Multicore CPU-based parallel computing accelerated digital image correlation for large soil deformations measurement

This paper presents a new approach for measuring large deformations in geotechnical experiments employing digital image correlation (DIC) or particle image velocity (PIV) techniques. The proposed method is based on the Eulerian analysis scheme, allowing for the application of multicore central processing unit (CPU)-based parallel computing to expedite the processing of experimental images. The displacement increments obtained through DIC analysis on the Eulerian mesh nodes (subset centers) are then mapped onto tracer particles (TPs), which are assigned by users to track material movement. Finally, accumulated displacements and strains are determined on these TPs. Two example applications are presented to showcase the capabilities of the proposed method: a centrifuge half model test of flat circular footing penetrating sand overlying clay and a full transparent soil model test (TMST) of conical pile penetration. A comparison with other standard first-order deformation algorithms and the Lagrangian analysis scheme demonstrates that the presented method offers comparable precision but significantly faster computation speed, with an improvement of over six times when processing a considerable number of (e.g. over 20) images. This enhanced computational speed can greatly reduce the time required for image post-processing. The proposed method is particularly suitable for large deformation experiments that involve the analysis of numerous images and require high precision.