Application of the RANS-DNS Framework to Post-processing the Mixing Layer DNS Data
Mohamed Abuhegazy, Omar Ahmed Mahfoze, Scott M. Murman, Svetlana Poroseva · AIAA AVIATION 2022 Forum · 2022
View Video Presentation: https://doi.org/10.2514/6.2022-3769.vid Typically, budget terms in the transport equations for statistical velocity moments (RANS equations) are extracted from DNS data at the time of simulations (on-the-fly). However, a larger set of statistics may later be required for modeling purposes. Some statistics may have to be recalculated for various reasons as well. Yet, resources may no longer be available for repeating DNS. Developing a procedure for extracting high-quality data of the budget terms in the Reynolds stress transport equations from previously generated instantaneous flow fields produced with the spectral-element code NEK5000 is one of the objectives of the current study. A set of 960 instantaneous flow fields in a turbulent mixing layer are used for this purpose. In DNS, the mixing layer spatially develops between two co-flowing laminar boundary layers formed on two sides of the sharp-ended splitter plate of a finite thickness. The flow conditions were specified to closely match those for the untripped boundary layers before their mixing in the wind tunnel experiments of Bell and Mehta (1990) with Re_δ=3930 and 2412 on the high- and low-speed sides of the splitter plate, respectively. In the current paper, the post-processing procedure is tested at a location downstream the splitter plate, where the flow is found to be turbulent exhibiting self-similar behavior for the mean velocity components and the Reynolds stresses. In the procedure, the mean velocity components, Reynolds stresses, and the budget terms in their transport equations are calculated using the central finite-difference scheme of a 2nd order accuracy on a uniform grid. The quality of the extracted statistics has been analyzed by comparing with the available statistics collected when conducting original DNS and also by using the RANS-DNS framework proposed in Poroseva et al. (2016). The RANS-DNS simulations are conducted using the open-source CFD solver OpenFOAM.