CFD Online Discussion Forums - Reactive flow : characteristic times

Dear CFD experts,

I am puzzled about different characteristic times that intervenes in a reactive case simulation. I am using OpenFoam and my case is a high speed (compressible) reactive flow. A modified version of rhoCentralFoam, whish is reactive, is used. What I do first is to run the simulations without chemistry and once I have the convergence for the flow, I add the chemistry. Lets say the residence time is bigger than the chemical characteristic time. The following cases are set :

1. initialChemicalTimeStep 1e-7s + deltaT in controlDict 1e-5 s

2. initialChemicalTimeStep 1e-7s + deltaT in controlDict 1e-5 s

3. initialChemicalTimeStep 1e-7s + deltaT in controlDict 1e-8 s

The fist setup was very very slow. Each iteration takes 10 seconds. The second is fast but the mass fraction of species are oscillating around the steady state solution (please see the figure MassFraction_secondCase in the attachment). The third case is fast and steady state case for species happens rapidly (please see the figure MassFraction_thirdCase in the attachment).

I am really puzzled about the influence of parameters in controlDict on the chemistry. I thought that the controlDict is for the flow part and the chemProperties in constant file for chemistry. However, by changing the parameters in flow field, I see huge impact on the chemistry part. I though decreasing the courant number and the deltaT it would be destructive for convergence. However I see the opposite. And another issue that I should mention is the fact the residual of the mass fraction in log file, does not show anything about the convergence that I see in the paraview. The residual in first and second case remain very high. In the third case, for some species the residuals are very high but for few of them very low. So, let's say the third case, for the moment, is the fastest and most accurate in terms of results. However, by looking at residual, I don't have the impression that steady state is reached. But by looking at paraview, I see that nothing changes after a while.

So, my questions are :

For solving mass fraction, we have an ODE solver + transport equation. The ODE solver solves a matrix for whole species. And after the mass fractions are transported. No diffusion and no viscosity in may case. So the transport of species is due to the convection. Why the third configuration is the best case for me ? Why is good to have such a low courant number and such a low deltaT ? Why the third case is the fastest case ? It should not be the slowest ? Why residuals remain so high but the steady state reaches ?

Thanks for helping me to solve these fundamental questions.

Mary