Hacker News new | ask | show | jobs
by seb1204 541 days ago
Would be good to elaborate and not let us hang out in the salt. ... Please
2 comments

To elaborate, the failure rate does not change with time, but with some other stressor like temperature. The failure distribution may be modeled as a random process, instead of a time-based one. An exponential failure distribution is an example of a time-invariant process, while a Weibull or Gamma distribution would be time-variant, because the failure rate changes with the age of the item.
I misinterpreted the equations of those models based on a paper that stated some corrosion models use a random, rather than time variant process.