Team proposes AI-powered approach to establishing a ‘carbon-neutral energy city’

A joint research team has developed key technologies to realize “Urban Electrification” using artificial intelligence (AI). Their findings have been published in the journal Sustainable Cities and Society. The team includes researchers from the Renewable Energy System Laboratory and the Energy ICT Research Department at the Korea Institute of Energy Research (KIER)

Urban electrification aims to reduce the use of fossil fuels and introduce renewable energy sources, such as building-integrated solar technology, to transform urban energy systems. While this concept is relatively unfamiliar in the Republic of Korea, it is being promoted as a key strategy in the U.S. and Europe for achieving carbon neutrality and creating sustainable urban environments.

In traditional urban models, energy supply can be easily adjusted using fossil fuels to meet electricity demand. However, in electrified cities, the high dependence on renewable energy leads to greater variability in energy supply due to weather changes. This causes mismatches in electricity demand across buildings and makes the stable operation of the power grid more challenging.

In particular, Low-Probability High-Impact Events (LPHI), such as sudden cold snaps or extreme heat waves, can cause a sharp increase in energy demand while limiting energy production. These events pose a significant threat to the stability of the urban power grid, potentially leading to large-scale blackouts.

The research team developed an energy management algorithm based on AI analysis to address power grid stability issues and implemented it into a system. The demonstration of the developed system showed an 18% reduction in electricity costs compared to conventional methods.

The research team first used AI to analyze energy consumption patterns by building type and renewable energy production patterns. They also unraveled how complex variables, such as weather, human behavior patterns, and the scale and operational status of renewable energy facilities, affect the power grid.

Notably, they discovered that Low-Probability High-Impact Events, which occur on average only 1.7 days per year (around 0.5% of the time), have a decisive impact on the overall stability of the power grid and its operational costs.

The analyzed content is developed into an algorithm and a system. The developed algorithm optimizes energy sharing between buildings and effectively manages peak demand and peak energy production. In addition to maintaining daily energy balance, the system is designed to respond to Low-Probability High-Impact Events, ensuring the stability of the power grid even in extreme situations.

When the developed system was applied to a community-scale real-world environment replicating urban electrification, it achieved an energy self-sufficiency rate of 38% and a self-consumption rate of 58%. This is a significant improvement compared to the 20% self-sufficiency and 30% self-consumption rate of buildings without the system. This application also resulted in an 18% reduction in electricity costs and greatly improved the stability of the power grid.

Particularly, the annual energy consumption applied in the demonstration was 107 megawatt-hours (MWh), which is seven times larger than simulation-based studies conducted by leading international institutions. This significantly enhances the potential for applying the system in real urban environments.

Dr. Gwangwoo Han, the lead author of the paper and a researcher at the Energy ICT Research Department, stated, “The results of this study demonstrate that AI can enhance the efficiency of urban electrification and address power grid stability issues, while also highlighting the importance of managing Low-Probability High-Impact Events.”

He further predicted that “by applying this system to various urban environments in the future, we can improve energy efficiency and enhance grid stability, ultimately making a significant contribution to achieving carbon neutrality.”