A Recursive Optimization and Simulation Approach to Analysis with an Application to Transportation Systems

Abstract

Modeling large systems with either an optimization or a simulation method has several disadvantages. Simulation is usually expensive if adequate detail and experimental designs are employed. Complete detail in optimization models may press the bounds of computability. A recursive approach integrating both types of models is presented. The recursive approach involves, an allocation of resources by optimization models at an aggregate level. At this level computation is not difficult and broad alternatives can be easily explored. Simulation models can then be designed to address detailed questions of productivity of resources, discreteness, and complex relationships. The simulation can use the particular schedules generated by the optimization so that experimental designs can be limited in size. The revised productivities can then be input to the optimization model for a more refined optimal solution. The recursive approach has been applied to the strategic mobility system problem in reaching decisions for the size of transportation forces for the DOD. A linear programming model provides optimal allocation of overall mobility system vehicles and schedules. Two simulation models, an airlift model and a sealift model, provide productivity estimates and tests of capability.