This is a clear exposition of a procedure which can be used to estimate the reliability of complicated systems. It is based on a combination of predicted failure rates and measured failure rates, where the system configuration under test does not necessarily correspond to the system configuration for which the prediction is required. Alternatively, and equivalently, the procedure is used to estimate reliability for different operational modes, where an operational mode is defined by the equipment used for that mode. All models are single-thread (i.e., serial) reliability models and, as presented, do not apply to computer complexes with redundant elements and self-repair.
The procedure should be useful to those who wish to estimate the reliability of a large system, taking advantage of a combination of theoretical data (e.g., MIL STD 217) and field measurements of different configurations of the same or similar equipment. As such, the method should be especially useful in proposal work.