Can artificial intelligence help renewables surge on the demand-side?

This article is the second of a three-part series on how artificial intelligence is already shaping the global energy system and could revolutionize it in the future.

Of all the intriguing new technologies in the works today, artificial intelligence (AI) may become the largest disrupter of the global energy system. AI’s ability to aggregate, sift through, and analyze data is already changing the business model for energy producers and could revolutionize how energy is delivered and consumed in the future. In 2017, Bill Gates advised college graduates to enter one of three fields: AI, energy, or biosciences.

More than any area, energy will affect the future of local communities, states, and the planet at large. AI will make energy markets smarter, cheaper, and more efficient, but the question remains: will AI simply reinforce old energy habits or nurture new ones?

Two weeks ago, I argued that artificial intelligence is already helping fossil fuels on the supply-side by reducing exploration and production costs and streamlining delivery. Yet AI ultimately presents the most upside for renewables because power demand is driving global energy demand growth by digitalizing and decentralizing the production, distribution and consumption of power. Renewables, moreover, are at the vanguard of new, smart technologies, to which AI can be developed in complementary fashion. Whereas oil and gas has a fixed model, renewables can furnish a new model for producing, distributing and consuming power, which cuts across all sectors and layers of the global economy. This is in contrast to fossil fuels, which are confined to the transportation and chemicals sectors.

You can predict the weather

Just like with oil and gas, AI helps producers of renewable energies decide where to build wind- or solar-power plants. Researchers at the Massachusetts Institute of Technology have developed a machine-learning system that can predict variations in wind speeds over time to help power companies more quickly evaluate potential locations for wind farms. Traditionally, a power company will gather 12 months of wind-speed data to evaluate a potential wind-farm location. Machine-learning systems, on the other hand, can produce more accurate models with just three months of data by correlating data from multiple sites and weather stations.

The ability to predict weather also helps in delivery. In Colorado, where wind generation has doubled since 2009, the National Center for Atmospheric Research is helping power companies decide when to use wind by providing AI-powered forecasts. One of these companies, Xcel Energy, used to oppose renewables; now it is a champion. Solar forecasts are intuitively easier, but are less seamless because of the lack of data on home units that feed into the grid. IBM has also developed a machine-learning system called Self-Learning Weather Model and Renewable Energy Forecasting Technology (SMT) to do this. SMT analyzes data from 1,600 weather stations, solar plants, wind farms, and weather satellites to generate weather forecasts 30 percent more accurately than the National Weather Service and predict renewable energy availability up to weeks in advance. 

Making smart grids smarter

The most potent application of AI, however, will be in remaking the power grid. Grids across the global are ageing and centralized, which creates an incentive to upgrade and expand them. Moreover, the intermittent nature of renewables – the sun isn’t always shining and the wind blowing – and the lack of adequate storage creates the ideal circumstance for a new distribution paradigm.

AI-driven smart grids, underpinned by wireless communications and the Internet of Things (IoT), will facilitate dynamic distribution management. The ability to monitor supply, storage, and demand in real time will make power cheaper and more reliable. These new systems can also manage consumers who are becoming low-scale producers of energy through their own solar panels. Swiss company Alpiq is already employing an AI system called GridSense to understand user behavior for optimal use of energy, one of many such innovative companies.

New production planning and control concepts, including manufacturing execution systems (MES) and enterprise resource planning (ERP) can decrease time-related risks and inefficiencies in energy consumption. The ability to network all components of industrial or manufacturing data, including operating and machine data, product-related master data and machine-related energy data, preferably in real time, more flexible and cost-efficient energy planning can permit flexible energy and cost-efficient planning. These systems have the added value of transparency in the production process. From an energy perspective, it synchronizes energy supplies and process.

The data-AI-renewables nexus

A symbiosis between data, AI, and renewable energy has already emerged with the growth of technology companies needing to store and access increasingly vast amounts of data. In 2014, U.S. data centers used 70 billion kWh of energy. From 2010 to 2014, energy consumption grew by 4% and is forecast to do the same until 2020. Google began applying DeepMind’s machine learning to reduce its energy usage on its data centers, managing to reduce energy use by 40% in 2016. The cheaper it becomes to power data centers with AI, the more data can be collected. Data giants such as Google, Amazon, and Apple are going into the renewable energy business. Locating data centers in the vast and sun-drenched West and Southwestern United States allows them to harness off-grid solar and wind power.

With the threat of cyber actions towards centralized grids, power-supply security takes on new meaning, incentivizing militaries and states to embrace renewables to protect operations but also their own growing stores of data. The largest employer and consumer of energy in the world, the U.S. Department of Defense, is also looking to renewables to reduce its energy bills and is even building its own power plants to do so for the first and second reasons listed above.

The dragon moves with electricity

The frontier in which AI might also capture the demand-side is automated electric vehicles. Many in the West fear trusting machines to move us, but this is misplaced. Human-induced accidents far outweigh technical malfunctions, while locomotives have moved us for nearly 200 years. The ability to take distracted humans off the road, opening up time to work or leisure, is attractive to any consumer. What only rich people can afford – having a chauffeur – could become a reality for the middle class.

One country that has no such problems with irrationality is China, which is far outpacing the United States and Europe in adopting of electric vehicles. Chinese carmaker Nio went public in August 2018 while Volkswagen announced an investment of €15 billion in China in August.

China is making in investments in AI that far outpace any other country and will drive the future of global electricity demand. Beginning in the 2010s, it emerged as the leader in production and export of renewable energy technologies. Now, it is surging as a leader on the demand-side. It dwarfs the rest of the world in installation of smart meters and high-voltage transmission and interconnection, and is the largest investor in smart distribution networks.

Oil and gas companies also recognize that power is the future, and are going into the electricity business. In April 2018, Shell announced that it wants electricity including from renewable sources to become the fourth pillar of its business.

Winning the demand-side is the goal of any business, but the energy terrain is now far more competitive than in the last two hundred years. Fossil fuel industries are harnessing AI more quickly to make their products cheaper, but renewables have the potential to be even cheaper. Whether they become more attractive than fossil fuels will depend on whether they create greater productivity, a topic to which I will turn next.