Route planning is a key focus in operations research, with the standard vehicle routing problem (VRP) serving as the classical formulation. Practical scenarios introduce additional complexities, leading to various rich VRPs. Numerous solution algorithms, particularly metaheuristics, have been developed for these variants. However, routing software must align with specific business rules to effectively support decision-making, and literature on customizing VRP heuristics is scarce. This thesis addresses this gap by presenting a flexible framework for developing custom metaheuristics applicable to a wide range of rich VRPs. The initial section offers background on vehicle routing and metaheuristic methods, emphasizing local search and large neighborhood search techniques that form the basis of the proposed framework. The main section details the framework's design, highlighting its flexibility, simplicity, accuracy, and speed, while promoting structured development and code reuse. It includes a library of established heuristics and mechanisms for adapting them to specific VRPs. The final part illustrates the framework's application through customizations for five rich VRPs, showcasing adaptations in data structures and neighborhood search methods. Computational results demonstrate the framework's effectiveness and efficiency, highlighting the robustness of heuristics for practical use.
