Search

A better understanding of the distribution of nonaqueous phase liquid (NAPL) plumes is of great importance to groundwater pollution remediation and control. However, the efficiency of surrogate models in simulating the transport is still not well addressed. Selecting a leakage problem as an example, 50 sets of random permeability distributions are generated using the Monte Carlo method, and a numerical model is used to obtain benchmark data of NAPL transport. Four machine learning methods are employed to simulate dense NAPL transport under point leakage sources across spatiotemporal scales. The validation of the models demonstrate that the random forest model can also effectively capture the spatial-temporal distribution of the plume in heterogeneous aquifers, with a maximum mean absolute error and root mean square error smaller than 5.55 x 10-4 and 5.88 10-5, respectively. Meanwhile, the multiple phase outcome from the random forest model fits well with the numerical results under the scenarios of linear leakage sources and light NAPL transport. The total time consumed in the computation is reduced by over 150 times after using the surrogate models. The results suggest that surrogate models can provide a prom)ising way to understand NAPL transport in heterogeneous aquifers [ABSTRACT FROM AUTHOR]

The review aims to provide a common understanding of the use of late-time drawdown to interpret aquifer pumping tests. The first part of the review provides an overview of the use of the late-time drawdown in literature to illustrate where and how the term is being used. A discussion on the practical implications of using the term and its significance is then presented. The review shows the use of the late-time drawdown in three main ways: the application of the Cooper and Jacob time-drawdown method, the description of the third segment of the unconfined aquifer drawdown-time curve, and when trying to estimate representative/ effective transmissivity parameters in heterogeneous aquifers. Unlike the other two situations, the use of late-time data in typical unconfined aquifers is supported by the groundwater flow principles and hence has a meaningful application. The aspects highlighted in this review are important to improve the theoretical and practical knowledge required for analysing and interpreting aquifer pumping test data. [ABSTRACT FROM AUTHOR]

[Display omitted] • Proposing CBLA as a remediation reagent for contaminated heterogeneous aquifers. • Pore-scale remediation simulation of organic pollutants with two-phase flow. • CBLA could improve the sweeping and extraction efficiency for PCE. • CBLA could improve the mobility and emulsification of PCE droplets. • The remediation efficiency of PCE was almost 100% at sandbox and micromodel scale. The heterogeneity of the aquifer seriously affects the remediation efficiency of dense non-aqueous phase liquid (DNAPL) due to the low contact efficiency for remediation reagents and DNAPL at macro scale and the restricted moving of DNAPL droplets by capillary forces in pore at micro scale. In this work, colloidal biliquid aphron (CBLA) as a remediation reagent was proposed, it has the rheological properties of shear thinning and the ultra-low interfacial tension between DNAPL. This work investigated the mechanisms and efficiencies of remediation DNAPL-contaminated heterogeneous aquifers using CBLA by sandbox and micromodel scale. The results indicated that CBLA could improve the mass transfer efficiency by both increasing the sweeping and extraction efficiency. The remediation efficiency of tetrachloroethylene (PCE) in aquifers with different permeability was above 99 % in sandbox, with the highest effluent concentration of 1.89 × 105 mg/L. Furthermore, at micromodel scale, CBLA could raise the remediation efficiency by improving the mobility and emulsification of PCE droplets. There was no PCE residue in the fine pore throats, and the remediation efficiency was 99.49 % in micromodel experiment and simulation. This work provided a promising remediation reagent for removing DNAPL from heterogeneous aquifers and laid a theoretical foundation for its practical application. [ABSTRACT FROM AUTHOR]

Subsurface heterogeneity could influence groundwater flow with implications on structure and productivity of aquifer ecosystems. Here we investigate effects of multifractal heterogeneity on topology of groundwater flow through MODFLOW simulations within a Monte Carlo framework. The results show that heterogeneity leads to focused groundwater advection and the creation of hotspots of bending‐type vortex flow in the fields. We demonstrate for the first time that the vortex structures characterized by Q criterion are 3‐D distributed and greatly deform surrounding pore‐water flow. The structures exhibit scale‐invariant features in multifractal fields and in stationary fields below the correlation scale, indicating that such vortex flow might be widely present with no characteristic scale. Complex spatial patterns of kinetic energy dissipation rate are identified for pore water flowing through heterogeneous porous media and correlate strongly with preferential flows. These findings are important for understanding solute fate and transport in aquifer systems. Plain Language Summary: The findings of this study provide insight into influences of heterogeneity on groundwater flow and solute transport processes in aquifers. Our results reveal that multifractal heterogeneity leads to focused regions of groundwater advection and creates hotspots of bending‐type vortex flow that are 3‐D and greatly deform surrounding pore water. It has strong implications for groundwater contamination and associated implementation of effective bioremediation strategies. This is because solutes/contaminants would be spun around these structures and therefore have less interaction with the soil matrix in the inner part of the vortices. We also demonstrate that the structures of the vortex flow are scale invariant in multifractal K fields and below the correlation scale of stationary K fields. It indicates that these vortex flow structures may be widely present in the field with no characteristic scale. Substantial energy dissipation along the high‐velocity zones also has strong implications for understanding activity, diversity, and the distribution of subterranean microorganisms. Key Points: Deformation and rotation of pore‐water flow occur near high vorticity zones, formed as blobs and correlated strongly to preferential flowsVorticity‐dominated blobs exhibit scale invariance in both multifractal K fields and up to the correlation scale in variogram K fieldsHeterogeneity dissipates substantial kinetic energy within large pore‐water velocity zones [ABSTRACT FROM AUTHOR]

Flow of mean velocity U takes place in a heterogeneous aquifer of random spatially variable conductivity K . A solute plume is injected instantaneously along a plane normal to U , over a large area relative to the logconductivity integral scale I (ergodic plume). Transport is by advection by the spatially variable Eulerian velocity. The study is focused on the derivation of the mean plume velocity in the four modes set forth by Kreft and Zuber [1978] for one dimensional flow in a homogeneous medium. In the resident injection mode the mass is initially distributed uniformly in space while in the flux mode it is proportional to the local velocity. In the resident detection mode the mean velocity pertains to the plume centroid, whereas in flux detection it is quantified with the aid of the BTC and the corresponding mean arrival time. In agreement with the literature, it is shown that U RR and U FF , pertaining to same injection and detection modes, either resident or flux, are equal to U . In contrast, in the mixed modes the solute velocity may differ significantly from U near the injection plane, approaching it at large distances relative to I . These effects are explained qualitatively with the aid of the exact solution for stratified aquifers. [ABSTRACT FROM AUTHOR]

We propose a computationally efficient probabilistic modeling methodology to estimate the adverse effects on humans of exposure to contaminated groundwater. Our work is aligned with the standard suggested by the regulatory agencies and allows to propagate uncertainty from hydrogeological, toxicological and behavioral parameters to the final health risk endpoint. The problem under consideration consists of a contaminated aquifer supplying water to a population. Contamination stems from a continuous source that feeds a steady plume which constitutes the hazard source . This scenario is particularly suited for NAPL pollutants. The erratic displacement of the contaminant plume in groundwater, due to the spatial variability of hydraulic conductivity, is characterized within the Lagrangian stochastic framework which enables the complete probabilistic characterization of the contaminant concentration at an environmentally sensitive location. Following the probabilistic characterization of flow and transport, we quantify the adverse health effects on humans. The dose response assessment involves the estimation of the uncertain effects of the exposure to a given contaminant while accounting for the exposed individual’s metabolism. The model integrates groundwater transport, exposure and human metabolism in a comprehensive probabilistic framework which allows the assessment of the risk probability through a novel simple analytical solution. Aside from its computational efficiency, the analytical features of the framework allows the assessment of uncertainty arising from the hydrogeological parameters. [ABSTRACT FROM AUTHOR]

The direct effect of changes in applied total stresses on fluid-saturated geologic materials is reflected in fluctuations in pore pressure. The quantification of these pressure changes has traditionally been based on the classical poroelasticity theory, developed ignoring the natural variability in aquifer properties (parameters). Uncertainty of excess pore pressure prediction is therefore expected to be large when applying the classical model to field aquifer systems which generally have high degree of heterogeneity. Very limited attention to this issue has been paid. The motivation for this work is to quantify the uncertainty associated with the classical model, the spatial variability of the predicted excess pressure head in the field, where the groundwater flow is produced by the changes in total stress. The stochastic methodology used to develop the results of this study is based on the nonstationary spectral approach. It is found that the heterogeneity and correlation length scale of the log hydraulic conductivity process, and the pore compressibility parameter play essential roles in enhancing the variability in excess pressure head. In addition, the pore pressure head distribution predicted using the classical poroelasticity theory is subject to large uncertainty at a large depth in heterogeneous aquifers. [ABSTRACT FROM AUTHOR]

Heterogeneity in aquifer permeability, which creates paths of varying mass flux and spatially complex contaminant plumes, can complicate the interpretation of contaminant fate and transport in groundwater. Identifying the location of high mass flux paths is critical for the reliable estimation of solute transport parameters and design of groundwater remediation schemes. Dipole flow tracer tests (DFTTs) and push-pull tests (PPTs) are single well forced-gradient tests which have been used at field-scale to estimate aquifer hydraulic and transport properties. In this study, the potential for PPTs and DFTTs to resolve the location of layered high- and low-permeability layers in granular porous media was investigated with a pseudo 2-D bench-scale aquifer model. Finite element fate and transport modelling was also undertaken to identify appropriate set-ups for in situ tests to determine the type, magnitude, location and extent of such layered permeability contrasts at the field-scale. The characteristics of flow patterns created during experiments were evaluated using fluorescent dye imaging and compared with the breakthrough behaviour of an inorganic conservative tracer. The experimental results show that tracer breakthrough during PPTs is not sensitive to minor permeability contrasts for conditions where there is no hydraulic gradient. In contrast, DFTTs are sensitive to the type and location of permeability contrasts in the host media and could potentially be used to establish the presence and location of high or low mass flux paths. Numerical modelling shows that the tracer peak breakthrough time and concentration in a DFTT is sensitive to the magnitude of the permeability contrast (defined as the permeability of the layer over the permeability of the bulk media) between values of 0.01–20. DFTTs are shown to be more sensitive to deducing variations in the contrast, location and size of aquifer layered permeability contrasts when a shorter central packer is used. However, larger packer sizes are more likely to be practical for field-scale applications, with fewer tests required to characterise a given aquifer section. The sensitivity of DFTTs to identify layered permeability contrasts was not affected by test flow rate. [ABSTRACT FROM AUTHOR]

Summary Due to the irregular distributions of aquifer hydraulic properties, the detail on the characterization of flow field cannot be anticipated. There can be a great degree of uncertainty in the prediction of heat transport processes anticipated in applying the traditional deterministic transport equation to field situations. This article is therefore devoted to quantification of uncertainty involving predictions over larger scales in terms of the temperature variance. A stochastic frame of reference is adopted to account for the spatial variability in hydraulic conductivity and specific discharge. Within this framework, the use of the first-order perturbation approximation and spectral representation leads to stochastic differential equations governing the mean behavior and perturbation of the temperature field in heterogeneous aquifers. It turns out that the mean equation developed in this sense is equivalent to the traditional deterministic transport equation and the temperature variance gives a measure of the prediction uncertainty from the traditional transport equation. The closed-form expression for the temperature variance developed here indicates that the controlling parameters such as the correlation scale of specific discharge, which measures the spatial persistence of the flow field, and the periodicity of the source term tend to increase the variability in temperature field in heterogeneous aquifers. The uncertainty of the traditional heat transport model increases as the penetration depth of thermal front through the aquifer increases. This suggests that prediction of temperature distribution using the traditional heat transport model in heterogeneous aquifers is expected to be subject to large uncertainty at a large depth (in the downstream region). For the management purpose, the variance of temperature could serve as a calibration target when applying the traditional model to field situations. It may be more reasonable to make conclusions from, say, the mean temperature with one or two standard deviations rather than only the mean temperature drawn from the traditional heat transport equation. [ABSTRACT FROM AUTHOR]

Highlights: [•] Pumping start is modeled by an immediate shift between two steady-state flow fields. [•] Ages deduced from CFCs and SF6 concentrations evolve in two distinct phases. [•] Transient flow patterns affect ages just after the pumping start but quickly vanish. [•] Atmospheric concentrations then transiently weight the residence time distribution. [•] Similar in terms of magnitude, these two regimes are helpful for models segregation. [Copyright &y& Elsevier]

We perform global sensitivity analysis (GSA) through polynomial chaos expansion (PCE) on a contaminant transport model for the assessment of radionuclide concentration at a given control location in a heterogeneous aquifer, following a release from a near surface repository of radioactive waste. The aquifer hydraulic conductivity is modeled as a stationary stochastic process in space. We examine the uncertainty in the first two (ensemble) moments of the peak concentration, as a consequence of incomplete knowledge of (a) the parameters characterizing the variogram of hydraulic conductivity, (b) the partition coefficient associated with the migrating radionuclide, and (c) dispersivity parameters at the scale of interest. These quantities are treated as random variables and a variance-based GSA is performed in a numerical Monte Carlo framework. This entails solving groundwater flow and transport processes within an ensemble of hydraulic conductivity realizations generated upon sampling the space of the considered random variables. The Sobol indices are adopted as sensitivity measures to provide an estimate of the role of uncertain parameters on the (ensemble) target moments. Calculation of the indices is performed by employing PCE as a surrogate model of the migration process to reduce the computational burden. We show that the proposed methodology (a) allows identifying the influence of uncertain parameters on key statistical moments of the peak concentration (b) enables extending the number of Monte Carlo iterations to attain convergence of the (ensemble) target moments, and (c) leads to considerable saving of computational time while keeping acceptable accuracy. [ABSTRACT FROM AUTHOR]

Summary: Environmental tracers have been extensively used to get characteristic groundwater velocities. When sampled in discharge zones, we show that they can also be efficiently used to get more integrative information on aquifer structure, which is especially valuable in complex geological settings like hard-rock contexts. Used in combination with generally available hydraulic data, they give useful prediction of the residence time distribution. Our analysis is illustrated on the produced crystalline aquifer of Plœmeur (Brittany, France) formed at a dipping contact zone between granites and micaschists. Although constrained on the same hydraulic head and CFC-12 data sampled at the produced well, full steady-state hydrogeological models and simpler exponential-like models yield predictions of the renewal rate and of the cumulative residence time that may differ by a factor 2. Differences come predominantly from the km-scale geological structures and topographical features that cannot be resolved by statistical models and are nonetheless essential within the characteristic dating range of the CFCs (0–70years). Our analysis highlights the interest of combining hydraulic and age information for the prediction of residence time distributions within full hydrogeological models, and the capacity of identifying global hydrogeological structures from point-like data. [ABSTRACT FROM AUTHOR]

Abstract: Large-scale advective transport through highly heterogeneous 3D formations is investigated using highly resolved numerical simulations and simple analytic models. Investigations are focused on impacts of two types of contaminant injection on transport through isotropic formations where flow conditions are uniform in the average. Transport is quantified by analyzing breakthrough curves for control planes at various distances from the injection zone. In flux-proportional injection mode local mass in injection zone is proportional to local groundwater flux; this setup models many practical cases such as contaminant injection through wells. In resident concentration mode local concentration in injection zone is constant. Results show that impacts of injection mode on breakthrough curves and their moments are strong and they persist for hundreds of correlation scales. The resident concentration mode leads to a fatter tails of the breakthrough curves, while the peaks are generally underpredicted. For a synthetic porous medium with logconductivity variance of 8, dispersivity computed using resident concentration mode at control plane 100 integral scales away from the injection zone was about 10 times larger than corresponding one for flux-proportional mode. Hence, injection mode impacts on transport through highly heterogeneous formations are strong and they persist for large distances from the injection zone. [ABSTRACT FROM AUTHOR]

Characterisation of transmissivity heterogeneity is critical for groundwater flow and solute transport. The heterogeneity of transmissivity is studied through variogram-based techniques. Conventionally, the parameters in variogram are obtained by fitting measurements to a theoretical variogram. However, conditioning to the sampled geological variables neglects the effects of observed concentration data. This paper presents a coupled inverse modelling system conditioning to both types of measurement. The results of a hypothetical two-dimensional steady flow indicated that the description of transmissivity and solute concentration field was improved when both measured transmissivity data and solute concentration data were combined in the coupled inverse modelling system. [ABSTRACT FROM AUTHOR]

• Analyses of gravity changes while pumping in 3D randomly heterogeneous aquifers. • Probability distributions of head and gravimetric variations during well operation. • Randomly heterogeneous domains consider uncertain geostatistical system descriptors. • The variance of heterogeneous fields affects mean and variance of gravity changes. • The correlation scale of the field affects extreme values of gravity changes. We evaluate the relative importance of the uncertainty related to parameters characterizing partially saturated groundwater flow on head and gravity changes associated with pumping tests taking place in homogeneous and heterogeneous porous media. We frame our study in a Global Sensitivity Analysis setting and assess the way imperfect knowledge of such parameters influences the probability distribution (pdf) of head and gravimetric variations recorded during well operation. We rely on a set of detailed computational analyses and conceptualize uncertain model parameters as random quantities. Randomly heterogeneous domains are treated by considering main geostatistical descriptors (i.e., variance and correlation scale) of three-dimensional spatial distributions of system properties as affected by uncertainty. We quantify the effects of the latter on the resulting pdf of (ensemble) mean and variance of head and gravity changes through a numerical Monte Carlo approach. While all uncertain parameters are influential to gravity changes in both homogeneous and randomly heterogeneous scenarios, consistent with the integral nature of gravity observations, our study enables us to quantify their relative importance. Values of Ensemble mean and variance of head and gravity changes associated with randomly heterogeneous fields are generally more influenced by the variance rather than by the correlation scale of the spatially heterogeneous parameters considered. Uncertainty in the correlation scale is more influential to the shape, and hence on extreme values, of the probability distribution of these moments. [ABSTRACT FROM AUTHOR]

Summary: The influence of the initial plume scale on the early time variability in cumulative solute mass across a control plane for advection dominated nonreactive solute movement in a heterogeneous three-dimensional aquifer is investigated. Under an assumption of a Gaussian velocity distribution, the closed-form expressions for the early time probability distribution function of travel times and coefficient of variation of cumulative solute mass are derived for the case of statistically isotropic media as functions of the initial plume size, the variance of the log hydraulic conductivity, the correlation scale of the log hydraulic conductivity, and the mean longitudinal velocity. These results are achieved by invoking a modified variance of longitudinal velocity, accounting for the influence of a finite domain size. It is found that at early time, the initial plume size has significant influence on the variability in cumulative solute mass in the case of a finite volume source. The presented methodology can be extended to statistically anisotropic media. [Copyright &y& Elsevier]

Abstract: Flowing fluid electric conductivity logging provides a means to determine hydrologic properties of fractures, fracture zones, or other permeable layers intersecting a borehole in saturated rock. The method involves analyzing the time-evolution of fluid electric conductivity (FEC) logs obtained while the well is being pumped and yields information on the location, hydraulic transmissivity, and salinity of permeable layers. The original analysis method was restricted to the case in which flows from the permeable layers or fractures were directed into the borehole (inflow). Recently, the method was adapted to permit treatment of both inflow and outflow, including analysis of natural regional flow in the permeable layer. A numerical model simulates flow and transport in the wellbore during flowing FEC logging, and fracture properties are determined by optimizing the match between simulation results and observed FEC logs. This can be a laborious trial-and-error procedure, especially when both inflow and outflow points are present. Improved analyses methods are needed. One possible tactic would be to develop an automated inverse method, but this paper takes a more elementary approach and focuses on identifying the signatures that various inflow and outflow features create in flowing FEC logs. The physical insight obtained provides a basis for more efficient analysis of these logs, both for the present trial and error approach and for a potential future automated inverse approach. Inflow points produce distinctive signatures in the FEC logs themselves, enabling the determination of location, inflow rate, and ion concentration. Identifying outflow locations and flow rates typically requires a more complicated integral method, which is also presented in this paper. [Copyright &y& Elsevier]

The local scale dispersion tensor, Dd, is a controlling parameter for the dilution of concentrations in a solute plume that is displaced by groundwater flow in a heterogeneous aquifer. In this paper, we estimate the local scale dispersion from time series or breakthrough curves, BTCs, of Br− concentrations that were measured at several points in a fluvial aquifer during a natural gradient tracer test at Krauthausen. Locally measured BTCs were characterized by equivalent convection dispersion parameters: equivalent velocity, veq(x) and expected equivalent dispersivity, 〈λeq(x)〉. A Lagrangian framework was used to approximately predict these equivalent parameters in terms of the spatial covariance of loge transformed conductivity and the local scale dispersion coefficient.The approximate Lagrangian theory illustrates that 〈λeq(x)〉 increases with increasing travel distance and is much larger than the local scale dispersivity, λd. A sensitivity analysis indicates that 〈λeq(x)〉 is predominantly determined by the transverse component of the local scale dispersion and by the correlation scale of the hydraulic conductivity in the transverse to flow direction whereas it is relatively insensitive to the longitudinal component of the local scale dispersion. By comparing predicted 〈λeq(x)〉 for a range of Dd values with 〈λeq(x)〉 obtained from locally measured BTCs, the transverse component of Dd, DdT, was estimated. The estimated transverse local scale dispersivity, λdT=DdT/U1 (U1=mean advection velocity) is in the order of 101–102 mm, which is relatively large but realistic for the fluvial gravel sediments at Krauthausen. [Copyright &y& Elsevier]

• Multiscale characterization of the effective conductivity distributions. • Improved analytical results using the energy dissipation formulation. • Efficient numerical estimator valid at all scales using that formulation. • Compare scale dependence of full pdf, mean and variance with two other estimators. In large scale heterogeneous aquifer simulations, determining the appropriate coarsening scale λ to define an effective hydraulic conductivity K eff is a challenging task, that involves a trade-off between accuracy and cost. Efficiently adjusting the scale λ is then key, in particular for uncertainty quantification. In this paper, we obtain improved analytical results for the variance of K eff , valid at any scale, in the context of energy dissipation formulation. Using this formulation, we then derive an efficient K eff numerical estimator, and compare it with those of the potential-flow average and permeameter formulations in 2D, for lognormal and binary media, over a wide range of λ and of heterogeneity. We analyze the probability density function (pdf), mean, and variance, of these estimators, comparing them with the analytical results. In the lognormal case, the pdf's are rather similar for the three estimators, and remain lognormal at all scales. In the binary case, slow convergence to an asymptotic regime is observed close to the percolation threshold. Image, graphical abstract [ABSTRACT FROM AUTHOR]

• A numerical model incorporating aquifer heterogeneities was developed to evaluate virus removal during ASR. • The sensitivity of virus removal to key operational and physicochemical parameters was analyzed. • The relative importance of physical heterogeneity increased under physicochemical conditions that reduced virus removal. • Sticking efficiency of viruses was found to significantly influence virus concentration in the recovered water. A quantitative understanding of virus removal during aquifer storage and recovery (ASR) in physically and geochemically heterogeneous aquifers is needed to accurately assess human health risks from viral infections. A two-dimensional axisymmetric numerical model incorporating processes of virus attachment, detachment, and inactivation in aqueous and solid phases was developed to systematically evaluate the virus removal performance of ASR schemes. Physical heterogeneity was considered as either layered or randomly distributed hydraulic conductivities (with selected variance and horizontal correlation length). Geochemical heterogeneity in the aquifer was accounted for using Colloid Filtration Theory to predict the spatial distribution of attachment rate coefficient. Simulation results demonstrate that the combined effects of aquifer physical heterogeneity and spatial variability of attachment rate resulted in higher virus concentrations in the recovered water at the ASR well (i.e. reduced virus removal). While the sticking efficiency of viruses to aquifer sediments was found to significantly influence virus concentration in the recovered water, the solid phase inactivation under realistic field conditions combined with the duration of storage phase had a predominant influence on the overall virus removal. The relative importance of physical heterogeneity increased under physicochemical conditions that reduced virus removal (e.g. lower value of sticking efficiency or solid phase inactivation rate). This study provides valuable insight on site selection of ASR projects and an approach to optimize ASR operational parameters (e.g. storage time) for virus removal and to minimize costs associated with post-recovery treatment. [ABSTRACT FROM AUTHOR]