Accelerated Testing
for predicting and improving product reliability

Instructor: Dr. Wayne Nelson

Many products last for years under normal use conditions. However, product tests must quickly yield reliable information for management and engineering decisions. Accelerated testing quickly yields such information. Test specimens are subjected to higher than normal levels of temperature, voltage, humidity, vibration, etc., and fail much sooner. Then a physical-statistical model is fitted to the early failure data, yielding estimates of product reliability under normal use conditions, including the failure rate, the population percentage failing on warranty or during design life, the mean time to failure, etc. This course shows how to measure and improve the reliability of diverse products.

Benefits

• Determine test purposes.
• Accelerate tests through overstress, high-usage rate, termination of the test before all specimens fail, specimen design, degradation, cycling, etc.
• Plan efficient tests that yield the most accurate product information for the allotted test time, cost, and constraints.
• Model and understand how product life or degradation depends on the accelerating stresses and on design, manufacturing, materials, and operating variables – thereby learning how to improve product reliability.
• Analyze data with simple and informative plots to estimate reliability and to assess the data and presumed model.
• Analyze data with a mix of failure modes 1) to estimate product reliability at a use condition when all failure modes act or 2) to predict reliability if some failure modes were eliminated through proposed design or manufacturing changes.
• Compare designs, manufacturing methods, vendors, materials, etc., with respect to their effect on product reliability.
• Analyze data with modern computer programs that fit complex models to data and provide estimates and corresponding confidence limits, which indicate the accuracy of the estimates.

Who should attend?

This course will benefit engineers, statisticians, and others working in product development, reliability, testing, manufacturing, procurement, and data analysis. You will learn to plan efficient accelerated tests and to accurately estimate product reliability and improve reliability using test data. To benefit fully, you need a working knowledge of a basic Statistics course.

Applications

The methods are applied to a variety of products including:

• Electronics, including microprocessors, semiconductors, solid-state devices, conductors, encapsulants, connections, and capacitors.
• Metal components undergoing fatigue, creep, creep-rupture, and corrosion.
• Electrical insulations and dielectrics in cables, motors, capacitors, electronics, transformers, etc.
• Ceramics, plastics, composites, and other materials.
• Pharmaceuticals, foods, paints and protective coatings, rubber and elastic materials, and chemicals.
• Munitions, propellants, and other energetics.

You will apply the methods to actual applications, including your own data, which you are asked to bring.

Text

The course text Accelerated Testing: Statistical Models, Test Plans, and Data Analyses, published by Wiley (2004), was written by Wayne Nelson. The textbook, plotting papers, computer program information, and other reference materials are furnished at the course.

Instructor

Dr. Wayne Nelson consults privately on and teaches engineering applications of Statistics for many companies. Formerly with General Electric Co. for 23 years, he consulted across many company divisions. He was elected a Fellow of the Inst. of Electrical and Electronic Engineers (IEEE), the Amer. Soc. for Quality (ASQ), and the Amer. Statistical Assoc. (ASA) for his contributions to Reliability data analysis and Accelerated Testing. In addition to the course text, he authored Applied Life Data Analysis (Wiley 1982), Recurrent Events Data Analysis for Product Repairs, Disease Recurrences, and Other Applications (ASA-SIAM 2003), and two ASQ booklets. The ASQ awarded him the 2003 Shewhart Medal for his technical innovations. The IEEE Reliability Society presented him the 2004 Lifetime Achievement Award for his contributions to Reliability. Among his 120+ literature publications, he received the Brumbaugh Award, Wilcoxon Prize, and Youden Prize of ASQ and eight Outstanding Presentation Awards of ASA. He has a Ph.D. in Statistics and an M.S. in Physics from the Univ. of Illinois and a B.S. in Physics from Caltech. Contact him at WNconsult@aol.com and www.members.aol.com/WNconsult.

Continuing Education Units

Upon completion of the course, participants will be eligible to receive 1.7 Continuing Education Units (CEUs).

Schedule

The following schedule includes morning, afternoon, and lunch breaks.

Day 1

Survey: applications, types of data, types of accelerated tests and stress loading, practical engineering and statistical considerations, and common tests (overstress, ESS, HALT, HASS, burn-in, single test condition, elephant).

Models: statistical life distributions (exponential, Weibull, lognormal, and others) and physical life-stress relationships (Arrhenius, inverse-power, Coffin-Manson, Peck’s, fatigue, etc.), including multivariable models for the effect of design, materials, manufacturing, operating, and other engineering variables.

Day 2

Graphical Analysis: simple probability plots (Weibull, lognormal, etc.) and relationship plots (Arrhenius, inverse-power, etc.) that yield estimates of product life under use conditions and assessments of the data and proposed model (distribution and relationship).

Computer Analysis: maximum likelihood fitting of models to censored data with runouts to obtain estimates and confidence limits for product reliability and model parameters, and to assess the data and model.

Test Plans: how to choose test conditions and the number of specimens at each, including optimum, traditional, and good compromise plans.

Day 3

Competing Modes: the series-system model for products with a number of failure modes, graphical and computer analysis of such data, and the effect of specimen and product size.

Step and Varying Stress: models for cumulative exposure and data analyses for step and varying stress.

Degradation Testing: basic models and data analyses for accelerated performance degradation data.

Better Accelerated Testing: an overview of important points.