Extension of POD for Feature Extraction from Time Resolved Reacting Flow Data Sets

Recent improvements in computational and experimental combustion have led to an increased availability of high fidelity data sets. Hence, the reduction and analysis of this data become a key aspect in improving our understanding of highly dynamic, reacting flow fields. This work utilizes simultaneous high speed measurements of fuel distribution, flame location and flow velocity in a high pressure, liquid fueled swirl combustor. The objective is to extract information from these experimental data sets, which can be directly compared with large scale computations. Due to the highly dynamic environment encountered in gas turbine combustors, an instantaneous snap-shot of the flow field rarely resembles the time-average. It is therefore essential to develop methods which allow us to analyze and compare unsteady flow features. While a number of both data driven and physics-based approaches have been previously presented in the literature, the purpose of this work is to extend one of the commonly used modal analysis techniques, POD, utilizing existing flow physics to condition the algorithmic detection of dominant flow structures.