You are on page 1of 21

^ .

t*
x :A<f5
ORNL/NUREG/TM-27

Analytical Relations Between


Elastic-Plastic Fracture Criteria

J. G Merkle

MASTER
i r:r
ai—<• MTr~f"-"-T** *&&

BLANK PAGE
Primed in the United States of America. Available from
National Technical Information .Service
U.S. Depigment of Commerce
5 2 8 5 Port Royal Road. Springfield. Virginia 22161
Price: Printed Copy 84-86 Microfiche S2.25
___,
i_£_
:

This report was prepared as an account r>< work sponsored bv (he United States
Government Neither the U n : e J States nor the Fnergy Research and Development
Administration United States N t " ' " r Regulatory Lcmrmssion. n c any ot thei'
employees, nor any o' their contractors, subcontractors, or t.ieir employees makes
any warranty, express or inpned. or assumes any iegai iiat»i-ty or responsibility for
the accuracy, completeness or usefulness of any information, apparatus, product or
process disclosed, or represents tr>at its use would not infringe privately owned rights
CfiHL/NUREG/*rM-S?7
NRC-5

Contract W-7*»05-eng-26

Reactor Division

ANALYTICAL RELATIONS BETUEEK ELASTIC-PLASTIC


FRACTURE CRITERIA

J. 3 . Merkle

Work funded by the Nuclear Regulatory Coaaicsion


under Interagency Agreement 1*0-1*95-75

JULY 1976

OAK RIDGE NATIONAL LABORATORY


Oak Ridge, Tennessee 37830
operated by
UNION CARBIDE CORPORATION
for the
ENERGY RESEARCH AND DEVELOPMENT ADMINISTRATION Atatfvv
• ^
r —
iii

CONTESTS

FoaEwoi© v
i
1
ABSTRACT i

nrrRooucTioN i
T
THE JK PROPOSED C X 3ES11Z CURVE 2

! ASALrTICAL REPRSSSSTATI>?: OF THE SOCIALIZED C X DESIGS

| CURVE 6

APPLICATIONS . 9

DISCTJSSIOK i:

COKCLUSICK il

REFERENCES 12

riOKESCLATURE Ij

i
i

I.

i
.-REWORD

The work .-*«:.-*•* iicrei.'. was perforaed at the Oak Ridge National
Laboratory, under .budget Activity "o. TA 01 19 03 2 , Project T i t l e , "Frac­
ture and Irradiation E f f e c t s . " This work i s sponsored by the U.S. Iluclear
Regulatory .'omission, Office <*i Standards Development, and i s directed
a t 3RNL by J. Z. Whitaar.. ~he :3>.C cognizant engineer for t h i s work i s
P. :.'. Randall of the Structures and Cceponents Standards Branch.
ANALYTICAL RELATIOB BETWEEN ELASTIC-PLASTIC
FRACTURE CRITERIA

J. G. Meriae

ABSTRACT

The equation of the normalized COD design curve recently


proposed in the UK as a basis for determining allowable crack
sizes is derived from the Equivalent Energy approximation for
the J Integral. It is also shown that another approximation
for the J Integral recently proposed by Meetinghouse is mathe­
matically equivalent to the normalized COO approach.

IHTRODUCTION

There are now in existence several proposed criteria for fracture in


the elastic-plastic range, each of which has a certain amount of experi­
mental evidence to support its validity. Prominent among these criteria
are the Crack Opening Displacement, the J Integral and the K I c d parameter
of the Equivalent Energy method. Progress has been made in developing
practical experimental and analytical procedures for applying each of
these criteria separately. Nevertheless, future progress in the develop­
ment of codes and standards could be ft. ilitated by a demonstration that
these criteria have a partially common analytical basis and should there­
fore approximately agree with each other. In the U.S., efforts are being
made to standardize the procedures for obtaining fracture toughness values
from nonlinear load-displacement test records, using either the J Integral
or the Equivalent Energy method of analysis. In the United Kingdom, a
proposal has already been made to establish allowable defect sizes in
1
structures by means of a Crack Opening Displacement analysis, despite the
fact that problems of definition and measurement still exist with respect
to determining critical COD values. The purpose of this discussion is to
present a derivation of the proposed UK relationship between allowable de­
fect sizes and critical COD values, using both the Equivalent Energy method
and the J Integral as bases for the analysis.
$

BLANK PAGE
2

THE UK H10P06ED COD DESKS CURVE

In developing n basis for an inelastic fracture analysis method, one


approach is to assume that a parameter defined on the basis of linear
elastic fracture mechanics still has essentially the sane meaning in the
1
plastic range. Burdekin and Daves proposed the graphical relationship
between allowable effective defect size, a a a x , and critical COD, 6 , showr.
C

t h e
in Fig. 1, by relating <p, the normalized COD, to «/«y, ratio of the
nominal strain to the uniaxial yield strain. The definition of <p is based
on linear elastic fracture mechanics, beginning with the general expres­
sion for the crack tip stress intensity factor, which is

Kj = CO y/0* . (1)

Ine effective defect size, a", is defined by

a = C'a , (2)

so that, combining Eqs. (l) and (2) gives

r~
K x = a vWa X
(3 .

For plane stress, the plastic zone radius, r , is estimated by the formula v

Y -*{$ '
r = ^:l~) - «'M

In addition, the e l a s t i c strain energy release rate, 6T, i s related t o


Kl by

^ = *i f (5)

3
and 6i i s related by 6 by

* T - «."6 • f (6)
S u b s t i t u t i n g F.q. (6) i n t o Eq. ( 5 ) , and then Eq. (5) i n t o Eq. (k) gives

r = -i— .'71
r
Y 2TTCV '

where * v is the uniaxial yield strain. Dividing both sides of Eq. (?)
by a gives the defi.nitior of y, the normalized COD, which is

8
* = rr = — = • ' )
a 2TTCya

The normalized TOT design curve shown in Fig. 1 is roughly parabolic


ir. the elastic range, and becor.es a straight line in the plastic range.
The linear shape of the curve in the plastic range is semi-empirical, as
discussed in Ref. 1 ani ^. The calculations for the two labeled points
in the elastic range in Fig. I were done by two different methods. The
calculation for design conditions was done by a conventional linear elas­
tic fracture mechanics formula. The calculation for proof test conditions
was done with the equation for the strip yield model in fracture me­
8
chanics. '"* For design conditions, the relationship used is

Kj = 1.2 c Vna , (9)

where the factor l.<: provides an adjustment for both plastic zone size
and free surface effects. For c c v = 2 •,, which represents the r.S. 1515
and ASMH "ode Section III primary membrane stress limits, with respect to
yield stress, for normal service conditions, Eq. (9} gives

- ft! (10)

Combining Eqs. (5) and (6) gives

Kj = Eo 6 , Y (11)
I*

from which it follows that

= C (12;
W Y*
Thus, substituting Eq. (12) into Eq. (10) gives, for the case of o c . ^ 2 M ,

= • <>•' ft) • «u'


which agrees with Fig. 1. Rearranging Eq. (8) gives

'ft)
2n a

1
and substituting Eq. (13) *nto Eq. (lU) gives, for the case of a/c. { = 2 ';,

9 = 0.32 , (15)

which also agrees with Fig. 1.


1 —3
The working equation of the s t r i p y i e l d model is

80^ /TT o \
6 = * fasec _ - ) , (16)

which rearranged gives

a = (17)
*— wi secFl^ *
TT
y

Referer.ee 1 ass uses that under proof -.est conditions, the pressure i s 1-5
tLr.es the lesigr. p r e s s - r e . and t h a t cor.sequently z z.. - Z.zT. For t h i s
7
value of z z v t Eq. ' l " ' gives

whicr. agrees vi^h F i ; . "..


e 3 -ir.stitu*ir^ E3. . I ^ ir.to Eq- 'I*.) gives

V = :•.•>• , •i* •

whtch also agrees with Fig- 1.


:-eca use two different equations were used in ?ef. 1 for the calcula­
tions pertaining to the labeled points or. the elastic portion of the curve
i-. Fu:. 1, the equation of this part of the curve is uncertain. In addi­
tion, it appears that the slopes of the two portions of the curve in Fig.
do not quite satch at their point of intersection. Instead, the elastic
range curve appears to lie slightly lelov the downward projection of tht
plastic rir.ge curve as the intersection point of the two parts of the
curve is approached. Froc the information given in Fig. 1, it car. .e
detemined that the equation of the linear portion of the curve in Fig. 1
is

If Fq. (2^) is extrapolated back into the elastic range, the value of $
for design conditions, for which c c . = 2 \, is .!ul*2, which exceeds the
elasticsUy calculate^ value of Z.tf. Thus the two parts of the curve
shown, in Fig. I prorarly do not aeet at a point of tangency.
6

ANALYTICAL REPRESENTATION' OF THE NORMALIZED


COD DESIGN CUR..E

The anomalies in the shape of the normalized COD design curve men­
tioned above have recently been remedied by Dawes, who suggested that
the curve should be represented by the equation

? = (e/Cyf , for e/ty £ 0.5 , (21)

ani by Eq. (20), for «/*,, ^ J.5. Eqv^tior (21) defines the parabola that
joins the straight line defined by Eq. (20) at a point of tangency, at
which e/iy = 0-5-
5
Equation (21) was derived by Dawes by writing the first term in the
series expression for the quantity on the right h£.r<1 side of Eq. (l6), as
discussed in Ref. 2, and then multiplying the result by a safety factor
of 2.0. While "vhe analytical basis of Eq. (21) is thus clear, it would
still be desirable to find a single analytical model, even though it might
be approximate, that would produce the entire normalized COD design curve,
and therefore lead to the derivation of both Eqs. (20) and (2l). As it
turns out, such an analyt:.cal model does exist.
6 7
It has been established that the Equivalent Energy Method provides
an approximate estimate of the -J Integral, by means of the equation

J = 2nd* aU (|-) , (22)

where U is the nominal work per unit volume at a reference point near the
location oi' the flaw, E ' is the initial slope of the nominal stress versus
noainal strain curve at the same point, and the other terms in Eq. (22) are
as previously defined. If uniaxial tension is assumed, then

f - 1 . (23)

Therefore, substituting Eqs. (?) and (2-0 into Eq. (22) gives

J = 2n"an . (2U)
7

For nonlinear conditions, J i s considered t o be a generalization of * j ,


so that

J = OyG - (25)

Combining Eqs. (2k) and (25) gives

6 = 222 , (26)
°Y

and substituting Eq. (26) into Eq. (8) gives

<P - 7T7- • (27)


o-Y*Y

Since

oy = ECy (28)

Eq. (27) can a l s o be written in the form

U
9 = . (29)
E«Y

Thus the expression for cp depends on the expression for U, which depends
on the shape of the stress-strain curve. The shape of the stress-strain
curve for an elastic-ideally plastic material is shown by the solid curve
in Fig. 2. If U is expressed as a function of strain, then a factor of
safety can be applied to the flaw size by overestimating the value cf U
for a given value of strain. This can be done by applying the safety fac­
tor to the elastic modulus, without changing the yield stress, as indi­
cated by the dashed curve in Fig. 2. The effect is to establish a reduced
yield strain which is the actual yield strain divided by the factor of
safety. Below the reduced yield strain,

u = § E<* , (30)
p

where n is the factor of safety on flaw size below the reducei yield
strain. Substituting Eq. (30) into Eq. (29) gives

i3

m- "7-
n 1 «

Above the reduced yield strain

OyCy
U = OyC , (32)

and substituting Eq. (32) into Eq. (27) gives

9 ---

For a value of n = 2, Eqs. (31) and (33) give

V 2
9 - ({$ > « * 0.5 c ,y (21)

and

9 = V 25
(«?) " " ' « * 0-5 «Y » (20)

which are the sane as the equations for the normalized COD design curve
4
proposed by Dawes. Prom Eq. (27) it follows that

do a
Q

Consequently, Eqs. (31) *nd (33) represent a parabola and a straight line
that seet at a single point of tangency, and there are no ancaalies in
the shape of the normalized COD curve proposed by Daves/

APPLICATIGHS

For a safety analysis, it is usually necessary to estimate an allow­


able crack size for a given strain level, and an allowable ctrain level
for a given crack size. Suppose that the following information is given:

K.^ = 200 ksi JixT

Oy = 65 ksi

o = 36 ksi

aj. = 1.0 in.

where a. is the estimated final effective flaw size. Substituting Eq.


(12) into Eq. (8) gives

#
9 =— z — > (35)
2fl
»»ax

which rearranged gives

m
Gtf (36)
•** TT"
Since «/« = o/o = 36/65 = 0.55 *, 9 i8 determined from Eq. (20) to be
Y y
1

0.30U. Then, using Eq. (36), the value of a ^ ^ is U.95 in. Thus the
final effective flaw size is smaller than the allowable effective flaw
size by a factor of U.95. For the allowable strain corresponding to the
10

estimated final effective flaw size, the value of <p, from Eq. (35), is
1.50. Then, from Eq. (20), the value of s/Cy is 1.73, indicating that
yielding should precede fracture. In this case, tbe allowable strain ex­
ceeds the applied strain by a factor of 3-l6.

DISCUSSION

Another approach to this problem, that produced a mathematically


8
identical result, was the analysis of Begley, Landes and Wilson, based
on the J Integral. The estimating equations for the J Integral given in
Ref. 8, when written in terms of the effective flaw size, are

J = e* En* , « i Cy , (37)

and

J = Ena« (2c - cy) ,


y c 2 «y • (33)

Using Eqs. (25) and (8), Eqs. (37) and (38) reduce to

* = ?{$' ****' (39)

and

9 = (J-) - 0.5 , C 2 «y (UO)

which are identical to the equations that result from using n = 1 in Eqs.
(31) and (33)> Thus the result of the analysis developed in Ref. 8 is
simply a normalized COD curve with a factor of safety of 1.0. Unfortunately,
there is no mention of this fact in Ref. 8.
The fact that an approximate analytical relation has been established
between tbe J Integral, the Equivalent Energy Method and the Normalized
COD approach implies that t&ese methods may share a partial!/ common basis

1
11

and also that they may be subject to SOBS similar limitations. The close
relation between the J Integral and the value of 6 is further substan­
tiated by the elastic-plastic finite eleaent calculations performed by
Suapter and Turner,* for flaws r—nsting frost a circular hole in a flat
plate. These calculations indicate that J/c_ and 6 do follow very similar
trend curves with o/cy, at least for the geometry investigated. However,
10
Egan has also shown experimentally that different flaw geoaetries may
actually follow different q> vs «/«y trend curves, the most noticeable ex­
ample studied being the trend curves for surface cracked and center
through cracked tensile specimens. The Equivalent Energy Method nay be
subject to the same kind of Limitation.

CONCLUSION

It has been shown that an equation proposed recently in the UK for


estimating an allowable flaw size as a function of the critical crack
opening displacement can be derived from the equation that gives the
Equivalent Energy approximation for the J Integral. It has also been
shown that a method proposed more recently by Vestingbouse for estimating
the J Integral reduces mathematically to the same approach.
13

KEFERfXCES

1. F. N. Burdekin and M. G. Dawes, "Practical Use of Linear Elastic and


Yielding Fracture Mechanics with Particular Reference to Pressure
Vessels,'* pp. 26-37 in Practical Application of Fracture Mechanics
to Pressure-Vessel Technology, the Institution of Mechanical Engineers,
London, 1971.
2 . F. M. Burdekin and D.E.U. Stone, "The Crack Opening Displacement Ap­
proach to Fracture Mechanics in Yielding Materials," Journal of Strain
Analysis, 1(2) 1U5-153, 1966.

3. J. H. Goodier and F. A. Field, "Plastic Energy Dissipation in Crack


Propagation," pp. 103-118 in Fracture of Solids, edited by D. C.
Drucker and J. J. Oilman, Interscience, 19627

It. M. G. Dawes, "Fracture Control in High Yield Strength Steel weldments,"


The Welding Journal, 53(9) 3€3s-379«, September 197W.

5. M. G. Dawes, personal communication, February 11, 1975.

6. J. G. Merjcie, "Analytical Applications of the J Integral," pp. 26V-


2&0 in Progress in Flaw Growth and Fracture Toughness Testing, ASTM
STP 536, American Society for Testing and Materials, 1973-

7. F. J. Witt, "The Application of the Equivalent Energy Procedure for


Predicting Fracture in Thick Pressure Vessels," pp. 163-167 in Prac­
tical Application of Fracture Mechanics to Pressure Vessel Technology,
the Institution of Mechanical Engineers, London, 1971.

8. J. A. Begley, J. D. Landes and V. K. Wilson, "An Estimation Model for


the Application of the J Integral," pp. 155-169 in Fracture Analysis,
ASTM STP 560, American Society for Testing and Materials, 197*.

9. J.D.G. Sumpter and C. E. Turner, "Fracture Analysis in Areas of High


Nominal Strain," Proceedings of the Second International Conference
on Pressure Vessel Technology, ASMS, Vol. II, pp. 1095-1103, 1973,
and Vol. Ill, pp. 17**-177, 197U.

10. G. R. Egan, "The Application of Fracture Toughness Data to the Assess­


ment of Pressure Vessel Integrity," Proceedings of the Second Inter­
national Conference on Pressure Vessel Technology, ASME, Vol. II,
pp. 1037-10U8, 1973, and Vol. Ill, pp. lW»-l>>5, 197U.

I
13

.NCMHfCLATURE

a Defect size, generally half-length, in.


a Effective defect size, in.
a^ Pinal effective flav size, in.
8^,^ Allowable effective defect size., in.
C Shape factor, dimensionless
E Modulus of elasticity, ksi
E' Initial tangent modulus of the curve of nominal stress versus
nominal strain at the location of a flav, ksi
*> T Elastic strain energy release rate for plane strain conditions,
a
in.-kips/in .
3
J The J Integral, in.-kips/in.
1 a
Kg The elastic crack tip stress intensity factor, ksi'in. ^
Kj c Critical value of tha elastic crack tip stress intensity factor
1 8
for plane strain conditions, ksi'in. '
n Safety factor on flaw size below the reduced yield strain,
diacnsionless
r„ Plastic zone radius, in.
U Nominal strain energy density, ksi
6 Crack opening displacement, in.
6 C Critical value at the crack opening displacement, in.
s Strain, in./in.
« Y Yield strain, in./in.
a Stress, ksi
0y Yield stress, ksi
9 Normalized crack opening displacement, diaensionless
Ik

OMtL-OWG 75-5577

— 1 — —\ 1 1 • 7 m— T i
* Fro«f test •»•«!« ,-"*
« A*» e/> =7 0 . ^ = 6 75
r

5 ff««=0024^/f r

» 5
v. I!
*;Pre*r test iwzie ^
•i5to«.=O03*r/*
- ' S - i * / e = * 0 . ^ = 5 75

Shell
r r

• 4 - •e •Sl r
</>
o
o
_, 3
V
«/»
=*
^c
^2 i•
§s o
k
• -275
«-•» = •s-
-i ^y^
^ 1
V 5J o 0OM /' c

5 si «*i75
c
:
* 2 -
m
- © s
«*«. 0 096,/'
j
r i
•• ~f w° ^ ^ i

i
IM
•»
5 *
m

%
1 |

A» ** e i
i ™
- ©•* Oft «, k-
O m.

1
>
• ^ ^ ^
^^£" _ J « J l i , •
05 10 15 20 2S 30 35 40
STRAIN e/«y

Fig. 1. Horaalized COO design carve proposed in Ref. 1. The


letters A.W. stand for "as welded," and the letters S.R. stand for
"stress relieved."

OMUL -DUG 75 57S7

Fig. 2. Elastic-ideally plastic stress-strain curves used to de­


rive the equations of the Modified COD design curve.

You might also like