Our new paper, “Model-based time super-sampling of turbulent flow field sequences,” introduces a powerful method to overcome the technological limitations of current Time-Resolved (TR) PIV systems. Our approach allows researchers to recover high temporal resolution from standard, non-time-resolved (NTR) measurements.
We leveraged POD-Galerkin models by applying Proper Orthogonal Decomposition (POD) to PIV snapshots. We then projected the Navier-Stokes equations onto this reduced space. By performing time integration of the resulting low-order dynamical system, we achieve a temporally continuous reconstruction of the flow evolution between the discrete PIV snapshots. This highly accurate process effectively fills in the temporal gaps.
Our model-based, data-driven method trained on experimental data offers physics-informed super-sampling. It is practical to use as it is directly applicable to planar PIV setups, while at the same time it is capable of reconstructing dynamics across multiple convective time scales.
Read the publication, co-authored by Qihong Lorena Li Hu, Patricia García Caspueñas, Andrea Ianiro, and Stefano Discetti here. You can also access the data sets on Zenodo and the code on Github.
