11.02.2013 Views

Computational Methods for Debonding in Composites

Computational Methods for Debonding in Composites

Computational Methods for Debonding in Composites

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

306 H. Miled et al.<br />

Eqs. 15.28 and 15.30 lead to the Advani and Tucker model [1] which expresses<br />

the effective stiffness tensor as a function of the second and fourth orientation<br />

tensors:<br />

C c � � � �<br />

ijkl p = b1aijkl+ b2 (aijδkl + aklδij)+b3 aikδ jl + ailδ jk + a jkδil + a jlδik<br />

� �<br />

+b4 (δijδkl)+b5 δikδ jl + δilδ jk<br />

(15.31)<br />

An averag<strong>in</strong>g on the compliance tensor gives an expression of the anisotropic<br />

tensor which is similar to Eq. 15.31, with five constants m1,...,m5:<br />

S c ijkl = m1aijkl+<br />

� �<br />

m2 (aijδkl + aklδij)+m3 aikδ jl + ailδ jk + a jkδil + a jlδik<br />

� �<br />

+m4 (δijδkl)+m5 δikδ jl + δilδ jk<br />

(15.32)<br />

In spite of SUD be<strong>in</strong>g the <strong>in</strong>verse of CUD , the effective compliance tensor is not<br />

the <strong>in</strong>verse of the stiffness tensor given by Eq. 15.31. This represents the weakness<br />

of the two levels approach. Many authors [18, 34], prefer to use the stiffness<br />

tensor to describe the anisotropic properties because of his better agreement with<br />

experiments.<br />

The effective thermal expansion tensor can be also expressed as a function of the<br />

second order orientation tensor [1]<br />

α c = P1a + P21 (15.33)<br />

P1 and P2 are two constants related to the unidirectional thermal properties:<br />

15.4 Results and Discussion<br />

P1 = α1 − α2<br />

P2 = α2<br />

(15.34)<br />

15.4.1 Choice of a Micromechanical Model <strong>for</strong> the Unidirectional<br />

Properties<br />

In order to compare the three models listed <strong>in</strong> Sect. 15.3, a composite re<strong>in</strong><strong>for</strong>ced<br />

by spherical <strong>in</strong>clusions is considered (β = 1). The predicted longitud<strong>in</strong>al Young<br />

modulus is compared to experimental measurements conducted by Simth [38].<br />

The Young’s modulus of the matrix and particles are respectively Em = 3GPa,<br />

νm = 0.4, E f = 76 GPa, and ν f = 0.23. The comparison between predicted and<br />

effective longitud<strong>in</strong>al Young modulus is done <strong>for</strong> <strong>in</strong>creas<strong>in</strong>g fiber volume fractions<br />

(Fig. 15.9).<br />

Figure 15.9 shows that Mori-Tanaka model gives the better prediction of the<br />

longitud<strong>in</strong>al Young modulus <strong>for</strong> volume fraction go<strong>in</strong>g up to 20–30%.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!