[Nek5000-users] turbChannel: Questions about targetCFL and dealiasing

[nek5000-users at lists.mcs.anl.gov]

The code implements OIFS (operator integrator splitting factor) not an implicit fractional step method! A convection problem is integrated in substeps with RK4. For a targetCFL of 3.5 you need at least two substeps to be stable. You can increase the targetCFL even further depending on the desired accuracy.  If this helps to lower time to solution depends on the cost of the pressure solve.

If you want to run the solver as fast as possible turn off dealiasing. In all other cases turn it on (lxd has to have the right size e.g. 3/2 rule).

On 10 Feb 2018, at 18:28, "nek5000-users at lists.mcs.anl.gov <mailto:nek5000-users at lists.mcs.anl.gov> " <nek5000-users at lists.mcs.anl.gov <mailto:nek5000-users at lists.mcs.anl.gov> > wrote:
 Dear Nek users,

 
I have some fundamental questions in terms of CFL and dealiasing in the example of turbChannel. In the .par file, I noticed that the default setting of targetCFL is as high as 3.5, which really makes the code run very fast. But, typically, this CFL value is around 0.5 using implicit fractional time-step approach based on the FV method. Why CFL could be this high here while not compromising the stability? Is it because of the spectral element method?

 
And Could you also please tell me when shall I turn on or off dealiasing?

 
Regards,

 
Emily 


_______________________________________________

Nek5000-users mailing list

Nek5000-users at lists.mcs.anl.gov <mailto:Nek5000-users at lists.mcs.anl.gov> 

https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20180211/34157e3a/attachment.html>