Abstract
Dense non-aqueous phase liquids (DNAPLs), which persist as a separate phase in the subsurface and are heavier than water, have become a major environmental problem throughout the world because of their long term persistence at deep locations in the subsurface acting as a source for ground water contamination. This is because the small dissolved concentration in ground water from the source zone is often more than sufficient to exceed the maximum acceptable levels in consumption water. Therefore, there is a need to remediate the contaminated source zone in order to create a safer environment.
At present, available in situ remediation technologies can only remove a certain amount of the DNAPLs in the subsurface leading to incomplete removal. This limitation necessitates the need of performing a risk evaluation at the post-remediation stage with consideration of the remaining DNAPL source. In this study, physical modelling of source zone remediation by surfactant flushing was performed to investigate the mass transfer behaviour at different stages of DNAPL removal.
A series of column experiments of surfactant flushing were performed to investigate DNAPL removal by solubilisation and mobilisation. Results show the need of evaluating this remediation technology in two-dimensional flow conditions if the total trapping number exceeds 1x10-5 and the mobilisation effect becomes the dominant removal mechanism. The steady state mass flux values at different stages of surfactant flushing were also measured to examine the effect of partial NAPL removal on aqueous mass flux variations. Rapid decrease in mass flux was achieved only when the DNAPL was removed to the level that the DNAPL saturation is less than 5%. Some rate-limited effects on mass flux behaviour were also observed. The measured data were used to develop an empirical correlation to estimate mass transfer rate coefficient between NAPL-aqueous phases under one-dimensional flow conditions.
Two-dimensional tank experiments were conducted to examine the effect of NAPL entrapment, source zone geometry and NAPL saturation on mass removal by surfactant flushing in two dimensional flow conditions. Two scenarios were considered: they are (1) a DNAPL source zone placed above an impermeable soil layer and (2) a DNAPL source zone placed above a low permeability soil layer. Similar to the column experiment results, the tank experiment results revealed that the decrease in mass flux is not proportional to the degree of NAPL removal. The mass flux decreased rapidly only when the average DNAPL saturation became 8% for case (1) and 13 % for case (2).
In the two-dimensional flow tank experiments, the incoming groundwater can partially flow around the DNAPL source, whereas in the one-dimensional flow column experiments, the groundwater has to go through the DNAPL source zone. This flow bypass effect can reduce the overall mass flux. Comparison between the one-dimensional column and two-dimensional tank test results show that the flow bypass resulted in a 10-40% reduction of the mass flux.
Quantification of the flow bypass effect in two-dimensional flow conditions was investigated by numerical simulations of the tank experiments. The multiple analytical source superposition technique by Sale and McWhorther, (2001) was adopted for uniform flow analysis, whilst the finite element method was used for non-uniform flow analysis. Comparison between the mass flux values measured by the physical models and those computed from the numerical models gave the degree of error associated with the uniform flow model over the non-uniform flow model. In general, 20-40% overprediction in mass flux was obtained when the uniform flow model was used instead of the non-uniform flow model. The change in mass transfer rate coefficient due to DNAPL mobilisation that occurred during surfactant remediation was also addressed and a modified form of the mass transfer model is proposed.