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

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Chapter 2<br />

Material and Failure Models<br />

<strong>for</strong> Textile <strong>Composites</strong><br />

Raimund Rolfes, Gerald Ernst, Matthias Vogler, and Christian Hühne<br />

Abstract The complex three-dimensional structure of textile composites makes<br />

the experimental determ<strong>in</strong>ation of the material parameters very difficult. Not only<br />

the number of constants <strong>in</strong>creases, but especially through-thickness parameters are<br />

hardly quantifiable. There<strong>for</strong>e an <strong>in</strong><strong>for</strong>mation-pass<strong>in</strong>g multiscale approach <strong>for</strong> computation<br />

of textile composites is presented as an enhancement of tests, but also<br />

as an alternative to tests. The multiscale approach consists of three scales and<br />

<strong>in</strong>cludes unit cells on micro- and mesoscale. With the micromechanical unit cell<br />

stiffnesses and strengths of unidirectional fiber bundle material can be determ<strong>in</strong>ed.<br />

The mesomechanical unit cell describes the fiber architecture of the textile composite<br />

and provides stiffnesses and strengths <strong>for</strong> computations on macroscale. By<br />

comparison of test data and results of numerical analysis the numerical models are<br />

validated.<br />

To consider the special characteristics of epoxy res<strong>in</strong> and fiber bundles two<br />

material models are developed. Both materials exhibit load dependent yield behavior,<br />

especially under shear considerable plastic de<strong>for</strong>mations occur. This non-l<strong>in</strong>ear<br />

harden<strong>in</strong>g is considered via tabulated <strong>in</strong>put, i.e. experimental test data is used<br />

directly without time consum<strong>in</strong>g parameter identification. A quadratic criterion is<br />

used to detect damage <strong>in</strong>itiation based on stresses. Thereafter soften<strong>in</strong>g is computed<br />

with a stra<strong>in</strong> energy release rate <strong>for</strong>mulation. To alleviate mesh-dependency this <strong>for</strong>mulation<br />

is comb<strong>in</strong>ed with the voxel-mesh<strong>in</strong>g approach.<br />

Epoxy res<strong>in</strong> is modeled with the first, isotropic elastoplastic material model<br />

regard<strong>in</strong>g a pressure dependency <strong>in</strong> the yield locus. As the assumption of constant<br />

volume under plastic flow does not hold <strong>for</strong> epoxy res<strong>in</strong>, a special plastic potential<br />

is chosen to account <strong>for</strong> volumetric plastic stra<strong>in</strong><strong>in</strong>g.<br />

To describe the material behavior of the fiber bundles, the second, transversely<br />

isotropic, elastoplastic material model is developed. The constitutive equations<br />

<strong>for</strong> the description of anisotropy are <strong>for</strong>mulated <strong>in</strong> the <strong>for</strong>mat of isotropic tensor<br />

R. Rolfes, G. Ernst, M. Vogler, and C. Hühne<br />

Institute <strong>for</strong> Structural Analysis, Leibniz University of Hannover, Appelstraße 9a, 30167 Hannover,<br />

Germany, e-mail: r.rolfes@isd.uni-hannover.de<br />

27

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

Saved successfully!

Ooh no, something went wrong!