Evaluation techniques for optical analysis of hybrid rocket propulsion

This article summarizes recent results of data evaluation techniques obtained with optical investigations on the combustion behavior of hybrid rocket fuels. Tests are performed in a 2D slab burner configuration with windows on two sides. Liquefying paraffin-based fuels are tested in combination with gaseous oxygen (GOX). High speed videos of combustion tests are recorded in order to investigate the combustion phenomena of this kind of fuels. Hybrid combustion of liquefying fuels is dominated by transient flow dynamics like Kelvin-Helmholtz instability and vortex shedding, also due to the characteristic turbulent diffusion flame. In order to better evaluate these flow phenomena, characteristic frequencies and wavelengths of the main structures of the flow field and of the combustion flame appearing in the video data have to be found. In this work, a spatial and temporal analysis of these structures is carried out by using two different techniques, applied within an automated video evaluation routine. First of all, the Proper Orthogonal Decomposition ( POD) technique is used. Its results deliver linearly uncorrelated variables, which are the principal components of the flow field. This method enables to recognize the main structures of the flow field and the combustion flame appearing in the video data. Secondly, the Independent Component Analysis (ICA) technique is applied to the same data. It is able to search for statistically independent, or as independent as possible, structures hidden in the data. It increases the independence to higher statistical orders with respect to POD. The basis functions found with the ICA are expected to describe the essential structure of the data and to resemble some physical processes involved in the combustion. With both methods it is possible to compute spatial and temporal coefficients, which can be later analyzed by applying a Power Spectral Density (PSD) in order to obtain the excited frequencies and wavelengths during the combustion. Finally, the results of the two methods are compared in order to better understand and interpret them. The results collected so far and the comparison of both techniques show that their application is consistent and useful for the automated evaluation of combustion data.