WELDING RESEARCH

SUPPLEMENT TO THE WELDING JOURNAL. OCTOBER, 1985

Sponsored by the American Welding Society and the Welding Research Council

Readers are advised that all papers published in the Welding Journal's Research Supplement undergo
Peer Review before publication for: 1) originality of the contribution; 2) technical value to the welding
community; 3) prior publication of the material being reviewed; 4) proper credit to others working in the
same area; and 5) justification of the conclusions based on the results.
The names of the more than 160 individuals serving on the AWS Peer Review Panel are published
periodically. All are experts in their respective technical areas, and all are volunteers in the program.

Failure Investigation of Eddystone

Main Steam Piping
Cause of through wall cracks in 316 stainless steel is established

BY ). F. DELONG, ). E. BYNUM, F. V. ELLIS, M. H. RAFIEE, W. F. SIDDALL, T. DAIKOKU AND H. HANEDA

ABSTRACT. In March 1983, personnel at

Philadelphia Electric's Eddystone No. 1
power plant discovered a through wall
leak in the main steam outlet piping. This
pipe was designed to carry steam at a
pressure of 5300 psi (36,538 kPa) and a
temperature of 1210F (654C). The pipe
was made of 316 stainless steel and had
been operated approximately 130,000
hours at the time that failure was discovered. Subsequent inspection revealed
Paper presented in a technical session sponsored by the Metal Properties Council at the
65th annual A WS Convention held April 9-13,
1984, in Dallas, Tex.

that many O D cracks existed in this piping

system.
This paper details the investigation into
the cause of the failure. The following
elements are highlighted: the in-place
metallography which successfully used
the plastic replica technique; the elasticplastic stress analysis and life prediction
techniques carried out to assess probable
failure modes and loadings; and the
experimental stress analysis which was
conducted to confirm analytical hypotheses.

Introduction

The discovery in March 1983 of extenI F. DELONG is senior metallurgical engineer, I.sive cracking in the main steam lines at
E. BYNUM is supervisor, elevated temperature Philadelphia Electric's Eddystone Unit No.
testing, and F. V. ELLIS is principal research
1 power plant led to a forced outage of
engineer with Philadelphia Electric Co., Phila- the plant and the present failure investidelphia, Pa. M. H. RAFIEE is supervisor, metal- gation. This investigation was conducted
lography. CE Metallurgical & Materials Lab,
by Philadelphia Electric Co. (PECO), ComChattanooga, Tenn. W. F. SIDDALL is managbustion Engineering, Inc., and Mitsubishi
er, engineering development, CE Fossil Power
Systems, Windsor, Conn. T. DAIKOKU is man- Heavy Industries, Ltd.
ager, research laboratories, and H. HANEDA is Eddystone Unit No. 1 commenced
manager, boiler development, for Mitsubishi
commercial operation on February 5,
Heavy Industries, Tokyo, lapan.
1960. The plant was designed for turbine

conditions of 5000 psi (34,470 kPa) at

1200F (649C) temperature with two
reheats to 1050F (566C). The turbine
was rated at 325 M W . The unit is a coal
fired, supercritical monotube design with
a once through twin furnace manufactured by Combustion Engineering, Inc.
From the first startup to the discovery
of the main steam line crack, the 316
stainless steel piping system accumulated
130,520 hours of high temperature operation. During this period, the unit experienced 326 startup cycles, many of which
were due to forced shutdowns. The main
steam line, including the welds, performed satisfactorily until the first indication of a problem in 1980. At that time, a
small area of outside diameter cracking
was discovered by visual examination
near a hanger lug location. Immediately,
one-third of the piping system, including
all pipe sections from the same heat as
the cracked section, was examined with
dye penetrant. No other cracks were
found. The cracked section was removed
and high temperature stress rupture tests
were performed on it. The rupture data
was analyzed by parametric methods

WELDING RESEARCH SUPPLEMENT I 271-s

Table 1Chemical Composition of 4B

Steam Line from Eddystone Unit 1
Element

Wt. %

C
Mn
P
S
Si
Ni
Cr
Mo
V
Cb/Ta
Ti
Co
Cu
Al
B
N2

0.035
1.72
0.014
0.020
0.36
12.62
17.10
2.30
0.03
<0.01
<0.01
0.12
0.10
<0.01
<0.001
0.033

a n d indicated a d e q u a t e remaining c r e e p
r u p t u r e life. H e n c e , the unit w a s p e r m i t t e d t o o p e r a t e at n o r m a l conditions. It
w a s w i t h c o m p l e t e surprise that a
t h r o u g h wall leak in t h e m a i n steam
piping system w a s d i s c o v e r e d subseq u e n t t o these tests. T h e failure w a s in a
vertical section o f o n e of the steam lines
a n d w a s n o t w e l d related. T h e safety
p r o b l e m m a n d a t e d that the unit b e
repaired immediately; t h e n e e d t o insure
that t h e p r o b l e m w o u l d n o t recur

Fig. 7 Typical OD longitudinal cracking in 4B

main steam line

Fig. 2 Section of 4B steam line having nearly

through wall cracking

r e q u i r e d that the cause of t h e cracking be

f o u n d . T h e r e f o r e , an extensive failure
investigation w a s u n d e r t a k e n . This p a p e r
will discuss the details of this failure investigation in chronological o r d e r .

ness. T h e selection o f t y p e 316 p i p i n g

material a n d 16 Cr-8 Ni-2 M o w e l d i n g
e l e c t r o d e w a s based o n a metallurgical
study b y C a u g h e y and Benz (Ref. 1). All
the cracking w a s c o n f i n e d t o the piping
material w i t h n o w e l d s i n v o l v e d . Cracks
f o u n d b y the d y e p e n e t r a n t inspection
m e t h o d w e r e c o n f i n e d t o f o u r of the 14
heats used in t h e piping system a n d only
in that piping u p s t r e a m o f the boiler s t o p
valves. T h e main steam line consists of 8
parallel lines n u m b e r e d 1A, 1B, 2A, 2B,
3A, 3B, a n d 4A, 4B, originating at t h e
eight superheater outlet headers.

Preliminary Failure Assessment

T h e steam line is a f o r g e d and b o r e d
t y p e 316 stainless steel having a 9.05 in.
(229.87 m m ) outside d i a m e t e r ( O D ) and
2.525 in. (64.14 m m ) m i n i m u m wall thick-

All l a b o r a t o r y specimens
reported
here w e r e taken f r o m an 11 ft (3.35 m)
section of t h e 4B steam line containing
the leak p o s i t i o n . The chemical c o m p o s i t i o n of t h e piping is g i v e n in Table 1.
Specimens w e r e also sent b y PECO t o
Failure Analysis Associates (Ref. 2), U n i versity of Pennsylvania (Ref. 3), Sumitom o M e t a l (Ref. 4), General Electric C o .
(Ref. 5), Central Electricity Research Labor a t o r y (Ref. 6), O a k Ridge National Labor a t o r y (Ref. 7), a n d Ishikawajima-Harima
Heavy Industries. Standard light optical

Vh

Fig. 3 As received ring section with three major cracks

272-s | OCTOBER 1985

Fig. 4 Intergranular oxide filled cracks perpendicular

to hoop stress direction
(Unetched)

<
(

Fig. 6 Oriented causation in OD and midwallposition while ID region is undamaged

Fig. 5 Damage adjacent to main crack consisting of cavities and single grain facet (sgf)
cracks (Unetched)
and scanning electron microscopy (SEM)
techniques were used to examine sections of the failed pipe. In addition to the
metallographic
examination,
tensile,
creep, Charpy, and compact tension
tests were performed.
Metallographic Examination
The longitudinal cracking on the outside of the failed pipe is shown in Fig. 1, A
crack nearly through the wall and the
surface appearance of the cracking are
shown in Fig. 2. Figure 3 shows a transverse ring which has three major
cracks.
The damage in all specimens examined

was similar. At low magnification, intergranular oxide-filled cracks perpendicular

to the hoop stress direction (Fig. 4) were
found near the O D . Figure 5 shows single
grain facet (sgf) cracks and grain boundary cavities in the region adjacent to the
main crack. Use of a multiple etch polishing procedure (Ref. 8) facilitated detection of the cavities which occurred at
grain boundaries oriented either normal
or nearly normal to the hoop stress
direction. This oriented cavitation is
observed on the O D and midwall positions, but the inside diameter (ID) region
is free of damage. The density variation
of these cavities' through the wall thickness is remarkably consistent from one
circumferential location to another. Figure 6 shows this through wall variation at
a circumferential position 90 from the
major crack.
In every piece examined, the cracking
starts at the O D or immediately beneath
the surface, as shown in Fig. 3. It is
presumed that the oriented cavities link

Fig. 7-Oriented cavities on outside surface of pipe (50X)

up and form the sgf cracks, which precede the major cracks and subsequent
failure. Figure 7 shows the oriented cavities detected on the outside using a
tangential section rather than the more
common transverse sections. The combination of O D crack initiation and detection of oriented cavitation on tangential
sections led to the use of plastic replica
techniques for field inspection of the
main steam line at the plant.
Examination at higher magnification
revealed that the microstructure of the
failed pipe consists of matrix precipitates
(probably M23Q) and large sigma phase
particles in the grain boundaries.
Although decohesion at the sigmaaustenite interface is cited as a potential
failure mode for this sigma phase morphology, the cavitation found during this
investigation occurs primarily in the ligament region between the sigma phase
particles, as shown in Fig. 8.
It is possible to produce grain boundary fracture in the highly damaged micro-

Fig. 8 SEM photomicrograph showing grain boundary cavities in

ligament between sigma phase particles

WELDING RESEARCH SUPPLEMENT I 273-s

A?i

I4SEC

air<iNT

tv/ci

o e" "
u.

"

1L

A. J

1it

COAX
f/g. 9 Grain boundary cavities and small par- Fig. 10 EDAX showing high sulfur content of small grain boundary particles
ticles on cavity walls
structure even at room temperature and
relatively slow loading rates. Figure 9
shows multiple cavities on a grain boundary facet produced by room temperature
impact loading of a thin slice taken from
the O D of the failed pipe. The structure
consists of cavity walls (the ridges) and
small particles on the base of the cavities.
The largest particle in the center of the
photograph was analyzed by Energy Dispersive X-ray Analysis (EDAX) and shown
to contain sulfur (Fig. 10the molybdenum peak occurs at nearly the same
energy as sulfur, but is a double, while
the observed peak is symmetrical). At
high magnification, the cavities have the
faceted appearance shown in Fig. 11.
Similar observations have been made by
other investigators (Refs. 2-5). At
present, the interpretation of the source
and role of sulfur in the failure is controversial. Definitive answers to these questions are not provided by this study.
However, based on the work of Yoshihara, Kurahashi and Nii (Ref. 11), it is clear
that (for iron) native sulfur will segregate
at free surfaces (or grain boundaries)
given sufficient aging time.

100X Glyceregia

A transverse section of the impact

loaded specimen is shown in Fig. 12 and
cavities with small particles between large
sigma phase particles can be seen on the
grain boundary facets. The compact tension specimens were loaded slowly at
room temperature and also showed the
same small particle decorated grain
boundary facets in the creep damaged
region. A careful re-examination of Fig. 9
also reveals the presence of the small
particles in the cavities adjacent to the
sigma phase particles.

the intergranular cracks and cavitation

found in a creep specimen from the ID of
the steam line. Figure 15 shows fracture
faces exposed to low temperature deformation of the creep specimen. Numerous
sigma phase particles are seen on the
grain boundary facets, but the small particles found closer to the O D of the pipe
are missing.

The 2B steam line material removed in

1980 for creep rupture testing (Ref. 12)
was re-examined using the same techniques applied to the 4B line material. A
piece of material removed from the O D
of the 2B line and fractured at room
temperature is shown in Fig. 13. The
same small particles and cavity morphology found in the 4B line material are
present.
Creep specimens removed from the ID
of the 2B steam line had 3 to 4 times the
rupture strength of specimens taken at
other radial locations and subjected to
identical test conditions. Figure 14 shows

500X Glyceregia

Fig. 12 Transverse view of fracture face showing cavities with small particles on grain boundary facets

274-S | OCTOBER 1985

1500X Unetched

I-

Table 2Tensile Properties of 4B Steam Line from Eddystone Unit 1

z
UJ

Property

DE-1

Specimen Code
DE-2
DE-3

DE-4

Yield Strength (ksi)

Tensile Strength (ksi)
Elongation ("..)

44.1
60.0

42.9
47.3

44.1
66.3

Q.

42.8
83.6
10.0

UJ
Q

o
<

Table 3Charpy Impact Properties of 4B Steam Line from Eddystone Unit 1

Fig. 13 Room temperature fracture face of

2B steam line material removed in 1980 at
position near OD of pipe

Specimen
Location

Test Temperature
( = F)

Absorbed Energy
(ft-lb)

Shear

OD
1/4t

77
77
77
77
77
77
77
77
535
535
535
535
1150
1150
1150
1150

3
4
6
23
4
6
8
24
4

0
0
0
5
0
0
0
5
0
0
0
99
0
0
0
100

3/4t
ID
OD

1/4t
3/4t
ID
OD

Creep and Tensile Testing

1/4t
3/4t

A c r e e p test o n the 4B material r u p t u r e d in 270 hours at 1382F ( 7 5 0 C ) and

6.8 ksi (468.8 MPa). At t h e same c o n d i tions, the 2B material (Ref. 12) r u p t u r e d in
648 hours f o r a d i f f e r e n c e in log r u p t u r e
t i m e o f 0.38. Both c r e e p curves exhibited
similar p r i m a r y a n d s e c o n d a r y creep
b e h a v i o r u p t o strains o f a b o u t 0.5%.
Subsequently, the 4B material e n t e r e d the
tertiary c r e e p stage a c c o u n t i n g f o r the
shorter r u p t u r e life.

ID
OD
1/4t

3/4t
ID

9
10
38
5

7
10
36

(%)

>
o

LU

Lateral Expansion
(mils)

I-

UJ

1
3
18

1
4

>

5
21
2
10
15
39
6
7
11
41

Q.
I
UJ

o
rr
<
UJ

</>

UJ

UJ

S
a.
O
-J

steam line w h i c h c o n t a i n e d
several
cracks. T h e specimens w e r e p o s i t i o n e d
so their f r a c t u r e planes a n d t h e crack
planes h a d similar orientations.

Four circumferential tensile specimens

w e r e cut f r o m the O D of the 4B steam
line. T h e specimens had a 0.252 in. (6.4
m m ) diameter, a 1 in. (25.4 m m ) gage
length and w e r e tested at r o o m t e m p e r ature. T h e test results are listed in Table 2.
T h e relatively l o w ductility values reflect
the d a m a g e d c o n d i t i o n o f t h e material.

T h e C h a r p y impact testing w a s perf o r m e d in a c c o r d a n c e w i t h A S T M Stand a r d M e t h o d E-23. The data, w h i c h are

s h o w n in Table 3, indicate a significant
increase in toughness as the inside d i a m eter is a p p r o a c h e d . This coincides w i t h
t h e radial d a m a g e distribution revealed
b y the m e t a l l o g r a p h y . Increasing the test
t e m p e r a t u r e f r o m 77 t o 1150F (25 C
t o 6 2 1 C ) d o e s increase the toughness
s o m e w h a t , but n o t e n o u g h t o indicate
that any ductility o r fracture m o d e transit i o n takes place.

Fracture Testing
In o r d e r t o assess the susceptibility of
cracks t o p r o p a g a t e rapidly in the steam
lines, several f r a c t u r e toughness tests
w e r e c o n d u c t e d . C h a r p y i m p a c t specimens a n d 1 in. (25.4 m m ) thick c o m p a c t
tension specimens w e r e cut f r o m the

T h e c o m p a c t tension specimens w e r e
p r e p a r e d a n d tested in a c c o r d a n c e w i t h

UJ

>
X

o
<
UJ
(A
UJ

UJ

S
o.
O
_i

Fig. 15 Grain boundary facet, sigma phase

particles and sgf cracks on fracture surface for
specimen shown in Fig. 14. Specimen was
fractured at LN? temperature

u
>

a
rr
<
UJ
(A
UJ

tr

>
y

' < *.
s

^ ,

yTiT
.i'

' A

H
Z

I >C *

'< '-, Vv> 9M '> .-v..mh

$

oa.
I
UJ

>

<>

<

UJ
Ui

Fig. 14 Collage of creep specimen from 2B steam line. Specimen fractured at 750C (1382F) and 47 MPa (6.8 ksi)

WELDING RESEARCH SUPPLEMENT 1275-s

Table 4Fracture Toughness Properties of

4B Steam Line from Eddystone Unit 1

Specimen
Code
PBB
PBD
PBF
PBE

Test
Temperature

Fracture'3'
Toughness-KQ

(F)

(ksi \/1n7)

76
77
76
270

41.2
38.3
35.8
35.6

(a)

K Q was determined using the loaci at the intersection of a

45"., secanl offset line and the load-deflection plot. This is
considered a conservative estimate of the onset of stable
crack extension.

ASTM Standard Method E-399. The

numerical test data are shown in Table 4.
These specimens, which were loaded
slowly, did not exhibit any evidence of
fast fracture. Only slow stable crack
growth took place. Consequently, none
of the tests met the linear elastic behavior
requirements of the Standard Method.
The toughness values in Table 4 are
considered conservative estimates of the
stress intensity for crack propagation
because they were calculated using the
load at the intersection of a 95% secant
offset line and the load-deflection plot.
To prove that this technique does, in this
case, produce conservative values, specimen PBF was loaded to a point just
beyond where a 95% secant offset line
would intersect the load-deflection plot.
The specimen was then heat tinted and
broken open. Microscopic examination
of the fracture surface failed to locate
any evidence of crack propagation in the
heat tinted region. This indicated that the
stable crack growth started at a stress
intensity higher than the point used for
toughness calculations.
One compact tension specimen (PBC)
was located astride a crack found in the
steam line. This test was designed to load
and open up a "natural" crack and determine if there was anything at the crack
front that might initiate a fast fracture.
This specimen behaved the same as the
standard compact tension specimens
which had fatigue sharpened notches as
crack fronts.
Examination of the fracture surfaces
revealed only grain boundary separation
near the steam line outside diameter. As
the inside diameter was approached, on
either the succession of Charpy specimens or on the compact tension, some
grain distortion accompanied the intergranular separation.
Leachants from Insulation Polythionic Acid
Because the cracks were found to be
intergranular in boat samples removed
from three failed areas of the piping
system, insulation material was obtained
from locations in original insulation directly from the surface of the pipe. The

276-s OCTOBER 1985

insulation was taken from locations

where pipe cracking was evident to
determine contaminant levels for elements which may cause intergranular
fracture in austenitic stainless steel.
The insulation was analyzed to determine its pH value and weight percent
leachable elemental composition for
sodium, potassium and lithium. Also, the
chloride and sulfur concentrations were
obtained by colorimetric and x-ray fluorescence spectrometry, respectively. Results obtained from a typical analysis are
shown in Table 5.
Because of the basicity of the solution
and the low or non-existent leachable
amount of chloride or sulfur in the insulation, intergranular stress-assisted corrosion cracking was discarded as a cause of
failure.
Conclusions from Preliminary Assessment
The following statements summarize
the findings of the preliminary assessment
portion of the investigation:
1. The longitudinal cracks in the 4B
steam line originate at the outside surface.
2. The crack planes and direction are
predominantly normal to the direction of
the hoop stress.
3. In addition to the major cracks,
creep damage in the form of grain
boundary cavities is found fairly uniformly around the pipe circumference and to
within approximately V2 in. (12.7 mm) of
the ID.
4. The cavities occur primarily in the
ligament region between large sigma
phase particles.
5. On fresh fractures of material away
from the ID, the cavities are decorated
with small particles which have a high
sulfur content. It has been confirmed that
the sulfur did not come from external
sources.
6. The cavities are faceted.
7. The creep damaged material did
not exhibit any propensity for rapid crack
propagation.

Field Metallography
Technique Development
Field metallography of the steam pipes
was initiated with metallographic examination of the spool piece shown in Fig. 2.
A number of specimens were removed
from this section and metallographicaliy
prepared for microscopic examination, as
described earlier. Der, et al. (Ref. 8) had
developed a repeated polish/etch procedure for the detection of cavities in low
alloy steels. For the TP316 piping, the
specimen remote from the fracture was
experimented with using a repeated polishing/etching technique. After a number
of trials, a suitable combination was

Table 5Leachants from Insulation

Polythionic Acid pH of 50% Water
Slurry9.4 Basic
Element

Wt-%
+I

Sodium, as Na
Potassium, as K+1
Lithium, as Li +1
Chloride, as Cl~
Sulfur, as S0 4 " 2

0.032
0.019
0.000
0.014
0.000

developed, whereby the evidence of

creep damage, oriented cavitation, could
be detected.
Field Application
In order to adapt the repeated polishing/etching technique to field metallography, a series of experiments was
conducted on the O D surface of the test
piece. These experiments consisted of
using portable grinding and polishing
equipment, swab-etching, and repeated
polishing/etching, followed by plastic
replication of the etched surface. Confidence in use of this method of preparation was achieved when surface areas
void of any indications by visual and/or
dye penetrant examination showed clear
evidence of oriented cavitation known to
be present.
Using these procedures, a total of 72
locations at Eddystone Unit No. 1 was
examined and documented. These
inspections showed creep damage for all
piping upstream of the boiler stop valve
and no indication (oriented or sigmatized
isolated cavitation) for those pipes downstream of the boiler stop valves, even in
those pipes with the blocky sigma microstructure.
Validation of Data
In order to further insure the integrity
of the data obtained from field metallography, four boat samples were removed
from the outlet headers for laboratory
examination. The boat samples were
about V2 X V2 X 3 in. (12.7 X 12.7
X 76.2 mm) long and were removed with
the long dimension along the long'tudinal
axis of the header. Each boat sample
contained a surface previously examined
by the replicating technique. Two specimens were removed from each boat
sample and metallographicaliy prepared,
using standard laboratory technique, for
microscopic examination. One specimen
from each boat sample was mounted in
such a manner to permit viewing of the
surface at the same location as that from
which the replica was taken, while the
other specimen showed a transverse
view of the piping material at the same
location. The similarity of the three microstructures (replicating technique, direct
viewing and transverse view) is demon-

Fig. 76 Comparison of microstructure revealed by OD plastic replication and metallographic mount of boat sample for the superheater outlet header.
Left replicating technique; center direct viewing; right transverse view
strated in Fig. 16 for one location. The
other locations from which the boat samples were taken exhibited the same similarity.
Metallography Conclusions
From the tests conducted and the data
collected, it was determined that the
microstructure observed at the surface of
the components was a representative
structure and not merely a surface phenomenon. The material upstream of the
boiler stop valve showed creep damage
when sigma phase was present, and the
material downstream of the boiler stop
valve showed no creep damage, even in
pipe sections which had a blocky sigma
microstructure.

Failure Scenario
At this point in the failure investigation,
we had established the following facts
and observations:
1. Cracking started from the pipe O D
and propagated axially along the pipe
and radially through the wall thickness.
2. Cracks were intergranular.
3. Creep voids had opened in grain
boundaries.
4. No cracks or creep voids were
detected downstream of the boiler stop
valves.
5. Ductility tests showed the outer
two-thirds of the pipe wall to have low
ductility.
6. The pipe showed no evidence of
creep swelling.
7. Chemical analysis of the 316 stainless steel material indicated compliance
with the original requirements of the
specification.
Facts one, t w o and three taken together would indicate the classic case of
creep rupture of a pressurized cylinder
under internal pressure. In order to
accept this failure scenario, one would
have to conclude that: one, the material

was substandard, or t w o , the allowable

stress in the ASME Code was very optimistic.
Fact seven established that the material
was not substandard, but it should be
noted that cracking did correlate with the
weaker heats of material. Because of
facts four, five, and six, we became
convinced that we were not dealing with
the classic case of stress rupture.
If this material was merely weak in
creep, it would have swelled before it
ruptured. The stress rupture conclusion
also did not explain the fact that no
cracking or creep voids had been found
downstream of the boiler stop valves. It
was these inconsistencies which forced
us to abandon failure scenarios which
assumed bad material or incorrect allowable stresses.
The fact that the failed pipe had not
swelled and that a definite ductility gradient was present across the wall thickness
directly led to the following speculations:
1. Residual stresses combined with primary stresses to cause premature failure.
2. Since a residual stress field must be
self-equilibrating, the outer two-thirds of
wall thickness must have been tensile and
the inner third compressive.
3. The most likely source of this resid-

2.0

2,42

2.83

ual stress field is a severe thermal downshock.

4. Since the material downstream of
the stop valves appears undamaged, this
down-shock must occur when the boiler
stop valve is closed.
This failure theory was arrived at
because it appeared to be the only
hypothesis which was consistent with the
facts and observations determined during
the preliminary assessment phase of this
investigation. In order to test the likelihood of this theory being correct, analytical studies were carried out using
assumed thermal loadings. The objective
of these studies were twofold: one, to
numerically demonstrate that the failure
scenario was in agreement with established mathematical laws and theories;
and two, to assess the severity of thermal
shock required to predict failure due to
creep rupture in the known time frame.
Stress Analysis
A transient heat transfer and thermal
stress analysis was performed for a
forced cooldown of 500F (278C) in 5
minutes using the computer program,
CREPLACYL. CREPLACYL is a CE program
which calculates the plastic and creep
strains induced in thick wall cylinders due
to transient thermal and pressure load-

3.25

3.67

4.03

4.50

RADIUS(inches)
Fig. 17- Thermal gradient main steam pipe 500F in 5 minutes down-shock transient

WELDING RESEARCH SUPPLEMENT 1277-s

and the outside diameter is in tension.

The present analysis indicates bore stresses of - 3 0 ksi (-207 MPa) and O D
stresses of +20 ksi (138 MPa).
The next loading step was to reapply
the internal pressure and heat the pipe
slowly to 1140F (616C). The resulting
stress field is shown in Fig. 20. The solid
lines show the stress distribution at the
beginning of a 2000 hour hold period and
the dotted lines indicate the relaxation
which has taken place due to creep. It
should be noted that the average tensile
stress during the hold period is in excess
of the value which would cause stress
rupture in 100,000 hours.

CO

LU

cc
tto

Conclusion

Q.

oo
I

-20-F/g. 18 - Elastic-plastic hoop stress resulting from thermal gradient Fig. 17

ings. The program features temperature
dependent material properties, kinematic
hardening based on Prager's hardening
rule, and solutions for heat transfer, compatibility and equilibrium by finite difference methods. Material properties for
316 stainless steel were taken from the
Liquid Metal Fast Breeder Materials Handbook.
The operational cycle which was simulated in this analytical study was a 200F/
hr (111 C/hr) normal start-up followed
by a 2000 hour hold period at 1140DF
(616C). At the end of the hold period,
the pipe was cooled slowly to 1000F
(538C). At this time, the thermal shock
was imposed on the pipe.

stress field which is necessary to balance

the plastic moment that has been
induced in the cylinder. It should be
noted from Fig. 19 that at room temperature the inside bore is in compression

The results of the elastic/plastic/creep

analysis described above lead directly to
the following conclusions:
1. Repeated severe thermal downshocks could explain this failure.
2. The magnitude of this down-shock
was less than 500F (278C) but greater
than 150F (83C). The exact number is
indeterminate because of the temperature dependency of the material properties.
3. If this failure scenario is correct, the
failed pipe should have significant residual
stresses at its inside and outside diameters. Furthermore, the O D residual stresses should be tensile and the ID residual
stresses should be compressive.

to
LU
CC

Results
Figure 17 is a plot of the maximum
temperature gradient induced in the pipe
due to the assumed thermal shock loading. This gradient was approximately
390F (217C). Figure 18 shows the elastic-plastic peak hoop stress distribution
which results from this gradient. A stress
of +28 ksi (193 MPa) was calculated at
the inner bore and - 1 9 . 8 ksi (-136.5
MPa) at the outer diameter. This loading
condition resulted in a plastic strain field
being induced in the cylinder. We calculate 0.15% plastic-strain at the inner bore
and 0.08% at the outer diameter.
Upon removal of the thermal gradient,
the vessel was cooled to room temperature and the pressure removed. Figure 19
plots the room temperature residual

278-s | OCTOBER 1985

co
Q-

o
o
X

Fig. 19 Main steam pipe residual stress field 100"F, 14.7 psi

STRESS LEVEL TIME TO RUPTURE

300,000 hrs
STRESS LEVEL TIME TO RUPTURE
100,000 hrs

Residual Stress Measurements

The laboratory and field residual stress
measurements on the Eddystone TP-316
piping were made using a modified airabrasive hole drilling technique (Ref. 13).
In this technique, a special 45 degree
rectangular strain gage rosette is bonded
to the surface of the material to be
tested. A small hole, approximately 0.070
in. (1.8 mm) in diameter and depth, is
then drilled in the center of the rosette
pattern such that each of the three elements is equidistant from the edge of the
hole. The strain changes measured by
each element after stress redistribution in
the vicinity of the hole are used to
compute the magnitude and direction of
the maximum and minimum principal
residual stresses.
The hole is drilled by propelling 50jit
aluminum oxide particles through a
0.0625 in. (1.6 mm) ID tungsten carbide
nozzle using compressed gas or air as a
carrier. The particles impinge on the
material being measured and abrade a
sharply defined hole of proper diameter
and depth in approximately 8-9 minutes.
Air-abrasive drilling is a low energy,
essentially "stress free" method of introducing a hole and offers a distinct advantage over mechanical drilling methods. In
mechanical drilling, the stresses induced
from the machining operation can actually be greater than the residual stresses to
be measured, depending on a number of
factors, such as type and sharpness of
cutter, operator technique, and material
characteristics. These variables are eliminated with air-abrasive drilling. The overall measurement error using the modified
air-abrasive hole-drilling technique has
been shown to be less than 10 percent
(Ref. 13).

BEFORE CREEP

2000 HOURS CREEP

-Main steam pipe steady state conditions 1140CF, 4500 psi

For the Eddystone TP-316 piping, residual stress measurements were first made
in the laboratory on a cylindrical section
from the 4B main steam line. The results
of these measurements, shown in Table
6, indicated ID compressive and O D
tensile hoop stresses were present. From
these results, the decision was made to

conduct on-site measurements at various

O D locations in the plant. The results of
these field measurements are also shown
in Table 6.
The tabular results are arranged by
floor level beginning at the penthouse on
the eleventh floor and continuing downward to the fourth floor, first floor, and
the basement. In all but one instance, the
maximum (most positive) principal residual stress was nominally oriented in the
axial direction with the minimum principal
residual stress being nominally oriented in
the circumferential direction. The one
exception was in line 3B on the fourth
floor where the two principal stresses
were very nearly equal. The maximum
principal stress direction is shown in the
table by the angle, 9, which is the direction relative to element 1 of the strain
gage rosette. In all cases, element 1 was
aligned with the axis of the pipe or
header.

and 2B at the stop valves on the first

floor. These were areas of complex
geometry where spool pieces containing
thermocouple wells had been welded in
the lines.
In contrast to the axial residual stresses,
the circumferential residual stresses
formed a much different pattern. Measurements on t w o outlet steam headers
on the eleventh floor showed small hoop
stresses, one being compressive and the
other tensile. Measurements above the
stop valves on the fourth floor and at the
stop valves on the first floor showed
large tensile hoop stresses. Finally, a measurement well below the stop valves on
the 1A steam line at the basement level
(before the line enters a junction header)
showed a small compressive stress.

Relatively large tensile axial residual

stresses were measured at every location
except in outlet steam header 4B in the
eleventh floor penthouse. The largest
measured axial stresses were in lines 2A

The cause of this premature piping

failure was creep rupture. This failure
mode occurred because of residual
stresses induced in the piping by a
repeated thermal down-shock. It should

Summary

Table 6Laboratory and Field Residual Stress Measurement Results on Philadelphia Electric, Eddystone Unit 1 Headers and Steam Lines

Location

Component
Identification

Max Principal
Stress ksi

Min. Principal
Stress ksi

0 Direction
degrees*3'

Laboratory
Measurements

Pipe ID
Pipe O D

4B Steam Line
4B Steam Line

-8.1
3.1

-14.0
-3.6

166.5
95.2

On-Site Measurements in Unit 1

Floor
Floor
Floor
Floor
Floor
Floor
Basement

lA Outlet
4B Outlet
1B Steam
3B Steam
2A Steam
2B Steam
1A Steam

13.7
0.9
17.0
11.5
21.2
30.3
12.5

3.8
-6.6
13.0
8.8
13.6
20.2
-5.2

160.4
14.3
10.1
92.7
170.9
4.0
0.0

(a)

11
11
4
4
1
1

Header
Header
Line
Line
Line
Line
Line

Direction of maximum principal residual stress measured clockwise f r o m rosette gage element l. Element I is axially oriented in all cases.

WELDING RESEARCH SUPPLEMENT 1279-s

also be noted that the material with

sigmatization failed first.
The commercial impact of these findings was to replace all piping from the
outlet headers to the boiler stop valves.
The material downstream of the boiler
stop valves was judged to be acceptable
for service because there were no signs
of cracking, creep damage, or dangerous
tensile residual stress. It can also be concluded that the results of this study indicate that 316 stainless steel does not have
any inherent liabilities to high temperature operation. However, the specification for the replacement piping required
a positive nickel balance to minimize the
formation of sigma phase.

References
1. Caughey, R. H. and Benz, |r., W. G.
October 1960. Material selection and tabulation, main steam piping for Eddystone No. 1,
1200F and 5000 psi service. ASME, journal of
Engineering tor Power, pp. 293-314.

2. Wells, C H., Letter to ). F. DeLong dated

May 26, 1983. Failure Analysis Associates, Palo
Alto, California.
3. McMahon, C. I., |r. 1983. Private communication, University of Pennsylvania, Philadelphia, Pennsylvania.
4. Sumitomo Metal, Ltd. lune 1983. The
investigation of leaked TP316 stainless steel
main steam pipe of Eddystone No. 1 unit.
Tokyo, lapan.
5. W o o d f o r d , D. A., Letter to |. F. DeLong
dated April 15, 1983. General Electric, Corporate Research & Development, Schenectady,
New York.
6. Lai, ). A. 1983. Private communication,
Central Electricity Research Laboratory, Leatherhead, Surrey, U.K.
7. Gray, R. J. 1983. Private communication,
Oak Ridge National Laboratory, Oak Ridge,
Tennessee.
8. Der, T. )., Lonsdale, P., Stevens, R. A.,
Chestell, D. I., and Flewitt, P. E. |. October
1981. Methods of sampling service components for metallurgical evaluation of remanent
creep life. CEGB Report SE/SSD/RN/81/067.
9. Davidson, M. |. May 1983. Header/
mainsteam pipe evaluation, Philadelphia Electric, Eddystone Station. Engineering Develop-

ment and Code Coordination, Department

Report No. EDD-83-12, Combustion Engineering, Windsor, Connecticut.
10. Bynum, I. E., Letter to W . F. Siddall
dated lune 9, 1983. MML, Combustion Engineering, Chattanooga, Tennessee.
11. Yoshihara, K., Kurahashi, M and Nii, K.
July 1980. The segretation behavior of sulfur,
oxygen, and phosphorus at the (100) surface
of iron single crystal, Trans, lapan Institute of
Metals, 21(7): 424-432. Tokyo, lapan.
12. Ellis, F. V. and Bynum, ). E. 1983. Evaluation of TP316 stainless steel main steam line
after 125,000 hours of service. Advances in
Life Prediction Methods, pp. 337-345. ASME,
New York, New York.
13. Bynum, |. E. lanuary 1981. Modifications
to the hole-drilling technique of measuring
residual stresses for improved accuracy and
reproducibility,"
Experimental
Mechanics,
21(1): 21-33.
14. McGough, T. C , Pigford, |. V., Lafferty,
P. A., Zielke, H. E., Ward, C. T., Tomasevich,
S., and DeLong, |. F. 1984. Material selection &
fabrication of the replacement main steam
piping for the Eddystone No. 1 supercritical
pressure unit. Welding journal 64(1): 29-36.

WRC Bulletin 306

July 1985
PVRC Flanged Joint User Experience Survey
By J. R. Payne
Over 180 completed questionnaires were received f r o m a distribution list of 295 flanged joint users.
The results of the survey provided guidance in establishing research needs and priorities for flanged joint
improvement programs.
The survey and the publication of the report were sponsored by the Task Group on Gasket Testing of
the S u b c o m m i t t e e on Bolted Flanges of t h e Pressure Vessel Research C o m m i t t e e .
The price of WRC Bulletin 306 is $14.00 per copy, plus $5.00 for postage and handling. Orders should
be sent with payment to the Welding Research Council, Rm. 1 3 0 1 , 345 E. 4 7 t h St., New York, NY
10017.

Call for Railroad Welding Papers

The AWS Railroad Welding C o m m i t t e e has announced t h a t the Third Railroad Welding Conference will
be held at the Hilton Hotel, Chicago, III., on September 25-26, 1986.
Papers are being solicited f r o m the railroad c o m m u n i t y on the general topics of "Failure Analysis and
P r e v e n t i o n " and " W h a t ' s New in Railroad Welding."
Abstracts, limited to 500 words, should be submitted no later than January 1, 1986, to: Secretary,
D15 Railroad Welding C o m m i t t e e , American Welding Society, 550 N. W. LeJeune Rd., P. 0 . Box 3 5 1 0 4 0 ,
Miami, FL 3 3 1 3 5 .
Authors will be notified concerning acceptance of their papers no later than February 15, 1986.
Deadline for submission of the final manuscript for pre-meeting publication is June 1, 1986.

280-s | OCTOBER 1985