly ring configuration, shown previously
in Fig. 2, is a 0.50-in radial thickness member containing a 0.30-in
deep x 0.60-in high relief groove directly behind the joint inter-
face. This configuration evolved from the initial design configura-
tion, a 0.125-in radial thickness strip containing a 0.050-in deep x
0.250-in high relief groove, as a result of experiencing occasional
weld puddle blowouts with the latter configuration late in the
development program.
 422

The concern over internal pressure buildup led to an

experimental setup transition from the lid/body mockup assembly
resting on a baseplate penetrated only by purge gas inlet and purge
gas outlet/pressure relief ports, to the mockup assembly gasket-
sealed to a flange of a "full volume" simulator. The latter is
simply an evacuable/backfillable tank whose internal volume is
approximately equal to the net free internal volume of a canister
loaded with a spent PWR fuel assembly (less the internal volume of
the lid/body mockup). This allowed a more realistic assessment of
the magnitude of internal pressure buildup effects due to the con-
tinuous injection of hot plasma gas into the assembly ring relief
groove as closure welding progressed. Provision for an internal
heater also allowed an assessment of the additive effect of heat
generated by the canister contents.
Results of weld experiments made with the simulator indicated
that puddle blowout near the end of the main weld (or during the
weld termination sequence) was a credible possibility. To alleviate
this, a 0.5-in long x 0.010-in high venting slot, located at the end
of the main weld travel and interconnected with the semicircular
keyhole starting bole, was incorporated into the closure lid weld
preparation. (This slot was illustrated previously in Fig. 11.)
Approximately 20 complete closure welds were made on the simulator
with mockups containing this venting slot, with no blowouts being
experienced.

Automatic AVC vs. Fixed Arc Gap - Through most of the program,
automatic AVC was used to maintain constant arc voltage as the main
weld progressed. As discussed earlier, the AVC control was locked
out until a stable, propogating keyhole was established (using a
preset initial arc gap). It was then automatically switched on, and
functioned to increase or decrease the arc gap in response to
workpiece ovality, thermal distortion or other geometric pertur-
bations as required to maintain arc voltage at the preset value. At
the point of weld current downslope initiation, the AVC control was
automatically switched off, allowing weld termination to proceed
under (final) fixed arc gap conditions.
Excellent results were obtained with the use of the AVC system
in this manner, and it appeared that the system could adequately
compensate for much larger geometric perturbations than were allowed
or experienced in the canister closure hardware. Late in the pro-
gram, however, It became necessary to abandon its use. As described
further in a later section, this resulted from the unsuccessful
attempt to develop a completely autogenous keyhole closure
procedure, and the subsequently determined necessity for adding
filler wire prior to the initiation of current downslope.
The filler wire feed unit is fixed to the plasma torch mounting
bracket, and consequently is fixed to the AVC drive unit. The wire
feed nozzle thus moves in concert with the torch as the AVC system
functions. During the main weld, observations indicated that a
continuous small, cyclic variation in keyhole size occurred as the
weld progressed, and that the AVC system responded to this variation
by producing a small, cyclic variation in arc gap. The effect of
 423

this variation on the shape and acceptability of the autogenous weld

bead was completely negligible; however, its effect on filler wire
placement proved to be intolerable.
Because of this, rather than design and implement an alternate
wire feed unit mounting scheme or attempt to reduce the cyclic arc
gap variations tborugh procedure parameter adjustments, the decision
to abandon use of the AVC system was made. No significant problems
were introduced by this decision for the subject application. The
main weld procedure was shown to be tolerant to arc gap variations
of +_ 0.125 in from the specified nominal value, whereas maximum
canister hardware ovality is of order 0.020 in and the thermal
distortions measured during welding were 0.005 in at most. The
final main weld procedure is thus based on a fixed initial arc gap,
and automatic AVC is not used.

Base Metal Chemistry Variations - In the course of main weld

procedure development, it was determined that small but vital
adjustments were required in certain critical parameters to accom-
modate the differences in melting temperature and weld puddle
viscosity and surface tension among the various canister body and
closure lid base metal combinations investigated. Table 5 gives the
parameter variations required to yield acceptable main weld beads
for three base metal combinations. Based on these results, it
appears necessary to separately qualify a detailed welding procedure
for et -n base metal combination allowed in the canister design and,
quite possibly, to qualify on a material beat basis if substantial
heat-to-heat chemistry variations are encountered.

Table 5. Main weld parameter variations required for various

base metal combinations

Closure Lid/ Weld Arc Torch/Work PIa sma Ga s

Canister Body Current Voltage Separation Flow Rate
Base Metals (amp) (volts) (in) (cfb)

A515 Grade 70/ i ao- 25.5 0.19 ^ 5 (? 40 in H 2 0

A106 Grade B ias
A516 Grade 70/ 27.0 0.38 ^ 5 (3 35 in H 2 0
A106 Grade B 145

A516 Grade 70 26.0 0.25 T> 5 <§ 38 i n H 2 0

A53 Tp. S Gr. B 140

Base Metal Temperature - Heatup of the closure l i d and canister

body before and during welding will result from three sources:
decay beat generated by the canister contents, normal weld beat
buildup in the joint region, and the continuous injection of hot
plasma gas into the assembly ring relief groove as the weld
progresses. Test results on preheated mockups indicated that a
 424

given main weld procedure produces acceptable results over an

initial weld joint temperature range from room temperature to
150°F. Higher temperatures were not investigated, but it appeared
that a reduction in weld heat input would be required for preheat
temperatures much above 150°F.

Final Main Weld Procedure - The final main weld procedure, as

qualified for the A516 Grade 70/A53 Type S Grade B base metal
combination, is summarized as follows:

• Shielding Gas Supply Pressure: 30.0 psig*

• Shielding Gas Flow Rate: 36 cfh*
• Plasma Gas Supply Pressure: 29.8 psig*
• Plasma Gas Console Pressure: 38 in H 2 0
• Plasma Gas Flow Rate: 5 cfb
• Torch/Work Separation: 0.250 in*
• Arc Gap: 0.466 in*
• Arc Voltage: 26.0 volts
• Weld Current: 140 amp
• Weld Travel Speed: 4 in/min
As noted earlier (see Table 5 ) , appropriate adjustments in weld
current and plasma gas (console) pressure would be made for the
other allowed base metal configurations. Figure 13 illustrates the
appearance of a typical mockup welded using one of these main weld
procedures.

Weld Termination

Three basic approaches to closure weld termination were investi-

gated, only one of which yielded acceptable results. These are
described and discussed in the following sections.

Continguous Autogenous Termination - The simplest and most

desirable termination procedure would be a contiguous continuation
of the main weld procedure in which some optimum combination of weld
current, plasma gas flow rate and/or travel speed dovoslope (or,
conceivably, upslope) would produce an autogenous closure of the
keyhole and a final solidified closure region bead which meets all
of the previously stated acceptance criteria. A large number of
trial terminations was performed in which these parameters were

* These values also used for weld initiation and termination

procedures.
 425

Fig. 13. Appearance of typical closure welded mockup.

varied and their interrelationships studied. Two main effects were

observed to work in opposition to achievement of the objective:

• After initiation of current downslope, as soon as the

keyhole closed (which closure always initiated at the
root side of the weld), the weld puddle changed from a
stable, conical layer to a turbulent, closed-bottom
pool agitated by the incident plasma gas. In most
caseSj this turbulence resulted in the development of
unacceptable porosity during final solidification.

• Due to the conical shape of the keyhole (small end at

the root side), and the root-to-face direction of
solidification during termination, a certain amount of
face-side undercut (actually, underfill) was inevitably
present after solidification was complete.

The most successful attempt at a termination of this type was

performed using the procedure shown graphically in Fig. 14. The
termination region was free of porosity, but the face-side undercut
remained excessive. At this point, schedular requirements dictated
 426

200 WELD TERMINATION

WELD
CURRENT 100
(amp)

PLASMA
FLOW RATE
(cfh)

TRAVEL
SPEED
(in /min) r 4 6
TIME (sec)
10

Fig. 14. Contiguous autogenous weld termination procedure.

that another termination approach be investigated, as discussed in

the next section.

Contiguous Termination with Filler Addition - The next level of

procedural complication was the incorporation of a filler metal
addition into the keyhole closure region during termination. In
this way, it was envisioned that an undercut condition such as that
just discussed could be overcome in the most expedient manner.
Again a large number of trial terminations was performed in
which the previously discussed parameters plus filler wire feed rate
and wire feed position were varied and their interrelationships
studied. The most successful procedural approach of this type is
shown graphically in Fig. 15. Simultaneously weld travel was
terminated and weld current was reduced in a step change (as opposed
to a down siope). Also, at the same point, plasma gas down slope was
initiated at the maximum rate allowed by the control system, and
filler wire feed was initiated. This approach minimised the problem
of weld pool turbulence, but the undercut problem persisted. The
general results of this approach can be summarized briefly: when
porosity was eliminated, undercut was excessive; and when undercut
was controlled, unacceptable internal porosity was invariably
detected. Accordingly, this approach was also abandoned in favor of
the technique discussed in the next section.

Two-Pass Termination with Filler Addition - The final, successful

approach to weld termination was developed by Westinghouse AESD/
Nevada personnel after the author's participation in the project
had ended. This and other later aspects of the program will be re-
ported separately from this paper.
 427

200 WELD TERMINATION

-H
WELD
CURRENT 100
(amp)
I i I I

5
PLASMA
FLOW RATE
(cfh)
0

100 I -
FILLER WIRE
FEED RATE
(in./min)

TRAVEL
SPEED
(in /rnin)
_L _L
0 2 4 6 8 10 12
TIME (sec)
Fig. 15. Contiguous weld termination procedure with filler addition.

The final termination procedure consists of the following steps:

• Stop torch travel, and simultaneously initiate plasma gas
flow rate downslope, weld current downslope and filler wire
feed. Pulsed weld current (a capability not described in
this paper) is used to help minimize be»t input.

• The above results in a closed .. ytiole region free of

porosity but having excessive undercut.

« Back the torch *\» 0.5 in (circumferentially) along the weld

travel path.

• Perform a second, *\» 0.5-in long melt-in filler pass to

build up the undercut region to an acceptable level.

Although this approach is operationally more complex than the

contiguous approaches described previously, it enabled a successful
overall procedure qualification to be performed on the A516
Grade 70/A53 Type S Grade B base metal combination in accordance
with program requirements.
 428

WELD EVALUATION

At various points in the program, complete welded mockup

assemblies were selected for comprehensive evaluation. Following
visual examination and x-ray radiography, each assembly was sec-
tioned in accordance with a detailed sampling pattern, and con-
verted into appropriate specimens for tensile and bend testing,
hardness measurements and metallographic examination. Results of
the evaluation of two typical assemblies will be described here;
these are representative of results obtained on all other mockups
evaluated throughout the program.

Tensile and Bend Tests

Tensile and bend test specimens were machined to (essentially*)

the procedure qualification configurations specified in Section IX
of the ASME Code. The curved (closure lid) portion of each trans-
verse tensile specimen was flattened with a press prior to machining;
since this portion was part of a grip section of the final specimen,
the strain hardening introduced by flattening had no effect on %be
test results.
Each transverse tensile specimen was scribed with a 1.25-in gage
length and then tested (at room temperature) to failure in a
Wiedemann tensile machine using a constant crosshead speed of
0.05 in/min. The strength and elongation values calculated from the
test results are given in Table 6, together with the corresponding
minimum values allowed for the closure lid and canister body base
metals in their respective material specifications.
All specimens tested fractured it) the canister body material
section, at yield and ultimate strength values well in excess of the
specification minimums for that material. The results are thus in
conformance with the qualification acceptance criteria of Section IX
of the ASME Code. The variations among individual test results are
typical for welded steels of these general types. The reduced
elongation values (as compared to the base material specification
minimums) reflect the weld fusion zone hardening effects described
below. These elongation values do not affect the acceptability of
the tensile test results with respect to the Code criteria.
Bend test specimens were tested in a die/punch bending fixture
mounted on a Wiedemann tensile machine. A crosshead speed of
1.0 in/min was used for all specimens. The results are compiled in
Table 7, and the surface appearance of four representative specimens
after bending is shown in Fig. 16. No cracks were seen in any
specimen, so the results are in conformance with the Section IX
qualification acceptance criteria.

*Some bend test specimen widths deviated slightly.

 429

Table 6. Transverse tensile results and base material

specification minimums for typical welded mockups

Specimen Yield Strength (psi) Ultimate Tensile Elongation in

Number 0.2 pet. Offset Strength (psi) 1.25 in (pet.)

55T3 47,800 66,800 18.3

55TT67* 50,100 66,500 14.1
55T7 58,100 67,900 18.8
88T1 57,300 67,100 17.1
88T5 51,000 71,400 14.0
88TT67* 50,900 69,000 16.4

Base Material
Specification

Canister Body 35,000 min 60,000 min 27.0 min

A106 Grade B

Closure Lid 38,000 min 70,000 min 21.0 min

A516 Grade 70

*Specimen located in tack weld region

Table 7. Bend test results for typical welded mockups

Specimen Number Type of Bend Test Test Results

55B1 Transverse face bend No cracks

55B2 Transverse root bend No cracks
55BT45 Transverse face bend No cracks
55B6 Transverse root bend No cracks

88B2 Transverse root bend No cracks

88BT45 Transverse face bend No cracks
88B6 Transverse root bend No crack s
88B8 Transverse face bend No cracks
88LB34 Longitudinal root bend No cracks
 430

Face of Weld Face of Weld

(a) 55B1 (Transverse (b) 55BT45 (Trans-

Face Bend) verse Face Bend)

Root of Weld

(c) 88B2 (Transverse (d) 88LB34 (Lonqi

Root Bend) tudinal Root
Bend)
Fig. 16. Appearance of typical as-tested bend specimens.
 431

Hardness Measurements

Diamond Pyramid (Vickers) hardness measurements were made on

specimens from both assemblies to characterize the softening/harden-
ing effects of the plasma arc welding process on the canister mate-
rials. The Diamond Pyramid technique permits the determination of
accurate hardness values and the mapping of hardness gradients
through the relatively small transition zones found in welded struc-
tures of this type. The test specimens were metallographically
polished and etched, and hardness traverses made along the prepared
surfaces beginning in the canister body base metal and proceeding
through the body heat-affected zone, weld fusion zone, closure lid
heat—affected zone and lid base metal.
Plots of the resultant hardness profiles are shown for four
representative specimens in Fig. 17. In each case, the weld fusion
zone hardness values are at or near the highest level in the weld-
ment; this reflects a combination of alloying effects (the body and
lid base metal compositions are somewhat different, as indicated
previously in Table 1), microstructural effects and hardening due to
phase transformations during weld cooling. The body base metal i s
the softest region in each case; this follows directly from the
minimum strength specification values given previously in Table 6,
and the fact that hardness and tensile strength are closely corre-
lated in steels of this type. The hardness gradient in the body
heat-affected zone forms a monotonic transition between the low-
hardness body and high-hardness weld fusion zone; the hardening in
this region is due to transformation effects, as evidenced by the
fact that the gradient increases sharply as the weld fusion zone
boundary (where welding-induced temperature levels and subsequent
cooling rates are highest) i s approached.
The hardness gradient in the lid heat-affected zone is similar
to that in the body side of the weldment, and a similar transforma-
tion effects explanation seems appropriate. The difference in
closure lid hardness levels between the two assemblies i s within the
range expected from a combination of heat-to-heat chemistry varia-
tions and differences in detailed fabrication/processing parameters
during ellipsoidal lid manufacture. No significant differences were
observed between hardness profiles in tack welded versus non-tack-
welded regions within a given assembly.
Although the details of the hardness measurement results are of
considerable interest from the viewpoint of characterization and
understanding of the metallurgical effects and resultant properties
produced by the welding operation, the hardness variations observed
are of no consequence in evaluation of the acceptability of the weld
joint. All of the conditions described above are acceptable from
the mechanical, metallurgical and ASME Code criteria viewpoints.
 432

! 260 I I
1 1 1 1 i i i 1 1 I I I I I I I I ! I
i 240
: 220
' 200
k
180
160 —
140
120
\J
HAZ
FZ

HAZ LID —
BODY
100,
} 11
<
5
r i
0.8
i
0.6
i
0.4
i
0.2
1 1 1
| | 0.2 0.4
0 0.6
1 1
0.6 0.4 0.2 0 0.2 0.4 0.6
DISTANCE FROM FUSION ZONE CENTER LINE (IN ) DISTANCE FROM FUSION ZONE
SPECIMEN 55M5. NON-TACK-WELD REGION CENTER LINE (IN I
SPECIMEN 55MT23. TACK WELD REGION

0.60.4 0.2 0 0.2 0.4 0.6 0.8 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6
DISTANCE FROM FUSION ZONE DISTANCE FROM FUSION ZONE CENTER LINL (IN I
CENTER LINE (IN ) SPECIMEN 88MT2. TACK WELD REGION
SPECIMEN 88M7. NON-TACK-WELD REGION

Fig. 17. Hardness profiles across typical closure weld joints.

Metallographic Examination

Prior to making hardness profile measurements, metallographic

examination was performed on specimens taken from each assembly.
The specimens were metallograph.cally polished, etched to display
the microstructures of the various weld zones, and examined by
optical microscopy at various magnifications.
The cross-sectional features of a typical weld section are shown
in Fig. 18. The weld bead face of all specimens examined exhibited
slight reinforcement toward the canister body (lower) material, and
most exhibited slight undercut toward the closure lid (upper) mate-
rial. All of the reinforcement and undercut conditions seen in all
of the specimens examined fall within the allowable limits specified
in the Code.
 433

Fig. 18. Cross-sectional features of typical closure weld joint.

The tnicrostructures of the base metals, heat-affected zones and

weld fusion zones in typical specimens are shown in Fig. 19. No
significant differences were seen either between assemblies or
between tack welded and non-tack-welded regions within a given
assembly. No porosity, weld-created inclusions or other undesirable
microstructural features were seen in any of the specimens examined;
all were judged to be metallurgically sound and completely accept-
able for this application.

REMOTE ENCAPSULATION PROCESS SEQUENCE

An a e r i a l view of t h e experimental e n c a p s u l a t i o n facility

("E-MAD" F a c i l i t y ) a t t h e Nevada Test S i t e i s shown i n F i g . 20.
This f a c i l i t y , b u i l t i n t h e 1 9 6 0 ' s a s p a r t of t h e NERVA ( n u c l e a r
r o c k e t engine) program, i s c u r r e n t l y operated for t h e U . S . DOE by
Westingnouse AESD/Nevada O p e r a t i o n s . An a p p r e c i a t i o n of t h e s c a l e
of t h e h o t bay i n t e r i o r may be gained from F i g . 2 1 , which shows a
c a n i s t e r e d spent fuel assembly (from an e a r l i e r program) being
remotely loaded i n t o a s h i e l d e d s t o r a g e cask for an above ground dry
storage demonstration. The canister closure welding station is
shown in Fig. 22, where a spent fuel assembly is being remotely
loaded into an earlier design canister prior to closure welding.
 434

IIS
Fig. 19. Microstructural features of typical closure weld joints.

The remote encapsulation process sequence for the canister and

plasma arc closure welding system described in this paper is shown
in Fig. 23(a) through 23(1). A brief description of each
illustrated item or step is given below:

23 (a) Weld pit general arrangement, with adapter for the

canister described in this paper.
23 (b) Canister body subassembly installed in weld pit.
23 (c) Evacuation/backfill collar installed on canister body.
23 (d) Photograph of evacuation/backfill collar.
23 ( e ) Spent PWR fuel assembly remotely installed into
canister body subassembly.
23 ( f ) Automatic welding bead/closure lid assembly remotely
installed onto canister body.
23 <g) Evacuation/backfill collar remotely positioned and
attached to backfill system. (Canister is then
 435

evacuated and backfilled with helium, after which

collar is deactivated and stripped down away from
closure weld joint.)
2i \h}> Automatic plasma arc closure weld made remotely.
23 (i), Automatic welding bead removed remotely from welded
canister.
23 (j), Evacuation/backfill collar removed remotely from
welded canister.
23 (k). Vacuum chamber hood remotely installed. (Chamber is
then evacuated through mass spectrometer leak detector
to perform helium leak inspection of welded canister.)
23 <1). (After remotely removing vacuum chamber hood), welded
canister is removed remotely from weld pit and
transferred to shielded storage or transport cask.

Fig. 20. Experimental encapsulation facility.

 436

Fig. 21. Hot bay interior.

 437

Fig. 22. Hot bay canister closure welding station.

 438

WELD PIT WITH VACUUM CHAMBER AND

INSTALLED WELD PIT ADAPTER FOR
12.75 O.D. CANISTER

INSTALLATION OF 12.75 O.D. CANISTER

•ODY INTO WELD PIT ADAPTER

INSTALLATION OF EVACUATION/MCKFILl COLLAR

(DOWN POSITION) ON 12.7S O.D. CANISTER BODY

Fig. 23. Remote encapsulation process sequence.

 439

INSTALLATION OF PWR FUEL ASSEMtLY INTO

12.75 O.D. CANISTER iODY (COLLAR IN DOWN POSITION)

INSTALLATION OF PLASMA ARC WELDING MACHINE

AND 12.75 O.D. CANISTER CLOSURE LID
(COLLAR IN DOWN POSITION)

EVACUAriON/IACKFILL COLLAR ATTACHMENT TO

HELIUM FILL CART FOR 12.75 O.D. CANISTER
•ACKFILL OPERATION

PLASMA ARC WELDING OF 12.75 O.D.

CANISTER BODY AND CLOSURE LID

Fig. 23. Remote encapsulation process sequence.

 440

ftfMOVAl OF PLASMA ARC WHDING MACHINI

FROM 12.75 O.D. CANISTER ASSIMILY

RIMOVAL OF EVACUATION/BACKFILL COLLAR

FROM 12.75 O.D. CANISTCR ASSEMM.Y

LEAK INSPECTION OF COMPLETED

12.75 O.O. CANISTER ASSEMBLY

REMOVAL OF COMPUTED 12.75 O.D.

CANISTER ASSEMBLY FROM WELD PIT

Pig, 23. Remote encapsulation process sequence.

 441

CONCLUDING REMARKS

The selection of plasma arc welding for t h i s remote, automatic

closure application has been thoroughly substantiated by the results
described herein. The advantages of the single-pass autogenous butt
weld made using the keyhole technique are numerous in view of the
apparent difficulties :n controlling groove shape/volume and f i l l e r
metal placement with the alternative G M AW and GTAW processes in the
2G position. Much has been learned regarding the equipment and
procedural requirements for such a demanding application which would
make the design and implementation of a second generation system a
more straightforward undertaking. Nonetheless, the only significant
disadvantage of the process is the present necessity to use a
two-pass, filler-added termination procedure. We feel that a
contiguous termination procedure is definitely achievable through
further development.
Hardware having much less precisely controlled joint fitup
dimensions than described here can definitely be accommodated
through the use of automatic AVC, so long as the system i s locked
out during the transient stages of keyhole initiation and weld
termination. In applications where bead shape requirements are less
r e s t r i c t i v e , the applicability of the process (and of a simpler
termination procedure) appears even more favorable.
Although the nominal carbon steel thickness of the subject
design is 0.25 in, mockups having up to 0.32-in wall thickness were
welded successfully. For significantly heavier thicknesses a
two-pass approach would be adopted. Using a part-through groove
configuration, an autogenous butt weld would be made for the f i r s t
pass, thus rigidly fixing the geometry of the remaining groove which
would be filled with a second pass using the melt-in filler-addition
technique. Similar 2G welding of the other base metals, incuding
stainless steels, nickel-base and titanium alloys, appears eminently
feasible.

ACKNOWLEDGMENT

The authors wish to acknowledge contributions made by the

following personnel t o the work described h e r e i n : Jane M. Branscomb,
M. K. Brecbel, D. E. Dickey, L. R. E i s e n s t a t t , H. J . Grubicb,
D. F. Moran and F. L. Przywarty, Westinghouse AESD/Pittsburgh, PA;
and D. B. Casey, J. S. Curamings, H. H. Irby, H. C. Snow and
K. N. Stingley, Westingbouse/AESD Nevada Operations.
 442

REFERENCES

1. "Canister Closure Welding Process S e l e c t i o n , " AESD-TME-3034,

Meetinghouse Advanced Energy Systems D i v i s i o n , P i t t s b u r g h , PA
(March 1980).

2. "Development of a Remote Automatic Plasma Arc Closure Welding

C a p a b i l i t y , " AESD-TME-3048, Westinghouse Advanced Energy Systems
D i v i s i o n , P i t t s b u r g h , PA (August 1980).
a+4
 PERFORMANCE OF THE CaF2-CaO-SiO2 SYSTEM
AS A SUBMERGED ARC WELDING FLUX FOR
A NIOBIUM BASED HSLA STEEL
S.LIU K.C., C.B.DALLAM, & D.L.OLSON
Center For Welding Reseach
Department of Metallurgical Engineering
Colorado School of Mines
Golden, Colorado 80401

ABSTRACT
Using fused reagent grade CaF2-Ca0-Si02 submerged arc
welding fluxes, welds were made on niobium based HSLA steel
plate. Weld metal oxygen content was correlated to the flux
composition and used to interpret the weld metal
microstructure, weld bead penetration and arc stability
behavior. The nature of the flux-metal reaction associated
with the various alloying elements was also investigated.

INTRODUCTION

The submerged arc welding process is often selected for

applications requiring high deposition rates and for
materials not limited by its high heat inputs. Traditionally,
commercial submerged arc welds were chosen for situations
where mechanical property requirements, particularly
toughness, were less stringent. The recent demand for
tubular products for arctic service has promoted researchers
to produce excellent high toughness submerged arc welds.
These advances were achieved through proper selection of
materials, such as HSLA steels, welding consumables and
parameters. The increase in weld toughness was brought about
by changes in flux compositions from high silica content,
usually based on the MnO-SiO2 or CaO-MgO-SiO2 systems, to
fluxes of lower silica contents, commonly based on the
Lime-Fluorspar-Silica system. A typical flux composition
range used (1) to achieve high toughness low carbon steel
welds for low temperature service is 10-25 % SiO 2 , 20-30 %
CaF 2 , 20-25 %(A1 2 O 3 +MnO), and 30-35 % (CaO + MgO).
The improvement in weld metal toughness was related to a
reduction in the weld metal oxygen content, which is a
function of flux composition (1,2,3). Reduction of the weld
metal oxygen content can be related to a reduction in the
total oxide content of the flux achieved by adding fluorides
or reducing the silica content of the flux. Also, decreasing
the amount of transition metal oxide in a flux promotes a

445
 446

reduction in the weld metal oxygen content (lr2) . Figure 1

illustrates the large variations in weld metal oxygen content
with both FeO and CaF 2 additions to manganese silicate and
calcium silicate flux systems. The high weld metal oxygen
contents are typical of manganese silicate fluxes. Notice the
major reduction in weld metal oxygen contents for calcium
silicates compared with manganese silicates of the same
silica content, and the further decrease with the
substitution for silica by CaF 2 . These flux modifications
illustrate the conceptual approach to the development of
advanced submerged arc fluxes.

Tuliani et al (4) showed that higher basicity fluxes

resulted in a lower weld metal oxygen, sulfur, and silicon
which was confirmed by Garland and Kirkwood (5). Several
definitions of basicity have been proposed (4,6,7). None of
which are completely adequate because the basicity equation
was originally defined for sulfur concerns in the iron and
steel making processes. North et al (8) pointed out that it
is important to differentiate between basicity and the oxygen
potential during welding. Indacochea et al (9) also observed
violations of a general rule relating basicity to weld metal
oxygen contents. The question of whether CaF2 should be
included as a basic component is also under debate.

Weld metal inclusion density is related to the weld

metal oxygen content (10,11). A high inclusion density has
been found to promote the nucleation of ferrite, particularly
at the higher temperatures during the cooling cycle (11,12).
The resulting higher temperature ferrite morphology can be
blocky ferrite or Widmanstatten side plate ferrite.
Inclusions are commonly found in the center of blocky ferrite
grains (1). As would be expected, each type of inclusion, and
thus flux type, has its own relationship between inclusion
density and weld metal oxygen content. Since it is the
inclusion density which is basic to the nucleation process of
ferrite, it should be fundamental to any microstructural
property relationship and not weld metal oxygen. Weld metal
oxygen is only indirectly related to the tendency for a
formation of a specific ferrite microstructure.
Blocky and side plate ferrite are not as desirable as
acicular ferrite in improving weld metal toughness. If
nucleation of the ferrite can be avoided at the higher
temperatures of the weld thermal cycle, the undercooling for
ferrite formation increases until numerous nucleation events
occur within the austenite grain, resulting in a fine
non-equiaxed ferrite microstructure.

The transformation of austenite can give different

microstructures depending on the cooling rate and the
chemistry of steel. When the cooling rate is extremely slow,
 447

O FeO-MnO-40% SiO,
0 16 V CaF,-MnO-4O% SiO8
A CaF-,-Ca0-40% SiO,
a CoFj-CoO-30% SiOj
014 CoFj-CoO-20% SiOj
-o
012

010
z
o
008
o

UJ 0 0 4 •
s
a
1^0.02
MILL PRODUCTi

10 20 30 40 50 60 70 80
CaFj or FeO in FLUX

Fig.I. Weld metal oxygen as a function of

CaF2 and FeO additions in the flux.

1600 - 8 A Fe
\ \ T il V / / sK
u.
1400
c\
\ w,</
:
•URE

r-
X
1200
\/VA/\ AciculorX
Y /\ A ¥ / X
"EMPE

1000
:
800

10 20 40 60 80
TIME (sec)

Fig.2. Cooling rate effect on resulting

microstructure of a weld (12):
Curve A - lOC/s - coarse blocky
ferrite,
Curve B - 20C/s - acicular ferrite
with patches of
blocky ferrite,
Curve C - 45C/s - acicular ferrite
with patches of
martensite.
 448

the ferrite nucleates at austenite boundaries, growing into

blocky ferrite. Widgery (10) showed that the microstructure
is related to the temperature in which the decomposition of
austenite occurs. Large amounts of acicular ferrite results
when the transformation occurs over the range 630 - 450C,
while welds tranforming at a higher temperature range 710 -
545C showed only a small percentage of acicular ferrite (10).
This effect is shown schematically with a hypothetical
continuous cooling, transformation diagram (12) in figure 2.
He showed that upon slow cooling, blocky ferrite formed,
while a faster cooling rate produced acicular ferrite.
Increasingly rapid cooling rates produces bainitic and
ultimately martensitic microstructures.

Alloying elements affect the hardenability of steels; a

relatively large degree of which is necessary to form
acicular ferrite. However, depending on the carbon
concentration of the steel, bainite or lath martensite may be
formed, increasing the strength of the weld while
sacrificing toughness. Recent research is directed at finding
the proper balance of Mn, Nb, 0, B, Ti, Mo, V and Zr to
maximize the amount of acicular ferrite in the weld metal
(13,14,15,16,17). Figure 3 shows that an increase in oxygen
shifts the CCT curve to the left, increasing the amount of
resulting proeutectoid ferrite (18) . Several researchers
(13,18,19) have stressed that this oxygen effect is due to
oxide population rather than oxygen in solution, and the
proposed mechanism for this is that oxides act as nucleation
sites for allotriomorphic ferrite. Therefore, to increase the
amount of acicular ferrite, low oxygen will be desirable.
Cochrane and Kirkwood (11) used dilatometric techniques for
continuous cooling transformation studies and found that the
transformation temperature for the lower oxygen welds was
lower by 50C than the high oxygen welds, and that the
transformation proceeded at a slower rate.

Fine interlocking , laths of acicular ferrite provide the

maximum resistance to cleavage whereas blocky ferrite
provides a flat and continuous path for easy crack
propagation. Cottrell and Petch (20,21) independently derived
equations from dislocation theory which predict that
decreasing the grain size would increase the toughness of a
material. In addition, a decreasing grain size also increases
the yield strength. This is one reason that control rolled
microalloyed steel is used.

The ability of a specific flux to perform the many

necessary roles which include weld pool refinement, arc
stabilization, weld pool protection and bead morphology
control is directly related to the physical and
pyrometallurgical behavior of the flux. It was the purpose of
this investigation to characterize the influence of
 449

^^_ Submerged arc weld Proeutectoid

Ferrlte at
Grain Bdrles

Ul
a:
-rr7'rrS'+
<
a: SHe PloW;.|g*| ** Aclcular
UJ Ferrlta'
;£
s ^^^*-~~" ' '"

p
[ Balnlte /
< >
End of transformation
I /
MartensltB

LOG TIME (b) LOG TIME

Fig.3. Schematic CCT diagrams, showing for some general weld metal
composition, influence of inclusions in nucleating acicular
ferrite (18):
a) few nucleantes, eg. oxygen content 0.01Z,
b) effective nucleantes, eg. oxygen content 0.03%.

40%Si02

CaO
10 20 30 40 50 CoF,
60 70 80 90

Fig.4. Concentration contours in ppm showing the

variation of weld metal oxygen as a
function of flux composition.
 450

variations of CaF 2 and CaO additions on the behavior of the

CaF 2 -Ca0-Si0 2 flux system and to determine the relationships
between weld metal composition and its microstructure.

MICROALLOYED HSLA STEEL

High strength, low alloy steels, or microalloyed steels,
are low carbon (0.06-0.15%) steels which contain small
amounts of elements such as Niobium, Vanadium, or Titanium.
The concentration of these elements is rarely above 0.1%. In
spite of this minimal amount of addition, their ability to
form nitrides and carbonitrides limits the grain growth very
efficiently, producing a very fine grained microstructure.
This resulting microstructure shows superior strength and
toughness when compared to plain carbon steels.

Together with microalloying, thermomechanical processing

( controlled rolling at low finishing temperatures ) provides
a further step of upgrading the final quality of the product.
The strength of such a product is around 500 MPa ( 72.5 ksi )
and the impact transition temperature is as low as -80C. Of
course, such improvement in mechanical properties are
beneficial.. However, good weldability is another essential
property for applications such as arctic line pipes, offshore
structures and pressure vessels. These applications commonly
use submerged arc welding because of the high heat input
conditions and high productivity characteristic of the
process. Due to the high heat input, several problems
concerning toughness degradation have been attributed to
niobium in the steel (22,23,24,25).

Current literature shows no unified opinion on niobium

effects on weld metal toughness. Some authors (22,23,24)
found that niobium was always harmful to impact toughness,
while others (25) reported that up to certain limit, niobium
is beneficial. Garland and Kirkwood (13) have proposed a
model concerning niobium behavior in weld metal toughness. In
principle, nicbium deteriorates the toughness by
precipitating fine Nb(C,N) particles, increasing the
strength. At the same time, niobium improves the toughness by
promoting acicular ferrite formation. The net effect of the
niobium addition depends on which contribution is greater.
Post weld heat treatments of niobium steels usually result in
considerable toughness degradation due to the carbonitride
precipitates. Therefore, multiple pass welding is a serious
concern in welding niobium based microalloyed HSLA steels.

With the CaF 2 -Ca0-Si0 2 system, the oxygen level of the

weld metal is significantly lowered, and so is the amount of
proeutectoid ferrite. The effect of niobium in such a low
oxygen system is not known. The influence of additions of
elements such as titanium and zirconium in the presence of
 451

niobium is also not totally clear.

MATERIALS AND PROCEDURE

The CaF2-Ca0-Si02 fluxes were produced using reagent

grade chemicals of nominal compositions at every 10%
composition interval on the ternary phase diagram. The
restrictions imposed were: 1) The flux must have a melting
point of 1450C or less; 2) less than 40% by weight silica.
Melting was carried out in an induction furnace using
graphite crucibles with nitrogen as a cover gas. A two color
optical pyrometer was used to determine the temperature of
the melt. The temperature in each case was raised to at least
1550C and viscosity measurements were made at 1450C upon
cooling using a Brookfield viscometer.

The molten flux was tapped into water. This quenching

broke the flux into pieces which were easily reduced in size.
Before sizing to -14, +100# the fluxes were dried for 2 hours
at 250C. After crushing all fluxes were baked at 750C for at
least 3 hours to insure moisture removal, but more
importantly to remove residual graphite - intrinsic
contamination of the process. Fluxes were processed further
under a magnetic separator prior to welding. This was to
remove any iron oxide which may have been introduced during
baking.
Lukens Frostline, a niobium microalloyed steel plate was
used in this study. The plate was received in quenched and
tempered condition. AWS grade E70S-3 welding wire of 2.38 mm
diameter was used as the filler. Compositions are shown in
Table I.

TABLE I. Compositions of base plate and welding wire.

Material C Mn Si Nb P S Al Ni Mo Cu
Lukens
Frostline ,14 1.38 .41 .042 .005 .03 .041 .14 .06 .18
E70S3
Wire .11 1.15 .53 —

Bead on plate welds were made using each of the fluxes

on 100 x 225 x 12.5 mm plates. The welding was done on a
Hobart submerged arc welding system using DCRP (Direct
Current Reverse Polarity) at 30 volts, with a travel speed of
4.8 m/min and the filler metal wire speed was adjusted to
maintained a current of 500 amp. The heat input was 1.87
kJ/mm. Chemical and metallographic analyses were performed on
the resulting weld metal.
 452

RESULTS
The performance of 28 reagent grade fused submerged arc
welding fluxes of the CaF 2 -Ca0-Si0 2 system was characterized
here in terms of their physical, chemical, and
microstructural influences. The weld penetration, interfacial
tensions, arc instability, refining ability of the fluxes ,
weld metal oxygen content and microstructural features were
quantitatively measured and analysed.

Figure 4 shows the variation of the weld metal oxygen

content with flux composition. Welds made with 0% SiO 2 fluxes
showed extremely low oxygen contents.The weld metal oxygen
was seen to increase with increasing silica content.
Increasing the CaO content also increased the weld metal
content but to a lesser degree than SiO 2 .

Patchett et al (26) have related oxygen activity to weld

bead penetration. They suggested that cathodic electron
emission from the weld pool is strongly influenced by the
oxygen activity. The oxygen activity of the cathode surface
is related to the weld metal oxygen content and indirectly to
the flux composition. Increasing the weld metal oxygen is
believed to decrease the arc instability and thus increase
the penetration.

Arc instability is defined as the average voltage

fluctuation of the arc from the set machine voltage. A lower
voltage fluctuation results in a more stable arc or lower
instability. Arc instability is obtained from the area under
a voltage-time curve generated during welding divided by the
time. Arc instability is important to weld quality, as large
fluctuations in voltage would lower the average heat input
density which in turn results in decreased and perhaps
irregular penetration.

The arc instability was first analyzed as a function of

weld metal oxygen content at various silica content in the
flux. There were general indications, as seen in figure 5 for
fluxes with 40 % SiO2 that the arc instability actually
increased with increasing oxygen. This is just opposite of
the expected behavior suggested by Patchett et al (25) . This
contradiction has been observed in other systems involving
CaF 2 (1), such as MnO-SiO 2 -CaF 2 . It is interesting that there
is still reasonable correlation between arc instability and
oxygen. If Patchett1s model is correct, then this would
suggest that some element besides oxygen may have affected
the arc behavior. For this reason, plots of arc instability
versus weld metal oxygen at constant CaF2 contents are
examined. Each of these plots shows that arc instability is
decreased with oxygen, figure 6, in agreement with Patchett's
 453

d 6
m

o
IE

200 300 400 500

OXYGEN CONCENTRATION (ppm)

Fig.5. The variation of arc instability

with weld metal oxygen content for
the 40 w/o SiO,-, fluxes.

100 200 300

OXYGEN CONCENTRATION (ppm)

Fig.6. The variation of arc instability

with weld metal oxygen content for
the 60w/o CaF 2 fluxes.
 454

model. This seems to indicate that,CaF2 (or fluorine) must

influence the arc behavior along with oxygen.
The variation of penetration with oxygen content in the
weld metal was measured. The constant silica plots show
diverse effects. The 40% silica flux welds showed increasing
penetration with oxygen. The 30% silica flux welds showed the
same effect to a lesser degree. No oxygen effect could be
observed in the 20% silica flux welds, while a decreasing
penetration with increasing oxygen was seen for the constant
10% silica fluxes, figure 7. Again, plots were made of
penetration variation with weld metal oxygen at a constant
CaF2 content. With the exception of the 50% CaF2 fluxes, each
plot showed the trend in increasing penetration with
increasing oxygen as seen in figure 8. This also seems to
agree with Patchett's model.

Penetration of the weld bead has been seen to be

affected by the composition of the flux (1,27) and Schwemmer
et al (27) explained this behavior in terms of flux
viscosity, arc instability, and capillarity. They found
linear relationships between penetration and these physical
properties and proposed an empirical equation to describe
penetration behavior. This relationship was not tested in
detail in this study because of problems in obtaining
reproducible viscosity data at 1450C. For this flux system,
several compositions melt near this temperature, which makes
many of these fluxes very temperature sensitive, causing
problems for the accuracy and reproducibility of the
viscosity measurements. No apparent correlation between
penetration and capillarity was observed in this work, but
there is some apparent relationship between penetration and
arc instability as seen in figure 9. This correlation is
plotted for a constant fluoride flux and is consistent with
observations of Schwemmer.

In Gas Tungsten Arc welds, a model has been developed

(28) proposing that fluid flow is the major factor in
determining the shape of the fusion zone. Fluid flow 'n the
weld pool can be caused by surface tension gradients on the
weld pool surface and this phenomena is known as the
Marangoni effect. Temperature differentials across the weld
pool bring about these surface tension gradients. These
gradients are sensitive to small amounts of surface active
elements in the weld pool such as oxygen, sulfur, or selenium
in liquid iron. The surface tension is reduced by these
elements and its temperature dependence is altered. For
example, sulfur and oxygen in iron cause the surface tension
to increase with temperature. Since fluid flows from low to
high surface tension regions, an inward flow is produced,
increasing the penetration. In the case of submerged arc
welds, the presence of molten flux on the liquid metal
 455

0 4 !• • • •

10% SiO2

o
•4 03
cr
l- O O^-\
LU
z
UJ
a
o
0.2
100 200 300
OXYGEN CONCENTRATION (ppm)

Fig.7. The variation of penetration with

weld metal oxygen content for the
10 w/o Si0 2 fluxes.

0.4

0.2
150 250 350
OXYGEN CONCENTRATION (ppm)

Fig.8. The variation of penetration with

weld metal oxygen content for the
30 w/o CaF fluxes.
 456

04 -

0.2
5 6 7 8 9
ARC INSTABILITY (volts)

Fig.9. Penetration plotted as

a function of arc iusta-
bility for the 40 w/o
CaF_ fluxes.

40%Si0. 40% SiOj

CaO CaF,
10 20 30 40 50 60 70 80 90

Fig.10. Part of CaF^CaO-SiO, ternary phase diagram

showing the silicon Behavior in the weld
metal comparing analytical and nominal com-
positions.
 457

surface modifies the interfacial tension. This effect cannot

be directly observed in submerged arc welds as in GTA welds,
but it serves as another possible explanation for penetration
variations with weld metal oxygen and flux composition. It
may also explain the influence of fluorine since it would
also be expected to be a surface active element.
WELD METAL CHEMISTRY
The weld metal chemistries were determined. The nature
of the flux-metal reaction was determined by comparing the
measured weld metal composition with the nominal composition
based on the dilution of the wire and base metal
compositions. If the difference is positive, the flux
contributes to the weld pool. On the contrary, a negative
deviation indicates a loss of the specific element from the
weld pool to the flux. Negative deviations were seen for
carbon, manganese and niobium, Table II.

TABLE II. Ranges of analytical and nominal compositions for

C,Mn,Si,Nb,P and S in the weld metal.

Chemical Analytical Nominal Deviation(*)

Element Low| Hiqh Low iHigh Low Hiqh
09 .12 - — —
c L « J
.77 ' 1.05 1.24 1.31 -.22 -.54
Mn
Si i .20 .61 .45 .50 -.18 + .16
Nb .008 .025 .01 _*03 -.003 -.015
P j .020 .031 — _ _ _
S 1 .011 .022 - - - —
Deviation = %analytical - % nominal

Silicon, however, had both positive and negative

deviations as shown in Figure 10 .When the SiO 2 content is
approximately equal to the CaO content, no silicon loss is
observed. Above that region, % SiO 2 /% CaO greater than 1,
there was an increase of silicon in the weld pool indicating
that silicon was transfered from the the flux to the weld.
When % SiO2/% CaO is less than 1, a silicon loss was seen.
The reason for the negative deltas is not clear. However, in
the presence of high CaO content fluxes (calcium silicates
fluxes), the SiO 2 activity is lowered, increasing the
tendency for silicon removal from the weld pool to the slag.
This behavior of negative and positive deviations was not
observed in the MnO-SiO2-CaF2 system. Figure 11 shows a
uniformly positive silicon deviations indicating the transfer
of the element from the flux to weld metal.
 458

40% SiO,

in 0.3
to
2
•a: MnO-Si0 2 -CoF 2
o 0.2 FLUX WELDS
o
i-
<t 0.1
cr
2
IU
o
CaO-SiO2-CoF2
o FLUX WELDS
o

-0 1
10 20 30 40 50 60
CoF2 in FLUX («/«)

Fig.11. The deviation between analytical

and nominal weld metal silicon
content plotted as a function of
w/o CaF ? in the flux.

10
40% SiO2
08 --

0.6 -

FLUX WELDS
< 0 4 -
I
o

2 0 2 -. o
o
o
O
7 •
o
o 0
O

Hi
o C Q 0 'Si Op C 0 ^p
z
S -0.2
FLUX WELDS
~7 -
c
S
D /
-0.4 0 u
Q * a

a
-0.6 — — j _

10 20 30 40 50 60
Cofj .n FLUX 0"7o)

Fig.12. The deviation between analytical

and nominal weld metal manganese
content plotted as a function of
w/o CaF in the flux.
 459

The presence of MnO in the manganese silicate fluxes

contributed to an all positive manganese deviation. In welds
made with calcium silicate fluxes, a uniform manganese loss
was observed. Since in this latter case, the only sources of
manganese are the base plate and the welding wire, positive
deltas would not be expected. The manganese behavior for the
two systems is shown in figure 12.
As a first approximation, the chemistry of the weld
should be equal to the base metal chemistry (disregarding
microstructural and solidification effects). In order to
accomplish this, manganese and niobium must be introduced to
the weld metal. This can be done by alloy additions to the
filler wires , or by adjustment in the flux composition. The
weld metal silicon content can be controlled, in this case,
quite adequately by adjusting the flux composition. This can
be important in cases where chemistry must be held in a
narrow range.

WELD METAL MICROSTRUCTURE

Representative microstructures are shown in figure 13.
The main constituents are blocky proeutectoid ferrite,
Widmanstatten side plate ferrite and acicular ferrite.
Quantitative metallography was done on samples across the
%Si02/%Ca0 = 1 line which showed a variation of oxygen
contents. Figure 14 indicates that with an increase in
oxygen, the amount of proeutectoid ferrite (both blocky and
Widmanstatten side plates) also increases. This is discussed
earlier as the nucleation of allotriomorphic ferrite on
pre-existing oxide particles.

Veining of proeutectoid ferrite is shown in figure 13a.

Both the amount and distribution of blocky ferrite are
important to mechanical properties. Veining should be held to
a minimum to maximize toughness. However, veining will always
be present due to the solidification nature of the weld pool.
The relationship between proeutectoid ferrite veining and
toughness is not very easy to quantify, since it depends on
the orientation of the veins with respect to the crack
propagation direction.

DISCUSSION

If we are to select welding consumable compositions

which consistently produce weld metal meeting the stringent
requirements of todays engineering materials, it is essential
that we first characterize consumable behavior, then
establish models and empirical relationships so that optimum
consumable
consumaDie compositions
compositions can
can be
oe developed.
developed. This
This attempt
attempt to
to
characterize the CaF2-Ca0-SiC>2 welding flux system was found
more difficult than previous investigations on the manganese
 460

Fig.13. Typical microstructures of welds made by

using fluxes of CaF^-CaO-SiOj system:
a) Proeutectoid ferrite v e i n m g , acicular
ferrite,
b) Widmanstatten side plates, blocky grain
boundary ferrite and acicular ferrite.

100 200 300 400

OXYGEN CONCENTRATION (ppm)

Fig.14. The variation of proeutectoid ferrite with

weld metal oxygen content.
 461

silicate system. Only with careful selection of the data,

such as observing the behavior with constant CaF2 content,
does any order appear in the results. This problem should be
expected from studying the CaF 2 -Ca0-Si0 2 ternary phase
diagram. Figure 15. The diagram is very complex with numerous
liquidus projections and four phase reactions. It becomes
apparent that the typical requirement of a flux melting below
1450C leaves only specific compositional regions available
for welding. This problem showed up during viscosity
measurements, which were done at 1450C. When fluxes with
melting points near this temperature were tested, the
viscosity became very sensitive to temperature. This resulted
in large variations in measured viscosity values. In the case
of Weld metal made with fluxes that have temperature
sensitive viscosity behavior, the weld metal chemistry and
microstructure would be expected to be very sensitive to the
heat input control.

A two liquid phase region appears on the silica-

fluorspar side of the phase diagram. This region suggests
that a light liquid is on top of a heavier liquid, which is
involved in the flux-weld metal reactions. It is not known
which flux composition should be correlated with weld pool
behavior, weld metal chemistry, and microstructure for fluxes
lying in the two phase region.

Fluxes with less than 10% silica were found to be

insufficient in silica network to be stable in a glass phaae
on cooling, crystallizing and losing much of their
effectiveness as a protective flux. During the drying
operation, these fluxes degraded into a fine powder, and the
welds made with these powders showed irregular bead shapes
and severe porosity.
Figure 16 illustrates, on a ternary phase diagram, the
suggested flux compositional regions which show the greatest
promise. Fluxes on the CaO-CaF2 side of the ternary would not
be advisable because of its lack of a silicate network. A
region on the CaF 2 -Si0 2 side should be avoided because of the
two liquid regions. Welds made in the upper left region
should be avoided in the welding of HSLA steels because the
excessive weld metal oxygen content would yield a higher
amount of proeutectoid ferrite than desired for maximum
toughness. Regions on the left side ought to be avoided
because of steep ridges in the melting temperature . For the
CaF 2 -Ca0-Si0 2 flux system, it becomes apparent that a strict
compositional control on the fluxes is necessary.

CONCLUSIONS
1
1. The CaF2-CaO-SiO2 welding flux system is found to be more
complex in physical behavior than the manganese silicate flux
 462

Two Liquids

CaO SiO 2

3CaO ?SiO 2
Liquids
2CaO SiO
2O7(
3CaOSiO 2

CaO
(-2570°)

Fig.15. CaF 2 -Ca0-Si0 2 ternary phase equilibrium

diagram showing the liquidus lines, 2
liquid regions and the isothermal
contours.(29)

4 0 % SiOa 40 % SiO,

2 LIQUIDS
REGION

CaO
70 80 90

Fig.16. Part of CaF^-CaO-SiC^ ternary phase equi-

librium diagram showing the flux compositions
which resulted in good quality welds.
 463

system.
2. A definite compositional range was found to produce good
quality niobium microalloyed HSLA steel welds.
3. This flux system produced welds with extremely low oxygen
content (100 - 460 ppm) and also a fine microstructure,
predominantly acicular ferrite.
4. At constant silica content, penetration and stability of
arc were found to decrease with oxygen. However, when the
amount of CaP2 was held constant, those properties showed an
increase with increasing oxygen, indicating that along with
oxygen, fluorine also influences the arc behavior and
penetration.

5. Niobium and manganese showed negative deviations in

chemistry during welding, indicating a loss from the weld
pool to the slag. For silicon, a definite pattern of positive
and negative deviations in chemistry was seen.

ACKNOWLEDGEMENTS
The authors acknowledge and appreciate the research support
of United States Army Research Office, the fellowship support
of CNPq-Conselho Nacional de Desenvolvimento Cientifico e
Tecnologico of one of the authors (S.L.K.C.), material
support from Lukens Steel Corporation, and equipment support
of Hobart Brothers Company.

REFERENCES

1. INDACOCHEA J.E. & OLSON D.L., " Relationship of Weld Metal

Microstructure and Penetration to Weld Metal Oxygen Content"
ASM Conf. on "Welding Consumables and Process Developments",
Peoria, Illinois (Sept. 1981).

2. INDACOCHEA, J.E., Ph.D thesis, Colorado School of Mines

(1981).

3. EAGAR, T.W., Welding Journal, v.57, 76s (1978).

4. TULIANI, S.S., BONISZEWSKI,T. , EATON, N.F., Weld and Metal
Fabrication, v.37,327 (1969).

5. GARLAND, J.G. & KIRKWOOD, P.R., Weld and Metal

Fabrication, v.44,217 (1976).

6. EAGAR,T.W. & CHAI, C.S., Met.Trans.B, V . 1 2 B , 5 3 9 (1981).

7. HAYES,M.D., HICKEY, J.T., JACKSON,C.E., "The Influence of

 464

Flux Basicity on Submerged Arc Welding Performance" COSOL -

6th Encontro Nacional sobre Tecnologia de Soldagem, Salvador,
Bahia, Brasil, (Oct.1980).

8. NORTH,T.H., BELL,H.B.,NOWICKI,A.,CRAIG, I., Welding

Journal, v.57,G3s (1978).
9. INDACOCHEA J.E. & OLSON, D.L., Unpublished Work, Colorado
School of Mines (1981).
10. WIDGERY, D.J., "New Ideas on Submerged Arc Welding" Intl.
Conf. on "Trends in Steel and Consumables for Welding" Paper
26; Welding Institute, London (Nov. 1978).
11. COCHRANE, R.C. & KIRKWOOD, P.R., "The Effect of Oxygen on
Weld Metal Microstructure" Intl. Conf. on "Trends in Steel
and Consumbales for Welding " Paper 35; Welding Institute,
London (Nov.1978).
12. THOMAS,Jr.D.R., Metal Progress, v.Ill,30 (1977).
13. GARLAND,J.G. & KIRKWOOD, P.R., Metal Construction,
v.l ,215 (1975) .
14. ENTREKIN,Jr,C.H., Metallography, v.12, 295 (1979).

15. LEVINE,E. & HILL,D.C, WRC Bulletin 213, 20 (Feb.1976).

16. LEVINE,E. & HILL,D.C, Metal Construction, 346 (1977).

17. KOHNO,R., TAKAMI,T., MORI,N., NAGANO,K., "New Fluxes of

Improved Weld Metal Toughness For HSLA Steels" AWS Annual
Meeting, (April 1981).

18. ABSON,D.J., DOLBY,R.E4, HART,P.H.M., "The Role of

Non-Metallic Inclusions in Ferrite Nucleations in Carbon
Steel Weld Metals" Intl. Conf. on "Trends in Steel and
Consumables for Welding" Paper 25;Welding Institute, London
(Nov. 1978) .
19. KIRKWOOD, P.R., Metal Construction, v.10,260 (1978).
20. COTTRELL, A.H., Trans. Met. S o c , AIME, v.212,192 (1958).

21. PETCH,N.J., JISI,v.l73,25 (1953).

22. HANNERZ,N.E., VALLAND,G., EASTERLING,K., IIW-I1612-72
(1972).
23. JESSEMAN,R.J., "Columbium Pick Up in High Dilution,
Submerged Arc Welding Deposits" Proceedings of MICROALLOYINGS
75, 74 (1975).
 465

24. SAWHILL,J.M., Report L-176-115 Climax Molybdenum Company

(1973).
25. GRAY,J.M.r STUART,H., HEISTERKAMP,F., "The Control of
Toughness in Submerged Arc Weldmer.ts in Niobium Containing
Steels" Intl. Conf. on "Trends in Steel and Consumables for
Welding", Paper 31; Welding Institute, London (Nov.1978).

26. PATCHETT,B.M., DANCY,E.A., SCHWEMMER,D.D., OLSON,D.L.,

WILLIAMSON,D.L., Welding Journal,v.59,31s (1980)
27. SCHWEHMER,D.D., OLSON,D.L., WILLIAMSON,D.L., Welding
Journal, v.59,153s (1979).
28. HEIPPLE,C.R. & ROPER,J.R., Welding Journal, v.6C(8),143s
(1981) .

29. HILLERT,L., Acta Chem.Scandinavian,v.20(2),295 (1966).

 TOUGHNESS OF SUBMERGED ARC WELD METALS OF CONTROLLED ROLLED
NB BEARING STEEL

Tadamasa Yamaguchi, Atsushi Shiga, Dr. Akio Kamada

and Dr. Junichiro Tsuboi

Welding Section
Research Laboratories
Kawasaki Steel Corporation
Kawasaki-cho 1 Chiba JAPAN 260

ABSTRACT

Single pass and multipass submerged arc weldings of controlled

rolled Nb bearing steel plates have been carried out. At first, the
influence of Nb on the tnicrostructures of single pass weld metal have
been investigated. Then the effects of such factors as maximum
reheating temperature and stress relieving time on the impact
toughness of single pass weld metal have been examined. Finally the
hardness behavior of multipass weld metal has been studied in
relation to the change of alloy content in each pass.
It is found that acicular ferrite which improves weld metal
toughness is formed only when Ti is contained in weld metal.
Nb seems to have the effect to suppress the formation of
proeutectoid ferrite which nucleates at prior y grain boundary, but is
of no use promoting acicular ferrite formation.
The toughness and the hardness of reheated weld metals depend on
the maximum reheating temperature. When the maximum reheating
temperature is 500°C to 700°C, the hardness of single pass weld metal
increases and the toughness decreases because of fine Nb- and V-
—areaAnitride precipitation. When the maximum reheating temperature is
over 800°C, the hardness and the toughness remain almost unchanged.
The stress relieving treatment of single pass weld metal at 600°C
for 1 up to about 100 hours causes the increase in hardness and then
decreases the hardness gradually. It needs over 500 hours to obtain
the same hardness value as that of as-welded metal.
The addition of Ti to weld metal is very effective to improve the
toughness, however excess Ti increases the hardness of stress relieved
weld metal by precipitating as fine Ti-carbonitride. Therefore Ti
addition should be restricted within the lowest limit required to
improve as-welded metal toughness. The optimum Ti content is about
0.020% in the case of weld metal of which oxygen content is 350ppm or
so.

467
 468

In multipass welding, the hardness of weld metal affected by

subsequent weld heat cycle varies from pass to pass, because Nb and V
content change with the passes as the result of the change in
dilution from base metal. The most hardened zone is observed in the
reheated first pass weld metal, in which Nb and V content are the
highest.
Good weld metal toughness would be obtained by lowering dilution
from base metal and taking advantage of grain refinement by sub-
sequent passes.

1. Introduccion

Controlled rolled Nb bearing steel plates have been widely used in

manufacturring line pipes for low temperature service because of their
high strength combined with good toughness.
When such plates are welded by submerged arc welding process,
about 70% of Nb in base plate is diluted into weld metal so that
influences the mechanical properties.
Although a number of investigations on the effect of Nb and V on weld
metal toughness have been carried out in both as-welded and stress
relieved conditions 1 ^ 5 ), the results reported have not always coincided.
Some investigators2'3'have reported that small amount of Nb (-0.02%) in
weld metal can be beneficial to toughness, while others'*)5) have shown
only detrimental effect of Nb and V. One of the reasons for such
discrepancy seems to be attributed to the difference in materials or
welding conditions used. In order to avoid or reduce the deterioration
of toughness, it is important to know what factors are most significant
for the toughness of weld metal containing Nb and V.
This paper deals with the influence of wire and flux combination,
maximum reheating temperature, stress relieving time at 600°C and Ti,
oxygen and nitrogen content on the toughness of single pass weld metal
containing Nb and V.
And the toughness of multipass submerged arc weld metals is also
examined from the view point of the Nb and V content in each pass.

2. The Effect of Nb on the Microstructure and the

Toughness of Submerged Arc Weld Metals.

Among many remarks on the effects of Nb on microstructure and the

toughness of weld metal, the following points can be said as common.
(1) Nb influences the toughness indirectly through the microstructure
change because Nb lowers the starting temperature of ferrite
transformation.
(2) When Nb precipitates as fine carbonitride which has the coherency
with the matrix, the hardness increases and the toughness decreases.
In discussing the role of Nb on the microstructure, it is
important to consider the content of alloying elements such as Mn and Mo,
which lower the transformation temperature, and inclusions (oxygen
content) or TiN which seems to be the acicular ferrite former.
 469

This chapter deals with the effects of Nb on the microstructure

and the toughness of submerged arc weld metals by changing the content
of alloying elements, the amount of inclusions and TiN by combining
different kinds of fluxes and wires.

Materials and Experimental Procedure

Five kinds of steels given in Table 1 were used. Fluxes used are
listed in Table 2. The presence of TiO 2 , and the basicity are the
significant factors which influence weld metal micrstructure.
And the chemical compositions of wire used .ire shown in Table 3.
They have various content of alloying elements such as Mn, Ni, Mo and Ti.
Single pass submerged arc welds were laid in 7 mm deep, 70°V groove
with 48Kj/cm heat input except
Table 1 Chemical compositions of
steel 5 which was welded with
steel plate used.
30Kj/cm heat input.
Steel Plan Charpy impact test, hardness
No C Si Mn P S Cu Ni Nb V Al N " thick lm mi
SI 012 Q16 1.15 0014 0006 — - - - 3042 N D 19 measurement, microstructure
S2 012 0.42 144 0013 Q010 - - - — X>33 NO 19 observation and chemical
S3 0.07 0 25 1.78 3015 3 004 - - 0D56 - 0033 NO. 19 analysis of weld metals were
S4 006 026 1 66 00W 0005 0 30 0 29 0.051 0072 0024 46 19 carried out.
S5 006 0 32 158 0.018 0.005 001 0 52 0170 0005 0010 80 13

Table 3 Chemical compositions of Table 2 Chemical compositions

wires used. of fluxes used.
Wire No. C Si Mn P S Ni Mo Ti Al B* N"
Flux No SiO2 TiO2 AUCB MnO MgO|CaO CQF2
W1 Q10 0.30 1 50 0007 Q012 - - - 0020 - N.D.
F1 34 5 6 2 2 36 15
,W2 0 0 7 0 25 1.10 Q007 0006 085 0.42 0 2 0 0040 - 83
W3 0.07 0.25 1 8 0 Q007 0.005 - 0.45 0.21 0.040 - N.D. F2 15 - 15 — 34 16 17
W4 Q06 0 2 5 1 51 0.010 0.005 - 1.00 - Q030 - N.D F3 11 1 17 1 32 7 24
W5 004 0 0 5 133 0.005 0005 0.09 0.32 0033 0017 45 91
»ppm

Results and Discussion

The combinations of welding materials which are classified into four

series A, B, C, and D according to the content of alloying elements, the
chemical compositions of weld metals and the test results obtained are
listed in Table 4.
The microstructure of each series are given in Fig. 1, and the main
changes of the microstructure and the toughness due to the combination of
welding materials are sumarrized in Table 5.
From the results described above, the followings are obtained about
the changes in microstructure and toughness.
(1) Oxide inclusions promote the formation of proeutectoid ferrite.
(2) Ti changes the microstructure from lath to acicular ferrite and
improves toughness
(3) B suppresses the formation of grain boundary ferrtie, arJ improves
toughness by raising acicularity.
(4) Nb also suppresses the formation of grain boundary ferrite, but is no
use in promoting acicular ferrite formation. The excess addition of
 470

Table 4 Combination of welding materials and test results

Wrld Materials Chemical composition <*/.) Toughness Hard-
No. Slew Flu» Wir* c Si Mn P S Ni Mo Nb V Ti Al a"3 N . 3 0** vE-25
(K&m) (HvW)
A1 SI Fl Wl 010 032 1 14 Q015 Q008 - - - <O0O5 0.016 - 47 381 30 • 25 168
A2"' SI Fl Wl 010 0J3 1 10 0016 Q008 — - 0020 — <O005 0015 - 50 376 20 .40 190
A3 SI Fl W2 008 0 35 1 OS 00)4 0008 0 2 9 0.14 - - Q031 0020 — 46 385 10 8 -25 192
A4"1 SI F1 W2 009 036 1 W 0017 0008 0 26 014 0020 - 0030 0017 - 45 377 ISO -30 208
81 S2 F2 W1 010 0.33 1 15 0017 0007 - — - - «MO5O015 - 55 214 18 • 20 190
B2 S3 F2 Wl 0.07 0.28 1.54 0.016 0008 — - Q034 - <QO05Q01» — 52 220 21 .30 218
B3 S2 F2 W4 0D7 0.31 1 10 0018 0008 — 032 - - <O0O5OOI5 - 56 210 15 .30 212
B4 S3 F2 W4 0.06 0 » 1.49 D015 0007 - 0.34 0036 — <Q.00500K - 53 225 13 .30 236
65 S3 F2 W3 006 0 3 0 1 61 0.017 0006 — 0 I S 0034 — QO33OOI6 - 55 218 12.4 -20 233
C1 S4 F3 ws 0 0 5 0.21 1.49 Q017 0004 0 21 0.19 0023 0045 0014 0018 i i 78 245 18.6 -80 219
C2 S5 F3 W5 0.06 0.25 1.37 0017 0005 0 3 4 012 0079 0006 00150014 9 B0 251 8.3 -36 221
D1 S4 F1 W3 005 041 1.51 0022 0004 0 2 0 016 0024 0050 0 0 3 3 0 0 0 — 63 374 82 -35 216
D2 S5 Fl W3 0.06 0 4 3 1.44 0021 0006 0.32 0 10 0072 0008 0035 Q006 - 86 389 40 -10 2 23

or* S4 Fl W3 004 0.37 154 00190006 0.22 0.28 0024 00490O32 0016 37 91 373 130 -80 238
S5 Fl mt 0.05 0 J 8 144 0 0 » 0008 0 . 2 * 0.21 0070 0008 0 f l « l 0.010 32 92 395 72 -25 236

Fig. 1 Microstructure change of weld metals.

 471

Table 5 Effect of alloy addition on microstructure and toughness

of weld metal
Se- Weld Addition of alloying
ries No Microstructure Toughness
element
Al standard weld metal proeutectoid ferrite and low toughness
lath like ferrite
A2 Nb addition tc Al increase in lath like low toughness
A ferrite
A3 Ni, Mo, Ti addition proeutectoid ferrite and increase in
to Al acicular ferrite toughness
A4 Nb addition to A3 decrease in proeutectoid increase in
ferrite toughness
Bl standard weld metal lath like ferrite low toughness
B2 Nb addition to Bl lath like ferrite low toughness
B B3 Mo addition to Bl lath like ferrite low toughness
B4 Mn,Nb addition to B3 lath like ferrite low toughness
B5 Ti addition to B4 promotion of acicular increase in
ferrite toughness
Cl standard weld metal acicular ferrite high tough-
containing Ni,Mo,B ness
C and Nb
C2 Nb increase in Cl change into lath decrease in
like ferrite toughness
Dl base weld metal acicular ferrite and better tough-
containing Ni,Mo, lath like ferrite ness
Ti, and Nb
D2 Nb increase in Dl increase in lath decrease in
D like ferrite toughness
D3 B addition to Dl increase in acicular increase in
ferrite toughness
D4 Nb increase in D3 change into lath decrease in
like ferrite toughness

Nb to the weld metal with high acicularity causes lath ferrite

formation and lowers toughness.
The more detailed investigations on the state of Ti seem to be
required in connections with cooling rate, the amount of inclusions,
nitrogen content, Al content and the amount of alloying elements in order
to clarify the mechanism of acicular ferrite formation.

3. The Effects of Maximum Reheating Temperature and Stress Relieving

Time on the Hardness and the Toughness of Nb and V Containing Weld
Metals

Materials and Experimental Procedure

Single pass submerged arc welds were laid in the same groove with
the same heat input as described in previous chapter, and two kinds of
weld metals with different microstructure, lath and acicular type,
were obtained by changing fluxes containing TiO 2 .
The effect of simulated thermal cycle, the cooling time of which
 472

from 800°C to 500°C is 60 sec, and stress relieving time at 600°C on the
hardness and the toughness changes were examined.

Results and Discussion

The chemical compositions and the microstructures of the two weld

metals are shown in Table 6 and Fig. 2.
Lath type weld metal contains 231 ppm oxygen and presents lath
like microstrueture, while acicular type weld metal contains 373 ppm
oxygen and shows acicular ferrite structure with a little proeutectoid
ferrite.
The hardness and the toughness of weld metal reheated by simulated
thermal cycle are greatly influenced by the maximum reheating temperature
as shown in Fig. 3.
The hardening up to 700°C is caused by the precipitation of fine
Nb- and V-carbonitride, and the softening between 800°C to 1000°C seems
to be attributed to the followings : (1) The microstructures change into
polygonal ferrite as given in Fig, 4 and dislocation density in these
structures is lower than that of original structure. (2) Nb precipitates
become large and lose coherency with the matrix as seen in Fig. 5, (3) V
exists as solid solution in the same way as in the as-welded metal.
Because the weld metal reheated over 1200°C presents the same
microstructure as as-welded metal, and Nb and V remain in solid solution,
the hardness and the toughness also remain unchanged.
Fig. 6 shows the relationship between stress relieving time at 600°C
and the hardness or the toughness of both weld metal.

Table 6 Chemical compositions of

weld metal used. 10
Main St. C Si Mn P s Ni MO AI Nb V H* 0*
Lath 0.06 0.28 1.62 0.01? 0006 022 0.12 0030tos XK2 39 231
Acicular 008 0.38 M4 00)6 Q007 022 013 0032 3029 0051 373
«ppn
type

l
'V.' ;'.•'.• • ' . ' '•.,..'•

JZ
o

1 2*°
5
3
U
HIRi '.'••••'';.

j
'::yj-:

. . . .
5
|
|
I
230
220
210
200
AW 600 800 TOOO 1200
'•-i:••!••.'•.',.'• • -, sow
P e a k temperature (*C)

Fig. 2 Microstructure of weld metal Fig. 3 Effect of maximum reheat-

ing temperature on hard-
ness and toughness of weld
metal.
 473

Lath Aclcular

eodb
* . • ' / - • : • • [ , : . • > .

70CTC

80(fc

£200
, |l I *J 1—U
AW 1 10 100 1000
Holding time at 600*C (hr )
120* Fig. 6 Effct of stress releiving
time on hardness and tough-
ness of weld metal.

Fig. 4 Effect of maximum reheating

temperature on microstructure.

• TMHK Lath Acicular

100
•i* •. . " V " " OA Acicular
A.W. fyv' •••••• • A Lath
&.. ••> H

7OCfC

AW 1 10 100 1000
Holding time at 600*C ( hr )
KXXfc
p. * I •' afr'P Fig. 7 Effct of stress relieving
time on the ratio of precipi-
tation of Nb and V in weld
Fig. 5 Effcts of maximum reheating metal.
temperature on microstructure
(electron micrograph, extraction
replica)
 474

It takes over 100 hours at 600°C for the fine Nb- and V-carbonitride
to grow to lose coherency with matrix. Though the toughness of the
acicular type weld metal recovers gradually with the hardness fall, the
toughness of lath type weld metal still remains low in spite of hardness
decrease.
To examine the reason for these, the analyses of Nb and V precipi-
tates were carried out, but no difference is observed between the two
microstructres as shown in Fig. 7.
After all it is uncertain why the difference in toughness change
during long time stress relief has occured between the two micro-
structure.

4. The Effect of Ti, Oxygen and Nitrogen on the Hardness of Reheated

Weld Metal

From the results previously described, the toughness of reheated

weld metal is almost the same as that of as-welded metal if the reheating
temperature is over 800°C, or the stress relieving time at 600°C is over
500 hours. But these conditions are impractical to reduce the toughness
deterioration of reheated weld metal.
In this chapter, the effect of Ti which also tends to form carbo-
nltride and seems co contribute to acicular ferrite formation was
examined and the effects of nitrogen and oxygen were also investigated
in the presence of Ti.

Materials and Experimental Procedure

A variety of single pass submerged arc weld metals were obtained by

changing the combination of fluxes and wires (Table 7) to vary the
content of Ti, oxygen and
Table 7 Chemical compositions of nitrogen. Mo content in all weld
base metal and wires used. metals were kept nearly constant
c Si Mn P S Ni Cu Mo Al Ti Nb V (about 0.15%).
Base Metal 005 0.26 1.59 0019 Q002 0.21 0.21 -
Charpy impact tests, hardness
0028 0.017 0051 0036
measurements and microstructure
A 0.09 0.30 V53 0.010 0006 — _ 0.53 — - - —
observations of weld metals were
B 0 08 035 1.50 0.009 0.006 0.80 - 0.45 - 0002 - -
Wire performed in both as-welded and
C 0.07 0.25 199 0.016 0005 - - 0.50 0.007 0.180 - -
stress relieved (600°C x 45 min)
0 007 0.24 1.10 Q007 Q0O6 0.65 - 0.42 0038 0.200 - -
conditions.

Results and Discussion

Fig. 8 shows the relation between AHv(=Hv stress relieved - Hv as

welded) and ATrs (= Trs stress relieved - Trs as welded). Generally
speaking, ATrs increases with increasing AHv, and this suggests that it
is important to suppress the increase in hardness for preventing stress
relieved weld metal from toughness deterioration.
The effects of Ti, nitrogen and oxygen content on AHv are presented
in Fig. 9, which can be classified into four groups according to Ti and
nitorogen contents as follows.
 475

Group Ti content nitrogen content

(a) low low
(b) low medium
(c) medium low
(d) high high

AHv becomes Jarge in the order of (a) to

(d) under the condition of the same
oxygen content, namely AHv rises with
increase in Ti and nitrogen content.
10 20 30 40 One of the features seen in Fig. 9
AHV(=HV.SR-HVAW) is that AHv presents high value in case
Fig. 8 Relation between AHv and of low oxygen weld metals regardless of
ATrs Ti and nitrogen content.
From the above results, it can be
said that to keep AHv small Ti and
Ti ('/.) ; N (ppm) nitrogen content should be reduced
0012-0.020 50-68
0012- Q018 77-84
without lowering oxygen content too much.
0030-0033 ' 50-56 Ti content required to improve
50
0033-0042 j 100-110 weld metal toughness is about 0.020%
when oxygen content is 350 ppm or so.
30 °\ "•*-—«
" - - * . . _ Excess Ti has detrimental effect
on the stress relieved weld metal
10 toughness as shown in Fig. 10, because
200 300 400 500 600 700
Ti as well as Nb and V precipitates as
Oxygen content in weld metal (ppm) fine carbonitride.
As mentioned above, AHv of low
Fig. 9 Effect of T i , nitrogen
oxygen (~250ppm) weld metal presents
and oxygen content on very large value and this phenomenon has
AHv little relation with Ti and nitrogen
content.
10
0
• SR
Fig. 11 shows an example of the
electron microstructure of weld metals.
When oxygen content is low, few
-10
304 - 380 spherical oxide are found, but large
-20 o 5 0 - 55 squarish precipitates which seems to be
-30 A1N or Ti(C,N) are observed. When
\
-40
-50 V ^~°-o °
AW
oxygen content is high, only spherical
oxides, which are the complex oxides of
Mn, Si, Al and Ti, are observed.
Q01 0.02 0.03 0.04
These facts suggest that Ti or Al
Ti content in weld metal {*/.)
tend to precipitate as Ti(C,N) or A1H in
F i g . 10 Effect of Ti content in low oxygen weld metal.
weld metal on transition The observation by large
temperature magnification reveals the fact that fine
particles precipitate around the large
squarish precipitates after stress
relief as shown in Fig. 12.
The qualitative analysis given in Fig. 13 indicates that large
precipitate consist of Al and Ti, and fine precipitate is Nb compounds.
(Cu peak is caused by Cu mesh used for the preparation of samples.)
 476

As Wtlded SR(60CTCx45min)
TirOOW.'/.
41:0030 V.
N.Mppm
O.mtfn,

Ti.Q016
At .0026 Squarish Coarse Precipitate
N:100 ,- : , , " / * -
O:«5 ' ; - ^

Ti:00U
Ai O02S
N: 77 • - ^
0:632
•. " . •'
Fig. 11 Effct of oxygen content on
raicrostructure (electron micro-
Fine Precipitate
graph, extraction replica)
Fig. 13 Results of qualitative
analysis of precipitates

As W«ld«d SR (60CfC x

'"••• S

Fig. 12 Electron micrographs of low oxygen weld metal (oxygen : 237ppm)

North et al 6 ) has reported that Al in low oxygen weld metal

containing Nb promotes the precipitation of Nb-carbonitride and
deteriorate the toughness. The similar results are obtained in our
experiments, namely large Ti(C,N) or A1N are produced in low oxygen weld
metal, and these precipitates accelerate the precipitation of Nb,
consequently AHv presents large value.
In conclusion it is important to control Ti, nitrogen and oxygen
content in weld metal in order to reduce AHv, and Ti should be added by
the lowest amount necessary for the improvement of as-welded metal
toughness.
 477

5. Multipass Submerged Arc Welding of Nb and V Containing Controlled

Rolled Steel

When Nb and V containing controlled rolled thick plate are applied

to shipbuildings or marin constructions, multipass submerged arc we:lding
process would be adopted.
In the study on simulated HAZ (weld Heat Affected zone) of Nb
bearing steel by Shiga et a l 7 ) , it has been revealed that the toughness
deterioraion of the steels due to double thermal cycle is negligible if
Nb is less than 0.06%, because Nb expands fine grained region of HAZ.
In this chapter, the hardness and the toughness changes of multipass
weld metals containing Nb and V were investigated.

Materials and Experimental Procedure

Multipass submerged arc welding were performed with 1 pass 1 layer

welding method in the groove shown in Fig. 14 by using the steels and
wire listed in Table 8 and SiO2-CaO-Al2O3type flux.
Not only Nb,V steel but also Nb, V free steel was used for
comparison and heat input in each passes was 45Kj/cm and oxygen content
in weld metal was about 350 ppm.

Table 8 Chemical compositions of

base metals and wire used
in multipass welding.
C Si Mn p S Cu Ni Mo Nb V Ti Al
Nb, V
s!«tl 0.08 0.25 ! . J 9 0024 0003 0.19 0.16 - 0035 3052 - 0.028
Nb.V Ire* 0.15 0 3 6 1.51 - - 00*2
steel 0019 0.012 0 0 2 - - -

Nb.V steel Nb,V free steel Wire 0.07 0 3 6 1.72 0017 0003 - - 0.29 - - 0.1K -

Fig. 14 Groove configuration of base

plate used in multipass
welding.

Results and Discussion

In multipass welding, the chemical composition vary from pass to

pass because of changes in dilution with base metal. Consequently the
hardness of each weld pass reheated by subsequent weld heat cycle or
stress relief treatment also changes.
Figs. 15 and 16 show the changes in contents of microalloying
elements in each passes and the hardness distribution of weld metal
measured in the direction of plate thickness respectively.
The changes of the maximum hardness value in each passes correspond
well to the change of Nb and V content as shown in Fig. 17, which was
drawn based on the results of Figs. 15 and 16. (Nb+1/2V) in abscissa
means that the contribution of V to hardness is half that of Nb as
indicated by Shiga et a l 8 ) .
The maximum hardness value in the hardened zone decreases as the
number of passes increases because the contents of Nb and V decrease.
 478

0.05 270 o As Welded

260 Nb, V steel • Stress Relieved
<600*C « 90 min)
250
240
230 • » m

220
210
200
190
-i U— 1mm Aff XJ ng F ;
pass 4 th 1st 1
180 pass pass pass pass
170
3rd 4th Final
Pass Pass Pass
Fig. 15 Change in contents of Fig. 16 Hardness distribution in
microalloying elements in each multipass weld metal containing Nb
passes of multipass weld metal. and V.

250 Nb, V tree st*»l

o As Welded
• Stress Relieved
(6OO'C« 90mm)
240
i
13
230
/ ^ (1st pass)
/ / ( 2 n d pass)

'•(3rd pass)

220 A
(tlnal ,
•th pass)

pass) iidr —Jk— Backing Side

o As welded Fina 2nd-+-lst-J
210 pass pass pass '
• Stress relieved
Fig. 18 Hardness distribution in
200 multipass Nb, V free weld metal.
001 002 0.03 Q04 0.05

Fig. 17 Effect of Nb and V content

In
in each passes on maximum hard- case of Nb, V free steel, the
ness observed in the reheated increase in hardness by the preci-
zone. pitation of Ti-carbonitride is taken
place a little as indicated in Fig. 18,
but the amount of extent is fairly small compared with Nb and V
containing weld metal.
Fig. 19 shows the macrostructure of multipass weld metal containing
Nb and V, and it can be classified into four zone according to the
maximum reheating temperature of the subsequent weld thermal cycle as
indicated in the figure.
The microstructure of each zone is almost the sarae as acicular type
microstructure presented in Fig. 4 in chapter 3.
Charpy impact test specimen taken from multipass weld metal includes
the above mentioned zones, that is to say specimen consists of softened
zone (B zone) and hardness zone (C zone), and two other zones (A and D
zones).
 479

25
O M u l t i p a s s weld m e t a l (AW)

• Multipass weld m e t a l (SB)

i 20 A Final pass weld metal (AW)

Nb,V steel
n 15
o

10
k_
Or
C
UJ

A : HAZ, Coarse grained zone ( £ 1200*0

B: HAZ, Fine grained zone (800*C-1100BC)
-60 -40 -20 0 20
C . Hardened zone ( 500'C - 700°C )
Temperature (°C )
D Unaffected zone <
Fig. 20 Toughness of multipass
Fig. 19 Macrostructure of multipass weld metal containing
weld metal containing Nb Nb and V.
and V.

Fig. 20 shows the toughness of multipass weld metal of which charpy

specimens were machined from the center of the plate thickness containing
2nd, 3rd and 4th passes, and the toughness of final pass weld metal of
which specimens were taken from the location 2 mm under plate surface and
contained only final pass.
Though both the hardened and softened zone are included, the
toughness of multipass weld metal in as-welded condition is nearly the
same as that of final pass weld metal. This result suggests that the
toughness deterioration in hardened zone is just counteracted by the
toughness improvement in softened zone.
But A and D zone? also turn into hardened zone by precipitation of
Nb and V after stress relieving treatment, therefore the deterioration in
toughness of stress relieved weld metal is very large.
In case of Nb and V free steel, no change in the toughness of
multipass weld metal was observed before and after stress relief though
the results are not presented in the paper.
As are shown in Figs. 15 and 16, Nb and V content in the first pass
is much higher than those in the other passes, and consequently the
maximum hardness value in reheated weld metal is also larger than that in
others.
In order to suppress the deterioration in toughness of multipass
weld metal, it is important to reduce Nb and V content in the first or
the second passes by decreasing the dilution with the base metal. For
that reason, the welding technique or the welding conditions which melt
less base metal are required.
 480

In conclusion, good toughness would be obtained in multipass weld

metal by lowering the dilution with base metal and taking the advantage
of grain refinement with subsequent passes.

6. Conclusions

The effects of Nb on microstructure of single pass weld metal, and

the toughness of single pass and multipass submerged arc weld metal
containing Nb and V were investigated under various conditions.
The following conclusions were obtained
(1) Nb suppresses the formation of proeutectoid ferrite, but has no
effect on acicular ferrite formation. The role of Ti is very
important in the formation of acicular ferrite structure.
(2) The toughness of Nb and V containing weld metal reheated by
subsequent thermal cycle depends on the maximum reheating temperature.
Nb and V, which exist as solid solution in as-welded condition,
precipitate as very fine carbonitride in the weld metal reheated
around 500°C 700°C, and deteriorate toughness but the toughness of
weld metal reheated over 800°C remains almost unchanged.
(3) It takes over 100 hours in stress relief at 600°C that fine Nb- and
V-carbonitride become large and lose the coherency with matrix.
(4) Though Ti addition is very effective in improving weld metal
toughness, excess Ti decreases toughness of stress relived weld metal
by precipitating as fine carbonitride, therefore Ti should be added
within the lowest limit required to improve as-welded metal toughness.
Nitrogen and oxygen should also be controlled in connection with
Ti in order to reduce the toughness deteroration of stress relieved
weld metal.
(5) In multipass welding, Nb and V content vary from pass to pass
because of changes in dilution with base metal, and lowers with the
number of passes. The most hardened zone in the weld metal reheated
by subsequent weld thermal cycle is observed in the first pass where
Nb and V content are the highest.
(6) The multipass weld metal containing hardened zone and softened zone
presents the same toughness as final pass weld metal toughness in
as-welded condition, but after stress relief the toughness of
multipass weld metal is decreased by precipitation hardening.
(7) In order to suppress the deterioration in toughness of multipass
weld metal, it is important to reduce the content of Nb and V diluted
into weld metal by decreasing the melting ratio of base metal and to
utilize the advantage of grain refinement with subsequent pasres.

7. Reference

1) Garland J.G. and Kirkwood P.R. "Towards improved submerged arc welds"
Metal Constr. 1975, ]_. (5) p.275, (6) P320

2) Levine E and Hill D.C. "Toughness in HSLA weldments" Metal Constr.

1977 9 (8) p. 346
 481

3) Yoshino Y and Stout R.D. "Effects of microalloys on the notch

toughness of line pipe seam welds" Weld J. 1979 58/3) p.59S

4) Hannerz N, Valland G and Easterling K. "Influence of Nb on the

microstructure and mechanical properties of submerged arc weld metals
for C-Mn steels" IIW Doc. Nos 1-612-72, IX-798-72

5) Wegzyn J and Dziubinski J. "The effect of minor additions of vanadium

and niobium in the submerged arc weld metal on its mechanical
properties and resistance to brittle fracture" IIW Doc XIIA-145-77,
Dec. 1976

6) T.H.North, H.B.Bell " Deoxidation during submerged-arc welding" Trends

in Steels and Consumables for Welding, paper 21 (1978) Welding
Institute.

7) C.Shiga, N.Tabata et al "Effect of Nb and V on Mechanical Properties

of Controlled-Rolled Plates and Their Welded Portions for X-65 X-80
Line Pipe" Kawasaki Steel Technical Report 10/1978) No. 1, P 1 (in
Japanese)

8) A Shiga, H.Imura, J.Tsuboi "Effect of Niobium and Vanadium on

Toughness of Submerged-Arc Weld Metal" IIW Doc. IX-1049-77
 CREEP CHARACTERISTICS OF BUTT WELDED JOINTS FOR
THICK 2 1/4 Cr - 1 Mo PIPEWORK AND PRESSURE VESSELS*

G. Cumi.no and F. Rinaldi

Research Laboratory - DALMINE S.p.A.

Italy

ABSTRACT

Research programme aiming to verify the possibility for asses

sing the creep strength of heavy wall thickness welded joi'its by
separately analysing different zones of weldments using new type
of specimens with rectangular section. Experiments carried out
on 80 mm thick 2 l/4Cr-lMo steel plates and 50 mm thick AISI 316
steel plates. Creep testing with strain measurements at 550°C,
in a range of stresses from 133 to 53 N/mm , for times up to 27,000
hours- Statistical treatment of results on times-to-rupture, %
e1 jngation and particularly on times—to-a-given-elongation (0.5
and 1 % ) . Comparison of the resulting behaviour with simulating
heat treatments. Example of application of experimentally measu
red creep parameters for the different zones considered in a sui-
table mathematical model to predict the inelastic behaviour of
welded joints.

*This work was developed with the financial support of the European
Community of Steel and Coal (ECSC) and involving eleven laboratories in
Belgium, France, Italy, Germany, Holland and one official quality con-
trol Agency

483
 484

INTRODUCTION

During the sixties a trend, later confirmed, towards the construc-

tion of increasingly larger power plants occurred; therefore the thickness
of pipework and pressure vessels had to be increased well beyond the tradi_
tional limits. At the time, there was some concern about consequences
of this change of scale, when it was found that transverse cracks had
occurred in weldments of high temperature spherical pressure vessels
For this purpose an extensive research programme was developed to
find whether it was possible to assess the strength of a welded joint
using uniaxial tests and separately analysing the creep characteristics
of different zones of the joint, taking into account the large scatter
that characterizes experimental procecures in the elevated temperatures
field. The main part of the programme was devoted to welded joints in
ferritic steel 2 l/4Cr - IMo, due to its wide u^e in power plants.
As a part of this programme, the creep behaviour of welded joints
in AISI 316 steel, was also investigated. All the results of the total
programme are collected in the final ECSC report
The results of the research on AISI 316 steel have been published
elsewhere by Etienne and others ' .
In this paper the main results of the investigation on the creep
behaviour of welded joints in ferritic steel are described in detail;
moreover, some informations about the research on AISI 316 welded joints
are given, in order to widen the scenario considered and draw a common
methodological inference out of all the results.

MATERIALS AND TEST PROCEDURES

Material specification and welding procedures

. .Ferr_^t_ic_stee_l_ - For the investigation a Cr-Mo steel according to th-.

Euronorm 43-72 classification (12Cr Mo 9 10) was chosen. The material
was manufactured in one of the ECSC Countries, as rolled plate with a
thickness of 80 mm, a width of 1500 mm, and a length of 8400 mm.
The heat treatment of the plate was austenitizing at 940°C-40 min/air
to r.t. followed by tempering at 740°C-30 min/air.
The plate was cut into several pieces (length 2000 mm and width
450 m m ) , which were welded together. A conventional double-slope
V-groove was chosen to get results representative of fabricated parts for
pipework drums or vessels. The welds were made by automatic submerged
 485

arc. The welding procedure is givnn in table I and fig.l shows a cross
section of the joint made.

Table !. Wi-iding procedure adopted for' Ihe ferrilic steel

- Prehea'ing : .'00 + ,<00"C

- Welding : a) f i r s t p a s s : TIG p r o c e d u r e - argon - .' mm i> w i r e ,

I1. V - HO-s-l/'O A
b) second and t h i r d p a s s e s : manual a r c - :3.?S and
4 mm li c o a t e d e l e c t r o d e s , type AWS L..5 E9016-B3,
d t i e d at ?SO°C for two hours - J?+?A\ - 90fl40A

• ) f i l l i n g ( f o u r t h to n i n e t i e t h p a s s ) : automatic
submerged a r r p r o c e d u r e , A mm <t> wire and fused
flux d r i e d at I00°f for two h o u r s , type AWS b.PA -
K80 EB<-B.1, /8+iOV, 400+')00A - Rate of welding
r'.GG to 0.4 ^ m/mi n

- Ir:terpa:J« IrmpiTS'un 1 : i.:'O * 1'/0"C

- PWHT : h e a t i n g t o 4')0"t.', c o o l i n g t o 300"C. P u t t i n g i n t o a
furnace a t :*00nC - I n c r e a s i n g by 70°C p e r hour
up t o 'VO" j_ 10°i'. M a i n t a i n i n g for 3.fj h o u r s -
b ur-nare e(joi i ng

Fig.l
Macrographic
section of the
joint made in
ferritic steel

x 1
 486

From one of the pieces (not welded) of the original plate, four 20 mm
thick blank plates were cut. These samples were submitted to appropriate
heat treatments, chosen to obtain specimens entirely constituted by homoge
neous microstructures, characteristic of four different sub-zones of the
heat affected zone of a welded joint. The thermal cycles adopted are re
ported in table II. After these heat treatments, the four plates under
went the same PWHT
Table 11. Thermal cycles applied on blank plates 20 mm applied to the actual
thick aid resulting "simulated" structures
joint after welding
Time (sec) (tab.I). The struc-
5' at Tnax
Cool ing between Structure obtained
<°C)
800 and 500 °C
tures obtained repr£
sent a coarse-grained
none (overheated
850 oil 15 ferrite + 6 0 % bainite
layer) and a partial
martensite + 20+30 lower ly re-austenitized
1300 oil 15
baini te zone for two diff£
rent welding prucedti
i'errite + carbirixed
850 ai r 300
zones » 8 0 % baini t.e res (manual arc welci
ing and vertical elec
upper bainitr with coarse
1J00 air 300 troslag welding),
grains
which represent the
*) characterised by the maximum temperature rear hed and by a
ends of the range of
standardised cooling r a t e .

procedures practically utilized for 2 l/4Cr - IMo steel.

. A u £ t £ n i t i c _ s i e £ J _ ~ T n e details of the plate manufacture were already

described by C.Etienne et a l . 1 4 . The choice of a U-groove derived from
common use in fabrication shops. The welds were made by a manual arc pro
cedure with coated electrodes (see tab.III). Fig.2 shows a cross section
of the weld.

Table III. Welding procedure adopted for the austenitic steel

- No preheating : interpass controlled at. ;'00°C or less

- Welding : - manual arc

- electrode type E316 - AWS A54 - 1978
- first pass : 3,25 mm 0 coated electrode
24V - 110A (completely ground before the back
welding)
- second to ninetieth pass : 4 mm 0 coated elec-
trode - 25V - 150A
- five backing passes : 4 mm 0 coated electrode
25V - 150A

- no PWHT
- ferrite % admitted : 7,5 * 10
 487

Chemical composition and mechanical properties

An independent quality control Agency

(ASS.VINCOTTE - Belgium) was entrusted to carry
out all the QC business involved. The chemical
composition and mechanical properties of the
base metal and weld metal are given in tables
IV + VI for both steels.

Non-destructive examination

Fig.2 The welded plates, both on austenitic and

Macrographic section ferritic steels, were subjected to accurate
of the joint made in non-destructive testing by the aforementioned
austenitic steel quality control Agency, using X-ray, ultrasonic
x 1 and penetrant techniques.
Places affected by NDE indications were
avoided when subsequently preparing the creep test specimens. Aftor
machining, the test bars were inspected again to assure freedom from weJd
defects.

Type and position of the creep test specimens

In order to evaluate the creep behaviour of a weldment in heavy wall

thickness manufacts, it appeared necessary to separately investigate on
samples representing the different zones of the joint. For this purpose,
three different types of test specimens were chosen :

a) test specimens cut across the weld bead in the traditional way, but
monitored to evaluate the elongation in great detail, zone by zone;

Table IV. Chemical composition (wt %) of the parent metal (PH) and weld metal (WM)
(ferritic and austenitic steels)
Zone

;
N
1) C Si Mn P S Cr Mo Ni Al Cu Ti B No
I0 ppm
J
-1 PM 0.14 0.?3 0.56 0.012 0.007 2.26 0.99 - 0.018 0.11 <0.01 - 0.005 130
•

ij
WM 0.06 0.55 0.70 0.013 0.009 2.76 1.03 - <0.003 0.29 <0.01 - 0.009 120

. j
PM 0.046 0.41 1.87 0.029 0.021 16.95 2.22 13.2 0.003 0.24 <0.01 <0.0001 <0.002 340
Auste

WM 0.052 0.28 1.66 0.022 0.013 18.25 2.16 11.45 0.001 0.05 <0.01 0.0003 <0.002 375
 488

Table V. Mechanical properties for the parent metal and weld metal (ferritic and
austenitic steels)

Position of Rolling Temperature Rp 0.2 Rm A Z

Steel
test bars direction °C N/mm2 N/mm2 * %

20 316 515 30 75
As delivered Parent metal transverse
500 227 367 28 77

20 333 533 30 77
300 299 459 21 70
parallel
400 280 437 21 68
After welding
500 262 372 26 77
Ferritic and stress
relieving Parent metal
20 314 531 32 76
300 276 443 21 71
transverse
400 267 435 21 69
500 243 363 27 77

Weld metal - 20 463 599 26 68

20 225 530 49 55
As delivered Parent metal transverse
650 101 361 37 43

20 226 527 60 73
500 119 402 43 70
Parent metal parallel
AusLenitic 600 99 339 44 70
700 96 247 44 48
After welding
Weld metal - 20 478 593 38 51

Across the weld - 20 243 520* 34 54

Fracture in parent metal

Table VI. Charpy V and DVM impact tests (J) at room temperature
(ferrit ic and austenitic steels)

Rolling Heat affected

Steel Parent metal Weld metal
direction zone
DVM - -
As delivered transverse
153 * 171 - -
Ferritic
KV KV KV
After welding transverse
and stress 182 * 216 195 * 215 92 * 148
relieving
longitudinal 214 H- 248 175 + 196 107 + 154
KV - -
As delivered transverse
95 - -
KV
Austenitic transverse - -
56
After welding
KV
transverse - -
67
N.B.: Average values of three specimens tested in different plate pieces
 489

b) test specimens cut parallel to the weld bead not only in the weld
metal itself but also in the heat affected zone, the adjacent tempered
:ione and the unaffected parent metal;
c) test specimens cut from the blank plates, subjected to thermal cycles
given in table I[.

For the type a) tests, circular section test specimens were used
(fig.3); the elongation in the various zones was determined by measuring
the distance among Vickers indentation of diagonal 1.^ mm, spaced at re-

I i

Kig.o - Circular section test specimen taken across the weld bead

gui ar intervals along the test specimen and used as gauge marks.
The measurements were carried out by optical means during programmed
interruptions in the creep test.
For the type b ) , test specimens with circular (fig.4) and rectangular
section (fig.5) were cut.

Fig.4 - Circular section test specimen with shoulders for the

application of extensometers

Fig.5 - Longitudinal test specimen with rectangular section used

to investigate the non-fused zones of the joints
 490

The first type of specimen was used to investigate the creep beha
viour of the parent metal (PM) and the weld metal (WM).
The specimen shown in fig.5 (whose rectangular cross section is
5 x 12 mm) was adopted to represent the non-fused zones. As the heat
affected zone (HAZ) width is 2.5 to 3.5 mm, the specimen width will
therefore include the whole HAZ plus a small strip of the next zone
(see fig.7). Particular care was taken just to exclude the weld metal,
adopting a sequential mill and etch procedure (see ^ — annex E ) .
Further consideration was given also to the parent metal portion in-
volved in welding thermal cycles with peak temperatures lower than A-^
(no y -phase formed during welding). To test this part of the joint,
specimens conventionally called tempered zone (TZ) were cut, baking care
just to exclude the HAZ (see fig.7).
The specimens conventionally called HAZ and TZ include, if taken to
gether, all microstructures different from the PM which in the as welded
condition are characterized by very rapidly changing hardnesses.
The* c]ongation of these test specimens was measured during the tests
by means of extensometcrs applied to the shoulders.
For the type c ) , test specimens with circular section, shown in
fig.6, were used. The elongation was measured by the traditional exten
sometric method.

Fig.6 - Test specimen used for creep tests on microstructures

obtained by thermal cycles of table II

In order to make the results as comparable as possible in these

three methods, the length between gauge marks was in all cases taken as
five times the diameter (actual or equivalent) and the same cross-section
(60 mm ) was used for round and rectangular test specimens. The creep
tests were performed according to Euronorm 123 , appropriately modified
to duly control the temperature and the elongation.
Fig.7 and 8 show the sampling sites in the welded joints for fer-
ritic and austenitic steels.
 491

PM

Fig.7
Sampling sites in
the ferritic steel
joint
PM = parent metal
HAZ = heat affected
zone
TZ - tempered zone
WM = weld metal

400

Fig.8
Sampling sites in
the austenitic steel
joint
 492

RESULTS OF CREEP TESTS

Ferritic Steel

The programme involved creep tests of long duration at S'.OT and the
stresses ranged from 133 to 53 N/mm'1 • The test conditions were selected
on the basis of preliminary results obtained on the plate used as parent
metal; some specimens were employed to analyse the creep behaviour up to
1% elongation. The remnant tests had to be carried out to rupture,but
several specimens of this group did not fracture despite the length of
the test period (up to ^7,000 hours). For this reason it was not. pos-
sible to fully exploit data according to the established experimental
design. Nevertheless statistical analyses were conducted on rupture da
ta (time and % elongation), but on these results several limitations arc
present on account of the high stress levels involved.
By analysing the numerous results for the specimens parallel to tin-
bead at 133 N/mm , a significant interaction was detected between t.hi•
non-welded zones of the joint and the position in the wai ! thickness from
which the test specimens were taken. This interaction gives an exper_i_
mental evidence of the different behaviour of the material at the outer
surface and that in the middle of the joint in the HAZ (see' -^-annexe N ) .
This difference, whose relevance is considerable at the higher stresses,
seems to decrease at larger lifetimes; this effect may be related to the
evolution of the micros true ture due to a tempering effect at the test.
temperature.
Tentative exploitation, based on minimum creep rate, in order to
obtain the times to rupture by extrapolation of interrupted creep tests
values, didn't give gJod results due t.o the relevant scatter of creep d_a
ta (see - annex H ) .
It appeared advisable to consider al .1 the results obtained without
dropping any long term indication; therefore as main variable, the tj_
me-l.o-O.o or to-1% elongaiion was chosen.
However, it was decided to drop data obtained at stresses higher
than 9S N/miTi'" • This choice was made to avoid confusion among d i f 1'IT-MI t
creep mechanisms and to uti1i ze in all cases the very lonp duration data
of unbroken specimens.
A simple •; tatist .leal method was appl ied 1^ to discriminate the beha
viour of the different zones of the joint. For this purpose the curves
t - ac " were separately considered for every group indicated in tables
VII *• IX. Practically, straight. lines log t vs. log a were calculated,
each characterised by its slope and a V—intercept with a constant
 493

stress line *' • Confidence limits at. 9l,Z gave the resolution power of
the comparison (see °- annex N ) .

Tabli- V I I . Avuilnb;'- c : p d s i l J I w . y • \ m i - - . - ' i— i ) . • %

Tr: 1 S p ' ^ i ,.-n:; p;i) • ; i l i . - l • ! • . : ; • : f>- • • . n\> •;:

POL- i :.i tin
to ' h r w.-ld b. li>- w l d b- ad
Of ' I-.1!.
spi-.-im..nB
I'M IIM. r~ I'M H/i/. WM

Ollt.T 1 8 6 a 8 8

Middl" of 8 8 8 8

the wall
'..hi rkn-n: 1 8 H H 8
1

T , - i b i •• V I I I - /-v:ii 1,-ibl'' .!-•••[> J . T . I : -• • . ; n - -.:-• -'%

Tt-b< spt*r i mrns para)!*1] p r , :m<-r^ ai-f o 3 s

Por ;. t i on
t.u Lh'- weld b«-ad -he wri d b e a d
Of Lest
jspt'C i mentj 1
PH HA;: TZ WM ( . liAZ | WM

Outer .• - 0 8 8 8 /

Midrili: o f 6 9 8 8 8
the wail
(Jiickness 8 6 - 8 8 8

Table IX. A v a i l a b l e c r c p r u p t u p f ri;iV.n a1. s'..r'*ssf'M

b,' + 1 '. •> N/ mm'

Test specimons parallel Test sperirni-'ns across

Position
to the wel d bead the weld head
of test
specimens
PH HAZ TZ WM PM HAZ

Ojter r} 8 8 11 <' 8
-

Middle of 6 O
11 Jl 11 - -
the wall
thickness 4 11 3 8
- -

*) Selected In order to avoid confusing extrapolations beyond

the stress range actually analised.

I
 494

The available creep data were collected according to the factors

adopted in the programme. Furthermore a thorough check for bias showed
the experimental design had been able to give unbiased data: only four
data clearly appeared to be biased by metallurgical factors (defective
zones present in only two pieces whose specimens had been analysed in
four different laboratories). These data were rejected.
A comparison with ISO DATA 8 showed uhe creep behaviour of the
thick plate utilized in the research was similar to that of lower thick
ness materials, cooled in furnace, after austenitisation, at a rale of
30°C per hour or less. In each of tho following figures the regression
line of these data (times-to-1% elongation) is drawn as a reference.
After these preliminary evaluations, the available creep data, c-
ported in tables VI1-VIII-IX were exploited by the statistical method
already referred. The results of this analysis for times- to-0.!J% eJon
gation, for the different zones of the joint, are reported,as an example,
in tab.X.

T a b l e X. S^ope <b) and Y-i n t . e r c p t ( a ) w i t h t h e 70 N/mnT l i n e of r e g r e s s i o n lines

log t v s . logo" f o r t i m e s - t o - 0 . 'J% e l o n g a t i o n obtained a', s t r e s s e s of S3+9P N.mnr
f-.r t h e di !".''•:-<••*.• :'^r.'>3 o f t h e weldment considered
1

Spt"' i in* ns para 1 " ! to the . e:d bead Sp.- ~i mens across the w - I d bead
!
Co:it"i dence a Confidonre a Confidence
/.dii'- • -o b Zone b
1 imi I s 9'>% limits 9'« ] i mi t s 9r>% (hi 1 imi ' s 9\.%

-;•-. VG 9,1/
UAZ -A. ' 6 1 -,0-1
-f.. .'6 /r J9".' - i .66 768
-. 48 -'",.08 106/
-6.-19 118,'
- / • . S6 :
I'.go [
- ! , .61. < 1 i1 PM
-6.3-1 .V8'.6

-1.8". .590 -,'.-'B 1 49 -.

- d 8.- 99o -'A -A. 6A
-7.81 -6.r. 1

200

200 500 K>00 50O0 100O0 30000

timt( hours)
Fig.9 - Parent metal results at bbO°C:time-to-rupture and
time-to-1% elongation (dot and dash line)
 495

The single data for each zone of the joint are shown in figg.

200. 1 1
pirs"I
tt(
r, 1)

•-
A r •-. m
»~«
>*
a k. A .
•g
••V
ft PJ»-.
—
"5Ca ifi
stress (

CM
•«•

4 t 1 •
5

/
200 500 1OOO sobo io ) 0 0 300(
timt(hours)

Fig.10 - Heat affected zone r e s u l t s at lV>O°C. Ti m:'-Lo-rupture (tR)

and time-to-0.b% elongation ( t o . . j ) - Creep t e s t specimens
p a r a l l e l to the weld bead

'/ft e |
a a A £ k
Li M M

ft
1
M M

/. -
- — .
»—.
1"" m a a mLa
a 1
-M 1« •
z
S
F I S LA 11
•M
- « =

-»
s

/«* T ^.
200 5C 0 1(OOO 5O'OO KX >00 300*
tim« (hours)
r i g . 1 1 - Tempered zone r e s u l t s a t bbO°C. T i m e - t o - r u p t u r e ( t R )
and time-to-0. r 3% e l o n g a t i o n ( t o . f , ) - Creep t e s t specimens
p a r a l l e l to the weld bead

Legenda : A preliminary tests

• outer surface
0 middle of the wall thickness - parallel to rolling
direction
A middle of the wall thickness - across rolling
d i r e c t i on
—*• t e s t s interrupted
 496

200i 1 I I 1

/i i
>are nt rti(sta
)
•mm
• -

\ — «u^— tt ? •

"B
z
Hi t
—c
a I :
• i_
I
: * • > •Mi ft ft

—J
1
• i , 1.—- I
A
I

:r
i

t /j T
0) J.5

u
\

•g
200 500 1000
D | o r *
5000 10000
1 3000
time (hours)

1'ig.L' - We i d m v t a l r c i i u l ' : s at. (•bO"U. Timc.--Lo-rupt.uri. arid

tirni'-to-O.'-.% c LijngaLion. Ci-cep ti-st sp'-cimenr. pnraJli-l
r
.o t h o weid bead

By a n a l y s i n g t , h e p a r a m e t e r s o f t h e regression l i n e s f o r t h>- v a r 1 i oil

c a s e s o f t a b . V I 1 - V I I I - I X , i t a p p e a r e d t h a t n o d e t e c t a b l e d i l ' f e r e n • • • ' wa
p r e s e n t among t h e c o n v e n t i o n a l z o n e s o f t h e j o i n t c o n s i d e r e d , e x cep;. ; h
WM for w h i c h a d i f f e r e n c e i n t h e y - i n f e r c e p t ( n ' s l s p e c i m e n s t ak < • t i a c r o
the weld bead) and a d i f f e r e n c e both i n t h e Y - i n t e r c e p c andt h s l o p
( t e s t s p e c i m> TIB t a k e n p a r a l l e l t o t h ' - w e l d b e a d ) f o r t i m r - r i - t o - 1% nga
t i o n were found ff'Lg.K-1).
Kurfhe-rmore d u f - f i i i t y was s a t i s f a < ' t o r y in a l l zoivs exeepi ill the
we I d m e t a l ( f i g . l.-i) .
I n more- d<' t a i 1

a) P M , T Z a n C H A Z sliowed a r e g r e s s i o n line with the same; slope- ISO

rupture data for p i p e a n d s h e e t m a t e r i a l T h e sit i h t . (fab.Xi). ] in.
J ies n a t u r a l l y in a d i f f e r e n t p o s i t i o n , d u e t o the c o n s i d e r : t i . i l l o f

t or' t i n s t e a d o f the r u p t u r e t i m e
0.'.% 1%

T a b l . - X ! . S;>.p.- ( b )r m r t i - i n l - r - p ' I,-,, w - I , •.;•,-.•:

i ,;,. ; i : ' ) . -N / m m .,f r-.-fi": .-...:...:!•. , . ,. • h

i .•:;,,..),,,. -
N imbi i
" ; ; ; , ; • •

9
'"°

ISO -G. : 8
- -

-b.A-' ] o: 16
PM +HAZ+TZ 0'. -6. , Il^^'J
9
-/.A0 l.'9,.l
1
 497

icf 5 10° 5 10 5
time Chours)
i ' ' i g . l 3 — % Reduction of a r e a v s . t.i me — t o - r u p t u r c . T ' >' >0" • •
Lop : specimens p a r a l l ( ; l to the weld bead
bottom : specimens a c r o s s the wold bead

Nota Beno : Each point, r e p r e s e n t s the average v a l u e for the

r e d u c t i o n of a r e a (%) d e t e r m i n e d in test, specimens
which f r a c t u r e d within a time c l a s s , the amplitude
of which was s e l e c t e d as O.?66 ( l o g a r i t h m i c value).
The f i g u r e next to each p o i n t r e p r e s e n t s the number
of v a l u e s included in t h e time c i a s s in question.
 498

b) In comparison to the other zones, the WM showed a different behaviour.

This result was confirmed both by test specimens taken longitudinally
in the weld bead and by those taken across the weld bead. In the
last, elongations were separately measured in each zone (examples are
reported in fig.14).
As shown in fig. 12, the WM regression line log tp, v s . log CT was
characterized by a slope significantly different from those of the
other zones and from that representative of ISO data. This evidence
helps to explain the ruptures in the PM at high stresses (- 10^; N/mm1-)
and the ruptures in bhe WM at low stresses occurring for specimens
taken across the wold boad.
This fact permits to predict with high probability that, in case of
failure, fracture should occur in the we]d metal, unless improvements
are introduced in the welding procedure. Also for the creep ductility
a notable gap was found (fig.L3).
Furthermore the creep data (Li mc-t.o-0. b% or to-1% elongation) showed
that the different behaviour between the material, at Lhe outer surface
and the middle1 of the wall thickness is present also in the WM.
The material in the middle of the wall thickness showed a belter
behaviour; this fact could be explained by the effect of the welding
multipass procedure.

c) tn the experimental design, consideration was given t.o the detection

iJ
of minor effects, if present (for details see l Fur example the
scatter of data in the WM (timos-to-1% elongation) is more marked in
the specimens cut along the axis of the weld bead than in those "ut.
across. Besides, no difference was found between the groups of test
specimens taken parallel to or across the rolling direction of the
plate; whereas the thickness of the weldment led In differences in the
behaviour between the outer surface and the middle for the wall thick
ness .

Ferritic steel - Simulating heat treatment

This type of creep tests allowed to characterize the1 behaviour of the

PM with homogeneous and well definite microstrueture; the results obtainru,
although without practical applications, may nevertheless useful, as
borderline events, Cor wolding engineers.
As an example, fig.IS shows a remarkable reduction in ductility of the
overheated coarse-grained mirrostrucl.ures,
 499

tr--933 h
A--2SJV.
Z=90 '/„

0,1

10
B
0/
<7=53
/. 3 t-20066h
A»p.tey.
1

0,3

0,1

0,03

0,01
20 40

Fig.14 - Examples of creep elongation measurements for the

various zones of the weld at given times on specimens
taken across the weld bead :
A) at a high stress
B) at a low stress
 500

100
90
/o
80
TO
60-
5O O 850'C air
A 850'C oil
40-
• 1300*C oil
30- 1300*C air
o
20-
10-

10 10 5 10' 5 10* 5
tima (hours)
Fig.15 - % Reduction of area vs. time-to-rupture, at T=55O°C, for
the various microstructures tested.

Nota Bene : Each point represents one measurement.

Austenitic steel

For space reasons it's not possible to report here all the results.
For this purpose see .
In any case it seams interesting to remember the most important
results; notwithstanding the reduced number of creep test specimens,
some differences in behaviour of the different zones of the joint were
found.
At 600°C large differences in the times-to-rupture for a given
stress were measured: the time-to-rupture of the WM was about 1/3 of
that of the PM at the same temperature, and the HAZ showed a creep rup_
ture strength higher than that of the PM.
At 550°C the creep rupture strength of the weld metal seems to be
at least equal to that of the PM. But no definitive conclusion can
be drawn because of the limited number of creep tests carried out.
The differences in creep behaviour of the zones of the joint could
be explained by differences in microstrueture.
 501

F i n i t e I'lomt'iit a n a l y s i s u l " w o j d ' - d j o i n ' s

A f i n i t e - e ] e m < > n t a n a l y s i s , b a s e d o n a ran: h e m a ' i ' a ! m< . d - • t (<>r ' lv •

i n ^ L a y ' i ; - b e h a v i o u r l - ' r e e p , p Ia s; ic I ;y ) >•!' ' h e v a i ' . ' U r f /.•>!:. : i s ; d " ' l i e

j u i n t I ' a s b.-ej; : - ; n r ! i ' d o u t i ' o j '; h e f e . s !f;p.-:i».!.;; 'al-:-!i . T T ' i s s - h e v.-- • i d

b e a d . i-'ig. 1 6 s h e w s ' h e e l e m e n t tni-sh i i d o ^ r e d : i : ' h . ; i ; , - s i ;•/:• . s .

- 99 mm -
i

WELD. F ! g . I b
rtoiIETAL- - H A Z - -PARENT METAL
Me;,|i d . ' w f !•; b u • :• r. :
;
i n i h e s< ••• : o i . < !" a • !r

p i a t f ; .,)• ' h . - I ! : : ' '

< ] ' in • •! i • ' ir i a I y s : : ; .

50 mm

1
™6 mm

The r e s u l t s oV the.se cvnl u a I. i .cn:s have been •(;mpai--d with ',h<- e x p e -

r i m e n t a l r e s u l t s on auyt.c-niLi.i- s t e e l , .showing good a g r e r m e , - , •, (:;••'• an

!
•x a m p It: in Tig. 17). S i m i l a r l y the R M T - S S an
P. d s t r a i n d :.••;'. i- i b u ion

* EXPERIMENTAL DATA AFTER

0.6 260 HOURS OF CREEP j o n g i ' . u d i naj ' - r e e p s t r a i n
di.s l-ri b u t J on a l o n g ; ,he
wf^ld at. d i f f e r e n t i.i,,!••«,
a f t e r out.or s i d e o f t h e
03 spocrirtifii.
Stress 16'. N/mm
0,2
T - 600"('
0.1

0
_WELD
METAL
 502

across a welded joint was calculated for the case of a large thick pla
te. The details of this analysis are reported in the ECSC Final Report
(see - annex Q and ). Naturally it is not possible to apply these
evaluations to creep test specimens taken parallel to the weld bead.

GENERAL DISCUSSION

Ferritic steel (2 1/4% Cr - 1% Mo)

The experimental programme adopted and the simple statistical me-

thod (*) applied to exploit all low stress data for the time-to-C.b% or
1% elongation, have shown a resolution power sufficient to detect dif-
ferences existing in the behaviour of the different parts of heavy wall
thickness welded joints. In fact, some differences and interactions
have been pointed out in the preceding paragraphs; the presence of such
significant differences permits to assess, with a little risk, that, ail
non-fused zones of the joint, affected by welding thermal cycles and PWH1.
behave homogeneously in a relatively uniform way.
Numerical differences are not only not significant, bur. also smal1,
of the same order of magnitude of common metallurgical differences pre-
sent in heavy wail plates of the same steel.
Creep ductility is likewise satisfactory in all parts of the consi-
dered zones, Included the coarse—grained zone, normally prcs"nt near I.he
fusion line. This result is apparently i r. contrast with the behaviour
shown by artificially simulated coarse-grained micros trueture obtained
by five minutes austenitising at ].i00°C, oil quenching and tempering at
720°C (see fig.13).
In the simulated specimens at low stresses, creep rates and the
consequent ductility resulted lower than expected. However, specimens
cut from the actual joint didn't show such a behaviour. Taking into a^
count the non homogeneity of the structure present In the section, the
specimen with rectangular section, taken from the real joint, shows the
ability of the adjacent layers to absorb the presence of narrow layers
of a less favorable structure.
Metallographic examinations have shown that the simulation treat-
ment developed coarse grains, whose size was larger than that normally
found in submerged arc welded joints. Ilie good suitability of the HAZ
to follow the creep behaviour of the adjacent, zones is confirmed by the
results obtained by specimens taken across the weld bead.

(*) Differences were calculated by considering confidence limits at

9b% prob-ibil i tv
 503

The WM obtained by multipass procedure (approximatly 90 passes)

clearly behave in a less satisfactory way than the non-fused zones con
sidered. This result points out the need to select consumables and
procedure qualifications in a safe manner to avoid premature ruptures
at lovu creep ductility.
For the practical application to qualification and control prucc_ .
dures, it is interesting to remark that the specimens taken across the
weld bead, whose elongation was evaluated by Vickers indentations, gave
data with reduced scatter in comparison with data obtained from specimens
taken in the WM and in the middle of the wall thickness. This fact
could be explained by the tipical dendritic structure of the WM: in fact
the intercrystalline segregations are arranged across the axis of this
type of specimen. Conventional specimens taken completely from weld
deposits obtained by superimposing many layers at 90° eachother are ther£
fore to be preferred to specimens taken from actual weld beads.

Austenitic steel (17% Cr - 12%Ni - 2% Mo)

Larger differences has been reported for this steel, notwithstanding

the reduced number of specimens utilized. The HAZ shov/ed a higher creep
rupture strength than the PM, and the WM gave very poor r suits (see p.18).
Furthermore the behaviour shown at temperatures in the range from !>SOnC
to 650°C, appeared to be quite different particularly for the WM. This
situation led to apply a finite element analysis to calculate stress and
strain distributions across the welded joint in a round bar, with the help
of the experimental data obtained in the uniaxial creep strain measure-
ments for the present work.
The good agreement shown for the round confirmed that it should be
possible to predict the inelastic behaviour of an arbitrary structure
containing a weld. And this 'may have a remarkable practical importance.

Methodological aspects

The whole set of results obtained on both steels examined, gives the
possibility to draw some general considerations concerning the test me-
thods applied as an approach to the problem of the behaviour of welded
joints in heavy wall thickness pipe and vessels.
The present work has demonstrated that metallurgical effects of
welding on creep may be adequatly measured when tests are carried out with
sufficient accuracy (eg: suitable temperature control).
The three test methods used, produced complementary results, that
help very much to clarifv mechanisms bound to street and strain distri-
butions during service. The uniaxial creep strain measurements, carried
 504

out p a r a l l e l t o t h e w e l d b e a d , giv:> an e f f e c t i v e - b a s i s to develop sui-

table mathematical models f o r inelastic behaviour of welds, considered
as c o m p o s i t e s t r u c t u r e s . Accurate 1 measurements made i n t. he various", :'.->nes
of" s p e c i m e n s t a k e n a c r o s s t h e we L d bead g i v e {.he p o s s i b i I i t y to verify
the soundnrss o f the model developed.
The a p p l i c a t i o n o f s i m u J a t e d tliermaJ cycles gives :eaL piece;: wi',i,
a uniform structure throup'.h the s e c t i o n . This f e a t u r e p<>rmits 'o
by-pass d i f f i c u l t i e s arising from steep micros t r u e t u r e gradien's n i ' IIA/'.;
i n f a c t , c r e e p d a t a o b t a i n e d by s i m u l a t e d s p e c i m e n s show a r e d u c e d sea •.." I T
However, it would not be a d v i s a b l e t o recommend t o use t h e information
inferred by s i m u l a t i o n s to p r e d i c t the b e h a v i o u r o f welded j o i n t s . This
purpose is b e l t e r c o v e r e d by t h i procedure mentioned before.
Thi. c r e e p r e s u i t s o b t a i n e d on s i m u J a t e d s p e c i m e n s may have a g..ed
value i n p r e d i c t i n g what may happen if the w< J dec push'?, t h e operai.iv
conditions to the limit, leading i.o an anomalous mi e r o s t r u e t u r e in i he
t'M. Likewise t l i i ' s i m u l a t e d a p p r o a c h may have u s e f u l appi i r a i . i o n s .11 di
v e l o p i n g welding processes f o r new s t e e l s , because it- may p e r m i t , t o d< :•-.'
s a f e area;" a b l e to minimize the r i s k of failures.

i'ONi." I . U S I O N S

'he present invest igatioii carried out on i.wo types of steel gave • he
possibility 'o evaluate 'he creep behaviour" of welded join's in heavy wall
steampipes and vessels from several view—points.
in synthesis the to!lowing conclusions cmld be drawn :

— No s i g n i f i c a n t difference was d e t e c t , d among the various, mi c r o s ' . ru<" —

iural layers of the non-fused ;;ones affeefed by welding thermal '-ycii-:-
and I'WHT.
— The WM o b t a i n e d by a commonly used multipass submerged arc procedure
showed a less sal is f a c t o r y behaviour.
— The experimental programme carried out p e r m i t t< d to deter!., some minor
e f!er ts due to me I a I 1 u r g i e a l factors .

. A l l s t(T]J_I.J_e^s t ee l_

— l'he r • v i d e n c e o t sf 'fn- • di i feretires be tv/een I'M and HA/, w a s s h o w n , r K i w i • h

standing the reduced number of specimens tested.
— The rupture time o f the WM was about 1/e of tha: of the I'M at 600"!';
in other' w o r d s the c r e e p rupture s'reng'h of the WM appeared to be
O.Q x Lft for the P M . This situation seems to improve at ' '>0T.
 505

'J'h'- '.tii-i-i d 1 I T - - f n T a p p i > . a ' h 's a p p i ' . d g a \ ' . ••. . m p i r r n " . \ ' ;t:-y : • < • : ; ( ! ; - ; •. .

1
'ill' s p - ' i m * II:'. : ,- i k - •: i '. : i d i I T ir < i : ' .'.'.in.'S • >!' t . h . . j u l i r , p a r a - i ' i ' < > '> i -

v. 1 - i d h - a d , p i - r m ' ' " . ' <• n v a s u r i •:•••• p p , - i r ; " T i ' • ' • >'•'•• : • . > : ' ' ; i b ! > '•. s u p p i . : '

n ; i ; . i i* m a ' ; •• . - 1 ! ' a ! ' i n a ' I >- l i s i ".'id', njj; '-.•< p r ' d ; - ' ' ! ) • • b - h a v i " i : .-;' a •, r i

1 :
p i - - ' s ' r i i * L 1 J -• -'• i i i ' a : n "^' a v . ' " i i i .

:
'I • • : . ' : ' i.jj, .'".'. i n n ; a ' < d u;i :'• n ' n i m ' •. : ' ' i s ' i ' l ; ' ' i i : ' i •:; m a y lj. -i p ' :: p ; - . d ! ' ' ' \.\\ a / .

m a ! " i i s :; • : u a ' i . >n:", d u < <• > i > p i - r ' a ' '. \ ' < - • • • : i d i : " • •! i ; - ; p u s h ' d ' • ' l i - i n. ' ,

a a d :• : i d ' ' < ^ : ' . \ \ p : . . ; a l ' " a r ^ a s :i w < i d ' . a p , > >i i d ' ' '. •. i n : ' - ' < • t > a p p . - d !.

n i w :;: ' • i :• .

J
S. v i - a i p . r s . . ( : s j ; n v ' h < i r ' / a i u a b J • a f . dd > • • ( > ! . d i . a• s i h u• i ;. •• h

;
V A ' : ' k a iici ' h ' -• • . u ' I n - I • ; ; V.' s h '• i ' ' x p i ' ' •;:; ' h « u £ s a • ' 'J<i' ' ' • a t <.'. • t:• in.

A p a i " i ' l U a r ;t''ki'.<'W1. < < 1 g m " i r . i s . '•- I ) - -g i ' . ' . ' . ; ' \-- . H a ' •' m ! ' - i 1
h i s . S ' i : •• :•••

y.l'lti i !!' it*::- a s 'i :mir,i ' ' > • i ' l i a i r m a i : '.i ' s i : i h ! i s l , a i : t i u s ' . p . ••!, a t y • n i 's '

a n i . i : : g c i i T t ' ( ' : ' i •s s . v . i vvs. a n d ' I n 'K X [ J - i * ' : : ^'h; > p a s ' i ' . p a ' ' (1 i: ntai,,\' i : s , t ! n

d i s ' u s s ' : i:iis i : i pi-' p a r : ii g , ' a s s y i n ^ m i ! a ; , d '• H i ' - i a di a g ' h i W ' l - k :

M . ' ' ; u i b w ( ' H I , ., . . i . - i q u i: : ; - i U M i s s a ( f i ) , H . C s a ' m b a i h ('i i , K . I I. H a p , - d* ,]•:. •. i l ) ,

W . U - b r i ' ! D ! , .'.!•'. I v ' .•!<:.'• ( N . . ) .

T h i s v.'i't'k w a s li.'V'• 1 1i p r d K i : h M v I i ii.aii.'i a i s u p p . s i i . f ' ) i . E i n - : j " a : ,

v,\ I ' y i /! S i > •• i a n d •'<, ; ) j .

 506

REFERENCES

1
i. F . R i n a l d i , . J o u r n e o s d ' In f o r m a t . I o n i'Et.7, - AViER, L u x e m b o u r g , n u v. 1 9 / '

VoL . i E U R ' ^ 0 1 d , c, f, J 7 S - . v l .

E . J a h n , V c r h a l i t i i w a r m i V s t . e r S t a b l e i n L a n g z e i ;.; S ' a n d v - r s u r h b'-i' ' 0 0

bi-:; / 0 0 ' T . T o i l i : i : E r g < b n i t - s < • cl'-r 7,<- i t . ; ; ' . a n d v i ' s u r h f ' a n f< I T : t : s i - h >
1
Hohi-crisLahJ'-n, Ar:-h . EJs - n h ! J t L'TIW.T.. n ^ S( i 9 - / '• H ' • / £ > , ' . 9 -6 /

i. M . i ' a u b o , I m p r o v e d f'crrilir- s t e e l s | V , r ; ;ijprrh< a i m t u b i n g , A S M E , W; n : i•]'

n
Annual M e e t i n g , P h i I a d o l p h i a , i n / - ; . 1 9 6 '., p a p r r N 6 i - W A - . '1(>

4. l : .B e t * zirf-b.!-, Q i , . |'i\-ij{!.i:;--h" VcrwctHiung v . n i.ang/.f.- i :,W"iM_.(i:, Mi '.i-ii.

d<T VGB(d<".-. L 9 6 ) , H81 - 4 I 6 - . 1 . .

1
. 0 . 1 . .B ; h ' - i . , !•'. • r-1 • i ( i r aili^yoci s ' j - i H 1'i.ir f.upi r l r a ' ' l';;, .:.;iii'. i n ' r i , a : .
f
Con!', uii Cr''-i'p, N - wY o r k a n d . . o n d o n (' I 9 6 i I pap'-.-- N " 6 6

6. I, . M . I ' . H u p k : n - D . M u r r a y - D . P u v a : , : 'i-> -i - p ; ; r n p r r ! . i ' r . . r :-; • .in- b u' • - i v .d s

in s t ^ a r n p i p ' K , .' . i . S . i . ( a u g i l J 6 ' , . J a i 9 - 8 . '

/' . E . .i i s f . t - r — J . M i r k J " r ' a i t . h - A . il i g g i n b u ' r < i m , I I : g h — '< m p ' r a ' i i i 1 ' p r o p e r 1

'.••;,

u I' s t e c t m p i p i • w e l d s , B!S K A /IS i i o u f ' . , E n r . * , b < I ' M M I C ( a p r. 1 9 6 6 i p a p > ] - N' '

8. W . K u 1lin;:im - I 1
. [*• ' ' / i <• -\i< - F . J a h : - K.O.MUI I ' - r - l : . i i r h i • • f < r s • • •; n ,
1
/.<• i ! . ; ; ' a n d v < r:;n - h e a n S r h v i " i :;;;'*•'-i'b i n d u n g o n /.v;i:. \\f,i warm!"' ; : • • r . S i f S M J • ;.
^•.•hA'ei;;:-;,.,, ,,. S ;' I I I 1 . •i d i ' I I , ^ I : 9 f >! • ) i I !, H- [ .'

9. i-.H-ToT! - T . H. K u w b . r r y - i . n o w l ' . y - H .(J.W'-1 J< . T - .! . U a r T o r d , T!.- •- i -

r ' ' ) ill i o n i l l ' s e r v i o e :; i m u ! a ! \ o n "••;-;'.!; • J ( ' w •! (h; :u a •]•< . p c •:; : :; ' i n f
:.;!."•-I !:t";iRi|)ipi' wi'.h :;'•:•'.•:'••• f i e i -I ' o r m a u •• • , ''i'Ui'. i n I V • •! d : r i p > . | • •.-•••\)
H.-sisl.rai'. Hle< 1 : ; , Noui"£is-|. ' ! < b 1 9' 0 ' impel- N "

L0. T . B o n i :;/.ewski - R . I ' . K - n ' - 11.W . K i r k b . v , i •'• - a s i h i , i : y .,( v ; . •; d i i ; f , i - x - S O O

p i p e w i t h m a ' r h i n g '\l'••.: • r -dr:; , i b i d e m , p a p e r N " •.

1
1 1 . J . W i I. 1 i a m s o n — G . D . B r a n c h , Int.i :-na] p r e s ; u r e ••[ p 'i-:-.'ing o f b u '

w r i d e d l/.'%Cr-\ /,'%Mo-l /'A%y t i i b e ^ , i b i d e m , p a p e r N ' /

1 2 . K.Sachs - A.J.Baldock - D.G.Evar;;, The rr-eep st-r-'-ngt h of we J d s in

thick steampipes, ibidem, paper N" 8

 507

l j . KC'jC F i n a l r e p o r t : i ' n c p u f w r i d " d j " : n t r , , K<'S'-ar'b Agi''-< nvii' ;-J'

6.JLO-8'.», ' c> b f p u b l i s h e d b y Lhi-OM'K-i- I'oi-o f f i c i a l l'i:b:: a ' , .it.:; . '
1
t h e Eui'iipran I ' o m m u n i !, i • . ; ; , b u f l c p u s : m ,0 0 i , !.uxi'inh"iii'g

\ A . •'.!•'. E ; i i ' i m • " • — 0 . v a t ; R >. s s i i m - :• . K < M > d i - , • ' r •• p ' •!' K< •: c i > M ) j .• ! ; , • : • ' • .••!.'

o i 6 , l i i U - r n a : . i ' u n t ' . o t ; E r . g ; n < < T i ; i | ; , - i r . p 1 •;•'.-. .•!' . - j •< •> . p , : ' ' ! r ' : .• > i

iiji-pt. 1 9 8 0 ) p ; j p c r N " C i 8 / 8 0

1 i. . K . R i K ' d e - ' ' . F . E t . U'llll'' - O . v a n R i i s s u m , S t j . : ; - ' a n d:;;;-;:it> a t ; . ' : . V : ' ' •;-. t " '•
:
ci-.'np a n dpin.-:: : ;.• i t . y o f W ' d f d j u i n ' : ; i n , \ ; S ! - : 6 , '. b i d . p a p . r ."•)'. 'i Hi

1
1 6 . i,.S;j:"hs, S ; a * i s ' . i a . ' h ' A n s w . ' i " unp,:;mi' L h n d f i ; , S p r iiif,'-r V ' - r t .b i ' i ';.' I'.) '•

]'/. E U K 0 N 0 R M I . ' o - / ' . : T(\'-.i.s a - i •, < • • • : ; , ' i - d - . ' H i p . r a i i i ' - r - . ' r . - ' p •••:•.•:: , u .•;i"!-:

i8. ISO/TC 17/WG 10 E'lT - SG N " •] • - d o c .'..4 , iioct. . J , fi;',._::3

50$
 THE INFLUENCE OF FLUX COMPOSITION ON THE
CHEMISTRY AND CLEANLINESS OF 2 l/4Cr-lMo
STEEL ELECTROSLAG WELD METAL

J.J. Oldani, R.H. Frost, and G.R. Edwards

Center for Welding Research

Department of Metallurgical Engineering
Colorado School of Mines
Golden, Colorado 80401

ABSTRACT

The chemistry and cleanliness of 2 l/4Cr-lMo electroslag

weld metal have been correlated with the flux composition.
Reverse polarity electroslag welds of 2 l/4Cr-lilo steel were
produced without dilution using an oxide based flux in a
specially designed crucible. Specific additions of C r 2 O 3 and
MnO were made to the flux, and resulting changes in weld
metal chemistry were determined at various heights from the
starting tab. Results indicated that the composition of the
weld metal was strongly dependent on flux composition, but
moved toward steady state concentrations.
Electrochemical reactions have been postulated to
dominate certain thermochemical equilibria in D.C.
electroslag welding processes. These reactions include: the
anodic oxidation of alloy elements such as chromium,
manganese, and silicon; oxygen pickup at the anode; and the
reduction of metallic cations from the flux at the cathode.
The alloy element oxides present in the flux were reduced at
the cathode and depleted from the flux, and fluxes which had
high oxygen activities provided continuous anodic oxygen
pickup by the discharge of oxygen anions.
The results illustrate the fallacy of making batchwise
flux additions to a weld in progress, and also infer methods
of weld metal chemistry control. For example, non-steady
state, continuous additions of flux to the slag pool could be
utilized to reduce concentrations of temper embrittling
elements such as manganese and silicon.

INTRODUCTION

Electroslag welding is a high deposition rate welding

process in which a consumable wire electorde is mexted
through an ohmically heated flux pool to produce single pass
welds. The chemistry of the weld is controlled by the
electrode and flux compositions, and by thermochemical and
electrochemical reactions which occur in the flux pool. A
distinguishing characteristic of the electroslag welding

509
 510

process is that the flux pool moves upward with the weld;
hence, both the flux pool and the weld deposit are subject to
time dependent composition changes. In submerged arc, coated
electrode, or flux cored wire welding processes, the flux is
continuously added and freezes off behind the weld pool.
Thus, time dependent composition changes are not significant,
and the flux does not approach a steady state composition. A
major objective of this research is to evaluate the time
dependent chemical changes which occur in the flux pool
during electroslag welding, and to investigate the use of
continuous flux additions to control the weld metal chemistry
and cleanliness. An example of this practice is the use of
continuous aluminum additions to deoxidize the slag bath in
electroslag ingot making processes.
Traditional electroslag welding fluxes have been oxide
based, with compositions inherited from submerged arc welding
fluxes. These fluxes have high oxygen activities, and allow
both oxygen pickup and alloy element oxidation losses. Few
fundamental studies of electroslag welding fluxes have been
made. Reviews of electroslag refining flux properties and
requirements which bear on the chemistry of D.C. electroslag
welding processes are given by Nafziger et. al. , Latash and
Medovar^, and Duckworth and Hoyle^. Oxygen content in
electroslag welds was investigated by Nakano et. al.^ Other
reviews on the influence of flux composition on the chemistry
of electroslag welds are given by Zeke^ and Edwards and Frost**.
Current electroslag welding procedures utilize two
practices which may be undesirable. First, alloy element
oxidation losses are generally counteracted by increasing
their concentrations in the wire electrode. This practice is
expensive, and it does not account for time dependent changes
in the flux pool chemistry which tend to limit oxidation
losses. Second, the practice of replenishing the flux pool
by batch additions of cold flux can cause serious chemical as
well as thermal transients. These transients can result in
the production of weld defects or in poor mechanical
properties. Based on data presented in this study, it is
proposed that continuous flux additions during electroslag
welding would provide for control or modification of weld
metal chemistry and enhance metal cleanliness.

EXPERIMENTAL PROCEDURE
Experimental D.C. electroslag welds were produced in a
cylindrical water cooled crucible to eliminate composition
changes resulting from dilution by the base metal. A weld
diameter of 58 mm (2.3 in.) was used to match the fill ratio
and weld velocity of a 100 mm (4 in.) thick electroslag plate
weld. The weld was initiated on a 2 l/4Cr-lMo starting tab
to minimize bottom contamination, and contamination from the
 511

crucible was prevented by a 2 mm thick slag skin which formed

between the weld surface and the crucible wall. The slag
skin also facilitated removal of the weld from the crucible.
The welds were produced in the reverse polarity mode at
a constant power input of 20 KW. The currents varied between
500 and 600 amperes, and the potentials varied between 30 and
40 volts. A consumable guide tube was used, and the
electrode velocity was held constant at 115 ram/s (4.5 in./s).
The guide tube dilution of the metal was less than 1.7 volume
percent. A single 500 gram flux charge was added at the
beginning of the weld. The finished welds were completed in
about 34 minutes, and were about 400 mm (15.7) in length.
The influence of the flux composition on weld metal
chemistry was investigated by making systematic reagent grade
manganese(II)oxide or chromium(III)oxide additions to a
commercial oxide base flux. The approximate composition of
the base flux is given in Table I, and the experimental
matrix of manganese oxide and chromium oxide additions is
given in Table II.
TABLE I. Commercial oxide base flux composition in
weight percent.

SiO2 CaO MnO MgO CaF A1 2 O 3 miscellaneous

2
35 20 6.5 5 15 15 3.5

The welds were sectioned at distances of 25, 50, 100,

200, and 350 mm (1, 2, 4, 8, and 14 in.) from the starting
tab at the bottom of the crucible. Chemical analyses for
alloy elements were performed by emission spectroscopy, and
analyses for oxygen was performed using a Leco interstitial
analyzer. All chemical analyses were made at points between
1/4 and 3/4 of the crucible radius to avoid center and edge
effects. The chemistry of the wire electrode was also
determined by the same methods.

RESULTS AND DISCUSSION

Analysis of the experimental welds shows substantial
variations in weld pool chemistry with changes in flux
composition and with distance along the weld. Table III
gives the chemical analyses of the welds at a height of 50 mm
(2 in.) from the starting tab, and Table IV gives the
chemical analyses at a height of 350 mm (14 in.) from the
start of the weld. The most significant chemical variations
were found for the elements chromium, manganese, silicon, and
oxygen. Figure 1 shows the concentration changes for these
elements as a function of distance from the starting tab for
 512

TABLE II. Matrix of Experimental Fluxes.

%Cr2O3 FLUX ADDITION

WELD
No 0 5 10 15
/o MnO FLUX ADDITION

0 1 2 3 4

5 5 6 7 8

10 9 10 1! 12

15 13 14 15 16
w

0.4 , , . , , • • • r • • —

400
WE LD No.l - COMMERCIAL OXIDE BASE FLUX
300
OXYGEN-.
•L 0.2 - O - 2 0 0 -7
"75—
a.
LU a.
-- 100
Si'> ^
X u • ~~^z- saa>_ n - 0 i-
o 2
_ Cr Ul
z V^ z
o o
A
_ o
|-0.2 2
i- LU
z CO
LU o Cr
U
A Mn X
|-0.4 a Si o
o O 0
1 I i i i i i i —i : 1 L___I

100 200 300

DISTANCE FROM STARTING TAB (mm)

Figure 1. Chemical analysis for weld metal, plotted as a

function of distance. The flux used was the oxide flux whose
composition is given in Table I.
 513

TABLE III. Chemical Analysis of Weld Metal (w/o at 50 m m ) .

ppm
WELD1 Cr Mn Si c Mo Ni s P Sn O
1 2 .67 0 .90 0 .33 0 .11 1 .04 0 .11 0 .016 0 .016 0 .011 166
2 3 .12 0 .62 0 .17 0 .11 1 .02 0 .13 0 .018 0 .016 0 .011 177
3 3 .49 0 .48 0 .10 0 .11 0 .11 0 .C19 0 .018 0 .011 377
4 3 .74 0 .38 0 .09 0 .12 1 .02 0 .11 0 .022 0 .019 0 .010 714

5 2 .76 1 .27 0 .27 0 .12 1 .08 0 .11 0 .014 0 .016 0 .011 146
6 2 .88 0 .80 0 .14 0 .12 1 .07 0 .13 0 .011 0 .018 0 .013 164
7 3 .30 0 .59 0 .08 0 .12 1 .08 0 .13 0 .013 0 .019 0 .013 398
8 3 .70 0 .56 0 .07 0 .13 1 .06 0 .11 0 .017 0 .017 0 .010 633

9 2 .66 1 .68 0 .22 0 .13 1 .06 0 .11 0 .012 0 .014 0 .011 128
10 3 .07 1 .13 0 .12 0 .14 1 .08 0 .11 0 .014 0 .014 0 .010 227
11 3 .28 0 .89 0 .09 0 .12 1 .07 0 .11 0 .015 0 .013 0 .010 442
12 3 .61 0 .77 0 .08 0 .13 1 .06 0 .11 0 .018 0 .013 0 .011 633

13 2,.62 1 .86 0..17 0 .12 1 .08 0 .11 0 .012 0 .012 0 .010 146
14 3 .01 1 .38 0..09 0 .12 1 .07 0 .11 0 .014 0 .013 0 .011 222
15 3,.28 1,.00 0.,07 0..12 1 .06 0 .11 0 .015 0 .014 0 .011 497
16 3,.63 0,.87 0.,05 0,.12 1 .04 0,.11 0 .013 0 .014 0 .010 750

wire 2..81 1..11 0.,20 0,.12 1..04 0,.14 0 .01 0 .013 0 .010 70

rABLE IV.
i
Chemical Analysis of Weld Metal (w/o at 350 mm)

ppm
WELD Cr Mn Si C Mo Ni S P <3n O

1 2. 76 0.,89 0. 18 0..11 1..04 0.,13 0,.014 0..015 0,.010 236

2 2. 80 0.,96 0. 17 0.,10 1.,03 0.,11 0..015 0.,015 0,.011 284
3 2. 83 0.,96 0. 17 0.,11 1.,03 0.,11 0..015 0.,016 0,.010 329
4 2. 83 0. 99 0. 16 0. 10 1. 02 0. 10 0..015 0.,017 0.,011 360

5 2. 60 0. 97 0. 16 0. 13 1. 10 0. 14 0.,009 0. 016 0.,013 287

6 2. 57 0. 97 0. 15 0. 13 1. 08 0. 13 0.,009 0. 016 0.,012 239
7 0. 89 0. 15 0. 11 1. 09 0. 13 0. 009 0. 017 266
8 2. 84 0. 94 0. 12 0. 12 1. 06 0. 10 0.,013 0. 015 0.,010 312

9 2. 76 1. 07 0. 15 0. 14 1. 10 0. 11 0. 012 0. 013 0. 010 236

10 2. 64 0. 98 0. 13 0. 15 1. 11 0. 11 0. 013 0. 011 0. 010 261
11 2. 73 1. 12 0. 14 0. 13 1. 11 0. 11 0. 013 0. 013 0. 010 316
12 2. 65 1. 12 0. :.6 0. 11 1. 11 0. 11 0. 013 0. 011 0. 010 312

13 2. 69 1.11 0. 15 0. 12 1. 09 0. 11 0. 013 0. 011 0. 010 260

14 2. 82 1.27 0. 15 0. 11 1. 06 0. 11 0. 012 0. 013 0. 012 243
15 2. 87 1.36 0. 15 0. 12 1. 08 0. 11 0. 014 0. 012 0. 011 255
16
 514

weld number l f which was produced using the unaltered

commercial flux specified in the experimental procedures.
The results show chromium and manganese oxidation losses, and
silicon and oxygen pickup from the flux. Examination of the
time dependence shows that the concentration changes for
chromium and silicon decrease to zero during the first 200 mm
of weld, while the changes in concentration for manganese and
oxygen are approximately constant over the whole weld length.
The chromium oxidation loss decreases from a 0.4 w/o
loss at the start of the weld to zero at a distance of 200 mm
from the starting tab. This appears to be the result of
chromium oxide buildup to a neutral concentration at which no
further gain or loss is observed. Manganese also shows an
oxidation loss, but the loss is constant. Based on
thermochemical reasoning, the manganese loss would be
expected to drop as the manganese oxide concentration in the
flux increased. The fact that thermochemical eqilibrium is
not approached indicates that electrochemical reactions may
control manganese behavior.
Silicon shows a 0.2 w/o pickup from the flux at the
start of the weld with the silicon pickup falling to zero
after 200 mm of weld metal has been deposited. The behavior
of silicon is opposite that of chromium, but it shows a time
dependence which can be explained by thermochemical
arguments. The initial silicon pickup indicates that the 35
w/o silica initially present in the flux is in excess of the
neutral concentration. Thus, silicon partitions from the
flux to the weld pool until the silica content of the flux
falls to neutrality. Oxygen shows a constant pickup of about
200 ppm over the whole length of the weld. As with
manganese, the reactions controlling weld metal oxygen
concentration do not appear to approach equilibrium. Thus f
oxygen behavior may be the result of electrochemical rather
than thermochemical reactions.
The high current densities present in electroslag
welding favor a large electrochemical influence on weld
chemistry. Previous divided cell electrochemical experiments
at the Colorado School of Mines using a CaF2-10w/oAl 2 03 flux
have shown that alloy elements are oxidized at the anode, and
are reduced from the flux at the cathode . Also, oxygen
anions are discharged and picked up at the anode, and some
oxygen refining occurs at the cathode. Electrical conduction
was found to be ionic, with a large fraction of the current
at both the anode and cathode being carried by iron
reactions. Chemical analysis of buttons recovered from both
sides of the electrochemical cell shov/ed that all of the
alloy elements were oxidized into the flux at the anode,
while the less reactive alloy elements such as molybdenum and
chromium were preferentially reduced at the cathode. The
more reactive elements such as silicon and manganese showed
greater oxidation at the anode than reduction at the cathode.
Thus, substantial net oxidation losses of these elements were
 515

observed. The overall electrochemical behavior can be

interpreted using a tentative electrochemical series shown in
Table V.
TABLE V. Tentative Electrochemical Series for
Reactions in Electroslag Welding Fluxes.
Ca = Ca + 2 + 2e
Al = A l + 3 + 3e
Ti = T i + 2 + 2e
Si = S i + 4 + 4e
Mn = Mn +2 + 2 e
(0" 2 ) = (0) + 2e
Cr = C r + 2 + 2e
Fe = F e + 2 + 2e

Cr +2 = C r + 3 + e
Co = C o + 2 + 2e
Ni = N i + 2 + 2e
Mo = Mo + 3 + 3 e

The more reactive elements are listed at the top of the

table, and the least reactive elements are listed at the
bottom. In reverse polarity electroslag welds, the wire
electrode is the positive anode, and the weld pool is the
cathode. The current density at the wire anode can be as
high as 1000 amperes per square cm (10 A/mm.). This causes a
high anodic overpotential and the oxidation of elements on
the electrochemical series as low as molybdenum. The current
density at the cathodic weld pool is of the order of 10 to 50
amperes per square cm.(0.1 to 0.5 A/mm.), and the lower
cathodic overpotential results in the preferential reduction
of the less reactive elements. Thus, silicon and manganese,
which are high in the electrochemical series, are oxidized
into the flux by the high overpotential at the anode, but are
not reduced to the same extent at the cathode. Based on this
model, the silicon concentration should eventually fall to a
net loss, and it will be shown that this occurs when chromium
and manganese oxides are added to the flux.
The same reasoning can be used to explain the behavior
of oxygen which does not appear to approach thermochemical
 516

equilibrium. Oxygen anions present in the flux are

discharged and picked up at the anode because of the high
overpotential and the high oxygen activity in the flux.
Oxygen refining reactions at the cathode are not carried out
to the same extent because of the low cathodic overpotential
and the lower oxygen activity in the weld metal. The net
result is a continuous pickup of oxygen over the entire
length of the weld.
Figures 2 through 5 show the influence of chromium oxide
flux additions on the changes in chromium, manganese,
silicon, and oxygen concentrations. Figure 2 shows a
substantial increase in weld metal chromium content with
increasing chromium oxide additions to the flux. The
chromium pickup is greatest in the early portion of the weld,
and it falls to approximately zero as the chromium partitions
from the flux to the weld metal. Based on these data, it
appears that a 2.5 w/o chromium oxide addition to this
specific commercial flux would achieve approximate chromium
neutrality, and maintain a constant chromium content along
the length of the weld.
Figure 3 shows that the manganese oxidation losses are
substantially increased by chromium oxide additions to the
flux. The losses are greater at the beginning of the weld
while the chromium oxide content of the flux is high, and a
net loss of about 0.2 w/o manganese is approached with
increasing distance along the weld. A 15 w/o chromium oxide
addition to the flux causes a 0.76 w/o manganese loss at the
start of the weld. This represents the loss of a substantial
fraction of the manganese present in the alloy. This
apparent non-equilibrium behavior of manganese with chromium
oxide additions can be explained in terms of electrochemical
reactions. Chromium is lower in the electrochemical series
than manganese, and the presence of a large concentration of
chromium cations in the flux blocks the reduction of
manganese at the cathode. After the high chromium
concentration in the flux has been depleted, the reduction of
manganese is still partially blocked by the presence of iron,
which is also below manganese in the electrochemical series,
and a net manganese oxidation loss of about 0.2 w/o is
sustained.
Figure 4 shows the behavior of silicon with chromium
oxide additions to the flux. Increasing chromium oxide
additions shift the silicon concentration change at the
beginning of the weld from a pickup of 0.2 w/o with the
unaltered commercial flux to a loss of 0.12 w/o with a 15 w/o
chromium oxide addition. The silicon composition change
approaches a net continuous loss of about 0.03 w/o with
increasing distance along the weld. This behavior is similar
to that shown by manganese, and it indicates that silicon
reduction at the cathode is hindered by the presence in the
flux of chromium, which is lower in the electrochemical
series. For the unaltered commercial flux, the silicon
 517

WELD No. FLUX ADDITIONS

%Cr g 0 s "AMnO
o 1 0 0
A 2 5 0
D 3 10 0
o 4 15 0

-o

-0.6
100 200 300
DISTANCE FROM STARTING TAB (mm)
Figure
g The chromium composition as a function of
distance for welds produced with various flux additions

1.0
WELD No. FLUX ADDITIONS
0.8 «/ o Cr 2 0 3 "ZoMnO
o 1 0 0
A 2 5 0
D 3 10 0
o 4 15 0

-0.8
100 200 300
DISTANCE FROM STARTING TAB (mm)
Figure 3. The manganese composition as a function of
distance for welds produced with various C^O-, flux additions.
 518

t i l l 1 1 t i 1 1 1 • 1

0.2 - Q WELD No. FLUX ADDITIONS

\ *»/oCr203 VoMnO
• \
O 1 0 0
A 2 5 0
x O.I D 3 10 0
o
o 4 15 0

CE
0 "

z
UJ .——— ^ - — • — ^

O
^
-
i-o. o ^ o—
CO

-0.2 i i i
1 1 1 ...J 1 1 1 L I '

100 200 300

DISTANCE FROM STARTING TAB (mm)
Figure 4. The silicon composition as a function of
distance for welds produced with various C ^ ? flux additions,

800 WELD No. FLUX ADDITIONS

w
/oCr 2 0 3 */oMnO
O 1
OOOO
0
A 2 5
t • 3 10
£600 O 4 15

UJ
(- 400
z
o
o
z
UJ

>-
200
X
o

0
100 200 300
DISTANCE FROM STARTING TAB (mm)
Figure 5. The oxygen content as a function of distance
for welds produced with various Cr 2 O 3 flux additions.
 519

concentration change is shown to cross the line from a pickup

to a net loss with increasing distance along the weld. This
confirms the apparent non-equilibrium behavior of silicon and
supports the hypothesis that electrochemical reactions are
over-riding thermochemical equilibria to control process
chemistry.
The influence of chromium oxide additions on weld metal
oxygen content is shown in Figure 5. The actua] oxygen
content is plotted rather than the change in oxygen content,
and the dashed line indicates that the electrode initially
had an oxygen content of 70 ppm. A 15 w/o chromium oxide
addition to the flux shifts the oxygen content at the start
of the weld from 200 ppm tc 800 ppm. Beyond 200 mm (8 i n . ) ,
a constant oxygen pickup of 200 to 350 ppm is observed for
welds produced with the unaltered flux and for chromium oxide
containing fluxes.
The dramatic increase in oxygen pickup to 800 ppm with
chromium oxide additions can be explained by the
chermochemical partitioning of trivalent chromic oxide to
divalent chromous oxide plus oxygen:

Cr2O 3 = 2 CrO + 1/2 O 2 (1)

This reaction would substantially increase the oxygen

activity in no flux in the same manner that ferric oxide
raises the oxygen activity in steelmaking slags. Rankin and
Biswas 7 conclude that divalent chromium is more stable in
high silica slags with the CaO/SiO 2 ratio less than about
2,0, and that trivalent chromium is more stable in highly
basic slags. The commercial flux used in these experiments
is approximately neutral with a ratio of (CaO+MgO+MnO)/SiO2
of about 1.0. Thus the reaction to form chromous oxide is
probable, and the increase in oxygen activity of the f]ux
accounts for the substantial oxygen pickup.
Figures 6 through 9 show the influence of manganese
oxide flux additions to the commercial flux on the changes in
chromium, manganese, silicon, and oxygen contents. Figure 6
shows that the addition of manganese oxide to the flux has
only a small influence on chromium behavior. Chromium shows
an average net loss of 0.1 to 0.2 w/o with a slightly greater
loss at the beginning of the weld. The lack of influence of
manganese oxide on chromium behavior may be attributed to the
fact that manganese is higher in the electrochemical series
than chromium, and chromium is preferentially reduced at the
cathode in spite of the increased manganese cation
concentration.
Figure 7 shows that manganese oxide flux additions
substantially increase the weld metal manganese
concentration, just as chromium oxide flux additions
increased the weld metal chromium concentration. The
unaltered commercial flux, which contained about 6.5 w/o
manganese oxide, showed a consistent manganese loss along the
 520

^ 1.2 i i i • • i i

WELD No. FLUX ADDiTIONS

% Cr2 0 3 %MnO
CO
2 0.8 O 1 0 0
A 5 0 5
X
o • 9 0 10
O 13 0 15
N0I1

0.4
cc
CONCEN

o
-0.4

0 100 200 300

DISTANCE FROM STARTING TAB (mm)
Figure 6. The chromium concentration as a function of
distance for welds produced with various MnO flux additions.

1.0 -i 1 r 1 1 1 1 1—
WELD No. FLUX ADDITIONS
0.8
w/oMn0
-~ 0.6
LU
% 0.4
<
" 0.2

a:
t- -0.2
O _
0.4
o
o
-0.6

-0.8
0 100 200 300
DISTANCE FROM STARTING TAB (mm)
Figure 7. The manganese concentration as a function of
distance for welds produced with various MnO flux additions.
 521

length of the weld. Manganese oxide flux additions caused a

manganese pickup of up to 0.8 w/o at the start of the weld;
however, the change in manganese concentration crosses over
from a pickup to a loss with increasing weld distance. This
crossover from gain to loss indicates that electrochemical
considerations will prevent the formulation of a flux
composition neutral to manganese by means of manganese oxide
additions. It appears that the only methods of insuring a
particular weld metal manganese content are to increase the
concentration in the electrode and accept a constant
manganese loss during welding, or to add manganese oxide to
the flux on a continuous basis.
Figure 8 shows the effect of manganese oxide flux
additions on silicon concentration changes. The influence of
manganese oxide additions on silicon is similar to the
influence of chromium oxide additions, because both manganese
and chromium are below silicon in the electrochemical series.
The initial silicon gain with the unaltered flux crosses over
to a silicon loss with increasing weld distance, and
increasing manganese oxide flux additions move the crossover
point closer to the start of the weld. The presence of a
high manganese concentration in the flux appears to hinder
the reduction of silicon at the cathode, causing a net
silicon loss. The fact that silicon approaches a net
electrochemical loss with increasing distance along the weld
indicates that a flux composition which is neutral to silicon
may not be feasible.
Figure 9 shows that manganese oxide flux additions have
almost no influence on the oxygen content of the weld metal.
The average oxygen content increases from about 150 ppm to
about 250 ppm over the length of the weld, but the effects of
manganese oxide additions of up to 15 w/o can not be
distinguished. Manganese oxide additions, therefore, do not
appear to increase oxygen activity in the flux. This
behavior is reasonable, since manganese(II)oxide contains
manganese in its lowest oxidation state.
The experimental data presented in Figures 1 through 9
show that flux additions have a strong influence on weld
metal chemistry. This influence decreases with time after
the flux addition; therefore, a long term modification of
weld metal chemistry would require continuous flux additions.
The maximum chemical influence of the flux additions is
observed at a height of 50 mm (2 in.) from the starting tab.
Thus, an estimate of the influence of continuous additions
can be made by comparing the chemistries of all sixteen welds
at this point. This comparison is provided by Figures 10
through 13, which summarize the effects of chromium oxide and
manganese oxide additions on the chromium, manganese,
silicon, and oxygen contents of the weld metal. The
concentration changes for each element are plotted versus the
w/o manganese oxide addition, and the influence of chromium
oxide is shown parametrically by plotting separate curves for
 522

1 1 I f 1

~ 0.2 - WELD No. FLUX ADDITIONS

\ «/o Cr 2 O s -ZoMnO
O 1 0 0
o \ A 5 0 5
• 9 0 10
I O 13 0 15
o
o
H 0 ~ ^ — A
o
z
UI
o
o
tn
-0.2
i t i i i i i I i i i

0 iOO 200 300

DISTANCE FROM STARTING TAB (mm)
Figure 8. The silicon concentration as a function of
distance for welds produced with various MnO flux additions,

800 WELD No. FLUX ADDITIONS

w/o Cr 2 0 3 <«/oMnO
E
a o 1 0 0
A 5 0 5
600
• 9 0 10
2 O 13 0 15

0 J_ JL
100 300 200
DISTANCE FROM
TAB (mm) STARTING
Figure 9. The oxygen content as a function of distance
for welds produced with various MnO flux additions.
 523

0, 5, 10, and 15 w/o additions.

Figure 10 shows that chromium oxide additions increase
the chromium content of the weld metal, and that manganese
oxide additions have a small negative effect on chromium
concentration. Figure 11 shows that manganese concentration
changes are strongly influenced by both manganese oxide and
chromium oxide additions. Manganese oxide additions increase
the manganese content of the weld metal, and chromium oxide
additions lower the manganese concentration because of the
preferential reduction of chromium cations at the cathode.
Additions of equal amounts of chromium and manganese oxides
produce a manganese loss of about 0.2 w/o, which is about the
same as that observed for the unaltered commercial flux.
Figure 12 shows that manganese and chromium oxide
additions both lower the weld metal silicon content, and that
the influence of chromium oxide is much stronger than that of
manganese oxide. The combined addition of 15 w/o of each
oxide lowers the silicon concentration by 0.15 w/o. This
represents a 75 percent reduction in the weld metal silicon
content. Figure 13 shows that manganese oxide additions have
a negligible influence on weld metal oxygen pickup, but that
chromium oxide additions result in oxygen pickup to levels as
high as 800 ppm oxygen.
Figures 10 through 13 show that continuous chromium
oxide additions to the flux will increase the chromium and
oxygen contents of the weld metal and will decrease the
manganese and silicon concentrations. Continuous manganese
oxide additions to the flux will increase the weld metal
manganese concentration, decrease the silicon concentration,
and will have little or no effect on chromium and oxygen
contents.
The concept of utilizing continuous flux additions to
control weld metal chemistry has several important
implications for electroslag welding practice. First, the
concentration of the alloy element lowest in the
electrochemical series can be maintained or increased by flux
additions, without having to increase the initial
concentration in the electrode. This applies to chromium for
the case of 2 l/4Cr--lMo steel investigated in the present
experiment. Second, the weld metal concentrations of
elements higher in the electrochemical series will be
difficult to maintain by flux additions, because of the
preferential cathodic reduction of the least reactive alloy
element. This applies to silicon and manganese in the
present experiments. Where concentrations of reactive alloy
elements must be maintained, it appears that the best method
is to increase their contents in the original electrode.
An important corollary for 2 l/4Cr-lMo steel is that a
reduction of manganese and silicon concentrations in the weld
metal may be desirable to reduce temper embrittlement
susceptibility. Swift 8 , Watanabe et. a l . 9 , and Murakami et.
al. °, have used a temper embrittlement factor based on alloy
 524

E t

O - Cr 2 0 3 FLUX
10
ADDITIONS

UJ
o • 0-—
0.8 - V
D
." — - —
D B
0.4 -
A
^ 5w/o

< A
(K 0 O
^0%
O
u
i-0.4
o
o i
-0.8
5 10 15
MnO FLUX ADDITION (v%)
Figure 10. The chromium concentration for the initially
deposited weld metal for combinations of both flux additions.

£
Cr,0 3 FLUX ADDITIONS
O
in
o

-0.4 -

I
5 ro 15
MnO FLUX ADDITION (w/o)
Figure 11. The manganese concentration for the initially
deposited weld metal for combinations of both flux additions.
 525

Cr_0. FLUX ADDITIONS

5 10 !5
MnO FLUX ADDITION (•/„ )

Figure 12. The silicon concentration for the initially

deposited weld metal for combinations of both flux additions.

•^1000
Q.
Q.

800 Cr,03 FLUX ADDITIONS

o
in
600

UJ 10%^
o 400-
•z.

o
»*\ A
^200
0w/o"
>- o
j^-wire
X
o
5 10 15
MnO FLUX ADDITION (w/o)
Figure 13. The oxygen content for the initially deposited
weld metal for combinations of both oxide flux additions.
 526

chemistry to predict the temper embrittlement susceptibility

of 2 l/4Cr-lMo steel:

J = (w/o Mn + w/o Si)(w/o P + w/o Sn) x 104 (2)

A J-factor of less than 200 indicates a resistance to temper
embrittlement as measured by changes in the Charpy impact
transition temperature after an embrittling heat treatment.
Swift shows that a reduction of the J-factor from 300 to 100
reduces the AT54 by about 30°C (AT54 is the increase in the
54J transition temperature caused by an embrittling heat
treatment). Such a decrease in temper embrittlement
susceptibility represents significant improvement in weld
metal toughness.
Equation 2 shows that the manganese and silicon contents
have a significant influence on the J-factor. Therefore, the
use of continuous chromium oxide additions to reduce weld
metal manganese and silicon contents can be expected to lower
the temper embrittlement susceptibility of 2 l/4Cr-lMo weld
deposits and improve their toughness. Figure 14 shows the
influence of chromium and manganese oxide additions on the
J-factors for the sixteen experimental welds. The J-factors
were calculated based on the chemical analyses at a height of
50 mm from the starting tab.

500 Cr203 FLUX ADDITIONS

£ 400

INITIAL J (FILLER WIRE) -\

A

o —•
f- S5 200
o +

~Z>L~J I O O

5 10 15
MnO FLUX ADDITION (w/o)

Figure 14. The J-factor for the initially deposited weld

metal for combinations of both oxide flux additions.
 527

The chemistry at this location is indicative of that which

could be obtained by using continuous additions to maintain
the respective chromium and manganese oxide concentrations in
the flux. The J-factor can be reduced to below 200 by
maintaining a chromium oxide concentration of 10 w/o in the
flux. With a manganese oxide content of 11 w/o ( 5 w/o
addition plus 6 w/o present in the commercial flux) the
chromium oxide content in the flux must be maintained at 15
w/o to keep the J-factor below 200.
An observation provided by these experiments is that
some method of flux pool or weld pool deoxidation must be
employed with oxide based fluxes to achieve low oxygen
content welds. Initial oxygen pickup of at least 100 ppm and
steady state oxygen pickup of at least 200 ppm were observed
for all the flux compositions studied, and much larger oxygen
increases were found to result from chromium oxide additions.
Two deoxidation techniques appear to be feasible. One is to
deoxidize the flux pool by continuous additions of aluminum
or magnesium, and the second is to increase the aluminum
content in the welding wire to provide a self killing
electrode.

CONCLUSIONS

1. Thermochemical and electrochemical reactions cause the

composition of electroslag welding fluxes to change with time
or distance along the weld. The flux approaches a steady
state composition which may promote the constant pickup or
loss of some alloy or tramp elements.

2. Alloy element concentration changes in the weld metal are

influenced by their oxide concentrations in the flux, by
their relative electrochemical reactivity, and by the flux
concentration of less reactive alloy oxides.

3. In D.C. electroslag welds, alloy elements low in the

electrochemical series such as chromium are oxidixed at the
anode and reduced from the flux at the cathode. Their
concentrations in the weld metal can be controlled by the
addition of their oxides to the flux.

4. In D.C. electroslag welds, alloy elements high in the

electrochemical series such as manganese and silicon are
oxidixed at the anode, but their reduction at the cathode is
hindered by the preferential reduction of iron and less
reactive alloy elements. These elements tend to show
constant oxidation losses during welding, and their weld
metal concentrations are difficult to control by flux
additions.
 528

5. Oxygen pickup was observed for all oxide based flux

compositions examined in these experiments. Oxygen pickup
increases dramatically when higher oxidation state alloy
oxides are added to the flux.

6. For 2 l/4Cr-lMo steel electroslag welds, continuous

chromium oxide additions to the flux can be used to reduce
silicon and manganese contents in the weld metal, and hence,
to reduce the susceptibility to temper embrittlement as
measured by the the J-factor.

ACKNOWLEDGEMENTS

The support of the Department of Energy, Advanced

Research and Development Fossil Energy Materials Program, is
gratefully acknowledged. Materials for this project were
supplied by Stearns-Roger Corporation, Chicago Bridge and
Iron, and Lukens Steel Company. The interest and assistance
of Dr. Robert A. Swift of Lukens Steel and Mr. James Sims of
Chicago Bridge and Iron are especially acknowledged.

REFERENCES

1. R. H. Nafziger, et. al., The Electroslag Remelting

Process, U.S. Bureau of Mines Bulletin 669 (1976)

2. Ya. V. Latash and B. I. Medovar, Electroslag Remelting,

Moscow, (1970)

3. W. E. Duckworth and G. Hoyle, Electro-Slag Refining,

Chapman and Hall (1969)

4. S. Nakano, K. Tamaki, and J. Tjuboi, Slag-Metal Reactions

And Origins Of Oxygen In ESW Process, IIW/XII-J-67-78 (1978)

5. J. Zeke, Quality Of The Electroslag Weld Metal As A

Result Of Wire/Flux Combination, IIW/XII-J-64-78 (1978)

6. G. R. Edwards and R. H. Frost, Characterizing and

Improving the Toughness of 2 l/4Cr-lHo Electroslag Weldments,
ORNL/SUB 72 19/1, Final Report, DOE Contract No.
DE-ASO5-OR12313 (October 1981)

7. W. J. Rankin and A. K. Biwas, The Nature and Behavior of

Chromium in Stainless Steelmaking Slags, Mineral Science
Engineering, vol. 7, no. 2, pp. 89-98 (April 1975)
 529

8. R. A. Swift, The Effects Of Composition And Heat

Treatment On The Mechanical Properties Of 300 MM Gage 2 1/4
Cr - 1 Mo Steel Plate, ASTM STP 755, pp. 166-188 (May 1980)
9. J. Watanabe, et. al., Temper Embrittlement of 2 l/4Cr-lMo
Pressure Vessel Steel, 29th Annual Petroleum Mechanical
Engineering Conference Proceedings, pp. 15-18, Dallas,
American Society of Mechanical Engineers (1974)
10. Y. Murakami, T. Nomura and J. Watanabe, Heavy-Section 2
l/4Cr-lMo Steel for Hydrogenation Reactors, ASTM STP 755, pp.
383-417 (May 1980)
 Session J.

: HEAT EXCHANGER APPLICATIONS :

Co-Chairmen
Wpatrtorca C. R. Felmtey, Jr.
Oakfila^NaHonaliabaator/ W * l n o Research Counci
OakWdae,T«Tnos»©© New Ywk, NewYork
 EFFECT OF JOINT DESIGN AND FABRICATION
LIMITATIONS ON TUBE-TO-TUBESHEET WELDING
SAMUEL D. REYNOLDS, JR., FASM
USNRC, Region I
Washington, D. C.

BACKGROUND
A tube-to-tubesheet (T/TS) weld is the essential part of
a system that is designed to form a barrier in a heat exchanqer
between fluids of different pressures. In some cases, the
weld joint additionally provides structural support to the
heat exchanger by acting as a "stay". As indicated previously
in the literature(l), a T/TS weld is not an ASME Code pressure
retaining or pressure boundary weld. It does not form a
barrier between the internal pressure of the heat exchanqer
and the environment outside the heat exchanqer. The ASME
(e.g., SCIII, NX-5000) classifies a T/TS weld in a "special
weld" category rather than a category "A, B, C or D" nressure
boundary weld. T/TS weld qualification requirements are not
stated in ASME SCIX, but are found in the "book" Sections of
the Code. The qualification and fabrication requirements of
the ASME B & PV Code do not include volumetric inspection, but
rather only surface examination and statistical data indicating
metallurgical soundness and effective throat size.
Service conditions and the consequence of leakaqe determine
the requirements for the weld joint design. The service
conditions include temperature, differential pressure, erosion
effects of the fluid, mechanically- and thermally-induced
loadinq, and corrosivity of the fluids. There are cases where
the differential conditions across the tubesheet, including

533
 534

fluid compatibility, are i.urh that minor leakage (except for

loss of thermal performance) is inconsequential. There are
other cases, e.g., high pressure feedwater heater" (2, 3 ) ,
where leakage of one t/TS weld, if not quickly repaired, may
lead to catastrophic erosion of the tubesheet (TS) ligament
and/or other tubes. T/TS weld leakage is also unacceptable in
the case of petrochemical or chemical heat exchangers where
the two fluids may be total incompatible. In the sodium heat
exchangers for the breeder proqram (4,5), T/TS leakaqe can be
explosive or, at a minimum, a serious caustic corrosion problem.
T/TS leakage in a pressurized water reactor steam generator
results in introduction of radioactivity in the secondary side
of the power plant.

T/TS welds are specified for one or more of the following

engineering purposes: (6)
a. Mechanical expansion has been determined to be
unreliable due to service experience or anticipated
based on temperature/pressure conditions.
b. The engineering design requires the strength of
a welded joint to support the tube bundle. (In
a straight tube design, credit may be taken for
the T/TS weld as a tubesheet stay). (32)
c. T/TS welding may be the least expensive leak-free
joint method.
d. A completely crevice-free joint is required as
may be produced with a back side of the tubesheet
weld.
The historic methods and current sophisticated mechanical
joint T/TS connections have been previously discussed in the
literature and will not be reiterated (1, 7, 8, 9 ) . T/TS
welds are required where thermal conductivity differences
between che tubes and tubesheet, elevated temperature properties (39)
of the tube material or pressure drop across the tubesheet
dictate. Recently, the leak free advantages of T/TS welds
have been employed on low pressure drop designs for power
plant surface condensers to minimize contamination of the feed
water by the condenser cooling water.
 535

As previously stated, the T/TS weld is a part of the

tube-to-tubesheet connection system. The total system often
includes post weld roller expansion, explosive expansion or
hydraulic expansion. This system may introduce yield strength
level residual stresses in the weld area and/or expansion
transition area, and adverse metallurgical structures in the
weld HAZ. The post weld expansion may or may not completely
close the T/TS hole crevice.
The T/TS system may be relatively complex to permit third
fluid leak detection systems (4,5,10) as employed in some
of the breeder reactor systems.
The ability to make a reliable T/TS weld under fabrica-
tion conditions imposed is a function of the joint design and
cumulative (stacked) tolerances associated with all of the
parts employed, including those of the welding device. These
dimensional tolerances may be as important as the metallurgical
considerations involved in the often dissimilar metal weld
combinations.
T/TS joint systems may require cladding (16) of the
tubesheet for corrosion or metallurgical compatibility reasons.
As a variety of welding processes may be used for the cladding
or T/TS welding, each process or combination of processes
introduces its specific metallurgical characteristics. (18)
Figure 1 shows the weld joints to be discussed in this
paper. They are basically two types. The first six are welds
made to the primary or face side of the tubesheet and the last
three are secondary or back of the tubesheet welds. The first
six contain geometric root cracks and tube hole crevices.
SELECTION OF T/TS WELD JOINT GEOMETRY
The optimum weld joint is that joint which meets all
governing specification requirements and engineering requirements
and is capable of being reliably and reproductively made to
meet soundness requirements. It should meet these requirements
at the lowest cost. The optimum joint is dependent on the
following items:
 536

TYPICAL TUBE WELD JOINT PREPARATION

SHOWING FINISHED WELD ON THE RIGHT SIDE

Flush Tube Weld Flush Tube Weld

Cladding

Tube Wall ' Tubesheel

Ligament
Trepanned Tube Trepanned Tube
Sheet Fillet Weld Sheet Fillet Weld Added-Ring Weld

B2

Fillet Weld Fillet Weld Recessed Tube Weld Recessed Tube Weld

D2
-1

Explosive Tube Weld Explosive Tube Weld Tube To Boss Weld Tuhe To Boss Weld

F
2. G2I

>O Born BacfcsW*

i Ptu«
UU r
* \7/Fi7; "**

j O ous Or
I Filler Weiil U M ) Fillet Me
Filki W.UJ Fillet WekJ

Figure 1
 537

TABLF 1

TUBE ICIDIW

Weld geo>retry Tjrpc of welding Tube M t e r i a l s Tube sheet u t e r i a l s

1. SKA Cu-RI Carbon Steel (C.S.) FVH

2s A6TAV TitantL* Tttanlua
»!
3!
4. *1 AETAV
ASTAV with
F i l l e r Metal
Stainless
70-30 Cu-an
Stainless
70-30 Cu-m
Condenser
Brine Healer

5. Bj AETAV Snconel tnconel Clad C.S. Mucl. S U . t e n .

Sodlua Sta. Gen. Model
6. »2 AETAV Stainless Stainlns Clad C.S. • u c l . S U . Cen.
7. »1 MSTAW with Altainua Al ua1n4aa A1r Coaled Cond.
Filler
B. C AETAV Cu-Kt C.S. with Rtctel «1ng FM4
Manual GTAtf Cu-Rl C.S. with Ktckel Ding FVH
9. C AETAV Carbon Steel C.S. with Alloy Steel •xicl. S U . Sen. Model
King
10. D- AC7AV with S u i n l e s s Steel Stainless Clad C.S. « u c l . S U . Gen.
F i l l e r Metal
11. r>2 A£TAU H/0 Stainless Steel Stainless G a d C.S. « u c l . S U . Gen.
F i l l e r Metal
12. D
2 AETAW with RI-Cr-Fe t i - C r - F e Clad C.S. ftjcl. S u . Gen.
F i l l e r Metal
13. D
2 MSTAV with 11-Cu arl-Cu Sxrb. Tube Evap.
F i l l e r Metal
14. 0j M5TAV with Stainless Stainless M u d . Slurry Can
F i l l e r Metal
IS. Ej AETAV AdairaUjr Silicon Bronze Condenser
16. E^ AST*. «1-Cr-Ft •ti-Cr-Fe FVH
Inconel Clad C.S.
17. Ej AETAU •H-Cr-Fe RI-Cr-Fe Clad C.S. Mucl. KI
ia. Ej A£TAV RI-Cv IIUc'l-Honel D a d C.S. FUN
19. E2 ASTAU 70-30 1 90-10 m u t l - M o n e l Clad C..S. Wild
Cv-RI
20. *2 AETAV Stainless Inconel Clad C.S. FkH
21. r2 ASTAV Carbon Steel IncDnel Clad C.S. FVH
22. Eiplos1>e Adatralty Silicon Bronze ConiS. (Dev.)
23. ?2 Explosive Inconel Inconel Clad C.S. Nucl. S U . Gen. (Her.)
1 Tube Plugging
24. AfiTAlf «1-Cf-Fe IH-Fe-Cr Iitconel Clad C $ . U Stau Gen. Model
Inconel 1 Incolojr
25. Fj •ETAH Monet Honel Clad C.S. Sut. Tube Etap.
26. Aj AETAV with 3003 Alclad 1100 Alclad C-S Otec «
27. E2 ASTAV 4 » S.S. Inconel Clad C.S. KM
28. •CTAV (1) with 2-1/4 Cr-1 Mo 2-1/2 Cr- 1 Mo l a . Stm. Gen. (Dt».)
Filler Metal
M. >J AETAV with 2-1/4 Cr-1 No 2 - 1 / 2 Cr-Wo • a . S U . Gen. *rotatjrpe
Filler Metal
30. *2 AETAV Carbon Steel Carbon Steel Clad S U . Gen. (Exp.)
C.S.

(1) Mechanized Manual

"ETAW - Manual Gal Tungsten Arc abiding
A6TAW - Automatic h i Tungsten Arc
SIM - S h i e l d * ) ( t e t i l - Arc
 538

1. Tube and tubesheet (or tubesheet cladding), metallurgical

compatibility for fusion welding.
2. Service requirements, including consequences of
leakage.
3. Repairability requirements.
4. Number of tube welds to be made and availability of
skilled manual welders.
5. Tube weld accessibility and position.
6. Shop environmental conditions and previous shop
experience.
7. Post weld heat treatment requirements.
8. Inspection methods and acceptance criteria.
Each of the weld joints shown in Figure 1 has specific
characteristic advantages and disadvantages. An example of
how one manufacturer utilized these weld joints is shown in
Table 1.
Flush Tube Weld (Al and A2)
This weld joint is the simplest and least expensive joint
to prepare for welding. It is a joint better suited for
automatic welding than for manual welding, as the resultant
leak path (effective throat) cannot be determined or adequately
estimated visually. The joint can be welded autogenously
where the tube and tubesheet have adequate metallurgical
weldability or can be welded with filler-metal-added techniques.
There is no problem with fuse through or melt through (40) and
resultant metallurgical degradation of the tube. The geometry
provides for easy 360° overlap fusion. It is the easiest to
repair and, when welded with automatic equipment, requires the
least complex controls and devices.
The flush tube weld, when welded autogenously, is severely
limited in leak path size. When filler metal is added, this
is less of a problem. The joint is generally limited to
thinner wall tubes. When welded autogenously, this joint is
intolerant of eccentric tungsten rotation as this can produce
a localized area with a negligible leak path dimension. Where
tube weld "roll over" (or reduction of tube ID by the tube
weld) is unacceptable, this joint may present a welding problem.
 539

Dimensionally, this joint can present some problems. The

weld is situated at the "bell mouth" tube location in the tube
hole area and often requires a flaring tool to close this
void. The we7d is normally welded autogenously automatically
with center post welding devices which, if not closely controlled,
can make very eccentric welds.
Joint applications have been thin wall stainless and titanium
condensers, Inconel steam generators, evaporators, and other
relatively low pressure heat exchangers.
Trepanned Tube Sheet Fillet WeTd (B7 & B2)
This weld has been produced by two methods. It has been
welded with post weld trepanning or has the trepanning performed
prior to welding. When welded by the former method, the joint
has been selected because the trepan added joint flexibility
where there is concern for thermally-induced fatique failures.
When welded by the latter sequence, the joint is normally
selected because of heat sink considerations and/or flexibility.
In the cases of aluminum or copper, the heat sink of the
tubesheet may necessitate the selection of this joint. When
s-mall effective throat welds are required and when metallurgical ly
suitable, the joint can be welded autogenously. Where larger
throats are required, filler metal is added. Where the trepan
is relatively deep, the joint can be radioqraphically inspected.
A defect that is through the wall of the trepan is equal
to a T/TS weld leak. Slag or oxide entrapments in clad tubesheets
(which cannot easily be volumetrically inspected) can present
problems in the trepan. The use of the trepan, therefore, may
decrease total reliability. Detection of dye penetrant
indications in the trepan area after completion of the T/TS
weld may cause a serious repair welding problem. Machinina
the trepan with a milling cutter or fly cutter leaves; "pips"
between the trepans which present dye penetrant problems or
further machining problems, if they must be removed. Post
weld T/TS inspection is difficult with this joint design.
 540

Dimensional outages are especially bad with this joint

geometry because whether the weld is made before or after
trepanning, it must be uniform in leg lengths to either "fit"
on top of the trepan or to prevent a portion of the weld from
being removed by trepanning. If the tube protrudes very far
above the trepan, there is a potential problem with fuse
through or burn through, and for melting off the tube end.
Applications for this joint have been aluminum heat
exchangers and early Navy steam generators.
Ring Weld (C)
This weld joint appears, at first glance, to be an economic
solution to an expensive T/TS welding problem where a special
chemistry ring can be used as pre-placed filler metal for a
dissimilar tube-tubesheet weld joint combination. It can be
used where it has been determined that cladding the tubesheet
is too expensive or not considered to be necessary. The joint
can be welded manually or automatically. Special chemistry
(other than standard filler metal chemistry) rings can be
developed.
There are a considerable number of welding problems
associated with this joint. The rinq presents additional
surfaces which must be kept clean to prevent porosity. Also,
additional surfaces present more oxides that must be reduced
or slagged out. It is difficult to properly direct the arc to
produce complete fusion between the ring and the tubesheet
face at the root of the fillet joint. When welding starts,
the ring expands and moves away from the tube, making welding
more difficult.
The ring weld joint presents considerable difficulties
even with perfect dimensional controls, but cumulative (stacked)
tolerances, lack of fusion problems, hidden porosity and
localized areas with small effective throats often result in
an unreliable joint.
This joint was utilized relatively unsuccessfully for
early copper-nickel tubed high pressure feedwater heater (HP
FWH) applications.
 541

Face Side Fillet Weld (Dl and D2)

This geometry offers the maximum flexibility for producing
various T/TS weld effective throat sizes. Single or multiple
pass fillets can be made. The joint can include a partial
penetration joint bevel at the tube hole to further increase
effective throat size. When material compatibility characteristics
permit, the root pass in a two step multi-pass weld can be
made autogenousiy which permits dye penetrant and/or visual
inspection prior to application of the second pass. This is
an excellent welding method as the sound deposition of the
root pass greatly increases the potential for the sound second
pass (and completed weld). In some sizes, combinations of
materials and shielding gas selections, the entire fillet weld
can be made autogenousiy.
The fillet weld geometry can produce fuse through, burn
through or complete melting of the tube edge when the arc runs
eccentrically around the joint (with varying arc length).
This is one of the most commonly used T/TS joints because of
its capability of producing various weld sizes. It is, however,
most sensitive to dimensional variations in fit up and welding
device dimensional outages.
This joint is normally welded with cold wire feed added, a
factor which introduces another spatial relationship problem.
The filler is often fed through a bent snorkel. The snorkel
must have three dimensional adjustment and, when set, must
maintain its position in the hot welding arc area. The
filler metal must have a controlled cast (not variable within
the tolerance range), controlled hardness and essentially zero
helix, if the wire feed spatial relationship is to be maintained.
Even arc voltage controls are difficult with the electrode in
the fillet position as the arc will always find the easiest
low resistance path and may wander to the wrong location.
One of the disadvantages of the fillet weld is that the
protruding tube creates a greater inlet end vena contracta
velocity and potential erosion-corrosion problem than the
rounded entrance for flush or recessed geometry T/TS welds.
 542

Equipment used to make this weld varies from relatively

crude fixed non-expandable center post devices with poor
control of the filler metal addition to extremely sophisticated
devices. One such sophisticated device, for example, has a
set-up flip-flop indexing pin, tubesheet pattern air lock
expanding mandrels, water-cooled torch body, double floating
electrode-locating fingers, AVC electrode arc length control,
tachometer feedback filler metal control, and can make
continuous multi-pass welds.
Applications for this joint include HP feedwater heaters,
many types of nuclear heat exchangers and petrochemical heat
exchangers.
Recessed Tube Weld (El and E2)
This weld joint (3, 6, 11) can accommodate thin wall
tubes similar to the flush weld. There is less chance for
excessive "roll over" with this weld than the flush weld. The
weld joint is inexpensive to prepare and very easy to repair.
There is no chance for burn through, fuse through, or burning
off the tube end as in the fillet weld. The major advantage
is that it is metallurgical^ simple for autgenous welding. A
technique can be employed where the face of the tubesheet is
melted down over the top of the recessed tube with minimal
melting and dilution of the tube material in the tube weld.
This technique is used, for example, for welding Admiralty
Metal (30% Zn - brass) tubes to silicon bronze tubesheets
(12). Another advantage is that the weld produces a rounded
tube entrance which minimizes inlet end erosion-corrosion
problems. It is easier to visually determine if the proper
size weld is made than it is with the flush tube geometry.
The joint can be welded with or without filler metal, but is
normally made without filler metal. Overlap refusion is easy
with this joint.

The major disadvantage of the recessed tube joint is its

size limitation which can be estimated as shown on Figure 2
for autogenous welds. The cost of the preparation for the
recessed joint is low, but more than the flush joint.
 543

FIRST APPROXIMATION

Filler Metal Available

Tube Weld "Throat"

Clad
1
ft\\\\\N

Tube
T
Dilution From Tube

figure 2.
 544

The major application is for feedwater heaters, condensers,

and petrochemical heat exchangers.
Explosive Tube Weld (Fl and F2)
The explosive weld joint (13, 14) requires a beveled hole
stand off to produce a weld. (It is normally produced in T/TS
pluq welding by a stand off angle on the end of the thimble
plug). The almost complete lack of melting makes possible the
welding of ductile tubes to tubesheets that are unweldable by
fusion welding techniques. It is possible to develop a long
leak path with this process. Although cleaning must be employed
for good welds, the cleaning is much less than required for
fusion welds which can suffer contamination porosity problems.
The process can combine welding and post weld expansion. It
can be used in limited access locations and requires only a
short setup and welding time.
The explosive tube weld obviously requires special safety
requirements and licenses (specific to each State) for handling
explosives. Only a tube-in-hole configuration can be welded,
as a protruding tube will be blown off. The ligament thickness
must provide sufficient stiffness to resist plastic deformation
from the explosive force. There is a difficulty in determining
by MDE methods the difference between a tight expansion and a
metal!urqically-welded joint. The residue from the explosives
and plaLtic holders require removal. Often the welding must
be done in special locations due to the noise factor.
The dimensional requirements are less than required for
most fusion T/TS welding applications. The applications for
this weld are normally heat exchangers with material combinations
not compatible for fusion weldinq, e.g., titanium tubed HX
with stainless tubesheets.
 545

Tube to Boss Weld (Gl and R2)

The tube to boss weld is normally a back side of the
tubesheet joint. It can produce 100% weld joint efficiency
with a geometry that produces no crevice, has no stress
concentration, no designed-in root-crack. The finished weld
can be volumetrically inspected. Shorter tubes can be used
with smaller tube holes and thinner tubesheets than must be
used for face type welds. If post weld heat treatment is
required, it can be accomplished with local PWHT methods.
There are no inlet end erosion problems. The added flexibility
of the joint (which is a function of boss length) permits
greater thermal cycling gradients than face side tube welds.
This weld is normally made autogenously with an ID (bore)
welding device. This method produces the maximum heat input
on the small ID diameter, but rotation cannot proceed until
melting through to the larger OD diameter is achieved. If it
is possible, this joint should be welded from the OD. Welding
from the ID produces excessive heat and large molten pool on
the ID, and it is often difficult to control the weld contour
to produce a uniformly-shaped 360° weld. It is extremely
difficult to add filler metal on the ID. If no filler metal
is added, there is no way, if desired, to adjust the chemistry
of the weld. If the arc is not located exactly at the joint,
a weld can be made without complete joint fusion. Welding is
conducted "in the blind" as the technique normally requires a
clamping and gas shielding fixture on the OD. The tube boss
in this design must also be completely sound since it also
represents a leak path. Penetrant testing of the ID is often
impossible.
The joint fitup is normally not a problem, but care must
be taken in machining not to chamfer the edges of the boss, as
it eliminates some of the weld metal. In a straight tube
design, the tubes must be cut short and lengthened normally by
localized thermal expansion to produce the butt tight conditions
for tube welding. One advantage to the design is that it can
be cleaned on a joint-by-joint basis.
 546

This joint is used for many special applications, including

heat exchangers for the breeder program.
Back Side of the Tube Sheet Socket Butt Weld (HI and H2)
This weld also has no inlet end erosion problems and can
use shorter tubes with small tube holes and thinner tubesheets
than face side welds. Two techniques are used to weld this
joint. One technique employs an ID bore autogenous process to
make the complete weld and depends on surface tension to
produce the OD contour. The other technique adds a wrap-around
filler-metal-added fillet on the OD after the ID bore weld.
Radiography is possible, but difficult.
The problems with this weld is that since you must weld
"in the blind," it is possible to miss the joint and the joint
is difficult to heat treat due to the heat sink mismatch.
When the tubesheet is not clad, there is no way to control the
chemistry of the ID weld. Visual, ID penetrant and radiographic
inspection are difficult.
The joint is often welded by inserting the tube into a
slight counter bore which makes the dimensional control of the
joint very good. The OD of the tube is usually tightly
controlled and the ID of the counter bore can be easily
controlled.
The applications for this joint include British (CEGB)
and a few domestic carbon steel feedwater heaters and some of
the British breeder heat exchangers.
Back Side Fillet Weld on Boss (II and 12)
This is a very specialized joint which has been discussed
in the literature (1, 4, 5, 10). It has the worst cumulative
(stacked) tolerance problems of any of the T/TS joints and
requires the most sophisticated equipment to weld the joint.
 547

FABRICATION DIMENSIONAL CONSIDERATIONS

The ability to produce reliable reproducible sound T/TS
welds with constant control of the weld contour, weld
penetration pattern, and effective throat (minimum leak path)
is directly related to the control of the optimum spatial
relationship between the filler metal, electrode end base
metals through the complete weld cycle. In the case of manual
welding, the manipulative skills and experience of the weldor
can theoretically compensate for dimensional variations
encountered. In automatic welding, the total weldinq system
must consistently control the proper arc lenqth, electrode
attitude and impingement point during the 360° rotation and
overlap portion of the weld. In continuous multi-pass T/TS
welds, the problems are greatly compounded. Unfortunately for
automatic weldinq, there are an extremely large number of
dimensional tolerances which, when stacked together, may
present difficult control problems. A list of the dimensions
of interest and an estimate of typical manufacturing tolerances
(where known) is shown in Table 2. The purpose of this table
is to indicate the large number of dimensions that may affect
the welding process for automatic welding. The actual values
may vary depending on how the materials and devices are specified
in purchase documents and how well fabrication operations can
be controlled. The important point concerning tolerances is
that they can have an additive effect.

There are many cases where all that is required of the

T/TS weld is that it provide a seal with no special requirements
for strength, thermal cycling, erosion or severe corrosion
resistance. In these cases, if the T/TS weld is eccentric
with very little effective throat at one point, it may be of
no concern. The opposite case may be true for critical T/TS
welds, as in the breeder program.
 548

TABLE 2
T/TS WELDING DIMENSIONAL CONSIDERATIONS
(TYPICAL VALUES GIVEN WHERE KNOWN)
1. Tube Tolerances
(A - 3/4" OD x 12 BWG Carbon Steel)
(B - 1" OD x 18 BWG Stainless Steel)
a. Wall thickness A: +0.022
-0.00
B: +0.010
-0.00
b. ID/OD eccentricity A: + 0.015
B: + 0.020
c. Diameter + 0.004"
2. Tube Hole Tolerances
a. True location and relative location +_ 0.014"
b. Diameter + 0.0035"
c. Roundness +_ 0.0003"
d. Perpendicularity (run out) - 0.0015 in/in
e. Bell mouthinq + 0.020
f. Surface rouqhness 63-125 AA
3. Welding Device Tolerances
T. Rotation (bearing slop and gear lash) ?
b. Electrode location and extension settinq- +_ 0.03"
c. Filler metal snorkel tube position +_ 0.050"
d. Electrode collet centering +_ 0.020"
e. Electrode grind geometry +_ 10%
f. Filler wire cast, helix, and hardness Note 1
g. Filler metal diameter Note 2
h. Tube ID centering Note 3
i. Tube 0D centering Note 3
j. Tube pattern centering Note 3
(1) adjacent hole
(2) numerical tape
k. Boss 0D centering - Note 3
 549

TABLE 2 (Continued)
4. Other Factors Affecting Rotational Tolerances
a. Preweld expansion tolerances
1. Percent wall reduction
2. Length of roll
3. Location or roll and contour
b. Tube cuttinq tolerances (recessed T/TS weld)
1. Depth
2. Parallelism with tubesheet face
5. Boss Tolerances
a. Height + 0.010"
b. ID or OD + 0.003"
c. Wall thickness + 0.010"
d. Eccentricity - effected by b and c

Note 1. Can cause wire wander to several wire diameters.

Note 2. Effects wander of filler metal protruding from snorkel
tube
Note 3. Determined by individual manufacturer's design.
 550

The tolerances for purchased tubing often are percentage

tolerances which can result in large absolute numbers for
heavy wall tubes. For example, if a weldinq device with an ID.
centering post is used for 3/4" OD x 12 BWG (0.108") carbon
steel tube, the combination of eccentricity and tube ID variations
can produce an electrode rotation that greatly chanqes the arc
length with respect to the OD. This can occur even if the ID
of the tube holes are identical, which is often not the case.
The tubing tolerances must be added to the tube hole tolerances
and to tube pre-weld expansion tolerances.
Tube hole location tolerances may have a minor effect on
T/TS welds made with a fixed ID centering device, but have a
pronounced effect on a tube welding device that is a fixed
tube pattern location system. Unfortunately, there are some
T/TS welds which are made on tubesheets that are drilled from
the opposite side of thick tubesheets (e.g., 20" tubesheets).
In these cases, the hole perpendicularity (or drill run out)
can be considerable (up to 0.040"). This effects the tube
pattern, the distance between tubes and the perpendicularity
with respect to the tubesheet face welded.
"Bell mouthing" is the increase in tube hole diameter at
the drilling face of the tubesheet caused by initial drill
wobble. This can affect the contact of the tube in the hole
at the tubesheet face and create a small crevice at this
point. In the case of thin wall tubes with fillet type welds,
this lack of tube contact at a critical point of the weld
joint can produce a poor localized heat sink. The results of
this can be burn through and/or increased porosity problems.
In some cases, hydraulic or urethane expansion methods are
used to deform the tube to the bell mouth contour to restore
the heat sink.
The effect of out-of-round drilled holes causes problems
for all devices that do not automatically follow the tube 0D.
As previously discussed, there are a wide variety of + ube
weldinq devices with different types of fixed and floating
tube joint followers. The significant difference in design
 551

is the method used to track the tube joint. Even with the
tube OD centering device, there are various methods and position
points used to center the welding head. Unfortunately, most
of the ID centering devices average out the tolerances (including
pre-weld expansion) based on a variety of centering rings or
splined expandable sleeves.
The tube welding device itself has a number of parts that
affect the spatial relationship with the weld joint. Bearing
slop (radial alignment) affects the centering of the device.
The gear lash affects the smoothness of rotation. The filler
metal snorkel design, materials and adjustment method affect
the relationship of the arc to the filler. The filler metal
diameter, cast, and helix affect the ability of the filler
metal to stay in the conical envelope necessary for proper
melting. The electrode diameter, methods used for grinding
the electrode, and electrode collett centering device affect
the arc location.
In the case of flush and recessed tube welds, the device
used for cutting the tube end and its ability to remain in
axial alignment with the tube affect the tolerances of the
machining operation. This results in variable dilution and
non-uniform effective throat size.
Preweld expansion methods and equipment affect the contour
of the tube ID and, consequently, affect the "fit" of a fixed
ID centering device location.
Visual inspection of a T/TS weld can provide a considerable
amount of information on the ability of the process to control
weld contour and the effective throat size in the face of
dimensional tolerance stacking problems. In the flush or
recessed T/TS weld, the eccentricity of the electrode rotation
can be easily observed (width of the weld bead). The consistency
of ID rollover weld bead shape can also be easily observed.
The presence of a "square corner" at one point (narrow bead
width location) is visual evidence that the effective throat
at this point is minimal. Variable leg length of fillet welds
also shows that somethinq (it could be a variety of
parameters) is out of control.
 552

SPECIALIZED WELDING EQUIPMENT

A wide variety of specialized equipment is used for
T/TS welding. In the early days of T/TS welding (in the
1950's and I960 1 s), the limited availability of'sophisticated
welding equipment required many fabricators to design and
build their own welding devices. Currently, a wide variety
of specialized programmable power sources and welding devices
are available. These include pulsing, synchronized travel/filler
addition/ pulsing, timer-based and position-based pre-set
weld schedule changes, and full microprocessor control
programmable power sources.
Current technology also makes robotic techniques
available for T/TS welding. Using robotics, the indexing of
the welding device can be determined by the numerical tape
used for drilling the tubesheet or by a locked indexing
position set in during the pre-weld i;xpansion operation.
The robotics technique has been developed for repair welding
or re-tubing of nuclear steam generators to provide less
radiation exposure for ALARA purposes.
CLEANLINESS
GTA T/TS welding combines the disadvantages of the
slagless GTA molten pool (with no inherent slag/molten metal
cleaning reactions) with built-in crevices (15). Cleaning
problems start with the contamination of the tubesheet with
the deep hole drilling lubricants. The tubesheet should be
cleaned immediately after drilling by vapor degreasing or
steam-detergent methods. The tubesheet must be placed in a
position such that during cleaning the contaminants will
leave the tubesheet and not "bead up" on one surface. The
tube support plates must be similarly cleaned. After cleaning,
these items must be either immediately dried and used, or
protected from typical fabrication floor airborne contamination
and rusting. When temperature-humidity clean room environments
are impractical, the tubesheet and tube support plate
structure should be heated (from initial fabrication through
 553

completion of tube welding and leak testing) to a temperature

slightly above the maximum expected ambient temperature to
prevent thermal cycling atmospheric condensation. When
weldinq is not conducted in a clean room environment, the
entire structure should be protected by a "tent" to prevent
accidental contamination, such as "droppings" from an overhead
crane or airborne oil or other hydrocarbons from sources
such as nearby machining operations. In "U" tube heat
exchangers, it is not often possible to develop an individual
tube-hole solvent cleaninq-tubing-welding sequence that does
not contaminate untubed holes. The tubes are normally
ordered to special cleanliness requirement, then individually
re-cleaned for a distance at least equal to the tubesheet
thickness, often by mechanical polishing followed by wipinq
with a clean cloth dampened in a solvent..
The easiest thinq in the world is to write, in a specification
for T/TS welding, a staterent such as "Remove all qrease,
oil, moisture, crayon marks, soap, grinding dust,, and other
foreign material that can produce porosity from the area to
be welded". The toughest thing is to achieve the cleanliness
level required under shop conditions. In T/TS weldinq, the
deck is stacked against the fabricator. Very highly volatile
solvents may dissolve contaminants effectively, but tend to
concentrate in "beads" and produce 350° of cleaned hole and
10° of highly contaminated hole (tubeshpet ax^s horizontal),
unless they are blown ary. The exhaust from air-operated
tools also can contaminate T/TS joints and must be modified
to prevent exhaust ports from blowing on the weld ?rea.
Normal compressed air should not be used for cleaning or
cutting tools unless filters are installed to completely
remove the normal moisture and oil contamination of the air.
Dry nitrogen is often used to replace compressed air.

Hydraulic expansion is academically superior to roller

expansion in many cases, but leakage from this method can
cause contamination.
 554

An example of typical spiralling T/TS weld porosity

illustrated by successive metal!ographic sectioning is shown
in Figure 3. The biggest problem with porosity in T/TS welds
is that it can be quite large and still be subsurface, but
service requirements can fracture the "skin" and cause a leak.
EFFECT OF GEOMETRY ON FATIGUE PERFORMANCE
Successful T/TS welds are those that meet service
requirements for the entire service life of the heat exchanger.
The ASME appears to have minimal interest in the mechanical
properties of a T/TS weld (as determined from the requirements
for the procedure qualification) unless the T/TS weld acts to
structurally support the tubesheet (16). Nevertheless, the
T/TS must, be capable of surviving thermal cycles which may
exercise the foot of the weld in fatigue, as shown in Figure
4. (11)
The face side of T/TS weld joint geometries shown in this
paper contain (within the design) a geometric root crack. In
order to mechanically evaluate T/TS weld design, it is necessary
to determine a realistic value for stress concentration at the
root of the weld, since maximum bending stress occurs at this
location. Theoretical approaches to the effect of the root
crack could result in an infinitely large stress concentration
factor. Experimental analyses have been conducted on some
specific T/TS metal combinations to demonstrate that the
effective stress (or strain) concentration takes into account
the cumulative effects of plastic flow, redistribution of
stress, weld geometry, size effects and material properties.
The results of these tests were the development of a rational
fatigue strength reduction factor which was determined to be
between 1.5 and 2.0 for carbon steel HP FWH tube welds made on
Inconel clad tubesheets.
 555

n
n

—Transverse consecutive sections showing extent of

tube weld porosity. Cyclic test model—carbon steel clad,
carbon steel tube sheet—gas tungsten-arc fusion tube weld
porosity. Approx. cycles to failure: A—110: B—230; approx.
defect length. In.: A—0.032; B—0.091. Unetched photo-
graph)

FIGURE 3
 556

INTERACTION BETWEEN TUBE

AND TUBE SHEET
FROM THERMAL AND PRESSURE LOADS

• Weld Root Notch

Clad Light
Pre-Weld Expansion
Fully
Ligament Expanded
Length
Recessed-Tube-To-Tube Sheet
Welded and Expanded Joint
Tube Flexure At
( \ Weld Root Notch

Contraction of
Tube Relative To
Tube Sheet Perforation

I Free Relative
Deformation

Axial Pressure
Load

Expansion of Tube
Relative To Tube Sheet
Perforation

Free Relative
Deformation

Figure*-
 557

WELD JOINT CREVICE

It is not unusual to have tubes and tubesheets of dissimilar
metals. Weld clad carbon and low alloy steel tubesheets are
most common (10). Most T/TS weld joints are on the face side
of the tubesheet which geometrically produces a crevice (annular
space between the tube and tube hole ID). Normal fabrication
tolerances produce a crevice annul us of 4 to 6 mils. The type
of heat exchanger, heat flux direction, thermal-hydraulic
conditions and shell side water chemistry determine the potential
for chemical hideout in the crevice area. If the heat exchanqer
contains the hotter fluid on the ID of the tubes, there is
potential for chemical concentration in the crevice area.
Where there is boiling on the OD of the tube, chemical
concentration will not occur as long as the areas where the
bubbles form is immediately re-flooded with the bulk chemistry
fluid. In the crevice area, the geometry inhibits re-flooding
and, therefore, promotes chemical hideout.
Where the heat flux is in the opposite direction (cooler
fluid on ID), there is little or no potential for chemical
hideout. In cases without chemical hideout where the tube is
noble, the carbon steel tubesheet may essentially provide
galvanic protection to the tube.
Where chemical hideout is present in the crevice, it may
result in adverse corrosion effects on the tube, e.g., stress
corrosion cracking, dissolution, pitting, or intergranular
attack. The crevice concentration problem can be minimized or
eliminated by post weld expansion by mechanical, hydraulic or
explosive methods. Careful evaluation of crevice-closing
techniques must be made to assure that this process does not
produce unacceptable residual stress in the tube due to "over
rolling" past the tubesheet. Where total elimination of the
crevice is required, the back side of the tubesheet weld
joints must be employed.
 558

RDT STANDARDS FOR T/TS WELDING

The RDT Standards of the DOE Division of Nuclear Power
Development are rarely specified as being applicable for T/TS
welding, except for heat exchangers for the breeder program
and other specialized next-generation nuclear equipment. The
RDT Standards are normally considered by fabricators of
standardized HX applications to be severe overkill. The
philosophical approach indicated in the RDT Standards is
correct. These standards recognize T/TS welds as being legitimate
welds (not just seal welds) that should be tested in all of
the ways possible and applicable to their intended service.
RDT Standard E15-2NB-T supplements the requirements for ASME
Section III, NB and invokes all of the requirements of NB. It
requires that T/TS welding be qualified in accordance with RDT
F6-5T which supplements ASME Section IX. (Part 73 provides
additional requirements to SCIX). It limits production welding
to the same position as used for procedure and performance
qualifications. It requires visual, liquid penetrant and
radiographic examination of qualification test assemblies.
For tubesheet face joints, radiographic examination is required
to be conducted on H inch thick slabs from the test assemblies.
This requirement for a radiographic test is an excellent
development tool and a good procedural requirement for statistical
confidence in the procedure, but only if the realistic shop
floor cleaning procedures and difficulties are precisely
duplicated.
When sensitive production leak testing is a requirement
of fabrications, this test is required of the test assemblies.
 559

Tension tests and bends tests are required for some T/TS
welds in the test assembly. If the weld joint geometry is
incompatible (face joints) with mechanical test specimen
geometries, a butt weld shall be made to demonstrate the
effect of welding on mechanical properties, including ultimate
strength and bend ductility.
RDT F6-5T also adds a number of additional parameter
requirements and controls to the standard SCIX WPS essential
and non-essential variables. The number of welds required in
the test assembly is greatly increased. RDT F6-5T Part TO
requires four test assemblies, each incorporating a minimum of
12 T/TS welds made consecutively for procedure qualification,
regardless of the production inspection requirements. Where
volumetric inspection is not possible or feasible, the 48
consecutive "perfect" welds represents a better statistical
evaluation of the procedure than the Standard Section IIT
requirement for 10 welds. In the cases where each individual
weld during HX fabrication will be accepted by 100% volumetric,
surface, and leak testing inspections, the requirement for 48
consecutive "perfect" procedural T/TS welds is an unnecessary
requirement. Repair welding requires separate qualification
for complete replacement and partial repairs.
The author does not imply that all T/TS welding should be
qualified in accordance with RDT Standards, but that these
standards should be reviewed and thoroughly understood by
personnel conducting procedure and performance qualification
tests and portions of the standard used judiciously for
"information purposes," as the document represents a thorough
engineering understanding of T/TS weld requirements.
The requirements that exceed Section IX for documentation
of the WPS and extension of the essential variables for
requalification represent good quality control and engineering
practice which should be given careful consideration.
 560

T/TS WELDING SYSTEMS EMPLOYING THIRD FLUID LEAK

DETECTION SYSTEMS
Developmental and prototypic double-wall-tubed heat
exchangers have employed (or are employing) third fluid leak
detection as a precursor to provide warning of impending
problems caused by the leakage of the primary fluid to the
secondary fluid. In the case of the sodium steam generator,
this system is designed to prevent the sodium to water reaction
by monitoring contamination in a helium third fluid. This has
previously been discussed in the literature (4, 5). The third
fluid leak detection systems introduce additional concerns in
the design and testing of T/TS welds. Obviously, the T/TS
welds must be helium leak tight. In straight double-wall tube
units, the third fluid system design may require wrap-around
fillet welds. With the advent of rod anode radiography, every
production weld can be volumetrically inspected; however, in
the case of fillet welds, the interpretation may be difficult.
Some additional design concepts for more easily repairable
third fluid leak detection system T/TS welds is indicated in
U.S. Patent 4,246,958. These methods permit localization of
leaks and repair from the face side of the tubesheet.
PREWELD AND POSTWELD EXPANSION
Tube-to-tube hole expansion operations serve multiple
functions. There are certain manufacturers that purposely
produce face type T/TS welds with the tube loose in the tube
hole. Their theory is that the loose tube joint permits
contamination gassing porosity to be vented in both directions.
Other manufacturers believe that, as the tube weld is being
formed, the pre-weld expansion is broken and, therefore, the
pre-weld expansion does not increase porosity problems. In
the case of the flush and recessed tube welds, it is often
required to roll the tube in the tube hole to hold the tube in
place for mechanically cutting the tube to the dimensions
required. It is also desirable to have the tube in contact
with the hole so that the tubesheet can act as a heat sink.
Some manufacturers use a flaring tool to "bell" the tube in
 561

the hole rather than to roll or hydraulica'ly expand the tube

in the hole. Typical pre-weld expansion is 2-3?£ wall or area
reduction for 3/8-%" length starting at the tubesheet face.
If the tube weld is to be adequately leak tested, it is required
that the roll should be broken by the welding operation so
that it doesn't mask the testing of the tube weld. This can
be demonstrated by stopping welding just before the overlap or
drilling a hole through the weld after welding.
In the case of fillet type welds with the tube protrudinq
past the face of the tubesheet, there is often a gap or crevice
caused by the "bell mouthing" of tube hole at the face of the
tubesheet. This crevice can create welding problems and can
be closed with "urethane" hydraulic jack type expansion.
Post weld expansion is often required to provide a damping
action to minimize mechanically- or thermally-induced fatigue
loading on face type tube welds. If there is no reason to
completely close the crevice, this expansion is nominally done
for ]h to 2" expansion length with a 6-10% wall or area reduction,
depending on the materials expanded. In the case of a T/TS
weld leak, the post weld tube expansion acts to choke flow
from the primary to secondary side and, therefore, minimizes
erosion by metering leakage. The longer the axial length of
the tube to tube hole expansion, the bettpr the "belt and
suspenders" effect. This expansion should not be utilized to
permit a lower quality level T/TS weld.

Methods used for short length post weld expansion are

roller expansion or hydraulic expansion. When the entire
crevice is to be closed, the hydraulic expansion or explosive
expansion techniques are normally employed.
In any case, there is a complex residual stress left in
the tube at the rolled-to-unrolled transition, which may cause
SCC problems. This appears to be more severe for roller
expansion than for hydraulic or explosive expansion, as the
latter methods normally taper out the transition better than
the roller expansion method. Hydraulic and explosive expansion
introduce much lower levels of cold work into the tube than
roller expansion.
 562

Preweld roller expansion is normally done without a

lubricant to minimize tube weld porosity problems. The roller
is normally cooled in alcohol between rolling operations.
LEAK TESTING
A T/TS weld has limited value if it leaks fluid while in
service. Traditionally, the ASME has considered the hydrostatic
test as satisfactory for leak detection. As the shell side of
heat exchangers is most often limited in design to a relatively
modest pressure, it is traditional to limit T/TS weld hydrotest
to 1.25 or 1.50 times the design pressure. This is often
nowhere near the primary to secondary pressure drop (across
the tubesheet).
Leak tests using dry air are much more sensitive than
hydrostatic tests. These tests include (in increasing sensitivity)
the soap bubble (or equivalent) test, "Leak Tec" (or equivalent),
air-under-water test with bundle in horizontal position,
air-under-water with bundle in vertical position and air-under-
water with bundle in vertical position with a glass collector
plate. The air-under-water test with a bubble collector
plate, run for about 8 hours, may be as much as 2 orders of
magnitude more sensitive than the soap bubble test.
The next type of leak testing is the halogen test. This
test utilizes a halogen tracer gas detected with an ion sniffer
probe. It requires a detailed controlled procedure and a very
high level of cleanliness to prevent outgassing of hydrocarbon
contaminants, producing a false sign.
The highest level of sensitivity is the Helium leak test.
This test uses Helium as a tracer gas detected with a mass
spectrometer sniffer probe. This requires the highest level
of procedure control and cleanliness.
It is important to remember that all of the gas tests
must be run with the tube-to-tube hold crevice completely dry.
 563

If there is water or condensation in the crevice, the gas test

will be in effect a gas pressurized hydrostatic test. Some HX
manufacturers keep the tubesheet heated slightly above maximum
ambient temperature from the time tubing commences until final
gas tests are finished.
The approximate relative sensitivity of the various test
methods is as follows:
Minimum Detectable Leak
Method Std cm-ysec
Air bubble test 10- 1 to 10- 2
Collector glass air under , r
water test 10-;? to 10-?
Halogen leak test 10-^b to 10-°
Helium leak test 10- to 10- 9
The actual leak rate detection levels are a function of
exactly how the procedures are conducted.
ASME Section V, Article 10 describes methods and requirements
for the performance of leak testing and should be reviewed for
engineering information. ASTM:E432 is written as a guide to
Selection of a Leak Testing Method. ASTM.-E515, E427, and E499
cover Bubble Emission, Halogen and Helium Leaking respectively.
The ASM Handbook Volume 11 (Pages 260 to 270) is the best
reference on the comparison of various leak testing methods.
It must be remembered that passing a leak test only
guarantees that the T/TS weld does not have a leak greater
than the detection limit.
Leak testing is similar to welding in that, to get good
results, all specific details of the test must be followed.
In the author's opinion, the best leak test for most T/TS
welds is the air-under-water test with a collector plate.
 564

This test can be accomplished with great sensitivity in many

shops and does not pose the cleanliness and background noise
problems associated with the halogen or Helium tests. It is
difficult to utilize for large units as it requires that the
unit be placed in the vertical position.
ONE FABRICATOR'S HISTORY OF T/TS WELDING
An example of one fabrictor's history of T/TS welding
(Shown in Figures 5 and 6) indicates the relationship between
increasing technical requirements, development of tubesheet
cladding techniques, and development and use of more sophisticated
welding devices. The first development was high deposition
series submerged arc stainless cladding for tubesheets. This
was used for a two pass (first pass autogenous) GTA T/TS weld
with an X-Y gantry heat-sink center-post welding device for
Navy steam generators. As automatic cladding techniques were
not developed by the fabricator for high nickel alloy cladding
in 1957-8, manual SMAW techniques were used for HP feedwater
neater welds (with poor service results). In an effort for
automatic welding, preplaced nickel rings were employed for
the feedwater heater (FWH) T/TS joints (with only fair service
results). In 1960, when the Navy changed to Inconel steam
generators, new filler metals and the automatic oscillated
GMAW process were used for tubesheet cladding and were applied
to Inconel T/TS welding using the same equipment as was used
for stainless T/TS welds. This technique was quickly utilized
to deposit nickel and Monel for feedwater heater tubesheet
cladding (16), therefore making it possible for recessed
autogenous Monel and Cu-Ni FWH T/TS welds.

By utilizing the deeper penetrating characteristics of

helium-argon shielding gasses and the high surface tension of
the austenitic stainless steel weld metal, an autogenous
stainless steel T/TS weld was made for commercial nuclear
steam generators. When interest increased in carbon steel HP
FWH's in 1964, a technique was developed for lowering the
carbon content of the tubesheet by cladding with very low
carbon SAW techniques for autogenous T/TS welds. When it was
determined that increased metallurgical characteristics could
 565

ONE FABRICATOR'S HISTORY OF

TUBE-TO-TUBESHEET WELDS

s • mini
Sic •1 (S-S.)

c . son
Sle tl (C.S.I

ClaCCing (309-31? F.nl L i r t i .

30BL Seconfl Layer)

Ring to GMAW Cl-Sdmg on Cuban

|fWM)
^•Cpitcaiion

Monet

Inconel

- AUIOT*JITC GTAW [Wilh Aulog«nouS GTAW

Aulomjuc Filler AcjanionJ TuDe Jo.nt (f *VM)
T u M Wctd tor N»«y Nuclt>f
Slcam Genrraion

EL.C.
C.S

lncon<»|
—^
— C.S
1963 — Automai-c Auloge-nouS GTAW c Lo«Cii6On
Wiin Recessed Tub* Jomt tof eO Arc CUCCog Wilfl W.lh B r t f 5*ea T
A FWH T«&*retJ Tut>f on
t i r f GWAW c
Larer SAW Ci'd

Figure 5
 566

ONE FABRICATOR'S HISTORY OF

TUBE-TO-TUBESHEET WELDS

CS

1965 — Automatic Autogenous GTAW 1967 — Aiiiom*!<c Auiogenou* 10 Boff 1968 — Automatic Autogenous GTAW
Desalting Application Weffl lo> Nil Sim Gen Test Uoo With Recessed Tube Joint 1or
Slamiesi Sleei FWH on OjcHUtrd
3 L*re« CMAW mconel Oeposrt
o< Osciiuted ? L» r e f SAW
Slaintess 309L-30BL Depoxt

J-Cf
1 Mo

I
1970 — Automatic Auiogenous GTAW 1977 — Automatic AulogrnDuS GTAW 19S0 — Automatic Filler Wrap Arou"0
Flush Tub* Jomi lor Thin Will With Flush Type Joint tor Nuclear Finn Weld on Backuce of
Stainless SI eel Tub* io Wrought Steam Generators TuOesheet [Boss)
Stainless Sleet Tubesheeis or
Oscillated 309X-30SL Two Ljyer
SAW Clad Tuftesheeis

Figure 6
 567

be obtained with Inconel welds on carbon steel HP FWH's, the

GMAW cladding was used for autogenous T/TS welding. The
autogenous recessed T/TS welding process was used for titanium
desalting plant T/TS welds. In 1967 Inconel was clad on 316
stainless steel tubesheets and a simple bent tungsten constant
current autogenous GTAW ID bore process was used to join
Icoloy 800 tubes to the Inconel cladding. This was used for a
breeder reactor test model.
The success of the Inconel clad carbon steel FWH recessed
tube T/TS weld lead to the use of this cladding process for
stainless tubed FWH's, but later these FWH's were changed to
use stainless clad tubesheets for economic reasons. An economic
chanqe to Inconel steam generator T/TS welds lead to the use
of recessed tube joints and, eventually, flush T/TS welds.
The need for redundant T/TS welds for a double wall three
tubesheet straight tube 2% Cr - 1 Mo breeder steam generator
with third fluid leak detection required considerable
development (5). A wrap around back side of the tubesheet
multipass weld was made with pulsing techniques. It was
determined that arc remelted tubesheets would be more reliable
than clad tubesheets. Special extra low carbon filler metal
was employed.
SUMMARY
The engineering requirements for the T/TS welds for the
complete service life including startup, emergency shutdown,
off line layup and any unusual partial load conditions should
be made prior to selecting tube weld geometry, po?t weld
inspection and post weld expansion. After selecting the weld
joint, tube weld size, and material combinations, the need for
equipment should be determined. Wherever possible, the most
simplistic tube weld geometry and equipment commensurate with
the requirements should be selected. In the author's opinion,
this fs most often the autogenous automatic recessed T/TS
weld.
 568

Two areas most often not given adequate early consideration

are actual shop cleaning methods and cumulative dimensional
tolerances. The more complex welding joints normally have the
worst cumulative tolerance problems and require the more
sophisticated welding equipment.
The use of GTA pulsing techniques is often an advantage.
The use of synchronized pulsing-travel techniques must be
carefully evaluated for complex joints with filler-metal-added
techniques, as it may minimize the amount of filler metal that
can be provided and add unnecessarily to the complexity of the
operation,

ACKNOWLEDGEMENTS
The author wishes to acknowledge the technical assistance of
Messrs. Jay L. Kratz and Walter G. Loyal of the Westinghouse Electric
Corporation, E. Harold Gray of Region I of the USNRC, Joe Kiefer of
Raymond Fabricators and Dr. B. J. Charles in the preparation of
this paper.
 569

REFERENCES

1. S. D. Reynolds, Jr. et al, "Successful Welding of

Tube-to-Tubesheet Joints," 2nd Symposium on Shell
and Tube Heat Exchangers," Sept. 1981, Houston,
Texas (ASM).
2. A. Lohmeier and S. D. Reynolds, Jr.. Carbon Steel
Feedwater Heater Tube-to-Tubesheet Joints, ASME
65 WA/PWR-10.
3. S. D. Reynolds, Jr. et al, Development of Tube
Welding Techniques for Carbon Steel Feedwater
Heaters, Welding Journal, January 1967.
4. C. E. Sessions et al, Materials Development for
a Fast Breeder Reactor Steam Generator Concept,
ANS Steam Generator Symposium, St. Petersburg
Beach, FL., October, 1980.
5. J. H. Kiefer, "Welding Equipment and Process
Development for Inaccessible Tube-to-Tubesheet
Welding on a Sodium-Heated Steam Generator,"
International Conference on Welding Technology
for Energy Applications, May 1982 (AWS/ORNL/ASM).
6. S. D. Reynolds, Jr., Tuoe Welding for Conventional
and Nuclear Power Plant Heat Exchangers, ASME 69-
WA/HT-24.
7. S. Yokel!, "Hydraulic Methods in Construction and
Repair of Shell and Tube Heat Exchangers," 2nd
Symposium on Shell and Tube Heat Exchangers,
Sept. 1981, Houston, Texas (ASM).
8. J. W. Smith, "Considerations in the Utilization of
Tube Expanding," 2nd Symposium on Shell and Tube
Heat Exchangers, Sept. 1981, Houston, Texas (ASM).
 570

9. C. J. Gaffoglio and E. W. Thiele, Jr., "Tube-to-

Tubesheet Joint Strengths," ASME Paper 81-JPGC-
Pwr~7.
10. J. M. Duke et ai, Materials Considerations for
Large Sodium Heated Steam Generators, ANS Winter
Meeting, Nov. 27 - Dec. 2, 1977.
11. A. Lohmeier and S. D. Reynolds, Jr., Carbon-Steel
Tube-to-Tubesheet Joints for High Pressure Feedwater
Heaters, Westinghouse Engineer (magazine), 1965.
12. B. J. Sirois and E. W. Hartsell, "Large Surface
Condensers Benefit from New Welding Techniques,"
Power, September, 1966.
13. R. Hardwick, Methods for Fabricating and Plugging
of Tube-to-Tubesheet Joints by Explosion Welding,
Welding Journal, April, 1975.
14. R. H. Wittman, "The Explosive Bonding Process:
Applications and Related Problems," paper AD67-177
delivered to ASTME at 1967 Meeting.
15. S. D. Reynolds, Jr., Tube-Tube Sheet Porosity:
Cause and Cure, Welding Design and Fabrication,
June 1976.
16. S. D. Reynolds, Jr., et al, "Development of Nickel
Alloy Surfacing Techniques for Feedwater Heaters,"
Welding Journal, 42(8), 635-643, (1963).
17. S. D. Reynolds, Jr., Tube-to-Tubesheet Welding of
Straight Chromium Stainless Steels for Moisture
Separator Reheater Applications, Presentation to
16th National Heat Transfer Conference, August 8-11,
1976, St. Louis, Mo.
 571

18. S. D. Reynolds, Jr., Metallurgical Aspects of Tube-

to-Tubesheet Welding, Presentation to 20th AWS Midwest
Welding Conference, November 13, 1974.
19. "Welded-In Tubes - A New Technique for Stronger
High Pressure Heaters," Mech. Topics, 16 (1),
The International Nickel Co., Inc.
20. S. D. Reynolds, Jr., A Metallurgist Looks at Tube-
to-Tubesheet Welding, Presentation ASM Wilmington
Chapter, April 9, 1973.
21. E. J. Barsness and P. D. Ritland, Hydraulic Effects
on Inlet End Wastage in Feedwater Heaters, ASME
Winter Annual Meeting, December 1-5, 1968.
22. J. L. Kratz, "Material Considerations for the
Aluminum-Tubed OTEC Test Article Heat Exchangers,"
8th OTEC Conference, June 9, 1981.
23. A. A. Omar and C. D. Lundin, "Pulsed Plasma-Pulsed
GTA Arc: A Study of the Process Variables," Welding
Journal, Volume 58, No. 4, p. 97s, April, 1975.
24. N. T. Burgess, "Welding of Tubes to Tube Plates,"
The Institute of Welding, 1963.
25. B. D. Hackney et al, "A Study of a Welded Tube to
Tube Sheet Joint Design, General Electric Co.
Atomic Power Equipment Department Report APED-4612,
May 1, 1964.
26. J. F. Sebald and L. W. Hawthorne, "Mechanical
Effects Resulting from Welding Copper Base Alloy
Tubes to Tubesheets," presented at ASME Winter
Annual Meeting, December 1-6, 1957.
27. F. X. Brown and W. T. Lindsay, Environment Effects
in the Design and Operation of Feedwater Heaters,
American Power Conference 29th Annual Meeting,
April 25-27, 1967.
 572

28. J. R. Spence et al, "The Development and Production

of High Pressure Feed Heaters for Modern Central
Power Stations," Combustion, March, 1969.
29. E. P. Vilkas, "Microprocessors and CNC GTAW Systems,"
AD81-268, Westec 81 Conference, March 23 to March 26,
1981.
30. J. F. Key, "Anode/Cathode Geometry and Shielding
Gas Interrelationships in GTAW," Welding Journal,
Volume 59, No. 12, p. 365S, December, 1980.
31. G. M. Slaughter et al, "Welding and Brazing of High
Temperature Radiators and Heat Exchangers," Welding
Journal (AWS), January, 1968.
32. ASME Boiler and Pressure Vessel Code, Section VIII,
Appendix A.
33. K. S. Brundige, "Modern High Pressure Steel Tube
Feedwater Heater Design," paper delivered to
Pacific Coast Electrical Association: March 4, 1966.
34. A. C. Pergola and A. Phillips, "History of Foster
Wheeler Rolled and Welded Feedwater Heaters,"
Heat Engineering (Foster Wheeler), September 1965.
35. J. Tillinghast, "Steel Tubes vs. Copper Alloy
Tubes for Feedwater Heaters in Sub-Critical and
Super-Critical Power Plants," Discussion in ASME
Power and Fuels Division, February 19, 1968.
36. E. W. Hartzell, Jr., "Welding Drawn Stress-
Relieved Copper-Nickel Tubes to Steel Tube Sheets,"
Welding Journal 41 (6), 541-547 (1962).
37. "Automatic Tungsten Inert Gas Tube Joint Welding,"
Griscorn-Russell News, winter 1968.
38. ASM Handbook Volume II, pp 260-270.
 WELDING EQUIPMENT AND PROCESS DEVELOPMENT FOR INACCESSIBLE
TUBE-TO-TUBESHEET WELDING ON A SODIUM-HEATED STEAM GENERATOR

J. H. KIEFER

Materials and Processes

.'--' —-Breeder Reactor Component Project
'"'• Westinghouse Nuclear Component Division

i<rf' ABSTRACT

Since 1977 Westinghouse Nuclear Components Division at Tampa,

Florida, has been actively involved in a U.S. Department of Energy
contract to design and develop a steam generator for future
breeder reactor plants. The contract includes preliminary design
of a plant-size steam generator and detail design, development,
fabrication, testing, and post-test examination of a 70 Mwt
prototype of the plant unit.

A unique feature of the Westinghouse steam generator is the

reliability afforded by the use of a double-wall tube and also a
double tube-to-tubesheet (T/TS) weld barrier between the
potentially reactive fluids-sodium and water. Each tube is welded
in two places on each end for a total of four T/TS welds per
tube. Two of these four are conventional tubesheet face side
welds. The other two are termed "inaccessible" welds because the
welding must be done on the tube side of the tubesheet. For a
sodium/water reactor to occur, both a face side weld and a tube
side weld must fail on the same tube, which makes the event very
improbable. The design also includes a leak detection system
which will sense and locate a leak at any weld and between the
double tube walls.

The inaccessible welds are unique because the welding must be

done completely from the outside of the tube with very limiting
space between the tubes. Internal bore welding is not an option
because the double-wall tube would be fused into a single
boundary, precluding leak detection. Specialized gas tungsten arc
welding equipment was developed which could fit within the small
space between the tubes and perform the welding automatically with
cold-wire feed. Filler metal addition presented some challenging
equipment design problems, but it provided the opportunity for
controlling weld metal chemistry and the flexibility of
accommodating a greater range of weld sizes.

573
 574

Rod-anode radlographic inspection of each T/TS weld will be

performed during fabrication of the steam generators to guarantee
volumetric soundness. Conventional GTAW equipment will be used to
repair manually any accessible defects. Special weld repair
equipment was developed for removing weld defects in the
inaccessible areas and performing local repair welds
semiautomatically.

This paper describes the development work performed in

connection with the T/TS welding, the challenges associated with
this welding because of the unique design of the steam generator,
the welding equipment which resulted, and the peripheral equipment
that made this specialized welding possible.

Performed under U.S. Department of Energy Contract

DE-ACO2-77ET372O1
 575

INTRODUCTION

Since February, 1977, Westinghouse Nuclear Components Division

(Tampa Plant) has been actively involved in a U.S. Department of
Energy program to design, develop, and fabricate a prototype of a
Large Scale Steam Generator for future commercial breeder reactor
applications. The main design features of the Westinghouse steam
generator are a straight shell, three tubesheets, and straight
double-wall tubes, with an integral leak detection system
utilizing grooves at the interface of the double-wall tubes.
These features of the Uestinghouse design are shown schematically
in Figure 1. There are four tube-to-tubesheet welds on each tube
which maintain the double boundary between the heat exchange
fluids. Two of the four welds (Welds No. 1 and No. 4) are located
at the terminal ends of the tube joining them to the "face" of the
tubesheet. The other two are within the tube bundle on the
backside of the tubesheets and are termed the inaccessible welds.
This paper focuses on the inaccessible tube-to-tubesheet (T/TS)
welding development effort including process selection, equipment,
techniques, and parameter development.

BACKGROUND

The Westinghouse steam generator has three important factors

to be considered with the manufacture of its T/TS welds. First,
the tubes and tubesheets are 2-l/4Cr-lMo low alloy steel which is
air hardenable. Consideration must be given to both the post-weld
hardness resulting from the phase transformation and the residual
stress level from the welding cycle. Secondly, the leak detection
design and double-boundary concept of the steam generator dictates
that the tubes be welded on both sides of one of the three
tubesheets. Ordinarily, T/TS welds made on the back side of the
tubesheet are done so with the internal bore welding (IBW1
technique. However, the double-boundary design requires equipment
and techniques for producing these welds completely from the
"inaccessible" side. The IBW technique is inappropriate in this
situation since it would fuse the inner and outer tubes together
thereby reducing them to a single boundary between heat exchange
fluids. The third requirement is that the T/TS welds must be free
from volumetric defects as determined by rod anode radiographlc
inspection to stringent acceptance criteria.

PROCESS SELECTION

Gas tungsten arc welding (GTAW) was chosen because of the high
weld quality possible and its superior control over heat input for
welding thin materials. A commitment to the cold wire feed
process was made early in the program. A wider range of weld
sizes is possible with filler addition. The weld metal chemistry
can be modified to control weldabillty and weld metal hardness.
 576

STEAM

STEAM TUBESHEET
TUBE-TO-TUBESHEET
WELD NO. 1
HELIUM CHAMBER
TUB E- TO- TUB ESH EE T— k j M
N SODIUM TUBESHEET
WELD NO. 2 W ^

Na IN

CONVOLUTED SHELL
EXPANSION JOINT

TUBE-TO-TUBESHEE-f
WELD NO. 3 WATER TUBESHEET
TUBE-TO-TUBESHEtT-
WhLD NO. 4

FIGURE 1 . OVERALL VIEW OF WESTINGHOUSE DOUBLE-WALL TUBE STEAM

GENERATOR
 577

It was also reasoned that if effective manual repair techniques

could be developed for the inaccessible welds, then automatic
(mechanized) welds could also be made in these inaccessible areas
with filler addition. No equipment had yet been developed which
could reliably add filler metal between the tube spacing imposed
on the inaccessible Welds No- 2 and Mo. 3 (approximately 15 mm).
To insure that an effective wire feed technique was developed,
work on both repair and automatic welding was conducted in
parallel. A visualization of the access problem is shown in
Figure 2.

The T/TS welds are made with essentially no preheat to

simplify fabrication. Since 2-l/4Cr-lMo is air hardenable, the
effect of a preheat on reducing the hardness of the weld is
negligable. With a joint of relatively low restraint, and a
no-hydrogen process such as GTAW, the concern for underbead
cracking is also reduced.(1) For other processes such as
submerged arc and shielded metal arc welding this is not the
caue. Particular attention must be paid to joint cleanliness and
dryness, however. The base materials are maintained at a
temperature above ambient to avoid moisture condensation. The
T/TS welds are individually post weld heat treated with a high
frequency induction process to reduce the weld metal and heat
affected zone hardness-

EQUIPMENT SELECTION

Selection of a supplier for the automatic T/TS welding

equipment was based primarily on the previous experience with
orbital T/TS equipment designs. The key "factor was the addition
of filler metal. Due to the close spacing of the tubes, it would
not be possible to orbit a filler wire drive system around the
weld joint which is the common approach with orbital pipe welding
equipment using filler metal. RTA Welding S. A. of Rome, Italy
had prior experience with orbital welding gun designs which
utilized a wrap-around wire guide tube. They had produced several
tube-to-tube and tube-to-tubesheet welding machines with
restrictive space limitations for nuclear and non-nuclear T/TS
weld applications. Further, a prototype machine for operating in
space limitations similar to the W_ design for a sodium steam
generator application had also been accomplished. For this reason
an RTA welding system was selected.
 578

y Back row tubes

welded and inspected

Tube-to-tubesheet
weld
Most inaccessible
location, no\ Row being welded
direct line
of sight libes not installed
until prior row
is completed

FIGURE 2. ILLUSTRATION OF LIMITED ACCESS AND TUBE INSERTION

SEQUENCE FOR INACCESSIBLE T/TS WELDS

AUTOMATIC WELDING DEVICE

The RTA 195 shown in Figure 3 is designed to operate within

the minimum tube space envelope of approximately 15 mm. This
device is secured to the joint by two clamps located at each end
of the head* Its axial position is controlled by using the
surface of adjacent bosses as a repeatable reference. Another
adjusting mechanism is provided to control the radial electrode
setting. The electrode and gas nozzle are mounted on a spring
loaded swingarm- At the end of this bracket is an adjustable
"feeler" which rides on the outside diameter of the boss just
below the electrode. This establishes the electrode position off
the tube wall. Figure 4 shows the torch end of the welding
device. The 195 torch is water-cooled. The cooling water
shielding gas, filler wire, and welding current cables are brought
to the torch through hoses which orbit the tube to be welded on a
helical take-up spool. Proper tension and placement of the hoses
are critical and are maintained by a matching spool inside the
head body from which the hoses unwind.
 579

FIGURE 3. RTA MODEL 195 INACCESSIBLE SIDE T/TS WELD HEAD

FIGURE 4. TORCH END OFRTA 195 APPLIED TO T/TS WELD MOCKUP

 580

Several refinements of the RTA 195 were required to make it

reliable for production application. The initial torch design had
an electrode position adjustment which was set to follow the tube
adjacent to the weld. The location of the follower between the
electrode and the pivot point served to amplify iny misalignment
of the torch at the electrode. By moving the "feeler" to the side
opposite the weld onto the tubesheet boss, accurate joint tracking
was achieved. Improvements in the diameter and contour of the
open-sided take-up helix were made. Other minor changes included
simplifying the method of assembling and disassembling the torch,
improving the smoothness of the rotation drive gears, and
increasing the stiffness of the wire guide tube.

The two most important improvements in the device were a

change to tachometer feedback control of the wire feed system, and
an improvement to the shielding gas flow. Prior to welding, the
hoses and wire guide tube are prewrapped around the spool which
surrounds the tube to be welded. Welding is accomplished while
the hoses unwind. At the beginning of the weld, the filler wire
travels an additional wrap around the tube compared to the end of
the weld where the hoses are unwound. Without servo control, this
changing load condition was corrected by decreased speed settings
on the programmer as the weld progressed. Day-to-day variations
in these control settings occurred, however, due to changes in
equipment temperature, drive roll setting, wire stiffness, and
wear-in of the wire guide tube. Installing tachometer feedback
motors eliminated the variations in wire feed rate and total wire
input from weld to weld.

Due to the very stringent radiographic acceptance standard, it

was necessary to optimize every aspect of jhe welding operation.
A type of "smoke" test was used to investigate the adequacy of the
shielding gas cover. Dry ice iii water was chosen because the
materials are readily available, are non-toxic, and do not cause
permanent contamination of the gas system. The condensed water
vapor, or fog, which is produced is very dense and can be seen
with a bright auxiliary light source such as those used with
fiberoptic equipment. The distances from the fog source and the
nozzle was kept to a minimum (8-inches) to avoid loosing the fog
due to condensation in the lines. Sublimination of the dry ice
alone was enough gas evolution to provide the flowrate typical of
GTAW welding. The flow characteristics of the nozzle were tested
with combinations of screens of different mesh, without any flow
diffuser, and with a combination of steel wcol and screen.
Condensation on the screens limited the effectiveness of each test
 581

to about 10-15 seconds. By far the worst condition was produced

with a single loosely placed screen. Gaps around the electrode
and between the screen and nozzle wall created jets of gas leading
to a great amount of turbulence. The least amount of turbulence
was produced with two pairs of screens, each with one fine and one
course mesh screen. One pair was placed on each side of the 40°
angle (electrode angle) in the nozzle channel. The stainless
steel wool was effective, however, it was discarded because small
loose particles of the wool could potentially fall out into weld.

PROCESS CONTROL

Fabrication of the steam generator is row-by-row (vertical

rows) with the tubes in the horizontal position. The T/TS welds,
therefore, are made out-of-position (5F). RTA supplied
Westinghouse with a relay logic, position control concept coupled
to a Miller Analogue 300 power supply to vary the welding
parameters. Position control, as opposed to timer control, is
ideally suited for parameter development. Quick in-process
adjustments to any welding parameter can be made without affecting
the weld tie-in position. By comparison, changes during welding
to the travel speed in a time based system will shift the position
of all subsequent levels in the program. During development of
the welding techniques and parameters, it was found that a
multipass technique would be required for each weld. For each
pass a significant number of control changes would be required on
the RTA programmer. This was seen as a potential source of
equipment setup errors which could leak to costly repairs. It was
decided to substitute the relay logic programmer with a
microcomputer which could permanently store the parameters of all
passes for each weld. The effort necessary to change the weld
profile could be reduced to simply calling up a coded weld pass
identification number. In addition to this improvement to the
system, closed-loop servo control was added to the rotation and
wire feed as previously mentioned. The microcomputer, power
supply, and data logging system have been supplied by System
Technika Inc. of Huntington Beach, California. These were
interfaced with the RTA welding gun and orbital welding head
retaining the position control upon which the parameters were
originally developed. Figure 5 shows the complete microcomputer
programmer, switching type transistor power supply, magnetic
cassette parameter storage, and data logging system.

POWER SUPPLY

The System Technika power supplies are significantly different

in operation from the Miller Analogue 300. The Miller supply is
an SCR controlled constant current type whereas the System
Technika is a switching transistor controlled type. The first has
a 360 Hz superimposed signal on the DC component, and the second
produces a 17KHz superimposed signal. In practice it was found
 582

FIGURE 5. SYSTEM TECHNIKA T/TS WELD CONTROL SYSTEM AND RTA

MODEL 180 WELD GUN

that within the range of operation for this process, the welding
characteristics, i.e., bead width and penetration, were
equivalent. This is only possible with machines with closed-loop
controlled current and essentially square waveform pulsing
response. The pulsed current process parameters developed from
one system were directly transferred to the other with no
redevelopment of parameters required.

TECHNIQUES AND PROCESS VARIABLES

It is estimated that over 4000 welds were made throughout the

weld development program. Most were done in the preliminary
development phase on accessible type welds (No. 1 and No. 4) and
during equipment development. By the time the prototypic welding
equipment was available, the basic ranges where the welding
parameters would fall were well understood. The most practical
approach to establish the final values for each weld was by
trial-and-error based on engineering judgment and practical
welding experience. The three most important considerations were
the size of weld which could be controlled out of position, the
penetration into the tube thickness, and the rate at which filler
metal could be added.
 583

Shielding Gas

A shielding gas composed of 100 °/o argon was selected for

all welds based on its documented penetration characteristics
which are well described in a recent Welding Journal
article.(2) Compared to helium, the penetration profile of
argon is more suited for controlling the weld penetration at the
tube wall for the inaccessible welds. With filler metal addition,
the size of the weld can be controlled by the amount of wire added
to the puddle. Deep penetration into the base metal is not
required to produce a throat of adequate size. Better wetting at
the toe of the weld is also achieved with argon over helium.
Ultra pure quality argon (e.g. 1 ppm H2O) is used to insure
that the shielding gas is not a source of hydrogen contamination.
Oxygen and moisture analysis is used to setup and monitor the
quality of the gas from the bottle and at the nozzle exit.

Welding Current

A pulsed-arc current is being used for added control over the

penetration and puddle for out-of-position welding. A pulse
frequency of approximately 1 Hz is being used. Studies have shown
that a ratio of the peak, pulse time to background pulse time
(pulse ratio) of approximately 3/1 enhances penetration for the
same average current.(3) L O W pulse ratios (1/4) afford an
increase in puddle control out-of-position. These effects were
used whenever possible to develop the final welding parameters.
For filler metal welds where filler is synchronized with the peak
pulse, the low pulse ratios which were desired were not
practicable. It was found that the longest background time needed
to freeze the puddle was approximately 0.5 seconds. To achieve a
1/4 pulse ration the peak pulse time would be too short
(0.12 seconds) for the wire drive motor to be effective even at
maximum speed.

Travel Speed

A continuous travel or rotation speed was used. Although the

programmer provided for rotation syncronized with the background
pulse time, it does not have any significant advantages for this
procedure. A disadvantage is that during development of the pulse
parameters, changes to the background time would affect the
average travel speed, which would exaggerate thn change to heat
input and affect the amount of filler metal added.

In order to hold down production costs it was desired to keep

the weld time to a minimum. High travel speeds, however, require
proportionally more filler wire and welding current for a given
weld size. The maximum wire feed motor speed and the maximum tube
 584

wall penetration were limiting factors. Heat buildup in the weld

joint limited the minimum travel speed. Since the welds are made
on a relatively long trepanned boss, the heat buildup due to long
weld times makes out-of-position puddle control difficult.

Welding Sequence

The welding parameters for the inaccessible welds (Nos. 2 and

3) were the most difficult to develop. Single pass techniques
were tried but were quickly abandoned in favor of multiple
passes. The problem here was not with controlling the puddle
out-of-position, but in limiting the penetration into the outer
tube wall. In order to preserve the leak detection grooves in the
outer tube, no penetration into the top of the groove is allowed.

A multi-pass technique was developed consisting of autogenous

and filler passes. The small arc gap which can be used when no
filler is added concentrates the heat into the root of the joint
on the first pass. This minimizes the problem of forming a bridge
of fillermetal across the root at the weld start which can then be
carried around the joint. The usual means of preventing this type
of lack of root fusion is by increasing the heat input. Again,
penetration into the leak detectioj grooves limited the amount of
current hich could be applied. The resultant contour in the root
from the first autogenous pass also greatly improved the puddle
formation conditions for the first filler pass.

Although the first autogenous pass contributed some initial

penetration to the amount of weld throat which was ultimately
needed, a subsequent single filler pass could not be made large
enough without fusing into the grooves. A third filler pass was
required to meet the minimum leak path requirement. Figures 6 and
7 show the typical inaccessible weld appearance and cross section.

FILLER METAL

A 0.035" diameter low carbon 2-l/4Cr-lMo filler metal is

specified for all T/TS welds including repair welding. The low
carbon type (.05 °/o max.) was selected over the regular carbon
(0.15 °/o max.) because of its superior weldability and low weld
metal hardness in che as-welded and post-weld heat treated
condition.
 58S

FIGURE 6. TYPICAL SURFACE APPEARANCE OF WELDS NO. 2 AND 3

FIGURE 7. TYPICAL CROSS-SECTION APPEARANCE OF WELDS NO. 2 AND

NO. 3 (SECTIONED BETWEEN LEAK DETECTION GROOVES.)
 586

The supplier of the weld wire is United States Welding of

Chatsworth, California. They have a unique, patented wire
manufacturing process that prduces a relatively contamination free
wire surface. The wire is rolled to size as opposed to drawn. No
lubricants are used in the process which can become imbedded in
the wire surface and cause weld contamination.(4) The wire is
vacuum annealed between the rolling steps to eliminate work
hardening. The cost of the wire is relatively high, but weighed
against the potential cost to repair an inaccessible T/TS weld, it
is justified.

TUBE-TO-TUBESHEET WELD REPAIR

Repair of conventional T/TS weld defects during fabrication

typically involves removal of the defect by grinding and
backfilling the cavity with the manual GTAW process using filler
metal. Standard equipment was not available, however, for repair
of the inaccessible side T/Ts welds. It was necessary to develop
both the equipment for removal of the weld defect, and for welding
with the addition of filler metal. Without repair welding
capabilility, the alternative would be to cut out the entire
double-wall tube, and prepare the boss surfaces for automatically
welding a new tube. This is undesirable for obvious reasons,
especially if the rejectable defect is a single isolated pore
which would be a simple repair on the most accessible area.

A manual approach was selected over automatic repair.

Although it might be possible to preprogram automatic equipment
for a range from partial to complete reweld, the primary approach
was to retain the flexibility of a manual process.

Defect Removal

The first logical step in developing a repair process is to

identify a method for removal of the defect. A survey was taken
of the commercially available industrial grinders. At that time,
the smallest high-speed grinder found was 15mm in diameter, which
is too large to be operated between the tube spacing. The
possibility of complete weld removal with a clamshell type of end
mill was investigated. Again, no such cutter was readily
available.

Two types of dental hand pieces were tried, the flexible cable
drive type, and the ultra high-speed air drive. The flexible
drive was able to reach to the most inaccessible area of the weld,
but the cutting speed was poor and the right angle drive gears at
the head would overheat and quickly deteriorate.
 587

The ultra high-speed cutter not only removed metal much faster
than the mechanical drive, but it was better suited for
application with a positioning fixture. With the handle cut to a
short stub, it can be clamped into a positioning sleeve for
accurate control of the cutting burr (Figure 8 ) . The cutter can
be angled toward the tube to control the radial cut depth. The
cutting depth in the direction of the tubesheet boss is controlled
by an adjustable collar which clamps onto the outside diameter of
the boss. The air turbine operates from a standard shop air
supply filtered and regulated to approximately 40 psig.

Repair Welding Torch

The inaccessible weld repair equipment was developed in-house

by W_ personnel. No manual GTAW equipment was available which
would fit between the tube spacing limitations. Three major
hurdles had to be overcome to produce a viable manual welding
torch. The first and most important was to have a GTAW torch
which could be controlled manually and produce a weld of high
quality. The second was to be able to add filler metal where
needed with enough control to avoid electrode contamination. The
final problem was providing the welder with clear visibility of
the weld puddle.

A high-temperature plastic was selected for the torch body

material. It is easily machined, has good dielectric and
high-temperature properties, and is more resistant to breakage
than ceramic. To protect the area nearest the arc against heat
damage, a section of the plastic was replaced with a ceramic piece
machined from Grade A lava. The open window provided by this
removable section also greatly increased the access for electrode
adjustment. Figure 9 shows the patented torch design applied to a
joint with the heat shield removed. Water cooling is delivered to
the end of the electrode holder to increase power cable and
electrode life.

With an enclosed chamber, the only method of introducing

filler wire is through a guide tube similar to automatic systems.
The force required to push the wire through the guide tube
typically is too high for a manual addition technique. To solve
this a foot-switch operated wire drive motor was installed, since
no visual system ad yet been developed, a special fixture was
required so that the "mechanized manual" wire feed system could be
tested. A torch body without the gas shielding enclosure was used
inside of a clear plastic, argon purged chamber. Good control
over the amount and placement of filler metal added was possible
in the areas where there was clear visibility. Once a fiberscope
mounted into the repair torch could provide clear viewing for the
 588

FIGURE 8. WELD REPAIR CUTTER APPLIED T/TS WELD JOINT MOCKUP

FIGURE 9. T/TS WELD REPAIR TORCH APPLIED TO T/TS WELD JOINT MOCKUP
 589

entire length of the inaccessible weld. The tube diameter and

pitch preclude any direct line of sight to the rear of the weld.
Welding with mirrors has been used for manmy other applications
where direct vision is not possible, but for this intricate repair
operation, it was found to be impractical.

A fiberscope with a custom viewing format was built by

American Optical to BRCP specifications. This instrument combines
components from several off-the-shelf models. The objective lens
was selected to provide maximum magnification at the focal
distance compatible with the weld torch design. The protective
sheath has to be free to twist axially so that the objective end
could rotate around the weld joint with the viewing end fixed.
The image is projected for viewing with an American Optical
Welding Monitor. The Welding Monitor use;: an iris type of
aperature to reduce the light intensity instead of a fixed-density
filter plate. The welder can adjust the brightness level to his
preference. The image remains in color, and both eyes can be used
on the screen which reduces eye fatigue.

Optical filters were tested with the Welding Monitor Co

enhance visibility of the weld puddle. There was no significant
improvement of contrast with any of the filters which were
selected to remove peak intensities associated with the argon
shielded arc. Since the images of the weld area are produced by
reflected arc light, any filtering of this light will produce a
corresponding reduction in weld area brightness. Selective
brightness reduction of the image would not be possible,
therefore, with uniform optical filters. Backlighting would
increase visibility but is impractical for this application. The
fiberscope and Welding Monitor viewing system are shown in
Figure 10.
 590

FIGURE 10. T/TS WELD REPAIR SUPPORT APPARATUS

After establishing that the repair equipment would produce

sound filler metal welds with adequate «reld puddle visibility, two
additional improvements were made in the area of penetration and
filler metal control which greatly improved the process
reliability. Until this time, a very high level of skill was
required on the part of the welder to prevent excessive weld
penetration into the leak detesction grooves and to control the
amount of filler metal added. The problem of weld penetration
control was solved with a water cooling mandrel which circulates
water over th^ inside diameter of the double-wall tube at the weld
zone. Constant water flow is circulated between 0-rings by a
standard water cooler/circulator normally used for water-cooled
welding torches. This device is not required for automatic
welding where heat input can be accurately controlled. Figure 11
shows seven completed welds using this equipment on a Westinghouse
model sodium steam generator.
 591

FIGURE 11. INACCESSIBLE WELDS ON THE WESTINGHOUSE FEW TUBE MODEL

STEAM GENERATOR.

Improvement in the control over filler addition was

accomplished with a process patented by this author which falls
into the category of semiautomatic gas tungsten arc welding. 'Che
control and wire feed unit in Figure 10 identified as "AVC Wire
Feed Controller" was built to BRCP specifications by CRC Welding
Systems Inc. (formerly The Merrick. Corp.). It cycles the wire
feed drive unit similar to the foot switch described earlier. The
welder judges the amount of filler metal to be added by the arc
gap over the weld puddle. This clearance, which was previously a
human judgment, can be measured as a function of arc voltage. For
a constant current level, a change la the arc voltage is an
indirect measurement of the amount of filler metal being added to
the joint. The arc voltage, therefore, can be used to regulate
the weld puddle height. Arc voltage is most commonly used to
control electrode position for the gas tungsten arc process,
welding current for gas metal arc process, and electrode feed
speed for the constant current submerged arc welding process.
 592

SUMMARY

Microcomputer controlled GTAW equipment and process techniques

have been developed capable of producing high quality
tube-to-tubesheet welds within tube spacings as low as 15 mm.
Maximum flexibility in the final weld size and tube wall
penetration is achieved through using cold wire feed. Pulsed
welding current is used to provide an effective means of
controlling bead shape and penetration for the out-of-position
welds.

Significant improvements to the process included the use of

extremely clean filler wire, addition of servo control to the wire
feed system, and improvement to the shielding gas flow lenses by
injecting fog from dry ice in water through the nozzle. Permanent
magnetic storage of the welding parameters was added to simplify
procedure changes during production.

Tube-to-tubesheet Weld

Repair equipment and techniques were also developed. A

special defect removal tool was designed based on a miniature air
driven dental drill. A semiautomatic GTAW process using arc
voltage to control filler metal addition was developed to restore
weld areas where defects have been removed. An enclosed plenum
torch is used with fiberoptics for remote viewing. Repairs can be
made with this equipment ranging from refusion of the weld surface
to complete weld replacement.
 593

8.0 REFERENCES

1. Provoznik, Louis J. "Welding 2-l/4Cr-lMo Steel Heavy Wall

Piping Using Minimum Preheat Requirements", Journal of Metals,
Volume 30, No. 10, pp. 25-27, October 1978.

2. Key, J. F. "Anode/Cathode Geometry and Shielding Gas

Interrelationships in GTAW", Welding Journal, Volume 59,
No. 12, p. 365S, December 1980.

3. Omar, A. A. and Lundin, C. 0. "Pulsed Plasma-Pulsed GTA Arc:

A Study of the Process Variables", Welding Journal, Volume 58,
No. 4, p. 97S, April 1975.

4. York, J. W. "Plasma Cladding Dye Penetrant Indications are

Related To Wire Cleanliness", Welding Journal, Volume 59,
No. 12, p. 32, December 1980.
 •
•r , \O
of?
 CREEP RUPTURE AND CREEP CRACK GROWTH
BEHAVIOUR OF TRANSITION JOINTS

B. Nath

Central Electricity Generating Board

Kelvin Avenue, Leatherhead, Surrey
KT22 7SE, UK

ABSTRACT

The creep rupture and creep crack growth behaviours of two commonly
used, as welded, transition joints viz. 2|CrlMo/Inconel 182/AISI 316,
and 2iCrlMo/AISI 316/AISI 316 have been studied at 580°C. It has been
shown that the dissimilar metal welds are inherently weak in rupture and
that the rates of crack growth along interfaces are rapid. Empirical
equations defining relationships between stresses and rupture lives of
the two joints have been formulated, and can be used to calculate long
term rupture strengths.
Three different and independent Types of failures have been
identified, namely {Type I) interfacial fracture, {Type II) brittle
failures, partly along the interface and partly through the HAZ and
{Type III) ductile rupture in the ferritic steel. A mechanism for creep
deformation of dissimilar metal welds has been put forward to account
for the behaviours observed. Roles of various factors, e.g. C-denuded
zones, mismatch in coefficient of thermal expansions, relative strengths
and ductilities of the ferritic steels, the HAZ and the weld metal, the
microstructural features of interfaces and the oxidation of the low
alloy steel adjacent to interfaces are also discussed.

INTRODUCTION

In coal fired and nuclear power plants it becomes necessary to join

ferritic and austenitic steels together. The three most common designs
of such dissimilar metal welds in thin section tubular components are
(i) Flash Butt Welding or Pressure. Welding (ii) Welding with an austenitic

*The paper is published by permission of the CEGB.

597
 598

filler and (iii) Welding using a nickel based filler. It is noteworthy

that the first two methods result in ferritic/austenitic interfaces and
the last one in a ferritic/nickel alloy interface. Premature, brittle
failures of all three types of joints have occurred in service. In order
to understand the high temperature behaviour of dissimilar metal joints
many creep rupture investigations have been carried out on different
combinations. Most of the tests have been conducted ^550 C, therefore,
the following observations refer particularly to these temperatures.
The long term rupture strengths of cross weld transition joints
appear to be lower than those of the low alloy parent metal (1-10) .
Furthermore, depending upon the test parameters used, failures have been
reported to occur within the low alloy component or through the HAZ near
the interface in different combinations e.g. in ferritic/austenitic
joints(2~4,10,11)j £ n ferritic/nickel based filler combination(l~10>12)?
and in low alloy ferritic/high alloy ferritic welds(13). The rupture mode
changed from ductile parent metal to brittle interface fracture as the
stress was lowered.
The joints made with nickel based filler appear to be marginally
stronger in rupture than those fabricated with austenitic f illerd , 3,4,10) _
This difference could be ascribed to the presence of a soft carbon
depleted zone adjacent to the interface of a ferritic/austenitic weld.
However, it has been shown(3) that at low stresses such specimens with
^Imm wide, decarburised zones had rupture lives similar to those without
the preformed C-denuded region. This could be the result of C-migration
which might have occurred in ferritic/austenitic joints within the
duration of tests on specimens which were not aged prior to testing.
But in addition to 2|CrlMo/austenitic combinations(2,3,9)} similar
effects of post weld heat treatments have also been noted in 2jCrlMo/
nickel based filler welds. Further, brittle interface failures take
place in a variety of dissimilar metal joints including those fabricated
with nickel based fillers. It is noteworthy that C-deplete.i zones do not
form in the latter combination. Thus, the effect of a C-denuded band,
per se, on the rupture behaviour of transition joints has not b a n
established.
A research programme has been undertaken to understand the failure
mechanisms and to identify the metallurgical factors responsible. This
paper reports the findings on as welded joints.

EXPERIMENTAL

The details of the plate/plate butt welds and the position of the
creep rupture specimens (6mm dia.) are shown in Fig. la. All specimens
were individually V -radiographed to ensure that the interfaces were
free from defects. All creep tests were carried out in air at 58O°C
and at 135 to 45 MPa. Notches were introduced along interfaces in
25mm compact tension specimens, by spark erosion (Fig. lb). Creep crack
growth tests were also conducted in air at 58O°C. The analyses of the
parent metals and the weld deposits are given in Table I. Only one
cast of the ferri <_ic plate which had been normalised at 93O°C and
 599

tempered at 700°C by the manufacturer, has been used throughout the

investigation.

TABLE I: Chemical Analyses in Weight Percent

Material Cr Fe Mn Mo Ni Nb Si Ti

2jCrlMo 0.11 2.06 Bal. 0.62 1.08 0.13 0.01 0.23 <0.01
AISI 316 0.04 17.2 Bal. 1.82 2.19 9.18 <0.02 <0.2 <0.02
Austenitic
0.03 17.7 Bal. 0.89 2.29 10.0 <0.02 0.33 <0.02
Weld
Inconel
14,3 8.68 7.74 <0.1 Bal. 1.2 0.5 1.0
182

ROLLING
DIRECTION

QO:
201

FILLERS B AND tt S IKCONEL 18!

ELECTRODE SIZE 4 00 rror DIA J 97 mm DIA
PREHEAT !00°C MIN 200°C MIN
INTERPASS TEMP 3O0°C MAX 300°C MAX
POLARITY DC • VE DC • VE
CURRENT RANGE 115- ItS Amps 9S - 1 IS Amps
WELDING POSITION OOWNHANl OOWNHAND

f l 6 . H i ) PLATE BU1I W M i . t » 8 CUPS; w a n CHEEP S K t W t H S lb) WELP PADS

AMD 25mm » S SfECIMEHS WITH IHITUL i / w =0,1,31, A M NOTCH MG1
RADIUS Of 0.2mm. WELDS FABRICATED USING t m m d l l . ELECTR0OE5 Of
EITHER AISI 316 OS IHtOHEl 1(2 C0MP0S11IOH5.

K
 600

RESULTS

As Welded Structures

The structure of the parent 2|CrlMo plate consisted of a mixture

of ferrite and bainite, with a grain size of ASTM7. The heat affected
zone ("v2mm wide) revealed a mixture of fine and coarse grain bainite.

FIG. 2 AS WELDED 2;CvlMo/INC0NEL 182. NOTE CARBIDES WITHIN ^10 pm

OF (a) SHARP INTERFACE (b) DIFFUSE INTERFACE

Sharp as well as diffuse interfaces were seen in the 2sCrlMo/

Inconel 182 weld (Fig. 2 ) . Some precipitate particles were also
visible in the interface area. The EDAX spectrum and the electron
microprobe analysis (Fig. 3) revealed Fe to be the major substitutional
constituent and confirmed the presence of carbon in these particles.
Carbon peaks corresponding to the precipitate phase appeared in the
2£CrlMo within 10 ym of sharp interfaces and in the weld metal within
10 ym of the diffuse interface. (Interface is defined as the location
where Ni count begins to rise above the level typical of the 25CrlMo).
The concentrations of Fe, Ni and Cr changed gradually from values
typical of the 2|CrlMo to those of the weld metal over a distance of
50 to 200 ym on the weld metal side.
The microhardness variation across the interface is shown in
Fig. 4. The hardness increased gradually across the HAZ of the
2|CrlMo from a value of 185 VPN, typical of the plate, reaching a
peak of 300 to 360 VPN at ^200 ym from the interface. At sharp
interfaces it gradually fell off to a value of 200 to 240 VPN
 601

representative of the weld metal, on the Inconel 182 side of the

interface. At diffuse interfaces, however, the finger like structure
(Fig. 2) was found to possess a hardness of 330 to 360 VPN giving rise
to & sharp peak on the weld metal side of the interface.

., =

IHTERFACE

USTANCE .n

FIG. 3 MICROANALYSIS ACROSS AN FIG. 4 MICROHARDNESS ACROSS

AS WELDED SHARP 2jCrlMo/INCONEL 2jCrlMo/INCONEL 182. NOTE HARDNESS
182. NOTE CARBON PEAK WITHIN PEAK AT DIFFUSE INTERFACES.
M.0 vm OF THE INTERFACE.

Thin foil microscopy revealed a clean interface between lath

structure and equiaxed grains (Fig. 5 ) . Using selected area diffraction
patterns the laths were identified as b.c.c. and the equiaxed grains
as f.c.c. No other phase was found in the interface areas of the foils
examined. The 2£CrlMo HAZ structure was bainitic.
The as welded structure of the 2|CrlMo/AISI 316 interface was
different, in that no precipitate was visible at the interface (Fig. 6)
and that the non etching zone present on the weld metal side of the
interface had high hardness (Fig. 7 ) . Optically the interface was
not clearly delineated. Regional variations in the width of the non
etching zone were observed depending upon the extent of mixing.
Electron microprobe analysis revealed that the composition of the
2jCrlMo adjacent to the interface remained unaltered (Fig. 8 ) . On the
weld metal side the Ni and Cr levels increased and that of Fe decreased.
The hardness peak at ^20 um from the interface represents a composition
roughly midway between the two sides. The carbon counts were similar
on both sides of the interface and no evidence of carbon depletion
was found. Lath as well as acicular martensites corresponding to the
non etching zone were seen in thin foils (Fig. 9 ) .
 WELDED 2lCrlMo/INC0NEL o

FIG. 6 AS WELDED 2|CrlMo/AISI 316 INTERFACE. THE STRUCTURE CORRESPONDING

TO THE NON-ETCHING ZONE IS SHOWN IN FIG. 9
 60 3

FIG. 7 MICROHARDNESS ACROSS FIG. 8 MICROAXALYSIS ACROSS AX AS

2|CrlMo/AISI 316. NOTE HARDNESS WELDED 2jCrlMo/AISI 316
PEAK ON THE WELD METAL SIDE. INTERFACE.

FIG. 9 AS I'fELDED 2|CrlMo/AISI 316. (a) LATH MARTENS ITE AND

[ill] DIFFRACTION PATTERN (b) ACICULAR MARTENSITE
 604

Creep Rupture Behaviour

The creep properties of the parent metal arid the two transition
joints are summarised in Table II. The primary and the secondary
creep behaviours of cross weld specimens were similar to the 2;CrlMo
steel at stresses 3100 MPa (Table II, Fig. 10). At lower stresses
minimum creep rates of the ferritic steel were lower than those of the
dissimilar metal weir's. With the exception of the 2|CrlMo/Inconel 182/
AISI 316 specimen tested at 125 MPa the tertiary stages were small in
all cross weld specimens in comparison to the parent metal data.
Furthermore, at a given stress, the tertiary stage initiated earliest
in welds made with the austenitic filler, followed by the specimens
joined with Inconel 182, and then by the ferritic steel.

TABLE II: Summary of Creep Rupture Tests at 580 C

Stress Rupture Strain at Minimum Type of

Specimen a in Life, Fracture, Creep Rate, Parameter Failure
Material
No. MPa t in h e in h~l 10 2 (see text)
c
f s Vf * ~
P-01 125 667 0.38 1.35x10"^ 8.998
2iCrlMo Plate P-02 100 2758 0.2817 1.43xlO"5 3.944
(as received)
P-03 90 5790 0.1764 4.64xl0"6 2.687 ^ /
P-04 80 12316 0.2113 1.92xl0"6 2.359

S-07 135 284 0.1120 2.86xlO~4 8.114 I

4
S-06 125 380 0.0708 1.45xlO" 5.518 I
S-02 100 724 0.0183 1.79xl0"5 1.296 I
2JCrlMo/
6 II
AISI 316/ S-01 90 1434 0.02 8.43xlO" 1.209
AISI 316
(as welded)
S-03 80 1991 0.0179 4.66xlO"6 0.9278 II
60 6 II
S-04 6692 0.0184 1.28xl0" 0.8523
S-05 50 9949 0.0223 9.79xl0"7 0.974 II
S-48 45 12754 0.0176 4.78xl0"7 0.6098 II
N-06 125 669 0.2618 1.35xl0"4 9.032 III
N-07 115 962 0.1379 5.19xlO"5 4.993 II
2}CrlMo N-48 100 1850 0.0435 1.00xl0~5 1.850 II
Inco 182/ N-01 90 2460 0.059 8.81xlO~6 2.167 II
AISI 316
(as welded) N-02 80 3857 0.0393 3.38xlO"6 1.304 II
N-03 60 7906 0.0391 1.7OxlO~6 1.345 II
N-04 50 15040 0.0360 l.O7xlO"6 1.615 II
 605

:.c-iMo I H C O X E L is:

FIG. 10 CREEP CURVES FOR 241CrlMo AND CROSS WELD SPECIMENS AT 580 C.
NOTE SIMILAR PRIMARY AND SECONDARY REGIMES AT 125 MPa AND DIFFERENCES
IN MINIMUM CREEP RATES AT 80 MPa.
o :.CHMo *S RECEIVED
> ! . O ! M O iNCOHEL IS? AS OELDEO
C'iMo A1SI l i t AS "ELDED
PE Of f MLURt ; '.I OB I H

\,
\i J n ~ "jT~

FIG. 11 RUPTURE (a) STRENGTH AND (b) DUCTILITY OF THE DISSIMILAR

METAL JOINTS AND THE ijCrlMo PLATE.

The stress rupture curves for the three types of specimens are-
shown in Fig. lla. The short term rupture strengths of the as received
2|CrlMo were marginally lower than the ISO mean properties but the
differences vanished at longer rupture lives. In comparison, both
transition joints were significantly weaker in rupture. The differences
between the rupture lives of the dissimilar metal welds and the
2|CrlMo specimens increased as the stresses were lowered, At low
stresses, typical of those attained in service (<50 MPa), the
transition joints were failing in times only 0.1 to 0.2 of the expected
rupture life for the ferritic steel. At all stress levels, test pieces
made with the austenitic weld metal exhibited shorter rupture lives
than those fabricated with Inconel 182 - the ratio being 0.5 to 0.7.
But this difference was small in comparison to the divergence from
the properties of the low alloy steel. '
Three different types of failures were observed in cross weld
specimens (Fig. 12, Table II).
 606

2|CrlMo AISI 316 2zCrlMo INCONEL 182 2'CrlMo INCONEL 182

5 mm
i
i)tJ Tup,: II Type
i'i.G. 12 MACROGRAPHS ILLUSTRATING THE THREE TYPES OF FAILURES. TYPE I
IN 2|CrlMo/AISI 316/ATSI 316 AT 125 MPa, TYPE II IN 2|CrlMo/INC0NEL 182/
AISI 316 AT 90 MPa, TYPE III IN 2£CrlHo/lNC0NEL 182/AISI 316 AT 125 MPa

6 ^ :V" " ;""' SURFACE

SPECIMEN
SURFACE „
FIG. 13 WELD METAL HALVES AT CRACK NUCLEATION SITES. INITIAL PROPAGATION
OCCURS CLOSE TO INTERFACES, (a) 2{CrlMo/AISI 316/AISI 316 AT 80 MPa
(b) 2{CrlMo/INCONEL 182/AISI 316 AT 90 MPa

.:.;'.:• £ ''•;;^r^m$m

FIG." 14 FERRITIC HALF OF"THE 2|CrlMo/INCO i.R2/Al'S*I 316 AT^90*M!;a"

(a) THICK LAHELLAR OXIDE AT CRACK NUCLEATION SITE (b) THTN OXIDE
AND GRAIN FLOW AT THE FINAL FRACTURE LOCATION
 607

Type I - The entire fracture path lay in the 2|CrlMo HAZ (within 10
to 20 um of the interface) and closely followed weld head contours.
This mode of fracture was seen in the 2jCrlMo/AISI 316/AISI 316
specimens tested at 100 to 135 MPa.
Type II - The essential feature of a ?:/;'.-? II failure was that the crack
followed the interface for only a part of the specimen diameter. The
rest of the path traversed across the ferritic HAZ leaving a f err i tic-
ligament attached to the weld metal half of the test piece. This type
of fracture was seen in cross weld specimens made with the austenitic
filler at $90 MPa, and in those fabricated with the Inconel 182
electrodes at $115 MPa. These failures were always associated with
surfaces where the weld metal made an acute angle of 45 (Fig. 1 3 ) .
Thick lamellar oxides were seen on the fracture surface of the 2|CrlMo
on this side of the specimen and only limited cavitation was observed
in the ferritic component (Fig. 14a). On the other side (where the
weld metal made an obtruse angle) the oxide scales were thin and the
ferritic grains exhibited clear signs of deformation (Fig. 14b).
Type III ~ The failure occurred in the parent 2£CrIMo steel, away
from the interface, and the HAZ. This mode of fracture was observed
in the iJCrlMo/Inconel 182/AISI 316 specimen at 125 MPa, and was
associated with ductilities* comparable to the value exhibited by
the 2|CrlMo steel (Fig. lib). However, in this specimen interracial
cracks were also observed (Fig. 1 5 ) .

•«* «,v- ->&•:

* > , * , . ' - *

..- A'' •' -!'.*•'*••

i^y7'! STRESS
-•"•- !il 0.1mm r".*•• •'

H^^teff SPECIMEN^ iS^i

SPECIMEN M
^2 SURFACE

FIG. 15 CRACK AT THE 2jCrlMo/ FIG. 16 FERRITIC HALF OF

INCONEL 182/AISI 316 INTERFACE 2|CrlMo'/AiSI 316/AISI 316 F~RACTURED
IN THE SPECIMEN WHICH FAILED IN AT 100 MPa SHOWING A NARROW ZONE OF
THE FERRITIC STEEL AT 125 MPa. CREEP DAMAGE IN THE 2jCrlMo HAZ.

+Throughout this document strain values pertaining to dissimilar metal

joints have been normalised to take into account the observation that all
creep strain is accumulated only in the ferritic part of the gauge length.
All other parameters related to strain, e.g. minimum creep rate, strain
at rupture, are normalised accordingly.
 608

The elongations at rupture in the 2|CrlMo ranged from 17 to

38% (Fig. lib). At >100 MPa strains to failure were >5% for both types
of joints, but decreased with decreasing stress and increasing rupture
life. At £100 MPa the welds made with AISI 316 failed with very low
elongations of the ordsr of 2%. For the same range of stresses the
2|CrlMo/Inconel 182/AISI 316 joint exhibited higher ductilities than the
austenitic filled welds.

FIG. 17 THE 2lCrlMo/AISI 316/AISI FIG. 18 MICROANALYSIS ACROSS

316 SAMPLE AT 45 MPa. C-DENUDED THE 2jCrlMo/AISl 316 INTERFACE
ZONE NOT EVIDENT BUT WELD METAL AGED FOR 12,754h AT 58O°C. NO
ENRICHED WITH CARBIDES. CAVITA- C-DENUDED ZONE WAS OBSERVED.
TIONAL DAMAGE SEEN IN THE FERRITIC C-PEAK IN THE WELD AT ^100 um
HAZ. ARISES DUE TO CARBIDE ENRICHMENT.

The fracture appeared to be intergranular (although it is difficult

to be certain due to oxidation) and cavities were visible at prior
austenite grain boundaries (Fig. 16). The hard non etching band ir
welds made with AISI 316 filler transformed into a dark etching
acicular structure during exposure at 58O°C and did not exhibit any
cracks or cavities. Metallography (Fig. 17), microhardness
measurement (Fig. 7) and microprobe analysis (Fig. 18) revealed that
C-denuded zone did not develop in any of the test pieces, including
the one exposed for 12,754h. During exposure at 58O°C the hardness
of the HAZ decreased towards the hardness of the parent metal
(Figs. 4 and 7 ) .
Preferential oxidation of the 2lCrlMo steel has been observed to
lead to two morphologies near weld interfaces. On some specimen
surfaces oxide filled notches/cracks were seen, as illustrated in
Fig. 19a, and on others a step was observed (Fig. 19b). It is
noteworthy that only steps of oxides were seen in coupons aged \<dth
creep test pieces.
 609

•V r ./ '• , •. . /
SPECIMEN •*—-.•,
SURFACE

0.1 m m

FIG. 19 (a) OXIDE NOTCH FORMED WHERE BENDING OF THE 2*Cr1Mo WAS EVIDENT,
e.g. 2|CrlMo/INC0NEL 182/AISI 316 AT 90 MPa. (b) THE STEP MORPHOLOGY WAS
NOT ASSOCIATED WITH OBVIOUS GRAIN FLOW, e•gT~2jCrlMo/AISI 316/AISI 316 AT
80 MPa.

Creep Crack Growth

Marked differences in the creep crack growth behaviour of the two

types of weld interfaces were observed. Specimens of each weld
combination, containing the same geometry of artificial notch, were
loaded to the same reference stress levels of 45 MPa^ and K (14,15) o f
12 MPa m~i . In the 2|CrlMo/AISI 316 joint the crack propagated
immediately after loading, and became unstable in <100h (Fig. 20).
It should be noted that the uniaxial data at 45 MPa indicates a rupture
life of 12,754h (Fig. lla). Metallographic examination showed that
(i) the crack front was bowed, (ii) throughout the thickness of the
specimen the crack was located in the HAZ, approximately 1 grain away
from the interface, and had followed the weld bead contours of the
interface, and (iii) the creep damage extended to a depth of M.3 mm
from the tip of the spark erosion notch, and ^5 mm from the tip of the
advancing crack front (Fig. 21). Creep cavities ahead of the crack
front were revealed by breaking open the damaged specimen at liquid
nitrogen temperatures. MnS particles were found to be associated with
the cavities.

/ It is recognised that this reference stress calculated by Haigh

and Richards (15) method applies to homogeneous specimens.
 TIME, h
FIG 12 CBEf r WflCB GBOICTH AlOIVG A 2'.ClHJo IHCOHEL 187 HELD S5 MPJ A.VO K-• J2«Pjn; 1 '"

flG.21 CREEP CRACK GROWTH AT THE 2 'A Cf IMo/AISl 3W INTERFACE. CREEP DAMAGE
AT PRIOR AUST6MITE GRAIN BOUNDARIES OBSERVED AT UPTO 5mm AHEAD OF
THE CRACK TIP.
 611

The expected rupture life (>20,000h for the uniaxial specimen,

Fig. lla) was also not realised under creep crack growth conditions for
the 2jCrlMo/Inconel 182 specimen. The crack in this case did not
initiate until after 86Oh and did not become unstable until after 4000h
(Fig. 22). Bowing of the crack front, although observed, was smaller
in comparison to the specimen discussed above. The crack extension was
similar (>8 mm) to that seen in the 2|CrlMo/AISI 316 specimen but only
limited damage was seen ahead of the crack tip. The specimen could not
be broken open at liquid nitrogen temperature. It was noted that on
both surfaces of the specimen the crack had propagated along the
interface or at 1 grain from it (Fig. 23a). However, in the centre of
the specimen thickness a network of fine cracks was observed to Cake
the shortest path rather than closely follow weld bead contours
(Fig. 23b). Consequently, at some locations cracks were seen at
distances of up to 1 mm from the interface.

DISCUSSION

As Welded Structure

For both types of welds the hardness variation across the HAZ
(Fig. 4 and 7) reflects the change from the duplex structure of the
2£CrlMo parent through fine grain duplex structure at the edge of the
HAZ and fine grained bainite to coarse grained bainite. Softening
during exposure to 580°C is caused by precipitation of carbides.
In the vicinity of the interface itself the situation is more complex.
At sharp 2|CrlMo/Inconel 182 interfaces the hardness changed over
^ 40 pm to reach a value of 200 to 240 VPN, typical of the weld metal.
No 'light etching phase' was observed in these regions, ilthough such
structures have been reported by a number of workers in welds made
using nickel based filler(6~8, 16) ancj h a s been shown to be martensite(8),
There are two probable reasons for this difference. Firstly, the
composition of the weld metal adjacent to the interface could be such
that the Ms temperature is below room temperature. This is unlikely
because immersing an as welded specimen in liquid nitrogen did not
produce martensite. The second probable reason is that the high Ni
and Cr concentrations (and correspondingly low Fe level) produces stable
austenitic structure. Microanalysis results confirmed that the Ni and
Cr contents were high, and that very little dilution had occurred during
welding (Fig. 3 ) . Metallographic and microhardness measurements are
also consistent. Carbides formed in the HAZ and in the weld metal during
ageing at the test temperature, but no significant change occurred in
the nature of the interface itself. Carbides in the weld metal were
responsible for carbon peaks in microanalysis profiles (e.g. Fig. 23a).
At diffuse interfaces the narrow feathery structure (^ 20 um) gave rise
to a hardness peak on the weld metal side of the interface. On ageing
it transformed into a dark etching acicular structure (e.g. Fig. 19a).
It is, therefore, concluded that it is tnartensite(8,17).
FIG.23 CRACKING AT THE 2% Cr1Mo/lNCONEL 182 INTERFACE, (a)AT SPECIMEN
SURFACE UNDER PLANE STRESS CONDITION (b) AT SPECIMEN CENTRE (b) 0.2mm
WHERE PLANE STRAIN PREVAILS. NOTE THE DIFFERENCES IN DAMAGE
IN THE T W O CASES.
 613

Unfortunately direct identification of the precipitates at the

interface could not be carried out because the foils examined did not
_ontain this phase. However, they were constituted of mainly Fe and C
(Fig. 3 ) . They are, therefore, believed to cementite. M23C6 particles
have been identified at interfaces of welds made with nickel based
filler after long term serviced^, 19) . it appears that during welding
carbon atoms in the HAZ migrated towards the weld, due to the temperature
gradient. However, they could not diffuse into the weld metal because
of the high Ni content of the pool adjacent to the interface. On
solidification the excess carbon precipitated out as carbides in the
interface region. Other investigators who have observed continuous
band of martensite along interfaces of welds made with nickel based
fillers, indicating greater dilution, have not observed carbides,
presumably because the excess carbon could dissolve in the larger volume
of the diluted pool.
A hardness peak was observed on the weld metal side of the 2^CrlMo/
AISI 316 interface (Fig. 7). This non-etching thin band (^ 20 um)
contained more Cr and Ni and less Fe as compared to the 2|CrlMo (Fig. 8 ) .
Furthermore, exposure at 58O°C changed it into a dark etching acicular
structure (e.g. Figs. 17 and 21). It is there-fore concluded to be
martensite.

High Temperature Mechanical Properties

The results demonstrate that under uniaxial creep rupture

conditions dissimilar metal joints fail in the parent metal if the
stresses are high and along the interface if they are low. Such a
transition in the failure location has been reported in the literature
at 555O°C. A mechanism for creep deformation of transition joints is
being put forward below and the results are discussed in this light.
The 2|CrlMo halves of the composite specimens accommodate all
creep strains because the weid metals and the austenitic material are
much stronger in creep. It has been demonstrated that all the strain
is confined to the ferritic component^">''. Further support for this
conclusion comes from the observation that the primary and the
secondary creep regimes of both transition joints (calculated on the
basis of this assumption) are similar to those of the ferritic steel
at >100 MPa (Fig. 10).
The elongation in the stress axis is accompanied by a reduction
in the cros? sectional area. However, at the interface the material
is physically constrained from such a contraction by the rigid weld
metal. This introduces triaxial stresses at the interface with a
significant stress concentration where the weld metal makes an acute
angle (^45°) with the surface. The following experimental evidence
indicates that the cracks nucleate at the interface in this location.
Plastic flow of grains in this part of the HAZ was not apparent,
although some creep cavities were seen. Thick laminated oxides were
present on the 2|CrlMo fracture surfaces associated with this side of
specimens (Fig. 14a). On the other side ( here the weld metal makes
 614

135°C angle) the 2|CrlMo grains were clearly elongated, and extensive
creep damage including coalesced cavities were seen. Furthermore,
only thin oxides were present on fracture surfaces (Fig. 14b). This is
consistent with the observation that in all specimens which displayed
Type I and Type II failures little (< 20 um) or no ferritic material was
found attached to the weld metal halves of test pieces, near the crack
nucleation sites (Fig. 13).
Concurrent with the nucleation and growth of interface cracks, the
parent metal and the HAZ continue to deform. The overall creep rupture
behaviour of a dissimilar metal joint can be viewed as the result of
three independent but competing processes, resulting in three types of
fracture.
Type I - Stable growth of the interfacial crack occurs until it
reaches a critical size. The crack then becomes unstable and propagates
rapidly along the interface.
Type II - As a consequence of the growth of the interfacial cracks,
the net cross sectional area decreases thus increasing the stress on
the remaining ligament. In the as welded condition the region where
the increased stress acts will be untempered bainite. This structure
undergoes creep deformation at increasing stress levels (as the crack
continues to grow). At some stage the reduced cross sectional area is
unable to sustain the load and failure occurs.
Type III - Development of the necking instabilities in the weakest
component is determined by its creep behaviour. If a neck can develop
in the parent metal before either of the above two processes have
reached critical stages then the failure takes place in the low alloy
steel.
The third failure mode is seen in the 2|CrlMo/Inconel 182/AISI 316
specimen at 125 MPa (Fig. 12) but the specimen also showed a small
crack nucleated at the interface (Fig. 15). Mellor (6) also observed
interfacial cracks in a 2|CrlMo/Inco Weld 'A'/Esshete 1250 samples,
which failed in the ferritic steel while Summers and Ham (9) found that
in some tests at moderately high stresses, necks were developing in
the low alloy component even though the final failure sites were adjacent
to the interfaces. These results clearly demonstrate the competitive
nature of failures in the vicinity of interface and those in the parent
metal.
Published literature classifies all failures near interfaces as
one type - brittle interface fracture. However, two different modes
have been identified in the present investigation.
Instances where the entire fracture path follow the weld bead
profiles, Type I, are observed in the 2jCrlMo/AISI 316/AISI 316 welds
tested at 100 to 135 MPa. The fact that this fracture mode was not seen
in the welds made with the nickel based filler is due to the slower
rates of interfacial crack growth in these welds compared to those in
the ferritic/austenitic combination. All welds made with Inconel 182
displayed a Type II at £115 MPa, due to the build up of damage in the
bainite before the interfacial crack reached a critical size. Type II
fractures were also observed in the welds made with AISI 316 at
stresses s90 MPa (Table II).
 615

The ductilities of cross weld specimens tested at >100 MPa were

significantly higher than those tested at £100 MPa, and depended upon
the stress (Fig. lib). This is due to a change in creep deformation
mode with decreasing stress. At high stresses where the stress
exponent in the equation e = Aa11 was equal to 9.87, matrix creep
makes a particularly large contribution to the creep strain (Fig. 24).
At lower stresses grain boundary processes giving rise to cavitational
damage become more important.

Mo AS RECEIVED
Mr INCONE. ' 3 ' AS "ELDED
Mo aiSI J(6 AS WELDED

FIG 75 DEPENDENCE OF THE PAflAMETEH t ), ON RUPTURE TIDE FOR THE TRANSITION JOINT!
ANCI~THE~F£RRmC STEEL

STRESS MP.i

FIG H MINIMUM CREEP RATE AS A FUNCTION OF STRESS

The creep crack growth results also demonstrate inherent

weaknesses of dissimilar metal welds and highlight significant
differences in the cracking characteristics of the two combinations.
At a reference stress of 45 MPa the 2|CrlMo/Inconel 182 failed in
^4000h, and the 2iCrlMo/AISI 316 in -v200h, as compared to the
uniaxial rupture times of >20,000h and 12,754h respectively. The
two welds exhibited quite different initiation and growth behaviour.
For the 2lCrlMo/Inconel 182 weld the incubation period for crack
initiation was ~86Oh whereas for the 2lCrlMo/AISI 316 weld no
incubation period was observed. Furthermore, the rates of crack
growth was much higher in the latter combination.
Bowing of crack front in the two crack growth -tests indicates
plane stress states at specimen surfaces and plane strain conditions
in the centre. It would therefore appear that under plane stress
conditions the 2|CrlMo/Inconel 182 welds is creep brittle and forms
a single interfacial crack (Fig. 23a) but under plane strain situations
it is creep ductile resulting in widespread damage in the vicinity of
the crack (Fig. 23b). The other weld, however, is creep brittle and
 616

consequently displays extensive damage, but confined within 1 grain

from the interface ahead of the crack tip.
Cavities ahead of the crack tip in the 2|CrlMo/AISI 316 weld were
associated with MnS particles. The role of this sulphide in cavitational
process has been discussed before (20). i n the present case, the HAZ in
both pads must be equally susceptible to cavitational damage due to MnS.
This cavitation process cannot, therefore, explain the observed
differences.
It is instructive to extrapolate the creep rupture data to longer
times. Many materials are found to obey the empirical Monkman-Grant
relationship given as: istf = Constant. If the dimensionless parameter
£stf is plotted against log tf (Fig. 25) it becomes apparent that this
parameter is not a constant for the two dissimilar metal welds. At
stresses of $100 MPa, estf is a function of time, i.e. estf = ej + m^
log tf. Values of the constants e^ and m^ for the two welds are listed
in Table III.

TABLE III: Creep Rupture Strengths of As Welded Transition Joints at 58O°C in MPa
e = 8.911 x 10~l3o3'534 for o « lOOMPa
Rupture lime (tf ) in h.

Weld Parameter £ t. 3xlO 3

5xlO 3 io4 3x10" 5x1 105

2{CrlMo/ £ tf = 2 .469xlO~2 - 4.211xlO~3 logt. 72.5 61.1 48 1 32 .4 26. 7 20 .2

AISI 316/
AISI 316
£ t, =
S f
9 .783 x 10~ 3 72.0 62.3 51 2 37 .5 32. 5 26 .7

2{CrlMo/ £ t = 3 .52xl0"2 - 5.0a7xl0"3 logt. 84.9 72.1 57 6 40 .2 33.8 26 .7

Inconel 182/
Vf=
AISI 316 1.656 x 10 83.5 72.3 59 4 43 c 37. 7 31 .0

Further, a two slope model is found to describe the relationship

between the minimum creep rate and the stress (Fig. 24). The transition
in the value of n in the relationship, £s = Aa n , takes place at ^100 MPa.
Assuming that the value of A (= 8.911 x 10-13) anc j n (= 3.534) remain
valid for all stresses $100 MPa, it is possible to calculate rupture
strengths from the above two equations, i.e. e^ + mi l°g tf = A0 n tf.
Extrapolated rupture strengths for the two joints calculated on this
basis are also given in Table III and are shown in Fig. lla. It should
be noted that the 105h rupture strengths for the 2iCrlMo/Inconel 182
and the 2|CrlMo/AISI 316 welds are 27 MPa and 20 MPa respectively, as
compared to the ISO mean value of 47 MPa for the 2|CrlMo parent metal.
(Assuming constant values of estf at £100 MPa the respective rupture
strengths would be 31 MPa and 2/ MPa.) (Table III).
The 10^h rupture strength of the two welds are therefore only
about half the rupture strength of the parent 25CrlMo. This is
important in the context of transition joints in superheater tubings
since these are designed on the basis of the hoop stress which is
approximately twice the axial stress. From these results therefore,
 G17

premature failures in tubular transition joints can be expected almost

entirely on the basis of creep alone. Additional contributions irorn
thermal transients, bending stresses and overheating will exacerbate
the situation and reduce the time to failure.
A comparison of the results from this investigation with published
rupture data at 58O°C is made in Fig. 26. This shows significant
differen es in the rupture behaviour of different types of joints im!
probable reasons lor these are discussed below.

.i.e. * , i»co. ! l 0 . u.i.aio JK1

ON AS WELDED S A I T L 1 1 »«D i n n v * R l C u ! lHDlV(DUA t COUPON! HT^

^ C t . J S OF .'.DINc U^VE; OI*U'.E<1 3* ' H :

>*NS!T'iW ;C!NH « ' OiMERENI '."RESSE1

(i) Carbon depleted zones:

Mo evidence of carbon denuded regions, massive grain growth or
drastic softenirg adjacent to the interface was observed in the
present welds. Furthermore, the published data show that brittle
interface failures take place in all combinations of transition joint.1
including those welded with nickel based filler, where carbon denuded
zones are not expected although carbon migrates across the interface.
Hence, carbon depletion, per se, cannot be considered solely
responsible for either the brittle interface failures nor for the
differences in the rupture strengths of the two welds investigated.
(ii) Differences in coefficients of thermal expansion:
Mismatch in coefficients of thermal expansions of the ferritic
steel and weld metals is of the order of 4.5 x 10~^/°K for the AISI
316 weld, 2 x 10~6/c'K for Tnconel 182 and 0.2 x 10~ 6 /°K for Inco
Weld 'A'. This is expected to generate different levels of residual
stresses near interfaces at the test temperature and contribute to
the differences observed. Tentative support for this is provided
by the data that the 2sCrlMo/Inco Weld 'A' may be marginally stronger
to the 2|CrlMo/Inconel 182 in terms of rupture strengths and the life
fractions (with respect to the 2|CrlMo) realised( g ) (Figs. 26 and 2 7 ) .
It should be noted that the low alloy steels used in both joints
possessed similar rupture and creep strengths and rupture ductility
 618

(Figs. 24 and 26). Furthermore the rupture strengths (and presumably

creep strengths) of the two nickel based fillers are also similar and
both alloys exhibit rupture ductility of <^5% (21, 22). However, the
use of a weld metal with a coefficient of thermal expansion similar to
that of the ferritic steel will transfer the high mismatch at the weld
metal/austenite interface, and may lead to cracking in this region,
(iii) Strength and ductility of the ferritic steel:
The creep properties of the ferritic steel is considered to be
important. The 2{CrlMo used in the present investigation is weaker in
creep and in rupture and is more ductile than that used by Summers and
Ham (9) (Figs. 24 and 26). This results in significantly larger life
fraction for the 2|CrlMo/Inconel 182 welds studied here as compared to
the published data (Fig. 27). The observed difference could not be
explained on the basis of the interfacial angle because of the following.
It has been found that the 2sCrlMo/Inco Weld 'A'/Esshete 1250 specimens
with different interfacial orientation ranging from 45° to 90° have
similar rupture strengths (5, 23). It is considered unlikely that in
this respect the 2|CrlMo/Inconel 182/AISI 316 combination would behave
differently. Thus a weak ferritic component appears to improve
rupture lives of transition joints in relation to the rupture lives
of the parent metal. Further, Zemzin (1) observed that the transition
from ductile parent metal failure to brittle interface failure took
plact. at longer times when the ferritic component (12CrMoVW) had been
softened prior to welding. He also observed that a joint between the
weak CrMoV (Cast 1) and 15Cr25Ni6Mo shows the transition to occur at
200h at 550°C and 150h at 600°C but all the specimens from the same
combination made using the stronger ferritic steel (Cast 2) failed
along the interface at 58O°C.
(iv) Properties of the HAZ:
The significance of the strength and the ductility of the HAZ has
important bearings on the effect of post weld heat treatments, and will
be discussed in a later report dealing with the creep rupture behaviour
of transition joints after heat treatment.
(v) Creep behaviour of the weld metal:
According to the mechanism proposed the cre?p behaviour of the weld
metal is expected to be an important parameter. A weld metal with creep
strength and ductility similar to the HAZ will accumulate creep strain
at a rate similar to the HAZ and thus minimise the stresses responsible
for the nucleation and initial growth of interface cracks.
A comparison of the data for as welded 2|CrlMo/Inco Weld 'A' and
2|CrlMo/50Ni50Cr joints show that (i) the parent metals in both cases
have similar creep strengths (Figs. 24 and 26) and ductilities,
(ii) the coefficients of thermal expansion mismatch are similar for
both welds and (iii) both weld metals exhibit low creep ductility.
Extrapolation of the available data on the properties of the two weld
metals show that the creep and rupture strengths (24, 25) of the
5ONi-5OCr weld metal is likely to be similar to that of the ferritic
steel whereas the Inco Weld 'A' is expected to be much higher (21).
It is, therefore, reasonable to infer that the high values of life
fraction observed is due to the lower creep and rupture strengths of
the weld metal. It should be emphasised that at 58 MPa this joint
 619

survived for >49,219h in the stress relieved condition and thus

exceeded the expected ISO mean life of approximately 45,OOOh for
2|CrlMo and that to the best of the author's knowledge this is the only
combination to have done so at such low stresses.
(vi) Interface structure;
Differences in microstructures (e.g. interface precipitation,
the martensite band) and microanalysis profile (indicating the extent
of dilution) between the two joints could possible have resulted in the
difference in properties observed. However, no particular fearure could
be correlated with the properties.
(vii) Role of the oxide notch:
It has been argued th?t oxide notches initiate cracking of
transition joints during service (26) and in laboratory tests (27, 28).
However, it should be remembered that Mellor (6) observed interfacial
cracks on specimens tested in vacuum. This and the observations
described above make it more likely that oxides form on cracked surfaces
and fill notches if localised stresses adjacent to interfaces can
initiate cracking. But if the stresses do not form cracks at interfaces
than a step of oxide is seen due to preferential oxidation of the
ferritic steel.

CONCLUSIONS

1. At 580 C, transition joints are inherently weaker in rupture

as compared to the ferritic steel and at <100 MPa exhibit ductilities

2. Dissimilar metal welds made vith nickel based fillers are

superior to those fabricated with austenitic fillers with respect to
rupture strength, rupture ductility and crack propagation behaviour.
3. The 2sCrlMo/Inconel 182 interface is more resistant to
initiation as well as propagation of cracks in comparison to the
ferritic/austenitic combination.
4. Crack propagation in these welds is not controlled by thf
reference stress.
5. The 2£CrlMo/Inconel 182 interface is creep ductile under
plane strain conditions but brittle in plane stress situations. The
weld fabricated with AISI 316 is brittle under both plane strain and
plane stress coaditions.
6. At 580°C and stresses <100 MPa, rupture lives of the
2lCrlMo/Inconel 182 and the 2^CrlMo/AISI 316 interfaces are given by:
3.52 - 0.509 log t f = 8.911 x 10" 1 1 a3.534 t f
and 2.469 - 0.421 log tf = 8.911 x lO" 11 a3.534 t f respectively.
These questions can be used to calculate long term rupture strengths
of the two welds.
7. Three distinct and independent types of failure have been
identified namely Type I where entire fracture paths is adjacent and
parallel to interfaces, Type TI where brittle failures occur partly
along interfaces and partly through the HAZ, and Type III where ductile
fracture occurs in the ferritic component.
 620

8. A mechanism for creep deformation of dissimilar metal joints

has been put forward. It successfully explains premature failures and
the three types of fracture. It also accounts for the generally reported
observation that dissimilar metal failures undergo a transition from the
ductile parent metal fracture to brittle interface cracking as the
stresses are lowered.
9. The importance of the relative strengths and ductilities of
the ferritic steel, the HAZ and the weld metal in determining the rupture
lives and the fracture modes of transition joint is discussed. Another
important parameter is considered to be the mismatch of the coefficients
of thermal expansion across the interface.
10. Two types of oxide morphologies have been observed in the
vicinity of interfaces. Oxide filled interfacial notch form under the
influence of external stresses. If no external load is applied then
the preferential oxidation of the ferritic steel creates a step at the
interface.

REFERENCES

1. V.N. Zemzin, The long time strengths of welded joints between

austenitic steels and pearlitio or chromium steels, Welding
Production, No. 7, 1 (1961).
2. R.T. Hobby, A survey of the stress rupture properties of austerdt-
ferritic joints, The English Electric Co. Report No. W/LGU 136
(1966).
3. J. Tanaka and T. Hano, A study of dissimilar >v-_ tal welding
between Cr-'Ao and austenitie stainless steels, Nippon Kokan
Technical Report No. 44, 428 (1968).
4. Yu.N. Gotalskii and V.V. Snisar, Certain special features of
the failure of welded joints between dissimilar steels when
subjected to loads for long periods at high temperatures,
Auto. Welding, No. 7, 5 (1969).
5. A.A.P. Summers, Drax - dissimilar- weld thermal cycling and stress
rupture tests, Babcock & Wilcox Report No. 2/71/685 (1971).
6. H.G. Mellor, Mechanical properties of a 2liCrlMo-Inco Weld 'A'-
Esshete 1250 welds, CERL Note No. RD/L/N 153/72 (1972).
7. J. Barford and K.S. Probert, Interfacial effects in dissimilar
steel joints, Proc. Conf. on Welding Research Related to Power
Plant, Paper 30, 412 (1972).
8. W.K.C. Jones, The effect of heat treatment on the miorostructure
and properties of a 2CrMo joint welded with a Ni-based weld
metal, MEL Report No. R/M/R 165 (1974).
9. A.A.P. Summers and W.M. Ham, The Welding Institute National
Seminar NS78/1 on Welding Dissimilar Metals, Leicester, UK
(June 1981).
10. R.D. Nicholson, The creep rupture properties of austenitic and
nickel-based transition joints, MEL Note No. RD/M/1176N81
(1981).
 621

11. R.I.. Jackson, Discussion on cyclic heating test of main steam

pining joints between ferritic and au.sicnit.ia steels, Sewaren
generating station by H. Weisberg, Trans. ASME, ]l_, 643 (1949).
12. Yu.N. Gotalskii, The problem of welding dissimilar steels in
structures i.ntcridcd for1 use for long periods at h-'gh temperatures,
A u t o . Welding, No. 1 2 , 34 ( 1 9 6 4 ) .
13. A.A.P. Summers and W.M. Ham, Dissimilar weld: carbon mrigrazior.
studies. Pswi (b) - The effect of carbon migration en the
stress rupture properties of %kCrlh'c/i>Cr/?Crly.o manual and
>-%!'!rl;-'o/iiCrlMo flash weld comr iuations, Babcock & Wilcox
Report No. 2 / 7 5 / 7 6 9 ( 1 9 7 7 ) .
14. D.J.F. Eving and C.E. Richards, The yield point loads of singl.;-
noznked pin-loaded Unsile i-trips, CERL Note No. RD/L/N 237/72
(1973).
15. J.R. Haigh and C.E. Richards, Yield point loads arid compliance
functions of fracture mechanics specimens, CERL Memo No.
RD/L/M 461 "(1974).
16. X. F. Eaton and B.A. Glossop, ~<,v welding of dissimilar jreep
resisting femtie steels, Conf. on Welding Dissimilar Metals,
Met. Const*. Br. Weld. J . , Paper 9, ]_, 6 (1969).
17. D. Wood, Discuss:o>: Session Z - Transition joints for high
1-c»pcra:ure service, M e t . C o n s t . B r . W e l d . J . , 1_', 134 ( 1 9 6 9 ) .
18. R . J . Cray, J . F . King, J.M. L e i t n a k e r and C M . S l a u g h t e r ,
r'.ramination ~f a failed transition weld joint and the associated
: asc metals, Micros t r u c t u r a l S c . , 5_, 115 ( 1 9 7 7 ) .
19. R.I). Nicholson, Inter facial structures in nickel-b'O.sed transitien
joints after long term service, MEL Note No. RD/M/N 1131 (1980).
20. B . J . Cane and C.J. Middle ton, Intcrgranutar creep cavity form ition
in low alloy cainitic steels, CERL Note No. RD/L/N 195/80 ( 1 9 8 1 ) .
21. A.A.P. Summers and W.M. Ham, Assessment of f>0CrS0Ni alloys and Inco
Weld 'A ' weld m-. tal from long term stress rupture and
embrittlement tests, Babcock & Wilcox Report No. 2 / 7 8 / 8 2 5 ( 1 9 7 9 ) .
22. A.N.R. Hunter and R.M. W e s t l a n d , The elevated temperature tensile
and creer> rupture properties of 2%CrlMo} Inconel 182 and
Metrode I4.7i> P-JnNb weld -:ctals, Babcock Power L t d . , Report t o
be published.
23. K . S . P r o b e r t , Private Communication (1976).
24. R. Kossowsky, Creep beliaviour of Ni-Cr lamellar eutectic alloy,
Met. T r a n s . , 1_, 1909 (1970).
25. International Nickel Ltd., Summary of experimental data alloy
r-i-657 (1971).
26. R.L. Klueh and J.F. King, Austenitic-ferritic weld joint failures,
ORNL Report No. 5637 (1981).
27. O.R. Carpenter, N.C. Jessen, J.L. Oberg and R.D. Wylie, Some
considerations in the joining of dissimilar metals for high
temperature high pressure service, Proc. ASTM, 50, 809 (1950).
28. J.T. Tucker and F. Eberle, Development of a ferritie-austenitic
weld joint for steam plant applications, Weld. J., p. 529-s
(1956).
 DAMAGE ACCUMULATION AND FAILURE CRITERIA IN
'CREEP BRITTLE' FERRITIC WELDMENT STRUCTURES

B. J. Cane

Central Electricity Generating Board,

Technical Planning & Research Division,
Central Electricity Research Laboratories,
Kelvin Avenue, Leatherhead, Surrey, U.K.

ABSTRACT

Life expectancy of weldments in plant undergoing elevated temperature

service is commonly dictated by the susceptibility of the associated
thermally induced structural states to low ductility intergranular fracture
This paper reports on the criteria determining the kinetics of creep
cavitation leading to such failures in heat affected zone (HAZ) structures
in low alloy ferritic steels. The micromechanisms responsible for cavity
formation and growth are briefly described and the predicted cavitation
rates are considered in relation to an upper bound approach based on
accommodational or constrained kinetic control by the creep response of
the material. The upper bound approach is used as the basis of a model
to define the stress and stress state dependence of intergranular cavity
growth. Multiaxial stress rupture criteria governing rupture life and
ductility are developed on this basis. Accordingly a representative rup-
ture stress is proposed which can be applied to uniaxial data to predict
the lifetime under a known state of stress.
The predictive model is compared with results of uniaxial and multi-
axial creep and rupture tests at 565°C on thermally simulated HAZ struc-
tures in 2sCrMo and jCrMoV steels. Application of the model is illustrated
for the case of failure analysis of butt-welded tubes and remanent life
assessment of welded components by in service inspection.

INTRODUCTION

The design stress permitted by design codes for pressure vessels

623
 624

operating in the creep range is in general based simply on the rupture

strength of the base material modified by a safety factor (the exception
is the ASME Code Case N-47 developed for nuclear components). The rupture
stress for a given design lifetime (10^-1.5xlCPh for fossil plant) is
obtained from uniaxial data at the design temperature. This approach
has proved conservative for the case of parent material, where the few
failures that have occurred are attributable to excursions from design
conditions. Service experience within the CEGB reveals that premature
failure is almost invariably associated with the weldments which are not
directly considered in the codes for non-nuclear vessels - it is merely
specified that the weld properties should match those of the parent mat-
erial. In practice the presence of welds leads to variability in met-
allurgical structure in the weld metal and across the heat affected zone
(HAZ) with an associated variation in creep response. This can lead to
significant local fluctuations in stress and stress state thereby in-
fluencing the failure life of the component. Thus, although the codes
implicitly assume that the safety factor incorporates a sufficient margin
to accommodate these effects, there is a degree of uncertainty introduced
and a consequent cause for concern regarding operational integrity
assurance. There is thus a need to examine the current design approaches.
Additionally in view of the growing importance of extending the operational
life of plant, there exists a requirement for improved accuracy in
remanent life predictability.
Reference stress techniques^'-' are becoming increasingly popular in t
component integrity assessment and have been shown to yield accurate es-
timates of deformation response of high temperature structures. Yet, it
is important to recognise that they can only be confidently applied to
rupture prediction when the multiaxial stress criterion for rupture is
the same as that for deformation and the strain tolerance of the material
is sufficient to enable long range stress redistribution. In practice
these techniques are more usually applied to homogeneous structures but
not for welds which generally exhibit lower creep ductilities. Whilst in
many cases weld failure need not necessarily limit the life of plant
there is a substantial cost benefit in predicting the operational life-
time before major repair or inspection requirement. An alternative
approach for low ductility materials is to use the maximum stationary
state creep stress^. However we must still take account of the stress
state and consequently accurate life predictability will require knowledge
of the multiaxial stress rupture criterion for the material and
conditions of interest.
Tha low ductility failure mode often associated with welds is attri-
butable to the 'creep brittle' response of thermally induced microstructural
states within the weld metal and the HAZ. This arises due to an associated
high susceptibility of such structures to intergranular cavitation which
can accumulate during the weld stress relieving cycle or during elevated
temperature service exposure. Macroscopic cracking often results and
depending on component geometry and microstructure, may lead to rapid
failure or necessitate continuous monitoring or repair. Studies of
cavitational damage kinetics are thus fundamental in formulating improved
predictive schemes for plant integrity and defect assessment.
Low-alloy ferritic steels (e.g. ^CrMoV and 2|CrMo) are particularly
 625

prone to intergranular cavitation in post-welded structural states. This

has frequently been associated with the existence of coarse grained
bainitic microstructures in both the weld metal and HAZ and methods of
microstructural control have been suggested on this basis .
By considering the criteria controlling the kinetic development of
intergranular cavitation leading to failure this paper proposes an
analytical route for failure prediction of 'creep brittle' weldment con-
stituents .

CREEP DAMAGE ACCUMULATION

Initial Structure

Owing to the high temperatures (>1300°C) and rapid cooling rates ex-
perienced during the weld thermal cycle for low alloy ferri tic .steels the
resultant HAZ structure adjacent to the fusion boundary general!y consists
of coarse grained bainite. In this condition and after standard post-
weld heat treatment the structure possesses a very high strength con-
ferred by a fine dispersion of carbide precipitates and a high dislocation
density within the prior austenite grains. An example of the micro-
structure obtained is illustrated for 2|CrMo steel in the simulated
tempered HAZ condition in Fig. l(a). A significant feature is the exis-
tence of precipitate free zones (PFZ's) at the prior austenite grain boundaries.
The relative weakness of these PFZ's enables shear deformation to con-
centrate in regions adjacent to the prior austenite grain boundaries.

-'•'.•t- ''•

Fig. 1. (a) Tempered HAZ microstructure in 2|CrMo steel, (X600).

(b) Cavity formation at sulphide inclusions on prior aus-
tenite grain facets in |CrMoV steel (X3000).
 62 6

Cavity Formation

During creep, cavities nucleate and grow at the prior austenite grain
boundaries predominantly on those boundaries orientated normal to the maxi-
mum principal stress ( a ^ ) . The number of cavities nucleated is known to be
dependent on the purity of the material" and the applied stress or stress
state^. The former can determine the cavitational susceptibility by con-
trolling the number of potential cavity nucleation sites while the latter
can facilitate cavity stability. The formation and subsequent growth of
cavities is also sensitive to thermal history, particularly the degree of
temper. This has been a common observation and has been attributed to
'softening' of the matrix and simultaneous disappearance of the PFZ's.
Alternatively it is possible that the tempering treatment alters the grain
boundary chemistry such that otherwise potential nucleation sites become
less effective.
Recent studies o > ' > ° on welded ferritic structures and weld-simulated
material subjected to high temperature austenitizing cycles (>1300°C) have
suggested that cavities form at non-wetting inclusion particles. These
have been identified as sulphide particles . (Fig. l(b)). It has further
been shown that the cavity population, N, is determined by the maximum
principal stress, a^, rather than the shear stress component and for
2|CrMo steel at 565°C has the form:

N = ^o\ ... (1)

where k^ and v(=l) are material constants. Under these circumstances it

can be inferred that whilst strain induced cavity nucleation can still
occur throughout the creep life the presence of non-wetting inclusions
ensures that the majority of cavities are stabilized during the initial
stages. The work suggests that the inclusion particles act as pre-existing
cavity nuclei. The process is consistant with classical nucleation whereby
stability is brought about by a o^ induced vacancy supersaturation when
a o > 2y/o-^ where y is the surface energy and a 0 is the nucleus size. The
observations can also be explained in terms of a plasticity or continuum
nucleation process controlled by the hydrostatic stress, OJJ . In the
presence of PFZs and for high PFZ/matrix strength differentials charac-
teristic of coarse grained bainitic structures a 'contact adhesive effect1
or local stress triaxiality exists at grain boundaries normal to a^ such
that the local ajj = 0]_. The critical size for cavity stability is thus
equivalent to above, however, unlike the classical vacancy process, the
critical size will increase as the matrix/PFZ strength differential de-
creases and the local stress triaxiality is lost'. Under these circum-
stances the population can, therefore be related to the microstructural
state and associated creep strength. Whatever the process responsible
for cavity stability the o^ dependence of the cavity population
(equation 1) can be rationalised in terms of a model in which the size
distribution of cavity nuclei may be related to the size distribution of
incoherent particles residing in the grain boundaries. Equation 1 con-
sequently represents a general form in which the kjj ancj v values will be
determined by the form of the cumulative size distribution curve''.
 627

Cavity Growth

In coarse grained bainitic structures the growth of cavities to pro-

duce discrete grain boundary cracks is the dominant stage in the creep
damage process. In homogeneous coarse grain size structures the linkage
ofboundary cracks and final fracture propagation involves a relatively small
fraction of the overall life. Consequently, to a first approximation for
a fixed cavity population, the kinetics of cavity growth govern the rup-
ture lifetime.

The basic micromechanisms responsible for cavity growth in ferritic

steels are not entirely clear. For a given range of stresses, temper-
ature and times it is likely that more than one basic mechanism pertains .
The two distinct micromechanisms are based on (i) diffusion control (ii)
continuum or plasticity control. These are considered briefly below.
Assessments of cavity growth in ;CrMoV and 2;CnMo steels at service
temperatures*-® have indicated that the kinetics are considerably slower
than those predicted by classical grain boundary diffusion controlled
models ' . It is arguable, however, that the reduced growth rates are
consistent with a suppressed diffusion controlled process whereby the
grain boundaries behave as imperfect vacancy sources. This will be the
case if vacancies arecreated at grain boundary dislocations and the move-
ment of the dislocations is impeded by grain boundary precipitates or
solute interaction effects. Under these circumstances for spherical
growth the volume growth rate of cavities may be expressed

vD = KD(a1-2y/a - % ) ... (2)

where a is the cavity radius, K Q is a parameter dependent upon the grain

boundary diffusion coefficient, the cavity size and spacing and a o is a
threshold stress derived from the interaction of grain boundary dis-
locations and precipitates. It is assumed in equation 2 that the grains
remain essentially rigid. If this were not the case then power law creep
within the grains would lead to an increased diffusion controlled growth
rate^ . It is also noteworthy that the above situation will only pertain
under conditions where dislocation climb in the grain boundary is rate
controlling. If, on ';he other hand climb in the boundary is limited by
the availability of dislocation sources then the basic mechanism will be
governed by the local strain rate.
In contrast to the principal stress controlled diffusional
mechanism cavities can grow under the action of the hydrostatic stress,
a^j, by the continuum growth process. For homogeneous flow conditions
the volume growth rate may be expressed

v = a v e sinh (@a /a) ... (3)

where v is the_cavity volume, a and g are functions of the creep ratg -

stress index, o is the effective or Von Mises equivalent stress and E is
the corresponding effective strain rate. Equation 3 predicts very low
growth rates at representative engineering strain rates and has con-
 628

sequently been associated with high strain rate tests. It has recently
been shown however that the presence of PFZ's at the grain boundaries can
lead to enhanced continuum growth rates owing to the locally induced
stress triaxiality on boundaries orientated normal to the maximum stress
direction^. For HAZ structures where large strength differentials exist
between matrix and PFZ the term o~ft/o in equation 3 may be replaced by the
local stress triaxially given by

1 /p -
a Jo = <\>'(aJa) ... (A)
n I
where <j> is the ratio of PFZ creep rate to matrix creep rate and p is the
creep rate - stress exponent for PFZ material. Taking cj> values
typically of the order 10-10 and p = 5 then from equations 3 and 4 the
continuum growth rate is increased by a factor around 10" - 10 .
It may be postulated that either or both of the above basic mechan-
isms can govern the cavity growth process. To a first approximation the
predicted growth rates VD and v p are additive so that we can describe the
overall growth rate v by the general equation

Accommodation of Cavity Growth

In the treatment of the cavity growth mechanisms described above it

has been implicitly assumed that the dilation due to cavity growth can be
readily accommodated by concurrent strain occurring within the polycrystal.
This may well be a valid assumption for pure metals and high strain rates,
but may not be the case in the present highly creep resistant HAZ structures
at relevant stress levels. If conditions are such that the cavitated
regions dilate at a rate (due to equation 5) faster than the surrounding
matrix can creep then cavity growth becomes constrained to follow the
creep deformation kinetics and the local stress near the cavitated regions
becomes less than the applied stress^"*»^•->. The constrained growth rate
may be given by the general form

v = iTC2de.. = (37r/2)c2dBan~1S. . ... (6)

c ij ij
where c is the cavity half spacing, d is the grain size, £^; is the
strain rate normal to the cavitating boundary, S-- is the deviatoric
stress in the same direction and B, n are creep parameters defined by the
material creep law, e = Bo . If conditions are such that dilation strain
rates are greater than the creep rate (as is likely for ferritic HAZ struc-
tures) then compatibility requires that stresses must redistribute and in
the limit the normal stress will approach zero at cavitated grain bound-
aries. We can thus obtain an upper bound on the constrained growth rate,
which for the uniaxial case is
vc = Trc2dB[a/(l-A)]n ... (7)
 629

where A is the area fraction of cavitating grain boundaries.

The cavity growth processes described by equations 2-7 are
compared in Fig. 2 using 565°C data from 2jCrMo steel in simulated
tempered HAZ condition. It is assumed that o o = 0 , the grain boundary
diffusion coefficient is given by that for pure a -Fe, (2.1x10 "1-um*"h
the cavity spacing c = 2um and the PFZ/matrix creep rate factor, i = 10.3.
The growth rates predicted by each model are shown as a function of
applied uniaxial stress, a and cavity radius, a.

io o; o< oi

DIFFUSION CONTROLLED
GROWTH

_1 I
wo 40a

Fig. 2: Control Regimes for Cavity Growth in 2jCrMo Steel

HAZ Structure at 565°C

Initial growth is diffusion controlled. As growth continues, at low

stresses, the growth becomes constrained. At higher stresses the
control mechanism undergoes a gradual transition from diffusion-
controlled to enhanced continuum growth as creep continues. During the
final stages constrained growth again becomes operative. It may be
concluded from Fig. 2 that although the two unconstrained mechanisms
(equations 2-5) and their variations may well occur, subject to test
conditions the constrained growth process will always provide an upper
bound on the damage accumulation rate. It will also be favoured at low
stresses and for large cavity populations (i.e. when c is small) which
is generally characteristic of coarse grained 'creep brittle' structures.

CREEP FRACTURE MODEL

An upper bound cavity growth rate given by constrained growth

concept provides us with a basis for a predictive failure model, which
will be conservative in the event of kinecic control by the unconstrained
mechanisms.

Since the majority of cavities nucleate in the initial stages of

 630

creep we can assume the cavity spacing is constant. It is also assumed

that the creep rate remains constant throughout the life which is not
unreasonable in view of the 'creep brittle' nature of HAZ material. On
this basis we can integrate equation 7 to yield the uniaxial rupture
life, occurring when a = c:

tR = 4c(l-A) n / 3dBan ... (8)

Taking c = 1/2N2 then from equation 1 we have

tR - [2(l-A)n/3k| d B ]/a n + v / 2 ... (9)

tn = (2/3k 2 d)/a e ... (10)

R N I m
where £m is the minimum creep rate. Thus for small values of v ^he
Monkman-Grant relationship, emt£ - const, is shown to hold. Neglecting
the primary strain and any tertiary creep a lower bound rupture
ductility can be defin2d by
eR = 4c/3d = (2/3k^d)^0 V / 2 ... (11)

The ductility is thus proportional to the cavity spacing described by an

inverse function cF. stress having a stress sensitivity as defined by v/2.
Under multiaxial stressing conditions we must take the general form
of the constrained growth equation (equation 6) so that:
n —1 l / / n —1

/
a S£. = (4/9dBk|)/a 1 v/ 0 Si.
(12)
By assuming that stress redistribution from cavitating to non-
cavitating regions occurs predominantly in the a^ direction we can define
the effective stress and deviatoric stress as:

= o(A) = (1/2) 2 (o'-a.,) +(0o-a,) +(o,,-01) ... (13)

S^ = S1(A) = (l/3)(2oj-02-03) ... (14)

aj = Oj/(1-A) ... (15)

In the presence of cavitation, therefore, the effective or Von-Mises

equivalent strain rate becomes stress state sensitive.
In order to predict the rupture life under multiaxial stress
conditions from uniaxial data we can now obtain a representative rupture
stress (cij^jp) by equating equations 9 and 12, i.e. :
r, - fi/T\ l/mc i «s n / m (m-n) /m _(n-l)/m c 1/m ... (16)
RUP ~ ^'2> (1-A) a1 0 S1
 631

where m = n + v/2 and c, Sj are given by equations 13-15.

If it is assumed that constrained growth occurs on all grain
boundaries orientated normal to c>\ then A ^ 0.33. Taking v = 1
(typically the case for 2{CrMo and :lCr>toV steels at 565°C) Fig. 3
shows ^R[jp derived from equation 16 as a function of the creep-stress
index, n for various applied stress states.

IRIAXIAU'
STRESS STATE <T , ' f f

NOTC1

PRESSURIZED <T (T I-
• rilBE SURFACE

oal—
i

Fig. 3: Variation in Representative Rupture Stress (

with Creep Rate-stress index (n) for Various
Stress States.

The rupture stress is given in normalised forn by dividing by the

effective stress in the absence of cavitation (o Q ). For a given value
of n the representative rupture stress increases as the stress
triaxially, (L'I/O ) , increases from torsion to the notch situation,
For example taking n = 4 then for a given o o , L1 RUP is a factor of about
2 greater in the Bridgman notch case than in torsion and accordingly
the rupture life in torsion will be longer by a factor 2 m where
m(=n+v/2) is the slope of the uniaxial stress rupture plot, i.e.
fc =
R C/o111. It is also noteworthy from Fig. 3 that for a given stress
state and n > 3, ajj^'p i s n o t strongly sensitive to the n value chosen
which makes it a useful engineering parameter.

COMPARISON WITH EXPERIMENT

The above model will now be considered in the light of uniaxial and
multiaxial creep damage and rupture data obtained on 2jCrMo and jCrMoV
steels in simulated coarse grained HAZ condition, (grain size = 150um)
Fig. 4 shows the stress dependence of rupture ductility obtained
experimentally for the two steels. Lower bound predictions are obtained
from equation 11 using mean cavity spacings (c) measured on the fracture
surface.
 632

CONSTRAINED EXPERIMENTAL

PREDICTED G R l = 4c/3di

ll = 150 jjm
c values given in

0I
S

Fig. 4: Stress Dependence of Measured and Predicted Uniaxial

Ductility (eR) for 2|CrMo and |CrMoV Steel HAZ
Structures

For the lower ductility jCrMoV steel th.2 stress dependence of £„ is

consistent with the constrained cavity growth model, i.e. the ductility
reduces with increasing stress due to higher cavity populations (lower
cavity spacings) at high stresses. For 2|CrMo steel a similar situation
pertains at low stress levels but at higher stresses the ductility
increases in a manner not directly related to the cavity spacing. This
may be rationalised in terms of a transition from constrained growth to
unconstrained diffusion controlled or continuum growth with increasing
stress. Under these conditions the cavity growth rate decreases
relative to the strain rate and so the strain to failure increases.
(Fig. 2 ) . The data for the two steels thus emphasises the fact that the
constrained growth model is favoured by small cavity spacings.
Fig. 5 illustrates the stress state dependence of the cavity growth
rate for 2|CrMo steel. Results are given for uniaxial tension, torsion
and Bridgman notch tests. Model predictions of growth rates for notch
and torsion situations using tension data are also shown. Good
agreement is found with the multiaxial data. Fig. 6 shows creep curves
obtained in uniaxial tension tests and torsion tests on thin-wslled
tube of jCrMoV steel at a given effective stress o o . In addition to the
longer life and greater ductility obtained in torsion, the data
indicates a strong stress state dependence of strain rate as predicted
by the model. The effect of stress state on life is greater than that on
r.train rate owing to the a^ dependence of the cavity population
(equation 1 ) . For a given effective stress, the a^ value is a factor of
/3 lower in tension than in torsion and consequently the cavity spacing
is greater in torsion.
The applicability of the model as a means of life assessment under
multiaxial stress is summarised in Table 1 where experimental and
predicted values of the representative rupture stress are given for the
various stress states tested. Excellent agreement is found for |CrMoV
steel.
 633

NOTCH 4,

/ TENSION
EXPERIMENTAL • TENSION
[1ATA a MOTCH
.1 TORSION /

PREDICTED
~ FROM TENSION
DATA

100 200
a;

•'ig. 5: Variation in Experimental and Predicted Cavity Growth

Rates with Applied Effective Stress (a )

Fig. 6: Creep Curves in Uniaxial Tension and Torsion of Thin-

Walled Tubes in ^CrMoV HAZ Structure.

For 2jCrMo steel the agreement is better at the lower applied stresses
where the constrained model is more likely to pertain throughout the
creep life.
 634

TABLE 1: Experimental and Predicted a R U p in |CrMoV and

2iCrMo Steels under Multiaxial Stresses at 565°C

., . ,, o Exptal. a_IT_ Predicted an,TTI

K
Material/ o n RUP RUP
stress state (MPa) (Wa) (MPa)
ICrMoV
Torsion 250 8.5 205 205
200 8.5 165 164

2ICrMo
Shear 147 4 107 114
124 4 98 97
108 4 81 84
Torsion 178 12 138 146
Notch 124 4 184 175
178 12 255 240

NB. The shear test gives the same stress state as torsion
and is performed by means of a direct shearing assembly
fitted to a uniaxial test machine-^. The notch test
(bar radius/notch throat radius = 1 . 9 ; notch radius/
notch throat radius = 0.5) gives a reasonably constant
stationary state triaxial stress situation at the centre
of the notch throat. Changes in notch geometry and
stress redistribution during creep can render the test
less definitive than in the case of torsion.

APPLICATION TO FAILURE ANALYSIS OF BUTT WELDED

TUBES AND PIPES

The above failure criteria can be incorporated into an analysis of

failure in welded pressure vessels containing 'creep brittle'
constituents. By virtue of their thermal simulation coarse grained HAZ
structures generally exhibit a higher creep strength than that of parent
base material. As a consequence, for cylindrical butt welds in tubular
components, the pressure stress is off loaded from parent material and
weld metal on to the HAZ. By utilizing uniaxial creep rate data for
parent pipe, weld metal and HAZ finite element computations can be
employed to calculate the stress redistribution. Such calculations
have recently been performed in analysing failure of ^CrMoV butt welded
tubes incorporating a range of weld materials.19 The extent to which
the pressure stress is redistributed is dependent on the relative creep
strengths of parent pipe, weld metal and HAZ. Fig. 7(a) shows the
computed variation in peak stationary state creep stresses in each of
the weldment constituents as a function of weld metal/parent pipe creep
rate ratio. As the latter increases the peak hoop stress (ai) increases
 635

in the HAZ and correspondingly decreases in the weld metal. However

the stress redistribution is more sensitive to the relative creep
strength of the HAZ rather than the weld metal since for matching creep
rates in the parent pipe and weld metal a significant stress enhancement
still exists in the HAZ. By way of example Fig. 7(b) illustrates the
cross weld variation in stationary state hoop and axial stress components
at the outer surface for a particular weld:parent: HAZ ratio of 10:1:0.25.
The principal stress components have been used to compute a representative
rupture stress, cr^jp in the HAZ based on multiaxial rupture data on the
simulated material.

Fig. 7: (a) Variation of peak hoop stress with weld metal/parent

metal creep rate ratio.
(b) Cross-weld stress distribution along outer surface of
tube. (after Browne et all9) .

The stationary state stress levels given in Fig. 7 were based on

material creep data in the form e =_Ba _where a is considered
independent of stress state, (i.e. a = a 0 ) . This is at variance with
the present model which requires that £ is stress state dependent. The
computed principal stresses will consequently be approximate and will
give rise to an approximate o^up value. A more rigorous analysis
incorporating a(A) and S^(A) (equations 13-15) is thus needed.
Nevertheless the approach demonstrates the principle that computed
stress components can be used to determine the variation in aRyp across
the weldment.. Failure prediction can then be made by considering each
material within the structure separately and assuming that the shortest
life predicted in any one material determines the failure position and
life of the weldment. Results of the butt-welded tube programme1^ have
demonstrated that failure usually initiates by axial cracking in the
HAZ. This has further been shown to be predictable with respect to
 636

both time and position by means of the stationary state stress analysis.
A flow diagram giving the analysis route and input data requirement is
shown in Fig. 8.

UNIAXIAL UNIAXIAL
STEADY STATE CREEP RUPTURE PARAMETERS
PARAMETERS FOR FOR W.M..HAZ, PARENT
W.M..HAZ, PARENT
Cm
B.n

CONSTITUTIVE RELATION
P - 3 R =. n " 1 c
ij " 2 'J

FINITE ELEMENT I
ANALYSIS

-1

X ELEMENTAL
POSITION
SS STATIONARY
STATE

Fig. 8: Flow Diagram for Stationary State Creep Stress Analysis

APPLICATION TO COMPONENT LIFE ASSESSMENT BY

IN SERVICE' INSPECTION

Damage assessment at welds in complex geometry components operating

at high temperatures is an important area of study with respect to
remanent life estimation. The present model provides us with ar.
assessment route which may be quantified by means of currently available
metallographic techniques. Such techniques include the plastic replica
method. This is advantageous since it can be applied to welds in
critical areas of limited access where relatively large areas can be
non-destructively sampled.
Iti addition to providing quantitative information on the general
development and distribution of cavitation and cracking the replica
method also allows us to predict the local strain accumulated by-
utilizing the present constrained cavity growth model. Rearranging
equation 6 gives us the effective strain increment (<5e) corresponding
to change in cavity size (6v),

fie = (2O/3S.... )(Sv/c d) (17)

 637

Since the model proposed, in general, predicts an upper bound on the

cavity growth rate for a given stress level <Sc will represent a lower
bound effective strain increment. Successive assessments of cavity size
can thus enable us to estimate the strain accrued over a given
operational period. The measurements can be made at planned overhauls
and would involve estimates of mean cavity size and spacing together
with grain size determination. With a knowledge of the lower bound
failure strain for the particular stress state (i.e. 4c/3d; equation 11)
the assessment route can provide some confidence with regard to life
usage. Alternai ively, successive strain assessments can be translated
into an effective strain rate. Then, with some information on the local
stress state this may be modified (via equations 13-15) to give the
uniaxial strain rate which, in turn can be applied to relevant
uniaxial creep data to obtain the local stress. Given the relevant
stress rupture data we can thus predict the time to local failure.
In the case of welds sited at changes in component geometry (e.g.
nozzle intersection welds) anoi_her approach would be to utilize the
local strain rate estimation in conjunction with the stress concentration
factor (SCF) for the geometry change to estimate the strain rate in
regions remote from the weld. This is a useful measure since in many
cases welds can be repaired while the base material must remain in service
The local strain rate may alternatively be used to estimate the effective
SCF associated with the geometry. Comparison with the calculated SCF
can give an indication of secondary stresses arising either due to poor
design or severe operational conditions and thus provide information on
which to base the decision of whether to modify operational practice or
repair the weld or replace the component.

CONCLUSIONS

A creep damage model based on the concept of constrained or

accomodational intergranular cavity growth has been proposed as a
means of defining the multiaxial stress criteria for damage
development and fracture of 'creep brittle' weldment constituents.
In general the model provides an upper bound on the intergranular
cavitation rate such that life estimates will tend to be
conservative under circumstances where other cavitational control
mechanism pertain.
The model has been shown to give good agreement with uniaxial and
multiaxial creep damage and rupture data on thermally simulated HAZ
structures in 2|CrMo and ^CrMoV steels.
The model has indicated an analytical route to failure production :
(i) by utilizing a representative rupture stress (ORUP) in
association with creep stress analysis it is possible to
predict the time to fracture of the HAZ from a knowledge of
the uniaxial creep rupture properties of the weldment
constituents.
(ii) by assessing cavity sizes and spacings by non-destructive
sampling at welds during overhauls the local strain may be
estimated which in turn provides a basis for residual life
prediction.
 638

ACKNOWLEDGEMENT

This paper is published with the permission of the Central

Electricity Generating Board.

REFERENCES

1. D.J. Walters and R.D.H. Cockrcft, Stress Analysis and Failure

Criteria for High Temperature Butt Welds, Int. Inst. of Welding,
(Toronto) (1972)

2. I.W. Goodall, R.D.H. Cockcroft and E.G. Chubb, An Approximate

Description of the Creep Rupture of Structures, Int. J. Mech.
Science, 17: 351 (1975)

3. F.A. Leckie, Approximate Procedures in High Temperature Design,

Recent Developments in High Temperature Design, Inst. Mechanical
Engineers, (London) 49-57 (1977)

4. I.W. Goodall and R.D.H. Cockcroft, On Bounding the Life of

Structures Subjected to Steady Load and Operating within the Creep
Range, Int. J. Mech. Sci., 15: (1973)

5. B.L. King, C.J. Middleton and R.D. Townsend, Prevention of Heat

Affected Zone and Weld Metal Cracking through Control of Micro-
structure in 5CrMoV-2|CrMo Weldments, CEGB Report RD/L/R1919 (1975)

6. C.J. Middleton, Cavity Nucleation in Bainitic Low Alloy Creep

Resisting Steels, Metal Science, 15: 154, (1981)

7. B.J. Cane and C.J. Middleton, Intergranular Creep Cavity Formation

in Low Alloy Bainitic Steels, Metal Science, 15: 295-301 (1981)

8. N.G. Needham and T. Gladman, Intergranular Creep Cavitation in

2jCrlMo Steels, Conf. on Advances in the Physical Metallurgy and
Applications of Steel, (Liverpool) (Sept. 1981).

9. B.J. Cane, Creep Fracture of Dispersion Strengthened Low Alloy

Ferritic Steels, Acta Met. 29: 1581-1591 (1981)

10. B.J. Cane, Mechanistic Control Regimes for Intergranular Cavity

Growth in 2.25Cr-lMo Steel under Various Stresses and Stress
States, Metal Science, 15: 302-310 (1981)

11. M.V. Speight and W. Beere, Vacancy Potential and Void Growth on
Grain Boundaries, Metal Science, 9: 190 (1975)
 639

12. A. Needleman an^ J.R. Rice, Plastic Creep Flow Effects in the
Diffusive Cavitation of Grain Boundaries, Acta Met., 28:
1315-1332 (1980)

13. K. Hellan, Int. J. Mech. Science, 17: 369 (1975)

14. B.F. Dyson, Constraints on Creep Cavity Growth, Metal Science,

10: 349 (1976)

15. R. Raj and A.K. Ghosli, Stress Rupture, Met. Trans. A., 12A:
1291-1302 (1981)

16. B.F. Dyson, A.K. Verma and Z.C. Szkopiak, The Influence of Stress
State on Creep Resistance: Experiments and Modelling, Acta Met.
29: 1573-1580 (1981)

17. J. Henderson and J.D. Sneddon, J. Inst. Metals, 100: 163 (1972)

18. D.R. Hayhurst, F.A. Leckie and J.D. Henderson, Creep Stress
Redistribution in Notched Bars, Int. J. Mech. Science, 19: 147,
(197/)

19. R.J. Browne, B.J. Cane, J.D. Parker and D.J. Walters, Creep
Failure Analysis of Butt Welded Tubes, Conf. on Creep Fracture
of Engineering Materials and Structures, (Swansea) 645-650 (1981).
 DISSIMILAR WELD FAILURE ANALYSIS AND DEVELOPMENT PROGRAM

K. H. Holko, Consultant, San Diego, California 92117

and
C. C. Li, General Atomic Company San Diego, California 92138

ABSTRACT

The problem of dissimilar weld cracking and failure is examined.

This problem occurs in boiler superheater and reheater sections as well
as main steam piping. Typically, a dissimilar weld joins low-alloy
steel tubing such as Fe-2-1/4 Cr - 1 Mo to stainless steel tubing such
as 321H and 304H. Cracking and failure occur in the low-alloy steel
heat-affected zone very close to the weld interface. The 309 stainless
steel filler previously used has been replaced with nickel-base fillers
such as Inconel 132, Inconel 182, and Incoweld A. This change has
extended the time to cracking and failure, but has not solved the
problem.
To illustrate and define the problem, the metallography of damaged
and failed dissimilar welds is described. Results of mechanical tests
of dissimilar welds removed from service are presented, and factors
believed to be influential in causing damage and failure are discussed.
In addition, the importance of dissimilar weldment service history is
demonstrated, and the Dissimilar Weld Failure Analysis and Development
Program is described.

INTRODUCTION

The dissimilar metal weld (DMW) failure problem has plagued the
fossil-fired power industry since the early 1950s. The weldments of
most interest and significance for failure are tube-to-tube weldments
in the superheater and reheater sections of boilers. Main steam line
dissimilar welds external to the boilers have also been cause for con-
cern. Failures in boiler DMWs occur in both furnace and vestibule
(nonheat transfer) locations. Typically, boiler tubing has about a
6.35-cm (2-1/2-in.) o.d. and a 1.27-cm (1/2-in.) wall thickness.
Most of the emphasis in research and failure studies has been
given to boiler DMWs, because a boiler may contain hundreds to thou-
sands of DMWs in the superheater and reheater sections. It is often

641
 642

difficult to gain access to these weldments for repair, ami the

consequence of failure is frequently a forced outage. The monetary
consequence of a forced outage is severe, and the problem can be very
costly.
The dissimilar weldments of interest in this paper are between
low-alloy steel tubing, typically SA 2L3 T-22 (2-1/4 Cr - 1 Mo), and
austenitic stainless steel tubing such as SA 213 321H (Table I ) . Vari-
ous filler materials have been used over the years, ranging from aus-
tenitic stainless steel to nickel-base filler materials. There have
also been a number of induction pressure welds formed without a filler
material by solid state welding. This category of weldment will net be
described and is outside the scope of this paper.
The reason for the use of dissimilar welds is to provide a
transition from low-alloy steel tubing to stainless steel tubing with a
higher operating temperature capability. The requirement for high
boiler efficiency necessitates high steam and tube metal temperatures.
The DMW is typically placed in a region with a metal temperature of
about 427° to 593°C (800° to 1100°F). This region is a compromise
between the maximum operating temperature for the low-alloy steel tub-
ing and conservation of the more expensive stainless steel tubing.
This transition is made in the furnace. Another transition is then
made outside the furnace, in the vestibule, or nonheat transfer region,
back to the low-alloy steel material so that the connection to the low-
alloy steel header can be made with a similar weld. The internal pres-
sure in the tubing varies from about 3.5 MPa (500 psi) in the reheater
section to as high as 26.2 MPa (3800 psi) in the superheater section of
a supercritical unit.
The external surface, or o.d. , of the welds of interest are
typically in contact with coal, oil, and gas heating environments.
Both reheated and superheated steam are in contact with the internal,
or bore, side of the DMWs.
The dissimilar weld failure problem is international in scope.
Failures have occurred and research work has been conducted in the
United States, United Kingdom, Canada, Japan, and the Netherlands.1~6
This paper presents the background of the problem, provides
general and specific illustrations, and describes some of the previous
and ongoing experimental work in the United States and elsewhere to
define and solve the problem. Recommendations for future work are
made.

BACKGROUND OF DISSIMILAR WELD PROBLEM

A large body of literature, including over 200 published and

unpublished papers, has been generated on the DMW problem. This is
due, no doubt, to the seriousness of the problem as well as the length
of time it has existed. One paper7 dates back to 1935, and other early
papers provide good descriptions of the problem along with initial
^'^. More recent papers^""'10~13 describe failures and
 643

results of research programs. The reader ic referred to a '-ecent

literature review* for a more complete reference list.
In summary, sorae of the major causes of failure that researchers
have cited over the years are:

1. Differences in the coefficients of expansion between the

austenitic stainless tubing, the filler material, and the
low-alloy steel tubing.
2. Effects of carbon migration and resultant carbon depletion in
the low-alloy steel heat-affected zone (HAZ).
3. Strength mismatch between the stronger filler material and
the weaker low-alloy steel HAZ.
4. Oxide notch formation in the low-alloy steel HAZ near the
weld line.
5. Presence of backing rings at the dissimilar weld and
associated stress concentration and cracking effects.
6. Improper joint design.
7. Formation of carbide phase(s) in the low-alloy steel HAZ near
the weld line.

Recent work, discussed below, has shown that some of these factors are
not as influential as previously thought.
Several changes in filler material have been made over the years
in an attempt to eliminate some of the causes of failure. Prior to
1950, columbium stabilized 309 stainless steel filler material
(Table II) was used to avoid martensite in the weld metal from dilution
by the low-alloy steel. In the early 1950s, a significant number of
DMW failures were observed with this filler material. These failures
typically occurred in the low-alloy steel HAZ, very close and parallel
to the weld line. Cracking was intergranular, and the failures showed
little evidence of ductility.
lor a short period of tir\e (about two years) in the early 1950s,
DMWs were made with a 2LC filler material, which was essentially a weak
ferritic filler material intended to alleviate the strength mismatch
(see Table II). However, failures occurred within one or two years.
Fracture occurred in the weld metal rather than in the low-alloy steel
HAZ.
Around the mid-1950s, some manufacturers changed to nickel-base
filler materials; Table II lists some of these. The most popular were
Inconel 132 and Inconel 182. The intention was to match the expansion
coefficient of the low-alloy steel more closely and minimize carbon
migration from the low-alloy steel HAZ to the weld metal. However, in
the mid-1970s, failures began to occur in plants with nickel-base
filler DMWs. The times to failure were typically fifteen to seventeen
years. Failures were experienced by some but not all utilities; the
welds with the longest service seem more prone to DMW failures.
Because thousands of welds had been made with nickel-base fillei"
 644

materials since the mid-1950s, an international task group was fronted

to solve the problem.
The first meeting of the Dissimilar Metal Weld Task Group, formed
under the auspices of the American Society of Mechanical Engineers
(ASME), the American Society for Testing Materials (ASTM), and the
Metal Properties Council (MPC) was held in June 1977. This task group
reported to the Steam Power Panel of the ASME Joint Committee on Effect
of Temperature on the Properties of Metals. Its membership consisted
of representatives from utilities in the United States and other coun-
tries (including Canada and the United Kingdom), boiler manufacturers,
universities, material producers, and allied industries. The extent of
the problem was identified, and ideas were exchanged concerning work
toward its solution. Three task forces were formed: Performance Sur-
vey, Experimental Joint, and Long-Range Planning. The Survey Task
Force identified the extent and common factors of the dissimilar weld
problem throughout the fossil-fired industry by sending out question-
naires and talking to utilities vhich were experiencing this problem as
well as those which were not. This survey resulted in a report*4, some
of whose findings are summarized in Ref. 1.
Along with the OMW Task Group, the Experimental Joint Task Force
idpntified possible improvements in the design and fabrication of dis-
similar welds. With the cooperation of task group members, over 200
weldments were made and placed in two boilers foe accumulation of serv-
ice time. These weldments were to be periodically examined to study
their performance. A report on this effort is being prepared.
The Long-Range Planning Task Force established a contract research
program to determine the causes of the problem and develop a solution.
This program, entitled the Dissimilar Weld Failure Analysis and Devel-
opment (DWFAD) Program, is currently funded by the Electric Power
Research Institute (EPRI) and jointly managed by EPRI and MPC. The
research is presently being conducted at General Atomic Company (GA)
and Combustion Engineering, Incorporated (CE).
Before describing this contract effort, the problem will be
illustrated and recent work described.

ILLUSTRATION OF PROBLEM

The DMWs of interest are located in the high-temperature super-

heater and reheater sections of boilers as well as in the main steam
piping from boilers to turbines. The problem exists in both horizontal
and pendant (vertical) boiler designs. This paper, and the DWFAD Pro-
gram, is concerned only with the boiler DMWs and does not address main
steam piping DMWs. The DWM failures of interest occur both in the
furnace (heat transfer region) and vestibule, or "penthouse" (nonheat
transfer region). Figure l(a) provides examples of these regions, and
Fig. l(b) presents a photograph of furnace DMWs.
Table I gives the base metal tubing compositions commonly used in
boilers. The "H", or high-temperature, grades of stainless steel are
shown in Table I, even though the plain lower-temperature grade has
 been used in some cases in the past.. The low-alloy steel tubing is
made of Cr-Mo material. Table II lists the filler metals commonly used
in DMWs. The types of filler materials most frequently used can be
divided into the stainless steel and the nickel-base filler categories.
Because there are considerable differences in both (Ju1 time to failure
and the type of failure occurring with these two general categories,
they will be treated separately.
The stainless steel filler DMWs typically fail within two to seven
years of service, although many weldments of this type have lasted for
over 20 years . Cracking and fai lure occur by prior austenite grain
boundary cracking in the low-alloy steel HAZ, as illustrated in Fig. 2.
The weld quality is usually adequate, and cracking almost always occurs
away from the fusion line. As seen in Fig. 2, cracking is not neces-
sarily associated with the formation of the oxide notch at the inter-
section of the weld metal and the low-alloy steel HAZ. However, the
loss of tube wall cross section may aggravate cracking in the HAZ. The
dark etching structure at the "weld line" in Fig. 2(c) is identified in
the literature as tempered martensite. As can be seen from the scan-
ning electron microscope (SEM) photograph in Fig. 2(d) , the carbides in
the band corresponding to the tempered marter.site zone have become
coarse with service exposure time, and micrcvoid formation can occa-
sionally be seen in the carbide band. The majority of the cracking is,
however, located in prior austenitic grain boundaries in the low-alloy
steel HAZ. The cracks eventually link up, and final failure in the
form of a steam leak or complete separation occurs. When one DMW fails
in a boiler, the DMWs removed from nearby welds may show extensive
cracking without steam leakage.
Nickel-base filler DMWs in normal service may fail after fifteen
to seventeen years of operation, but many DMWs have lasted for 200,000
hours. Failures have been reported after two to seven years of serv-
ice, but these have been associated with either overstress or overtem-
perature operating conditions. Because of the better general perform-
ance of nickel-base filler DMWs, many utilities and manufacturers have
replaced their stainless steel DMWs with nickel-base DMWs in existing
plants and original construction. The filler materials preferred for
repair and construction are Incoweld A, Inconel 82, and Inconel 182, as
seen in Table II. However, there is some doubt about the thermal
stability of Inconel 182. 15
Figures 3 through 6 show cracking and failures observed in
nickel-base DMWs. These examples show that cracking and failure in
nickel-base DMWs occur primarily at a semicoauinuous phase which form
very close to the weld line and secondarily at prior austenite grain
boundaries near the weld line. The fracture is entirely on the low-
alloy steel HAZ side. The amount of HAZ cra-.king in failed or damaged
nickel-base DMWs is considerably less than that in stainless steel
filler DMWs. In addition, the nickel-base filler DMWs fail closer to
the weld line. Ferritic material on the fracture surface is often dif-
ficult to find, because this material is consumed by oxidation. Final-
ly, when observed visually, the weld pass penetration contour on the
low-alloy steel side is usually apparent on the fracture surface, as
Illustrated in Fig. 3(a).
 646

RECENT OBSERVATIONS OF NICKEL-BASE DMWs

Most of the current researcV has been devoted to nickel-base DMWs,

because recent failure experience has shown that the use of this mate-
rial is not the complete solution to the problem, as was previously
thought. Although nickel-base DMWs have a lower failure frequency than
the stainless steel DMWs, the majority of the nickel base DMWs gener-
ally have accumulated shorter service times. Many of these nickel-base
DMWs are approaching service times at which failure can occur.
Failure analysis reports'^ illustrate the importance of a seraicon-
tinuous phase near the weld line in the fracture and failure of nickel-
base DMWs. Figures 5 and 6 show this for a superheater nickel-base DMW
failure after seventeen years of service. Fracture follows this phase
and prior austenite grain boundaries near the weld line. This phase
has recently been identified by Nicholson^ and Hilton^" to be primar-
ily M23C5 carbide. This carbide phase has been found in other reheater
and superheater DMWs examined after long periods of service, as illus-
trated in Figs. 3, 4, 7, and 8. The microvoids formed around these
carbide particles during service are thought to eventually link up and
cause failure.
The reheater DMW shown in Fig. 4 actually failed because of
propagation of the oxide notch through the T-ll tube wall. This led to
steam leakage and cutting between adjacent DMWs, causing failure. How-
ever, extensive microvoiding was observed at interfacial and grain
boundary carbides in the T-ll HAZ near the weld line, as shown in Fig.
4(d). When microvoids are small enough, they may be missed by conven-
tional optical metallography, as can be seen by comparing Figs. 4(c)
and 4(d). The SEM is better suited for detecting this type of damage.
The amount of cracking damage before final failure in nickel-base DMWs
does not progress to the amount observed for stainless steel filler
DMWs.
To determine the role of the interfacial carbide phase in
nickel-base DMW failures, unfailed superheater DMWs were examined after
approximately 172,000 hours of service. Samples were removed from a
superheater that had suffered failures. After 126,000 hours of serv-
ice, all the DMWs were given a normalizing heat treatment to redistrib-
ute the carbon. The amount of carbon was thought to be low due to car-
bon depletion in the T-22 HAZ. One effect of this heat treatment was
to produce a fine grain size in the T-22 HAZ near the weld line, as
illustrated in Fig. 8. Extensive microvoid formation at the interfa-
cial carbides was also observed. These microvoids probably formed dur-
ing service. Electron microprobe line scans across the weld line show
that carbon depletion is not present in most locations, as illustrated
in Fig. 9. Isolated instances of carbon depletion were found, but they
were not representative. Carbon migration occurred, as evidenced by
carbon reductions to approximately 0.1% to 0.09% in the HAZ and carbon
enrichment (carburization) in the weld metal.

Defined as 0.03% carbon or less.

 647

Mechanical test specimens were removed and tested to determine the

significance of the carbides and microvoids. One such test was a
four-point bending test conducted at elevated temperature with constant
load on a reduced section of this 172,000-h-old service weldment.
Failure occurred after 354 hours at 593°C (1100°F) with an initial
outer fiber bending stress of 62 MPa (9 ksi), as illustrated in
Fig. 10. The visual appearance of the fracture is typical of service
failures, as shown in Fig. 10 by the penetration contour of the initial
passes in the T-22 HAZ. Conventional elevated-temperature tensile and
stress rupture specimens were machined transverse to 172,000-h-old
service weldments. Testing at 593°C (1100°F) in air also caused frac-
ture in the typical service failure fashion, as illustrated in Fig. 11.
Figure 12 shows a significant reduction in the rupture time for service
weldments compared with the original expected minimum rupture times.
This reduction was location dependent, and the weldments operating at
somewhat higher-temperature regions of the superheater failed sooner
than those from lower-temperature regions (fourth row compared with
sixth row in Fig. 12).
Cross sections through typical tensile and stress-rupture failures
showed that fracture occurred through the interfacial carbides and
microvoids, as seen in Fig. 13. Figure 14 illustrates this further by
showing the stress-rupture fracture surface after testing in helium to
prevent oxidation. Carbide particles initiated the failure and are
present in most of the dimple facets on the fracture surface. The
small grain size is probably responsible for the ductile nature of the
fracture in Fig. 14. Hence, it becomes apparent that interfacial car-
bides and microvoids can be instrumental in DMW fracture location.
Interfacial carbides can also be produced by laboratory aging of
DMWs, as illustrated in Fig. 15. This capability is expected to play
an important role in the DWFAD Program in producing typical service
failures in short time periods.

THE DISSIMILAR WELD FAILURE ANALYSIS AND DEVELOPMENT PROGRAM

As previously mentioned, through the efforts of EPRI, MPC, and the

Dissimilar Metal Weld Task Group and its task forces, a research effort
was initiated to determine the causes of DMW failures and to initiate
work toward a solution(s) to the problem. The objectives of the DWFAD
Program are: (1) to obtain a clear understanding of the underlying
causes of dissimilar weld failures; (2) to develop a methodology for
estimating the residual life of service welds; (3) to devise a critical
discriminatory test to predict the relative performance of welds; and
(4) to establish guidelines for improved dissimilar welds. The program
is divided into six tasks, which are briefly described below. The
specific results of this program will be published in the near future.
 648

TASK 1: THE IDENTIFICATION OF THE CAUSES OF DISSIMILAR WELD FAILURES

Work, on this task will concentrate on studying and correlating

previous experience, reported in published and unpublished reports,
with dissimilar weld failure analyses. Emphasis will be given to
service factors which may be influential in causing failures.
Samples of DMWs with aoproximately 100,000 hours of service will
be obtained from horizontal and vertical heat exchanger sections;
stainless steel and nickel-base filler material DMWs wil] be retrieved.
Cracked and uncracked samples will be removed, compared, extensively
analyzed by current metallurgical techniques, and mechanically tested.
Important service factors and detailed service histories will be
determined, and the results of the metallurgical and mechanical tests
will be correlated with the histories to enable the cause(s) of the
actual failures to be established.

TASK. 2: ANALYSIS OF DISSIMILAR METAL WELDMENTS WITH FOSSIL-FIRED

BOILER SERVICE

Dissimilar weldments in as-fabricated samples and in samples with

50,000 and 150,000 hours of service will be metallurgically examined.
Initially, only nominally sound weldments made with nickel-base filler
materials will be examined. The objective of these examinations is to
characterize the evolution of damage in nickel-base weldments. The
resuts will be essential for estimating residual life in other parts of
the program. In addition, the information obtained from the metallur-
gical analysis combined with service histories and operational factors
will be instrumental in helping determine the cause(s) of dissimilar
weld failures.

TASK 3: SIMULATION OF METALLURGICAL CHANGES THAT OCCUR IN DISSIMILAR

WELDS DURING SERVICE

Dissimilar metal weldments will be fabricated to simulate service

weldments and to represent possible improvements based on information
derived from this program, available literature, and other experience.
Various combinations of filler and base materials will be subjected
to thermal aging cycles to simulate service exposure, and realistic
temperature/time parameters will be developed to produce service micro-
Structures. Once this accelerated procedure(s) is developed, the
weldments which represent possible improvements can be evaluated based
on their aged microstructures. The information from this task will be
helpful in determining the cause of DMW failures and will help provide
age samples for discriminatory testing.
 649

TASK 4: STRESS ANALYSIS OF DISSIMILAR WELDS

Currently available stress analysis and life prediction techniques

will be used to determine the types and magnitudes of internal and
external stresses on dissimilar welds. The information generated will
be helpful in determining the residual life of dissimilar weldments.
The results may be important in establishing the causes of failure and,
in fact, may explain why some DMWs fail after short periods of time
while the same types of DMWs outlast the boiler.

TASK 5: DEVELOPMENT OF AN ACCELERATED LABORATORY TEST TO DISCRIMINATE

BETWEEN WELDS

Laboratory tests will be conducted on full-size DMWs to produce

service-type damage and failure in an accelerated time frame. Existing
DMWs and possible improvements will be evaluated, and the test results
will form a critical part of residual life determinations of DMWs in
service. The task consists of using higher temperatures or stresses
and more frequent operational cycles to cause early damage and repre-
sentative DMW failure. Pre-aging will be used to simulate prior service
exposure. Crack initiation and propagation to final failure will be
monitored using acoustic emission equipment.

TASK 6: DEVELOPMENT OF GUIDELINES FOR FABRICATION OF IMPROVED

DISSIMILAR WELDS

Based on the results of the above six tasks, guidelines will be

established for future improved dissimilar weldments. Optimum combin-
ations of materials, prior heat treatment, fabrication practice, and
design will be identified.

OTHER PROGRAMS

Development of an accelerated DMW test has been pursued at

Ontario-Hydro in Canada and at KEMA in the Netherlands. These
programs'"-* have successfully produced typical failures in stainless
steel filler DMWs. Typical failures have been produced in an acceler-
ated time frame by increasing the stress on the dissimilar weldment^
and increasing the number of cycles and test temperature^* »^. Neither
program has produced typical failures in nickel-base DMWs.
A considerable effort has been expended on the dissimilar weld
problem at the Central Electricity Generating Board (CEGB) in the
United Kingdom. Work has been done 2 which shows that the dissimilar
weld problem is statistically significant, as differentiated from
 650

failures in similar weldments. Test work directed at failures in

stainless steel and nickel-base weldments has also been performed, and
typical failures have been brought about in stainless steel DMWs in
shortened time frames by increasing the test temperature^"^. Nickel-
base filler DMW failures have been produced by prior austenite grain
boundary cracking in the low-alloy HAZ. Typical interfacial carbides
were not present in the DMW microstructure in these tests.
Work on the analysis of dissimilar welds returned from service has
been done at the University of Tennessee^. Research on DMWs is also
being performed at the Colorado School of Mines and the University of
Michigan. Recent work at the Colorado School of Mines has focused on
the significance of expansion coefficients*'.

FUTURE WORK

The most effective work toward the solution of the dissimilar weld
problem will be done when the causes of dissimilar weld failure are
completely understood. It is important to know, for example, if oper-
ational factors such as high local stresses in certain boiler locations
can cause a dissimilar weld to fail before the expected base material
life. It is important to know if there are metallurgical phenomena
which can be controlled by proper DMW heat treatment which will prolong
DMW life beyond the life of the parent metal under any service condi-
tions. The development of an accur'te accelerated test is essential so
that future improvements can be evaluated in an accelerated time frame.
Exposure to actual service conditions, such as the work being done on
the experimental joint programs, is also valuable, and the results or
periodic examinations will be important in deciding future improvements
in DMWs. Reference 18 describes some improvements being evaluated in
the United Kingdom. Some of the current efforts on improvements which
are under way, such as continously graded joints and automatically
welded joints, should be evaluated in service environments and in an
accelerated test and compared with state-of-the-art dissimilar
weldments that have accumulated long service times.

ACKNOWLEDGMENTS

The authors would like to acknowledge the efforts of G. M.

Slaughter of Oak Ridge National Laboratory, Chairman of the DMW Task
Group, its members, and EPRI and MPC personnel in creating the DWFAD
Program. This work was partially supported by MPC Subcontract J80-6
under EPRI Contract RP-1874.
 651

REFERENCES

1. C D . Lundin, "Dissimilar Metal Welds - Transition Joints

Literature Review," Weld J. 61 (2): 58-s - 63-s (1982).

2. R. D. Nicholson and A. T. Price, "Service Experience of

Nickel-Based Transition Joints," paper presented at the Welding
Institute Seminar on Dissimilar Joints, Leicester, United Kingdom,
June 1981, to be published by Welding Institute in 1982.

3. R. B. Dooley et al., "Ontario Hydro Experience with Dissimilar

Metal Welds in Boiler Tubing," Weld. J. 6I_ (2): 45-s - 49-s
49-s (1982).

4. M-. D. C. Moles et al. , "Accelerated Testing of Three

Austenitic-Ferritic Superheater Tube Transition Welds Under
Simulated Two-Shift Conditions," Metals Technol., April 1981.

5. J. T. W. Pastoors, and J. van Liere, "KEMA Research on Dissimilar

Joints," N. V. KEMA, Arnhem, Netherlands, Report WSK/98000-4,
October 1981, presented at ATSM-ASME-MPC Joint Committee
Dissimilar Weld Task Group Meeting, October 8-9, 1981, La Jolla,
California.

6. 0. Abiko and A. Yamaura, "Dissimilar Weld in Thermal Plant," paper

presented at ASTM-ASME-MPC Joint Committee Dissimilar Weld Task
Group Meeting October 8-9, 1981, La Jolla, California.

7. F. Rapaty, and TTumitzsch, "Structure of Transition Fusion Zone

Obtained in Fusion Welding," Arch. Eisenhuttenw. 8 (12): 555-556
(1935) (Henry Brutcher, translation No. 2202).

8. 0. R. Carpenter et al., "Some Considerations in the Joining of

Dissimilar Metals for High-Temperature High-Pressure Service,"
Proc. ASTM 50: 809-860(1950).

9. R. U. Blaser, F. Eberle, and J. T. Tucker, "Welds Between

Dissimilar Alloys in Full-Size Steam Piping," Proc. ASTM 50:
789-808 (1950).

10. R. D. Nicholson, "Interfacial Structures in Nickel-Based

Transition Joints After Long Terra Service," CEGB (Marchwood)
Report RD/M/N1131, September 1980.

11. R. D. Nicholson, "The Creep Rupture Properties of Austenitic and

Nickel-Based Transition Joints," CEGB (Marchwood) Report
RD/M/1176N81, June 1981.
 652

12. V. C. Jolliffe and J. A. Williams, "Thermal Cycling Behavior of a

Cracked Transition Joint Between Austenitic and Ferritic Steels
Welded With an Austenitic Weld Metal," CEGB (Marchwood) Report
R/M/R271, October 1978.

13. J. F. King et al., "Examination of a Failed Weld Transition

Joint," Oak Ridge National Laboratory Report ORNL-5223, January
1977.

14. P. E. Haas, "Results of Industry-Wide Survey on Dissimilar Metal

Weld Performance," American Electric Power Service Corporation,
Canton, Ohio, 1978-

15. E. Siegel, "Paramagnetic-Ferromagnetic Transition in Aging,

Precipitation Hardened Nickel-Iron Solid Solution Alloys," Journal
of Magnetism and Magnetic Materials 7: 312-316 (1978).

16. Hilton, S., Combustion Engineering, private communication, 1978.

17. J. W. Elmer et al., "Thermal Expansion Characteristics of

Stainless Steel Weld Metal," paper presented at American Welding
Society 1981 Annual Conference, Cleveland, Ohio, to be published
in Weld. J.

18. B. I. Bagnall, T. J. Rawberry, and J. A. Willams, "Service

Experience With Heavy Section Dissimilar Joints Operating at
Elevated Temperature," presented at the Welding Institute Seminar
on Dissimilar Joints, Leicester, United Kingdom, June 1981, to be
published by Welding Institute in 1982.
 Table I. Base metal compositions used in dissimilar metal

Cb +
Type Specification Fe Ni Cr Mo C Ti Ta Si Mn s P

T-22 ASME SA-213 Balance — 1.9- 0.87- 0.15 — — 0.5 0.3— 0.03 0.03
2.6 1.13 0.6
T-11 ASME SA-213 Balance — 1.0- 0.44- 0.15 — — 0. c;_ 0.3- 0.03 0.03
1.5 0 65 1.0 0.6
T-21 ASME SA-213 Balance — 2.65- 0 8- 0.15 — — 0.5 0. 3- 0.03 0.03
3.35 1 06 0.6
304H ASME SA-213 Balance 8.0- 18.0- — 0.04- — — 0.75 2.0 0.03 0.04
11.0 20.0 0. 1
321H ASME SA-213 Balance 9.0- 17.0- — 0.04- >4 x C — 0.75 2.0 0.03 0.04
13.0 20.0 0. 1 <0.60
347H ASME SA-213 Balance 9.0 17.0- — 0.04- — >8 x C 0.75 2.0 0.03 0.04
13.0 20.0 0. 1 «1.0
316H ASME SA-213 Balance 11.0- 16.0- 2 .0- 0.04- — — 0.75 2.0 0.03 0.04
14.0 18.0 3 .0 0. 1

'a)Unless otherwise indicated, single values are maximuras.

 Table II. Filler mecal compositions used for dissimilar metal welding' 3 ^

Cb + |
Type Specification Ni Cr Fe Ta Mn Co Mo Ti Si C Cu S P
!
E3O9 Cb AWS/ASME SFA- 12.0- 22. 0- Balance 0. 7- 2 5 • — 0. 5 -- 0.9 0. 12 0.5 0.03 0.04
5.4 - E309 Cb 14.0 22. 5 1.0
E310 AWS/ASME SFA- 20.0- 25. 0- Balance
2
5 — 0 5 -- 0.75 0.20 0.5 0.03 0.03
5.4 - E310 22.5 28. 0
2LC AWS/ASME SFA- — — Balance 0 .9 • — 0 4- —— 0 6 0.05 0 04 0.03
5.5 - E7015 - 0 65

Inconel 132 AWS/ASME SFA- 62.0 13 0- 11.0 1 5- 3.5 — - -- 0 75 0.08 0.5 0 02 0.03
5.11 - ENiCrFe-1 minimum 17 0 4
Inconel 182 AWS/ASME SFA•• 59.0 13 0- 10.0 1 0- 5.0- (d) — 1.0 10 0. 1 0.5 0 015 0.03
5 . 1 1 - ENlCrFe-3 miniraura 17 0 2 c( C) 9.0
Incoweld A AWS/ASME SFA- 62.0 13 0- 12.0 0 1.0- (d) 0 — 0 75 0 1 0 5 0 02 0 03
5.11 - ENiCrFe-2 minimum 17 0 3 3.0 2 .5
Inconel 82 AWS/ASME SFA- 67.0 18 0- 3.0 2 .0- 2.5- (d) 0 75 0 .5 G 1 0 5 0 .015 0 03
5.14 - ERNiCr-3 miniraura 22 0 3 .0 3.5
Inconel 92 AWS/ASME SFA- 67.0 14 .0- 8.0 — .0- — — 2. 5- 0 .5 0 08 0 5 0 .015 0 03
5.14 - ERNiCrFe-6 minimum 17 .0 .7 3 5

'a)llnless otherwise specified, single values are maximums.

'b^ for low carbon.
= 0 . 3 maximum when specified.
12 maximum when specified.
 655

INLET OUTLET
HEADER HEADER

'LOW-ALLOY STEEL TUBING (T-22)

•DISSIMILAR METAL WELD

LOW-ALLOY
STEEL TUBING FURNACE ROOF
(T-22)-
DISSIMILAR METALWELD

STAINLESS STEEL TUBING

(321H)

(a)

Fig. I. Locations of dissimilar metal weldments (DMWs): (a) schematic of

single tube circuit in a pendant superheater or reheater; (b)
photograph of DMWs in service in a boiler superheater furnace
section
 656

WELD ^CRACK/OXIDE WELD O.D.

METAL NOTCH METAL.

;—OXIDE
NOTCH

T-22 HAZ

CRACKING

Fig. 2. Boiler tube DMW made with 309 filler material. Weldment was in
service at 538°C (1000°F) for 55,000 h in a supercritical pendant
superheater: (a) exterior of weldment; (b) photomacrograph
illustrating typical HAZ cracking and oxide notch
 657

WELD
METAL

TEMPERED
MARTENSITE

CRACK

HAZ

>;?/. >av'.; •• . •'•, -."• V ' .-..?

* t.1

(c)

/
y
CRACK

^•- -^.-v-
'*•*
• • • ^
"V*
-CARBIDE
BAND IN
TEMPERED
MARTENSITE
REGION

Fig. 2. Boiler tube DMW made with 309 filler material. Weldment was in
service at 538°C (1000°F) for 55,000 h in a supercritical pendant
superheater: (c) light photomicrograph illustrating typical HAZ
cracking and weld interface structures (dilute aqua regia etch)
(475X); (d) scanning electron microscope (SEM) photomicrograph
similar to (c), illustrating HAZ and weld interface carbide
distribution (2000X)
 658

STAINLESS
.OXIDE NOTCH
STEEL TUBING
AND CRACK
T-22 TUBING

(a)

BACKING RING
WELD FRACTURE
META .^SURFACE

(b)

Fig. 3. Boiler tube DMWs made with nickel-ba^e filler material (Inconei
132). Weldments were in service at 565°C (1050°F) for ~ 2 0 yr:
(a) exterior view of weldment with oxide notch and partial
separation; (b) failure with complete separation
[6.35-cm (2-1/2-in.) o.d.] (note weld bead contour on fracture
surfaces)
 &:- 9

WELD STAINLESS
T-22 CRACK METAL STEEL

\ \

WELD LINE

T-22 HA2

WELD
METAL

-•<. ,

Fig. 3. Boiler tube DMWs made with n i c k e l - b a s e f j J l e r m a t e r i a l (Ineonel

132). Weldments were in s e r v i c e at 565°C (105fl°K) for ~20 y r :
(c) a s - c u t s e c t i o n through weldroent shown in (fa); (d) SEM photo-
micrograph of i n t a c t s e c t i o n from ( b ) , i I l u s t r a t i n g microvoid
damage (410X)
 660

321 STAINLESS STEEL 1-1/4 Cr-1/2 Mo

ENiCrFe-1
(132) WELD
METAL

(a)

STEAM
CRACK CUTTING
(b)

4. N i c k e l - b a s e f i l l e r DMW f a i l u r e f r o m p e n d a n t r e h e a t o r a f t e r
1 ^ 0 , 0 0 0 h o f s e r v i c e ' a t 5 1 8 ° t o 5'J8°C ( 9 5 0 ° u> 1 0 0 0 ° r ) : (a)
o v e r a l l v i e w ; (b) c l o s e - u p
 661

WELD
METAL

CARBIDE
PHASE

WELD
METAL

SEIMICONTINUOUS
CARBIDE PHASE

VOID/CRACK
FORMATION

WELD
LINE

Fig. t. NickeJ-base filler DMW failure from pendant reheater after

160,000 h of service at 518° to 538"C (950° to 10O0°F): (c)
light photomicrograph at weld interface, 2% nital etch (500X);
(d) SEM photomicrograph near ( a ) , liftlit 2Z nital etch
 662

' 2-1/4 Cr-1.0 Mo STEEL

(W
Fig. 5. Failed DMW made with nickel-base filler material. WeJdment was
in service in a horizontal superheater at 565"C (1050°F) for
~17 yr (from Ref. 13): (a) segment from failed DMW; (b) photo-
macrograph of (a)
 WELD METAL (INCONEL 132 FILLER) FUSION LINE "

CROLOY 2 - 1 / 4 ; J .
(2-1/4 Cr-1.0 Mo) HAZ i'- ' • ' . . . ^ O ; V

(a)

S ^ f l M w E L d ' METAL (INCONEL 132 FILLER)

•v FUSION L I N E ' ' : ? ^ ,

\ g * = - ^ , GRAIN BOUNDARY
FISSURESV • " '
• ?> •:.• • L - ,

\ ' - GRAIN BOUNDARY CRACK..

CROLOY (2-1/4 Cr- 1.0 Mo)HAZ

(b)

F i g . 6. P h o t o m i c r o g r a p h s of c r a c k t i p from unfai. led p o r t i o n of DMV shown in I" i j». 5 (from

Ref. 1 3 ) : ( a ) f r a c t u r e a l o n g s e m i c o n t i n u o u s p h a s e ; (b) Krain boundary f r a c t u r e
n e a r weld l i n e
 664

ENiCrFe-1
(132) WELD METAL

321H

Fig. 7. DMW from horizontal superheater location after 172,000 h of

service at 593°C (1100°F). After 126,000 h of service, weldment
was normalized at 940°C <1725°F) for 1 h
 665

V
•:• 'V,

2% IMITAL 2-1/4 Cr- Mo WELD ENiCrFe-1 (INCONEL 132)

ETCH HAZ INTERFACE

SEMICONTINUOUS
CARBIDE PHASE

WIICROVOIDS

2%I\IITAL 2-1/4 Cr - 1 Mo WELD ENiCrFe-1 (INCONEL 132)

ETCH HAZ INTERFACE

(b)

Fig. 8. SEM photograph ul" 2-1/4 Cr - I Mo HAZ and void interface ^t D>fW
shown in Fig. 7: (a) micrograph (450X); (b) higher magnification
(1800X)
 666

CO

%
 667

,WELD METAL (INCONEL 132 FILLER)

3Z1H
T-22 HAZ FRACTURE

lig. 10. Bending t e s t conducted on service DMW similar to that shown in

Fig. 7; t e s t conducted at 593°C (1IOO°F) with an i n i t i a l outer
fiber bending s t r e s s of 62 MPa {9 ksi) and 354 h to complete
failure
 668

ENiCrFe-1 FRACTURE
FRACTURE (132) WELD
METAL

2-1/4 Cr- 1 A/lo

321H

50E-2 ' . 50E-4

STRESS-RUPTURE TENSILE
ENiCrFe-1
VIEW OF FRACTURE SURFACES (132) WELD
METAL
TA Cr-1Mo

50 E4 50E-2

TENSILE STRESS-RUPTURE
(b)

Fig. 11. Mechanical t e s t specimens machined from D

M W similar to that
shown in Fig. 7; tested at 593°C (MOOT) in a i r : (a) side view;
(b) isometric view
 669

o
o
o

o
o
.—.
UJ .—.
en - — • •

C-T.
[ •—(
O- o L^,
=3
§§ EC
O
n
_~"

DC DC
|— c
UJ
2 j£
LJJ CO tn
a <n 1-
in in

D
DO
u
a.

J L J L

if
[(ISM) SS3U1S
 670

ENiCrFe-1 (132) FRACTURE SURFACE

WELD METAL

ETCH: 2% NITAL +
ELECTROLYTIC

(iOg CHROMIC OXIDE

10ml ACETICACID
200 ml H 2 0)

(a)

2-1/4 Cr - 1 Mo HAZ

FRACTURE SURFACE

PHASE
EWiCrFe-1 ASSOCIATED
(132) WELD WITH
METAL FRACTURE

ETCH: 2%
(b) NITAL ONLY

Fig. 13. T e n s i l e f r a c t u r e c r o s s s e c t i o n a t 593°C (M00°F) from .specimen

shown in Fltf. 10: (a) photomacrograph (5.5X); (b) p h o t o m i c r o -
graph (1000X)
 671

as*

SEM fracto^caph o\ creep-rupture test conducted in helium to

preserve fracture surface; test condition and yample similar to
those shown in Figs. W and 12 (274OX)
 672

INTERFACIAL
WELD METAL. / CARBIDE PHASE /2-1/4 CR - 1 MO HA2

Fig. 15. Interfacial carbide phase produced in Alloy 800/Inconel 82/2-1/

Cr - 1 Mo dissimilar weldment by aging for 10,000 h at 593°C
(1100°F); It nital etch (566X)
 673

SOME OBSERVATIONS ON THE PHYSICAL

METALLURGY OF NICKEL ALLOY WELD METALS

C. G. Skillern and A. C. Lingenfelter

Huntington Alloys, Inc.

Huntington, WV 25720

ABSTRACT

Numerous nickel alloys play critical roles in various energy-

related applications. Successful use of these alloys is almost
always dependent on the availability of acceptable welding methods
and welding products. An understanding of the physical metallurgy of
these alloys and their weld metals and the interaction of weld metal
and base metal is essential to take full advantage of the useful
properties of the alloys. To illustrate this point, this paper
presents data for two materials: INCONEL alloy 718 and INCONEL
Welding Electrode 132.

INCONEL* alloy 718

INCONEL alloy 718 is used in a number of nuclear and

conventional power systems. Alloy 718 is an age-hardenable
high-strength alloy used at cryogenic temperatures as well as high
temperatures ranging up to 1300°F. The sluggish aging response of
the alloy is an asset from a welding standpoint, since it avoids the
problems of strain-age cracking.

Table I shows the nominal chemical analysis of alloy 718. The

I1' and )"" aging constituents develop because of the presence of the
elements columbium, aluminum, and titanium.

*INCONEL and INCOLOY are registered trademarks for products of

Huntington Alloys, Inc.
 674

Table 1. Nominal Chemical Analysis ot INCONEL alloy 718

c Mn Fe S Si Cu Ni Cr Al Ti Mo Cb+Ta
0.04 0.18 18. S 0.005 0.18 0.15 52.5 19.0 0.50 0.90 3.05 5.13

Alloy 718 is normally welded by gas-tungsten-arc welding (GTAW)

with a matching-composition filler metal. Figure 1 shows a photo-
micrograph of the structure of GTAW weld metal. The segregation
typical of the weld-metal structure is evident. The white phase in
the interdendritic areas is Laves phase, a brittle intermetallic com-
pound. Laves phase precipitates primarily because of the segregation
of columbium to the interdendritic areas.

In most applications, the alloy is used in one of two condi-

tions: welded plus direct age or welded plus 1750°F anneal plus age.
With these heat treatments, the all-weld-metal elongation and
reduction in area will be of the order of 10% and 15%, respectively.
This level of ductility has proven adequate for most applications.
However, many of the energy-related applications have specifications
which require a weld bend test as a quality measure. This level of
ductility will not pass a 5T U-bend test. It is known that a post-
weld solution anneal will dissolve the Laves phase and markedly
improve the weld-metal ductility. It is also well established that
heir treatment affects the physical properties'- and weldability-»^>4>5
of the alloy. Thus, the objective was to improve the weld ductility
sufficiently to meet the bend-test requirement without unduly
damaging the other desirable properties of the alloy.

Figure 1. INCONEL alloy 718 GTAW weld given Heat Treatment 3. Laves
phase is present in Interdendritic channels. M ^ C b precipitates
during heat treatment. Etchant: 5% chromic acid electrolytic. 500X.
 675

Effect of Heat Treatment on Weldablllty

Coarse grain size increases sensitivity to heat-affected-zone

microfissuring. Figure 2 shows microfissuring in material with an
average grain size of ASTM #1. Generally, grain sizes larger than
ASTM #5 show increased sensitivity to heat-affected-zone roicro-
fissuring. In the case of a postweld heat treatment which causes
grain coarsening, the repair weldability is the property damaged.
Alloy 718 recrystallizes in the 1750-1800°F temperature range.
Temperatures above 19OO~195O°F dissolve the majority of the phases in
the microstructure. Without these phases present to pin the grain
boundaries, grain growth occurs. Temperatures at or above 1950°F are
needed to dissolve the Laves phase present in the weld metal and
improve the weld-metal ductility. Annealing at 1950°F will generally
produce a grain size of ASTM #4 or finer.

Figure 2. Intergranular fissure in fusion line and heat affected

zone of J-groove weld in coarse-grained INCONEL alloy 718. Etchant;
5% chromic acid electrolytic. 100X.
 676

Effect of Heat Treatment on Base-Metal Properties

Annealing in the 1750-1800°F temperature range produces a fine

grain size (ASTM #7 or finer), the best room-temperature tensile
strength, the best fatigue strength, and notch ductility in stress
rupture. It also produces low stress-rupture strength, reduced
transverse ductility, and notch brittleness in the room-temperature
tensile test, and it does not dissolve the Laves phase in the weld
metal.

Annealing in the 1900-1950°F temperature range produces the best

smooth-bar stress-rupture strength, better room temperature tensile
ductility, some degree of grain coarsening, and a relatively clean
microstructure. The high temperature dissolves the Laves phase,
improving weld-metal ductility. The anneal also produces a notch-
brittle condition in stress rupture. Earlier work" showed that a
furnace-cool from 1950°F to 1750°F reprecipitated carbides and/or
NijCb in the grain boundaries. The presence of these grain-boundary
phases improved the notch-bar stress-rupture strength although not to
the point of being notch ductile.

Proposed PoBtweld Heat Treatment

Table II summarizes the heat-treatment combinations which were

explored. Table III shows all-weld-metal room-temperature tensile
properties for manual GTAW welds made in 1/2 in. alloy 718 plate.
The 1950°F anneal, as would be expected, improves the weld-metal
ductility. It also improves the weld-metal yield and tensile
strengths since it dissolves the Laves phase. The columbium tied up
in the Laves phase becomes available for strengthening as y1 and yll .

Table II. Heat Treatments Explored

(1) As-welded.

(2) 1325°F/8 hr FC 100°F/hr to 1100°F/8 hr, AC.

(3) 1750°F/l hr, AC plus 1325°F/8 hr FC 100°/hr to 1100°F/8 hr, AC.

(4) 1750°F/l hr, AC plus 1400°F/5 hr FC 100°/hr to 1200°F/8 hr, AC.

(5) 1950°F/l hr, AC plus 1400°F/5 hr FC 100°F/hr to 1200°F/8 hr, AC.

(6) 1950°F/l hr FC 200°F/hr to 1750°F/l hr, AC plus 1400°F/5 hr FC

100°F/hr to 1200°F/8 hr, AC.
 677

Table III. All-Weld-Metal Roo«-Temperature

Tensile Properties

Heat Treatment 0.2% YS TS % El % RA

(1) 84,000 124,500 32.0 30.5

(2) 163,300 186,500 11.0 16.0
(3) 147,000 178,800 11.0 13.0
(4) 148,800 180,600 14.0 14.5
(5) 173, 100 195,300 22.0 32.4
(6) 174,400 195,600 20.0 28.3

Table IV summarizes the face-bend and side-bend test results.

Prior to bending, the specimens were scribed such that estimated
elongati^" .alues could be obtained from the 5T bend tests. Figure 3
shows face-bend tests representing four of the heat-treatment
conditions.

Table IV. 5T Face-Bend and Side-Bend Test Results

Heat Treatment
#1 #2 #3 #4 #5 #6
Estimated Estimated Estimated Estimated Estimated Estimated
% El % El % El % El % El % El
8-12-15-C* 8-12-15-C 8-12-15-C 8-12-15-C 8-12-15-C 8-12-15-C

Face Bend X p** F F X X

Side Bend X F F F X X

*C - full 5T U-bend.

**F - 5T bend failed before reaching this level of elongation.

Table V shows all-welrl-metal stress rupture properties for three

heat treatments.

Table V. All-Weld-Metal INCONEL alloy 718

Stress, Temperature Life

•
Heat Treatment psi °F h 7 El 7 RA
(4) 100,000 1200 34. 3 8 .0 5.0
(5) 100,000 1200 140.0 3.3 4 .0
(6) 100,000 1200 203. 7 1.4 11.5
 678

Figure 3. 5-T longitudinal face bends from 1/2-in.-thick weldments

made with INCONEL Filler Metal 718. Bend A, Heat Treatment 1.
Bend B, Heat Treatment 2. Bend C, Heat Treatment 3. Bend D, Heat
Treatment 6.

From these data, It was concluded that Heat Treatment 6 provides

the best compromise. It produces <. afficient ductility to meet a 5T
bend test requirement, provides reasonable tensile and rupture
properties, and retains repair-weldii e; capability. If the
application requires particular physical properties, the heat
treatment can be tailored to meet the properties. Unfortunately,
optimizing particular properties usually means sacrificing other
useful characteristics.

INCONEL Welding Electrode 132

The original INCONEL Welding Electrode 132 was introduced in the

late 1940's. It was one of the first coated electrodes available for
welding INCOtfEL alloy 600. The early version of the electrode,
although quite adequate for most applications, did not prove suffic-
iently crack-resistant to meet the defect-f?:ee standards for
nuclear-power construction. The original version also had problems
with porosity when used for out-of-positlon welding.

These deficiencies lead to the development ot the INCONEL

Welding Electrode 182 which successfully met the standards as an a l l -
position electrode. This product continues to be the industry
standard for joining INCONEL alloy 600 and other alloys.
 679

Evaluation of the Improved Electrode

INCONEL Welding Electrode 132 has been modified a number of

times over the years, but the most notable improvements have resulted
from a just-completed redevelopment program. The electrode has none
of its previous shortcomings and may be superior to INCONEL Welding
Electrode 182 for some applications.

Table VI shows the nominal chemical composition of the

redeveloped electrode. The composition of INCONEL Welding Electrode
182 is shown for comparison.

Table VI. Nominal Chemical Compositions, Z, of

INCONEL Welding Electrodes 132 and 182

C_ M n F e £ S j . O i N i C r _ Ti Cb+Ta

Old 132 0.04 0.75 8.50 0.005 0.20 0.10 73.0 15.0 - 2.10
New 132 0.04 2.40 8.25 0.005 0.55 0.10 70.0 15.^.5 - 2.JO
182 0.05 7.75 7.50 0.005 0.50 0.10 67.0 14.0 0.40 1.75

The manganese content of INCONEL Welding Electrode 132 was

significantly increased (from 0.75 to 2.4%). However, the manganese
level remains well below that of Electrode 182. Manganese, in com-
bination with columbium, has long been used in the Ni-Fe-Cr coated
electrodes to provide resistance to weld-metal fissuring. The new
electrode iias excellent fissuring resistance as evidenced by the 2T
side-bend tests shown in Figure 4. The electrode was evaluated using
a variety of base metals: INCONEL alloy 600, INCOLOY alloy 800,
HUNTINGTON alloy HA-330, INCOLOY alloy DS, and carbon steel
(surfacing). Crack-free welds of extremely high quality were
achieved in all cases. The electrode has proved capable of meeting
the most demanding porosity standards in all welding positions.
Operability and slag removal are excellent. Additional testing is
under way.

Physical Properties

The improved electrode deposits weld metal having high strength

at room and elevated temperatures. Table VII shows all-weld-metal
tensile test data. Figure 5 shows a plot of stress-rupture
properties. Table VIII compares the 100- and 1000- hour estimated
stress-rupture strengths for Electrodes 132 and 182.
 680

Table VII. All-Weld-Metal As-Welded Tensile Data

Test
Temperature 0.2% YS TS % I11 % RA

Room 55,200 9«,300 43,.0 43.5

400 °F 46,900 87,700 42..7 50.0
600 "F 44,400 85,500 44.,3 ',4.7
800°F 44,400 83,900 46.,8 38.3
1000°F 42,900 78,500 47.,4 47,3
1200°F 39,900 65,800 38.,5 37.8
1400°F 36,100 45,600 53.J 55.0
1600°F 19,900 22,500 47. 3 68.8
1800°F 10,500 13,300 12. 6 17.3

Table VIII. Estimated 100-Hour and 1000-Hour Stress-Rupture

Strength (All-Weld-Metal Testa)

Estimated 100-Hour Estimated 1000-Hour

Strength (psl) Strength (psi)
1200"F 1400°F 1600°F 1800°F 1200°F 1400°F 1600°F 1800°F

132 36,000 17,000 5,700 1,950 24,000 10,000 2,700 1,150

182 34,000 - 3,850 1,700 23,000 - 1,900 840

Figure 4. Side-bend tests from a 1/2-in.-thick INCOLOY alloy 800 butt

joint welded with INCONEL Welding Electrode 132.
 681

100,000

1200°F

10,000

Stress
(psil

1,000

10 100 1,000 10,000 100,000

Stress Rupture Life (hours)

Figure 5. All-weld-metal stress-rupture results for INCONEL Welding

Electrode 132.

Oxidation Resistance

Figure 6 shows results of cyclic oxidation tests at 1800°F for

Electrodes 132 and 182 and INCONEL alloy 600. The redeveloped
Electrode 132 appears to be similar to alloy 600 and an improvement
over Electrode 182. Manganese is known to be detrimental to
oxidation resistance, and the higher manganese content of Electrode
182 may explain the electrode's lower oxidation resistance.

Metallurgical Stability

The redeveloped Electrode 132 does not appear to undergo any

sort of significant aging reaction other than the expected precipi-
tation of carbides. Figure 7 shows a side bend test taken from a
A-in.-thick butt joint in INCONEL alloy 600. The butt joint had been
stress relieved in a fully restrained condition.

Table IX shows all-weld-metal Charpy impact data for as-welded

samples and samples exposed to 1200°F and 1400°F for up to 2000
hours. There is some loss of impact strength; however, the level of
toughness retained after 2000 hours of exposure is excellent.
 682

+20

INCONEL alloy 600

IIMCONEL Welding Electrode 132

-20

-40
WEIGHT
CHANGE
mg/cm^
-60 -

-80

INCONEL Welding Electrode 182

I I I I
-100
200 400 600 800 1000
CYCLIC EXPOSURE TIME, h

Figure 6. Cyclic oxidation test results - 1800°F/15-5 cycle.

Table IX. All-Weld-Metal Impact Data

Heat Treatment Condition Impact Energy (ft. lbs)

As Welded 116-114-114
1200°F/500 hours 92-96-101
12OO°F/IOOO hours 86-88-91
1200 °F./2000 hours 80-80-79
I400°F/500 hours 89-90-87
1400°F/1000 hours 81-79-82
1400°F/2000 hours 71-67.5-82.5
 683

Figure 7. Side-bend test from a 4-in.-thick INCONEL alloy 600 butt

joint welded with INCONEL Welding Electrode 132. Stress relieved at
1050°F/50 hours in the fully restrained condition.

The stability of the improved Electrode 132 is a further

advantage over Electrode 182, which undergoes an aging reaction at
the common stress-relieving temperatures (1050-1250'F). The weld
deposit loses sufficient ductility that it will not pass a 2T bend
test in the stress-relieved condition. An example is shown in
Figure 8. The problem is more pronounced in heavy-section weldraents
(over about 2 in. thickness). The aging mechanism is not understood,
but is thought to be an ordering phenomenon in the areas of manganese
segregation.

Data Review

Manganese seems to play a key role in determining the properties

of Electrode 132. Weld-metal cracking resistance benefits from
higher manganese content. However, even higher manganese contents
apparently lead to some decrease in oxidation resistance and may be
detrimental to the metallurgical stability of the alloy system.

The mechanical properties of Electrode 132 appear to be

comparable to those of Electrode 182. The high strength along with
the other apparent advantages should make this welding electrode a
useful addition to the products available for joining INCONEL alloy
600 as well as dissimilar materials.
 684

Figure 8. Side-bend test from 2-l/2-in.-thick INCONEL alloy 600 butt

joint welded with INCONEL Welding Electrode 182. The weld had been
stress relieved in a fully restrained condition.

BIBLIOGRAPHY

1. H. L. Eiselstein, "Metallurgy of a Columbium-Hardened

Nickel-Chromium-Iron Alloy," Advances in Technology of Stainless
Steels and Related Alloys, ASTM Special Technical Publication
No. 369, Philadelphia, 1965.

2. D. J. Bologna, "Metallurgical Factors Influencing the Micro-

flssuring of Alloy 718 Weldments," Metals Engineering Quarterly,
Nov. 1969, pp. 37-43.

3. M. J. Lucas, C. E. Jackson, "The Welded Heat Affected Zone in

Nickel Base Alloy 718," Welding Journal, Feb. 1970, pp. 46s~54s.

4. J. Gordine, "Some Problems in Welding INCONEL alloy 718,"

Welding Journal, Nov. 1971, pp. 480s-484s.
 685

5. R. A. Douty, H. Schwartzbart, "Cobalt Base Surfacing of INCONEL

a l l o y 718," Welding J o u r n a l , Dec. 1973, pp 550s-556s.

6. Private communication, R. E. Crickmer, Huntington Alloys, I n c . ,

Huntington, West V i r g i n i a .

7. C. E. W i t h e r a l l , "Welding of Nickel-Chromlum-Iron Alloy for

Nuclear Power S t a t i o n s , " Welding J o u r n a l , Nov. 1960, pp
473s-478s.

i-U.S. GOVERNMENT PRINTING OFFICE: 1982-5A6067/148