Learning Forum

[nc952]

Hi,

I am trying to find flow over a 2D square at 24.83m's inlet and using a Laminar flow-

Method: Fractional step-Non-iterative Time Advancement.

ALso, running it for transient state.- pressure-based.

It's a 2D square- 0.2m by 0.2m.

[nc952]

[Rob]

Check the cell quality, the above looks OK but I can't tell by looking.
If you're getting that level of back flow something's gone very wrong, I'd not expect any/much in the above. Turn off the prevent back flow: you don't need it. What is the time step relative to the time it takes the flow to cross a cell?

[nc952]

Thank you. Yes, I did that. THat helped, then I also changed sphere of influenece to body instead of face. That worked. But now, It's diverging after 680 iterations.
I am using 0.1 time step size. I want to capture the vortices behind the square.
This is my mesh quality.
Mesh Quality:

Minimum Orthogonal Quality = 1.79371e-02 cell 252753 on zone 3 (ID: 114212 on partition: 0) at location ( 1.97455e-01, 2.00001e-01)

Maximum Aspect Ratio = 2.99920e+03 cell 252753 on zone 3 (ID: 114212 on partition: 0) at location ( 1.97455e-01, 2.00001e-01)

Any idea why is it diverging?

[nc952]

[Rob]

That's not converging at any part of the run, and then fails. I suspect it's the mesh quality: get the ortho skew over 0.1 and I suspect it'll look at lot better. You've also fallen into the classic trap: flow is separating so just looking at y+ means the streamwise mesh isn't checked, and with separation you want an aspect ratio in the 10's rather than (at most) a few hundred.