Weird results


#1

Hi Jousefm, I got really weird result when simulation finished can you plz have a check?
thank you!


#2

https://www.simscale.com/projects/czhang/ubr21_yaw/


#3

Hi @czhang,

I’ve identified a few things that may help you in achieving more accurate results, I’ll go through in detail on each of them afterwards:

  • Adhering to y+
  • Consideration of general grid size
  • Extension of refined area
  • Double checking K and Omega values
  • Determining convergence

I have to head out right now so I’ll comment more soon on this specific post.

Cheers.

Regards,
Barry

EDIT:

I’m back so let me break down each pointer.

Your current y+ is not small enough for a 2 layered approach using k-omega SST. Generally, you would want a y+ of less than 1 and to obtain the first layer size of your y+ you can a y+ calculator which will give you the maximum size of the first layer.

I’ve taken your current boundary layer inflation values and computed them here. From a rough calculation your expected y+ is somewhere around 3E-4 m if i’m not wrong, so do redo the y+ calculation and work from there. It is important to have a sufficient y+ value in order to capture the wall effects which in your case is also very important since you’re looking data at surfaces.
image

Right now your grid size looks rather coarse, you might want to check through some literature to see if there is any form of guidelines or “best practices” for determining the average grid size needed. You can then adjust this by the dimension of the box divided by the no of cells that can be changed.

You want to ensure flow propagating towards the car is of better accuracy as well. Extending or adding an additional refinement box in front of the car should help your accuracy.

Make sure you have calculated these values properly and in the correct units. Best to find some literature that provides a formula for your K and omega values.

What is your convergence criteria? At the moment your residuals are reaching a steady-state but they don’t seem to be converging well if you consider that 1E-4 is where a simulation is typically considered converged. You might want to double check on this. Oh yes your force values are completely off, so I suspect it is a mesh issue.

Hope this helps!