Vehicle Routing with Time Window for Regional Network Services - Practical Modelling Approach Iman Niroomand 1 , Amir H. Khataie 2 ,Masoud Rahiminezhad Galankashi 3 1 Department of Network Design, Canada Post, Ottawa, Ontario, Canada 2 Department of Network Design, Canada Post, Ottawa, Ontario, Canada 3 Department of Material, Manufacturing and Industrial Engineering, UniversitiTeknologi Malaysia, Johor, Malaysia Abstract- Vehicle Routing Problem with Time Window (VRPTW) has demonstrated an excessive application in postal service network design. In this paper we elaborate on how VRPTW modelling approach can benefit both firm and contractors. The business objective is to facilitate the process of retendering the contracts. Accordingly, the novel modeling approach has been developed in order to improve the practicality of the results. The Ant Colony Optimization (ACO) is used for solving the VRPTW for regional post office services. Keywords-Vehicle Routing Problem, Postal Service, Ant Colony Optimization I. INTRODUCTION Vehicle Routing Problem (VRP) is a combinatorial optimization problem that has been investigated for many years by scholars. VRP has extensive application in supporting decision making process in transportation, distribution, and logistics domains. The problemcan be defined based on different type and size of vehicles and it could vary from air to ground services. The basic VRP includes a distribution hub or center and a set of customers who needs to be visited. Vehicle starts the trip from the base (distribution center) to delivers committed itemsto each customer [1]. After the last visit, the vehicle must be repositioned at the base for the next day trip. Vehicle routing has several applications and can be customized for different type of transportation and logistics problems. Vehicle Routing with Time Windows (VRPTW) is auniversal problem in postal delivery service. In VRPTW, vehicle must visit each customer within the defined time window and must return to its based at a certain time [2].In order to solve VRPTW in real world not only the timing of visiting customer is important but also many other restrictions are imposed from both service provider and customers. These issues need to be fulfilled in order to enhance the practicality of the suggested solution [3]. In postal service distribution/collection network,each Postal Office (PO) might have different business physical and operational characteristics. For example, each PO has a different processing shifts, facility restrictions (e.g. high or low dock, parking limit),equipment limits (single stack vs. double stack possibility), and different road condition. Each of these restrictions can impact the truck type, truck size, and truck loading profile. Moreover, the suggested loop needs to respect certain legal and contractual obligations such as, the maximum routing length, minimum or maximum number of stops per route, and layover time between collection and delivery process. The VRP main algorithms could be divided to three categories: Exact, Heuristics and Meta Heuristics [5]. Exact algorithms mainly use branch and cut method. This type of algorithm requires asophisticated mathematical programming unless the problem is defined on small size (i.e. Hong Kong post office [6]) or the algorithm is used at strategic level.Such a problem can’t be solved easily at tactical or operational level. In addition to that and in a real worldmore assumptions are imposed by operation which makes the problem more complicatedto model or rarely ended in successful solving [7].Therefore, developing a comprehensive mathematical model which answers all the design aspects is almost impossible and such solution is neither cheap nor quick. Furthermore, the model needs to be modified time to time based on new rules and regulations.In reality and for successful implementation, many scenarios must be verified with operators and need to be approved by different shareholders either at operation level or at upper management level. In this matter, VRPTW model needs to be validated through different what-if scenario analysis and should also be solved during reasonable computational time. In this paper, the ACO algorithm is applied and modified for VRPTW problem. The criteria of developing such an algorithm are clearly discussed in order to have reliable operational network. In the end, the solution is compared with existing regional network to explain how the new route construction would benefit the postal corporation in retendering process. II. PRACTICAL ROUTESUSING META- HEURISTIC ALGORITHM In general, there are two criteria to evaluate the superiority of a model;computational time vs. the solution accuracy. Computational time refers merely to methodology of solving the problem and quality of the solution refers to simplicity of implementation and flexibility of adapting new terms and constraints [4]. 978-1-4799-6410-9/14/$31.00 ©2014 IEEE