User Manual for Credible Reliability Prediction

The ASQ Reliability Division (RD), copyrighted the 2003 monograph “Credible Reliability Prediction” (CRP) but lost all copies circa 2014. I pestered the RD to let me republish CRP, because people asked “How do I make credible reliability predictions?” Copyright reversion to authors is accepted practice when a publisher no longer supports a document.

The RD management decided they’d prefer to republish CRP and asked me for a copy, which I gave. Nothing happened, so after 6 months I pestered again. Dan Burrows, Chair of the RD asked ASQ to allow me to publish CRP. ASQ did not object. [CREDRP2020.PDF]

Harold Williams, editor of CRP, proposed substituting the word “reliability” for the word “quality” in these ASQ Strategy Themes circa 2014.

Support reliability professionals and practitioners in their efforts to grow in value in the workplace and community.
Prove and communicate the economic case for reliability.
Assure that a vital, growing body of knowledge is accessible to everyone.
Become the community of choice for reliability.
Grow the use and impact of reliability in every segment of the economy.
Make sure the world knows the importance and value of reliability.

“Credible reliability prediction” should be part of reliability strategy. To help with CRP, I wrote the “User Manual to CRP” [CRPUSM1.*]. Thanks Harold.

Reliability is One Minus a Probability Distribution Function

“Apparently, if you want to make a prediction about the future, you should base it on hard data, and that data should be completely divorced from any hunches or biases. You should deal in probabilities.” [“Superforecasting: The Art and Science of Prediction,” Philip E. Tetlock and Dan Gardner]

Here is a simple example of how to make a “Credible Reliability Prediction”. Predict a new product’s failure rate function a(t; new) as a comparable old-product’s failure rate function a(t; old) times MTBF(old)/MTBF(new), where the MTBFs are predictions. This is known as a proportional hazards model, justifiable because generations of products have proportional failure rate functions. A “Credible Reliability Prediction” is P[Life > t] = exp[–INTEGRAL[a(s; new)ds]], where the integral is from 0 to age t. The same prediction method applies to parts and to new product made up of parts. This uses all information available at the time a new product is designed. The CRP monograph also tells how to estimate old products’ failure rate functions; even if there’s no life data, using ships and returns counts, population data required by generally accepted accounting principles.

The CRPUSM1.DOCX version of the CRP User Manual contains active images of Excel spreadsheets described in CRP and some new spreadsheets. Double-click on the spreadsheet images to use the spreadsheets and see formulas. Unfortunately, copying the spreadsheet images and pasting them into Excel pastes values, not formulas. I will upload the workbooks and active spreadsheets described in this User Manual to https://sites.google.com/site/fieldreliability/ and put them in a “List of Files”. Or send your data and describe it. I will plug it in and send you the active spreadsheets and the results.

In addition to help with CRP, the User Manual contains a correction to CRP [sorry], recent developments, and chapters on what to do with credible reliability predictions:

MTBF prediction with storage followed by operation,
MTBF prediction with variable operating hours per calendar hour,
Spares required for a finite mission with repair,
Reliability management of failure rates,
MTBF and reliability prediction for redundant systems,
- without constant failure rates,
- with warm standby,
- with Cox’ proportional hazards model or with General Method of Data Handling (www.GMDH.net),
Reliability demonstration testing without unwarranted assumptions, and
Reliability centered maintenance and risk-based inspection.

“Credible Reliability Prediction” includes reality, and reality trumps incredible reliability predictions.

Table 1. Cross reference between CRP and CRPUSM chapters. [*.*] are in https://sites.google.com/site/fieldreliability/ “List of Files”

CRP	User Manual for CRP
1 The Objective of “Credible Reliability Prediction”	1 What is Credible Reliability Prediction?
2 MTBF and Reliability Predictions Are Part of Reliability	2 Credible MTBF Prediction [CRPExamp.xlsx]
3 State of the Art of MTBF Prediction	3 Field Reliability Estimation without Life Data [Npmle2.xlsm]
4 Extensions to MTBF Prediction	4 Reliability Prediction from comparison products {MH217F2.xls]
5 What Are Age-Specific Reliability Predictions For?	15 and 16 RCM classification, risk-based inspections, and actuarial forecasts
6 Age-Specific Reliability Prediction	4 and 6 for repairable systems or products
7 Credibility Is Possible	15 Reliability Demonstration Testing
8 Spreadsheets and Software	{reality.xlsx] and others
	7 Storage followed by Operation
	8 Variable operating hours per calendar hour [Convert.xlsx]
	9 Finite mission time and spares
	10 Reliability management?
	11 MTBF prediction and optimization for redundant systems
	12 Reliabilityprediction and optimization for redundant systems
	13 With warm standby [WrmStby.xlsx]
	14 PH and GMDH proportional hazards