Developing activity duration specification limits for effective project control

We propose a new structure for guiding project control decisions to ensure that a project is completed on schedule when activity durations are uncertain and modeled by random variables. This structure consists of specifying a specification limit for each activity duration. During the project, if the time to complete an activity is going to exceed its specification limit, actions are taken, at some cost, to bring the time down to that limit. We present an algorithm that selects specification limits to achieve targeted on-time probabilities at minimum cost. The method involves estimating the effect of small changes in specification limits on the probability of completing a project on time and on the cost of control actions. The required simulation-based estimates for all activities are obtained in a single set of simulation runs. Computational results show the algorithm to be efficient to apply and, when compared to a more ad hoc approach of using activity importance measures (specifically, activity criticality), the use of the resulting specification limits to be of significant benefit in guiding project control decisions.