IT3 2005 Conference

45Lemieux

Paper for presentation at the 2005 Conference on Incineration and Thermal Treatment
Technologies, Galveston, TX, May 9-13, 2005

A Pilot-Scale Study on the Combustion of Waste Carpet in a Rotary Kiln: Dioxin and

Furan Emissions

Paul M. Lemieux
U.S. EPA
Office of Research and Development
National Homeland Security Research Center
Research Triangle Park, NC 27711

Chris Winterrowd
ARCADIS G&M
Durham, NC 27709

Matthew Realff, Jim Mulholland
Georgia Institute of Technology
Atlanta, GA 30332-0100

ABSTRACT

Post-consumer carpet is a potential substitute fuel for high temperature thermal processes such as
cement kilns and boilers. In addition, cleanup of contaminated buildings can result in the need to
dispose of potentially significant quantities of carpet, which may or may not be contaminated,
and will possibly have decontamination chemicals present. Data gaps exist regarding the
potential for the production of increased levels of oxides of nitrogen (NOx), organic pollutants
such as polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans (PCDDs/Fs), and
engineering issues such as pre-sizing requirements for the carpeting in order to achieve effective
combustion. To respond to these data gaps, US EPA, in collaboration with the Carpet and Rug
Institute (CRI) and the Georgia Institute of Technology, performed experiments to address some
of these data gaps. This paper reports on results examining emissions of PCDDs/Fs from a series
of pilot-scale experiments performed on the EPA's rotary kiln incinerator simulator facility in
Research Triangle Park, NC.

INTRODUCTION

Building decontamination and cleanup efforts from a biological warfare (BW) or chemical
warfare (CW) agent terrorist attack typically result in a significant quantity of building
decontamination residue (BDR). This BDR consists mainly of porous materials, such as
carpeting or ceiling tiles, which were removed from the building either before or, after
decontamination efforts. The BDR is likely to have been decontaminated but due to its porous
nature and the limitations of sampling methodologies, the possibility exists for the presence of
trace quantities of agents, as well as the likelihood of the presence of varying quantities of
decontamination chemicals (e.g., bleach solutions). One likely disposal technique for the BDR is
high temperature thermal incineration. Regardless of the issue of whether or not residual agent is

1


-------
IT3 2005 Conference

45Lemieux

present in the BDR, disposal facilities must be able to operate within relevant permit restrictions
while processing BDR, and data gaps exist as to the behavior of BDR in high temperature
combustion devices. In addition, certain types of materials that are found in BDR, such as waste
carpeting, may be useful as auxiliary fuels for high temperature combustion devices, and similar
data gaps exist relative to operational and permit issues for the purposes of the use of those
materials for that purpose. The EPA instituted a pilot-scale test program to investigate issues
related to the thermal destruction of contaminated BDR (1) including carpeting, ceiling tile, and
wallboard.

In the US, approximately 2.2-2.7 billion kg (5-6 billion lbs) of carpet is sold annually, of which
60% is for replacement (2). In spite of considerable effort in the past decade to develop recycling
technologies for carpet wastes, most carpet continues to be disposed of in landfills (3). The
development of economically viable, environmentally sound, high volume, robust systems for
dealing with carpet waste would move the carpet industry closer to its goals of environmental
stewardship and protection. Since carpet has a heating value similar to that of coal, the
application of carpet as a fuel for high temperature combustion devices such as cement kilns or
boilers is potentially attractive, but there are potential environmental and operational issues that
need to be addressed in order to promote this as a viable practice for industry. For example,
some of the elemental components of carpeting (e.g., nitrogen) could potentially result in the
formation of pollutants of concern (e.g., nitrogen oxides [NOx]). In response to this data gap, the
US EPA performed testing on a pilot-scale rotary kiln, which showed only a slight increase in
NOx emissions from co-firing carpeting with natural gas (4). This study also showed only minor
increases in organic pollutants and no measurable emissions of mercury (Hg). A follow on study
showed that no other nitrogen-containing species such as NH3 or N20 could be accounting for
the fuel nitrogen, and that the burnout characteristics of the carpet were relatively independent of
the cut size of the carpet (5).

This paper describes experiments that were performed in a pilot-scale rotary kiln incinerator
simulator to evaluate the combustion characteristics of carpeting as a component of
BDR in an effort to aid in the selection of appropriate disposal facilities and to aid facilities in
maintaining permit compliance while processing potentially contaminated carpet. This paper
reports on emissions of polychlorinated dibenzo-/>dioxins and polychlorinated dibenzofurans
(PCDDs/Fs) while combusting carpet with and without a simulated decontamination chemical
(in the form of a 10% bleach solution) present on the carpet.

EXPERIMENTAL

Testing was performed at the EPA's Rotary Kiln Incinerator Simulator (RKIS) facility located in
Research Triangle Park, NC. The RKIS has been used in the past to test a wide variety of solid
and liquid wastes (4, 6, 7). The RKIS (shown in Fig. 1) consists of a 73 kW (250,000 Btu/hr)
natural gas-fired rotary kiln section and a 73 kW (250,000 Btu/hr) natural gas-fired secondary
combustion chamber (SCC). Following the SCC is a long duct that leads into a dedicated flue
gas cleaning system (FGCS) consisting of another afterburner, baghouse, and wet scrubber. The
RKIS is equipped with continuous emission monitors (CEMs) for oxygen (O2), carbon dioxide
(CO2), carbon monoxide (CO), nitrogen oxides (NOx), and total hydrocarbons (THCs). A series
of Type-K thermocouples monitor the temperature throughout the system. For the initial tests,

2


-------
IT3 2005 Conference

45Lemieux

the rotary kiln combustion air was flowing at a rate of 85.0 sm3/hr (3000 scfh) and the main
burner natural gas fuel was flowing at a rate of 5.66 srnVhr (200 scfh). The static pressure in the
rotary kiln section was maintained at -0.05 in. w.c. For the purposes of these tests, the SCC was
not operated, and the temperature in the transfer duct was maintained at approximately 300-350
°C (572-662 °F) to promote the formation of PCDDs/Fs so that the differences between the test
conditions could hopefully be maximized.

LNlulion Air

Kiln Sectior Transition Section

Fig. 1. Rotary Kiln Incinerator Simulator

A series of experiments were performed where approximately 0.45 kg (1 lb) bundles of 7.6 cm (3
in.) square pieces of carpeting, banded together using 1.3 cm (0.5 in) polypropylene straps, were
fed into the RKIS every 10 minutes over a 3-hour period, while samples were acquired to
measure PCDDs/Fs (8). The purpose of these tests were to evaluate the potential for combustion
of carpeting resulting from building decontamination operations to result in an increase in
PCDD/F emissions. The bundles of carpet were wetted with deionized water (approximately
50% of the total charge mass) in one set of experiments and they were wetted with a 10% bleach
(sodium hypochlorite) solution to a similar degree of wetness in another set of experiments. The
wetting was performed by quickly dunking the banded bundle of carpeting in a plastic bucket
containing the desired wetting agent, and allowing the bundle to drain until drops of liquid no
longer dripped from the bundle. The bundles of wetted carpet were then manually charged into
the RKIS using the ram feeder at 10-min intervals. Method 23 sampling was initiated at the time
of the first charging event and terminated 10 minutes after the final charging event, for a total of
3 hours of sampling. A 3-hour combustion blank test was also performed on the RKIS burning
natural gas only prior to the other testing. Analyses were performed using high-resolution gas
chromatography (HRGC) with high-resolution mass spectrometry (HRMS). The test matrix is
shown in Table I.

3


-------
IT3 2005 Conference

45Lemieux

Table I. Test Matrix

Run

Feed

Time Fed to Kiln

Dry Carpet Mass
(g)

Banded Carpet
Mass (g)

Water or 10%
Bleach Solution
Mass (g)

1

Natural Gas

-

-

-

-

2

Natural Gas
Carpet
DI Water













1134

453.4

478.7

429.3





1144

453.0

477.1

439.6





1155

448.5

472.1

367.9





1205

451.8

475.8

455.3





1215

458.3

483.8

515.2





1225

457.0

481.6

515.4





1235

451.0

474.6

452.4





1245

456.2

480.1

379.6





1255

453.8

477.2

419.3





1305

460.0

486.0

385.7





1315

450.0

476.3

460.5





1325

456.7

484.9

378.0





1335

456.1

482.8

410.9





1345

452.3

477.1

396.7





1355

456.9

482.5

426.7





1405

456.0

480.7

397.4





1415

455.1

479.9

385.2





1425

454.6

479.2

380.9

3

Natural Gas
Carpet
DI Water













1021

450.0

472.3

484.7





1031

459.0

481.3

509.4





1041

453.0

475.6

476.3





1051

454.0

476.7

459.7





1101

454.5

476.5

474.9





1111

458.7

480.8

459.9





1121

452.8

474.3

452.8





1131

457.2

478.9

436.7





1141

454.6

476.4

403.7





1151

451.8

474.3

447.8





1201

454.0

476.7

480.4





1211

454.7

476.6

444.9





1231

451.9

474.4

445.4





1241

459.4

482.3

482.9





1251

457.2

480.1

485.2





1301

459.0

481.7

410.7





1311

455.2

478.3

414.8





1321

451.4

475.2

485.8

4

Natural Gas
Carpet
10% Bleach













1055

455.2

478.5

531.7





1104

452.0

477.8

571.0





1114

454.0

476.8

555.4





1124

457.0

484.7

535.6





1134

453.7

482.0

473.7





1144

456.1

478.9

479.6





1154

448.8

474.6

502.8





1204

452.1

471.7

510.0





1214

449.0

470.1

472.7

4


-------
IT3 2005 Conference

45Lemieux

Run

Feed

Time Fed to Kiln

Dry Carpet Mass
(g)

Banded Carpet
Mass (g)

Water or 10%
Bleach Solution
Mass (g)





1224

455.3

475.0

511.3





1234

458.2

477.9

485.1





1244

451.3

469.8

476.1





1254

455.3

474.1

502.7





1304

457.4

475.0

498.2





1314

450.2

472.6

496.4





1324

452.8

475.4

506.8





1334

454.9

477.4

500.4





1344

454.1

476.5

443.0

5

Natural Gas
Carpet
10% Bleach













1202

457.3

480.1

462.5





1212

458.5

482.3

522.8





1222

457.7

482.1

493.4





1232

455.8

480.6

524.8





1242

453.2

477.2

355.8





1252

457.1

481.5

506.4





1302

450.6

474.6

470.1





1312

455.5

479.8

495.1





1322

455.0

478.1

468.3





1332

450.0

474.2

402.6





1342

451.8

475.5

487.7





1352

452.5

477.4

418.0





1402

458.6

483.6

460.6





1412

454.5

483.2

526.4





1422

453.0

480.4

388.9





1432

453.7

479.4

395.3





1442

454.5

479.0

388.8





1452

460.2

484.0

396.4

RESULTS

Table II lists the kiln temperature, the duct O2 and CO2 concentrations, and the flue gas
temperature and moisture conditions at the Method 23 sampling point. It must be noted
however, that due to the batch feed nature of the experiments, there were a series of transients in
gas species concentrations and temperatures associated with each charging event. Fig. 2 shows a
sample from one of the run days (Run 2) showing the O2 and CO2 concentrations, the CO
concentrations, and the kiln and duct temperatures. This shows the perturbations from baseline
conditions associated with the batch charging of carpet.

Table II. Operating Conditions

Run

Average O2
(% dry)

Average CO2
(% dry)

Average
Moisture
(%)

Average Kiln
Temperature
(°C)

Sampling

Temperature

(°C)

1

13.4

3.8

10.7

892

293

2

11.4

5.3

10.2

957

292

3

11.8

5.1

10.2

953

303

4

10.7

5.7

11.4

1000

291

5

11.5

5.3

9.51

1010

308

5


-------
IT3 2005 Conference

45Lemieux

02
C02

CO

i i i r

o

0
Q.

E
0

1000 -

800 -

600 -

400 -



Kiln T
Duct T

o
o

o
CO

o
o

o
CO

o
o

o
CO

o
o

o
CO

Time

Fig. 2. Sample CEM Traces from Run 2.

Table III lists the PCDD/F results from Runs 1 through 5, as well as the concentrations in terms
of the international toxic equivalency (TEQ) units (9). It must be noted that these data were
generated at conditions specifically intended to maximize formation of PCDDs/Fs, by not
operating the SCC of the RKIS facility, and by adjusting the transition duct temperature to be
within the optimal temperature window for PCDD/F formation (300-350 °C). In addition, the
samples were acquired prior to any flue gas cleaning devices. These experiments are not
designed to duplicate concentrations that might be seen in practice, but rather to elucidate
relevant qualitative trends that might be seen in practice. Although there were variations
between the duplicate runs, the variations between the duplicates were significantly less than the
variations between run conditions.

6


-------
IT3 2005 Conference

45Lemieux

Table III. PCDD/F]

Results (ng/dscm).

Analyte

Run 1

Run 2

Run 3

Run 4

Run 5

2,3,7,8-TCDD

0.000263

0.00158

0.000626

0.0369

0.00548

1,2,3,7,8-PeCDD

0.000294

0.00401

0.0012

0.222

0.0431

1,2,3,4,7,8-HxCDD

0.000228

0.00231

0.0012

0.354

0.0911

1,2,3,6,7,8-HxCDD

0.00119

0.003

0.00126

0.745

0.342

1,2,3,7,8,9-HxCDD

0.00119

0.00283

0.0012

0.689

0.361

1,2,3,4,6,7,8-HpCDD

0.0019

0.00828

0.00356

8.27

4.58

2,3,7,8-TCDF

0.000917

0.0302

0.0275

0.805

0.12

1,2,3,7,8-PeCDF

0.00119

0.0229

0.0093

1.49

0.232

2,3,4,7,8-PeCDF

0.00119

0.0641

0.0376

3.88

0.834

1,2,3,4,7,8-HxCDF

0.00119

0.0222

0.00915

3.49

0.759

1,2,3,6,7,8-HxCDF

0.00119

0.0242

0.00605

4.36

1

2,3,4,6,7,8-HxCDF

0.00119

0.0377

0.00903

12.5

4.17

1,2,3,7,8,9-HxCDF

0.000294

0.0124

0.00243

4.19

0.914

1,2,3,4,6,7,8-HpCDF

0.000349

0.0319

0.00915

24.1

7.54

1,2,3,4,7,8,9-HpCDF

0.000349

0.00933

0.002

10.7

3.22

PCDD/F I-TEQ (ND=0; EMPC=0)

0.000147

0.0509

0.0251

5.37

1.41

PCDD/F I-TEQ (ND=0; EMPC=EMPC)

0.000152

0.0509

0.0255

5.37

1.41

PCDD/F I-TEQ (ND=DL/2; EMPC=0)

0.00101

0.0509

0.0253

5.37

1.41

PCDD/F I-TEQ (ND=DL/2; EMPC=EMPC)

0.00101

0.0509

0.026

5.37

1.41

PCDD/F I-TEQ (ND=DL; EMPC=EMPC)

0.00187

0.0509

0.0265

5.37

1.41

Mono-Di~Tri-CDDs

0.0564

0.13

0.0817

6.48

0.145

TCDDs

0.0506

0.0916

0.0771

9.98

0.354

PeCDDs

0.00573

0.0461

0.0161

9.86

1.03

HxCDDs

0.00423

0.0326

0.0113

14.7

3.83

HpCDDs

0.00432

0.0167

0.00699

15.9

8.56

OCDD

0.00499

0.0117

0.00831

12.3

6.3

Mono-Di-Tri-CDFs

0.118

2.4

0.506

130

5.72

TCDFs

0.0204

0.847

0.241

46.5

5.48

PeCDFs

0.0044

0.585

0.208

43.9

10.2

HxCDFs

0.0037

0.273

0.0718

59.3

17.1

HpCDFs

0.00485

0.0648

0.0171

71.8

24.7

OCDF

0.00461

0.0165

0.00759

62.7

19.2

Fig. 3 shows the total PCDD/F emissions and Fig. 4 shows the PCDD/F emissions in terms of
the International Toxicity Equivalency (I-TEQs) for the 3 conditions with results from duplicate
conditions being averaged. The combustion of the wetted carpet resulted in PCDD/F emissions
only slightly higher than the natural gas blank, in terms of both the total PCDDs/Fs and TEQs.
Addition of the 10% bleach solution, however, resulted in a significant increase of PCDD/F
concentrations both in terms of the total PCDDs/Fs and the TEQs. This suggests that
combustion facilities that may process BDR that has been decontaminated with a chlorinated
decontamination agent should be aware of the potential for increased emissions of PCDDs/Fs. In
general, good combustion practices including minimization of CO emissions and operating the
flue gas cleaning equipment at temperatures below 250 °C will effectively minimize emissions
of PCDDs/Fs (10). Other operating practices such as rapidly quenching the flue gases so that
they spend as little time as possible in the PCDD/F formation temperature window will also
minimize emissions of PCDDs/Fs (11). There has been no conclusive study to implicate water
concentration in the formation mechanism, probably because at flue gas conditions, water is
present in high concentrations.

7


-------
IT3 2005 Conference

45Lemieux

I Blank

I Carpet/DI Water
I Ca rpet/Bleach

Total PCDDs (ND=0; EMPC=0)

Total PCDFs (ND=0; EMPC=0)

Fig. 3. Total PCDD/F Emissions

I Blank

I Carpet/DI Water
I Carpet/Bleach

Condition (ND=0; EMPC=0)

Fig. 4. PCDD/F I-TEQ Emissions

Fig. 5 shows the normalized distribution of the homologue groups. In all cases the furans were
present at higher concentrations than the similarly chlorinated dioxin species. The homologue
distribution for the natural gas combustion blank is heavily weighted towards the mono-tri
chlorinated dioxins and furans, with some tetra-substituted dioxins contributing to the
distribution. Most of the higher chlorinated species were present at very low levels relative to
the lower chlorinated species. The samples with the carpet/DI water conditions showed a greater
diversity in homologue groups present in significant concentrations, although the distribution
was monotonically decreasing with higher degree of chlorination. The samples with the
carpet/bleach conditions showed a relatively flat homologue distribution. The octa-chlorinated
species were present at levels similar to the lower chlorinated species.

8


-------
IT3 2005 Conference

45Lemieux

0.6

0.5

0.4

0.3

0.2

0.1















II II1

1 > 1 1

I1VRHI r



O

o

u

Q
Q

U

Q
Q

U

Q
O

U

Q
Q

U

O
Q

U
O

Q
U

Q

U

Q

U

O

U
X
X

Q
U

Q
U
O

I Blank

I Carpet/DI Water
Carpet/Bleach

Homologue Group

Fig. 5. PCDD/F Homologue Group Distribution.

Fig. 6 shows the normalized distribution of the isomers (the dioxin "fingerprint"). Of the species
with the toxic 2,3,7,8 substitutions, there were significant qualitative differences between the
fingerprints of the 3 test conditions. The toxic species associated with the combustion blank
were almost totally contained in the octa-substituted dioxin and furan. Note that of the
homologue groups, the combustion blank showed very low relative concentrations of the higher
substituted species. This helps explain why the TEQ levels were very low for the combustion
blank. The test condition with the carpet and deionized water showed a significant increase in
the 2,3,7,8-TCDF and 2,3,4,7,8-PeCDF. These 2 species have a fairly high toxicity equivalency
factor (TEF) which explains the significant increase in the TEQ emissions when compared to the
combustion blank. The distribution for the conditions with the bleach solution showed a
significant contribution from the higher substituted species, both dioxins and furans.

9


-------
IT3 2005 Conference

45Lemieux

0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05

o
o

u

o

Q

U

Q
O

U

Q
Q

U

I

.1.

771

u

I Blank

I Carpet/DI Water
Carpet/Bleach

ii1 tiiin n

Q
Q

U

Q
Q

U

Q
O
U
O

Q
U

O

U

Q

U

Q

U
X
X

Q

U
X
X

o
u

X
X

o
u

X
X

o
u

Q

u

Q
U
O

CONCLUSIONS

Isomer

Fig. 6. PCDD/F Isomer Distribution

Testing was performed on a pilot-scale rotary kiln incinerator simulator to evaluate the potential
for formation of PCDDs/Fs from the combustion of carpeting. Five runs were performed at three
test conditions (combustion blank, carpet wetted with deionized water, and carpet wetted with a
10% bleach solution). In order to maximize the differences in measured emissions between the
various run conditions, the combustor was operated in such a way as to maximize formation of
PCDDs/Fs. In all conditions, emissions of the furan species were higher than emissions of the
analogous dioxin species. Emissions of PCDDs/Fs in the combustion blank were extremely low,
and exhibited a homologue distribution heavily favoring the lower chlorinated species, although
the majority of the toxic 2,3,7,8-substituted isomers were the octa-chlorinated dioxin and furan
species. Emissions of PCDDs/Fs from the carpet wetted with deionized water were somewhat
higher than the combustion blank, but were still low. The homologue distribution for the
deionized water condition was monotonically decreasing as degree of chlorination increased,
although unlike the combustion blank, there were significant levels of the higher chlorinated
species. The condition with the deionized water showed a significant increase in the 2,3,7,8-
TCDF and the 2,3,4,7,8-PeCDF isomers. The homologue distribution for the test condition with
the bleach solution was flat, and the contribution of the octa-chlorinated species to the TEQs was
significant. These results suggest that although combustion of clean carpet is not likely to
increase emissions of PCDDs/Fs from solid fuel-burning facilities, the combustion of carpeting

10


-------
IT3 2005 Conference

45Lemieux

that has been decontaminated with a chlorinated decontamination agent such as bleach, may
require care to prevent an increase in PCDD/F emissions.

ACKNOWLEDGMENTS

The authors would like to thank Richie Perry and Steve Terrll of ARCADIS and Marc Calvi of
EPA/NRMRL for their invaluable help in making these tests happen. The authors would also
like to thank John Conyers and Steve Bradfield of Shaw Industries for providing the carpet
samples.

REFERENCES

1.	Lemieux, P. (2004), "EPA Safe Buildings Program: Update on Building Decontamination
Waste Disposal Area," EM, Vol. 29-33.

2.	Statistical Report, 2001, Floor Covering Weekly 50 (18).

3.	US EPA, June 2002. Municipal Solid Waste in the United States: 2000 Facts and Figures,
Office of Solid Waste and Emergency Response, EPA530-R-02-001. Washington, D.C., 2002.

4.	Lemieux, P.; Stewart, E.; Realff, M.; Mulholland, J.A. (2004), "Emissions Study of Co-
firing Waste Carpet in a Rotary Kiln," Journal of Environmental Management, Vol. 70, pp. 27-
33.

5.	Realff, M., Lemieux, P., Lucero, S., Mulholland, J., and Smith, P., "Characterization of
Transient Puff Emissions from Burning of Carpet Waste Charges in a Rotary Kiln Combustor,"
Paper to be presented at the IEEE 47th Cement Industry Technical Conference, Kansas City,
MO, May 15-20, 2005.

6.	Stewart, E.S.; Lemieux, P.M., "Emissions from the Incineration of Electronics Industry
Waste," IEEE International Symposium on Electronics and the Environment & the IAER
Electronics Recycling Summit Electronics Goes Green 2003 International Congress and
Exhibition: Life-Cycle Environmental Stewardship for Electronic Products, Boston, MA, May
19-22, 2003.

7.	Lemieux, P.M.; Stewart, E.S. (2004), "A Pilot-Scale Study of the Precursors Leading to
the Formation of Mixed Bromo-Chloro Dioxins and Furans," Environmental Engineering
Science, Vol. 21, pp. 3-9.

8.	U.S. EPA, 1991, EPA Test Method 23, "Determination of Poly chlorinated Dibenzo-p-
dioxins and Poly chlorinated Dibenzofurans from Stationary Sources" in Code of Federal
Regulations, Title 40, Part 60, Appendix A, U.S. Government Printing Office, Washington, DC,
July 1991.

9.	US EPA, 1998. The inventory of sources of dioxin in the United States, Review Draft,
EPA/600/P-98/002Aa, Washington DC, April 1998.

11


-------
IT3 2005 Conference

45Lemieux

10.	U.S. EPA, 1996 Proposed Rule-Hazardous Waste Combustors; Maximum Achievable
Control Technologies Performance Standards (Performance Specifications). Fed. Regist. 1996,
61, 17499-17502.

11.	Rigo, G.H.; Chandler, A.J.; Lanier, W.S. The Relationship between Chlorine in Waste
Streams and Dioxin Emissions from Waste Combustor Stacks; ASME Research Report CRTD;
American Society of Medical Engineers: New York, 1996; Vol. 36.

12


-------