Documente Academic
Documente Profesional
Documente Cultură
CHAPTER 3
based criteria category. Examples for Strain based criteria are Wolfe and
Eckold criteria. Tsai, Rotem, Puck, Hashin and Edge belong to interactive
theory category. The proposed modified criterion is also interactive in nature.
The conventional failure theories like Maximum stress and maximum strain
criteria assume no stress interaction, hence not considered for this evaluation.
Failure criteria for composite materials are often classified into two
groups: based on the stress interaction considered,
ε whenε11 ≥ 0 ε 2t whenε 22 ≥ 0
ε11 = 1t and ε 22 = (3.2)
ε 1c whenε11 ≤ 0 ε 2 c whenε 22 ≤ 0
1. Polynomial Theories,
2. Direct mode determining theories,
3. Strain Energy theories.
Tsai-Hill Theory
It takes form from 2D Von-Misses yield Criteria
σ 11 2 + σ 22 2 − σ 11σ 22 = σ 2 yp (3.4)
Hill modified this Criteria for ductile material and based on that
Azizi and Tsai formulated one Criteria for orthotropic composite materials,
σ 112 σ 22 2 τ 12 2 σ 11σ 22
2
+ 2
+ 2
− =1 (3.5)
F1 F2 F6 F12
1 1
f 66 = 2
; f 12 = − [ f11 f 22 ]1 / 2
S LT 2
1 1 1 1 1 1
− σ 11 + − σ 22 + σ 112 + σ 222 +
X L XT YL YT X L XT YLYT
1 2 σ 11σ 22
2
τ 12 − =1 (3.8)
S LT X L X T YLYT
Failure Modes
σ 11 τ 12
Combined Shear and fiber compression: + = 1 (3.15)
X T S LT
Hashin’s criteria is the base of all theories of this type, first it was
formulated based on failure modes as fiber failure and matrix failure, later had
been extended into tension and compression for each modes.
2 2
σ τ 12
Fiber Tension (for σ 11 ≥ 0 ): 11 + = 1 (3.16)
XL SL
σ 11
Fiber Compression (for σ 11 ≤ 0 ): =1 (3.17)
XT
2 2
σ τ 12
Matrix Tension (for σ 22 ≥ 0 ): 22 + = 1 (3.18)
YL SL
Matrix Compression (for σ 22 ≤ 0 ):
2 2
σ 22 YT σ τ
+ − 1 22 + 12 = 1
2ST 2 S T Y
T SL
(3.19)
The translation from the fiber failure surface to the lamina strain
surface depends on the assumption made by Hart-Smith that the longitudinal
and transverse strains in the lamina are the same as those in the fiber. There
are certain issues that need to be resolved in connection with the Hart-Smith
approach to failure. The most crucial is that of whether or not there are 450
cutoffs of the lamina failure envelope in the second and fourth quadrants. As
indicated previously, without the cutoffs the Hart-Smith criterion reduces to
the maximum Strain or Maximum Stress criterion. In addition, the evidence
for shear failures in fibers under normal stress loading as well as the
implications of low shear strength in ±450 laminates need to be fully explored.
The experimental results cited by Hart-Smith for these phenomena are
somewhat limited and have not been confirmed to any extent by other criteria.
1 γ f 12
Fiber Tension (for ε 11 ≥ 0 ): ε 11 + mσ f σ 22 = 1 (3.20)
ε 1T Ef1
Fiber Compression (for ε 11 ≤ 0 ):
1 γ f 12
mσ f σ 22 = 1 − (10γ 21 )
2
ε 11 +
(3.21)
ε 1C Ef1
87
In order to find the fracture plane one has first to determine the
stresses σ n (α ) , τ nt (α ) , τ nl (α ) on a plane parallel to the fibers. In dependence of
σ n being positive or negative the stresses are inserted into the valid equations.
This has to be repeated for a sufficiently large number of angles (α ) , between
-900 and +900 until that cutting angle (α fp ) are found for which the highest
The final failure predictions for multi directional laminates are also
conservative in nature. The stress strain response is generally truncated at
much lower strains than the final strain observed in the experiments. They
generally coincide with the initial failure range alone and they do not predict
the final failure properly.
89
σ 11
=1 (3.27)
XL
2 2
σ 22 τ
+ 12 = 1 (3.28)
YL S LT
σ 11
Fiber Tension (for σ 11 ≥ 0 ): =1 (3.29)
XL
σ 11
Fiber Compression (for σ 11 ≤ 0 ): =1 (3.30)
XT
σ 2
τ
2
(segment 3-4 of the diagram), the modulus of the ply completely regains its
initial value. Repeated deformation of the unidirectional ply under transverse
tension (positive σ 2 values) follows along the 3-2 segment and further along
the 2-21 segment of the diagram. The deformation diagram in Figure 3.6(b) is
plotted as a function of the modified strain ε 2 = ε 2 + γ 12 ε 1 . Hence σ 2 can be
written as
E11 γ
σ 22 = γ 21 ε 11 + 12 ε 22 (3.35)
(1 − γ 12 γ 21 ) γ 21
−1
ε 22 γ 122 *
σ 22
E22 = * + = (3.36)
σ 22 E11 γ 2σ *
ε 22 + 12 22
E11
The starred values are the largest algebraic values during the
history of deformation. If the stress σ 22 reaches its ultimate value YT, the ply
is then considered to be broken. The ply behaviour under shear is, in many
ways, similar to that under deformation in the direction transverse to the fiber
direction (Figure 3.6(c)). The stress/ strain curve of the ply is linear elastic
within the 0-1 segment. Segment 1-2 corresponds to the stage of matrix
cracking. Unloading process (segment 2-3) takes place with the unloading
~ τ 12*
shear modulus G12 = . The process of shear deformation does not depend
γ 12*
on the sign of the stress τ 12 , which is why ply deformation within the 3-4
~
segments also occurs with the unloading modulus G12 . Repeated deformation
of the ply under positive τ 12 values follows along the 4-3-2 segment and
further along the 2-21 segment where the process of matrix cracking resumes.
τ T eff = τ T + ηT σ n (3.40)
τ L eff = τ L + η Lσ n (3.41)
The above set of equations has the effect of applied biaxial stress
field σ 22 , τ 12 and inclination of critical section. It is proposed that matrix
failure occurs under compression due to interaction between effective shear
stress components acting on the faces of the critical section.
τ T eff
=
2
[ (
σ 22 − sin θ cos θ + η T cos 2 θ )]
2
(3.45)
ST S T2
[
A = η T cos 2 θ − sin θ cos θ ]2
+ η L2 Cos 4θ (3.48)
C = Cos 2θ (3.50)
equation (3.43) and condition 3.44. Hence APC’s can be computed using
equations (3.48) to (3.50). Substituting the values of APC’s in equation
(3.47), the magnitude of τ 12 for the next load case can be computed. Utilize
the computed value of τ 12 for the next increment of σ 22 to calculate θ . Using
the calculated θ , revise the τ 12 using (3.47). Repeat this procedure till σ 22
reaches the transverse compressive strength. The same procedure can be
utilized for evaluating the failure envelope points in the longitudinal
compressive region also.
To avoid the sudden jump in strain at ply failure seen in the Parallel
Spring Model, a model resembling the bilinear hardening rule in classical
plasticity can be formulated. Laminate stiffness reduction is achieved similar
to the Parallel Spring Model. However, it is assumed that the reduced
100
such as [±45]s may fail without fiber breakage. Others have suggested a “last
ply” definition in which the laminate is considered failed if every ply has been
damaged. For this project, the laminate failure is defined as occurring when
either fiber breakage occurs in any ply or the reduced stiffness matrix
becomes singular (Reddy et al 1987).
This FPCL Code has been generated using C++ language. The
purpose of this program is to provide a thorough analysis of the failure
progression leading to ultimate failure in laminated composites. The program
utilizes 2-D classical laminated plate theory with a Ply-by-Ply Discount
laminate analysis method. The code utilize two kinds of inputs namely Inbuilt
input and interactive input. The inbuilt input includes the mechanical and
thermal properties of four fibrous and matrix materials and are enlisted in
Tables 3.1 and 3.2.
103
Once the programme is initialized, the user can enter the required
type of fiber or matrix material. Type 1 if the fiber type is E-glass 21×K43
Gevetex, similarly type either 2 or 3 or 4 if the fiber type is Silenka E-Glass
1200tex or AS4 or T300. For the case of matrices 1, 2, 3 and 4 corresponds to
MY750/HY917/ epoxy, LY556/DY063 epoxy, 3501-6 epoxy and BSL914C
epoxy respectively. In order to invoke the required material properties from
the inbuilt input “Switch-Case-Break” option is employed. The other kind of
input employed in the coding is Interactive input. The following parameters
are specified as interactive input.
E-glass Silenka
Fibre type AS4 T300 21×K43 E-Glass
Gevetex 1200tex
Longitudinal modulus, Ef1 (GPa) 225 230 80 74
Transverse modulus, Ef2 (GPa) 15 15 80 74
In-plane shear modulus, Gf12 (GPa) 15 15 33.33 30.8
Major Poisson's ratio, ν f 12 0.2
Transverse shear modulus, Gf23 (GPa) 7 7 33.33 30.8
Longitudinal tensile strength, XfL (MPa) 3350 2500 2150 2150
Longitudinal compressive strength, YfL 2500 2000 1450 1450
(MPa)
Longitudinal tensile failure strain, 1.488 1.086 2.687 2.905
ε f 1L (%)
Longitudinal compressive failure strain, 1.111 0.869 1.813 1.959
ε f 2 L (%)
Longitudinal thermal coefficient, -0.5 -0.7 4.9 4.9
α f 1 (10-6/0C)
Transverse thermal coefficient, 15 12 4.9 4.9
α f 2 (10-6/0C)
104
LY556/ MY750/
3501-6 BSL914C
Matrix type DY063 HY917/
epoxy epoxy
epoxy epoxy
Modulus, Em (Gpa) 4.2 4.0 3.35 3.35
Shear modulus, Gm (Gpa) 1.567 1.481 1.24 1.24
Poisson's ratio, ν m 0.34 0.35 0.35 0.35
Tensile strength, YmT (MPa) 69 75 80 80
Compressive strength, YmC (MPa) 250 150 120 120
Shear strength, Sm (MPa) 50 70 68 68
Tensile failure strain, ε mT (%) 1.7 4 5 5
Thermal coefficient, α m (10-6/0C) 45 55 58 58
1. Hart Smith
2. Sun
3. Zinoviev
4. Eckold
5. Hart Smith
6. Wolfe
105
vector and stiffness matrix of that lamina. The computed stress can be
substituted in the failure criteria under consideration to check out the
condition of the particular lamina (Kere et al 2001). For interactive theories
there will be set of failure envelope equations as discussed in the previous
topic. If the failure is encountered in matrix, ply by ply discount procedure is
adopted. On the other hand, if the failure is in fiber the code will declare it as
ultimate failure of the laminate (Jones 1998). The flowchart for the failure
envelope generation of a lamina is given in Appendix 2.
106
100
τ12 (MPa)
SR = -1.81:1 Hart - Smith[54]
Sun[117]
Zinoviev[143]
80 Modified
Experimental data[28]
SR = 1.52:1
60
40
20
0
-175 -150 -125 -100 -75 -50 -25 0 25 50 75 100
σ22 (MPa)
The plotted graph shows the response of shear stress due to the
application of either tensile transverse stress or compressive transverse stress.
From the experimental data plotted in the figures, it is understood that the
107
100
SR = -1.81:1 Eckold[34]
τ12 (MPa)
Hart-Smith[54]
Wolfe[137]
80 Modified
Experimental data[28]
SR = 1.52:1
60
40
20
0
-175 -150 -125 -100 -75 -50 -25 0 25 50 75 100
σ22 (MPa)
100
τ12 (MPa)
SR = -1.81:1 Edge[37]
Hart Smith[54]
Hashin[57]
80 Puck[101]
Rotem[109]
Tsai[133]
Modified
60 Experimental data[28]
40 SR = 1.52:1
a
20
σ22 (MPa)
0
-175 -150 -125 -100 -75 -50 -25 0 25 50 75 100 125
Puck’s criteria in addition to Modified criteria fit well with experimental data
and they predict the maximum shear stress point. Since the interactive criteria
predict the failure in a relatively better manner, a comparison among the
interactive criteria is also made in terms of standard error (SE) between
analytical and test results and is obtained using the following expression.
2
1 N Analysis Re sult
SE = ∑ 1 −
N i =1 Test Re sult i
(3.51)
N is the number of test data points. Tables 3.3 to 3.5 shows the error
comparison for prediction made by various stress based, strain based and
interactive criteria with experimental data points. Predictions made by
Modified criteria are good with a SE of 0.0497 whereas the SE for Puck and
Edge are 0.0596 and 0.0608 respectively. Hence, it can be concluded that
Modified criteria along with Puck and Edge can be recommended for
reasonable conservative predictions of the lamina failure envelopes for
composite wind turbine blades. Figures 3.14 and 3.15 show the bar chart
comparison in failure prediction at stress ratio of 1.52:1 which is in the first
quadrant and at -1.81:1 which is in the second quadrant.
1.5 1.19
1.07 1.07 1.11 1.11 1.12
theo /
1.02
1
0.5
0
v
i
i th
d
ge
k
em
ed
fe
a
ie
Su
l
c
Ts
ko
ol
Ed
i fi
Sm
Pu
ov
t
Ro
Ec
od
n
t-
Zi
M
r
Ha
Figure 3.14 Bar Chart showing the ratio of predicted and experimental
strength for lamina of Wind turbine blade material in
τ 12 − σ 22 field at stress ratio of 1.52:1
110
1.20 1.06
1.00 0.92 0.92
0.77 0.77 0.79 0.80
0.80 0.70 0.73
σtheo /σexp
0.60
0.40
0.20
0.00
v
ith
i
ed
em
ge
fe
a
ie
Su
c
Ts
Sm
ol
if i
Ed
Pu
ov
t
Ro
W
od
n
t-
Zi
M
ar
H
Figure 3.15 Bar chart showing the ratio of predicted and experimental
strength for lamina of wind turbine blade material in
τ 12 − σ 22 field at stress ratio of -1.81:1
112
113
Q 11 Q 12 Q 16
Qij = Q 21 Q 22 Q 26 (3.52)
Q 61 Q 62 Q 66
4 4 2 2
Q11 = Q11 C + Q22 S +2(Q12+2Q66) S C (3.53)
E11 ν 21 E11 ν E
Q11 = ; Q12 = ; Q16 = Q26 = Q61 = Q62 = 0; Q21 = 21 22 ;
1 − ν 21ν 12 1 − ν 21ν 12 1 − ν 21ν 12
E 22
Q22 = ; Q66 = G12
1 − ν 21ν 12
114
0
N xx A11 A12 A66 ε xx B11 B12 B66 k xx
0
N yy = A21 A22 A26 ε yy + B21 B22 B26 k yy (3.62)
A66 γ xy B61 B66 k xy
0
N xy A61 A62 B62
0
M xx B11 B12 B66 ε xx D11 D12 D66 k xx
0
M yy = B21 B22 B26 ε yy + D21 D22 D26 k yy (3.63)
B66 γ xy D61 D66 k xy
0
M xy B61 B62 D62
Here,
n
Aij = ∑ Q ij [ ] (h
k k − hk −1 ) ;
k =1
1 n
Bij = ∑ Q ij
2 k =1
[ ] (h k
2
k − h 2 k −1 ; ) (3.64)
1 n
Dij = ∑ Q ij
3 k =1
[ ] (h k
3
k − h 3 k −1 )
0
ε0 xx N xx M xx k xx N xx M xx
ε yy = [ A1 ] N yy + [B1 ] M yy ; k yy = [C1 ] N yy + [D1 ] M yy (3.65)
0 N xy M xy k N xy M xy
γ xy xy
Here,
[A1 ] = [A −1 ] + [A −1 ][B] (D *)−1 [B][A −1 ] ;
[B1 ] = −[A −1 ][B] (D *)−1
[C1 ] = [B1 ] T ; (3.66)
After calculating the mid plane strain and curvature, induced strain
in each lamina due to the applied force or moment can be found out using the
following relation.
0
ε
ε xx xx k xx
0
ε yy = ε yy + z k yy (3.67)
γ 0 k
xy γ xy xy
Induced stresses in each lamina can be found using the stress strain
relationship.
120
τxy (MPa)
SR = -2.35:1
100 SR = 1:1
80
60
40
Hart-Smith[54]
Sun[117]
20
Zinoviev[143]
Modified
σyy (MPa)
0
-275 -225 -175 -125 -75 -25 25 75
120
τxy (MPa)
Eckold[34]
Hart-Smith[54]
Wolfe[137] 100 SR = 1:1
Modified SR = -2.35:1
80
60
40
20
σyy (MPa)
0
-275 -225 -175 -125 -75 -25 25 75
120
τxy (MPa)
SR = -2.35:1
100 SR = 1:1
80
60
40
Edge[37]
Hart-Smith[54]
Hashin[57]
Puck[101]
20
Rotem[109]
Tsai[133]
Modified
σyy (MPa)
0
-275 -225 -175 -125 -75 -25 25 75
matrix failure. If the failure is due to matrix shear Q 22 and Q 66 are reduced.
This is due to the fact that a transverse matrix failure doesn’t necessarily
inhibit the ability of the lamina to carry loading in the shear direction. The
119
500
τxy (MPa)
Hart-Smith[54] SR = 1:1
Sun[117]
Zinoviev[143]
400
Modified
Experimental Data[28]
300
Working Stresses
SR = -2.35:1
200
100
σyy (MPa)
0
-500 -300 -100 100 300 500 700
Working Stresses corresponds to Wind Velocities 5, 10, 15, 20 & 25 m/sec
500
τxy (MPa)
Eckold[34]
SR = 1:1
Hart-Smith[54]
Wolfe[137] 400
Modified
Experimental Data[28]
Working Stresses
300
SR = -2.35:1
200
100
σyy (MPa)
0
-500 -300 -100 100 300 500 700
Working Stresses corresponds to Wind Velocities 5, 10, 15, 20 & 25 m/sec
500
τxy (MPa)
Edge[37]
SR = 1:1
Hart-Smith[54]
Hashin[57]
400
Puck[101]
Rotem[109]
Tsai[133]
Modified 300
Experimental Data[28]
Working Stresses
SR = -2.35:1 200
100
σyy (MPa)
0
-500 -300 -100 100 300 500 700
Working Stresses corresponds to Wind Velocities 5, 10, 15, 20 & 25 m/sec
Figures 3.22 and 3.23 shows the bar chart comparison between
various prediction at stress ratio of 1:1 and -2.35:1 respectively. Tables 3.6 to
3.8 show the error in prediction of failure for various criteria. For predicting
the final failure of the wind turbine blade material, Modified criteria has a
least SE of 0.0581. Puck, Hart-Smith and Zinoviev are also having a
minimum value of SE with 0.0625, 0.0664 and 0.0752 respectively. Hence
Modified criteria along with Puck Hart-Smith and Zinoviev can be considered
for reasonable conservative predictions of the final failure of wind turbine
blade materials.
122
1.2 1.03
0.99 1.00
1
0.80
0.71 0.73 0.73
0.8
σtheo /σexp
0.6 0.44
0.38 0.38 0.39
0.4
0.2
0
ev
ith
ld
n
i
ed
m
ge
fe
hi
Su
c
Ts
vi
ko
te
Sm
ol
ifi
Ed
Pu
as
no
Ro
W
Ec
od
H
t-
Zi
M
ar
H
Figure 3.22 Bar chart showing the ratio of predicted and experimental
strength for laminate of wind turbine blade material in
τ xy − σ yy field at stress ratio of 1:1
1.2
1.02
0.97
1
0.82 0.82
0.8 0.70 0.71 0.72
σtheo/σexp
0.63
0.6
0.47
0.2
0
ith
ev
ld
d
em
n
ai
ck
ge
fe
ie
hi
Su
Sm
vi
ko
Ts
ol
Ed
Pu
if
as
ot
no
W
od
Ec
t-
Zi
M
ar
H
Figure 3.23 Bar chart showing the ratio of predicted and experimental
strength for laminate of wind turbine blade material in
τ xy − σ yy field at stress ratio of -2.35:1
123
σ yy τ xy τ xy RE τ xy RE τ xy RE τ xy RE
(%) (%) (%) (%)
520.0 10.0 -- -- -- -- -- -- 9.6 -4.0
450.0 175.0 -- -- -- -- 162.5 -7.1 260.4 48.8
430.0 140.0 -- -- -- -- 187.7 34.1 267.3 90.9
410.0 160.0 -- -- -- -- 212.1 32.6 274.2 71.4
400.0 250.0 -- -- -- -- 224.7 -10.1 277.7 11.1
480.0 250.0 -- -- -- -- 51.2 -79.5 250.0 0.0
350.0 280.0 -- -- -- -- 225.0 -19.6 296.8 6.0
300.0 300.0 -- -- -- -- 225.0 -25.0 307.3 2.4
250.0 300.0 9.6 -96.8 -- -- 225.0 -25.0 321.5 7.2
150.0 320.0 106.4 -66.8 105.3 -67.1 225.0 -29.7 323.4 1.1
0.0 275.0 95.0 -65.5 60.0 -78.2 225.0 -18.2 274.3 -0.3
-50.0 270.0 63.8 -76.4 40.1 -85.1 225.0 -16.7 264.6 -2.0
-70.0 260.0 51.3 -80.3 34.8 -86.6 225.0 -13.5 255.9 -1.6
-100.0 250.0 41.4 -83.4 30.4 -87.8 225.0 -10.0 243.2 -2.7
-130.0 250.0 35.9 -85.6 24.4 -90.2 205.6 -17.8 231.1 -7.6
-150.0 225.0 32.3 -85.6 22.2 -90.1 184.9 -17.8 225.9 0.4
-220.0 200.0 -- -- 10.5 -94.8 -- -- 200.3 0.2
-280.0 180.0 -- -- 0.0 -- -- -- 178.9 -0.6
-300.0 150.0 -- -- -- -- -- -- 164.2 9.5
-320.0 150.0 -- -- -- -- -- -- 149.2 -0.5
-350.0 110.0 -- -- -- -- -- -- 111.7 1.5
-380.0 80.0 -- -- -- -- -- -- 69.5 -13.1
SE 0.2851 0.1610 0.0752 0.0581
Table 3.8 Relative error and standard error comparison of interactive criteria for laminate failure of wind turbine
blade material under τ xy − σ yy field
125
126
The stress strain curves for the unidirectional loading case for stress
based criteria, strain based criteria and interactive criteria are shown in
Figures 3.20 to 3.22. It was observed that, all the curves are very similar in
shape for predictions made by stress based criterion. Among them, prediction
made by Hart-Smith is shallower than the others. Among the strain based
criteria, prediction made by Eckold extends only up to initial failure region.
Prediction made by Wolfe is very much conservative and not capable of
predicting the final failure. Predictions made by all interactive criteria are
similar in shape except for those of Rotem. In Rotem's analysis, there is an
abrupt increase in strain associated with initial failure, especially in
longitudinal direction. The slope of curve predicted by Sun was shallower
than the others.
-150
σ yy (MPa)
Prediction
Final
-100
Hart Smith[54]
-50 Sun[117]
Zinoviev[143]
Prediction
Initial
Modified
Experimental Data[28]
-150
σyy (MPa)
Prediction
Final
-100
Eckold[34]
-50 Hart Smith[54]
Prediction
Initial
Wolfe[137]
Modified
Experimental Data[28]
-1.5 -1 -0.5 0 0.5 1 1.5 2
0
εxx (%) ε yy (%)
-150
σyy (MPa)
Prediction
Final
-100
Edge[37]
Hart Smith[54]
Hashin[57]
-50 Puck[101]
Rotem[109]
Prediction
Tsai[133]
Initial
Modified
Experimental Data[28]
1.2
1.07
1.03
1.00 1.00
0.95 0.96 0.96
1
εtheo /εexp
0.90
0.8
0.64
0.6
0.4
0.2
v
ith
i
n
n
ed
ge
ck
fe
ie
hi
Su
Ts
ol
Sm
if i
Ed
Pu
ov
as
W
od
H
n
t
Zi
ar
M
H
Figure 3.27 Bar chart showing the ratio of predicted and experimental
strength of E-glass epoxy lamina for wind turbine blade
under uniaxial load in the loading direction
1.2 1.10
1.05
1.00
1 0.95
0.89 0.89 0.89 0.90
ε theo /ε exp
0.77
0.8
0.68
0.6 0.51
0.4
0.2
0
ai
n
ie v
d
n
ge
i th
k
tem
ed
fe
Su
l
i
c
Ts
ko
sh
ol
Ed
Pu
if i
Sm
ov
Ro
Ha
W
Ec
od
n
Zi
rt
M
Ha
Figure 3.28 Bar chart showing the ratio of predicted and experimental
strength of E-glass epoxy lamina for wind turbine blade
under uniaxial load in the transverse direction
129
The stress strain curves under biaxial tension (SR -1.81: 1) for
various criteria were shown in Figures 3.29 to 3.31.
1.2 1.15
1.03 1.05 1.05
1.02 1.02 1.02
0.99
1 0.94
εtheo /εexp
0.8
0.59
0.6
0.4 0.30
0.2
0
v
ith
i
d
n
n
ed
em
ge
ck
fe
a
ie
l
hi
Su
Ts
ko
ol
Sm
if i
Ed
ov
Pu
as
t
Ro
W
Ec
od
H
n
t
Zi
ar
M
H
Figure 3.32 Bar chart showing the ratio of predicted and experimental
strength of E-glass epoxy lamina for wind turbine blade
under biaxial load in the loading direction
1.6 1.60
ε theo /ε exp
1.4
1.2 1.10
1.00 1.00 1.00 1.00 1.00
1
0.77 0.80
0.8
0.6
0.4
0.2
0
Wolfe Hart Puck Modified Zinoviev Edge Sun Tsai Hashin
Smith
Figure 3.33 Bar chart showing the ratio of predicted and experimental
strength of E-glass epoxy lamina for wind turbine blade
under biaxial load in the transverse direction
132
Figures 3.34 to 3.36 shows the stress strain curves for the
unidirectional loading case for stress based criteria, strain based criteria and
interactive criteria of [0/ ± 45/90]s Laminate.
-550
σyy (MPa)
-500
Prediction
-450
Final
-400
-350
-300
-250
Sun[117]
Initial
-150 Zinoviev[143]
-100 Modified
Experimental Data[28]
-50
-1.5 -1 -0.5 0 0.5 1 1.5 2
0
ε xx (%) ε yy (%)
It was observed that, all the curves are very similar in shape for
predictions made by stress based criterion. Among them, prediction made by
Hart-Smith is shallower than the others. In transverse direction all criteria are
slightly unconservative. Among the strain based criteria, prediction made by
Eckold extends only up to initial failure region. Prediction made by Wolfe is
very much conservative and not capable of predicting the final failure.
Predictions made by all interactive criteria are similar in shape except for
those of Rotem. In Rotem's analysis, there is an abrupt increase in strain
associated with initial failure, especially in longitudinal direction. The slope
of curve predicted by Sun was shallower than the others.
-550
-500
σyy (MPa)
Prediction
-450
Final
-400
-350
-300
-250
Eckold[34]
-200 Hart Smith[54]
Prediction
Wolfe[137]
Initial
-150
Modified
-100 Experimental Data[28]
-50
-1.5 -1 -0.5 0 0.5 1 1.5 2
0
ε xx (%) ε yy (%)
Tsai and Modified and in four stages by Sun. In almost all the theories where
more than one stage of failure is predicted, the first stage was by transverse
tension in the matrix in the plies that were perpendicular to the load direction
and the final stage was tension along the fibers in the plies parallel to the
loading direction. Depending upon the predictive capabilities and error
involved in predicting the failure, it can be concluded that Puck, Modified,
Zinoviev, Hashin and Tsai relatively made good predictions.
-550
-500
σyy (MPa)
Prediction
-450
Final
-400
-350
-300
Edge[37]
-250 Hart Smith[54]
-200 Hashin[57]
Prediction
Puck[101]
Initial
-150 Rotem[109]
-100 Tsai[133]
Modified
-50 Experimental
-1.5 -1 -0.5 0 0.5 1 1.5 Data[28] 2
0
ε xx (%) ε yy (%)
The stress strain curves for [0/ ± 45/90]s Laminate under biaxial
tension (SR 1: 1) for various criteria were shown in Figures 3.37 to 3.39.
Similar to that of unidirectional loading, prediction made by various theories
were also remarkably similar to one another. In this case also Rotem showed a
step in his predictions. In this case also slope of curve predicted by Sun was
shallower than the others. Prediction made by Eckold terminates with initial
failure region itself.
135
900
SR = 1:2
SR -1:3
600
SR =1:1
σxx(MPa)
300
σyy (MPa)
0
-300 -200 -100 0 100 200 300 400 500 600
Hart Smith[54]
Sun[117]
Zinoviev[143]
-300 Modified
Experimental data[28]
SR = 1:-1
-600
800
σxx(MPa)
SR = -1:3 SR = 1:2
500
SR = 1:1
200
σyy (MPa)
-300 -200 -100 -100 0 100 200 300 400 500 600
-400 SR = 1:-1
Eckold[34]
Hart Smith[54]
Wolfe[137]
Modified
Experimental data[28]
-700
The standard error for all strain based criteria is more and is about
0.0992 for Hart-Smith and 0.1198 each for Wolf and Eckold’s prediction.
Among the interactive criteria (Figure 3.42), prediction made by modified
criteria, Puck, Hashin and Edge are fit well with experimental data points and
these criteria are capable of predicting maximum stress points. Predictions
made by Tsai and Rotem lie only in initial failure region. Prediction made by
Hart-Smith is partially good in one region and is unconservative in other
regions. The error comparison also favors modified criteria as a better one in
139
σxx (MPa)
SR = 1:2
800
SR = -1:3
SR = 1:1
500
200
σyy (MPa)
-300 -200 -100 -100 0 100 200 300 400 500 600
Edge[37]
Hart Smith[54]
Hashin[57]
-400 SR = 1:-1 Puck[101]
Rotem[109]
Tsai[133]
Modified
Experimental data[28]
-700
142
143
1.6 1.43
1.4 1.31
0.64
0.6
0.4 0.21
0.2
0
ai
n
ev
d
n
ge
ck
it h
tem
fe
ed
Su
l
i
Ts
ko
sh
vi
ol
Ed
Pu
if i
Sm
Ro
no
Ha
W
Ec
od
Zi
rt
M
Ha
Figure 3.43 Bar chart showing the ratio of predicted and experimental
strength for laminate of wind turbine blade material in
σ xx − σ yy field at stress ratio of 1:1
1.2 1.14
1.08
1.00 1.01 1.02 1.02 1.03
1
0.77
0.8 0.71
0.66
exp
σ /σ
0.6
theo
0.4
0.2 0.14
0
iev
i
n
d
ith
ge
k
em
ed
fe
Su
l
hi
c
Ts
ko
ol
Ed
ifi
m
Pu
ov
as
t
Ro
W
Ec
tS
od
H
n
Zi
r
M
Ha
Figure 3.44 Bar chart showing the ratio of predicted and experimental
strength for laminate of wind turbine blade material in
σ xx − σ yy field at stress ratio of 1:2
144
1.4 1.25
1.2 1.10
0.99
1 0.90 0.91
0.82
exp
σ /σ 0.8 0.73 0.73
0.59
theo
0.6 0.45
0.4
0.23
0.2
0
ie v
i
n
d
i th
ge
k
em
ed
fe
Su
l
c
Ts
ko
sh
ol
Ed
Pu
ifi
Sm
ov
t
Ro
Ha
W
Ec
od
n
rt
Zi
Ha
Figure 3.45 Bar chart showing the ratio of predicted and experimental
strength for laminate of wind turbine blade material in
σ xx − σ yy field at stress ratio of 1:-1
2.5
2.00
2
exp
1.10
1.00 1.01 1.02
theo
1 0.75 0.80
0.67
0.5
0
ai
v
ge
n
ck
ld
tem
ith
e
ed
Su
v ie
i
Ts
o lf
sh
ko
Ed
Pu
ifi
Sm
Ro
no
Ha
W
Ec
od
Zi
rt
M
Ha
Figure 3.46 Bar chart showing the ratio of predicted and experimental
strength for laminate of wind turbine blade material in
σ xx − σ yy field at stress ratio of -1:3
145