Monitoring brine leakage from CO2geological storages (CGS) is necessary to protect shallow aquifers against contamination. A framework for designing CGS monitoring systems that optimally use both easily available shallow zone data and hard‐to‐obtain deep zone observations is developed and validated. This framework is based on calibrating a transport model using monitoring data to determine leakage source conditions and then predict the subsequent brine plume that potentially contaminates shallow aquifers. As cost considerations are expected to limit monitoring deep formations, the framework is developed to minimize the number of deep observation points (e.g., deep sensors). The best monitoring locations that yield the most worthful data for reducing predictive uncertainty is selected by integrating linear uncertainty analysis with Genetic Algorithm under this framework. Due to practical challenges, testing such a framework in the field is not feasible. Thus, the framework was tested in an intermediate‐scale soil tank, where monitoring data on brine leakage plume development from the storage zone to the shallow aquifer were collected. Predictions made by a transport model calibrated on these data were then compared with experimental measurements to evaluate data informativity and thus validate the framework's applicability. The results demonstrate the framework ability to select the optimum monitoring locations for leakage detection and model calibration. It was also found that not only deep observations, but also shallow zone data are worthful to determine source conditions. Moreover, the results showed the possibility of identifying the likely areas to be impacted in the shallow aquifer using early stage monitoring data.
Natural fissures/faults or pressure‐induced fractures in the caprock confining injected CO2have been identified as a potential leakage pathways of far‐field native brine contaminating underground sources of drinking water. Developing models to simulate brine propagation through the overlaying formations and aquifers is essential to conduct reliable pre‐ and post‐risk assessments for site selection and operation, respectively. One of the primary challenges of performing such simulations is lack of adequate information about source conditions, such as hydro‐structural properties of caprock fracture/fault zone and the permeability field of the storage formation. This research investigates the impact of source condition uncertainties on the accuracy of leaking brine plume predictions. Prediction models should be able to simulate brine leakage and transport in complex multilayered geologic systems with interacting regional natural and leakage flows. As field datasets are not readily available for model testing and validation, three comprehensive intermediate‐scale laboratory experiments were used to generate high‐resolution spatiotemporal data on brine plume development under different leakage scenarios. Experimental data were used to validate a flow and transport model developed using existing code FEFLOW to simulate brine plume under varying source conditions. Spatial moment analysis was conducted to evaluate how uncertainty in source conditions impacts brine migration predictions. Results showed that inaccurately prescribing the permeability field of storage formation and caprock fractures in models can cause errors in leakage pathway and spread predictions up to ∼19% and ∼100%, respectively. These findings will help in selecting and characterizing storage sites by factoring in potential risks to shallow groundwater resources.
more » « less- NSF-PAR ID:
- 10367582
- Publisher / Repository:
- DOI PREFIX: 10.1029
- Date Published:
- Journal Name:
- Water Resources Research
- Volume:
- 57
- Issue:
- 9
- ISSN:
- 0043-1397
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
more » « less
Abstract -
more » « less
Abstract Quantitative predictions of natural and induced phenomena in fractured rock is one of the great challenges in the Earth and Energy Sciences with far‐reaching economic and environmental impacts. Fractures occupy a very small volume of a subsurface formation but often dominate fluid flow, solute transport and mechanical deformation behavior. They play a central role in CO2sequestration, nuclear waste disposal, hydrogen storage, geothermal energy production, nuclear nonproliferation, and hydrocarbon extraction. These applications require predictions of fracture‐dependent quantities of interest such as CO2leakage rate, hydrocarbon production, radionuclide plume migration, and seismicity; to be useful, these predictions must account for uncertainty inherent in subsurface systems. Here, we review recent advances in fractured rock research covering field‐ and laboratory‐scale experimentation, numerical simulations, and uncertainty quantification. We discuss how these have greatly improved the fundamental understanding of fractures and one's ability to predict flow and transport in fractured systems. Dedicated field sites provide quantitative measurements of fracture flow that can be used to identify dominant coupled processes and to validate models. Laboratory‐scale experiments fill critical knowledge gaps by providing direct observations and measurements of fracture geometry and flow under controlled conditions that cannot be obtained in the field. Physics‐based simulation of flow and transport provide a bridge in understanding between controlled simple laboratory experiments and the massively complex field‐scale fracture systems. Finally, we review the use of machine learning‐based emulators to rapidly investigate different fracture property scenarios and accelerate physics‐based models by orders of magnitude to enable uncertainty quantification and near real‐time analysis.
-
more » « less
Abstract Injecting anthropogenic CO2into the subsurface is suggested for climate change mitigation. However, leakage of CO2from its target storage formation is a concern. In the event of leakage, permeability in leakage pathways such as faults may get altered due to mineral reactions induced by CO2‐enriched water, thus influencing the migration and fate of the CO2. An example of such fault permeability alteration is found in the Little Grand Wash Fault zone (LGWF), where the fault outcrops show fractures filled with calcium carbonate. To test the nature, extent, and time‐frame of such fault ‘self‐sealing’, we developed reactive flow simulations based on hydrogeological conditions of the LGWF. We measured LGWF water chemistry and conducted an x‐ray powder diffraction (XRD) analysis of the fault‐rock to constrain the model geochemistry. We hypothesized that the choice of parameters for relative permeability, capillary pressure, and reaction kinetics will have a huge impact on the fault sealing predictions. Simulation results showed that precipitation of calcite in the top portion of the fault led to a decrease in porosity of the damage zone from a value of 40% to
. Over a simulation time of 1000 years, porosity in the damage zone showed self‐enhancing behavior in the bottom portion and self‐sealing behavior in the top portion of the fault. We found that the results were most sensitive to the relative permeability parameters and the fault architecture. A major conclusion from this analysis is that, under similar conditions, some faults are likely to seal over time. © 2017 Society of Chemical Industry and John Wiley & Sons, Ltd. -
Shallow groundwater resources overlaying deep saline formations used in carbon storage applications are subjected to a potential contamination threat by CO2/brine leakage via natural or anthropogenically-induced conductive pathways in the confining caprock. Identifying the leakage source location and rate is critical for developing remediation plans and designing corrective actions. Owing to limited information about the flow and transport characteristics of deep regimes and high cost of obtaining data on their response to CO2 injection operation, estimating accurate source settings (i.e., location and rate) can be extremely challenging. Under such conditions, Bayesian inverse frameworks become useful tools to help identify potential leakage patterns. This study tests and validates an ensemble-based data-assimilation approach that reduces the uncertainty in the prior knowledge about source settings through conditioning forward transport models using relatively inexpensive easy-to-acquire shallow zone data. The approach incorporates the newly developed ensemble smoother tool in the inversion code “PEST++” with the transport code “FEFLOW” to perform history matching and uncertainty analysis. A novel parameterization method that allows the disposition of potential source was used to search the leakage location during calibration process. In the absence of field data, the approach was validated using experimental data generated in ~8 m long soil tank simulating leakage from storage zone migrating to the shallow aquifer. The results show that source location uncertainty can be reasonably reduced using shallow zone data collected from near-surface aquifers. However, more prior information about the system and deeper data are essential to estimate a practical probability range for the leakage rate.more » « less
-
more » « less
Abstract Solving multiphysics-based inverse problems for geological carbon storage monitoring can be challenging when multimodal time-lapse data are expensive to collect and costly to simulate numerically. We overcome these challenges by combining computationally cheap learned surrogates with learned constraints. Not only does this combination lead to vastly improved inversions for the important fluid-flow property, permeability, it also provides a natural platform for inverting multimodal data including well measurements and active-source time-lapse seismic data. By adding a learned constraint, we arrive at a computationally feasible inversion approach that remains accurate. This is accomplished by including a trained deep neural network, known as a normalizing flow, which forces the model iterates to remain in-distribution, thereby safeguarding the accuracy of trained Fourier neural operators that act as surrogates for the computationally expensive multiphase flow simulations involving partial differential equation solves. By means of carefully selected experiments, centered around the problem of geological carbon storage, we demonstrate the efficacy of the proposed constrained optimization method on two different data modalities, namely time-lapse well and time-lapse seismic data. While permeability inversions from both these two modalities have their pluses and minuses, their joint inversion benefits from either, yielding valuable superior permeability inversions and CO2plume predictions near, and far away, from the monitoring wells.
