A numerical model to optimize LNAPL remediation by multi-phase extraction

Abstract

Light non-aqueous phase liquid (LNAPL) contaminated sites pose a risk to human health and the natural environment. Multi-phase extraction (MPE) is one of the most widely used technologies to remediate these sites. Thus, it is important to optimize MPE systems to improve their effectiveness and cost-efficiency. In this study, we developed a numerical model to optimize LNAPL mass removal by MPE, in which the aquifer domain was simplified as a cylinder with a single MPE extraction well located at the center. A dual-pump extraction system was applied to the model, which involved vacuum enhanced recovery to remove volatilized gaseous phase contaminants and a submerged pump to remove NAPL and contaminants in groundwater. After the model was validated with field data, the results showed that the contaminant extraction rate varied with the LNAPL thickness and submerged pump position. For benzene selected as the contaminant of concern, greater LNAPL extraction rates were achieved when the initial LNAPL thickness was large (>1.5 m) or in aquifers of high permeability (>2.45 × 10−10 m2). Importantly, it was discovered that in highly permeable coarse sand and gravel, the submerged pump ought to be placed within the LNAPL layer, whereas the pump should be placed below the water-NAPL interface in fine to medium sand aquifers. It was also found that an optimal liquid pumping rates exist, beyond which contaminant mass removal rates do not increase. Furthermore, it was found that in aquifers contaminated with thin LNAPL layers, mass transfer modelling that assumes equilibrium between the phases may greatly overestimate the accumulated mass of contaminants removed and, therefore, non-equilibrium modelling should be adopted. Finally, a cost analysis was carried out to compare the costs of remediating a contaminated site with MPE and by an alternative chemical oxidation approach. The MPE technology was found to be more cost effective when the initial thickness of LNAPL was relatively thin. In summary, the numerical model developed in this study is a useful tool for optimizing MPE system design.