A data-driven approach for the evaluation of seismicity risks associated with CO2 injection

Abstract

This work examined the application of Bayes’ theorem in evaluating the risk of induced seismicity associated with CO2 sequestration in the Arbuckle Group, which extends across the southern Mid-Continent of the US. Geological storage can effectively contribute to reducing emission of CO2, otherwise released into the atmosphere, achieving the climate goals committed in the 2021 United Nations Climate Change Conference (COP26). However, concerns about risks associated with CO2 injection along with economic challenges of infrastructure required to execute the Carbon Capture Utilization and Storage projects stand against full realization of remarkable potentials. The main goal is usually for CO2 to be stored over geologic time; hence, geomechanical risks such as the seismicity in the field or potential CO2 leakage through seals cannot be ignored and is considered as one of the requirements to determine success of the project. This work elaborated on the risk of potential seismic events that can impact the longevity and success of projects. Accurate risk estimation is key for environmental, economic, and safety concerns and is also one of the requirements to get class VI permits from the US Environmental Protection Agency. The increase of fluid injection in the Arbuckle Group and how it has increased seismicity risks was first demonstrated, and then utilizing the Bayesian approach, a statistical model where a random probability distribution is used to represent uncertainties within the model, including both input/output parameters to evaluate the seismicity risks was used to estimate these risks. Using the Arbuckle Group as a case study, established physics-based models of the system and the details from past observed/monitored failures was utilized to evaluate future risk potential for the area. In this approach, the current probability for the state of stress for the area under investigation was established, then the evolution in the state of stress was monitored. The stress state probability distribution was calculated to evaluate the probability of activating a critically oriented fault over a range of specified pore pressures. The results suggest seismicity risk is directly a function of fluid injection and that the probability of inducing seismicity in the formation can be estimated. Based on the modelling results, at initial injection pressures there is a 24% risk of introducing seismicity in the Arbuckle Group when a critically oriented fault exists. Based on these results, sensitivity analysis was conducted to determine the features that can impact the risk level. Introducing the stress state constraints from the Arbuckle Group in Kansas State, the risk of seismicity reduced to 12%. Considering the results from this work, operators can optimize the site screening and collect additional data to constrain inherent uncertainties in geomechanical risk evaluation and make informed decisions during operations. The result from this work shows that geological storage of CO2 with attention to seismicity risks in the Arbuckle formation can be a feasible safe strategy towards achieving climate goals in selected areas and there is value of information in obtaining stress data in these areas.