A system dynamics model of a large R&D program

Abstract

Organizations with large R&D activities must deal with a hierarchy of decision regarding resource allocation. At the highest level of allocation, the decision is related to the total allocation to R&D as some portion of revenue. The middle level of allocation deals with the allocation among phases of the R&D process. The lowest level of decisions relates to the resource allocation to specific projects within a specific phase. This study focuses on developing an R&D model to deal with the middle level of allocation, i.e., the allocation among phases of research such as basic research, development, and demonstration. The methodology used to develop the R&D model is System Dynamics. Our modeling concept is innovative in representing each phase of R&D as consisting of two parts: projects under way, and an inventory of successful but not-yet-exploited projects. In a simple world, this concept can yield an exact analytical solution for allocation of resources among phases. But in a real world, the concept should be improved by adding more complex structures with nonlinear behaviors. Two particular nonlinear feedbacks are incorporated into the R&D model. The probability of success for any specific project is assumed partly dependent upon resources allocated to the project. Further, the time required to reach a conclusion regarding the success or failure of a project is also assumed dependent upon the level of resources allocated. In addition, the number of successful projects partly depends on the inventory of potential ideas in the previous stage that can be exploited. This model can provide R&D management with insights into the effect of changing allocations to phases whether those changes are internally or externally driven. With this model, it is possible to study the effectiveness of management decisions in a continuous fashion. Managers can predict payoffs for a host of different policies. In addition, as new research results accumulate, a re-assessment of program goals can be implemented easily and allocations adjusted to enhance continuously the likelihood of success, and to optimize payoffs. Finally, this model can give managers a quantitative rationale for program evaluation and permit the quantitative assessment of various externally imposed changes.