Subsurface CO₂ sequestration is a promising method to advance carbon neutrality and support the shift toward sustainable energy. However, the unique behavior of CO₂ in these operations, particularly for cold CO₂ injection in depleted hydrocarbon reservoirs, poses challenges to wellbore injectivity, reservoir containment, and reservoir capacity. These challenges necessitate the development of a numerical model to better understand and optimize the interplay between wellbore dynamics and reservoir processes. In this work, we present the development of an open-source coupled wellbore-reservoir numerical model, named DARTS-well, which is tailored to CO₂ disposal in subsurface reservoirs. To this end, a multi-segment, multi-phase, non-isothermal wellbore model is first developed using the Drift-Flux Model (DFM), and its results for selected CO2 injection scenarios are validated against the commercial transient wellbore simulator OLGA. The multi-segment wellbore model is then coupled with the Delft Advanced Research Terra Simulator (DARTS) which is used in this study as the reservoir simulator. DARTS is widely used and validated for energy transition applications. The coupled model utilizes the Operator-Based Linearization (OBL) technique, employing state-dependent operators for thermodynamic properties interpolated from predefined tables or generated on the fly. This OBL parametrization approach addresses challenges associated with complex physics and reduces computational time, making it well-suited for modeling subsurface CO₂ sequestration.

Direct Use Geothermal Systems (DUGS) are rapidly and densely deployed to meet the growing demand for renewable energy with less carbon emissions globally. The simulation of DUGS can provide a reservoir-scale understanding of geothermal resource assessment, where the geothermal system's lifetime and the injection well Bottom Hole Pressure (BHP) are used as performance indicators. However, there are inherent errors from numerical simulations of any engineering problems, due to approximating continuous partial differential equations by their discretized approximation in time and space. In this work, we establish an optimal numerical setup with reduced errors across the homogeneous, stratified and heterogeneous models for the simulation of a geothermal system. Next, we develop a standardized method for calculating recoverable Heat In Place (HIP) and an analytical solution for evaluating the HIP recovery factor across various geological models using a single forward simulation. We present reference examples on the design of DUGS simulations using the open-source software Delft Advanced Research Terra Simulator (open-DARTS). The open-DARTS platform enables accurate and efficient sensitivity and uncertainty analysis. Using Distance-Based Generalized Sensitivity Analysis (DGSA), we identify reservoir depth and discharge rate as the most influential parameters for geothermal projects across all three types of geological models.

The effective management of geo-energy systems heavily relies on robust modeling frameworks that integrate diverse simulation capabilities, including flow and transport, phase equilibrium, geochemistry and geomechanics. While a multiphysics simulation engine within a unified framework has its advantages, integrating specialized modeling packages often enhances viability. Efficient and seamless communication between these engines be- comes crucial for improving the performance and scalability of the integration. Advanced parametrization tech- niques can facilitate this integration by efficiently approximating and interpolating coupling data, ensuring both speed and accuracy. In this study, we compare the efficiency of different interpolation techniques used for the parametrization of complex many-component fluid systems in compositional simulation. We employ an Operator- Based Linearization (OBL) framework that leverages the general formulation of corresponding conservation laws. OBL effectively learns the operators required for assembly of the laws while interpolation delivers fast evalua- tion of operators and their derivatives for all physical states in a simulation domain. Multilinear interpolation is a simple and robust approach, yet it has poor scaling properties with respect to the dimension of the physical state. To alleviate interpolation costs in multiple dimensions, we study the performance and accuracy of other interpolation techniques, including linear interpolation with standard and Delaunay triangulation. Overall, this approach provides great flexibility, saves development costs and simplifies the incorporation of thermodynamics and geochemistry engines for precise modeling of phase equilibrium, reactive transport, dissolution-precipitation and kinetics of chemical reactions. This research extends the scalability of the OBL framework and addresses the challenges of high dimensionality in compositional modeling. Consequently, this approach holds significant potential for integrating various complex multiphysics problems, enabling the creation of more comprehensive digital twins for geo-energy systems management.

Three-phase equilibrium calculations for water-CO2-hydrocarbon mixtures are required in the compositional simulation of various applications in CO2 storage, geothermal systems, and enhanced oil recovery. The very low solubility of hydrocarbon components in water leads to a special mathematical structure of the problem. Several techniques were suggested, such as the free-water flash (FWF) and the augmented free-water flash (AFWF); in the former, the aqueous phase is pure water, while in the latter only certain components, CO2 or methane for example, are dissolved in the aqueous phase. However, only the first-order successive substitution method was used in the previous published approaches, making them unattractive for compositional simulations in which a significant number of phase equilibrium calculations are performed. In this work, a robust and efficient AFWF method is proposed, using combined successive substitutions-modified Newton iterations. The new method is general, allowing partial solubility of any selected component in the water-rich phase, depending on the specific compositions and operating conditions. A detailed description of second-order methods in a Gibbs energy minimization framework for the general AFWF is presented. In the AFWF, the dimension of the problem and the number of function evaluations (thus the computation time) are significantly reduced. Moreover, it is shown that the augmented method always has better convergence properties than its conventional multiphase flash counterpart, in both first- and second-order methods. The new AFWF method is tested for various hydrocarbon-water-CO2 mixtures and proved to be robust and efficient, systematically outperforming the conventional approach. Unlike in previous AFWF formulations, the number of components soluble in water is not limited, leading to a controlled accuracy with respect to a full three-phase equilibrium, even at high pressures and/or large amounts of CO2.

Modeling of multiphase flow and reactive mass transport in porous media remains a pivotal challenge in the realm of subsurface energy storage, demanding a nuanced understanding across varying scales. This review paper presents a comprehensive overview of the latest advancements in multiscale modeling techniques that address the inherent complexity of these processes. Three cutting-edge approaches are presented: hybrid multiscale simulation, which leverages both continuum and discrete modeling frameworks to enhance model fidelity; approximated physics, which simplifies complex reactions and interactions to expedite computations without significantly sacrificing accuracy; and machine-learning-assisted multiscale simulation, which integrates predictive analytics to refine simulation outputs. Each method presents distinct advantages and hurdles, collectively advancing the precision and computational efficiency of subsurface modeling. Despite the substantial progress, we recognize the persistent challenges, such as the need for more robust coupling techniques, the balance between model complexity and computational feasibility, and effectively combining machine learning with traditional physical models. Promising directions for future work are discussed to address these challenges, aiming to push the boundaries of current multiscale modeling capabilities.

Physics-Informed Neural Networks (PINNs) gains attentions as a promising approach for applying deep neural networks to the numerical solution of nonlinear partial differential equations (PDEs). However, due to the challenging regions within the solutions of 'stiff' PDEs, e.g., shock front of CO₂ immiscible flooding, adaptive methods are essential to ensure the neural network accurately addresses these issues. In this work, we introduce a novel method for adaptively training PINNs, named Self-Adaptive PINNs (SA-PINNs). This approach employs fully trainable adaptation weights that are applied individually to each training point. Consequently, the neural network autonomously identifies challenging regions of the solution space and focuses its learning efforts on these areas. This method is hereby used to simulate a two-phase immiscible flooding in a low-permeability oil reservoir, with considering gas dissolution and the threshold pressure gradient of oil phase in low-permeability oil reservoirs, i.e., modified Buckley-Leverett (B-L) problem. The model is capable of generating a precise physical solution, accurately capturing both shock and rarefaction waves under the specified initial and boundary conditions, though the introduction of complicated physics increases the nonlinearity of the governing PDEs. The self-adaptive mechanism modifies the behavior of the deep neural network by simultaneously minimizing the losses and maximizing the weights. It, thus, can effectively capture the non-linear characteristics of the solution, thereby overcoming the existing limitations of PINNs. In these numerical experiments, the SA-PINNs demonstrated superior performance compared to other state-of-the-art PINN algorithms in terms of L2 error. Moreover, it was also achieved with a reduced number of training epochs. SA-PINNs can effectively model the dynamics of complex physical systems by optimizing network parameters to minimize the residuals of the PDEs.

The SPE11 comparative solution project presents a benchmark for geological carbon storage in an aquifer, as the development of sufficiently accurate CO₂ sequestration models is critical for predicting the distribution of CO₂ during and after injection. In this paper we present a convergence analysis of the SPE11 benchmark simulation using the Delft Advanced Research Terra Simulator (open-DARTS). Open-DARTS, an open-source simulation framework designed for forward and inverse modeling, as well as uncertainty quantification, employs a unified thermal-compositional formulation and operator-based linearization. In our convergence analysis the SPE11b (2D - reservoir conditions) starts to converge at a grid resolution of 1340 × 240, after which added resolution provides diminishing returns. In addition the three-dimensional SPE11c benchmark is simulated with 8M grid blocks. However, 2D results from SPE11b suggest that a greater resolution is required for a truly converged solution. Furthermore, we extend the SPE11b benchmark to include H₂S as a trace impurity in the injection stream.

Thermal-Hydro-Mechanical-Compositional analysis is crucial for addressing challenges like wellbore stability, land subsidence, and induced seismicity in the geo-energy applications. Numerical simulations of coupled thermo-poromechanical processes provide a general-purpose tool for evaluating these phenomena across laboratory and field scales. However, efficient integration of the coupled equations for fluid mass, energy and momentum poses multiple numerical and implementation difficulties, such as combining different numerical methods on staggered grids and associated limitations on admissible grids. This paper introduces a novel fully-implicit Finite Volume Method (FVM) for modeling thermal compositional flow in thermo-poroelastic rocks. The scheme employs gradient-based, coupled multi-point approximations of fluid mass, momentum and heat fluxes.

The novelty of the scheme lies in its integration of temperature as a parameter in the flux approximation process. The scheme supports a wide range of cell topologies, arbitrary heterogeneity and anisotropy as well as various boundary conditions, while respecting local flux balance under temperature gradients. Overall, the scheme represents a unified FVM-based approach for the integration of all conservation laws relevant to geo-energy applications on a cell-centered collocated grid. Additionally, the implemented two-stage block-partitioned preconditioning strategy enables the efficient solution of obtained linear systems.

The framework, implemented in the open-source Delft Advanced Research Terra Simulator (open-DARTS), leverages the Operator-Based Linearization (OBL) technique for flexibility in compositional fluid properties. Rigorous validation demonstrates the framework’s capabilities in capturing advanced phenomena, including thermal expansion, thermo-poroelastic effect and compositional flow with phase transitions. The performance of preconditioning strategy is assessed using the mechanical extension of the SPE10 benchmark model.

In this work, we present a thermal-compositional simulation framework for modelling of CO₂ sequestration in de- pleted hydrocarbon reservoirs. The parametrization technique utilizes thermodynamic state-dependent operators expressing the governing equations for the thermal-compositional system to solve the nonlinear problem. This approach provides flexibility in the assembly of the Jacobian, which allows straightforward implementation of advanced thermodynamics. Taking advantage of the flexibility of operator-based linearization (OBL), multiphase thermodynamic modelling at arbitrary state specifications is implemented. The use of a hybrid-EoS approach to combine equations of state for aqueous and hydrocarbon phases and advanced initialization schemes for multi- phase equilibrium calculations improves the accuracy and efficiency of the simulation. Careful phase identifica- tion is required for the simulation of multiphase flow, in particular with the potential occurrence of multiple liquid phases in CO₂-hydrocarbon mixtures. We apply the simulation framework to model a set of CO₂ injection cases at conditions typical for depleted hydrocarbon fields. We demonstrate that important thermophysical phenomena resulting from the interaction of CO₂ and impurities with reservoir fluids can be accurately captured using the OBL approach. The consistency of compositional simulation is supported by robust and efficient modelling of multiphase equilibria between brines, hydrocarbons and CO₂. The method is shown to be robust for capturing the thermal effects related to expansion, mixing and phase transitions. This work presents a highly flexible and efficient framework for modelling of multiphase flow and transport in CCUS-related subsurface applications. Ro- bust modelling of thermodynamic equilibria at arbitrary state specification captures the complex thermophysical interactions between CO₂ and reservoir fluids.

Pore pressure fluctuation in subsurface reservoirs and its resulting mechanical response can cause fault reactivation. Numerical simulation of such induced seismicity is important to develop reliable seismic hazard and risk assessments. However, modeling of fault reactivation is quite challenging, especially in the case of displaced faults, i.e., faults with non-zero offset. In this paper, we perform a systematic benchmarking study to validate two recently developed numerical methods for fault slip simulation. Reference solutions are based on a semi-analytical approach that makes use of inclusion theory and Cauchy-type singular integral equations. The two numerical methods both use finite volume discretizations, but they employ different approaches to represent faults. One of them uses a conformal discrete fault model (DFM) while the other employs an embedded (non-conformal) fault model. The semi-analytical test cases cover a vertical frictionless fault, and inclined displaced faults with constant friction and slip-weakening friction. It was found that both numerical methods accurately represent pre-slip stress fields caused by pore pressure changes. Moreover, they also successfully cope with a vertical frictionless fault. However, for the case with an inclined displaced fault with a constant friction coefficient, the embedded method can not converge for the post-slip phase, whereas the DFM successfully coped with both constant and slip-weakening friction coefficients. In its current implementation, the DFM is therefore the model of choice when accurate simulation of local faulted systems is required.

Injection of high-pressure CO2 into depleted gas reservoirs can lead to low temperatures promoting formation of hydrate in the near wellbore area resulting in reduced injection rates. The design of effective mitigation methods requires an understanding of the impact of crucial parameters on the formation and dissociation of CO2 hydrate within the porous medium under flowing conditions. This study investigates the influence of water saturation (ranging from 20 % to 40 %) on the saturation and kinetics of CO2 hydrate during continuous CO2 injection. The experiments were conducted under a medical X-ray computed tomography (CT) to monitor the dynamics of hydrate growth inside the core and to calculate the hydrate saturation profile. The experimental data reveal increase in CO2 hydrate saturation with increasing water saturation levels. The extent of permeability reduction is strongly dependent on the initial water saturation: beyond a certain water saturation the core is fully blocked. For water saturations representative of the depleted gas fields, although the amount of generated hydrate is not sufficient to fully block the CO2 flow path, a significant reduction in permeability (approximately 80 %) is measured. It is also observed that the volume of water + hydrate phases increases during hydrate formation, indicating a lower-than-water density for CO2 hydrate. Having a history of hydrate at the same water saturation leads to an increase in CO2 consumption compared to the primary formation of hydrate, confirming the existence of the water memory effect in porous media.

CO2 storage in deep saline aquifers is an effective strategy for reducing greenhouse gas emission. However, salt precipitation triggered by evaporation of water into injected dry CO2 causes injectivity reduction. Predicting the distribution of precipitated salts and their impact on near-well permeability remains challenging. Therefore, a detailed investigation of the interactions between salt precipitation and porous domain is essential for of revealing the mechanisms of pore blockage due to salt crystallization. Through series of microfluidic experiments, direct observations, coupled with detailed imaging processing, form the basis for explaining these phenomena and provide a relationship between water and salt saturations, highlighting the critical roles played by local capillary-driven flow and water film along grains in influencing water relocation. The results reveal two distinct types of salt crystallization: occurring inside the brine with smooth edges and at the CO2-brine interface with rough edges. Furthermore, the impact of local heterogeneity and surface wettability on salt precipitation patterns is discussed. The transition region between the porous domains and inlet/outlet channels exhibits brine backflow and a larger amount of salt accumulation. This paper presents a comprehensive analysis of the dynamic process of salt dry-out occurring during CO2 injection at the pore scale.

The reduction of temperature caused by Joule-Thomson effect during injection of CO2 in low pressure reservoirs combined with presence of water can lead to formation of hydrates, which in turn reduces rock permeability and hence CO2 injectivity. This paper introduces an empirical model to evaluate impact of hydrate formation on injectivity of CO2 injection wells. Experiments were also conducted to validate the model. The model was then used to simulate injection of CO2 into a multi-layered depleted gas field. The results indicate that operational parameters, particularly CO2 injection rate and temperature, have a large influence on hydrate formation. This is because a higher CO2 injection rate leads to a greater pressure drop within the injection well, potentially triggering conditions conducive to hydrate formation. It is also shown that the dynamics of the competition between the dry-out and temperature fronts play an important role in the final saturation of the hydrate within porous media. For large evaporation rates, the evaporation of water reduces water saturation near wellbore and hence formation of hydrates is limited.

Carbon dioxide capture and storage in subsurface geological formations is a potential solution to limit anthropogenic CO₂ emissions and combat global warming. Depleted gas fields offer significant CO₂ storage volumes; however, injection of CO₂ into these reservoirs poses some potential challenges for the injectivity, containment and well/facility integrity due to low temperatures caused by isenthalpic expansion of CO₂. A key injectivity risk is due to possible formation of hydrates at the low expected temperatures. This study aims to address main causes of CO₂ hydrate formation and its impact on permeability of porous media. This review highlights the current state of knowledge in the literature while emphasizing the need to bridge existing gaps in derisking CO₂ injection into (depleted) low-pressure gas reservoirs. In summary, according to the existing literature, the potential for hydrate formation is assessed to be credible. Current industry solutions exist to manage this risk; however, they are costly and energy intensive. Future research will be needed to provide capabilities to manage this risk more efficiently.

This study investigates the integration of machine learning (ML) and data assimilation (DA) techniques, focusing on implementing surrogate models for Geological Carbon Storage (GCS) projects while maintaining the high fidelity physical results in posterior states. Initially, we evaluate the surrogate modeling capability of two distinct machine learning models, Fourier Neural Operators (FNOs) and Transformer UNet (T-UNet), in the context of CO₂ injection simulations within channelized reservoirs. We introduce the Surrogate-based hybrid ESMDA (SH-ESMDA), an adaptation of the traditional Ensemble Smoother with Multiple Data Assimilation (ESMDA). This method uses FNOs and T-UNet as surrogate models and has the potential to make the standard ESMDA process at least 50% faster or more, depending on the number of assimilation steps. Additionally, we introduce Surrogate-based Hybrid RML (SH-RML), a variational data assimilation approach that relies on the randomized maximum likelihood (RML) where both the FNO and the T-UNet enable the computation of gradients for the optimization of the objective function, and a high-fidelity model is employed for the computation of the posterior states. Our comparative analyses show that SH-RML offers a better uncertainty quantification when compared to the conventional ESMDA for the case study.

In this study, we conduct a comprehensive history matching study for the FluidFlower benchmark model. This benchmark was prepared and organized by the University of Bergen, the University of Stuttgart, and Massachusetts Institute of Technology, for promoting understanding of the complex physics of geological carbon storage (GCS) through in-house experiments and numerical simulations. This paper synthesizes the experiences of history matching the benchmark data encountered by the Delft-DARTS and CSIRO participants. History matching is first performed based on a low-dimensional-zonated structured model using a simple Poisson-like solver. The permeability of six facies and two faults is inferred in this stage to match the digitized concentration data. The history matching is then further enhanced to richer spatial and physical models to capture the spatial variation of permeability and buoyancy effects, using an unstructured grid. Efficient adjoint methods are used to evaluate the gradient used in the optimization of data misfits or equivalent Bayesian log-likelihoods. With efficient optimization methods available for both maximum a posteriori model inference and Randomized Maximum Likelihood methods for model uncertainty, we perform history matching using both binary and continuous concentration observations. The results show that the tracer plumes in the enriched model match the experimental plumes more accurately compared with the results from the parsimonious-zonated model. The history matching results based on the concentration observations provide more similar plume shapes compared with the case based on the binary observations. The permeability difference between the model before and after history matching reveals that the tracer plume zone and the high permeable zone are the regions of high sensitivity in terms of data misfit between the model response and observations. Surprisingly, CO ₂ concentration plume forecasts based on these history-matched models were not especially sensitive to the improvements observed in the enhanced model.

Carbon capture and storage is an essential technology to mitigate anthropogenic CO2 emissions from carbon-intensive industries. To model CO2 injection, physics-based numerical methods are computationally intensive due to the nonlinear nature of the governing equations. Therefore, several data-driven deep learning methods have been developed to serve as proxies and replace numerical simulations. These proxies have demonstrated significantly faster runtimes while maintaining comparable accuracy to numerical simulations. This makes them suitable for high-fidelity models and ensemble-based techniques that require a large number of forward runs. Our method utilizes physics-informed neural networks (PINNs) to parameterize the solution space of immiscible compositional problems. The PINN parameterizes the forward solution of the compositional problem based on the composition of the upstream grid block at the updated time step, the composition of the current grid block at the current time step and the total velocity at their interface. The neural network is trained in the entire solution space and is used in a sequential, cascading solver. In this approach, we obtain the pressure solution first before solving for transport by treating the reservoir as a series of two-cell problems. The resulting transport solver is applicable to all problems with different initial/injection conditions and different heterogeneous reservoirs. We demonstrate our approach for binary and multicomponent problems and furthermore use multilinear interpolation to compare and validate the solution method.

The data assimilation process for geothermal reservoirs often relies on well data, which primarily offer insights into the immediate vicinity of the borehole. However, integrating geophysical methods can provide valuable information beyond well proximity, possibly enhancing reservoir predictions. Current methods of monitoring geothermal reservoirs struggle to maintain a good signal-to-noise ratio for deep reservoirs. Diffusive electromagnetic (EM) methods can be sensitive to the decreasing conductivity from heat extraction in geothermal reservoirs and offer promising additional value. To test their potential effectiveness, numerical examples are simulated. A scheme to incorporate diffusive EM observations into a data assimilation process for geothermal reservoirs is presented and implemented in this study. First, an ensemble of prior models representing the reservoir uncertainty is used to determine the moments of the resulting temperature field using a forward geothermal simulation. Subsequently, a conductivity model is calculated from the temperature field using an empirical relationship. The expected electric field response can then be simulated using an EM forward model. EM sources are placed on the surface around the expected cold plume location. The receiver is placed at reservoir depth. To assimilate the data, the ensemble smoother with multiple data assimilation method is used. The findings demonstrate that the incorporation of EM data provides valuable information regarding the temperature field. This improves the accuracy of the temperature forecast of the entire reservoir when combined with the localized data from the production well and, therefore, helps to resolve the complex migration of the cold front. These results highlight the monitoring potential of EM observations for geothermal reservoirs.

The role of Thermal-Hydro-Mechanical-Compositional analysis in the development of geo-energy resources has been amplified in recent years. As an example, challenges such as wellbore stability, land subsidence and induced seismicity highlight the necessity for comprehensive geomechanical evaluations which are then coupled with thermo-hydrodynamical processes within the reservoir. Numerical simulations of the coupled thermo-poromechanical processes provide a general-purpose tool capable of performing these evaluations at both continuum laboratory and field scales. However, efficient integration of the coupled system of fluid mass, energy and momentum conservation equations poses multiple numerical and implementation difficulties, such as combining different numerical methods on staggered grids and associated limitations on admissible grids. This paper introduces a new fully-implicit scheme of the Finite Volume Method (FVM) for modeling thermal compositional flow in thermo-poroelastic rocks. The scheme uses the gradient-based variant of coupled multi-point approximations of fluid mass, momentum, heat convection and conduction fluxes, which are derived from their respective local balances. The novelty of the scheme is that it incorporates temperature into the approximation of these fluxes. Consequently, the approximation of displacement gradients depends on temperatures, while the approximation of temperature itself is derived from the balance of heat conduction fluxes. At the same time, we utilize a single-point upstream weighting for the temperature-dependent terms in heat convection fluxes. The resulting scheme respects the local balance of fluxes in the presence of temperature gradients. Besides, it also supports star-shaped and various boundary conditions. Overall, the scheme represents a unified FVM-based approach for the integration of all conservation laws relevant to geo-energy applications on a cell-centered collocated grid. Furthermore, the implemented two-stage block-partitioned preconditioning strategy enables the efficient solution of obtained linear systems. The proposed modeling framework has been implemented in an open-source Delft Advanced Research Terra Simulator (DARTS). Moreover, the flexibility regarding compositional fluid properties is reinforced by the Operator-Based Linearization (OBL) technique incorporated into DARTS. The proposed modeling framework has undergone rigorous validation in convergence study, and comparisons against established analytical and numerical solutions. The framework covers advanced physical phenomena including thermal expansion and contraction, porosity dependent on pressure, temperature and strain, and multiphase flow with phase changes and chemical alterations. The framework capabilities and the performance of the preconditioning strategy have been assessed in the mechanical extension of the 10th SPE Comparative study (SPE10) model.

In this comprehensive study, we discuss a novel approach to enhance data assimilation and uncertainty quantification in the field of Geological Carbon Sequestration (GCS). We specifically address the complexities of channelized reservoirs, which pose significant challenges due to non-Gaussian permeability distributions and the intricate non-linear physics of CO2 injection processes. Our innovative method integrates Fourier Neural Operators (FNOs) and Transformer UNet (T-UNet) with advanced data assimilation techniques - the Surrogate-based Hybrid Ensemble Smoother with Multiple Data Assimilation (SH-ESMDA) and the Surrogate-based Hybrid Randomized Maximum Likelihood (SH-RML). These techniques make use of the very efficient computation of gradients that neural networks provide and they not only improves the speed of data processing but also enhances the accuracy of predictions in synthetic data assimilation experiments for GCS applications. A key element of our approach is the use of proxy models alongside high-fidelity simulations, ensuring the consistency and reliability of physical posterior distributions. We utilized Alluvsim for detailed geological modeling and the Delft Advanced Research Terra Simulator (DARTS) for comprehensive fluid flow simulations, providing a comprehensive understanding of reservoir dynamics. A synthetic case study on a channelized reservoir model for CO2 sequestration demonstrates the effectiveness of these methods, with improvements in predicting CO2 plume migration and pressure dynamics within the reservoir. The results of our study show that the integration of FNOs and T-UNet with SH-ESMDA and SH-RML leads to enhanced prediction capabilities, particularly in the challenging context of channelized reservoirs. The SH-ESMDA method proves to be highly efficient in speeding up the data assimilation process without compromising accuracy, while SH-RML demonstrates a more effective history matching compared to standard Ensemble Smoother with Multiple Data Assimilation (ESMDA) techniques, indicating a robust strategy for assimilating complex data. This research not only contributes to the realm of GCS but also presents a novel solution for the integration of artificial intelligence with traditional methodologies that can be applied in various fields where data assimilation and uncertainty quantification are crucial. Our study paves the way for future advancements in this domain, highlighting the potential of AI-driven techniques in enhancing data assimilation and uncertainty quantification for GCS projects.