	

Re: Pb designations modeling comments/questions

Sarah VanderWielen 

to:

Mary Portanova

05/07/2010 12:36 PM

Cc:

Andy Chang, Michael Compher, "Hunter,	Jennifer", Randall Robinson

 

 







From:	"Sarah VanderWielen" <Sarah.VanderWielen@epa.state.oh.us>





	To:	Mary Portanova/R5/USEPA/US@EPA





	Cc:	Andy Chang/R5/USEPA/US@EPA, Michael Compher/R5/USEPA/US@EPA,
"Hunter,	

Jennifer" <Jennifer.Hunter@epa.state.oh.us>, Randall
Robinson/R5/USEPA/US@EPA

	

Hi Mary, 

 

Below are responses to the questions on the lead modeling submitted for
the Lead NAAQS designations. 

 

Please let Jennifer Hunter know if you have any additional questions (I
will be out of the office until June 2, 2010). 

 

Thanks, 

 

Sarah 

 

Sarah VanderWielen

Meteorologist

Ohio EPA

Division of Air Pollution Control

50 W. Town Street 

Columbus, Ohio 43215

Phone - 614-644-3632

Please consider the environment before printing this email. 

>>> <Portanova.Mary@epamail.epa.gov> 5/6/2010 12:13 PM >>>

Sarah, here is some of what we sent to Tyler Fox's group at HQ when they

requested info on how the State designations modeling had been done.

This refers to the Pb designation modeling for Logan and Fulton

Counties.  Perhaps you can address some of the points here.  The intent

of these comments were to address the question, "Was the modeling done

in accordance with App W?"  Ideally, where there is spotty source

information, the modeling should use conservative estimates where

possible, and explain/justify the estimates.  Thanks for rechecking the

source locations; it's good to know that those were just typoes in the

table, not in the model runs.

I've also attached part of the draft TSDs with HQ's comments on the

modeled sources.

Logan Co/Bellefontaine

The Daido area source emission rate was higher than the NEI, but the

source of the data was not identified.

 

- Ohio used three sources of information regarding emissions: NEI, TRI
and Ohio's FER (Fee emissions report).  Typically the FER is used to
populate the NEI; however we have found that post FER reporting, changes
have been made for unknown reasons. Ohio worked extensively with the
companies, USEPA (Motria) and others to develop accurate numbers for
each facility.  When necessary Ohio submitted corrections to USEPA to
these numbers (e.g., if the NEI was too high).  But in the end, the
analysis of who should receive a monitor, who should be modeled for the
need, or who was out was based solely on the worst case emissions. And
subsequently, all modeling conducted was based again on the worst case
numbers.  In some cases when we began working with the individual
companies that were to be modeled and requesting emissions rate/stack
heights, etc.; rates would not always match the total actual TPY. 
Because of the tight time frame for submittal; Ohio chose not to spend
more time working out the details of the rates if they were
overestimating emissions.  Ohio felt this would result in a more
conservative approach that would result in a larger boundary if
anything.  It should be noted that the modeled plume boundary was
extensively smaller than the boundary requested.

Three other Pb sources were identified.  One was modeled as an area

source because the State believed its lead emissions come from an area

source.  Actual source info was not provided.  Building data was used as

the area source size.  The modeled emission rate matched the tpy data of

the 2005 NEI and 2005 TRI.

 

-Quincy: the vast majority of lead emissions (which are minor in
themselves from this facility which is located nearly 12 miles west of
Daido) are fugitive (>95%). The fugitive sources are :

 

 

The other much more minor source of lead is from the following:

 

 

Which is controlled by a fabric filter baghouse.  It should also be
noted that Quincy is "inactive" as of 3/5/10.  They have chosen not to
permanently shutdown but before resuming operations or selling the
company they would need to re-contact Ohio EPA.

The other two sources (two Honda facilities) were modeled as single

point sources.  Ohio had stack information for the lead-emitting sources

at one facility.  This information was averaged to create a single point

source.  Analysis gave no information on the actual stacks or how the

averaging was done.  The other facility did not provide stack

information to Ohio, so the stack data for the other Honda facility was

used.  The analysis provides no further justification for this.  (See

attachment for HQ comments re the source information given in NEI.)

 

Ohio was relying on Ohio's own internal databases and direct contact
with the facilities for emission source information. Ohio did not
realize stack information was available in the NEI. Recently reported
information shows that for the Honda transmission company the stack is
45' (NEI showed 42) and account for about 54% of the NEI emssions
(remaining fugitive). For the Honda East Liberty plant the stacks are
between 85-90 (NEI showed 70) and account for about 27% of the NEI
emissions.    Ohio used 88 ft as the stack height for both facilities
because that was the average of the East Liberty stacks for which we had
available information at that time.   Honda Transmissions is about 10
miles NW of Daido while Honda East Liberty is about 10 miles East. 
Winds are predominantly out of the SW.  Given the distance, wind pattern
and low emissions of the two Honda sources, it is extremely unlikely any
changes to stack heights would change any impact, also considering the
offseting affect of applying 75% of the Honda East Liberty emissions as
fugitive area source emissions. 

Were Pb emission rates also averaged?

 

No the lead emissions were not averaged.  The total reported amount by
Honda was modeled  (0.0075 tpy)

No information was given about land use characterizations besides "one

sector, four seasons."  Did this modeling use the same AERSURFACE run as

in the monitor siting work, and was it the corrected AERSURFACE run, if

it was one where R5 had identified an issue during the monitor siting

process?  What year or years of met data were used?

 

AERSURFACE was run using one sector, four seasons for Logan County.  The
same AERSURFACE file was used in the monitor siting analysis.  Met data
was from 1984-1988.  

200m receptor spacing is less dense than recommended.  (If the selected

boundaries were well beyond the hot spot areas, receptor spacing might

not be a big issue in the case of designations, but there was no

discussion of why 100m resolution was not used or not considered

necessary.)

 

Ohio feels that 200m spacing, although less dense than recommended, was
adequate to model the maximum impact from these facilities.  When
reviewing the modeled results, it appears that Daido contributed 99.98%
of the maximum modeled concentration.  The next largest contribution was
from Quincy Casting which was 0.05% of the maximum modeled
concentration.  Both Honda plants had little to no impact on the maximum
modeled concentration. 

 

As mentioned above, Ohio was very conservative in expanding its boundary
beyond the impact area.  From a big picture perspective - the only other
sources of lead in the county are E and NW (two Honda - recall also they
are 50% or more fugitive) at 0.0075 TPY each and a fugitive area source
(Quincy) at 0.035 TPY while Daido was modeled at 0.121 TPY and Daido no
longer operates.  A larger area need not be part of the boundary.  The
entire county is comprised of four facilities emitting lead- one
permanently shutdown, one idle and two with 0.0075 TPY emissions each. 
Based on the monitor at Daido the last 3-month violation was in May
2007.  We have seen a drop in monitor values since the June shutdown.
Here are the 3-month averages from June to October of 2009:

 

 

Here is June to October of 2008, while still operating

 

 

If this trend continues through May 2010 (the end of this month) we will
have three complete years of no violations, albeit they aren't calendar
years.  If there continues to be any violations in this area it is
likely a result of lead entrained in the dirt.

Background concentrations were not included.   Are there monitors that

could have provided background concentrations?  (Local monitoring data

was used in the analysis, but in a different way.)

-No we do not have background lead monitors.   Background values were
not included because all lead sources in the are were included in the
modeled inventory.   And also - we asked USEPA for guidance with respect
to this and were told to set the background at 0.

   Fulton Co/Delta

Three other sources (other than Bunting Bearing) were included in the

modeling.  They were modeled as single point sources using 2005 NEI and

2005 TRI emissions.  Stack parameters were averaged for two facilities.

Need some explanation of the stack parameter averaging.  There was no

stack information for the fourth facility, so Bunting Bearing stack

parameters, deemed to be the most conservative, were used for this

facility.  (conservative is good...no NEI source info available?)  Were

fugitives considered?  (See also HQ comments in attachment.)

 

Northstar - West 2 miles - stack 89' (NEI all emissions stack)

Multicast - West (slightly south) 7.5 miles, stack 48' (emissions from
stack and fugitive are about 50/50 in the NEI)

Sauder - SW over 16 miles - 2 are 80' stacks accounting for 90% of
emissions and one 40' stack. 

 

Upon review, it was discovered that a mistake was made when converting
the latitude and longitude into UTMs for Sauder.  It is believe that
this error occured because this modeling includes some sources in UTM
zone 16 and some in UTM zone 17.  The inventory was converted to a
single zone so that modeling can be preformed.  As a result of this
error, Sauder was actually modeled much closer to the Bunting Bearings
facility, and thus the monitor itself.  Because this source was moved
closer to the monitor, it should be considered a conservative modeling
approach.  Even with Sauder moving closer to the monitor, it still
showed very little modeled impact.  

 

Emissions from Multicast and Sauder are inconsequential considering
their distance and low levels; however all emissions were modeled as
stack so as to provide the most conservative estimate of potential
impact (fugitive lead emissions are less likely to transport such great
distances and we had no record (just like the NEI) of fugitive sources).
 Northstar emissions were modeled at about 1/2 that of Bunting and all
stack (fugitive emissions added into the stack) - this is conservative
as if they were modeled as fugitive they would have even less
opportunity to be transported east (winds predominantly out of the SW). 

 

Were Pb emission rates averaged?

 

No, the emission rates were not averaged.  The maximum emission rates
reported were modeled for each facility. 

Same question about the surface characteristics and met data.

 

AERSURFACE was run using one sector, four seasons for Fulton County. 
The same AERSURFACE file was used in the monitor siting analysis.   Met
data was from 1984-1988. 

Same comment on the receptor resolution, which in this case was 150m.

 

Ohio feels that 150m spacing, although less dense than recommended, was
adequate to model the maximum impact from these facilities. When
reviewing the modeled results, it appears that Bunting Bearing
contributed 99.6% of the maximum modeled concentration.  The next
largest contribution was from Sauder which was 0.66% of the maximum
modeled concentration.  Northstar and Multicast had little to no impact
on the maximum modeled concentration. 

Same comment on the background.

 

-No we do not have background lead monitors.   Background values were
not included because all lead sources in the are were included in the
modeled inventory.   And also - we asked USEPA for guidance with respect
to this and were told to set the background at 0.

