DESC0024790

While geothermal is the only renewable energy source that can operate 24 hours a day and 7 days a week, our countryĺs geothermal accounts for merely 0.4% of the electricity generation and there were fewer operating plants in our country in 2020 than 2010 [1]. This is due to uneconomic performance forcing plant closures and high risks impacting the decision to invest in a new geothermal plant. There is a lack of understanding of energy and heat recovery from each well that feeds a geothermal plant and what decisions are needed to intervene before the plant costs more to operate than it produces in energy. The project will use arti?cial intelligence and machine learning to create a very accurate, low cost and high reliability system to monitor and improve high temperature geothermal systems that can be used across all types of geothermal systems. Accurate estimates of ?ow, energy and heat characteristics are needed for all the individual parts of the geothermal sites including the wells, pipelines, and facility to operate e?ciently. In Phase I, the project will train a computer model to predict characteristics to determine how to measure thermodynamic flow using distributed acoustic and thermal sensors in the geothermal plant. This is accomplished by training a software program on existing geothermal data and systematically removing data points to predict the results. This creates a model with the minimal amount of data needed for sufficient accuracy. The software developed in Phase I will be paired with hardware to prepare for Phase II field trials. Geothermal electricity generation could increase more than 26-fold from today, reaching 60 gigawatts of installed capacity by 2050, according to our countryĺs energy department in 2019. [2] In Phase II, the project will field test the technology to enable several sites to make rapid interventions and continue operations. This technology will also reduce the risk of underperformance and lead to improved economics for new site builders.