United States
Environmental Protection
Agency
Air and Energy Engineering
Research Laboratory
Research Triangle Park NC 27711
Research and Development
EPA/600/S8-91/032 July 1991
vvEPA Project Summary
MOBILE4 Sensitivity Analysis
Mark G. Smith and Terry T. Wilson
The purpose of this analysis is to iden-
tify the MOBILE4 input variables that can
have significant impacts on highway ve-
hicle emissions inventories and to de-
velop priorities for the development of
improved guidance for specifying MO-
BILE4 inputs. Two major factors are con-
sidered: (1) the likelihood and potential
range of variability in values for each
MOBILE4 input; and (2) the potential
magnitude of the effect of these varia-
tions, in terms of impact on typical mobile
source inventories. This exercise uses
ranges that illustrate: misinterpretation
of guidance; use of MOBILE4 defaults
where local conditions are significantly
different; and use of assumptions about
local conditions which may not accu-
rately reflect actual conditions.
This Project Summary was developed
by EPA's Air and Energy Engineering Re-
search Laboratory, Research Triangle
Park, NC, to announce key findings of the
research project that is fully documented
in a separate report of the same title (see
Project Report ordering information at
back).
Introduction
This analysis updates previous work
based on MOBILES by using MOBILE4 for
the sensitivity analysis and by adding new
MOBILE4 variables. The approach used in
previous work is modified to address the
specific concerns of this project (the state
implementation plan [SIP] and National
Emissions Data System [NEDS] inven-
tory/guidance context). An additional level of
detail is included for two critical variables
(speed and temperature). Sensitivity to basic
inspection/maintenance (I/M) program spec-
ifications (waiver and compliance rate) is
also considered.
The primary sensitivity analysis presented
here is structured around two base cases
representing ozone and carbon monoxide
(CO) season conditions. Table 1 summa-
rizes the protocol for the primary sensitivity
analysis, including all relevant MOBILE4
input variables. In general, the base cases
and primary ranges were chosen to be par-
allel to the previous sensitivity analysis and
to relate to specific cities or national aver-
ages from available work by the EPA Office
of Mobile Sources (OMS). The pollutants,
regions, and calendar years were chosen to
cover the areas, periods, and pollutants of
interest in SIP inventories and other typical
inventory applications. To provide additional
data about the importance of accuracy in
estimating key variables, secondary sensi-
tivity ranges were tested around the base
cases as well as around the primary speed
and temperature ranges specified in the pro-
tocol.
MOBILE4 Input Values
for the Base Cases
Emissions of hydrocarbons (HC), nitrogen
"oxides (NOx), and CO were analyzed for the
summer ozone season situation. Only CO
was considered for the winter CO season
base case. Both low and high altitude situa-
tions were included. The years 1990 and
2005 were selected as typical base and pro-
jection years for SIP inventories, to provide
some perspective on the relative importance
of the individual variables over time. Both
base cases and all sensitivity runs include a
typical basic I/M program. The individual pro-
gram specifications were selected to be typ-
ical of current programs.
MOBILE4 Input Values
for Sensitivity Analysis
Table 1 lists the MOBILE4 input values for
the ozone and CO season ranges used in
the primary sensitivity analysis, which are
described below. The low speed of 7.1 mph
is from the New York City cycle, a testing
protocol that represents driving in highly con-
Printed on Recycled Paper
-------
United States
Environmental Protection
Agency
Air and Energy Engineering
Research Laboratory
Research Triangle Park NC 27711
Research and Development
EPA/600/S8-91/032 July 1991
vxEPA Project Summary
MOBILE4 Sensitivity Analysis
Mark G. Smith and Terry T. Wilson
The purpose of this analysis is to iden-
tify the MOBILE4 input variables that can
have significant impacts on highway ve-
hicle emissions inventories and to de-
velop priorities for the development of
improved guidance for specifying MO-
BILE4 inputs. Two major factors are con-
sidered: (1) the likelihood and potential
range of variability in values for each
MOBILE4 input; and (2) the potential
magnitude of the effect of these varia-
tions, in terms of impact on typical mobile
source inventories. This exercise uses
ranges that illustrate: misinterpretation
of guidance; use of MOBILE4 defaults
where local conditions are significantly
different; and use of assumptions about
local conditions which may not accu-
rately reflect actual conditions.
This Project Summary was developed
by EPA's Air and Energy Engineering Re-
search Laboratory, Research Triangle
Park, NC, to announce key findings of the
research project that Is fully documented
in a separate report of the same title (see
Project Report ordering information at
back).
Introduction
This analysis updates previous work
based on MOBILES by using MOBILE4 for
the sensitivity analysis and by adding new
MOBILE4 variables. The approach used in
previous work is modified to address the
specific concerns of this project (the state
implementation plan [SIP] and National
Emissions Data System [NEDS] inven-
tory/guidance context). An additional level of
detail is included for two critical variables
(speed and temperature). Sensitivity to basic
inspection/maintenance (I/M) program spec-
ifications (waiver and compliance rate) is
also considered.
The primary sensitivity analysis presented
here is structured around two base cases
representing ozone and carbon monoxide
(CO) season conditions. Table 1 summa-
rizes the protocol for the primary sensitivity
analysis, including all relevant MOBILE4
input variables. In general, the base cases
and primary ranges were chosen to be par-
allel to the previous sensitivity analysis and
to relate to specific cities or national aver-
ages from available work by the EPA Office
of Mobile Sources (OMS). The pollutants,
regions, and calendar years were chosen to
cover the areas, periods, and pollutants of
interest in SIP inventories and other typical
inventory applications. To provide additional
data about the importance of accuracy in
estimating key variables, secondary sensi-
tivity ranges were tested around the base
cases as well as around the primary speed
and temperature ranges specified in the pro-
tocol.
MOBILE4 Input Values
for the Base Cases
Emissions of hydrocarbons (HC), nitrogen
"oxides (NOx), and CO were analyzed for the
summer ozone season situation. Only CO
was considered for the winter CO season
base case. Both low and high altitude situa-
tions were included. The years 1990 and
2005 were selected as typical base and pro-
jection years for SIP inventories, to provide
some perspective on the relative importance
of the individual variables over time. Both
base cases and all sensitivity runs include a
typical basic I/M program. The individual pro-
gram specifications were selected to be typ-
ical of current programs.
MOBILE4 Input Values
for Sensitivity Analysis
Table 1 lists the MOBILE4 input values for
the ozone and CO season ranges used in
the primary sensitivity analysis, which are
described below. The low speed of 7.1 mph
is from the New York City cycle, a testing
protocol that represents driving in highly con-
Printed on Recycled Paper
-------
Table 1. Protocol for MOBILE4 Sensitivity Analysis
Pollutant
Region
Calendar Year
Avg. Speed (mph) b
Avg. Temperature (°F)C
Hot/Cold Start*
VMT Mix and Mileage Accumulation
Vehicle Age Distrib.
ASTM Class
Diurnal Temp. (°F)
Base RVP (psi)
In-use RVP (psi)
I/M Program f
Compliance
Waivers
Base Case 1:
Ozone Season
HC, CO, A/Ox
Low + high altitude
1990, 2005
19.6
78.1
20.6/27.3/20.6
MOBILE4 Default
MOBILE4 Default
C
60-84° F
10.5
9.0 in 1992
Basic I/M
95%
8%
Base Case 2:
CO Season
CO only
Low + high altitude
1990, 2005
19.6
43.2, no diurnal
20.6/27.3/20.6
MOBILE4 Default
MOBILE4 Default
E
No diurnal
13.7
13.7
Basic I/M
95%
8%
Ozone Season
Rangesa
..
..
7. 1 (low) - 35 (high)
86.0 (low) -91. 7 (high) d
5.0/5.0/5.0 (low) -
5.0/55.0/5.0 (high)
Fairbanks - California
Fairbanks - Phoenix
CorBc
72-90; 62-1 02d
10.5;9.0d
9.0;7.8d
90-1 00% 9
0-1 6%9
CO Season
Ranges9
7. 1 (low) - 35 (high)
1 1.3 (low) - 66. 1 (high)
57.0/0/57.0 (low) -
5.0/5.0/5.0 (high)
Fairbanks - California
Fairbanks - Phoenix
„
No diurnal
..
„
90-1 00% 9
0-1 6%9
In addition, secondary ranges of 5 mph and 5°F around the base cases and ranges for speeds and temperatures were simulated.
Readers more familiar with metric units may use the factors listed at the end of this project summary.
0 These are trip- and emission-weighted average temperatures as calculated by MOBILE4.
^ Temperature, ASTM class, diurnal range, and RVPs varied jointly for Muskegon and Sacramento cases.
* Percent of VMT accumulated by: non-catalyst vehicles in cold-start mode/catalyst vehicles in hot-start mode/catalyst vehicles in cold-start mode-
other fractions calculated by MOBILE4.
f Inspection/maintenance.
9 Ranges for I/M combined into two cases: 100% compliance with 0% waivers (high); 90% compliance with 16% waivers (low).
gested urban traffic. The high speed of 35
mph is based on the highest-speed urban
driving cycle used by OMS. Since tempera-
ture and RVP are interrelated, two cases
based on recent OMS analyses for Muske-
gon and Sacramento were used to create
two logical joint scenarios for the CO and
ozone season primary sensitivity analyses.
Muskegon represents a case in which the
diurnal temperature range is 18°F rather
than the 24°F of the Federal Test Procedure
(FTP) (with corresponding trip- and emis-
sion-weighted MOBILE4 average tempera-
ture of 86.0°F rather than the 78.1°F FTP
average). The RVP-related parameters for
Muskegon are unchanged from the base
case (10.5 psi in 1990). Sacramento is in a
different ASTM region, and has incremen-
tally lower RVPs (9.0 in 1990), as well as a
larger diurnal range and higher trip- and
emission-weighted average temperature.
Variations in the hot/cold start fractions
were selected to be parallel to the previous
MOBILES sensitivity analysis. For the ozone
season, the set of input fractions labeled as
"low" represents low levels of both cold and
hot starts (5%), resulting in a high level of
stabilized emissions. The set labeled "high"
represents a high level of hot starts (55%)
and low levels of cold starts. For CO season,
the "low" input set includes 57% cold starts
and the "high" set has a minimum level of hot
and cold starts (5%), resulting in a high per-
centage of stabilized emissions. These des-
ignations were used to be consistent with the
previous MOBILES analysis and the "high"
and "low" terms do not refer to the resulting
emissions levels.
Ranges for VMT mix, mileage accumula-
tion, and vehicle age distribution were based
on input variable sets used in the previous
MOBILES analysis to represent significant
variations from the MOBILES defaults. Fair-
banks, Alaska, was selected for its high pro-
portion of light duty trucks, low mileage ac-
cumulation, and relatively low numbers of
older vehicles. The VMT mix and mileage
accumulation for California were chosen for
the relatively high proportion of cars in the
California fleet and the higher levels of mile-
age accumulation reported. The vehicle age
distribution for Phoenix was used as an ex-
ample of an area where vehicles have rela-
tively long lives.
Sensitivity of MOBILE4 to I/M program
parameters was limited to joint variation of
compliance and waiver rates from the base
case (95% compliance, 8% waivers). The
"low" case used 16% waivers and 90% com-
pliance, and the "high" case used 100%
compliance and 0% waivers (an ideal pro-
gram with respect to these variables).
A second set of sensitivity runs was made
to illustrate the effects of smaller changes in
speed or temperature. These runs were
made to illustrate the potential effects of
inaccuracies in these two critical variables
across their potential ranges. These second-
ary sensitivity analyses were made by vary-
ing temperature and speed for each of the
high, low, and default cases in Table 1 by
5°F and 5 mph, respectively.
Sensitivity Analysis Results
Figure 1 presents the results of the sensi-
tivity analysis for the primary ozone and CO
season cases for 1990 at low altitude. Figure
2 presents the results of the secondary sen-
sitivity runs for speed and temperature, also
for 1990 and low altitude.
Results for Low Altitude in 1990
Ozone Season
The first three graphs in Figure 1 show that
speed and the combination of RVP and tem-
perature have the largest and most consis-
tent effects. CO is the pollutant most affected
by speed, followed by HC, and then NOX.
The joint variation of temperature and RVP
to simulate Muskegon and Sacramento re-
sulted in very significant increases in HC and
CO for both cases, with NOX being reduced
slightly. This overall effect is due primarily to
the differences in temperature, with the HC
and CO results for Sacramento being tem-
pered by the lower RVP.
-------
Ozone season MOBIUE4 sensitivity results
1990, low altitude, HC
(2)
I/M program (low)
RVP/Temp (Sacramento)
RVP/Temp (Muskegon)
Vehicle age (Phoenix)
Vehicle age (Fairbanks)
(3)
Mileage accum. (Fairbanks)
VMTmix (California)
VMT mix (Fairbanks)
(2)
(11) I
(25)
Speed (low)
I/M program (high)
2
31
Mileage accum. (California)
3
Hot/Cold start (high)
Hot/Cold start (low)
Speed (high)
s38^ 93
L
-150 -100 -50 0 50 100
Percent change from the base scenario
150
200
Ozone season MOBILE4 sensitivity results
1990, low altitude, CO
(6)
I/M program (low)
RVP/Temp (Sacramento)
RVP/Temp (Muskegon)
Vehicle age (Phoenix)
Vehicle age (Fairbanks)
0
0
VMTmix (California)
VMT mix (Fairbanks)
(7) I
(16) m
(45)
Speed (low)
_L
I/M program (high)
5
19
75
I 1
me
Mileage accum. (California)
Mileage accum. (Fairbanks)
0 3
2
Hot/Cold start (high)
Hot/Cold start (low)
Speed (high)
151
-150 -100 -50 0 50 WO
Percent change from the base scenario
150
200
Figure 1. Primary MOBILE4 sensitivity analysis results (low altitude - 1990).
-------
Ozone season MOBILE4 sensitivity results
1990, low altitude, NO*
0
0
(5)
(3)
(2)
Vehicle age (Fairbanks)
Mileage accum. (California)
Mileage accum. (Fairbanks)
(D
VMT mix (Fairbanks)
(2)
(12) I
(6)
Speed (low)
I/M program (high)
I/M program (low)
RVP/Temp (Sacramento)
RVP/Temp (Muskegon)
Vehicle age (Phoenix)
47
36
VMT mix (California)
36
Hot/Cold start (high)
Hot/Cold start flow)
Speed (high)
23
J_
•150 -100 -50 0 50 100
Percent change from the base scenario
150
200
CO season MOBILE4 sensitivity results
1990, low altitude, CO
(5)
I/M program flow)
Vehicle age (Phoenix)
Vehicle age (Fairbanks)
(1)
(2)
VMT mix (California)
0
(30)
Hot/Cold starts flow)
(19)
Temperature flow)
(44)
Speed flow)
I/M program (high)
4
Mileage accum. (California)
Mileage accum. (Fairbanks)
VMT mix (Fairbanks)
Hot/Cold starts (high)
68
Temperature (high)
53
Speed (high)
146
•150 -100 -50 0 50 W0~
Percent change from the base scenario
150
200
Figure 1. (Continued)
-------
Ozone season MOBILE4 sensitivity results
Secondary temperature and speed ranges for HC
Temperature (high+5)
(13) I
Temperature (low+5)
(16)$
Temperature (base+5)
(9)
(5)
Speed (high-5)
(31)
Speed (low-5)
(11) I
Speed (base-5)
_L
915
Temperature (high-5)
16
Temperature (low-5)
20
Temperature (base-5)
Speed (high+5)
7
Speed (low+5)
Speed (base+5)
18
134
-150 -100 -50 0 ^50"TOO"
Percent change from the base scenario
150
200
Ozone season MOBILE4 sensitivity results
Secondary temperature and speed ranges for CO
Temperature (high+5)
(11)
Temperature (low+5)
(9)
Temperature (base+5)
(2)
(12)
Speed (high-5)
(39)
Speed (low-5)
(20)
Speed (base-5)
8
Temperature (high-5)
Temperature (low-5)
I 10
Temperature (base-5)
Speed (high+5)
17
Speed (low+5)
Speed (base+5)
29
J L
170
-150 -100 -50 0 50 100
Percent change from the base scenario
150
200
Figure 2. Secondary MOBILE4 sensitivity analysis results (low altitude -1990).
-------
Ozone season MOBILE4 sensitivity results
Secondary temperature and speed ranges for NO*
(2)
Temperature (high-5)
(2)
Temperature (low-5)
(2) I
Temperature (base-5)
Speed (high+5)
0
(10)
Speed (low-5)
Speed (base-5)
L
Temperature (high+5)
I 2
Temperature (low+5)
I 2
Temperature (base+5)
2
1
Speed (high-5)
Speed (low+5)
15
Speed (base+5)
\6
_L
-150 -100 -50 0 50 100
Percent change from the base scenario
150
200
CO season MOBILE4 sensitivity results
Secondary temperature and speed ranges for CO
(5)
Temperature (high-5)
(7)
Temperature (low-5)
(4)
Temperature (base-5)
(12)
Speed (high-5)
(39)
Speed (low-5)
(20)
Speed (base-5)
J I
Temperature (high+5)
5
Temperature (low+5)
7
Temperature (base+5)
5
Speed (high+5)
m 17
Speed (low+5)
170
Speed (base+5)
28
I L
-750 -100 -50 0 50 100
Percent change from the base scenario
150
200
Figure 2. (Continued)
-------
Both of the alternate ozone season
hot/cold start ratios reduced start emissions
for all pollutants. Changes in the I/M program
waiver and compliance rates have a rela-
tively small effect on HC emissions, a slightly
larger effect of CO emissions, and no effect
on NOX emissions (typical I/M programs are
not intended to reduce NOx).
The three variables associated with local
vehicle fleet characteristics (VMT mix, mile-
age accumulation, and vehicle age distribu-
tion) show relatively little effect on overall
emissions in 1990. The major exception is
the effect of the Fairbanks VMT mix on NOX
emissions, which is mainly due to an in-
crease in heavy duty diesel VMT of 130%
over the base case. The Fairbanks VMT mix
was also used in conjunction with the Fair-
banks vehicle age and mileage accumula-
tion cases, so the NOX results for these
cases are affected similarly. The net result is
thatthe Fairbanks mileage accumulation ap-
pears to have no effect and the Fairbanks
vehicle age distribution has about 9% addi-
tional effect on NOX emissions. Of the other
fleet-related variables, the only results over
5% are the effect of the Fairbanks vehicle
age distribution on HC and CO emissions
and the effect of the California mileage ac-
cumulation on NOX emissions. The effects of
fleet-related variables can vary among pol-
lutants. For example, the California mileage
accumulation reduces HC by 3%, increases
NOX by 8%, and has no effect on CO.
CO Season
The last graph in Figure 1 shows the vari-
ation of CO fleet composite emission factors
from the CO season base case inputs to the
CO season ranges shown in Table 1. For the
cooler CO season temperatures, speed re-
mains a major variable, and the tempera-
tures and hot/cold start ranges for the CO
season also have dramatic effects on CO
emissions. Effects are especially pro-
nounced for low speed (7.1 mph), low tem-
perature (11.3°F), and the cold-start-domi-
nated hot/cold start mix. The I/M program
ranges have slightly less effect at CO sea-
son temperatures than at ozone season
temperatures. The vehicle-fleet-related vari-
ables also have less effect at CO season
temperatures.
Secondary Sensitivity Runs
for Speed and Temperature
Figure 2 shows the results of further vari-
ation of speed and temperature around the
base case and ranges described in Table 1.
Secondary ranges of 5°F and 5 mph were
used around both the ozone and CO season
base cases and the cited primary ranges.
This exercise illustrates the sensitivity of
MOBILE4 results to small potential errors in
these key variables.
In general, the results in Figure 2 indicate
that relatively small differences in average
speed and temperature can have significant
effects on the MOBILE4 composite fleet
emission factors for ozone season HC and
for CO in both seasons. NOX is much less
sensitive to speed and temperature.
Results for 2005
and for High Altitude
The main purpose of these runs was to
identify situations in which the basic conclu-
sions obtained for low altitude in 1990 might
change in the future or in the few areas
considered high altitude for MOBILE4 mod-
eling. For all future year and high altitude
cases, the effect of changing I/M program
compliance and waiver rates is marginally
less than for low altitude in 1990. For 2005
at low altitude, the following ozone season
parameters showed some significant varia-
tion from the 1990 base year:
• Effect of the specified RVPAemperature
combinations is somewhat smaller for HC
and their slightly negative effects for NOX
in the base year become slightly positive
for 2005.
• Effects of vehicle fleet parameters and
hot/cold start ratios are marginally larger
for HC and CO, and are mixed for NOX.
• Effects of the speed range become a little
larger for HC and NOX, but are reduced for
CO.
For the CO season, the results of the
specified variations in vehicle fleet parame-
ters are mixed, and results for hot/cold start
mix, temperature, and speed are marginally
smaller than in 1990.
Going from low to high altitude for the two
analysis years results in only a few notice-
able changes. The general effects of the
vehicle fleet characteristics on ozone sea-
son HC and CO at high altitude are some-
what greater for 1990 and are considerably
greater in 2005. The Fairbanks VMT mix
appears to be responsible for the biggest
changes, resulting in roughly double the
changes seen for HC at low altitude. Sensi-
tivity to speed is slightly lower in 2005, for
ozone season HC and CO emissions as well
as for CO emissions in the CO season.
Conclusions
The following general conclusions can be
drawn from this analysis:
The most consistently significant variables
identified in this analysis are speed and the
combination of Reid vapor pressure (RVP)
and temperature (temperature alone for the
CO season). This is true for current and
future years, and for high and low altitudes.
The increasing sensitivity of emissions at
lower speeds indicates that methods of in-
corporating speeds in inventories should be
oriented toward better reflection of the true
distribution of speeds.
The combinations of RVP and tempera-
ture used for the ozone season cases can
cause significant variations between areas
(about 20 to 30% difference in HC and CO).
The effect of RVP is less pronounced in
future years due to Federal RVP control
mandates.
The overall results for the CO season point
out the need for explicit consideration of
speed, temperature, and vehicle starts in
analyses of CO exceedance situations and
for the development of better methods and
more accurate data for these variables if
possible.
Secondary sensitivity analyses around the
base and primary ranges for temperature
and speed indicate that relatively small dif-
ferences in these variables can have signif-
icant effects on the MOBILE4 emission fac-
tors for ozone season HC and for CO in both
seasons.
Results for vehicle fleet characteristics
and VMT mix were significant in only a few
of the cases in this analysis. The most dra-
matic result was a 36% increase in NOX due
to a 130% increase in heavy duty diesel VMT
in the VMT mix for Fairbanks. The specific
fleet-related variables were taken from a pre-
vious analysis and no independent attempts
were made to determine whether they actu-
ally represent appropriate alternative cases
for the 1990 and 2005 analysis years. More
detailed study of this area could provide
inventory managers a better perspective on
the value of developing area-specific inputs
for these variables and on methods for their
development.
Metric Conversions
Readers more familiar with metric units
may use the following factors to convert to
that system.
Nonmetric Times
°F 0.556(T-32)
gm/mi 1.609
Ib 0.454
mph 1.609
psi 6.895
Yields Metric
°C
gm/kg
kg
km/h
kPa
.S. GOVERNMENT PRINTING OFFICE: Ml - 548-02B/40043
------- |