oscillation
 

Hi everyone,

I am simulating a case using standard k-e model with standrad wall function.From the residual monitor,all the parameters perform well druning 1,0000 interations,however after that only the continuity exhibites oscillation with small scale range goes up and down and maintain till now(1,3000),others keep steady in the height of 1e-05.

Does anyone can tell me where is the problem and how can i check whcih part is wrong?

Thanks very much for your reply!:)

Hi!

there might be a problem with the grid. Check the cell skewness (e.g. in Gambit) - it should not be to high (below 0.6) ...
Also, in areas, with high gradients, the cell size should be fine.

Another problem might be, that your yplus is not in the right range for the std wall function ( 3 < yplus < 300)

Best wishes
Ralf

Hi Ralf,

Thanks very much for your answer, i chech the y plus and all them are within the range 0.3-120,maybe it is better to use Non equilibrium wall functions instead of SWF.Like you said,the cell skewness is also a factor.

By the way,i am sorry to mention that the continuity has already decreased to 1e-08,does this is low enough to say the results is good even there is small osillation?

Best wishes

Huijuan

Hi!

turbulence modeling and the choice of the right turbulence model is a big thing.

It is difficult to guess, what model will be appropriate for your case - depends on fluid properties, Re number, .....
I suggest near wall treatment models for resolving the BL and correctly displaying heat and momentum transfer. ....

For convergence: the residuals give you a good idea, whether your simulation is "finished" or not - and 10e-08 is a very low value...
I would suggest to look at one additional thing: use a surface monitor of the parameter, that is interesting.

e.g. you want to simulate integral heat transfer, monitor that value. When it is (almost) constant for a number of iterations, your simulation is good enough.
Further iterating might lower the residuals, but not change the heat transfer any more and thus will not contribute to solution accuracy!

If you are interested in the flow field, create a number of points in areas with high velocity gradients and monitor the velocities on that points...

Best wishes
Ralf

Hello...

You may restart the case after smoothing-swapping the grid just in case you haven't done it before. Your solution appears to have reached convergence. You may also go for the second order discretization schemes.

It won't say 'solution converged' if you have set unreachable criterion. In any case, to comment further some solution details are required.

Hi!

i am just searching the old fluent forum entries... and i found an interesting discussion about that issue:

http://www.cfd-online.com/Forums/flu...tml#post127748

Thanks very much for all the kind help and information!