Characterizing stochastic and dynamic forward/closed-loop supply chains by using simulation optimization

A supply chain network, due to its nonlinear, stochastic, time-dependent nature, and presence of complex interactions between supply chain members, can become quite challenging to optimize and requires a more complex model. At this point, simulation optimization gains a better understanding of the complex and messy phenomenon of supply chain problems. In this thesis, two types of supply chain problem are solved by using simulation optimization. The first one is a forward supply chain problem (FSCP), and the second one is a closed-loop supply chain problem (CLSCP). Simulation optimization consists of two-phase, namely (1) optimization phase, and (2) simulation phase. Optimization phase tries to find out answers for both strategic and operational decisions. Inventory control system parameters and strategic decisions (selecting a supplier, opening a plant etc.) are determined by using genetic algorithm. Then, determined decision variables are evaluated by using simulation model. Moreover, extensive statistical analysis is given to enhance the understanding of system behavior. Computational results showed that not only cost functions but also other performance measures (average service levels etc.) can be improved by using simulation optimization. Proposed simulation optimization provides a remarkable contribution to the uncertain, dynamic and complex real-world supply chain problems.