CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   K-eps what should converge first? what last? (http://www.cfd-online.com/Forums/fluent/44573-k-eps-what-should-converge-first-what-last.html)

 Nick May 2, 2007 06:43

K-eps what should converge first? what last?

When converging on the residuals is there any order in how convergence should occur. Basically before i did a grid analysis to reduce the aspect ratio of elements near the airofoil i had "k" being the last to converge followed by epsilon and then continuity. Now i have continuity being the last to converge?? is there any problem with this??

Also when i try and scale my mesh evenly in X and Y obtain a finer mesh all my residuals go unstable and everything begines to rise from the first iteration? What is going on

My coarse mesh has lengths down about .0001 my next step mesh i wante lenghts down around .00005 Is this to small??

 Phil May 2, 2007 06:52

Re: K-eps what should converge first? what last?

You shouldn't worry which order they converge, instead be thankful that they all did!

In my opinion if continuity converges last then the model isn't quite as good as if it converges first but you still get the same result both ways.

You may need to use relaxation factors if the solution is unstable or there may be a problem with your model. Try using the model you got to converge and gradually refining the mesh. The more cells you use the harder to converge and the more likely you need RF's.

I am however unsure what you mean by 'evenly in X and Y' and 0.001 means nothing without knowing more about the dimensions of the geometry. You should take your model to a lecturer/researcher.

ONE ALTERNATIVE IS THAT THE SOLUTION IS UNSTEADY (VARIES WITH TIME) AND YOU NEED THE UNSTEADY SOLVER.

Check all boundary conditions and the quality of the mesh.

Phil

 All times are GMT -4. The time now is 17:45.