Unveiling the Invisible: Mapping Basement Faults Using Microseismic Clouds

ABSTRACT: The Illinois Basin Decatur Project (IBDP) has piloted the process of large-scale injection of CO2 into a deep saline reservoir. During the project, microseismic clouds were observed from locations that were not identified in any previous geological database. Based on the injection depth, it is surmised that the pressure plume reached previously unidentified basement faults, altered their stress conditions, and triggered microseismic events. Understanding the presence and behavior of basement faults is critical for monitoring and managing induced seismic risks. However, basement faults with near-vertical dips can be challenging to detect using conventional seismic imaging methods due to poor incident angles. In this study, sponsored by the U.S. Department of Energy's SMART initiative, we developed a machine learning workflow using Long Short Term Memory (LSTM) recurrent neural network (RNN) architecture that maps basement faults using microseismic clouds. The LSTM-RNN is trained on synthetic microseismic catalogs produced by physics-based geomechanics models that simulate fluid injection into a reservoir above a faulted basement. Microseismic catalogs are passed to the LSTM-RNN to predict the location of the basement faults and shear stress state. The network aims to take real-time microseismic data and progressively produce an updated fault map. Our approach shows the potential of ML-based methods for mapping previously invisible basement faults, providing information to enable rapid decision-making during CO2 injection to prevent and mitigate induced seismic risks. However, there is still much room for improvement in the accuracy of the machine learning model, which we are actively working on. LLNL Release Number: LLNL-CONF-860823 1 INTRODUCTION Carbon sequestration is a process that involves capturing CO2 emissions from various sources such as industrial plants and power generation facilities, and storing the captured CO2 in subsurface reservoirs to help mitigate the buildup of greenhouse gases in the atmosphere (Halmann and Steinberg (1998); Pires et al. (2011); Herzog et al. (2000)). As nations and industries strive to reduce their carbon footprints and limit global warming, large-scale carbon sequestration has emerged as a significant, potential contributor toward meeting ambitious net-zero targets (Budinis et al. (2018); Meys et al. (2021)).