[Anna Tian]

Hi

Here is my approach to detect the flow regime for flow in complex geometry:

1) Run the simulation by SST model (or other turbulence model)

2) Look at the turbulence intensity value to know the flow regime


Question:

1) Does this approach make sense?

2) What is the range of the turbulence intensity value corresponding to laminar flow? And transitional flow?

[MachZero]

Anna, what is your application? What kind of model is this. Internal flow or external? There is no strict guideline for laminar vs turbulent, it is model dependent. Tell more about the model and what you are trying to get out of it

[Anna Tian]

Originally Posted by MachZero

Anna, what is your application? What kind of model is this. Internal flow or external? There is no strict guideline for laminar vs turbulent, it is model dependent. Tell more about the model and what you are trying to get out of it

This is for internal flow in heat exchanger for electronics cooling. I need to find out the flow is laminar or turbulent. Then I can choose an appropriate model (If it's detected to be laminar, then laminar model. If's it's detected to be turbulent, then SST model.) to resolve the overall pressure drop.

[FMDenaro]

Originally Posted by Anna Tian

This is for internal flow in heat exchanger for electronics cooling. I need to find out the flow is laminar or turbulent. Then I can choose an appropriate model (If it's detected to be laminar, then laminar model. If's it's detected to be turbulent, then SST model.) to resolve the overall pressure drop.

make an estimation of your main non-dimensional parameters (Re, Pr, Ra), laminar conditions are very rare in practical applications, the reality is that laminar, transitional and turbulent flow regions often cohexist.

[Anna Tian]

Originally Posted by FMDenaro

make an estimation of your main non-dimensional parameters (Re, Pr, Ra), laminar conditions are very rare in practical applications, the reality is that laminar, transitional and turbulent flow regions often cohexist.

For complex geometry, Re always doesn't make a lot of sense. I have Re = 100 with very good grids quality but the simulation doesn't converge any more with laminar model. How to use Pr and Ra to tell the flow regime?

Could turbulence intensity value also give a hint?

[Anna Tian]

Originally Posted by FMDenaro

make an estimation of your main non-dimensional parameters (Re, Pr, Ra), laminar conditions are very rare in practical applications, the reality is that laminar, transitional and turbulent flow regions often cohexist.

Could anyone get a hint about the flow regime from the turbulence intensity value?

[MachZero]

I don't work with electronics much so my opinion definitely doesn't count as fact here. Are you calculating re based on channel height or the height of an object that air is moving around? 100 is low enough that I would use a laminar model personally but would make sure my boundary layers are well resolved. Just my two cents

[Anna Tian]

Originally Posted by Anna Tian

For complex geometry, Re always doesn't make a lot of sense. I have Re = 100 with very good grids quality but the simulation doesn't converge any more with laminar model. How to use Pr and Ra to tell the flow regime?

Could turbulence intensity value also give a hint?

Could anyone get a hint about the flow regime from the turbulence intensity value?

[triple_r]

As someone mentioned in previous answers turbulence intensity that you get from simulation is model dependent. In my experience SST was good for internal flows in shutting itself off when flow was laminar, giving very low turbulence viscosity ratios (I ran a case with laminar and SST model, for low enough Reynolds numbers, SST results were close enough to the laminar solution).

However, my suggestion would be to try a transitional model. SST has such an extension (look for Menter's paper on a transitional model for CFD applications). There are also other transitional models and similar to turbulence modeling you might want to try a few and pick the best for your application. These models are going to add a few extra equations to your simulation, so they will be slightly slower than a pure turbulence model.