You are on page 1of 67

Solution Manual

for

Fracture Mechanics

by

C.T. Sun and Z.-H. Jin

Updated April 15, 2013

1
Chapter 2

Prob 2.1:
a)

2
P 1 P
P 1 3 P

4 P 3 P

No load is carried by Part 2 and Part 4 . There is no strain energy stored in them.

Constant Force Boundary Condition


The total strain energy per unit width stored in Part 1 and Part 3 is
L P
2
a P
2
P 2 ( L − a) P 2a
U =∫ dx + ∫ dx = +
a 2A E 0 2A E 2 A1 E 2 A3 E
1 3

A1 = h , A3 = h / 3
Thus the total strain energy is
3aP 2 P 2 ( L − a)
U= +
2 Eh 2 Eh
Finally, the energy release rate is
1 ∂U P2
G= =
2 ∂a 2 Eh

Fixed End Boundary Condition


dWe = 0 since dδ = 0, then energy released due to the crack extension is
1
dWs = −dU = P 2 dS
2
2da
dS =
hE
P 2 da
⇒ dWs =
Eh

Finally, the energy release rate is


1 dWs P2
G= =
2 da 2hE

2
b)

P/2 P/2 P/2


2 2, 4

P 1 3 P
P P
3
4 P/2
No load is carried by Part 1. There is no strain energy stored in it.

Constant Force Boundary Condition


The total strain energy per unit width stored in Part 1 and Part 3 is
2
a P ( P 2)2 P2a P2a
U =∫ +2 dx = +2
0 2 AE 2 AE 2 AE 8 AE
A= h/3
Thus the total strain energy is
3aP 2 3aP 2 9 P 2 a
U= +2 =
2 Eh 8Eh 4 Eh
Finally, the energy release rate is
1 ∂U 9 P 2
G= =
2 ∂a 8 Eh

Fixed End Boundary Condition

Based on the fixed end boundary condition, we can have dWe = 0 ⇒ dWs = -dU
Thus the energy released due to the crack extension is
1 P 1
dWs = − dU = 2 ( ) 2 dS1 + P 2 dS 2
2 2 2
−3da 3da
dS1 = , dS 2 =
2hE hE
2
3P da
⇒ dWs =
4 EA
Eventually, the energy release rate is
1 dWs 9 P 2
G= =
2 da 8Eh

3
Prob 2.2:
a) Symmetric case

The strain energies stored in each part are


2 2
a M a ( Px ) P 2a 3 th 3
U1 = U 2 = ∫ dx = ∫ dx = , I1 =
0 2 EI 0 2 EI 6 EI 12
So the total strain energy is
P 2a 3 4P 2a 3
U = U1 + U 2 + U 3 = =
3EI Eth 3
The energy release rate is given by
1 dU 12 P 2 a 2
GI = =
t da Eh 3t 2
b) Anti-symmetric case

The strain energies stored in each part are


a ( Px )
2
P2a3 th 3
U1 = U 2 = ∫ dx = , I1 =
0 2 EI
1 6 EI 1 12
L ( 2 Px ) 2 2P 2 3 2th 3
U3 = ∫ dx = (L − a3 ) , I3 =
a 2 EI 3 3EI 3 3
So the total strain energy is
P2
U = U1 + U 2 + U 3 = ( L3 + 3a 3 )
12 EI
The energy release rate is given by
1 dU 9 P 2 a 2
GII = = 2 3
t da Et h

4
Prob 2.3:
Since the structure is symmetric (both loading and geometry), we can represent the given
cracked beam as two cracked beams as shown in Figure 1.

M is the moment needed to ensure zero slope at the location where we made the free
body cut.

Assuming that the beam marked "1" is a cantilever beam with a tip load P/2, the slope
at the tip is given by

θP =
( P 2) a2
2 EI
The slope at the tip of a cantilever subjected to a tip moment is given by
Ma
θM = −
EI
To ensure that the system shown in Figure 1 accurately represents the actual center
cracked structure with a mid-section load, we need to satisfy
θ p + θM = 0
( P 2 ) a 2 − Ma = 0 Pa
⇒M =
2 EI EI 4
The moment distribution along the length of beam 1 is
P Pa
M ( x) = x − (x=0 being the tip of the cantilever beam).
2 4
The strain energies can be written as
a 2 a 2 2 3
M ( x) dx 1  Px Pa  P a
U1 = U 2 = ∫ = ∫  −  dx =
0
2 EI 2 EI 0  2 4  96 EI
and U3 = 0. The total strain energy is therefore,
2 3
Pa
U = U1 + U 2 + U 3 =
48EI
Strain energy release rate can be calculated as

5
2 2
1 dU P a
G= =
t da 16 EIt
−10 4
The moment of inertia for the specified dimensions is, I = 4.7 ×10 m
The crack will propagate if G ≥ Gc . Hence, we can calculate the minimum load Pmin
required to propagate the crack as given below.
2 2
Pmin a
G = Gc =
16 EIt
16 EItGc
⇒ Pmin = 2
= 216 N
a

6
Prob 2.4:

Let us assume that the crack extends by da as shown above


Total distance d moved by the loading point is given by
dδ = Elastic extension + peeled off length
= ε ( L + da ) − ε L  + da
= da (1 + ε )
 P
= da 1 +  ; assuming unit width
 At 
Work done by the external force is therefore,
 P
dWe = Pd δ = P 1 +  da
 Et 
Change in strain energy is given by

dU = U L + da
−U L

1  P  P  1  P  P 
=    ( L + da ) A −    ( L) A
2  A  EA  2  A  EA 
P 2 da
=
2 Et
Hence,
dWs = dWe − dU
 P 
= P 1 +  da
 2 Et 
Strain energy release rate is given by

7
dWs  P 
G= = P 1 + 
da  2 Et 

Note that in this particular problem, dWs ≠ dU

8
Prob 2.5:

Let the crack extend by da as shown above. The incremental displacement can be written
as
Pda
dδ = , taking unit width
Et
Work done by external force,
2
P da
dWe = Pdδ =
Et
Change in strain energy is given by
dU = U L + da − U L
1  P  P  1  P  P 
=    ( L + da ) A −    ( L) A
2  A  EA  2  A  EA 
P 2 da
=
2 Et
Hence,
2
P da
dWs = dWe − dU =
2 Et
Strain energy release rate is
2
dWs P
G= = .
da 2 Et
In this case, dWs = dU .

9
Comparing Prob 2.4 and 2.5

As seen in the two problems above, the strain energy release rate and the strain energy
gained by the film are not equal for prob 2.4 and are equal for prob 2.5. This is due to
nature of the P-δ curves shown in Figure 4. In prob 2.4, the nature of the loading makes
the system non-conservative. It should also be noted that for the same load P, prob 2.4
has a higher G and hence it is easier to propagate the crack in prob 2.4 than in prob 2.5.

P P

Pda
Pda
da Et
Et

δ δ
PL P ( L + da ) PL P( L + da)
+ da
Et Et Et Et
Prob 2.4 Prob 2.5

10
Chapter 3

Prob 3.1:

Given the Airy stress function as


φ = Ay 2
By definition, we can get the stresses
∂ 2φ
σ xx = 2 = 2 A
∂y
∂ 2φ
σ yy = 2 =0
∂x
∂ 2φ
σ xy = =0
∂x∂y
Based on the stress-strain relation
1  λ*  1  λ* 
exx = σ xx − ( σ + σ )  , e = σ − ( σ + σ ) 
2µ 
 2 ( λ *
+ µ ) xx yy


yy
2µ 

yy
2 ( λ *
+ µ ) xx yy

3 − 4ν for plane strain
1 3−κ 
exy = σ xy , λ = *
µ , κ =  3 −ν
2µ κ −1  1 + ν for plane stress
The strain can be derived as
A λ * + 2µ A λ*
exx = , e = − , exy = 0
2µ λ * + µ 2µ λ * + µ
yy

By the strain-displacement relation


A λ * + 2µ
ux = x + f1 ( y )
2µ λ * + µ
A λ*
uy = − y + f2 ( x )
2µ λ * + µ
Both f1(y) and f2(x) are the rigid body terms. Check by exy = 0
∂f1 ( y ) ∂f ( x )
=− 2 = C1
∂y ∂x
Eventually,
A λ * + 2µ
ux = x + C1 y + C2
2µ λ * + µ
A λ*
uy = − y + C1 x + C3
2µ λ * + µ
C1 shows the rigid body rotation, C2 and C3 show the rigid body displacement.

11
Prob 3.2:

The Airy Stress Function can be expressed as:


φ = Re { zψ ( z ) + χ ( z )}

For Mode I
ZI is defined as
Z I = 2ψ ' ( z ) − A
and
χ ''( z ) = − zψ ''− A
Do the integration with respect to z
χ '( z ) = ∫ − zψ '' dz − Az = −  zψ '( z ) − ∫ψ '( z )dz  − Az = − zψ '( z ) + ψ ( z ) − Az
1
χ ( z ) = ∫ − zψ '( z )dz + ∫ψ ( z )dz − ∫ Azdz = − zψ + 2∫ψ dz − Az 2 = − zψ + Z I
2
Substitute the equations above into the Airy stress function, we can get
φI = Re { zψ ( z ) − zψ ( z )} + Re{Z I } = Re {−i 2 yψ ( z )} + Re{Z I }

= y Im {2ψ ( z )} + Re{Z I } = y Im{ Az + Z I } + Re{Z I }

= Re{Z I } + y Im{Z I } + Ay 2

For Mode II
ZII is defined as
Z II = 2iψ ' ( z ) − B
Also we have
χ ' ' = −2ψ ' ( z ) − zψ ' −iB
Similarly, do the integration with respect to z
B
χ = − zψ − i z 2
2
Substitute the equation above into the Airy stress function, we can get
φII = − y Re{Z I }
Eventually, the Airy Stress Function can be written as
φ = φI + φII = Re{Z I } + y Im{Z I } + Ay 2 − y Re{Z II }

12
Prob 3.3:

Stress function is given as


φ = Re{Z I } + yI m {Z I }
According to the chain rule and the Cauchy-Riemann relation
∂ Re {Z I } ∂ Im {Z I }
= Re {Z I′ } , = Im {Z I′ }
∂x ∂x
∂ Re {Z I } ∂ Im {Z I }
=− = − Im {Z I′}
∂y ∂x
∂ Im {Z I } ∂ Re {Z I }
= = Re {Z I′}
∂y ∂x

∫ Re {Z }dx = Re {Z } , ∫ Im {Z }dx = Im {Z }
I I I I

∫ Re {Z }dy = Im {Z } , ∫ Im {Z }dy = − Re {Z }
I I I I

The first-order derivatives of Airy stress function are


∂φ
= Re{Z I } + y Im{Z I }
∂x
∂φ
= − Im{Z I } + Im{Z I } + y Re{Z I } = y Re{Z I }
∂y
The stresses can be written as
∂ 2φ
σ xx = 2 = Re {Z I } − y Im {Z I′ }
∂y
∂ 2φ
σ yy = 2 = Re {Z I } + y Im {Z I′}
∂x
∂ 2φ
σ xy = − = − y Re {Z I′}
∂x∂y
Stress-Strain relations are given by
1  λ*  1 κ −1 
exx =
2µ 
σ xx −
2(λ + µ )
( σ + σ )
yy  =  Re {Z I } − y Im {Z I′ }
 2 µ  2
xx

*

1  λ*  1 κ −1 
eyy =
2µ 
σ yy −
2 (λ + µ )
( σ + σ yy )  =  Re {Z I } + y Im {Z I′ }
 2 µ  2
xx

*

Do the integration to get the displacement
ux =
1 κ −1

2µ  2
{ }
Re Z I − y Im {Z I }


uy =
1 κ + 1

2µ  2
{ }
Im Z I − y Re {Z I }

13
Prob 3.4:

The given problem of a pressurized crack can be thought of as a superposition of two


separate problems as shown below.

P0

P0
= + P0

(a) (b) (c)


The Westergaard’s Function of (b) is
P0 z
ZI =
z 2 − a2
Boundary conditions for (c) can written as
σ xx = − P0 , σ yy = − P0 , σ xy = 0 at r → ∞
The Airy’s stress function for this case is given by
−P
(
φ c = o x2 + y 2
2
)
Hence, the stress function for the given problem of pressurized crack can be written as
P
(
φ = φ b + φ c = Re{Z I } + y Im{Z I } − o x 2 + y 2
2
)
The stresses can be calculated using superposition from the following equations
σ xx = Re {Z I } − y Im {Z I′ } − Po
σ yy = Re {Z I } + y Im {Z I′} − Po
σ xy = − y Re {Z I′ }
The whole field stresses are hence given by,
Pr   θ θ  a2 3 
σ xx = o  cos  θ − 1 − 2  − sin θ sin (θ1 + θ 2 )  − Po
r1r2   2 2  r1r2 2 

14
Po r   θ1 θ 2  a 2 3 
σ yy = cos
  θ − − + sin θ sin (θ1 + θ 2 )  − Po
r1r2   2 2  r1r2 2 
Po r  a 2 3 
σ xy =  sin θ cos (θ1 + θ 2 ) 
r1r2  r1r2 2 
Note the coordinate system is the same as that in the course notes.

Checking the boundary conditions


For remote stresses
@ x2 + y2 → ∞
σ xx = Po − Po = 0 ⇒ Satisfied
σ yy = Po − Po = 0 ⇒ Satisfied
σ xx = 0 ⇒ Satisfied

At the crack surface


@ y = 0and x ≤ a we have, θ1 = π ,θ = θ2 = 0
σ yy = − Po ⇒ Satisfied
σ xy = 0 ⇒ Satisfied

Hence the stress function that solves the pressurized crack problem is

{} { }
φ = Re Z I + y Im Z I − o ( x 2 + y 2 ) , Z I =
P
2
P0 z
z2 − a2

Alternative Solution
For (c), the stress can be calculated as
σ xx = Re {Z I } − y Im {Z I′} + 2 A
σ yy = Re {Z I } + y Im {Z I′ }
σ xy = − y Re {Z I′ }
Where A = Po/2. Substituting the boundary conditions of (c) into the stress equations
above, the Westergaard function for (c) can be derived as
Z Ic = − Po
The total Westergaard function based on superposition is
Po z
Z I = Z Ib + Z Ic = − Po
z 2 − a2

15
Prob 3.5:

The coordinate setup is shown below:

The Westergard function is


P a2 − b2
ZI =
π ( z − b) z 2 − a 2
By the polar coordinate shown above, the Westergard function can be rewritten as
θ θ 
P a 2 − b 2 i 21 + 22 +θ3 
ZI = e
π r3 (r1r2 )1/ 2
The first derivatives is
 1 − i θ21 + θ22 +θ3  r iθ − 23θ1 − 23θ2 −θ3  
P a2 − b2
Z I′ = −  e  
+ e 

π r3 r1r2
 r3 r1r2 
The stress component are obtained as
P a 2 − b 2   θ1 θ 2  r  θ1 θ 2 
σ xx = cos  + + θ3  − sin θ sin  + + θ3 
π r3 r1r2   2 2  r3 2 2 
r2  3 3 
+ sin θ sin  θ − θ1 − θ 2 − θ3  
r1r2  2 2 
P a 2 − b2   θ1 θ 2  r  θ1 θ 2 
σ yy = cos  + + θ3  + sin θ sin  + + θ3 
π r3 r1r2  2 2  r3 2 2 
r2  3 3 
− sin θ sin  θ − θ1 − θ 2 − θ3  
r1r2  2 2 

16
P a 2 − b2  r  θ1 θ 2 
σ xy =  − sin θ sin  + + θ3 
π r3 r1r2  r3 2 2 
r2  3 3 
+ sin θ sin  θ − θ1 − θ 2 − θ3  
r1r2  2 2 
Checking the boundary conditions:
(1) As x2 + y2→ ∞ ⇒ σxx = 0, σyy = 0 and σyy = 0
(2) As |x| ≤ a, x ≠ b and y = 0 ⇒ σxy = 0
(3) As |x| ≤ a and y = 0 ⇒ σxy = 0
(4) As |x| ≤ a and y = 0+ and y = 0- ⇒ ∫ σyy dx = -P

It is easy to confirm BCs (1), (2) and (3). For BC (4), it can be proven by improper
integral.
If f (z) has a simple pole at z=b on the real axis, then
lim ∫ f ( z )dz = π i Res f ( z )
r →0 c2 z =b

So we have
a b−r a
∫ σ yy dx = ∫ σ yy dx + lim ∫ σ yy dx + ∫ σ yy dx
−a −a r → 0 c2 b+r

= 0 + π i Res f ( z ) + 0 = − P
z =b

Hence the given Westergaard function is the solution of the given problem.

Calculation of the stress intensity factors:


At right crack tip, we have
r = a, r2 = 2a, r3 = a − b, θ = θ2 = θ3 = 0
σyy is given by
P a2 − b2 θ
σ yy = cos( 1 )
π (a − b) 2ar1 2
The stress intensity factor is
P a+b
K I ,right = lim 2π r1σ yy =
r1 →0,θ1 →0 πa a −b
Similarly at left crack tip, the stress intensity factor is
P a −b
K I ,left = lim 2π r3σ yy =
r2 →0,θ 2 →π πa a +b

17
Prob. 3.6

y y
P z

x r1 r
P θ1 θ x
b
-b
The Westergaard function is given by
P b
ZI =
π (b + Z ) Z
The first derivative is
P b P b
Z I′ = − −
π (b + Z )
2
Z 2π ( b + Z ) z 32

In polar coordinates, the Westergaard function and its derivative can be written as
 θ
− i  θ1 + 
 2
P b P ae
ZI = =
π r1 eiθ1 re iθ
π r1 r
 − i 2θ1 + θ2  − i  θ1 + θ  
 3 
   2 
P b e e 
ZI ' = −  + 
π r r2
2r1r 2 
3
 1 
Hence, we have
P b  θ
Re {Z I } = cos  θ1 + 
π r1 r  2
P b 1  θ 1  3θ  
Re {Z I′ } = −  cos  2θ1 +  + cos  θ1 +  
π r1 r  r1  2  2r  2 
P b 1  θ 1  3θ  
Im {Z I′ } =  sin  2θ1 +  + sin  θ1 +  
π r1 r  r1  2  2r  2 
The whole stress fields can be written as
σ xx = Re{Z I } − y Im{Z I′ }
P b   θ 1  θ 1  3θ   
= cos  θ1 +  − y  sin  2θ1 +  + sin  θ1 +  
π r1 r   2  r1  2  2r  2   
P b   θ  θ 1  3θ  
= cos  θ1 +  − sin θ1 sin  2θ1 +  − sin θ sin  θ1 +  
π r1 r   2  2 2  2 

18
σ yy = Re{Z I } + y Im{Z I′ }
P b   θ  θ 1  3θ 
= cos  θ1 +  + sin θ1 sin  2θ1 +  + sin θ sin  θ1 + 
π r1 r   2  2 2  2 
σ xy = − y Re{Z ′}
P b   θ 1  3θ  
= sin θ1 cos  2θ1 +  + sin θ cos  θ1 +  sin θ 
π r1 r   2 2  2  
Near the crack tip r1 ≈ b, θ ≈ 0, and r/b << 1. Substituting these in the whole field
solutions, we have
P θ θ 3θ 
σ xx = cos 1 − sin sin 
π br 2 2 2
P θ θ 3θ 
σ yy = cos 1 + sin sin 
π br 2 2 2
P θ θ 3θ
σ xy = sin cos cos
π br 2 2 2
Stress intensity factor KI
P 2
K I = lim 2π rσ yy |θ =0 = 2π r =P
r →0 π br πb
The crack surface displacement is given by
(κ + 1) Im Z − y Re Z + 1 κ − 3 Ay
2µu y =
2
I { } { I} (
2
)
Along the crack, y = 0. Also, since our coordinate system has an origin at the crack tip,
the region behind the crack tip has negative x coordinate. Therefore, we can write z = ≈
−|x| along the crack behind the crack tip. Hence, the Westergaard function becomes

P b
ZI =
i π (b − x ) x
and we have
P b 1 1
Z I = ∫ Z I dz = − ∫ d x
iπ a− x x
Consider transformation
1 1
u= x ⇒ du = d x
2 x
So that
iP b 2du iP b  1 1  2du P u+ b P x + b
ZI =
π ∫ b−u 2
=
π ∫  − 
u + b u − b  2 b
= i ln
π u− b
= i ln
π x − b
The crack surface displacement is

19
κ + 1 P  x + b 
2u y =
κ +1

{ }
Im Z I =
2µ π 
ln
x − b


 
where κ = (3− υ) / (1+ υ) for plane stress and µ is shear modulus. The displacement for
the upper and bottom crack surface are symmetric with respect to x axis. Therefore, the
crack surface displacement is
4P x + b
δ = 2u y = ln
πE x − b

20
Prob 3.7:

The stress intensity factors due to the arbitrary crack face loading are
1 a a +ξ
KI = ∫ p (ξ ) d ξ at the right crack tip (x = a )
πa −a a −ξ
1 a a −ξ
p (ξ ) dξ
πa ∫
= at the left crack tip (x = −a )
−a a +ξ
For this problem, we have
p(ξ ) = p − a1 < x < a1
At the right crack tip, the stress intensity factor can be represented as
p a1 a + ξ p a1 1 + ξ a
∫ dξ = dξ
π a ∫− a1 1 − (ξ a )2
KI =
πa − a1 a −ξ
With the new variable η defined as
ξ = a sin η , dξ = a cosη dη
the integral can be evaluated as
a1 1+ ξ a sin −1 ( a1 a ) 1 + sin η
∫−a1 1 − (ξ a)2 d ξ = ∫−sin−1 ( a1 a ) cosη a cosη dη
sin −1 ( a1 a )
= a∫ 1 + sin η dη
− sin −1 ( a1 a )

a1
= 2a sin −1
a
The stress intensity factor at the right crack tip is
2 sin −1 ( a1 a )
KI = p πa
π
At the left crack tip, the stress intensity factor is
a −ξ p a1 1 − (ξ a )
2
p a1
dξ = dξ
π a ∫− a1 a + ξ π a ∫− a1 (1 + ξ a)
KI =

Following the same step shown previously, we can define the new variable η as
ξ = a sin η , dξ = a cosη dη
The integral can be solved as
a1 1 − (ξ a) 2 sin −1 ( a1 a ) cosη
∫ dξ = ∫ −1 a cosη dη
− a1 (1 + ξ a) − sin ( a1 a ) 1 + sin η

sin −1 ( a1 a )
= a∫ 1 − sin η dη
− sin −1 ( a1 a )

a1
= 2a sin −1
a
Thus, the stress intensity factor at the left crack tip is
2 sin −1 ( a1 a )
KI = p πa
π

21
Prob 3.8:

The stress intensity factors due to the arbitrary crack face loading are
1 a a +ξ
KI = ∫ p (ξ ) d ξ at the right crack tip (x = a )
π a −a a −ξ
1 a a −ξ
p (ξ ) dξ
πa ∫
= at the left crack tip (x = −a )
−a a +ξ
For this problem, we have
ξ
p (ξ ) = p 0< x<a
a
ξ
= −p
−a < x <0
a
At the right crack tip, the stress intensity factor can be represented as
p  0 ξ a (1 + ξ a) a ξ a (1 + ξ a ) 
KI = ∫ − dξ + ∫ dξ 
π a  − a 1 − (ξ a) 2 0
1 − (ξ a) 2 

p  − a ξ a (1 + ξ a) a ξ a (1 + ξ a ) 
= ∫ dξ + ∫ dξ 
π a  0 1 − (ξ a) 2 0
1 − (ξ a) 2 
With the new variable η defined as
ξ = a sin η , dξ = a cosη dη
the integral can be evaluated as
ξ a (1 + ξ a) sin η (1 + sin η )
∫ 1 − (ξ a)2 dξ = ∫ cosη a cosη dη
= a ∫ sin η (1 + sin η )dη
η 1 
= a  − cosη − sin 2η  + C
2 4 
where C is the integration constant. The stress intensity factor at the right crack tip is
p 2
KI = a (1 + 1) = p π a
πa π
At the left crack tip, the stress intensity factor is
p  0 ξ a 1 − (ξ a) a ξ a 1 − (ξ a ) 
2 2
KI = ∫ − dξ + ∫ dξ 
π a  − a 1+ ξ a 0 1+ ξ a 

p  − a ξ a 1 − (ξ a) a ξ a 1 − (ξ a ) 
2 2
= ∫ dξ + ∫ dξ 
π a  0 1+ ξ a 0 1+ ξ a 
Following the same step shown previously, we can define the new variable η as
ξ = a sin η , dξ = a cosη dη
The integral can be solved as

22
ξ a 1 − (ξ a ) 2 sin η cosη
∫ 1 + ξ a dξ = ∫ 1 + sin η a cosη dη
= a ∫ sin η (1 − sin η )dη
 η 1 
= a  − − cosη + sin 2η  + C
 2 4 
where C is the integration constant. Thus, the stress intensity factor at the left crack tip is
p 2
KI = a (1 + 1) = p π a
πa π

23
Prob 3.9:

The boundary conditions are


σ xy = σ yy = 0 in − a + 2nb ≤ x ≤ a + 2nb, n = 0, ±1, ±2,... and y = 0
σ xx = σ yy = σ 0 , σ xy = 0 at x 2 + y 2 → ∞
The Westergaard function is
πz πz
σ 0 sin σ 0 sin
ZI = 2b = 2b
πz πa π ( z + a) π ( z − a)
sin 2 − sin 2 sin sin
2b 2b 2b 2b
The first derivative is given by
π πa πz π π a  π ( z + a) π ( z − a) 
σ 0 sin 2 cos σ 0 sin  sin − sin
2b 2b 2b = − 4b 2b  2b 2b 
Z I′ = −
 2 πz πa   π ( z + a) π ( z − a) 
32 32

 sin − sin 2   sin sin


 2b 2b   2b 2b 
Note a complex sine function can be divided into real part and imaginary parts such as
sin z = sin( x + iy ) = sin x cosh y + i cos x sinh y
Thus,
π ( x + iy ) πx πy πx πy
sin = sin cosh + i cos sinh
2b 2b 2b 2b 2b
Let
 πx πy  πx πy
2 2

r =  sin cosh  +  cos sinh 


 2b 2b   2b 2b 
πx πy πx πy
tan θ = cos sinh sin cosh
2b 2b 2b 2b
Similarly,
 π ( x − a) πy  π ( x − a) πy
2 2

r1 =  sin cosh  +  cos sinh 


 2b 2b   2b 2b 
π ( x − a) πy π ( x − a) πy
tan θ1 = cos sinh sin cosh
2b 2b 2b 2b
 π ( x + a) πy  π ( x + a) πy
2 2

r2 =  sin cosh  +  cos sinh 


 2b 2b   2b 2b 
π ( x + a) πy π ( x + a) πy
tan θ 2 = cos sinh sin cosh
2b 2b 2b 2b
So Westergaard function and its first derivatives can be rewritten as
σ r  1 1 
Z I = 0 exp i  θ − θ1 − θ 2 
r1r2  2 2 
π π a r2 exp iθ 2 − r1 exp iθ1  3 
Z I′ = −σ 0 sin 32
exp  −i (θ1 + θ 2 ) 
4b 2b (r1r2 )  2 

24
At y = 0, σxy is equal to zero automatically. The normal stress σyy is obtained as
σ r  1 1 
σ yy = Re{Z I } = 0 cos  θ − θ1 − θ 2 
r1r2  2 2 
If we focus on the central crack surface at right side(0 < x < a), θ can be described as
πx πy πx πy
tan θ = cos sinh sin cosh
2b 2b 2b 2b
where
πx πy πx πy
cos → +, sinh → 0, sin → +, cosh →1⇒θ = 0
2b 2b 2b 2b
Similarly, for the left side,
πx πy πx πy
cos → −, sinh → 0, sin → −, cosh →1⇒ θ = π
2b 2b 2b 2b
Meanwhile, θ1 can be described as
π ( x − a) πy π ( x − a) πy
tan θ1 = cos sinh sin cosh
2b 2b 2b 2b
where
π ( x − a) πy π ( x − a) πy
cos → −, sinh → 0, sin → −, cosh → 1 ⇒ θ1 = π
2b 2b 2b 2b
θ2 can be described as
π ( x + a) πy π ( x + a) πy
tan θ 2 = cos sinh sin cosh
2b 2b 2b 2b
where
π ( x + a) πy π ( x + a) πy
cos → +, sinh → 0, sin → +, cosh → 1 ⇒ θ2 = 0
2b 2b 2b 2b
Furthermore, the value of θ, θ1 and θ2 can be applied for other cracks since the period of
the sine and cosine functions is 2b. Substituting the value of θ, θ1 and θ2 into σyy, we can
have
σ yy = 0 for the arbitrary crack surface
2 2
At x + y →∞, it can be found that
r1 = r2 = r → ∞, θ1 = θ 2 = θ
and that results in
σ xx = σ yy = σ 0 , σ xy = 0
In order to determine the stress intensity factor, we use
πa
σ 0 sin
K I = 2π lim z − aZ I = 2π z − a 2b
z →a π a π ( z − a)
sin
b 2b
πa 2b πa
= σ 0 2b tan = tan σ 0 π a
2b πa 2b

25
Prob 3.10:
The Westergaard function is
σ0z
ZI =
z 2 − a2
The stresses components are given by
σ xx = Re {Z I } − y Im {Z I′ } + 2 A
σ yy = Re {Z I } + y Im {Z I′ }
σ xy = − y Re {Z I′ }
where
σ0
σ xx − σ yy = 2 A ⇒ A =
2
The stress components are
σ r   θ θ  a2 3 
σ xx = o  cos  θ − 1 − 2  − sin θ sin (θ1 + θ 2 )  − σ 0
r1r2   2 2  r1r2 2 
σ or   θ1 θ 2  a 2 3 
σ yy = cos
  θ − − + sin θ sin (θ1 + θ 2 ) 
r1r2   2 2  r1r2 2 
σ or  a2 3 
σ xy =  sin θ cos (θ1 + θ 2 ) 
r1r2  r1r2 2 
Along y = 0, we have θ = θ1 = θ2 = 0, σyy is
σ r σox
σ yy = o = x ≥a
r1r2 x2 − a2
and
σ yy = 0 x < a
For the crack surface opening displacement, we have
{ }
4µu y = (κ + 1) Im Z I
which gives
1
4µu y = (κ + 1)σ 0 r1r2 sin (θ1 + θ 2 )
2
At the upper surface, θ1 = π, θ2 = 0, the displacement along the crack surface is
4µu y = (κ + 1)σ 0 a 2 − x 2

26
Prob. 3.11
From Prob. 3.4, we have
a a +ξ
P0
KI = ∫ d ξ = P0 π a
πa a −ξ
−a

According to the superposition in Prob. 3.4, the total stress ahead of the crack (y=0) can
be expressed as:
( x + a) P0
σ yy = − P0
x ( x + 2a )
Degree of K-dominance is defined as
σ singular
Λ=
σ singular + σ non-singular
The singular and non-singular parts of the stress field are
KI KI
σ singular = , σ non-singular = σ yy − σ singular = σ yy −
2π x 2π x
Therefore the degree of K-dominance is
KI
Λ= 2π x =
1
KI Po ( x + a ) KI 2( x a + 1) x
+ − Po − 1+ − 2 −1
2π x x( x + 2a ) 2π x x a+2 a
From the class notes, the degree of K-dominance for the case of remote tensile case can
be obtained as
x a+2
Λ tensile =
2 ( x a + 1)
A comparison of the degree of K-dominance is shown below.

27
Let us calculate the size of the K-dominance zone by assuming that Λ = 95%. The
distances for each of the cases are shown below:

The pressurized crack: x/a ≈ 0.00145


The remote tensile stress: x/a ≈ 0.07

Clearly for the pressurized crack the K-dominance zone is very small and hence the non-
singular contribution cannot be neglected.

28
Prob 3.12:

a) Center crack with remote tensile stress P0


The crack opening stress is
P0 a
σ yy =
x ( x + 2a )
Using William’s eigen-function expansion, we obtain b0 as
b0 = ( d
dx
σ yy x) x =0
=
3P0
4 2a
b) Central crack with internal pressure P0
Using William’s eigen-function expansion, we can write the crack opening stress as
P0 a KI 3P0
σ yy = − P0 = − P0 + x + ......
x ( x + 2a ) 2π x 4 2a
It can be found that bo is the same for two cases.

29
Prob. 3.13
Check the accuracy of the formula given by equation (3.82) by FEA or the boundary
collection method.
The specimen setup to test Eq. (3.83) in the class notes is represented as

where σ = 100MPa, 2H = 400mm, 2b = 300mm, 2a = 30, 60, 120, 180, 240mm. The
material properties are E = 300GPa, ν = 0.3. The meshes around the crack tip for 2a =
120mm (element size is equal to 0.4mm) is shown as

Extract the σyy distribution along the crack line (red dots) and plot σyy√2 vs. x to
determine the stress intensity factors

30
The stress intensity factors with respect to different ratios between the crack length (2a)
and the width (2b) can be obtained by the projection method, listed in the following table.

a/b Crack length (mm) KI (MPa√m)


0.1 30 21.6298
0.2 60 31.2225
0.4 120 48.2002
0.6 180 69.9048
0.8 240 110.8697

The comparison with Eq.(3.83) is shown as

Based on the figure above, it can be found that Eq.(3.83) can provide satisfied results if
a/b is below 0.6.

31
Prob. 3.14
Consider Problem 3.4. Replace the internal pressure with a pair of concentrated forces
applied in opposite directions at the midpoint of the crack. Assume that these forces are
applied to open the crack surfaces. Compare the K-dominance zone size of this loading
with that of the remote stress loading case. If these two loading conditions are used to
measure the fracture toughness of the material, which specimen would yield a higher
toughness value? Explain.

The stress intensity factor for the case with a pair of concentrated forces applied in
opposite directions at the midpoint of the crack can be written as
P
KI =
πa
According to the Westergaard function, the total stress ahead of the crack (y=0) is
Pa
σ yy =
π ( x + a ) x ( x + 2a )
Degree of K-dominance is defined as
σ singular
Λ=
σ singular + σ non-singular
The singular and non-singular parts of the stress field are
KI KI
σ singular = , σ non-singular = σ yy − σ singular = σ yy −
2π x 2π x
Therefore, the degree of K-dominance is
KI
Λ= 2π x =
1
KI Pa KI 2
+ − 1+ −1
2π x π ( x + a ) x( x + 2a ) 2π x ( x a + 1) x a + 2
From the class notes, the degree of K-dominance for the case of remote tensile case can
be obtained as
x a+2
Λ tensile =
2 ( x a + 1)

32
A comparison of the degree of K-dominance is shown below.

The distance ahead of the crack tip corresponding to Λ = 0.95 can be calculated as:

The point force: x/a ≈ 0.042


The remote tensile stress: x/a ≈ 0.07

The smaller K-dominance zone means the influence of non-singular part is larger and
vice versa.

It is noted that the total fracture criterion is contributed by both singular (apparent
fracture toughness) and the non-singular part. Theoretically, the total fracture criterion
can be treated as a material property (a constant). For the case with extremely large K-
dominance zone, the contribution from the non-singular part can be neglected and the
total fracture criterion is equal to the apparent fracture toughness. Once the non-singular
contribution cannot be neglected, the apparent fracture toughness decreases or increases
dependent on whether the contribution of the non-singular stress is positive or negative.
Consider σyy(2πx)0.5 / KI whose slope is the first non-singular term as following

33
where the blue line is for the remote tensile case
σ o ( x + a) 2π x
σ yy 2π x x( x + a) x  2
= =  + 1
KI σo π a a  x a+2
and the green line is for the concentrated force case
Pa 2π x
σ yy 2π x π ( x + a) x( x + 2a) 1 2
= =
KI P ( x a + 1) x a + 2
πa
As we can see, the slopes of two distributions are opposite, which represents the
contribution of the non-singular stress is positive for the remote tensile case but negative
for the concentrated force case. Hence, the apparent fracture toughness tends to decrease
for the remote tensile case but increase for the concentrated force case.

Chapter 4

Prob. 4.1:

34
(a) J-integral
σo

Γ4
Γ5 y
h
x Γ3

Γ1 h
Γ2
σo

a b
J-integral is given by
 ∂u 
J = ∫  Wdy − Ti i d Γ 
Γ
 ∂xi 
where Γ = Γ1+ Γ2+ Γ3+ Γ4+ Γ5.
From the symmetry of the problem, we have
J = 2( ∫ + ∫ )+∫
Γ1 Γ2 Γ3

Along Γ1: W = 0, n1 = -1, T1 = 0, T2 = 0


Along Γ2: n1 = 0, T1 = 0, T2 = σo
Along Γ3: n1 = 0, T1 = 0, T2 = 0

Therefore, we can get


For Γ1:
 ∂ui 
∫Γ1 
 Wdy − Ti
∂xi
d Γ1 =0

For Γ2:
 ∂u  ∂u b ∂u
∫  Wdy − Ti i d Γ 2  = ∫ −Ty y d Γ 2 = ∫ σ o y dx = σ o u y (b) − u y (− a ) 
Γ2
 ∂xi  Γ2 ∂x −a ∂x
Based on simple beam theory, we obtain
σ oa4
u y (−a) = −
8 EI
Also considering simple tension, we obtain

σ oh
u y (b) = −
E
Hence,

35
 ∂ui   σ oa4 σ oh 
∫Γ2  Wdy − T d Γ 2 = σ o − 
∂xi
i
  8 EI E 
For Γ3:
The strain energy density is
1 σ2
W = σ xxε xx = o
2 2E
Therefore
 ∂ui  σ o2 h σ o2 h
∫Γ3  Wdy − Ti
∂xi
d Γ 3 = ∫
 2E − h
dy =
E
The final solution is
σ o2 a 4 σ o2 h
G = J = 2( ∫ +∫ )+∫ = −
Γ1 Γ2 Γ3 4 EI E
(b) Compliance method

σo

(1) y h
x (3)

(2) h

σo

a b
The strain energy stored in section (1) and (2) is
a M ( x)
2
σ o2 a 5
U1 = U 2 = ∫ dx =
0 2 EI 40 EI
where I = h3/12.
The strain energy stored in section (3) is
2 h P( x)
2
(σ b) 2 (2h) σ o2 h
U3 = ∫ dx = o = ( L − a)
0 2 EA 2 Eb E
The total energy is U = U1+U2+U3. The energy release rate can be calculated by
∂U σ o2 a 4 σ o2 h
G= = −
∂a 4 EI E
The energy release rates obtained from two methods are identical.

Prob. 4.2:
Given in Prob. 4.1, we have

36
σ o2 a 4 σ o2 h
GI = −
4 EI E
Let the thickness of beam to be unity,
3σ 2 a 4 σ 2 h 3(106 ) 2 (0.1) 4 (106 )2 0.01
GI = o 3 − o = − = 17142.29 N/m
Eh E 70 × 109 (0.01)3 70 × 109
In FEM, the meshes around crack tip are shown as

The symmetric condition is applied along the crack line and a quadratic element size is
0.00833 × 0.01 mm2.

(a) Modified crack closure method


The reaction forces and displacements are

The corresponding energy release rate is


1
GI = 54598(2 × 2.2749 × 10−6 ) + 33948(2 ×1.4427 × 10−6 )  = 20781.82 N/m
−6 
2(8.33 ×10 )

(b) Direct projection method


The crack opening stress is shown as

37
KI = 39.0105 MPa√m

Through the G-K relation in plane-strain condition, we can get


1 −ν 2 2 1 − 0.252
GI = KI = (39.0105 ×106 ) 2 = 20381.51 N/m
E 70 × 109
Two results from the FEM solution show agreement but don’t satisfy with the analytic
solution in Prob. 4.1. The ratios between the analytical and FEM solutions against
different crack lengths varied from 5 cm to 50 cm are shown as

It can be found that the assumption of double-cantilever beam becomes valid as the crack
length increases.

38
Prob 4.3:
2P

y
h
x

L
The split beam with a vertical load can be considered as a superposition of the following
two cases:
P P

Γ4 Γ4
Γ5 (1) y
Γ5 (1) y

x (3) Γ3 x (3) Γ3

Γ1 (2) Γ1 (2)
Γ2 Γ2

P P
(a) Case 1 (b) Case 2

The symmetric loading structure in case 1 gives GI and the anti-symmetric loading
structure in case 2 gives GII. The moment of inertia I1 = I2 = h3/12, I3 = 2h3/3.

(a) GI for Case 1


J-integral:
J-integral is given by
 ∂u 
J = ∫  Wdy − Ti i d Γ 
Γ
 ∂xi 
where Γ = Γ1+ Γ2+ Γ3+ Γ4+ Γ5.

Along Γ1 and Γ5: W = 0


Along Γ2 and Γ4: dy = 0, Ti = 0
Along Γ3: W = 0, Ti = 0

Hence,

39
J =∫ +∫
Γ1 Γ5

For Γ1 and Γ5:


 ∂ui  −h ∂u y 0 ∂u y h ∂u y  h ∂u y 
∫Γ1 +Γ5  ∂xi  ∫0 ∂x
 −Ti d Γ 2  = Ty dy + ∫h ∂x
Ty dy = − ∫− h ∂x
Ty dy = − 2  ∫0 Ty
 ∂x
dy 

From section (1), we have
∂u y Pa 2 6 Pa 2
=− =−
∂x x at free end 2 EI1 Eh3
It is a constant with respect to y. So the J-integral is calculated as
 h ∂u  12 Pa 2 h 12 P 2 a 2
GI = J = −2  ∫ Ty y dy  =
Eh 3 ∫0
T dy =
∂x
y
 0  Eh3

Compliance method:
The strain energy stored in section (1) ,(2) and (3) is
2
a ( Px ) P 2 a3
U1 = U 2 = ∫ dx = , U3 = 0
0 2 EI 6 EI1
1
The total energy is U = U1+U2+U3. The energy release rate can be calculated by
∂U P 2 a 2 12 P 2 a 2
GI = = =
∂a EI1 Eh3

(b) GII for Case 2


J-integral:
J-integral is given by
 ∂u 
J = ∫  Wdy − Ti i d Γ 
Γ
 ∂xi 
where Γ = Γ1+ Γ2+ Γ3+ Γ4+ Γ5.

Along Γ1 and Γ5: W = 0


Along Γ2 and Γ4: dy = 0, Ti = 0
Along Γ3: uy / x = 0, Tx = σxx

Hence,
J =∫ +∫ +∫
Γ1 Γ5 Γ3

For Γ1:
∂u y ∂u y ∂u y Pa 2  2 P( L − a ) 2 2 Pa ( L − a ) 
= + =− − + 
∂x x at free end
∂x sec 2
∂x sec 3
2 EI1  2 EI 3 EI 3 
Pa 2
P( L − a 2 ) 2
6 Pa 2 3P( L2 − a 2 ) 3P ( L2 + 3a 2 )
=− − =− − = −
2 EI1 EI 3 Eh3 2 Eh3 2 Eh3
So

40
 ∂u y  3P( L2 + 3a 2 ) − h 3P 2 ( L2 + 3a 2 )
∫Γ1  y ∂xi 1 
−T d Γ =
2 Eh3 ∫0 y T dy =
2 Eh3
For Γ5:
 ∂u y  3P( L2 + 3a 2 ) 0 3P 2 ( L2 + 3a 2 )
∫Γ5  y ∂xi 5 
−T d Γ =
2 Eh3 ∫h y
T dy =
2 Eh3
For Γ3:
 ∂u x   σ xx2  h σ xx2 h  1   2 PLy  3P 2 L2
∫Γ3  W − T  dy = − σ ε
∫Γ3  2 E xx xx  ∫−h 2 E
dy = − dy = ∫− h  2 E   I3 
− dy = −
∂xi 
x
Eh 3
The final solution is
9P2a 2
GII = J = ∫ + ∫ + ∫ =
Γ1 Γ5 Γ3 Eh3

Compliance method:
The strain energy stored in section (1) ,(2) and (3) is
2
a ( Px ) P2a3 2P2a3
U1 = U 2 = ∫ dx = =
0 2 EI
1 6 EI1 Eh3
(2 Px) 2L P2
U3 = ∫ dx = 3 ( L3 − a 3 )
a 2 EI Eh
3
The total energy is U = U1+U2+U3. The energy release rate can be calculated by
∂U 9 P 2 a 2
GII = =
∂a Eh3

41
Prob 4.4:

J-integral is given by
−ε ∂u y
J = 2 lim ∫ σ yy dx
ε →∞ ε ∂x y =δ → 0 +

The near tip stress and displacement field are


KI θ θ 3θ 
σ yy = cos 1 + sin sin 
2π r 2 2 2
K  θ 3θ 
u y = I 2π r ( 2κ + 1) sin − sin 
8µπ  2 2
By chain rule
∂u y ∂u y ∂r ∂u y ∂θ
= +
∂x ∂r ∂x ∂θ ∂x
By x=rcosθ, y=rsinθ, we have
∂r ∂θ sin θ
= cos θ , =−
∂x ∂x r

Hence,
∂u y 2π  θ 3θ
( 2κ + 1) sin − sin  cos θ
KI
= 
∂x 16 µπ r  2 2
KI  θ 3θ   sin θ 
+ 2π r ( 2κ + 1) cos − 3cos   − 
16 µπ  2 2  r 
KI 2π  θ 3θ 
=  − ( 2κ + 2 ) sin + 2 sin θ cos
16 µπ r 2 2 
The J-integral is given by

42
−ε ∂u y
J = 2 lim ∫ σ yy dx
ε →∞ ε ∂x y = δ → 0+

π K I2  θ 3θ   θ 3θ 
= 2∫ 1 + sin sin  (κ + 1) − 2 cos cos  dθ
0 8µπ
 2 2  2 2
The above integration can be estimated by using computer software package. We
therefore can obtain
κ +1 2
J= KI

For plane stress condition,
3 −ν E
κ= ,µ=
1 +ν 2(1 + ν )
The energy release rate for mode I is
K2
GI = J = I
E

43
Prob. 4.5

44
Prob. 4.6:
(a) Direct projection method – Crack opening stress
The crack opening stress is shown as

KI = 1.111 MPa√m

Through the G-K relation in plane-strain condition, we can get


1 −ν 2 2 1 − 0.252
GI = KI = (1.111×106 ) 2 = 16.53 N/m
E 70 × 10 9

(b) Direct projection method – Crack opening displacement


The crack opening displacement is shown as

KI = 1.112 MPa√m

The corresponding energy release rate is


1 −ν 2 2 1 − 0.252
GI = KI = (1.112 × 106 ) 2 = 16.58 N/m
E 70 ×10 9

45
(c) Modified crack closure method

The energy release rate is calculated as


1
GI = 1509(2 × 6.4824 × 10−8 ) + 1003(2 × 4.1996 × 10 −8 )  = 16.79 N/m
2(8.33 × 10−6 ) 
The corresponding stress intensity factor is
E 70 × 109
KI = GI = (16.79) = 1.119 MPa m
1 −ν 2 1 − 0.252

46
Chapter 5

5.1 Prove that Eq. (5.10) can be obtained from dσθθ/dθ = 0.

It follows from the first equation in Eq. (5.9) that

dσ θθ 1  1 θ 2θ 3  θ θ θ 3 
= − sin  K I cos − K II sin θ  + cos − K I cos sin − K II cos θ  
dθ 2πr  2 2 2 2  2 2 2 2 
1  1 θ 2θ 3  θ 1 3 
= − sin  K I cos − K II sin θ  + cos − K I sin θ − K II cos θ  
2πr  2 2 2 2  2 2 2 
1 θ 1 θ θ 3 θ  1 3 
= cos − sin  K I cos − K II 2 sin  + − K I sin θ − K II cos θ  
2πr 2 2 2 2 2 2  2 2 
1 θ 1 3 θ 1 3 
= cos − K I sin θ + K II sin 2 − K I sin θ − K II cos θ 
2πr 2 4 2 2 2 2 
1 3 θ 1 θ 
= cos − K I sin θ + K II sin 2 − K II cos θ 
2πr 2 2 2 2 
1 3 θ 1  1 − cos θ 
= cos − K I sin θ − K II  cos θ − 
2πr 2 2 2  2 
1 3 θ 1 3 cos θ − 1
= cos − K I sin θ − K II 
2πr 2 2 2 2 

Hence, dσ θθ / dθ = 0 leads to

θ
cos {− K I sin θ − K II (3 cosθ − 1)} = 0
2

which is Eq. (5.10)

Prob 5.2:

Plot σθθ(2πr)0.5/KI versus θ for KII/KI = 0.5 and 2.0, respectively, to verify that Eq. (5.11)
gives the orientation at which σθθ reaches the maximum value.

Based on the Maximum tensile stress criterion, the failure angle can be determined by
K I sin θ + K II (3cos θ − 1) = 0
Rearrange the equation above, we can get

47
K II sin θ
=
K I 3cos θ − 1
Consider KII/KI = 0.5 or 2.0 and use MATLAB to obtain the numerical solutions
θ = −40.2 when KII/KI = 0.5
θ = −61.4 when KII/KI = 2.0
The near tip solution is
σ θθ 1 θ θ 3 K II 
= cos cos 2 − sin θ 
KI 2π r 2 2 2 KI 
When KII/KI = 0.5

θ = −40.2

When KII/KI = 2.0

θ = −61.4

48
5.4 Show that the angle determined by kII = 0 is along the maximum energy release rate
direction.

It follows from Eq. (5.21) that α satisfies the following equation when k II = 0

C21 (α ) K I + C22 (α ) K II = 0
C (α )
or K II = − 21 KI (1)
C22 (α )

Substituting the energy release rate in Eq. (5.25) into Eq. (5.26), we have

 dC dC  2  dC dC  2
 2C11 11 + 2C21 21  K I +  2C12 12 + 2C22 22  K II
 dα dα   dα dα 
 dC dC dC dC 
+ 2 C11 12 + C12 11 + C21 22 + C22 21  K I K II = 0
 dα dα dα dα 

Substituting (1) into the equation above yields

dC11 dC  dC dC  C 2
C11 + C21 21 +  C12 12 + C22 22  21
dα dα  dα dα  C222
 dC dC dC dC  C
−  C11 12 + C12 11 + C21 22 + C22 21  21 = 0
 dα dα dα dα  C22
or
 dC dC C   dC dC C 
 C11 11 − C12 11 21  +  C21 21 − C22 21 21 
 dα dα C22   dα dα C22 
 dC C 2 dC12 C21   dC22 C212 dC C 
+  C12 12 21 − C 
 + 
 C − C21 22 21  = 0
dα C22 dα C22   dα C22 dα C22 
2 11 22 2

or
 C11C22 − C12C21  dC11  C12C21 − C11C22  dC
  +  C21 12 = 0
 dα  dα
2
 C22 C22 
or
 
(C11C22 − C12C21 ) dC11 − C21 dC12  = 0
 dα C22 dα 

Use of Eq. (5.24) yields (C22 dC11 / dα − C21dC12 / dα ) = 0, which means k II = 0 and
dGkink / dα = 0 yield the same angle. Hence, the angle determined by kII = 0 is along the
maximum energy release rate direction (it can be verified that k II = 0 does not
correspond to the minimum energy release rate).

49
Prob 5.5:

1. Maximum Stress Criterion

The near tip solution is


1 θ θ 3 
σ θθ = cos  K I cos2 − K II sin θ 
2π r 2 2 2 
1 θ θ 1 
σ rθ = cos  K I sin + K II (3cos θ − 1) 
2π r 2 2 2 
Maximum σθθ is at σrθ = 0, so
K I sin θ + K II (3cos θ − 1) = 0
Substitute KI = σo(πa) sin β, KII = σo(πa)0.5sin βcos β into the equation above, we can get
0.5 2

KI 3cos θ − 1
= tan β =
K II sin θ
Failure occurs when
K
(σ θθ )max = Ic
2π r
Substitute it into the σθθ distribution and let KI = σo(πa)0.5sin2β, KII = σo(πa)0.5sinβcosβ, we
can obtain
K 1
σ ult = Ic
π a cos θ (sin 2 β cos 2 θ − 3 sin β cos β sin θ )
2 2 2
where KIc = σ90o(πa)0.5

2. Minimum Strain Energy Density


The strain energy density factor is
S = a11K I + 2a12 K I K II + a22 K II
where a11, a12, a22 are function of θ (Eq. (5.14) in class notes). Rearrange the equation
above, we can get
2
S  KI  KI
2
= a11   + 2a12 + a22
K II K
 II  K II

Substitute KI = σo(πa)0.5sin2 β, KII = σo(πa)0.5sin βcos β into the equation above and use
MATLAB to find the θ where the equation above to reach the minimum value.
Failure occurs when
S = Sc = a11 K I + 2a12 K I K II + a22 K II
⇒ σ ult
2
a(a11 sin 2 β + 2a12 sin β cos β + a22 cos 2 β ) sin 2 β = Sc
If we take

50
1 − 2ν 2
Sc = K Ic
4 µπ
We can get
1
K 1 − 2ν  1 2
σ ult = Ic  
π a 4µπ  (a11 sin 2 β + 2a12 sin β cos β + a22 cos 2 β )sin 2 β 
where KIc = σ90o(πa)0.5

3. Maximum Energy Release Rate


The energy release rate at the kink tip is
κ +1 2 2
G= (k I + k II )

where
k I = C11 (α ) K I + C12 (α ) K II
k II = C21 (α ) K I + C22 (α ) K II
In class notes, there are three different kinds of expressions introduced by Hussain et al.,
Nuismer, Cotterell and Rice.
Rearrange the equation above, we can get
G κ + 1   
2 2
KI   KI
=  C11 + C12  +  C21 + C22  
K II2 8 µ  K II   K II  

Substitute KI = σo(πa)0.5sin2 β, KII = σo(πa)0.5sin βcos β into the equation above and use
MATLAB to find the α where the equation above to reach the maximum value.
Failure occurs when
κ +1
G = GIc = (C11 sin 2 β + C12 sin β cos β ) 2 + (C21 sin 2 β + C22 sin β cos β ) 2 

If we take
κ +1 2
GIc = K Ic

We can get
1
K  1 2
σ ult = Ic  2
π a  (C11 sin β + C12 sin β cos β ) + (C21 sin β + C22 sin β cos β ) 
2 2 2

where KIc = σ90o(πa)0.5

The failure angle and failure load with respect to different oblique crack angles are
plotted as the followings

51
52
For min. strain energy density criterion, only plane strain condition is considered in these
two plots. It can be found that the results of max. tensile stress criterion and max. energy
release rate criterions (Nuismer, Cotterell and Rice) are nearly identical, since these three
methods are based on the local mode I condition.

53
Chapter 6

Prob 6.1:

Irwin’s model:
Since the size of plastic zone is known, the stress intensity factor can be obtained as
0.002 K I2
rp = = ⇒ K I = 39.6 MPa m
2 2π (σ )y
* 2

For the DCB specimen, the stain energy release rate is given by
12 P 2 aeff 2
G=
Et 2 h 3
where aeff = a + rp = 0.1 + 0.001 = 0.101 m.
By the G-K relation, the stress intensity factor can be calculated as
12 P 2 aeff
2

K = GE = = 39.6 MPa m
t 2 h3
Hence, P = 226.5 N.

Dugdale’s model:
Here we assume small scale yielding, i.e., KI (a +Ro) ≈ KI (a).
The plastic zone size is given by
2
π K  8 Ro
Ro =  *I  ⇒ K I = σ *y

8 σy  π
The stress intensity factor can also be calculated from
12P 2 a 2
K = GE =
t 2 h3
Hence the load according to the Dugdale model is, P = 206.0 N.

Elastic solution
Based on the Von Mises (plane stress) and Tresca, the size of plastic zone is
2
1  KI 
Rp =  
2π σy 
Following the procedures listed above, the load is, P = 323.6 N.

54
Prob 6.2:

The stress intensity factors for the above problem are given by:

 P a+b
 , right tip
 πa a−b
KI = 
 P a −b
 , left tip
πa a+b
Irwin’s model
Let aeff = a + rp and the plastic zone size can be estimated as

 P2  aeff +b
   , right tip
( ) − b 
2
KI 2 
 2π 2
aeff σ *
y  aeff
rp = =
2π (σ *y ) 
2
P2  aeff +b
 2   , left tip
 2π aeff (σ y )  eff
* 2
a − b 
Iteration is necessary to get the converged solution.

Dugdale model for small scale yielding


Using the Dugdale model, we can split the crack problem with a plastic zone into two
problems as shown below.

σy
P P

= +

P P σy

55
(1) (2)
In order to remove the singularity, we need
K I + K I(2) = 0
We have the stress intensity factor for (2) as
2 Ro
K I(2) = −2σ *y
π
Therefore using equations (4) and (5) we have the plastic zone size as
 a + b  P 2
   , Right tip
 8a( a − b)  σ *y 
Ro =  2
 a −b  P 
  *  , Left tip
 8a( a − b)  σ y 

Dugdale model for large scale yielding


In the case of large scale yielding, we have KI (a +Ro) ≠ KI (a). Simultaneously, we still
assume (for simplicity) that the plastic zone size is the same at both the crack tips.
However, a more rigorous approach would be to let the plastic zone size to be different at
the left and right crack tips.

We will split the problem into three cases as shown below

σy σy
P P

= + +

P P

(1) (2) (3)


We have
P a + Ro + b
K I(1) =
π a a + Ro − b
For (2) and (3), the stress intensity factors are equal to
−σ y a + Ro a + Ro + η
K I(2) = ∫ dη
π ( a + Ro ) η = 0 a + Ro − η
−σ y a + Ro a + Ro − η
K I(3) = ∫ dη
π (a + R ) ηo
=0 a + Ro + η

56
For the right tip, in order to remove the singularity,
K I(1) + K I(2) = 0
Substitute the stress intensity factors into the equation above, we have
P a + Ro + b a + Ro  a 
= 2σ *y cos −1  
π a a + Ro − b π  a + Ro 
Therefore the plastic zone size is given by
  P a+ R +b 1 
sec  * o
 − 1, Right tip
Ro   2σ y a + Ro − b a a + Ro 
=
a   P a+ R −b 1 
sec  2σ * a + R + b
o
 − 1, Left tip
  y o a a + Ro 

57
Chapter 7

Prob 7.1:

The net section stress can be evaluated as


σ (2W ) = σ net (2W − 2a) ⇒ σ = σ net (1 − a W )
Considering σnet = σyield = 400 MPa, we can have the applied stress at failure as

Based on Irwin’s plastic zone adjustment method, the fracture toughness Kc (plane stress)
can be obtained as
σ πa a
Kc = F 
1 − 0.5(σ σ Y ) 2  W 
400 π a (1 − a W )  a 
2 3
a a
= 1 + 0.128   − 0.288   + 1.525   
1 − 0.5(1 − a W ) 2  W  W   W  
Setting a = 10 cm, we can have the fracture toughness Kc as a function of a/W like

It can be found obviously that Kc is specimen size dependent.

58
Prob 7.2:
Based on Irwin’s plastic zone adjustment method, the stress intensity factor KI can be
obtained as
 a + ∆aeff 
K I = σ π (aeff + ∆aeff ) F  eff 
 W 
  a   a 
2
 a
3
 a  
4

F ( a W ) = 1.12 − 0.231  + 10.55   − 21.72   + 30.39   


 W  W  W   W  
where the effective crack length aeff and effective crack extension ∆aeff are
a ∆a
aeff = , ∆aeff =
1 − 0.6289(σ σ Y ) 2
1 − 0.6289(σ σ Y ) 2
for plane stress. Considering the yield stress equal to 40 Ksi (from Alcoa) and varying the
applied stress σ, we can show the comparison between KI as a function of ∆aeff and the
resistance curve in Figure 7.6

It shows the critical applied stress σc = 14.4 Ksi = 99.3 MPa.

59
Chapter 8

-----------------------------------------------------------------
Prob. 8.1: Calculate the two Dundurs’ parameters for the following two pairs of
materials: (1) Glass/epoxy with E1 = 70 GPa, ν1 = 0.2 (glass) and E2 = 2 GPa, ν2 = 0.4
(epoxy); (2) Al2O3/aluminum with E1 = 350 GPa, ν1 =0.25 (Al2O3) and E2 = 70 GPa, ν2 =
0.33 (aluminum).

Solution

(1) Glass/epoxy interface

The shear moduli are:

E1 70 E2 2
µ1 = = = 29.17 GPa, µ 2 = = = 0.7143 GPa
2(1 + ν 1 ) 2(1 + 0.2) 2(1 + ν 2 ) 2(1 + 0.4)

The parameters κ1 and κ2 in Eq. (8.22) are

Plane strain: κ1 = 3 − 4ν 1 = 3 − 4 × 0.2 = 2.2 , κ 2 = 3 − 4ν 2 = 3 − 4 × 0.4 = 1.4


3 − ν 1 3 − 0 .2 3 − ν 2 3 − 0 .4
Plane stress: κ1 = = = 2.333 , κ 2 = = = 1.857
1 + ν 1 1 + 0 .2 1 + ν 2 1 + 0 .4

Using Eq. (8.22), the Dundurs’ parameters can be calculated as

Plane strain: α = 0.9368 , β = 0.1495


Plane stress: α = 0.9444 , β = 0.2806

(2) Al2O3/aluminum interface

The shear moduli are:

E1 350 E2 70
µ1 = = = 140 GPa, µ 2 = = = 26.32 GPa
2(1 + ν 1 ) 2(1 + 0.25) 2(1 + ν 2 ) 2(1 + 0.33)

The parameters κ1 and κ2 in Eq. (8.22) are

Plane strain: κ1 = 3 − 4ν 1 = 3 − 4 × 0.25 = 2 , κ 2 = 3 − 4ν 2 = 3 − 4 × 0.33 = 1.68


3 − ν 1 3 − 0.25 3 − ν 2 3 − 0.33
Plane stress: κ1 = = = 2 .2 , κ 2 = = = 2.008
1 + ν 1 1 + 0.25 1 + ν 2 1 + 0.33

Using Eq. (8.22), the Dundurs’ parameters can be calculated as

60
Plane strain: α = 0.6523 , β = 0.1517
Plane stress: α = 0.6667 , β = 0.2167

---------------------------------------------------------------

61
Prob. 8.2: Derive the stress intensity factors for a semi-infinite interface crack subjected
to uniform pressure p along part of the crack faces as shown in Figure 8.12.

Solution

Based on Rice and Sihs’ definition, the stress intensity factors for a concentrated load
acting at a distance x are given by
1 2
dk1 =  Pdx cos ( ε ln x ) 
π x
2 1
 − Pdx sin ( ε ln x ) 
dk2 =
π x
The complex form of the stress intensity factor is
dk = dk1 − idk2
Pdx 2
exp {iε ln x}
=
π x
The stress intensity factor for the distributed pressure loading can be calculated as
k = k1 − ik2
P 2
exp {iε ln x} dx
b
=∫
0 π x

=
2 2bP
{ }
cos ( ε ln b ) + 2ε sin ( ε ln b )  − i  2ε cos ( ε ln b ) − sin ( ε ln b ) 
π (1 + 4ε 2 ) 
Therefore, the stress intensity factors for an interface crack loaded with crack face
pressure are:
2 2bP
k1 = cos ( ε ln b ) + 2ε sin ( ε ln b ) 
π (1 + 4ε 2 ) 
2 2bP
k2 =  2ε cos ( ε ln b ) − sin ( ε ln b ) 
π (1 + 4ε 2 ) 
If you want to use Sun and Jihs’ definition, it should be noticed that the stress distribution
in Eq.(8.36) involves (2a)-iε term. Hence, taking superposition with respect to Eq.(8.39)
directly is not appropriate.

---------------------------------------------------------

62
Prob. 8.3: Derive the relationship between the energy release rate and the stress intensity
factor for Mode III interface cracks.

Solution

The crack tip displacement and stress fields in Eqs. (8.20) and (8.21) are first copied here

θ µ µ θ
w1 = r A( 0) sin , w2 = 1 w1 = 1 r A( 0 ) sin
2 µ2 µ2 2

(σ θz )1 = (σ θz )2 = 1 µ1 A(0) r −1 / 2 cos θ
2 2

Because no stress oscillation exists for the Mode III problem, the Mode III stress
intensity factor can be defined as

K III = limr → 0 2πr (σ θz (r ,0) )1 = limr →0 2πr


1
µ1 A( 0) r −1 / 2
2

The constant A(0) is related to the stress intensity factor by

2
A( 0 ) = K III
µ1 2π

The crack tip fields (8.20) and (8.21) can thus be rewritten as

1 2r θ µ 1 2r θ
w1 = K III sin , w2 = 1 w1 = K III sin
µ1 π 2 µ2 µ2 π 2

(σ θz )1 = (σ θz )2 = 1 K III cos θ
2πr 2

According to the definition of the energy release rate

∆a
1
σ θz ( x,0)∆w(∆a − x, π )dx
2∆a ∫0
G3 = lim ∆a → 0

Substituting the crack tip fields into the equation above, we have

∆a
1 1  1 2(∆a − x) 1 2(∆a − x) 
G3 = lim ∆a → 0 ∫0 2πx  µ1
2∆a
K III 
π
K III +
µ2 π
K III  dx

∆a
K 2
1 1  µ 
= III lim ∆a → 0
2πµ1 ∫
∆a 0 x   ∆a − x + 1 ∆a − x  dx
µ2 

63
2
K III  µ  1 ∆aπ
= 1 + 1  lim ∆a → 0
2πµ1  µ 2  ∆a 2
1 1 1 
=  +  K III2
4  µ1 µ 2 

64
Chapter 9

--------------------------------------------------------------
Prob.9.1: Prove that the cohesive energy densities for the trapezoidal model Eq. (9.13)
and the exponential model Eq. (9.15) are given by Eq. (9.14) and (9.16), respectively.

Solution

The cohesive energy density is the area under the cohesive traction – separation curve.
For the trapezoidal model in Eq. (9.13) and Fig 9.6(c),
δc
Γc = ∫ σ (δ )dδ = σ cδ1 + σ c (δ 2 − δ1 ) + σ c (δ c − δ 2 )
1 1
0
2 2

= σ c [δ1 + 2(δ 2 − δ1 ) + (δ c − δ 2 )] = σ c (δ c + δ 2 − δ1 )
1 1
2 2

For the exponential model in Eq. (9.15) and Fig. 9.6(d),


∞ ∞ ∞ ∞
δ δ δ 
Γc = ∫ σ (δ )dδ =σ c ∫ e1−δ / δ dδ =eσ cδ 0 ∫ e −δ / δ
0 0
d   = eσ cδ 0 ∫ xe − x dx
0
δ
0 0
δ
0 0  δ0  0

 ∞ −x 
= eσ cδ 0  ∫ xe dx = 1
 
0 

----------------------------------------------------------------

65
Prob. 9.2: Find the relation between the critical separation displacements δc in the linear
softening model Eq. (9.11) and δ0 in the exponential model Eq. (9.15), assuming identical
peak cohesive traction σc and identical cohesive energy density Γc in the two models.

Solution

Based on the assumptions that the cohesive energy densities and the peak tractions are the
same for the two models, we have
1
σ cδ c = eσ cδ 0
2
Hence,
δc
δ0 =
2e
In ductile fracture applications, the cohesive energy density and peak cohesive traction
are more important parameters than the shape of the cohesive curve, and the separation
parameters (such as δ0) may be selected based on the equality of cohesive energy density
and peak traction between the different models.

---------------------------------------------------------------------

66
Prob.9.3: Derive the integral expression of the cohesive energy density Eq. (9.31) from
Eq. (9.30). Assume that the opening, sliding, and effective separations reach their critical
values δ nc , δ sc , and δ effc simultaneously.

Solution

Using Eq. (9.29), Eq. (30) can be written as


δ nc δ sc δ nc
δ
σ σ σ
c

Γc = ∫ σ n dδ n + ∫ σ s dδ s = ∫ eff δ n dδ n + ∫η 2 eff δ s dδ s = ∫ eff (δ n dδ n + η 2δ s dδ s )


s

0 0
δ
0 eff 0
δ eff δ eff
From the definition of δ eff in eq. (9.27), we have
δ ndδ n+η 2δ sdδ s δ ndδ n+η 2δ sdδ s
dδ eff = =
δ n2 + η 2δ s2 δ eff
Hence,
δ eff
c
δ eff
c

σ eff
∫ δ dδ = σ dδ
δ eff eff eff ∫0 eff eff
Γc =
0

67

You might also like