How Artificial Intelligence Will Revolutionize the Energy Industry

by Franklin Wolfe
figures by Franklin Wolfe and Kimia Mavon

Earlier this year, Bill Gates, founder of Microsoft and the richest man on Earth, wrote an essay online at “The blog of Bill Gates,” to college students graduating worldwide in 2017. He stated, “If I were starting out today… I would consider three fields. One is artificial intelligence (AI). We have only begun to tap into all the ways it will make people’s lives more productive and creative. The second is energy, because making it clean, affordable, and reliable will be essential for fighting poverty and climate change.” The third field he mentioned was biosciences.

What is inspiring for individuals who are dedicated to improving living conditions today and for future generations to come is that AI and energy are not mutually exclusive career paths. In fact, they are becoming increasingly interconnected as computing power, data collection, and storage capabilities scale exponentially on an annual basis. According to Dan Walker, who leads the emerging technology team in British Petroleum’s (BP) Technology Group, “AI is enabling the fourth industrial revolution, and it has the potential to help deliver the next level of performance.”

Although AI is in its early stages of implementation, it is poised to revolutionize the way we produce, transmit, and consume energy. At the same time, AI is also limiting the industry’s environmental impact at a time when demand is steadily growing, our energy production portfolio is diversifying, and we are witnessing the ramifications of fossil fuel consumption on biodiversity, air quality, and quality of life.

Why does the energy grid need to be modernized?

In 1882, Thomas Edison opened America’s first power plant at Pearl Street Station in lower Manhattan to deliver power to 59 customers. The customer base has since swelled to hundreds of millions of users, but the overall structure has yet to receive a modern overhaul. It consists of a vast network of power plants, transmission lines, and distribution centers (comprising roughly 5,800 power plants and over 2.7 million miles of power lines).

High costs for infrastructure and distribution lines, as well as stringent governmental regulations, naturally create opportunities for monopolies to develop in the market. As a result, three separate U.S. grids produce and transmit power under the mandate to provide low-cost, reliable energy as a public good.

In the U.S., the average age of power plants is over 30 years and of power transformers is over 40 years. This deteriorating transmission system led to the 2003 Northeast blackout, the largest failure in U.S. history according to the federal task force charged with its investigation. It left 50 million people without power for several days when an overloaded transmission line sagged and struck a tree. Instances like these can have cascading effects on the entire regional grid and pose a difficult task for utility companies to manage.

Figure 1: This figure demonstrates the rising trend of U.S.
renewable energy supply over the past decade. Hydropower includes conventional hydroelectric power only, and excludes pumped storage generation. Liquid biofuels include ethanol and biodiesel. Other renewables includes biofuels production losses and co-products. Data retrieved from the U.S. Energy Information Administration.

An additional challenge is the rise of distributed generation, where private users generate and use their own electricity from renewable sources, such as wind and solar. This complicates supply and demand and forces utility companies to buy excess energy from private users, who generate more electricity than they use and send the excess energy back to the grid. Since 2010, solar use has more than tripled, and this trend is poised to continue into the future as photovoltaic cells, the devices that generate electricity from sunlight, decrease in cost and increase in efficiency.

The current system was not built to accommodate this diversification in energy sources, especially not the rise in renewable resources. Rather, when demand outpaces supply, utilities turn on backup fossil fuel-powered plants, known as ‘peaker plants’, at a minute’s notice to avoid a cascading catastrophe. This procedure is the most expensive and wasteful part of business for these companies, manifesting itself in higher electricity bills for consumers and enhanced greenhouse gas emissions into the atmosphere. These problems will be exacerbated as the U.S. energy demand is projected to steadily increase into the future.

How can the energy grid be modernized?

To combat these problem, the U.S. Department of Energy (DOE) has made supporting the ‘smart grid’ a national policy goal, which entails a “fully automated power delivery network that monitors and controls every consumer and node, ensuring a two-way flow of electricity and information.” Since 2010, the DOE has invested $4.5 billion in smart grid infrastructure and installed over 15 million smart meters that monitor energy usage per device and alert utilities of local blackouts. It is estimated that while total U.S. energy demand is expected to increase 25 percent by 2050, this program will limit the rise in peak electricity load on the grid to only 1 percent.

Figure 2: Past and projected U.S. energy consuption in quadrillion Btu. By 2040, world energy consumption is expected to increase by 15.3%. Data retrieved from the U.S. Energy Information Administration.

AI will be the brain of this future smart grid. The technology will continuously collect and synthesize overwhelming amounts of data from millions of smart sensors nationwide to make timely decisions on how to best allocate energy resources. Additionally, the advances made from ‘deep learning’ algorithms, a system where machines learn on their own from spotting patterns and anomalies in large data sets, will revolutionize both the demand and supply side of the energy economy.

As a result, large regional grids will be replaced by specialized microgrids that manage local energy needs with finer resolution. These can be paired with new battery technologies that allow power to continually flow to and between local communities even when severe weather or other outages afflict the broader power system.

On the demand side, smart meters for consumers, including homes and businesses, and sensors along transmission lines will be able to constantly monitor demand and supply. Further, briefcase-sized devices known as ‘synchrophasers’ would measure the flow of electricity through the grid in real time, allowing operators to actively manage and avoid disruptions. These sensors would communicate with the grid and modify electricity use during off-peak times, thereby relaxing the workload of the grid and lowering prices for consumers. Google recently applied this AI technology to reduce its total data center power consumption, which translated to millions of dollars in savings.

On the supply side, AI will allow the U.S. to transition to an energy portfolio with increased renewable resource production and minimal disruptions from the natural intermittency that comes with these sources due to variable sunlight and wind intensity. For example, when renewables are operating above a certain threshold, either due to increases in wind strength or sunny days, the grid would reduce its production from fossil fuels, thus limiting harmful greenhouse gas emissions. The opposite would be true during times of below-peak renewable power generation, thus allowing all sources of energy to be used as efficiently as possible and only relying on fossil fuels when necessary. Additionally, producers will be able to manage the output of energy generated from multiple sources to match social, spatial, and temporal variations in demand in real-time.

Are there concerns with the future smart grid?

One of the major concerns with the smart grid is the increased use of Information and Communication Technology, which relies on the Internet as well as computing and processing power to run. This industry has become a large contributor of greenhouse gas emissions in recent years as companies shifted to machine-run operations, and the use of the Internet has increased by 30-40 percent per year. To process the amount of data necessary to run the smart grid, additional machines and computing power will be needed, and the impact of energy consumption on the environment from further greenhouse gas emissions is sure to increase. Therefore, players in the AI energy grid industry will need to address this problem.

Fortunately, industry leaders are aware of this challenge and are already taking steps to in the right direction. The three leading greenhouse gas emmitters in this industry, computer makers, data centers, and telecoms are looking to reduce emmissions in many ways. For example: Computer makers are investing in new hard drives, screens, and fuel cells; data centers are monitoring temperatures, pooling resources and researching cloud computing; and telecoms are looking into network optimization packages, solar-powered base stations, and fibre optics.

If the smart grid is able to use fossil fuels in the most efficient way possible through increased incorporating of renewable resources as those technologies advance in sophistication and capability, the entire system may be able to reduce its carbon footprint. Despite this uncertainty associated with future technological innovation, we can be optimistic in expecting the smart grid system to lower electricity bills and prevent catastrophic blackouts by optimizing supply and demand at local and national levels.

For those looking to make a difference in shaping the future of society, the interface between AI and energy is a great place to start. Technological innovation is drastically changing the way we think about these two industries and their integration is in its early stages. Their synergy may change the world like we never knew it, and they are primed for innovative thinkers to make their mark.

Franklin Wolfe is a graduate student in the Earth and Planetary Science program at Harvard University.

This article is part of a Special Edition on Artificial Intelligence.