Reliability, Life Testing, and Shelf Life

Abstract

Everything will fail, eventually. Reliability is a probability that a system (or a component) will fail no sooner than t time units from the time it begins operating. Reliability has many manifestations, or one might consider reliability as a special case of probabilities that some specific type of event will occur no sooner than t units from some initial reference time. Other manifestations include survival of patients having a particular disease, or the shelf life of drugs. For simplicity, we will refer to reliability in terms of time to failure, with the understanding that this time could be the time from a reference point to the occurrence of some specific type of event (such as death, progression of disease, a drug concentration in a human body drops below some threshold, or potency/reactivity loss). The point is to derive a model by which the probability of interest can be predicted, and to incorporate design parameters into this model. In this way, we hope to help the EAS design a system so that it will have a desired reliability for a particular time-to-event.