Abstract Understanding CO2 nanobubble formation in water‐saturated sandstone is critical for understanding fluid behavior in CO2 storage systems. Here, CO2 exsolution from an aqueous phase in a sandstone was investigated using small‐angle neutron scattering at 50°C during cyclic depressurization from 12 to 0.7 MPa. Nanoscale heterogeneities consistent with CO2 clusters and nanobubbles (5–200 nm) were resolved during pressure reduction. Although bulk phase diagrams predict exsolution at ∼8 MPa at 50°C, detectable exsolution emerged only at 2.4 MPa, indicating strong confinement and surface effects. A progressive loss of signatures associated with nanobubbles <∼15 nm suggests preferential disappearance of nanobubbles, consistent with curvature‐driven coarsening (e.g., Ostwald ripening). Repeated cycling revealed partial qualitative reversibility, implying nucleation and saturation‐history (hysteresis) effects relevant to operational pressure transients in CO2 storage. Our findings improve assessment of CO2 mobility, trapping, and leakage risk under dynamic pressure in CO2 storage systems.

Flow and transport in fractured geological media are strongly controlled by aperture heterogeneity and uncertainty in subsurface characterisation, yet most upscaling approaches rely on deterministic representations of fracture permeability. This study presents a scalable probabilistic workflow that bridges image-based fracture geometry and uncertainty-aware hydraulic predictions across scales. The approach integrates Bayesian correction of aperture-permeability model misspecification, a deep learning surrogate for predicting spatially distributed permeability statistics, and Darcy-scale flow upscaling to propagate uncertainty to effective transmissivity. The workflow is applied to natural shear fractures from core material in the Little Grand Wash Fault damage zone (Utah) and to simplified geometries derived from the same datasets. The Bayesian component quantifies uncertainty due to measurement errors and imperfect constitutive relations, while a Residual U-Net learns the effects of local heterogeneity and spatial correlation on predicted permeability uncertainty. Together, these components generate ensembles of permeability fields that are subsequently upscaled to probabilistic macroscopic flow responses. Results show that common empirical aperture-permeability relations are systematically biased for natural fractures, whereas the proposed probabilistic workflow yields uncertainty-aware permeability estimates consistent with physics-based behaviour. The method captures the impact of channelisation, connectivity, and complex 3D void geometries on transmissivity while quantifying the resulting uncertainty bounds. Computational efficiency arises from the proposed hybrid strategy for probabilistic upscaling, which combines physics-informed and data-driven approaches, preserves Stokes-flow consistency and supports uncertainty propagation without repeated high-fidelity simulations.

This study extends the vertical equilibrium (VE) modeling framework to simulate multi-phase flow involving CO₂, methane, and brine in depleted gas reservoirs. Methane presence introduces complexity not captured in traditional VE models. The proposed model integrates a black-oil approximation with VE assumptions, reducing dimensionality and enabling rapid simulation of large-scale CO₂ injection. By representing CO₂ and methane as separate compressible phases, and treating brine as incompressible, the model balances computational speed and physical realism. Pressure-dependent density and viscosity relationships for CO₂ and methane enable accurate property estimation without flash calculations. Three case studies validate the model: (1) a flat, layered reservoir benchmark demonstrating accurate CO₂ plume behavior under buoyancy control; (2) a dual-anticline scenario demonstrating that the VE model reproduces the large-scale plume geometry and inter-anticline mass redistribution observed in compositional simulations; and (3) a field-scale application to the Hugin Formation East, where the VE model replicates key features of CO₂ and methane migration observed in compositional models. The VE model runs orders of magnitude faster than full 3D simulations while capturing large-scale displacement dynamics. Although the formulation neglects explicit mixing, dissolution, and capillary fringe effects, it captures first-order displacement behaviour with speedups up to 100×. This makes it a practical tool for screening, uncertainty quantification, and optimisation in CCS projects targeting depleted gas reservoirs.

The comparison between laboratory-induced and subsurface fractures, and their corresponding flow is still unclear. Here, we examine three natural shear fractures and two induced tensile fractures from the same low-permeability lithology. Using high-resolution synchrotron imaging, we extracted three-dimensional fracture void geometries to analyse aperture distributions, surface roughness, and spatial correlation patterns. We then compared measured fracture transmissivities against theoretical predictions from parallel-plate models (cubic law) and direct numerical simulations (DNS) of flow to evaluate consistency and uncertainties. We find that despite differences in heterogeneity, induced tensile and natural shear fractures can generally yield similar flow in caprocks. Our comparisons further indicate that the choice of flow estimation method can introduce more uncertainty than the fracture opening mode.

Simulating the fluid flow along fault zones at different scales is essential for predicting the CO2 leakage and containment during injection and storage. However, this can be challenging, especially in the early stages of a storage project when knowledge of the reservoir and caprock is limited and the cost of obtaining the relevant data is high. This study proposes a tool for fast screening of fault leakage at the site screening stage. The tool uses a vertically integrated reservoir model coupled with an upscaled fault leakage function based on source/sink relations. The fault is conceptualized as an increased vertical permeability through the caprock due to the presence of a fracture network in the damage zone and a reduced horizontal permeability in the reservoir due to fault throw and presence of a low-permeability fault core. Simulation results of various CO2 injection scenarios in a reservoir with potential for fault leakage demonstrate that the tool can produce physically consistent leakage predictions. The computationally efficient model presented in this study is a valuable tool for quantifying uncertainties in key fault parameters, and other constitutive relations that affect the behavior of the storage reservoir and potential fault leakage. By incorporating this tool into the site screening stage, stakeholders can quickly screen the risk of CO2 leakage along faults across a range of possible storage sites and subsequently design targeted data acquisition campaigns to better characterize and model the faults. Overall, the proposed tool is a cost-effective and efficient method for screening fault leakage risk during CO2 injection and storage, helping to ensure safe and effective carbon storage.

Abstract Accurately capturing the complex interaction between CO2 and water in porous media at the pore scale is essential for various geoscience applications, including carbon capture and storage (CCS). We introduce a comprehensive dataset generated from high-fidelity numerical simulations to capture the intricate interaction between CO2 and water at the pore scale. The dataset consists of 624 2D samples, each of size 512 × 512 with a resolution of 35μm, covering 100 time steps under a constant CO2 injection rate. It includes various levels of heterogeneity, represented by different grain sizes with random variation in spacing, offering a robust testbed for developing predictive models. This dataset provides high-resolution temporal and spatial information crucial for benchmarking machine learning models.

Over the past decade, geological carbon storage (GCS) has developed into a deployable technology, with modeling and simulation playing a central role in site permitting, operations, and risk assessment. This paper synthesizes a decade of progress in CO2 storage modeling, with a focus on field-scale modeling and incorporation into decision support and regulatory workflows. We summarize advances across the modeling spectrum, from reduced-order screening tools to compositional flow simulation and coupled-multi-physics models, and discuss how uncertainty quantification, data assimilation, and emerging machine-learning methods are being integrated to support decision-making and regulatory filings. Particular attention is given to boundary condition implementation, upscaling from core to field scale, dynamic fault modeling and reactivation assessment, trapping mechanisms, and CO2 storage in depleted hydrocarbon reservoirs. Case studies from major GCS projects illustrate how simulation tools are applied and refined through monitoring data. Despite significant progress, remaining challenges relate to how they are applied, particularly overconfidence, insufficient sensitivity analyses, limited probabilistic uncertainty quantification, and inconsistencies in regulatory expectations. We conclude by identifying key research gaps and priorities needed to support the safe, reliable, and scalable deployment of GCS.

Different operations dealing with the subsurface, such as subsurface CO2 disposal, hazardous waste disposal, geothermal energy extraction, underground hydrogen storage, etc., can change the fluid/flow system underground. The injection of fluids with thermodynamic and chemical properties different from those of the reservoir fluid can trigger a series of chemical reactions, which may affect the fluid and/or rock properties. Depending on the system under study, these changes may be advantageous or unfavorable. Reactive transport modeling is a choice for investigating how these changes can alter the system. In this study, a reactive transport solver is developed in the MATLAB Reservoir Simulation Toolbox using the sequential fully-implicit approach. The developed reactive transport solver is illustrated using reactions and geometries using reactions and geometries relevant for assessing the sealing capacity of a fractured caprock of a deep saline aquifer used for underground CO2 disposal, and the limitations and advantages of the approach are stated. Moreover, the results of the simulation for two fracture models, the discrete fracture matrix and embedded discrete fracture matrix models, are compared. The simulations demonstrate that hydrogen ion concentration or pH is the primary parameter affecting the extent of dissolution, while the other aqueous species concentrations are less influential. It is also shown that at higher flow rates, dissolution substantially occurs in the vicinity of the main fracture, along the flow direction, while at lower flow rates, because the injected fluid becomes fully buffered closer to the inlet, dissolution only occurs in the vicinity of the inlet over the course of the simulation. Applying the discrete fracture matrix and embedded discrete fracture matrix models to one of the scenarios demonstrates that both yield equivalent results.

The UNet-enhanced Fourier Neural Operator (UFNO) extends the Fourier Neural Operator (FNO) by incorporating a parallel UNet pathway, enabling the retention of both high- and low-frequency components. While UFNO improves predictive accuracy over FNO, it inefficiently treats scalar inputs (e.g., temperature, injection rate) as spatially distributed fields by duplicating their values across the domain. This forces the model to process redundant constant signals within the frequency domain. Additionally, its standard loss function does not account for spatial variations in error sensitivity, limiting performance in regions of high physical importance. We introduce UFNO-FiLM, an enhanced architecture that incorporates two key innovations. First, we decouple scalar inputs from spatial features using a Feature-wise Linear Modulation (FiLM) layer, allowing the model to modulate spatial feature maps without introducing constant signals into the Fourier transform. Second, we employ a spatially weighted loss function that prioritizes learning in critical regions. Our experiments on subsurface multiphase flow demonstrate a 21% reduction in gas saturation Mean Absolute Error (MAE) compared to UFNO, highlighting the effectiveness of our approach in improving predictive accuracy.

Structural uncertainties and unresolved features in fault zones hinder the assessment of leakage risks in subsurface CO2 storage. Understanding multi-scale uncertainties in fracture network conductivity is crucial for mitigating risks and reliably modelling upscaled fault leakage rates. Conventional models, such as the Cubic Law, which is based on mechanical aperture measurements, often neglect fracture roughness, leading to model misspecifications and inaccurate conductivity estimates. Here, we develop a physics-informed, AI-driven correction of these model misspecifications by automatically integrating roughness effects and small-scale structural uncertainties. Using Bayesian inference combined with data-driven and geometric corrections, we reconstruct local hydraulic aperture fields that reliably estimate fracture conductivities. By leveraging interactions across scales, we improve upon traditional empirical corrections and provide a framework for propagating uncertainties from individual fractures to network scales. Our approach thereby supports robust calibration of conductivity ranges for fault leakage sensitivity analyses, offering a scalable solution for subsurface risk assessment.

Simulating reactive dissolution of solid minerals in porous media has many subsurface applications, including carbon capture and storage (CCS), geothermal systems and oil gas recovery. As traditional direct numerical simulators are computationally expensive, it is of paramount importance to develop faster and more efficient alternatives. Deep-learning-based solutions, most of them built upon convolutional neural networks (CNNs), have been recently designed to tackle this problem. However, these solutions were limited to approximating one field over the domain (e.g. velocity field). In this manuscript, we present a novel deep learning approach that incorporates both temporal and spatial information to predict the future states of the dissolution process at a fixed time-step horizon, given a sequence of input states. The overall performance, in terms of speed and prediction accuracy, is demonstrated on a numerical simulation dataset, comparing its prediction results against state-of-the-art approaches, also achieving a speedup around 10^4 over traditional numerical simulators.

The Malay Basin has received significant attention for geological carbon dioxide storage (GCS), but there are no published studies addressing the selection of appropriate deep saline aquifers. This study closes this gap. We process spatial data and use geological modelling and cluster analysis to identify optimal areas for GCS, considering various subsurface characteristics such as temperature, pressure, porosity and thermophysical CO2 properties. It is found that the basin contains numerous Cenozoic aquifers suitable for GCS including locally thick, but low net-to-gross (NTG), stacked formations. Pliocene aquifers are too shallow to offer storage for CO2 in large quantities, but upper Miocene aquifers located in the northwest of the basin contain promising intervals with significant porosities and conditions favouring denser CO2. Middle Miocene aquifers, while low NTG, are thick, and optimally located around the margins of the basin. They also have significant storage capacity and could be developed as a stacked GCS site. Lower Miocene aquifers are higher NTG, but deeply buried across many areas of the basin, yet the oldest aquifer evaluated still holds substantial storage capacity, where subject to minor burial at the margins of the basin. Overall, this study provides a novel first assessment of aquifer GCS potential in the Malay Basin, while also contributing to wider efforts to evolve screening workflows for other geological basins.

Determining the (in)efficiency of wetting phase displacement by an invading non-wetting phase (drainage) in a single fracture is key to modelling upscaled properties such as relative permeability and capillary pressure. These constitutive relationships are fundamental to quantifying the contribution, or lack thereof, of conductive fracture systems to long-term leakage rates. Single-fracture-scale modelling and experimental studies have investigated this process, however, a lack of visualization of drainage in a truly representative sample at sufficient spatial and temporal resolution limits their predictive insights. Here, we used fast synchrotron X-ray tomography to image drainage in a natural geological fracture by capturing consecutive 2.75 μm voxel images with a 1 s scan time. Drainage was conducted under capillary-dominated conditions, where percolation-type patterns are expected. We observe this continuously connected invasion (capillary fingering) only to be valid in local regions with relative roughness, λb ≤ 0.56. Fractal dimension analysis of these invasion patterns strongly aligns with capillary fingering patterns previously reported in low λb fractures and porous media. Connected invasion is prevented from being the dominant invasion mechanism globally due to high aperture heterogeneity, where we observe disconnected invasion (snap-off, fragmented clusters) to be pervasive in local regions where λb ≥ 0.67. Our results indicate that relative roughness has significant control on flow as it influences fluid conductivity, and thus provides an important metric to predict invasion dynamics during slow drainage.

Geological carbon capture and storage (CCS) is a critical technology for mitigating greenhouse gas emissions, but the risk of leakage remains a significant concern. Fault and fracture networks across sealing intervals are potential pathways for CO2 to escape from storage reservoirs, necessitating accurate assessment of their permeability and connectivity. Our study presents an integrated approach for modelling geological leakage in fault zones, combining single fracture stress-permeability laboratory measurements with detailed fracture outcrop data to simulate in-situ conditions for carbon storage. We studied caprock sequences cut by a normal fault in the Konusdalen West area (Svalbard, Norway), a regional seal for the reservoir of the Longyearbyen CO2 Laboratory, and an analogue to Barents and North seas caprock formations. Digitising the outcropping fracture network, we explored the variations in fracture size distribution and their connectivity in different portions of the fault zone. These parameters are fundamental to establish if the fracture network provides permeable pathways. Integrating outcrop analysis with laboratory measurements allows us to create coupled hydromechanical models of the natural fracture network and to evaluate their upscaled permeability. We found that fracture network geometries vary across the fault zone, resulting in different upscaled permeability models, thus highlighting the importance of including detailed fracture network information into permeability simulations. Our study provides a framework for incorporating fracture permeability measurements and outcrop analysis into the modelling of geological leakage in fault zones, which can inform the design and operation of CCS projects and help mitigate the risks associated with geological storage of CO2.

The Malay Basin is a mature hydrocarbon province that is being re-assessed for CO2 storage. Selecting appropriate storage sites requires a comprehensive understanding of the structural and stratigraphic history of the basin, but previous studies of the basin have been limited to observations from either regional, 2D seismic lines or individual 3D seismic volumes. Here, we access and utilize a basin-wide (~40,000 km2) 3D seismic and well database to describe the structural and stratigraphic features of the basin, particularly those within the uppermost ~ 4 km (Oligocene to Recent), to gain new insights into the evolution of the basin. E-W transtensional rift basins first developed because of sinistral shear across a NW-SE strike-slip zone. The NW-SE basin morphology seen today was generated during the late Oligocene – early Miocene during which time dextral motion across marginal hinge zones created en-echelon antithetic, extensional faults and pull-apart basins, especially well preserved along the western margin of the basin. Collisional forces to the southeast during the early-middle Miocene resulted in shallowing of the basin, intermittent connection to the South China Sea and a cyclic depositional pattern during the Miocene. During the late Miocene, significant uplift of the basin resulted in a major unconformity with up to 4 km of erosion and exhumation in the southeast of the basin. In the centre and northwest region, the inversion of deeper E-W rifts resulted in the folding of Miocene sequences, and the formation of large anticlines parallel to the rift-bounding faults. The Pliocene-Pleistocene history is more tectonically quiescent, but some extensional faulting continued to affect the northwest part of the basin. Larger glacio-eustatic sea-level fluctuations during this time resulted in major changes in sedimentation and erosion on the Sunda Shelf, including the formation of a middle-Pliocene unconformity. These structural events have created a variety of hydrocarbon traps across the basin, of different ages, including transpressional anticlines, rollover anticlines and tilted fault blocks. Each of these has discrete and distinct trap elements, with important implications for their CO2 storage potential.

The development of hydrocarbon (HC) fields with high CO2 content requires Carbon Capture and Storage (CCS) due to the prohibition of CO2 venting in Malaysia. Mitigating CO2 emissions is also important for addressing climate change concerns. There are several potential offshore storage sites in offshore Peninsular Malaysia (PM), such as depleted fields and saline aquifers. The currently identified depleted fields are insufficient to provide the required storage capacity volume for current development plan of HC fields with high CO2 content, and eventually for wider-scale storage of CO2. Therefore, saline aquifers are required to provide additional storage sites. Identifying the most suitable regional saline aquifer for carbon capture and storage (CCS) presents many challenges due to the limited understanding of their distribution and characteristics within the basin. The primary challenge, compared to hydrocarbon fields, lies in ensuring containment and sealing effectiveness, as hydrocarbon fields have already demonstrated entrapment within their structures. To mitigate the risks and uncertainties in identifying regional saline aquifers, it is crucial to analyze the seals and assess the suitability of potential storage sites. This includes evaluating the distribution, continuity, and properties of the storage reservoirs, as well as the distribution, continuity, integrity, and effectiveness of the seals.

Carbon capture and storage (CCS) is vital to reducing greenhouse gas emissions and mitigating climate change. Most CCS projects rely on the permanent geological storage of CO2 within deep sedimentary rock formations, but accurately constraining the capacity of these reservoirs usually involves detailed and computationally demanding reservoir modelling and simulation of the pressure evolution and CO2 plume migration. In the absence of this, efficiency factors are often used within volumetric capacity estimates, but this often results in overestimations of storage capacity. As an alternative, we propose a workflow harnessing various, existing, reduced complexity models that account for the surface topography and dynamic fluid behaviour in a computationally efficient manner. We first undertook a static analysis using algorithms available within MRST-co2lab. The reservoir topography is used to determine the locations of structural traps, the trapping routes that link them and downdip filling areas that feed a given trap. This analysis provides indications of the optimal well placement and helps us refine the total capacity of the area into the capacity available just from structural trapping. We followed this with a dynamic analysis, also within MRST-co2lab, using computationally efficient Vertical Equilibrium models. This efficiency allowed us to performing hundreds of simulations and use these results to map storage efficiency and determine the optimal well placement where efficiency is greatest. We tested this workflow within an area of the Malay Basin with illustrative reservoir parameters and estimated storage efficiency, capacity and the optimal well placement within the area without performing any full-physics simulations. The results from VE modelling indicate that the amount that can be contained within this area is 15 times less than the predictions using static storage efficiency factors. The advantage of such a light approach is that sensitivity and uncertainty analysis can be carried out at speed, before targeting certain parameters/areas for more detailed study.

Carbon capture and storage is vital for reducing greenhouse gas emissions and mitigating climate change. Most projects involve the permanent geological storage of CO2 2 within deep sedimentary rock formations, but accurately constraining storage capacity usually involves detailed and computationally demanding reservoir modeling and simulation. Efficiency factors can also be used but these often lead to capacity overestimations. To address this, a workflow is proposed harnessing various existing, reduced complexity models that account for the surface topography and dynamic fluid behavior in a computationally efficient manner. This workflow was tested in an area of the Malay Basin mapped from three-dimensional seismic data but with illustrative reservoir parameters. A static analysis was first undertaken using algorithms within MRST-co2lab. Structural traps, spill paths and spill regions were identified using the reservoir topography. This provided initial indications into optimal well placement and led to refinement of the total capacity of the area into the capacity available within structural traps. This was followed with a dynamic analysis, also within MRST-co2lab, using computationally efficient Vertical Equilibrium models. Hundreds of simulations were undertaken and the optimal well placement was determined based on the maximum storage efficiency achieved. The results indicated that the amount that can be contained within this area is 15 times less than equivalent predictions using static storage efficiency factors. The advantage of such a light approach is that sensitivity and uncertainty analysis can be carried out at speed, before targeting certain parameters/areas for more detailed study.