Degree Type

Dissertation

Date of Award

2019

Degree Name

Doctor of Philosophy

Department

Geological and Atmospheric Sciences

Major

Geology; Environmental Science

First Advisor

William W. Simpkins

Abstract

Previous work characterizing till in Iowa has identified the extensive fracture networks that can reach depths of 30 m. Preferential flow in those fractures can result in groundwater velocities that are 1-4 orders of magnitude higher than those in the matrix and facilitate rapid advective transport. The effects of till fractures have been quantified by column tracer tests in the laboratory, but their effects on flow and transport at larger scales (e.g., field, watershed) are unknown, primarily due to the lack of data on fracture properties and a computationally-feasible method for including fractures in large-scale models. This research provides a methodology to address this problem and allow fractures to be incorporated into models at larger scales. Field fracture measurements, quantification of the Representative Elementary Volume (REV), development of a hydraulic conductivity (K) tensor estimation program (FracKFinder) and numerical modeling with the dual-continuum (D-C) model in HydroGeoSphere (HGS) were used to characterize the hydraulic properties of the matrix and fractures of the late Wisconsin Dows Formation till in central Iowa.

Determining bulk direction K values of ten different volumes of fractured till, ranging from 0.125 m3 to 7 m3, was the first step in this analysis. HGS simulations suggested that the REV of the Dows Formation ranges from 4 to 5 m3 at depths of one m where fractures are densest (P32 ≥ 27.5 m2/m3) and most transmissive, to 2 to 3 m3 at depths of 3.3 m where fracture transmissivity and fracture density are lower (P32 ≤ 24.4 m2/m3). A MATLAB program, FracKfinder, was used to compute a six-component K tensor from the REV. The resulting tensors showed that the till matrix is approximately anisotropic, while the fracture network had slightly greater (40 to 60 percent) K values in the z-direction, due to the predominantly vertical orientation of the fractures. Using the tensors, a D-C simulation of a previous large-column tracer experiment showed superior agreement to other well-documented modeling approaches (i.e., equivalent porous media; EPM; and discrete fracture network-matrix; DFM-M) with simulation times of under eight seconds. Sensitivity analysis using PEST showed that parameters K and porosity associated with the matrix are the most important to constrain with field data. Fracture spacing was shown to be important for computing input parameters for the second (fracture) continuum in the D-C model. When input parameters are taken from site-specific field and laboratory data, the D-C model is able to predict the experimental results with a high degree of accuracy (modified index of agreement > 0.95). Additionally, the ability of the D-C model to calibrate to empirically- or theoretically-derived parameters without good initial parameter estimates suggests that the D-C model is capturing the physics of groundwater flow and solute transport in fractured till.

Modeling results were supported by laboratory data from bromide tracer experiments conducted on samples (16-cm-wide by 16-cm-tall and less than the REV) from a Dakota Access pipeline trench. Samples containing well-connected fracture networks produced breakthrough curves with rapid first arrival times and shapes characteristic of solute transport in a fractured medium. Samples containing fewer or no connected fractures produced slower breakthrough with curves similar to unfractured till. Furthermore, a ± 540% range of breakthrough times was observed in cores with nearly identical K values, demonstrating that the effects of fractures on transport behavior is unpredictable even when K data suggest that well-connected fractures are present. These results suggest that column tracer tests in fractured till will not produce groundwater flow and transport parameters representative of in-situ field conditions unless the test volume is greater than the REV.

Copyright Owner

Nathan Lee Young

Language

en

File Format

application/pdf

File Size

154 pages

Available for download on Sunday, April 18, 2021

Share

COinS