Multiteam, Multiproject Research and Development Planning with GERT

Abstract

This paper reports on a simulation study of multiple research and development projects that are worked on concurrently and sequentially by more than one research team. The technique employed in the modeling and simulation effort was GERT, which was used because of its capability to incorporate the probabilistic outcomes and feedback loops common to R&D projects. The GERT (Graphical Evaluation and Review Technique) includes the modeling of systems in network form and analysis through simulation. It is an important example of technological spin-off from space science research. GERT provides the capability to model and analyze networks of a very general form. Some of the features included in GERT networking are probabilistic branching (stochastic models), network looping (feedback loops), multiple sink nodes (multiple outcomes), multiple node realizations (repeat events), and multiple probability distributions (assigned to activity times). GERT networks are typically analyzed using the GERTS IIIZ simulation package, which provides statistics on project times and cost. Results of the simulation included statistical data on individual project duration and cost as well as overall network time and cost. These results were employed to provide management with an evaluation of different model configurations, prepare overall time and cost estimates as inputs to contract negotiations and to plan and schedule manpower, equipment and capital. Sensitivity analysts was employed to determine key stages in the R&D process where changes in success-failing probabilities might reduce network time and cost. Implementation experiences validated the model as satisfactory with the exception of several minor problems.