Appendix I describes the methodology used to develop the CCW HBNs and provides the HBNs used in the screening analysis.
Sections 3.4 and 3.5), were much more effective at reducing risk for all constituents; 90th (and 50th) percentile risks with composite liners for landfills were zero 7 for arsenic and metals and
4.4 Variability and Uncertainty
4.4.3 Parameter Uncertainty and Variability
4.4.3.1 Waste Concentrations
The CCW constituent database used to represent CCW total waste and waste leachate concentrations is arguably the most important data set in terms of driving the risk assessment results. The constituent data are subject to two primary uncertainties beyond the normal sampling and analysis uncertainty associated with environmental measurements: (1) the appropriateness of the landfill leachate data used in the analysis and (2) high percentages of nondetect analyses for some CCW constituents.
Appropriateness of Leachate Data. The CCW leachate data were collected from a varying number of sites using a variety of methods. The available landfill data were largely derived from the TCLP, a laboratory test designed to estimate leachate concentrations in municipal solid waste (MSW) landfills. The TCLP has been shown to both over- and
underpredict leachate concentrations for other waste disposal scenarios, so the use of the TCLP data to represent CCW leachate is another source of uncertainty. However, as noted below, the TCLP data do appear to encompass the range of variability in CCW leachate concentrations that have been measured in more recent studies.
Surface impoundment leachate is represented by porewater measurements taken beneath actual impoundments, which should more closely represent the leachate seeping from the bottom of the impoundment than would bulk surface impoundment waste concentrations. The porewater is in direct contact with the waste, so these concentrations should typically be at least as great as concentrations in the bulk surface impoundment. However, although these porewater data arguably should better represent leachate concentrations, they are fewer in number than the landfill data and therefore subject to uncertainty as to how representative they are of all CCW
wastes. Results for surface impoundments for antimony, mercury, and thallium are not presented due to the paucity of leachate data (1 or 2 sites, and 11 or fewer values).
Since the CCW risk assessment was conducted in 2003, EPA-sponsored research
conducted by Vanderbilt University has improved the scientific understanding of the generation of leachate from CCW, in particular for mercury, arsenic, and selenium (U.S. EPA, 2006c; U.S. EPA, 2008c). Figure 4-2 plots the results from U.S. EPA (2006c) for arsenic and selenium, along with data from EPA’s Leach2000 database and EPRI (as provided in U.S. EPA, 2006c) against the data used for landfills and surface impoundments used in the CCW analysis.
Arsenic 0.0001 0.001 0.01 0.1 1 10 Brayton Point Pleasant Prairie Salem Harbor Facility C St. Clair Facility Facility L CCW - LF EPRI - LF EPA - LF CCW - SI EPRI - SI
Vanderbilt Study Other Data Sets
As ( m g /L ) max Natural pH min 95th % 50th % 5th % Selenium 0.0001 0.001 0.01 0.1 1 10 Brayton Point Pleasant Prairie Salem Harbor Facility C St. Clair Facility Facility L CCW - LF EPRI - LF EPA - LF CCW - SI EPRI - SI
Vanderbilt Study Other Data Sets
Se (m g /L ) max Natural pH min 95th % 50th % 5th %
Key to data sets:
Vanderbilt = U.S. EPA (2006c)
CCW = CCW Constituent Database (this analysis) EPRI = EPRI Leachate data (from U.S. EPA, 2006c) EPA = Leach 2000 data (from U.S. EPA, 2000) LF = landfills
SI = surface impoundments
Figure 4-2. Comparison of CCW leachate data with other leachate data in U.S. EPA (2006c).
For the 2006 Vanderbilt leaching study report, data are provided for each ash tested, with the minimum, maximum, and value at natural pH plotted on the chart. Percentile values (95th, 50th, 5th) are plotted for the compiled data sets (EPA, EPRI, and CCW), and mercury was not modeled for landfills because of a high number of nondetects.
For arsenic, the CCW values bracket the range found in the other studies. Selenium values also agree fairly well for CCW landfill data, although the CCW landfill values appear to be lower than some of the values from the other studies, suggesting that selenium risks may have been somewhat underestimated for landfills in this analysis. This is significant even though selenium risks from landfills were not above an HQ of 1 in this analysis, because selenium is often reported as a constituent of concern (along with arsenic and boron) in CCW damage cases (U.S. EPA, 2000, 2003e; Lang and Schlictmann, 2004; Zillmer and Fauble, 2004).
U.S. EPA (2008c) extends the work in U.S. EPA (2006c) to include laboratory leaching studies of 23 CCWs sampled from 8 coal combustion power plants. Wastes tested included fly ash, scrubber sludges, and gypsum. All of the metals addressed in this risk assessment were measured in the laboratory leaching tests.
Similar to Figure 4-2 above, Figures 46–59 on pages 77–86 in U.S. EPA (2008c)
compare constituent concentration ranges in their laboratory CCW extracts to ranges reported by other CCW leachate data compilations, including the constituent data from this risk assessment. These graphs are not repeated here, but the conclusions are similar to the U.S. EPA (2006c) comparisons, in that the ranges of metals concentrations generally plot within the range reported for the laboratory tests, especially with fly ash and flue gas desulfurization sludges. For ash codisposed with coal refuse metal, concentrations tend to be an order of magnitude or more greater than the wastes studied in U.S. EPA (2008c), which did not include such codisposed wastes. Only two CCW metals plot largely outside the range for fly ash. Barium fly ash concentrations from the CCW risk assessment are an order of magnitude or more lower than those reported by U.S. EPA (2008c), and lead concentrations in the fly ash and FGD wastes modeled in this risk assessment are one to two orders of magnitude above those plotted in U.S. EPA (2008c). The latter may be an artifact of the predominance of TCLP measurements in the CCW constituent database, because the acetate buffer in the TCLP can be especially effective in complexing lead compounds into the extract solution. Finally, a few of the Vanderbilt
measurements for molybdenum and selenium are above the range modeled in the CCW risk assessment.
The fact that the 2006 and 2008 Vanderbilt results are in general agreement with the CCW arsenic and selenium levels does help allay concerns that the TCLP CCW leachate values used in the analysis markedly overestimate or underestimate the concentrations actual CCW leachate.
Mercury and Nondetect Analyses. For certain of the CCW constituents addressed in this analysis, the CCW leachate database contains a large number of nondetect measurements (concentrations below an analytical instrument’s ability to measure). Table 4-30 illustrates this point by showing, by WMU type and chemical, the overall percent of nondetect values for each
chemical and the percent of site-averaged values11 that are composed entirely of nondetect measurements. Although some constituents have a large number of nondetect values, many of those could still be modeled (substituting half the detection limit for nondetect values). Where there are detections for a chemical, the specific substitute value used for nondetect values does not affect the upper percentile risks, because the upper percentile risks are associated with the higher, detectable source concentrations in the distribution rather than the lower source concentrations associated with nondetect values. Values for nondetects will be in the lower percentiles whether they are half the detection limit or some other value.
Table 4-30. Proportion of Nondetect Analyses for Modeled CCW Constituents
Chemicala
Measurements Sites
Number % nondetects Number % with all nondetects
Landfills Aluminum 397 18% 61 5% Antimony 496 50% 66 41% Arsenic 1,182 49% 128 20% Barium 1,225 11% 126 5% Boron 930 8% 83 2% Cadmium 1,237 50% 124 31% Cobalt 559 56% 52 19% Lead 1,109 60% 125 30% Mercury 974 91% 101 58% Molybdenum 373 24% 58 10% Nitrate/Nitrite 141 48% 20 15% Selenium 1,227 49% 131 17% Thallium 402 60% 40 45% Surface Impoundments Aluminum 158 10% 16 6% Antimony 11 100% 2 100% Arsenic 155 16% 16 6% Barium 161 14% 16 13% Boron 164 7% 171 6% Cadmium 164 68% 16 50% Cobalt 49 59% 4 50% Lead 138 78% 14 36% Mercury 1 100% 1 100% Molybdenum 161 37% 17 24% Nitrate/Nitrite 267 59% 14 7% Selenium 140 33% 15 20% (continued)
Proportion of Nondetect Analyses for Modeled CCW Constituents (continued)
Chemicala
Measurements Sites
Number % nondetects Number % with all nondetects
Thallium 11 100% 2 100%
a Results for constituents shown in bold italics were not presented in this report because of high detection limits or limited data.
Constituents that could not be addressed in this analysis because of a very high number of nondetects (i.e., more than 90 percent of measurements) included mercury (for landfills and surface impoundments) and thallium and antimony (for surface impoundments only). Mercury is of particular interest because it is the only constituent with significant concern through the fish consumption pathway, and because there is the potential for mercury concentrations in CCW to increase as flue gas mercury controls are installed on coal-fired power plants in response to the Clean Air Interstate Rule (CAIR) and the Clean Air Mercury Rule (CAMR). However, analysis of the effect of mercury emission controls was outside the scope of the risk assessment, which was to evaluate current waste disposal conditions, not potential future changes due to emission controls.
Recent work by Vanderbilt University (U.S. EPA, 2006c, 2008c) sheds some light on mercury concentrations in leachate from some CCWs. Figure 4-3 plots the CCW distribution of mercury concentrations (assuming half the detection limit for mercury values below detection) against results from the Vanderbilt work and recent data collected by EPRI (from U.S. EPA, 2006c; results are similar in U.S. EPA, 2008c). Assuming half the detection limit, the CCW mercury leachate values are about an order of magnitude or more higher than the Vanderbilt or EPRI data. With a single CCW leachate analysis available for surface impoundments, it is difficult to draw firm conclusions, but the concentration value is above the maximum value shown in the other studies. In short, the mercury levels in the CCW database are not useful because of high detection limits. In addition, the Vanderbilt study found that older mercury analyses, such as the ones in the CCW database, could be biased high because of cross- contamination issues.
Finally, U.S. EPA (2006c) and preliminary results of ongoing EPA studies (e.g., U.S. EPA, 2008c) suggest that both mercury levels and mercury leachability in CCW can vary depending on the flue gas mercury controls used at a power plant.
Mercury 0.0000001 0.000001 0.00001 0.0001 0.001 0.01 0.1 Brayton Point Pleasant Prairie Salem Harbor Facility C St. Clair Facility Facility L CCW - LF EPRI - LF CCW - SI EPRI - SI
Vanderbilt Study Other Data Sets
Hg ( m g /L ) max Natural pH min 95th % 50th % 5th %
Key to data sets:
Vanderbilt = U.S. EPA (2006c)
CCW = CC Constituent Database (this analysis) EPRI = EPRI Leachate data (from U.S. EPA, 2006c)
EPA = Leach 2000 data (from U.S. EPA, 2000, as cited in U.S. EPA, 2006c)
LF = landfills
SI = surface impoundments
Figure 4-3. CCW mercury concentrations compared with other leachate data. 4.4.3.2 WMU Locations and Characteristics
The locations of the specific sites in the United States where CCW is disposed are known, and EPA used the soil and climatic characteristics of these sites in the Monte Carlo analysis. Because most locations were facility front gates or centroids, the exact location of the CCW landfill or surface impoundment was not known. To account for this uncertainty, soil data were collected for an area around the plant and soil type distributions were sampled in the Monte Carlo analysis. Climate center assignments were combined with the soil texture distributions to select infiltration and recharge rates to use in the analysis.
WMU area, depth, volume, and liner type were not varied in the Monte Carlo analysis because values for these variables were known from the EPRI survey data. More uncertain parameters, such as depth below grade, were varied within reasonable ranges. These data were used in the source model calculations to generate the distribution of environmental releases used by the fate and transport modeling.
Three standard WMU liner scenarios (clay, composite, and unlined) were assigned to each facility based on best matches to data in the EPRI survey on liner type. Infiltration through these liners was then modeled using assumptions, models, and data developed in support of EPA’s Industrial Subtitle D guidance. How well these assumptions and models represent the performance of CCW WMU landfills and surface impoundments is an uncertainty in this analysis.
With respect to the clay liners, the 2009 risk assessment used the assumption that compact clay liners were designed to have a hydraulic conductivity of 1×10-7 cm/sec. This is consistent with EPA’s Industrial D Guidance, which states that “clay liners should be at least 2 feet thick and have a maximum hydraulic conductivity of 1×10-7 cm/sec”(U.S. EPA, 2006d). However, clay liners designed to meet a 1×10-7 cm/sec hydraulic conductivity could perform differently in practice. In one liner study (Moo-Young et al., 2004), a small set of clay-lined landfills were found to have field hydraulic conductivities ranging from 2×10-9 to
4.4×10-8 cm/sec and a small set of surface impoundments were found to have field hydraulic conductivities ranging from 3×10-6 to 3.2×10-5 cm/sec. Thus, the assumption of clay liners performing at 1×10-7 cm/sec could lead to an under- or over-estimate of actual risks.
Composite liners would also not be expected to perform consistently over 10,000 years as was assumed in the model. Instead, the liner would eventually perform at the level of the clay layer once the synthetic layer had deteriorated. This simplification is likely to lead to an underestimate of composite liner risks.