Üçüncü-parti bir lojistik taşıyıcısının karayolu nakliye operasyonlarının optimizasyonu

Planning of daily freight operations is one of the main challenges in transport logistics. Our study originates from a complex real-life routing problem faced by a third-party logistics (3PL) firm. Most of the 3PL firms employ a decentralized approach to ease the planning process and manage the complexity in their operations. This leads to suboptimal decisions due to the lack of integrated use of their resources as well as missed consolidation opportunities. The aim of this study is to develop a solution methodology to optimize the planning of day-to-day road transport operations of the firm in a centralized manner within an acceptable time frame. We address three novel problems related to road freight operations in this thesis. We first examine the less-than-truckload (LTL) planning problem, which involves pickup and delivery orders with deadlines and cross-docking opportunities. We assume that an unlimited number of trucks of different capacities can be hired from the spot market on a short-term basis to fulfill the LTL transportation requests. The objective is to find a minimum cost assignment of orders to feasible route-vehicle pairs while satisfying practical constraints. We propose a decomposition-based matheuristic algorithm to tackle this problem. Our solution approach demonstrates a significant reduction in the freight costs over the long term. The 3PL firm also operates a limited number of dedicated trucks which are either directly owned or hired through a long-term contract. The second problem we address deals with the assignment of dedicated vehicles to less-than-truckload (LTL) and full truckload (FTL) routes. The LTL routes obtained in the first part can be re-scheduled in this phase while the spot-hired vehicles are replaced with dedicated ones to maximize savings via an Mixed Integer Linear Programming (MILP)-based solution approach. We finally examine the last-mile delivery operations of the 3PL firm where different types of products are delivered to customers within a certain geographic region from a central depot. We model this part as a Rich Vehicle Routing Problem (RVRP) with heterogeneous vehicle fleet, time windows, selectiveness and several other practical requirements while allowing multiple trips per vehicle. We propose a parallelized hybrid evolutionary algorithm for solving the problem. The contribution of our study is twofold. On one hand, it creates actual value for the business through reducing freight costs with the proposed solution techniques. On the other hand, it introduces novel problems and solution algorithms to the optimization literature which can be extended to a wide range of industry problems in the future.