iAL WOUNI

by

P.D. STEVENS-GUILLE

Chaik River Nuclear Laboratories

Chcik River, Ontario

August 1973

AECL-443O
 SUCCESSFUL APPLICATION OF
SPIRAL WOUND GASKETS
REQUIRES ASME CODE REVISIONS

by

P.D. Stevens-Guille

ABSTRACT

Spiral wound gaskets are widely used in piping and pressure vessel construction.
However, there is little published data on their characteristics. The ASME Code rules for
gaskets contain design data which have not been revised for the last thirty years. Test results
and other specification values indicate that they do not reflect current gasket manufacturing
practice. Their use can result in designs which provide insufficient bolt area to seat gaskets
less than about 5 inches diameter. Recommended design values are given for gaskets from
2V4 to 20% inches diameter. In the long term, the ASME Code Subcommittee on Pressure
Vessels is encouraged to consider the adoption of the design rules in the German Standard,
DIN 2505.

Chalk River Nuclear Laboratories

Chalk River, Ontario
August 1373

AECL-4430
 L'utilisation reussie de joints enroules en
spirales necessite la revision du Code ASME

par

P.D. Stevens-Guille

Resume

Les joints enroules en spirales sont tres employes dans la fabrication des cuves de
pression et des canalisations. Cependant, peu de donnees ont ete publiees quant a leurs
caracteristiques. Les normes du code ASME pour les joints comportent des donnees de
calcul qui n'ont pas ete revisees depuis une trentaine d'annees. Les resultats des essais et
d'autres valeurs montrent que ces donnees ne correspondent pas a la pratique actuelle de
fabrication des joints. Leur emploi peut donner lieu a des conceptions fournissant
insuffisamment d'espace pour loger les joints ayant un diametre inferieur a environ 5 po.
Les valeurs recommandees pour le calcul s'appliquent aux joints dont le diametre va de 2'/2
a 20 3/4 po. Eventuellement, le Sous-comite des Cuves de Pression charge de mettre a
jour le Code ASME devrait envisager l'adoption des regies de calcul figurant dans la norme
allemande DIN 2505.

L'Energie Atomique du Canada, Limitee

Laboratoires Nucleates de Chalk River
Chalk River, Ontario

Juillet 1973

AECL-4430
 TABLE OF CONTENTS

Page

1. INTRODUCTION 1

2. THE ASME CODE 1

2.1 Background 1
2.2 ASME Code Rules for Bolted Flanges 2
2.3 Discussion 2

3. NEW INFORMATION 2
3.1 Results of Tests 2
3.2 Information from Other Sources 4

4. SUMMARY 5

5. REFERENCES 5

APPENDIX A: GASKET CONSTRUCTION AND TEST METHOD 6

TABLES

3
1. RESULTS CALCULATED FROM TESTS
4
2. RECOMMENDED GASKET CONSTANTS
3. RESULTS CALCULATED FROM MIL-G-21032B 4
Al. CROSS SECTION OF SPIRAL WOUND GASKET
6
and GASKET CONSTRUCTION DATA

FIGURES

1. TYPICAL STRESS - DEFLECTION CURVE FOR A SPIRAL WOUND GASKET 3

AECL4430
 SUCCESSFUL APPLICATION OF SPIRAL WOUND
GASKETS REQUIRES ASME CODE REVISIONS

1. INTRODUCTION
hydrostatic end load progressively reduces the .'om-
Spiral wound gaskets are composite structures pressive load on the gasket until it is a minimum
made by winding a preformed metal strip and filler when the pressure is at a maximum. The gasket must
material over a mandrel. They are widely used in therefore be capable of expansion, and it is this
pressure vessel and piping construction, and parti- property of resilience which is least understood. The
cularly in nuclear service. They are suited for use in complex behaviour of gaskets in flanged joints is
assemblies subject to pressure and temperature simplified in design methods by specifying only two
cycling, shock and vibration. Power utilities and properties. These are the initial seating stress and a
nuclear vessel manufacturers use them in preference ratio indicating the eompressive stress to seal against
to any other type of gasket [1]. pressure.
An important disadvantage to the use of spiral In 1943 Rossheim and Markl [3] presented a table
wound gaskets is the lack of published data on their of gasket loading constants which was adopted by the
characteristics. Another drawback is the ASME Boiler ASME Boiler and Pressure Vessel Code, Section VIII
and Pressure Vessel Code design rules for gaskets. in toto. Their paper presented no experimental data
These rules contain gasket design data which has been or reasons as to their choice of the constants. This
published without revision in successive editions of important paper attracted only two written dis-
the code for the last thirty years. cussions. One strenuously criticized the authors for
This report attempts to provide new design theii lack of presenting any experimental evidence to
information for the application of spiral wound support the new constants, and the other suggested
gaskets. Attention is drawn to the empirical Code they had over simplified the problem and suggest/'d a
design data and preferred values are recommended precompression of doubie the seating stress value.
with the objective of obtaining leak tight joints. Rossheim and Markl's constants appear unchanged
in the latest ASME Codes:
2. THE ASME CODE Section VIII, Division 1, Appendix II (1971
edition with aridenda), Pressure Vessels.
2.1 Background
Section VIII, Division 2, Appendix 3 (1971
The search for leak tight joints is as old as edition with addenda), Pressure Vessels. Alternate
mechanical engineering. Serious analytical and experi- Rules.
mental studies were started in Germany, Britain and - Section III, Appendix XI (1971 edition with
the U.S.A. in the 1930's. Progress was made, Dut the Summer 1972 addendum), Nuclear Power Plant
main findings were that the problems were intrac- Components, Bolted Flange Connections for
table. A recent text book on the stress analysis of Containment Vessels.
pressure vessels confirms that the problems remain
largely unsolved and the major uncertainties are still - Section III, Nonmandatory Appendix E (1971
the gasket and joint behaviour [2]. edition),Nuclear Power Plant Components.

Initial tightening of the bolts is the key to leak This report limits discussion to stainless steel and
tight joint performance. A preload sufficient to seat asbestos spiral wound gaskets wholly confined within
the gasket and cause intimate contact with the flanges the flange bolt circle, subject to internal pressure
is essential. When fluid pressure is applied, tne only. External loads are not considered.

- 1 -
2.2 ASME Code Rules for Bolted Flanges The minimum bolt area is then calculated
from the largest bolt force found by dividing
Code rules for bolted flange connections with W -^ by the allowable stress at operating
gaskets wholly inside the bolt circle require calcu- temperature and W m 2 by the allowable stress
lations to be made for two independent conditions: at room temperature.
(1) Operating Conditions
2.3 Discussion
A bolt load W m j shall be determined by
Appendix S of Section VIII, Division l,Pressure
3.14 G 2 P
2b3.14GmP (D* Vessels states that the prime purpose of the rules for
bolted flange connections is to ensure safety. The
This equation states that the minimum impression created is that safety is proof against
required bolt load for operation is the sum of sudden rupture which could be hazardous to people
the hydrostatic end force (calculated on an or equipment.
area with a mean diameter G) plus a residual The current awareness in environmental protection
load acting on (approximately) the gasket demands that safety involves not only proof against
area times a multiple, m, of the design rupture but also the prevention of leakage of toxic
pressure. The recommended (i.e. non- chemical or radioactive materials during service.
mandatory) m value is 3.0 for stainless steel As shown below, the Code recommended value of
spiral wound gaskets. y, in particular, is so much lower than commercial
(2) Gasket Seating practice that joint designers may be in danger of
providing insufficient bolt area by using this value.
A minimum bolt load to seat the gasket at Insufficient bolt area will reduce the safety of the
room temperature without pressure is joint when the bolts are overloaded to achieve leak
determined by: tight joints.
Wm2=3.14bGy (2)* The ASME Code lists m and y values for stainless
steel and asbestos spiral wound gaskets as 3 and 4500
This equation states that the seating stress y
psi respectively. It points out that they have generally
is applied to (approximately) one half of the
proved satifactory in service. They are suggested
contact area, subsequently called the Code
values only. There is little published information
area in this report. The recommended y value
available to the joint designer so it is natural to resort
is 4500 psi for stainless steel spital wound
to the ASME Code values.
gaskets.
The Code rules condense the complex behaviour
*where: of gaskets to two equations and empirical design
properties. It is reasonable to suggest that the values
Wml minimum required bolt for opera- for these two properties be subject to periodic
ting coiidiuons, pounds scrutiny and revision as the state of gasket technology
W,m2 minimum required bolt load for demands.
gasket seating, pounds
G diameter at location of gasket load 3. NEW INFORMATION
reaction
3.1 Results of Tests
P design pressure, pounds per square
inch A series of 900 lb rating gaskets ranging from
b effective gasket or joint-contact- 2-7/32 in. to 20% in. outside diameter were tested in
surface seating width, inches compression to determine their stn-is-deflection
characteristics. Figure 1 shows a typical result
m gasket factor (dimensionless) obtained from one size of gaskets. AH gaskets were
y gasket, or joint-contact-surface compressed to the manufacturer's recommended
unit seating load, pounds per thickness; 0.130 0.005 in. for gaskets of 0.175 in.
square inch. nominal thickness and 0.095 0.005 in. for 0.125 ia

- 2 -
nominal thickness. Appendix A details gasket con-
struction and test methods.
Table 1 shows data calculated from the test
results. Column 1 shows the seating stress, designated
^ I (based on contact area calculated to Code rules).
Column 2 shows the ratio of y i to the Code y value.
The most striking result is that the actual seating
stress values are from 2.8 to 7.9 times the Code value
of4500!b/in 2 .
GASKET: FLEHTALLIC CO.
TIPE SPEC. Hi
This result is also reflected in the low values of 1 2 " li 0.D . H ' , , 0 . 0.175 .n. THICK
W m 2 calculated by the Code formula (equation 2) for LOflDED BETuEEf ?U1 PLarfT.

the case of pressure, P = 1600 lb/in2 (the primary

system design pressure of CANDU reactors). The
values are shown in column 4. Column 5 shows the
actual load required to compress the test gaskets to
recommended thickness. A comparison of columns 4 Figure 1Typical Stress-Deflection Curve for a Spiral
and 5 shows that th Code seating stress value is of Wound Gasket
no use in estimating actual seating loads. It misleads a
joint designer who has no other information. Values diameter W m j < actual seating load. Thus the joint
of Wm-^ calculated to the code formula (equation 1) designer may provide insufficient bolt area to cor-
are shown in column 3. In all cases for the design rectly seat these gaskets.
pressure chosen for this illustration, W m ^ > W m 2,
Appendix S of Section VIII discusses conside-
and for gaskets 12-13/16 to 20% in. diameter,W ml >
rations for the design of bolted flange connections. It
actual seating load. However for gaskets < 5Vi in.
states that for the great majority of designs, accep-

TABLE 1 RESULTS CALCULATED FROM TESTS

Gasket Data Bolt Loads for P -1600 Ib/in2 Gasket Factor

900 Ib rating Stress Code Calculations Actual
on Code
Load
on ID Thickness Area y , ' 1 '
(in) (in) (lb/uAclO3) (UndO3) (IbidO3) (ltalO3)

1 2 3 4 5 6

20 3/4 191/4 0.175 23.6 5.2 681 83 436 6.2

18 161/2 0.175 23.3 5.2 528 72 370 6.2

17 9/16 16 1/16 0.175 21.9 4.9 506 70 339 6.0

14 7/16 13 7/16 0.175 12.8 2.8 352 44 126 4.7

12 13/16 11 9/16 0.175 18.3 4.1 283 46 185 5.5

51/2 4 3/4 0.125 26.7 5.9 62 12 71 6.6

4 3/4 3 3/4 0.175 27.5 6.1 57 14 88 6.7

41/8 3 3/8 0.125 25.6 5.7 39 9 50 6.5

2 3/8 19/JS 0.175 35.5 7.9 18 5 43 7.5

2 7/32 111/16 0.175 21.4 4.8 12 3 14 6.0

(1) 1/8 ta.subtracted from nominal OD for gasket bead.

(2) Calculated from equation (3).

_ Q
table practice should be to use the Code rules for Table 3 lists date from this specification for 900 lb
design and tighten the bolts sufficiently to withstand rating gaskets. Values for y, and m, were calculated
the test pressure. in the same way as in Table 1. The y, seating stress
values range from 40,100 to 102,400 lb/in 2 . The
It is evident that the small test gasket bolts will be
average is over ten times the recommended Code
overloaded in practice in an attempt to seat the
value. Likewise, m, values calculated from equation
gasket. This is undesirable and may be unsafe.
(3) range from 8.0 to 12.4 and average about 3 times
Appendix S goes on to discuss bolt conditions where
the recommended Code value.
unusual conditions exist. But it is clearly not the
intent of the Code that common, small gasket joints Joint designers working to the ASME Code who
should be considered as unusu A. use MIL specification gaskets may expect difficulties.
It is not possible to calculate the m value The reasons for the large discrepancy between
required by the test gaskets as they are empirical calculated values from the test, results and the MIL
facturs determined by good engineering practice and specification compared with the Code recom-
used to compensate for external loads. However, an mendations are not easily explained. It is possible
approximate calculation is made which shows that that, since Rossheim and Markl's paper in 1943,
the Code values are low and maybe non-conservative gasket manufacturing methods have changed and now
from the leak tight viewpoint: result in higher seating loads. It is also possible that
the recommended values were too low even for 1943.
In Rossheim and Markl's original paper a relation-
The paper's lack of experimental data does not allow
ship between y and m is given:
speculation. The possibility that the test gaskets are
= 180(2m-l) 2 (3) not representative of current manufacturing practice
cannot be dismissed. However as they were selected
Assuming that this relationship is derived from arbitarily this does not appear likely. Other gaskets
gaskets that have proven satisfactory in service, tested at Chalk River Nuclear Laboratories have also
in values calculated from actual seating stresses and shown characteristics similar to the test gaskets.
designed AH, are shown in column 6. They range
from 4.7 to 7.5; ail are larger than the Code valuta
of 3.
i \QLE 3 RESULTS CALCULATED FROM M1L-G-2IU32B

Recommended gasket constants for the test

Uaskel I),ata
gaskets are shown in Table 2. The average of the
S'.iUSS
calculated y, values was ~ 20,000 !b/in2 for the large y()0 !b rating 0.175 in thick Gasket
Factor
t>D in ?'ip ftiw V^cuite
gasKets. (Values for one size of gasket fel! signifi- jinj (in) (ml ilb'in-\10 I 3

cantly below the average.) The average for >e small

gaskets was ~ 27,000 lb/in 2 (with one size ^ gaskets 1 7 '32 3/4 1/2 1-J2.4 22.B 12.)

significantly above average). Values of m, were 1 17; 13 1 J ! 67.4 ,5.0 10.2

calculated from equation (3). 1 27,32 1 1/4 1 66.3 ,1.7 10.U

2 11/32 111,16 1 1/4 5-1.3 12.1 9.2

2 23/32 2 '- 1/2 54.9 12.2 'J.2

TABLE 2: RECOMMENDED GASKET CONSTANTS
3 11/32 2 9/16 i 60.y US 9.7

3 27 32 2 15/16 21 2 44.4 9.U ri.4

1 23,32 3 3/4 3 4.1.4 mi 8.3

5 27.32 1 3/4 4 10.3 9.0 J.O

Gaskets 12-13/16 to 20% in. diameter 20000 5.8 ti 31/32 5 13/16 5 .l 9.8 8.3

Gaskets to 5VJ in. diameter 27000 6.6 a 7/32 6 7,8 G 40.1 S.9 6.0

10 11/32 1*7/8 8 47.4 10.5 8.6

12 1513?. 1013(16 10 48.7 10.8 B.7

14 23/32 12 7/8 12 48.3 10.7 8.7

3.2 Information from Other Sources

U.S. Military Specification MIL-G 21032B [4]

(lj _uaded between steel plates with concentric 0.003 luserratioiis 32/in. Area calculated by subtracting
covers spiral wound gaskets for standard ANSI pipe 1/8 in. bom nominal OD for gasket bead.

flanges and other gaskets for valves, pumps etc. (2) Calculated from equation (3).

(boiler manway and handhole gaskets are excluded).

- 4 -
 The Flexitallic Gasket Company have recognized bolt area to seat small gasjets. and hence the
the drawback of the Code y value. In a 1971 overall joint safety may be lowered when the
publication [5] they state, "our experience indicates bolts are overloaded to seat the gasket.
the minimum "y" value ... coupled with the effective
gasket seating width " b " ... is unrealistic". A y value 4.6 Recommended m and v values for the test gaskets
of 9000 lb/in2 is recommended for asbestos filled are given in Table 2. In the absence of published
spiral wound gaskets. This value is much lower than design data for other sizes of gaskets it is
that recommended in Table 2. They make no recommended that load-deflection tests be made
recommendations as to changing the m value. on gaskets. The v and m values may then be
calculated from:
The method used in Germany for the design of
flanged joints is specified in standard DIN 2505. It Load to deform to manufacturer's recommendation
differs from the ASME Code method by its relatively 3.14 bG (A)
simple approach to flange stress analysis. However, it
employs a sophisticated approach to leak tight and
behaviour. As structural failures in flanges are most
unusual, but sealing difficulties quite common, the /
DIN method focusses attention on this critical
feature. Gasket design values in the DIN method are
not directly comparable to the ASME Code method. 4.7 To design reliable leak tight joints, the designer
However it is instructive to note that the DIN requires more information than these simple tests
standard states, "gasket characteristics correspond to can provide. Such information concerns the
the latest state of the technology and will be time-dependent behaviour of spiral wound
continuously updated". gaskets under various service conditions. One
such program is underway at Chalk River Nuclear
Laboratories.
4. SUMMARY
4.8 In the long term the ASME Code Committee
4.1 The ASME Code reduces the complex behaviour dealing with flanged connections should be
of gaskets in bolted flange connections to two encouraged to give consideration to the design
equations and two design properties. It is methods of the Germany Standard, DIN 2505.
reasonable to subject them to periodic scrutiny
and revision when technological conditions 5. REFERENCES
demand.
1. "Static seals and their application in water-cooled
4.2 Results of tests on large (12-13/16 to 20% in.
nuclear reactor systems", ORNL-TM-3308, March
diameter) and small ( < 5Va in. diameter) spiral
1971.
wound gaskets show the actual seating stresses
are from 280 to 790% of the Code recommended 2. S.S.Gill, editor, "The stress analysis of pressure
value of y = 4500lb/in 2 . vessels and pressure vessel components", p. 314,
Pergamon Press,1970.
4.3 A paper by Rossheim and Markl published in
1943 is the basis of the Code values. A relation- 3. D.B. Rossheim, A.R.C. Markl, "Gasket-Loading
ship suggested in this paper was used to calculate Constants", Pressure Vessel and Piping Design, p.
values of gasket factor m from the test results. 87, Collected Papers, 1927-1959, American
They range from 4.7 to 6.2 for 12-13/16 to 20% Society of Mechanical Engineers, 1960.
in. diameter gaskets and 6.0 to 7.5 for gaskets <
5'/2 in. diameter compared to the Code value of 3. 4. Military Specification, "Gaskets, Metallic As-
bestos, Spiral Wound (For ASA Commercial
4.4 Calculated m and y values for gaskets conforming Flanged Joints in Piping Systems)", MIL-G-
to MIL-G-21032B support the test findings and 21032B Amendment 3, February 1963.
exceed the ASME Code values by large margins.
5. "Spiral Wound Gaskets: Design Criteria", Flexi-
4.5 A joint designer using the recommended Code tallic Gasket Company Inc., Camden, New Jersey,
values may be in danger of providing insufficient U.S.A., 1971.

- 5 -
 APPENDIX A
GASKET CONSTRUCTION AND TEST METHOD

WIDTH

-I U- PLY WIDTH

CROSS SECTION OF SPIRAL WOUND GASKET

TABLE AJ GASKET CONSTRUCTION DATA

1 2 3 4 6
OD ID Thickness Manufacturer's Width Pljd) Construction' '
in. width
in. in. in. Code No.
in^ply

20 3/4 191/4 0.175 Special 0.688 0.031 5-20-4

18 16 1/2 0.175 Special R 0.688 0.033 5-19-3

17 9/16 16 1/16 0.175 M-Round 0.688 0.038 2-17-4

14 7/16 13 7/16 0.175 Special R-3 0.431 0.022 6-16-4

12 13/16 11 9/16 0.175 Special R-3 0.562 0.022 6-22-2

51/2 4 3/4 0.125 R3-9K 0.308 0.025 6-10-4

4 3/4 3 3/4 0.175 CG-9J 0.437 0.022 6-16-5

4 1/8 3 3/8 0.125 R3-9H 0.320 0.024 5-11-3

2 3/8 1 9/16 1 0.175 CG-9E 0.350 0.025 6-11-5

2 7/32 111/16 0.175 Special 0.190 0.025 3-6-2

(1) 1 ply = 1 asbestos and 1 metal winding. The metal varied from 0.007 to 0.0085 in. thick.

(2) 5-20-4 construction = 5 inner metal windings, 20 metal and asbestos windings and 4 outer metal
windings.

- 6 -
A.1 Gasket Construction measured with an overall accuracy of + 0.5%. The
Tinius Olsen machine is calibrated periodically and
All gaskets used in the tests were supplied by the
the large Balwin machine is regularly calibrated with
Flexitallic Gasket Company. Table Al lists gasket load cells.
dimensions and construction before testing.
Deflections of the large gaskets were taken as the
A.2 Test Method average of the 4 dial indicator readings. Differences in
the readings showed that some parts of the gaskets
A 1.2 million pound Balwin Universal Testing compressed a greater amount than others. The maxi-
machine was used to compress the large gaskets. The mum slope of the loading plates was 7.7 x 10"" inVin.
small gaskets were compressed in a 120,000 pound which did not seem to have an effect on the test
Tinius Olsen machine. The test procedure was similar results. The 4 indicators were mounted on the head
to that specified in reference [4]. Three gaskets in of the Baldwin machine and one on the base of the
each size were tested. They were gradually com- Tinius Olsen. Deflection measurements were made
pressed (0.025 in/minute) between hard flat plates with little error as it was unlikely that the testing
with a surface finish of 63 M'n. or less. The plates for machines deflected to influence the readings.
the large gaskets were spherically seated to equally
The gasket seating stress (column 1, Table 1) was
distribute the load. Deflection of the large gaskets
calculated from the Code formula (2):
was measured under load by 4 equi-spaced dial
indicators graduated in 0.0001 in. divisions. The zero
setting on the indicators was made with a load of
5000 lb for the large gaskets (equivalent to a
maximum of 500 lb/in 2 ). The plates for the small 3.14 bG
gaskets were not spherically seated and only one dial
indicator was used to measure deflection. The
nominal 0.175 in. thick gaskets were compressed in at
least 7 steps to the manufacturer's recommended where W 2 1S t n e ' o a ^ to con >press the gasket to
value of 0.130 0.005 in. Nominal 0.125 in. thick 0.130 in. for 0.175 in. thick and 0.095 in. for 0.125
gaskets were similarly loaded to 0.095 + 0.005 in. in. thick gaskets. Load-deflection curves were extra-
They were then unloaded, reloaded and unloaded polated to the required compression for some gaskets
again. which were not quite fully compressed. Values of y,
were calculated from nominal gasket dimensions.
Although the large gaskets were out of round by an
A.3 Measurements and Accuracy
average of 0.1 in. before test, their areas were within
The compression load applied to the gaskets was 2.5% of the areas based on nominal dimensions.

- 7 -
 i

A d d i t i o n a l copies of this d o c u m e n t
may be o b t a i n e d f r o m
S c i e n t i f i c Document D i s t r i b u t i o n O f f i c e
A t o m i c Energy of C a n a d a Limited
C h a l k River, O n t a r i o , C a n a d a
KOJ 1J0

Price - 50C per copy

1869-73