Data-driven stability analysis in a multi-element supercritical Liquid Oxygen-methane combustor

Abstract: Thermoacoustic instability (TAI) is a pressing problem in rocket combustors. TAI can cause significant damage to a combustor, resulting in mission failure. Therefore, stability analysis is crucial during the design and development phases of a rocket combustor. Stability analysis during the design phase can be substantially aided by the rocket combustor's large eddy simulation (LES). However, the computational cost of LES for full-scale rocket combustors is high. Therefore, using a small set of data from a large eddy simulation of a multi-element full-scale combustor, we investigated the effectiveness and computational needs of many data-driven and physics-driven tools for the classification of the stable and unstable regimes in the current study. Recurrence network analysis (RNA), reservoir computing (RC), and multi-scale permutation entropy (MPEA) analysis are the instruments employed in this study. The regime categorization task is unsuitable for RNA and MPEA, according to the results. With little input data, RC-based metrics may map the stable and unstable regimes and are thought to be computationally inexpensive and straightforward to use. In order to help with the design and development of rocket combustors, the combined LES-RC method to stability analysis is therefore anticipated to result in a notable decrease in processing needs.