Small profit margins and cost pressure in the road transportation business require an efficient planning of operations on the part of trucking companies. Complicating aspects such as legal restrictions, specific customer requirements, or the necessity of responding quickly to unforeseen events disqualify manual planning entirely, and even off-the-shelf route planning software can be inappropriate for setting up both feasible and cost-effective transportation plans in a certain business context. Vendors of sophisticated, customized decision support systems that provide support on different planning levels are confronted with the task of tailoring embedded vehicle routing algorithms to the specific demands and business rules of their clients. For such purposes we proposed the concept and design of a flexible metaheuristic framework to facilitate and accelerate the development of solution methods for a wide range of rich vehicle routing problems (VPR) arising in diverse industries. Part I provided an introduction into the field of VRPs and metaheuristic solution methods. Our framework is based on solution methods for the standard VRP and consequently, we started with defining this basic variant of the VRP class. Then, we presented an overview of real-world extensions to the VRP - the class of rich VRPs. Partly, this overview served as an illustration of the relevance of flexible and adaptable solution approaches, but it also served as the foundation of our requirements analysis in the second part of the thesis. We introduced common metaheuristics from the literature and explained those local search and large neighborhood search techniques in detail upon which our framework is based. Part II was dedicated to the framework itself. Reviewing existing publications on VRP frameworks we first documented that there is a lack of flexible approaches that explicitly address the solution of wide ranges of VRPs. We formulated requirements for a good framework, including the four attributes flexibility, simplicity, accuracy, and speed, as well as characteristics of rich VRPs that it should be able to take account of. Aiming to meet these requirements we presented the architecture and main concepts of our framework, mainly on a pseudocode level: we defined the five implemented standard VRP heuristics and the mechanisms available for adapting these heuristics to rich VRPs. The central aspect is the separation of standard VRP operations and specific VRP operations by user-definable checking and post-processing functions. Part III finally dealt with the evaluation of our framework, mainly regarding flexibility, accuracy, and speed, by developing customized metaheuristics for five rich VRPs.
A flexible metaheuristic framework for solving rich vehicle routing problems
Wirtschaftsinformatik und Operations-Research ; 17 ; 1-185
2011
185 Seiten, Bilder, Tabellen, 111 Quellen
Hochschulschrift
Englisch
General Metaheuristic Algorithm for a Set of Rich Vehicle Routing Problems
Transportation Research Record | 2016
|Solving Practical Vehicle Routing Problem with Time Windows Using Metaheuristic Algorithms
DOAJ | 2012
|Solving Full-Vehicle-Mode Vehicle Routing Problems Using ACO
British Library Conference Proceedings | 2018
|Ant Metaheuristic with Adapted Personalities for the Vehicle Routing Problem
British Library Conference Proceedings | 2015
|A unified-adaptive large neighborhood search metaheuristic for periodic location-routing problems
Online Contents | 2016
|