This paper presents an extension of existing Proper Orthogonal Decomposition (POD) based methods for gappy data reconstruction to allow its use with experimental data without a priori knowledge of the ‘True’ solution, and applies it time-resolved DPIV data taken in a transonic turbine cascade. The method is based on minimizing the residual of a divergence criterion, in order to determine the optimum number of modes required for reconstruction. This serves as a convergence criterion for termination of Venturi and Karniadakis’s iterative Gappy POD method. Gappy flow fields were created using DNS data from a near wall turbulence simulation. Gappyness levels of 5%, 11%, 20%, 50% and, 80% were created with gap sizes 3×3, 7×7, 11×11, and arbitrary N×M vector spaces. The method is shown to closely predict the optimum required modes for a minimum-error reconstruction, and the errors associated with this convergence criterion are shown to be on or below the error associated basic DPIV velocity uncertainty measurements. Finally, this method is tested on gappy experimental data from a transonic turbine cascade. The resulting reconstructed flow fields demonstrate clearly observable vortical structures and dominant frequencies.

This content is only available via PDF.
You do not currently have access to this content.