Title Berth Template Design Problem with Stochastic Ship Arrival Times
Year 2021
Degree Master
School Department of Transportation and Logistics Management,National Chiao Tung University
Author Chi-Yang Lin
Summary        Berths are the most crucial resource of a container terminal, and the associated planning
decisions play an important role in terminal efficiency. Given ever-increasing demand at
terminals, this study focuses on the berth template problem (BTP), a tactical (midterm) design
problem to support terminal operators for the contract negotiation processes with shipping
lines and the facilitation of short-term berth assignment operations. With the objective of
minimizing the cost of lost revenue and operational penalty, the terminal operator determines
the berth assignment and the service sequencing of the potential calling ships on a cyclical
basis with respect to a fixed length of planning horizon (e.g., a week). Meanwhile, the operator
has the flexibility of denying a ship at the price of a ship-dependent revenue loss or including
it in the template.
      In particular, in order to consider ship arrival uncertainty, this study will focus on the
approach of the two-stage stochastic programming (SP) model, for which the berth assignment
and sequencing decisions are determined in the first stage of the SP model. After the realization
of uncertain ship arrivals, the exact berthing time for each ship is made as the re-course
decision in the second stage, and the associated deviation with respect to the arrival time is
modeled as an operational penalty to take into account the negative impact on shipping lines.
For this difficult stochastic optimization problem, A Genetic Algorithm is designed to generate
approximate solution to improve the efficiency of the solution in this study.
      Based on the numerical experiments, the total operating cost and the risk of the original
plan be interrupted can be efficiently decreased after considering ship arrival uncertainty. In
addition, the quality gap of the GA solution is only 2% with respect to the optimal solution,
and the computational time is also relatively short.
Hashtags
View count: 125