Engineering Fundamentals of Weibull Analysis
Weibull life data analysis (standardized under IEC 61649) is the international engineering benchmark for modeling component time-to-failure behavior. Unlike simplistic constant failure rate (MTBF) models which assume equipment never degrades or ages, the Weibull distribution incorporates aging, infant mortality, and wear-out dynamics into a versatile probabilistic framework.
Mathematical Formulations
Where β is the dimensionless shape parameter (Weibull slope), η is the characteristic life or scale parameter (at which exactly 63.2% of components have failed), and Γ(·) is the complete Gamma function.
Worked Industrial Example
A plant engineer tests 7 heavy-duty centrifugal pump mechanical seals. Five seals fail at 120, 250, 410, 600, and 850 operating hours. Two pumps are still running smoothly at 1,100 and 1,450 hours when the inspection window closes (right-censored suspensions).
Results from Rank Regression on X (RRX):
- Shape Parameter (β): ~1.42 (Indicates early mechanical wear-out / abrasion).
- Scale Parameter (η): ~985 operating hours.
- B10 Life: ~202 hours. (10% of seals will fail before 202 operating hours).
- Recommendation: Do not use calendar-based overhaul beyond 500 hours; schedule seal replacements during planned outages near B10 life.
Standards & Engineering Limitations
Adheres to IEC 61649 (Weibull Analysis) and ISO 19973. Rank Regression on X (RRX) is recommended for small sample sizes (N < 20) with suspensions because it minimizes errors in time measurement. For large automated CMMS datasets (N > 50), Maximum Likelihood Estimation (MLE) provides asymptotically unbiased estimates and robust confidence intervals.
Frequently Asked Questions
What is the difference between complete data and suspended data?
Complete data means every unit in the test group was run to failure. Suspended (right-censored) data means some units operated up to a certain inspection hour without failing. The Johnson adjusted order method ensures suspensions contribute valuable survival information without skewing median ranks.
Why is B10 life so important for bearing and seal reliability?
B10 life represents the point in time at which 90% of the population remains surviving and only 10% has failed. In ISO 281 standards for rolling element bearings, basic rating life L10 corresponds mathematically to Weibull B10 life.
When should I use a 3-parameter Weibull model?
Use a 3-parameter model when physics dictates a guaranteed failure-free period (location parameter γ > 0), such as zero fatigue cracking before a known threshold of mechanical stress cycles.