Hello everybody,

I need to simulate the whole thermal cycling bench we use to test or products (heat exchangers), in order to investigate flow behavior around the tested parts. Indeed, we suspect some flow recirculation in the bench.

You can find a description of the bench here: http://www.uploadimages.com/myalbum/7652

Here are my main questions: 1. the fans: how would you guys set up a simplified model of the fan? I just have no experience at all when it comes to fan modeling. after checking Fluent help /tutorials, it seems that several strategies are available (moving frame, fan model ...). I think it is important in my model to get a fairly good idea of the swirl.

2. the grills: as you can see in the legend, the grills are not homogeneous. I'm afraid this could affect the flow patterns pretty badly (specially the one the fan rings are welded on! Is there a simple way to account for that? a porous media, I would guess? Is there a standard method to calibrate such a thing?

Thanks in advance for your help.

/JB Leydet

hmm... anybody there?

I can give you some ideas. But be warned that I have little actual experience in implementing these.

The fans:

I'd try to model the fans as source terms in the momentum equations very near the boundary. And I'd start with 'simple' airfoil theory to try and quantify those sources. Also, have you tried to locate design procedures or texts for fans and propellers?

The grills:

The grills represent a flow resistance across the boundary. Depending on the mind set of the modeler, they might be represented as a porous barrier or as a loss coefficient that varies with location. Either way, I think there's a bit of empiricism here.

Maybe this speculation will stir those with more experience to write?

Hopefully!

Thanks for your contribution, anyway.

I think I will start with a rough 2D model, modelizing the fan as a source term, with no swirl. Then I'll play with the swirl, in order to check how much it can affect my results.

What do you mean by "very near the boundary" ?

"very near the boundary"

That is a bit ambiguous!

Two choices you might consider.

Represent the fans (as sources/sinks of the various momenta) in the first row of computational cells next to the boundary (or perhaps a few rows if the fan is 'thick' relative to a cell thickness). Or

Covert your source/sink models into boundary fluxes. This seems a bit easier, in that you don't have to mess with the coding for the equations in the interior.

I'd probably start by making a special boundary condition (second option above) because that puts the fan outside the computing region - which is where it resides on your diagram.

Whichever way you go, it'll probably not work as expected the first time.

Good luck!