You are on page 1of 32

Elastic-Plastic Fracture Mechanics

Introduction
When does one need to use LEFM and EPFM?
What is the concept of small-scale and large-scale yielding?

Background Knowledge
Theory of Plasticity (Yield criteria, Hardening rules)
Concept of K, G and K-dominated regions
Plastic zone size due to Irwin and Dugdal

Contents of this Chapter


The basics of the two criteria used in EPFM: COD (CTOD), and J-Integral (with H-R-R)
Concept of K- and J-dominated regions, plastic zones
Measurement methods of COD and J-integral
Effect of Geometry
LEFM and EPFM
LEFM
In LEFM, the crack tip stress and displacement field can be uniquely characterized by K, the
stress intensity factor. It is neither the magnitude of stress or strain, but a unique parameter that
describes the effect of loading at the crack tip region and the resistance of the material. K filed is
valid for a small region around the crack tip. It depends on both the values of stress and crack size.

We noted that when a far field stress acts on an edge crack of width a then
for mode I, plane strain case
LM1 sin( ) sin( 3 )OP
R
| U
| M 2 2
P
M 3 P
xx

S
|
yy V
|

KI
2r
cos
2 M
1 sin( ) sin(
2 P
)

T xy W MM
2
3
PP
MNsin( 2 ) sin( 2 ) PQ
zz 0 for plane stress; zz ( xx yy ) for plane strain

LMcos (k 1 2 sin ( ))OP


RSu UV K
2
x I r
2 M
M 2 2
P
P
Tu W 2
y
MNsin 2 (k 1 2 cos ( 2 ))PQ
2
LEFM cont.

For =0
R| xx
U| LM1 OP
S| yy V| KI
2r MM1 PP
Singularity dominated region
T xy W N0 Q

For = , all ij 0
2

LEFM concepts are valid if the plastic zone is much smaller than the singularity zones.
1 KI 2
Irwin estimates rp ( )
2 ys
1 KI 2
Dugdale strip yield model: rp ( )
8 ys
KI 1
ASTM: a,B, W-a 2.5 ( )2 , i.e. rp of specimen dimension.
ys 50
EPFM
In EPFM, the crack tip undergoes significant plasticity as seen in the following diagram.

sharp tip

Load ratio, P/Py


1.0
Fracture

Ideal elastic brittle behavior


cleavage fracture
P: Applied load
Py: Yield load Displacement, u

Blunt tip

Load ratio, P/Py


1.0
Fracture

Limited plasticity at crack


tip, still cleavage fracture
Displacement, u
Blunt tip

Load ratio, P/Py


Fracture
1.0

Void formation & coalescence


failure due to fibrous tearing
Displacement, u
large scale
blunting

Load ratio, P/Py


1.0 Fracture

Large scale plasticity


fibrous rapture/ductile
failure Displacement, u
EPFM cont.
EPFM applies to elastoc-rate-independent materials, generally in the large-scale plastic
deformation.
Two parameters are generally used:

(a) Crack opening displacement (COD) or crack tip opening displacement (CTOD).
(b) J-integral.
Both these parameters give geometry independent measure of fracture toughness.

y
x
Sharp crack

ds
Blunting crack
EPFM cont.
Wells discovered that Kic measurements in structural steels required very large thicknesses for
LEFM condition.
--- Crack face moved away prior to fracture.
--- Plastic deformation blunted the sharp crack.
4 K I2
2u y
2ys E
3
Note: k and E 2 (1 )
Sharp crack 1
2
4 G K
Blunting crack CTOD since G I
ys E

Irwin showed that crack tip plasticity makes the crack behave as if it were longer, say from size a to a + rp
1 KI
rp ( )2 -----plane stress
2 ys

From Table 2.2, KI r
uy sin( )[k 1 2 cos 2 ( )]
2 2 2 2
k 1 ry a ry
Set = , uy KI
2 2
CTOD and strain-energy release rate
4 G
Equation CTOD relates CTOD ( ) to G for small-scale yielding. Wells proved that
ys

Can valid even for large scale yielding, and is later shown to be related to J.

can also be analyzed using Dugdales strip yield model. If is the opening at the end of the strip.

Consider an infinite plate with a image crack subject to a


ys
2u y
8 ys a
lin sec(

)
E 2 ys

Expanding in an infinite series,


8 ys a 1 2 1 4
[ ( ) ( ) ...]
E 2 2 ys 12 2 ys

K I2 1 2
If ys
0 ( ys ), then =
K 2I

G
ys E ys
, and can be given as:
ys E
[1 (
6 2 ys
) ]

In general, G , m = 1.0 for plane stress; m = 2.0 for plane strain


m ys
Alternative definition of CTOD

Sharp crack
Blunting crack

Blunting crack

Displacement at the original crack tip Displacement at 900 line intersection, suggested by Rice
CTOD measurement using three-point bend specimen displacement
Vp

z
p
a

r p( W
- a) expanding
W
'
P
'
rp (W a)Vp
'
rp (W a) a z
pl
Elastic-plastic analysis of three-point bend specimen

rp (W a)Vp

loa d
K I2 V,P
el pl
m ys E rp (W a) a z

Where pl is rotational factor, which equates 0.44 for SENT specimen.


p e

Specified by ASTM E1290-89 Mouth opening

--- can be done by both compact tension, and SENT specimen


Cross section can be rectangular or W=2B; square W=B

K I2 (1 2 ) rp (W a)Vp
el pl
2 ys E rp (W a) a z
P a
KI is given by KI f( )
B W W
CTOD analysis using ASTM standards
loa d
Pc Pu Pm
Pi
Pi
fracture fracture

(a) (b) (c)

Mouth opening

Figure (a). Fracture mechanism is purely cleavage, and critical CTOD c <0.2mm, stable crack growth,
(lower transition).
Figure (b). i --- CTOD corresponding to initiation of stable crack growth.
u --- Stable crack growth prior to fracture.(upper transition of fracture steels).
Figure (c) i and then m---CTOD at the maximum load plateau (case of raising R-curve).
More on CTOD
K I2 J
COD or
yE y
The derivative is based on Dugdales strip yield model. For
Strain hardening materials, based on HRR singular field.
n
J n1 1
ui y r ui , n
n1

y y I n
By setting =0 and n the strain hardening index based on
n 1
y 3 e ij

y 2 y y
*Definition of COD is arbitrary since u y x,0 u y x,0

1
A function x
n1 as the tip is approached
*Based on another definition, COD is the distance between upper
and lower crack faces between two 45o lines from the tip. With this
Definition J
d
COD n
y
Where d n d n , y , n ranging from 0.3 to 0.8 as n is varied from
3 to 13 (Shih, 1981)

*Condition of quasi-static fracture can be stated as the


Reaches a critical value COD . The major advantage is that this
provides the missing length scale in relating microscopic failure
processes to macroscopic fracture toughness.

*In fatigue loading, tip continues to vary with load and is a


function of:
k 2
(a) Load variation I
2 y
(b) Roughness of fracture surface (mechanisms related)
(c) Corrosion
(d) Failure of nearby zones altering the local stiffness response
3.2 J-contour Integral

By idealizing elastic-plastic deformation as non-linear elastic, Rice proposed J-integral, for regions
beyond LEFM.
In loading path elastic-plastic can be modeled as non-linear elastic but not in unloading part.
Also J-integral uses deformation plasticity. It states that the stress state can be determined knowing
the initial and final configuration. The plastic strain is loading-path independent. True in proportional
load, i.e. d 1 d 2 d 3 d 4 d 5 d 6 k
1 2 3 4 5 6
under the above conditions, J-integral characterizes the crack tip stress and crack tip strain and
energy release rate uniquely.
J-integral is numerically equivalent to G for linear elastic material. It is a path-independent integral.
When the above conditions are not satisfied, J becomes path dependent and does not relates to any
physical quantities
3.2 J-contour Integral, cont.

y
x


ds

Consider an arbitrary path ( ) around the crack tip. J-integral is defined as

z u
J (wdy Ti i ds),

xi
w z 0
ij
ij d ij

where w is strain energy density, Ti is component of traction vector normal to contour.


It can be shown that J is path independent and represents energy release rate for a material where ij
is a monotonically increasing with ij
Proof: Consider a closed contour: J * ( wdy Ti
*

z ui
x i
ds)
A *
*
Using divergence theorem: J * z
A*
(
w
x


x i
ij
ui
x
)dxdy
Evaluation of J Integral ---1
w ui
J ( ij ) dxdy
*

x x j
A*
x
w w ij ij
Evaluate ij
x ij x x
w
Note ij is only valid if such a potential function w exists
ij
w 1
Again, ij [ (ui , j ) (uij,,ij )]
x 2 x x
1 ui u j
ij [ ( ) ( )]
2 x j x x i x
Since ij ji

ui
ij ( )
x j x
ij
Recall 0 (equilibrium)
(equilibrium) leads to
leads to
x j
ui ui
ij ( ) ( )
x j x x j x
ij
Evaluation of J Integral ---2
Hence, J * 0. Thus for any closed contour J * 0.

Now consider 2
4

1
3

J J1 J 2 J 3 J 4 0

z
1 2 3 4

w
Recall J* ( wdy t ii ds)

x

On crack face, ti i 0, dy 0 (no traction and y-displacement), thus


J3 J4 0, leaving behind J1 J2
Thus any counter-clockwise path around the crack tip will yield
J; J is path independent.
Evaluation of J Integral ---3

y ti
a
x
2D body bounded by '

'
In the absence of body force, potential energy
z
A'
wdA z
''

t i i ui ds

Suppose the crack has a vertical extension, then


d
da
z
A'
dw
da
dA i
'
z
dui
da

ds (1)

Note the integration is now over '


Evaluation of J Integral ---4
d x x
Noting that since 1
da a a x a x a
d
da
(
'
w w

a xz
A
)dA t ii (
'
w ui

a x
)ds z

(2)

w w i j ui
ij ( )
a ij a x j a
d
Using principle of virtual work, 0 for equilibrium, then from
da
eq.(1), we have
z ij
ui
(
x j a
) dA z t ii
ui
a
ds

zt z
'
A '

Thus, d ii i ds dw dA
du
da '
dx dx
A'
Using divergence theorem and multiplying by -1

d
da
z
'
( wn x t ii
dui
dx
)ds z
'
wdy t ii
w
x
ds
Evaluation of J Integral ---5
d
Therefore, J is energy release rate da
, for linear or non-linear
elastic material
In general U F and J
A
Potential energy; U=strain energy stored; F=work done by
external force and A is the crack area.
p dU * dU

Load
u d
-dP

Displacement

p
U P U * Complementary strain energy = dP
0
Evaluation of J-Integral

dU *
For Load Control J
da p
dU
For Displacement Control J
da

The Difference in the two cases is 1 .dp.d dU and hence J for


2
both load Displacement controls are same

D p

J .dp .dp
a 0 p 0
a p
or

p
J pd .d
a 0 a
K I2
J=G and is more general description of energy release rate J '
E
More on J Dominance

J integral provides a unique measure of the strength of the singular


fields in nonlinear fracture. However there are a few important
Limitations, (Hutchinson, 1993)
(1) Deformation theory of plasticity should be valid with small strain
behavior with monotonic loading
(2) If finite strain effects dominate and microscopic failures occur, then
this region should be much smaller compared to J dominated region
1
Again based on the HRR singularity n1 I
J
ij y ij , n
y y I n r
Based on the condition (2), we would
like to evaluate the inner radius ro of J
dominance. Let R be the radius ro
where the J solutions are satisfied
within 10% of complete solution.
FEM shows that ro 3 COD R
However we need ro should be greater than the forces zone
(e.g. grain size in intergranular fracture, mean spacing of voids)
Numerical simulations show that HRR singular solutions hold
good for about 20-25% of plastic zone in mode I under SSY
Hence we need a large crack size (a/w >0.5) . Then finite strain
region is 3 COD , minimum ligament size for valis JIC is
25 J IC
b
y
For J Controlled growth elastic unloading/non proportional loading
should be well within the region of J dominance

dJ J
and a R
da R

Note that near tip strain distribution for a growing crack has a
logarithmic singularity which is weaker then 1/r singularity for a
stationary crack
Williams solution to fracture problem

Williams in 1957 proposed Airys stress function


R r
As a solution to the biharmonic equation
2
1 1 2
4 0 where 2 2 2 2
r r r r
For the crack problem the boundary conditions are
r 0 for

Note will have singularity at the crack tip but is single valued
r 2 p r , q r ,
Note that both p and q satisfy Laplace equations such that
2 p 2 q 0
Now, for the present problem.

p A1r cos Az r sin


q B1r 2 cos 2 Az r 2 sin 2
Then
r 2 A1 cos B1 cos 2
r 2 A2 sin B2 sin 2
Consider only mode I solution with
r 2 A1 cos B1 cos 2
2
2 1 2 r A1 cos B1 cos 2
r
1
r
r r
1 r A1 sin 2 B1 sin 2
Williams Singularity3
Applying boundary conditions,
A1 +B1 cos 0
A1 2 B1 sin 0

cos 0 2Z 1
Case (i) , Z=0,1,2...
2
or,
B1 A
2 1
Case (ii) sin 0 Z B1 A1

Since the problem is linear, any linear combination of the above two will also be
acceptable.
Thus Z2 with Z= ... 3, 2, 1,0,1,2,3...
Though all values are mathematically fine, from the physics point of view, since
ij r and ij r
Williams Singularity4
U 12 ij ij r 2
2 R
= 1
ij ij rdrd
0 r0 2

r
2A R 2 1

0 r0
drd

Since U should be provided for any annular rising behavior r0 and R ,

U as r0 0, 1 ( 1 makes ij 0)
Also, ui r r 1 needs > 1. Thus
=- 12 ,0, 12 ,1, 23 , 2... with = Z2 where Z=-1,0,
positive number.

The most dominant singular form


=- 12 and B1 A1
3
Williams Singularity4
Now r A1 cos 2 13 cos 32
3
2


+ r 2 ...

r ...
5
2

and 2

ij A1r ij I ij r 0 ij r
1 1
2

where indicates the order of


Note the second term in ij ij r 0 is a non-singular
and non-vanishing term. However, higher order vanish as r 0
KI
with A1
2
KI
ij ij I T ix jx (no sum on x)
2 r
Williams Singularity5
Now
r A1 cos 2 13 cos 32
3
2


r 2 ...

r ...
5
2

and I

ij A1r ij ij r ij r
12 0

1
2

where indicates the order of


Note the second term in ij ij r 0 is a non-singular
and non-vanishing term. However, higher order vanish as r 0
KI
with A1
2
KI
ij ij I T ix jx (no sum on x)
2 r
Williams Singularity6

For in-plane stress components,


xx xy K I xx xy T 0
I I

I
yx yy 2 r yx yy 0 0
I

Second-term is generally termed as "T-stress" or


"T-tensor" with xx T
For brittle crack of length 2a in x-z plane y
with yy & xx

applied x
K I yy a and
2a

z
T= yy xx

HRR Singularity1
Hutchinson, Rice and Rosenbren have evaluated the character of crack tip
in power-law hardening materials.
Suppose the material is represented by Ramberg-Osgood model,
n


0 0 0
0 Reference value of stress=yield strength

0 0 , strain at yield
E
dimensionless constant
n strain-hardening exponent
Note if elastic strains are negligible, then
n


y
y
n 1
ij 3 eq ij 3
; eq ij
y 2 ij y 2
HRR Singularity2
Then
4 f , 0 , r , n,
C1 r s 2 r t (similar to Williams expression)
k 0 r s

Applying the appropriate boundary conditions


1
EJ n 1
ij 0 ij n,
2

0 n I r
n
0 EJ n 1
ij ij n,
E 0 2 I n r
I n Integration constant
, Dimensionless functions of n and

You might also like