[TheVictorVS]

Hello everyone

I would like to consult with the forum about a problem:

I want to 2D model a confined flow of air inside a structure, exactly a cooling tower of 2 meters high, well the difficulty comes when it comes to specifying the flow in some areas along the tower that have small elements very close between yes.

The air flow is propelled by a fan on top, the air flow of the fan is directed from the inside to the outside of the tower, the incoming flow is sucked from the base of the tower, attached a figure on this.

According to what I read about turbulence models, the most robust are the models of the k-epsilon family, within these the k-w model can handle flows with high disorder and complex geometry.

But what I am trying to do is obtain the most realistic approximation of the flow within this geometry, if possible with the flow resolved to the walls but maintaining an approximation to reality in the areas far from the wall.

I have been using the software Comsol Multiphysics 5.3 in license evaluation for this purpose, the software has implemented this model k-w but I would like to hear your opinions on how to do this as best as possible, or recommendations of other models of turbulence or software for that purpose.

I attach details about the aforementioned geometry and the results I have obtained from the aforementioned software.

Speed field captures produced by the model k-epsilon and k-w are also attached

I hope I could express myself with the necessary precision, I am new to the forum

Atte,
Victor

[Ravindra Shende]

Different results with different turbulence models is expected for such complex geometry problems.

The best that you can do is, compare the results of the two models with experimental measurements and conclude which model is better suited for the problem.

If you are working in academia, I would suggest to get a result using LES (if comsol has it). But if you are working in industry, you have to make decision on doing LES based on the time and money you have to spend.

[TheVictorVS]

I appreciate the focus of your response, however I do not use LES because I do not have the resources to use it, my consultation with the forum is if there is another model apart from K - w, suggestible for my model, in your experience you can recommend me some information additional on this, or the difference between the wall functions and the resolution of the flow to the same wall. I must add that this work is my thesis project, so I appreciated support material for the theoretical foundation.

Atte,
Victor