Chaotic mixing of tracer and vorticity by modulated travelling Rossby waves

Abstract

We consider the mixing of passive tracers and vorticity by temporally fluctuating large scale flows in two dimensions. In analyzing this problem, we employ modern developments stemming from properties of Hamiltonian chaos in the particle trajectories; these developments generally come under the heading “chaotic advection” or “Lagrangian turbulence.” A review of the salient properties of this kind of mixing, and the mathematics used to analyze it, is presented in the context of passive tracer mixing by a vacillating barotropic Rossby wave. We then take up the characterization of subtler aspects of the mixing. It is shown the chaotic advection produces very nonlocal mixing which cannot be represented by eddy diffusivity. Also, the power spectrum of the tracer field is found to be k ^{? l} at shortwaves—precisely as for mixing by homogeneous, isotropic two dimensional turbulence,—even though the physics of the present case is very different. We have produced two independent arguments accounting for this behavior.

We then examine integrations of the unforced barotropic vorticity equation with initial conditions chosen to give a large scale streamline geometry similar to that analyzed in the passive case. It is found that vorticity mixing proceeds along lines similar to passive tracer mixing. Broad regions of homogenized vorticity ultimately surround the separatrices of the large scale streamline pattern, with vorticity gradients limited to nonchaotic regions (regions of tori) in the corresponding passive problem.

Vorticity in the chaotic zone takes the form of an arrangement of strands which become progressively finer in scale and progressively more densely packed; this process transfers enstrophy to small scales. Although the enstrophy cascade is entirely controlled by the large scale wave, the shortwave enstrophy spectrum ultimately takes on the classical k ^{? l} form. If one accepts that the enstrophy cascade is indeed mediated by chaotic advection, this is the expected behavior. The extreme form of nonlocality (in wavenumber space) manifest in this example casts some doubt on the traditional picture of enstrophy cascade in the Atmosphere, which is based on homogeneous two dimensional turbulence theory. We advance the conjecture that these transfers are in large measure attributable to large scale, low frequency, planetary waves.

Upscale energy transfers amplifying the large scale wave do indeed occur in the course of the above-described process. However, the energy transfer is complete long before vorticity mixing has gotten very far, and therefore has little to do with chaotic advection. In this sense, the vorticity involved in the enstrophy cascade is “fossil vorticity,” which has already given up its energy to the large scale.

We conclude with some speculations concerning statistical mechanics of two dimensional flow, prompted by our finding that flows with identical initial energy and enstrophy can culminate in very different final states. We also outline prospects for further applications of chaotic mixing in atmospheric problems.     

Quantitative analysis of numerically induced mixing in a coastal model application

In recent years, various attempts have been made to estimate the amount of numerical mixing in numerical ocean models due to discretisation errors of advection schemes. In this study, a high-resolution coastal model using the ocean circulationmodel GETM is applied to the Western Baltic Sea, which is characterised by energetic and episodic inflows of dense bottom waters originating from the Kattegat. The model is equipped with an easy-to-implement diagnostic method for obtaining the numerical mixing which has recently been suggested. In this diagnostic method, the physical mixing is defined as the mean tracer variance decay rate due to turbulent mixing. The numerical mixing due to discretisation errors of tracer advection schemes is defined as the decay rate between the advected square of the tracer variance and the square of the advected tracer, which can be directly compared to the physical variance decay. The source and location of numerical mixing is further investigated by comparing different advection schemes and analysing the amount of numerical mixing in each spatial dimension during the advection time step. The results show that, for the setup used, the numerically and physically induced mixing have the same orders of magnitude but with different vertical and horizontal distributions. As the main mechanism for high numerical mixing, vertical advection of tracers with strong vertical gradients has been identified. The main reason for high numerical mixing is due to bottom-following coordinates when density gradients, especially for regions of steep slopes, are advected normal to isobaths. With the bottom-following coordinates used here, the horizontal gradients are reproduced by a spurious sawtooth-type profile where strong advection through, but not along, the vertical coordinate levels occurs. Additionally, the well known relation between strong tracer gradients and high velocities on the one and high numerical mixing on the other hand is approved quantitatively within this work.     

Measurement of LNAPL Flux Using Single‐Well Intermittent Mixing Tracer Dilution Tests

The stability of subsurface Light Nonaqueous Phase Liquids (LNAPLs) is a key factor driving expectations for remedial measures at LNAPL sites. The conventional approach to resolving LNAPL stability has been to apply Darcy's Equation. This paper explores an alternative approach wherein single‐well tracer dilution tests with intermittent mixing are used to resolve LNAPL stability. As a first step, an implicit solution for single‐well intermittent mixing tracer dilution tests is derived. This includes key assumptions and limits on the allowable time between intermittent mixing events. Second, single‐well tracer dilution tests with intermittent mixing are conducted under conditions of known LNAPL flux. This includes a laboratory sand tank study and two field tests at active LNAPL recovery wells. Results from the sand tank studies indicate that LNAPL fluxes in wells can be transformed into formation fluxes using corrections for (1) LNAPL thicknesses in the well and formation and (2) convergence of flow to the well. Using the apparent convergence factor from the sand tank experiment, the average error between the known and measured LNAPL fluxes is 4%. Results from the field studies show nearly identical known and measured LNAPL fluxes at one well. At the second well the measured fluxes appear to exceed the known value by a factor of two. Agreement between the known and measured LNAPL fluxes, within a factor of two, indicates that single‐well tracer dilution tests with intermittent mixing can be a viable means of resolving LNAPL stability.     

Feasibility of High Pressure Injection of Chemicals into the Subsurface for the Bioremediation of the Exxon Valdez Oil

Delivery of dissolved chemicals to bioremediate oil from the Exxon Valdez oil spill was investigated at Beach EL056C of Eleanor Island, Prince William Sound, Alaska. The delivery technique was high pressure injection (HPI) of an inert tracer, lithium, at the approximate depth of 1.0 m into the beach near the mid‐tide line. The results revealed that the maximum injection flow rate was 3.0 L/min and the associated pressure was around 196 kPa. Therefore, exceeding any of these values would probably cause failure of the injection system. The injected tracer was monitored at multiple depths of four surrounding observation wells, and the results showed that the tracer plume occupied an area of 12 m² within 24 h. The tracer plume traveled at the average speeds of 10 m/d in the seaward direction and 1.7 m/d in other directions. The rapid movement under HPI and the large diameter of influence (3.0 m) indicated that bioremediation of the Exxon Valdez oil on this beach via injection of chemicals is logistically feasible.     

Spatial variability of in situ microbial activity: biotracer tests

Biotracer tests have been proposed as a means by which to characterize the in situ biodegradation potential for field-scale systems. In this study, field experiments were conducted at two sites to evaluate the utility of the biotracer method for characterizing the spatial variability of microbial activity. The first site is a mixed waste-contaminated surficial aquifer in Utah, and the second site is a chlorinated solvent-contaminated regional aquifer in Tucson, Arizona. Mass recovery of the biotracer decreased approximately linearly with increasing residence time for the Tucson site. Similar behavior was observed at the Utah site, except in the region adjacent to the injection zone, where percent recoveries were much lower than those predicted using a correlation determined using data collected downgradient of the injection zone. First-order biodegradation rate coefficients obtained from model calibration of the tracer data varied between 0.2 and 0.5/day for the Tucson site. For the Utah site, the values varied between 0.1 and 0.6/day downgradient of the injection wells, and between 0.7 and 2.6/day near the injection wells. Considering the large range over which biodegradation rate coefficients can vary, the rate coefficient exhibited relatively minimal spatial variability (factor of 2.5) for the Tucson site. Conversely, the spatial variability of the rate coefficient was an order of magnitude greater for the Utah site. These differences in variability are consistent with conditions associated with the respective sites. For example, the greater microbial activity observed in the vicinity of the injection wells for the Utah site is consistent with the biomass distribution determined from analysis of core samples, which shows larger bacterial cell densities for the region near the injection wells. These results illustrate the utility of biotracer tests for in situ characterization of microbial activity (e.g., biodegradation potential), including evaluation of potential spatial variability.     

Time-scales of hybridisation of magmatic enclaves in regular and chaotic flow fields: petrologic and volcanologic implications

This paper describes numerical models of advection/diffusion between enclaves and host magmas, applied with the aim of estimating time-scales during which enclaves can be homogenised. In particular, advection was simulated using a numerical system consisting of regular and chaotic regions. Results indicate that the homogenisation time of enclaves in chaotic regions is several orders of magnitude faster than in regular regions. For instance, an enclave with a diameter of 100 cm may be homogenised in the chaotic region in ∼ 380 years, assuming an advection velocity of 10 cm/year, whereas in the regular region it would require 6.5×10⁵ years for complete homogenisation. This implies that, in the same magmatic system, large differences in the degree of homogenisation may co-exist, generating magmatic masses with large spatial and temporal inhomogeneities. The results of this study may have significant petrological and volcanological implications. From a petrological point of view, mafic enclaves dispersed in felsic host rocks are regarded as portions of mafic magma which, trapped inside regular regions, survived the hybridisation process. Instead, host rocks are regarded as regions where efficient mixing dynamics generated hybrid magmas. The fact that a single magmatic mass may display large compositional differences at the same time undermines the assumption of most geochemical models, which assume the temporal and spatial homogeneity of the magma body. From the volcanological perspective, the presence of magmatic enclaves in volcanic rocks allows us to estimate the mixing times of magmas by analysing chemical diffusion patterns between host rocks and enclaves. Editorial responsibility: D. Dingwell     

Modeling effects of multinode wells on solute transport

Long-screen wells or long open boreholes with intraborehole flow potentially provide pathways for contaminants to move from one location to another in a ground water flow system. Such wells also can perturb a flow field so that the well will not provide water samples that are representative of ground water quality a short distance away from the well. A methodology is presented to accurately and efficiently simulate solute transport in ground water systems that include wells longer than the grid spacing used in a simulation model of the system and hence are connected to multiple nodes of the grid. The methods are implemented in a MODFLOW-compatible solute-transport model and use MODFLOW's Multi-Node Well Package but are generic and can be readily implemented in other solute-transport models. For nonpumping multinode wells (used to simulate open boreholes or observation wells, for example) and for low-rate pumping wells (in which the flow between the well and the ground water system is not unidirectional), a simple routing and local mixing model was developed to calculate nodal concentrations within the borehole. For high-rate pumping multinode wells (either withdrawal or injection, in which flow between the well and the ground water system is in the same direction at all well nodes), complete and instantaneous mixing in the wellbore of all inflows is assumed.     

A spreadsheet program for two-well tracer test data analysis

Two-well tracer tests are often conducted to investigate subsurface solute transport in the field. Analyzing breakthrough curves in extraction and monitoring wells using numerical methods is nontrivial due to highly nonuniform flow conditions. We extended approximate analytical solutions for the advection-dispersion equation for an injection-extraction well doublet in a homogeneous confined aquifer under steady-state flow conditions for equal injection and extraction rates with no transverse dispersion and negligible ambient flow, and implemented the solutions in Microsoft Excel using Visual Basic for Application (VBA). Functions were implemented to calculate concentrations in extraction and monitoring wells at any location due to a step or pulse injection. Type curves for a step injection were compared with those calculated by numerically integrating the solution for a pulse injection. The results from the two approaches are similar when the dispersivity is small. As the dispersivity increases, the latter was found to be more accurate but requires more computing time. The code was verified by comparing the results with published-type curves and applied to analyze data from the literature. The method can be used as a first approximation for two-well tracer test design and data analysis, and to check accuracy of numerical solutions. The code and example files are publicly available.     

Characterization of a multilayer aquifer using open well dilution tests

An approach to characterization of multilayer aquifer systems using open well borehole dilution is described. The approach involves measuring observation well flow velocities while a nearby extraction well is pumped by introducing a saline tracer into observation wells and collecting dilution vs. depth profiles. Inspection of tracer profile evolution allows discrete permeable layers within the aquifer to be identified. Dilution profiles for well sections between permeable layers are then converted into vertical borehole flow velocities and their evolution, using an analytic solution to the advection-dispersion equation applied to borehole flow. The dilution approach is potentially able to measure much smaller flow velocities that would be detectable using flowmeters. Vertical flow velocity data from the observation wells are then matched to those generated using a hydraulic model of the aquifer system, "shorted" by the observation wells, to yield the hydraulic properties of the constituent layers. Observation well flow monitoring of pumping tests represents a cost-effective alternative or preliminary approach to pump testing each layer of a multilayer aquifer system separately using straddle packers or screened wells and requires no prior knowledge of permeable layer depths and thicknesses. The modification described here, of using tracer dilution rather than flowmeter logging to obtain well flow velocities, allows the approach to be extended to greater well separations, thus characterizing a larger volume of the aquifer. An example of the application of this approach to a multilayer Chalk Aquifer in Yorkshire, Northeast England, is presented.     

Chaotic advection at the pore scale: Mechanisms,upscaling and implications for macroscopic transport

The macroscopic spreading and mixing of solute plumes in saturated porous media is ultimately controlled by processes operating at the pore scale. Whilst the conventional picture of pore-scale mechanical dispersion and molecular diffusion leading to persistent hydrodynamic dispersion is well accepted, this paradigm is inherently two-dimensional (2D) in nature and neglects important three-dimensional (3D) phenomena. We discuss how the kinematics of steady 3D flow at the pore scale generate chaotic advection—involving exponential stretching and folding of fluid elements—the mechanisms by which it arises and implications of microscopic chaos for macroscopic dispersion and mixing. Prohibited in steady 2D flow due to topological constraints, these phenomena are ubiquitous due to the topological complexity inherent to all 3D porous media. Consequently 3D porous media flows generate profoundly different fluid deformation and mixing processes to those of 2D flow. The interplay of chaotic advection and broad transit time distributions can be incorporated into a continuous-time random walk (CTRW) framework to predict macroscopic solute mixing and spreading. We show how these results may be generalised to real porous architectures via a CTRW model of fluid deformation, leading to stochastic models of macroscopic dispersion and mixing which both honour the pore-scale kinematics and are directly conditioned on the pore-scale architecture.     

Tracer mass recovery in fractured aquifers estimated from multiple well tests

Forced-gradient tracer tests in fractured aquifers often report low mass recoveries. In fractured aquifers, fractures intersected by one borehole may not be intersected by another. As a result (1) injected tracer can follow pathways away from the withdrawal well causing low mass recovery and (2) recovered water can follow pathways not connected to the injection well causing significant tracer dilution. These two effects occur along with other forms of apparent mass loss. If the strength of the connection between wells and the amount of dilution can be predicted ahead of time, tracer tests can be designed to optimize mass recovery and dilution. A technique is developed to use hydraulic tests in fractured aquifers to calculate the conductance (strength of connection) between well pairs and to predict mass recovery and amount of dilution during forced gradient tracer tests. Flow is considered to take place through conduits, which connect the wells to each other and to distant sources or sinks. Mass recovery is related to the proportion of flow leaving the injection well and arriving at the withdrawal well, and dilution is related to the proportion of the flow from the withdrawal well that is derived from the injection well. The technique can be used to choose well pairs for tracer tests, what injection and withdrawal rates to use, and which direction to establish the hydraulic gradient to maximize mass recovery and/or minimize dilution. The method is applied to several tracer tests in fractured aquifers in the Clare Valley, South Australia.     

Field Tracer Test for the Design of LNAPL Source Zone Surfactant Flushing

A field tracer test was carried out in a light nonaqueous phase liquid (LNAPL) source zone using a well pattern consisting of one injection well surrounded by four extraction wells (5‐spot well pattern). Multilevel sampling was carried out in two observation wells located inside the test cell characterized by heterogeneous lithology. Tracer breakthrough curves showed relatively uniform flow within soil layers. A numerical flow and solute transport model was calibrated on hydraulic heads and tracer breakthrough curves. The model was used to estimate an average accessible porosity of 0.115 for the swept zone and an average longitudinal dispersivity of 0.55 m. The model was further used to optimize the relative effects of viscous forces versus capillary forces under realistic imposed hydraulic gradients and to establish optimal surfactant solution properties. Maximum capillary number (N_Ca) values between injection and extraction wells were obtained for an injection flow rate of 16 L/min, a total extraction flow rate of 20 L/min, and a surfactant solution with a viscosity of 0.005 Pa?s. The unconfined nature of the aquifer limited further flow rate or viscosity increases that would have led to unrealistic hydraulic gradients. An N_Ca range of 3.8 × 10^?4 to 7.6 × 10^?3 was obtained depending on the magnitude of the simulated LNAPL‐water interfacial tension reduction. Finally, surfactant and chase water slug sizing was optimized with a radial form of the simplified Ogata‐Banks analytical solution (Ogata and Banks 1961) so that injected concentrations could be maintained in the entire 5‐spot cell.     

An Investigation of the Pressure Pulsing Reagent Delivery Approach

Pressure pulsing technology is an innovative method that has been developed with the aim of overcoming preferred flow paths associated with remediation techniques that rely on the injection of reagents. Numerical and field experiments were conducted to assess how pressure pulsing affects groundwater flow and solute transport during reagent injection. A series of field experiments were performed at two field sites where a monitoring network designed to capture the breakthrough of solutes delivered from an injection well was installed. Pressure pulsing and conventional injection methods were used at each site. One site was comprised of fine sand with low heterogeneity, while the other was moderately heterogeneous with discrete layers varying from fine sand to silt. The data suggest that breakthrough was more uniform for the pressure pulsing injections; however, this difference was minor and complicated by sorption of some of the tracers employed. The groundwater flow and solute transport modeling exercise simulated the rapid boundary pressure modulation that occurs in association with pressure pulsing. Two‐dimensional (2D) simulations revealed that repeated sudden onset of injection cessation produces brief periods of gradient reversal and the development of a mixing zone near the injection well. The spatial extents of this mixing zone were found to be highly dependent upon the hydraulic diffusivity of the medium, with medium heterogeneity and pulsing frequency playing secondary roles. Three‐dimensional (3D) numerical simulations were used to benchmark the observations from one of the field sites. The results from the modeling effort showed that solute breakthrough from a pressure pulsing injection is more dispersed relative to a conventional injection as a result of the mixing zone phenomenon; however, we were unable to directly observe this mixing zone using the instrumentation deployed at the two field sites.     

Grain Size Analysis and Permeametry for Estimating Hydraulic Conductivity in Engineered Porous Media

Grain size analysis and permeametry are common methods for estimating the hydraulic conductivity (K) of porous media. It is well known that these methods have limited accuracy when they are used to characterize natural sediments. However, hydrogeological research has increasingly introduced technologies dependent on engineered porous media that may be less problematic because complex geologic structures are eliminated in the lab and field-scale packings. The recently introduced Horizontal Reactive Media Treatment Wells (HRX® Wells), for in situ, passive remediation of groundwater is one such example. The HRX Well passively collects groundwater and directs it through a horizontal pipe packed with an engineered porous medium. In this project, grain size analysis was conducted for sand and sand-iron mixtures to estimate K using the 16 algorithms provided in the HydrogeoSieveXL2.3.2 software. The results were compared to K determined by permeametry and a field-scale column, 30 cm long and 25 cm in diameter, representing an HRX Well. The best comparability of K estimates from grain size analysis and permeametry were obtained using the USBR, Slichter, and Shepherd K estimation methods. These also showed good agreement between lab-scale and field-scale K estimations, with reproducibility within the range ±20%. This study shows that laboratory K estimations can be representative across various relevant scales, including the field-scale, for engineered porous media. This finding extends to filter packs, and other engineered porous media design methods by emphasizing and demonstrating one case of accuracy in lab-scale permeability estimation for field-scale implementations.     

Investigation of Small‐Scale Preferential Flow with a Forced‐Gradient Tracer Test

A new tracer experiment (referred to as MADE‐5) was conducted at the well‐known Macrodispersion Experiment (MADE) site to investigate the influence of small‐scale mass‐transfer and dispersion processes on well‐to‐well transport. The test was performed under dipole forced‐gradient flow conditions and concentrations were monitored in an extraction well and in two multilevel sampler (MLS) wells located at 6, 1.5, and 3.75 m from the source, respectively. The shape of the breakthrough curve (BTC) measured at the extraction well is strongly asymmetric showing a rapidly arriving peak and an extensive late‐time tail. The BTCs measured at seven different depths in the two MLSs are radically different from one another in terms of shape, arrival times, and magnitude of the concentration peaks. All of these characteristics indicate the presence of a complex network of preferential flow pathways controlling solute transport at the test site. Field‐experimental data were also used to evaluate two transport models: a stochastic advection‐dispersion model (ADM) based on conditional multivariate Gaussian realizations of the hydraulic conductivity field and a dual‐domain single‐rate (DDSR) mass‐transfer model based on a deterministic reconstruction of the aquifer heterogeneity. Unlike the stochastic ADM realizations, the DDSR accurately predicted the magnitude of the concentration peak and its arrival time (within a 1.5% error). For the multilevel BTCs between the injection and extraction wells, neither model reproduced the observed values, indicating that a high‐resolution characterization of the aquifer heterogeneity at the subdecimeter scale would be needed to fully capture 3D transport details.     

Mass Discharge in a Tracer Plume: Evaluation of the Theissen Polygon Method

A tracer plume was created within a thin aquifer by injection for 299 d of two adjacent “sub‐plumes” to represent one type of plume heterogeneity encountered in practice. The plume was monitored by snapshot sampling of transects of fully screened wells. The mass injection rate and total mass injected were known. Using all wells in each transect (0.77 m well spacing, 1.4 points/m² sampling density), the Theissen Polygon Method (TPM) yielded apparently accurate mass discharge (M_d) estimates at three transects for 12 snapshots. When applied to hypothetical sparser transects using subsets of the wells with average spacing and sampling density from 1.55 to 5.39 m and 0.70 to 0.20 points/m², respectively, the TPM accuracy depended on well spacing and location of the wells in the hypothesized transect with respect to the sub‐plumes. Potential error was relatively low when the well spacing was less than the widths of the sub‐plumes (>0.35 points/m²). Potential error increased for well spacing similar to or greater than the sub‐plume widths, or when less than 1% of the plume area was sampled. For low density sampling of laterally heterogeneous plumes, small changes in groundwater flow direction can lead to wide fluctuations in M_d estimates by the TPM. However, sampling conducted when flow is known or likely to be in a preferred direction can potentially allow more useful comparisons of M_d over multiyear time frames, such as required for performance evaluation of natural attenuation or engineered remediation systems.     

A modular injection system, multilevel sampler, and manifold for tracer tests

Ground water injection and sampling systems were developed for bacterial transport experiments in both homogenous and heterogeneous unconsolidated, surficial aquifers. Two types of injection systems, a large single tank and a dynamic mixing tank, were designed to deliver more than 800 L of amended ground water to the aquifer over 12 hours, without altering the ground water temperature, pH, Eh, or dissolved gas composition. Two types of multilevel samplers (MLSs) were designed and installed. Permanent MLSs performed well for the homogenous surficial aquifer, but their installation procedure promoted vertical mixing, which could obfuscate experimental data obtained from vertically stratified, heterogeneous aquifers. A novel, removable MLS was designed to fit in 2- and 4-inch wells. Expandable O-rings between each sampling port hydraulically isolated each port for sample collection when a nut was tightened at the land surface. A low-cost vacuum manifold system designed to work with both MLS designs used 50 mL centrifuge tubes to efficiently sample 12 MLS ports with one peristaltic pump head. The integrated system was developed and used during four field campaigns over a period of three years. During each campaign, more than 3000 ground water samples were collected in less than one week. This system should prove particularly useful for ground water tracer, injection, and push-pull experiments that require high-frequency and/or high-density sampling.     

Application of ERT,Saline Tracer and Numerical Studies to Delineate Preferential Paths in Fractured Granites

Accurate quantification of in situ heterogeneity and flow processes through fractured geologic media remains elusive for hydrogeologists due to the complexity in fracture characterization and its multiscale behavior. In this research, we demonstrated the efficacy of tracer-electrical resistivity tomography (ERT) experiments combined with numerical simulations to characterize heterogeneity and delineate preferential flow paths in a fractured granite aquifer. A series of natural gradient saline tracer experiments were conducted from a depth window of 18 to 22 m in an injection well (IW) located inside the Indian Institute of Technology Hyderabad campus. Tracer migration was monitored in a time-lapse mode using two cross-sectional surface ERT profiles placed in the direction of flow gradient. ERT data quality was improved by considering stacking, reciprocal measurements, resolution indicators, and geophysical logs. Dynamic changes in subsurface electrical properties inferred via resistivity anomalies were used to highlight preferential flow paths of the study area. Temporal changes in electrical resistivity and tracer concentration were monitored along the vertical in an observation well located at 48 m to the east of the IW. ERT-derived tracer breakthrough curves were in agreement with geochemical sample measurements. Fracture geometry and hydraulic properties derived from ERT and pumping tests were further used to evaluate two mathematical conceptualizations that are relevant to fractured aquifers. Results of numerical analysis conclude that dual continuum model that combines matrix and fracture systems through a flow exchange term has outperformed equivalent continuum model in reproducing tracer concentrations at the monitoring wells (evident by a decrease in RMSE from 199 to 65 mg/L). A sensitivity analysis on model simulations conclude that spatial variability in hydraulic conductivity, local-scale dispersion, and flow exchange at fracture-matrix interface have a profound effect on model simulations.