CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Decreasing residuals (http://www.cfd-online.com/Forums/fluent/31658-decreasing-residuals.html)

 Robert June 15, 2003 12:56

Decreasing residuals

How can I decrease the residuals and obtain a convergent solution if my cfl is O -3 and my under relaxation factors are 0.15 or lower. Increasing the cfl will diverge the solution. This is a scramjet 2D model using ethylene as the fuel in a three step reaction.

 ap June 15, 2003 15:35

Re: Decreasing residuals

What residuals are not decreasing?

If you are using a high order discretization scheme, try to use a lower one for the first iterations. For example, you can obtain a converged solution with the first order scheme, and refine it using a higher one.

However, your under-relaxation factors are really low and this increses the number of iterations you need to obtain a converging solution. What happens if you increse them?

Hi

ap

 Robert June 15, 2003 17:52

Re: Decreasing residuals

All residuals are remain approximately constant. My scheme is first order. Increasing them can cause divergence. Especially the reactions underrelaxation. I am trying to find the most efficient way to find a convergent solution. Any suggestions would be greatly appreciated.

 Jim Clancy June 15, 2003 20:38

Re: Decreasing residuals

Robert,

For some cases It is not necessary to monitor residuals as the only convergence criterion. If they became constant but they close to the convergence level, you may check other things like mass flux, or some force in certain areas of the domaoin..

Jim

 ap June 16, 2003 15:57

Re: Decreasing residuals

If your residuals stay constant after a high number of iteration, maybe you gave an initial solution close to the converged one. In that case, residuals can't reduce as much as you're asking. You can check this changing your initial solution.

However, I agree with Jim Clancy. You should check if other parameters are correct. Pay great attention to reaction data, which seem to be the most difficult to manage in your case. If they're ok, you can consider your results as a solution.

Hi

ap

 zxaar June 18, 2003 09:05

Re: Decreasing residuals

yes,

a solution should be stable, means once u say solution is reached it shall not change with further iterations. (after u sau the convergence criteria with residuals is met)

 Alamgir June 25, 2003 01:26

Re: Decreasing residuals

This small residual means u r taking values only 15% from the previous iteration... so if u still on a straight line.. that means there is something wrong.. so I wanna say try this way... in the control (solutions) box just deselect other equations keeping main one (flow).. solve only for this, after getting convergence.. select others with same urf.

Alamgir

 All times are GMT -4. The time now is 23:39.