An Inverse Approach for Reconstruction of Local Scalar Fields in Turbulent Flows from Line-of-Sight Measurements

Abstract

Turbulent flows pose some significant challenges for line-of-sight (LOS) measurement techniques, like two-color radiation thermography or line-of-sight absorption spectroscopy. The desired local scalar properties often have a non-linear dependence on the path-averaged measured signal and sophisticated reconstruction techniques are required. In the case of turbulent flows, a time-averaged axisymmetric flame shape can often be assumed. However, due to the non-linear dependency, deconvolution with state-of-the-art techniques leads to a significant overestimation of the desired scalar. Motivated by this, we describe an inverse method based on a Monte Carlo simulation for the tomographic reconstruction of statistical properties in turbulent flows. The method optimizes a statistical model that mimics the stochastic nature of the turbulent flow field by estimating the local scalar distributions and an integral length scale, which is used for spatial correlations of the scalar properties and to generate intensity distributions in LOS direction. The performance of the method is demonstrated on a numerical two-color radiation thermography test case, which is based on a large-eddy simulation (LES) of a hydrogen burner. The mean temperature is reconstructed with reasonable accuracy, with a maximum deviation of 10% observed in a small annular volume. This represents an improvement in accuracy in comparison to current deconvolution techniques. Furthermore, the method provides an estimate of the standard deviation of temperature and an integral length scale, which are valuable additional information. Possibilities to improve the accuracy of this method are presented and its limitations.