# 108

Chapter 5: Failure Theories

5.1 Comparison of Isotropic and Composites

1) Isotropic – usually one material property

failure in x direction
u
σ

failure in y direction
u
σ

Shear Strength
2
max
u
σ
= τ .

2) Composites

a) Several failure modes possible

together) (humps Topic Current
Debonding iii)
Microcrack Matrix ii)
Break Fiber i)
¦
)
¦
`
¹
iv) Delamination} Elasticity/Fracture Mechanics.

b) Different strengths in different directions

failure in x direction
T
X

failure in y direction
T
Y

109
Shear Strength S

c) Different strength in tension and compression

especially n compressio ← σ

≠ σ
Tu Tu

c T
X X ≠

d) Sign on shear is important

e) Interaction between stresses is important

x
σ and
y
σ combine to cause failure.

5.2 Use of Lamina Strengths in the Analysis of Laminates

110

1) Use CLT to find stresses in the nth ply due to

2) Use these stresses to analyze the nth ply independent of
the other plies.

3) Use “weakest link” concept to predict strength of the
laminate:

i.e., If none of the plies fail, then the laminate does not fail.

If one or more of the plies fail, then the laminate has failed.

Note: This procedure requires

i) Knowledge of the stresses within the laminate

ii) A failure theory to predict at what stress level failure
occurs.

5.2 General Features of Composite Failure Theories

- Most theories begin with lamina properties in the
principle material directions

i.e.,
LTu Tu Lu
τ σ σ
,
, etc.

S X X
T T

111
- These properties are transformed into other coordinate
systems as needed.

- Most have the form: 1 ≤
σ
σ
iu
i

This means that if the stress in the lamina,
i
σ , is less
than the failure stress,
iu
σ , then the lamina is okay.

- All failure theories are a variation on this theme.

5.3 Determination of , ,
Tu Lu
σ σ etc.

5.3.1 Micromechanics

1) Recall
m f m f fu Lu
V E V
*
∈ + σ = σ

where
f
f
E
fu
σ
= ∈
*

This is not very accurate in general. It has assumed that
all the fibers break at once.

Tu
σ and
LTu
τ ?

There are not accurate micromechanics failure
theories for these values.

112
At least none that are commonly used.

Solution? Go to the Lab.

b) Picture Frame

Problem with premature failures in the corners.

c) Rail Shear

113
d) Isoipescu Specimen

Advantage: Pure Shear in the center of the beam.

Disadvantage: Shear distribution is not uniform across the
center.

Solution: Notch the beam.

114

The problem with this is that the notch angle is a function of
material stiffness.

i.e., 0° specimen

90°

115

Edge Damage in a
s
] ) 10 [(
2
± Celion 6000/PMR-15
Graphite Polyimide Laminate at 99% of Ultimate Strain

116

Edge Damage in a
s
] ) 30 [(
2
± Celion 6000/PMR-15
Graphite Polyimide Laminate at 99% Ultimate Strain

117

Edge Damage in a
s
] ) 45 [(
2
± Celion 6000/PMR-15
Graphite Polyimide Laminate at 95% of Ultimate Strain

118

Edge Damage in a
s
] ) 60 [(
2
± Celion 6000/PMR-15
Graphite Polyimide Laminate at 99% of Ultimate Strain

119

Edge Damage in a
s
] ) 90 [(
2
± Celion 6000/PMR-15
Graphite Polyimide Laminate at 92% of Ultimate Strain

120

Edge Damage in a
2
] 90 / 0 / 45 [± Celion 6000/PMR-15
Graphite Polyimide Laminate at 37% of Ultimate Strain

121

Edge Damage in a
s
] 45 / 90 / 45 / 0 [ − ± Celion 6000/PMR-15
Graphite Polyimide Laminate at 80% of Ultimate Strain

122

Edge Damage in a
s
] 45 / 90 / 45 / 0 [ − ± Celion 6000/PMR-15
Graphite Polyimide Laminate at 80% of Ultimate Strain

123

124

Composite
δ
σ r K
I ij
~

Metal
2 / 1
~

σ r K
I ij

125

126
5.4 Failure Theories (from Simplest to most Complex)

5.4.1 Maximum Stress Theory (pg 133)

1) For a material loaded in principle material direction:

¦
¹
¦
´
¦
=
σ
σ
σ < σ <
σ
σ
1 failure at
) ( 1 failure No for
Direction Longitudal In
1u
1
10 1
1

Similarly , 1
2
2

σ
σ
u
1
12
12

τ
τ
u

Each component must be checked

i.e., , 1
1
1
<
σ
σ
u
, 1
2
2
<
σ
σ
u
1
12
12
<
τ
τ
u

for no failure

127
2) For an off axis lamina:

¦
)
¦
`
¹
¦
¹
¦
´
¦
σ
(
(
(
¸
(

¸

=
¦
)
¦
`
¹
¦
¹
¦
´
¦
τ
σ
σ
0
0
12
2
1 x
T

⇒ θ σ = σ
2
cos
x L

θ σ = σ
2
sin
x T

θ θ σ = τ cos sin | |
x LT

Failure Theory Becomes: 1
cos
2
<
σ
θ σ
Lu
x

1
sin
2
<
σ
θ σ
Tu
x

1
cos sin
<
τ
θ θ σ
LTu
x

or ,
cos
2
θ
σ
< σ
Lu
x
,
sin
2
θ
σ
< σ
Tu
x

θ θ
τ
< σ
cos sin
LTu
x

128
Plot of
x
σ versus θ (Tension)

Problems with Max Stress:

i) No stress interaction

ii) Does not fit data extremely well.

5.4.2 Maximum Strain Theory

1) Similar to Max Stress:

, 1 <

Lu
L
, 1 <

Tu
T
1 <
γ
γ
LTu
LT

129
If the material is linear elastic
L
Lu
Lu
E
σ
= ∈ ,
T
Tu
Tu
E
σ
= ∈ ,
LT
LTu
LTu
G
τ
= γ

Recall Hooke’s Law: (Orthotropic Material)

T
L
LT
L
L
L
E E
σ
ν
− σ = ∈
1
) (
2 12 1 11
σ + σ = s s

T
T
L
T
TL
T
E E
σ + σ
ν
− = ∈
1

LT
LT
LT
G
τ
= γ

Now: 1
1
<
σ
σ
ν
− σ
=

L
Lu
T
L
LT
L
L
Lu
T
E
E E

or 1 <
σ
σ ν − σ
Lu
T LT L

Similarly 1 <
σ
σ + σ ν −
Tu
T L TL

130
and 1 <
τ
τ
LTu
LT

failure envelope

2) For an off axis lamina:

θ θ
τ
< σ
θ ν − θ
σ
< σ
θ ν − θ
σ
< σ
cos sin
cos sin
sin cos
2 2
2 2
LTu
x
TL
Tu
x
LT
Lu
x

131
Not much difference from max stress theory. Still no stress
interaction, this doesn’t work as well as we want.

5.4.3 Tsai-Hill Theory (Max Work)

1948 – Hill proposed anisotropic yield theory for metals

1965 – Tsai modified Hill’s theory and applied it to composites

1) 1
2 2 2
<
|
|
.
|

\
|
τ
τ
+
|
|
.
|

\
|
σ
σ
+
|
|
.
|

\
|
σ
σ
|
|
.
|

\
|
σ
σ

|
|
.
|

\
|
σ
σ
LTu
LT
Tu
T
Lu
T
Lu
L
Lu
L
4 4 3 4 4 2 1

Stress interaction term, but relatively weak (i.e.,
Lu
T
σ
σ
is
small)

2) For an off axis lamina ) 0 , 0 ( = τ σ ≠ σ
xy y x

2 2
2 2
2
4
2
2 2
2
4
1 cos sin
sin sin cos cos
x LTU
Tu Lu Lu
σ
<
τ
θ θ
+
σ
θ
+
σ
θ θ

σ
θ

132

3) For multi-axial stress state

i) Compute
xy y x
τ σ σ , , for a given ply

ii) Transform the stresses to the principle material
directions ) , , (i.e.,
LT T L
τ σ σ
¦
)
¦
`
¹
¦
¹
¦
´
¦
τ
σ
σ
=
¦
)
¦
`
¹
¦
¹
¦
´
¦
τ
σ
σ
xy
y
x
LT
T
L
T] [

iii) Check the sign on the principle (material) stresses if
0 < σ
L
(compressive) use
Lu
σ

0 < σ
T
(Compressive) use
Tu
σ

LT
τ doesn’t matter

133
iv) Let:

2 2 2
) (
|
|
.
|

\
|
τ
τ
+
|
|
.
|

\
|
σ
σ
+
|
|
.
|

\
|
σ
σ
|
|
.
|

\
|
σ
σ

|
|
.
|

\
|
σ
σ
= σ
LTu
LT
Tu
T
Lu
T
Lu
L
Lu
L
F

or
Lu
σ

or
Tu
σ

v) Check F if F < 1 no failure

if F ≥ 1

vi) Repeat process for each ply

ii) Still easy to use
iii) Stress interaction term

2
) ( , therefore
differences between tension and
compression are not accurately
portrayed

5.4.4 Tsai-Wu Tensor Polynomial (1971)

1) Deviation of Formulae

i) Assume a failure function F(σ)
ii) Expand in a series:

134
... 2 ) (
3
3
2
1 0
+ σ + σ + σ + = σ f f f f F

or in Tensor Form:

... ) (
0
+ σ σ σ + σ + = σ
mn k ij ijk ij ij ij
f f f F
l l

iii) Arbitrarily say that failure occurs when
0
2 ) ( f F
ij
= σ

Then at failure

... 1 2
0 0 0
+ σ σ σ + σ σ + σ +
mn k ij
mn ijk
k ij
ijk
ij
ij
f
f
f
f
f
f
l
l
l
l

Define
ij
ij
F
f
f

0
, etc.

... + σ σ σ + σ σ + σ
mn k ij mn ijk k ij ijk ij ij
F F F
l l l l

iv) Tsai and Wu used two terms:

failure for 1 = σ σ + σ
l l k ij ijk ij ij
F F

if 1 < σ σ + σ
l l k ij ijk ij ij
F F no failure

In reduced notation
1 11
σ = σ
6 12
σ = σ etc.

1 = σ σ + σ
j i ij i i
F F

135
2) Features of this failure theory

i) It is a tensor theory; therefore it lends itself well to
computer

ii) It can handle a fully 3-D stress state

iii) It contains many stress interaction terms

iv) The
i
F terms are sight sensitive and therefore can
tell the difference between tension and compression

v) It is not as simple as other theories

vi) It requires more tests to determine
i
F and
ij
F (i.e.,
LTu Tu Lu
τ σ σ and , , are not enough.)

5.4.5 Review of Failure Theories

1) These failure theories area used for three failure types

i) Fiber Break

ii) Matrix Microcrack

iii) Debonding

2) All of the theories rely heavily on experimental data

136

3) Delamination cannot be accounted for with these