[Nek5000-users] convective boundary conditions

Thu Mar 10 06:35:15 CST 2016

Dear Philipp,

thank you very much. I am running a 3D numerical simulation of a jet 
with time dependent velocity conditions at the inlet surface 
(periodically ux=uy=uz=0). Since it is an open flow, I am having some 
troubles with the boundary conditions. Specially at the outlet surface, 
where I have a couple of vortex traveling across the boundary. The flow 
is laminar, however I have used ON boundary conditions and I have some 
reflections. I have also tried with turb_outflow subroutine, rq=2 (the 
simulation crash) and rq=100 (it is affecting the flow and it also has 
some reflections). I think that the solution to solve this problem could 
be using convective boundary conditions at the outlet...

Regards,
SL

El 10-03-2016 23:16, nek5000-users at lists.mcs.anl.gov escribió:
> Dear all,
> indeed we studied different way to have "better" outflow conditions
> for Nek. In particular we were interested in open (or semi-open) flows
> in which the nozzle condition does not properly work (at least to our
> experience, as the free-stream was disturbed). We have thus studied
> two different alternatives: the convective condition and the
> stabilized stress-free condition. The former follows essentially
> standard text books, whereas the latter was presented in a paper by
> Dong et al. in JCP (2014). If you are interested in any of these,
> please let me know a little bit better what case you are studying,
> such that we could prepare a code that suits your situation.
> 
> Best,
> Philipp
> KTH Mechanics
> 
> On 2016-03-10 07:21, nek5000-users at lists.mcs.anl.gov wrote:
>> Because you impose a divergence in the last layer of elements, it
>> emulates an acceleration (or suction) only in that last layer of
>> elements without affecting the rest of your domain.
>> 
>> The effect the two produce is similar in the sense that it ensure
>> that turbulent vortices don't recirculate back into your domain
>> through the outflow boundary.
>> 
>> Regards
>> 
>> Goran
>> 
>> ________________________________________ From:
>> nek5000-users-bounces at lists.mcs.anl.gov
>> <nek5000-users-bounces at lists.mcs.anl.gov> on behalf of
>> nek5000-users at lists.mcs.anl.gov <nek5000-users at lists.mcs.anl.gov>
>> Sent: Thursday, March 10, 2016 1:14 AM To:
>> nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users]
>> convective boundary conditions
>> 
>> Hi Goran,
>> 
>> thank you for your answer. However I still not understand very well
>> what happens in your simulation if you are using unphysical boundary
>> conditions... Activating turb_outflow subroutine, are you emulating
>> a suction (from the boundary to the inner domain)? In this case, is
>> it not affecting your physical results from your numerical
>> simulation? Please, correct me if I am wrong...
>> 
>> One more question, the effect that you are producing with the
>> turb_outflow subroutine, is it similar to applying convective
>> boundary condition?
>> 
>> Thanks again. SL
>> 
>> 
>> 
>> El 10-03-2016 04:27, nek5000-users at lists.mcs.anl.gov escribió:
>>> SL,
>>> 
>>> I had the same problem before. I've had to set rq = 200. If your
>>> flow is very turbulent, just turn up that value a lot.
>>> 
>>> No, it's not physical because you're imposing a divergence on the
>>> last layer of elements. You can just exclude that part of your
>>> solution in your post processing.
>>> 
>>> Regards,
>>> 
>>> Goran
>>> 
>>> 
>>>> On Mar 9, 2016, at 21:58, "nek5000-users at lists.mcs.anl.gov"
>>>> <nek5000-users at lists.mcs.anl.gov> wrote:
>>>> 
>>>> Hi Neks,
>>>> 
>>>> regarding my question about convective boundary conditions:
>>>> 
>>>> In the problem that I am solving, there are vortex traveling
>>>> across the boundary. Using O boundary conditions, the solution
>>>> blows up (due to the negative flux of the vortex). On the
>>>> contrary, if I use ON, the simulation continues running, but I
>>>> get very strong reflections.
>>>> 
>>>> In addition, although my flow is laminar, if I use turb_outflow
>>>> subroutine (with O boundary conditions) with rq=2, the
>>>> simulation works better, but at a certain moment, it also
>>>> crashes. I am wondering if I should increase rq to, let say 3, or
>>>> maybe I should use a different boundary condition, i.e. non
>>>> reflective.
>>>> 
>>>> I would like to ask, if I am using turb_outflow subroutine I am
>>>> imposing grad U>0 in the boundary (please correct me if I am
>>>> wrong). The question is, is it physical? What I am doing is
>>>> -pn+(1/Re)(n*grad(U))>0 ? What does it mean?
>>>> 
>>>> Finally, I have seen that in 2015 you implemented some
>>>> non-reflective boundary conditions and they were working very
>>>> well. Please, could you help me with this issue? (i.e.: a piece
>>>> of advise or some procedure to follow...)
>>>> 
>>>> Thank you very much in advance. SL
>>>> _______________________________________________ Nek5000-users
>>>> mailing list Nek5000-users at lists.mcs.anl.gov
>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>> _______________________________________________ Nek5000-users
>>> mailing list Nek5000-users at lists.mcs.anl.gov
>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>> 
>> 
>> _______________________________________________ Nek5000-users mailing
>> list Nek5000-users at lists.mcs.anl.gov
>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>> _______________________________________________ Nek5000-users mailing
>> list Nek5000-users at lists.mcs.anl.gov
>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>> 
> _______________________________________________
> Nek5000-users mailing list
> Nek5000-users at lists.mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users