I worry a lot when I hear about new technologies (especially computer technologies) being added to cars. The main problem I have is that computer parts can fail fairly suddenly.
I think my family's old ’84 Cavalier had computer parts fail a few times. The last time it happened, the failure caused the fuel pump to shut down. I could only drive as long as there was gas in the fuel line. Well, the car had been hesitating a bit as I drove it earlier that day, so maybe the failure could have been anticipated. Fortunately, I was only a few miles from home when the car broke down.
I had a similar problem with the ’88 Cavalier I have now. I'm not sure if the computer ever outright failed (the fuel pump always worked, at least), but the car would hesitate intermittently. Apparently it was no longer doing proper fuel mix calculations. At least this failure was accompanied by a “Service Engine Soon” idiot light on the dashboard. I could drive the car, but it behaved funny, and in theory the engine could have been damaged if I continued to drive it in that condition.
Back to the ‘84 for a moment: I once had the car overheat while I was driving in light city traffic. I tried to mitigate the problem by turning on the heat, and I probably started driving slowly as well. However, my mother (who has an Electrical Engineering degree) looked at the car's repair manual and discovered that turning on the air conditioning would have temporarily fixed the problem. The thermostat that failed controls the fan behind the radiator, but turning on the A/C makes the fan run continuously. Anyway, the problem went away once I got on the highway, since there was actual airflow through the radiator..
In my opinion, the scenario with the ’88 turned out the best. There was a failure—but the car fell back to a simpler mode of operation, informed the driver that something was wrong, and kept the vehicle drivable.
I'm not impressed with how the other parts failed. Of course, it's possible that the first problem I mentioned could not have been worked around. Also, the last scenario with the failed thermostat might have turned out the best way—the car could have been set up to keep the fan on if the thermostat failed, but then you'd have to take it into a service station to have them communicate with the computer to determine the problem... That's way more expensive than just going down to the parts store to pick up a $10 thermostat.
Well, I guess there's a whole other discussion of open formats and protocols for communicating with onboard computers in cars that I could get into here, but I'll hold back on that for now.
The main point I'm trying to get to is that, when designing a system that needs to be robust (like a car) you need to think about how to react when parts of that system break. And things will break on cars—they're hot and cold, dry and wet. They produce and accept lots of shock and vibration, and are just dowright dirty.
Now, new cars are getting all sorts of sensors and doodads. Suspensions and brake systems are computer-controlled (or at least computer-influenced). Some steering wheels aren't even connected to anything. Some cars are bristling with miniature radar systems for parking assistance, collision avoidance, and even controlling airbag inflation. Continuously variable transmissions are neat new toys too.
However, what happens if nothing is in control of these things anymore?
I'm sure many of the engineers who have worked on these projects have thought this out, but there are probably more than a few products that can fail in very strange ways.
Posted by mike at June 19, 2003 12:18 PM | Car , Hardware | TrackBack