Projects

SUMMER 2026 PROJECTS

Project 1: A Data-Driven Digital Twin for Container Dwell Time Prediction in Terminal Yard Operations

Background: This project aims to develop a data-driven digital twin framework for predicting container dwell time in a terminal yard. Container dwell time refers to the amount of time a container remains inside the terminal from arrival until departure. Accurate dwell time prediction can help terminal operators improve yard planning, reduce congestion, support better stacking decisions, and reduce unnecessary container reshuffling.

The proposed digital twin will represent the terminal yard as a dynamic system where container status, vessel information, cargo characteristics, yard location, weather conditions, and operational timestamps are continuously updated. Using historical terminal data, machine learning models will be developed to predict the remaining dwell time of containers and identify containers that are likely to stay in the yard for a long period. These predictions can provide early warning signals for yard planners and help them make better decisions before congestion or inefficient stacking occurs.

The digital twin component will connect the prediction model with a periodically updated representation of the yard. As new container information becomes available, the digital twin can update the predicted dwell time and provide decision-support insights. This creates a practical framework for proactive yard management rather than reactive problem-solving.

Project Team: Hasini Balasuriya, Kelly Gorman, and Rachel Black

Student Activities: The IRES students will leverage generative AI techniques and digital twins to improve Container Dwell Time Prediction in Terminal Yard Operations. The students will create digital twins that predict the remaining dwell time of containers and identify containers that are likely to stay in the yard for a long period.  

Project 2: Risk-Sensitive Digital Twin Architecture for Quay Crane Assignment

Background: Quay cranes are massive gantry cranes that load and unload container ships in ports, moving along a shared rail. Their routing is bounded by the shared rail and a safety distance. In the Eurogate terminal at the port of Hamburg, a given quay crane makes around 20-25 moves per hour, and a large ship may require around 5000 total moves during its service time. While cranes may be assigned in advance with long time horizons, plans can quickly fall apart when plans do not match real-world states. In the face of delays, outages, or other mismatches between plans and reality, quay crane assignment systems must be set up to keep operations on track and minimize downtime. 

This research focuses on creating dynamic, Digital Twin-based systems for quay crane reassignment that can respond to disruptions in real time. Digital Twins are granular representations of real-world systems, constantly exchanging real-time data to track and respond to changes, allowing for more flexible decision-making. To this end, the project investigates the use of exact solvers, heuristics, and metaheuristics within a fast-and-slow optimization approach, incorporating time constraints and risk management in decision-making to develop an assignment system that can handle the noise and idiosyncrasies of realistic environments without unnecessary downtime, slack, or inefficiency.

Project Team: Evelyn Ray, Steven Fung, and Hasini Balasuriya

Student Activities: The student will develop a Digital Twin framework for quay crane assignment, collect and analyze real-time operational data, design and implement exact, heuristic, and metaheuristic optimization methods, and evaluate risk-sensitive reassignment strategies to improve system resilience and minimize downtime under disruptions.

Project 3: A Digital Twin for Public Transport Route Optimization in Urban Mobility Systems

Background: This project aims to develop a digital twin for modeling and optimizing public transport routes in a city. Public transport systems are dynamic urban networks where passenger demand, traffic conditions, travel times, and route performance change throughout the day. A digital twin can help represent this system in a computational environment and support better decision-making for route planning, passenger movement, and system management.

The project is designed in two main phases. In the first phase, the team will develop a graph-based representation of a city, where nodes represent important locations such as bus stops, neighborhoods, transfer points, or activity centers, and edges represent travel connections between them. Population flow over time will be used to mimic passenger demand across the city. Using this graph structure, an optimization problem will be formulated to identify bus routes that move the maximum number of people through the city while minimizing total travel time or system delay.

The second phase of the project will expand the graph-based optimization model into a digital twin architecture. This digital twin will connect the city network model, passenger-flow data, routing decisions, and system performance measures into a continuously updated framework. 

Project Team: Evelyn Ray and Avi Shekhar

Student Activities: The student will develop a graph-based model of the city, analyze passenger demand and traffic data, formulate and solve route optimization problems, and build a Digital Twin framework that continuously updates system information to improve public transport efficiency and reduce travel time and delays.