CREEP-FATIGUE DAMAGE DEVELOPMENT DURING SERVICE-CYCLE

THERMO-MECHANICAL FATIGUE TESTS OF 1CrMoV ROTOR STEEL

Stuart R Holdsworth ¥, Edoardo Mazza § & Arnd Jung ¤

¥
ALSTOM Power, Rugby CV21 2NH, UK
§
ETH Zurich, CH-8092
¤
ALSTOM Power, CH-5401 Baden
[email protected]

ABSTRACT
Service-cycle thermo-mechanical fatigue (TMF) tests have been performed on a 1CrMoV
steam turbine rotor steel, with unexpected consequences in terms of the observed deformation
and damage development characteristics.
The tests were conducted on two heats of the rotor steel using service-like cycles without
and with hold times at peak temperature. The temperature and mechanical strain control
profiles of the cycles were devised to represent conditions which could occur at critical
locations in a high temperature steam turbine rotor during operation, both with respect to
shape and transient-rate. Details of the service cycle TMF test control regime are described.
A key feature of all the tests was the development of necking due to ratchetting, despite the
control of mechanical strain. Clues to the experimental cause of the effect are given in this
paper. A more detailed analytical explanation is provided in a companion paper in the
conference proceedings. The observations question what are currently regarded as acceptable
temperature control limits for TMF testing.
The local deformation phenomenon had a significant influence on the fatigue and creep
damage development characteristics exhibited by the two steels, as revealed by a
comprehensive post-test inspection activity. A range of physical damage mechanisms were
encountered, ranging from fatigue-dominated to creep-dominated, and these are rationalised
with reference to a creep-fatigue damage summation diagram for the steam turbine material.

INTRODUCTION
Critical locations in steam turbine rotors may be subject to the combined influences of cyclic
damage arising from strain transients generated during start-up and shut-down and creep
damage resulting from primary (directly applied) and secondary (self equilibrating) stresses
during operation. Traditionally, the risk of thermal fatigue cracking at such locations has
been assessed using the results from isothermal tests conducted at (or close to) the peak
operating temperature (e.g. [1-4]). 1CrMoV steel is widely used for HP/IP steam turbine
rotors operating at temperatures up to ∼565°C, and there is an extensive knowledge base
covering isothermally determined LCF and cyclic/hold properties of the steel (e.g. [5-9]).
One purpose of performing the service-cycle thermo-mechanical fatigue (TMF) tests
described in the following paper was to understand the practical implications of applying such
an approach to an anisothermal service situation.
Prior to the present study, there was only limited experience of TMF testing 1CrMoV rotor
steel. A Japanese study had shown that creep-fatigue endurances resulting from out-of-phase
300↔550°C TMF cycles were similar to those for isothermal tests at 550°C, for the same
mechanical strain range. Endurances resulting from in-phase 300↔550°C TMF cycles were
longer than those from equivalent isothermal tests at strain ranges below 1% [10]. This
evidence appeared to confirm the belief that the current practice of using isothermal test data
to predict service behaviour was conservative. However, in a second study employing
temperature acceleration to enhance the effect of 10 minute hold times by means of in-phase
300↔667°C TMF tests, endurances were lower than the equivalent isothermal tests, with
necking and off-centre barrelling experienced at high strain ranges (up to 9%) [11].
The new tests involve 350↔565°C TMF cycles comprising both in-phase and out-of-phase
components with characteristics which more closely represent the thermo-mechanical strain
histories experienced by rotor features in practice. Two unexpected features of the present
study were the development during the course of the service-cycle TMF tests of i) testpiece
necking (despite the adoption of a practically relevant strain range and thermal cycle) and
ii) apparently non-interacting creep and fatigue damage for one of the steel heats. The
following paper considers these features further. The results of analytical studies performed
to underpin the explanations for the observed behaviour are considered elsewhere [12-14].

EXPERIMENTAL DETAIL
Materials
Tests were performed on two heats of 1CrMoV rotor steel. The chemical compositions and
mechanical properties of the two production forgings are summarised in Table 1. Both steels
were oil quenched from 950/970°C and tempered at 695/700°C. The ambient monotonic
yield and tensile strength properties of Heat-1 were higher than those of Heat-2.
Uniaxial testpieces were taken from locations close to the periphery of each production
forging with a longitudinal orientation with respect to the axis of the forging.

Table 1 Material Details

C Cr Mo Ni V RP,0.2 RM
% % % % % MPa MPa
Heat-1 0.25 0.88 0.76 0.69 0.33 660 803
Heat-2 0.22 0.90 0.65 0.40 0.32 577 719

Testing
The TMF tests were conducted with a uniaxial testpiece having a parallel length of 18mm and
a diameter of 9mm.
The thermal cycle was applied using an induction heating coil linked to a 200kHz power
source (Figure 1). Temperature control was by means of a PtRh-Pt thermocouple wound
around the circumference of the testpiece at mid gauge length. During peak temperature hold-
periods, the temperature at the control thermocouple position was held to within ±1°C
throughout test. Mechanical strain control was achieved using a side-entry extensometer with
a datum leg spacing of 15mm.
The tests were performed in accordance with the requirements of the current draft ISO
Standard for TMF testing [15].
 testpiece

side-entry
control extensometer legs
thermocouple

Induction
heating coil

Figure 1 View of testing arrangement (temperature control is by means of spring loaded

thermocouple at middle of gauge length)

0.0 600 0.0

6 1 1 6
-0.2 1 -0.2
4 5 500
-0.4 -0.4 4,5
TE MP ERA TUE, C
o

-0.6
STRAIN, %

-0.6 400
S TRA IN, %

-0.8 -0.8
300
-1.0 -1.0

-1.2 200
-1.2
-1.4
-1.4
100 -1.6 2 3
-1.6
2 3 -1.8
-1.8 0
300 400 500 600
0 50 100
o
TIME , mins TE MP ERATURE , C

Figure 2 Service-like thermo-mechanical fatigue cycles (NB: the times given are those for
Cycle-1)

TMF Cycle Types

Three service-cycle types were applied, with the same cycle shape (Figure 2). In each case
the mechanical strain range was 1.4% and the temperature varied between 350 and 565°C.
For all cycle types, the mechanical strain was ramped from -0.2% to -1.6% (1-2), was held
at -1.6% (2-3) and was then ramped to -0.4% (3-4) while the temperature was increased from
350 to 565°C (3-4). The time to 565°C (1-4) was 50 minutes for Cycle-1, and 15 minutes for
Cycle-2 and Cycle-3 (Table 2).
In the case of Cycle-2, there was no hold time at the peak temperature (4-5), whereas for
Cycle-1 and Cycle-3 there were hold times at 565°C of 60 and 160 minutes respectively.
 The mechanical strain was increased from -0.4% to -0.2%, as the temperature was reduced
from 565 to 530°C (5-6), and held at -0.2% as the temperature was returned to 350°C (6-1).

Table 2 Details of TMF Service-Cycles

POSITION TEMP STRAIN CYCLE-1 CYCLE-2 CYCLE-3

o
IN CYCLE C % min min min
1 350 -0.2 0 0 0
2 393 -1.6 10 51 51
3 522 -1.6 40 10 2 10 2
4 565 -0.4 50 15 15
5 565 -0.4 110 15 185
6 530 -0.2 120 20 190
1 350 -0.2 130 30 200
1 2
421°C, 493°C

0.2

H eat-1, C ycle-1
H eat-1, C ycle-2
H eat-1, C ycle-3
H eat-2, C ycle-1
A RE A RE D UC TION

H eat-2, C ycle-2
H eat-2, C ycle-3
0.1

0.0
0.0 0.5 1.0

END URA NC E FRA C TION

Figure 3 Necking at centre of gauge length Figure 4 Variation of area reduction with
of service-cycle TMF testpiece endurance fraction and cycle type

POST-TEST INSPECTION
Necking
After test, the TMF testpiece gauge lengths were visibly necked (e.g. Figure 3). This is not a
characteristic feature of isothermal cyclic/hold tests on 1CrMoV rotor steel at practical
application temperatures and prompted a detailed post-test inspection activity.
At the end of test, local area reductions were typically ∼10%, slightly more for Heat-2,
slightly less for Heat-1 (Figure 4). Typically, necking occurred over the central ∼6mm of the
gauge length. Photographic observations made during the course of a Cycle-1 test indicated
that necking developed almost linearly with cycle number throughout the test (Figure 4).
Fatigue Damage
At the end of test, small transgranular fatigue cracks typically to a depth of ∼50µm were
distributed almost uniformly along the parallel length of the testpiece. Deeper transgranular
cracks were also observed, but only within the central necked section of the gauge length (e.g.
Figure 5). Quantitative measurement details of the fatigue damage assessment have been
reported elsewhere [16].

2mm 75µm

Figure 5 Fatigue-dominated cracking in Heat-1 (TMF Cycle-2 test)

2mm 100µm

Figure 6 Creep-dominated cracking in Heat-2 (TMF Cycle-3 test)

Creep Damage
As a generality, the creep damage observed in the post-test inspected TMF testpieces occurred
on prior austenite grain boundaries. Significantly, creep damage was not distributed
uniformly along the entire parallel length of these service-cycle TMF testpieces. Quantitative
measurement details of the creep damage assessment have been reported elsewhere [16].
Creep damage was mainly concentrated in a ∼7mm diameter x ∼10mm long central core in
the testpiece gauge section. The intensity of the damage was greatest in those testpieces
subject to Cycle-1 and Cycle-3 transients, and in particular the latter (with a 3h hold time at
565°C, Figure 6). For a given cycle type, creep damage was more developed in Heat-2
testpieces relative to Heat-1 testpieces.

Crack Development
The way in which cracking developed was significantly different for the two heats [16]. In
the case of Heat-1, crack development was always from the surface, being fatigue-dominated
in the Cycle-2 testpiece (e.g. Figure 5), close to fatigue dominated in the Cycle-1 testpiece
and by creep-fatigue interaction in the Cycle-3 testpiece. In contrast for Heat-2, cracking
originated at the surface and propagated by creep-fatigue in the Cycle-2 testpiece (Figure 7),
but developed internally in Cycle-1 and Cycle-3 testpieces (e.g. Figure 6).

2mm 40µm

Figure 7 Creep-fatigue cracking in Heat-2 (TMF Cycle-2 test)

DISCUSSION
Testpiece Necking
Necking of the order of magnitude observed in the service-cycle TMF tests (∼10%) is not
commonly experienced in cyclic/hold tests on 1CrMoV rotor steel at temperatures of ∼500-
550°C. Published experience of isothermal cyclic/hold tests with hold times up to 16h does
not acknowledge the presence of necking at crack initiation [5-9].
Significant necking has previously been observed during in-phase 300↔667°C TMF tests
on a 1CrMoV steel, but notably the peak temperature was within 30°C of the tempering
temperature of the alloy and mechanical strain ranges were up to 9% [11]. In such
circumstances, the incidence of necking is not surprising.
There is a notable difference between the temperature control specified for isothermal LCF
tests and TMF tests. In the case of isothermal tests, a temperature gradient over the gauge
length of the testpiece of ≤3°C is permitted for temperatures up to 600°C, although limits for
creep-fatigue tests of ≤±1°C are recommended (e.g. [17]). This compares with a permissible
gradient of ≤±0.01.T in TMF tests [15], or within 11°C at the peak test temperature in the case
of these service-cycle 350↔565°C tests. It has been demonstrated by analysis that the
observed gradient of ∼11°C at the gauge length extremities in these TMF tests can lead to
ratcheting and consequent necking in a 1CrMoV rotor steel [12,14].
 For a given nominal mechanical strain range, the observed necking leads to a significant
reduction in the Heat-2 endurances in Cycle-1 and in particular Cycle-3 TMF tests relative to
those observed in equivalent isothermal tests at the maximum test temperature (Figure 8).
The reduction is apparent but not as great in the Heat-1 TMF tests. The respective endurances
are rationalised by assessing the TMF test data as a function of local mechanical strain range
[12,14].
The direct application of the endurance results from these service-cycle TMF tests to
practical circumstances requires some care in view of the complexities introduced by the
necking arising from the high but apparently acceptable temperature gradient [12,14]. For
Heat-2 in particular, the evidence indicates that necking is responsible not only for locally
increasing the magnitude of the cyclic strain but also for influencing the damage mechanism
(Figure 8). It is almost certain that these complications would not have arisen in the absence
of a high gauge-length temperature gradient.

10

1C rMoV Rotor S teel

S TRA IN RA NGE , %

o
LCF/H eat-2/565 C
Fatigue dom inated
cracking
1
TM F Cycle-2 o
Cyclic/hold(1h)/Heat-2/565 C
TM F Cycle-1 Creep-fatigue cracking
TM F Cycle-3
Heat-1 - sm all sym bols
Heat-2 - large sym bols
O pen points - fatigue dom inated cracking
Light filled points - creep-fatigue interaction
Solid points - creep dom inated cracking

0.1
10 100 1,000
C YC LE S TO INITIA TION

Figure 8 Comparison of endurances and cracking mechanisms resulting from service-cycle

TMF and isothermal LCF and cyclic/hold tests for 1CrMoV steels

Damage Development
The development of creep-fatigue damage in isothermal cyclic/hold tests on 1CrMoV rotor
steel depends on temperature, strain range, hold time and the creep ductility of the material [5-
9]. In the absence of a significant hold time, crack initiation and growth is fatigue-dominated,
even at temperatures of ∼538-565°C (e.g. Figure 5). With increasing hold time and
decreasing strain range, the creep damage condition within the testpiece becomes increasingly
influential, to the limit beyond which crack development becomes fully creep-dominated (e.g.
Figure 6). At intermediate hold times and strain ranges, fatigue cracking interacts with creep
damage resulting in accelerated crack growth (e.g. Figure 7). The extent of any interaction is
likely to increase with decreasing creep ductility [7].
 In the service-cycle TMF tests, the crack development characteristics of Heat-1 are notably
different to those of Heat-2, for a given cycle type (Figure 8). In Heat-1 tests with hold time
at peak temperature, crack development is by a creep-fatigue interaction mechanism whereas
in Heat-2 tests with hold time, crack development is creep-dominated. In comparable
isothermal cyclic/hold tests of Heat-2 at 550°C (e.g. [6,9]), crack development occurs by a
creep-fatigue interaction mechanism. It is speculated, but still to be confirmed, that the creep
ductility of Heat-2 is sufficiently lower than that of Heat-1 to result in the observed damage
mechanism change as a consequence of the local change in stress state due to necking.

1.2

1C rMoV Rotor Steel Cycle 2

Cycle 1
1.0 Cycle 3
Heat-1 - sm all s ym bols
Heat-2 - large sym bols
O pen points - fatigue dom inated cracking
C YC LIC D AMAGE FRA C TION

0.8 Light filled points - creep-fatigue interaction

Solid points - creep dom inated crack ing

0.6

0.4

0.2

D am age sum m ation

locus [9]

0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2

C REEP D AMAGE FRAC TION

Figure 9 Creep-fatigue damage summation diagram

The TMF test results summarised in Figure 8 are re-plotted in a creep-fatigue damage
summation diagram in Figure 9. In this data comparison, cyclic damage fractions are
determined with respect to the respective Cycle-2 endurances for the two heats. Creep
damage fractions are based on strain (ductility exhaustion) fractions [9], determined from the
strains accumulated during the respective Cycle-1 and Cycle-3 peak temperature hold times,
and assuming upper and lower bound ductility properties respectively for Heat-1 and Heat-2
to represent the circumstances postulated above. The limiting damage summation locus
shown in Figure 9 was established using isothermal LCF and cyclic/hold test data for Heat-2
[9]. By assuming a significant difference in creep ductility properties, it is possible to account
for a difference in interaction characteristics between the two heats in these service cycle
TMF tests (and for a difference in interaction characteristics for Heat-2, as determined in
isothermal tests at 550°C [9] and the 350↔565°C TMF tests). Interestingly however, even
this analysis does not predict the observed creep dominated cracking in the Heat-2 Cycle-1
and Cycle-2 tests (i.e. creep damage fractions close to unity).

CONCLUSIONS
A series of service-cycle thermo-mechanical fatigue tests have been performed on two
1CrMoV rotor steels. Two unexpected features of the results from these tests were the
development during the course of test of i) testpiece necking and ii) apparently non-
interacting creep and fatigue damage for one heat of the steel. Both features are attributed to
the magnitude of the temperature gradient along the testpiece gauge length at peak
temperature during TMF cycling.

REFERENCES
1 Timo, D.P., 1969, 'Designing turbine components for low-cycle fatigue', Proc. Int. Conf.
on Thermal Stresses and Fatigue, Berkeley, 453.
2 Mayer, K-H. & Tremmel, D., 1979, 'The thermal fatigue of components in steam power
plant', Proc. Int. DVM Symp. on Low Cycle Fatigue Strength and Elasto-Plastic
Behaviour of Materials, Stuttgart, Oct-8/9.
3 Härkegard, G., 1984, 'Designing steam turbines for transient thermal loading', Proc. Int.
Conf. on High Temperature Structural Design, ESIS 12, Mech. Eng. Publ., London, 21.
4 Dawson, R.A.T., 1989, 'Monitoring and control of thermal stress and component life
expenditure in steam turbines', Proc. Intern. Conf. on Modern Power Stations, AIM, Liège,
September.
5 Bhongbhibhat, S., 1979, 'Untersuchungen über das werkstoffverhalten in Gebiet der
Zeitfestigkeit zur Erstellung von Berechnungsunterlagen für überwiegend thermisch
beanspruchte Bauteille', Techn.-wiss. Ber. MPA Stuttgart, Heft 79-02.
6 Thomas, G. Dawson, R.A.T, 1980, 'The effect of dwell period and cycle type on high
strain fatigue properties of 1CrMoV rotor forgings at 500-550°C', Proc. I.Mech.E. Conf.
on Engineering Aspects of Creep, Sheffield, Paper C335/80.
7 Miller, D.A., Priest, R.H. & Ellison, E.G., 1984, 'A review of material response and life
prediction techniques under fatigue-creep loading conditions', High Temperature Materials
and Processes, 6(3,4), 155-194.
8 Bicego, V., Fossati, C. & Ragazzoni, S., 1988, 'Low cycle fatigue characterisation of a
HP-IP steam turbine rotor', in Low Cycle Fatigue, ASTM STP 942, 1237-1260.
9 Holdsworth, S.R., 1996, 'Prediction of creep-fatigue behaviour at stress concentrations in
1CrMoV rotor steel', Proc. Conf. on Life Assessment and Life Extension of Engineering
Plant, Structures and Components, Churchill College, Cambridge, 09/96.
10 Kuwabara, K. & Nitta, A., 1977, 'Isothermal and thermal fatigue strength of CrMoV steel
for turbine rotors', CRIEPI Report E277005.
11 Westwood, H.J. & Lee, W.K., 1981, 'Creep-fatigue crack initiation in ½CrMoV steel',
Proc. Int. Conf. on Creep and Fracture of Engng. Mat. and Struct., Swansea, Pineridge
Press, 517-30.
12 Colombo, F., Masserey, B., Mazza, E. & Holdsworth, S.R., 2003, 'Service-like thermo-
mechanical fatigue tests for the lifetime assessment of turbine components', Proc. Conf. on
The Mechanical Behaviour of Materials, ICM9, Geneva, 25-29/05/03.
13 Colombo, F., Masserey, B., Mazza, E. & Holdsworth, S.R., 2003, 'Simple modelling of
the constitutive behaviour of a 1%CrMoV rotor steel in service-like thermo-mechanical
fatigue tests', Materials at High Temperature, 20(1).
14 Masserey, B., Colombo, F., Mazza, E. & Holdsworth, S.R., 2003, 'Endurance analysis of a
1%CrMoV rotor steel in service-like thermo-mechanical fatigue tests', Fatigue & Fracture
of Engineering Materials & Structures, submitted for publication.
15 ISO/TC164/SC5/WG9, 2002, 'Thermo-mechanical fatigue testing method', Working
Document.
16 Holdsworth, S.R., Mazza, E. & Jung, A., 2003, 'The Response of 1CrMoV Rotor Steel to
Service-Cycle Thermo-Mechanical Fatigue Testing', ASTM J. Test & Evaluation,
submitted for publication.
17 PrEN3988, 1995, 'Test methods for metallic materials - Constant amplitude strain-
controlled low cycle fatigue testing'.