| Title | A Progress Report on GeoThermalCloud Framework: an Open-Source Machine Learning Based Tool for Discovery, Exploration, and Development of Hidden Geothermal Resources |
|---|---|
| Authors | Bulbul AHMMED, Maruti K. MUDUNURU, Luke FRASH, and Velimir V. VESSELINOV |
| Year | 2023 |
| Conference | Stanford Geothermal Workshop |
| Keywords | geothermal exploration, machine learning, NMFk, Great Basin |
| Abstract | GeoThermalCloud is a Department of Energy’s Geothermal Technologies Office funded project to develop an open-source tool (https://github.com/SmartTensors/GeoThermalCloud.jl) to discover hidden geothermal resources using machine learning and prospecting enhanced geothermal systems (EGS). We named the geothermal resources exploration component GeoThermalCloud-RE while the EGS prospecting component GeoThermalCloud-EGS. GeoThermalCloud-RE utilizes unsupervised machine learning (ML) to automate data analyses and interpretations by extracting hidden signatures to elucidate geothermal prospects. Also, it enables the identification of critical measurements needed to identify geothermal resource signatures. GeoThermalCloud-RE can be applied to (1) analyze large sparse field datasets, (2) assimilate model simulations, (3) perform transfer learning (between sites with different exploratory levels), (4) label geothermal data types, resources, and processes, (5) identify high-value data acquisition targets, and (6) guide geothermal exploration and production by selecting optimal exploration, production, and drilling strategies. GeoThermalCloud-EGS is a machine learning-based alternative to GeoDT, a fast, simplified multi-physics solver to evaluate EGS designs in uncertain geologic systems. This paper will briefly update our progress from the project's onset to the present. |