Regularized virtual fields method for mechanical properties identification of composite materials

This paper presents a new material parameters identification method based on the Virtual Fields Method (VFM) where a regularization term coming from a micromechanical model has been added. In this Regularized Virtual Fields Method (RVFM), the system of equations associated to the VFM is solved within a constrained optimization framework. The method was applied to virtual Scanning Electron Microscope in-situ tests aiming at determining the local properties of uni-directional composites loaded in the transverse direction and for which the strain fields were obtained through Digital Image Correlation (DIC). The method was tested for composites with various volume fractions and properties contrasts and under different boundary conditions. The potentials and limitations of the algorithm in the presence of noisy images were investigated and compared to that of the VFM and the other finite element based methods. The corresponding results indicate that the proposed RVFM is more efficient than the other identification techniques in terms of both accuracy and computational time. An optimum size of region of interest was also determined by taking into account DIC magnification requirements and based on the accuracy of the identified parameters. This kind of study can be very useful for determining, a priori, the level of resolution in the imaging system required for a favorable accuracy of the identified parameters.