SPE and Project Innerspace are organizing the first Geothermal AInnovation Competition, hosted by the SPE Gulf Coast Section and the SPE Data Science and Engineering Analytics Technical Section. Teams from around the world are invited to participate in this exciting virtual competition aimed at showcasing the potential of AI-assisted work flows in the geothermal life cycle. This contest is the third collaboration between SPE and Project Innerspace, after the SPE Geothermal Datathon organized at PIVOT2022 and the SPE Geothermal Datathon that is closing this month with results to be announced at PIVOT2023. SPE and Project Innerspace, together with Geothermal Rising, have been selected by the US Department of Energy to be the administrators of the GEODE project, a 5-year initiative to facilitate technology transfer from oil and gas into geothermal, announced in May.
The inaugural Geothermal AInnovation Competition challenges the participating teams to propose large language model (LLM)-based work flows for any discipline (geophysics, geology, drilling and completions, reservoir engineering, production engineering, process engineering, and HSES) that would support accelerated assessments to facilitate the implementation of geothermal projects, paving the way for a greener and more sustainable future.
This competition aims to highlight the disruptive force of LLM to accelerate geothermal, from field development to research and development of new enablers. Participants are allowed to utilize any of the existing pre-trained models in the public domain to present conceptual ideas of applications of LLM-models into the life cycle of geothermal developments.
Registration is open to all AI enthusiasts that want to explore opportunities in the geothermal world. This is a good opportunity for students and young professionals to utilize advances of AI to excel in a new energy domain.
The competition is scheduled to start on 19 July and allows participation in teams of 1 to 4 members. The participants will have until 31 August to develop an LLM-based work flow applicable to a geothermal energy topic. Participants will submit a video of maximum 3 minutes where they will describe the target problem, the selected information and datasets to apply the LLM model, and a visualization of how the coded work flow works in one geothermal example. To ensure homogeneity among proposals, participants are requested to use only open access pre-trained LLM models. The winning team will be announced at PIVOT2023 on 20 September.
Authors: Pushpesh Sharma, SPE; Dani Merino-Garcia, Project Innerspace; Silviu Livescu, SPE: John Boden, SPE; John Reed, Project Innerspace; Sushma Bhan, SPE; Meisong Yan, SPE; Jerjes Porlles, SPE; and Balnur Mindygaliyeva, SPE.