Novel DNNs for Stiff ODEs with Applications to Chemically Reacting Flows
This addresses high CPU demands in engineering applications like combustion, but it is incremental as it compares existing learning approaches for ODEs.
The paper tackles the computational bottleneck of simulating chemically reacting flows with many reactions by introducing novel deep neural networks to approximate stiff ODEs, showing that accounting for physical properties improves generalization.
Chemically reacting flows are common in engineering, such as hypersonic flow, combustion, explosions, manufacturing processes and environmental assessments. For combustion, the number of reactions can be significant (over 100) and due to the very large CPU requirements of chemical reactions (over 99%) a large number of flow and combustion problems are presently beyond the capabilities of even the largest supercomputers. Motivated by this, novel Deep Neural Networks (DNNs) are introduced to approximate stiff ODEs. Two approaches are compared, i.e., either learn the solution or the derivative of the solution to these ODEs. These DNNs are applied to multiple species and reactions common in chemically reacting flows. Experimental results show that it is helpful to account for the physical properties of species while designing DNNs. The proposed approach is shown to generalize well.