01FL - 470
Ammonia Emissions from the EPA's Light Duty Test Vehicle
Richard C. Shores*, John T. Walker Jr., Larry G. Jones
U. S. Environmental Protection Agency, Office of Research and Development, National Risk Management Research
Laboratory, Research Triangle Park, NC
M. O. Rodgers, James R. Pearson
Air Quality Laboratory, Georgia Institute of Technology, Atlanta, GA
Ronald B. McCulloch
North Carolina Division of Air Quality, Raleigh, NC
ABSTRACT
Ammonia (1NH3) emissions were measured from the
EPA's Light Duty Test Vehicle while operated on a
chassis dynamometer. The vehicle's (1993
Chevrolet equipped with a three-way catalyst)
emissions were measured for three transient (urban
driving, highway fuel economy, and hard
acceleration) cycles and steady state operation.
Previous research L 2 has shown that NHi is
predominately emitted from vehicles with a catalyst
(three-way or dual-bed). The vehicle's catalyst is
designed to reduce nitrogen oxides (NOx) to
nitrogen (N2) and oxygen (O2) during normal
operation. The reduction of NOx to NH3 occurs
during periods of reducing conditions when
insufficient O; is available. NH3 emissions were
measured during fucl-rich/reduced-O? conditions
(open-loop control scheme). The results
demonstrate that NH3 production is correlated to
combustion conditions3. The results also show that
the amount of NH3 produced correlates with the
amount of time that the vehicle remains in the open-
loop control scheme. The significance of this
finding is that NH3 production can be predicted for
a .fleet based on the frequency of enrichment of
vehicles equipped with a three-way catalyst. The
results also provide a means of determining the
location of roadway links and/or specific locations
where NH3 production can be anticipated based on
predicted engine power.
INTRODUCTION
North Carolina and other states have been studying4
the overloading of nitrogen compounds that has
occurred in rivers and waterways, resulting in algae
plumes and fish kills. Traditionally these sources of
nitrogen compounds have been blamed on runoff
and overflow from agricultural operations. Acid
rain monitoring networks, such the National Acid
Deposition Program/National Trends Network
(NADP/NTN), have indicated a continuous increase
in ammonium (NH4+) concentration within the rain
samples collected3. If these samples are considered
as a surrogate for air concentrations, nitrogen
compounds in the atmosphere are also increasing.
In 1989, 85 percent of the on-road vehicles were
equipped with a three-way catalyst. Remote
sensing data collected in 1999 in Raleigh, NC,
indicated that 97 percent of the on-road vehicles
were equipped with a three-way catalyst. From
1989 to 1999, there was an 11.3 percent increase6,7'8
in vehicles operated on North Carolina roads. The
combination of a greater percentage of the on-road
vehicle fleet being equipped with a three-way
catalyst and an increase in vehicle fleet size resulted
in a 20 percent increase of vehicles with three-way
catalysts being operated on North Carolina roads.
The chemistry required to generate NH3 is a l'uel-
rich/reduced-Oi condition at the catalyst. These
conditions are controlled by the engine's computer
and exist for power demand and start-up operation.

-------
For example, n-octane (a common component of
gasoline) could be reduced to NH? by reactions such
as:
2NO + C8Hi0 + 4O2 —> 2NH3 + 8CO + 2H20 (1)
When an excess of 02 is available, the catalyst
would function as designed to destroy the nitric
oxide (NO), and the following reaction would
predominate:
2NO + C8H10 + 9.502 -> 8C02 + 5H20 + N2 (2)
To date, NH3 emissions have been calculated based
upon a mass emission for a distance traveled,
resulting in evenly distributed emissions across the
modeled region. Geographical information systems
(GlS)-based modeling efforts allow modal and
temporal distribution of emissions onto a road
network rather than over a region. The resultant
data of GIS-based modeling provide emissions data
with better spatial resolution to support ambient air
deposition and exposure assessments for
populations affected by mobile emissions.
EXPERIMENTAL METHODS
The EPA's light duty test vehicle. A 1993
Chevrolet Lumina was purchased by EPA's Air
Pollution Prevention and Control Division in 1998.
The vehicle has a computer-controlled, fuel-injected
3.1-liter 6-cylinder engine and a three-way catalyst.
The vehicle's engine control, engine, and exhaust
system have not been modified and are as they were
when manufactured.
The purpose of this test vehicle was to determine
the functional relationship between the vehicle's
engine and control system and the emissions
produced. The engine and emissions control system
in this test vehicle, and in most light duty vehicles
now in service, are computer controlled. Emissions
were measured while the vehicle was operated on a
chassis dynamometer under a variety of operating
conditions. The inputs to the computer from the
sensors for O2, mass air flow, engine revolutions per
minute, and other parameters were monitored to
determine when the computer commanded a shift
from closed-loop to open-loop, fuel-enrichment
operation. Under closed-loop operation, the catalyst
functions as designed to produce effective
emissions control. Under short periods of open-
loop operation, primarily to boost the power ouput
of the engine, the catalyst operating conditions are
drastically affected and the emissions are greatly
increased.
Vehicles are generally designed to minimize open-
loop operation when being tested on a chassis
dynamometer in accordance with the federal
certification test cycle procedure. The test
procedure was revised to include more high-speed
operation, and manufacturers are thereby limited in
the programmed use of open-loop operation in
vehicle control schemes. However, use of fuel
enrichment in vehicle control schemes is not
entirely eliminated for a variety of customer
satisfaction and technical (e.g., catalyst cooling)
reasons. These new Supplemental Federal Test
Procedures arc being phased in between 2001 and
2004 model years. However, vehicles can have
significant periods of open-loop operation when
high power demands (e.g., heavy acceleration) are
commanded by the vehicle operator, especially in
vehicles built in earlier model years.
In this study, the vehicle engineering parameters
correlated with the NH3 emissions arc recorded and
analyzed with an instrumented vehicle. The
instrumented vehicle was tested on a number of
commonly used dynamometer test cycles as well as
evaluated on the roadway. These data will be useful
in assessing the emissions impact at specific
roadway segments where fuel enrichment is
prevalent.
An additional purpose of this test vehicle was to
establish a relationship between emissions
measured on the road and on the dynamometer.9
This is important because the majority of emissions
data to date have been collected while the vehicles
were operated on dynamometers, yet the emissions
are modeled for vehicles operated on the road.
The vehicle contains four deep-cycle 12-volt
batteries and a true sinusoidal inverter to power
instrumentation and operate data-recording
computers. The vehicle is equipped with a receiver
type hitch and a lightweight trailer rated for 204
kilograms.
Measurement methods. Gaseous measurements
are made for carbon monoxide (CO), carbon
dioxide (C02), hydrocarbons (HCs), nitrogen

-------
oxides (NOx), and NH3. The exhaust gas is
sampled at three ports. The gas sampled from the
first port is cooled to remove water and delivered to
the CO, CO2, HC, and NOx monitors under a slight
pressure [-12 centimeters water (H2O)]. Gas
sampled from the second port is diluted, cooled
with conditioned ambient air, and delivered to the
NH3 and CO2 monitors. The conditioned ambient
air has been dried and cleaned using silica gel,
charcoal, and a mol sieve. The second CO2 monitor
measures the diluted sample concentrations to
ensure that a dilution ratio of -15:1 is being
maintained. The dilution ratio was measured to
prevent moisture condensation in the sample lines
and to calculate the true exhaust gas NH3
concentration. The third port is used for direct
exhaust gas sampling. The vehicle is also equipped
with hardware and software to record the engine
computer data stream [i.e., revolutions per minute
(RPM), fuel flow, and commanded enrichment].
This data stream, along with a wide-range O2 sensor
mounted in the exhaust system, is used to determine
if the vehicle is operating in a fuel-rich/02-reduced
condition. Exhaust gas concentrations, engine
computer data, and other data of interest are
recorded using a personal computer (PC).
Two techniques were used to measure NH3: (1)
analysis of NH/ ions from impingers containing
dilute sulfuric acid (H2SO4),10'11 and (2)
I ^
chemiluminescence analyzer ~ determination of
NH3 as NO. The acid impinger or gas washing
technique represents an average concentration over
a period of time, usually 10-20 minutes. Gas was
sampled from the third port, pulled through a gas-
washing vessel, equipped with a diffusing cylinder
and filled with -0.01 N H2SO4, at a rate of
approximately 1 liter per minute. Gaseous NH3 in
the sample is absorbed into solution as bubbles from
the gas sample ascend through the liquid column.
The concentration of NH3 in solution is then
determined (as NH4+) by colorimetric flow injection
analysis or by ion chromatography using a Lachat
Quikchem Model 8000 FIA Auto-analyzer with
three channels [ortho-P, NH4+, and nitrate (NO3")].
The method is based on sodium salicylate (not
alkaline phenol) and has a 10 ppb detection limit for
NH/.
The TECO Model 17C chemiluminescence NH3
analyzer is designed to sample at intervals of no less
than 40 seconds. This is reasonable because, when
7
operated in the automatic mode, the analyzer has to
sample NO, NO2, NOx, and total nitrogen (N,)
before there is enough information to calculate an
NH3+ concentration. To overcome this sampling
time problem, two chemiluminescence analyzers
were operated in the manual mode, one indicating
total nitrogen (N,= NO + NO2+ NH3) continuously
and the other indicating NOx [NO + nitrogen
dioxide (NO2)] continuously. The difference
between these two analyzers represented the NH3
diluted concentration. The manually operated
analyzers updated the indicated concentrations
every 0.5 second and operated in a range of 0-100
ppm. True NH3 gas concentration was determined
based upon the CO2 dilution ratio.
Validation of ammonia measurements. Sampling
from NH3 gas cylinders analyzed by Scott
Technologies validated the gas washing technique.
Cylinder gas was diluted with humidified zero air
and sampled using the gas washing technique.
Dilution flow rates were regulated using rotameters
and measured using a DryCal flow measurement
device. The comparison between the gas washing
technique and diluted cylinder gas concentrations
indicated a correlation coefficient of 0.93 (r) and
Equation (3).
Gas Washing Technique, ppm = 1.7515 + 0.8561
(Diluted NH3 Cylinder Gas, ppm) (3)
Diluted cylinder gas concentrations ranged from 2
to 28 ppm, and 36 individual tests were conducted.
Figure 1 provides a graphical representation of this
comparison.
The dual analyzer technique was calibrated using
NO and NH3 gas. The N, calibration data exhibited
a correlation coefficient of 0.9993 (r~) and Equation
(4).
Analyzer Response [Nt], ppm = 0.6999 + 0.9844
(Cylinder Gas [Nt], ppm)	(4)
Calibration concentrations ranged from 0 to 85.6
ppm, and the analyzer response was characterized
before and after data collection. Equation (4) is the
result of all input concentrations regressed onto all
response concentrations. Converter efficiency was
evaluated with both NO and NH3 present.
Converter efficiency exhibited a correlation
coefficient of 0.9995 (r2) and Equation (5).

-------
Analyzer Response [Nt], ppm = -1.2973 + 0.9808
(Cylinder Gas [Nt], ppm)	(5)
During calibrations, analyzer response was adjusted
to ensure that the concentrations recorded were
accurate. Figure 2 compares these response data
and is considered to represent an estimate of
precision and accuracy of the concentrations
recorded. These data include the calibration
responses, converter efficiency, and calibration
checks conducted between tests. Figure 2 also
shows that the analyzers remained stable throughout
the testing.
The gas washing technique is an accepted method,
whereas the dual chemiluminescence analyzer
technique is an unconventional technique. The gas
washing technique was used to validate the dual
analyzer technique. Automotive exhaust gas
samples were collected simultaneously using the
gas washing and dual analyzer techniques. Average
concentrations were calculated for each of the
collected samples for both the gas washing and dual
analyzer techniques. The comparison between gas
washing and dual analyzer techniques indicated an
average difference of 1.6 ppm, a correlation
coefficient of 0.8604 (r), and Equation (6).
Gas Washing Technique, ppm = 1.6528 + 0.9630
(Dual Analyzer Technique, ppm) (6)
Average concentrations ranged from 0 to 71 ppm,
and there were 14 tests. Figure 3 provides a
representation of this comparison, and Table 1
provides the average concentrations determined for
each of the 14 tests.
Air flow rate calculations. A significant source of
error for mobile source emission rate calculations is
associated with the volumetric flow rate through the
engine. Dynamometer testing facilities provide the
opportunity to compare the total exhaust gas
volume measured by the test facility to the total
exhaust gas volume calculated for the test vehicle.
Test vehicle volumes are the summation of second-
by-second flow rate calculations. These total
exhaust gas volume calculations contain numerous
sources of error including sensor response times,
changing volumetric efficiency, temperature and
pressure changes, and changing chemistry within
the exhaust gas. Test vehicle flow rates were
calculated using three methods. The first took
advantage of the mass airflow rate data included in
the engine computer data stream. The second
method used RPM, the internal combustion volume
of the motor, and appropriate corrections for
' temperature and pressure. The third method used
the volume of fuel injected and the engine-
computer-data-stream-commanded air/fuel ratio.
The first method used the engine computer data
channel, and mass air flow (MAF) rate and was
calculated using Equation (7).
MAFair = (AFR)(VolE)(Pi/Ps)(Ts/Ti)(M/28.9)(C|)
(7
where:
MAFair = Mass air flow, liters/sec.
AFR= Engine computer flow rate, g/sec.
VoIe = Volumetric efficiency, used 15%.
Pi= Intake manifold pressure, mmHg.
Ps= Pressure, standard (760 mmHg).
Ts= Temperature, standard (298 K).
Ti= Intake manifold temperature, K.
M= Moles at 28.9 grams/mole.
C| = Constant, 24.45 liters/mole.
The second method used the known volume of the
engine, the RPM, and the intake air manifold
conditions. RPM airflow rate through the engine
was calculated using Equation (8).
RPMa,r=(REV)(Evo1)( I/2)(VolF.)(Pi/Ps)(Ts/Ti) (8)
where:
RPMA|r = RPM air flow rate, liters/sec.
REV = Revolutions per second,RPM/60.
Evoi = Engine internal volume, 3.1 liters.
Vi = 2 revolutions for engine volume.
VoIe = Volumetric efficiency, used 75%.
Pi= Intake manifold pressure, mmHg.
Ps= Ambient pressure, mmHg.
Ts= Ambient temperature, K.
Ti= Intake manifold temperature, K.

-------


p-


a 30"
~


~

c


o


'•= 25



~ \

V
~ / ~

c 20


¦0


cy
.2" ^


c
f-
t

"3


H 10
	

CL
A/


~ /J

1 5 "
~

£
X
| ~

r


0 5 10 15 20 25 30
Diluted Cylinder Concentrations, ppm
Figure 1. Comparison of the gas washing technique measured concentrations to
the ammonia gas standard diluted concentrations.
100 -
~ Calibration Points ;
~ Converter Eff.
0 20 40 60 80 100
Input Nt Cpncentration, ppm
Figure 2. Comparison of the Scott Specialty Gases calibration standards and
the Teco analyzer response. These data include the calibration responses,
converter efficiency checks, and calibration checks conducted between tests.

-------
Test Conducted
Gas
Washing,
ppm
Dual
Analyzer,
ppm
Average
MAP,
mmHg
Dyno Test
Volume,
m3
Engine
Computer
Air Flow,
m3
RPM
Technique
Volume,
m3
Injector PW
Technique
Volume,
m3
FTP
18.6
23
352.(X)
12.46
12.22 '
14.198
13.48
FTP
12.8
17.4
351.85
13.39
13.85
14.888
13.98
505
8.9
15.3
367.31
6.24
7.10
6.344
5.82
Steady State
5.7
10
300.54
7.97
6.11
8.881
5.78
Hwy Fuel
16
38.8
394.08
10.90
11.30
12.212
8.80
FTP
4.7
3
352.97
13.38
14.02
15.226
14.35
505
0
7
369.53
5.63
6.34
6.009
5.33
Steady State
1.6
12.2
302.44
8.03
5.84
8.568
5.58
Hwv Fuel
2.4
5.7
392.89
1 1.04
12.24
12.337
12.35
Hwy Fuel
6
1 1.4
395.24
10.54
12.03
12.168
11.55
Hard Acceleration
51.1
52.6
425.48
11.12
13.24
8.205
8.11
Hard Acceleration
70.7
68.8
434.56
1 1.10
13.22
8.355
7.80
Hard Acceleration
40.3
61.8
432.80
10.38
13.28
8.388
8.17
Hard Acceleration
52.7
37.9
439.50
10.72
13.78
8.346
8.33
MAP: Manifold absolute pressure.
FTP: Federal test procedure, 40 CFR. Part 86, Urban Driving Cycle.
505: Represents the third sample bag (505 seconds) of the FTP, urban driving cycle, referred to as the "hot
505."
Hwy Fuel: Federal Test Procedure, 40 CFR, Part 86, Highway Fuel Driving Cycle.
Hard Acceleration: Driving cycle included with the Clayton dynamometer software package, described in
Figure 4.
Table 1. Comparison of average ammonia concentrations, manifold pressures, and test cycle volumes
calculated using the three flow method calculation techniques.
The third method used the injector pulse width data
channel provided by the vehicle's computer.
Dependent upon commanded power, the engine
control strategy changes the ratio of pulse width
frequency to MAP sensor voltage. This makes an
accurate calibration (using total fuel volume) of the
fuel injectors nearly impossible. This response is
shown in Figure 5. Figure 5 also demonstrates how
the calibration between injector pulse width and
MAP sensor volts changes during commanded fuel
enrichment. The volume of gas injected per pulse
width was determined by recording the total fuel
volume combusted and the total number of pulse
widths over the same period of time. Fuel density
and carbon content values were taken from
References 12, 13, and 14, and no actual fuel
analysis was conducted. Injector-pulse-width-
calculated airflow rate through the engine was
calculated using Equation (9).
InjAiR = [(PW)(A/F) + (PW)](C2) (9)
where:
InjAiR = Airflow rate, liters/sec.
PW = Injector pulse width, g/sec.
A/F = Engine computer air-to-fuel ratio.
C2 = Constant, 0.84602 liter/g.
The greatest contribution to the uncertainties of
flow rate is the constantly changing volumetric
efficiency of the engine. Volumetric efficiencies
ranged from 89 to 66 percent, respectively, for low
to high engine flow rates with a correlation
coefficient of 0.77(f). An average volumetric
efficiency of 75 percent was chosen for the
calculations to provide a reasonable median of

-------
80
60
Q.
a. 50
di
!E 40
0
10
20
30
40
50
60
70
80
Dual Analyzer, ppm
Figure 3. Comparison of the dual analyzer technique measured ammonia concentrations to the
gas washing technique measured ammonia concentrations.
140
A

I
Time, seconds
Figure 4. Speed versus time trace for the hard acceleration cycle.
differences across the three flow method
techniques. This data collection effort included
significant loads on the engine, resulting in
enrichment events. Volumetric efficiency is a
function of load, and the results can be seen in
Figure 6. Figure 6 shows how the total test volume
calculated by each of the three methods compares to
the volume of exhaust gas determined by the
Georgia Institute of Technology dynamometer
facility. Figure 6 also shows how the repeated tests
exhibited very similar differences from the
dynamometer facility (i.e., data points are clustered
1

-------
12
10-
X
£
j£
¦o
C/5
3
CL<

. ¦¦ >
• a
¦
v
¦v^
i-	1 1
.4'
I I	I I
-i	1—i	1—i	r
1.0 2.0 3.0 4.0
MAP Sensor, volts
Figure 5. The relationship between injector pulse width and the intake manifold
absolute pressure. The break in the graph is where enrichment combustion conditions
were being commanded by the engine computer.
16.00
12.00
8 00
4.00
4.00
Volume
1.07Cx
volurr.e
1.00
12.00
16.00
Dynamometer Test Volumes, m3
I ~ MAF
I
I RPM
I1"	 i
i A Injector PW .
Figure 6. Comparison of the air volumes determined using the instrumented vehicle
and the air volumes determined by the dynamometer facility.

-------
together by the test cycle operated). Table 1
contains the total volume of exhaust gas as
determined by the dynamometer facility and each of
the three flow rate calculation techniques used for
the test vehicle engine.
Ammonia mass emissions. NH? mass emissions
are the product of NH? measurements
(concentrations) and the air flow rate (volume/time)
calculations. The concentrations for these
dynamometer tests are determined using the dual
analyzer technique, and the mass air How rate was
obtained from the engine computer data stream
(MAFair) on a second-by-second basis, as shown in
Equation (10).
NH3 Emission = (N|Analyzerl — NOxAnalyzer2)
(DR)(MAFair)(C.O	(10)
where:
NH;, Emission = Ammonia emission, g/sec.
^tAnalyzerl- NOxAnalyzer2 — AmmOIlid
concentration, ppm.
DR = Dilution ratio as indicated by two CO2
analyzers.
MAFair = Mass air flow, mole/sec.
C? = Constant, 28.9 g/mole.
The total grains per test value was calculated by
summing the grams per second values for each test,
and the grams per mile was calculated by dividing
the total grams per test by the total distance traveled
during the test. The signal-to-noise ratio for the
calculated NH? concentrations was a concern. The
N, and NOx data were smoothed using rolling
averages, a macro was used to resolve differences
when NOx was greater than N, and additional
smoothing functions were applied to the final NH?
concentrations. Dilution ratio is the ratio between
the CO2 analyzer (concentrations) sampling the raw
exhaust gas and the CO2 analyzer (concentrations)
sampling the gas being delivered to the NH?
analyzers.
Test matrix. The hypothesis was that the test
vehicle would produce NH? only when the exhaust
gas O2 concentrations were low enough to facilitate
reduction chemistry within the catalyst.
Dynamometer test cycles were chosen based upon
two criteria. The first criterion was to collect data
using a test cycle frequently used by other testing
laboratories. The FTP is the most common test,
providing a large data set for comparison, and this
comparison might also provide some insight as to
why NH? has only recently become recognized as a
significant mobile emission. The second criterion
was to collect data using a test cycle that included
significant enrichment opportunities. The goal of
this work was to demonstrate the NH? emission
potential for the test vehicle (passenger light duty
vehicle) and develop a relationship between the
load (percent of time in enrichment) and the amount
of NH? produced. The test vehicle was used to
collect these data while operating through three
transient dynamometer cycles and one steady state
(72 km/hr) cycle. The transient cycles included the
FTP. highway fuel economy, and hard acceleration.
The FTP consists of a cold-start urban driving
cycle, a 10-minute soak, and a hot-start repeat of the
initial 505 seconds. The steady state data collection
began after engine temperature and emissions had
stabilized. The highway fuel economy cycle is
conducted at higher speeds and less acceleration
modes than FTP. The FTP and highway fuel
economy cycles are EPA test cycles described in 40
CFR. The hard acceleration cycle was included
with the Clayton dynamometer software package.
The ability of the catalyst to form NH? was also
considered to be an age factor. The testing included
the original catalyst (with ~ 120k kilometers) and a
new General Motors - Original Equipment
Manufacturer (GM-OEM) catalyst (with ~ 800
kilometers).
RESULTS AND DISCUSSION
The results have shown a correlation between NH?
emissions and enrichment events. Table 2 provides
the percentage of time that the test vehicle was
operating in enrichment conditions and the average
emissions for each of the test cycles and catalyst.
Average values reported in Table 2 have been
calculated by combining the old and new catalyst
test results for each of the cycles tested. Although
emissions of NH? were typically higher with the old
catalyst, not enough tests were made to define a
difference.
Data indicating fuel-rich/rcduced-02 conditions
were analyzed for correlation to average
<1

-------
Cycle
GM OEM 02
Sensor
Enrichment Time,
%
EPA 02 Sensor
Enrichment
Time, %
Test
Emissions,
g/km
Predicted Emissions
[GM/EPA],
g/km
ftpa
6.8
2.5
0.040
0.052 / 0.030
ftpa
6.3
3.7
0.037
0.047/0.040
505a
11.3
6.3
0.035
0.090 / 0.062
Steady StateA
0
0
0.009
0.000 / 0.009
Hwy FuelA
4.8
2.3
0.050
0.034 / 0.029
FTP6
6.1
3.4
0.007
0.045 / 0.037
505b
6.6
2.4
0.027
0.050 / 0.029
Steady State8
0
0
0.011
0.000 / 0.009
Hwy Fuel8
3.0
0.7
0.008
0.019/0.015
Hwy Fuel6
3.4
2.0
0.015
0.022 / 0.025
Hard Acceleration"
22.9
22.4
0.145
0.189/0.197
Hard Acceleration"
18.7
21.7
0.204
0.153/0.191
Hard Acceleration"
1 1.6
13.1
0.186
0.093/0.119
Hard Acceleration"
16
13.1
0.135
0.130/0.119
Average Values
Steady Slate
0
0
0.010
0.000 / 0.009
505
9.0
4.4
0.031
0.070 / 0.046
FTP
6.4
3.2
0.028
0.048 / 0.035
Hwy Fuel
3.7
1.7
0.024
0.025 / 0.021
Hard Acceleration
17.3
17.6
0.168
0.141 / 0.157
"A" denotes the use of the original catalyst with 120k km and "B" denotes the use of a new catalyst.
Table 2. The relationship between percent enrichment and ammonia emissions.
range of 43.8 to 57.4 percent of the time indicating
rich operation. The engine computer did maintain
an indication of rich conditions during the short
periods of significant NH3 production, but the
percentage of time indicating rich conditions was
not sufficient to establish a correlation to average
NH? emissions.
The second engine computer indicator of
enrichment was the GM-OEM O2 sensor voltage.
The sensor voltage ranged between 0 and 1 volt,
with 0 indicating lean conditions and 1 indicating
rich conditions. A graphically determined voltage
was derived and used as an indicator of enrichment.
Any values greater than 0.85 volt were considered
to be rich combustion mixtures. Figure 7 provides
an indication of this sensor's performance while
NH3 concentrations were being measured for one of
the hard acceleration test cycles. Figure 7 shows a
direct correlation between the O2 sensor's
maximum voltage reading and NH3 production. A
regression analysis of percent enrichment and
dynamometer emissions and modal (second by
second) emissions. The average dynamometer
emission analysis was to verify that the production
of on-road NH3 emissions could be predicted, and
modal analysis was to develop a method of
predicting on-road NH3 emissions. NH3 average
emission rate (g/km) per test cycle was regressed
onto the percentage of cycle time that enrichment
conditions existed. Enrichment indicators included
the engine computer data stream, the GM-OEM
sensor volts, and the EPA-installed wide-range O2
sensor.
The engine computer data stream included an
indicator of either "rich or lean" conditions as
calculated by the engine computer using the GM-
OEM Oi sensor. Engine combustion control is
dithered between rich and lean conditions for
drivability and catalyst operation. Hesitation-free
acceleration requires a slightly rich combustion
mixture, and complete oxidation requires a slightly
lean combustion mixture. The rich and lean
indicator reflected this dithered operation with a

-------
Figure 7. Change in oxygen concentration in relationship to the production of ammonia. Also an
indication of the signal-to-noise ratio between the original GM-OEM oxygen sensor and the EPA-
installed oxygen sensors.
Time, seconds
average cycle (g/km) emissions indicated a
correlation coefficient (r2) of 0.835.
The third indication of enrichment was the EPA-
installed wide-range O2 sensor mounted between
the engine and the catalyst. Again, a graphically
determined voltage was derived and used as an
indicator of enrichment. The sensor ranged
between 1.5 and 4.5 volts with 4.5 indicating lean
conditions and 1.5 indicating rich conditions. Any
values less than 2.8 volts were considered to be rich
combustion mixtures. Figure 7 provides an
indication of this sensor's performance while NH-?
concentrations were being measured for one of the
hard accelerations test cycles. Figure 7 also
provides a graphical indication of the signal-to-
noise ratio and response rate for the two O2 sensors.
The GM-OEM sensor was faster to respond to
changes in 0? concentrations and the signal-to-noise
ratio was significantly greater than the EPA-
installed wide-range O2 sensor.
Figure 8 shows the average test cycle percent
enrichment regressed onto average cycle NH3
emissions (g/km). These results show a correlation
coefficient (r) of 0.919 and also show that the
emissions with no enrichment are 0.009 g/km when
averaging across both the GM-OEM and the EPA-
installed percent enrichment data. Table 2 contains
the predicted emissions for both the EPA and GM-
OEM O2 sensors. These predicted emissions show
that NH3 production is inversely proportional to 02
concentration. When evaluating data that contain
both enrichment and no enrichment, the emissions
were found to be bimodally distributed for O2
concentration. Figures 9 and 10 show this bimodal
distribution for all of the tests and one of the hard
acceleration cycles, respectively. These graphs
represent average test cycle percent enrichment of
2.4 and 13.1, respectively. Modal emissions were
analyzed using the EPA-installed wide-range 02
sensor as the indicator of rich combustion mixtures.
Modal NH3 emission analysis required time
alignment for sample system volumes, analyzer
analysis, and the response time of the different
channels within the chemiluminescent analyzers.
Even after taking these delays into account, the
I*

-------
0.250
0	5	10	15	20	25
Percent of Time in Enrichment, %
Figure 8. Percentage of time that the engine was operated in an enrichment mode as
indicated by both the EPA and GM-OEM oxygen sensors.
Ill .06 -
U .US
o

V)
c
O 0.03 -
w
x
E
m
0.04 -
0.02 -
0.01 -
-0.00
1.5
:*rsV
2.5 3.0
Oxygen Sensor, volts
4.0
Figure 9. The relationship of ammonia emisions for oxygen concentration, reflecting
results from all 14 cycles tested.

-------
0.06
0.05
CJ
-r.
\ 0.04
ir.
E
C
1 0.03
c
"2- 0.02
Z
0.01
0.00
1.5 2.0 2.5 3.0 3.5 4.0 4.5
Oxygen Sensor, volts	j
!	I
L._			______			J
Figure 10. The relationship of ammonia emissions for oxygen concentration,
reflecting data from only one of the hard acceleration cycles tested.
modal analysis indicated a third delay that is
thought to be associated with the reduction
chemistry reactions within the catalyst. The delay
between rich combustion conditions within the
engine and NHi emissions is estimated to be 14
seconds or 0.4 km later at 96 km/hr. Second-by-
second analysis of NH;, emissions shows that NH;,
emissions are well correlated to reduction in Ot
concentrations. Figure 11 shows the percent
enrichment regressed onto total NH-? (grams)
measured for each of the 14 cycles tested for both
the EPA-installed and the GM-OEM 02 sensors.
Previous instrumented vehicle emissions analysis
has shown acceptable correlation to mass air flow
rate (MAF) through the engine. Regression
equations were developed for both all data and hard
acceleration test cycles, and sorted for enrichment
and non-enrichment events. Neither analysis
provided acceptable confidence, but on average the
enrichment data showed base line emissions at
0.006 g/km and a slope of 0.00074[MAF
(mole/sec)]. These results have shown that
predicting NH3 emissions must include the
prediction of enrichment events and that O?
concentration is a direct indication of when
enrichment is in effect.
CONCLUSIONS
An automotive exhaust gas NH3 measurement
. technique was demonstrated, and significant NH;,
emissions have been shown to correlate to fuel
enrichment events. This supports the original
hypothesis that significant NH-? emissions will
occur only when insufficient oxygen at the catalyst
allows reduction chemistry to predominate within
the catalyst.
The dual chemiluminescent analyzer technique
showed acceptable agreement with the exhaust gas
concentrations measured by the gas washing
technique. The total NH3 mass emissions for the
standard FTP cycle agreed with work completed by
other researchers. Modal data analysis shows that

-------
be predicted using percentage of time of
enrichment. Predicting the percentage of time of
enrichment is the subject of on-going EPA research.
The average base line emissions with no work-
induced enrichment events (e.g., steady state) are
not significantly different from previous data and on
average range from 0.009 to 0.011 g/km. Fuel
enrichment events resulted in NH-? emissions that on
average ranged from 0.135 to 0.204 g/km. Two
tests of emissions from tunnels have been
performed to address fleetwide NH-? emissions from
vehicles. The San Francisco Bay area tunnel16
results indicated an adjusted NH3 emission factor of
0.05 g/km. The Sherman Way tunnel study1
indicated an adjusted NH-? emission factor of 0.06
g/km. Tunnel studies measure the change in air
concentrations, and these results have been adjusted
to g/km units by estimating a fleet economy of 23
miles per gallon. The vehicle reported in this paper
gets 30 miles per gallon. Given the uncertainty in
fuel consumption and the percent enrichment of the
tunnel study fleet, these results are in reasonable
agreement with the results reported in this paper.
Data presented here were collected using only one
test vehicle, but the engine control strategy and
catalyst design are consistent for a large percentage
of on-road vehicles. It is therefore reasonable to
believe that the results presented here can be
applied to many of the on-road vehicles today.
25
Q.
~ GM-OEM 02 Sensor
¦ EPA 02 Sensor
0
0.5
1
1.5
2
2.5
Total Ammonia per Test, grams
Figure 11. Percentage of enrichment per test and the total grams of ammonia produced.
\4

-------
REFERENCES
(1)	Fraser, Matthew P.; Cass, Glen R.,
Detection of Excess Ammonia
Emissions from In-Use Vehicles and the
Implications for Fine Particle Control,
Environmental Science & Technology,
Volume 32, Number 8, pages 1053-
1057, 1998
(2)	Klingenberg, H.; Schurmann, D.,
Unregulated Motor Vehicle Exhaust Gas
Components, Volkswagon AG report,
Wolfsburg, Germany, 1989
(3)	Shores, Richard C., Proceedings: 10th
CRC On-Road Vehicle Emissions
Workshop, Presentation Visuals:
Measurements of Ammonia Emissions
from EPA's Instrumented Vehicle,
March 27-29, 2000
(4)	Aneja, Viney P.; Murray, George;
Southerland, James, Proceedings:
Workshop on Atmospheric Nitrogen
Compounds If, 1999
(5)	Walker, John T., Atmospheric Transport
and Wet Deposition of Ammonia in
North Carolina, Atmospheric
Environment, Volume 34, pages 3407-
3418,2000
(6)	Teets, Mary K., Highway Statistics
Summary to 1995, U.S. Department of
Transportation, Report Number FHWA-
PL-97-009, Washington, DC, 1997
(7)	North Carolina Division of Motor
Vehicles registration data base, Raleigh,
NC. 1999
(8)	American Automobile Manufacturers
Association, Motor Vehicle Facts &
Figures, Washington, DC, 1997
(9)	Shores, Richard C., Proceedings: 9lh
CRC On-Road Vehicle Emissions
Workshop, Presentation Visuals:
Comparison of Dynamometer and
Remote Sensing Emissions
Measurement of the APPCD
Instrumented Test Vehicle, April 19-21,
1999
(10)	Methods of Air Sampling and Analysis,
Second Edition, Method Number 401,
Tentative Method of Analysis for
Ammonia in the Atmosphere
(Indophenol Method), American Public
Health Association, Washington, DC,
1977
(11)	Methods of Air Sampling and Analysis,
Second Edition, Method Number 801,
Analytical Method for Ammonia in Air,
American Public Health Association,
Washington, DC, 1977
(12)	Thermo Environmental Instruments Inc.,
Model 17C, Instruction manual,
Franklin, MA. 1997
(13)	Obert, Edward F.,Internal Combustion
Engines, International Textbook
Company, Scranton, PA, 1965
(14)	Hey wood, John B., Internal Combustion
Engine Fundamentals, McGraw-Hill,
Inc., New York, NY, 1988
(15)	Obert, Edward F., Internal Combustion
Engines and Air Pollution, Harper &
Row, Publishers, New York, NY, 1973
(16)	Kean. Andrew J.; Harley, Robert A., On-
Road Measurement of Ammonia and
Other Motor Vehicle Exhaust Emissions,
Environmental Science and Technology,
Volume 34, pages 3535-3539, 2000

-------
TECHNICAL REPORT DATA
NRMRL-RTP-P-6 2 3 (Please read Instructions on the reverse before completin
1. REPORT NO
EPA/600/A-01/116
2.
3. RECIPI
4. TITLE AND SUBTITLE
Ammonia Emissions from the EPA's Light Duty Test
5. REPORT DATE
Vehicle


6. PERFORMING ORGANIZATION CODE
7.authors r.c.Shores and J.T.Walker Jr. (EPA); M,0.
Rodgers and J.R.Pearson (GIT); and R.B.McCullough (NC)
8. PERFORMING ORGANIZATION REPORT NO
9 PERFORMING ORGANIZATION NAME AND ADDRESS
Air Quality Laboratory North Carolina Division of
10. PROGRAM ELEMENT NO.
Georgia Institute of Air Quality
Technology Raleigh, NC 27699
Atlanta, GA 30332
11. CONTRACT/GRANT NO
CR828285 (GA Tech)
12. SPONSORING AGENCY NAME AND ADDRESS
U. S. EPA, Office of Research and Development
Air Pollution Prevention and Control Division
Research Triangle Park, North Carolina 27711
13. TYPE OF REPORT AND PERIOD COVERED
Published paper; 1999-2000
14. SPONSORING AGENCY CODE
EPA/600/13
is. supplementary notes ^ppQ) project officer is Richard C. Shores, Mail Drop 61, 919/541-4983
Presented at Int. Fall Fuels and Lubricants Meeting and Exhibition, San Antonio, TX,
9/14-27-01.
16.abstract paper discusses measurements of arrmonia (NH3) emissions from EPA's light-
duty test vehicle while operated on a dynamometer. The vehicle's (1993 Chevrolet equip-
ped with a three-way catalyst) emissions were measured for three transient (urban dri-
ving, highway fuel economy, and hard acceleration) cycles and steady state operation.
Previous research showed that NH3 is predominantly emitted from vehicles with a catalyst
(three-way or dual-bed). The normal operation of the vehicle's catalyst is to reduce
nitrogen oxides (NOx) to nitrogen (N2) and oxygen (02). The reduction of NOx to NH3
would have to occur during periods of operation when insufficient 02 is available. NH3
emissions were measured during fuel-rich/reduced-02 conditions (open-loop control
scheme), and the results indicated that NH3 production is correlated to combustion con-
ditions. The results also indicated that the amount of NH3 produced correlates with the
amount of time that the vehicle remains in the open-loop control scheme. The signifi-
cance of this finding is that NH3 production can be predicted for a fleet based on the
frequency of enrichment of vehicles equipped with a three-way catalyst. The results also
provide a way to determine the location of roadway links and/or specific locations
where NH3 production can be anticipated based on predicted engine power.
17
KEY WORDS AND DOCUMENT ANALYSIS


a. DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Pollution
Mobile Equipment
Ammonia
Dynamometers
Catalytic Converters
Internal Combustion Engines
Pollution Control
Test Vehicles
13B
15E
07B
14B
07A, 131
2 IK
18. DISTRIBUTION STATEMENT

19. SECURITY CLASS (Tft/s Report)
21. NO. OF PAGES


20. SECURITY CLASS (This Page)
22. PRICE
EPA Form 2220-1 {Rev, 4-77 ) PREVIOUS EDITION IS OBSOLETE	forms/admin/tectirp! frm 7/8/99 pad

-------