Difference in turbulence model
 

Hi Guys,

I am solving for flow analysis in waterjacket in IC engine. I am studying, "effect of different turbulence model on flow physics in waterjacket".
I have simulated this with Std ke, RNG k-epsilon and EARSM k-epsilon turbulence model. Results with the EARSM model is close to the baseline results.
My question is whether is it recommended to use EARSM over std ke? what are the advantages and disadvantages of EARSM over other two turbulence model.
Why this model is not so popular if it accounts for the secondary flow?

Thanks in advance !

I generally recommend the SST turbulence model as the default model unless you have a good reason to do otherwise. So no, I would not generally recommend EARSM over SST.

Advantages:
* Can capture secondary flows
* Can capture the effect of streamline curvature

Disadvantages:
* Often is MUCH harder to converge
* Additional variables and simulation time and memory requirements
* For flows which do not have secondary flows or streamline curvature it adds no additional benefit
* It introduces new empirical constants for tuning which in some flows makes it less accurate than the simpler SST model.

It has many problems - as you can see above. So it is not recommended unless you have a flow which requires the secondary flow or curvature abilities it possesses.

Thanks Glenn for information.

Well, the problem with SST is that the mesh count is more than double compared to k-epsilon. Also, simulation time is much higher for SST. Results of k-epsillon are in good agreement with SST values, infact the results of pressure drop and mass flow of EARSM model are more close to SST results compared to std ke and RNG ke.

You said in disadvantages of EARSM, " in some flows it is less accurate"
Can you put light on which flow you were talking?

Umm, what? You can run SST on any mesh you like. And if you are saying you require y+=1 then you are mistaken, SST works fine with wall functions for y+>11 just like k-eps.

I cannot quote exactly which flow EARSM is less accurate. I have come across several examples over the years. I would expect it to have difficulties in things like boundary layer growth, bluff body flows and things like that. k-eps and SST are more suitable in those situations for most applications.