A comparison of PM2 and crow extended growth planning models

Reliability growth planning is an important part of the reliability growth process. Its purpose is to provide achievable benchmarks throughout developmental testing in order to evaluate whether the project is on track to meet its requirement. Different models produce different growth curves. This paper compares two popular models, the Planning Model Based on Projection Methodology (PM2) and Crow Extended (CE), and highlights several features and patterns that are inherent to each of them through examples. Crow Extended was found to consistently model more aggressive growth based on fewer expected problem failure modes than PM2 in the early phases of a test. Given that design for reliability techniques are implemented specifically to mitigate the highly-significant problem failure modes that lead to aggressive early growth, a program that has designed for reliability should not exhibit the type of growth curve modeled by CE. Such a model will show the program to be constantly failing its benchmarks. In contrast, PM2 curves tend to be more plausible, with relatively consistent, moderate growth. A non-aggressive approach would be preferable for most planning purposes, as both moderate and aggressive growth programs can achieve the benchmarks. The shape parameter used in CE is based on historical values, and as such, has a certain amount of subjectivity in its estimation. Varying the parameter within the acceptable bounds of error had a considerable effect on the amount of test time required for the CE model to meet the reliability goal.