Abstract
Traditional shipping logistics planning often relies on the subjective experience of planners, leading to potential inefficiencies in fleet composition and resource allocation. This paper proposes a mathematical framework to replace experience-based planning with a rigorous integer programming (IP) model. By quantifying task constraints—such as cargo dimensions, sailing conditions, and fleet formation rules—the authors demonstrate a systematic method to minimize the number of ships required for complex logistics operations while ensuring operational viability.
The Challenge of Experience-Based Planning
In large-scale logistics, determining the exact "shipping demand scheme" (which ships and how many) has traditionally relied heavily on the intuition of planners. This "gut-feeling" approach often leads to underutilized capacity or failure to meet specific mission constraints, such as speed requirements or physical hatch limits. The authors aim to solve the shipping demand calculation problem by transforming qualitative transport requirements into a quantitative model: given a specific set of goods and a specific route, what is the most efficient fleet composition that satisfies all physical and operational constraints?
Methodology: A Multi-Stage Framework
The authors propose a structured methodology to move from raw transport tasks to a finalized shipping scheme. This process involves decomposing the problem into manageable constraints and solving them through mathematical optimization.
1. Sensitivity Analysis
Factors affecting shipping are categorized into three primary buckets:
- Transport Objects: Physical attributes like weight, volume, and specialized needs (e.g., Ro/Ro ships for vehicles).
- Sailing Conditions: Route-specific limits including required speed, endurance (fuel range), and the ability to handle specific sea states.
- Formation Conditions: Rules for fleet cohesion, such as ensuring speed and tonnage differences between ships are minimal for safe maneuvering.
2. Integer Programming (IP) Modeling
The core of the framework is an IP model where the objective function is to minimize the total number of ships. Constraints include total volume, weight, area, hatch sizes, and sailing requirements. To solve this, the authors utilize a standard Branch and Bound algorithm to generate a set of "initial schemes."
3. Marshalling Adjustments
Initial mathematical solutions are subjected to a final filtering step called Marshalling Adjustment. This ensures that the resulting fleet is operationally viable by checking solutions against formation rules (e.g., ships in the fleet must have similar speeds to travel as a unit).
Case Study: USMC 2nd Expeditionary Demand
The paper validates its model using a case study based on the U.S. Marine Corps (USMC) 2nd Expeditionary demand. The input data included requirements for 380 vehicles, 15,300 personnel (~9,592 tons), and 3,958 tons of ammunition. The model drew from a dataset of 8 specific ship models (e.g., T-AKR9, T-AE-33) with varying displacement, payload, and hatch sizes.
Key Insights
- Bottleneck Identification: The authors found that for different cargo, different constraints dominate. For vehicles and personnel, loading area is the bottleneck; for ammunition, loading volume is the primary constraint.
- Physical Feasibility: The model incorporates hatch size constraints, ensuring that cargo dimensions do not exceed the physical openings of the assigned ships.
- Fleet Homogeneity: A critical finding was the necessity of the Marshalling Adjustment. While a pure mathematical model might suggest a fleet of one very fast and one very slow ship to meet capacity, the authors enforced a constraint that speed differences cannot exceed 3 knots for the fleet to remain cohesive.
Results and Optimized Schemes
After applying formation constraints (tonnage difference ≤ 2000 tons, speed difference ≤ 3 knots), the model narrowed down 12 possible combinations into two final viable programs:
- Scheme 1: 1x T-AE-26 (Ammunition) + 2x T-AKR9 (Vehicles/Personnel).
- Scheme 2: 2x T-AK294 (Ammunition) + 2x T-AKR9 (Vehicles/Personnel).
Conclusion and Future Work
The proposed model provides a foundational step toward automated, optimized shipping planning. Future work aims to incorporate more granular "particle transport material simulation" to improve packing precision and to introduce dynamic constraints such as changing weather patterns and port congestion into the optimization loop.
Summary updated