You are on page 1of 20

Int. J. Pres. lies.

& Piping 25 (1986) 393-412

IHSI Application to the Welded Joint with Small


Cracks

A. Okamoto, H. Wada and T. Umemoto


Nuclear Power Division, Ishikawajima-Harima Heavy Industries Co. Ltd,
1 Shin-Nakahara-Cho, Isogo-Ku, Yokohama 235, Japan

A B S T R A CT

Induction Heating Stress Improvement (IHSI) has been proved as


a strong countermeasure .for intergranular stress corrosion cracking
(IGSCC). I H S I is also expected to suppress further crack extension o f
existing cracks. The applicability of l H S I to a flawed pipe was investigated
using a thermo elasto-plastic F E M program. IGSCCs with depths up to
50% of the pipe thickness are considered.
It is found that there is a large margin for the extension of the crack
during IHSI. Effectiveness of I H S I as an eternal countermeasure for
IGSCC depends upon the level of service stress as well as the crack size.
A long circumferential crack will be prevented from further extension by
I H S I if the depth is under 30% of the thickness and the service stress is
under O"75 S m.
For practical use, a simplified procedure to predict I H S I effectiveness
for preflawed pipe is proposed making use of calculation of K b), the
superposition principle.

INTRODUCTION

Induction Heating Stress Improvement (IHSI) has become known


worldwide as one of the most effective countermeasures for intergranular
stress corrosion cracking (IGSCC). There has been extensive experimen-
tal and operational experience of the strong effectiveness of IHSI against
393
Int. J. Pres. Ves. & Piping 0308-0161/86/$03"50 © Elsevier Applied Science Publishers
Ltd, England, 1986. Printed in Great Britain
394 A. Okamoto, H. Wada, T. Umemoto

IGSCC. Some experience of applying IHSI to preflawed pipe has


revealed that this stress-related countermeasure was also effective
in these circumstances and could prevent further stress corrosion
cracking. 1,2
Up to the present, it has been set forth as a premise that, before
applying IHSI, one should make sure that there is no cracking in the
vicinity of the weldment by means of non-destructive examinations. But
it seems from the above experience that this restriction can be waived
to some extent if the cracking is small enough.
The objective of this paper is to investigate the limit of the crack size
for effective IHSI. It is also intended to study the possibility of using
IHSI as a good temporary measure until permanent countermeasures
for IGSCC, such as the replacement of the degraded pipe by a new one,
can be implemented.

A P P L I C A T I O N OF IHSI TO F L A W E D PIPES

The process of IHSI is detailed in many references. 3'4 Here, we will


only briefly describe the concept of IHSI so far as is necessary for the
discussion of the problem being considered.
A large temperature difference can be introduced by heating a pipe
from the outside with an induction coil, while cooling water is supplied
to the inner surface (ID) of the pipe simultaneously, as shown in
Fig. 1.
When a linear temperature gradient is attained across the pipe wall,
the thermal stress induced can be expressed approximately as follows:
ath = E c t A T / 2 ( 1 - v) (1)
WELDMENT INDUCTION CQIL

,~,
\"~_ ",/ ~._, ,//PIPE

Fig. 1. Outline of heating and cooling.


I H S I application to the welded joint with small cracks 395

where
tTth ----thermal stress in axial and circumferential directions,
--linear thermal expansion coefficient,
AT = temperature difference between inner and outer surface,
E -- Young's modulus,
v = Poisson's ratio.
If the thermal strain is sufficiently high and the thermal stress on ID
exceeds the tensile yield strength, the subsequent removal of the
temperature difference AT induces a compressive residual stress as the
thermal strain comes back to the original point (Fig. 2). On the outer
pipe surface (OD), the compressive thermal stress by AT exceeds the
compressive yield strength and as a result a tensile residual stress is
induced. The effect of IHSI is controlled by certain factors called
'essential variables' which have already been discussed and established 4
for unflawed pipe components.
Now, when we want to discuss IHSI application to a preflawed pipe,
it is necessary to answer the following questions:
(1) Does the inside surface crack remain as it was throughout the
duration of IHSI treatment and not extend under the tensile
stress induced by the heating process?

THERMALSTRAIN
' i12,, ',T

1__

Fig. 2. Stress-strain diagram.


SPECIFIC HEAT, CO (xlO-b kcallnm~J°C)

THERMAL CONDUCTIVITY, ,(x10-3 kcallmmh°C)

o s ~ 8 8

Fn

--4

r--

u~
u~

CD 8 8 g 8 g
FREE THERMAL EXPI~NSION,
Y(xIO-6/°C)

ELASTIC MODULUS, E (xlO 4 kgflmm 2)

YIELD STRENGTH, ;y (kgf/rrlll


2)

/ A' / // ~'% ~

/-, L \ I' /
/5 /~ \ / /

•' / ~ / / ~ ' A //
/ //# / ~,.'/
m

1111/'~
, ." # Z J ' \
%

" i
8

PO[SSON'S RATfO,
I H S I application to the welded joint with small cracks 397

(2) What is the profile and how is the magnitude of the residual
stress related to IHSI in the flawed section?
(3) As for the stress in the operating condition, it is in compression
along the inside surface of the pipe, including at the crack tip?
Or, is the stress low enough to prevent IGSCC?
The effect is expected to depend on crack size. So the investigation
should be done taking the crack depth as a parameter.
In this study, the analytical examination is made assuming a pipe in
a BWR primary loop which has long circumferential IGSCC on the
inside surface. The length is considered to be long enough to idealize as
a complete circumferential crack. This assumption holds true in some
cases of IGSCC.

CALCULATION P R O C E D U R E

The following Ishikawajima-Harima Heavy Industries domestic FEM


computer codes for axisymmetrical structures were used for the calcu-
lation: ITEMP and IEPTC. ITEMP is a code for thermal conduction
analyses. Temperature dependent material constants and thermal boun-
dary conditions can be given in the user subroutines that enable flexible
use of this code. IEPTC deals with elasto-plastic stress analyses.
Incremental theory with isotropic stress hardening rule is used in the
program. Material constants including the stress-strain relationship are
easily specified in the user subroutines.
Material constants used in the analyses are shown in Fig. 3. These
values are obtained from Ref. 5. The same material constants are used
for the weld material and the heat affected zone.
First, a full model with coarse meshes was analyzed in order to find
how to make an adequate zooming model with fine meshes around the
cracked portion.
The finite element idealization with the boundary conditions for
thermal and stress analyses are shown in Fig. 4(a). The depth of the
crack was assumed to be 30% of the pipe thickness (t). The heating
density and the duration were chosen by trials to give an equivalent
temperature history to be observed in the usual IHSI works. The
calculated temperature histories at some selected locations are shown
in Fig. 5.
Referring to the above analysis, the zooming model was formed
taking a pipe length of t ( = 30 ram), as shown in Fig. 4(b).
398 A. Okamoto, H. Wada, T. Umemoto

-I
350.0

(b) ~ 3o.o If) __

IIII I ,

iiii i i i m
IIIII I I
IIIII I
ill i l
Ih,,, !
II]lI I
[111] I
IIIII I
OD
Fig. 4. Finite element model (30% t crack): (a) coarse mesh; (b) fine mesh.

i LOCATION

I. 3= 5 OD

c~

i
~0 5 i0 15 20
TIME (MIN)
Fig. 5. Temperature history.
I H S I application to the welded joint with small cracks 399

i00 T I "r "I T -


O HOOPSTRESS
X AXIALSTRESS
5O
CRACK
(30%t)

~~FINE MESH

-50

- i00
5 10 15 20 25 30

DS
I TANCEINTHC
I KNESSDIRECTION(mm)
Fig. 6. Comparisonof calculated residual stresses.

The results of this verification study are shown in Fig. 6. In this figure,
calculated residual stress distributions through the cross section at the
cracked location for the two models are compared with each other.
Note that the numerical errors are found in the stress distribution on
the cracked part owing to the strong singularity at the crack tip. There
would not actually be oscillation of the stress at that part. Nevertheless
it can be concluded that both results are identical, and a more precise
residual stress distribution must have been obtained by the zooming
model.
Second, a sensitivity analysis was performed with respect to the crack
depth in the fine mesh model. Crack depths of 10%, 30% and 50%
thickness (t) were assumed. The model with no crack was also considered
for comparison. The same thermal cycle according to standard IHSI
was loaded to each model. Then axial tensile loads simulating a service
load were applied up to S m at design temperature, which is the limit of
primary membrane stresses.

CRACK OPENINGS D U R I N G IHSI

The calculated crack openings are illustrated in Fig. 7. As expected,


cracks opened most widely at the maximum temperature.
400 A. Okamoto, H. Wada, T. Umemoto

0,4 •"-" MAX TEMPERATURE


o-o AFTER IHSI

alt% i12 COD


10% 0,024
30% 0.069
0,3 50% 0.085

50%t
w

..J
c/)
0,2

30%t
c~_
o

0,1
50%t

lO%t

0 5 10 15
DISTANCE IN THICKNESS DIRECTION (mm)

Fig. 7. Crack opening shapes and crack tip opening displacements.

The crack tip opening displacements 6 were obtained graphically.


They were converted to J values according to the following approximate
equation:
J = 2 ,~oo (2)
where a o denotes flow stress at the crack tip temperature, determined
as the average of the tensile strength a u and the yield point ay. Material
data obtained experimentally are used here. The J values were obtained
as 3.6, 10.5 and 12.2kgfmm +1 for 10%, 30% and 50% t cracks
respectively. An empirical study 6 showed that Jic for Type 304 austentic
stainless steel was around 100 k g f m m - 1 . Therefore it can be concluded
that crack initiation is not likely to occur in the process of IHSI.
In Fig. 7, crack opening contours at the end of IHSI are also shown.
Note that the 10% t crack is closed and actually both the surfaces come
IHSI application to the welded joint with small cracks 401

into contact with each other. But the effect on computational error
seems to be small because the displacement is not so significant.
Cracks more than 10% t remain open after IHSI. This implies that
the calculated residual stress does not include the error due to the
assumed boundary condition at the crack side.

R E S I D U A L STRESS D I S T R I B U T I O N A F T E R IHSI

The residual stress distributions are shown in Fig. 8(a)-(d) including


the uncracked model in Fig. 8(a) which is for reference. Stress concentra-
tions at the crack tip caused deep residual stress near the tip. Triaxial
tensile stress state brought the stresses higher than the uniaxial yield
point. It can be seen that the residual stress becomes deeper as the crack
depth increased up to about 30% t. Then the trend is reversed due to
the reduction of nominal thermal stress by IHSI.
This compressive residual stress at the crack tip does not imply
sufficient resistance against IGSCC by itself. But it still shows that the
application of IHSI improves the stresses so as to mitigate the cause of
IGSCC. For a semi-elliptical artificial flaw on the butt weld of 4in

i00 i

(o) o HOOPSTRESS
X AXIAL STRESS

N 50

(NO CRACK)

(: -50

-ioo a i a i J

0 5 10 15 20 25 30

DISTANCE IN THICKNE~SSDIRECTION (mm)

Fig. 8. Residual stress distribution after IHSI: (a) no crack; (b) 10% t crack; (c) 30%
t crack; (d) 50% t crack.
402 A. Okamoto, H. Wada, T. Umemoto

100 I I i I I I
0 HOOPSTRESS
(b)
X AXIAL STRESS

50

o:
-50

- i00 I I I I I
5 10 15 20 25 30

DISTANCE IN THICKNESS DIRECTION (mm)

I00,

(C) 0 HOOPSTRESS
X AXIAL STRESS

50

~
CRACK (30%t) ~I
X~

-50

I I I I
-100
0 10 15 20 25 30
DISTANCE IN THICKNESSDIRECTION (r~n)

Fig. 8.--contd.
IHSI application to the welded joint with small cracks 403

100 i

(cl) 0 HOOPSTRESS
x AXIAL STRESS

GRACK (~0%t) =i x------

0 X

-50

I I I I I
-i00
5 i0 15 20 25 30
DISTANCE IN THICKNESSDIRECTION (rnm)

Fig. 8--contd.

sch. 80 pipe, the effectiveness was demonstrated by an experiment. 7 It


is evident that IHSI will delay further IGSCC.

STRESSES IN T H E O P E R A T I N G C O N D I T I O N

To ensure that the stress at the crack tip would remain in compression
under operating conditions, axial tensile load was applied in succession
to the IHSI sequence. The temperature o f the pipe was assumed to rise
to 300°C as the load increased to 8.0 x 105 kgf, producing nominal
membrane stress o f S m (12.3 k g f m m - 2 ) . For reference, the service load
was also applied to the same model but subjected to no IHSI process.
The stress distributions through the cross section at the cracked
location are shown in Fig. 9(a)-(c) and Fig. 10(a)-(c) for axial and
hoop stresses respectively. The axial stresses just at the crack tip are
summarized in Fig. 11 for crack depth and nominal membrane stress
as parameters. It is seen that the crack tip stress does not become a
tensile stress if the crack depth is smaller than 30% t and the nominal
m e m b r a n e stress is under 0.75S m. For the nominal stress o f S m, IHSI
cannot keep the crack tip stress in compression. But the level of the
stress is well below the other case in which no IHSI was applied.
404 A. Okamoto, H. Wada, T. Umemoto

100 I T T

(a) x NO IHSI + Sm
@ IHSI + Sm
0 IHSI + 0,75Sm
• IHSI + 0,5Sm
50 A AS IHSI

ClOXt)

-50

-100
5 10 15 20 25 30
DISTANCE IN THICKNESS DIRECTION (nTn)

100 ~ 7 T
(b) X NO IHSI + Sm
@ IHSI + Sm
0 IHSI + 0,75S m
• ISHI + 0,5S m
50 CRACK (30%t) .. A AS IHSI

-50

-i00
5 i0 15 20 25 30

DISTANCE IN THICKNESS DIRECTION (mm)


Fig. 9. Change in axial stress with tensile load: (a) 10% t crack; (b) 30% t crack; (c)
50% t crack.
IHSI application to the welded joint with small cracks 405

100 r r T
' x NO IHS~ + Sm
(c) • IHSI + Sm
RACK (50 ) 0 IHSI + 0,75Sm
I • IHSI + 0.SSm
50 A AS IHSI

-50

-100
5 10 15 20 25 30
DISTANCE IN THICKNESS DIRECTION (mm)

Fig. 9--contd.

100
x NO IHSI + Sm
(o)
• IHSI + Sm
0 IHSI + 0.75Sm
• IHSI + 0.5Sm
50 - A AS IHSI

-50

L I I
-i00 5 ~0 115 20 25 30

DISTANCE IN THICKNESS DIRECTION (mm)


Fig. !0. Change in hoop stress with tensile load: (a) 10% t crack; (b) 30% t crack; (c)
50% t crack.
406 A. Okamoto, H. Wada, T. Umemoto

100
X NO IHSI + Sm
(b)
• IHSI + Sm
0 IHSI + 0.75Sm
• IHSI + 0.5Sm
50 A AS IHSI

x x

-50

- i00
5 10 15 20 25 30
DISTANCE IN THICKNESS DIRECTION (mm)

100 r r l m 7

(c) X NO IHSI + Sm
• IHSI + Sm
CRACK (50%t) -H 0 IHSI + 0.75Sm
- - - - i • IHSI + 0,5Sm
5O I A AS IHSI


-50

.L L L L
-i00
0 5 10 15 20 25 30
DISTANCE IN THICKNESS DIRECTION (ram)

Fig. lO---contd.
I H S I application to the welded joint with small cracks 407

loo

50

e~

0 J

-50
AS IHSI

-i00 i i
0% 10% 30% 50%
CRACK DEPTH
Fig. !1. Axial stress at the crack tip.

S I M P L I F I E D P R O C E D U R E OF T H E E V A L U A T I O N

Nakamura et al. 7 showed that analysis using the superposition principle


with calculation of K by weight functions gave a good prediction of
fatigue crack extension under residual stress field. The same procedure
applied to the problem being considered here.
The superposition principle can sometimes be utilized to derive stress
intensity factors. Figure 12 shows an explanation of this principle from
a textbook on fracture mechanics. 8 By this theory, we can calculate the
stress intensity factor if we know the stress distribution in the structure
without the crack.
The K factor due to IHSI in a preflawed pipe was obtained using the
calculated residual stress for the uncracked pipe which is shown in Fig.
8(a). The residual stress distribution through the thickness was fitted by
a third-order polynomial expression as follows:
tr(IHSI) = - 2 4 . 8 5 - 1.971X+O.3519X2-O.OO7701X 3 (3)
where X is the distance from the pipe inside surface.
o

J
3

K (kgflmm2)
@- li
c~ o-
t i

./. ~ ./ i, ~ 'q
p~ /J /p t
e~
J P~
J
/
/ 0

Ii
/
i

\
\ 0
\
a ---t
.n.

~. -
I I i L I I I
+
IHSI application to the welded joint with small cracks 409

The calculation o f K factors for various crack sizes was performed


making use of the influence function derived by Harris and Lim 9 for
complete circumferential interior surface cracks. The influence functions
are given for the point load on the circumferential crack so that non-
linear stress distribution can be dealt with.
The calculated K for the IHSI residual stress has a negative value and
is fictitious. But as the residual stress is produced through reversed
thermal bending stress, the K acts as an index for the residual stress
distribution.
The superposition principle was also assumed in order to obtain K
at the operating conditions. K for uniform tensile load was calculated
in the same way and added to K of the residual stress. The calculation
results are shown in Fig. 13. If the combined K is negative, it means
that the crack tip stress is also negative in the operating condition and
vice versa. Thus the crack value at the cross point of the K curve and
the abscissa is the limit of the effective IHSI.
The curves (b) in the figure denote K values for the flawed pipes
subjected to tension without IHSI. Before combining these values with
K due to IHSI, i.e. curves (a), the following modification was made to
obtain curves (c):
K(c) = K(b) × 2/(1 - a/t) (4)
where a/t means the fraction of the cracked part to the pipe cross
section. This modification is necessary for the above case because the
yielding in tension along the outer surface contributes to the overall
stress level increase near the crack tip. The effect is evident when we
look at Figs 9(a)-(c).
The dotted lines (a) + (c) show the combined K value for the service
loading after IHSI. The crack depth limit of the effective IHSI can be
obtained as 5% t for S m, 30% t for 0.75S m and 40% t for 0.5S m. These
limits are a little smaller than those obtained in the foregoing F E M
analyses, but almost agree with them. So it can be concluded that this
simplified method works well.

E F F E C T I V E IHSI F O R E L L I P T I C A L S U R F A C E C R A C K

Using the above simplified method, a prediction was made for elliptical
surface cracks to determine the limit crack size o f effective IHSI. The
410 A. Okamoto, H. Wada, T. Umemoto

ID ' ' ' 0D


250 ~ (
200 b)TENSO
I I(Sm)

i00 / ~ i/~) +(b)


50/ / /
,/ f
/
F
0 '~ I,~ / I ~'~ I I

~C
\\ /
-5o \ \ A/
\\ /
\\ \ \ /.
-i00 \\ \ \ / / "x~(a)IHS

"~ a:c=113
0'2 0'4 o'6 o'8 1.o
o/T
F i g . 14. Prediction of effective IHSI for elliptical surface crack.

same IHSI residual stress distribution expressed by the eqn (3) was
assumed. The influence functions developed by Shiratori e t al. ~° are
used to calculate K values for the elliptical surface cracks. Modification
was made by the following equation which was derived from the same
point of view as in the previous case:
K(modified) -- K x 2/(1 - A) (5)
where A is again the fraction of the cracked part to the pipe cross
section.
The results for elliptical cracks with a : c - - 1:3 are shown in Fig. 14.
F r o m this figure, we can conclude that IHSI to this type of crack will
IHSI application to the welded joint with small cracks 411

change the crack tip stress into compression in the operational condition,
so long as the crack depth is not greater than 25% t or so.
It is worth noting that, for the combined K in Fig. 14, the value at
the deepest point A of the crack goes positive faster than that at the
surface point B. This implies that the crack extension, if any, will tend
to grow into the thickness direction rather than along the surface. This
is desirable from the standpoint of the leak-before-break design concept.

CONCLUSIONS

The following conclusions are obtained from the analytical results.


1. There is a large margin for the initiation of cracking from a
pre-existing flaw during IHSI so long as the crack depth is not
greater than 50% of the thickness.
2. The residual stress after IHSI can be in compression at the
crack tip which mitigates sensitivity to IGSCC.
3. Effectiveness of IHSI as a permanent countermeasure for
IGSCC depends upon the level of service stress in the operating
condition as well as on the crack depth and its configuration.
A long circumferential crack will be prevented from further
extension by IHSI if the depth is under 30% of the thickness
and the service stress is under 0.75S m, which seems to be likely
in normal circumstances.
4. A simplified procedure to predict IHSI effectiveness for pre-
flawed pipe was proposed, making use of calculation of K by
the superposition principle.

REFERENCES

I. Ando, Y., Yagawa, G. et al., Theoretical and experimental study on


Induction Heating Stress Improvement (IHSI) of nuclear primary piping,
Trans. 5th S M i R T Conf., Berlin, 1978, Paper F2/8.
2. Umemoto, T., Yoshida, K. and Okamoto, A., Residual stress improvement
for pipe weld by means of induction heating pre-flawed pipe. 4th Int. Conf.
Pressure Vessel Technology, Paper C24/80, London, May 1980.
3. Umemoto, T. and Tanaka, S., Residual stress improvement by means of
induction heating, I H I Engng Rev., 11(4) 0978).
4. Amano, M., Nojima, K. and Umemoto, T., Theoretical and practical
412 A. Okamoto, H. Wada, T. Umeraoto

aspects of Induction Heating Stress Improvement, 6th S M i R T Conf., Post-


Conf. Seminar 3, Paris, August 1981.
5. EPICC data book of material properties, Japanese Society of Mechanical
Engineers, August 1974 (in Japanese).
6. Survey report on reliability proof tests on welding heat affected zone.
Appendix IlL Fracture toughness testing, Nuclear Power Engineering Test
Center, 1984 (in Japanese).
7. Nakamura, H., Matsushima, E., Okamoto, A. and Umemoto, T., Fatigue
crack growth under residual stress field in low-carbon steel, 8th SMiRT
Conf., Brussels, 19-23 August 1985, Paper G5/2.
8. Broek, D., Elementary Engineering Fracture Mechanics, 3rd revised edition,
Martinus Nijhoff, The Hague, 1982, p. 77.
9. Harris, D. O. and Lim, E. Y., Stress-intensity factors for complete
circumferential interior surface cracks in hollow cylinders, Fracture Mechan-
ics: 13th Conf., R. Roberts (ed.), ASTM STP 743, American Society for
Testing and Materials, New York, 1981, pp. 375-86.
10. Shiratori, M., Miyoshi, T. and Tanikawa, K., Analysis of stress intensity
factors for surface cracks subject to arbitrarily distributed surface stresses,
to be published in Proc. JSME, 1985 (in Japanese).

You might also like