8 replies to this topic

[Steve McGahey]

I often hear questions about the interpretation of OREDA data for RAM (Reliability, Availability, Maintenance) Studies, because there are no clear guidelines on what should be considered.

Well, I've long had my own opinions on the answers to these questions; answers formed from some of the training courses I've been on (where people sometimes have different opinions) and also from my experience and specific questions I've asked while working on projects.

I'm going to summarise a few of these here, in the hope of helping people with similar questions, and/or also in the hope of prompting some discussion on the subject.

For now, I'm just going to say that all of the failure mode distributions used in OREDA are exponential, and that the repair times are based on a smattering of data with no real distribution... so it can be argued that repair times should be modelled as constant, or rectangular distributions.

More to come, including a blurred-out (to protect OREDA's copyright) labelled picture of a typical OREDA data page, and some notes on the different elements on that page.

- Steve

[gvdlans]

Steve,

Sounds interesting to me. I am looking forward to your posts!

[Steve McGahey]

More to come, including a blurred-out (to protect OREDA's copyright) labelled picture of a typical OREDA data page, and some notes on the different elements on that page.

- Steve

Well, here are the attachments. That's all, for now. Might come back if there are any questions.

 Example_OREDA_page.png   122.67KB   142 downloads

 Explanation_of_OREDA_database.doc   28.5KB   171 downloads

[gvdlans]

Steve,

The picture of the OREDA page is quite vague. Is it possible to put a higher quality picture in the Word document?

[Steve McGahey]

Steve,

The picture of the OREDA page is quite vague. Is it possible to put a higher quality picture in the Word document?

Unfortunately, no - I have made the picture vague by design, not by accident. I am worried about being accused of publishing parts of an expensive database, which is why I've blurred the numbers.

If you can suggest a way around this problem, then I'm happy to help.

[Steve McGahey]

I've seen people asking how to convert OREDA data into a Weibull distribution (Wiki: Weibull Distrbutions).

Well, the OREDA data follows an exponential distribution, which is a Weibull distribution with a shape factor (k) of 1.

So, assume that the shape factor is 1.
The characteristic life remains the same.

There's your answer.

[gvdlans]

Ah OK, now I understand why the red numbers were sharp while the black numbers were vague.

Maybe you can make your own table of a hypothetical piece of equipment?

[Steve McGahey]

Ah OK, now I understand why the red numbers were sharp while the black numbers were vague.

Maybe you can make your own table of a hypothetical piece of equipment?

Are you really keen to see this? I'll do it if you are, but not otherwise.

[gvdlans]

No, it is OK. Thank you for the effort!