A hybrid subgrid model for large-eddy simulation of compressible wall-bounded flows

点击翻译按钮获取中文摘要

0933377 MaheshThis award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5). The PI plans to develop a novel hybrid wall-modeling strategy for large-eddy simulation (LES) of compressible wall-bounded flows, After formally developing the model, hybrid LES of attached compressible boundary layers will be performed. The model will be evaluated against both direct numerical simulation (DNS) and experiment. Unlike conventional hybrid approaches, the PI does not blend near-wall RANS (Reynolds-averaged Navier-Stokes) regions with LES regions away from the wall, Instead, he solves the LES equations down to the wall. He defines a new error that is a weighted sum of the Germano identity error (LES error) and an error that relates the time-average of the LES subgrid stresses to the Reynolds stresses (RANS error). This composite error is minimized to obtain the model coefficient. The weight in the composite error is defined such that it skews the error toward the RANS component in regions where the LES error is high (near walls). As a result, the model does not explicitly depend on distance from the wall. LES is rapidly progressing towards becoming the method of choice for a wide range of engineering flows. This work addresses a key limitation of LES that is hindering its acceptance -- inaccurate predictions of attached wall-bounded flows. It recognizes important limitations in the prevalent approach towards this problem, and plans to construct a hybrid model applicable to subgrid models beyond those considered in this work. This work will be performed by one PhD student and one undergraduate student. The PI will participate in outreach programs involving high school students from the twin cities area, and summer interns drawn from across the country. This work will form part of the PI's teaching material, and presentations and articles intended for a broad audience.