Multi-Objective Optimization For Environmentally Friendly Logistics Network

Congratulations to Dr Irina Harris who has recently achieved her PhD. Her research was supervised by Dr Christine Mumford of the School's Distributed and Scientific Computing research group, and Professor Mohamed Naim of the Cardiff Business School

Thesis Abstract

Traditionally, infrastructure modelling of logistics network design is driven by a need to reduce costs. However, many real-world cases may involve dealing with multiple and sometimes conflicting objectives, especially when climate change and environmental concerns have been increasingly discussed worldwide. In this thesis we devise and investigate a multi-objective evolutionary optimization framework together with Lagrangian Relaxation to solve a large size Facility Location Problem (FLP) where 'green issues' (CO2) and traditional objectives are solved simultaneously, offering the decision maker a choice of trade-off solutions. Lack of benchmark data for multi-objective FLP with environmental objectives created initial difficulties in our research. However, the opportunity to work with a leading UK supermarket supply chain provided a good basis for generating large artificial data sets and to test our techniques with a good range of parameter setting. The analysis of the research indicates that more facilities could be desirable to reduce the environmental impact and that it is possible to offer the decision maker good compromise solutions.

Two variants of the FLP are considered during the investigation for building multi-objective decision tools: the uncapacitated and the capacitated. Additionally, we investigate the optimization of a single source assignment problem as part of our collaborative work with industry. In this way we explore exact and heuristic approaches based on cost optimization as well as considering the environmental impact from vehicles in the two-objective approach on the models with realistic constraints. The trade-off solutions demonstrate to the decision maker how a small increase in cost could equate to a considerable decrease in the distance travelled by the vehicle.