Technical Summary: Optimizing Container Loading and Unloading through Dual-Cycling and Dockyard Rehandle Reduction

1. Problem Statement

The paper addresses the NP-hard problem of optimizing container handling operations at seaports. Specifically, it targets the critical interdependence between two distinct but coupled operational challenges:

1. Quay Crane Dual-Cycling (QCDC): Maximizing the number of simultaneous loading and unloading cycles to reduce vessel turnaround time. In a standard Single Cycle (SC), a crane unloads a stack and returns empty before loading. In Dual Cycling (DC), the crane unloads a stack and immediately loads a prepared container, eliminating the empty return trip.
2. Dockyard Rehandle Minimization: Reducing the number of times containers must be relocated (rehandled) in the yard to access a target container that is not at the top of its stack.

The Core Conflict: The paper identifies that these two objectives are often optimized in isolation in existing literature. An unloading sequence designed to maximize DCs may inadvertently demand containers from the dockyard in an order that forces excessive rehandles. Conversely, a dockyard layout optimized for minimal rehandles may not align with the unloading sequence required to maximize DCs. The authors argue that treating these as separate, sequential stages creates a "feedback gap" where optimizing one metric degrades the other.

2. Methodology: QCDC-DR-GA

To resolve this conflict, the authors propose QCDC-DR-GA (Quay Crane Dual Cycle – Dockyard Rehandle Genetic Algorithm), a hybrid Genetic Algorithm (GA) that performs joint optimization of both the unloading sequence and the dockyard configuration within a single search loop.

2.1. Mathematical Formulation

The problem is modeled as a two-machine flow-shop scheduling problem. The objective is to minimize the total completion time ($T$), defined as a weighted sum of:

• $w_s$: The number of Single Cycles.
• $w_d$: The number of Dual Cycles.
• $R$: The total number of dockyard rehandles.

$$T = \alpha w_s + \beta w_d + \gamma R$$

Where $\alpha$, $\beta$, and $\gamma$ represent the time costs (in seconds) for single cycles, dual cycles, and rehandles, respectively.

2.2. Hybrid Genetic Algorithm Structure

The algorithm employs a mixed chromosome representation to handle the distinct nature of the two decision variables:

• Component A1 (1D Chromosome): Encodes the unloading sequence of ship stacks (a permutation of stack indices).
• Component A2 (2D Chromosome): Encodes the dockyard container arrangement (a matrix representing stack positions and tiers for containers to be loaded).

2.3. Specialized Genetic Operators

To maintain feasibility and diversity, the GA uses tailored operators for each chromosome component:

• Crossover:
  • 1D (Unloading): Uses Two-Point Crossover with a repair mechanism to ensure valid permutations (no duplicate stack indices).
  • 2D (Dockyard): Uses 2D Substring Crossover, inspired by aircraft gate assignment problems. This involves row-wise and column-wise exchanges followed by a repair step to ensure every container is assigned exactly once and stack height limits are respected.
• Mutation:
  • 1D: Uses Swap Mutation (exchanging two genes).
  • 2D: Uses 2D Two-Point Swapping Mutation, exchanging genes at specific row/column coordinates to introduce spatial diversity in the yard layout.
• Selection & Termination: The algorithm uses Roulette Wheel Selection with an Elite Preservation strategy (20% of the population). Termination occurs after a maximum number of generations or if the best solution remains unchanged for 100 consecutive iterations.

2.4. Operational Assumptions

The model assumes:

• Containers for loading are available at the dockside upon request.
• Operations are conducted row-wise (dual cycling is restricted to containers within the same ship row).
• Dockyard rehandles follow a "nearest-lowest stack" strategy for displaced containers.
• No external delays (e.g., vehicle unavailability) are considered.

3. Key Contributions

The paper outlines seven specific contributions:

1. Empirical Validation: It establishes a correlation between the unloading sequence of stacks and the occurrence of dockyard rehandles, specifically within dual-cycling strategies.
2. Integrated Framework: It develops a comprehensive algorithm that jointly optimizes dockyard and QCDC operations, moving beyond sequential or isolated optimization.
3. Novel Hybrid GA: It introduces a specific GA architecture capable of handling mixed 1D and 2D chromosome structures simultaneously.
4. Specialized Operators: It proposes tailored crossover and mutation operators (2D Substring Crossover and 2D Swapping Mutation) designed specifically for container handling constraints.
5. Parameter Analysis: It conducts an extensive analysis to identify the most influential computational parameters (e.g., population size, mutation rates) within the GA framework.
6. Methodological Rationale: It provides clear justifications for the selection of specific strategies at various stages of the optimization process.
7. Statistical Benchmarking: It validates the proposed strategy against four state-of-the-art algorithms using rigorous statistical paired t-tests.

4. Experimental Results

The authors evaluated QCDC-DR-GA across six scenarios involving small, medium, and large container ships (ranging from 5 to 30 stacks per row). The algorithm was compared against:

• Greedy Upper Bound: A heuristic maximizing DCs but ignoring rehandles.
• QCDC-bi-level-GA: Optimizes QCDC but treats rehandles separately.
• GA-ILSRS (Scenario 1): Optimizes DCs but ignores rehandles.
• GA-ILSRS (Scenario 2): Optimizes rehandles but uses Single Cycling.

Performance Metrics

• Total Operation Time: QCDC-DR-GA reduced total operation time by up to 30.1% compared to existing methods. The average improvement across large-ship scenarios was 20.1%.
• Statistical Significance: Using a two-tailed paired t-test ($\alpha = 0.05$), the proposed method showed statistically significant improvements in 20 out of 24 pairwise comparisons.
• Scaling: The performance advantage of the joint optimization approach grew monotonically with ship size. For large ships (30 stacks), improvements over single-objective baselines reached 20.1%, whereas for small ships, the advantage was smaller but still present.
• Ablation Study: A controlled experiment where the dockyard plan (A2) was "frozen" (not optimized) demonstrated that while the unloading sequence optimization alone provides a competitive baseline, the co-optimization of the dockyard layout provides a significant additional gain, particularly in reducing rehandles (up to 98% reduction in small scenarios).

5. Significance and Claims

The paper claims that its primary significance lies in demonstrating the inefficiency of isolated optimization in port operations. By treating the unloading sequence and dockyard layout as a single, coupled decision problem, the algorithm resolves the trade-off between maximizing crane productivity (DCs) and minimizing yard inefficiencies (rehandles).

• Practical Impact: The approach offers a cost-effective solution for ports to decrease vessel turnaround times without requiring new infrastructure or advanced hardware investments. It is designed to be integrated into existing Terminal Operating Systems (TOS) as a pre-arrival planning module.
• Theoretical Contribution: The paper formalizes the joint optimization of QCDC and dockyard rehandles as a unified NP-hard combinatorial problem with a search space of $S! \times N!$ (where $S$ is the number of stacks and $N$ is the number of containers), arguing that previous "integrated" approaches were merely sequential rather than truly joint.
• Scalability: The authors assert that the row-wise decomposition of the problem allows the algorithm to scale to large vessels (e.g., 2,000+ containers) by solving independent subproblems for each row, making it feasible for real-world deployment on standard hardware.

The paper concludes that while it relies on simplifying assumptions (e.g., immediate container availability, no dynamic disruptions), the proposed QCDC-DR-GA represents a significant step toward holistic port optimization, offering a robust method to enhance resource utilization and operational efficiency.