If you assume an exponential failure model, then the number of units that are still good = the reliability R, would be
R = exp[-lambda t] = exp[-t/MTTF]
Take the ln of both sides and do a little algebra
ln(R) = - t/MTTF
MTTF = -t/ln(R)
So of you want <= 1% failure each year, you need R >= 0.99, so
MTTF = -2080 hr/ln(0.99) = 207000 hr
- If you wanted 1% failure over the entire 12 yr (not 1% each year), then MTTF = -(2080*12)/ln(0.99)
- The the failures do not follow an exponential model, then you would need to analyze it for the specific model.
Tim