Reliability Growth Analysis

MarkovVaribles1

Hello,

In the training video https://community.jmp.com/t5/Mastering-JMP/Analyzing-Reliability-for-Repairable-Systems/ta-p/483465 when the analysis is performed, no information about the duration of maintenance work is provided. For example a single fix may have been done on day 3 and the next one on day 10. However this fix may have taken 2, hours, 10 hours, 2 days and etc. My understanding is that the time stamp (in his example days) refer to the hours of operation (service hours). If my understanding is true, then this analysis may be misleading. Am I missing something here?

Also, can one use reliability growth platform for corrective maintenance scheduling/forecasting?

 

Thanks,

TurbineEngineDesign1.jmp

peng_liu

For the first part, you concern is a good point. The software can be used in a way to address your concern.

The underlying model builds a model based on time to event, and assumes the event duration is zero, or at least negligible while calculating time to the next event.

If your data has the situation that you described, you need to understand your options. The platform supports two formats: time-to-event format, and date format.

If you carefully calculate operation time, and remove maintenance duration, then use time-to-event format to analyze.

If you ignore maintenance duration, and use date timestamps directly to avoid detailed data cleaning work. That should be fine if maintenance duration is relatively short and infrequent during operation. Otherwise, you should use the previous approach.

 

For the second part. I don't understand the question. But the model itself does not make assumption on the type of maintenance. If the maintenance type is corrective (i.e. replace with new), then it is likely the beta of the fitted model is close to one.