Reliability growth and demonstration: The Multi-Phase Reliability Growth Model (MPRGM)

The Multi-Phase Reliability Growth Model (MPRGM) has two key distinguishing features when compared with extant reliability growth models. Firstly, it accurately incorporates complex reliability growth test phases based on `design dates,' allowing observed reliability characteristics to be attributable to specific points in reliability growth. The incorporation of `design dates' allows complex, overlapping and dissimilar phases of reliability growth testing to be accurately incorporated into subsequent analysis. In contrast to the MPRGM, many existing models assume immediate implementation of design improvements (corrective actions or fixes) on discovering failure, or uniform application of `design delays' in implementing reliability growth test phases. Secondly and perhaps most importantly, the MPRGM involves an `abstract time,' 10, associated with preliminary design. This time implies that a system's reliability is `zero' at the point where preliminary design commences and not the start of reliability growth testing. This overcomes the problem of many Power Law reliability growth models that imply a nonsensical mean time between failure of zero at the commencement of reliability growth. By overcoming these shortcomings, the MPRGM is equally applicable to reliability growth planning, tracking and projection. It also allows data from multiple reliability growth test phases to be aggregated in a way that provides assurance on demonstrated reliability growth, to the extent that it can be used in an acceptance testing framework.