I was playing around a bit this weekend and there’s something I found:
I don’t know wether it is a wise thing to do but I found that you can use Slip walls instead of Pressure outlet boundary condition in simple cases. I mean in my current project where I simulate flow around a sphere (https://www.simscale.com/workbench?publiclink=ae4d6967-4840-4f82-a1a3-a38b0aa8b052) I had a setup with 1 inlet, 1 outlet, 4 slip walls and 1 inlet, 5 slip walls as well. The results and the runtime were identical in each cases.
These are my findings regarding to this topic:
-With Hex-dominant automatic mesh generation the drag force is:
- accurate (37% error with moderate, 13% with very fine mesh) at Re=10^2 (laminar) flow
- around twice as much as calculated at Re=10^6 (turbulent) flow
-With manual mesh (by trying to implement similar approach like @AsadAli used in the Parachute project)
- around twice as much as calculated at Re=10^2 (laminar) flow
- accurate (32% error with moderate mesh) at Re=10^6 (turbulent) flow
(The settings are identical only the velocity was changed from 0.00155 m/s (laminar) to 155m/s (turbulent))
@pfernandez; since my model is only a one-surface part I experimented with a split model based on your suggestion. The funny thing is that in the laminar case I expect for the drag force:
And with the split model I got:
In this case it is not twice the value but around half.
So with all this information this phenomenon gets more and more interesting for me. Although I tried I couldn’t find out what went wrong what should be done differently.
If some “more-experienced-than-me” CFD guy would review this topic I’m sure it’d be useful for all of us.
And for sure I’ll also investigate it further.