A Three-Phase Constitutive Model For Macrobrittle Damage of Composites

You might also like

Download as doc, pdf, or txt
Download as doc, pdf, or txt
You are on page 1of 40

A THREE-PHASE CONSTITUTIVE MODEL

FOR MACROBRITTLE DAMAGE OF


COMPOSITES

Ever J. Barbero* and Gasser Abdelal**


West Virginia University, Morgantown, WV 26506-6106

Abstract

Characterization of composite material behavior is currently limited to its elastic and


strength properties. Such information is not enough to study the remaining life of a
structure component or to estimate the likelihood of failure as a function of the time in
service. In the framework of Continuum Damage Mechanics, a micromechanical damage
model is proposed to study damage evolution in polymer-matrix laminate under
monotonic load. The micromechanical model accounts for damage in the matrix, fibers,
and the fiber-matrix interphase regions. A new damage surface is proposed and a step by
step model identification is presented. Theoretical results are compared with
experimental ones for verification of the micromechanical model.

* Associate Professor, West Virginia University.


** Graduate Research Assistant.

1
1 Introduction

In macrobrittle damage the only irreversible deformation due to damage. The analysis of
composite materials involves two categories, micromechanical and macromechanical.
Micromechanics is concerned with the mechanical behavior of constituent materials, the
interface of these constituents and the basic composite. Macromechanical is concerned
with the gross mechanical behavior of the composite materials and the structure without
regard for the constituents. Composite materials are usually anisotropic. Objectives of
present work are to develop a micromechanical orthotropic damage model for composite
material assuming elastic behavior of the constituents (small strain), and verification of
the theoretical model by comparing the numerical results with experimental data.

2 The Damage Model

Anisotropic damage is represented by second-order tensor. The effective stress is

  M ijkl kl

     kl I   ijkl  1 1 
M ijkl  ijkl kl  ijkl  ijkl   
1  ij 1  ij 2  1  C n ii 1  C s ii 
where the primed system is aligned with the principal directions of 

   Tik T jl  kl
  Tik T jl  kl

X  1 if X ij  0
X  0 if X ij  0

 ijkl = 1 if i = j = k = l
 ijkl = 0 otherwise

2
Here, Cn and Cs are coefficients that account for the effect of closure of voids and cracks
due to compressive stress or shear stress respectively. Cn = Cs = 0, if the crack closure
prevents further damage and Cn = Cs = 1, if the crack closure does not prevent further
damage at all.

Using the hypothesis of energy equivalence states the damaged elastic modulus is
1 1
Eijkl  M ijmn Emnpq M klpq

The main advantage of applying energy method is the symmetrization of the E-tensor
regardless of the nature of the M-tensor.

4.3Application to Elastic-Brittle Materials

The elastic-brittle material is damaged by the development of distributed microscopic


cracks and leads to the final fracture without significant inelastic deformation. A second
rank orthotropic damage tensor  will be employed to model the damage characterized
by these cracks. The internal state variables will be,
Vk = { ,  }
Where  is the increment of damage strengthening threshold (damage hardening). The
associated variables are in terms of the free energy potential 
1
 1 E ijkl
Y     ij  kl
 2  mn

  


Where the minus sign of the scalar conjugate force  is taken for the convenience of
further formulation. From Barbero and De Vivo [4],  is related to the damage hardening
parameter  through the relation
 (

)
  c1 1  e c2 
 

3
where c1 and c2 are the material parameters. A procedure will be discussed in the next
chapter to determine these values, so that the theoretical curve matches the experimental
data with an acceptable error. This is a very versatile hardening function, as we shall see
in later section.
The second rank symmetric tensor Y represents the energy release of the material due to
the development of damage. Then using eqn.

1  M pqij M rskl  1
Ymn   ij  M rskl  M pqij  E pqrs kl
2    mn 

In the absence of plastic deformation, the free energy potential of a damaged material
can be defined as
1
   ij Eijkl
1
 kl
2

5 Damage Criterion

A surface, defined in the Y-space exists from which damage flow can occur. For states
contained within this surface, the material behavior is entirely elastic. This g = 0 surface
is called the loading surface. The loading-unloading damage flow occurs only if the
following two conditions are satisfied:

(1) The representative point of Y-state is located on the loading surface,


g(Y, Vk) = 0.
(2) The point of the Y-state does not leave the loading surface. During the
continuous flow, the consistency condition must be met
dg = 0.

On the other hand, variation of Y is allowed to displace the state point towards the
interior of the loading surface. This is the case of unloading characterized by (dg < 0).
According to this scheme no damage flow can occur during unloading. In summary:
 g<0: elastic behavior.

4
 g = 0 and dg = 0: damage flow.
 g = 0 and dg < 0: elastic unloading.
Figure 1 illustrates the consistency condition (dg=0). In the case of loading the material
to a point outside the flow surface (dg > 0), the point will be projected back on the flow
surface using the inelastic strain.

5.1Nonassociated Damage

The generalized normality hypothesis associated with the instantaneous dissipative


phenomena allows us to write

Barbero and De Vivo [2] suggest the following formulations for the loading surface g,
and the potential surface F,

g (Y ,  )  Y  J d  Y  H d . Y  (   o )

F (Y ,  )  Y  J d  Y  (   o )

Where Jd is a fourth rank symmetric tensor and Hd is a second rank symmetric tensor,
called the damage characteristic damage tensors, with components

 J 11 0 0 
J d   0 J 22 0  ,
 0 0 J 33 

H d   H1 , H 2 , H 3 

The evaluation of these damage symmetric tensors is described by a step by step in


section 8. o is a material constant representing the initial damage threshold.

5
F
  
Y
F
  
 (  )
2F  F  2Y  2 F  Y  2 
 i       
 2  Y  2 Y 2    
 

From the consistency condition,


g g d 
dg  dY  d  0
Y  d 
the damage multiplier is defined as,
g
dY
   Y
2
 g  
   
 
The incremental relations in the undamaged and damaged configurations are, respectively

d ij  Eijkl d  kl
d ij  Dijkl d  kl

d ij  dM ijkl kl  M ijkl d kl
d  ij  dM ijkl  kl  M ijkl d  kl

From eqns. and , and assuming,


  n d

Where n is a proportional factor. Equation is exact in case the increment of strain is


constant during the damage process.

The stiffness tensor is determined by the following expression

1 1
D pqxz  M pqij Vijrs Z rsxz

6
where,
1
Vijrs  Eijrs  n dM ijkl M klxz E xzrs

Z uvrs  n dM uvrs  M uvrs

7 MICROMECHANICAL MODEL

The model presented in this work is a micromechanical approach that is similar to


Voyiadjis [9], but different in many aspects, as it will be discussed in detail later.

The unidirectional composite material consists of an elastic matrix with continuous and
aligned elastic fibers. The strain and the deformation are assumed to be small. The stress
(strain) in the composite can be related to the stress (strain) in the fiber and the matrix
applying the relations

 ij  c f  ijf  c m  ijm
 ijf  Bijkl
f
 kl ,  ijm  Bijkl
m
 kl

  ij  c f  ijf  c m  ijm 
 f 
  ij  Aijkl
f
 kl ,  m  Aijkl
m
 kl 

where cm and cf are the volume fractions for the fiber and the matrix respectively. These
volume fractions are defined by

Am and Af are the elastic strain concentration factors, and Bm and Bf are the elastic stress
concentration factors, determined by Micromechanics [2].
The micromechanical model presented in this research is built on a three-mode damage
concept. The first mode is concerned with the fiber breaking, the second one is concerned
with the matrix cracking, and the third one is concerned with the fiber-matrix debonding
damage.

7
Three configurations will be assumed to model the three damage modes. These three
configurations are shown in Figure 2. The first configuration is the effective one, where
all the damage has been removed and the constituents are in their virgin state. That means
the damaged area (volume) is removed, which will lead to change in the volume fractions
of the constituents. This configuration will be designated by the bar (  ) symbol. The
second configuration is the partially damaged one, where the only damage modes that
take place are in the fiber and (or) the matrix and no debonding damage. This
configuration will be designated by the tilde (  ) symbol. The third configuration is the
damaged (real) one, where debonding damage is assembled to the partial damaged
configuration. Each configuration has a number of assumptions that will be discussed in
the next sections.
The equations that relate the stress (strain) in the effective configuration and the partial
damaged one are
 ijf  M ijkl
f
 klf
 ijm  M ijkl
m
 klm

  ijm  M ijkl
f
klf 
 m 
  ij  M ijkl
m m 

 kl 

where Mf and Mm are the fourth-order damage tensor for the fiber and the matrix
respectively (see eqn. (2.6)). The equations that govern the relation between the partial
damaged configuration and the damaged one are
 ij  M ijkl
d
 kl

  ij
d
 M ijkl  kl 
where Md is the fourth-order damage tensor for the debonding interface between the fiber
and the matrix.

The effective configuration represents a state of the material where damage has been
removed and the material behaves elastically. The stress (strain) in the constituents is
related to the overall stress in the composite through the equations

8
 ijf  Bijkl
f
 kl ,  ijm  Bijkl
m
 kl

 ij
f
 Aijkl
f
 kl ,  m  Aijkl
m
 kl 

where A and B are the strain and stress concentration factors in the effective
configuration respectively.
The relation between the stress and the strain in the constituents (composite) is governed
by the equations
 ijf  Eijkl
f
 klf ,  ijm  Eijkl
m
 klm

  ij  Eijkl
c
 kl 
 
 E  c f E f A f  c m E m Am 
 ijkl ijmn mnkl ijmn mnkl 
where E f , E m and E c are the fourth-order elasticity tensor for the fiber, matrix and the
composite respectively in the effective configuration. These tensors are assumed to
constant through the damage process. As the norm of the second-order damage tensor 
change from zer to one, through the damage process, the elasticity tensor of the
constituents (composite) will remain unchanged.

The volume fraction ( c f , c m ) of the constituents will change with the damage process as
the damaged volume is assumed to be removed in the effective configuration. Caceres [1]
proposed the following equations to relate the volume fraction of the constituents in the
effective configuration to the second-order damage tensor  as a function of the real

(damaged) volume fraction ( c f , c m ),

 
0.5
 req   rij  rij , r  f ,m

c f

 1   f
eq
m

 1     cc  1   
f
eq f
m
eq

c m

 1   m
eq
f

 1     cc  1   
m
eq m
f
eq

9
The partially damaged configuration is concerned with the damage in the fiber and the

matrix with no debonding (interface) damage. The volume fractions ( c f , c m ) are


assumed to be constant during the damage process and have the same values of the

volume fractions ( c f , c m ) in the damaged configuration. The fourth-order elasticity


tensor of the constituents (composite) in the partially damaged configuration is related to
the elasticity tensor in the effective configuration through the equations
r r
r
E ijkl  M ijmn r
Emnpq M klpq , r  f ,m

 E ijkl  c f E ijmn
f
A mnkl
f
 c m E ijmn
m
A mnkl
m

Where A ,( B ) are the strain (stress) concentration factors in the partially damaged
configuration that can be related to the concentration factors in the effective configuration
through the equations

 
1
m f f
A ijkl
f
 c m M ijpq m
Apqrs Arsmn M mnkl  c f I ijkl

A ijkl
m

I ijkl  c f A ijkl
f

m
c


 B f  c m M  m B m B  f M f  c f I  
1

 ijkl ijpq pqrs rsmn mnkl ijkl





I  c f B ijkl
 B m  ijkl
f
 

 ijkl 
 c m 

where I ijkl is a fourth-order identity tensor defined by


1
I ijkl  ( ik  jl   il  jk )
2
 ij  1 if i  j
 ij  0 if i  j

The damaged configuration the debonding damage is determined and assembled with the
damage in the fiber and the matrix evaluated in the partially configuration to determine
the overall damage in the composite. The constituents properties are the same in both the
partially damage and the damaged configurations. Thus,

10
 r   r ,  r   r , r  f,m
The transformation of the fourth-order elasticity tensor of the composite from the
partially damage configuration to the damaged one is governed by the equation,

d d
Eijkl  M ijmn E mnpq M klpq

Where Md is the fourth-order tensor that represents the debonding (interface) damage
between the fiber and the matrix. Stress and strain in the composite can be determind-
applying eqn. .

For Finite element application in ABAQUS the stiffness tensor D for the composite is
needed to be determined at each strain interval. Eqn. will be applied to evaluate the
stiffness tensor for the composite

t 1
D pqxz  M pqij Vijrs Z rsxz

where,
t
Vijrs  E ijrs
c
 n dM ijkl
t
M klxz E cxzrs ,

Z uvrs  n dM uvrs
t
 M uvrs
t

t
M ijrs 
 c f M ijkl
f f
B kluv  c m M ijkl
m m
Bkluv  d
M uvrs

and Mt is the total damage in the composite.

7.1 Numerical Formulation

The previous theoretical section will be applied in a numerical algorithm to evaluate the
stress-strain curve of the composite material under different applied loads. The numerical
algorithm consists from four stages. In the first stage, the material properties and the
material parameters are defined for the constituents and the debonding interface. In the
second stage, the stress and the damage in the constituents are evaluated. In the third

11
stage, the debonding damage is determined. Finally in the fourth stage, the total damage
and the total stress in the composite are evaluated. An iteration procedure is performed on
the d in stages two and three. The first advantage of applying this iteration is the
improved numerical results.

A set of flowcharts will be used to illustrate the numerical algorithm. Figure 3 illustrates
the four stages. Then Figure 4 will illustrate in detail the second and the third stages. The
program will stop if it achieved one of the following failure criteria,
 Tsai-Hill failure envelope,

 1    1  2     2     6     4    5 
2 2 2 2 2

 1
 F1   F1   F2   F6   F4   F5 
2 2 2 2 2 2

  CR m  CR
f f m
or .

8 Material Parameters and Numerical Simulation

The damage theory that was discussed in the previous chapters implies a number of
material parameters that need to be determined. For each phase, they are
 E material modulus of elasticity.
 cr Volume fraction of the constituents. (r = f, m)
  Poisson ratio of the constituents.
 Jd1, Hd1 damage characteristic tensor.
 C1, C2 Hardening parameters.
 Cn, Cs the coefficients of closure.
 Tol the tolerance of the constituent that needed for the iteration on d (see
flowchart B).

E, cr, and  are the material constants, while the rest are the material parameters. The
material constants and parameters for the debonding interface are
  Poisson ratio of the composite.
 Jd, Hd damage characteristic tensor.
 C1, C2 Hardening parameters.
 Cn, Cs the coefficients of closure.

12
 Tol the tolerance of the constituent that needed for the iteration on d (see
flowchart B).
The fiber and the matrix are assumed to be isotropic materials while the debonding
interface is assumed to be transversally isotropic.
The characteristic damage tensors Jd and Hd are function of the damage critical values
and the tensile and the compressive strength of the material. These values can be found
from experimental or based on reasonable assumptions.

The sign of the material parameters C1 and C2 from eqn. will be determined by the
prediction of how the material will harden during the damage process. Figures 5 and 6
illustrate two types of hardening behavior. Minimization of the error between the
experimental and the theoretical results will determine their values.

8.1 Critical Damage and Failure Strength Values

Damage domain is defined by the damage characteristic tensors J and H.These damage
tensors with the hardening law will define the initiation, evolution and fracture of the
material. These tensors were derived in a way that makes the damage domain resemble
the Tsai-Wu failure criteria for composites.

 C 11  C 11
J 11  6  F12t  H1 6
 1
  1t   1t

 C 11  C 11
J 11  6  F12c  H1 6
 1
  1c   1c

 C 22  2 C 22
J 22 
  F2 t 
6 
H2
 2t
6
 1
 2t 
H2   r1 H 1

where,

13
 F1t, F2t are the longitudinal and transverse tensile strength respectively. F1c is
the longitudinal compressive strength.
 C11 and C22 are the first two diagonal elements of the compliance 2-D tensor

E .
1

 J11, J22 are elements from the damage characteristics tensor Jd. H1 and H2 are
elements from the damage characteristic tensor Hd, and r1 is a ratio between
these two elements.
  is the damage integrity tensor,

  1

and 1t, 2t and 1c are the critical damage values in the longitudinal, transverse tensile
directions and longitudinal compressive direction respectively.
The experimental data available for the fiber, the matrix and the debonding interface are
not sufficient to determine the damage critical values and the failure strength properties.
The following assumptions are made for the prediction of these values

 The critical damage of debonding interface for tension in the longitudinal


direction is equal to that of the fiber since loading in that direction is
dominated by the fiber properties. The critical fiber damage under tension by

Kelly and Barbero [34] is 1t  0.1 for fiber weibull modulus m=9.
 The critical damage of fibers and debonding interface for compression in the
longitudinal direction is equal to that of the composite. This is based on the
assumption that the fibers depend on the matrix for lateral support. In essence,
the critical damage is
  
1c  1  erf  CR 
 2

14
where erf ( ) is the error function,  is the standard deviation of fiber
misalignment, and the critical misalignment angle CR is given by Barbero
[44]

 The critical damage of the debonding interphase for the transverse directions
equals that of the matrix. This is based on the assumption that loading in the
transverse direction is dominated by the matrix properties. Based on the
Janson-Hutt model [37] the critical damage value for any matrix is

 mt  0.5

The numerical values are in Table 1.

8.2 Damage Characteristic Tensors

The loading surface expressed by eqn. is a function of the damage characteristic tensor
(Jd, Hd). There is no exact procedure for the evaluation of these tensors. The available
ones in the literature offer only approximate values for these tensors. The procedure
proposed by Barbero and De Vivo [2] assumes 2-D state of stress and transversely
isotropic behavior of the composite as a continuum. This is not the case for the
micromechanical model presented in this work, which represent 3-D state of stress and
isotropic behavior for the fibers (matrix) and transversely isotropic behavior for the
debonding interface. Barbero and De Vivo procedure is a good approximation for the
case of unidirectional composite, which is the focus of our work. at the previous
procedure has to be simplified for the case of isotropic materials (fiber and matrix).

8.2.1 Isotropic Material (fiber, matrix)

In case of longitudinal uniaxial load, Barbero and De Vivo [2] proposed the following
equations in terms of the tensile and compressive strength (F 1t, F1c) and the critical
damage values in the tensile and compressive loading conditions (1t, 1c). The critical
damage values were estimated in the previous section.

15
Equations and will be solved for J11 and H1 values. The same two equations will be
solved in the transverse direction to determine J 22 and H. But as the material is assumed
to be isotropic,

J 11  J 22  J 33

H1  H 2  H 3

A program written in MapleV5.0 to evaluate the damage characteristic tensors for


isotropic materials, which is listed in appendix A. The resulting values for M40/949 are
presented in table 2.

8.2.2 Transversely isotropic material (debonding interface)

The determination of the damage characteristic tensors (Jd, Hd) of the debonding interface
between the fiber and the matrix will be based on the assumption that it has transversely
isotropic behavior.

Similar to the procedure described by Barbero and De Vivo [2] for continuum composite,
the damage characteristic tensors (Jd, Hd) will be evaluated. First, the critical damage
values for the debonding interface will be determined according to section 8.1. Then
eqns. , , , and will be solved simultaneously to evaluate ( J 11 , J 22 and H 1 , H 2 ) . The
transversely isotropic behavior of the debonding interface with respect to direction (1)

will lead to ( J 33  J 22 , h3  h2 ) . A program written in MapleV5.0 to evaluate the

damage characteristic tensors for transversely isotropic material is listed in appendix A.


The resulting values for M40/949 are presented in table 2.

16
8.3 Model Identification

The numerical algorithm is tested with the experimental test results of the (M40/949)
composite material. The material constants are listed in the table 1. There are 20 material
parameters needed to be determined and they are,

 C1, C2, o, Cn, and Cs for the fiber.


 C1, C2, o, Cn, and Cs for the matrix
 C1, C2, o, Cn, and Cs for the debonding interface.
The following procedure will be applied to determine these parameters,
1. Let ,
C nf  C nf  C nf  1
C sf  C sf  C sf  1
C1d  1, C 2d  2*104 ,    7.0
Based on experimental results the damage in debonding interphase will be assumed to
have no effect on the composite performance. So, set the debonding interphase
parameters to a value so that debonding damage is minimum.
2. The experimental tension test in the longitudinal direction results will be used to

determine the sign and the value of C1 , C 2 ,  o . Thus these three parameters will be
f f m

adjusted until the theoretical results will match the failure criteria of the fiber, as it is
assumed that the composite material behavior in the longitudinal direction is
dominated by the fiber properties. That is, at failure we should get
 The stress in the fiber is the strength of the fiber bundle.
 The damage coincides with the critical value and Tsai-Hill surface is almost
reached.

Figures 7 and 8 show strain-stress and strain-damage curves respectively of the


composite under longitudinal tension load and it shows that the fiber fails at   0.1

and  f  4 GPa . Under these condition we set

17
C1f   1, C 2f  8*105 ,  Of   7

3. The experimental tension test in the transverse direction results will be used to

determine the sign and value of C1 , C 2 ,  O . Both the matrix failure and the Tsai-
m m m

Hill criteria will be used to adjust these parameters, as it is assumed that the
composite material behavior in the transverse direction is dominated by the matrix
properties. The previously determined parameters are held constant.

Figures 9 and 10 show strain-stress and strain-damage curves respectively of the


composite under transverse tension and it shows that the matrix fails at   0.5 and

 f  50 MPa .

C1m  1, C 2m   2.15*107 ,  om  1.95

The resulting values for M40/949 are presented in table 2.

8.4 Model Verification

The above model identification will be used in a shear test for the M40/948 and
compared with results that was obtained from experimental shear Barbero and Wen, Ad
[43].

9. Conclusions

A micromechanical model is proposed to study damage in polymer-matrix laminate.


Damage in the composite in the longitudinal direction is dominated by damage in the
fibers. While damage in the composites due to transverse or shear loading is dominated
by damage in the matrix. Debonding damage is a small factor in composite damage,
although it can be used as an extra parameter for model identification but it can not be
verified easily through experimental results. The micromechanical model is similar to
Voyiadjis model but different in many aspects. The first aspect is the new damage

18
surface, which give better results in the shear direction. The second aspect is the model
identification procedure. The third aspect is the failure criteria, which depends on both
the Tsai-Hill criteria and damage accumulation in the composite. The proposed model
can be implemented in a finite element program to study damage in laminated composite.
Thus the damage evolution through each layer and load transfer from one layer to another
can be simulated. This application will be proposed in a subsequent paper.

Appendix A

Elastic Stress and Strain Concentration Tensors

The expressions for the elastic stress and strain concentration tensors given here are
based on Luciano and Barbero [5]. The equivalence principle of Nemat-Nasser [6] is used
to find the stresses in the fibers where the stress field in the fiber material is the same as
the stress of the inclusion.

The stress concentration tensor for the fibers is given by

1
  m
Pmnrs Erspq  f 
f
Bijkl   I ijkl  c Eijmn  I mnpq 
m m
 c m  m
 E pqkl  E pqkl  
   

Where the fourth-order tensor P is a modified version from Nemat-Nasser.

1 
 1 
P
m
 t ( )  2* sym(  I ( 2)
 ) 
2(1   m )
      
  

The following series will be defined,

19
  
S1   t ( ) 12 , S2   t ( )  22 , S 3   t ( )  32 ,
  
  
S4   t ( )  14 , S5   t ( )  14 , S6   t ( )  34 ,
  
  
S7   t ( )  22  32 , S8   t ( )  12  32 , S9   t ( )  12  22
  

All odd powers, such as   1  3 , vanish because of the orthogonality of the Fourier
expansion in Lucano-Barbero [5].

In terms of the series Si, the P-tensor can be expressed in the reduced (6x6 matrix),

 2 S1 S4 S9 S8
   2 (1   ) 
2 m (1   m )

2  m (1   m )
 m m m

 2 S2 S5 S7
 symm  
 m 2  m (1   m ) 2  m (1   m )
P 2 S3 S6
 symm symm 
 m 2 m (1   m )
 symm symm symm

 symm symm symm
 symm symm symm

0 0 0 
0 0 0 

0 0 0 

S3  S2 S4 
 0 0
3 m 2 m (1   m ) 

S1  S 3 S8 
symm  0 
3 m 2  m (1   m )

S1  S 2 S9 
symm symm 
3 m 2  m (1   m ) 

For unidirectional composites with long circular cylindrical fibers, the following
expressions apply

S1  S4  S8  S 9  0
S2  S 3
S5  S 6

20
and according to Luciano-Barbero [5]

S3  0.49247  0.47603 c f  0.02748 c f 2


S6  0.36844  0.14944 c f  0.27152 c f 2
S7  0.12346  0.32035 c f  0.23517 c f 2

The stress concentration tensor for the matrix can be determined with

B m

I ijkl  c f Bijkl
f

ijkl m
c

The strain concentration tensor in the fiber is

 
1
f
Aijkl   I ijkl  Pijmn Emnkl
f
 Emnkl
m

And the strain concentration tensor for the matrix can be determined with

A m

I ijkl  c f Aijkl
f

ijkl m
c

21
REFERENCES

[1] Barbero, E. and Caceres, A., “Local Damage Analysis of Fiber Reinforced
Polymer Matrix Composites”, Ph.D., Mechanical Engineering, West Virginia
University, (1998).

[2] Barbero, E. and De Vivo, L., “A Constitutive Model for Damage in Fiber-
Reinforced Laminae”, Ph.D.

[3] KACHANOV, L. M., ''On the Time to Failure under Creep Conditions', Izv, AN
SSSR, Otd. Tekhn. Nauk, No. 8, pp. 26-31, (1958)

[4] RABOTNOV, Y. N., ''Creep Rupture'', in Proceedings XII International Congress


on Applied Mechanics, Stanford, Springer, Berlin, (1968).

[5] LEMAITRE, J., and CHABOCHE, J. L., ''A Non-linear Model of Creep-Fatigue
Damage Cumulating and Interaction'', Proceedings IUTAM Symposium on
Mechanics of Visco-elastic Media and Bodies}, Springer, Gothenburg, (1974).

[6] HULT, J., On Topics in Applied Continuum Mechanics, (J. L. Zeman and F.
Ziegler, eds.), Springer, New York, (1974).

[7] LECKIE, F. A., ''The Constitutive Equations of Continuum Creep-Fatigue


Damage Cumulation and Interaction”, Philosophical Transactions Royal Society
of London}, vol. A288, pp. 27, (1978).

[8] MURAKAMI S., ''Notion of Continuum Damage Mechanics and its Applications
to Anisotropic Creep Damage Theory'', Journal of Engineering Materials
Technology}, vol. 105, pp. 99-105 (1983).

22
[9] VOYIADJIS, G. Z., and PARK, T., ''Local and Interfacial Damage Analysis of
Metal Matrix Composites'', International Journal of Engineering Sciences}, vol.
33, No. 11, pp. 1595-1621, (1995).

[10] MROZ, Z., ''Mathematical Models of Inelastic Material Behavior'', University of


Waterloo, pp. 120-146, (1973).

[11] KACHANOV, L.M., ''Introduction to Continuum Damage Mechanics'', Martinus


Nijhoff, Dordrecht, The Netherlands, (1986).

[12] LEMAITRE, J. and CHABOCHE, J. L., ''Mechanics of Solid Materials'',


Cambridge University Press, (1990).

[13] HAYHURST, D. R., ''Creep Rupture under Multiaxial State of Stress”, Journal of
Mechanics and Physics of Solids}, vol. 29, No. 6, pp. 381-390, (1972).

[14] ARNOLD, S. M. and KRUCH S., ''Differential Continuum Damage Mechanics


Models for Creep and Fatigue of Unidirectional Metal Matrix Composites,''
NASA Technical Memorandum} TM-105213, (1991).

[15] MURAKAMI, S., ''Mechanical Modeling of Material Damage”, Transactions of


the ASME, vol. 55, pp. 280 Design,'' Taylor & Francis, (1989)

[16] LUCIANO, R. and BARBERO, E. J. ''Formulas for the Stiffness of Composites


with Periodic Microstructure'', International Journal of Solids and Structures, vol.
31, No. 21, pp. 2933-2944, (1994).

[17] TALREJA, R., ''A Continuum Mechanics Characterization of Damage in


Composite Materials”, Proceedings Royal Society of London, A399, 195, (1985).

[18] CHRISTENSEN, R. M., ''Tensor Transformation and Failure Criteria for the
Analysis of Composite Materials'', Journal of Composite Materials, Proceedings
Royal Society of London, A399, 195, (1985).

[18] CHRISTENSEN, R. M., ''Tensor Transformation and Failure Criteria for the
Analysis of Composite Materials'', Journal of Composite Materials, 22, 874,
(1988).

[19] HILL, R., ''A Self-Consistent Mechanics of Composite Materials'', Journal of the
Mechanics and Physics of Solids, 13, 213, (1965).

[20] DVORAK, G. J., and BAHEI-EL-DIN, Y. A., ''Elastic-Plastic Behavior of


Fibrous Composites'', Journal of Mechanics and Physics of Solids, 48, 809,
(1979).

[21] NEEDLEMAN, A., ''A continuum model for void nucleation by inclusion

23
debonding'', Journal of Applied Mechanics, Vol. 54, pp.525-531, (1987).

[22] TVERGAARD, V., ''Effect of fiber debonding in a whisker-reinforced metal'',


Materials Science and Engineering}, Vol. A125, pp. 203-213, (1990).

[23] ABOUDI, J., ''Constitutive Equations for Elastoplastic Composites with Imperfect
Bonding'', International Journal of Plasticity, Vol. 4, pp. 103-125, (1988).

[24] TEPLY, J. L., and DVORAK, G. J., ''Dual Estimates of Overall Instantaneous
Properties of Elasto-Plastic Composites'', 5th International Symposium on
Continuum Models of Discrete System}, Nottingham, July 14-20, pp. 205-216,
(1985).

[25] CHRISTMAN, T., NEEDLEMAN, A., NUTT, S., and SURESH, S., ''On the
Microstructural Evolution and Micromechanical Modeling of Deformation of a
Whisker Reinforced Metal Matrix Composite”, Materials Science and
Engineering A, Vol. 107, pp. 49-61, (1989).

[26] SHERWOOD, J. A., and BOYLE, M. J., ''Investigation of the Thermomechanical


Response of a Titanium-Aluminide/Silicon-Carbide Composite Using a Unified
State Variable Model and the Finite Element Method”, Microcracking-Induced
Damage in Composites}, AMD-Vol. 111., G. J. Dvorak and D. C. Lagoudas, eds.
American Society of Mechanical Engineers, pp. 151-161, (1990).

[27] BARBERO, E., and GANGARAO, H. V. S., “Structural Applications of


Composites in Infrastructure. Part I.”SAMPE Journal, Vol. 27, No. 6, pp. 9-16,
(1991)

[28] BARBERO, E., and GANGARAO, H. V. S., “Structural Applications of


Composites in Infrastructure. Part II.” SAMPE Journal, Vol. 28, No. 1, pp. 9-16,
(1992).

[29] DAVIDSON, L., and STEVENS, A. L., “Thermomechanical constitution of


spalling elastic bodies.”Journal of Applied Physics, Vol. 44, pp. 667-674, (1973).

[30] KRAJCINOVIC, D., and FONSEKA, G. U., “The Continuous Damage Theory of
Brittle Materials.” Journal of Applied Mechanics, Vol. 48, pp. 809-824, (1981).

[31] OWEN, D. R. J., and HINTON E., “Finite Element in Plasticity: Theory and
Practice.” Pineridge Press Limited, pp.249-253, (1980).

[32] STUMVOLL, M. and SWOBODA, G., “Deformation Behavior of Ductile Solids


Containing Anisotropic Damage.” Journal of Engineering Mechanics, Vol. 119,
No. 7, pp. 1331-1352, (1993).

[33] DVORAK, G. J., “Dual Estimates of Overall Instantaneous Properties of Elasto-

24
Plastic Composites.” 5th International Symposium on Continuum Models of
Discrete System, Nottingham, July 14-20, pp. 205-216, (1985).

[34] KELLY, K. W., and BARBERO, E., “The Effect of Fiber Damage on the
Longitudinal Creep of a CFMMC.” International Journal of Solid Structures, Vol.
30, No. 24, pp. 3417-3429, (1993).

[35] ROSEN, B. W., “Tensile Failure of Fibrous Composites.” AIAA Journal, Vol. 2,
No. 11, (1964).

[36] LEMAITRE, J., “A Course in Damage Mechanics.” Springer-Verlag, (1992).

[37] JANSON, J., and HULT, J., ''Damage Mechanics and Fracture Mechanics - A
Combined Approach.” Journal de Mechanique Applique, Vol 1, pp.69-84, (1977).

[38] BARBERO, E., “Introduction to Composite Materials Design.” Taylor & Francis,
(1999).

[39] IWAKUMA, T., and NEMAT-NASSER, S., “Composites with Periodic


Microstructure.” Computers and Structures, Vol. 16, No. 1-4, pp. 13-19, (1983).

[40] Voyiadjis, G. Z. and Echle, R. “ A Micromechanical Fatigue Damage Model for


Unidirectional Metal-Matrix Composites.” Applications of Continuum Damage
Mechanics to Fatigue and Fracture, ASTM STP 1315, D.L. McDowell, Ed.,
American Society for Testing and Materials, pp. 97-115, (1996).

[41] Cailletaud, G. and Levaillant, C., “Creep-Fatigue Life Prediction: What about
Initiation?.” Nucel. Engng. Design., Vol. 83, pp. 279-292, (1984).

[42] Socie, D.F., Fash, J.W. and Leckie, F.A., “A Continuum Damage Model For
Fatigue Analysis of Cast Iron.” ASME, Conf. in Life Prediction, Albany, N.Y.,
(1983).

[43] Barbero, E. and Wen, AD. (1999) Compressive Strength Prediction for Composite
Unmanned Aerial Vehicles, MSAE Thesis, West Virginia University.

25
Table 1. Material constants for the (M40/948) composite Material.

FIBER MATRIX DEBONDING INTERPHACE


E Modulus, (Gpa) 379.2 3.1
Poisson’s ratio  0.2 0.38 0.49
Initial volume fraction 0.6 0.4
Ft, (Gpa) 4.8 0.0572 0.0572
Fc, (Gpa) 1.35 0.1876 1.096
t 0.105161 0.5 0.105161
c 0.110945 0.5 0.110945
F6, (Gpa) 0.08616
G12, (Gpa) 204.166 1.2929 5.99

Table 2. Material Parameters for the (M40/948) composite Material.

FIBER MATRIX DEBONDING


INTERPHACE
-14 -14
H1 .287913975*10 .130654*10 .2583*10-11
H2 = H3 .287913975*10-14 .130654*10-14 .2168108*10-13
J11 -.31148207*10-6 -.201863*10-6 -.00003439
J22 = J33 -.31148207*10-6 -.201863*10-6 .156764*10-6
C1 -1.0 1.0 1.0
C2 8*105 -2.15*106 -9*106
o -7.0 1.95 -7.0
Cn 1.0 1.0 1.0
Cs 1.0 1.0 1.0

26
Loading surface, g = 0

E t
Y
D i
3

Y2

Y1

Fig. 1. Schematic diagram to illustrate the consistency condition (dg = 0).

27
  c   c 
f f m m
 m Matrix damage
 m
 m  Bm :  START

  Debonding
damage The material 
Input the parameters are

  M :  d
material defined ichapter
4.
parameters
  B :
f f

 f Fiber damage

Evaluate Ef, Em
f

  M : 
f f f
and Ec in the
effective
configuration.

 i 1   i   
Fig. 2. Micromechanical Model Representation of the Damage in Composites.

Evaluate
stress in the
 i 1  E  i 1
composite

 ,M f f
 m, Mm
Illustrated in Illustrated in
details in Evaluate Evaluate details in
flowchart B. stress and stress and flowchart B.
damage in damage in
the fiber the matrix

28
To be continued
in the next page
Applying eqn.
Assemble (3.1).
constituents
stresses

Evaluate Flowchart B.
damage in
debonding
interphase.

 c , M t , E c , Dc
Evaluate new states
Applying eqns.
of the composite. (3.12), (3.13),
and (3.14).

Fig. 3. Flowchart ‘A’.


Save the stress-
strain curve in a file

29
Evaluate stress Applying eqn. (3.1).

 or 
f m
or  c in the
constituent
The debonding the stresses are
assembled first to evaluate c in the
partially damaged configuration.

Evaluate Y- Applying eqn.


(2.35)
tensor

Evaluate Applying eqn.


hardening  (2.32)

Applying
eqn. (2.37)
Check g

Fig. 3. Flowchart ‘A’(Continued).

g<0 g>0

 r   r  Dr  i
d ,  ,  i

E r  M  r E r M  r Evaluate damage Applying eqns.


increment, (2.36). Iterate on
Update the E
d,
Return and the stress inelastic strain
(dnew-dold) < 
tensors. and hardening .
30
Fig. 4. Flowchart ‘B’.

31
10

6
 ( )

d (  )
4

0
0 0.5 1 1.5 2

Fig. 5. The hardening law and its derivative in case of (C1  0, C 2  0) .

 ( )

d (  )

2
0 0.5 1 1.5 2

Fig. 6. The hardening law and its derivative in case of (C1  0, C 2  0) .

Fibers

Composite
32
Matrix

Fig. 7. Theoretical Strain-Stress curves for the M40/949 longitudinal tension test.

Fibers

Matrix

Interphase

Fig. 8. Theoretical Strain-Damage curves for the M40/949 longitudinal tension test.

33
Fibers

Composite

Matrix

Fig. 9. Theoretical Strain-Stress curves for the M40/949 Transverse tension test.

Matrix

Debonding

Fibers

Fig. 10. Theoretical Strain-Damage curves for the M40/949 Transverse tension

34
Theoretical Composite Shear
Fibers

Experimental composite
Experimental Compositeshear
Shear

Matrix

Fig. 11. Theoretical Strain-Stress curves for the M40/949 Shear test.

Matrix

Debonding Fibers

35
Fig. 12. Theoretical Strain-Damage curves for the M40/949 Shear test.

36
37
38
39
40

You might also like