Article Preview
TopIntroduction
Maritime transportation has been evolving at a rapid pace for decades, and the sector is directly impacted by global concerns. Safety, security, and environmental management have been at the forefront of these developments and difficulties. By imposing new laws or regulations, the maritime authorities aim to create a safety culture as well as clean waters for shipping. On the other hand, optimization solutions for stakeholders ensure safe, secure, and cost-effective maritime transportation choices.
One of the research areas of port management is the expected time of arrival time prediction (Dobrkovic et al., 2016; Meijer, 2017; Yu et al., 2018). Although voyage optimization has been a maritime topic for decades, port operations optimization is also a popular topic (Cammin et al., 2020). Vessels could transport cargo efficiently from one port to another if cargo handling time was minimized. Furthermore, optimized cargo operations and voyages result in a lower carbon footprint, more environmentally friendly transportation, and increased profits.
This research aims to analyse container port operations using historical data in this manner. The study dataset was gathered from several types of container vessels that conducted one or more visits to the sample container terminal in Turkey. Four mobile gantry cranes and two berthing quays are available at the sample container port. Because of its geographic position, the port's operation is influenced by a variety of conditions, including humid, hot, and occasionally windy weather.
The objective of this study is to investigate several machine learning algorithms for analysing container handling time based on operational differences including crane properties, container cargo condition, container weight, operation types (loading, discharging, or shifting), and weather conditions. In summary, the study's goals are as follows:
- •
Providing an overview of potential weather effects on container handling operations.
- •
Providing high-accuracy machine learning techniques that could be used to analyse container port operations with historical data.
- •
Exploring the key factors that could cause a delay in cargo operation time.
- •
Exploring possible solutions to reduce vessel port stay time.
Multiple Linear Regression (MLR), Ridge Regression, LASSO Regression, Principal Components Regression (PCR), Partial Least Squares Regression (PLS), Support Vector Regression (SVR), Random Forest Regression (RF), and Multilayer Perceptron Regression (MLP) methods are used in this paper to determine the effects of weather conditions. Furthermore, Fuzzy C-Means clustering (FCM) methods are used to analyse vessels’ port of calls in terms of operational benchmarks, such as cargo completion time, cargo handling capacity, and vessel dimensions.
The rest of the paper is organized as follows: previous studies in the literature are discussed in the second part. The methodology of the investigation is presented in Section 3. The container terminal case study is examined in the fourth part. The findings are evaluated, and the discussion is given in Section 5. Finally, conclusions are discussed.