In order to accurately simulate laminar cases where diffusion plays a significant role, the solver should account for backward diffusion of species at the inlet, which has a major impact on the inlet mass fractions. Other CFD solvers capable of simulating laminar reacting flows provide an option to consider diffusion at the inlet.
It would be beneficial to implement a new boundary condition that includes diffusion flux. This addition would complement DeepFlame's existing capability to accurately calculate diffusion coefficients.
Hello,
In order to accurately simulate laminar cases where diffusion plays a significant role, the solver should account for backward diffusion of species at the inlet, which has a major impact on the inlet mass fractions. Other CFD solvers capable of simulating laminar reacting flows provide an option to consider diffusion at the inlet.
It would be beneficial to implement a new boundary condition that includes diffusion flux. This addition would complement DeepFlame's existing capability to accurately calculate diffusion coefficients.
Please refer to the BC used in this paper: https://www.sciencedirect.com/science/article/pii/S0010218016000699
Or, LaminarSmoke repository that has this BC: https://github.com/acuoci/laminarSMOKE/tree/master/libs/boundaryConditionsOpenSMOKE%2B%2B/backwardDiffusion
Or, the definition of the inlet B.C. of Cantera one-D flames: https://cantera.org/science/flames.html
Or, Fluent solver: https://www.afs.enea.it/project/neptunius/docs/fluent/html/ug/node578.htm