What Is the Hazard Function?

The hazard function is a mathematical construct in reliability engineering and survival analysis that expresses the instantaneous rate of failure for a component or system at any given point in time, given that no failure has yet occurred. It is defined as the ratio of the probability density function of failure time to the survival function, and it carries units of failures per unit time. Because it conditions on survival to time t, the hazard function characterizes risk dynamically rather than simply summarizing the total probability of failure over a fixed interval.

The concept originated in actuarial science and statistics, where it was used to model human mortality, and was adopted by reliability engineers in the mid-twentieth century to describe component and system lifetimes. In engineering contexts, the hazard function is also called the failure rate function or the instantaneous failure rate.

Mathematical Definition and Properties

For a random variable T representing the time to failure with probability density function f(t) and cumulative distribution function F(t), the hazard function h(t) is given by h(t) = f(t) / [1 - F(t)], where the denominator is the survival function S(t). The cumulative hazard function H(t), the integral of h(t) from zero to t, is related to the survival function by S(t) = exp(-H(t)). This relationship means that if the hazard function is known at all times, the full distribution of failure times can be recovered.

The NIST/SEMATECH e-Handbook of Statistical Methods provides detailed derivations of the hazard and reliability functions and their relationships for the commonly used lifetime distributions, including the exponential, Weibull, and lognormal families.

Failure Rate Patterns and the Bathtub Curve

The shape of the hazard function over time is diagnostic of failure mechanisms. A constant hazard function corresponds to the exponential distribution, which implies that failures occur at a steady, memory-less rate regardless of age. This describes purely random failures, such as those from random electrical transients or manufacturing defects that survive burn-in.

An increasing hazard function, as in the Weibull distribution with shape parameter greater than one, indicates wear-out: the probability of failure increases as the component ages. A decreasing hazard function indicates early-life failures, where components with latent defects fail quickly, and survivors become increasingly reliable over time.

The combination of all three patterns in the lifetime of a product population produces the characteristic bathtub curve: a decreasing early-failure rate, a roughly flat random-failure region, and a rising wear-out tail. The Wiley Online Library chapter on reliability and hazard functions discusses how these regime boundaries are estimated from field data and used to inform warranty policies and maintenance scheduling.

Hazard Severity and Risk Integration

The hazard function quantifies the rate of failure but does not by itself capture the consequences of that failure. In safety-critical engineering, hazard severity assessment assigns consequence categories (ranging from catastrophic to negligible) to each failure mode. Risk is then a function of both the hazard rate and the severity: a high hazard rate with negligible consequences may be acceptable, while a low hazard rate with catastrophic consequences may not be. Six Sigma programs use reliability data and hazard function models to drive defect-reduction targets in manufacturing, connecting statistical failure rates to quality cost.

Survival analysis methods, closely related to the hazard function, are applied in clinical trials and biomedical research by NIH-supported investigators to model time-to-event outcomes such as disease recurrence or device failure in implanted medical devices.

Applications

The hazard function has applications across engineering and quantitative science, including:

• Reliability prediction and warranty cost estimation for electronic components
• Maintenance scheduling and inspection interval optimization
• Life testing and accelerated aging analysis
• Survival analysis in clinical trials and medical device evaluation
• Quality control and defect rate modeling in Six Sigma programs
• Insurance and actuarial modeling for equipment and liability products