A final report to the U.S. Environmental Protection Agency
on the project:
EFFECTS OF AMBIENT CONDITIONS ON THE EMISSIONS
OF TWO- AND FOUR-STROKE AIR-COOLED ENGINES
for the period: 5/27/95 to 2/28/97
carried out by Michigan State University
under Cooperative Agreement: CX824083-01-0
by
Professor Giles Brereton
Engine Research Laboratory
Department of Mechanical Engineering,
Michigan State University, East Lansing, MI 48824-1226
July 2nd, 1997
Principal Investigator: Prof. Giles J. Brereton (517) 432-3340
i
-------
Table of Contents
Table of Contents i
1. Background and Cooperative Agreement Overview 1
2. Objectives for the Project 1
3. Achievements during the Project 2
3.1 Facility/Equipment Construction and Improvement 2
3.2 Technical Achievements 3
3.3 Educational Benefits 4
4. Suggestions for Future Work 5
Appendix A. SAE Paper 97P382 7
Appendix B. SAE Paper 961739 19
-------
1. BACKGROUND AND COOPERATIVE AGREEMENT OVERVIEW
The Clean Air Act Amendments of 1990 require the EPA to determine whether non-
road engines are significant contributors to ozone or carbon monoxide levels in U.S. non-
attainment areas. As a result, EPA is currently studying the impact of non-road engine
and equipment emissions on non-compliance areas, and the potential for cost-effective
emission control of these sources. In support of this, EPA is investigating emissions from
such currently unregulated sources such as outboard marine engines, small gasoline
engines, and other non-road engines and equipment, through a cooperative agreement
between MSU and EPA's National Vehicle and Fuel Emissions Laboratory.
The intention of this particular agreement was to: t) develop a climate-controlled
small engine engine research facility; and u) provide EPA with access to the expertise
and resources needed to investigate emissions from non-road engine sources through
cooperation with researchers and students from the College of Engineering at Michi-
gan State University. This program, originally proposed as a 12-month program, was
extended to 18 months. A report on the completed 18 month program follows.
2. OBJECTIVES FOR THE PROGRAM
The technical objectives which were set for the report period are listed below:
• to establish a climate controlled engine test facility for emissions tests of small
air-cooled engines;
• to measure exhaust emissions (hydrocarbons, oxides of nitrogen, and carbon
monoxide) from small utility engines using SAE recommended practice J1088
under controlled climatic conditions, to assess the sensitivity of emissions to dif-
ferent ambient conditions;
• assessment of emissions reduction technologies and evolving low-emissions small
engines.
At the beginning of the project, it became clear that a number of items whose pur-
chase was planned under earlier related EPA Cooperative Agreements at University of
1
-------
Michigan were not purchased and so were not transferred to Michigan State University
for their planned use on this project. Consequently, a revised objective was to build
a number of those items from scratch (data acquisition unit, dynamometer modified
to replace an existing inadequate torque sensor with an inline torque sensor). The
time required to carry out this work, together with construction delays in installing
heating, chilling, and humidifying capabilities in the MSU engine test cell, resulted in
the project lasting six months longer than planned.
In the course of the project, the outcome of the EPA's negotiated regulation discus-
sions de-emphasized the need to assess emissions reduction technologies. Consequently,
little progress was made towards this original objective and efforts were, instead, fo-
cused on understanding better the cause and effect of changing ambient conditions
and emiossions. Thus, the working objectives of this cooperative agreement underwent
some revision as the project proceeded.
3. ACHIEVEMENTS DURING THE PROJECT
3.1 Facility/Equipment Construction and Improvement
In the course of this project, a climate controlled small-engine test facility, with
controllable ambient temperature and humidity (costing approximately $40,000), was
built at MSU as the university's contribution in cost sharing. The facility was equipped
with a Horiba emissions bench (CO, CO2, HC, CH±, NOx) and a heated FID analyzer
(1985 model year), also secured during the project period. A Micro-Dyn 15 hydraulic
dynamometer was rebuilt and fitted with an in-line torque meter, allowing accurate
dynamic torque measurements to be made. A fuel-flow meter and delivery system
was also constructed, built around a Micro-Motion DM-6 Coriolis meter, as a further
contribution to project instrumentation. A personal-computer data acquisition system
was also designed and built to control and sample data from these instruments during
the project.
2
-------
3.2 Technical Achievements: New Emissions Measurements and Analyses
The apparatus described above was used to make series of measurements of the
performance and emissions of two hand-held engines, a functions of changing ambient
temperature and pressure. In addition, a considerable amount of theoretical analysis
and modeling of small-engine thermal-fluid processes and emissions was carried out.
This combined work resulted in two SAE papers, presented at the 1996 and 1997
Off-Highway and Powerplant Congress in Indianapolis and Milwaukee. The particular
engines tested were the Homelite Super 2, two-stroke chainsaw engine and the Ryobi
Pro-4-Mor four-stroke utility engine. The two-stroke engine was a Model #246Y with
a 3.642 cm bore, 3.12 cm stroke, 32.4 cm3 displacement and so a Class IV engine
in the hand-held category for federal emissions standards. It featured reed valves,
four transfer ports for loop scavenging, and a flat-top piston, with fuel metered by a
diaphragm carburetor. The nominal power output of the engine was about 1 kW at
wide-open throttle. The four-stroke engine was1 an overhead valve Ryobi Pro-4-Mor
with a 3.2 cm bore, 3.26 cm stroke, 26.2 cm3 displacement rated at 0.75 kW at 7000
rpm, with fuel metered by a Zama diaphragm carburetor. This engine was also a Class
IV engine in the hand-held category for federal classification purposes.
Details of the work are described in the attached pre-prints of those papers, presented
on the following pages. Some highlights of this work include:
(») a new correction factor for production of NOx at standard humidity, presented as
Kjj = Kjj(u, AFR), as an improvement on the existing Kjj = Kjj(u) correction
described in the Code of Federal Regulations ( motivated by the observation that,
for the four-stroke engine tested, NOx halved, as w was increased from 0.006 to
0.014);
(it) new analyses which describe the effects of ambient conditions on carburation
(the air-fuel ratio metered by fixed-jet carburetors) and formation of combustion
products;
3
-------
(tit) the strong dependence of two-stroke hydrocarbon emissions on ambient humidity
(50% increase as w increases from 0.006 to 0.014), which suggest the need for a
tighter acceptable range of certification-testing conditions for two-stroke engines
than for four-stroke engines;
(tu) a procedure for tuning carbureted engines to design air-fuel ratios, using the
exhaust-gas %CO and the prevailing ambient temperature and pressure;
(u) a reasonably large data set of emissions measurements for these two engines, under
a range of controlled climatic conditions.
The raw and processed emissions data for these engines have been stored in EXCEL
file format and will be forwarded to EPA. The references for these papers are:
Brereton, G. J., DeAraujo, A. & Bertrand, E., "Effects of ambient conditions on the
emissions of a small carbureted four-stroke engine," SAE Paper No. 961739.
Brereton, G. J., Bertrand, E. & Macklem, L. "Effects of changing ambient humidity
and temperature on the emissions of carbureted two- and four-stroke hand-held
engines," SAE Paper No. 97P382.
Bertrand, E. B. "Effects of changing ambient humidity and temperature on the
emissions of a small carbureted overhead valve four-stroke utility engine," Master's
Degree Thesis, Dept. of Mechanical Engineering, Michigan State University, July
1997.
3.3 Educational Benefits
The project has been beneficial both as a research/development project and as a
means of educating students and researchers on engine emissions and their measure-
ment. In the course of the project, 5 students have participated on different aspects
of the project. Undergraduate Christopher Gross contributed to much of the initial
laboratory setup, while Masters candidate Michael deSouza participated in the design
of the climate control facility, and the fuel-flow cart. Undergraduate Alex DeAraujo
4
-------
undertook a directed-study project for two semesters, in which he carried out prelimi-
nary engine tests and analyzed 'round-robin1 test data measured in different laborato-
ries under different climatic conditions. Graduate students Loren Macklem and Erik
Bertrand carried out nearly all the varying-humidity and varying-temperature emis-
sions tests, with Erik specializing in the four-stroke studies, while Loren focused on the
two-stroke engine. Erik was also instrumental in rebuilding and redesigning a hydraulic
dynamometer, a fuel cart, and a data acquisition system. One of these students partici-
pated in the study as a Master degree project, while the other wrote his Masters Degree
Thesis on this research. This Thesis will be available shortly. In addition, Michigan
State University ran several feature stories on the work carried out under this project,
in publications at the College of Engineering levels, and at the University-wide level.
4. SUGGESTIONS FOR FUTURE WORK
Under the program summarized in this report, much has been learned about the
emissions of existing hand-held engines under changing ambient conditions. While the
findings of this study will apply to future generations of simple carbureted engines, they
would not be valid for proposed technologies such as direct and port injection systems,
or for engine-out emissions with exhaust after-treatment systems (i.e. catalytic con-
verters combined with air injection and cooling), and may require modification for use
with proposed alternate fuels. The theoretical approaches developed in this work are
likely to be even more applicable to larger off-road and utility engines, in which charge
homogeneity is likely to be greater. It would be particularly interesting to apply these
approaches to study of emissions from two-stroke engines used in personal watercraft
as well as conventional marine engines.
To assure carefully developed regulatory control, it will be essential to have flexible
laboratories which can adapt rapidly to study of the ever-changing technologies and
products which are of concern to global emissions levels. It is also likely that more
detailed measurements of emissions may be required, such as concentrations speciated
by molecular weight, to allow the emissions of particular substances (i.e. butadienes,
5
-------
benzenes) to be analyzed, or the fate of additives like two-stroke oils to determined. As
concern has grown about small particulate emissions, preliminary research would seem
to indicate that correlations between the size of the particulate matter and its compo-
sition characterize certainly varies with engine size (in diesel engine studies). Similar
studies could be initiated for off-road and utility engines, as practically no work seems
to have been done in this area. The complexity of hydrocarbon combustion processes in
engine cylinders are such that capabilities for measuring individual substances remain
much greater our ^ability to predict them reliably from models and computations of
combustion processes. Therefore laboratory measurement programs appear to be an
essential way of examining these emissions issues.
6
-------
Effects of Changing Ambient Humidity and Temperature
on the Emissions of
Carbureted Two- and Four-Stroke Hand-Held Engines
G. J. Brereton, £. Bertrand & L. Macklem
Engine Research Laboratory, Michigan State University
Abstract
Effects of changing ambient humidity and temperature
have been studied on the performance and emissions of
a hand-held two-stroke and a hand-held four-stroke en-
gine. The main effect of changes in ambient conditions is
to change the intake air density and therefore the air-fuel
ratio metered by the carburetor. TVends in the effects of
humidity and temperature on emissions are predicted rea-
sonably well by theoretical thermodynamic models. They
suggest an improved correction for the dependence of NOx
on ambient conditions, as a function of both humidity and
operational air-fuel ratio, which appears to collapse NOx
production data better than the existing Kh correction
factor. They also suggest a simple procedure for tuning en-
gines to design air-fuel ratios using the measured exhaust-
gas %CO, which takes into account the prevailing ambient
conditions.
1. Introduction
The dependence of performance and emissions of dif-
ferent internal combustion engines on ambient conditions
can vary widely and is closely related to the precision with
which the air-fuel ratio can be controlled. In modern auto-
mobile engines equipped with electronic fuel injection and
mass airflow sensors, with feedback from exhaust-gas oxy-
gen sensors, charge preparation and spark timimg are re-
fined as often as every engine cycle to provide the optimal
performance which has been programmed for the prevail-
ing driving conditions. In contrast, today's small engines
developed for use in hand-held applications typically me-
ter air and fuel using only a fixed main-jet carburetor, of
the float or diaphragm type. While the ambient conditions
under which automobile engines operate are compensated
for by sophisticated on-board computer control, the re-
sponse of small-engine emissions to changing ambient con-
ditions depends on the characteristics of the carburetor,
heat transfer to/from the engine and other factors, many
of which differ from one engine to another. For this reason,
ranges of acceptable ambient conditions are specified for
emissions-compliance tests in Federal Regulations,1 and
corrections have been proposed to relate measurements
of emissions/performance under prevailing ambient con-
ditions to those expected at reference conditions. As emis-
sions regulations for off-road and utility engines are devel-
oped, one part of the rulemaking procedure is to specify
the acceptable ranges of ambient conditions for certifica-
tion tests and any corrections to be used. Since the effects
on emissions of changing ambient conditions on small en-
gines are not well known, and the effects potentially so
complex, experimental studies are necessary to establish
acceptable ranges of test conditions and valid engineering
models for these effects.
The main premise followed in this research is that the
principal characteristics of hand-held engines that distin-
guish them from automotive engines are: their carbura-
tion; the design air-fuel ratios at which they operate; and
the reduced control over combustion temperatures (on ac-
count of the relatively poor mixing of the charge and the
availability only of air cooling). In principle, ideal fluid
mechanics and equilibrium thermodynamic analyses can
be used to estimate how ambient conditions determine the
charge composition, and subsequent idealized compression
and combustion events. The extent to which these theo-
retical approaches can be applied to thermo-fluid processes
under boundary conditions as complex as those of small en-
gines is not well known. However, when these approaches
are reliable, theoretically derived models for small-engine
emissions and performance under different ambient condi-
tions could be used to identify acceptable ranges of test
conditions which are straightforward to control, and to
develop corrections for 'effects of changing test conditions
which are difficult or expensive to control.
Several previous studies have been carried out on the
dependence of emissions on ambient conditions. The stud-
ies of Brown et a/.2 and Robison3 in the 1970's concerned
experimental evaluations of corrections for NOx emissions
tested when ambient humidity differed from its standard
value (10.71 grams H^Ofkg dry air). The correction pub-
lished in the Code of Federal Regulations1 shortly after
these studies still takes the form
K" = 1 - 0.0329 (H - 10.71) ^
where H is the specific humidity in grams H^O/kg dry air.
The basis for this correction was that it fitted experimental
data for water-cooled, multi-cylinder automobile engines of
the late 1960's on typical driving cycles, with basic carbu-
retor settings in the 15.5:1 to 16.5:1 range. It may not be
valid for today's hand-held engines (or even today's auto-
mobile engine) for several reasons. Oxides of nitrogen are
-------
non-equilibrium products of combustion, frozen during ex-
pansion, after formation through a complicated series of
decomposition/formation reactions. The rates at which
these reactions proceed depend strongly on the flame tem-
perature. On average, this temperature depends on the
composition of the charge (humidity, air-fuel ratio), its in-
take temperature/pressure, the compression ratio, and the
cylinder heat transfer. However, in gasoline engines, NOx
production is thought to result predominantly from the
charge in the cylinder that burns first (t.e. the charge at
the spark plug [Blumberg k Kummer4|), in which case
the composition (or cycle-to-cycle distribution in composi-
tions) of the charge at that location might be of principal
importance. Because of these complexities, an empirical
NOx correction developed for one class of engine might
only be expected to apply to others either fortuitously
or because the variables upon which flame temperatures
are most sensitive match those of the empirical correction.
Moreover, most hand-held engines operate at air-fuel ra-
tios and with flame temperatures very different from those
in the engines for which (l) is appropriate. While it is ex-
pected that CO emissions reflect the metered air-fuel ratio
and, in turn, its dependence on the ambient, it is quite
likely that emissions of HC have some dependence on am-
bient conditions. However, little research has been carried
out in this area.
A number of studies have considered the likely global
emissions production from small engines (Donohoe et at.,5
Hare et a/.6), typical emissions levels and engine duty cy-
cles (Morgan Ac Lincoln,7 Coates & Lassanske,8 Laimbock
ic Landerl,9 Brereton et a/.10) and the effectiveness of
strategies to reduce those emissions (White et af.,11 Sun et
al.,12 Mooney et ai.13). However, relatively little attention
has been paid to testing procedures, beyond the work of
the SAE J1088 task force, though Brereton et a/.10 have
assessed the usefulness of transient test cycles chosen to
mimic the duty cycles of engines in actual use.
Recently Brereton et a/.14 analyzed emissions data of a
Kohler CV 12.5 S 4-stroke single-cylinder overhead-valve,
gasoline-fueled engine used in a 'round-robin' emissions
test in 8 different laboratories over 12 months, spanning
a range of different ambient conditions. They concluded
that one third of the scatter in observed air-fuel ratio and
CO emissions could be attributed to effects of ambient
pressure, temperature, and humidity on the carbureted
air-fuel mixture, and could be removed by a correction.
While this study was a first attempt to characterize ambi-
ent effects on emissions of small engines, the'round robin'
nature of the testing introduced considerable uncertainty
through the use of 8 different dynamometers and sets of
emissions measurement instrumentation. For this reason,
this second study was carried out using two engines and
dynamometers with the same measurement equipment in a
single, climate controlled laboratory. The main objectives
of this study were:
(») to measure the emissions of hand-held two-stroke
and four-stroke engines under controlled ambient
conditions, in which each of the humidity and tem-
perature were varied systematically;
(it) to interpret the observed emissions and performance
data in terms of thermodynamic models for carbu-
ration and combustion; and
(ut) to propose improved corrections and guides to the
effects of ambient conditions on hand-held engine
emissions and performance.
2. Experimental Approach
Emissions measurements of a two-stroke chainsaw en-
gine and a four-stroke hand-held utility engine were made
according to the SAE J1088 test procedure, in the climate-
controlled test cell at Michigan State University's Engine
Research Laboratory. The two-stroke engine was a Home-
lite Super 2, Model #246Y with a 3.642 cm bore, 3.12 cm
stroke, 32.4 cm3 displacement and so a Class IV engine
in the hand-held category for federal emissions standards.
This engine featured reed valves, four transfer ports for
loop scavenging, and a flat-top piston, with fuel metered
by a diaphragm carburetor (Zama). The nominal power
output of the engine was about 1 kW at wide-open throt-
tle. The four-stroke engine was an overhead valve Ryobi
Pro-4-Mor with a 3.2 cm bore, 3.26 cm stroke, 26.2 cm3
displacement rated at 0.75 kW at 7000 rpm, with fuel me-
tered by a Zama diaphragm carburetor. This engine was
also a Class IV engine in the hand-held category for federal
classification purposes.
Emiisioru Monitoring Equipment
HC NO* CO C02
Fuel
FlcrwCan
lT
Dynamometer
Engine
t y t
tzt
Data Processing PC
Dilution
Tunnel
Voltages
Temperature and Humidify
Controlled Room
Exhaust Dueling
Fig. 1. Experimented facility.
The emissions of these engines were measured in a test
cell in which the ambient temperature and humidity could
be controlled in several ways. Make-up air (to compensate
for exhausted products of combustion) could be heated or
chilled to any desired temperature T between -7°C and
40°C and humidified (above the ambient level) to specific
humidities w as high as 0.016 kg water/kg air. In typical
experiments, either: i) T and u> were held constant; t»)
-------
T was held constant while u was varied through its at-
tainable range; or ut) ui was held constant while T was
varied through the desired range. In the emissions test
described in this paper, T and cii were varied in the sec-
ond and third modes described. Measurements of humid-
ity and temperature were made at the engine intake using
a Vaisala HMP233 humidity meter and a thermocouple.
The two-stroke engine was operated on a Homelite elec-
tric dynamometer/controller at idle and wide-open throt-
tle, under constant-speed control. The four-stroke engine
ran at the same modes of the SAE J1088 test procedure on
a Micro-Dyn 15 hydraulic dynamometer. The fuels used
were either Indolence Clear or Amoco 93 Octane gaso-
line. In both cases, fuel was supplied to the engine from
a constant-head delivery system which measured (using a
Micro Motion DM6 Coriolis meter) the fuel-flow required
to maintain a float at constant height in the delivery cham-
ber.
Emissions were measured using a dilute sampling ap-
proach, with a critical-flow venturi nozzle in a dilution
tunnel metering a constant flow-rate of the sum of the
engine exhaust gas and room air. A continuous sample of
the diluted exhaust was fed to a Horiba emissions bench
equipped with CO, CO2, HC, NOx, and CH+ analyz-
ers. Typical dilution factors in these tests ranged from 10
to 30. Emissions measurements were reduced to specific
and brake-specific values of each species as recommended
in the Code of Federal Regulations.1 Air-fuel ratios were
inferred using Spindt's method.15 The implicit assump-
tion of negligible O2 content in the exhaust gas, used in
dilute sampling approaches, probably led to air-fuel-ratio
estimates of one-half to one below their true values.
The test procedure followed was to: ») allow the room
temperature and humidity to equilibrate to the desired ini-
tial level; and «) steadily raise/lower one of the temper-
ature or humidity while holding the other constant. In
tests of this kind, temperatures could be varied by about
AT = 30°C and humidities by Aw = 0.01 in a time pe-
riod of about 2 hours. This time period was thought to
be long enough for engines to maintain quasi-steady-state
operation during transients, while sufficiently short that
changes in atmospheric pressure were very small. Thus
the effects of changing ambient temperature and humid-
ity on emissions could be studied under varying tempera-
ture at constant humidity and pressure, and under varying
humidity at constant ambient temperature and pressure.
Tests of this kind were run for each engine three or four
times, for operation at idle and at wide-open throttle. It
is worth noting that, in the course of tlu3e tests, continu-
ous engine operation for longer periods occasionally led to
minor breakdowns or partial seizures, which raised some
questions about whether the engine thermal processes had
reached a truly steady state in the course of these tests.
3. Thermodynamic Analysis
Thermodynamic analyses of iso-octane combustion in
air can be used to infer effects of ambient conditions on
flame temperatures and products of combustion when as-
sumptions of thermodynamic equilibrium are valid. In par-
ticular, for given ambient humidities, pressures and tem-
peratures, the air-fuel ratio metered by the carburetor and
the subsequent compression and combustion processes can
be modeled (under certain idealized conditions) to yield
the theoretical dependence of flame temperatures and com-
bustion products on ambient conditions. While there is no
question that thermodynamic processes in small engines
are far too complicated to be treated exactly by analy-
ses of this kind, these analyses can be useful in explaining
some trends in measured data, and may be able to predict
quite accurately details of some combustion products.
3.1 Carburation
As a first stage in the thermodynamic analysis, the me-
tering of air flow through the carburetor was modeled as
a frictionless, iso-energetic, compressible, one-dimensional
process. Using subscripts i and t to refer to the venturi
intake and throat respectively, for the usual case of intake
air velocities being much lower than throat velocities, it
can be shown from first principles16 that
''"moist air = CD At \JtPi{Pi - Pt) • $ ("7. Pt/Pi) (2)
Here m is the mass flow rate, $ is a factor which incor-
porates effects of compressibility and 7 is the ratio cp/c„
of the moist air. In typical single-cylinder small engines,
with strongly pulsating intake flows, 1 might possibly be
an average property of a moist air/fuel mixture, since there
may be periodic backflow of charge 'upstream' of the car-
buretor. Effects of friction and flow non-uniformity are
compensated for by introducing a steady-flow discharge
coefficient Co, which typically varies smoothly with air-
flow Reynolds number.
For operation at wide-open throttle (WOT), when P, -
Pt is likely to be constant and the carburetor throat is the
flow-limiting area, the carburetor regulates the quantity:
mmoiat air I\fpl- Assuming air and water vapor to be ideal
gases, with p
-------
transient temperature tests, the (measured) fuel tempera-
ture remained almost at its initial temperature throughout
each test, so the fuel density and m
-------
customized STAN J AN code. In Figs. 6, 7 and 8, the depen-
dence of flame temperature (and likely NOx production)
is plotted against ambient humidity u and temperature T.
In Fig. 6, the conventional behavior of flame temperature
against operating air-fuel ratio is shown, with Tf|omt ris-
ing with increasing Tambimt, and 7)(ame reaching its peak
value (also the NOx production peak) about an air-fuel
ratio rich of stoichiometric. The effect of metering fuel
with a carburetor, and so operating at a fixed design air-
fuel ratio, is to lower the flame temperature (and NOx)
with increasing ambient temperature when, the design air-
fuel ratio is less than 13 — the reduction in flame temper-
ature caused by the carburetor's richening of the charge
outweighs the increase in flame temperature on account of
the increased ambient temperature, leading to the behav-
ior shown in Fig. 7. A uniform effect of reducing Tfiatne
by raising w is shown in Fig. 8 — this ambient effect is
scarcely changed by carburation.
For fuel-rich combustion, the equilibration of CO and
COi during expansion and subsequent exhaust processes is
usually modeled as the equilibrium result of the simplified
'water-gas3 reaction
co2 + h2^co + h2o
with equilibrium constant K, as in
[co][H2o\
(COa |[tf3
= K
(6)
(7)
Spindt15 suggests a value of K = 3.5 which matches avail-
able laboratory measurements. FYom equilibrium thermo-
dynamic theory, K takes this value for this reaction when
T = 1700 K, from which it may be inferred that the equi-
librium CO and COi concentrations at 1700 K remain
frozen at those levels when the temperature is reduced.
Thus, if this 'water-gas' reaction is the only one that af-
fects CO and C02 levels, then the calibration of K = 3.5
is effectively a calibration of the temperature at which CO
and COi are frozen at their equilibrium levels in 1C en-
gine expansion processes. This calibration may be a robust
one since it is concerned primarily with the limits of equi-
librium thermodynamics and only secondarily with engine
processes. Using equilibrium STANJAN calculations for
the combined analysis at 1700 K, with the products limited
to: N2, CO, CO2, H2 and H^O, or by combining carbon,
hydrogen, and oxygen balances in the moist-air/iso-octane
combustion equation:
C&Ui% -4- a(^2 3.767^2) -f- bH20 — ...
with (7) (as in Heywood16), the variation of the ratio of
CO and CO2 concentrations with ambient conditions (a
and £>) may be computed as a function of design air-fuel
ratio. It is plotted against u> and T in Figs. 9 and 10.
While the theoretical variation of C0jCO2 with chang-
ing humidity is slight, the dependence on ambient temper-
ature can be significant, particularly at low air-fuel ratios.
This result is especially relevant to the practice of tuning
carbureted engines to a target exhaust-gas CO concen-
tration, since, to achieve the same design air-fuel ratio,
the target % CO must take a new value whenever ambi-
ent conditions change. In theory, an ambient tempera-
ture/pressure dependent correction to target %CO could
be devised based on the analyses of this paper.
4. Experimental Results
The climatic conditions achieved under: ») a typical
constant T, varying u> test; and it) a representative con-
stant iij, varying T test; are shown in Fig. 11. While the cli-
mate control equipment experienced some difficulty main-
taining constant humidity at low temperatures, the ob-
served te3t conditions were very close to their prescribed
values. The measured performance and emissions results
for the two- and four-stroke engines under these two tests
are presented below for operation at wide-open throttle.
The corresponding data at, idle are difficult to interpret
theoretically and often make a relatively small contribu-
tion to weighted averages of brake-specific emissions, so
are not reported in this paper.
The torque produced by each engine is shown in Fig3.12
and 13 as functions of varying ambient humidity and tem-
perature. In all plots of experimental measurements, filled
symbols denote four-stroke engine data whereas unfilled
ones indicate two-stroke engine results. Since these tests
were performed at constant engine speed, the power out-
puts depend identically 011 changing ambient conditions.
There is a clear trend of decreasing torque with increas-
ing temperature for the four-stroke engine, while the com-
panion two-stroke data seems to vary erratically. Because
torque and power tend to scale on the mass of oxygen sup-
plied to the cylinder, for fuel-rich combustion,21 it is likely
that the decrease of air density with temperature explains
this trend. Measurements of torque also show a trend to-
wards lower values with increasing humidity over the range
tested, for the two-stroke engine, though this trend is only
followed by the four-stroke engine at high humidities. This
result is also consistent with a reduction in oxygen supplied
to the cylinder with increasing w, as expressed in equation
(4).
The dependence of operating air-fuel ratio (inferred by
Spindt's method15) on ambient conditions is shown in Figs.
14 and IS for the two- and four-stroke engines tested. In-
creasing ambient temperature and humidity both appear
to lower the air-fuel ratio slightly, presumably reducing
the amount of air entering the cylinder by: t) reducing
its density; and ii) lowering the proportion of dry air in
the charge. This result is particularly significant since the
air-fuel ratio plays such an important role in determining
the combustion products. These trends are consistent with
predictions based on carburetor model equation (4), and
Figs. 2 and 3, though the slopes of the theoretical curves
do not match exactly those of the experimental data.
-------
Production of the regulated emission species CO (on
a brake-specific basis) is plotted against changing ambi-
ent conditions in Figs. 16 and 17. There is a slight trend
towards increasing brake-specific CO with increasing tem-
perature/humidity, caused partly by the reduced power
and partly by the increased air-fuel ratio. As was ob-
served for torque and air-fuel ratio, the two-stroke data
vary much more erratically during temperature transients
than humidity transients, possibly on account of a com-
plex charge-homogeneity dependence on temperature. The
emitted levels of oxides of nitrogen (NOx) are shown in
Figs. 18 and 19. In the case of the four-stroke engine op-
erating in the 13 < AFR < 14 range, the effect of in-
creasing humidity from w = 0.006 tow = 0.015 is to halve
NOx emissions. For the two-stroke engine operating in
the 7.5 < AFR < 8.5 range, over which NOx produc-
tion is almost negligible, the effect of increasing humidity
is greatly reduced. The companion dependences of brake-
specific NOx on ambient temperature are only slight and
do not reveal am obvious trend.
Emissions of hydrocarbons (HC) are plotted in Figs. 20
and 21 as functions of ambient temperature and humidity.
In the case of the four-stroke engine, HC emissions scarcely
vary with temperature. They do, in fact, increase signif-
icantly with rising humidity, though this effect is hidden
when plotted on the same graph as the (scavenging-loss
dominated) two-stroke HC emissions. In the case of the
two-stroke engine, the same erratic dependence on tem-
perature is seen (and attributed to varying charge inho-
mogeneity). The two-stroke HC emissions increase signif-
icantly with rising humidity, growing by over 50% aa u>
varies from 0.006 to 0.014. This trend might be explained
partly by the reduced air-fuel ratios metered by the car-
buretor at high humidities, though there is likely some ef-
fect of the reduced power (torque) and cylinder pressures
changing the engine's scavenging and flow characteristics.
Experimental measurements of the ratio CO/CO-2, as
functions of ambient humidity and temperature, are shown
in Figs. 22 and 23. They reflect the same trends (with
CO/CO2 inversely related to -4.F.R} as the inferred air-fuel
ratios in Figs. 14 and 15 and are plotted separately for ease
of comparison with the predicted CO/CO2 ratios of Figs. 9
and 10. The measured experimental values of T and w and
the design air-fuel ratios of 14.25:1 for the four-stroke and
10.85:1 for the two-stroke were used to compute the the-
oretical CO/CO2 ratios, which are also shown in Figs. 22
and 23. While the theoretical CO/COi ratio and the mea-
sured four-stroke data sire in excellent agreement, the ad-
ditional scatter in the two-stroke data makes it hard to tell
if the trends are correct. It is noteworthy that the design
air-fuel ratios which result in theoretical CO/CO2 ratios
of the correct size are significantly higher than those in-
ferred from Spindt's method and reported earlier (Figs. 14
and 15). This is probably caused, in part, by the inabil-
ity to measure emitted O2 accurately in dilute sampling
approaches and raises questions about the desirability of
using Spindt's method at all in such circumstances.
5. Interpretation of Results
Perhaps the most general observation on the experi-
mental data of section 4 is that the variation of emissions
and performance data with changing humidity is consid-
erably smoother than with temperature, particularly for
the two-stroke engine. This is noticeable in Figs. 13, 15,
17, and 21, in which the temperature dependences follow
scattered paths about the general trend in the data. Af-
ter checking the performance of thermocouples, then re-
covering almost identical 'scatter' on re-running several
tests, and on replacing ambient temperature with carbu-
retor venturi-inlet temperature, it was inferred that the
'scatter' was a real effect. In hand-held four-stroke engines,
the close coupling of the carburetor to the cylinder restricts
opportunities for mixing of air and fuel downstream of the
throttle plate; in two-stroke engines, wall-wetting within
the crankcase and transfer ports is also likely to inhibit
homogeneous charge formation. Since the charge in the
cylinder was likely to be quite inhomogeneous at ignition,
it was thought that changes in ambient temperature caused
changes in fuel evaporation, diffusion and transport rates,
with resultant changes in charge inhomogeneity and so per-
formance and emissions levels. Because charge homogene-
ity need not be a monotonic function of temperature, it
is plausible that the performance and emissions measure-
ments need not change monotonically or smoothly with
temperature either!
On the other hand, water vapor is almost perfectly di-
luted in air, has minimal effect on fuel evaporation, and
makes negligible changes to momentum or heat transfer
to/from air, since it has almost no effect on either the air
viscosity or Prandtl number. Therefore, changes in am-
bient humidity should make no change to charge homo-
geneity. We believe this is the reason that performance
and emissions vary smoothly with changing humidity, but
irregularly with changing ambient temperature. Conse-
quently, theoretical trends in emissions with changing hu-
midity might be a better match to experimental data than
those with changing ambient temperature.
The smooth variation of emissions with changing ambi-
ent humidity suggests that correction functions may suc-
cessfully relate emissions measured at an arbitrary humid-
ity to those which would be observed at a reference hu-
midity. Of the emissions species measured, NOx clearly
shows the greatest sensitivity to changing humidity. Since
NOx formation depends strongly on the flame temper-
ature, and therefore, from Fig. 6, on the operating air-
fuel ratio, it seems logical that a correction of the form
Kh = Kh(u, AFR] should be attempted. For carbu-
reted (and water-cooled) engines operating at AFRs close
to 16.5:1, the existing multiplicative correction of NOx to
its expected value at « = 0.0107 is:
K" ~ 1 - 32.9 (w - 0.01071) ^
From Fig. 8, it is clear that the dependence of flame tem-
perature on humidity is greater at small AFRs than at
-------
large ones, so that a larger correction is needed when cax-
buration is richer. A simple way to model this behavior is
to re-express the scaling factor 32.9 as constant/AFR, in
which case, constant = 32.9 x 16.5 and the revised, AFR-
dependent correction is:
K" = 1 - ^ (w - 0.01071) ^
Plots of NOx measurements corrected using the Federal
Kh{<>>) correction (6) and the proposed Kh[<», AFR) cor-
rection for the two-stroke engine (in the 7.5 < AFR < 8.5
range) and the four-stroke engine (over the 13 < AFR <
14 range) are shown in Figs. 24 and 25. The proposed
correction used in Fig. 25 clearly collapses these hand-held
engine data quite well, with corrected data a better ap-
proximation to a horizontal line than those of Fig. 24 (the
Federal correction).
Experimental and predicted variations of the CO/CO2
ratio were shown in Figs. 22 and 23, using the precise ex-
perimental values of « and T and a single assumed design
air-fuel ratio to make the prediction. The agreement be-
tween experiment and theory is quite reasonable for both
the changing temperature and changing humidity cases.
This agreement tends to validate the use of the combined
thermodynamic analysis outlined in 3.4 and particularly
the assumption that products are effectively frozen at their
equilibrium composition at about 1700 K (or equivalently
that K in the 'water-gas' reaction is approximately 3.5).
The same analysis can then be applied to the question of
how to select the target CO concentration to which an
engine should be tuned, under arbitrary ambient condi-
tions, to achieve a desired design air-fuel ratio. The raw
%CO predicted by the combined thermodynamic analysis,
for combustion of iso-octane in moist air in carbureted en-
gines, is plotted in Fig. 26 as a function of T (with w and
P held constant) and indicates an almost constant shift in
target %CO with ambient temperature rise above a stan-
dard value To (298 K), regardless of the design air-fuel
ratio chosen. This result suggests that a simple correction
of the form:
%CO(T) = %CO{T0) + 14.6 (10)
would allow tuning of carbureted engines at wide-open
throttle to a specified design air-fuel ratio, at an arbitrary
ambient temperature. The same analysis showed almost
negligible effects of changing humidity, and that ambi-
ent pressure changes produced equal and opposite effects
to those of temperature, implying the combined ambient-
condition correction:
%CO(T,P) = %CO{T0, P0) + 14.6 ~
° (11)
using absolute values for pressure and temperature. While
this correction may be useful for fuel-rich combustion in
carbureted four-stroke engines, it may not be as accurate
for some two-stroke engines for the reasons outlined in sec-
tion 4.
6. Concluding Remarks
For operation of hand-held engines at wide-open throt-
tle, many trends in performance and emissions with vary-
ing ambient humidity and temperature can be estimated
quite well using theoretical approaches. ¥Vom a regula-
tory perspective, hand-held two-stroke engines are particu-
larly hard to characterize, since all performance/emissions
measured seemed to vary erratically with changine am-
bient temperature, and their HC emissions increase very
strongly with rising humidity. In general, the theoretical
approaches developed in this paper work well in describ-
ing trends with changing humidity because water vapor is
mixed perfectly in ambient air and has minimal effect on
other factors which influence formation of emissions (i.e.
fuel evaporation, transport and engine heat transfer. How-
ever, in the case of the hand-held two-stroke engine, the
apparently complex dependence of the charge homogene-
ity on slight changes in ambient temperature is inconsis-
tent with assumptions of the theory and it is not clear if
theoretical approaches are useful. However, it is possible
to draw some conclusions which may of general value in
describing effects of varying ambient conditions on small
carbureted engines.
To correct for effects of ambient humidity on the forma-
tion of NOx, the effect of operating air-fuel ratio can be
incorporated in a simple multiplicative correction factor:
ir — 1 (Q^
" 1 - ^ (w - 0.01071)
This correction factor produces a good collapse of data for
hand-held engines over a wide range of air-fuel ratios, while
reducing to the original Federal correction at AFR = 16.5.
While this correction seems to apply to ambient humidity,
it may not be as valid for local humidification processes
such as water-injection into intake ports, when the homo-
geneity of water vapor in the air might be worse.
In tuning of engines to operate at a prescribed design
air-fuel ratio, it appears possible to use the exhaust-gas
CO percentage and the prevailing ambient temperature
and pressure to adjust a carbureted engine to operate at
the desired air-fuel ratio at standard conditions. The cor-
rection:
%CO(T,P) = %CO{T0,Po) + 14.6 ~
0 (11)
then allows the target %CO to be selected for tuning on
any given day. This simple guide may prove helpful in
recalibration of in-service engines with adjustable-jet car-
buretors or for checks on the calibration of engines with
fixed-jet carburetors.
-------
The question of what range of temperatures, pressures
and humidities should be permitted for an acceptable emis-
sions test would seem to depend on the kind of engine and
emissions species of most concern. For two-stroke engines
similar to the one studied, a narrow range of humidities
would be necessitated by the high sensitivity of HC to
w. For four-stroke engines, it may be plausible to base a
reasonable range of ambient conditions on an acceptably
small variation in HC emissions, and use corrections for
NOx and even CO if necessary.
Acknowledgements
The authors gratefully acknowledge the support of the
National Vehicle and Fuel Emissions Laboratory of the
U.S. Environmental Protection Agency, administered by
T. Trimble and W. Charmley.
References
1. Code of Federal Regulations, 40 CFR Parts 86 to 99,
Office of the Federal Register National Archives and
Records Administration, Washington, DC, 1992.
2. Brown, W. J., Gendernalik, S. A., Kerley, R. V., fe
Marsee, F. J., "Effect of engine intake-air moisture on
exhaust emissions," SAE Paper No. 700107.
3. Robison, J. A., "Humidity effect on engine nitric ox-
ide emissions at steady state conditions," SAE Paper
No. 700467.
4. Blumberg, P. M. k. Kummer, J. T. "Prediction of NO
formation in spark-ignited engines — an analysis of
methods of control," Combust. Sci. Tech., 4, 73, 1971.
5. Donohoe, J. A., Hardwick, G. C., Newhall, H. K., San-
vordenker, K. S. k. Woelffer, N. C., "Small engine ex-
haust emissions in the United States," SAE Paper
No. 720198.
6. Hare, C. T. &c White, J. J., "Towards the environmental
-ly-friendly small engine: fuel, lubricant, and emission
mea-
surement issues," SAE Paper No. 911222.
7. Morgan, E. J. ic Lincoln, R. H., "Duty cycle for recre-
ational marine engines," SAE Paper No. 901596.
8. Coates, S. W. k. Lassanske, G. G., "Measurement and
analysis of gaseous exhaust emissions from recreational
and small commercial marine craft," SAE Paper
No. 901597.
9. Laimbock, F. J. k Landerl, C. J., "50cc two-stroke en-
gines for mopeds, chainsaws and motorcycles with cat-
alysts," SAE Paper No. 901598.
10. Brereton, G. J., Morrison, K., Chishty, H. A., Schwartz,
G. & Patterson, D. J., "Measured emissions of small
engines under steady state and transient operation,"
SAE Paper No. 941806.
11. White, J. J., Carroll, J. N., Hare, C. T. k. Lourenco,
J. C., "Emission control strategies for small utility en-
gines," SAE Paper No. 911807.
12. Sun, X., Brereton, G., Morrison, K. k Patterson, D.,
"Emissions analysis of small utility engines," SAE Pa-
per No. 952080.
13. Mooney, J. J., Shinn Hwang, H., Daby, K. 0. k Win-
berg, J. R., "Exhaust emission control of small 4-3troke
air cooled utility engines — an initial R fc D report,"
SAE Paper No. 941807.
14. Brereton, G. J., DeAraujo, A. k Bertrand, E., "Effects
of ambient conditions on the emissions of a small car-
bureted four-stroke engine," SAE Paper No. 961739.
15. R. S. Spindt, "Air-fuel ratios from exhaust gas analy-
sis," SAE Paper No. 650507.
16. J. B. Heywood, "Internal combustion engine fundamen-
tals," McGraw-Hill, 1988.
17. Reynolds, W. C., "STAN J AN - chemical equilibrium
analysis by the method of element potentials," Dept.
of Mechanical Engineering, Stanford University, 1987.
18. Gordon, S. & McBride, B. J., "Computer program for
calculation of complex chemical equilibrium composi-
tions, rocket performance, incident and reflected shocks
and Chapman-Jouguet detonations," NASA SP-273,
1971.
19. Woschni, G., "A universally applicable equation for the
instantaneous heat transfer coefficient in the I.C. en-
gine," SAE Paper No. 670931.
20. Bertrand, E. B., "Effects of ambient conditions on the
emissions of a small carbureted overhead valve four
stroke utility engine," M.S. Thesis, Dept. of Mechan-
ical Engineering, Michigan State University, July 1997.
21. Automotive Engine Test Code, General Motors Corpo-
ration, Seventh Edition (Metric), 1994.
16
15
g 14
<
oo
¦= 13
C3
U
Q.
O 12
I!
10
5
t £
3
;asi
1g
o
o
it
3
f
ii
o
o
i/i
10
11
12
13
14
15
16
Design AFR
Fig. 2. Theoretical variation of carbureted AFR with hu-
midity, for different design AFRs.
8
-------
Fig. 3. Theoretical variation of carbureted AFR with tem-
perature, for different design AFRe.
594
p
592
u
2
590
S3
£¦
588
c
u
586
H
t>
aa
584
£
-c
582
•Si
580
tS.
578
3
576
574
oj int
reasing
A
r
r m =
0.011
10 11 12 13 14
Design AFR
15
16
Fig. 4. Theoretical variation of compressed-charge tem-
perature with design air-fuel ratio, at different ambient
humidities.
2950
2900
y.
1? 2850
=j
2
| 2800
£
g 2750
3
U.
2700
2650
4.
4—
!
icreasing
^ Ti =
7 ~C
Ti = 47 '(
10 11 12 13 14
Op«ratiag AFR
15
16
Fig. 6. Theoretical variation of flame temperature with
operating air-fuel ratio, at different ambient temperatures.
2950
2900
V
2850
3
n
%>
c.
2800
x>
fr
2750
Li.
2700
2650
A
1 1
ri increasing
W
Tt increasing
i 1 -
i * *
! i '
Ti =7*C
! ! 1
! | :
Ti = 47'(
i 1
10
15
L6
II 12 13 14
Design AFR
Fig. 7. Theoretical variation of flame temperature with
design air-fuel ratio, at different ambient temperatures.
Design AFR
Fig. 5. Theoretical variation of compressed-charge tem-
perature with design air-fuel ratio, at different ambient
temperatures.
¦x.
2920
2900
2880
2860
| 2840
g 2820
| 2800
0 2780
1 2760
2740
2720
2700
(j a) increasing
i
/Iff
r » i
:T ;
!
//,
/
i
I
///
j
///
'
///
///
i
/ CD — (J.C
15
1 i
10
12
13
14
16
Design AFR
Fig. 8. Theoretical variation of flame temperature with
design air-fuel ratio, at different ambient humidities.
-------
3
2.5
.2
2
CO
Od
1.5
o
o
o
1
0.5
\ CD
>0.015
0) o 0.00
en ncre
»«ig
10
11
12 13
Design AFR
14
15
Fig. 9. Theoretical variation of CO/CO? concentration
with design air-fuel ratio, at different ambient humidities.
0.90
0.85
0.80
0.75
0.70
0.65
0.60
0.55
0.50
0.45
0.40
0.
003 0.005 0.007 0.009 0.011 0.013 0.015
CO
Fig. 12. Experimental variation of torque with ambient
humidity. In this and successive graphs, filled symbols de-
note four-stroke engine data and unfilledd symbols signify
two-stroke engine results.
o
u
10
14
15
11 12 13
Design AFR
Fig. 10. Theoretical variation of CO/COa concentration
with design air-fuel ratio, at different ambient tempera-
tures.
0.015
0.013
0.011
3 0.009
0.007
0.005
0.003
10
15 20
TCC)
•25
30
Fig. 11. Ambient conditions achieved in typical constant
temperature and constant humidity tests.
0.90
0.85
0.80
0.75
2
0.70
u
3
0.65
OfiO
H
0.55
0.50
0.45
0.40
10
15 20
TCC)
25
30
Fig. 13. Experimental variation of torque with ambient
temperature.
14
13
12
11
10
9
i
t
V .
•
:.
i
j
.1 .
1
j
0.003 0.005 0.007 0.009 0.011 0.013 0.015
Fig. 14. Experimental variation of operating AFR (deter-
mined from Spindt's method) with ambient humidity.
-------
8.
o
10
15 20
T(-C)
25
30
Fig. 15. Experimental variation of operating AFR (deter-
mined from Spindt'3 method) with ambient temperature.
1400
1200
1000
£ 800
^t
s 600
o
O
400
200
0.003 0.005 0.007 0.009 O.OU 0.013 0.015
(0
Fig. 16. Experimental variation of brake-specific CO with
ambient humidity.
1400
1200
1000
800
ob
600
G
U
400
200
0
!• -J /
•*
•
j
i
i
10
15 20
T CC)
25
30
Fig. 17. Experimental variation of brake-3pecific CO with
ambient temperature.
4.0
3.5
3.0
-C
=f
2.5
00
2.0
X
O
l.t
1.0
0.5
0.0
e ° " «
0.003 0.005 0.007 0.009 0.011 0.013 0.015
Fig. 18. Experimental dependence of brake-specific NOx
on ambient humidity.
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
10
15 20
TCC)
25
30
Fig. 19. Experimental dependence of brake-specific NOx
on ambient temperature.
700
600
500
400
300
o
200
100
0
0.003 0.005 0.007 0.009 0.011 0.013 0.015
CO
Fig. 20. Experimental variation of brake-specific HC with
ambient humidity.
-------
700
600
500
JZ
%
400
£
00
300
0
T
200
100
0
10
15 20
TCC)
25
30
Fig. 21. Experimental variation of brake-specific HC with
ambient temperature.
1.8
1.6
1.4
| 1.2 i
32 - ^ ! * CO/C02 (Measured 2-Strokei
O ! i * C0/C02 (Theoretical 2-Stroke. AFRd= 11.3)
0.8 i j ° CO/C02 (Measured 4-Stroke)
I * CO/CQ2 (Theoretical 4-Stroke. AFRd=l4.25)
0.6
0.4 J
0.2 T
0.003 0.005 0.007 0.009 0.011 0.013 0.015
Fig. 22. Experimental variation of CO/CO2 with ambient
humidity.
2.0
1.3
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0.003 0.005
I *
0.007 0.009
CO
0.011
0.013 0.015
Fig. 24. Corrected values of brake-specific NOx at differ-
ent ambient humidities, using the Federal Kh[uj) correc-
tion factor.
2.0
1.8
1.6
f 1.4
J
U
§ '•=
t 1.0
!> 0.8
6
4
Z °
0.-
0.2
0.0
0.003
0.005
0.007
0.009
a)
0.011
0.013
0.015
Fig. 25. Corrected values of brake-specific NOx at differ-
ent ambient humidities, using the proposed Ku (<*>, AFR)
correction factor.
1.9
1.8
1.7
1.6
0 l° |
1 1.5 j-
Ol
8 '¦4JT
o
O I
0.1
10
• CO/CO2 (Measured 2-Stroke) -
¦ C0JC02 (Theoretical 2-Siroke. AFRd=l0.85)
° C0/C02 (Measured 4-Stroke) _
• CO/CQ2 (Theoretical 4-Stroke. AFRd=14.5) ^
TFTT
15
20
TfC)
25
30
Fig. 23. Experimental variation of COjCOi with ambient
temperature.
II 12 13
Design AFR
Fig. 26. Theoretical variation of %CO with ambient tem-
perature at different design air-fuel ratios.
-------
SAE TECHNICAL
PAPER SERIES
961739
Effects of Ambient Conditions on the
Emissions of a Small Carbureted
Four-Stroke Engine
G. J. Brereton, A. DeAraujo, and E. Bertrand
Michigan State Univ.
Reprinted from: Design, Modeling, and Emission Control for
Small Two- and Four- Stroke Engines
(SP-1195)
r The Engineering Society International Off-Highway &
For Advancing Mobility Powerplant Congress & Exposition
I N TE Bn'atTonVL* Indianapolis, Indiana
//V I tz ri P4 A ! I u Pi A\ L August 26-28, 1996
400 Commonwealth Drive, Warrendale, PA 15096-0001 U.S.A. Tel: (412)776-4841 Fax:(412)776-5760
-------
961739
Effects of Ambient Conditions on the Emissions
of a Small Carbureted Four-Stroke Engine
G. J. Brereton, A. DeAraujo, and E. Bertrand
Michigan State Univ.
Copyright 1996 Society of Automotive Engineers, Inc.
Abstract
The exhaust-gas emissions of a small four-stroke, carbu-
reted, single-cylinder spark-ignition engine have been stud-
ied as functions of ambient conditions, using gasoline as the
fuel. In steady-state dynamometer tests at fixed engine
speeds/loads, carried out under different climatic condi-
tions, the concentrations of exhaust-gas components have
been measured. Their dependence on ambient conditions
has been analyzed principally in terms of the influence of
ambient temperature, pressure, and humidity on the air-
fuel ratio metered by the carburetor. While the air-fuel
ratio of carbureted utility engines at fixed loads varies by
only a small percentage during modest changes in ambient
air conditions, these changes can correspond to significant
changes in the production of regulated pollutants. Using a
correction for air mass flow and fuel density at wide open
throttle, the scatter in observed air-fuel ratio and % CO
data could be reduced by about one third.
1. Introduction
It is well known that the performance of internal com-
bustion engines can depend on ambient conditions through
their influence on: the air-fuel ratio metered during intake;
the rate of evaporation of fuel droplets; other aspects of the
charge composition of that affect combustion; and ambi-
ent effects on engine heat transfer. In the case of automo-
bile engines, procedures have been proposed which would
correct the observed engine performance to that expected
under standard atmospheric conditions (20°C, 101.3 kPa,
10.71 grams H^O/kg dry air). One example is the correc-
tion factor for observed brake power at wide-open throttle,
C*brake
CjFbrake ( pQ )
1.2
To
^o.ref >
0.6
(1)
which, when multiplied by the observed brake power, gives
an estimate of the expected brake power under standard
conditions. It involves assumptions of steady, frictionless
intake-air-flow without heat transfer, a direct proportion-
ality between the power developed in the cylinder and the
mass-flow rate of dry air, and the assumption that mechan-
ical friction always accounts for losses of a fixed 15% of the
power delivered at the piston face. Although the correc-
tion is based on well-established principles, almost all the
assumptions are violated to some degree in real IC engine
operation. However, repeated application of the correction
has tended to indicate that, for conventional designs of and
operation of four-stroke automotive engines, the correction
is quite accurate for applications at wide-open throttle over
a wide range of ambient temperatures and pressures.1
A less satisfactory model for ambient effects is the Kh
correction
Kh =
1 -0.0329 (H - 10.71)
(2)
where H is the specific humidity in grams H^O/kg dry air,
which would rescale an observed NO/NOx concentration
at ambient specific humidity H to that expected when the
ambient was at standard conditions. The origin of this
correction is experimental, based on the performance of a
fleet of test automobiles on a Federal Test Cycle.2 While
it is of relevance to the average NO/NOx emissions of
water-cooled, multi-cylinder automobile engines on typical
driving cycles, with basic carburetor settings in the 15.5:1
to 16.0:1 range, its validity for other classes of engines in
different applications has not been established.
For reasons of this kind, it is of great interest to study
the effects of ambient conditions on performance and emis-
sions of air-cooled, carbureted off-road and utility engines,
which are now the subject of increasingly stringent federal
emissions regulations. In contrast to the multi-cylinder,
water-cooled automobile engine, for which these effects
have been documented extensively, little has been reported
about ambient effects on emissions of single-cylinder, air-
cooled utility engines. A preliminary study of effects of
ambient temperature, pressure, and humidity was there-
fore carried out on a typical utility engine of this class. In
this paper, we describe the tests and ranges of conditions
explored, analyse experimental results, interpret them in
77
-------
terms of likely physical processes involved, and discuss ap-
propriate forms of corrections which might apply to this
class of engine.
Before describing experiments and their results in de-
tail, it is useful to present a hypothesis for the most likely
effects of ambient conditions on small-engine emissions:
the charge preparation and the subsequent in-cylinder pro-
cesses are the determinants of emissions which are most
likely to be affected by changes in ambient conditions and
so are processes which should be studied closely. For this
reason, it is extremely useful to be able to measure the
air-fuel ratio metered by the carburetor at each set of am-
bient conditions, and to know the properties of the charge
prior to compression and combustion. Since the primary
objective of this study is to assess effects of ambient con-
ditions on emissions, changes in concentrations of emis-
sions are provisionally viewed as consequences of changes
in the air-fuel ratio and changes in the conditions under
which combustion takes place. Thus engine performance
measurements which indicate air-fuel ratio (i.e. ratios of
CO and CO2 concentrations), and air intake and cylinder
temperatures are equal in importance to measurements of
ambient pressure, temperature and moisture content. Ef-
fects of ambient conditions on production of NO/NOx,
which arises through non-equilibrium chemical reactions
and plainly depends on in-cylinder processes, is deferred
for a later study.
2. Experimental Overview
The experiments were carried out as part of a series of
emissions measurements made according to the SAE J1088
test procedure,3 on a Kohler CV 12.5 S 4-stroke single-
cylinder overhead-valve engine (with bore: 87 mm; stroke:
67 mm; compression ratio: 8.5:1; displacement: 398 cm3;
rated power: 9.33 kW at 3600 rpm). The gasoline fueled
version of this engine employed a fixed main-jet carbure-
tor with air-bleed compensation. Using Indolene as the
fuel, this engine was tested at over 8 different laboratories
throughout the US within a 12-month period beginning
in April 1993. The engine was operated for a minimum
of three J1088 cycles at each laboratory and was broken
in for a period of 100 hrs by its manufacturer, prior to
these tests. The subsequent tests amounted to approxi-
mately 70 hours of further engine use. In the course of
these tests, the emissions of a newly broken-in utility en-
gine, of a size/class which might be expected to provide
repeatable performance for a large number of hours of op-
eration, were measured under a range of different climatic
conditions.
Emissions measurements were all made on a raw ba-
sis, using the same sampling probe built into the exhaust
manifold. To provide a check on the relative accuracy of
emissions measurement equipment used in the different
laboratories, a container of an unidentified gas mixture,
with concentrations of engine exhaust gases typicaJ of those
found in utility engines, was also analyzed in each labora-
tory. In nearly all cases, the measured concentrations of
this gas mixture were almost identical, from which it could
be inferred that the use of different emissions equipment
at different laboratories was not a major source of uncer-
tainty when data sets were compared. During these tests,
the ambient temperature, pressure and specific humidity
spanned the ranges: 21.5 —~ 31°C; 97.97 —~ 99.56 kPa;
and 4.8 —* 16.7 grams H^Ofkg dry air. For purposes of
comparison, emissions tests for diesel engine certification
(for which restrictions on acceptable ambient conditions
have been developed quite extensively) must be carried at
ambient temperatures between 20 and 30°C and at pres-
sures between 96.6 and 105 kPa, with additional restric-
tions on the fuel temperature at the fuel-pump inlet (40
CFR 86.341-79).4
The SAE J1088 test procedure requires measurements
of concentrations of the products of hydrocarbon combus-
tion: CO, NO/NOx, HC, and usually also CO2 to facil-
itate estimation of the air-fuel ratio. These measurements
are made at idle, at wide-open throttle, and at 75%, 50%,
25% and 10% of the wide-open-throttle torque, at 85%
of the rated speed (t.e. 3060 rpm for a 3600 rpm rated
engine). In addition to these exhaust gas concentrations,
temperatures of the spark-plug seat, intake air, crankcase
oil and exhaust gas in the muffler were recorded in all tests.
This information was used to assess the validity of theo-
retical models for the dependence of engine emissions on
ambient conditions, and thus develop a clearer understand-
ing of the true influence of ambient conditions on typical
air-cooled, carbureted, gasoline-fueled utility engines.
3. Experimental Results and Data Reduction
Experimental results for engine performance and the
ambient conditions at each of the loads: 100%; 75%; 50%;
25% and 10% are shown in Tables 1-5. Data recorded at
the idle condition were not included in this study because
of the difficulties of reproducing repeatable idling condi-
tions on different dynamometers in different laboratories.
Tables 1-5 also include the computed specific humidity w
and the air-fuel ratio, as estimated by Spindt's method.5
Of these data, the most straightforward to analyze are
those measured at 100% load (wide-open throttle). At this
condition, effects of friction within the intake system are
lowest and the assumption of frictionless flow, which leads
to several convenient fluid-mechanic and thermodynamic
relationships, is most likely to be valid. Also, in typical dy-
namometer tests, the wide-open-throttle mode is the most
repeatable and does not depend on the relative accuracy of
torque sensors or the ability to set a throttle-plate at a pre-
cise position, as does the repeatability of engine operation
at other loads in J1088 tests.
From inspection of the data at wide-open throttle, it is
evident that over the range of ambient conditions tested,
the air-fuel ratio only varies between extremes of approxi-
mately 12.0 and 12.3 — a relative variation of only 3.0%.
The variation is slightly less (2.8%) at 75% and 50% loads,
but then increases with decreasing load, reaching 5% at
25% load and 5.5% at 10% load. These variations between
-------
maximum and minimum air-fuel ratio may be systematic
effects of changes in ambient conditions, or they may repre-
sent increased scatter in experimental data, as it becomes
progressively more difficult to operate the engine at pre-
cisely the load specified.
At wide-open throttle, there are trends in the data of
Table 1 such as the tendency of the charge to become
leaner (air-fuel ratio increasing) as the specific humidity
increases (Fig. 1). There is also a companion increase in
spark-plug seat temperature — most likely a consequence
of combustion at higher temperatures as the air-fuel ra-
tio increases towards its stoichiometric value (Fig. 2) —
though one set of measurements (K-l, K-2, K-3) disputes
this trend. There appears to be a weak correlation be-
tween increasing air-fuel ratio and rising ambient or intake
temperatures, though it is arguable that this might be lit-
tle larger than the uncertainty/scatter in the experimental
data (Fig. 3). The relation between air-intake tempera-
ture and air-fuel ratio is similar to that of Fig. 3 except
for one set of measurements in which the temperature rise
from the surroundings to the intake is almost double that
observed in all others (B-l, B-2, B-3).
At 75% load, there is a comparable variation in the ob-
served air-fuel ratio despite variations of 14°C in intake
air temperature and 8 grams .HjO/kg dry air in the spe-
cific humidity. At 50% load, an equally narrow range of
air-fuel ratio variation is observed as similar changes in
room/intake air temperature and humidity are observed.
Based on these observations, plausible cases might be made
for either carburation being insensitive to changing condi-
tions, or for changes in ambient conditions (i.e. increas-
ing humidity and air temperature) producing effects which
compensate for one another.
The 50%, 25% and 10% load data are similar to those
at wide-open throttle, with high values of specific humid-
ity accompanying the highest ambient/intake air temper-
atures. However, the peak air-intake temperatures are al-
most all lower than at wide-open throttle while the specific
humidities are unchanged, suggesting a possible correla-
tion with the (lower) cylinder temperatures, which might
arise through heat transfer from the cylinder-head to the
intake air. Since, according to the data of Table 2, almost
the same air-fuel ratios are observed at both the highest
combinations of intake temperature and humidity, and the
lowest combinations of these properties, it is plausible that
the two have opposite effects on carburation, and that they
may indeed partially compensate for each other at 100%,
75%, and 50% load.
Apart from the general dependences on specific humid-
ity and ambient/intake temperature noted above, there are
also dependences on other properties such as ambient pres-
sure. In Fig. 4, there is a trend of spark-plug seat tem-
perature rising with increasing ambient pressure. This is
possibly a consequence of the increasing peak combustion
temperatures with the leaner charges which result from the
increased air density at higher ambient pressures. If the
air-fuel ratio is plotted against ambient pressure (Fig. 5),
a systematic rise in air-fuel ratio with increasing ambient
pressure is seen, if one set of measurements (K-l, K-2,
K-3) is disregarded. At a given load, variations in the
spark-plug seat temperature might then be more sensitive
indicators of changes in air-fuel ratio than methods based
on exhauBt-gas analysis. Although other factors may well
influence the metered air-fuel ratio in small carbureted en-
gines, they are believed to be relatively small compared to
humidity, ambient pressure, ambient and cylinder temper-
atures, and the experimental uncertainty in these data.
4. Data Reduction and Model Assessment
Corrections were developed by taking models proposed
for the dependence of air-fuel ratio on ambient conditions
and testing them against measured data, using a statisti-
cal regression technique. Over a wide range of engine op-
erating conditions, the characteristic (AFR vs. APVenturi)
curve of an air-bleed compensated carburetor is quite flat
(Fig. 6), and small changes in air-fuel ratio on account of
ambient changeB at a given load can be modeled as if the
pressure drop from the carburetor venturi to its throat re-
mained constant. Models for the dependence of air-fuel
ratio on ambient conditions are therefore developed only
to account for changes in air-fuel ratio, and not for any re-
sultant changes in venturi-throat pressure drop. They are
expressed as the correction factor for ambient effects which
rescales observed air-fuel ratios to the design air-fuel ratio
at standard ambient conditions, i.e.
AFR . CFafr — AFRq (3)
where the subscript o denotes the design or reference condi-
tion. Since the design air-fuel ratio was not measured, and
its precise value would be subject to uncertainty in both
experimental measurements and Spindt's method, AFRo
is taken as the value which best fits the AFR data cor-
rected by each model. The deviation in corrected AFR
about AFRo is then a measure of the effectiveness of a
proposed model, with the most faithful corrections yield-
ing the smallest deviations about the fitted value of AFRo.
In this study, all fitting was carried out by minimizing the
sum of the absolute deviations between the data and the
modeled dependence.6 This kind of minimization is more
robust than least-squares minimizations, being less sensi-
tive to the influence of single outlying points on fits to
data.
The air-fuel ratio is written as: AFR = mdry air/^fuei
where = mdry air(l+w)- We can therefore write:
AFRo = ((mmoi" °) (-^1-) . AFR
\1+W0/ \ mrooist air / V»T»fuel, 0 /
= CFafr ¦ AFR (4)
where CFAFr is the correction for metered air-fuel ratio
according to changes in ambient conditions. Several mod-
els for moist-air and fuel flows in small-engine intakes and
-------
carburetors are expressed in the form of a correction fac-
tor CFafr al|d tested against experimental data in the
following sections.
5. Carburetor Modeling
5.1 Fuel-flow rate
Under steady-state conditions, the fuel flow rate mfuei
can be modeled as an incompressible flow described by a
Bernoulli equation, in which effects of friction and orifice
shape are compensated for with a discharge coefficient Cp,
leading to a model of the form
""^uel = Cd Ay/2pfueiAPf
(5)
where APf is the pressure drop experienced by the fuel
flowing through the jet orifice of area A. In the absence of
air-bleed compensation, the fuel-flow pressure drop APj is
APj — APair Pfuel 9 A/lfuei
(6)
where APa,r is the pressure drop of air flowing from the
ambient to the carburetor throat and A/ijue] is the ver-
tical elevation the fuel must travel from the fuel surface
in the float chamber to the point where it mixes with air.
Air-bleed compensation acts to reduce the proportional-
ity of mfuei to APair as APait increases. Thus, instead
of over-richening the charge at wide-open throttle, a close
approximation to a constant AFR is achieved over a range
of values of APa;r.
While A is obviously constant in a fixed main-jet car-
buretor, Co varies smoothly with the fuel-flow Reynolds
number, A Pj changes according to the air mass-flow rate,
and PfUei depends primarily on temperature. In typical
applications at a fixed engine load, with power scaling ap-
proximately on mfUel
Pfuel, 0
= \/l - £(7fuel - Tfuel, o)
(8)
5.2 Air-flow rate
When the air flow through the venturi of a carburetor is
modeled as if a quasi-steady process, assuming frictionless,
iso-energetic, compressible, one-dimensional flow, the air
mass-flow rate can be written as:7
fHnoist air —
At Pi
y/RT<\
-«)
(i-i)Ar
Here the subscripts
7 ~ -l V fi J
(9)
and t refer to the venturi intake
and throat respectively, P is the absolute pressure, and R
and 7 are average properties of the moist air. In typical
single-cylinder small engines, with strongly pulsating in-
take flows, R and 7 should really be average properties of
a moist air/fuel mixture, since there is periodic backflow
of charge 'upstream' of the carburetor. Effects of friction
and flow non-uniformity are compensated for by introduc-
ing a steady-flow discharge coefficient Co, which typically
varies smoothly with air-flow Reynolds number. Whereas
Ti and P, might be related closely to ambient pressures and
temperatures in automotive carburetors, positioned some
distance from the engine, in small-engine applications the
ducting of intake air put hot surfaces could lessen a de-
pendence between the ambient and venturi-intake temper-
atures. For the usual case of intake air velocities being
much lower than throat velocities, equation (9) can be re-
expressed as
air = CDATy/2pi{Pi - Pr) • * [l, Pr/Pi) (10)
where $ is a factor which incorporates effects of compress-
ibility. Since -y depends only weakly on w7 and (P, —
Pj-)/P, IZH 0.1 over the normal operating range (mak-
ing $ very close to 1), a first-order model for moist-air
flow at a given engine load is:
air ^ \/( Pi Pr)
(11)
When expressed as a correction for air mass-flow rate, it
becomes
'7*moiBt air, 0 J Pi,0 (-^i Pr)o
rilmoi.t air V Pi ~ Pt)
1 + <^0 Pi, dry air, 0 (P. ~ Pt)o
1 -f- Ci/ p-lt dry ajr (.p Pf )
(12)
80
A number of variations on (12) may then be derived, de-
pending on the assumptions most appropriate to engine op-
eration. For example, in the Automotive Engine Test Code
-------
of General Motors Corporation,1 it is assumed that the ra-
tio of partial pressures of dry air: Pu dry air/^T, dry sir re-
mains constant for operation over a range of ambient con-
ditions at wide-open throttle and that the moist-air gas
constant R is the same as that for dry air. It is also as-
sumed that frictionless, iso-energetic, one-dimensional flow
is a reasonable approximation to air-flow conditions from
the surroundings, and through the intake system as far as
the limiting restriction on the flow. In the case of au-
tomobile engines, experimental data are well correlated
with this variation on (12) provided none of the ambi-
ent conditions vary from standard values by more than
about 5%.1 In the case of single-cylinder air-cooled utility
engines, with close-coupled carburetors which may experi-
ence engine-body heat transfer, and intake ducts of smaller
cross-sectional area, the most appropriate simplifying as-
sumptions have still to be established.
6. Assessment of carburetor models from exper-
imental data
As a baseline test for proposed corrections, we consider
first the coefficient of variance of air-fuel ratio when no cor-
rection is applied. This coefficient of variance is the root-
mean-square or standard deviation from the mean air-fuel
ratio, as a percentage of the mean value. From the data of
Tables 1-5, corresponding to loads of 100%, 75%, 50%,
25% and 10% on the J1088 test procedure, the baseline
coefficients of variance have been computed and are listed
in Table 6. To explore the effectiveness of the possible cor-
rections, we consider first the case of wide-open throttle,
for which the data is believed to be the most trustworthy.
Using different corrections discussed below, coefficients of
variance in AFR at wide-open throttle were computed and
are given in Table 7. Because of concerns cited earlier over
selected measurements, the air-intake temperature data of
B-l, B-2 and B-3 and the air-fuel ratio/cylinder tempera-
ture data of K-l, K-2 and K-3 were not used for the pur-
poses of assessing correction models.
6.1 Dry-air mass-flow-rate correction
If the automotive-engine dry-air mass-flow rate model
(10) is developed,1 so that the dry-air mass-flow rate is
proportional to the intake dry-air partial pressure and in-
versely proportional to the square root of intake air tem-
perature, the correction factor is
r, ¦Pi, dry air. 0 / ^i. dry air ,-,1
^*AFR = —£ \ ¦=, (13)
•* i, dry air V -m, dry air, 0
where o denotes the reference' state. It effectively cor-
rects for the dry-air flow rate assuming air-flow from the
surroundings through the intake system to the carburetor
throat are frictionless, iso-energetic and one-dimensional,
and that there is no ambient effect on the fuel flow rate.
Taking intake conditions as the room conditions, with the
partial pressure of room air as the total pressure minus the
partial pressure of water vapor, the coefficient of variance
in corrected air-fuel ratio may be computed from the data
of Table. 1. Using only this correction, the uncorrected
baseline coefficient of variance of 0.98% is nearly doubled
to 1.95%. Clearly, it is better to apply no correction than
to use this one alone, since it either over-corrects or cor-
rects in the wrong direction. Possible explanations for its
poor performance include the absence of a fuel-flow cor-
rection, or the possibility that effects of heat transfer and
friction within small-engine intake systems are more sig-
nificant than in typical automobile engines and violate the
assumptions on which the correction is based.
6.2 Fuel-flow-rate correction
While fuel-flow rates are. controlled using mass-airflow
sensors, rpm and air-density correlations, and feedback
from exhaust-gas oxygen sensors in automobile engines,
it is the volume flow rate of fuel which is metered in car-
bureted applications. To account for fuel-density changes
with temperature, the first-order temperature-dependent
fuel density correction developed above was:
.mfUCl = y^l — /9(7fuel — Tfuel, o) (14)
^uel. 0
If it is assumed that the gasoline is in the purely liquid
phase, with ~ 0.0018^ 0, and is taken as Troomi
then the correction factor is
CFafr = \J 1 - A)(7fUel, room - rfuei. o) (15)
When tested against the data of Table 1, for operation
at wide-open throttle, this correction reduces the coeffi-
cient of variance in AFR to 0.80% from the uncorrected
baseline value of 0.98% — an improvement of 19%.
If it is assumed that the fuel density at the carburetor
main jet is not at room temperature but at the air-intake
temperature, owing to heat transfer within the body of the
carburetor, then
OFAfR = \Jl - Po{Ttue\, intake - 7fUel. o) (16)
and the coefficient of variance in AFR is reduced below
its uncorrected value of 0.98% to 0.65%, removing 35% of
the scatter in observed AFR. If the modulus of volumet-
ric expansion f) is treated as a free parameter in (15), the
coefficient of variance in AFR is reduced further to 0.64%
when [i takes the value: 0.0036/°C. Since the possibility
that fuel vapor bubbles were suspended in the liquid can-
not be discounted, particularly since Ti„take is v«ry close to
the initial boiling point of Indolene (31.5°C), it is not un-
reasonable for the true value of to exceed the pure-liquid
value of 0.0018/°C. Alternatively, if (Tfuel, intake - Tfuei. o)
is an underestimate of the temperature rise experienced by
the fuel (owing to, for example, heat transfer to the car-
buretor body from the engine cylinder), a larger effective
value of P might be appropriate.
These results imply that there may be a strong correla-
tion between the fuel density (and its temperature within
the carburetor) and the metered AFR, which can be cor-
rected simply. It is noteworthy that this effect has not
-------
been included in corrections proposed for automobile en-
gines, which form the basis for most previous performance
and emissions corrections for ambient effects. It may well
be because the more sophisticated design of automobile
carburetors allowed some compensation for varying intake-
air temperatures, whereas today's automobile engines use
fuel injection systems with feedback from an exhaust-gas
oxygen sensor, together with manifold mass-air-flow mea-
surements, to determine the appropriate mixture strength.
6.3 Combined air/fuel flow rate corrections
Having established a correlation between fuel density
and the metered AFR as fuel-flow-rate correction, it is
instructive to seek broader corrections which also account
for the air-flow rate. If (13) is combined with (16) to yield a
correction factor appropriate for frictionless, iso-energetic,
one-dimensional air flow, then
CFafr =
¦Pi, dry air, 0 / 2i, dry air /I a(rr ^ *
p \l rp y intake ¦'fuel, Oj
M, dry air V -M, dry air. 0
(17)
This correction reduces the coefficient of variance in AFR
from its uncorrected value of 0.98% to 0.69%, accounting
for 30% of the scatter in AFR data. Although it might
appear that the advantages of the fuel-flow correction are
slightly diminished by combination with a dry-air flow cor-
rection, the 0.69% coefficient of variance also includes ad-
ditional scatter introduced by the greater propagation of
uncertainties when two multiplicative corrections are used.
Thus it appears that the paper's original hypothesis of a
combined air- and fuel-flow correction is a promising one.
There are a number of alternatives to the dry-air flow
correction which involve slightly different intake-flow as-
sumptions. In particular, the moist-air correction of (12)
may be simplified in a number of different ways. If the
intake-to-throat pressure drop ratio (P, — Pr)/Pi is as-
sumed constant, at steady-speed wide-open throttle oper-
ation, then (12) can be written as
mmoi»t air, 0 /1 ^ ^0 ^i. dry air Fj, dry air.O Pi, 0
™moist air V 1 + W T<'• dry air 0 P*' dry air p
(18)
Combining (18) with (3) yields a correction which results in
a coefficient of variance of 0.81%, which accounts for 18%
of the data's scatter. If the dependence of dry-air density
on pressure (which is slight over the range considerd) is
removed from (18), in the interests of simplification, the
correction becomes
CFafr =
/ 1 + W Tl, dry air P>, 0 f. \
\ T~, D~ V 1 ~~ P^tuel, intake ~ Jfuel. o)
y 1 + W0 1 i, dry air, o Pi v
(19)
This correction results in a coefficient of variance of 0.68%,
which accounts for over 30% of the AFR scatter. It is
about as successful as the fuel-flow corrections described
earlier, even though it is more vulnerable to increased un-
certainty owing to the use of more variables in the cor-
rection. A graph of the uncorrected and corrected AFR
data using (19) is given in Fig. 7. It appears that all mea-
surements but one are corrected towards a constant design
air-fuel ratio which takes a value of approximately 12.10.
6.4 Part-load corrections
The question of how to correct for changing ambient
conditions when a carbureted engine operates at part-load
is a much more difficult one. Typically, to regulate op-
eration at a prescribed load, the throttle plate would be
adjusted to meter an almost constant mass-flow rate of dry
air (power being, to a first approximation, proportional to
"idry air, o) while ambient conditions changed. While (10)
can be re-expressed in terms of mdry air, o by dividing by
1+w, and, to correct for slight differences in observed load,
nidry air, o/t°rque can be set to a constant value, no clear
correlation could be found to reduce variance in AFR or
to confirm the model of constant dry-air mass-flow rate.
When applied to load conditions for each of the data in
Tables 1-5, none of the corrections discussed reduced
the total variance in AFR. Each would reduce variance
at some loads but increase it at others. The correction
of (19) increased the scatter in AFR over its uncorrected
value by 5%; others increased it by as much as 50%. It
is thought that a general correction for ambient effects on
AFR at all loads could not be found because: t) the ex-
perimental imprecision in trying to run engines at specified
torques may lead to greater changes in AFR than ambient
effects; and «) different model assumptions may apply at
part-load, when the cylinder temperature is much lower
and intake flow is controlled by the flow resistance past a
throttle plate.
7. Concluding remarks
The air-fuel ratio of a 9.33 kW air-cooled, carbureted
utility engine has been studied over a range of ambient
temperatures, pressures and humidities. It was found that,
at any given load, variations in the air-fuel ratio were small
(of the order of few percent) despite changes in ambient
temperatures of up to 10° C and in humidity of 12 grams
H-zOjkg dry air. Nonetheless, for typical gasoline combus-
tion in today's utiility engine, a variation in AFR between
11.7 and 12.3 corresponds to changes in exhaust concentra-
tions (dry) of CO from 5.2% to 6.8%, which is a relatively
large change in a regulated pollutant. The significance of
such variations in CO concentration begs the question of
what proportion of this change in CO concentration can
be correlated with changes in ambient conditions.
Attempts to correlate variation in AFR were only suc-
cessful when the engine was operated at wide-open throt-
tle. In this case, variation in AFR could be reduced by
82
-------
approximately one third when a correction based on the
temperature-dependence of fuel density was applied. Such
a correction might reduce variation in CO concentration to
between 5.5% and 6.5%. While more sophisticated correc-
tions could be proposed, the multiplicative effect of ad-
ditional terms in corrections simultaneously propagated
more experimental uncertainties, which in turn increased
uncertainty in the correction. For this reason, weighted
averages or summations of single-variable corrections may
be more appropriate when applied over a range of loads.
At part-load, no corrections based on carburetor mod-
els were found which reduced the variance in AFR at
all loads. It is not clear if this finding reflects the inap-
plicability of the proposed models, a need for more pre-
cisely controlled experiments, or the inherent variability
in operation of a typical utility engine. If variations in
part-load AFR are simply particularly random in typi-
cal engine tests, and not subject to systematic effects of
changing external conditions, these finding may suggest a
need for tests such as the SAE J1088, to either include
multiple measurements at each mode (i.e. more than the
three of this study) to further reduce variability in the final
weighted-average result, or to attempt to reduce variabil-
ity in emissions/performance data in other ways. Issues of
this kind are likely to grow in importance in the future, as
this important and popular class of engine evolves and its
characteristics become better understood.
Acknowledgments
The authors gratefully acknowledge the support of the
National Vehicle and Fuel Emissions Laboratory of the
U.S. Environmental Protection Agency, and particularly
the assistance of T. Trimble and W. Charmley. The labo-
ratories and corporations who participated in the tests of
this engine included: Walbro; Onan; Tecumseh; Kohler;
Briggs & Stratton; U.S. E.P.A.; and University of Michi-
gan.
References
1. Automotive Engine Test Code, General Motors Corpo-
ration, Seventh Edition (Metric), 1994.
2. Brown, W. J. et al. "Effect of engine intake-air moisture
on exhaust emissions," SAE Paper No. 700107, 1970.
3. "Test procedure for the measurement of gaseous ex-
haust emissions from small utility engines," SAE J1088
Recommended Practice, 1983.
4. Code of Federal Regulations, 40 CFR Parts 86 to 99,
Office of the Federal Register National Archives and
Records Administration, Washington, DC, 1992.
5. R. S. Spindt, "Air-fuel ratios from exhaust gas analy-
sis," SAE Paper No. 650507, 1965.
6. W. H. Press, B. P. Flannery, S. A. Flannery
& W. T. Vetterling, "The art of scientific computing,"
Cambridge University Press, pp. 539 - 546, 1986.
7. J. B. Heywood, "Internal combustion engine fundamen-
tals," McGraw-Hill, pp. 284 - 290, 1988.
8. J. B. Maxwell, "Data book on hydrocarbons," Van Nos-
trand Inc., p. 143, 1950.
Test
T (room)
T(intake)
T(cyl)
P (room)
P (vapor)
CO
C02
02
0)
Spindt
Deg C
Deg C
Deg C
mm Hg
mm Hg
%
%
%
AFR
W-1
25.5
30.2
215.0
739.14
11.61
6.81
10.36
0.38
0.00992
12.086
W-2
28.6
33.1
218.0
739.14
11.81
6.58
10.47
0.36
0.01010
12.156
W-3
24.5
29.2
213.9
741.68
12.97
6.41
10.54
0.41
0.01107
12.244
0-1
25.0
30.6
212.8
735.84
7.00
6.91
10.55
0.27
0.00597
12.026
0-2
25.6
30.6
212.8
735.84
7.00
6.91
10.56
0.26
0.00597
12.022
0-3
25.6
31.1
212.2
735.84
5.65
7.04
10.48
0.27
0.00482
11.979
T-1
25.0
28.9
212.2
734.31
10.22
7.06
10.57
0.40
0.00878
12.062
T-2
26.1
32.2
213.9
734.31
10.46
7.12
10.42
0.40
0.00899
12.022
T-3
25.0
30.0
212.2
734.31
12.64
7.06
10.40
0.30
0.01089
11.978
K-1
26.7
31.1
220.6
746.25
9.16
7.19
10.78
0.30
0.00773
12.005
K-2
25.6
29.4
221.7
746.00
9.17
7.08
10.82
0.32
0.00774
12.051
K-3
27.2
31.1
221.1
743.71
11.79
7.28
10.71
0.28
0.01002
11.959
B-1
27.2
37.2
215.0
739.90
17.15
6.17
11.01
0.15
0.01476
12.232
8-2
28.9
40.6
217.8
739.39
18.31
5.99
11.09
0.15
0.01579
12.294
B-3
31.1
42.2
219.4
738.38 I 19.35
5.90
11.10
0.16
0.01674
12.327
83
Table 1.
Observed exhaust gas concentrations, inferred air-fuel ratios, and ambient conditions; engine operation
at 85% of rated speed, wide-open throttle.
-------
Table 2.
Observed exhaust gas concentrations, inferred air-fuel ratios, and ambient conditions; engine operation
at 85% of rated speed, 75% of load at wide-open throttle.
Test
T (room)
T(intake)
T(cvD
P (room)
P (vapor)
CO I C02
02
<0
Spindt
Deg C
Deg C
Deg C
mm Hg
mm Hg
% | %
%
AFR
W-1
25.7
29.7
202.5
739.14
11.61
5.01
11.48
0.38
0.00992
12.774
W-2
27.7
31.7
204.3
739.14
11.81
5.03
11.45
0.37
0.01010
12.759
W-3
25.0
29.2
201.2
741.68
12.97
4.71
11.60
0.41
0.01107
12.902
0-1
23.9
28.9
202.8
735.84
7.00
5.19
11.65
0.23
0.00597
12.650
0-2
23.9
28.9
203.3
735.84
7.00
5.22
11.63
0.24
0.00597
12.644
0-3
27.8
32.2
205.0
735.84
5.65
5.19
11.66
0.22
0.00482
12.645
T-1
25.6
30.6
204.4
734.31
10.22
5.26
11.59
0.40
0.00878
12.722
T-2
23.9
27.2
200.6
734.31
10.46
5.13
11.48
0.40
0.00899
12.749
T-3
25.6
28.9
202.8
734.31
12.64
5.21
11.42
0.30
0.01089
12.657
K-1
25.6
30.0
207.2
746.25
9.16
5.17
12.02
0.28
0.00773
12.729
K-2
26.7
30.0
207.8
746.00
9.17
5.21
11.98
0.28
0.00774
12.713
K-3
26.1
30.6
206.1
743.71
11.79
5.23
11.95
0.29
0.01002
12.709
B-1
27.2
35.0
200.6
739.90
17.15
5.04
11.69
0.16
0.01476
12.659
B-2
29.4
39.4
203.3
739.39
18.31
5.33
11.51
0.14
0.01579
12.537
B-3
31.1
40.6
204.4
738.38
19.35
5.24
11.53
0.14
0.01674
12.567
Table 3.
Observed exhaust gas concentrations, inferred air-fuel ratios, and ambient conditions; engine operation
at 85% of rated speed, 50% of load at wide-open throttle.
Test
T (room)
T(intake)
T(cyl)
P (room)
P (vapor)
CO
' CO2
02
a
Spindt
Deg C
Deg C
Deg C
mm Hg
mm Hg
%
%
%
AFR
W-1
25.7
29.4
179.3
739.14
11.61
5.25
11.35
0.34
0.00992
12.660
W-2
27.9
31.6
182.0
739.14
11.81
4.98
11.51
0.33
0.0101 Oj
12.757
W-3
25.0
28.6
178.8
741.68
12.97
4.72
11.68
0.36
0.01107
12.877
O-l
22.8
27.2
181.7
735.64
7.00
4.67
11.99
0.22
0.00597
12.843
0-2
22.8
27.2
182.8
735.84
7.00
4.74
11.96
0.22
0.00597
12.818
0-3
29.4
33.9
18B.3
735.84
5.65
5.24
11.64
0.22
0.00482
12.628
T-1
23.9
26.1
181.1
734.31
10.22
4.92
11.83
0.30
0.00878
12.795
T-2
23.9
26.7
179.4
734.31
10.46
5.30
11.42
0.30
0.00899
12.630
T-3
26.7
30.6
182.8
734.31
12.64
5.32
11.42
0.30
0.01089
12.624
K-1
26.7
28.9
185.6
746.25
9.16
5.13
12.09
0.21
0.00773
12.709
K-2
26.1
29.4
185.0
746.00 j
9.17
5.12
12.07
0.22
0.00774
12.715
K-3
27.2
30.0
185.0
743.71
11.79
5.26
12.01
0.23
0.01002
12.672
B-1
27.2
28.3
176.7
739.90
17.15
5.57
11.42
0.13
0.01476
12.449
B-2
29.4
37.8
180.6
739.39
18.31
5.53
11.41
0.12
0.01579
12.454
B-3
31.1
37.8
180.0
738.38
19.35
5.51
11.38
0.12
0.01674
12.455
Table 4.
Observed exhaust gas concentrations, inferred air-fuel ratios, and ambient conditions; engine operation
at 85% of rated speed, 25% of load at wide-open throttle.
Test
T (room)
T(intake)
T(cyl)
P (room)
P (vapor)
CO
C02
02
CO
Spindl
Deg C
Deg C
Deg C
mm Hg
mm Hg
0/
to
%
%
AFR
W-1
26.1
29.4
158.3
739.14
11.61
4.57
11.83
0.26
0.00992
12.881
W-2
28.3
31.1
160.0
739.14
11.81
4.61
11.79
0.25
0,01010
12.858
W-3
25.6
28.3
157.2
741.68
12.97
4.26
12.05
0.28
0.01107
13.016
O-l
22.2
26.1
159.4
735.84
7.00
4.15
12.40
0.14
0.00597
13.002
0-2
21.7
25.6
160.0
735.84
7.00
4.10
12.44
0.14
0.00597
13.022
0-3
30.6
33.9
169.4
735.84
5.65
4.24
12.34
0.15
0.00482
12.974
T-1
25.6
29.4
162.2
734.31
10.22
4.56
12.12
0.20
0.00878
12.880
T-2
26.1
30.6
160.6
734.31
10.46
5.53
11.33
0.20
0.00899
12.490
T-3
24.4
26.7
160.0
734.31
12.64
4.85
11.73
0.20
0.01089
12.746
K-1
26.1
28.3
161.1
746.25
9.16
5.55
11.90
0.12
0.00773
12.512
K-2
25.6
28.9
160.6
746.00
9.17
5.72
11.77
0.13
0.00774
12.453
K-3
26.1
27.8
161.7
743.71
11.79
5.78
11.71
0.12
0.01002
12.422
B-1
27.8
32.6
155.0
739.90
17.15
4.63
12.05
0.05
0.01476
12.761
B-2
29.4
36.1
158.3
739.39
18.31
4.93
11.83
0.04
0.01579
12.638
B-3
31.1
36.7
160.0
738.38
19.35
4.88
11.86
0.04
0.01674
12.657
84
-------
Table 5.
Observed exhaust gas concentrations, inferred air-fuel ratios, and ambient conditions; engine operation
at 85% of rated speed, 10% of load at wide-open throttle.
Test
T (room)
T(intake)
T(cyl)
P (room)
P (vapor)
CO
C02
02
CD
Spindt
Deg C
DegC
Deg C
mm Hg
mm Hg
%
%
%
AFR
W-1
26.1
29.4
149.4
739.14
11.61
4.10
12.16
0.22
0.00992
13.043
W-2
27.8
30.6
150.0
739.14
11.81
3.75
12.34
0.23
0.01010
13.183
W-3
25.5
28.9
147.2
741.68
12.97
3.71
12.41
0.25
0.01107
13.215
0-1
23.3
26.7
149.4
735.84
7.00
3.63
12.73
0.13
0.00597
13.196
0-2
22.8
26.1
149.4
735.84
7.00
3.57
12.78
0.12
0.00597
13.214
0-3
31.1
34.4
158.3
735.84
5.65
3.95
12.55
0.12
0.00482
13.069
T-1
24.4
26.7
148.9
734.31
10.22
4.32
12.27
0.10
0.00878
12.911
T-2
23.3
25.0
146.1
734.31
10.46
4.79
11.74
0.10
0.00899
12.705
T-3
25.6
28.9
149.4
734.31
12.64
4.51
11.94
0.10
0.01089
12.816
K-1
26.1
28.3
150.0
746.25
9.16
4.56
12.53
0.09
0.00773
12.859
K-2
26.1
28.9
150.0
746.00
9.17
4.65
12.47
0.09
0.00774
12.825
K-3
25.0
27.8
150.0
743.71
11.79
4.45
12.58
0.09
0.01002
12.898
B-1
27.8
32.8
137.8
739.90
17.15
5.28
11.69
0.01
0.01476
12.499
B-2
29.4
35.0
141.7
739.39
18.31
5.19
11.71
0.00
0.01579
12.522
B-3
30.6
35.0
141.7
738.38
19.35
5.06
11.77
0.01
0.01674
12.574
Table 6.
Range in air-fuel ratio variation at different loads over the considered range of climatic conditions.
Engine load
Torque/Peak Torque
AFRjmuc - AFRmia
as a % of AFR
Coeff. of Var.
in AFR
100%
3.0%
0.98%
75%
2.8%
0.67%
50%
2.8%
1.06%
25%
5.0%
1.63%
10%
5.6%
1.87%
Table 7.
Coefficients of variance for proposed corrections for ambient effects on air-fuel ratio.
Corr.
Correction applied to
Coeff. of var.
no.
observed AFR at 100% load
in AFR
1
No correction applied
0.98%
2
Dry-air mass-flow rate correction
1.95%
3
Fuel-density correction, 0 = 0.0018/° C, Tfuei = Troom
0.80%
4
Fuel-density correction, 0 = 0.0018/°C, Tfuei = 7^,^
0.65%
5
Fuel-density correction, = 0.0036/°C, Tfuei = Intake
0.64%
6
Corr. 3 combined with corr. 2
0.69%
7
Corr. 3 combined with a moist-air mass-flow-rate corr. (18)
0.81%
8
Corr. 7 with pair ^ pair{P)- (19)
0.68%
85
-------
"12.8
224
12.4
12
11.6
I S
• •
220
216
11.2
10.8
.004
.008
.012
.016
212
• •
732 736
740 744
748
Fig. 1. Variation of observed air-fuel ratio with ambient specific humidity u,
at wide-open throttle.
/'aim (nun IIk)
Fig. 4. Variation of s|mrk-|>liig scat U:iup«.'rat»in: with aml>i(!iil pressure, ul
wide-open throttle.
OS
Tc„i "C
Fig. 2. Variation of observed air-fuel ratio with spark-plug seat temperature,
at wide-open throttle.
ft:
Pair* (mm Hg)
Fig. 5. Variation of observed air-fuel ratio with ambient pressure, at wide-
open throttle. ^
**
4/P-suJ-
6 "C
Fig. 3. Variation of observed air-fuel ratio with ambient temperature, at
wide-open throttle.
Fig. 6. Typical relation between air-fuel ratio and venturi-throat pressure drop
in an air-bleed compensated carburetor.
I2.3
J 2.2
12.1
Fig. 7. Air-fuel ratio data at wide-open throttle, before and after correction
by a combined fuel-density and moist-air mass-flow model; • , uncorrected
data; O , corrected data.
------- |