Calcium carbonate, which is an important mineral in both anthropogenic and biological sys- tems, is a very critical component in nature. Numerous studies have shown that precipitation of calcium carbonate in fractured and porous media influences solute transport. Previous studies have used two-dimensional computational models to mimic the three-dimensional fluid flow and reactive transport of dissolved mineral. With the goal of improving under- standing of mineral precipitation and transport process and to test the accuracy of results from the two-dimensional models, a three-dimensional, two-step reactive transport computa- tional model is developed with OpenFOAM, an open source, C++ based CFD toolbox. This thesis presents the processes of customizing OpenFOAM to solve reactive flow and transport in a three-dimensional variable aperture fracture. Several parameters, including resolution in aperture height direction, aperture fields, and relative roughness of fracture surfaces, are modified to assess sources of errors and to test the soundness of the model. Results are then directly compared with those from the two-dimensional model. Also, a comparison view of different discretization schemes for representing advection in the transport model is presented. All together, our research results suggest that even though false diffusion is observed, the upwind scheme is chosen because of its boundedness and stability.

It is known that earthquake and seismic activities changes behavior of a hydrological system. These changes are attributed to changes in permeability of the medium due to pore pressure oscillations induced by seismic waves. Multiple studies suggest that fine particle mobiliza- tion inside fractures due to these oscillation, may be the possible mechanism that enhances permeability; however, direct evidence is lacking. Here I introduce a set of experiments to further analyze e ects of pore pressure oscillations on fine particle mobilization inside a transparent analog fracture. I performed a controlled perturbation, using the designed oscillator, with peak flow rate of 1.9 ml/min and frequency of 1 Hz for 5 minutes. E ective aperture or hydraulic aperture is widely used in aperture studies to represent permeability of a fracture. In this study I observed that e ective aperture enhanced by pore pressure oscillations from 58.77 μm to 73.23 μm. This enhancement was correlated to the particles that were mobilized during the oscillations. The particle mobilization was also visible in the animation created from the images that were taken during the experiment.

Suspensions of non-Brownian particles are ubiquitous in natural processes such as debris flows, magma flows, and sediment transport in river systems. Non-Brownian suspensions are used in engineering applications such as the injection of slurries for environmental remediation, mud injection during well drilling, cosmetics, and hydraulic fracturing. Suspensions are especially important in hydraulic fracturing, a technique developed in the 1940's by the oil industry to increase production rates from wells. The goal of hydraulic fracturing in oil and gas applications, is to enhance the permeability of subsurface rock formations by injecting a pressurized fluid to create a network of conductive pathways, i.e., fractures. The fluid used in hydraulic fracturing is composed of solids, chemical additives, and water. The solids (e.g., proppants) are used to \textit{prop} open the fracture and allow hydrocarbons to flow through the fracture. The hydrocarbons flow through the pore spaces between proppant particles, which means the fracture conductivity is limited by the amount and distribution of proppants delivered to the fracture.

The work presented in this dissertation explores the behavior of two different proppant mixtures (suspensions). The first suspension is a conventional proppant mixture (sand only) with a high solid volume fraction, $\phi_s=0.5$, which may increase fracture permeability by increasing the amount of sand delivered to the fracture. The second suspension is made up of sand ($\phi_s = 0.177$) and polymeric fibers ($\phi_f=0.0038$), which may create a proppant distribution that may enhance fracture permeability. I used transparent, laboratory-scale (15.2 cm $\times$ 15.2 cm), fractures to understand the behavior of these suspensions as they flow through and settle inside hydraulic fractures.

Concentrated suspensions flowing through idealized parallel-plate fractures exhibited complex flow behavior. Concentrated suspensions flowing through parallel fractures developed a non-uniform velocity distribution across the fracture width (in the plane of the fracture). As the concentrated suspension flowed through a confined rectangular channel (e.g., fracture) the suspension developed regions of high velocity near the no-flow boundaries where velocity was expected to be the lowest. These high-velocity regions were $\sim$2cm wide and were observed regardless of upstream boundary conditions. Furthermore, the observed non-uniform velocity distribution persisted irrespective of flow rate and fracture geometry. Through further experiments it was observed that these high-velocity regions were due to non-uniform $\phi$ distribution in the plane of the fracture. Additionally, the pressure gradient across the fracture, $\nabla P$, exhibited behavior that defied simple relationships between applied pressure gradient and flow rate. It was observed that the timescale required for $\nabla P$ to reach a steady state was significantly longer than expected. The cause of the transient $\nabla P$ was explored using a two-dimensional numerical model of concentrated suspensions flowing through a fracture of the same dimension as our experiments.

I explored the flow and settling behavior of multi-component (sand-fiber) proppant mixtures inside a fracture subjected to an applied stress. The experiments presented here show that adding fibers to conventional proppants leads to heterogeneous proppant distribution inside the fracture. This heterogeneous proppant distribution led to the formation of sand-fiber \textit{clusters} capable of supporting the applied stress and maintained the fracture open. Additionally, it was observed that injecting solids-free fluid (flowback) led to the mobilization of some solids within the proppant pack. Solid mobilization and flowback was explored through the use of a non-Newtonian flow solver that simulates flow through a mixed fracture-porous matrix medium. The experimental and numerical results suggest that adding polymer fibers to conventional proppants leads to highly heterogeneous proppant distribution which may lead to higher fracture permeability.

Fractures are often leakage pathways for fluids through low-permeability rocks that otherwise act as geologic barriers to flow. Flow of fluids that are in chemical disequilibrium with the host rock can lead to mineral precipitation, which reduces fracture permeability. When fracture surfaces contain a single mineral phase, mineral precipitation leads to fast permeability reduction and fracture sealing. However, the feedback between precipitation and permeability may be disrupted by mineral heterogeneities that localize precipitation reactions and provide paths of low-reactivity for fluids to persist over relatively long time-scales. In this dissertation, I explore the role of mineral heterogeneity on precipitation-induced permeability reduction in fractures. To do this, I use a combined experimental and numerical approach to test three hypotheses: (1) Mineral heterogeneity prolongs fracture sealing by focusing flow into paths with limited reactive surface area, (2) Precipitation-induced transport alterations at the fracture-scale are controlled by three-dimensional growth dynamics at the grain-scale, and (3) The effects of mineral heterogeneity become more pronounced as mineralogy and surface roughness become autocorrelated over similar length-scales.

Direct measurements of mineral precipitation using transmitted light methods in a transparent analog fracture show that mineral heterogeneity can lead to the progressive focusing of flow into paths with limited reactive surface area, which is in support of (1). In this experiment, flow focusing led to a 72\% reduction in the max precipitation rate; measurements of the projected mineralogy show that this was due to focusing of large dissolved ion concentrations into regions that contained 82\% less reactive surface area than the fracture-scale average. Results from a newly developed reactive transport model that simulates precipitation-induced fracture surface alterations as a three-dimensional process are in good qualitative agreement with these experimental observations. Comparison of these results with a reactive transport model that represents precipitation as a 1D alteration of the fracture surfaces show that this flow-focusing process is driven by lateral growth of reactive minerals across the fracture-plane, which supports (2). Lastly, results from simulations in fractures that contain varied degrees of heterogeneity show that precipitation leads to a competition between two feedbacks: (i) precipitation-induced reactive surface area enhancement, which increases precipitation rates, and (ii) precipitation-induced permeability reduction, which decreases precipitation rates. When surface roughness and mineral heterogeneity provide persistent paths of limited surface area, the reactive transport becomes very sensitive to local permeability reduction. Simulation results show that this prolongs the fracture-sealing process and can lead to a reduction in fracture-scale precipitation rate, which supports (3). Furthermore, the results presented in this dissertation demonstrate that predictions of fracture sealing by mineral precipitation can be easily misinformed by studies that ignore small-scale mineral heterogeneity and neglect the three-dimensional nature of precipitation-induced fracture surface alterations.

Pflotran was used to model the water table response to tides and wave effects in the first couple hundred meters inland from the shoreline. Unlike previous analytical solutions and many numerical models, Pflotran can take into account both the fully and partially saturated zones, irregular boundary shapes, and time-varying boundary conditions such as wave setup and overtopping in addition to the regular tidal harmonics. This research demonstrates the effect of several soil and beach characteristics and the importance of considering all forcings coming together to create boundary conditions. Specifically, the influence of waves and wave overtopping represented by ponding on the beach surface prove to be significant. Field data from Cardiff State Beach quantifies the significant improvement gained by including all of these effects.

Harmful algae blooms associated with toxic cyanobacteria have been increasing in both frequency and severity as a result of global climate change. These blooms are responsible for the release of biotoxins, the most common and toxic of which are the microcystins (MCs), that are recalcitrant to the operations of conventional drinking water treatment plants (DWTPs). Bio-based technologies targeting MC removal, such as biofiltration systems, have been proposed as cost-effective and sustainable alternatives to advanced treatment technologies (i.e., ozonation). Biofilters rely on native bacterial communities endemic to the source water to metabolize microcystin as a carbon and energy source. However, biofilter treatment variability is an ongoing challenge, arising from variations in environmental conditions including temperature, pH, and the presence of other bioavailable nutrients. To effectively address this treatment variability, biofilters must be evolved into “engineered” systems, in which MC degrader bioactivity is promoted and treatment tightly controlled.

This dissertation has amalgamated a series of system modelling, experimental data collection efforts, unstructured kinetic modelling, as well as sensitivity and optimal experimental design (OED) analyses to arrive at an improved predictive understanding of MC biodegradation. The results of these efforts first indicated that the removal of biotoxins within biofilters is highly dependent on the pretreatment system and operations employed by the DWTP. The kinetics of MC biodegradation by degrading communities were bi-phasic, where the taxonomy of the communities and the kinetics of MC degradation were altered in the presence of an alternative carbon source. The kinetics of MC metabolism and bacterial growth of isolated degrading populations were well predicted by the Moser kinetic model, where up to 5 out of 6 parameters could be identified. A novel approach to global sensitivity analysis was developed to improve the accuracy and convergence efficiency of the sensitivity indices ranking the most influential parameters of the Moser model. Finally, the OED procedure designed a fed batch reactor experiment that drastically improved the practical identifiability of the parameters of the Moser model. The culmination of these analyses has laid the foundation for a comprehensive and practical kinetic model describing degrader growth and MC removal in bio-based treatment systems.

This paper describes a novel experiment in which two very different methods of underwater robot localization are compared. The first method is based on a geometric approach in which a mobile node moves within a field of static nodes, and all nodes are capable of estimating the range to their neighbours acoustically. The second method uses visual odometry, from stereo cameras, by integrating scaled optical flow. The fundamental algorithmic principles of each localization technique is described. We also present experimental results comparing acoustic localization with GPS for surface operation, and a comparison of acoustic and visual methods for underwater operation.