SimScale CAE Forum

Courant Number, Mesh, relaxation factors and schemes issues

LINK OF THE PROJECT: https://www.simscale.com/workbench/?pid=622516419259523012&mi=run%3A395%2Csimulation%3A331&mt=SIMULATION_RUN

Actual simulation: Compressible 9

SUMMARY: I’m simulating a transient analysis about an Intake manifold for an internal combustion engine, It has a turbocharger so the mass flow parameter it’s high, I established the pressure pulse in excel with the RPM but my issues are:

  1. The numerics, I’m trying to make a cost effective analysis, but I tried to set up the relaxation factors for fast convergence, but I have issues with the solvers and the schemes I don’t know which ones I have to use for my analysis.

  2. The simulation control, I tried with adjustable time step: false, but I identified high courant number, in transient test 14. Also, I tried an adjustable time step: true with a courant number of 0.7 but It’s too slow.

  3. Using a her-parametric mesh with a level of refinement of 1, I tried a simulation with a level of refinement of 2 but the simulation crashed.

Best regards, I will appreciate your help.

Hello @agredafuentes1,

Have you by any chance already checked this webinar on setting up cost-effective transient analyses? It goes through numerics, simulation control, mesh settings and some other useful hints.

Cheers

Hello Ricardo @RicardoParis Yes I checked the webinar yesterday and I identified that I didn’t write the courant number in the proper way, so I introduced a CFL max of 1000 as the webinar but overcame the limit CFL >1000, and diverged.

So, I tried to reduce the time step but it didn’t work, I used a refinement in the mesh of level 2.
Transient test 33, timestep 1e-5 s, diverged at Courant max 2,571.
Transient test 34, timestep 1e-8 s, diverged at Courant max 6,142
Transient test 35, timestep 1e-12 s, diverged at Courant max 22,004

Also, I identified that the mean Courant number overcame 1 at low Courant max for example 25.
I will do some test with a mesh refinement of level 1 but, with level 1 looks a little bit bad.
This is the mesh, level 2:

Mesh level 1, The issue near the outlet:

No problem. It’s definitely not a good idea to have mean Courant too high.

From inspecting the results of your latest runs, the simulations are diverging around the outlet that has a varying pressure (at the time that the simulation fails):

One question:

Is it possible for you to set the velocity outlet via a table input for mass flow rate?

@RicardoParis Yes, you’re right it’s diverging at the outlets but how do I fix that, Do you think this is the problem? (screenshot)

And answering your question, the Intake Manifold works with pressure pulse based in the fire order so, I did the pressure pulse P vs Time according to the plenum pressure near 145 kPa and the ambient pressure near 1 atm but due to vacuum pressure drops a little I wrote 93 kPa, the pressure pulse it’s a representation of the real pressure behavior, so If i make a table input for mass flow rate to velocities outlets I will avoid the design goal that is the velocity because if I make that i will chain the behavior.
Also I didn’t find the option of potencial foam initialization in simulation control I think It’s no more available to community plan.

I did some test in transient test 36, 37 and 38, with a coarse mesh level 1, but diverged in the outlet, the last one I tried with a courant max of 10k. But It seems that the problem maybe is the mesh :thinking:

1 Like

A max Courant number of 10000 definitely sounds too much. The velocities are already pretty high in this application.

The mean Courant number is already over 1 after a couple of timesteps (with max Co around 30). I’d still try to heavily limit the max Courant (e.g. max of 5) and try different BC configurations to try and achieve a stable initial run.

1 Like