United States Air and Energy Engineering EPA-600/8-89-063
Environmental Protection Research Laboratory August 1989
Agency Research Triangle Park, NC 27711
Research and Development
N>EPA Municipal Waste
Combustion
Assessment:
Technical Basis For
Good Combustion
Practice
Prepared For
Office of Air Quality Planning and Standards
Prepared By
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
This document is printed on recycled paper
-------
EPA-600/8-89-063
August 1989
MUNICIPAL WASTE COMBUSTION ASSESSMENT:
TECHNICAL BASIS FOR GOOD COMBUSTION PRACTICE
Prepared by
P.J. Schindler
L.P. Nelson
Energy and Environmental Research Corporation
3622 Lyckan Parkway. Suite 5006
Durham, NC 27707
Under Contract No. 68-03-3365
Work Assignment No. 1-05
EPA Project Officer James D. Kilgroe
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
Prepared for
U.S. Friv •• onmen le i P-otection Agency
•Jf*:ct . Kese.ir : '•.'.•> Development
-------
REVIEW NOTICE AND DISCLAIMER
The information in this document has been funded wholly by the United
States Environmental Protection Agency under Contract No. 68-03-3365 to Energy
and Environmental Research Corporation. It has been subject' to the Agency's
peer and administrative review (by both the Office of Research and Development
and the Office of Air Quality Planning and Standards), and it has been
approved for publication as an Agency document. Mention of trade names or
commercial products does not constitute endorsement or recommendation of a
commercial product by the Agency.
-------
ABSTRACT
The EPA's Office of Air Quality Planning and Standards (OAQPS) is
developing emission standards and guidelines for, respectively, new and
existing MWCs under the authority of sections lll(b) and lll(d) of the Clean
Air Act (CAA). The EPA's Office of Research and Development (ORD) is
providing support in developing the technical basis for good combustion
practice (GCP), which is included in the regulatory alternatives considered in
selecting the proposed standards and guidelines. This report defines GCP and
summarizes the approach used to implement GCP into the proposed MWC standards
and guidelines. The report identifies the minimum subset of GCP operating
parameters that can be continuously monitored to ensure that the goals of GCP
are achieved. Finally, the report provides a detailed description of the data
and rationale used to establish quantitative operating limits for each of the
continuous operating parameters.
-------
FOREWORD
Based upon its analysis of Municipal Waste Combustors (MWCs), the
Environmental Protection Agency (EPA) has determined that MWC emissions may
reasonably be anticipated to contribute to the endangerment of public health
and welfare and warrant further regulation. As a result, EPA's Office of Air
Quality Planning and Standards is developing emission standards for new MWCs
under Section lll(b) of the Clean Air Act (CAA) and guidelines for existing
MWCs under Section lll(d) of the CAA.
In support of these regulatory development efforts, the Air and Energy
Engineering Research Laboratory in EPA's Office of Research and Development
has conducted an assessment of combustion control practices to minimize air
emissions from MWCs. The results of'this assessment are documented in the
following reports:
Municipal Waste Combustion Assessment: Combustion Control at New
Facilities, August 1989 (EPA-600/8-89-057)
Municipal Waste Combustion Assessment: Combustion Control at
Existing Facilities, August 1989 (EPA-600/8-89-058)
Municipal Waste Combustion Assessment: Fossil Fuel Co-Firing,
July 1989 (EPA-600/8-89-059)
Municipal Waste Combustion Assessment: Waste Co-Firing, July 1989
(EPA-600/8-89-060)
Municipal Waste Combustion Assessment: Fluidized Bed Combustion.
July 1989 (EPA-600/8-89-061)
Municipal Waste Combustion Assessment: Medical Waste Combustion
Practices at Municipal Waste Combustion Facilities, July 1989 (EPA-
600/8-89-062)
Municipal Waste Combustion Assessment: Technical Basis for Good
Combustion Practice. August 1989 (EPA-600/8-89-063)
Municipal Waste Combustion, Multi-Pollutant Study, Emission Test
Report, Maine Energy Recovery Company, Refuse-Derived Fuel
Facility, Biddeford, Maine. Volume I, Summary of Results, July
1989 (EPA-600/8-89-064a)
Municipal Waste Combustion, Multi-Pollutant Study, Emission Test
Report, Mass Burn Refractory Incinerator, Montgomery County South,
Ohio, Volume I, Summary of Results, August 1989 (EPA-600/8-89-
065a)
-------
The lll(b) New Source Performance Standards (NSPS) and the lll(d)
Emission Guidelines for new and existing municipal waste combustors (MWCs)
include two specific emission control strategies: combustion controls, or
good combustion practice (GCP), and add-on flue gas cleaning controls. The
specific objectives of this report, "Municipal Waste Combustion Assessment:
Technical Basis for Good Combustion Practice," are to identify the combustor
design and operating parameters that are necessary components of GCP, to
provide the rationale for designating specific components as continuous
compliance parameters in the standards and guidelines, and to present the data
and supporting rationale used to select operating ranges and limits for the
continuous compliance parameters.
111
-------
CONTENTS
SECTION PAGE
1.0 SUMMARY 1-1
2.0 GOOD COMBUSTION PRACTICES 2-1
2.1 Waste Feeding 2-1
2.2 Adequate Combustion Temperature 2-3
2.3 Amount and Distribution of
Combustion Air 2-5
2.4 Mixing : 2-5
2.5 PM Carryover 2-7
2.6 Downstream Temperature Control 2-7
2.7 Combustion Monitoring and Control 2-8
3.0 SELECTION OF GCP OPERATING PARAMETERS TO BE
CONTINUOUSLY MONITORED 3-1
4.0 DETERMINATION OF COMBUSTOR-SPECIFIC CONTINUOUS
COMPLIANCE REQUIREMENTS 4-1
4.1 Carbon Monoxide in Flue Gases 4-1
4.1.1 CO Emission Data Analysis 4-3
4.1.1.1 Long Term CEM Data Analysis
Procedures 4-4
4.1.1.2 Averaging Time 4-5
4.1.1.3 Categorization of MWC
Technologies 4-6
4.1.2 Mass Burn Waterwall MWCs 4-6
4.1.2.1 Millbury Long Term CEM Program 4-7
4.1.2.2 Millbury Compliance Test Data 4-9
4.1.2.3 Commerce, CA Long Term CEM Data 4-11
4.1.2.4 Long Term Data Analysis
Conclusions 4-15
4.1.2.5 Review of Additional Short Term
CO Data 4-16
4.1.2.5.1 Quebec City, Quebec
Combustion Retrofit
Program 4-18
4.1.2.5.2 Hampton, Virginia
Combustion Retrofit
Program 4-21
4.1.2.5.3 Commerce, California
Overfire Air
Optimization Tests 4-24
4.1.2.5.4 Conclusions 4-27
4.1.3 Modular Starved Air MWCs 4-27
4.1.3.1 Oswego County, NY Long Term CEM
Data 4-28
4.1.3.2 Oswego County, NY Parametric Test... 4-29
4.1.3.3 Prince Edward Island (PEI)
Parametric Test 4-31
4.1.3.4 Red Wing, MN Compliance Test 4-31
4.1.3.5 Conclusions 4-31
-------
4.1.4 RDF Spreader Stoker MWCs 4-33
4.1.4.1 Mid-Connecticut Long Term CEM Data.. 4-34
4.1.4.2 Penobscot, ME Long Term CEM Data 4-35
4.1.4.3 RDF Compliance Test Results 4-39
4.1.4.4 Mid-Connecticut Parametric Test .... 4-41
4.1.4.5 Conclusions 4-48
4.1.5 Mass Burn Refractory Wall MWCs 4-48
4.1.5.1 Dayton, OH Parametric Test 4-49
4.1.5.2 Mass Burn Refractory
Compliance Test Results 4-51
4.1.5.3 Conclusions 4-52
4.1.6 Mass Burn Rotary Waterwall MWCs 4-53
4.1.6.1 Dutchess County, NY Compliance
Test ' 4-53
4.1.6.2 Conclusions 4-54
4.1.7 Modular Excess Air MWCs 4-54
4.1.7.1 Pittsfield. MA Parametric Test 4-56
4.1.7.2 Pope/Douglas County, MN
Compliance Test 4-56
4.1.7.3 Conclusions 4-56
4.1.8 Fluidized Bed Combustors 4-58
4.1.8.1 Western Lake Superior Sanitary
District (WLSSD), Duluth. MN 4-58
4.1.8.2 Northern States Power French
Island Facility, La Crosse, WI 4-59
4.1.8.3 Sundsvall, Sweden CFB Test
Program 4-61
4.1.8.4 Conclusions 4-61
4.2 Operating Load 4-61
4.3 Downstream Temperature Control 4-66
5.0 REFERENCES 5-1
APPENDIX A EXPECTED EXCEEDANCE LEVELS FOR CO DATA
FROM MUNICIPAL WASTE COMBUSTION FACILITIES A-l
-------
FIGURES
2-1 Effect of Waste Feed on CO from a 2-4
Circulating Fluidized Bed Combustor
2-2 Relationship Between Excess 02 and CO 2-6
4-1 Millbury Long Term CO Emissions 4-8
4-2 Millbury CO Data Plots 4-10
4-3 Effect of Steam Load on CO Emissions 4-12
from a Mass Burn Waterwall MWC
4-4 Commerce Long Term CO Emissions 4-13
4-5a Quebec MWC - Unmodified Design 4-19
4-5b Quebec MWC - Modified Design 4-19
4-6 Effect of Overfire Air on CO Emissions 4-26
from a Mass Burn Waterwall MWC
4-7 Mid-Connecticut Long Term CO Emissions - 4-36
Unit 11
4-8 Mid-Connecticut Long Term CO Emissions - 4-37
Unit 12
4-9 Mid-Connecticut Long Term CO Emissions - 4-38
Unit 13
4-10 Penobscot Long Term CO Emissions 4-40
4-11 Mid-Connecticut CO Emissions - PT8 4-44
4-12 Mid-Connecticut CO Emissions - PT9 4-45
4-13 Mid-Connecticut CO Emissions - PT11 4-46
4-14 CO Versus 02 - Mid-Connecticut - PT9 4-47
4-15a Dutchess County CO Emissions - Unit 1 4-55
4-15b Dutchess County CO Emissions - Unit 2 4-55
4-16 Fluidized Bed Combustor - CO Emissions 4-60
4-17 Effect of Feed Rate on Temperature for 4-63
a Modular Excess Air MWC
4-18 Effect of Operating Load on CO Emissions 4-64
from a Mass Burn Waterwall MWC
4-19 Relationship Between PM Carryover and 4-65
CDD/CDF Emissions
4-20 CDD/CDF Removal Efficiency as a Function of 4-68
ESP Inlet Temperature
VI
-------
TABLE
TABLES
1-1
2-1
4-1
4-2
4-3
4-4
4-5
4-6
4-7
4-8
4-9
A-l
A-2
A-3
A-4
GCP Operating Standards for MWCs 1-2
GCP Components and Objectives 2-2
Millbury Compliance Data 4-10
Mass Burn Waterwall MWCs - Short 4-17
Duration Test Emissions Summary
Summary of Emissions From Quebec City 4-22
MWC Performance Tests
Test Matrix for Commerce, CA MWC 4-25
Combustion Optimization Program
Oswego County. NY Parametric Test Results 4-30
PEI Parametric Test Results 4-32
Mid-Connecticut Performance Test - Summary 4-42
South Dayton MWC Parametric Test - Summary 4-50
of CO Emission Concentrations
Pittsfield. MA Parametric Test Results 4-57
Millbury MWC - Expected Exceedance Rates A-4
Penobscot MWC - Expected Exceedance Rates A-5
Mid-Connecticut Unit 12 - Expected A-6
Exceedance Rates
Mid-Connecticut Unit 13 - Expected A-7
Exceedance Rates
VI 1
-------
1.0 SUMMARY
The objectives of this report are to define good combustion practices
(GCP) as they apply to prevention and control of air pollutant emissions from
municipal waste combustors (MWCs), to summarize the approach used to implement
GCP in the MWC standards and guidelines, and to provide the rationale and
supporting data used to establish numeric GCP operating limits in the
regulation. GCP are defined as the MWC system design and operating techniques
which, when applied with appropriate flue gas cleaning techniques, are
expected to minimize trace organic emissions. The GCP control strategy
includes a number of combustor conditions which are applied collectively to
achieve this goal. Because of the interrelationship among these conditions,
it is possible to provide verification of GCP on a continuous basis by
monitoring a select subset of combustor operating parameters.
Based on this rationale, the MWC regulation establishes numeric limits
for three specific combustor operating parameters:
• CO i n f1ue gases
• Maximum operating load
• PM control device flue gas temperature
Table 1-1 summarizes the recommended operating limits for each of the three
combustor operating parameters. The CO emission limits vary according to
combustor technology. The operating load limits apply only to those MWCs that
generate steam. All MWCs are subject to the PM control device flue gas
temperature limits. These three parameters were selected for inclusion in the
regulation for two reasons. First, each can be monitored on a continuous
basis to provide verification of good combustion conditions. Secondly, the
parameters as a group comprise the minimum set of combustor operating
conditions that can be related either directly or indirectly to the GCP
components that must be addressed in order to satisfy three broad objectives
of GCP:
• Maximize in-furnace destruction of organics
• Minimize particulate matter (PM) carryover out of the furnace
• Minimize low temperature reactions which promote formation of
polychlorinated dibenzo-p-dioxins and dibenzofurans (CDD/CDF)
1-1
-------
ct:
CD
oo
a
OL
Q
z
<
h-
OO
Q.
CD
*
O)
Cn
O3
S-
CVJ 01
0 >
O3
1^ NX
U
+-> O
03 i —
-Q
E
Q. S-
Q- =3
O
i
£_
3
O
>-
C3
O
1
O
z
UJ
1—
O
3
s:
CDOOCDCDCDOO
oOLOLomLOLncD
i — 1 T— 1 « — t *— 1 i — 1 i— 1
r-
i — O3 (/)
03 S £-
3 s- 0
OJ 4_>
>^ _l_i (/j
i — £- O3 t- t- 3
I O 3 -r- S_ Ol -Q
03 4-> •<•!-.* E
5 0 >> •< O 0
£- 03 S_ T3 4-> C_>
Ol S— 03 Ol co t/)
4-> t- 4-> > CO T3
(DOIO£-O1£-"OO1
±> X -<3 S-
C C t= OO UJ ro-i— "O
t- t- I- Ol U. O)
=3 3 3 t- t- t- i M
CQCOCa O3 O3Q.O-r-
r— i — oo <_> -a
CO CO CO 3 3 -i—
CO CO CO "O "O U_ U_ 3
O3 03 O3 O O ' ' ' * r^—
z""Bfifl£u-
co
.tf
E
•i —
1
c
o
• 1 —
CO
CO
'1 —
E
UJ
O
0
ai
CD cn
CD 03
r-H t-
Ol
•a >
Ol 03
Ol
o ^.
X 0
Ol O
,
4-> -Q
o
c: t-
^
i— O
1 — t~
03 1
J^ £-
CO 3
0
oi
i— -a
0 O
J3 r—
>> E
t— fO
01 ai
O CO
o
Ol C
£- cn
• 1 —
4-» CO
O3 Ol
01 -a
j=
M-
x: o
•c- 4J
2 c
01
CO O
o s-
3 01
2: Q.
"O
O3
O
_J
cn
c
• r-
4_)
03
t_
Ol
Q-
O
Ol
cn
O3
t-
Ol
>
03
-^
O
O
i —
-Q
t_
O
.C
1
£_
O
4—
*
E
3
E
•i —
X
O3
E
1 1
0
O
Lf)
-j
O
CM
CO
CM
"o
O
^:
0.
03
Ol
t_
^
03
t_
01
Q.
E +->
oi ai
I— r-
t~
co 1—1
03
CJ3 O>
O
Ol ••-
3 >
i — Ol
u_ a
-a
o
s_
o>
CL
CO
Ol
4) --v
O) •
(O E
£_
Ol 03
>
03 O
O
-C O
i +->
Ol
C +->
O J^:
Cn
o -a
01 ••
3 C\J
03 -
> .
cn
c •
03 O)
Ol -^
E
CO
Ol CD
-e E
co
O3 T3
Ol
T3 +->
Ol 03
c c
•i- cn
4- -I-
Ol CO
-a oi
-a
CO
•i- 4-3
03
Ol
cn co
03 T3
t- c
01 O)
>
03 T3
c
.*: 03
O
O CO
•— Ol
jQ O
c
i. O)
3 E
O E
JZ O
i O
£_
3 .C
O (J
4- -i-
1-2
-------
This report is organized as follows. Section 2 summarizes the potential
sources of trace organic emissions from MWCs. identifies the components of
GCP, and describes the interrelated roles of the components in the GCP control
strategy. Section 3 rationalizes how adherence to the CO, load, and
temperature limits ensures continuous achievement of GCP. Section 4 provides
the rationale and supporting data used to establish the achievabi1ity of the
numeric limits for the GCP operating parameters.
1-3
-------
2.0 GOOD COMBUSTION PRACTICES
The principles of good combustion have been incorporated to varying
degrees to provide the basis for improved efficiency and emissions performance
in all types of combustion systems, including MWCs. The application of these
principles to MWCs is intended to meet a specific goal: minimization of trace
quantities of potentially toxic organic emissions. Organic emissions are
products of incomplete combustion which result from the inability of MWCs to
achieve ideal combustion conditions. There are three potential sources of
organic emissions. First, organics are present in waste feed and can
potentially pass through the combustor undestroyed. Secondly, organics can be
formed from precursor compounds that evolve in the combustion process.
Finally, some organics (specifically CDD/CDF) can form as a result of
reactions that occur in low temperature regions of the MWC system. Additional
detailed discussion of each of these organic emission sources is included in
other reports published by EPA on GCP.1-2-3
The GCP control strategy is developed based on the need to minimize the
occurrence of all three sources of organic emissions. This is accomplished by
identifying a group of combustor design and operating conditions, or
components, which will achieve three broad GCP goals:
• Maximize in-furnace destruction of organics
• Minimize PM carryover
• Minimize the occurrence of low temperature CDD/CDF formation
reactions
Seven GCP components have been identified which collectively address the three
GCP goals. Each component provides a necessary contribution to the GCP
control strategy, and each can be directly related to one of the three GCP
goals. The following sections describe the role of each component in the
overall GCP strategy. Table 2-1 summarizes the objectives of each component.
2.1 Waste Feeding
Combustion stability can be affected significantly by sudden changes in
waste composition or feed characteristics (e.g., moisture content, heating
value, volatiles content). Excursions in waste feed rates or waste properties
can deplete local oxygen levels in the furnace, allowing organics to escape
2-1
-------
Table 2-1. GCP COMPONENTS AND OBJECTIVES
COMPONENT
OBJECTIVE
Waste feed control
Temperature at fully mixed
1ocation
Amount and distribution of
combustion air
Mixing
PM carryover
Downstream temperature control
Monitoring and control
o Avoid combustion instabilities
o Ensure destruction of gas-phase
organics
o Maximize particle burnout
o Provide proper local stoichio-
metries in waste drying,
ignition, and burnout zones
o Eliminate fuel-rich zones and
avoid quenching
o Provide proper excess air
margins
o Minimize PM entrainment
o Ensure availability of oxidant
to all organic material within
high temperature region
o Minimize escape of organics and
metals
o Minimize downstream formation of
organics
o Ensure that all GCP objectives
are continuously met
2-2
-------
complete oxidation. Alternately, combustion of low heating value fuel can
lead to quenching of combustion reactions, which also allow organics to escape
thermal destruction. Both of these conditions are accompanied by elevated
concentrations of CO. Therefore. MWCs should be designed and operated in a
manner that minimizes occurrence of waste feed excursions. Flue gas CO
concentrations can be monitored to provide a continuous indication of proper
waste feed conditions. The appropriate design and operating procedures to
minimize waste feed excursions vary according to combustor technology, and may
include measures such as fuel blending by the crane operator, use of ram
feeders, and use of air preheat.
For example, the importance of optimizing waste feed is well
demonstrated by two data sets generated at a circulating fluidized bed RDF
facility in Sweden, shown in Figure 2-1.4 The top data set (Figure 2-la)
indicates continuous CO emissions measured in the stack while firing an RDF
that was processed with a single-stage shredder. There are frequent CO spikes
in excess of 2000 ppmv during a period of just over 1 hour. The second test
was conducted while firing RDF which had been processed with two stages of
shredding, resulting in a more uniform waste feed with smaller nominal
particle size. The change in waste size distribution was the only difference
between the two tests. The CO data for the more uniform feed conditions
(Figure 2-lb) shows that the majority of spikes are eliminated and combustion
conditions are far more stable, resulting in a lower potential for trace
organic emissions.
2.2 Adequate Combustion Temperature
The ability of an MWC to achieve combustion temperatures that are
adequate to destroy organics is a fundamental requirement of GCP. The
occurrence of spatial and temporal temperature variations during normal
operating conditions necessitates that MWCs be designed and operated in a
manner that will ensure that all combustion products are exposed to the
minimum destruction temperature. Residence time is also an important
requirement to ensure that burnout of gaseous and solid phase organics occurs
in the furnace. Failure to achieve the necessary temperatures and residence
times will result in the escape of organics from the furnace, which will lead
to elevated concentrations of CO in flue gases.
2-3
-------
CO
2000 ppm
0
2000 ppm
A. Single-stage shredding.
CO
2000 ppm
30
0
MIN
2000 ppm
0
B. Uith secondary shredding.
Figure 2-1. Effect of Waste Feed on CO Emissions
From a Circulating Fluidized Bed Combustor
2-4
-------
2.3 Amount and Distribution of Combustion Air
The proper amount and distribution of combustion air are essential for
efficient combustion. The amount of excess air introduced to the combustor
must be sufficiently high to minimize the existence of fuel-rich pockets and
sufficiently low to avoid quenching of combustion reactions. The range of
excess air levels that will satisfy these objectives varies for each combustor
technology. Figure 2-2 illustrates the relationship between excess 02 and CO
emissions. CO emissions typically increase when insufficient 02 is available
to complete combustion, or when excessive amounts of 02 quench combustion
reactions. High excess air levels can also cause excessive PM carryover,
which can include adsorbed organics. A key objective of GCP is to ensure that
MWCs operate at excess air levels that fall in the trough of this U-shaped
curve.
Total combustion air is typically split between primary and secondary
supplies. The amount and distribution of primary (underfire) air controls bed
burning stoichiometry and waste burnout. Secondary air is used to adjust
local stoichiometries to levels needed to achieve complete combustion, to
control flame height, and to complete the mixing process (see Section 2.4).
Failure to distribute combustion air in the correct proportions to primary and
secondary supplies can result in elevated organics and CO emissions and
excessive PM carryover. The GCP recommendations for combustion air control
must also be applied in a technology-specific manner according to the design
objectives of each combustor type. Results from several full scale tests
which provide evidence of the effects of the amount and distribution of
combustion air on emissions are summarized in Chapter 4.
2.4 Mi xi nq
Mixing of combustion products and air is a key requirement to ensure
destruction of organics in the furnace. The most common method used to
achieve good mixing in MWCs is properly designed and operated secondary
(overfire) air systems. Mixing can also be effected through combustor design
measures such as bull noses, baffles, turns, and changes in duct shape or
cross-sectional area. Mixing failures are accompanied by spikes or bulk
increases in CO emissions. Several full scale tests provide data which
2-5
-------
I—I CD
OCC
ZLU
OCJ
CDZ
Q; CD
-»•*-
OXYGEN CONCENTRATION
A - INSUFFICIENT AIR C+|02—-CO
B - APPROPRIATE OPERATING REGION
C - "COLD BURNING"
Figure 2-2. Relationship Between Excess 0? and CO
2-6
-------
document the effect of a change in mixing design on CO emissions. These are
discussed further in Section 4.1.
2.5 PM Carryover
PM carryover is defined as the amount of entrained particulate matter,
both organic and inorganic, which is carried out of a combustor with flue
gases. The amount of PM carryover from MWCs can affect the ultimate emission
of metals and organic pollutants. In addition, correlations have been
established between the amount of PM emitted from MWC boilers and CDD/CDF
emission levels in the stack.5 Therefore, minimization of PM carryover is a
necessary requirement of GCP.
Parameters affecting PM carryover include operating load, excess air
levels, and primary/secondary air ratios. Combustor specific design features
can also influence rates of PM carryover. For example, RDF spreader stokers
exhibit higher PM carryover than many other designs due to the use of a semi-
suspension firing mode. Conversely, starved air modular MWCs are designed
with low air flows and velocities in the primary combustion chamber, and
typically exhibit relatively low PM carryover. Operating load limits can be
established to minimize PM carryover within the constraints of specific
combustor design characteristics. A more extensive discussion of test data
correlating PM carryover with emissions of organics is included in Section
4.2.
2.6 Downstream Temperature Control
Several full scale emission testing programs have produced data
indicating that, under certain operating conditions, CDD/CDF concentrations in
MWC flue gases increase between the combustor exit and the stack.6.7,8,9 This
low temperature CDD/CDF formation has also been investigated in bench scale
experiments, and results indicate the existence of formation reactions in a
temperature range of approximately 390-750°F (200-400°C).10,n This
temperature range typically occurs in the flue gas cleaning equipment of many
MWCs. Discovery of the low temperature reaction has created the need for
inclusion of the third GCP goal: minimization of conditions which promote low
temperature CDD/CDF formation. The recommended control strategy is to
minimize the retention time of flue gases and particulate concentrations in
the temperature window where CDD/CDF formation is maximized. The data leading
2-7
-------
to specification of the operating limits that achieve this goal are discussed
in detail in Section 4.3.
2.7 Combustion Monitoring and Control
Inclusion of the proper monitoring and control features in the design of
the combustion system is a vital part of GCP. A number of combustor operating
parameters can be incorporated into the combustion control network, including
steam flows, temperatures and pressures, air flows and distributions,
combustor operating temperatures, and flue gas oxygen concentrations.
Operating parameters are monitored either for purposes of verifying emission
performance or to maintain continuous operational stability. Three specific
parameters have been identified which must be monitored as GCP regulatory
requirements (CO in flue gases, operating load, and PM control device
temperature), because they provide direct or indirect verification of good
combustion conditions. Additional technology specific operating parameters
may be monitored to maintain operational stability and provide the operator
with information that can be used to control the combustion process.
2-8
-------
3.0 SELECTION OF GCP OPERATING PARAMETERS TO BE CONTINUOUSLY MONITORED
The GCP control strategy is implemented in the MWC regulation by
specifying numeric operating limits for selected combustion parameters and
requiring that these parameters be monitored on a continuous basis. These
parameters are:
• CO in flue gases
• Maximum operating load
• PM control device operating temperature
The parameters selected for continuous compliance are sufficient as a group to
provide verification that the three goals of GCP are continuously achieved.
These three operating parameters are direct measures of or surrogates for each
of the GCP components.
The first goal of GCP, maximization of in-furnace destruction of trace
organics, is accomplished by optimizing waste feeding procedures, achieving
adequate combustion temperatures, providing the proper amount and distribution
of combustion air, and optimizing the mixing process. A failure in any one of
these components will be accompanied by spikes or bulk increases in flue gas
CO concentrations. Therefore, establishment of a CO emission limit
commensurate with GCP ensures that waste feeding, combustion temperature,
amount and distribution of air, and mixing are addressed in the regulation.
The second goal of GCP, minimization of PM carryover, is satisfied by
maintaining appropriate operating load, combustion air flow rates, and air
distributions. For a given combustor design, total air flows are directly
related to operating load, because each combustor is designed to maintain a
relatively constant excess air level. *s load increases above design limits,
air flows increase proportionally, anc the potential for PM entrainment and
carryover increases. Because combustor specific design characteristics which
affect PM carryover vary extensively, and typical excess air levels also vary
on a combustor specific basis, operating load serves as the best measurable
parameter to address PM carryover.
The third goal of GCP, minimization of low temperature CDD/CDF formation
»
downstream from the furnace, can be addressed directly by establishing an
operating temperature limit for PM control devices. A detailed discussion of
3-1
-------
the rationale and data used to establish specific CO, load and PM control
device temperature limits is included in Section 4.
3-2
-------
4.0 DETERMINATION OF COMBUSTOR-SPECIFIC CONTINUOUS COMPLIANCE
REQUIREMENTS
As discussed in Section 3. the recommended GCP implementation approach
is the use of continuous compliance requirements. Three design/operating
parameters have been selected to serve as continuous compliance indicators for
GCP achievement in MWCs:
1. CO emission 1imit
2. Load limit
3. Maximum temperature at PM control device inlet
This section describes the impact of each parameter on achieving low trace
organic emissions. Data are provided to support the selection of the
continuous compliance parameter on a combustor technology-specific basis, and
the continuous compliance levels are detailed.
4.1 Carbon Monoxide in Flue Gases
The oxidation of carbon monoxide (CO) to carbon dioxide (C02) is the
final reaction step in the hydrocarbon oxidation chain. Thus, low bulk CO
concentrations in flue gases provide verification of hydrocarbon oxidation,
which is one of the goals of good combustion practice. Good combustion is
also correlated with low emissions of CDD/CDF and other trace organic
compounds.
High emissions of CO in MWCs may be due to a number of conditions,
includi ng:
• insufficient bulk oxygen levels resulting from feed and
stoichiometry variations
• insufficient temperatures that result in quenching of reactions
• poor air distribution and/or inadequate mixing which results in
localized oxygen deficient conditions
• excessive carryover of particulate-bound organics from the furnace
into lower temperature regions of the combustion system prior to
completion of combustion
4-1
-------
Many of these conditions can also contribute to elevated emissions of CDD/CDF.
The available emissions data base confirms that good correlations exist
between CO and CDD/CDF when CO emissions are high, since high CO generally
indicates poor combustion. However, the fact that oxidation of CO to C02 is
the final reaction step may lead to instances where the cause of the high CO
occurs after all the organic species have been destroyed (e.g., furnace
residence time is sufficient to destroy all organic species, but too short for
complete CO burnout before quenching occurs). The correlation between CO and
CDD/CDF is typically not as strong when CO levels are low (i.e., low CO alone
is not sufficient to ensure that CDD/CDF is minimized). The absence of a
strong correlation between low CO and low CDD/CDF is due to the fact that
multiple mechanisms may contribute to CDD/CDF formation, some of which occur
independent of the CO oxidation process. Selection of CO as a continuous
compliance measure is based on the rationale that maintaining low CO in flue
gases provides verification of good overall mixing and combustion stability,
both of which are necessary requirements for achieving good combustion.
Although low CO is not sufficient to ensure low CDD/CDF, preventing the
occurrence of high CO concentrations minimizes the potential for high CDD/CDF
emissions from the combustor.
The ability to maintain low CO concentrations in MWC flue gases is
dependent on combustor design features and operating practices. A review of
emissions data from MWCs confirms that design limitations may make it
challenging for some combustor types to achieve CO emission levels that are
routinely attained by other units. For example, semi-suspension fired RDF
systems may have more difficulty maintaining low CO levels than mass burn
units due to the effects of carryover of incompletely combusted materials into
low temperature portions of the boiler, and in some cases due to combustion
control instabilities which result from fuel feeding characteristics. A
number of key combustor design characteristics (e.g., secondary air design)
can impact CO emission levels. Based on these considerations it was necessary
to establish the achievable CO emission levels on a technology specific basis.
However, the achievable CO emission concentrations represent performance
levels that correspond to the use of good combustion practices for each MWC
design type.
Variations in CO concentrations from a single unit or from units of a
similar design are usually caused by differences in operating conditions.
4-2
-------
Parametric test data from several MWCs confirm that CO emissions are greatly
impacted by combustor operating conditions and that optimization of process
operations results in minimization of CO emissions. Fuel properties may also
affect the variability of CO emissions from a given MWC. Of particular
concern are the waste moisture content and volatility. It may be challenging
for a combustor to maintain all the conditions necessary for low CO emissions
and good combustion practice when firing extremely wet fuel. Operating
problems can also occur with extremely dry, or highly volatile fuels which can
ignite quickly and deplete available oxygen in localized portions of the
furnace, resulting in unmixed fuel rich pockets. Design and operating
measures (e.g.. availability and use of air preheat) must be incorporated into
the system to accommodate high moisture content and other waste properties
caused by seasonal variation. Blending of wastes by crane operators can also
help to alleviate some of these types of problems.
The combustion control system provides the link between design and
operation of the combustor. The sophistication of the combustion controller
can significantly impact the ability of a unit to maintain stable combustion
conditions and low CO emission levels. Advances in combustion control
hardware and software have evolved so that many new state-of-the-art units now
use computerized controllers that automatically adjust multiple combustion
control variables to maintain very stable combustion conditions. State-of-the-
art combustion controls are considered an available and demonstrated
technology which will contribute to optimization of combustion conditions and
minimization of emissions. The degree of sophistication in combustion control
networks varies greatly for the MWCs included in the emissions data base, and
the variations in emissions performance between data sets may, in some cases,
be attributed to different combustion control designs or operating
philosophies.
4.1.1 CO Emission Data Analysis
The existing CO emissions data base was analyzed to establish an
achievable CO emission level for each MWC technology. This analysis includes
an examination of several long duration CEM data sets and additional short
duration CO emissions measurements gathered during parametric and compliance
tests (typically 1-6 hours per test run). The majority of available long term
continuous emissions monitoring data can be characterized as representing
"normal operating conditions." The majority of the parametric tests provide
4-3
-------
an examination of the effects of combustor operating variations on emissions
performance. The remaining short term CO data in EPA's emissions data base
are from compliance tests, and the majority of these tests represent "best
operating conditions."
In some cases, local permit conditions or State regulatory requirements
provide little incentive to minimize CO emissions. Thus, normal operation for
some MWCs is represented by those operating conditions which result in
achievement of current permit and regulatory requirements, and may not
necessarily represent optimal operating conditions. Recognition of this
difference is important prior to initiation of a data analysis to establish an
achievable CO emission level. Application of good combustion practice seeks
to optimize the combustion process. The achievable CO emission limit is
defined in this analysis by the performance levels that well designed and
operated MWCs in a combustor class can achieve on a continuous basis using
good combustion practice, and not simply the levejs demonstrated by "normal
operation."
Parametric test data provide a glimpse of the extent to which operation
of the combustor affects short duration CO concentrations. A few of the units
for which long term CEM data are available were also involved in parametric
testing programs. These data sets provide the basis for concluding that some
MWCs can improve CO emissions performance from normal operating levels to
levels representing good combustion by making design and operational changes.
Special emphasis is given to these data sets in an attempt to characterize the
difference between normal operation and optimum combustion conditions.
4.1.1.1 Long Term CEM Data Analysis Procedures
Long term continuously monitored CO data provide the best indication of
the process variations experienced during normal operation, and long term
continuous data from well designed and operated units provide support for
establishing achievable emissions performance. Operators from the majority of
MWCs that monitor CO on a continuous basis were contacted, and five sets of
data were acquired in addition to one long term data set that was generated by
EPA. The six data sets comprise long term data from three RDF facilities, two
mass burn waterwall facilities, and one modular starved air MWC. The data
were screened to identify episodes of abnormal operation (e.g.. start-up and
shutdown conditions, etc.) and all data were normalized to 7 percent 02 and
4-4
-------
converted to hourly averages. A decision was made that only those data
measured at operating load levels between 80 and 100 percent of design load
would be included in the analysis. The data were also subjected to an EPA
QA/QC review to verify CEM instrument performance. Results of the data
analysis are presented in the following sections.
4.1.1.2 Averaging Time
Process upsets will occur to some extent in all combustion systems.
Many of these upsets result in loss of one or more of the operating conditions
associated with GCP and, therefore, may be accompanied by spikes in CO
concentration. They also represent a significant potential for the formation
and release of organic species. Thus, one requirement of good combustion
practice is to operate the combustor in a manner that minimizes the frequency
and magnitude of CO spikes, both by incorporating practices that minimize
their occurrence and by initiating prompt corrective action in the event that
upsets and CO spikes do occur. This goal influences the selection of an
averaging time for the emission limit. The averaging times considered in this
analysis included one-hour, four-hour, and eight-hour averages. Eight-hour
and longer duration averaging times were excluded from consideration because
they provide the option of simply "averaging out" an extended period of poor
operation rather than initiating prompt action to correct the upset.
Some combustor upsets that result in high CO levels may not be
correctable immediately (e.g., ram feeder failure will affect the
characteristics of the burning waste bed). The time required to move waste
through the combustor at normal operating rates should be sufficient to
complete any corrective action needed to respond to an upset and reestablish
normal CO levels. Any more extensive corrections may necessitate unit
shutdown. Since the majority of mass burn waterwall MWCs have a waste
retention time on the grate of up to one hour, selection of a one-hour average
CO emission limit does not provide adequate time for an operator to make a
good faith effort to correct upsets and still achieve an emission limit
representative of good combustion practice. Therefore, four hours were
selected as an appropriate averaging time. A four-hour average is expected to
allow operators to achieve the standard even with occasional upsets, provided
that action is taken to correct the upset, yet is sufficiently restrictive to
encourage prompt corrective action so that periods of poor operation do not
4-5
-------
continue unattended. The four-hour averaging time selected is a block average
rather than a rolling average.
4.1.1.3 Categorization of MWC Technologies
Different MWC design types have inherent design and operating
limitations which impact their ability to minimize CO emissions by control of
the combustion process. The definition of achievable CO concentrations
therefore requires that data be evaluated on a technology specific basis. The
CO data analysis was organized to establish achievable emission limits for the
following specific MWC design types:
• Mass burn combustors
Conventional waterwall
Refractory wall
Rotary waterwal1
• Modular combustors
Starved air
Excess air
• Refuse-derived-fuel (RDF) fired combustors
• Fluidized bed combustors (FBC)
The results of the analysis are presented in the following sections.
4.1.2 Mass Burn Waterwall MWCs
The existing emissions data base from mass burn waterwall MWCs is more
extensive than that from any other class of combustor. Two long term CEM data
sets were available from the Millbury, MA and Commerce. CA MWCs for use in the
analysis. Numerous compliance tests and several parametric data sets are
available for mass burn waterwall MWCs, and results of two combustion
retrofits at the Quebec City, QE and Hampton, VA facilities provide supporting
data to evaluate GCP retrofits to existing units. All of these data
contributed to establishment of the achievable CO emission limits. A
discussion of the data analysis is provided below.
4-6
-------
4.1.2.1 Mi 11 bury Long Term GEM Program.
The Millbury. MA MWC comprises three 750 tpd (682 Mg/day) mass burn
waterwall combustors. Each unit is equipped with a computerized combustion
controller and an individual spray dryer/ESP control device. All three units
entered commercial operation in 1987. A 63-day field test program was
conducted by EPA in 1988, during which continuous emissions of CO and other
pollutants (S02, HC1 , opacity, etc.) were measured using source-installed
continuous monitors and instrumentation provided by EPA.12 Flue gas CO
concentrations were measured using a non-dispersive infrared (NDIR) gas
analyzer. Continuous oxygen and C02 levels were monitored to provide the
basis for normalizing the emissions data to 7 percent 02 or 12 percent C02.
Extensive process monitoring was also conducted in the program. During the 63-
day monitoring program a total of 42 valid data days were obtained. A valid
data day was defined as >18 valid hours of monitoring data from all systems
concurrently, with a valid hour defined as >50 percent data availability.
The Millbury MWCs are balanced draft units, which automatically adjust
forced draft and induced draft fan speeds based on steam demand and furnace
pressure, respectively. The units typically operate at or near full rated
capacity to the maximum extent possible. The primary control loop
automatically adjusts combustion air flow to maintain steam setpoints and
exhaust gas oxygen concentrations. Grate speeds and ram feed speeds are
established manually and maintained automatically. Approximately 40-50
percent of total combustion air is typically supplied by the secondary
(overfire) air systems. All of these design and operating practices
contribute to the ability of the system to maintain relatively uniform furnace
stoichiometry, good mixing, and steady state operation. This is reflected in
the ability of the combustor to maintain relatively low, stable CO emissions.
Figure 4-1 shows a time plot of all of the four-hour block averages in the
long term data set.
A statistical analysis of the Millbury data was initiated to establish
an emission limit that would result in various exceedance frequencies for one-
hour, four-hour, and eight-hour averaging times. The approach and results of
the statistical analysis are included in Appendix A. The results of the
statistical analysis of the Millbury data are summarized below:
4-7
-------
100
TIME
Note: Each data point represents a 4 hour block
Figure 4-1. Millbury Long Term CO Emissions
4-8
-------
ONE
EXCEEDANCE
PER
100 hrs
1 mo
3 mo
6 mo
1 yr
10 yr
CALCULATED
1 HR
ROLLING
/BLOCK
51.3
55.4
57.3
58.5
59.6
62.9
CO EMISSION
4 HR
ROLLING
49.8
53.4
55.2
56.2
57.2
60.2
LEVEL (ppm
4 HR
BLOCK
46.7
51.0
53.0
54.1
55.2
58.4
at 7% 02)
8 HR
ROLLING
48.9
52.3
54.0
54.9
55.8
58.6
8 HR
BLOCK
44.3
48.8
50.7
51.9
52.9
56.1
Based on the selection of a four-hour block average for the standard, this
analysis indicates that the unit would exceed a CO emission level of 58.4 ppm
once every ten years.
4.1.2.2 Millburv Compliance Test Data.
Additional short duration CO emission data are available from testing
that was performed by EPA at the Millbury facility in February 1988.13 Five
sample runs were performed at the spray dryer inlet for CDD/CDF concurrent
with stack compliance tests, and CO concentrations were measured continuously
during each of the sampling runs. CO concentrations are plotted against time
in Figure 4-2 for the five runs. Table 4-1 summarizes the mean CO levels for
each run. These short term averages are all approximately 5 to 7 hours in
duration, and the range of concentrations for the five runs is similar to the
emission levels measured during the long term CEM test. All testing was
performed on the same combustor (Unit #2).
These results indicate no significant differences between the short and
long duration monitored CO data from the Millbury unit. The Millbury MWC was
judged to have good combustion practice in place based on its design and
operating features, and this judgement is supported by the unit's emission
performance. It is probable that the combustion control network contributes
to the ability of the unit to maintain operational stability and relatively
low. uniform CO emissions.
4-9
-------
c
3
tr
c
3
o:
-------
4.1.2.3 Commerce. CA Long Term GEM Data.
Long term CO data were also provided to EPA for the Commerce, CA MWC, a
single 350 tpd (318 Mg/day) mass burn waterwall unit.14 The unit is equipped
with the same automatic combustion control system as that at the Millbury, MA
plant. Commerce uses selective non-catalytic reduction to control NOx
emissions and a spray dryer and baghouse flue gas cleaning system.
Three months of continuous CO emissions data were supplied for Commerce
along with the corresponding steam load data. All data were reported as one-
hour average values (24 values per day). The data are characterized by
significant fluctuations in steam load, and the CO emissions levels appear to
be strongly correlated to operating load levels. Figure 4-3 illustrates the
relationship between CO emissions and steam load for a one day period when
significant fluctuations in steam load occurred. The data indicate that the
ability of the Commerce combustor to minimize CO emissions is dependent on the
operating load maintained. The unit reportedly experiences load fluctuations
due to problems with fuel availability.
Commerce plant personnel reported that extensive problems existed with
the CO monitors during the first two months of the period for which the data
were supplied, and that the monitors were subjected to certification tests at
the beginning of the third month.is Based on this information the first two
months of data were excluded from this analysis, leaving only one month of
valid CO emissions data from the unit.
The variations in CO emissions expressed as four-hour block averages are
presented graphically in Figure 4-4. Comparison with Figure 4-3 reveals that
the Commerce data are more variable than those from Millbury. The Commerce
data were also subjected to a statistical analysis to determine exceedance
frequencies for various averaging periods. The results of this analysis are
presented below: J
4-11
-------
a
D
a
D
m
a
E]
D
a
o
o
o
a>
o
co
o
o
CD
o
LO
o
co
o
o
OJ
o
o
o
CO
a.
co
o
0)
to
c
o
o
o
Q
o
UJ
U)
o
(O
OJ
CO
CO
-------
200
0
TIME
Note: Each data point represents a 4 hour block
Figure 4-4. Commerce Long Term CO Emissions.
4-13
-------
ONE
EXCEEDANCE
PER
100 hrs
1 mo
3 mo
6 mo
1 yr
10 yr
CALCULATED
1 HR
ROLLING
/BLOCK
107.7
128.0
137.8
143.5
149.0
165.6
CO EMISSION
4 HR
ROLLING
80.9
93.5
99.5
103.3
106.4
116.8
LEVEL (ppm
4 HR
BLOCK
70.0
85.0
91.8
95.8
99.5
110.7
at 7% 02)
8 HR
ROLLING
73.3
83.8
88.7
91.7
98.5
103.0
8 HR
BLOCK
59.0
72.8
78.9
82.4
85.6
95.4
The statistical model predicted that the Commerce unit could achieve a CO
limit of 110.7 ppm, four-hour block average, with a once in ten years
exceedance based on its performance during the one month period of continuous
data. Review of the CO time line in Figure 4-4 reveals that this predicted
emission level was influenced by several short lived episodes during which a
CO spike resulted in a relatively high four-hour average. This type of upset
was not experienced by the Millbury unit. A comparison of the Commerce and
Millbury units was initiated to determine potential causes for the difference
in performance.
The variation in CO emissions performance between the Commerce and
Millbury units may be influenced to some extent by differences in unit design
and operation. For example, the number and arrangement of secondary air
nozzles and the amount of total combustion air supplied as secondary air
varies between the two units. Millbury reportedly supplies 50-60 percent of
total air through three rows of overfire air nozzles while Commerce typically
supplies 20-40 percent of total air as overfire through two rows of nozzles.
These differences in design and operation affect mixing characteristics, of
which CO concentrations are an indicator. However, the major difference
between the two data sets is the apparent stability of the combustion process.
The Millbury data are characterized by relatively steady CO concentrations
while the Commerce data show much larger fluctuations around the mean CO
value. Both units incorporate state-of-the-art automatic combustion control
systems which were supplied by the same manufacturer and which provide control
of the same combustion parameters. Comparison of the mean one-hour emission
levels from Millbury and Commerce confirms that over a long duration (1-2
months in this instance) both units can achieve comparable mean CO emissions
(37 and 38 ppm. respectively).
4-14
-------
Follow up contacts were initiated with the Commerce plant personnel to
investigate potential reasons for the high CO spikes. The periodic episodes
of high CO emissions were attributed mainly to problems that are experienced
when firing wet refuse.15-16 The unit was not equipped with air preheat
capabilities during the period when the data were generated. Installation of
steam coil air preheat will be completed in 1989, and this modification is
expected to minimize the occurrence of CO spikes and lead to improved
combustion stability. The ability to modify operating procedures in response
to changing waste characteristics is an important part of good combustion
practice. The absence of air preheat at Commerce is judged to be the primary
contributing factor which resulted in the CO spikes. These operating
conditions can be minimized by installing air preheat, which is a necessary
design requirement of good combustion practice. These design features are in
place at the Millbury plant, and there is less evidence of problems associated
with wet waste firing based on the CO data available from the CEM study.
Based on a judgement that the CO spikes result from conditions that are
preventable, the statistical analysis was repeated for the Commerce data set.
Seven additional one-hour averages were deleted from the available data which
comprised 524 one-hour averages. Each of these points represented a one-hour
average value which was greater than 3 standard deviations above the mean CO
concentration, and each point was judged to be due to a preventable upset
condition. The values ranged from 130-480 ppm. Calculations with this
altered data set indicated an achievable emission level of 76 ppm for a four-
hour block averaging period and one exceedance in 10 years. If the upset
conditions are limited to those data with CO hourly averages above 200 ppm,
only three data points are omitted (202. 232, and 480 ppm) and the achievable
emission level is 80 ppm at 7 percent 02 for a four-hour averaging period with
one exceedance in 10 years. These emission concentrations are representative
of the achievable CO emission levels when good combustion practices are
applied.
4.1.2.4 Long Term Data Analysis Conclusions.
The two data sets provide an example of the distinction between "optimal
emissions performance." as demonstrated by Millbury and "normal operation," as
demonstrated by Commerce. The Millbury data confirm that mass burn waterwall
MWCs can be designed and operated in a manner that results in continuous
4-15
-------
achievement of moderately low CO emissions. The data from Commerce indicate
that the unit has the ability to achieve long term average CO emissions
comparable to those at Millbury. However, periods of combustion instability
occurred at Commerce during the measurement period which resulted in
occasional high short term average concentrations. These upset conditions are
due to conditions which are deemed to be preventable by applying good
combustion practices. Therefore, based on the results of the above analysis
it is judged that mass burn waterwall MWCs which utilize good combustion
practices can achieve a CO emission level of 100 ppm, normalized to 7 percent
02, four-hour block average.
4.1.2.5 Review of Additional Short Term CO Data.
It is necessary to evaluate additional short term CO data to address the
following issues:
(1) How do the CO emission performance data from other mass burn
waterwall units compare to the data from the Millbury and Commerce
units?
(2) In the event that some units in the population of mass burn
waterwall combustors have CO emission performance levels much
higher than those demonstrated by Millbury and Commerce, can
combustion retrofit measures be specified and implemented which
will bring the CO performance of these units into compliance with
the recommended emission limit?
The first issue is addressed by review of the available short term data
from other facilities in the population. Table 4-2 provides a summary of
these data.
All of the CO emissions data in Table 4-2 are short duration test data,
generally 1 to 6 hour averages. There are short term CO data available from
15 mass burn waterwall facilities, which represent more than half of the total
population using this technology in the U.S. in August 1989. Nearly all of
the short term data were gathered during compliance tests under normal steady
state operating conditions. With two exceptions these facilities have
reported average short term CO concentrations less than 100 ppm at 7 percent
02. Included among these facilities are Millbury and Commerce, for which long
4-16
-------
TABLE 4-2. MASS BURN WATERWALL MWCS - SHORT DURATION
TEST EMISSIONS SUMMARY
FACILITY
OPERATING CONDITIONS
CO
(ppnw at 7% 02)
Millbury, MA
Pinellas County. FL
Westchester County, NY
Saugus, MA
North Andover, MA
Commerce. CA
Marion County. OR
Alexandria. VA
Tulsa, OK
Chicago, IL
Hampton, VA
Claremont. NH
Long Beach, CA (SERFF)
Quebec City, Quebec
Portland, ME, North Unit
Portland, ME. South Unit
Normal
Normal
Normal, end of campaign
Normal, start of campaign
Low load
High load
Normal
Normal
Normal
Residential/Commercial
Commercial
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Low load/good
Design load/good
High load/good
High load
Low temperature
Poor air distribution
Normal
Normal
38
4
7
24
21
36
40
43
16
50
22
18
18
22
215
24
55
118
40
33
55
82
121
204
41
75
4-17
-------
term CEM data are available. Although individual combustor performance will
vary to some extent based on site specific design and operating conditions the
available data confirm that the majority of units in the existing population
can achieve CO emission levels of 100 ppm, four-hour average, in their current
configuration.
The second issue regarding combustion retrofit options is an important
one for any mass burn waterwall facility that cannot achieve CO concentrations
less than 100 ppm. Combustors that do not currently meet the goals of good
combustion practice will be required to implement changes to existing design
and/or operating practices in order to minimize emissions. There are three
case studies available at this time which provide supporting information and
data to relate the effects of combustor design and operating modifications to
emission reductions at mass burn waterwall MWCs. Two of the studies (Quebec
City, QE and Hampton, VA) involved rather extensive modifications to existing
design and operating procedures. The third case study at Commerce, CA
provides supporting data concerning the effects of operational changes on CO
emissions achievabi1ity. The scope and results of each case study are
summarized below.
4.1.2.5.1 Quebec Citv. Quebec Combustion Retrofit Program. The Quebec
City MWC facility comprises four 250 tpd (227 Mg/day) mass burn waterwall
combustors. All four units have been operating since 1975. The goal of
Environment Canada's retrofit program at the Quebec City MWC was to determine
the optimum design and operating conditions to minimize air emissions from the
unit and to retrofit the system to meet these conditions.*7 A profile of the
unmodified design is shown in Figure 4-5a. In their original configuration.
each combustor had a vibrating feeder-hopper and a water-cooled chute that fed
the waste by gravity. There were three grates (drying, burning, and
finishing) in each unit. The grates had a 15° slope and contained vertical
drops between each section. The furnaces were membrane waterwall construction
with a refractory-lined burning chamber and a mechanically-rapped convective
section with superheater and economizer tube sections. Each unit controls PM
emissions with a two-field ESP which operated at temperatures between 392-
504°F (200-280°C). Bottom ash was discharged from the grates to a wet quench
tank and removed with a drag chain.
In 1979 a waterwall arch (shown in Figure 4-5a) was installed above the
drying and burning grates. Existing side wall overfire air ports were
4-18
-------
£=
CD
to
.0
o-
(O
tn
i
CD
•r—
Li_
4-19
-------
abandoned in favor of 20 new ports located on the front wall beneath the
waterwall arch. An auxiliary oil burner was located in the upper front
furnace; however, it was not used. The underfire air fan supplied
approximately 90 percent of the total air flow through five plenums beneath
the grates. The control scheme was largely manual, with the exception that
total underfire air flows were adjusted automatically to maintain steam flow
setpoints.
In 1985 emission testing was conducted on one of the pre-modified
boilers. A slipstream arrangement was used to route untreated flue gases from
the boiler outlet to a pilot scale dry scrubbing/fabric filter control device
in order to examine the performance of the flue gas cleaning technology. Data
from this program can be used to characterize the emissions performance of the
pre-modified combustor design. Thirteen test runs were evaluated and average
CO emissions varied from 198 to 362 ppm corrected to 7 percent 02.
The first step in the retrofit program was the completion of flow
modeling studies to examine the existing furnace flow patterns. The objective
of the modeling studies was to select a configuration where furnace geometry
and air flows could provide the best mixing of combustion products and
adequate retention times in the furnace for good combustion to occur. The
following modifications were made to the combustor following analysis of the
flow modeling results. A profile of the modified configuration is shown in
Figure 4-5b.
A lower bull nose was added on the rear furnace wall to maximize the
radiation reflection onto the burning and finishing grates, thus providing
improved ash burnout. The bull nose was also designed to pinch the flow of
combustion gases from the finishing grate to mix the combustion products and
complete the burning process. The upper bull nose reduced gas vortices in the
upper portion of the furnace, improving gas distribution and reducing
stratification at the inlet to the convective section. New overfire air
nozzles were installed to the pinched wall section to improve mixing. Various
front-to-rear overfire air ratios were examined and a 1:1 ratio was chosen
because it resulted in the optimal vertical mixing and least amount of
stratification at the inlet to the convective section. The reconfiguration
also prevented high velocities in the upper furnace, which helped to reduce PM
carryover.
4-20
-------
The underfire air supply was redesigned to include nine separate
plenums. The arrangement provided a single plenum under the drying grate, six
individual plenums beneath the burning grate, and two plenums beneath the
burnout grate. Each of the underfire air supplies is individually controlled
to maintain a preset distribution. Total underfire air flows are controlled
to maintain steam production rates. The underfire air system supplies 65
percent of total combustion air under normal operating conditions and the
overfire air system supplies the remaining 35 percent.
A state-of-the-art automatic combustion controller was installed. The
system automatically controls grate speed in response to boiler steam flow
with an excess air feedback loop to the grate speed controller. Underfire air
flows and distributions are maintained automatically and there are provisions
in the control system to vary overfire air flow rates in response to
temperature readings in the upper furnace.
Following completion of the modernization program, a parametric testing
effort was conducted to evaluate the effects of the retrofit on emission
levels. The first phase, characterization testing, investigated the effects
of feed rate, excess air rates, combustion temperatures, and
overfire/underfire air ratios on emissions of CO and other continuously
measured gases. From the results of characterization testing, a series of
performance testing conditions were selected for manual sampling of CDD/CDF,
and other organic and inorganic pollutants. All sampling was conducted at the
ESP exit location. Table 4-3 summarizes the CO emissions measured during each
performance condition. These data confirm that changes to design and
operation can result in significant reductions in CO emissions.
4.1.2.5.2 Hampton. Virginia Combustion Retrofit Program. The Hampton,
VA MWC comprises two 100 tpd (91 Mg/day) mass burn waterwall combustors. The
units started up in 1980. Each unit has an individual electrostatic
precipitator. The Hampton MWC has been tested for trace organic emissions at
least five times, and prior to the completion of a combustion retrofit the
units were characterized by relatively high emissions of CDD/CDF and CO.3 In
the 1984 test report average CO emissions reportedly varied from 900 to 1400
ppm, corrected to 7 percent Oz. The units had a history of unstable operation
characterized by large fluctuations in excess oxygen and furnace temperature.
4-21
-------
TABLE 4-3. SUMMARY OF EMISSIONS FROM QUEBEC CITY MWC
PERFORMANCE TESTS
TEST NO.
PT1
PT2
PT3
PT4
PT5
PT6
PT7
PT9
PT10
PT11
PT12
PT13
PT14
PT15
OPERATING LOAD
Low
Low
Design
Design
Design
Design
High
High
Low
Low
Design
High
Design
Design
COMBUSTION
CONDITIONS
Poor
Good
Poor
Poor
Good
Good
Good
Good
Good
Good
Good
Good
Poor
Poor
CO
(ppm at 7% 02)
99
20
89
86
21
29
46
50
28
31
35
82
165
202
4-22
-------
Following the completion of the 1984 emissions test, the plant operators
initiated a retrofit program to modify the design and operation of the
units.18 This was not only due to concerns related to emissions, but also due
to the need for corrective action to address operating problems that were
plaguing the boilers. These problems were mainly related to chloride
corrosion. The original cast iron grate bars were replaced with high alloy
chrome-nickel grates and the life of the grates was extended from 4-6 months
to 2-3 years. High alloy blocks were retrofitted on the lower side walls of
the furnace, replacing existing silicon carbide refractory, and resulting in
improved heat transfer and reduced clinker formation. Steam coil air
preheaters were also added to the units for operation during periods of wet
waste firing.
The major improvements that were made to reduce emissions were primarily
related to combustion airflows and distributions. First, it was determined
that the forced draft fan supplying the overfire air was providing less than
half its design capacity. The fan blades were modified and the discharge duct
size was increased, making the flow more aerodynamic. These modifications
restored the overfire air supply to its original design capacity (45 percent
of total air). The plant personnel also realized that mixing was not
optimized, so they began to evaluate the size and orientation of the overfire
air nozzles. There are four rows of overfire air nozzles (two rows on each of
the front and rear walls). The orientation of the lower two rows was changed
based on visual observations made in the furnace. The angle of the front row
was raised from -45° (from the horizontal) to -22.5°. The angle of the rear
wall nozzle row was changed from -20° (from the horizontal) to 0°
(horizontal). Now the overfire air jets converge at a point approximately
five feet (1.5 meters) above the grate rather than directly on the grate.
Modifications were also made to the operation and combustion control
system. The grate speeds, which were automatically controlled, were switched
to manual, which allowed the speed to be varied from 0-80 percent rather than
40-80 percent. This provided more flexibility to deal with varying waste
characteristics (particularly wet waste), and resulted in improved burnout. A
15 point CO profile was performed at the economizer outlet and it was
determined that CO was highest when active burning occurred on the lower
burnout grate sections. An oxygen trim loop was installed which modulates the
distribution of air to the burnout grate based on the 02 content of the flue
4-23
-------
gases. Setpoints are maintained between 7-9 percent 02 and the proper burnout
of waste on the grates is maintained.
Lastly, the existing economizer was replaced with new tube banks which
drop the flue gas temperature to 425°F (218°C) before entering the ESP.
Previously the ESP operated at approximately 550°F (288°C). where the
potential for CDD/CDF formation was relatively high. Installation of the new
economizer has reduced total fuel consumption on an hourly basis, but this has
been offset by increased system availability, which has actually increased
overall steam output and waste throughput. The most recent emission test
performed at Hampton resulted in CO stack concentrations of 24 ppmv. The
modifications were successful in reducing CO emission concentrations by more
than an order of magnitude.
4.1.2.5.3 Commerce. California Overfire Air Optimization Tests.
Results from a combustion optimization study conducted at the Commerce, CA MWC
provide data correlating emissions of CO and NOx with changes in overfire air
firing rate and excess air operating levels.19 The study was initiated in an
attempt to define the optimal levels for these operating variables. Gaseous
emission and unit operating characteristics were monitored and observed
continuously, providing a set of criteria for determining optimal operating
conditions. Sixteen separate test runs were performed which included three
overfire/underfire air ratios (30/70, 40/60, and 50/50) and three excess 02
setpoints (5.5%. 7.0%, and 8.5%). The test matrix is presented in Table 4-4.
The thermal de-NOx control system was shut off during the tests in order to
examine the effects of the operational changes on NOx emissions. Some of the
major conclusions in the study were:
• CO emissions were less than 39 ppm (corrected to 7% 02) for all 02
levels when operating at 40 and 50 percent overfire air. At 30
percent overfire air, CO levels ranged from 39 to 140 ppm
(corrected to 7% 02). This relationship is illustrated in Figure
4-6.
• At 40 and 50 percent overfire air CO increases slightly with
excess air.
4-24
-------
TABLE 4-4. TEST MATRIX FOR COMMERCE. CA MWC
COMBUSTION OPTIMIZATION PROGRAM
TARGET OFA
30%
40%
50%
5.5%
3, 15
5
12*
FURNACE 02 SET POINT
7.0%
1. 2. 4
6. 9
11
(wet)
8.0%
8
7. 16**
10. 13. 14**
Numbers in table designate individual test runs.
*Actual setpoint 6.0%
**Actual setpoint 8.0%
4-25
-------
c\j c\j cvj
O O O
§ CD 0)
El
O
LO
^ o
O
CO
O
CM
O
O
CM
O
O
LU
DC
Li.
DC
UJ
o
4-
o
•p
o
OJ
10
I
CLI
s_
3
CD
Lo_
(ZO
iiidd) OO
4-26
-------
• At 30 percent overfire air, a parabolic curve of CO versus 02 was
demonstrated, with CO increasing sharply at both low and high 02
1evels.
The authors concluded that there is a threshold level above which mixing and
completion of combustion at the overfire air injection location are optimized.
Visual observations confirmed that flame height in the furnace was also lower
during operation at high overfire air levels. The study confirms the ability
of MWCs to minimize CO emissions by making changes to combustor operating
conditi ons.
4.1.2.5.4 Concl usi ons. The Quebec City and Hampton combustion
retrofit studies provide conclusive evidence of the ability for existing mass
burn waterwall MWCs to minimize CO emissions through changes in combustor
design and operating practices. It is doubtful that there are many existing
mass burn waterwall units that will require design changes as extensive as
those implemented at the Quebec unit. In addition, it is difficult to assess
the contribution that each individual design change made to the overall
improvement in emissions performance by the combustion system. However,
results of these programs provide strong support for the judgement that
combustors can retrofit successfully to reduce CO emission limits. The
results of the combustion optimization study at the Commerce MWC also provide
evidence that it is possible to implement operational changes that result in
improved emissions performance with little or no resultant capital
expenditures. The testing programs provide support for the achievabi1ity of
100 ppm CO, four-hour average, by way of applying good combustion practices.
4.1.3 Modular Starved Air MWCs
The available CO emissions data base for modular starved air MWCs
consists of one set of long term continuous data, two parametric tests, and
one compliance test. The long term data demonstrated that modular starved air
combustors can achieve CO emissions less than 10 ppm over extended periods of
time at normal steady state operating conditions. These findings were
verified by data gathered during one parametric test and one compliance test.
However, due to the occurrence of periodic combustor upsets during the course
of normal operation, an emission limit of 50 ppm, four hour average, was
4-27
-------
determined to be continuously achievable for new and existing modular starved
air units. The test results used to reach this conclusion are summarized
below.
4.1.3.1 Osweqo County. NY Long Term GEM Data.
The Oswego, NY MWC plant comprises four starved air combustors, each
rated at 50 tpd (45 Mg/day). Each unit is equipped with a separate waste heat
boiler and electrostatic precipitator (ESP) controls. Individual 02 and CO
monitors are in place at each of the four units.
Process and emissions data were obtained for a 60 day period from March
17 to May 15, 1989. which reportedly represents a period of normal
operation.20 All four units were in operation to some extent during this
period, although normal operation consists of three units on line and one on
standby. The units normally operate at 80 to 90 percent of rated capacity.
Emissions and process data are measured at 3 minute intervals.
Carbon monoxide emissions data from all four units were relatively
consistent over the 60 day period. In general, during steady-state operating
conditions, the CO concentrations were 0 to 2 ppm. A number of spikes were
traced to CO instrument calibration, which occurred automatically every eight
hours. These values usually varied from 10 to 50 ppm for one 3 minute
reading, depending on the timing between the calibration procedure and data
recording. The consistent frequency of these readings permitted them to be
identified as calibration values.
There were also periodic episodes when CO spikes occurred during normal
steady state process conditions. In most cases these were isolated instances
for one or two readings (3-6 minutes) which could be traced to abnormal
operating conditions (e.g., low secondary chamber temperatures). These upsets
were usually corrected in less than an hour by taking specific action such as
turning on auxiliary fuel burners.
In a few instances the duration of a CO spike extended for periods of
several hours, resulting in one-hour and four-hour average CO concentrations
of 100 ppm or greater. Review of process data (secondary temperature, steam
flow, flue gas flow rate, etc.) provided no indication that the spikes are a
direct result of process variations. Therefore, two potential sources of the
4-28
-------
high CO levels are suggested. First, some of the high CO levels may result
from incorrect instrument readings. Plant personnel identified several
specific time periods during which problems were experienced with one or more
of the emission monitors.20 There may have been additional periods of monitor
malfunction that were not specifically identified. The unexplained CO
excursions were confined almost exclusively to Unit #2. Alternately, the high
CO may have been a result of combustor operating procedures which could not be
pinpointed by review of available process data. For example, several episodes
of high CO occurred with regularity between midnight and 8:00 a.m. Although
the process data do not provide an explanation of why the higher emissions
were occurring, it is possible that the pattern reflects the operating
procedures employed during that 8-hour shift. Additional periods of high CO
were experienced during combustor start-up and shutdown conditions as
expected. Data gathered during start-up and shutdown were excluded from the
analysis.
Despite the occasional occurrence of flue gas CO concentrations as high
as 100 ppmv, nearly all of the data measured during normal operating
conditions were less than 10 ppm. resulting in extremely low one-hour and four-
hour average values. The data represent "best combustion conditions" during
the majority of time that the units are in operation.
4.1.3.2 Oswego County. NY Parametric Test.
The emission performance of one of the Oswego units has also been
verified independently in a parametric emissions test sponsored by New York
State Energy Research and Development Authority (NYSERDA).21 The program
included four testing conditions carried out at beginning and end of campaign
cycle and at varying secondary combustion chamber temperatures. Table 4-5
summarizes the results of the CO data for each test run. The data clearly
confirm the ability of the unit to operate in a consistent fashion which
minimizes CO emissions. The data suggest a slight degradation in CO oxidation
at lower secondary chamber operating temperatures. However, the consistently
low CO emissions confirm that the unit achieves good mixing and adequate
residence time at the temperatures necessary to complete CO burnout, and that
the low CO emissions obtained in the long term CEM study are representative of
normal operating conditions. The data indicate that burnout of CO continues
in the high temperature portion of the boiler, reducing CO concentrations
between the boiler inlet and outlet sampling locations.
4-29
-------
TABLE 4-5. OSWEGO COUNTY, NY PARAMETRIC TEST RESULTS
TEST CONDITION
Start of
Campaign
Mid-Range
Secondary
Temperature
End of Campaign
Low Secondary
Temperature
RUN
1
2
3
Avg
4
5
6
Avg
7
8
9
Avg
10
11
12
Avg
SECONDARY CHAMBER
, TEMPERATURE
1875 1024
1851 1011
1837 1003
1854 1012
1752 956
1741 949
1738 948
1744 951
1817 992
1822 994
1834 1002
1824 995
1634 890
1627 886
1617 881
1626 885
CO
(ppm @ 7% 02)
BOILER BOILER
INLET OUTLET
NA 0
3 1
4 0
0
NA 0
4 1
3 5
2
18 2
4 4
20 3
14 3
21 6
20 6
18 3
20 5
4-30
-------
One important finding in the parametric testing program was that
periodic high, short duration CO spikes coincided with activation of the ash
ram in the primary combustion chamber. It was assumed by the author of the
report that the cause of the CO spike was flashing of unburned refuse as it
was pushed off the hearth into the ash discharge. This may have also caused
the periodic short lived CO spikes that were observed in the long term
continuously monitored data.
4.1.3.3 Prince Edward Island (PEI) Parametric Test.
An earlier parametric test was conducted at the Charlottetown, PEI
facility by Environment Canada in 1984.6 The plant includes three 36 tpd (33
Mg/day) units with a single waste heat boiler and no add-on pollution control
equipment. The primary goal of the program was to study the relationship
between combustion conditions and emissions. The available CO and CDD/CDF
data are summarized in Table 4-6. Although the CO emissions data measured at
PEI are relatively low, they are significantly higher than the Oswego data.
Because the two units are of a similar design, it was assumed that the higher
values reflected a difference in combustor operating procedures.
4.1.3.4 Red Wing. MN Compliance Test.
The Red Wing. MN MWC includes two 36 tpd (33 Mg/day) units with a waste
heat boiler and ESP controls. The unit was tested for a host of pollutants in
September 1986.22 Average CO concentrations for three runs were less than 2
ppm. Limited process data are available for the units. Average primary
chamber temperatures ranged from 1400-1590°F (760-866°C) and mean secondary
chamber temperatures ranged from 1750-1960°F (954-1071°C).
4.1.3.5 Conclusions.
The available data from the Oswego. PEI, and Red Wing facilities confirm
the ability of modular starved air MWCs to achieve very low CO emissions.
However. PEI had a significantly higher mean CO concentration than the other
units. The Red Wing and PEI combustors use the same combustor design. Thus,
it is judged that PEI is not limited by its design from achieving CO emissions
on the order of those demonstrated by the Oswego and Red Wing units, and the
slightly higher emissions must be attributed to differences in combustor
4-31
-------
TABLE 4-6. PEI PARAMETRIC TEST RESULTS
CONDITION
Normal
Long Cycle
High Secondary
Temperature
Low Secondary
Temperature
RUN
2
3
4
Avg
5
6
7
Avg
8
9
10
Avg
11
12
13
Avg
SECONDARY CHAMBER
TEMPERATURE
(°F) (°C)
1834 1001
1832 1000
1839 1004
1835 1002
1830 999
1837 1003
1788 976
1818 992
2055 1124
2079 1137
2095 1146
2076 1136
1617 881
1656 902
1656 902
1643 895
STACK EMISSIONS*
CO 02
(ppmv) (%)
59 12.3
68 12.3
78 12.2
68 12.3
47 12.6
48 12.6
29 12.3
41 12.5
47 9.9
15 9.5
39 9.6
34 9.7
40 13.6
74 13.6
45 13.2
53 13.5
*CO emissions corrected to 7 percent 02.
02 percentages as measured.
4-3?
-------
operating procedures, feed characteristics, or both. Contacts were made by
telephone with the unit supplier to discuss operating practices which can
affect CO emission levels, and one suggestion was that slag buildup around
secondary air ports or in the breeching between the primary and secondary
chambers can often disrupt air flow and mixing patterns to the extent that
higher CO levels result.23 This effect is implied by the data from the Oswego
parametric test in which higher CO concentrations are observed at the boiler
inlet during the end of campaign test runs than at the beginning of campaign.
If mean CO emissions increase during the end of an operating campaign due to
slagging problems it may be necessary to increase the number of scheduled
shutdowns to perform maintenance and cleaning in order to maintain low CO
concentrations in flue gas.
Based on the available emissions data, a CO emission limit of 50 ppm,
four-hour block average, corrected to 7 percent 02, is judged to be achievable
for modular starved air MWCs. It is expected that all new units can achieve
this limit by incorporating the proper mixing design and residence time in the
secondary combustion chamber. The majority of existing modular starved air
units are expected to be able to achieve the limit by optimizing process
operations. In a very few cases the emission limit may force combustors to
modify existing designs to satisfy the criteria of good combustion practice.
4.1.4 RDF Spreader Stoker MWCs
The CO concentrations from RDF spreader stokers are historically higher
than from other MWC design types.24 This is partially a result of some
inherent design features unique to RDF spreader stokers. For example, semi-
suspension firing results in a higher percentage of particulate matter
carryover from the furnace. Particle bound organic matter can be swept out of
the primary combustion zone prior to completion of the combustion process,
resulting in increased CO concentrations. The design of the combustion
control system also influences combustion stability and CO emissions. RDF
spreader stokers typically use fuel input rate as the primary operating
variable to control steam flows, while mass burn combustors automatically
adjust the rate of primary combustion air supplied to the burning waste bed.
Both combustion control modes attempt to maintain uniform heat release rates
by adjusting overall furnace stoichiometry. However, RDF units are subject to
more rapid changes in fuel properties and heat release rates due to semi-
suspensibn feeding, and many RDF units do not have control features which can
4-33
-------
respond to these changes in stoichiometry by automatically adjusting air flow
rates. Thus, excess oxygen levels in RDF units experience wider variations
than mass burn waterwall MWCs, and greater difficulties are encountered in
maintaining stable combustion conditions and low CO concentrations. Secondly,
RDF units are generally designed to operate at lower excess oxygen levels than
mass burn waterwall units (3-9 percent 02. dry basis, in flue gases, versus 6-
12 percent 02 for mass burn).i Lower 02 availability provides a greater
potential for the existence of starved air pockets of gas in the combustion
process, which can lead to elevated CO levels.
Two sets of long term data were available for inclusion in this analysis
from RDF combustors located in Hartford, CT and Orrington, ME. Both data sets
are characterized as representing "normal operating conditions", which include
significant fluctuations in oxygen and CO concentrations. Recent data from a
parametric testing program at the Mid-Connecticut (Mid-Conn) RDF fired plant
in Hartford, CT provide evidence that, despite differences in the design of
RDF combustors which may result in high CO, emissions can be reduced
significantly by modifying combustor operating conditions. These results,
along with other limited compliance data from two new facilities, comprise the
supporting data for the proposed CO emission limit for RDF fired combustors.
A discussion of the available data is provided below.
4.1.4.1 Mid-Connecticut Long Term CEM Data.
The Mid-Connecticut (Mid-Conn) RDF facility includes a waste processing
plant and three spreader stoker boilers with a combined capacity of 2000 tpd
(1818 Mg/day). The units have the ability to fire 100% coal in addition to
RDF. Emissions control is achieved by spray dryers and fabric filters. Three
months of continuous emissions and processing data were submitted to EPA for
each of the three units (#11, #12, and #13).25 An initial screening of the
data revealed that the three month period was characterized by frequent unit
start-up and shutdown. Normal operating procedures at the plant include two
units on line while the third is down for maintenance. However, the longest
periods of continuous operation for each of the boilers were 9 days, 22 hours
for Unit #11, 11 days, 20 hours for Unit #12. and 5 days. 23 hours for Unit
#13. Periodic problems with instrumentation resulted in exclusion of
additional data from these periods. Periods of 100 percent coal firing were
also excluded. The remaining data were included in a statistical analysis for
purposes of determining exceedance frequencies. Results from the statistical
4-34
-------
analysis are presented below for Units #12 and #13. The analysis predicted
the level at which one exceedance per ten years would occur was 338 ppm for
Unit #12 and 333 ppm for Unit #13. Figures 4-7, 4-8, and 4-9 illustrate the
variation in CO with time for each of the three units.
UNIT 12
ONE
CALCULATED CO EMISSION LEVEL (ppm at 7% 02)
1 HR
XCEEDANCE
PER
100 hrs
1 mo
3 mo
6 mo
1 yr
10 yr
ROLLING
/BLOCK
299
338
357
368
379
411
4 HR
ROLLING
211
297
311
319
328
352
4 HR
BLOCK
211
276
292
302
311
338
8 HR
ROLLING
253
279
292
299
306
328
8 HR
BLOCK
217
251
267
276
294
308
UNIT 13
ONE
EXCEEDANCE
PER
100 hrs
1 mo
3 mo
6 mo
1 yr
10 yr
CALCULATED
1 HR
ROLLING
/BLOCK
332
368
385
395
404
433
CO EMISSION
4 HR
ROLLING
283
340
314
320
326
343
LEVEL (ppm
4 HR
BLOCK
265
290
302
308
314
333
at 7% 02)
8 HR
ROLLING
247
258
263
266
269
278
8 HR
BLOCK
232
247
253
256
260
270
4.1.4.2
Penobscot. ME Long Term CEM Data.
Five months of continuous CEM data from the Penobscot Energy Recovery
Company (PERC) RDF fired facility were submitted to EPA for inclusion in this
analysis.26 The PERC plant, located in Orrington, ME, utilizes two 360-tpd
(327 Mg/day) spreader stoker boilers with spray dryer/baghouse controls. The
boilers are capable of'burning 100 percent oil in addition to RDF. The carbon
monoxide monitor is located in the stack, which serves both boilers. The
units have a CO emission limit in their operating permit of 400 ppmv,
corrected to 12% C02, four-hour rolling average. A statistical analysis was
performed on a representative segment of the data. The results are presented
below:
4-35
-------
CM
O
500
400
300
8 200
100
5 days 16 hours
mean - 188ppm
TIME
Note: Each data point represents a 4 hour block average
Figure 4-7. Mid-Connecticut Long Term CO emissions - Unit 11
4-36
-------
300
200
E
Q.
Q.
O
O
100
0
10 days 20 hours
mean = 125 ppm
max = 253 ppm
TIME
Note: Each data point represents 4 hour block average
Figure 4-8. Mid-Connecticut Long term CO Emissions - Unit 12
4-37
-------
300
CM
o
E
Q.
a.
O
O
200 -
100
0
5 days 9 hours
mean = 205 ppm
max = 256 ppm
TIME
Note: Each data point represents 4 hour block average
Figure 4-9. Mid-Connecticut Long Term CO Emissions - Unit 13
4-38
-------
CALCULATED CO EMISSION LEVEL (ppm at 7% 02)
ONE 1 HR
XCEEDANCE
PER
100 hrs
1 mo
3 mo
6 mo
1 yr
10 yr
ROLLING
/BLOCK
293
339
361
374
386
423
4 HR
ROLLING
265
303
321
332
342
373
4 HR
BLOCK
233
278
298
310
321
355
8 HR
ROLLING
253
287
304
313
322
350
8 HR
BLOCK
206
251
271
283
293
325
The predicted concentration that would result in one exceedance every ten
years was 355 ppm, four-hour block average. The data are plotted for a two
week period in Figure 4-10.
4.1.4.3 RDF Compliance Test Results.
Numerous compliance tests have been performed on new RDF boilers that
have come on-line in the past two years. Short term data are available from
the following plants: Mid-Conn; PERC; Maine Energy Recovery Company (MERC) in
Biddeford, ME; and Red Wing, MN. 27,28.29,30 The test results are summarized for
each of the facilities below.
Mean CO Concentration
Faci 1 itv
Mid-Conn
PERC
MERC
Red Wing
(ppm @ 7% 02)
198
191
81
99
By comparison, the mean four hour block concentrations from the continuous Mid-
Conn data were 188, 125, and 205 ppm (see Figures 4-7, 4-8, and 4-9), and the
mean four hour block concentration from the long term PERC data was 144 ppm.
Although the short term compliance data may not necessarily reflect the
variations in CO concentration that occur over longer operating periods, the
short term mean concentrations do not vary significantly from the long term
4-39
-------
E
Q.
a.
O
O
600
500
400
300
200
100
0
TIME
(2 weeks duration)
Figure 4-10. Penobscot Long Term CO Emissions
4-40
-------
mean values. Based on this finding, it is judged that long term data from
MERC or Red Wing would likely provide lower predicted exceedance levels than
those from Mid-Conn or PERC.
4.1.4.4 Mid-Connecticut Parametric Test.
An extensive set of combustor operating conditions were investigated in
a parametric testing program that was conducted on Unit #11 at the Mid-Conn
facility in late 1988 and early 1989.31 The goals of the program were to
determine the design and operating practices which influenced emissions of
organic, metals, acid gases, and criteria pollutants. Simultaneous
characterization of combustor and flue gas cleaning device performance were
examined. Carbon monoxide was monitored continuously during all test
conditions.
Table 4-7 summarizes the test conditions examined in the program and the
mean CO concentrations that were measured during each test. The data are
organized based on four operating load levels (low, intermediate, normal, and
high). The units have rated capacities of 231.000 Ib/hr (105,000 kg/hr) of
steam. However, normal steam flow conditions during the performance tests
ranged from 209,000 to 223,000 Ib/hr (95,000 to 101.400 kg/hr), and high load
conditions were only 1 to 2 percent above rated capacity [234.000 to 235.000
Ib/hr (106.400 to 106.800 kg/hr)]. The test conditions were characterized as
good, poor, and very poor by target CO levels «100 ppm, 200 to 400 ppm, and
>400 ppm, respectively). These conditions were established primarily by
varying combustion air distributions.
Although the test data cannot be used to characterize the long term
performance of the combustor, they provide important evidence of the effect of
operating conditions on CO emissions. For example, the units are equipped
with two separate overfire air systems: there are three elevations of
tangential overfire air (TOFA) which are normally used when burning RDF, and
rows of conventional overfire air nozzles on the boiler'walls, which are used
when firing coal. Various combinations of TOFA and wall air were examined in
the study. A pressure setting of 35 inches of water (8715 Pa) for the wall
air was determined to be the optimal pressure for minimizing CO emissions in
the characterization tests, which were performed prior to the performance
tests. Tests PT3 (normal load) and PT5 (intermediate load) were conducted
with one bank of TOFA in service and 35 inches of wall air. It is clear that
4-41
-------
>-
oi
<
oo
UJ
o:
o
u_
C£
UJ
O-
o
(-3
ca
•>
O t-
O
+J 4J
C C
1^ VO
• •
t^ CO
oo oo
co •«>-
01 oo
I—I r-H
CM m
m
CO
CO
V
^
m
i— i
V
i
i
CM
CM
in
CM
CD
i— 1
CD
r—l
-a
o
0
CO
^
c:
i— i
CO
•
l^^
oo
CM
CD
i— 1
CD
i — I
in
CO
r^
CO
in
i— i
in
i — i
oo
•*~s
in
CO
<— i
^
oo
VO
oo
in
CO
t_
o
0
O-
^~
(-
o
r*^
•
^^
CD
in
i — i
CM
CO
in
CO
p^
CO
in
i—i
i
i
CO
in
VO
^
01
I— 1
CM
c
O
o
Q_
^
C-
o
VO
•
oo
CD
^
i— |
CM
r^
CO
CM
CO
CO
r—l
i
i
CO
1— 1
in
CM
r--
CD
^~
CO
c
0
o
a.
^~
t_
o
"S
I
I — 1
0
1— 1
CO
CM
CM
p^.
CO
CM
CO
CO
i— 1
in
r^
oo
- — *
m
CO
CM
00
in
r*"
CO
CTi
-r^
O
O
CO
--~
£
t-
o
•z
*^-
•
VO
CD
CM
r-l
CM
OO
l~^
CO
CvJ
CO
CO
r-l
in
• — i
CO
x~"'
in
CO
CM
i — i
in
in
r"*
vo
CD
-a
0
o
CO
^
t-
o
CD
•
in
CD
CD
CD
CM
CD
in
CO
r^
CO
in
r-H
in
i — i
00
>~^
in
CO
CM
CM
in
CD
I^'
1 — 1
•^
-a
o
0
CO
—~
^
0
•z
CD
•
in
CD
•—i
r—l
CM
r—l
r-l
•* CO
CO CO
CD CM
CO •*
vo r^
r-H I— 1
in
i — i
CO
'
i in
i CO
CM CM
1 — CO
in in
CD in
oo vo
CM CD
0 0
^r * — i
i- -a
O 0
o o
Q_ CD
— ~~~
cn cn
n: i
•**• oo
* •
VO VO
CD 0
r-l I— 1
•«3- in
CO CO
CM CM
VO CM
r— 1
cu
-a
V) CU
o CM -r-
-P i £_
-r- ^J- CU
~O l/)
o cx c
O CX-r-
cn - C
•r- X O
4-» CU -r-
+->
VTJ CU >
cu a> cu
CX in i—
O ^—' '''
•K
•K -K
4-42
-------
this is an undesirable operating condition based on the CO levels (358 and 901
ppm, respectively). A single bank of TOFA does not provide sufficient mixing
in the upper furnace to control CO emissions. Two additional tests were
performed with all three banks of TOFA in service, PT4 and PT7 (both normal
load). Neither test included wall air, and average CO emissions were 219 ppm
and 340 ppm.
The remaining nine test conditions were conducted with two banks of TOFA
in service, which is a normal combustion air distribution for the unit. All
four operating loads were examined. Two tests were performed at low load
conditions. No wall-fired combustion air was used in either case. Mean CO
levels were 199 ppm during PT13 and 75 ppm during PT14. At intermediate load,
the CO levels were 131 ppm during PT2 and 101 ppm during PT10, without wall-
fired air.
At normal load conditions, triplicate runs were conducted with two banks
of TOFA and 35 inches (8715 Pa) of wall-fired air (PT8, 9, and 11). Mean CO
levels were less than 100 ppm for all tests, with a three-run average
concentration of 87 ppm. Time plots are provided for each of the runs in
Figures 4-11. 4-12. and 4-13. The data include periodic CO spikes which
generally last for several minutes duration, with magnitudes on the order of
300-650 ppm at 7 percent 02. Figure 4-14 shows the relationship between flue
gas oxygen levels and CO concentrations for PT9. The figure clearly shows
that when bulk 02 levels in flue gases are maintained between 6 and 12
percent, the CO concentrations are below 150 ppm, corrected to 7 percent 02.
The CO spikes are entirely associated with operating conditions for which 02
levels are below 6 percent or above 12 percent. One of the requirements of
good combustion practice is to maintain operating conditions (excess 02
levels) in a range which results in low CO emissions. This combustor, and
many other existing RDF units, are lacking the control features which provide
this ability, or are simply not operated in a manner which achieves this goal.
It is suggested by these data that if control setpoints were established which
would prevent operation outside of an acceptable range of excess 02 levels,
that CO emissions would be significantly reduced.
Additional support for the benefit of good operation on CO emission
levels is provided by review of data measured during high load conditions.
Two tests (PT6 and PT12) were conducted at high load, both while firing two
banks of TOFA. However, wall-fired air was added during PT12, and average CO
4-43
-------
E
Q.
Q.
CO
<
O
CD
•+- '
3
c
E
CO
c
CD
Q.
E
4— >
C
"o
CL
CO
(3
•
co
(X
i
en
c
o
CO
(/I
o
o
o
o
CD
S=. o
o V
CD -a
LU -£
CD
i ?
O)
S-
3
CD
oo
4-44
-------
E
Q.
Q.
CD
CD
D)
<
CO
CD
E
CO
CO
•I— «
c
CD
CO
CD
Q_
CD
Q.
o
03
LLJ
0)
O
CTl
Q.
I
Z5
O
O
O)
c
c
o
o
CM
.—I
I
O)
t.
CD
Ll_
o
o
00
I
o
o
CD
O
O
o
o
CM
(2O
OO
4-45
-------
E
Q.
Q.
O)
<
CD
•*->
U
C
"E
CO
co
*—•
C
CD
CO
CD
QL
CD
UJ ^
^ S
p a
CO
Cd
o
03
LJLJ
CD
-i—•
O
UJ
o
o
o
o
OJ
o
o
I
T3
ro
i—i
I
-------
CM
O
E
Q.
Q.
>••*
O
O
800
600 -
400 -
200 -
0
8 10 12 14
02
Figure 4-14. CO Versus Op - Mid-Connecticut PT9
4-47
-------
emissions were reduced significantly, from 402 ppm during PT6 to 109 ppm.
Thus, the use of wall-fired secondary air appears to be the optimal operating
condition for minimizing CO emissions. This is currently not part of normal
operating practice at the Mid-Conn units, and the long term CEM data which
were discussed previously do not reflect this mode of operation.
4.1.4.5 Conclusions.
It is concluded that by establishing the proper controls to maintain
excess air levels within a prescribed range, and by incorporating some key
changes in combustion air distribution, that the Mid-Conn units can greatly
reduce CO emissions from current normal operating levels. These recommended
changes in operation (or design, if necessary) are consistent with achievement
of the goals of good combustion practice. The recommended design and
operating practices can be implemented for new RDF units from the initial
design stages. Modification to current operating practices, and possibly
design changes, will be necessary for many existing RDF units to achieve
comparable CO emission limits. However, these changes are judged to be
necessary as a part of the application of good combustion practices. The
average CO concentrations that resulted from "good operating conditions"
during the Mid-Conn performance tests ranged from 71 ppm at normal load (PT11)
to 199 ppm at low load (PT13). The average steam flow rate during PT13 was 68
percent of rated capacity and included no rear wall air. The data from "best
operating conditions", which are characterized by two rows of TOFA and 35
inches of rear wall air, ranged from 71-109 ppm. Based on these data, it is
judged that 150 ppmv, four-hour block average, is an achievable CO emission
limit for new and existing RDF boilers, and that the performance level is
representative of good combustion.
4.1.5 Mass Burn Refractory Wall MWCs
There are currently no long term data available from mass burn
refractory wall MWCs that can be used to establish an achievable CO emission
limit. Short term data are available from one parametric testing program and
several compliance tests. However, the majority of data were gathered at
units that do not satisfy key requirements of good combustion practice, and
are therefore not useful in establishing an achievable CO emission level.
There are limited data available from two units that are comparable to the
performance data routinely measured at well designed and operated mass burn
4-48
-------
waterwall MWCs. Based on engineering judgement and these limited data it was
concluded that well designed and operated mass burn refractory wall MWCs
should be capable of achieving the same CO emission limits as mass burn
waterwall units. The following sections provide a discussion of the available
data and rationale used to reach this conclusion.
4.1.5.1 Dayton. OH Parametric Test.
A parametric emissions test was conducted by EPA at the South Dayton, OH
municipal incinerator to investigate the effects of ESP temperature and
sorbent use on the control of trace organic and acid gas emissions. The plant
comprises three 300 tpd (273 Mg/day) refractory wall combustors with rotary
kilns. There is no energy recovery capability currently in place at any of
the units. Flue gas temperature reduction is accomplished by water quench
chambers and emissions control is achieved by three-field ESPs. Six test
conditions were examined in the program as shown in Table 4-8.32 Triplicate
testing runs were conducted at each test condition.
As shown by the test matrix, conditions 1 through 5 represent normal
combustor operating conditions with a mixing chamber temperature setpoint of
1800°F (982°C). Despite attempts to maintain consistent combustor operating
conditions for the first five test conditions, average CO concentrations
experienced more than an order of magnitude variation (single run average
concentrations ranged from 17 ppm during Run 1 to 369 ppm during Run 16). The
data indicate that the unit was able to maintain the lowest CO concentrations
when furnace limestone injection was shut off (conditions 1 and 2).
The specific reasons for the wide variation in CO emissions are not
clear at this time, but are probably due to poor mixing and an overall lack of
combustion control. For example, the unit was operated solely on induced
draft, which is manually set and automatically adjusted to maintain negative
pressure setpoints in the furnace. There was no primary air supplied through
the existing forced draft fan. Multiple points of air inleakage are present
throughout the system at access doors, ports, and at both the entrance and
discharge of the rotary kiln. Temperature setpoints in the ignition chamber
(upstream of the kiln) and the mixing chamber (downstream of the kiln) are
maintained by adjusting refuse feed rates, the speed of the reciprocating feed
grates and to a lesser extent, the kiln speed. There is no secondary air
injection with the exception of side wall ports in the ignition chamber.
4-49
-------
(_3
Z
O
o
l~-l
oo
CJ
h—I
o:
Q-
<_>
O
oo
CO
CO
<
^
OO CM
Z. O
0
OO l~^
00
i— i 4->
UJ
E
O Q._
O Q-
"^ C£
Z.
O
1 — 1
1
1
UJ
•D
^y
i — i
1—
Z.
UJ
CO
CtL
O
OO
Oi
UJ
CO UJ
s: a;
<: =D i—
3: >— uj
(_J tf. CD
o: oi
CJ3 LU -<
z CL I—
•-i 2:
X UJ
1— 1 |— —
UJ
1— DC ,
UJ ;3 1—
1 J__ Z
, . ^^ c i
UJ°-
PL_ p^ r"~
H-
O
»-H
1—
0
0
CJ
l~-^ CM •*
i— 1 ^ OO
r-H CM CO
4—
4-
0
u_
o
0
0
oo
1 — t
1
^^
' — t-J
03 o
E CM
£- 00
o cn
^
u_ o
o C\J
LT) O
r^ co
u-> ~^
I— 1
i-i r-^CMLD^H CDOCOCO
CO COir>OO^" CO CM t— 1 CM
cn cy sj>
> > CD i-l CM >
C
O
••-
0
OJ
•I-J
c~
Ol HH
C
o cu
+-> 0
vo r*^« co ^
r-H i — 1 i~H -^C »~H «~H i-H ^^
c
0 TD
•r- CU
O ro O
O) t- •!-
•1-3 -O -I-5
C >, 0
cu i— i ^n a>
C -r-j
o a> +-> c
4-> O O >— i
in to 3
a> c a cu
E s- ' E
•r-3 C. ••-
—1 U_ i— i _l
i — i —
n3 fO
E E
t_ t-
O O
Z Z
^ ^
U_ O 1 i O
O ^* O ^J"
CD rH O t— 1
CO —r CO -^
^ m
CO t-~ CM
i-H ^f i— 1
>— I CO ^t"
r-x oo cn
c:
O
••-
CJ
O)
•r-3
f~
ai i— i
C"
O O)
4-> O
(/) to
<3J C
E 5-
•i — 13
—1 U_
1 1
o
CD
CD
<-<"> .-^
•-I c_)
o
i VO
1 — 1
SCO
0
1 *^
^-*
1 i O
0 CVJ
LT5 CD
r^ co
10
CM
cn
CM
CJ>
4-50
-------
The facility fails to satisfy a key requirement of GCP (good mixing at
adequate temperature) with its current design. The absence of good combustion
air distribution and control results in relatively high fluctuations in 02 and
CO concentrations. Although the units can clearly achieve the necessary
temperatures to destroy trace organics, failure to provide good mixing at the
necessary temperature results in poor CO oxidation. It is probable that the
units could make the necessary modifications in design and operation to
correct these problems.
4.1.5.2 Mass Burn Refractory Compliance Test Results.
The McKay Bay MWC, located in Tampa, FL comprises four 250 tpd (227
Mg/day) units that began operating in 1985. The units are equipped with waste
heat boilers and ESP controls. Compliance tests were performed in 1986 and
average CO emissions were 32-35 ppm for each of the four units.24 Limited
information is available describing the process control data during the tests.
The units reportedly operate on a manual control scheme to maintain 02 levels
in flue gases between 8-12 percent, dry basis. This range of excess oxygen is
lower than concentrations reported for most non-heat recovery refractory wall
MWCs and it corresponds to the range established for good combustion practice
recommendations. The very low CO levels indicate that the units have the
ability to achieve good mixing at adequate furnace temperatures. It is not
possible to speculate on the stability of the combustion process over long
term operating periods. However, the units are equipped with steam air
preheaters which provide the ability to address operating problems that may be
encountered with wet waste. These operating provisions, which are rarely
found at refractory wall MWCs, are necessary to ensure good combustion
practice.
Additional compliance test results are available from three old
refractory wall incinerators (Philadelphia Northwest and East Central plants
and Grosse Point/Clinton, Michigan plant). 24-33 None of these plants satisfies
the design and operating requirements of GCP, and the data are not useful for
establishing emission levels representing good combustion practice. Both of
the Philadelphia plants were permanently closed in 1988. Short term average
CO emissions from the plants ranged from a low value of 51 ppm at Philadelphia
East Central #2 to a high of 821 ppm at Philadelphia Northwest #2. Emissions
from the Grosse Point/Clinton MWC were reported to be 376 ppm.
4-51
-------
4.1.5.3 Conclusions.
The McKay Bay data provide some evidence that new refractory wall
combustors can be designed and operated to achieve CO emissions in the range
typically reported for well designed and operated mass burn waterwall units.
Although basic differences in furnace design exist between waterwall and
refractory wall mass burn systems, the principles that control CO burnout
(good mixing at adequate temperature) can be incorporated into the design of
new units in both combustor classes. Thus, no design limitations are expected
to prevent new refractory wall combustors from achieving CO emission levels
comparable to those from mass burn waterwall units, provided that the
refractory wall units incorporate good combustion practice. This requirement
may force some manufacturers to consider inclusion of additional design
features such as automatic combustion controls to provide for combustion
stability or to consider modifying mixing designs. However, by designing the
combustor to satisfy a basic goal of GCP, good mixing at adequate temperature,
refractory wall combustors are expected to attain CO emission performance
levels similar to those of mass burn waterwall MWCs.
The population of existing refractory wall MWCs includes incinerator
units that vary in age from more than 30 to less than 5 years. Nearly all of
these units will require retrofits to improve operating performance to a level
representing good combustion practice. A discussion of potential combustion
retrofit strategies is included for refractory wall combustors in a separate
report published by EPA.34 In this report it was estimated that existing
refractory wall MWCs could reduce CO emissions to 150 ppm, four-hour average,
by retrofitting to meet good combustion practice requirements. This estimate
was made prior to the completion of the South Dayton parametric test. Results
from that testing program indicate that the estimate was overly conservative
because the unit achieved very low CO levels for many test runs despite
lacking some key components of good combustion practice. It is therefore
judged that implementation of the necessary design and operating modifications
to satisfy good combustion practice would result in the ability of the units
to achieve CO emission levels comparable to new and existing mass burn
waterwall MWCs. It was concluded that the mass burn waterwall CO emission
levels (100 ppm corrected to 7 percent 02, four-hour block average) are also
achievable for new and existing mass burn refractory wall MWCs that apply good
combustion practices.
4-52
-------
4.1.6 Mass Burn Rotary Waterwall MWCs
Limited emissions data are available to characterize the CO emission
performance of mass burn rotary waterwall MWCs, partly due to the few number
of operating facilities in the existing population. However, the units are
sufficiently distinct in their design from conventional mass burn waterwall
combustors that a technology specific performance analysis is warranted. For
example, rotary waterwall combustors are typically designed to operate at
excess air levels near 50 percent, which is a value more commonly associated
with RDF spreader stokers than conventional mass burn waterwall MWCs.1 Grate
burning waterwall units are typically designed to operate at 80-100 percent
excess air. Lower excess air levels reduce the availability of oxygen to
complete the combustion process, which can lead to higher CO emissions. The
available data from a new mass burn rotary waterwall MWC confirm that CO
performance levels are also more typical of RDF spreader stokers than
conventional mass burn waterwall MWCs. The data support establishment of an
achievable CO emission limit equal to that of RDF spreader stokers, 150 ppm at
7 percent 02, four-hour block average. A discussion of the data is presented
below.
4.1.6.1 Dutchess County. NY Compliance Test. The Dutchess County
MWC is designed to burn 510 tpd (464 Mg/day) of MSW in two rotary waterwall
combustors. Each combustor has its own boiler and flue gas cleaning system, a
dry sorbent injection system and a baghouse. Both units were tested for CO as
part of an initial compliance requirement in February 1988.35
The permitted CO emission rate is 13 Ib/hr for each unit. Both units
exceeded the permitted limit during the initial compliance test. Average
emissions were 13.9 Ib/hr (161 ppm at 7 percent 02) for Unit #1 and 13.8 Ib/hr
(156 ppm at 7 percent 02) for Unit #2. Both values are average emission rates
for three one-hour tests.
Two specific design changes were reportedly made to the units which were
intended to improve combustion conditions and lower CO emissions.35 First,
the axial seals, which are used to seal the combustor windbox sections, were
redesigned and replaced. Second, a deflector plate was installed to spread
the bottom ash evenly across the afterburning grate. Following completion of
these modifications, the units were retested in May 1988. Average CO
4-53
-------
concentrations for the units were reported to be 144 ppm and 127 ppm,. both
corrected to 7 percent 02.
A plot of the CO data from the retest is included for both units in
Figures 4-15a and 4-15b. The data include three individual one-hour runs per
unit. Two of the hourly runs include CO spikes which exceeded 600 ppm at 7
percent 02. Review of the 5-minute average 02 concentration confirms that the
CO spikes occurred at low 02 conditions. High heating value fuel can cause a
rapid release of volatiles which depletes local 02 concentrations and results
in the occurrence of oxygen starved, fuel rich gas pockets. If these fuel
rich gases escape the furnace unmixed, high CO spikes will occur. The flue
gas 02 levels at the Dutchess County plant are controlled manually. The
absence of an automatic 02 control loop which can respond to low flue gas 02
concentrations probably contributes to the high CO levels. The new rotary
waterwall designs are expected to incorporate this control feature, which
should lead to more stable CO performance.
4.1.6.2 Cone!usions
The CO data available for mass burn rotary waterwall MWCs are
insufficient to firmly establish a long term achievable emission limit.
However, it is judged that the units can maintain relatively low CO levels by
providing good control of excess air levels and a good mixing design, both of
which are necessary requirements of good combustion practice. Based on the
similarity between rotary waterwall MWCs and RDF spreader stokers with regard
to typical excess 02 design and operating levels, it is judged that rotary
waterwall units can achieve the same long term CO emissions as RDF spreader
stokers, 150 ppm at 7 percent 02, four-hour block average.
4.1.7 Modular Excess Air MWCs
There are currently no long term continuous CO emissions data available
for excess air modular MWCs. The available data base does include short term
data gathered at the Pittsfield, MA facility during a parametric testing
program and data measured during compliance testing at the Pope/Douglas
County, MN MWC. An extensive set of data from Pittsfield confirmed that the
unit was capable of achieving CO concentrations comparable to those from well
designed and operated modular starved air MWCs. Based on the similarities in
design, operation, and emissions performance between excess air and starved
4-54
-------
to
c:
o
O
o
3
o
o
to
V)
JD
LO
O)
t.
3
cn
•r—
U_
ZO %L 00
to
o
t/1
i
o
o
3
O
o
CO
to
-r-
3 C
IT)
i—I
I
*3-
OJ
5-
3
cn
30 %L (2> OO
4-55
-------
air modular MWCs. identical CO emission levels were recommended for the two
technologies. The test results used to establish this conclusion are
summarized below.
4.1.7.1 Pittsfield. MA Parametric Test. The Pittsfield MWC
comprises three units, each rated at 120 tpd (109 Mg/day), with two waste heat
boilers and electrified gravel bed filter controls. Table 4-9 summarizes
average CO emission results from the Pittsfield, MA parametric test on a run-
by-run basis.7 The units were tested while operating at temperatures varying
from 1300-1800°F (704-982°C), and while firing a number of waste fuels with
different characteristics. Two units operated during the test and the third
was maintained on standby. The key operating variable used to control
combustion temperature was the feed input rate. Varying amounts of combustion
air and recirculated air were also used to a lesser extent to vary operating
temperatures.
The data in Table 4-9 clearly confirm that the Pittsfield MWC is capable
of maintaining CO emissions below 20 ppm while operating at design conditions.
Only when the unit was operated at very low temperatures did the CO
concentrations show a noticeable increase above baseline levels.
4.1.7.2 Pope/Douglas County. MN Compliance Test.
The Pope/Douglas County facility includes two modular excess air units,
each rated at 38 tpd (35 Mg/day). Compliance tests were performed in July
1987. One hour carbon monoxide samples were measured and ranged from 20-40
ppm at 7 percent 02 during normal operating conditions. 36
4.1.7.3 Conclusions.
The short duration test results from the Pittsfield and Pope/Douglas
MWCs are not sufficient by themselves to firmly establish a continuous CO
emission limit for modular excess air MWCs. However, the data demonstrate
that excess air modular units and starved air modular units have demonstrated
similar levels of CO emission performance. The two technologies are extremely
similar in their design in that both typically include multiple refractory
lined combustion chambers into which waste is fed and burned in a primary
chamber and the off gases are mixed with air and burned out in a separate
4-56
-------
TABLE 4-9. PITTSFIELD. MA PARAMETRIC TEST RESULTS
TESTING CONDITION
1300°F - MSW
1400°F - MSW
1550°F - MSW
1550°F - MSW + H20
1800°F - MSW
1800°F - MSW, low 02
1800°F - MSW + PVC
1800°F - PVC free
1800°F - PVC free + PVC
1800°F - PVC free + H20
RUN
11
22
28
10
16
15
21
9
14
23
26
18
25
12
17
13
19
24
29
TERTIARY
DUCT
CO*
(ppmv)
152
251
43
8
11
8
25
5
3
3
12
7
4
0
2
4
7
11
6
BOILER OUTLET
02
(%)
13.5
14.8
12.1
12.0
10.8
10.8
12.3
8.7
10.1
8.3
8.0
9.5
10.1
9.4
10.3
10.8
11.2
10.0
10.6
CO*
(ppmv)
111
187
21
13
13
12
16
7
16
8
10
7
8
5
13
20
8
3
10
*CO presented corrected to 7 percent 02.
4-57
-------
secondary chamber. The major distinction between the two technologies is the
stoichiometry in the primary chamber. This oxidation of CO to C02 is achieved
by both combustor types in the secondary combustion chamber, so the
distinction in the primary chamber stoichiometry is not significant in
determining the extent to which the units can achieve low CO levels.
Therefore, the similarity between starved air and excess air modular MWCs in
design and in emissions performance provides the rationale that both combustor
types can achieve CO emission levels of 50 ppm at 7 percent 02, four-hour
block average.
4.1.8 Fluidized Bed Combustors
The use of FBC units in the MWC industry is not widely practiced in the
U.S. at this time. There are currently only two operating facilities, both of
which use bubbling bed technology, in Duluth, MN and La Crosse, WI. Two
facilities are in the advanced planning or construction stage and several
others are in the feasibility study or early planning stage. The available CO
performance data are limited to short term emission compliance tests from the
Duluth and La Crosse units, and short term data from a circulating fluidized
bed unit located in Sundsvall , Sweden. The majority of the data are
comparable to or lower than short term CO concentrations typically measured at
well designed and operated mass burn waterwall MWCs. These results led to the
conclusion that well designed and operated FBC units could achieve a
continuous CO emission limit of 100 ppm, corrected to 7 percent 02, four-hour
average. A discussion of the data used to reach this conclusion is presented
below.
4.1.8.1 Western Lake Superior Sanitary District (WLSSD). Duluth. MN.
The WLSSD facility consists of two identical bubbling bed FBCs with
individual waste heat boilers. Each combustor is capable of firing 120 tpd
(109 Mg/day) of fluff-RDF and 345 tpd (314 Mg/day) of sewage sludge (18
percent solids); thus, RDF represents approximately 26 percent of the total
waste input at full load. A detailed description of the facility design and
operating conditions is provided in a separate FBC technology report which is
published as part of EPA's MWC Technology Assessment.37
4-58
-------
An emissions performance test was conducted on Unit 2 in November
1987.38 During these tests, RDF was fired at 4.7 ton/hr (4.3 Mg/hr) and sewage
sludge was injected at 14.4 ton/hr (13.1 Mg/hr). Available CO data are
summarized in Figure 4-16. Four separate test runs were performed on
consecutive days. Each test run was four hours in duration. Average four-
hour block CO concentrations for the four runs were 16 ppm, 56 ppm. 5 ppm, and
7 ppm, all corrected to 7 percent 02. The 56 ppm mean value included two
measured values of 346 and 369 ppm. The cause of the CO excursions are not
identified. The available data from the unit confirm the ability of the
system to achieve good CO burnout. With the exception of the one run. all
four-hour block averages are below the CO values typically measured at mass
burn waterwall MWCs.
4.1.8.2 Northern States Power French Island Facility. La Crosse. WI.
The NSP French Island Generating Facility in La Crosse. Wisconsin,
comprises two bubbling bed FBCs which were retrofitted to existing boilers in
1981 (Unit 2) and 1987 (Unit 1). NSP began co-firing RDF and wood chips in
the two units (previously fueled with coal) in November 1987. Each of the FBC
units has capacity to fire about 12 ton/hr (10.9 Mg/hr) of RDF and 11 ton/hr
(10.0 ton/hr) of wood chips; with the current operation pattern (16 hr/day. 5
day/wk), the daily capacity of each unit is about 185 tpd (168 Mg/day) of RDF
and 175 tpd (159 Mg/day) of wood waste (approximately 51 percent RDF by
weight). The facility operating permit limits the RDF to a maximum of 50
percent of the heat input.
An emissions performance test was conducted on Unit 1 in May 1988.39
During the test, the unit heat input was evenly distributed between RDF and
wood chips. Sampling was conducted in the stack, downstream of the gravel
bed electro-scrubber. Average CO emissions from the unit were 296 ppm,
corrected to 7 percent 02. It is probable that design constraints resulting
from the retrofit of the boiler limit the ability of the unit to achieve good
CO burnout. Less than 9 feet (2.7 meters) of furnace height separates the top
of the bed from the entrance to the first convective section. This relatively
short freeboard limits the time at which combustion gases emitted from the bed
are subjected to high temperatures. By contrast, the WLSSD freeboard is
approximately 30 feet (9.1 meters) high. It is probable that CO oxidation
reactions are quenched when the combustion products reached the convective
4-59
-------
f-
co
o>
2
o
Z
I
X
o
_l
13
Q
O SB
- - =
" a
ra
co
o>
(O
I
0>
Z
I
X
_l
Q
a
X^
o
CM
.9
E
in
a
nj
•g
n)
uu
£
o
8
a
s
to
o
(u»dd) ZQ%1 OO
O
o
oo
O
o
CM
e
in
i
a>
T"
«
I
O
o
I
X
13
o
4->
c/)
o
o
8
(ludd) jQXt @ OO
4-60
-------
sections of the boiler, resulting in incomplete CO oxidation. This condition
is unique to the French Island boilers and is not expected to be encountered
in units that are designed to fire RDF in the future. The data were not
considered in determining the achievable CO emission limit for FBC units.
4.1.8.3 Sundsvall. Sweden CFB Test Program.
The Sundsvall CFB is a 20 MW boiler manufactured by Gotaverken. The
unit is equipped with an ESP and a baghouse in series. Results from 17
performance tests at the unit were supplied to EPA by Gotaverken during which
multi-pollutant emissions testing was performed at various operating loads.37
The tests ranged from 1-3 hours duration. Average CO emissions in the stack
ranged from 5-59 ppm, corrected to 7 percent 02. Documentation describing
process operating conditions for the tests is very limited. However, the data
indicate that CFB units are capable of achieving low CO levels over short
operating periods.
4.1.8.4 Conclusions.
The CO data available from FBC units are very limited. However, the
short term data from one bubbling bed unit and one CFB unit provide evidence
that the units can be designed and operated to achieve low CO concentrations.
The achievable CO emission limit is established for FBC units largely by use
of these limited data and by using engineering judgement. The short term CO
data are comparable to data from other units for which long term continuous
data are available (mass burn waterwall MWCs). It is judged that by providing
the necessary design, operation and control measures associated with good
combustion practices, FBC units can achieve CO levels of 100 ppm, corrected to
7 percent 02, four-hour block average, on a continuous basis.
4.2 Operating Load
A large number of combustor operating parameters, including furnace
temperature, mixing, and combustion air flow rate, are impacted by operating
load levels. Because these parameters have a direct effect on pollutant
emission levels, it is necessary to maintain an appropriate operating load
range as part of good combustion practice. Low load operation is limited by
minimum temperature and mixing constraints. For example, as waste feed rates
are reduced, it is necessary to make corresponding reductions in combustion
4-61
-------
air flow rates to maintain stoichiometry and furnace temperature. As load
continues to drop furnace temperature also decreases (Figure 4-17), and as air
flows are reduced, a point is reached at which the overfire air system does
not provide adequate mixing. The combined deterioration in mixing performance
and temperature will lead to increased CO and organic emissions.
Low load operating limits are incorporated indirectly in GCP by way of
the CO emission limits. Figure 4-18 illustrates the relationship between
steam load and CO flue gas concentrations for a mass burn waterwall unit. As
the operating load is reduced, the temperature and mixing characteristics
diminish, resulting in increased flue gas CO concentrations. This
relationship applies in all MWCs. Therefore, the existence of a CO emission
limit precludes the need for a lower operating load limit.
At operating loads above design ratings the furnace may be overcharged,
volumetric flow rates are increased, solid and gas residence times are
reduced, and increased amounts of unburned gas and solid organic materials may
be carried out of the furnace prior to completing combustion. Recent research
on CDD/CDF formation mechanism indicates that emissions of CDD/CDF are
correlated with the amount of particulate matter carryover from the furnace.
Figure 4-19 illustrates the relationship between PM carryover and CDD/CDF in
the stack for a mass burn waterwall combustor and an RDF spreader stoker.40
The amount of PM carryover will vary based on a number of parameters,
including fuel properties and feeding method, excess air levels, and
primary/secondary air ratios. No direct method exists to continuously monitor
PM carryover. Therefore, a maximum operating load level in effect provides a
surrogate limit to address control of PM carryover.
Maximum design loads are established based on calculated volumetric heat
release rates and on established boiler design codes and standards. An upper
load limit is established for MWCs to prevent operation at loads that exceed
design criteria. Limited data are available to characterize the effect of
high load operation on emissions. However, a fundamental requirement of good
combustion practice is that combustors be operated within limits established
by design criteria. This requirement provides the basis for establishing a
maximum operating limit of 100 percent of design steam flow.
4-62
-------
o
o
o
CM
O
o
00
O
O
CD
O
O
Tfr
LL
o
UJ
DC
DC
LU
a.
2
LU
H
O
Z)
o
DC
DC
LU
O
O
CM
O
X
UJ
t.
rtS
•a
o
2:
(O
o
M-
-------
0
Q
n.
ni|»i
EEa
nigf
a
EL_
^"^•^^
LIU
El
_
cP
kJ
LJ
a
D D
a
a
i , • ,
V — »
•
—
_ 0 ±i
o^ ^
as
a.
03
o
TJ
CD
^•^
O CO
CO ^
c
CD
5f
0
a.
x^^x
. o a
i^ <
0
_j
m ^^^
^>
~j
^^
in
c3 i
CD (/)
*
O
un
D O O O
D O 0
o eg T-
—^
Ji.
S
fd
s_
rO
OJ
Q.
0
H-
O
-P
O
O)
M—
M-
UJ
00
1
s_
CD
•r~
Ll_
COAV dnOH 3NO 'ZO
QO
4-64
-------
(/I
c
o
in
in
Q
o
Q
Q
O
-a
c
O>
>
O
>,
?^
O)
a>
CQ
Q.
•i —
jr
co
c
o
•I —
-M
res
ai
OO
LT\ "3-
(spu-esnom)
9Snj3y/SUIXOIQ
4-65
-------
The operating load limit can be verified in heat recovery units by
continuously monitoring steam production rates. All MWCs that generate steam
are expected to continuously monitor steam flow rates and/or pressures and
temperatures. Steam flow or pressure setpoints are maintained automatically
by adjusting combustion air flows and feed rates. No convenient continuous
verification method exists for facilities that do not produce steam. Thus,
the operating load can be measured only at the following MWC design types:
• mass burn waterwall boilers
• RDF spreader stoker boilers
• rotary waterwall boilers
• RDF co-fired boilers
• FBC boilers
• CFB boilers
• mass burn refractory and modular combustors with waste heat
boilers
Normal steam production rates are subject to slight variation from
setpoints despite the automatic controls in place at most MWCs. Therefore, it
is necessary to include an averaging time with the operating limit to allow
for initiation of corrective action in the event of upset conditions. A four-
hour block average is selected because it is of short enough duration to
encourage prompt corrective action to be taken in the event of an upset
condition, and it is consistent with other data averaging periods in the
combustor operating standards.
4.3 Downstream Temperature Control
Recent research findings indicate that CDD/CDF formation can occur on
fly ash particles in the presence of excess oxygen at temperatures in the
range of 392-752°F (200-400°C), with maximum formation rates occurring near
572°F (300°C).10-11 A number of variables, including oxygen content, fly ash
carbon content, catalysts, moisture, residence time, etc., appear to influence
the rate of the formation reactions. The effects of these parameters on
CDD/CDF formation have been measured in the laboratory in controlled
experiments. In addition, multi-point emission testing results from full
scale MWCs provide further evidence of the occurrence of low temperature
CDD/CDF formation.6.7,8.9 The full scale data suggest that the particulate
matter (PM) control device operating temperature plays a key role in
4-66
-------
determining the extent to which CDD/CDF is either generated and released to
the atmosphere or captured on fly ash and removed in the control device.
Figure 4-20 illustrates the relationship between PM control device
temperature and CDD/CDF capture efficiency for four MWCs. Each unit is
equipped with ESP controls. The mean ESP operating temperatures range from
435°F (224°C) to 590°F (310°C). The temperature values are averaged over the
duration of the sampling runs, which varied in duration from three to six
hours. The data clearly show that negative removal efficiencies were measured
(CDD/CDF concentrations increased through the ESP) during runs with ESP
temperature greater than 500°F (260°C). At ESP operating temperatures between
450 and 500°F (232 and 260°C). both positive and negative removal efficiencies
were observed. When the ESP operating temperatures were maintained below
232°C (450°F), positive CDD/CDF removal efficiencies were measured for all
runs.
Based on the data in Figure 4-20, it is recommended that all MWCs
maintain flue gas temperatures below 450°F (232°C) at the inlet of particulate
control devices in order to minimize the potential for occurrence of CDD/CDF
formation, and that the flue gas temperature be monitored continuously at this
location. The operating temperature limits will not totally eliminate CDD/CDF
formation; the formation temperature window will be shifted to upstream
portions of the system. However, the temperature limit will help to minimize
the CDD/CDF formation reaction process by avoiding increased residence times
such as would occur in the ESP or fabric filter at temperatures where
formation of CDD/CDF has been observed. In fact, the available data indicate
that operating a PM control device at temperatures below 450°F (232°C) can
provide removal of CDD/CDF.41
Although several MWCs currently operate ESPs at temperatures below 450°F
(232°C), there is little information available on long term performance of low
temperature operation. In general, ESP corrosion concerns should not exist
provided that gas temperatures are maintained above 350°F (177°C).42 Normal
PM control device operating temperatures will typically be lower than 300°F
(149°F) for those MWCs equipped with acid gas controls. The 450°F (232°C)
limit is appropriate for systems that do not use acid gas controls because it
provides a 100°F (56°C) operating window above the temperatures where
corrosion concerns may exist, thus allowing for temperature fluctuations which
result during a normal operating campaign cycle.
4-67
-------
®
, <-
4
<• ^
•4
•« 4 •«
I 1 1
A
v
•
A
V
•
•
•
•
V •
o e
«
A C -
• v
ft
«
0
1 1 1 1 1 1 1 1 1 1
O
- O)
10
o
- f-
to
-
o
- to
to
0
- o
U)
o
- T-
to
o
r °*
^
o
»~
*— p*»
^
o
- to
«r
0
L 0
^J-
3OOOOOOOOOOOOOOO
OtD^fM NfCDCOOCM^COCOOCM
1 1 1 1 V V V- T- V- CM 01
I I I I I I 1
ul
R)
V
0
•o
*^
0
_c
a.
CO
UJ
a
e
L.
3
(8
^
e
a
£
e
1 _
r~
*
•
•
*
•
O '
e
=
•
jf
•
•
a.
<4
..
>
0
c
<
c
z
•
c
3
O
U
•
•
•
£
&
0
o
k_
a
"to
*_
«
a
E
0)
1-
<—
^o
JC
a.
en
UJ
**•
o
c
JD
O
C
3
li-
re
(A
re
>»
o
c
jot
o
z+z
••—
UJ
"re
o
0
CC
U.
O
o
^.
o
o
0
o
CM
1
-------
Continuous achievement of the temperature limit will pose no problem for
new MWCs because the temperature limit can be incorporated into the design of
the system, and in most cases the design will include acid gas controls.
Existing MWCs equipped with water spray chambers or lime slurry spray dryers
can maintain the PM control device inlet temperature by establishing adequate
temperature setpoints in the existing control system. However, a number of
existing MWCs without acid gas controls are designed with economizer outlet
temperatures above the recommended temperature limit. Design retrofits would
be required for many of these existing units to achieve the temperature limit.
An averaging time must be included with any limit due to the
fluctuations that occur in economizer exit gas temperature during normal
operating conditions. Changes in operating load, use of air preheat and
cleanliness of heat transfer surfaces will result in economizer exit gas
temperature variations over a normal campaign cycle. As fly ash adheres to
boiler tubes the system heat transfer characteristics diminishes, resulting in
less heat removed by the boiler and slightly higher flue gas exit
temperatures. Periodic sootblowing removes fly ash from heat transfer
surfaces. A four-hour block average is recommended because it is short enough
in duration to encourage relatively prompt response action in order to meet
the operating requirement. Therefore, an operating temperature limit of 450°F
(232°C) maximum, four-hour block average, at the PM control device inlet is
achievable for all MWCs. The four-hour block average is also consistent with
other averaging times for parameters included in the combustion operating
standards.
4-69
-------
5.0 References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
Seeker. W.R., W.S. Lanier, and M.P.
Combustion Study: Combustion Control
Minimize Emission of Trace Organics."
1987.
Heap. "Municipal Waste
of MSW Combustors to
EPA/530-SW-87-021c. May
Schindler,
Combustion
1989.
P.J. "Municipal Waste
Control at New Facilities.
Combustion Assessment -
EPA-600/8-89-057 . August
Schindler, P.J. "Municipal Waste Combustion Assessment -
Combustion Control at Existing Facilities." EPA-600/8-89-058.
August 1989.
Kullendorf, A., B. Oscarsson, and C. Rollan. "Gotaverken CFB
Boiler: An Environmentally Safe Solution to Our Waste Disposal
Crisis." Fourth Solid Waste Management and Materials Conference,
New York, NY, January 1988.
Barton, R.G.. W.
Emissions During
Meeting - Western
1988.
Clark, W.S. Lanier, and W.R. Seeker. "Dioxin
Waste Incineration." Presented at 1988 Spring
States Section/The Combustion Institute. March
Environment Canada. National Incinerator Testing and Evaluation
Program. "Two Stage Combustion." Summary Report, EPS 3/UP/l
September 1986.
Midwest Research Institute. "Results of the Combustion and
Emissions Research Project at the Vicon Incinerator Facility in
Pittsfield, MA." Prepared for New York State Energy Research and
Development Authority. June 1987.
Radian Corporation. "Municipal Waste Combustion Multipoll utant
Study - Summary Report." North Andover RESCO, North Andover, MA.
EMB Report No. 86-MIN-02a. March 1988.
Entropy Environmentalists. "Stationary Source Sampling Report
Pinellas County Resource Recovery Facility." St. Petersburg, FL.
February and March 1987.
Stieglitz, L. and G. Vogg. Formation and Decomposition of
Polychlorodibenzodioxins and -furans in Municipal Waste. Report
KFK4379. Laboratorium fur Isotopentechni k. Institut fur Heize
Chemi, Kernforschungszentrum Karlsruhe. February 1988.
Hagenmaier. H. et al . "Catalytic Effects of Fly Ash from
Waste Incineration Facilities on the Formation and Decomposition
of Polychlorinated Dibenzo-p-dioxins and Polychlorinated
Dibenzofurans." Environmental Science Technology. November 11,
1987, Vol. 21, 1080-1084.
5-1
-------
12. Entropy Environmentalists, Inc. "Emissions Test Report -
Municipal Waste Combustion Continuous Emission Monitoring Program.
Wheelabrator Resource Recovery Facility, Millbury. Massachusetts."
EMB Report 88-MIN-07c. January 1989.
13. Entropy Environmentalists. "Municipal Waste Combustion
Multipol1utant Study: Emission Test Report - Wheelabrator
Millbury, Inc. Millbury. MA." EMB Report No. 88-MIN-07. July
1988.
14. Letter and Attachments from Moon S. Chung, Los Angeles County
Sanitation Districts, to Jeff Telander, U.S. EPA/OAQPS, dated June
12,1989.
15. Telecon. Peter Schindler, Energy and Environmental Research
Corporation, and Ed Wheless, Commerce Refuse-to-Energy Facility.
June 29. 1989.
16. Telecon. Peter Schindler. Energy and Environmental Research
Corporation, and Moon Chung. Los Angeles County Sanitation
District. July 7, 1989.
17. Environment Canada. NITEP. "Environmental Characterization of
Mass Burning Incinerator Technology at Quebec City." Summary
Report. EPS 3/UP/5. June 1988.
18. Schindler. P. "Site Visit Report Summary - Hampton. VA Steam
Plant." Submitted to U. S. EPA/OAQPS on December 22. 1988.
19. McDannel , M.D.. and B.L. McDonald. "Combustion Optimization
Study at the Commerce Refuse-to-Energy Facility." Volume
I. ESA 20528-557 Prepared for County Sanitation
Districts of Los Angeles County. Los Angeles. CA. June
1988.
20. Letter and Attachments from Frank Visser, Osewego County
Department of Public Works, Fullton, NY. to Susan Agrawal. Energy
and Environmental Research Corporation, dated June 1. 1989.
21. Radian Corporation. "Results From the Analysis of MSW Incinerator
Testing at Oswego County, New York." Prepared for New York State
Energy Research and Development Authority. March 1988.
22. Savage, G.M., D.L. Bordson. and L.F. Diaz. "Important Issues
Related to Air Pollution at Municipal Solid Waste Facilities."
Environmental Progress. Vol. 7, No. 2, May 1988.
23. Telecon. Dick Scales, Consumat Systems. Inc., and Peter
Schindler, Energy and Environmental Research Corporation. July
19. 1989.
24. Midwest Research Institute. "Municipal Waste Combustion Study:
Emission Data Base for Municipal Waste Combustors." EPA/530-SW-87-
021b. May 1987.
5-2
-------
25. Letter and Attachments from Mike Hartman, Combustion Engineering.
to Susan Agrawal, Energy and Environmental Research Corporation,
dated June 1. 1989.
26. Letter and Attachments from Carlos White, Penobscot Energy
Recovery Company RDF Plant, to Peter Schindler, Energy and
Environmental Research Corporation, dated June 1. 1989.
27. Radian Corporation. "Municipal Waste Combustion Multi-Pol 1utant
Study: Refuse-Derived-Fuel Summary Report." Mid-Connecticut
Resource Recovery Facility. Hartford. CT. Prepared for U.S. EPA,
ORD and OAQPS. EMB Report No. 88-MIN-09A. January 1989.
28. Roy F. Weston, Inc. "Source Emissions Compliance Test Report
Incinerator Units A and B." Prepared for GE Company Penobscot
Energy Recovery Company Facility, Orrington, ME. September 1988.
29. "Municipal Waste Combustion. Multi-Pol 1utant Study. Emission Test
Report. Maine Energy Recovery Company, Refuse-Derived Fuel
Facility, Biddeford ME, Volume I. Summary of Results." EPA-600/8-
89-064a. July 1989.
30. Interpoll Laboratories. "Results of the March 21-26, 1988 Air
Emission Compliance Test On the No. 2 Boiler at the Red Wing
Station." Prepared for Northern States Power Company. Report
Number 8-2526. May 10, 1989.
31. Alliance Technologies Corporation. "Field Test Report - NITEP
III. Mid-Connecticut Facility. Hartford. CT." Volume II
Appendicies. Prepared for Environment Canada. June 1989.
32. Radian Corporation. "Site-Specific Test Plan and Quality
Assurance Project Plan for the Screening and Parametric Programs
at the Montgomery County Solid Waste Management Division South
Incinerator- Unit #3." Prepared for U.S. EPA. OAQPS and ORD,
Research Triangle Park, NC. November 1988.
33. Swanson Environmental. Inc. "Emission Compliance Tests - Grosse
Pointes-Clinton Refuse Authority." July 1982.
34. "Municipal Waste Combustors, Background for Proposed Guidelines
for Existing Facilities. " EPA-450/3-89-027e. August 1989.
35. ETS. Inc. "Compliance Test Report for Dutchess County Resource
Recovery Facility." May 1989.
36. Response to Clean Air Act Section 114 Information Questionnaire.
"Results of Non-Criteria Pollutant Testing Performed at Pope-
Douglas Waste to Energy Facility." July 1987. Provided to EPA on
May 9. 1988.
37. Nelson. L.P. "Municipal Waste Combustion Assessment. Fluidized
Bed Combustion." EPA-600/8-89-061. July 1989.
5-3
-------
38. Interpoll Laboratories. "Results of the November 3-6, 1987
Performance Test on the No. 2 RDF and Sludge Incinerator at the
WLSSD Plant in Duluth, Minnesota." Interpoll Report No. 7-2443.
April 25, 1988.
39. Clean Air Engineering. "Results of Diagnostic and Compliance
Testing at NSP French Island Generating Facility Conducted May 17-
19, 1989." July 1989.
40. Maly. P.M., G.C. England, W. R. Seeker. N.R. Soelberg. and D.G.
Linz. "Results of the July 1988 Wilmarth Boiler Characterization
Tests." Gas Research Institute Topical Report No. GRI-89/0109,
June 1988-March 1989.
41. Radian Corporation. "Results from the Analysis of MSW Incinerator
Testing at Peekskill, NY." Prepared for New York State Energy
Research and Development Authority. DCN:88-233-012~21. August
1988.
42. Sedman, C.B. and T.G. Brna. "Municipal Waste Combustion Study:
Flue Gas Cleaning Technology." EPA/530-SW-87-021d (NTIS PB87-
206108). June 1987.
5-4
-------
APPENDIX A
EXPECTED EXCEEDANCE LEVELS FOR CO DATA
FROM MUNICIPAL WASTE COMBUSTION FACILITIES
Prepared by M.R. Ledbetter
University of North Carolina at Chapel Hill
for
U.S. EPA, Air and Energy Engineering Research Laboratory
EPA Contract 68-02-4269
1. Introduction
The purpose of this analysis is to estimate levels which will be
exceeded at given rates (in the long run) by CO emissions at selected
municipal waste incineration facilities. In each case the basic data
(provided by HER) consists of hourly averages .and it is desired to consider a
variety of exceedance rates including 1/10 yrs, 1/yr, 1% for hourly, 4-hourly
and 8-hourly "rolling" and "block" averages. The three facilities considered
are:
(1) Wheelabrator Resource Recovery Facility, Millbury, MA (here
referred to as "Millbury")
(2) Mid-Connecticut Resource Recovery Facility (here referred to as
"Mid-Conn")
(3) Penobscot Energy Recovery Company, Penobscot, ME (referred to as
"Penobscot")
The most complete analysis has been carried out for the Millbury data
(I). For cases (2) and (3) sample calculations have been done for small
sections of the data, which nevertheless demonstrate the general conclusions.
Since the desired exceedance rates (1% or less) are small, direct
estimates of the corresponding levels by counting exceedances will be
unreliable and indeed virtually impossible for 1/yr rates when the available
data extends over 2 months. Hence the approach taken is to model the data as
a "stationary, normal process" and to calculate the exceedance rates
theoretically from this estimated model. The calculations use the mean,
variance and "lag correlations" estimated from the data. It should perhaps be
pointed out that previous similar studies for S02 emissions have used the more
special so-called "AR-1" models. The advantage of so doing lies only in that
estimation of one parameter (the lag 1 serial correlation) determines all the
A-l
-------
other correlations required, thus providing a neat computation. In the cases
dealt with here, the simple AR-1 model does not fit the data well, but the
correlations needed are readily estimated individually.
A number of data points are either missing or regarded as aberrant for
various indicated reasons. A special program has been written so that the
estimates do not involve these data points, for the complete Millbury
analysis.
The Millbury results are presented in Section 2, and Section 3 contains
the sample results obtained for Penobscot and Mid-Conn. The methods used for
the calculation are detailed in Section 4 of this Appendix.
2. Millbury Data
Hourly CO averages were reported for a 9-week period involving
potentially 1512 data points. Missing and aberrant values were coded as zeros
(by EER, who supplied the data) and a special program was written to obtain
serial correlations using the zeros only to preserve the time sequencing and
not otherwise in the calculations. The values of the mean (p.), variance (a)
and lag correlations ri. r2 ... n, were calculated (technically, estimated)
from the data giving
M. = 37.112 o2 = 37.176 (a - 6.0972)
ri - .78036
r2 - .70094
r3 - .15510
r4 - .59370
r& - .52109
re - .45512
r? - .41337
The standard deviations 01, a4, as of each 1-, 4- or 8-hour average were
calculated by Eqn A.2 using the above correlation giving
oi - 6.0972 a4 - 5.4527 as - 5.0791
A-2
-------
Using these values, the levels corresponding to exceedance rates ranging
from 1 in 100 hrs to 1 in 10 yrs were calculated from Eqn A for 1-, 4-, and 8-
hour (both "rolling" and "block") averaging, giving the values in Table A-l.
A point should be made clear concerning the relationship between the
rolling and block averages. It is commonly supposed that for a given level
the expected "rolling rates" should exceed the block rates for the reason that
one high average will tend to be followed by another. However the argument is
not relevant since also low averages tend to be followed by further low
averages, reducing the rolling rate in a compensatory manner. In fact, for a
given level, the percentage of expected exceedances is precisely the same for
rolling and block averages. The differences in the table arise since the
exceedance rates are reported not as percentages but as 1/100 hr, I/month,
etc. and there are e.g., four times as many rolling as block 4-hour averages
in any given period. For example there are the same number of 4-hour block
averages in a year as there are rolling averages in 3 months so that the same
level 55.2 in Table 1 gives a 4-hour rolling rate of 1 per 3 months and 4-hour
block rate of 1 per year. In both cases this corresponds to the percentage
100 100
3 x 30 x 24 (12 x 30 x 24)/4
assuming 30 day months for simplicity.
As a consequence, it may be regarded as better to report levels for
exceedance rates expressed as percentages, so that the "rolling" and "block"
levels would coincide. For example the level for a 4-hour average at a 1%
rate would be 49.8 for both rolling and block averages. The figure 46.7
corresponds to a percentage exceedance rate of 4% (1/25) since there are 25 4-
hour blocks in 100 hours.
3. Penobscot and Mid-Conn Data
As noted, short periods (approximately 6 days) were analyzed for
Penobscot and Mid-Conn data. Tables A-2, A-3, and A-4 summarize the results
obtained.
A-3
-------
TABLE A-l.
Estimated levels corresponding to stated expected exceedance rates
for 1-, 4-, 8-hour (rolling and block), Millbury CO data
[i - 37.112. al - 6.0972
EXPECTED
EXCEEDANCE
RATE
1/100 hr
1/mo
1/3 mo
1/6 mo
1/yr
1/10 yr
1-HOUR
ROLLING
/BLOCK
51.3
55.4
57.3
58.5
59.6
62.9
4-HOUR
ROLLING
49.8
53.4
55.2
56.2
57.2
60.2
4-HOUR
BLOCK
46.7
51.0
53.0
54.1
55.2
58.4
8-HOUR
ROLLING
48.9
52.3
54.0
54.9
55.8
58.6
8-HOUR
BLOCK
44.3
48.8
50.7
51.9
52.9
56.1
A-4
-------
TABLE A-2.
Estimated levels corresponding to stated expected exceedance rates
for I-, 4-, 8-hour (rolling and block). Penobscot CO data
M. - 134.57, al - 68.1863
EXPECTED
EXCEEDANCE
RATE
1/100 hr
1/mo
1/3 mo
1/6 mo
1/yr
1/10 yr
1-HOUR
ROLLING
/BLOCK
293
339
361
374
386
423
4-HOUR
ROLLING
265
303
321
332
342
373
4-HOUR
BLOCK
233
278
298
310
321
355
8-HOUR
ROLLING
253
287
304
313
322
350
8-HOUR
BLOCK
206
251
271
283
293
325
A-5
-------
TABLE A-3.
Estimated levels corresponding to stated expected exceedance rates
for 1-, 4-, 8-hour (rolling and block). Mid-Conn Unit 12 CO data
M. = 161.75. CTI - 58.908
EXPECTED
EXCEEOANCE
RATE
1/100 hr
1/mo
1/3 mo
1/6 mo
1/yr
1/10 yr
1-HOUR
ROLLING
/BLOCK
299
338
357
368
379
411
4-HOUR
ROLLING
211
297
311
319
328
352
4-HOUR
BLOCK
211
276
292
302
311
338
8-HOUR
ROLLING
253
279
292
299
306
328
8-HOUR
BLOCK
217
251
267
276
294
308
A-6
-------
TABLE A-4.
Estimated levels corresponding to stated expected exceedance rates
for 1-, 4-, 8-hour (rolling and block). Mid-Conn Unit 13 CO data
H = 209.79, al - 52.715
EXPECTED
EXCEEDANCE
RATE
1/100 hr
1/mo
1/3 mo
1/6 mo
i/yr
1/10 yr
1-HOUR
ROLLING
/BLOCK
332
368
385
395
404
433
4-HOUR
ROLLING
283
340
314
320
326
343
4-HOUR
BLOCK
265
290
302
308
314
333
8-HOUR
ROLLING
247
258
263
266
269
278
8-HOUR
BLOCK
232
247
253
256
260
270
A-7
-------
4. Model and Calculations
Let Xn denote the measured average CO for hour n. n-1. 2.... Assume Xn
is a stationary normal series with mean n, variance a2 and log correlation
rj = corrn (Xi, Xi+j). Write
1 n
Xn - --- IX i - average of Xi Xn
n 1
a2 - var of Xn (ai - a)
za - (1-a) percentile of standard normal distribution (e.g.. z.05=1.65,
z.oi = 2.33)
Xn - desired exceedance rate for averages of length n = 1, 4. 8 hrs
(e.g., an - .01 for rate of 1/100 hrs. rolling basis)
Cn = level giving the exceedance rate of an
The purpose of the analysis is to express Cn in terms of an, p., a2 and
the correlations rj. Now
an - expected number of exceedances of Cn per unit time by Xn
Cn - u.
PIXn > Cn} - 1 - fl-
where O is the standard normal distribution function. Hence. (Cn - n)/an =
Zan or
Cn - U, + On Zan (A. 1 )
To complete the calculation an is expressed in terms of a and the lag
correlation as follows:
1 n
on2 - var — Z Xi
n 1
a2 2a2 n
+ I (n-j)rj
n n2 j-1
a2 2 n
I + ... j; (n-j)rj (A.2)
n n j-1
A-8
-------
The procedure used to calculate the entries in the tables is to
(a) estimate \i, a2, rj, l
-------
Mid-Conn Data. Unit 12
M. = 161.745 ol2 - 3.470
Estimated lag correlations
j 1 23 4 56 7
rj .4659 .4478 .3790 .2952 .1925 .1704 .2224
01 - 58.91, a4 - 45.02, c8 - 39.19
Mid-Conn Data. Unit 13
H = 209.786 o2 - 2778
Estimated lag correlations
j 1 2 3 4 56 7
rj .2647 .1150 -.1562 -.4269 -.1793 -.2135 -.1364
dl - 52.71. a4 - 31.56, a8 - 1601
The listed exceedance rates correspond to the quartiles zan as follows:
Exceedance rate: one per
Type of Avg: 100 hr 1 mo 3 mos 6 mo 1 vr 10 .yr
8 hr rolling
4 hr block
8 hr block
2
1
1
.3263
.7507
.4051
2.
2.
2.
9955
5440
2918
3.
2.
2.
3160
9068
6826
3.
3.
2.
5050
1171
9068
3
3
3
.6853
.3160
.1171
4
3
3
.2351
.9126
.7419
A-10
-------
TECHNICAL REPORT DATA
rraJ /nilmftiom on ifir r(\rn< t>rjr>fr c
i •>« PO«T f.O
EPA-60P/8-89-063
AND SUI'I f I 1
3 H« C'«"'« NT S ACCtSSlCN NO
Municipal Waste Combustion Assessment:
Technical Basis for Good Combustion Practice
August 1989
P.O. Schindler
L.P. Nelson
8 PIHIQHMINC, C"G*Mr'GN H I »> C f T N. C
9. Pl«^O«MING ORGANISATION IsrAMt
Energy and Environmental Research Corporation
3622 Lyckan Parkway. Suite 5006
Durham. NC 27707
NO
II CCNT«ACT/GAANT NO.
68-03-3365
12. SPONSORING *G£NC> NAM« AND AOO«tSS
13 TV»>£ OF
cove*»eo
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park. NC 27711
SPONSORING AGtNCv COCt
IS. SUPPLCMeNTARV NOTES
Project Officer
- James D. Kilgroe
16. ABSTRACT
The EPA's Office of Air Quality Planning and Standards (OAQPS) is
developing emission standards and guidelines for, respectively, new and
existing MWCs under the authority of sections lll(b) and lll(d) of the Clean
Air Act (CAA). The EPA's Office of Research and Development (ORD) is
providing support in developing the technical basis for good combustion
practice (GCP). which is included in the regulatory alternatives considered in
selecting the proposed standards and guidelines. This report defines GCP and
summarizes the approach used to implement GCP into the proposed MWC standards
and guidelines. The report identifies the minimum subset of GCP operating
parameters that can be continuously monitored to ensure that the goals of GCP
are achieved. Finally, the report provides a detailed description of the data
and rationale used to establish quantitative operating limits for each of the
continuous operating parameters.
17.
K£Y WO«OS AND OOCUM€NT ANALYSIS
b.lO€NTI*l£BS/OPEN CNOCO TtflMS C. COSATI \ Kld.GlOup
. DISTRIBUTION
19 StCU«iTv CLASS i Tint Krponi
Jl NC C'
JO StCU«iTv
, Tun
PA f •— 2:30-1 if)... 4-771 «"-«v,oo» »ci tiox •» o«»oc* '«
------- |