This thread has been a very interesting read. I will fuel the fire by adding my 2 cents
.
Professionally coming from an experimental test background, I would always tease the analytical guys by saying that their models were only good at predicting anything only after they were calibrated with real world data
.
However Swamp is right in saying that there are a lot of variables to control in the real world. Experimental data needs to be statistically averaged to have any significant meaning. In other words, you cannot rely on a single run/test to prove or invalidate a theory.
Good calibrated models are extremely useful in developing almost any product. They have significantly reduced the cost and lead-time to develop products by reducing the required quantity of testing and trial and error aspect of it.
I also partially agree with Swamp that relatively small changes in gearing do not have significant impact on the total spectrum of the performance envelope. However, if one wants to improve performance in a specific range, gearing can have an important impact. Any racing team can attest to that, where they will pick optimal gearing for every single track. Most often, simulation models are used to figure out that optimal gearing (analytical models are useful
)
If gear ratios would have no impact on performance, why would we even bother having multiple ratio gearboxes, we could simply go with a direct drive from engine to wheel and reduce drivetrain loss
.