A proposed model for accelerated‐stress life‐test data exhibiting two failure modes

Abstract

AbstractAccelerated‐stress life‐testing involves testing of a material or system at a series of stresses higher than that which is normally seen in service. An estimate of the service life is then obtained by extrapolating the failure time vs stress relationship to use stress. A common problem with this methodology is that a change in failure mechanism may occur if stresses become too high. When this occurs, the data above and below the failure mode transition point are generally treated separately. This paper discusses a model that allows for a change in failure mechanism and characterizes the transition by a probability density function over stress. Advantages of the model are: it allows representation of all test data by a single equation; it allows probability statements to be made about the probability of failure of either failure mode; it allows optimal experiments to be designed; and it allows data in the area of the failure mode transition to contribute to the precision of the estimated service life.