Intake manifold transient simulation

Hi, I’m working with a CFD transient simulation of an intake manifold, so with the specs of the engine (4 cylinders) I set up a pressure pulse for each cylinder, so the time lapse for the valve open-close cycle for the 4 cylinders is 0.02668 seconds with a delta t of 0.00006667 seconds (the four cylinders) but the end time is 0.10672 seconds (I put 4 for times the cycle of the open-close valve), so as you can see in the steady state in all cylinders are almost the same velocity ( like 58 m/s) but in the transient analysis I runned looks some unstable in the velocity (Uy) in the beginning and I was reading about the convergence with the residuals so in the transient run i guess the residuals must be in the range of (1e-5 and 1e-6) , so my questions are:

  1. Do I have to change the relaxation factor of the velocity and pressure to have a better convergence?
  2. What set up do you recommend me with: numbers of outer correctors, number of correctors, number of non-orthogonal correctors (mesh quality, non, orthogonality between 60-50), solvers and schemes set up?

Residuals:

Pressure pulse:

Uy velocity:

Link of the project: https://www.simscale.com/workbench/?pid=1967226731207844362&mi=spec%3Aee154b95-6f6f-4ce3-a337-2a8cc7a55a5a%2Cservice%3ASIMULATION%2Cstrategy%3A201

1 Like

Hi there, my first thought would be to improve the mesh quality, as your criteria have some high values, have a look here:

Have you considered using the standard mesher instead of the hex dominant?

FYI: Simulation convergence problem - #3 by Get_Barried

Best regards,
Fillia

1 Like

Hi, improving the mesh quality helps me, but my residuals still high.

1 Like

Maybe you could also try ‘Mean value’ as pressure type, and see whether it helps with convergence :slight_smile: