Statistics and Reliability Meeting: August 9
Please forward this invitation to your friends and colleagues.
The ASQ Silicon Valley Statistics and Reliability Discussion Group cordially invites you to attend an outstanding FREE presentation. Welcome IEEE and ASA members.
TOPIC: Random-Tandem Queues and Reliability Estimation Without Life Data
SPEAKERS: Dr. Larry George
Assistant Chair: Dr. Don Mintz
Corporate Sponsor: Applied Materials
Sponsor: Dr. Charles Chen from Applied Materials
ABSTRACT:
Ships and return counts are statistically sufficient to make nonparametric estimates of field reliability and failure rate functions. But life data contains more information. Is it worthwhile to track everything by serial number to get life data? Compare alternative data estimators’ bias, variances, covariances, entropy, (–∑p(i)log(p(i)) quantifies information in probability distributions better than variance), and cross-entropy (Kullback-Leibler divergence) guides portfolio allocation to reliability data alternatives!
What if failures are not dead forever; i.e., repairable, renewal, or recurrent processes? Superposition of staggered-start-cohort recurrent processes’ failure counts is statistically sufficient. M88-A1 engines, Nigerian refinery pumps. Compare alternative data.
Is cost of s(â(t|ships and returns)) > Cost of s(â(t)|life data)) (â(t) is failure rate function estimate.)? Ships and returns counts data required by GAAP costs ~$0.00 and preserves privacy (e.g., AIDS patients’ names). Cost of uncertainty in an actuarial forecast, ∑â(t)n(t), (n(t) is installed base of age t) and depends on
s2[∑â(t)n(t)] = ∑s2(â(t))n(t)2 + ∑∑n(s)n(t)Cov(â(s), â(t))
Risk equity: shouldn’t consumer risk = producer risk!? The Producer has the advantage of better info from life data. The Consumer has to depend on consumer groups, lawyers, and government.
Progress in Artificial Stupidity reports on “How to Make Data Fit Weibull;
Verification and Validation of Alternative Facts” and
“How to make a constant failure rate = 1/MTBF”
BIOGRAPHY:
Larry George has a PhD from UC Berkeley in Industrial Engineering, an MBA from UCLA and 11 years of teaching and training experience including Texas A&M. He worked at Lawrence Livermore National Laboratory for 11 years and at Apple and Applied Materials. He has more than 30 years of field reliability experience including employment, contracts, consulting, and free service.
Intense pressure and lack of life data forced Larry to derive age-specific field reliability estimators from “ships and returns” counts. Fortunately, GAAP requires data containing population ships and returns. I.e., everybody can estimate their products’ and parts’ reliability and failure rate functions and do something about them (if not satisfactory).
Larry is a fellow of the ASQ, ASA, and International Double Reed Society.
Please send an RSVP to johnflaig@yahoo.com at least two days prior to the meeting if you plan to attend.
statnrel.wordpress.com 