Assessment of chemical scalars for heat release rate measurement in highly turbulent premixed combustion including experimental factors

Abstract

This study uses direct numerical simulation (DNS) data from two turbulent premixed flames to explore the variables involved in experimental measurement of the heat release rate, h˙. Several commonly used markers of h˙ are compared, namely the CH, OH, and HCO radicals, as well as the spatial overlap of OH and CH2O and of H-atom and CH2O. The DNS data were also used to generate synthetic Planar Laser-Induced Fluorescence (PLIF) images that account for confounding experimental factors such as laser sheet thickness, optical blur, and shot noise. The overlap of OH and CH2O and the HCO radical appear to be the most promising markers of h˙, as they show excellent correlation with the high-intensity heat release. The synthetic-PLIF images show that increasing the laser sheet thickness and optical blur progressively de-correlates the measurement and the underlying DNS structures. However, for a sufficiently thin laser sheet (δsheet/δL ≲ 1), the correlation between the synthetic-PLIF and DNS remains strong even with significant optical blur. Similar results are observed in the measured layer thickness and the peak-to-peak distance between the PLIF structure and the heat release layers. The resolution of h˙ features was evaluated by measuring the average turbulent flame length, the spatial power spectral density, and the flame surface curvature. As the laser sheet thickness or optical blur is increased, the measured flame surface area progressively decreases, less fine-scale features are resolved, and regions of high curvature are eroded. Shot noise strongly affects spatial correlation with the DNS data and the layer thickness, but has relatively small impact on the turbulent flame length. Significantly greater deviations are observed for the overlap techniques requiring simultaneous measurement from two cameras, as optical blur and/or shot noise present in the two camera systems will both influence the resulting overlapped PLIF signal.