[ishan_ae]

For the past two months I am trying to use rhoSimpleFoam( OF v6) to get results for my compressible subsonic internal flow. The solver refuses to not crash.

I have tried every trick I know; initialization using in compressible solution/inviscid solution, changed the boundary conditions, first order schemes, low relaxation factors, etc.

I know that the solver is sensitive to the values and settings. But having this much sensitivity makes it almost un-usable for me. Also, my case not complicated geometry wise as well as flow wise. It’s a simple duct with p_tot/t_tot at inlet and mass flow rate at outlet.

To investigate more on this, I have also tried to run a simple flow through a diffuser with a maximum Mach number of 0.15. I took this case from NASA’s validation archive and tried to run rhoSimpleFoam on it: https://www.grc.nasa.gov/www/wind/va.../fraser01.html

The solver crashes after 3 iterations.

Does anyone know why this solver is so sensitive? Has anyone used this solver in their studies ? If someone can provide some general guidelines for using this solver it will be helpful.

This will help other people as well, as going by the posts here I am not the only one encountering such problems.

[dlahaye]

Dear Ishan_ae,

There is truth and comfort in the statement that nothing is easy.

It is with interest that I read your posts in this forum. We have had our share of difficulties in making rhoSimpleFoam converge for our application. I much appreciate your tenacity in making the rhoSimple work.

I have two suggestions:

1/ please share details of the conical diffuser test case: geometry, mesh, boundary conditions, initial conditions etc

2/ let us discuss via conference call. I am available for instance on Friday, November 27th, Monday, November 30 th and Tuesday December 1st. I am in time zone GMT+1 (Amsterdam).

There is a chapter on compressible flow solvers in the book of Moukalled e.a. Not sure whether this helps.

Best wishes, Domenico Lahaye.

[ishan_ae]

Hello Domenico.

Thank you.

Of course we can connect via conference call on Friday Nov 27th Amsterdam time.This would be really helpful for me

We can discuss a mutually suitable time via PM. I will attach the files here as well in a while.

Regards,
Ishan

[dlahaye]

Wonderful. Pls. indicate time slot and make files available.

Thx, Domenico.

[ishan_ae]

Hello


I have attached here the case files as well the reference paper: https://drive.google.com/drive/folde...pW?usp=sharing

It includes the mesh as well as the setup files. The setup and geometry resembles that of "Study#1" mentioned in the archive.

In its original form the case has been setup as 2D axisymmetric, total pressure and total temperature specified at inlet, and mass flux at outlet.

I have used a full 3D model with the same conditions at the inlet, but mass flow rate imposed at outlet. I did this for two reasons:

1. I was not able to find the mass flux value that the authors have used.
2. I don't know how in OpenFOAM we can impose any mass flow quantity when a 2D case is being simulated.
3. A full 3D model simulation will allow me to understand on a very basic level what sort of settings work for rhoSimpleFoam solver since the case is not transonic/supersonic but subsonic. I could then replicate them on my actual model which has a max Mach number of 0.7.
4. In the original numerical setup the authors have applied outlet BC at 0.15 metres downstream of the outlet. I have instead applied the BC at the outlet which corresponds to x=1.10 metres location.

I got the mass flow rate value from the "README" file mentioned on the archive page.

With regards to the time slot I PMed you.

EDIT - I have added to the list some more changes I did not menion.

[dlahaye]

See my reply on PM.

[HPE]

Might using the following fvOptions help to avoid crashes:

- limitTemperature (particularly this one)
- limitVelocity

Note that `rhoSimpleFoam` is not a coupled solver. That explains myriad of reasons why the solver is somewhat numerically fragile. Have a review on such solver paradigms, please.

[ishan_ae]

Hello HPE.

Thanks for the info on the solver not being coupled. I have used both the quantity limiters in my original case.

With regards to the solver being fragile, which applications should be suitable for this solver ?

I think aerospace applications with M>0.6 will cause troubles. I have also observed that when flowRateOutletVelocity is imposed, the solver almost immediately becomes unstable if the relaxation factors are > 0.2.

[HPE]

May be you wanna try:

https://hisa.gitlab.io/

[ishan_ae]

Hi HPE.

Thanks for the info on that solver. Looks interesting. I will try to run it once I sort out my current issue.