This dissertation presents the results of a series of experimental and numerical studies designed to advance knowledge of the fundamental mechanisms controlling colloidal particle transport in saturated porous media. That colloidal particles facilitate contaminant transport in porous media, or act as contaminant sources, is well known, and also widely recognized as important to environmental and health issues around the world. Many prior and ongoing studies are aimed at understanding particle transport and deposition behavior in saturated porous media, and these studies have generated a broad range of knowledge regarding particle fate and transport mechanisms. However, the prediction of particle transport behavior still remains challenging, not least because the particle transport processes themselves still include many unknown factors. The goal of the work reported in this dissertation, was to advance understanding of the influence of varying flow velocity conditions, flow direction, particle size and mixed particle populations on particle transport processes. In order to meet this goal, a new numerical model for particle transport was developed, and standard laboratory column test protocols were modified to enable the imposition of varying flow conditions during a test, as well as visualization of particle concentrations within the interior of a column. In addition, and in collaboration with other researchers, numerical modeling work was also undertaken to provide insight into the processes governing particle transport at an instrumented field site.
Numerical models have been used extensively to investigate a wide variety of engineering and applied science problems, including those involving colloidal particle transport in saturated porous media. For the research presented in this dissertation, a new numerical model, termed the Kinetic Colloid Transport Model (KCTM), was developed and implemented using the Matlab platform. The KCTM is based on a one-dimensional (1-D) advection-dispersion-sorption equation coupled with different kinetic sub-models for simulating particle interactions with the solid phase of a porous medium, including irreversible and reversible attachment mechanisms, as well as two-attachment site and two-particle population behaviors. The KCTM is capable of directly simulating particle transport behavior for a given set of initial and boundary conditions, and also inversely solving for the sub-model kinetic parameters based on particle concentrations observed during column or field experiments. To validate the KCTM, KCTM results were compared with analytical solutions generated by the STANMOD program and numerical solutions generated by HYDRUS-1D. Simulation of particle breakthrough concentrations during a hypothetical column experiment with fourteen different case studies, involving a range of particle dispersion coefficients as well as attachment and detachment rates, was used for the validation. Agreement between the KCTM results and those generated by STANMOD and HYDRUS-1D, as defined by corresponding R squared values (all above 0.999), was considered acceptable across all ten case studies. The KCTM has the advantage of modeling a range of particle transport mechanisms, many of which are not accounted for in current open-source or commercially available codes.
Fluctuating or varying velocity conditions are common under many real-world scenarios involving colloidal particle transport, yet are often neglected in laboratory column experiments designed to investigate particle transport behavior. To understand the influence of varying velocity conditions on particle transport, a series of traditional and modified laboratory column experiments was conducted. For the modified column experiments, a protocol was developed to enable the simulation of both increasing and decreasing velocity conditions during a test, as well as conditions involving an increase followed by a decrease in velocity. Laboratory column experiments were performed to examine the downward transport of 2 micron diameter microspheres through a saturated bed of 100 micron diameter glass beads under both constant and varying velocity conditions. The KCTM was simultaneously fit to observed particle concentration breakthrough curves, as well as measured particle concentrations retained in the column at the end of each constant velocity experiment, to obtain a relationship between a dimensionless irreversible kinetic attachment coefficient Ki* and transport velocity. This relationship was then used to model the results of the varying velocity tests, with limited success. A comparison of the Ki* values obtained from direct fitting of the varying velocity tests using the KCTM, with the Ki* values derived from the results of the constant velocity experiments, revealed a potential dependence of Ki* on the rate of change of transport velocity, which is currently not accounted for in any particle transport model. Overall, the results of this experimental and numerical investigation pointed to the need for better understanding of how varying velocity conditions impact fundamental particle transport mechanisms.
A visualization technique was used to examine the effects of particle size and flow direction on particle transport in a saturated porous medium comprised of 500μm diameter glass beads. Packed column experiments with uniform (100% 1μm or 100% 6μm) and mixed (90% 1μm with 10% 6μm and 90% 6μm with 10% 1μm) polystyrene latex microspheres were performed in one-dimensional upward, horizontal and downward flow fields at a constant velocity of 1.7m/day. Particle concentrations were recorded over time in the interior of a column and at the column exit. Experimental results showed that upward flow conditions generally gave rise to higher retained particle concentrations and lower particle breakthrough concentrations than horizontal and downward flow conditions, indicating that gravitational settling decreases particle transport distances and enhances particle deposition mechanisms. Consistent with prior studies, results also showed increasing particle retention with increasing particle size. The 1μm particle tests results were successfully modeled using a first order, irreversible particle attachment model, indicating little filtration of this particle size within the glass bead columns during transport. Modeling of the 6μm particle tests required a two-site kinetic modeling approach that accounted for particle interactions with the surfaces of the glass beads as well as straining of particles at bead-bead contact points. The presence of a second particle population had little impact on the transport of the 1μm particles. For the 6μm particles, the presence of the second particle population increased particle attachment rates, with the greatest impact observed during downward flow conditions. Overall, the results of this study confirm that particle size and flow direction impact particle transport processes. The study also reveals that particle size heterogeneity could also impact particle transport under certain conditions. Both of these findings have implications for field-scale modeling of particle transport.
The up-scaling of results obtained from laboratory column experiments to predict particle transport at the field scale is generally reported to under-estimate particle transport distances observed in the field. The over-simplification of column experimental conditions, in comparison to field conditions, or the use of improper kinetic models are two possible reasons leading to such inaccurate predictions. In order to explore the possible hurdles to current up-scaling methods, the KCTM was used to analyze a series of Escherichia coli based column experiments using aquifer sand obtained from a field site in Bangladesh, which are described in the collaborative work presented in Appendix A. Four E.coli breakthrough curves (BTCs) and two profiles of spatially retained E.coli concentrations at the end of an experiment were generated by the column test series. The KCTM successfully modeled the BTC results using a two-population kinetic sub-model. Both one-site and two-site particle attachment sub-models failed to reproduce the observed BTCs. None of the kinetic sub-models could reproduce the observed particle retention profiles, although the two-population sub-model generated similar hyper log-linear profiles to those seem in the experiment results. Low mass recovery rates in the column experiments is one possible reason why the KCTM failed to fit the retained profiles. The kinetic parameters obtained from the KCTM fits to the column experimental results were incorporated into a two-dimensional transport model, HYDRUS-2D, to predict E. coli transport observed at an instrumented field in Bangladesh. Predictions obtained using only irreversible attachments kinetics, reversible attachment kinetics and both reversible and irreversible attachment kinetics performed with RMSE values of 1158, 826, and 99, respectively. The dramatic decrease in RMSE with the application of the two-site kinetic model indicates that E. coli transport at the field site likely involves both reversible and irreversible attachment. An important conclusion of this work was the significance of designing laboratory column experiments that can enable the extraction of kinetic parameters relevant to field scale transport processes.
The numerical and experimental studies presented in this dissertation examined some factors that influence particle fate and transport in saturated porous media, which are commonly overlooked in many conceptual and numerical models of particle behavior. The results of these studies point to a need to better understand how varying velocity conditions, flow direction, particle size and mixed particle populations influence particle fate and transport. The results of these studies also prompt out several recommended future works. For the developed numerical model, current kinetic sub-models imperfectly reproduced experiment results, also inadequately described the particle transport in microscale observations, indicating the simplified first-order kinetics are inaccurate for describing actual particle transport behaviors. A non-log-linear kinetic sub-model and corresponding micro-scale experiments are needed for better predictions. Moreover, the effects of particle-particle interaction were proven significant in certain conditions, however, the processes is still unclear. Visualization technique introduced in this research is capable to explore the controlling mechanisms in micro-scale and further provides the foundations for developing non-log-linear kinetic model, quantifying the effects of particle-particle interactions, acceleration, and other uncovered physical/chemical factors on particle transport in porous media. Advancing understanding of these factors has potential for improving the prediction of colloidal particle transport under real-world, field conditions, which can benefit many programs aimed at reducing the environmental and health impacts of colloid facilitated contaminant transport.
| Identifer | oai:union.ndltd.org:columbia.edu/oai:academiccommons.columbia.edu:10.7916/D8P84C02 |
| Date | January 2016 |
| Creators | Liu, Po-Chieh |
| Source Sets | Columbia University |
| Language | English |
| Detected Language | English |
| Type | Theses |
Page generated in 0.0032 seconds