HF Modeling Task
Mike Williams
November 19, 2013
Objective
• Objective: To estimate and compare short-term
concentrations of HF associated with emissions
from the Intel’s Rio Rancho facility to the Texas
Commission on Environmental Quality (TCEQ)
short-term Environmental Screening Level (ESL)
for Hydrogen Fluoride (HF) (15 micrograms per
cubic meter at the altitude of Intel). The
screening level was developed by the TCEQ
based on studies by Lund in 1999 (see
http://www.tceq.state.tx.us/assets/public/implem
entation/tox/dsd/final/october09/hydrogen_fluori
de.pdf).
Background
• “Spikes” in contaminant concentrations are a
concern to some people in the community
• HF (Hydrogen Fluoride) is emitted by chip
production facilities and may be a health
concern at low concentrations
• HF was mentioned in the draft ATSDR report on
community health issues at Rio Rancho
• Short-term, elevated concentrations of HF are of
more concern than are long-term average
concentrations
Background (cont.)
• Texas Commission on Environmental Quality
(TCEQ) short-term Environmental Screening
Level (ESL) for Hydrogen Fluoride (HF), 15
micrograms per cubic meter (μg/m3), was
selected as an appropriate yardstick for
assessing HF concentrations
• The TCEQ ESL was one of the standards that
was suggested by Kowalski of ATSDR when we
asked what he thought would be appropriate for
this work
Background (continued)
• The TCEQ ESL was based on a study involving
exposures of 25 healthy, male volunteers to
three levels of HF for an hour period. The lowest
level was divided by a factor of 30 to adjust for
the small sample size and the selected
population, which was healthy males aged
between 20 and 50. The screening level is used
as a planning tool to decide if further studies or
different options are required for a new facility. If
modeled levels are below the screening level no
further studies or options need to be used.
Approach
• Estimate short-term HF concentrations
with an EPA approved air quality model
• Use on-site meteorological data and
model parameters consistent with EPA
approved procedures
• Use median measured emission rates for
HF
• Use three years, 2010, 2011, and 2012, of
measured meteorological data
Approach (cont.)
• Use source parameters consistent with the
most recent permit applications
• Once the results are obtained, examine
the variations from one receptor to the
next to assure that the receptor spacing is
appropriate
• Examine the variation in measured
emission rates to see if the conclusions
are sensitive to fluctuations in emissions
Task Apportionment
• Class 1 prepares meteorological inputs
• Kurt Parker provides median emission
levels from HF stack measurements
• Class 1 prepares basic source inputs
• Mike Williams runs the model & analyzes
the results
Model-choice AERMOD
• AERMOD is an EPA approved model for
estimating concentrations from industrial
sources
• It includes parameterizations for treating
the dispersion produced by building wakes
• It is particularly appropriate for near
source estimations where travel times are
short
Receptor locations
Receptor Locations
3901500
3901000
3900500
UTM-Northing
3900000
3899500
Series1
3899000
3898500
3898000
3897500
3897000
347500
348000
348500
349000
349500
UTM -Easting
350000
350500
351000
Boundary-Line Receptors
Line Receptors
3900200
3900000
3899800
3899600
3899400
3899200
3899000
3898800
3898600
3898400
3898200
348600 348800 349000 349200 349400 349600 349800
Line Receptors
Source Characteristics
• 22 emission points
• Stack heights and emission rates vary
greatly from one source to the next
• Emissions also vary with time so that the
highest emissions from one of the major
sources are 35% higher than the median
emissions used in the modeling
Source Locations
UTM Northing
Boundary Receptors & Sources
3900200
3900000
3899800
3899600
3899400
3899200
3899000
3898800
3898600
3898400
3898200
348600 348800 349000 349200 349400 349600 349800
UTM Easting
Series1
Model Options
• Appropriate for desert, semi-rural
conditions
• Used one “beta” option designed to
improve estimates in low wind conditions
• “Beta” options must be approved for
regulatory applications. While these
meteorological data sets have not been
reviewed by the regulatory agency, it
followed the same methodology previously
approved by the department
Results
• Highest concentration is 7.5 μg/m3 on
11/25/2012 @4 am. The screening level is
15 μg/m3.
• Next Highest is 7.3 at a nearby receptor
• Highest concentrations in 2011 and 2010
were 5.7 and 5.3 μg/m3
• Nearby concentrations were very similar
Source Locations & Highest Hits
UTM Northing
Boundary Receptors & Sources
3900200
3900000
3899800
2010
3899600
3899400
2011
3899200
Series1
3899000
3898800
3898600
3898400
3898200
2012
348600 348800 349000 349200 349400 349600 349800
UTM Easting
Boundary receptors, sources, & highest
concentrations during 19 events
Boundary Receptors, Sources, & Hit
Receptors
UTMY (meters)
3900500
3900000
Boundary Receptors
3899500
Sources
3899000
Hit Receptors
3898500
3898000
348500
349000
349500
UTMX (meters)
350000
ILLUSTRATION OF “SPIKES”
with constant emissions
HF Concentrations on 11/25/2012
Modeled for a single receptor
8
Concentrations (ug/m3)
7
Environmental
Screening
Level=15μg/m3 at the
altitude of Intel
6
5
4
HF levels
Odor threshold is 30 to
110 μg/m3
3
2
1
0
0
5
10
15
Hour
20
25
30
Conclusions
• EPA approved model estimates HF
concentrations lower than the screening level
• Receptor locations were sufficiently close to one
another that the results are not sensitive to the
choice of receptor locations
• Measured emissions showed more variation with
time than expected, but not enough to change
the conclusions
• However, the modeled concentrations did not
provide a large margin for error so that further
examination of the role of model options might
be advisable
Possible Next Steps
• Rerun model without “beta” option
• Use old tracer measurements to check
validity of plume downwash model
• Examine other pollutants released by the
scrubbers and scale to obtain model
estimates for them
Why is downwash model important
• The manner in which buildings cause dilution is very
important in near source concentrations
• Risk Assessment reports maximum HF hourly estimated
concentrations of 12ug/m3 for Prime downwash (used in
our modeling) versus 33ug/m3 for Schulman-Scire
downwash (tables A-4 & A-5)
• Risk Assessment describes comparison between
modeled and measured concentrations and prime model
underestimated measurements by a factor of 2.2 while
Shulman-Scire underestimated by 1.12 (for highest
tracer measurements, table 4-3)
Differences between Risk
Assessment and Current Work
• RA used occupational limits divided by a factor of 100 for
acceptable level of 25ug/m3 versus 15ug/m3 (at altitude)
• RA had different source configueration
• RA used higher emissions, RA had a rate equivalent to
12,603 pounds per year (table 3-4) while our rate is 5287
pounds per year
• RA used CalPuff model, I used AERMOD
• I used 3 years of recent met. data and they used older
data
• I got 7.5ug/m3 and they got 12ug/m3 for the prime
downwash
Possible Next Steps
• Examine other contaminants from
scrubbers
• Rerun without beta model
• Compare new AERMOD with old tracer
measurements
• Conduct & compare HF measurements
with AERMOD estimates
Estimations for other contaminants
• Examine risk assessment & ATSDR
reports to find candidates
• Choose appropriate screening levels
• Ratio emissions to HF & estimate
concentrations
• Downwash concerns remain
Rerun without Beta Options
• Requires rerunning met. model with new
inputs
• Probably not much change
• Downwash modules still a concern
Compare AERMOD results to old
tracer measurements
• Make revised model inputs – emissions &
meteorological inputs
• Represents only 2 months meteorology
• Source configueration is much different
Measure HF concentrations in the
future & compare to future
estimates
• HF emissions must be estimated
• Can the FT-IR sampling be done with
adequate minimum detection levels
• Where could the resources to support
such a study be obtained?