Would you believe you don't need life data?

Publications  Work Play
Please choose

Reliability Decision-Making Flowchart

This a pseudo-program that implements a flowchart...
  1. Do I want to know my products' and parts' age-specific field reliability?
    1. No thanks, ignorance is bliss
    2. Yes, field reliability is fundamental to our business and to our suppliers' and our customers' businesses.
  2. Do I have to track products and parts by name and serial number to get their ages-at-failures and survivors' age?
    1. Yes. 
      1. The FAA requires tracking of ~75 safety-critical system by tail number and serial number. 
      2. The NIH and CDC require "Universal Device Identifier" for all lFDA act 201(h) medical devices.  
      3. Automotive OEMs track using www.reyrey.com under warranty. At least 3 Oracle-based service dbs do it by part name, although they don't distinguish whether multiple failures of same part name were first failure of second part or second failure of first part.
    2. No. 
      1. Ships and returns counts are statistically sufficient to make nonparametric estimates of age-specific field reliability and failure rate functions for products and their parts. 
      2. For parts, you need BoMs to convert product installed base into part installed base =P(1-N)^(-1) where P is product installed base (vector) by cohort and N is gozinto matrix of parts per product or next assembly. 
  3. Do I want to buy a mainframe or cloud computer; major service db, data-mining software, and reliability statistics programs, and employ many people tediously keeping records, making data entries, getting bored, and entering errors; and learning field reliability only for tracked parts using possibly incorrect assumptions built into the reliability software?
    1. Yes, it's good for the economy
      1. Do you need help justifying assumptions built into reliability statistical software? See How to Make Data Fit Weibull? "Verification and Validation of Alternative Facts."
      2. Or would you like a constant failure rate = 1/MTBF? 
    2. No. I prefer to send installed base and field reliability data to pstlarry077@gmail.com, describe it, and ask questions, at least to see whether nonparametric, age-specific field reliability and failure rate function estimates, without life data,  (from data required by GAAP) are good enough. How much bang-per-buck do you get for information from life data vs. essentially free reliability information
  4. What if failures are dependent: successive failures or repairable systems or failures of parts within same system (e.g. Tesla Model S battery, charger, and drive unit)? Make nonparametric, multivariate distribution estimates of TTFF, TBF1, TBF2,...
    1. Are you kidding? We assume independence and fudge forecasts if they don't agree with observations (Lab 126).
    2. Wow. Can I do that even if I don't have life data TBFs for repairable systems? Yes, send your data. 
If you would like comparisons of field reliability estimates, with vs. without life data, ask for "Random-Tandem Queues and Field Reliability Estimation, Without Life Data" now on this site 

Introduction to this web site

This is a test. Nov. 12, 2015 to see whether I can upload and use *.html, workbooks, and so on. Links are not active, nor have the files been uploaded, yet. Please contact me if you want them, or take a look at the articles and files in the left column. Cleaned up July 10, 2016 FInally figured out how to make links work. More coming
This is an experiment to use the NPmle.cdf file to estimate reliability without life data. The following HTML code needs to be executed by your browser to obtain the cdf-player, wolfram.com cdf-player.  If that doesn't work, you could use NPmle.nb in Mathematica.
 <script type="text/javascript" src="http://www.wolfram.com/cdf-player/plugin/v2.1/cdfplugin.js"></script>
<script type="text/javascript">
var cdf = new cdfplugin();
cdf.setDefaultContent('<a href="http://www.wolfram.com/cdf-player/"><img  src="NPmle.png"></a>');
cdf.embed('NPmle.cdf', 1210, 1014);
Estimate field reliability, without life data
Make actuarial forecasts and estimate prediction limits
Correct homophone errors "Yeltsin Eeks Out Victory" local newspaper. "Stocks Eek out Gains," Yahoo Finance, Dec. 27, 2019
MTBF with redundancy
Reliability with redundancy
Reliability estimation with hidden renewals
Production genie says... True! 
MIL-HDBK-217F predict MTBF
Kaplan-Meier Reliability estimation
Reliability allocation
Weibull estimation, with or without life data
Spares? Finite mission!
Predict MTBF, Variable op. hours
85th %ile is biased! Max. likelihood
Dies in wafer? Exact!
Cold standby workbook
FMEA + risk & diagnostics 
Multivariate P[g(Stress, strength)=failure | Eq] old 
US NRC NUREG Observed or subjective correlations 
Design and Analysis of Accelerated Relaibility Tests 
           Recent pubs
"Random-Tandem Queues and Reliability Estimations, WIthout Life Data"
"User Manual for Credible Reliability Prediction"
Credible Reliability Prediction, second Edition is https://sites.google.com/site/fieldreliability/credible-reliability-prediction/
CRP Preface New!
Design ALT Newer!
Opportunistic Replacement Old!
Fix RNs Fix bias and variance
Firestone tire reliability
Firestone could have known in '97
Reliability is Free!
Evaluate field reliability applications
Fables of automotive reliability
FAQs about reliability
Courses on reliability statistics
Virtual reliability?
Virtual statistics? Cheat!
Newsletters Links work
Biostatistics (mis)applications
Read Bicycle news\
Progress in Artificial Stupidity
Chamber Music Workshop scheduling software AUTHOR AT PLAY!
Low B-Flat for EH
Uncertainty in the URC Uncertainty in the URC 
"Artificial Reliability: Is There Reality in Reliability?"