WCEN - Technology integration in maritime and ports operations

Demand forecasting in third-party logistics (3PL) networks is increasingly complex due to the heterogeneity of data and variability of systems protocols, requiring accurate, cost-effective, and privacy-preserving solutions. Federated Learning (FL) with Domain-Adversarial Neural Networks (DANN) enables 3PL warehouses and customers to collaborate on demand forecasting without sharing raw data. However, the cost-effectiveness of integrating FL with digital twin platforms under zero trust security remains uncertain. Compared to traditional models, this study develops a cost-benefit framework to assess FL-based forecasting through the DANN method. The results show that FL reduces forecasting errors, saving total inventory costs. While the initial computational costs are high, medium and long-term gains include high order fulfillment rates and logistics efficiency, thus increasing revenues. Integrating digital twins improves forecasting adaptability despite adding security and operational complexity. A contribution-weighted model is proposed to ensure fair benefit distribution, which rewards participants based on contributed dataset quality. Findings highlight that FL-based forecasting can enhance decision-making, resilience, and cost efficiency in 3PL networks and the potential for other complex supply chain networks like maritime.

In the logistics sector, the ability to obtain real-time information about cargo location and condition presents a significant opportunity to differentiate containerized transport services, which are typically considered a commodity. In a period marked by continuous disruptions, the use of smart containers could increase client's willingness to pay more for a transport service that offers higher levels of safety, integrity and security. This study explores the key characteristics relevant to different operators involved in the container supply chain, examining available technologies on the market and how liner carriers are investing in this segment. The integration of IoT sensors and technologies enables real-time monitoring of cargo conditions, improving transport reliability and providing greater transparency. The paper also looks at how this evolution could change customer expectations and pricing models, fostering the rise of a market more focused on services quality.

Maritime digitalization presents unique challenges at the intersection of physical and digital systems, particularly in critical operations like heavy-lift vessel cranes. This study examines how maritime crane failures affect vessel operations, environmental impact, and user acceptance of digital systems. By integrating Moravec's Paradox and Polanyi's Paradox with the Unified Theory of Acceptance and Use of Technology (UTAUT), we address theoretical gaps in understanding technology acceptance in physical-digital interfaces. The research employs a mixed-methods approach to analyze operational data from maritime vessels, contributing to both UTAUT extension and practical maritime operations management.

Battery-powered automated guided vehicles play an important role in achieving environmental sustainability by minimizing emissions and noise pollution for automated container terminals. Nevertheless, the limited driving range and long recharging time of battery-powered automated guided vehicles can pose obstacles to terminal performance. How to select a reasonable charging mode and optimally solve operation scheduling problem to maintain high terminal performance is prominent for sustainable port operation. In this paper, en-route charging mode is considered, and a mixed integer programming model for this complicated scheduling problem is constructed, which aims to reduce terminal cost and enhance terminal efficiency. An adaptive large neighborhood algorithm is adopted to solve practical-sized instances. Numerical experiments are conducted to identify the value of en-route charging mode to automated container terminals when compared to off-route charging mode.