Dissolution of Entrapped DNAPLs in Variable Aperture Fractures: Experimental Data and Empirical Model SARAH E. DICKSON* AND NEIL R. THOMSON Department of Civil Engineering, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada An appreciation of the dissolution from entrapped nonaqueous phase liquids (NAPLs) in fractures is essential as we attempt to understand and predict the fate of NAPLs present in fractured rock systems. Eight long-term dissolution experiments using 1,1,1-trichloroethane and trichloroethylene were conducted in two laboratory-scale dolomitic limestone variable aperture fractures under various conditions. Between 560 and 2600 fracture volumes of water were passed through the fractures resulting in the removal of 10-60% of the initial mass trapped. The effluent concentration profiles revealed three distinct and characteristic stages of dissolution: an initial pseudosteady stage, a transient stage, and a tailing stage. On average, 8% of the initial volume of NAPL present was removed during the initial pseudosteady stage. Data from the dissolution experiments were used in conjunction with statistical techniques to develop a continuous empirical model describing the initial pseudosteady and transient stages of dissolution. The model was used to successfully replicate effluent concentration data from two separate and independent dissolution experiments. The experimental results provide an indication of the expected dissolution behavior of entrapped NAPLs, while the developed model is a useful tool for characterizing mass transfer rates in variable aperture fractures. Introduction Dense nonaqueous phase liquid (DNAPL) releases to the subsurface often occur over fractured geologic deposits and can migrate through fracture networks under the influence ofgravitational,viscous,and capillaryforces.Capillaryforces act to resist the forward movement of flowing DNAPL and are responsible for separating and immobilizing DNAPL masses from the continuous DNAPL phase. Once released into the subsurface,the DNAPLwillcontinue to migrate until the entire volume is trapped as residual ganglia or until it reaches a capillary barrier that it cannot penetrate. The trapped DNAPLdissolvesinto the surroundingaqueousphase and eventually forms a plume within the fracture network and the porous matrix and represents a potential long-term source of contamination. The remediation offractured rockmassesisexceptionally difficult to achieve because the complex fracture network makes it nearly impossible to predict or accurately charac- terize the spatial location and distribution (residual vs continuous) of the DNAPL. Additionally, there has been no general agreement among researchers regarding mass trans- fer and contaminant transport model formulations that correctly account for all dominant processes occurring in fractured porous media systems (1). Remediation strategies currently used in fractured rock are similar to those used in porous media (2) and many of these technologies are fundamentallybased on the masstransferprocess(e.g.,pump and treat, soil vapor extraction, in-situ air sparging), which presently is poorly understood in fractured rock systems. In the absence of remedial actions, the mass transfer process influences the length oftime that a DNAPLentrapped in fracture planes will persist and the magnitude of the dissolved phase plume. It has been shown that DNAPL disappearance from a fracture is only significantly affected by matrix diffusion in relatively porous environments (3, 4). Esposito and Thomson (5) developed a numerical model to simulate nonequilibrium dissolution and aqueous phase transport in a variable aperture fracture and coupled it with an existing two-phase flow model (6). They showed that the time frame for NAPL dissolution is clearly sensitive to the mass transfer coefficient. They also found that increasing the hydraulic gradient across the fracture plane can aid in the removaloffree phase mass;however,diffusion-controlled mass transfer is necessary to dissolve NAPL that is trapped in regions of the fracture plane which are not accessible to aqueousphase flow.Glassand Nicholl(7)simulated a fracture plane using etched glass to study the dissolution of an entrapped air phase and were not able to explain their results using any existing conceptual dissolution models. Based on both experimental and simulation results, Detwiler et al. (8) used a simple exponential relationship with a constant bulk mass transfer coefficient to model the temporal change in DNAPL saturation in a variable aperture fracture. Although the various investigations discussed above were motivated by the ultimate need to understand DNAPL behavior in fractures at the field scale (i.e.,fracture networks), thiscannot be accomplished without first understandingthe behavior and developing tools to characterize dissolution at the scale of a single fracture. With this goal in mind, the investigation presented in this paper had two primary objectives: (1)to conduct a series oflong-term single variable aperture fracture physical model experiments to yield representative dissolution profiles subject to a range offluid flow and entrapment conditions and (2) to use these experimentaldata to develop a continuous empiricalmodel describing the bulk mass transfer rate from entrapped DNAPLs in single variable-aperture fracture planes. Materials and Methods Experimental Apparatus. The fracture planes employed in these experiments originated from a dolomitic limestone outcrop located near Kingston, Ontario. This formation was selected due to the presence of prominent stylolites and bedding planes, which provide convenient planes of weak- nessforinducingfractures.Rocksampleswere removed from the outcrop using a Quick Cut saw (TS 350, Stihl) equipped with a diamond-tipped blade and returned to the laboratory where tension fractures were induced using the method described by Reitsma and Kueper (9). This method of inducing tension fractures in rock samples is similar to the manner in which natural extension fractures are formed by weathering or stress relief. Two fractures were used in this *Corresponding author phone: (905)525-9140 ×24914; fax: (905)529-9688; e-mail: sdickso@mcmaster.ca. Current address: Department of Civil Engineering, McMaster University, 1280 Main Street West, Hamilton, ON L8S 4L7, Canada. Environ. Sci. Technol. 2003, 37, 4128-4137 4128 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 18, 2003 10.1021/es026275r CCC: $25.00  2003 American Chemical Society Published on Web 08/12/2003