Multiple failures of the same equipment

[zahid]

I'm currently trying to compile reliability database for pumps for my company's refinery using Weibull ++ to come out with the weibull parameter. It is hoped that from the analysis, we can tell whether the pumps are getting better or worst in terms of reliability.

There are 100 pumps and the observed failure period is 1 year. My definition of failure is inability to deliver fluid at the required head and flow. When pumps failed, they will be fixed and return back to operation.

As expected, there are more than 300 failure data from the 100 pumps (some pumps failed more then the rest, some didn't fail at all). The question is, is the method of analysing those life data is the same with equipment that only failed once in its lifetime which is common in manufacturing industry (i.e once failed it will be discarded)? I’m thinking of inputting those 300 failures as separate data.

[Pantelis]

Well it depends on what you are assuming. If you assume that after repair its as good as new then you can go on with all the data, if not – then that would not be appropriate. (If you assume as-good-as-old then other models/methods -RGA- may be more appropriate.)

Do you know the failure mode (component responsible) of each failure? If so analysis could be done that way using Weibull methods.

[usara_reliable]

I agree with Pantelis,

In addition i couldn't get u why u r considering failure data as separate form. If the100 Pump of same type have different failure mode then you can separate in that form for getting paramete in individual case.

Just find whether pump impeller fail,suction chock etc for your problem case. Now differentiate the data with failure mode and find the root cause and corrective action for your problem. Reliability growth could not be possible without knowing failure mode.


Hope this help u