Mat-1.3656 Numeerisen analyysin ja laskennallisen tieteen seminaari.


Ma 8.10. 2007, sali U356

Sven Bossuyt
Mixed numerical-experimental characterization of deformation behaviour
involving discontinuous displacement fields


The inelastic deformation behaviour of many materials is non-uniform on a mesoscopic scale. Models assuming uniform deformations at the macroscopic scale, using homogenized properties, are often good enough for practical applications. But in some cases, the uniform deformation is unstable with respect to non-uniform perturbations, and highly localized deformation is observed on the macroscopic scale. Three examples of such behaviour will be presented: shear banding in bulk metallic glasses, multiple cracking in textile-reinforced cement-matrix composites, and delamination in polymer-matrix composites. For each of these materials, optimal use of their capabilities in engineering applications requires predictions of the non-uniform deformation behaviour, so this is a problem of practical interest.

A solution may be found in recent advances in optical measurement techniques for experimental mechanics, enabled by continuing evolution of computers and digital cameras. "Full-field" optical measurement techniques are of particular interest: optical imaging is then used to do measurements simultaneously at each point in the image. As a result, localized deformation can be observed even if it is not a priori known where in the image it will occur. In principle, at least, full-field measurements could be combined with a numerical model of non-uniform deformation behaviour, to determine appropriate material parameters via an inverse method.

This seminar will focus on ongoing research, using displacement fields measured by "digital image correlation", to quantitatively characterise localised deformation behaviour of materials. In digital image correlation, displacement fields are obtained from the coordinates of corresponding subsets in sequences of images. Digital image processing algorithms are used to match subsets of the image of the deformed material, to corresponding subsets in an image of the material before deformation. Shortcomings of existing implementations of this technique will be highlighted, especially with regard to highly localised deformations, and possible improvements building on ideas from other fields will be discussed.