We recently had the pleasure of welcoming Dr Siddharth Arora to our lecture organized with The Oxford India Centre for Sustainable Development (OICSD). Dr Arora is a Departmental Lecturer in Management Science at Saïd Business School and Programme Director at the OICSD. His is an expert in developing statistical models for time series forecasting. Dr Arora’s research is not only academic but also highly applicable in various sectors. He has contributed to healthcare by modeling disease symptom severity using wearable technology, in energy through predicting electricity consumption with smart meters, in macroeconomics by modeling GNP, and in climate studies by comparing forecasts from General Circulation Models with statistical time series models. In this blog post, we aim to share our learnings from Dr Arora's research.
The Shift in Energy Forecasting: From High Voltage National Analysis to Low Voltage Local Focus
Meeting the Paris Agreement's goal of limiting global warming to 1.5 degrees Celsius requires an average annual investment of $3.5 trillion USD from 2016 to 2050, focusing on renewable energies like wind, solar, and nuclear power (de Coninck et al., 2018). However, a key part of these discussions also includes the often less-talked-about aspects of energy infrastructure, namely transmission, distribution, and storage. These areas are essential for the effective implementation of decarbonization strategies and have implications for the smart grid. One of the important research topics in this area is low voltage load forecasting. It marks a significant shift in energy forecasting from large-scale national or regional analysis to a more detailed focus on individual households, buildings, and streets. This approach reflects the increasing decentralization in energy production and use. Before 2008, research in this field was mainly focused on larger scales (e.g. national level or high voltage level), but there has been a notable shift towards analyzing energy needs at a more localized level (e.g. household level or low voltage level). This is especially important in the context of developments like solar panels, electric vehicles, and other low voltage technologies.
Before delving into the specifics of low voltage load forecasting, Dr Arora took a moment to examine the high voltage, national-level energy scenario in Great Britain. The pattern observed in the energy load revealed a clear intra-year seasonality. The demand peaks during the winter months of January and February due to increased heating needs, then declines around March and remains lower during the summer. As winter approaches, the demand rises again. This seasonal pattern is a typical reflection of energy usage variations throughout the year. However, there is another, intriguing trend: overall demand seemed to be decreasing, and the variability in demand was increasing. Why so? Are we consuming less energy? The answer lies in the small-scale renewable energy generation, which is more distributed. Traditional methods of measuring demand focus on how much energy passes through the transmission grid. However, with more distributed sources of generation at the local level, which often go unaccounted for in central measurements, the demand seems more variable and appears to be reducing.
Dr Arora then shared his thoughts regarding the growth in low voltage load forecasting. Firstly, the high penetration of renewables like solar, wind, and hydro has significantly influenced energy utilization and distribution. This has major implications for decarbonizing energy systems and is driving policies and financial investments towards clean energy goals. The second reason for the growth in low voltage load forecasting is digitization. This includes technologies like machine learning and smart meters, which are important for integrating new, renewable, and distributed sources of generation into the smart grid reliably. Smart meters are a key component in this evolving topic, and they not only facilitate the integration of renewable sources but also fuel research in the area by providing detailed and accurate data on energy usage at a granular level.
Smart Meter Technology: Socio-Economic Implications and Ethical Considerations
Smart electricity meters have many advantages. Smart meters not only records and reports electricity consumption to the supplier but it also provides real-time feedback on consumption and costs to the consumer through a user-friendly in-house display. A study conducted in Ireland by the Commissioner of Energy Regulation (CER) demonstrated that using smart meters, as opposed to conventional meters, led to a significant reduction in electricity usage—3.2% overall and 11.3% during peak times (Commission for Energy Regulation, 2011). This shows the potential of smart meters in managing energy sources efficiently and reducing carbon emissions. In the UK, the rollout of smart meters is estimated to cost around 13.5 billion pounds, with anticipated benefits valued at 19.5 billion pounds (2011 prices), according to the Department of Energy, Security and Net Zero. The rollout is expected to reduce 45 million tons of CO2 equivalent emissions in the UK.
Dr. Arora then shared that there are 6.5 million fuel-poor households in the UK (National Energy Action, 2024). Fuel poverty is defined as a situation where a household has to spend more than 10% of its income on energy to maintain an adequate level of warmth in the home. Dr. Arora then realized that about 90% of households in the lowest income bracket use prepayment methods for their energy bills (Waddams Price, 2002). However, these pay-as-you-go models are more costly to operate and the unit prices for energy are typically higher compared to those on credit plans. This is a paradox: why would these fuel-poor households choose a more expensive payment method? The answer lies in the perceived control these households have over their energy consumption. However, Dr. Arora asks, if we can forecast a household's energy consumption for the coming week or month, can we also convert that consumption forecast into a cost forecast?
In the last portion of his talk, Dr. Arora raised some concerns alongside the benefits of smart meter data. He presented an average intraday profile of electricity consumption. For instance, one user’s consumption spiked from 8am, peaking during office hours, and tapering off by early evening, with minimal activity during weekends. Another profile showed consistent usage throughout the week, including nights, hinting at possible work-from-home scenarios. Dr. Arora pointed out that smart meter data can now be decoded to identify which appliances are used and when. This level of detail can reveal daily routines and behaviors within a household and can raise significant ethical and privacy concerns. This creates critical questions about AI and ethics. On the smart meter front, addressing privacy concerns, the potential for hacking, data safety, and the need for transparent regulations are important. The ability to make inferences from smart meter data is not just a technological advancement but also a challenge that requires careful consideration of the ethical implications and the establishment of robust safeguards to protect consumer privacy and data security.
References
Commission for Energy Regulation (2011). Electricity smart metering customer behaviour trials findings report. Dublin: Commission for Energy Regulation.
de Coninck, H., Revi, A., Babiker, M., Bertoldi, P., Buckeridge, M., Cartwright, A., ... & Wollenberg, L. (2018). Strengthening and implementing the global response. ISO 690
National Energy Action. (2024, January 4). What is fuel poverty?. National Energy Action (NEA). https://www.nea.org.uk/what-is-fuel-poverty/#:~:text=We%20estimate%20that%20there%20are,on%20the%20UK%20energy%20crisis
Waddams Price, C. (2002). Prepayment meters: the consumer perspective, Energy Action, 86, 14-15.
Yunus Isik, VP (Academics), Chief Editor
Comments