Dear Experts,

Is it possible to show in a single simulaiton to show all three types of flow (laminar, transition and turbulent) using FLUENT or STAR-CD or CFX? For example in a Reynolds apparatus, we can see all three types of flow in a single tube. How can I show this using CFD simulation?

Thanks,

Harendra

Only with DNS.

Up to now, all the transition models using RANS methods have prooved insuccessful in predicting transition. CFX claims to be able to predict transition (http://www.ansys.com/assets/tech-briefs/PDF0422.pdf) for industrial applications with the RANS approach, but I would not bet much on that ...

The straight-forward answer to your question is: Yes, of course, you can do that. There is no big problem showing laminar, transitional, and turbulent flow, provided that you have some way of telling which behavior is prevalent at which location in the flow field.

However, that's quite a lot to ask: How do you know the position and extent of the transition? If you're happy with specifying the transition point by yourself, I think it should be fairly easy to do, the procedure depending on the code and turbulence model. I would also assume that many of the commercial CFD codes offer some kind of transition model that will determine the transition point based on some semi-empirical theory.

So the short answer to the question you posed is: Yes, there are possibilities to do that.

How realistic and accurate this is going to be, is a different question, that depends on a lot of other issues.

As I know, In the fluent user's guide, It is mentioned that you can specifiy the laminar zone and turbulent zone if you know where the trantional point is. I am also looking for how to do that.

As mentioned by other people, you are in a DNS-like approach. In principle, you can do that with a numerical solver as Fluent or others. The main challenge is to have a grid sufficiently fine to capture all the scales which are going to develop on the grid without dissipation introduced by the schemes which are not accurate enough. Through the consistency theorem, you can do that with a very fine grid. In practice, I am afraid that dissipation or round-off errors may jeopardize your approach.

Give a try (be carefull, no turbulence models: you must use laminar + time dependent and the less dissipative scheme you can have). If you manage to get something, check if it is quantitatively correct (as the good transition position,...)

Many thanks to all of you and my special thanks to Mr. Guillaume for giving me working guidelines

-Harendra