On Sep 23, 2005, at 5:26 AM, graywolf wrote:
MTBF (Mean Time Before Failure) usually expressed in hours or cycles is the design life of equipment. It is an average, some may fail immediately some may go 10 times that long, but it does give some idea of how long something is intended to keep on working.
The MTBF calculation assumes that all failures occur randomly and that the failure rate is constant over any period of time, which means that it doesn't specifically account for out-of-the-box failures or wear-out failures as they aren't truly random (they're generally caused by manufacturing defects and wear&tear respectively).
The random failures are really a combination of how normal usage affects the components and materials within the context of that particular design, within its intended lifespan. There are *lots* of things that can cause "random" failure - I've even seen papers published about failures induced by cosmic rays at ground-level.
MTBF applies statistically over a population rather than individual units, so a product with 10,000,000 hours (or cycles, or whatever) MTBF means that if you sell 10,000 units then you will expect, on average, to observe one failure every 1,000 hours. It doesn't mean that each product will last for an average of 10,000,000 hours of use.
Oh no, now I'm having flashbacks of manufacturing process statistics and 6-sigma...
Cheers, - Dave (time to go - Futurama is on in a few minutes)
