Organic contaminants in soil and ground water pose a serious problem to both human health and the environment. Spilled compounds penetrate the soil media by advectiive and dispersive processes and sorb to soil particles. Significant concentrations of the compound transform the soil into a "contaminated mass." Dealing with this type of soil contamination can be difficult if not impossible and/or expensive.
Solvent flushing is a technique to remove subsurface pollutants. The Augustijn et al model provides an accurate representation of contaminant behavior. Solvent flushing is simple. Wells are dug on opposite sides of a contaminated site. Alcohol is added to a water solvent to increase its contaminant solubility. The solvent is pumped into the injection well upgradient. As the solvent travels through the soil it carries the contaminant with it to the recovery well downstream where it is pumped out to the surface. The solvent and contaminant solution is treated and disposed of.
The solvent flushing technique has many benefits:
The following steps are used in implementing solvent flushing:
(click the diagram to view a larger image)
The solvent flushing technique utilizes the injection of a solvent mixture, commonly a water / alcohol mix. This mixture:
The solvent flushing system works best if the retardation factor in water is small and the cosolvent fraction is high, so the contaminant is recovered faster. Some of the constraints that affect the success and efficiency of the solvent flushing system are nonequilibrium conditions, soil heterogeneity, and type of cosolvent that ultimately influences the time required to recover the contaminant.
The objective of the Augustijn et al model is to understand and quantify the relationship between retardation, (amount of contaminant left in the soil), sorption (the amount of contaminant take out), flow rate and concentration of organic compounds in soil when being flushed by water and a cosolvent.
The support for using the solvent flushing method is established by studies that have been done earlier on the subject of "water-miscible organic cosolvents for enhanced in situ remediation":
Elution curves for the simulation model assume:
To obtain their data, a small tube was packed with Eustis sand. Naphthalene, an organic contaminant that adsorbs to soil, was pumped into the tube and was allowed to reach equilibrium. The concentration of the naphthalene was recorded. Once equilibrium was reached, a solution of water and methanol (solvent) was also pumped into then tube. The flow rate and concentration of this solvent was recorded. The effluent discharged at the other end was put in test tubes and tested for naphthalene concentration using liquid chromatography. The amount of naphthalene in the effluent equals the amount of naphthalene taken out of the sand. This procedure was repeated several times for different flow rates and concentrations of the water/methanol solvent and various mixtures of anthracene and naphthalene contaminants.
From the data collected in the experimental procedure, and knowing the properties of Eustis sand, the volume of solvent required to remove arbitrary amounts of naphthalene were derived. The volumes were plotted against concentration of methanol in the solvent, and concentrations of the organic contaminants in the effluent. The results obtained from the data concluded that the rate and concentration of the pollutant removed from the soil increased with both the flow and concentration of methanol in the solvent.
The Augustijn et al model for "solvent flushing" is a one dimensional, theoretical model whose accuracy depends on the "uncertainty in input parameters." For field purposes a more comprehensive model that entails three dimensional flow, including soil heterogeneity and unsaturated flow conditions needs to be developed. Other factors that need to be considered are the different types of contaminants, hydrologic considerations (coarse or fine textured soil), climatic conditions, etc. Also on a field scale one needs to be careful about the leaching of the cosolvent outside the hydraulically controlled zone that may cause offsite transport of the contaminants. Conceptually the model needs to be more comprehensive and "realistic" and after that the details about field application need to be worked out. For now the Augustijn et al model provides a crude estimation of the interaction between the cosolvent and the contaminant using the "solvent flushing" system.
Prepared By Murad Pandit & Peter Benchimol,
Undergraduate Students at Virginia Polytechnic Institute & State University, Department of Civil Engineering, November, 1996
Send comments or suggestions to:
Faculty Advisor: Daniel Gallagher, dang@vt.edu
Copyright © 1998 Daniel Gallagher
Last Modified: June 7, 1998
