Distribution Identification Calculator
Before fitting parameters you have to choose a distribution, and that choice affects extrapolated life far more than the parameter estimates do. Distribution identification ranks candidate distributions against your data using goodness-of-fit statistics — but the physics of the failure mode should always get a vote alongside the statistics.
Formula
Anderson-Darling statisticLower AD indicates a better fit; it weights the tails more heavily than Kolmogorov-Smirnov.
Correlation coefficientρ from the probability plot; closer to 1 is better.
Likelihood-based comparisonAIC = 2k − 2·ln(L)
Penalises extra parameters, so a 3-parameter Weibull must earn its third parameter.
How to interpret the result
| Weibull | Weakest-link failures, wear-out, most mechanical life data. The flexible default. |
| Lognormal | Multiplicative degradation — crack growth, diffusion, corrosion — and repair times. |
| Exponential | Constant hazard only. Justified for some electronics in useful life, rarely elsewhere. |
| Normal | Wear-out with a well-defined mean life and symmetric spread; less common for life data. |
Common mistakes
- Choosing on goodness-of-fit alone with a small sample. Below roughly 20 failures, several distributions will fit acceptably while extrapolating to very different B10 lives.
- Fitting one distribution to data containing multiple failure modes. A curved probability plot is the tell — segment by mode first.
- Defaulting to exponential because it is mathematically convenient. It assumes no wear-out at all, which is a strong physical claim.
- Ignoring censored data in the fit comparison.
Related tools
Weibull SolverLife Data & Field Analysis
Lognormal SolverLife Data & Field Analysis
Field ReliabilityLife Data & Field Analysis
Fleet FailureLife Data & Field Analysis
WarrantyLife Data & Field Analysis
Reliability ForecastLife Data & Field Analysis
About Reliability Pro
Distribution Analyzer is part of a suite of 126 reliability, maintenance and quality engineering tools covering life data analysis, accelerated testing, system reliability, FMEA and root cause, SPC, and design for reliability. It runs in the browser and as native iOS and Android apps, so the same calculation is available at a desk or in front of the asset.