SIMULATION-BASED OPTIMIZATION FOR RESOURCE ALLOCATION AT TPL SYSTEMS

Allocating resource at TPL systems differs significantly from traditional private logistics systems. The resources are considered commodities sold to customers of different types. Total yield suffers when over-allocate to lower-rate or price-sensitive customers; but the resource become “spoiled” when reserve too much for full-rate or time-sensitive customers that do not arrive as expected. Uncertain order characteristics make the optimization of such decisions very hard, if not impossible. In this paper we proposed a simulation-based optimization to address related issues. A genetic algorithm based optimization module is developed to generate/search good solutions; and a discrete-event simulation model is created to evaluate the solutions generated. The two modules are integrated to work in evolutionary cycles to achieve the optimization. The study also compared GA/Simulation model with more traditional approach such as response surface methodology via designed experiments. The models were validated through experimental analysis