System Analysis with Left Censored Data

[A. Rowland]

I have a system that was originally placed in service in late 70's, but computerized maintenance records are not available until June 1991. Also, the records are not typically detailed enough to bin by failure mode.

I have considered cumulative time to repair request with t0 = June 1, 1991. The other approach I have considered is time between repair requests, since these times would be independent of actual system age. This would, then, allow me to determine the recurrence rate. I could use this recurrence rate in subsequent modeling.

Is there an optimal methodology for handling system repair data such as mine?

Thanks.

[Adam]

I assume that you mean cumulative time to failure with t0=june 1, 1991. I also assume that you will probably fit a Power Law (NHPP) model to this data. In my opinion this will be the best approach, since the power law assumes that the system is As Bad As old after the repair, where the time-between-failures approach assumes that the system is As Good As New after the repair.

[RS Support]

You may want to take a look at ReliaSoft's new RGA tool for this type of data/analysis.

http://rga.reliasoft.com/

[RS Support]

You may want to take a look at ReliaSoft's new RGA tool for this type of data/analysis.

http://rga.reliasoft.com/