In my experience with car physics I've done some "advanced trial and error" like dividing the car's speed (the difference between its position in two instants) by the wheel speed (the angular ammount of movement it accomplishes in in the same time) to calculate wheel diameter. In possession of the diameter you can calculate car speed from wheel speed (and vice-versa), increasing the ammount of control over the car (I have, for instance, written code to compare the 'known' linear/angular ratio with the real-time ratio to detect skidding and control traction in a 4x4 vehicle).