Article Preview
TopIntroduction
Planning for energy availability is critical in this era of rapid global economic and technological growth and climate change. Energy use is predicted to increase by approximately 15% and 70% by the year 2050 for OECD and non-OECD countries respectively (U.S. EIA, 2019). Researchers estimate the cost to construct infrastructure to support peak load increases of 18% in the U.S. would be approximately 180 billion dollars, assuming current infrastructure pricing (Auffhammer, Baylis, & Hausman, 2017). Nevertheless, the main issue is not even whether we will have enough fossil fuel, but whether we can afford to produce and burn the fossil fuel we have (Alexander, 2015). Experts warn the World will pass the crucial 2’C global warming limit and the carbon pledges from 147 nations to the Paris Climate Summit are not enough to stop such a temperature rise. As such, finding ways to not only plan for but also reduce future energy consumption are key.
While the industrial and transportation sectors will continue to consume the most energy globally, residential consumption also drives increased energy demand (U.S. EIA, 2019). Attitudinal and behavioral changes that reduce energy consumption could create needed ‘quick wins’ while longer-term systematic infrastructure upgrades are designed and implemented to address anticipated increased energy loads.
At the same time, data collection tools available for understanding current energy utilization have advanced. The use of smart meters enables the real-time data collection and more accurate billing, and facilitates unique insight on energy and water consumption behavior (Bhattarai et al., 2022; Ratner et al. 2022; Bhardwaj et al. 2022; Gohil et al. 2021; Dhanwani et al. 2021; Monks et al. 2021; Kumar et al. 2021; Ahuja and Khosla, 2019; Cango et al., 2018; Chaudhari and Mulay, 2021; Khosrojerdi et al., 2021; Kuhe and Bisu, 2020; Sharma and Singh 2015; Siepermann et al., 2021; Rathod and Garg, 2017; Agwu 2021; Baccarne & De Marez, 2021; Croci & Molteni 2021).
While Canada and the U.S. led the way in smart meter installation (30-40% current usage across North America), Europe is estimated to have around 72% installation in 2020 (IoT Analytics, 2020). This has led to the availability of large volumes of smart-meter generated data. For this reason, there is value in complementing traditional theory-driven research in energy consumption with research that uses machine learning techniques, or more broadly by artificial intelligence (AI)-driven approaches. Prior to the smart-meter era, studies tended to examine electricity consumption behavior using traditional theory-driven research methods, incorporating a limited number variables. The practical implications of the findings were predominantly tailored towards suppliers and businesses rather than residents.