Inverting Injection-Induced Microseismic Monitoring Data with Coupled Flow and Geomechanical Models: Application to CO2 Injection

Abstract

Abstract High pressure fluid injection into geologic formations can trigger rock deformation and failure and lead to microseismic events. Injection induced seismicity has been proposed as a monitoring technique that can be used to constrain the description of rock flow and mechanical properties. In this paper, we present a stochastic geomechanics-based approach to establish physical correlation between rock flow and mechanical properties and microseismic monitoring data during CO2 injection. The resulting correlations are then used to estimate rock properties from observed microseismic clouds. To establish the correlations between rock properties and microseismic clouds, the developed framework combines coupled flow and geomechanics simulation outputs with Mohr-Coulomb's failure criterion to describe the spatiotemporal distribution of seismicity potential in the formation (i.e., a map that quantifies the probability of geomechanically-induced seismic events at different times and locations in the formation). The resulting seismicity potential map depends on the stress conditions and mechanical strength of the formation rock and establishes a relationship between predicted microseismic clouds and rock property distributions. This model is then used with an ensemble-based data assimilation (i.e., ensemble smoother with multiple data assimilation, ES-MDA) to estimate rock properties form observed microseismic data. Since the ensemble smoother is designed for assimilation of continuous data, kernel density estimation (KDE) is applied to convert discrete microseismic events to a continuous map of seismicity density. Furthermore, for estimation of discrete geologic facies, prior to data assimilation, a distance transformation is used to convert facies descriptions to constinuous parameters. We present the developed formulation and discuss its application to inversion of microseismic data for characterization of reservoir flow and mechanical properties. Several examples are presented to evaluate the performance of the method for estimation of rock properties from microseismic data. Our preliminary results, based on synthetic experiments, suggest that the developed method can combine coupled physics models with microseismic monitoring data to constrain the description of rock flow and mechanical properties.