[galap]

Dear all,

I am struggling now for weeks to stabilize my simulation and I would like to ask you about your procedure to damp pressure waves: My goal is to simulate a disperse transonic choked flow. Meanwhile I was able to simulate the choked gaseous flow (using LES with rhoPimpleFoam). The simulation is very difficult because the simulation is very unstable. Though I was able to obtain a stable simulation finally with a totalPressure/waveTransmissive BC combination initialized with a totalPressure ramp.

Now proceeding to the disperse flow I switched to the compressibleInterFoam but I had to realize that my grid is too coarse to properly resolve the phase interface with VOF. In order to use a finer grid I had to shorten the CFD domain (around 100 mio cells). With that kind of fine mesh it seems that I am resolving now all kind of pressure waves and I am not able anymore to obtain at least a stable simulation of the gaseous choked flow.

My setup:

waveTransmissive inlet pressure BC
pressureDirectedInletVelocity inlet U BC
almost only upwind for div schemes, cellMDLimited for gradSchemes
Smagorinsky sgrid model
adaptive time step (max Co fixed to 0.4)

On the attached image you see the pressure profile in the vicinity of the inlet. The pressure should be there around 5e7 Pa.

What is your procedure? Use a very large inlet entry path? Do you coarse the mesh at the inlet? Use a baffle to avoid that pressure waves reach the inlet?

[lukasf]

Hi,


were you able to fix your problems? I would be curious to hear what you did.


Thank you

[kin3062]

As far a I know in high dynamic, explicit, solid mechanical simulations, there may be high frequency numerical oscillations. Solvers can include an artificial numerical bulk viscosity to damp out these oscillations.


Don't know if similar tricks can also be done on CFD? And how?

[galap]

Not really. At this time, actually, I had no experiences with compressible choked flows. The "standard" pressure based solvers in OpenFOAM with their equations solved in a sequential manner are not good for those kinds of flow.

For me it helped a bit to initialize the case on a very coarse mesh and then I interpolated the soluation on a much finer grid.

In the end, for steady state flows I switched to CFX. Because the equations are solved in a coupled way there, it is much more stable.

Eventually, for transient compressible flows with shocks (as the case I encountered at the time I started this thread), I was using (and still using) an explicit density based solver which is absolutly stable for this kind of problems.

[kin3062]

It seems to me that OpenFOAM's compressible solvers are not very good anyway ...

I have tried running rhoPimpleFoam but not successful, solver just keep on blowing up or just turn to be very unstable. Neither do the buoyantPimpleFoam ....

But normal PimpleFoam and buoyantBoussinesqPimpleFoam works properly ... It seems to me that, once your calculations involving thermophysical Properties, calculations doesn't go well.

Maybe OpenFOAM should go and look more carefully about their compressible solvers and thermophysical properties I think.

Originally Posted by galap

Not really. At this time, actually, I had no

experiences with compressible choked flows. The "standard" pressure based solvers in OpenFOAM with their equations solved in a sequential manner are not good for those kinds of flow.

For me it helped a bit to initialize the case on a very coarse mesh and then I interpolated the soluation on a much finer grid.

In the end, for steady state flows I switched to CFX. Because the equations are solved in a coupled way there, it is much more stable.

Eventually, for transient compressible flows with shocks (as the case I encountered at the time I started this thread), I was using (and still using) an explicit density based solver which is absolutly stable for this kind of problems.