I'm running a race car for which there are several wing elements and configurations available(single element, dual element, slotted flap, various wing sections, etc)

As an engineer I'm interested in maximizing lift (downforce) and reducing drag. After spending time adjusting mesh sizes, turbulence parameters etc on a 2D airfoil model, I find out pretty quickly that CFD is not very good predicting lift and drag. Lift numbers are kind of in the ball park, but I don't know if they are useful or not. Drag numbers are completely out to lunch.

In terms of making a race car go faster, these are things that would be useful to able to do:

1) Predict in absolute terms lift and drag of different configurations. These numbers can also be used in simulations of car performance.

2) Predict lift and drag of different configurations for comparison in relative terms in order to select the optimum combination- in this case I don't care what the actual forces are, I just want to know which configuration is better

3) Optimize the section and geometry of a single configuration- for example can varying the slot gap geometry or the relative positions of the main element and the flap improve performance?

Since it's already been established that 1) is not feasible, what about 2) and 3)? If CFD can't be used to predict lift and drag forces, for what IS it suited other than making pretty pictures of dubious results? Pressure distributions? Flow viz? Anything? A 2% reduction in lap time is significant. In what manner can CFD be applied to improve the performance of a race car? Or am I wasting my time?

Any insight is most appreciated.

Hi Mr Chu,

I just posted one message on the main forum half an hour ago about my frustation on 2D multi elelement airfoil type analysis. My Cl is 50% off and my Cd is 300 off.

Since I am a rookie, I don't have much to add but indeed what i know is CFD accuracy on external flow problem using PC is not really good. If you look at the NASA drag workshop every year, you can find that the results submitted from various CFD company, they all use clusters of network PC to solve the problem. If you only have one PC or a couple more, I just don't think you will ever get result accurate to within "a few percents". Well this is just my thought. Everyone out there please correct me if I were wrong on this. I too wish CFD is much more accurate.