UMD-Developed AI Tool Advances Building Decarbonization and Compliance

The clock is ticking for building owners in Maryland. Beginning in 2030, buildings larger than 35,000 square feet must meet state carbon-emissions targets or face financial penalties. Help is on the way. Harnessing the power of machine learning (ML), the UMD Center for Environmental Energy Engineering (CEEE) has upgraded its custom-developed energy-auditing software to help building owners in Maryland and beyond predict energy usage and emissions.

Developed by CEEE’s Smart and Small Thermal Systems (S2TS) laboratory, in conjunction with the State of Maryland Department of General Services, the Rapid Energy Auditor (REA) software identifies areas for improvement, quickly pinpoints which buildings could benefit most from energy efficiency and decarbonization measures, and forecasts energy usage and emissions using its AI and ML algorithms that have and continue to be used on large clusters of commercial buildings.

The building sector accounts for over 40% of energy consumption and 75% of electricity consumption in the United States. Given that scale, the UMD-developed Rapid Energy Auditor software could offer a powerful tool for combating climate change and reducing energy costs.

“REA provides a transformative solution for conducting virtual energy audits and planning decarbonization strategies for large building portfolios,” says CEEE Co-Founder and Minta Martin Professor Michael Ohadi, who spearheads REA development.

Going virtual

Since 2018, S2TS has conducted onsite energy audits on over 10 million square feet of state-owned facilities in Maryland. Working with the Maryland Department of General Services’ Office of Energy and Sustainability, the S2TS team examines each building and identifies energy efficiency measures and decarbonization strategies. The effort has saved the state more than $72 million annually and reduced annual energy consumption at those facilities by 11%.

“Clearly, traditional on-site audits play an important role in compliance and decarbonization in certain circumstances,” Ohadi says, “but ML and AI-driven tools are replacing traditional time-consuming and resource-intensive methods, and their accuracy is continuously increasing.”

Ph.D. candidate Aditya Ramnarayan, who started his graduate studies in 2021, recalls a two-hour car ride to Salisbury, Maryland, for an on-site energy audit, when he first joined the audit team — only to find a facility with little room for improvement. “That wasn’t the best use of our resources,” he says. There had to be a more efficient approach.

Skyline at Baltimore Inner Harbor on a blue background with the words,

The custom-developed Rapid Energy Auditor uses AI and machine learning algorithms to quickly assess the energy efficiency of large clusters of commercial buildings and forecast their energy usage.

In 2022, the S2TS team began developing the REA software tool, Rapid Energy Auditor, to provide a preliminary audit in minutes rather than weeks. The software taps into utilities data to create a quick snapshot of energy performance. The goal isn’t to eliminate all in-person audits, but rather to provide immediate feedback and flag facilities that could benefit from a follow-up onsite assessment. Over time, the team developed a multi-criteria ranking algorithm to rate building performance based on three factors: energy-use intensity, carbon emissions per square foot and dollar-saving potential per square foot.

“The idea was to combine these three metrics to rank the facilities from the worst-performing to the best-performing ones. Then we know where to prioritize our attention and where to target, so that we can get the most savings,” says Ramnarayan, who presented a paper on the multi-criteria ranking approach at the 2025 annual conference of the American Society of Heating, Refrigerating and Air-Conditioning Engineers.

With REA, the team can review far more facilities each year, which means more savings. So far, they have conducted virtual audits on 40 million square feet of state facilities including the 10 million audited in person — achieving an estimated $25.6 million in annual energy savings. REA identifies which buildings would benefit most from a follow-up onsite audit, maximizing potential for energy savings and greenhouse gas (GHG) reductions.

Enter artificial intelligence

The state of Maryland has enacted one of the most ambitious climate plans in the country. The Climate Solutions Now Act of 2022 requires buildings larger than 35,000 square feet to achieve net-zero GHG emissions by 2040 or pay a fee. Starting in 2030, buildings must meet interim targets or face penalties.

To avoid these fees, building owners need accurate projections — which is where AI comes in. The latest version of REA uses an AI-driven analytical engine to predict energy usage and carbon emissions, and even determines related penalties. “Through machine learning, we leverage historical data along with ambient weather data, building characteristics and equipment age to predict building performance,” says Ramnarayan. “REA also calculates the cost of inaction, the fee building owners will pay if they don’t make any upgrades,” he explains. “What are the monetary repercussions? Because we all know that if you don't do anything, your building is going to perform worse with age.

“We provide the analytics. Then, the choice is in the hands of building owners. Do you want to pay for upgrades, or do you want to pay the penalties? At the end of the day, everything comes down to numbers,” Ramnarayan says.

In a study published last year in Energies, the S2TS team used machine learning to forecast energy consumption and GHG emissions on a portfolio of over 100 buildings. “The study demonstrated that robust machine learning models can generate accurate forecasts to evaluate carbon compliance and guide prompt action in decarbonizing the built environment,” says Ramnarayan, the paper’s lead author.

The predictive feature currently supports operations across 45 million square feet of state-owned buildings in Maryland and can be customized for any facility using its historical energy data and building-specific information. REA has already proven to be an effective tool for the State of Maryland, advancing its ambition to rank among the nation’s leading states in emissions reduction and energy cost efficiency.

“Beyond its impact within Maryland, REA has the potential to become a Maryland-born global energy optimization platform, serving facilities worldwide and benefiting millions,” Ohadi says.

Published March 13, 2026