Electron donor (ED) addition can be very effective in stimulating enhanced reductive bioremediation (ERB) of a wide variety of groundwater contaminants. However, ERB can result in secondary water quality impacts (SWQIs) including decreased levels of dissolved oxygen (O₂), nitrate (NO_3-), and sulfate (SO₄^2-), and elevated levels of dissolved manganese (Mn²⁺), dissolved iron (Fe²⁺), methane (CH₄), sulfide (S^2-), organic carbon, taste and odor concerns, and naturally occurring hazardous compounds (e.g., arsenic [As]). Fortunately, this ‘plume’ of impacted groundwater is usually confined within the original contaminant plume and is unlikely to adversely impact potable water supplies.

The overall objective of this research was to develop an improved understanding of the near- and long-term impacts to groundwater quality after implementation of in situ anaerobic bioremediation processes. Specific objectives achieved during this project include:

• Formulate a general modeling approach appropriate for simulating the production and subsequent natural attenuation of SWQIs for a wide range of sites. The model approach and validation was based on data from the Bemidji Crude Oil Spill and Cape Cod Wastewater Plume.
• Assemble a database of SWQIs at ERB sites to determine the range of SWQI concentrations and evaluate trends over time and distance downgradient.
• Use the validated modeling approach and results from the database analysis to develop a general conceptual model of SWQI formation, mobilization, and attenuation. This conceptual model can be used to identify sites where SWQI could be an issue and develop monitoring and management approaches to minimize adverse impacts.

A reactive transport model was developed to simulate the production and natural attenuation of important electron acceptors and SWQI parameters including organic carbon, O₂, NO_3-, Mn²⁺, Fe²⁺, SO₄^2-, S^2-, CH₄, and CO₂. The behavior of these and other reactive species in groundwater have been intensively studied for many years by the USGS Toxics Substances Hydrology Program (Toxics Program). Model simulation results were compared to an extensive database on monitoring results from two Toxics Program sites: (a) Bemidji Crude Oil Spill, MN; and (b) Cape Cod MMR Wastewater Plume, MA.

SWQI data were collected from 47 active and former ERB sites across the United States, including different contaminants of concern (COCs), injection strategies, organic substrates, and geographic locations. Monitoring data was compiled on SWQI parameters in 917 different monitoring wells from regulatory monitoring reports, DOD (i.e., AFCEE, ESTCP, SERDP) project reports, journal articles, and personal correspondence with site project managers. Summary statistics (mean, median, standard deviation, etc.) and cumulative frequency distribution plots were generated for upgradient, treatment zone, and various downgradient locations to define the range of SWQIs impacts and evaluate trends over time and distance downgradient.

The reactive transport model and SWQI database were then used to produce indicator simulations that illustrate the major processes control the formation and natural attenuation of SWQIs at ERB sites. The model was used to simulate the fate of the added organic carbon and how the reduced products affected the aquifer over a period of 40 years following subustrate addition.

Results from the reactive transport model development, SWQI database, and indicator simulations were integrated to develop a general conceptual model of the major processes controlling SWQI production and attenuation. This conceptual model can be used as a guide in understanding the magnitude, areal extent, and duration of SWQIs in ERB treatment zones and the natural attenuation of SWQI parameters as the dissolved solutes migrate downgradient with ambient groundwater flow.

Reactive Transport Model Development

Reactive transport models were created for USGS long-term study sites where reducing conditions were created in shallow aquifers by crude oil (Bemidji) or wastewater disposal (Cape Cod). The goal of the modeling was to understand the creation, natural attenuation, and longevity of the SWQIs created by the reducing conditions. The model simulations described flow and reactions along two-dimensional vertical cross sections of the study aquifers. Steady-state models of the flow systems were created using MODFLOW (Harbaugh, 2005) and reactive transport simulations were performed with PHT3D (Prommer et al., 2003). For the Bemidji Crude Oil Spill, the goal of the modeling was to understand the generation and fate of the most important SWQIs at the site, especially CH₄, aqueous Fe(II), and dissolved organic carbon (DOC). The simulation covered the time period from 1979 to 2008 along a 260-m horizontal by 7-m vertical cross-section. Other aspects of the model including initial concentrations, grid discretization, permeabilites, dispersivities, dissolution, equilibrium and kinetic reactions, outgassing, and sorption are described in Ng et al. (2015). For the Cape Cod Wastewater Plume, the modeling goal was to understand what controlled the recovery rate of aerobic conditions in the aquifer after the cessation of wastewater disposal. The time frame of recovery depended on re-oxidation rates of two pools of reduced species that had accumulated on the aquifer sediments during the 60 years of wastewater disposal pond operations.

SWQI Database

Analysis of the SWQI database results provided significant insights into the production and attenuation of SWQI parameters. Maximum TOC concentrations decline rapidly with distance downgradient of the injection zone. As a result, CH₄ production and SO₄^2- reduction are largely restricted to the injection zone and immediately downgradient. While SWQIs were detected in downgradient wells, the large majority of these wells were within the primary contaminant plume. In wells where chlorinated ethenes were previously below maximum contaminant levels (MCLs), SWQIs were generally within background levels or within a few meters of the injection zone. As a result, SWQI production is unlikely to adversely impact potable water supplies.

This research has improved our understanding of the near- and long-term impacts to groundwater quality after implementation of in situ anaerobic bioremediation approaches. Project results were used to develop a general Conceptual Model and guidance document “Extent and Persistence of Secondary Water Quality Impacts after Enhanced Reductive Bioremediation”. This document provides information on the microbiological and geochemical processes controlling production and attenuation of secondary water quality impacts and typical characteristics of secondary water quality impact plumes and can be used by project managers, consultants, and regulators to evaluating the duration, extent, and magnitude of secondary water quality impacts and to develop monitoring and management approaches to minimize these impacts.

Several results from this study show that natural attenuation of SWQIs can be effective. The first concerns the fate of aqueous Fe[II] mobilized during the active phase of ERB. Modeling of the Bemidji crude oil spill site demonstrated that 91% of mobilized Fe[II] is readsorbed to the sediments in the reducing zone, preventing migration from the treatment zone (Ng et al., 2015). In addition, a field study of the fate of mobilized arsenic at the Bemidji site shows that the arsenic mobilized by natural attenuation of crude oil is re-adsorbed over the same interval as the Fe[II] (Cozzarelli et al., 2015). Production of CH₄ and transport from the treatment zone can also be a concern during ERB. The Bemidji modeling results show that 70% of the produced CH₄ enters the gas phase and is oxidized before reaching the surface. This greatly limits the migration of CH₄ in the groundwater. The recovery time of aerobic conditions in an aquifer following completion of ERB was examined with modeling of the Cape Cod wastewater study site. The results showed that recovery of aerobic conditions can take decades, but dissolved Fe²⁺ does not appear to migrate long distances (Ng et al., in prep.).