Electric Power Generation Using A
Phosphoric Acid Fuel Cell On A
Municipal Solid Waste Landfill Gas Stream

Technology
Verification Report

by

Sushma Masemore and Steve Piccot
Greenhouse Gas Technology Verification Center
Southern Research Institute
P.O. Box 13825
Research Triangle Park, NC 27709

EPA Cooperative Agreement CR 826311-01-0

EPA Project Officer: David A. Kirchgessner
Air Pollution Prevention and Control Division
National Risk Management Research Laboratory
Research Triangle Park, NC 27711

Prepared For:

U.S. Environmental Protection Agency
Office of Research and Development
Washington, D.C. 20460


-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY	^

Office of Research and Development	//

^	Washington, D.C. 20460	EjL V/

ENVIRONMENTAL TECHNOLOGY
VERIFICATION PROGRAM

ENVIRONMENTAL TECHNOLOGY VERIFICATION PROGRAM
VERIFICATION STATEMENT

TECHNOLOGY TYPE:

LANDFILL GAS CLEANUP AND PHOSPHORIC ACID FUEL CELL SYSTEM

APPLICATION:

POWER PRODUCTION FROM WASTE LANDFILL GAS

TECHNOLOGY NAME:

GPU AND PC25 ™ 200 kW FUEL CELL

COMPANY:
ADDRESS:

INTERNATIONAL FUEL CELLS CORPORATION

195 GOVERNORS HIGHWAY

SOUTH WINDSOR, CONNECTICUT 06074

PHONE:

(860) 727-2388

PROGRAM DESCRIPTION

The U. S. Environmental Protection Agency's (EPA) Office of Research and Development has created the
Environmental Technology Verification (ETV) Program to facilitate the deployment of promising environmental
technologies. Under this program, third party performance testing of environmental technology is conducted by
independent Verification Organizations. Their goal is to objectively and systematically evaluate technology
performance under strict EPA quality assurance guidelines. The EPA's Air Pollution Prevention and Control
Division has selected Southern Research Institute as the independent Verification Organization to operate the
Greenhouse Gas Technology Verification Center (the Center). With the full participation of the technology
developer and users, the Center develops testing plans, and conducts field and laboratory tests. The test results are
analyzed, peer reviewed, and then distributed to industry, regulatory, vendor, and other groups interested in the
data.

TECHNOLOGY DESCRIPTION

For several years, International Fuel Cells (IFC) Corporation has employed the commercially available phosphoric
acid fuel cell (PC25™) to generate electricity from natural gas. This fuel cell unit can also be used at municipal
solid waste landfills to convert landfill gas into electric power. This application requires a supplemental gas
treatment unit (GPU) to remove sulfur and halide compounds present in the landfill gas (LFG). The combined
GPU and PC25™ Fuel Cell system provides a means for utilizing waste landfill gas, thus, reducing methane
emissions and other air pollutants.

The design of the GPU is dictated by the gas purity requirements of the fuel cell, and the composition and physical
properties of the incoming LFG. The cleaned waste gas is then converted into electric power for on-site use or
distribution to an electric grid. In the GPU, hydrogen sulfide is first removed via adsorption on an activated
carbon bed, which is used to catalyze the conversion of H2S into elemental sulfur. Additional water, heavy
hydrocarbons, sulfides, and other contaminants are removed through the removal system consisting of a low
temperature cooler, carbon bed, dryer bed, and particulate filter. A heat exchanger is used to ensure the gas
temperature meets fuel cell inlet requirements. The PC25™ fuel cell consists of a fuel processing system, an
electrical conversion system, and a thermal management system. In the fuel processing section, treated LFG is

EPA-VS-GHG-01

The accompanying notice is an integral part of this verification statement

ii

August 1998


-------
converted to hydrogen and carbon dioxide for introduction into the fuel cell stack. The fuel treatment process
consists of a low temperature fuel preprocessor which removes the residual contaminants from the treated gas, a
fuel reformer, and a low temperature shift converter where the exhaust from the reformer is further processed. The
hydrogen from the process fuel stream is then combined electrochemically with oxygen from the air to produce
electricity in the fuel cell stacks. The DC current produced is converted into AC in a power-conditioning package.
The PC25™ is designed to produce 200 kW of electric power from natural gas. With LFG, the PC25™ unit
generates less power due to lower heating value of LFG.

VERIFICATION DESCRIPTION

This verification statement summarizes the results of tests conducted to verify the performance of a combined
GPU and PC25™ fuel cell system operating on LFG. These tests were conducted at two sites where the LFG flow
rates, composition, heating value, and contaminant levels are representative of the U.S. landfill population. The
performance of the GPU was evaluated by comparing the sulfur and halogen concentrations in the GPU outlet gas
with the levels required to effectively operate the fuel cell. The GPU operating availability was determined by
dividing the length of time the unit was available by the total operating time of the GPU. The emissions
characteristics of the GPU flare, which is used to combust the contaminants collected by the GPU, were measured
to evaluate hazardous air pollutants emitted into the atmosphere. The performance of the fuel cell was evaluated
by demonstrating the LFG to energy conversion process and by quantifying the power output. Total energy
conversion efficiency of the power generation equipment, fuel cell availability, and fuel cell exhaust emissions
were also measured.

The first verification test was executed at the Penrose site in Los Angeles, California. This test addressed
contaminant removal efficiency by the GPU, flare destruction efficiency, and the operational availability of the
cleanup system. The system was then relocated to the Groton Landfill in Connecticut where its performance was
verified under different operating conditions. Details of the verification may be found in the report titled Electric
Power Generation Using A Phosphoric Acid Fuel Cell On A Municipal Solid Waste Landfill Gas Stream (EPA-
600/R-98-105). The verification report may be ordered through the National Technical Information Service or
downloaded from the ETV Program or Center websites (www.epa.gov/etv or www.sri-rtp.comV

VERIFICATION OF PERFORMANCE

Performance Factors for the GPU:

•	Halide and Sulfur Removal Efficiency: The fuel cell requires total halogen and total sulfur levels to be <3
ppmv in the GPU outlet stream. At Penrose, the GPU exceeded the removal requirements of both
contaminants, with total halides reduced from 60 ppmv to [0.032 ppmv, and total sulfur reduced from 113
ppmv to [0.047 ppmv. The Groton performance results were similar, with total sulfur levels reduced to
[0.022 ppmv and total halides to [0.014 ppmv.

•	Estimated Flare Destruction Efficiency and CO/NO Concentrations: The destruction efficiencies of non-
methane organic compounds and volatile organic compounds were estimated to be 99 percent. The
conversion efficiency of sulfur compounds is also estimated to be about 99 percent. These efficiencies are
based on an estimation of flare gas exhaust flow because the measured flow rate was below the EPA
Method 2 detection limit. The NOx and CO concentrations at the flare outlet averaged 10.4 ppmv and 3.0
ppmv, respectively.

•	Operational Availability: The GPU logged 2,297 hours at Penrose and 4,168 hours at Groton (6,465
hours total). The GPU availability for the Penrose test was 87 percent. At Groton, the GPU availability
decreased to 45 percent because of leaks caused by relocating the test equipment from California to
Connecticut, and a malfunctioning gas compressor added at Groton to provide pressurized inlet gas. Once

	these mechanical failures were corrected, the GPU availability increased to 70 percent.	

EPA-VS-GHG-01

The accompanying notice is an integral part of this verification statement

iii

August 1998


-------
Performance Factors for the PC25™ Fuel Cell:

•	Electrical Output: At the Penrose site, a nominal output of 140 kW was expected to be generated from
the waste gas containing 44 percent methane (heating value of 446 BTU/scf). The test verified a
maximum output of 137 kW. The heating value of the Groton LFG was higher, 581 BTU/scf and 57
percent methane, resulting in higher power production from the fuel cell (165 kW).

• Energy Conversion Efficiency: The fuel cell system energy conversion efficiency, based on lower
heating values, was determined to be 37.1 percent at Penrose and 38.0 percent at Groton.

•	Operational Availability: The adjusted availability for the fuel cell, which compensates for shutdowns
not caused by the fuel cell, was over 96 percent at both test sites.

•	Stack Emissions: The emissions from the fuel cell exhaust are consistent with the data measured from
16 other PC25™ units operating on natural gas. The average emissions were measured as follows (dry
gas, corrected to 15 percent 02): NOx = 0.12 ppmv or 0.29 g/hr, S02 = non detectable (0.23 ppmv
detection limit) or <0.78 g/hr, and CO = 0.77 ppmv or 1.15 g/hr.

The results of these tests satisfy the requirements set forth in the testing plan for the GPU and
the fuel cell system. The GPU functioned according to its design specifications, purifying LFG to

a level which was more than suitable for fuel cell use. The fuel cell produced power with no
forced outages and provided consistently low secondary emissions. The electricity produced at
both sites were connected to a local grid system and sold to utility companies.

Although the PC25™ 200 kW fuel cell system has been used on natural gas, this verification was the first
application on LFG. This required the process design and engineering of a new GPU system to clean up the
contaminants not present in natural gas. The costs for the GPU were higher at Penrose ($2,450/kW), and lower at
Groton ($l,655/kW) due to reduced labor and start-up requirements. The cost for the PC25™ fuel cell was
$3,000/kW. The vendor estimates that, with system simplifications, dedicated production facilities, and other cost
reduction options, the GPU costs may eventually be reduced to $264/kW ($180/kW for equipment and material,
$84/kW for labor). Similarly, the vendor estimates that $l,500/kW may eventually be the mature phase cost of the
fuel cell. These cost estimates have not been independently verified.

E. Timothy Oppelt
Director

National Risk Management Research Laboratory
Office of Research and Development

Stephen Piccot
Director

Greenhouse Gas Technology Verification Center
Southern Research Institute

NOTICE: GHG Center verifications are based on an evaluation of technology performance under specific, predetermined criteria
and the appropriate quality assurance procedures. The EPA and Southern Research Institute make no expressed or implied
warranties as to the performance of the technology and do not certify that a technology will always, under circumstances other
than those tested, operate at the levels verified. The end user is solely responsible for complying with any and all applicable
Federal, State, and Local requirements.

EPA-VS-GHG-01

The accompanying notice is an integral part of this verification statement

iv

August 1998


-------
TABLE OF CONTENTS

Verification Statement	ii

List of Figures	vii

List of Tables	vii

Abstract	viii

Acknowledgments	ix

List of Abbreviations and Acronyms	x

Units and Conversion Factors	xi

Section No.	Page No.

1	EXECUTIVE SUMMARY

1.1	ETV Overview	1-1

1.2	Verification Objectives	1-1

1.3	Technology Description	1-2

1.4	Verification Approach	1-4

1.5	Verification Results and Performance Evaluation	1-6

1.5.1	GPU Performance	1-7

Contaminant Removal Performance	1-7

GPU Flare Emissions	1-7

GPU Availability	1-8

1.5.2	Fuel Cell Performance	1-8

Power Output and Efficiency	1-8

Fuel Cell Availability	1-10

Fuel Cell Exhaust Emissions	1-10

1.6	Data Quality Assessment	1-10

1.7	Technology Modifications to Achieve Full Power Production	1-12

1.8	Vendor Supplied Technology Costs	1-12

2	INTRODUCTION TO LFG AND FUEL CELL TECHNOLOGY

2.1	Methane Control Technology Challenge	2-1

2.2	Fuel Cell Power Plant Technology Description	2-2

2.2.1	Gas Pretreatment Unit Process Description	2-2

2.2.2	Fuel Cell System Process	2-4

2.3	Applicability of the Fuel Cell Technology	2-5

3	VERIFICATION TEST DESIGN AND DESCRIPTION

3.1	Background	3-1

3.2	Technology Performance Verification Objectives	3-1

3.3	Site Selection Criteria	3-2

3.4	Characteristics of the Candidate Sites and Selection	3-3

3.4.1	Initial Verification Test	3-3

3.4.2	Follow-On Verification Test	3-3

3.5	Description of Selected Sites	3-4

3.5.1	Penrose Site	3-4

3.5.2	Groton Landfill	3-4

3.5.3	Comparison of Gas Quality	3-4

V


-------
TABLE OF CONTENTS (cont.)

3.6	Site Specific Engineering Design and Layout	3-5

3.6.1	Penrose Landfill	3-5

3.6.2	Groton Landfill	3-5

3.7	Schedule	3-8

4	QUALITY ASSURANCE AND QUALITY CONTROL MEASURES

4.1	Objective and Road Map	4-1

4.2	Listing of Measurements Conducted	4-1

4.3	Sampling and Analytical Procedures	4-2

4.3.1	GPU Exit Gas Composition	4-2

4.3.2	GPU Flare and Fuel Cell Exhaust Emissions	4-3

4.3.3	Heat Content Measurements	4-5

4.3.4	Fuel Cell Power Output and Fuel Flow Rate	4-6

4.3.5	GPU and Fuel Cell Availability	4-6

4.4	Calibration Procedures	4-7

4.4.1	Manual Sampling Equipment	4-7

4.4.2	Fuel Cell and GPU Flare Continuous Monitoring	4-7

4.4.3	Other Equipment	4-7

4.5	Quality Control Checks, Audits, and Corrective Actions	4-8

4.6	Data Reduction, Validation, and Reporting	4-8

4.6.1	Calculations	4-8

Overall Calculations	4-8

Calculation of Data Quality Indicators	4-9

4.6.2	Data Validation	4-10

4.6.3	Identification and Treatment of Outliers	4-11

5	TECHNOLOGY RESULTS AND EVALUATION

5.1 Summary of Verification Results	5-1

5.1.1	GPU Contaminant Removal	5-1

5.1.2	GPU Flare Emissions	5-4

Flare Destruction of VOCs	5-6

Flare Destruction of Total Non-Methane Organics	5-7

Flare Outlet Concentration of NOx, C02, and Particulate Matter5-7
Condensate Analyses	5-7

5.1.3	GPU Operation and Availability	5-7

5.1.4	GPU Exit Heat Content	5-8

5.1.5	Fuel Cell Power Output and Efficiency	5-9

5.1.6	Fuel Cell Availability	5-10

5.1.7	Fuel Cell Emissions	5-11

6	REFERENCES	6-1

vi


-------
LIST OF FIGURES

Figure No.	Page No.

1-1	Landfill Gas Fuel Cell System	1-3

1-2	Verification Test Measurement Locations	1-5

2-1	Flow Diagram of the GPU and Fuel Cell System	2-3

2-2	GPU Process Operation	2-4

3-1	Penrose Landfill Test System Layout	3-6

3-2	Groton Landfill Test System Layout	3-7

4-1	Flow Diagram of GPU Flare Test Locations	4-4

5-1	GPU Exit Contaminant Concentration Vs. Time	5-3

LIST OF TABLES

Table No.	Page No.

1-1	Verification Performance Goals	1-2

1-2	Summary of Measurement Plans	1-5

1-3	Summary of Performance at the Penrose and Groton Landfills	1-6

1-4	GPU Exit Gas Heat Content Averages	1-9

1-5	Fuel Cell Efficiency Results	1-9

1-6	Fuel Cell Emissions Summary - Penrose Landfill	1-10

1-7	Summary of Quality Assurance Goals and Test Results	1-11

2-1	Potential for Power That Can Be Produced from Fuel Cells at U.S. Landfills	2-5

3-1	Verification Performance Goals	3-2

3-2	Landfill Gas Characteristics	3-3

4-1	Typical Concentrations, Detection Limits, and Blank Samples	4-3

4-2	GPU Flare Emission Test Target Compounds	4-5

5-1	Summary of Performance at the Penrose and Groton Landfills	5-2

5-2	GPU Exit Contaminant Levels - Penrose Landfill	5-3

5-3	GPU Exit Contaminant Levels - Groton Landfill	5-5

5-4	GPU Flare Emission Levels - Penrose Landfill	5-6

5-5	GPU Operation Summary During Fuel Cell Operating Periods	5-8

5-6	GPU Exit Heat Content Measurements - Penrose Landfill	5-9

5-7	GPU Exit Heat Content Measurements - Groton Landfill	5-9

5-8	Fuel Cell Power Output Results	5-9

5-9	Fuel Cell Efficiency Results	5-10

5-10	Summary of Fuel Cell Operation During Test Periods	5-10

5-11	Fuel Cell Emissions Summary - Penrose Landfill	5-11

vii


-------
ABSTRACT

The U.S. Environmental Protection Agency (EPA) and International Fuel Cells
Corporation conducted a test to verify the performance of a landfill gas treatment unit and a
phosphoric acid fuel cell system. The complete system removes contaminants from landfill gas
and produces electricity for on-site use or connection to an electric grid. The verification test
was the first use of fuel cell technology at a municipal solid waste facility. The test design was
subjected to extensive review and comment by the EPA's National Risk Management Research
Laboratory, landfill test site operators, and the technology developers.

Performance data were collected at two sites determined to be representative of the
U.S. landfill market. The Penrose facility, located in Los Angeles, California, was the first test
site. The landfill gas at this site represented waste gas recovered from four nearby landfills,
comprised primarily of industrial waste material. It produced approximately 3,000 standard
cubic feet (scf) of gas per minute, and had a higher heating value of 446 BTU/scf at about 44
percent methane concentration. The second test site, Groton Landfill, was located in Groton,
Connecticut. This was a relatively small landfill, but with greater heat content gas (methane
levels were about 57 percent and average heating value was 585 BTU/scf).

The verification test addressed contaminant removal efficiency, flare destruction
efficiency, and the operational capability of the clean-up system, and the power production
capability of the fuel cell system. The test verified that the clean-up system is capable of
reducing total halogen and total sulfur levels to less than 3 ppmv, which are the minimum levels,
required to operate the fuel cell. The GPU flare met emission requirements for sulfur, volatile
organic compounds, and other hazardous air pollutants. The GPU exceeded the minimum
operating requirement and logged more than 6,465 hours between the two sites. Based on the
landfill gas quality, it was expected that the PC25 fuel cell would produce a minimum of 140 kW
power. The power produced at Penrose was slightly below this value and peaked at 137 kW.
The Groton landfill produced a maximum power output of 165 kW due to higher BTU gas. The
overall fuel cell efficiency was determined to be 37.1 percent and 38.0 percent at Penrose and
Groton, respectively. Additional performance results determined for the fuel cell include
adjusted availability of over 96 percent at both sites, and low NOx, SO2, and CO emissions.

viii


-------
ACKNOWLEDGMENTS

The Greenhouse Gas Technology Verification Center wishes to thank Ronald J. Spiegel
of EPA's Air Pollution Prevention and Control Division for assisting in preparing this report. He
provided valuable information in accounting for the events and conclusions reached in the initial
study. Thanks are also extended to SCS Engineers, Inc. for providing guidance in representing
the landfill industry and for evaluating measurement results reported in this document. Finally,
the technical expertise provided by Larry Preston of International Fuel Cells, Inc. played a
critical role in evaluating the technology performance data.

ix


-------
LIST OF ABBREVIATIONS AND ACRONYMS

AC

Alternating Current

ASTM

American Society for Testing and Materials

BTU

British Thermal Units

DC

Direct Current

dscf

Dry Standard Cubic Feet

EPA

Environmental Protection Agency

ETV

Environmental Technology Verification

°F

Degrees Fahrenheit

g

Grams

GC/FPD

Gas Chromatography / Flare Photometric Detectors

GC/MS

Gas Chromatography / Mass Spectrometry

GHG

Greenhouse Gas

GPU

Gas Pretreatment Unit

h

Hours

IFC

International Fuel Cells Corporation

kW

Kilowatt

LFG

Landfill Gas

NIST

National Institute for Science and Technology

NMOC

Non-Methane Organic Compounds

NRMRL

National Risk Management Research Laboratory

NSPS

New Source Performance Standards

ORD

Office of Research and Development

ppmv

Parts per million, volume based

psi

Pounds Per Square Inch

QAPP

Quality Assurance Project Plan

RSD

Relative Standard Deviation

SCAQMD

South Coast Air Quality Management District

scf

Standard Cubic Feet

scfm

Standard Cubic Feet Per Minute

x


-------
UNITS AND CONVERSION FACTORS

Power

MW
kW

To Convert To

Multiply By

Megawatt
Kilowatt

Mass

Mg
Tg

Megagrams(10 grams)	pounds

Terragrams (106 grams)	pounds

2,205
2,204,600

Volume

SCMD
SL/M

Standard cubic meters per day
Standard liters per minute

scfd (standard cubic feet/day) 35.3
scfm (standard cubic feet/min) 0.0353

Pressure

Pa

Pascal

psi

1.45x10"'

Heating Value

kcal/SL	Kilocalories per standard liter

BTU/scf

112

XI


-------
SECTION 1

EXECUTIVE SUMMARY

1.1	ETV OVERVIEW

The U. S. Environmental Protection Agency's Office of Research and Development
(EPA-ORD) has created a program to facilitate the deployment of innovative technologies
through performance verification and information dissemination. The goal of the Environmental
Technology Verification (ETV) Program is to further environmental protection by substantially
accelerating the acceptance and use of improved and more cost-effective technologies. The
ETV program is specifically funded by the Congress in response to the belief that there are
viable environmental technologies which are not being used for the lack of credible third party
performance testing.

The Greenhouse Gas Technology Verification Center (the Center) is one of twelve
Centers currently operating under the ETV program. Together with EPA's partner verification
organization, Southern Research Institute, the Center provides verification testing capability to
GHG technology vendors, buyers, exporters, and others that have a need for performance data.
The Center develops test protocols, conducts field tests, collects and analyzes data, and reports
findings. Performance evaluations are conducted according to a rigorous verification plan and
established protocols for quality assurance to ensure objective and systematic evaluation of
innovative GHG technologies.

The Center is guided by a volunteer group of Stakeholders that have a stake in the GHG
emission reduction area. The group consists of national and international experts in the areas
of climate change science, policy, technology, and regulation. It also includes industry trade
organizations, technology vendors and buyers, environmental technology finance groups,
government research organizations, and government sponsored GHG mitigation outreach
programs. These groups help the Center develop strategic plans, establish credible technology
evaluation strategies, review verification results, distribute results widely, and coordinate with
other GHG programs and regulatory efforts.

This document summarizes the results of a verification test conducted on a technology
which generates electric power from waste landfill gas. The individual components of the
technology which were verified include: (1) a landfill Gas Pretreatment Unit (GPU) and (2) a
PC25™-200 kW phosphoric acid fuel cell system.

1.2	VERIFICATION OBJECTIVES

The objective of the verification test was to verify the performance of the GPU and the
PC25™-200 kW fuel cell system which is manufactured by ONSI Corporation, a subsidiary of
International Fuel Cell (IFC) Corporation in South Windsor, CT. The GPU performance was
evaluated by measuring sulfur and halogen concentrations in the GPU outlet gas stream and
then comparing these values with fuel cell operating specifications. The operational reliability of
the GPU was verified by logging the continuous and total duration of operation on landfill gas.
Emissions testing of the GPU flare was conducted to determine the contaminant levels of
hazardous air pollutants such as sulfur dioxide, carbon monoxide, and nitrogen oxides. Fuel
cell performance was evaluated by verifying the LFG to energy conversion process and by

1-1


-------
verifying its power output, energy efficiency, availability, and exhaust emissions. Table 1-1
presents the performance claims set out to be verified.

Table 1-1. Verification Performance Goals

GPU Parameter

Performance Goal

Exit Total Sulfur Concentration

Exit Total Halogen Concentration

Total Duration of Operation on LFG

Longest Continuous Run on LFG

Adjusted Availability

GPU Enclosed Flare Exhaust Emissions

<	3 ppmv

<	3 ppmv

>	500 hours

>	200 hours

No Initial Goal (to be determined from test)
No Initial Goal (to be determined from test)

Fuel Cell Parameter

Performance Goal

Maximum Power Output
Stable Power Output
Energy Efficiency
Duration of Operation
Adjusted Availability
Exhaust Emissions

140 kW or more

No Initial Goal (to be determined from test)
No Initial Goal (to be determined from test)
No Initial Goal (to be determined from test)
No Initial Goal (to be determined from test)
Equal to or less than those produced from
fuel cells operating on natural gas
S02 = negligible
NOx ~ 0.5 ppmv
CO ~ 1.1 ppmv

1.3 TECHNOLOGY DESCRIPTION

Municipal solid waste landfills are regulated to control air emissions (61 CFR 49). The
current standards correspond to emissions of non-methane organic compounds (NMOCs),
comprising some 100 volatile organic compounds and hazardous air pollutants contained in
landfill gas (LFG). These pollutants generally represent less than 1 percent of the total
composition of LFG which is primarily methane (35 to 60 percent) and carbon dioxide (40 to 55
percent), with a heat content equal to roughly one-half that of natural gas. Landfills emitting
greater than 50 metric tons per year of NMOCs are required to install a LFG collection system
and a treatment system capable of destroying 98 percent of the NMOCs in the gas or reducing
their concentration to less than 20 ppmv. In this process, the potent greenhouse gas methane
is also converted to carbon dioxide, or utilized to produce electricity or heat.

LFG collection is a mature technology and typically involves installation of vertical wells
into the landfill mass, using perforated plastic pipe. These pipes are usually connected to a
manifold, and a vacuum is applied for central collection and treatment and/or utilization. The
collected gas may be vented, flared, used to generate electricity and heat, or used to produce
pipeline quality gas. Electric power generation with fuel cells is one such method of utilizing
LFG (Roe etal. 1998).

The landfill gas to energy system design, offered by IFC, is based on providing a modular,
packaged, energy conversion fuel cell system which can operate on landfill gases with varied
compositions. The energy conversion system requires a LFG collection system, and consists of
a modular gas pretreatment system and a PC25™ natural gas fuel cell power plant modified for
LFG operation. The wells and collection system collect raw LFG and deliver it at low pressures

1-2


-------
to the gas pretreatment unit (GPU). Raw LFG is treated in the GPU to remove moisture,
particulates, and other contaminants to a level suitable for the fuel cell. The fuel cell power
plant converts cleaned LFG to electricity and useful heat. Figure 1-1 illustrates a simplified
diagram of this system, and the remaining paragraphs discuss its process description.

Gas wells and
collection system

Landfill

Figure 1-1. Landfill Gas Fuel Cell System

LFG exiting the GPU must contain very low concentrations of total sulfur and halogens to
properly operate the fuel cell. The GPU is designed to provide the contaminant removal
capability needed to meet fuel cell operating specifications. The system incorporates one non-
regenerable step, plus two stages of refrigeration combined with two regenerable adsorbent
steps. The first active bed of the GPU is a carbon adsorber designed to remove hydrogen
sulfide (H2S). A first-stage refrigeration condenser then removes most of the water contained in
the saturated LFG and some of the heavier hydrocarbon and contaminant species.

LFG exiting the first-stage refrigeration condenser is sent to a dryer bed where its water
content is reduced. This bed is regenerated every eight hours with heated, clean LFG. During
regeneration, a second, fully regenerated bed takes over the identical function. The
regeneration gas is subsequently incinerated in an enclosed flare. The LFG flare achieves
destruction of NMOCs by maintaining the combusted regeneration gas at a temperature of at
least 14000 F for at least one second.

Following the dryer bed, the LFG proceeds to a second stage low-temperature cooler to
enhance the performance of the downstream activated carbon bed. The activated carbon bed
adsorbs the remaining NMOCs (including organic sulfur and halogen compounds.) This bed is
regenerated every eight hours, and the regeneration gas is incinerated in the enclosed flare.
Output gas is filtered to reduce carryover of dust from the regenerable beds. The treated LFG
exits the filter for consumption in the fuel cell, and a small fraction of the gas is extracted and
used as the regeneration gas.

1-3


-------
In the fuel cell power plant, treated LFG is converted into electric power and heat. The
fuel cell converts the chemical bonding energy of a chemical substance directly into electricity.
The general PC25™-200 kW fuel cell system consists of three major subsystems: fuel
processing, direct current (DC) power generation in the fuel cell stack, and DC to alternating
current (AC) power conditioning by the inverter. The fuel cell extracts hydrogen from the clean
LFG, and electrochemically combines with oxygen from the air to produce DC electricity and by-
product water. The by-product water is recovered and used in the reformer. The heat
generated in the cell stack is removed to an external heat rejection system. The DC power is
converted to utility grade AC power in a power-conditioning package. The AC power can be
used on site or transformed and put into the utility grid.

1.4 VERIFICATION APPROACH

Two landfill test sites were identified where verification test results would be
representative of U.S. landfills. Factors for landfill site selection included LFG flow rates,
composition, heating value, and contaminant levels. The Penrose site, located in Los Angeles,
California, was the first location where the performance test was conducted. This site
represents gas collected from a total of four nearby landfills, which is comprised primarily of
industrial waste material. The LFG produced at Penrose is pressurized, thus the GPU was
manufactured to accept pressurized inlet gas. All power produced by the unit was fed into the
existing electrical grid for sale to the local electrical utility, the Los Angeles Department of Water
and Power. Approximately 3,000 standard cubic feet (scf) of LFG per minute is produced at
Penrose, with approximately 44 percent methane concentration and a higher heating value of
446 BTU/scf.

The entire GPU and fuel cell system was then relocated to the Groton landfill in
Connecticut to further verify the suitability of the energy conversion equipment under different
landfill site conditions. The LFG at Groton was not pressured, and since the GPU was built to
accept pressurized gas, a compressor was added to meet the inlet gas pressure requirements.
The gas flowrate at Groton was about 400 scfm, with methane concentration of 57 percent and
higher heating value of 585 BTU/scf.

The verification tests at Penrose and Groton occurred between September 1993 and
July 1997. The system was tested by IFC, TRC Environmental, Pacific Energy, and Northeast
Utilities, with QA and other oversight from EPA's Air Pollution Prevention and Control Division.

The verification test at the Penrose landfill consisted of the installation, operation, and
testing of the GPU and PC25™-200 kW fuel cell by IFC. The Penrose test began in September
1993 and was completed in February 1995. The first objective of the Penrose test was to verify
the performance of the GPU (i.e., reduce total halides and total sulfur levels to specified
amounts, and obtain emission measurements data for the GPU enclosed flare). Upon
verification of the GPU, the fuel cell was installed and performance data for the complete
system were obtained. Figure 1-2 illustrates the schematic of major measurement locations
where test data were collected. An EPA Category II Quality Assurance Project Plan (QAPP)
and a site-specific test plan were developed to address data quality requirements,
measurements, calculations, and audit requirements. Table 1-2 summarizes the key
measurement parameters and methods which were followed to obtain the verification data.

1-4


-------
Exhaust

Exhaust

Figure 1-2. Verification Test Measurement Locations

Table 1-2. Summary of Measurement Plans

Parameter

Method

Sample Location

Frequency





(see Figure 1-2)



Sulfur Compounds1

EPA 16 & 18

A2, A3, A4, B

Prior to start,







then monthly

Volatile Organic Compounds2

EPA TO-14

A2, A3, A4, B

Prior to start,

(including halides)





then monthly

GPU Flare Exhaust Emissions1





3 samples on

Total NMOC

CARB Method 25.2

A2, A4

Oct. 21, 1993

Particulate Matter

EPA 5 and 202

A2

(Penrose landfill

SO2, NOx, CO, and Diluents

EPA6C, 7E, 10, 3A

A2

only)

GPU Heat Content







Input Gas

On-Line Analyzer

A3

Monthly

Output Gas3

ASTM D-3588

B

Monthly

Cumulative Gas Flow Rate

Process Monitor

A2, A3, A4, B

Weekly

Fuel Cell Electrical Output

kWh Meter

C

Weekly

Fuel Cell Exhaust Emissions1







S02

EPA 6C

A1

6 samples on

NOx

EPA7E

A1

Feb. 17, 1995

CO

EPA 10

A1

(Penrose landfill

C02

EPA 3A

A1

only)

02

EPA 3A

A1



Cumulative Gas Flow Rate

Continuous Monitor

A1

Continuous

Availability, Maintenance Req., Operation Req.

Operator Log

N/A

Monthly

1 Source: 60 CFR 40







2 Source: Riggin 1988 and Winberry et at. 1988







3 Source: ASTM 1991







1-5


-------
Upon completion of the Penrose test, the GPU and fuel cell were relocated to the Groton
landfill to verify their performance under a different range of LFG composition. Northeast
Utilities conducted the follow-on testing, including the engineering design, construction,
installation, operation, and maintenance of the equipment. Quality assurance guidelines
prepared for the Penrose test were generally followed at this site (GPU flare and fuel cell
exhaust emission measurements were not conducted). The testing was performed between
June 1995 and July 1997. Additional details on the test, including data summary and discussion
of results, may be found in Trocciola and Preston, 1998a and 1998b.

1.5 VERIFICATION RESULTS AND PERFORMANCE EVALUATION

Table 1-3 summarizes the performance results. The following discussion highlights
each performance parameter.

Table 1-3. Summary of Performance at the Penrose and Groton





Landfills







Units

Goal

Penrose

Groton

GPU

Exit Total Sulfur (as H2S)

ppmv

<3

~0.047

~0.022

Exit Total Halides (as CI)

ppmv

<3

~0.032

~0.014

GPU Flare Emissions1









NOx

ppmv

No Initial Goal

7.5 to 14.9



CO

ppmv

No Initial Goal

1.6 to 5.8



NMOC

ppmv

No Initial Goal

6.8 to 11.7



Destruction Efficiency of Sulfur

%

No Initial Goal

>99

[A]

Compounds









Destruction Efficiency of VOCs

%

No Initial Goal

>99



Particulate Matter

grains/dscf

No Initial Goal

0.015



Total Duration of Operation On

hours

>500

2,297

4,168

LFG









Longest Continuous Run On LFG

hours

>200

342

827

Gross Availability

%

No Initial Goal

87.3

45 (total)









70 (last 6 months)

Fuel Cell

Maximum Power Output

kW

>140 kW

137

165

Stable Power Output

kW

No Initial Goal

120

140

Efficiency at Stable Output2

%

No Initial Goal

37.1

38.0

Total Duration of Operation on

Hours

No Initial Goal

707

3,313

LFG









Adjusted Availability

%

No Initial Goal

98.5

96.5

Exhaust Emissions









S02

ppmv

negligible

<0.23



NOx

ppmv

0.5

0.12

[A]

CO

ppmv

1.1

0.77



[rtJ Performance measurement was not required at this site.





1 Results represent emission measurements conducted during hot bed regeneration (worst case conditions) and

cold bed regeneration.









Represents the efficiency of the fuel cell only. Based on lower heating value measured per ASTM method.

1-6


-------
1.5.1 GPU Performance

Contaminant Removal Performance

At both test sites, the GPU consistently reduced contaminants in the LFG to levels
significantly below the initial goals of < 3 ppmv total sulfur and < 3 ppmv total halides.

At Penrose, the GPU reduced total halides from inlet levels of 45 to 65 ppmv in the raw
LFG to very low or undetectable levels at the outlet. At about 200 hours of operation, four
samples showed no detectable levels (0.002 ppmv detection limit), while one sample tested at
0.008 ppmv and one sample tested at 0.032 ppmv. At 2,235 hours of operation, six samples
showed no detectable halides. The halide removal performance of the GPU enabled IFC to
eliminate the addition of a halide guard bed in the fuel cell power plant.

The GPU met the performance goal of less than 3 ppmv total sulfur from an initial
concentration of 110 ppmv. The exit concentration ranged between non-detectable to 0.385
ppmv. The elevated level of 0.385 ppmv represents an atypical condition resulting from a
break-through occurring in the non-regenerable H2S removal bed. This breakthrough condition
causes carbonyl sulfide formation in the dryer beds. Some of the carbonyl sulfide passes
through the final carbon bed, resulting in increased sulfur at the GPU exit. After a fresh H2S
removal bed was installed, the exit total sulfur level returned to non-detectable levels. This
experience at Penrose established an operating procedure which requires switching the non-
regenerable H2S removal bed to a fresh bed when GPU exit total sulfur concentrations
increases rapidly.

At the Groton landfill, the GPU showed acceptable sulfur and halide compound removal.
Data from standard Summa canisters at normal GPU operating conditions showed no
detectable sulfur or halides at about 5,805 hours total GPU operation. The continuing low exit
levels of sulfur and halide compounds indicated that the original GPU bed design life of 8,000
hours is likely to be achieved.

GPU Flare Emissions

As part of the South Coast Air Quality Management District's (SCAQMD) permit
requirements, 18 permit conditions were required to construct and operate the GPU enclosed
flare. Details on these conditions may be found in Trocciola and Preston, 1998b. The GPU flare
met mechanical and operational permit requirements. Specifically, pollutant measurements
conducted on the GPU flare demonstrated that the destruction efficiency in the flame was
greater than 99 percent for both non-methane organics and sulfur compounds during the hot
regeneration cycle of the carbon bed. Flare emissions of nitrogen oxides (NOx) averaged about
0.056 lb/million BTU. This is below the 0.6 lb/million BTU NOx emissions currently required by
the SCAQMD. Total particulate matter, including back-half organic and inorganic fractions,
averaged 0.015 grains/dscf.

GPU Availability

The GPU operated for 2,297 hours at Penrose and 4,168 hours at Groton. The total
operating time of 6,465 hours well exceeded the minimum verification goal of 500 hours. The
longest continuous runs at the Penrose and Groton were 342 and 827 hours, respectively.

1-7


-------
The GPU gross availability was determined to be 87.3 percent for the Penrose landfill.
During this test, the GPU experienced a total of three shutdowns which resulted in lost time.
Reasons for these shutdowns include: the loss of coolant temperature control in the d-limonene
loop, loss of flare flame sensor, condenser tank overfill from high condensate influx at the site,
and lockup to the programmable logic (PLC) controller due to a control valve position switch
being out-of-limits. All causes were identified and corrected, and no outstanding operational
issues remained.

The gross availability of the GPU for the Groton test was determined to be 45 percent.
Most of the GPU shutdowns occurring during the first six months were one-of-a-kind mechanical
failures, and were attributed by leaks resulting from the equipment being moved from California
to Connecticut. All leaks and mechanical failures were corrected, and did not reoccur. The
primary cause of periodic failures was the added compressor which provided the needed
pressurized gas. It is estimated that a downtime of about 1,050 hours resulted from the
malfunctioning compressor valves. If these downtimes are removed from the GPU operating
hours, because the compressor is not part of the GPU system design, the gross availability
increases to 56 percent.

The remaining system issues diagnosed during the Groton test include: recurring high
GPU pressure drop (corrected by adding two new coalescing filters and water traps to prevent
LFG condensate from entering the small H2S removal bed), and periodic freeze-ups of the
refrigeration system (eliminated by adding an in-line dryer to the d-limonene refrigerant in
addition to the d-limonene air vent dryer installed at Penrose). After these improvements were
made, the GPU availability for the second half of the Groton test improved to 70 percent.

1.5.2 Fuel Cell Performance

Power Output and Efficiency

At Penrose, the fuel cell was operated at maximum power of 137 kW, which was 3 kW
below the goal for operation on LFG. The lower power output was due to less than expected
heating value of the inlet LFG stream. The power output at the Groton landfill improved to a
peak value of 165 kW. This was due to a 31 percent increase in higher heating value of the
Groton LFG.

Fuel cell efficiency was calculated by dividing the energy power output measured at the
fuel cell outlet with the GPU exit gas flow rate and lower heating value (LHV). The electric
power was measured using a utility grade meter calibrated by the Los Angeles Department of
Water and Power. The GPU exit gas flow rate was measured using a Yokogawa calibrated gas
flowmeter. The GPU exit gas heat content was determined by averaging the hourly on-line gas
chromatograph (GC) samples collected at the GPU exit for the time periods corresponding to
fuel cell efficiency measurement periods. These heat content measurements were based on
ASTM Method D-3588 (ASTM 1991).

The heat content measurement results, reported on a dry gas basis at 60 °F and 14.696
psia, are summarized in Table 1-4. The average higher heating value (HHV) at Penrose (based
on 7 measurements) was 445.8 Btu/scf versus 580.6 Btu/scf at Groton (based on 6
measurements). The most significant difference was the lower nitrogen content and higher
methane content in the Groton gas. Both sites contained very little higher hydrocarbons, with
Groton measuring none heavier than methane, and Penrose measuring about 0.02 percent
ethane.

1-8


-------
Table 1-4. GPU Exit Gas Heat Content Averages



Penrose

Groton

Average Composition (ASTM method D-1945)1

Nitrogen (%)

17.31

1.16

Carbon Dioxide (%)

37.88

41.21

Methane (%)

44.11

57.30

Ethane (%)

0.02

< 0.01

Propane (%)

nd2

< 0.01

Butane (%)

nd

< 0.01

Pentane (%)

nd

< 0.01

Hexanes (%)

nd

< 0.01

> Hexanes(%)

nd

< 0.01

Higher Heating Value (Btu/scf)

445.8

580.6

Lower Heating Value (Btu/scf)

401.3

522.8

1 Source: ASTM 1996





2 nd = non-detected





Fuel cell efficiency at Groton was calculated over a nine-day period while the fuel cell
was operating at a constant 140 kW. Efficiency during this period of continuous operation was
38.0 percent on a lower heating value basis (See Table 1-5). This calculated efficiency was
higher than the 36.5 percent efficiency calculated over a continuous six-day period at 120 kW at
the Penrose landfill.





Table 1-5.

Fuel Cell Efficiency Results







Period of
Steady
Power
Output

Steady Power
Output
Achieved
(kW)

Net Energy
Output Per
Electric Meter
(kW)

Gas Flow
Consumed
(ft3)

GPU Exit
LHV by
ASTM
Method
(BTU/scf)

Efficiency3

Penrose

1/24/95 to
1/30/95

120

16,800

3.92E+5

401.51

37.1%

Groton

6/10/97 to
6/19/97

140

28,682

4.87E+5

529.62

38.0%

1	Average of on-line analyzer measurements taken during steady power output.

2	Average of two measurements taken on 6/19/97.

3	Represents efficiency for the fuel cell only. Based on lower heating value.

Fuel Cell Availability

Total operating time for the fuel cell was determined to be 4,020 hours which included
707 hours at Penrose and 3,313 hours at Groton. During the Penrose test, the fuel cell was
operational at all times except a period of about 11 hours when the shutdown occurred due to a
failed sensor module. The longest Groton run lasted for 825 hours, with one forced outage
caused by the fuel cell.

1-9


-------
At the Penrose landfill, the adjusted availability was determined to be 98.5 percent,
which includes about 11 hours of down time caused by the fuel cell. The adjusted availability for
the Groton test was determined to be 96.5 percent, with total downtime of 119 hours.

Fuel Cell Exhaust Emissions

Fuel cell exhaust emission measurements were not performed at the Groton landfill.
The emissions for the Penrose site are summarized in Table 1-6. Also included are
comparative data taken from other fuel cell units operated on natural gas (Trocciola and
Preston, 1998a). Based on these emission levels, fuel cells can operate on LFG while
maintaining low emissions, as experienced in natural gas applications.

Table 1-6. Fuel Cell Emissions Summary - Penrose Landfill



Penrose Average

Average Based on 16 Units





Operating on Natural Gas

Nitrogen Oxides (ppmv)

0.12

0.46

Sulfur Dioxide (ppmv)

< 0.23

negligible

Carbon Monoxide (ppmv)

0.77

1.1

Oxygen (%)

7.9

Not available

Carbon Dioxide (%)

12.5

Not available

Note: dry measurements, corrected to 15% oxygen

1.6 DATA QUALITY ASSESSM ENT

A QAPP was developed for the Penrose test, and data quality indicators were
determined to address accuracy and precision of key performance parameters. A discussion of
sampling procedures, calibration procedures, analytical procedures, and other QA requirements
is provided in Section 4. A QAPP was not prepared for the Groton test. However, procedures
outlined at Penrose for data handling, sampling, and analyses were continued to maintain data
quality. Table 1-7 presents the quality assurance goals and test results for the GPU and the
fuel cell.

The quality assurance measurement for accuracy of hydrogen sulfide and three of the
four tested halogenated volatile organic compounds did not meet the 15 percent goal. Since the
GPU removed all of these selected compounds to below the detection limit in the GPU exit gas,
these errors are not significant to the conclusions regarding the overall effectiveness of the GPU
for sulfur and halide removal.

The accuracy of the ASTM D-3588 method for calculating GPU exit gas heat content did
not meet the 2 percent goal for all constituents, most notably methane at -3.5 percent. All other
hydrocarbon species were negligible in these tests. The impact of the 3.5 percent error could
be an overstatement of the apparent fuel cell efficiency by 3.5 percent (i.e., the reported 36.5
percent efficiency could be 35.3 percent). A comparative analysis between a series of
measurements conducted using an on-line analyzer in the GPU inlet stream and the ASTM
measurement results in the GPU exit stream revealed a close agreement (within 1 percent)
between the two methods. This indicates that the real error is probably less than 3.5 percent.

1-10


-------
The accuracy and precision goals and results for the fuel cell exhaust emissions are also
presented in Table 1-7. The quality assurance tests of the emissions monitors showed that the
S02, CO, C02, and 02 measurements generally met or exceeded the QA goals. The NOx
emissions tests met the stated QA goals for accuracy and precision.

Table 1-7. Summary of Quality Assurance Goals and Test Results



Precision

Accuracy



Measurement | Method

Goal | Results

Goal | Results

Effect On Data Conclusions

GPU

Sulfur ComDounds1
Hydrogen Sulfide

EPA 16 & 18

5%

0.6%

15%

30.7% 4

H2S was not detected at GPU exit
during GPU performance test, so
accuracy was not significant to
conclusion regarding GPU
effectiveness.

Volatile Oraanic
ComDounds2
Vinyl Chloride

Cis-1,2-dichlorethane

1,1 dichlorethane
tertrachloroethene

EPA -1014

15%
15%

15%
15%

19.0%
5.8%

6.9%
6.4%

15%
15%

15%
15%

54.5%
17.6%

13.2%
31.3%

These species were never detected at
GPU exit, so effect of not meeting
precision (vinyl chloride) or accuracy
goal (vinyl chloride, cis-1, 2-
dichloroethane, tetrachloroethene) was
not significant to conclusion regarding
GPU effectiveness.

GPU Input Gas
Heat Content3

On-Line
Analyzer

2%

N\A

2%

1.1%

Meets QA goal for accuracy

GPU Output Gas
Heat Content

ASTM
D-3588

2%

0.11%

2%

N2, CO2,
C3H8 within 2%
CH4 - 3.5%
C3H8, C4H10
C5H12>10%

Accuracy does not meet QA goals for
some species. Net effect on heat
content is possibly 3 to 4%.

FUEL CELL

S02 Emissions1

EPA-6C

5%

Zero drift =
-2.1 to +0.9%
Span drift =
-1.2 to+1.3%

5%

-4.0%

Meets QA Goals

NOx Emissions1

EPA-7E

10%

Zero drift =
-28 to +35.2%
Span drift =
-32 to +21.5%

15%

-22.4%
-20.7%

Low absolute NOx values make higher
uncertainty less significant

CO Emissions1

EPA-10

10%

Zero drift =
2.8 to+1.9%
Span drift =
-30 to +2.1%

10%

-5.4%

All but 1 span drift meets QA goals

C02 Emissions1

EPA-3A

5%



5%

1.3%

Meets QA goal

02 Emissions1

EPA-3A

5%



5%

0.8%

Meets QA goal

1	Source: 60 CFR 40

2	Source: Riggin 1988 and Winberry et al. 1988

3	Source: ASTM 1991

4	Determined from analysis of one hydrogen sulfide audit

1.7 TECHNOLOGY MODIFICATIONS TO ACHIEVE FULL POWER PRODUCTION

Based on the experience gained from the first application of fuel cells at landfills, the
vendor anticipates that the PC25™ power plant can be modified to achieve full rated power
when operating on LFG. The following paragraphs briefly discuss these recommendations.

The PC25™ power plant was designed to produced 200 kW of net power when
operating on natural gas having a higher heating value range of 976.6 kcal/SL to 1196.2 kcal/sL.
LFG with 50 percent methane and a higher heating value of 498.4 kcal/SL was projected to

1-11


-------
produce 140 kW of net power. To increase the net power, higher flows of LFG would be
required to obtain an equivalent natural gas fuel content and heating value.

One approach for achieving higher flows consists of increasing the fuel flows without
producing unacceptable pressure drops. The PC25™ fuel delivery train to the hydro-
desulfurizer contains two fuel isolation valves and a check valve having inside diameters (ID) of
2.54 cm and a fuel control valve having an ID of 1.27 cm. By increasing the ID of the isolation
valves and the check valve to 3.81 cm, the ID of the fuel control valve to 2.54 cm, and the
connecting plumbing to 3.81 cm, flow increases would be accommodated with acceptable
pressure drops. The manufacturer expects that these changes would not require major
modifications to the power plant.

The high flow rates of LFG would also increase the pressure drop across the fuel flow
fields within the cell stack assembly (CSA), and would result in a corresponding increase in the
fuel inlet operating pressure. This increased fuel side operating pressure affects the desired
pressure differentials between the fuel and air sides of a cell. This differential is known as
"reactant cross pressure". The increase in fuel inlet pressure by itself does not result in an
unacceptable increase in fuel delivery system pressure drops, but the corresponding increase in
reactant cross pressure exceeds established operating limits. One approach capable of
lowering the fuel inlet pressure consists of adjusting the flow directors in the CSA fuel manifolds
with a corresponding change in the size of the cathode exit flow orifice. This change would
produce flows sufficient to generate 175 kW of power. For flows to produce 200 kW, the size of
the cathode exit flow orifice would be reduced, and would result in the need for a larger cathode
air blower. This approach has a minimum impact on CSA production costs and CSA height.

1.8 VENDOR SUPPLIED TECHNOLOGY COSTS

The equipment verified at the Penrose and Groton landfills represents a single 200 kW
production capacity module. According to IFC Corporation, the installed cost for the GPU tested
was about $2,450/kW for the Penrose test (labor plus material cost equals $490,000) and
$1,655/kW for the Groton test. The cost for the PC25™ fuel cell was $3,000/kW. The operating
and maintenance costs are 0.40/kW-hr for the GPU and 1.50/kW-hr for the fuel cell. These cost
factors are based on estimates provided by the vendor, and were not independently verified by
the Center.

IFC Corporation projects that a full-scale implementation of the energy conversion
technology would incorporate at least four modules capable of generating 800 kW power. They
expect the installed cost to be significantly reduced for both the GPU and fuel cell in the future.
The GPU scale-up cost from a single 200 kW unit to four units may decrease to $264/kW with
system simplifications, dedicated production in a manufacturing facility, and other cost reduction
options. For the fuel cell, IFC estimates that a cost factor of $1,500/kW may be offered for the
mature technology as routine and production scale quantities are manufactured. These costs
have not been verified by the Center, and more accurate and current costs can be obtained
from the vendor.

SECTION 2

INTRODUCTION TO LFG AND FUEL CELL TECHNOLOGY

1-12


-------
2.1 METHANE CONTROL TECHNOLOGY CHALLENGE

Municipal solid waste landfills are one of the largest anthropogenic sources of methane in
the United States, contributing approximately 36 to 40 percent of total U.S. methane emissions
annually. The EPA estimates that in the absence of gas recovery and utilization projects, 10 to
14 million metric tons of methane will be emitted by landfills in the year 2000 (EPA 1993).
Methane, which is produced by the anaerobic decay of organic material contained in buried
solid waste, is a potent greenhouse gas with a global warming potential estimated at 21 times
that of carbon dioxide (IPCC 1995). In addition, the explosive nature of methane poses a
significant safety hazard from migrating LFG through the side slopes of landfills.

Methane is the major component of LFG produced in landfills. Typical LFG is composed of
35 to 60 percent methane and 40 to 55 percent carbon dioxide, with Non-Methane Organic
Compounds (NMOCs), nitrogen and sulfur compounds, oxygen, moisture, and other trace
compounds comprising the remaining 0 to 15 percent.

LFG generation is affected by a number of factors, including solid waste density, moisture
content, pH, and temperature. Consistent LFG generation is generally achieved within several
months of waste placement, and LFG production is a function of the amount of waste in place.
Dry landfills typically have a lower annual generation rate than wet landfills. EPA modeling
assumes that a typical solid waste landfill with a density of 44.44 lb/ft3 will produce 2.0 cubic feet
of methane per pound of compacted waste over time.

Until recently, surface landfill emissions were not regulated at the Federal level. New
Source Performance Standards (NSPS) and Emission Guideline regulations, promulgated in
March 1996, require solid waste landfills with greater than 2.5 million metric tons of waste in
place and emissions of over 50 metric tons of NMOCs annually to install LFG control systems.
Collected LFG is to be treated at a minimum 98 percent destruction efficiency of NMOCs. These
regulations, affecting those landfills that were open on or after November 8, 1987 (61 CFR 49,
9905, March 12, 1996), may drive the development of LFG flares and utilization projects
because LFG collection is mandated.

At a minimum, the method for mitigating methane and NMOC from LFG is the installation of
a gas collection and flare system. The fuel cell energy conversion system provides the
opportunity for converting methane in LFG to useful energy. There are two significant
differences between mitigation by flaring and mitigation with the fuel cell energy conversion
system. First, the fuel cell energy conversion system produces electric energy and thermal
energy which may be used to generate revenues from the LFG mitigation system. Secondly,
the fuel cell efficiently converts methane to electricity, has lower emissions at the site, and can
provide potential emission offsets due to the reduction in emissions from the electric utility which
would otherwise be providing the energy.

Fuel cells are efficient converters of chemical energy to electrical energy. Up to 80 percent
of energy available through the fuel supply can be converted into electrical and heat energy
because the conversion process does not have an intermediate conversion step (i.e., the
combustion step required by internal combustion engines). The efficiency of conventional fossil-
fueled power plants is dependent on the load at which they operate. At non-peak loads, the
efficiency of these commercial systems is significantly reduced. Fuel cells, on the other hand,
operate at a relatively constant efficiency under varying loads. Because fuel cell power plants
are modular, they can be pre-assembled and installed with relative ease. Based on natural gas

2-2


-------
applications, low emission levels of nitrogen oxides, carbon monoxide, and sulfur dioxide can be
achieved. In addition, low noise and vibration characteristics enable the units to be used in
areas where such conditions must be met. Since LFG contains contaminants which affect fuel
cell operation, a gas cleanup system is required to reduce impurities such as sulfur and halide
compounds.

2.2 FUEL CELL POWER PLANT TECHNOLOGY DESCRIPTION
2.2.1 Gas Pretreatment Unit Process Description

A simplified schematic of the GPU and fuel cell system is shown in Figure 2-1. The GPU
process consists of ambient temperature H2S removal followed by cooling, condensation,
drying, further cooling, hydrocarbon removal and final filtration. It is comprised on three primary
subsystems: Clean gas production process, Regeneration process, and Refrigeration process.

Clean Gas Production Process

The clean gas production process operates on raw LFG which is regulated down to
1.52x10s Pascal (Pa) from the Penrose plant compressor. This process incorporates H2S
removal, refrigerated cooling, and condensation to remove water, adsorption drying, cooling,
and hydrocarbon adsorption process units to remove contaminant from the LFG (see Figure 2-
2).

The H2S removal bed reacts H2S with 02 present in the LFG to produce elemental sulfur.
This bed contains 119 liters (43 cm diameter by 81 cm deep) of activated carbon impregnated
with potassium hydroxide, from Westates Carbon. It is not regenerated on-site, but it can be
regenerated off-site if desired, and is replaced periodically. The first stage cooler condenser
operates at approximately +2 °C and the 2nd stage cooler operates at -28 °C. The 1st stage
cooler removes water, some heavy hydrocarbons, and sulfides which are discharged as
condensate to the Penrose plant's existing gas condensate pretreatment system. Since the
GPU operates on a small slip stream from the Penrose site compressor and gas cooler, some of
the water and heavy hydrocarbons species are removed prior to the GPU. Most of the
contaminant halogen and sulfur species are lighter and remain in the LFG to be treated in the
GPU.

All remaining water in the LFG, as well as some sulfur and halogen compounds, are
removed in a regenerable dryer bed which has a capacity for adsorbing the remaining water
vapor in the LFG. The bed is 119 liters total volume (43 cm diameter by 81 cm deep) filled with
71 liters of Alcoa F200 alumina, followed by 48 liters of Davidson 3A mole sieve. There are two
dryer beds so that one is always operational while the other is being regenerated.

The dry LFG is then fed to the second stage cooler. This cooler can be operated as low
as -32 °C, and potentially can condense out heavy hydrocarbons if present at high levels. In
addition, the second stage cooler reduces the temperature of the carbon bed which enhances
its adsorption performance (Graham and Ramaratnam, 1993). The downstream hydrocarbon
adsorption unit whose temperature is controlled by the second stage cooler, is conservatively
sized to remove all heavy hydrocarbon, sulfur, and halogen contaminant species in LFG. The
unit consists of two beds, each containing 119 liters of activated carbon (Barneby and Sutcliffe,
type 209C). This way, one bed is always operational while the other bed is being regenerated.
Both the regenerable dryer bed and hydrocarbon removal beds operate on a nominal 16 hour
cycle with each set of beds operating in the adsorption mode for eight hours and regeneration

2-3


-------
mode for eight hours. The gas then passes through a particulate filter and is warmed indirectly
by an ambient air finned tube heat exchanger to insure a fuel inlet above 0 °C before being fed
into the fuel cell. The GPU process operating pressure is nominally 1.38x10s Pa with minimal
pressure loss across the equipment. A final regulator reduces the LFG pressure to the fuel cell,
which operates at 1x103 to 3.5x103 Pa inlet pressure.

FLARE
EXHAUST

CONDENSATE
RETURN TO SITE
CONDENSATE
TANK

REDUCES WATER REMOVES
& HYDROCARBONS WATER, SOME
SULFUR AND
HALIDES



REMOVES TRACE
HYDROCARBONS
SULFUR &
HALIDES

J

REMOVES HEATS GAS
PARTICULATES

REGENERATE
BEDS

NATURAL

GAS

FOR

START

BURNERS

-GAS PRETREATMENT SECTION-

¦ FUEL CELL SECTION—~

Figure 2-1 Flow Diagram of the GPU and Fuel Cell System
Regeneration Process

The regeneration process, shown in Figure 2-2, heats clean product LFG from the
production process and regenerates the dryer and hydrocarbon adsorption beds in the reverse
flow direction, and destroys the spent regenerant gas in an enclosed flare. An electric heater is
used to heat the recycled clean LFG to 288 °C. This heated, regeneration gas is used first to
regenerate the hydrocarbon adsorption bed, and then the dryer bed is regenerated. The
regeneration gas heater is then bypassed and the dryer bed is cooled down with cold
regeneration gas. Lastly, the hydrocarbon adsorption bed is cooled down. Each heating and
cooling period lasts about two hours for a total regeneration cycle of eight hours. At all times,
the regeneration gas flows to the enclosed flare ensuring continuous operation of the flare and
continuous thermal destruction of the contaminants and regeneration gas prior to atmospheric
dispersion.

2-4


-------
Raw LFG at 1.2xl05 Pa

L

Non-Regenerable H2S Removal Bed



1

32 °C





First Stage Cooler Condenser





1

r +2 °C



1st Stage Liquid Coalescing Separator

From LFG Production Process

I

Regenerable Dryer Beds (2)

. +4 °C to 10 °C

2nd Stage Refrigeration Cooler

^ -28 °C

2nd Stage Liquid Coalescing Separator
(Not Required)



Regeneration Gas Heater

288 °C (Hot Regeneration)
+10 °C (Cool Down)

Dryer Adsorption Beds (2)

Hydrocarbon Adsorption Beds (2)

Regenerable Hydrocarbon Adsorption Beds (2)

BedHeats Up To 232 °C

Particulate Filter

	

Vapor/Liqu

f.	

d Separator

i

f

Enclose

d Flare

Compressor



r

Finfan Condener





Liquid

Receiver



r

Filter/Dryer







M

r

Evaporator

d-limonene from
1st & 2nd stage
refrigeration coolers

To d-limonene
surge tank 1st &
2nd stage
refrigeration coolers

-26 °C

Ambient Air Finned
Tube Heat Exchanger

i Regeneration Process

1.24x10s Pa

Flare Exhaust

r +io°c

To Fuel Cell Supply Regulator

Clean Gas Production Process

Refrigeration Process

Regeneration Process
Figure 2-2. GPU Process Operation

Refrigeration Process Unit

The refrigeration process uses R-22 refrigerant in the cycle which provides refrigerated
d-limonene coolant at a nominal 2 °C to the first stage cooler and -28 °C to the second stage
refrigeration cooler. The d-limonene refrigerant is accepted as an environmentally benign
organic extracted from orange peels and pressed pulp. The refrigeration process incorporates a
double-stage compressor and plate-type evaporator. The refrigeration cycle operates to
maintain the d-limonene coolant temperature setting at its discharge from the evaporator. The
compressor is driven by a 7.5 kW motor drive and operates continuously to recirculate R-22
refrigerant in the refrigeration process. The two refrigerants R-22 and d-limonene coolants are
completely recycled and are not purged or vented from the process.

2.2.2 Fuel Cell System Process

The fuel cell system, normally used with natural gas, was slightly modified by IFC to
accept LFG. It consists of three major components as shown in Figure 2-1: the fuel processor,
the fuel cell stack, and a DC-to-AC inverter. The fuel processor converts methane from LFG
into hydrogen by the steam reforming process. It also removes low levels of oxygen and sulfur
in the LFG stream that may have remained following treatment in the GPU.

2-5


-------
Fuel cells produce power and heat by combining hydrogen and oxygen electro-
chemically. Natural gas or LFG is reformed within the power plant to provide the hydrogen fuel
while ordinary air is used as the oxygen source. The fuel cell stack consists of two electrodes
(cathode and anode), a phosphoric acid cell, and an external circuit for the conduction of
electricity. Input hydrogen gas is oxidized at the anode to produce hydrogen ions and electrons.
The electrons flow through an external circuit to the cathode. Hydrogen ions flow through the
electrolyte to the cathode and react with the introduced oxygen in the presence of electrons to
produce water and heat. The external direct current (DC) is then converted into 60-cycle
alternating current (AC) by the inverter. This power could then be conducted through a
transmission line via an interconnect point. Excess heat that is generated can be vented to the
air through a cooling module. The fuel cell exhaust maintains low levels of NOx, S02, CO, and
NMOCs, allowing the system to offer significant environmental benefits.

2.3 APPLICABILITY OF THE FUEL CELL TECHNOLOGY

To determine the potential power generation market available for fuel cell energy
recovery, the current population of municipal solid waste landfills in the U.S. was examined. It
was assumed that each fuel cell would consume 70 scfm of LFG with a heating value of 498
BTU/scf to generate 200 kW of power. The evaluation was based on using an EPA estimate of
methane emissions in the year 1997, and an estimate of LFG production rate of 3.08 liters per
Mg per year of refuse in place. As shown in Table 2-1, approximately 4,370 MW of power could
be generated from the 7,480 existing and closed sites. The largest number of potential sites
greater than 200 kW occurs in the 400 to 1,000 kW range, representing a market of 1,700 sites.

Table 2-1. Potential Power That Can Be Produced From
Fuel Cells at U.S. Landfills

(Based on Landfill Size and Electric Power Output)

Individual Site Power
Rating (kW)

Number Of Potential
Landfill Sites

Total Estimated Power
Output (MW)

<200

3700

220

201-400

1100

330

401-1000

1700

1010

1001-1500

380

480

1501-2000

220

380

2001-2500

90

190

2501-3000

60

160

> 3000

230

1600

TOTAL

7480

4370

Based on Table 2-1, it can be assumed that fuel cell applications at landfills will likely
require installations of more than one PC25-200 kW units (i.e., modular units can be installed to
produce greater power).

To identify which landfills are candidates for application of this technology, several
assumptions were made based on industry practice: (1) a typical U.S. landfill with 1 million tons
of municipal waste in place generates about 400 scfm LFG, and (2) landfills smaller than 1
million tons of waste in place are less likely to have installed extensive LFG collection systems
(due to depth and economic constraints). A query calculation in the EPA's Landfill Methane

2-6


-------
Outreach Program Database (Profiles Database) suggests that an estimated 820 landfills,
located in 30 states, are characterized by these assumptions (EPA 1997). This number is in
approximate agreement with PC25-200 kWfuel cell manufacturer's estimate of 980 landfill sites.
As a result, it is reasonable to assume that the fuel cell technology could be potentially applied
at about 800 - 1,000 sites.

Potential applications of the fuel cell technology should also be adjusted to account for
those locations where the technology is best suited from an economic and/or environmental
standpoint. To this end, certain localities and/or regions of the country may be preferred where
a developer can obtain a sales agreement for the generated power on the order of $0.08/kWh to
$0.10/kWh (or more). Currently, sales agreements of this kind are difficult to obtain, in part, due
to current pricing mechanisms for electrical power. However, favorable market conditions may
exist in such areas as California, New York, and parts of New England, and may expand further
as a result of deregulation.

Similarly, certain localities and/or regions of the country which are classified as non-
attainment for ozone (i.e., NOx and VOCs) and where offsets are expensive or difficult to obtain,
may be well suited for low/negligible emission technologies such as fuel cells. Generally, non-
attainment areas include most major population centers and are where larger landfills are sited
as well. Thus, a reasonable estimate of 70 percent of the applicable 1,000 landfills may be
located within the non-attainment areas, and are likely to be candidates for implementing this
technology.

For commercial operation, fuel cells will be required to produce full power on LFG (i.e.,
200kW). Based on Table 2-1, the most favorable market appears to be sites which produced a
total of 800 kW or 4 fuel cells. In addition, this market segment of landfills is unique because
large gas turbines or internal combustion engines are likely not to be used at landfills that have
the potential to produce greater than 1 MW power.

2-7


-------
SECTION 3

VERIFICATION TEST DESIGN AND DESCRIPTION

3.1	BACKGROUND

The verification test described in this report is based on an evaluation of a landfill gas-to-
energy test sponsored by the U.S. Environmental Protection Agency's Air Pollution Prevention
and Control Division, "Demonstration of Fuel Cells to Recover Energy from Landfill Gas"
(Trocciola and Preston, 1998a and 1998b; and Preston and Trocciola, 1998). It represents the
first application of a commerical PC25™-200 kW fuel cell at two municipal solid waste landfills.

Technology verification testing was first conducted at the Penrose site in Los Angeles,
CA from September 1993 to February 1995. The test was extended to the Groton landfill in
Groton, CT to further test the performance of the LFG-to-energy conversion equipment under
different conditions. This follow-on test started on June 1995, and was completed in July 1997.
The equipment is currently in operation at the Groton landfill.

This section presents the performance verification goals and site characteristics of the
Penrose and Groton landfills. Section 4 describes testing and quality control measurement
methods, sampling techniques, and calibration procedures followed to ensure data quality.
Section 5 presents results, and evaluates the performance of the GPU and the fuel cell system.

3.2	TECHNOLOGY PERFORMANCE VERIFICATION OBJECTIVES

Operational goals and objectives were established to confirm performance of the LFG-
to-energy conversion technology through verification testing (see Table 3-1). In brief, the GPU
verification objectives were to:

•	Remove LFG contaminants to levels required to operate the fuel cell (<3 ppmv of
total sulfur and halides);

•	Demonstrate up to 500 total hours and 200 hours of continuous operation; and

•	Measure emission levels and calculate destruction efficiencies of the GPU flare.

The fuel cell system performance goals included continuous operation to achieve 140 kW power
output, estimate fuel cell efficiency, availability, and to measure fuel cell exhaust emission
levels.

The verification tests were conducted under the direct management of EPA's Air
Pollution Prevention and Control Division. IFC manufactured the GPU and the fuel cell
equipment. The test team at Penrose landfill consisted of IFC, Pacific Energy, Southern
California Gas, Los Angeles Department of Water and Power, and the TRC Environmental
Corporation (TRC). IFC was responsible for coordination of participants, construction and start-
up activities, plant operations, measurements and monitoring, and record-keeping. Pacific
Energy provided the site, LFG supply and facilities, and staff to operate the GPU and monitor
and document LFG quality and quantity. TRC conducted the emission tests, collected and
analyzed GPU gas samples to estimate performance, and prepared the emission test report.

3-1


-------
Additional laboratory analyses were conducted by Performance Analytical, Inc. and Texas
Oiltech Laboratories, Inc.

Table 3-1. Verification Performance Goals

GPU Parameter

Performance Goal

Exit Total Sulfur Concentration

Exit Total Halogen Concentration

Total Duration of Operation on LFG

Lowest Continuous Run on LFG

Adjusted Availability

GPU Enclosed Flare Exhaust Emissions

<	3 ppmv

<	3 ppmv

>	500 hours

>	200 hours

No Initial Goal (to be determined from test)
No Initial Goal (to be determined from test)

Fuel Cell Parameter

Performance Goal

Maximum Power Output
Stable Power Output
Energy Efficiency
Duration of Operation
Adjusted Availability
Exhaust Emissions

140 kW or more

No Initial Goal (to be determined from test)
No Initial Goal (to be determined from test)
No Initial Goal (to be determined from test)
No Initial Goal (to be determined from test)
Equal to or less than those produced from
fuel cells operating on natural gas
S02 = negligible
NOx oo 0.5 ppmv
CO oo 1.1 ppmv

The test team at the Groton landfill consisted of IFC, TRC, Performance Analytical, Inc., and

Northeast Utilities operated the test equipment.

3.3 SITE SELECTION CRITERIA

The EPA and IFC sought candidate landfill sites to conduct the fuel cell verification.

Several criteria were used to guide the selection of the initial test site (Penrose):

•	The landfill site was to have an existing LFG collection system, stable gas flows
and composition, available LFG blower capacity, available flare capacity, and
excess LFG to operate the test equipment.

•	The site was to have available space/facilities suitable for the verification tests.
Ideally, the site(s) would have an available natural gas supply and contracts for
electricity sales.

•	Local codes and environmental regulations at the test site needed to be stringent
enough (i.e., the South Coast Air Quality Management District) so that a
successful verification would be readily accepted at most potential landfill sites in
the U.S.

The Groton landfill also satisfied these conditions while providing different LFG test
conditions, and a local utility agreed to host the verification test. As Table 3-2 shows, both test
sites are representative of typical U.S. landfills based on size, composition, LFG flow, and
moisture content.

3-2


-------
Table 3-2. Landfill Gas Characteristics



Units

Range For Most

Penrose

Groton





U.S. Landfills1





LFG Flow Conditions

Total LFG Produced at Site

scfm

70-5000

3000

400

LFG Higher Heating Value

BTU/scf

349 to 598

446

585

Moisture

%

1 to 10

dry

wet

LFG Composition (by volume)

Methane

%

35 to 58

44.0

57.0

Carbon Dioxide

%

40 to 55

38.0

41.0

Nitrogen

%

Oto 15

17.6

1.3

Oxygen

%

0 to 2.5

0.4

0.41

Total Halides (as CI)

ppmv

not available

45 to 65

7 to 45

Total Sulfur (as H2S)

ppmv

1 to 700

111

182

NMOCs (as methane)

ppmv

237 to 14,294

130 to 475

not available

1 Source: EPA 1991; and Augenstein and Pacey, 1992

3.4 CHARACTERISTICS OF THE CANDIDATE SITES AND SELECTION

3.4.1	Initial Verification Test

All candidate landfill sites considered were in California, since they were located in a
state which was expected to provide a significant portion of the total U.S. market for fuel cells.
Landfill sites initially recommended by Pacific Lighting Energy Services included Oxnard
Station, Penrose, Toyon Canyon, and Otay Station.

Oxnard and Otay landfills were eliminated from consideration primarily because they
were located at a greater distance from Los Angeles than other sites (and thus, not under the
purview of stringent regulations of the SCAQMD), and because natural gas service was not
available onsite (as required for GPU and fuel cell tests). The Toyon Canyon site was dropped
from consideration because of unstable LFG flows and composition, especially during the
summer months.

The Penrose site was selected because: (1) it was representative of typical U.S. landfills
based on size and LFG composition, (2) it was located within the SCAQMD region (3) an
existing LFG system was available and well maintained; (4) natural gas supplies were
accessible onsite; (5) contracts to sell electricity and purchase equipment were available; and
(6) there was sufficient space for a verification test equipment to be installed. Contracts for
electricity sale existed because of operation and subsequent power sales from IC engines
operating on LFG at the site.

3.4.2	Follow-On Verification Test

The purpose of the follow-on test of the fuel cell power plant was to test the consistency
of operation using LFG that differed from the initial test conditions. That is, LFG with higher
heating value than Penrose was expected to be tested and operated for a year. The project
was moved to the East Coast and narrowed down to a New England site since the Northeast
Utilities Company agreed to provide significant financial contribution to become the host utility
for the test. The Groton landfill site was chosen because it had recently been closed and

3-3


-------
capped, had an active LFG collection system, and was supported by Groton town officials. The
site also had access to a natural gas supply which was considered necessary for preliminary
tests.

3.5 DESCRIPTION OF SELECTED SITES

3.5.1	Penrose Site

The Penrose site is located approximately 15 miles northwest of downtown Los Angeles,
in Sun Valley, California. It is one mile west of the Golden State Freeway at Penrose Avenue.
The Penrose site is actually a power station where the LFG from four different non-hazardous
municipal waste landfills is brought, and processed to produce electricity with six reciprocating
compressors.

The refuse is filled over an area of 72 acres with an average depth of 200 feet. The
facility has been closed since 1983, and approximately 9 million tons of municipal solid waste is
in place. Typical LFG collection rates at Penrose are greater than 3 million standard cubic feet
per day. The collection system consists of 85 wells (55 are single pipe wells and 30 are double
pipe wells). The wells are interconnected with above and below ground header systems. The
LFG collected is conveyed to six 150-horsepower reciprocating compressors that compress the
LFG to about 90 to 100 psig before delivery to six 2 kW Cooper Superior IC engines. The site
typically generates 9.4 MW. The electricity is sold to the Southern California Edison Company
under a 20-year contract. Labor resources for the power plant include a three-man crew to
operate and maintain the facility.

3.5.2	Groton Landfill

The Groton landfill, located in Groton, Connecticut, is a 45-acre closed landfill, with
approximately 2 million tons of waste in place. Prior to the selection of the site for the fuel cell
verification test, LFG generated at the landfill was collected and burned in an open flare at an
approximate rate of 400 scfm. On the basis of this flow, it was estimated that 1 MW of electric
power could be generated on site. The fuel cell system currently uses a maximum flow of 80
scfm LFG; excess LFG is conveyed to the on-site flare.

3.5.3	Comparison of Gas Quality

LFG collected at Penrose landfill contained 44 percent methane compared with
approximately 57 percent methane measured from the Groton landfill. The nitrogen content in
Groton LFG was about 1.3 percent compared to a relatively high 17.6 percent nitrogen content
at Penrose. The nitrogen content of the Penrose site is higher than the range (0 to 15 percent)
expected from most U.S. landfills (with a typical value of 5 percent). Carbon dioxide and oxygen
contents were comparable at the two sites. A concentration range for total halides is not
available for U.S. landfills. However a typical value of 132 ppmv has been presented in
Trocciola and Preston 1998a. Due to the lack of variability expected in halide levels, it can not
be assessed whether the halide levels are representative of U.S. landfills. Additional
comparison of LFG quality and composition is detailed in Table 3-2.

Based on differences in the LFG methane content, the net energy recoverable from a
unit volume of Penrose LFG would be lower than that recoverable from a corresponding volume

3-4


-------
of Groton LFG, (i.e., the heating value of Penrose LFG was lower than the heating value of LFG
from the Groton landfill).

3.6 SITE-SPECIFIC ENGINEERING DESIGN AND LAYOUT

3.6.1	Penrose Landfill

The power plant layout for the Penrose landfill is shown in Figure 3-1. Located in the
northern portion of the landfill site, the gas pre-treatment skid and refrigeration unit were sited in
the middle of the test area. The fuel cell power plant and cooling module were located directly
south of the gas pre-treatment unit. The fuel cell pad was made of reinforced concrete,
approximately 3.35m by 7.62m, while the cooling pad was approximately 2.4m by 3.66m. The
typical area required for such a project is about half an acre.

The control panel for the GPU was located in a utility building to the north, outside the
testing area. To the right of this utility building was the flare for the gas pre-treatment system.
The entire test area was enclosed by a chain-link fence. Electricity generated at the Penrose
Landfill was sold to the Los Angeles Department of Water and Power.

3.6.2	Groton Landfill

The site layout for the fuel cell verification test at Groton is shown in Figure 3-2. The
surface area required was about 13 m by 41m and was enclosed by a chain link fence. Since
the potential for freezing was not a concern in Los Angeles, the GPU was not designed for cold
weather operation. However, to account for cold temperatures of the northeast, an enclosed
building was constructed to house the GPU at Groton. The housing was a pre-engineered, all-
weather building with aluminum siding and insulated walls and roof. The building was heated to
prevent potential freezing of LFG and LFG condensate. The area within the building was
classified as a Class 1, Division 2 location, and associated electrical equipment and fixtures
were built to be explosion proof.

The LFG moisture separator, hydrogen sulfide adsorber vessels, gas compressor, GPU,
and refrigeration unit were sited in the gas pretreatment unit building. The building was
equipped with a combustible gas detector to monitor the interior atmosphere and shut down the
gas compressor if methane gas was detected. The existing LFG flare was located in the
southern portion of the site along with an underground storage tank to collect condensate that
comes from the LFG and the GPU. The GPU control room housed the GPU control panel,
refrigeration unit purge air compressor, nitrogen bottles for activating the GPU pneumatic
valves, and project documentation. A GPU flare was used to combust the regeneration gas.

Start-up burner fuel for the fuel cell and for the GPU flare was stored on a compressed
natural gas bottle rack. The switchgear contained the distribution bus and breakers for the fuel
cell and other site equipment. The step-up transformer took the 480-volt power from the fuel cell
output and increased it to 13,800 volts for use on the utility grid. The equipment and site layout
were designed for unmanned operation. The Groton landfill project sold the electricity generated
to the Northeast Utilities system.

3-5


-------
Figure 3-1. Penrose Landfill Test System Layout

The major modification for this test was the addition of a continuous duty LFG
compressor to pressurize the gas to 40 psig, at a flow of 80 scfm. The GPU had been designed
for LFG under high pressure due to the requirements of the original IC engine-based electricity
generation system at Penrose. Since no such compression system existed at the Groton site, a
new unit was installed to compress the LFG to levels suitable for GPU functioning.

3-6


-------
Figure 3-2. Groton Landfill Test System Layout

The capacity of the hydrogen sulfide removal beds was increased relative to the
Penrose site, because initial data indicated significantly higher H2S concentrations (500 ppmv)
in Groton LFG. The last modification was the addition of compressed natural gas cylinders to
the site to serve as start-up fuel for the reformer burners. This was necessitated since the site
did not have a pre-existing natural gas supply.

3-7


-------
3.7 SCHEDULE

The test at the Penrose landfill was conducted between September 1993 and February
1995. The Groton landfill test was performed from June 1995 to July 1997. The GPU shake-
down archives were performed in the period of October 1993. During these start-up archives,
the thermal, mechanical, and electrical performance of the GPU was confirmed on nitrogen gas
at the IFC facility in Connecticut and then on LFG at Penrose landfill. The final tests involved
the documenting GPU flare performance, and conducting LFG contaminant removal
performance checks. The GPU was installed at Penrose landfill in April 1993 and tested for
operational performance between September 1993 and December 1993. The complete fuel cell
power plant was tested at Penrose landfill from December 7, 1994 to February 19, 1995.

The second test was conducted at Groton landfill beginning with initial set up in June
1995. After LFG quality tests and site preparations were conducted, the verification test was
initiated in July 1996 and monitored through July 1997.

3-8


-------
SECTION 4

QUALITY ASSURANCE AND QUALITY CONTROL MEASURES

4.1	OBJECTIVE AND ROAD MAP

The objective of Section 4 is to briefly present measurement methods and quality
assurance measures for the two verification tests. For the Penrose test, a Quality Assurance
Project Plan (QAPP) and Test Plan were developed to ensure that performance and emission
measurements were conducted by qualified individuals using proper equipment and written
procedures. The QAPP followed the guidelines presented in EPA's Quality Assurance
Handbook for Air Pollution Measurement Systems (EPA 1979) for generating accurate and
defensible data.

The plan at Groton was to continue the verification test that began at the Penrose site.
The operating power output was to be increased to 140 kW if possible. Gas analyses were
conducted per the quality assurance guidelines developed at Penrose to maintain a consistent
data set. With the exception of the GPU flare and fuel cell exhaust emissions, all
measurements conducted at Groton were identical to the data collected at Penrose. An EPA
Quality Assurance Officer observed the sampling techniques to ensure measurements were
conducting per QAPP guidelines.

Section 4.2 lists the measurements conducted, Section 4.3 discusses sampling
techniques, sampling frequency, and analytical procedures, Section 4.4 discusses calibration
techniques, and Section 4.4 discusses quality control checks, audits, and corrective actions.
The final section summarizes data reduction, validation, and reporting procedures. Additional
detail on the QA/QC requirements for each test may be found in Trocciola and Preston, 1998a
and 1998b; and Preston and Trocciola, 1998.

4.2	LISTING OF MEASUREMENTS CONDUCTED

The following parameters were measured to verify the performance of the GPU and the
fuel cell system at Penrose:

Performance Measurements (GPU and Fuel Cell)

GPU Exit Gas Composition and Removal Efficiency

Analysis for sulfur and target list VOCs (see Table 4-1) including halides

Fuel Cell Efficiency, determined from the following measurements:

GPU Inlet Heat Content (on-line method)

GPU Exit Heat Content (manual method)

GPU Output Gas Flowrate
Fuel Cell Electrical Output

GPU and Fuel Cell Raw Availability and Adjusted Availability

4-1


-------
Emission Measurements (Fuel Cell Exhaust and GPU Flare Exhaust)

Sulfur Dioxide (S02)

Nitrogen Oxides (NOx)

Carbon Monoxide (CO)

Carbon Dioxide (C02)

Oxygen (02)

Flowrate
Moisture

4.3 SAMPLING AND ANALYTICAL PROCEDURES

4.3.1 GPU Exit Gas Composition

Samples were collected from the GPU exit to verify the GPU's ability to remove LFG
contaminants. The sampling location was under 24 psig pressure which did not require
sampling pumps. The sampling port was comprised of a gate valve with a %" tube Swagelok
connector. The GPU exit and raw LFG sampling locations were in 11/4" pipes.

It was expected that breakthrough of organic compounds would most likely occur at the
end of an on-line cycle. Therefore, sampling was conducted at the end of the cycle to assess
GPU performance. Samples were collected during the last hour of an eight-hour GPU bed
"make" cycle (after seven hours of on-line operation; before regeneration commences at eight
hours).

Tedlar bag samples were collected twice per week from the GPU exit during the one-
month performance test period. The tedlar bags were collected over approximately five-minute
periods using a stainless steel valve to regulate the flowrate. Integrated samples were collected
and analyzed off-site by gas chromatography/mass spectrometry (GC/MS) and gas
chromatography/flame photometric detector (GC/FPD). The target compound list is shown in
Table 4-1.

The samples collected in tedlar bags were analyzed for seven sulfur compounds and
total reduced sulfur as hydrogen sulfide utilizing a GC/FPD according to the procedures outlined
in EPA Method 16 (60 CFR 40). An initial calibration curve with a minimum of three points was
established using calibration gas standards containing the analytes of concern. The calibration
curve spanned the expected concentration of the samples. The initial calibration was verified at
least once at the beginning of each 24-hour period with the analysis of a mid-level Continuing
Calibration standard. The percent difference of the continuing calibration response factors was
within ±15% from the initial calibration mean response factor. One field sample per analytical
sequence was analyzed in duplicate to demonstrate the precision of the analytical technique on
the sample matrix.

The samples collected in tedlar bags were also analyzed by GC/MS for VOCs. The
analyses were performed according to the methodology outlined in EPA Method TO-14 from the
Compendium of Methods for the Determination of Toxic Organic Compounds in Ambient Air
(Riggin, 1988). The method was modified for using Tedlar bags. The analyses were performed
by GC/MS utilizing a direct cryogenic trapping technique.

4-2


-------
Table 4-1. Typical Concentrations, Detection Limits, and Blank Samples for
Targeted Compounds in the GPU Exit



Typical Value in Untreated
Landfill Gas

Detection Limit
Objective

Blank
Samples

Sulfur Compounds (ppmv)

H2S

102.0

0.04

< 0.002

Methyl mercaptan

3.0

0.04

< 0.002

Ethyl mercaptan

0.5

0.04

< 0.002

Dimethyl sulfide

6.5

0.04

< 0.002

Dimethyl disulfide

< 0.07

0.02

< 0.002

Carbonyl sulfide

0.2

0.04

< 0.002

Carbon disulfide

< 0.07

0.02

< 0.002

Total sulfur as H2S

109.0

0.28

< 0.002

Volatile Organic Compounds (p

pmv)

Dichlorodifluoromethane

0.3-0.9

0.009

< 0.001

1,1 dichloroethane

1.2-2.9

0.002

< 0.001

Benzene

1.1-1.7

0.002

< 0.001

Chlorobenzene

0.6-1.4

0.002

< 0.001

Ethylbenzene

4.5-12.0

0.002

< 0.001

Methylene chloride

4.0-11.0

0.003

< 0.001

Styrene

0.5-1.1

0.003

< 0.001

Trichloroethene

1.3-2.4

0.001

< 0.001

T richlorofluoromethane

0-0.6

0.004

< 0.001

Toluene

28.0-47.0

0.002

< 0.001

Tetrachloroethene

2.4-4.8

0.002

< 0.001

Vinyl chloride

0.1-1.4

0.005

< 0.001

Xylene isomers

5.0-28.0

0.005

< 0.001

Cis-1,2-dichloroethene

3.9-5.9

0.003

< 0.001

Total halides as CI

47.0-67.0

0.086



4.3.2 GPU Flare and Fuel Cell Exhaust Emissions

The flare was used to control emissions from the GPU during bed regeneration periods.
The flare stack was a 32-inch-diameter refractory lined stack with two sampling ports located
90° apart - one diameter upstream from the outlet and approximately three diameters
downstream of the nearest flow disturbance. Flare tests at the GPU exhaust were conducted
with sampling times correlating to specific events in the bed regeneration cycles. Flare inlet and
outlet samples were collected during hot regeneration of the carbon bed and the dehydration
bed, and cold regeneration of the dehydration bed (see Figure 4-1).

Triplicate 60-minute test runs were conducted for each compound listed in Table 4-2 for
the flare inlet and outlet. These samples were collected during the first hour, the middle six
hours, and the final hour of bed operation. The flare inlet gas flow was measured with an in-line
process monitor which sends a signal to the control room chart recorder. The following
paragraphs discuss individual measurement techniques.

Inlet and Outlet VOC Emission Concentration - Determined from the triplicate one-hour
samples collected simultaneously at the inlet and outlet in Tedlar bags using the evacuated
canister technique according to EPA Method 18 (60 CFR 40). The samples were analyzed
by gas injections on a GC/MS according to EPA Method TO-14 (60 CFR 40). The samples

4-3


-------
FLARE
EXHAUST

Figure 4-1 Flow Diagram of GPU Flare Test Locations

(Shown With "B" Beds on Line and "A" Beds Being Regenerated)

were concentrated with a cryogenic trap prior to analysis. Each compound was quantified
by external calibration curves prepared from gas standards.

Inlet and Outlet Sulfur Compounds Concentration - The same flare inlet and outlet bag
samples collected for VOCs were also analyzed by GC/FPD for seven target compounds
(see Table 4-2). Samples were analyzed by gas injection on a Hewlet Packard 5890
GC/FPD with a 60 m by 0.53 mm ID capillary column (crossbonded 100 percent dimethyl
polysiloxane). These analyses were conducted off-site by an approved, independent testing
laboratory. A multilevel calibration was performed for each compound.

Outlet NOv, CO?, and O? Emission Concentrations - Triplicate one-hour tests were
conducted according to EPA Methods 7E, 10, and 3A (60 CFR 40). The reference method
analyzers were housed in a mobile CEM laboratory parked at the base of the stack. Sample
gas was transported to the system through 50 feet of heated Teflon sample line to a sample
gas conditioner in the laboratory. Calibrations were conducted with EPA Protocol I gases.

Outlet Particulate Emissions - Particulate emissions were measured according to EPA
Methods 5 and 202 at the flare outlet (60 CFR 40). Triplicate one-hour tests were
conducted using non-isokinetic sampling. Samples were collected non-isokinetically
because the gas velocity in the stack was below the detection limit of the pitot
tube/manometer and hot wire anemometer methods. Total particulate matter was
determined as "front half" which included material collected in the probe wash and filter, and
"back half" which included both inorganic and organic material collected in the impingers.

4-4


-------
Table 4-2. GPU Flare Emission Test Target Compounds

Flare Inlet

Flare Outlet

Methane

Total Non-Methane Organics
Hydrogen Sulfide
C1 through C3 Sulfur Compounds
Carbon Dioxide

Methane

Total Non-Methane Organics
Oxides of Nitrogen
Carbon Monoxide
Carbon Dioxide
Total Particulates

Toxic Air Contaminants (Including but not limited to)

Benzene

Chlorobenzene

1,2 Dichloroethane

Dichloromethane

Tetrachloroethylene

Tetrachloromethane

Toluene

1,1,1 Trichloroethane
Trichloroethylene
Trichloromethane
Vinyl Chloride
Xylene

Benzene

Chlorobenzene

1,2 Dichloroethane

Dichloromethane

Tetrachloroethylene

Tetrachloromethane

Toluene

1,1,1 Trichloroethane
Trichloroethylene
Trichloromethane
Vinyl Chloride
Xylene

Oxygen
Nitrogen

Oxygen
Nitrogen

Moisture Content

Moisture Content

Temperature

Temperature

Flowrate

Flowrate

Emissions from the fuel cell exhaust were measured over one day. S02, NOx, CO, C02,
and 02 and exhaust flowrate were monitored for six 1-hour periods on February 17, 1995.
Pollutant measurements were conducted according to EPA Methods 6C, 7E, 10, and 3A (60
CFR 40). Exhaust flow rate was also measured according to EPA Methods 1 and 2 (60 CFR
40).

4.3.3 Heat Content Measurements

For the Penrose test, heat content measurements were conducted at the GPU exit and
inlet. ASTM Method D-3588, which covers procedures for calculating heat content from
compositional analyses of gas samples, was followed to determine the heat content of GPU exit
gas (ASTM 1991). In addition, Pacific Energy operated a continuous fuel heat content analyzer
(gas chromatograph) on the raw landfill gas (a sample was analyzed every four minutes). The
continuous analyzer data were compared with the ASTM measurements to quantify the
difference in the fuel heat content and to allow using on-line measurements when exit heat
content data were not obtained.

For the ASTM method, samples were collected in steel canisters by purging the
canisters with at least 12 volumes of sample gas. Compositional analyses of the samples were
conducted using a gas chromatograph equipped with a thermal conductivity detector to
measure the concentrations of N2, 02, CH4, and C02, and a gas chromatograph equipped with a
flame ionization detector to measure the concentrations of C1 through C6 hydrocarbons. For
each gas chromatograph method, an initial calibration curve with a minimum of three points was
analyzed using calibration gas standards containing the analytes of concern. The calibration
curve spanned the expected concentration of the samples. The initial calibration was verified at
least once at the beginning of each 24-hour period with the analysis of a mid-level Continuing

4-5


-------
Calibration standard. The percent difference of the continuing calibration response factors was
targeted to be within ±15 percent from the initial calibration mean response factor. The heat
content of the samples was then calculated using the equations presented in ASTM Method D-
3588 from the measured chemical composition.

The on-line analyzer was automatically calibrated daily using a certified gas. The
calibration gas contained CO, 02, N2, and CH4. The data system recorded the response factor
of each compound, compared it to the certified reference, and reported a deviation.

Accuracy calculations for the ASTM method derived heat content values indicate that the
methane concentration was 3.5 percent lower than the certified value. Nitrogen, carbon dioxide,
and propane measured concentrations were within 2 percent of the certified values. The
remaining compounds (propane, butanes, and pentanes) had a variation greater than 10
percent. The results of this audit indicated that performance was less than QAPP
specifications. However, the net effect on heat content analyses was not significant. A
comparison study between the on-line Pacific Energy analyzer and ASTM method
measurements showed that the two methods were consistently within 2 percent.

4.3.4	Fuel Cell Power Output and Fuel Flow Rate

Fuel cell power output was measured continuously with a calibrated utility-grade
digital electric meter. The meter was a digital-display-type meter (Transdata EMS, Model PMG
30018-15) calibrated according to the American National Standard Code for Electricity Metering
(ANSI C12). It was installed at the outlet of the fuel cell, following DCDAC conversion
equipment, and thus reflects energy loses at the connection into the electric grid.

Fuel cell fuel inlet flowrate was measured continuously with a temperature and pressure
calibrated process monitor, a Yokogawa YFCT Flow Computing Totalizer (Style B). Calibration
of the gas meter installed on the GPU exit was performed by the manufacturer.

4.3.5	GPU and Fuel Cell Availability

Fuel cell availability was determined by compensating for the times when the unit was
not operating due to factors caused by other equipment or operating conditions. Factors which
account for adjusted availability include: unforced outages not due to the power plant,
shutdowns due to operator error, waiting time for replacement parts where parts were
recommended the customer should have on hand, and periods of time when power plant could
be worked but manpower was not available such as weekends or vacations. Adjusted
availability was not calculated for the GPU. Instead, gross availability, without compensation for
downtimes was reported.

4.4 CALIBRATION PROCEDURES

4.4.1 Manual Sampling Equipment

The sampling and measurement equipment, including continuous analyzers, recorders,
pitot tubes, dry-gas meters, orifice meters, thermocouples, probes, nozzles, and any other
pertinent apparatus was uniquely identified, underwent preventive maintenance, and was
calibrated before and after each field measurement, following written procedures and
acceptance criteria. Most calibrations were performed with standards traceable to the National
Institute for Science and Technology (NIST). These standards include wet test meters, standard

4-6


-------
pitot tubes, and NIST Standard Reference Materials. Records of all calibration data were
maintained in files.

4.4.2	Fuel Cell and GPU Flare Continuous Monitoring

The continuous measurement analyzers were calibrated before and after each test for
zero and span drift according to EPA Methods 6C, 7E, 10, and 3A (60 CFR 40). EPA Protocol I
gases were used. The calibration gas was introduced to the system at the probe outlet using a
three-way tee. An excess flow of calibration gas was metered to the tee with the excess flowing
into the stack through the probe. A calibration error test was also conducted once by initially
conducting a zero and span calibration, followed by introducing a zero, high and mid point
calibration gas to the system.

The exhaust emission monitors were calibrated before and after each one-hour test with
EPA Protocol I gases and the drift performance specifications were within the method
specifications for each parameter except for NOx (the NOx analyzer was operated at the 0-2.5
ppm range which was too low to meet the method drift specification).

In the analysis of VOCs, verification of the GC/MS was checked at the beginning of
every 24-hour analytical sequence by the direct injection of 50 nanograms (ng) of
bromofluorobenzene. The calibration range of the target compounds was determined by a
three-point curve. Linearity was established over the range of the three-point curve if the
percent relative standard deviation of the response factors was less than 30% for each analyte.
A continuing calibration was considered to establish the same conditions of linearity and range
as the initial calibration if the response factor for each analyte was within 20% of the average
response factor for the initial calibration. A continuing calibration was performed at the
beginning of each 24-hour period. A blank was analyzed following calibration as a sample to
demonstrate that the analytical system was free from contamination.

4.4.3	Other Equipment

The calibration of the gas meter installed on the GPU Exit was performed by the
manufacturer. The calibration for the electrical power measurements were conducted by the
Los Angeles Department of Water and Power. The calibration reports for these equipment and
the on-line heat content analyzer are provided in Trocciola and Preston, 1998b.

4.5 QUALITY CONTROL CHECKS, AUDITS, AND CORRECTIVE ACTIONS

Continuous emission monitoring quality control checks included zero and span drift
tests, calibration error tests, system bias checks, and audits. All continuous monitoring zero
and span gases were delivered to the probe outlet to challenge the entire sampling system.
The QC data was recorded on a data logger chart. A brief discussion of the quality control
checks is provided below:

Blanks for both sulfur and VOC analyses were conducted with each set of samples
delivered to the analytical laboratory. The blank concentration of target sulfur compounds was
less than 2 ppbv and the blank concentration of target VOCs was less than 1 ppbv.

Audit samples for this verification were purchased for target volatile compounds, sulfur
compounds, and heat content analysis. The audits were used to determine the accuracy of the
results determined from the tests.

4-7


-------
Instrument calibration verifications for GC and GC/MS were performed for target volatile
compounds, sulfur compounds, and heat content analysis.

Laboratory duplicates were performed for each analytical parameter for each analytical
sequence. The percent difference determined was used to evaluate matrix effect on the
precision of the analytical technique. The precision objective for laboratory duplicates is 10%
relative percent difference.

4.6 DATA REDUCTION, VALIDATION, AND REPORTING
4.6.1 Calculations
Overall Calculations:

Pollutant Mass Emission Rate (SO?. NCX. and CO)

= C x F x MW x 0.0025

hr

where:

C = Concentration, ppmvd
F = Flowrate, dscm/m

MW = Molecular Weight (S02 = 64, NOx = 46, CO = 28)

GPU Performance (total sulfur and halides) - The performance limit was 3.0 ppmv of
total sulfur and 3.0 ppmv of total halides. Total sulfur was computed by summing the
products of each sulfur species times the number of sulfur atoms per mole. The results
were plotted vs. operating hours. Total halides were computed by summing the
products of each halide species times the number of halide atoms per mole of species
(e.g., CCI4 = 4). The results were plotted vs. operating hours.

Fuel Cell and Flare Emissions

7-	Mass Emission RateClb/hr)

Emissions(lb / kWh) =	

Power Output (kWh)

Concentration and flowrate measurements were used to calculate a mass emission rate
of NOx, S02, CO, and C02 from the flare stack and power plant. Emissions from each
source were summed and converted to mass emissions per energy output as follows:

Fuel Cell Efficiency (reference Figure 4-1 for measurement locations)

kWh at fd x 3413 BTU/kWh

Efficiency 		 r 	

scf at[B]xBTU/scf

4-8


-------
where:

scf = measured GPU exit gas at position [B]; based on flow, temperature, and
pressure

BTU/scf = hourly average heat content measured with Pacific Energy's on-line
analyzer and a correction factor (1.01) developed from a comparison of six GPU
exit ASTM measurements to six hourly averages from the Pacific Energy
analyzer

Fuel Cell Availability (adjusted to compensate for outages which are not caused by the
power plant system)

a i i a iii	Operating Hours

Adjusted Availability =	

(Elapsed Time - Adjustment)

Calculation of Data Quality Indicators

Precision: Continuous Emission Monitoring - (determined before and after each test
period using a zero and span calibration drift test). The drift was calculated as a
percentage of instrument range as follows:

n/ , .„ monitor value - certified concentration ,

% drift =	x 100

span value

Precision: Sulfur and Halide Compounds in the GPU Exit samples - (calculated for each
detectable compound from a series of three samples collected simultaneously):

RSD = —
x

where:

RSD = relative standard deviation
s = standard deviation
x = mean value

Precision: GPU Exit Gas Heat Content Analysis - (calculated from the RSD of a series
of three replicate samples per above defined equation).

Accuracy: Continuous Emission Monitoring - (determined by analyzing audit gases for
each parameter). The audit cylinders were EPA Protocol I (±1%) or equivalent.
Accuracy was calculated as follows:

Accuracy =	x 100

Ca

4-9


-------
where:

Cm = monitor response

Ca = certified audit concentration

Accuracy: Sulfur and Halide Compounds - Audit samples were prepared gravimetrically
by a specialty gas manufacturer and certified for ±5% accuracy. The audits were
analyzed with the first set of samples submitted to the laboratory. The sulfur audit gases
contained hydrogen sulfide and the halide audit gases contained six target compounds.
Accuracy was determined as previously described for continuous monitoring.

Accuracy: GPU Exit Heat Content Analysis - One BTU audit cylinder gas audit was
purchased from a specialty gas manufacturer and analyzed with the heat content
samples by the ASTM method.

4.6.2	Data Validation

Each one-hour period of continuous emission data was reduced on a separate Lotus
worksheet file. Copies of the raw data logger charts and the worksheet printouts are provided in
Trocciola and Preston, 1998b. Laboratory data were submitted to the designated laboratory
personnel for QA evaluation. The QA specialist examined the data, checked the precision and
accuracy of the results (duplicate analyses and audits), and reported the findings.

4.6.3	Identification and Treatment of Outliers

Continuously monitored parameters did not change significantly throughout the program.
Responses for CEM monitors and Pacific Energy process monitors were evaluated during the
emissions testing and no unusual activities were observed. Similarly, the analytical values for
halide and sulfur compounds concentrations of the GPU exit gas were constant over the course
of the program.

The heat content of the GPU exit sample collected on February 9 was unusually low and
was considered to be caused by a sampling error. It was expected that the sampling bulb was
not completely purged with sample gas.

4-10


-------
SECTION 5

TECHNOLOGY RESULTS AND EVALUATION
5.1 SUMMARY OF VERIFICATION RESULTS

Table 5-1 presents performance measurements data obtained at the Penrose and Groton
landfills. The GPU exceeded the stated performance goals. Total halogen and total sulfur
levels were well below the minimum requirement. The GPU flare met all mechanical and
operating requirements of the SCAQMD, and its performance was determined to be in
compliance.

The GPU exceeded the performance goal of 200 hours of continuous operation, with a
total of 6,465 hours logged at the two landfills. The longest continuous runs of 342 hours and
827 were recorded at Penrose and Groton, respectively. The gross availability at the Penrose
site was about 87 percent. At Groton, several shutdowns of the GPU resulted in overall gross
availability of 45 percent; about half of these shutdowns occurred in the first six months of the
testing period, and were related to one-of-a kind mechanical failures of the GPU equipment
parts which were corrected and did not reoccur because they were related to moving the
equipment from California to Connecticut. Significant downtime was also caused by the
compressor which was added to increase the LFG pressure as required for the GPU. Several of
these failures required considerable trouble-shooting time to determine the root cause and to
design a suitable solution to the problem, and then procure and install the appropriate fix. It is
estimated that over one thousand hours were lost due to failed compressor valves. Once the
mechanical failures were resolved, the GPU gross availability for the second half of the Groton
test period improved to 70 percent.

The fuel cell met all performance goals, with the exception that the maximum power
generated at Penrose was 3 kW less than expected. This was primarily because the LFG
heating value was less than expected (methane composition was about 44 percent). At Groton,
the fuel cell achieved higher power (165 kW) because of higher heating value gas (methane
concentration was over 57 percent). The fuel cell demonstrated the LFG-to-energy conversion
concept with energy efficiency values between 36.5 and 38.0 percent (based on lower heating
values), adjusted availability of over 96 percent, and over 4,000 operating hours logged. The
fuel cell exhaust emissions were consistent with those reported for natural gas applications.
The remaining discussion in this section focuses on the measurement and evaluation of each
verified parameter.

5.1.1 GPU Contaminant Removal

The GPU removed contaminants in the LFG by a factor of about 64 times below the 3
ppmv performance goal for total sulfur, and by a factor of about 94 times below the 3 ppmv
performance goal for total halides.

Figure 5-1 illustrates time series results for samples collected throughout the GPU
operation. The concentration data represent measurement results obtained from tedlar bag
samples collected at the GPU exit. With the exception of pre-test check out testing,
simultaneous samples were not collected at the GPU inlet (i.e., during actual Penrose and
Groton testing). Therefore, average concentration measurements produced from pre-test
sampling were used to estimate removal efficiency.

5-1


-------
Table 5-1. Summary of Performance at the Penrose and Groton Landfills

| Units Goal Penrose | Groton

GPU

Exit Total Sulfur (as H2S)

ppmv

<3

~0.047

~0.022

Exit Total Halides (as CI)

ppmv

<3

~0.032

~0.014

GPU Flare Emissions'

NOx
CO

NMOC

Destruction Efficiency of Sulfur

Compounds
Destruction Efficiency of VOCs
Particulate Matter

ppmv
ppmv
ppmv
%

%

grains/dscf

No Initial Goal
No Initial Goal
No Initial Goal
No Initial Goal

No Initial Goal
No Initial Goal

7.5	to 14.9

1.6	to 5.8
6.8 to 11.7

>99

>99
0.013

[A]

Total Duration of Operation On
LFG

hours

>500

2,297

4,168

Lowest Continuous Run On LFG

hours

>200

342

827

Adjusted Availability

%

No Initial Goal

87.3

45 (total)
70 (last 6 months)

Fuel Cell

Maximum Power Output

kW

>140 kW

137

165

Stable Power Output

kW

No Initial Goal

120

140

Efficiency at Stable Output^

%

No Initial Goal

36.5

38.0

Total Duration of Operation on
LFG

hours

No Initial Goal

707

3,313

Adjusted Availability

%

No Initial Goal

98.5

96.5

Exhaust Emissions

so2

NOx
CO

ppmv
ppmv
ppmv

negligible
~0.5
~ 1.1

<0.23
0.12
0.77

[A]

[A] Performance measurement was not required at this site.

Results represent emission measurements conducted during hot bed regeneration (worst case conditions) and
cold bed regeneration.

2 Calculated at lower heating values.

During the GPU pre-start check out testing at Penrose, four samples showed no
detectable halides while one sample detected 0.008 ppmv and another detected 0.032 ppmv.
These outlet concentrations reflect greater than 99 percent removal efficiency (based on
simultaneous measurement of 60 ppmv total halogens at the GPU inlet). The removal efficiency
for total sulfur was also greater than 99 percent (based on measured inlet concentration of 113
ppmv). Total NMOC (as methane) showed a reduction from 5,700 ppmv at the inlet to 13.8
ppmv at the GPU outlet, for an overall removal efficiency of 99.8 percent.

During the Penrose test, the sample at 1,685 hours detected 0.009 ppmv total halides.
All six remaining samples collected through 2,235 hours of operation showed no detectable

5-2


-------
2.5
2
1.5
1

0.5
0

Maximum Allowable Total Sulfur and Total Halides <3 ppmv

r'enrose Testi

pre-start check
out & GPU flare
testing ^

i



Groton Test

~¦switched to fresh H2S removal bed

-o—o—&

0	1000	2000	3000	4000	5000	6000	7000

Total GPU Operating Hours

I Total Sulfur ~ Total Halides

Figure 5-1. GPU Exit Contaminant Concentration Vs. Time

halides. Because of this degree of halide removal, IFC was able to eliminate a halide guard bed
initially planned for integration into the PC25™ power plant. A direct calculation of removal
efficiency could not be conducted because simultaneous concentration measurements were not
taken at the GPU inlet. However, if it is assumed that the GPU inlet total halide concentration
was about 60 ppmv, as measured in the initial pre-test analysis, the removal efficiency is greater
than 99.98 percent. Table 5-2 presents the data for individual samples collected at Penrose.

Table 5-2. GPU Exit Contaminant Levels - Penrose Landfill







December 1994 - February 1995

GPU Operating Time (Hours)

1685

1701

1710

1826

2046

2069

2235

Hydrogen Sulfide

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

Methyl Mercaptan

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

Ethyl Mercaptan

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

Dimethyl Sulfide

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

Dimethyl Disulfide

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

Carbonyl Sulfide

<0.004

<0.002

0.071

0.077

0.173

0.385

0.061

Carbon Disulfide

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

Total Sulfur (as H2S)

nd

nd

0.071

0.077

0.173

0.385

0.061

Dichlorodifluoromethane

<0.02

<0.02

<0.001

<0.001

<0.02

<0.02

<0.02

1,1-dichloroethane

<0.001

<0.001

<0.001

<0.001

<0.001

<0.0012

<0.001

Benzene

0.001

<0.002

<0.002

<0.002

<0.002

<0.0016

<0.002

Chlorobenzene

<0.001

<0.001

<0.001

<0.001

<0.001

<0.0011

<0.001

Ethyl Benzene

<0.001

<0.001

<0.001

<0.001

<0.001

<0.0012

<0.001

Methylene Chloride

0.005

<0.002

<0.002

<0.002

<0.002

<0.0015

<0.002

Styrene

<0.001

<0.001

<0.001

<0.001

<0.001

<0.0012

<0.001

Trichloroethene

<0.001

<0.001

<0.001

<0.001

<0.001

<0.0009

<0.001

Toluene

<0.002

0.003

0.002

0.001

0.004

0.0041

0.002

T etrach loroeth ene

<0.001

<0.001

<0.001

<0.001

<0.001

<0.007

<0.001

Vinyl Chloride

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

Xylene Isomers

0.001

0.003

0.001

<0.001

<0.002

0.0042

0.004

cis 1-2-dichloroethene

<0.001

<0.001

<0.001

<0.001

<0.001

<0.0013

<0.001

Total Halides (as CI)

0.009

nd

nd

nd

nd

nd

nd

nd - non-detected.

All GPU exit samples were collected during the last hour before regeneration.

5-3


-------
Total sulfur concentrations in the GPU exit ranged between non-detectable to 0.385
ppmv. The elevated level of 0.385 ppmv represents an atypical condition resulting from a break-
through occurring in the non-regenerable H2S removal bed. The increased concentration
results from carbonyl sulfide being formed in the dryer bed according to the following chemical
reaction: H2S + C02 = COS + H20. The low temperature carbon bed does not remove carbonyl
sulfide. As shown in Figure 5-1, the exit H2S levels returned to non-detectable after a fresh non-
regenerable H2S bed was installed,. This experience at Penrose established an operating
procedure which required switching the non-regenerable H2S removal bed to a fresh bed when
GPU exit total sulfur concentrations increases. Carbonyl sulfide levels measured shortly after
this switch decreased from 0.385 ppmv to 0.061 ppmv. In conclusion, the GPU consistently
maintained total sulfur levels well below the goal under normal operating conditions.

Based on the experience with elevated total sulfur concentrations, the useable life of the
H2S removal bed was estimated at 21 days. This yields an apparent capacity of 0.12 grams of
sulfur per gram of carbon in the bed. The 119 liter bed volume of the two beds used in the test
was too small. Future installations would incorporate larger tanks designed for low cost, ease of
servicing, and longer changeout times.

At Groton, GPU exit gas samples were taken from both parallel sets of beds in the GPU
(dryer bed and carbon bed). Total halide results also showed no detectable levels for most
samples, with the exception of those collected in the last month (see Table 5-3). The
concentration for these samples ranged from 0.012 ppmv to 0.019 ppmv because of a change
in GPU operating conditions. Similar to the Penrose test, simultaneous GPU inlet gas samples
were not collected at Groton. Assuming the inlet total halide concentration is 7 to 45 ppmv, as
measured from initial LFG analysis, the removal efficiency is estimated to be greater than 99
percent.

Total sulfur remained below detection limits when measured with summa canisters.
However, duplicate analyses from Tedlar bags indicated 0.017 ppmv carbon disulfide (0.033
ppmv total sulfur as H2S) at 5,803 hours, and 0.014 ppmv carbon disulfide (0.027 ppmv total
sulfur as H2S) at 5,805 hours. These data suggest that low levels of carbon disulfide may have
adsorbed onto the summa canister walls. Although adsorption of carbonyl disulfide is a
concern, the fuel cell preprocessor has the capability to remove organic sulfur, provided the
concentration is less than 30 ppmv.

5.1.2 GPU Flare Emissions

The flare was tested during regeneration of Bed A. Samples were collected during hot
regeneration of the carbon bed and the dryer bed, and cold regeneration of the dryer bed. The
highest concentrations of VOCs and sulfur compounds were measured during the hot
regeneration of the dryer bed. The data demonstrated that the flare reduced total NMOC levels
from 21,100 ppmv to 11.5 ppmv during the worst-case hot dehydration bed regeneration (see
Table 5-4).

5-4


-------
Table 5-3. GPU Exit Contaminant Levels - Groton Landfill

June 1996 - July 1997

GPU Operating Time (hr)

2320

4645

4651

5514

5518

5803

5805

62864

62874

SULFUR COMPOUNDS (ppmv)

Hydrogen Sulfide

N.A.

N.A.

N.A.

N.A.

N.A.

N.A.

N.A.

N.A.

N.A.

Methyl Mercaptan

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

Ethyl Mercaptan

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

Dimethyl Sulfide

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

Dimethyl Disulfide

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

Carbonyl Sulfide

0.022

<0.004

<0.004

<0.004

<0.004

<0.004

<0.004

0.010

0.080

Carbon Disulfide

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

Total Sulfur (as H2S)

0.022

nd

nd

nd

nd

nd1

nd1

0.010

0.080

VOLATILE ORGANIC COMPOUNDS

ppmv)

Dichlorodifluoro methane

N.A.

N.A.

N.A.

N.A.

N.A.

N.A.

N.A.

N.A.

N.A.

1,1-dichloroethane

<0.002

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

Benzene

<0.003

<0.002

<0.002

<0.002

<0.002

<0.002

<0.002

<0.001

0.00042

Chloro benzene

<0.002

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

Ethyl Benzene

<0.002

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

Styrene

<0.002

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

0.0001

Trichloroethene

<0.002

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

Toluene

0.007

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

0.00039

T etrach loroethene

<0.002

<0.002

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

Vinyl Chloride

<0.004

<0.001

<0.002

<0.002

<0.002

<0.002

<0.002

<0.001

0.0022

Xylene Isomers

0.001

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

0.002

<0.001

Cis 1-2-dichloroethene

<0.002

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

0.00015

Total Halides (as CI)

0.014

nd

nd

nd

nd

nd

nd

0.0122

0.0193

nd = non-detected.

N.A. = data not available.

1 Carbon disulfide was not detected in Summa Canister sample, but was detected in a tedlar bag sample at 0.017 ppmv (0.033
ppmv totalsulfur) at 5803 hours, and at 0.014 ppmv (0.027 ppmv total sulfur) at 5805 hours.

Chloromethane detected at 0.012 ppmv and Bromomethane at 0.00044 ppmv.

3	Also detected (ppmv): Chloromethane = 0.013, Bromomethane = 0.00046, Chloroethane = 0.0016, and
Trichlorofluoromethane = 0.00033.

4	Data taken with final carbon beds operating warmer than normal, as part of testing for reduced cost GPU.

In order to calculate flare destruction efficiencies, volumetric gas flow rate at the flare exhaust
was required. Although an exit gas flow rate measurement was conducted according to EPA
Method 2, the rate was determined to be below the method's detection limit. As a result,
destruction efficiency had to be estimated. It was estimated based on the sum of the methane
and non-methane gas entering the flare, the stoichiometric combustion air required to oxidize
the methane entering the flare, and a measured excess air factor of 2.3 based on oxygen
content of the flare exhaust (60 CFR 40). The calculated flare exhaust flowrate was 368 scfm
based on 25 scfm total gas flow entering the flare at 44.8 percent methane concentration, the
stoichiometric air, and the excess air. Based on these calculations, there was a 14.7 times
more gas flow at the outlet sampling location than there was at the inlet sampling location; a
factor of 14.7 was used to calculate destruction efficiency. Based on this estimation method,
the destruction efficiencies during hot regeneration of the dryer bed was calculated to be greater
than 99 percent: dichloromethane at >99.98 percent, tetrachloroethylene at >99.85 percent,
dimethyl sulfide at >99.2 percent, and total nonmethane organics at about 99.2 percent. Flare
test data for these individual compounds are summarized in Table 5-4, and discussions of the
data are provided below.

5-5


-------
Table 5-4.

GPU Flare Emission Levels (ppmv)







Penrose Landfill (October 23, 1993)







Equipment Tested

GPU

Flare

Flare

Flare

Time

1000

-1700

1030

- 1130

1230

- 1330

1730

- 1830

Process Activity

Bed B: on-line,

Carbon Bed A

Dryer Bed A

Dryer Bed A



Bed A: on

reqeneration

Reqeneration (Hot)

Reqeneration (Hot)

Reqeneration (Cold)

Sample Location

GPU Inlet

GPU Outlet

Flare Inlet

Flare Outlet

Flare Inlet

Flare Outlet

Flare Inlet

Flare Outlet

Methane

472,000

483,000

440,000

<1

448,000

<1

463,000

<1

Total Non-Methane Organics

5,700

13.8

1,860

11.7

21,000

11.5

250

6.8

Oxides of Nitrogen4

NR 1

NR

NR

7.5

NR

8.9

NR

14.9

Carbon monoxide4

NR

NR

NR

5.8

NR

1.7

NR

1.6

Hydrogen Sulfide

106

<0.004

<0.004

NR

<0.016

NR

<0.004

NR

Carbon dioxide (%)

2

2

2

6.36

2

6.26

2

7.76

Oxygen (%)

2

2

2

14.9

2

15.03

2

13.5

Nitrogen (%)

2

2

2

78.86

2

78.86

2

78.86

C1 through C3 Sulfur CPDS (Total As H2S)

117

0.017

0.254

NR

80.4

NR

0.05

NR

Carbonyl Sulfide

0.16

0.017

0.061

NR

<0.016

NR

0.014

NR

Methyl Mercaptan

2.79

<0.004

<0.004

NR

0.087

NR

<0.004

NR

Ethyl Mercaptan

0.44

<0.004

<0.004

NR

0.016

NR

<0.004

NR

Dimethyl Sulfide

6.57

<0.004

0.042

NR

73.9

NR

0.031

NR

Carbon Disulfide

<0.04

<0.002

0.146

NR

<0.008

NR

<0.002

NR

Dimethyl Disulfide

<0.04

<0.002

<0.002

NR

0.908

NR

0.005

NR

Toxic Air Contaminants

















Benzene

1.7

<0.002

0.03

<0.002

16

<0.002

<0.04

<0.002

Chlorobenzene

1.4

<0.002

<0.02

<0.002

3.8

<0.002

0.07

<0.002

1,2 Dichloroethane

<0.35

<0.002

<0.02

<0.002

<2.5

<0.002

<0.04

<0.002

Dichloromethane (Methylene chloride)

4.1

<0.002

0.28

<0.002

110

<0.002

0.07

<0.002

Tetrachloroethylene (tetrachloroethene)

4.8

<0.002

0.17

<0.002

19

<0.002

0.1

<0.002

Tetrachloromethane

<0.23

<0.001

<0.02

<0.001

<1.6

<0.001

<0.03

<0.001

Toluene

47

<0.002

1.2

0.007

230

<0.004

0.83

0.0025

1,1,1 Trichloroethane

<0.26

<0.001

<0.02

<0.001

<1.9

<0.001

<0.03

<0.001

Trichloroethylene

2.4

<0.002

0.02

<0.002

17

<0.002

<0.03

<0.002

Trichloromethane

<0.29

<0.001

<0.02

<0.001

<2.1

<0.001

<0.03

<0.001

Vinyl Chloride

1.4

<0.002

1.5

<0.002

<3.9

<0.002

<0.05

<0.002

Xylene

28.2

<0.002

0.04

<0.002

43.8

<0.002

1.8

<0.002

Additional Contaminants

















Dichlorodifluoromethane

0.26

<0.002

3.6

<0.002

<2.0

<0.002

<0.03

<0.002

Cis - 1,2 - Dichloroethene

5.8

<0.002

<0.02

<0.002

62

<0.002

<0.04

<0.002

1,1- Dichloroethane

2.8

<0.002

<0.02

<0.002

32

<0.002

<0.04

<0.002

Ethyl Benzene

12

<0.002

0.04

<0.002

25

<0.002

0.76

<0.002

Styrene

1.1

<0.002

<0.02

<0.002

<2.4

<0.002

<0.03

<0.002

Acetone

15

<0.005

<0.07

<0.005

150

0.065

<0.12

0.02

2 - Butanone

3.7

<0.004

<0.06

<0.004

28

<0.004

<0.99

<0.004

Ethyl Acetate

10.8

<0.002

<0.04

<0.002

5.4

<0.002

<0.04

<0.002

Ethyl Butyrate

8.4

<0.002

<0.04

<0.002

2.1

<0.002

<0.04

<0.002

Alpha - Pinene

18

<0.002

0.05

<0.002

3.6

<0.002

1.8

<0.002

d-Limonene

18

<0.002

0.07

<0.002

1.4

<0.002

3.6

<0.002

Tetrahydrofuran

2

<0.002

<0.04

<0.002

0.99

<0.002

<0.04

<0.002

Total Particulates (lag/m3^

NR

NR

NR

41,900

NR

41,000

NR

20,300

Front Half

NR

NR

NR

0.0069

NR

0.0135

NR

0.0072

Back Half (Organic)

NR

NR

NR

0.0005

NR

0.0010

NR

0.0011

Back Half (Inorganic)

NR

NR

NR

0.0108

NR

0.0033

NR

0.0005

Moisture (%)



<0.01

<0.1

9.2

<0.1

9.1

<0.1

8.6

Temperature (°F)





80

1186

80

929

79

990

Flowrate (slpm)

2260

1560

700

11,2403

700

11,4403

700

91703

NOTES

















1 NR = Not Required

















' Typical landfill gas values are: 43.9% CH4, 40.1% C02, 15.6% N2, 0.4% O













Calculated based on the sum of the methane and non-methane gas entering the flare, the stoichiometric combustion of air to oxidize the methane entering the flare,

and the excess air based on 02 content of the flare exhaust.













Ambient air concentration <1.0 PPMV

















8-hour ambient air sample collected within 20 feet of flare measured 267 |ag/m (equivalent to 0.000116 gr/dscf)







Calculated based on the flare inlet CO? plus complete combustion of orqanics in flare to CO?. Percent nitroqen calculated as 100% minus sum of O? and CO?.

Flare Destruction of VOCs

The highest VOC concentration entering the flare occurred during hot regeneration of the dryer
bed. One-hour samples were collected in Tedlar bags simultaneously at the inlet and outlet
during each phase of regeneration. The samples were analyzed for target compounds by
GC/MS according to EPA Method TO-14 (60 CFR 40).

Toluene and acetone were the highest concentration VOCs entering the flare, at 230
ppmv and 150 ppmv, respectively. Inlet halide concentrations were also significant, with
methylene chloride at 110 ppmv; cis-1,2-dichloroethane at 62 ppmv; 1,1-dichloroethane at 32
ppmv; trichloroethane at 17 ppmv; tetrachloroethane at 19 ppmv; and chlorobenzene at 3.8

5-6


-------
ppmv. Flare outlet concentrations of these compounds were below the GC/MS detection limit of
0.002 ppmv.

The destruction efficiency of methylene chloride was greater than 99.97 percent based
on a calculated flare exhaust flow of 368 scfm and inlet flow of 25 scfm. The destruction
efficiency of tetrachloroethene, which is relatively difficult to oxidize, was greater than 99.85
percent.

Flare Destruction of Total Non-Methane Orqanics

The highest concentration of NMOCs was also measured during hot regeneration of the
dehydration bed. Inlet concentration was 21,100 ppmv (as carbon) and the outlet concentration
was 11.5 ppmv. Based on a 14.7-fold increase in air flow at the outlet, the destruction efficiency
was 99.2 percent.

Flare Outlet Concentration of NOv. CO?, and Particulate Matter

The nitrogen oxides (NOx) and carbon monoxide (CO) concentrations at the flare outlet
averaged 10.4 ppmv and 3.0 ppmv, respectively, over the three test periods. Particulate matter,
based on the front-half catch, averaged 0.009 grains/dscf over the three test runs. Particulate
matter, based on front-half and back-half catches, averaged 0.013 grains/dscf.

The ambient concentrations of NOx and CO were below the detection limits of the
analyzers. Detection limits were 1.0 ppmv for each compound. Particulate matter was
measured with one eight-hour sample collected within 20 feet of the flare on the day of the flare
emission testing. The particulate matter concentration was 267 micrograms per cubic meter
(ug/m3).

Condensate Analyses

One condensate sample was collected from the first cooler condenser during the first
hour of each cycle for a total of three samples. There was no condensate in the second
condenser; as a result, no sample could be collected. Each sample was analyzed for the target
sulfur compounds by GC/FPD and the target VOCs by GC/MS.

The highest concentration of VOCs were measured for acetone and 2-butanone, which
were detected in each sample. The average concentrations were 16.7 mg/l of acetone and 12.7
mg/l of 2-butanone. The highest concentration of a target sulfur compound was 1.7 mg/l of
dimethyl sulfide. However, an unknown sulfur compound was also detected in each sample
which increased the average total sulfur concentration to 33.0 mg/l.

5.1.3 GPU Operation and Availability

A summary of GPU operation at the Penrose and Groton landfills is provided in Table 5-
5. At the Penrose landfill, the GPU exceeded the performance goal of 200 hours of continuous
operation with the longest continuous run of 342 hours. It also exceeded the goal of 500 total
hours of operation with 1,782 hours logged with the fuel cell operating and a total operation of
2,297 hours, which includes initial GPU shakedown periods. The shutdowns at Penrose landfill
were caused by a loss of the flare UV flame sensor, loss of temperature in the cooling process

5-7


-------
for moisture elimination, condensate tank overflow due to high condensate influx at the site,
electronic lock-up of control valves, and loss of LFG pressure to the GPU. Modifications were
made to prevent such occurrences in the future.

Table 5-5. GPU Operation Summary During Fuel Cell Operating Periods

Landfill Test
Site

Test Period

Total
Hours

Total
Adjusted
Hours

Total
Operational
Hours

Total
Shutdown
Hours

Gross
Availability

Penrose

12/7/94 -
2/19/95

1782

810

707

103

87.3%

Groton

6/17/96-
7/14/97

9408

9262

4168

5094

45.0%
70% (last 6
months)

The gross availability of the GPU for the entire Groton test was 45 percent. Of the 21
GPU related shutdowns, about half were due to one-of-a-kind mechanical failures which were
corrected and did not reoccur. Other shutdowns were due to three major system issues: high
pressure drop across the GPU, periodic freeze-ups of the refrigeration system, and an under
performing LFG compressor exhaust valve. Recall, the compressor was added at Groton to
pressurize the incoming LFG. An estimated down time of approximately 1,050 hours was
experienced as a result of malfunctioning compressor valves.

Since the compressor was not a component of the overall GPU system design, it can be
generalized that shutdowns resulting from compressor are not representative of GPU's
performance. When these downtimes are removed, the gross availability improves from 45
percent to 56 percent. After the remaining mechanical failures were corrected, the gross
availability increased to 70 percent over the second half of the demonstration period. The
longest operating period recorded was 827 hours.

5.1.4 GPU Exit Heat Content

Heat content measurements of the GPU exit gas were conducted to provide a basis for
determining fuel cell efficiency. Table 5-6 and Table 5-7 summarize the data for the Penrose
and Groton landfills. The average higher heating value at Penrose was 445.8 Btu/scf, versus
580.6 Btu/scf at Groton. The most significant difference was the lower nitrogen content and
higher methane content in the Groton gas. Both sites contained low levels of higher
hydrocarbons.

5-8


-------
Table 5-6. GPU Exit Heat Content Measurements - Penrose Landfill



















Sampling date

1/19/95

1/20/95

1/25/95

1/26/95

2/9/95

2/10/95

2/17/95

Averaa

Sampling time

16:44

09:27

16:09

08:31

10:37

09:26

13:33

GPU Exit Gas Composition (Measured by ASTM Method)

Nitrogen (%)

16.27

17.25

16.24

16.34

23.89

17.66

20.10

17.31

Carbon dioxide (%)

35.54

38.90

39.56

39.53

36.04

38.86

34.91

37.88

Methane (%)

44.17

43.81

44.14

44.09

40.07

43.48

45.00

44.11

Ethane (%)

0.024

0.029

0.049

0.037

nd1

nd

nd

0.02

Propane (%)

nd

nd

nd

nd

nd

nd

nd

nd

Hexanes

nd

nd

nd

nd

nd

nd

nd

nd

> Hexanes

nd

nd

nd

nd

nd

nd

nd

nd

Higher Heating Value

450.5

447.4

451.5

450.5

409.1

443.4

458.4

445.8

(BTU/scf)

















Lower Heating Value

406.0

403.0

406.0

405.0

367.6

398.0

413.1

401.3

(BTU/scf)

















nd = non-detected.

Table 5-7. GPU Exit Heat Content Measurements - Groton Landfill



















Sampling date

3/20/97

5/19/97

5/19/97

6/19/97

6/19/97

7/9/97



Sampling time

14:30

10:15

14:53

10:15

12:30

14:15

















Average

GPU Exit Gas Composition (Measured by ASTM Method)

Nitrogen (%)

0.07

1.68

1.68

0.93

1.45

1.42

1.16

Carbon dioxide (%)

42.14

41.24

41.00

41.46

39.19

40.19

41.21

Methane (%)

57.78

56.59

56.86

57.32

58.70

57.97

57.30

Ethane (%)

nd

nd

Nd

nd

nd

nd

nd

Propane (%)

nd

nd

Nd

nd

nd

nd

nd

Hexanes

nd

nd

Nd

nd

nd

nd

nd

Higher Heating Value

585.5

573.4

576.0

580.9

595.5

587.4

580.6

(BTU/scf)















Lower Heating Value

527.2

516.3

518.7

523.0

536.2

528.9

522.8

(BTU/scf)















nd = non-detected.















Standard conditions at 60

°F and 14.7 psia.











5.1.5 Fuel Cell Power Output and Efficiency

The fuel cell did not deliver the 140 kW power expected at the Penrose landfill. This was
attributed to the lower heat content LFG measured than initially expected. The Groton test
results indicate a maximum power production of 165 kW, with 140 kW of steady production.
The increase in power is due to about a 10 percent increase in the LFG heating value.

5-9


-------
Table 5-8. Fuel Cell Power Output Results

Landfill Test
Site

Fuel Cell
Capacity

Fuel Cell
Performance
Goal

Maximum
Power
Output
Achieved

Steady Power
Output
Achieved

Penrose

200 kW

140 kW

137 kW

120 kW

Groton

200 kW

140 kW

165 kW

140 kW

Fuel cell efficiency was calculated using lower heating values (LHV) measured at the
GPU exit per ASTM method (ASTM 1991 and 1996). The efficiency for the two tests ranged
between 37.1 and 38.0 percent (see Table 5-9). They were derived by using the average LHV
measurements corresponding to the days when steady state power production occurred.

Table 5-9

Fuel Cell Efficiencv Results









Period of
Steady
Power
Output

Net Energy
Output Per
Electric Meter
(kW)

Gas Flow
Consumed
(ft3)

GPU Exit LHV
by ASTM
Method
(BTU/scf)

Efficiency''

Penrose

1/24/95 to
1/30/95

16,800

3.92E+5

401.51

37.1%

Groton

6/10/97 to
6/19/97

28,682

4.87E+5

529.62

38.0%

' Average of two measurements taken on 1/25/95 and 1/26/95.

2	Average of two measurements taken on 6/19/97.

3	Based on lower heating value.

5.1.6 Fuel Cell Availability

A summary of the fuel cell availability for the two test periods is presented in Table 5-10.
As with the GPU, adjusted availability discounts shutdown periods that were not directly
attributable to the fuel cell. At both tests, the adjusted availability was determined to range
between 96 and 98 percent.

Table 5-10. Summary of Fuel Cell Operation During Test Periods



Test Period

Total
Hours

Total
Adjusted
Hours

Total
Operational
Hours

Total
Shutdown
Hours

Adjusted
Availability

Penrose

12/7/94 -
2/19/95

1782

718

707

11

98.5%

Groton

7/15/96-
7/14/97

8760

3432

3313

5328

96.5%

During the Penrose test, a single shutdown due to a bad sensor module in the fuel cell
control system was experienced. A single shutdown due to the fuel cell also occurred at
Groton. The cause of this shutdown was due to a mechanical failure within the fuel cell power
plant resulting from the failure of several electrical space heating elements inside the unit. This

5-10


-------
in turn damaged a pump, valve, and flow switch due to freezing of the parts. The mechanical
components and heaters were replaced and normal operation was resumed.

5.1.7 Fuel Cell Emissions

The fuel cell emissions data were based on six, one-hour continuous monitor
measurements conducted at the Penrose landfill. It was anticipated that the results of the
emission testing conducted during this period would be representative of longer term continuous
emissions. This is because the fuel cell power plant controller continuously adjusts the fuel and
air mixture to maintain constant temperature inside the reformer burner where carbon monoxide
and NOx are generated.

Table 5-11 summarizes the measurements data, reported as actual dry concentrations
in ppmv. The power plant S02 emissions (0.23 ppmv) were below the method detection limit.
NOx emissions averaged 0.12 ppmv, and CO emissions were near the detection limit, averaging
0.77 ppmv.

Table 5-11.

Fuel Cell Emissions Summary - Penrose Landfill







February 17, 1995















Fuel Cell Operatinq Time (Hours)





660-661

662-663 | 664-665

666-667

667-668



669-670

Averaqe

Concentration (dry measurements)

Nitrogen oxides (ppmv)

0.3

0.17



0.31



0.17



0.41



0.18

0.26

Sulfur dioxide (ppmv)

<0.5

<0.5



<0.5



<0.5



<0.5



<0.5

<0.50

Carbon monoxide (ppmv)

1.5

1.8



2.1



2.3



0.6



1.9

1.70

Oxygen (%)

7.96

8.01



7.88



7.8



8.03



7.91

7.93

Carbon dioxide (%)

12.5

12.6



12.7



12.3



12.4



12.5

12.50

Concentration (dry measurements, corrected to 15% oxyqen



Nitrogen oxides (ppmv)

0.14

0.08



0.14



0.08



0.19



0.08

0.12

Sulfur dioxide (ppmv)

<0.23

<0.23



<0.23



<0.23



<0.23



<0.23

<0.23

Carbon monoxide (ppmv)

0.68

0.82



0.95



1.04



0.28



0.86

0.77

Volumetric Flow Rate (dscm/m)1

10.1

10.1

9.4

9.4

9.7

9.7

9.7

Stack Temperature (°C)

56.7

56.7

43.3

43.3

42.8

42.8

48

Mass Emission Rate (q/hr)

Nitrogen oxides

0.35

0.20



0.33



0.18



0.46



0.20

0.29

Sulfur dioxide

<0.80

<0.80



<0.75



<0.75



<0.78



<0.78

<0.78

Carbon monoxide

1.06

1.27



1.37



1.51



0.41



1.29

1.15

Mass Emission Rate (q/kW-hr)

Nitrogen oxides

0.0029

0.0016

0.0028

0.0015

0.0038

0.0017

0.0024

Sulfur dioxide

<0.0067

<0.0067

<0.0062

<0.0062

<0.0065

<0.0065

<0.0065

Carbon monoxide

0.0088

0.0106

0.0115

0.0125

0.0034

0.0107

0.0096

1 dscm/m = dry standard cubic meters per minute 20 °C

The LFG fuel cell results agree well with other emissions data measured from natural
gas fuel cells. Average emissions for 16 fuel cell power plants tested at IFC's manufacturing
facility using natural gas fuel are: 0.46 ppmv NOx and 1.1 ppmv CO. Emission tests were also
conducted on a fuel cell unit operating at the SCAQMD Headquarters building in Diamond Bar,
CA. The results showed 0.45 ppmv NOx and 1.1 ppmv CO. The Diamond Bar site results were
confirmed by two independent laboratories, and were used by SCAQMD as the basis for a
blanket exemption from air permit requirements for fuel cells in the Los Angeles basin.

5-11


-------
REFERENCES

60	CFR 40. 1997. Federal Register, U.S. Environmental Protection Agency, "Standards of
Performance for New Stationary Sources," Archives and Records Administration, Appendices A
and M, 1997.

61	CFR 49. 1996. Federal Register, U.S. Environmental Protection Agency, "Standards of
Performance for New Stationary Sources and Guidelines for Controls of Existing Sources,
Municipal Solid Waste Landfills," Final Rule and Guidelines, pp. 9905-9944, March 12, 1996.

ANSI C12. 1995. American National Standards Institute, "Code For Electric Metering," National
Electrical Manufacturers Association, Rosslyn, VA, 1995.

ASTM D-1945. 1996. American Society For Testing and Materials, "Standard Test Method for
Analysis of Natural Gas by Gas Chromatography," West Conshohocken, PA, 1996.

ASTM D-3588. 1991. American Society For Testing and Materials, "Standard Practice for
Calculating Heat Value, Compressibility Factor, and Relative Density (Specific Gravity) of
Gaseous Fuels, West Conshohocken, PA, 1991.

Augenstein, D., and J. Pacey. 1992. U.S. Environmental Protection Agency, Landfill Gas
Energy Utilization: Technology Options and Case Studies, EPA-600/R-92-116 (NTIS PB92-
203116), Office of Research and Development, Research Triangle Park, NC, June 1992.

EPA 1979. U.S. Environmental Protection Agency, Quality Assurance Handbook for Air
Pollution Measurement Systems, Volume III, EPA-600/4-77-027b (NTIS PB80-112303),
Environmental Monitoring and Support Laboratory, Research Triangle Park, NC, May 1979.

EPA 1991. U.S. Environmental Protection Agency, Air Emissions for Municipal Solid Waste
Landfills- Background Information for Proposed Standards and Guidelines, EPA-450/3-90-011a
(NTIS PB91-197061), Office of Air Quality Planning and Standards, Research Triangle Park, NC
March 1991.

EPA 1993. U.S. Environmental Protection Agency, Opportunities to Reduce Anthropogenic
Methane Emissions in the United States, Report To Congress, EPA-430-R-93-012, Office of Air
and Radiation, Washington, D.C., October 1993.

EPA 1997. U.S. Environmental Protection Agency, Energy Project Landfill Gas Utilization
Software (E-Plus) Version 1.0a, Atmospheric Pollution Prevention Division, Washington, D.C.,
1997.

Graham, J., and M. Ramaratnam. 1993. "Recover VOCs Using Activated Carbon," Chemical
Engineering, pp. 6-12, February 1993 (Supplement).

IPCC 1995. International Panel On Climate Change, Climate Change 1995: The Science of
Climate Change Summary for Policy Makers and Technical Summary of the Working Group I
Report, IPCC Working Group I Technical Support Unit, Bracknell, United Kingdom, 1995.

6-1


-------
Preston, J.L., and J.C. Trocciola. 1998. U.S. Environmental Protection Agency, Demonstration
of Fuel Cells to Recover Energy From Landfill Gas: Groton Landfill Final Report, EPA/NRMRL-
RTP-129, Office of Research and Development, Research Triangle Park, NC, In Press.

Riggin, R.M. 1988. U.S. Environmental Protection Agency, Compendium of Methods for the
Determination of Toxic Organic Compounds in Ambient Air, EPA-600/4-84-041 (NTIS PB87-
168688), Office of Research and Development, Research Triangle Park, NC, May 1988.

Roe, S., J. Reisman, R. Strait, and M. Doom. 1998. U.S. Environmental Protection Agency,
Emerging Technologies For the Management and Utilization of Landfill Gas, EPA-600/R-98-021
(NTIS PB98-127913), Office of Research and Development, Research Triangle Park, NC,
February 1998.

Trocciola, J.C., and J.L. Preston. 1998a. U.S. Environmental Protection Agency,
Demonstration of Fuel Cells to Recover Energy From Landfill Gas: Phase III and Phase IV,
Volume 1, Technical Report, EPA-600/R-98-002a (NTIS PB98-127368), Office of Research and
Development, Research Triangle Park, NC, January 1998.

Trocciola, J.C., and J.L. Preston. 1998b. U.S. Environmental Protection Agency,
Demonstration of Fuel Cells to Recover Energy From Landfill Gas: Phase III and Phase IV,
Volume 2, Appendices, EPA-600/R-98-002b (NTIS PB98-127376), Office of Research and
Development, Research Triangle Park, NC, January 1998.

Winberry, W.T., N.T. Murphy, and R.M. Riggin. 1988. U.S. Environmental Protection Agency,
Second Supplement to Compendium of Methods for the Determination of Toxic Organic
Compounds in Ambient Air, EPA-600/4-89-018 (NTIS PB90-127382), Atmospheric Research
and Exposure Laboratory, Research Triangle Park, NC, June 1988.

6-2


-------