[Robbebatte]

I am currently attempting to run some CFD simulations for the rear wing of a formula student car, however i am having some issues getting the residuals to converge properly, specifically the Tdr residual.

I am using steady state solver and segregated flow, with K-epsilon turbulence. I am quite certain that my setup should be steady state with no time dependent variables. Also i have 5 prism layers, with a total prism layer thickness of 0.008m.

I have a very fine mesh so i don't believe my issue lies here. Also i am using a velocity inlet with 20 m/s and a pressure outlet at 0 Pa located 25 meters back from the inlet.

I have attached a copy of the residuals i have received. This is the second running of the sim so i know that the Tdr residual doesn't converge more with further iterations.

I would greatly appreciate any help/ideas on how i can improve the residuals.

[arvindpj]

I don’t see any major issues, as these are scaled residuals. For Tdr convergence, here are a few suggestions:

1. Skip the first 50 iterations since turbulence is still developing early on.
2. Adjust normalizing factors in STAR-CCM+; you can either set them manually or let STAR-CCM+ handle them automatically, which may stabilize residuals.
3. Lower under-relaxation factors for momentum and turbulence variables.
4. Refine mesh in high-gradient areas or switch to a coupled solver if needed.

These steps should help improve Tdr residual convergence. Let me know if you need more assistance! More info in this video series.

[SESP]

I agree, the results don't look too bad. I would not rely on those too much but rather on the quantities that you are interested in.
Add some monitor points and observe mass flow rates, pressure drop, max + avg quantities, etc. Use those to judge your convergence. Residuals are nice for quick info, but I would always analyze the relevant quantities.