Local search heuristics for pollution-routing problem with multiple vehicle types and deadlines


Thesis Type: Postgraduate

Institution Of The Thesis: Orta Doğu Teknik Üniversitesi, Faculty of Engineering, Department of Industrial Engineering, Turkey

Approval Date: 2013

Student: ONUR CAN SAKA

Supervisor: SİNAN GÜREL

Abstract:

Vehicle Routing Problem (VRP) is one of the most widely studied problems in logistics literature. Up to now, many different types of exact solution methods and heuristics have been developed in order to deal with various variants of this computationally complex optimization problem. However, only a few researchers have included the concepts of speed control, fuel consumption and greenhouse gas (GHG) emissions in their studies. The first part of this study is dedicated to a special variant of VRP called the Pollution-Routing Problem (PRP), which includes a comprehensive cost function that takes into account fuel consumption, GHG emissions and driver wages. An extension of PRP incorporating multiple vehicle types and deadlines is considered. Throughout this part, firstly two alternative exact solution methods are proposed: a Mixed Integer Programming model with a piecewise linear cost function and a Mixed Integer Second Order Cone Programming model, followed by local search heuristics with a special initialization algorithm and optimal travel time determination procedure. Results of experiments are interpreted in an extensive computational study. In the second part (See Appendix E), the report of an applied project is represented. The project took place in a third-party logistics (3PL) company in the Netherlands with the aim of investigating the possible improvements that can be achieved via employing a multi-depot and automated planning approach.