Assessment of LES and RANS-LES Hybrid Models for Heat Transfer Predictions in Effusion Cooled Combustor Liners

Abstract

Abstract Accurate numerical predictions of surface heat flux on combustor liners in the presence of effusion cooling involve appropriate resolution of turbulent boundary layers and mixing of two different streams. Precise surface heat flux and wall temperature predictions are necessary for the optimal design of combustor liners to avoid burnout and damage to the combustor. Reynolds Average Navier Stokes (RANS) model has shown superior wall heat transfer predictions for steady flows; however, in combustor liners involving complex effusion jet mixing patterns, it fails. On the other hand, Large Eddy Simulation (LES) can capture to a good extent core flow mixing in such situations, but it requires very high-resolution near-wall meshes for accurate surface heat flux predictions. To overcome these issues, a hybrid model using RANS in the near-wall region and LES in the core region have been proposed for better wall heat transfer predictions. In this study, a numerical analysis is carried out to test the capability of RANS, LES and hybrid models (SBES, WMLES) for wall heat transfer predictions. The computational setup is a flat plate where freestream high-speed flow approaches a thirty-five degree inclined jet. The study is divided into two regions of interest, one before the jet freestream interaction and another post-interaction. We demonstrate with the SBES approach, surface heat flux can be predicted to much better agreement with the test data in both the regions of interest. Also, it is shown that such results can be obtained with much coarser mesh resolution, hence less computational cost, with hybrid models than pure LES.