Fractal travel time estimates for dispersive contaminants

Alternative fractional models of contaminant transport lead to a new travel time formula for arbitrary concentration levels. For an evolving contaminant plume in a highly heterogeneous aquifer, the new formula predicts much earlier arrival at low concentrations. Travel times of contaminant fronts and plumes are often obtained from Darcy's law calculations using estimates of average pore velocities. These estimates only provide information about the travel time of the average concentration (or peak, for contaminant pulses). Recently, it has been shown that finding the travel times of arbitrary concentration levels is a straightforward process, and equations were developed for other portions of the breakthrough curve for a nonreactive contaminant. In this paper, we generalize those equations to include alternative fractional models of contaminant transport.     

Effective macroscopic transport parameters between parallel plates with constant concentration boundaries

A macroscopic transport model is developed, following the Taylor shear dispersion analysis procedure, for a 2D laminar shear flow between parallel plates possessing a constant specified concentration. This idealized geometry models flow with contaminant dissolution at pore-scale in a contaminant source zone and flow in a rock fracture with dissolving walls. We upscale a macroscopic transient transport model with effective transport coefficients of mean velocity, macroscopic dispersion, and first-order mass transfer rate. To validate the macroscopic model the mean concentration, covariance, and wall concentration gradient are compared to the results of numerical simulations of the advection–diffusion equation and the Graetz solution. Results indicate that in the presence of local-scale variations and constant concentration boundaries, the upscaled mean velocity and macrodispersion coefficient differ from those of the Taylor–Aris dispersion, and the mass transfer flux described by the first-order mass transfer model is larger than the diffusive mass flux from the constant wall. In addition, the upscaled first-order mass transfer coefficient in the macroscopic model depends only on the plate gap and diffusion coefficient. Therefore, the upscaled first-order mass transfer coefficient is independent of the mean velocity and travel distance, leading to a constant pore-scale Sherwood number of 12. By contrast, the effective Sherwood number determined by the diffusive mass flux is a function of the Peclet number for small Peclet number, and approaches a constant of 10.3 for large Peclet number.     

The Horizontal Reactive Media Treatment Well (HRX Well®) for Passive In‐Situ Remediation

A new in‐situ remediation concept termed a Horizontal Reactive Media Treatment Well (HRX Well®) is presented that utilizes horizontal wells filled with reactive media to passively treat contaminated groundwater in‐situ. The approach involves the use of large‐diameter directionally drilled horizontal wells filled with granular reactive media generally installed parallel to the direction of groundwater flow. The design leverages natural “flow‐focusing” behavior induced by the high in‐well hydraulic conductivity of the reactive media relative to the aquifer hydraulic conductivity to passively capture and treat proportionally large volumes of groundwater within the well. Clean groundwater then exits the horizontal well along its downgradient sections. Many different types of solid granular reactive media are already available (e.g., zero valent iron, activated carbon, ion exchange resins, zeolite, apatite, chitin); therefore, this concept could be used to address a wide range of contaminants. Three‐dimensional flow and transport simulations were completed to assess the general hydraulic performance, capture zones, residence times, effects of aquifer heterogeneity, and treatment effectiveness of the concept. The results demonstrate that capture and treatment widths of up to tens of feet can be achieved for many aquifer settings, and that reductions in downgradient concentrations and contaminant mass flux are nearly immediate. For a representative example, the predicted treatment zone width for the HRX Well is approximately 27 to 44 feet, and contaminant concentrations immediately downgradient of the HRX Well decreased an order of magnitude within 10 days. A series of laboratory‐scale physical tests (i.e., tank tests) were completed that further demonstrate the concept and confirm model prediction performance. For example, the breakthrough time, peak concentration and total mass recovery of methylene blue (reactive tracer) was about 2, 35, and 20 times (respectively) less than chloride (conservative tracer) at the outlet of the tank‐scale HRX Well.     

Stochastic evaluation of subsurface contaminant discharges under physical,chemical, and biological heterogeneities

A finite element 2D Monte Carlo approach is used to evaluate the sensitivity of groundwater contaminant discharges to a Damkohler number ω and spatial variability in aquifer hydraulic conductivity, initial microbial biomass concentrations, and electron acceptor/donor concentrations. Bioattenuation is most sensitive to spatial variations in incipient biomass and critical electron donors/acceptors for ω ≥ 1 (i.e., when pore-water residence times are high compared to the time needed for microbial growth or contaminant attenuation). Under these conditions, critical reaction processes can become substrate-limited at multiple locations throughout the aquifer; which in turn increases expected contaminant discharges and their uncertainties at monitored transects. For ω ≤ 0.2, contaminant discharge is not sensitive to incipient biomass variations. Physical heterogeneities expedite plume arrival and delay departure at transects and in turn attenuate peak discharges but do not affect cumulative contaminant discharges. Physical heterogeneities do, however, induce transect mass discharge variances that are bimodal functions of time; the first peak beings consistently higher. A simple stream tube model is invoked to explain the occurrence of peaks in contaminant discharge variance.     

Concentration rebound following in situ chemical oxidation in fractured clay

A two-dimensional, transient-flow, and transport numerical model was developed to simulate in situ chemical oxidation (ISCO) of trichloroethylene and tetrachloroethylene by potassium permanganate in fractured clay. This computer model incorporates dense, nonaqueous phase liquid dissolution, reactive aquifer material, multispecies matrix diffusion, and kinetic formulations for the oxidation reactions. A sensitivity analysis for two types of parameters, hydrogeological and engineering, including matrix porosity, matrix organic carbon, fracture aperture, potassium permanganate dosage, and hydraulic gradient, was conducted. Remediation metrics investigated were the relative rebound concentrations arising from back diffusion and percent mass destroyed. No well-defined correlation was found between the magnitude of rebound concentrations during postremedy monitoring and the amount of contaminant mass destroyed during the application. Results indicate that all investigated parameters affect ISCO remediation in some form. Results indicate that when advective transport through the fracture is dominant relative to diffusive transport into the clay matrix (large System Peclet Number), permanganate is more likely to be flushed out of the system and treatment is not optimal. If the System Peclet Number is too small, indicating that diffusion into the matrix is dominant relative to advection through the fracture, permanganate does not traverse the entire fracture, leading to postremediation concentration rebound. Optimal application of ISCO requires balancing advective transport through the fracture with diffusive transport into the clay matrix.     

Analytical model for fully three‐dimensional radial dispersion in a finite‐thickness aquifer

Analytical solutions for contaminant transport in a non‐uniform flow filed are very difficult and relatively rare in subsurface hydrology. The difficulty is because of the fact that velocity vector in the non‐uniform flow field is space‐dependent rather than constant. In this study, an analytical model is presented for describing the three‐dimensional contaminant transport from an area source in a radial flow field which is a simplest case of the non‐uniform flow. The development of the analytical model is achieved by coupling the power series technique, the Laplace transform and the two finite Fourier cosine transform. The developed analytical model is examined by comparing with the Laplace transform finite difference (LTFD) solution. Excellent agreements between the developed analytical model and the numerical model certificate the accuracy of the developed model. The developed model can evaluate solution for Peclet number up to 100. Moreover, the mathematical behaviours of the developed solution are also studied. More specifically, a hypothetical convergent flow tracer test is considered as an illustrative example to demonstrate the three‐dimensional concentration distribution in a radial flow field. The developed model can serve as benchmark to check the more comprehensive three‐dimensional numerical solutions describing non‐uniform flow contaminant transport. Copyright © 2009 John Wiley & Sons, Ltd.     

Error analysis of finite difference methods for two-dimensional advection–dispersion–reaction equation

In this paper, the numerical errors associated with the finite difference solutions of two-dimensional advection–dispersion equation with linear sorption are obtained from a Taylor analysis and are removed from numerical solution. The error expressions are based on a general form of the corresponding difference equation. The variation of these numerical truncation errors is presented as a function of Peclet and Courant numbers in X and Y direction, a Sink/Source dimensionless number and new form of Peclet and Courant numbers in X–Y plane. It is shown that the Crank–Nicolson method is the most accurate scheme based on the truncation error analysis. The effects of these truncation errors on the numerical solution of a two-dimensional advection–dispersion equation with a first-order reaction or degradation are demonstrated by comparison with an analytical solution for predicting contaminant plume distribution in uniform flow field. Considering computational efficiency, an alternating direction implicit method is used for the numerical solution of governing equation. The results show that removing these errors improves numerical result and reduces differences between numerical and analytical solution.     

Estimating Reaction Rate Coefficients Within a Travel‐Time Modeling Framework

A generalized, efficient, and practical approach based on the travel‐time modeling framework is developed to estimate in situ reaction rate coefficients for groundwater remediation in heterogeneous aquifers. The required information for this approach can be obtained by conducting tracer tests with injection of a mixture of conservative and reactive tracers and measurements of both breakthrough curves (BTCs). The conservative BTC is used to infer the travel‐time distribution from the injection point to the observation point. For advection‐dominant reactive transport with well‐mixed reactive species and a constant travel‐time distribution, the reactive BTC is obtained by integrating the solutions to advective‐reactive transport over the entire travel‐time distribution, and then is used in optimization to determine the in situ reaction rate coefficients. By directly working on the conservative and reactive BTCs, this approach avoids costly aquifer characterization and improves the estimation for transport in heterogeneous aquifers which may not be sufficiently described by traditional mechanistic transport models with constant transport parameters. Simplified schemes are proposed for reactive transport with zero‐, first‐, nth‐order, and Michaelis‐Menten reactions. The proposed approach is validated by a reactive transport case in a two‐dimensional synthetic heterogeneous aquifer and a field‐scale bioremediation experiment conducted at Oak Ridge, Tennessee. The field application indicates that ethanol degradation for U(VI)‐bioremediation is better approximated by zero‐order reaction kinetics than first‐order reaction kinetics.     

Uncertainty propagation of hydrodispersive transfer in an aquifer: an illustration of one-dimensional contaminant transport with slug injection

It is evident that the hydrodynamic dispersion coefficient and linear flow velocity dominate solute transport in aquifers. Both of them play important roles characterizing contaminant transport. However, by definition, the parameter of contaminant transport cannot be measured directly. For most problems of contaminant transport, a conceptual model for solute transport generally is established to fit the breakthrough curve obtained from field testing, and then suitable curve matching or the inverse solution of a theoretical model is used to determine the parameter. This study presents a one-dimensional solute transport problem for slug injection. Differential analysis is used to analyze uncertainty propagation, which is described by the variance and mean. The uncertainties of linear velocity and hydrodynamic dispersion coefficient are, respectively, characterized by the second-power and fourth-power of the length scale multiplied by a lumped relationship of variance and covariance of system parameters, i.e. the Peclet number and arrival time of maximum concentration. To validate the applicability for evaluating variance propagation in one-dimensional solute transport, two cases using field data are presented to demonstrate how parametric uncertainty can be caught depending on the manner of sampling.     

Biocurtain Design Using Reactive Transport Models

Bioremediation is an attractive alternative to traditional remediation methods for a variety of ground water contaminants. However, widespread implementation of bioremediation is currently limited by the complexity of the dynamic chemical and biological processes that need to be understood and incorporated into the design approach. Reactive transport models provide a powerful tool to simulate these complex interactions and, thus, can be used to improve and guide the design of bioremediation systems. We present a remediation design approach for intermittently stimulated biodegradation using multicomponent reactive transport models, parameterized using a series of nondimensional Damkohler numbers. Designs were based on either (1) a target aqueous phase concentration at the exit of the treatment system, or (2) the total contaminant mass fraction removed from a region of interest. The equation set used to develop this design approach is specific to the case of intermittent electron donor addition to drive cometabolic transformations. We illustrate the design procedure for a biocurtain that removes carbon tetrachloride. Our results for this case indicate that intermittent injection is significantly more efficient than strategies based on continuous pumping. Example design parameters include the length of the biologically active zone (i.e., biocurtain), the effective rate of degradation in this zone, and the interval between electron donor injection cycles. The presented dimensionless parametric approach can be used to design bench-scale column studies and should be helpful for scale-up to field-scale remediation systems.     

Analysis of a vertical dipole tracer test in highly fractured rock

The results of a vertical dipole tracer experiment performed in highly fractured rocks of the Clare Valley, South Australia, are presented. The injection and withdrawal piezometers were both screened over 3 m and were separated by 6 m (midpoint to midpoint). Due to the long screen length, several fracture sets were intersected, some of which do not connect the two piezometers. Dissolved helium and bromide were injected into the dipole flow field for 75 minutes, followed by an additional 510 minutes of flushing. The breakthrough of helium was retarded relative to bromide, as was expected due to the greater aqueous diffusion coefficient of helium. Also, only -25% of the total mass injected of both tracers was recovered. Modeling of the tracer transport was accomplished using an analytical one-dimensional flow and transport model for flow through a fracture with diffusion into the matrix. The assumptions made include: streamlines connecting the injection and withdrawal point can be modeled as a dipole of equal strength, flow along each streamline is one dimensional, and there is a constant Peclet number for each streamline. In contrast to many other field tracer studies performed in fractured rock, the actual travel length between piezometers was not known. Modeling was accomplished by fitting the characteristics of the tracer breakthrough curves (BTCs), such as arrival times of the peak concentration and the center of mass. The important steps were to determine the fracture aperture (240 microm) based on the parameters that influence the rate of matrix diffusion (this controls the arrival time of the peak concentration); estimating the travel distance (11 m) by fitting the time of arrival of the centers of mass of the tracers; and estimating fracture dispersivity (0.5 m) by fitting the times that the inflection points occurred on the front and back limbs of the BTCs. This method works even though there was dilution in the withdrawal well, the amount of which can be estimated by determining the value that the modeled concentrations need to be reduced to fit the data (approximately 50%). The use of two tracers with different diffusion coefficients was not necessary, but it provides important checks in the modeling process because the apparent retardation between the two tracers is evidence of matrix diffusion and the BTCs of both tracers need to be accurately modeled by the best fit parameters.     

Contaminant Spread and Flushing in Fractured Rocks Near Oak Ridge, Tennessee

Liquid wastes, including metals dissolved in nitric acid, were discharged into the S-3 Ponds from 1951 to 1983. During this period, contaminants in ground water spread along shallow fracture flow paths toward nearby streams. Also, a high concentration of nitrate in one well at a depth of 110 to 240 in shows that some contaminants may have moved downdip because of differences in fluid density. Neutralization of the ponds in June 1983 caused a dramatic decrease in the contaminant concentrations of Bear Creek about three months later. Since then, the contaminant concentrations of Bear Creek have decreased at a first-order exponential rate. This average rate, which is the same for both more reactive and less reactive constituents, can be interpreted to show that the contaminant reservoir consists of the unfractured rock matrix.
Flushing caused by the natural recharge and discharge of ground water is occurring at all locations, but contaminant concentrations are controlled by the relative rates of molecular diffusion from the rock matrix and advection along the fracture flow paths. Hushing has thus been most effective near the water table. If the exponential decrease in contaminant concentrations continues, water in Bear Creek will meet drinking water standards by 2012: regardless of any remedial action, contaminants will remain in the rocks for many years.     

Non-uniqueness of inverse transmissivity field calibration and predictive transport modeling

Recent work with stochastic inverse modeling techniques has led to the development of efficient algorithms for the construction of transmissivity (T) fields conditioned to measurements of T and head. Small numbers of calibration targets and correlation between model parameters in these inverse solutions can lead to a relatively large region in parameter space that will produce a near optimal calibration of the T field to measured heads. Most applications of these inverse techniques have not considered the effects of non-unique calibration on subsequent predictions made with the T fields. Use of these T fields in predictive contaminant transport modeling must take into account the non-uniqueness of the T field calibration. A recently developed ‘predictive estimation’ technique is presented and employed to create T fields that are conditioned to observed heads and measured T values while maximizing the conservatism of the associated predicted advective travel time. Predictive estimation employs confidence and prediction intervals calculated simultaneously on the flow and transport models, respectively. In an example problem, the distribution of advective transport results created with the predictive estimation technique is compared to the distribution of results created under traditional T field optimization where model non-uniqueness is not considered. The predictive estimation technique produces results with significantly shorter travel times relative to traditional techniques while maintaining near optimal calibration. Additionally, predictive estimation produces more accurate estimates of the fastest travel times.     

An Unsteady State Tracer Method for Characterizing Fractures in Bedrock Wells

Evaluating contaminants impacting wells in fractured crystalline rock requires knowledge of the individual fractures contributing water. This typically involves using a sequence of tools including downhole geophysics, flow meters, and straddle packers. In conjunction with each other these methods are expensive, time consuming, and can be logistically difficult to implement. This study demonstrates an unsteady state tracer method as a cost‐effective alternative for gathering fracture information in wells. The method entails introducing tracer dye throughout the well, inducing fracture flow into the well by conducting a slug test and then profiling the tracer concentration in the well to locate water contributing fractures where the dye has been diluted. By monitoring the development of the dilution zones within the wellbore with time, the transmissivity and the hydraulic head of the water contributing fractures can be determined. Ambient flow conditions and the contaminant concentration within the fractures can also be determined from the tracer dilution. This method was tested on a large physical model well and a bedrock well. The model well was used to test the theory underlying the method and to refine method logistics. The approach located the fracture and generated transmissivity values that were in excellent agreement with those calculated by slug testing. For the bedrock well tested, two major active fractures were located. Fracture location and ambient well conditions matched results from conventional methods. Estimates of transmissivity values by the tracer method were within an order of magnitude of those calculated using heat‐pulse flow meter data.     

Relationship between dual-domain parameters and practical characterization data

Dual-domain solute transport models produce significantly improved agreement to observations compared to single-domain (advection-dispersion) models when used in an a posteriori data fitting mode. However, the use of dual-domain models in a general predictive manner has been a difficult and persistent challenge, particularly at field-scale where characterization of permeability and flow is inherently limited. Numerical experiments were conducted in this study to better understand how single-rate mass transfer parameters vary with aquifer attributes and contaminant exposure. High-resolution reference simulations considered 30 different scenarios involving variations in permeability distribution, flow field, mass transfer timescale, and contaminant exposure time. Optimal dual-domain transport parameters were empirically determined by matching to breakthrough curves from the high-resolution simulations. Numerical results show that mobile porosity increases with lower permeability contrast/variance, smaller spatial correlation length, lower connectivity of high-permeability zones, and flow transverse to strata. A nonzero non-participating porosity improves empirical fitting, and becomes larger for flow aligned with strata, smaller diffusion coefficient, and larger spatial correlation length. The non-dimensional mass transfer coefficient or Damkohler number tends to be close to 1.0 and decrease with contaminant exposure time, in agreement with prior studies. The best empirical fit is generally achieved with a combination of macrodispersion and first-order mass transfer. Quantitative prediction of ensemble-average dual-domain parameters as a function of measurable aquifer attributes proved only marginally successful.     

An interpolation-corrected modified method of characteristics to solve advection-dispersion equations

Numerical dispersion, numerical oscillation, and peak clipping are common numerical difficulties in solving advection-dispersion equations. The development of numerical approaches that can handle these numerical difficulties with reasonable computational efforts is an ongoing challenge. In this paper, an interpolation-corrected modified method of characteristics (ICMMOC) is proposed for solving advection-dispersion equations. The ICMMOC is an improved version of the modified method of characteristics (MMOC). It uses a high-order (second-order or higher) interpolation scheme to reduce numerical dispersion and an interpolation-correction procedure to eliminate numerical oscillation. A simple peak capturing scheme to overcome the peak clipping problem is also developed in this study. Simulation results show that the ICMMOC is able to overcome the aforementioned numerical difficulties for a large range of grid Peclet numbers.     

Virtual rate and mean distance of travel of individual clasts in gravel-bed channels

Travel distances in gravel-bed rivers of tagged particles of various sizes are related to excess stream power estimated for peak discharge. Mean distance of movement, irrespective of grain size, is weakly correlated with stream power, especially near the threshold of movement. There may be several reasons for the weak correlation, including variable effects of bed structure, varying magnitudes of sediment mobilizing events and sampling problems. Grain size itself is of marginal significance. The virtual rate of travel calculated using total time for which the flow is larger than that needed to initiate clast movement also bears a weak relation to the excess stream power over the period. Better results are obtained by relating the virtual rate of travel to the first peak of the flow event only. This implies that the initial seeding of the tagged particles dominates the observations. Nonetheless, an underlying general relation is indicated by the data, which are derived from a wide range of flow regime types.