Year: 2024
Type: conference-paper
Abstract: Current exploration of geothermal involves much of focus on fracture imaging and mapping because this is one of the crucial factors influencing the geothermal production and recovery.There are many studies based on various datasets from drilling, completion, monitoring, testing, and production.However, how to leverage such multiple datasets to observe fracture from different "angle" of the datasets is still a challenge due to limited data availability and complex nature of the datasets especially in geothermal sector.This paper, we are going to apply advanced machine learning techniques to such various datasets to image fracture based on available datasets from a geothermal field frontier observatory.At National Energy Technology Laboratory (NETL), U.S. Department of Energy (U.S. DOE), a multi-tiered analysis has been started to transfer its applications from oil & gas and carbon storage to geothermal studies.Introducing advanced machine learning techniques to analyze each dataset for fracture image is the first part of our methodology. Such datasets are offered from U.S. DOE sponsored Frontier Observatory for Research in Geothermal Energy (FORGE) initiative. The analysis includes: 1) Apply multiple unsupervised machine learnings to analyze passive seismic dataset on 16A well for quantifying of the b-value and diffusivity based on the events from survey and monitoring, and then map the fracture geometry from the quantifications. 2) Leverage well logs (vertical and horizontal), core testing, stimulation and pumping data, and tracer records with supervised machine learnings to cross-check what the fracture understanding from passive seismic dataset. 3) Simulate completion and stimulation process with treatment, injection, and testing pressure measurements. The simulation results are used to validate what the fracture imaging obtained from 1) and 2). The second part of the methodology is to apply data/information fusion techniques to couple the fracture understanding comprehensively.
DOI: 10.2172/2477230
Authors: Gavin Liu, Abhash Kumar, Scott Beautz, Mackenzie Mark-Moser, Luciane Cunha, Diego Gonzalez, Jeffrey Nguyen, Saif Qawasmeh, Dylan Vaughn, HuiHui Yang, Jay Chen, Dustin Crandall, Mark McKoy, Alexandra Hakala
Language: English
Subfield: Mechanical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4404442159