Airports are like small cities. They never actually shut down and maintaining them takes a tremendous amount of energy. From lighting terminals and runways to powering heating, cooling and countless electronic systems, energy costs can account for as much as 10% to 15% of an airport’s operating budget.
For regional airports such as Columbia Metropolitan, those costs can have a greater impact. Every dollar spent on energy is one that cannot be invested elsewhere, whether expanding services, upgrading infrastructure or improving operations.
That’s one reason why rising junior Farid Tavakoli is taking on a project that could help airports better understand and manage their energy use. Supported by the University of South Carolina’s Magellan Scholar Grant, the computer science major is developing a machine learning system that can forecast airport energy consumption and costs, helping airport leaders make smarter decisions and potentially save money.
The project combines two of Tavakoli’s interests: aviation and artificial intelligence. He has worked in the lab of Department of Computer Science and Engineering Chair Homayoun Valafar, and when he learned there was an opportunity to collaborate with Columbia Metropolitan Airport, he immediately wanted to be involved.
“By working in his lab, I’ve become more interested and exposed to artificial intelligence and integrated machine learning,” Tavakoli says. “This is a project that has the best of both worlds.”
Today, airports use different methods to estimate energy usage, but most focus on only one piece of the puzzle. While machine learning has improved forecasting accuracy in recent years, there is no widely adopted system that brings together all the factors affecting airport energy consumption into a single forecasting tool.
Tavakoli believes that many airports simply do not prioritize detailed energy forecasting because it can be expensive and difficult to implement. His goal is to create a system that can automatically analyze data and generate useful predictions.
“The critical component is his devotion to learning,” Valafar says. “He is dedicated and committed and combined with his intelligence, he absorbs material easily. He’s going to be successful.”
The first phase of his yearlong project began in May with the development of an interactive dashboard designed for airport operations staff. The dashboard will store and organize more than a decade of data, from 2014 through 2026, and display historical energy consumption, costs, and trends. Users will also be able to compare that information with variables such as weather conditions and passenger activity.
Once the dashboard is complete, Tavakoli will build a machine learning model capable of predicting total airport energy consumption and costs up to a week in advance. The model will use weather data, passenger-related flight activity and historical energy records to generate forecasts. Those predictions will then be integrated directly into the dashboard, giving airport staff a practical planning tool.
“I want to integrate those models into the dashboard so it’s easier to use instead of just maintaining data,” Tavakoli says. “It’s also simpler because the data can be entered into the dashboard, which would handle everything itself.”
While reducing costs is an important goal, Tavakoli says the potential benefits go further. More accurate forecasts can help airports plan resources more effectively, manage strain on infrastructure and prepare for periods of high energy demand. By anticipating peak loads, airports can also coordinate more efficiently with utility providers and emergency response systems.
For the Columbia native, the project offers Tavakoli a chance to apply cutting-edge technology to a challenge that affects his local community.
“Working with a local airport to apply AI to a real-world energy problem is motivating since it connects advanced computer science techniques with meaningful community impact,” he says. “The project will enhance my undergraduate research experience by enabling hands-on work with real operational data, machine learning model development and system design.”
If successful, his work could provide Columbia Metropolitan Airport with a clearer picture of its energy future and a roadmap for operating more efficiently in the years ahead.