Schedule a Software Walkthrough
Software Reliability Growth Models
SRGM Reference — Model 4 of 14

The Rayleigh Model

The Rayleigh model is the special case of the Weibull model with its shape parameter fixed at 2, producing the classic rise-then-fall defect curve long used to describe how defect discovery rises through early testing, peaks, and tapers off toward release.

Fixed-Shape NHPP
Weibull model with shape parameter c = 2
Classic Defect Curve
Rooted in Norden's 1950s-60s staffing/defect profile work
Practical uses
The Rayleigh model is most useful for identifying the "peak" of the defect discovery. The Weibull model is more flexible.
Overview

The defect curve behind decades of lifecycle planning

Long before it was formalized as a software reliability growth model, the Rayleigh-shaped curve was already familiar from Peter Norden's work on project staffing and defect discovery profiles: activity ramps up, peaks, and tapers off in a predictable rise-then-fall pattern. As an SRGM, it captures the same behavior in cumulative failure data.

Model Family

Weibull, Shape Parameter = 2

Rayleigh is not a separate model family so much as a fixed, well-known special case of the more general Weibull curve.

Origin

Norden-Rayleigh Lineage

Traces back to Peter Norden's staffing and defect-profile research, later adapted directly into SRGM form.

Requs Trend

Covered — Rayleigh

This model is one of the five named model families in Requs Trend's ensemble averaging.

01 — The Math

How the Rayleigh Model works

Because Rayleigh is a fixed-shape Weibull curve, its formulas are the general Weibull equations with the shape parameter locked at 2, leaving two free parameters to fit.

§1Mean Value

Mean Value Function

Formula
μ(t) = a (1 − e^(−bt²))
Cumulative expected failures — the Weibull form with c fixed at 2

The squared exponent produces the characteristic rise-then-fall shape in the failure intensity, unlike the strictly concave exponential model.

What the Terms Mean
  • μ(t) — expected cumulative number of failures by time t
  • a — total expected number of faults (the asymptote)
  • b — a scale parameter controlling where the peak of the curve falls
§2Intensity

Failure Intensity Function

Formula
λ(t) = 2abt e^(−bt²)
Instantaneous failure intensity, peaking partway through testing

λ(t) rises from zero, peaks at t = 1/√(2b), then falls back toward zero — the signature rise-then-fall Rayleigh shape.

What the Terms Mean
  • λ(t) — instantaneous failure intensity at time t
  • a, b — same parameters as the mean value function
  • The peak location and height are both determined by b alone
§3Assumptions

Key Assumptions

Model Assumptions

Because the shape is fixed, Rayleigh assumes the data actually follows a symmetric rise-then-fall pattern rather than fitting whatever shape the data shows.

Assumption List
  • A finite total fault count a exists in the software
  • Failure intensity rises, peaks, and then falls in the specific rise-then-fall shape fixed by c = 2
  • Fault repair is instantaneous and perfect
  • Testing effort follows a roughly consistent ramp-up-then-taper profile, mirroring the classic Norden staffing curve
§4Estimation

Parameter Estimation

Method

With the shape parameter fixed, only a and b need to be estimated, typically via nonlinear least-squares regression or maximum likelihood estimation — making Rayleigh easier to fit reliably than the general Weibull model on smaller data sets.

What's Being Estimated
  • a — the eventual total fault count, from where the curve flattens
  • b — the location of the peak in the failure intensity curve
02 — Strengths & Limitations

Strengths, limitations, and best fit for the Rayleigh Model

STRENGTHS

Fewer parameters than the general Weibull

With the shape fixed, Rayleigh needs less data to produce a stable fit than the full three-parameter Weibull model.

LIMITATIONS

Locked into one specific shape

If the real defect curve doesn't actually follow the classic rise-then-fall pattern, the fixed shape parameter can't adapt the way the general Weibull model can.

BEST USED FOR

Lifecycle-phase defect planning

A longstanding standard for projecting defect density across development phases, from early testing through release.

REQUS TREND

Included in the ensemble

Listed directly among Requs Trend's five named model families, alongside Exponential, NHPP, Weibull, and Logarithmic.

See this model fit against real project data.

Start with the Requs Trend demo or a discussion of your current reliability growth modeling approach.