Model for predicting risk in scheduling proposed R&D tasks

This paper develops a model that can be used to predict a potential variance from planned schedule for a given R&D task. The task is defined as the net increase in knowledge required to move from the initial level of basic knowledge about the problem to the level specified for its accomplishment. Variables are identified that affect the time required for attainment of successive knowledge- state levels, assuming that an adequate technical solution is feasible and adequate funds are available. The variables are defined and evaluated for sixteen historical R&D tasks from an industrial laboratory through use of comparative value estimation techniques, and a correlation-regression analysis is performed with time as the dependent variable. This information is used to structure a predictive model for time to attain intermediate knowledge states given values for the variables associated with these subactivities. Total task time is obtained by addition of predicted times to attain the intermediate knowledge states. The predicted task time is normally distributed with variance including variance and covariance of the intermediate knowledge-state times. The validity of the model is tested on a task representative of a laboratory activity beginning in its planning stage and continuing through the knowledge states of design concept, design, unit assembly, and developing the unit. Faced with evaluation of a new R&D task, management can then determine the optimum method to lower the risk commensurate with its goals/objectives by reassignment of resources or developing alternative technical approaches.