United States
Environmental Protection
Agency
Air and Energy Engineering
Research Laboratory
Research Triangle Park, NC 27711
Research and Development
«rEPA Project Summary
EPA/600/SR-94/059 July 1994
Analysis of Real-time Vehicle
Hydrocarbon Emissions Data
J. Philip Childress and James H. Wilson, Jr.
Analyses using real-time dynamom-
eter test emissions data from 13 pas-
senger cars were performed in a study
to examine variations in emissions dur-
ing different speeds or modes of travel.
The resulting data provided a means
for separately identifying idle, cruise,
acceleration, and deceleration emis-
sions for examining how emissions dif-
fer by vehicle speed during cruise
mode.
To select a set of vehicles for the
study, the hydrocarbon/time relation-
ship was established for several ve-
hicles operating on summer-grade base
fuel. Federal Test Procedure (FTP) re-
sults were then produced and exam-
ined to identify normal emitters (clean
vehicles). After these vehicles were se-
lected, an intensive analysis of their
second-by-second emission character-
istics was conducted.
The FTP runs for cold start:, hot start,
and hot stabilized emissions (Bags 1,
2, and 3 of the FTP) were performed for
each of the four driving cycles—accel-
eration, deceleration, idling, and cruise
—and the fraction of overall emissions
contributed by each mode was com-
puted for the warmed-up portion of the
driving cycle. A protocol was then de-
veloped for review of the FTP real-time
data.
The study results showed significant
emissions differences related to travel
mode: (1) cruise mode emissions are
invariant with speed when expressed
on a grams-per-second basis; (2) emis-
sions resulting from acceleration from
a stop to cruise speed are similar to
those resulting from acceleration from
cruise speed to a higher speed; (3)
acceleration emissions were the high-
est of all the modes; and (4) cruise
emissions are very similar to idle emis-
sions.
This Project Summary was developed
by EPA's Air and Energy Engineering
Research Laboratory, Research Tri-
angle Park, NC, to announce key find-
ings of the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).
Introduction
The purpose of this project was to in-
vestigate whether it is feasible to develop
new motor vehicle emission inventory pro-
cedures using modal (second-by-second)
data, either exclusively or as a supple-
ment to data that are more routinely col-
lected. One issue related to the analysis
is the potential for using these data to
develop a future motor vehicle emissions
model. At a minimum, modal emission
data hold promise for validating motor ve-
hicle emission factors.
Background
This summary presents the results of
an analysis performed to examine varia-
tions in emissions during different speeds,
or modes of travel, using real-time dyna-
mometer test emissions data from 13 pas-
senger cars. The primary data sets used
for the project, developed by one of the
EPA's research laboratories, included
emission measurements from 1986
through 1990 model year vehicles with
accumulated mileage between 17,000 and
55,000 miles.
Printed on Recycled Paper
-------
Modal emission data, such as the data
sets examined in this study, provide a
means for separately identifying idle,
cruise, acceleration, and deceleration
emissions. These data can also be used
to study how emissions differ by vehicle
speed during cruise mode, which is im-
portant when trying to determine ways in
which motor vehicle emissions modeling
can be made simpler (such as assuming
that cruise emissions per unit time are
constant with speed).
It is important to note that FTP mea-
surements are typically made in phases,
or bags, and the results are expressed as
averages over the measurement period.
The FTP begins with a cold start, and Bag
1 represents the first 505 seconds of ve-
hicle operation. Seconds 506 through 1372
of the FTP are known as Bag 2 and rep-
resent the hot stabilized phase of the cycle.
Following Bag 2 is a 10-minute engine-off
period. Then, there is a hot start and the
vehicle is operated using the same speed/
time trace as in Bag 1. This phase is
known as Bag 3.
Procedure
The analysis described in this summary
is based on work performed in two phases.
Phase 1 focused on examining the hydro-
carbonAime relationship for several ve-
hicles operating on summer-grade base
fuel. Overall FTP results were produced
and examined to identify vehicles that had
emissions that were considered normal
(or clean); these were used for further
analysis. For this subset of clean vehicles,
hydrocarbon (HCytime traces were devel-
oped for the long arterial road cruise sec-
tion, also known as Hill 11 of the FTP.
Phase 2 of this study involved intensive
analysis of the second-by-second emis-
sion characteristics of the selected nor-
mal-emitting vehicle models. First, the re-
producibility of emission patterns was ex-
amined for particular acceleration, decel-
eration, and cruise modes. Second, the
fraction of overall emissions contributed
to by acceleration, deceleration, idling, and
cruise modes was computed for the
warmed-up portion of the driving cycle for
these modes. Finally, a protocol was de-
veloped for review of FTP real-time data.
Methodology
Phase 1:
Real-time regulated emission and fuel
economy data were acquired and archived
for 229 out of the 273 FTPs run during the
period between January 1989 and Janu-
ary 1990. The real-time data included FTPs
on 20 passenger cars using 13 fuels (or
fuel blends) at five test temperatures, which
yielded a representative cross section of
emission levels.
Table 1 lists the vehicles for which emis-
sion measurements were used in this
study. [Note: Because vehicle CO174G
only had two FTP runs, it was not in-
cluded in the data set that was analyzed
for this study.] While measurements were
made for the study vehicles while they
were operating on a number of different
fuels, only vehicles operating on summer-
grade gasoline were included in this analy-
sis. Table 2 shows the bag-specific re-
sults for each of the vehicle/run combina-
tions where the fuel was summer-grade
gasoline. While this table is useful by it-
self for determining the emission charac-
teristics of the vehicles tested, these data
were used primarily to identify normal emit-
ters, where a normal emitter emits less
than twice the applicable standard. The
applicable standards for the model year
vehicles tested are: 0.41 g/mi HC; 3.4 g/
mi carbon monoxide (CO); and 1.0 g/mi
nitrogen oxides (NOX).
Table 2 lists selected Bag 1, Bag 2, and
Bag 3 emissions for the 80 FTP runs
performed using summer-grade gasoline.
From this data set, 47 runs were identified
as being normal emitters. It is interesting
to note that of the 13 vehicles tested, 7
were always normal emitters, 1 was never
a normal emitter, and the rest were nor-
mal emitters on some runs but not on
others. It is possible that the variation in
emissions between runs was caused by
the temperature differences among tests
and driver variability.
The FTP results were used to select
two clean cars: vehicles SO756B and
LS612B. The test data from these two
cars were analyzed by plotting HC emis-
sions in parts per million (ppm) versus
Table 1. Real-Time Data Test Vehicles*
vehicle speed (mph). One observation
made from these plots was that the spikes
in HC emissions relate to changes in speed
(i.e., accelerations produce the highest
emissions but cruise did not produce
changes). It is also notable that accelera-
tions from a cruise speed to a higher
speed appear to be as important as ac-
celerations from a stop in producing high
HC emission values. Some runs show
these phenomena more clearly than oth-
ers, however.
Figure 1 shows the lagged 2 second
acceleration (L2ACC) and instaneous hy-
drocarbon emission (HOTFID) variable
time series for the best run of vehicle
LS612B. For this vehicle, as well as for
most of the "well-behaved" vehicles, the
graphs of HOTFID and L2ACC show
spikes in the first few seconds of positive
acceleration, though they become more
erratic and less prominent during other
times (modes). Figure 2 plots the instan-
taneous HC emission rate (on a per-sec-
ond basis) versus vehicle speed during
cruise mode for vehicle SO756B. As the
figure shows, most of the higher emission
values were observed at low speeds,
though the data seem to support the hy-
pothesis that cruise mode emissions do
not change much as speed increases or
decreases. Other data collected in this
analysis support the conclusion that cruise
mode emissions are invariant with speed
as well.
Phase 2:
The analyses performed for this effort
were divided into three parts:
1. Standard FTP analysis for the cold
start, hot start, and hot stabilized
phases (Bags 1, 2, and 3) of the
FTP
Vehicle
ID
CA365B
CO174G
CO322G
CO665W
CO710B
CV924W
ES707R
LA127B
LA392W
LS612B
SA333B .
3O7S6B
TA207G
Engine
displacement
Vehicle
GM Chev. Caprice Classic
GM Corsica
GM Corsica
GM Corsica
GM Corsica
Ford Crown Victoria
Ford Escort
Chrysler LeBaron
Chrysler LeBaron
GM Buick LeSabre
Ford Mercury Sable
GM Buick Somerset
Ford Taurus
(liters)
5.0
N/A
2.0
2.0
2.8
5.0
1.9
2.5
3.0
3.8
3.0
2.5
3.8
Accum.
VIN
1G1BN69H9GY1 00365
N/A
1G1LT5116HY1 02322
1G1LT5111JY616665
1G1LT51W9HY1 04710
2FABP73F8HX1 83924
1FAPP2599HW328701
1C3CJ41KOJG324127
1C3XJ4538LG418392
1G4HP14C6JH482612
1 MEBN5048HA6 15333
1 G4NM14 V7HM078756
1FABP524XJA 148207
Model
mileage
39,970
N/A
34,364
16,935
34,268
39,242
44,559
36,418
20,087
54,802
44,360
45,136
20,465
Year
1986
1987
1987
1988
1987
1987
1987
1988
1990
1988
1987
1987
1988
Includes gasoline-fueled vehicles only. Cars listed above are those tested with summer-grade gasoline.
N/A = Data not available.
-------
Figure 1. HC emissions and lagged acceleration (Bag 2:506-1006 sec; vehicle: LS612B, run: 31116).
Q.
o
o
T3
C
CO
Q.
Q.
Q
O
-30
-40
-50
500
600
700 800
Time (sec)
900
1000
Figure 2. Cruise mode speed vs. HC emissions (Bags 2 and 3 only; vehicle: S0756B).
5 000
8
-CO
1? 4 000
c 3,000
o
'to
-------
Table 2. Weighted Bag 1, Bag 2, and Bag 3 Emissions (grams per mile)
Run
Vehicle ID number HC NOX
CA365B
CO174G
CO322G
CO665W
CO710B
«•>
CV924W
ES707R
LA127B
LA392W
31065
31061
31060
31057
31063
31062
31064
31034
31036
30936
30974
30978
30972
30948
30945
30946
30934
30975
30973
30971
30935
30989
30854
30870
30869
30853
30861
30862
30992
30999
31002
30996
30994
30997
30993
31001
30995
30990
31114
31107
31106
31115
31113
31020
31025
31024
31019
31026
31021
31045
31042
31043
31044
31041
31046
31122
31123
31124
1.47
1.47
1.33
1.28
1.69
1.39
1.77
0.22
0.21
0.42
0.43
0.63
0.42
0.38
0.36
0.38
0.34
0.45
0.42
0.43
0.32
0.26
0.33
0.58
0.53
0.56
0.25
0.27
0.95
0.85
0.99
0.88
0.92
0.79
1.10
0.90
0.84
1.15
0.82
0.91
0.86
0.95
0.82
0.52
0.46
0.51
0.58
0.26
0.54
0.20
0.26
0.22
0.19
0.26
0.22
0.59
0.64
0.52
1.13
0.89
0.83
0.76
1.06
0.99
1.18
0.26
0.28
0.63
1.00
2.11
0.67
0.53
0.63
0.57
0.62
0.90
0.69
0.73
0.69
0.60
0.37
0.47
0.52
0.41
0.30
0.41
0.43
0.41
0.37
0.37
0.36
0.43
0.47
0.42
0.38
0.42
0.95
0.85
0.87
1.01
0.91
0.72
0.91
0.77
0.74
0.59
0.80
0.81
0.57
0.71
0.78
0.63
0.50
1.05
1.06
1.29
CO
15.32
15.80
13.42
12.75
19.94
15.16
25.17
0.69
1.06
8.23
10.30
15.90
8.74
6.54
6.37
7.71
5.61
11.15
8.73
8.80
5.51
7.79
5.31
11.36
10.34
4.86
4.76
4.32
6.63
6.23
9.45
6.06
6.35
6.01
6.87
7.13
5.93
8.39
2.70
2.85
2.79
3.09
2.78
11.16
12.30
13.42
12.82
5.57
12.27
4.09
3.80
3.17
3.24
3.77
3.14
4.85
4.46
4.82
(Continued)
-------
1
Table 2. (Continued)
Run
Vehicle ID Number
LS612B 31117
31116
31119
31118
31120
31121
CV924W 31114
31107
31106
31115
31113
ES707R 31020
31025
31024
31019
31026
31021
LA127B 31045
31042
31043
31044
31041
31046
LA392W 31122
31123
31124
LS612B 31117
31116
31119
31118
31120
31121
SA333B 31035
31037
31038
31040
31032
31039
SO756B 31017
31015
31013
31018
31016
31014
TA207G 31050
31047
31048
31049
NOTES: Bold areas indicate normal emitters.
2. Modal analyses of the four driving
modes within Bags 2 and 3; and
3. Statistical analysis of second-by-
second HC emissions.
These analyses focused on a subset of
12 vehicles, representing 76 FTP runs,
selected for data robustness (at least 4
FTP runs each), fuel (summer-grade gaso-
HC NOX CO
0.31 0.21 2.26
0.32 0.18 2.19
0.28 0.25 2.12
0.22 0.18 1.88
0.31 0.27 2.71
0.29 0.26 2.55
0.82 0.95 2.70
0.91 0.85 2.85
0.86 0.87 2.79
0.95 1.01 3.09
0.82 0.91 2.78
0.52 0.72 11.16
0.46 0.91 12.30
0.51 0.77 13.42
0.58 0.74 12.82
0.26 0.59 5.57
0.54 0.80 12.27
0.20 0.81 4.09
0.26 0.57 3.80
0.22 0.71 3.17
0.19 0.78 3.24
0.26 0.63 3.77
0.22 0.50 3.14
0.59 1.05 4.85
0.64 1.06 4.46
0.52 1.29 4.82
0.31 0.21 2.26
0.32 0.18 2.19
0.28 0.25 2.12
0.22 0.18 1.88
0.31 0.27 2.71
0.29 0.26 2.55
0.39 1.16 4.42
0.43 1.23 4.89
0.41 1.22 4.77
0.51 1.49 6.31
0.39 1.09 4.91
0.41 1.52 6.13
0.21 1.06 2.13
0.19 1.04 2.00
0.22 0.84 2.11
0.20 1.03 1.95
0.21 0.98 1.88
0.24 0.81 2.19
0.51 0.88 3.04
0.49 0.75 2.94
0.43 0.84 2.85
0.39 0.79 2.27
line only), and emissions behavior (show-
ing average emissions in grams per mile
of not more than twice the applicable Fed-
eral emission standard of any regulated
pollutant).
Modal analysis by FTP phase for Bags
1 , 2, and 3 wase performed by assigning
each second of the FTP to one of four
modes (idle, acceleration, cruise, or de-
celeration). Modes were taken from an Ur-
ban Dynamometer Driving Schedule Mode
table. As a result, 81 distinct modes were
identified over the complete driving cycle.
In the modal data analysis, HC emis-
sions in grams were calculated for each of
the modes of the FTP runs for the 76 FTP
cases selected. In modal analysis, the con-
cept of grams per mile is meaningless,
since speed, or miles traveled per sec-
ond, is used as a denominator; thus, grams
per mile would be infinite in idle mode. To
overcome this problem, a normalized mea-
sure of relative emissions across driving
modes was developed. HC emissions were
then analyzed for each of the vehicle FTP
tests for the 12 vehicles studied.
Bag 3 results were selected for display
because Bags 1 and 3 appear to have
higher speeds than Bag 2, and there was
concern that Bag 1 results might be bi-
ased by emissions in the first 100 sec-
onds of the test (when the engine is cold).
In general, emission rates for Bag 3 (in
milligrams per second) are highest during
acceleration and lowest during decelera-
tion. Cruise emission rates appear to be
slightly higher than those during the idle
mode. Results for vehicle TA207G appear
to be anomalous, since idle emission rates
for that vehicle are higher than its emis-
sions in any other mode; these results are
consistent for all four test runs.
A comparison of idle emission rates
among the three bags showed that, for
most cars and FTP runs, the majority of
the idle emissions are in the cold start
mode. Of the 41 simulations with normal
emission rates, the Bag 1 gram-per-sec-
ond emission rates at idle are 19 times
the Bag 2 rates, and 5 times the Bag 3
rates. Bag 3 idle emission rates are about
twice the average Bag 2 idle emission
rates.
Results
Speed Versus Emissions
The hypothesis that there is only slight
correlation between instantaneous re-
corded vehicle speed (in miles per hour)
and HC emissions rate (in grams per mile)
was investigated. HC emission rate was
calculated on a per-second basis, for well-
behaved gasoline vehicles operating in
the cruise portions of the FTP cycle. Three
FTP runs per vehicle were chosen, result-
ing in a total of 18 FTP runs.
Only Bag 2 (seconds 506-1372) and
Bag 3 (seconds 1973-2477) were used
for this analysis to eliminate the potential
cold start biases of Bag 1. Bag 3 is of
interest because it contains highway
speeds in excess of 50 mph. Hydrocar-
bon emissions in grams per second were
-------
calculated using a VHB, number calculated
from the data base tor each FTP run,
representing cubic feet of total diluted mix
passing per second (e.g., 10.3).
No discernable correlation was found
between speed and the per-second HC
grams-per-mile rate. The Pearson corre-
lation coefficients, -0.12 to -0.25 for the
six vehicles, are not sufficient to indicate
that a more sophisticated model could be
developed to test the hypothesis.
FTP Run Statistical Analysis
Statistical analysis was also performed
for 76 FTP runs covering 12 gasoline-
fueled vehicles. The purpose of the analy-
sis was to see if there appeared to be a
reasonable model to use for estimating
instantaneous HC emissions (HOTFID)
using speed, acceleration (taken to be the
first difference of speed), or lagged repre-
sentations thereof. A delta HOTFID de-
pendent variable was also developed, al-
though it did not show much improvement
in correlation. Only simple correlations
were performed — not model regression
or factor analyses.
The highest simple correlation seemed
to be between the lagged (2 second) ac-
celeration, the variable L2ACC, and
HOTFID. This correlation is not consistent
across vehicles, however. Nevertheless,
8 out of the 12 vehicles had correlation
coefficients greater than 0.3, showing that
there is a relationship between the vari-
ables (though probably a weak one).
The relationship between speed and HC
emissions was further investigated for the
normal emitters in the data set. This analy-
sis was restricted to cruise mode emis-
sions to allow full comparison of emission
results, expressed on a grams-per-sec-
ond basis.
Conclusions
Cruise mode emissions are invariant
with speed when expressed on a grams-
per-second basis. Accelerations produce
the highest emissions. Accelerations from
a cruise speed to a higher speed appear
to be as important as accelerations from a
stop in producing high HC emission val-
ues. In general, emission rates for Bag 3
(in grams per second) are highest during
acceleration and lowest during decelera-
tion. Cruise emission rates appear to be
nearly the same as those during the idle
mode.
-------
-------
J.P. Childress and J.H. Wilson, Jr., are with E.H. Pechan and Associates, Inc.,
Springfield, VA 22151
Carl T. Rlpbergeris the EPA Project Officer (see below).
The complete report consists of paper copy and four diskettes.
Paper Copy—(Order No. PB94-180494; Cost: $27.00, subject to change)
Diskettes—(Order No. PB94-501780; Cost: $140.00, subject to change-includes
paper copy-wilt be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Air and Energy Engineering Research Laboratory
US, Environmental Protection Agency
Research Triangle Park, NC 27711
United States
Environmental Protection Agency
Center for Environmental Research Information
Cincinnati, OH 45268
Official Business
Penalty for Private Use $300
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
EPA/600/SR-94/059
------- |