I love this stuff. As a mechanical engineering degree holder, I find it worthwile to examine any problems like those discussed and hope the library comment didn't offend. I was just trying to bring light to such a topic brought forth by such terrible outcomes. But, this needs to be done, and no participant needs to be silent as any ideas are food for thought.
Anyway, keep in mind folks, that the car is acting in its own coordinate reference and the driver's reference coordinate is the car around him/her. Whatever the car is going, the driver is not. The resultant forces delivered to the driver are products of the acceleration to the finish, the braking decceleration opposite the first after finish and the driver's momentum as a seperate body still acting in line of the intial acceleration. I'd think i'd start with the max force that the body can safely take and add a factor of safety (say 10%) and aim to find out what it takes to stop the car (i.e. distance, friction surface, chute area, other means) within that max.