Permanganate flushing of DNAPL source zones, experimental and numerical investigation

Abstract

In site chemical oxidation (ISCO) is a potentially effective technology for remediating aquifers contaminated with dense, non-aqueous phase liquids (DNAPLs) such as trichloroethene (TCE) and perchloroethene (PCE). These contaminants dissolve slowly in groundwater; however, the dissolution rate is enhanced by the addition of a concentrated oxidant, such as potassium permanganata, which rapidly degrades these contaminants to cardbon dioxide and chloride. The research presented in this thesis focuses on the application and effectiveness of ISCO at the field scale. As part of this research, an ISCO field experiment was completed at an experimental site at Canadian Forces Base Borden. A small DNAPL source zone containing both residual PCE and TCE was flushed with a 8 g/L permanganate solution for 485 days. In addition, a numerical model was developed to simulate field-scale application of the remediation process.

Recycling of the residual permanganate in the extraction wells was used to minimize the mass of the oxidant injected into the treatment zone and the volume of effluent permanganate solution requiring treatment. In total, 892 kg of permanganate were injected into the treatment zone while 303 kg were recaptured by the recycling system. A permanganate mass balance was completed which suggested that much the injected oxidant solution was lost from the treatment zone. The oxidant loss was primarily attributed to downwards vertical flow caused by the density difference between the permanganate solution and the background porewater.

Groundwater monitoring during the oxidant flush focussed on characterizing the spatial distributions of permanganate and chloride in groundwater. While the injection system successfully flushed the source zone with the concentrated permanganate solution, the monitoring data suggested that some of the oxidant migrated below the capture zone of the extraction wellls. Measurements of the chloride concentration down-gradient of the source zone were used to indicate the removal of DNAPL mass. Interpretation of the chloride concentrations measured in the extraction wells was complicated by the recycling of chloride along with the residual oxidant and provided little information on the rate of DNAPL mass removal; however, after ~300 days of oxidant injection, the concentration of chloride, suggesting that the rate of DNAPL removal had slowed considerably. Discrete-depth sampling immediately down-gradient of the source zone was used to identify the presence of a high-concentration chloride signature. After 485 days of oxidant injection, high chloride concentrations were only found in samples collected right on the down-gradient edge of the source zone, suggesting that the chloride plume produced by DNAPL removal was small.

The performance of the oxidant flush was assessed by comparing the extent of groundwater contamination before and after the oxidant flush. Specific performance measures included the reductions in DNAPL mass, peak solvent concentration, and solvent plume load. While the groundwater data indicated that some DNAPL was present in the source, DNAPL was not detected in soil samples collected from the source, emphasizing the difficulty of evaluating DNAPL mass in the field. Plume loading was identified as the most useful performance measurement. The oxidant flush reduced the loads of TCE and PCE in the groundwater plume by 99% and 90% respectively.

A numerical model was developed to simulate the kinetic physico-chemical processes through which enhanced mass transfer occurs. The model was applied to several synthetic scenarios to demonstrate the processes which influence the effectiveness of ISCO. A mass transfer enhancement in a scenario where dissolution was limited by diffusion across a stagnant path length was demonstrated; two-way diffusion of the oxidant towards the DNAPL zone increased the mass transfer rate in inverse proportion to the path length. The effect of several parameters on the effectiveness of ISCO was evaluated by applying the model to a synthetic field scale problem. Dispersivity and oxidation rate had no affect on the performance of ISCO; mass transfer rate, injected oxidant concentration, and hydraulic gradient had varying affects. Remediation appeared to be limited by the oxidant flux into the source zone and the mass transfer rate.