View Single Post
      06-07-2013, 10:57 AM   #102
Lieutenant Colonel
drob23's Avatar

Drives: S4
Join Date: Oct 2012
Location: Michigan

iTrader: (1)

Originally Posted by swamp2 View Post
CarTest is a physics based simulator, certainly many effects are missing from the enormous range of subtleties in any real world car, however, the dominant effects that allow accurate absolute predictions and even better relative predictions of all common vehicle performance metrics are all present. That includes but is not limited to power, torque (full rpm dependent curves for both), weight, weight distribution, weight transfer, gearing, parasitic losses (drive train, tires and others), traction, friction and wheel spin, tire size, wheel growth with speed, environmental effects such as pressure and air density, road slope, shift times and last but not least aerodynamics including drag coefficient and frontal area.

Now that being said, I think you misunderstood what I mean by "eliminating variables". What that means is that if you compare car A and car B you can change one and only one parameter between the cars and ALL other inputs (variables if you like) are held completely fixed and identical. It is this sort of control and parameter isolation that is not at all possible in real world test, no matter how much care is taken.
Seems like the biggest source of uncertainty in a 0-60 or 1/4 mile sprint would be the tire model. But maybe you can correlate your model and known benchmarks for the M5?

Also, I'd guess suspension modeling plays a big part with the weight transfer, how is the suspension modeled? I know there are 2 types in the industry, one such as Carsim which uses parametric models and 2nd order approximations (look up tables), while others actually model the linkages and multi-body stuff (is it called ATOM model?). I'm no expert on this as you can probably tell...