EPA-460/3-77-006
February 1977
EMISSION FORMATION
IN HETEROGENEOUS
COMBUSTION
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Waste Management
Office of Mobile Source Air Pollution Control
Emission Control Technology Division
Ann Arbor, Michigan 48105
-------
EPA-460/3-77-006
EMISSION FORMATION
IN HETEROGENEOUS COMBUSTION
by
G.L. Borman, P.S. Myers, O.A. Uyehara,
L. Evers, M. Ingham, and D. Jaasma
University of Wisconsin
Department of Mechanical Engineering
Madison, Wisconsin 53706
Grant No. R803058-01-1
EPA Project Officer: John J. McFadden
Prepared for
ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Waste Management
Office of Mobile Source Air Pollution Control
Emission Control Technology Division
Ann Arbor, Michigan 48105
February 1977
-------
This report is issued by the Environmental Protection Agency to report
technical data of interest to a limited number of readers. Copies are
available free of charge to Federal employees, current contractors and
grantees, and nonprofit organizations - in limited quantities - from the
Library Services Office (MD-35), Research Triangle Park, North Carolina
27711; or, for a fee, from the National Technical Information Service,
5285 Port Royal Road, Springfield, Virginia 22161.
This report was furnished to the Environmental Protection Agency by the
University of Wisconsin, Department of Mechanical Engineering, Madison,
Wisconsin 53706, in fulfillment of Grant No. R803058-01-1. The contents
of this report are reproduced herein as received from the University of
Wisconsin. The opinions, findings, and conclusions expressed are those
of the author and not necessarily those of the Environmental Protection
Agency. Mention of company or product names is not to be considered
as an endorsement by the Environmental Protection Agency.
Publication No. EPA-460/3-77-006
11
-------
CONTENTS
Page
Abstract ii
List of Figures iii
List of Tables vi
Acknowledgements vii
Sections
I Conclusions 1
II Recommendations 3
III Introduction 5
IV The Delayed Mixing Stratified Charge
Engine Concept 7
V The Formation of NO by Liquid Fuel 12
Combustion x
VI The Divided Chamber Engine Evaluation and
Texaco Engine Projects 51
VII Publications 104
VIII Glossary 105
IX Appendix 107
-------
ABSTRACT
Three research projects are reported under the grant.
The first is an investigation of a stratified engine con-
cept in which spark ignited combustion in an engine with a
homogeneous rich charge is completed and then air is in-
jected during the expansion stroke giving a leaner overall
fuel-air ratio. The study showed substantial reduction of
nitric oxides without increasing other emissions. Combus-
tion efficiency was not increased and, because substantial
work was needed to supply the compressed air, the engine
efficiency was decreased. The second project is an investi-
gation of nitrogen oxides produced by burning of liquid nor-
mal heptane from a fuel wetted porous cylinder in a cross
flow of air. Variation of free stream air velocity and cy-
linder diameter showed the moles of nitric oxide per mole of
fuel burned to be a weak function of Reynolds number. Soot
produced by the flame and collected downstream has been
identified as giving off significant amounts of nitric oxide
indicating a carbon, nitric oxide interaction in the flame
envelope. The third project consisted of a study of the
part load operation of the Newhall divided chamber engine
previously developed at U.W., Madison, an emissions and
fuel economy evaluation of this engine relative to other
engines and initiation of a study of the formation of hydro-
carbons in a Texaco engine by utilization of an in-cylinder
sampling technique. Reasons for abandonment of the divided
chamber engine were its higher hydrocarbons and lower fuel
economy relative to other engines and its sensitivity to
knock.
This report was submitted in fulfillment of Grant
Number R803858-01 by the University of Wisconsin under the
partial sponsorship of the Environmental Protection Agency.
Work was completed as of February 9, 1977.
ii
-------
FIGURE
No. Page
1 Schematic Representation of the Experiment. 14
2 Air Supply System. 15
3 Cross Section of Porous Cylinder Model. 17
4 Fuel Supply System. 19
5 Collection System. 20
6 Apparatus for NO Profile Determination. 23
7 Typical NO/NO Concentration Measurement. 27
J\.
8 Determination of Noise Source. 28
9 NO Concentration Following Removal of Porous
Cylinder Model. 30
10 Results of Experiments With Sooted and Clean
Collectors. 31
11 Apparatus Used to Measure Amount of NOX in
Soot. In 1, Soot is Collected in Inlet Tube
and on Filter. In 2, Filter and Tube are
Heated to Drive Off NO . 33
.A.
12 Appearance of Flame at Different Fueling Rates. 37
13 Effect of Fuel Flow Rate on Molar Emission
Indices at Constant Cooling Water Flow Rate
for .53 x 1.3 cm Porous Cylinder Model in
32 cm/sec Air Stream. 38
14 Apparatus for Determination of Where NO,, is
Produced. 40
15 Effect of Fuel Center Temperature in NO and
NOX Emission Indices and Burning Rate. Center
Temperature Varied by Changing Cooling Water
Flow Rates Air Velocity = 32 cm/sec, .52 x
1.3 cm Cylinder. 41
16 NO Concentration Profile 50° From Stagnation
Point of a 1.3 cm Porous Cylinder Model.
Air Velocity = 30 cm/sec. 46
17 Molar Emission Indices for Porous Cylinder
Models Plotted Against Reynold Number Based
on Free Stream Properties. Mean Values are
Plotted With Bars Showing Highest and Lowest
Measured Values. Numbers Under Lower Bar
Indicates Number of Points Averaged. 48
111
-------
No,
18 Divided Chamber Engine Cylinder Head (Top
View) - Final Part Load Hardware Configura-
tion. 60
19 Measured NO Concentration Versus Overall
Equivalence Ratio - 105° BTDC Injection. 62
20 Measured NO Concentration Versus Overall
Equivalence Ratio 70° BTDC Injection. 63
21 Measured CO Concentration Versus Overall
Equivalence Ratio 105° BTDC Injection. 65
22 Measured CO Concentration Versus Overall
Equivalence Ratio - 70° BTDC Injection. 66=
23 Measured HC Concentration Versus Overall
Equivalence Ratio 105° BTDC Injection. 68
24 Measured HC Concentration Versus Overall
Equivalence Ratio - 70° BTDC Injection. 69
25 Calculated Specific Fuel Consumption ISFC
Versus Overall Equivalence Ratio 105°
BTDC Injection. 70
26 Calculated Specific Fuel Consumption (ISFC)
Versus Overall Equivalence Ratio - 70° 72
BTDC Injection.
27 Calculated Mean Effective Pressure (IMEP)
Versus Overall Equivalence Ratio - 105°
BTDC Injection. 73
28 Calculated Mean Effective Pressure (IMEP)
Versus Overall Equivalence Ratio 70°
BTDC Injection. 74
29 Calculated Specific NO Versus IMEP at Con-
stant Equivalence Ratios 105° BTDC
Injection. 76
30 Calculated Specific NO Versus IMEP at Con-
stant Equivalence Ratios - 70° BTDC Injection. 77
31 Calculated Specific CO Versus IMEP at Con-
stant Equivalence Ratios - 105° BTDC Injec-
tion. 78
52 Calculated Specific CO Versus IMEP at
Constant Equivalence 70° BTDC Injection. 79
53 Calculated Specific HC Versus IMEP at
Constant Equivalence Ratios - 105° BTDC
Injection. 80
IV
-------
No. Page
34 Calculated Specific HC Versus IMEP at
Constant Equivalence Ratios 70° BTDC
Injection. 81
35 Calculated ISFC Versus IMEP at Constant
Equivalence Ratios 105° BTDC Injection. 82
36 Calculated ISFC Versus IMEP at Constant
Equivalence Ratios 70° BTDC Injection. 83
37 Typical Primary Chamber Combustion Pressure
Trace. 85
38 Calculated Specific NOX. CO, HC and Fuel
Consumption Versus IMEP For Dual Ignition
Tests. 87
39 Specific NOX Versus IMEP Divided Chamber
Engine Comparison. 90
40 Specific CO Versus IMEP Divided Chamber
Engine Comparison. 91
41 Specific HC Versus IMEP - Divided Chamber
Engine Comparison. 93
42 Specific Fuel Consumption Versus IMEP
Divided Chamber Engine Comparison. 94
43 Sampling Valve Design for TCCS Hydrocarbon 99
Study.
v
-------
TABLES
No.
1 Amounts of NO and NO in Soot and Bulk Gases
2 Molar Emission Index Test Data Cylinder
Size .36 x 2.1 cm 43
3 Molar Emission Index Test Data Cylinder Size
.52 x 1.3 cm 44
4 Molar Emission Index Test Data Cylinder Size
0.52x1.3 cm 45
5 Molar Emission Index Test Data Cylinder Size
0.67 x 1.3 cm 46
6 Divided Chamber Engine - Operating Conditions
Tested 56
VI
-------
ACKNOWLEDGEMENTS
We wish to thank Mr. William Tierney, Mr. Edward
Mitchell, Mr. Robert Canup and most particularly Mr.
Martin Alperstein of Texaco for their continuing sup-
port and technical assistance with the TCCS hydrocarbon
sampling project.
We are grateful to the U.S. Army TARADCOM for pro-
viding the M-151 TCP engine for this project, and we
would like to thank Mr. Edward Rambie and Mr. Walter
Bryzik of TARADCOM and Mr. Andrew Lennert, Mr. Richard
Sowls and Major Brian Harwood of the Arnold Research
Organization whose efforts were instrumental in obtain-
ing this engine for us.
vn
-------
SECTION I
CONCLUSIONS
DELAYED MIXING ENGINE CONCEPT PROJECT
a. Computer comparison of two mixing strategies for
stratified charge combustion, i.e., gradually
mixing successive portions of the rich products
into air or vice versa, indicates that successive-
ly mixing portions of the air into the rich pro-
ducts gives reasonable efficiency, low nitric
oxide concentration, and high specific power for
operation near stoichiometric.
b. An experimental study of this concept using a
laboratory engine with outside compression of the
air to be mixed with the rich products verified
the low nitric oxide concentration and high
specific power but showed relatively poor fuel
economy due to the work required to compress
the air to be mixed with the rich products.
FORMATION OF NO BY LIQUID FUEL COMBUSTION PROJECT
a. There is preliminary, but convincing, experimental
evidence suggesting that carbon absorbs significant
quantities of NO thus serving as a sink for NO .
.A. ' .A.
Work should be done to conclusively verify this
observation and to understand how NO is absorbed
J\.
and released since carbon is usually produced
during heterogeneous combustion.
b. Preliminary data indicate that molar emission
indices for diffusion type flames surrounding a
porous cylinder are relatively insensitive to
changes in the Reynolds number based on approach
stream properties.
-------
THE DIVIDED CHAMBER ENGINE EVALUATION AND TEXACO ENGINE
PROJECTS
a. The divided combuster chamber engine produces lower
NO emissions than conventional engines but has
A
higher specific fuel consumption than competitive
hybrid engines having comparable NO emissions.
.?c
b. Hybrid engines have better fuel economy but their
hydrocarbon emissions are high. An in-cylinder
sampling study should be conducted on a hybrid
engine to determine the mechanism and location of
hydrocarbon formation since this could lead to a
low emission engine having good fuel economy.
-------
SECTION II
RECOMMENDATIONS
THE DELAYED MIXING STRATIFIED ENGINE CONCEPT
a. That no additional work be done on the delayed mixing
concept unless a practical method of compressing the
required air which requires significantly less com-
pression work than that required in the experimen-
tal set up is found.
THE FORMATION OF NOV BY LIQUID FUEL COMBUSTION
A
a. That sufficient additional research work be done to
determine whether particulate emissions from en-
gines and stationary combustion sources carry with
them significant amounts of nitric oxide. If the
preliminary findings that there is significant
transport of NO by particulates are confirmed
A.
the mechanisms of formation and/or absorption and
release of the NO should be determined.
JC
b. That the effects of fuel composition and free
stream temperature and composition on emission
indices be evaluated.
c. That emission indices be determined for situations
simulating multiple droplets (probably multiple
porous or absorbent cylinders) with the objective
of producing emission criteria for dense and/or
dilute sprays.
THE DIVIDED CHAMBER ENGINE EVALUATION AND TEXACO ENGINE
PROJECT
a. That work on HC formation be done on the Texaco
engine rather than on the divided chamber engine
because of the fuel economy advantage of the
Texaco engine.
-------
b. That samples be taken at different times and loca-
tions in the clearance and displacement volumes of
of the Texaco engine in an attempt to determine
where and how hydrocarbons fail to be oxidized and
are consequently emitted through the exhaust.
-------
SECTION III
INTRODUCTION
The work done under this grant consists of three dif-
ferent projects each being the doctoral thesis topic of a
graduate student. Dr. L. Evers worked on the project
titled, "The Delayed Mixing Stratified Charge Engine Con-
cept" and upon graduation joined the ERDA staff at
Bartlesville. Because his project work has been published
as a separate report, only a brief discussion of the re-
sults are given here. Mr. D. Jaasma is working on the pro-
ject titled, "The Formation of NO By Liquid Fuel Combustion".
This project is concerned with the burning of liquid fuels
in a diffusion flame which is supported by a fuel wetted
porous cylinder. The cylinder is in a cross flowing air
stream. The aim of the project is to produce correlations
of moles of NO per mole of fuel burned as a function of
.A.
porous cylinder diameter, fuel type and flow stream velocity,
composition and temperature. Mr. M. Ingham is working on
the project entitled, "The Divided Chamber Engine Evaluation
and Texaco Engine Project." The first phase of this project
was to evaluate the emission and fuel economy performance
of the Newhall-U.W. divided chamber engine at part load
conditions. After extensive testing and comparisons with
other engine data it was concluded in consultation with
the EPA Project Officer that continuation of the second
phase would be more meaningful if carried out on a more
promising engine. The second phase of the project was to
deal with an investigation of the sources of unburned
hydrocarbons in the engine. It was decided therefore to
obtain a Texaco TCCS single cylinder engine and conduct
sampling probe studies on it to try to determine the
mechanisms resulting in unburned hydrocarbon emissions.
-------
Because of this switch, probe data from the Texaco engine
had not yet been obtained at the termination of the grant,
Although all three projects concern heterogeneous
combustion they are different enough that presentation of
each in a separate section seemed desirable. Thus the
following three sections deal with the three projects in
turn.
-------
SECTION IV
THE DELAYED MIXING STRATIFIED CHARGE ENGINE CONCEPT
One portion of the work conducted under this (and pre-
vious) EPA grant was completed during this grant period.
(See Section VII of this report.) Five publications have
been presented covering the work done. In order that the
reader of this report can assess for himself whether or
not he wishes to read these publications to obtain a de-
tailed understanding of the work done a brief synoposis of
the work done will be presented here.
Computer studies conducted under a previous EPA grant
indicated that, in an engine having a stratified charge
consisting of a rich mixture plus a lean mixture or, in
the limit, air, the method of burning the rich mixture and
mixing the resulting products with the remaining air to
obtain an overall lean mixture markedly influenced the
variation in specific NO (grams per KW) produced as the
.A.
air-fuel ratio was varied.
Two strategies were investigated analytically. In
both strategies combustion was started in the rich region
of the charge. In the first strategy, called instantaneous
mixing, as soon as the rich products of combustion of a
portion of the mixture were formed they were thoroughly
mixed with the lean mixture (air). Especially during the
early portion of the process, the resultant product mix-
ture was cool and lean, consequently limiting the nitric
oxide formation. This strategy was very effective for
very lean overall fuel-air mixtures. As the overall fuel
air mixture approached stoichiometric proportion, however,
the product mixture temperature increased. Higher tempera-
tures and the presence of excess oxygen resulted in higher
nitric oxide concentrations.
-------
The disadvantage of the instantaneous mixing approach
was that low nitric oxide concentrations were associated
with low specific power because of the lean overall fuel-
air mixture. The advantage of this approach was that,
assuming burning rates could be maintained, lean overall
mixtures resulted in both low nitric oxide concentrations
and higher efficiencies.
In the second strategy, called delayed mixing, the
rich products of combustion were not mixed with the gas
(air) in the lean region until combustion was complete.
A major difference, however, was that the lean region was
mixed into the rich products instead of vice versa as for
the instantaneous mixing concept. Thus the product mixture
was initially rich and was gradually diluted to the final
overall fuel-air ratio. When the final fuel-air ratio was
stoichiometric, the nitric oxide formation was low because
of the lack of oxygen in the early high temperature part
of the expansion and because of the low temperature in the
later part of the expansion. However, if the overall fuel-
air ratio was made very lean, high temperatures and excess
oxygen would be present early in the expansion stroke and
would result in high concentrations of nitric oxide. The
disadvantage of the delayed mixing strategy was that high
efficiency (lean overall operation) was associated with
high nitric oxide concentration. The advantage indicated
by the computer model was reasonable efficiency, low nitric
oxide concentration, and high specific power for operation
near stoichiometric.
When the nitric oxide emissions of various types of
internal combustion engines were examined, it appeared that
some engines employed a limited mixing concept. The pro-
cess began with rich combustion followed by a period of
limited mixing. The final stage involved mixing of air
into the products, or products into the air, or air and
products into each other. This group of engines had trends
8
-------
in nitrogen oxide emissions similar to those of the delayed
mixing concept. The nitrogen oxide emissions were low for
operation near stoichiometric fuel-air ratios and not sen-
sitive to changes in fuel air ratios.
An experimental program was set up to simulate the com-
bustion processes of the delayed mixing stratified-charge
engine concept. The experimental engine was not intended
to represent a practical engine configuration , but was
simply a means for testing the delayed mixing concept. A
single-cylinder spark-ignited carbureted CFR engine was
modified to include an air injection valve which could in-
ject air at the desired time in the engine cycle. The ex-
perimental engine was operated at a variety of overall fuel
air ratios and air injection timings to evaluate the de-
layed mixing concept.
The high pressures associated with combustion required
high pressure air. This air came from storage tanks; through
a pressure regulator; through a flow'restriction used for
metering purposes; through a flow controlling valve; through
a surge tank; and finally through a short duration (40 crank
angle degrees) variable-timing cam-operated valve to a noz-
zle and to the combustion chamber.
In general, the engine was operated near 800 rpm at a
compression ratio of seven using iso-octane as a fuel with
the maximum air flow rate through the carburetor. The
spark timing was varied in an attempt to operate at minimum
advance for best torque (MET) but these conditions were not
always obtained. The high pressure air was normally intro-
duced after or during the last part of combustion. However
some runs had variable injection timing (including well
before spark) to show the influence of the entire range of
air-injection timings.
To establish a baseline the engine was first operated
without air injection over a range of fuel-air ratios with
power output, fuel economy and emissions being measured. As
-------
expected, the maximum NO emissions occured at slightly lean
A.
conditions.
Using air injection into a rich mixture to produce an
overall stoichiometric mixture NO emissions were essential-
J\.
ly equivalent to the homogeneous case whenever air injec-
tion started before about 30 degrees BTDC but decreased
rapidly as air injection was retarded. Starting air injec-
tion at TDC or later produced low NO emissions, i.e., 80-
j\.
100 ppm. This suggests that when injected before combus-
tion the injected air mixes quickly with the rich mixture
to form essentially homogeneous mixture having the equiva-
lence ratio of the overall mixture and producing comparable
NO emissions when burned. When air was injected during
JC
the latter part of or after combustion very little NO was
J\.
formed from the additional oxidation because of the lower
temperatures.
In general, the CO concentrations using air injection
were directly comparable to those observed when operating
the engine on a homogeneous mixture having an equivalence
ratio equal to the overall fuel-air ratio using air-
injection. The two exceptions to this were that, when air
injection occurred very late (92 degrees ATDC) or during
the middle of the combustion period (near TDC), higher CO
concentrations were observed. In the first case it was
thought that the air was injected too late to destroy the
CO from both a time and temperature standpoint. In the
second case it appeared that the injected air interferred
in some way (possibly by cooling) with combustion and
produced excessive CO.
Speaking of HC's, with the exception of an air-injection
timing of 8° BTDC, the values observed with air injection
were generally lower than those observed with homogeneous
combustion at the same overall equivalence ratio. The high
observed values at 8° BTDC are undoubtedly the result of
disturbing combustion by air injection and are probably
10
-------
related to the high CO observed with this air injection
timing.
The indicated efficiency with air injection (not taking
into account the work of compressing the air) increased
when using leaner overall mixtures partly for thermodynamic
reasons and partly because more compressed air was put in-
to the cylinder. When the work of compressing the required
high pressure air was taken into account, the indicated
efficiency decreased as the overall equivalence ratio de-
creased, i.e., became leaner. This indicates that the work
used in this set up to compress the air was large and re-
quired more energy than could be obtained from the addition-
al burning of the rich mixture. While a practical engine
might not use air compression outside the cylinder and cer-
tainly would not include a metering system, these data raise
serious questions as to the viability of the process from a
fuel economy standpoint.
Speaking of engine power, when the IMEP was corrected
by subtracting the estimated work required to compress the
injected air, the magnitude of the corrected IMEP is equal
to or less than the corresponding homogeneous operation.
Thus again emphasizes the importance of the work of com-
pressing the air.
In summary, the delayed mixing stratified-charge en-
gine concept was an effective method of controlling the
nitric oxide emissions. The carbon monoxide emissions were
equivalent to those of homogeneous engines and the HC's
were lower. The disadvantage of the delayed mixing con-
cept was its low efficiency brought about primarily be-
cause of the large amount of work required to produce the
injected air.
11
-------
SECTION V
THE FORMATION OF NOX BY LIQUID FUEL COMBUSTION
INTRODUCTION
Liquid fuel diffusion flames are common phenomena occur-
ring in oil fired burners such as home furnaces, gas turbine
combustors, and utility boilers. Oxides of nitrogen (NO and
N02, hereinafter termed NO ) are produced in such flames,
.X
and because of their environmental impact there is interest
in reducing the amount of NO which is formed in combustion
.A
processes.
One solution to the problem is to find ways to burn
fuels so that minimal amounts of NO are formed. In the case
x
of premixed flames NO formation is a post flame occurrence.
J\.
This has allowed analytical models to be used to seek out
ways of burning fuels in an environmentally acceptable man-
ner. Models are also being used to study the effects of
parameters on NO formation in liquid fuel diffusion flame
devices, but this type of work is hindered by the fact that
these systems are heterogeneous in nature and analytical
solutions for the space and time dependent temperatures and
concentrations are not easily obtainable. Even the case of
a single drop of fuel under many simplifying assumptions is
difficult to solve for the amount of NO produced.
Due to the desirability of using an analytical approach
to find cleaner ways to burn liquid fuels as diffusion flames
and the difficulties involved in applying such an approach,
it was deemed reasonable to follow the same procedure used
in heat transfer and fluid dynamics when the analytical pro-
blems become intractable, namely, to form an empirical cor-
relation. The work reported herein represents the first
steps of an effort to experimentally correlate the molar
emission index (molar emission index = gram mol of pollutant
12
-------
formed/gram of fuel burned) for burning droplets as a function
of independent variables such as fuel, drop size, velocity of
air relative to the drop, composition of the air around the
drop, and temperature of the air stream around the drop.
Such a correlation is of interest over a range of Reynolds
numbers from 0 to the point where transition to wake flame
occurs.
EXPERIMENTAL APPARATUS
Figure 1 shows a schematic diagram of the experimental
apparatus used in this investigation. The air supply system
provides a jet of air in which the experiments are performed,
and a porous cylinder model is used to simulate the burning
droplet. The collection system is used to catch all of the
products of combustion from the experiment, along with the
inevitable dilution air from the surroundings. The systems
are described separately in detail.
Air Supply System
The purpose of the air supply system is to provide a
region of uniform air velocity, the temperature of which
can be varied over a range of 20-1100°C. The design of
the system includes a cross flow heat exchanger and is
shown in Figure 2. The amount of air to be heated is deter-
mined by varying the pressure upstream of a critical flow
orifice, which was calibrated using a positive displacement
gas meter. The metered air enters the steel shell of the
heat exchanger at one end, is forced by baffles to pass
through a series of fourteen cross flow core sections, and
passes out of the shell through a vertically oriented con-
verging nozzle with a rectangular outlet of size 5.6 x 4.9
centimeters. An industrial burner using propane as fuel
provides hot products of combustion which pass in a straight
line through the fourteen core sections and then out an ex-
haust pipe at the other end of the exchanger. All internal
13
-------
BLOWER
SHARP EDGED ORIFICE
EXHAUST
SAMPLE
COLLECTION
SYSTEM
NO/NOX ANALYSER
ENVELOPE FLAME
MEASURED*o
sfflBv
POROUS CYLINDER DROPLET MODEL
AIR SUPPLY
SYSTEM
Figure 1. Schematic Representation of the Experiment
14
-------
REGULATOR
rCRITICAL FLOW
\ ORIFICE
HOUSE AIR
REFRACTORY ,^TEST AREA
NOZZLE A |ff
©GAGE
- -3
PRODUCTS
o
COMBUSTION
BURNER
CROSS FLOW CORE SECTIONS (14)
STEEL SlHELL
Figure 2. Air Supply System.
15
-------
parts of the heat exchanger are made of refractory materials,
with insulating fire brick being used for the support and
baffling materials. The brick is coated with refractory
mortar to reduce its porosity, and the air outlet nozzle is
a refractory casting. The steel shell of the heat exchanger
is supported by a steel angle frame and made air tight, the
only inlets and outlets being for the air and propane pro-
ducts. A great deal of effort was spent trying to produce
acceptable sealing between the air and products side of the
heat exchanger, but the internal leakage was never reduced
to acceptable levels. Consequently, it was decided to
operate the heat exchanger as a heat storage device by first
capping off the air inlet and outlet and running the propane
burner long enough to heat up the core sections and surround-
ing bricks to a very high temperature. The burner inlets and
exhaust could then be sealed off and the air supply turned
on to flush out the products of combustion of the burner and
to provide the hot air for the experiments. Using this
technique, an air outlet temperature of over 500°C at 7.8
kilograms per hour was achieved, with the temperature of the
air falling off so slowly that it could be assumed constant
for data taking purposes. Due to problems with warpage of
the stainless steel at the nozzle outlet, the system has
not yet been tested to determine the maximum possible air
temperature it can deliver. As mentioned previously, the
design condition has an outlet temperature of 1100°C, and
it is expected that temperatures in this range can be
achieved once the nozzle outlet problem is solved.
Experimental Models of Burning Droplets
A cross section of the porous cylinder experimental
model of the burning droplet is shown in Figure 3. The
porous section is made of a sintered bronze material which
is commonly impregnated with oil and sold as a pre-lubricated
bearing. Some sizes used were available off the shelf, while
16
-------
-WATER RETURN JACKETS-
PLUG
WATER INLET TUBE
^THERMOCOUPLE
FUEL INLET TUBE \uPOROUS BRONZE
THREAD WRAPPING
THERMOCOUPLE
PROTECTION
TUBE
Figure 3. Cross Section of Porous Cylinder
Model.
17
-------
the others were machined from bars of this material. The
cylinders are washed thoroughly in trichlorethylene to re-
move the oil, and then soft soldered at the ends to four
tubes, two for cooling water return, one for a fuel feed,
and another to insert a thermocouple for measurement of the
fuel temperature at the center of the cylinder. The larger
diameter cylinders also require end plugs as shown to keep
the water and fuel separated. An additional tube is inserted
from each end around the innermost tube and inside the water
return tube. Cooling water is fed down the additional (water
feed) tubes, turns around at the ends of the porous cylinder,
and exits through the water return tubes. The cooling water
is necessary to allow the fuel to reach the cylinder in a
liquid condition, particularly when the cylinder is placed
in a high temperature air stream. Cooling water flow through
each side is separately controlled and measured by means of
rotameters. In order to insure that the entire porous cy-
linder is fuel covered, a .2 mm diameter cotton thread is
wrapped around the cylinder and secured at each end by a
small amount of epoxy cement. The capillary action of the
thread is sufficient to cause the entire cylinder to be
wetted with the fuel.
Fuel Supply System
The fuel supply system for the experimental models is
shown in Figure 4. Fuel flow is controlled by varying the
air pressure above the fuel and is measured using a rota-
meter which was calibrated using a knife edge balance to
weigh the amount of fuel passed in a given amount of time.
Collection System
The products of combustion from the droplet model are
collected by a water cooled system as shown in Figure 5.
The cooling is required primarily for the heated air experi-
ments. The products of combustion enter at the bottom of
18
-------
HOUSE
AIR -
SUPPLY
PRESSURE
REGULATOR
TV
GAGE
(7)
FUEL TANK
FUEL TO
MODEL
ROTAMETER
[_CAPILLARY TUBING TO
CREATE PRESSURE DROP
Figure 4. Fuel Supply System.
19
-------
SOLENOID
VALVE TO
ADD MAKE
UP WATER
COOLING WATER OUT
T\
)
t
i
»•—
\
Av
\
^^^-WATER JACKET
ORIFICE
^PRODUCTS OUy EXHAUSTv
T"j ^~
^
-, n , FLOW
-i U 1 CONTROL
U VALVE
i •
J NO/NOX SAMPLE
^PRODUCTS IN
-rnm IMC UWATCD IM
-OVERFLOW
-ELECTRIC HEATING ELEMENT
RESERVOIR
Figure 5. Collection System,
20
-------
the cooler and pass through 4 meters of copper tube which is
surrounded by cooling water. Problems with condensation of
products in the cooler led to installation of an electric
heating element to keep the cooling water at a minimum of
49°C. The temperature is controlled by a sensor which turns
on a solenoid operated valve when the cooling water reaches
the preset temperature level, adding room temperature water
to the system. In this way the cooling water temperature
is maintained to within ± 2°C of the nominal set point,
which can be anywhere from 49 80 °C. The cooled products
and excess air pass through a sharp edged orifice which had
been calibrated using a positive displacement gas meter. A
blower is used to pull the products through the system, and
the gas sample is drawn off downstream of the orifice.
Instrumentation
The independent variables in the experiment are the
physical description of the porous cylinder (diameter and
length), air stream velocity, air stream temperature, cool
ing water flow rate (through the porous cylinder supports),
and collector flow rate. The air stream velocity is cal
culated from a knowledge of the pressure upstream of the
critical flow orifice and the temperature of the air at
the nozzle exit. The pressure is measured with a bourdon
tube gage, and the temperature is measured with a chromel
alumel thermocouple located in the exit plane of the nozzle.
Cooling water flow is varied by use of needle valves and
measured with rotameters. The temperature of the fuel at
the center of the cylinder depends on the amount of cooling,
and can be varied over the range 50-95°C. The collector
flow is measured using a sharp edged orifice and a water
manometer. A positive displacement gas meter was used to
calibrate several different size orifice plates so that
no matter what the collector flow rate is the differential
pressure read on the manometer is in the range of 6-50
centimeters of water when the appropriate orifice is used.
21
-------
The dependent variables are the concentrations of NO and
NO in the collector stream, the burning rate of the fuel,
A,
and the temperature and pressure of the product stream as
measured at the orifice. Concentrations of NO and NO are
J\.
measured continuously using a chemiluminescent analyzer,
with the sampling probe location downstream of the orifice
as shown in Figure 1. The sampling probe is located in
this position because it was believed that the products/
air mixture entering the cooler could be somewhat strati-
fied, but would be well mixed by the time it got to the
sample point. The accuracy of the reading is limited to
± 5%, since the calibration gas is known only to this
accuracy. A strip chart recorder is used to record the
readings as a function of time. A thermocouple is located
near the orifice, and the manometer used for measuring the
differential pressure could also be used to measure the
difference between the upstream pressure of the orifice
and the ambient. This difference is small enough to be
neglected in the subsequent computations of emission in-
dices .
The chemiluminescent analyzer is also used to obtain
nitric oxide concentration profiles by sampling flame gases
directly into the analyzer through a quartz microprobe. A
vacuum pump is used in place of the low pressure bypass
pump and the sample pressure regulator is adjusted so that
a choked flow condition is maintained at the probe inlet.
The set up is shown in Figure 6.
EXPERIMENTAL TECHNIQUE
In most of the tests a single 'independent parameter is
varied through its range of interest, while all the other
independent variables are held constant. The sequence of
events for a run is as follows.
The collector-cooler is warmed up to at least 49°C to
prevent condensation of products before they become well
22
-------
QUARTZ
MICROPROBE
VACUUM
PUMP
POROUS CYLINDER
Figure 6. Apparatus for NO Profile Determination,
23
-------
mixed with excess air. A porous cylinder model is placed
over the outlet of the air nozzle and the dimensions are
noted. The blower is then turned on and the collector flow
rate adjusted by means of a valve. Air stream velocity and
the cooling water flows for the .porous cylinder model are
then adjusted. The fuel supply is turned on and the cylin-
der is lit with a match when the fuel appears on its sur-
face. The air pressure in the fuel tank is then adjusted
so that the thread appears wet with fuel. Readings of NO
or NO concentrations, fuel flow, collector flow, and pro-
X
duct gas temperature downstream of the orifice are made
after the strip chart recorder shows a constant signal level
for 1-2 minutes. Noise in the recorder signal necessitates
a visual interpretation of the average signal level during
this time. The parameter which is under study for the
particular test run is then changed and new readings are
taken after the NO and NO readings are again stabilized.
A.
When the air velocity and amount of cooling are varied, a
readjustment of the fuel flow is required following each
change, since the mass burning rate of the fuel is dependent
on these parameters.
In the NO profile determination, the quartz probe is
adjusted to the desired position by means of a micromanipulator.
The flow rate through the analyzer is very low in this sit-
uation due to the small size of the probe orifice (about 50
micrometers}, and it takes up to 5 minutes for the instru-
ment to respond completely to a change in probe position.
COMPUTATION OF EMISSION INDICES
In order to compute the molar emission indices for a given
set of experimental conditions, it is necessary to know the
mass burning rate of the fuel and the molar rate of forma-
tion of NO and NO . The mass flow of fuel is known by means
&
of the calibrated rotameter, while the net molar rate of
formation of NO and NO is obtained from measurements of the
j\.
24
-------
total volumetric flow rate through the system (using the sharp
edged orifice) and concentration measurements for NO and NO
Jv
downstream of the orifice. This information may be converted
to a molar rate of NO by the equation
\0 = Q ' " * YNO
x x
where
MNQ = molar rate of NO , mols/second
.X
Q = volumetric flow rate through orifice,
liters/second
n = molar density of gases at the orifice,
mols/liter
Y _ = mol fraction of NO
NO x
x
The molar emission index for NO is then given by
EINO = ^NO ^fuel
x x
where
EI,,n = molar emission index of NO , (gram mols NO /gram fuel)
'NO
x
m
,. , = mass rate of fuel burned, grams/second
The molar emission index for NO may be computed from the same
formulas, if the mol fraction of NO is substituted for the
mol fraction of NO . Thus we have
x
MNO = Q ' n ' YNO
where
MNO = molar rate of NO formation
There are at least two reasons for using molar emission in-
dices (mols of pollutant/mass of fuel) instead of mass emis-
sion indices (mass of pollutant/mass of fuel). The environ-
mental impact of the emission of one mol of NO is the same as
25
-------
the impact of a mol of N02- The molar emission index gives
the same index to a drop which produces a mol of NO as to a
drop that produces a mol of NO-. The second reason is that
the molar emission index allows the relative amounts of NO
and N02 produced by a given flame condition to be compared
by looking at the emission indices for NO and NO^. If the
mass emission index is used, one must do one of two things
to compare relative amounts of NO and N02> Either one of
the emission indices must be divided by the ratio of the
molecular weights of NO and N07, or the emission index for
LJ
NO could be computed using the molecular weight of N09.
Lt
The use of the molar emission index thus eliminates
ambiguity and lends itself to easy evaluation of what is
happening in the flame.
DISCUSSION
Preliminary tests using unheated air gave NO and NO
A
concentrations on the order of 2-10 parts per million (PPM).
A typical trace on the strip chart recorder appears as shown
in Figure 7. Two tests using calibration gas and with no
flame were run to determine the cause of the noise in the
signal, and the resulting strip chart recordings are shown
schematically in Figure 8. The results indicate that the
cause of the fluctuations is in the fluid dynamics of the
product gases as they enter the cooler, rather than being
associated with the flame itself or a pulsation induced by
the blower.
During the course of one of the early experiments, the
following sequence of events occurred:
1. The experiment was run for about a half hour
at various conditions.
2. The flame was blown out and fuel supply turned
off.
3. The blower unit was turned off. while the NO
' x
analyzer was left on.
26
-------
10
9
•! 8
1 ,
0)
Q.
5
CHANGE AIR VELOCITY rNO,
AND FUEL FLOW /
*" uMMiy\ SWITCH TO NO MODE
W P — \ Vou/iTru rn Kin Mnncr
O
1
SWITCH TO NO, MODE
-NO
100 200 300 400 500
TIME, seconds
Figure 7. Typical NO/NO Concentration Measurements
A.
27
-------
COLLECTOR
/CALIBRATION GAS
CALIBRATION GAS
o.
Q.
10
UJ
o
o
o
0 50
TIME, seconds
UJ
o
o
o
0 50
TIME, seconds
Figure 8. Determination of Noise Source
28
-------
This produced the strip chart recording shown in Figure 9.
The measured concentration of NO decreased to near zero
shortly after the removal of the flame, but then increased
following shut down of the blower which had been pulling
fresh air through the collection system. In fact, the mea-
sured concentration of NO increased to levels which were
much greater than those measured during the actual running
of the experiment. It became apparent that somewhere in
the system NO was being released and that this occurred over
a period of hours, as evidenced by letting the system sit
for an hour, purging the system with the blower, shutting
off the blower, waiting five minutes, and then sampling
the collector-cooler gases. Inspection of the system re-
vealed condensed water in the cooler and orifice tubing (un-
expected) and that the entire system was coated with soot
(expected). It appeared that there were three possible
causes of this "residual NO", namely, the water, the soot,
or the copper tubing from which the collector-cooler was
made. Since NO is insoluble in water, the water was con-
sidered as a highly unlikely source. Rerunning the experi
ment with higher collector flow rates (to collect enough
excess air to prevent condensation) ruled out the water
hypothesis, as the qualitative behavior of the system was
the same.
In an attempt to determine whether the soot or the
copper was the cause of the residual NO, a mock up was
made of the collector-cooler system using clean copper
tubing. The two experiments shown schematically in Figure
10 with their resulting strip chart outputs were then con-
ducted. The first experiment used the sooted collector-
cooler and a porous cylinder flame, while the second experi
ment used the clean collector and a premixed propane-air
nonsooting flame. In the first experiment, the sooted
collector was purged with room air until there was no longer
any residual NO being measured. The porous cylinder model
29
-------
120
100
E
£ 80
O 60
z
O
r? 40
UJ
O
O
O
20
BLOWER
TURNED OFF
FLAME
REMOVED
50 100 150
TIME , seconds
200
250
Figure 9. NO Concentration Following Removal of Porous
Cylinder Model.
30
-------
SAMPLE
CLEAN
COLLECTOR
SOOTED
-COLLECTOR-COOLER
[O] POROUS CYLINDER
MODEL (SOOTY FLAME)
v-' »*/^>
SAMPLE
VPREMIXED
PROPANE-AIR
FLAME
(NON-SOOTING)
EXPERIMENT I
EXPERIMENT 2
I"
Q.
QL
H-
Z
LU
O
Z
O
O
FLAME
REMOVED
VALVE
^CLOSED
£10
r>
Q.
o"
0
55
(T
H-
Z
Ul
O
z
O
O
0
_— ,
_ FLAME
REMOVED
,_ VALVE
ACLOSED
/
\ /
\ /
\ / . .
^~~1 — 1
0 50 100 150
TIME, seconds
0 50 100 150
TIME, seconds
Figure 10.
Results of Experiments With Sooted and Clean
Collectors.
31
-------
was then used to give a bulk gas concentration of NO ty .5 PPM
for a period of 15 minutes with the blower running. Removal
of the flame was followed by a drop in the measured concen-
tration of NO, and at the end of 20 seconds the blower in-
let valve was closed. The measured concentration of NO then
increased to a value approximately seven times higher than
it had been when the porous cylinder was burning.
In the second experiment, the clean collector system
was run for 15 minutes with a propane air (non sooting)
flame at a collector flow which gave a measured concentra-
tion of NO around 5 PPM. The flame was removed, and 20
seconds later the valve controlling collector flow was again
closed. The measured concentration of NO remained very close
to zero.
Comparison of the results of these two experiments in-
dicates that the residual NO is associated with the soot,
and not the material of the collector cooler. The fact that
the measured residual NO concentrations exceeded the NO con-
centrations during the experiment is not evidence that more
NO is associated with the soot, since the concentrations
alone do not give a measure of total amount of a specie
present. Since it is of great interest to know how much NO
is bound to the soot (and hence escapes measurement by the
chemiluminescent analyzer) compared to how much NO is in
the gas phase, an experiment was performed to estimate
the ratio of the NO in the soot to the NO in the gas phase.
The experiment is shown schematically in Figure 11.
An uncooled collector was set up which allowed sampling of
the product gases 13 centimeters above the point where the
collector is attached to the product gas piping. The sample
line had a 7 micrometer filter at the end of it which seemed
adequate to trap all of the soot which entered the sampling
system. The .52 cm diameter cylinder was used to provide a
flame for 30 minutes, during which time NO and NO measure-
J\.
ments were made in the manner described earlier. Since the
32
-------
ROTAMETER
PRODUCTS
AND
EXCESS AIR
If
FILTER
FILTER INLET
ROTAMETER
ROOM AIR
OVEN AT 250°C
Figure 11.
Apparatus Used to Measure Amount of NOX in the Soot. In
1, Soot is Collected in Inlet Tube and on Filter. In 2,
Filter and Tube are heated to Drive Off NO .
-------
filter and filter inlet tube were initially clean, the soot
which accumulated in them was due entirely to the combustion
gases which were pulled through the sampling system. The
sampling system flow rate was measured by rotameter to be
43 liters per hour, and the concentrations of NO and N0x were
5.2 and 7.1 parts per million. The amount of NO and NO^
which is associated with the gas portion of the sample
stream can then be calculated, and is given in Table 1. The
sooted filter was then placed in an oven and raised to a
temperature of 250°C, during which time a flow of 40 liters/
hour of room air was passed through the filter and the con-
centration of NO was measured by the strip chart recorder.
The concentration reading rose to a maximum of 4 PPM and
fell to room level in about two hours.
The resulting curve was integrated over time to yield
the amount of NO which was associated with the soot, and
x
this value is also given in Table 1. The amount of NO
shown in the table is an estimate based on the fact that
in this type test the ratio of concentrations of NO/NO
A,
appeared to be about .9. Since it was only possible.to
monitor either NO or NO as a function of time, spot checks
-A.
of the second specie (NO) are the basis of the estimate.
It may be argued that not all the NO was driven off
J\,
of the soot by heating it to 250°C, but for lack of a hotter
oven this is the temperature the data are for. Assuming
that the sum of the measured gas phase NO and the measured
J\.
soot bound NO accounts for all the NO produced by the
x x
flame, the percentage of the total NO which is "missed" by
A.
using the gas phase estimates alone is 16%. Since the mea-
surements in this report are based on measured gas phase
NO/NO only, it is clear that they would need to be adjusted
JX
upwards by about 19% to account for all the NO produced.
This assumes that the ratio NO gas phase/NO soot is a con-
stant over all experimental conditions measured, which is
not known to be true at this time.
34
-------
Table 1. AMOUNTS OF NO AND NOX IN SOOT AND
BULK GASES
g mols from gas
phase measurements
g mols from sooted
filter and inlet tube
NO
5.0 x 10"6
1.2 x 10"6
(ESTIMATE)
NOY
A
6.8 x 10"6
1.3 x 10"6
35
-------
Several tests were fun to verify whether the parameters
which are assumed to be immaterial to the experiment did in-
deed have negligible effects on the measured emission in-
dices. These tests are discussed in this section.
The flow rate through the collector-cooler was varied
by means of a valve shown in Figure 5 which was located
between the blower inlet and the sharp edged orifice. If
this flow rate is very small, it seems clear that not all
the products of combustion will be pulled into the collector,
and the measured emission indices will be too low. As col-
lector flow is increased it is expected that less of the
products will escape the collector, and the measured index
will increase. Assuming that the collector flow rate does
not affect the amount of NO and N02 produced by the flame,
the emission index should become constant with respect to
collector flow rate as the flow rate is increased to the
point where all the products are collected. This behavior
was observed and periodic checks were made during data.
taking to insure that the apparatus was operating on the
flat portion of the curve.
A second parameter which could conceivably influence
the emission index was the rate at which fuel was supplied.
to the cylinder. Three conditions could be qualitatively
identified when cooling water flows to the model were held
constant and are shown in Figure 12. In the underfueled
condition, the thread appeared dry and the flame did not
cover the entire cylinder. The correct fueling rate caused
the thread to appear wet and the flame to cover the entire cy-
linder. In the overfueled condition, the fuel and flame
spread out onto the water return jackets. Measurements of
the emission indices for these conditions gave the results
shown in Figure 13. The NO emission index was virtually
JC
unaffected by the fuel flow, but the NO emission index de-
creased markedly with increasing fuel flow. As far as
36
-------
UNDERFUELED CORRECT FUELING OVERFUELED
Figure 12. Appearance of Flame at Different Fueling Rates
37
-------
5.0
4.5
4,01—
o
I"
»
ID
g
x 30|—
X
UJ
o
2 2.5
CO
CO
UJ
o: '•
o
2 |.o
0.5
o
o
NO o
NO x
5 10
FUEL FLOW,mg/sec
15
Figure 13
Effect of Fuel Flow Rate on Molar Emission Indices
at Constant Cooling Water Flow Rate For .53 x 1.3
cm Porous Cylinder Model in 32 cm/sec Air Stream.
33
-------
environmental considerations go, it is not very important
whether NO or N02 is produced, and so it can be argued that
the sensitivity of NO emission index to fueling rate is not
a reason to abandon the use of experimental models of this
type.
It was considered to be of interest to determine where
the N02 is being formed, and to this end the experiment
shown in Figure 14 was performed. A baseline condition was
established with the 5.2 x 12.7 mm porous cylinder, and
concentration readings of NO and NO were taken. The experi-
.A.
ment was repeated as case 1 with a measured flow of calibra-
tion gas (NO in N2) being added to the collected products at
the inlet to the cooler. The measured values of NO and NO
x
both increased by the same amount, indicating that none of
the added NO was converted to N02 between the cooler inlet
and the analyzer. Moving the tube to a position near the
bottom edge of the collector gave a similar result, leading
one to believe that the production of N02 was associated
with the flame or the gases immediately surrounding the
flame.
The amount of cooling due to the water jacket on each
end of the porous cylinder was another parameter which
could conceivably affect the results of the experiment.
To determine what effect the cooling has, the flow of
cooling water was varied and the temperature of the fuel
at the center of the porous cylinder was measured with an
iron constantan thermocouple. The results are shown in
Figure 15. Although the burning rate of fuel increases
with temperature as would be expected, the emission indices
of NO and NO were virtually unaffected at the cooling rates
X
which gave these fuel temperatures.
The results of the tests on the porous cylinder models
using N-Heptane as fuel at ambient air temperatures with no
correction for soot bound NO are presented graphically in
Figure 16 and in tabular form in Tables 2-5. The averages
39
-------
BASELINE
CALIBRATN
GAS CASE I
CALIBRATION
GAS
/I
CASE 2
Figure 14. Apparatus for Determination of Where NO,
is Produced. '
40
-------
4.0
"o
E
o>
..50
O
X
X
LJ
O
22.0
O
V)
LJ
o:
< 1.0
o
0
1 1 1 1 1 1 1 1 1 1 1
_ x x x _
x x x
O
o o o
A 0
A*
A
_^_ .
A
A
NO o
NOX x
BURNING RATE A _
1 1 1 1 1 1 1 1 1 1 1
C.*J.\J
20.0
15.0 «
o>
ul
H
10.0*
e>
o:
CD
5.0
50 55 60 65 70 75 80 85 90 95
CENTER TEMPERATURE ,°C
100 105
Figure 15.
Effect of Fuel Center Temperature on NO and NO
Emission Indices and Burning Rate. Center x
Temperature Varied by Changing Cooling Water
Flow Rates. Air Velocity = 32 cm/sec,
.52 x 1.3 cm Cylinder.
41
-------
5.0
o»
"o 3.0
I.
X*
LU
O
Z 2.0
O
LJ
o:
A
4 •
NO
A
A
CYLINDER DIAMETER x LENGTH
.36x2.1 cm •
.52x|.3cm o
.67x1.3 cm A
1
100 200
REYNOLDS NUMBER
300
Figure 16. Molar Emission Indices for Porous Cylinder Models
Plotted Against Reynolds Number Based on Free
Stream Properties. Mean Values are Plotted with
Bars Showing Highest and Lowest Measured Values.
Number Under Lower Bar Indicates Number of Points
Averaged.
42
-------
Table 2. MOLAR EMISSION INDEX TEST DATA
CYLINDER .36 x 2.1 cm.
Air
Velocity,
cm/sec
0
32
32
32
46
46
46
60
60
60
60
60
71
71
Reynolds
Numb e r
0
79
79
79
112
112
112
146
146
146
146
146
173
173
Fueling
rate ,
mg/sec
8
11
10
10
12
11
12
14
13
14
14
14
14
13
.3
.4
.2
.4
. 5
.7
.0
.4
.7
.0
. 0
.0
.4
.1
EINO'
gmol/g
3.
2.
3.
3.
2.
3.
2.
2.
2.
2.
2.
2.
2.
56 x 10 5
26
38
00
15
12
89
32
63
83
68
46
98
EINOx'
gmol/g
4
4
4
4
3
4
3
3
3
3
3
3
3
4
.64 x 10"5
.23
.62
.15
.95
.25
.96
.85
.92
.79
.97
.82
.92
.09
43
-------
Table 3. MOLAR EMISSION INDEX TEST DATA
CYLINDER SIZE .52 x 1.3 cm.
Air
Velocity ,
cm/sec
f32
variable cooling I 32
water flows J 32
with "correct" A 32
fuel feeds
32
^32
cooling water f 32
flow constant,
32
variable <^ 32
fuel feed
32
32
32
\ 32
Reynolds
Number
114
114
114
114
114
114
114
114
114
114
114
114
114
Fueling Fuel ^NO'
rate, temperature,
mg/sec °C gmol/g
7
8
11
13
12
12
7
9
5
7
8
8
5
.7 54
.3 63
.4 81
.4 89
.2 92
.0 91
.7
.1
.2
.5
.6
.6
.5
2.
3.
2.
2.
2.
2.
2.
1.
2.
2.
2.
2.
3.
90 x 10"5
02
92
90
85
68
55
50
97
82
02
02
15
EINO '
x
gmol/g
3
4
4
3
3
4
4
4
4
4
4
4
4
.89 x 10"5
.03
.01
.88
.82
.03
.05
.03
.34
.12
.16
.16
.43
-------
Table 4. MOLAR EMISSION INDEX TEST DATA
CYLINDER SIZE 0.52 x 1.3 cm.
Air
Velocity ,
cm/sec
46
46
46
46
60
60
60
60
73
73
73
Reynolds
Number
165
165
165
165
214
214
214
214
263
263
263
Fueling
rate ,
mg/sec
8.1
8.1
8.1
7.9
9.3
9.3
9.3
9.3
10.2
10.2
10.2
2
2
2
2
2
2
2
2
2
2
2
EINO'
gmol/g
.93 x 10"5
.86
.86
.92
.72
.71
.73
.73
.62
.67
.62
4
4
4
4
4
4
4
4
4
3
4
EINOX '
gmol/g
.26 x 10"5
.07
.10
.21
.25
.28
.14
.37
.03
.96
.12
45
-------
table 5. MOLAR EMISSION INDEX TEST DATA
CYLINDER SIZE 0.67 x 1.3 cm.
Air
Velocity,
cm/ sec
0
32
32
32
32
32
32
32
32
32
39
66
80
Reynolds
Number
0
146
146
146
146
146
146
146
146
146
178
303
366
Fueling
rate,
mg/sec
5.8
6.7
9.1
9.3
9.5
9.5
9.5
9.5
9.3
16.1
9.3
11.0
11.7
EINO'
gmol/g
3.11 x 10"5
3.02
2.91
2.85
2.88
3.16
2.88
3.16
3.19
2.72
3.00
2.67
2.69
EINOX'
gmol/g
4.11 x 10 5
4.20
4.17
4.01
4.11
-
-
-
4.20
3.95
4.23
3.84
3.79
46
-------
of the data are plotted against a Reynolds number based on
the properties of air at the approach stream condition.
Within the range and repeatability of the data, the emission
indices are relatively unaffected by Reynolds number and do
not seem to vary significantly due to diameter or air stream
velocity alone.
It is regrettable that due to equipment problems with
the air heating system we were unable to obtain data for
the high temperature air stream. The emission index mea-
surements to date do not seem to be highly sensitive to the
independent parameters varied during our investigation. It
will be of great interest to see if the sensitivity to am-
bient temperature (predicted by simple analytical models)
can be demonstrated experimentally, and also to see what
effect the addition of product gases to the ambient air
stream has on emission indices.
Attempts to obtain concentration profiles for NO around
porous cylinder models produced unexpected results. The
profiles appear to have two maxima, one near the surface of
the model and the other on the air side of the luminous zone,
as shown in Figure 17. This type of profile has also been
reported by Hart et al. for measurements along the stagna
tion line of porous cylinders burning N-Heptane doped
with pyridine. If the data are correct, then there is pro-
duction of NO in the cool and relatively oxygen free region
near the surface of the simulated droplet. The dip in con-
centration between the two maxima then indicates that a
destruction reaction or absorption mechanism exists. Both
these phenomena are not predicted by the familiar Zeldovich
kinetics and leads one to question whether the dip in the
profile is an antifact due to the disturbance introduced by
the probe. However, the discovery of the interaction
between soot and NO providesa possible basis for a NO gas
phase concentration reduction mechanism, since soot forma-
tion and the reactions leading up to it may be expected to
47
-------
LUMINOUS
O\J
70
60
-
o. 50
°-
2 40
ct:
H
Z 30
LU
0
§20
10
0
ZONE |
0
___ __
• . :•
o o
00 —
00
O 0
0
o
o
— —
0
0
1 °
5 10
DISTANCE FROM SURFACE.mm
Figure 17.
NO Concentration Profile 90° From Stagnation
Point of a 1.3 cm Porous Cylinder Model. Air
Velocity = 30 cm/sec.
48
-------
take place in the region of the minimum. If this is true,
the soot may then be thought of as a "sink" for gas phase
NO in the region where the minimum in NO concentration
occurs, and such a dip in the profile would be expected to
occur. Some portion of the NO trapped by or reacted with
the carbon may be driven off as the carbon goes through
the wake region and is oxidized and/or heated. The NO re-
maining with the carbon that escapes from the flame may
thus be only a small portion of the NO which has reacted
with carbon inside the flame envelope. Clearly the entire
area of carbon NO interaction within the flame and its
resulting products deserves considerably more study.
49
-------
REFERENCES FOR SECTION V
1. Hart, R., M. Nasralla, and A. Williams. The Formation
of Oxides of Nitrogen in the Combustion of Droplets and
Sprays of Some Liquid Fuels. Combustion Science and
Technology. 11:57-65, July 1975.
50
-------
SECTION VI
THE DIVIDED CHAMBER ENGINE EVALUATION AND
TEXACO ENGINE PROJECTS
INTRODUCTION
Recent years have seen the expenditure of large quanti
ties of research time and money in the quest for alternative
internal combustion engine configurations. The primary
motivation for these efforts has been the ever increasing
pressure of federal legislation for cleaner and more fuel
efficient personal transportation. Proposed standards
for exhaust emissions and fuel economy were straining the
capability of conventional homogeneous charge and diesel
engines, prompting the search for hybrid combustion systems.
The original purpose of the research carried out under
the current grant (R-803858-01-1) was to complete the
analysis of one such hybrid, the divided combustion chamber
engine, and to begin a study of the mechanism of unburnt
hydrocarbon formation peculiar to the divided chamber com-
bustion system. •
Prior research on the divided chamber concept, carried
out at the University of Wisconsin, demonstrated its ability
to reduce emissions of the oxides of nitrogen (NO ),and
J\.
carbon monoxide (CO) to levels below those of conventional
engines. However, its emission of high concentrations of
unburnt hydrocarbons (HC), and its high specific fuel con-
sumption demanded further attention.
All of this prior divided chamber work had been run
under naturally aspirated unthrottled intake conditions
and consequently over a limited load range. In order to
gain a more complete picture of the engines overall flexi
bility, as well as its future potential, part load testing
was initiated. Three methods for achieving a broader load
51
-------
range were tried: fuel stratification within the prechamber,
reduction of prechamber volume and intake air throttling.
As will be explained in the report, the first two approaches
proved unsatisfactory while intake throttling, in conjunction
with fuel injection and ignition parameters, enabled a rea-
sonably broad range of operation.
Emissions and fuel consumption data from these tests,
(to be presented later in this report) along with the un-
throttled test results, were compared with available data
for a number of other promising hybrid engines as well as
data from conventional homogeneous charge spark ignition
and diesel engines. This comparison showed that while emis-
sions from the divided chamber engine are comparable to
those from other hybrid combustion systems such as the Ford
PROCO and the Texaco TCCS, its specific fuel consumption is
20 to 50 percent greater throughout its useable load range.
The comparison also brought out the fact that the upper out-
put limit (IMEP) available from the divided chamber engine
in its naturally aspirated form was substantially lower than
conventional homogeneous charge and diesel engines and most
of the hybrid engines as well.
Further tests (presented in a later section of this
report) using a modified ignition system showed the fuel
consumption picture somewhat improved at light to medium
loads but still some 15 to 30 percent higher throughout the
load range than the other engines studied in the comparison.
In addition to these tests, a number of unthrottled runs
were conducted using fuel blends having octane numbers in
the range of currently available motor fuels. These tests
indicated a degree of octane sensitivity in the divided
chamber engine, and examination of combustion pressure
traces showed what appeared to be incipient knock. This
condition seemed to preclude turbocharging as a means for
extending the output limit of the divided chamber engine.
52
-------
The results of all of these tests, as well as the com-
parison study, were presented to the EPA grant monitor at
the grant review meeting. The merits of further study of
the divided chamber engine were weighed and it was decided
that because of its poor fuel consumption characteristics
and its limited load range, the divided chamber engine held
limited prospect as an alternative vehicle powerplant.
Accordingly, it was concluded that there would be minimal
interest in a study of the mechanism of unburnt hydrocarbon
formation peculiar to the divided chamber engine and that
the intent of the grant would be better served by a study
of the origins of unburnt hydrocarbons in a stratified
charge combustion system of recognized potential.
It was suggested that the Texaco TCCS stratified charge
engine be given serious consideration as a candidate for
this study. Vehicle applications of the TCCS had clearly
demonstrated its excellent fuel economy and low emissions
qualities as well as its multifuel capability.
An extensive search of heterogeneous combustion litera-
ture was conducted to determine the extent of knowledge of
hydrocarbon formation in such systems. Literature dealing
with sampling techniques which might lend themsleves to such
a study were also reviewed.
Because of the rapid spatial and temporal variation of
burning stoichiometry occuring in heterogeneous combustion
systems, it was felt that the most informative technique by
which some understanding of unburnt hydrocarbon formation
occuring in such systems might be obtained would be a high
resolution, in-cylinder sampling device which was not limited
to use solely in the clearance volume of the engine. In
order to obtain the degree of resolution desired and avoid
the possible masking effect of quench layer formation at
the sampling orifice, it was felt that such a device should
utilize the continuous flow concept first developed by Rhee.
It was agreed that the development of such a device could
53
-------
proceed in conjunction with and as a means to enable the
study of hydrocarbon formation in the Texaco combustion
process.
The next step taken was to contact the U.S. Army TACOM
and seek their assistance in determining the feasibility of
using a TCCS engine for the hydrocarbon study. They pro-
vided detailed drawings of two versions of the TCCS engine
which were then studied with a view to accessibility for
installation of the proposed sampling device. It was found
that the original single cylinder M-151-TCP engine, designed
by Texaco for TACOM and based on the L-141 cylinder block,
held the most promise for our purposes.
TACOM was contacted again and queried about the pos-
sibility of obtaining one of these engines. After a series
of negotiations, a single cylinder M-151-TCP engine was
transferred from TACOM to the University of Wisconsin.
The engine has since been completely overhauled. An
engine support stand has been fabricated and the engine has
been installed into an engine test cell and coupled to an
electric dyanmometer. The installation and instrumentation
of the engines' peripherial systems is nearing completion
as of this writing and initial run in of the engine will
begin shortly.
EXPERIMENTAL APPARATUS FOR DIVIDED-COMBUSTION CHAMBER ENGINE
STUDIES
The design and construction of the experimental set up
used for the divided chamber engine tests has been described
2
in detail by El-Messiri . However, since a number of hard-
ware changes were subsequently incorporated, a brief des-
cription will be given here. The engine used for these
tests was a Waukesha CFR-48 cetane engine. The secondary
"quench" volume for the divided chamber design was created
by inserting a surface ground spacer plate of appropriate
thickness between the cylinder block and head. The thick-
ness of this plate was calculated after selecting the
54
-------
desired compression ratio and the ratio of primary to total
combustion chamber volume. For all of the part load testing
these values were set at 8:1 and 0.65:1 respectively.
These values were selected as a judicious compromise between
desired emissions levels versus fuel economy and power out-
put. They do not reflect a truly optimized configuration.
The orifice connecting the quench region with the standard
cetane combustion chamber, hereafter designated the primary
chamber, was enlarged by El-Messiri to an orifice to piston
diameter ratio of 0.10- El-Messiri's tests were run with
a Bosch model ADN-30-S3 Pintle type nozzle mounted in the
standard cetane engine location. A standard CFR ignition
system was used in conjunction with an extended tip 14 mm
spark plug mounted in a sleeve designed to replace the
variable compression plug assembly.
The following hardware changes were made prior to the
final part load test program. The Bosch injection nozzle
was replaced with a proprietary "soft spray" vibrating
pintle nozzle having a spray cone angle of 80° and an in-
jection pressure of ^ 1700 kPa. The nozzle was moved from
its original location to the opposite end of the primary
chamber in place of the spark plug. The ignition system
was replaced with a high energy capacitive discharge unit
in order to accomodate the use of larger spark plug gaps
(^ 1 mm) and provide a more reliable source of ignition.
An adaptor was built to permit installation of a 10 mm motor-
cycle spark plug in the original injection nozzle location.
The once through cooling system used by El-Messiri was re-
placed with the standard CFR evaporative cooling tower so
that a more stable coolant temperature could be maintained.
All results reported here were obtained with the en-
gine configuration just described. Table 6 summarizes the
engine operating conditions maintained during the tests
except as noted, all tests were run with iso octane as the
engine fuel.
55
-------
Table 6. DIVIDED CHAMBER ENGINE - OPERATING
CONDITIONS TESTED
Compression Ratio C.R. = 8:1
Primary to Clearance Volume Ratio 3 = 0.65
Orifice to Piston Diameter Ratio a = 0.10
Speed 900 RPM
Ignition Timing MET
Injection Timing 105° § 70° BTDC
Overall Fuel-Air Equivalence
Ratio 0.5 - 1.0
Degree of Air Throttling 0 - 601
56
-------
TEST RESULTS FROM PRELIMINARY PART LOAD CONFIGURATIONS
Three methods of achieving part load operation were ex-
amined during the exploratory phase of the testing program.
Charge stratification within the primary chamber was be-
lieved to be a possible approach because earlier undocumen-
ted testing by El-Messiri had implied that this phenomena
was occuring. The engine configuration selected for the
part load evaluation was essentially the same as that used
by El-Messiri at the time he noted apparent stratification,
so it seemed expedient to examine this approach first. The
Bosch Pintle injection nozzle was retained in its original
location and the spark gap was positioned in the location
specified by El-Messiri. The engine was run unthrottled
over a range of speeds (800-1800 RPM) and equivalence
ratios but the resulting data were not indicative of the
type of stratification sought. The experiences of Ford
and Texaco with stratified charge engines have shown that
stratification is extremely difficult to maintain over the
range of operating loads imposed on a vehicle powerplant.
Accordingly, it was decided to abandon this approach and
explore more conventional methods of part load operation.
The second technique tried was throttling of the in-
take air. The procedure was to set the throttle for the
desired degree of air flow restriction and to vary the in-
jection pump rack setting to obtain a spread of fuel-air
equivalence ratios. Again the results proved somewhat dis-
appointing. As the degree of throttling was increased, mis-
firing occured with greater frequency causing extremely
high concentrations of unburnt hydrocarbons in the exhaust.
Examination of the spark plug insulator and the surrounding
combustion chamber surfaces showed a large build up of
soot. What appeared to be happening was that the reduced
air density in the primary chamber due to throttling was
allowing increased penetration of the injection spray with
subsequent wetting of the chamber walls and the spark plug.
57
-------
Hardware changes were indicated, but it was decided to pro-
ceed with the third approach before altering any equipment.
The third method used to achieve part load operation
was to reduce the volume of the primary combustion chamber
while maintaining the secondary "quench" volume constant.
The net effect was to increase the compression ratio while
reducing the relative proportion of combustible mixture.
This was accomplished by increasing the depth of the spark
plug mounting sleeve. It was hoped that the increase in
compression ratio would alleviate some of the fuel spray
penetration problem, but the test results indicated that
this effect was more than offset by the shorter distance
between the spark plug and the injection nozzle. The mis-
fire problem was more severe than it had been with air
throttling resulting in loss of power, increased specific
fuel consumption and high hydrocarbon emissions. It was
decided to discard this technique altogether since it was,
from the viewpoint of practicality, the least feasible of
the three methods and to concentrate on the throttling
approach to part load operation.
The hardware changes made in an effort to cure the
problems encountered during the throttling phase of the
initial testing program were detailed in the section on
experimental apparatus. However, in the interest of con-
tinuity, they will be reviewed again briefly. In order to
reduce fuel spray penetration related problems, a nozzle
having both a lower operating pressure and a broader spray
cone angle was needed. The Ford Motor Company was kind
enough to loan us a proprietary vibrating pintle nozzle
having both a low operating pressure and good quality of
atomization as well as a broad spray cone angle. In order
to improve fuel spray distribution, this nozzle was mounted
on the primary chamber centerline axis at the end of the
chamber opposite the original injector nozzle. The spark
plug was mounted in the original injector location, direct-
ly over the connecting orifice, and the ignition system was
58
-------
replaced with a high-energy, long-duration capacitive dis-
charge unit. Figure 18 shows the general hardware layout
used for the remainder of the part load testing. Initial
tests after these modifications indicated improved con-
sistency of combustion and generally smoother operation.
The final test procedure was worked out during the
initial testing sequence on the engine. It was decided to
run tests at five different throttle settings and two
different injection timings. Table 6 lists the operating
conditions selected. The procedure was to start the en-
gine and allow it to run unthrottled at the desired test
speed until conditions stabilized, i.e., coolant and lubri
eating oil temperature and ambient air temperature. At
this point, the atmospheric pressure and wet and dry bulb
temperature in the test cell were measured and recorded.
The rate of air consumption of the engine was measured
using a Meriam laminar flow element and the throttling
valve was then set to give the desired degree of throttling.
For example, if the test was to be run with the engine 30
percent throttled, the throttling valve was set to achieve
a rate of air consumption equal to 70 percent of the un-
throttled rate at the same speed. This done, the throttle
was locked in place and the injection pump rack was ad-
justed to determine the leanest operating point. The igni-
tion timing was adjusted to obtain the best torque at this
rack setting and the engine was allowed to stabilize at
this condition.
Fuel consumption was measured using an electro-optical
gravimetric weighing system. Engine RPM and load were re-
corded as were emissions data. Exhaust samples were taken
downstream of an exhaust mixing tank having a volume of
approximately ten times the cylinder displacement. The
sample was first run through a particulate filter, thence
through an ice bath and finally through a dessicant tube
before entering the emissions analysis equipment.
59
-------
EXHAUST
INTAKE
SFfcRK PLUG
ADAPTOR
PRESSURE
TRANSDUCER
PROCO
FUEL
INJECTOR
ASSEMBLY
Figure 18. Divided Chamber Engine Cylinder Head flop
View) Final Part Load Hardware Configuration
60
-------
Hydrocarbon concentration was measured with a Beckman
Model 109A unheated flame ionization detector (FID), carbon
monoxide was measured using a Beckman model IR-15A non
dispersive infrared detector with a 0 to 3 percent range and
a Thermo-Electron Model 10A chemiluminescent analyzer was
used to measure NO and NO .
x
Once the data points were recorded, the injection pump
rack was reset to give a slightly richer fuel-air ratio and
the ignition-timing was adjusted to give maximum torque.
After stabilization, a new set of data were recorded. This
procedure was repeated to yield a spread of fuel-air equiva-
lence ratios from 0.5 to 1.0.
Each set of data were processed by means of a computer
program to obtain specific fuel consumptions, mean effective
pressures, specific emissions rates and other quantities of
interest for purposes of plotting and final analysis.
PRESENTATION AND DISCUSSION OF TESTS OF FINAL PART LOAD
CONFIGURATION
Figures 19 through 28 are plots of recorded emissions
concentrations of NO, CO and HC and calculated values of
specific fuel consumption (ISFC) and mean effective pres-
sure (IMEP) versus overall fuel-air equivalence ratio cf>.
Figures 19 and 20 show the variation of NO with both <}>
and throttle for the 105° and 70° BTDC start of injection
respectively. Recall that the region of the combustion cham-
ber to which fuel is initially introduced constitutes only
the fraction 3 (0.65 for these tests) of the total combustion
chamber volume. If it can be assumed that the fuel-air
equivalence ratio in the primary chamber can be reasonably
approximated by 4> primary = 4> overall/6, then an interesting
point about these graphs can be noted. (It is recognized
that this assumption is not strictly correct. In all prob-
ability, a larger fraction of residuals remains in the pre-
chamber than in the main chamber. Hence, the resulting fuel
air mixture in the prechamber will be slightly richer than
that predicted by the above assumption. However, it is
61
-------
1000
800
INJECTION (5) 105° BTDC
D WOT
O 15% THROTTLED
O 30% «
A 45% "
O 60% »
600
E
Q_
Q.
I
O
400
200
Measured NO Concentration Versus Overall
Equivalence Ratio - 105° BTDC Injection.
62
-------
1000
800
600
E
Q.
O.
I
O
400
200
INJECTION (5)70° BTDC
D WOT
O 15% THROTTLED
O 30%
A 45%
O60%
.4 .5
Figure 20.
.9
1.0
OVERALL
Measured NO Concentration Versus Overall
Equivalence Ratio 70° BTDC Injection.
63
-------
believed that this shift will be quite minor and will not
significantly alter any of the results presented here.)
Unlike homogeneous charge engines, the peak in NO concen-
tration for the divided chamber occurs slightly on the rich
side of stoichiometric. This seems to indicate that NO for-
mation is occuring in the secondary chamber. A reasonable
explanation for this behavior can be offered. As the hot
products leave the primary chamber and begin to mix with
the secondary air a mixing interface is formed. It seems
reasonable to expect that NO formation reactions will occur
at this interface, at least early in the expansion. If we
were able to correctly calculate the portion of air in the
secondary chamber which takes part in these NO reactions, we
expect that a plot of NO concentration versus fuel-air
equivalence ratio, calculated for the primary chamber plus
this additional air, would show that the peak NO concentra-
tion, would occur on the lean side as it does in the homo-
geneous engine.
We note too that the concentration of NO increased with
the later injection timing. This rise is due to an increase
in power output which resulted from the later timing. It
is believed that less of the injected fuel was sprayed on
the chamber walls at the later timing due to the higher air
density. Figures 23 and 24 show that hydrocarbon emissions
decreased as a result of the retarded timing, which seems
to bear out this theory. Therefore, more of the fuel burned
during the early part of the combustion process with the
later timing producing higher pressures and temperatures
which resulted in increased NO formation.
Figures 21 and 22 display measured percent CO versus
and degree of throttling for the two injection timings.
We note that the minimum percent CO occurs at roughly the
same equivalence ratio for both timings and that the shape
and position of the curves are approximately the same also.
This seems to agree with the general consensus that CO is
a function primarily of fuel-air ratio alone.
64
-------
2.5
2.0
INJECTION (5) 105° BTDC
D WOT
O 15% THROTTLED
O 30%
A 45%
O60%
1.5
55
O
O
1.0
.5
.5
.6
.8 .9
COVERALL
1.0
I.I
1.2
Figure 21.
Measured CO Concentration Versus Overall
Equivalence Ratio 105° B.TDC Injection.
65
-------
INJECTION© 70
D WOT
O 15% THROTTLED
O 30%
A 45%
O 60%
Measured CO Concentration Versus Overall
Equivalence Ratio - 70° BTDC Injection.
66
-------
If wo again assume that the equivalence ratio in the
primary chamber can be approximated by the overall ratio
divided by the volume fraction $, we can note another mea-
sure of the reactions occurring in the secondary chamber.
At an overall equivalence ratio of 0.8, the ratio in the
primary chamber would be 1.23. Reference to Obert3 shows
that for iso octane there should be about 5.5 percent CO
in the products entering the secondary chamber. This is
more than ten times the value measured in the exhaust from
the divided chamber engine. Clearly, substantial oxidation
is occurring in the secondary chamber as intended.
One last point might be mentioned. As the fuel-air
ratio is leaned beyond the minimum CO point, the percent CO
begins to rise. It is thought that this is due to reduction
in the destruction of CO in the quench region caused by lower
overall gas temperatures resulting from combustion. This
occurs because the already slow burning rate of the lean mix-
tures is further slowed by dilution of the combustible mix-
ture with the residual fraction of exhaust products remaining
in the primary chamber from cycle to cycle.
Figures 23 and 24 present measured HC concentration
versus overall equivalence ratio as a function of throttling.
We can see that at all throttle settings, HC emissions in-
crease rapidly as is decreased from 0.7 to 0.5. It is
thought that this increase in HC is due primarily to the
decreasing average gas temperature resulting from combustion
and the attendant slowing of hydrocarbon oxidation reaction
rates. Note also that the equivalence ratio at which this
rapid increase in HC begins increases with decreasing intake
manifold pressure. This occurs because the exhaust residual
fraction increases with decreasing intake pressure, lowering
the flame temperature and slowing down the oxidation re-
actions. As the overall fuel-air ratio is increased and
approaches stoichiometric, HC emissions again begin to rise.
It is believed that this is due to increased deposition of
fuel on the surface of the primary chamber and resultant wall
67
-------
10
8
INJECTION @ 105° BTDC
D WOT
O I5%THROTTLED
O 30% »
A 45% "
60% »
to 6
O
x
E
Q.
a.
I
O
X
.5
.6
,
C
-9
OVERALL
1.0
I.I
1.2
Figure 23.
Measured HC Concentration Versus Overall
Equivalence Ratio 105° BTDC Injection.
68
-------
10
8
INJECTION (3)70° BTDC
D WOT
O 15% THROTTLED
O 30% «
A 45% "
O 60% »
ro
06
x
E
Q.
Q.
I
O
I
.4
.5
.6
.7
Figure 24.
<#>
.8
OVERALL
.9
1.0
I.I
Measured HC Concentration Versus Overall
Equivalence Ratio 70° BTDC Injection.
69
-------
quenching phenomena. Lastly, note that retarding the injec-
tion timing resulted in a decrease in HC emissions through-
out the operating range. This is probably a result of the
generally improved combustion regularity noted at this set-
ting, as evidenced by the increase in mean effective pres-
sure at all equivalence ratios as shown in Figure 28.
Figures 25 and 26 are plots of indicated specific fuel
consumption (ISFC) versus for the two injection timings.
We note that ISFC is extremely sensitive to overall fuel-air
ratio, much more so than conventional homogeneous charge en-
gines. As is increased from 0.5 to the value corresponding
to the minimum ISFC at any given throttle setting, ISFC de-
creases almost linearly. Figures 27 and 28 show us that the
reason for this is that the power output of the engine in-
creases almost linearly over this same range of equivalence
ratios. As
is increased. It is believed that increasing quantities of
fuel end up on the primary chamber walls, vaporizing slowly'
and taking part in wall quenching reactions which lead to in-
creased hydrocarbon emissions rather than useful power output,
Lastly, Figures 27 and 28 present calculated values of
indicated mean effective pressure (IMEP) versus $ for the
various throttle settings. Note the increase in power at
the leaner fuel-air ratios which occurred when the fuel in-
jection timing was retarded from 105° to 70° BTDC. The in-
creased air density in the chamber at the later timing re-
duces the fuel spray penetration, cutting down on the amount
of fuel sprayed directly onto chamber surfaces. Such fuel
vaporizes at a rate determined by the chamber surface tem-
perature and takes part in wall quenching phenomena leading
to increased hydrocarbon emissions, but making little con-
tribution to power output.
An injection timing of 50° BTDC was tried, but com-
bustion became erratic and hydrocarbon emissions rose
70
-------
.5
o>
I
O
LL
CO
/,
.4-
INJECTION© 105° BTDC
D WOT
O 15% THROTTLED
D
I
.4 .5
Figure 25.
.6
.9
1.0
I.I
>
OVERALL
Calculated Specific Fuel Consumption (ISFC
versus Overall Equivalence Ratio - 105°
BTDC Injection.
71
-------
.5
i
O
LL.
to
INJECTION © 70° BTDC
D WOT
O 15% THROTTLED
O 30% n
A 45% «
O 60%
.4 .5
Figure 26.
.7,
'OVERALL
.9
1.0
I.I
Calculated Specific Fuel Consumption (ISFC)
Versus Overall Equivalence Ratio - 70° BTDC
Inj ect ion.
72
-------
600
500
o
Q.
XL
I
Q_
LJ
400
300
200
100
D
D
D
INJECTION (a) 105° BTDC
D WOT
O 15% THROTTLED
O 30%
A 45%
O 60%
A -5
Figure 27.
.6
COVERALL
.9
1.0
Calculated Mean Effective Pressure (IMEP)
Versus Overall Equivalence Ratio 105°
BTDC Injection.
73
-------
600
500
400
o
Q.
J£
I
Q_
LJ
»-<300
200
100
INJECTION © 70° BTDC
D WOT
O 15% THROTTLED
O 30% «
A 45% "
O 60% v
.5
.6
-8
OVERALL
.9
1.0
I.I
Figure 28.
Calculated Mean Effective Pressure (IMEP) Versus
Overall Equivalence Ratio - 70° BTDC Injection.
74
-------
sharply so no data were recorded at this setting. Apparent-
ly, this injection timing did not allow sufficient time for
adequate fuel vaporization and mixing with the air to occur
before ignition.
Figures 29 through 36 are cross plots of calculated
specific emissions and fuel consumption versus IMEP, at
constant fuel-air equivalence ratios, for the two injection
timings. The data were plotted in this fashion to display
the load flexibility of the divided chamber engine under a
variety of operating regimes. It was thought that such
plots might identify a set of operating parameters by which
emissions could be minimized while maximizing fuel economy
and load flexibility. However, study of the plots did not
indicate any such clear cut choice and in fact presented
the same type of trade offs peculiar to the homogeneous
charge engine.
Figures 29 and 30 present indicated specific nitric
oxide (ISNO) versus IMEP. If we wished to minimize ISNO
while maintaining the broadest possible load range, we
might reasonably choose to run the divided chamber engine
at a constant fuel-air equivalence ratio of = l with in-
jection starting at 70° BTDC. Examination of the corres-
ponding plots for ISCO, ISHC and ISFC (Figures 32, 34 and
36) shows that this set of operating conditions would
maximize specific emissions of CO and HC while yielding
the worst possible fuel economy.
Figures 31 through 36 show that if we wished to mini
mize specific CO, HC and fuel consumption, we could do so
by running the engine at an equivalence ratio between 0.7
and 0.8 with injection starting at 70°BTDC. However, the
penalty of operating this way would be a narrowed operating
load range and near maximization of specific NO emissions.
A casual comparison with available data showed that
the divided chamber engine was capable of relatively low
exhaust emissions levels of NO , CO and HC. Further, no
75
-------
D>
I
CO
H 2
INJECTION START (S> 105° BTDC
100
200
300 400
IMEP-kPa
500
600
Figure 29. Calculated Specific NO Versus IMEP at Constant
• Equivalence Ratios - 105° -BTDC Injection.
-------
i
O
Z
co 2
0
INJECTION START Co) 70° BTDC
100
Figure 30.
200
300 400
IMEP-kPa
500
600
Calculated Specific NO versus IMEP at Constant
Equivalence Ratios - 70° BTDC Injection.
-------
CO
90
80
70
-c 60
<* 50
i
O
o
£40
30
20
10
INJECTION START 105° BTDC
100
200
300 400
IMEP-kPa
500
600
Figure 31. -Calculated Specific CO Versus IMEP at Constant
Equivalence Ratios - 105° BTDC Injection.
-------
INJECTION START (5) 70 BTDC
100
200
300 400
IMEP-kPa
500
600
Figure 32. .Calculated Specific CO Versus IMEP at Constant
Equivalence Ratios - 70° BTDC Injection.
-------
.9
00
o
.8
.7
I-6
INJECTION START O 105° BTDC
I
o
M
.5
.4
.3
.2
I
I
0
100
Figure 33.
200
300 400
IMEP-kPa
500
600
Calculated Specific HC Versus IMEP at Constant
Equivalence Ratios - 105° BTDC Injection. •
-------
8
INJECTION START (5)70° BTDC
oo
O
100
Figure 34
200
300 400
IMEP-kPa
500
600
Calculated Specific HC Versus IMEP at Constant
Equivalence Ratios - 70° BTDC Injection.
-------
oo
N)
.40
.35
O
UL
in
^.30
.25
INJECTION START (3) 105° BTDC
100
Figure 35
200
300 400
IMEP-kPa
500
600
Calculated ISFC Versus IMEP at Constant Equivalence
Ratios - 105° BTDC Injection.
-------
oo
.40
.35
jx:
\
o>
O
u_
CO
.30
INJECTION START (o) 70° BTDC
.25
100
Figure 36.
200
300 400
IMEP-kPa
500
600
Calculated ISFC Versus IMEP at Constant Equivalence
Ratiqs - 70° BTDC Injection.
-------
recognizable operating regime would achieve a minimum level
of NO, CO and HC simultaneously. However, this casual com-
parison suggested that regardless of the operating parameters
used, the divided chamber engine as it was presently con-
stituted was not capable of power output or fuel economy
comparable to a conventional homogeneous charge engine.
One possible approach toward improving these two as-
pects of the divided chamber engines' performance seemed
to be turbocharging. However, several factors encountered
during the part load testing prompted us to delay explora-
tion of the effects of simulated turbocharging in favor of
some additional tests with the existing experimental set up.
Examination of a typical primary chamber combustion
pressure trace, as shown in Figure 37, shows that combus-
tion in the divided chamber engine occurs somewhat later
in the cycle than in a conventional homogeneous charge en-
gine. It became apparent during the part load testing that
as the degree of throttling increased, spark had to be fur-
ther and further retarded to secure ignition. The reason
for this behavior was simply that combustible mixture was
not entering the region of the connecting orifice and spark
plug until later and later in the expansion process. It
was felt that this late combustion was at least partially
responsible for the low power and poor fuel economy.
In order to test this theory, a second spark plug
was installed in the region of the prechamber furthest
from the orifice. It was thought that by igniting the
mixture in the region of this plug at some point before
TDC, the resulting expansion would force combustible mix-
ture into the region of the orifice earlier in the cycle.
By using a long duration O 2 msec) ignition system to
fire the spark plug at the orifice; we could fire both
plugs simultaneously and ignite the mixture as it
approached the orifice so that no unburned fuel would be
blown into the quench region. Tests were run under
84
-------
TDC -~
15° CA
900 RPM, UNTHROTTLED, MBT SPARK
INJECTION & 70° BTDC
Figure 37.
Typical Primary Chamber Combustion
Pressure Trace.
-------
unthrottled conditions using both single ignition at the ori-
fice and dual ignition. The results of these tests are pre-
sented in Figure 38. It can be seen that specific fuel con-
sumption at light load was improved some 20% by this dual
ignition scheme, but the improvement decreases as load is
increased. HC emissions -are lower throughout the load
range using the dual ignition, but NO and CO are increased
J^.
at light to medium loads. No improvement in either emissions
or fuel consumption was realized at peak load with this
modified ignition scheme and in fact, the peak load output
was unchanged from its previous level.
As a preliminary to experimenting with simulated tur-
bocharging, it was decided to run some tests to evaluate
the octane sensitivity of the divided chamber engine.
Blends of iso octane and N-Heptane, having RON's in the
range of currently available motor fuels, were run in the
engine under unthrottled conditions. Pressure traces re-
corded while running on these fuels showed an increasing
development of the pressure fluctuations associated with
knock as the octane rating of the fuel was reduced. Audible
knock was not observed but may have been masked by the high
ambient noise level associated with the operation of the
divided chamber engine.
Several inferences were drawn from the results of
these additional tests. First, although the specific fuel
consumption at light loads was improved by the dual igni-
tion modification, it appeared that fuel consumption was
still substantially greater than current conventional homo-
geneous charge engines operating in the same load range.
Second, the apparent onset of knock under naturally aspi-
rated running conditions seemed to rule out turbocharging
as a means to improve the divided chamber engines'
efficiency.
Taken together, these two points raised serious doubts
as to the future potential of the divided chamber engine as
-------
ro
01
ISFC-kg/kWh
OJ
o
04
O1
ISNOx-g/kWh
o
o
o
o
00
m
T)
T)
Q
O
o
\
\
\
\
ISHC-g/kWh
ISCO-g/kWh
ro
O
Figure 38. Calculated Specific NOX, CO, HC and Fuel Consumption
Versus IMEP For Dual Ignition Tests.
-------
an alternative vehicle powerplant. It was decided to post-
pone the proposed study of hydrocarbon formation in the
divided chamber engine and instead, to gather together com-
parable emissions, power and fuel consumption data from
conventional homogeneous charge, diesel and other hybrid
engines for the purpose of making a comparison with the
divided chamber engine.
Data from six other engines encompassing four different
combustion systems were examined for purposes of comparison.
To illustrate the conventional throttle controlled approach,
the results of Wimmer and Lee for their homogeneous charge,
spark ignited CFR engine are used throughout the graphical
comparison presented in Figures 39 through 42. The data
on this engine were gathered in conjunction with data from
a second CFR equipped with a prechamber and set up to run
as a torch ignited homogeneous lean engine- The prechamber
was designed to be installed in the spark plug opening and
was equipped with an intake valve and spark plug. The
volume of this prechamber was approximately 10% of the
total clearance volume, and it received premixed fuel and
air at an equivalence ratio of 5. Lean mixture was sup-
plied to the cylinder during the intake stroke and subse-
quent dilution of the prechamber charge during the compres-
sion stroke produced a readily ignitable mixture. Upon
ignition, a torch of flame issues from the prechamber in-
flaming the lean main chamber mixture. This engine was
similar in many ways to the Honda CVCC, so where applicable,
data from Date, et al ' were also used to illustrate the
state of the art for torch ignited engines.
Two hybrid stratified charge spark ignited engines
representing two techniques of combustion control were also
included in the comparison. The Ford PROCO, which uses
early injection of fuel and controls combustion by flame
speed and mixture quantity, was represented by data ex-
cerpted from the paper by Simko, et al , while data for
88
-------
the Texaco TCCS engine, which achieves combustion control
through control of fuel injection rate, similar to an open
chamber diesel, was taken from the paper by Mitchell, et al
Lastly, the modern open chamber diesel engine was felt
to be relevant to the comparison, so data for the high out-
put Labeco-TACOM single cylinder were taken from the ex-
n
tensive results published by Bolt, et al .
The test conditions for the various engine data used
in this comparison are not identical to the WOT conditions
maintained for the divided chamber engine tests. However,
it was felt that the data were taken under similar enough
conditions that a valid comparison could be made.
The various data are presented in the form of plots
of indicated specific emissions and fuel consumption versus
indicated mean effective pressure. Indicated values were
used because the majority of the data were taken from
single cylinder engine tests.
Figure 39 displays ISNO versus IMEP for all of the
X
engines included in the comparison. Note that the worst
case seems to be the conventional homogeneous charge en-
gine. It is believed that if data could have been found
for a modern engine utilizing spark retard and EGR, this
type of engine would have looked much better from an emis-
sions standpoint. The divided chamber engine looks better
than all of the others from the standpoint of NO . How-
X
ever, note that this is data for
-------
35
30
25
20
X
O
CO
10
0
0
n DIVIDED CHAMBER-
900 RPM ,<*> = !
? TACOM O.C. DIESEL-
2000 RPM,5lkPa BOOST
x PRECHAMBER TORCH IGN
CFR -1000 RPM
O CVCC-80km/h, =.755,
ISFC = .24kg/kWh
o HOMOGENEOUS CFR-
1000 RPM
^ PROCO-1500 RPM
W/15% EGR
o TCCS -1500 RPM
W/12% EGR
200
800
1000
400 600
IMEP-kPa
Figure 39. Specific NOX Versus IMEP - Divided Chamber Engine Comparisons
-------
40
30
O
O 20
10
0
n DIVIDED CHAMBER -
900 RPM,INJ.(cDI050BTDC
o HOMOGENEOUS CFR -1000 RPM
* PRECHAMBER TORCH IGN.
CFR-1000 RPM
A PROCO-1500 RPM W/15% EGR
o TCCS-1500 RPM W/12% EGR
0
200 400
IMEP-kPa
600
800
Figure 40. Specific CO Versus IMEP - Divided Chamber Engine Comparison,
-------
divided chamber engine begins to approach its' output limit,
demanding increasingly rich mixtures which result in a sharp
rise in CO emissions.
Figure 41 compares specific hydrocarbon emissions for
the various engines versus power output. Note the decrease
in hydrocarbon emissions from the divided chamber engine
which occurred when the injection timing was retarded from
105° to 70° BTDC. At the later timing, the HC emissions
from the divided chamber engine are better than the other
hybrids at light loads up to 400 kPa IMEP, but fall midway
between the bracketing engines at higher loads.
Lastly, Figure 42 presents indicated specific fuel
consumption versus IMEP. It is felt that this particular
plot is the most important one with regard to the divided
chamber engine. It shows that the specific fuel consump-
tion of the divided chamber engine is 15 to 30 percent
greater throughout its load range than the two hybrid
stratified charge engines and the diesel. In addition, it
shows clearly the limited peak output of the divided cham-
ber engine when compared to the conventional homogeneous
charge and diesel engines, as well as the other hybrids.
The results of the part load tests and the additional
tests described previously, along with this comparison,
were presented to the EPA grant officer at the grant
review meeting. The merits of continuing the study of the
divided combustion chamber concept, and specifically the
mechanism of unburnt hydrocarbon formation peculiar to it,
were discussed and it was decided that the original goals
of the project would be more profitably accomplished by
studying the origins and causes of hydrocarbon emissions
from a combustion system of more recognized potential.
It was suggested at this time that we consider the
possibility of studying the hydrocarbon formation process
in the Texaco Controlled Combustion System (TCCS) strati-
fied charge engine. Recent results from the TCCS engine
92
-------
8
I
O
0
n DIVIDED CHAMBER-900 RPM,INJ.(a)l05°(TOP) a70°BTDC
? TACOM O.C.DIESEL-2000 RPM,5lkPa BOOST
x PRECHAMBER TORCH IGNITION CFR -1000 RPM
o HOMOGENEOUS CFR-IOOO RPM
PROCO-1500 RPM W/I5%EGR
o TCCS-1500 RPM W/12% EGR
0
200
400 600
IMEP-kPa
800
Figure 41. Specific HC Versus IMEP - Divided Chamber Engine Comparison
-------
o
Li.
CO
.6
.5
.4
.3
.2
.1
0
0
n DIVIDED CHAMBER-900 RPM,INJ.(o)l050 a 70° BTDC
o HOMOGENEOUS CFR-IOOO RPM
x PRECHAMBER TORCH IGNITION CFR-IOOO RPM
A PROCO-1500 RPM W/15% EGR
o TCCS-1500 RPM W/12% EGR
* TACOM O.C. DIESEL-2000 RPM ,51 kPa BOOST
200
400 600
IMEP-kPa
800
1000
Figure 42. Specific Fuel Consumption Versus IMEP - Divided Chamber Engine
Comparison.
-------
development program funded by TACOM had demonstrated ex-
cellent fuel economy and low NO emissions levels in vehicle
X.
applications. However, these installations had utilized
catalytic converters to reduce HC emissions and it was felt
that the hydrocarbon problem of the TCCS was severe enough
to warrent a detailed study.
Before making a final decision regarding the TCCS, an
extensive search of the literature pertinent to the hetero-
geneous combustion process occurring in the TCCS was con-
ducted. Possible mechanisms of hydrocarbon formation were
studied and a reasonable scheme of the mechanisms peculiar
to the TCCS was proposed. In addition, literature dealing
with in-cylinder gas sampling techniques was reviewed in an
effort to determine the possible role of such techniques in
studying and verifying the proposed mechanism of hydrocarbon
formation in the TCCS.
The complete literature search appears in the appendix
of this report. The following conclusions were drawn from
the survey.
1. Literature dealing with hydrocarbon formation
in heterogeneous combustion systems is sparse,
and the mechanisms of formation involved are
not yet well understood.
2. Based on available knowledge, we can postulate
a reasonable hypothesis of the origins and
causes of hydrocarbon formation in the Texaco
combustion system.
3. A continuous flow, in-cylinder sampling technique
would be the most straightforward way to investi-
gate hydrocarbon formation in heterogeneous com-
bustion.
4. Such a technique could be developed in conjunction
with, and as a means to, the study of hydrocarbon
formation in the Texaco combustion process.
95
-------
The next step taken was to determine the feasibility
of using an in-cylinder sampling technique in the TCCS
engine. Two designs of the TCCS were subsequently ex-
amined for accessibility for such a device and it was
found that the single cylinder M-151 TCP engine was the
most suitable for this purpose. A request was made to
TACOM for one of these engines and, after a prolonged
series of negotiations, a single cylinder M-151 TCP engine
was officially transferred from TACOM to the University of
Wisconsin.
The next section of this report details the develop-
ment of the engine test apparatus and the in-cylinder
sampling device.
EXPERIMENTAL APPARATUS FOR TEXACO ENGINE STUDIES
The M-151 TCP engine obtained from TACOM for use in
the proposed hydrocarbon study had some 2000 hours of pre-
vious test time on it. Upon disassembly, it was found that
the engine was in need of a complete overhaul. The main
and rod bearing journals of the crankshaft were badly
scored and the bearing support surfaces of the connecting
rods were nicked and generally in poor condition. In
addition, the valve tappets of the working cylinder had
apparently ceased to rotate at some point in the engines'
prior life. Their faces were severely galled as were the
matching cam lobes.
A new crankshaft and two new connecting rods were ob-
tained along with a complete set of new bearings and the
engine was reassembled. A new camshaft and tappets pro-
vided by Texaco were also installed.
An engine support cradle was fabricated for the engine
and moved into one of the engine test cells along with a
General Electric 30 kW dynamometer. The dynamometer has
since been tied into the Ward-Leonard motor-generator system.
96
-------
The engine was next installed in the cradle and, after
alignment, was coupled to the dynamometer using a Falk
model 8F tapered grid coupling.
A pair of 15 gallon (.057 m ) vertically mounted air
receiver tanks were ordered from Kargard Industries, Inc.,
in Marinette, Wisconsin for use as intake and exhaust surge
tanks. This volume is approximately 100 times the engines'
displacement. The tanks were installed in the engine test
cell on the port side. The exhaust tank has been coupled
to the cylinder head exhaust port by means of a 2.5 in
(6.35 cm) diameter flexible stainless steel conduit and
a stainless mating flange. A bronze valve has been in-
stalled at the exhaust tank outlet so that the back pres-
sure can be regulated and the outlet side of this valve
has been connected to the building exhaust system with 2
in. (5 cm) diameter aluminum pipe. Chromel-alumel thermo-
couples have been installed at the cylinder head exhaust
port and at the outlet of the surge tank for monitering
purposes. The intake surge tank was connected to the
cylinder head intake port by means of an 18 in. (.46 m)
section of 1.25 in (3.2 cm) diameter pipe. Eighteen
inches was the length recommended by Texaco to give best
overall performance.
It was decided to set the air system up so that the
engine could be run with boost pressures up to 1 atmosphere
and air temperatures to 200°F (93°C). The maximum air flow
required was calculated and the heating requirement was
worked out. Two chromalox model KSEF-24 heating elements,
rated at 1950 W each, were subsequently ordered. In order
to control the cycling of these heating elements, a West
Instrument Model 802M temperature controller and a Crydom
Model A2425 solid state zero crossing relay were purchased.
The controller senses the intake air temperature at the in-
take port entrance by means of an iron-constantan thermo-
couple, switching the elements on and off as required by
97
-------
means of the high capacity relay. A housing to enclose the
two heating elements has been designed and is presently
under construction. This assembly will connect directly
to the inlet opening of the intake surge tank.
A set of three critical flow orifices were designed to
match the anticipated air flow requirements of the engine.
However, it was determined that to machine such orifices to
the recommended ASME specifications would be both expensive
and time consuming. It was discovered that synthetic
sapphire jewel bearings having suitable orifice configura-
tions could be purchased in the required sizes at a very
attractive price. Since any nozzle would require calibra-
tion before it could be used, this approach did not seem
to entail any additional work and enabled a cost savings
as well. The jewel bearings have been ordered from the
Swiss Jewel Company in Philadelphia, Pa. The related air
line network and nozzle holders have been designed and are
under construction.
The W.D. Ehrke Company, Inc., in Milwaukee was con-
tacted for help in selecting an appropriate air regulator
to control the pressure upstream of the critical flow
nozzles. A Fischer type 95 HD differential air regulator
has been ordered and delivery is expected shortly.
The cooling system of the engine was the next order
of business. A Perfex Model B-310-121-A counterflow heat
exchanger was obtained with the engine and plumbed into
the system. Flow of cooling water through the exchanger
is controlled by a Sterlco Model R-151 F reverse acting
temperature control valve. The sensor from this valve
is installed in the engine coolant expansion tank and the
valve can be adjusted to vary the coolant temperature
between 175° and 215° F (80°-100°C). A Doerr Electric
Corporation 1/3 hp (1/4 KW) centrifugal pump is used to
circulate the engine coolant through the heat exchanger.
98
-------
ID
10
Figure 43. Sampling Valve Design for TCCS
Hydrocarbon Study.
-------
No steps have been taken to modify either the piston or
the cylinder head for the accomodation of the sampling de-
vice. It was felt that a substantial data base should first
be acquired for the unmodified engine so that the effect of
future modifications could be more clearly understood.
The sampling valve design has progressed to the point
shown in Figure 43. The valve will utilize the continuous
flow concept described previously, with the majority of the
flow being wasted and the sample being snatched from the
flow at the desired sampling point. The valve will be
actuated electro-magnetically in response to a trigger
signal picked up by an optical sensor on the engines' cam-
shaft. The positioning of the sensor is adjustable so that
the timing of the triggering signal can be varied over the
desired range of sampling points.
DISCUSSION OF CONCLUSIONS
The divided chamber engine has demonstrated emission
levels of the oxides of nitrogen (NO ) considerably lower
J\,
than conventional homogeneous charge spark ignition engines.
However, other hybrid engines have displayed similar emis-
sions levels while yielding substantially better fuel
economy and power output than the divided chamber engine.
It was concluded, therefore, that the study of hydrocarbon
formation in an engine showing more promising fuel efficiency
would be more productive.
This conclusion is based on the following findings:
1. The specific fuel consumption (kg/kWh) of
the divided chamber engine is 20 to 25 per-
cent greater throughout its useful load
range than conventional homogeneous charge,
diesel and hybrid stratified charge engines
currently being considered for production use.
2. The maximum output (IMEP) of the divided
chamber engine under naturally aspirated
100
-------
operation is 15 to 30 percent lower than most
of the conventional and hybrid engines under-
going serious study.
3. Octane sensitivity of the divided chamber
design rules out turbocharging as a means
of increasing the output and improving the
fuel efficiency of the engine to an acceptable
level.
DISCUSSION OF RECOMMENDATIONS
The study of the origins of unburnt hydrocarbon emis-
sions in the Texaco combustion system, which was initiated
under the present, should be continued.
The study of combustion phenomena peculiar to hybrid
combustion systems is an important and necessary step in
the development of these engines to their fullest ecological
and economical potential.
Hydrocarbon emissions have been and continue to be a
serious problem in most hybrid combustion systems. Con-
siderable attention has been given to the problems of NO
-A.
and carbon monoxide formation in such systems, but little
has been done about the hydrocarbon problem. The increasing
emphasis on fuel economy demands that the hydrocarbon pro-
blem be attacked in the combustion chamber rather than
eliminating it catalytically as has been the recent practice
Many studies detailing the sources and causes of hydro-
carbon emissions from homogeneous charge spark ignition en-
gines are available in the literature. However, the same
cannot be said for stratified charge engines. It is be-
lieved that the understanding of hydrocarbon formation in
heterogeneous combustion to be gained through this study
would provide a useful and much needed contribution to the
body of knowledge of stratified charge combustion.
Lastly, discussions with members of the various or-
ganizations connected with the development of the Texaco
101
-------
combustion system (Texaco, TACOM, M.I.T.) have indicated a
high degree of interest in the results of such a study.
The concentration of unburnt hydrocarbons in the untreated
exhaust of the Texaco engine is recognized as one of its
aspects which needs to be improved before serious considera
tion is given to widespread application. It is thought
that the results of this hydrocarbon study might point the
way to the necessary design changes.
102
-------
REFERENCES FOR SECTION VI
1. Rhee, K. T., G. L. Borman, 0. A. Uyehara, and P. S.
Myers. A Continuous-Flow Gas Sampling System for
Engines. Combustion Science and Technology. 12:
105-109, 1976.
2. El-Messiri, I. A. The Divided Combustion Chamber
Concept and Design for Control of S.I. Engine
Exhaust Air Pollutant Emissions. University of
Wisconsin. Madison, Wisconsin. Ph.D. Thesis. EPA
Office of Air Programs. January 1973. 286 p.
3. Obert. E. F. Internal Combustion Engines, Third
Edition. Scranton, Pennsylvania, International Text-
book Company, February 1970, 736 p.
4. Wimmer, D. B., R. C. Lee. An Evaluation of the Per-
formance and Emissions of a CFR Engine Equipped With
a Prechamber. Research Center, Phillips Petroleum
Co. 730474. Society of Automotive Engineers. May
1973, 15 p.
5. Date, T., S. Yagi, A. Ishizuya, and I. Fugii. Research
and Development of the Honda CVCC Engine. Honda Re-
search and Development Co., Ltd. 740605. Society of
Automotive Engineers. August 1974. 18 p.
6. Date, T., S. Yagi, and K. Inoue. NOX Emissions and
Fuel Economy of the Honda CVCC Engine. Honda Research
and Development Co., Ltd. 741158. Society of Automotive
Engineers. October 1974. 13 p.
7. Simko, A., M. A. Choma, and L. L. Repko. Exhaust
Emission Control by the Ford Programmed Combustion
Process-PROCO. Ford Motor Company. 720052. Society
of Automotive Engineers. January 1972. 22 p.
8
Mitchell, E., M. Alperstein, J. M. Cobb, and C. M .
Faist. A Stratified Charge Multifuel Military
Engine - A Progress Report. Research and Technical
Dept., Texaco Inc. 720051. Society of Automotive
Engineers. January 1972. 9 p.
Bolt, J. A., M. F. El-Erian, J. Duerr, and A. Miklos
Diesel Engine Combustion and Emissions TACOM
Research Engine. University of Michigan. 11796.
USATACOM. April 1973. 302 p.
103
-------
SECTION VII
-PUBLICATIONS
1. Evers, L. W., P. S. Myers and 0. A. Uyehara. A Search
for a Low Nitric Oxide Engine. Presented before the
1973 Spring Meeting of the Midwest Section of the
Combustion Institute.
2. Evers, L. W., P. S. Myers, 0. A. Uyehara. A Search
for a Low Nitric Oxide Engine. SAE 741172.
3. Evers, L. W. Nitrogen Control With the Delayed
Mixing Stratified Charge Engine Concept. PhD Thesis.
University of Wisconsin. 1976.
4. Evers, L. W., P. S. Myers and 0. A. Uyehara.
Nitrogen Oxide Control With Delayed-Mixing Stratified
Charge Engine Concepts. EPA-460/3-76-022.
5. Evers, L. W., P. S. Myers and 0. A. Uyehara. An
Experimental Study of the Delayed Mixing Stratified
Charge Engine Concept. SAE 770042.
104
-------
SECTION VIII
GLOSSARY
ATDC
BTDC
CO
CVCC
HC's
IMEP
ISCO
ISFC
ISHC
ISNO
ISN°x
L-141
MET
M
FUEL
M.
NO
M-151-TCP
NO
PROCO
After top dead center
Before top dead center
Carbon monoxide
Honda Controlled Vortex Combustion Chamber
Molar emissions index
Unburnt hydrocarbons
Indicated mean effective pressure, psi
Indicated specific carbon monoxide, g/kWh
Indicated specific fuel consumption, kg/kWh
Indicated specific unburnt hydrocarbon, g/kWh
Indicated specific oxide, g/kWh
Indicated specific oxides of nitrogen, g/kWh
Designation of standard "Jeep" engine
Minimum advance for best torque
Mass burning rate of fuel, g/sec
Molar rate of NO, moles/sec
Molar rate of NO , moles/sec
.X
Designation of single cylinder "Jeep" engine
converted to Texaco Combustion Process
Molar density, moles/liter
Nitric oxide
Combined oxides of nitrogen
Parts per million
Ford Programmed Combustion Process
105
-------
TACOM
TCCS
TCP
TDC
—
3
Volumetric flow rate, liters/sec
U.S. Army Tank-Automotive Command
Texaco Controlled Combustion System
Texaco Combustion Process
Top Dead Center
Orifice to piston diameter ratio
Primary to clearance volume ratio
Fuel-air equivalence ratio
106
-------
SECTION IX
APPENDIX
LITERATURE SEARCH FOR TCCS HYDROCARBON STUDY
107
-------
In order to evaluate the nature of the hydrocarbon formation pro-
cess in the Texaco Controlled Combustion System, a study of the litera-
ture pertinent to the heterogeneous combustion process peculiar to
the TCCS has been conducted. The purpose of this study was to
identify the probable mechanisms of hydrocarbon formation operating
in the TCCS and to determine what technique for studying these
phenomena would be most applicable.
The first part of this appendix reviews general heterogeneous
combustion literature and discusses its relationship to the Texaco
combustion process. Based on this review, the second part of
the appendix explains the scheme of hydrocarbon formation thought to
be operating in the TCCS and reviews literature dealing with the
specific mechanisms believed to be involved. The final part
deals with the experimental problem of conducting the proposed
study and the possible role of available in-cylinder sampling
techniques in verifying the hypothesis.
BACKGROUND
The Texaco Controlled Combustion System (TCCS) is a hybrid,
displaying characteristics in common with both Diesel and Otto cycles.
As such, it seems logical to begin with a general description of the
TCCS, pointing out these similarities, as well as its peculiarities,
before proceeding to a detailed analysis of the combustion process.
Table 1 presents the salient features of Diesel and S.I. engine
operation. The characteristics common the the TCCS (TCP) are denoted by
an asterisk. One notes immediately that the TCCS appears to be much more
closely related to the Diesel than the S.I. engine. Its use of direct in-
jected, stratified charge combustion, with locally rich burning but over-
all lean operation and load control by variation of injected fuel quantity.
108
-------
Table A Combined Features of the Compression Ignited
and the Spark Ignited Engine, which are
Present in the TCP Engine
Compression Ignited Sna:k Ignited*
Fuel burned in excess air* Fuel/air ratio limited neai
chemically correct
Smoke limited full power* Air limited full power
Unthrottled air* Throttled ai:
Fuel injector required* Fuel injector <,-. carburetor.
optional
Rate of pressure rise, Rate of pressure rise.
100 psi/deg 40 psi/des*
•Features of TCP engine
along with smoke limited output, clearly relate it to the Diesel concept.
In addition, the TCCS employs moderate air swirl and a toroidal combustion
chamber much like that used in numerous small displacement open chamber
Diesels. However, the use of a positive ignition source in the TCCS elim-
inates the variable ignition delay period of the Diesel and its attendant
fuel quality requirements. Use of spark ignition and compression ratio
comparable to conventional homogeneous charge S.I. engines, as well as
the more moderate rate of pressure rise characteristic of the TCCS, relate
the concept to the Otto cycle as well. It should be pointed out that
since only air is compressed in the TCCS and residence time for combus-
tible mixture within the cylinder is quite short, the S.I. engine phenom-
enon of knock and its attendant fuel quality requirements are also
elimated.
109
-------
With this description of the general features of the TCCS in mind,
one can now begin to fill in the detail regarding the various combustion
events.
COMBUSTION
Figure 1 presents a plot of PVY versus crank angle for the Texaco
process. The data were taken on an early model hemispherical combus-
f3l
tion chamber Texaco engine1- J running at full load on a variety of fuels.
The plot has been normalized using the value of PVY of the compressed
air just prior to the start of injection. Injection starts at 20° BTDC
and is followed by a period of apparent inactivity, or ignition delay if
you will, of approximately 10° (-1.3 msec) before appreciable pressure
rise begins. This delay has also been observed by Lazarewicz'- -" on a
more recent TCCS engine configuration using a toroidal bowl combustion
chamber, implying that the delay period is peculiar to the process and
not the specific engine. Between 10° BTDC and 10° ATDC, the product PVY
rises linearly. Since volume is changing little during this period, and
V/V ~ 1, this rise is due primarily to pressure increase. This suggests
that the rate of heat release is essentially constant and is probably
proportional to the rate of fuel injection. Similar control of heat re-
lease rate is observed in diesels after the initial spike due to pre-
/
mixed burning. An interesting feature regarding Figure 1 suggests itself
at this point. The data for the three fuels were taken while holding the
total energy input constant (i.e. mass injected x LHV = constant). This
was accomplished solely by varying the injection duration which means that
the rate of energy input was different for each fuel. The experimental
110
-------
o gasoline
JP-4
A diesel
60
BTC
40 20 TOC 20
40 60 80 100
CRAM ANGLE DEGREES
120 140
ISO
ATC
Figure 1 PVY plot using experimental data on HCC engine - 1200 RPM, full load (ref. 9)
-------
data seem to bear this out in that, during the injection period, the
curves corresponding to the individual fuels have slope differences pro-
portional to the differences in their rates of energy input. Barber,
et al ^ -* theorize that the combustion process follows the description
given pictorially in Figure 2. Fuel is vaporized and mixed with air to
FUEL
INJECTOR
DIRECTION OF
AIR SWIRL
I-FUEL SPRAY
2-COMBUSTIBLE AIR-FUEL MIXTURE
3-FLAME FRONT
4-COMBUSTION PRODUCTS
Fig. £-Texaco combustion process - air-swirl method (ret.
combustible proportions before reaching the spark plug. A quasi-stationary
flame front is established at the spark plug by ignition of the first mix-
ture flowing into the gap and persists for the duration of fuel injection
due to the steady influx of fresh mixture. Combustion products are car-
ried downstream, mixing with available air as they go and yielding addi-
tional heat release until all products have been oxidized to the overall
112
-------
equivalence ratio. This explanation, although not accounting for the
initial delay period, seems to be consistent with the remainder of the
injection period. Nagao, et al •- -" have demonstrated that a stationary
flame front can be maintained in a combustion chamber utilizing moder-
ate rates of swirl and a homogeneous mixture, so long as the ignition
Fl?l
source is continuous. Heywood1- J suggests that the spark plug acts as
a flame holder during the injection period with the fuel-air mixture
burning in its turbulent wake. Martin'- -" and Wong'- -* have recorded
high speed color photographic sequences of TCCS combustion simulated on
a rapid compression machine. Their films show that the first visible
burning occurs just downstream of the spark plug and that burning con-
tinues in this region throughout injection. The luminosity given off
by the flame during this period is an intense white-yellow. According
to Rife and Heywood'- ^ and Scott*- ^, this coloration is associated
with a turbulent diffusion flame, rich in carbon particulates, with a
temperature as high as 4200°F. Both Martin and Wong noted a delay
period (1.0-1.8 msec) between start of injection and first appearance
of visible flame. Wong ascribed this to the first portion of combus-
tible mixture being too cool (due to vaporization and air entrainment)
to ignite. This explanation is unsatisfactory in that the first portion
of fuel-air mixture should be the easiest to ignite since its tempera-
ture would be higher than all succeeding portions if no ignition occurred.
Two other explanations seem more probable. It is conceivable that in its
initial stages, the developing fuel jet from the nozzle misses the spark
gap altogether. Another possibility is that premixed burning is occur-
ring during the delay period. The energy release due to this burning
would be partially offset by the heat transfer required for fuel
113
-------
vaporization. This would explain the lack of appreciable pressure change
during the delay period. Curiously, neither Martin nor Wong followed the
recommendation of Scott"-16^ and Alcock^-17-" regarding the addition of cop-
per oleate to the fuel. This measure would have assured that any pre-
mixed burning would have sufficient luminosity to show up on the color
film. Between 10° ATDC and 70° ATDC, the expansion appears to be charac-
terized by the competing effects of further heat release due to entrain-
ment and mixing of the remaining air with the swirling plume of rich
combustion products and heat transfer, probably radiative as well as
convective, to the combustion chamber surfaces. Heat release dominates
until about 40° ATDC and is thereafter balanced by heat transfer. The
movies of Scott*- ^ taken on a compression swirl chamber diesel have
shown that the centrifugal force field resulting from the swirl causes
the hot combustion products to move toward the center of the chamber
while the cooler air migrates to the periphery. One might expect the
same sort of segregation to occur in the TCCS (although clearly not on
the same order of magnitude due to the differences in the swirl rate)
creating an interface across which mixing and oxidation processes take
Fl3l
place. MartinL J observed localized areas of high vorticity, embedded
in the swirling mixture, during the period of combustion following in-
jection cutoff. Such random turbulence would probably promote mixing
between the regions of products and air. Lyn and Valdmanis,'- •" using
a Schlieren photographic technique on a swirl chamber diesel, reported
microturbulence in the form of small eddies, superimposed on the main
swirl, which caused rapid diffusion of soot particles into the oxidizing
Fl4l
region. WongL J observed a rather curious effect of swirl rate on the
duration of visible combustion following the injection period in the TCCS.
114
-------
With all other parameters held constant, the duration of visible flame
(Scott1- J reports that the last visible portion of diesel combustion
is a dull red color corresponding to a temperature of about 1850°F.)
first increased and then decreased as the BDC swirl rate was increased
from 500 to 1100 radians per second. The influence of swirl rate on
mixing and combustion is apparently not straightforward. The remainder
of the expansion after 70° ATDC is dominated by heat transfer effects
until the blowdown process begins, somewhere around 150° ATDC, at which
point the resulting expansion causes both the pressure and temperature
of the combustion products to decrease rapidly.
Let us now retrace our steps and review, with the aid of the block
diagram given in Figure 3, what is known about the heterogeneous combus-
tion process occurring in the Texaco engine, pointing out as we go the
various points at which hydrocarbon compounds may be formed. The discus-
sion will follow the letters a, b, c, . . ., on the right side of the
block diagram.
a) Air Intake
roT
Davis, et a!L J describe the TCCS as a "coordination of fuel
injection and positive ignition with swirling air." This last is
brought about by use of an intake port and shrouded intake valve con-
figuration (although the most recent version of the TCCS has abandoned
the shrouded valve in favor of a swirl port) designed to cause uni-
directional flow into the cylinder. Upon compression, the swirl set
up by the constraint of the cylinder wall is intensified as the air is
forced into the smaller diameter combustion chamber bowl. The design
swirl rate at TDC is such that the smoke limited maximum power fuel
injection duration is approximately 80 percent of the time necessary
115
-------
Air Motion,
Swirl
(Port, Piston
iConfiguration)
Droplet, Decelera-
tion, Vaporization
d Delay
Period
Surface
Imp ingement
Mixing & Fur-
ther Oxidatio4
Rapid
Combustion
Vaporization
and/or Partial
Oxidation
J
,
k
ixJng
Mixing
Controlled
Combustion
i_
FIGURE 3. TCCS Block Diagram
-------
for one complete revolution of the swirling air. Some squish occurs,
but its effect on the general air motion is negligible according to Alcock
and Scott. J The resulting air motion prior to fuel injection has been
characterized by Watts and Scott'- ^ as somewhere between free and forced
vortex and by Dent and Derham"- * as solid body rotation. Apparently,
more study is needed in the area of air motion in the cylinders of engines,
both prior to and during combustion, before the motion can be adequately
described.
Fuel Injection
Fuel is injected through a single hole nozzle utilizing a flat seat-
ing arrangement and a short length to diameter ratio. Several adequate
models are available for predicting fuel injection flow rates as a func-
tion of injection system design and operating parameters and fuel
properties.
b) Breakup
F211
According to El Wakil, et al, J breakup of the fuel jet in diesel
engines occurs within a few orifice diameters of its leaving the nozzle.
[221
The spray chamber observations of Burt and Troth, J and the rapid
F151
compression machine study of Rife and HeywoodL J indicate that this
will also be the case for injection and combustion chamber pressures
T231
typical of the TCCS. The spray analysis of Jain, et al, J based on a
continuum model of the fuel jet, predicts that the combination of in-
jection parameters used in the TCCS is sufficient to insure "complete and
immediate" disruption of the fuel jet for a wide variety of fuels ranging
from methanol to a wide boiling range distillate.
c) Droplet Deceleration, Vaporization and Mixing
The interaction of the atomized fuel spray with the swirling air,
117
-------
in the time period between its leaving the injection nozzle and reaching
the plane of the ignition source, is probably the most critical part of
the TCCS. Other than the model of Jain, et al,^23-' no analysis of the
fuel spray-air interaction peculiar to the TCCS has been found in the
literature. By equating the change in momentum of fuel droplets to the
corresponding aerodynamic drag force, Jain calculated that the droplets
would decelerate to the velocity of the entrained air within a very short
distance (-1.5 mm) of leaving the nozzle. Using an equation for the
evaporation time of a single stagnant drop in an infinite atmosphere,
modified to include the effects of forced convection, Jain next calculat-
ed that all the fuel would be vaporized before it reaches the ignition
source. Although it is stated at the outset that the analysis is strictly
for performance modeling, it is not at all clear that the authors recog-
nize the inaccuracies introduced by neglecting the interaction of proxi-
mate droplets and most particularly by the assumption of a monodisperse
(10 ym droplets) spray. Sass^ -" reports that typical droplets from
diesel injection nozzles vary in diameter from 2 ym to 50 ym. More re-
F251
cently, Hiroyasu and KadotaL J have put forth an experimentally deter-
mined drop size distribution based on an empirical expression for the
Sauter's mean diameter (roughly, the ratio of the integral of droplet
volume distribution to the integral of droplet surface area distribution).
However, no verification of this expression has as yet been found in the
literature. Clearly, if different droplet sizes exist in the TCCS spray,
then we also need to know something of their velocity distribution.
F26l
Borman and Johnson,1 J in their single droplet study, have shown that
initial droplet diameter affects both the droplet's path of motion and
its vaporization history and conclude that accurate simulation of the
118
-------
spray must include consideration of the reduction in vaporization rate
caused by the presence of fuel vapor from nearby droplets. Jain treats
the mixing of the fuel jet with the swirling air by adopting the analyt-
ical scheme of Rife and Heywood.'- ^ This scheme was based on turbulent
entrainment parameters developed from observation and modeling of air
entrainment into smokestack plumes and gave good agreement with spray
tip penetration and spray centerline trajectory observations made by the
f27l
authors in their rapid compression machine studies. Khan and WangL J
believe that the rate of preparation of combustible mixture is controlled
by two scales of mixing. They define these as macromixing, the gross
entrainment of air into the fuel jet, and micromixing, the small scale
mixing of fuel and air occurring within the fuel jet. They indicate that
macromixing is increased by jet impingement on combustion chamber sur-
faces and that micromixing is dependent on the rates of droplet evapora-
tion and turbulent diffusion of fuel vapor within the jet. Micromixing
is also thought to control the distribution of equivalence ratio and
temperature within the jet. This latter has been reported by El Wakil,
et aU ^ to increase by several hundred degrees between the central spray
core and the outer fringe of the spray. Clearly, numerous concepts of
fuel spray-air interaction and mixture formation are available. However,
none is universally accepted and further basic studies are needed to
provide the foundation for an accurate, flexible model.
d) Surface Impingement
F231
Although the TCCS fuel spray analysis of Jain, et alL J predicts
fl4l
that all the fuel will be vaporized before reaching the spark plug, Wong,
in his rapid compression machine studies of the TCCS, has observed liquid
fuel impingement on the combustion chamber bowl surface. Some of this
119
-------
fuel was deflected toward the center of the bowl where it quickly
T281
vaporized and mixed with the swirling air. According to Henein, the
fuel remaining on the bowl surface will vaporize slowly, due to the rel-
atively low temperature of the surface, and may contribute to hydro-
carbon emissions by taking part in wall quenching and partial oxidation
reactions during the remainder of combustion. These possibilities will
be discussed further in the appropriate section as indicated on the block
diagram.
e) Fuel Droplets
It seems probable that liquid fuel droplets still exist at the start
of combustion in the TCCS, particularly in the core of the fuel jet where
T291
the temperature is lowest and vaporization rates slowest. AlpersteinL J
indicates that injection tailings may be a source of unburned hydrocarbons
in the TCCS. These large droplets are introduced into the combustion
chamber at the end of the injection period when the pressure differential
across the nozzle is insufficient to cause breakup and penetration of the
fuel. Numerous experimental and analytical studies of droplet combustion
have appeared in the literature over the years. However, Williams, ^
in his review of recent developments, concluded that further study is re-
quired before an adequate description of combustion of a moving droplet
in a high pressure, high temperature environment can be formulated.
f) Fuel Vapor Plus Air
In addition to liquid fuel droplets, we also expect that a substan-
tial portion of the injected fuel will have vaporized completely prior
to the onset of combustion. The Schlieren pictures of Lyn and Valdmanis^ *
suggest that mixing and evaporation are controlled by air entrainment at
the fuel jet boundary where turbulent eddies are seen to form and roll
120
-------
away from the jet to be carried along by the swirling air. These eddies
contain small packets of fuel vapor and the premixed portion of the
charge is thought to form in this way. The remainder of the vaporized
fuel spray will burn as a turbulent diffusion flame with the burning rate
controlled by the rates of turbulent diffusion of fuel vapor within the
jet and entrainment of oxidant into the burning zone.
g) Beyond Lean Limit
We might expect a portion of the premixed charge formed at the fringe
of the fuel jet to become so thoroughly mixed with air prior to ignition
f28l
that it passes beyond the lean flammability limit of the fuel. Henien1 J
indicates that such regions of premixed charge are an important source of
unburned hydrocarbon emissions in open chamber diesels, particularly at
F311
light loads. In addition, Barnes1- J correlated odor intensity with lean
flammability limit, and thus with the proportion of fuel existing in pre-
mixed regions with fuel-air ratios beyond the lean limit, by replacing
the intake air with mixtures of oxygen and a variety of inert diluents
such that the lean flammability limit, using the same fuel, would be dif-
ferent. Odor compounds are generally identified as unburned, decomposed
and partially oxidized fuel fractions and as such also enter the hydro-
carbon emissions picture.
h) Pyrolysis
Droplets existing at ignition will burn as droplets with the rate
of heat release governed by the rates of diffusion of fuel vapor and
oxidant into the burning zone surrounding the drop. If the temperature
of the droplet becomes high enough, due to convective and radiative heat
transfer, thermal cracking (endothermic breakdown of the fuel molecules
to species of lower molecular weight) may occur yielding new non fuel
121
-------
hydrocarbons into the reaction scheme. Pyrolysis reactions are also
observed in the region between the droplet surface and the burning zone.
These reactions lead to the formation of small carbon particles (50-2500 A)
composed of carbon and 1 to 3% hydrogen. The hydrogen probably appears
in such small quantities since it is more readily oxidized than the carbon
in the fuel. Although the reaction kinetics of soot formation are for
all practical purposes unknown, it is recognized that the previously men-
tioned carbon particles figure heavily in the overall scheme. Soot is
also thought to form in the rich core of the turbulent diffusion flame.
[321
Khan, et alL J have suggested that the overall rate of soot formation
can be adequately predicted by an Arrhenius equation of the form
jjf « 4>n exp(-E/RT)
where t)jn accounts for the influence of fuel-air equivalence ratio and n
is estimated by comparison of predictions with experimental data. How-
ever, this approach is an interim one and more fundamental study is needed
before the actual mechanism comes to light.
i) Quenching
We expect some contribution to the overall hydrocarbon emissions
from quenching phenomena occurring during the combustion process. Specif-
ically, we would expect a flame propagating into a region of premixed
charge adjacent to the combustion chamber surface to be quenched in much
the same manner as quenching occurs in conventional premixed charge S.I.
engines. This situation might occur at the surface of the bowl where fuel
spray impinged. The diesel movies of Scott'- -" show that near TDC the-
large velocity gradient near the cup surface tends to shear the liquid
fuel off, vaporizing and mixing it with air. If the resulting mixture
122
-------
ratio is within the flammability limits, flame will burn through this
premixed layer to within a small distance of the surface and then be
quenched due to heat transfer effects and radical destruction in the
cool unburned mixture. This so-called quench distance will be deter-
mined by the instantaneous pressure, by the local mixture temperature
and by fuel-air ratio.
In addition, we expect that stratification of the fuel-air mixture
will be extreme in certain regions, particularly at light loads, and
that a flame will cease to propagate through such a gradient when the
energy released in the flame is no longer sufficient to sustain the
reaction. This occurrence is often called air quenching.
Lastly, if combustion duration extends late into the expansion
process, the reactions which complete oxidation to the overall equiva-
lence ratio may be quenched by the cooling effect of the expansion.
j) Mixing and Furtheir Oxidation
At the end of the injection period, the combustion chamber is filled
with a rich burning plume surrounded by the remaining unused air. As
the expansion process continues, the rate of heat release available from
the rich products, and the duration of combustion, will be controlled by
the rate of entrainment and mixing of this remaining air with the plume.
Jain, et alL J have modeled this portion of combustion by using the tur-
[331
bulent eddy entrainment parameters derived by Blizzard and KeckL J for
their turbulent burning model. Agreement between Jain's model predic-
tions for PVY versus crank angle and the available experimental data is
reasonably good. However, the turbulent burning model of Blizzard and
Keck was developed for a flame propagating through a homogeneous, pre-
mixed charge in a non swirling environment. The possibility that Jain's
123
-------
model's agreement with experimental results is partially fortuitous should
probably be taken under consideration.
k) Vaporization and/or Partial Oxidation
As stated previously, Wong'- -" has observed fuel impingement on the
surface of the combustion chamber cup. The portion of the liquid fuel
which remained on the cup surface and did not vaporize early in the com-
bustion cycle and take part in premixed burning, as described in (i),
will find it increasingly difficult to encounter sufficient oxygen for
complete combustion. Henein*- -" suggests that this late vaporizing fuel
will decompose, forming unburned hydrocarbons, partial oxidation products
and carbon particulate. This occurrence is clearly demonstrated in Scott's. J
movies when the swirling combustion products spill out of the cup and be-
gin to diffuse into the clearance volume between the edge of the cup and
the cylinder wall. Adjacent to the points of spray impingement, large
F291
smoke clouds appear indicating insufficient oxygen. AlpersteinL J reports
that unburned hydrocarbon reduction in the cylinder of the TCCS can be
enhanced by promotion of small scale turbulence between the piston and cyl-
inder head and by increased swirl rate.
Premixed charge whose fuel-air ratio was beyond the lean flammability
limit prior to ignition may, if sufficiently isolated, pass through the
combustion process relatively intact or it may be partially oxidized by
the high temperatures attained during the combustion period. Either way,
it will enter the overall hydrocarbon emissions picture.
1) Slowdown and Exhaust Process
Extensive research with homogeneous, premixed charge, spark ignited
engines has shown that hydrocarbon compounds are selectively exhausted
from the cylinder, depending on their physical origin. Daniel and
124
-------
Uentworth'- ^ were the first to document the variation in exhaust HC
concentration with crank angle by sampling at the exhaust port.
F35"i
Tabaczynski, et alL J carried this one step further by measuring in-
stantaneous mass flow rate of HC during the exhaust process. Both
papers explained the variation in HC in terms of the physical origin
of the gases leaving the chamber. They also demonstrated that a large
proportion of the unburnt hydrocarbons formed during premixed combus-
tion are not subsequently entrained into the flow leaving the cylinder,
but remain in the cylinder as residuals, primarily because of their
physical locations (i.e. combustion chamber surfaces far removed from the
exhaust valve and also the piston face). We expect that hydrocarbon com-
pounds formed in the TCCS by mechanisms, and in locations, similar to
those occurring in conventional spark ignition engines will behave in the
manner predicted by these studies.
This concludes the review of the heterogeneous combustion process
occurring in the TCCS. Due to the lack of literature dealing specifical-
ly with the TCCS, much of the material presented was drawn from studies
of open chamber, direct injection diesels utilizing combustion chamber
configurations and air swirl rates similar to those used in the TCCS.
The points at which hydrocarbons appearing in the exhaust are thought to
originate have been identified and a brief description of the particular
mechanism(s) involved has been presented.
The next section will construct the hypothesis of hydrocarbon forma-
tion in the TCCS and review the literature dealing with the specific
mechanisms involved in the scheme.
125
-------
HYDROCARBON FORMATION IN THE TCCS
We are now in a position to begin formulating the mechanisms respon-
sible for hydrocarbon emissions from the TCCS. Figure 4 is a plot of
specific hydrocarbon emissions (grams C/ihp-hr) versus fuel-air equiva-
lence ratio taken from the experimental data of Lazarewicz. -1 The
inset plot for a direct injected, open chamber diesel was extracted from
f36l
Yumlu and Carey, J and the plot for the homogeneous charge S.I. engine
[371
is from Jackson, et al. J The order of magnitude or more difference
in the rates of hydrocarbon emission between the TCCS and the diesel can
be attributed to the higher average temperatures attained during combus-
tion in the diesel as a result of its higher compression ratio (-16:1 vs
10:1 for the TCCS). This difference in emission levels probably accounts
for the lack of reference to hydrocarbon emissions in the majority of the
diesel literature; they are just not a serious problem in diesels. How-
ever, present vehicle applications of the TCCS are able to meet current
HC emission standards only with the addition of an oxidation catalyst.
With more stringent standards on the horizon, an understanding of the in-
cylinder HC formation mechanisms occurring in the TCCS will be fundamental
to improving its emissions characteristics.
One feature of Figure 4 worth noting is that the minimum specific
hydrocarbon emission for each type of engine occurs at approximately the
same fuel-air equivalence ratio. The shape of the homogeneous S.I. curve
is reasonably well understood at this point. Flame quenching theory pre-
dicts that for premixed burning, the minimum quench layer thickness (and
thus minimum HC emissions) should occur at fuel-air ratios slightly richer
than stoichiometric where the flame speed is greatest. However, as
126
-------
80
.OO61 . OVi
FUEL-AIR RATIO
T 'T 'T
.061
70
60
50
.c
I
0.
J=
u
i
s
UJ
o
30
u
o
CE
Q
z 20
1,0
•1500 RPM
A.20OO RPM 0007.
• 2500 RPM
Diesel Engine
ENGINES POINTS
002 003 004
FUEL-AIR RATIO
oo» o • \S A
Homogeneou
S.I. Engine
LOG R£FZR£NCZ: JULY 9,1975
4 .5 .6 .7 .8
EQUIVALENCE RATIO
FIGURE 4 HYDROCARBON CONCENTRATION VS EQUIVALENCE RATIO (ref. 10)
127
-------
TOOT
Huls, et a!L J point out, quench theory does not account for any post
combustion oxidation which may occur during the remainder of the expan-
sion and exhaust processes in an engine. Exhaust temperature remains
quite high as the fuel-air ratio is leaned beyond stoichiometric. Figure
5 is a plot of average exhaust temperature versus fuel-air equivalence
ratio for the TCCS. The dashed line represents typical exhaust tempera-
tures for a homogeneous charge S.I. engine. The high exhaust temperature,
in conjunction with the excess air available on the lean side of stoi-
chiometric, causes the minimum hydrocarbon emission to occur slightly
lean of stoichiometric. Leaning the mixture further causes the quench
layer thickness to increase, although the concentration of fuel in it is
lower, and the lower exhaust temperatures slow oxidation reactions by a
factor greater than the increase in oxygen concentration can balance.
The net result is an increase in hydrocarbon emissions. At mixture ra-
tios richer than the ratio corresponding to the minimum quench thickness,
both the quench thickness and the fuel concentration in the quench layer
increase. Also, fuel penetrates into crevices such as the region be-
tween the piston crown and top compression ring and the cylinder wall.
Burning is quenched before entering these regions and the fuel remains
intact. Exhaust temperatures remain quite high, but there is insufficient
oxygen available to promote oxidation reactions. The result is that hydro-
carbons rise sharply on the rich side of stoichiometric. Referring again
to Figure 5, we can see that when operating at similar equivalence ratios,
exhaust temperatures in the TCCS approach those in the homogeneous charge
S.I. engine (as we expect since their compression ratios are similar).
We can assume then that exhaust destruction of hydrocarbons in the TCCS
under these operating conditions is probably as important as it is in the
128
-------
o
o
u_
o
I
LJ
CC
Ul
CL
Ui
in
X
Ul
• I50O RPM
A200O RPM
• 25OORPM
LOG REFEREN'CE: JULY 9,1975 _
.1 .2
.3
1.0 I.I
.4 .5 .6 .7 .8 .9
EQUIVALENCE RATIO
FIGURE 5 EXHAUST TEMPERATURE VS EQUIVALENCE RATIO (ref. 10)
•129
-------
homogeneous charge engine. However, as we reduce the load in the TCCS,
its exhaust temperature drops rapidly. It is not yet clear whether the
rapid increase in hydrocarbon emissions from the TCCS which occurs as
load is reduced, as shown in Figure 4, is due primarily to reduced
destruction in the exhaust, or to increased formation during combustion,
or some combination of these effects.
Spray Tailings
One problem occurring in engines which utilize direct, high pressure,
fuel injection systems is the injection of large fuel droplets at the end
of the fuel injection period because of the reduced pressure differential
across the nozzle. The kinetic energy imparted to such drops is so low
that they do not atomize or penetrate far from the nozzle. Rife and Hey-
fl5l
woodL J observed spray tailings at all fueling levels in their rapid
f29l
compression machine study of diesel combustion. AlpersteinL J has in-
dicated that fuel spray tailings are a probable source of hydrocarbon
emissions in the TCCS although he indicates that post injections, another
possible source of HC, have been successfully designed out of the TCCS
injection system. The length to diameter ratio (1/d) of the TCCS injec-
tion nozzle is quite small so we expect minimal dribble of fuel after the
end of injection. Consequently, we state at the outset that injection of
large fuel droplets at the end of the injection period may be a source of
exhaust hydrocarbons at all loads. Whether these droplets burn completely
or not will be a function of local temperature and oxygen availability.
In addition, their burning will be diffusion controlled and thus time
dependent.
130
-------
Light Loads
Air Quenching of combustion reactions may be an important mechanism
of hydrocarbon formation in the TCCS at idle and light loads. Light load
combustion in both the TCCS and the diesel requires extreme stratifica-
tion of the fuel-air mixture because of the small quantities of fuel in-
jected. Overall fuel-air ratios at idle approach .01:1. Since the delay
period between the start of injection and appreciable pressure rise in
the TCCS has been observed by Wong*- -" to be relatively invariant with
load, we can assume that as load is reduced, a greater proportion of the
injected fuel will have been vaporized, mixed with the swirling air and
swept past the spark plug before combustion starts. That this charge is
premixed seems a reasonable assumption since most investigators agree
that the initial "spike" of heat release in diesels results from premixed
burning and since the time available for fuel vaporization and mixing in
the TCCS is comparable to the ignition delay period in diesels. Upon
ignition, the flame will propagate through this mixture until the energy
released by the flame no longer exceeds the heat transferred to the sur-
rounding mixture. This so-called air quenching of the flame leaves a
portion of the injected fuel in an unreacted state. How much fuel is
left will depend on how far the flame propagates through the mixture gra-
dient. This will be a function of the heat transferred to this mixture
from prior combustion, insofar as the mixture temperature affects the
lean flammability limit. Average gas temperatures resulting from com-
bustion at idle and light load are relatively low. This is reflected in
the low exhaust temperatures shown in Figure 5. The low combustion tem-
perature, in conjunction with the diffuse concentration of the remaining
fuel molecules, causes the post flame oxidation reactions to proceed slowly.
131
-------
The result is that some of this fuel winds up in the exhaust in an un-
burned or partially oxidized state. We don't expect much destruction of
such products in the exhaust because of its low temperature. Figure 6,
compiled from the data of Hurn,^-39-' shows that at idle hydrocarbon com-
pounds found in diesel exhaust have molecular weights similar to that of
the original fuel. Figure 7, taken from Johnson, et al, J shows that
for direct injected, open chamber diesels, both the specific rate of
hydrocarbon emission and the fraction of unburnt fuel are highest at the
lowest fueling level. Figure 8, also from Hurn, shows that the highest
concentrations of unburnt hydrocarbons and aldehydes (partially oxidized
hydrocarbon compounds identified as contributors to diesel exhaust odor)
occur at idle in diesel exhaust. Barnes"- ^ suggests that compounds
associated with odor result from partial oxidation reactions occurring
in the regions where the fuel-air mixture was beyond the lean flammability
limit prior to the start of combustion. Mitchell, et al^- -* confirm that
diesel-like odor is noticeable in the TCCS exhaust at idle and light loads.
Fuel Impingement on the bowl surface may or may not occur in the TCCS
f28l
at light loads. Henein1 J states that it does not occur in diesels, but
offers no experimental proof or logical reasoning to support his view.
T271
Khan and Wang J suggest that it is unimportant at light load but won't
Fl4l
go as far as claiming it does not occur. WongL J observed impingement
in his rapid compression machine study of the TCCS, but it is not clear
from his thesis that it occurred at all fueling levels. The spray pene-
f26l
tration theory of Borman and Johnson1 J says that succeeding droplets
penetrate further because of momentum transferred from preceding'drops
to the entrained air. This implies that ultimate spray tip penetration
is a function of the total momentum transferred from the droplets to the
132
-------
6 *
0
o
<-• e
D
°, 4
ID
5
O
I "7
^
° c
t:
o - ^
n
i .^
NA TC rpm ['
* * 1600 " -<
CU
n • 2000 ?-
-(
14 1C, | | | | | |
-I •* | 6 811
'
1
|| 16 20 TL 2A-
Carbon otoms per
molecule
^
o . 2400
-, * oonn ''"'• 6 "eel ol lo.ul on ilisli ihnlion ol inihiiineil ludio / -.
- * 2 BOO . . - ( K»OT
e:tihon emissions in ,i 1) I engine Vlcl
2
00
0
,. .004
£
c .003
;~
^^
o
z .002
K .001
u
z
i/> o
1 1 1 1 1
~ ^^8*^1 ^ UBF ~
o • * *
— X ' ^^
\
c \
"" \V ""
8 "~^~- _!_ ISHCR
1 1 1 « 1 ~-~'\'
Fig. 7 - Effect of fuel delivery on
specific hydrocarbon rate and per cent
0 20 40 60 80 100 120 unburned fuel (1600, 2000, 2400, and
2800 rpm; naturally aspirated and
FUEL DELIVERY , cu. mm/stroke turbocharged conditions) (ref. 38)
JIYY1
MUkAJ [
()
3000 S
o
o
w-l
o
CK ?000
(X
0
u>
^
o. 100°
\
\
\ /— Unburned Hydrocarbons
\^* I As Carbon 1
\
>\
, — Aldetiydes
•^-H —
*-,
i>^
'
o • — ~ — •
0 25 ^0 75 100
% IOAO
I K,. O I iKx'l ol lo.ul on unlHiinol huh oc.u hum .ind
:iljL'li)ilc emissions in :i I) I engine (r6f. 39)
133
-------
air. Therefore, decreasing injection duration should reduce the penetra-
tion of the spray. It is conceivable then that at idle and light loads
penetration of the fuel spray is insufficient to reach the bowl surface.
However, should the spray reach the bowl surface we would then have addi-
tional possibilities for hydrocarbon emissions. The velocity gradient
near the bowl surface is large and, according to Scott, -* tends to shear
off liquid fuel, vaporizing it and mixing it with air. If this mixture
is subsequently ignited, we would expect the flame to propagate in close
to the bowl surface and be quenched in the same way that burning is
quenched in homogeneous charge engines. This quenching would leave a
layer of unburned mixture which might or might not leave the cylinder
during the exhaust process. At light loads, ignition of this mixture
might not occur and vaporization of the fuel would continue well into
the expansion. Both of these possibilities could contribute substantially
to overall hydrocarbon emissions. Spray tailings occurring at idle and
light load probably represent a substantial part of the injected fuel.
If these drops descend into the bowl before being ignited, they will be
spread across the bowl surface by the swirling air and will undergo the
same processes as impinged fuel. If they ignite, they will burn at a
diffusion controlled rate. At the low combustion temperatures attained
at light load, they may not burn to completion in the time available.
Moderate Loads
As load is increased, several competing factors come into play. Less
time is available for combustion of the last portion of the fuel spray and
proportionately less oxygen is available to complete this combustion.
However, these two conditions are more than offset by the increased rates
of the oxidation reactions, promoted by the higher temperatures resulting
134
-------
from combustion of larger quantities of fuel. The higher temperatures
extend the lean flammability limit, so we expect a smaller quantity of
unburned fuel resulting from air quenching. In addition, the higher post
flame reaction rates will oxidize much of this unburned fuel further
reducing the influence of air quenching on overall hydrocarbon emissions.
Fuel impingement is more assured at moderate loads and Figure 9, taken
F271
from Khan and Wang,L J demonstrates clearly its effect on hydrocarbon
emissions. It shows that at a constant fueling rate, advance of injec-
tion timing (i.e. earlier in the compression stroke) results in increased
hydrocarbon emissions. Khan and Wang attribute this rise to increased
fuel impingement and resultant wall quenching. This seems reasonable
since earlier injection occurs against lower combustion chamber air den-
sity and spray penetration should increase. With this in mind, it seems
appropriate to discuss more fully the phenomenon of wall quenching and
its effect on hydrocarbon emissions.
Wall Quenching of flames in spark ignited engines was first observed
by Daniel."- -" He measured quench distances (thickness of the layer of
unburned fuel-air mixture between the flame front and chamber surface at
the time of flame extinction) ranging from .002" to .015", depending on
cylinder pressure, local gas temperature and stoichiometry, and obtained
good agreement with state-of-the-art quench distance predictions. Daniel
concluded by postulating that a large proportion of hydrocarbons appear-
ing in homogeneous S.I. engine exhaust originate in the quench layers
formed by flame quenching. This theory received support from the combus-
tion bomb study of Shinn and Olson.'- -* To avoid the cyclic variability
associated with engine combustion, they constructed a combustion bomb and
control apparatus which allowed simulation of the engine combustion cycle,
135
-------
•400
50
30 Z5 20 15 10 5 t.d.c.
DYNAMIC INJECTION TIMINC-°b.t.d.c.
Injection shape Injection period Rate
A Steep front 17° c.a. 3-35
O Pilot front 22° c.a. 2-73
• Steep front 22° c.a. 2-73
Engine speed, 2000 rev/min; 0 = 0-72.
Nozzle: 4 hole x0-28 mm diameter.
Fig. 9 Effect of timing and rate of injection on .
hydrocarbon emissions (Engine B) (ref. 27)
ci
r 0.6
** O.4
0 0.2
n
o
,,
1
1
1
s - S
UNSATURATEO HYDROCARBONS
' CCMB. CHAMBER SAMPLE
a EXHAUST GAS RES. SAMPLE
i I
« i* 1 -
100
150 200 250 300 350
COMBUSTION CHAMBER WALL TEMP (fl
2.6
2 4
2.0
1.8
1.6
1.2
I.C
° 8
OG
O.fl
tl 2
O
••10
SATURATED HYDROCARBONS
« COMB. CHAMBER SAMPLE
« EXHAUST GAS RES. SAMPLE
:00 150 20O V!"0
COW'.'HOf'OM Oi'.v.iER W.'J I. rtW. Cf)
I I'j 'J'OiMilJiJII O'li!i:lil ill |),ouU':' : I''
-------
but with independent control of many variables. Figure 10 shows their
measurements of hydrocarbons resulting from isooctane combustion. It
can be seen that the proportion of hydrocarbons in the combustion cham-
ber sample, which was taken by means of a sampling valve mounted flush
with the surface of the compustion chamber, is roughly 2.5 times greater
than the proportion contained in the bulk exhaust gases. The predomi-
nance of saturated (fuel-like) compounds in the chamber sample suggests
that the fuel-air mixture near the wall was left unburned by the quench-
ing process. In addition, it was noted that condensation of fuel becomes
important when the combustion chamber surface temperature approaches the
dew point temperature of the fuel-air mixture. Shinn and Olson accounted
for the difference in hydrocarbon concentration between the bulk exhaust
gas and the chamber sample by theorizing that quench layers far removed
from the exhaust valve are unlikely to leave during the exhaust process
and that viscous drag will retard entrainment of quench regions near the
T341
valve into the bulk flow. Daniel and WentworthL J sampled combustion
gases near the head surface in an operating engine by means of a small,
inwardly opening poppet valve. The general layout of the sampling valve
is shown in Figure 11 and the measured sample hydrocarbon concentration
versus sampling rate (controlled by adjusting valve lift in this study)
is shown in Figure 12. The results show that as sampling rate is reduced,
the concentration of unburned hydrocarbons in the sample increases. The
reason for this is that small sample flow rates contain greater propor-
tions of gas withdrawn from the region close to the surface. They iden-
tified the reason for this gradient in hydrocarbon concentration as wall
quenching. Further, they showed that the concentration of hydrocarbons
in well mixed residual gases remaining in the combustion chamber at the
137
-------
HYOkAUilC liflC*
Fig. 11 Quench zone sampling valve actuating mechanism (f*6f. 34)
r i i i
j^l llf.', /AlVt If It AC t
Hi.'. 12 ! !;. Jrocirr- :i C(»n,?^i'irji?on of g;.Jf:j .-vi
qii' u'-ii :• >ne sin:(jlin>; valve (ref. 34)
with
138
-------
end of the exhaust process is some eleven times greater than the measured
average exhaust hydrocarbon concentration. This confirms Shinn and Olson's
theory that much of the hydrocarbon-rich quench regions remain in the
combustion chamber as residuals.
Numerous additional literature dealing with the effects of wall
quenching phenomena on hydrocarbon emissions could be referenced. How-
ever, the portion presented is sufficient to show that wall quenching
resulting from vaporization and burning of fuel which impinged on combus-
tion chamber bowl surface is probably an important hydrocarbon formation
mechanism at moderate to heavy loads. Fuel in the bowl which does not
vaporize and burn early in the combustion process will continue to
vaporize during expansion. However, the higher gas temperatures and
excess oxygen available will promote oxidation of at least part of this
fuel vapor. Whether or not fuel vapor remaining in the bowl at the start
of the exhaust process will leave is open to some question. It would
appear from Daniel and Wentworth's observations that the vapor would
probably not be entrained into the exhaust flow.
Spray tailings should ignite quickly at moderate loads. Heating of
these droplets from surrounding combustion may cause some pyrolysis of
the fuel, but the droplet should burn to completion in the excess air.
Lastly, exhaust temperatures rise rapidly as load is increased, as shown
in Figure 5, and we expect that destruction of hydrocarbons in the exhaust
will become more important. The net result is a rapid decrease in hydro-
carbon emissions as load is increased from light to moderate levels, as
can be seen in Figures 4, 7 and 8.
139
-------
Heavy Loads
As load is further increased, the fuel-air ratio approaches stoi-
chiometric and the hydrocarbon emissions from the TCCS begin to increase.
The longer period of injection results in increased fuel impingement in
the bowl. If this fuel vaporizes early in the combustion process, it
will probably mix with air and burn. The resulting flame will be quenched
near the bowl surface and a thin layer of unburnt hydrocarbons will be
F28l
formed. If, however, this fuel vaporizes slowly, as HeneinL J suggests,
the fuel vapor will probably decompose forming unburnt hydrocarbons,
f39l
partially oxidized compounds and possibly some carbon particulate. Hum J
suggests that in diesels operating near full load, time factors begin to
limit oxidation processes. Since more fuel is injected, we expect a
larger proportion of the mixture at the core of the jet to be richer than
the rich flammability limit. Mass transport, mixing and heat transfer
from the surrounding combustion ultimately determine how much of this ini-
tially over rich mixture subsequently burns; these processes are all time
dependent. If droplets other than spray tailings exist, they will be
found in the core and upon ignition they will burn at a diffusion con-
trolled rate. We have previously noted that radiant and convective heat
transfer from nearby combustion can raise temperatures sufficiently to
cause thermal cracking of the fuel in both its liquid and vapor phases,
particularly in regions where there is insufficient oxygen to complete
combustion. Such cracking produces hydrocarbon compounds of lower molec-
ular weight than the original fuel. Fristrom and Westenberg*- •* suggest
that combustion of a typical fuel molecule in a rich flame proceeds
according to the following scheme.
H + CnH2n+2 ~^ H2 + CnH2n+l
140
-------
They indicate that radical concentrations resulting from this reaction
can be high enough to make recombination reactions important. If such
recombination occurs, it will result in production of hydrocarbon species
of higher molecular weight than the original fuel. These two phenomena
(thermal cracking and recombination reactions) may explain Figure 6 which
shows that at full load, the hydrocarbon composition of diesel exhaust
consists mainly of species having either lower or higher molecular weight
F311
than the original fuel molecules. Barnes1- J also noted that diesels
operating at heavy load on a pure fuel emit a broad spectrum of hydro-
carbon compounds having greater and lesser carbon atoms than the fuel.
Combustion of spray tailings at heavy loads is a probable source of hy-
drocarbon emissions also. The temperature of surrounding gases is high
and oxygen is not so readily available as it is at moderate loads. The
droplets may pyrolise, forming lower molecular weight compounds or car-
bon particles which then take part in soot formation. Soot may also
form in the rich core of the fuel jet at heavy loads. Exhaust tempera-
tures at heavy loads approach those of the conventional homogeneous S.I.
engine, so we expect comparable hydrocarbon destruction in the exhaust.
However, as the mixture ratio nears stoichiometric, oxygen will be scarce
in the exhaust and destruction may decrease somewhat.
Summary
Literature pertaining to the mechanisms of hydrocarbon formation
thought to be important in homogeneous charge S.I. engines and diesels
has been reviewed. Those mechanisms which appear to be operating in the
TCCS have been identified and their varying influence on overall hydro-
carbon emissions has been discussed. In conclusion, we can summarize the
mechanisms operating in the TCCS as follows.
141
-------
1. At all loads, large droplets of fuel introduced at the end of
injection may burn to completion or decompose depending on
local conditions of temperature and oxygen concentration and
the time available for their diffusion controlled combustion.
2. At idle and light loads, air quenching of the burning plume
leaves much of the injected fuel unburned. Post flame
oxidation rates are too slow to eliminate this fuel because
of the low average temperature resulting from combustion.
Impingement of fuel spray on the bowl surface may occur. If
it does, this fuel may vaporize quickly, mix with the swirl-
ing air and burn, leading to wall quenching of the flame near
the bowl surface. If this fuel vaporizes slowly and does not
burn, it may enter the exhaust system intact. Exhaust tempera-
tures are low at light loads and little destruction of hydro-
carbons is expected.
3. At moderate loads, higher average combustion temperatures
extend the lean flammability limit and increase post flame
oxidation rates, thus reducing the effect of air quenching
on overall hydrocarbon emissions. Fuel impingement on bowl
surfaces occurs. Gradual vaporization of this fuel promotes
wall quenching reactions early in the combustion process and
may add raw fuel vapor to the bulk gases late in the expan-
sion. Exhaust temperatures rise rapidly with load and oxida-
tion of hydrocarbons in the exhaust becomes important.
4. At heavy loads, longer injection periods result in more fuel
being deposited on the bowl surface. If this fuel vaporizes
early, it probably mixes with air and burns with quenching
142
-------
reactions resulting. If it vaporizes slowly, its chances
of finding adequate oxygen are reduced and the high tempera-
tures resulting from combustion will decompose it, forming
unburned and partially oxidized hydrocarbons and carbon
particles. Lack of sufficient oxidant in the core of the
jet and time limited diffusion, mixing and heat transfer
processes will lead to incomplete combustion. Also high
temperatures caused by heat transfer from surrounding
combustion will cause pyrolysis of the liquid and vapor
fuel in this region. Exhaust temperatures will be high
and destruction in the exhaust will be important until the
mixture ratio exceeds stoichiometric.
EXPERIMENTAL PROBLEM
Ideally, we would like to be able to formulate a kinetic model for
hydrocarbon formation in heterogeneous combustion. Such a model could
then be incorporated into an engine simulation program for use as a
design tool. However, such kinetic models appear to be a long way off
T121
even for homogeneous charge S.I. engines. HeywoodL J reports that
over 200 organic compounds have been identified in the exhaust from
such engines by means of gas chromatography. Many of these are trace
species, but even ignoring them, the number of compounds to be accounted
for is formidable. We don't know many of the reactions involved and for
those we do, much of the reaction rate data is questionable.
The current approach to understanding the mechanisms of hydrocar-
bon formation has been to study the effect on hydrocarbon concentration
143
-------
of variation of engine design and operating parameters. Daniel and
F341
WentworthL J studied the variation, with engine operating parameters,
of hydrocarbon concentration both in the quench layer and the exhaust
by means of sampling valves. More recent works have investigated the
effects of piston design, cylinder wall temperature and blowby on
hydrocarbon emissions. Daniel'- -• has developed a model for hydrocar-
bon emissions from homogeneous charge S.I. engines based on empirical
functions for quench layer thickness, unburned fuel due to crevices,
post flame oxidation and exhaust destruction. It treats all hydro-
carbons as a single species and involves a trial and error fit of
various parameters in order to obtain an acceptable level of agreement
with experimental observations. However inelegant this technique may
appear, it is the only one currently available for predicting hydro-
carbon emissions from homogeneous charge engines.
Pollutant formation in heterogeneous combustion such as that Occur-
ring in the TCCS is considerably more obscure. Spatial and temporal
variations in burning stoichiometry cause rapid fluctuations in reaction
rates and species concentrations. In order to gain a more complete under-
standing of the reactions taking place, we would like to be able to do
two things: (1) to follow a small parcel of fuel from the time it leaves
the fuel injector until the time it passes out the exhaust port, noting
in detail its participation in combustion reactions; (2) to be able to
sit at a point in the combustion chamber and observe the reactions taking
place in the vicinity of that point. Recent studies of NO formation in
A
diesels have adopted the latter approach because of its relative sim-
F451
plicity. Nightingale's1- J study is typical of the work being done. He
utilized an inwardly opening sampling valve which could be inserted to
144
-------
various depths at a number of positions in a variety of combustion chamber
configurations. His results include measurements of hydrocarbon concen-
tration as well as NOX, CO and C02 and he notes variations of two orders
of magnitude in HC concentration between sampling points in line with and
far removed from the fuel spray. However, hydrocarbons are only of pe-
ripheral concern to the study and little comment is devoted to the subject.
It is not clear whether he recognized the possibility of a quench layer
forming over the sampling valve tip in the interim between sampling points
and how such a quench layer might affect the sample composition. Ben-
nethum, et al ^ -* have indicated that as much as 30% of the mass in a
given sample could come from the boundary layer formed at the valve tip.
F471
Rhee, -J in reviewing available sampling techniques, recognized this draw-
back of intermittent sampling devices and developed a design permitting
continuous flow of combustion products with the capability of snatching
part of this flow for sampling purposes at desired points in the cycle.
It would appear that such a continuous flow sampling device, with
the additional requirement that it be capable of insertion into the spe-
cific regions mentioned previously, could be used to determine spatial
and temporal variations in hydrocarbon concentration in the TCCS. Such
information would assist in establishing the existence and quantifying
the influence of the various hydrocarbon formation mechanisms ascribed to
the TCCS.
145
-------
REFERENCES
1. E.M. Barber, J. B. Malin and J. J. Mikita, "The Elimination of
Combustion Knock." J. Franklin Institute, Vol. 241, No. 4 (1946),
pp. 275-298.
2. E. M. Barber, B. Reynolds and W. T. Tierney, "Elimination of
Combustion Knock -- Texaco Combustion Process." SAE Quarterly
Transactions, Vol. 5, No. 1 (1951), pp. 26-42.
3. C. W. Davis, E. M. Barber and E. Mitchell, "Fuel Injection and
Positive Ignition — A Basis for Improved Efficiency and Economy."
SAE Transactions. Vol. 69 (1961), pp. 120-134.
4. W. A. Summerson and E. Mitchell, "Small Military Engine-Stratified
Charge Combustion Yields Multifuel Capability." Automotive In-
dustries. Vol. 142, No. 11 (1967), pp. 71-76.
5. E. Mitchell, J. M. Cobb and R. A. Frost, "Design and Evaluation
of a Stratified Charge Multifuel Military Engine." SAE. Trans-
actions. Vol. 77 (1968), pp. 118-131.
6. E. Mitchell, M. Alperstein, J. M. Cobb and C. H. Faist, "A
Stratified Charge Multifuel Military Engine — A Progress Report."
Paper 720051 Presented at the SAE Automotive Engineering Congress,
Detroit, January 1972.
7. M. Alperstein, 6. H. Schafer and F. J. Vi11 forth, III, "Texaco's
Stratified Charge Engine -- Multifuel, Efficient, Clean, and
Practical." Paper 740563 Presented at the SAE Southern Cali-
fornia Section Meeting, May 1974.
8. R. E. Canup, "The Texaco Ignition System -- A New Concept for
Automotive Engines." Paper 750347 Presented at the SAE Auto-
motive Engineering Congress, Detroit, February 1975.
9. B. C. Jain, J. C. Keck and J. M. Rife, "A Performance Model for
the Texaco Controlled Combustion, Stratified Charge Engine."
Paper 760116 Presented at the SAE Automotive Engineering Congress,
Detroit, February 1976.
10. M. L. Lazarewicz, "Performance of Texaco Stratified Charge Engine
on Gasoline." M.S. Thesis, M.I.T., August 1975.
11. F. Nagao, T. Kobayakawa and F. Adachi, "Stationary Flame Front
in a Swirl Combustion Chamber." Bulletin of JSME. Vol. 1, No. 2
(1958), pp. 150-155.
12. J. B. Heywood, "Pollutant Formation and Control in Spark Ignition
Engines." 15th Symposium (International) on Combustion, Tokyo,
Japan, August 1974.
146
-------
13. M. Martin, "Photographic Study of Stratified Combustion Using a
Rapid Compression Machine." M.S. Thesis, M.I.T., January 1975.
14. V. W. Wong, "A Photographic and Performance Study of Stratified
Combustion Using a Rapid Compression Machine." M.S. Thesis,
M.I.T., June 1976.
15. J. Rife and J. B. Heywood, "Photographic and Performance Studies
of Diesel Combustion with a Rapid Compression Machine." Paper
740948 Presented at the SAE Automobile Engineering Meeting,
Toronto, Canada, October 1974.
16. W. M. Scott, "Looking in on Diesel Combustion." SAE Paper
SP-345. December 1968.
17. J. F. Alcock and W. M. Scott, "Some More Light on Diesel Combus-
tion." Proceedings, Automobile Division, Inst. of Mech. Engineers,
1962-3, No. 5.
18. W. T. Lyn and E. Valdmanis, "The Application of High Speed
Schlieren Photography to Diesel Combustion Research." The
Journal of Photographic Science, Vol. 10 (1962), pp. 74-82.
19. R. Watts and W. M. Scott, "Air Motion and Fuel Distribution in
High Speed Direct Injection Diesel Engines." Inst. of Mech.
Engineers, Symposium on Diesel Engine Combustion, April 1970,
Paper No. 17.
20. J. C. Dent and J. A. Derham, "Air Motion in a Four-Stroke Direct
Injection Diesel Engine." Proc. Inst. of Mech. Engineers, Vol.
188, No. 21 (1974).
21. M. M. El Wakil, P. S. Myers and 0. A. Uyehara, "Fuel Vaporization
and Ignition Lag in Diesel Combustion." SAE Transactions, Vol.
64 (1956), pp. 712-729.
22. R. Burt and K. A. Troth, "Penetration and Vaporization of Die-
sel Fuel Sprays." Inst. of Mech. Engineers, Symposium on
Diesel Engine Combustion, April 1970, Paper No. 15.
23. B. C. Jain, J. C. Keck and J. M. Rife, "A Model for Fuel-Air
Mixing in the Texaco Controlled Combustion, Stratified Charge
Engine." M.I.T. Fluid Mechanics Laboratory Report No. 76-1,
January 1976.
24. F. Sass, "Kompressorlose Dieselmaschinen." Berlin, Julius
Springer, 1929.
25. H. Hiroyasu and T. Kadota, "Droplet Size Distribution in Die-
sel Engines." SAE Paper No. 740150.
26. G. L. Borman and J. H. Johnson, "Unsteady Vaporization Histories
and Trajectories of Fuel Droplets Injected into Swirling Air."
Paper 598C Presented at the SAE National Power Plant Meeting,
Philadelphia, Pa., November 1962.
147
-------
27. I. M. Khan and C. H. T. Wang, "Factors Affecting Emissions of
Smoke and Gaseous Pollutants from D.I. Diesel Engines." Inst.
of Mech. Engineers, Symposium on Air Pollution Control in Tran-
sient Engines, November 1971, Paper No. C151.
28. N. A, Henein, "Combustion and Emission Formation in Fuel Sprays
Injected in Swirling Air." Paper 710220 Presented at the SAE
Automotive Engineering Congress, Detroit, January 1971.
29. Private Communication with M. Alperstein, Texaco, Inc.
30. A. Williams, "Combustion of Droplets of Liquid Fuels: A Review."
Combustion and Flame, Vol. 21 (1973), pp. 1-31.
31. G. J. Barnes, "Relation of Lean Combustion Limits in Diesel
Engines to Exhaust Odor Intensity." SAE Paper No. 680445.
32. I. M. Khan, G. Greeves and D. M. Probert, "Prediction of Soot
and Nitric Oxide Concentrations in Diesel Engines." Inst. of
Mech. Engineers, Symposium on Air Pollution Control in Trans-
port Engines, November 1971, Paper No. C142.
33. N. C. Blizzard and J. C. Keck, "Experimental and Theoretical
Investigation of Turbulent Burning Model for Internal Combus-
tion Engines." SAE Paper No. 740191.
34. W. A. Daniel and J. T. Wentworth, "Exhaust Gas Hydrocarbons —
Genesis and Exodus." Paper No. 486B Presented at the SAE
National Automobile Week, Detroit, March 1962.
35. R. J. Tabaczynski, J. B. Heywood and J. C. Keck, "Time-Resolved
Measurements of Hydrocarbon Mass Flowrate in the Exhaust of a
Spark Ignition Engine." Paper No. 720112 Presented at the SAE
Automotive Engineering Congress, Detroit, January 1972.
36. V. S. Yumlu and A. W, Carey, Jr., "Exhaust Emission Charac-
teristics of Four-Stroke, Direct Injection, Compression Igni-
tion Engines." Paper 680420 Presented at the SAE Mid-Year
Meeting, Detroit, May 1968.
37. M. W. Jackson, W. M. Wiese and J. T. Wentworth, "The Influence
of Air-Fuel Ratio, Spark Timing and Combustion Chamber Deposits
on Exhaust Hydrocarbon Emissions." Paper 486A Presented at SAE
National Automobile Week, Detroit, March 1962.
38. T. A. Huls, P. S. Myers and 0. A. Uyehara, "Spark Ignition
Engine Operation and Design for Minimum Exhaust Emission."
Paper 660405 Presented at the SAE Mid-Year Meeting, Detroit,
June 1966.
39. R. W. Hum, "Air Pollution and the Compression-Ignition Engine."
12th Symposium (International) on Combustion (1969), pp. 677-687,
148
-------
40. J. H. Johnson, E. J. Sienkki and 0. F. Zeck, "A Flame lonization
Technique for Measuring Total Hydrocarbons in Diesel Exhaust."
Paper 680419 Presented at the SAE Mid-Year Meeting, Detroit,
May 1968.
41. W. A. Daniel, "Flame Quenching at the Walls of an Internal Com-
bustion Engine." 6th Symposium (International) on Combustion
(1956), pp. 886-894.
42. J. N. Shinn and D. R. Olson, "Some Factors Affecting Unburned
Hydrocarbons in Engine Combustion Products." SAE Technical
Progress Series, Vol. 6, "Vehicle Emissions" (1964), pp. 137-145.
43. R. M. Fristrom and A. A. Westenberg, Flame Structure , McGraw-
Hill (1965).
44. W. A. Daniel, "Why Engine Variables Affect Exhaust Hydrocarbon
Emission." Paper 700108 Presented at the SAE Automotive Engineer-
ing Congress, Detroit, January 1970.
45. D. R. Nightingale, "A Fundamental Investigation into the Problem
of NO Formation in Diesel Engines." Paper 750848 Presented at
the SAE Off-Highway Vehicle Meeting, Milwaukee, September 1975.
46. J. E. Bennethum, J. N. Mattavi and R. R. Toepel, "Diesel Combus-
tion Chamber Sampling -- Hardware, Procedures and Data Interpre-
tation." Paper 750849 Presented at the SAE Off-Highway Vehicle
Meeting, Milwaukee, September 1975.
47. K. T. Rhee, G. L. Borman, 0. A. Uyehara and P. S. Myers, "A Con-
tinuous-Flow Gas Sampling System for Engines." Combustion Science
and Technology, Vol. 12 (1976), pp. 105-109.
149
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse In-fore completing}
1. REPORT NO.
EPA-460/3-77-006
2.
4. TITLE AND SUBTITLE
Emission Formation in Heterogeneous Combustion
3. RECIPIENT'S ACCESSION"NO.
5. REPORT DATE
February, 1977
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
G.L. Borman, P.S. Myers, O.A. Uyehara, L. Evers,
. Ingham. D. Jaasma
8, PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Department of Mechanical Engineering
University of Wisconsin
Madison, Wisconsin 53706
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
R-803858-01-1
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Protection Agency
Emission Control Technology Division
2565 Plymouth Rd.
Ann Arbor, Michigan 48105
13. TYPE OF REPORT AND PERIOD COVERED
11/10/76 - 2/9/77
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16, ABSTRACT
Three research projects are reported under the grant. The first is an investi-
gation of a stratified-charge engine concept in which spark ignited combustion in an
engine with a homogeneous rich charge is completed and then air is injected during
the expansion stroke giving a leaner overall fuel-air ratio. The study showed sub-
stantial reduction of nitric oxides without increasing other emissions. Combustion
efficiency was not increased and, because substantial work was needed to supply the
compressed air, the engine efficiency was decreased. The second project is an inves-
tigation of nitrogen oxides produced by burning of liquid normal heptane from a fuel
wetted porous cylinder in a cross flow of air. Variation of free stream air velocity
and cylinder diameter showed the moles of nitric oxide per mole of fuel burned to be
a weak function of Reynolds number. Soot produced by the flame and collected down-
stream has been identified as giving off significant amounts of nitric oxide indica-
ting a carbon, nitric oxide interaction in the flame envelope. The third project
consisted of a study of the part load operation of the Newhall divided chamber engine
previously developed at U.W., Madison, an emissions and fuel economy evaluation of
this engine relative to other engines and initiation of a study of the formation of
hydrocarbons in a Texaco engine by utilization of an in-cylinder sampling technique.
Reasons for abandonment of the divided chamber engine were its higher hydrocarbons and
lower fuel economy relative to other engines and its sensitivity to knock.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
COSATI Irield/Group
exhaust emissions, combustion products,
internal combustion engines, nitrogen
oxides, valves, burning rate
stratified-charge
engines
0702, 1311,
2102, 2107,
2111
18. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
20. SECURITY CLASS (Thli page)
Unclassified
21. NO. OF PAGES
tPA Form 2220-1 (9-7J)
------- |