Investigations on the accurate prediction of supersonic shear layers for detached eddy simulation

One of the current problems of the detached eddy simulation (DES) and delayed DES (DDES) methods is the inaccuracy in prediction of the separated shear layer flows, which mainly reflects in the delayed development of the shear layer. To correctly simulate the separated shear flow, various modified definitions of the DDES length scale are investigated and the improved definition of the subgrid eddy viscosity is introduced to deal with the delayed development problem for the first time. In this paper, the shear layer without the upstream boundary layer is studied in order to assess the subgrid model of the DDES in the simulation of shear layers. Large eddy simulation (LES) and experiment results are introduced to test the performance of different definitions of the DDES parameters. According to the results, both the definitions of the length scale and subgrid eddy viscosity coefficient should be improved to get the best prediction. DDES with the LES length scale and vorticity-correlated length scale as well as the modified definition of the subgrid eddy viscosity coefficient predict a more accurate development of the shear layer than DDES with the original definitions, which provides a possible solution to the delay development problem of the shear layer. Furthermore, the LES mode of DDES performs considerably differently from the Smagorinsky-LES, which shows the former subgrid model is essentially different from that of Smagorinsky-LES. Theoretical analyses are made in order to find out the quantitative difference of the subgrid eddy viscosity between the LES mode of DDES and Smagorinsky-LES.