The Acceleration of the Diffusion-Limited Pump-and-Treat Aquifer Remediation with Pulsed Pumping that Generates Deep Sweeps and Vortex Ejections in Dead-End Pores

Clean water is a critical natural resource. We do not have much available: only 2.5% of water on Earth is freshwater and of that only 31% is in liquid form. 96% of the liquid fresh water is groundwater. Unfortunately that resource is subject to contamination by hazardous materials accidentally or illicitly spilled, leaked, or deposited in or on the ground. Among the methods to remediate these disasters, pump-and-treat (P&T) is the most common. The vertical circulation well (VCW) is a P&T configuration with extraction and injection sites within the same well. It can be adapted to many remediation techniques and has been gaining popularity since the 1990s and is often a better alternative to conventional P&T. Conventional P&T and VCWs are typically run with steady flow.

The major bottleneck to steady flow remediation is that contaminants become trapped in dead-end pores. In an aquifer there are two types of pores: pass-through pores and dead-end pores. The flow in former completely sweeps through the pore space while the flow does not enter the later; however, the flow through the pass-through pore induces a vortex in the dead-end pore. Under steady flow the only mechanism for contaminants to escape the dead-end pores is molecular diffusion.

A similar problem is encountered in the removal of surfactants in the manufacture of semiconductor and the removal of oil residue build-up in small ducts. Manufacturers discovered that pulsed flow would accelerate the mass transfer between the cavities and grooves on these surfaces and the external flow. This was because the unsteady ramp-up in flow rate initiated a deep sweep of the cavities. The unsteady ramp-down in flow rate initiated a vortex ejection where the sequestered vortex is no longer constrained and protrudes from the cavity.

We hypothesized that just as pulsed flow improves cleaning of grooved surfaces in several manufacturing procedures, rapidly pulsed pumping (with a period on the order of a second rather than weeks or months) in pump-and-treat groundwater remediation would boost the diffusion-limited removal of contaminants trapped in dead-end pores by generating transient deep sweeps and vortex ejections in these pores. These processes have not yet been exploited in groundwater remediation to any significant degree.

We tested our hypothesis in a series of numerical and laboratory experiments. We considered unwashed and washed media. For unwashed media (Chapter 1) we used as a square pore in the numerical domain and crushed glass (for its negligible sorption capacity) in laboratory column studies. For washed media (Chapter 2) we used a smooth dead-end pore constructed with two tangential quarter circles as the pore in the numerical domain and glass spheres in the laboratory column studies. In all our laboratory experiments we used a fluorescent dye, Fluorescein, as a conservative tracer. We used the same parameters in our numerical experiments. However, in some we also considered immiscible contaminants such as NAPLs (Chapter 4).

All numerical experiments were conducted with the computational fluid dynamics software, FIDAP. In numerical experiments we studied the contaminant removal from interacting dead-end pores connected to both a straight pass-through pore and a divergent pass-through pore. The latter with the flow somewhat analogous to the radial spreading encountered around a around a well in field applications (Chapter 5).

To elucidate the dead-end pore dynamics (Chapter 3), we performed numerical experiments and used a physical model to obtain a relationship between the rapidly pulsed flow frequency and length of the pore. Our dimensional analysis pointed to the change in pressure as the key component in the initiation of transient deep sweeps and vortex ejections, two new pore-cleaning mechanisms.

We conclude that the rapidly pulsed flow improves the recovery of contaminants from unwashed, or rough, porous media. In numerical experiments with a pore system consisting of just a single square dead-end pore and a single pass-through pore, at 100 pore volumes pumped the rapidly pulsed flow improved cleanup of the dead-end pore alone by approximately 40%. This translates into a 10% improvement of the cleanup of the pore system (dead-end and pass-through pore). Since the dead-end pore is the bottleneck of the current groundwater remediation, it the first measure that is relevant.

In corresponding laboratory column experiments with crushed glass, the dead-end pore volume alone is not known. The cleanup of the whole pore space was improved by roughly 10% with the rapidly pulsed pumping, which corresponds nicely to our numerical results.

Our numerical experiments demonstrate that there exists an optimal pulsed pumping frequency that is a function of the local flow velocity and the pore geometry (size and morphology).

The contaminant recovery from washed, or rounded, media was not as pronounced in the laboratory experiments and the numerical experiments showed no improvement. While both rapidly pulsed and steady flow recovered all of the contaminant in the laboratory column tests, the difference in the time between the two pumping schemes was approximately 0.9 pore volumes pumped. This improvement is likely to be amplified with sorbing contaminants.

Many contaminants are non-aqueous phase liquids (NAPLs), which do not readily dissolve in water. We showed in numerical experiments that rapidly pulsed flow can recover NAPLs with viscosity lower than water, but is not as effective with higher viscosity materials; however, these results were based on a model that did not account for interfacial tension and wetting; therefore we will require additional numerical and laboratory experiments.

In practice, a flow through porous media is significantly more complex than the one-directional dominated flows considered in our numerical and laboratory column experiments. Around a well the flow is typically three-dimensional and largely radially dominated. We constructed two numerical domains to study the interactions between the cleanup of three square pores: one in a straight channel and one in a divergent channel to study the radial spread that would be experienced around a well. For a series of three dead-end pores, there was a 35% improvement by rapidly pulsed flow over steady flow in the straight channel and a 33% improvement in the divergent domain. The optimal frequency was different in the divergent flow even though the pores were the same size as in the previous study. Since the divergent channel reduced the flow velocity, the pulses reached the pores at a decreasing rate. Due to this divergence and the range of pore-sizes in a natural aquifer, implementation of rapidly pulsed flow should likely include a range of frequencies.

We concluded that the rapidly pulsed flow on the time scale of one-second would greatly enhance the cleanup of contaminated aquifers by P&T or VCW approaches. We measured significant improvements in the time to recovery. For our preliminary VCW experiment showed that rapidly pulsed pumping recovers 50% of the contaminant four times faster than steady pumping. P&T and VCW remediation typically use a steady flow; there are some methods that change the flow rate in P&T and other configurations, such as the VCW. These periodic changes in rate are on the scale of months to years. Some VCWs and air sparging technologies pulse oxygen, surfactants, and/or nutrients into the aquifer to oxidize, mobilize, or bioremediate the contaminants. As reviewed in chapter 6 in detail, all pulsing so far applied in remediation is on the time scale of a day or longer. Such low pulsing frequency does not produce sufficiently many deep sweeps to make a significant difference in cleaning dead-end pores.

Implementation of rapidly pulsed technology will utilize the same extraction and injection wells currently used in pump-and-treat remediation but will require replacement or significant modification of the pumps.

There are public health and financial implications of this research. In the dissertation conclusions section we reinterpret our numerical experiments with the multiple interacting dead-end pores and a divergent pass-through pore and laboratory experiments with a vertical circulation well chamber by calculating and plotting the ratio of times needed to reach a specified fraction recovered (specified cleanup level) in the steady and rapidly pulsed pumping modes, \tau_{s} / \tau_{p}. This ratio represents the speedup factor, i.e., the factor by which the time needed to reach the specified cleanup level with the conventional remediation (with steady pumping) would be reduced. From our experiments it appears that with the increasing level of targeted cleanup (contaminant fraction recovered), the speedup factor increases and may even exceed an order of magnitude. As we demonstrate in the dissertation conclusions section, this could translate into tens of billions of dollars in savings. Whether or not the laboratory speedup factors would hold in the field cannot be established without field-scale experiments.