4th Party Logistics Problem Optimizer

Abstract

This thesis considers a pickup and delivery problem with multiple time windows, a complex cost structure and factory constraints. We formulated the problem as a mathematical model and created an instance generator based on real data. We also implemented a heuristic solution method for the problem and ran extensive statistical tests. The mathematical model shows the complexity of the problem and is implemented in AMPL to give a benchmark for the proposed solution method. The instance generator was created based on real anonymized data from a 4th party logistics (4PL) company. The proposed solution method, called the 4th Party Logis- tics Optimizer, is a meta-heuristic approach with industry specific implementations. The solution method is refined through extensive statistical experiments. The ex- periments determine which parts of the solution method have a significant positive impact on the objective value. This leads to a final composition of our solution method. The final solution method is robustly giving near optimal solutions to re- alistic sized instances in seconds, and is a powerful tool for companies facing the proposed adaptation of the pickup and delivery problem.