Pro-active Dynamic Vehicle Routing

Transcription

1 Francesco Ferrucci Pro-active Dynamic Vehicle Routing Real-Time Control and Request-Forecasting Approaches to Improve Customer Service Physica-Verlag A Springer Company

2 Introduction Motivation Categorization of the Considered Problem Aims and Contributions of the Book Central Research Questions to Be Answered in This Book Contributions of This Book Outline of the Book 13 Introduction to Tour Planning: Vehicle Routing and Related Problems General Task and Definitions Representation of Requests and General Types of Routing Problems Node-Based Routing Problems The Traveling Salesman Problem (TSP) The Vehicle Routing Problem (VRP and CVRP) The General Pickup and Delivery Problem (GPDP) and Related Variants Complexity of Node-Based Routing Problems Solution Methods for Node-Based Routing Problems Extensions to the Vehicle Routing Problem Time Windows Backhauls Simultaneous Pickup and Delivery Multi-Depot Problems Open Routing Problems : Vehicle Scheduling Multiple Compartments Multiple Trips Time-Dependent Travel Times Legal Driving Time Regulations Heterogeneous Fleet Objective Functions in Vehicle Routing Problems 36

3 xviii Contents 2.5 Information Revelation in Routing Problems Revelation of Relevant Information in Vehicle Routing Problems Approaches in the Literature for Characterizing Relevant Information and Distinguishing Between Static and Dynamic Routing Problems A Unified Approach for Classifying Routing Problems with Regard to Characteristics and Processing of Relevant Information Dynamic Routing Problems Which Utilize a Centralized Coordination Typical Objectives in Dynamic Routing Problems Dynamic Events in Dynamic Routing Problems Technologies Utilized in Real-Time Control Approaches The Degree of Dynamism Three-Echelon Classification of Dynamic Routing Problems Evaluating the Performance of Solution Approaches for Dynamic Routing Problems A General Classification Scheme for Routing Problems Summary 79 3 The Considered RDOPG Applications Problem Description of RDOPG Applications Existence of a Real Urban Road Network Presence of Vehicle On-Board Units and a Central Dispatching Center Availability and Utilization of Past Request Information Allowing Vehicle En-Route Diversion Integration of Vehicle Scheduling Decisions Characteristics of RDOPG Applications Differences from Other Dynamic Routing Problems Computational Complexity Classifying RDOPG Applications with Regard to Characteristics and Processing of Relevant Information General Classification of RDOPG Applications Summary 87 4 Review of the Literature Related to the Considered RDOPG Applications First Papers on Dynamic Routing Problems in the Literature Selected Reactive Real-Time Control Approaches for Dynamic Routing Problems Strategies for Increasing Flexibility in Dynamic Routing Problems Waiting Strategies Relocation Strategies Request Assignment Strategies 100

4 4.4 Flexibility in Dynamic Routing Without Stochastic Knowledge Flexibility in Dynamic Routing Using Stochastic Knowledge Strategy-Oriented Approaches Approaches Using Exact Solution Methods Stochastic Modeling Based Approaches Approaches with Manually Provided Stochastic Knowledge Sampling-Based Approaches Related Objective Functions Other Relevant Factors for RDOPG Applications En-Route Diversion of Vehicles Road Network Simulator Real-Time Control City Logistics-Related Decision Support Systems and Communication Technologies Conclusion of the Literature Review 145 A New Deterministic Real-Time Control Approach for RDOPG Applications General Categorization of Different Control Concepts in Logistic and Production Systems The Deterministic Real-Time Control Approach The Applied Real-Time Control Concept Update Handling Vehicle Scheduling Strategy Idle Vehicle Waiting Strategy The Mathematical Model of Individual Problem Instances Road Network and the Derived Digraph The Deterministic Dynamic Mathematical Model Characterization of the Deterministic Real-Time Control Approach Utilization of the Real-Time Control Approach in RDOPG Applications Summary 165 A New Forecasting Approach for Generating Stochastic Knowledge from Past Request Information and Utilizing the Stochastic Knowledge Integrating Stochastic Knowledge into the Proposed Real-Time Control Approach ' Generation of Dummy Customers Using Segment-Based Clustering Segment-Based Cluster Generation Cluster Selection and Dummy Customer Generation Handling Dummy Customers in the Pro-Active Real-Time Control Approach Dynamic Parameter Updating During the Transportation Process 179

5 6.3.2 Extended Vehicle Scheduling Strategy with Regard to Dummy Customers Extended Vehicle Waiting Strategy Using Dummy Customer Information Summary 183 The Proposed Tabu Search Solution Method General Concepts of the Implemented Solution Method The Tabu Search Metaheuristic Description of an Iteration in the Solution Method Representation of a Tour Plan in the Solution Method Efficient Calculation of Changes in the Objective Function Value Neighborhood Operators Within Tour Insertion (WTI) Relocate (REL) MultiRelocate (MREL) Large Neighborhood Search (LNS) Exchange Between Tours (XBT) The Implemented Tabu List Construction of the Initial Tour Plan Stage-Based Neighborhood Operator Selection Scheme Integration of the Solution Method into the Real-Time Control Approaches Summary 200 Computational Results The Discrete Event-Based Simulator Simulation of the Requests Simulation of the Vehicles Test Environment, Parameter Values, and Request Data Classes System Environment and Real-Time Control Parameter Values Tabu Search Solution Method Parameter Values Dummy Customer Request Parameter Values Request Parameter Values and Request Data Classes Number of Conducted Computational Experiments Performance Evaluation Using the Additionally Attainable Improvement Performance of the Tabu Search Solution Method Performance Evaluation of Deterministic Real-Time Control Approaches Performance Evaluation of Allowing Vehicle En-Route Diversion Measuring Structural Distortion Using the Goodness-of-Fit Test Results of Generating Stochastic Knowledge for S REAL and 5 GEN. 231

6 xxi 8.8 Performance Evaluation of the Pro-Active Real-Time Control Approach Results on S^^ Results on S GEN The Degree of Structural Diversity Driver Inconvenience Aspects in the Real-Time Control Approaches Summary and Answers to the Three Central Research Questions Summary and Outlook on Future Work Summary Outlook on Future Work 276 Appendix A Additional Facts Related to Chap Appendix B Additional Information on Request Data Sets of Request Data Class S GEN 285 B.I Parameters for Generating Selected Request Data Sets of S GEN B.2 Visualization of All Request Data Sets of S GEN 288 Appendix C Request Response Times of the Examined Request Data Classes 289 C.I Request Response Times of S REAL 289 C.2 Request Response Times of S GEN 291 About the Author 301 References 303 Index 317