AI Enhances Building Efficiency and Cuts Emissions

By Dr Silpaja Chandrasekar, PhDReviewed by Susha Cheriyedath, M.Sc.Jul 29 2024

In a paper published in the journal Nature Communications, researchers evaluated artificial intelligence's (AI) potential in the building sector, focusing on medium office buildings in the United States (US). They developed a methodology to assess emissions reductions, identifying key areas such as equipment, occupancy influence, control and operation, and design and construction.

Six scenarios were used to estimate energy and emissions savings across climate zones. The study showed that AI could reduce energy consumption and carbon emissions by a significant amount by 2050 and, when combined with energy policy and low-carbon power generation, by even greater percentages.

Background

Past work has shown that climate change poses significant challenges, with governments setting ambitious targets to reduce energy consumption and carbon emissions. With rapid urbanization, the global building stock is projected to double by 2060, highlighting the critical need for enhanced energy efficiency in buildings. AI has potential in various domains, including building design and operation, but its full impact on energy savings and emissions reduction in buildings still needs to be clarified.

Building Energy Optimization

The study assumed that buildings' technical energy-saving potential is the difference between a median US office and a zero-energy verified building. Four key categories—imperfect design/construction, subpar controls/operation, occupancy influence, and equipment efficiency—were identified to account for this potential.

The sum of these four categories describes the total technical energy saving. Annual building energy simulations using the US Department of Energy (DOE's) Energyplus tool and literature reviews helped quantify each category's saving potential.

The study utilized the American Society of Heating, Refrigeration, and Air-conditioning Engineers (ASHRAE)standard 90.1 prototype model, focusing on a medium office building. It represents various building types and is well-calibrated based on real data. The selected medium office building has three floors and a total floor area of 4980 square meters, divided into core and perimeter zones. A packaged air conditioning unit was assumed for cooling and heating.

EnergyPlus models were developed for design variations due to the four key categories. These models considered different energy potentials from equipment and building design/construction. Additionally, discrete choice models were used to estimate the market share of varying building types over time, factoring in the net present value (NPV) of adoption, which includes construction and development costs.

The study utilized the American Society of Heating, Refrigeration, and Air-conditioning Engineers (ASHRAE) standard 90.1 prototype model, focusing on a medium office building. Cost premiums for high energy efficiency buildings (HEEBs) and net zero energy buildings (NZEBs) were expected to decline over time due to autonomous or policy-induced changes, further reduced by AI utilization.

AI Boosts Efficiency

Understanding AI's impact on energy consumption involves recognizing the average building energy efficiency and current best practices. The 2012 Commercial Buildings Energy Consumption Survey (CBECS) by the Energy Information Administration (EIA) found that office buildings were the most common commercial structures. These buildings exhibited the highest electricity consumption. Their median energy use intensity (EUI) was recorded at 167 kWh/m². A review of 67 low-energy verified buildings showed a median EUI of 57 kWh/m², indicating significant energy-saving potential. The US DOE's study modeled a prototype medium office building based on ASHRAE standard 90.1 to assess energy-saving potentials through design, construction, equipment, and operation variations.

The Pacific Northwest National Laboratory (PNNL) analyzed the impact of control measures on energy savings, revealing a typical total energy saving from controls for a medium office of 27.2%. Evaluations of equipment efficiency improvements showed energy savings between 11.5% and 17.3%, while design and construction changes offered savings from 5.9% to 9.1%. These evaluations included increased efficiency in heating, cooling, lighting, and equipment power density and variations in building orientation, insulation, and window-to-wall ratio. 

To achieve carbon neutrality by 2050, AI and policy-driven scenarios promoting HEEBs and NZEBs could significantly reduce energy consumption and CO2 emissions. AI adoption could cut energy use by ~21% compared to the Frozen scenario and ~8% compared to BAU by 2050, while AI combined with policy measures could reduce CO2 emissions by ~40% compared to BAU and ~60% compared to Frozen scenarios, achieving near-zero emissions with low-emission power generation.

Summary

To sum up, AI demonstrated significant promise for enhancing building energy efficiency and reducing carbon emissions. By focusing on key areas such as equipment, occupancy influence, control, operation, and design, AI lowered cost premiums and increased the penetration of high energy efficiency and net-zero buildings.

Projections showed that AI could reduce energy consumption and carbon emissions by 8% to 19% by 2050. AI's impact grew significantly along with supportive energy policies and low-carbon power sources. It led to a 40% decrease in energy consumption and a 90% reduction in carbon emissions compared to standard business-as-usual scenarios.