Computational modelling and experimental characterisation of heterogeneous materials

Beveridge, A. J. and Wheel, M. A. and Nash, D. H. (2009) Computational modelling and experimental characterisation of heterogeneous materials. In: 3rd International Conference on Advanced Computational Engineering and experimenting, ACE-X 2009, 2009-06-22 - 2009-06-23.

[thumbnail of strathprints016330]
Preview
Text. Filename: strathprints016330.pdf
Accepted Author Manuscript

Download (453kB)| Preview

Abstract

Heterogeneous materials can exhibit behaviour under load that cannot be described by classical continuum elasticity. Beams in bending can show a relative stiffening as the beam depth tends to zero, a size effect. Size effects are recognised in higher order continuum elastic theories such as micropolar elasticity. The drawback of higher order theories is the requirement of addition constitutive relations and associated properties that are often difficult to establish experimentally. Furthermore the finite element method, of great benefit in classical elasticity, has shown limitations when applied to micropolar elasticity. The determination of additional constitutive properties and the computational modelling of micropolar elasticity will be discussed in the context of a model heterogeneous material loaded in simple 3 point bending. The model material was created by drilling holes in aluminium bar in a regular pattern, with the hole axis normal to the plane of bending. The bending tests show that a size effect is present. These results are compared against modelling the detailed beam geometries in the finite element package ANSYS, which again shows the size effect. These two bending test are used to extract the additional micropolar elastic material properties. A comparison is then made against analytical solutions,numerical solutions using a micropolar beam finite element and a micropolar plane stress control volume method.It will be shown that the need for extensive experimental testing to determine the additional constitutive properties may not be necessary with the appropriate use of numerical methods.