Vehicle routing problem with pickup and delivery and with time windows

Abstract

This work proposes a metaheuristic method for solving Vehicle Routing Problem with Pickup and Delivery with Time Windows (VRPPDTW). The mathematical model which is developed to solve this problem tries to fulfill a set of transportation requests within time intervals of related nodes of their request by a heterogeneous vehicle fleet based at multiple depots so as to minimize the total travel cost. A mixed integer model is developed for this problem, and a Simulated Annealing algorithm (SA) is designed as the solution procedure which employs two types of moves, namely Request Exchange and Merge. The algorithm is experimented on a set of randomly generated instances. The instances are also solved by CPLEX solver within GAMS suite v22.0. The results are compared. It is observed that CPLEX could find solutions for only small-size instances in reasonable computing time. The algorithm developed in this thesis is capable of solving large size problems in a reasonable amount of time. Then the SA algorithm is applied to Boğaziçi University (BU) Personnel Transportation System to determine the optimal schedule for transporting groups of personnel from specific pickup locations in different parts of the Metropolitan city of İstanbul to the delivery location, i.e. the main university campus in Bebek.