Performance comparison of activity sensitivity metrics in schedule risk analysis

Abstract

In Schedule Risk Analysis (SRA), activity sensitivity metrics measure the importance of activities in a project schedule. Highly sensitive activities are those more likely to increase project duration variability and/or cause project duration extensions. Several activity sensitivity metrics have been proposed over the years, but a comparison of all of them has never been made. This has made it difficult to know which metrics perform better and under what circumstances. In this paper, an extensive comparison of all relevant SRA activity sensitivity metrics is performed using a set of 4100 artificial projects. Unlike previous studies, the comparison framework is decoupled from corrective actions (e.g. activity crashing) which allows the merits of each metric to be assessed individually. Additionally, a new metric that performs better for overall sensitivity ranking is proposed. Results show that most sensitivity metrics do not perform well unless they are applied iteratively (the sensitivity of the remaining scheduled activities has to be recalculated whenever the duration variability of at least one activity has been restricted). However, if applied iteratively, most metrics can enhance project monitoring and control, while significantly shortening project duration.