EPA420-F-00-050
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
ANN ARBOR, Ml 48105
January 2, 2001
OFFICE OF
AIR AND RADIATION
MEMORANDUM
SUBJECT: Emission Data and Procedures for Large SI Engines
FROM: Alan Stout, Mechanical Engineer
Assessment and Standards Division
Chuck Moulis, Engineer
Assessment and Standards Division
THRU: Glenn Passavant, Nonroad Center Director
Assessment and Standards Division
TO: Docket A-2000-01
In an Advance Notice of Proposed Rulemaking from the Environmental Protection Agency, we announce our
intent to develop a program to regulate emission levels from nonroad spark-ignition engines rated over 19 kW ("Large
SI engines"). This category of engines generally includes all nonrecreational land-based engines that are not installed
in motor vehicles or stationary applications. This memorandum reviews available technical information relevant to
setting emission standards for these engines. The memorandum first considers various ways to define test procedures
for measuring emissions, then presents emission data that will shed light on appropriate emission standards.
I. Emission Testing to Support the Development of Emission Standards
In 1998, California ARE conducted testing with Southwest Research Institute (SwRI) to determine the
feasibility of achieving low emission levels with a three-way catalyst system that could be installed in nonroad
equipment. They were successful in showing that an engine with a new emission-control system was capable of
operating effectively, both to control emissions and to provide satisfactory performance in the equipment.1 These data
are summarized in Section III below.
In 2000 SwRI did further testing on Large SI engines with funding and/or consultation from the California
ARE, the South Coast Air Quality Management District, and EPA. This effort focused on three principal remaining
areas. First, we obtained transient duty cycle information by monitoring several engine, equipment, and ambient
variables from normal forklift operation. Second, we removed these two engines from the forklifts and operated them
-------
in the laboratory. Since these engines had each accumulated several thousand hours of operation with a functioning
catalyst system, this showed us how emission-control systems perform over extended normal operation in an industrial
application. Third, emission testing on a full range of steady-state test points and several transient duty cycles, with
optimized engine calibrations as appropriate, provided a good indication of the level of emission-control possible from
the installed technologies.
The selected engines had been retrofitted with the emission-control systems in Spring 1997 after having
already run for 5,000 and 12,000 hours. Both engines are in-line four-cylinder models operating on liquefied petroleum
gas (LPG)—a 2-liter Mazda engine rated at 32 hp and a 3-liter GM engine rated at 45 hp. The retrofit consisted of a
new, conventional three-way catalyst, electronic controls to work with the existing fuel system, and the associated
sensors, wiring, and other hardware. The electronic controller was the more basic type that allows only a single
adjustment for controlling air-fuel ratios across the range of speed-load combinations.
Testing occurred in laboratory conditions typical for the test location. Ambient temperatures ranged from 70
to 86° F. Barometric pressures were in a narrow range around 730 mm Hg. Humidity levels ranged from about 4 to 14
g of water per kg dry air, but all emission levels were corrected to a reference condition of 10.7 g/kg. Most testing
occurred at humidity levels above 10.7, in which case actual NOx emission levels were up to 7 percent lower than
reported by SwRI. In the driest conditions, measured NOx emission levels were up to 10 percent higher than reported.
The SwRI report include a further description of these forklifts, the facility where they were operating, and the
results of the laboratory testing.2 These data are summarized in Section III below.
II. Emission Measurement
Before considering the appropriate emission standard for these emission-control technologies, it is important to
consider the procedures for measuring emissions.
Defining emission standards based on steady-state duty cycles is a common approach. The Large SI emission
standards set by California ARE, for example, are based on the ISO C2 and D2 duty cycles. The following sections
describe a rationale for supplementing these steady-state duty cycles with additional ways to measure emission-control
effectiveness.
A. Why measure emissions during transient operation?
An engine designed for low emissions from a small number of discrete test points may not effectively control
emissions in use. We have measured engine operation in forklifts and found that, except for extended engine idling,
there is very little steady-state operation over the course of normal operation. Similar measurements in a wide range of
-------
diesel equipment have also shown that steady-state operation in the field is quite rare for those engines. Engines that
power arc welders are governed to operate at a constant speed, but even in this application, engine load varies
significantly.
SwRI testing has shown a wide variation in transient emission levels from an engine with a consistent steady-
state calibration that gives low emission levels on the C2 duty cycle. Table 1 shows a variety of measured transient
emission levels corresponding with a single steady-state calibration. Transient HC+NOx levels were up to 20 times
higher than steady-state levels on the same engine." Transient CO levels were sometimes lower than steady-state levels,
but were in some cases higher by a factor of five. This shows that the steady-state test results are a poor predictor of
emissions during transient operation.
Table 1
Mazda Engine Emission Levels for Various Settings (g/hp-hr)
Test Engine
Mazda
C2 Emission Levels
THC+NOx
0.51
CO
3.25
Transient Emission Levels
THC+NOx
10.1
3.3
CO
0.07
17.1
Some of the difference between steady-state and transient emission levels results from the fact that the
transient operation traverses a wide range of steady-state test points, many of which may have emission levels
significantly higher than those included in the seven modes of the C2 duty cycle. Optimizing the engine for its best
control of transient emissions generally resulted from adjusting the control of air-fuel ratios to keep the engine at
stoichiometry at idle and other low-speed operating points.
Requiring manufacturers to measure emissions on a prescribed transient duty cycle would be an effective way
of ensuring that engines are able to control emissions in real-world operation. The design engineer can investigate
control strategies to address high emission levels from transient operation, primarily by focusing on careful control of
air-fuel ratios at idle and low-speed operation.
B. What are off-cycle emissions?
As described earlier, a steady-state duty cycle with a discrete number of test points fails to capture a wide
range of in-use operation. A transient duty cycle helps address these concerns by exercising the engine as it goes
throughout this memorandum, measured hydrocarbon emissions are either nonmethane
hydrocarbon (NMHC) or total hydrocarbon (THC), which includes methane emissions.
-------
through many different speeds and loads of real operation. However, a transient duty cycle based on measured
operation from a few pieces of equipment still fails to include many kinds of operation from other engines. Similar
equipment operating in a different facility may undergo significantly different operation in the field. Also, different
types of equipment performing widely varying tasks will likely have engine operation that is unique to the application.
No single, defined duty cycle can include the whole range of engine operation. Not-to-exceed testing
acknowledges this by setting a standard for any normal operation, taking into account the natural variation in emissions
from varying engine operation at different speeds and loads. Similarly, engines in the field are not restricted to
operating in controlled laboratory conditions. Any time engines are designed only to control emissions under a narrow
range of operation or ambient conditions, there is a significant risk that the anticipated emission reductions will not
materialize.
Emission measurements from the SwRI testing highlight this concern. Figure 3 shows that an engine can have
very low NOx emission levels at some points, with a 10-fold increase at other points under the engine map. This
variation is not the result of an intent to "beat" the cycle with a design that operates at low levels only for a certification
test. Rather, this reflects the normal variation for a first-iteration effort to calibrate the engine for low emissions.
Figure 4 provides an example of a problematic calibration. This engine has very high CO levels at all the full-
load points, indicating the likely need for a software adjustment to better control air-fuel ratios at those points. A
transient duty cycle may include very little engine operation at these high-emission points, and would therefore not
require manufacturers to optimize emission levels there. If this engine is installed in an application where it operates
more often on the engine's lug curve, it will have much higher emissions than would be predicted by the steady-state or
transient duty cycles.
-------
Figure 3: Truck 29, Steady-State NOX Emissions Results over
Normalized Speed and Load
STEADY STATE NOx EMISSIONS [g/hp-hr]
10 o
TRUCK 29
GM 3.0 L ENGINE
NEW CAT
Figure 4: Truck 16, Steady-State CO Emissions Results
STEADY STATE CO EMISSIONS [g/hp-hr]
TRUCK 16
MAZDA 2.0 L ENGINE
24
23
22
21
20
19
18
17
-14 TT
13
12
11
10
-------
One of the SwRI test engines was found to be unable to provide enough fuel flow at one engine operating
mode. The air-fuel ratio at this point was so high that there was no NOx reduction occurring in the catalyst. A simple
adjustment to enable increased fuel flow allowed the engine to operate at stoichiometry, which brought the NOx
conversion efficiency up to about 80 percent at that mode.
Adopting a not-to-exceed standard would require the design engineer to control an engine's emissions under
the whole range of normal in-use operation. An additional important advantage of not-to-exceed testing relates to the
ability to test engines in the field. If testing were limited to discrete steady-state or transient duty cycles, it would not be
possible to conduct a valid emission test without a laboratory dynamometer. The next section addresses this in further
detail.
C. Why do we want to measure emissions in the field?
Measuring an engine's emissions as it undergoes normal operation in the field, powering a specific piece of
equipment on any given day, is the best way of verifying that an engine is achieving the intended level of emission
control during real operation. The alternative to field testing is to remove engines and ship them to a laboratory for
dynamometer testing. The high cost of this effort limits the amount of in-use testing we can reasonably conduct or
expect manufacturers to conduct.
The tools for field measurement have advanced to the point that we are able to consider new equipment and
procedures to do a valid emission test without removing the engine. The equipment in which the engine is installed
serves as the dynamometer. The portable devices for measuring emissions in the field cost significantly less than full-
size laboratory equipment.
Engineers can design an engine's onboard computer to monitor speed and load at all times. Identifying and
recording speed is straightforward. For torque values, the manufacturer would need to monitor a surrogate value such
as intake manifold pressure or throttle position, then rely on a look-up table programmed into the onboard computer to
convert these torque indicators into foot-pounds or other equivalent torque units. Manufacturers may also choose to
program the torque-conversion tables into a remote scan tool.
Such a method would allow manufacturers to test the required number of in-use engines at a much lower cost
than in a laboratory. EPA or manufacturers could also do additional testing to better understand the in-use performance
of engine designs certified to meet emission standards.
To use field-testing equipment and procedures as a tool to confirm that engines comply with emission
standards, we would want to adopt test procedures, including calibration and measurement procedures, definitions, and
other specifications. This would help ensure sufficient accuracy to rely on the measurement as a valid indication of an
-------
engine's emission level.
Designing an engine to monitor and report its instantaneous speeds and loads also has an additional benefit.
Manufacturers could easily monitor engines to develop cycles of normal operation for individual applications. This
could be used to verify that engine designs are able to control emissions in various applications, thereby reducing the
risk of noncompliance from field measurements. This duty-cycle information could also be used for product-
development purposes to ensure that engines will perform well in specific applications.
D. Why would we keep steady-state test procedures?
As described above, a steady-state duty cycle provides limited assurance that in-use engines will adequately
control emissions over their useful life. However, we would likely propose including steady-state operation as part of
the test program for four reasons.
First, measuring emissions on the available steady-state duty cycles addresses intermediate-speed operation
that is not covered well by the transient segment. Currently available data for developing the transient duty cycle do not
include engines used in vehicles like sweepers, aerial lifts and airport service vehicles that have more intermediate-
speed operation. Thus, inclusion of the C2 test points adds to the overall representativeness of the duty cycle.
Second, using the C2 and D2 duty cycles would allow for better harmonization of our standards with those of
California ARB. Worldwide harmonization is also at issue, since European and Japanese regulators have long relied on
the steady-state ISO duty cycles to define emission standards.
Third, identifying emission levels on the C2 or D2 duty cycles provides a useful benchmark for later testing.
For example, the goal of production-line testing is primarily to evaluate the effect of production variability on emissions
performance. Routinely testing production-line engines under transient operation should therefore not be necessary.
Simpler tests with the steady-state testing modes should be sufficient to quantify the effect of production variability. If
steady-state testing reveals a problem, however, it may be best to do transient emission measurements before presuming
noncompliance.
Fourth, the C2 and D2 duty cycles have been widely used as duty cycles for evaluating emission performance.
Measuring emissions on these steady-state cycles provides consistency with emission requirements adopted by
California ARB. This also helps for emission modeling purposes, where historical data is all based on steady-state
emission levels.
E. How did EPA develop transient duty-cycle segments?
The primary goal in defining a transient duty cycle is to include a wide range of real engine operation. A well-
7
-------
designed duty cycle would provide assurance that engines used in a wide variety of applications are able to control
emissions effectively. As a result, a good transient duty cycle would include a variety of challenging, real in-use
operation to minimize the risk of surprisingly high emissions from not-to-exceed testing.
Our approach in constructing a transient duty cycle was to include measured engine operation from various
sources. Forklifts are the dominant application, so it is clearly important to include operation from that application.
Besides forklifts, Large SI equipment applications fall into two broad categories. First, other variable-speed
applications usually have an operator who "drives" the equipment (sweepers, ice surfacing machines, etc.). We expect
most of these other variable-speed applications to include a subset of the kinds of operation measured on the forklifts
and therefore believe that the forklift operation may adequately cover these applications. This group of non-forklift
engines is also relatively small, accounting for less than 10 percent of Large SI engines. Second, the remaining engines
are used in a wide range of portable (or transportable) applications that experience much less variation in engine speeds
than motive equipment. Examples include generators, welders, pumps, compressors, saws, and chippers. These
applications are typically governed to operate within a narrow band of engine speeds.
We developed most of the data for a Large SI transient duty cycle in a contracted effort with Southwest
Research Institute (SwRI), with assistance and consultation from California ARE and the South Coast Air Quality
Management District. To do this, we selected a pair of forklifts operating at an apple-processing facility. The report
describing this effort includes a further description of these forklifts and the facility where they were operating.3 After
a statistical characterization of the approximately 10 hours of measured engine operation, we selected one five-minute
segment that had the most typical operation from each forklift.4 Additional five-minute segments from each lift truck
captured the most highly transient activity.
Using the two "typical" segments provides 10 minutes of test operation. Selecting the first half of each of the
high-transient segments from each forklift adds another 5 minutes. This allows us to incorporate the real operation
from the high-transient segment without overemphasizing it in the total measurement.
To include the operation of engines in portable applications, we borrowed a 5-minute segment of typical
operation measured from a diesel welder. SwRI developed this under an earlier work assignment. The welder,
governed to operate at rated speed, had engine loads generally varying between 10 and 50 percent. This aligns well
with the modal distribution of the ISO D2 cycle. Also, we would expect diesel and spark-ignition engines operating an
arc welder to have very similar engine operation. People using the welder would not likely change their welding
methods for the different engine types. Welder manufacturers may size engines somewhat differently based on the type
of engine, but the observed engine loads for the diesel welder appear to fall within the expected range for spark-ignition
engines.
Combining the forklift and welder segments into a single 20-minute cycle results in a procedure that we
believe will cause manufacturers to design emission-control systems that are effective at reducing emissions from a very
-------
wide range of engine applications. Adding in elements from other applications may improve the statistical
representativeness of the duty cycle, but will not likely yield a significant degree of additional emission control.
The 20-minute transient duty cycle described above should not generally apply to engines that rarely or never
operate that way. We are therefore considering a different transient duty cycle for all engines designed to operate only
at constant speed. This alternate transient cycle is a 20-minute set of measured engine operation from the diesel welder.
Half of this is from measured typical operation and half is from a high-transient segment. We expect to propose this
same constant-speed transient cycle for nonroad diesel engines in a later rulemaking.
The warm-up sequence preceding the transient duty-cycle segment for emissions measurement is also
important. The warm-up begins with a cold-start. This means that the engine should be very near room temperature
before the test cycle begins. Once the engine is started, it would be operated over the first 3 minutes of the specified
transient duty cycle without emission measurement. The purpose of the warm-up segment is to bring the engine up to
normal operating temperature in a standardized way. The 3-minute warm-up period allows enough time for engine-out
emissions to stabilize, for catalyst light-off to occur, and for the engine to start closed-loop operation. This serves as a
defined and achievable target for the design engineer to limit cold-start emissions to a relatively short period. SwRI
testing has shown that a wide variety of Large SI engines have stabilized engine-out emissions after 3 minutes of
operation or less.5
F. How would field testing work?
The goal of field testing is to characterize the emission levels resulting from normal operation. With
established equipment and procedures for measuring emissions, the field test can consist of direct measurement of
exhaust concentrations without removing the engine. As with any in-use emission testing, all properly maintained
engines would be expected to meet the emission standards.
We are considering constraints on several operating variables to ensure that emission sampling occurs during
normal engine operation, and that measured emissions include no spurious data. We request comment on the following
parameters:
Average power should be above some threshold to avoid very high brake-specific emission levels (as
a result of dividing by near-zero power). As much as possible, such a threshold should include
normal operation without reaching unusually high brake-specific emission levels.
The emission sampling period should not include a continuous idling period so long that catalyst
temperatures cool enough to significantly decrease conversion rates. Longer idling periods could cool
the catalyst below light-off temperatures, resulting in unrepresentatively high emission levels. Since
idle operation is included in the steady-state test, it isn't necessary to include this in not-to-exceed
testing.
The emission sampling period should start after the engine has reached stable operating temperatures
-------
and should not include engine starting. Cold-start effects are addressed at certification and need not
be included in not-to-exceed measurements.
We are considering a maximum frequency of acceleration events. A maximum number of
accelerations per minute (or a comparable limit) should be based on the upper end of real-world
operation from high-transient activity. More severe transient activity could pose an unrealistic
challenge to consistently provide the right quantity of air to the cylinders.
The sampling period should be longer than some minimum interval. We are considering a minimum
sampling time between 30 and 120 seconds.
Severe-duty engines are unable to operate for extended periods at wide-open-throttle without
overheating. We are therefore considering a limit on this operation for severe-duty engines. For
example, we could restrict operation to be above 90 percent of maximum engine power less than 10
percent of the emission sampling period. We also request comment on the need to apply this type of
provision to water-cooled gasoline engines.
In addition, field testing should include emission measurement under a wide range of ambient conditions
representative of conditions engines experience in the field. Based on similar programs with other engine categories,
we are considering a range of ambient temperatures and pressures to reflecting normal variations in the conditions in
which engines operate.
During the emission sampling period, manufacturers would operate the engine on a representative commercial
fuel or on a fuel that meets the specifications for certification testing. We would allow the engine to operate on any
commercially available fuel for service accumulation.
G. What about not-to-exceed testing in the laboratory?
The provisions described above for field testing would apply to not-to-exceed testing in the laboratory. Any
steady-state operation consistent with the above criteria would be a possible test point. To measure transient not-to-
exceed operation, the engine would need to operate over a sequence of speeds and loads that could be characterized as
normal operation. This could come from any segments of measured engine operation, consistent with the listed criteria.
Some examples of valid engine operation would include:
A subset of the transient duty cycle for certification.
Other segments of the lift truck operation measured by SwRI.
New measurements of engine speeds and loads from another Large SI application.
We would not base emission measurements on engine operation in a given application if the manufacturer
didn't sell engines into that application. Conversely, if engines from an engine family power equipment in several
different applications, any in-use engine tested on a dynamometer should comply with emission standards based on
operation from any other type of equipment in which the engines are installed.
10
-------
III. Selecting Emission Standards
The Clean Air Act requires that standards achieve the greatest degree of emission reduction achievable
through the application of technology that will be available, giving appropriate consideration to cost and other factors.
This section describes how we will consider applying this to Large SI engines.
Engine manufacturers are currently developing technologies and calibrations to meet the 2004 standards that
apply in California. We expect manufacturers to rely on electronically controlled, closed-loop fuel systems and three-
way catalysts to meet those emission standards. As described below, emission data show that water-cooled engines can
readily meet the California ARB standards (3 g/hp-hr NMHC+NOx; 37 g/hp-hr CO). The rest of this memorandum
addresses the potential to meet more stringent emission standards with new test procedures.
The California ARB emission standards described above would be effective in reducing emissions from Large
SI engines, but we believe these levels don't fulfill our obligation to adopt standards achieving the "greatest degree of
reduction achievable" from these engines.
The biggest uncertainty in adopting emission standards for Large SI engines has been the degree to which
emission-control systems deteriorate with age. While three-way catalysts and closed-loop fueling systems have been in
place in highway applications for almost 20 years, there is very little information showing how these systems hold up
under nonroad use. To address this, we participated in an investigative effort with SwRI, the California ARB, and the
South Coast Air Quality Management District, as described in Section I above.
Laboratory testing consisted of measuring steady-state and transient emission levels, both before and after
taking steps to optimize the system for low emissions. This testing provides a good indication of the capability of these
systems to control emissions over an engine's full useful life. The testing also shows the degree to which transient
emissions are higher than steady-state emission levels for Large SI engine operation. Finally, the testing shows how
emission levels vary for different engine operating modes. Much of the emissions variability at different speeds and
loads can be attributed to the basic design of the controller, which has a single, global calibration setting.
A. Steady-state testing results
Testing results from the aged engines at SwRI showed very good emission control capability over the full
useful life. Test results with new hardware on the aged engines lead to the conclusion that the systems operated with
relatively stable emission levels over the several thousand hours. As shown in Table 2, the emission levels measured by
SwRI are consistent with results from a wide variety of measurements on other engines. This data set supports
emission standards significantly more stringent than those presently established in California.
11
-------
Table 2
Steady-State Emission Results from Various Engines
Test engine
Mazda 2L6
GM3L7
Engine B8
Engine E9
GFI10
Toyota/ECS 2L11
GM/Impco 3L12
Fuel
LPG
LPG
LPG
gasoline
LPG
LPG
LPG
HC+NOx*
g/hp-hr
0.51
0.87
0.22
0.28
0.52
NMHC+NOx
1.14
0.26
CO
g/hp-hr
3.25
1.84
2.79
42.2
2.23
0.78
0.21
Notes**
4,000 hours
5,600 hours
250 hours
250 hours; air-cooled; ISO D2 duty
cycle
5,000 hours
zero-hour; ISO Cl duty cycle
zero-hour
*Measurements are THC+NOx, unless noted otherwise
"Emissions were measured on the ISO C2 duty cycle, unless noted otherwise.
B. Transient testing results
The SwRI testing is currently the only source of information available for evaluating the transient emission
levels from Large SI engines equipped with emission-control systems. Table 3 shows the results of this testing. The
transient emission levels are higher than those measured on the steady-state duty cycles. This probably results primarily
from the fact that the transient duty cycle includes operation at engine speeds and loads that have higher steady-state
emission levels than the seven modes constituting the C2 duty cycle.
Table 3
Transient Test Results from SwRI Testing
Engine*
Mazda
GM
Duty Cycle
Variable-speed, variable-load
Constant-speed, variable-load
Variable-speed, variable-load
THC+NOx
g/hp-hr
1.1
1.5
1.2
CO
g/hp-hr
9.9
8.4
7.0
*Based on the best calibration on the engine operating with an aged catalyst.
12
-------
C. Not-to-exceed testing results
Engines operate in the field under both steady-state and transient conditions. Although these emission levels
are related to some degree, they are measured separately. This section therefore first considers steady-state operation.
Figures 5 through 10 show plots of emission levels from the test engines at several different steady-state
operating modes. The plotted emission levels show the emissions at each normalized speed and normalized load point.
The 100-percent load points at varying engine speeds form the engine's lug curve, which appears as a straight line
because of the normalizing step.
Figure 5 shows the THC+NOx emissions from the Mazda engine when tested with the aged catalyst. While
several points are higher than the 0.51 g/hp-hr level measured on the C2 duty cycle, the highest levels observed from
the Mazda engine are around 2.3 g/hp-hr. The highest emissions are generally found at low engine speeds.
CO emissions from the same engine had a similar mix of very low emission points and several higher
measurements. The CO levels along the engine's lug curve range 12 to 22 g/hp-hr, well above the other points, most of
which are under 4 g/hp-hr. The corner of the map with high-speed and low-load operation also has a high level of 9
g/hp-hr. These high-emission modes point to the need to address control of air-fuel ratios at these extremes of engine
operation.
If CO emissions at these points are an inherent problem associated with these engines, we could take that into
account in proposing the standard. Figure 8 shows, however, that the GM engine with the same kind of aged emission-
control system had emission levels at most of these points ranging from 0.7 to 4.7 g/hp-hr. The one remaining high
point on the GM engine was 11.6 g/hp-hr at full load and low speed. A new high-emission point was 28 g/hp-hr at the
lowest measured speed and load. Both of these points are much lower on the same engine with the new catalyst
installed (see Figure 10). These data reinforce the conclusion that adequate development effort will enable
manufacturers to achieve broad control of emissions across the engine map.
Figure 7 shows the THC+NOx emissions from the GM engine when tested with the aged catalyst. Emission
trends across the engine map are similar to those from the GM engine, with somewhat higher low-speed emission levels
between 2.3 and 4.4 g/hp-hr at various points. Operation on the new catalyst shows a significant shifting of high and
low emission levels at low-speed operation, but the general observation is that the highest emission levels disappear,
with 2.3 g/hp-hr being again the highest observed emission level over the engine map (see Figure 9).
13
-------
Figure 5
Mazda/old cat.--NOx+HC
g/hp-hr
100
80
as 60
o
40
20
0
1
1.89
1.19
1.61
1.92
2.08
2.28
1.43
0 20
C2 = 0.51 g/hp-hr
0.6
0.46
0.95
0.87
1.11
1.67
2.26
30 40
0.77
0.53
0.41
0.35
0.33
0.14
1.24
50 60 70
Speed
0.57
0.31
0.31
0.5
0.62
0.72
0.28
80 90
0.25
0.27
0.25
0.43
0.63
0.81
0.54
100 110
Figure 6
Mazda/old cat--CO
100 2224
80 1'07
0.23
§ 6° 0.33
_i
40 0.64
20 °'51
1.3
10 20
C2=3.25 g/hp-hr
11.52
2.28
1.27
0.88
0.56
0.04
0.19
30 40
g/hp-hr
15.24
8.07
4.06
2.44
0.91
0.79
0
50 60 70
Speed
18.98
4.17
3.01
3.87
3.61
2.89
1.61
80
2.49
3.87
3.88
3.9
4.47
7.6
9.08
90 100 110
14
-------
Figure 7
GM/old cat.--NOx+HC
100 3.5
80 16
1.6
as 60 1R
o 1 .6
_i
40 2.3
20 23
4.4
10 20
C2=0.87 ghp-hr
1.3
1.2
1.1
1.2
0.9
0.6
1.6
30 40
g/hp-hr
0.8
0.7
1.2
0.9
0.7
0.5
1.1
50 60 70
Speed
0.8
0.7
0.6
0.6
0.7
0.5
0.7
80
0.9
0.9
0.6
0.5
0.4
0.7
0.4
90 100 110
Figure 8
GM/old cat.--CO
100 11.6
3 9
80 Jy
4.3
1 6° 4.1
_i
40 6.0
20 35
28.0
10 20
02=1.84 ghp-hr
0.7
2.1
2.4
3.5
3.6
3.9
5.1
30 40
g/hp-hr
4.7
0.6
1.7
1.6
2.1
1.1
1.4
50 60 70
Speed
4.5
0.7
0.6
0.8
0.3
2.8
10.3
80
0.7
0.7
1.3
1.8
1.4
6.2
4.3
90 100 110
15
-------
Figure 9
GM/new
100 0.57 0.92
80 2'25 °'75
2.25 0.82
1 6° 1.93 0.79
_l
40 1.61 0.83
2Q 1.33 0.66
1.47 1.17
10 20 30 40 50
C2=0.35 ghp-hr
cat.-NOx+HC
g/hp-hr
0.32
0.28
0.19
0.25
0.30
0.13
0.25
60 70
Speed
0.26 0.14
0.18 0.11
0.19 0.08
0.20 0.05
0.06 0.04
0.13 0.08
0.65 0.16
80 90 100 110
Figure 10
GM/new cat.
100 4.08
80 °'55
0.33
03 60
o 0.33
_i
40 0.24
20 °'11
0.45
10 20
C2=0.28 ghp-hr
0.16
1.03
0.92
0.70
0.72
1.04
0.44
30 40
g/hp-hr
2.65
0.81
0.37
0.00
0.93
0.29
0.73
--CO
1.78
0.23
0.47
0.65
0.10
0.23
6.70
50 60 70 80
Speed
0.06
1.15
0.44
0.21
0.12
0.28
0.26
90 100 110
16
-------
Not-to-exceed testing is designed to also include transient emission measurement. As described above, this
might include any segment of normal operation above some minimum sampling period. We would not intend for this to
include engine starting, extended idling, or other cold-engine operation. Table 4 shows a wide variety of transient
emission levels from the two test engines. These could be considered as not-to-exceed measurements to evaluate
whether an engine is meeting emission standards. Several segments included in the table include emission
measurement after starting an engine that had soaked for up to 20 minutes in the lab. This can significantly increase
emission levels, depending on how long the engine runs in open loop after starting. This seems to be especially
important for CO emissions. Even with varied strategies for soaking and warming up engines, emission levels are
generally between 1 and 2 g/hp-hr THC+NOx and between 4 and 13 g/hp-hr CO. Emission levels don't seem to vary
dramatically between cycle segments, even where engine operation is significantly different.
Table 4
Transient Emission Measurements from SwRI Testing
Engine
Mazda
GM
Test Segment
"typical" forklift (5 min.)
"high-transient" forklift (5 min.)
highway certification test
backhoe/loader cycle
"typical" forklift (5 min.)
"high-transient" forklift (5 min.)
highway certification test
backhoe/loader cycle
THC+NOx
g/hp-hr
2.0
1.3
1.2
1.3
1.3
2.0
1.0
1.0
CO,
g/hp-hr
5.7
4.3
4.6
9.1
9.5
12.6
4.4
3.8
Notes
hot start
hot start
hot start
20-minute soak before test
hot start
hot start
3-minute warm-up; 2-minute soak
3 -minute warm-up; 2-minute soak
D. Durability of Emission-Control Systems
SwRI tested engines that had already operated for several thousand hours with functioning emission-control
systems. Before being retrofitted with catalysts and electronic fuel systems, these engines had already operated for
5,000 and 12,000 hours, respectively. The tested systems therefore provide very helpful information to show the
capability of the anticipated emission-control technologies to function over a lifetime of normal in-use operation.
The testing effort required selection, testing, and re-calibration of installed emission-control systems that were
not designed specifically to meet emission standards. These systems were therefore not necessarily designed for
simultaneously controlling NOx, HC, and CO emissions, for lasting 5,000 hours or longer, or for performing effectively
under all conditions and all types of operation that may occur. The testing effort therefore included a variety of
17
-------
judgments, and adjustments to evaluate the emission-control capability of the installed hardware. This effort
highlighted several lessons that should help manufacturers design and produce durable systems.
Selecting engines from the field provided the first insights into the functionality of these systems. Tailpipe
ppm measurements showed that several engines had catalysts that were inactive (or nearly inactive). These units were
found to have loose catalyst material inside the housing, which led to a significant loss of the working volume of the
catalyst and exhaust flow bypassing the catalyst material. This very likely resulted from a straightforward production
error of improperly assembling the catalyst inside the shell.13 This is not an inherent problem with catalyst production
and is easily addressed with automated or more careful manual production processes. The catalyst from the GM engine
selected for testing had also lost some of its structural integrity. Almost 20 percent of the working volume of the
catalyst had disappeared. This catalyst was properly re-assembled with its reduced volume for further testing. This
experience underscores the need for effective quality-control procedures in assembling catalysts.
Substituting a new catalyst on the aged system allowed emission measurements that help us estimate how
much the catalysts degraded over time. This assessment is rather approximate, since we have no information about the
zero-hour emissions performance of that exact catalyst. The new catalysts, which were produced about three years later
under the same part numbers and nominal characteristics, generally performed in a way that was consistent with the
aged catalysts. Not surprisingly, the catalyst with the reduced working volume showed a higher rate of deterioration
than the intact catalyst. Both units, however, showed very stable control of NOx and HC emissions. CO deterioration
rates were generally higher, but the degree of observed deterioration was very dependent on the particular duty cycle
and calibration for a given set of emission measurements.
Measured emission levels from the aged catalysts shows what degree of conversion efficiency is possible for
each pollutant after several thousand hours of operation. The emission data from the new catalysts suggest that
manufacturers would probably need to target low enough zero-hour CO emission levels to account for significant
deterioration. The data also show that catalyst size is an important factor in addressing full-life emission control. The
nominal sizes of the catalysts on the test engines were between 50 and 55 percent of total engine displacement. We
would expect manufacturers to reduce catalyst size as much as possible to reduce costs without risking the possibility of
high in-use emissions.
Another important issue relates to degradation associated with fuel impurities, potential lack of maintenance,
and wear of oxygen sensors. Fuel system components in LPG systems are prone to fuel deposits, primarily from
condensation of heavy hydrocarbon constituents in the fuel. The vaporizer and mixer on the test engines showed a
typical degree of fuel deposits from LPG operation. The vaporizer remained in the as-received condition for all
emission measurements throughout the test program. Emission tests before and after cleaning the mixer give an
indication of how much the deposits affect the ability of the closed-loop fueling system to keep the engine at
stoichiometry. For the GM engine, the combined steps of cleaning the mixer and replacing the oxygen sensor improved
overall catalyst efficiency on the C2 duty cycle from 50 to 62 percent for NOx. CO conversion efficiency improved
18
-------
only slightly. For the Mazda engine, the single step of cleaning the mixer slightly decreased average catalyst efficiency
on the C2 duty cycle for both NOx and CO emissions (see Table 5). These data show that closed-loop fueling systems
can be relatively tolerant of problems related to fuel impurities.
Table 5
Average C2 Catalyst Conversion Efficiencies Before and After Maintenance
Engine
GM
Mazda
Pollutant
NOx
CO
NOx
CO
before maintenance
50.4 %
95.5 %
62.9 %
99.4 %
after maintenance
61.7%
96.0 %
60.0 %
99.0 %
Manufacturers may nevertheless be concerned that some in-use operation can cause fuel deposits that exceed
the fuel system's compensating ability to maintain correct air-fuel ratios. Two technologies are available to address this
concern. First, the diagnostic system we are considering would inform the operator if fuel-quality problems are severe
enough to prevent the engine from operating at stoichiometry. A straightforward cleaning step would restore the fuel
system to normal operation. Manufacturers may also be able to monitor mixer performance directly to detect problems
with fuel deposits, rather than depending on air-fuel ratios as a secondary indicator. In any case, by informing the
operator of the need for maintenance, the diagnostic system reduces the chance that the manufacturer will find high in-
use emissions that result from fuel deposits.
The second technology to consider is designed to prevent fuel deposits from forming. A commercially
available device regulates fuel temperatures by heating the vaporizer with varying engine coolant flow. Keeping the
fuel temperature above a given temperature reduces the likelihood that the heavy hydrocarbons will condense out
downstream in the vaporizer or other fuel system components.
Manufacturers have also raised the question of whether oxygen sensors are durable enough to provide reliable
performance for most of the engine's useful life. Oxygen sensors for highway engines have undergone significant
improvements in reliability and durability to provide better long-term control of emissions. We request comment on
how nonroad engines may cause oxygen sensors to be more susceptible to premature aging than oxygen sensors in
highway vehicles.
Maintaining the integrity of the exhaust pipe is another basic but essential element of keeping control of air-
fuel ratios. Any leaks in the exhaust pipe between the exhaust valves and the oxygen sensor would allow dilution air
into the exhaust stream. The extra oxygen from the dilution air would cause the oxygen sensor to signal a need to run at
a air-fuel ratio that is richer than optimal. If an exhaust leak occurs between the oxygen sensor and the catalyst, the
19
-------
engine will ran at the right air-fuel ratio, but the extra oxygen would affect catalyst conversion efficiencies. As
evidenced by the test engines, manufacturers can select materials with sufficient quality to prevent exhaust leaks over
the useful life of the engine.
20
-------
REFERENCES
1. "Three-Way Catalyst Technology for Off-Road Equipment Powered by Gasoline and LPG
Engines," by Jeff White, et al, Southwest Research Institute, prepared for California ARE,
California EPA, and South Coast AQMD, (SwRI 8778), April 1999.
2. "Evaluation of Emissions Durability of Off-Road LPG Engines Equipped with Three-Way
Catalysts," by Vlad Ulmet, Southwest Research Institute, SwRI 08.03661, November 2000,
(Docket A-2000-01, document U-A-07).
3. SwRI 08.03661.
4. See SwRI 08.03661 for further description of the statistical techniques for selecting transient
cycle segments.
5. SwRI 8778.
6. SwRI 08.03661.
7. SwRI 08.03661.
8. SwRI 8778.
9. SwRI 8778.
10."Durability Experience with Electronic Controlled CNG and LPG Engines," A. Lawson et al.,
February 2, 2000 (Docket A-2000-01, document U-D-02).
11."Exhaust Controls Available to Reduce Emissions from Nonroad Heavy-Duty Engines," by
Kevin Brown, Engine Control Systems, in Clean Air Technology News, Winter 1997 (Docket A-
2000-01, document U-A-02).
12. "Case Study: The Results of EVIPCO's GM 3.0 liter Certified Engine Program," presented by
Josh Pietak, February 2, 2000 (Docket A-2000-01, document II-D-11).
13.See SwRI 08.03661 for a further description of the catalyst damage observed.
21
------- |