Wednesday  |  Salon 7  |  05:40 PM–06:00 PM

#15854, Temporally Adaptive Digital Image Correlation for Resolving Large Image Sequences of Non-uniform Temporal Events

In most Digital image correlation (DIC) tests, it is common to take a sequence of digital images of a specimen during deformation and analyze each frame's displacement and strain field. With the rapid development of high- and ultra-high-speed imaging technology, DIC can measure quickly evolving phenomena where the sampling frequency can reach 1 kHz to 10 MHz using commercial high-speed cameras, and the recording length for a single test can easily exceed one hundred frames and might span 1000 to 1 million frames. While all these advancements enable new capabilities of DIC to investigate unknown material's dynamic behavior, they also result in a vast number of images being post-processed, which can be computationally expensive. Recently, we leveraged the emerging data-driven reduced order modeling (DD-ROM) method and developed a new algorithm, spatiotemporally adaptive quadtree mesh (STAQ-) DIC, to achieve temporal adaptivity for analyzing DIC image sequences. However, the STAQ-DIC algorithm hasn't been verified for resolving non-uniform temporal evolutions, such as inertial cavitation or high-rate crack events where bubble collapse and crack nucleation happen within extremely short time windows but subsequent bubble oscillations or crack propagation feature relatively longer characteristic time scales. In addition, DIC measurement difficulties are further exacerbated by the large deformations experienced during these non-uniform temporal evolutions. One challenge in tracking these motions is that conventional DIC cumulative tracking (all the subsequent frames are always compared with the first reference frame) might fail. Instead, the incremental tracking DIC (every two consecutive frames are compared to solve incremental deformation fields first and then interpolate and add together to obtain the cumulative displacement field) is able to track extremely large deformations but accumulates tracking and interpolation errors. To further improve DIC performance, we have devised a novel strategy to temporally adaptively determine either cumulative or incremental tracking modes to analyze DIC image frames. The accuracy and efficiency of the developed novel temporally adaptive DIC method for resolving large image sequences of non-uniform temporal events are further assessed using both synthetic and experimental datasets.

Kaixin Zhan   The University of Texas at Austin
Vito Rubino   École Centrale de Nantes
Alexander McGhee   University of Wisconsin-Madison
Christian Franck   University of Wisconsin-Madison
Jin Yang   The University of Texas at Austin