The conference will take place at the EICC (Edinburgh International Conference Center) in the heart of Edinburgh, Scotland.
InterPore2023 will include both oral and poster presentations which can be given online and in-person. In-Person oral and poster presentations will be presented in sessions at the conference in Edinburgh. Online presentations will not be scheduled into live sessions, but presenters will have the opportunity to include digital materials and a pre-recorded presentation, with dedicated chat channels available for anytime interaction. More information on the conference format can be found online.
Topics and applications
|
|
Providing clean, safe water reliably in an affordable manner is a major global challenge. A wide variety of water pollutants, including heavy metals, dyes, pesticides, and pharmaceutical compounds pose a threat to public and environmental health. Existing water treatment technologies do not adequately meet water quality standards for removal of the diverse range of contaminants; thus, technological innovation is needed to enhance water security and accessibility. Engineered nanomaterials, such as graphene oxide (GO), offer tunable multifunctionality for effective removal of a diverse range of contaminants from water. However, the practical implementation of nanomaterials such as GO in water treatment requires their immobilization into three-dimensional macrostructures which may impair their performance. Unlike colloidal nanomaterials, solid macrostructures of GO can be easily stored, transported and manipulated. Despite the progress on forming high surface area and multifunctional GO macrostructures, synthesizing mechanically robust porous macrostructures, especially for wet applications, is a challenge. This talk will describe approaches for the preparation of GO-based macrostructures that can be used in water treatment. The functionalization of macrostructures of engineered nanomaterials with antimicrobials for prevention of biofouling or removal of pathogens from contaminated waters will also be discussed.
A safe and efficient hydrogen storage mechanism will be crucial for the successful transition towards a green hydrogen economy. Underground storage of hydrogen can be a viable option for short to long-term storage to meet the fluctuations in energy demand. However, there is limited understanding of the pore-scale displacement and trapping mechanisms for hydrogen-brine systems, especially in heterogeneous rocks at reservoir conditions. Our recent experimental study [1] allowed us to understand the trapping of hydrogen within the pore space of a homogeneous sandstone rock and showed dissolution of hydrogen in the resident brine after injection and production of hydrogen at subsurface conditions. In this work, we build on these findings and use X-ray micro-tomography to study the pore-scale fluid displacement processes during cyclic injection of hydrogen in a layered sandstone rock sample. We investigate how the presence of a thin and low permeability layer between two high permeability zones in the rock sample affects fluid displacement processes and hydrogen trapping. The results indicate that hydrogen preferentially occupies the higher permeability zones, and the residual hydrogen saturation increases in subsequent cycles. The findings from this experiment contribute towards the selection of the most suitable subsurface formations for underground hydrogen storage. Extending our research to perform time-resolved synchrotron X-ray imaging experiments will provide additional insights into the dynamics of pore-scale processes in layered reservoirs during underground hydrogen storage.
The goal of reducing carbon emissions relies heavily on the world’s energy sectors to undergo significant energy transformations. The hydrogen economy plays a critical role in achieving that goal by harvesting hydrogen and using it as an energy carrier. The current storage options limit hydrogen's large-scale adaptation to a major energy form. For that reason, underground hydrogen storage has been an alternative that appealed to the scientific community and prompted multiple studies to explore its feasibility in several aspects. One of those aspects is evaluating the potential of remobilising trapped gases in a porous medium through the Ostwald ripening phenomenon. As such, we examined the phenomenon of Ostwald ripening by leaving the H2-brine system in a sandstone sample uninterrupted for 12 hours to observe any hydrogen re-distribution. The sample was scanned with a micro-CT twice: before and after. Additionally, we demonstrated the derivation of a simple equation that estimates the timescale for disconnected gas ganglia to reach partial equilibrium over a given length scale. Finally, we explored whether changes in the interfacial curvature, in-situ contact angles, saturation distribution, and gas ganglia size distribution occurred during the 12 hours.
We observed the re-distribution of gas ganglia and the emergence of multiple larger new gas ganglia with a maximum extent of about 2 mm. This confirmed the length scale estimation made by the derived equilibrium timescale equation. Additionally, a slight increase in curvature was observed after 12 hours, and the mean contact angle slightly increased at the bottom part of the sample. Overall, our experimental study of the Ostwald ripening phenomenon presented significant remobilisation of trapped gas ganglia and gas re-distribution at the pore scale, which could mean that residually trapped hydrogen can be less than what is thought.
Objective:
The success of large-scale geological storage of gases requires proper understanding of the interfacial behavior among the participating phase. In this work a systematic study on the impact of pressure and brine salinity on the interfacial tension (IFT) of binary H2-brine systems as well as wettability within ternary systems comprising H2-brine-shale are investigated. Furthermore, the shale adsorption capacity of H2 is measured at elevated pressures up to 30 MPa. The conversion of the organic matter at elevated temperatures and elevated pressures under H2 atmosphere has also been examined to understand the role of hydrogenation in upgrading shale oil products.
Methods:
A high-pressure high-temperature view cell with a Pmax of 69 MPa and a Tmax of 200 °C (Eurotechnica GmbH,Germany) was employed to measure the IFT using the pendant drop method. The view cell was also employed to measure the wettability using the sessile drop method. A magnetic suspension balance (MSB) with Pmax of 40 MPa, Tmax of 150 °C (Rubotherm GmbH, Germany) was used to measure the adsorption of H2 on shale based on the gravimetric method. Thermal gravimetric analysis (TGA) was conducted using an MSB with Pmax of 15 MPa and 400 °C (IsoSORP, Waters TA instruments, Germany). The products of the TGA were analyzed using Nuclear Magnetic resonance (NMR) (Avance III 600 MHz-Bruker, Czech Republic) and Gas Chromatography (GC) (Varian 320-Agilent, United States).
Results, Conclusions:
The reduction in IFT upon increasing the pressure was insignificant. Further on, wetting tests suggest that the system is water wet under all experimental conditions. Both these findings guarantee the structural storage integrity of the shale. It is also found that adsorption plays a role in H2 storage within the shales. The NMR and GC analyses reveal that aliphatic compounds are excessively present in comparison to aromatics and olefinic compounds. In an H2 environment, it was observed that aromatic substitution by aliphatic hydrocarbons took place.
Novelty:
This work presents information on H2 wettability and H2 adsorption capacity of shale at conditions relevant to gas storage which are severely lacking in the literature. Furthermore, and up to the best of the author’s knowledge, the conversion of organic matter at H2 pressures relevant to gas-storage is introduced for the first time.
The decarbonization of energy mainly requires the substitution of fossil fuels with low-carbon alternatives. Heavy industries require high-temperature heat that cannot be supplied through electricity. Moreover, the production of renewable electricity requires a storage medium to compensate for their intermittent behaviour. Hydrogen is a favorable medium for storing the excess low-carbon electricity and can accommodate the high temperature requirements. Subsurface hydrogen storage provides the mean to safely store and re-use the hydrogen gas. A successful storage project requires accurate modeling of the hydrogen movement and the extent of its loss.
Flow of hydrogen in porous media containing water is affected by hysteresis in flow properties, mainly relative permeability and capillary pressure. This hysteretic behaviour is a consequence of changes in contact angle and capillary trapping of non-wetting phase in porous media. As a result, the amount of hydrogen trapped in underground increases overtime, causing significant hydrogen loss. Most of the available literature have only considered the hysteresis in relative permeability and have not studied the path dependency of capillary pressure. Moreover, studies focused on the impact of hysteresis on hydrogen storage and hydrogen loss due to trapping is still scarce.
The overall aim of this study is to model the hysteresis effect during the two-phase flow of hydrogen-brine. The outstanding contribution of this work would be considering the capillary pressure hysteric behaviour through generating the full scanning curves during the injection/production cycles for each grid cell. At the next stage, we will look into the relative permeability hysteresis to investigate their individual and mixed influence on the hydrogen trapping in subsurface. Finally, we aim to perform a sensitivity analysis on the controlling parameters (rates, shut-ins and so on) to derive the most optimized scenarios for a successful storage operation. Due to the cyclic nature of the system, we speculate hydrogen loss because of increased trapped hydrogen during the consecutive cycles. We also expect the capillary pressure to be a less contributing parameter compared to relative permeability as the field-scale nature of the system is less affected by capillarity [1].
Hydrogen is considered a low-carbon fuel that can potentially contribute to the large-scale decarbonization of different sectors, including power generation, heating, transportation, and industry. Blending hydrogen into national gas distribution networks can also help decarbonize distributed carbon emissions from domestic consumers where carbon capture is not feasible. More pilot projects worldwide, such as H100 Fife and HYDeploy2 in the UK and HyGrid in Long Island, US [1–3], are showcasing the use of hydrogen in the national gas networks. The reliable and robust operation of a gas network at the national scale (12 to 51TWh of hydrogen in the case of the UK national gas network [4]) would critically require safe and efficient large-scale hydrogen storage. As an example, to meet the UK's seasonal demand and production variations from intermittent renewable energies, 0.37-1.58 billion cubic meters of hydrogen storage capacity is required which represents 25-105% of the current UK strategic natural gas storage capacity [5]. This means large-scale repurposing of current storage sites (both salt caverns and depleted gas reservoirs) and the development of new sites.
To unlock the large-scale storage capacity of depleted gas reservoirs and saline aquifers for efficient hydrogen storage, understanding the flow and trapping mechanisms of hydrogen at the pore scale in contact with resident fluids is essential. To achieve this, a set of experiments is designed to explore the fluid distribution in cyclic fluid displacements representative of seasonal storage and production of hydrogen in subsurface reservoirs. These cyclic flow experiments are performed in a sandstone core and the fluid distribution at the pore scale is imaged by an X-ray micro-computed tomography (micro-CT) rig with a cubic voxel size of 3x3x3 microns. To achieve the highest imaging resolution with our in-house testing rig, a core sample size of 5mm diameter and 10mm length is selected. The outlet pressure is maintained by the receiving pump at 7.0 MPa. The fluids are injected through the core at a constant flow rate of 5 mL/hr to ensure capillary-dominated flow. Potassium iodide (KI) salt is dissolved as a dopant in the brine to provide effective contrast between the brine, hydrogen, and rock sections of the images. To ensure gravity-stable fronts, hydrogen is injected from the top and brine from the bottom of the core holder. The alternate injection of hydrogen and brine is then performed until no significant change in saturations is observed along the sample. After gas injection an average hydrogen saturation of 32% is observed in the core sample, however, about 6% fluctuation in saturations is observed. The core is then isolated for 30 days, and the imaging experiment was repeated. The amount of the displaced hydrogen by buoyancy effect and its distributions is then discussed. This provides insight into hydrogen displacement by gravity forces. Ongoing studies are investigating the trapping of hydrogen at similar flow conditions at the pore and core scale.
Gas injection and withdrawal in the subsurface can be considered as a long-term energy storage solution. Green gas can be produced from the excess electricity during peak production and can subsequently be injected into the surface reservoir and withdrawn during times of high demand. Repeated injection and withdrawal of gas causes capillary pressure hysteresis – in this work we use X-ray tomography to understand the hysteresis phenomenon, which can be applied in operating underground hydrogen storage processes. Two experiments were performed at an unsteady state to investigate gas and water distribution in different pore space geometries during drainage and imbibition cycles. Gas phase was injected into 6 cm long samples of Bentheimer sandstone and Estaillades carbonate at ambient temperature and a pore pressure of 1 MPa, followed by water flooding in three cycles. The gas flow rates decreased from 2 ml/min to 0.08 ml/min while the brine injection rate was keeping constant during the three cycles. We observe and quantify several interesting phenomena including (I) capillary pressure hysteresis, and (ii) hydrogen migration by Ostwald ripening through diffusion of gas dissolved in the brine. We characterise these phenomena by analysing interfacial curvature and area, along with wettability and pore occupancy analysis. This work provides pore-scale insights into hydrogen storage and withdrawal and uses image-based analysis to quantify multiphase flow properties for input into the reservoir-scale simulation.
Underground hydrogen storage (UHS) in porous media offers a long-term and large-scale storage solution which is vital for a sustainable H2 economy. Despite growing interest in the topic, the understanding of the physical processes during cyclic H2 flow is not yet adequate. Here we use microfluidics to experimentally investigate multiple cycles of H2 injection and withdrawal under a range of injection rates at shallow reservoir storage conditions. Our analysis is aimed at qualitative and quantitative description of H2 reconnection mechanisms and hysteresis. We find that H2 storage capacities increase with increasing injection rate. The residual H2 saturation is reproducible between cycles, but its distribution in the pore space visually appears to be hysteretic. In most cases, the residually trapped H2 reconnects in the subsequent injection cycle, predominantly in proximity to the large pore clusters. Our results provide valuable experimental data to advance the understanding of multiple H2 injection cycles in UHS schemes.
Coupled flow and deformation in fractured media is often modeled by the classical dual-porosity poroelasticity theory. The latter is based on the Barenblatt hypothesis of pressure equilibrium inside the rock matrix. This is a reasonable assumption if the characteristic time scales for pressure propagation in the matrix are comparable or smaller than the characteristic fracture time scales. Under large permeability contrasts between the fracture and matrix domains, these conditions may not be met, and the flow and deformation behaviors are dominated by non-equilibrium effects, which manifest in long-tails in flux responses. Using volume averaging, we derive a multicontinuum approach that accounts for pressure non-equilibrium in the rock matrix, and compare it to the classial dual porosity approach. We use explicit analytical solutions to identify the dominant time scales and time regimes, and to evaluate the scaling behaviors of the flux response in consolidation and production scenarios. The flux evolution at a production well is characterized by decay behaviors that are different from the classical dual porosity approach. These behaviors are related in the proposed multicontinuum theory to the permeability contrast and the permeability distribution across the matrix blocks.
Shear displacement of fractures in porous rock leads to fracture dilation influencing the flow field. This is an important mechanism in e.g. enhanced geothermal systems, where the fracture aperture determines the heat extraction performance of the reservoir. To predict shear dilation in a fractured reservoir, the shear displacement needs to be calculated first, since the dilation directly depends on it. This can be done using analytical solutions depending on far field stresses, or with mechanical solvers. Analytical solutions exist for simple test cases of isolated fractures, and approximations from far field stresses calculate the local shear and normal stress on the fracture with Cauchy’s equation. However, we expect that using these two leads to wrong results in complicated fracture patterns, because the interaction between fractures is neglected. This is the main reason why mechanical solvers like boundary element methods, extended finite element methods (XFEM) and extended finite volume methods (XFVM), all of which resolve the mechanics locally by solving for stress equilibrium, were developed.
We compare results based on approximating the local stress at the fractures by the far field stress with those relying on spatially resolved stress fields obtained with a mechanical solver. While the former are computationally much cheaper, the latter are more accurate and more flexible. Our goal is to describe the accuracy and range of application of current cheap approximations regarding shear displacement. To obtain reference solutions we used a solver based on XFVM, in which the fractures are embedded manifolds of lower dimension represented by special discontinuous basis functions. These functions have the property that the displacement gradient is continuous over the fracture segments, which simplifies the computation of traction and compressive forces across the manifold.
The results show that the shear displacement of a single fracture in a rock matrix is well represented by far field stress approximations. In two intersecting fractures the behavior of the fracture slipping at higher pressures is approximated well by using an adaptation depending on the fracture length. In conjugate fractures, on the other hand, the far field approximation overestimates the shear displacements. The importance of locally resolved stresses is highlighted by simulation results of a model with a layer-restricted fracture pattern mapped in the Hornelen basin in Norway, that is, large differences can be observed in the resulting aperture distributions obtained with resolved vs. global stress field approximations.
Modelling of multiple fractures in hydraulic fracturing is of importance for creating a complex fracture network and enhancing the productivity of resources in underground reservoirs. In this work, multiple hydraulic fracturing in low permeability media is studied by extended finite element method (XFEM) and the governing equations for fluid flow and elastic rock are introduced. Two robust algorithms are presented to couple the two media (rock and fracturing fluid) for discretized model in plane strain condition. The algorithms include: 1) stress transfer from fluid to rock matrix and 2) evolution of fracture opening width, resulting in the change of fluid pressure. An iterative process is demonstrated for the interaction of the two media to promote convergence. The coupled model for multiple hydraulic fracturing is developed to express the interaction between the model parameters, combined with the process of fracture propagation. To verify the results, the shadow effect between fractures is analysed by showing the stress change alongside the propagating fractures.
In this work, a three-scale poromechanical model for naturally fractured coalbed methane reservoir is developed. The coal seam reservoir is composed of a coal matrix mainly containing nanopores saturated by adsorbed gases and natural fracture network (called cleats). Beyond the empirical Langmuir law, the adsorption isotherm of the fluid mixture (CH4 and CO2) is rigorously constructed by using the Density Functional Theory (DFT) applied to a Lennard-Jones fluid [1, 2, 3], allowing to compute the fluid distribution in the pores and the adsorption-induced force (solvation force) exerting on the solid phase by the adsorbed fluid. It is highlighted that the solvation force magnitude is much higher than the bulk pressure leading to an important impact on the mechanical properties at higher scale. A first homogenization procedure of the nanopore scale model is performed to derive the mechanical response of the continuum matrix, characterized by a modified Biot-Willis parameter depending on the solvation force magnitude. Such system of governing equations in the matrix is coupled with the fluid pressure in the discrete cleat system with dependency of aperture with the normal stress dictated by the hyperbolic Barton-Bandis model [4]. The problem is strongly non-linear and coupled with the hydrodynamics due to the rapid increase of the joint stiffness and the dependence of the fluid pressure. Moreover, the cleat stiffness is directly related to the cleat closure, which controls the permeability of the reservoir. A second homogenized procedure is pursued and capable of providing the constitutive response of the homogenized poromechanical parameters on gas pressure at the reservoir scale. In this context, increase in the normal BB-stiffness of the cleats tends to reduce the jumps of characteristic functions at the matrix/cleat interfaces which are propagated to the macroscale in terms of perturbations in the macroscopic poromechanical parameters. In addition to the overall three-scale decomposition of the total macroscopic stress, we constructed a new constitutive law for the Lagrangian cleat porosity. The dependence of the two- and three-scale homogenized poromechanical coefficients on the gas pressure is reconstructed numerically quantifying precisely the influence of the solvation force and the non-linear elastic behavior of the natural fractures.
Finally, the poromechanics is coupled with the multiscale hydrodynamic model in order to simulate the enhanced coalbed methane reservoir by CO2 injection. The interplay between the solvation force due to the adsorption effect and the non-linear elastic response of the fractures is numerically analyzed during the CH4 production and the CO2 injection procedure, underlying the increase in fracture stiffness at the injection well due to the matrix swelling stemming from the preferential CO2 adsorption in coal. Moreover, the fracture permeability tends to decrease in the vicinity of the injection well due to the same effect.
Fractures in subsurface shale formations serve multiple purposes, for example, in the recovery of resources in hydraulic fracturing or as potential harmful leakage passages through caprocks that may contribute undesired fluids to the atmosphere or functional groundwater aquifers. A proposed method to seal or influence fracture properties is Ureolysis-Induced Calcium Carbonate Precipitation (UICP), a bio-mineralization technology driven by the enzymatic hydrolysis of urea, resulting in the formation of calcium carbonate. Sporosarcina pasteurii is a common microbe used as the source of the urease enzyme that catalyzes the chemical reaction. The resulting calcium carbonate can bridge the gaps in fractured shale and reduce fluid flow through fractures. However, there is little information on how this process affects the mechanical properties of the resulting biomineralized shale. This study represents the first step toward determining the influence of UICP treatment on shale material and its subsequent mechanical strength properties. This methods development study aims to determine the effect that temperature has on the tensile strength of intact, unfractured shale cores (2.54 cm (1 in) diameter, 5.08 cm (2 in) long). Tensile strength was determined indirectly using a modified Brazilian test where the splitting tensile strength is attained by applying a compressive load onto the core. Shale cores from Eagle Ford and Wolfcamp formations were tested at both room temperature and 60°C to determine if increased temperature influences the tensile strength of the rock. This data will help to assess the necessity of testing biomineralized cores at temperature. Though 60°C may not mimic subsurface temperatures of the shales used in this study, it was chosen due to limitations of the UICP process while still approaching temperatures of shale formations. This project aims to evaluate what effect temperature has on the mechanical properties of intact shale cores so that engineered or natural rock fractures that are sealed by biomineralization can be better understood.
Hydro-mechanical coupling in deforming porous media has been the subject of studies in mechanical, energy, geology and environmental engineering. In our work, following Griffith’s theory [1] and Francfort and Marigo’s [2] variational approach to fracture, we develop a generalised phase-field-based formulation for predicting the fluid-driven fracture propagation in porous media across different time scales. The advantage of the phase-field method is that the complex fracture behaviour, such as initiation, propagation, branching and merging, is the natural outcome of simulations without prior knowledge of propagation path. A macroscopic framework is proposed for phase-field modelling of dynamic fracture to couple the physics of flow with the mechanics of fracture, including the deformation behaviour of solid skeleton, the crack propagation and fluid flow within pores and cracks. The effect of fluid properties such as viscosity and permeability is also discussed. The numerical algorithm is implemented in ABAQUS by user-defined subroutines. We compare numerical results against several analytical and experimental solutions and also demonstrate the approach’s ability to predict complex fluid-driven fracture systems.
[1] Griffith Alan Arnold & Taylor Geoffrey Ingram. The phenomena of rupture and flow in solids. Philosophical Transactions of the Royal Society of London. Series A, Containing Papers of a Mathematical or Physical Character 221, 163–198 (1921).
[2] Francfort, G. A. & Marigo, J.-J. Revisiting brittle fracture as an energy minimization problem. Journal of the Mechanics and Physics of Solids 46, 1319–1342 (1998).
Hydroshearing, or shear stimulation, is recognized as the main method to exploit geothermal energy in hot low-permeability crystalline rocks at depth. It consists of enhancing permeability via injection-induced shear slip and dilation of preexisting fractures. Hydroshearing usually causes some induced microseismicity, sometimes of sufficient magnitude to be felt on the surface. Thus, high-pressure fluid injection to enhance fracture permeability should be made carefully to avoid inducing earthquakes above the acceptable magnitude.
The process of hydroshearing is theoretically well understood and numerical models are capable of simulating it. However, fundamental investigations at the field scale are limited. This study focuses on the modeling of a hydraulic stimulation carried out at the Bedretto Underground Laboratory for Geosciences and Geoenergies (BULGG), in Switzerland, to investigate hydro-mechanical coupled processes due to fluid injection into fractured granite. We examine three numerical models with increasing complexity (a model with calibrated time-variable permeability, a model with strain-dependent permeability, and a model incorporating viscoplasticity with strain weakening and dilatancy) to improve the simulation and capture the hydro-mechanical response of the fractured rock mass.
The first model yields a reasonable fitting to measured overpressures at the injection borehole. Yet, the pore pressure distribution and the corresponding poromechanical response of the rock are not well captured. Employing an embedded model to calculate permeability changes as a function of volumetric strain improves the temporal evolution of overpressure at the injection borehole at the early stages of stimulation, but overestimates it once the fracture undergoes shear slip. Using a viscoplastic constitutive law with strain softening and dilatancy results in an additional enhancement of fracture permeability and thus a better reproduction of the monitored overpressures. The results show that the timestamps of monitored microseismic events correlate well with the times when permeability enhancement surpasses the previously maximum amount in each injection cycle.
The upscaling of 2-phase flow in porous media from pore to Darcy scale is a long-standing problem. While several approaches have been published in the literature, there remains no consensus on what the right approach is and what the correct Darcy scale transport equations are. Many approaches assume explicitly or implicitly that a length scale exists where pore scale dynamics average out such that equipartitioning of energy in the different degrees of freedom holds, which is sometimes referred to as the multi-phase representative elementary volume (REV). The implicit assumption is ergodicity where spatial, temporal, and ensemble averages are equivalent.
Many theoretical approaches explicitly require ergodicity, and few general strategies have been advanced to treat non-equilibrium thermodynamic behavior associated with non-ergodic systems. We develop a non-equilibrium theory using time-and-space averaging; assuming that ergodic conditions hold only at very small length scales. For instance, at late times Haines jumps travel beyond the range of diffusive mixing (which would be required for equipartitioning). Since the timescale for mixing is fast at small length scales, many non-ergodic systems can be described based on this approach. We show that fluctuations are constrained by the internal energy dynamics, deriving quasi-ergodic requirements that must hold for any stationary process due to conservation of energy. Since these requirements are formulated in terms of observable quantities, they can be used to explicitly identify the timescale where valid transport coefficients can be obtained. This result is significant because it provides a straightforward way to homogenize the dynamics of multiphase flow in porous media which does not obey equipartition of energy, particularly with slow fluctuations.
We apply our theory to derive transport coefficients for immiscible fluid flow through porous media, demonstrating that pressure fluctuations observed in experiments can be non-Gaussian due to cooperative effects that are caused by capillary events. We show that the macroscopic dynamics can still be homogenized if the timescale for averaging is chosen such that these fluctuations perform no net work on the system. We further demonstrate that changes to fluid topology are responsible for non-ergodic effects, and that time-and-space averages provide a natural mechanism to account for discrete changes based on the topological residence time associated with a particular micro-state of the system.
Following that methodology, we derive Darcy’s law for single phase flow and the 2-phase Darcy equations for multiphase flow at stationary (“steady-state”) conditions. The equations have the same form as the 2-phase Darcy equation introduced as a phenomenological extension to Darcy’s law. Cross terms in relative permeability arise from having experimental access to phase pressures independently.
Abstract: Lithium-ion (Li-ion) batteries play a major role in the electrification of many business sectors as well as public and private transportation. Although being a mature battery technology, the manufacturing of Li-ion batteries has still room for optimization. A relevant example is the process step of electrolyte filling that comes with unwanted pore-scale effects such as gas entrapment. However, the underlying physics can be hardly studied using experiments, leading to the necessity of enhancing mesoscopic modeling and simulation methods.
In this context, the lattice Boltzmann method has recently gained importance for studying flow in complex porous media. It comes however at the cost of large computational expenses, especially when simultaneously simulating flow in structurally resolved pores at different length scales. Therefore, homogenization methods have been developed to circumvent the explicit modelling of pores at the smallest length scale, but describe the flow by a Darcy-Brinkman-type approach instead where only the mean permeability of the medium is considered.
In this work, we present such a homogenized lattice Boltzmann method (HLBM) that combines a grayscale approach with the multi-component Shan-Chen model. It enables simulations of multi-phase flow in heterogenous porous media by physically modelling fluid-fluid and fluid-solid interactions even at sub-resolution scales. The HLBM presented here shows special advantages: the interfacial tension and wetting conditions are not affected by the homogenization and physical properties are continuous across interfaces between different porous media.
The model was validated using different test cases for single- and two-phase fluid flow. The results are in excellent agreement with the corresponding analytical solutions where available. In addition, the HLBM was applied to electrolyte filling of Li-ion batteries. On the one hand, it was used to study the influence of the nanoporous and partially permeable carbon-binder domain on the electrolyte flow. On the other hand, it was used to study flow in a fully homogenized separator microstructure with local heterogeneities.
All in all, it is shown that the HLBM can be applied to study multi-phase flow in porous media from which the pore sizes differ by orders of magnitude, without fully resolving the microstructure. This speeds up simulation times significantly. Thus, the HLBM is an efficient approach that can be applied to energy storage materials, but is not limited to it.
Acknowledgement: This work received funding from the European Union’s Horizon 2020 Research and Innovation Programme within the project DEFACTO [grant number 875247]. The simulations have been carried out on the Hawk at the High-Performance Computing Center Stuttgart (HLRS) [grant LaBoRESys], and on JUSTUS 2 at the University Ulm [grant INST 40/467-1 FUGG].
Keywords: Lattice Boltzmann method, porous media, Li-ion batteries, electrolyte filling
While irreversible thermodynamics has proven his relevance in modelling of flows through porous media, explicit and usable formulations based on thermodynamics remain few and hard to extend. Yet thermodynamics offers powerfull concepts to achieved the coupling of the wide variety of processes occuring in porous media. This work proposes to apply the principles of irreversible thermodynamics to compositional multiphase flows through porous media and to derive from them a formulation that encompasses the most usual models.
Starting from a clean decomposition of the porous media system between the volume of matrix, the volumes of fluids and the interfaces, the entropy is assumed to be a Euler homogeneous function of first order of the total internal energy, the total mole numbers of each component, the volumes of each fluid, the areas of each interface. The entropy balance equation is then derived from the energy and mole balance equations.
The local entropy production brings out a term that summarizes the exchanges of mechanical work between fluids and interfaces. Such a dissipation term, counting for the interface displacements and deformations, gives a proper definition for the interface equilibrium assumption : interface dissipation is zero, interface transformations are reversible. We show how that interface equilibrium assumption is fundational for the common idea of capillary pressure curve. Indeed, it appears that all the capillary pressure curves in a given porous media derive from a unique convex potential. Although the two phase case is trivial, for three and more phases, this result is strongly structuring, possibly conflicting with some capillary models proposed in the litterature.
The other entropy production terms are related to energy and mass transfers. They come in the form of flow-force products and are compatible with usual laws like Darcy's law for the fluid velocities and Fourier's law for the thermal conduction. The exception is the molecular diffusion where the more classical Fick's law (driven by the gradient of concentration) is not compatible with the positivy of entropy production expected by the second principle. Instead, we have to consider generalizes Fick's laws or Maxwell-Stefan diffusion that are driven by the gradients of the chemical potentials.
In order to ensure the usability of such of modeling, we conclude by proposing a closed formulation, based on a fixed set of primary variables related to the temperature and the chemical potentials. The data of the model are the coefficients for flux-gradient laws, the equations of state for the fluids and the "capillary" potential. The total entropy of the local system is provided and the entropy balance equation derivation is natural thanks to the choice of primary variables.
Understanding and modelling contaminant transport is necessary to assess pollution sources’ lifetime and severity and optimize the remediation strategies. The transfer of contaminants from the NAPL (Non-Aqueous Phase Liquids) phase to the aquifer is a multi-scale problem with different transport mechanisms within the various phases and at the interfaces. This two-phase flow problem is, in particular, driven by mass transfer between both phases, and is generally described by local non-equilibrium models. Such models at the macroscopic scale include transport equations for each phase which are coupled through one or several mass exchange coefficients. While these coefficients, which integrate the impacts of different pore-scale features (pore geometry, phase distribution, flow velocity), play a key role in the fate of the pollution source, it is usually approximated, for a given phase saturation, by a constant value estimated from empirical correlations (Quintard and Whitaker 1994, Soulaine et al., 2011). However, it generally shows a transient behaviour and can evolve with NAPL phase composition and relative solubilities, which remains poorly studied (Shafieiyoun & Thomson, 2018).
In this work, we start with the numerical modelling of the problem at the pore scale using COMSOL Multiphysics. The NAPL phase is considered an immobile blob at the pore centre and is composed of a non-soluble component and one or more soluble components. At this scale, the dissolution is implemented using Raoult’s law at the interface between phases. The solubilities evolve in a complex and coupled way as a function of the mass fractions of the considered components, themselves dependent on time. The change in the NAPL’s volume is modelled using an ALE approach and the derived equation for the interface velocity. Flow and transport in the water phase are considered and transport in the NAPL may be taken into account. We study the impact of different factors (number of soluble components, component diffusions, interface evolution, Péclet number) on the form and behaviour of the mass exchange coefficient. In the second step, we upscale from the pore scale to the Darcy scale using a numerical approach, with a focus on the mass exchange coefficient. The potential implications of replacing this time-and-space-dependent mass transfer coefficient in a Darcy-scale model with a constant and unique value as well as with a function of different state variables are discussed.
The formal derivation of the macroscopic mass and momentum balance equations for two-phase, creeping incompressible and Newtonian flow in rigid and homogeneous porous media is proposed in this work, assuming separation of length-scales and the existence of a (periodic) representative elementary volume, both classical for upscaling. The development is performed by making use of elements of the volume averaging method, combined with the adjoint technique and a Green’s integral formulation [1, 2]. The macroscopic mass balance equation in each phase is identical to that already reported in the literature [3, 4]. The macroscopic momentum balance equation expresses the seepage velocity in each phase under the form of a pair of Darcy-like terms, involving a dominant and coupling permeability tensor, respectively related to viscous effects in the phase under concern and viscous coupling through the interfaces. Importantly, it includes an additional term resulting from capillary effects. The later has not been obtained so far as a result of a priori assumptions that this term should be negligible, in particular for small capillary numbers [4, 5, 6]. The effective coefficients present in this macroscopic model are all obtained from the solution of two coupled closure problems that coincide with those already reported in the literature [4, 5, 6, 7]. The performance of the model is illustrated with numerical simulations carried out in a model two-dimensional configuration using a boundary element method. Average velocities, resulting from direct numerical simulation, are compared to the predictions of the macroscopic model obtained from the closure problems solution, showing excellent agreement over extended ranges of the capillary number, viscosity ratio and wetting-phase saturation. The additional capillary term present in the average momentum equation is shown to have a very important contribution is some situations. Extensions to other flow situations are briefly discussed.
The problem of immiscible and incompressible two-phase flow in porous media can be recast in terms of the average seepage velocity of the wetting- and non-wetting fluids and a novel velocity called the co-moving velocity, which has the potential of simplifying the theoretical description of macroscopic flow properties [1]. The theory is based on degree-1 Euler homogeneity of the total volumetric flow rate through the porous medium, and the framework takes on the appearance of a thermodynamic theory. The co-moving velocity is the quantity that bridges the gap between the measurable seepage velocities and the abstract thermodynamic velocities that appear in the thermodynamic theory. It has been shown both numerically and experimentally that the co-moving velocity has a particularily simple behaviour [2], and understanding the role of this quantity in a more general theoretic setting might aid our intuition for this abstract velocity.
We will present different interpretations of the transformation from the seepage velocities to the total seepage- and co-moving velocity in the context of a pseudo-thermodynamic theory with as few variables as possible. This will lead us to make the connection between the flow quantities and differential geometry, and show that one is able to regain results from previous works and uncover new descriptions of the flow by convenient coordinate-changes on the space of extensive pore-areas. We discuss a general description of the flow-quantities based on vector fields, and the relation between our framework and the broader field of geometric thermodynamics.
Immiscible displacements in porous media have been extensively explored, both experimentally and numerically, in the last decades, and, for two-phase Newtonian flows, it was shown that the competition between the characteristic forces involved, like viscous and capillary forces, determines the structure of the invasion pattern [1]. In this work, we try to extend these studies taking into account some non-linear behaviors, namely the presence of yield stress and the formation of compact displacement regions. If the displaced fluid consist in a non-Newtonian Bingham liquid, flowing like a Newtonian fluid only above a finite stress, a yield stress dominant pattern emerges, characterized by needle-like paths of low dimensionality [2]. On the other hand, compact invasion regions, in which every pore is invaded by both fluids in a rapid succession, emerge when imposing a constant pressure drop throughout the medium, as recently shown experimentally in a porous Hele-Shaw cell [3]. We perform numerical simulations of a two-dimensional porous medium, in the framework of the dynamical Pore-Network model [4]. An algorithm, part of the Augmented Lagrangian methods class and already implied successfully for the study of non-Newtonian flow, was adopted for solving the non-linear relation between the flow rate and the pressure at the pore level [5]. We then characterize the structure of the invasion patterns related to these non-linear effects, measuring quantities like the saturation of the invading phase and the fractal dimension of the corresponding paths, and comparing these values with the ones present in literature for the two-phase fully-Newtonian flow. The domains of validity of these patterns are finally mapped onto a plane with axes the ratio between the different forces into play, obtaining a 'phase-diagram' for these immiscible invasion displacements.
Trapping of fluid in porous media by capillary forces is a key process in many subsurface processes. It can be favorable to store carbon dioxide in deep saline aquifers, or unfavorable for groundwater remediation and in petroleum production, where droplets of contaminants/oils are trapped in the pore-space by capillary forces. Wettability properties at the vicinity of three-phase contact region is a key parameter to describe trapping mechanisms as well as the long-term stability of trapped droplets. Although the concept of contact angle – i.e. the angle visually measured between the solid surface and the fluid-fluid interface – is widely used to model two-phase flow processes at the pore-scale, it does not accurately describe wettability alteration due to change in pH and salinity. The latter arises from inter-molecular interactions. Furthermore, the variation of contact angle with the velocity field at the three-phase contact region gives rise to more complications. This region is created by spreading of a thin film of one phase on the solid as oppose to the other fluid and is thin enough to be in the range of inter-molecular forces. We intend to model the wettability by investigating the evolution of this film.
To do so, we developed a lubrication model for the thin film evolution on the solid surface. The model is physically rooted and replaces the concept of contact angle. It accounts for inter-molecular forces by introducing the different components of the disjoining pressure, notably the van der Waals and electric double layer potentials. The developed framework relates molecular interactions to pore-scale simulations through the boundary conditions, paving the way to more realistic pore-scale simulations with wettability alterations. It also can be used as a tool to investigate other phenomena governed by inter-molecular forces, such as film stability and streaming potential.
For the particular case when the film contains electrolytes, the different ions will get adsorbed on the solid surface. Even though the fluid charge is neutral, a distribution of ions will form close to the solid surfaces. Movement of fluid in this region will create an electrical voltage called streaming potential. The effect of streaming potential can be noticeable, especially in the case of moving contact line. We would like to study the movement of the ions in this region in order to find out the range of the importance of streaming potential.
The study of mass-transfer in confined geometries is extremely important in many engineering and biological systems. In the context of geological carbon sequestration, carbon dioxide is injected into subsurface reservoirs leading to the formation of elongated bubbles that can either be trapped, move, or interact with the solid matrix. The presence of carbon dioxide has the effect of increasing the acidity of the in-situ brine, boosting a series of chain reactions that enhance rock dissolution. This may threaten the long-term integrity of the storage process due to the formation of leakage pathways for carbon dioxide.
Although the hydrodynamics of elongated bubble has been object of several studies, the case where a solute is transported in the surrounding liquid and surface mass-transfer mechanisms act on the solid wall or the bubble-fluid interface is much less understood. To fill this gap, we investigate the transport problem around a confined Taylor bubble to access the competition between advection, diffusion, and surface mass-transfer in the different regions of the bubble. To this aim, we derive a one dimensional Advection-Diffusion-Mass-Transfer equation where the transport mechanisms are described through an effective velocity, an effective diffusion coefficient, and an effective Sherwood number. Our model generalises the Aris-Taylor dispersion to the case of a Taylor bubble and clarifies the impact of surface mass-transfer in the advection and diffusion dominated regimes for both the front and rear menisci.
Despite the fact that the motivation of our work is oriented to microfluidics applications that involves solute transport and mass-transfer, its ramifications are relevant also in scenarios where the presence of a solute affects the surface tension (i.e., Maragoni effect) or even drives the flow (i.e., diffusioosmosis).
Pore scale imaging and modelling have played an enormous role in advancing knowledge in complex transport phenomena within porous media. We discuss new challenges and directions in pore-scale research by integrating artificial intelligence. These include the recreation of porous media images at a super-resolution, multimineral segmentation and prediction of petrophysical properties with applications in underground reservoir simulation, ore characterisation and groundwater modelling.
Ganglia (bubbles, or droplets) are widespread in porous media of various industrial applications. Thermodynamic properties of a ganglion, such as its morphology, free energy, capillary pressure, surface energy, etc., are crucial in determining its transport and reactive performance. Although these in homogeneous porous media have been recently resolved [1, 2], it is still challenging to quantitatively describe the thermodynamic properties of ganglia in heterogeneous media [3-5].
We develop a pore-scale numerical algorithm for determining the thermodynamic properties of hydrostatic ganglia in heterogeneous porous media. We track cycles of quasi-static growth and quasi-static shrinkage of a ganglion in a two-dimensional heterogeneous porous media, as shown in Figure.1(a). The algorithm is as follows:
(1) Create a heterogeneous porous medium and set the initial capillary pressure (Pc) and pore occupancy of the ganglion.
(2) Find the hydrostatic morphology of the ganglia with set Pc and pore occupancy, and record its properties including ganglion volume (V), free energy (F), surface area (A), etc.
(3) Make a small change in Pc while keeping pore occupancy unchanged, and then check whether a new hydrostatic morphology can be achieved. If so, the change is reversible and we go back to step (2) and continue. If not, an irreversible event emerges that changes pore occupancy while keeping V unchanged, and we search for the new stable morphology.
This algorithm may be used to simulate degassing and dissolution process in heterogeneous porous media and enhance our understanding of these processes. In addition, this algorithm may be used to construct the energy landscape of the entire heterogeneous porous media. We believe that this work helps better understand the behaviors of the dispersed phase in heterogeneous porous media.
Transition to renewable energy sources, due to their naturally intermittent production, requires developing large-scale storage technologies. Underground Hydrogen Storage (UHS) in porous formations is a promising approach to providing a giant storage capacity. To ensure the efficiency of the storage operation, multiscale modeling and simulation strategies are essential. Since micro-scale physics controls macro-behavior. Therefore studying the flow behavior in porous rocks at the micro-scale is insightful for UHS projects in porous reservoirs.
The present study develops a dynamic pore-network modeling (D-PNM) approach to simulate the immiscible two-phase flow of hydrogen and water through representative digital network models of different porous structures. As the key feature of UHS, the model is developed for the cycles of injection and production of hydrogen into a porous media. The model input parameters are based on the experimentally obtained static and dynamic wettability analyses as presented in the literature. As for the rock, digital networks are constructed based on 3D X-ray images of porous samples. The topology of the pore space geometry is translated to a representative pore-network model. To preserve the simulation stability, the developed D-PNM solves the transient multi-phase Stokes equations fully implicitly, for pressure and phase volume saturation. Through several test cases, we analyze the transport characteristics of the hydrogen/water interface, especially the fingering and spreading physics. These results shed new light on how a representative continuum-scale model should be created to study the process at the field scale.
We developed a thermodynamically-based pore network model to simulate fluid intermittency during two-phase flow through porous media. Relationship between pressure gradient and flow rate during multiphase flow in porous media have been observed to transition from linear to non-linear at intermediate flow rates in recent studies. With the aid of high resolution X-ray tomography, intermittent filling of the pore spaces by the phases has been observed resulting in a nonlinear relationship between the pressure gradient and flow rate. Existing pore network models have not been able to reproduce this phenomenon. We first develop a quasi-static pore-network model to simulate the drainage and imbibition processes where capillary forces dominate. We then modify this quasi-static model by introducing a probability distribution of filling inspired by the thermodynamic formulation of multiphase flow proposed by Hansen and colleagues. The probability distribution is formulated by drawing an analogy between thermodynamics and fluid flow in porous media. We have shown that a simple thermodynamically-based pore network model can simulate the nonlinear intermittent fluid behaviours during two-phase flow through porous media.
Evaporation from a porous medium into a free flow is one of the fundamental processes
in environmental systems (e.g. the evaporation of water from soil into the atmosphere
[1]). In technical systems self-pumping transpiration cooling can be realized with the
help of porous materials where the combination of capillary action and phase change
is a promising approach to cool structures due to its high cooling efficiency [2]. The
distribution of liquid in the porous material, namely the existence of continuous liquid
pathways to the surface of the porous medium influences significantly the evaporation
rate [3]. Furthermore, the condition of the turbulent boundary layer in which the vapor
is transported away from the surface is of great importance.
Hybrid-dimensional models are successfully used for the efficient modeling of
such systems under laminar flow conditions [4]. These models use coupling conditions
to ensure the continuity of mass, momentum and energy between the pore network
model (PNM) and the free-flow domain. But these coupling conditions comprise
unknown parameters (e.g. the slip length) and their validity for turbulent flows is
unclear. One possibility to evaluate the validity of coupling conditions and to derive
closures for the unknown parameters is to fully resolve the Navier–Stokes equations in
the free flow and the pore space.
In this talk results of such pore resolved calculations are presented for a porous
medium with different water saturation levels. The focus will be on the momentum
balance at the interface. It will be discussed (i) how the rough, permeable surface
influences the turbulent boundary layer, (ii) how the fluid distribution will influence
the effective roughness and (iii) how the pore wall wettability influences the fluid
distribution.
Finally a possible approach for a coupling condition for the momentum balance
of a turbulent flow with a porous medium under different saturation levels is presented.
Evaporation studies focus on the identification and characterization of heat transfer and flow dynamics in the vicinity of the solid-liquid-vapor contact line. The meniscus is often characterized by the following three regions: non-evaporating adsorbed layer, thin-film, and capillary regions. The adsorbed layer, which has a thickness on the order of nanometers, is traditionally believed to be non-evaporating due to the strong intermolecular forces producing a strong disjoining pressure that suppresses evaporation. Despite this classical view, recent molecular dynamics (MD) simulations have shown that adsorbed layer plays a significant role during thin film evaporation [1]. Utilizing a new energy-based interface detection method [2], we present nonequilibrium MD simulation results of thin film evaporation of liquid argon sandwiched between two parallel platinum plates. One end of the platinum channel is heated by energy addition, while the other end is cooled at the same rate to ensure constant energy of the simulation system. Liquid argon evaporates in the heater and travels to the condenser region. As a result, the utilized MD simulation system exhibits statistically steady transport. Here we present the shapes of the evaporating menisci for 4 different channel heights varying from 2 nm, 4 nm, 8 nm, and 16 nm, at three different wall-fluid interaction parameters and under several different heating/cooling rates. Depending on the surface wettability and applied heat flux the meniscus can be in the pinned or receding regimes. The latter case creates adsorbed layers suitable for investigating its dynamics. The higher wettability cases exhibit thicker and more stable adsorbed layers, with reduced radius of curvature (ROC) and reduced evaporation rate. They are more stable and can handle higher heat fluxes. The lower wettability cases exhibit more evaporation but can easily lead to dry out. For channel sizes less than 10 nm, the adsorbed layer and evaporating thin film regions are intertwined, and evaporation from the adsorbed layer can contribute up to 80% of the total evaporating mass flowrate. Even for the largest channel case (16 nm), the adsorbed layer contributed about 10% of the total evaporating mass flowrate [3]. The talk will focus on these findings and gear towards consolidation of our findings towards a universal behavior of adsorbed layer transport in nanoscale confinements.
Wicking is the spontaneous imbibition due to the negative capillary pressure created at the liquid-air interface 1. The wicking of simple fluids, such as water and organic solvents is well understood for a long time [2]. This well defined situation becomes more complicated in the case of complex fluids with an internal structure on the nanoscale. Then, competitive wetting and confinement effects play an important role and influence both, the behavior of the complex fluid inside the porous matrix and the inner structuring of the fluid. The latter can result in changes in phase behavior and other fundamental properties. To explore these effects we use bicontinuous microemulsions in the ternary phase system (water/octane/C\textsubscript{10}E\textsubscript{4}) as model complex fluid and controlled-pore glasses as confining matrices (CPG). Understanding the influence of geometrical restrictions yields both, fundamental insights and importance for applications, e.g. decontamination and enhanced oil recovery.
For a deeper understanding of these effects, of the spontaneous imbibition of a bicontinuous microemulsion and its components into the CPGs is investigated. In our study, we explore the wicking with the Washburn approach. In this approach the wicking of a test liquid is monitored gravimetrically, as shown on the right side in figure 1. Effects of the traversed matrices are studied by using various CPGs with pore diameters between 75 – 1000 Å and porosities from 55% to 80 %. The naturally hydrophilic surfaces of the CPGs were hydrophobically modified to analyze the impact of the surface polarity. The bicontinuous microemulsion shows a more universal wicking behavior than the tested pure liquids.
Imaging techniques (cryo-SEM) and small angle scattering (SANS, SAXS) are used to investigate the microemulsion phase structure inside the porous matrices. In this talk, the results of these combined experiments will be presented and discussed.
NMR relaxation time measurement is a well-known, non-destructive method to probe all states of protonic liquid such as water, in porous media, at different pore scales. In contrast with MRI which can get local information but is blind with respect to most liquid in nanomaterials, standard NMR relaxation measurements can provide information on the liquid content over six decades of relaxation times typically corresponding to pore scales from the millimeter to the nanometer. Here, we propose a simple though powerful technique which provides various direct, quantitative information on the liquid distribution inside nanoporous porous structures and its variations over time due to fluid transports and/or phase changes. It relies on the analysis of the details of the NMR (nuclear magnetic resonance) relaxation of the proton spins of the liquid molecules and its evolution during some process such as imbibition, drying, phase change, etc, of the sample. We present a few applications of this technique to nanoporous materials such as a silica glass [2], cellulose fibers [3], or nanoporous glass beads [1]. We show that this approach allows to observe and quantify a variety of possible dynamic phenomena such as: a progressive homogeneous or inhomogeneous emptying of pores, and isotropic or differential shrinkage of the pores, the possible existence of liquid films along the pore walls, transfers between bound and free water.
Imbibition is important physics in nanoporous media, related to many energy and technology areas, e.g., fuel cells, water desalinization, bio-sensor, hydrology, hydrocarbon recovery, CO2 geo-storage, underground hydrogen storage, etc. The classic theory to describe the spontaneous imbibition dynamics is the Lucas-Washburn (L-W) equation, while the classic theory to describe the electrocapillary imbibition is the Lippmann equation and the Young – Lippmann (Y-L) equation. However, whether these classic theories are still valid at nanoscale have not been rigorously examined yet.
Therefore herein, we experimentally investigate the dynamics of spontaneous and electrocapillary imbibition in nanoporous media. For spontaneous imbibition in hydrophilic nanoporous media in the absence of evaporation, spontaneous imbibition height is linear with square root of time and a larger pore size causes a faster imbibition, which are consistent with the L-W equation; in contrast, for spontaneous imbibition in hydrophilic nanoporous media in the presence of evaporation, this linear relationship is deviated from linearity at early stages and a modified L-W theoretical model is derived to incorporate the evaporation effect. For electrocapillary imbibition in hydrophobic nanoporous media, counterintuitive voltage polarity dependence and electro-dewetting phenomena are observed, indicating that the Lippmann and the Y-L theory are invalid to describe the fundamentals of electrocapillary imbibition at nanoscale. Hence, the underlying mechanisms responsible for these two novel physics are explained by electrical double layer charging, Faradaic reactions and others.
These insights will provide significant guidance on various applications relevant to energy transition, such as energy storage and conversion devices, water desalination, batteries and fuel cells.
Many applications of nanoporous materials require their porosity to be filled with liquid. This is notably the case in heterogenous catalysis or in electrochemistry. In all cases, it is essential to determine whether the porosity is uniformly filled or whether the liquid is excluded from specific pores. In a macroscopic context, wetting is well predicted in terms of the different energies of the wet and dry surfaces. By contrast, the conditions for wetting of nanoporous solids are still poorly understood. It is unclear whether or not macroscopic physical concepts apply at scales close to molecular dimensions. Moreover, the geometry of porous media can be complex with pores with a variety of sizes and connectivity. Besides the spontaneity of the pore space liquid invasion, other important questions concern the kinetics of the problem in relation with the permeability of the different pores, which makes the question even more challenging.
In the present study, we investigate the nanometer-scale wetting of nanoporous materials. Specifically, we focus on carbon xerogels with two families of pores, namely, mesopores with sizes around 20 nm coexisting with micropores having almost molecular dimensions. We perform capillary-rise experiments of water in these materials, and we use synchrotron Small Angle X-ray Scattering (SAXS) to investigate the process at nanometer scale in a space- and time-resolved way. Different materials are considered with different meso- and micro-porous structures. We also report capillary-rise experiments on materials with water-saturated micropores by preliminary adsorption of water vapour. All experiments were performed at the Belgian DUBBLE station (BM26) at the European Synchrotron Radiation Facility.
Our results reveal a two-stage wetting process, with a diffuse water front coming first, followed by a sharp front lagging a few millimetres behind. The SAXS data shows that the diffuse front corresponds to the early filling of the molecular-sized micropores, while the sharp front corresponds to the later filling of the mesopores. The two water fronts propagate according to a √t law, which is typical of a Washburn model whereby the wetting kinetics is limited by viscous dissipation. We use independent water adsorption experiments to estimate the capillary suction into the micropores, from which we infer their permeability.
Water condensation and evaporation from saline porous materials has attracted the attention of scientists for years due to a large field of applications: salt weathering of buildings, desalination of water, CO2 sequestration, soil decontamination, etc… [1, 2, 3]. A complete understanding of related nanoscale processes is however lacking, in particular concerning the coupling between evaporation/condensation and crystallization/deliquescence in confinement [4]. While the comprehension of the phenomenon has progressed in the past few years [4, 5, 6], there are still some challenges remaining in characterizing and understanding these processes.
Here we carried out thermodynamic experiments coupling sorption isotherms of nanoporous media containing salt, to dynamical measurement of the evaporation of salt solution droplets from the surface of the same nanoporous media. We show that we can account for both thermodynamic and dynamical experiments by using a minimal model involving coupled phase change of the solvent (water evaporation) and the solute (salt crystallization).
[1] Huber P., Journal of Physics: Condensed Matter 27:103102 (2015)
[2] Steiger M., Journal of Crystal Growth 282 (2005) 455–469
[3] Scherer G.W., Cement and Concrete Research 34(9), 2004, 1613-1624
[4] Jain P. et al., Langmuir 2019, 35, 3949−3962
[5] Vincent O. et al., Langmuir 2017, 33, 1655−1661
[6] Talreja-Muthrejas T. et al., Langmuir 2022, 38, 10963−10974
In recent years, substantial research and development endeavors are dedicated to enhancing the performance of polymer electrolyte fuel cells (PEFC) as a promising candidate for transport application. However, the power density still needs improvement for large-scale operations with one of the core issues being water management. The convoluted water balance in the PEFC requires proper water management to ensure high cell performance. Excess water retention in the pores results in blockage of the gas diffusion to the active sites of the electrochemical reaction, thus reducing performance and efficiency. Appropriate membrane humidification is needed for high proton conductivity. The deployment of a microporous layer (MPL) located between the gas diffusion layer (GDL) and the catalyst layer (CL) has been demonstrated to enhance water management. Operando scanning small and wide angle X-ray scattering (SAXS&WAXS) is utilized in this study as a practicable tool to explore the water level in the nanoscale CL and correlate it with the neighboring MPL material design at the cSAXS beamline of the Swiss Light Source. The use of a small beam size (≈7x30 microns) and ≈0.6 microns vertical step size resolves the bulk CL (≈7 microns thick) and additionally allows for a precise registration in case of membrane movement during operation. Due to the electron density difference between solid-void and solid-liquid interfaces, the water saturation level is quantifiable (see Fig. 1a). Relevant PEFC operating conditions (80⁰C, relative humidity 100%, 1.7 and 3 bar abs.) were enforced to compare the impact of two different MPL modifications (pore former inclusion, and higher PTFE content) to the water saturation of a base case MPL (Li100, 20% PTFE). The presentation will detail the material modification consequences on the CL and MPL saturation levels to reveal pore size-specific filling mechanisms. (see Fig. 1b-c).
Appropriate modeling approaches to quantify gas migration in low-permeability porous media can assist appraisal of sealing efficiency of caprocks, with key applications in sustainable use of underground energy resources. A variety of models depicting gas movement across low-permeability geomaterials are available (Wu et al., 2016; Sun et al., 2017, Rani et al., 2018). Some of these models represent gas migration in low-permeability media as a weighted sum of diverse mechanisms taking place across the porous system. Parameters associated with these models are envisioned to embed the chemical, mechanical, flow, and transport features governing feedbacks between gas and the host rock matrix. Such parameters cannot be easily and unambiguously evaluated via experimental investigations and are always affected by uncertainty. In this context, modern sensitivity analysis techniques enable us to diagnose the behavior of a given model through quantification of the importance and role of model parameter uncertainties onto a target model output.
Here, we rely on two global sensitivity analysis approaches and metrics (i.e., variance-based Sobol’ indices and moment-based AMA indices) to assess the behavior of a recent interpretive model that conceptualizes gas migration as the sum of a surface diffusion mechanism and two weighted bulk flow components (i.e., Slip flow and Knudsen diffusion). We quantitatively investigate the impact of each uncertain model parameter on the evaluation of methane flow, which is, in turn, conceptualized as a random quantity. Considering the paucity of available information, we consider three diverse characterizations of the probability density function describing the uncertain model parameters: (a) all parameters are described by uniform distributions; (b) all parameters are represented through truncated normal distributions; and (c) the reference pore radius is described by a truncated log-normal distribution while the remaining parameters are associated with uniform distributions. We then derive analytically the structure of an effective diffusion coefficient embedding all complex mechanisms of the model considered and rely on the global sensitivity analysis results to quantify the relative contribution of each flow mechanism to the overall gas flow.
Our results suggest that, in decreasing order of importance, reference pore radius, reference porosity, pore pressure, tortuosity, and temperature are the model parameters driving the major features of the gas flow probability density function. These results remain essentially unaffected by the choice of probability density function characterizing model uncertain parameters.
Computational models can predict and improve our understanding of multiphase flow in porous media. In this field, the task of uncertainty quantification is of paramount importance when developing and evaluating mathematical models aimed at the design and prediction of complex processes such as enhanced oil recovery techniques. One promising Enhanced-Oil-Recovery technique is the injection of foam in the porous medium, since foam injection reduces gas mobility and increases apparent viscosity, thus improving reservoir sweeping and increasing recovery efficiency. This work focuses on parameter estimation and uncertainty quantification of the foam flow in porous media. In particular, we present an uncertainty quantification approach based on surrogate models and Bayesian inference to evaluate how these techniques can reduce uncertainties and improve physical understanding and parameter estimation of foam flow in porous media. Our results suggest that the new framework based on Bayesian inference and surrogate models enhances parameter estimation and improves the uncertainty quantification of the foam flow in porous media.
Acknowledgements. The current work was conducted in association with the R&D project ANP n 20715-9, “Modelagem matemática e computacional de injeção de espuma usada em recuperação avançada de petróleo” (UFJF/ Shell Brasil/ANP). Shell Brazil funds it in accordance with ANP’s R&D regulations under the Research, Development, and Innovation Investment Commitment. This project is carried out in partnership with Petrobras.
Reduced-order models (ROMs) can be used in place of a high-fidelity model (HFM) to alleviate the computational cost associated with HFM simulations. Emulators or surrogates are a class of ROMs whose aim is to reduce the complexity of a given HFM by learning the dynamics of the state variables directly from the model’s output, i.e. they are trained on a dataset generated by running the HFM multiple times. As such, the number of simulations required to train a ROM is a measure of its effectiveness. Here, we use dynamic mode decomposition (DMD), a powerful data-driven method to construct ROMs of complex dynamical systems [1,2]. DMD employs singular value decomposition (SVD) and pursues the computation of the best-fit linear operator to approximate the relationship between time-shifted snapshots in time of the state variable [2]. Variants of the standard DMD algorithm exist, including the residual, generalized, and extended DMD [2,3]. In this study, we assess the accuracy of different DMD algorithms when mimicking flow and transport in porous media. We consider both interpolation and extrapolation (i.e. to get short-time future prediction) scenarios. The DMD has proven its utility in approximating systems of partial differential equations (PDEs); however, it doesn’t handle the possible variability in model parameters. As such, we explore how to combine DMD with the Polynomial Chaos Expansion (PCE), a family of ROMs used to approximate the response surface of a HFM in the random parameter space; this allows to obtain a ROM in terms of a polynomial relationship explaining the model response of interest as a function of the uncertain parameters, properly represented as independent random variables [4,5].
Data-driven approaches, among them machine learning tools, have garnered increasing interest in porous media research and offer alternatives to traditional numerical methods to improve the predictive modeling based on observation data. Recently, the idea of incorporating prior physical principles within measurements to better rely on experimental data has been successfully immersed into Bayesian inference as a valuable tool for uncertainty assessments.
The emergence of the Bayesian Physics-Informed Neural Networks (BPINNs) paradigm offers the opportunity to query the confidence in the predictions, the uncertainty in the measurements, and the model adequacy by providing posterior distribution of the neural network predictions [1]. In this presentation, we will show how to make BPINNs auto-weighted in order to address multi-objective IA problems, even when relying on shadow quantities.
Classical BPINNs mostly rely on Markov Chain Monte Carlo methods to sample from a weighted multi-objective target distribution, whose weights are related to the scaling of the tasks, the noise magnitude, and ultimately the inherent tasks’ uncertainties. While these parameters are recognized as critical, they are mostly hand-tuned in the applications leading to pathological behaviours or biased estimation, in the sense that one of the objectives will be prevailing in the posterior distribution exploration.
Actually, when dealing with real-world complex systems which involve heterogeneities and multi-scale phenomena in addition to uncertainties in the measurements, the setting of these weights can remain particularly challenging and require considerable energy in tuning. Furthermore, such dynamics also raise scale imbalances that highly disrupt the usual approaches, generating instabilities that make the sampling inoperative.
We focus here on a novel adaptive strategy for unbiased uncertainty quantification in BPINNs, based on an inverse Dirichlet weighting [2] of the target posterior distribution, which remedies to the failure modes previously identified. Our approach provides enhanced convergence, stability, and balanced conditions between the different tasks which ensure an efficient exploration of the Pareto front throughout the sampling procedure. While reducing the bias in the sampling, we show that this strategy is able to automatically adjust the weights, with them the uncertainties, according to the sensitivity of each task.
It then offers an interesting framework to study complex multi-scale dynamics from the Bayesian inference perspective and incorporate uncertainty quantification in multi-objective and stiff inverse problems.
In this direction, we aim to capture and quantify unresolved features arising from noisy X-Ray micro tomography measurements by adding information on the predictive physical-chemical model and then compensate for the lack of knowledge on the rock matrix structure with PDE-based priors, established according to a Darcy-Brinkmann two-scale porosity model [3].
Hence, we apply our auto-weighted methodology to the simultaneous identification of chemical parameters and morphological uncertainties on the porosity field, in a reactive inverse problem at the pore scale.
Altogether with the approach developed in [4] to determine the absolute permeability deviation, we will be able to provide uncertainty quantification on the main macro properties of a porous media sample based on its micro tomography and then perform more relevant direct numerical simulations with respect to the experimental data.
The reservoir rock in subsurface applications such as geothermal or hydrocarbon reservoirs, geological carbon sequestration or nuclear waste deposition is often fractured. When fractures are present, they can potentially dominate flow and transport in those applications. It is therefore necessary to characterize the relevant fracture parameters, particularly the fracture apertures, as good as possible to reliably predict performance and assess risks. However, direct measurement of fracture parameters is difficult. Usually, only data from sparsely located wells and seismic measurements is available. Therefore, indirect methods such as outcrop analogues, geological models and production data become crucial.
Ensemble-based data assimilation is a widely used technique in subsurface applications to match production history, reduce uncertainties in model parameters and improve simulation results. In this work, we use the ensemble smoother with multiple data assimilation (ES-MDA) (Emerick & Reynolds, 2013). As an iterative ensemble smoother, ES-MDA is suited for (at least weakly) nonlinear systems (Evensen, 2018) and various studies have successfully applied it for reservoir characterization (e.g. Emerick, 2016; Ranazzi & Sampaio 2019; Todaro et al., 2021).
In this study, we use a 2D fracture geometry with more than 3500 individual fractures obtained from aerial photographs of an outcrop (Odling, 1997). We therefore assume that the fracture geometry is known a priori except for the fracture apertures. In our model, each fracture has a different aperture which is constant over the fracture length. We consider a scenario where all fractures have an initial fracture aperture which is a function of the fracture roughness. We then apply a constant far field stress, such that fractures open due to shear dilation and close due to normal stress. The exact fracture apertures are however unknown due to uncertain model parameters (e.g. fracture roughness or rock properties).
We use ES-MDA based on flow and transport data to reduce the uncertainties regarding fracture apertures and study the influence of the prior ensemble on the performance of the data assimilation framework. Calculating the individual realisations of the prior ensemble with a geomechanical simulator is expensive. A purely stochastic approach on the other hand does not incorporate all geological knowledge. As a compromise between those two methods, we propose to generate the prior ensemble based on geomechanical far field approximations which do not rely on geomechanical simulations, while geological knowledge still is incorporated to some degree. Compared to the purely stochastic approach we expect that the required number of realisations is smaller, if such a prior ensemble is employed, since it tends to be closer to the reference.
Identifying the final fate of treated wastewater is sometimes a challenging task because not always receiving surface water bodies are available in the neighboring. Consequently, the water treatment agencies often resort to infiltration pond facilities for discharging effluents from the treatment plant. This technical solution is considered extremely favorable because it recharges the groundwater bodies, increasing the availability and improving the qualitative status of the natural reservoirs. The daily operation of such infiltration pond facilities is often based on heuristic rules simply aimed at discharging the treated water volumes. Nevertheless, the functioning of these infrastructures strongly relies on some hydraulic and hydrogeological features of the hosting site. The knowledge of parameters such as the average vertical and horizontal hydraulic conductivity in the area surrounding the basins would allow for optimizing the water flows to the groundwater. This study compares some techniques for modeling clusters of interconnected infiltration ponds with the aim of estimating the average values of the hydrogeological parameters involved therein. The overall inverse model is based on a dynamical system derived from mass balance and Darcy’s law. Within this general computational framework, several techniques have been implemented and tested, such as different Kalman filter versions. It is worth highlighting that the considered model is intrinsically ill-conditioned, and the right-hand side of the ODEs system is discontinuous: these issues somehow affect the accuracy of the tested techniques. This study has been conducted on synthetic data, partially based on the results of a previous study.
The generalized physics-based scaling curve method proposed by Patzek et al. (2013) is an excellent alternative to the decline curve methods that forecast gas production from shale reservoirs. However, it still neglects the multiphase flow effects and may lead to unreliable hydrocarbon production prediction from mudrock reservoirs. In this study, we perform a global sensitivity analysis using a compositional reservoir simulator to analyze the sensitivity of the scaling factors describing the physics-based method to multiphase flow effects varying selected input factors. We built a conceptual reservoir model of a typical, hydraulically fractured shale condensate gas well using a commercial reservoir simulator. We select the fluid input factors and their range of possible values over which we analyze the scaling curve. We perform a space-filling design using the MaxiMin Latin Hypercube sampling method. We run our simulation tests and estimate the scaling parameters: characteristic time of pressure interference between neighboring hydraulic fractures ($\tau$) and hydrocarbon mass in place in the stimulated reservoir volume ($\mathcal{M}_{SRV}$). We then calibrate a surrogate model to map the relationship between the multiphase flow properties and the scaling parameters using Bayesian optimization. Finally, we identify the key parameters affecting the shale condensate gas mudrock plays forecasting using global sensitivity analysis (``Sobol'' indices). Our results show the relative contribution of the multiphase flow input factors of the reservoir simulator to the variance of the physics-based curve scaling parameters. We demonstrate the importance of reservoir permeability, initial condensate/gas ratio (CGR), initial reservoir pressure, wet-gas phase behavior, and hydraulic fracture spacing in the variations of $\mathcal{M}$ and $\tau$. We show that the mudrock ultimate recovery factor (EUR) prediction when the condensate saturation around the wellbore is below a critical saturation may be accurately estimated using the single-phase solution. Finally, we highlight the limitations of using the single-phase physics-based scaling curve method to forecast condensate gas production from low-permeability reservoirs.
Continental shale oil in China is mainly of low-medium maturity, filled with heavy oil of low mobility and organic matter that unconverted. Horizontal drilling and hydraulic fracturing are insufficient to obtain economic production in such reservoir, thus in-situ heating and transform technology should be applied. A multiphase multicomponent hydro-thermal coupled numerical model considering multistage kinetic reactions is developed to describe the decomposition of solid organic matter, cracking of heavy hydrocarbon, phase behavior and rock property evolution.
During the in-situ process, organic matter (kerogen) decomposition and heavy oil cracking happens, enhancing hydrocarbon mobility. The research focuses on the development of multiphase multicomponent hydro-thermal coupled numerical model, with the evolution of porosity and permeability considered. The finite volume method is used for the space discretization of flow and heat transfer equation, and solved by the fully coupling method. Finally, the impact of important parameters on cumulative production are analyzed.
The compositional flow model is validated by comparing the results with those of CMG, and the coupled hydro-thermal model is validated against COMSOL Multiphysics. The impact of parameters including heating temperature, kerogen concentration, well bottom hole pressure, heater pattern and initial water saturation on cumulative production is analyzed. The results are summarized as: kinetic reaction rate is controlled by temperature and different reactions take place at variety heating temperature, influencing the fluid composition; higher kerogen concentration can enhance cumulative hydrocarbon production after in-situ conversion, making it an important parameter to evaluate before production; low bottom hole pressure can extract hydrocarbon products in time to prevent from further cracking and coking; different heater pattern has impact on the ratio of energy output to energy input, and hexagon heater is the most benefit; high water saturation will enhance energy consumption to heat water and reduce the utility ratio of energy, thus dewater process is required to reduce water saturation. It can be concluded that the in-situ conversion process is feasible in low-mid maturity shale oil reservoir, during which kerogen decomposition and hydrocarbon cracking happens. Besides, the operating parameters should be investigated to make the heating process economical.
The proposed model provides an efficient tool for modeling the in-situ conversion process of low-mid maturity shale oil reservoirs. In this paper, the reservoir fluid property variation, in-situ porosity and permeability evolution, and production characteristics are illustrated, which could provide insights on heater design and well operational management. With multiple transport mechanisms and multi stage kinetic reactions incorporated, the hydrocarbon production characteristics and formation property evolution of shale reservoirs can be both accurately captured.
Thermal regimes of arctic soils are strongly correlated to the presence of perennially frozen soil layers (permafrost). These soils undergo a cyclic annual freeze-thaw phenomenon with the formation of an active layer during summer. A complex patchwork of low vegetation layer consisting of Sphagnum moss, lichen, and peat covers this active layer. Such profiles are found in boreal regions for millions of km².
The latest IPCC reports show that arctic regions are highly vulnerable to climate change.
This vegetation cover proved to be crucial for modelling thermal soil regimes, both at watershed scale [1] and at continental scale [2]. This layer is the main interface between the atmosphere and the geosphere, through which energy and matter fluxes are mainly occurring by evapotranspiration [3]. Assessing morphological, hydraulic and thermal properties of this vegetation layer is thus compulsory to enhance predictive climate change impact models on boreal regions.
However, field measurements are difficult to conduct properly due to the large scale and the poor accessibility of the study area. To do so, some usual porous media study techniques (Representative Elementary Volume study, Pore Network modelling) has been applied in order to quantify morphological properties and hydraulic properties [4]. This first study showed the existence of Representative Elementary Volumes and that the bryophytic cover is highly porous and water conductive.
In the present work, the assumption to consider arctic vegetation cover as a porous medium is extended to thermal properties' assessment. A coupled experimental and numerical approach is set up to cross-validate the results found using both methods.
For this work, 12 dried samples extracted in 2018 at Khanymey Research Station (Siberia) are studied as well as some alive samples extracted from Clarens (Upper-Pyrenees). These samples consist of eight Sphagnum moss samples, two lichen samples and two peat samples.
The experimental setup is based on an enhanced version of the EN 12667 [5] norm for the assessment of thermal conductivity of highly thermal resistive material. Effective thermal conductivity and thermal diffusivity are extracted from thermocouple data and heat flux data coupled with infrared thermography. The values are then averaged to a continuous medium by bisection method.
A two-phase numerical simulation is after conducted on a macroscale tridimensional reconstruction of samples obtained by X-ray tomography.
Samples’ thermal conductivity is then fitted to cope with the averaged continuous medium and leads to the cross-validation of the experiments.
The preliminary results show that most of the studied vegetal cover samples are thermally resistive, in-line with field measurements [6]. Infrared thermography shows high heterogeneity in thermal response. Yet, some further work is needed to better understand the linkage between water saturation and hydraulic and thermal properties’ variability. Such study allows the generation of computationally-efficient boundary conditions of this bryophytic layer for large scale climate change impact models.
The safety against radioactive waste stored deeply in the underground is principally at risk where groundwater can attack the metallic waste canisters. For a performance assess-ment of a geological repository, it is therefore imperative to know the groundwater flow system over the projected lifetime of the repository. According to the current legislation in Germany, this will be a million years /STA 17/.
Over the past million years, several cold ages have occurred and have brought permafrost conditions basically to all potential sites for a radioactive waste repository in Germany (e.g. /VAN 93/). It can therefore be expected quite safely that any conceivable repository will sooner or later be subject to these conditions again. This is significant because permafrost will have a considerable impact on groundwater flow as the ground freezing tends to sepa-rate aquifers in the underground hydraulically from the surface.
However, it is also known that even under permafrost conditions there are local volumes of unfrozen ground, called taliki, connecting the surface waters with unfrozen aquifers. Flow of contaminated waters from a possible leakage in the repository could thus be directed towards such taliki and reach the surface concentrated in single spots (e.g. /JOH 16/). Taliki are thus a key feature in the assessment of a possible exposure of the biosphere to harmful radioactive substances.
Talik formation is presently quite intensively investigated in the framework of global warm-ing and thereby refers to melting processes (e.g. /PAR 18/). In case of geological storage of radioactive waste, by contrast, the question is rather, where open taliki will remain in an otherwise increasingly freezing ground. Taliki are not accessible to direct observation even though they can be detected by laborious field work. Insight into the processes of talik formation might therefore alternatively be gained by numerical modelling.
A surprisingly large variety of mathematical formulations can be found in the literature that describe groundwater flow under freezing conditions including ice formation and may be applied to the problem at hand. To ensure that all relevant processes are appropriately addressed in an own model of choice, though, general balance equations for groundwater and heat flow are developed from scratch without prematurely introducing assumptions and restrictions. These balance equations are supplemented by constitutive equations and equations of state (EOS) covering also sub-zero temperature conditions. Additionally, a computationally less demanding set of EOS valid in the temperature range between 20°C and +60°C and hydraulic pressure up to 20 MPa has been developed. The ensuing math-ematical model is then numerically realised in the framework of the code COMSOL Mul-tiphysics.
For realistic boundary conditions at the model surface, the air temperature evolution over the last 400,000 years determined from ice cores from Antarctica /JOU 07/ has been adapted to the location of present Germany. Heat flux from inner earth can be shown to be approx-imately constant over this time period of time. First results from modelling ground temperatures during the beginning of an ice age confirm a thermal shadowing of the cooling ground under large aquatic surface features.
When a temperature difference is applied over a porous medium soaked with a fluid mixture, two effects may be observed, a component separation (the Ludwig-Soret effect, thermodiffusion) [1,2] and a pressure difference due to thermo-osmosis [3,4]. In this work, we have studied both effects using nonequilibrium thermodynamics and molecular dynamics. We have derived expressions for the two characteristic parameters, the Soret coefficient and the thermo-osmotic coefficient in terms of phenomenological transport coefficients, and we show how they are related [5]. Numerical values for these coefficients were obtained for a two-component fluid in a porous matrix where both fluid and solid are Lennard-Jones/spline particles. We found that both effects depend strongly on the porosity of the medium and weakly on the interactions between the fluid components and the matrix. The Soret coefficient depends strongly on whether the fluid is sampled from inside the porous medium or from bulk phases outside, which must be considered in experimental measurements using packed columns. If we use an equimolar methane/decane mixture in the bulk as an example, the results for the Soret coefficient give that a temperature difference of 10 K will separate the mixture to about 49.5/50.5 and give no pressure difference. In a reservoir with 30 % porosity, the separation will be 49.8/50.2 whereas the pressure difference will be about 15 bar. Thermo-osmotic pressures with this order or magnitude have been observed in frost-heave experiments [6].
A detailed study of a two-component fluid in slit pores revealed that the thermo-osmotic effect was driven by a thermal Marangoni effect, by the gradient in a temperature dependent surface tension along the pore walls [7].
Image segmentation is a prominent process in Digital Rock Physics analysis. It directly affects the accuracy of the fluid-flow simulation in porous mediums. The accuracy of conventional global thresholding segmentation depends on image resolution. Coarse-resolution digital rock images are normally used to avoid high computation costs in processing high-resolution images that usually have more accurate results. But the image segmentation implemented on the low-resolution images becomes more arbitrary and thus generates less satisfying simulation results. This work develops a novel local-minima 3D image segmentation method that can improve the accuracy of simulations of fluid flow in low-resolution rock images. It adopts two global threshold values to capture the convinced pure pore and solid phase. Voxels having greyscale values between the two thresholds are assigned to a temporarily uncertain phase. A search algorithm is then applied to find the local minima in the uncertain region. These local minima are pores while the rest are solids. Indiana Limestone and Bentheimer Sandstone digital rock images scanned at different resolutions are studied to validate the local-minima segmentation method. We apply the conventional global thresholding and the proposed method to these digital models and study the pore- and throat-size distributions by extracting pore networks by a maximal-ball algorithm. The result shows that the local-minima method yields networks that are more accurate than those generated by the global thresholding method. In addition, we calculate the permeabilities of these models by a Lattice-Boltzmann method. The local-minima segmentation method yields an average absolute permeability error of 23% for the Indiana Limestone and 13% for the Bentheimer Sandstone, whereas the global thresholding method yields 202% and 83.67% errors, respectively. The result demonstrates that the carbonate rock is impacted more by the coarsening of image resolutions. Our segmentation technique can improve the overall accuracy of fluid flow simulations in low-resolution digital images.
The performance of Ground source heat pump systems (GSHP), and heat losses along well bores of high-temperature aquifer thermal energy storage systems (HT-ATES) and geothermal wells is strongly affected by conductive properties$^{(1,2)}$.
The anticipated growth of the geothermal and HT-ATES in line with global potential of geothermal energy production of between 125 and 1793 EJ/yr $^{(3)}$ augments to the relevance of in-depth understanding of conductive properties. Apart from heat losses affecting performance, shallow groundwater used for drinking water could be affected by the heat transfer$^{(4,5)}$. This can result in (bio)chemical changes in the water composition$^{(5,6)}$. This could potentially reduce the amount of suitable drinking water reserves for future use. In addition, too much heat loss in the cold subsurface could yield a risk of the formation of thermal plumes, which could in turn compromise the production of neighboring geothermal wells in urban areas$^{(7)}$.
Uncertainties in the thermal conductivity of an aquifer can affect the efficiency estimations of a single HT-ATES doublet. Using DoubletCalc$^{(8)}$, it was determined that this especially plays a role during the initial three loading cycles with a difference of up to 3.5% in efficiency. Others have shown even larger impacts, where an increase of 12.5% of the thermal conductivity reduces the total aquifer technical potential with 25–33%, while decreasing by 12.5% results in a 29–49% increase$^{(3)}$.
Understanding and being able to measure and predict the thermal properties both on the centimeter- and meter scale is challenging. Experimental determination is typically on the millimeter scale, most numerical simulations use solid rock-, oil and gas- and construction industry values and well thermal response tests lumpsum many different sediment types into one value.
A semi-automated setup was developed based on the needle-probe method to create an experimentally based understanding of the influence of different interrelated physical properties of unconsolidated sediments on the thermal properties of such sediments. Through a series of experiments, the impact of several sediment configurations have been investigated. Sediments were selected with various amounts of complexity in the composition or layers or layered orientation. For parameter isolation and model calibration, the impact of grain size, shape, porosity and water content was determined using amorphous soda-lime glass.
A numerical model with a radial symmetric finite volume formulation was used to determine the various thermal properties of the sediment sample, using an ensemble smoother with multiple data assimilation (ES-MDA) to inverse fit the model to the experimental data.
The combined experimental-numerical approach provides a reliable and reproducible method for determining the thermal properties of unconsolidated sediments and porous media in general and a means to determine the validity of the numerical model calculations. In oncoming research projects, the experimentally validated results will provide the input for upscaling and validation in a real life ground source heat pump setup.
Porous or fibrous complex medias are widely used for energy applications such as heat storage, thermal insulation, solar absorbers, heat exchangers... There is a need to develop methods that are relevant to solve the heat equation in those complex medias.
Monte Carlo method can be used to solve parabolic partial differential equations such as heat equation in complex geometries [1,2,3] or porous media. It relies on reformulating the thermal model first as an integral and then as an expected value introducing a probability density function. An important point is that this method does not require a volumic mesh which makes it relevant for complex geometries.
Randomly generated paths carry information (known temperature or flux on a boundary, volumetric heat source...) in their weights.The observable - local temperature, mean temperature on a given surface - is then evaluated by computing the arithmetic mean of the weights, based on the Law of Large Numbers. It is noticeable that Monte Carlo method does not evaluate a temperature field but only the observable. Therefore, it reduces the amount of data to handle for post-treatment. The Monte Carlo algorithm can easily be parallelized since each path is independently computed on a single processor. Based on the Central Limit Theorem, the result is always given with its variance and then with the associated uncertainty.
In this work, we solve the thermal model in a diphasic complex porous media.
Geometry has been obtained through tomography technique and is composed of $8 \times 10^6$ triangles. This sample has been chosen for its complexity: large range of spatial scales, hollow fibres... Computations have been performed with the free and open-source software Stardis (https://www.meso-star.com/projects/stardis/stardis.html) which is suitable to take conduction, convection and radiation transfers into account. Based on recent work of Tregan [4], Stardis has been extended to non-linear cases to take the radiative term - difference of temperatures to the power four - into account without linearization which is crucial when the difference of temperatures is high.In the present work, the thermal model has been successfully solved to determine the apparent conductivity tensor with and without radiative transfers.
Further work is required to investigate how to solve other advection-diffusion equations with this Monte-Carlo method.
Nanopore arrays, fabricated by the track-etching technique, mostly in Polycarbonate (PC) or Polyethylenterephthalate (PET) foils, are commercially available. These are formed by irradiation with highly energetic ions, such as Ar, from e.g. a cyclotron, and chemically etching the damage tracks up into cylindrical nanopores. For the present study, a linear heavy ion accelerator was applied, and, next to PC and PET, also Polyimide with a better chemical and thermal stability than the polyesters was used. The foils were up to 30 µm thick, the nanopores with an areal density up to 10-9 cm-2 had a diameter of down to 10 nm, corresponding to very high aspect ratio tubes.
Uniformly coating the inner walls of those nanopores with a thin reactive metal film constitutes a challenging process, since the thin film material transport into the nanopores has to be well controlled, particularly to avoid clogging of the apertures. By means of a kinetically controlled electroless deposition process, the nanopore inner walls were uniformly coated with thin films of gold, platinum, palladium, platinum-ruthenium, platinum-palladium, and palladium-coated nickel, leading to respective nanotubes embedded in the polymer foil. This has been evidenced by cross-sectional SEM and TEM investigations.
Embedded Pd nanotubes were tested for their performance as catalyst for a flow-through reactor. In UV-Vis absorption spectrometric measurements, they showed a very high efficiency towards the nitrophenol reduction to aminophenol, used for the fabrication of paracetamol (acetaminophen), the well-known analgetic and antipyretic [1]. The nanotubes show an advantage over comparable nano particle based systems: in contrast to the latter, they are more mechanically stable and do not agglomerate, leading to a better long-term stability of the catalyst.
When the ion beams tracks in the polymer foil are crossed and the resulting nanopores are coated with metal, the dissolution of the polymer foil leads to free standing porous nanotube networks. Due to the cross-linking, they are mechanically very stable, despite the very small diameter of the individual nanotubes. With their large internal surface area, they constitute highly efficient catalysts for the methanol oxidation in Direct Methanol Fuel Cells. This has been shown for Pt-Pd alloy and Pd-doped Ni networks by means of electrochemical half-cell methanol oxidation measurements [2, 3].
In electrochemical devices, porous transport layers (PTLs) bridge the gap between flow fields and catalyst layers. They provide pathways for liquids and gases to be distributed over and removed from the catalyst layer, provide mechanical support as well has thermal and electrical conductivity. As the electrical current drawn from such devices is directly linked to the flow of reactants, the mass transport capabilities of the PTLs become especially critical at high current density operation. State of the art PTLs are primarily fiber-based (e.g. carbon, titanium) as they satisfy the complex requirements posed by their operating environment. However, in two phase counter flow operation, undesirable liquid accumulations and subsequent blockage of gas transport pathways can occur. Excessive accumulations of reaction products or a lack of fresh reagents stall out the electrochemical conversion, limiting the achievable power density. To overcome these limitations, alterations and improvements to the base material have been investigated to guide liquids and gases into dedicated pathways within the porous structure[1–4]. While they show varying degrees of success, they require additional processing steps, adding cost, and are still limited by both phases having to compete for the same pore network.
We explore the synthesis of hierarchical PTLs containing dedicated pathways for the transport of liquid and gas, realized by a difference in pore sizes on two distinct length scales. While 3D printing is positioned as a promising manufacturing route, currently this method lacks the production speed and resolution. Therefore our approach is focused on an alternative synthesis route whereby the porous material is generated by co-depositing a metal and a gas from a solution containing the metal salt and a source of protons to form hydrogen gas. The hydrogen acts as dynamic template which, together with the deposition kinetics of metal at high overpotentials, forms a structure containing macroscopic and microscopic pores (Figure 1a). This type of structure has in the past been used successfully to improve boiling heat transfer[5], and has been postulated to find application in other electrochemical devices such as batteries or fuel cells[6].
In this talk, I will discuss the necessary development steps to adapt this material for the use in electrochemical systems. At the core of the synthesis route we conceptualized an approach to manufacture self-standing PTLs from this material while preserving its hierarchical microstructure. This enables its use as transport layer and allowed for the application of a wide range of characterization methods to link the synthesis parameters to the resulting material microstructure. Structural information (Figure 1b) obtained through X-ray tomographic microscopy was used to perform transport simulations and ascertain the potential of the material as PTL. The simulations showed increased diffusive transport in dry and liquid filled state compared to state-of-the-art materials. Through careful tuning of synthesis parameters and post treatment steps, the mechanical stability was improved substantially to the point where the integration in electrochemical systems is possible. If successful, this could open the door to a new class of PTLs tailored to the transport requirements of a given system.
Additive manufacturing, commonly called 3D printing, is increasingly applied in numerous disciplines. The most common type, Fused Deposition Modelling, manufactures 3D-printed parts by extruding a filament of molten material layer upon layer. Upon solidification of the molten filament which cross-section has rounded corners, air gaps are created between each layer (Biswas, Guessasma, and Li 2020). Given the presence of those air gaps, this 3D-printed material can be defined as a porous material, for which mechanical properties are then dictated by the classical laws of poromechanics. Considering the internal length scale introduced in the system via porosity, we postulate that these manufacturing imperfections influence the 3D-printed material mechanical size effect, which has been shown to exist in various studies (Bell and Siegmund 2018; Wu, Chen, and Cheeseman 2021). Here we show that this size effect can effectively vanish if air gaps and sample size are simultaneously scaled. By fine-tuning certain printing parameters such as printing speed (Lanzotti et al. 2015) and printing temperature (Afonso et al. 2021), we find it feasible to maintain the shape and distribution of air gaps while varying sample size. Given the possibility of scaling the 3D-printed material’s microstructure along with the sample size, we are left to check whether this is enough to effectively remove the size effect phenomenon previously observed. From our results on cubic samples of 3D-printed polylactic acid (PLA) (Figure 1), we obtain similar stiffness (3.1% differences) and uniaxial compression strength (3.2% differences) when the microstructure is scaled with the sample size, whereas 19.8% differences in stiffness and 12.6% differences in strength are obtained when the microstructure is fixed, see Figure 2. With this study, 3D-printed material mechanical size effect can be linked to the printing parameters straight-forwardly, which is a starting point towards predicting more directly the mechanical behaviour.
Low cost hydrogen production is essential to meet global hydrogen production targets by 2050. Therefore research into alternative water electrolysis device design may lead to reduction in capital costs. One emerging alternative to the capital cost is the use of membraneless electrolysers. In particular we focus on diverging flow through membraneless devices which utilise cell and porous electrode geometry to separate hydrogen and oxygen without a separator or membrane. They can also utilise alkaline conditions allowing for lower cost catalyst and construction materials.
However, the technology is not commercial and the influence of the design of the device and the manufactured porous electrode properties are unknown. Computational fluid dynamic simulations (OpenFOAM) using the volume of fluid method is used to model the two-phase flow of hydrogen and oxygen bubbles coupled to electrochemistry. Different device geometry, porous electrodes, flow and current density are varied to investigate their impact on the cell potential.
Electrolyte flow distribution and scaling of the devices are investigated, which are highly dependant on the sizes of the pores, electrode gaps and higher Reynolds number flows. The initial results show that there is an interplay between the pore size and the electrode length in order to maintain uniform flow across the electrode, which is important to reduce bubble blockage of the electrode surfaces. The effect of the electrode microstructure on the current density distribution are evaluated and strategies to avoid bubble accumulation are discussed. Changes to the porous electrode morphology through advanced manufacturing techniques, along with the wettability and device flow geometry could lead to higher efficiency, low capital cost water electrolysis.
A new automated method for the fabrication of functional 3D porous structures directly on planar standard silicon wafers has been developed [1,2]. A typical approach comprises the filling of a mold pattern with micron sized particles of the desired material, and their fixation via atomic layer deposition (ALD). It has been demonstrated that it is possible to manufacture for example micromagnets from NdFeB powders that can be used for energy harvesting.
In order to achieve smallest dimensions and highest filling factors, the utilization of dry powder as the starting material is beneficial. The new approach utilizes the superimposition of high- and low frequency oscillations for particle mobilization in order to achieve optimum mold filling. Additionally, rubber balls are applied for densification of the powder packing.
For verification of the application properties, micromagnets were created from 5 µm NdFeB powder on 8” Si wafers, using the novel automated mold filling technique, as well as an existing manual one for benchmarking purposes. Subsequent atomic layer deposition were utilized to agglomerate the loose NdFeB particles into rigid microstructures. The magnetic properties and inner structure of the NdFeB micromagnets were investigated. It is shown that the novel automated technique outperforms the manual one in major terms. In addition, examples for further materials and applications will be briefly discussed.
The adaptation of additive manufacturing for chemical flow reactors has recently gained momentum as the manufacturing methods become more advanced and manufacturing equipment is increasingly affordable.[1] Periodic open cellular structures (POCS) from additive manufacturing have lately received growing attention. Compared to randomly structured substrates such as metal foams, POCS offer an ordered structure, which promises improved flow control and homogeneous flow profiles at comparably low pressure losses. However, the achievable minimum cell size is still limited to few millimeters, which results in low specific surface areas compared to conventional metal foams.[2] Moreover, finely-resolved 3D models that define these POCS require an extensive amount of computing power with increasing resolution and corresponding large file sizes. A scale-up to industry-relevant sizes is therefore limited. Consequently, the direct definition of micro-features in large 3D models is infeasible, as STL files that directly define POCS will reach gigabyte sizes for reactors with outer dimensions of few centimeters and with POCS cell sizes of few millimeters and strut sizes in the millimeter range.
We have recently shown that the porosity and thus the specific surface area in selective laser melting (SLM) can be globally controlled via the laser energy density.[3] In this work, we now present a method to create defined microfeatures, without the need to explicitly define the features in a CAD model, thus avoiding large file sizes. Our method requires no definition of micro-features in 3D models such as STL files, but rather makes use of the scan lines, i.e., the path with which the laser proceeds through the powder bed. By actively controlling the scan line pattern of each layer, repeating structures are created implicitly. The scan line pattern is defined in the print job file, requiring only a 3D model of the macrostructure, which may be as simple as a cylinder, thus only amounting to a file size of few kilobytes.
As example for the simplest form of POCS, a cubic structure is created with cell sizes as small as 400 µm and strut thicknesses of approximately 100 µm. The resulting structures are analyzed by reconstruction of 3D models from micro-computed X-ray tomography.[4] From the model, the strut thickness, cell size and specific surface area can be derived, which presents as beneficial compared to other imaging methods such as SEM, where the analysis is limited to the surface.
Figure 1 presents a model POCS structure, consisting of cubic cells with a cell size of 800 µm. The strut sizes and specific surface area are approximately 120 µm and 5.6 x 10³ m²/m³, respectively, which is well in the range of commercial metal foams.
The presented method therefore shows great promise to propel the design of highly active, open-porous reactor systems to industrially relevant scales with small feature sizes that have sparsely been reported in literature so far.
Vegetation like trees and grass is known to have a cooling effect that naturally enhances the thermal comfort for pedestrians in cities. While shadowing and transpiration by vegetation cool the urban environment during daytime, the blocking of cooling to sky by longwave radiation during night and increase in relative humidity might have an adverse effect. Also, while urban wind cooled by trees may have a cooling effect at downwind urban places lacking vegetation, dense trees may decrease the heat removal by ventilation from streets due to wind blocking effect. To study these complex interactions, a detailed vegetation model is needed to take into account the momentum, heat and moisture transfer processes taking place at different scales.
The authors developed over the years an urban microclimate model, coupling (1) a computational fluid dynamics (CFD) model for air, heat and moisture flow in the air domain, (2) a longwave and shortwave radiation model for radiative exchange between urban surfaces, sun and sky, (3) a coupled heat and moisture transport model for moisture transport in building facades, soil, pavements, and (4) a wind driven rain model. The set of equations is solved in OpenFOAM and the model is open source (urbanMicroclimateFoam at the Chair of Building Physics). Vegetation is modelled as a porous medium by introducing sink and source terms in the momentum, heat and moisture transport equations. The momentum sink is modelled by introducing a drag coefficient that depends on leaf area density. The heat and moisture transport from leaves is modelled by a leaf model depending on leaf area density in each vegetation cell, and the model accounts for convective and latent heat, and vapor transport depending on stomatal resistance. Solar radiation shadowing is modelled using a radiation attenuation model, while longwave radiation is modelled using a view factor method. Special shapes for trees based on Lidar information are introduced, limiting element discretization and taking into account the growth of trees. A special interface model is introduced for the modelling of heat and moisture exchange between grass, air and soil.
The model has been validated and applied to different case studies. In a first case study we analyze the influence of tree size on the pedestrian thermal comfort in street canyons. An optimal tree age of around 20-30 years is found, maximizing shadowing and transpirative cooling, while not blocking heat removal by air flow. A second example shows that, depending on wind direction, trees can cool down city parts down-windward, even when no trees are present at these locations. A third example shows the redevelopment of parking spaces as green areas on St. Helene island in Montreal and the possible cooling effects during heatwaves.
Salt precipitation from evaporation is a key factor for soil degradation in arid and semi-arid regions. Evaporation-induced water movement transports dissolved salt ions to the surface of the porous medium where they accumulate. When the solubility limit is reached, salt starts to precipitate and forms crusts on top (efflorescence) or inside (subflorescence) of the porous medium depending on the type of solute in solution. The aim of this study was to non-invasively investigate the development of subflorescent MgSO$_4$ crusts in evaporating porous media. In particular, micro-X-ray computed tomography (XRCT) was used to investigate the development of the volume fraction of precipitated salt, brine, and air and single-sided unilateral nuclear magnetic resonance (NMR) measurements were used to determine high-resolution near-surface water content profiles during evaporation. In a first step, sand packings with deionized water and MgSO$_4$ solution with an initial concentration of 0.96 mol/L were evaporated while periodically making XRCT and NMR measurements. It was found that void, brine, salt, and sand could not be segmented in the XRCT images because of limited contrast between the brine and the salt phase. However, a downward movement of the evaporation front was observed using unilateral NMR, which involved salt precipitation that deformed the top of the sand. To avoid deformation, porous sintered glass with similar porosity, intrinsic permeability, and internal surface area as the sand packings were prepared in a second step. It was found that evaporation of deionized water was similar for sand and sintered glass, which was related to the similar evaporation conditions and properties of the porous media. In contrast, evaporation of saline solution and salt crust formation differed in both porous media. The delayed crust formation in sintered glass was attributed to the smooth surface and to the highly supersaturated magnesium sulfate solution, which reduced nucleation and thus hindered crust formation. XRCT measurements on the sintered glass sample showed that salt crystals grew into the void space that was not occupied with liquid before. This suggests that film flow supported crystal growth, which needs to be analyzed in more detail in future studies. It is concluded that the surface properties of the porous medium and properties of the highly supersaturated solution (i.e., viscosity) significantly affect evaporation of MgSO$_4$ solution and the formation of subflorescent MgSO$_4$ crusts.
Recent years have recorded an increase of the rainfall intensity which is met by a reduced infiltration capacity causing severe surface runoff, flooding, and groundwater depletion. Thus, the ecosystem functions of groundwater aquifers are at risk. The project Smart-SWS couples flood protection and drought management by infiltrating flood waves into porous aquifers close by (Flood-MAR). The water held back in flood retention basins is conditioned in the infiltration ditch to meet the quality criteria for groundwater recharge. Laboratory experiments to select suitable materials for the conditioning of the infiltrated waters are run. Here, the focus is on the removal of colloids and particles. As the infiltration is at irregular time intervals, we expect and test extended drying periods. A GIS-based site selection scheme has been developed and applied to select pilot sites. It respects (hydro)geological features as well as all kinds of protection zones and agricultural, municipal, and industrial use cases. To meet the pronounced asymmetry of infiltration (rapid infiltration requires high hydraulic conductivity) and long term storage (works best with low flow velocities) geotechnical measures like sheet pile walls or sand/cement injections are required. The concept is based on a minimum invasive approach: groundwater flow should not change at normal conditions and stagnation of groundwater has to be prevented. A hydrogeochemical model is developed to assess the reactions along the infiltration path and their effects on the integrity and stability of the aquifer.
Deterioration of water quality has become a critical global challenge. Commonly contamination of groundwater is caused by human activities at the surface including petroleum leakage from fuel stations, leakage of substances used and produced within manufacturing and chemical industries, and also importantly substances used in the farming industry [1]. The contaminants of significance include a vast array of chemicals such as polyfluoroalkyl substances, fertilizers, pesticides, and antibiotics. Investing in new technologies to improve the quality of water resources is the key to resilience in a changing world.
Biochar is a stable and porous carbon-rich adsorbent material that is used to remove contaminants from water [2]. Biochar is produced through the pyrolysis process using relatively inexpensive and sustainable material (biomass) as feedstock. The performance of biochar depends on the biomass properties and the parameters of the pyrolysis process (such as pyrolysis temperature and heating rate). Pristine biochar with low surface functionality and small pore sizes offers limited adsorption capacity. A key consideration in improving biochar adsorption efficiency is to choose suitable biomass and activation of biochar [3,4]. Chemical activation is commonly used in which chemicals (e.g., acid or alkali, metal oxide or metal salt) are used to activate biochar.
Here we focus on contaminants introduced to the environment as a result of processes used within the farming and textile industries. Methylene blue (MB) is a dye which causes contamination in textile industry wastewater streams. Nitrate (NO3-) is a major contaminant that is caused by excessive use of fertilizers in the farming industry. MB removal from water has been researched significantly, in particular, sorbents such as activated carbon and biomass have been studied extensively. NO3- removal is more challenging, therefore, in most cases nitrate levels in water are reduced through dilution rather that removal. Here we report on the efficiency of clay-biochar composites for removal of NO3- and MB.
Synthesis of functional clay biochar offers an economical method to remove contaminants from water [5]. We use wheat straw as the feedstock since it is abundantly available in the northeast of England and is considered as a waste product of farming industry in the region. To prepare the wheat straw was dried and milled. Kaolinite, montmorillonite, and bentonite clay powder was suspended in deionized (DI) and mixed with milled wheat straw. The mixture was stirred for 24h and then separated through sieving. The clay feedstock was oven dried and pyrolyzed at 600°C in a tube furnace to get clay-biochar. The characterization of functionalized biochar was performed using SEM, XRD, TGA and FTIR. We show that the clay-biochar can remove MB with over 99.5% efficiency at pH of 11. Our experiments show that clay-biochar cannot remove nitrate from water. We compare our results against pristine biochar and the clay minerals used. In conclusion, modification of biochar causes changes in its surface functional groups which contribute to biochar hydrophilicity, negative charge, and sorption ability for cations. This experiment shows that clay-biochar is a cost-effective and high-performance material for removal of MB in high pH environments.
This study aims to investigate the interplay between mixing in porous media and mineral precipitation from groundwater. These processes are fundamental in salt lakes/lagoons, soil salinization, land desertification, alteration of soil’s mechanical properties and reliability of wastewater disposal. These contexts are mainly found in arid and semi-arid regions, where the evaporation plays a key role in the coupling of those processes, as it drives upward the groundwater flow and reconcentrates the solutes at the exposed aquifer surface. The resulting increase in shallow water density consists in a gravitationally unstable condition that relaxes through the formation of saline fingers and the development of a free convective regime in the entire aquifer. In this way, solutes sink while they are precipitating as mineral phases in the intergranular pores, in case of saturated waters. Then, effects on the evaporation rate and aquifer geochemistry derive from the changes in porosity and relative amounts of chemical species. Here we present a variable-density flow model coupled to reactive transport to replicate a typical evaporite environment and the aquifer beneath, in fully saturated conditions. The numerical model simulates well the aquifer recharge according to realistic values of evaporation and permeability, meanwhile, the reconcentration of solutes and the resulting fingers of saline water that grow, merge, diffuse, and sink. Evidence of the above-mentioned diluting mechanism is visible in the periodic oscillation of maximum density in the system (i.e. linear function of the concentration) with time, which drastically decreases after a local maximum value corresponding to finger formation event. This sensitivity is observed in the mixing indicators too, while the system, started at a homogenous condition, goes through an increasing segregation to progressively recover to better mixed stages, when the salinization of the deeper aquifer is observed, and the inflow of fresh water is limited to a superficial wedge. Indeed, changes in the fingering dynamics lead to substantially different evolution of the aquifer flow and geochemistry, that is, the combination of evaporation rate and permeability, which define the boundary layer Rayleigh number (Wooding, 1997), enhancing or not the convection. We observe that for higher values of this number, the maximum density reached in the system decreases, consistently with a more dynamic evolution, faster and stronger aquifer salinization, and a thinner freshwater wedge. The model simulates adequately the precipitation of minerals within the most superficial soil layer and a complementary decrement of porosity, showing a heterogeneous spatial distribution depending on the fingering dynamic occurring beneath as well as a different amount depending on the strength of the convective regime. Thus, the numerical simulations can be implemented as a predictive reactive transport tool applied to geo-engineering and agricultural studies.
By 2030, a third of the population in developing countries will reside in areas where the gap between water demand and supply is predicted to be over 50%. Agriculture is responsible for over 71% of annual water withdrawals worldwide (currently ~ 3,100 billion m3 and predicted to be ~4,500 billion m3 by 2050). The amount of water held as groundwater is more than 100 times the amount collected in rivers and lakes. Globally, the per capita irrigated area has been decreasing for 30 years. Population growth and increased irrigation requirements have resulted in groundwater mining with a universal increase in water table depth.
A preliminary study aimed at developing a bioinspired pump capable of passively lifting subsurface water is presented. The bioinspired system uses emerging materials and concepts in geotechnical engineering to mimic the wicking mechanisms that plants use for transpiration. Upon droughts and dry periods, soil desaturates, its hydraulic conductivity drops and less water is made available to plants. As water inflow to the plant decreases, stomata close to stop transpiration and prevent plant dehydration, photosynthesis ceases, and plants wilt. Upon dry conditions, to delay soil desaturation, preserve a high hydraulic transmissivity and extract water for longer, plants secrete a gelatinous substance named mucilage around root tips. This naturally engineered “grout” fills the pore space by reducing the pore diameter and increasing the soil air-entry value (creating smaller capillaries).
A capillary network aimed at mimicking plant mucilage secretion is formed by the injection of colloidal silica-based hydrogel (CS) into the soil. It is shown that the presence of hydrogel enhances soil hydraulic conductivity and water retention capacity thus enabling better water uptake during periods of drought. A hydromechanical characterization of the hydrogel-soil system and preliminary experimental results of the hydrogel interaction with the roots of various species of shrubs will be presented.
Keywords: Colloidal silica hydrogel; plants; drought; capillary rise; roots
Plants acquisition of soil resources such as nutrients and water will be severely impeded in the near future as a consequence of climate change. Root hairs, tubular extensions of epidermal root cells, substantially increase the contact area between roots and soil and are hence considered a key rhizosphere trait increasing the capacity of plants to capture soil resources. While their pivotal role in the uptake of immobile nutrients such as phosphorus is well accepted, their effect on root water uptake remains controversial as it varies across plant species.
By means of image-based modelling, our objective was to identify environmental conditions (e.g. soil water content) and hair traits (e.g. root hair length and density) that determine the effectiveness of root hairs in root water uptake. Furthermore, we investigated the effect of drought stress-induced root hair shrinkage on root water uptake.
Using synchrotron radiation X-ray CT, we scanned root compartments of 8 days old maize seedlings (Zea Mays L.) grown in loamy soil, a complex porous medium. The acquired image-data served as a basis for our image-based 3D root water uptake model. By solving Richards equation numerically, we computed the propagation of water potential gradients across the root-soil continuum. The high spatial resolution of the acquired images allowed to explicitly take rhizosphere features, such as root hairs, root-soil matrix contact and aggregate structure into account. To determine the key parameters governing the effectiveness of root hairs in water uptake, we compared a set of six maize root compartments of approx. 1.4mm length before and after digitally removing their hairs. The quantification of root hair turgor-loss in response to progressive soil drying allowed us to implement hair shrinkage within our model.
We found that the effectiveness of root hairs in root water uptake mainly depends on 1) the root hair induced increase in root soil contact and 2) root hair length. Furthermore, our results suggest that root hairs potentially facilitate root water uptake under dry soil conditions (< -0.1MPa). However, in the dry range, root hair shrinkage severely impairs the effect of hairs. Depending on the turgor-loss curve, root hairs may still provide a positive effect on root water uptake in a narrow range of soil matric potential.
In summary, the effect of root hairs on root water uptake is determined by soil water content, root-soil contact, root hair length and the turgor-loss point of hairs.
The atmosphere-soil system forms a highly coupled system, which makes key processes such as evaporation complex to analyse as the mass, energy and momentum transfer is influenced by both domains. To enhance the understanding of evaporation processes from soils, stable water isotopologues are suitable tools to trace water movement within these systems as heavier isotopologues enrich in the residual liquid phase. Due to the complex coupled processes involved in simulating soil-water evaporation accurately, quantifying fractionation during flow and transport processes at the soil-atmosphere interface remains an open research area. In this work, we present a multi-phase multi-component transport model that resolves flow through the near surface atmosphere and the soil, and models transport and fractionation of the stable water isotopologues using the numerical simulation environment DuMuX. Using this coupled model, we simulate transport and fractionation processes of stable water isotopologues in soils and the atmosphere by solving compositional flow equations and by using suitable coupling conditions at the soil-atmosphere interface instead of commonly used parameterization.
In a series of examples of evaporation from bare soil, the transport and distribution of stable water isotopologues are evaluated numerically with varied conditions and assumptions, including different atmospheric conditions (turbulent/laminar flow, wind speed) and their impact on the spatial and temporal distribution of the isotopic composition. Building on these results, we observed how the enrichment of the isotopologues in soil is linked with the different stages of the evaporation process. A qualitative study is conducted to verify single fractionation processes in our approach. As an outlook, we will present how the coupling of the free-flow and the porous medium domain allows us to use atmospheric measurements (which are often conducted at 2 m above the soil surface) and account for convective transport in the free-flow region. Thus, we can validate our results by using field-scale lysimeter experiments.
We attempt to formalise the relationship between the poroelasticity theory and the effective medium theory of micromechanics. Assumptions of these approaches vary, but both can be directly linked by considering the undrained response of a material. To analyse the linkage between poroelasticity and micromechanics, we do not limit ourselves to the original theory of Biot. Instead, we propose a concise extension of anisotropic poroelasticity, where pore fluid pressure may vary within the representative volume element. As a consequence, the inhomogeneities are not necessarily interconnected---they may form separated pore sets that are described by distinct poroelastic parameters and pore pressure. Further, we attempt to incorporate the effective methods inside Biot-like theory and investigate the poroelastic response of various microstructures. We show the cases where such implementation is valid and the others that appear to be questionable.
A domain decomposition approach for flow simulations in poro-fractured media using non-conforming meshes is presented. Fractures in a porous medium can either act as preferential flow path, either represent barriers for the flow. When a geometrical reduction approach is used, as, e.g. in a Discrete Fracture and Matrix (DFM) model, fractures are represented as planar interfaces embedded in a three dimensional porous matrix. A formulation with six independent pressure variables and an additional field for the flux at fracture intersections is proposed to de-couple the problems on each fracture and in the bulk domain. A suitable cost functional is then minimized to recover the global solution. Each field can be discretized independently from the others, and on an independently built mesh. As the pressure solution in the porous matrix can be discontinuous across barrier interfaces, the eXtended Finite Element method with discontinuous enrichment functions is used to describe this kind of irregular behavior on a mesh non conforming to the irregularity interface.
The proposed approach has the advantage of a strong robustness to geometrically complex configuration and allows to take advantage of parallel computing techniques.
In this work, we present a mixed finite element formulation of the Biot problem based on the rotation and displacement for the elasticity, and Darcy velocity and pressure, for the fluid phase flow. The discretization of the problem is based on exact discrete complexes and also on a suitable choice of a quadrature rule to localize, in a multi-point fashion, and thus algebraically eliminate the rotation and flux variables. The resulting method has fewer degrees of freedom than the original one, leading to a cost-effective formulation of the poroelastic problem with a two fields formulation. Indeed, we consider lowest order Raviart-Thomas finite elements for the displacement and piece-wise constants for the fluid pressure, thus the number of degrees of freedom is equivalent to a mixed formulation of a single-phase flow. Numerical results show the expected convergence rates for the errors for all the variables, for both the four and two fields formulations. We extend this discretization strategy to account for faults, which are three-dimensional physical objects where one of the dimension (their thickness) is orders of magnitude smaller than the others, and their material properties might be very different than the surrounding porous media. To avoid excessive mesh refinement, in the discrete setting, we represent them as objects of codimension one and consider a new set of equations based on a dimensional reduction strategy. We can thus reformulate the problem as mixed-dimensional and exploit the properties of the discrete approximation to obtain reliable solutions at an affordable computational cost. The reference paper for this contribution is [1], see also references therein.
The control volume finite element (CVFE) method is inherently flexible for modelling flow and transport in complex geological features such as faults and fractures. The finite element method that captures complex flow characteristics is combined with the control volume approach known for its stability and mass conservative properties. The classical CVFE approach exploits two meshes: the element mesh that represents the petrophysical properties element-wise and the control volume mesh, centered on the element vertices, representing the saturation solution in the medium. The discrepancy between those two meshes introduces inconsistency in the transport solution especially along material discontinuities or abrupt material interfaces.
In this work, we present an original discontinuous formulation based on the CVFE method for modeling multiphase flow and transport in porous media. We introduce the element pair $P_{1,DG}-P_{0,DG}$ denoting a linear discontinuous Lagrangian velocity approximation and an element-wise pressure approximation, respectively. The formulation enables the use of a single mesh that, in return, does not exhibit the inconsistency issues described earlier. We validate the method and demonstrate the effectiveness of the approach with numerical examples of complex fractures in highly heterogeneous domains.
Given the high uncertainty of fracture characteristics in subsurface porous media, we focus in our work on the prediction of the mean or Ensemble Averaged Flow (EAF) field. Typically fractures can cover distances comparable to the size of the domain of interest. While classical homogenization only is valid for representative elementary volumes (REV) much larger than all embedded structures, the presented approach does not rely on such restrictions. The new model, which is formulated at this point for many isolated fractures, relies on a nonlocal multi-media description based on coupled integro-differential equations. It is shown how a previous description for fractures of equal length and aperture can be extended for much more realistic scenarios with multiple fracture families. With a series of numerical studies and comparisons with Monte Carlo reference data it is demonstrated that also for such more complex scenarios the devised sub-REV model accurately captures mean flow rates and pressure profiles for arbitrary domain sizes.
Objectives/Scope
Understanding the fundamental mechanisms of fracture-matrix fluid exchange is crucial for the modeling of fractured reservoirs. Traditionally, high-resolution simulations for flow in fractures often neglect the effect of matrix contribution on the fracture hydraulic behavior. In this study, we develop a multi-scale approach to capture the matrix-fracture leakage interaction and its impact on the hydraulic properties of roughed fractures.
Methods, Procedures, Process
Because of the multiscale nature of the fracture and matrix rocks, full physics Navier-Stokes (NS) simulation in both matrix and fracture media is not feasible. For such multiscale phenomena, we use NS equations to describe the flow in the fracture, and Darcy’s law to model the flow in the surrounding porous rocks. The hybrid modeling is achieved using the extended Darcy-Brinkman-Stokes (DBS) equation. With this approach, a unified conservation equation for flow in both media is applied. We use an accurate Mixed Finite Element approach to solve the extended DBS equation. Analytical solutions were used to verify the numerical method.
Results, Observations, Conclusions
Various sensitivity analyses were conducted to explore the leakage effects on the hydraulic aperture of rock fractures by varying the permeability of the surrounding medium, fracture roughness, and Reynolds number (Re). A series of pore-scale simulations for flow through roughed fractures were performed, and the results were used to develop a relationship between the flow rate and pressure loss. Streamline profiles show the presence of back-flow phenomena, where in- and out-flow are possible between the matrix and the fractures. Further, zones of stagnant (eddy) flow are observed within locations of large asperities of sharp roughness within the fracture and high Re. This implies the presence of dynamic trapping mechanisms that may impact the relative permeabilities and residual saturations within the fractures. Numerical results show the side-leakage effect can create non-uniform flow distribution in the fracture that deviates significantly from the flow with impermeable wall conditions. The proposed friction factor has the potential to be used as an upscaling tool to estimate the hydraulic properties of roughed fractures within permeable rocks in fractured reservoir simulations.
Novel/Additive Information
We develop a high-resolution approach to investigate the flow exchange behavior between the fracture and rock matrix. We investigate static and dynamic effects, including variable Reynold numbers, mimicking flow near and away from the wellbore. We show that local fracture characteristics such as roughness and tortuosity may impact the flow, which is often not accounted for in dual-porosity simulations. We propose a new upscaling friction factor to account for these mechanisms in field-scale reservoir simulations.
In the subsurface, fractures are discontinuities in the medium in the form of narrow zones. Fractures are very numerous and present at all scales, with highly varying sizes and permeabilities. The permeability of the neighboring rock matrix is generally about two orders of magnitude lower than that of the fractures. This is why fractures are preferential channels for flow and, therefore, play a vital role in a large number of industrial and environmental applications.
One commonly used geometrical representation of fractured porous media is the discrete fracture matrix model (DFM) in which fractures are represented as manifolds of codimension 1. The model for single-phase flow in DFMs is described in [1], where Darcy's law in the fractures includes an additional source term that takes into account the coupling with the rock matrix.
Meshing the fracture network is carried out thanks to a specialized surface mesh generator called MODFRAC [2]. The surface mesh is then used as input for a volume mesh generator named GHS3D [3]. We developed nef-flow-fpm, a mixed hybrid finite element (MHFE) code for simulating steady-state incompressible single-phase flow in 3D DFMs. The MHFE method is conservative and leads to a square, sparse, symmetric, positive and definite linear system. Both direct [4, 5] and iterative [6, 7] solvers are integrated in nef-flow-fpm. Our code has been validated on a test case from the benchmarks in [8].
Because of the growing geometric complexity in large fracture networks, test cases recently proposed in the literature are mainly 2D, or 3D but with a limited number of fractures. In this talk, with the help of nef-flow-fpm, we analyze the computational costs from simulations with fracture networks of increasing complexity. The goal is to assess the performance of the linear solvers mentioned before and the challenges they face. We propose large-scale test cases, up to 87 329 fractures, generated with a genetic algorithm [9]. As expected, direct solvers suffer from large memory consumption, while iterative solvers may need a large number of iterations. Thus, we conclude that it is necessary to develop a dedicated, robust and efficient linear solver for even larger networks with more than one million fractures [10].
We performed a set of numerical simulations to characterize the influence of mesh refinement and upscaling on flow and transport properties in fractured porous media. We generated a set of generic three-dimensional discrete fracture networks at various densities, where the radii of the fractures were sampled from a truncated power-law distribution, whose parameters were loosely based on field site characterizations. We also considered five network densities, defined using a dimensionless version of density based on percolation theory. Once the networks were generated, we upscaled them into a single continuum model using the upscaled discrete fracture matrix model presented by Sweeney, Gable, Karra, Stauffer, Pawar, and Hyman (2020). We considered steady, isothermal pressure-driven flow through each domain and simulated passive/conservative, decaying, and adsorbing tracers using a pulse injection into the domain. We calculated the effective permeability and solute breakthrough curves for each simulation as quantities of interest to compare between network realizations. We found that selecting a mesh resolution such that the global topology of the upscaled mesh matches the fracture network is essential. If the upscaled mesh has a connected pathway of fracture (higher permeability) cells, but the fracture network does not, then the estimates for effective permeability and solute breakthrough will be incorrect. Local false connections between fractures due to a coarse mesh result in more solute dispersion in the transport behavior, but to a smaller degree than if there is a mismatch in global connectivity. False connections cannot be eliminated entirely, but they can be managed by choosing the appropriate mesh resolution and refinement for a given network. Adopting octree meshing to obtain sufficient levels of refinement leads to fewer computational cells (up to a 90% reduction in overall cell count) when compared to using a uniform resolution grid and can result in a more accurate continuum representation of the true fracture network.
Dissolution mass transfer of trapped phases in porous media is an important phenomenon in various fields, such as groundwater contamination [1–3], groundwater remediation [4], geological carbon sequestration [5], and energy storage in geological formation [6]. In the process of dissolution mass transfer, a fluid, which could be a liquid, gas, or supercritical fluid, is trapped in porous media by a capillary force. In the presence of another flowing fluid the trapped phase then gradually dissolves with solubility as the driving force.
The difficulties in investigating this phenomenon is the method to observe the dissolution process inside the porous medium. Earlier studies were mainly conducted using upscaled mass transfer approaches [1–4], which was modeled as a single grid block of mass transfer without knowing the pore scale processes in the porous media. As a result, the interfacial area, which the place for the mass transfer to occur, was unknown. For that reason, the developed mass transfer models [1–4] were phenomenological to the porous media.
In this work, we used X-ray CT microtomography (micro-CT) to observe the pore-scale process of dissolution mass transfer inside a porous medium. The main goal of using the micro-CT is to measure the capillary trapping interfacial area during the dissolution process, and thus, mass transfer model that is non-phenomenological to the porous media characteristics can be developed. To generate the mass transfer model, we performed comprehensive experimental investigations [7–12] by using various porous media characteristics (unconsolidated porous media particle size and wettability), various trapped phase types (non-aqueous phase liquids (NAPLs) and gases), and various water velocity.
Along with the development of this models, additional phenomena that affect the mass transfer rate were elucidated. The first phenomenon is the dissolution fingering [9], which occur due to slight differences in local permeability caused by the spatial distribution of the trapped phase in the porous media. Dissolution fingering was found to reduce the mass transfer coefficient to a third. Another phenomenon is the two-stage dissolution process [7,8] that occurs due to the rapid dissolution rate of the capillary trapping represented by the dissolution ratio (ratio between solubility and density). As a result, dissolution occurs much faster than the solute advection, resulting in the accumulation of high local solute concentrations that hinder the mass transfer process.
Eventually, a non-phenomenological mass transfer model based on Sherwood, Reynolds, and Schmidt numbers was developed. To the best of our knowledge, this is the first non-phenomenological model of capillary trapping dissolution mass transfer based on Sherwood, Reynolds, and Schmidt numbers. We believe that this work could provide valuable insights for the porous media community, especially interfacial phenomena across scales.
Gas bubbles can form and grow in otherwise liquid-saturated granular media due to various physical processes, such as corrosion or the microbial decomposition of organic matter. These gas bubbles are typically non-wetting to the solid grains; as such, it is energetically costly for the gas to invade the narrow pore throats between grains. If the solid skeleton is sufficiently soft and/or the confining stress is sufficiently low, the gas can instead displace the solid grains to open macroscopic cavities. These gas cavities form in a variety of soft porous media, including seabed sediments, industrial waste ponds, and peatlands. An increase in the confining stress can trigger the collapse of these cavities, forcing the gas into the pore space. A quantitative understanding of cavity collapse is thus important for characterising the macroscopic mechanics of this three-phase system and for predicting the rate of gas venting to the surrounding environment. Here, we investigate this problem experimentally using a packing of hydrogel beads as a model soft porous medium. We complement our experimental observations with a novel phase-field model that captures the competing effects of elasticity and gas-liquid-solid interactions (capillarity). We study the deformation-driven collapse of gas cavities in a 1D setting, identifying the confining stress at which cavities collapse and investigating the reversibility of cavity formation and collapse under fluctuating confining stress.
Bubbles in subsurface porous media spontaneously coarsen to reduce free energy. Bubble coarsening dramatically changes surface area and pore occupancy, which affect the hydraulic conductivity, mass and heat transfer coefficients, and chemical reactions. Coarsening kinetics in porous media is thus critical in modeling geologic CO2sequestration, hydrogen subsurface storage, hydrate reservoir recovery, and other relevant geophysical problems.
We show that bubble coarsening kinetics in porous media fundamentally deviates from classical Lifshitz-Slyozov-Wagner theory, because porous structure quantizes the space and decouple the mass transfer coefficient from the bubble size. We develop a new coarsening theory that agrees well with numerical simulations. We further identify a pseudo-equilibrium time proportional to the cubic of pore size. In a typical CO2 sequestration scenario, local equilibrium can be achieved in 1s for media consisting of sub-micron pores so local equilibrium can be presumed, while in decades for media consisting of 1 mm pores so capillary equilibrium fails.
This work provides new insights in modeling complex fluid behaviors in subsurface environment. In addition, along with our preceding works, we demonstrate that the porous media rescale mass transport of discrete fluid systems, by 1) modifying the free energy vs. volume correlation, and 2) decoupling the mass transfer kinetics from blob size.
Geologic carbon storage is the most readily available technique that can store the carbon in a relatively larger volume. The higher injectivity and the larger storage capacity are required for the efficient and economical geological carbon storage. The storage capacity of reservoirs highly depends on the multiphase flow properties of the reservoirs, especially the residual saturation of CO2 . The CO2-compatible surfactants can enhance the storage capacity by increasing the residual saturation of CO2, due to the reduced interfacial tension between brine and CO2 . A few groups of CO2-compatible surfactants have been tested for their performance in controlling interfacial tension in various pressure and temperature conditions. The interfacial tension and contact angle on commercially available substrate, such as quartz mineral and sandstones, were experimentally measured. The pressure and temperature ranges 25~40℃ and 4~10MPa, respectively. Non-ionic surfactants perform better than ionic surfactants in the brine-CO2 system, whereas the performances were similar to each other in the brine-air system. The contact angel alteration become more prominent in higher pressures. Lower interfacial tension and higher contact angle, induced from surfactants, will lower capillary pressure, and thus increase the residual saturation of CO2. The lower capillary pressure also enhance the sweep efficiency during the geologic carbon storage operation. Further study with various and systematic measurement conditions are required for the operation design in field scale.
For a wide range of engineering applications such as transpiration cooling, filtration processes, heat exchangers and geothermal engineering, understanding how porous media with different topologies interact with turbulent free flows is crucial. For this purpose, the focus is on the efficient design, operation and optimisation of such engineering applications and relies on comprehensive understanding of the exchange of mass, momentum and energy across the interface between the porous medium and the free flow.
A controversial point of discussion is whether the transport at the interface is mainly diffusion or advection controlled. It is demonstrated that a viscous sublayer exists close to the permeable wall only when the intrinsic permeability of the porous medium is small compared to the viscous length scale of the fluid [1]. If this requirement is fullfilled, a direct proportionality between the strain rate at the interface and the velocity difference between the interfacial velocity and the Darcy velocity inside the porous domain can be derived as an empirical boundary condition for coupling the porous media region and the free flow at the interface [2]. With the increase of the permeability Reynolds number [3] turbulent fluctuations start to penetrate into the porous media region, which is known as turbulent pumping, while simultaneously the porous media topology also affects the turbulent flow in the free flow region [4]. Due to visual accessibility inside of the porous model and the possibility of achieving high porosity, a triply periodic minimal surface topology is used as porous medium. Thus investigating both the penetration of turbulent fluctuations and the dependence of the porous media topology on the turbulent free flow is possible.
In this talk results of highly resolved PIV measurements are presented for such a porous periodic topology adjacent to a turbulent fluid flow, for both inside the porous medium region and at the interfacial region. It will be discussed (a) how the periodicity of the triply periodic minimal surface model influences the interfacial layer of the free flow, (b) how the flow field behaves inside the periodic porous region and (c) which flow phenomena occur at the interface of the porous medium and the free flow, specifically emerging vortices.
We consider the Hele-Shaw model of porous media flows involving two immiscible upper convected Maxwell fluids [1, 2]. Linear stability analysis shows that singularities up to three types can occur including resonance and fracture, the latter one consistent with the experimental results of Mora and Manna [3]. The resonance occurs when one of these two fluids is air and is removed when air is replaced by a Newtonian fluid. The Oldroyd-B case currently in progress will also be discussed. This is joint work with Zhiying Hai.
Foam-assisted water alternate gas injection (FAWAG injection) could be a promising technology to assist gas mobility control, consequently gas management in surface facilities of ultra-deepwater fields in the Brazilian offshore (Vieira et al., 2020). In these fields carbon dioxide (CO2) concentration in the gas stream varies, which can negatively impact foam stability, accelerating its destruction (Abdelaal et al., 2020). However, the impact of CO2 concentration in the gas stream is seldomly evaluated. In this work, we evaluate the impact of CO2 concentration in the gas phase on foam generation and foam strength (apparent viscosity) in porous media.
To this end, we conducted coreflood experiments where gas phase and surfactant solution were co-injected through Indiana limestone under relevant conditions (temperature – 65C, pressure – 10 MPa and superficial velocity – 3.5 x 10-6 m s-1). Four different gas compositions were used for the tests, namely nitrogen (N2), CO2 and two CO2-N2 mixtures (10 mol%, 50 mol%). Two commercial zwitterionic surfactants (CAHS – cocamidopropyl hydroxysultaine and CB – cetyl betaine) were used at 0.5 wt.% concentration (active matter) as foaming agents. The experiments were carried out with the core mounted in vertical position and the injection direction was bottom-to-top.
Our results have shown that increase in CO2 concentration in the gas phase decreased overall apparent viscosity (foam), but no linear correlation has been found. Maximum foam values for CB ranging from 30 mPa s-1 for CO2-foam to 170 mPa s-1 for N2-foam. For CAHS, maximum foam values ranged between 70 mPa s-1 and 170 mPa s 1 for the same conditions. Gas composition also impacted transition foam quality (fg), however, the changes in foam behavior as a function of gas fraction seemed to be dependent of the structure of zwitterionic surfactant. For CB surfactant, increased CO¬2 concentration in gas phase (10 mol% and 50 mol% CO2-N2) shifted foam transition quality to the lower gas fractions (from 0.6 to 0.4), indicating that coalescence was favored under these conditions. The apparent viscosity of the foam formed between CO2 and CB surfactant solution did not present a clear transition, and it remained constant as a function of foam quality. Under the same conditions, fg for CAHS was shifted to the right (0.5 to 0.7), indicating that resistance to coalescence increased as a function of CO2 concentration on the gas phase for this surfactant. Another important observation was that from a gas fraction of 0.9, foam apparent viscosity was independent of gas composition. That meant that N2-, gas mixture- or CO2-foam had the same foam apparent viscosity (CAHS ~ 20 mPa s-1, CB ~ 10 mPa s-1), indicating that this regime was dominated by limiting capillary pressure.
These findings suggest that for FAWAG project gas stream has considerable variation in CO2 concentration, the surfactant chosen not only needs to tolerate salinity and temperature of the reservoir, but also needs to maintain a high fg* under these changing gas composition conditions.
The injection of foams into porous media has gained importance as a method of controlling gas mobility. The multilayer structure of the porous medium raises a question about its efficiency in dealing with layers of different permeabilities. The present work shows the existence of a single traveling wavefront in a two-layer porous medium for a simplified model derived from a realistic two-dimensional one. Besides the necessary conditions for the solution's existence, we prove that the traveling wave velocity is a weighted average of the velocities as if both layers were isolated. All theoretical estimates were validated through one- and two-dimensional simulations. Finally, we estimated the order of magnitude of the characteristic time the traveling wavefront needs to stabilize.
Field measurements of apparent geochemical weathering reaction rates in subsurface fractured porous media are known to deviate from laboratory measurements by multiple orders of magnitude. To date, there is no geologically based explanation for this discrepancy that can be used to predict reaction rates in field systems. Proposed correction factors are typically based on ad hoc characterizations related to geochemical kinetic models. Through a series of high-fidelity reactive transport simulations of mineral dissolution within explicit 3D discrete fracture networks, we are able to link the geo-stuctural attributes with reactive transport observations. We develop a correction factor to linear transition state theory for the prediction of the apparent dissolution rate based on measurable geological properties. The modified rate law shows excellent agreement with numerical simulations, indicating that geological structure could be a primary reason for the discrepancy between laboratory and field observations of apparent dissolution rates in fractured media.
Rock-groundwater interactions may substantially alter the shape and size of voids in the rocks comprising Earth’s upper crust. In carbonate aquifers, these interactions often lead to intense dissolution and the formation of extensive karstic cave systems. Recent studies show that a large portion of the known karst systems was formed by groundwaters ascending from depth (“hypogenic karst”) rather than by $\rm{CO_2}$-loaded meteoric water that infiltrated from the surface (“epigenic karst”). The hypogenic karstic cave systems often make up giant and complex mazes of caves with passages reaching hundreds of kilometers and have significant hydrogeological implications. Despite the importance of the hypogene karstic cave systems, the mechanisms of their formation have remained elusive and ill-constrained [1-2]. To address this issue, we provide herein geological, geochemical, and theoretical evidence that many hypogene karst systems were most likely formed by the interaction of carbonate country rocks with $\rm{CO_2}$-rich geothermal groundwater that rapidly ascended from depth. As the water cools, carbonate solubility increases (due to its retrograde solubility), inducing rock dissolution and cave formation on relatively short geological timescales. A numerical simulation based on this scenario produces maze-like hypogenic karst cave systems very similar to those observed in field studies and constrains the range of feasible hydrological, geological, and geochemical conditions. These conditions are very common in Earth’s crust, suggesting that the scenario proposed herein for the formation of extensive hypogene karstic caves may be ubiquitous worldwide. Finally, we demonstrate the large and relatively rapid impact of these rock-groundwater interactions on the global $\rm{CO_2}$ cycle.
Understanding flow, transport, chemical reactions, and hydro-mechanical processes in fractured geologic materials is key for optimizing a range of subsurface processes including carbon dioxide and hydrogen storage, unconventional energy resource extraction, and geothermal energy recovery. Flow and transport processes in naturally fractured shale rocks have been challenging to characterize due to experimental complexity and the multiscale nature of quantifying exchange between micrometer-scale fractures and nanometer-scale pores. In this study, we use positron emission tomography (PET) to image the transport of a conservative tracer in a naturally fractured Wolfcamp shale core before and after exposure of the core to low pH brine conditions. Image-based experimental observations are interpreted by fitting an analytical transport model to every fracture-containing voxel in the core. Results of this analysis indicate subtle increases in matrix diffusivity and a strong reduction in fracture dispersivity following exposure to low pH conditions. These observations are supported by a multi-component reactive transport model that indicates the capacity for a 10% increase in porosity at the fracture-matrix interface over the duration of the low pH brine injection experiment. This porosity enhancement is the result of exposure of carbonate minerals in the shale matrix to low pH conditions. This workflow represents a new direct approach for quantifying fracture-matrix transport processes and provides a foundation for future work to better understand the role of coupled transport, reaction, and mechanical processes in naturally fractured rocks.
Underground storage of gas (H2, CO2, etc.) and geothermal energy has become a major research area in the ongoing energy transition. In this context, it is important to model and simulate single- and multiphase flows in highly heterogeneous porous media, characterized by very irregularly distributed permeability profiles featuring fractures, channels and macropores.
Flows in these media might not follow Darcy’s law; Forchheimer’s quadratic law is more adequate in the high-Reynolds zones, and applying it globally in the domain is very accurate but costly numerically because of the nonlinearity introduced. Instead, keeping Forchheimer’s law only where necessary should improve computing cost without losing much accuracy. The difficulty with coupling the two laws lies in determining which regions of the porous medium require Forchheimer’s model and which ones can be treated linearly, a question with no clear answer yet.
Two adaptive models have been recently proposed to couple the two laws and answer the above question; given a fixed threshold on the flow velocity’s magnitude, these models locally select the more appropriate law as they are being solved. At the end, each mesh cell is flagged as being in the Darcy or Forchheimer subdomain.
In the first model [1], the interface separating the two subdomains is tracked throughout a fixed-point algorithm. More precisely, the velocity is iteratively re-evaluated and, comparing the velocity’s magnitude to the fixed threshold, the cells are reflagged as being Darcy or Forchheimer cells. Also, a remeshing is performed: if the opposite edges of a cell have velocities with higher and lower magnitudes than the threshold, then the interface is moved at the center of the cell and a new mesh is generated such that the interface coincides with the edge of two newly created neighboring cells.
In the second model [2], the interface is not localized sharply. Instead, a regularized law is introduced resulting from a smooth average of Darcy’s and Forchheimer’s laws; this law gradually passes from Darcy’s to Forchheimer’s, and vice-versa, in so-called transition zones which surround the interface. A classical fixed-point algorithm is then directly run on the regularized law.
In this presentation, we will define these two models in detail, prove their well-posedness using tools from monotone operator theory and variational calculus, and illustrate their behavior via some numerical results obtained on simple, preliminary one- and two-dimensional test cases.
We developed a theoretical and numerical model to study dispersion effects in two-dimensional porous media gravity currents experiencing drainage along their bottom boundary. The need for including dispersion comes from experimental observations of miscible gravity currents experiencing either local or dispersed drainage. In either case, it is found that significant dispersion may arise leading to the appearance of distinct bulk and dispersed phases. For the case of local drainage, we derive an analytical model starting from mass- and buoyancy-balance in both bulk and dispersed phases. The dispersion severity is characterized by quantifying the amount of fluid that appears in the dispersed phase or, equivalently, the spatial separation of leading fronts of the bulk and dispersed phases. Results for gravity currents with local drainage show that the severity of the dispersion depends on flow conditions upstream of the (local) fissure, as well as the fissure dimension and permeability. The extension of our results to the case of distributed drainage shall also be discussed.
The theoretical model is corroborated with reference to complementary COMSOL numerical simulations. COMSOL results are used to specify, in the theoretical model, the value of entrainment parameters that characterize mass transport across the bulk and dispersed phase interfaces. The COMSOL simulations are performed for various source and drainage conditions. Generally, a good agreement between theory and numerics is found.
Finally, the implications of our work to real geological flows in energy sectors i.e. H_2 storage in depleted gas reservoirs are briefly highlighted.
Funding acknowledgment: NSERC
We use experiments and simulations to investigate the mixing dynamics of a convection-driven porous media flow. We consider a fully saturated homogenous and isotropic porous medium, in which the follow is driven by density differences induced by the presence of a solute. In particular, the fluid density is a linear function of the solute concentration. The configuration considered is representative of geological applications in which a solute is transported and dissolves as a result of a density-driven flow, such as in carbon sequestration in saline formations or water contamination processes. The mixing mechanism is made complex by the presence of the rocks (solid objects), which represent obstacles in the flow and make the solute to further spread, due to the continue change of the fluid path. Making predictions on the dynamics of this time-dependent system is crucial to provide reliable estimates of the evolution of subsurface flows and in determining the controlling parameters, e.g., the injection rate of a current of carbon dioxide or the spreading of a pollutant in underground formations. To model this process, we consider here an unstable and time-dependent configuration defined as Rayleigh-Taylor instability, where a heavy fluid (saturated with solute) initially sits on top of a lighter one (without solute). The fluids are fully miscible, and the mixing process is characterised by the interplay of diffusion and advection: initially diffusion controls the flow and is responsible for the initial mixing of solute. At a later stage, the action of gravity promotes the formation of instabilities, and efficient fluid mixing takes place over the entire domain. The competition between buoyancy and diffusion is measured by the Rayleigh-Darcy number (Ra), the value of which controls the entire dynamics of the flow. With the aid of experiments in bead packs (optical measurements) and pore-resolved numerical simulations (immersed-boundary method), we analyse the time-dependent evolution of this system at high Ra, and we quantify the effect of the Rayleigh-Darcy number on solute transport and mixing. The results are analysed at two different flow scales: i) at the Darcy, where the buoyancy-driven plumes control the flow dynamics, and ii) at the pore-scale, where diffusion promotes inter-pore solute mixing. Numerical and experimental measurements are used to design simple physical models to describe the mixing state and the mixing length of the system. The results obtained are compared against previous experimental and numerical works.
The determination of realistic rates of CO2 dissolution associated with geological CO2 storage in deep saline aquifers requires an understanding of the mixing process that takes place during the emplacement of CO2 into these formations. The mixing process is triggered by the local density increase in the ambient brine following the CO2 dissolution. As a result, gravitational instabilities occur, and perpendicular elongated finger-like patterns form that are enhancing the mixing between CO2 and water compared to a purely diffusive process. This density-driven mixing process is important because it accelerates the CO2 dissolution into brine and could eventually form a stable stratification in the aquifer, thereby reducing the chances of leakage.
Owing to the difficulty of imaging the time-dependent convective process, experiments so far have largely focused on two-dimensional systems (e.g., Hele-Shaw cells), which inherently limit the lateral spreading of the downwelling plumes. Here, we present the development of an experimental approach to investigate the evolution of the convective mixing process in three-dimensional porous media using X-ray Computed Tomography. To this end, we have considered consolidated rock samples (two sandstones, two carbonates), for which observations have thus far been lacking.
We characterize the rocks based on the different scales of heterogeneities using different measures such as the representative elementary volume (REV), the coordination number and the pore size distribution.
To imitate the dissolution process of CO2 in brine in the rocks under laboratory conditions, a salt is used with a high X-ray attenuation coefficient that dissolves in water and creates a heavier solution than pure water. We observe that the mixing structures, that arise upon dissolution in the consolidated rock samples, differ among those and are strongly impacted by heterogeneities, especially by macro-heterogeneities such as fractures and vuggy pores.
A key advantage of the three-dimensional X-ray CT images is the possibility to monitor and compare the temporal evolution of individual plume structures between the different rock types.
Further, we compute the temporal evolution of the spatial moments of the vertical concentration distribution, including the cumulative dissolved mass, the location of the centre of mass and the spreading length. We find correlations between the scaling of the moments with the heterogeneities of the pore space. This suggests that apart from characteristics of the advective transport (such as permeability and porosity, included in the Rayleigh number), other micro- and macro-structural features are influencing the overall mixing.
These observations provide therefore more representative information towards the investigation of convective mixing in the context of CCS as well as the selection and evaluation of sequestration sites.
The reduction of atmospheric greenhouse gas concentrations, for which CO2 contributes to 70% of the greenhouse effect, involves securely trapping CO2 in the subsurface. This is done by one of the four main mechanisms, namely structural, residual, dissolution, and mineral trapping [1-3], in the order of their storage security. Dissolution trapping in deep saline aquifers occurs when the supercritical CO2 trapped below the cap rock dissolves into the brine underneath. The CO2-enriched brine has a higher density than the ambient aquifer fluid, which causes it to form a gravitationally-unstable layer between the pure brine and the supercritical CO2. This unstable layer’s destabilization develops into a natural convection that brings the dissolved CO2 to the lower regions of the aquifer while providing fresh brine to the brine-supercritical CO2 interface, in which the latter can further dissolve [4,5].
This convective dissolution of CO2 in a brine saturating a granular porous medium was recently investigated by Brouzet et al. [6] using refractive index matching and planar-laser-induced fluorescence. In their study, the growth dynamics of the instability was significantly different from Darcy-scale theoretical predictions. They explained this discrepancy by the coupling of heterogeneous advection and solute mixing at the pore scale, which cannot be accounted for by Darcy scale models, unless they take local porosity fluctuations into account. These results suggest that Darcy scale models of convective dissolution may underestimate the typical time scale of dissolution trapping by up to several orders of magnitude.
In line with the work of Brouzet et al., we focus here on experimentally chacterizing the Rayleigh-Darcy instability and resulting convection inside a three-dimensional (3D) granular porous medium. That is, we decorrelate the convection from the dissolution, and use analog fluids to study the former alone. The miscible light and heavy analog fluids’ (solutions of Triton X-100, water, and zinc chloride) refractive index is matched to that of the porous medium’s transparent PMMA grains, to render the medium transparent. The density difference between the fluids is achieved by adding a different amount of ZnCl2. The heavier fluid initially carries a uniform colouring dye (Nile blue) concentration. We control the Rayleigh (Ra) number quantifying the initial strength of the instability, and the Darcy number (Da) quantifying the model aquifer’s vertical size by changing the densities of fluids and the size of the grains. A custom-made optical tomography scanner is used to reconstruct the 3D dye concentration field from horizontal cross-sections. The convection dynamics are analyzed from the growth rate of the fingers and the finger number density. Measurements are performed for various values of Ra, andand, independently, for each of them, for various values of the number Ra√Da, which quantifies the typical size of the most unstable instability mode with respect to the typical pore size. The results seem to be consistent with the findings by Brouzet et al.
Residual bubbles in porous media, initially emerging at non-equilibrium state by direct injection, phase changes or imbibition, spontaneously coarsen towards a thermodynamic equilibrium state. During coarsening process, bubbles’ morphology and pore occupancy change that affects hydraulic conductivity, mass & heat transfer coefficients, and chemical reaction kinetics. The kinetics from initial distribution to equilibrium is critical in determining physically-correct models for predicting CO2 subsurface sequestration and gas condensate production.
Based on our earlier theoretical approaches on bubbles’ coarsening [Xu et al., PRL, 2017; Xu&Mehmani et al., GRL, 2019] and on bubbles’ stability analysis [Wang et al., PNAS, 2021], we apply recently-developed pore-network modeling (PNM) tool [Mehmani & Xu, JCP, 2022; Mehmani & Xu, AWR, 2022] to investigate the kinetics of bubble coarsening in porous media, and reveal the final state in both homogeneous and heterogenous media. The time scale of coarsening is also theoretically derived and numerically validated [Yu & Wang, et al., GRL, 2023].
We first study the local equilibrium state of a two-bubble system in two connected pores. Without external field, there are three different equilibrium states when increasing the initial bubble volume: (a)the smaller one is eaten by the larger one, (b)both bubbles survive but of different sizes, and (c)both bubbles survive and of the same size. When there is an external field, only (a) and (b) are found with the growth of initial bubble volume. Analytical solutions matches the simulation well.
We then simulate the bubble coarsening in a 200×200 homogeneous pore-network model. The results show that some bubbles survive and are all finally of the same volume while others disappear. Bubble coarsening kinetics in porous media deviates from Lifshitz-Slyozov-Wagner theory, showing a much slower radius – time scaling. We attribute this new scaling to that porous structure quantizes the space and decouple the mass transfer coefficient from the bubble size. We accordingly develop a new theory for bubble coarsening in porous media, that matches the theory well.
Finally, we investigate bubble coarsening in heterogeneous systems. Although slightly affected by initial condition, we note that survival bubbles at equilibrium statistically fill from the largest pore to smaller pores. We plot capillary pressure – saturation curve and pore-occupancy – saturation curve at equilibrium, that can well match our theory considering pore-size distribution. The time scale for reaching equilibrium can also be estimated by the homogenous media theory, with necessary modification of perfector.
Convective drying of porous media is central to many engineering applications, ranging from spray drying over water management in fuel cells to food drying. To improve these processes, a deep understanding of drying phenomena in porous media is crucial. Therefore, detailed simulation of multiphase flows with phase change is of great importance to investigate the complex processes involved in drying porous media.
In this contribution, we propose a Navier-Stokes Cahn-Hilliard model coupled with balance equations for heat and moisture to simulate the two-phase flow with phase change. The phase distribution of the two fluids air-water is modelled by the Phase Field equation [1].
The focus of this contribution is on the validation and application of the numerical model. While many studies aim to access the phenomena by simulations, here we succeed to compare comprehensively simulations with an experimental methodology based on microfluidic multiphase flow studies in engineered porous media [2]. Comparisons with experiments are rare in literature and usually involve very simple cases. We compare our simulation with convective drying experiments of porous media [3]. Experimentally, the interface propagation was visualized in detail in a structured microfluidic cell made from PDMS. The drying pattern and the drying time in the experiment is very well reproduced by our simulation.
Porous electrodes are performance- and cost-defining components in modern electrochemical systems as they determine the hydraulic resistance, facilitate mass transport, conduct electrons and heat, and provide surfaces for electrochemical reactions [1]. Thus, electrode engineering is an effective approach to improve cost competitiveness by increasing power density. In convection-enhanced technologies, currently used porous electrodes are fibrous substrates developed for low-temperature fuel cells, but their microstructure and surface chemistry limit the performance of emerging electrochemical systems. Microstructure-informed multiphysics simulations can be leveraged to aid the theoretical design of advanced electrode architectures [2]. However, they have only recently been deployed for the bottom-up design of electrode microstructures [3]. The combination of microstructure-informed multiphysics with evolutionary algorithms could accelerate progress in the optimization of porous electrodes for a given application. In this work, we combine three-dimensional simulations with a genetic algorithm for the bottom-up design of porous electrodes for redox flow batteries.
In the first part of the talk, I will describe a methodology to couple an experimentally validated microstructure-informed, electrolyte-agnostic pore network modeling framework [4] with an evolutionary algorithm [5]. This genetic algorithm is used to optimize electrode microstructures by evolving the structure driven by a fitness function that minimizes pumping power requirements and maximizes electrochemical power output, where the optimization only relies on the electrolyte chemistry and initial electrode and flow field geometries as inputs. The analyzed proof-of-concept employs a flow-through cubic lattice structure with fixed pore positions and shows significant improvement of the fitness function over 1000 generations. The fitness improved by 75% driven by a reduction in the pumping requirements by 73% and an enhanced electrochemical performance of 42%. The evolutionary design resulted in a bimodal pore size distribution containing longitudinal electrolyte flow pathways of large pores and an increased surface area at the membrane-electrode interface.
In the second part, I will discuss our latest progress on the genetic algorithm by implementing integrated flow field geometries, commercial fibrous electrodes as offspring networks, and extended evolutionary freedom during the optimization. Coupling the genetic optimization to the flow field geometry affects the fitness evolution, shifting the balance between electrochemical and hydraulic performance, emphasizing the interaction between flow fields and electrodes. By including additional evolutionary freedom (i.e., by allowing merging and splitting of pores outside fixed coordinates), commercial electrodes can be enhanced by reducing their pumping losses. The presented genetic algorithm offers potential for the predictive design of electrode microstructures tailored for specific electrochemical systems. While applied to flow batteries in this study, this methodology can be leveraged to advance electrode microstructures in other electrochemical systems by adapting the relevant physics.
Nuclear Magnetic Resonance (NMR) is a powerful tool to assess physical quantities that characterize porous media, offering detailed information about the fluid molecules confined in the pore space. This work presents a computational implementation of image-based simulations of NMR experiments in porous media using the Random Walk method with a particular focus on reservoir rocks. We explore and discuss the computational challenges of running such simulations on personal hardware instead of using multicore clusters, which is the conventional approach. In that sense, the proposed solution includes a scheme for data compression and a strategy for massive parallelization in the graphics processing unit. Moreover, we present applications simulating NMR diffusometry and relaxometry experiments. In the first study, the time-dependent apparent diffusion coefficient is measured by simulating the Pulsed-Field Gradient NMR technique. This quantity's asymptotic behavior in both short and long-time ranges is then used to recover the surface-to-volume ratio and the tortuosity of the underlying porous medium. We explore the correlation between the recovered parameters and the absolute permeability in a set of synthetic granular media and segmented microtomographic images of sandstones and carbonates. In the second study, we explore the influence of the diffusive relaxation mechanism in the transverse relaxation time, $T_2$. The relevance of this mechanism arises in the presence of strong internal magnetic field gradients induced by a pronounced contrast between the magnetic susceptibility of fluid molecules and mineral components of the solid phase. These simulations require a two-step workflow: in the first step, we compute a spatial description of the magnetic field inside the pore space by solving Maxwell equations under zero-current condition using an image-based finite elements implementation; second, we feed our random walk simulations with the computed field map, incorporating its dynamic effect in the magnetized spins. We perform such simulations in sintered models of glass beads containing localized concentrations of iron oxides and sandstones of varying mineral compositions. Not only will this enhanced relaxation alter the otherwise straightforward interpretation of $T_2$ relaxation times into pore sizes, but it may also indicate the presence of clay components in the mineral phase. In both studies, experimental data is provided for comparison purposes.
Extended research is necessary in view of delivering safe, sustainable and publicly acceptable solutions for the management of radioactive waste across Europe now and in the future. In light of this, a full understanding of the migration behavior of corrosion gases in clay rock environment is of fundamental importance for the reliability of scenarios predicting the long-term safety of geological repositories. The Callovo-Oxfordian clayrock, studied in France as a potential hostrock, presents the complex aspect of having a pore size distribution predominantly mesoscopic (nm), transition scale where different processes occur and interplay. Due to the low permeability of clayrock, the produced gas is expected to accumulate as a distinct gas phase which may attaint important pressure. The pressurized gas phase may desaturate the surrounding clayrock by displacement of pore water along gas flow paths, but also by the diffusion of water vapor throughout the gas phase.
In order to better understand the impact of key transport processes occurring in gas migration in clay material, pore-scale direct numerical simulations taking into account the capillary-dominated two-phase flow, the evaporation and condensation at liquid - gas interfaces, the diffusion of water vapor in the gas phase as well as the specific feature of nanoporous materials (Kelvin effect) are proposed. The work has been carried out using the Smoothed Particle Hydrodynamics (SPH) method, a Lagrangian and meshless method which has emerged as an efficient and reliable tool for simulating complex fluid flows, like those found in porous media at the mesoscopic scale. A novel drying algorithm with Kelvin effect, which drives the local thermodynamic equilibrium between the fluid phase and the gas phase at nanoscale, has been implemented in a two-phase flow and non-deformable solid phase SPH code, initially developed at IRSN.
Different flow conditions will be first investigated for a 2D isolated pore, with and without capillary effect and/or Kelvin effect, and validated against analytical solutions. To highlight key drying behaviors occurring within pores as a function of pore geometry and throat size, we set up a series of standardized simulations over representative geometries of pore doublet. Dynamic capillary effects on pore refilling will be also discussed. Then evaporation-diffusion-condensation model will be used for simulation of drying of 2D heterogeneous pore networks. The impact of the Kelvin effect and of some dynamic capillary effects on the desaturation of the porous material will be investigated. This development should indeed serve as frameworks for upscaling. We will discuss more particularly, and compare to similar studies, the impact of drying and kelvin effect on gas drainage patterns.
Pore-resolved direct numerical simulations (DNS) are performed for turbulent open channel flow over a randomly packed porous sediment bed over a range of permeability Reynolds numbers of $Re_K = {\mathcal O}$(1-10) representative of aquatic systems. A fractional time-stepping based fictitious domain method (Apte et al. 2008) is used to simulate flow over spherical sediment particles on Cartesan grids by enforcing the rigidity and no-slip condition on the particle boundaries. The DNS predictions are compared with the experimental data of Voermans et al. (2017) to show excellent agreement of mean and turbulent flow quantities. A space-time averaging methodology is used to compute the Reynolds stresses, form-induced stresses, and pressure fluctuations. Shear layer and turbulent shear stress as well as Reynolds and form-induced bed-normal stresses increase with $Re_K$. The peak values of the form-induced stresses were found to occur within the top layer of the sediment bed for the Reynolds numbers studied. The sum of turbulent and form-induced pressure fluctuations at the zero-displacement planes are statistically similar and can be well approximated by a $t-$ location-scale distribution fit based on high-order statistics, providing with a model that could potentially be used to impose boundary conditions at the SWI in reach scale simulations. A continuum model based on the volume-averaged Navier-Stokes (VANS) equations is developed by defining smoothly varying porosity across the bed interface and modeling the drag force in the porous bed using a modified Ergun equation with Forchheimer corrections for inertial terms (Wood et al., Annual Review of Fluid Mechanics, 2020). A spatially varying porosity profile generated from the pore-resolved DNS is used in the continuum approach. Mean flow and Reynolds stress statistics and net momentum exchange between the free-stream and the porous bed are compared to show very good agreement. The continuum VANS approach allows for significant reduction in computational costs, thereby allowing to study hyporheic exchange of mass and momentum in large scale aquatic domains with combined influence of bedform and bed roughness.
Funding from US Depart of Energy, Office of Basic Energy Sciences (Geosciences) under award number DE-SC0021626, Pacifict Northwest National Laboratory's internship program, as well as US National Science Foundation award #205324 are gratefully acknowledged.
Understanding the mechanical behavior and fluid flow properties of porous media composed of packed particulate has numerous applications within the physical sciences and engineering, being pertinent to the study of naturally occurring geo-materials, such as sedimentary rocks, and engineered media, such as fuel cells and catalysts. Both manmade and geologic granular porous media often exhibit pronounced spatial variability in their component particle sizes, which in turn, imparts internal heterogeneities in porosity, permeability, capillarity and mechanical strength within the particle column. Numerical methods for the simulation of pore-scale fluid flow or granular mechanics (e.g., Lattice-Boltzmann, finite volume and discrete element methods:) have enjoyed widespread application over the past decade, owing to the proliferation of both commercial software and open-source libraries through which such models can be readily deployed. Coupled with advances in volume imaging of real porous media (i.e., x-ray microcomputed tomography), workers are now able to probe such processes numerically within highly heterogeneous pore networks, providing a wealth of insights into the key physical properties of a wide range of porous materials [e.g., 1-3]. Despite these developments, objective methods for the generation of synthetic porous media characterized by grain-scale heterogeneities in particle size, mirroring those observed in both naturally occurring and manmade porous materials (i.e., layering, grading, lenses, nested pore structures / intragranular and matrix porosity) remain limited. The availability of such a framework is conceptually attractive, as it provides the means to introduce conditional heterogeneity into computational fluid dynamics and discrete particle scale mechanical simulations, providing experimental control over spatial variability in particle size for such studies.
In this work, we present a method capable of simulating granular media which can represent pore-scale heterogeneities commonly observed within a wide variety of natural (i.e., geologic) and manmade porous materials alluded to above. Here, we utilize a modified implementation of the sequential deposition algorithm from the classic material physics literature towards the generation of highly heterogeneous 2D and 3D particle beds (sphere packs), amenable to computational fluid dynamics (CFD) and the initialization of discrete element model (DEM) based mechanical simulations. The presented approach utilizes closed form analytical solutions for the detection of linear and rotational collision between the mobile and static particle pack, meaning that it is computationally efficient and amenable towards the generation of granular porous media containing large numbers of elements. We demonstrate the power of the approach using finite volume CFD simulations of immiscible fluid flow using a range of heterogeneous pore systems generated using our novel framework.
Triple periodic minimal surfaces can be approximated to three-dimensional cell structures, which are found in many forms in nature, such as on butterfly wings or on the skeletal plate of a sea urchin. The structures are representable by a mathematical periodic function. For sheet-based structures, the result is two disjoint, intertwined channels with a uniformly curved surface. The three most common sheet-based structures are D-gyroids, Schwartz diamonds, and Schwartz primitive structures.
The three-dimensional regular periodic structure makes them attractive for various research areas, such as in the medical field for tissue engineering or as a possible heat exchanger, due to their high surface to volume ratio, bionic and mechanical properties.
In this work, the sheet-based gyroid structures with different porosity-levels are topology optimized with respect to their mechanical stability at constant volume using the inhouse micro structure simulation framework "Pace3D". The optimized structures and the original structures are simulated and compared with respect to their mechanics in the linear-elastic range, and other properties such as the surface-to-volume ratio are also investigated.
Simulations of mechanical load in the linear elastic regime are carried out on both the optimized as well as the original structures and the mechanical properties are compared. Furthermore, micro structure characteristics such as the surface-to-volume ratios are evaluated.
In this work, we use a combination of formal upscaling and data-driven machine learning for explicitly closing a nonlinear transport and reaction process in a multiscale tissue. The classical effectiveness factor model is used to formulate the macroscale reaction kinetics. We train a multilayer perceptron network using training data generated by direct numerical simulations over thousands of microscale examples. Once trained, the network is applied in an algorithm for numerically solving the upscaled (coarse-grained) differential equation describing mass transport and reaction in two example tissues. The network is described as being explicit in the sense that the network is trained using macroscale concentrations and gradients of concentration as components of the feature space.
Network training and solutions to the macroscale transport equations were computed for two different tissues. The two tissue types (brain and liver) exhibit markedly different geometry and spatial scale (cell size and sample size). The upscaled solutions for the average concentration are compared with numerical solutions derived from the microscale concentration fields by a posteriori averaging.
There are two outcomes of this work of particular note: 1) we find that that the trained network exhibits good generalizability, and it is able to predict the effectiveness factor with high fidelity for realistically-structured tissues despite the significantly different scale and geometry of the two example tissue types; and 2) the approach results in an upscaled PDE with an effectiveness factor that is predicted (implicitly) via the trained neural network. This latter result emphasizes our purposeful connection between conventional averaging methods with the use of machine learning for closure; this contrasts with some machine learning methods for upscaling where the exact form of the macroscale equation remains unknown.
Precise 3D demonstration of heterogeneous porous materials while critical is still a challenge. The advantage of having such models includes for example more accurate characterization and estimation of transport properties. Realistic 3D representations can be achieved using several high-resolution 2D samples. We applied a deep learning algorithm to utilize 2D images and reconstruct 3D models of complex materials such as lithium-ion battery electrodes. The deep learning algorithm was trained using 2D images for generating 3D samples. The results of testing the trained network with new samples show the capability of the algorithm for reproducing important structural properties. The reconstructed samples also reproduce the results for flow and heat properties in an acceptable range.
Accurate prediction of solid mineral dissolution during reactive flow in porous media is vital for a wide range of subsurface applications (e.g., CO2 sequestration [1] and geothermal systems [2]). Detailed numerical modelling of mineral dissolution at the pore-scale is generally expensive [3] and that limits our ability to perform comprehensive uncertainty quantification studies to explore the various sources of uncertainties and its impact on the dissolution process.
In this work, we develop efficient deep learning emulators for geochemical reactions. We build on an earlier work on reduced order modelling (ROM) using Deep residual recurrent neural network [4] to develop highly predictive ROM using limited training data. We utilize a U-net architecture to perform approximate explicit time stepping for the dynamical system. The input features for the deep learning model are the discrete components of the physical residual which are known to correlate well with the solution updates over the training samples as well as the unseen validation dataset. This correlation is governed by the physical equations controlling the evolution of the system. The second component of the DL emulator is a hierarchical architecture of neural networks, where a stack of U-Nets is used at every timestep to mimic fixed point iterations in numerical schemes. In order to stabilize the training, the algorithm starts with a single step update, after which, the first level U-Nets are frozen, and the next level U-Net is trained and so on.
The developed algorithm is demonstrated on a dataset with different Peclet and Kinetic numbers. The pore-scale dissolution training and validation datasets are generated using detailed numerical simulations using the improved Volume-of-Solid method in GeoChemFoam [3] and are available as an open access repository for a range of dissolution regimes [5].
Simulating fluid flow in reservoir models is an expensive and time-consuming task. Given the inherent uncertainty in most measurements used as inputs for these models, it is customary to perform stochastic modeling in order to reduce and quantify the uncertainty. Recently, a new class of machine learning algorithms referred to as operator learning has been developed. These algorithms, such as DeepONets and Fourier Neural Operators, can learn mappings between two infinite-dimensional spaces. However, these approaches suffer from data inefficiency as they require thousands of training observation pairs in the input and output domains which is computationally prohibitive. Physics-informed DeepONet has been proposed as a remedy to this problem. In this paradigm, DeepONets are regularized by underlying physical laws in a manner similar to Physics Informed Neural Networks (PINNs), hence the name. Physics-informed DeepONets can learn the solution operator mapping between a set of initial and boundary conditions to the full spatio-temporal solution making it a powerful tool for parametric PDE learning. Here, we investigate the applicability of Physics-informed DeepONets to an immiscible two-phase fluid flow problem through a 1D porous medium. We provide two test cases. First, we attempt to learn the solution operator mapping of an initial condition to the entire spatio-temporal solution of all possible initial conditions. Second, we test Physics-informed DeepONets to learn the solution operator mapping of a boundary condition to the entire spatio-temporal solution of all possible boundary conditions. Our results show that with a small sacrifice in accuracy, enormous gains in speed can be achieved, as this approach can solve thousands of PDEs in a fraction of a second.
Understanding the reservoir behavior is vital knowledge required for various aspects of the reservoir management cycle such as production optimization and establishment of the field development strategy. Reservoir simulation is the most accurate tool for production forecast, but often it is very expensive from the aspects of computational time and investment in the model building process. In this work, the machine learning methods for accurate production forecast that honor the material balance constraints are presented.
The presented approach uses two machine learning methods and one semi-analytical approach namely Capacitance Resistance Model (CRM). The first machine learning method is the powerful Generalized Additive Models (GAM) approach, which uses splines as basis functions for the representation of the solution. The advantages of splines are the smoothness of the underlying functions with continuous derivatives, and the easy way of constraining splines to monotonic and convex shapes. Another advantage of GAM is its explainability capabilities, which are inhered from the Generalized Linear Models. The second machine learning approach is a combination of Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN), which are proven to be a good choice for time series predictions. The common extension of the two methods is the material balance constraints in the form of a CRM model, where rates are the corresponding machine learning solutions. Such constraints are necessary during the training process to avoid unphysical solutions and to honor conservation laws. The constrained GAM approach belongs to a broad category of Physics Informed Machine Learning (PIML) methods, while LSTM-CNN with constraints is part of Physics Informed Neural Networks (PINN).
The implemented approach was applied to the publicly available data with an existing history-matched reservoir model for the offshore field with several injectors and producers. This allowed us to thoroughly analyze the results of the study for communication between wells. Splines used in the GAM model have the option to be a function of one or multiple features, while Neural Networks naturally define communication between features through hidden layers and weights. Such flexibility allows taking into account inter-well connectivity, using inter-well distances, production and injection rates, and average reservoir properties, which are analogs of transmissibilities in a simulation model. The average properties are obtained through the construction of Voronoi grids around wells.
Machine learning is improving at solving difficult problems, while it often suffers from nonphysical solutions and unexplainable models. The presented machine learning methods hold the properties of explainable statistical regression models, in the case of GAM, and highly tunable time series predictors, in the case of LSTM-CNN. Both methods provide powerful predictability capabilities within material balance constraints. By no means does it try to replace the reservoir simulation but offers a complementary solution, which is reliable and necessary in cases where there is no full reservoir model available.
Simulation of CO$_2$ utilization and storage (CCUS) in subsurface reservoirs with complex heterogeneous structures requires a model that captures multiphase compositional flow and transport. Accurate simulation of these processes necessitates the use of stable numerical methods that are based on an implicit treatment of the flux term in the conservation equation. Due to the complicated thermodynamic phase behavior, including the appearance and disappearance of multiple phases, the discrete approximation of the governing equations is highly nonlinear. Consequently, robust and efficient techniques are needed to solve the resulting nonlinear system of algebraic equations. Machine learning (ML) techniques have recently been applied to a wide range of nonlinear computational problems. Recently, Physics informed neural network (PINNs) has been proposed for solving partial differential equations. Unlike typical ML algorithms that require a large dataset for training, PINNs can train the network with unlabelled data. The applicability of this method has been explored for the flow and transport of multiphase in porous media. However, for strongly nonlinear hyperbolic transport equations, the solution degrades significantly. In this work, we propose a sequential training PINNs to simulate two-phase transport in porous media. The main concept is to retrain neural network to solve the PDE over successive time segments rather than train for the entire time domain at once. We observe that sequential training can capture the solution more accurately concerning the standard training method. Furthermore, we extend the sequential training approach for compositional problems in which nonlinearity is more significant due to the complex phase transition.
Hydrogeological properties are very important to enhance the modeling of physical and chemical processes related to various geoscience and environmental applications such as geologic carbon storage, subsurface energy recovery, and environmental fate and transport. One critical component of subsurface characterization for prediction of flow and reactive transport is how accurately we can estimate heterogenous permeability (and porosity) fields. In this work, we will compare physics-informed machine learning methods such as physics-informed neural network (PINN) and Bayesian PINN to estimate heterogenous permeability fields with spatial and temporal observation data of tracer concentrations in 3D sandbox experiments. Emphasis will be placed on comprehensive state-of-the-art datasets obtained using magnetic imaging resolution approach that provide non-reactive tracer transport over time in well controlled laboratory sandbox experiments. This work will provide outstanding benchmark datasets that can be used for validation of machine/deep learning approaches.
SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525.
Per- and poly-fluoroalkyl substances (PFAS) are emerging contaminants of great importance, because of their spreading in subsurface, gradual bioaccumulation and toxicity. Assessing exposure risk, developing management strategies, and implementing remediation scenarios require an accurate understanding of the fate of PFAS in subsurface. In the present work, PFAS transport in saturated and unsaturated soil columns has been studied under varying initial concentration for two type of PFAS: perfluorooctanoic acid (PFOA) and perfluorodecanoic acid (PFDA). The PFAS surface tension and PFAS / n-dodecane (n-C12) interfacial tension were measured as functions of PFAS concentration and salinity with static (DuNouy ring) method, and fitted to Langmuir-Szyszkowski equation (Fig.1a,b). The capacity of PFAS solutions to emulsify non-aqueous phase liquids (NAPLs), commonly trapped in the saturated zone, was investigated by mixing PFAS solutions with n-C12 at various volume ratios with the aid of an ultrasound probe, inspecting their stability optically, and measuring transient changes of the shear viscosity. Moreover, the effects of PFAS on wetting properties were analysed by measuring the contact angles of PFAS drops surrounded by either air or n-C12 on glass surfaces (Fig.1c,d). A dried sandpack was evacuated and saturated with NaCl solution with free imbibition. Unsaturated and NAPL-polluted conditions were created by injecting air or NAPL at constant flow rate, reinjecting NaCl solution and monitoring the axial distribution of water saturation with a multiple-electrode resistivity meter [1]. PFAS flow tests were then conducted, and the concentration of PFAS was measured in aqueous samples collected at the outlet port with the methylene blue active substances (MBAS) method and UV-Vis spectrophotometry [2]. The spreading of PFAS in the soil column was simulated with a 3D field scale Computational Fluid Dynamic model which simulates the PFAS transport by solving the Navier Stokes equation inside infinite domain (field scale), where convection, dispersion and adsorption (on solid interfaces and air/water interfaces) terms were included. The simulations were developed on Comsol Multiphysics platform [3] and the numerically predicted PFAS concentration breakthrough curves under saturated and unsaturated conditions were compared with corresponding datasets of PFAS flow tests in soil column.
Acknowledgements
This work was performed under Grant Agreement 101037509 — SCENARIOS — H2020-LC-GD-2020 / H2020-LC-GD-2020-3 (project title: “Strategies for health protection, pollution Control and Elimination of Next generAtion RefractIve Organic chemicals from the Soil, vadose zone and water” - acronym “SCENARIOS”) supported by the European Commission.
Iodinated Contrast Media (ICM) are organic compounds, widely used during X-ray procedures for medical imaging [1]. ICM are connected with diseases, such as hyperthyroidism or hypothyroidism, due to the iodine ions bound to the chemical molecule [2]. Moreover, in aquifer system, ICM can form toxic intermediate products during photodegradation or Managed Aquifer Recharge processes [1].
Nano-scale zero-valent iron (nZVI) can efficiently dehalogenate ICM and turn ICM into non-toxic products [1]. The chemical reaction between ICM and nZVI is impacted by several factors, such as the initial concentration of nZVI, the presence of oxygen in the subsurface environment (i.e., the occurrence of anaerobic versus aerobic conditions) and the pH of fluid phase. Although several experimental studies have analyzed the reaction kinetics between ICM and nZVI at laboratory scale under batch conditions ([3],[4],[5]), few studies have investigated the interaction between ICM and nZVI under flow conditions. In this framework, Zhou et al. [6] performed both batch and column experiments in order to identify the reaction mechanism and kinetics of diatrozate (DTA) dehalogenation using sulfide-modified nZVI (S-nZVI) under anaerobic conditions. The authors also proposed a pseudo-first-order kinetic model to interpret the batch experiment outcomes. However, the proposed model cannot adequately reproduce the experimental results obtained under flow condition.
Here, we cast the batch experiment of Zhou et al. [6] within a stochastic framework and (i) provide Maximum Likelihood estimates and associated uncertainties of characteristic parameters driving the underlying kinetic mechanisms and (ii) assess the way uncertainty associated with model parameters propagates into uncertainty in quantifying the temporal evolution of DTA concentration. Finally, we propose a new kinetic model able to describe the interaction between DTA and S-nZVI under flow conditions. The new kinetic model which includes advective-dispersive transport, sorption and desorption to and from the reactive surface (S-nZVI), dehalogenation of DTA by S-nZVI, adequately reproduces the experimental results.
Chlorinated solvents, such as trichloroethylene (TCE), have caused groundwater and soil contamination for years due to their massive and uncontrolled use adopted in the past [1]. Once released in the subsoil, these compounds are characterized by high mobility and low biodegradability with a consequent persistence in the environment [2]. For these reasons, in most industrialized countries, groundwater bodies are currently characterized by diffuse contamination by chlorinated compounds, which can cause potential long-term risks to human health [2-4]. In particular, the most critical migration pathway for chlorinated solvents is the volatilization from the subsoil into overlying buildings (i.e. vapor intrusion) [5-6]. Traditional remediation techniques in sites characterized by diffuse contamination by chlorinated solvents are not technically and economically sustainable as they typically require significant amounts of reagents or energy [4]. In this scenario, it is thus more indicated to adopt risk management strategies aimed at interrupting the migration pathway of chlorinated solvents vapors to air ambient or into buildings [3-4]. Recently, it was proposed to use horizontal permeable reactive barriers (HPRBS) placed in the unsaturated zone to treat upward volatile organic compounds vapors [3-5, 7-8]. Zero-valent iron (ZVI) was proposed as reactive material for HPRBs and tested for TCE degradation in the vapor phase through reductive dehalogenation [3-4]. In the last years, ZVI bimetals have also been widely studied for the enhancement of the degradation of chlorinated compounds via iron corrosion or hydrogenation in contaminated groundwater [9-11]. However, such bimetals were poorly investigated to treat chlorinated solvents in the vapor phase [12-13]. In this study, we examine the feasibility of using zero-valent Fe-Cu ad Fe-Ni bimetals for the degradation of TCE vapors at partially saturated conditions. Different bimetals were synthesized by mixing Fe and Ni or Cu powders using disc milling and then characterized. The produced bimetals were then tested in anaerobic batch TCE vapors degradation tests at different reaction times to evaluate their reactivity towards dechlorination. The disc-milled bimetals produced were characterized by a homogenous distribution of Ni or Cu in the Fe phase and micrometric size. In all the experiments, complete degradation of TCE vapors was achieved in maximum 4 days with zero-order degradation kinetics. Fe-Ni bimetals have shown better performances in terms of TCE removal than Fe-Cu bimetals leading to a complete degradation of TCE in the vapor phase after 2 days of reaction. These results showed a significant enhancement in TCE removal compared to ZVI alone, which was found to entirely degrade TCE vapors after minimum 2 weeks of reaction in previous studies [3-4]. The only detectable reaction byproducts in the tested conditions were C3–C6 hydrocarbons. No vinyl chloride (VC) or dichloroethylene (DCE) peaks were observed. In view of using the tested bimetals in HPRBs to treat chlorinated solvent vapors emitted from contaminated groundwater, the experimental results obtained were integrated into an analytical model to simulate the reactive transport of vapors through the barrier. It was found that an HPRB of 20 cm could ensure a complete reduction of TCE vapors.
Foam has been studied and applied for enhanced oil recovery (EOR) for many decades. There is a large body of research on this topic (Kovscek and Radke, 1994; Rossen, 1996), from pore-level mechanisms of creation and destruction of bubbles and foam mobility to modelling foam processes on the laboratory and field scale. Foam is also increasingly receiving attention as a means of improving soil remediation (Bertin et al., 2017). This raises the question: what findings and modeling approaches that apply to foam for EOR apply to soil remediation, and which require major modification?
In EOR applications in the relatively deep subsurface, foam stability is controlled by capillary pressure. Bubbles are as large as pores, because of inter-bubble diffusion. As a result, foam exists in two flow regimes depending on flowing gas fraction (Alvarez et al., 2001). This is key to modelling foam for EOR.
In soil remediation, as in EOR, foam’s primary purpose is to redirect the flow in the formation. In soil remediation, permeability is much greater, which means capillary pressure is less than in EOR. Foam bubbles are not trapped as easily as in EOR foam. Experiments show bubbles smaller than pores, and wet conditions in aquifers, make the effect of diffusion uncertain. As a result, whether the two flow regimes found for EOR foam apply to soil remediation is moot. The goal of EOR is to make a measurable increase in oil recovery; in soil remediation 100% recovery of toxic waste (NAPL) is the goal. Moreover, the remediation fluids must be recovered, not left in the formation. Foam for soil remediation is usually pregenerated before injection, but injection pressure is severely limited. As a result, aquifer flow and gravity play a much larger role in soil remediation than EOR.
Many aspects of foam apply to both applications. Gas mobility is greatly reduced in both cases, and gas trapping is significant in both. Capillary forces are critical at the pore scale, though they are likely to be less dominant at the higher permeabilities in soil remediation. The basic mechanisms of bubble creation are the same. Foam generation in gas flow across layer boundaries was found to be critical to the success of foam in one application to aquifer remediation (Hirasaki et al., 1997). The presentation will discuss how the physico-chemical processes described in EOR apply to environmental application, and if new phenomenon need to be considered specifically for soil remediation.
Groundwater remediation is a pressing issue in the modern world. Some regions of the UK, such as Southeast England, take more than 75% of their public water supply mainly from Chalk aquifer (Groundwater resources in the UK, 2022). In Brazil, almost 37% of the cities are supplied exclusively with groundwater. Study by Lunardi et al., (2021) highlighted high susceptibility to groundwater pollution of regions, where pollutant source, such as industry, is present. The studies performed on methods similar to ones studied in this work usually do not study what is happening on pore-scale. Such study is performed by Pandey, Sharma and Saha (2022) on nZVI nanoparticle production techniques, or by Chen et al., (2021) on slow-release potassium permanganate. This highlights a knowledge gap in the modern understanding of these remediation techniques.
In this work, a dataset on nZVI nanoparticle reaction with TCE (trichloroethane) is studied. TCE is a DNAPL – Dense non-aqueous phase liquid. These compounds are challenging to be removed from groundwater reservoirs via conventional means, as they are almost immiscible in water, and are difficult to remove from the porous medium. Therefore, nanoparticles used for remediation of such reservoirs have to be able to reach the contaminant. The dataset was obtained via X-ray microtomographic scanning (X-ray micro-CT) and allows for 4D (3D + time) study of the processes, happening on the pore-scale. The dataset was captured at Diamond Light source by Dr. Tannaz Pak. This study is performed via specialised software, such as Fiji (ImageJ), Avizo and MatLab.
In addition to this, a new setup has been developed for column experiments. This setup gave us the possibility to investigate liquid and particle dynamics on a larger scale, across a column of approximately 36 cm long 3.5 cm in diameter. With this setup we were able to measure particle distribution through the column after several nZVI injections on different porosities. Particle distribution was assessed via magnetic susceptibility sensor. Each material loaded into the column is initially characterised by a breakthrough curve, evaluated via conductivity of NaCl brine. Subsequent breakthrough curves for nanoparticle injections are evaluated via magnetic susceptibility sensor. The experimental data of the breakthrough curves is then analysed with MnMs, to confirm the results of the experiment and get a better insight into nanoparticle mobility.
In addition to that, the experiment on porosity influence on the efficiency of nanoparticle remediation was performed. This experiment involved saturating porous medium with nitrate-contaminated water and injecting nanoparticle suspension into the column. Parallel to the column experiment, a reference batch test was performed, with the same concentrations of nitrates and nZVI, but in absence of any porous medium. The results of this experiment were then compared with each other, to assess the influence of presence of porous medium and porosity on decontamination efficiency.
Reactive transport is a multi-scale and multi-disciplinary process used to study various environmental and subsurface applications including geothermal utilization, carbon dioxide storage, well acidizing, and contaminant remediation. Much research has been conducted to simulate reactive transport using the Eulerian approach, Lagrangian particle tracking models, and various pore-scale models. To simulate the process on larger scales, continuum models have been used due to their fewer resources demanding nature compared to other approaches.
Eulerian (continuum) models are usually expressed with a PDE describing the transport and the interconnected reaction in the form of the Advection-dispersion-reaction equation (ADRE). In cases with homogeneous reactions, which is the purpose of this study, using the reaction constant derived from the well-mixed batch reaction results in the over-prediction of the product formation in continuum models. The reactants segregation, incomplete mixing, non-Fickian transport, and the fact that ADRE is limited in considering the effect of the local fluctuations and heterogeneity of the transport and the reaction at the pore level, have been introduced as the underlying reasons for the differences.
This discrepancy has been tackled by considering a time-dependent effective rate coefficient, a smaller dispersion coefficient different from the conservative experiment, as well as non-Fickian diffusion in time or space, using an effective constant reaction rate, and considering beta distribution for the mixing ratios within the representative elementary volume (REV). Continuum models usually need more than one calibration parameter to match the experimental results which may lead to inconsistencies if the initial or boundary conditions change which undermines the generality of these models. Instead, using pore-scale models that are capable of replicating realistic variations of velocity and reaction at the pore level is of great importance. In this manner, direct numerical simulations such as Lattice Boltzmann Method (LBM) or Pore Network Modeling (PNM) are viable tools to carry out the pore-scale simulations.
The upscaled reaction rate extracted from the pore-scale simulation by volume averaging, which reflects the effect of structural heterogeneity, and preferential flow pathways, can be utilized as an input to continuum models to amplify the reliability of the Darcy-scale, and field-scale predictions. In such a way, one can include the pore-scale chemical reactions and fluctuations in the large-scale transport in subsurface heterogeneous porous structures.
This study leverages pore network modeling for pore-scale simulations due to its fast and accurate enough computation of reactive transport phenomenon, and much less computationally demanding nature compared to direct methods. The pore-scale model incorporates incomplete mixing at the pore level by the relationship between the effective reaction constant (Keff) and the Peclet number, validated by experimental results. PNM simulations were executed on a network with a size of a REV. Upscaled reaction rates were derived by performing volume averaging on the pore-scale reaction rates and were used in the continuum model to predict the experimental results. Furthermore, the capability of this upscaling method in predicting the reactive process in a 2D heterogeneous porous medium was inspected.
Iodinated X-ray contrast media (ICM) agents are a class of pharmaceuticals and personal care products (PPCPs) of growing concern in environmental science as emerging contaminants [1, 2]. ICMs are the primary source of adsorbable organic iodine responsible for the formation of iodinated disinfection byproducts (I-DBPs), highly toxic compounds commonly found in surface waters [3]; as such, prevention and remediation strategies are necessary for the removal of these contaminants from natural waters [4, 5]. Nonetheless, there currently does not exist a scalable, environmentally safe, and efficient strategy for their removal from natural waters. Adsorption systems are routinely used in water treatment and purification from contaminants, and have the major advantage of preventing the formation and release of toxic byproducts [6]. In addition, adsorption of ICM agents to potential absorbents is an approach for purification of water resources from this emerging contaminant. To this end, studies investigating the sorption dynamics of these compounds to candidate sorbent materials, in different experimental conditions, are needed.
In the present study, the adsorption capacity of the non-ionic ICM iopamidol and the ionic ICM diatrizoate onto candidate sorbent materials has been investigated by means of single sorption batch tests, under both equilibrium and kinetic conditions. Column tests have also been performed to highlight the influence of contact time and dynamic flow conditions on the sorption process, with experimental conditions selected based on their relevance for field applications.
In batch equilibrium tests, neither ICM agents showed sorption to soils, thus highlighting their peculiar nature as non-lipophilic contaminants remaining in the water phase [7]. A novel iron oxide-based colloidal suspension was also incapable of removing the target compounds, thus highlighting the need for sorbent materials with specific characteristics. A pelletized activated carbon sorbent material was successful in removing the target ICM agents in different experimental conditions. Effect of a variety of influencing factors such as amount of sorbent and sorbates, contact time, and hydrodynamic condition on the removal rate of the ICM agents were analysed. The results, gathered with the aim of understanding the key principles and dynamics behind ICM agents adsorption, offer a new set of data that inform one how to design tailored removal strategies and industrial treatment processes.
We present a theoretical asymptotic solution for high-speed transient flow through micro-porous media in this work by considering the inertia effect in the high-pressure-difference pulse decay process. It includes all three gas related effects, that are the inertia effect, the slippage effect, and the compressibility effect. Capillaric model, in which a bundle of straight circular tubes whose radius is much smaller than length is used to represent the internal structure, is adopted and the flow is described by the unsteady-state incompressible Navier-Stokes equation with mean density in two-dimensional case, capturing the main characteristic of mass flow rate. By order of magnitude analysis and asymptotic perturbation, our inertial solution along with its dimensionless criterion for high-pressure-difference pulse are derived. The theoretical results are verified using our self-built experimental platform, by comparing the permeabilities calculated by our inertial solution and the benchmark steady-state measurement. Our inertial solution can shorten the measurement time and is expected to be used in measurement of extremely low-permeability samples.
Exchange processes at the interface between a porous medium and a turbulent flow field are relevant in a wide range of natural and industrial systems: Prominent examples for technical applications range from food drying up to processes within fuel cells. In the environment, the exchange of mass within the hyporheic zone is vital for the health of aquatic ecosystems, whereas the evaporation from soils must be considered for sustainable land use. Despite the apparent heterogeneity of these fields, scalar transport across the interface is driven by a common set of mechanisms, which can be distinguished in the double-averaging framework (e.g. [1,2,3]): Whereas transport due to molecular diffusion requires gradients in the scalar concentration field, turbulent scalar transport results from correlated fluctuations of the flow and scalar field in time. In contrast, dispersive scalar transport is caused by correlated fluctuations of the mean flow and scalar field in space.
The objective of our research is to contribute to a more comprehensive mechanistic understanding of scalar transport in the interface region. We (i) identify the regions of influence of individual scalar transport processes and (ii) analyze the interaction between the fundamentally different processes within the double-averaging framework.
For the numerical investigation with our in-house code MGLET [4,5], the porous medium is represented by a static random pack of spheres with uniform diameter. While the advection-diffusion equation is solved for a passive scalar with a Schmidt number of $Sc=1$, the flow field is obtained from solving the incompressible Navier-Stokes equations. By means of a single-domain Direct Numerical Simulation (DNS), all temporal and spatial scales are resolved both in the free flow region and in the pore space of the porous medium, which avoids any model assumptions. For a representative case, the flow field is validated against experimental data [6]. In total, we consider eight different simulation cases with shear Reynolds numbers in the range of $Re_\tau = 150 – 500$ and permeability Reynolds numbers of $Re_K = 0.4 – 2.5$.
Instantaneous fields (please, find figure attached) provide an intuitive impression of the processes, which is supported and quantified by the double-averaged statistics of the results: Whereas turbulent scalar transport dominates in the free flow region, dispersive scalar transport takes the leading role in the topmost sediment layers below the interface, before molecular diffusion becomes most relevant in deeper regions. The results confirm that the relative importance of different processes is determined by $Re_K$ [7]. Observing that turbulent and dispersive scalar transport hardly co-exist in any regions, we evaluate budget equations for temporal and spatial fluctuations in the scalar concentration field to explain the interaction between the two processes.
The obtained insight is meaningful for the development of hyporheic scalar transport models, as the described interaction prohibits addressing the problem as a mere superposition of two transport processes. Also, ecological implications can be derived, as the presence or absence of strong spatial mean concentration gradients influences the biocenosis in a habitat.
Transport of species in porous media plays a crucial role in a variety of fields, including environmental engineering, geochemistry, and biology. Understanding the movement of species such as contaminants, nutrients, and microorganisms through porous media is essential for predicting and mitigating the impacts of human activities on the environment, as well as for developing effective remediation strategies [1]. For example, in the field of environmental engineering, knowledge of species transport in porous materials is essential for designing effective remediation strategies for contaminated soil and groundwater. In other applications such as water treatment, understanding the transport of nutrients and contaminants in porous media can help in the design of filtration systems and the optimization of treatment processes [2]. Additionally, research on transport of $CO_{2}$ in porous media is crucial for developing and implementing effective strategies for carbon sequestration in aquifers, which is a key component of efforts to mitigate climate change [3]. In general terms, the study of transport of species in porous media has the potential to significantly improve the effectiveness and efficiency of a wide range of environmental and engineering technologies.
In this study, we investigate the behavior of $A+B \to C$ reaction-diffusion chemical fronts in a finite radial geometry where the chemical species $A$ and $B$ are initially separated in space. In addition to the time properties found for rectilinear ([4]-[6]), and radial ([7]) geometries, we describe the dynamics of the spatial position of the reaction front ($AB$) which strongly depends on the initial parameters such as the ratio of initial concentrations ($\gamma = B_{0}/A_{0}$), ratio of diffusion coefficients ($\delta=D_{B}/D_{A}$), and the size of the geometry ($R$). We performed simulations and numerical analysis to predict the dynamics of the front and compared our results to experimental observations performed in gel and liquid systems.
Unlike previous studies done in infinitely extended domains, our results show that the reaction front could remain stationary at any spatial position depending only on the initial conditions. With the aforementioned numerical analysis, we are able to predict the short, medium, and long-time dynamics of the reaction front.
Our findings provide new insights into the behavior of $A+B \to C$ chemical fronts in finite radial geometry. By better understanding these dynamics, we can improve our ability to control and manipulate chemical reactions in more complex settings.
Darcy’s Law is a classic example of a scientific paradigm in the study of fluid migration through porous media. However, many authors have reported deviation of flow regimes from Darcy’s law at low flow velocity [1] when analyzing fluid flow in rocks, often called pre-Darcy flow. In addition to the velocity dependence of rock permeability, the heterogeneous nature of geomaterials deserves equal attention when analyzing the hydraulic behavior of rock masses. When performing conventional permeability tests, little is known regarding fluid behavior within the sample. In this context, neutron imaging provides an ideal full-field technique for better understanding these phenomena. This work employs neutron imaging to investigate the influence of sample heterogeneity and flow rate on flow paths and permeability by performing flow tests with Idaho Gray sandstone cores.
In-situ experiments were performed at the Neutra instrument at Paul Scherrer Institut (Switzerland) using the setup described in Vieira Lima et al. [2]. The flow tests were carried out on samples saturated with heavy water (D2O) or normal water (H2O) by pressure-driven percolation of the respective opposite fluid. Before the neutron experiments, x-ray tomograms were acquired to provide a reconstructed 3D image with a cubic voxel width of 13 µm for a detailed characterization of the pore and grain structures. During each test, neutron radiographies were acquired with an exposure time of 1 s, generating 2D images with a 200 µm pixel size. The difference in neutron attenuation between D2O and H2O was exploited to track the advance of the infiltration front; as D2O and H2O are otherwise very similar fluids, a near single-phase fluid condition was assumed. Boundary fluid flow-rate and pressure measurements were recorded and correlated with the images. The raw neutron images were processed using in-house python codes, generating maps of saturation-time and -degree plus flow-speed fields. 3D porosity fields and pore network models (using Porespy software [3]) were generated from the x-ray tomograms. Simulations of the visualized phenomena were performed using the “Invasion Percolation” and the “Stokes Flow” algorithms from the OpenPNM package [4] based on the 3D pore networks.
Pre-Darcy flow was observed from the boundary measurments in all samples (Fig.1) with a reduction in the bulk permeability with the flow rate. The results from the neutron radiography in the form of 2D maps of the flow-field evolution showed that the percolation followed preferential paths due to the natural heterogeneity of the samples (Fig.2), which could be correlated with the heterogeneity in the porosity fields (Fig.3). In the injection of either of the percolating fluids (H2O and D2O), the flow rate increase generated a flattening of the advancing flow fronts and reduced spatial heterogeneity of the infiltration, indicating that more pores were accessed at higher injection rates and suggesting threshold rates/pressures exist to access different parts of the pore network. The simulations of the fluid infiltration using the pore network models reproduced well the observed flow patterns and showed a good performance in predicting the change in bulk permeability for each flow rate.
The understanding of water transfer in heterogeneous porous media such as soils is at the center of many issues such as water resource management. In the macroporosity as opened cracks, earthworm burrows free surface flow can be a dominant (Sammartino et al., 2012). Very little is known about the physical processes involved in infiltration, whether it is the form of streamflow, continuous films in the macropore (Keven and Germann, 2013) or the "active" macroporosity during a flow and also the exchange mechanisms at the macropore interface (Katuwa et al., 2015). Answering these questions on a small scale is crucial because, for example, the filling rate of a macropore is closely related to colloid filtering, water retention phenomena. The last decades many models have been developed to model flow in macroporosity such as dual porosity with Kinetik Dispersive Waves models (Di Pietro et al., 2013) or conceptual approaches to film flow as in Nimmo et al. (2010). However, these models still do not explain all of the observed flows made by imaging techniques (Sammartino et al., 2015 ; Lissy, 2019).
In this paper, we focus on the modelling and simulating free surface flow in a cylindrical microtube taking into account the physicochemical properties of the matrix and at the interface between the matrix and the macropore. Indeed, organic matter is known to generally impart hydrophobic properties to soils. In an impervious microtube surface, a rich range of flow shapes has been identified: droplets, thin films or rivulets and notably, there is a regime leading to complete wetting (Beltrame, 2018). In the present work, the mesopore surface is porous and fluid transfer may appear through the interface between the macropore and the soil matrix.
The model is based on the long-wave approximation with a free surface. The soil matrix wettability is taking into account using disjoining and conjoining pressures as presented in Beltrame (2019). The linear classical flux condition on the liquid/porous interface as used in Ding and Liu (2011) does not yield if a hydrophobic coating is present: the flux depends on the matrix moisture too (Doerr et al., 2000). Our present model takes into account wettability at the surface and also in the porous matrix (Beltrame and Cajot, 2022). Thus, the dynamics, both in the matrix and in the macropore, are governed by a gradient type equation (Thiele, 2018) where free energy terms characterize the wettability.
Using numerical simulation and bifurcation diagrams, a rich behavior is brought to light. Notably, several flow regimes in a microtube are in competition and complex spatial organization appears showing clusters of annular drop trains. In addition, the interaction between the flow in micro-tube and the imbibition in porous matrix leads to non-linear phenomena. In particular, decreasing the saturation in the porous matrix may slightly increase the flow rate in the microporosity for specific parameters. This non-linear analysis highlights the crucial role of wettability in the fluid transfer.
We present a foam displacement model with a separate balance equation for the surfactant concentration in the aqueous phase. We consider the gas mobility that depends on the surfactant concentration and the dynamic behavior of foam as Newtonian. We study traveling wave solutions for the proposed model considering a high initial water saturation (drainage scenario) and varying the injected water saturation. The traveling wave solutions are studied using phase portrait analysis and validated with direct numerical simulations. For surfactant concentration at the injection and initial conditions above the Critical Micelle Concentration (CMC), we only found traveling wave solutions in the case when these concentrations are equal. For surfactant concentration at the injection and initial conditions below the CMC, we found traveling wave solutions whenever surfactant concentration at the injection is greater or equal to that at the initial condition.
Non newtonian fluids in porous media flow offers complex interplays that are not fully understood. The Bingham rheology is an approximation of the rheology of a non-Newtonian fluid presenting yield stress, which are useful in several engineering applications, as reinforcement of soils by injection of slurries [1] and in the timely topic of fracking processes [2]. The subject is notoriously hard to study numerically, as we have a nonlinear rheology in a complex porous structure, but there has been recent advances in the field, for instance in characterizing a Darcy law [3]. This work aims to investigate the flow conditions of the Bingham body in complex geometries by using a Pore Network Model with a fairly novel numerical solver in the Augumented Lagrangian Method -- a method recently introduced by Talon and Hansen [4]. We are using the model to describe the qualitative behaviors of the yield stress effect, and have characterized a power law behavior that deviates from existing litterature, as found in [3] and [5].
Salt lakes occur worldwide in arid environments and are spectacular geological features, displaying breathtaking patterns on their surface. In these lakes, the only outflow of fluid is due to evaporation and dissolved salts in the groundwater precipitate at the surface, leading to the growth of a salt crust. Under the right conditions, ridges can be observed in the crust, resulting in a remarkable polygonal pattern. Understanding the formation of these distinct polygonal patterns is key to monitoring the dust emission potential of salt lakes. We model salt lakes using a 3D porous medium which is subject to a uniform through-flow, parameterised by the Rayleigh number and the lake depth. This leads to a base state characterised by exponentially-distributed salinity that is unstable for large enough Rayleigh numbers and whose instability leads to buoyancy-driven convection supported by salinity plumes. We simulate the dynamics numerically and analyse the sequential stages of the instability using characteristic properties of the system (e.g. average salinity fluxes, average and dominant pattern wavenumbers). Initially, linear growth away from the base state develops and patterns emerge in the surface flux of salinity. As nonlinearity becomes important, a net transport of salinity away from the surface builds. Eventually, plumes penetrate deep into the domain and the dynamics approach a chaotic but statistically-steady end-state, characterised by patterns which are strikingly similar to those observed in situ.
In this talk, I will discuss how liquid-liquid interfaces can be stably locked into periodic structured porous substrates and used for controlling the transfer of momentum, heat and mass with an external flow. First, we address the behavior of liquid-liquid interfaces locked in textured surfaces and exposed to an external shear flow. When the liquid-liquid interface remains stable, these surfaces can enhance heat and mass exchange with a bulk flow and reduce flow drag and biofouling. We demonstrate how shear stresses and soluble surfactants modify the dynamics of the liquid-liquid interface, resulting in waves, drainage and Marangoni stresses, all of which significantly affect transport processes with the external flow. Second, we introduce systems with liquid-liquid interfaces locked in three-dimensional periodic porous scaffolds. By tuning the wettablity and introducing appropriate “fluid traps”, we can immobilize interfaces of different morphologies, including spherical droplets or diamond-shaped structures. These multi-phase materials are remarkably stable and provide a very high area-to-volume ratio. We will discuss their potential as flow-continuous heterogeneous catalysis and for other applications that require optimizing mass transfer across interfaces, such as CO2 capture.
Granular materials exhibit a broad range of intricate dynamic behaviours. The study of their hydrodynamics is extremely relevant in the chemical and process industries, where those materials are widely handled and produced. Understanding how internal (e.g., particles size and shape) and external (e.g., applied stresses, moisture content) physical properties impact on the flow behaviour of solid particles helps industrial practitioners handle and produce particulates in an efficient and less costly way. Building upon previous findings applied to the process industry, the talk explores the challenges associated with the dynamic behaviour of dry and wet granular material and discusses recent experimental and modelling efforts on the flowability of pyroclastic powders. Pyroclastic powders are investigated with the aim of predicting and managing the hazard resulting from volcaniclastic debris flows, natural phenomena which occur when a mixture of pyroclastic fallout/current deposits and water move down slopes under the action of gravity.
The bulk of renewable energy production varies in a way that typically does not align with time-dependent energy consumption. This can lead to an energy surplus or a shortage of the energy supply causing a challenge in supporting the baseload requirements. Enormous storage capacity is required to accommodate these fluctuations in supply and enable large-scale storage of excess renewable energy. Gas as chemical energy carrier, and hence energy, can be stored in such large amounts in subsurface structures of depleted reservoirs. Especially hydrogen gas is an excellent energy carrier and can be produced via electrolysis from the surplus of renewable energy.
First pilot tests in the field were carried out to estimate the risks of hydrogen loss and a reduction of the stored energy due to physical, chemical or biological processes. It turned out that most notably microbial processes lead to a decline of hydrogen and hence to loss of energy. In these processes, microorganisms convert hydrogen and carbon dioxide into methane. This observation, which is detrimental in terms of hydrogen storage security, yield to a new approach to efficiently store excess renewable energy in the form of “renewable” methane. Such operations include a (possibly cyclical) usage of CO2 and can therefore be treated as CCU projects (Carbon Capture and Utilization). The possibility of using these processes as in-situ bioreactor to generate and store “renewable” methane is investigated. The biochemical reactions which convert the gases also lead to a growth of biomass in the pore space. The expansion of biomass will reduce the available pore space for gas storage and likely the permeability of the reservoir rock as well. Consequently, biomass may compromise storage capacity and injectivity substantially.
This work aims to investigate crutial processes of the in-situ methanation on different time and length scales containing numerical simulations and laboratory experiments. With core flooding experiments on the meter scale, we want to get a macroscopic insight into the reactive transport mechanisms which are governing (a) the hydraulic properties, and (b) the gas conversion rate and hence the overall performance of the subsurface reactor. Numerical field scale simulations are used to study the dynamics of the macroscopic conversion process under assumptions concerning conversion rates and field geometries. By using simplified models, mechanisms are identified and their performance is investigated by sensitivity analysis. The approach is to perform generic field scale simulations in order to understand the flow and reaction kinetics and their coupling.
The aim of the presented study is to develop a workflow that contains both, numerical and experimental components to gain a holistic understanding of the physical and biochemical mechanisms of the in-situ methanation. With the resulting comprehensive datasets, we expect deep insights into bio-reactive hydrogen transport and the controlling parameters to be applied to future field cases.
Over the last years the interest in molecular hydrogen (H2) has soared: in many countries an accelarating energy transition is considering hydrogen as the main energy carrier of the future. To enable large scale storage for hydrogen, research on subsurface storage options in geologic structures as artificially created caverns in evaporites (e.g. salt domes) or depleted gas fields mainly in sandstones is pivotal. Some countries as Germany have large volumes available in existing and additionally mineable caverns being the preferred option, but many regions in Europe will continue to use porous rocks – and there mainly sandstone reservoirs of depleted natural gas fields – as underground storage option. In these sandstones geochemical reactions of the molecular hydrogen with dissolved ions in the pore water – and foremost – on reactive mineral surfaces have to be considered in risc assessments for selecting the most suitable rock formations. Some minerals may oxidize hydrogen resulting in a loss of hydrogen and the production of either water or hydrogen sulfide. These products may enhance the alteration of the rock by increasing dissolution-precipitation reactions – or impart on the recoverable gas quality e.g. by traces of hydrogen sulfide (Heinemann et al. 2021).
In this contribution the results from geochemical experiments at in situ conditions (elevated pressures and temperatures as in the subsurface in presenc of a liquid water phase) of one common reactive mineral in many sandstones –the iron oxide hematite Fe2O3 – will be presented. The discussion will focus on details on the processes observed, the kinetics of the overall oxidation of hydrogen by hematite, and parameters limiting the extent of the oxidation of hydrogen. An interesting link to ongoing research results in photo(catalytic) water splitting on hematite surfaces has emerged during the study. In addition, for a set of five different natural sandstone samples from the Bunter formation with different contents ot hematite the extent of the reaction was investigated over one month at a pressure of 120 bars and a temperature of 120°C. The overall oxidation of hydrogen was low, but significant differences were apperent between the rocks. Spatially resolved analyses by optical microscopy and Raman spectroscopy allowed to document the reaction products in the intact pore space in the sandstones during and after experiments, pointing to mineral matrix effects on geochemical reactions and hence changes in porosity, pore throat diameters and permeability. These findings will help in delineating guidelines for selecting the formations best suitable for storing hydrogen for extended times in subsurface sandstone reservoirs.
Geological porous reservoirs are seen as an attractive solution for large scale underground hydrogen storage (UHS). Microbes are expected to be abundant in these reservoirs and could have a significant impact on the UHS process as the stored hydrogen can be used in their metabolism. Next to adverse effects such as hydrogen loss, H2S formation and clogging, microbial activity could alter the wettability of the hydrogen/brine/rock system and, consequently, the hydrogen transport behavior during UHS.
To effectively exploit these reservoirs for UHS, a good understanding of the impact of microbial activity on the hydrogen transport behavior inside porous rock is crucial. In this work, we characterize hydrogen transport properties in a microbial active environment from the pore- to the core-scale using several experimental techniques: Wettability is characterized using the captive-bubble cell approach and microfluidics, while relative permeability and capillary pressure are measured during core-flood tests at the core-scale. The activity of the living brine used in the experiments, which contains the sulphate reducing bacteria “Oleidesulfovibrio alaskensis” is continuously monitored through the pH. Our preliminary results show that high microbial activity increases the contact angle with around 5º, making the hydrogen/brine/rock system less water-wet.
We use high-resolution three-dimensional X-ray imaging to study hydrogen injection and withdrawal in the pore space of Bentheimer sandstone. The results are compared with a replicate experiment using nitrogen. We observe less trapping with hydrogen because the initial saturation after drainage is lower due to channelling. Remarkably we observe that after imbibition, if the sample is imaged again after 12 hours, there is a significant rearrangement of the trapped hydrogen. Many smaller ganglia disappear while the larger ganglia swell, with no detectable change in overall gas volume. For nitrogen, the fluid arrangement seems largely unchanged. We suggest that this rearrangement is facilitated by concentration gradients in the aqueous phase – Ostwald ripening – and provide an estimate of the time-scales for the effect to be significant, which are consistent with the experimental observations. The work implies that there is less capillary pressure hysteresis in hydrogen storage, promoting hydrogen withdrawal efficiency.
Subsurface fluid flow primarily transpires in porous rocks, however, in low-permeability formations, interconnected rock fractures can govern fluid flow. Synonymous with fracture flow is the immiscible displacement of a wetting phase (e.g., brine) by a non-wetting phase (e.g., CO2), a process called drainage, which is fundamental to many subsurface engineering applications. Robust modelling of fracture drainage on the field scale is required to effectively predict and manage the risk of fault-related leakage. Despite this, the controls on leakage through a single fracture are only partially understood. Fluid transport through a natural fracture is complicated by aperture heterogeneity, which arises from opposing rough walls and the presence of discrete contact points related to chemical/stress alterations. For two-phase flow, phase interference is high in fractures as flow predominantly transpires in 2D rather than the 3D pore space of a rock matrix. Recent modelling and experimental studies have provided insight into how drainage progresses through fractured materials, however, a lack of investigation using a truly representative sample (natural rough fracture) at sufficient spatial and temporal resolutions limits the predictive insights of such studies. Here, we used fast synchrotron X-ray tomography to image drainage in a natural geological fracture (6 mm diameter & 18 mm length) obtained from the Carmel Formation, a regional caprock sequence overlying a naturally leaking CO2-charged reservoir in Green River, Utah (USA). Drainage was imaged continuously over ~3 hrs by capturing consecutive volumes at 2.75 μm voxel size with a 1 s scan time. The experiment was performed with analogue fluids (brine and decane) at a controlled fluid flux (capillary regime) analogous to that anticipated during CO2 fracture leakage. In this contribution, we will discuss the results obtained, which provide new insight into the micrometre-scale displacement processes that directly impact global fracture saturations (and leakage rates), and the key challenges associated with imaging drainage in such small fractures using synchrotron imaging.
Fluid flow through fractured rock masses determines groundwater resource utilization, contaminant transport and remediation, resource recovery (oil and gas, geothermal), and energy waste storage (CO2 geological storage). While the matrix determines storativity in most cases, fractures with high transmissivity control fluid flow. Fluid flow through fractures may be accompanied by particle transport, including detached native fines or injected proppants and lost circulation materials. Small-scale experiments fail to capture the radial-dependent inertial effects and particle clogging patterns that can emerge away from injection or extraction wellbores.
This research explores divergent particle-laden fluid flow through large-scale fractures. We designed and built a large-scale parallel-plate setup (diameter=900 mm) to mimic fractures with different surface topographies and apertures. The device is instrumented with multi-physics sensors, while the transparent plates facilitate real-time visualization and particle-tracking. We explore particles with different sizes, shapes and specific gravities (including quasi-buoyant and dense particles).
Experimental results, numerical simulations and energy-based analytical solutions highlight the development of an annular zone with negative pressure away from the central injection point (previously reported in very few publications in other fields). Annular depressurization is more apparent as the fluid flow rate increases, i.e., at high Reynolds numbers, and it is anticipated under field conditions during drilling (particularly while traversing high aperture fractures) or when imposing high fluid injection rates.
Quasi-buoyant particles follow the fluid streamlines. However, local changes in the fluid velocity field during radial flow can enhance particle retardation, which changes the local particle concentration and enhances the probability of clogging. Dense particles transported along horizontal fractures settle to form an annular “dune” during divergent radial flow. Experimental and numerical results show the interplay between particle concentration, fracture aperture and injection flow rate on the dune topology and its radial distance to the injection port. Particle deposition patterns and the resulting dune topology become more complex in fractures with rough surfaces or shear-induced anisotropic transmissivity. Analytical and numerical studies investigate the relative role of the various parameters involved.
Many subsurface processes feature mechanically closed fractures elastically deforming in response to stress changes. In cases involving temperature contrasts, such as geothermal reservoirs, these changes are typically due to thermal stresses as well as pore pressure. In turn, changes in hydraulic fracture apertures impact the flow field and thereby also heat transport, resulting in a strongly coupled system of governing equations. We study this interplay drawing on a series of numerical simulations using the fracture simulation toolbox PorePy. The thermoporomechanical system is solved fully coupled in both fractures and matrix, ensuring a rigorous numerical representation of the modelled processes.
Depending on the geographical and geological setting, geothermal energy is one of the few renewable energy sources that can supply a constant and reliable source of low-carbon heat and electricity. In the UK, the greatest potential for power generation from geothermal resources is limited to high-heat producing granites in SW England. It has long been known that there is the potential for geothermal energy extraction from Cornish granites. However, until recently, project development has been slow, with the United Downs Deep Geothermal Power (UDDGP) project being the first one to be developed in the UK. The UDDGP will operate on the principal of producing hot fluids (>170°C) from the Porthtowan Fault Zone (PFZ), which is hosted in the Carnmenellis Granite, at ~4.5km depth and re-injecting the fluids using a subsequent well intercepting the same fault zone at ~2.5km depth. Project viability is dependent on how much fluid is stored within the PFZ and surrounding rock mass (porosity), and the ability of fluid to flow through it (permeability).
Granite is considered a low porosity and permeability rock, where fluid flow through it is usually controlled by fractures. The presence of mineral veins, as well as hydrothermal alteration of the surrounding rock mass, is evidence that hot fluids have passed through the PFZ (and other similar fault zones in Cornwall) in the past. Although hydrothermal alteration is common, its importance in granite-hosted geothermal resources has received little attention to date. In this research, we aim to understand how hydrothermal alteration may affect the transport properties of the PFZ, by measuring the petrophysical properties of samples of unaltered and hydrothermally altered Carnmenellis Granite.
Samples were collected from a fault zone at Holman’s test mine (a mine which is situated within the Carnmenellis granite), analogous to that of the PFZ. Porosity was measured using Nuclear Magnetic Resonance (NMR), and permeability was measured using a combination of steady state and unsteady state methods, on both intact and fractured material. All experiments were conducted at room temperature, and permeability experiments were conducted at confining pressures between 4 – 34 MPa.
We found that the porosity of the hydrothermally altered material (10%) is much higher than that of the unaltered material (1%), and that the matrix permeability of the hydrothermally altered material is ~4 orders of magnitude higher than that of the unaltered material. However, the permeability of fractured hydrothermally altered samples is lower than that of fractured unaltered samples. We suggest that the results from the fractured samples is due to the altered material being weaker and more ductile than the unaltered material, and under confinement, asperities deform more, which leads to a reduced aperture and consequently permeability.
Our results demonstrate that the altered material has the potential to store significant amounts of hot fluid in the subsurface, and that there may be an important contribution from the matrix in terms of flow, outwith the fractures. Crucially, our results highlight the importance of fully characterising the reservoir so that accurate resource predictions can be made.
Bacterial spreading through motility and growth plays a central role in agriculture, biotechnology, the environment, and medicine. These processes are typically studied in the lab in liquid cultures or on flat surfaces; however, many bacterial habitats—e.g., soils, sediments, and biological gels/tissues—are more complex and crowded 3D porous media. In this talk, I will describe my group's work unravelling how confinement in a 3D porous medium changes how bacteria behave. We have developed the ability to (i) directly visualize bacteria from the scale of a single cell to that of an entire population, and (ii) 3D-print precisely structured multi-cellular communities, in crowded 3D porous media more akin to their natural habitats. Our experiments using this platform have revealed previously unknown ways in which crowding fundamentally alters how bacteria move and grow, both at the single cell and population scales. Guided by these findings, we have developed theoretical models to more accurately predict the motion and growth of bacterial populations, and other forms of "active matter", in complex porous media. Taken together, these findings help to reveal new principles to predict and control the organization of bacteria, and active matter in general, in complex and crowded environments. They could also potentially help provide quantitative guidelines for the control of these dynamics in processes ranging from bioremediation and agriculture to drug delivery.
Groundwater contamination caused by nonaqueous phase liquids (NAPLs) is a significant environmental concern as NAPLs are ubiquitous and persistent pollutants, remaining recalcitrant to bioremediation due to their low solubility and limited bioavailability. Chemotaxis, the biased migration of motile bacteria toward chemical gradients, may facilitate remediation of NAPLs by transporting pollutant-degrading bacteria to residual contaminant sources trapped within the soil matrix. Greater accumulation of chemotactic bacteria was observed in Gao and coworkers’ study near NAPL contaminants at the juncture between different permeability regions in a heterogeneous micromodel [1]. Bacterial distributions in the pore space were influenced by chemical gradients and fluid flow, whose combined effect on bacterial transport is not well characterized in porous media. In this work, we aimed to investigate the transport mechanism of chemotactic bacteria from moving pore water into stagnant micropockets formed by oil-phase contaminant ganglia.
Chemotactic bacteria (Pseudomonas putida G7) were introduced at varying fluid flow rates (0.2-56 m/d) into a dual-permeability microfluidic device contaminated by NAPL. Bacterial suspension flowed preferentially through the highly permeable area while the low-permeability regions retained NAPL, which served as contaminant sources. Bacteria showed accumulation in micropockets near junctures of high- and low-permeability zones due to chemotaxis. However, accumulation in micropockets was initially increased and then decreased as pore fluid velocity increased in each trial. Convection in porous media did not simply override chemotaxis as previously observed in bulk liquid [2]; instead, higher pore velocity brought bacteria closer to NAPL sources than diffusion alone and triggered stronger chemotactic response by creating steeper chemical gradients. The optimal pore velocity, in terms of maximum bacterial accumulation, depended on the time scale of exposure to chemicals. Bacterial exposure time to chemical gradients was estimated to be τe=L⁄Vp , where L was characterize pore dimension and Vp fluid velocity. When exposure time exceeded response time (~ 2 s in Pseudomonas putida [3]), bacteria in bulk flow did not have sufficient time to bias their swimming directions in response to the presence of NAPL contaminants. Our results indicated that in heterogeneous porous media chemotaxis could be enhanced by fluid flow rather than merely being suppressed and chemotactic bacteria would lose their advantage when their exposure time to chemicals was below the threshold of response time.
Results from this study suggest that accumulation of NAPL-degrading bacteria in porous media micropockets will facilitate biofilm formation and enhance bioremediation. The dimensionless group of parameters comparing response time to exposure time will aid practitioners in determining an appropriate pore water velocity to use in delivering chemotactic bacteria for in situ bioremediation.
Non-aqueous phase liquid (NAPL) trapped in stagnant or low permeability regions, such as a dead-end fracture or rock matrices, are hard to remediate because they are mostly inaccessible by groundwater flow. In this study, we utilize branching fungus to remediate NAPLs immobilized in low permeability regions. Hyphae of fungi are known to generate tremendous turgor pressure on their tips [1] and produce surfactants [2] that allow them to navigate through small pores and air pockets in porous media and even penetrate rock matrix [3]. However, to the best of our knowledge, there has been no direct visualization of fungal hyphal penetration into oil-water interfaces, and its implication on the remediation of NAPL has been unclear.
This study reports the active removal of NAPL by fungi using microfluidic experiments. We isolated naphthalene-degrading colonies from a local coal-tar-contaminated site, and through the microbiome analysis, we identified and selected the fungal colony which constituted the major fungal populations in biofilms sampled from the site. The fungi were suspended in a minimal salt medium, and the solution was injected into a PDMS microfluidic chip with a flow channel surrounded by NAPL-saturated low porosity regions (Figure A). Vegetable oil with 10 g/L of naphthalene was used as the model NAPL. The fungal growth and the change of oil-water interfaces were recorded through a scientific CMOS camera at the pore scale. Our results showed the active removal of NAPL by fungi over 65 hours. We observed that clogging of the preferential flow path by fungi induced flow instability which led to a fingering-like displacement of trapped NAPL (Figure B). Moreover, fungal hyphae effectively penetrated water-oil interfaces and significantly enhanced the oil removal from low porosity regions (Figure C). In this contribution, we will further discuss the mechanisms behind the effective removal of NAPL by fungi.
Active in-situ microbial reduction of nitrate and soluble selenate to selenite and elemental selenium (less mobile) was induced by subsurface methanol injections and can stabilize selenium (Se) in mined waste rock. Biogeochemical processes require careful balancing of oxidants (oxygen and nitrate) and reductants (methanol). Pulsed nutrient injection strategies were used in the field in attempts to minimize near-well biofouling. Molecular biology and biological engineering methods have been used to characterize the microbial ecology and metabolic capacity of waste rock to treat mine-affected water for mining operations in the Elk Valley, located in southern British Columbia, Canada.
Laboratory scale batch and column studies with native microbes demonstrated the capacity to reduce nitrate and Se in saturated waste rock and showed that oxygen and nitrate inhibition of Se reduction was overcome via carbon addition. Biofilm grown on waste rock in saturated aerobic column tests was capable of 50 to 99% nitrate reduction followed by 40 to 95% Se removal; Se was sequestered in the biofilm predominantly in the zero-valent state. Denitrification and Se reduction was most rapid and efficient under suboxic conditions, and as high as 99% removal.
These results were scaled up to a pilot test and ultimately to a full scale in-situ saturated rock fill bioremediation system treating over 20 million L/d. In-situ biofilm coupons were deployed to track the microbial community structure using 16S rRNA gene sequencing. Applying the tools of molecular biology, bioengineering, geochemistry, and principles of microbial ecology to the understanding of biomineralization/bioprecipitation has been effective for management of nitrate and Se in mining settings.
Multiphase flow in porous media is widely studied and impacts countless applications in many natural and industrial processes, such as geologic CO2 sequestration, water infiltration into soil, and particle filtration. However, many questions remain, particularly with regard to the effect of the confinement and the geometry of the porous medium on the transport of dispersions.
We address these issues experimentally using controlled porous media: micromodels. We designed polydimethylsiloxane (PDMS) micromodels consisting of regular networks of vertical cylindrical posts, at the centres of which we injected water droplets in a continuous oil phase. A priori, no preferential paths are expected, except in a stochastic manner. However, we show that the radial alignment of the posts, i.e. the geometric tortuosity of the network, varies angularly in a periodic manner and plays a key role in droplet transport by generating reproducible preferential paths. By systematically varying the geometrical configuration of the posts, injection capillary number, droplet size, and droplet concentration, we characterise the droplet transport and the conditions for droplet breakup. At low capillary numbers, radial droplet transport is homogeneous. By increasing the capillary number, droplets initially follow the least tortuous paths before transitioning to a stable flow regime whereby droplets flow primarily in the most tortuous paths. Through large-scale droplet tracking, we demonstrate the influence of the geometric tortuosity of the media on the resulting droplet flow patterns and the counter-intuitive responses that can arise. Through this analysis, we emphasise the role of local geometrical configuration and propose a new metric for droplet transport which is the tortuosity of the porous media.
During drainage in porous media, film flow through networks of corners and capillary bridges can establish connections between seemingly isolated defending fluid clusters. Coupled with the drainage through the bulk of pores and throats, the flow through these networks constitutes a secondary drainage mechanism that can significantly affect final fluid configuration and residual saturations. We propose a simple numerical model that incorporates such mechanism by modifying the cluster identification algorithm in an invasion percolation model for drainage. In the model, which represents quasi-2D porous media, wetting-phase-filled sites are considered available to invasion when connected to the liquid outlet directly through successions of pores and throats, or through chains of interconnected capillary bridges. Within the available sites, the order of invasion follows a hierarchy of local capillary pressure thresholds that can be perturbed to accommodate gravitational and viscous effects. With the proposed model, recently obtained experimental data of drainage of Hele-Shaw cells filled with spheres were reproduced, showing good qualitative agreement. In particular, we investigated the existence of an active zone where film-flow-related events are more likely to occur, the capillary bridges size and spatial distributions, and the impact of film-flow drainage on the residual saturation.
Under the current climate change, assessing water transfer and infiltration in soil, considered as complex porous medium, is a crucial point for estimating consequences of either heavy rain on runoff or of drought on plant water uptake. In both cases, variations in soil wettability due to amphiphilic materials is an overlooked point, but can greatly affect the infiltration and water transfer, such as water repellency in soil [Doerr2000,Orfánus2021,Bens2007].
A macroscopic model of the infiltration of a water drop into a porous medium is developed and applied to a soil containing amphiphilic molecules such as Exopolysaccharides (EPS) found in soil near plant roots [Bérard2020]. These molecules present a hydrophobic or hydrophilic property depending on the water content in soil. Experiments found in literature [Liu2012,Hapgood2002] or performed in our laboratory show two main behaviors :
i) When the soil is sufficiently moist, imbibition is immediate and rapid as in hydrophilic soils.
ii) In contrast, for a dry soil, the drop does not infiltrate immediately and the subsequent imbibition is slower and depending on the soil hydrophobicity, the drop may never infiltrate.
Models based on Richards Equation [Richards1931] in the soil and its variants [Beljadid2020,Landl2021] can only reproduce the rapid infiltration of regime i). We propose here to derive new equations describing the hydrophilic and hydrophobic interactions both in the soil and on the soil surface in contact with the water drop to describe all water infiltration regimes. In place of a contact angle to characterize the wettability of the soil surface, we introduce a free energy term which includes attractive and repulsive interactions, derived from the modeling of drop dynamics on a substrate [Thiele2018] and include the dependence of the surface wettability on the water saturation in the porous matrix [Doerr2000]. Concerning the soil, we recently developed a water-dependent hydrophobicity model [Beltrame2022] which has been extended to the case of amphiphilic molecules. In order to reproduce to interactions between water at soil surface and in the soil volume, and to be consistent with thermodynamic principles, we show that it is necessary to add a term inside the porous matrix that depends on both the saturation and the film height at the surface. The resulting equation system is a fourth order PDE system similar to the lubrication model with wettability. To our knowledge, it is the first time that wettability, both in the soil and on the soil surface, is accounted for to represent water infiltration. The numerical simulation of developed coupled equations is in agreement with the experiments of the infiltration of a drop on a thin layer of sand containing EPS. We retrieve the dependence of the Water Drop Penetration Time (WDPT) test with the concentration of amphiphilic molecules and soil moisture.
Moreover, we are able to reproduce the two regimes of the infiltration dynamics: instantaneous infiltration and progressive and slow infiltration depending on the initial water saturation of the soil.
The capillary entry pressure is a fundamental quantitative parameter in two phase flows in porous media. The entry pressure is set primarily by the interfacial tension between the invading and defending fluids, the relative wetting properties of the fluids on the solid skeleton, and the length scale of the pore throats. In rigid porous media, all of these quantities are typically fixed, meaning the entry pressure is a constant for a particular choice of medium or fluids.
Here, we consider entry pressure for very soft porous media: specifically, those with elastic moduli comparable to the characteristic capillary pressure scale of the system. In such media, the pore geometry may undergo significant deformations due to the injection of an invading phase. As a consequence, the size of the pore throats and, hence, the entry pressure may evolve dynamically. We investigate how entry pressure is impacted by deformation using an idealised model experimental system comprising a quasi-2D column of water-saturated hydrogel beads, which we compress using a "capillary piston": a pressure-controlled bubble of non-wetting gas that squashes the column along its length, driving liquid from the far end of the column via a permeable barrier. These experiments are complemented with simple analytical models of analogous systems.
We begin by considering quasistatic loading and then extend our study to rapid dynamical loading. The latter scenario introduces a viscous pore pressure, which further opposes the entry of gas into the pore space. We discuss how the transient consolidation flow driven by the capillary piston introduces a spatial distribution of pore and solid stresses and, hence, a spatially-distributed capillary entry pressure within the system. We consider how we may decouple the different forces at work to extract quantitative knowledge of deformation-dependent entry pressure from experimental observations of percolation in soft porous media.
The spreading of Brownian particles in space, in a macroscopically one-dimensional domain, is described by a Gaussian law for the probability density function. But deviations from Brownian motion are widespread across disciplines, and diffusion frequently exhibits a power-law dependence $⟨x^2(t)⟩ ≃ K_βt^β$, in terms of the anomalous diffusion exponent $β$ and the generalized diffusion coefficient $K_β$ (with physical dimension length$^{2}$/time$^{β}$). Examples of anomalous diffusion arise in charge carrier motion in amorphous semiconductors, passive tracer particle motion and molecular motor-driven motion in biological cells, motion of particles in crowded environments such as biological membranes or dense liquids, and transport in gels. And yet, in naturally occurring porous media such as soils and rocks, as well as in natural and engineered pore structures such as membranes and in catalytic systems, diffusion of chemical species is almost invariably modelled as a Brownian process in terms of Fick’s law; these disciplines completely ignore the possible—or likely—occurrence of anomalous diffusion in such heterogeneous, disordered media. Here, we develop the continuous time random walk (CTRW) framework to anomalous diffusion (with no advective velocity component) in disordered and porous media. For an effectively one-dimensional, semi-infinite disordered system connected to a reservoir of tracer particles kept at constant concentration, we provide the dynamics of the concentration profile. We develop a formulation for the concentration profile $C(x,t)$ in a semi-infinite space for the boundary condition $C(0,t) = C_0$, using a subordination approach. From this, we deduce the tracer flux and breakthrough curve at a given distance from the tracer source. For the "residual" breakthrough curves, given by $1-C(x,t)$, we demonstrate a long-time power-law behavior that can be compared conveniently to experimental measurements, which are currently in progress. For completeness, we also derive expressions for the moments in this constant-concentration boundary condition.
We study the mechanisms of advective trapping in composite porous media that consist of circular inclusions of distributed permeability embedded in a high conductivity matrix. Advective trapping occurs when solutes enter a low velocity zone in the porous medium. Current multirate mass transfer (MRMT) models consider slow advection and diffusion but do not separate these processes, which makes parameterization difficult. Transport is analyzed in terms of breakthrough curves measured at the outlet of the system. We observe that the volume fraction occupied by the inclusion controls the curve's peak behavior, while the distribution of permeability is responsible for the shape of the tail. Using the continuous-time random walk framework, we derive a Lagrangian trapping model parameterized in terms of volume fraction and the distribution of conductivites in the inclusions. Then we show that this model is equivalent to a first-order MRMT and to a non-local partial differential equation for the mobile solute concentration derived by volume averaging of the microscale transport equation. The upscaled approach, parameterized by medium and flow properties captures all features of the observed solute breakthrough curves, and sheds new light on the modeling of advective trapping in heterogeneous media.
Human activity influences largely the unsaturated vadose zone. Located above water tables, the vadose is impacted by pollution, typically from agriculture and industrial activities. Therefore, understanding contaminant transport in the vadose zone is crucial for water resources management. However, there is still a lack of comprehension on dispersion in unsaturated porous media, and the subject remains an active research topic. Classical models such as advection-diffusion equation often fails to predict the dispersion, notably because of the increased heterogeneity in the multiphase system. Particularly, the link between the nature of the multiphase flow, the phase configuration in the porous medium and the dispersion stays unclear. Notably, experimental techniques often struggle to gather significant number of data and to consider long time dispersion. Therefore, we propose a multiscale multipoint statistic algorithm (MPS) to generate porous media images at different saturation of immiscible fluids. Generated images are based on experimental observations of immiscible multiphase flow air/water in a complex porous structure. To evaluate the representativeness of MPS generated images, we first analyze structural properties like the grains and air clusters size and geometry. These properties, compared to the experimental image’s ones, show a good match. Then, flow and transport are computed using Lattice-Boltzmann simulation in both experimental and generated images for different saturation. The resulting velocity distribution and concentration profile are very comparable. Particularly, the variances of the concentration profiles are very well reproduced. These results shows that MPS algorithm are willing to capture and reproduce the main pore scale features that govern flow and transport in a complex porous media. Therefore, the MPS algorithm could be used to generate a large number of images based on experimental images to study transport in unsaturated porous media. Notably, it allows more statistical coherence that leads to a better understanding of the link between two-fluid phase configuration and transport. Furthermore, we generate larger images (in comparison to experimental data) which allow us to get more insight on long time dispersion.
The unsaturated zone, including soil and vadose zone, controls the exchange of water, heat, and chemical substances between the soil surface and aquifers. It also hosts several processes involved in the transfer of nutrients, playing a key role in the availability of life-sustaining resources. Anthropogenic actions, such as agriculture, urban waste management, and industrial activities, add substances to the soil that might compromise the quality of fresh groundwater resources. Being able to predict the fate of such substances in the subsurface through an assessment of flow and transport processes is essential for mitigating their negative effects and for designing more effective remediation measures. We analyze flow and transport processes in unsaturated media at pore-scale using high spatio-temporal resolution X-ray computed micro-tomography (synchrotron). 3D transport experiments through a synthetic sand-like porous medium using a contrast solution were performed at different saturation degrees. Experimental data allowed the reconstruction of the plume’s advancing front and the tracking of its deformation over time, i.e., variation in the surface area of the 50% concentration plane. Results indicate an enhancement of the solute front deformation at lower saturation degrees and at larger flow rates, showcasing the role of the system’s heterogeneity in shaping solute dispersion. This is explained by a better connectivity of the system at lower saturation degrees, expressed through more negative Euler characteristic values, which highlights the better performance of the system at connecting initially separated parcels of fluid through the formation of preferential paths and a larger number of stagnation zones. To also link the observed solute front deformation rates with the hydrodynamics in the pore space, the average helicity density in the pore space was computed. Lower saturation degrees resulted in a larger helicity density, indicating a more heterogeneous flow field characterized by larger tortuosity and more complex streamlines, which explains the observed stronger solute front deformation at lower saturation degrees. Implications of these results on transport were assessed via estimation of the Okubo-Weiss parameter, which indicated a stronger control of both shearing and vorticity on solute plume deformation at lower saturation, potentially hinting at an enhancement of mixing rates. These findings represent a major step towards understanding the control of saturation on the hydrodynamic landscape within the pore space and on the deformation rate of solute plumes and fronts, both essential to understand mixing dynamics in unsaturated porous media.
The drying of heterogeneous porous materials is accompanied by capillary pumping from large to small pores which results in the surface remaining partly wet, guaranteeing an almost constant drying rate. At certain degree of saturation, the capillary pumping is turned off and the material experiences a decreasing drying rate. A two-component two-phase Lattice Boltzmann model [1] is used at pore scale to simulate the convective drying process of a dual porosity layered porous material showing the influence of inflow air speed (Re number), inflow vapor concentration difference from the liquid-vapor interface and contact angle. Using these parameters, a universal scaling law is derived which allows predicting the drying rate during the constant drying period [2].
The conditions for capillary pumping are derived based on simulation of the drying of a system of two (and four) connected channels of different size. Sequential drying of the channels from large to small guarantees a maximal drying rate, and is controlled by the capillary pressure difference between the channels and the fluid permeability of the connecting pores. An analytical model at pore scale is developed based on this interaction between capillary channels where the drying across the boundary layer is modelled with a mass transfer coefficient. This analytical model is applied to the drying of real porous materials, like ceramic brick and calcium silicate stone. The former material shows from start a high drying rate and dries out over several days, while the second material needs hundreds of days to dry. Using the developed drying model and a known pore size distribution, the drying curve for these two materials can be predicted with good agreement. Finally, we use this new model as a toy model to design the pore structure of materials to meet expected drying patterns, showing that the presence of well-connected coarse pores of different sizes promotes a fast drying of porous materials.
Freeze-drying is investigated based on a non-isothermal pore network model of coupled heat and mass transfer [1]. Simulations were carried out using image data from X-ray tomography (µ-CT) of freeze-dried maltodextrin, which was originally prepared with a solid content of c = 0.2 w/w solved in water [2]. Freeze-drying was conducted at a shelf temperature of -18°C and a chamber pressure of 10 Pa [3]. The experimental parameters were used in the pore network simulation in which a domain size of 100x100x250 µm3 was considered. The pore network simulation provides data about the dynamics of the pore scale resolved sublimation front propagation as well as local temperature and pressure evolution and vapor diffusion rates. It can be shown and analyzed for the first time how the sublimation front travels through the pore network in dependence of pore size distribution and various different process conditions. For this purpose, different temperature and pressure conditions were applied at the boundaries of the pore network. Besides µ-CT image data, also regular pore networks with different pore size distributions (monomal and bimodal) were implemented. The latter option is faster than imaging and image processing and allows to study more fundamentally different scenarios. This way, the evolution of the sublimation front can be studied at the limits of i) heat and ii) mass transfer controlled freeze-drying regimes as well as intermediate situations. As a result, the conditions for the formation of either flat or structured sublimation fronts can be provided. The outcome of this study can thus be used as a base for the prediction of material collapse.
The flow of non-Newtonian fluids in porous materials can be found in many industrial applications such as chemical engineering, subsurface engineering (de-contamination, energy production), and the food industry.
The relation between the shear stress and viscosity in non-Newtonian fluids is not linear and it is time-dependent, making it difficult to understand their behaviour. Due to the complex microstructure of pores in the porous media, the shear stress in each pore will be spatially variable, and thus the rheology of the non-Newtonian fluid would spatially vary along the flow pathways. Thus, it is very challenging to know how to upscale the shear-stress to estimate the upscaled porous medium-based rheology. Experimental characterization of non-Newtonian fluid flow inside three-dimensional porous media is not feasible; however, pore-scale modelling offers a versatile tool to understand and simulate non-Newtonian fluid flow in porous media. The pore-scale modelling offers a better understanding of fluids rheology, viscosity, thermostability and flow diversion.
The present study examines the feasibility of upscaling the non-Newtonian shear-thinning fluid bulk rheology to porous medium rheology using the pore-network modelling approach. Laboratory work was done to obtain the fluid’s bulk rheology. Then, a pore network model was constructed based on the Meter model equation and Hagen-Poiseuille law to simulate the porous medium rheology. The numerical results provided the pressure drop across the pore network for a given flow rate, fluids bulk rheology and pore-network geometrical and topological properties. As a result, the upscaled viscosity was back calculated using Darcy’s law and compared to the bulk rheology. Due to the high cost and time consumption of the laboratory and modelling work, it is essential to find a simple way to predict the porous medium rheology from the bulk rheology. Thus, based on several simulation scenarios for various pore-size distributions, permeability and flow rates, an empirical equation was proposed to predict the porous medium rheology based on the bulk rheology and porous medium properties.
Subsurface CO2 storage is a means to limit emissions to the atmosphere and global warming. Residual trapping, which occurs when brine invades the pore space occupied by the migrating CO2 plume and creates disconnected CO2 ganglia, is one of the mechanisms by which significant amounts of CO2 can be stored safely in the subsurface [1]. Experiments on rock samples show that larger amounts of CO2 can be residually trapped in the presence of both oil and water [2], suggesting depleted hydrocarbon reservoirs are suitable sites for CO2 storage. However, reservoir conditions and fluid compositions vary widely, leading to different three-phase displacement mechanisms and residual trapping from miscible to immiscible conditions. Further, mass transfer between phases may change the amount of residually trapped CO2 over time.
A residually trapped gas-bubble distribution will undergo mass exchanges through Ostwald ripening. It is a process that leads to mass transfer from bubbles having a higher chemical potential to bubbles with a lower chemical potential. Our previous study on three-phase ripening [3] has highlighted that the ripening of gas bubbles in the presence of oil and water can lead to different residual gas volumes in the two liquids and different order of bubble loss during evolution. In this work, we will analyse the impact of partial gas miscibility on ripening evolution.
This study uses a chemical-potential difference and level-set based methodology [3-5] that calculates mass transfer between bubbles through diffusion paths in oil and water and across oil/water interfaces. We use the Peng-Robinson equation of state to calculate gas bubble fugacity at reservoir conditions. We perform simulations on different idealised 2D homogeneous and heterogeneous porous media. In the 2D heterogeneous medium, we simulate oil-water-gas-water invasion cycles to generate residual phase volumes. We also use 3D pore-space images of a water-wet sandstone to simulate Ostwald ripening in near-miscible conditions on residual three-phase fluid configurations with isolated oil and gas ganglia obtained after a water-alternate-gas invasion cycle. We quantified the evolution of pressure, volume, surface area, and the number of residual bubbles, for different initial fluid distributions and saturations.
Our results show that the gas-liquid interfacial tensions, gas-liquid contact angles, and oil-water capillary pressures determine the residual gas bubble sizes in each liquid phase. Specifically, we find that the equilibrium volume of bubbles in the oil phase, as well as the range of bubble volumes in oil and water, are smaller for near-miscible conditions than for immiscible conditions. This decrease is due to larger gas-liquid contact angles in the near-miscible case creating smaller gas bubble pressure differences and less mass transfer (and lower mass transfer rate) even though the ratio of gas-water to gas-oil interfacial tensions increases with partial miscibility. During fluid redistribution, we also identify cases where the bubble coarsening leads to capillary instabilities and three-phase double displacements (e.g., oil displaces a gas bubble that displaces water), which can lead to lower residual gas trapping.
Geologic carbon storage (GCS) is a viable technology that can reduce carbon emissions to the atmosphere and mitigate the impact of climate change. Undeniably, a better understanding of the seismic potential in a GCS site and developing a corresponding mitigation strategy for risk management plan is required to ensure operational safety and minimize environmental impacts. To achieve this, a fast surrogate model is very powerful since many geological structures and operating conditions must be evaluated for their sensitivity and uncertainty in model prediction, while maintaining an acceptable accuracy level of high-fidelity models. In this work, we present Barlow Twins deep operator networks to be used as a surrogate for geologic carbon storage, with application for the Illinois Basin Decatur Project (IBDP) site, where a million metric tonnes of CO2 have been injected. The proposed data-driven framework is built upon a combination of deep operator networks (DeepONets) [1], and Barlow Twins reduced order models (BT-ROM) as well as its variations [2, 3]. To elaborate on this, we use DeepONets' architecture of branch and trunk nets in combination with a projector from BT-ROM. The loss function is constituted of point-wise differences (mean squared errors) and redundancy reduction terms. Our goal is to enhance DeepONets' capability by achieving better-reduced manifolds through an information bottleneck principle and a joint embedding architecture of BT-ROM. Our parameter space (input) contains heterogeneous hydrogeological properties (permeability and porosity) and operation constraints (e.g., varying injection rates and bottom hole pressure). We will compare a surrogate model performance with a high fidelity model in terms of fluid pressure and CO2 saturation.
SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525.
[1] Lu, L., Jin, P., Pang, G., Zhang, Z., & Karniadakis, G. E. (2021). Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators. Nature Machine Intelligence, 3(3), 218-229.
[2] Kadeethum, T., Ballarin, F., O’Malley, D., Choi, Y., Bouklas, N., & Yoon, H. (2022). Reduced order modeling for flow and transport problems with Barlow Twins self-supervised learning, Scientific Reports, 12, 20654.
[3] Kadeethum, T., O’Malley, D., Ballarin, F., Ang, I., Fuhg, J. N., Bouklas, N., ... & Yoon, H. (2022). Enhancing high-fidelity nonlinear solver with reduced order model. Scientific Reports, 12, 20229.
Abstract:Shale oil is a valuable unconventional oil and gas resource. It has a complex mineral composition, and rapid and accurate prediction of core flow parameters is crucial for its exploration and development strategy. At present, researchers predict flow parameters such as speed,pressure,apparent permeability,etc. through core experiments that require specific experimental conditions and methods, which are difficult and time-consuming. Conventional simulation methods for predicting flow parameters require considerable computational resources. Therefore, deep learning can be used as a pore-scale simulation prediction method. Considering that the mineral properties of the nanopore wall of shale oil have a large influence on the flow, a core dataset with organic distribution.
We predict the flow parameters of shale oil porous media by two methods. First,we designed a convolutional network for the dataset, adopted the structure of SE-ResNet, added the squeeze-and-excitation (SE) module to the double-layer residual module of ResNet18, and combined the characteristics of the SE block with the attention mechanism and ResNet to effectively obtain the information between channels and avoid the problem of gradient disappearance or explosion. Using SE-ResNet for directly predicting the apparent permeability from images. Another method attempts to couple a point cloud residual network with flow equations, reconstruct the flow field and predict apparent permeability. The coordinates of the porous media pore space are used as input to the point cloud network. Different slip conditions are set for organic and inorganic matter. The loss function of the neural network is constructed by the NS equation, the continuity equation, and the flow boundary between organic matter and inorganic matter.Based on the principle of PINN, optimization algorithms such as gradient descent are used to obtain the weight parameters of neural network connection and the physical parameters of partial differential equations. Only sparse data acquisition points are required to predict microscopic flow parameters. The above two methods are well applied in shale porous media.
Accurate modeling of water flow and solute transport in unsaturated soils are of significant importance for precision agriculture and environmental protection. However, traditional modeling approaches are considerably challenging since they require well-defined boundaries and initial conditions. Harnessing machine-learning techniques, specifically deep neural networks (DNNs), to detect water flow and solute transport in porous media have recently gained considerable attention [1]. In traditional DNNs, an artificial neural network with several hidden layers is trained solely using data to approximate parameter and state estimation, e.g., the spatiotemporal distribution of water content and pore-water salinity. However, data is extremely limited and sparsely available in subsurface applications. Physics-informed neural networks (PINNs) have recently been developed to learn and solve forward and inverse problems constrained to a set of partial differential equations (PDEs). Unlike traditional DNNs, PINNs are confined to physics and do not require" big" data for training [2]. However, hydrological applications of PINNs only considered an in-silico environment with spatial measurements of hydraulic head, water content and/or solute concentrations well distributed in the subsurface [3]. Such measurements are hard to obtain in real-world applications since they require drilling to extract soil samples or installing in-situ measurement devices at depth which also violets the soil's natural structure. As opposed to conventional subsurface characterization and monitoring techniques, non-invasive geoelectrical methods can provide continuous, extensive, and non-invasive information of the subsurface [4]. Nevertheless, the sensitivity of the measured electrical signal to various soil parameters, mainly water content and pore-water salinity, as well as inversion errors, could result in biased hydrological interpretations.
This work adopted the PINNs framework to simulate two-dimensional water flow and solute transport during a drip irrigation event and the following redistribution stage, using time-lapse geoelectrical measurements with unknown initial conditions. For that manner, a PINNs system containing two coupled feed-forward DNNs was constructed, describing the spatiotemporal distribution of both water content and pore-water salinity. The system was trained by minimizing the loss function, which incorporates physics-informed penalties, i.e., mismatch with the governing PDEs and boundary conditions, and measurement penalties, i.e., mismatch with the geoelectrical data. Two-dimensional flow and transport numerical simulations conducted with the Hydrus 2D/3D software [5] were used as benchmarks to examine the suitability of the described approach.
Results have shown that the trained PINNs system was able to reproduce the spatiotemporal distribution of both water content and pore-water salinity during both stages, i.e., irrigation and redistribution, with high accuracy, using five time-lapse geoelectrical measurements conducted with 59 electrodes placed at the surface. The trained PINNs system also reconstructed the initial conditions of both state parameters for both stages. It was also able to separate the "measured" electrical signal into its two components, i.e., water content and pore-water salinity. In addition, the subsurface geoelectrical tomograms were significantly improved compared to those obtained from a classical inversion of the raw geoelectrical data.
The existence of fracture network in porous media can have positive or negative effects on matrix-fracture transfer depending on the flow rate. In higher flow rates the efficiency of heat extraction decreases in fractured geothermal reservoirs due to preferential flow through the fractures. The fracture plugging can be considered as a solution to cope with it by diverting flow through porous matrix, resulting in more heat extraction. In present study, the effect of fracture plugging on heat extraction was investigated by cold water injection through the single fractured core plug including an obstacle at different flow rates. Present study indicated that the presence of obstacle in the fracture contributes high heat extraction compared to the absence of it because of fluid penetration into matrix from fracture. The analysis of heat transfer in porous matrix by Local Thermal Equilibrium (LTE) conditions leads to overestimated outlet temperature, thermal analysis with Local Thermal Non-Equilibrium (LTNE) conditions is more accurate, especially, at higher flow rates. In LTE condition both solid and fluid phases are at the same temperature and the average porous media temperature can be scaled with porosity and heat capacities of the fluid and solid phases[1]. In LTNE condition fluid and solid phases have different temperatures. Energy balance equations of the phases are coupled with heat source terms described with interstitial heat transfer coefficient [2]. Numerical studies indicated that in Darcy scale the main controlling parameter is interstitial heat transfer coefficient between solid and fluid phases. Minkowycz et. al [3] investigated the effect of rapid heat source change on NLTE conditions via analytical solutions in Darcy scale. Wang et. al [4] studied NLTE conditions in porous media with trapped fluid - solid matrix system. At LTNE conditions, interstitial heat transfer coefficient of Darcy scale problem can be calculated by averaging heat flows over the solid - fluid interface in pore scale.
In the present study the single fracture-matrix system was selected for cold water injection mimicking geothermal system. Since we are focusing on geothermal energy, there is a flow in our present study, where the Darcy scale parameters are extracted from pore scale model. There is constant injection rate at the inlet and constant pressure at the outlet of the fracture. Other boundaries are no-flow boundaries. Coating the system with epoxy resin allows no flow conditions at the outer surface. Temperature at the outer surface is kept constant.
Fig. 1. Matrix- fracture model domain for thermal transport problem
At the same flow rate temperature of the solid matrix is lower in LTE compared to LTNE condition, which clearly shows an overestimate for heat extraction (Fig. 2). Whereas at the fracture outlet, temperature of LTE is larger, since in LTNE condition there is still a transient heat transfer between the solid and fluid phases. As the flow rate increases, temperature at fracture outlet decreases in both LTE and LTNE conditions (Fig. 3).
Fig. 2. Matrix temperature values for a) LTE and b)LTNE conditions
Fig 3. Fracture temperature output for LTE and LTNE conditions
Existing literature suggests the importance of the thermo-osmosis (TO) for an accurate simulation of pore pressure evolution in heater tests for nuclear waste disposal in clay rock. However, there is limited consensus regarding the appropriate choice of parameters controlling TO and the extent of its physical impact. This study will use the ATLAS in-situ heating experiment, a full-scale experiment from an underground research laboratory in Mol in Belgium, to investigate the impact of TO on the thermal pressurisation in Boom Clay.
The ATLAS experiment was simulated using the open-source code OpenGeoSys. A fully coupled thermo-hydro-mechanical model combined with an inelastic constitutive model for the host rock was used. After comparison to published data, a parameter study, using an Assisted-History-Matching
workflow (Buchwald et al., 2020), was performed to obtain a good representation of the in-situ measurements without taking into account TO. Next, the same procedure was repeated with a model extended to account for TO. The comparison of both groups allows a clearer discussion of the influence of TO on temperature and pressure evolution in the studied system. The final step - uncertainty quantification of the TO parameterisation, puts the results in the context of large uncertainty of parameters documented in the literature. The impact of the said uncertainty will be illustrated by a range of plausible model predictions.
Li-Ion batteries are widely used for energy storage mediums because of their high volumetric and gravimetric energy capacities and proven mature technology level suitable for mass production. However, they have one key problem which is the heat generation during charging and discharging cycles. As the cells are getting too hot or too cold, battery life and performance decreases. If the heat generated from the batteries are not dissipated, there is even risk of explosion. To model the thermal behavior of the battery package one first needs to figure out the heat dissipated from a single cell, which can be set as an heat source value for the thermal modeling for the battery package. This can also be achieved with electrode scale continuum scale models, where Li transport in electrolyte and charge balance and Li diffusion in porous electrodes are modeled. However, electrode scale model would be over detailed to combine with a system level model.
First, cell level 2nd degree Thevenin’s equivalent circuit model [1] was developed under Matlab (fig. 1 and fig.2).
Figure1. Thevenin’s equivalent circuit model concept
Figure2. Thevenin’s equivalent circuit model developed under Matlab
The model requires Open Circuit Voltage (Uoc) vs State of Charge (SOC) relationship as an input. The parameters Ro represents contact/ohmic resistance of the cell, R1-C1 represents cell polarization, R2-C2 represents diffusion process [2], which are determined by fitting terminal voltage (Ut)-SOC measurement. Different from the lead acid batteries the Li-Ion cells have exponential decrease in the terminal voltage when the terminal voltage is approaching the cut-off voltage. To mimic this behavior, the equivalent circuit components (R,C values) are not set to constant values, they are varying as a function of SOC.
The developed Thevenin Model is able to calculate SOC, Uoc, Ut, state of health (SOH), remaining capacity, useful capacity [3], thermal power and generated heat. The main purpose of building an equivalent circuit model is to calculate thermal power generated by the cell (Thermal_Power=(Uoc-Ut)* I_current) [4,5].
Battery Package level simulation is carried out with setting thermal power of Thevenin model as a heat source to model temperature distribution. Thermal conductivity properties are taken from [6].
This is achieved with Comsol finite element simulation software (fig. 3).
Figure 3. Temperature distribution within the battery package with active air cooling
Total system model is created to model system dynamics with varying terminal current. Total system model is composed of Grid Connection, Auxiliary Load, Power Control System, Power Management System, Convertors, HVAC System, Battery Management System and Battery Package. The developed equivalent circuit cell module is set in the heart of the battery package module. During operation the main heat sources are battery packages. The temperature distribution of the total system is calculated with FEM simulator.
The advantage of this approach is that the experimental data can be perfectly fitted to the model data. The drawback of this approach is that the heat that is generated during charge/discharge process is assumed to be homogenously distributed at the outer surface of the cell.
Enhanced geothermal systems (EGS) are typically tight and naturally fractured like unconventional oil and gas (UOG) reservoirs, so the leading technology being evaluated for their commercial development is also multistage fractured horizontal wells (MFHW). The state-of-the-art approach of thermal recovery from EGS involves injecting cold water into a multiply fractured horizontal/deviated well and producing hot water from a parallel well above the injector, as in the ongoing Utah FORGE project. Considering the negligible control on hydraulic fracture size and orientation, the actual injection and production wells may not intersect planar and bi-wing hydraulic fractures in the ideal and optimum configurations they are simulated. This, coupled with the well-known risk of short-circuiting certain parts of the fracture network, could result in lower heat recovery from the field compared to the simulated MFHW recoveries. To address this problem, we present an alternative technology that employs unique configurations of mechanically cut fractures to recover heat efficiently from all parts of hot rocks in the subsurface. The precise control over these fractures’ location, size, orientation, and conductivity facilitates the design of suitable configurations of intersecting fractures.
This paper presents high-resolution numerical studies of thermal recovery from both MFHW and the proposed approach. We simulated several cases with and without stochastic natural fractures to evaluate the performance of these technologies in such systems. To facilitate a reasonable comparison between the MFHW and the proposed technology, we ensure that the total fracture surface area is the same. The results from the natural stochastic fracture systems studied indicates that the contribution of natural fractures to heat recovery is minimal in the proposed approach. This is due to the flexibility in designing the mechanically cut fractures to avoid being short-circuited by large natural fractures or faults known to be present in the subsurface. We simulated several cases, including one based on the published model parameters of the Utah FORGE project. All these simulation results show that the proposed approach can recover 50% to 140% more thermal energy than the state-of-the-art approach based on MFHW.
The temperature profiles after simulating 50 years of thermal recovery show that the precise control over the location of the fractures allows the reliable and efficient recovery of heat from all parts of the EGS, which could be the key to their commercial development. Finally, the control over the location, size, orientation, and aperture of the mechanically cut fractures provides more reliability in comparing the system modeled to the actual EGS in the subsurface. In contrast, the actual MFHW system could be much less efficient than the simulated system because of the lack of control over the hydraulic fractures' size, orientation, and geometry. There is also no guarantee that the injection and production wells will intersect all the hydraulic fractures.
Cellular membranes serve as selective barriers for regulation of molecular transport between interior and exterior of cells. Under the effect of electric field pulses of very short duration (from several hundred of nanoseconds to several milliseconds) with pulse amplitude from 100-300 V/cm to 100-300 kV/cm, the biological membrane is electrically pierced and loses its semi-permeability temporarily or permanently. The electrical permeabilization of biological membranes (called electroporation) may be reversible or irreversible. It was shown that electroporation can serve to introduce into cells or extract from cells small and/or large molecules. This phenomenon has been applied to amplify the insertion of nucleic acid molecules in genetic modifications, to enhance drug transport in cancer treatment or for the killing of microorganisms. Electroporation can also be used to enhance extraction of valuable cell compounds (polyphenols, carbohydrates, proteins,..) from biological media (plant tissue and biomass materials). Biological tissue with electroporated cell membranes, but with a preserved cell wall network, is selectively permeable. For the purpose of mass transport, electroporated cell tissue presents a porous network with improved permeability and diffusivity characteristics.
This lecture presents the mechanisms of cell electroporation, its impact on the physical properties of biological media, and gives examples of mass transfer enhancement in electroporated cell network. Different methods to detect and quantify electroporation phenomena in porous network of biological tissue are presented. Impacts of electroporation on the mechanical, diffusional and electrophysical properties of biological media are illustrated by numerous examples. Physical models of liquid expression and compounds diffusion in compressible electroporated biological tissue are presented. Several innovative green technologies based on the pulsed electric energy induced electroporation are presented, including selective extraction, filtration, pressing, and drying of plant materials and biomass.
Numerical modelling with commercial software (CMG) was used to analyse of the effect of contrasting permeabilities on fluid flow and hydrogen plume development in subsurface, porous media employed in underground hydrogen storage. Increasing heterogeneities were introduced to reservoir-scale simulations, based upon the Navajo sandstone, Utah in an aquifer-supported system. Initial investigations into the effects of well placement on reservoir pressure, cumulative hydrogen and water production in a homogeneous and heterogeneous model were used as baseline simulations to benchmark the performance of scenarios containing further permeability contrasts.
The results show, in terms of well placement, that production well placement at the top of the reservoir is the most important factor to maximise hydrogen production, due to the buoyancy of hydrogen. The relationship between permeability and viscosity in Darcy’s equation of flow provides a rudimentary guide to the behaviour of hydrogen in relation to contrasting permeabilities. However, reservoir heterogeneities affect fluid pathways, linking the effects of previous permeabilities, creating compartments and impacting upon the flow of other fluids. These characteristics, coupled with hysteretic effects, affect local pressure gradients, the other variable in Darcy’s equation, and determine the hydrogen migration. As a result, forecasts of plume development and reservoir performance need to consider the whole system.
As a strategy to match renewable energy supply and demand, surplus energy can be converted into hydrogen gas and stored in the pore space of geological subsurface formations such as saline aquifers and depleted gas reservoirs. Although similar operations with natural gas and CO2 are well studied, H2 has unique chemical and physical properties which, combined with cyclic injection and withdrawal, may cause complex phenomena that affect the efficiency and safety of storage operations. In this study, we investigate the risk of H2 injectivity impairment due to salt precipitation in the pores, driven by the interplay between the evaporation of water from the brine (originally present in the reservoir) into the injected gas, salt diffusion, and capillary fluid flow. To do so, we investigate the pore-scale salt concentration distribution and resulting precipitation patterns during gas injection in sandstone using micro-CT imaging. We present the first of such dry-out experiments performed with hydrogen gas under reservoir pressure and temperature, supplemented with N2 experiments to evaluate the influence of the original salt concentration and flow rate on the salinity gradient during drying, as well as on the resulting salt precipitation. In order to explain and predict the associated permeability impairment, we set up a pore network model with pore structure modification due to precipitated salt in individual pores, measured during the experiment. Our experimental results indicate that salt precipitation can cause up to a 30% reduction in permeability, which was also supported by numerical model outputs. The results of this study provide useful information on the impact of salt precipitation, and other geochemical and microbiological effects driven by concentration gradients on storage efficiency.
Hydrogen for clean energy is in the national and international spotlight. Offshore wind presents an extensive renewable energy source in the UK, and a large green hydrogen resource, positioning the UK to be a major player in the emerging global hydrogen market. In the UK and around the globe there’s a handful of likely subsurface hydrogen storage sites and it is widely recognised that hydrogen storage in porous media (rocks) will be necessary to support the scale of production, storage and use anticipated for a global hydrogen economy.
A key component of subsurface risk management is the suite of geological controls needed to ensure that storage is efficient and secure (i.e. that injected fluids do not leak from the storage formation). Storage security is closely related to caprocks and their capacity to hold the stored hydrogen at the place for the needed period of time. The work characterizes and describes the Kimmeridge Clay. A caprock widely spread across the Central and Norther North Sea and which has acted as an effective seal for numerous hydrocarbon fields. Two key phenomena defining caprock ability to seal are capillary pressure (CP) threshold and displacement pressure (DP). Capillary pressure of the caprock needs to be sufficient to resist the upward buoyant forces of the hydrogen that is built up beneath the caprock and displacement pressure rules the flow of leaking hydrogen. Both capillary and displacement pressure are effected by pores/throats size distribution and wettability. The work focuses on porosity & wettability determination and an effect of these parameters on capillary pressure. The aims is to better understand caprock compatibility to hydrogen stored underground and to examine the processes conditional to caprock integrity and its sealing capacity.
Understanding of geological controls is critical to inform the selection of appropriate reservoir sites as well as designing safe and effective storage and recovery schemes. The work outcomes will inform (a) security, monitoring and assessment approaches for hydrogen geological storage, and (b) potential for engineered barriers for enhanced containment or leak remediation.
To meet the global commitments for net zero carbon emissions our energy mix must transition away from fossil fuels. Hydrogen is gaining increasing recognition as a low carbon energy option to support this energy transition, tackling the hard to abate sectors such as decarbonising domestic and industrial heat, power generation and heavy-duty transport. It can also promote increased renewable energy uptake by acting as an energy store to balance supply and demand. For hydrogen to be deployed at the scales required for net zero, we will need access to large-scale geological storage. Depleted porous gas fields provide both the required TWh storage capacity and production rates that can be delivered over many months. Interseasonal hydrogen storage in underground porous formations involves complex displacement and trapping mechanisms that can influence recovery efficiencies over time and as such the economic feasibility of any underground porous formation hydrogen storage operation.
The talk will present the findings from our ongoing research into hydrogen displacement and trapping in porous media during multiple drainage and imbibition cycles, undertaken using x-ray computed micro-CT, micromodels and conventional core flooding experimental equipment. Our results indicate that hydrogen behaves as a non-wetting fluid filling the centre of the pores, with residual brine in the pore corners and throats. During multiple injection and withdrawal cycles we demonstrate that hydrogen trapping occurs via snap-off of hydrogen ganglia. Our work also demonstrates that the magnitude of the trapping depends on flow rate, pore fluid pressure and pore size distribution. This suggests that appropriate site selection and management of hydrogen injection and withdrawal rates can create the opportunity to minimise hydrogen trapping, optimising recovery efficiencies and the economic feasibility of underground porous formation hydrogen storage operations.
The existence of fractures in porous media has a strong impact on the characteristics of the flow behavior. In geological rocks, fractures occur both naturally as well as intentionally induced as in geothermal applications. Thus, accurate modeling and simulation of flow and transport in fractured media is vital for many industrial applications.
Mixed-dimensional models have been widely used for modeling flow in fractured media. The high aspect ratio of the fracture width as compared to their remaining dimensions allows for representing them as lower-dimensional manifolds. By combining the fundamental principle of mass conservation and Darcy’s law on each subdomain and mass transfer inbetween domains, underlying equidimensional models can be conveniently replaced. Yet, despite the large interest in mixed-dimensional models for flow and transport in fractured media, direct comparisons to high-quality lab experiments have been missing.
In this talk, we present such a comparison study based on PET experiments of tracer transport in fractured sandstone and corresponding numerical simulations of mixed-dimensional flow and transport (PorePy). In addition, we present tailored image analysis (DarSIA) used to transfer PET images to Darcy-scale images and to compare different Darcy scale images by suitable metrics.
Modeling subsurface flow and reactant transport on large (km) scales necessarily involves statistical descriptions of the underlying pore space. Pore-scale models are one route to the constitutive models needed to close the macroscopic transport equations, but when the reaction rate is high, classical upscaling methods fail. Here we describe a different upscaling scheme based on concentration fluxes rather than concentration fields. The upscaling is based on two observations from finite-volume (OpenFOAM) simulations of dissolution in perioidic porous materials. First, that the concentration field in each unit cell of a periodic array can be mapped to a universal spatial distribution that depends only on the incoming concentration flux. Second, that the shape (and therefore the porosity) of a dissolving unit cell in one position can be mapped onto a different unit cell at a different
time. These two observations can be combined into an ansatz for the time-dependent
concentration field in a dissolving (initially periodic) array of grains. I will present numerical results in support of this ansatz over a range of Peclet and Damkohler numbers.
Based on the proposed ansatz, we have developed an REV-scale model for the dissolution of a porous matrix, which is valid for all Damkohler numbers. The predicted porosity evolution is compared with pore-scale simulations in the Figure, shows results for a square array of disks at Peclet numbers of 20 (left column) and 200 (right column); the rows have Damkohlet numbers of 0.02, 2, 200, and infinity. (transport limited kinetics). The symbols indicate the model predictions in different unit cells, and the solid lines are pore-scale simulations.
At low Damkohler numbers (Da < 1), the REV model can be approximated by a continuum
theory. In both cases (REV and continuum) a single constitutive model is all that is required. It accounts for the fraction of the incoming flux to the unit cell that is absorbed by the solid. It can be determined by pore-scale simulations of small samples.
Recently, we have extended the REV model to include a spatially varying macroscopic flow. The key idea is that the concentration fluxes leaving a unit cell (or REV) are distributed in proportion to the fluid volume flux. This approximation is valid whenever the REV-scale Peclet number is greater than 1, or when the reactant within the unit cell is well mixed. The fluid velocity can be derived from the porosity-dependent permeability of the unit cell, which can be determined along with the effective mass-transfer coefficient from pore-scale simulations on small samples.
This work was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, Chemical Sciences, Geosciences, and Biosciences Division under Award Number DE-SC0018676, and by the National Science Center (Poland) under
research Grant No. 2012/07/E/ST3/01734.
Matrix diffusion is a critical process to capture in many subsurface applications in low permeability fractured rocks. In most discrete fracture network (DFN) models, a semi-analytical description of matrix diffusion is adopted in conjunction with Lagrangian particle tracking methods. However, the solutions to semi-analytical matrix diffusion are based on simple fracture networks, i.e., single fractures or multiple parallel fractures where fracture spacing is well-defined. Natural fracture networks can have spatially variable fracture spacing, orientations, and lengths, which could impact the accuracy of semi-analytical matrix diffusion, but has not been quantified to date. In this work we leverage new developments in the dfnWorks software suite to generate discrete fracture matrix (DFM) models to compare solute breakthrough in DFNs with semi-analytical matrix diffusion and DFMs with an explicit description of matrix diffusion. We are able to generate high resolution DFM meshes using a 3D Poisson disc sampling algorithm, where the faces of the fracture mesh conform to the faces of the matrix mesh. Governing equations for flow and transport in the coupled system are solved using the Amanzi Multiphysics code. We first verify the implementation of the explicit matrix diffusion in Amanzi for a single fracture against the semi-analytical solution. Then we systematically increase the complexity of the underlying networks to determine how well the semi-analytical matrix diffusion compares with the explicitly captured matrix diffusion and link the accuracy back to quantifiable fracture network properties.
Counter-current flow can be encountered under quite general conditions in two D or 3D two-phase flows in fractured medium. It may describe the imbibition of a wetting fluid replacing a non-wetting fluid inside the rock matrix. As the first order term of the driving transport equation drops-out, the resulting transport equation is a singular non-linear diffusion equation.
Although the early time solution of such equations is quite well known, it turns out that the long-time asympotics describing the asymptotic decay of the overall non-wetting fluid saturation is less known.
In that contribution, we develop a general self-similar solution, the time dependance of which is a power law of time, with an expoennt related to the NAPL rel perm decay at law NAPL saturation. The spatial part of the solution can be computed using a suitable fixed point algorithm that solves a non-linear eigenvalue problem. In stratified media, a complete analytical solution can be developped. In the genral case, numerical tests performed with accurate complete simulations confirm the relevance of that solution for many matrix block shapes. Consequences about the matrix to fracture overall fluw are also expected.
As the convergence of the solution to its asymptotic appears to be quite slow, a perturbative approach was developped to get a firther understanding of that observation. That allows us to develop an asymptotic solution under the form of a series of time power-laws that may describe the NAPL overall saturation decay once lower than 40% of its initial value.
Such solutions may be used to look for physically based proxys of matrix to fracture exchanges well-suited for developping an averaged description when considering a population of matrix blocks of random sizes and shapes.
In that talk, we will present previous findings published in following paper, as well as more recent results. Douarche, F., Braconnier, B., Momeni, S., Quintard, M., & Nœtinger, B. (2022). Counter-current imbibition and non-linear diffusion in fractured porous media: Analysis of early-and late-time regimes and application to inter-porosity flux. Advances in Water Resources, 169, 104319.
Microbially induced carbonate precipitation (MICP) is a promising biogrouting method for ground improvement and leakage remediation. Most studies to date have focused on MICP treatment of uniform clean sands, with few studies having been conducted at large-scale on well-graded soils more representative of in situ deposits. This study presents a laboratory meter-scale MICP test on medium-graded very gravelly sands consisting of 3.9% fines (< 63 µm) from field. The MICP treatment was conducted in a radial flow cell (diameter: ~1m; thickness: ~0.15 m) with an injection well located at the center and a constant hydraulic head at the outer boundary to replicate field conditions. Aqueous chemistry of the effluent samples at the middle of the central injection well and the outer boundary was continuously monitored, and transport of tracer and bacteria breakthrough in the radial flow cell and in separate 1-dimensional columns was modeled and simulated for a better understanding of the MICP process. The MICP-treated soil was subjected to a series of hydro-mechanical tests and microstructural analysis. Transport modeling and effluent sampling monitoring of the electrical conductivity and pH show that there was an overall good delivery and reaction of the bacteria and chemicals in the radial flow cell, but there also existed preferential flow paths due to soil heterogeneity and fines migration, which caused significant variations in permeability. Interestingly, compared to previous studies, the biocemented core samples with well-graded angular particles in this study had higher strengths (2.6-7.4 MPa) for a given calcite content (9.2-15.1%) than those in comparable studies treating uniform soils. This is likely due to a higher density of particle contacts as a result of both increased particle angularity (as suggested by backscattered electron imaging and X-ray imaging analyses) and higher packing efficiency in the medium-graded very gravelly sand. Consolidated-drained triaxial compression tests on two samples near the injection well showed a peak deviatoric strength of ~5 MPa under an effective confining stress of 500 kPa and a clear shear band was observed upon failure. To summarize, we have successfully achieved an overall good bio-cementation in the radial flow cell, despite the great soil heterogeneity. The study also suggests that migration of fines and the subsequent formation of preferential flow paths may be a challenge for producing uniform biocementation in field applications of MICP.
Microbial urease catalyses the hydrolysis of urea to ammonium and carbonate, which results in an increase of the environmental pH value. Addition of calcium ions then leads to calcium carbonate precipitation. Microbial Induced Calcite Precipitation (MICP) is successfully applied for, e.g., restoration of construction materials, soil reinforcement, or metal and radionuclide bioremediation. However, the precipitation process requires further optimization to make industrial application of MICP more efficient.
A high precipitation rate of CaCO$_3$ in the pore space of consolidated sand samples is necessary to increase the compressive strength. Multiple parameters as e.g. the urea and calcium ratio and concentration have been described in literature to have an influence on the precipitation process. However, most studies do not monitor the precipitation reaction itself, but perform application experiments on sand matrices and only check for the outcome data, typically compressive strength or the calcite content. Nevertheless, a small number of studies do try to derive strategies to improve the overall MICP process by getting an insight on the calcite precipitation kinetics and/or the crystal formation. These studies can mainly be grouped in two categories; there are simple beaker experiments with frequent manual sampling, as well as more advanced microfluidic experiments. Besides other disadvantages, both experimental strategies fail to provide a high amount of data points from a large number of parallel set-ups. Therefore, only the screening of individual parameters can be evaluated and possible interactions between parameters are disregarded.
Here, a new high-throughput microplate assay is presented, enabling online monitoring of calcite precipitation kinetics with a measurement interval of only 150 seconds. This assay was realised by making use of the automated high-throughput microbioreactor BioLector, which is able to measure a backscatter signal of 48 wells of a microplate in parallel while shaking at high speeds. The backscatter signal, intended for biomass estimation, corresponds to the turbidity in each well. When bacterial suspension and a cementing solution containing urea and a calcium source are mixed, calcium carbonate forms and precipitates, causing the backscatter signal to increase over time. As multiple precipitation kinetics can be measured in parallel by this system, the influence of multiple parameters on the precipitation rate can be easily compared. Interactions of multiple parameters influencing the MICP kinetics can be described as well by applying a Fractional Factorial Design (FFD) experimental approach.
In this study, the parameters OD600, pH, urea- and calcium concentration, type of calcium salt and culture washing were analysed. Three settings, which showed distinct calcite precipitation kinetics, were chosen to be adapted for quartz sand cubes solidification experiments to find a correlation between compressive strength and the precipitation rate. The results showed that very fast as well as delayed calcite precipitation is disadvantageous to solidify samples with high compressive strength.
Overall, the microbioreactor system can be successfully used to measure increasing suspension turbidity. This enables an easy systematic screening of a multitude of parameters influencing the precipitation rate and could help to optimize MICP applications, e.g. for building restoration to improve porous or deteriorated building components.
Worldwide production of concrete is estimated to be responsible for approximately 8.6% of all CO2 emissions originating from human activity [Miller et al., 2016]. Many countries, including the UK, now have ambitious targets to achieve net zero greenhouse gas emissions. To achieve these targets, the construction industry needs to transform its use of materials and approaches to asset management, with a shift towards extending the lifespan of existing structures, rather than constructing new ones.
Microbially Induced Carbonate Precipitation (MICP) is a novel engineered process in which ureolytically active bacteria trigger the catalysis of urea, resulting in the formation of calcium carbonate crystals. MICP shows promise for a wide range of engineering applications including rock fracture grouting, soil strengthening and for stone and concrete repair.
The aim of this research is to develop a mesoscale Finite Elements Model (FEM) to predict the mechanical behaviour of MICP-treated concrete. In order to calibrate the FEM model, MICP treatment and tensile strength tests were conducted on concrete cores.
Seven cylindrical concrete specimens were drilled from a caisson acquired from docks in Devonport, England. Subsequently, the cores were artificially cut along their vertical length creating a single fracture within each core. A variety of filling scenarios were investigated: (i) open fracture with glass bead spacers (500μm in diameter) only present at corners, (ii) patches of glass beads within the centre of the fracture, (iii) fully packed with glass beads, (iv) fully packed with silica sand grains, and (v) fully packed with carbonate sands.
Cores were subjected to multiple treatments of MICP. Each treatment included the injection of Sporosarcina Pasteurii (highly ureolytically active bacteria) followed by injection of a cementing solution consisting of calcium chloride and urea. Core permeability was monitored after each treatment cycle. Treatment was stopped once a 2-order of magnitude reduction in permeability was observed. After treatment, the cores were subjected to X-ray Computed Tomography (XCT) scanning and image analysis was conducted to evaluate the amount and spatial distribution of contact points created by calcium carbonate precipitation bridging across fracture surfaces. Following XCT imaging, the cores were loaded under Brazilian test conditions to evaluate tensile strength. After failure, the patterns of calcium carbonate precipitation on the surfaces of the fracture were inspected, validating the results derived from image analysis.
The experimental results show that the mechanical strength of the MICP-treated cores is governed by the amount of calcium carbonate precipitation which bridges across from one fracture surface to the other. A FE model simulating tensile loading has been developed which can be used to predict the mechanical behaviour of MICP-treated concrete as well as to better understand the influence of MICP treatment strategies on mechanical strength recovery.
Microbially-induced carbonate precipitation (MICP) has demonstrated promise in a variety of subsurface applications including immobilization of groundwater contaminants and remediation of leakage pathways associated with CO2 sequestration. In order to implement MICP at the field scale, however, the injection strategy must be tailored for efficacy in natural, heterogeneous porous media. Specifically, the overlapping effects of varied mineralogy and pore geometry on bacterial attachment, growth, and mineralization must be fully resolved. While the affinity of microorganisms for certain minerals (e.g. carbonates and clays) over others (e.g. silicates) is established, and biomass growth rate is known to be mediated by variables such as pH, such insights must be synthesized to develop injection strategies that produce desirable quantities and distributions of precipitate.
In this study, we investigate four questions pertaining to the influence of mineralogy on final precipitate distribution in a typical MICP injection consisting of separate attachment, growth, and mineralization phases. First, we assess to what extent initial biomass distribution is correlated with mineralogy. To this end, we construct modular columns to resolve average attachment rate versus distance from inlet; attachment rates are then determined experimentally for a set of common minerals including silica sand, kaolinite, Na-montmorillonite, and natural limestone grains, with parameters including grain size distribution, flow rate, and pH held constant. Second, we examine the correlation between final precipitate distribution and initial biomass distribution; this is accomplished by post-MICP characterization of columns via X-ray computed microtomography (XCT) for spatially-resolved precipitate distribution. Third, we attempt to decouple final precipitate distribution from initial biomass distribution through two modifications to the growth stage of injection. These include mechanical redistribution of biomass through rapid flow-induced shear sloughing, and slowing of biomass growth rate near the inlet via influent media acidification. Finally, we determine whether these modifications remain effective when applied to natural cores of clay-rich sandstone, whose pore size distribution differs from the engineered columns. Taken together, these experimental results elucidate the influence of mineralogy on the distribution of precipitates for typical MICP processes, and suggest avenues for optimizing injection strategy given mineralogy.
We study the synergistic impact of wettability and viscosity on immiscible fluid displacements in heterogeneous porous media. Direct Numerical Simulations are performed for viscosity ratio M (of invading vs defending fluid) ranging several orders of magnitude and contact angles ranging from very small to very large i.e. from completely wetting to completely non-wetting. The capillary number is kept constant at 𝐶𝑎=1×10−6 for all the investigations.
We notice different fluid displacement patterns such as fingering and compact displacements when the Ca is maintained low and by varying the viscosity ratios and the contact angles. For viscosity ratios greater than 1, the morphology of the displacement patterns is observed to be compact and is hardly affected by the wettability. On the other hand, at viscosity ratios lower than 0.1, we observe viscous fingering during imbibition and drainage. When the viscosity ratio moves towards 1, capillary fingering emerges. This intriguing observation suggests that one cannot use the knowledge about the displacement patterns to comment on the wettability states of the porous medium.
We further quantify the pore occupancy by the invading fluid during imbibition and drainage. Though we notice similar displacement patterns that occur at lower (M<1) and higher viscosity ratios (M>1), we notice differences in the pore filling mechanisms by the invading fluid. For M>1, the ‘co-operative pore filling’ is prominent irrespective of the wettability state. For M<1, we notice the dominant pore invasion mechanism during drainage is ‘channelling’ whereas for imbibition, the wetting phase propagates over the surface of the solid grains gradually.
It has become increasingly common to examine multi-phase flow systems in the context of thermodynamics, with the aim of expanding the traditional capillary pressure-saturation (Pc-Sw) relationship to remove its dependence on the history of the system. A commonly used multiphase flow theory based on rational thermodynamics introduces specific interfacial area of fluid-fluid interfaces, anw (interfacial area per unit volume of the porous medium), as a separate thermodynamic entity to extend the Pc-Sw relationship and better describe the the system, including hysteresis. Past pore-network models and 3D imaging experiments have verified that the Pc(Sw, anw) relationship can uniquely describe two-phase flow under quasi-equilibrium conditions, but very limited work has considered three-phase-flow systems, and in particular the issue of interfacial area formation under three-phase-flow conditions for systems of varying wettability.
In this study, we examine the impact of porous medium wettability on three-phase-flow systems. High resolution three-dimensional images, allowing us to measure and analyze capillary pressure, saturation, and interfacial area throughout water and gas invasion (imbibition and drainage scenarios), were generated using x-ray microtomography. The experimental data allows us to evaluate the contact angle behavior for the various fluid pairs under both water-wet and oil-wet conditions, and demonstrated a significant difference in the three-dimensional capillary pressure-saturation-interfacial area relationship as wettability was altered from water-wet to a fractionally-wet.
Wettability plays an important role in many natural and industrial processes, like mineral processing, hydrocarbon production and ground water remediation. For multiphase fluid flow in porous media, extensive research has been performed on the influence of the wettability on the phase distribution and morphology in both, static and dynamic conditions (1,2,3,4,5). Wettability is also crucial for the topics of increasing interest - hydrogen and carbon dioxide storage, for example for the description of injection dynamics and assessment of caprock stability (6).
In order to investigate wettability effects on multiphase flow in a controlled and standardized manner, researchers utilize model systems like bead-packs or micromodels. Glass is one of the materials used for the creation of such models, being transparent, chemically inert and easily formable. Additionally, the wettability of the glass surfaces can be altered from its original hydrophilic to more hydrophobic state by reaction of silane/siloxane groups with the hydroxyl groups of glass known as silanization (10).
The degree of wettability alteration by silanization reaction depends on numerous variables (7,8,9) such as the reaction time, temperature, concentration, the nature of the solvent and the prior glass cleaning procedure. Although silanization is widely used for glass wettability modification, comparable detailed systematic approaches over a large range of geometries, treatment conditions and measurement systems are scarce in the literature (7,8,9).
In this work, dichlorooctamethyltetrasiloxane (Surfasil) treatment was investigated with the purpose of systematically obtaining and providing a guide for achieving a wide range of contact angles. Secondly it was investigated whether different geometries display comparable contact angles under similar treating conditions using independent methods of contact angle determination.
Wettability was quantified through contact angle measurements on glass plates, beads and 2D micromodels. Initially, the influence of the solvent, treatment time and Surfasil to solvent ratio on plates was investigated using the sessile drop method. After establishing a clear relationship between the parameters and contact angles, the same treatment parameters were applied to single bead, microchip and multiple glass beads, the latter to form a bead pack. Contact angles from single beads and micromodels were obtained using image analysis of projections, while contact angles within the bead-pack were extracted from segmented 3D micro-CT images using algorithms (11).
By varying treatment times and the Surfasil to heptane ratio, it was possible to achieve a wide range of comparable and repeatable contact angles; from the initial 20 to 100 degrees as ultimate non-wetting state measured for air-water systems; for plates and individual beads, see figure attached. The flooding treatment in the micromodel was so far limited to the ultimate non-wetting state, showing comparable results to the plate and individual beads within limitations of the measurement.
Contact angle derivations from the bead pack using the 3D micro-CT images showed higher contact angles in comparison to the single bead, but it confirms a larger spread of the contact angle as observed in the literature (12).
Figure caption: The dependency of the contact angle on the volume ratio. The contact angle increases until it reaches a plateau value.
In various subsurface systems, the interactions between multi-phase flow and mineral reactions play an important role in controlling the evolution of porous media. These interactions - especially the impacts of multiphase flow dynamics on mineral reaction rates - are rarely accounted for in continuum scale models, or are simply corrected via reactive surface area and saturation of the aqueous phase. However, the relations (e.g., power laws between reactive surface area and water saturation), used for the correction are not based on pore-scale dynamics. Our previous study of a single channel with different levels of roughness showed that the mineral reaction rate in a gas bubble flow is significantly reduced compared to a single-phase flow system at the same flow rate. The extent of reduction in reaction rate follows a non-monotonic relationship with respect to water saturation, in contrast to the traditionally-used monotonic relationship (i.e., a power law relationship). In this study, we extend our investigations to pore-doublet geometries, to examine how two-phase flow dynamics arising from competing channels as would be expected in complex porous media affect mineral reaction rates. For our investigations, we control the two-phase flow dynamics by varying the air and water flow rates (i.e., capillary number) and the relative difference between the competing channels. Calcite dissolution rate in these channels is quantified for the two-phase flow cases with different saturations and the corresponding single-phase flow case. The relationships between the changes in dissolution rate in two-phase flow cases and the wetted surface area, the interfacial area, and water saturation are examined to provide insights on improving reaction rate descriptions in multiphase continuum scale models.
Miscible phase flow in porous media plays a significant role in many natural and industrial processes, such as CO2 sequestration, aquifer salinization, and soil pollution. In these processes, a less dense and less viscous invading phase mixes with a more dense and more viscous defending phase at the interface between the two phases. The resulting mixture at the interface has an intermediate density and viscosity based on the mixing ratio of the phases. The invasion pattern is determined by the rate of mixing and displacement between the phases, which is influenced by the miscibility ratio, viscosity, and geometry of the phases. Most previous research on miscible multiphase flow has been conducted using 2D Hele-Shaw cells, in which a resident phase is displaced by an invading miscible phase introduced at the center of a circular plate. However, these studies do not account for the complexity of porous media structure at the pore scale, where the uneven advancement of the invading phase due to capillary or viscous forces is dominated by the heterogeneity of the porous structure. In this research, we will address the gap between the pore scale and volume scale by examining how the inner structure of the porous medium, or heterogeneity, leads to various mixing patterns for different inlet pressures and heterogeneity levels. We will use a low viscosity fluid invading and mixing with a high viscosity fluid in a 2D porous media at various flow rates and heterogeneity levels to investigate the impact on fingering patterns and displacement to mixing patterns. Our results will show that these variations in displacement to mixing have a unique signature at the Darcy scale as measured by flux measurements, demonstrating that the pore scale phenomenon for miscible phase flow in porous media can propagate to the Darcy scale.
Microorganisms can establish organized biofilms in many natural and engineered porous media systems with significant advantages to applications such as biofilm barriers to groundwater pollution. The formation of thick biofilms can change the pore structure and consequently alter the hydrodynamics and reactive transport in porous media. Yet, the impact of preferential flow path formation and spatiotemporal rearrangement on overall system reactivity in bioclogged systems remains poorly understood. A two-dimensional pore-scale numerical model was developed to examine the effect of mixing and reaction efficiency upon biofilm development, biomass growth scheme, and preferential flow path stability. Simulations of water flow and solute transport in the porous medium were coupled to a biomass growth and attachment model scheme for a period of 400 hours. Four biomass growth models were tested, including i) no decay, ii) kinetic decay, iii) degradation, and iv) mechanical detachment. Our results indicate that i) permeability reduction and variations in the biomass fraction reached a similar and quasi-constant value after 100 hours for all growth models, ii) the shifting location of preferential flow paths only occurred when biomass growth was overcome by the combination of shear forces with biomass decay and/or degradation, iii) flow stagnation zones enhance the formation of strong concentration gradients, and iv) the preservation of high overall reactivity within the system requires intermittent shifting in the location of preferential flow paths.
Microplastic fibers (MPFs) are the largest kind of microplastic in the environment by mass and their presence has been identified on every continent and ecosystem on the planet. MPFs are known to pose a threat to aquatic species and worms but impacts on larger animals and humans are largely speculative, in no small part to the difficulty in quantifying the dynamics of how these non-dissolved, colloidal masses move. The objective of this work was to advance our basic knowledge regarding how simple kinds of MPFs move through porous media using a combination of experimentation and numerical simulation. Pseudo-2d flow chambers, termed “meso-models,” were created as oversized analogs of micro-models that were big enough to inject fibers into a controlled flow field. High fluorescence MPFs were injected into the flow between an opaque backing and a clear polycarbonate top sheet, the flow was subjected to UV light, and the MPF paths through the periodic pore meso-model were captured directly using HD video. Image processing extracted the trajectories providing position and time from which both breakthrough curves and velocity statistics could be extracted. Numerical simulations of the experiments using the known pressure gradients and flow rates from the experiments were constructed and a bead-rod chain model of MPF migration [1] was tested against the experimental results. The numerical and experimental results showed strong similarities, differing mainly by variations that can likely be attributed stochastic fluctuations. These results are the first direct capture of MPF dynamics in any porous media and the encouraging agreement with the numerical results suggests that, despite their extraordinary complexity, predictive modeling of MPF dynamics is feasible, which will be essential for realistic risk assessment of any direct or indirect hazards posed by MPFs.
Wettability has an enormous impact on the effectiveness of enhanced oil recovery (EOR) techniques and geologic carbon storage (GCS). Water flooding or carbon dioxide injection in EOR or GCS imposes pore pressure on the media, which can induce deformation or even failure of porous media. Experimental studies have shown that wettability is a key factor in determining flow patterns along with fluid characteristics and injection conditions. However, most previous research has been conducted in non-deformable media. As some studies performed in deformable media are limited to macroscopic scale, pore-scale study on simultaneous deformation or failure by immiscible flow is still lacking. This study attempted to examine the effect of wettability on hydromechanical behavior using pore-scale modeling. Fluid injection pressure alters the porous structure by mechanical deformation, causing a change in flow characteristics with the re-distribution of pore pressure. Therefore, a two-way coupling between hydraulic and mechanical behavior is required to mimic this process. To achieve two-way coupling, force equilibrium at each node was assumed in pore network model and block-spring model. Flow pattern and mechanical behavior under various wettability were explored using a two-way hydro-mechanically coupled pore network model.
Acknowledgment
This work was supported by the Korea Institute of Energy Technology Evaluation and Planning (KETEP) grant, funded by the Korean Ministry of Trade, Industry & Energy (No. 20212010200020)
Quasi two-dimensional approximations of interfacial curvature, present in current network models of multi-phase flow in porous media, are extended to three dimensions. The effect of each principal radii of curvature on displacement is analysed using high resolution direct numerical simulations on synthetic geometries, for both uniform and mixed-wet wetting states, and the analysis is used to calibrate network model extensions. A fully three-dimensional consideration of interfacial curvature is shown to be a key step in improving the physical accuracy of network models. Finally, the calibrated network model is used to obtain predictions of relative permeability and capillary pressure for a water-wet and a mixed-wet Bentheimer sandstone, and compared to experimental measurements, where the inclusion of three-dimensional interfacial curvature yields more accurate predictions.
A large portion of shale gas is stored in the kerogen matrix as an adsorbed phase, responsible for the slow production after primary recovery. During CO2 injection, the preferential adsorption of CO2 over CH4 in the shale organic matrix facilitates the desorption of CH4; therefore, gas recovery can be potentially enhanced. In this study, the Navier-Stokes equation and the advection-diffusion equation are coupled in the Lattice Boltzmann method to simulate the CO2-CH4 displacement in two-dimensional dual-porosity porous media. The Langmuir adsorption kinetics is implemented at solid surfaces for mass exchange between the free space and solid matrix. The adsorbed gas is assumed to diffuse within the solid matrix homogeneously. The coupling scheme is validated by comparing the simulation results with the analytical solutions for mass transfer. A convergence study is performed for the lattice resolution and the number of extended layers at the inlet/outlet. The lattice Boltzmann model is robust and efficient in porous media of irregular complex geometries. Preliminary results show that the CO2-CH4 displacement is controlled by the inter-solid and intra-solid mass transfer as well as the mass exchange rate between them. The gas diffusion coefficient, adsorption/desorption rate constants, and pore geometry can affect the concentration and adsorption evolutions of CO2 and CH4.
The foundation of homogenisation methods rests on the postulate of Hill-Mandel, describing energy consistency throughout the transition of scales. The consideration of this principle is therefore crucial in our discipline of Digital Rock Physics which focuses on the upscaling of rock properties. For this reason, numerous studies have developed numerical schemes for porous media to enforce the Hill-Mandel condition to be respected. The most common method is to impose specific boundary conditions, such as periodic ones. However, the recent study of Thovert and Mourzenko (2020) has shown that most boundary conditions still result in the same intrinsic effective physical property if the averaging is applied outside the range of the boundary layer. From this discovery, it becomes logical to question the status of Hill-Mandel condition in porous media when homogenising away from the boundary. In this contribution, we simulated Stokes flow through random packings of spheres and a range of rock microstructures. For each, we plotted the evolution of the ratio micro- vs macro-scale of the energy of the fluid transport outside the boundary layer, for increasing subsample size of our porous media. Here, we prove that we naturally recover energy consistency across scales when reaching the size of the Representative Elementary Volume (REV), which is a known condition for rigorous upscaling. Furthermore, we show that this ratio for the energy consistency is a more accurate indicator of REV convergence since the mean value is already known to be unitary, which adds to the initial advantage of not having to impose any specific boundary conditions.
The trade-off between the field of view (FOV) and resolution of micro-computed tomography (micro-CT) is a hardware bottleneck limits capturing both heterogeneity and micro-structure detail. Efficient super resolution methods combine the upper limits of both FOV and voxel resolution, while efficient modelling permits analysis of the large domain. A low resolution image of a 1-inch sandstone core plug and an unregistered high resolution (1-micron) sample trains an efficient and world-first 3D un-paired super resolution convolutional neural network, Dual-CycleSR. The resulting 25,000 x 25,000 x 50,000 domain provides unprecedented geometric fidelity over a full-sized core plug and reveals spatial heterogeneity that is captured by pore-to-core upscaling, with which forward-modelling produces a close match with unsteady-state core flooding production curves.
Due to the complexity of rock structure with features on scales ranging ten orders of magnitude, the multiscale fractured carbonates require more complex analysis tools than the macroscopic simulation approaches currently used which have been developed for and validated against siliciclastic rocks primarily. Digital Rock Technology (DRT) offers a means to determine the system’s transport properties on a pore-scale basis. DRT allows the investigation of fluid flow mechanisms in a comprehensive setting (e.g. single/multiphase flow, reactive flow). Furthermore, pore network models (PNM) have been used to model complex rocks due to their superior efficiency (improved computational speed, carbon footprint, and scalability) and ease of characterising the porous media using representative geometrical and topological statistics. However, current PNM open-source codes rely on strict definitions of the pore elements with symmetrical convex bodies and pores which can be challenging or even impossible to define the necessary irregular pore geometries in carbonate. Moreover, PNM does not consider heterogeneous multiscale features like fractures and vugs, which are difficult to discriminate and segment but have distinctive flow properties that can critically change the overall system behaviour.
In this talk, we will present a novel machine-learning algorithm for the semantic segmentation of rock matrix, porous/vugular elements, fractures, and secondary mineralogy, which was optimised considering its accuracy, complexity (measured using the total number of parameters, number of operations, run-time, energy consumption, and carbon footprint), and explainability based on the Green-AI philosophy. After comparing several techniques, shallow machine learning methods were preferred due to their superior computational efficiency and explainability whilst achieving comparable segmentation accuracy. The workflow proposed is a hybrid algorithm relying on both region-based and filter-based techniques to achieve the best accuracy and speed. Firstly a 2.5D (slice-by-slice) analysis is performed to separate pores from larger features, with the size threshold selected via a Gaussian mixture model. Subsequently, the micro-fractures and pores are separated via watershed, and the resulting elements are separated into pore elements and over-segmented fracture elements. Following this procedure, pixel-level segmentation is performed to distinguish and potentially separate large fractures and vugs, using this more computationally intense method only on the uncertain areas, hence optimising performance. The matrix class is also analysed to identify secondary mineralogy, which has the potential to alter wettability.
Each feature class is further segmented into instances and idealised, creating a multiscale network. The pre-existing open-source codebase is expanded by increasing its flexibility to complex geometries and introducing multiscale features, which will remove one of the main weaknesses in current approaches. Moreover, fracture elements are also modelled distinct from pores, observing the typical flow regime. Validation of the algorithm against the solution via direct simulation methods (namely the finite volume method) and experimental results of known samples is ongoing.
The outcome of this research is that of a resource-efficient and explainable algorithm that can discriminate between pores, fractures and vugs (and optionally secondary mineralogy) enabling automatic linking of the semantic segmentation result with the pore-fracture-vug network extraction, hence improving the modelling accuracy.
Analyzing the physics under the same imaging condition is hampered by the domain difference between digital rock images from micro-computed tomography (micro-CT). Different scan devices, scan conditions, and sample conditions (dry/wet samples) are frequently to blame for domain differences in micro-CT rock images. Unpaired domain transfer by Generative Adversarial Network (GAN) is a method that reduces domain differences by transferring the image style from one to another without the requirement of paired images. Herein, we develop a pseudo-3D domain transfer network, Pseudo-3D Semantic CycleGAN (3D-PSCycleGAN) that transfers the rock domains with the user-defined semantic information in a 3D manner while only requiring 2D computational resources. The 3D stacking effect that is present in 2D networks without fail is eliminated by the pseudo-3D transmission. The 3D-PSCycleGAN opens up a way to analyze digital rock images under the same condition to avoid any bias or inconsistency.
The types, content changes and distribution of mineral components in unconventional reservoirs are complex and diverse, which brings great difficulties to reservoir evaluation. In order to more accurately carry out numerical simulation of rock physical properties based on digital cores, it is necessary to accurately identify and divide the skeleton, establish multi mineral 3D digital cores, and determine the types and distribution forms of each mineral component. The image multi-threshold segmentation of grayscale core images is a common method to obtain multi-component core images, but the thresholds need to be manually adjusted, which is time-consuming and may cause large errors. This study reveals the application advantages of semantic segmentation method in digital core, and establishes a multi mineral 3D digital core automatic semantic segmentation model based on depth learning. Firstly, the mineral distribution characteristics in the study area are obtained through QEMSCAN and the typical mineral types are determined. Secondly, it is proposed to use the typical mineral (glauberite) in the study area for feature similarity analysis to achieve automatic and accurate image registration. Then, the sample data is determined by image alignment, interpolation, image filtering, brightness averaging, sharpening and other preprocessing methods. Finally, a multi mineral 3D digital core semantic segmentation model based on depth learning is established. The obtained multi mineral 3D digital core is highly consistent with the QEMSCAN mineral analysis results, retaining the layered distribution of shale minerals, and is applied to conductive simulation and electrical property research.
We review challenges to accurate simulation of processes of foam injection into geological formations for CO2 storage, aquifer remediation and enhanced oil recovery, with a focus on numerical issues (Rossen, 2013). Foam responds in an abrupt, nonlinear way to changes in water saturation, surfactant concentration, and oil saturation, in ways that cause fluxes to fluctuate in time and space. For instance, in simulations of foam with oil, consecutive grid blocks can lie on opposite sides of a strong foam/weak foam boundary on the composition diagram. The fluctuations can be suppressed by including capillary diffusion in the simulation. In addition, difficulty in representing shock fronts can lead to an increase in the foam-swept zone in simulations. As the grid is refined these effects have smaller impact on the overall process but execution of the simulation slows. Consideration of the 1D fractional-flow solution for the same displacement can determine whether the increase in foam sweep is a numerical artifact (Lyu et al., 2021a,b).
The representation of near-well effects on injectivity can require an impractical level of grid resolution near an injection well (Gong et al., 2020a,b). In addition, some near-well effects are not yet represented in foam simulators. An imperfect solution is to refine the grid to the extent practical and simply disregard the rise in injection-well pressure predicted by the Peaceman in the injection-well grid block.
Because by definition foam is an interaction between gas and water, the naming of phases (gas or oil) in a compositional simulation of a miscible EOR process can have significant effect on the simulation of a foam displacement. Numerical dispersion of surfactant concentration is also a problem, but attempts to minimize its effect can lead to other numerical artifacts. Because foam is so sensitive to water saturation and capillary pressure, capillary effects are important, especially in finely laminated formations.
"Population-balance" foam simulators, which represent the complex dynamics of bubble creation and destruction along with the effect of foam on gas mobility, face additional challenges with instability and slow run times, especially for models that represent the multiple steady states seen in the laboratory. A minimum velocity for foam generation in co-injection of gas and liquid can be represented by at least two simulation models (Yu and Rossen, 2022), but the implications for foam propagation may not be fully resolved.
We collect and review the various numerical challenges to foam simulation. Some of these problems are largely cosmetic, giving for instance fluctuating fluxes and pressure gradient but no significant effect on sweep and final recovery. Others do severely influence the whole progress of the flood. We discuss the origin of the challenges, how to recognize them, how they can be mitigated, and whether they arise from a correct representation of foam physics or the unintended result of attempts to solve other numerical problems.
An important challenge in up-scaling inertial and compressible flows is the treatment of the non-linear terms remaining in the closure problems [1]. As a consequence of these non-linearities, current approaches require to solve closure problems that are themselves dependent on the local averaged flow [2], thus limiting the benefits of the upscaling procedure.
Here, a methodology is proposed wherein the non-linear closure problems are linearized according to relevant dimensionless numbers using power series. Indeed, dimensional analysis of the closure problems arising in the volume averaging procedure of inertial and slightly compressible flows indicates that non-linear terms are controlled by dimensionless parameters such as the Reynolds number and the dimensionless compressibility coefficient [3]. For each order of the power series decomposition, linear and intrinsic closure problems are determined. Finally, the effective properties of the medium for small values of the dimensionless numbers are obtained by truncating the developments to the appropriate order. One of the main advantages of the proposed method is that it does not require to solve the full closure problems for each value of the local averaged flow.
After assessing the validity of this approach against numerical solution of the corresponding non-linear closure problems, the global permeability tensor, including contributions from inertia (Forchheimer term) and compressibility, is determined. Finally, generalization of this methodology to other types of non-linear flows such as flows with temperature-dependent properties is examined.
Due to the limited resources of alluvial aggregates, the use of recycled aggregates has become a growing practice in the construction industry today. These recycled aggregates come from the recycling of building or road demolition waste. They differ from natural aggregates in their composition and structure. The high porosity of this hardened cement paste implies a high-water absorption, which can reduce the workability and modify the properties of hardened concrete.
This presentation investigates the water transfers between porous aggregates and fresh cement paste. First, sintered glass beads are used as a model porous media; similarly, the cement paste is replaced by water to identify the dominant physical phenomena better.
We thus show the influence of the geometry and microstructure of the porous medium on the imbibition kinetics by immersion in water. Therefore, the commonly used Washburn model must be adapted to describe the imbibition kinetics. Geometry-specific imbibition models are then developed.
In the second step, we characterize the water transfer between the fresh cement paste and the porous aggregates. By Nuclear Magnetic Resonance spectrometry (NMR), we show that the absorption of aggregates in fresh cement paste is lower than in pure water. This decrease in absorption is a consequence of the contraction of the fresh cement paste during imbibition. Furthermore, the absorption kinetics is also slowed down compared to the measurements in water.
While there are many methods available for the characterization of pore sizes of soils and other geological materials, most of them are expensive or destructive or, in fact, both. Non-Newtonian fluids have been utilized recently for that purpose, providing not only a cheaper and more accessible alternative to the classical porosimetry techniques but also a method that does not disturb the sample and can be used repeatedly. In particular, the so-called ANA method [1] derives the effective pore size distributions of the porous sample based on a set of saturated flow experiments with different shear-thinning fluids, in our case the aqueous xanthan gum solutions of different concentrations.
We will discuss a methodology to measure the progressive changes in the pore size distribution of a sample of sand that is placed in the standard triaxial test chamber and subject to a drained compression. After every compression step (i.e. after increasing the pressure level maintained in the chamber, thus further compressing the sample), a sequence of permeability measurements with fluids of different rheology is performed and the effective pore size distribution is approximated. The ANA approach is used in our case since the similar yield-stress method [2] requires using larger hydraulic gradients, which would disturb the effective stress imposed on the compressed sample.
[1] Hauswirth, S.C., Abou Najm, M.R., Miller, C.T., 2019. Characterization of the Pore Structure of Porous Media Using non-Newtonian Fluids. Water Resources Research 55, 7182–7195. https://doi.org/10.1029/2019WR025044
[2] Rodríguez de Castro, A., Agnaou, M., Ahmadi-Sénichault, A., Omari, A., 2020. Numerical porosimetry: Evaluation and comparison of yield stress fluids method, mercury intrusion porosimetry and pore network modelling approaches. Computers and Chemical Engineering 133. https://doi.org/10.1016/j.compchemeng.2019.106662
Mixing occurs when two miscible fluids are brought into contact. Dispersion is a homogenized manifestation of the mixing process, which averages velocity and concentration fluctuations that cannot be resolved at the scale of observation. Shear dispersion, the process of solute spreading in pipe flow, originates from the non-uniform velocity profile in a pore cross-section. Taylor dispersion is the asymptotic limit of shear dispersion, i.e., when the pipe is long enough.
In the limit of long times, Taylor [1] derived the advection–dispersion equation for the cross-sectionally averaged concentration with an effective dispersion coefficient. This effective dispersion coefficient is analytically given as a function of Pe, where Pe=UR⁄D_m is the Péclet number expressed in terms of the mean velocity U, tube radius R, and molecular diffusion coefficient D_m. For a tube of finite length, the dispersion exhibits the pre-asymptotic behavior, where the dispersion coefficient, defined by the temporal derivative of the mean square displacement of tracers, increases with flow time and eventually converges to the Taylor dispersion coefficient. There are numerous studies that seek to find the early-time solution for dispersion in straight tubes. However, due to the difficulties of theoretical analysis, most of the studies focused on tubes of circular cross-sections [2]. Since the cross-section of pore structures is highly irregular, it is difficult to upscale these pore-scale studies to the porous media scale.
In this study, we propose a correlation of dispersion coefficient in tubes of different cross-sections. The proposed correlation relates the dispersion coefficient to Pe and flow time. The present study can be easily combined with other simulation methods, such as pore network models, for upscaling the pore-scale shear dispersion to porous media scale.
The interaction of fluids with different types of porous media plays an important role not only on our daily lives, but also understanding natural and industrial processes. Detailed studies of evaporation processes in porous materials are required not only to increase the fundamental understanding but also to enhance performance in engineering terms. The efficient design, operation and optimization of such engineering applications rely on detailed and thorough understanding of the interaction in terms of exchange of mass, momentum and energy. Several different techniques have been implemented to study this behaviour experimentally. Most experimental investigations are conducted using weight measurements, where a completely saturated porous probe is placed on a balance. However, local values of the surface evaporation flux are difficult to determine using this measurement technique. For this reason, we want to use an existing measurement technique, which is the interferometry, to estimate these local evaporation rates at the interface of a porous medium. This measurement technique has already been used to investigate drying processes on porous media [1, 2], but also to determine concentration gradients on evaporating droplets [3] and the evaporation of binary water-ethanol mixtures [4].
In this work, the evaporation of different fluids in a fully saturated porous medium is examined with a Mach-Zehnder interferometer. The latter uses the phase shift between two collimated light beams that results from splitting the light from a single light source due to a change in density or refractive index. The evaporation of moisture from the porous surface causes a deflection of the interference fringes, which thus leads to a phase shift. From this phase shift $\Delta \phi$, the change of refractive index $\Delta n$ is computed using $\Delta n = \frac{\Delta \phi \lambda}{2 \pi t}$ where $\lambda$ is the wavelength of the light source, and $t$ the depth of the measurement region. To extract the two-dimensional phase shift from the interferogram, the Fourier transform based approach by [6] is used. However, one of the problems of the approach used is that the phase-retrieval technique give the detected phase wrapped into the interval $[-\pi, \pi]$. This is due to the non-linear wrapping function involved in the phase-estimation process. Unwrapping is the process by which these discontinuities are resolved and the result is transformed into the desired continuous phase $\phi_{con}(x,y) = \phi (x,y) + 2 \Pi k(x,y)$ where $ k(x,y)$ is an array of integers. The unwrapping problem has been an important research topic for over decades [5]. For phase maps composed from consistent phase maps fringe data, there are many different algorithms, but there is none used for our type of problem. The questions that needs to be addressed to resolve the unwrapping problem is: under what circumstances can this lost information be recovered? The main objective of this work are these phase discontinuities and how they can be solved to reproduce the local evaporation rates at the surface of the porous models as accurately as possible.
The efficient exploitation of coalbed methane (CBM) plays an important role in reducing outburst hazards, securing energy supply and reducing carbon footprint. Horizontal well cavity completion performs well in Zhengzhuang Block, China, with stable daily gas production of 10,000 m3, which was four times more than that of the adjacent fractured well. However, the stimulation mechanism of horizontal well cavity completion is not clear and it is a challenge to directly reveal the evolution of stress and strain. This study proposes a method, finite discrete element method (FDEM), for characterizing fracture and describing stress and strain evolution. A visualization experimental device based on digital image correlation (DIC) was proposed to measure the strain field. Then the established FDEM model is calibrated based on observations from the experiment. And the effects of cavity diameter and in-situ stress on short-term response of stress evolution and fracture extension were investigated. The results show that numerical simulations are in good agreement with the experimental observations. The stress concentration occurs first around the cavity, and then induces fracture propagation, which further leads to stress release. The fracture extension and stress relief is limited in the vertical direction as the vertical stress decrease. The fracture length grows linearly with cavity diameter. The key findings of this study provide insights into the progress of stress reconstruction and fracture extension in CBM horizontal well cavity completion.
Optimal Bayesian Experimental Design is one of the methods for data acquisition system optimization that is frequently utilized in subsurface flow problems. In this method utility function that measures expected quality of the experiment is derived from the first principles of probability and statistics as a function of design parameters (aka sensor location). Therefore, the optimal experimental scheme can be found as a maximum of that utility function.
The most significant advantage of this method is solid Mathematical foundations.
Unfortunately, direct calculation of the Utility Function is computationally expensive, because it requires nested MCMC integration [1].
In the author's previous work [2] it was shown that Polynomial Chaos Expansion (PCE) can be utilized to accelerate Bayesian Experimental Design significantly. Basically, a novel approach for the utilization of PCE to avoid nested MCMC integration was proposed. The key idea for developing of that novel technique was the orthogonality of basis polynomial functions in PCE.
Despite the advantages of PCE, Decsiion Trees and Gradient Boosting methods seems to be an attractive altermative to PCE due to high popularity and simplicity in tuning. Therefore, the present work shows how orthogonality ideas can be extended to Decision Trees or Gradient Boosting like methods to provide acceleration of Bayesian Experimental Design. In other words, the novel algorithm for construction of the Gradient Boosted trees with specific orthogonality constraint was developed and examined on several test cases that include flow in porous media. Additionally, validation against classical methods is provided.
Real-time subsurface flow simulation is desirable for managing groundwater resources, geothermal exploitation, carbon dioxide geological sequestration, or underground hydrogen storage. Data assimilation methods are developed to achieve this goal. However, assimilation models usually use mesh-based numerical methods. Remeshing is frequently required whenever new data to be integrated into the model are not located at the existing computational nodes. This study aims to develop an adaptation algorithm to accommodate node layout to the exact positions of additional data. For flexibility, we chose a mesh-free numerical method. We combined it with a fast node generation technique called the advancing front method to adjust meshless node placement before assimilation by ensemble Kalman filter. A hypothetical flow problem was used to test the proposed approach. The results show that the adaptive node adjustment works effectively for the real-time updating model. The accuracy and precision of modeling states and parameters were improved when integrating additional data.
50 % waste (quarry sand) is generated during the production of ashlar. In this study, this waste material is used to produce biosandstone as new and sustainable construction material. The quarry sand used in this study is delivered by the local natural stone plant (Picard) in Krickenbach (Germany). Viewed globally, there is a high need to investigate new construction material as alternatives to concrete, because with a share of about 8.6 % of anthropogenic CO2 emissions, concrete is a major contributor to global warming. Additional, not every sand like for example sand from the desert can be used for the production of concrete because the supporting grain is missing. Microbiologically induced calcium carbonate precipitation (MICP) offers the potential of a more sustainable alternative in construction. Additional, pretrials showed that a consolidation of sand from different deserts is possible allowing desert sand to be used as an alternative building material. During MICP calcium carbonate is formed by microbiological activity and can serve as a binder between mineral particles. This calcium carbonate can be an alternative binder to conventional cement mortar used for concrete. The most commonly used mechanism for MICP is ureolysis. In this process, a cell suspension and a calcination solution (urea + calcium ions) are applied alternately in a cyclic process, whereby calcium carbonate crystals are formed by the metabolism of the cells, which bind the aggregate (e.g.: sand) together.
In this study, Sporosarcina pasteurii is used as ureolytic microorganism to consolidate the quarry sand from the local natural stone plant in Krickenbach. In a first step, the quarry sand was consolidated by MICP to check if this is possible at all. Since this was successful, a deeper understanding of the influence of particle size on consolidation was investigated. Therefore, the quarry sand was classified into four different fractions of particle sizes and consolidated using MICP. In addition, the consolidated samples were scanned by micro computed tomography. Contact points and pore space depending on various parameters during the treatment of MICP were investigated. However, this study shows what influence the grain size has on the strength of the biosandstone. Furthermore, it will be shown what influence the pore volume has on the strength of the samples and whether the strength can be optimized by an optimal composition of the different fractions of particle sizes. Finally, a demonastrator will be presented produced from quarry sand using the interlocking principle. The interlocking principle is an adaptable modular structure based on the building block principle, which is functional without mortar.
This project is financially supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project-ID 172116086 – SFB 926, the “Landespotentialbereich NanoKat” and “Lehre Plus”.
Hydrocarbon spills into the subsurface can lead to the formation of light non-aqueous phase liquids (LNAPL), i.e., a separate phase, immiscible with water, representing a long-term environmental threat. Traditionally, the presence of mobile LNAPL is evaluated by installing monitoring wells in the area of potential concern [1]. This approach, however, provides only qualitative evidence of the thickness of LNAPL observed in the wells, thus not providing information on the distribution of residual LNAPL in the subsurface [2].
As an alternative, in the last decades, radon (Rn) has been widely proposed as a naturally occurring tracer for light non-aqueous phase liquids in the soil (e.g., [3], [4]) since it has been shown to tend to partition into LNAPL. Rn concentration in soil gas is expected to decrease in the impacted area compared to the value observed at background locations (not impacted by LNAPL), creating a measurable Rn deficit. This work examines the feasibility of using soil gas data collected in unsaturated soil at some distance from the source zone to apply the Rn deficit technique to identify and quantify LNAPL contamination. To this end, we developed a steady-state 1-D analytical solution based on a 3-layer model that simulates the transport and distribution of Rn in the source zone, capillary fringe, and overlying unsaturated soil [5]. The analytical solution was first validated against a more detailed numerical model available in the literature [6]. Then, a series of simulations were carried out to evaluate the vertical concentration profiles of Rn in soil gas above the source zone and in a background location not impacted by LNAPL. Simulation results showed that the parameters that most influence the migration and distribution of Rn in the subsurface are the distance of the soil gas probe from the source zone and, to a lower extent, the type of contamination (e.g., diesel or gasoline) and soil type. Based on these results, to aid the determination of LNAPL saturation, some nomograms have been developed that can be used to apply the Rn-deficit technique from Rn concentration data in soil gas collected at a certain distance from the LNAPL source zone. The developed nomograms show that the Rn deficit is more evident as the measurement point approaches the source area. According to the obtained results, the Rn deficit technique is a feasible method for qualitatively identifying residual LNAPL when Rn in soil gas is measured at distances lower than 2m from the contaminated zone. However, for an accurate quantitative estimation of the LNAPL phase content, soil gas probes should preferably be located at distances lower than 1m from the source zone.
The nomographs provided in this work, which allow the estimation of LNAPL saturation as a function of the distance of soil gas probes from the LNAPL source zone and of the type of soil and contamination, are generally applicable to all sites involving relatively homogenous soils. Conversely, more sophisticated numerical models should be preferred (e.g., [6]) in the case of heterogeneous soils involving geological barriers or stratified contaminations.
Nanofluids possess great application potential in enhanced oil recovery (EOR). However, the EOR effects and mechanisms of nanofluids with specific nanoparticles (NPs) are not clear. In the study, the molecular dynamics (MD) simulation is thus adopted to explore the displacement of trapped oil in the rough channel by various nanofluids. Our results indicate that nanofluids with hydrophilic NPs and Janus NPs hold a greater EOR effect (9.7% and 7.1%, respectively), while hydrophobic ones are not suitable for oil film (with EOR effect of 2.3%). Specifically, hydrophilic NPs increase the viscosity and the sweeping scope of injected fluid. Janus NPs are prone to stay at the oil-water interface to reduce the interfacial tension. Most of them adsorb on the bulge, alter the surface wettability, and squeeze the trapped oil while others remobilize the trapped oil by sliding along the interface. Due to the entering of a large number of hydrophobic NPs inside the oil clusters, the influence of oil molecules being bound by NPs greatly reduces the effect of volume replacement, which leads to a poor displacement effect and even a risk of plugging the channel. Among the nanofluids, the ones with Janus NPs can maintain a stable oil displacement performance under low pumping force, thanks to sufficiently long contact time between Janus NPs and the oil phase. Further analysis on capillary number highlights the applied prospect of Janus NPs in actual oil reservoirs. Our findings are favorable to understanding the mechanism of nanofluids in EOR and provide guidance on the screen of NPs.
Understanding of heat transfer processes in subsurface fractured rocks is critical for the development of geothermal resources. One of the challenging tasks is to build models that can adequately represent the complexity of the formation geometry and subsurface processes without the extensive computation cost. Single porosity models with effective parameters are commonly used for this purpose. However, these models are often too simplistic and inadequate to represent complex fractured rocks. The aim of this study is to evaluate when a single porosity model can adequately represent heat flow in fractured rocks. Our study uses numerical modelling to simulate heat flow and upscale hydraulic and thermal properties. We then use thermal breakthrough curves generated from the simulation results to evaluate the performance of the upscaling. Embedded discrete fracture model (EDFM) with explicit fracture and rock properties provides the base solution with which we evaluate the performance of the single porosity model. Our sensitivity analysis includes fracture density, connectivity, fracture lengths as well as the permeability contrast with the background matrix. The results indicate that single porosity models are mostly inadequate to reproduce the thermal breakthrough of fractured rocks except for cases where the permeability contrast between the fractures and the matrix is less than three order of magnitude. Overall, this study demonstrates when a single porosity model can be useful to represent heat flow in fractured rocks.
Slow Sand Filters are the last step of producing drinking water in the Netherlands which play a crucial role in removing microorganisms. A biolayer formed on top few centimeters of the sand, called Schmutzdecke, plays an effective role in colloid removal. A multi scale study is performed to investigate removal efficiency of this layer and attachment mechanisms inside these filters. Pilot plant/ column experiments are done at the meter/centimeter scale to study the effect of different operating conditions such as grain size and flow velocity. These experiments are done by seeding Escherichia coli WR1 as a model bacteria into the filters. At the micro scale, microfluidics are used to directly observe colloid-biofilm interactions, biofilm growth and colloid transport inside the porous media. Result shows that biofilm growth can clog the throats, make preferential flow paths, and decrease filters conductivity, while it can increase removal efficiency.
Inverse modeling plays a fundamental role in the subsurface characterization of aquifers, given the scarcity of available data. Several techniques have been proposed in the literature and tested using synthetic examples. However, one of the big criticisms of these techniques is the lack of demonstrations in real cases. In this context, this study presents the application of two of the most advanced inverse modeling techniques: the Ensemble Smoother with Multiple Data Assimilation (ES-MDA) and Deep Learning-based inverse modeling (DL), for the characterization of the non-Gaussian hydraulic conductivity field of a 2D tank model of an aquifer. The experiment consisted of the release of a fluorescent solution from a point source on a horizontal flow field (constant head imposed to the left and right boundaries of the model). The physical model was built with glass beads of two sizes, forming a homogeneous low hydraulic conductivity matrix with sub-horizontal high conductivity channels embedded. The inverse problem pursued the identification of the hydraulic conductivity from measurements of the solute concentration at given locations and times. Prior field realizations were generated using multiple-point geostatistics to resemble the channel patterns observed on the physical model. The efficiency and accuracy of both techniques in terms of computational time and error/dispersion in hydraulic conductivity and solute concentration are evaluated.
DeepAngle uses machine learning to determine contact angles between different phases in the tomography images of porous materials. The measurement of these angles in 3D can be inaccurate and time-consuming due to the discretized space of image voxels. A computationally intensive solution involves fitting and vectorizing all surfaces using an adaptable grid to measure angles between the desired vectorized planes. However, the present study offers an alternative low-cost technique that utilizes deep learning to estimate interfacial angles directly from images. DeepAngle was tested on synthetic and realistic images and was found to improve the r-squared of predicted angles by 5 to 16%, while reducing computational costs by 20 times. This rapid method is particularly useful for processing large tomography data and time-resolved images that are computationally intensive. The developed code and the dataset are available in a public repository on GitHub at [https://www.github.com/ArashRabbani/DeepAngle].
Note: An extended version of this poster has been accepted for publication by in Journal of Geoenergy Science and Engineering.
For the performance assessment of nuclear waste repositories, a thorough analysis of the uncertainty and sensitivity of the underlying processes is necessary. Whereas a detailed experimental investigation of the final repository site is infeasible due to numerous reasons, the verification and validation of the numerical tools under realistic conditions using experimental data of underground research laboratories are all the more important.
One of such experiment is the FE-experiment at the URL site in Mt. Terri - a full-scale multiple heater test in the Opalinus clay in Switzerland which simulates "as realistically as possible, the construction, waste emplacement, backfilling and early post-closure evolution of a spent fuel/vitrified high-level waste disposal tunnel according to the Swiss repository concept." [H. Müller et al. 2017]
In our contribution, we present an application of design-of-experiment-based history matching as an approach to reduce and investigate parameter uncertainties in finite-element models for repositories of high-level radioactive waste [Buchwald 2020]. We combine experimental data from the FE-experiment at the Mt. Terri site in Switzerland with thermo-hydro-mechanical modeling using the open-source package OpenGeoSys. The parameter space was reduced by an initial parameter screening to find heavy hitters and an experiment-matching procedure using Monte-Carlo sampling on a Gaussian proxy model to fit modeling responses. The resulting parameter bounds were used in a subsequent uncertainty quantification and global sensitivity analysis based on the proxy model demonstrating the impact of parameter sensitivities.
The increase in plastic production is expected to exacerbate plastic waste disposal in terrestrial ecosystems. Soil represents a large reservoir for plastic wastes. Once disposed into the soil, plastic wastes interact with soil particles and biota and affect chemical, physical, and biological processes in soil (Jannesarahmadi et al., 2023). Microplastics (MPs) with distinct thermal and radiative properties and filling characteristics can alter energy partitioning over the surface of drying porous media and thus subsurface thermal regimes. The present study aims to quantify impacts of MPs on latent heat loss and temperature dynamics in drying sandy media. We conducted a series of evaporation experiments on sand columns (height: 20 cm – diameter: 8 cm) with grain size ranging from 0.4 to 0.8 mm and density of 2.65 g/cm³. Two types of microplastics with different characteristics and concentrations were used: Polyethylene (PE) with 34 to 50 μm particles and density of 0.94 g/cm³ and Polyvinylchloride (PVC) with particles ranging from 80 to 200 μm and density of 1.4 g/cm³. Mass loss rates from sand samples with different concentrations of MPs (i.e., 0.5, 2, and 5%) were compared with drying rates of the sand column without MPs serving as a reference. An array of thermocouples continuously measured vertical temperature profile in drying sand columns subjected to different wind and radiative boundary conditions. Airflow was generated by an adjustable fan and shortwave radiation flux was mimicked using halogen lamps with different intensities. Our preliminary results show that the presence of MPs with different characteristics alter evaporative loss and vertical temperature profiles in drying sand samples with the surface accumulation of PE particles (with lower density relative to water) influencing the thermal and radiative properties at the surface of drying porous media. The study provides new insights into the impact of MPs on energy partitioning dynamics over drying terrestrial surfaces and subsurface thermal regimes that could potentially affect various hydrological and biological processes in soil.
Hypothesis: in-situ recovery is an alternative to conventional mining, relying on the application of an electric potential to enhance the subsurface flow of ions. The governing physics of electrokinetic transport are electromigration and electroosmotic flow, which depend on the electric potential and excess charge adhered to mineral surfaces, respectively. Hence, mineral occurrence and its associated zeta potential should be the governing parameters that affect the efficacy of EK-ISR. Theory and Simulations: The governing model includes three coupled equations: (1) Poisson equation, (2) Nernst--Planck equation, and (3) Navier--Stokes equation. These equations were solved using the lattice Boltzmann method within X-ray computed microtomography images. The effects of mineral occurrence, zeta potential, and electric potential on a complex 3-dimensional synthetic iron ore were evaluated. Findings: Although the positive zeta potential of chalcopyrite can induce a flow counter to the direction of electromigration, the net effect is dependent on the occurrence of chalcopyrite. However, the ion flux induced by electromigration was the dominant transport mechanism, whereas electroosmosis made a lower contribution. Overall, Electrokinetic in-situ recovery is a promising technique that can be controlled because the dominant ion transport mechanisms are electromigration and diffusion. The former term depends on the applied external electric potential, and the latter term depends on the lixiviant injection.
Understanding reactive solute transport in natural media is critical for many applications (e.g., groundwater remediation, carbon storage, and enhanced oil recovery). It has already been confirmed that solute mixing can be significantly enhanced when decreasing the saturation, which ultimately increases effective reactivity. Most studies have been conducted in steady state conditions, i.e., constant flow rate and immobile immiscible phase (e.g., gas or oil) within the pore space. However, in a dynamic multiphase flow system, the motion of the immiscible phase constantly alters the effective flow paths and increases the complexity of the flow field. The impacts of dynamic multiphase flow on reactive solute transport remain an open question. To this end, we build up a quasi-2D porous medium using a 3D printing technique. The new device allows the injection of the reactants together with a steady multiphase flow. We directly evaluate the evolution of a mixing limited reaction by capturing the light emission from an optimized chemiluminescence reaction. Direct numerical simulations are used to infer the velocity field within the liquid phase. In steady state conditions, after an initial increase, the effective reaction rate decreases monotonically. However, while multiphase flow enhances mixing, the effective reaction rate fluctuates in time. Immiscible phase displacements suddenly put two reactants in contact, changing dramatically the local reaction rates in space and time.
Even though geologic carbon storage could reduce carbon emissions to the atmosphere and mitigate the impact of climate change, there are potential seismic risks and uncertainties associated with a GCS operation. Hence, we need to understand this process better before it becomes a reliable technology. However, the system of partial differential equations used to describe an induced seismicity event is highly nonlinear. Subsequently, we need substantial computational resources to approximate this system, making this process unsuitable for handling large-scale uncertainty quantification in which an extensive set of simulations must be explored. Kadeethum et al. [1] have illustrated the use of reduced order modeling (ROM) to enhance full order modeling (FOM) solvers. In this study, we apply the framework to an induced seismicity high-fidelity solver (coupled hydro-mechanical (HM) processes) proposed by Chang et al. [2]. Our goal is to investigate the improvements of ROM-assisted FOM performance with emphasis on (1) computational cost reduction and (2) a convergence rate. Our systematic approach is:
1. Substitute the high-fidelity hydro (H-FOM) solver with the low-fidelity hydro (H-ROM) model but still use high-fidelity mechanics (M-FOM) solver: H-ROM-M-FOM.
2. Substitute the M-FOM solver with low-fidelity mechanics (M-ROM) model: H-FOM-M-ROM.
3. Use HM-FOM with HM-ROM initialization.
SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525.
[1] Kadeethum, T., O’Malley, D., Ballarin, F., Ang, I., Fuhg, J. N., Bouklas, N., ... & Yoon, H. (2022). Enhancing high-fidelity nonlinear solver with reduced order model. Scientific Reports, 12, 20229.
[2] Chang, K. W., Yoon, H., & Martinez, M. J. (2022). Potential seismicity along basement faults induced by geological carbon sequestration. Geophysical Research Letters, 49(13), e2022GL098721.
Phase change materials (PCMs) can store and release heat at a relatively constant temperature. Incorporating PCMs into mixtures for backfill materials may improve their thermal energy density and thus contribute to the enhancement of borehole thermal energy storage and shallow geothermal energy systems. However, PCMs might reduce the overall heat transfer between the borehole and the surrounding ground because of their low thermal conductivity. Thus, other additives may be needed to improve the effective thermal conductivity of the backfill-ground system, while maintaining desirable heat capacity and then the corresponding efficiency of the shallow geothermal systems. This study incorporates encapsulated PCMs (EPCMs) and graphite into glass fines with the purpose of using the mixtures as backfill materials. The heat capacity and thermal conductivity of mixtures with different content of each component are measured in the laboratory. In addition, a computed tomography slice, which contains EPCMs, graphite, glass, and air phases, is selected to study the heat transfer at the particle scale. The experimental data agree with the simulated results. The findings in this study can be used in geothermal system design and contribute to the transition from fossil-based energy systems to renewable energy sources.
Large-scale fracture network and fracturing fluid retention in shale reservoir will inevitably affect the stress sensitivity. We focused on different areas of shale gas flow, and design several stress sensitivity tests of matrix, unsupported fracture, supported fracture and water-bearing fracture cores. The influence of different types of fractures and fracture water content on permeability is clarified. Further more, a characterization model describing the change of permeability with stress is established. The result shows: The stress sensitivity of different areas varies greatly. The loss of conductivity of support fracture is small under high stress conditions. But the loss rate of unsupported fracture permeability under the effect of effective stress is up to 97%, and the permeability recovery rate after stress recovery is less than 20%. Permeability decreases by 3-4 orders of magnitude after micro-fracture water cut, which is more sensitive to stress change. The research support the optimization of fluid filling intensity and flowback system of shale gas wells.
The mechanical properties of highly porous materials are expressed by fabric tensors. But the structure of highly porous materials is complicated if it is not arranged periodically. In micro-mechanical schemes or homogenization, the Eshelby tensor is widely used to estimate porous materials' mechanical properties. However, the relationship between these two tensors is not known. The result will be given by the equation, but some experiments are required to obtain numerical results.
Considerable experimental evidence in a variety of porous materials (concrete, ceramics, bones, rocks) that undergo infinitesimal deformations makes it clear that material moduli depend on the density (identified as density dependence of Young's modulus).
The phenomean of material damage is a consequence of the inhomogeneity of the body as material properties detieriorate with deformation.
This is especially relevant for subsurface flows through geological media which have experienced deformation over millenia.
However, the most widely used models for subsurface flow (such as Darcy/Brinkman model) typically assume that the porous medium is a rigid skeleton. Even if one moves onto theories of poroelasticity, the porous solid is assumed to be a linearized elastic solid with constant values of material properties such as Young's modulus.
Within the framework of linearized elasticity, modelling damaged porous media with a density dependent Young's modulus is inconsistent and untenable due to the assumption of infinitesimal displacement gradient. However, it is possible to incorporate density-dependent material moduli in the framework of implicit constitutive theories where the kinematic variable (strain) is expressed as a function of the stress.
Such a viewpoint remains unexplored in the context of flow of a fluid through porous media, and we shall
illustrate the effect of damage by comparison with existing theories that ignore it completely.
In the work of Lundström et al [1], a new concept of stormwater storage in sponge-like porous bodies (SPBs) is suggested: down-flow and up-flow SPB storage. The analytical and numerical results of the analysis based on the first principles argue that the studied up-flow model can capture and control the stormwater runoff for various conditions of Swedish design rainfalls. In the present research study additional work on the existing model is carried out.
The model presented [1] consists of a solid cylinder (radius δ) surrounded by an inner and outer porous annulus (radii a and b respectively, where b>a). The inner and outer porous media is made of thin vertical cylindrical fiber rods with dimensions Ri,Ro << a, b, δ, where Ri and Ro represent their respective radii. For such a model of the water uptake, the governing equation is the Darcy Law, and the flow is mainly driven by the capillary action (∆p ∝ 1/R). Further advances of the model are presented in this research including the diffusion of water into the dry soil, for which the diffuse-front modeling, as done by Zarandi and Pillai [2] (Richard’s equation), is applied. The corresponding set of equations for the motion of liquid fronts for each of the channels with the corresponding boundary conditions are given and the pressure quantities are averaged over the cross-section. Numerical integration is carried out in MATLAB. The diffuse-front model is resolved with COMSOL Software using Porous Media Flow Module. The numerical simulation results will be validated against the experimental measurements planned on a physical up-flow model in the laboratory setting. Similarly, to the work of Lundström et al [1], the model storage inflow rates and volume absorption will be plotted against time and compared to the Swedish design rainfall data.
References:
[1] Lundström, T. Staffan, et al. "Dynamic distributed storage of stormwater in sponge-like porous bodies: Modelling water uptake." Water 12.8 (2020): 2080.
[2] Zarandi, Amin, and Krishna M. Pillai. "Application of Sharp-and Diffuse-Front Models for Predicting Mass Gain and Saturation in Fibrous Wicks." (2018).
In underground hydrogen storage, pore pressure cycling due to annual or more frequent gas production (depletion) and injection leads to changes in the stresses acting on the reservoir rock, which in turn lead to rock deformation. Although inelastic deformation has an important effect on the physical properties of the rock, its effect on rock mechanical and transport properties is not well understood. To investigate the effect of inelastic strain, triaxial cyclic axial compression experiments were carried out on Castlegate and St Bees sandstone samples with 26% and 20% porosity, respectively. This was done using the Harpers THMC Flow Bench at Heriot-Watt University at 4.5 MPa, 10.5 MPa, and 19.5 MPa confining pressure. Permeability tests were carried out at key differential stress points throughout the compression tests. 3D images of the whole specimens before and after the cyclic loading experiments were obtained by performing X-ray micro-computed tomography scans, and digital core models were established to quantitatively characterize the geometric topological features of the two sandstones. The results show that the total axial strains of the two sandstones after cyclic loading ranges from 0.98–1.42% and 0.82–1.17%, respectively. The more porous Castlegate sandstone shows greater inelastic strain than the St Bees sandstone (0.47 to 0.85%, compared to 0.23–0.62%, respectively). However, upon stress changes, the Castlegate permeability shows lower permeability loss (19-46%) compared to St Bees Sandstone (~70%). In both sandstone samples, the first cyclic loading event produced the most significant inelastic strain and therefore permeability loss. Microstructural evidence based on CT analysis indicates that inelastic compaction in the Castlegate sandstone is controlled by a combination of intergranular cracking and intergranular slip, with the former dominating. Some of the large pores were compacted to form smaller pores due to intergranular slip, causing a decrease in the permeability of this sandstone, and the inelastic compaction became more pronounced as the confining pressure increased. In contrast, for the St Bees sandstone, inelastic compaction is mainly controlled by intergranular and intra/transgranular cracking. In addition, broken grains in the pores and throats were responsible for the decrease in permeability.
In the current times, the protection of the environment is becoming more and more important in all sectors. This also includes agriculture, which has to overcome particularly large hurdles in this respect, since on the one hand the world population is steadily increasing and thus more people have to be fed, but on the other hand the usable land is limited. Therefore, the yield must be increased, which is mainly done through the excessive use of mineral fertilizers, which are extremely energy-intensive to produce and are harmful to the environment. Therefore, natural, living fertilizers are searched that form a mutualism with crop plants. Thus, the problems of over-fertilization can be overcome. Cyanobacteria, ubiquitous phototrophic prokaryotes, are a possible source of biological fertilizer, mainly because of their ability to fix elemental nitrogen from the atmosphere and to release it in a usable form into the environment. Among other organisms, cyanobacteria are able to enter into symbiosis with plants, whereby not only nitrogen but also other nutrients or growth-promoting substances can be exchanged. Furthermore, cyanobacterial biofilms contribute to an improvement of the soil condition. By producing extracellular polymeric substances, which consist largely of polysaccharides, it can positively influence both soil aggregation and soil water retention and thus reduce soil erosion. In addition, the biofilm can also change the nutrient composition or availability in soils. Cyanobacteria thus represent a promising environmentally friendly alternative to traditional fertilizers.
Wheat is one of the most important food grains in the world, so this work investigates the co-cultivation of common wheat (Triticum aestivum) and cyanobacteria. Diazotrophic strains isolated from the temperate zone are used as cyanobacteria to investigate the effect of nitrogen fertilization by cyanobacteria on the growth of wheat. In addition, the influence on nutrient availability will be investigated by analyzing the pore water. Furthermore, it will be determined whether the use of cyanobacteria can lead to an increase in water retention in the soil. All experiments are conducted in typical agricultural soils for a complete growing season of wheat.
Nuclear energy will play a key role in the UK’s strategy to achieve net zero carbon by 2050. However, the high cost and intergenerational burden of decommissioning and waste management remains high and there is a need to reduce the costs of decommissioning and clean-up. Nuclear site decommissioning involves the retrieval and handling of various radioactive waste forms. Removal of particulate wastes, such as contaminated concrete and soils, represents a potential hazard in terms of radiation exposure for the workforce and the surrounding environment. This may be due to the accidental release of airborne or groundwater-borne radioactive particulates during waste recovery and transport, or to the loss of radioactive debris upon retrieval. The development of innovative techniques to reduce hazard in decommissioning operations is therefore a critical aspect of site decommissioning.
This study explores the suitability of colloidal silica, in combination with in-situ electrokinetics, to remediate contaminated soils by promoting migration of radionuclides into grouted soil volumes, prior to their removal. Colloidal silica is an aqueous suspension of silica (SiO2) nanoparticles, with average particle size <100 nm. The creation of siloxane bonds (Si – O – Si), typically triggered by the addition of an electrolyte accelerator, leads to the formation of a solid-like network of silica nanoparticles in the form of a hydrogel. Previous work on colloidal silica gel has proved its potential to form low-permeability hydraulic barriers against fluid migration, and to inhibit the diffusion of radionuclides through the gel, making it a promising material for use in retrieval operations.
Here we present research to determine the potential for electrokinetics, in combination with colloidal silica grouting, as a low energy remediation technique for radioactively contaminated soils. Experiments were carried out using electric field gradients ≤ 1 V/cm, to satisfy the low-energy requirements that make electrokinetic remediation advantageous over other remediation methods. The effect of i) applied voltage and ii) groundwater chemistry on the mobility of two types of radionuclides, namely Cs and Sr, was assessed. These small-scale laboratory experiments demonstrate that electrokinetics can be used to mobilise radionuclides (Cs and Sr) within the ground and trap them within a relatively small volume of grouted soil that can be readily removed. As well as inhibiting groundwater flow, and thus advective migration of radionuclides in the soil, the grout also increases the sorption capacity in the ground, reduces the risk of airborne migration of radioactive particulates during excavation, and can be readily incorporated into cementitious or vitrified wasteforms.
Redox flow batteries are a promising option for large-scale energy storage, but their stringent cost requirements hinder widespread deployment. One option to increase cost competitiveness is by improving the power density of the electrochemical cell by enhancing the performance of the porous electrode microstructure, which determines the available surface area for electrochemical reactions, electrolyte transport, and fluid pressure drop [1]. Conventional porous electrodes are fibrous mats assembled in coherent structures repurposed from fuel cell gas diffusion electrodes [2]. And, while functional, these materials have not been tailored to sustain the requirements of liquid-phase electrochemistry. Hence, new manufacturing techniques need to be developed affording a higher control over the electrode microstructure and resulting properties. Additive manufacturing, or 3D printing, is an emerging approach to manufacture controlled and deterministic architectures, enabling the tuning of electrochemical performance and hydraulic resistance [3].
In this study, we manufacture model grid structures using stereolithography 3D printing followed by carbonization (Figure 1a) and explore their application in redox flow batteries. We employ microscopy, tomography, spectroscopy, fluid dynamics, and electrochemical diagnostics to investigate the impact of the electrode structure on the fluid and mass transport of ordered lattice structures in non-aqueous redox flow cells. We investigate the influence of the flow field, printing orientation, and pillar geometry on mass transport (Figure 1b). We elucidate correlations between the electrode structure and performance metrics including pressure drop, surface area, and mass transfer correlations. We find that the printing orientation influences the electrode performance through a change in electrode morphology caused by surface roughness and resin spreading, impacting the shrinking direction after carbonization, internal surface area, and therefore the charge transfer, mass transfer, and hydraulic resistances. Moreover, mass transfer rates within the electrode are enhanced by using an interdigitated flow field or by altering the pillar shape to a helical or triangular design, which could improve mixing. Compared to commercial carbon-fiber electrodes, the pressure drop is significantly reduced (Figure 1c) because of the larger pore sizes (~500 μm for the 3D printed electrode vs. 2-100 μm for the Freudenberg H23 paper electrode and 2-300 μm for the ELAT Cloth electrode). Whereas the commercial electrodes feature a superior internal surface area, their area-normalized mass transfer coefficients are lower compared to the printed electrodes (Figure 1d). Going forward, the use of additive manufacturing enabling finer features combined with carbonization at elevated temperatures can be utilized to manufacture multiscale electrodes concurrently providing excellent electrochemical performance and low hydraulic resistance. Combining additive manufacturing with emerging computational topology optimization approaches could enable the bottom-up design of advanced electrode materials for electrochemical devices [4].
With a share of about 8.6 % of anthropogenic CO2 emissions, concrete is a major contributor to global warming. Microbiologically induced calcium carbonate precipitation (MICP) offers the potential of a more sustainable alternative. During MICP calcium carbonate is formed by microbiological activity and can serve as a binder between mineral particles. This calcium carbonate can be an alternative binder to conventional cement mortar used for concrete. The most commonly used mechanism for MICP is ureolysis. In this process, urea is enzymatically degraded to ammonium and carbonate. In the presence of calcium ions and in an alkaline environment, calcium carbonate precipitates. Since the compressive strength after MICP is related to the amount of precipitated calcium carbonate, multiple cycles of treatment with cell suspension and calcination solution are necessary if high compressive strengths need to be achieved. It is therefore of interest to improve treatment times by obtaining knowledge of the reaction speed of ureolysis. Various studies have investigated the rate and kinetics of MICP regarding the concentration of cells, urea and calcium ions. However, only low concentrations of calcium ions (up to 500 mM) and cells (up to OD600 1) have been investigated so far. In order to obtain insight into the efficiency of MICP under conditions during the production of biocement, this study investigated MICP for calcium and urea concentrations up to 1391 and 1492 mM, and cell concentrations with an OD600 up to 10. It was shown that the rate of MICP continuously decreases with the addition of calcium ions. Furthermore, it could be observed that under these conditions the free calcium ions are degraded by formation of calcium carbonate within a few hours. During this time Sporosarcina pasteurii cells encapsulate in calcium carbonate while still maintaining ureolytic activity. Depending on the parameters reaction times under 3 hours were achieved which is significantly shorter than the reaction time of 24 hours often used in literature protocols for MICP treatment. Therefore, these findings make it possible to determine an optimum reaction time for the production of biocement depending on the cell concentration and the composition of the calcination solution used during MICP. Based on these results silica sands with different particle size distributions were treated with several injection methods under these optimised conditions. The resulting samples were scanned by micro computed tomography. Contact points and pore space depending on various parameters during the treatment of MICP were investigated. These findings can be used to further investigate if a reduced reaction time between cycles is applicable concerning compressive strength and homogeneity of the samples.
We focus on the evaluation of the adsorption energy (EAd) of 24 Iodinated contrast media agents (ICMs) on activated carbon through Density Functional Theory (DFT) in silico simulations. The study is motivated by the emergence of concerns related to the impact of pharmaceuticals on the environment and human health [1,2]. Iodinated contrast media agents are typically used in radiology, primarily in CT scans for soft tissue imaging [3]. Their presence in the environment can be the source of hazard [4]. With an annual global consumption of 3.5 × 106 kg [5], ICMs have been frequently detected in surface water bodies as well as in groundwater systems with levels up to 100 µg/L [3]. While conventional water treatment technologies are unable to contain the release of ICMs into the environment, some studies document the possibility that toxic by-products resulting from the transformations of ICMs can affect water and soil systems [3,6]. Thus, there is a growing need for robust technologies for the removal of ICMs from aquatic environments. Most of the available studies focus only on a few selected ICMs such as diatrizoic acid, iopamidol, or iohexol [3]. Considering that there are more than 30 ICMs available commercially [7], our study is the first one providing a comprehensive analysis of a variety of these compounds. Due to its tunable physiochemical properties, activated carbon is a porous material of remarkable interest in the context of groundwater remediation practice [8]. Because of the complexity and uncertainty associated with the structure of activated carbon, we use monolayer graphene as a proxy model [9,10]. We find that overall strong adsorption energies (EAd) can be documented through our DFT studies. These range from -114 Kcal/mol for Iophendylate to -13 Kcal/mol for Methiodal. To enhance our knowledge about the fundamental mechanisms underpinning the adsorption of ICMs on activated carbon, we rely on a Quantitative structure-activity relationship (QSAR) regression modeling approach. The latter yields quantitative correlation between (a) chemical structure information which is, in turn, represented in terms of molecular descriptors, and (b) adsorption [11]. Our results suggest that descriptors such as the topological charge index and the Van Der Waals surface area (of aromatic atoms) are positively corelated to adsorption energy. Otherwise, other descriptors, such as, e.g., the ‘average Randic-like index from Burden matrix weighted by ionization potential’, are characterized by a negative correlation to EAd. Our results are intended as a first step to assist the assessment of the role of intermolecular interactions governing adsorption of ICMs on activated carbon surfaces and to enhance our ability to further improve (and possibly design) a new generation of porous media to be effectively employed as sorbents in this context.
As countries around the world are trying to transition away from fossil fuels to renewable energy sources, short- and long-term storage of an increasing, yet unsteady, renewable energy supply becomes a major challenge. Further, as provision of heat is a major part of industrialized countries’ energy needs, storing heat energy, in applications such as the capturing of excess heat from industrial processes or concentrated solar power plants, has the potential for great increase in energy efficiency.
Among the available heat storage technologies, thermochemical heat storage provides a large energy capacity for short- and long-term storage. To further develop the technology, DLR is developing models and simulations as well as experimental characterization methods for thermochemical heat storage. More specifically, storage in the CaO/Ca(OH)2-System is investigated because of the low price and environmental friendliness of the reactants.
However, a major challenge to modelling such systems, is the restructuring of the powder bed during repeated cycling, i.e., repeated charging and discharging of the reactor. This happens through mechanical and chemical alteration of the powder bed. The three dominant effects are, the compaction of the bed from the gas flow, the expansion/shrinkage of the powder particles through water uptake/release and the agglomeration of powder particles, where bonds between the particles form, solidifying the bed.
To model the compaction and solidification of the powder bed during cycling, we present an elasto-plastic mechanical model based on the Drucker-Prager-Cap yield surface, which has been used previously for powder compaction, see e.g. [1]. The changes in the powder bed during cycling are modeled by hardening mechanisms, i.e., a changing yield surface, corresponding to powder compaction and agglomeration, respectively. While the exact mechanism of the agglomeration is yet unknown, it can be characterized by mechanical measurements.
Then, the plastic model is coupled to a reactor model, simulating the heat and mass transport, as well as the thermochemical reaction using a model, similar to [2]. This enables the study of the powder bed dynamics under different boundary conditions during cycling, such as pressure drop, water vapor fraction and reactor geometry.
In this contribution, we will present a parameterization of the model based on experimental data, that was obtained from a test reactor, and the parameterization of the mechanical model, i.e. the plastic yield surface, is done via flow tester experiments.
Then, we will show simulation results with an emphasis on investigating the irreversible effects of continuous cycling on the powder bed. This includes the compaction of the powder bed during the pressurization of the reactor, the possible emergence of hysteresis effects in the deformation of the powder bed under repeated cycling, as well as degradation through irreversible structural changes, such as powder agglomeration.
Despite being major contributors to global CO$_{2}$ emissions, fossil feedstocks are finite natural resources frequently used to produce high value goods including fuels and plastics. One alternative is to replace fossil feedstocks with renewable agricultural feedstocks due to their ability to sequester carbon during growth. While a promising alternative, the use of food crops as feedstocks brings its own set of challenges. Recent emphasis has been placed on deconstruction of agricultural residues, such as corn stover, into fuels and chemicals. Polysaccharides from lignocellulosic plant cell walls can be converted to glucose, but biomass recalcitrance to enzymatic hydrolysis presents a practical challenge to this pathway. Pretreatment steps help improve enzymatic access to plant cell walls and once optimized, allow for these processes to be scaled. Nuclear magnetic resonance (NMR) relaxometry is applied to corn stover to gain a better understanding of these systems and the impacts of pretreatment. These measurements directly measure water adsorption in anatomical fractions of corn stover. NMR transverse T$_{2}$ relaxation time distribution measurements indicate multiple water populations, which vary with anatomical fraction and water adsorption. Measured T$_{2}$ data are used to calculate thermodynamic properties of Brunauer-Emmet-Teller (BET) adsorption theory using a model to estimate mono and bilayer relaxation. T$_{2}$ data are used directly to determine rotational diffusion correlation times indicating adsorption interaction strength. T$_{1}-$T$_{2}$ longitudinal-transverse relaxation time correlation measurements quantify differences in the molecular level structural order of the adsorbate surface water as a function of water activity, i.e. relative humidity or water vapor partial pressure. The T$_{1}$/T$_{2}$ ratio provides a measure of the surface energy related to the adsorption strength and surface diffusive mobility of the water adsorbate, and differentiates the anatomical fractions. The results indicate that direct measurement of NMR relaxation times can be used to characterize corn stover biomass water adsorption, which are data relevant to biomass processing and handling. These procedures may be extended to pretreated lignocellulosic materials to study how morphological changes impact adsorption, and applied to monitor enzymatic hydrolysis progress in situ.
Figure 1. Corn Stalk MRI. A 1 mm thick transverse slice taken of a hydrated corn stalk with a 25x25 mm field of view over 128x128 pixels for a resolution of 195 µm/pixel.
In the context of climate change, studies of water transfer in bio-sourced materials are becoming essential in order to meet the multiple challenges of developing high-performance materials over the long term and preserving resources while limiting greenhouse gas emissions. One of the keys to supporting these studies is access to the "water status" in these porous materials [1]. However, the monitoring of their performance over time is often limited to destructive studies or relies on techniques that are too local or average, or even intrusive.
Here, we present the development of an innovative methodology based on nuclear magnetic resonance (NMR) relaxometry and imaging (MRI) at low and high magnetic fields, respectively, to study water content and transport during water stress in living materials, from the leaf to the whole plant scale. The results obtained by the combination of these approaches will be compared with water transfers in model porous materials such as wood, cellulose, glass beads, etc. Indeed, thanks to NMR approach, it has been possible to identify the drying mechanisms in wood and to show that bound water plays a fundamental role, transporting, by diffusion, free water from the interior of the material to the free surface, and this during all the drying phase [2]. Another interesting result concerns the behavior of NMR signal and relaxation times in two contrasted genotypes in term of cellulosic ratio. This dependence of relaxometry provided an important information on molecular dynamics directly related to plant resistance [3].
This study has demonstrated the potential and the versatility of NMR relaxation as a means to characterize the microstructure of living porous material and model water transport mechanisms under different environmental constraints.
Controllable but realistic representation of subsurface gas storage scenarios are needed to assess the impacts that microbial processes could have on behaviour of stored gas in hydrogen storage. Current experimental approaches may misrepresent the type and magnitude of microbial activity as they do not consider the limited residual water available as habitats for microorganisms. Understanding microbial processes is necessary as microorganisms could consume and produce gases (including converting hydrogen to corrosive hydrogen sulphide or to methane) or block flow pathways with biomass or precipitates. These processes can only be properly assessed if the gas filled, residual water state can be recreated in the laboratory.
We have developed an experimental set-up that allows us to study microbial processes in both saturated and unsaturated conditions and change between them to represent fluid movement occurring at the fringes of stored gas during storage cycles. The fringes of stored gas have been identified as a potential hotspot for microbial processes.
The apparatus comprises two 1000 ml syringe pumps attached to a core sample held in a pressurised vessel. The system can be operated at pressures of up to 130 -500 Bar (depending on configuration) which is representative of pressures expected during hydrogen storage cycles. The sample can be heated up to at least 90 °C to cover the expected activity range of subsurface microorganisms. The system allows the core to be saturated from the base of the sample by flowing a suitable groundwater mimicking the environment in a saline aquifer prior to gas storage. Using the second pump, groundwater can be displaced from pore spaces by injecting hydrogen into the top of the core sample, leaving a residual volume of water more closely replicating the conditions that would occur in a storage reservoir. Continuous logging of flow, pressure and volumes allow the degree of saturation and flow properties to be calculated. Sampling ports allow collection of both gas and water samples, allowing microbial gas consumption and conversion and impact on water chemistry to be monitored.
The apparatus is currently undergoing proof-of-concept testing in which stimulation of methanogens within sandstone cores is being investigated. Although the system most closely resembles hydrogen storage in saline aquifers, it is also relevant to storage in depleted hydrocarbon fields and storage of carbon dioxide.
Geothermal energy is a renewable resource that may help to provide a green energy supply, although the low rock permeability at the required depth prevents an energy-efficient use of this resource. Enhanced Geothermal Systems (EGS) allow to increase permeability by means of the hydraulic stimulation of the fractures of underground formations. However, it implies risks as, for instance, induced seismicity or an early thermal exhaustion of the reservoir.
We perform numerical simulations to delve into the thermo-hydro-mechanical and frictional phenomena that control that risks in EGS projects. Our computational model solves the fully coupled equations of thermo-poro-elasticity, together with rate-and-state friction at faults. The methodology allows us to simulate both fluid-injection induced earthquakes and long-term behavior of geothermal reservoirs.
In the short them, we explore the optimization of the injection protocols to avoid or delay fault reactivation. Our results arise that the injection protocols with an early increase of flow rate could delay fault reactivation depending on the frictional properties of the contact. In the long term, we observe that the permeability stimulation can induce an early thermal decline of the reservoir. This may affect to energy production over years, depending on the increase of permeability achieved with hydraulic stimulation.
The methodology proposed in this work may help to improve the competitiveness of geothermal energy, as it can be a useful tool to manage the seismic risk and the long-term operation of geothermal reservoirs.
Acknowledgements
This research was supported by the “Agencia Estatal de Investigación” and “Ministerio de Ciencia de Investigación” (10.13039/501100011033) through grant “HydroPore” (PID2019-106887GB-C33). S.A. gratefully acknowledges funding from the Spanish “Ministerio de Ciencia, Innovación y Universidades” through “Programa de Formación del Profesorado Universitario FPU 2018”. D.S. thanks the financial support from the Comunidad de Madrid through the call Research Grants for Young Investigators from Universidad Politécnica de Madrid under grant APOYO-JOVENES-21-6YB2DD-127-N6ZTY3, RSIEIH project, research program V PRICIT.
In this presentation we are interested in operational applications and new numerical approaches for modeling the heterogeneous mucus bio-film of human lungs for the monitoring of cystic fibrosis (CF) therapies. At an operational level, we aim at predicting whether a therapy has a significant impact of the mucociliary clearance or not, that is to say predicting the ability of the respiratory mucus to be functional, i.e. to move together with the motion of the surrounding cells. By opposition, a non-functional mucus will not move sufficiently to clear the lung wall from allergens, toxic agents, viruses, bacteria and their residual products (DNA filaments and altered mucoid elements).
In this biological configuration, the mucus is itself a porous media made of Newtonian periciliary fluid (PCL) and highly concentrated mucins produced by the goblet cells, whose motion in the mucus will allow a mixture between the mucins and the PCL leading by reaction to a polymerized mucus with a particular rheology. Among the rheological features such as visco-elasticity, visco-plasticity, yield stress and shear-thinning, we focus on this last one which has been shown to be the dominant feature leading to non-functional mucus [3]. Moreover, the PCL is produced by the respiratory epithelium covering the lung membrane, another porous media that allows the transcytosis mechanism producing the PCL. Indeed, the PCL is not present or not working properly when the cystic fibrosis transmembrane conductance regulator protein CFTR presents a mutation responsible of CF.
The numerical simulation of such configurations has two main objectives. On the one hand, one can predict whether a mucus is functional or not, with respect to the rheological features that are measured from samples [4]. On the other hand, the numerical simulation allows to adjust the parameters of an upscaled model, including the mucus permeability and the tortuosity index that relates the effective diffusion and the molecular diffusion of chemical species by means of a power of the porosity.
The mucus mixing is modeled by $-div(2\mu(c,D)\, D(u))=f-\nabla p $, the non-Newtonian stationary Stokes equation, where $f$ is the driving force induced by the epithelial cell, $\mu$ is mucus viscosity, $D= (\nabla u+\nabla u^T)/2$ is the shear-rate of the velocity $u$, $p$ is the pressure, and the incompressibility is satisfied by $div (u)=0$. The mucin concentration $c(x,t)$ follows
$\displaystyle \frac{\partial c}{\partial t} + div(uc) - div(\sigma\nabla(c)) = 0$
and the shear-thinning rheology is driven by the relation
$\displaystyle\mu(c,D)= \mu_\infty + (\mu_0(c)-\mu_\infty) \left ( 1 + 2\beta(c)^2 |D|^2 \right )^{\frac{q(c)-2}{2}}$
which makes all these equations strongly coupled. We will show that the solutions of this system can be expressed by a Lagrangian formulation [1,2], called particle method, that the numerical result are compatible with clinical resume of the patients whose sputum rheology has been characterized [4], and that the upscaled tortuosity index can be carried out.
This work has been funded by French National Agency of Research, project MucoReaDy ANR-20-CE45-0022-01, and by Carnot Institute ISIFoR project MicroMineral P450902ISI.
Fractured sorptive geomaterials (FSG) are ubiquitous in geological systems such as coal, shale and chalk. The solid matrix of FSG can adsorb species in gas or liquid form, the process of which is often accompanied by the deformation and micro- structural alternation of the matrix. Such coupling is further obscured by the presence of fracture network, introducing complex fracture–matrix interactions. Predicting the hydromechanical properties of FSG is of particular importance for the production of coalbed methane (CBM) which requires the assessment of coal permeability under varying pressure and stress conditions. This study attempts to investigate the interplay between adsorption, deformation, and permeability evolution of coals. The novel concept of adsorption stress popularized in material science research is adopted here to construct a mechanistic theory describing sorption-induced deformation of coals. The constitutive theory is implemented in a finite element (FE) scheme and then adopted for describing coal matrix in a FE model of coal–fracture system. The model is calibrated for San Juan coals and applied to simulate a typical methane depletion test. It is observed that, depending on the competing effect between desorption-induced fracture opening and poroelastic compaction, the predicted permeability curve may be monotonically increasing (rising type) or decreasing (decline type), or may exhibit reduction first and then increase (rebound type) during gas depletion. Such competition is found to be controlled by the volume ratio, the permeability ratio, and the stiffness ratio between the matrix and the fracture elements. The prediction covers a wide range of permeability data obtained from laboratory tests and field observations.
Ion-exchange membranes(IEMs) have been widely used for desalinated and energy conversion processes. Since the IEMs determine the efficiency of the above process, it is necessary to develop them with impoved separation performance and durability. Novel composite-type anion- or cation- exchange membranes were prepared as follows; first, pore-filling of monomer mixtures (styrene/ divinylbenzene (DVB), vinylbenzylchloride(VBC/DVB)) and an initiator was done in commercial polytetrafluoroethylene(PTFE) porous films, respectively. Thermal polymerization was followed in high temperature oven for the formation of precursor membranes. Post-sulfonation was done with chlorosulfonic acid in methylene chloride to give -SO3H for the preparation of cation exchange membranes. Post-amination was performed in trimethylamine (TMA) in acetone to give -N+(CH3)3- for the preparation of anion exchange membranes. SEM analysis confirmed these membranes were successfully prepared.
The electrochemical properties of the resulting membranes - ion exchange capacity, electric resistance and water content - were studied in terms of the ratio of dope compositions of monomers (Styrene/DVB, VBC/DVB).The composite membranes showed excellent electrochemical properties – electric resistance, water content and IEC value - depending on the monomer dope compositions (Styrene/DVB ratio and VBC/DVB ratio). These membranes showed lower electric resistances, lower water contents and higher IECs than commercial membranes thanks to thin PTFE supports. These results showed our composite membranes could be applied to the desalinatings electro-dialysis and energy conversion process.
Quantitative characterization of pore structure and analysis of seepage characteristics of tight reservoir based on digital core and NMR
Meng Du1,2,3, Zhengming Yang*1,2,3, Weifeng Lv2,4, Xinliang Chen2,3, Wen Li2,4
1 University of Chinese Academy of Sciences, Beijing 100049, China;
2 Institute of Porous Flow & Fluid Mechanics, Chinese Academy of Sciences, Langfang 065007, China;
3 Research Institute of Petroleum Exploration & Development, PetroChina, Beijing 100083, China;
4 State Key Laboratory of Enhanced Oil Recovery, Beijing 100083, China
Abstract: The pore-fracture structure characterization and seepage characteristics analysis are the keys to the effective development of tight reservoirs. However, it is difficult to accurately characterize the pore and fracture structures of different scales by conventional methods, which makes it difficult to analyze the seepage characteristics. In this study, combined with CT scanning technology and advanced mathematical algorithms of AVIZO visualization software, a three-dimensional digital core of tight reservoirs was constructed, and the comprehensively quantitative characterization of microscopic pore-fracture structure from multiple dimensions was carried out. On this basis, nuclear magnetic resonance (NMR) centrifugation experiments were conducted to monitor the fluid migration dynamics in tight reservoirs, and mobile fluid migration characteristics were analyzed based on NMR T2 spectra. The results show that the average porosity of the reservoir in this area was 11.2%, and the average permeability was 1.573mD, which belongs to low porosity and low permeability tight reservoirs. The distribution of pore throats was mainly contiguous and isolated. The connected pores were mainly distributed in enriched bands, which was due to the interconnection of gas pores, intergranular pores, and dissolution sheet fractures, while the disconnected pores were mainly distributed in isolated form, which was related to the development of inter-gravel dissolved pores and matrix dissolved pores, and the contribution of pore connectivity to seepage was greater than that of pore scale. The pore radius in this area was mainly 4.31-32.17 μm, the throat radius was mainly 3.42-13.29 μm, and the pore and throat cross-sectional shapes were mostly triangular, meanwhile, the fracture types could be divided into 3 types according to the occurrence and opening, which were mainly high-angle structural fractures and vertical fractures, indicating that pore-fracture structure had strong heterogeneity and fractures could play a better role in the infiltration of oil and gas. The connectivity of pore throats in the tight matrix was poor, which made the imbibition exchange effect weak and prone to water sensitivity. Subsequently, the movable fluid saturation increases with the increase of permeability, and the fractures and micropores had less flow resistance and were more conducive to water flow compared with small pores. This case study provides new insights into the exploitation of similar tight reservoirs.
Key works: NMR; dual porous medium; CT scanning; cross scale; digital core; quantitative characterization of pores fractures; fluid mobility
Water in natural environments consists of many ions, which exert electric forces on each other. We discuss whether the coulombic effects are relevant in describing mixing and reaction processes in natural environments or laboratory experiments. A typical model for electric interactions in dilute aqueous solutions is the Nernst–Planck equation.
Using FEniCS (fenicsproject.org) and Reaktoro (reaktoro.org), we solve the Nernst–Planck transport and equilibrium reactions of the ionic species in water. By comparing numerical simulations to reaction-driven flow experiments performed in a Hele-Shaw cell, we show that the electric interactions between ions can be relevant in mixing and reaction processes. We further discuss the numerical techniques in solving the Nernst–Planck system. In microfluidic experiments considering the mixing of aqueous fluids or electrokinetic effects, the Nernst–Planck equation can be essential to describe fluids' behavior.
Over the past decade non-destructive imaging methods for materials have been increasingly advanced. Two most notable imaging methods include X-ray computed microtomography (μCT) that can image materials at sub-micron scale to millimeter scale resolutions and focused ion beam-scanning electron microscopy (FIB-SEM) that can image at a nano-meter scale. Hence, the segmentation of images obtained from different imaging techniques is a critical step towards quantitatively describing various features of geomaterials over a range of scales. In this work we evaluate various deep learning methods (e.g., U-Net, Attention U-Net, Efficient net, transformer, VGG16, ResNet, and MultiResUnet) to segment both μCT and FIB-SEM images. Four independent datasets including sandstone, carbonate chalks, and shale are evaluated. Each of these datasets is composed of three-dimensional image stacks and corresponding ground truth segmentation labels obtained using various traditional image processing techniques. Our preliminary results indicate that deep learning architectures can successfully be applied to the task of semantic segmentation for individual dataset with frequency weighted accuracy between 94% and 99% (on testing data) and can perform better than manual segmentation to recover the natural morphology of original images. However, performance is significantly deteriorated by ~ 10-30% in accuracy when mixed images from different imaging methods are used as training data. Here, we will demonstrate the improvement of semantic segmentation of multiple rock images from both μCT and FIB-SEM through transfer learning of transformers and other deep learning methods.
SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525.
Aiming at the changes of temperature field and velocity field caused by high temperature and high pressure fluid injected into reservoir porous medium during heavy oil thermal recovery process, based on Darcy's law, thecoupled equation of heat flow in porous media was established by the finite volume method. Based on the REV-scale porous media model, the effect of permeability and volume fraction of porous media and the injection pressure of hot fluid during thermal recovery on the heat flow coupling heat transfer process in porous media was studied. The results showed that the effect of increasing the permeability on the heat transfer of porous media was better than increasing the inlet pressure, increasing the solid volume fraction and increasing the thermal resistance, resulting in a decrease in heat transfer. Increasing the pressure of the injected hot fluid could increase the heat transfer rate.
Progress towards a net zero carbon economy involves subsurface activities, such as geothermal energy production and geological storage of carbon dioxide, hydrogen and radioactive waste, that disturb tectonic stresses in the Earth’s crust. Seismicity induced by such stress perturbations is associated with risk from damage due to ground motion, fluid leakage and pollution due to increased permeability, and the potential loss of public confidence. Safe operation of these activities therefore requires effective management to minimise induced seismicity. Failure in brittle, porous materials initiates when structural damage, in the form of smaller-scale fractures, localises along an emergent failure plane or 'fault' in a transition from stable crack growth to dynamic rupture. Due to the extremely rapid nature of this critical transition, the precise micro-mechanisms involved are poorly understood and difficult to capture. However, these mechanisms are crucial drivers for earthquakes, including induced seismicity, and other devastating phenomena.
Here we observe these micro-mechanisms directly by controlling the rate of micro-seismic events to slow down the transition in a unique triaxial deformation apparatus that combines acoustic monitoring with contemporaneous in-situ x-ray imaging of the microstructure. We compare the seismic signatures from this experiment with those from a sister experiment carried out under constant strain rate loading. The results [1, 2] provide the first integrated picture of how damage and associated micro-seismic events emerge and evolve together during localisation and failure and allow us to directly constrain the partition between seismic and aseismic deformation at the micro-scale.
The evolving damage imaged in the 3D x-ray volumes and local strain fields undergoes a breakdown sequence involving several stages: (i) self-organised exploration of candidate shear zones close to peak stress, (ii) spontaneous tensile failure of individual grains due to point loading and pore-emanating fractures within an emergent and localised shear zone, validating many inferences from acoustic emissions monitoring, (iii) formation of a proto-cataclasite due to grain rotation and fragmentation, highlighting both the control of grain size on failure and the relative importance of aseismic mechanisms such as crack rotation in accommodating bulk shear deformation. Dilation and shear strain remain strongly correlated both spatially and temporally throughout sample weakening, confirming the existence of a cohesive zone, but with crack damage distributed throughout the shear zone rather than concentrated solely in a breakdown zone at the propagating front of a discontinuity.
Contrary to common assumption, we find seismic amplitude is not correlated with local imaged strain. The seismic strain partition coefficient is very low overall and locally highly variable. Local strain is therefore predominantly aseismic, explained in part by grain/crack rotation along the emergent shear zone. Reactive loading to maintain a constant micro-seismic event rate increases the seismic b-value, decreases the maximum event magnitude, and reduces the seismic strain partition coefficient compared with loading under a constant strain rate. Adding event rate control to that of maximum recorded magnitude may therefore be more effective than the current ‘traffic light’ system (based on maximum magnitude alone) for managing the risk of induced seismicity.
In the study of solute transport in porous media, it is common to rely on the Advection Dispersion Equation (ADE) model to interpret effluent breakthrough curves (BTCs) post unidirectional tracer laboratory tests. However, this approach is not suitable for porous rocks, as these are characterised by transport processes that occur over a wide range of length- and temporal scales. To deliver better subsurface engineering solutions in complex porous media, whether it be for groundwater contaminant tracking, carbon capture and storage (CCS), or geothermal/petroleum resource extraction, a workflow that integrates an updated experimental approach, and a novel means of data processing is necessitated.
Here, we deploy a numerical optimisation routine to fit experimental BTC data measured on Bentheimer Sandstone (BS), Ketton Limestone (KL), and Edwards Brown Carbonate (EB), at different flowrates and published previously (Kurotori et al. 2020). Although we use the ADE model to fit the BTC data of BS (a homogeneous sandstone), the Multi-Rate Mass Transfer (MRMT) model was used to fit the BTC data of KL and EB, two highly heterogenous carbonates. The analysis includes the estimation of parameters’ uncertainty by Bayesian inference. These parameter values (and their uncertainties) are then used to evaluate the first four spatial moments of the internal concentration distribution, representing the temporal evolution of total mass (0th), centre of mass (1st), variance (2nd), skewness (3rd) and kurtosis (4th). Unique to this study, the predicted moments are compared to their experimental counterparts, which have been estimated from 4D solute concentration measurements obtained by Positron Emission Tomography (PET) imaging.
We demonstrate that PET can be used to precisely measure the spatial moments of the solute concentration and that these present unique features depending on the rock type. We show that for BS the spatial moments are insensitive to flow rate when plotted as a function of pore volumes injected (PVI). However, for the two carbonate rocks, they feature a flow rate dependency, due to the presence of microporosity and vugs, which introduce porous regions of virtually stagnant flow - where transport is largely dominated by diffusion. For the two carbonate samples, both 0th and 1st moment yield earlier breakthrough, and greater tailing of the solute mass with increasing flowrate. The 2nd moment takes much larger values for KL and EB than BS, indicating greater spreading of the tracer pulse and less mixing due to the larger contrasts in activity between the immobile and mobile zones. This is further exacerbated at higher flowrates. For BS, the 3rd and 4th moments prior to breakthrough take constant values at 0 and 3, respectively, indicating that the tracer plume is normally distributed. Yet, lower values are observed for the carbonates, reflecting an evolving skewness of the tracer plume during transport.
The risk of environmental pollution, particularly groundwater contamination, has increased over the last century as a result of the growth of industry. Light non-aqueous phase liquids (LNAPLs) are one of the most common contaminants and refined petroleum hydrocarbons (RPHs-diesel, gasoline, motor oil, etc.) are typical examples [1, 2]. Heterogeneity in the subsurface represents one of the main issues for LNAPLs remediation, conventional pump-and-treat method rarely exceeds 60% of efficiency [3]. However, some studies demonstrating the non-Newtonian shear-thinning behavior of foam in highly permeable porous media point to the promising potential of foam to improve remediation yields [4, 5].
The use of aqueous foam in environmental remediation (ER) was inspired by enhanced oil recovery (EOR), and it has already proven to be an excellent displacing fluid for in situ remediation of NAPLs [6 – 8]. However, contact with petroleum compounds tends to deteriorate the stability of foam significantly and thus it is a challenge for both foam applications [9, 10]. Many researchers are currently focused on strategies to enhance foam employing numerous additives: i.e. co-surfactants [11, 12], polymers [13, 14], and nanoparticles (NPs) [15, 16]. It is worth noting that all of these studies mainly address foam applications in EOR. The main objective of our work is to evaluate experimentally the stability of foam generated with two or more additives in the presence of RPHs, both in bulk and in porous media.
In order to implement the concept, two environmentally friendly surfactants (sodium dodecyl sulfate (SDS) and cocamidopropyl hydroxysultaine (CAHS)) were experimentally investigated for their foaming ability and stability in the presence of diesel oil using the bulk foam screening method. Stability of complex foam formulations including a combination of surfactants, polymers and NPs were then examined. Two one-dimensional columns packed with sand and coupled in series were used to (i) generate a fully developed foam flow, (ii) evaluate foam stability and recovery efficiency of diesel initially at residual saturation. Mass balance and differential pressure were measured during each injection experiment.
The bulk foam study demonstrated an apparent increase in foam stability in the presence of Diesel when complex foaming solutions were used. The mixture of SDS and CAHS (SC) at a ratio of 1:1 could improve 7.5 times the bulk foam stability in contrast to SDS alone. The presence of NPs enhanced the bulk foam stability up to a factor of 2.6 for the SDS alone and 1.2 for the SC. Among the three types of tested environmentally friendly polymers, xanthan gum (XG) showed the best stabilizing properties compared to carboxymethyl cellulose and guar gum, with increased stability by factors of 3.4, 2, and 1.3 times respectively. Concerning the performance of foam in porous media, complex foaming solutions ranked as SC+XG > SC+NPs > SC > SDS.
Further studies are ongoing to explain how the addition of NPs and polymers affects recovery. Nevertheless, advanced foam formulations clearly exhibited promising perspectives to develop an efficient remediation technique for highly permeable soils contaminated by RPHs.
Abstract: There is still more residual oil in heterogeneous reservoirs after water injection, and it is difficult to recover them by further water flooding. As a common gas flooding medium, CO2 can further improve the recovery of heterogeneous reservoirs after water flooding, so it is of great significance to study the laws of CO2 miscible flooding under different injection methods in heterogeneous reservoirs. In this study, according to the heterogeneity characteristics of Lunnan Oilfield in Tarim, China, a single 1 m double-layer long core model was designed and prepared, and CO2 miscible displacement experiments with four different injection methods were carried out. Through the comparison and analysis of the experimental data, the displacement effect of CO2 miscible flooding under different displacement methods is obtained, and the laws of CO2 miscible flooding under different injection methods in heterogeneous reservoirs are summarized. The research shows that: ① The oil displacement efficiency of different injection methods of CO2 miscible flooding in heterogeneous reservoirs from high to low is: CO2-water alternate injection, continuous CO2 flooding, periodic CO2 flooding, and CO2-hydrocarbon gas alternate injection. ② CO2 miscible flooding in heterogeneous reservoirs mainly relies on convective diffusion and miscible mass transfer to recover crude oil. ③ The effect of convective diffusion mainly depends on the plugging of the dominant channels in high-permeability areas and the control of injection-production pressure differential. The effect of miscible mass transfer mainly depends on the degree of displacement in the early stage and the strength of gas miscibility. ④ To improve the recovery efficiency of CO2 miscible flooding in heterogeneous reservoirs, on the one hand, gas channeling should be slowed down, dominant channels should be blocked, and displacement pressure differential should be improved. On the other hand, the miscibility of CO2 should be improved.
Keywords: Different injection methods; Heterogeneous reservoir; CO2 miscible flooding;
Plume deformation and mixing determines the effective reaction in porous media
chracterised by internal heterogeneous reaction. Via pore-scale simulations, we show the dynamic of a passive scalar injected in a packed bed consisting of a mixture of chemically inert and reactive spherical particles (adorbers), to mimic, e.g., the contaminat uptake by a fraction of grains in the soil matrix [1].
The scalar plume deformation is a consequence of the different mechanisms of transport characterising the transport of molecules in the proximal and remote pores relative to the adsorbers, diffusion and advection, respectively. The scaling laws governing stretching and broadening of isoscalars are quantified and discussed in relation to medium characteristics, such as the mean adsorbers' interparticle distance.
We show that a transition from diffusion- to advection- dominated macroscopic adsorption is determined by the amount of adsorbers within the medium, with diffusion and advection dominating at high and low fractions, respectively.
At high fractions the temporal evolution of the macroscopic adsorption scale as $\propto \sqrt{t}$. while at law fractions it follows $\propto t$. The transition shows that more rapid adsorption is taking places in areas of soils where the fraction of adsorbers is lower, leading also to a faster saturation of contaminant uptake capacity.
With the recent COVID-19 pandemic, different viruses in the environment and their disastrous effects has drawn the sharp attention of researchers and scientists worldwide. In recent years, the release of viruses has been a serious concern around the globe.
This inspired us to investigate more about what happens once a virus is released. Deciphering how various virus species act in a system of environmental mobility will be extremely interesting. This will enable us to understand and forecast their fate and transport behavior in various subsurface environments.
The release and migration of viruses in aqueous environments is a primary focus of this investigation. A short examination of the literature reveals that the majority of articles focus on representative viral species, such as bacteriophage MS2 and PhiX174. This might be an earlier made hypothesis on common viral behavior. Here, it is assumed that model viruses show similar transport and retention phenomena as all hazardous viruses. This might be due to various regulatory constraints and challenges that come into the picture when working with viruses that are lethal to humans. We assume that different viral species may behave differently based on their surface chemistry and physical morphology. Which has not been accounted distinctively in the literature, as most of them use model virus strains. Therefore here, we put out our hypothesis that natural viruses or more specifically their surfaces can be mimicked by utilizing engineered nanoparticles. Such surfaces can be further compared with natural viruses in terms of their transport and retention behavior in a saturated porous media environment.
Here, we are using a novel approach for synthesizing, surface-modified silica nanoparticles to closely resemble the physicochemical characteristics of virus surfaces. Physical characteristics like size, shape, and surface morphology are closely considered during the synthesis and post-modification processes, as well as surface chemistry characteristics including surface potential, particle density, and soft framework. These particle surficial features will be achieved in several stages of modification and optimization in the synthesis process. This will let us study the effect of individual elements on nanoparticle transport and retention behaviors. The results from the column sorption experiments will be studied under different environmental conditions and interpreted using numerical modelling tools.
Hierarchical porosities consist of small, often nano-scaled pores as well as large, macroscopic pores to simultaneously achieve large inner surfaces in combination with optimized mass transport. The investigation of the capillary dynamics within optically opaque hierarchically porous membranes necessitates sophisticated microscopy techniques. First hints to unveil the dynamics are obtained from theoretical thoughts and lab-scale experiments, e.g. mass-uptake as a function of time or the mechanical response depending upon wetting and drying in dilatometry. However, these techniques do not spatially resolve on the rising liquid front, which we achieved with transmission X-ray microscopy at DESY’s beamline P05. The samples are scanned in radiography (2D) and tomography (3D) to resolve both the static structure and the capillary dynamics. The findings from those experiments can help to tailor hierarchical porous materials for their designated application and to tune the dynamics in wetting and drying depending on the needs.
The characterisation of multiphase flow properties is key to predict large-scale fluid behaviour in the subsurface, such as the migration of a carbon dioxide (CO2) plume at a Carbon Capture and Storage (CCS) site. Many CCS sites have displayed unexpected fluid flow behaviour, where the CO2, once injected, migrated away from injection wells at significantly higher rates and in different orientations to what had been predicted with reservoir simulations. Recent studies have demonstrated that conventional reservoir models are not incorporating the impact of small-scale heterogeneities in multiphase flow properties, such as capillary heterogeneity. In this work, we combine experimental and numerical methods to model the impact of capillary heterogeneity on CO2 plume migration at the proposed Endurance storage site. The site supports the Northern Endurance Partnership (NEP) serving the Zero Carbon Humber and Net Zero Teesside projects in the UK. We build small-domain, fine-scale models, populated with well and experimental data from the Endurance site. These models are used to infer the impact of heterogeneity on CO2 flow in 3D with the full physics represented. Our results show that capillary heterogeneity can lead to a 3-fold increase in the relative CO2 migration speed, underscoring the importance of characterising and incorporating it within reservoir models. Using the results, we then build a full field-scale 3D model of the Endurance site. We apply a novel upscaling scheme, originating in the work of Jackson & Krevor (2020), to model the impact of heterogeneity, buoyancy and structure on CO2 migration. Our results emphasize the prevalent impact of small-scale capillary heterogeneities on CO2 plume migration.
In sequestered CO2 underground wells or in permanently abandoned wells, a cement plug is typically used as a well barrier material to ensure long-term zonal isolation of the well [1]. Proper plugging should prevent any flow of fluids from the well to the surrounding environment and the surface. It is assumed that the ability of the cement plug to fulfil its barrier function over the long term depends on its behaviour from the early age to the hardened state [2]. A good knowledge of this initial stage might allow prediction of potential crack (microannuli) formation between the cement and casing. In this study we investigate the impact of curing time during the early age of the cement plug, on its sealing properties in a scaled-down configuration but under relevant downhole conditions.
In the first stage we use an Ultrasonic Cement Analyzer (UCA) to monitor the evolution of cement strength over time, under appropriate downhole conditions. From this test, we mainly detect 3 phases: before the setting time (zero and no cement strength), a second phase where we have a strong increase of cement strength (transient state) and a third phase where the cement strength increase is negligible (steady). This allows us to select times we use as curing times before performing a plug integrity test in the transient and steady states.
In the second stage, we use a custom-built set-up [3] with which we simulate and evaluate the integrity of the plug by inducing several differential pressures and monitoring the resulting flow rates. The cement slurry is always identically mixed and placed but cured for different durations. The slope of the curve of flow rate as a function of differential pressure, namely the effective permeability of the cement-casing system falls in two distinct regions.
For curing time in the steady state, the neat Portland G cement exhibits a rapid gas breakthrough and relatively high flow rates compared to the time in the transient state. However, for curing time in the transient state, the pressure breakthrough of the gas is delayed, and the flow rates are very low or almost not observed. This suggests a degradation of the bondings and thus
of the sealing properties of cement with time despite the significant increase in cement strength.
Carbon capture and storage (CCS) is a key technology to reduce CO2 emissions and reach long term climate goals, aiming to limit the temperature rise to 1.5 ◦C above pre-industrial levels. CCS consists of capturing CO2 from large industrial processes or from burning fossil fuels in power generation. The captured CO2 is thereafter transported via pipelines or ships and stored in appropriate geological formations, such as depleted oil and gas reservoirs, unminable coal beds, and deep saline aquifers [1]. The existence of infrastructure, the wealth of reservoir data, and revenue from incremental oil recovery make depleted oil and gas reservoirs the best option for underground CO2 storage [2]. Carbonate reservoirs may be candidates for CO2 sequestration through CO2-EOR since those reservoirs hold more than 50% of the known petroleum reserves worldwide [3]. The main issue with CO2 injection into carbonate reservoirs is the formation of carbonic acid that ionizes to form hydrogen ions and bicarbonate ions. This weak acidic environment could alter the performance of carbonate CO2 storage reservoirs due to dissolution processes that lead to the formation of highly porous and conductive wormholes [4].
Long-term CO2 sequestration in carbonate reservoirs requires comprehensive assessments of CO2-reservoir fluid-mineral interactions. However, this is not an easy task due to the complexity of the reservoir
fluid and rock-forming minerals. In the present work, the interfacial-phenomena at rock-aqueous interface during CO2 injection in carbonate rocks is evaluated by a highly sensitive microcalorimetry technique called that is Isothermal titration calorimetry (ITC). Cobos et al. [5] reported that accurate adsorption enthalpy values for complex rock-fluid systems can be obtained by microcalorimetry. In the ITC experiments, 100 mg of Edwards limestone powder (< 100 μg) was placed in a reaction vessel and 200μL of North Sea formation brine (NFB) was added to the particles. The titration ampule containing the rock-brine slurry was lowered step by step into the calorimeter and equilibrated for 1 hour at 40 ◦C. Seven injections of 9.948 μL of 3.5 wt% NaCl brine saturated with CO2 (BCO2) were injected independently into the limestone+NFB system with an interval time of 420 seconds between injections. Fluid-fluid experiments consisted in injecting BCO2 into NFB. Baseline rock-fluid-fluid and fluid-fluid tests were performed similarly but without CO2.
This work shows that dynamic rock−fluid and fluid−fluid interactions take place upon CO2 injection into carbonate rocks due to composition variation “waves” that alters the equilibrium in the system. The results from the Isothermal Titration Calorimetry (ITC) indicate that the dissolution process due to the formation of a weak acidic environment is driven by entropy. This dissolution process is unfavorable with respect to enthalpy change but thermodynamically favorable with respect to entropy change (increment of cations and hydrogen carbonate in the brine). A large perturbation in the water-water network was observed when BCO2 was injected into the reservoir fluid. This alteration is a result of the salinity difference and also the presence of CO2 in the injection fluid.
Expansion and compression of fluids by injection or production in the reservoir leads to cooling or heating effects due Joule-Thomson and adiabatic processes. This effect on the near-wellbore temperature becomes significant in some applications such as carbon dioxide storage in a depleted gas reservoir. Commercial reservoir simulators using a compositional approach can model these effects. However, setting up and running compositional simulations can be very cumbersome and computationally costly, making it difficult to incorporate these simulations in workflows for uncertainty quantification, history matching and optimization. In order to allow incorporation of carbon dioxide injection modelling in workflows and networks, while keeping the relevant physics, we have derived two simplified models that account for Joule Thomson effects. The first is an analytical model based on solving for temperature and pressure equations with variable rate assuming a cylindrically shaped homogenous reservoir. In this case temperature and pressure are decoupled. This model calculates first the temperature profile based on the energy balance, and then computes the bottom hole pressure by integrating the pressure gradient given by Darcy’s law for the different viscosity regions in the reservoir. We found an excellent match of this analytical model with a commercial reservoir simulator. The second model, a numerical model, allows for more complex geological features and fluid properties modelling. We have extended an open-source fully implicit black-oil thermal reservoir simulator to account for Joule-Thomson effects. We have compared the two-models against an industry-standard compositional simulation.
Brine leakage resulting from induced fractures during CO2 storage in deep geologic formation from pressure buildup during injection or existing faults in the far field risk contamination of the shallow aquifers used for drinking and other economic activities. Our past investigations studied the effects of uncertainties in the hydraulic parameters of the storage zone on the plume development and methods for optimal monitoring of leakage and pressure release through brine extraction. One of the main challenges in validating such methods is the unavailability of field data, as no such events have occurred. In our past and ongoing research, for these types of problems, we have used intermediate-scale testing systems where some field complexities can be mimicked to generate high spatially and temporally resolved data under highly controlled laboratory environments. This paper presents a study of the effects of uncertainties of the caprock fractures on brine plume development using this approach. A novel intermediate-scale testing system was developed to couple a fractured caprock to a geologic formation with two overlaying aquifer layers representing the shallow and intermediate zones over the caprock. The fracture network was designed using predefined geostatistical parameters, and a realization of the random network was etched into a plexiglass sheet. The etched channels were filled with sand to obtain the needed transmissivities to create a hydraulic leakage pathway. By injecting tracers at four different fracture initiating points, it was possible to activate different leakage pathways and plume configurations. The 1.3m x1.3m fracture zone was hydraulically connected to the eight-meter-long soil zone with the shallow and intermediate aquifer layers. The aquifer was packed with six well-characterized test sand types to create spatially correlated random fields for the two zones with different mean and variance of the log hydraulic conductivities. Bromide was introduced as a tracer representing the leaking brine, and the plume along the intermediate stratigraphic zone was tracked through aqueous sampling. Aqueous sampling in the shallow zone was performed using a high spatial resolution grid with 448 ports. The data collected for four leakage scenarios were used to validate and verify a new numerical model that couples the fracture and the aquifer zones to simulate the migration of the leakage plume. The model was used to conduct the uncertainty analysis by varying the parameters of the fracture zone represented as an equivalent porous medium. This paper presents the experimental system design, the coupled model, and the results from the uncertainty analysis.
CO2 geo-sequestration is a practical approach to achieve net-zero carbon target. Coal has become an optimal geological storage option due to its large adsorptive capability for CO2. However, one of the main challenges for successful CO2 geo-sequestration is the reduced injectivity that are caused by adsorption-induced swelling of coal matrix. In addition, its complex and heterogenous internal pore and fracture structure make the processes of gases adsorbing, desorbing, and transporting more complicated compared with conventional rocks. This work aims to gain insights about the gas transport behaviours in coal by developing a coupled model to simulate gas flow multiphysics as well as dynamic coal deformation.
This work develops an image-based 3D fracture network model, called Fracture Box Model (FBNM), which is directly derived from 3D images of real coal samples. In this model, each fracture is described by arrays of box elements such that the regional change of fracture opening widths can be preserved. FBNM is used to simulate viscous flow in fractures, gas diffusion in matrix micropores, gas exchange on coal surface, coal matrix deformation (due to sorption and thermal expansion). Compared with other fracture models (e.g. discrete fracture network), FBNM can simulate such complicated multiphysical gas transport more efficiently, but also be able to simulate corresponding coal matrix deformation. By comparing permeability results between direct simulation method with FBNM, it is found that FBNM can effectively estimate the permeability of original fracture networks, but requiring significantly less computational cost. To study the implications of gas types, effective stress, gas adsorption, and thermal expansion on coal permeability, gas injection pressures, gas types, coal seam temperatures are varied and investigated in the simulations.
The FBNM developed in this work is more preferable for complicated flow transport simulations where direct simulation methods are still challenging. It provides a promising framework which could be further developed for multiphase and multicomponent flow simulations for CO2 geo-sequestration projects.
In this study, we illustrate key features of transport in heterogeneous fractured aquifers, when the fluid-rock diffusive exchange is a significant player, like in the case of heat transport. This advective-diffusive behavior is determined by the combined effects of flow velocity heterogeneity in the fracture system, and diffusive exchange between the fluid in the fractures and the rock matrix. In this context, the temporal evolution of the response to a pulse injection exhibits a post-peak pre-asymptotic regime, with a slope that deviate from the traditional signature of matrix diffusion. This deviation is driven by the variability of both velocity field and fracture aperture field. We illustrate the impacts of these two factors, under different conditions of heterogeneity and fracture network connectivity. We derive theoretical models that predict the pre-asymptotic tail under three extreme cases that can be related with specific network structures, i.e., networks dominated by large or small fractures, networks with highly or poorly channelized flow. These theoretical predictions are compared with results from numerical simulations in different sets of three-dimensional discrete fracture networks. Based on the numerical and theoretical results, we determine that the combined observation of solute and heat transport responses allows classifying the network in terms of connectivity structure, and partially characterizing the fracture aperture variability in terms of upscaled parameters.
Massive hydraulic fracturing has made economical production from well-compacted deep geological formations such as shale possible. Despite the commercial success, the physical mechanisms for the significantly enhanced production rate versus the conventional models remains unresolved. The mass flow from the matrix blocks to their adjacent fractures are the sources fed to the fracture network as well as the bottleneck of this semi-sealed production system. Earlier studies attribute the enhanced flow rate from the matrix blocks to the fractures to slip flow and Knudsen diffusion within the matrix blocks. Patzek [1], on the other hand, has argued that Knudsen-like scaling model is inappropriate for describing gas flow in tight formations such as shale. He demonstrated that the effect of slip on gas flow in tight formations is very weak. Both slip flow and Knudsen diffusion are only suitable for describing rarefied gas flows, but not for shale gas, which is under very high pressure in deep formations. Similar arguments on the unsuitability of the slip model for shale gas were also provided by Chen & Shen [2]. Gas flow from the matrix block to the adjacent fractures is a problem of production from a semi-sealed system. For such a system, our previous works based on the pore-scale compressible Navier-Stokes equations have shown that the motion of a viscous compressible gas is governed by a damped wave equation, and it exhibits a slip-like mass flow rate with a no-slip velocity profile [2-6]. Here we numerically and experimentally investigate how the rarefaction wave initiated at the start-up of gas flow affect the gas production from a semi-sealed dense porous plug. When a wave tries to penetrate a dense random porous medium, it loses its coherence and degenerates to a diffusion front beyond the so-called penetration length, resulting from repeated random reflections of the wave from the solid surfaces in a dense porous medium. Effective diffusion models have been long used by the physics community to describe such gas transport [7-14]. In our work, rarefaction wave induced gas transport at the pore-scale is first numerically simulated in randomly distributed porous media. By matching the computed macroscale mass flow rate with the one computed using a macroscale diffusion equation, the effective diffusion coefficient and its structure can be identified. With a large number of such computations, a machine learning model is established to extract the dependence of the effective diffusion coefficient on the mean radius and the variance of the solid grain, the porosity of the porous medium and the ratio of the outlet to the mean radius of the solid grain. A laboratory scale gas production experiment is then carried out to validate the effective diffusion model. Comparison between the model and the experiments shows that the wave-mediated effective diffusion model provides significantly better predictions for the gas production rate than that based on the Darcy’s law. The newly proposed wave-mediated effective diffusion model is therefore promising for applications to gas production from semi-sealed systems with fractured networks.
A large class of porous media consists of consolidated grains. If there is a mixture of different grain types, capillary forces may be strongly affected under immiscible two-phase flow. We have studied the effect of a random mixture of two types of grains having different wetting properties on the transport properties of immiscible two-phase flow in porous media under steady-state flow conditions using a dynamic pore-network model.
Immiscible fluids $A$ and $B$ flow through pores between two types of grains denoted "$+$" and "$-$". Fluid $A$ is fully non-wetting with respect to grain type "$+$" and is fully wetting with respect to grain type "$-$". Fluid $B$ is fully wetting with respect to grain type "$+$" and is fully non-wetting with respect to grain type "$-$". We model the pore structure as the links in a square lattice. The nodes of the dual lattice is populated by the grains. The grains of type "$+$" are assigned with a certain probability and the rest of the grains are assigned type "$-$". There are no spatial correlations among the grains. If a link passes between two "$+$" type grains, the capillary force at fluid interfaces in the link will point in the direction of fluid $B$. If a link has type "$-$" grains as neighbors, the capillary force at fluid interfaces in the link will point towards fluid $A$. If the link lies between type "$+$" grains on one side and type "$-$" grains on the other side, we assume the capillary force to be zero between the two fluids.
For a window of grain occupation probability values, a percolating regime appears where there are active connected paths with zero capillary forces. Due to these paths, no minimum threshold pressure is necessary to start a flow in this regime. Furthermore, while varying the pressure drop across the porous medium from low to high in this regime, the relation between the volumetric flow rate in the steady state and the pressure drop goes from being linear to a power law with an exponent $2.5$, then being linear again. The linearity in the initial low pressure drop regime is due to the active connected paths with zero capillary pressures, which remains the same with small increase in the pressure drop. The non-linearity at the intermediate regime is due to the opening of new paths with the increases in the pressure drop whereas the linearity in the high pressure regime is essentially due to the entire network being active.
We also measure the mobility of the system at the percolation threshold of the grain occupation probability, which exhibits a critical behavior reminiscent to the conductivity of a random resistor network. We measure the critical exponent related to this mobility and find it approximately equal to $5.7$.
The conversion of anthropogenic CO2 emissions into valuable products in gas-fed electrochemical reactors using electricity from renewable sources is a promising solution to combat global warming. To move the chemical industry towards a closed carbon cycle, the usage of gas diffusion electrodes (GDEs) will help overcome mass transport limitations in electrochemical CO2 reduction.[1][2]
Most gas-fed electrochemical reactors suffer from the flooding of GDEs within a few hours of operation, which effectively prevents stable long-term operation.[3] However, parameters which favor or prevent flooding events are not yet fully understood In addition, investigation and visualization of these parameters is challenging in conventional electrochemical reactors. In this work, we present a microfluidic model structure with multi-scale porosity featuring heterogeneous surface wettabilities to represent the behavior of a conventional GDE realistically. We establish a gas-liquid-solid phase boundary within a conductive, highly porous structure. A literature-known catalyst layer composed of silver nanoparticles and Nafion binder enables the realistic reproduction of conditions at gas-liquid-solid-interfaces seen on GDE surfaces. Especially conditions in which electrodes are partially or fully flooded can be readily investigated by our in-operando visualization method, allowing the study of wetting phenomena with confocal laser scanning microscopy. We show, that wetting of the catalyst layer is not fully reversible and demonstrate the influence of different pore sizes on GDE flooding. Application of electric potential results in the destabilization of the phase boundary and partial flooding of the electrode, thus, electrowetting is shown to have a major influence in the durability of GDEs. The influence of catalyst and binder on the advancing wetting front was investigated seperately using 3D saturation curves. This allows insights into the wetting state of electrodes based on correlations between the course of the saturation curve and the actual visualized wetting state.
Moreover, fluorescence lifetime imaging microscopy facilitates the observation of reactions on the surface of the model electrode, for the first time enabling the identification of active GDE areas, while at the same time visualizing the wetting state of the electrode. The presented results lay the foundation for the optimization of GDEs towards long-term operation of full-scale gas-fed electrolyzers.
With the aid of our microfluidic model, in-depth investigations on multi-phase wetting phenomena as well as reaction mapping are made possible, both of which are challenging or even impossible to obtain in conventional reactors. In addition, our findings may advise the design of and process conditions for larger-scale electrochemical processes. Parts of these results were recently published by our group.[4]
The successful deployment of carbon dioxide (CO2) geological sequestration in porous media is reliant on the sealing efficiency of the overlying clay-rich caprock to act as a physical barrier. Clay-rich caprock formations are considered as favourable materials to act as a seal due to them characteristically consisting of small pores providing high capillary entry pressures, hence preventing the intrusion of a non-wetting fluid (e.g., CO2).
In relation to CO2 sequestration, past experimental campaigns have traditionally focused on determining the capillary breakthrough pressure of caprock geomaterials. Only until recently have experimental results demonstrated that CO2 breakthrough is dominated by the creation of very localised channels (e.g., cracks) across the sealing barrier (Espinoza & Santamarina, 2010; Harrington et al., 2012; Busch et al., 2016 and Gonzalez-Blanco & Romero, 2022). The underlying hypothesis of this experimental work is that pore size heterogeneity governs the micro-mechanisms that ultimately control crack formation and thus, eventually, CO2 breakthrough. Therefore, this experimental campaign aims to provide evidence at the micro-scale to develop our understanding of the micro-mechanisms that lead to (or underly) the formation of large, localised channels (e.g., cracks) that pressurised CO2 is generating, causing an early breakthrough. An innovative experimental set-up which allowed for the onset of surface crack formation to be captured during gas injection into clayey geomaterials is presented. Post-mortem assessment of the aperture, volume and internal nature of these localised pathways was then visualised using the non-invasive and non-destructive xCT imaging technique.
Preliminary data on different cracking patterns when non-wetting gas (i.e., air) is injected into consolidated clay show the formation of large cracks that nucleate from the centre of the sample. Upon air pressurisation, before crack formation, the sample undergoes volumetric deformation, as the resulting action of the vertical stress applied at the air-water interface (menisci). Once a crack forms, volumetric deformation stops, and breakthrough occurs. Changing the particle size distribution, by using clay-silt mixtures, shows the potential effect of pore size heterogeneity on breakthrough and cracking patterns. Clay-silt mixtures with higher silt mass fraction result in earlier and larger crack formation, subsequently lowering the breakthrough pressure. This is opposed to the uniform pore size distribution of clay and silt materials alone, which display smaller cracking patterns. Our results, therefore, indicate that heterogeneity at the particle and subsequent pore-scale is a controlling parameter in the formation of localised pathways (e.g., cracks) in clayey geomaterials. Gas invasion into the tested clayey geomaterials occurred at lower pressures than the expected air-entry-values traditionally recorded throughout the literature. The mechanisms of air intrusion are expected to be of a similar nature as CO2 intrusion. Understanding the parameters which control the formation of localised pathways is therefore paramount when assessing the security of a geological CO2 reservoir.
Fluid-fluid displacement in porous media occurs in many natural and engineering processes such as geological CO2 storage and enhanced oil recovery. It has been recognized that wettability plays an important role in the displacement process. Thanks to decades of research, we now have a good understanding of fluid-fluid displacement in porous media with uniform wettability. In contrast, our knowledge of fluid-fluid displacement in porous media with heterogeneous wettability (i.e., mixed-wet) is much less complete, even though mixed-wet conditions are common in many subsurface processes.
Here, we study fluid-fluid displacement in simple mixed-wet micromodels. The micromodels are made of an oil-wet polymer whose wettability can be locally tuned to become water-wet via deep UV exposure. Our experiments show the mixed-wet pores exert fundamental control over the macroscopic displacement pattern and that the incorporation of the capillary entry pressures at mixed-wet pores into a dynamic pore-network model reproduces the experiments. Using the pore-network model, we systematically vary the fraction of water-wet to oil-wet regions and obtain a variety of displacement patterns over a wide range of Ca. We find that the impact of mixed-wettability is most prominent at low Ca, and it depends on the complex interplay between wettability fraction and the intrinsic contact angle of the water-wet regions. Mixed-wettability is also manifested in the injection pressure signature, which exhibits fluctuations at low wettability fractions. Finally, we demonstrate that scaling analyses based on a weighted average description of the overall wetting state of the mixed-wet system can effectively capture the variations in observed displacement pattern morphology.
Understanding and controlling transport through complex media is central for a plethora of processes ranging from technical to biological applications. Yet, the effect of micro-scale manipulations on macroscopic transport dynamics still poses conceptual conundrums. Here, we will demonstrate the predictive power of a conceptual shift in describing complex media by local micro-scale correlations instead of an assembly of uncorrelated minimal units. Specifically, we will show that the non-linear dependency between microscopic morphological properties and macroscopic transport characteristics in porous media is captured by transport statistics on the level of pore junctions instead of single pores. Probing experimentally and numerically transport through two-dimensional porous media while gradually increasing flow heterogeneity, we find a non-monotonic change in transport efficiency. Using analytic arguments, we built physical intuition on how this non-monotonic dependency emerges from junction statistics. This suggests the value of a shift in perspective towards larger-level structural elements that can broadly affets our understanding of transport within the diversity of complex media.
This study uses an experimental approach to estimate the average longitudinal
and transverse dispersion coefficients in a homogeneous, non-uniform, and anisotropic porous medium during miscible displacement. Traditionally, most miscible displacement studies have focused on recovery factor and recovery mechanism and the Peclet number is used to ?find the dispersion and diffusion coefficients from mathematical correlations. This study employs a unique method to estimate the longitudinal and transverse dispersion coefficients. A unique image processing tool is developed and used to analyze the developing mixing zone. Concentration pro?files from the processed images are then used to collaborate with Bayesian estimator tool, which is developed to ?find the dispersion coefficients in the analytical solution of the Convection-Diffusion Equation (CDE). The results con?firm that both longitudinal and transverse dispersion coefficients strongly depend on the velocity of the displacing fluid. The effects of anisotropy on miscible mass transport are investigated in this study using this unique method and longitudinal and transverse dispersion coefficients are estimated.
Scaling up renewable energy production needs to be accompanied by a concomitant scaling of storage technologies. In this regard, hydrogen (H2) is an attractive energy carrier due to its large specific energy capacity and its clean combustion products. However, its low mass density requires gigantic volumes (billion m3) to store energy in the order of TWh. Geological formations such as depleted gas reservoirs conveniently provide these large volumes. To maintain a safe operational pressure range, a cushion gas is introduced into the reservoir, which expands and compresses during the storage cycles of H2. A large compressibility of this gas maximizes the storage capacity of H2 during injection, and minimal mixing facilitates an efficient recovery of pure H2 during production. Among many options such as nitrogen, methane and H2 itself, carbon dioxide is also considered a suitable cushion gas due to its large compressibility at supercritical conditions[1].
On timescales relevant to storage (weeks to years), the cushion gas will mix with H2 through molecular diffusion and flow-induced mechanical dispersion. This work focuses on molecular diffusion, as the first step towards quantifying mixing of the stored H2 and the cushion gas. According to Fick’s law, the diffusing mass flux is a product of gradient in mole-fraction and the (molecular) diffusion coefficient D. Fick diffusion coefficients exhibit strong dependencies on thermodynamic variables such as the mixture composition, temperature T, and pressure P. In this work, we compute D at various mixture compositions using equilibrium molecular dynamics[2] for P ∈ [20,300] bar and T ∈ [250,350] K for various binary gas mixtures. The analytic expression for diffusion coefficients based on kinetic theory of gases deviates significantly from our predictions - thus emphasizing the need to account for molecular interactions. Furthermore, we provide fit functions to enable fast and accurate prediction of diffusion coefficients at reservoir conditions, which are beneficial for reservoir flow simulators. Finally, the phase equilibria of these gas mixtures are also predicted using molecular simulations.
Fate and transport of colloids and bio colloids in structurally heterogeneous porous media are known to exhibit anomalous behaviours such as non-Gaussian breakthrough curves. Classical approaches, like Colloid Filtration Theory, relies on spatial averaged quantities, neglecting flow topology heterogeneity brought about by both local pore scale surface irregularities and broad pores size distribution: two potential triggers for super diffusive effects and broad trapping time distributions. Recent theoretical work has tried to address these deficiencies by modeling deposition and flow variations as stochastic processes (Miele et al., Phys. Rev. Fluids 2019; Bordoloi et al., Nat. Commun. 2022). However, experimental evidence to demonstrate its validity for 3D geologic structures is still lacking. We thus design a novel experimental set-up to assess colloid fate transport under realistic structural heterogeneity with controlled laboratory conditions. Heterogeneous pore structures are first obtained from X-ray tomography of field samples and are subsequently 3D-printed at high resolution. Column transport experiments with gold (Au) nanoparticles are then conducted at different flow regimes, from which effluent concentration (at the macro scale) and colloid deposition (at the pore scale) are collected. These empirical data are complemented with pore network analysis that parametrizes the co-presence of preferential channels and stagnant cavities and, further, validates the stochastic model of interest. The findings shed light on the main drivers and structural hotspot for colloid filtration in realistic porous media.
Spontaneous imbibition of brine at nonzero initial water saturation is an important mechanism for recovering crude oil from mixed-wet heterogeneous carbonate rock. Many studies focus on studying or modeling spontaneous imbibition of brine into fully oil-saturated (i.e., without connate water) or water-wet porous media. As a result, adequate models describing spontaneous imbibition process into rock with nonzero initial water saturation and mixed-wettability do not exist.
First, we review the experimental variables that are important for spontaneous imbibition in water-wet and mixed-wet rocks in the presence of connate water. We show that the classic Amott experiment, broadly used to evaluate ultimate oil recovery, masks several flaws that hinder the interpretation of recovery dynamics and thus the development physical and predictive recovery models. The key aspects are 1) contribution of the buoyancy-driven oil production; 2) wettability-dependent oil hold-up at the core surface; and 3) inconsistent outer surface wettability of mixed-wet limestone core plugs.
We then modify the classic Amott testing procedure to minimize experimental artifacts in the recovery dynamics. The main modifications include the following: 1) capping the top and bottom faces of core plugs with glass discs to eliminate the axial flow and enforce only 1D radial two-phase flow; 2) continuous shaking throughout the entire experiment to eliminate the oil external-surface hold-up with different core wettability states; 3) degassing of both brine and oil prior to any experiments. We show that the modified Amott experimental procedure obtains smooth and reproducible oil-recovery histories for oil-saturated core plugs with different wettability conditions of limestone rock. Figure 1 compares the cumulative oil recovery versus square root of time in the classic Amott test and with the introduced modifications. Note the smooth oil recoveries compared to the classic procedure.
Finally, we show that the resulting smooth recovery profiles of oil production can be described by a statistical model that fits the data very well. For the first time, we demonstrate that generalized extreme value (GEV) distribution can be applied to model spontaneous brine imbibition into water-wet and mixed-we cores in the presence of connate water. Figure 2 is an example of the GEV scaling of cumulative recovery versus square root of dimensionless time for water-wet core plugs.
The next step in developing our new approach to modeling spontaneous imbibition in the presence of connate water is to elaborate on how the distinctions of the physical processes during oil recovery by spontaneous imbibition are reflected by the GEV statistics. We believe, that our GEV modeling approach will serve as a foundation for the development of a next-generation predictive model of oil-recovery dynamics from mixed-wet carbonates.
Steady-state transitions in porous media, here defined as a discontinuity in one or more macroscopic observables as a function of Reynolds number while the flow remains steady, are known to occur for a multitude of different types of porous media. In previous studies, it has been discovered that these transitions coincide with the development of inertial cores and a reduction in the spatial variance of the velocity field [1]. Recently, flow through an ordered porous media cell consisting of a staggered cylinder packing was investigated using tomographic particle image velocimetry [2]. The results reveal a complex three-dimensional steady-state flow pattern, occurring in the region where inertial effects are expected to become dominant. The peculiar flow pattern, which significantly increases the vorticity and flow resistance, indicates that the transition may need more explanation than the development of inertial cores as suggested by earlier studies.
The investigation is confined to single-phase, fully saturated Newtonian flow through porous media. By making use of an in-house GPU implementation of an artificial compressibility finite difference method the transition from the stokes flow region, to the end of the steady inertial region, is performed on three types of ordered porous media. These are a staggered packing of mono-radii cylinders, a staggered packing of quadratic cross-section rods and a body-centred cubic packing of mono-radii spheres as disclosed in the figure. Here, also the flow regions are presented where the Reynolds number increases from left to right and the velocity magnitude is visualized by a volume rendering. When increasing the Reynolds number it is concluded that in addition to the transition resulting in an increase of the spatial variance of the velocity, as known from earlier studies, it coincides with a rise in the absolute value of the pressure integral across the solid surfaces. These observations, together with an observed increase in the absolute value of the pressure velocity coupling, indicate that the flow tends to alternative flow paths which reduces impingement on the solid surfaces.
In the eye of climate change, Carbon Capture and Storage (CCS) gained importance as a large-scale option to permanently sequester CO2. To ensure storage safety, it is crucial to understand trapping mechanisms and the trapping potential. To do so in a time-efficient way, the application of Digital Rock Physics Simulation has become a major tool. The presentation will focus on capillary trapping by re-imbibition simulated by means of the morphological approach and benchmarked by pore-scale flooding experiments.
Using the data of flooding experiments with in-situ pore scale imaging, fluid configurations in digital twins of various sandstone samples were simulated and compared to experimental results. To compute the trapping potentials of those rocks, simulations of primary drainage and subsequent imbibition processes were performed based on the morphological method. From the results, land trapping model constants were calculated depending on the initial and residual CO2 saturations of the imbibition processes. For obtaining realistic results, we make use of recent developments in the frame of the morphological approach in digital rock physics by Arnold et al. Prior to that, only spontaneous imbibition processes could be simulated, now extended to the whole imbibition branch including forced imbibition. Also, the effect of wettability and contact angle variation is considered in a stochastic and deterministic way. Simulations of multiple scenarios with varying input parameters were performed to benchmark against experimental results and as first step towards stochastic input for reservoir modeling.
The definition of an optimal reservoir management strategy is fundamental for the primary production of oil and gas, Enhanced Oil Recovery, Underground Gas Storage, Underground Hydrogen Storage, CO2 storage, and geothermal systems. Due to the complexity of geological formations, the uncertainty associated to the fluid-rock interaction parameters must be estimated and possibly mitigated by the acquisition of further information at all stages of reservoir life. The characterization and analysis of fluid flow phenomena at the pore scale can contribute to minimizing such uncertainties. Thanks to micro-CT images, a realistic representation of the reservoir rock is obtained, and it can be used as input for further analyses of pore space characteristics. In this work, the A* search algorithm (Hart et al., 1968) is used to compute the shortest connected paths across micro-CT images of rocks in the three main flow directions. This information is employed to calculate tortuosity, effective porosity, constriction factor, pore size distribution, permeability and anisotropy ratio of the rock. This process was introduced by Salina Borello et al. (2022) but was not applied to real 3D images of rocks obtained by micro-CT techniques. In this work, a sandstone and two carbonate rocks are analyzed. Fluid flow is intrinsically influenced by all the features of the pore space (i.e. tortuosity, effective porosity, constriction and pore radius) and the porous medium is usually characterized by a single parameter, the permeability, which takes into account all these properties. In fluid flow simulation, permeability can be calculated directly by inverting the Darcy’s law. In the geometrical analysis instead, all pore space characteristics are evaluated individually and combined through the Kozeny-Carman equation. The geometrical tortuosity is calculated as the average length of the shortest (or geometrical) paths divided by the edge length of the sample. The effective porosity is computed as the portion of pore space crossed by the geometrical paths. The constriction factor expresses the variation of the pore cross-section orthogonal to the path. The pore size is estimated as the distance between the pore walls locally orthogonal to the path. In order to calculate the permeability, the Kozeny-Carman equation is used by including the geometrical tortuosity, effective porosity, and a representative pore radius in the equation. Finally, the anisotropy ratio is calculated using permeability values in the three main directions. Results are compared with those obtained by single-phase CFD simulation directly in the pore space using OpenFOAM, with the exception of pore size distribution and constriction factor. Geometrical analysis and CFD simulations are run at the pore-scale directly on binary images of rocks. The values of tortuosity, effective porosity, permeability, and anisotropy ratio calculated with the geometrical analysis and CFD simulation are in good agreement in all the cases. The outcome of this investigation evidences that the geometrical analysis used in this research can provide a reliable characterization of rocks.
The rapid advancement in digital core analysis has greatly promoted the research development of flow and transport in porous media. However, the field of interest revealing pore level information that can be processed through standard digital core analysis workflow is rather limited for practical purposes. The integration of pore-scale information into continuum scale is widely concerned as it associates deeply with the future development of digital core analysis. For hierarchical porous structure, pore-scale rock-typing and upscaling of petrophysical properties is a promising solution to bridge the gap between microscale and continuum scale. Morphological and topological parameters associating data clustering methods are popularly utilized for the pore-scale rock-typing on 3D digital samples. However, the size of regional support window through which the fields of the parameters are generated greatly affects the descriptive capacities of the parameters on the structural characteristics, thus the classification using traditional unsupervised clustering methods such as Gaussian Mixture Models (GMM) is hard to deliver optimal performance. Towards the issue, we propose in this work to apply a supervised method called Support Vector Machine (SVM) for rock type classification. Minkowski functionals are determined as robust descriptors for the morphological and topological characteristics of porous structures, and a fast computational method utilising Fast Fourier Transform (FFT) has been applied for the generation of the fields of the regional Minkowski measures. On the basis of the Minkowski fields generated through different regional support sizes from the target porous structures, comparative experiments between the two different classification methods SVM and GMM have been conducted on two complex artificial porous systems and one digital image of a laminated sandstone. Throughout the tests, SVM has illustrated obvious advantage on overcoming regional support size effect even with limited labelling information. The combination of regional morphological and topological descriptors with SVM method could provide extraordinary convenience for the realization of pore-scale rock-typing on large 3D digital images with excellent computational efficiency.
Investigating the operation dynamics of energy storage systems, like lithium-ion batteries, is becoming increasingly important in a range of applications, from automotive to grid-connected energy storage. To simulate battery discharge we performed finite element simulations in unstructured grids. The transient simulations are focused on the cathode of the battery (i.e.: half-cell simulations), and the discharge at different C rates is explored [1]. The mass and charge equations of balance are solved at the microscale on three-dimensional reproductions of cathodes geometries. Three phases are considered in the electrochemical model: active material, carbon binder domain, and electrolyte. The creation of the cathode geometries is based on the fabrication process of the electrodes which allows us to produce a wide range of geometries to perform numerical simulations on. Different amounts of active material at different degrees of compression (calendering) of the electrodes have been explored to create electrodes geometries that produce different battery discharge profiles. The results of these simulations are then interpolated back to a regular 3D grid. This whole process is very costly and cannot be integrated into optimization workflows.
The multi scale neural network (MSNet) has been employed to train data-driven models to perform different tasks related to transport in porous media. Notable examples include prediction of flow fields [2], electric potential [3], and concentration fields in reactive flows [4]. In all these applications, the target of interest is the steady-state solution of the field. Nevertheless, in the case of electrochemical transport in lithium-ion batteries we are interested in knowing the discharge profile in time at different rates. Hence, the data-driven model requires modifications to learn to model the temporal dynamics.
In this work we trained an autoregressive MSNet on a dataset of time-dependent 3D simulations of electrochemical transport in lithium-ion batteries. The network takes as input the geometrical descriptors and the operating conditions, together with a temporal feature, which is: the initial condition for the prediction of the first time-frame, or the previous time-frame prediction for the prediction of the following one. Using this approach, the transient nature of the dataset is preserved, since MSNet is provided with the complete ordered samples (from initial conditions to final time) at each epoch and learns how to dampen the prediction errors along the time-frames. The predicted fields are integrated in order to reproduce the discharge curves of lithium-ion batteries. The performance of the trained network is tested on new cathodes to show its generalization capabilities. We envision this work as a proof-of-concept of the feasibility of this approach leading to the employment of the autoregressive MSNet in optimization frameworks on a wider and more challenging datasets.