6

Remediation of Fractured Rock

The most problematic contaminated sites are those with persistent contaminants—that is, chlorinated solvents that are recalcitrant to biodegradation in hydrogeologic settings—found in settings characterized by large spatial heterogeneities or the presence of fractures (NRC, 2013b). Furthermore, the deeper below the surface the contamination is, the more problematic remediation tends to be. In the four decades since remediation began in earnest at contaminated groundwater sites, remediation success has been achieved mostly when contamination was shallow, localized, in simple hydrologic settings, and comprised of a single or moderate number of compounds.

In 2004, the U.S. Environmental Protection Agency (EPA) estimated that more than $209 billion would be needed to mitigate these hazards over the next 30 years (EPA, 2004)—likely an underestimate because this number did not include sites where remediation was under way or had transitioned to long-term management. The majority of technologies developed for porous media remediation are not applicable in fractured rock settings (NRC, 1994; SERDP, 2001), and the time frames for remediation can be longer than estimated for unconsolidated porous media sites (without significant contaminant storage) because of slow release of contaminants stored in the rock matrix. Remediation of fractured rock is complicated by, among other things, adequately characterizing the flow system, obtaining representative measurements of spatial and temporal contaminant concentration distributions in groundwater, and the fact that the majority of contaminant may reside in the fractured rock matrix, making the contaminant essentially inaccessible.

Pump-and-treat containment strategies¹ are applied often to remediate many fractured rock sites, but often without detailed knowledge of site hydrogeology, hydraulics, or contaminant location. Although employed frequently, pump and treat is better in the context of containment rather than remediation. Technical details, designs, and monitoring strategies for pump-and-treat remediation are readily found in other works (e.g., Gorelick, 1993; Cohen et al., 1994, 1997; EPA, 1996, 2002, 2008).² Over the past decade, in situ techniques such as bioremediation, chemical oxidation, and thermal treatment have been employed increasingly at fractured rock sites. The majority of these approaches (with the exception of thermal treatment) require relatively uniform and predictable contact with the dissolved contaminants, increasing greatly the level of site investigation required as compared to pump-and-treat approaches.

Perhaps the greatest impediment to remediation in fractured rock settings is the lack of common framework, understanding, or expectations related to assessment and realistic remediation endpoints. For example, chemical site investigation data are presented often as contours based on measured concentrations in mobile fluids in fracture porosity (i.e., fractures that transmit fluid). Contours are generated using some algorithm that assumes a linear and structured relationship between data points. Such an approach, however, does not honor the true relationship between discrete data points and is inconsistent with the current understanding of contaminant location if it

___________________

¹ Pump-and-treat strategies often involve withdrawal of contaminated water from the ground for treatment. A broader definition of the term was adopted by EPA (1996) to include any remediation system that includes withdrawal of or injection into groundwater.

² Information searches on a range of topics are easily performed using EPA’s Contaminated Site Clean-Up Information search engine at http://clu-in.org.

is being used to represent remediation performance. A second example is the use of numerical modeling in the design of remedies and prediction of their performance. Historically, little thought has been given to the ability of the underlying equations and assumptions to accurately represent the complex physics of the processes occurring in fractured rock (see Chapter 4). Design underpinned by the use of single porosity models (as an example) is inherently faulty, although that limitation may not be recognized by all appropriate parties.

DIFFICULTIES OF REMEDIATION IN FRACTURED ROCK

Characterization

As discussed in Chapter 5, the level of detail in characterization required for remediation of fractured rock sites exceeds significantly that required at most unconsolidated media sites and may not be possible given budgetary and technology limitations. At the most basic level, any use of source-zone in situ treatment approaches will be challenged by the need to determine the three-dimensional treatment volume. Current hydrostructural models for fractured rock (see Chapter 4) and conceptual models for contaminant distribution indicate the possibility of contaminant penetrations at depths where measurements of depth of penetration of contaminants at fractured rock sites is not trivial. Investigations to hundreds of feet in depth may be required (Parker et al., 2010; see Figure 6.1). Core sample analyses for the entire length of boreholes, in combination with strategically applied non-invasive techniques as described in Chapter 5, may be needed to determine flow distribution and transport and better characterize the depth of the treatment zone. This would be applicable when considering relatively near-surface contamination, as opposed to deep geologic storage or impacts from deep geologic waste disposal. In certain situations characterization that defines treatment volumes may be more costly than the remedy or the point-of-use or point-of-extraction water treatment systems.

In addition to accurate estimates of the spatial extents of the treatment zone, most remedial approaches that target source zones require an understanding of hydraulics at the individual fracture level for hydraulically important fractures (see Chapter 5). Remediation approaches such as conductive heating can be rendered ineffective if a single highly transmissive fracture is not detected during characterization. Similarly, methods that rely on injection and advection of treatment fluids or amendments will be compromised if a highly transmissive zone is undetected. Rapid migration of injectants and amendments away from the treatment zone can result.

The transmissivity measurement interval used within a borehole can be related directly to estimates of transmissivity. Figure 6.2 shows the measured transmissivity along a borehole through dolomite obtained at different test intervals (Novakowski et al., 2000). Measurements were obtained by packing off different lengths of the borehole and performing the test. Smaller packer spacings may test only a single fracture (or no fractures), whereas larger spacings will average a number of fractures and the interspersed matrix. The figure indicates that measured transmissivity is a function of measurement spacing and demonstrates the extreme variation in hydraulic conductivity that exists in fractured rock settings. This occurs between not only fractures and matrix, but also fractures themselves. This is one of the most critical factors in the design of remediation and ultimately of performance monitoring systems.

Understanding Plumes

The highly variable nature of flow paths in fractured rock settings presents challenges for plume mapping that are rarely of concern in unconsolidated media. In contrast to flow in sands, the migration of contaminants in fractured rock pathways is not resolved easily from coarse monitoring data, and plume migration can dissociate from the averaged direction of groundwater flow interpreted from boreholes with longer screen lengths. Site investigations designed to delineate contaminant plumes, therefore, need to account for the strike and dip of fracture sets when, for example, identifying the location and screen intervals of boreholes. Transport in sedimentary rocks is often along horizontal bedding plane fractures, where “stair-stepping” of contaminant plume is common. A single or series of highly transmissive vertical connecting fractures can result in deviation of some of the plume to higher or lower planes. Vertical gradients and their spatial variations are key controls on the mechanics of plume migration and need to be mapped in conjunction with horizontal gradients.

The depth of contaminant migration in fractured rock (particularly at dense non-aqueous phase liquids [DNAPL] sites) needs to be known to predict the ultimate discharge point of the plume and the extent of groundwater resources that have been impacted. Experience has shown that the majority of plumes in unconsolidated media where the water table is relatively shallow (with the possible exception of sources located within deep groundwater recharge zones) have readily definable discharge points, usually to surface water features such as streams, rivers, or lakes in close proximity to the source of the plume. Investigations of groundwater ages with depth in fractured rock (e.g., Gascoyne, 2004; Palcsu et al., 2007; Takahashi et al., 2013) show that water contained in fractures at depths of greater than 100–150 meters is often on the order of tens of thousands of years old and imply that flow paths from recharge to discharge are potentially basin sized. The process of determining the location, and mapping the flow path, therefore, would require investigations at a regional scale.

The process of diffusion of contaminant into the rock matrix (see Chapter 3) along a flow path results in retardation of the plume migration velocity and a reduction in contaminant concentration. This reduction is often exaggerated as a result of averaging the measured concentrations in samples from monitoring wells with long screen lengths. In such cases, flow in the borehole can be dominated by fractures that are not the most contaminated. The combination of these processes, in conjunction with the difficulty identifying plume centerlines, often results in plumes in fractured bedrock being described as “dilute” or “diffuse.” Treatment approaches for plumes in fractured rock conceptualized as large, diffuse, and unstructured are limited if not non-existent.

Access to Contaminants

Remediation at most fractured rock sites is confounded by the large ratio of contaminant in the matrix as opposed to fractures (see Chapter 3). Parker et al. (1994, 1997) demonstrated that the lifespan of a DNAPL in a fracture system can be measured in days to years in many cases, which results in most sites being categorized as late stage by the terminology of Parker (2007; see Figure 6.3). A site in a late-stage configuration has essentially identical contaminant distributions in the source and the plume, and the benefit of contaminant removal from fractures alone is limited.

Remediation approaches such as in situ chemical oxidation (ISCO; e.g., Werner and Helmke, 2003; Goldstein et al., 2004; Shaefer et al., 2012) and enhanced in situ bioremediation (EISB) rely on achieving contact between the oxidant and the contaminant (for ISCO) or the degrading

microorganism, the electron donor, and the contaminant (for EISB). Hydraulic injection and advective migration of amendments (such as electron donors or oxidants) will preferentially follow fractures, particularly those of higher permeability. The effectiveness of these approaches is therefore limited because it is difficult to achieve uniform distribution within the fracture system where the contaminants may reside. Furthermore, because much of the contaminant is ultimately stored in the matrix, contact between amendments and contaminants relies on diffusion of the amendment into the matrix—a rate-limiting step in the remediation process.

Numerous techniques have been proposed for overcoming the rate-limiting diffusion step in oxidation and bioremediation approaches (e.g., use of shear-thinning additives, multiple amendment injections, high concentration amendment injections, use of electrical fields to rapidly migrate amendments into the matrix, hydraulic fracturing, creation of biofilms on fracture surfaces to treat back-diffusing contaminants [see Hill and Sleep, 2002; Charbeneau et al., 2006; Reynolds et al., 2008; Zhong et al., 2011]). Most of these techniques remain unproven in field settings. The use of biological approaches for treatment of contaminant that resides in the matrix may be limited by the ability of degrading bacteria to access, migrate, and survive within the matrix porosity of the rock (see Chapter 3).

POTENTIAL TECHNOLOGIES TO REMEDIATE ORGANIC COMPOUNDS

Kinner et al. concluded in 2005 that the limitations of remediation technologies that are applied to porous media are not known for application in fractured rock. Challenges remain in the form of delivery and distribution of injected material and in remediation of microfractures and low-flow zones in the rock matrix or at very large scales. The same language about the states of knowledge and practice used in 2005 can be applied almost 10 years later. Excluding containment and management remedies such as pump-and-treat and encapsulation techniques (i.e., slurry walls), four remediation approaches remain of potential interest for selected remedial goals at fractured rock sites: bioremediation, ISCO, thermal methods, and monitored natural attenuation.

Biological Approaches

Cost and sustainability indicators make the use of biological remediation in fractured rock attractive. More than 55 percent of sedimentary fractured rock sites contaminated with chlorinated solvents registered by EPA have detectable levels of biodegradation products in groundwater (EPA, 2014a). A significant amount of research and testing has shown that biological treatment is possible within fractures themselves (Hohnstock-Ashe et al., 2001; Lenczewski et al., 2003; Macbeth et al., 2004; Darlington et al., 2008; Bradley et al., 2009; Schaefer et al., 2010). However, limited work has been done on the ability of biological approaches to perform effectively in the matrix porosity of fractured rock.

Historical speculation has been that matrix pore sizes are a limiting factor for microbial growth and transport within sedimentary rock matrices (Lima et al., 2012). However, microbial growth in the matrix porosity of sandstones has been shown in both laboratory and field investigations (Jenneman et al., 1985; Krumholz et al., 1997; Lima et al., 2012). The existence of cells within the matrix porosity of sandstones is not evidence of their ability to contribute to contaminant degradation, and the potential for microbial existence in less porous rocks with smaller pore sizes and less pore interconnectivity is so far unproven. The observed decline in microbial abundance with increasing depth has been associated with various environmental factors; however, the role of geometrical constraints and soil-bacteria mechanical interactions remains poorly analyzed (Rebata-Landa and Santamarina, 2011). Pore sizes may restrict habitable pore space and traversable interconnected porosity (see Chapter 3), and sediment-cell interaction may cause puncture or tensile failure of cell membranes. In addition, because of nutrient and electron donor diffusion limitations, there may be a finite depth to which microorganisms can penetrate into the rock matrix before growth requirements cannot be met (Yu and Pinder, 1994).

Lima et al. (2012) identified three dechlorinators within the matrix porosity of a sandstone contaminated with chlorinated compounds at a distance of 64 centimeters from the nearest fracture. Greater heterogeneity in the microbial population was observed closer to fractures. In general, Lima et al. (2012) found that heterogeneity in microbial populations was present across all samples, even those from closely spaced intervals, indicating the heterogeneity of the rock matrix itself and possibly the heterogeneity of nutrient distribution and availability.

In fractured rock environments, biofilm development has been shown to have a potentially significant impact on network fluid flows (Ross et al., 2001; Hill and Sleep, 2002; Charbeneau et al., 2006; Smith, 2010) through the clogging of transmissive fractures and the resultant diversion of flow into less permeable areas (see Figure 6.4). A biofilm growth layer that is limited to the surface of fractures, and does not clog or significantly reduce the fracture permeability, has the potential to degrade contaminants during the process of back diffusion from the matrix, providing a treatment “barrier” within the individual fractures themselves.

Despite many years of research and field testing, the use of biological remediation approaches in fractured rock environments is still in embryonic stages. Little is known or written about biological remediation for fractured rock at great depth. It is reasonable, however, that biological approaches are possible and should be considered at fractured rock sites. A combination of engineered biostimulation and natural attenuation may provide effective solutions in many cases. The optimal approaches, engineering difficulties, time frames, and realistic endpoints for contaminant bioremediation within fractured rock sites are still poorly understood and represent large research needs.

In Situ Chemical Oxidation and Reduction Approaches

In situ chemical oxidation (ISCO) has a long history of enhancing contaminant removal in unconsolidated porous media. Common oxidants include permanganate, persulfate, Fenton’s reagent, and to a lesser extent, ozone. Few Records of Decision³ include the use of chemical oxidation for source zone remediation in fractured bedrock. Laboratory experimentation on the use of ISCO in fractured settings has shown little impact on dissolved concentrations exiting fractures following flooding; however, contaminant discharge from the fractures decreased, likely because of buildup of oxidation byproducts and resultant decreases in fracture permeability and contaminant transfer rates (Cho et al., 2002; Tunnicliffe and Thomson, 2004; Schaefer et al., 2012). The rate of NAPL removal within fractures during oxidant flooding has been shown to decrease quickly, indicating that NAPL removal is controlled by mass transfer at the NAPL-water interface. However, the overall limits of NAPL removal are controlled by both the mass-transfer rate and the aperture distribution (Schaefer et al., 2012). Research on the interaction of oxidants with matrix-held contamination is extremely limited and has not included physical experimentation. Mundle et al. (2007) found no well-defined correlation between the magnitude of rebound concentrations arising from back diffusion and the percentage of total contaminant destroyed within the fracture. Pang (2010) highlighted that greater than 90 percent of oxidant was consumed by natural organic matter, and the effectiveness of the process was limited by the rate of diffusion of the reaction front into the rock matrix.

___________________

³ A Record of Decision is a public document describing the cleanup strategy to be employed at a Superfund site.

The inability of chemical-oxidation approaches to adequately remove contaminant from fractures where DNAPL is present indicates that ISCO approaches are likely best suited to late-stage scenarios (see Figure 6.3). However, the rate-limiting step of the diffusive transport of the oxidant into the matrix porosity will limit the application of the technology, requiring either multiple injections over many years, or delivery methods that can accelerate the migration rate of oxidant into the matrix.

Multiple studies have been published on the use of chemical oxidation in fractured rock at the field scale (Werner and Helmke, 2003; Goldstein et al., 2004, 2007; Helsen et al., 2007). In all cases, difficulties were encountered achieving adequate distribution of the oxidant throughout the treatment zone, rapid rebound of contaminant concentrations to pre-treatment levels, or plugging of fractures and inability to deliver design levels of oxidant.

Gefell et al. (2002, 2003, 2004, 2007, 2008) present the design and preliminary results of a multiple-injection ISCO-remediation pilot project in fractured siltstone and shale bedrock. The intent of the project is to provide permanganate to the fracture porosity through multiple (up to 20) injections occurring approximately every 6 weeks, such that concentrations remain elevated (target in fracture porosity is greater than 20 mg/L) to maximize the potential for diffusive flux into the matrix porosity. Although final results have not been published as of this writing, preliminary results indicate that permanganate transport has occurred up to 100 feet from the injection locations, with peaks appearing from 1–6 weeks following injection. Contaminant concentrations in the target zone and downgradient are decreasing. However, no formal assessment of rebound or degree of penetration and treatment in the rock matrix has been documented.

The use of oxidation and reduction remediation approaches in fractured rock environments is more advanced than biological approaches, but there is still a lack of well-documented case studies, particularly for large-scale implementations. Oxidation and reduction approaches are plausible

and need to be considered for remediation of fractured rock sites, but optimal approaches, time frames, and realistic remediation endpoints still need to be determined, and significant effort to obtain permits may be required.

Thermal Approaches

In situ thermal treatment (ISTT) approaches have been developing and applied, primarily in unconsolidated porous media settings, since early trials in the late 1980s. The most commonly implemented versions of ISTT include electrical resistance heating (ERH), thermal conductive heating (TCH), and steam-based heating (Triplett Kingston, 2008). Technical descriptions of those can be found in detail in Johnson et al. (2009). All ISTT approaches involve elevating the temperature of the soil and groundwater within the impacted zone (i.e., source zone or plume) such that volatilization, mobilization, and direct destruction of the chemical compounds occur. Relative to other technologies, some in situ thermal treatment technologies (e.g., ERH) result in preferential heating and contaminant removal from lower permeability media (NRC, 2013b). To date, there are no reported uses of ERH at sites where contaminants are primarily within fractured rock.

The majority (93 percent) of published applications of ISTT since 2000 have been in unconsolidated geologic settings (Triplett Kingston et al., 2010). Steam heating was employed at four of the six fractured rock sites, with TCH and other approaches each employed at one site. Of 14 sites with sufficient data to assess performance in terms of concentration and contaminant flux reduction identified by Triplett Kingston et al. (2010), only one was in fractured rock (described as competent but fractured bedrock), and measured reductions in source zone concentration and contaminant flux immediately downgradient of the source zone were less than a factor of 10.

In comparison to fluid flushing technologies (e.g., oxidant flushing and surfactant flushing), ISTT offers distinct advantages. Heat migration is not as adversely affected by geological heterogeneity as is fluid migration. In comparison to other thermal technologies, TCH has the advantages of (1) not relying on fluid injection (e.g., steam flooding) to deliver heat; (2) being able to achieve temperatures above boiling (impossible with steam flooding or ERH); and (3) being able to destroy contaminants in situ as a result of the high temperatures achieved, thereby reducing the need for ex situ produced fluids treatment. To date, TCH has been implemented in four fractured rock settings:

• A pilot-scale demonstration of TCH in fractured chalk (at the United Kingdom Atomic Energy Authority site in Harwell, United Kingdom) completed in 2005 demonstrated a fourfold increase in soil vapor extraction rate in chalk when the unsaturated zone was heated to ~100°C (CL:AIRE, 2010).
• A pilot-scale test at the National Aeronautics and Space Administration Marshall Space Flight Center Source Area 13 in Huntsville, Alabama, was undertaken in 2007 (Cole et al., 2008). Only the bottom 5 feet of the treatment volume was limestone bedrock, and the majority of the treatment volume was unconsolidated materials.
• At another site in the southeast United States, the total saturated thickness of approximately 25 feet of soil and partially weathered bedrock overlying fractured bedrock was the focus of the treatment zone, with conductive heating occurring approximately 10 feet into the fractured gneiss bedrock (Heron et al., 2008). The effectiveness of TCH on the removal of contaminants from the fractured rock was not demonstrated because of the lack of pre- or post-treatment samples within the rock to quantify contaminant removal.

• TCH in fractured rock was performed at the Naval Air Warfare Center research site in Trenton, New Jersey (Lebrón et al., 2012). Bedrock sample analyses indicate the average reduction in trichloroethylene (TCE) concentrations was 41–69 percent. However, the rock matrix did not achieve targeted temperature in all locations (mostly because of contaminated groundwater influx through existing fractures). Discrete sampling at 5-feet intervals, correlated with observed fractures from borehole video logs, allowed identification of the depth where heating was incomplete and lower levels of contaminant reduction occurred.

Modeling studies indicate careful attention to groundwater influx into a target treatment zone is warranted to determine whether and how long boiling water temperatures can be reached in all locations (Lebrón et al., 2012). Given potential variability of individual fracture flow rates, accurate assessment of the influence of inflowing cold groundwater at the fracture scale is required. Lebrón et al. (2012) indicate that in the case of low groundwater inflow, only quantification of the total groundwater influx through the treatment zone is necessary, rather than characterization of discrete fractures.

Low matrix permeability, high matrix porosity, and wide fracture spacing can contribute to boiling point elevation in the rock matrix. Consequently, knowledge of these properties is important for the estimation of treatment times. This is particularly relevant in low matrix permeability rock where thermal expansion of groundwater leads to pressure increases, which in turn result in elevated water boiling points. Because of the importance of fracture spacing in determining the pressure rise in the matrix, a discrete fracture model is more appropriate than an equivalent porous medium model for simulating boiling.

Predicting Remediation Performance in the Context of Feasibility Studies

Under the Comprehensive Environmental Response, Compensation, and Liability Act, a critical step in the investigation and remediation of contaminated sites is the remedial investigation/feasibility study (RI/FS), which is designed to develop a site-specific, scientifically sound, rigorously supported site remediation strategy. The investigation stage of an RI/FS at a fractured rock site is complex and difficult, given the challenges and costs of data acquisition, data integration, and conceptual model development. The feasibility study stage is additionally challenged given the relatively small number of remediation attempts in fractured rock and the lack of high-quality refereed studies that provide unbiased, critical assessment of remediation success and failure that help to quantify uncertainties associated with various remediation strategies.

The RI portion of the process involves characterization of the site and contaminant, evaluation of risk to human health and the environment, and treatability testing to evaluate the potential performance and cost of the treatment technologies under consideration. The use of treatability studies is well established and understood for unconsolidated sites, but it is less mature for contaminated sites in fractured rock. Properly implemented treatability studies provide key data for the detailed consideration of remedial alternatives, as well as indications of the potential limitation of the remedial alternative being tested. Common treatability studies include reactor testing for bioremediation rate determination (Shah et al., 2009), column flow-through studies for assessment of interactions with site soil or rock (Schaefer et al., 2011; Simmons and Neymark, 2012), batch studies for chemical oxidant effectiveness and consumption (Helsen et al., 2007), and thermal properties testing (Bastona and Kueper, 2009).

Treatability studies can provide the data required to optimize process engineering, determine likely duration of treatment, and forward the design to considerations of scaling at pilot and full scales. Implementation of treatability studies for fractured rock sites requires careful consideration of the differences between laboratory and field conditions. Treatability studies using crushed rock or artificial representations of the rock (induced fracturing) will not replicate the fracture surface area to rock ratio in situ, nor will any flow field established in a laboratory setting accurately reflect the conditions in the field. Treatability studies can and should form a part of any remedial alternatives investigation at fractured rock sites. However, they must be carefully planned and their limitations fully understood and documented.

The FS portion of the process is the mechanism for the development, screening, and detailed evaluation of alternative remedial actions. Pilot testing is sometimes undertaken in the FS stage, particularly if the technology is new or being implemented in a new environment or on a new contaminant. Although not documented, experience on the part of committee members indicates that pilot studies in fractured rock are often costly, difficult to design, and prone to less success than in unconsolidated porous media. Even well-designed pilot tests following significant site characterization and treatability testing (Goldstein et al., 2004; Pearson et al., 2004; Lebron et al., 2012) can fail to meet expectations.

Very little is being published in the peer-reviewed or publicly accessible literature on pilot tests at fractured rock sites. The lack of accessible material does not necessarily indicate that pilot studies are not being undertaken, but it might be taken to indicate a lack of success and, therefore, little drive to publish. In a review of 393 published case studies (FRTR, 2007), none was identified as fractured rock sites. Pilot studies at contaminated fractured rock sites are of value and need to be performed in a rigorous design and execution framework, subject to independent peer review and published to avoid repetition of previous mistakes by practitioners.

POTENTIALLY APPLICABLE APPROACHES FOR RADIONUCLIDES

Radionuclides are naturally decaying inorganic elements that migrate in the subsurface similarly to inorganic materials such as salts and metals. Positively charged radionuclides tend to react with and be sorbed by mineral constituents of rock and, therefore, migrate more slowly than flowing groundwater. Some radionuclides, such as tritium, are nonreactive and migrate by advective transport at essentially the same rate as flowing groundwater. Some radionuclides may sorb to colloidal particles and undergo transport relatively unaffected by geochemical processes. Still other radionuclides are disposed of in the subsurface with high or low pH and may undergo a myriad of reactions that result in precipitation or dissolution. Thus, essentially the full range of geochemical phenomena that one might expect for inorganic contaminants applies to radionuclides. The fact that certain elements might be radioactive affects their potential to affect human health, but in and of itself has little effect on mobility in the subsurface. Treatment of radionuclides removed from the subsurface during remediation may require special technologies and requirements for ultimate disposal.

NATURAL ATTENUATION

Natural attenuation is the reduction in concentration of contaminant in groundwater as a result of naturally occurring biological or chemical processes, dilution, or evaporation that may occur under the correct environmental conditions. Monitored natural attenuation (MNA) is a strategy often applied in the cleanup of chlorinated solvents, petroleum hydrocarbons, and other hazardous wastes in conjunction with source zone cleanup. Biodegradation is an important form of natural attenuation, relying on naturally occurring microorganisms to cause or result in the breakdown of various contaminants into less harmful or inert components.

Matrix diffusion can be considered a natural attenuation process, because it attenuates the rate at which contaminants migrate in the forward direction, and it attenuates the contaminant discharge into the mobile fluid in the reverse direction. Dispersion is also a significant process occurring within fractured rock settings, particularly in crystalline rocks containing multiple hydraulically connected fracture sets resulting in the dispersion of a given contaminant through an ever increasing volume along the flow path.

Natural attenuation through matrix diffusion and dispersion may be less strong in weathered sedimentary systems than in crystalline systems, particularly where fractures are laterally extensive and have large apertures, yet matrix porosities remain low. Plumes generated from strong, high-contaminant sources may grow to significant lengths under such circumstances and only terminate at natural discharge points such as surface water bodies, or at artificial discharge points such as extraction wells. An example of this situation is the Cayuga Springs Superfund site in New York State, where a high strength TCE source exists in a highly transmissive limestone unit approximately 150 feet below the ground surface (EPA, 2012). The resultant plume from this source extends more than 5 miles to a discharge point at a surface water body. Both biotic and abiotic transformation of TCE to daughter products is occurring, primarily because of the high levels of natural organic carbon in the limestone unit. However, the levels of natural attenuation are not sufficient to reduce the contaminant discharge at the receptor to a point where there is no unacceptable risk to human health.

IMPORTANT CONSIDERATIONS IN FRACTURED ROCK REMEDIATION

A conundrum in setting remedial goals, criteria, or even objectives for fractured rock remediation is the potential that the most basic underlying assessment metric—point concentrations in mobile fluids in fracture porosity—is entirely inappropriate. Proper characterization of risk, long-term liability, and performance may require the use of alternate metrics, or additional metrics in conjunction with point source concentration measurements.

Meeting drinking water standards at the points of monitoring, or at actual or potential receptors, is the long-term goal of remediation at most hazardous waste sites in the United States. However, established drinking water maximum contaminant levels (MCLs) are not likely to be met for decades or perhaps centuries for contaminated sites in fractured rock settings. Because U.S. regulatory bodies are becoming more accepting of such long time frames, it is also becoming practical to include alternate approaches to setting goals and objectives for remediation, particularly at fractured rock sites (EPA, 2011). EPA has issued draft guidance that attempts to address setting remedial objectives and formalizes the transition from active remediation to long-term monitoring (EPA, 2014b). The National Research Council (NRC) also discussed this issue, referring to it as a “transition assessment” (NRC, 2013b).

Underlying the difficulties of setting scientifically based, reasonable, and appropriate objectives is the lack of reliable, open-access, peer-reviewed data from previous experience to support the process of constructing objectives. There have been few technical demonstrations of remediation approaches—and particularly as they vary between different settings—to provide a baseline for understanding the performance of the various technologies. Similarly, there have been insufficient controls available to assess the actual long-term performance of remediation efforts in fractured rock, particularly as the performance assessment relates to incremental change in secondary performance indicators (such as contaminant transfer between porosities downgradient of the treatment volume). The confluence of these limitations results in conclusions that are often based on inference as opposed to adequate and appropriate quantitative measures of performance.

This chapter presents a series of key concepts important at an overview level (as opposed to a detailed technical level). Multiple reference sources exist for detailed engineering design, scientific theory, and policy considerations, and the reader is directed to these other more appropriate sources for detailed considerations.

Concept 1. Remediation focused solely on removal or destruction of contaminant in fractures is futile.

As has been discussed in this report, the majority of contamination is often found in the matrix of fractured rock, and this contamination will not be remediated with traditional approaches that focus on remediation of groundwater in rock fractures. Matrix diffusion will dominate the source term in most contaminated settings and will result in plume lifespans that could reach thousands of years. To reduce contaminant concentrations at a given location, remedial approaches must focus on the contaminant held in the matrix porosity of the fractured rock (Parker et al., 2010).

Concept 2. Contaminant source zone remediation has little or no impact on plume lifespans in settings with high matrix porosity.

A remediation effort focused solely on the contaminant source zone in a fractured rock setting is not likely to have a discernible impact on concentrations in downgradient plumes for tens to hundreds of years in rock with high matrix porosity. Parker et al. (2010) used numerical modeling to demonstrate the impacts of source removal on downgradient plumes (see Figure 6.5). A comparison of panels (c) and (d) in Figure 6.5, where (c) represents contaminant concentrations 50 years after complete source removal and (d) represents contaminant concentrations at the same point in time without source removal, demonstrates the calculated marginal impact on the plume. A similar conclusion was reached by Lipson et al. (2005), the main point of which is shown in Figure 3.6 in Chapter 3. Plume lifespans are shown to exceed 1,000 years following complete source removal.

Concept 3. Significant benefits in terms of plume lifespan reduction are more likely through treatment of contaminant within the fractured rock matrix.

Current conceptual models of contaminant distribution in fractured rock settings indicate the importance of contaminant remediation in both the matrix and fracture porosities. This extends beyond source zone remediation to include plume remediation and is a significant difference from remediation in porous media. To achieve significant reductions in concentrations within the mobile

fluid in fractured rock at distances from the initial contaminant source (i.e., at a receptor downgradient from the point of contaminant release), contaminant removal from the fracture porosity along the flow path from the point of release to the receptor is necessary.

The requirement for treatment of fracture porosity along flow paths has a direct and significant correlation with complexity of remedial design, cost, and resistance from responsible parties. In the majority of cases, given the resources required for such extensive remediation and the technical difficulties of even achieving such remediation, a careful and considered assessment of the ultimate remedial goals is necessary.

Concept 4. Better characterization leads to better remediation design, better monitoring, and, ultimately, more successful remediation.

The difficulties inherent in fractured rock characterization (see Chapter 5) and the increased cost as compared to characterization in unconsolidated media combine to increase the potential for poor characterization in all stages of investigation at fractured rock sites, especially at large depth. It is well recognized that conceptual site model development needs to be a dynamic process, with new information feeding back to the existing conceptual model and revised conceptual models informing additional investigations (see Chapter 7). This process is critically important in fractured rock settings, and failure to embrace the process is likely to lead to poor remediation design and failure to meet remediation objectives.

A dynamic conceptualization process should be the default practice among practitioners at fractured rock sites (see Chapter 7). However, evidence indicates that either the practice is not followed or the understanding of the process endpoint in terms of uncertainty reduction and provision of adequate, representative, and fit-for-purpose data is limited. An overall indication of the lack of adequate characterization (in both fractured and unconsolidated media) is provided by Triplett Kingston et al. (2010), where only 14 of 182 thermal remediation projects had adequate infrastructure to assess success, indicating that initial conditions for the rest of the projects were not resolved with sufficient certainty to provide a baseline.

The potential impacts of very small-scale site features in fractured rock settings increase the importance of adequate characterization at relevant scales. A single fracture in an anomalous location, or with an “outlier” transmissivity not readily apparent from core examination, has the ability to severely degrade the performance of a remediation system (such an occurrence was documented during a thermal pilot test; Lebrón et al., 2012). A dynamic conceptualization process enhances understanding of where remediation efforts could fail as a result of incomplete characterization, allows examination of failure mechanisms unique to an individual remedy, and can inform a more cost-effective detailed site investigation that, in turn, leads to more effective remedy design.

Concept 5. Better characterization and performance monitoring leads to more honest appraisal of success and failure.

Sufficient characterization of the treatment zone and the larger likely monitoring zone surrounding it is critical for realistic assessment of remediation success and allows for realistic discussion with owners, operators, regulators, and other stakeholders. Degradation of remediation performance that results from a single feature, as described above, is a logical performance monitoring point during and following remediation. Failure to monitor such a feature, which may transmit a high percentage of contaminated flow through the treated volume, would bias performance monitoring and would provide non-representative data for remediation performance assessment.

Concept 6. Natural attenuation is a required component of all remediation solutions in fractured rock.

Because of the skewed distribution of stored contaminant toward the matrix porosity, the difficulty of treating contaminant within the matrix porosity, and the difficulties inherent in the identification and location of fluid flow within the fracture porosity, natural attenuation is likely to be

a required component of any remedial approach undertaken in fractured rock. Natural attenuation processes occur in fractured rock to greater and lesser degrees than in porous media, and an understanding of the effectiveness, magnitude, and implications of these processes are crucial to the proper incorporation of natural attenuation into remedial approaches. Effective characterization and parameterization, consideration, and explicit inclusion of matrix diffusion and dispersion as a component of remedial approaches will increase the potential for achieving remedial goals and may even play a role in the setting of realistic and achievable remedial goals at fractured rock sites.

Concept 7. Monitored natural attenuation may not be a suitable sole remedy in some fractured rock settings where risk to human or ecological health is severe.

The occurrence of natural attenuation processes within fractured rock systems, and their demonstration at a particular site, is not necessarily sufficient for the use of monitored natural attenuation as a sole remedy (e.g., EPA, 1999a). Reaction processes (i.e., biodegradation and chemical) that reduce contaminant within a system and that are fundamental to achieving a steady-state plume configuration are potentially less common and likely less strong in the majority of fractured rock settings as compared to porous media settings (Chapelle et al., 2012; Lima et al., 2012). Organic carbon, nutrients, and electron donors are often found in insufficient quantities in fractured rock environments to promote and promulgate biodegradation, which results in less dominant reaction processes and longer and continually expanding plumes. The reduced magnitude of these processes in fractured rock is often masked by matrix diffusion over relatively long timescales (as compared to porous media) imparting an artificially increased importance to reactive processes.

A high level of confidence in monitoring is required to effectively use MNA. Principles of MNA require a suitable, representative, conservative, and fit-for-purpose monitoring regime (Wiedermeier et al., 1998). The level of uncertainty in monitoring programs at many fractured rock sites may exceed that allowable for MNA to be a protective approach, and the capacity to reduce that uncertainty at some sites may be beyond that of the responsible party. It is entirely conceivable that a line of evidence within the MNA framework may show that the plume has dissipated or depleted, where in fact the monitoring infrastructure is not properly positioned within the flow paths and is no longer monitoring the plume. The level of characterization required for acceptable uncertainty in the delineation of the plume in fractured rock systems is likely orders of magnitude above that required in most unconsolidated systems, and application of MNA will likely require demonstration of adequate characterization of fate and transport processes and pathways (e.g., EPA, 2001). Furthermore, natural attenuation may not reduce contaminants to levels sufficiently or quickly enough to avoid unacceptable risk to health, as demonstrated at the Cayuga Springs Superfund site (EPA, 2012).

Concept 8. Current metrics are not well suited for remedial objectives at fractured rock sites.

NRC (2013) proposed that better metrics for demonstrating remediation progress are needed that are based on quantifiable, transparent metrics of remedial performance and human health reduction rather than regulatory milestones. Contaminants typically cannot be removed from an interconnected fracture system to meet MCL cleanup goals unless some percentage of the contaminant mass in the matrix is removed. The contaminant flux through back diffusion needs to be reduced to a point where concentration-based standards in groundwater sampled from monitoring

wells (which is, in most cases, derived primarily from fracture flow) are met. A more suitable metric for remedial performance may be the reduction in concentration of contaminants within the matrix itself, coupled with demonstration of contaminant source removal (e.g., DNAPL) in the fractures. Such a matrix cleanup goal requires consideration of a coupled system of transmissive fractures and low permeability matrix (Rodriguez and Kueper, 2013).

Given the heterogeneous nature of transmissivity and the often complex flow patterns in fractured rock, the use of point measurements of concentration in monitoring wells as performance objectives is potentially flawed. As described in Chapter 5, monitoring wells in fractured rock may intersect a single, highly contaminated flowing fracture, and the measured concentration may not represent the average system concentration. The value, therefore, is not suitable for comparison to a risk-based criterion that assumes an exposure pathway (be it through dermal contact, ingestion, or as a source for vapors) comprised of only that water. Measurement of contaminant flux across a control plane is potentially a more applicable metric for determining the risk to human and environmental health in fractured rock environments.

Concept 9. At some point, active remediation is no longer appropriate, and long-term management should be formally implemented.

NRC (2013) has suggested that the transition to long-term management should be recognized formally as the point when further active remediation results in little or no decrease in contaminant concentration and when the unit cost of remediation increases much faster than reductions in contaminant concentrations. Under this approach, the transition point to long-term management (i.e., monitored natural attenuation) is likely to occur far earlier in fractured rock than in unconsolidated systems. However, both the suitability and the acceptability of this occurrence are less clear in fractured rock settings. For example, based on the predictions of Parker et al. (2010), any active remediation of the initial source mass (note that the initial source contaminant is representative of the classic unconsolidated source and does not consider back diffusion from the matrix porosity within the “plume” as a source) could be reasonably argued to have effectively little change on contaminant concentrations and thus not be required. Such an approach is not suitable as a default position for remediation at fractured rock sites, but it may be suitable as a considered position at some sites.

THE PATH FORWARD

The differential between the state of practice and state of the art in fractured rock investigation and remediation is greater than in unconsolidated porous media. The reasons for this are numerous and include less experience among practitioners, the greater difficulty of investigation in fractured rock, a much shorter history of detailed scientific understanding, and a residual impression among most stakeholders that the RI/FS process between fractured rock and unconsolidated porous media is similar, and so there is a high level of transferability of approaches and outcomes between the two. Reducing this differential in remediation will require new and increased effort in several areas:

1. Recognition and understanding of the differences between fractured rock and unconsolidated porous media.
2. Collaborative efforts between researchers, practitioners, and regulators to develop meaningful and obtainable outcome frameworks for remediation.

3. Scientifically rigorous and peer-documented demonstrations of various remediation technologies in different geologic settings.
4. Increased strategic characterization prior to remediation design utilizing appropriate techniques.
5. Development of monitoring strategies based on valid conceptual models, targeting appropriate locations and times, and specific parameters whose changes are related to remediation performance success or failure.

Improving practice will require aligning the goals and resources of researchers, funding agencies, regulators, site owners, and practitioners. Reduced research funding for fractured rock remediation studies leaves few options for implementing and managing a program or process to increase knowledge and improve practice. Without a governing body to focus and direct efforts on this process, and without inclusion of the oil and gas industry in the process, the knowledge gaps will continue to be a significant impediment to progress on effective management of the nation’s contamination issues in fractured rock.