About this title - Engineering Geology of Groundwater in Design and Construction: Engineering Group Working Party Report

The understanding of the orientation of pre-existing discontinuities is crucial for subsurface engineering applications, such as drilling, well completion, and hydraulic fracturing. A variety of methods have been developed to identify fracture orientation, including wellbore imaging and monitoring of microseismic events in reservoirs. Recent research has utilized shear waves to evaluate fracture orientation and density in porous media by analyzing the shear wave splitting (SWS). The aim of this study is to assess the fracture orientation in two rock types, synthetic Hydrostone and Eagle Ford shale, under triaxial loading conditions using the SWS phenomenon. Shear wave velocities were measured using two orthogonal transducers at each end of the fractured sample for transmitting and receiving shear wave travel times. The results showed that SWS, the velocity difference between the two shear waves, was the highest at fracture orientations of 0° and 90°, when one of the transducers was parallel with the fracture. Whereas SWS was the lowest for the sample with fracture at 45°, when the split shear waves attenuated during re-orientation with their preferred propagation directions. The findings of this study could help identify fracture orientation near the wellbore and contribute to minimizing various drilling and completion issues. INTRODUCTION Natural fractures are abundant mechanical structures that exist in most rock types, especially in carbonates and shale (Moore and Wade, 2013). These pre-existing discontinuities could cause several drilling issues including mud losses (Razavi et al., 2017) and drilling breaks (Narr et al., 2006). In production engineering, the presence of natural fractures may enhance the local permeability, and hence the production rate (Luffel et al., 1993; Sakhaee-Pour and Bryant, 2011; Ben et al., 2012; Al-Rubaye et al., 2020). However, in the meantime, it can also increase the gas-oil ratio and water cut impairing the hydrocarbon productivity (Aguilera, 1980; Wennberg et al., 2016). Furthermore, in hydraulic fracturing applications, these natural fractures are known to have significant impact on the hydraulic fracture trajectory and the Stimulated Reservoir Volume (SRV) depending on their relative orientation to the hydraulic fracture propagation direction (Gale, et al., 2007; Zhou et al., 2008; Bahorich et al., 2012; Lee et al., 2015). The orientation of natural fractures is determined by the paleo-stress field of the most active tectonic period (Maerten and Maerten, 2006; Zhang et al., 2022). Nevertheless, this orientation may not always align with the current stress field (Abul Khair et al., 2013; Abul Khair et al., 2015; Likrama et al., 2019). Thus, characterizing the orientation of natural fractures is crucial for effective design of various subsurface engineering applications.

Understanding heat transfer in rock fractures is crucial for optimizing geothermal energy extraction, nuclear waste storage, and other subsurface engineering applications. In geothermal systems, the understanding of thermal behaviour in fractured media is still challenging, due to the complexity of fracture geometry, heterogeneous properties of the fractures and the host rock, and varying fluid flow dynamics influenced by temperature-dependent fracture aperture. Considering that the aperture and shape of fractures can promote preferential transport of fluids and heat, several numerical and experimental studies have demonstrated that these preferential paths, or “flow channeling,” significantly impact heat transfer. However, there is no clear consensus on the effects of flow channeling on the thermal exchange between the fluid and the rock matrix, as some authors observed a decrease, due to increased flow velocity and shortened transit times in the channeled regions, while others report an increase, as radial conduction from the channel to the matrix is more efficient for heat transfer than the linear conduction assumed in a parallel plate model. This study explores the relationship between fracture roughness and heat transfer mechanisms, focusing on advective and diffusive processes under saturated conditions. Finite element numerical models are employed to simulate fluid flow and heat transfer in a set of simplified fracture geometries in which the fracture walls are represented through a sinusoidal function. These models include three scenarios: a fully-mated fracture geometry formed by two aligned sinusoidal surfaces, a fully-unmated configuration, and an intermediate geometry that transitions between the two mentioned geometries. Preliminary results indicate that surface roughness influences convective heat transfer by inducing localized flow channeling. This effect is quantified by observing the thermal attenuation and the lag time of the induced cold pulse imposed over the system. Notably, depending on the fracture geometry, distinct temperature peaks and varying heat recovery tailing profiles are observed across different scenarios. Further work is needed to define appropriate model dimensions, select suitable heat and flow parameters, and refine the time discretization. Additional numerical experimentation is required to determine the optimal approach for modelling the fracture, such as choosing between a function-based or fracture-based representation.

Wettability alteration of oil-wet carbonate by silica nanofluid

Deep-learning (DL) models have been used for production forecasting in subsurface engineering applications, but it is often assumed that each well operates independently. Graph convolutional networks (GCNs) can incorporate data from neighboring wells. However, existing spatial-temporal (ST) GCN (ST-GCN) methods are mainly used for autoregressive tasks and face limitations in predicting newly developed wells with no prior history. In this study, we introduce an ST-graph-level feature embedding (GFE) (ST-GFE) method that fully utilizes temporal neighbor interactions for newly developed wells. It enhances forecasting by integrating a non-autoregressive encoder-decoder structure and aggregating the historical data from neighboring wells into a single feature vector. This aggregated vector, merging local and contextual information, contains richer information about the studied region. We evaluate ST-GFE using a data set of 6,605 Montney shale gas wells, incorporating formation properties, fracture parameters, and production history. ST-GFE significantly improves prediction accuracy for newly developed wells compared with the purely temporal models, such as recurrent neural network–based and transformer models. ST-GFE adapts to production changes in adjacent wells, including shut-in and infill drilling activities. Additionally, the ST-GFE model treats each well and its surrounding wells as a graph, enabling batch training and significantly reducing memory usage compared with transductive GCN approaches. Furthermore, the model dynamically updates its forecasts with real-time production data, enhancing precision and relevance. The xperimental results confirm that ST-GFE effectively leverages spatio-temporal dynamics and interactions between adjacent wells, broadening its applicability to various drilling and production scenarios.

Anisotropy has a significant impact on the elastic and mechanical properties of rocks. Misidentifying a rock formation as isotropic can lead to significant errors in predicting stress distribution and mechanical deformation in the subsurface. Sandstone-shale laminated rocks are intrinsically anisotropic and are of great interest in subsurface engineering applications, as they constitute important assets in global oil and gas reserves. Under specific conditions (layer thickness, property contrast), these rocks can be effectively represented by an equivalent homogeneous transversely isotropic (TI) medium. The elastic moduli of the TI medium are calculated as the product of the properties of each layer and their respective thicknesses. We examine the latter concept through laboratory testing and angle-dependent ultrasonic reflection-coefficient measurements. Experimental data acquired from synthetic 3D-printed layered samples with varying layer thicknesses (smaller or greater than the receiver size) are compared with semi-analytical and numerical simulations. This comparative analysis yields valuable insights into the resolution of the method and helps to determine the conditions under which spatially heterogeneous samples can be accurately represented by effective-medium models of elastic rock behavior. Laboratory measurements acquired in a controlled environment confirm that samples characterized by weak anisotropy and layer thickness smaller than the receiver diameter can be accurately represented by an equivalent vertical TI medium. However, the receiver size-to-layer thickness condition is only valid under the assumption that the receiver aperture spans several wavelengths and that the measurements are subject to spatial averaging. Noteworthy differences arise when measurements are taken parallel and perpendicular to the sample bedding plane. Measurements acquired perpendicular to the layering reflect the properties of the effective medium. Reflection coefficients acquired parallel to the layers can effectively capture the elastic properties of the layer with differences below 5% compared with the homogeneous material.

Suspended particle transport is frequently involved in many geophysical processes and subsurface engineering applications. Although common and important, the effect of liquid cohesion on particle clogging has been overlooked in previous studies. We conduct visualized experiments of dilute suspension flow in a rough fracture and find a dramatic enhancement of clogging by a tiny amount of additional immiscible wetting liquid, even at weight percentage ω ≤ 0.5%. An experimental phase diagram of clogging patterns is obtained in the space of secondary liquid content and flow rate. The combined effect of suspension composition and hydrodynamic condition on the clogging behavior is analyzed to explain transitions of clogging regimes. A theoretical model of agglomerate size is proposed to quantify the capillary cohesion effect. This work improves the understanding of fines migration and particle‐clogging behaviors in the subsurface and paves the way for possibility of controlling particle transport and clogging in various applications.

The dynamics of droplet detachment from a surface is a fundamental topic studied in coating engineering, fluid mechanics, and subsurface engineering applications. This topic has direct relevance to wettability alteration using the modified ionic composition of water in contact with oil droplet, low salinity waterflooding (LSWF). Previous experimental studies of LSWF have shown a very long timescale in wettability alteration which cannot be explained using bulk diffusion coefficient. In the present study, we address both the time scale of detachment, as well as the impact of buoyancy and interfacial forces (referred to as Bond number) on droplet detachment by proposing an advanced GPU-enhanced lattice Boltzmann model. In this model, the immiscible two-phase flow has been coupled with wettability alteration due to the salinity dilution. After full validation of the model against the former experiments, a set of computational setups with distinct Bond numbers were designed to investigate the effect of interfacial tension and droplet size on the dynamics of droplet detachment. Results demonstrate that droplet detachment from a surface is not a unique function of Bond number and diffusion length scale is critical in detachment time. In summary, this study provides a fully validated model of LSWF and delineates the significant difference in impacts of interfacial tension and bubble size on detachment time, which can be used to predict capability for LSWF, to determine favorable conditions, as well as to observe and explain low-salinity-effect.

The interaction between CO2 and water in subsurface environments plays a critical role in altering the geomechanical properties of rocks, with significant implications for carbon sequestration, reservoir integrity, and underground storage applications. This study evaluates the impact of CO2-water exposure on the strength, elasticity, porosity, and mineralogical composition of different rock types, including sandstone, limestone, dolomite, basalt and shale. Laboratory experiments were conducted to characterize the initial petrophysical and geomechanical properties of the rock samples before subjecting them to CO2-saturated water under controlled pressure and temperature conditions. Post-exposure analyses were performed using nanoindentation SEM-EDS and X-ray diffraction (XRD) to assess mineralogical and structural changes. The results indicate that CO2-water interaction leads to varying degrees of mechanical weakening, with carbonate rocks showing significant dissolution effects and reduced elastic modulus. In contrast, silicate-rich rocks like sandstone exhibited comparatively lower degradation due to their mineralogical stability. These findings highlight the importance of rock-specific evaluations in subsurface engineering applications, particularly in optimizing CO2 storage strategies and ensuring long-term stability. Further studies incorporating extended exposure durations and field-scale validation are recommended to enhance predictive models for rock behavior in CO2-rich environments.

Fluid injection into rock masses is involved during various subsurface engineering applications. However, elevated fluid pressure, induced by injection, can trigger shear slip(s) of pre-existing natural fractures, resulting in changes of the rock mass permeability and thus injectivity. However, the mechanism of slip-induced permeability variation, particularly when subjected to multiple slips, is still not fully understood. In this study, we performed laboratory experiments to investigate the fracture permeability evolution induced by shear slip in both saw-cut and natural fractures with rough surfaces. Our experiments show that compared to saw-cut fractures, natural fractures show much small effective stress when the slips induced by triggering fluid pressures, likely due to the much rougher surface of the natural fractures. For natural fractures, we observed that a critical shear displacement value in the relationship between permeability and accumulative shear displacement: the permeability of natural fractures initially increases, followed by a permeability decrease after the accumulative shear displacement reaches a critical shear displacement value. For the saw-cut fractures, there is no consistent change in the measured permeability versus the accumulative shear displacement, but the first slip event often induces the largest shear displacement and associated permeability changes. The produced gouge material suggests that rock surface damage occurs during multiple slips, although, unfortunately, our experiments did not allow quantitatively continuous monitoring of fracture surface property changes. Thus, we attribute the slip-induced permeability evolution to the interplay between permeability reductions, due to damages of fracture asperities, and permeability enhancements, caused by shear dilation, depending on the scale of the shear displacement.

Understanding particle transport in rough-walled fractures is essential for predicting flow behavior, clogging, and permeability evolution in natural and engineered subsurface systems. This study develops a fully coupled CFD–DEM framework to investigate how self-affine fractal roughness, represented by the Joint Roughness Coefficient (JRC), governs fluid–particle interactions across multiple scales. Nine fracture geometries with controlled roughness were generated using a fractal-based surface model, enabling systematic isolation of roughness effects. The results show that increasing JRC introduces a hierarchy of geometric perturbations that reorganize the flow field, amplify shear and velocity-gradient fluctuations, and enhance particle–wall interactions. Particle migration exhibits a nonlinear response to roughness due to the competing influences of disturbance amplification and the formation of preferential high-velocity pathways. Furthermore, roughness-controlled scaling relations are identified for mean particle velocity, residence time, and energy dissipation, revealing JRC as a fundamental parameter linking geometric complexity to transport efficiency. Based on these findings, a unified mechanistic framework is established that conceptualizes fractal roughness as a multiscale geometric forcing mechanism governing hydrodynamic heterogeneity, particle dynamics, and dissipative processes. This framework provides new physical insight into transport behavior in rough fractures and offers a scientific basis for improved prediction of clogging, proppant placement, and transmissivity evolution in subsurface engineering applications.

Estimating multiphase flow properties, particularly permeability, is critical for addressing critical challenges in subsurface engineering applications such as CO2 sequestration, efficient oil and gas recovery, and groundwater contaminant remediation. At the sub-core scale, accurate determination of permeability is vital for understanding flow dynamics and reservoir characterization. However, traditional estimation methods, which rely heavily on numerical simulations, are computationally expensive and time-intensive, limiting their scalability for large-scale or real-time applications. Deep Neural Networks (DNNs) have emerged as a promising alternative due to their ability to learn complex input-output relationships, enabling rapid predictions. Despite their potential, standard data-driven deep neural networks (DNNs) encounter substantial challenges when data availability is limited, often resulting in suboptimal performance and unreliable predictions. Additionally, these models heavily rely on the quality of the measurements, making them sensitive to noise and inaccuracies in the dataPhysics-Informed Neural Networks (PINNs), a class of DNNs that incorporate physical laws as soft constraints, have demonstrated exceptional robustness in addressing inverse problems under data-scarce conditions. By embedding the governing equations into the learning process, PINNs bridge the gap between data-driven and physics-based modeling approaches. Nevertheless, the application of PINNs to inverse problems is often scenario-specific, requiring retraining when transitioning to new conditions or settings. While recent studies have begun leveraging PINNs as surrogate models to efficiently solve forward problems across varying conditions, their full potential in generating datasets for coupled systems remains underexplored. In this study, we present an innovative framework that integrates a PINNs-based surrogate model with a data-driven DNN to accurately and efficiently estimate a 1D heterogeneous permeability profile using sub-core saturation measurements. The surrogate PINNs system was pre-trained to solve a 1D steady-state two-phase flow problem, incorporating capillary pressure heterogeneity and spanning a wide range of flow conditions. This pre-trained PINNs system was subsequently employed to generate an extensive dataset for training a DNN, which establishes a direct mapping between permeability, flow conditions, and measured saturations at the sub-core level. By coupling these two systems, our approach enables the rapid prediction of permeability profiles based on observed flow conditions and saturation measurements, bypassing the computational burden of traditional numerical simulations. The coupled framework demonstrated remarkable accuracy and robustness, achieving average misfits below 1% when validated against actual permeability profiles. Its computational efficiency also facilitated the development of a stochastic extension, allowing the system to handle noisy or contaminated data while quantifying uncertainties. This enhanced solution, capable of delivering results in less than 15 seconds, significantly improves the reliability and applicability of the method for real-world scenarios. Furthermore, the approach successfully reconstructed 1D permeability structures from 3D datasets and generated 1D saturation profiles under varying conditions, achieving an average misfit of approximately 3%. These findings highlight the potential of integrating PINNs with data-driven models for high-fidelity, efficient estimation of flow properties in heterogeneous systems. The proposed method offers a powerful tool for advancing subsurface flow characterization, with broad implications for both scientific research and practical applications.

Viscoelastic strain rate-dependent behaviour of coal is critical in several subsurface engineering applications especially coal seams gas production. Such rate dependency is controlled by the interaction between coal bulk and gas sorption (a sorbing gas) or gas pressure (a non-sorbing gas). Despite the research conducted to date, the gas pressure effect (non-sorbing) on the viscous behaviour of sediments in particular coal remains unexplored. We, therefore, investigate the strain rate-dependent mechanical behaviour of coal under isotropic loading to specifically explore the effect of gas pressure (Helium) on its rate dependency eliminating the sorption effect. We perform a set of triaxial experiments on coal specimens at dry and pressurised gas (Helium) conditions under different strain rates under isotropic loading. The experimental results show that all coal specimens have viscoelastic strain rate dependency at a dry condition where viscous effect increases with strain rate. As a result, the bulk modulus of the specimens increases with the increase in strain rates. This strain rate dependency response, however, reduces with an increase in pore pressure and vanishes at a certain pore pressure under the same effective stress to that of dry specimens. We further employ X-ray micro-Computed Tomography (XRCT) to 3D scan a coal specimen saturated with Krypton gas undergoing different loading rates to shed light on the micro-mechanisms of gas pressure effect on specimens’ rate dependency. The XRCT results show that gas can be trapped in small-scale fractures and pores during the loading process leading to a localised undrained response that can stiffen the specimen and reduce its ability to show viscous rate dependency. The obtained results are significant in optimizing coal seam gas production and coal seam gas drainage applications.

The flow velocity in metre-scale natural fracture networks readily exceeds centimetres per second, the threshold for non-stationary flow. However, despite widespread evidence of such dynamics, these are rarely considered in subsurface engineering applications, where steady-state simulation approaches dominate. Here, we compare Reynolds-averaged Navier-Stokes (RANS) and Detached-Eddy Simulation (DES) methods for the transient Navier-Stokes equation applied to fracture flow. These models are validated with experimental data of flow through fracture intersections. DES is then applied to a metre-scale fracture pattern with hundreds of discrete fractures, examining flow dynamics at velocities up to metres per second (m/s). DES accurately captures the temporal flow fluctuation and multiscale eddy formation, especially when a fine computational mesh is used in wake regions. By contrast, unsteady RANS fails to capture flow-field variations and produces results similar to steady RANS. DES reveals significant network flow periodicity (∼40 Hz) at m/s velocities, unlike the low-frequency results (∼0.4 Hz) from RANS. We also explore the impact of unsteady flow on particle transport by integrating mixture-multiphase and rheological models into DES. Corresponding results indicate that inertia alters the concentration of transported solids, mixture viscosity, and particle dynamics such as clustering.

Spatial estimation of the thickness of low permeability topsoil materials by using a combined ordinary-indicator kriging approach with multiple thresholds

The biogeochemical process known as microbially induced calcite precipitation (MICP) is being investigated for engineering and material science applications. To model MICP process behavior in porous media, computational simulators must couple flow, transport, and relevant biogeochemical reactions. Changes in media porosity and permeability due to biomass growth and calcite precipitation, as well as their effects on one another must be considered. A comprehensive Darcy-scale model has been developed by Ebigbo et al. (Water Resour. Res. 48(7), W07519, 2012) and Hommel et al. (Water Resour. Res. 51, 3695–3715, 2015) and validated at different scales of observation using laboratory experimental systems at the Center for Biofilm Engineering (CBE), Montana State University (MSU). This investigation clearly demonstrates that a close synergy between laboratory experimentation at different scales and corresponding simulation model development is necessary to advance MICP application to the field scale. Ultimately, model predictions of MICP sealing of a fractured sandstone formation, located 340.8 m below ground surface, were made and compared with corresponding field observations. Modeling MICP at the field scale poses special challenges, including choosing a reasonable model-domain size, initial and boundary conditions, and determining the initial distribution of porosity and permeability. In the presented study, model predictions of deposited calcite volume agree favorably with corresponding field observations of increased injection pressure during the MICP fracture sealing test in the field. Results indicate that the current status of our MICP model now allows its use for further subsurface engineering applications, including well-bore cement sealing and certain fracture-related applications in unconventional oil and gas production.