I have a spray model that I validated on STAR CD. This code is getting pretty old and in my lab, we want to go to STAR CCM+.
I imported my mesh, put in place the parameters as in STAR CD to make validations between the two codes. However, the simulations always diverge. I checked my model item per item 6 or 7 times, tried several things (reduce time step by 5, increase inner iterations, decrease relaxation factors). I got some improvements by changing the initial conditions on the turbulence but it ended diverging as well.
Is it normal to have such differences between two codes? Difference on conditions needed for the model to converge, like time step etc? Is it possible that the mesh of the first code is not adapted at all for the second one?
To be clear, I understand that the differences between the two codes will bring differents sensitivity to mesh and other parameters but I would expect them to be able to run the same model, was I wrong?
Thank you for your help.
If you have the STAR-CD case and the STAR-CCM+ case, check the set-up is equivalent (boundary conditions, properties, mesh, timestep/URF, etc..) and if so I would send it to the local support. The new code should work as well or better than the old in principle. So you have ground for requesting a visit to the garage.
Yes, as I said, I checked X times... As I'm running out of options, I will send a request to the local support.
Thanks
I`m so sorry for my poor english but I could give you a little help.
First of all, You have to check that spray model setup is correct with frozen another solve(just active time solver and lagrangian solver) and if you not concerd vaporized, it is not important mesh defendency than time step. And as for me, I will be regenerate the mesh so, imported STAR-CD mesh have some different qulity.
Hi,
Since last time, I got some new and forgot to update my post...
So today, I am able to run a simulation with a quality equal to my STAR-CD simulation but I still have to check results in details (I got to stop this activity during 6 months...).
So basically, I left the mesh from STAR-CD in a corner and started with a new mesh made in STAR-CCM+ which allowed me to have a simulation running but still with "bad" convergence.
I said "bad" because in reality, it is not. Indeed, in STAR-CCM+, we all got convergence problems in my lab but for different types of application. And finally, someone became aware that STAR-CCM+ was not using the classical idea of residuals we usually use. By default, it uses normalized residuals which makes impossible to reach residuals of 1e-4 or 1e-6. So turning off this normalization, it seems it solves the problem for now. When I finish my current test, I will try to use the same simulation with the STAR-CD mesh and see what happen...
I will keep this thread updated!
- Badges
- Science topic