High Re K Epsilon vs Low Re K Epsilon
 

Hi folks,

I am somewhat new to CFD (only been doing for some months :o), and I have a doubt which I hope some can clear...

I am trying to implement a Low Re K-Epsilon turbulence model to a backstep and compare it to a High Re K-Epsilon turbulence model, to show that, in this particular case, the Low Re K-Epsilon turbulence model represents the situation better.

However, I get a pressure drop that is higher for the Low Re K-Epsilon turbulence model than the pressure drop for the High Re K-Epsilon turbulence model...

Is this correct? shouldn't it be the opposite? :confused:
Or perhaps I made a mistake in the modelling step?

Hope someone can help!

Best Regards,

Rui Silva

Anyone?

I'm really stuck in this matter...

Hi Rui,

In this specific case, you will need to ensure that the mesh is appropriate to the model employed (and ensure that the wall treatment is correct).

Low-Reynolds k-epsilon turbulence model:
- The first cell y+ should be around 1 (or less);
- the wall treatment is based on first cell being the viscous sub-layer.

High-Reynolds k-epsilon turbulence model:
- the first cell y+ should be around 15-30;
- A high-Reynolds wall treatment is to be used.

From your post, I am unsure that you are using two different mesh that are consistent with the employed model.

Kind regards,
Julien

Hello Julien,

thank you for your reply.

What I am doing is modifying the standard high Reynolds k-epsilon (according to Wilcox) that comes with a cfd software (comsol) so that it becomes the
Launder-Sharma Low Re k-epsilon Turbulence model.

So I modified the equations to incorporate the new set of damping functions and boundary conditions near the wall.

After reading some posts and papers I have refined my mesh.
But since I am using a modified standard k-epsilon model, I cannot see y+!

I have attached pics of my mesh near the wall and some size info!
Hope the pics help...

Is there a way to evaluate y+ for a low re k-epsilon model??

Best Regards,

Rui

Have a look at: http://www.cfd-online.com/Wiki/Dimen..._wall_distance for the definition of y+ - which can be evaluated for a low Reynold k-epsilon model.

There is also a simple y+ estimator on the web page.

Also, you could use the other calculations to estimate the friction velocity (under the assumption that the other calculations friction velocity is accurate).

The rule y+ around 1 is a guidelines - do not get stuck with it.