DSHW-2012-003591 - 0901a068802dd459

Division of
Solid and HazardousWaste
JUN 0 7 2012
^-OlQi -0036^2-
1 June 2012
8200-FY13-010
Mr. Scott T. Anderson, Executive Secretary
State of Utah Department of Environmental Quality
Division of Solid and Hazardous Waste
195 N.1950 W.
P.O. Box 144880
Sah Lake City, Utah 84114-4880
Re: ATK Launch Systems-Promontory EPA ID number UTD009081357
Response to Comments on the Preliminary Air Dispersion Modeling Assessment
Draft Report March 23, 2012
Dear Mr. Anderson:
Enclosed are responses to the comments provided by the UDSHW and TechLaw, Inc.
regarding the Preliminary Air Dispersion Modeling Assessment Draft Report for the
Open Bum and the Open Detonation Treatment Units at ATK Launch Systems
Promontory facility dated March 23, 2012. This information is necessary to conduct the
Human Health and Ecological Risk Assessments for our OB/OD operations.
Please contact me if you have any questions conceming this report. My telephone
number is [phone] or you can contact Blair Palmer at [phone].
Sincerely
^y^^if^Y^ 6^C7^eyt^
George E. Gooch, Manager
Environmental Compliance
cc: Jeff Vandel
ATK LAUNCH SYSTEMS RESPONSES TO TECH LAW COMMENTS REGARDING
REVIEW OF ATK LAUNCH SYSTEMS AIR DISPERSION MODELING REPORT FOR
USE IN THE HUMAN HEALTH AND ECOLOGICAL RISK ASSESSMENTS
JUNE, 2012
FACILITY EPA ID# UT009081357
GENERAL COMMENTS
1. ATK Launch Systems Revised Preliminary Air Dispersion Modeling Assessment Draft Report
for Open Burn and Open Detonation Treatment Units, Appendices D and E
In reviewing the OBODM input files for annual averages at Adams Ranch, two issues were noted that
impact the values that should be generated for comparison to annual NAAQS and use in the risk
assessment. First, ATK did not generate the annual average of peak concentrations for use in the
risk assessment. The files indicate that only the annual time average was generated. Utah DEQ
prefers the more conservative annual average of peak concentrations be used in the risk-based
analyses.
The second issue involves the specification of Concentration Averaging Time in OBODM's first
Control/Print Options menu. ATK has specified this parameter as 8760 hours, presumably in
reflection of the number of hours in a year. However, Appendix A, OBODM Menus, of Volume 1 of
the OBODM Users Guide states that the Time-Average Concentration represents a concentration
average over a user specified time interval that should be less than or equal to 3600 seconds (i.e., 1
hour). In fact, when ATK's annual average input files are loaded into OBODM, the value of 8760
hours for Concentration Averaging Time is flagged by a dialog box that states: "Warning, data value
8760.0 is questionable."
While the OBODM Users Guide recommends the value of this parameter to be less than or equal to
3600 seconds, setting its value to 600 seconds, or 10 minutes (the recommended value for the
measurement time for the standard deviation of the wind direction angle entered by the user in
OBODM's meteorological data menus), results in annual time-average concentrations similar to the
values obtained for annual average peak concentrations. This occurs because the measurement
time parameter is considered in estimating the value of dispersion coefficients. Thus, many air
dispersion modeling references recommend setting the value of the averaging time parameter equal
to or close to the value of the measurement time for the standard deviation of the wind direction
angle.
Please revise the air modeling files for annual averages to ensure the annual average modeling runs
generate annual average peak concentrations for use in subsequent risk-based analyses. The
currently modeled annual time-average concentrations are too low for this application and were
generated with a suspect parameter value. Further, the annual average air modeling should be
repeated using an appropriate value for the Concentration Averaging Time. As discussed during a
conference call on May 2, 2012, it is recommended that a value of 3600 seconds be entered for
Concentration Averaging Time in OBODM's first Control/Print Operations menus.
Response: The OBODM input files for annual averages have been updated to include peak
concentrations for use in risk-based analyses in addition to the time-averaging period being reduced
from 8760 hours to 3600 seconds per the conference call that was conducted on May 2, 2012. The
annual average concentrations presented in Tables 3-1 through 3-17 for Discrete Receptors will be
revised on the basis of these new model calculations. The General Grid Receptor summary tables
will also include average annual concentrations for comparison to NAAQS based on these new model
calculations. Separate tables will be prepared and included in Appendices D (Gas), E (Particulate),
and F (Deposition) of the Final Report that will present maximum peak annual concentrations and
deposition for each receptor for use in the risk-based analyses.
2. ATK Launch Systems Revised Preliminary Air Dispersion Modeling Assessment Draft Report
for Open Burn and Open Detonation Treatment Units, Section 3.0, Air Dispersion Modeling
Results, and Appendices D, E, G, H, J, and L
OBODM modeling was performed for utilized emissions so that a single modeling result can be
applied to all constituents emitted from the modeled treatment process (i.e., open burning, open
detonation). This is analogous to running the ISC model or AERMOD using a unit emission rate.
One of the OBODM source menus allows the user to input either the pounds of constituent emitted
and pounds of waste being treated OR the fraction of the plume consisting of the constituent being
modeled. Utilized results are obtained by specifying 1.0 as the fraction ofthe plume consisting ofthe
constituent being modeled. This means that the unitized modeling results will represent the total
mass of all constituents emitted during treatment per unit of plume volume (e.g., pgwaste stream
emissions/m^). Thus, to Obtain the concentration of a specific constituent, the modeling result must be
multiplied by the fraction of the total plume represented by the constituent of interest. This fraction is
represented by the constituent-specific emission factor. Thus, the specific air concentration of PMio
would be calculated as the product of the emission factor for PMio (EFRMIQ in micrograms of PMio per
micrograms of waste stream treated) and the unitized air concentration dispersion factor for
particulates (DFpartide in Mgwaste stream emissions/m^):
X"*"^" (^gpMlOemitted/in^) = EFpMio (^gpMlO emitted/) X DFpartide (/m^).
As discussed during a conference call on May 2, 2012, ATK used the following formula to calculate
the concentration of PMIO in air:
X*""" = EFpMlO (klgPMlO emitted/) X MaSS of Treated per hour (/hr) X DFpartide (Hgwaste stream emissions/m').
Thus, it appear that ATK's approach results in incorrect units for constituent-specific air concentration
(i.e.,( pgpMio emitted' kigwaste stream emissions/(hrm^)) and an Overestimate of its magnitude by a factor equal
to the assumed constituent emission rate.
Please review the calculations of all constituent-specific air concentrations contained in the air
modeling report to ensure that the correct value of air concentration has been compared to the
appropriate NAAQS and TSLs.
Response: It is agreed that the equation used in the draft report to compute air concentrations for
comparison to HAAQS and TSLs was incorrect. Constituent-specific air concentrations (Xi) for
Discrete and General Grid Receptors have been recalculated using the product of the emission factor
(EF|) for each specific constituent (jjg emitted//|jg treated) and the unitized air concentration
dispersion factor (DF,) (pg/m^):
Xi (|igem,tted/m^) = EF, (HgpMlOemitted/) X DP, (/m^)
where i = the specific constituent
Note: In the case of short term concentration averages greater than 1-hour (3-hour, 8-hour, 24-hour,
and 3-month, but not annual), the calculation will also include the appropriate screening conversion
factor.
3. ATK Launch Systems Revised Preliminary Air Dispersion Modeling Assessment Draft Report
for Open Burn and Open Detonation Treatment Units, Tables 3-1 through 3-17
Tables 3-1 through 3-17 define 1 hour CO NAAQS as 10,000 |jg/m^ and 8-hour CO NAAQS as
40,000 pg/m^. These are reversed; 40 CRF 50.8 defines the 1-hour CO NAAQS as "35 parts per
million (40 milligrams per cubic meter) for a 1-hour average" and the 8 hour NAAQS as "9 parts per
million (10 milligrams per cubic meter) for an 8-hour average concentration." Some or all of the 1-
hour CO NAAQS exceedances identified in the tables and discussed in Section 3.2.1 probably are
not predicted exceedances, but, they may have 8-hour predicted exceedances they missed because
they used a level 4 times too high. ATK performed comparisons against the incorrect values and
noted the results in the text when a constituent-specific air concentration exceeded its NAAQS. It will
be necessary to revise Tables 3-1 through 3-17 to list the correct value for each CO standard, and
repeat the comparisons. Any exceedances will need to be noted in Section 3.2.1, Compliance with
NAAQS, of the text. Ensure the CO air concentrations compared against the standards have been
calculated correctly. A technical review comment on this issue will be drafted and sent to ATK.
Response: The 1-hour CO NAAQS and 8-hour CO NAAQS will be swapped and air concentrations
will be compared to the correct standard. Text will be revised per these changes to determine
compliance with the applicable CO NAAQS.
4. ATK Launch Systems Revised Preliminary Air Dispersion Modeling Assessment Draft Report
for Open Burn and Open Detonation Treatment Units, Appendix D, Gas Modeling Input and
Output Files, and Appendix E, Particle Modeling Input and Output Files
Appendices D and E contain most of the OBODM modeling files needed to verify the modeling
methodology for gas and particle phase modeling, respectively. However, the hourly source strength
files used in modeling annual average impacts have not been provided. Please provide these files as
part of Appendices D and E.
Response: Appendices D and E have been updated to provide the hourly source strength files and
will be included in the Final Report.
SPECIFIC COMMENTS
5. Introduction, Page 1-2
The last paragraph on page 1-2 states that "ATK typically treats three different waste type packages;
one hundred percent pure propellant, 85 percent pure propellant with 15 percent waste materials,
and 65 percent pure propellant with 35 percent waste materials." Please revise the text to clarify
that most ofthe reactive waste that is treated by open burning is represented by the three pure
propellant/waste material mixtures that are cited in this paragraph and that the three waste-stream
classifications are based on an analysis of reactive wastes that have been generated at the facility in
the past.
Response: The last paragraph on Page 1-2 has been revised to read as follows:
ATK typically treats three different waste type packages; one hundred percent pure propellant; 85
percent pure propellant with 15 percent waste materials, and 65 percent pure propellant with 35
percent waste materials. Historically, most of the reactive waste that is treated by open burning is
represented by the three mixtures (one hundred percent pure propellant; 85 percent pure
propellant with 15 percent waste materials, and 65 percent pure propellant with 35 percent waste
materials.) These three waste-stream classifications are based on an analysis of reactive wastes that
have been generated at the facility in the past. In this modeling analysis, it was conservatively
assumed that the lower level of propellant (65 percent) and higher waste levels (35 percent) were
burned, which would overestimate the potential impacts because these trash burns would have
lower plume temperature and result in higher air dispersion factors, and this mixture would
potentially contain more ofthe by-products than 100 percent pure propellant.
6. Section 2.12, Receptor Networks, Page 2-14
The second paragraph of Section 2.12 indicates that Universal Transverse Mercator (UTM) northing,
easting, and terrain elevation data were obtained from United States Geological Survey (USGS)
Digital Elevation Maps (DEM). This source of information is not referenced or listed in the Reference
section of the Air Modeling Report. Revise the list of References to include a citation for the USGS
OEMs used in the air modeling analysis so stakeholders can locate this information. In addition,
revise Section 2.12 to include a reference to the citation.
Response: The reference for the Universal Transverse Mercator (UTM) northing, easting, and terrain
elevation data was obtained from U.S. Department of the Interior | U.S. Geological Survev URL:
http://www.usqs.qov/pubprod/maps.html. This reference will be added to Section 2.12 text and the
Reference section of the Final Report.
7. Section 2.13.1, Surface Data, Page 2-16
Section 2.13.1 states: "ATK operates the on-site monitoring station approximately 1.5 km southwest
of the M-225 treatment unit at an elevation of about 5,000 feet amsl (see Figure 2-1)." However,
Figure 2-1 appears to indicate about 700 meters between the two locations. Please review the
locations depicted on Figure 2-1, and revise the figure and/or text to describe and illustrate the
distance between M-225 and the on-site meteorological monitoring station (M-245) in an accurate
and consistent manner.
Response: The locations for M-225 and M-245 have been reviewed and the figure and text are now
consistent and accurately represent the location of the unit and meteorological station. A revised
version of Figure 2-1 will appear in the Final Report.
8. Section 2.14.1, Discrete Receptors, Page 2-19
Section 2.14.1 indicates "A database of the 250 worst case 1-hour events for each discrete receptor
and source is provided in Appendix B." A database containing the 250-wrost case 1-hour event was
not found in Appendix B. The hours producing the worst case events are contained in the 5 year
meteorological data files submitted as part of Appendix B on the CD of air modeling files. However,
no database was found. It is not known if ATK plans to present all worst-case 1-hour events in a
database once the air modeling results for the general receptor grid are submitted in May 2012. If a
database of the 250 worst case 1-hour events will not be prepared and presented, remove the
reference to the database from the text.
Response; A database of the 250 worst case 1-hour events for each source and each discrete
receptor and general grid maximum impact receptor will be provided in a separate folder within
Appendix B of the Final Report.
9. Section 2.14.1, Discrete Receptors, Page 2-19
The last paragraph on Page 2-19 begins: "...M-136 Source 1 will conduct treatment twice daily, three
times a week for 52 weeks, which is equivalent to 156 annual treatment events." However, the
numbers provided in this sentence result in 312 separate treatment events over 156 days per year.
Revise this sentence to indicate that 312 treatment events are assumed for M-136 Source in the air
modeling analysis.
Response: This sentence has been edited to read, "...M-136 Source 1 will conduct treatment twice
daily, three times a week for 52 weeks, which is equivalent to 312 treatment events."
10. Section 3.2.1, Compliance with NAAQS, Pages 3-3 and 3-4
Incomplete modeling results are reported for two discrete receptors in the air modeling report. For
example, page 3-3 states: "Autoliv is a complex terrain receptor. As a result, only gas modeling
results are available." Page 3-4 presents an identical description for the Boundary 1 discrete
receptor. It appears that no comparison to particulate matter (PM, PM10, or PM2.5) emission
standards were performed for these two discrete receptors. A conservative estimate of the air
concentrations of PM10 and PM2.5 can be estimated by multiplying the gas phase unitized dispersion
factor by the emission factor for PMIO, and PM2.5, respectively. Revise the air modeling report to
address compliance with the particle-based emissions standards at these two discrete receptors by
applying the approach outlined above. In addition, discuss the inherent conservatism in this
approach in Section 3.4.5, Uncertainty Due to OBODM, ofthe Air Modeling Report.
Response: Gas modeling has been performed for Autoliv and Boundary 1 and those air dispersion
factors have been multiplied by the emission factors for PMIO and PM2.5 to provide a conservative
estimate for particle phase concentrations. These concentrations will appear in the final report.
11. Section 3.2.1, Compliance with NAAQS, Pages 3-3 and 3-4
As stated in Section 2.0 of the HHRA Protocol, criteria pollutants will be evaluated in the air
dispersion modeling report by comparing modeled air concentrations to the NAAQS. The criteria
pollutants will not be evaluated in the HHRA. As of June, 2010, the EPA has set primary one-hour
averaging time standards for nitrogen dioxide and sulfur dioxide. Comparisons of modeled air
concentrations to these specific NAAQS are not included in the air modeling report. Please calculate
the appropriate air concentration values and revise the report to include comparisons to these
NAAQS.
Response: The final report will include comparisons to the primary one-hour averaging time
standards for nitrogen dioxide and sulfur dioxide established by USEPA in June 2010 in all NAAQS
summary tables.
12. Section 3.2.2, Compliance with Utah 1-Hour Acute Gas TSLs, Page 3-8
The discussion entitled 1-Hour Acute Gas TSL Summary at the top of page 3-8 indicates
formaldehyde was not detected in the ODOBi testing study. Further, the text states that
formaldehyde is not believed to be an emission product associated with the OB and OD treatment
processes at M-136 and M-225. However, Tables 2-5 and 2-6 as well as the tables in Appendix L
indicate that formaldehyde was detected during testing. Please review the discussion in Section
3.2.2 and the table entries and revise the text and tables as necessary to provide accurate and
consistent information regarding whether formaldehyde was detected during emissions testing.
Response: It has been confirmed that formaldehyde was detected in the ODOBi testing study. As a
result, compliance with the 1-Hour Acute Gas TSL for formaldehyde will be included in the TSL 1-
hour summary tables. Also, the text in Section 3.2.2 will be revised to remove the statement
"However, formaldehyde was not detected in the ODOBi testing study and therefore is not
considered to be an emission product from ATK OB or OD". Final dispersion modeling calculations
will be reviewed to determine compliance with the 1-Hour Acute Gas TSL for formaldehyde.