[Nek5000-users] Effect of CFL number.

Tue May 27 17:43:05 CDT 2014

Hi Kamal,

Not only different numerical methods have different transitions, but different
experiments as well.   The point here is that the sensitivity you observe
(across numerical methods, or across spatial/temporal resolution for a
given method, or across the initial condition, or across different experiments)
is due to physical phenomena, modulated by the method of investigation.

If we take, say, an experiment as a gold standard, the question is: which one?

Even Osborne Reynolds experimental apparatus does not give the same 
transition point today as it did when he did his experiments in the 1800s.
Why?   Vibrations due to truck traffic in the street...    That input is not
accounted for in any numerical simulations of pipe flow.

The problem of course is that this is a subcritical bifurcation, unlike
Rayleigh-Benard, for which the transition is triggered by a classic
linear instability and is therefore repeatable.

Paul

________________________________________
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: Tuesday, May 27, 2014 4:27 PM
To: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Effect of CFL number.

Hi Kamal,
Depending on what you want to do, I would stay away from the
transitional region altogether, as you also might see spatial
localisation which is maybe not what you want. The standard case for
low-Re turbulence, which is sustained even in small boxes, is the
Re_tau=180 case, which corresponds to Re_b=5300 or ReD=3464
(approximately). At that Re you also have a good deal of data to compare to.
Best regards,
Philipp

On 2014-05-27 23:11, nek5000-users at lists.mcs.anl.gov wrote:
> Thanks Paul and Philipp. I have small cross wise velocity of v =
> 0.001 as a perturbation at the inlet to trigger the turbulence but as
> you said the sensitivity of the system determines the transition.
>
> Mr.phillip If you don't mind, could you please tell me at above which
> Re I can expect a sustained turbulence independent of the Mesh and
> order , Because I tend to see different Numerical methods (FVM,FEM)
> have different critical point of transition for a similar case .
>
> Thanks in advance for your replies
>
> Kamal
>
> On 27 May 2014, at 23:01, nek5000-users at lists.mcs.anl.gov wrote:
>
>> Dear all, just to add also the domain length to Paul's answer;
>> Re=2000 for pipe flow is in fact slightly below the "critical"
>> Reynolds number for sustained turbulence. In a shorter pipe,
>> turbulence will decay even faster. This indicates, as Paul
>> mentions, the sensitivity to initial conditions and geometry, even
>> though I would expect all DNS to eventually decay to laminar at
>> that Re.
>>
>> Best regards, Philipp
>>
>> On 2014-05-27 18:35, nek5000-users at lists.mcs.anl.gov wrote:
>>>
>>> Hi Kamal,
>>>
>>> In my view, you are trying to run in a regime that is actually
>>> not reproducible. You are right at the transition Re, which
>>> implies strong sensitivity to initial conditions and under such
>>> circumstances you should not expect two simulations at two
>>> different resolutions to converge to the same result, just as you
>>> would not expect two turbulent simulations to follow the same
>>> trajectory point-by-point, only on average.
>>>
>>> Paul
>>>
>>> ________________________________________ 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: Tuesday, May 27, 2014
>>> 11:23 AM To: nek5000-users at lists.mcs.anl.gov Subject:
>>> [Nek5000-users] Effect of CFL number.
>>>
>>> Dear Neks,
>>>
>>> I would like to know what is the effect of CFL number on a
>>> turbulence simulation in SEM.
>>>
>>> I use polynomial order of lx1 = 5 and lxd = 8 and get a C = 0.176
>>> for a Re = 2000 with dt = 0.001  in pipe flow where I find a
>>> transition to turbulence. when I decrease my time step 'dt =
>>> 0.0001 ' further my pipe becomes fully turbulent for the same
>>> mesh.
>>>
>>> When I increase the order lx1 = 7 and lxd = 10, I find a
>>> completely steady solution, instead of a transition to
>>> turbulence.
>>>
>>> It would be nice what if some could please tell me what
>>> parameter determines the accuracy of the simulation in terms of a
>>> turbulent flow because I am not able to reproduce the result with
>>> a different mesh or with higher order.
>>>
>>> Thank you,
>>>
>>> Kamal _______________________________________________
>>> 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