Advances in Water Resources最新文献

The dynamics and aggregation of microplastics in marine environments are investigated through high-fidelity direct numerical simulations with Lagrangian point-particle tracking. The properties of microplastics and biogenic particles, including size, density, and concentration, align with scenarios typical of seawater systems. The stickiness nature of microplastics, induced by biofilm formation (biofouling), is modeled through coagulation efficiency (stickiness parameter), which represents the probability of aggregation following a collision event. Two main aspects are at the core of the present work: analyzing the mechanisms of collision and coalescence between microplastics and biogenic particles, along with their spatial distribution, and characterizing the emerging aggregates. The results indicate that particles stickiness, concentration and (especially) size impact on the collision and coalescence rates. Furthermore, microplastics exhibit a strong tendency to accumulate near biogenic particles, leading to the creation of hetero-aggregates whose tendency to sink supports the general hypothesis of “missing microplastics”. Particularly, in cases where microplastics and biogenic particles are evenly concentrated, microplastics primarily contribute to the formation of aggregates. The stickiness mainly determines the most complex and large aggregates, which are less than 1% of the total.

Physics-informed neural networks (PINNs) are receiving increased attention in modeling flow in porous media because they can surpass purely data-driven approaches. However, in heterogeneous domains, PINNs often face convergence challenges due to discontinuities in rock properties. A promising alternative is the mixed formulation of PINNs, which utilizes pressure head and velocity fields as primary variables. This formulation introduces a multi-term loss function whose terms must be carefully balanced to ensure effective convergence during training. The main goal of this work is to identify the most suitable weighting technique to overcome convergence issues and enhance the applicability of the mixed formulation of PINNs for modeling flow in heterogeneous porous media. Thus, we implement and adapt different global and local weighting techniques and evaluate their performance through multiple test scenarios, involving stochastic and block heterogeneity. The results reveal that the most appropriate weighting strategy is the max-average technique. In the case of stochastic heterogeneity, this technique allows for improving the convergence of the training algorithm. In the case of discontinuous heterogeneity, the max-average method is the only strategy that achieved convergence, highlighting its robustness. The results also show that under high heterogeneity, using an appropriate weighting technique becomes imperative because baseline PINN failed to converge. Implementing an optimal weighting strategy can improve convergence and yield accurate solutions with fewer learnable parameters, thereby enhancing overall model performance.

A sudden surge of data has created new challenges in water management, spanning quality control, assimilation, and analysis. Few approaches are available to integrate growing volumes of data into interpretable results. Process-based hydrologic models have not been designed to consume large amounts of data. Alternatively, new machine learning tools can automate data analysis and forecasting, but their lack of interpretability and reliance on very large data sets limits the discovery of insights and may impact trust. To address this gap, we present a new approach, which seeks to strike a middle ground between process-, and data-based modeling. The contribution of this work is an automated and scalable methodology that discovers differential equations and latent state estimations within hydrologic systems using only rainfall and runoff measurements. We show how this enables automated tools to learn interpretable models of 6 to 18 parameters solely from measurements. We apply this approach to nearly 400 stream gaging sites across the US, showing how complex catchment dynamics can be reconstructed solely from rainfall and runoff measurements. We also show how the approach discovers surrogate models that can replicate the dynamics of a much more complex process-based model, but at a fraction of the computational complexity. We discuss how the resulting representation of watershed dynamics provides insight and computational efficiency to enable automated predictions across large sensor networks.

Freshwater lenses in otherwise saline environments contain an important source of fresh water for natural vegetation and agricultural crops. Such lenses are regularly found in areas where both upward seeping saline groundwater and downward infiltrating fresh recharge water occur simultaneously during part of the year, resulting in shallow freshwater lenses which are highly susceptible to changes in recharge or seepage. In a series of two papers, we consider the water – and solute transport in a 2D cross-section between two parallel outflow faces. In this second part of the series, we build upon expressions presented in the first part to consider a situation where the density of seepage water exceeds that of recharge water, as typical for many deltaic areas around the world. Analytical expressions and approximations are given to obtain the steady state position of the interface between the two types of water using a sharp interface approximation, with a focus on the position midway between two outflow faces. Results show that the effect of a heterogeneous density distribution is limited when the seepage flux exceeds the density difference induced flux, but increases rapidly for ratios of the seepage flux over the density flux falling below 1. The heterogeneous density distribution then results in a decrease in freshwater lens thickness and, correspondingly, a decrease in fresh water availability. We also consider time-variant, oscillatory boundary conditions, and show that for heterogeneous density distributions the interface approaches its equilibrium position faster than for a corresponding situation with a homogeneous density distribution, indicating a higher vulnerability for changing boundary conditions. We also demonstrate that heterogeneous density distributions have limited effect on the amplitude of the interface oscillations. Analytical results obtained with the simplified model are validated using the numerical code SUTRA, which solves the full model for a numerical grid.

The present study investigates the quality of four three-dimensional (3D) printing technologies to accurately reproduce the complex pore structure of a real undisturbed soil sample for laboratory experiments of transport in porous media at a 1:1 scale. Four state-of-the-art 3D printing technologies were evaluated (digital light synthesis, PolyJet with gel support material, low-force stereolithography, and PolyJet with water-soluble support material) using a combination of 3D image analysis from microtomopraphy and flow simulations of the pore structure produced with each 3D printing technique. Accuracy, as determined by matching solid and void volumes, permeability, connected porosity, specific surface area, and pore size distribution of the print against the original digital soil structure, was found to be substantially better for digital light synthesis, as compared to the other tested technologies. Repeatability, as determined by the same metrics but compared between identical prints, was found to be comparable across all printing technologies and did not significantly improve for prints at greater magnification (1.5$\times$). Wettability of the samples was improved by plasma treatment of the prints. The thorough analysis herein presented demonstrates that advanced, yet relatively inexpensive 3D printing approaches can be used to generate real-scale high quality analogs of soils/rocks that are much needed for experimental laboratory work. Such a method can open countless opportunities for studying the coupling of pore-structure and hydrodynamics on reactive mass transport in environmental science and engineering, soil science, and other subsurface related fields.

Thin water lenses floating on top of the main groundwater body are important for many natural and agricultural systems, owing to their different properties in terms of chemical composition or density compared to the surrounding groundwater. In settings with upward seeping groundwater, lenses may form that have thicknesses ranging from tens of centimeters to a few meters, making them prone to changing conditions in the short (seasonal) or long term (climate change). Knowing their thickness, shape, movement and mixing zone width may help in managing these lenses.

In a series of two papers, we present a mathematical description of the flow of water and transport of solute in a 2D cross-section between two parallel outflow faces and compare a simplified model to a complete model as described by the numerical code SUTRA. In this first paper of the series, we consider situations with a homogeneous density distribution. In the simplified model we employ the sharp interface approximation to obtain an expression for the stream function, the interface between the two types of water and the corresponding maximum lens thickness in steady state in the domain considered. This steady state description is used for travel time analyses and forms the basis for the transient analyses. For a typical example of oscillatory (e.g. seasonal) fluctuations in boundary conditions, we obtain expressions of the movement of the interface midway between two outflow faces by separating the problem into two timescales using the interface motion equation. This analysis provides insight into the importance of parameters on the vulnerability of water lenses under changing conditions, and may easily be extended to situations with abrupt or gradual changes in boundary conditions reflecting changes in land use or climate, respectively. Finally, we derive an analytical approximation of the mixing zone midway between the drains for steady state solutions, stepping away from the sharp interface approach. For a variety of examples, we validate the obtained expressions of the simplified mathematical model against the numerical model code SUTRA, which solves the fluid and solute mass balances explicitly.

Deep limestone aquifers are potential CO₂ storage sites, but CO₂-saturated brine reacts with the carbonate rock, changing its transport and storage properties. This study provided a preliminary investigation of the optimal injection rate of CO₂-saturated brine in carbonate rocks. Indiana limestone cores were subjected to CO₂-saturated brine injection at varied rates using an HPHT coreflooding setup with X-ray CT monitoring. The samples were characterized pre- and post-treatment in terms of porosity and pore size distribution using a gas porosimeter and NMR T₂ measurements. Moreover, the reaction was evaluated by measuring the aqueous effluent calcium ions concentration as a function of throughput using ICP-OES analysis. A high-resolution micro-CT scan was used to capture the dissolution post-treatments and characterize the wormhole's size and patterns. Results showed that the wormholes broke through to the sample exit face after injecting 160, 48, and 36 pore volumes at 0.5, 1, and 2 cm³/min, respectively thereby revealing the importance of injection velocity. The ICP-OES analysis revealed that a larger dissolution rate was achieved at 2 cm³/min, which explained the fast wormhole propagation. An increase in rock porosity and the pore-size distributions was observed after coreflooding on all samples with minimum precipitation, as concluded from the NMR T₂ relaxation time. A universal optimum Damköhler number can be obtained that enables calculating the optimum injection rate of CO₂-saturated brine at different rock and fluid conditions. We speculated that the optimum Damköhler number could be different from the value of 0.29 proposed by Fredd and Fogler (1998). This study provides a preliminary understanding of the optimal CO₂-saturated brine injection velocity that has an application for CO₂ storage, water alternating gas (WAG) operations, and acid stimulation of carbonate formations.

Mixing-limited reactions are central to a wide range of processes in natural and engineered porous media. Recent advances have shown that concentration gradients sustained by flow at the pore-scale influence macroscopic reaction rates over a large range of reactive transport regimes. Yet, resolving concentration gradients driven by fluid mixing at the pore-scale is challenging with current simulation methods. Here, we introduce a computational methodology to resolve concentration gradients at the pore scale in mixing-limited reactions. We consider a steady-state reactive transport problem characterized by reactive fluids flowing in parallel in a porous material. Given a mesh representation of the pore space and a steady velocity field, we solve the steady advection-diffusion equation for conservative scalar transport using a stabilized finite-element method combined with mesh refinement adapted to local scalar gradients. Based on this solution and assuming instantaneous reaction kinetics in the fluid, we infer the distribution of species involved in an irreversible bi-molecular reaction. We validate the method by comparing our results for uniform flow with analytical solutions and then apply it to simulate mixing-limited reactions in a three-dimensional random bead pack and Berea sandstone sample. Chaotic flow within the pore space leads to sustained concentration gradients, which are captured by our numerical framework. The results underscore the ability of the methodology to simulate transverse mixing and mixing-limited reactions in complex porous media and to provide bottom-up numerical data to improve the prediction of effective reaction rates at larger scales.

Non-Darcy flow through porous media is of great significance in hydraulics, oil and gas engineering, biomedical science, chemical and civil engineering etc. However, it is difficult to fully grasp the nature of fluid flow through porous media from macroscopic scale alone. Based on the statistically fractal scaling laws of pore structures, a new fractal pore-throat chain model (FPTCM) for non-Darcy flow through the isotropic porous media is developed. The analytical expressions for the Darcy and non-Darcy permeability as well as non-Darcy coefficient are derived accordingly. In order to explore the local flow field of high-speed non-Darcy flow through porous media, the finite element method is also carried out on an equivalent pore-throat unit. The predicted permeability by FPTCM shows better agreement with present numerical results and available experimental data, compared with commonly used semi-empirical formulas including Kozeny-Carman and Ergun equations. It has been found that both Darcy and non-Darcy permeability as well as non-Darcy coefficient strongly relate to the pore structures of porous media. The non-Darcy permeability is positively correlated to porosity, pore fractal dimension and Darcy permeability, while it is negatively related to tortuosity fractal dimension and pore size range. The non-Darcy coefficient shows opposite correlation with these parameters. The present work can provide theoretical basis for oil and gas development, nuclear waste treatment, carbon dioxide geological sequestration etc.

We unravel the nonmodal stability of a three-dimensional nonstratified Poiseuille flow in a saturated hyperporous medium constrained by impermeable rigid parallel plates. The primary objective is to broaden the scope of previous studies that conducted modal stability analysis for two-dimensional disturbances. Here, we explore both temporal and spatial transient disturbance energy growths for three-dimensional disturbances when the Reynolds number and porosity of the material are high, based on evolution equations with respect to time and space, respectively. Modal stability analysis reveals that the critical Reynolds number for the onset of shear mode instability increases as porosity increases. Moreover, the Darcy viscous drag term stabilizes shear mode instability, resulting in a delay in the transition from laminar flow to turbulence. In addition, it demonstrates the suppression of three-dimensional shear mode instability as the spanwise wavenumber increases, thereby confirming the statement of Squire’s theorem. By contrast, nonmodal stability analysis discloses that both temporal and spatial transient disturbance energy growths curtail as the effect of the Darcy viscous drag force intensifies. But their maximum values behave like $O\left(R{e}^2\right)$ for a fixed porous material, where $Re$ is the Reynolds number. However, for different porous materials, the scalings for both temporal and spatial transient disturbance energy growths are different. Furthermore, increasing porosity also suppresses both temporal and spatial disturbance energy growths. Finally, we observe that temporal transient disturbance energy growth becomes larger for a spanwise perturbation, while spatial transient disturbance energy growth becomes larger for a steady perturbation when angular frequency vanishes. The initial disturbance that excites the largest temporal energy amplification generates two sets of alternating high-speed and low-speed elongated streaks in the streamwise direction.