[TommyM]

Hi All,


I am trying to simulate supersonic flow in a nozzle (rhoCentralFoam) and I don't know why residuals are always zero. Anyway, simulation proceeds and looking at the flowfield by Paraview the results seem to be good.
I really don't know how it is possible. Does anyone know what is the problem?
Files are attached.



Thanks in advance.

[TommyM]

I noticed that this happens for inviscid simulation but not for viscous ones.
I know that my question could be trivial but I have very little knowledge in density-based solvers.

[TommyM]

I have found the solution to my problem. It is a trivial problem but I post here the solution so that it could be useful for someone new to OpenFOAM, just like me.
RhoCentralFoam solves explicitly the inviscid part of governing equations, and then correct them implicitly when it comes to viscosity. Thus, if you are solving for an inviscid flow, then you will not have residuals since the calculation is all explicit.
You can also try to change this in fvSolution, but rhoCentralFoam will continue to use diagonal solver (explicit) for the (inviscid) variables.
Hope that all is correct.

[as020002]

Hi Tommy,

Thanks for the info, I recently implemented a reacting rhoCentralFoam solver to address problems with reacting flow. I also found out that the solver itself is not quite stable (not easy to converge) when solving cases from initial time, especially in supersonic flow. I always have to use pressure-based(e.g. sonicFoam, reactingFoam, rhoReactingFoam) to obtain a decent initial solution then go to the density-based.

In addition, when using different -based solver, they manifest quite different result(e.g. max temperature, flow pattern).

Does anyone have an idea to give me a inspiration why it happens like that?

Thanks,
Rick

[TommyM]

Hi Rick,

I don't know the physics of your case but, as far as I know, rhoCentralFoam is the right solver if you have a supersonic flow, especially when strong shockwaves are present.

I tried to use rhoPimpleFoam and reactingFoam (in their 'transonic' forms) to solve internal supersonic flows and they worked, but this did not happen for external flows with complex waves interaction (e.g. Mach diamond structure) or for internal flows with strong changes in geometry (e.g. nozzles with sharp edges), where the pressure-based solvers gave me unphysical results.

About rhoCentralFoam, it is highly unstable and the only way to overcome this issue (to my knowledge) is to reduce the Courant number.

In my opinion, first you should find which of these two solvers produces the correct solution and:
- if it is rhoCentralFoam, then pressure-based solvers could be not appropriate for your case.
- if it is rhoPimpleFoam, then your implementation of reacting rhoCentralFoam or the settings of your simulations could have some errors.
I hope this comment could help you, but I am not an expert so I might be wrong in something.

Tommy

[m_zanichelli]

Originally Posted by TommyM

I have found the solution to my problem. It is a trivial problem but I post here the solution so that it could be useful for someone new to OpenFOAM, just like me.
RhoCentralFoam solves explicitly the inviscid part of governing equations, and then correct them implicitly when it comes to viscosity. Thus, if you are solving for an inviscid flow, then you will not have residuals since the calculation is all explicit.
You can also try to change this in fvSolution, but rhoCentralFoam will continue to use diagonal solver (explicit) for the (inviscid) variables.
Hope that all is correct.

Hello, I am having the same problem as I am simulating the inviscid flow around an airfoil in transonic conditions. From your answer I couldn't understand whether there is a way to "make the residuals appear" in this case by changing the solver.

[Tobermory]

Tommy explains it in this sentence:

Thus, if you are solving for an inviscid flow, then you will not have residuals since the calculation is all explicit.

What does he mean by explicit? He means, that in this case you can solve the equation explicitly, without iteration (like solving the equation 3x = 2 for x). That is why the solver reports zero iterations and zero values for the residuals. You'll see that in other solvers too, eg for the density/continuity equation.

So - to your question - can you "force the residuals to appear"? No, not in that solver, since there are no residuals to show.