Stiffness parameters in bending identified using strain imaging and optimization
Stiffness coefficients in granular micromechanics determined with three-point bending experiments and digital image correlation for a bimodulus material
Computational Engineering, Finance, and Science
Summary
Materials can behave differently when stretched versus compressed, making it hard to measure their stiffness accurately. The authors performed bending tests on a special granular material and used pictures to track how it deforms. They combined this with computer models that adjust parameters until the simulations match the real measurements closely. This approach helps find detailed stiffness properties to improve material models for engineering uses.
What this means in practice
- •For mechanical engineers: Determine detailed stiffness properties of composite and granular materials to improve structural simulations and design accuracy.
- •For materials testing laboratories: Enhance experimental protocols by combining three-point bending tests and imaging to reliably identify material stiffness asymmetry.
Authors
Mohammad Mahdi Nayeban, Bilen Emek Abali, Reza Afshar, Francesco Fabbrocino, Anil Misra, Luca Placidi
Abstract
This study presents an inverse analysis to determine the stiffness parameters of a bimodulus constitutive model formulated within a granular micromechanics framework. Full-field normal strain measurements are obtained by Digital Image Correlation (DIC) during three-point bending tests and are used as experimental input to a hybrid numerical optimization procedure. The experimental campaign is designed to ensure that the measured strain fields remain within the elastic regime. Since the identification is formulated as a nonlinear regression problem, a physically informed initial parameter set is constructed to guide the optimization toward mechanically meaningful values. An initial estimate of the compressive stiffness is first obtained from the available manufacturer data. The tension-compression stiffness asymmetry is then estimated from the experimentally measured neutral-axis shift, which provides a macroscopic signature of bimodulus material behavior in bending. These estimates define the initial parameter set, which is subsequently refined through a hybrid optimization procedure based on the full-field comparison of the experimental and numerical normal strain distributions. The identified parameter set enables the numerical model to reproduce more closely both the through-depth profile and the spatial distribution of the normal strain component over the selected surface region. These results demonstrate that full-field normal strain measurements provide an effective basis for identifying the stiffness parameters of bimodulus granular micromechanics models. The largest differences between the numerical and experimental strain fields are observed near the load-application region, where localized strain variations are induced by the concentrated load.