Clarification of node representation in generalized activity networks for practical project management

The activity network (e.g. PERT, CPM) has long been established in industry as a tool for planning and managing projects. One of its shortfalls is its rigid analysis and specification of the project structure-a point that most managers take for granted. Applying flexibility to the network structure in the form of stochastic activity nodes results in a generalized activity network. This type of activity network allows more rigorous analysis of project plans by predicting and mirroring more accurately the events that occur in real development situations. This enables managers to forecast risks more easily, and plan for project variations proactively. Generalized activity networks are particularly suited to projects in which uncertainties exist, for example research and development projects and software-development projects. Previous literature in this field provided either ambiguous node definitions or overly complex representation and analysis. The paper clarifies the limitations imposed by existing systems, and provides a smooth transition for project managers currently using standard PERT-type methodologies. Examples of each node type in a practical project scenario are provided.