Comparative study of Vehicle Routing Problem’ solving methods

Abstract

The decision-making process in distribution operations has a significant impact on minimizing distance, time, and costs, making it a crucial topic in the field of logistics. Efficient vehicle routing is essential in scenarios such as school bus routes, pizza delivery, and distribution of goods. The Vehicle Routing Problem (VRP) represents the core challenge we aim to address and solve. Understanding the complexities and nuances of VRP allows us to develop strategies and methodologies to optimize routing decisions and achieve minimal distances, reduced time, and lowered costs. The objective of this dissertation is to conduct a comprehensive study of methods to solve the Vehicle Routing Problem (VRP) and evaluate their effectiveness and suitability. In this dissertation, we employ a mixed-methods approach. Firstly, we conduct a liter ature review to understand the definition and characteristics of VRP and gather existing methodologies. We then develop a methodology by creating a dataset and modifying existing codes. These modifications aim to address specific VRP requirements and opti mize route planning, resource utilization, and cost efficiency. The proposed methodology serves as a practical approach to solving VRP. For the experimental part, we evaluate the effectiveness of ACO, NN algorithm, and PABCW using benchmark datasets provided by [9] and a real-world case. We compare the solutions obtained by these algorithms across different instances and evaluate their performance. The comparative study reveals that the modified version of the k-Nearest Neigh bors (kNN) algorithm achieves the highest accuracy percentage among NN, ACO, and PABCW. The evaluation based on benchmark datasets and a real-world case demon strates the effectiveness of the modified kNN algorithm in solving the VRP. However, further evaluation and validation are necessary to confirm its effectiveness and robustness in real-world applications, considering potential cost savings and improved operational efficiency. This dissertation provides valuable insights into the performance and suitability of different algorithms for addressing the Capacitated Vehicle Routing Problem. The study highlights the significance of efficient vehicle routing in minimizing distance, time, and costs in distribution operations. The developed methodology, based on a modified kNN algorithm, demonstrates promising results in solving the VRP. Future research should III focus on further evaluation and validation of the algorithm in real-world applications, considering the potential for cost savings and improved operational efficiency.