Experimentally induced errors in Digital Image Correlation measurement of small strains with large gradients

Abstract

Digital Image Correlation or DIC is an already widespread and commonly used technique to perform full field strain measurements. Usually, a so called 'speckle pattern' is put on the specimen and then, pictures are taken first in the undeformed stage, which is the reference and then in various deformed states. The technique itself consists of comparing the image of a deformed pattern with a reference image and determining the displacements of the so called 'subsets'. The mathematical algorithm eliminates rigid rotations of the subset and is able to determine the strain field. This technique was originally designed for large-strain measurements and as such, it works very well when large strains are present, but when determining (very) small strain fields, especially in combination with large (rigid body) deforma- tions or large strain gradients, this technique becomes a lot more sensitive to the boundary conditions of the experimen- tal setup. This manuscript will illustrate the errors induced by experimental factors such as in-plane rotation, in-plane rigid body translation, out-of-plane rigid body rotation, which are in fact all related to how the specimen is gripped in the ten- sile machine, on the eventual derived strain field. Furthermore, processing parameters such as subset and step size and the used strain window on numerically and experimentally induced transformations will be assessed.