CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Convergence Question (http://www.cfd-online.com/Forums/main/86104-convergence-question.html)

racerdude777 March 14, 2011 15:59

Convergence Question
 
5 Attachment(s)
I am running an alpha sweep on a serrated leading edge wing (top view attached entitled "Wing"). Basically setting the wing at varying angles of attack with respect to the free stream in order to obtain lift and drag coefficients. However the simulation was having trouble converging at a low value and i'm wondering if anybody can comment on why this is and/or how to ensure it converges lower.

The simulation is run at a Reynold's Number of 3 million and the number of cells are listed on each picture with the Spalart-Allmaras as my turbulence model. You can see that as cell density increases, some of the residuals smooth out but they don't drop more than three orders of magnitude. The flow domain, as shown in "Slide1" is divided into a higher density region (Body#1) surrounding the Wing and a lower density region (Body#2) in the outer region.

This simulation is in Star-CCM+ but i'm hoping this question about residuals and convergence is generic enough to apply to CFD as a whole. I would also like to avoid playing with under-relaxation values in order to ensure accuracy.

Martin Hegedus March 14, 2011 21:54

Where is your maximum residual error occurring?

cfd_newbie March 14, 2011 23:07

Quote:

Originally Posted by racerdude777 (Post 299384)
I am running an alpha sweep on a serrated leading edge wing (top view attached entitled "Wing"). Basically setting the wing at varying angles of attack with respect to the free stream in order to obtain lift and drag coefficients. However the simulation was having trouble converging at a low value and i'm wondering if anybody can comment on why this is and/or how to ensure it converges lower.

The simulation is run at a Reynold's Number of 3 million and the number of cells are listed on each picture with the Spalart-Allmaras as my turbulence model. You can see that as cell density increases, some of the residuals smooth out but they don't drop more than three orders of magnitude. The flow domain, as shown in "Slide1" is divided into a higher density region (Body#1) surrounding the Wing and a lower density region (Body#2) in the outer region.

This simulation is in Star-CCM+ but i'm hoping this question about residuals and convergence is generic enough to apply to CFD as a whole. I would also like to avoid playing with under-relaxation values in order to ensure accuracy.

There is no fixed way for saying weather a simulation (based on residual fall) is converged or not, it's dependent on the problem, grid and solver too. But if your residuals are swingeing then you should consider creating a better mesh for your problem. Also you should look at is weather your quantity of interest is converging or not (ex. C_L or C_D)


All times are GMT -4. The time now is 06:11.