H11I-01 INVITED
Hydraulic Tomography in Fractured Granite: The Mizunami Underground Research Laboratory Site, Japan
Two large-scale cross-hole pumping tests were conducted at separate locations in deep boreholes at the Mizunami Underground Research Laboratory (MIU) construction site in central Japan. We analyze the two cross- hole tests using the Transient Hydraulic Tomography (THT) code of Zhu and Yeh [2005] to compute the hydraulic conductivity (K) and specific storage (Ss) distributions, as well as their uncertainties in three-dimensions. The equivalent K and Ss obtained using asymptotic analysis served as the initial parameter estimates for the 3D stochastic inverse modeling effort. Results show several, distinct high K and low Ss zones that are continuous over hundreds of meters, which appear to delineate fault zones and its connectivities. The fault zones imaged through THT correlate well with available geological data and drawdown records. The THT analysis also identified a low K zone which corresponds with a known fault zone trending NNW and has been found to compartmentalize groundwater flow at the site. The results are evaluated through available geological information, drawdown records, arrival times of drawdown in response to pumping, velocities of drawdown pulses, and coseismic groundwater level responses during several large earthquakes.
H11I-02
Outcrop-Based Lidar Imagery to Develop Millimeter-Scale Models of Heterogeneity
Dispersion in ground water systems is largely believed to be due to variable velocity fields caused by aquifer heterogeneity; however, models that capture realistic heterogeneities are often elusive. In order to evaluate the influence of outcrop-scale heterogeneities (e.g., 2-5 meters), we interpret lithofacies from detailed outcrop scans using high-resolution (3-5 mm-scale) terrestrial lidar. To delineate lithofacies from the outcrop images, we apply both automated and semi-automated segmentation methods in a hierarchical approach. We first define bounding surfaces that surround lithofacies units. We then segment lithologic classes (e.g., gravel, sand, or silt) within these units. By applying reasonable hydraulic conductivity values to each lithology and populating a grid at the resolution of the lidar scan with these conductivities, we can build 2D groundwater flow and particle tracking models that capture this ‘realistic' heterogeneity as observed in outcrop. We apply this approach to several outcrops in order to evaluate flow characteristics through different facies assemblages. The resulting groundwater simulations show that cross-bedding within a unit focuses flow and transport along select pathways, typically coarse-grained units. In some cases, flow may be focused inside less than 10% of the total aquifer volume. Additionally, truncated gravel cross-beds can focus flow through adjacent, finer-grained beds. Transport through these heterogeneous realizations is markedly non-Fickian.
H11I-03
Inference of 3-D Hydraulic Conductivity from Flowmeter and Pumping-Test Data
We jointly apply field data of flowmeter and multiple pumping tests in fully screened wells to estimate hydraulic conductivity using a geostatistical inversion method. We use the steady state drawdowns of pumping tests and the discharge profiles of flowmeter tests as our data in the inference. The discharge profiles are not converted to "measurements" of hydraulic conductivities. The flowmeter profiles are indicative of the relative vertical distribution of hydraulic conductivity in the direct vicinity of the boreholes, while drawdown measurements of pumping tests provide information about horizontal fluctuation of the depth-averaged hydraulic conductivity. For inversion, we use the quasi-linear geostatistical approach of Kitanidis (1995), accelerated by spectral methods for the evaluation of cross-covariance matrices (Nowak et al., 2003) and stabilized by a modified Levenberg-Marquardt method (Nowak and Cirpka, 2004). We apply the method to data obtained at the Krauthausen test site of the research center Jülich, Germany. In the field, multiple pumping tests were conducted in a hydraulic tomographic format, stressing and monitoring the aquifer at different locations. We determine the most likely estimate of hydraulic conductivity and the associated posterior uncertainty. The resulting estimate of our joint three-dimensional geostatistical inversion shows an improved three-dimensional structure in comparison to the inversion of pumping test data only. The corresponding uncertainty field shows a considerable decrease near the wells where both tests were conducted.
H11I-04
Characterization of spatial variability of hydraulic parameters in fractured rocks: Interpretation of pumping tests at the Altona Flat Rock Experimental Site
We present the results of the interpretation of multiple hydraulic tests conducted at the
Altona Flat Rock experimental site, located near Plattsburgh (NY). The purpose of these
tests was to establish the nature of flow connectivity in a single sub-horizontal bedrock
fracture. The geology of the area is dominated by the Potsdam sandstone which is
characterized by sub-horizontal bedding-plane fractures that extend over the scale of
kilometres. Seven open boreholes with a diameter of 15 cm have been drilled to a depth of
12.2 m in the formation at reciprocal distances ranging from 7 to 15.8 m. Packer
injection tests show the presence of a saturated horizontal fracture at 7.3 meters of
depth, which intersects all the wells. The single fracture is characterized by a highly
variable aperture which leads to a wide range of hydraulic transmissivity (T) and
storativity (S) estimated from slug tests. In order to characterize the hydraulic
properties of the fractured rock, a series of pumping tests were performed. The pumping
tests were executed at constant rate of 7\;e-5\; m
H11I-05
Effective Conductivity Of Highly Heterogeneous Anisotropic Formations: Does The Self- Consistent Solution Work?
Three-dimensional simulations of flow in highly heterogeneous anisotropic porous formations are conducted to estimate horizontal and vertical effective conductivities of aquifer formations. Multi-Indicator conductivity model, developed by Dagan et al (2003), was employed to test validity of the classical self-consistent solution (Dagan 1989) and the solution based on the exponential conjecture (Gelhar and Axness 1983). In a Multi-Indicator model for statistically anisotropic formations, conductivity statistics (mean, variance and 2-point correlation) of the formation are reproduced using spheroidal inclusions of different hydraulic conductivities and sizes. In the present study, hydraulic conductivity of constant-size inclusions follows a log-normal distribution and the anisotropic covariance function has spherical structure with different integral scales in the horizontal (Ih) and vertical (Iv) directions. Using large domain sizes (over 100 conductivity integral scales long) and high volume fraction of inclusions (0.5 and 0.7), we examine accuracy of the self-consistent solution over a range of logconductivity variances (up to 8) and anisotropy ratios (Ih/ Iv up to 20). We present two main findings: 1) effective conductivities (both horizontal and vertical) of the simulated Multi-Indicator Models are exactly equal to those predicted by self-consistent solution; 2) the exponential conjecture is not valid for examined Multi-Indicator model. Modeling technology developed for the present study can be used in other flow and transport studies, including large-scale transport through highly heterogeneous anisotropic formations.
H11I-06
Using Wavelet Analysis to Identify Dominant Scales of Hydraulic Conductivity and Head Fields
Wavelet analysis involves an integral transform of, for example, a permeability data set, using a wavelet as the kernel of the transform. A wavelet is a function that is non-zero only over a finite region; therefore the wavelet transform analyzes only a subset of the data set. The wavelet is shifted to analyze different subsets of the data set, and it is scaled to analyze different scales of the data set. We generate sets of bounded one dimensional, statistically homogeneous permeability fields with exponential distributions, and we run numerical flow simulations using these permeability fields. We use the global wavelet energy spectrum (GWES) to analyze the dominate scales in both the permeability fields and resulting head distributions, and we explore the relationships between dominant scales. We develop analytical solutions for the GWES to corroborate these relationships.
H11I-07
Steady-state two phase flow in porous media: experiments and simulations
We report on experimental and numerical studies of steady-state two-phase flow in a two-dimensional porous medium. Experimentally the wetting and the non-wetting phase are injected simultaneously from alternating inlet points into a porous Hele-Shaw cell initially filled with wetting fluid. Transient behavior is observed in time and space, such that a certain distance behind the front, fully developed fragmented steady-state flow develops. Numerical studies confirm this transition and support the conclusion that the steady-state phase is fragmented and dominated by bubble dynamics, allowing for a statistical description on the macroscopic-scale in terms of effective behavior and on the pore-scale in terms of cluster-size distributions. The latter is shown to obey a scaling-law with the Darcy velocity.