You are on page 1of 10

Int. J. Pres. Ves.

& Piping 20 (1985) 55-64

Crack Instability in a Multiple-cracked System

E. Smith

Joint Manchester University/UMIST


Department of Metallurgy and Materials Science,
Grosvenor Street, Manchester MI 7HS, Great Britain

(Received: 31 October, 1984)

ABSTRACT

The paper analyses the behaviour of a solid containing two symmetrically


situated deep cracks, and with tension of the small remaining ligament.
The results are used to support the view that when a piping system
fabricated from a ductile material, such as 304 stainless steel, contains
more than one crack, then the presence of other cracks has an insignificant
effect on the condition for instability of growth of a given crack.

INTRODUCTION

Against the background of the technological problem of the cracking of


304 stainless steel Boiling Water Reactor piping, the theoretical analysis
of Tada et al.l showed that circumferential growth of a through-wall
crack should be stable for the pipe run lengths that are typical of a reactor
coolant system. Their approach, based on the tearing modulus procedure
developed by Paris and co-workers, 2 assumed the material to be
elastic-perfectly plastic, with the plastic deformation being confined to
the cracked cross-section which is fully yielded. This cross-section thus
behaves like a plastic hinge with a region of tension immediately adjacent
to the crack tips and above the neutral axis, and a region of compression
below the neutral axis. It has now become apparent that if a crack forms,
by a stress corrosion mechanism, in one part of a piping system, then
there is a real possibility that there will be similar cracks in other parts of
55
Int. J. Pres. Ves. & Piping 0308-0161/85/$03.30 ~ Elsevier Applied Science Publishers
Ltd, England, 1985. Printed in Great Britain
56 E. Smith

the system. With this situation in mind, an earlier analysis by the author 3
has extended the approach o f T a d a et al. 1 and shown that, when there are
several circumferential cracks together with their associated plastic hinges
in a stainless steel piping system, then the tendency towards instability of
circumferential growth of any given crack is no greater than for the case
where the given crack exists in isolation. Instability was considered from
the basis that though there is a moment reduction at a cross-section
containing a non-growing crack, as another crack grows under a fixed
displacement applied to the system, the plastic rotation recovery at a non-
growing crack cross-section can be ignored in comparison with the elastic
recovery of the system.
This paper focuses on this particular point, and uses the results from
an analysis of the behaviour of a solid containing two symmetrically
situated deep cracks and with tension of the small remaining ligament, to
show that if the tension is due to a load applied at a point away from the
ligament and there is a small load reduction AP below the general yield
load, then the elastic displacement recovery is proportional to AP, and the
plastic displacement recovery is proportional to (AP) 3. It is thereby
demonstrated, for a material with a high resistance to crack growth such
as 304 stainless steel, that, when there are several such cracks and their
associated fully plastic ligaments, the tendency towards fracture
instability of a given ligament is essentially unaffected by the presence of
the other cracks. This result is therefore supportive, and indeed important
back-up, to the conclusion reached in Ref. 3.

T H E O R E T I C A L ANALYSIS

Figure 1 shows the model representing the plane strain deformation of a


solid containing two symmetrically situated deep cracks, and with tension
of the small remaining ligament which is fully yielded; the ligament width
and thickness are respectively 2L and B, while the solid width is 2h. The
displacement is A at a great distance D/2 (i.e. D >>L) from the ligament,
while P is the magnitude of the load associated with this displacement.
This model simulates the behaviour of a growing crack, but, in order to
assess the effect of a non-growing crack on this growth behaviour, it is
supposed that the system is connected with another similar system
containing a non-growing crack. In other words the overall system is
subjected to a total displacement Av and the displacement of the second
Crack instability in a multiple-cracked system 57

P,A

Y Y

Y Y

< 2L-- :~

< 2h-- ~,-

Fig. 1. The plane strain deformation of a solid containing two symmetrically situated
deep cracks. A displacement A is applied to the solid at a great distance from the ligament,
while P is the corresponding load.

system is A, with AT -- A, + A. The stability of the system will now be


assessed for a fixed value of the total displacement AT (see Fig. 2).
The analysis is based on the Dugdale-Bilby-Cottrell-Swinden (DBCS)
representation of plastic deformation, 4'5 whereby yield is confined
entirely to the planes containing the cracks. (In this respect the model is
analogous to that used by Tada et al. 1 and also to that used by the author
in Ref. 3, where plastic deformation is confined to the cracked cross-
section of a pipe.) The tensile stress within the plastic zones between the
crack tips is assumed to have the value Y which is representative of the
material's tensile yield stress. For the particular configuration in Fig. l,
the magnitude of the J-integral is given by the relation (see Appendix eqns
(A. 7) and (A. 8)):
J = JEL + JPL + JGY
4(1 -- v2)y2L 4(1 -- v2)y2L (2In
= 1rE -~ rtE 2 - 1) + YAov (1)

where E is Young's modulus, and v is Poisson's ratio. The first term is the
elastic component JEL due to the application of the limit load P = 2BLY,
presuming that there is no plasticity; the second term is the additional
58 E. Smith

P, A T

D Non-growing
cracks

D Growing
cracks

L J
I~ 2h "-1
Fig. 2. A composite system consisting of a section containing growing cracks, and
another section with non-growing cracks. The ligaments in both sections are assumed to
be fully yielded.

(plastic) contribution JPL when the general yield state is just attained;
while the third term J~v is due to the post-general yield plastic
deformation; A~v is the post-general yield plastic contribution to the
displacement A. The total load point displacement AT = A, + A, assumed
to remain fixed during crack growth, is given by the expression (see
Appendix eqns (A. 15) and (A. 16)):
A = A, + AEL,NCR+ AEL,Ca + ApL + AGy
(1-vg)LDY 8(1-_v2)LYln(2h ~
= A, 4 hE ¢" nE \nL/
8(1 -- vz)YL
-~ (1 - ln2) + A~v (2)
nE
where the first term A, is the displacement associated with the section
containing the non-growing cracks; the second term AEL,NCRis the elastic
contribution to A assuming that there is no crack and no plastic
deformation; the third term AEL,CR is the elastic contribution to A due to
the crack's presence again assuming that there is no plastic deformation;
the fourth term ApL is the additional (plastic) contribution to A when the
Crack instability in a multiple-cracked system 59

general yield state is just attained; while the fifth term AGv has already
been defined.
The original Paris-Tada-Zahoor=-Ernst approach 2 is based on the
assumption that the J-crack growth resistance curve is specific for a given
material, with the tearing modulus TMaT being related to the slope of the
J-crack growth curve:
E dJ
TMAT -- y2 dc (3)

where 6c is an increment of crack length. Instability under a displacement


control condition (i.e. d AT/dC = 0) is presumed to occur when TApP >
TMAT, TApp being related to J by an expression similar to eqn (3), and J
being given by eqn (1). It follows from eqns (1), (2) and (3), noting that
fc = - ilL, that the condition for crack growth instability is
E/dA, dAELNCR dAELCR dApLX E //dJEL dA~_~
d; di +
(4)
after elimination of AGv. Of the various terms on the left-hand side of
eqn (4), eqns (1) and (2) show that with the possible exception of the terms
involving A. and AEL,NCa,i.e. the displacement associated with the non-
growing crack section and the elastic displacement in the crack's absence
for the growing crack section, all the terms will be ~ 10 if realistic values
are assumed for h and L. Consequently if the material has a high crack
growth resistance, and 200 has been suggested1 as being an appropriate
value for the tearing modulus TMAT for 304 stainless steel, then it is only
the terms involving A, and AEL,NCRthat need be considered, and the
instability condition given by eqn (4) simplifies, by use of eqn (2), to

Y \ dL + _ - Y dL + h > TMA x (5)

Provided that the tearing modulus of the material exceeds ,-~ 100, neglect of
the other terms is justified and the ensuing instability condition (eqn (5))
should be accurate. Even if the tearing modulus is as low as 50, neglect of
the other terms should still give a usable instability criterion.
The next stage in the analysis involves a consideration of the term
involving A, in the instability criterion of eqn (5) i.e. the term concerned
with the displacement associated with the non-growing crack section.
60 E. Smith

Now as the cracks grow in the growing crack section, there will be a
reduction in the load P. This produces a reduction in both the elastic and
plastic contribution to A,. If the load reduction is AP, the reduction in the
elastic contribution can be obtained by use ofeqn (A. 16) in the Appendix.
The reduction is
(1-v2)DAP 8 ( 1 - v 2 ) A P , {2h~
(6)
2BhE + ~ffE mt~£ )
the first term being due to elastic behaviour with no crack present, and the
second term due to elastic behaviour associated with the crack's presence.
As regards the reduction in the plastic contribution to A,, this stems from
the formation of a compressive zone (stress-Y) in the immediate vicinity
of the crack tips. The magnitude of this reduction is readily obtained from
eqn (A. 14) in the Appendix, with Y replaced by 2Y and 2 now being equal
to AP/4BLY. Remembering that AP is small, this reduction is equal to
8(1-v:)YL 1( AP ~3
)E 3 \4BLYf (7)

Because the reduction in the elastic contribution is proportional to AP


(see expression (6)), while the reduction in the plastic contribution is
proportional to AP 3 (see expression (7)), the latter can be ignored in the
limit as AP-~ 0. It immediately follows from expression (6) that the first
term on the left-hand side of eqn (5) becomes
E dA, (1 - v2)D 8(1 - v2) In ( 2 h ' ]
Y dL - h + (8)

Using the same arguments as those used earlier for the growing crack, one
can ignore the second term on the right-hand side of eqn (8), i.e. that
associated with the elastic contribution due to the crack's presence. It then
follows from eqns (5) and (8) that the crack instability condition becomes
2(1 - v2)D
> TMA T (9)
h
which is the same instability condition as for the case where the non-
growing crack does not exist. In other words the presence of the non-
growing crack has no effect on the instability criterion for the first crack,
and the instability criterion may therefore be determined from the basis
that the second crack does not exist in the system.
Crack instability in a multiple-cracked system 61

DISCUSSION

The preceding section's analysis has clearly shown that provided a


material's crack growth resistance is sufficiently great, and this should be
the case with 304 stainless steel, then the tendency towards failure
instability of a given ligament, when there are two symmetrically situated
deep cracks, should not be affected by the presence of other cracks and
their associated plastic ligaments. (By implication, the same conclusion
holds when the other ligaments are not fully yielded.) In the context of the
behaviour of circumferential through-wall cracks in 304 stainless steel
pipes, the work described in this paper provides valuable support for the
conclusion reached by the author in Ref. 3. There it was shown that if
several circumferential through-wall cracks, together with their as-
sociated plastic hinges, are present in a stainless steel piping system, then
the tendency towards instability of circumferential growth of a given
crack is no greater than for the case where the given crack is present in
isolation. A basic assumption in the analysis in Ref. 3 was that the plastic
rotation recovery at a non-growing crack cross-section can be ignored,
and this assumption is vindicated by the preceding section's analysis.
It is important to appreciate that a prime reason why this conclusion is
likely to be valid is that the material's crack growth resistance is
presumed to be high and in this context a lower limiting value of 50 for the
material's tearing modulus has been identified. If the tearing modulus
falls below this value, neglect of the various terms in the preceding
section's analysis cannot be justified and the situation must then be
assessed in a completely different light. Indeed, in this new type of
situation, the tearing modulus approach is unlikely to be valid as has been
demonstrated elsewhere ;6 one will then have to resort to some alternative
approach, such as, for example, an approach based on the maintenance of
a constant crack tip opening angle at the tip of a propagating crack.

ACKNOWLEDGEMENTS

The work described in this paper is part of the Electric Power Research
Institute Program on Pipe Cracking. The author thanks Drs T. U.
Marston and D. M. Norris for valuable discussions in this problem
area.
62 E. Smith

REFERENCES

1. Tada, H., Paris, P. C. and Gamble, R. M., Proceedings oJ' the Twelfth
National Symposium on Fracture Mechanics, ASTM STP 700, July 1980,
p. 296.
2. Paris, P. C., Tada, H., Zahoor, A. and Ernst, H., Proceedings of Symposium
on Elastic Plastic Fracture, ASTM STP 668, March 1979, p. 5.
3. Smith, E., The instability of growth of a through-wall crack in a piping system
when other cracks are present, Int. J. Pres. Ves. & Piping, 20 (1985), this issue.
4. Dugdale, D. S., J. Mechs. Phys. Solids, 8 (1960), p. 100.
5. Bilby, B. A., Cottrell, A. H. and Swinden, K. H., Proc. Roy. Soc., A262
(1963), p. 304.
6. Smith, E., Int. J. Fracture, 17 (1981), p. 373.
7. Bilby, B. A., Cottrell, A. H., Smith, E. and Swinden, K. H., Proc. Roy. Sot.,
A279 (1964), p. 1.
8. Paris, P, C., Ernst, H. and Turner, C. E., Fracture Mechanics." Tweff?h
Confi'rence, ASTM STP 700, July 1980, p. 338.
9. Rice, J. R., Mechanics and Mechanisms o[' Crack Growth, Proceedings of
Conference at Cambridge, M. J. May Ed., British Steel Corporation Physical
Metallurgy Centre Publication, 1974, p. 14.

APPENDIX

In analysing this paper's model, the results ~ for a periodic system of


coplanar cracks in an infinite solid will be used, the solid being cut along
vertical surfaces so as to give the model of a solid of total width 2h
containing two symmetrically situated cracks of depth c, the solid
deforming under plane strain conditions due to the application of an
applied tensile stress a (this cutting procedure is exact for the analogous
Mode I II model). Using the DBCS representation of yield,4"5 the results 7
for the contained yield situation show that s and J are given respectively
by the expressions

I
s = 2ec t.
sin na _ sin2 nc tan
(A.1)
sin [ ~ )

and
8(1 -- v2)h Ysin e ~./2 cos Z In [sin (Z + ~ ) ] dz
J= (A.2)
n2E J~, x/1 - sin2~sin2z
Crack instability in a multiple-cracked system 63

where (a - c) = R p is the size of the plastic zone at each tip and o~= na/2h
is given by the expression

sin(n~-~) (~__y)
cos = sin ~b (A. 3)
sin
For the special case where the cracks are deep in comparison with the
solid width, when the model reduces to the case where a load P is applied
at a point a great distance away from the ligament (thickness B), eqns
(A.1), (A.2) and (A.3) give
s = 2eL22 (A.4)
and
4(1-v2)y2L
J= [(1 + 2)ln (1 + 2 ) + ( 1 - 2)In(1 - 2 ) ] (A.5)
nE
where 2 = P / 2 B L Y ; 2 is small for small-scale yielding conditions while
2 = 1 at general yield. For the general yield state, the parameter s is given
by the relation
s = 2eL (A.6)
while the J-integral can be separated into an elastic component JEL (the
value of J assuming there to be no plasticity), i.e.
4(1 - v 2 ) y 2 L
JEL -- nE (A. 7)

and an additional (plastic) contribution JPL, given by eqns (A.5) and (A.7)
with 2 = 1, as
4(1 - vZ)YZL (21n 2 - 1) (A.8)
JPL ----- nE

The load point displacement A can be separated into respectively elastic


and plastic components AEL and ApL. Then using the same dimensional
arguments as Paris et al.8 for the contained yield situation, JPL can be
expressed in the form

JPL _ _ ).g(2) I ~g(e) de (A.9)


2LY 2 Jo
64 E. Smith

with AEL ~ Lg(2). Thus if g(2) = dI(2) d2, eqn (A.9) reduces to
dI 21 JpL(2)
(A.lO)
d2 2 LY2
which integrates to give

22 f~ JPL(e)de (A.1 1)
I=~ e3

Since eqn (A.5) gives the value of the J-integral, and because the elastic
component JEL is
4(1 -- v2)y2L22
JEL = roE (A. 12)

it follows that the plastic component JPL is given by the expression


4(1 - v2)y2L
JPL~- ~E [(1 + 2)ln (1 + 2 ) + ( 1 - 2)ln(1 --2)--22] (A.13)

whereupon substitution in eqn (A. 11) gives I. It then follows that


LdI
ApL = Lg(2) -- d2

4(1 - vz)YL
- [22-(1 + 2)In(1 + 2 ) + ( 1 - 2 ) l n ( 1 - 2 ) ] (A.14)
roE
At general yield (2 = 1), ApL is given by the expression
8(1 - v2)yL
ApL -- rtE (1 - In 2) (A. 15)

while the elastic component AEL of the load point displacement is given 9
by the expression

ApL=AELNcR+AELcR (I-v2)LDY 8(1-vZ)LYln(2h ~


' " - hE + ~E \~LJ (A.16)

where AEL,NCRis the contribution assuming there is no crack, and AEL,CRis


the contribution due to the crack's presence.

You might also like