How to predictmodel htol with 50%25 failures

[cwong668]

is it possible to predict/model the hours required in the htol test to have 50% failures?

if yes, please provide details.

is "mean time to failure" the same as "median time to failure"?

thanks,
cwong668

[Pantelis]

Yes – it involves use of quantitative accelerated life testing principles. There is an entire textbook on weibull.com devoted to this subject. See http://www.weibull.com/acceltestwebcontents.htm.


Mean and median are not the same (unless one assumes normality). Equivalently mean time to failure and median time to failure are not the same. For more details see http://www.weibull.com/LifeDataWeb/basic_statistical_definitions.htm

[Annandhan Mayappan]

Yes - you can use log graph to extrapolate your data and then reach 50% failure point. The time taken to reach 50% would be your prediction. The danger is, you have to make sure no other effect to your device as time increases. If yes, then your extrapolation would be wrong