let me describe one typical use case:
8:00am: drive to work (let's say 15 miles)
11:30am: drive to lunch (2 miles)
12:30pm: drive back to work (2 miles)
3:30pm: go do errand, return to work (3 miles round-trip, 10 minute to do errand, one stop)
4:30pm: drive from work to school to pick up kids (10 miles, 30 minute wait idling)
5:00pm: drive from school to grocery store (5 miles)
5:45pm: drive from grocery school to post office (0.5mi)
6:00pm: drive from post office to home (20 miles)
now let's count starts and stops:
8:00am: turbine starts, goes for 15-30 minutes, stops
11:30am: turbine starts, is on for five minutes, stops
12:30pm: turbine starts, is on for five minutes, stops
3:30pm: turbine starts and is on for five minutes; does it stop for the 10 minutes the driver is inside doing the errand? choices to make here! this is either one start/stop or two, depending on how you answer that question
4:30pm: turbine starts and is on for, what, 20 minutes? then it's on for another 30 minutes while you idle (doing nothing)
5pm: if the turbine never stopped, it needs to start again, runs for 10 minutes, then stops
5:45pm: turbine starts, goes for 1 minute, stops
6:00pm: turbine starts, goes for 30 minutes, stops
so, for this utterly representative usage pattern for ***one day*** of automobile usage, with a pure-turbine design, we have seven turbine cycles ***minimum***, and possibly as many as ten
we also have lots of "dead time" during which it's not doing anything useful, simply burning gas (possibly as much as an hour, depending on how bad your choices are)
we haven't even gotten into "oops I turned the car on and forgot something in the store so I have to go back in"
this gets stupid very quickly, especially when you're dealing with something that ideally ***should*** only be turned on one single time in a day, at most