I'm trying to generate a "survival" vs "age" curve for some component data similar to Exhibit C-10 in Appendix C of Nowlan & Heap's RCM report.

Rather than just throwing the data into some fancy software (that I don't have anyway), I'm trying to follow the logic explained in the report and do it in MS Excel.

Before I started putting my own numbers in, I attempted to replicate their example so I had some confidence in the steps to get the results.

Unfortunately (for me) I hit a wall in the steps going from Exhibit C-9 (which I can replicate) to Exhibit C-10 (which I cannot replicate). These are covered on page 406 of the appendix.

I could attach my Excel file but it is a real mess and will take a bit of time to clean up. Could anyone direct me to another reference that explains the steps in more detail, or, if you have completed a similar analysis and are willing to help, let me know and I will contact you off the board?

Regards, Matt.

Matt,

I think the spreadsheet that I attach herewith with an example of a fictitious accelerated life test can be of help. I use it in my classes to demonstrate to engineering students how to manipulate the resulting data in order to build the reliability functions R(t), f(t), F(t), h(t) and mean life until each interval. Columns J and K return the same values of the function h(t) although calculated differently (column K according to Nowlan and Heap RCM report).

Note the curve R(t) which you are particularly interested in.

Regards,

Life_test.xls (57 Kb, 33 downloads)

In the above example 300 electric switches are tested until all have failed. A great number fail at an early stage (infant mortality).

Thanks

Rui,
Many thanks for you help, I will digest your reponse over the weekend.

Matt.

Matt,
In Appendix 3-1, pages 47-53, I have recreated the N&H failure patterns using dummy data, in my book, details below.
Unfortunately I am on the road quite a bit, so I am not able to respond adequately to this or other posts, my apologies.

Hi Matt,

I recall something about the different exhibits and trying to put all of the data in one place. I think that what I ultimately did in the spreadsheet that I am uploading here.

I did this a while back that I use in training. Let me know if this fits the bill. Don't pay too much attention to the graphs they are used for specific reasons durin ghte training run I do with them.

In the meantime I will have a look through that appendix again this evening and get back to you on that and this spreadsheet.

I hope all is well where you are,

actuarial_analysis.xls (38 Kb, 20 downloads) Nolan and heap recreation

Matt,

Note that in the example above (in the attached spreadsheet) the function h(t) is the famous "bathtub curve", whose typical shape can be seen in the graph. In this graph, the presence of early failures, random failures and wear out failures as time elapses is well evident. The increase of the number of tested units would cause a smoothing effect on the h(t) curve.

Regards,

This message has been edited. Last edited by: <Rui Assis>,

Dear Rui,

Please have a look through this document also. It explains the figures in the preceeding spreadsheet.

Also, it is important to bear in mind that this is an analysis of a PRatt & Whitney 7T8D-7 engine on a Boeing 737. It is not a graph of a single failure mode within a population.

Actuarial_analysis.pdf (1,417 Kb, 19 downloads)

Gentlemen,
Many thanks for your help. In reviewing your sheets I discovered my confusion was simply in the terminology.

It appears the survival curve can be plotted using the data in Exhibit C-7 on page 401 using columns (1) "age interval" and (5) "cumulative probability".

I was under the impression that further intermediate steps were necessary but once I viewed the sheets from Rui and Daryl the light bulb flashed on above my head in a brief moment of clarity.

My sheet is attached showing my version of the charts in Exhibit C-9 (Age-Reliability) and C-10 (Age-Probability of Survival).

Thanks again,
Matt.

Actuarial_Analysis.xls (53 Kb, 15 downloads) Actuarial Analysis

Thanks Daryl for the part of the document from Nowlan and Heap.

Regards and good work (must be hot down there...)