I am working on a project where a machine has a part replaced every 500 hours it's been run or it fails (which ever comes first). I have been asked to determine if we should lower the replacement time to 100 hours to make the machine have better reliability. I have a lot of data of failures of the current situation but I don't have data on what happens if they replace it every 100 hours. My initial thought and research leads me to creating my own simulation with bootstrapping however I don't know if this is statistically valid. Experts - how would you determine if they should make the change? What statistical methods would you use?
A change in process is proposed. How do I determine if the change will be good without any data?
- Thread starter jackbio
- Start date
- Tags bootstrap simulation process change reliability analysis
Then you have the survival time when the time is less than 500 h. And for more that 500 h it will be censored. If you can find a distribution that fits the data you can estimate it with maximum likelihood. Then you can get an estimate of probabiity of less than 100 h. There are lots of programs and theory for survival analysis. Search for it!