Artificial intelligence, energy or biosciences
it’s telling that those were the three fields that Bill Gates advised college graduates to enter into when asked in May 2017. His opinion lends credence to the assumption made in this article, namely that renewable energy, artificial intelligence (AI) and machine learning (ML) are not only intricately linked, but that ‘smart’ renewable energy must be considered the most sustainable way forward for our energy needs.
AI defines business today. And it will increasingly continue to do so. Along with allied technologies such as ML, deep learning and advanced neural networks, AI has transformative potential for the global energy sector. This article will focus on three key issues with regard to AI/ML and renewable energy:
1. How machine learning is already transforming the renewables market;
2. AI and energy consumption considerations and; 3. AI and grid efficiency optimization.
An important caveat regarding the fossil fuel-renewable energy dichotomy will also be discussed at article’s conclusion.
Machine Learning and Renewables
ML is already entrenched in our everyday lives, from smart phone assistants like Apple’s Siri or Samsung’s Bixby, to voice and image recognition systems. More important even will be ML’s ability to assist in tackling some of the world’s most pressing physical and logistical problems, including that pertaining to the full potential of renewable energy.
This nexus between ML and renewable energy is already being harnessed by Germany, for example, which is using an ML-based early warning system that analyzes real-time data from wind turbines and solar panels across Germany in order to extrapolate the amount of predicted energy that the country will need over a two-day period.
In a similar vein, researchers at the Massachusetts Institute of Technology (MIT) have developed an ML system that can more quickly predict variations in wind speeds over a given period. This can help utility companies and renewable energy start-ups to more quickly pinpoint potential locations for wind farms. This data collection typically takes up to 12 months for a utilities company to gather. ML systems such as those developed by MIT can produce models based on just three months of data, which is considerably more accurate than conventional data-gathering processes.
The National Center for Atmospheric Research is also assisting utility companies to decide when to use wind energy by providing AI-determined forecasts, particularly in Colorado, a state in which wind generation has doubled since 2009. It is able to provide the same for solar energy forecasts.
IBM has developed its own ML system that can assist renewable energy companies in forecasting weather patterns. Named Self-Learning Weather Model and Renewable Energy Forecasting Technology (SMT), it analyzes data acquired from 1,600 weather stations, solar plants, wind farms, and weather satellites. SMT’s weather forecasts are up to 30% more accurate than those compiled by the National Weather Service. This is invaluable in helping to predict renewable energy capabilities for up to weeks in advance.
Google has trumpeted its own use of ML systems for its data centers, which are notoriously energy-hungry. Google has calculated that using its DeepMind ML system in its data centers has allowed the company to reduce the amount of energy used for cooling by up to 40%. The company estimates that the company has become 3.5 times more energy efficient in just five years, much of it thanks to initiatives such as DeepMind.
The ability of ML to assist data and IT companies in dramatically reducing their energy consumption cannot be underestimated. A study released in 2016 by the U.S. Department of Energy’s Lawrence Berkeley National Laboratory (the ‘Berkeley Lab’) titled the “United States Data Center Energy Usage Report” showed that data center energy usage in the IT industry in the period 2009-2014 had remained almost flat despite a huge growth in computing-related demand during said period.
Sarah Smith of the Berkeley Lab stated how the team found that one of the main reasons for these dramatic improvements in data center energy efficiency was that of the sharing of resources or “virtualization” fueled by the surge of cloud services. Smith commented how, “The cloud is certainly one of the drivers to more energy efficiency…Energy is one of the easier things to optimize.” ML and the cloud are already symbiotically linked.
Energy Consumption Considerations
Energy consumption in terms of financial and socio-environmental costs continues to be one of the leading mega risks for corporations and national economies alike. Hence the growing interest in renewable energy. There is an inevitability about it, as Kyle Harrison, a New York-based analyst at Bloomberg New Energy Finance, points out: “Companies are declaring 100 percent renewable energy targets. This will drive new-build regardless of government subsidies. Sustainability is a primary driver.”
The huge potential of AI/ML for renewable energy is not only relevant regarding improving weather prediction capabilities, as already discussed, but also regarding energy consumption factors generally, including energy storage potential. However, an important consideration is that one of the current biggest ‘drawbacks’ regarding renewable energy uptake in the global energy market is that of unpredictability. Ashiss Dash of Infosys makes this point when stating, “Among the leading sources of renewables, solar and wind are dependent on the weather, are unpredictable, and we still need to smoothen [sic] the flow of energy from generation to consumption.”
A solution must be in the ability of renewable utility companies to store excess energy whenever possible. This is where demand-side flexibility must be used intelligently, whereby renewable energy inputs can be maximized according to demand-side peaks and ebbs. One way to achieve that is with the deployment of AI/ML technologies can resolve this issue by applying measures such as ML applications for the data generated by advanced sensors, smart meters and intelligent devices beyond-the-meters.
Of critical importance is that this will allow utility companies and grid operators alike to estimate how the individual appliances of users are likely to behave over time, as well as at specific times. Swiss company Alpiq is already employing an AI system called GridSense so as to better understand user behavior to ensure optimized energy use, referring it to as an “intelligent Home Energy Management System (HEMS)”.
This is something that has proven very difficult to factor in given the unpredictability of renewables such as solar and wind. Furthermore, utility companies can also use these algorithms to predict the storage life of renewable energy, so that grid operators can more efficiently determine required energy ‘payouts’ accordingly.
Power Grid Efficiencies
It is contended that possibly the the most dynamic and game-changing application of AI regarding renewables will be with regard to national power grids. Commentators have noted how power grids across the globe are ageing and massively centralized. This has only exacerbated what has until recently been the biggest drawback for national grids regarding renewables, i.e. the ‘intermittent’ nature thereof, particularly regarding solar (‘the sun isn’t always shining’) and wind (‘the wind isn’t always blowing’). This in turn has hampered the adequate storage of renewable energy. The advent of AI-generated grid efficiency means that this ‘double whammy’ against the uptake of renewables is a non-issue. AI applications in renewable energy production creates the ideal circumstance for an entirely new distribution paradigm.
Swagath Navin Manohar, a research analyst on energy and the environment has stated how, “In combination with other technologies like Big Data, cloud, and Internet of Things (IoT), AI can support the active management of electricity grids by improving the accessibility of renewable energy sources.” As discussed, AI algorithms can assist energy companies in understanding and optimizing consumer behavior, combined with precise ML-generated weather prediction capabilities and climate modeling. This potential is duplicated where energy deliverables are most critical, i.e. with grid operators.
A further area where AI solutions can make a significant difference for a national energy grid is in the control and operation of what is termed ‘demand response’. This principle, already well-entrenched in most energy markets, allows large consumers of electricity to be rewarded when decreasing their energy requirements on short notice in order to stabilize the grid. AI raises this potential exponentially – and it will also save costs for grid operators.
‘Grid resilience’ is also a key factor as to why AI will ensure the ascendancy of renewables in national grid considerations. That’s why UK demand-side energy analyst Michael Phelan is of the opinion that, “High energy users must embrace AI and machine learning technology to advance towards a more resilient, flexible and decarbonized grid.”
A Caveat: AI and Fossil Fuels
The current reality of global energy needs, coupled with what was until recently the unpredictability of renewable energy, means that, at least for now and whether we like it or not, fossil fuels remain an integral part of the energy status quo. It should be noted that AI is already assisting fossil fuel companies on the supply side by reducing exploration and production costs, as well as streamlining delivery thereof. In short, AI can ‘empower’ fossil fuels as much as it can renewables. That is not a long-term ideal scenario, given the proven unsustainability of fossil fuels.
However, renewable energy should ultimately win that battle, as renewables are at the vanguard of new, smart technologies. AI, itself dynamic, innovative and smart, is undoubtedly a better fit with the philosophy of renewables.
John Bowlus has stated that, “…artificial intelligence (AI) may become the largest disrupter of the global energy system.” It is contended here that the only bigger ‘disrupter’ than AI to the global energy system will be the continued uptake of renewable energy. That uptake will surge once AI and the renewable energy industry are in perfect symbiosis.
As discussed in this article, AI is already revolutionizing and will continue to revolutionize supply-side renewable energy companies. On the demand side, an AI-driven smart grid will facilitate grid operators in offering optimized renewable energy distribution solutions to consumers. Renewable energy will become even cheaper and more reliable in what can only be considered a triumph for sustainability.