Characterization of induced microseismicity at a carbon dioxide (CO2) storage site is critical for preserving reservoir integrity and mitigating seismic hazards. We apply a multilevel machine learning (ML) approach that combines the nonnegative matrix factorization and hidden Markov model to extract spectral representations of microseismic events and cluster them to identify seismic patterns at the Illinois Basin-Decatur Project. Unlike traditional waveform correlation methods, this approach leverages spectral characteristics of first arrivals to improve event classification and detect previously undetected planes of weakness. By integrating ML-based clustering with focal mechanism analysis, we resolve small-scale fault structures that are below the detection limits of conventional seismic imaging. Our findings reveal temporal bursts of microseismicity associated with brittle failure, providing insights into the spatio-temporal evolution of fault reactivation during CO2 injection. This approach enhances seismic monitoring capabilities at CO2 injection sites by improving fault characterization beyond the resolution of standard geophysical surveys.
We numerically investigate the mechanisms that resulted in induced seismicity occurrence associated with CO2 injection at the Illinois Basin–Decatur Project (IBDP). We build a geologi-cally consistent model that honors key stratigraphic horizons and 3D fault surfaces inter-preted using surface seismic data and microseismicity locations. We populate our model with reservoir and geomechanical properties estimated using well-log and core data. We then performed coupled multiphase flow and geomechanics modeling to investigate the impact of CO2 injection on fault stability using the Coulomb failure criteria. We calibrate our flow model using measured reservoir pressure during the CO2 injection phase. Our model results show that pore-pressure diffusion along faults connecting the injection inter-val to the basement is essential to explain the destabilization of the regions where micro-seismicity occurred, and that poroelastic stresses alone would result in stabilization of those regions. Slip tendency analysis indicates that, due to their orientations with respect to the maximum horizontal stress direction, the faults where the microseismicity occurred were very close to failure prior to injection. These model results highlight the importance of accurate subsurface fault characterization for CO2 sequestration operations.
The transmission interference fringe (TIF) technique was developed to visualize the dynamics of evaporating droplets based on the Reflection Interference Fringe (RIF) technique for micro-sized droplets. The geometric formulation was conducted to determine the contact angle (CA) and height of macro-sized droplets without the need for the prism used in RIF. The TIF characteristics were analyzed through experiments and simulations to demonstrate a wider range of contact angles from 0 to 90°, in contrast to RIF's limited range of 0-30°. TIF was utilized to visualize the dynamic evaporation of droplets in the constant contact radius (CCR) mode, observing the droplet profile change from convex-only to convex-concave at the end of dry-out from the interference fringe formation. The TIF also observed the contact angle increase from the fringe radius increase. This observation is uniquely reported as the interference fringe (IF) technique can detect the formation of interference fringe between the reflection from the center convex profile and the reflection from the edge concave profile on the far-field screen. Unlike general microscopy techniques, TIF can detect far-field interference fringes as it focuses beyond the droplet-substrate interface. The formation of the convex-concave profile during CCR evaporation is believed to be influenced by the non-uniform evaporative flux along the droplet surface.
Estimating spatially distributed properties such as permeability from available sparse measurements is a great challenge in efficient subsurface CO2 storage operations. In this paper, a deep generative model that can accurately capture complex subsurface structure is tested with an ensemble-based inversion method for accurate and accelerated characterization of CO2 storage sites. We chose Wasserstein Generative Adversarial Network with Gradient Penalty (WGAN-GP) for its realistic reservoir property representation and Ensemble Smoother with Multiple Data Assimilation (ES-MDA) for its robust data fitting and uncertainty quantification capability. WGAN-GP are trained to generate high-dimensional permeability fields from a low-dimensional latent space and ES-MDA then updates the latent variables by assimilating available measurements. Several subsurface site characterization examples including Gaussian, channelized, and fractured reservoirs are used to evaluate the accuracy and computational efficiency of the proposed method and the main features of the unknown permeability fields are characterized accurately with reliable uncertainty quantification. Furthermore, the estimation performance is compared with a widely-used variational, i.e., optimization-based, inversion approach, and the proposed approach outperforms the variational inversion method in several benchmark cases. We explain such superior performance by visualizing the objective function in the latent space: because of nonlinear and aggressive dimension reduction via generative modeling, the objective function surface becomes extremely complex while the ensemble approximation can smooth out the multi-modal surface during the minimization. This suggests that the ensemble-based approach works well over the variational approach when combined with deep generative models at the cost of forward model runs unless convergence-ensuring modifications are implemented in the variational inversion.
Seismic imaging methods are critical for Global Security and Energy & Homeland Security missions and activities that rely on subsurface characterization, but traditional methods remain computationally expensive and require significant labor hours and expertise to execute. Within the past few years, machine learning (ML), namely deep learning (DL), has been used to develop data-driven end-to-end full waveform inversion (FWI) methods to estimate 2D P-wave velocity (Vp) models in a fraction of the time as conventional FWI. These methods, however, are trained on simplistic acoustic wave seismic data and Vp models that are not realistic nor representative of real-world observations, leaving a large gap between the state-of-the-art and deployable, feasible, and practical DL FWI methods. Here, we generate a synthetic active-source, 3D, elastic wave seismic data set and a variety of Vp models with realistic geologic structure for training DL FWI methods. We evaluate six different methods that have performed well for acoustic DL FWI or medical imaging tasks using our more realistic dataset. We find that these six trained models do not match the performance of published acoustic end-to-end DL FWI methods, indicating more training data may be needed, physics may need to be incorporated to achieve good accuracy at the sacrifice of the end-to-end advantage, and/or novel methods need to be developed to enable end-to-end DL FWI methods to perform well for real-world seismic data.
This project helped address the growing need for efficient and scalable models to support geological carbon and energy storage, which are crucial for achieving net-zero emissions. Traditionally accurate high-fidelity numerical models have been used to simulate relevant storage processes under a handful of processes, however such models are computationally demanding, making uncertainty quantification impractical. Consequently, we first developed a machine learning framework, based on Graph Neural Operators (GNOs), to improving the accuracy of model predictions for a fixed computational budget. We then developed an Ensemble of Improved Neural Operators (ENO), which uses bagging and Monte Carlo dropout techniques, to further improve prediction accuracy. Lastly, we developed the way to explain progressive transfer learning methods to reduce the amount of training data and computational cost of training (i.e., reduce trainable parameters) when using our models for multiple storage sites. Our numerical investigation, which used real-world case studies, demonstrated that our framework can significantly improve the safety and efficiency of geological storage operations, with potential applications in other domains such as geothermal reservoirs and climate modeling.
Morris, Joseph P.; Pyrak-Nolte, Laura J.; Yoon, Hongkyu; Bobet, Antonio; Jiang, Liyang
In this article, We present results from a recent exercise where participating organizations were asked to provide model-based blind predictions of damage evolution in 3D-printed geomaterial analogue test articles. Participants were provided with a range of data characterizing both the undamaged state (e.g., ultrasonic measurements) and damage evolution (e.g., 3-point bending, unconfined compression, and Brazilian testing) of the material. In this paper, we focus on comparisons between the participants’ predictions and the previously secret challenge problem experimental observations. We present valuable lessons learned for the application of numerical methods to deformation and failure in brittle-ductile materials. The exercise also enables us to identify which specific types of calibration data were of most utility to the participants in developing their predictions. Further, we identify additional data that would have been useful for participants to improve the confidence of their predictions. Consequently, this work improves our understanding of how to better characterize a material to enable more accurate prediction of damage and failure propagation in natural and engineered brittle-ductile materials.
Morris, Joseph P.; Pyrak-Nolte, Laura J.; Yoon, Hongkyu; Bobet, Antonio; Jiang, Liyang
In this article, We present results from a recent exercise where participating organizations were asked to provide model-based blind predictions of damage evolution in 3D-printed geomaterial analogue test articles. Participants were provided with a range of data characterizing both the undamaged state (e.g., ultrasonic measurements) and damage evolution (e.g., 3-point bending, unconfined compression, and Brazilian testing) of the material. In this paper, we focus on comparisons between the participants’ predictions and the previously secret challenge problem experimental observations. We present valuable lessons learned for the application of numerical methods to deformation and failure in brittle-ductile materials. The exercise also enables us to identify which specific types of calibration data were of most utility to the participants in developing their predictions. Further, we identify additional data that would have been useful for participants to improve the confidence of their predictions. Consequently, this work improves our understanding of how to better characterize a material to enable more accurate prediction of damage and failure propagation in natural and engineered brittle-ductile materials.
