Multiobjective Γ-Robust Port Service-Capacity Planning Under Demand Uncertainty: Evidence from Chilean Vessel Movements
- 1Departamento de Ingeniería Industrial, Universidad Tecnológica Metropolitana, Santiago 7800002, Chile
- 2Departamento de Ingeniería Industrial y de Sistemas, Facultad de Ingeniería, Universidad de Tarapacá, Arica 1010069, Chile
Abstract
Port authorities must commit service capacity before monthly vessel demand is known. This study develops a multiobjective linear programming model based on Bertsimas–Sim Γ-robustness that jointly determines lower-cost regular capacity and higher-cost surge capacity. Monthly vessel-call uncertainty is represented by a budgeted uncertainty set, while an independent full-stress constraint imposes a minimum service level when all modeled upward deviations occur. A lexicographic ε-constraint procedure generates Pareto-optimal cost–shortage policies, and a balanced policy is selected by distance to the ideal point. The model is evaluated using 2020 data from 28 Chilean maritime authorities and validated with 2018–2022 data. At Γ = 0.10, the balanced policy allocates 647.0 regular and 207.5 surge units and achieves 98.41% full-stress service, with CostIdx 21% above the minimum-cost policy. The 15% deviation matches the historical year-ahead bound, and Γ = 0.10 is the smallest tested budget meeting the 95% out-of-sample target in three of four transitions. Full-stress reliability is the main CostIdx driver, whereas the base-capacity quantile mainly determines the capacity mix. Among alternatives meeting the same full-stress standard, the proposed model has the lowest CostIdx without requiring an assumed demand distribution.
Related
- Time to AC-t: When AC-transmission changes capacity expansion planning
- LLM-Guided Transportation Hub Capacity Planning with Textual Business Inputs
- Solving Multiobjective Linear Set Optimization Problems
- Optimality and Exact Moment-SDP Formulations for Robust Two-Stage Convex Polynomial Multiobjective Optimization
- Expected Hypervolume Maximization for Multiobjective Optimization under Uncertainties
BibTeX
@article{duran2026multiobjective,
title = {{Multiobjective Γ-Robust Port Service-Capacity Planning Under Demand Uncertainty: Evidence from Chilean Vessel Movements}},
author = {Claudia Durán and Amir Karbassi Yazdi and Sebastián Bastías},
journal = {Mathematics},
year = {2026},
month = oct,
doi = {10.3390/math14213970},
url = {https://www.mdpi.com/2227-7390/14/21/3970},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Multiobjective Γ-Robust Port Service-Capacity Planning Under Demand Uncertainty: Evidence from Chilean Vessel Movements
AU - Claudia Durán
AU - Amir Karbassi Yazdi
AU - Sebastián Bastías
PY - 2026
DA - 2026/10/30
JO - Mathematics
DO - 10.3390/math14213970
UR - https://www.mdpi.com/2227-7390/14/21/3970
AB - Port authorities must commit service capacity before monthly vessel demand is known. This study develops a multiobjective linear programming model based on Bertsimas–Sim Γ-robustness that jointly determines lower-cost regular capacity and higher-cost surge capacity. Monthly vessel-call uncertainty is represented by a budgeted uncertainty set, while an independent full-stress constraint imposes a minimum service level when all modeled upward deviations occur. A lexicographic ε-constraint procedure generates Pareto-optimal cost–shortage policies, and a balanced policy is selected by distance to the ideal point. The model is evaluated using 2020 data from 28 Chilean maritime authorities and validated with 2018–2022 data. At Γ = 0.10, the balanced policy allocates 647.0 regular and 207.5 surge units and achieves 98.41% full-stress service, with CostIdx 21% above the minimum-cost policy. The 15% deviation matches the historical year-ahead bound, and Γ = 0.10 is the smallest tested budget meeting the 95% out-of-sample target in three of four transitions. Full-stress reliability is the main CostIdx driver, whereas the base-capacity quantile mainly determines the capacity mix. Among alternatives meeting the same full-stress standard, the proposed model has the lowest CostIdx without requiring an assumed demand distribution.
SN - 2227-7390
ER -