Learning Forum

[goktug.yilmaz]

Hello all,

I am conducting an analysis in Fluent software, focusing on flow separation at different angles of attack, such as 6 and 10 degrees. However, I have noticed that the residual values display fluctuating characteristics and the lift and drag coefficients do not converge to a certain value. I have attached an image showing the velocity vectors at 4, 6, and 10 degrees and residuals at 6 degrees for reference. For 4 degrees, I have converged results because there is nearly no separation. But when it separates, residuals start to oscillate shown in the image below. 

To obtain a more accurate analysis, I am seeking guidance on how to approach this issue. I have tried reducing the velocity at 6 degrees to decrease the flow separation and subsequently decrease the residual values, resulting in a slight convergence of the lift and drag coefficients. However, my analysis involves a Mach number around 0.8, and I am currently using the kw-SST method.

I would greatly appreciate any suggestions or solutions that you may have regarding this problem. Thank you in advance for your assistance.

Best regards,

[Federico Alzamora Previtali]

I suspect some inherent unsteadiness from the flow separation which prevents the solution to converge to steady-state.

[goktug.yilmaz]

Thanks for your answer. That is what I also suspect. However I do not know how to solve the problem. I tried to reduce time scale from 1 to 0.4, however it does not work.

 

[Federico Alzamora Previtali]

You can confirm if this is indeed the issue by inspecting where these larger residuals occur in your domain: #2067529 - How can I post-process my residuals to better understand convergence issues ? (ansys.com).

If the solution cannot converge in steady-state, you might need to switch to transient simulation.

[Rob]

To add, put some monitor points into the domain in regions of interest and look at (for example) velocity fluctuations on those points. Plot wall shear stress on the upper & lower wing surfaces every few iterations too. Then review the results to see what's changing, and by how much.