Chosing turbulence model
 

Hi all. I have been working on this project for about 6 months now and since i am new at using Fluent i want somebody to confirm that i am using the right model. I am interested in air flow over a car, trying to fvisualize the stream lines and get preasure drop. It is a school project. I am using 3ddp solver, rolling tires, sliding plance etc. I also have the case solved, but now we are adding some additional things to see the effects and this is where the problem comes from. It gives me the warning about TVR limited to 1e^5 in a large number of cells. I have got this problem before but it got fixed in mesh stage. So far I have been using RNS linear, with non equilibrium wall functions. I am mostly worried about using the wrong model. What would you use in this case?

hi,

if its something simple like a 2D simulation of streamlines over a car, just use a standard RANS k-epsilon model...what additional things did you add exactly? could you be a bit more specific?

Ak

Yes, I mentioned it is a 3d simulation. I haven't changed any of the default settings on the turbulence model, other than picking non-equilibrium conditions, on the near wall treatment setting. But what i want to know is if you had to do this what model would you chose. The case is supposed to be as close to real-life as possible. You have a formula car, and want to find out how much air flow is going through radiator, also down force etc. So you want to run a CFD simulation on the whole vehicle at velocity of say 30 m/s.

hi,

the model I would choose would depend on the Reynolds number of my system, in your case you are solving an industrial problem, which means high reynolds numbers...so RANS is definitely the model you should you use, you might get away with LES, but if you arent sure I would strongly recommend RANS. as far as your other problem, I cant help you unless you are more specific.

AK

thanks for the reply. I cannot tell you much about the other problem..all i know is that it says turbulent viscosity ratio limited to 1e5 in.. that many cells. their number goes up and then goes back to zero but the plot still doesn't look like it's converging. i plotted the ratio and the highest value on the scale is about 7.7e4, which is still lower than e5, and i don't understand why it is giving me the warning. I also tried refining the mesh, but still nothing.

hello again

It may be that the the TVR for your particular problem is infact higher than 1e5, this is possible for very turbulent problems, if that is the case all you have to do is raise the allowable limits, FLUENT has an upper limit for all solved variables (I think), just go to solve, solution, limits, and change the TVR limit to say to 1e6, and then see if your solution converges. I dont have experience with high reynolds number flows because I mainly do DNS, but you should try and validate that this high TVR is expected, which for your problem may well be because air has a low molecular viscosity. If you are solving an unsteady problem, try decreasing your timestep until your solution stops changing, and make sure you dont use a finer mesh then you have to, bad aspect ratios in certain areas can lead to ill calculated gradients which will affect the physics of your simulation.

hope that helps
ak