Scheduling Programs with Repetitive Projects Using Composite Learning Curve Approximations

Abstract

Programs that require executing a modest number of similar projects arise in several industries, such as aerospace, construction, and defense. The scheduling of such programs often involves deciding how many projects will run simultaneously (in parallel) to “optimally” trade-off resource utilization and penalty costs. One approach for scheduling such programs is to perform a quick approximation and following it later with a “full-blown” procedure. By assuming a common learning rate for all project activities, the quick approximation, although computationally inexpensive, is not sufficiently accurate to eliminate the need for the full procedure, which requires so much data tracking and so many calculations that it is discouraging to practitioners. The approximation proposed in this paper is significantly more accurate than the quick approximation, while requiring only slightly more computation, making it more valuable for program managers.