Contents I. Introduction
Actions on components/structures as well as structural properties are usually not constant, but will vary with time. This means that the probabilistic analysis does not include random variables only, as in the standard case, but also random functions of time, usually referred to as random processes. As a result the failure probability is no longer just a single number, but also a function of the time. In general one should always keep in mind that just mentioning a value for the failure probability does not make any sense without specifying the period of time for which it was derived. The two main classes of random processes are the stationary and non stationary processes. The basic feature of stationary processes is that their statistical properties (mean, standard deviation etc.) do not change with time. Stationary processes may be subdivided into ergodic and nonergodic processes. Ergodic means that all statistical properties can be inferred from a single observation, supposing it is long enough. Conventional reliability models seldom considered the times of load action. The reliability calculated by these models is actually the reliability when random load acts for specified times. For general components and systems, these models can't reflect the effect of times of load action on reliability explicitly. In the present paper different types of load models and the methodology for solving the time invariant and time variant reliability problems has been discussed. These models have been developed using probability differential equations. The relationship between reliability and time is discussed in this paper. The time varying reliability model of components without strength degradation, and that with strength degradation is also discussed.
