NC0001422_Final Duke Letter regarding Geochemical Modeling_20160819

Water Resources
ENVIRONMENTAL QUALITY
Mr. Harry Sideris
Senior Vice President
Environment, Health, and Safety
Duke Energy
526 South Church Street
Mail Code EC3XP
Charlotte, NC 28202
Subject:
PAT MCCRORY
Governor
DONALD R. VAN DER VAART
Secretary
S. JAY ZIMMERMAN
Director
August 19, 2016
Sutton Energy Complex
NPDES Permit No. NC0001422 New Hanover County, North Carolina
Comments on Geochemical Modeling (Corrective Action Plan Part 1, Appendix D,
and Corrective Action Plan Part 2, Appendix C)
Dear Mr. Sideris:
The Division of Water Resources (Division) received the Sutton Energy Complex Corrective Action
Plan (CAP) Part 1 on November 3, 2015 and the CAP Part 2 on February 1, 2016, both prepared by
SynTerra. These plans contain geochemical model reports (as Appendices) in support of the CAP.
The geochemical model reports submitted in the CAP Parts 1 and 2 are deficient for purposes of the
Coal Ash Management Act (CAMA) and/or 15A NCAC 02L.0106.
1. The CAP should provide a geochemical conceptual model and numerical model designed
specifically for the groundwater system at the Sutton facility.
2. These models should describe and document the current geochemical conditions (distribution of
observed pH, Eh, sorptive host content, etc.) across the Sutton facility and how those conditions
help explain the observed Contaminants of Interest (COI) concentration distribution.
State of North Carolina I Environmental Quality J Water Resources
1611 Mad service Center I Raleigh, North Carolina 27699-1611
919 707 9000

3. The conceptual model should also explain how geochemical conditions are expected to change
after the proposed corrective action has been implemented and how that is expected to effect the
COI distribution over time.
4. Finally, the CAP should provide predictive modeling, using 1-D PHREEQC transport code or
equivalent, of COI concentrations simulated along flow path transects in areas of the site where
groundwater exceeds 15A NCAC 02L groundwater standards at the compliance boundary. This
modeling should be done for current conditions and for conditions expected after corrective
action has been implemented.
The attached comments specify additional data collection, conceptualization, modeling, and
documentation needed in order for the Division to adequately review and subsequently approve a
CAMA and 15A NCAC 02L compliant CAP.
Please provide the revisions and documentation requested herein (one hard copy and one electronic
copy each) to WQROS Wilmington Regional Office and WQROS Central Office no later than
November 1, 2016. Additionally, if the same approach in the Geochemistry Model submitted for
Sutton Energy Complex was used for other SynTerra Geochemistry Models submitted (i.e.
Weatherspoon, H.F. Lee, etc.), please revise these documents to take into account these comments and
resubmit in like fashion (one hard copy and one electronic copy) to the appropriate Regional Office and
WQROS Central Office no later than November 1, 2016.
For questions concerning the Sutton facility, please contact Morella Sanchez -King at [phone].
For all other questions contact Steve Lanter at [phone].
Sincerely,
S. erman, P.G., Director
Division of Water Resources
cc: WQROS Regional Supervisors and Assistant Supervisors
WQROS Central File Copy
Attachment (1)

Attachment 1
General and Site -Specific Comments Sutton Geochemical Modeling
Attachment 1
General and Site -Specific Comments regarding the Geochemical Modeling
Presented in the SynTerra Corrective Action Plan Part 1, Appendix D, and
SynTerra Corrective Action Plan Part 2, Appendix C
General Comments
1. Provide a site -specific conceptual geochemical model for areas of the Sutton site associated
with contaminant concentrations above 15A NCAC 02L (2L). Modeling should describe, at the
local, monitor well scale, the geochemical conditions (and/or other factors) that explain the
horizontal and vertical distribution of observed contaminant concentrations along flowpath
transects upgradient of the ash source, within the ash source, and downgradient of the ash source.
As part of this effort, identify the dominant mechanism(s) that immobilize or otherwise reduce
concentrations of a given constituent along the transect (for example, adsorption/desorption and
mineral precipitation/dissolution processes) and discuss whether or not changes in pH, Eh,
unstable soil oxyhydroxides, or other conditions during or after the proposed corrective action
would be expected to result in higher or lower downgradient Contaminant of Interest (COI)
concentrations in the future and how quickly those effects would be expected to occur. The
descriptions should include tables of the specific well and boring data upon which the conceptual
model is based and any calculations (such as mineral saturation indices) that are made to develop
the site -specific model. This exercise should be done for each COI.
Separate localized conceptual modeling is needed to explain the occurrence of isolated elevated
COI concentrations at locations not along obvious flowpaths, such as selenium concentrations in
the area of MW-27B (north of the new ash basin) and arsenic concentrations in the area of MW-
21 C (south east of the old ash basin). The conceptual model should also explain why pore water
COI concentrations are lower than downgradient COI concentrations in certain areas.
If the generalized understandings gleaned from site -wide data/modeling are verified in specific
areas of the site (using actual monitor well pH/Eh/COI data) where particularly high
concentrations of a given COI are observed and/or where pH or Eh conditions vary substantially
from that of other areas of the facility, they may be used as the basis for CAP design. If the
understandings do not hold, efforts should be made to explain why and subsequently to correct or
Page 1of13

Attachment 1
General and Site -Specific Comments Sutton Geochemical Modeling
supplement them in those specific areas. Only those conceptual models using site data, to
accurately explain the causes and effects of COI concentration distributions and mobility in
individual areas of the site should be used as evidence in support of a proposed CAP remedy.
2. The localized conceptual models discussed in #1 above should be used, along with the
appropriate monitor well and boring data, to inform and develop 1-D PHREEQC transport
models along selected flowpath transects. These evaluations are needed in support of the
proposed CAP. Once the computer models are developed and their accuracy verified (by
comparing observed to simulated dissolved concentrations in wells along the transect), the
models should be used to predict future concentrations based on changes to the geochemical
setting (pH, Eh, etc) that are expected to occur as a result of the proposed corrective actions.
Sensitivity analyses may be used as needed to bracket the plausible downgradient concentrations
expected in the future; this is particularly important if there is only limited confidence in the
models, input data, or assumed future conditions.
3. In some cases a single value of Kd is used to represent sorption along very long flow paths.
However, measured and modeled Kds are shown in the reports to have some very large ranges
(up to orders of magnitude). Sensitivity analyses should be conducted and presented to
demonstrate the effects of Kd on predicted COI concentrations.
4. The primary contaminant attenuation mechanism at Sutton is adsorption onto Fe and Al
hydroxides present in the aquifer solid phase. As a consequence, the stabilities of these solids
under site conditions are important because they affect the available concentration of adsorbent.
The solubility of Fe hydroxide is strongly pH and Eh dependent, while the solubility of Al
hydroxide is pH dependent over the ranges considered for the site geochemical model. To this
end, because the sorption model is built upon the presence of HFO in the aquifer, it would be
beneficial to discuss the stability of this solid phase under the measured pH/Eh conditions found
in the aquifer(s). What is the saturation index of HFO (or ferrihydrite) for the range of water
chemistries found across the site? A discussion should be included in the report on Al and Fe
adsorbent stabilities, perhaps with accompanying Pourbaix diagrams for these elements. The
effects of various remediation methods on the solubilities of these minerals should also be
addressed. See Section 4 of the CAP Part 1 Appendix D for context.
Page 2 of 13

Attachment 1
General and Site -Specific Comments Sutton Geochemical Modeling
This evaluation of adsorbent stability is consistent with the following statement from the CAP
Part 2 Appendix C Introduction, last sentence: "Therefore, the primary emphasis was to quanta
how changes in the system conditions will alter the speciation and mobility of each constituent
(particularly changes in pH and EH). This will allow us to determine if changes occurring
during remediation could mobilize any particular constituent. " A critical component in this
process is to calculate the solubilities of the adsorbents as a result of changing pH and Eh, which
has not been discussed in the report and apparently was not part of the PHREEQC model
calculations (i.e., the Fe and Al adsorbents were not specified as Equilibrium Phases in the
PHREEQC simulations).
5. The following comment pertains to any additional sampling and analysis of solid samples that
may be needed to augment the CSA, CAP, or geochemical modeling. To extend the results of
the lab -derived Kd terms to other soils and geochemical conditions, soil samples were collected
and analyzed for hydrous ferric oxide (FIFO), which is believed to be the dominant adsorbent of
COPCs. Soil HFO is typically measured using a selective, sequential extraction technique that
dissolves the FIFO so that its concentration can be quantified. Because dissolving the HFO also
releases to solution the species adsorbed to the solid, it is recommended that those adsorbed
species also be measured during the HFO determination. This additional data can then be used
to calculate adsorbed concentrations on the HFO, which can be compared to the Kd-derived
concentrations to provide assurance that there is relationship between Kd and HFO
concentration. Also, the measured adsorbed concentrations can be used to validate computer
calculations of surface complexation models developed to simulate the adsorption process.
6. A range of Kd values for each contaminant is calculated by PHREEQC based on the
variable input parameters pH, Eh, and major ion concentrations (See CAP Part 2
Appendix C, Tables 3.1, 4.1, 5.2, and 6.1). For each contaminant, the Kd range covers
several orders of magnitude. It appears that for the transport model calculations only a
single Kd values was used for each contaminant at each site (Tables 3.1, 4.1, 5.2, and
6.1). It is very likely that the Kd value is not constant along the flowpath away from the
source to the leading edge of each plume. Why is the constant Kd transport modeling
approach appropriate for simulating the current conditions and possible future conditions
after remediation? Choosing a low Kd for transport modeling may be conservative from
Page 3 of 13

Attachment 1
General and Site -Specific Comments Sutton Geochemical Modeling
the standpoint of plume extent, but it is not at all conservative in terms of the solid phase
adsorbed contaminant concentration that will provide a long-term source of
contamination to groundwater following remediation. This source of contamination in
the solid phase of the aquifer will extend the amount of time necessary for the
groundwater to reach a target cleanup level.
7. Consideration should be given to the potential importance of groundwater residence time and
duration of mineral weathering in evaluating groundwater concentration levels. For example,
chemical weathering reactions generally consume H' thereby increasing groundwater pH. Also,
chloride concentrations increase in groundwater as a result of leaching from minerals, resulting
in higher concentrations correlated with groundwater age. The overall result is that groundwater
pH and chloride concentrations increase along a groundwater flowpath or between a shallow and
deep aquifer. See the CAP Part 1, Appendix D, Section 3 for context.
8. Molecular oxygen is rarely the dominant redox buffer in groundwater systems because of
slow reaction rates. Most likely, the redox buffers that have the greatest influence on Eh are Fe
or Mn. See CAP Part 1, Appendix D, Section 3 for context.
9. Discuss whether or not an anoxic environment was maintained during sample transport and
lab Kd testing (glove box) for solid samples collected from locations characterized by low
dissolved oxygen. Discuss whether or not the model results are sensitive to the chosen sample
collection and testing methodology. If they are sensitive to the collection and testing methods,
how does this affect the conclusions that have been drawn about COI mobility and model
predictions?
10. Please provide date and page numbers in the updated report.
Site -Specific Comments
1. CAP Part 1, Appendix D. The Executive Summary states "The capacity of the aquifer solids
to sequester the constituents of interest was estimated by assuming the aquifer solids contained
0.05 moles of sorption sites per mole of extractable iron. The number of moles of several
constituents of interest in the pore fluid was estimated assuming all constituents were present at
the NC2L standard levels. Assuming 100% sorption of the summation of the total moles of all
constituents, less than 1% of the total available sorption sites was occupied. Therefore, it appears
Page 4 of 13

Attachment 1
General and Site -Specific Comments Sutton Geochemical Modeling
the aquifer solids have sufficient sorption capacity for high concentrations of all constituents
though the actual sorbed concentrations will vary based on the sorption affinity (i. e. distribution
coefficient) of individual constituents".
This conclusion appears to erroneously neglect the fact that flowing groundwater has contributed
this amount of moles of constituents (mass flux) to available sorption sites for many decades and
will continue to contribute more over time. In other words, the sorption sites have been filling
for a very long time and will continue to do so. Only by accounting for the total mass of
constituents contained in the ash can an appropriate conclusion about the sequestration capacity
of the aquifer be drawn.
Also, sorption sites may also be filled with constituents that are not COIs (major ions, etc). For
all future solid sample extraction tests to quantify adsorption capacity, the concentrations of ALL
COIs in the extract should be measured along with the Fe or Al adsorbent concentrations. This
will allow the actual data to confirm and corroborate the modeled or lab -observed sorptions for
the various COCs from the collected solid samples.
2. CAP Part 1, Appendix D, Section 2, last paragraph states "In these Pourbaix diagrams, the Ex
and pH measurements from the Sutton site are shown as individual datapoints. A generic
groundwater chemistry containing 500 ppb of each constituent of concern was used in the
simulations ('fable 2.1). These concentrations are generally higher than the concentrations
observed in Sutton groundwater samples." This statement appears to be inaccurate, at least for
concentrations of boron, iron, and manganese. Actual maximum observed groundwater
concentrations should be used for these COIs.
3. CAP Part 1, Appendix D: Section 3, last paragraph. The report should provide additional
information on the filtration testing. Was filtering done in the field or in the lab? Were the
filtrates preserved with acid prior to analysis? How did the results of the 0.1 u filters compare to
those of the 0.45 u filters? What does this say about colloidal transport at the site? In looking at
filtered pore water results versus filtered down or side gradient groundwater results, are the
colloids associated with ash? Were filtered (dissolved) sample results used exclusively in the
PHREEQC model?
Page 5of13

Attachment 1
General and Site -Specific Comments Sutton Geochemical Modeling
4. CAP Part 1, Appendix D: Section 4, second to last paragraph, last sentence states: "This
assumed value is similar to value of 0.11 molsites/molwo (0.2 mo6t./mo1Fe) used by Dzombak
and Morel in a global model of ion sorption to HFO (discussed in section 6 below)". If all of the
sites are on HFO and HFO contains only one Fe in its formula, then the site densities on a mole
scale should be the same (0.20 molsites/molHuo)• The assumed fraction of extractable Fe that is
available for sorption sites should be 0.20 molsites/moliuo in the equation.
5. CAP Part 1, Appendix D: Table 5.1 should include all COIs.
6. CAP Part 1, Appendix D: Section 4, last paragraph, the conversion from pe to Eh is incorrect.
The correct relationship is Eh(volts) = 0.059 x pe or Eh(mV) = 59 x pe (Appelo CAJ and D
Postma. 1993. Geochemistry, groundwater, and pollution. page 246). A pe of -5 corresponds to
an Eh of -296 mV and a pe of 15 corresponds to an Eh of 888 mV.
7. CAP Part 1, Appendix D: Section 5. Assuming that the adsorption calculations were all run
over a pe range of -5 to 15, the Eh values in all the figures in this section are incorrect. The Eh
range should be -296 mV to 888 mV,
8. CAP Part 1, Appendix D: Section 6, 2"a paragraph. The relationship between pe and Eh is
given as pe = Eh x 59 mv. As discussed above, the correct relationship is pe = Eh(mV)/59mV.
The incorrect relationship is used throughout the geochemical modeling report, which means that
the conversion of the site -measured Eh values to pe for Table 6.1 and the use of that pe in
PHREEQC to calculate the modeled fractions of As(III) and As(V) in the table lead to incorrect
results. Figure 6.1 derived from the data in Table 6.1 also requires correction.
9. CAP Part 1, Appendix D, Section 6, Figure 6.2. The x-axis on this figure is pH + pe.
Because the pe was not calculated correctly from the measured Eh for these water samples, the
locations of the samples along the x-axis will change significantly. For example, the first water
sample has a pH of 7.9 and an Eh of -170 mV (pe =-2.87); therefore, the pH + pe = 7.9 + -2.87 =
5.03. However, the data are plotted at a pH + pe value of -2.10 because the incorrect conversion
of Eh resulted in a pe of -10.
10. CAP Part 1, Appendix D, Section 6. A counterpart table to Figure 6.2 should be provided
that includes the numerical percent differences between modeled values ofCOI species
concentrations and those that are observed in various locations across the site, including areas
Page 6 of 13

Attachment 1
General and Site -Specific Comments Sutton Geochemical Modeling
with COIs above 2L. This should be done for each COI. Expectations are that boron and
chloride concentrations should match to within about 20% and arsenic, selenium, and thallium
within about 50%.
11. CAP Part 1, Appendix D, Section 7, 1' paragraph. At the end of this paragraph, the gibbsite
adsorption site density is given as 0.41 molsnes/molAi. However, using the formula in this
paragraph to calculate this value with the provided data results in a site density of 0.033
molsicedmolAi. This lower value for gibbsite is more consistent with the HFO value of 0.2
molsaes/mo1Fe and with HFO's higher surface area of 600 m2/g (Dzombak and Morel 1990)
compared to the gibbsite surface area of 32 m2/g. If the correct gibbsite adsorption site density is
0.033 molsices/molAi, then the gibbsite adsorption capacities given in Table 7.1 will all decrease
by about a factor of ten, bringing them close to the HFO capacities. Table 7.2 for gibbsite will
also have to be revised.
12. CAP Part 1, Appendix D, Section 8. Summary.
• 2"d Bullet. The assumption that 5% of the extractable iron content is
available for adsorption and that this modeling assumption predicts Kd
values within an order of magnitude needs to be revisited after using
correct pe values for the Eh measurements.
• Yd Bullet. Arsenic speciation modeling and speciation modeling of other
COIs needs to be reviewed to confirm that correct pe values were used.
• Provide a summary of the conceptual site geochemical model for Sutton
developed from site data and the PHREEQC calculations.
13. CAP Part 2, Appendix C, Page 7, last paragraph. "The variability of the pH and EH
conditions at each site will essentially be "noise" considering the wide range of Kd values
predicted as a function of pH which are discussed below. Therefore, one "global" model which
shows the influence of Kd as a function of pH and EH within the selected range is appropriate
for all sites." How does a "global" site model deal with `outliers" that have been observed at the
sites. For example, there is localized elevated arsenic in groundwater in the area of monitoring
well MW-21 and elevated selenium in the area of monitoring well MW-27. The geochemical
Page 7 of 13

Attachment 1
General and Site -Specific Comments Sutton Geochemical Modeling
conditions that have produced these outliers need to be understood so that responses to
remediation can be estimated at these locations. Details in item 14.
14. CAP 2 Part 2- Site conditions described below seems not to be adequately simulated by the
geochemical model and the Global Model Input parameters listed in Table 2.3. Groundwater
data at the Sutton site reflect that there are three distinctive zones: North, Transition N-S, and
South zones.
North Zone (north of New Ash Basin)
Site Conditions: The "north zone" reflects low pH values, ranging from 4.5 to 5.6 and Eh
values ranging from 425 to 505 mv. The zone has been characterized by relatively low iron
concentrations (when compared to the rest of the site), high manganese concentrations, and one
isolated area where selenium has been consistently above the 2L standards (MW-27B). The
model provided did not include selenium as a contaminant of concern. Understanding of
selenium mobilization is required.
Input parameters in Table 2.3: The closest representation to site conditions would be: (1)
pH=4; Eh=482 (2) pH=5.1; Eh=372; and (3) pH=5.6; Eh= -21. Being pH a logarithmic
function, additional modeling ("with more pH resolution") would be required to represent the
site.
Transition N-S Zone
Site Conditions: Groundwater data from the "transition zone" have reported pH values
generally > 5.5 and Eh values from 350 to 460 mv. Notice that pH values in the same well at
different screened depths show marked differences in pH (i.e. MW-24 B and MW-24C). The
zone has reported high concentrations of boron, iron, and manganese. Thallium above the
IMAC value has been observed in well MW-24B and Cobalt above the IMAC value has been
reported in MW-24C.
Input parameters in Table 2.3: The closest representation to site conditions would be: pH=5.1;
Eh=372. Being pH a logarithmic function, small differences in pH would have a large impact
on the predicted Kd values. Additional modeling ("with more pH resolution") would be
required to properly represent the site. Moreover, differences in geochemical behavior can
vary at the same location at different depths. Include geochemical modeling of thallium and
cobalt as contaminant of concern. Understanding of thallium and cobalt mobilization and
potential attenuation mechanisms is required.
South Zone (South, South-east of Old Ash Basin)
Site Conditions: Groundwater data from the "south zone" have reported pH values generally >
6.3 and Eh values in the range of 350-400 my and around of 150 my (i.e. MW-21C). The
zone has reported high concentrations of boron, iron, and manganese. There is an area with
consistent high arsenic concentrations (i.e. MW-21 Q. Thallium and cobalt above IMAC
values have also been reported. Include geochemical modeling to understand the observed site
conditions.
Page 8 of 13

Attachment 1
General and Site -Specific Comments Sutton Geochemical Modeling
15. The PHREEQC modeling code used to make the speciation calculations and develop
the geochemical model requires that the pe of the system be used to input the redox
potential of the groundwater. Eh and pe are related by the following equation from
Stumm and Morgan (Aquatic Chemistry 3`d Edition, 1996, p. 444):
Eh(volts) = 2. F T pe
Using appropriate values for R (1.987 x 10-3 kcal/deg-mol), T (298 deg), and F (23.06
kcal/volt) results in the equation Eh(volts) = 0.059 volts x pe, which upon re -arranging
and with Eh in millivolts (mv) becomes pe = Eh(mv) / 59 mv. The Eh to pe conversion
used for the geochemical modeling calculations was, however, pe = Eh (V) x 59 mV,
note dimensional inconsistency. This was stated in an e-mail (subject: L.V. Sutton
Energy Complex - Corrective Action Plan- Part 1 (Appendix D) - Geochemistry Model:
Verify pe-Eh relationship) on *February 9, 2016* from Morella Sanchez King, notifying
Mr. Ed Sullivan of the error and implications for the model results. This results in a
calculation error that increases the actual pe by a factor of 3.5. Using the incorrect pe in
PHREEQC results in redox speciation calculations that do not represent the geochemical
system being modeled and in conclusions as to redox speciation, mineral equilibrium, and
adsorption of the redox-sensitive species that are not accurate. A geochemical model of
the aquifer system based on these calculations is not representative of actual site
conditions. This comment applies to the Sutton CAP Appendix D and to the Sutton
CAP2 Appendix C, but may also apply to the models developed for other Duke sites.
16. CAP Part 2, Appendix C: Page i and Table 2.2. The site density of the Al adsorbent is
given as 0.4 moles of Al sites per mole of solid phase Al and this value is used in Table 2.2 to
calculate the Al adsorbent concentration (1.16 E-05 molsites/gsoiid). The site density in
molsites/molAv can be calculated from values of site density (8 sites/nm2) and aluminum hydroxide
surface area of 32 mZ/g provided in Karamalidis and Dzombak (2010, Surface Complexation
Modeling — Gibbsite). The calculated site density from these values for the Al adsorbent is 0.033
molsites/molAv, not the 0.4 molsites/molm used to calculate the adsorbent concentration used in the
Page 9 of 13

Attachment 1
General and Site -Specific Comments Sutton Geochemical Modeling
PHREEQC modeling. Using the correct site density, the Al adsorbent concentration should be
9.34 E-07 molsites/gsoGd, which is a factor of 12 less than that used for the Al adsorption modeling.
This modeling should be redone with the correct Al adsorbent concentration.
Page 17. CAP Part 2, Appendix C: Page ii, Arsenic bullet, second sentence states: "l )
increased sorption of As(V) relative to As(III) which would remove all As(V) from the
groundwater and prevent As(V) measurements in samples". Sorption cannot remove all the
As(V) from groundwater. There will always be some dissolved in groundwater in equilibrium
with the adsorbed concentration. Sorption might remove As(V) to below the analytical detection
limit.
18. CAP Part 2, Appendix C: Page ii, Arsenic bullet, middle of paragraph. "...the minerals
scorodite (FeAsO4.2H2O) and mansfieldite (AIASO4.2H2O) are near saturation under some pH
and EH conditions examined in this model and measured in the field." Figure 3.8 shows that
scorodite is more than 100 times undersaturated (Saturation Index < -2) under the range of pH
and Eh conditions considered. Mansfeldite is even further from saturation under these
conditions. It is highly doubtful that they could form and limit dissolved arsenic in the aquifer.
19. CAP Part 2, Appendix C: Page 2, first paragraph. Why were both the WATEQ417 and
MINTEQ 0 databases used? Why not use only the more comprehensive MINTEQ v4 database,
which has all the necessary thermodynamic data?
20. CAP Part 2, Appendix C: Page 3, last paragraph. Why is the concentration of the solid
phase assumed to be 50 g/L? hi a typical aquifer, the concentration is on the order of 8,000 g-
solid/L-gw (based on a bulk density of 2 kg/L and a porosity of 0.25).
21. CAP Part 2, Appendix C: Page 5, first paragraph, third sentence states", modeling
groundwater concentrations of each constituent of interest at the 2L Standard Level and
conservatively assuming 100% sorption, the capacity of the solid phases to sorb the constituents
of interest was determined. In all cases, less than I % of the total sorption capacity of the solid
phases was occupied by the constituents of interest [28-34]." This assumes that adsorption only
occurs between the solid phase and constituents in one pore volume of groundwater. Flow of
groundwater replaces that pore volume of water with fresh dissolved constituents that have
adsorbed since contaminants began entering the aquifer decades ago. The method of calculating
Page 10 of 13

Attachment 1
General and Site -Specific Comments Sutton Geochemical Modeling
adsorption capacity used in the report is not at all conservative from the standpoint of estimating
consumption of the adsorption capacity by contaminants in groundwater over the past few
decades.
22. CAP Part 2, Appendix C: Page 23, first paragraph. This discussion concerns major ions
competing with arsenic for adsorption sites thereby lowering arsenic adsorption; however,
arsenic adsorption is not shown on Figures 3.6 and 3.7 called out in the paragraph. If the
concentration is too low to plot, at least provide the arsenic adsorption concentrations under the
different scenarios.
23. CAP Part 2, Appendix C: Page 23, middle of last paragraph. "...a saturation index of 1 or
greater indicates that the solution is saturated with respect to that ion and will precipitate." A
saturation index of 0 or greater indicates saturation. The conclusions in this paragraph based on
the saturation index of 1 should be revised.
24. CAP Part 2, Appendix C: Page 23, middle of last paragraph. The arsenic mineral scorodite
discussed in this paragraph as a possible solid phase limiting dissolved arsenic concentration
only forms under acidic, highly oxidizing conditions and would not form in the site
environments.
25. CAP Part 2, Appendix C: Page 25, Figure 3.7. The HAO adsorption calculations need to
be redone using the correct adsorbent concentration as discussed in the first Specific Comment
above.
26. CAP Part 2, Appendix C: Page 26, Section 3.3. Because of the discrepancy between
measured and modeled arsenic redox species, would it be more realistic to run PHREEQC using
the measured As(III) and As(V) concentrations and not couple arsenic redox to the pe entered for
the run? This method may more closely simulate site conditions in which dissolved arsenic in
the groundwater appears to be primarily As(III).
27. CAP Part 2, Appendix C: Page 28, second paragraph. "The calculated and measured EH
values are shown in Figure 3.10. From these data, it is clear that the expected EH values based
on the Fe redox couple are higher than the measured values." Because measured Eh values are
primarily lower than calculated Eh by 100 to 200 mV, confirm that field measured ORPs were
Page 11 of 13

Attachment 1
General and Site -Specific Comments Sutton Geochemical Modeling
correctly converted to Eh values. The conversion factor typically requires adding about 200 mV
to the field -measured ORP, and may bring the two values into closer alignment.
28. CAP Part 2, Appendix C: Page 29, third paragraph, first sentence. Briefly describe how the
Kds used in reactive transport modeling were derived if they did not come from lab
measurements or PHREEQC modeling of adsorption.
29. CAP Part 2, Appendix C: Page 30, last paragraph. "In the models of the average and
maximum major ion concentrations from Table 2.5, competition for (boron) sorption sites by
other major ions results in a decrease in the observed Kd values." What specific ions compete
for the boron adsorption sites? There are dramatic decreases in calculated boron Kds with
changing groundwater compositions suggesting that the boron Kd may change along
groundwater flowpaths away from the ash landfills. It appears from Table 4.1 that in most cases
a constant Kd was used for transport modeling across entire sites. How is this representative of
site conditions?
30. CAP Part 2, Appendix C: Page 31, second paragraph. "Sorption of boron to aluminum
hydroxides was predicted to be significantly higher than iron oxides as shown in Figure 4.4."
Revisit this conclusion after revising the HAO site concentrations discussed above in the first
Specific Comment.
31. CAP Part 2, Appendix C: Page 39, first paragraph. "Based on the groundwater
concentrations listed in Tables 2.4 and 2.5, only Fe2CrO4 is predicted to have a solubility product
greater than one (indicating precipitation is possible)." Text should read "...saturation index
greater than zero... ", instead of "solubility product greater than one."
32. CAP Part 2, Appendix C: Page 44, last paragraph. "However, rhodocrocite generally
occurs in hydrothermal systems and is unexpected to form under these site conditions."
Rhodochrosite is known to form under moderately reducing conditions caused by contamination
(e.g., beneath landfills) and may form under some of the site conditions.
33. CAP Part 2, Appendix C: Page 46, Arsenic Box. Reconsider checking the Chemical
Precipitation box because it is highly unlikely that any arsenic mineral will form in the site
Page 12 of 13

Attachment 1
General and Site -Specific Comments Sutton Geochemical Modeling
environments. The arsenic minerals are orders of magnitude undersaturated (Figure 3.8) in all
the pH/Eh conditions and should not be considered equilibrium phases in the model.
34. CAP Part 2, Appendix C: Page 47, Chromium Box. "This concentration range is
similar to what was modeled in PHREEQC and indicates that formation of mineral
phases containing Cr may occur under high pH conditions with relatively high Cr
concentrations." What chromium mineral phases might form? Why do they only occur
under high pH and high chromium concentrations? Amorphous Cr(OH)3 is known to
limit dissolved chromium to low concentrations over a wide range of typical groundwater
pH values.
Page 13 of 13