Turbulent flow in a plane channel gradually turns laminar

[Jundi He]

Hello Neks,

I am currently carrying out an LES of a periodic plane channel flow (Re=11,000) using the recycle inlet method. The whole flow field is initialised by random perturbations equilivant to 40% turbulent intensity, but the flow gradually turns laminar after running for some time. Have anyone met the same problem? Should I do some special treatment to maintain the turbulence? Thanks! 

Regards!
Jundi

[Yuan, Haomin]

Are you running in non dimensional ? 

Have you checked inlet velocity profile is mapped correctly from the interpolation plane? 

Get Outlook for iOS

---

From: 'Jundi He' via Nek5000 <nek...@googlegroups.com>
Sent: Saturday, September 16, 2023 6:58:44 PM
To: Nek5000 <nek...@googlegroups.com>
Subject: [nek5000] Turbulent flow in a plane channel gradually turns laminar

 

--
You received this message because you are subscribed to the Google Groups "Nek5000" group.
To unsubscribe from this group and stop receiving emails from it, send an email to nek5000+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/nek5000/2453e4ea-4870-45b5-bb60-aa7bf693fec8n%40googlegroups.com.

[Philipp Schlatter]

I guess the main questions are:
- is the Reynolds number correct and the problem correctly scales?
- is the domain size large enough to sustain turbulence
- is the streamwise forcing correctly turned on
- is the resolution sufficient to sustatin turbulence.
- are the initial conditions such that turbulence is really initiated.

How quickly is turbulence gone (in convective units)?

In particular at these Reynolds numbers, you need to make sure that you
force enough large scales, as small scales wtih wrong phases are quickly
disappearing. Higher amplitude does usually not help, it just kills the
initial time step, and is quickly dissipated away.

As an alternative, we made good experience with putting in a trip force
during the first 50 or so time units, rather than an initial conditon.
If that is an alternative for your case, I don't know.

Philipp

On 2023-09-17 17:30, 'Yuan, Haomin' via Nek5000 wrote:
> Are you running in non dimensional ?
> Have you checked inlet velocity profile is mapped correctly from the
> interpolation plane?
>

> Get Outlook for iOS <https://aka.ms/o0ukef>
> ------------------------------------------------------------------------
> *From:* 'Jundi He' via Nek5000 <nek...@googlegroups.com>
> *Sent:* Saturday, September 16, 2023 6:58:44 PM
> *To:* Nek5000 <nek...@googlegroups.com>
> *Subject:* [nek5000] Turbulent flow in a plane channel gradually turns

> laminar
> Hello Neks,
>
> I am currently carrying out an LES of a periodic plane channel flow
> (Re=11,000) using the recycle inlet method. The whole flow field is
> initialised by random perturbations equilivant to 40% turbulent
> intensity, but the flow gradually turns laminar after running for some
> time. Have anyone met the same problem? Should I do some special
> treatment to maintain the turbulence? Thanks!
>
> Regards!
> Jundi
>
> --
> You received this message because you are subscribed to the Google
> Groups "Nek5000" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to nek5000+u...@googlegroups.com

> <mailto:nek5000+u...@googlegroups.com>.

> To view this discussion on the web visit
> https://groups.google.com/d/msgid/nek5000/2453e4ea-4870-45b5-bb60-aa7bf693fec8n%40googlegroups.com

> <https://groups.google.com/d/msgid/nek5000/2453e4ea-4870-45b5-bb60-aa7bf693fec8n%40googlegroups.com?utm_medium=email&utm_source=footer>.

>
> --
> You received this message because you are subscribed to the Google
> Groups "Nek5000" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to nek5000+u...@googlegroups.com

> <mailto:nek5000+u...@googlegroups.com>.

> To view this discussion on the web visit

> https://groups.google.com/d/msgid/nek5000/DM8PR09MB74311658F82C021DF517F6B6A2F4A%40DM8PR09MB7431.namprd09.prod.outlook.com
> <https://groups.google.com/d/msgid/nek5000/DM8PR09MB74311658F82C021DF517F6B6A2F4A%40DM8PR09MB7431.namprd09.prod.outlook.com?utm_medium=email&utm_source=footer>.

[Jundi He]

Hi,

Many thanks for the suggestions and the thoughts, the problem has been solved. The flow I am trying to simulate is at Re=180 based on the friction velocity and I am simulating it using the dimensional way. The channel should be big enough with a streamwise length of 6d and a transverse length of 4d in which d is the distance between the two planes. Streamwise force is not used as the periodicity is implemented by the recycle method. The mesh resolution is very fine with near-wall y+ ~ 1. The flow is initialised using a uniform velocity with a random perturbation equivalent to 40% of the bulk velocity. 

The solution that I use is to initialise the flow using a higher velocity equivalent to a higher Reynolds number (Re=550) and run it for some time so as to 'flush' the plane channel, then gradually reduce the velocity to a lower value (Re=180). I guess Re=180 based on friction velocity is a very moderate Reynolds number and the flow might just go laminar even with initial perturbations. 

The comparison is shown in the below figure, the two flows have the same bulk velocity (ub=1.0). In the upper one, the whole channel is initialised with a uniform velocity (u=1.0, equivalent to Re=180) and random perturbations everywhere, it develops to a parabolic laminar flow after running for some time. In the lower one, the whole channel is initialised with a uniform velocity (u=5.0, equivalent to Re=550) and the high-Re flow is kept for some time, then the velocity is reduced back to ub=1.0 gradually. The lower one can maintain a turbulent state for a long time without changing to laminar. 

Regards!

Jundi