APPLICATION OF MIXED INTEGER LINEAR PROGRAMMING MODEL IN OPTIMIZING THE EXPORT WOOD TRANSPORTATION NETWORK FOR SMALL AND MEDIUM ENTERPRISES IN HANOI

N. MinhInternational Journal of Apllied Mathematics 2025 · DOI

Abstract

This study builds upon and develops the Mixed Integer Linear Programming (MILP) model for multi-objective transportation network optimization, aiming to evaluate the practical applicability in the wood export activities of small and medium-sized enterprises (SMEs) in the Hanoi area. The input data includes the locations of production facilities, collection points, export ports, consumption markets, transportation cost-time matrices, and storage limits based on the characteristics of wood products. The model is solved using CPLEX software with three scenarios: (1) cost optimization, (2) time optimization, and (3) multi-objective optimization. The results show a clear trade-off between transportation cost and time. Specifically, with a weight range of 0.6 to 0.75, the composite index Q achieves a low and stable value, indicating a reasonable balance between the two objectives. The model demonstrates high potential for application in designing logistics networks suitable for the scale, business strategy, and practical conditions of SMEs in the wood export sector.

Introduction

Vietnam is a global leader in wood processing, with export turnovers exceeding $15 billion. However, Small and Medium Enterprises (SMEs) in this sector face a significant "logistics tax." In Vietnam, logistics costs account for roughly 16% of GDP, significantly higher than the global average of 10-12%. For wood SMEs in the Hanoi area, the challenges are twofold: they lack the economies of scale enjoyed by larger firms, and their transportation networks are often organized in an ad hoc, fragmented manner. These businesses must navigate a complex web of production facilities, distribution centers, and international ports while satisfying international markets that demand both speed and strict traceability. This paper seeks to replace experience-based management with a rigorous mathematical framework to optimize these networks.

Methodology

The researchers utilized Mixed Integer Linear Programming (MILP) to find the best outcome in a system with complex constraints and both continuous and discrete variables. Key aspects of the approach include:

  • Multi-objective Optimization: Simultaneously evaluating total cost and total transportation time.
  • Scenario Analysis: Testing under three conditions: (1) Pure cost optimization, (2) Pure time optimization, and (3) Multi-objective optimization using Weighted Sum and Minimax methods.
  • Solver Technology: Problems were solved using IBM ILOG CPLEX, a high-performance industrial-grade mathematical programming solver.
  • Decision Variables: The model accounts for truck transport (15 tons) to distribution centers, 40’DC trucks to ports, and maritime shipping to international markets.

Network Configuration and Data

The study is grounded in real-world data provided by a wood export enterprise in Hanoi. The network configuration includes 12 production facilities, 5 distribution centers, and 2 main export ports serving 4 key international markets.

Summary of network components, demand, and constraints used in the MILP model.

The model accounts for a total export demand of 100 TEU (Twenty-foot Equivalent Units) and incorporates real-world parameters such as a 240-hour storage limit and specific cost-time matrices between all nodes.

Key Results and Analysis

The research explicitly maps the "price of speed," showing that prioritizing time reduces delivery duration by about 18.4% but increases system costs by 13.9%. A critical finding is the identification of an optimal weight range for multi-objective optimization. When the cost weight is set between 0.6 and 0.75, the composite index $Q$ (representing deviation from ideal values) becomes stable and low, suggesting a strategy that slightly favors cost over time provides the most efficient network for most SMEs.

Optimized route metrics showing cost and time performance for specific export markets.

Under the cost optimization scenario, the model achieved a minimum total cost of 5.137 billion VND, representing an estimated 7–12% reduction in logistics expenses compared to unoptimized structures. In contrast, the time-optimized scenario reached an average delivery time of 56.8 hours, significantly lower than the 70+ hours seen in cost-focused routes.

Conclusion

The MILP model demonstrates that SMEs can achieve high container fill rates and choose infrastructure-efficient routes previously overlooked by manual planning. Future work will focus on developing Dynamic MILP models to account for real-time traffic and port delays, as well as incorporating supply chain risk factors like fuel price volatility and shifting international carbon emission standards.

Summary updated

Abstract

This study builds upon and develops the Mixed Integer Linear Programming (MILP) model for multi-objective transportation network optimization, aiming to evaluate the practical applicability in the wood export activities of small and medium-sized enterprises (SMEs) in the Hanoi area. The input data includes the locations of production facilities, collection points, export ports, consumption markets, transportation cost-time matrices, and storage limits based on the characteristics of wood products. The model is solved using CPLEX software with three scenarios: (1) cost optimization, (2) time optimization, and (3) multi-objective optimization. The results show a clear trade-off between transportation cost and time. Specifically, with a weight range of 0.6 to 0.75, the composite index Q achieves a low and stable value, indicating a reasonable balance between the two objectives. The model demonstrates high potential for application in designing logistics networks suitable for the scale, business strategy, and practical conditions of SMEs in the wood export sector.
BibTeX
@article{minh2025application,
  author = {N. Minh},
  title = {{APPLICATION OF MIXED INTEGER LINEAR PROGRAMMING MODEL IN OPTIMIZING THE EXPORT WOOD TRANSPORTATION NETWORK FOR SMALL AND MEDIUM ENTERPRISES IN HANOI}},
  journal = {International Journal of Apllied Mathematics},
  year = {2025},
  doi = {10.12732/ijam.v38i10s.1080},
}
Figure