Hydrology [H]

H31A  CC:2  Wednesday  0830h

Characterizing Ground-Water Flow and Chemical Transport in Fractured-rock Aquifers - I


Presiding: P G Cook, CSIRO, Land and Water Division; A M Shapiro, U.S. Geological Survey

H31A-01

Recent Developments in Modeling Coupled Fluid Flow and Transport Processes in Fractured Porous Media

* Therrien, R (rene.therrien@ggl.ulaval.ca), Laval University, Department of Geology and Geological Engineering, Quebec, QC G1K7P4, Canada

The development of numerical models to simulate fluid flow and solute transport remains an active area of research in fractured rock hydrogeology. Some problems that still need to be resolved relate to the large computational effort associated with some large-scale simulations with uncertainty in model parameters, the representation of complex 3D fracture networks and the need for additional capabilities such as coupled physical and chemical processes. This talk aims at giving an overview of some current model developments in the context of the control-volume finite element model, FRAC3DVS/HydroGeoSphere, for applications related to deep geological repositories for spent nuclear fuel. The model can simulate 3D variably-saturated flow, heat and reactive mass transport in fractured porous media using an equivalent porous medium, a dual continuum or a discrete fracture approach. Some examples of fluid flow and transport simulations related to deep geological repositories will be presented to highlight model improvements required because of the very long temporal scale for simulations, the need to quantify uncertainty, as well as continual improvements in field characterization tools that generate new type of data. These model improvements will also be valuable for other applications, such as simulating the near-surface flow and transport dynamics in fractured aquifers used for water supply.


H31A-02

Three-dimensional Discrete Fracture Network Simulations of Flow and Particle Transport Based on Laxemar Site Data (Sweden).

* Frampton, A (frampton@kth.se), Royal Institute of Technology (KTH), Dept of Land and Water Resources Engineering, Brinellv 32, Stockholm, 10044, Sweden
Cvetkovic, V (vdc@kth.se), Royal Institute of Technology (KTH), Dept of Land and Water Resources Engineering, Brinellv 32, Stockholm, 10044, Sweden

We study particle transport in a 3D DFN scenario based on Laxemar site characterisation data in Sweden, which is a candidate repository site for high level radioactive waste in the Swedish nuclear waste management program. The site characterisation data has revealed several interesting geometric and hydraulic fracture properties, such as power-law distributed fracture sizes and transmissivities. A fundamental aspect towards understanding tracer migration in subsurface sparsely fractured rock formations is the relationship between the Eulerian flow field at a sub-fracture scale with the Lagrangian flow field at a characteristic (model domain) scale. In this work we present results from a new technique for upscaling particle transitions obtained from Eulerian flow statistics to predictions of tracer discharge at a characteristic transport scale, based on previously developed methods used for 2D DFN's. This includes a mapping algorithm for transforming Eulerian into Lagrangian flow statistics without a priori knowledge of network connectivity, and by retaining the correlation between the water residence time τ and the hydrodynamic control of retention β we present accurate tracer discharge predictions. These results are illustrated using the unlimited diffusion model, and for some hypothetical tracers with properties designed to capture the behaviour of many common radionuclides. Finally we emphasise the importance of capturing the early arrival and peak of tracer breakthrough curves, i.e. to capture the bulk of the tracer mass arrival, in order to make accurate and conservative predictions.


H31A-03

Primary and Secondary Controls on Fracture and Fracture Permeability Development

* Mortimer, L (luke.mortimer@flinders.edu.au), Centre for Groundwater Studies, Flinders University, GPO Box 2100, Adelaide, SA 5001, Australia
Love, A J (andy.love@flinders.edu.au), Centre for Groundwater Studies, Flinders University, GPO Box 2100, Adelaide, SA 5001, Australia
Aydin, A (aaydin@olemiss.edu), Dept Geology and Geological Engineering, University of Mississippi, PO Box 1848, University, MS 38677, United States
Simmons, C T (craig.simmons@flinders.edu.au), Centre for Groundwater Studies, Flinders University, GPO Box 2100, Adelaide, SA 5001, Australia

Present-day hydrogeological properties and patterns of fractured rock aquifers are the direct result of their geotectonic history and setting. In the case of layered sedimentary sequences, individual sedimentary units respond to applied stress regimes via different deformation mechanisms as determined by many factors such as their mechanical properties, strain rate, temperature, confining pressure, fluid pressure, unit thickness and position within a sequence. Overall, this process results in the development of characteristic fracture network patterns and conductivities specific to individual rock types and their geologic structures. Mapping of the fractured rock aquifers of the Clare Valley revealed that the simple, layered, sedimentary sequence responded to Cambrian-age, fold-thrust deformation by developing dissimilar fracture patterns across its various units. In particular, relatively strong competent rock units preferentially deformed through macroscopic folding and intense brittle fracturing whilst less competent rock units tended to deform via mesoscopic folding, flexural slip deformation and/or the development of a foliation. These different deformation mechanisms ultimately led to the development of distinct hydraulic characteristics across the various rock units that are strongly stratabound in nature and poorly inter-connected. This strong stratabound character and poor inter-connectivity between different rock units has significant implications for the prediction of groundwater flow paths ie refraction through different units. Within the Clare Valley, detailed in-situ fracture hydraulic conductivity mapping of selected observation bores suggest that an anisotropic fracture permeability direction exists, which favours vertical to steeply dipping bedding planes and joints. This anisotropic fracture permeability direction is inferred to be a secondary effect, which is superimposed upon the primary fracture sets by the present-day, in-situ stress field. These results are supported by basic geomechanical (UDEC) models, which demonstrate the effect on fracture network conductivity and connectivity for different fracture networks under certain applied stress regimes. This secondary enhancement of fracture permeability within vertical to steeply dipping fractures is inferred to be the direct result of active, horizontal, ground relaxation due to present-day uplift, unloading and erosion.


H31A-04

Modelling of groundwater flow in a narrow regional fault zone at a mine site in Victoria

Bringemeier, D (detlef_bringemeier@coffey.com), Coffey Geotechnics Pty Ltd, 47 Doggett Street, Newstead QLD 4006 Australia, Brisbane, 4006, Australia
* Wang, X (xuyan_wang @coffey.com), Coffey Geotechnics Pty Ltd, 47 Doggett Street, Newstead QLD 4006 Australia, Brisbane, 4006, Australia

It is generally understood that groundwater flow in hard rock is mainly within a discrete network of conduits developed along rock discontinuities such as fractures, faults, shears, and lithological boundaries. Groundwater management may become a challenge if mineralisations associated with highly permeable faults zones are mined. This paper describes a groundwater model of a mine site in Victoria with open cast and underground mining following mineralisations along a narrow regional fault zone. A three-dimensional transient finite-element FEFLOW model was developed for simulating groundwater flow and pressure regimes within and adjacent to the fault zone. A combination of discrete fracture elements and Darcy elements were employed to model the fault zone with fractures developed at various scales. The model was calibrated with 17 years of groundwater level records from monitoring bores spread over the modelling domain. Pump test results were used to calibrate the model locally by varying hydraulic parameters of the two types of fault zone elements. A sensitivity analysis was carried out to evaluate the effect of fault zone parameter variations on the modelling results. The calibrated model was successfully used for simulating the effects of in-pit tailings storage and underground mining on regional groundwater regimes.


H31A-05

Using Induced Radon Concentrations in Boreholes to Infer Sources of Water and Flow Paths During an Aquifer Discharge Test in a Fractured Rock Aquifer

* Harrington, G A (Glenn.Harrington@csiro.au), CSIRO Land & Water, Private Bag 2, Glen Osmond, SA 5064, Australia
Brown, K G, Aquaterra, Suite 4, 125 Melville Pde., Como, WA 6152, Australia
Love, A J, Flinders University, Sturt Rd., Bedford Park, SA 5042, Australia

Aquifer pumping tests conducted in fractured rock aquifers (FRA) are notoriously difficult to interpret because an infinite number of fracture orientation/connectivity/permeability combinations can satisfy the measured water level data. The integration of natural or applied tracers into conventional hydraulic techniques therefore can provide improved capability for characterisation of a FRA site at the scale of several 10s to 100s of metres. We have used an adaptation of the Radon-222 method originally developed for estimating ambient groundwater flow rates through nested piezometers in FRAs (Cook et al. 1999) to characterize flow conditions during an aquifer discharge test. The field site is located near Balhannah in South Australia, where a thick sequence of Proterozoic meta-sediments (Balhannah Shale) dominates the near-surface hydrogeology. The site is approximately 40m x 30m in area and has a total of 12 open wells installed to depths of 48-56 m below ground level (bgl). Two of these wells were diamond-cored to provide for detailed fracture mapping using an acoustic televiewer (ATV). The remaining 10 wells were drilled using rotary down-the-hole hammer methods, which resulted in rougher borehole walls and thus less accurate ATV fracture mapping. The central well at the site was used as a production well for a 7.35hr constant-rate aquifer discharge test while the 11 surrounding wells were used as observation points. Radon concentration was measured at multiple discrete depths in the production and observation wells both prior to commencing the test and once steady state drawdown had been achieved (immediately prior to ceasing pumping). In general radon concentrations in the production well increased by 1 to 5 times their ambient values over the duration of the test, and in the observation wells they increased by up to 3 times the ambient values. Comparing the ambient and "pumped" radon profiles resulted in rapid identification of the fractures providing most water supply into the production well, and where water was most actively moving through observation wells towards the production well. An electromagnetic borehole flow meter (BHFM) was installed in the production well to provide profiles of flow rate before and during the test. This in turn provided a semi-quantitative check on the relative contributions of water inferred from the radon profiles. Down-hole electrical conductivity (EC) profiling before and after pumping provided further evidence of the most active flow zones when natural salinity contrasts permitted. Hydraulic data collected throughout the test revealed a total drawdown of more than 13m in the production well and 3.3m or less in each of the surrounding observation wells, with minor anisotropy favoring a NE-SW orientation. Despite the widespread drawdown, and more than 25 000 L of water being removed from the aquifer over the duration of the test, the radon, BHFM and EC data showed that flow into the production well was dominated by three SE-dipping (41-51°) fractures, located at about 18m, 28m and 55m bgl. These fractures are well connected at shallower depths to at least two adjacent observation wells.


H31A-06

Groundwater Ages in Fractured Rock Aquifers

* Cook, P G (peter.g.cook@csiro.au), CSIRO Land and Water, Private Bag 2, Glen Osmond, SA 5064, Australia

There have been many studies in porous media aquifers in which groundwater ages estimated using environmental tracers have been used to determine groundwater flow velocities. Because hydrodynamic dispersion is small in many aquifers, groundwater ages determined using these tracers closely approximate hydraulic ages (subsurface residence times). In fractured rock aquifers, however, matrix diffusion processes cause mixing between water in fractures and the matrix, causing tracer ages within the fractures to be much greater than water ages. Apparent ages obtained using different tracers will usually be different. However, while water ages cannot be determined from individual tracer measurements, the use of multiple tracers sampled along flowlines can constrain groundwater flow rates. Examples will be presented from the Clare Valley, South Australia, where modelling of vertical profiles of 14C, 3H, 36Cl and CFC-12 obtained from nested piezometers enables estimation of aquifer recharge rates.


H31A-07 INVITED

Using Dipole Tracer Tests to Predict Solute Transport in Fractured Rock

* Weatherill, D (douglas.weatherill@flinders.edu.au), School of Chemistry, Physics and Earth Sciences Flinders University, GPO Box 2100, Adelaide, SA 5001, Australia
Simmons, C T (craig.simmons@flinders.edu.au), School of Chemistry, Physics and Earth Sciences Flinders University, GPO Box 2100, Adelaide, SA 5001, Australia
Cook, P G (peter.g.cook@csiro.au), CSIRO Land and Water, Private Bag 2, Glen Osmond, SA 5064, Australia

The highly heterogeneous nature of fractured rock makes characterisation of solute transport properties a complex task. Dipole tracer tests have been used previously to determine in-situ transport parameters for single isolated fractures (Novakowski et al., 2004) and fracture networks (Sanford et al., 2002). A dipole tracer test involves injecting tracer into one well and recovering it from another whilst maintaining a steady state hydraulic field. The dipole flow field allows sampling of a relatively large region of the aquifer in a short time period. This study examines the possibility of using data from dipole tracer tests to make predictions of solute transport under natural hydraulic gradients in fractured rock. When interpreting measured tracer breakthrough data, the choice of analytical model is critical to determining the transport parameters of the system. Typically a dipole model of a single fracture is used to analyse breakthrough curves. This idealised flow geometry is unlikely to occur in fractured systems and therefore the parameters obtained from such an analysis may have limited usefulness. Analytical modelling using a single fracture model found that predictions of solute transport characteristics (times to first inflection point and the peak, peak concentration) may be obtained with greater certainty in some cases than the individual fracture and matrix parameters (Weatherill et al., 2006). This result suggests that perhaps a greater emphasis should be placed on predictions rather than the individual parameter values used to make them. To assess the ability of analytical models to make accurate predictions of solute transport in complex fracture systems, a series of tracer tests were numerically simulated in 3-D fracture networks. Tracer breakthrough was fitted analytically under a high forced gradient and then predicted at several lower gradients. Comparison of predictions from two analytical models with the numerical results found that breakthrough characteristics were within two orders of magnitude of the actual (numerical) values. This study demonstrates the potential for dipole tracer tests to be used to predict natural gradient solute transport in fractured rock. Tracer breakthrough data must be interpreted carefully, considering that the assumptions utilised in most analytical models are a large simplification of reality, with potential to generate significant errors. A range of analytical interpretations may be beneficial in bounding predicted transport behaviour. Novakowski, KS, Bickerton, G, Lapcevic, P (2004), Interpretation of injection-withdrawal tracer experiments conducted between two wells in a large single fracture, J. Contam. Hydrol. 73, 227-247. Sanford, WE, Cook, PG, Dighton, JC (2002), Analysis of a vertical dipole tracer test in highly fractured rock, Ground Water 40, 535-542. Weatherill, D, Cook, PG, Simmons, CT, Robinson, NI (2006), Applied tracer tests in fractured rock: Can we predict natural gradient solute transport more accurately than fracture and matrix parameters?, J. Contam. Hydrol 88, 289-305.


H31A-08

Non-Unique Interpretations of Chemical Transport in Fractured-Rock Aquifers

* Shapiro, A M (ashapiro@usgs.gov), U.S. Geological Survey, 431 National Center, Reston, VA 20192, United States

Fractured-rock aquifers are widely regarded as one of the most complex geologic settings in which to characterize ground-water flow and chemical transport. Regardless of the rock type, fractures are not uniformly distributed and their hydraulic properties can vary over orders of magnitude, which can result in highly convoluted flow paths over distances of meters to kilometers. Because of the geologic and hydraulic complexity of fractured- rock aquifers, quantitative characterization of chemical transport usually entails conducting tracer experiments under hydraulically stressed conditions, where a known quantity of a tracer solution is introduced into the ground water and the recovery of the tracer is monitored at points of ground water discharge, or by extracting ground water at one or more locations. The resulting tracer breakthrough curves along with a conceptual model of ground-water flow are used to infer the physical and chemical processes that control the character of the breakthrough curves. Flawed conceptual models of flow, including ignoring the effects of the density of the tracer solution and failing to characterize the range of aquifer hydraulic properties, can result in flawed interpretations of chemical transport processes and properties. Tracer experiments conducted in a fractured crystalline rock using tracer solutions with a range of densities resulted in disparate breakthrough curves that yielded different estimates of the fracture porosity. The density of the tracer solution affected the flow path between the injection and recovery locations, resulting in different tracer residence times and estimates of the fracture porosity than varying by a factor of three. Other tracer experiments conducted in bedding plane fractures of sedimentary rocks exhibited extended tails of breakthrough curves that could be interpreted as being dominated by diffusion into and out of the rock matrix adjacent to the bedding plane fractures; however, subsequent sampling of the ground water at boreholes between the injection and recovery locations revealed a significant amount of tracer mass resident in the fractures, implying that highly heterogeneous hydraulic properties of the fractures were responsible for the diffusion-like behavior of the breakthrough curves. The potential non-uniqueness of interpretations of tracer experiments conducted under hydraulically stressed conditions in fractured-rock aquifers needs to be carefully considered. Results of tracer experiments are often extrapolated to ambient flow conditions over temporal and spatial scales larger than those in the experiment, and the results are often used to resolve issues such as waste isolation, contaminant transport, and the design of ground-water remediation strategies.