VIA E-MAIL

April 13, 2010

Mr. Jeffrey A. Telander

Sector Policies and Programs Division

Metals and Minerals Group (D243-02)

Office of Air Quality Planning and Standards

U.S. Environmental Protection Agency

109 TW Alexander Drive

Research Triangle Park, NC  27711

Mr. William Neuffer

Sector Policies and Programs Division

Metals and Minerals Group (D243-02)

Office of Air Quality Planning and Standards

U.S. Environmental Protection Agency

109 TW Alexander Drive 

Research Triangle Park, NC  27711

 

RE:	Representativeness of Wool Fiberglass and Mineral Wool Data

Dear Messrs. Telander and Neuffer:

INTRODUCTION

In the draft Information Collection Requests (“ICRs”) for both the
wool fiberglass and mineral wool insulation manufacturing industries’
residual risk review, the Environmental Protection Agency (“EPA”)
states that “[i]n lieu of new testing, existing and well documented
test data from your current operations that you consider representative
. . . can be submitted.”  This statement is repeated in the “Summary
of Test Procedures, Methods, and Reporting Requirements” for each ICR.

This letter sets forth the wool fiberglass and mineral wool insulation
manufacturing industries’ proposal to provide both existing test data
and new test data to respond to the draft ICR.  Specifically this letter
provides the details of the existing data that will be used to respond
to the ICR and the bases for our position that such test data should
constitute representative industry data.  This letter also identifies
where we propose that new testing be conducted and the bases for our
position that such test data for the selected point sources should
constitute representative industry data.

As we stated in our March 27, 2010 letter concerning the use of
alternative test methods for responding to the draft ICRs, EPA’s
acceptance of such alternative test methods will enable us to provide
EPA the emissions data it needs while both expediting the information
collection process and leveraging the use of existing emissions data. 
The industry recommended these alternative test methods not only because
it has extensive data collected by these alternative methods that are
relevant to EPA’s requests, but also because every company is
operating at reduced production due to the significant economic downturn
in the building industry.  New data could not be collected from the many
production lines that are not in operation, even though they are
currently permitted.  Use of existing data will be essential for
providing EPA with the requisite data for all point sources.

Moreover, we appreciate EPA’s willingness to accept as responsive the
test data generated by many of the alternative test methods we
identified.  However, the use of such data necessarily requires EPA’s
acceptance of at least some, if not all, such data that was generated
pursuant to tests run for the length of time prescribed by the
applicable test methods, not for some longer period of time.  For
example, one hour should be more than sufficient for all the organics to
be measured.  The majority of these test methods have a one-hour length;
detection limits are set to work within the designated timeframes.  If
the detected amount of the pollutant is one or two orders of magnitude
above the detection limit, extending the time and reducing the detection
limit does not provide any additional substantive or meaningful
information because the pollutant in question has already been detected
and measured.  We understand that it is our responsibility to make sure
sampling times are long enough to obtain results that are greater than
minimum detection limits (unless the EPA-recommended sampling times and
volumes are implemented where no-detect results will be accepted).

By this letter, the North American Insulation Manufacturers Association
(“NAIMA”) requests EPA’s approval of the selection of data below
as representative industry data so that we can prepare a plan and
schedule for the collection of the data and other information responsive
to both draft ICRs.

DESCRIPTION OF GLASS AND MINERAL WOOL MANUFACTURING PROCESSES

Fiber glass insulation starts with raw materials such as sand, soda, and
lime.  Broken or recycled glass, known as cullet, is added to the
mixture of raw materials to accelerate the melting process.  These
ingredients are mixed in the batch house and then transferred to and
melted in refractory-lined furnaces, or in some cases, special large
pots.  The heat from the furnace converts these raw materials into a
homogenous molten liquid.  The melting of glass and forming it into a
fiber consists of two distinct stages.  The first involves the chemical
reaction between the ingredients and results in a sticky molten mass
with many bubbles.  In the second stage, the glass is cooled until it
has the required viscosity for fiberization.  The molten glass flows
from the furnace through the channel to the forehearth, where, after
losing a good deal of heat along the way, the glass drains through
bushings into the fiberizing equipment.

Mineral wool follows a similar process, with the raw materials being
different.

EXISTING AND REPRESENTATIVE DATA

To identify and organize the data to be supplied to EPA, this letter
arranges by MACT Standard the point sources and pollutants for which we
will supply existing test data or newly collected data and our bases for
concluding that such data should be deemed representative.

	Wool Fiberglass

Furnaces

Process Unit:	Oxy Fuel

Pollutants Testing Requested:	PM2.5 (Filterable and Condensable), Total
PM (Filterable), Metal HAPs, Hex Chrome, HF, HC1

Data To Be Supplied:	There are 12 oxy fuel furnaces permitted in the
industry.  PM2.5, total PM, metal HAPs, hex chrome, and hydrogen
fluoride data will be collected from Owens Corning’s Kansas City,
Kansas, facility (K5 Line).  As described below, the data collected at
this plant will be representative of all 12 oxy fuel furnaces.  It is
critical that newly-collected data be accepted as representative data
because at least 6 of the 12 oxy fuel furnaces are idle.  A survey of
fiber glass companies indicates that there would be no possible source
for HC1.

Location/Lines Represented:	Knauf – Lanett, Alabama (621 Furnace (with
DEP) and 622 Furnace (with DEP))

	CertainTeed – Athens, Georgia (A2 Furnace); Chowchilla, California
(A1 Furnace); Kansas City, Kansas (K1 Furnace and K2 Furnace); Mountain
Top, Pennsylvania (M1 Furnace and M2 Furnace)

	Johns Manville – Cleburne, Texas (1902 Marble Furnace (has Scrubber
and ESP)); Defiance, Ohio (2601 & 2603 Marble Furnaces (have common ESP
and Baghouse in series and do not have oxy firing))

	Owens Corning – Delmar, New York (DM1 Furnace (with DEP) and DM2
Furnace (with DEP)); Kansas City, Kansas (J6/U3 Furnace (with Electric
Boost and DEP) and K5 Furnace (with DEP)); Waxahachie, Texas (VI Furnace
(with DEP)); Newark, Ohio (F6 Furnace (with Electric Boost and
Baghouse), A3 Furnace (with DEP))

Bases For Representativeness:	The data collected would be representative
to the extent the method of operation is the same, all oxy fuel furnaces
are powered by natural gas, all of the oxy fuel furnaces use a dry
electrostatic precipitator or similar control, all of the oxy fuel
furnaces are making borosilicate glass using essentially the same raw
materials, all of the oxy fuel furnaces are the same class size of
furnaces, and the Kansas City plant is at the midpoint of the range.

Process Unit:	Cold Top Electric – With Baghouse

Pollutants Testing Requested:	PM2.5 (Filterable and Condensable), Total
PM (Filterable), Metal HAPs, Hex Chrome, HF, HC1

Data To Be Supplied:	Heavy metal and hex chrome data will be collected
from Knauf’s Shasta Lake, California facility using EPA Method 29 and
the accepted test method for hex chrome, and this data will be
representative of all cold top furnaces with baghouses.  PM2.5 and Total
PM will be provided from Knauf’s Shasta Lake, California, facility’s
existing data.  This data will be representative of all cold top
furnaces with baghouses, as described below.  A survey of fiber glass
companies indicates that there appears to be no possible source for HF
or HC1 from this group of facilities.

Location/Lines Represented:	Knauf – Shelbyville, Indiana (611 Furnace
and 602 Furnace); Shasta Lake, California (641 Furnace)

	CertainTeed – Athens, Georgia (A1 Furnace)

	FiberTEK – Lakeland, Florida (Middle Melter and West Melter); Nephi,
Utah (Middle Melter and West Melter)

	Johns Manville – Richmond, Indiana (Melter); Defiance, Ohio (3
Melters, Lines 801-806)

Bases For Representativeness:	The data collected would be representative
to the extent the method of operation is the same, all of the furnaces
are operated with electricity, all of the furnaces are the same brand,
all of the furnaces are making borosilicate glass using essentially the
same material, and all of the furnaces have a baghouse.

Process Unit:	Cold Top Electric – Without Baghouse

Pollutants Testing Requested:	PM2.5 (Filterable and Condensable), Total
PM (Filterable), Metal HAPs, Hex Chrome, HF, HC1

Data To Be Supplied:	Heavy metal, hex chrome, and hydrogen fluoride data
will be collected from Owens Corning’s Kansas City, Kansas, facility
(70 furnace) using EPA Method 29 and the accepted test method for hex
chrome, and that data will be representative of all cold top furnaces
without baghouses.  New emissions tests for PM2.5 and Total PM will also
be conducted.  This data will be representative of all cold top furnaces
without baghouses, as described below.  A survey of fiber glass
companies indicates that there appears to be no possible source for HC1.

Location/Lines Represented:	Owens Corning – Fairburn, Georgia (FG1,
FG2, FG3 Furnaces (With Batch Wetting)); Newark, Ohio (C4 and B4
Furnaces (With Batch Wetting)); Kansas City, Kansas (70 Furnace (With
Batch Wetting)); Waxahachie, Texas (V2 Furnace, V3 Furnace (With Batch
Wetting)); Santa Clara, California (M and O Furnaces (With Batch
Wetting)) 

Bases For Representativeness:	The data collected would be representative
to the extent the method of operation is the same, all of the furnaces
are operated with electricity, all of the furnaces are the same basic
design, all of the furnaces are making borosilicate glass using
essentially the same material, all furnaces are operated without a
baghouse, and the capacity is at the midpoint of the range.

Process Unit:	Electric Steel Shell Melter – With Baghouse

Pollutants Testing Requested:	PM2.5 (Filterable and Condensable), Total
PM (Filterable), Metal HAPs, Hex Chrome, HF, HC1

Data To Be Supplied:	Existing PM2.5, total PM, and heavy metal data from
Johns Manville’s Willows, California, facility will be used as
representative data for all electric steel shell melters in the
industry.  Data for hex chrome will be collected at Johns Manville’s
Willows, California, facility and will be used representative data for
all electric steel shell melters in the industry.

Location/Lines Represented:	Guardian – Albion, Michigan (Line 1
Furnace, Line 2 Furnace, Line 3 Furnace, Line 4 Furnace); Kingman,
Arizona (Line 1 Furnace); Inwood, West Virginia (Line 1 Furnace, Line 2
Furnace); Winnsboro, South Carolina (Line 1 Furnace); Mineral Wells,
Mississippi (Line 1 Furnace, Line 2 Furnace)

	Johns Manville – Penbryn,  New Jersey (L58 Melter); Winder, Georgia
(Line 105 and 106 Melters); McPherson, Kansas (Line 122 and 123
Melters); Willows, California (Line 132 and 133 Melters); Cleburne,
Texas (Lines 91, Melter); Defiance, Ohio (Line 89 Melter)

	Owens Corning – Eloy, Arizona (Two Furnaces); Salt Lake City, Utah
(Two Furnaces); Mount Vernon, Ohio (Three Furnaces)

Bases For Representativeness:	The data collected would be representative
to the extent the method of operation is the same, all of the furnaces
are operated with electricity, all of the furnaces are steel shell
pot-type containers with refractory liners, all of the furnaces use a
baghouse, and all of the furnaces are making borosilicate glass using
essentially the same raw materials.

Process Unit:	Pot & Marble Flame Attenuated

Pollutants Testing Requested:	PM2.5 (Filterable and Condensable), Total
PM (Filterable), Metal HAPs, Hex Chrome, HF, HC1

Data To Be Supplied:	Johns Manville is the only member in NAIMA with a
Pot & Marble melting process.  They melt glass marbles in very small
pots using small gas burners (not electric).  It is a completely
different process than rotary and was subcategorized in the original
MACT standard.  Johns Manville will provide heavy metal, hex chrome,
PM2.5 and Total PM data from existing data.  A survey of fiber glass
companies indicates that there would be no possible source for HF or HC1
from this group of facilities.

Location/Lines Represented:	Johns Manville – Defiance, Ohio (Lines 81,
82, 84, 85, 86, 87, 88, 21, 23, 24, 25); Cleburne, Texas (L92, 93);
Waterville, Ohio (pot and marble lines (no binder)).  

Bases For Representativeness:	The data collected would be representative
to the extent the method of operation is the same, all furnaces are
making borosilicate glass using essentially the same material, and all
are pot & marble flame attenuation.  

Forming, Curing, and Cooling

For purposes of identifying representative data responsive to EPA’s
request, NAIMA has combined the forming, curing, and cooling sources
together into one source.  These three sources are more often than not
exhausted through one stack or common exhaust system, not separately and
individually.  Where a common stack or exhaust is not used, NAIMA will
combine the emissions data from the three sources into one data set. 
This is consistent with prior practices and eliminates a lot of
confusion in treating forming, curing, and cooling in separate
categories that ultimately undertake the same discussion and analysis
for each source.

Process Unit:		Forming, Curing, and Cooling

Pollutants Testing Requested:		Formaldehyde, Phenol, Methanol

Data To Be Supplied:	All plants with bonded lines using formaldehyde
will use existing data from Owens Corning’s Kansas City, Kansas,
facility for formaldehyde, phenol, and methanol with appropriate factors
to account for different control devices, LOI, and/or pounds per ton of
melt, as applicable, because this facility’s forming, curing and
cooling sources have no voc controls.  Formaldehyde, phenol, and
methanol data are from testing done in accordance with Method 316 for
formaldehyde, Method 308 for methanol, and Lab 8, BAAQMD Method ST-16
for phenol.  No formaldehyde, phenol, or methanol data will be supplied
for plants with bonded lines using a non-Phenol/Formaldehyde (“P/F”)
binder.  The data that will be provided to EPA is not representative of
these unbonded lines, and no data has been collected for these unbonded
lines as agreed upon by EPA.

Location/Lines Represented:	The following company facilities that use
phenolic/ formaldehyde binders for which data from Owens Corning’s
Kansas City facility will serve as representative data on formaldehyde,
phenol, and methanol are:

	CertainTeed – Athens, Georgia (All Lines); Chowchilla, California
(C11 Line); Kansas City, Kansas (K11 Line, K21 Line); Mountain Top,
Pennsylvania (M1 and M2 Lines)

	FiberTEK – Lakeland, Florida (BL Line); Nephi, Utah (BL Line)

	Guardian – Albion, Michigan (Line 1, Line 3); Kingman, Arizona (Line
1); Inwood, West Virginia (Lines 1 and 2); Winnsboro, South Carolina
(Line 1); Mineral Wells, Mississippi (Line 1)

	Johns Manville – Cleburne, Texas (Lines 91, L92, L93 Lines);
Defiance, Ohio (Lines 89, 801-806, 807, 81-88, 21-25)

	Knauf – Shelbyville, Indiana (Bonded Line)

	Owens Corning – Delmar, New York (DM1 and DM2 Lines); Fairburn,
Georgia (FG1, FG2, FG3 Lines); Newark, Ohio (C4, F6, B4, A4, A5 Lines);
Kansas City, Kansas (70, J6, K5 Lines); Waxahachie, Texas (V1, V2, V3
Lines); Santa Clara, California (M, O Lines); Eloy, Arizona (1 Bonded
Line); Salt Lake City, Utah (1 Bonded Line)

	

	The following member company facilities that do not use
phenolic/formaldehyde binders or use no binders (unbonded lines) for
which no such test data will be provided are:

	CertainTeed – Athens, Georgia (A12 and A23 Lines); Chowchilla,
California (C12 Line); Kansas City, Kansas (K11 and K21 Lines)

	FiberTEK – Lakeland, Florida (UL Line); Nephi, Utah (UL Line)

	Guardian – Albion, Michigan (Lines 2 and 4); Mineral Wells,
Mississippi (Line 2)

	Johns Manville – Penbryn, New Jersey (Line 58); Richmond, Indiana
(Lines 2 and 3); Waterville, Ohio (Pot/Marble Production); Winder,
Georgia (Lines 105 and 106); McPherson, Kansas (Lines 122 and 123);
Willows, California (Lines 132 and 133)

	Knauf – Shasta Lake, California (100% Non-P/F); Lanett, Alabama (100%
Non-P/F); Shelbyville, Indiana (Approximately 75% Non-P/F)

	Owens Corning – Kansas City, Kansas (U3 Line); Salt Lake City, Utah
(SL2 Line); Mount Vernon, Ohio (3 Unbonded Lines)

Bases For Representativeness:  	The data collected being provided would
be representative of forming, curing, and cooling sources at facilities
suing formaldehyde/phenolic binder because the method of manufacture is
the same, of the all products are being formed from borosilicate glass
using essentially the same raw materials, all of the products are
bonded, all of the binders are phenolic, and essentially the same
process controls are used.

Mineral Wool

Process Unit:	Cupola – With Baghouse and Incinerator

Pollutants Testing Requested:	Carbon Monoxide, Carbonyl Sulfide

Data To Be Supplied:	Existing carbonyl sulfide and carbon monoxide data
from a USG Interiors facility will be used as representative data for
all mineral wool cupolas with baghouses and incinerators as described
below.  The carbonyl sulfide data has been collected with EPA Method 10,
and the carbon monoxide data has been collected with EPA Method 320.

Location/Lines Represented:	Based on the bases/assumptions identified
below, this data is representative of all lines at the following plants
and locations:

	Industrial Insulation Group – Phenix City, Alabama

	Thermafiber – Wabash, Indiana (Bonded #2, Granulated #4)

	USG Interiors – Red Wing, Minnesota; Walworth, Wisconsin

Bases For Representativeness:	This data is representative to the extent
each cupola is coke-fired, melting almost identical raw material mix,
producing rock and slag wool products, and using controls of an
incinerator or baghouse.

Process Unit:	Cupola – With Baghouse Only

Pollutants Testing Requested:	Carbon Monoxide, Carbonyl Sulfide

Data To Be Supplied:	Existing carbonyl sulfide and carbon monoxide data
from the Amerrock Products, Nolanville, Texas facility will be used as
representative data for all mineral wool cupolas with baghouses and no
incinerators.  The carbonyl sulfide data has been collected with EPA
Method 15, and the carbon monoxide data has been collected with EPA
Method 10.  

Location/Lines Represented:	Based on the bases/assumptions identified
below, this data is representative of all lines at the following plants
and locations:

	Amerrock Products – Nolanville, Texas

	Isolatek International – Huntington, Indiana (Cupola #1, Cupola #2
– common baghouse/combined point source)

	Rock Wool Manufacturing – Leeds, Alabama (1 Cupola, Line #2)

Bases For Representativeness:	This data is representative to the extent
each cupola is coke-fired, melting almost identical raw material mix,
producing rock and slag wool products, and using controls of a baghouse.

Process Unit:	Cupola

Pollutants Testing Requested:	PM 2.5 (Filterable and Condensable), Total
PM (Filterable), Metal HAPs, Hex Chrome, HF, HC1

Data To Be Supplied:	Existing metal HAPs data from a USG Interiors
facility will be used as representative data for the entire industry, as
described below.  The data has been collected pursuant to EPA Method 29.
 Specific data on hexavalent chromium, mercury, and HF and HC1 need not
be provided to EPA.  Hex Chrome is identified in the wool fiber glass
industry because glass furnaces are lined with chrome refractories that
contain hexavalent chromium.  Mineral wool companies do not use chrome
refractories to line their cupolas.  To the extent that hex chrome may
appear in raw materials, NAIMA will provide data for the industry.  A
survey of mineral wool companies indicates that there would be only a
limited possibility for a source for HF emissions and apparently no
possible basis for HC1 emissions.  PM2.5 (filterable and condensable)
and Total PM filterable will have to be collected from new emissions
testing.  This data will be collected pursuant to EPA Method 27 or 28 at
one of the USG Interiors facilities.  The data collected from USG
Interiors will be used as representative data for the entire mineral
wool industry.

Location/Lines Represented:	Based on the bases/assumptions identified
below, this data is representative of all lines at the following plants
and locations:

	Amerrock Products – Nolanville, Texas

	Industrial Insulation Group – Phenix City, Alabama

	Isolatek International – Huntington, Indiana

	Rock Wool Manufacturing – Leeds, Alabama

	Thermafiber – Wabash, Indiana

	USG Interiors – Red Wing, Minnesota; Walworth, Wisconsin

Bases For Representativeness:	This data is representative to the extent
that all of the cupolas are coke-fired, melting almost identical raw
material mix, producing rock and slag wool products, and using controls
of an incinerator or baghouse.

Process Unit:	Collection Operation

Pollutants Testing Requested:	Formaldehyde, Phenol, Methanol

Data To Be Supplied:	Existing formaldehyde data will be supplied from
Thermafiber, Industrial Insulation Group, and Rock Wool Manufacturing. 
New methanol and phenol data will be collected from Thermafiber pursuant
to one the EPA approved methods (EPA Method 318 or 308 for methanol and
EPA Method 318 or Method 8270D for phenol) and be used as representative
for all three companies as described below.  The remaining companies,
USG Interiors, Amerrock Products, and Isolatek International, operate
unbonded lines, and, therefore, use no binder is applied during the
manufacturing of mineral wool.  In addition, Thermafiber line –
Granulated #4 is unbonded.  The data that will be provided to EPA is not
representative of these unbonded lines, and no data has been collected
for these unbonded lines as agreed upon with EPA.    

Location/Lines Represented:	As noted above, the data supplied will be
for existing or representative data for:

	Industrial Insulation Group – Phenix City, Alabama

	Thermafiber (Bonded #2) – Wabash, Indiana

	Rock Wool Manufacturing (Bonded Line #2) – Leeds, Alabama

Bases For Representativeness:	This data is representative to the extent
that all of the cupolas are coke-fired, melting almost identical raw
material mix, producing rock and slag wool products, and using controls
of an incinerator or baghouse.

Process Unit:	Curing Oven

Pollutants Testing Requested:	Formaldehyde, Phenol, Methanol

Data To Be Supplied:	Existing formaldehyde data will be supplied from
Thermafiber, Industrial Insulation Group, and Rock Wool Manufacturing. 
New methanol and phenol data will be collected from Thermafiber pursuant
to one the EPA approved methods (EPA Method 318 or 308 for methanol and
EPA Method 318 or Method 8270D for phenol) and be used as representative
for all three companies as described below.  The remaining companies,
USG Interiors, Amerrock Products, and Isolatek International, operate
unbonded lines, and, therefore, no binder is applied during the
manufacturing of mineral wool.  In addition, Thermafiber line –
Granulated #4 is unbonded.  The data that will be provided to EPA is not
representative of these unbonded lines, and no data has been collected
for these unbonded lines as agreed upon with EPA.  

Location/Lines Represented:	As noted above, the data supplied will be
for existing or representative data for:

	Industrial Insulation Group – Phenix City, Alabama

	Thermafiber (Bonded #2) – Wabash, Indiana

	Rock Wool Manufacturing (Bonded Line #2) – Leeds, Alabama

Bases For Representativeness:	This data is representative to the extent
that each cupola is coke-fired, melting almost identical raw material
mix, producing rock and slag wool products, and using controls of an
incinerator or baghouse.

Process Unit:	Cooling Operation

Pollutants Testing Requested:	Formaldehyde, Phenol, Methanol

Data To Be Supplied:	New formaldehyde, methanol, and phenol data will be
collected from Thermafiber pursuant to one of the EPA approved methods
(EPA Method 318 or 308 for methanol and EPA Method 318 or Method 8270D
for phenol) and be used as representative for all three companies as
described below.  The remaining companies, USG Interiors, Amerrock
Products, and Isolatek International, operate unbonded lines, and,
therefore, no binder is applied during the manufacturing of mineral
wool.  In addition, Thermafiber line – Granulated #4 is unbonded.  The
data that will be provided to EPA is not representative of these
unbonded lines, and no data has been collected for these unbonded lines
as agreed upon with EPA.  

Location/Lines Represented:	As noted above, the data supplied will be
for existing or representative data for:

	Industrial Insulation Group – Phenix City, Alabama

	Thermafiber (Bonded #2) – Wabash, Indiana

	Rock Wool Manufacturing (Bonded Line #2) – Leeds, Alabama

Bases For Representativeness:	This data is representative to the extent
that all of the cupolas are coke-fired, melting almost identical raw
material mix, producing rock and slag wool products, and using controls
of an incinerator or baghouse.

Sincerely,

Angus E. Crane

Angus E. Crane, Executive Vice President, General Counsel

 F.J. Terence Maloney, Glass in the Modern World, at 74 (New York:
Doubleday, 1968).

Messrs. Jeffrey A. Telander and William Neuffer

April 13, 2010

Page   PAGE  12 

 



3

4

7

8

:

6

z

µ

À

Ï

Ø

h¥

hö

h¥

h¥

hö

h¥

h

h¥

h

h¥

▪  Suite 310  ▪  Alexandria, Virginia  22314  ▪  Tel: (703)
684-0084  ▪  Fax: (703) 684-0427

