CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Optimize for laminar flow, assume it valid for turbulent flow? (http://www.cfd-online.com/Forums/cfx/93649-optimize-laminar-flow-assume-valid-turbulent-flow.html)

Chander October 21, 2011 20:14

Optimize for laminar flow, assume it valid for turbulent flow?
 
Hi,

I want to optimize a design for convective heat transfer and pressure drop.
I have four designs..design 1 ,2,3 and 4.
I want to judge how these designs compare at different inlet Re numbers.
The Re numbers are in laminar region at lowest level and sweep through transition range on the higher side.
I am finding it difficult to get convergence at higher Re both with or without using turbulence models.
However i am able to get converged results at low Re i.e. in the laminar flow regime.

I analyse through CFD and rank the available designs in terms of their performance at low Re i.e. in the laminar flow regime and get a ranking say 2>3>1>4 in terms of their performance for pressure drop and conjugate heat transfer. Now how safe will it be to assume that the same ranking of designs will also hold at higher Re i.e. in the transition and turbulent flow regimes ( and thus avoid CFD analysis at higher Re)?

mauricio October 22, 2011 01:52

-
 
Hello Chander, i Have never worked with Low-RE models and the selection of a turbulent would depend on many factors such fluids, your required accuracy, the geometry itself, etc...

My small input in this is that the rank you obtained probabably will not hold true at turbulent conditions. The fact that the speed is changing will change the convective heat transfer as well, since by chaning the design you're changing geometry then flow field would be different for each heat exchanger when under Re numbers. You should rank them under turbulent flow as well.
Change the boundary conditions that produce those Re numbers. try this guidelines. But generally i doubt that same rank would hold for the high Re numbers...

ghorrocks October 22, 2011 06:05

What turbulence model are you using? Is it turbulent at the entrance or does it transition in the domain? Any relaminarisation in the domain? How far into the turbulent regime do you go?

Chander October 24, 2011 08:05

5 Attachment(s)
Hi Glen,

This is the same problem as I discussed with you in the thread
http://www.cfd-online.com/Forums/cfx...s-numbers.html
I am reattaching the geometry figures here.

Essentially, the variation of Re is 600-4000 at inlet which reduces to about 150-750 at slot nozzle. This variation is shown in the figures attached with the next post

Chander October 24, 2011 08:10

2 Attachment(s)
Please refer to the figures.

Re_at_inlet.jpg : Variation of Re at inlet to inlet manifold
Re_at_slot_nozzle.jpg : Variation of Re at slot nozzle connecting inlet duct to porous medium.

Now convergence at conditions 2 and 3 is very difficult to get.

So the question is that if I optimize my geometry for pressure drop and heat transfer performance at condition 1, will the same optimization be valid for conditions 2 and 3?
In other words, will ranking of a given set of geomtrical designs in tyerms of their performance at condition 1 be valid at conditions 2 and 3?

If it is safe to assume so, than I can work on optimizing the geometry at condition 1 and then just analyse the performance of the optimized design at conditions 2 and 3 rather than separately optimizing for different Re conditions.

Chander October 24, 2011 08:23

Just for clarification:

The first figure above is the variation of Re at inlet and the second figure is the variation of Re at slot nozzle

ghorrocks October 24, 2011 18:32

What is the inlet condition? Laminar or turbulent? This depends on the conditions upstream of the inlet.

Chander October 24, 2011 19:19

Well, the conditions upstream of the inlet are not known as yet.
But it will definitely not be fully developed flow.
At present I am using uniform velocity inlet.
While simulating for conditions 2 and 3 with SST model, I assume low turbulence intensity (as available in CFX option) as the inlet boundary condition for turbulence.

Graham81 October 25, 2011 02:19

How do you define the optimum design (i.e. what is the relative importance of pressure drop and heat transfer to you, as both will be promoted in a turbulent flow)?

Chander October 25, 2011 05:23

@Graham81
The optimized design will of course be the one with minimal pressure drop and maximum heat transfer from solid to fluid.
However, which one of these is given more importance will depend later on how much we can improve each one of them..though heat transfer will most probably have more importance.

Chander October 27, 2011 05:12

@ghorrocks

looking forward to your kind inputs on this questions

Thanks!
Chander

ghorrocks October 27, 2011 06:02

Looks like you are still running with a turbulence model. I have already said I think this is a bad idea.

What can you change in the optimisation process?

Chander October 27, 2011 06:15

Thanks Glen for replying.

Yes, you had already pointed out that using turbulence models in low flow Re is not a good idea.

So I am
a) using laminar flow model for the lowest inlet Re of 600 i.e. condition 1 as shown in above Re plot
b) trying with both laminar model and SST turbulence model for the mmid-range inlet Re of ~2000 i.e. condition 2
c) trying with SST turbulence model for the highest inlet Re of ~4100 i.e. condition 3

During optimization, I plan to change the inlet/ outlet duct widths to change the mass flow rate through the slot nozzles. I may also vary the slot nozzle widths and locations.
So since I am having lot of difficulty in getting convergence in b) and c) above, I am thinking that if I perform the geometry i.e. shape optimization for (a) , then can I assume it to be valid for (b) and (c)?
I think that if for example I achieve a ranking for 4 designs A>B>C>D with A being the most optimal geometry for (a) and D being the least for (a), then this ranking should also hold for (b) and (c) though design A may not be THE optimal design for (b) or (c).
I am not sure if this approach is foolproof.

Chander October 30, 2011 18:46

@ghorrocks

eagerly awaiting your kind inputs on this topic

Thanks!
Chander

Chander November 5, 2011 20:51

Any inputs please.

Thanks!

ghorrocks November 6, 2011 06:06

The SST turbulence model should have no problems converging on Low-Re flow. If you can get laminar to converge but not SST then something is wrong in the setup of your model.


All times are GMT -4. The time now is 17:16.