DNAPL Source Depletion: 2. Attainable Goals and Cost–Benefit Analyses

It is difficult to quantify the range in source strength reduction (MdR) that may be attainable from in situ remediation of a dense nonaqueous‐phase liquid (DNAPL) site given that available studies typically report only the median MdR without providing insights into site complexity, which is often a governing factor. An empirical study of the performance of in situ remediation at a wide range of DNAPL‐contaminated sites determined MdRs for in situ bioremediation (EISB), in situ chemical oxidation (ISCO), and thermal treatment remedies. Median MdR, geometric mean MdR, and lower/upper 95 percent confidence interval for the mean were: 49x, 105x, 20x/556x, respectively, for EISB; 9x, 21x, and 4x/110x for ISCO; and 19x, 31x, and 6x/150x for thermal treatment. Lower MdR values were determined for large, complex sites and for sites with DNAPL pool‐dominated source zones. A feasibility analysis of partial DNAPL depletion is described for a pool‐dominated source zone. Back‐diffusion from low‐hydraulic conductivity units within a pool‐dominated source zone is shown to potentially sustain a secondary source for more than 1,000 years, indicating that aggressive source treatment may not reduce the remediation timeframe. Estimated plume response demonstrates there may be no reduction in cost associated with aggressive treatment, and little difference in risk reduction associated with the various alternatives. Monitored natural attenuation (MNA) for the source zone is shown to be a reasonable alternative for the pool‐dominated source zone considered in this example. It is demonstrated that pool‐dominated source zones with a large range in initial DNAPL mass (250 to 1,500 kg) may correspond to a narrow range in source strength (20 to 30 kg/year). This demonstrates that measured source strength is nonunique with respect to DNAPL mass in the subsurface and, thus, source strength should not be used as the sole basis for predicting how much DNAPL mass remains or must be removed to achieve a target goal. If aggressive source zone treatment is to be implemented due to regulatory requirements, strategic pump‐and‐treat is shown to be most cost effective. These remedial decisions are shown to be insensitive to a range of possible DNAPL pool conditions. At sites with an existing pump‐and‐treat system, a significant increase in mass removal and source strength reduction may be achieved for a low incremental cost by strategic placement of extraction wells and pumping rate selection. © 2014 Wiley Periodicals, Inc.     

Evaluating the longevity of a PFAS in situ colloidal activated carbon remedy

The remediation of per‐ and polyfluoroalkyl substances by injection of colloidal activated carbon (CAC) at a contaminated site in Central Canada was evaluated using various visualization and modeling methods. Radial diagrams were used to illustrate spatial and temporal trends in perfluoroalkyl acid (PFAA) concentrations, as well as various redox indicators. To assess the CAC adsorption capacity for perfluorooctane sulfonate (PFOS), laboratory Freundlich isotherms were derived for PFOS mixed with CAC in two solutions: (1) PFOS in a pH 7.5 synthetic water that was buffered by 1 millimolar NaHCO₃ (K_f = 142,800 mg^1‐a L^a/kg and a = 0.59); and (2) a groundwater sample (pH = 7.4) containing PFOS among other PFAS from a former fire‐training area in the United States (K_f = 4,900 mg^1‐a L^a/kg and a = 0.24). A mass balance approach was derived to facilitate the numerical modeling of mass redistribution after CAC injection, when mass transitions from a two‐phase system (aqueous and sorbed to organic matter) to a three‐phase system that also includes mass sorbed to CAC. An equilibrium mixing model of mass accumulation over time was developed using a finite‐difference solution and was verified by intermodel comparison for prediction of CAC longevity in the center of a source area. A three‐dimensional reactive transport model (ISR‐MT3DMS) was used to indicate that the CAC remedy implemented at the site is likely to be effective for PFOS remediation for decades. Model results are used to recommend remedial design and monitoring alternatives that account for the uncertainty in long‐term performance predictions.     

Application of an Adapted Version of MT3DMS for Modeling Back‐Diffusion Remediation Timeframes

Simulation of back‐diffusion remediation timeframe for thin silt/clay layers, or when contaminant degradation is occurring, typically requires the use of a numerical model. Given the centimeter‐scale vertical grid spacing required to represent diffusion‐dominated transport, simulation of back‐diffusion in a 3‐D model may be computationally prohibitive. Use of a local 1‐D model domain approach for simulating back‐diffusion is demonstrated to have advantages but is limited to only some applications. Incorporation of a local domain approach for simulating back‐diffusion in a new model, In Situ Remediation‐MT3DMS (ISR‐MT3DMS) is validated based on a benchmark with MT3DMS and comparisons with a highly discretized finite difference numerical model. The approach used to estimate the vertical hydrodynamic dispersion coefficient is shown to have a significant influence on the simulated flux into and out of silt/clay layers in early time periods. Previously documented back‐diffusion at a Florida site is modeled for the purpose of evaluating the sensitivity of the back‐diffusion controlled remediation timeframe to various site characteristics. A base case simulation with a clay lens having a thickness of 0.2 m and a length of 100 m indicates that even after 99.96 percent aqueous TCE removal from the clay lens, the down‐gradient concentrations still exceed the MCL in groundwater monitoring wells. This shows that partial mass reduction from a NAPL source zone via in situ treatment may have little benefit for the long‐term management of contaminated sites, given that back‐diffusion will sustain a groundwater plume for a long period of time. Back‐diffusion model input parameters that have the greatest influence on remediation timeframe and thus may warrant more attention during field investigations, include the thickness of silt/clay lenses, retardation coefficient representing sorbed mass in silt/clay, and the groundwater velocity in adjacent higher permeability zones. Therefore, pump‐and‐treat systems implemented for the purpose of providing containment may have an additional benefit of reducing back‐diffusion remediation timeframe due to enhanced transverse advective fluxes at the sand/clay interface. Remediation timeframes are also moderately sensitive to the length of the silt/clay layers and transverse vertical dispersivity, but are less sensitive to degradation rates within silt/clay, contaminant solubility, contact time, tortuosity coefficient, and monitoring well‐screen length for the scenarios examined. ©2015 Wiley Periodicals, Inc.