[Nek5000-users] CVODE and IOFS

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

Dear nek5000 team,

I have modified the makeq.f routine to include the line “call convab” into the condition “ifcvode”, just before "call wlaplacian…".
According to the test I performed with my vortex convection case as well as the “lowMach_test” case from the example directory of nek, now it seems to works as it should. I get the same results as with no-CVODE, or CVODE without the characteristic method.

If I am right, with that change, now convective and diffusive terms of heat and scalar equations are solved within CVODE, while the velocity in the hydrodynamic system is computed with the characteristic method (OIFS, sorry for the mistake in my previous email).

Could you confirm me that it is an appropriate fix for the problem (supposing it was initially a bug)?
Thank you for your help!


Best regards,
Emmanuel



On 22 May 2014, at 23:41, nek5000-users at lists.mcs.anl.gov<mailto:nek5000-users at lists.mcs.anl.gov> wrote:

Dear nek5000 team,

I have set-up a test case that consist of a free convection of a 2D vortex in a periodic domain. In the initial condition, I have put a line of passive scalars superimposed onto the vortex. If I use the standard solver for temperature and PS (p16 = 0 or 1), everything works fine. My vortex is convected and the line of passive scalars is distorted by the vortex and diffused progressivly. It works well even with IFCHAR= T or F. Perfect.

However when I switch to the CVODE solver (p16 = 3) for time integration of temperature and PS, it works only with IFCHAR=F. When IFCHAR is T, my line of passive scalars is not convected by the flow and doesn’t interact with the vortex. Only the progressive diffusion is taken into account.

A quick look at the routine makeq.f confirms this behaviour. If IFCHAR is F, the routine convab is called to compute the convective operator, and then the diffusion operator is computed with the call of the routine wlapacian. If IFCHAR is T, only this latter diffusion part is computed. In that case I didn’t find where the convection part is computed. It seems that it is missing.

My question is just: is this normal? Why only the diffusion operator is computed when CVODE and IOFS are used together? Is it a bug? Am I missing something in the theory?

Than you in advance for your explanations !

Best regards,
Emmanuel



_______________________________________________
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/20140523/572ac4b8/attachment.html>