COURSE DESCRIPTION

This class starts with foundational concepts to develop an understanding of reliability in the three parts of the bathtub curve. Reliability mathematics are explored using the exponential distribution and constant failure rate as the first example. Examples are sprinkled throughout to help reinforce concepts. The first day ends by illustrating the advantages in prototype quantity and test time of using HALT to discover relevant failure modes and to do a comparison demonstration of reliability for new product release.

The second day generalizes concepts and mathematics developed the first day into the Weibull distribution and non-constant failure rate. The vertical scale on a Weibull graph is de-mystified as are censored and interval testing. Applications of Weibull Analysis for minimal testing using WeiBayes and zero-failure test strategies to save time and money are presented. Field failure prediction and accelerated testing using Weibull Analysis culminate the two days.

WHO SHOULD ATTEND

This course is intended primarily for Engineers and Managers in New Product Development, Reliability, and Quality.

COURSE OUTLINE

DAY ONE:

• Definition of reliability & failure; strength/load interaction; and overview of the bathtub curve.
• Details of Wear-Out failures: Reservoir & process which is consuming it; Fatigue accumulation; Cycle testing with in-class example of metal fatigue using Weibull Analysis and confidence bounds.
• Details of Useful-Life failures: Strength/Load calculation; HALT to discover relevant failure modes; how to prove that HALT works in your product; preparing for HALT; using off-site test facilities.
• Details of Early-Life failures: Manufacturing anomalies; shipping, handling, and storage damage; run-in, burn-in, Environmental Stress Screening, and HASS to discover relevant failure modes; Safety of screen.
• Recap of discovering and curing failure modes in the three parts of the bathtub curve.
• Reliability mathematics: Reliability Function, Cumulative Distribution Function, Probability Density Function, and Hazard Function; general relationships among these functions.
• First example: Exponential distribution and Constant Failure Rate; confidence bounds, failure-truncated & time-truncated testing, Serviceability and Availability.
• Reliability Block Diagrams: series, parallel, & redundant configurations; reliability calculations for each; cut sets, tie sets, and classical reliability “predictions” (MIL-HDBK-217 and Telecordia).
• Appreciating HALT by comparing prototype quantity and test time with traditional testing.
• HALT and a non-parametric, ranked-sum method to “demonstrate” reliability before product release.

DAY TWO:

Weibull distribution: all four functions for two or three parameters; how a Weibull graph give a straight line; meaning of the slope and scale factor; establishing the vertical position on a Weibull graph.

Methods of fitting curves: Rank Regression, Maximum Likelihood Method, modified MLE; suspended testing; Median Ranks; multiply-censored data; Auth’s formula and Benard’s formula.

Curved Weibulls; mixed failure modes; misapplication of the three-parameter model; confidence intervals; Fisher Matrix, and Likelihood Ratio; Interval Testing; WeiBayes Analysis.

Minimal Testing with zero-failure test strategies; substantiating improvement either prototype or time constrained; reliability demonstration with known Beta; cost effectiveness.

Predicting future failures from test data or current field history; Accelerated testing Weibull Analysis.