EPA-600/R-94-166
September 1994
ESTIMATE OF METHANE
EMISSIONS FROM U.S. LANDFILLS
by
Michiel R.J. Doom
E.H. Pechan & Associates, Inc.
3500 Westgate Drive, Suite 103
Durham, North Carolina 27707
Leonard A. Stefanski and Morton A. Barlaz
North Carolina State University
Raleigh, North Carolina 27695
EPA Contract No. 68-D1-0146,
Work Assignments No. 15, No. 22, No. 31, and No. 34 (Pechan)
EPA Contract No. 68-D9-0147,
Work Assignment No. 34 (NCSU)
Project Officer:
Susan A. Thorneloe
Air and Energy Engineering Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Prepared for:
U.S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse be/ore complet
PEPOnT MO.
EPA-600/R-94-166
t. TITLE AMD SUBTITLE
Estimate of Methane Emissions from U.S. Landfills
7 authoris] M.R.J. Doom (E. H. Pechan); and I. A.
Stefanski and M. A. Barlaz (K. C. State Univ.)
S. REPORT DATE
September 1994
6. PfcHFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
E. .H. Pechan and Associates, Inc., 3500 Westgate
Drive, Suite 103, Durham, North Carolina 27707
North Carolina State University, Raleigh, North
Carolina 27695
10. PROGRAM ELEMtNT NO.
11. CONTRACT/GRANT NO.
68-Dl-0146, Tasks 15, 22, 31,
and 34 (Pechan)*
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT ANO PERiOD COVEREO
Task final; 12/92-1/94
14. SPONSORING AGFNCY CODE
EPA/600/13
15. supplementary notes project officer is Susan A. Thorneloe, Mail Drop 63,
919/541-2709. (*) Effort by N. C. State University under contract 68-D9-0141/, Task
21:
i6. abstract rep0rt describes the development of a statistical regression model
used for estimating methane (CH4) emissions, which relates landfill gas (I.FG) flow
rates to waste-in-place data from 105 landfills with LFG recovery projects. (NOTE:
CH4 flow rates from landfills with .1 .FG recovery systems can be used as surrogates
for CTI4 generation and successively for CII4 emissions.) The model has. three lin-
ear segments, each of which applies to a distinct landfill size class. Assumptions
were required to account for the recovery efficiency of LFG projects and for the
probable oxidation of CH4 in the top soil cover of the landfill. National CH4 emis-
sions may be estimated by applying the regression model to municipal-waste-in-
place data for U.S. landfills collected in 1986 by EPA's Office of Solid Waste (OSW).
This value is adjusted for CH4 emissions from industrial landfills and Cli4 which is
currently recovered or flared. For 1986, CII4 emissions from U.S. landfills were
estimated at 11 tg/yr with lower- and upper-bound values of 7 and 15 tg/yr, respec-
tively. For 1992, estimates were between 9 and 18 tg/yr. The report details uncer-
tainties which limit the quality of the above estimates. The report concludes with a
discussion of trends which will affect future LFG emissions, as well as 1/FG utili-
zation.
17.
KEY WORDS AND DOCUMENT ANALYSIS
a.
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Held/Group
Pollution
Regression Analysis
Pollution Control
13 B
Earth Fills
Wastes
Stationary Sources
13 C
Methane
Municipal Waste
07 C
Emission
14G
Estimating
Mathematical Models
12 A
13. DISTRIBUTION STATEMENT
19. SECURITY CLASS (This Report)
21. NO. OF PAGES
Unclassified
64
Release to Public
20. SECURITY CLASS (Thispage)
22. PRICE
Unclassified
EPA Form 2220-1 (9-73)
f
-------
NOTICE
This document has been reviewed in accordance with
U.S. Environmental Protection Agency policy and
approved for publication. Mention of trade names
or commercial products does not constitute endorse-
ment or recommendation for use.
-------
ABSTRACT
The estimation of U.S. methane (CH4) emissions from landfilled waste is part of a
bigger effort by the U.S. Environmental Protection Agency's Air and Energy Engineering
Research Laboratory (EPA/AEERL), Global Emissions & Control Division, to obtain global
greenhouse gas emissions data. Methane flow rates from landfills with landfill gas (LFG)
recovery systems were used as surrogates for CH4 generation and successively for CH4
emissions. AEERL collected data on 112 U.S. LFG recovery projects, 105 of which are
included in the "ORD Database."
The development of a regression model relating LFG flow rates to waste in place
data from the ORD Database, is described in this document. The model has three linear
segments, each of which applies to a distinct landfill size class. Correction factors were
used to account for the recovery efficiency of LFG projects and for the probable oxidation
of CH4 in the top soil cover of the landfill.
In 1986, the EPA's Office of Solid Waste (OSW) conducted a survey, in which
detailed information on 1,175 U.S municipal landfill facilities was collected in the "OSW-
Westat Database." This population was designed to be a stratified random sample so its
data can be extrapolated by means of scaling factors, to obtain total waste in place for
active U.S. municipal solid waste landfills. This database contains data which make it
possible to estimate waste in place by two different methods. The method which is based
on the difference between design capacity and remaining capacity of the landfill, appears
to be the more appropriate. The total waste in U.S. landfills in 1986 was 4.7* 1016 g
(5.2*10® tons). The yearly disposal rate (1986-1992) was estimated to be 248 tg/yr
(273 tons/yr).
Application of the regression model to the waste in place data calculated from the
OSW-Westat Database yields national CH4 emissions from landfills. Methane emissions
from industrial landfills were added to this value and CH4 emissions which are currently
recovered or estimated to be flared were subtracted. For 1986 CH4 emissions from U.S.
landfills were estimated to range from 7 to 15 tg/yr (8 to 16 tons/yr) with a mid-point of
11 tg/yr (12 tons/yr). Methane emissions in 1992 were estimated to be between 9 and 18
tg/yr (10 and 20 tons/yr) with a mid-point of 13 tg/yr (15 tons/yr).
The report details uncertainties which limit the quality of the above estimates.
The main uncertainty arises from the inability to perform quality assurance on the OSW-
Westat Database, as it exists today. The report concludes with a discussion of trends
which may affect future LFG emissions, as well as LFG utilization. Upcoming regulations
for controlling air emissions from new and existing landfills are expected to significantly
reduce LFG emissions.
i i i
-------
CONTENTS
Page
ABSTRACT ii
FIGURES vi
TABLES vi
ABBREVIATIONS, SYMBOLS, AND DEFINITION OF MSW Vl i
ACKNOWLEDGEMENTS i
INTRODUCTION 1
BACKGROUND 1
METHANE PRODUCTION FROM THE ANAEROBIC DECOMPOSITION
OF SOLID WASTE 4
METHANE POTENTIAL OF MUNICIPAL SOLID WASTE IN LANDFILLS . . 6
MODEL TO ESTIMATE MUNICIPAL LANDFILL METHANE EMISSIONS 9
REGRESSION MODEL 9
CONVERSION FACTOR 11
ESTIMATE OF TOTAL WASTE IN PLACE 12
MUNICIPAL LANDFILLS 12
Magnitude of Waste Quantity 14
INDUSTRIAL LANDFILLS 15
ESTIMATE OF U.S. LANDFILL METHANE EMISSIONS 16
DEVELOPMENT OF RANGES 18
UNCERTAINTIES 20
QUALITY OF OSW-WESTAT DATABASE 20
Comparison of RATE and DIFF Methods 20
Scaling Factors 25
UNCERTAINTIES ASSOCIATED WITH THE MODEL 26
Emissions from Small Landfills 26
Bias toward "Rich" Landfills 26
OTHER UNCERTAINTIES 27
Extrapolation to 1993 27
Industrial Waste 27
v
Preceding page blank
-------
Generation Time 27
Variables Used in the Conversion Factor Derivation 27
TRENDS IN WASTE MANAGEMENT AND THEIR IMPACT ON FUTURE
EMISSIONS 29
BACKGROUND 29
TRENDS IN SOLID WASTE GENERATION, COMPOSITION, AND
MANAGEMENT 29
TRENDS IN LANDFILL GAS RECOVERY AND UTILIZATION 32
REGULATORY ISSUES AFFECTING LANDFILL GAS RECOVERY 32
ESTIMATE OF FUTURE EMISSIONS 33
SUMMARY AND CONCLUSIONS 35
REFERENCES 37
APPENDIX A: STATISTICAL METHODS 41
INTRODUCTION 41
ESTIMATING CH4 USING REGRESSION RESULTS 44
Ratio Estimation 44
Regression Modeling 47
APPENDIX B: COMPARISON OF ESTIMATES OF METHANE EMISSIONS
FROM U.S. LANDFILLS 51
APPENDIX C: CALCULATIONS FOR THE DENSITY OF METHANE 52
APPENDIX D: FIELD DATA 53
vi
-------
FIGURES
Number Page
1. Landfill gas recovery sites in the United States 3
2. Landfill gas flow rates versus welled waste for 105 recovery sites and best
fitting regression curve 10
3. Comparison of waste in place data for RATE and DIFF methods for all landfills
from sample population 23
4. Comparison of quotient of waste in place data for RATE and DIFF methods
versus age of landfill from sample population 24
5. Trends in municipal solid waste management, 1960 to 2000 31
6. Landfill gas flow rates versus welled waste for 105 recovery sites and regression
lines for Ratio and Regression methods 45
TABLES
Number Page
1. COMPARISON OF WASTE GENERATION AND WASTE-TO-LANDFILL
RATES FOR EIGHT INDUSTRIALIZED COUNTRIES 14
2. METHANE EMISSIONS FROM INDUSTRIAL LANDFILLS (Metric Units) 17
3. METHANE EMISSIONS FROM INDUSTRIAL LANDFILLS (U.S. Units) 17
4. TOTAL WASTE IN PLACE AND TOTAL METHANE EMISSIONS FROM
MUNICIPAL LANDFILLS PER SIZE CLASS (Metric Units) 18
5. TOTAL METHANE EMISSIONS FROM U.S. LANDFILLS (Metric Units) 19
6. TOTAL METHANE EMISSIONS FROM U.S. LANDFILLS (U.S. Units) 19
7. COMPARISON OF DIFF AND RATE METHODS FOR WASTE IN PLACE
CALCULATIONS 22
8. COMPARISON OF DIFF AND RATE METHODS FOR WASTE IN PLACE
CALCULATIONS: STATISTICAL SUMMARY 22
9. ESTIMATES OF TOTAL WASTE IN PLACE; STATISTICAL
CONSIDERATIONS 43
10. RATIO A AND RATIO ESTIMATES OF METHANE EMISSIONS FOR 1986. ... 46
11. REGRESSION ESTIMATES OF METHANE EMISSIONS, STATISTICAL
CONSIDERATIONS 49
12. COMPARISON OF ESTIMATES OF METHANE EMISSIONS FOR
U.S. LANDFILLS 51
13. WASTE AND LFG FLOW RATE DATA FROM THE ORD DATABASE 53
14. SUMMARY OF LANDFILL DATA FROM PEER ET AL. (1992) 56
vi i
-------
ABBREVIATIONS, SYMBOLS, AND DEFINITION OF MSW
AEERL Air and Energy Engineering Research Laboratory
BMP Biochemical Methane Potential
CFM Cubic Feet per Minute
EPA Environmental Protection Agency
LFG Landfill Gas
MSW Municipal Solid Waste
ORD Office of Research and Development
OSW Office of Solid Waste
RCRA Resource Conservation and Recovery Act
U.S. United States
g gram
kg kilogram, 1,000 grams
Mg megagram, 106 grams
tg teragram, 1012 grams
ft3 cubic feet
1 liter
min minute
ml milliliter
m* cubic meter
yr year
CH4 methane
C02 carbon dioxide
CbH10O_ cellulose
C5H(i04 hemicellulose
DEFINITION OF MSW:
Municipal solid waste includes wastes from residential, commercial, and certain
industrial sources, and does not include construction and demolition wastes, sludges,
power plant ashes, hazardous wastes and industrial process wastes. (Industrial sources
producing MSW may be workshops and other small industries which are typically found
among commercial or residential sources, where the waste is comparable to commercial or
residential waste in composition, as well as in collection and treatment method.)
vi i i
-------
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the advice and assistance of the persons
identified below.
Susan A. Thorneloe, EPA Work Assignment Manager, Air and Energy Engineering
Research Laboratory, U.S. Environmental Protection Agency, Research Triangle Park,
North Carolina. Susan provided oversight and direction in the development of the
methodology and coordinated the review of the report. Susan also provided the
information on waste generation trends, soon-to-be published Clean Air Act
regulations and the potential benefits of landfill gas utilization.
Ross Leadbetter, Professor of Statistics, University of North Carolina at Chapel Hill,
Chapel Hill, North Carolina. Provided oversight in the statistical analyses that were
conducted for this report.
Allen J. Geswein, George A. Garland, and Andy L. Teplitzky, Office of Solid Waste,
U.S. Environmental Protection Agency, Washington, DC. Helped to provide overview
of the data being used in the report on waste generation quantities.
Kathleen B. Hogan and Cindy B. Jacobs, Office of Atmospheric Programs, Office of
Air and Radiation, U.S. Environmental Protection Agency, Washington, DC. Helped
to review the report and provided comments on the development of the methodology.
Mark Najarian, Office of Air Quality Planning and Standards, Office of Air and
Radiation, U.S. Environmental Protection Agency, Research Triangle Park, North
Carolina. Served as one of the peer reviewers for the report and provided information
on the status of the Clean Air Act regulations for municipal solid waste landfills
Robert D. Gibbons, Professor of Biostatistics, University of Illinois at Chicago,
Chicago, Illinois. Provided peer review of the statistical analyses that were conducted
for the report.
Rebecca L.Peer and David L. Epperson of Radian Corporation, Research Triangle
Park, North Carolina. Assisted in developing the initial methodology and emission
estimates.
Randy Strait, Bill Barnard, Dorothy Titus, and Kathy Manwaring, E.H. Pechan &
Associates, Durham, North Carolina. Provided editorial comments and general
assistance.
ix
-------
INTRODUCTION
BACKGROUND
Methane (CH4) produced via the anaerobic decomposition of waste buried in landfills
and open dumps is a significant contributor to global methane emissions, with estimates
ranging from 10 to 70 teragrams Itg or 1012 grams (g)J per year. Global anthropogenic
sources emit 360 tg/yr (IPCC, 1992) which suggests that landfills account for 3 to 19
percent of the total. Existing emission estimation methodologies for this source tend to
assume that optimal conditions for anaerobic decomposition exist within a landfill.
However, this is rarely the case as the information in the following section and an article
by Rathje (1991) indicate.
To address this concern, AEERL began a research program in 1990, aimed at using
field data from LFG recovery sites to develop an empirical model relating LFG flows to
waste in place. The research program started with a review of available models and data
and identified several theoretical models and laboratory experiments used to estimate
CH4 from individual landfills. However, these methodologies usually rely on site-specific
data which are difficult to extrapolate. Some emission estimation methodologies were
found to be reasonable, but the estimates were based on assumed values for certain
parameters, such as refuse generation rates and waste composition. In order to develop a
new and better model, AEEEL initiated a second phase in the program to identify key
variables that affect CH4 generation from buried waste and at developing an empirical
model based on those variables. (Thorneloe, 1994a.)
Landfills with gas recovery systems, where LFG is collected and measured by
personnel on site, offer a unique opportunity for studying CH4 emissions. LFG recovery
rates can be used to estimate CH4 generation which in turn can be related to CH4
emissions. However, in order to use this approach, the accuracy of such LFG data needed
to be verified. Furthermore, the availability of additional information on the landfills
from which LFG data were collected, including the amount and nature of the waste
present, needed to be examined.
The first step in developing this second phase program was a field study of six U.S.
landfills with LFG recovery systems (Campbell et al., 1991). This pilot study was aimed
at verifying the existence and accuracy of the waste in place and gas flow rate data. The
results of this pilot study were sufficiently encouraging such that a large-scale field study
was conducted at 30 U.S. landfills. The study and its findings are described in a report
entitled: "Development of an Empirical Model of Methane Emissions from Landfills,"
(Peer et al., 1992). The objective of this study was to develop statistical models of annual
landfill CH4 emissions as a function of climate, mass of buried refuse, age, waste
acceptance rate, composition, and compaction, and to develop an emission factor which
could be used to estimate both U.S. and global CH4 emissions from landfills. Sites were
chosen to represent a wide range of climatic conditions, as they occur in the U.S. The
research concluded that the mass of waste in place showed a significant correlation with
CH4 flow rates. The effect of refuse age on gas production was also analyzed. Gas flow
rates correlated most strongly with refuse age for 10 to 20 year old refuse. Although
1
-------
these results were not conclusive, they suggest that the generation time for gas
production is at least 20 years. This result is analogous to the generation time of 20 to 30
years with an average of 25 years suggested by Augenstein and Pacey (1990).
None of the climate variables—precipitation, average temperature and dewpoint—
proved to have significant correlations with the CH4 flow rate. Appendix D, Table 13
summarizes data from Peer et al. (1992). For the purpose of this report the regression
analysis was reproduced for annual rainfall (x values) and average CH4 recovery rate per
unit mass (y values). A linear regression was carried out (i.e., a line was fit through the
data using the least squares method), and an analysis of variance was performed. The
value for R2 according to this analysis was 0.0047, indicating 0.47% of the variability in
the observed values is explained by rainfall. Therefore, no significant correlation exists
between these two variables. In addition, an exponential curve [log (average CH4 recovery
rate per unit mass) versus annual rainfall] was calculated that fit the data. Results from
this analysis again revealed no significant correlation between the two variables, with an
R2 value of 0.066.
To relate CH4 flow rates from recovery projects to CH4 generation rates two
assumptions needed to be made. It was assumed that the average recovery efficiency of a
gas collection system is 75 percent (adapted from Augenstein and Pacey, 1990).
Furthermore, it was assumed that 10 percent of unrecovered methane is oxidized (adapted
from Whalen et al., 1990). Both assumptions are subject to discussion, as very limited
data exist.
Because a large amount of the variability remained unexplained in the field study
described above, a decision was made to try and refine the correlation between LFG flow
and waste mass. A larger LFG recovery data base was produced, which included data
from most U.S. LFG recovery projects. The data base described in the 1991-1992 Methane
Recovery From Landfill Yearbook (Governmental Advisory Associates, 1991) was used as a
starting point. This data base contains information on 170 U.S. LFG recovery projects.
The data quality is uncertain because the data are from survey results which had not
been verified. In addition, the gas flow rates in this data base are often modeled instead
of measured values. To develop more accurate data, the data base developed by
Governmental Advisory Associates was reviewed by AEERL and dubious information was
eliminated or corrected. This effort was conducted with the Solid Waste Association of
North America. In this report, the new data base, which now contains data on 105 U.S.
LFG recovery plants, is referred to as the "ORD Database." Pertinent data from the
ORD Database are included in Appendix D. The geographical distribution of LFG
recovery plants is depicted in Figure 1.
With the expanded and verified data set in the ORD Database, it became possible to
generate a regression function instead of an emission factor. This report describes the
development and application of a regression model with three linear segments. Appendix
A presents the statistical background of the model. The description of a second, more
simple model is also included in Appendix A. This model consists of a single linear
regression line and can be applied when no size-specific waste in place data are available,
as is the case for most countries other than the U.S.
2
-------
Figure 1. Landfill gas recovery sites in the United States.
-------
METHANE PRODUCTION FROM THE ANAEROBIC DECOMPOSITION OF
SOLID WASTE
The anaerobic decomposition of organic matter, as it occurs in a landfill is a complex
process which requires that several groups of microorganisms act in a synergistic manner
under favorable environmental conditions. Anaerobic refuse decomposition has been
reviewed in detail by Barlaz et al. (1990) and more detail on the microbiology of municipal
waste decomposition has been reported by Barlaz et al. (1989a).
Three trophic groups of anaerobic bacteria must be present to produce CH4 from
biological polymers such as cellulose, hemicellulose, and protein: (1) hydrolytic and
fermentative microorganisms, (2) obligate proton-reducing acetogens, and (3) methanogens
(Wolfe, 1979; Zehnder et al., 1982). The hydrolytic and fermentative group is responsible
for the hydrolysis of biological polymers. The initial products of polymer hydrolysis are
soluble sugars, amino acids, long-chain carboxylic acids, and glycerol. Following polymer
hydrolysis, the hydrolytic and fermentative microorganisms ferment the initial products of
decomposition into short-chain carboxylic acids, alcohols, carbon dioxide (C02), and
hydrogen. Acetate, a direct precursor of CH4, is also formed.
The second group of bacteria, 'obligate proton-reducing acetogens,' convert the
fermentation products of the hydrolytic and fermentative microorganisms to C02,
hydrogen, and acetic acid. The conversion of fermentation intermediates, such as
butyrate, propionate, and ethanol is thermodynamieally favorable only at very low
hydrogen concentrations. Thus, these substrates are utilized only when the obligate
proton-reducing acetogenic bacteria can function in syntrophic association with hydrogen
scavengers, such as CH4-producing or sulfate-reducing organisms. The third group of
bacteria necessary for the production of CH4 are the methanogens. Major substrates
utilized by methanogens for the production of CH4 are acetate, formate, methanol,
methylamines, and hydrogen plus C02 (Wolin and Miller, 1985).
While CH4 and C02 are the terminal products of anaerobic decomposition, C02 and
water are the terminal products of aerobic decomposition. Aerobic decomposition occurs
in management facilities where waste is exposed to air, such as when compost is turned
for aerating, and in uncontrolled dumps, such as when refuse is spread in thin layers or
otherwise exposed to oxygen (e.g., by scavenging). When refuse is buried in large piles,
whether at an open dump or in a sanitary landfill, the oxygen entrained at burial is
consumed rapidly, and substantial quantities of CH4 may be produced (Bhide et al., 1990).
Landfilled waste contains numerous constituents that have the potential to
biodegrade under anaerobic conditions. The traditional method of classifying municipal
solid waste (MSW) according to sortable categories [e.g., paper, plastic, food waste, yard
waste, glass, metals, rubber, wood, textiles, dirt, and miscellaneous (U.S. EPA, 1990)J is
appropriate for recycling studies and overall solid waste management planning. However,
data specific to the chemical composition of refuse are more applicable to analysis of
refuse decomposition. Refuse representative of typical MSW from Madison, WI, in 1987
was reported to contain 51.2 percent cellulose, 11.9 percent hemicellulose, no more than
4.2 percent protein, and 15.2 percent lignin (Barlaz, 1988). Measurements of the cellulose
concentration of Madison refuse taken from the period of 1984 through 1986 showed
4
-------
values of 40 to 48 percent (Barlaz, 1985). Cellulose plus hemicellulose accounted for
91 percent of the CH4 potential of refuse (Barlaz et al., 1989b).
The components of MSW that contain significant biodegradable fractions are food
waste, yard waste, and paper. Paper has a combined cellulose and hemicellulose content
of 50 to 100 percent. Lignin is the other major organic component of refuse; however,
lignin does not decompose significantly under anaerobic conditions (Young and Frazer,
1.987).
Methane formation does not occur immediately after refuse is placed in a landfill or
dump. It can take months or years for the proper environmental conditions and the
required microbiological populations to become established. Numerous factors control
decomposition, including moisture content, nutrient concentrations, presence and
distribution of microorganisms, particle size, water flux, pH level, and temperature.
Reviews of the effect of each of these factors on CH4 production are provided in Barlaz et
al., (1990); Pohland and Harper (1986) and Halvadakis (1983).
The two factors that appear to have the most impact on CH4 production are moisture
content and pH. The effect of refuse moisture content has been summarized by
Halvadakis (1983), although some of the data in the summary relate to manure and not
municipal waste. The broadest data sets are those constructed by Emberton (1986) and
Jenkins and Petus (1985). Emberton measured CH4 production rates in excavated landfill
samples under laboratory conditions. Jenkins and Petus sampled refuse from landfills
and tested how CH4 production was affected by the moisture content of refuse. In both
studies, the CH4 production rate exhibited an upward trend with increasing moisture
content, despite differences in refuse density, age, and composition. It is difficult to
translate the results of these laboratory studies to actual landfills. An attempt by AEERL
to identify a statistically significant correlation between landfill gas recovery and annual
precipitation found no such correlation, (Peer et al., 1992).
A second key factor influencing the rate and onset of CH4 production is pH. The
optimum pH level for activity by methanogenic bacteria is between 6.8 and 7.4. Methane
production rates decrease sharply with pH values below about 6.5 (Zehnder, 1982). When
refuse is buried in landfills, there is often a rapid accumulation of carboxylic acids; this
results in a temporary pH decrease and a long time-lapse between refuse burial and onset
of CH4 production which can range from months to years.
Neutralizing leachate and recycling it back through refuse has been shown to enhance
the onset and rate of CH4 production in laboratory studies (Pohland, 1975; Buivid, 1981;
Barlaz et al., 1987, 1989a). Given that moisture and pH have been reported as the two
most significant factors limiting CH4 production, the stimulatory effect of leachate
neutralization and recycling is logical. Neutralization of leachate provides a means of
externally raising the pH of the refuse ecosystem. Recycling neutralized leachate back
through a landfill increases and stabilizes refuse moisture content and substrate
availability. It also enhances mixing in what would otherwise be an immobilized batch
reactor. Field experience with leachate recycling systems is limited and more information
is needed to fully document its value. It is expected that new information will become
available in the next few years.
5
-------
The lapsed time preceding the onset of CH4 production in landfills is important when
considering the management of individual landfills for biogas recovery or emissions
mitigation. The age at which landfills and uncontrolled dumps begin to produce CH4 is of
lesser importance when evaluating global CH4 emissions from MSW management
systems. For estimating global emissions, it is the total CH4 production potential that is
more critical.
METHANE POTENTIAL OF MUNICIPAL SOLID WASTE IN LANDFILLS
Methane potential of landfilled refuse can be determined in three basic ways. The
theoretical CH4 potential of the main chemical constituents may be calculated or
laboratory tests may be conducted, imitating reality in various degrees. Also field tests
have been performed. All methods have in common the question whether the data are
representative or not because, even in field tests, waste composition and other parameters
that affect CH4 generation, may show unpredictable variety from one location to the next.
Knowledge of the chemical composition of refuse buried in a landfill makes it possible
to estimate the maximum volume of CH4 that may be produced. The mass of CH4 that
would be produced if all of a given constituent were converted to CH4, C02, and ammonia
may be calculated from Equation 1 (Parkin and Owen, 1986).
CnHa°bNc + [n-(al4)-(bj2) + 3(c/4)\H2o -
[(n/2) -(a/8) +(i>/4) +3(c/8)]C£?2+[(n/2) +(«/8)-(*/4)-3(c/8)]C/f4 +cNH3
Using this stoichiometry, potential CH4 production volume from cellulose (C6H10O5)
and hemicellulose (C5H804) is 415 and 424 liters per dry kilogram(l/kg) at standard
temperature and pressure, respectively. These methane potentials represent maximum
CH4 production if 100 percent of the cellulose and hemicellulose were converted to CH4.
However, decomposition of these constituents in landfills is well below 100 percent for
several reasons, but mainly because (1) some cellulose and hemicellulose is surrounded by
lignin or other recalcitrant materials (such as plastic) and, therefore, is not biologically
available; and (2) without active intervention, buried refuse is not evenly exposed to
moisture, microorganisms, and nutrients. Barlaz et al. (1989b) applied mass balances to
shredded refuse incubated in laboratory-scale lysimeters with leaehate recycle. Carbon
recoveries of 87 to 111 percent were obtained, where a perfect mass balance would give a
carbon recovery of 100 percent. Greater than 100 percent recoveries were obtained in
some cases due to sampling and analytical error. Mineralization of 71 percent of the
cellulose and 77 percent of the hemicellulose was measured in a container sampled after
111 days. Mass balances were useful for documenting the decomposition of specific
chemical constituents and demonstrating the relationship between cellulose and
hemicellulose decomposition and CH4 production.
Stoichiometry may also be used to estimate the CH4 potential remaining in a landfill
by sampling the refuse, performing the appropriate chemical analyses, and calculating the
CH4 potential. Ideally, the initial chemical composition and CH4 potential of the refuse
6
-------
would be known, in which case comparing that initial CH4 potential with the potential at
the time of sampling would provide information on the fraction of the refuse that has been
degraded. Indisputably, representative sampling of a full-scale sanitary landfill is not
realistic. Sampling size is limited to volumes that can be reasonably handled and reduced
by proven techniques. However, it is possible to obtain multiple samples at presumably
representative locations within a landfill to get an estimate of the range and extent of
decomposition.
Another technique for assessing the CH4 potential of refuse is the biochemical
methane potential (BMP) test (Shelton and Tiedje, 1984; Bogner, 1990). In the BMP test,
the anaerobic biodegradability of a small sample of refuse (5 to 10 g) is measured in a
small batch reactor [100 to 200 milliliter (ml)]. While the BMP represents an upper
bound of CH4 potential from refuse, it will be lower than the stoichiometric estimate
described above. BMP's also require representative sampling in landfills. A recent
application of the BMP test was presented by Wang et al. (1994).
Comparison of CH4 production data between field-scale landfills and laboratory
experiments is difficult because there are essentially no data in the open literature on
CH4 production rates in field-scale facilities. Interpretation of data from field-scale
landfills is complicated by questions regarding the mass of refuse responsible for
production of a measured volume of gas and the efficiency of gas collection. While
laboratory data are of higher quality due to the more closely controlled conditions, they
are not completely representative of the field. Also, data are not perfectly comparable in
that experimental conditions (e.g., moisture, particle size, temperature, etc.) are not
uniform between studies. In addition, most laboratory experiments were conducted to
explore techniques for enhancing CH4 production. The enhanced CH4 production rates
would not be expected at field-scale landfills unless certain enhancement techniques are
employed.
Methane yields of 42 to 120 1/kg dry refuse have been reported in laboratory tests
conducted with leachate recycling and neutralization (Barlaz et al., 1987; Barlaz, 1988;
Kinman, 1987 and Buivid, 1981). These studies show significant variation in CH4
production rate and CH4 yield. Some of the differences can be explained by differences in
experimental design. For example, the data reported by Barlaz et al. (1987) and Barlaz
(1988) differ in reactor volume (100 vs. 2 1), temperature (25DC vs. 41°C), and the rate of
leachate recycling. Also, Buivid (1981) used refuse with an abnormally high paper
content.
Methane yields were measured in field-scale test cells as part of the Controlled
Landfill Project in Mountain View, California (Pacey, 1989). Yields of 38.6 to 92.2 1
CH4 /dry kg of refuse were measured after 1,597 days. However, mass balance data
suggested that significant volumes of CH4 were not measured in certain test cells. A
number often used by the LFG industry as an estimate of CH4 production in field-scale
landfills is 0.1 cubic feet CH4 per wet pound per year (ft3/wet Ib-yr). Assuming refuse
buried at 20 percent moisture and a 15 year period for gas production, this converts to a
yield of 7.8 1 CH4/dry kg, a number comparable to some of the lower values reported in the
literature.
7
-------
Even in landfills with venting systems, some of the CH4 is likely to escape from the
landfill through the final cover. The fraction released through the final cover will be a
function of the type of gas venting system in place and the type of cover. Probably not all
the CH4 that escapes from landfills is released to the atmosphere. Some may be
converted to C02 as it passes through the cover soil by aerobic methanotrophic bacteria,
CH4 oxidation has been documented in landfill cover soil studied under laboratory
conditions (Whalen et al., 1990). However, there are no data on the quantitative
significance of CH4 oxidation above field-scale landfills. Methane escaping through cracks
in a landfill cover most likely will not reside in the cover for a period sufficient to undergo
significant oxidation.
8
-------
MODEL TO ESTIMATE MUNICIPAL LANDFILL METHANE EMISSIONS
REGRESSION MODEL
The ORD Database includes data on LFG recovery flow rates and welled waste from
105 LFG recovery sites. Welled waste is defined as the quantity of waste from which LFG
is extracted through the recovery wells. As mentioned in the Introduction, data in the
ORD Database are based on measurements and estimates from on-site personnel at
landfills with LFG recovery systems and were quality assured in telephone-interviews
conducted by AEERL staff.
To develop a model relating flow rates to welled waste, the ORD Database was
subjected to regression analysis. The objective was to let statistical criteria dictate the
shape and position of the regression curve. The only constraint was that the line had to
start in the origin. This resulted in a regression model with three different linear
segments, where each segment applies to a distinct landfill size class. Figure 2 shows the
regression curve, as well as the gas flow rate plotted against welled waste for the
105 sites from the ORD Database.1 The analysis is described in detail in Appendix A.
The size classes and equations for the three segments of the curve are:
I x < 1.128 y = 19.822 x
II 1.128 < x < 4.082 y = 1.652 * + 20.495
IH * > 4.082 y = 9.195 x - 10.294
where: x = welled waste, (tg).
y = LFG flow rate, (m3/minute).
Common sense would suggest that the curve is more likely to be smooth. However, due to
limitations inherent to the regression analysis the curve is segmented. The position,
inclination, and length of the segments are dictated by the data. Therefore, the fact that the
second segment is flatter than the first and third segment may not be interpreted as a CH4
rate drop for a certain size landfill. This second segment merely connects the two other
segments in such a way that the combination of all three segments best represents the data.
9
-------
Figure 2. Landfill gas flow rates versus welled waste for 105 recovery sites and best fitting
regression curve.
-------
CONVERSION FACTOR
In order to convert LFG flow rate (y) in m3/min to the actual mass flow of CH4
released to the environment (Y) in g/min, the following three steps must be taken:
1. Convert the initial LFG flow, expressed in units of volume per minute, into mass CH4
flow (Yw). Use the relative concentration of CH4 in LFG, c = 0.50 IPeer et al, (1992);
Roqueta, (1992) and Anderson, (1992)1, and the density p = 677 g/m3. The density
calculation is presented in Appendix C.
rw = y*c*P ] (2)
nun m m
2. Adjust for the efficiency (r) of the gas recovery system. This step actually converts
the CH4 flow rate to CH4 generation rate (Yg ). A recovery efficiency of 75 percent is
assumed (Augenstein and Pacey, 1990). Subsequently, an adjustment needs to be
made for the fraction of CH4 that does not reach the atmosphere because the CH4 is
oxidized on its way out of the landfill. This adjustment converts CH4 generation into
CH4 emission. The oxidation factor, (o) is estimated to be 0.10 (Mancinelli and
McKay, 1985).
(1-0) r 8
Y=y*c*p* [-*-] (3)
s r mm
3. By introducing a factor 525,600 to convert from minutes to years, the actual mass of
CH4 released annually (M) becomes:
Y = y*c*p* Q—^ * 525,600 [-£- * — ] (4)
r min yr
By employing the presented values for c, o, p, and r, the conversion factor is:
CF = 213 * 106 [^55] (5)
yrnr
The accuracy of this number is addressed on page 18 in the section entitled:
"Development of Ranges."
11
-------
ESTIMATE OF TOTAL WASTE IN PLACE
MUNICIPAL LANDFILLS
In 1988, the EPA's Office of Solid Waste (OSW) published the results of a survey in
which detailed information on 1,176 U.S landfill facilities was collected in a data base.
The results of this survey, the Subtitle D Municipal Landfill Survey, are described in
"National Survey of Solid Waste (Municipal) Landfill Facilities," (U.S. EPA, 1988)
henceforth referred to as "the Survey." The Survey includes site-specific waste in place
data, which can be used in the regression model described in Section 2.
The Survey was conducted in response to the Hazardous and Solid Waste
Amendments of 1984. Under these amendments, EPA was required to determine whether
the Resource Conservation and Recovery Act (RCRA) regulations were sufficient to protect
human health and the environment from ground-water contamination. In this report, this
data base will be referred to as the "OSW-Westat Database." For the Survey, a landfill
qualified as a municipal landfill if it received primarily household refuse and commercial
waste, and was not a hazardous waste facility (U.S. EPA, 1988). The target population
included all municipal facilities in the U.S. and its five territories that had at least one
active landfill unit as of November 1, 1986. At the time of the Survey, there were
approximately 6,500 landfills in the U.S. For the sub-population of 1,176 landfills
information was collected on ownership, location, operations, hydrogeology, waste
characteristics, landfill unit construction, monitoring systems, and operating cost. Not all
sites have complete data. For the purpose of this report, 142 sites had insufficient
information for estimating waste in place and were eliminated. More information on why
the 142 facilities were eliminated from this analysis is given in Section 5.
The following landfill data, which can be used to calculate waste in place, are
included in the data base sub-population for 1,034 landfills.
• The year that waste was first placed in the landfill.
• The average annual quantity of waste received in tons.
• The total design capacity of the landfill in tons.
• The total remaining design capacity in tons.
Responses to the first and second items can be used to estimate waste in place according
to what is referred to as the "RATE" method. The total waste in place was estimated by
multiplying the length of time that the landfill has been accepting waste in years and the
average annual quantity of waste received.
Waste In Place, = (1987 - Yearwastefirstplacedin landfill) * ^
(Average Annual Quantity of Waste Received)
12
-------
It is also possible to calculate waste in place for each landfill by taking the difference
of the results from the third and fourth items. This method is referred to as the DIFF
method.
Waste In Place-DIFF. = Total DesignCapacity - Remaining Design Capacity (7)
In this report, preference is given to the DIFF method. A detailed discussion and
comparison of the DIFF and RATE methods is included in Section 5. Appendix A includes
estimates based on both methods.
Through a stratified sample design, large facilities in the Survey had a higher
probability of being sampled than small facilities. The distribution of waste in place
obtained from the subset of landfills included in the municipal solid waste landfill survey
was applied to the whole population of active municipal landfills. This is done through
the application of scaling factors, also referred to as weights or raising factors. The
eligible sample of landfills was broken down into two strata. Stratum 1 was comprised of
large or active facilities that received at least 500 tons of waste per day. Stratum 2
consisted of small or less active facilities, those that received less than 500 tons of waste
per day. Fifty-two percent of the eligible active facilities were surveyed, while 13 percent
of the eligible, less active sites were investigated. (This is the optimal stratified sampling
plan for a sample size of 1,000, in that the variance of the estimate of the total quantity of
waste received is minimized.) This allocation resulted in a large, active site being roughly
four times more likely to be sampled than a small facility.
To extrapolate the sub-population to the total population of U.S. landfills, as well as
to account for the difference in likelihood to be sampled, the scaling factors for active and
less active landfills are 2.00 and 7.00, respectively. These particular scaling factors were
presented with the data base. By multiplying the scaling factor by waste in place for each
site from the sub-population and adding up the results, total waste in place for the nation
can be calculated. Section 5 addresses the quality of the scaling factors. Appendix A
gives the statistical background for the determination of scaling factors.
According to the DIFF method, the total waste in place in the U.S. in 1986, the year
the Survey was conducted, was 4,720 tg. The average landfill age was 19 years. The
total average annual quantity of waste received was 248 tg. The rate at which MSW is
landfilled has been declining by approximately 2% per year between 1987 and 1993 (U.S.
EPA, 1992). Because this decrease is small and information on changes in disposal rates
of other types of waste such as demolition and construction debris have not been found,
the annual disposal rate of 248 tg/yr was assumed to be steady between 1987 and 1993.
Hence, for 1992 the total waste in place for the U.S. was 6,200 tg. Estimates for the
RATE method are given in Appendix A.
13
-------
Magnitude of Waste Quantity
In the U.S., 248 tg/yr of waste is estimated to be placed in municipal landfills. This
amount is equivalent to a daily, per capita placement rate of 3.0 kg. (6.6 lb/cap/day) [U.S.
population of 226,505,500 (1990)]. It should be emphasized that this number is not
equivalent to the MSW generation rates computed by OSW and published in their bi-
annual "Characterization of Municipal Solid Waste" reports2 [U.S. EPA, (1990) and U.S.
EPA, (1992)]. Waste generation computed by the EPA/OSW study amounts to
approximately 2.0 kg/cap/day (1992) and does not include such wastes as demolition and
construction debris, nonhazardous industrial waste, sludges, fly ash, etc., all of which may
(in part) be landfilled. Instead, the 3.0 kg rate generated here represents the waste
placement rate which is actually landfilled in municipal landfills in the U.S. based on the
Survey and OSW data base. As Table 1 indicates, this placement rate is comparable to
those of other industrialized countries.
TABLE 1. COMPARISON OF WASTE GENERATION AND WASTE-TO-LANDFILL RATES FOR
EIGHT INDUSTRIALIZED COUNTRIES
Waste (kg per
capita per day)
MSW
Generated
Total Waste
to Landfill
Average %
to Landfill
Landfillable
Waste1
Source2
United States
2.0
3.0
70
4.3
This report
Japan
1.0
1.6
40
3.8
Cossu
United Kingdom
2.0
4.23
97
4.3
Cossu
Netherlands
1.6
2.5
51
5.14
Beker
Sweden
0.8
2.1
80
2.6
Nillson
Denmark
1.2
1.8
60
3.1
Christensen
Finland
1.4
24.16
varies
n/a
Ettala
Italy
0.8
3.4
100
3.4
Cossu &
Urbini
' Assumed that the average % to landfill (previous column) is 100 %.
2 All references from "International Perspectives on Municipal Solid Wastes and Sanitary Landfilling," (Carra and Cossu,
1990).
3 Excluding mining, agricultural, and power station wastes which amount to 16 kg/cap/day)
4 Mainly demolition, construction, and nonhazardous industrial waste.
5 Wastes from mining and wood harvesting and processing are also landfilled.
These waste generation estimates are based on a materials flow methodology using
production output figures from, for instance, the Department of Commerce.
14
-------
INDUSTRIAL LANDFILLS
The U.S. maintains separate records of industrial waste that is landfilled at specially
designated industrial landfills, instead of at municipal landfills. Estimates by Schroeder
et al., (1987) indicate that 15 tg of biodegradable industrial waste is landfilled annually at
industrial landfills. In this report it is assumed that the size as well as the age
distribution of industrial landfills is similar to that of municipal landfills. Because the
average age of a landfill is 19 years, the total amount of landfilled industrial waste is
285 tg. For 1992, this number would be 375 tg, which is 15 tg multiplied by 25 years.
15
-------
-------
ESTIMATE OF U.S. LANDFILL METHANE EMISSIONS
The ORD Database, which presents data from 105 LFG recovery sites, was used to
generate a model which describes CH4 gas flow rate as a function of refuse mass in place.
In order to use flow rates as a surrogate for CH4 emissions, certain assumptions needed to
be made. These assumptions are expressed in a conversion factor. This factor accounts
for the recovery efficiency of LFG projects and for the probable loss of CH4 due to
oxidation. Furthermore, the factor incorporates a conversion from LFG flow rate (m3/min)
to actual mass CH4 emissions. Application of the model to the waste mass data from the
OSW-Westat Database and multiplication of the results with the conversion factor yields
national CH4 emissions from landfills.
Since the developed regression model is tripartite, CH4 emissions for each landfill
depend on size classification. Therefore, it is not possible to calculate the total nation-
wide waste in place first and then apply the model. Instead, the proper approach is to
divide the landfill population into the three size classes. For each site the LFG flow rate
is calculated by applying the appropriate regression equation to the waste in place. This
rate is in turn multiplied by the appropriate scaling factor for each site. The estimated
values of LFG flow rates are then added together, yielding an estimate of total LFG flow
rate for all active municipal landfills. Ultimately, the total LFG flow rate is multiplied by
the conversion factor to convert LFG flow rate to mass of emitted CH4 in grams per year.
This estimate is then adjusted for the amount of CH4 being recovered or flared. In 1992,
1.2 tg of landfill methane was utilized (Thorneloe, 1992a). There are also landfills that
flare the gas instead of recovering the energy. As no quantitative data are available, the
amount of LFG that is flared is assumed to be 0.5 tg/yr (i.e. 40 percent of LFG recovered
for energy purposes). Therefore, the total amount of recovered CH4 for 1992 is estimated
to be 1.7 tg/yr.
It is estimated that an additional 15 tg of industrial waste is landfilled annually in
the U.S. at industrial landfill sites. Because limited information on industrial landfills is
available, it is assumed that industrial landfill characteristics, including size and age
distribution, are similar to those of municipal landfills. The average age of an industrial
landfill is then also assumed to be 19 years, equal to the age of a municipal landfill. Then
the total amount of landfilled industrial waste is 285 tg. The 285 tg was apportioned to
the three size classes used to classify municipal landfills. Tables 2 and 3 illustrate the
calculations for CH4 emissions from waste in industrial landfills.
Tables 4, 5 and 6 summarize the waste in place data as well as the nationwide
emissions for 1986 and 1992. In Tables 5 and 6 emissions are based on waste calculations
by the DIFF method. Emissions estimates making use of waste in place calculated by the
RATE method may be found in Appendix A. For 1986 CH4 emissions from U.S. landfills
were estimated at 11 tg/yr with lower- and upper bound values of 7 and 15 tg/yr
respectively. Methane emissions from U.S. landfills in 1992 were estimated at 13 tg/yr
with lower and upper bound values of 9 to 19 tg/yr. Appendix B compares these estimates
with other previously published estimates. The development of the lower and upper
bound values is discussed on page 18 in the section entitled: "Development of Ranges."
16
-------
-------
TABLE 2. METHANE EMISSIONS FROM INDUSTRIAL LANDFILLS (Metric Units)
Size Class
I
II
III
Total
Industrial Waste in Place, (tg)
1986
17
34
234
285
1992
23
45
307
375
LFG Flow Rate, (m3/min)
1986
339
77
2,138
--
1992
455
95
2,813
-
Conversion Factor, (106 g.min/yr.m3)
213
Emissions Estimate, (tg/yr)
1986
0.07
0.02
0.46
0.55
1992
0.10
0.02
0.60
0.72
Note: Mass industrial waste was calculated by the DIFF method, apportioning the total industrial waste in
place to the size classes, proportionate to the fraction of municipal waste in each size class.
TABLE 3. METHANE EMISSIONS FROM INDUSTRIAL LANDFILLS (U.S. Units)
Size Class
I
II
III
Total
Industrial Waste in Place,
(109 tons)
1986
0.019
0.037
0.258
0.314
1992
0.025
0.05
0.338
0.413
LFG flow rate, (CFM)
1986
11,934
2,727
7,5519
-
1992
16,065
3,348
99,333
--
Conversion Factor, (tons.min/yr. cu. ft)
4,860
Emissions Estimate,
(10e tons/yr)
1986
0.08
0.02
0.50
0.60
1992
0.11
0.02
0.66
0.79
Note: Mass Industrial waste was calculated by the DIFF method, apportioning the total Industrial waste in
place to the size classes, proportionate to the fraction of municipal waste in each size class.
17
-------
TABLE 4. TOTAL WASTE IN PLACE AND TOTAL METHANE EMISSIONS FROM MUNICIPAL
LANDFILLS PER SIZE CLASS (Metric Units) (Not corrected for industrial landfills, flaring or utilization)
Size Class
1
II
III
Total
Size Class Boundaries, (tg)
x< 1.13
1.13 s x < 4.08
* i 4.08
Number of Sites
1986
816
144
74
1,034
1992
781
154
99
1,034
Percent of Sites, (%)
1986
79
14
7
100
1992
76
15
10
100
Municipal Waste in Place,
(tg)
1986
898
1,169
2,639
4,720
1992
942
1,452
3,797
6,200
Percent of Waste in Place,
(%)
1986
19
25
56
100
1992
15
23
61
100
LFG Flow Rate,
(104 rtrVmin)
1986
1.8
1.3
2.2
5.3
1992
1.9
1.6
3.2
6.7
Conversion Factor, (10'' g.min/yr.m3)
213
Emissions Estimate, (tg/yr)
1986
3.8
2.8
4.7
11.3
1992
4.1
3.4
6.8
14.2
DEVELOPMENT OF RANGES
Due to the impossibility of estimating errors associated with the assumptions used to
estimate CH4 emissions from landfills a mathematical approach in which individual errors
are propagated is meaningless. Therefore, the following method was adopted. The
standard deviation in the emissions estimate is 12 percent (Appendix A). Approximate
95 percent confidence intervals are obtained by adding plus/minus two standard
deviations to the estimate. Consequently, it is assumed that errors in other parameters
amount to at least another 12 percent. Therefore, ranges are expressed by adding
plus/minus 36 percent to the emissions estimate.
18
-------
TABLE 5. TOTAL METHANE EMISSIONS FROM U.S. LANDFILLS (Metric Units)
1986
1992
Total Municipal Waste in Place, (tg)
4,720
6,200
LFG flow rate, (104 m3/min)
5.3
6.6
Estimated Emissions from Municipal Landfills, (tg/yr)
11.2
14.2
Emissions from Industrial Landfills, (tg/yr)
0.55
0.72
Methane Currently Recovered or Flared, (tg/yr)1
1.22
1.7
(1.1)
(1.5)
Lower bound
7
9
Estimated Total U.S. Emissions,
11
13
(tg/yr) Mid-point
Upper bound
15
18
1 Amount of recovered CH« needs to be adjusted for the fact that it would have undergone oxidation if it had not
been recovered. So it needs to be multiplied by 100 - 10 = 90 %. Adjusted amount is in parentheses.
2 Amount of CH„ recovered in 1986 assumed to be 70% of 1992 value.
TABLE 6. TOTAL METHANE EMISSIONS FROM U.S. LANDFILLS (U.S. Units)
1986
1992
Total Municipal Waste in Place, (10® tons)
5,200
6,830
LFG flow rate, (104 CFM)
187
233
Estimated Emissions from Municipal Landfills, (10s
tons/yr)
12.5
15.7
Emissions from Industrial Landfills, (10s tons/yr)
0.60
0.79
Methane Currently Recovered or Flared, (106 tons/yr)1
1.32
(1.2)
1.9
(1.7)
Lower bound
Estimated Total U.S. Emissions, lM., . ,
(106 tons/yr) Mid-point
Upper bound
8
12
16
10
15
20
' Amount of recovered CH4 needs to be adjusted tor the fact that it would have undergone oxidation if it had not
been recovered. So it needs to be multiplied by 100 - 10 = 90 %. Adjusted amount is in parentheses.
2 Amount of CH4 recovered in 1986 assumed to be 70% of 1992 value.
19
-------
UNCERTAINTIES
This section discusses areas of uncertainty in the estimate of CH4 emissions from
landfills. Specific uncertainties regarding the statistics applied in the calculations of
waste in place and CH4 emissions can be found in Appendix A.
QUALITY OF OSW-WESTAT DATABASE
Data from the 1986 Subtitle D Municipal Landfill Survey are included in the OSW-
Westat Database and the accompanying Survey document published in 1988. Although
this document is quite comprehensive and explicit, there are questions regarding whether
it is reasonable to extend the data base to 1992-1993. It appears that changes were made
to the data base which have not been documented sufficiently. For example, no
documentation was found that included the density used to convert from volume to mass
for landfilled waste in the OSW-Westat Database. Also, the Survey document reported
the total annual quantity of waste received as 208.8 million tons, whereas the quantity of
waste received now totals 273 million tons (248 tg). Geswein (1993) stated that this
difference was caused by an error in the 1986 data base, which was corrected after
publication. Geswein (1993) indicated that the 248 tg/yr estimate is the most accurate. A
number of the same magnitude was also used by EPA's Office of Air Quality Planning and
Standards in preparation of the new regulation for new and existing landfills under the
Clean Air Act.
Comparison of RATE and DIFF Methods
This section presents a comparison of the DIFF and RATE methods and summarizes
the advantages and disadvantages of each method. Conclusions drawn from a numerical
comparison of the DIFF and RATE methods seem to confirm the doubt regarding the
quality of the OSW-Westat Database.
Waste in place can be approximated using two different methods. The RATE method
uses information on the length of time that the landfill has been accepting waste (age of
landfill) and the average annual quantity of waste received [Equation (6)1. According to
the DIFF method, waste in place can be calculated based on the difference between total
and remaining design capacity in volume units [Equation (7)1.
Waste In Place.^^={19%! - Year waste first place in landfill) * ^
(Average Annual Quantity of Waste Received)
Waste In Place.D[Fr = Total Design Capacity-Remaining DesignCapacity (7)
20
-------
In order to compare the RATE and the DIFF methods, a sub-population of
1,034 landfills was identified for which data were available to calculate waste in place
using both methods. Table 7 presents the results of the comparison. The data base
contains 1,175 entries. Data needed to calculate waste in place using either method are
not available for 81 landfills. For 30 of the remaining landfills, no data are available to
estimate waste in place using the DIFF method, and an additional 30 landfills do not
have sufficient data to calculate waste in place using the RATE method. For the sub-
population of 1,034 landfills, waste in place calculated by the RATE method is
approximately one and a half times higher than waste in place calculated by the DIFF
method. Table 8 presents summary statistics for each method.
However, a comparison between the two methods of calculating waste in place for
individual landfills show significant inconsistencies, which are illustrated in Figures 3
and 4. In Figure 3, landfill sizes calculated by the DIFF method are plotted against
landfill sizes calculated by the RATE method. In Figure 4, for each site, the logarithm of
the quotient of waste in place estimated by the RATE method and the DIFF method is
plotted against the age of the landfill. In Figure 4, the methods should present identical
or at least comparable waste in place estimates, with the logarithm of the quotient being
zero or close to zero. Nevertheless, for several landfills, the waste in place calculated by
one method is a factor 10 or more than waste in place calculated by the other method, for
only 32 percent of landfills waste in place estimates by both methods are within a
40 percent range of each other. One could possibly expect that the two methods would
show different results for older landfills due to changes in compaction between waste and
varying or misinterpreted acceptance rates. However, the site-specific differences in
waste in place estimates also occur for relatively young landfills. For these landfills, it is
unlikely that the amount of waste is estimated erroneously by either of the methods. The
magnitude of the differences raises concern about the quality of the data.
From the comparison of the results of the two methods, it is not clear which of the
two methods provide a more accurate estimate of waste in place. The DIFF method is
based on the difference of only two parameters (Equation 7). Also it would take into
account the compaction of the waste. The RATE method on the other hand, is based on a
parameter (i.e., the average annual quantity of waste received) which has to be calculated
incorporating historical data over the life of the landfill (Equation 6). Also, it does not
adjust for compaction of waste already in the landfill.
A further consideration is the uncertainties associated with waste in place estimates
using the RATE method. The Survey asked respondents to report the average annual
quantity of waste received over the life of the landfill. No information is available on how
each respondent calculated the response to this question. In 1986, the average age of the
landfills in the data base was 19 years and 170 of the landfills had been operating for
30 years or longer. Particularly the older landfills may not have kept accurate historical
records on the annual amount of waste received. Also, it is unlikely that the average
annual quantity of waste for these older landfills was calculated with a high degree of
precision, by averaging out waste acceptance data from records (if available) spanning
several decades.
21
-------
TABLE 7. COMPARISON OF DIFF AND RATE METHODS FOR WASTE IN PLACE CALCULATIONS
DIFF
RATE
Total number of landfills in the data base
1,175
1,175
Landfills with no data for calculating waste in
place by either method
81
81
Landfills to which the indicated method
cannot be applied
60
60
Remaining sample population
1,034
1,034
Total population (sum of scaling factors)
6,223
6,223
Waste in place for sub-population of 1,034
landfills, (109 tons)
1.58
2.30
TABLE 8. COMPARISON OF DIFF AND RATE METHODS FOR WASTE IN PLACE CALCULATIONS:
STATISTICAL SUMMARY
DIFF
RATE
Mean (million tons)
1.52
2.23
Standard error (million tons)
0.26
0.31
Median (million tons)
0.12
0.14
Mode (million tons)
0.23
0.18
Standard deviation (million tons)
8.55
9.79
Variance
7.3*1013
9.6*1013
Kurtosis
374
632
Skewness
23
17
Range (million tons)
2.44
2.44
Minimum (tons)
46
70.00
Maximum (million tons)
244
244
22
-------
5.0E7
O
4.0E7-
O
3.0E7-
w
c
o
LU
H
<
DC
O
2.0E7-
0
1.0E7-
X)
o
o
o
o
o
Q>
O
o.ot
O o
o
o o
o
o
o
o
oO
o
o
oo
-+-
0.0
1.0E7
2.0E7
3.0E7
DIFF, tons
o
o
o
-o-
4.0E7
Note: values for six very large sites are not shown, but are taken into account in evaluations
5.0E7
Figure 3. Comparison of waste in place data for RATE and DIFF methods for all
landfills from sample population.
-------
Figure 4. Comparison of quotient of waste in place data for RATE and
DIFF methods versus age of landfill from sample population.
24
-------
Nevertheless, it is not unreasonable to suspect that certain respondents based their
estimate of the average annual quantity of waste received on the acceptance rate for the most
recent year or years prior to 1986 when the survey was conducted. If this is true, the
average annual acceptance rates used in the RATE method probably are too high, because
national annual waste generation has increased substantially over the past decades. (This
issue is discussed in Section 6, "Trends in Waste Management and Their Impact on Future
Emissions").
An issue that should be raised regarding the DIFF method is how respondents were
expected to come up with an estimate for the remaining capacity of the landfill. In reality,
landfill owners do not usually know how much waste there is in their landfill, nor how much
the remaining capacity might be. The survey questionnaire tried to circumvent this problem
by providing the definition for remaining design capacity as: "total design capacity minus
amount of waste currently in the landfill." If it is difficult to estimate the amount of waste in
a landfill directly. The only other way of estimating the quantity of waste currently in a
landfill is by recording the flow of waste going into the landfill. If this is true, respondents
would have had to use some version of the RATE method for their estimates of waste in
place. Although this would not explain the difference in the waste in place estimates using
the two methods, it does suggest that the DIFF method data in the data base may be based
on respondent estimates of total waste in place using the RATE method thus, implying that
both methods may be mathematically the same.
Scaling Factors
Table 7 shows that the original data base contained information from 1,175 sites. These
sites were a sub-population of the approximately 6,500 landfills in the U.S. Of the 1,175
surveyed sites, 1,102 were considered eligible for the intended purpose of determining the
risks as specified under RCRA, as described in section 3. The Survey states that the 1,102
sites in the sub-population would correspond to a total of 6,034 landfills. The applied scaling
factors are not provided in the Survey document. (Scaling factors merely describe the
mathematical procedure to go from 1,102 to 6,034; addition of all the scaling factors yields
the total population of landfills.) Instead, scaling factors were listed in the data base, having
values of either 2.00 for landfills receiving at least 500 tons of waste per day and 7.00 for
landfills receiving less than 500 tons of waste per day. (There are a few values of 1 and 3
which were considered erroneous and set equal to 2.00.) For lack of better information, it
was assumed that the scaling factors of 2.00 and 7.00 are based on the size of the original
populations of eligible and total landfills.
For the purpose of determining waste in place, it was not possible to use all 1,102
entries. Only 1,034 sites had sufficient data. Addition of the scaling factors for these
1,034 landfills corresponds to a total number of 6,223 landfills, which is 3 percent more
then the original count of 6,034 landfills. Therefore, the scaling factors of 2.00 and 7.00
may be too high. Consequently, the estimate of total waste in place may also be too high.
Since the original sub-population of 1,102 sites cannot be defined exactly, it is not correct
to assume that total waste in place or CH4 emissions is overestimated by the same
percentage. To eliminate this error, more about the history and validity of the current
scaling factors and sub-populations needs to be known.
25
-------
UNCERTAINTIES ASSOCIATED WITH THE MODEL
Emissions from Small Landfills
There is anecdotal evidence that emissions per ton of waste are higher from small
landfills than from larger landfills. (Small landfills are defined as landfills from Size
Class I with less than 1.128 tg of waste in place). This supposition is strengthened by the
fact that the slope of the first segment, belonging to Size Class I, in Figure 2 is steeper
than the slopes of the other two segments. A steeper slope relates to a higher emission
factor. This is detailed in Appendix A. However, the regression analysis for the first
segment is based on data from only 5 LFG recovery projects and should, therefore be
considered with caution.
The reason why there are only a few LFG recovery projects at smaller landfills is that
the feasibility of a LFG recovery operation will generally increase with the size of the
landfill. To have a feasible LFG recovery project at a small landfill, the LFG yield per ton
of waste has to be higher than at an identical project at a large landfill, i.e. it has to be a
"rich" landfill. In addition, a LFG recovery project at a small landfill would have to be
more efficient than would be necessary at larger sites. Hence, a bias may be introduced
to the data, causing the emission factor for smaller landfills to be relatively high.
Although more research would be necessary to resolve this issue, it is important to
point out that small landfills, though numerous, contain only a small fraction of the total
amount of refuse disposed of in the U.S. For instance, 86 percent of all active MSW
landfills in the U.S. has less than 1 tg of waste in place (1986). Nevertheless this
population only contains 17 percent of total waste in place.
Bias toward "Rich" Landfills
The reasoning from the previous section may be applied in broader terms. To ensure
sufficient return on investment LFG developers will pick those landfills with the highest
LFG potential ("rich" landfills). The methodology employed in this report extrapolates
LFG flow versus welled waste data (i.e., emission factors) to include all waste at U.S.
municipal landfills. Since the emission factors are only based on landfills that do have
LFG projects (the "rich" ones), a bias may be introduced causing the estimates to be too
high. Alternately, most LFG recovery projects are in the Northeast and in California
(Figure 1). Perhaps this is because the high population density in these areas leads to
bigger landfills. Also, the energy economics may be better in these geographical regions.
However, arguments are also available to suggest that the aforementioned bias may
be lower than would be expected. Especially in the past when waste with high CH4
potential may have been combined with waste with low CH4 potential. Also, CH4
potential assessments usually have a large margin of error, and developers may
inadvertently install wells in "poor" sections or layers of the landfill. In addition, LFG
feasibility studies will usually involve economics that is more complicated than straight
forward profitability calculations. If a landfill owner is required to control LFG emissions
it may be beneficial to choose LFG recovery, even if no profits are ever generated.
Returns from gas sales would help offset part of the required investment for the otherwise
26
-------
obligatory flare. In addition, there may be certain incentives or tax breaks that would
make the development of marginal landfills possible.
OTHER UNCERTAINTIES
Extrapolation to 1993
According to the DIFF method, the total waste in place in the U.S. at the beginning of
1987 was 4,720 tg. Total waste in place was updated to the beginning of 1993 by linear
extrapolation. Uncertainties in the waste estimates, as well as in methane emissions
estimates will be magnified by this multiplication. Section 6, "Trends in Waste
Management and Their Impact on Future Emissions," discusses factors which influence
the update of the 1986 data from the Survey.
Industrial Waste
It is estimated that an additional 15 tg of industrial waste is landfilled annually in
the U.S. in industrial landfills (Schroeder et.al., 1987). Compared to the annual waste
disposal rate of 248 tg/yr at municipal landfills this is only a small amount; 6 percent.
Very limited information on industrial waste, other non-MSW, and on industrial landfills
is available. It is assumed that industrial landfill characteristics, including size and age
distribution, are similar to those of municipal landfills. To reduce uncertainty, more
reliable data for industrial landfills would have to be obtained.
Generation Time
The generation time is the "life time" of a batch of waste during which it produces
CH4. Typically, a generation time of 20 to 30 years is assumed to be reasonable for
temperate climates (Augenstein and Pacey, 1990; EMCON Associates, 1982). The
calculations in this report do not explicitly consider the generation time. The reason is
that the regression model is developed from data from LFG recovery projects at landfills
which have waste of different ages. Therefore, the regression model already accounts for
generation time and differences in waste age. However, this holds true only if the age
distribution of waste in landfills in the U.S. was accurately represented by the age
distribution of landfills used to develop the regression model.
Variables Used in the Conversion Factor Derivation
The MASS of waste in place is expressed in metric tons. Data on mass of refuse are a
large source of error as they are gathered by site operators and are not always properly
documented. Mass of refuse can be calculated by several different methods. At a few
sites the trucks are weighed at the gates (U.S. EPA, 1992). In most cases, however, the
operators keep count of the number of trucks and estimate the load. Refuse density in
the U.S., has been estimated to range from 500 to 1,300 lbs/yard3 (300 to 750 kg/m3)
depending on the characteristics, compaction, and humidity of the waste.
27
-------
The GAS FLOW is measured in m3/min or ft3/min. Gas flow data need to be defined
at a certain standard pressure and temperature. Within the EPA, a standard pressure of
1 atmosphere (atm) and a standard temperature of 20°C (68°F) are often used.
Nevertheless, there is reason to adopt a standard temperature of 16°C (60°F), as this is
the temperature used throughout the LFG industry as well as the petroleum and natural
gas industries (Roqueta, 1992; Anderson, 1992 and American Gas Association, 1985).
The inaccuracy of flow meter readings plays a minor role in overall uncertainty of landfill
emissions compared to other parameters used in estimating CH4 emissions from landfills.
One source of error can be found in the conversion of actual gas flows to dry standard
conditions. This conversion is accounted for in the density calculation.
The DENSITY of CH4 at 0°C and 1 atm was calculated assuming that CH4 is an ideal
gas. The density of CH4 at 1 atm and 60°F (16°C) is 677 g/m3. This calculation is
demonstrated in Appendix C.
The average relative CONCENTRATION of CH4 in LFG for 21 field study sites was
50.1 percent while the averages for the individual landfills ranged from 40.2 to
58.1 percent (Peer et al.t 1992; Roqueta, 1992 and Anderson, 1992). In this report, a
concentration of 50 percent was assumed (factor = 0.50).
The OXIDATION of CH4 is estimated to be 10 percent, |i.e., o = 0.10 (Mancinelli and
McKay, 1985)]. Whalen (1990) found a large range of oxidation rates, leading him to
estimate that as much as 50 percent of landfill CH4 might be oxidized before it reaches
the surface. Due to the common occurrence of cracks and fissures in the landfill surfaces,
which would reduce contact between oxidizing organisms and the LFG, the lower value of
10 percent is used. More research is needed to determine the effect of oxidation on
potential emissions.
The RECOVERY EFFICIENCY of LFG projects is thought to be highest when the
projects are undertaken to comply with regulatory programs; however, this cannot be
assumed in all cases. The overall recovery efficiency is typically affected by well spacing
and the presence or permeability of the cover layer. The only published estimate of gas
recovery efficiency is based on expert judgements and gives a most probable value of
75 percent with lower and upper bounds of 50 to 90 percent (Augenstein and Pacey, 1990).
28
-------
TRENDS IN WASTE MANAGEMENT AND THEIR IMPACT ON FUTURE
EMISSIONS
BACKGROUND
Landfill practices in the U.S. are undergoing changes that will affect LFG emissions
and, consequently, CH4 emissions from landfills. Future landfill CH4 emissions will be
influenced by several factors:
¦ Amount of waste landfilled, which depends on:
• Population growth.
• Per capita waste generation, which depends in part on economic growth.
• Increased public awareness of the consequences and hazards of waste
generation and disposal.
• Regulatory requirements affecting waste landfill practices.
¦ CH4 recovery, which depends on:
• Regulatory requirements for controlling emissions from landfills.
• The price of energy.
• Measures to promote the use of CH4 from landfills.
• Availability of more cost-effective CH4 recovery technologies.
The influence of these factors will result in a number of changes, including:
¦ Changes in solid waste generation (e.g. resulting from source reduction).
¦ Changes in solid waste composition.
¦ Increased recycling, composting, and other methods of waste treatment.
¦ Changes in landfill waste management.
¦ Increased control or recovery of LFG as a result of regulatory requirements and
evolution of advanced technologies.
The trends resulting from these changes are discussed in more detail in the following text.
TRENDS IN SOLID WASTE GENERATION, COMPOSITION, AND
MANAGEMENT
Sewage sludge and distinct types of industrial waste contain degradable organic
carbon fractions that may contribute to LFG generation. Other types of non-MSW, for
instance, demolition and construction debris, fly ash, and also wood contain no or very
little degradable organic carbon. Based on data from other countries (Carra and Cossu,
1990) it seems that wastes with little or no organic carbon make up the majority of the
non-MSW stream. It is therefore assumed that MSW is the prime contributor to CH4
generation. In addition, no records on the generation and disposal of non-MSW in the
U.S. were found. Therefore, this section focusses entirely on trends affecting MSW
generation and disposal.
29
-------
Between 1960 and 1990, generation of MSW grew steadily, from 88 million tons to
over 195 million tons per year. Over this period, the per capita generation of MSW
increased from 2.7 pounds to 4.5 pounds per day. OSW has made projections for the year
2000, which indicate that per capita generation of MSW is expected to increase, but at a
substantially slower rate compared to 1992. The projection for the amount of MSW that
will be generated by the year 2000 is 222 million tons per year (U.S. EPA, 1992). MSW
generation is difficult to predict because it depends on various factors including consumer
preference which can change in response to social trends. For example, consumer
preferences can affect the demand for lighter packaging materials or, conversely, more
durable goods. Other factors which can affect the amount of MSW generated include
increased consumer awareness of environmental issues and new regulations.
Recycling has been increasing over the last three decades. In 1992, 17 percent of
generated MSW was recovered for recycling and composting. It is expected that this
percentage will continue to rise. Projected scenarios indicate that between 25 and
35 percent of MSW will be recycled by 2000. Another alternative to landfilling of MSW is
incineration. In 1990, 16 percent of MSW was incinerated. The fraction of MSW that will
be incinerated in the year 2000 is estimated to increase to 21 percent (U.S. EPA, 1992a),
although this increase would be contingent upon public acceptance, which has been an
issue recently. Due to trends in recovery (recycling and composting) and incineration, the
amount of MSW that is landfilled is projected to decrease 16 percent from 130 tons per
year in 1990 to an estimated 109 tons per year in 2000 (U.S. EPA, 1992a). Figure 5
shows MSW generation and waste management trends in the U.S. from 1960 to 2000.
The difficulty in siting new landfills will probably lead to fewer and generally larger
landfills than exist today. Due to increasing regulatory costs, many less efficient landfill
operators may be forced to close reducing the number of landfills even further. This trend
is illustrated by the annual acceptance and remaining design capacity data for landfills in
the OSW-Westat Database from the Survey. Assuming a constant annual waste
acceptance rate, these data suggest that by the year 2000, approximately 38 percent of
currently operational landfills will have closed. The trend toward fewer but larger
landfills should be favorable for the LFG recovery industry, because the feasibility of gas
recovery projects tends to increase with the amount of waste in place. Stricter landfill
design regulations may also benefit the CH4 recovery industry; for example, compulsory
liners would increase CH4 recovery efficiency if the liners reduce lateral gas leakage.
30
-------
Figure 5. Trends in municipal solid waste management, 1960 to 2000.
-------
TRENDS IN LANDFILL GAS RECOVERY AND UTILIZATION
Landfill gas recovery is a comparatively young industry: 75 percent of all landfill CH4
utilization projects are less than seven years old (Governmental Advisory Associates, Inc.,
1991). New technologies to utilize landfill gas are under development; these include
application of fuel cells and the production of compressed gas vehicle fuel and possibly
synthetic fuel. Fuel cells appear particularly promising and the EPA is funding further
research to facilitate development of this technology (Sandelli, 1992). The conventional
purification and combustion of CH4, as it is applied in LFG operations, is a well known
technology. It is unlikely that currently employed equipment can be further improved to
notably enhance efficiency. On the other hand, it is plausible that significant
improvements can be made upstream in the gas collection setup. The regulation of the
flows from different wells or groups of wells is the most important factor in LFG recovery
maximization. The lay-out and design of the collection system poses the biggest challenge
in optimizing overall collection efficiency (Augenstein and Pacey, 1992).
In certain cases, LFG does not have to be converted to electrical energy before it can
be sold. If there is an appropriate buyer, such as a nearby industrial plant, the gas can
be sold directly and used to fuel the plant's boilers to offset the consumption of
conventional fuels. This option requires the least capital investment since there is no
need to tie into the local grid as is done by projects that generate electricity. In these
situations, considerable gains in CH4 recovery efficiency can be made by planning the
project not only to maximize the profit, but also to maximize the amount of CH4
recovered.
REGULATORY ISSUES AFFECTING LANDFILL GAS RECOVERY
Several different regulations affect LFG recovery. Some of these regulations are
aimed at reducing hazardous emissions. Other regulations consist of federal or state
incentives to promote the use of CH4 from landfills as an energy source (Augenstein and
Pacey, 1992).
A regulation has been proposed, directed at reducing emissions from certain new and
existing landfills under the Clean Air Act Section 111(b) and 111(d) (Federal Register,
May 30, 1991). The final rule will probably be promulgated by Summer 1994. This rule,
requiring a gas collection system and add-on control device at affected landfills, is
expected to result in the control of 500 to 700 sites, reducing 5 to 7 tg of CH4 per year
(Thorneloe, 1994b). Although energy conversion is not likely to be required in the final
regulation, it would nevertheless be encouraged. Just as flares, LFG energy utilization
projects reduce toxics, non-methane organic compounds and CH4. Also, there are
additional benefits associated with the utilization of this non-fossil fuel source, such as
the potential offsets from coal-fired power plants.
To illustrate the significance of this potential energy source, comparisons were made
using recent Department Of Energy statistics (DOE, 1993). For the year 2000, the
potential energy of LFG for the sites affected by this rule is estimated to be equal to
2.7 percent of the annual U.S. coal consumption, comparable to 6.7 x 10° kWh. Electricity
32
-------
generation with LFG projects, instead of coal-fired power plants and thereby offsetting
fossil fuel consumption would result in a reduction of C02, S02, NOx, and other
pollutants. Assuming recovery of all the gas that is available, there is potentially a
savings of 4.2 million tons of C02, 1,300 tons of S02, and 12,000 tons of NOx (year 2000).
Currently, data are insufficient to calculate the net reductions more accurately. EPA has
research underway through its Office of Research and Development to develop a
methodology for use by States in considering the offset in emissions associated with LFG
utilization projects so that the overall environmental benefits of these projects can be
considered in permitting applications.
Many other state and federal regulations exist regarding the control of LFG. State
regulations proposed for California seem stricter than the federal regulation described
above. California's draft guidelines propose that energy conversion projects must meet
best available control technology (BACT) criteria (Augenstein and Pacey, 1992).
The U.S. Congress and several state legislative bodies have shown an intent to
encourage and facilitate the use of energy from small-scale alternative sources such as
LFG. This intent has resulted in various credits and incentives such as those in the
Public Utility Regulatory Policies Act (PURPA). This Act provides the structure in which
small-scale electricity producers can sell their power to utility grids. Under the act,
utilities are required to buy the alternative electricity at "avoided cost," which is the sum
of costs the utility would incur if it had to produce the energy itself.
Regulations also exist that hinder the development of LFG recovery projects. Under
these regulations, the recovery systems are considered as sources of themselves. These
regulations do not take into account the offset of pollutants from the landfill surface or
energy savings associated with LPG recovery systems (Thorneloe, 1992b).
ESTIMATE OF FUTURE EMISSIONS
In summarizing and comparing the different trends in MSW generation, waste
management, LFG utilization and pending legislation, two main trends can be identified:
less MSW will be landfilled and gas recovery projects should become more feasible in the
future. Due to legislative and economic pressure, there will be a tendency toward larger
and fewer landfills. As small landfills begin to close in favor of larger landfills, there will
be a shift from Size Class I into Size Class II and/or III. The regression lines developed
for Size Class II and III have slopes that are less steep than the segment for Size Class I
(Figure 2), so this would lead to a reduction of overall emissions.
The size classes and equations for the three segments of the Regression curve are:
I x < 1.128
II 1.128 4.082
where: x
= welled waste, million metric tons or tg.
-------
Innovation in gas collection systems and gas technology can be expected. This
change, as well as the trend toward larger landfills will make gas recovery projects more
feasible. Yet the main factor determining the feasibility of a LFG recovery project will
remain the price for which the LFG, or power derived from it, can be sold. As a result of
increased source reduction, recycling, composting and combustion the yearly amount of
landfilled waste will continue to decrease (U.S. EPA, 1992), which will eventually lead to
a reduction in annual CH4 emissions rates from landfills. However, this is somewhat
offset by increasing population.
However, the shifts towards larger landfills and the influence of changes in source
reduction, recycling, composting and combustion may be overshadowed by the effect of the
new landfill rule. This statement can be underwritten by the fact that, according to the
Survey, 14 percent of all landfills had 1 million or more metric tons of waste in place.
These sites produced 83 percent of all the landfill CH4 emissions (1986). This illustrates
that regulation of a small number of large landfills could possibly achieve a considerable
reduction in CH4 and other emissions from landfills.
34
-------
SUMMARY AND CONCLUSIONS
The estimation of U.S. CH4 emissions from landfilled waste is part of a bigger effort
by AEERL, to obtain global greenhouse gas emissions data (Thorneloe, 1994a). Methane
flow rates from landfills with LFG recovery systems can be used as surrogates for CH4
generation and subsequently for CH4 emissions. The AEERL in conjunction with the
Solid Waste Association of North America collected data on 112 U.S. LFG recovery
projects, 105 of which are included in the ORD Database (Appendix D).
The development of a regression model used for estimating CH4 emissions, which
relates LFG flow rates to waste in place data from the ORD Database, is described in this
document. The model has three linear segments, each of which apply to a distinct landfill
size class. A conversion factor was used to convert LFG flow rates to yearly CH4
emissions. The conversion includes assumptions to account for the recovery efficiency of
LFG projects and for the probable oxidation of CH4 in the topsoil cover of the landfill.
In 1986, OSW conducted a survey, in which detailed information on 1,175 U.S.
landfill facilities was compiled in a data base (OSW-Westat Database). This population
was designed to be a stratified random sample of all U.S. landfills, therefore, its data can
be extrapolated by means of scaling factors to obtain total waste in place for the U.S.
This data base contains data which make it possible to estimate waste in place by two
different methods, the RATE and the DIFF method. Preference is given to the DIFF
method which is based on the difference between Total Design Capacity and Remaining
Design Capacity of the landfill. According to the DIFF method, the total waste in U.S.
landfills in 1986 was 4.7*1015 g (5.2*109 tons). The yearly disposal rate (1986-1992) was
estimated to be 248 tg/yr.
Application of the regression model to the mass waste in place data calculated from
the OSW-Westat Database yields national CH4 emissions from landfills. This value is
increased by the estimate of CH4 emissions from industrial landfills and adjusted for CH4
which is currently recovered or flared. An attempt has been made to update the
emissions estimate for 1992. For 1986 CH4 emissions from U.S. landfills were estimated
at 11 tg/yr with lower and upper bound values of 7 and 15 tg/yr respectively. Methane
emissions from U.S. landfills in 1992 were estimated to be 13 tg/yr with lower and upper
bound values of 9 and 18 tg/yr.
This report details uncertainties which limit the quality of the above estimates. The
main uncertainty arises from the quality of the waste in place data from the OSW-Westat
Database. There are several indications that, since its publication, the data base has
been subject to alterations: the scaling factors seem too high, waste in place data do not
match up, and a density conversion has taken place. The original data base is not
available and none of the alterations to the original data base have been documented;
consequently, a quality assurance review of the data could not be performed. The possible
uncertainties are inflated by the update from 1986 to 1992. Regarding the conversion
factor, the efficiency of the gas recovery system appears to be the largest cause of
uncertainty.
35
-------
Due to regulatory and economic pressure, there will be a tendency toward larger and
fewer landfills. Landfill gas recovery projects become more feasible as the landfill size
increases, which should lead to a reduction in CH4 emissions. As a result of source
reduction, increased recycling, composting and incineration the yearly amount of
landfllled waste is projected to continue to decrease; this will also lead to a reduction in
annual CH4 emissions from landfills. The influence of changes in waste management will
likely be overshadowed by the effect of the anticipated new rule for controlling air
emissions from landfills. This rule, requiring a gas collection system and add-on control
device at affected landfills, is expected to result in the control of 500 to 700 sites, reducing
CH4 emissions by 5 to 7 tg per year.
36
-------
REFERENCES
American Gas Association, Orifice Metering of Natural Gas and Other Hydrocarbon
Fluids. A.G.A. Report No. 3. ANSI/API 2530, September 1985.
Anderson, C. Rust, Inc. Maperville, IL, 1992, Telecon.
Augenstein, D., and J. Pacey. 1990. Modeling Landfill CH, Generation. International
Conference on Landfill Gas: Energy and Environment, 10/17/90. Bournemouth, England.
Augenstein, D., and J. Pacey. Emcon Associates, Inc. 1992. Landfill Gas Energy
Utilization: Technology Options and Case Studies. Prepared for U.S. EPA, Air and
Energy Engineering Research Laboratory, Research Triangle Park, NC. EPA-600/R-92-116
(NTIS PB92-203116).
Barlaz, M.A. 1985. Factors Affecting Refuse Decomposition in Sanitary Landfills.
M.S. Thesis, Dept. of Civil and Environmental Engineering, Univ. of Wisconsin - Madison.
Barlaz, M.A., M.W. Milke, and R.K. Ham. 1987. Gas Production Parameters in Sanitary
Landfill Simulators. Waste Management and Research, 5, p. 27.
Barlaz, M.A. 1988. Microbiological and Chemical Dynamics During Refuse
Decomposition in a Simulated Sanitary Landfill. Ph.D. Dissertation, Department of Civil
and Environmental Engineering, University of Wisconsin - Madison.
Barlaz, M.A., D.M. Schaefer, and R.K. Ham. 1989a. Bacterial Population Development
and Chemical Characteristics of Refuse Decomposition in a Simulated Sanitary Landfill.
Appl. Env. Microbiol., 55, 1, p. 55-65.
Barlaz, M.A., R.K. Ham, and D.M. Schaefer. 1989b. Mass Balance Analysis of
Decomposed Refuse in Laboratory Scale Lvsimeters. ASCE of Environmental
Engineering, 115, 6, pp. 1088 - 1102.
Barlaz, M.A, R.K. Ham, and D.M. Schaefer. 1990. CH, Production from Municipal
Refuse: A Review of Enhancement Techniques and Microbial Dynamics. CRC Critical
Reviews in Environmental Control, Volume 19, Issue 6.
Bhide, A.D., S.A. Gaikwad, and B.Z. Alone. 1990. CH, from Land Disposal Sites in India.
Proceedings of the International Workshop on CH4 Emissions from Natural Gas Systems,
Coal Mining and Waste Management Systems. Environment Agency of Japan, the U.S.
Agency for International Development and the U.S. EPA, Washington, DC. 04/09-13/90.
Bogner, J.E. 1990. Controlled Study of Landfill Biodegradation Rates Using Modified
BMP Assays. Waste Management and Research, 8, p. 329-352.
37
-------
Buivid, M.G. 1981. Fuel Gas Enhancement bv Controlled Landfilling of Municipal Solid
Waste. Resource Recovery and Conservation, 6, p.3.
Campbell, D., D. Epperson, L. Davis, R. Peer, and W. Gray. Radian Corporation. 1991.
Analysis of Factors Affecting Methane Gas Recovery from Six Landfills. Prepared for U.S.
EPA, Air and Energy Engineering Research Laboratory. Research Triangle Park, NC.
EPA-600/2-91-055 (NTIS PB92-101351).
Carra, J.S. and R. Cossu. 1990. International Perspectives on Municipal Solid Wastes
and Sanitary Landfilling. Academic Press, New York, NY.
Emberton, J.R. 1986. The Biological and Chemical Characterization of Landfills.
Proceedings of Energy from Landfill Gas, Solihull, West Midlands, UK, 10/30-31/86.
EMCON Associates. 1982. Methane Generation and Recovery from Landfills. Ann Arbor
Science Publishers, Inc. Ann Arbor, MI.
Federal Register, May 30, 1991. Vol. 56, No. 104, page 24511-24522
Geswein, A. J. 1993. U.S. EPA, Office of Solid Waste and Emergency Response,
Municipal and Industrial Solid Waste Division. Telecon.
Governmental Advisory Associates, Inc. 1991. 1991-92 Methane Recovery from Landfill
Yearbook. Directory and Guide. G.A.A. New York.
Halvadakis, C.P. 1983. Landfill Methanogenesis: Literature Review and Critique.
Technical Report No. 271. Department of Civil Engineering, Stanford University.
IPCC. 1992. Climate Change 1992. The Supplementary Report to the IPCC Scientific
Assessment. Published for The Intergovernmental Panel on Climate Change (IPCC),
World Meteorological Organization/United Nations Environment Programme. Cambridge
University Press. Edited by J.T. Houghton, G.J. Jenkins, and J.J. Ephraums.
Jenkins, R.L. and J.A. Pettus. 1985. In Biotechnological Advances in Processing
Municipal Wastes for Fuels and Chemicals. A.A. Antonopoulos ed. Argonne National
Laboratories. Report ANL/CNSV - TM - 167, p. 419.
Kinman, R.N. 1987. Gas Enhancement Techniques in Landfill Simulators. Waste
Management and Research, 5, p. 13-25.
Mancinelli, R.L. and C.P. McKay. 1985. Methane-Oxidizing Bacteria in Sanitary
Landfills, in Proceedings of First Symposium on Biotechnical Advances in Processing
Municipal Wastes for Fuels and Chemicals. A.A. Antonopoulos ed. Argonne National
Laboratories.
Pacey, J. 1989. Enhancement of Degradation: Large-Scale Experiments, in Sanitary
Landfilling: Process Technology and Environmental Impact, T. Christensen, R. Cossu,
R. Stegmann, ed. Academic Press, London, p. 103-119.
38
-------
Parkin, G.F. and W.F. Owen. 1986. Journal of the Environmental Engineering Division.
ASCE, 112, 5, p. 867.
Peer, R.L., D.L. Epperson, D.L. Campbell, and P. von Brook. Radian Corporation. 1992.
Development of an Empirical Model of Methane Emissions from Landfills. Prepared for
U.S. EPA, Air and Energy Engineering Research Laboratory, Research Triangle Park,
NC. EPA-600/R-92-037 (NTIS PB92-152875).
Pohland, F.G. 1975. Sanitary Landfill Stabilization with Leachate Recycle and Residual
Treatment. Prepared for U.S. EPA, Municipal Environmental Research Laboratory,
Cincinnati, OH. EPA-600/2-75-043 (NTIS PB248-524).
Pohland, F.G. and S.R. Harper. 1986. Critical Review and Summary of Leachate and
Gas Production from Landfills. Prepared for U.S. EPA, Hazardous Waste Engineering
Laboratory, Cincinnati, OH. EPA/600/2-86/073 (NTIS PB86-240181).
Rathje, W.L. 1991. Once and Future Landfills. National Geographic. May. pp. 117-134.
Roqueta, A. 1992. Bryan A. Stirrat & Associates. Diamond Bar, CA. Telecon.
Sandelli, G.J. 1992. Demonstration of Fuel Cells to Recover Energy from Landfill Gas.
Phase 1, Final Report: Conceptual Study. Prepared for U.S. EPA, Air and Energy
Engineering Research Laboratory, Research Triangle Park, NC. EPA-600-R-92-007
(NTIS PB92-137520).
Schroeder, K., R. Clickner, and E. Miller. Westat, Inc. 1987. Screening Survey of
Industrial Subtitle D Establishments. Prepared for U.S. EPA, Office of Solid Waste and
Emergency Response, Washington, DC.
Shelton, D.R. and J.M. Tiedje. 1984. General Method for Determining Anaerobic
Biodegradation Potential. Appl. Env. Microbiol. 47, 4, p. 850-857.
Thorneloe, S.A. 1992a. Landfill Gas Recovery/Utilization - Options and Economics.
Published in Proceedings for the 16th Annual Conference by the Institute of Gas
Technology on Energy from Biomass and Wastes. Orlando, FL. March 15, 1992.
Thorneloe, S.A. 1992b. Landfill Gas Utilization - Options. Benefits, and Barriers.
Published in Proceedings for The Second United States Conference on Municipal Solid
Waste Management. Arlington, VA. June 4, 1992.
Thorneloe, S.A. 1994a. Emissions and Mitigation at Landfills and other Waste
Management Facilities. Published in Proceedings: the 1992 Greenhouse Gas Emissions
and Mitigation Research Symposium, U.S. EPA, Air and Energy Engineering Research
Laboratory, Research Triangle Park, NC. EPA-600/R-94-008 (NTIS PB94-132180).
Thorneloe, S.A. 1994b. Landfill Gas and Its Influence on Global Climate Change, in
Landfilling of Waste: Gas. Publisher Chapman and Hall. In press.
39
-------
U.S. DOE. 1993. Annual Energy Outlook with Projections to 2010. Department of
Energy, Energy Information Administration. January, 1993. DOE/EIA-0383.
U.S. EPA. 1990. Characterization of Municipal Solid Waste in the United States: 1990
Update. U.S. EPA, Municipal and Industrial Solid Waste Division, Office of Solid Waste,
Washington, DC. EPA-530/SW-90-042 (NTIS PB90-215112).
U.S. EPA. 1992. Characterization of Municipal Solid Waste in the United States. 1992
Update. U.S. EPA, Municipal and Industrial Solid Waste Division, Office of Solid Waste,
Washington, DC. EPA/530-R-92-019 (NTIS PB92-207166).
U.S. EPA. 1988. National Survey of Solid Waste (Municipal) Landfill Facilities. U.S.
EPA, Office of Solid Waste and Emergency Response, Washington, DC. EPA/530-SW-88-
034 (NTIS PB89-118525).
Wang, Y., C.S. Byrd, and M.A. Barlaz. 1994. Anaerobic Biodegradabilitv of Cellulose and
Hemicellulose in Excavated Refuse Samples Using a Biochemical Methane Potential
Assay. Journal of Industrial Microbiology, 13 (1994) 147-153.
Whalen, S.C., W.S. Reeburgh, and K.A. Sandbeck. 1990. Rapid CHt Oxidation in a
Landfill Cover Soil. Appl. Env. Microbiol. 56, 11, p. 3405 - 11.
Wolfe, R.S. 1979. Methanogenesis. In Microbial Biochemistry, International Review of
Biochemistry, Vol. 21. J.R. Quayle, ed. University Park Press, Baltimore, MD.
Wolin, M.J. and T.L. Miller. 1985. In Biology of Industrial Microorganisms, ed.
A.L. Dernsin, Benjamin/Cumings Publ. Co. Inc. Menlo Park, CA. p. 189-221.
Young, L.Y. and A.C. Frazer. 1987. Geomicrobiology J., 5, p. 261.
Zehnder, A.J.B. 1982. Microbiology of CH1 Bacteria. Anaerobic Digestion, D.E. Hughes,
ed., Elsevier Biomedical Press B.V., Amsterdam, p. 45.
40
-------
APPENDIX A: STATISTICAL METHODS
INTRODUCTION
The statistical methods employed to obtain estimates of total waste in place and
national methane emissions are described in this appendix.
There are two independent sources of data from which the estimates are derived. The
OSW-Westat Database is used to construct estimates of the total waste in place and the
national landfill size distribution. The ORD Database is required, in conjunction with the
OSW-Westat Database, to produce estimates of national methane emissions.
Landfill size and gas flow rate are denoted generically by x and y, respectively.
Statistical estimates are reported for two methods of calculating waste in place x, the
RATE and DIFF methods, as described in the text. The statistical theory is the same for
both methods so that no distinction between the methods is made henceforth.
The value of x for the landfill in the ilh stratum of the OSW-Westat Database is
denoted xv-, where j = 1,and i = 1,2; i = 1 denotes a small landfill and i = 2 denotes
a large landfill. These designations were determined for each landfill in a preliminary
classification of landfills reported in the Survey. Thus, nt is the number of landfills
surveyed in the ith stratum. Strata 1 and 2 consist of small and large facilities
respectively as determined by a preliminary classification of the landfills in the Survey.
Associated with each landfill in the OSW-Westat Database is a scaling factor, r^,
which is simply the inverse of the stratum sampling fraction. The Survey reports that
the target sampling fractions were 52 percent of the large' facilities and 13 percent of the
'small' facilities.
Thus the target sampling fractions suggest that the scaling factors for all landfills in
the first stratum (small landfills) should be 1/0.13 = 7.69, and for the second stratum
(large landfills), the scaling factors should be 1/0.52 = 1.92. The scaling factors reported
in the OSW-Westat Database are for large and small landfills are 7.00 and 2.00,
respectively (with a few values of 1 and 3 that must be erroneous). The source of the
discrepancy between the recorded and anticipated scaling factors is unknown. Possibly
the difference is due partly to a better-than-expected response rate, partly to rounding (for
convenience perhaps), and perhaps in part to a recalculation of the number of 'eligible'
landfills as described in the Survey.
41
-------
The scaling factors used in the formulas that follow are those recorded in the OSW-
Westat Database with the modification that the values of 1 and 3 were set equal to 2.00.
Thus, = 7.00 for small landfills (i.e., = 1) and = 2.00 for large landfills (i.e., i =2).
The following quantities figure prominently in the formulas that follow:
"i
Ni=HriP *' = 1.2;
y = l
(8)
N =NX + N2;
n = nt + n2.
Where nl and n.A are the number of small and large landfills, respectively, in the
Subtitle D Municipal Landfill Survey. With this notation, the total number of U.S.
landfills (also an estimate, as described in the Subtitle D Municipal Landfill Survey) is N;
similarly N, and N2 are estimated stratum sizes.
The estimate of total waste in place is
X = ATjJj + Njfo , (9)
where:
*i. = f £ V < » 1.2- (10)
ni j' 1
and: xtj = the amount of waste in place in landfill j in stratum i
n, = the target sampling fraction for stratum i.
This formula calculates total waste in place as the sum of the estimated waste at small
and large landfills. Estimated waste in place at each landfill type is the product of the
average waste in place of the surveyed landfills and the total number of landfills, by
stratum.
The standard error of the estimate is calculated form the formula
sx =
w.
«i) , NJNz-nJ
5 j + *5
(ii)
"2
where s \ is the sample variance of xn,...jcin(0, i = 1, 2. Applying these formulas to the data
produces the results in Table 9.
42
-------
TABLE 9. ESTIMATES OF TOTAL WASTE IN PLACE; STATISTICAL CONSIDERATIONS
Method
Comparison
DIFF
RATE
DIFF
RATE
Landfills in Data Base
1,175
Landfills with Complete
Records
1,064
1,064
1,034
Total Waste in Place (g)
(standard error)
4.86*1015
(4.73*1014)
6.61*1015
(5.04*1014)
4.72*1015
(4.72*1014)
6.36*1015
(5.03*1014)
Estimated Landfills (N)
6,363
6,376
6,223
6,223
Notes: • Standard error in parentheses. Approximate 95 percent confidence intervals are obtained by adding
plus/minus two standard deviations to the estimate. Thus for the Comparison/DIFF method the
estimate of total waste in place is 4,717 tg. The 95 percent confidence interval is (between 3,773 and
5,661 tg).
• Method denotes Ihe various ways of accommodating incomplete records as described in the text.
Remarks:
1. The statistical theory is based on the assumption that N, and N2 are known, not
estimates as they are here. The effect of using estimated strata sizes is to increase
the variability of the estimate. Consequently the standard errors reported in Table 9
are optimistic, i.e., they undervalue the variability in the estimate of total waste in
place.
2. The statistical theory also supposes that x is measured precisely, i.e., there are no
estimation errors or reporting errors, either systematic (bias) or random (noise) in the
values of x-., i.e., the estimates of waste in place. Random errors increase the
variability of the estimate although the increased variability should, to a great extent,
be reflected in the reported standard error. Systematic errors are more problematic.
It is not possible to evaluate the effect of systematic errors with the available data.
3. The estimate of total waste in place, X may also be calculated as:
2 «<
i=i j=i
This representation makes clear the role of the scaling factors as weights in a
weighted sum of the xtJ from the surveyed landfills.
43
-------
ESTIMATING CH4 USING REGRESSION RESULTS
The key variates in the ORD Database are y, the LFG flow rate, and x, the estimates
of waste in place. Since these data are not stratified, values of y and x for an individual
landfill are denoted with a single subscript, e.g., y} and xJtj = 1 ,...m where m is the
number of landfills in the ORD Database. The ORD Database contains information on
112 landfills; however, seven landfills have missing values for either x or y and thus data
from only 105 landfills are used in the statistical analyses.
Two methods were examined, Ratio estimation and Regression modeling. Ratio
estimation is the simpler of the two; it generates a straight line running through the
origin. There is some evidence that the Ratio method is not appropriate for the available
U.S. data. The Regression modeling approach was adopted to overcome deficiencies with
the Ratio method. It uses different emission factors for different landfill sizes.
Consequently, it can only be used when country-specific information about landfill size is
available, as is the case in the U.S.
Ratio Estimation
The ratio estimate of total landfill CH4 emissions is
where CF is the conversion factor described in Section 2; £ is the estimate of total waste
in place from (9), and
which is simply the average LFG flow rate emitted per ton of landfilled waste. Therefore
the Ratio method can be viewed as an emission factor.
y and x are calculated from
The regression line resulting from application of the Ratio method is shown in Figure 6.
The regression curve developed by the Regression method, as in Figure 2, is included for
comparison.
Y = CFRX
(13)
R =y I x
(14)
1 m 1 m
? = and x = -$>r
mpt J mp( 1
(15,16)
44
-------
3
C
300 t
3 250-
S 200 -
-------
The standard error of ¥ is computed according to the formula
sf = CF^sl + X2s\ + £2sj)
(r
where s% is the standard error of the estimate of total waste in place (11), and s,j is the
standard error of & calculated from
sr =
mPj-i
(li
Estimates of total yearly methane emissions obtained by ratio estimation are displayed
the Table 10.
TABLE 10. RATIO fi AND RATIO ESTIMATES OF METHANE EMISSIONS FOR 1986.
DIFF
RATE
Landfills in data base
1,175
Landfills with complete records
1,034
Total waste in place (g)
(standard error)
4.72*1015
(4.72*1014)
6.36*1015
(5.03*101")
Ratio R = y/x (m3/g min)
(standard error)
8.76*1012
(0.79*1012)
Gas flow rate (m3/min)
(standard error)
4.13*10"
(5.56*103)
5.59*104
(6.69*103)
Conversion Factor (g min/yr m3)
213*10®
Emissions estimate (g/yr)
(standard error)
8.81*10'2
(1.19*1012)
11.92*1012
(1.42*1012)
Emissions estimate for industrial landfills, (g/yr)
0.55*1012
Methane currently recovered or flared, (g/yr)
1.1*1012
Estimated total U.S. methane emissions, (g/yr)
8.3*1012
11.4*1012
Note: Standard errors are given in parentheses. The standard deviation in the emission factor,
developed from the landfill gas flow data is 9 percent. Approximate 95 percent confidence
intervals are obtained by adding plus/minus 18 percent
46
-------
Remarks:
1. Ratio estimation is justified statistically when either: (i) the landfills in the ORD
Database can be regarded as a simple random sample of all landfills in the U.S.; or
(ii) landfill gas flow rate is proportional to landfill waste in place, with the constant of
proportionality independent of landfill size.
2. The landfills in the ORD Database were not randomly chosen from among all landfills
in the U.S. and they are much larger than the averaged-sized landfill. Thus there is
ample reason to suspect that (i) does not hold.
3. The second condition, (ii), seems reasonable; however, in light of the anecdotal
evidence that CH4 generation is greater in smaller landfills, (ii) is certainly suspect.
Furthermore, the data tend to support the anecdotal evidence. Among the smaller
landfills the majority of gas flow rates lie below the line, ^ =&£. However, the
remaining smaller landfills vary substantially above the line, so much so that the
average generation rate of the smaller landfills lies well above the line. This is
evidence that the Ratio method may be inappropriate — the same behavior would
also result from a tendency for operators of smaller landfills to under-report the
amount of waste in place.
Regression Modeling
The ORD Database can be used to estimate a model for landfill gas flow rate as a
function of landfill size. The model is used to 'predict' gas flow rate for each landfill in
the data base, and then a national methane emission estimate is computed by summing
the predicted gas flow rates over all landfills and applying the conversion factor. A
weighted sum is employed as in the alternative definition of X equation (12).
Ratio estimation may be viewed as a special case of the method just described. If the
appropriate gas flow rate model has the form
y = P* (19)
that is, gas flow rate is proportional to the amount of waste in place, then an estimate of
P can be obtained by averaging both sides of (19) over the values of y} and x} in the ORD
Database. The resulting estimate would be
p = I (20)
x
which is just R, the ratio estimate described previously in Equation (14). Next every
landfill in the OSW-Westat Database is assigned the 'predicted' gas flow rate yu = $xtj.
41
-------
The weighted sum
2 "i
CF £ E Vr
(21)
i-i j=i
is an estimate of national emissions. Inspection of (21) shows that it is identical to ?
defined in Equation (13).
The problems with ratio estimation discussed previously can be alleviated to a great
extent by replacing the simple linear model (19) with a model that allows for differential
gas flow rates depending on the size of the landfill. This can be accomplished with any
number and variety of models. The particular model is not critical. The segmented model
is sufficiently flexible to reflect the higher gas flow rates of the smaller landfills while
providing an adequate lit to the larger landfills without unnecessary complexity.
The equation describing the piecewise linear model depends on five parameters, three
representing the slopes of the linear segments, (3 = ( p„ p^, pa ) and two determining the
Tiinge' points, 9 = (Bp 02 ). These parameters were estimated by weighted least squares
with weights proportional to the reciprocal of landfill sizes. In other words, if fi(x, p, Q)
denotes the function defining the piecewise linear model, then (5 and 8 minimize
over all values of p and 0. This is analogous to the mathematical criterion for estimating
P (equivalently R) when the simple linear model (19) is employed.
Given the similarity of the estimated piecewise linear model to the simple model
y = Ax, it is natural to question whether the difference is significant, either practically or
statistically. Table 11 displays estimates of national methane emissions derived from the
estimated piecewise linear model.
/-I X,
(22)
48
-------
TABLE 11. REGRESSION ESTIMATES OF METHANE EMISSIONS, STATISTICAL
CONSIDERATIONS
Method
DIFF
RATE
Comparison
DIFF
RATE
Landfills in Data Base
1,175
Landfills with Complete Records
1,064
1,064
1,034
Total Waste in Place (g)
(standard error)
4.86*10'5
(4.73*10'4)
6.61 *1016
(5.04*1014)
4.72*101S
(4.72*1014)
6.36*10'5
(5.03*10'4)
Gas Flow Rate (m3/min)
(standard error)
5.39*10*
(7.00*103)
6.83* 104
(8.20* 103)
5.25*104
(6.83*103)
6.66*104
(7.91 *103)
Conversion Factor (g min/yr m3)
213*10®
Emissions Estimate (g/yr)
(standard error)
11.49*1012
(1,50*1012)
14.95*1012
(1.75*1012)
11.21*1012
(1.46*1012)
14.21*10'2
(1.69*101Z)
Note: Standard errors are given in parentheses. Approximate 95 percent confidence intervals are obtained by adding
plus/minus two standard deviations to the estimate.
Remarks:
1. The estimates in Table 11 are noticeably greater than those derived from the ratio
method described previously. The reason is that the vast majority of landfills are
small and even the slightest difference between models for small landfills will induce
a large difference in national estimates.
2. The decision to use the piecewise linear model was based in part on the anecdotal
evidence of higher generation rates at smaller landfills. However, in principle it is
also possible to make a purely data-based comparison of the simple linear and
piecewise linear models, that is, to perform a statistical test comparing the fit of the
models. There is a problem in that the statistical theory for comparing piecewise
linear models is generally less well understood than the theory for comparing linear
models. Nevertheless the results of the test gives some guidance as to the
appropriate choice of model. The normal-theory likelihood ratio test for comparing
the linear and piecewise linear models indicates that the latter model fits the data
significantly better (p-value = 0.02) than the simple linear model. However, this
result should be interpreted cautiously, given the approximate nature of the
statistical theory.
3. Calculating standard errors for the emission estimates derived from the more
complicated model is difficult. The standard errors reported in Table 11 were
49
-------
calculated as 13 and 12 percent of the emissions estimates for the difference and rate
methods respectively. These percentages were derived from the standard errors and
emissions estimates from the ratio estimates. The standard errors derived in this
fashion should be reasonably accurate, although they are probably biased low.
4. The combination of sparseness of the data for large values of x and the
heteroscedastie variation of gas flow rates causes problems when estimating the
parameters of the piecewise linear model. The determination of the 'hinge' points is
particularly sensitive to these characteristics of the data. Problems were avoided by
minimizing (22) subject to an upper bound on the largest liinge' point.
5. The flexibility of the piecewise linear model also makes the estimated model more
sensitive to outlying or extreme data points, and there are certain landfills in the
ORD Database whose exclusion from the data have a pronounced effect on the
estimated model and on the final emissions estimates. When suspected outliers are
removed from the data the emissions estimates so calculated are generally in closer
agreement with the ratio estimates. Since it was not possible to confirm that any
suspected outliers were in fact erroneous, only the results for the complete data set
are presented.
50
-------
APPENDIX B: COMPARISON OF ESTIMATES OF METHANE EMISSIONS
FROM U.S. LANDFILLS
TABLE 12. COMPARISON OF ESTIMATES OF METHANE EMISSIONS FOR U.S. LANDFILLS
SOURCE
EMISSIONS ESTIMATE (tg/yr)
REMARKS
Lower bound
Mid-point
Upper bound
Bingemer & Crutzen1
(1987 estimate)
11
16
21
Pro-rated from global estimates.
Uses a mass balance
approach.
IPCC/OECD Method 2
(1990 estimate).
20
Uses same waste data as this
study (for comparison).
Augenstein3
(1990 estimate)
3
6
8
Uses a gas generation model,
based on decomposition
kinetics.
OAP, Global Change
Division4
(1990 estimate)
8
10
12
Uses a model similar to this
study with lower waste in place
estimates.
This study (1986)
7
11
15
(1990)
8
13
17
(1992)
9
13
18
1 Bingemer, H.G. and P.J. Crutzen. 1987. The Production o{ Met harm from Solid Wastes.
Journal of Geophysical Research, Vol. 92, No. D2.
2 OECD (Organization for Economic Cooperation and Development). 1991. Estimation of
Greenhouse Gas Emissions and Sinks. Final Report from OECD Experts Meeting, 18-21
February 1991, Paris, France. Prepared for Intergovernmental Panel on Climate Change.
OECD, Paris, France.
3 Augenstein, D.C. 1990. Greenhouse Effect Contributions of United States Landfill Methane.
GRCDA 13th Annual Landfill Gas Symposium, Lincolnshire, II.
4 U.S. EPA, Office of Air and Radiation (OAR), Global Change Division. 1993. Anthropogenic
Methane Emissions in the United States: Estimates for 1990. Report to Congress.
51
-------
APPENDIX C: CALCULATIONS FOR THE DENSITY OF METHANE
Ideal gas law: PV/nRT = constant,
where
R = 8314.41 Joule per Kelvin per kilomole [J/(K.kmole)]
for
P = pressure in Pascal
V = volume in m3
n - number of kmoles of gas
T = temperature in K
Therefore, V = nRT/P = 22.4138 m3 for:
with
n = 1 kmole
T = 273.15 K (which = 0°C)
P = 101,325 Pa (which = 1 atm)
The molecular weight of 1 kmole of CH4 = 16,040 g
So the density p = 16,040/22.4138 = 715.631 g/m3
To convert to a standard with T = 60°F = 288.71 K (and all other parameters unchanged),
multiply by 273.15/288.71 which gives p = 677.062 g/m3.
52
-------
APPENDIX D: FIELD DATA
TABLE 13. WASTE AND LFG FLOW RATE DATA FROM THE ORD DATABASE
Landfill
Welled Waste
Identification
(Millions of
Welled Waste
GAS
Gas Flow Rate
Gas Flow Rate
Code
U.S. Tons)
(tg)
(MCFD)
(CFM1)
(CMM2)
323
0.1
0.0
0.04
28
0.7
331
0.4
0.3
0.05
35
0.9
19
0.6
0.5
0.42
292
8.2
261
0.8
0.7
0.90
625
17.7
334
1.0
0.9
0.86
597
16.9
272
1.2
1.0
2.10
1458
41.3
298
1.2
1.0
0.48
333
9.4
320
1.3
1.1
0.72
500
14.1
208
1.4
1.2
4.36
3028
85.7
318
1.4
1.2
0.72
500
14.1
213
1.5
1.3
0.50
347
9.8
304
1.5
1.3
0.70
486
13.7
330
1.6
1.4
0.45
313
8.8
210
1.6
1.4
0.50
347
9.8
290
1.6
1.4
0.70
486
13.7
308
1.7
1.5
0.42
292
8.2
9
1.8
1.6
1.58
1097
31.0
232
1.8
1.6
0.80
556
15.7
244
1.9
1.7
0.40
278
7.8
284
2.0
1.8
0.90
625
17.7
255
2.0
1.8
2.40
1667
47.2
802
2.0
1.8
0.69
479
13.5
15
2.0
1.8
2.30
1597
45.2
337
2.0
1.8
0.70
486
13.7
360
2.1
1.9
0.80
556
15.7
278
2.2
1.9
0.80
556
15.7
302
2.5
2.2
3.00
2083
59.0
228
2.5
2.2
0.81
563
15.9
328
2.6
2.3
0.90
625
17.7
231
2.6
2.3
1.00
694
19.6
227
2.7
2.4
3.50
2431
68.8
263
2.7
2.4
0.30
208
5.9
342
2.8
2.5
1.00
694
19.6
313
2.8
2.5
2.50
1736
49.1
315
2.8
2.5
1.44
1000
28.3
53
(continued)
-------
TABLE 13. WASTE AND LFG FLOW RATE DATA FROM THE ORD DATABASE (Continued)
Landfill
Welled Waste
Identification
(Millions of
Welled Waste
GAS
Gas Flow Rate
Gas Flow Rate
Code
U.S. Tons)
(tg)
(MCFD)
(CFM1)
(CMM2)
317
2.9
2.6
1.73
1201
34.0
8
3.0
2.7
1.05
729
20.6
332
3.0
2.7
2.00
1389
39.3
221
3.0
2.7
2.20
1528
43.2
355
3.0
2.7
1.37
951
26.9
316
3.0
2.7
0.40
278
7.8
247
3.0
2.7
0.80
556
15.7
23
3.1
2.8
0.30
208
5.9
299
3.1
2.8
1.30
903
25.5
6
3.1
2.8
1.30
903
25.5
305
3.3
2.9
2.50
1736
49.1
307
3.4
3.0
0.50
347
9.8
11
3.5
3.1
1.08
750
21.2
257
3.5
3.1
2.00
1389
39.3
7
3.5
3.1
1.50
1042
29.5
239
3.5
3.1
0.86
597
16.9
240
3.5
3.1
1.08
750
21.2
258
3.6
3.2
2.00
1389
39.3
336
3.6
3.2
2.00
1389
39.3
218
3.7
3.3
0.85
590
16.7
295
3.8
3.4
0.80
556
15.7
238
3.8
3.4
1.20
833
23.6
209
4.0
3.6
1.00
694
19.6
16
4.0
3.6
2.10
1458
41.3
312
4.0
3.6
0.78
542
15.3
220
4.0
3.6
1.44
1000
28.3
285
4.0
3.6
1.00
694
19.6
279
4.3
3.9
0.85
590
16.7
233
4.5
4.0
1.20
833
23.6
241
4.5
4.0
0.60
417
11.8
286
4.8
4.3
1.40
972
27.5
274
5.0
4.5
1.00
694
19.6
219
5.0
4.5
1.70
1181
33.4
354
5.0
4.5
2.00
1389
39.3
10
5.8
5.2
2.71
1882
53.2
326
5.8
5.2
2.60
1806
51.1
300
6.0
5.4
1.30
903
25.5
54
(continued)
-------
TABLE 13. WASTE AND LFG FLOW RATE DATA FROM THE ORD DATABASE (Continued)
Landfill
Identification
Code
Welled Waste
(Millions of
U.S. Tons)
Welled Waste
(tg)
GAS
(MCFD)
Gas Flow Rate
(CFM1)
Gas Flow Rate
(CMM2)
234
6.1
5.5
1.90
1319
37.3
314
6.5
5.8
2.88
2000
56.6
321
6.8
6.1
1.94
1347
38.1
294
7.0
6.3
2.10
1458
41.3
24
7.3
6.6
2.00
1389
39.3
2
7.7
6.9
2.50
1736
49.1
1
8.2
7.4
4.00
2778
78.6
352
8.3
7.5
3.00
2083
59.0
229
9.0
8.1
4.30
2986
84.5
246
9.0
8.1
4.90
3403
96.3
237
9.7
8.7
1.90
1319
37.3
335
9.9
8.9
4.00
2778
78.6
216
10.0
9.0
5.00
3472
98.3
291
10.0
9.0
3.82
2653
75.1
18
11.3
10.2
6.90
4792
135.7
3
11.6
10.5
4.00
2778
78.6
289
12.0
10.8
0.80
556
15.7
207
12.0
10.8
3.00
2083
59.0
25
12.6
11.4
4.00
2778
78.6
245
13.4
12.1
3.00
2083
59.0
327
14.0
12.6
5.40
3750
106.2
343
14.0
12.6
8.00
5556
157.3
214
14.3
12.9
2.00
1389
39.3
4
15.3
13.8
10.00
6944
196.6
217
16.0
14.5
4.00
2778
78.6
347
17.5
15.8
2.02
1403
39.7
230
18.0
16.3
11.00
7639
216.3
275
20.0
18.1
6.00
4167
118.0
249
25.0
22.6
23.00
15972
452.3
14
28.0
25.3
9.50
6597
186.8
303
30.0
27.2
7.20
5000
141.6
251
60.0
54.4
37.50
26042
737.5
226
66.0
59.8
14.40
10000
283.2
1 CFM = cubic feet per minute.
2 CMM = cubic meter per minute.
55
-------
APPENDIX D: FIELD DATA (continued)
TABLE 14. SUMMARY OF LANDFILL DATA FROM PEER ET AL. (1992)*
Landfill
Number
Refuse
Mass
(106 Mg)
Average
Refuse
Age
------- |