Modeling and Methods for Container Transport Operation = 컨테이너 운송 최적화를 위한 모형 및 해법 연구

다국어 초록 (Multilingual Abstract)

As maritime trade growth is anticipated to decelerate to around 2% per year until 2030, container terminals—responsible for over 80% of worldwide seaborne traffic—are redirecting their operational emphasis from capacity augmentation to the efficient usage of current resources. In this evolving landscape, terminal competitiveness increasingly hinges on the proper operation of existing infrastructure and equipment, rendering operational decision-making under resource restrictions increasingly vital. Bottlenecks in container flows resulting from waterside activities associated with boats and landside operations related to inland transportation have emerged as critical factors limiting overall terminal performance.
This dissertation addresses two major operational decision problems to alleviate these bottlenecks: quay crane scheduling on the waterside and drayage truck routing on the landside. On the waterside, the focus is on dual-spreader quay cranes, which are increasingly adopted to improve handling speeds for large container vessels. While dual-spreader cranes can lift two containers simultaneously and increase potential quay productivity, actual operations require consideration of mode changeovers and weight constraints, making scheduling far more complex than for conventional single-spreader cranes. In particular, computational complexity increases as the number of combinations of feasible tandem-lift container pairs increases.
Previous studies have addressed this problem using sequence-based mixed-integer programming formulations that provide precise representations. However, for large vessels with hundreds of containers, model size grows exponentially, limiting the ability to obtain solutions within practical time frames. To overcome this limitation, this study proposes a phase-based approach that assigns containers to phases defined by mode changeovers rather than individual sequence positions. The proposed approach is proven theoretically equivalent to sequence-based models, thereby preserving both representational accuracy and solution optimality. Building on this foundation, a two-stage solution method is developed that first determines lift types and phases, and then constructs detailed sequences within each phase. Computational experiments demonstrate efficiency gains over state-of-the-art logic-based Benders decomposition, solving large bay problems with up to 500 containers in under 12 seconds.
On the landside, this study addresses truck routing under separation-mode drayage operations, which are used to enhance truck utilization flexibility. In separation mode, trucks can decouple from trailers during loading and unloading at customer sites, enabling more efficient routing. However, this also introduces compound challenges involving truck–trailer synchronization in time and space, along with empty container management. Despite these complexities, most existing studies either assume empty containers are always available or pre-plan empty container flows separately.
To overcome these limitations, this study proposes an integrated model that treats empty container sourcing and repositioning as endogenous decision variables. The proposed model explicitly captures shipping-line compatibility constraints and dual-transaction requirements within a multi-terminal environment. Recognizing that branch-and-bound approaches become increasingly difficult as problem size grows, an Adaptive Large Neighborhood Search (ALNS) algorithm is also designed to efficiently solve large-scale instances. Computational experiments show that the MIP formulation fails to find optimal solutions for instances with 36 or more requests within one hour, whereas the proposed ALNS solves instances with up to 180 requests in under 17 minutes, serving an average of 4.6 additional requests on the largest instances compared to the MIP.
The results of this dissertation contribute to faster and more cost-effective operational decision-making for terminal operators and trucking companies across both waterside and landside domains. The phase-based crane scheduling approach provides quantitative foundations for rapidly comparing and evaluating alternative operational scenarios in higher-level decisions such as crane assignment and berth allocation. The integrated approach to landside drayage, by explicitly incorporating realistic operational constraints including empty container availability and repositioning, demonstrates potential as an analytical tool to support dispatchers in day-to-day decision-making.

목차 (Table of Contents)

• Abstract i
• List of Figures vii
• List of Tables viii
• 1 Introduction 1
• 1.1 Background and Motivation 1
• 1.2 Research Challenges and Gaps 5
• 1.2.1 Dual-Spreader Crane Scheduling 5
• 1.2.2 Container Drayage with Separation Mode 6
• 1.3 Research Objectives and Contributions 7
• 1.3.1 Research Objectives 7
• 1.3.2 Contributions for the DSCSP 8
• 1.3.3 Contributions for the CDOP-SEP 8
• 1.3.4 Dissertation Organization 9
• 2 Tandem Quay Crane Operation 10
• 2.1 Introduction 10
• 2.2 Literature Review 13
• 2.2.1 Quay Crane Scheduling Problem 13
• 2.2.2 Evolution Towards Multi-Spreader Crane Technology 14
• 2.2.3 Dual-Spreader Crane Scheduling Problem (DSCSP) 15
• 2.2.4 Summary and Research Gap 16
• iii
• 2.3 Problem Description 16
• 2.3.1 Ship-Bay Structure and Crane Operations 17
• 2.3.2 Tandem-Lift Feasibility 18
• 2.3.3 Operational Phases and Time Components 18
• 2.3.4 Problem Objective 20
• 2.3.5 Key Assumptions 20
• 2.4 Mathematical Formulation 20
• 2.4.1 Notation 20
• 2.4.2 MIP Formulation 22
• 2.4.3 Phase Upper Bound 26
• 2.5 Solution Methodology 27
• 2.5.1 Equivalence with Sequence-Based Formulation 27
• 2.5.2 Two-Stage Solution Approach 31
• 2.6 Computational Experiment 32
• 2.6.1 Experimental Setup 32
• 2.6.2 Test Instances 33
• 2.6.3 Computational Results 34
• 2.7 Conclusion 38
• 2.7.1 Summary 38
• 2.7.2 Practical Implications 38
• 2.7.3 Future Research Directions 39
• 2.7.4 Closing Remarks 39
• 3 Container Drayage Operation 40
• 3.1 Introduction 40
• 3.2 Literature Review 43
• 3.2.1 Drayage Operations: From Stay-with to Separation Mode 43
• 3.2.2 Empty Container Management 46
• 3.2.3 Adaptive Large Neighborhood Search 47
• 3.2.4 Research Positioning 48
• 3.3 Problem Description 48
• iv
• 3.3.1 Operational Environment 48
• 3.3.2 Customer Orders and Request Structure 49
• 3.3.3 Operational Constraints and Policies 51
• 3.3.4 Problem Objectives 52
• 3.4 Mathematical Formulation 52
• 3.4.1 Task-Based Network Framework 54
• 3.4.2 MIP Formulation 58
• 3.4.3 Computational Complexity and Solution Approach 60
• 3.5 Adaptive Large Neighborhood Search 60
• 3.5.1 Initial Solution Construction 61
• 3.5.2 Destruction Operators 65
• 3.5.3 Repair Operators 70
• 3.5.4 Simulated Annealing Acceptance with Served-Request Priority 76
• 3.5.5 Adaptive Weight-Based Operator Selection 78
• 3.5.6 Reheat Mechanism for Stagnation Recovery 78
• 3.5.7 Overall ALNS Framework 79
• 3.6 Computational Experiments 80
• 3.6.1 Instance Generation 81
• 3.6.2 Parameter Setting 87
• 3.6.3 Computational Results 88
• 3.7 Conclusion 94
• 3.7.1 Contributions 94
• 3.7.2 Future Research Directions 95
• 4 Conclusion and Future Work 96
• 4.1 Summary 96
• 4.2 Future Work 97
• Appendix 100
• A DSCSP Computational Results 101
• B CDOP-SEP Computational Results 110
• v
• References 120
• Acknowledgement 125
• 국문초록 126
• vi