3 ms·
Sorry if off-topic, but this community may be able to help with a long-term problem. I live in Manhattan across the street from a firehouse. Every day they tes
by spodek 6y ago
Sorry if off-topic, but this community may be able to help with a long-term problem.
I live in Manhattan across the street from a firehouse. Every day they test their devices to get through locked doors that are basically chain saws. Imagine someone testing a couple chain saws every day at a random time between 9am and 7pm.
I've spoken to them to find out why daily ("It would be a shame if we showed up to a fire and found out it didn't work then."), politely pointing out that after testing it, they don't know if it works since that test may have broken it. They don't know why daily either.
Anyone have any ideas on how to optimize the testing schedule for an engine? I'd love data relating testing frequency with safety.
- kqr 6y agoThis is where the manufacturing monitoring guys know a lot of stuff. Maybe statistical process control can help, maybe some other type of fault modeling is more appropriate. Whichever way, rest assured the manufacturing industry has studied this over Styx and back. Look up what they found out!