You are on page 1of 39

Principle of Superposition

For a linear elastic system, the individual components of stress,


strain, displacement are additive. For example, two normal stress
in x direction caused by two different external load can be added
to get total stress
P M
x   y
A I P P

M M
On simillar lines stress intensity
factors can also added a

K total  K I (a)  K I (b) P P

= Ya a  Y b a M M

M a M

K total  K I (a)  0 P
P

= Ya a  0
Principle of Superposition
For example consider a plate having semicircular crack
subjected to an internal pressure
1.12 a
KI   g()

1/4 a
3  a 2
 2 a2 2 
  2
, g( )   sin   2 cos   2c
8 8c  c 

For Semi circular crack



  , g( )  1
2
K I (total)  K I (b)  K I (c)
2
 1.12 p a  0

2.24
= p a

Crack Tip Plasticity
LEFM assumes a sharp crack tip, inducing infinite stress at the
crack tip. But in real materials, the crack tip radius is finite and
hence the crack tip stresses are also finite. In addition, inelastic
deformations due to plasticity in metals, crazing in polymers leads
to further relaxation of stresses.
For metals with yielding, LEFM solutions are not accurate.
A small region around the crack tip yields leading to a
small plastic zone around it.

For moderately yielding metals, LEFM


solutions can be used with simple
correction.
For extensively yielding metals,
alternative fracture parameters like
CTOD, J-integral are to be used taking
into account material non-linearity.
Small Scale Plasticity
Irwin’s Approach
Normal stress yy based on elastic analysis is given by
KI    3 
 yy  cos   1  sin   sin   
2 r  2  2   2 

On the crack plane  = 0


KI
 yy 
2 r

As a first approximation yielding


occurs when yy = ys
2 2
KI 1  KI  a  
 ys  , ry    , or ry   
2ry 2   ys  2   ys 

r = first order estimate of plastic zone size. This is approximate


Irwin’s Approach
When yielding occurs, stresses redistribute in order to satisfy equilibrium conditions.
The cross hatched regions represents forces active in the elastic analysis that cannot
be carried in elasto-plastic analysis, because of the reason that the stresses cannot
exceed ys.To redistribute this excessive force, the plastic zone size must increase.
This is possible if the material immediately ahead of plastic zone carries more stress.

Irwin proposed that plasticity makes the crack behave as if it were


larger than actual physical size.
Let the effective crack size be aeff ,
such that aeff = a+
whereis the correction.

or
To permit redistribution of stresses , the areas A and B must be the same

2
 KI 
   plastic constraint factor
  ys 
2
1  KI 
effective crack size = a eff  a  2ry = a+  
   ys 
Effective stress in tensity factor K I  Y  a eff
A long rectangular plate has a width of 100 mm, thickness
of 5 mm and an axial load of 50 kN. If the plate is made of
titanium Ti-6AL-4V,(KIC=115 MPa-m1/2, ys=p10 MPa)
what is the factor of safety against crack growth for a 150KN

crack of length a = 20 mm.


K I  Y  aeff
1/ 2
  a     a
2
 a 
4
Y   sec     1  0.025    0.06    2a=20mm
  2b    b b 
trial 1. a eff = a = 0.01m 2b=100mm
1/ 2
 X 0.2  
Y=  sec       
2 4
  1  0.025 0.2  0.06 0.2
  2 
 1.02448
150 X 1000
  300 MPa
5 X 100
K I  1.02448 X 300 X 0.01
 54.525MPa  m1/ 2
K IC 115
Factor of Safety without plastic correction =   2.11
K I 54.525
trial 2.
2
1  KI 
2
1  54.525 
2 ry =       0.00114275
   ys    910 
effective crack size = a eff  a  2 ry = 0.01  0.00114275  0.1114275
Y  1.3007
Effective SIF K I  Y  aeff  1.0307 X 300 X X 0.01114275  57.85
trial 3.
2 ry  0.001286
a eff  a  2ry = 0.011286
Effective stres s in tensity factor K I  58.25
K IC 115
Factor of Safety with plastic correction =   1.975
K I 58.25
Strip Yield Method (Proposed by Dugdale and Barrenblatt)
They assumed a narrow long slender plastic zone ahead of crack tip
in a non-hardening (ideal plastic) material in plane stress for a
through crack in a infinite plate.

The strip yield plastic zone assumes a crack length of 2a + 2


Whereis length of plastic zone.
Approximate elasto-plastic behavior is obtained by superimposing two
elastic solutions (a) a through crack under remote tension (b) a
through crack under crack closure stress.

Concept: stress at the crack tip is no more a singular and it is a finite


value (ys), hence stress singularity term is zero. I.e. the length of the
plastic zone is such that the stress intensity factor due to remote
tension cancels with crack closure.
Strip Yield Method
SIF due to crack closure stress can be estimated by considering a normal
force P applied to the crack at a distance x from the center line of crack
SIF for two crack tips are given by
x P
P ax
K I (a) 
a a  x
P ax P
K I ( a )  2a
a a  x
Crack closure force = P = - ys dx
For the problem considered
SIF is obtained by replacing P by
- ys dx and a by a+, and adding the two
- ys  a+  x
a+ a+  x 
K I (closure)     dx
  a+  a  a+  x a+  x 
Strip Yield Method
a+ 1  a 
K I ( closure)  -2 ys cos   (A)
  a+ 
SIF due to remote tension
K I (tension )     a+  (B)
(A) = (B)
a   
 cos  
a+ 2 
 ys 
Expanding right hand side using Taylor's series
2 4
a 1    1   
 1       
a+ 2!  2 ys  4!  2 ys 
Considering first two term of the Taylor's series
2
  a   KI 
2 2
=    ...... length of the plastic zone
8 ys
2
8   ys 
Comparison of plastic corrections with LEFM

LEFM solutions are linear


Irwin and strip yield predictions deviate from LEFM theory at
stresses greater than 0.5 ys
Two plasticty predictions deviate at 0.85 ys
Plastic Zone Shape
In the earlier calculation plastic zone size was calculated for  = 0
(along the crack plane), the plastic zone shape will be quite
different when all angles are considered. Plasticity are based on
von Mises and Tresca’s yield criteria. For mode I problem stress
field are obtained using Westergaard’ stress function., then
principal stresses are given by
1/ 2
 xx   yy   xx   yy 2

1 , 2       xy 
2

2  2  
and hence
Using the principal stresses in von Mises yield
criteria

Tresca:
 max   ys / 2
1  2 1  3
 max  or : for PSN
2 2
1  2 1
 max    max  : for PSS
2 2
(a) Von Mises yield criterion (b) Tresca yield criterion
Plane stress plastic zone sizes are larger than plane strain
plastic zone size.
Tresca plastic zones are larger than von Mises plastic zones.
Plastic zone shapes for sliding mode and tearing modes
Plane strain or plane stress
In general, the conditions ahead of a crack tip are neither plane stress
nor plane strain. There are limiting cases where a two dimensional
assumptions are valid, or at least provides a good approximation.
The nature of the plastic zone that is formed ahead of a crack tip
plays a very important role in the determination of the type of failure
that occurs. Since the plastic region is larger in PSS than in PSN,
plane stress failure will, in general, be ductile, while, on the other
hand, plane strain fracture will be brittle, even in a material that is
generally ductile. This phenomenon explains the peculiar thickness
effect, observed in all fracture experiments, that thin samples exhibit
a higher value of fracture toughness than thicker samples made of the
same material and operating at the same temperature. From this it can
be surmised that the plane stress fracture toughness is related
to both metallurgical parameters and specimen geometry while the
plane strain fracture toughness depends more on metallurgical factors
than on the others.
Due to presence of crack
tip, stress in a direction to
normal to crack plane yy
will be large near the crack
tip. This stress would in
turn tries to contract in x
and z direction. But the
material surrounding it will
constraint it, inducing
stresses in x and z direction,
there by a triaxial state of
stress exists near the crack
tip. This leads to plane
strain condition at interior.
At the plate surface zz is
zero and zz is maximum.
This leads to plane stress
condition at exterior.
The state of stress is also dependent on size of plate thickness.
If the plastic zone size is small compared to the plate thickness,
plane strain condition exists.
If the plastic zone size is larger than the plate thickness, plane
stress condition prevails.
As the loading is increased, plastic zone size also increases
leading to plane stress conditions.
Effect of plate thickness on fracture toughness
Limits of LEFM
As per ASTM standard LEFM is applicable for components of size
2
 KI 
a  2.5 
  
 ys 
2
 KI 
t or B  2.5 
  
 ys 
2
 KI 
W or b  2.5 
  
 ys 
As per ASTM standard fracture toughness testing can be done on
Specimens of size
2
 K Ic 
a , B,W  2.5 
  
 ys 
Concept of Isoparametric Elements
(1,1)
Y  (x3, y3)
 (x7, y7)
(-1,1) (1,1)

(x4 , y 4) 
(-1,1) (x 6, y6)


(x 8, y8)

(x5, y5) (1,-1)


(x2 , y2 )
(-1,-1)
(x1, y1)

(-1,-1) (1,-1)
X

 Elements are defined in local coordinate system () with straight


well defined edges.
Elements defined in local coordinates have advantage that numerical
integration limits vary –1 to +1
Elements from local coordinated when mapped on to global cartesian
coordinates, it can be distorted to a new curvilinear set as shown in the
figure.
 In Isop. Formulation, with distorted elements, numerical integration
are still carried with limits of local coorinates i.e. –1 to +1
In the conventional FE formulations final stiffness
matrix is given by [K] =  [B]T[D][B] dv
For the given structural component and loading conduct FE analysis
for the initial crack length 'a' and find the total strain energy in the system U(a).
Now physically shift the crack tip by one element length by suitablly altering the
boundary condition, and find the new strain energy for the extended crack, U(a+a).
Now SE release rate is given by and dv = dxdy
U U(a a)  U(a)
G =-
a a
using the above computed value of G, SIF is evaluated using
GE
KI = GE for PSS and K I = for PSN cases.
1-2

If an appropriate transformation is obtained for [B] and dv from


(x,y) to () we have equation for [K] set in () system where
the integration can be done with in the limits –1 to +1
In order to achieve such a transformation ( [B] and dv) two sets of
nodes ate defined for each element.
One set of nodes (marked as )are used to
interpolate coordinates of a point with in the
element using nodal coordinates. For
4 7 3
example
If N1, N2, N3…… are the interpolation
8 X (x,y) 6
function used to interpolate shape variation,
such that
X = N1 X1+ N2 X2 + N3 X3 +……… 1 5 2
Nodes to define shape (coordinate
Y = N1 Y1+ N2 Y2 + N3 Y3 +……… interpolation with in the element)

For an eight node quadratic element


N1 = - ¼ .()
N5 = - ½ .(

Theinterpolation functions N1, N2, N3……


Another set of nodes (marked as )are used to
interpolate displacements of a point with in the
4 7 3
element using nodal displacements. For
example if N1, N2, N3…… are the interpolation X (u,v)
8 6
function used to interpolate shape variation,
such that
u = N1 u1+ N2 u2 + N3 u3 +……… 1 5 2

v = N1 v1+ N2 v2 + N3 v3 +……… Nodes to define displacement


(displacement interpolation
with in the element)

For a four node bilinear element


N1 = - ¼ .(
N2 = - ½ .(

Theinterpolation functions N1, N2, N3…… are


generally called here as ‘Displacement
function’ as they define the displacement
4 7 3

If number of nodes used to define shape (identified by


and Ni ) and the number of nodes used to define the 8 6

displacement variation (identified by and Ni ) are


equal, say for example 8 nodes for shape and 8 nodes 1 5 2
Isoparametric element
for displacement, then such elements are called
ISOPARAMETRIC ELEMENTS. 4 7 3

If number of nodes used to define shape are more than


the number of nodes used to define the displacement 8 6

variation say for example 8 nodes for shape and 4


nodes for displacement, then such elements are called Superparametric
1 5 element2
SUPERPARAMETRIC ELEMENTS.
4 7 3

If number of nodes used to define shape are less than


8 6
the number of nodes used to define the displacement
variation say for example 4 nodes for shape and 8
nodes for displacement, then such elements are called 1 5
Subparametric element
2

SUBPARAMETRIC ELEMENTS.
N i N i x N i y
 
 x  y 
N i N i x N i y
 
 x  y 
 N i   x y   N i 
        x 
   
 N i   x y   N i 
        y   N i N i 
 N i   N i     x i   yi 
   where  J    
 x   N i x N i 
    J   N  (A) 
   yi
 N i   
i
 i
    y   J  is called Jacobian matrix
For any curved element formulations (iso, super or sub) [J] is
obtained using shape functions (defined for shapes)
Re writing eqn. (A)  N i 
 0 
 N i   N i   x 
 N i 
 x  
1   
 note  B  0
y 
 N    J     
 i  N i   N i N i 
 y 
 y      x 

 N i 
  
note that   is obtained by displacement functions
 N i 
  
 N i 
 x 
The terms of strain-displacement matrix [B] are given by  
N
 i
 y 
of the prvious equation and dv  dxdy  det J d d 
Crack tip singular elements
4 7 3
Use of conventional elements near the crack
tip, even with very fine mesh near would not
simulate the stress field conditions of the
6
crack tip (1/ r ) . Barsoum showed that by
8
moving mid-side nodes of a 8 node
isoparametric element to quarter points 1
5 2
induced 1/ r singularity and improved the Crack tip Move mid-side
Nodes at 41 points
performance of the analysis enormously.

4 7 3 3 3

7 7
Collapse
side
1,8,4 1,8,4 6
8 6 6
Crack tip Crack tip
5 5
1 Move mid-side
5 2 2 Nodes at 41 points 2
Crack tip
In the FEM formulations strain vector
 N i 
is given by   0 
ui   
    J   B    ,
1
 N i 
 vi  where  B  0
 
J = jacobian matrix  
 N i N i 
,  parametric coordinates of a point    
 
on an element
The strain vector can be singular if either  J  or  B is sigular
-1

 (x,y)
 J -1
can be sigular when det J = =0 4 7 3

 (,)
consider the quadratic element as shown 6

let us evaluate the boundary nodes 8

1
between 1 and 2 in x direction Crack tip
5 2
Move mid-side
Nodes at 41 points
1 1
N1   (1  ), N1  (1  ), N1  (1  2 )
2 2 y
x
x  N1 x1  N 2 x2  N 5 x5 1 5 2

L
1 1 L/4
x   (1  ) x1  (1  ) x2  (1  2 ) x5
2 2
x1  0, x2  L, x5  L / 4
1 2 L
x  (1  ) L  (1   )
2 4
x
solving   1  2
L
x L x
Jacobian  J    (1  ) 
 2 L
as x  0; det J  0......leading to singularity
Let us evaluate the strain
1 1
u  N1u1  N 2 u2  N 5u5   (1  )u1  (1  )u2  (1  2 )u5
2 2
x
using   1  2
L
1 x  x 1 x  x 
u    1  2 2  2  u1   1  2 2  u2
2 L  L 2 L  L 
 x x
 4   u5
 L L
u 1 3 4 1 1 4  2 4
x      u1      u2     u5
x 2  xL L  2 xL L   xL L 
1 1
  ......leading to 1/ r singularity
x r
Virtual Crack Extension Method to Evaluate SIF
For a two-dimensional cracked
body in mode I, the total potential
in terms of FE solutions is given
by
U = ½ {u}T[K]{u}- {u}T[F]
The strain energy release rate
Is defined as
U
G=- crack tip FE mesh
a
  u 1 T   K T   F
T

= -
a
  K   u   F    2  u a  u   u a
The first term in above expression is zero (equilibrium condition).
In the absence of traction on crack face the third term is also zero.
Hence
U K I2 1 T   K The strain energy release rate
G=-     u  u is proportional to the
a E  2 a
derivative of the stiff ness

where E  E for plane stress
matrix with respect to crack
E length.
= for plane strain
 
1- 2

For a given FE mesh for a body with


crack length a, and to extend the crack
length by a it is not necessary to alter
entire mesh. This can be achieved by
moving a few elements near the crack
tip and keeping rest intact. If N
number of elements that are effected
then a Da

1 T  N   K 
G =   u     u
2  i 1 a 
Stress Approach to Evaluate SIF
KI    3 
 yy  cos   1  sin   sin   
2 r 2  2   2 
On the crack plane  = 0
KI
 yy 
2 r
Re arranging r1 r2 r3 r4

K I   yy 2 r y1
y2
y3
y4
At various points r1 , r2 ...measure  y1 ,  y2 ..
Hence a set of K I (r1 ), K I (r2 )... r1 r2 r3 r4

can be calculated.
Plot these values againist r/a.
KI
x
Extend the best fit line to meet x
x x
x
x
abcissa to get the required K I for x

the given structural conponent


and loading. r/a
Displacement Approach to Evaluate SIF
2(1  ) r    1  
v KI sin     cos2   
E 2  2   2  2 
where v = displacements in y directions
3  
  (3  4) for PSS and     for PSN
 1  
At some given angle =1 v1 v2 v3
v4

2
K I (r)    f  1  v(r) r4
r r2
r3

q= q1
At various points r1 , r2 ...measure v 1 , v 2 .. r
r1

Hence a set of K I (r1 ), K I (r2 )...can be


calculated. Plot these values againist r/a.
KI
Extend the best fit line to meet abcissa to x
x
x
x x
x
get the required K I for the given structural x

conponent and loading.


r/a
Strain Energy Release Rate to Evaluate SIF

1 2
1 2

a + a
a

F o r a s te e l b e a m s h o w n b e lo w , a c r a c k o f le n g th 7 .5 m m is d e te c t e d
b y N D T . F in d th e b e a m t h ic k n e s s to p ro v id e a fa c to r o f s a fe ty o f 2
(1 ) b y ig n o rin g p la s tic z o n e (2 ) b y ta k in g p la s tic z o n e in to a c c o u n t.
 ys= 1 0 0 0 M P a , K IC = 9 0 M P a - m 1 /2 P = 2 0 K N

3 P L a
K I  2
x
2 b e
  a   a 
2 e = ?

1 .9 6  2 .7 5    1 3 .6 6   b = 5 0 m m

  b   b 

 a 
2
 a 
4

 2 3 .9 8    2 5 .2 2   
 b   b  
Ig n o rin g p la s tic ity
 2 0 x 1 0 0 0 
3 
1 x 1 0 6   0 .9 0 .0 0 7 5
K    x
2 0 .0 5  e
I 2

1 .9 6  2 .7 5 0  0  0  0  
2 2 4
.1 5  1 3 .6 6 .1 5  2 3 .9 8 .1 5  2 5 .2 2 .1 5
 

You might also like