One time I was asked to do a durability test on an auto trans. I was not best pleased, so asked for the procedure, and grumpily went out to do it. (Durability tests are a pain in the arse). This particular one consisted of driving gently forward, put it into reverse, drive gently backwards, put it in to drive, drive gently forward. 50 times. It's an abuse test. How gently is gently when you don't want to do the test? Not very. Broke it.
Another fun fact is that if you are driving at 80 kph on a wet road or gravel, in a 4 speed Falcon if you floor the throttle and shift into reverse the wheels break traction and spin backwards,gently slowing the car. Then keep the throttle on and put it back in drive and they break traction the other way. I can't remember who showed me that one, but it was in heavy 3 lane traffic, they were driving down the road with rear wheels spinning continuously, throwing out steam and spray all over. Doesn't actually bang the car up. I should point out we used to have $1 cars that were no longer assets, basically we used to use them for transport and any old experiments we felt like (the rules have changed). One car was left instrumented for a year and every day the engineer went and repeated the exact same noise test. Absolutely useless information... until he demonstrated that our practice of adjusting the calibration of the sound level meters to a calibrated noise source every day was actually decreasing the accuracy of the measurements. On another car he built a duplicate of the ideas in a Lexus LS400 air intake, and ended up with a significantly quieter car for the cost of a few bits of plastic tube. Hardly any of our stuff got into production, but messing about with zero value assets at least made you think about techniques. I cut a big hole in the front of the the car and measured the velocity of the crankshaft pulley with a laser, that was pretty interesting (the pulley is a big, known, noise source, but you usually infer the vibration from the noise generated, this way we measured it directly). Sadly the new program schedules don't give development engineers any time to just play with the cars, not that we have $1 cars, or are allowed to drive experimental cars on trade plates, or any of that fun stuff. Somewhere on a program scheduling website there is a DAY by DAY diary of the decisions or activities that are supposed to happen on that day. Needless to say I sing a little song every time something comes out of left field and blows our schedules into tiny pieces. Last year i was working on one safety related job that would routinely took 2 weeks, but for strange reasons took 4 months. It was an absolute grind, but I learned a lot, and got put onto the special matlab license server for priority 24/7 access-usually there is a slight chance that a license will not be available as we share a limited pool. Still got that access, I suppose I should turn it back in. Nah.