Weibull applicability

[Brian Lewandowski]

If a product that is installed outside fails during the winter months when it gets cold, is there any value in performing a Weibull analysis; is it relevant? Also, since the failures are periodic, what time interval should be considered if a Weibull analysis is relevant: “all” time, or just the time when failures occur during the winter months (4 months every year).

[Student]

Hello,
Im a uni student current doing a thesis on forecasting technology. I'd like to know how does the weibull distribution give rise to the s-curve equation and where can i get more info on such this topic. If anyone could help me with this i'd be most appreciated. Thanks.

[tarik]

I will suggest 2 approaches (my preference is the second one):
- If you want to consider a year (with the variations in temperature that occur during a year) as your time unit, you can do a straightforward Weibull analysis (using the Weibull++ software).
- If you are interested in having a more accurate modeling of the variations in temperature and analyzing their effect on the life of your product, I would suggest using the time varying stress model, in which you can construct a profile of the temperature change throughout the year and then model the data (times to failure ) you’ve collected to obtain a model that describes your product’s failure behavior under those varying temperature levels and that can be used to derive the metrics you are interested in. For more info about the time varying stress model, review the following material (this analysis can be done in the ALTA software).

http://www.weibull.com/AccelTestWeb/model_formulation.htm
http://www.weibull.com/AccelTestWeb/time-varying_stress_models_chap_.htm

[tarik]

For answer to Student question, please review discussion in post titled ‘Need Help’