[Anand Sis]

I've been trying to model the flow around a car with an aim to determine the aerodynamic lift and drag induced on it.
I've used realisable k epsilon turbulence model and pressure based coupled solver. I used 1st order upwind setting for my 1st 100 iterations(so that the values would stabilise) and 2nd order for the remaining. I had an experimental value of Cd=0.34 as my reference. I got a Cd of about 0.35 for my first analysis at 290kmph, which i though was fairly correct. I had to run the same simulation at lower speeds of 250kmph and 200kmph with all parameters unchanged.

What i have been observing is that its showing a consistent drop in Cd values at lower velocities(0.262 at 250kmph and 0.1677 at 200kmph ).I know this shouldn't be happening but i have tried changing the turbulence model from rke to SA but with no luck.I didn't check the other models as i've never worked with those. Any suggestions anyone ?

[taxalian]

Originally Posted by Anand Sis

I've been trying to model the flow around a car with an aim to determine the aerodynamic lift and drag induced on it.
I've used realisable k epsilon turbulence model and pressure based coupled solver. I used 1st order upwind setting for my 1st 100 iterations(so that the values would stabilise) and 2nd order for the remaining. I had an experimental value of Cd=0.34 as my reference. I got a Cd of about 0.35 for my first analysis at 290kmph, which i though was fairly correct. I had to run the same simulation at lower speeds of 250kmph and 200kmph with all parameters unchanged.

What i have been observing is that its showing a consistent drop in Cd values at lower velocities(0.262 at 250kmph and 0.1677 at 200kmph ).I know this shouldn't be happening but i have tried changing the turbulence model from rke to SA but with no luck.I didn't check the other models as i've never worked with those. Any suggestions anyone ?

Hi Anand,
I think before you try different turbulence models, you need to make sure that the computational mesh is fine enough to predict the drag fairly accurately.
Secondly you mentioned your free-stream velocities that lies within the range of Mach Number (Ma) = 0.16 - 0.2 i.e. incompressible flow.
It seems you are using Fluent solver for the simulation. I would solve this problem using incompressible pressure based solver with SA Turbulence model. But once again make sure that you performed grid independent check. Then another important thing you need to make sure that you should use yplus of less than 1 other wise vice versa if you prefer to use the wall function to resolve the boundary layer. The mesh should be fine enough in the wake region downstream of the car.

It would be nice if you can share some of the sketches of your mesh to suggest more better approaches.

Good Luck.

[Anand Sis]

Originally Posted by taxalian

Hi Anand,
I think before you try different turbulence models, you need to make sure that the computational mesh is fine enough to predict the drag fairly accurately.
Secondly you mentioned your free-stream velocities that lies within the range of Mach Number (Ma) = 0.16 - 0.2 i.e. incompressible flow.
It seems you are using Fluent solver for the simulation. I would solve this problem using incompressible pressure based solver with SA Turbulence model. But once again make sure that you performed grid independent check. Then another important thing you need to make sure that you should use yplus of less than 1 other wise vice versa if you prefer to use the wall function to resolve the boundary layer. The mesh should be fine enough in the wake region downstream of the car.

It would be nice if you can share some of the sketches of your mesh to suggest more better approaches.

Good Luck.

Thanks for the reply. I know i had to perform a mesh dependency check but i was trying to meet a deadline. So all i could do was refer similar works and do a mesh which worked for them(i made it still finer) and hope it'd work for me too.I know its kind of crude but didn't have an option .For the wall treatment i used non-equilibrium wall function.
And i think i found out the problem .The thing is, the forces obtained by the fluent solver happens to be correct but the drag and lift coefficient calculations proved to be wrong. I reverse calculated the Cd from the obtained drag force and am getting Cd of 0.35 through out the velocity range (with only slight change in the 3rd decimal place, which i can live with ). So am hoping my meshing and turbulence models were fairly correct. Do you have any explanation for this discrepancy ? Am puzzled coz i've specified the correct reference area