Hi guys,

I am a new research student doing experimental work on boundary layer transition under oscillation. The case is 2d Flat Plate. Now, my experimental data need to be simulated or compare with CFD assessment. My problem is i am definetely 'blank' about CFD, particularly simulating and modelling the transition of the boundary layer (LES), plus it's characteristics under oscillation. My point is: 1. Does anybody would like to validated their software by comparing it with my experimntal data? 2. Is there any available 'commercial code' which can simulate my problem? 3. We have Fluent installed in our department (which LES possibility). Is Fluent accurate enough to perform my problmes?

Your attention, response, comments and suggestion are much appreciated.

cheers Jay

Dear Jay, is the plate horizontal or vertical?, your problem can be done using fluent.Initially you need to generate the geometry and mesh using a mesh generation software (like gambit). I suggest you to go through Fluent manuals.

Regards

Sridhar.

Hi Sridhar,

Thank for your response. The flat plate is put horizontally with semi ellipsoidal leading edge.. But how do i simulate the oscillation? So there are two cases here: transition simulation and it's chracteristics undrer oscillation.

Your further comment is much apreciated. By the way..Yes, I am just starting to learn Gambit..

cheers, jay

Transition simulations, as you know, are very difficult to do correctly.

Do you have a natural transiton, caused by instabilities in the laminar boundary layer, or a by-pass transition, caused by convection and diffusion of free-stream turbulence into the boundary layer?

In any case, Fluent is not really suitable for predicting transition. The numerical schemes are probably not good enough, the wall-treatments available with LES are very limited and the eddy-viscosity models in Fluent that I have tried (studying transition on flat plates and on turbine blades) will most often just give you a turbulent boundary layer from the leading edge.

You are probably better off working together with someone who has an in-house code specialized for transition predictions. I noticed that you had a .uk mail address - Mark Saville at Cambridge was a focal point for transition modeling a few years ago. I haven't followed this subject very closely in the last few years but he might still be a good person to contact. I'm sure others can give you better, or at least more current, hints on who to contact. Good luck!

Hi Jonas

What to say..Thanks a lot. It's very good comment and suggestion. My case is not bypass transition. It's a natural transition process, as you mentioned caused by instabilities in the laminar boundary layer. The transition location then be further investigated by oscillating the flat plate. My supervsior suggested me to use Fluent because we realised and simply because I do do not have experience in CFD coding especially for transition prediction. Developing new inhouse code would be a long process and much pain involved. Anyhow, I'll forward your comment to my supervisors and consider another alternative.

About co-operation with someone, yes that what i am really expecting..inviting folks from this CFD community..they probably would like to validate their code using my experiments, meanwhile i can simulate my experiment using their code..It's 'win-win' solution.. Or maybe someone has finished their transition prediction code and would like to offer me their code (in very reasonable good price) before 'selling' it to CFD developer company.. But, of course I must first validate the code.

Yes, I am in UK and I'll try to contact Cambridge.

Thanks again.. More suggestion from anybody would be much appreciated...

Jay

Predicting natural transition with a CFD code is very difficult - you most often need to do something more than just solve Navier-Stokes plus a turbulence model. The classical method is based on solving the Orr-Sommerfeld equations. I think also that there are other integral BL methods and some multi-scale approaches, DNS etc, which I don't really know much about. Hence, if you don't have any experienced people near by contact someone who knows these things - predicting natural transition is a science on its own.

Hi Jonas, (Your are indeed good looking man, Jonas. I noticed your picture standing infront of your '150 CPU Linux Cluster' in 'Fluent News' - Issue-1,2002) Jonas,

Thanks again for your comment. Yes, but for the first, i'dlike to use Fluent and see how good Fluent can actually simulate my case. I am studying it now. I am thinking to apply ' boundary layer' and 'transition' features in Gambit mesh and grid modelling and see what is the outcome in Fluent. For Fluent users: Do I have to make partision between laminar, transition, and turbulent sections in my modelling then use approriate solvers available in Fluent for processing?

Thank for any helps. Jay

Heh, yep, that's me.

The "boundary layer" and "transition" options in gambit are just meshing options (transition here refers to grid-topology and has nothing to do with transition from turbulent to laminar flow).

As I said before, you can not predict natural transition with fluent. You have to set your flow to be either laminar or turbulent.

Hi. Jonas, Of course I know, the the "boundary layer" and "transition" options in gambit are just meshing options, but at least these approaches may be used as a good way to have better meshing.

I have tried using 'laminar' solver and it's worked fine. Now, i want to try 'turbulent modelling' approach, but confusing me little bit--in which way of approaches that i have to use. There are quite some ways to do that e.g One-equation models, Spalart Allmaras, Two-equarion models, Standard k-e; RNG k-e, Realizable k-e, and Reynold stress model and the last of course, LES. In Fluent which modelling that you think is the more reasonalble way to use?. Actually, in my case there will be no transition may occur, because in my experimental investigation, i use low turbulence wind tunnel. In Fluent, i try to simulate the oscillation and expect to get transition by varied the incidence - in steady state condition. Is that sound to you?

Thanks for your further suggetion

Jay

None of the turbulence models in Fluent are capable of predicting natural transition. I'm quite sure that if you turn on a turbulence model you will get fully turbulent results where you have the model activated.

But please go ahead and try to prove me wrong ;-)

I hope I am not a bit late speaking about this subject only now !!! Just when Jonas says that it is not possible to predict transition with Fluent, it reminded me that some years ago I worked on transition on wings using a modified Wilcox k-omega turbulence model (not with Fluent though). I know that k-omega model are now implemented into Fluent 6. So does anyone know if this kind of model could be used to sucessfully model transition with Fluent ?? Or maybe some kind of modifications could be done to this turbulence model to do it correctly.

Hi, no, it's not too late, cause i am the one who really concern and rising this problem. Unfortunately, we do not have fluent 6,o installed in our system (university), so i could not implement the k-omega. Indeed i know that Fluent claim that in Fluent 6 can be used for transitional flows (though tends to predict early transition). However, i am still trying several turbulent models using Fluent 5.4 (Jonas may be right, but i am still trying since i do not have any other tools to use), but in this time i use 'unsteady' instead of steady state that i use previously. It was suggested by my supervisor to continue with k-e/RNG/2layer zonal and Spalart Allmaras, both unsteady, since no transition indication in steady results.

Any more suggestions/comments, please?

Thanks a lot Jayd