Wednesday  |  Conference Center B  |  11:30 AM–11:50 AM

#13705, A new Approach for Resolving Large Deformations Around Complex Geometries and Discontinuities: Spatiotemporally Adaptive Quadtree Mesh (STAQ) DIC

Digital image correlation (DIC) is a tool for measuring full-field deformations in experimental mechanics popular for its versatility. However, when analyzing the deformation field of samples containing complex geometries or discontinuities, or near sample boundaries, conventional DIC algorithms may cause a multitude of errors. To reduce errors, local DIC subsets that are near the sample edge or pass through the discontinuity path can be either split or augmented with the Heaviside jump shape function. Yet these adaptations are challenging to generalize and automate, often requiring significant user intervention. Here we present a new user-friendly automatic approach for resolving the deformation fields around complex geometries and displacement discontinuities that overcome the above-described limitations of traditional DIC approaches. The new method is referred to as spatiotemporally adaptive quadtree mesh (STAQ) DIC method. A key feature of this method is that (i) the adaptive quadtree mesh is automatically generated using a mask file of the digital image and (ii) subsets spanning such discontinuities are automatically split. To efficiently solve large image sequences and reduce computational costs, we also propose a data-driven reduced-order modeling (ROM) approach in which unnecessary image frames are skipped, thus achieving temporal adaptability. We show that our STAQ method has high accuracy in solving complex geometric and discontinuous deformation fields in an automated fashion. An important result is that our proposed data-driven ROM method can save 40% ~ 60% computational cost while maintaining the same level of accuracy compared to a fully processed image set. Finally, we maintain an open-source MATLAB implementation of the presented STAQ DIC approach that is freely available to download (https://github.com/FranckLab).

Jin Yang   University of Wisconsin-Madison
Vito Rubino   California Institute of Technology
Zhan Ma   University of Wisconsin-Madison
Jialiang Tao   University of Wisconsin-Madison
Yue Yin   Carnegie Mellon University
Alexander McGhee   University of Wisconsin-Madison
Wenxiao Pan   University of Wisconsin-Madison
Christian Franck   University of Wisconsin-Madison