Analysis of manufacturing supply chains using system dynamics and multi-objective optimization
CHALLENGE - Supply chains are complex networks composed of autonomous entities where multiple performance measures at different levels, have to be taken into account; hence decision making is much more complex than treating it as a single objective optimization problem. The aim of this work is thus to introduce a methodology to address supply chain management (SCM) problems within a truly Pareto-based multi-objective optimization.
SOLUTION - This thesis proposes a new simulation-optimization framework, in which the simulation is based on system dynamics (SD) and the optimization utilizes multi-objective meta-heuristic search algorithms. In order to connect the SD and MOO software, this thesis introduces a novel SD and MOO interface application which allows the modeling and optimization applications to interact. The DOE setting used is a Latin Hypercube Design and the optimization algorithm employed is the NSGA-II algorithm available in modeFRONTIER. The methodology is evaluated through 3 case studies, in which case study 1 and 2 are based on common supply chain problems, whereas case study 3 presents a MOO of a real-world internal supply chain. All case studies incorporate the so-called stock management problem.
BENEFITS - In this study, simulation software VensimInterface is coupled with modeFRONTIER. The three case studies clearly showed that single objective optimization would not be suitable for the problems in hand as all the optimization results, in terms of the Pareto frontiers, clearly revealed the conflicts between at least two of the optimization objectives in the experiments. The optimization carried out in environment of modeFRONTIER leads to optimal results in SCM decision-making.