Physics informed machine learning for chemistry tabulation

• Published in: Journal of computational science 2023-05, Vol.69 (C), p.102001, Article 102001
• Main Authors: Salunkhe, Amol, Deighan, Dwyer, DesJardin, Paul E., Chandola, Varun
• Format: Article
• Language: English
• Subjects: Chemistry tabulation; Combustion; Deep neural networks; Fluid dynamics; Physics informed machine learning
• ISSN: 1877-7503; 1877-7511
• DOI: 10.1016/j.jocs.2023.102001

Modeling of turbulent combustion system requires modeling the underlying chemistry and the turbulent transport. Solving both systems simultaneously is computationally prohibitive. Instead, given the difference in scales at which the two sub-systems evolve, the two sub-systems are typically (re)solved separately. Popular approaches such as the Flamelet Generated Manifolds (FGM) use a two-step strategy where the governing reaction kinetics are pre-computed and mapped to a low-dimensional manifold, characterized by a few reaction progress variables (model reduction) and the manifold is then “looked-up” during the run-time to estimate the high-dimensional system state by the turbulent transport system. While existing works have focused on these two steps independently, in this work we show that joint learning of the progress variables and the look-up model, can yield more accurate results. We build on the base formulation and implementation (Salunkhe et al., 2022) to include the dynamically generated Thermochemical State Variables (Lower Dimensional Dynamic Source Terms). We discuss the challenges in the implementation of this deep neural network architecture and experimentally demonstrate its superior performance.