[jhs]

I am simulating a Formula Student race car in Star CCM+.

Steady, Coupled Solver
12m/s inlet
Symmetry in car centre line

The convergence is quite low.
Of course, the problem is highly turbulent and a steady approach is not the optimal way (but is chosen because of time and hardware resources).



Using adaptive Meshrefinement and other meshing strategies shows a higher number of cells is needed for better convergence (grid convergence is not reached). 40M cells are about as high as I can go.


In literature generally calls for the wind tunnel size with 2-3 car length in front and a least 5 to the rear of the car.

I get why I need space in the front (to not disturb the inlet boundary condition) but not really why I need so much space behind the car.

The wake requires a lot of cells to get the global convergence down. To limit the number of cells and get them where detailed flow interests me the most (around the car) one idea would be to get the outlet closer to the car and lose the cells in the turbulent wake.

To the actual question:

How would that influence the solution close the car (lift, drag, ...) as I am not really interested in the flow 10m behind the car?

The literature I read never gives a reason for these dimensions, so maybe someone here has a good answer.

Thanks

[FMDenaro]

because numerical perturbation from the outlet can go back in the interior affecting the solution... Any numerical outlet is somehow defined by an approximate set of BCs therefore such a disturbance can disrupt the quality of the solution.

Have a look to the theory of characteristics for subsonic unsteady flows.