Send this article to a friend:


Machine Learning Could Make Geothermal Energy More Affordable
Felicity Bradstock

As governments and private companies pump funds into research and development, aiming to achieve the innovation needed to advance renewable energy operations, we are seeing greater progress in the global green transition. With major advances in artificial intelligence (AI) and other digital offerings in recent years, some energy experts believe that this technology can now be used to enhance energy production, boosting the potential of the world’s geothermal energy output.

Geothermal energy is a natural form of heat energy that is located within the Earth. There is enough geothermal energy on Earth to meet the world’s energy needs, but accessing this energy can be very challenging. Geothermal energy can be accessed in different ways, the first is direct use, which has been popular for hundreds of years. This involves using heated water near the Earth’s surface, such as from hot springs and geysers, which requires no drilling equipment, and the water or steam can be used to heat buildings close to the source. By contrast, the geothermal energy that is used for electricity production is accessed through drilling. Reservoirs of hot water can be found in various places around the world just a few miles beneath the Earth’s surface. They can be accessed through drilling, allowing steam to be extracted to rotate a turbine, which activates a generator that produces electricity. 

Drilling for geothermal energy remains unpopular due to technological restrictions that mean it can be expensive to explore potential geothermal sites. In addition, hard-to-reach reserves can be extremely difficult to access with existing drilling technology. At present, geothermal energy contributes less than one percent of U.S. electricity, despite the huge potential to tap into this abundant renewable energy source. 

Zanskar, a Utah-based startup, has built machine learning models to assess the optimal locations to drill for geothermal energy. The company’s models analyse extensive data to determine the best locations to drill for geothermal energy, which Zanskar believes will lead to a significant reduction in exploration costs in the coming years. This could encourage more companies to invest in the geothermal sector and help diversify the green energy sector. 

Carl Hoiland, Zanskar’s CEO, stated “We've now discovered more of these hidden geothermal resources in just the past year and a half than the entire industry combined had done over the prior decade.” This demonstrates the huge potential of the new technology and could encourage more energy companies to invest in the sector. This month, Zanskar announced it had raised $30 million in a series B funding round led by Obvious Ventures, which values it at $115 million. To date, Zanskar has raised $45 million in funding, which will contribute to greater exploration and the development of its first power plants. The firm plans to work with existing geothermal companies to develop new sites. 

There is great enthusiasm around the potential for developing the world’s geothermal energy as it could provide an endless renewable resource. However, at present, the cost of project development is around five times that of wind energy, meaning that it is often overlooked. It costs around $8.7 million per megawatt of geothermal-derived electricity. This price is so high largely because drillers often fail to find the correct spot to access reservoirs, meaning they may have to drill multiple holes before being successful if they find it at all before running out of time and money. 

Zanskar uses a vast array of data collected from satellites, geological surveys, the waves that travel through the ground after an earthquake, and other data points, to predict the best spots for drilling. The more data available in a region, the more accurate the machine learning programme can be. This technology could be coupled with other innovations, such as advanced drilling techniques, to make geothermal energy more accessible and therefore cheaper. 

The U.S. National Renewable Energy Laboratory (NREL) is also developing its AI and machine learning techniques to enhance the production of renewable energy. It has developed a suite of algorithms and tools to improve reservoir characterisation, economise drilling, and optimise geothermal steam field operations. Meanwhile, since 2018, the U.S. Geothermal Technologies Office (GTO) has funded early-stage research and development applications in machine learning to develop new technologies for exploration and operational improvements for geothermal resources. 

Rapid advancements in AI and machine learning technologies have enhanced renewable energy operations, and further progress over the coming decades is expected to help optimise energy operations and drive down costs even further. Some of the main deterrents to developing geothermal energy operations could now be a thing of the past, if machine learning algorithms are as accurate as Zanskar and other companies are promising, allowing us to tap into abundant renewable energy sources around the globe. 

By Felicity Bradstock for




Felicity Bradstock is a freelance writer specialising in Energy and Finance. She has a Master’s in International Development from the University of Birmingham, UK.

Send this article to a friend: