Including Generative Mechanisms in Project Scheduling Using Hybrid Simulation

Abstract

This research seeks to expand the range of concepts used in project scheduling and aligns with the broader research effort to rethink project management. It begins with the observation that current project scheduling practice is based on ideas from the 1960’s, whilst research has developed considerably since that time. It adopts the approach of identifying and challenging assumptions underlying both the research and practice of project scheduling, to develop an understanding of why practice seems to ignore research. It develops the view that the issue is fundamentally philosophical in nature.

The solution proposed involves relaxing the constraints, both ontological and epistemological, that a preponderance of positivistic thinking imposes on project scheduling. Relaxing ontological constraints involves expanding the range of things acknowledged to be real. Adopting the philosophy of Critical Realism is proposed as one way to achieve this. Linked to this are epistemological concerns regarding the clarity of the explanations we construct, using our knowledge of real things, for the events we observe. The adoption of mechanistic explanation, using Critical Realism’s generative mechanisms, is also proposed for its causal clarity.

Operationalisation of ontological expansion and causal clarity is pursued through a sequence of decisions that narrow down these theoretical concerns to ideas that can be implemented, with existing technology, for the purposes of resource constrained project scheduling. Design Science Research is adopted as the framework to guide this process. The novel idea of using the concepts of work-effort and resource-periods is proposed, as the basis for a new mathematical treatment of resource productivity that allows variation during schedule construction. Variable resource productivity is identified as a gateway technology supporting an expanded range of scheduling concepts that can be linked through non-linear relationships, feedback and time delays.

Operationalisation of these ideas proceeds through designing, building and testing a scheduling engine, as a proof of principle. Its scheduling capabilities are benchmarked against research algorithms and commercial software. Eight scheduling heuristics were tested, using 1,560 computer-generated project networks of 30, 60 or 120 tasks, comprising 12,480 resource constrained schedule duration calculations. The best heuristic, prioritising tasks with the earliest Critical Path Method parameter of ‘Late Finish’, compared favourably to the benchmark data. The development of the scheduling engine is presented as an example of how a design approach can be useful in management science research and its use is illustrated by modelling the impact of overtime induced fatigue on project schedule duration.