Relationship between Weibull and AMSAA models in reliability analysis : A case study

Historically, reliability of systems has been tracked based on a common assumption that, at the system level, the failure rate follows the exponential distribution, and is therefore assumed to be constant over the useful life of the system. However, this method, while adequate for many purposes, does not necessarily provide the early warning system that many companies need to stay ahead of expensive quality or reliability fixes. This paper presents a new method that provides the needed early warning, at a reasonable analysis cost, by combining the use of two reliability distributions for the purpose of analyzing fielded systems. In particular, this paper describes a hypothesized relationship between a key parameter contained in the Weibull distribution and within the Army Material Systems Analysis Activity (AMSAA) reliability growth model. Actual data from General Electric Transportation Systems (GETS) were used to explore this relationship. The results suggest that there indeed exists a significant relationship between the two models and both can be used in tandem to track reliability of systems.