Principles of Sequencing and Scheduling, 2nd Edition

19 PERT 21: Analytics‐Based Safe Project Scheduling

19.1 Introduction

Historically, CPM and PERT were not merely theoretical frameworks for project scheduling and resource allocation, but they were also marketed successfully as decision support systems (DSS). To this day they retain that function. Our coverage in the previous three chapters provides the theoretical underpinnings of project scheduling as reflected in most of those systems. In this chapter, we progress beyond the standard functionality. We do so mainly by introducing more advanced safe scheduling models. Furthermore, these models utilize reliable samples that reflect historical experience for simulation‐based analysis as discussed in Chapter 18. We refer to the framework we describe as PERT 21, because it adapts PERT/CPM for use in the twenty‐first century. The most important difference between PERT 21 and PERT/CPM as implemented in project management DSS today is that it is based on validated distributional assumptions. Those assumptions are justified by empirical evidence validating the use of lognormal processing times subject to linear association and possibly the Parkinson effect (see Chapter 18). Serendipitously, instead of triplets as in PERT, users need to input only estimates of mean durations. All other necessary parameters can then be estimated from the historical performance of similar single‐point estimates. It is even possible to bypass the need to estimate parameters explicitly and still achieve ...