Machine Learning based reduced models for the aerothermodynamic and aerodynamic wall quantities in hypersonic rarefied conditions

Recently proposed collision-specific parameters for direct simulation Monte Carlo simulations are tested for binary mixtures of nitrogen, oxygen, and argon. Near ambient conditions, the traditional collision-averaged parameters are highly accurate, whereas the collision-specific parameters are not. The simulated transport using the collision-averaged parameters for mixtures with helium, however, is found to be inaccurate. Hence, we propose a novel method to determine molecular parameters by combining the Chapman–Enskog theory with empirical mixing rules and experimental data. The optimized parameters are highly accurate for the binary mixtures of nitrogen, oxygen, and argon and greatly improve the simulated transport for the helium mixtures.

show abstract

Machine Learning based reduced models for the aerothermodynamic and aerodynamic wall quantities in hypersonic rarefied conditions

Cited by 5 publications

References 24 publications

Sensitivity analysis and uncertainty quantification for a three-dimensional rarefied ionized hypersonic flow

Sensitivity analysis and uncertainty quantification for a three-dimensional rarefied ionized hypersonic flow

Establishment and validation of a relationship model between nozzle experiments and CFD results based on convolutional neural network

Optimized collision-specific parameters for binary mixtures of nitrogen, oxygen, argon, and helium

Contact Info

Product

Resources

About