A data-driven subgrid scale model in Large Eddy Simulation of turbulent premixed combustion

摘要

We derive a data-driven model of a subgrid scale (SGS) closure for turbulent premixed combustion inthe context of Large Eddy Simulation (LES) using deep learning. We validate the model through LES ofthe direct numerical simulation (DNS) flame configuration and compare it to other subgrid models fromthe literature. The filtered DNS used as training data was provided by Lapeyre et al. (2019). The neuralnetwork developed in this study was designed to estimate the SGS flame surface density, using only localprogress variable values as a basis. A priori tests show that the results inferred from the frozen neuralnetwork were comparable to results obtained from the convolutional neural networks (CNNs) usingthe full nonlocal set of variables, and were in good agreement with the filtered DNS. A model-agnosticmethod for interpreting machine learning was employed to investigate the behavior of the trained neuralnetwork. A posteriori evaluation using the network as an LES subgrid model demonstrates that the proposeddata-driven modeling is more accurate than classical algebraic models in terms of the integratedflame area in the axial direction. This illustrates that the proposed data-driven subgrid model to representthe non-linear unresolved terms is a successful approximation from both an a priori and an a posterioriperspective and that only fully local values in the filtered domain suffice to yield good agreement withDNS results. This is in contrast to earlier attempts, which use the full LES domain dataset as input to thenetwork.