BULLETIN 302 FEBRUARY 1985

ISSN 0043-2326

Welling Research Council

bulletin
POSTWELD HEAT TREATMENT OF
PRESSURE VESSELS

R . D . Stout

RELAXATION STRESSES IN
PRESSURE VESSELS

P . S . Chen
W . A . Herman
A . W . Pense

A STUDY OF RESIDUAL STRESS

IN PRESSURE VESSEL STEELS

R . J . Zhou
A . W . Pense
M . L . Basehore
D . H . Lyons

These Bulletins contain final Reports from projects sponsored

by the Welding Research Council, important papers presented before
engineering societies and other reports of current interest.

WELDING RESEARCH COUNCIL

UNITED ENGINEERING CENTER
345 EAST 47th STREET, NEW YORK, N .Y . 10017
INTENTIONALLY LEFT BLANK
 WRC - The Welding Research Council brings together science and engineering specialists in
developing the solutions to problems in welding and pressure vessel technology. They exchange
knowledge, share perspectives, and execute R and D activities. As needed, the Council organizes
and manages cooperative programs.

MPC – A Council of the WRC, the Materials Properties Council is dedicated to providing
industry with the best technology and the best data that can be obtained on the properties of
materials to help meet today’s most advanced concepts in design and service, life assessment,
fitness-for-service, and reliability and safety.

PVRC – A Council of the WRC, the goal of the Pressure Vessel Research Council is to
encourage, promote and conduct research in the field of pressure vessels and related pressure
equipment technologies, including evaluation of materials, design, fabrication, inspection and
testing.

For more information, see www.forengineers.org

WRC Bulletins contain final reports from projects sponsored by the Welding Research Council, important
papers presented before engineering societies and other reports of current interest.

No warranty of any kind expressed or implied, respecting of data, analyses, graphs or any other
information provided in this publication is made by the Welding Research Council, and the use of any
such information is at the user’s sole risk.

All rights are reserved and no part of this publication may be reproduced, downloaded, disseminated, or
otherwise transferred in any form or by any means, including photocopying, without the express written
consent of WRC.

Copyright © 1985 The Welding Research Council.

All rights, including translations, are reserved by WRC.
Printed in the United States of America.

ISSN 0043-2326
Library of Congress Catalog Number: 85-647116

Welding Research Council

20600 Chagrin Blvd.
Suite 1200
Shaker Heights, OH 44122
www.forengineers.org
INTENTIONALLY LEFT BLANK


Postweld Heat Treatment of Pressure Vessel Steels

by R . D . Stout

CONTENTS

Introduction 1 sels . This review is an attempt to clarify the present

Methods of Postweld Heat Treatment 1 status of PWHT as a fabrication tool and to provide a
PWHT Parameters 1 background for defining the conditions under which
Heating Methods 2
postheating is necessary and those in which it may be
Mechanical Effects 2 detrimental.
Residual Stress Generation 2 The discussion of PWHT will be limited to treat-
Thermal Stress Relief 3 ments performed in the temperature range of 400C to
Benefits of Thermal Stress Relief 4
just below the transformation temperature of the steel
Metallurgical Effects of PWHT on Parent and to steels with minimum yield strengths up to 700
Metals 4 MPa (100 ksi) . Temperatures below this range largely
Carbon Steels 4 fail to accomplish the desired effects, while treatments
Unhardened Low Alloy Steels 5
Quenched and Tempered Steels 5 into or above the transformation range evoke an entirely
Cold Formed Steels 6 different set of responses in the steels that are accom-
Metallurgical Effects on Weldments 7 panied by an equally different set of procedures and
Tensile Properties 7 problems.
Hydrogen Behavior 8
Stress-Relief Cracking 8 Methods of Postweld Heat Treatment
Notch Toughness 9
Fatigue Resistance 12 PWHT Parameters
Bases for Refining Conditions Requiring The variables of PWHT that are significant to the
PWHT 12 results obtained are essentially (1) the heating rate to
Alternatives to PWHT 12 the desired temperature, (2) the maximum temperature
imposed, (3) the time during which the maximum
Need for Additional Research 13
temperature is maintained in the steel, and (4) the
References 13
cooling rate from the maximum temperature to room
temperature . Each of these phases has a bearing on the
final outcome and must be controlled to produce opti-
mum results from PWHT.
Introduction From the viewpoint of economy, the rate of heating
should be as rapid as practical, and in some steels the
Postweld heat treatment produces both mechanical time spent at intermediate temperatures may be
and metallurgical effects in pressure vessel steels that harmful to their integrity or properties . On the other
will vary widely with the composition of the steel, its hand, excessive heating rates are likely to generate
past thermal and mechanical history including welding, sizeable temperature gradients that can induce cracking
the temperature and duration of the postheat, and the or shape distortion in the as-welded structure . Also, in
heating and cooling rates accompanying postheat . When massive structures the heating rate is limited by the rate
the majority of welded pressure vessels were fabricated at which heat can be transferred to the interior by ra-
from carbon steels, postweld heat treatment (PWHT) diation, convection, and conduction.
was performed to relieve the locked-up stresses resulting The temperature of PWHT is selected by consider-
from fabrication, and the process was identified as ation of the desired changes that are sought in the
"stress relief." As higher-strength alloy steels were ap- structure . For example, stress relief increases progres-
plied to pressure vessels, it was realized that there were sively with temperature and may be inadequate unless
other effects involved, some of which were beneficial a high enough temperature is reached . Certain alloy
and others detrimental to the performance of the ves- steels are embrittled rather than toughened at tem-
peratures around 500° C (930° F), which should
therefore be avoided for these steels . At the upper limit,
R. D. Stout is a consultant and is with the Department of Metallurgy & Ma- PWHT temperatures cannot exceed the tempering
terials Engineering at Lehigh University, Bethlehem, PA . temperature of heat-treated steels if they are to retain

Postweld Heat Treatment of Steels 1



the specified strength level . The upper bound of tem- Residual Stress Generation
perature in any event is the initiation of phase trans- By its very nature the welding process invariably in-
formation which, depending on the steel composition troduces locked-in stresses in the welded structure . The
varies from 600 to 750° C (1100—1380° F). responsible factors are (1) a sharp temperature gradient,
Since most of the phenomena which are produced by (2) variation of the flow strength and dimensions of the
the PWHT process are not instantaneous, the holding parent and weld metals with temperature, (3) the pro-
time at temperature must be regulated to allow the gressive or stepwise deposition of the joining metal, and
desired time-dependent actions to take place . To a very (4) volume changes accompanying transformation of
limited degree holding time and temperature are in- the metal during cooling . Broadly the sequence of events
terchangeable, but small temperature changes have is as follows:
been shown to be equivalent to large changes in holding
1. Liquid metal is deposited or generated by the heat
times . Deleterious effects on ductility and toughness
1>2

of the arc and the immediately adjacent unmelted

can be produced by prolonged heating at lower or higher
metal is heated to a high temperature that falls off
than normal PWHT temperatures.
sharply with increasing distance from the fusion
The cooling phase is controlled primarily to avoid
zone.
thermal gradients in the steel that would regenerate
2. As the arc progresses along the joint, the weld zone
residual stresses or distort the structure harmfully.
left behind cools rapidly by heat flow into the
Heating Methods surrounding cold metal . The solidified weld metal
Because all phases of the thermal cycle of PWHT are and the adjacent heated metal contract in volume
with the decreasing temperature.
pertinent to satisfactory results, the most suitable
3. The natural contraction of the hot (and low
procedure is to heat the weldment in a totally enclosed
facility . The common method is to place the weldment strength) metal is impeded by the surrounding
in a furnace heated by electricity or fuel and capable of cold, stiff metal and as it cools is forced to adjust
to the dimensions of its environment by plastic
regulation that will ensure the specified uniform heat-
deformation.
ing, temperature hold, and cooling cycle . The weldment
4. As cooling proceeds the weld metal gains strength
must be supported in a way to avoid distortion due to
and exerts increasing resistance to the restraint of
sagging and to allow uniform application of heat to the
the surrounding metal, so that the stresses re-
exposed surfaces . In the case of very large structures or
quired to continue plastic flow correspond roughly
structures to be postheated in the field, temporary en-
with the yield strength of the metal at its existing
closures or insulation are built around the structure for
temperature . Preheating is beneficial in this con-
PWHT. Care must be taken to avoid nonuniformity of
temperatures in these installations . Another feasible nection, because it reduces the thermal gradient
technique is to provide an effective insulating enclosure at the weld and thus the plastic adjustment re-
quired.
for the pressure vessel and inject heat into the vessel
5. At the time cooling is completed the welded joint
interior .
3

Vessels of unusual length, such as kilns or refinery will contain residual stresses comparable to the
columns have been postheated by placing each end of metal yield strength at the final temperature.
the vessel into the furnace for the heating cycle in suc- While these are the events accompanying any given
cession with sufficient overlap to secure overall treat- welding pass, the consequences become more complex
ment . This method relieves stresses satisfactorily if the in welded fabrication that involves multiple pass
structure configuration is not complicated, but it welding, constraint imposed on the joint by the struc-
subjects the overlap section to intermediate tempera- ture configuration and previous welding, external loads
tures that may be unfavorable to the properties of some used to force fit-up, and local preheating . In service
alloy steels or welded joints. structures the distribution, signs and magnitudes of the
Local PWHT has sometimes been proposed for crit- residual stresses are affected by the interplay of these
ical areas of structures too large for total treatment. factors . If two large plates otherwise unattached are
Heating is accomplished by strip heaters, low-frequency welded together by a multiple-pass butt weld, the re-
induction coils, or gas burners, together with partial sidual stresses in the weld metal longitudinal and
enclosure or insulation . 4 The selected welded joint can
>5
transverse to the weld axis are about as shown in Fig . 1.
be tempered by local heating and peak residual stresses Note that (1) the peak residual stresses approach the
lowered, but unless precautions are taken there is the yield strength of the material (which is 375 MPa), (2)
risk of introducing new stresses when the locally heated the longitudinal stresses are higher than the transverse
region cools and contracts against the restraint of the and are tension in sign except for a small region near
parent structure . Accurate control of temperature is midthickness, and (3) the transverse stresses are bal-
more difficult in local heating, and possible shifting of anced in equilibrium between tensile and compression.
stresses to other structural details must be consid- If the plates to be butt-welded are mounted in a rigid
ered. frame (or structural configuration) so that they cannot
move inward when the butt joint contracts, the
Mechanical Effects transverse residual stresses acquire a tensile component

2 WRC Bulletin 302



MPa
MPa MPa
for alleviation of the stress by distortion . In very heavy
-100 100 300 300 100 100 300 -300 -100 100 300
sections the stresses may be skewed or highest in the
adjacent base metal . Increasing weld width or the
number of passes also creates more weld shrinkage and
with it more distortion or residual stress . For minimizing
residual stresses, the benefit of preheat is not always
recognized . By reducing the temperature span over
which weld contraction must take place preheating can
reduce the level of the final residual stresses if properly
applied . Since local preheat can also intensify reaction
A . LONGITUDINAL B . TRANSVERSE C . NORMAL TO SURFACE stresses when the local preheated region cools down and
STRESSES ,Cr, STRESSES, O STRESSES, 0 z - contracts against the surrounding metal, the preheating
must be applied in a pattern to avoid this effect.
Fig . 1—Distribution of residual stresses in the thickness direction in
the weld metal of a butt joint 29 Thermal Stress Relief
Basically the thermal relief of residual stresses is
accomplished by heating the welded structure to a
temperature range high enough to reduce the yield
as illustrated in Curve 2 of Fig . 2 . These additional
strength of the steel to a small fraction of its magnitude
stresses, called reaction stresses, may raise the surface
at ambient temperature . Since the steel can no longer
transverse stresses to tensile yield strength magni-
sustain the residual stress level, it undergoes plastic
tude.
deformation until the stresses are relaxed to the at-
The pattern of residual stresses is affected by the
temperature yield strength. Additional relaxation will
dimensions of the structure and the parameters of
take place by a creep action if postheating is continued.
welding. Generally residual stresses develop a higher
average level as the section thickness increases because Fig . 3 shows the degree of stress relaxation obtained in
a carbon steel as a function of temperature and time.
structural rigidity is higher and there is less opportunity
The interaction of time and temperature in relaxing
residual stresses by creep can be described satisfactorily
by a parameter P identified as the Larson-Miller pa-
rameter ) which takes the form:
P = T(C + log t)
where T is the absolute temperature of heating,
deg K
t is the time at temperature in hours
C is a constant equal approximately to 20

As will be referred to later, this relation implies that

Y temperature has a far stronger effect on stress relief
A. BUTT WELD than time . Thus doubling the time at temperature has
Y

STRESS RELIEVING TEMPERATURE, °F

600 800 1000 1200
Y TIME AT STRESS
B . DISTRIBUTION OF CTx ALONG YY RELIEVING TEMP.
I . I hr
2. 4 hr
TENSION— _____ Ty -CURVE 3. 6 hr
REACTION STRESS — CURVE I
X- . .. .11111111111111h._ _
X
I I "'" 4 41 W
80
COMPRESSION
_-
100315 370 430 480 540 595 650 705
C . DISTRIBUTION OF Cry ALONG XX STRESS RELIEVING TEMPERATURE, *C

Fig . 2—Typical residual stress distribution in a butt weld 3) Fig . 3—Stress relief in a carbon steel' )

Postweld Heat Treatment of Steels 3



about the same effect as raising the absolute tempera- and tensile stresses in the interior . If the surface layer
ture 1 .5%, or at 600° C (1110° F) doubling time would on one side is removed by machining or grinding, the
be equivalent to a rise of 13° C (23° F). balance of residual stresses will be disturbed and the
The effect of steel composition on the process of slab will tend to arc or bend toward the machined side
thermal stress relief is consonant with the effect of the causing it to become concave . Weldments likewise can
alloying elements present on elevated temperature yield distort under the action of local peak stresses and re-
strength and creep . 6 ' 7 The behavior of a representative action stresses when machining is performed.
group of steels is shown in Fig . 4 . The carbide-forming Ordinarily carbon and low alloy steels exposed to
elements such as Mo, Cr, V, and Nb enhance the neutral or basic solutions show a tolerable rate of cor-
strength of steel at PWHT temperatures and conse- rosion if the oxygen content of the solution is not high.
quently require higher temperatures for stress relief. If high tensile stresses are present at the steel surface,
The interplay of time and temperature of PWHT is not the general corrosion rate is only moderately raised, but
altered by the presence of alloying elements even though a localized action develops known as stress-corrosion
the relation of stress relief to the Larson-Miller pa- cracking . The most important instance is the type called
rameter may be considerably modified by them. " caustic embrittlement," which is stress-corrosion

Benefits of Thermal Stress Relief cracking encountered in hot hydroxide solutions used
in the paper industry, but it has also been observed in
While PWHT does not remove residual stresses
chloride-bearing solutions, in ammonia, and in nitrates.
completely, it can lower the peak stresses to 10–20% of
The mechanism is a cooperative action of local corrosion
their as-welded level depending on the material and the
PWHT temperature . This is sufficient to obtain several attack causing pits or crevices and tensile stresses
causing cracking at the root of these imperfections.
valuable effects on the mechanical behavior of the
Cracking accelerates corrosion at the crack tip and in
weldment . These include dimensional stability, re-
turn further cracking ensues . In weldments cracking
duction or elimination of stress-corrosion cracking, and
often occurs at right angles to the weld axis since the
improved load-carrying capability in the brittle-fracture
longitudinal residual stresses at the weld joint are
temperature range of service . The effects of PWHT on
usually the highest tensile stresses present . Thermal
fatigue resistance are discussed later.
stress relief is effective in reducing or preventing
The use of thermal stress relief to assure dimensional
stress-corrosion cracking because it greatly reduces the
stability in steel structures predates fabrication by
tensile residual stresses essential to the process.
welding . In the machining of large forgings and castings
The effect of residual stresses on the brittle fracture
it was recognized that removal of metal to obtain the
of steel weldments has long been a subject of contro-
final desired dimensions could result in intolerable
versy. 8- 11 Large scale tests, such as the wide-plate tests,
changes in the shape of the part, such as loss of flatness
have shown that residual stresses have no effect on the
or shifts in diameter or length . The source of these
transition temperature behavior of welded steel . At
changes was identified as residual stresses whose pat-
temperatures low enough to cause very low notch
tern was altered by removal of metal that had supported
toughness, residual stresses may trigger crack initiation
tensile or compressive stresses . As a simple example, a
at a flaw at low external loads . Also cracks may be
slab of steel cooled rapidly from a high temperature (not
triggered in locally embrittled metal in or adjacent to
necessarily above the transformation temperature) will
the weld at low loads or occasionally spontaneously.
develop residual compressive stresses on the surfaces
These cracks may be arrested if they travel into tougher
microstructures . Therefore it appears that relief of re-
sidual stresses is beneficial to avoiding brittle fracture
only when the operating temperature is below the
transition temperature of the steel structure, or when
300 partial cracks may be initiated in locally embrittled
material at the weld joint at low loads.
U)
U)
w
Metallurgical Effects of PWHT on Parent Metals
200
u)
Carbon Steels
Carbon steel grades are commonly used in weldments
in the as-rolled or normalized condition . The air cooling
through the transformation range produces a micro-
structure in these steels of pearlite and proeutectoid
ferrite formed at temperatures above the normal
0 PWHT range, and thus these steels are relatively stable
16 17 18 19 20 when postheated . Fig. 5 shows that the tensile proper-
LARSON-MILLER PARAMETER ties are little altered unless the time of heating is pro-
longed or higher than usual temperatures are employed.
Fig . 4—Influence of alloy content on thermal stress relief 31 The loss of strength is associated with a partial

4 WRC Bulletin 302



5 .0 1 I Unhardened Low Alloy Steels

• A537 CI .I STEEL A variety of low alloy steels has been developed to
provide improved yield strength, toughness, and
0 weldability in the as-rolled or normalized condition . If
the carbon content is restricted, preferably below 0 .15%,
PWHT of these steels is ordinarily not necessary unless
.` dimensional stability or stress-corrosion is of concern.
-5 .0 •
J •
W • If strengthening is obtained by small Ni, Cr, Cu, and/or
• Mo additions, the effects of PWHT on strength and
z
-10.0 notch toughness is not markedly different from those
w
O observed in carbon steels . Fig. 7 illustrates the magni-
z - --!
Q tude of the changes in several representative grades.
x In the past 20 years an increasing use has been made
U -15 .0
18 .0 19 .0 20 .0 of elements such as V, Nb, and Ti added to steels in
amounts less than 0 .1% to obtain HSLA steels with yield
strengths over 350 MPa (50 ksi) without special heat
= 0 0 0
C, treatment . While PWHT is seldom involved in struc-
z o\ •
\o tural applications, these steels may be postheated in
w
cc 60 N
e pressure-containing weldments where their response

q0 I 0 q
y -5 .0
to PWHT becomes significant . Steels containing V or
u)z • ° •q 6 Nb derive some of their strength from finely distributed
w • A537 CI .I •A A
-. 5 \1c. - carbides or carbonitrides of these elements . PWHT
-10 .0 --•-, °
z q A537 CI .I *0 softens these steels and decreases notch toughness by
o A537 CI .2 a coarsening of the carbides and migration to ferrite
w
co o A516 Gr .70 grain boundaries . As shown in Fig . 8, the effects of
1 , I
a -15 .0 PWHT become significant as holding time is in-
x 18 .0 19 .0 20 .0
U creased.
LARSON-MILLER PARAMETER
Quenched and Tempered Steels
Fig . 5—Effect of PWHT on the yield and tensile strengths of C-Mn
The behavior of quenched and tempered steels
steels 32
subjected to PWHT is strongly dependent on their
composition and the time and temperature of PWHT.
To preserve the strength requirements of quenched and
spheroidization of carbides in pearlite. tempered steels, the PWHT temperature must be kept
As indicated by Fig . 6, carbon steels exhibit a pro- safely below the tempering temperature specified . From
gressive loss of notch toughness with longer times and the Larson-Miller parameter, which is equally appli-
hgher temperatures of PWHT . 12,13 It is likely that mild cable to the tempering process as it is to the relief of
embrittlement occurs at the ferrite grain boundaries by residual stress, the effect of PWHT can be limited to a
migration of carbon or impurity elements . These effects small increment of the previous tempering effect . For
of PWHT are reversed only by normalizing. example, 10 hrs at 595° C (1100° F) PWHT has the
same parameter value as 8 min tempering at 675° C
(1250° F) . Nevertheless the risk of extending the time
or raising temperatures too close to the tempering
1 0 d temperature remains . Some specific examples of
q A285
o A5I6-70 A
o
•
A2I2 4 U
A36
O NOTES ARROWS INDICATE CHANGE IN PROPERTY FROM
I hr . TO 100 hr . PW HT _ 0 a
-r( I-
A Cn

A - 500 - -30 a
f-
E
-60'?

0 • a.
• FL

-90 x
▪ A2O3D A302B A387D HY-65 A2O3D A302B A387D HY-65 U
17 18 19 20 STEELS
LARSON-MILLER PARAMETER
Fig . 7—Effect of PWHT on strength and notch toughness of unhardened
Fig . 6—Loss of notch toughness in carbon steels after PWHT 32 HSLA steels 13

Postweld Heat Treatment of Steels 5



I PWHT cycles appropriate to thinner (under 35 mm)

o o A737 Gr .0 NORMALIZED
.
12)
• • : A737 Gr . B NORMALIZED and smaller structures do not alter the notch toughness
FAIT of any of the grades of quenched and tempered alloy
steels. However, in heavy sections requiring longer hold
times and sometimes repeated PWHT during fabrica-
FATT tion, significant embrittlement may be incurred in
-20 susceptible grades . As indicated by Fig . 10, grades such
as A203D, A302B, and A533-B2 show only slight
changes in notch toughness as a result of 100 hrs
PWHT, but some grades including A387D and A517F,
lose appreciable notch toughness . The difference in
these two sets of steel lies in the presence of strong
TS carbide formers in the second set . It should not be
80 overlooked that the heat-treated grades have high initial
toughness and therefore may have ample toughness
0
after PWHT for all but the most critical applications.
f 500
When the cooling rates following PWHT are slow, the
time of exposure in the 400–550° C (750–1025° F) range
may be long enough to induce temper embrittlement in
U) steels containing carbide formers. As Fig . 11 demon-
YS 60 }
strates, loss of notch toughness in susceptible steels is
observed after a few hours at these temperatures . 14
These losses may or may not be significant to service
50 performance.

AR 1 .0 2 .0 3 .0 (LOG HR) Cold Formed Steels

(2) (10) (30) (100)(300) (HR) Cold forming is used to produce pressure vessel sec-
tions, pipes, and special shapes for welded structures.
HOLDING TIME AT 620°C
The forming can introduce an amount of plastic defor-
Fig . 8—Effect of PWHT on the tensile and Charpy properties of A737 mation that depends on the applicable code and the
normalized plates 31 material. If the service temperatures are in the 200–350°
C (400–650° F) range, the strained regions experience
strain-aging . Both the tensile properties and notch
toughness are affected by straining and aging. PWHT
strength losses in quenched and tempered steels during of cold formed steels offers the opportunity for resolu-
PWHT are illustrated in Fig . 9. 13 tion of the precipitates and possibly recovery and partial
The effects of PWHT on the notch toughness of recrystallization of the strained ferrite . Fig . 12 13 shows
quenched and tempered steels require more detailed the effects observed in the laboratory of 5% plastic
consideration of compositional factors . Generally the straining, aging, and PWHT on the tensile properties
of a variety of steels, while Fig . 13 shows the corre-
sponding changes in notch toughness of these steels.

1000 NOTE . ARROWS INDICATE LOSS OF STRENGTH -

FROM I hr . TO 100 hr . PWHT

a U
a- °
NOTE ARROWS INDICATE CHANGE OF TOUGH-
a NESS FROM I hr . TO 100 hr . PWHT
750
x 2
H- w
H
C7
z -50
w z
a LA
I- 500
to I-
F
W
J
rA
n Lo E -100 Aa
z CD w CN
D CD CD CD

w 250 I- I- U
I-
x x x >- 0 0
3a CO 0
3
a
3
a
3
a
3
a
3
a
a.
tr w CD CD C0
r
-150
A203D A302B A533-B2 HY80 A517F A387-22 A203D A302B A533-B2 HY-80 A517F A387-22
V
STEELS STEELS

Fig . 9—Effect of PWHT on the tensile strength of quenched and tem- Fig. 10—Effect of PWHT on the notch toughness of quenched and
pered steels 13 tempered steels 13

6 WRC Bulletin 302



25 z
Q
r
1-

U
0

I-
I-

325 650 325 650

AGING TEMPERATURE , °C
N+T N+T
0+T N+T 0 +T
A204C 43028 A533 43878 A542 Fig . 13—Effect of 1 hr postheating on the notch toughness of steels
strained 5% 13
STEELS

Fig . 11—Effect of PWHT followed by step cool on notch tough-

ness 14

the plastic strain induced in the weld metal as its cools

and tries to contract under the restraint of the joint . The
fine grain size, finely dispersed carbides, and cold strains
The tensile properties are largely restored to unstrained
combine to produce these properties. Other than to
values by PWHT . On the other hand, the notch
reduce residual stress, PWHT generally is not very
toughness is completely restored by PWHT in some
beneficial to weld metal . The response of a variety of
steels but only partially recovered in others . It is be-
weld metals to PWHT with regard to tensile properties
lieved that failure to recover toughness is associated
is shown in Fig . 14 . In almost all cases the yield and
with the precipitates which form or persist during
tensile strengths are lowered with moderate gain in
postheating. The Nb or V-containing steels exhibit this
ductility . The effects are enhanced by higher temper-
behavior . Overall the strain-aging effect in the pres-
atures or longer holding times.
sure-vessel steels is not large and is satisfactorily
Heat-Affected Zone . The microstructures formed
counteracted by PWHT.
in the heat-affected zone (HAZ) of a weldment are a
reflection of the composition of the steel and the ther-
Metallurgical Effects on Weldments
mal cycle or cycles imposed by welding . The most sig-
Tensile Properties nificant changes to the parent metal occur in the regions
Weld Metal . As-deposited weld metal exhibits a heated above 1000° C (1830° F) where grain coarsening
high yield to tensile ratio and a strength level that is and partially or fully hardened structures are likely to
high for its composition. These properties result from occur. In multiple-pass welds, the HAZ becomes a
the fast cooling rates characteristic of arc welding and complex aggregate of heated and reheated zones con-

• 1000
2
TENS . STR.
YIELD STR.
ELONG.
-
POSTHEAT I h r. at 625°C
Z
EXCEPT AS NOTED.

J• 500
u)
zZzzZ 0 o 0
z r
z
w a
I-
x X X X X +r
a 250 x
z
4
a
J AW PH AW PH AW PH AW PH AW PH AW PH AW PH
w 0 w
E70I8 E8018 E70T-2 EHI4 E70S-3 EI2015 E9015-83
325 650 325 650 }
-
-CI SAW GMA
AGING TEMPERATURE - °C
WELD METALS
Fig . 12—Effect of 1 hr postheating on room temperature tensile strength
of steels strained 5% 13 Fig . 14—Effect of PWHT on the tensile properties of weld metals 24

Postweld Heat Treatment of Steels 7



carbides, displacing Fe 3 C, some forming a fine was limited to 1—2 hrs in the range of 550—600° C
dispersion within the grains thus strengthening (1025—1110° F) . Extending the time of PWHT leads
them, while a portion migrates to the grain uniformly to moderate loss of notch toughness in these
boundaries to form films or nodules. lower strength weld metals probably due to carbide
5. Because the boundaries become weaker than the coarsening at grain boundaries.
dispersion-strengthened grains, most of the creep When alloying elements are added to weld metal to
strain occurs in the boundary regions which are enhance strength, notch toughness, or elevated tem-
limited by the coarse-grain size. perature properties, the response to postweld heat
6. Rupture occurs in the boundaries by a triple-point treatment becomes complex . 24 As a general trend, those
or a cavitation process. filler metals that obtain their strength through hard-
enability and are softened by the tempering action of
Two groups of elements contribute to stress-relief
PWHT do not lose or may even gain notch toughness
cracking . The carbide formers V, Nb, Mo, and Cr act
when postheated . By contrast those filler metals with
with a potency decreasing in the order listed. Residual
compositions that resist softening during PWHT,
elements such as As, P, Sb, Sn, and trace amounts of Al
namely with carbide-forming elements, exhibit sensi-
appear to act adversely as they do in temper embrit-
tivity to embrittlement from PWHT (Fig . 18) . Two
tlement. Carbon may be expected to be harmful since
types of embrittlement have been observed in these
carbide formation is responsible for the cracking, but
weld metals, arising from related but distinct mecha-
its effect does not seem to be important above 0 .10% . A
nisms that may overlap or occur in succession depend-
number of steel grades involving various combinations
ing on the postheating times and temperatures.
of the carbide-forming alloying elements (particularly
Temper embrittlement is a phenomenon that occurs
V-bearing grades) have been found to be crack-sus-
in heat-treated alloy steels when they are held in or
ceptible in fabrications.
cooled slowly through the temperature range of 400—
The question that naturally arises is what can be done
575° C (750—1075° F) which results in loss of notch
to avoid stress-relief cracking (also called stress-rupture
toughness at low temperatures . It has been traced to the
or reheat cracking) . Very little can be done in terms of
action of certain impurity elements such as Sb, P, Sn,
manipulating the PWHT cycle itself, particularly for
and As coupled with the presence of Si and Mn and to
heavier-section weldments, since the cracking occurs
a lesser extent, Ni and Cr. The embrittlement is asso-
during the heating phase through 400—550° C (750—
ciated with intergranular fracture and is believed to be
1025° F) . Since cracking is the product of stress, hard-
due to segregation of the impurities and movement of
ened structure, carbide-forming elements and residual
alloying elements at former austenite grain boundaries
elements, these are the factors that must be controlled.
with concomitant local loss of cohesive strength . No
The stress level can be reduced by avoiding or elimi-
change in hardness or tensile properties is involved, and
nating stress raisers such as sharp weld toes, incomplete
the process can be reversed by heating to temperatures
root penetration or planar defects where crack initiation
above 600° C (1100° F) followed by rapid cooling . The
is almost always observed . Peening of the welds will
occurrence of temper embrittlement in weld metal
reduce stresses if properly applied . If it is permissible,
arises during the cooling phase of PWHT.
higher heat input has been found21 to be favorable in
The second type of embrittlement, variously called
steels that transform to higher temperature products
"stress-relief embrittlement," "precipitation embrit-
when cooling rates are lowered . Preheating is also
tlement" or "secondary hardening," is the dominant
helpful both for the slower cooling and reduced residual
mechanism by which weld metal notch toughness can
stresses . If there is the option, steel compositions and
be degraded during PWHT . In the as-welded condition,
lower strength weld metals can be selected to lessen the
weld metals containing carbide formers have cooled
likelihood of stress-relief cracking.
through the transformation range too rapidly to allow
Notch Toughness the Cr, Mo, or V to precipitate as carbides . Upon re-
Weld Metal . The effect of PWHT on the notch heating into the range above 425° C the carbon begins
toughness of weld metals varies widely according to to diffuse and form fine precipitates with the elements.
their composition and strength level, the flux or coating, These precipitates harden the steel or at least retard
the heat input, and the temperature and time of post- softening of the matrix but at the same time reduce
heating . Contradictions occur in published data which notch toughness at low temperatures and creep ductility
are difficult to resolve because of insufficient informa- at high temperatures . The time-temperature relations
tion about plate thickness, coatings, welding parameters controlling the process and the degree of embrittlement
and also the considerable scatter in results that is typical ensuing are illustrated in Figs . 19 and 20 . 25 At temper-
of weld metals. atures above 600° C (1110° F) the alloy carbides ag-
As an example, Armstrong and Warner22 reported glomerate, and the overaging process gradually restores
toughness losses in E6010, E10016 and E8015-C2 weld notch toughness . However some alloying elements delay
metals and no gain for E6013, E6020, and E7015 after the recovery so greatly that they cannot be used in weld
PWHT (Fig . 16) while Sagan and Campbell 23 indicated metals that are to be postheated . Vanadium is the most
appreciable toughening in E7016 and E7018 weld metals familiar example, but Nb behaves similarly . Extended
after PWHT (Fig . 17) . In both cases the PWHT time times at PWHT temperatures may undo the recovery

Postweld Heat Treatment of Steels 9



.n
J

E6010 E6013
u- 100 100 100 _ E6020
>-
a- 80 80 80
a
x 60 60 60
U SR
x 40 40 40
U
SR
o 20 20 - / 20
z
> 0 0 ~y I ,
-130 10 150 -130 10 150 -130 10 150
TEMPERATURE, °C TEMPERATURE, °C TEMPERATURE, °C
J
SR
100 100- E70 15 100
*-

>- 80 - - 80 - // - 80
x
60 - / / - 60 - V - 60

= 40 - / / - 40 - // - 40

0z 20 -
-
/ / - 20 - 20

0 I 0
0
> -130 10 150 -130 10 150 -130 10 150
TEMPERATURE, °C TEMPERATURE, °C TEMPERATURE, °C
Fig . 16—Effect of PWHT on the notch toughness of SMAW electrodes 22

process by carbide thickening at ferrite grain bounda- multiple-pass welding . In mild steels the HAZ will
ries. contain pearlite and ferrite which will be little affected
Heat-Affected Zones . There is a wide range of mi- by postheating . Carbon steels of higher carbon content
crostructures that are produced in the as-welded joint will form partially hardened structures in those
as a function of steel composition, cooling rates, and coarse-grained regions of the HAZ cooled moderately

I I I I I I' I I I I I ' I I I

E70I8 ELECTRODE-BRAND E E70I6 ELECTRODE-BRAND M

EFFECT OF HEAT TREATMENT EFFECT OF HEAT TREAT-
MENT

HEATED TREATED
120 621 ° C 2 hrs.
FURNACE COOLED
AS
0
H
WELDED
0 80 HEATED - 80
z 621°C
FOR 2 hrs,
> FURNACE
COOLED .
40
AS
WELDED
0 0 I , I I I
0 1

-84 -62 -40 -I8 0 24 -62 -40 -18 0 24

°C °C

Fig . 17—Response of E7018 and E7016 weld metals to PWHT23

10 WRC Bulletin 302



40 °L
U NOTE ARROWS INDICATE CHANGE BY I hr . PWHT
AT 625°C

a.
2
w
I

E
u)

0 .5 2 .0 8 .0 32
TIME AT TEMPERATURE, Hrs.

E8016-CI E9015-B3 SAW E11018-G E12015 Fig . 19—Effects of time and temperature of PWHT on the Charpy test
W-19 25
energy level at 10 0 C of 2 1/4% Cr-1 % Mo steel weld metal
WELD METALS

Fig. 18—Effect of PWHT on the notch toughness of higher strength weld

metals24

70 36
> 510°C 676°C
25 28

-2C .' 593°C 20

-65

482°C ♦
25 r'- .•
649°C

---•_ - ~'

565°C ,
-20
•
-6

454°C 537°C 621°C

25 AS DEPOSITED ♦

-20 _
20
AS DEPOSITED

-65 12

1 .0 10 .0 50 .0 1 .0 10 .0 50 .0 I .0 10 .0 50 .0
LOG TIME (hrs .)
• 15 FT . LB .TEMP .-LEAD POT TREATMENT
• 15 FT . LB .TEMP .- AIR FCE TREATMENT
♦ ROCKWELL C HARDNESS

Fig. 20—Effect of PWHT temperature on the notch toughness of Ni-Mo-V weld metal 33

Postweld Heat Treatment of Steels 11



rapidly and not reheat treated by overlying passes, and PLATE TESTS
-AS RECEIVED
these regions will be softened and toughened by PWHT. - NOTCHED
Steels in which alloying elements are used to increase
- --------------------------------------------------
strength and low temperature toughness must be
welded at controlled heat inputs producing bainitic or z 0
as
martensitic microstructures . If the carbon content is F0 1 .5 - ------------------------- ---------------
rnw
kept below 0 .15% the HAZ's of these steels display ex- w'
cellent as-welded notch toughness without need of J W
m 0
Q W
PWHT. As carbon content is increased the hardness of a 1 .0
ow
the HAZ rises and ductility becomes low, so that PWHT - Q
-a w
is essential to satisfactory notch toughness. 0
W 0 .5
The response of heat-affected zones in carbon and 0Q

alloy steels to PWHT is illustrated more specifically in o

-
Table 1 . 26 Postheating the 0 .15% carbon steel improves a 0
cc 1-
R N R N R N R N R N R N
the HAZ toughness slightly, while the 0 .25% C steel is A201 48s5 A302 HY-65 NES-70 A517F
considerably toughened . Likewise the Mn-Mo and Ni- STEEL
Cr-Mo steels with about 0 .20% C show large gains in
HAZ toughness . The vanadium-bearing steel heat-af- Fig . 22—Comparison of the fatigue resistances of carbon and high-
fected zones, however, are noticeably embrittled by strength steels in biaxial plate tests34

PWHT, particularly if 550° C (1025° F) is used . Thus

PWHT offers no metallurgical benefit to this class of
steels. endurance limit, tensile residual stresses may alter the
Fatigue Resistance fatigue resistance particularly in corrosive environ-
The high cycle fatigue resistance of steels is so ments, 27 and therefore PWHT may benefit fatigue
strongly dependent on the severity of stress raisers that performance by reducing them . Nonetheless stress-
occur in fabricated structures that it is little affected by raising details remain the dominant factor in high-cycle
steel composition, heat treatment or welding (except as fatigue service.
the welded configuration contributes stress-raisers) . In
smooth-contoured specimens prepared from pres- Bases for Defining Conditions Requiring PWHT
sure-vessel steels the presence of a weld had no dis-
Alternatives to PWHT
cernible effect on the allowable strain range for a life of
From the information that has been presented it is
100,000 cycles, Fig . 21 . By contrast, a mild notch in-
obvious that for a number of the important low alloy
serted into biaxial-stressed plate tests lowered the al-
steels postweld heat treatment can be a mixed blessing.
lowable strain range significantly, and the higher
There are many applications of high strength steels in
strength steels offered little advantage over the carbon
which sound engineering principles can be used to jus-
steel, Fig. 22.
tify the replacement of postheating by other technical
It follows from these observations that PWHT will
procedures, including applications for which code
have a negligible effect on the fatigue resistance of steels
specifications presently make postweld heat treatment
in the cycling range typical of pressure vessels, less than
mandatory. In the final analysis PWHT is used essen-
100,000 cycles. In the cyclic range at stresses close to the
tially for two purposes : (1) the relief of residual and
reaction stresses, and (2) the tempering of hardened
structures in the welded joint . If these factors can be
ameliorated sufficiently by other techniques, PWHT
CANTILEVER BEAM TESTS can be eliminated . These alternative techniques include
z
steel and filler metal composition, welding procedures,
warm overstressing, inspection, and peening.
Steel Composition . Improvements have been de-
veloped in the production of high strength steels that
enhance their weldability and their service properties.
First, carbon content has been lowered to reduce
cracking sensitivity and hardening during welding and
to depress the temperature range where brittle fracture
becomes possible . Second, the impurity content, such
as P and S, has been significantly lowered with corre-
A302 HY-65 NES-70 A517F sponding gains in notch toughness. As a result steels are
STEEL
available that provide a good margin of as-welded notch
toughness in low temperture service . Consequently
Fig . 21—Comparison of the fatigue resistances of carbon and high- postweld heat treatment is not needed except for di-
strength steels tested as received, as-notched, and as-welded 34 mensional stability or stress-corrosion resistance . The

12 WRC Bulletin 302



benefits to strength and toughness provided by car- 2 . Weldment characteristics

bide-forming alloying elements in the welded joint can a. section thicknesses?
be exploited in as-welded structures, but would be ad- b. tensile properties?
versely affected by PWHT. c. fracture toughness of parent metal, HAZ, and
Welding Procedures. The proper choice and exe- weld metal?
cution of welding procedures contribute importantly to d. max flaw size as controlled by inspection
the as-welded properties of weldments . Some examples methods used?
are the following: e. control of contours, warping, weld cycles in
critical regions of the weldment?
1. controlled bead size limits grain coarsening and
reheat-treats underlying heat-affected zones, thus If there are doubts about the matching of the weld-
enhancing the notch toughness of the welded ment properties to any of the service requirements, the
joint. final question then is—will postweld heat treatment
2. selected consumables, joint cleanliness, and pre- remedy the deficiency identified? As pointed out pre-
heat keep hydrogen levels below the threshold of viously, PWHT serves to reduce residual stresses and
delayed cracking. to temper the heat-affected zone, and essentially the
3. quality control of welding avoids stress-raising deficiency must be remedied by one or both of these
contours at weld toes and roots or planar defects actions . Otherwise PWHT will not effect the cure.
such as lack of fusion that degrade notch tough-
ness or fatigue resistance. Need for Additional Research
4. efficient inspection methods made possible by
As a consequence of the preparation of this report,
recent advances can reduce the maximum unde-
several subjects become evident that warrant further
tected flaw size and hence lessen the risk of frac-
investigation. They are listed below:
ture initiation in service.
5. residual stresses can be reduced as much as 50% 1. While the effects of PWHT have been reported for
by narrow-gap welding or by peening the inter- a wide variety of steels and weld metals, a much
mediate weld passes. larger volume of data is desirable to establish the
effects of heat or lot variations, time and temper-
Warm Overstressing . Structures such as pressure
ature relations during PWHT, and welding pa-
vessels, piping, and tanks are commonly submitted to
rameters on tensile and notch toughness proper-
a proof stress higher than the maximum value expected
ties.
in service to demonstrate integrity or to reveal locations
2. Since considerable stress relief and tempering of
that are faulty . It has been shown28 by large-scale lab-
hard heat-affected zones can be accomplished by
oratory tests containing a deliberate flaw that if the
short exposure at the PWHT temperature, an in-
overstressing is performed at a temperature safely above
vestigation should be made of the feasibility of
the brittle-fracture range of the steel, the test piece will
reducing the time requirements for PWHT and of
substantially withstand loads approaching the over-
establishing time-temperature combinations to
stress load even though applied at temperatures below
optimize toughness in alloy steels such as 2 1/4 Cr-1
brittle fracture transition . The effect of the warm ov-
Mo steel.
erstressing is thought to be the redistribution of peak
3. As a mean of obviating the need of PWHT, further
stresses at the flaw and a blunting of the flaw tip which
attention should be directed toward the use of
makes it a less severe stress raiser.
techniques such as (1) the fitness for purpose
Service Requirements concept, (2) the application of state-of-the-art
The determination whether the as-welded charac- methods of non-destructive examination to reduce
teristics of a pressure vessel or other weldments will be the maximum size of undetected flaws, (3) closer
suitable for a specific service is not a precise process, but control of welding parameters to minimize residual
it can be reduced to a series of questions which may be stress and HAZ embrittlement, and (4) develop-
answered satisfactorily by a sufficient knowledge of (1) ment of parent metals and filler metals whose
the combination of demands that the service will place mechanical properties are adequate for service
upon the structure, and (2) the properties that will be without PWHT.
exhibited by the as-welded structure predicted from
previous welded tests or concurrent ones . A sample set References
of questions might be the following: 1. Larson, F . R . and Miller, J ., "A Time-Temperature Relationship for
Rupture and Creep Stresses, " Trans . ASME, Vol . 74, p . 765, 1952.
2. Holloman, J. H . and Jaffee, L. D ., "Time and Temperature Relations
1 . Service conditions in Tempering Steel," Trans . AIME, Vol . 162, p . 223, 1945.
3. Cooper, P ., "Postweld Heat Treatment in the Field," Metal Progress,
a. max stress? Vol. 114, No. 4, pp . 36-39, Sept. 1978.
b. rate of loading? 4. Hormann, E ., " The Inductive Heat Treatment of Weld Seams, "
Schweissen and Schneiden, Vol . 22, No. 5, pp. 201-202, 1970.
c. cyclic loads and number? 5. "Methods of Post Weld Heat Treatment, " Japanese Standards Assoc.,
JIS, Z3700-1980.
d. lowest operating temperature? 6. Pense, A . W ., " Stress Relaxation in Pressure Vessel Steels, " Final Report
e. corrosive environment or contents? to Pressure Vessel Research Committee, April 18, 1982.
7. Ritter, J . C . and McPherson, R ., " Stress Relaxation in Steel, " Welding
f. consequences of a fracture? Research Abroad, Vol . 18, No . 4, pp. 20-31, April 1972.

Postweld Heat Treatment of Steels 13



8. Burdekin, F. M., "The Significance of Thermal Stress Relief as Protection 21. Meitzner, C. F., "Stress-Relief Cracking in Steel Weldments," WRC Bull.
Against Brittle Fracture in Mild Steel," Commonwealth Welding Conference, 121, Nov. 1975.
1960 . 22. Armstrong, T . N. and Warner, W. L ., "Effect of Preheating and Post-
9. Wells, A . A., "Effects of Thermal Stress Relief and Stress Relieving heating on Toughness of Weld Metal," Welding Journal, Vol. 37, No. 1, pp.
Conditions on the Fracture of Notched and Welded Wide Plates," British 27s—29s, Jan . 1958.
Welding Journal, Vol . 10, No. 5, pp . 270—276, 1963. 23. Sagan, S . S. and Campbell, H . C., "Factors Which Affect Low-Alloy
10. Provost, W., "Effects of a Stress Relief Heat Treatment on the Toughness Weld-Metal Notch Toughness," WRC Bull . 59, April 1960.
of Pressure Vessel Steels," Int . Journal of Pressure Vessels and Piping, Vol. 24. Dorschu, K . E., "Factors Affecting Weld Metal Properties in Carbon and
10, No. 2, pp . 93—118, 125—154, 1982. Low Alloy Pressure-Vessel Steels," WRC Bull . 231, Oct. 1977.
11. Takuchi, T., Fukaya, T ., Sato, M ., and Takano, G., "Study on the Ap- 25. Swift, R . A . and Rogers, H . C ., "Embrittlement of 2 1/4 -Cr-1 Mo Steel Weld
plication of Thickened Welds Without Post Weld Heat Treatment for Con- Metal by Postweld Heat Treatment," Welding Journal, Vol. 52, No. 4, pp.
tainment Vessels," Mitsubishi Juko Giho, Vol. 15, No . 5, pp. 564—572, 1978. 145s—153s, April 1973.
12. Gulvin, T . F ., Scott, D ., Haddrill, D . M ., and Glen, J ., "The Influence of 26. Watkins, B., Vaughan, H . G ., and Lees, G . M ., "Embrittlement of Sim-
Stress Relief on the Properties of C and C-Mn Pressure-Vessel Plate Steels," ulated Heat Affected Zones in Low-Alloy Steels," Brit . Welding Journal, Vol.
Journal of the West of Scotland Iron and Steel Institute, Vol . 80, pp. 149—175, 13, pp. 350—356, June 1966.
1972—73. 27. Boothe, G . S. and Wylde, J. G ., "Some Mean Stress Effects on the Cor-
13. Rubin, A. I ., Gross, J . H., and Stout, R . D ., "Effect of Heat Treatment rosion Fatigue Performance of Welded Joints," Proc. 2nd Int . Symp ., Paper
and Fabrication on Heavy Section Pressure-Vessel Steels, " Welding Journal, 13, Glasgow, Scotland, July 1—3, 1981.
Vol . 38, No. 4, pp . 182s—197s, April 1959. 28. Nichols, R. W., " The Use of Overstressing Techniques to Reduce the Risk
14. Swift, R . A . and Gulya, J. A ., "Temper Embrittlement of Pressure Vessel of Subsequent Brittle Fracture," Brit . Welding Journal, Vol . 15, pp. 21—42, 75,
Steels," Welding Journal, Vol . 52, No. 2, pp . 57s—68s, Feb . 1973. 84, Jan.—Feb. 1968.
15. Interrante, C . G ., "Interpretive Report on Effect of Hydrogen in Pres- 29. Gunnert, R., "Method for Measuring Triaxial Residual Stresses,"
sure-Vessel Steels ; Section 1," WRC Bull . 145, Oct . 1969. Welding Research Abroad, Vol . 4, No . 10, pp. 17—25, 1958.
16. Interrante, C . G . and Stout, R . D ., "Delayed Cracking in Steel Weld- 30. The Welding Handbook, Volume 1, 7th Edition, 1976, American Welding
ments," Welding Journal, Vol . 43, No . 4, pp. 145s—160s, April 1964. Society.
17. Capla, J . S . and Landerman, E . I ., "Preventing Hydrogen-Induced 31. Pense, A. W ., "Stress Relaxation in Pressure Vessel Steels," Final Report
Cracking After Welding of Pressure Vessel Steels by Use of Low Temperature to Fabrication Division of PVRC, April 1982.
Postweld Heat Treatments," WRC Bull . 216, June 1976. 32. Sprung, I ., " Effect of Postweld Heat Treatment on Mechanical Properties
18. Bates, J . F ., "Sulfide Stress Cracking of High Yield Strength Steels in of Carbon and Low Alloy Steels," Report to ASME Subgroup on Toughness,
Sour Crude Oils, " Materials Protection, Vol. 8, No. 1, 3, 1969. Sept . 13, 1982 Meeting Minutes.
19. Meitzner, C . F . and Pense, A . W ., "Stress-Relief Cracking in Low-Alloy 33. Puzak, P . P . and Pellini, W. S ., "Embrittlement of High-Strength Ferritic
Steel Weldments, " Welding Journal, Vol. 48, No. 10, pp . 431s–440s, Oct. Welds," Welding Journal, Vol . 31, No . 11, pp . 521s—526s, Nov. 1952.
1969 . 34. Stout, R. D . and Gross, J. H., "Mechanical Properties of Weldability of
20. Yamazaki, Y., Manago, Y ., Okabayashi, H., and Kamematsu, M., "Stress Six High-Strength Steels," WRC Bull . 27, May 1956.
Relief Cracking of High-Tensile Steels," Journal of the Japan Welding Society,
Vol . 32, p. 283, 1963 .

14 WRC Bulletin 302



Relaxation Stresses in Pressure Vessels

by P. S. Chen, W. A. Herman, and A . W. Pense

CONTENTS

Abstract 15 ments producing creep-relaxed stresses below 69 MPa

Introduction 15 (10 ksi) the strength decrease was about 5% for A737
Materials and Procedures 16 Grade B, about 10% for A543 Type B Class 2 and about
Experimental Materials 16 20% for A387 Grade 22 Class 2.
Experimental Procedures 16 Comparison of the data with true creep and true
Results and Discussion 17 stress-relaxation data show the values measured in the
Creep-Relaxation Stress Results 17 investigation to be intermediate between these two
Data Summary Techniques 19 limits, a result of the nature of the test . A semi-em-
Mechanical Property Changes 19 pirical computational method was used to construct
Comparison to Creep and Stress-Relaxation
approximate stress-relaxation curves from the data . It
Data 20
Application of the Data 21 is proposed that the data developed here represent a
Conclusions realistic approximation of those found in heavy weld-
21
ments of these steels.
References 22
Appendix I 22
Introduction
Approximation of Stress-Relaxation Curves
from Creep-Relaxation Data 22 The stress relaxation behavior of welded pressure
vessels is a matter of interest to steel suppliers, to fab-
ricators, and to owners who employ stress relief to re-
duce residual stresses in pressure vessel weldments . The
Abstract
use of stress relief is mandated by the ASME Boiler and
Creep-relaxation stresses were measured in A737 Pressure Vessel code in some cases, and even when not
Grade B, A543 Type B Class 2 and A387 Grade 22 Class required, it is considered desirable by many engineers
2 steel for times of 8 or 9 hrs in the 538° C (1000° F) to for stress reduction to prevent brittle fracture, to
677° C (1250° F) temperature range . The test proce- eliminate hydrogen induced cracking, to control sub-
dure used was a fixed grip tension test type . The data sequent machining distortion, to reduce the chance of
values were to be intermediate between standard creep stress corrosion, and to effect metallurgical changes in
test and stress-relaxation test results and simulate the weldment microstructures that improve their proper-
load maintaining behavior of a complex component ties.
during post weld heat treatment. From the stress relief viewpoint, the reduction of
Creep-relaxed stresses below 69 MPa (10 ksi) were stress during heat treatment, or more exactly, a suffi-
achieved after 8 hrs at 538° C (1000° F) for A737 Grade ciently low stress level after stress relief, is the most
B, and after 8 hrs at 593° C (1100° F) for A543 Type B. important result . Since the stress relief treatment in-
Stresses below 69 MPa (10 ksi) were not achieved for volves expense, however, it is not economical to spend
A387 Grade 22 for times up to 8 hrs at 649° C (1200° F) more time than necessary in the process . Therefore,
but were achieved after 2 hrs at 677° C (1250° F) . The data addressing the rate and extent of stress relief for
data curves had a nonlinear region in the yield point various treatments given to a representative group of
stress range at temperature after which creep-relaxation pressure vessel steels could be a great potential use to
occurred in a logarithmic manner with time. the pressure vessel industry . It is for this reason that the
The room temperature properties of the steels de- Pressure Vessel Research Committee has had a con-
creased with stress-relief out to 8 or 9 hrs . For treat- tinuing interest in this subject . The experimental in-
vestigation reported here is one of many sponsored by
the Thermal and Mechanical Effects Subcommittee of
the Fabrication Division, PVRC, on this general topic.
P . S . Chen, W. A . Herman, and Prof. A . W. Pense are with the Department The program was specifically aimed at measuring re-
of Metallurgy and Materials Engineering, Lehigh University, Bethlehem,
PA. sidual stress reduction rates during stress relief.

Relaxation Stresses in Pressure Vessels 15



There are a number of potential approaches to the remain will reflect this . It is from this viewpoint that the
study of stress relief. The first is to measure residual data from this program will be analzyed.
stress before and after stress relief and obtain a direct The overall program was sponsored by the Subcom-
measure of their level. This approach has the advantage mittee on Thermal and Mechanical Effects of the
that the final residual stress is well known at a specific Fabrication Division of the Pressure Vessel Research
position in a stress relieved weldment, and is directly Committee . A comparison work on residual stress
measured . It has the disadvantage that the rate of ap- measurements was also cosponsored by the Subcom-
proach to the final level is not known unless a large mittee on Thermal and Mechanical Effects and the
number of intermediate measurements are made. Subcommittee on Cyclic and Creep Behavior of Com-
Moreover, if the weldment is large, there may be a large ponents and is to be published elsewhere . 1 It is the
number of areas of interest, and the cost of measure- purpose of these studies to improve the ability of vessel
ment of residual stress in each is prohibitative. Another fabricators to predict the stresses and properties of in-
limitation is that the stresses in depth cannot be readily dustrial pressure vessels before, during, and after post
measured at the present time as only surface stresses weld heat treatment.
can be studied without destructive testing.
Another technique is to measure the creep stresses Materials and Procedures
that the steel in the weldment can support as a function
Experimental Materials
of temperature and time, and deduce an upper limit of
Three steels of different creep behavior were selected
residual stress from these creep stresses . This has the
for study. One of the materials was a high temperature
advantage that it can reveal the rate of stress reduction.
alloy commonly used for chemical and petrochemical
Moreover, the numbers obtained can be generally ap-
operations—A387 Grade 22, Class 2 (2 1/4 Cr — 1 Mo)
plied even to the interior of the weldment . They do
steel . This material has been extensively studied and
represent an extreme upper bound, however, since the
is known to be creep resistant, both at normal service
actual relieved stresses may be lower, and the sense of
temperatures, 370—600° C (—700—1100° F) and during
the stress cannot be known.
stress relief at 675° C (1250° F) . The material studied
A third approach is to measure stress-relaxation
was obtained as a nozzle cut out from a 75 mm (3 in .)
stresses. This is a technique that takes into account the
thick plate . The material was in the mill quenched and
elastic response of the weldment to creep and results in
tempered condition.
stress reductions beyond those predicted by creep alone.
The second material was A543 Type B, Class 2 steel
The stresses predicted by this method are general, but,
(Ni-Cr-Mo), a quenched and tempered steel normally
once again, their sense is not known.
employed for strength and toughness . This material was
The current study combines aspects of the latter two
being studied in the residual stress portion of the PVRC
types of investigations . From the standpoint of a pres-
program, and was a high strength chemistry normally
sure vessel being given stress relief treatments, it is not
used for heavy section applications . It was obtained in
entirely clear how it will respond . It is usually assumed
150 mm (6 in .) section . This steel is known to have in-
that the relief of stresses locked in from welding will
termediate creep resistance, and is normally stress re-
occur by the process of stress relaxation . This is best
lieved at 620° C (1150° F) . The third material was a
illustrated by a bolt inserted through a rigid plate . At
microalloyed steel, A737 Grade B . Since it was also
temperature, the bolt elongates under stress by creep,
under study by PVRC in the residual stress measure-
but the stress is not maintained, as it would be in a dead
ment program, it appeared appropriate to include it in
load creep test, and as a result the stress decreases
this study . This material is used for good toughness and
rapidly as the bolt becomes longer . This is to be con-
weldability in combination with intermediate strength,
trasted with the case of a bolt inserted through a plate
with a spring under the nut or head. As the bolt elon- and as such is usually employed for ambient and low
temperature applications. Its creep behavior is not well
gates, the spring maintains the load . This causes the
documented . It is usually stress relieved at 620° C
stress present to be higher, and is a function of the
(1150° F) . It was in the normalized condition.
spring constant of the system and the elongation of the
The chemical compositions and room temperature
bolt . As a limit, with a very compliant spring and a high
mechanical properties of the steels are shown in Tables
spring capacity, the bolt load stays about constant (al-
1 and 2, respectively. It should be noted that all three
though perhaps low), and the test becomes a creep
materials were relatively low in residual elements, and
test.
all were aluminum-silicon killed.
As will be described, the experiments performed were
such that a pure stress-relaxation test did not occur but Experimental Procedures
the load was partially maintained in the test . The actual The basic test procedure was to use a constant
case in a pressure vessel is probably a partially load- crosshead position tension test employing an initial load
maintained case as well . This is because the effect of the equal to the at-temperature yield point and allowing
geometry of the vessel is such that it can, in fact, act like creep of the specimen to relax the applied load . As such,
a large spring, which welding stresses serve to compress. this test is a hybrid between ASTM specification
Thus during stress relief there will be a tendency for the E150-75, " Conducting Creep and Creep-Rupture
load to be maintained and the residual stresses that Tension Tests of Metallic Materials Under Conditions

16 WRC Bulletin 302

 Table 1-Chemical Composition of the Steels (wt %)
Steel C Mn P S Si Ni Cr Mo Al Nb

A737 Grade B 0 .14 1 .44 0 .009 0 .006 0 .19 0.28 0 .22 0 .09 0 .030 0 .025
A543 Type B 0 .16 0.29 0 .007 0 .020 0 .32 3 .72 1 .73 0 .47 0 .032 -
A387 Grade 22 0 .13 0.40 0 .009 0 .004 0 .19 0.11 2 .25 1 .00 0 .027

of Rapid Heating and Short Times" 2 and ASTM this position and the load recorded for a period of up to
specification E320-78 "Stress-Relaxation Tests for 8 or 9 hrs . The temperature was held constant to within
Materials and Structures ." 3 It differs from these tests ±5° C (8° F) using a resistance wound furnace covering
primarily in the way in which the load is maintained the specimen, and extended grips, about 250 mm . The
during the test. In a true stress-relaxation test, the test specimen load was recorded on a load-time chart, and
fixture is perfectly rigid, i .e ., the load is not maintained. the creep-relaxation stress was determined from this
Under these conditions, creep of the specimen is directly plot and the specimen cross section . The load was re-
translated into relaxed stress . As described above, such corded continuously for the first 2 hrs and then hourly
a test is like a bolt in a rigid plate . In a creep-rupture for the final hours . The specimen used was a standard
test, the exact opposite is true, the load is fully main- 6 .4 mm (0 .252 in .) diameter tension test specimen 108
tained during the test, i .e ., in creep-rupture tests the mm (4.25 in .) long. After completion of the 8-hr test, the
stress is constant and the rate of strain is measured. tension test specimen was unloaded, removed and
In the test reported here, neither of these conditions tested at room temperature to determine residual ten-
specifically hold as the test machine itself acts as a kind sile properties.
of spring to provide some maintenance of load . The The test temperatures employed were based on the
spring constant of the machine is less than that of the normal stress relief cycles applied to the materials
specimen, and thus there can be considerable, but by studied . The A387 Grade 22, Class 2, was tested at 538°
no means complete, load recovery as the specimen C (1000° F), 593° C (1100° F), 621° C (1150° F), 649°
elongates . The result is that the load measured is C (1200° F) and 677° C (1250° F) . The A543 Type B,
controlled both by the spring action of the test appa- Class 2 and the A737 Grade B were tested at 538° C
ratus and the creep behavior of the steel . It is considered (1000° F), 593° C (1100° F) and 621° C (1150° F).
by the authors that this condition approximates the real
behavior of a weldment where some spring action of the Results and Discussion
structure acts to maintain load, but there is also inde-
terminancy in the results . This will be discussed Creep-Relaxation Stress Results
later. The results of these tests are shown in Table 2 and on
In the procedure used in this investigation, the Figs . 1-3 . For A543 Type B Class 2, the yield stresses at
specimen was held at the temperature of interest for 6 these temperatures ranged from 434 MPa (64 ksi) at
hrs to allow for thermal equilibration and was then 538° C (1000° F) to 310 MPa (45 ksi) at 621° C (1150°
loaded at 1 .27 mm/min (0.05 ipm) until the yield point F) . At all three temperatures, the stresses relaxed quite
was reached (as determined by a load-displacement rapidly (Fig . 1) during the first hour, falling to values
record) . The cross-head of the machine was locked in 35-50% below the original yield stress. After eight hours,

Table 2-Mechanical Properties of the Steels

0 .2% Yield Tensile
Offset Sth. Ultimate Sth . Elong . Red. of
Steel and Condition MPa (ksi) MPa (ksi) % Area %

A737 Grade B (Initial)

1 . Prior to test 403 .4 (58 .5) 598 .5 (86.8) 30.1 60 .0
2 . After test 8 hr at 538° C (1000° F) 444 .7 (64 .5) 562 .6 (81 .6) 30.9 74 .4
3 . After test 8 hr at 621° C (1150° F) 400 .6 (58 .1) 526 .1 (76 .3) 37 .3 76 .8
A543 Class 2 (Initial)
1 . Prior to test 705 .4 (102 .3) 815 .7 (118.3) 23 .7 63 .6
2 . After test 8 hr at 538° C (1000° F) 634 .3 (92 .0) 800 .5 (116.1) 24 .1 64 .3
3 . After test 8 hr at 598° C (1100° F) 625 .4 (90 .7) 772 .9 (112.1) - 60 .6
4 . After test 8 hr at 621° C (1150° F) 604 .0 (87 .6) 759 .1 (110.1) 21 .2 62 .4
A387 Grade 22 Class 2 (Initial)
1 . Prior to test 556 .4 (80 .7) 690 .9 (100.2) 29 .1 78 .8
2 . After test at 8 hr 538° C (1000° F) 514 .4 (74 .6) 647 .4 (93.9) - 78 .3
3 . After test at 9 hr 621° C (1150° F) 464 .0 (67 .3) 615 .0 (89.2) 27 .2 79 .6
4 . After test at 9 hr 649° C (1200° F) 433 .7 (62 .9) 583 .3 (84.6) 30.5 80 .1
5. After test at 8 hr 677° C (1250° F) 389 .6 (56 .5) 544 .7 (79 .0) - 80 .8

Relaxation Stresses in Pressure Vessels 17



00

A543 Type B Class 2 oV A737 Grade B

300 300 Q.
40 0 40
a
_
\0\ o~38°C (1000°F) N
30 - 200 N
p~ 30

~ N
Z
20 O 0
`
593 °C (1100°FI a
521°C (1150°F) < ` - 00 - 538°C (1 000°F)
W 593°C (11
10 C to
ff° ~•~
621 C (1150°F)

5 .0 n
0.1 0 .25 0.5 1 .0 2.0 10 20 0 .1 0 .25 0 .5 1 .0 2 .0 5.0 10 20
RELAXATION TIME, hrs RELAXATION TIME, hrs

Fig . 1-Creep-relaxation stress curve for A543 Type B Class 2 Fig . 2-Creep-relaxation stress curve for A737 Grade B

the remaining stresses at 598° C (1100° F) and 621° C The A387 Grade 22 steel shows much higher creep
(1150° F) were fairly low, 62 and 48 MPa (9 and 7 ksi) resistance with much less decrease in stress at 538° C
respectively, while at 538° C (1000° F) a significant (1000° F) for times out to 8 hrs . The initial yield
amount of stress still remained, 165 MPa (24 ksi) . Based strength of this steel at each temperature is similar to
on the slopes of Fig . 1, treatments for longer times at A543 Type B Class 2 (ambient temperature yield
these temperatures will have only a minimal effect on strength being much lower) and the strength retained
the prevailing stress level. after 8 hrs . is between 40 and 45% of the at-temperature
For A737 Grade B, the parallel study produced much yield at 598° C (1100° F) and 621° C (1150° F) as com-
lower stresses (Fig . 2) for each equivalent condition. At pared to 15-20% for the A543 Type B Class 2 and A737
any of the three temperatures, the yield stress was 138 Grade B . This is also consistent with previous work on
MPa (20 ksi) lower than the corresponding times for this steel . 4 The relatively small difference between the
A543 Type B Class 2 . Therefore, even though the curves 538° C (1000° F) and 621° C (1150° F) samples can
had the same general shape, the loss in stress with time probably be attributed to the fact that these tests are
was less rapid in A737 Grade B . For this material, even conducted in the secondary hardening range and creep
at the lowest test temperature of 538° C (1000° F) the may be inhibited by precipitation processes . The tests
remaining stress after 8 hrs was only 69 MPa (10 ksi) as for A387 Grade 22 steel were continued to higher tem-
compared to the 165 MPa (24 ksi) level in A543 Type peratures because the normal stress relief treatment for
B Class 2 for the same condition . This is, of course, not this grade is above 621° C (1150° F) . Results for treat-
surprising as the alloy content of the A737 Grade B is ments at 649° C (1200° F) and 677° C (1250° F) are
quite lean compared to the A543 Type B Class 2 and its shown in Table 3 and Fig . 3 . Only at 677° C (1250° F)
room temperature yield point is lower . However, as do the relaxed stresses reach levels typical of the other
previous creep studies on A543 Type B Class 2 indi- steels at 621° C (1150° F), 50 MPa (7 .2 ksi) after 8 hrs.
cated,4 its creep resistance is not exceptionally good Treatments at 649° C (1200° F) generally produce re-
when compared to nickel-free chromium-molybdenum sults similar to those obtained for other steels at 621°
steels . The A737 Grade B relaxed stress is probably C (1100° F), with relaxation stresses being above 83
similar to carbon-manganese steels while the A543 Type MPa (12 ksi) after 8 hrs.
B is only modestly higher.

Table 3-Basic Stress-Creep-Relaxation Date

Test Temp. Yield 0.25 hr 0 .50 hr 1 .Ohr 2.Ohr 3.Ohr 8.Ohr
Steel °C °F MPa (ksi) MPa (ksi) MPa (ksi) MPa (ksi) MPa (ksi) MPa (ksi) MPa (ksi)
A543 538 (1000) 435.8 (63 .2) 282 .7 (41 .0) 241 .3 (35.0) 220.6 (32 .0) 186.2 (27 .0) 173 .8 (25 .2) 165.5 (24 .0)
Type B 593 (1100) 324 .1 (47 .0) 161 .3 (23 .4) 144.8 (21 .0) 105.5 (15.3) 92.4 (13 .4) 75 .8 (11 .0) 62.1 (9 .0)
Class 2 621 (1150) 309.6 (44 .9) 133 .8 (19 .6) 117.2 (17.0) 100.7 (14 .6) 88.3 (12 .8) 59 .3 (8 .6) 48.3 (7 .0)
A737 538 (1000) 264.8 (38 .4) 150 .3 (21 .8) 132.4 (19.2) 115.1 (16.7) 98.6 (14 .3) 80 .0 (11 .6) 71 .7 (10 .6)
Grade B 593 (1100) 239 .9 (34 .8) 118 .6 (17 .2) 103.5 (15.0) 95.2 (13.8) 74.5 (10 .8) 57 .9 (8 .4) 49.6 (7 .2)
621 (1150) 151 .7 (22 .0) 55 .2 (8 .0) 48.3 (7 .0) 38.6 (5.6) 34 .5 (5 .0) 27 .6 (4 .0) 20.7 (3 .0)
A387 538 (1000) 434 .4 (63 .0) 337 .8 (49 .0) 324.0 (47.0) 306.8 (44.5) 289.6 (42 .0) 268 .9 (39 .0) 250.3 (30 .3)
Grade 22 593 (1100) 375 .8 (54 .5) 235 .1 (34.1) 217 .2 (31 .5) 207 .5 (30.1) 189.6 (27 .5) 166 .2 (24 .1) 155 .1 (22 .5)
Class 2 621 (1150) 341 .3 (49 .5) 214 .4 (31 .1) 199.3 (28.9) 181 .3 (26.3) 170.3 (24 .7) 158 .6 (23 .0) 144 .8 (21 .0)
649 (1200) 285 .4 (41 .6) 151 .0 (21 .9) 132 .4 (19 .2) 124 .8 (18.1) 111 .7 (16 .2) 99 .3 (14 .4) 86 .9 (12 .6)
677 (1250) 222 .0 (32 .2) 100 .7 (14.6) 83 .4 (12 .1) 77 .9 (11 .3) 68 .9 (10 .0) 58 .6 (8 .5) 49 .6 (7 .2)

18 WRC Bulletin 302



au 50
A387 Grade 22 Class 2
A387 Grade 22 Class 2
538°C (10\ 300

40 0
O\ °
a 300

N 40
593°C (1100°F) N
30 200 =
821°C (1150°F)
h

20 °
O
11200°FI
30 -
200
877°C (1250°0 °
10 G
° moo ono

0 2
0.1 0 .25 0 .5 1 .0 2.0

RELAXATION TIME, hrs

- \oi 100
P0 = T(C 2 + Log t)
Fig . 3—Creep-relaxation stress curve for A387 Grade 22 Class 2

Data Summary Techniques 32 33 34 35

1

36 37 38
0

Examination of the data for these steels reveals a

short time nonlinear portion at the yield point load CREEP RELAXATION PARAMETER, P 6
followed by a uniform logarithmic relationship between
Fig . 4—Larson-Miller type curve for creep relaxation of A387 Grade
stress and time for the rest of the test . Thus, the 22 Class 2
stress-time relationship is simply represented at each
temperature in the uniform region by:
Q=K—C 1 logt (1)
F) and 677° C (1250° F) reaching 30% in yield strength
where r is the relaxed stress, K and C 1 are constants and and 21% in tensile strength . This is, in a sense, surpris-
t is time in hours . It may also be represented by the ing, as this material is usually tempered at a tempera-
more common Larson-Miller expression, ture above 621° C (1150° F) . However the loss is uni-
form and consistent, and matched by increases in duc-
P a = T(C 2 + log t) (2)
tility . Thus this decrease must be considered due to the
where P a is a creep-relaxation parameter (similar to the extended exposure this treatment produces in a
Larson-Miller parameter), T is absolute temperature, quenched and tempered grade.
C 2 is a constant and t is time in hours . The data are It should also be noted that a 6 hour stabilizing ex-
sufficiently limited and times sufficiently short that a posure to the test temperature was used prior to appli-
Larson-Miller type curve can be constructed for only cation of load . Thus a 9-hr treatment actually involved
the A387 Grade 22 steel . Such a plot is seen in Fig . 4. a 15-hr exposure to temperature . Since tempering
The usefulness of this plot is limited because of the short phenomena normally follows a logarithmic relationship
time involved in the study. with time, the additional 9 hrs should not strongly in-
Of more direct application may be the simple rela- fluence the extent of tempering beyond the 6-hr treat-
tionships of the Eq . 1 type, and a table of the values of ment, thus creep processes may also be important in the
K and C are listed for each temperature and steel in strength loss.
Table 4 . Because of differences in precipitation response
and creep behavior between steels, and for a given steel
at different temperatures, further data reduction be-
yond Table 4 does not appear justified. Table 4—Stress Relaxation Data Analysis Parameter a K
and C 1
Mechanical Property Changes K C1
Temperature
Mechanical properties of the steels measured after °C (°F) S (E) S (E)
8 or 9 hrs of exposure to creep relaxation at each tem- (1000) 117 (17) 48 (7)
A737 Grade B 538
perature are seen in Table 2 . As a result of this exposure, 593 (1100) 48 (13) 41 (6)
the A543 Type B Class 2 and A387 Grade 22 had yield 621 (1150) 41 (6) 21 (3)
A543 Type B 538 (1000) 220 (32) 62 (6)
strength losses of about 15% for exposure up to 621° C Class 2 593 (1100) 117 (17) 62 (9)
(1150° F) while their tensile strength losses were vari- 621 (1150) 96 (14) 55 (8)
able between 6 and 10% . The smallest loss occurred in A387 Grade 22 538 (1000) 303 (44) 62 (9)
Class 2 593 (1100) 207 (30) 55 (8)
A737 Grade B which maintained its original yield 621 (1150) 186 (27) 48 (7)
strength but had a tensile strength decrease of 17% . In 649 (1200) 124 (18) 41 (6)
terms of actual strength, the loss was about the same as 677 (1250) 76 (11) 34 (5)

that of A543 Type B and A387 Grade 22 . The strength a Stress (a)= K — C 1 log time (hrs) (MPa) (S).
decline for A387 Grade 22 continues at 649° C (1200° a = K — C l log time (hr) (ksi) (E).

Relaxation Stresses in Pressure Vessels 19



Comparison to Creep and Stress-Relaxation Data ferred to later.

The results of the tests reported here are neither pure Since the testing apparatus itself acts as a load
stress-relaxation nor creep curves but something in maintaining device, the contribution of this effect
between . It is desirable, therefore, to place them in should be constant for a given test condition . Although
proper perspective by making a comparison with ex- the design of the experiment did not allow control of this
isting data of these two types . This is possible to do with variable, its effect could at least be estimated from the
the A543 Type B Class 2 steel because previous PVRC spring constant of the testing load train . Dr . M . J.
programs had produced creep-rupture data for this Manjoine has suggested a procedure 7 by which the
material. 4 Substantial data have also been generated load-time data obtained in this investigation can be
over the years on A587 Type 22, 4°5 because of its normal adjusted using the spring constants of the specimen and
service temperature range . The comparison between the the load to perform an iterative recalculation of ap-
program data and creep stresses is not entirely obvious, parent and real load decreases and time periods to
however, unless a creep rate can be established, or a produce approximate stress-relaxation curves . The data
relative amount of creep selected which will correspond of this investigation, as adjusted by this method, are
to the total extension in this test. It is clear that selecting shown in Figs . 5—7 . The procedure used is listed in Ap-
the stress to produce rupture failure in the time speci- pendix I.
fied for stress relief is not relevant, as rupture is not Since the design of the experiment did not include a
normally induced in stress relieving operations . This is consideration of this effect, the test apparatus was far
because deformation in stress relieving is usually limited from ideal for the calculation, and these recalculated
to strains on the micro- rather than the macro-level . One stress-relaxation data must be considered relatively
comparison that can be made is to the stress to induce crude . However, the above quoted literature and
1% total creep deformation, and such data have been stress-relaxation data obtained for A543 Type B and
tabulated for A543 Type B Class 2 and A387 Grade 22 A387 Grade 22 are in reasonable agreement with the
in a previous PVRC program data summary . 4 corrected stress-relaxation curves, Figs . 5—7, on which
As might be expected, the stresses for 1% creep de- the literature data are placed . The adjusted curves for
formation are considerably higher than reported here A387 Grade 22 are also in agreement with a more ex-
as existing after stress relief for 1—10 hrs . For example, tended unpublished data 8 for this grade and these are
a relieved stress of 62 MPa (9 ksi) is reported after 8 hrs compared on Fig . 8. In each case, as seen from Fig . 8, the
at 593° C (1100° F) for A543 Type B Class 2 . Previously initial stress level was higher than used in the Lehigh
1% measured creep stresses for this steel at 593° C were work, raising the whole curve somewhat . Within this
138 MPa (20 ksi) . Similarly, data for A387 Grade 22 puts limitation, the agreement is good and it is proposed that
1% creep stresses for 8 hrs at 538° C (1000° F) at 496 the corrected curves may at least be considered an en-
MPa (72 ksi) . The current tests puts the relaxed stress gineering approximation of the true stress-relaxation
value at 250 MPa (26 .6 ksi) . Thus the creep-relaxation stress levels.
stresses are less than half of the pure creep values. For the A737 Grade B, for which no creep or stress-
With respect to the stress-relaxation data, the liter- relaxation data could be obtained, some independent
ature data may be used for comparison but also with data derived from the residual stress program are
some ambiguity . This is mainly because the tests per- available .' These consist of residual stress measure-
formed here had very high initial stress values (the yield ments made on a weldment given a 2-hr stress relief at
point) whereas normal tests are at much lower initial 550° C (1022° F) . The relaxed stresses, for the most
stresses . Since the measured stresses are lower initially, part, were about 69—83 MPa (10—12 ksi), which falls just
they remain so during the test . However, for both A387 above the recalculated line for the 1000° F data from
Grade 22 and A543 Type B Class 2, some literature data
do exist which may be used for comparison .5 For ex-
ample, literature data are available for A387 Type 22,
5
at 600° C (1112° F) for an initial stress of 176 MPa (25 .5 I

A543 Type B Class 2

ksi) . After 5 hrs, the relaxed stress was 104 MPa (15.0 300
ksi) . These data may be compared to data obtained in
this investigation at 593° C (1100° F) and at 621° C
538°C (1000°F)
(1150° F) . For a heat of A543 Type B Class 2, at a tem-
perature of 538° C (1000° F), for an initial stress of 352 593°C (1100°F)
0
MPa (51 ksi), the 1-hr relaxed stress was 138 MPa (20 H 20
LITERATURE I-
ksi) and the 10-hr stress was 83 MPa (12 ksi) . Similarly, Sc 538 C (1000°F )
\ _ 00
for 593° C (1100° F), at an initial stress of 271 MPa (39 821°C `
10 LITERATURE cc
`93°C (1100°F
ksi), the 1-hr relaxed stress was 76 MPa (11 ksi) and the
5-hr was 34 .5 MPa (5 ksi) . When these data are com- 0
pared to the project data, the true stress-relaxation data 0 .25 0 .5 1 .0 2 .0 5 .0 10 20
RELAXATION TIME, hrs
fall at or below the data measured in this program . This
is as it should be because of the load-maintaining as- Fig . 5—Recomputed stress relaxation curves for A543 Type B Class
pects of these tests . These literature data will be re- 2

20 WRC Bulletin 302



.50
DATA POINTS IN PARENTHESIS
A737 Grade B ARE INRIAL VALUES A387 Grade 22 Class 2
300 300

40
a 40
— CALCULATED _
a
o 593 °C (1100 °F) • EXPERIMENTAL
N
N N
N
30 30
200 p~ •n321°C (1150°
N
0 650°C (1200°F) o
20
— 538°C (1000°F) - O 0
• I-
a o

00 j
593°C (1100°F
00 Q 0 n •
• o ° (LI
s
n 877 °C (1250°F) n _
---------------
-------
821°C(1150°F ------ _
0
0.1 0 .25 0.5 1 .0 2 .0 5 .0 10 20 0 .25 0.5 1 .0 2.0 5 .0 10 20

RELAXATION TIME, hrs RELAXATION TIME, hrs

Fig . 6—Recomputed stress-relaxation curves for A737 Grade B Fig . 8—Comparison of recomputed stress relaxation curves for ex-
perimental data for A387 Grade 22

this program.
The data from this investigation may be used as an
Application of the Data
approximation of minimum stresses (recalculated
The problem of residual stresses remaining after
stress-relaxation curves) or an estimate of stress re-
stress relief is a complex one, and difficult to measure
maining in complex weldments in which some elastic
either directly or indirectly except for simple cases . For
response is anticipated (data curves) after extended
a straightforward application, bolting being a classic
stress relief.
case, stress-relaxation data are most appropriate . For
a rather complex weldment, this may not be so, as other
Conclusions
factors will play a role . When examining the data from
this program, the adjusted data for stress-relaxation The results of creep relaxation tests on three pressure
form a lower bound on the residual stress expected . The vessel steels, A737 Grade B, A543 Type B Class 2 and
creep stress is the very upper limit but probably much A387 Grade 22, may be summarized as follows:
too high . It is the authors view that the curves obtained 1. After 15 min at temperature, substantial rapid
here, which incorporate a definite spring effect from the
creep relaxation occurred for the steels at the
testing procedure, provide a reasonable engineering
temperatures studied. Subsequent relaxation oc-
approximation for some of these more complex cases.
curred in a logarithmic manner with time.
For example, the initial stress in many stress relaxation
2. Creep relaxation at 538° C (1000° F) for times of
studies is less than the yield point . However, for weld-
8 hrs and beyond resulted in stresses less than 69
ments, initial stresses are closer to the yield point, which
MPa (10 ksi) for the A737 Grade B steel but re-
corresponds to the values used in this study . Unfortu-
sulted in stress levels above 138 MPa (20 ksi) for
nately, the exact load-maintaining nature of the test and
the other steels.
of the structure cannot be matched in most practical
3. Stresses below 69 MPa (10 ksi) were observed in
cases, and thus a range of stress can occur.
A543 Type B Class 2 steel for times of 8 hrs at 593°
C (1100° F) and in A387 Grade 22 Class 2 steel for
times of 2 hrs at 649° C (1200° F) . Stresses below
69 MPa (10 ksi) were not observed for in A387
5 Grade 22 Class 2 steel times out to 8 hrs at 1150°
A387 Grade 22 Cla ss 2
300 F.
538°C (1000°F)
4. The room temperature strength properties of the
2
593°C (1100°F) _ steels decreased as a result of the exposure treat-
H
N
200 pc
ments . At temperatures employed for creep re-
821°C (1150°F~
H laxation to below 69 MPa (10 ksi) residual
2 :2 11_ stress—538° C (1000° F) for A737 Grade B, 621°
849 °`
C (1200 F)
_
_ _LITERATURE
_ n 800°C (1112°F) -
C (1150° F) for A543 Class 2 and 649° C (1200° F)
677°C (1250°F _~
for A387 Grade 22 Class 2—strength reductions
cc 10 ---
– were mixed but about 5% for A737 Grade B, about
10% for A543 Type B Class 2 and about 20% for
0
0.25 0.5 1 .0 2.0 5 .0 10 0 A387 Grade 22 Class 2.
RELAXATION TIME, hrs 5. Comparison of these data to creep-rupture and
Fig . 7—Recomputed stress-relaxation curve for A387 Grade 22 Class
stress-relaxation data show the measured values
2 to be intermediate between these two limits, a

Relaxation Stresses in Pressure Vessels 21



result of the nature of the test . A semi-empirical This load drop is actually less than would occur under
computation method was used to construct ap- true stress-relaxation testing because of the behavior
proximate stress-relaxation curves from the of the testing machine which acts as a large spring to
data. maintain the load . The actual load drop should be cor-
rected by (K 1 + K 2 /K 1 ), to a new larger value, X2 . The
References
data point at X I should really be at X 2 so a new point,
1. Zhou, R . J ., Pense, A. W., Basehore, M . L ., and Lyons, D . H ., "A Study of
Residual Stress in Pressure Vessel Steels," Pressure Vessel Research Committee calculated from X 1 is plotted at X2 (point b) . This is the
Final Report, Oct . 1984. first point of the recalculated curve . Calculating the next
2. Zhou, R . J ., Pense, A . W ., Basehore, M . L ., and Lyons, D. H., ASTM
Specification E150-75 "Standard Recommended Practices for Conducting point requires that the creep-relaxation data curve at
Creep and Creep-Rupture Tension Tests of Metallic Materials Under Condi-
tions of Rapid Heating and Short Times," ASTM Annual Book of Standards, load X 2 (point c on curve) be examined . The next load
Part 10, p . 332 (1979) ASTM, Philadelphia, Pa. drop to be recalculated, which can again be X 1 in value,
3. Zhou, R . J ., Pense, A . W ., Basehore, M . L ., and Lyons, D. H., ASTM
Specification E328-78 "Standard Recommended Practices for Stress Relaxation is marked off on the creep-relaxation data curve and a
Tests for Materials and Structures," ASTM Annual Book of Standards, Part
10, p. 471 (1979) ASTM, Philadelphia, Pa. new load drop, which will again be X2 if X I load drop
4. de Barbadillo, J . J ., et al ., "Creep-Rupture Properties of Quenched and
Tempered Pressure Vessel Steels Data Summary, WRC Bull . 136, Dec. is used, is calculated . However, this data point is not
1968 . plotted at the time interval for X 2 , i .e ., point c on the
5. Bynum, E . E ., Ellis, F . V., and Roberts, B. W ., "Tensile and Creep Prop-
erties for an Annealed vs. Normalized and Tempered 21/4 Cr-1 Mo Steel Plate," time axis but is translated backward so that the time
Chrome-Moly Steel in 1976 ASME MPC-4, 1976.
6. Manjoine, M. J. and Voorhees, H . R ., Compilation of Stress-Relaxation interval between the first recalculated point, point b,
for Engineering Alloys, ASTM Data Series Publication DS60 ASTM, Phila-
delphia, 1982. and the new recalculated point, point e, is equal to the
7. Manjoine, M. J ., Westinghouse Electric Corporation, Research and De- time interval between points c and d . This is done be-
velopment Center, Private Communication.
8. Manjoine, M . J., "Final Report of Stress-Relaxation of Grade 22, 2 1/4 Cr-1 cause the time interval for the load drop is a function
Mo Steel Plate," Westinghouse Research and Development Center Report to
Chicago Bridge and Iron Co. To be published. of the load applied . If the true load is lower, the time
interval for the next drop is longer than would be true
Appendix I if the load is maintained . Thus the time interval for the
Approximation of Stress-Relaxation Curves from second true load drop must be determined by reference
Creep-Relaxation Data to the data curve at the true load and then retranslated
This empirical method provides for a recalculation back to the correct time interval on the recalculated
of creep-relaxation data to produce a stress relaxation curve . The remaining data points on the recalculated
curve by adjusting the load on the specimen for the curve are obtained by repeating the above procedure.
spring loading effect of the testing machine . To perform One of the results of this manner of building up the
this calculation, the spring constant of the testing ma- recalculated curve is that the curves produced by re-
chine and the specimen must be known . The procedure calculation can only cover a time period less than the
is an iterative one which requires that creep-relaxation data, and thus the recalculated curves always stop short
data over a reasonable time period be available to allow of the data curves.
for iteration . If the spring constant of the testing ma-
chine is K2 and that of the specimen is K 1 , the procedure
is illustrated below on a hypothetical data curve ob-
tained with a compliant testing machine such as used
in this investigation, Fig . Al . The first step is to select
a load drop of X I lbs . which occurs in the creep-relax-
ation (Lehigh method) data (point a) .

LOG TIME

Fig . A-1—Recalculation method for stress relaxation data (see text)

22 WRC Bulletin 302



A Study of Residual Stress in Pressure Vessel

Steels

by R. J. Zhou, A. W . Pense, M . L. Basehore, and D . H . Lyons

CONTENTS

Abstract 23 face (weld face) stresses both in the center of and at a

Introduction 23 cut face were determined . Stresses through the thick-
Experimental Procedure 24 ness of the weldment on a cut face were also measured.
Materials 24 Two techniques were used in this study.
Welding Process and Procedures 24 It was found that the residual stresses perpendicular
Methods for Measurement of Residual Stress 24 and parallel to the weld seam were high and tensile at
The Hole Drilling Method 24 the weld surface in the center of the weldment length,
Battelle Chip Removal Technique 25
Experimental Results 26 but these stresses decline sharply as the cut face is ap-
Distribution of Welding Residual Stresses in proached . The maximum tensile stresses were less than
the Weld of A543 Class 1 Steel 26 the yield strength of the steel . Large residual com-
Lehigh Hole Drilling Results 26 pressive stresses were found at the center thickness of
Battelle-Columbus Results 27 the weld on the cut face . Both measurement techniques
The Effect of Heat Treatment for Relief of
Welding Residual Stresses in A737 Grade B and gave essentially equivalent results.
A737 Grade C Steel Weldments 27 The two A737 microalloyed steel weldments were also
Discussion 28 made using the submerged arc process and a matching
Distribution of Welding Residual Stresses strength electrode . These were cut into 75 mm blocks
on the A543 Class 1 Weldment 28 prior to testing, so some of the as-welded stresses were
Lehigh University Data 28 changed . The stress relief experiments showed that
Battelle-Columbus Data 31
31 nearly full stress relief for these steels could be achieved
Comparison of Techniques
The Influence of Stress Relief on the after a treatment of 2 hrs at 550° C.
Distribution of Residual Stresses ii1 A737
Grades B and C 32 Introduction
Conclusions 32
References 32 Welding residual stresses are an important inherent
property of welded structures, especially heavy weld-
ments such as large pressure vessels . Not only can they
combine with stresses from service to alter loading and
Abstract
failure behavior but even before the weldment sees
This investigation determined the levels of residual service, welding residual stresses are one of the impor-
stress found in three pressure vessel steel weldments tant elements contributing to potential weld crack
using the blind hole drilling and Battelle Chip Removal formation . Research on the magnitude, direction and
techniques . The purpose of the program was to study distribution of welding residual stresses can therefore
residual stress levels in a high strength steel weldment, provide information not only useful for understanding
the effect of stress relief in two microalloyed weldments, of stresses in vessels but also for improving the quality
and to compare the results from the two measurement of weldments made as part of various fabrication op-
techniques. erations.
Surface residual stresses both longitudinal and A wide variety of procedures can be used to measure
transverse to a multipass submerged arc weld seam were residual stresses . These include sectioning followed by
measured in a heavy section A543 Class 1 weldment dimensional change determination, strain gaging and
made with a high strength welding electrode . Top sur- sectioning, hole drilling or machining near or between
strain gage arrays, X-ray diffraction and many others .'
Each of these methods has certain strengths and
weaknesses, and none has been found to be universally
R. J . Zhou and Prof. A . W . Pense are with the Department of Metallurgy and
Materials Engineering, Lehigh University, Bethlehem, PA . M. L . Basehore superior . In fact, there is still much uncertainty over the
and D. H . Lyons are with the Battelle-Columbus Laboratories, Columbus,
OH . accuracy of each of these methods, and about their

Residual Stress in Pressure Vessel Steels 23

comparability . One of the most uncertain areas is how Welding Process and Procedures
accurately local stresses, say at a point in a weld heat- The finished size of the weldment of A543 Class 1 was
affected zone, can be measured . X-ray diffraction 595 mm long, 256 mm wide, and 150 mm thick . This
techniques are able to measure these stresses in quite weldment was made by the submerged arc process with
a small area, say a 2 mm square, but the data produced Armco XW-28 weld metal using Linde 0091 flux . The
by such measurements are quite scattered . 2 The older weld preparation was a single V and the heat input was
sectioning methods cover more area which produces less 4 KJ/mm . After welding, the weldment was given a 315°
scatter, but are global in nature, thereby tending to C PWHT for 6 hrs to prevent cracking and was air
produce lower or average stresses and not peak ones. cooled.
The hole drilling and local machining methods are in- The finished weldments of A737 Grade B and A737
termediate in that they include a finite area but do give Grade C were 830 mm long and 400 mm wide . The
a reasonable measure of peak stress. Grade B weldment was 103 mm thick while the Grade
For this reason, gaging and hole drilling was one of the C was 76 mm thick . These weldments were submerged
methods used in this study . Chip machining of a strain arc welded with Armco W-19 weld metal using Linde
gaged area was used as a second method . This method 7095 flux. The weld preparation was "K" shaped and
was developed by Battelle-Columbus Laboratories and the heat input was 2 .77 KJ/mm. After welding, the
was applied by them . The hole drilling method was a weldment was air cooled without heat treatment.
standard technique frequently employed at Lehigh In order to measure mechanical properties for an-
University and was used by these investigators. other program, the A737 Grade B and A737 Grade C
The purpose of the study reported here was 2-fold. specimens were cut to provide cross sectional pieces 75
The first was to determine the level of residual stress mm thick . The cutting of the weldment probably re-
found in a high strength alloy steel weldment and to moved some of the constraint inherent in the original
determine the effects of thermal stress relief on the level weld . Thus, measurement of the as-welded residual
of residual stress in a lower strength microalloyed steel stresses was compromised as some of these stress pat-
weldment . The hole drilling method was used in both terns were probably altered . These pieces could be used
of these determinations . The second part of the study to study the influence of post-weld stress-relief heat
was to repeat the residual stress meaurements in the treatment on the reduction of the welding residual
high strength steel using the Battelle chip removal stresses in the 75-mm samples, i .e ., the extent of relax-
method so that a comparison could be made between ation of residual stress from initially quite high values,
the two methods and a check of their probable accuracy but possibly not peak values of stress.
obtained . The materials used were part of an ongoing A 2-hr stress relief heat treatment at 550° C was used
study of high strength steels conducted by the Pressure for A737 Grade B and A737 Grade C to study the effect
Vessel Research Committee . The work reported here of post-weld heat treatment on residual stresses.
was jointly sponsored by the Subcommittee on Cyclic
Methods for Measurement of Residual Stress
and Creep Behavior of Components and the Subcom-
Hole Drilling Method . The widely used technique
mittee on Thermal and Mechanical Effects of the
for measuring residual stresses, the "Blind Hole Drill-
Fabrication Division of the Pressure Vessel Research
ing" method (Fig. 1), was used . With this method, strain
Committee.
gages are installed in a rosette on the surface of the
strained sample and measurements of strain are made
before and after a small shallow hole is drilled between
Experimental Procedure the gages . The change in residual stresses is computed
from the change in strain recorded from the data . The
Materials holes in this investigation were drilled with a standard
The materials used were three pressure vessel steels: Photolastic Hole Drilling kit using a 1 .6 or 3 .2 mm drill
A543 Class 1, A737 Grade B and A737 Grade C . The and a RS-200-01 milling guide . The strains were mea-
chemical compositions of the steels are shown in Table sured with a Vishay model P-35A strain indicator using
1 and their mechanical properties are shown in Table 062 RE and 125 RE gages 3 (three 120 S2 gages mounted
2. in a rosette).

Table 1-Chemical Composition of the Steels and Weld Metals

Steel or
Weld Metal C Mn Si P S Cr Ni Mo Cu Al Other

A543 Class 1 0 .16 0 .29 0 .32 0.007 0.02 1 .73 3.27 0 .47 0 .13 0 .032 -
A737 Grade B 0 .14 1 .44 0 .19 0.09 0.009 0.006 0.22 0 .28 0 .27 0 .030 Cb .025
A737 Grade C 0 .18 1 .29 0 .30 0.005 0.007 0 .18 0.15 0 .08 0 .009 V .09
N .016
Armco XW-28 0.14 1 .78 0 .28 0.006 0 .006 0 .50 2 .12 1 .02 0 .09
Armco W-19 0.09 0 .80 0 .15 0.010 0 .008 0 .08 3 .50 0.03 0 .25

24 WRC Bulletin 302



Table 2—Mechanical Properties (Longitudinal) of the Steels and Weld Metals

Steel or Heat Yield Str . Tensile Str . Elong. Red . of

Weld Metal Treatment MPa (Ksi) MPa (Ksi) % Area

A543 Class 1 Q& T 703 (102 .) 813 (118 .) 23 .7 63 .3

A737 Grade B Q& T 444 (64.5) 568 (82.5) 31 .1 —
A737 Grade C Q& T 469 (68 .1) 566 (85.1) 30 .3 74.6
Armco XW-28 (as welded) 740 (107.0) 848 (123 .0) 20 .0 63 .8
Armco W-19 (as welded) 511 (74 .1) 636 (92.2) 22 .3 65 .3

The advantages of this method are simple equipment The pattern of holes drilled in the A737 Grades B and
and convenient operation . The disadvantage is the C weldments were those shown in Fig . 3 . Here the holes
reading can not be repeated at the same location . Be- were confined to cut faces on the reduced sections cut
cause of this, data repetition is impossible and care must from the weldment (75-mm wide segments) . Holes were
be used to produce accurate results. Some of the vari- drilled near the surface of each weld face and at the
ables influencing the measurement of strain are the center thickness of the weld . Only transverse weld
smoothness of the surface to be measured, the speed of stresses were measured . Measurements were taken both
drilling, the roundness of the hole, and the stability of before and after post weld heat treatment. Since two
the gage mountings . Accurate data can only be obtained measurements were needed at each point, two holes in
when all procedures are controlled and reproduced. roughly equivalent positions were drilled in each welded
One of the inherent problems with this method is the section (see Fig . 3).
fact that the machining (drilling) of the hole produces All of the hole drilling method work was done at Le-
some residual stresses from mechanical working . There high University . The area used for measurement was
is no method by which the error so introduced can be smoothed in the regions where the gages were placed by
calculated, but rather it must be empirically determined careful grinding. The weldment was shipped to Bat-
by experimental data . The stress introduced by ma- telle-Columbus Laboratories after these measure-
chining the hole has been estimated as being 40–50 MPa ments.
(6 Ksi) . This will be discussed later in the light of the Battelle Chip Removal Technique. The surface
data obtained. residual stresses were also determined using the Battelle
The pattern of holes drilled in the weldment of A543 Chip Removal Technique . This method involves cutting
is shown in Fig . 2 . It will be noted that stresses were a small pyramid of material (chip), with properly ori-
measured along the length of the weldment (line C-C) ented strain gages attached, from the specimen surface.
and transverse to the weldment (lines A-A and B-B) . As Removal of the chip relaxes the strain field and the re-
this weld was cut from a larger weldment, there was also sulting strain changes are measured by the strain gage
a cut face on which stresses in depth could be studied. attached to the chip . The relaxation of the stress field
Measurements along lines D-D and L-L were taken to is almost total for this technique, as opposed to the
explore these stresses . In each case, both longitudinal fractional relaxation for the hole drilling technique, if
and transverse stresses in relationship to the direction the chip is very thin.
of welding (along the X, Y, or Z axis, depending on the For this investigation, biaxial strain gage rosettes
line of holes) were measured . Since all measurements were used . The gages were installed at preselected sta-
were on surfaces, it is assumed that stresses transverse tions, with the gages being parallel and transverse to the
to the surface were zero at this location . weld line . The chips were then removed by cutting

ALIGNMENT HOLE MILLING

Collet
rleaible Coupling
Depth Control

Guide Bar
T
Type RE
Rosette
1

Fig . 1—Procedure and equipment for hole machining Fig . 2—Locations of points of measurement

Residual Stress in Pressure Vessel Steels 25



Q ,ksi

-20
o C-4
--30 Distance from weld center line, mm
-40

Fig . 7—Longitudinal surface residual stresses in A543 pressure vessel steel along line C-C

shows a plot of the distribution of residual stresses along comparison with the data from the hole drilling method.
line L-L, which is located 43 mm below the upper sur- It should be remembered that chips were removed from
face . Fig . 11 shows a plot of the distribution of residual lines offset from the hole lines by about 12 .5 mm which
stresses along D-D, located in a cross sectional surface may influence their values.
perpendicular to C-C.
Effect of Heat Treatment for Relief of Welding
Battelle-Columbus Results . The results of the Residual Stresses in A737 Grade B and A737 Grade C
chip removal method tests are shown on Figs. 7—11 for Steel Weldments

a, ksi + Along X axis

70 o Along Y axis
+ A-2

Fig . 8—Transverse surface residual stresses in A543 pressure vessel steel along line A-A

Residual Stress in Pressure Vessel Steels 27



Battelle data taken LEHIGH

12 .5 mm from line B-B + Along X axis
o Along Y axis

BATTELLE
• Along X axis
m Along Y axis

r S

Fig . 9—Transverse surface residual stresses in A543 pressure vessel steel along line B-B

The locations of points of measurement are shown in lumbus investigations, did the tensile residual stress
Fig . 2 . The residual stress relief data of A737 Grade B approach the yield strength, 700 MPa (102 Ksi).
and A737 Grade C are shown in Figs . 12 and 13 . In each On the other hand, the peak compressive residual
case, as-welded residual stresses were measured on one stresses are much higher, and do approach the steel
surface . Post weld heat treatment to remove residual yield point. As may be seen from Table 2, the weld
stresses for the 2 hr at 550° C were then applied . Re- metal, ARMCO XW-28, is actually stronger than the
maining residual stresses were measured at similar lo- base plate and it would be expected that base plate
cations on the opposite surface of the weldment after properties would control . Thus the high strength
such treatment. weldment does not follow the expected patterns of yield
strength level surface stresses . It has been shown,
Discussion however, that a thin compressive surface layer of weld
Distribution of Welding Residual Stresses in the A543 metal can exist on the weld face . The work of Macher-
Class 1 Weldment auch and Wohlfahrt4 indicated that these are due to
Lehigh University Data . From the distribution of other causes of residual stress . One is a rapid cooling of
welding residual stresses on the surface along line C-C the weld bead surface, which was called the "quenching
(Fig . 7), the center line of the seam (x = 0), the residual effect. " Another possible moderating effect is the "phase
stresses in the longitudinal direction are tensile . The transformation effect . "
maximum stress is in the middle of the seam and the Quenching effect residual stresses are caused by
stress decreases towards the edges of the weldment. In cooling which is not homogeneous throughout the depth
the transverse direction, the residual stress distribution of a weld bead. Actually, the bead surface may cool more
is similar to the longitudinal one . Stresses are tensile rapidly than the interior of the weld . Because of the
midway between the center line and the ends of the temperature differential, thermal stresses appear over
weldment . This result is the one to be expected. The last the weld cross-section which can cause local residual
passes to solidify are restrained by the preceding ones stresses . Typically, heterogeneous deformation occurs
and the surface stresses are tensile . It should be noted caused by rapid cooling and hardening of the surface
however, that the maximum tensile residual stress is less while the interior is still hot . This is followed by cooling
than 415 mPa (60 Ksi) and, in fact, in none of the de- of the center region, putting the surface in compres-
terminations, in either of the Lehigh or Battelle-Co- sion.

28 WRC Bulletin 302



ksi

L-6
+,
100 120
L-6

Distance from weld

center line, mm

Battelle data taken LEHIGH

12 .5 mm above line L-L + Along X axis
* Along Y axis

BATTELLE
o Along X axis
+ L 0 1 -9 0 • Along Y axis
L-1 L
-100
Fig . 10—Transverse residual stresses under surface in A543 pressure vessel steel along line L-L

If rapid cooling of the surface was the only source of Normally, however, these stresses reverse on continued
residual stresses, compressive stresses would arise at the cooling as the thermal contraction stresses predominate,
surface of weld passes and would be concentrated on the but the transformation stresses influence the final re-
surface of each pass . These stresses would be in equi- sult.
librium with the tensile stresses in the inner part of the An examination of the chemical composition of the
pass . Subsequent weld passes would alter this pat- XW-28 weld metal indicates that it is highly harden-
tern. able, i .e., that it readily transforms to lower bainite or
Depending on the cooling rate present in a welded martensite, which accounts for its substantial as-welded
plate, a phase transformation from austenite to ferrite, strength . Under these conditions, the compressive
bainite or martensite will occur . Because each phase has transformation strains will be greater than expected
a different specific volume, the strains associated with because transformation products more voluminous than
these transformations vary . The weld and HAZ which pearlite will form . These transformation strains are, of
are being transformed tend to expand, but the expan- course, superimposed on thermal contraction stresses,
sion is hindered by the cooler material not being but the end result is differential contraction and a re-
transformed . Thus the area being transformed is duction of residual stress from the level anticipated
subjected to compression stress, and tensile stress when higher temperature transformation products form
should exist in the regions not being transformed . in the weld . Thus both the quenching and transforma-

Residual Stress in Pressure Vessel Steels 29



Bottom Top
20
10 t
* .D-1 • 1
0 /140

-20 LEHIGH
+ Along X axis
* Along Z axis

-40
BATTELLE
/ o Along X axis
• Along Z axis
, -6b /
b
+ D-2
0
I
-80
+
D-1 D- 5
.+/D- 3
-100

-120
Fraction of thickness
Fig . 11—Distribution of residual stress across the thickness in A543 pressure vessel steel along line D-D

tion residual stress can work, in this weld metal, to lower this direction (— 250 mm as compared to -600 mm).
the overall residual stress . The fact that tensile residual Fig . 8 indicates that near line A-A, which is perpen-
stresses in high strength steels do not reach the yield dicular to the seam and located at the middle of the
strengths of the steel has been reported before. 5 weld, the longitudinal shrinkage effect is major and the
In fact, the various sources of residual stresses are not tensile stresses are dominant . Maximum tensile stress
independent of one another. They influence each other, is at the center of the seam (the width of the seam is
and lead to a complicated local state of stress, especially represented as "S" in the figure) and decreases towards
in the multipassiwelding of a heavy weldment . Gott 2 the edges of the plate . But at the end of the HAZ, the
pointed out that the residual stresses in the weld metal stresses became compressive . In the transverse direc-
of a heavy weldment can locally vary considerably from tion, the shrinkage length decreases and the restraint
100 to 200 MPa over a distance of 5 mm. or other effects may become less distinct, so the tensile
After the above general discussion, the distribution stresses are less, especially near the weld center line.
of residual stresses in Fig . 7 can be considered . In the Line B-B, for which the distribution of residual
center part of the weldment, the dominant residual stresses is shown in Fig . 9, is located at 3/4 length of the
stresses parallel to the seam are due to shrinkage . Since weld . The shrinkage effect is reduced at this location
the shrinkage residual stresses decrease as the distance such that the maximum tensile residual stresses do not
from the center point of the weld increases and the ends appear at the center line of the seam but at both edges
of the weld are approached, the tensile residual stresses of the seam . In the transverse direction, compressive
decay with the distance . In the X-axis direction, the stresses exist at the center line of the weld probably
magnitude of residual stress at middle point is about 1/3 from a sectioning effect, i .e ., a redistribution of stresses
to 1/2 of the Y-axis direction, which may be due to the on sectioning which reduces their value or changes their
more limited shrinkage in the transverse orientation, sign.
or due to the decrease in restraint in the weldment in Fig. 10 is the distribution of the residual stresses along

:30 WRC Bulletin 302



Along X axis Along X axis

Top 0 , ksi Top
Bottom Bottom
80
0
60
40
20 o--
o~-
+-

-50 -40 -30 -20 -10 10 20` 30 40 50

-20
Distance from center point, mm
-40
. 60
o. ` Distance from center point, mm
-60 o After Welding
o After Welding + After Heat Treatment
+ After Heat Treatment (T : 550 C, t :2hr.)
(T : 550 C, t :2hr.) Along Z axis a, ksi
Along Z axis 0 , ksi 80
60
_o 0_60
40 40 o
20 N
-50 -40 -30 -20 -10 10 20 30 --`*— 50
1-20 -20
-40 _40 Distance from center point, mm
Distance from-center point, mm
-60 -60 o After Welding
Fig . 12—Distribution of residual stresses after heat treatment in A737 + After Heat Treatment
Grade B steel ( T:550 C, t2 hr .)
Fig . 13—Distribution of residual stresses after heat treatment in A737
Grade C steel

line L-L, 43 mm under the surface on the cut face. Re-

sidual stresses in the weld and HAZ in the X and Z di-
rections are compressive and the maximum value is
always at the center line of the seam . Note that in the approach the yield strength of the steel.
X direction the compressive stress is 620 MPa (90 Ksi) Fig . 10 illustrates very good agreement between the
and that the stresses dramatically change into tensile hole drilling and chip removal results, except at the
stresses beyond the HAZ . Fig. 11 shows the distribution center of the weld . However, if we look at Fig . 10 in
of the residual stresses along the thickness line D-D conjunction with Fig . 11, the lower level compressive
(Fig . 6) . Except near the top and bottom surfaces, the stresses measured by the chip removal technique may
compressive residual stresses are present in all of the result from the fact that they were measured 12 .5 mm
weld . The stresses in the X direction are higher than the above line L-L, and therefore closer to the top surface
stresses in the Z direction . Once again, the compressive of the specimen . Fig. 11 shows that both the transverse
residual stresses approach the yield strength of the and longitudinal residual stresses are indeed increasing
steel. in that region . As a matter of fact, the chip removal re-
Battelle-Columbus Data. While the same general sults almost fall on top of the hole drilling results at this
trends observed during the Lehigh study are noted in location.
the Battelle-Columbus data, there are noticeable dif-
Comparison of Techniques
ferences between the data sets . For line B-B (once again
While it is not the intent of this investigation to
remembering that the Battelle data was actually taken
conduct an in-depth comparison of the two experi-
along a line 12 .5 mm from line B-B), the chip removal
mental techniques, a few conclusions drawn from this
results reveal higher longitudinal and transverse stresses
and earlier work s at Battelle should be stated.
in the center of the weld than measured by the hole
drilling technique . In addition, the magnitude of the (1) Both the chip removal and hole drilling methods
transverse stress at all locations along line B-B suggests are of similar accuracy in terms of the measured
a much less intense transverse stress field in the Bat- strains . However, since the magnitude of strains
telle-Columbus data, as can be seen from the data points measured for the hole drilling technique are only
in Fig . 9 . However, peak tensile stresses still did not a fraction of those measured for the chip removal

Residual Stress in Pressure Vessel Steels 31



method, the error in calculated stress may be Conclusions

greater for the hole drilling method.
The following conclusions can be drawn from the
(2) The hole drilling method was found to have a
study of welding residual stresses on the heavy section
constant imposed stress due to machining
welds of pressure vessel steels.
equivalent to -40µc (8 MPa).
(3) The chip removal method was found to be inde- 1. The residual stresses perpendicular and parallel
pendent of stress gradients (perpendicular to the to the weld are tensile at the surface or near the
surface) on the order of 10,000 MPa/mm . The surface and compressive in the center of the
hole drilling method was affected by stress gra- weldment thickness.
dients, averaging stresses over a depth of one hole 2. The largest residual stresses are perpendicular to
diameter unless analytically corrected. the weld (transverse direction) at the center
thickness of the weldment . They are compressive
Influence of Stress Relief on the Distribution of
Residual Stresses in A737 Grades B and C and have a value that is very close to the yield
The distribution of residual stress in A737 Grade B strength of the material.
and C are shown in Figs . 12 and 13 both in the as-welded 3. The largest residual tensile stress is parallel to the
condition and after post-weld heat treatment. The as- weld (longitudinal direction) at the center of the
welded residual stress patterns are somewhat similar seam . The tensile residual stresses are always
to those in the A543 weldment in that the surface layers substantially less than the yield strength of the
of the weld are in tension and the subsurface layers are steel in the high strength steel weldment.
in compression . These initial patterns are not without 4. The Battelle Chip-Removal technique produced
alteration due to sectioning of the weldment after residual stress level measurements essentially
welding and thus may not completely represent the equivalent to those by the hole-drilling technique
initial as-welded state . It is also interesting to note that although the results appeared more consistent and
tensile residual stresses in the A737 Grade C reach the less scattered.
yield strength in tension, but the compressive residual 5. The effect of stress relief treatment on specimens
stresses are not as great as would be expected in this of A737 Grade B steel and A737 Grade C steel
steel. The A737 Grade B has high compressive residual demonstrated that residual stresses were sub-
stress, but the tensile stresses do not approach the yield stantially reduced by a two-hour treatment at 550°
strength of the steel. This is seen as a disturbance of C.
normal patterns through sectioning. References
The post-weld heat-treatment is shown in these fig- 1. Masubuchi, K ., "Residual Stresses and Distortion," The Welding Hand-
ures to reduce these stresses to relatively low tensile book V 1, 6th ed., Chap. 5, pp. 5.1–5.44.
2. Gott, K . E., "Residual Stress in a Weldment of Pressure Vessel Steel,"
values, generally in the range of 35—70 MPa (5—10 Ksi). Residual Stresses in Welded Construction and Their Effects, International
Conference, The Welding Institute, London, 1977.
This demonstrates that the stress relief operation is 3. Gott, K . E ., "Measurement of Residual Stresses by the Drilling Method,"
effective, and that even high tensile or compressive Measurements Group, Visbay Intertechnology, Inc ., 1977.
4. Macherauch, E. and Wohlfahrt, H ., "Different Sources of Residual Stresses
stress can be reduced to modest levels in this steel as a Result of Welding," Residual Stresses in Welded Construction, Interna-
tional Conference, The Welding Institute, London, 1977.
through standard stress-relief treatments . The residual 5. Masubuchi, K ., "Models of Stresses and Deformations Due to Welding—A
stress level is within the range reported as the typical Review," Journal of Metals, 33, No . 12 (Dec . 1981), pp . 19-23.
6. Basehore, M. L. and Lyons, D . H ., "Investigation of the Residual Stresses
error introduced in the hole drilling method determi- Field in a Heavy Weldment, " Battelle-Columbus Laboratories, Feb . 1983.
nations, and since it is generally tensile, it may represent
merely the residual stress from the process of mea-
surement. In any case it is sufficiently low to eliminate
most of the concern about the effect of residual stresses
in service .

32 WRC Bulletin 302

INTENTIONALLY LEFT BLANK