|dc.description.abstract||In situ chemical oxidation (ISCO) with permanganate is a remedial technology that has been prevalent over the last decade. Permanganate is injected into the subsurface to oxidized reduced organic contaminants with the intent of mineralizing the organics to innocuous compounds such as water, oxygen, and carbon dioxide. However, the demand for permanganate from the naturally occurring reduced components associated with aquifer materials inhibits the ability of permanganate to effectively oxidize the target contaminants. This demand for permanganate is referred to as the Natural Oxidant Demand (NOD) and results from the presence of naturally occurring reduced aquifer species such as inorganic species containing iron, manganese, or sulfur, and natural organic matter. Traditionally, NOD has been considered to be an instantaneous sink for permanganate that required satisfaction before permanganate could propagate through the subsurface. However, recent research has suggested that NOD is kinetically controlled and not instantaneous resulting in the effectiveness of ISCO systems to be underestimated using traditional approaches. The objectives of this research were to develop a comprehensive NOD kinetic model from existing laboratory data of several aquifer materials, and then to use this model to estimate the impact of NOD kinetics on treatment efficiency.
The NOD kinetic model primarily was developed using results of bench-scale experiments performed on four aquifer materials, measuring the reduction of permanganate and oxidizable materials. Data analysis indicated that there are two bulk reactions occurring: a fast reaction and a slow reaction. For both of these reactions a second-order rate law was deemed to be appropriate; first-order with respect to each reactant. The slow reaction was subject to passivation and the reaction rate coefficient decreased hyperbolically as manganese oxide reaction by-products precipitated on grains. The developed NOD kinetic model was incorporated into a 1-dimensional transport model and was used to successfully simulate the results of NOD column studies.
Experimental efforts were completed to validate the 1-dimensional reactive transport model with data for organic contamination. A column study was completed to characterize the oxidation of an isolated trichloroethylene residual source zone. The chloride breakthrough data were used to represent the rate of TCE oxidation and a bromide tracer test was used as a conservative tracer to determine the dispersivity and porosity of the column. Both the simulated bromide and chloride breakthrough curves fit the experimental data well using published and calculated transport and chemical parameters.
The impact of NOD kinetics on treatment efficiency was evaluated through numerical simulations of four common organic contaminants using two injection schemes: vertical well flushing and inject-and-leave. The treatment efficiency was defined as the fraction of supplied permanganate used to oxidize the organic compound. Two aquifer materials were simulated representing a wide range of NOD characteristics. The results indicated that despite a great difference in the ultimate NOD (order of 15) the treatment efficiency only varied by 0-7% between the materials. In general, the treatment efficiency of the contaminant increased as the solubility and the reaction rate coefficient increased.
For treatment of organic compounds with a low solubility and reaction rate coefficient, the fast and slow NOD reaction kinetics should both be characterized since both exert a strong demand for permanganate in both the vertical flushing and inject-and-leave schemes. For organic compounds having moderate solubility and reaction rate coefficient the NOD species that require kinetic characterization depends on the injection scheme used: for a vertical well flushing scheme only the fast NOD requires characterization, whereas for the inject-and-leave scheme both the fast and slow NOD require characterization. For treatment of organic compounds with high solubility and reaction rate coefficient only the fast NOD requires characterization since the organic and fast NOD are depleted at the same time and the slow NOD does not play a significant role in permanganate consumption while free phase organic and fast NOD remain.
Traditional modelling approaches were compared, using the vertical well flushing scheme, to compare the treatment efficiency with the NOD kinetic model to past methods. The model was used to simulate ISCO treatment when NOD kinetics were not included and when the ultimate NOD was assumed. The simulations with no NOD term overestimated the treatment efficiency whereas the simulations with the ultimate NOD model underestimated efficiency. These findings further stressed the importance of the NOD kinetics on treatment efficiency.
The kinetics of the NOD kinetics must be characterized to determine if ISCO is a viable, cost-effective treatment option when considering ISCO as a redial strategy. Mischaracterization of these reactions could result in either over or underestimation of the treatment efficiency and poor design of pilot and full-scale treatment systems.||en