Learning Forum

[vikoviko3009]

Hi you all :)

I'm currently simulating a mixing process (3d) in an injector, something similar to a pinlet injector. My geometry is a section of a hollow cylinder and then the injector in the middle of it, but trouble with the boundary layer and I can not get a value of y+ less than one. I'm using the SA turbulence model, btw. 

I have tried the following things: 

I tried putting inflation layers and I got y+ values around 1, but the mesh quality is awful. All the mesh metrics are very bad, e.g. element sizing is between e-03 and 1. The convenience is also bad, and the results do not seem reasonable.

Another approach was to use local sizing, and the best I could get was a value around 45. But the number of cells is really high and if I decrease the local sizing further, it exceeds 1 million cells. Moreover, after a certain number, I get a message: "reversed flow" and the results are not okay. 

I have heard of wall functions, and I've read some previous threads here on this forum, but I didn't really get them. 

Could you pleaseeeeeeeeeeee suggest any solution?

Thank you in advance!

[Rob]

I'd start with some reading, not least to understand why you may want a different turbulence model. The meshing tutorials may also be of some use. Wall functions may well be covered in the Turbulence lecture in Learning (left tool bar on here). 

Depending on the size of the nozzle, and wider domain you may easily find the mesh exceeds 1M cells. That's fairly normal, but does conflict with the Student cell count limits.