Predicting reliability investment to achieve given reliability improvement

One of reliability engineering´s long standing needs has been a credible and repeatable method for estimating the investments required to satisfy stated reliability requirements. This paper reports on the development of a mathematical model that can be used to predict the investment in reliability required to achieve a given amount of reliability improvement. An appropriate analogy is the constructive cost model (COCOMO) used to predict software development cost. Similar to COCOMO, the reliability investment model comprises a set of sub-models. We have developed three thus far: 1) a basic model that computes reliability development effort and cost as a function of program size and desired reliability improvement. Its purpose is to provide rough order of magnitude estimates when very limited program information is available, as is typical at the early stages of programs; 2) an intermediate model that computes development effort as a function of program size, desired reliability improvement, and a set of relevant cost drivers; and 3) a production and support cost model to estimate the variation of investment in production and the change in operations and support cost due to an improvement in reliability.