EPA-460/3-74-002a
FOUNDATION FOR MODELING NOX
AND SMOKE FORMATION
IN DIESEL FLAMES
FINAL REPORT FOR PHASE I
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Water Programs
Office of Mobile Source Air Pollution Control
Emission Control Technology Division
Ann Arbor, Michigan 48105
-------
EPA-460/3-74-002a
FOUNDATION FOR MODELING NOX
AND SMOKE FORMATION
IN DIESEL FLAMES
FINAL REPORT FOR PHASE I
Prepared by
R. P. Wilson, Jr., C. H. Waldman, andL. J. Muzio
ULTRASYSTEMS, INC.
2400 Michelson Drive
Irvine, California 92664
Contract No. 68-01-0436
EPA Project Officers:
J. L. Bascunana andG. D. Kittredge
Prepared for
COORDINATING RESEARCH COUNCIL INC .
30 Rockefeller Plaza
New York, NY 10020
APRAC Project CAPE 20-71
and
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Water Programs
Office of Mobile Source Air Pollution Control
Emission Control Technology Division
Ann Arbor, Michigan 48105
January 1974
-------
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 - as supplies permit - from
the Air Pollution Technical Information Center, Environmental Protec-
tion Agency, Research Triangle Park., North Carolina 27711, or from the
National Technical Information Service, 5285 Port Royal Road, Spring-
field, Virginia 22151.
This report was furnished to the Environmental Protection Agency by ULTRA-
SYSTEMS, INC., Irvine, California, in fulfillment of Contract No. 68-01-
0436. The contents of this report are reproduced herein as received from
ULTRASYSTEMS, INC. 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 con-
sidered as an endorsement by the Environmental Protection Agency.
Publication No. EPA-460/3-74-002a
ti
-------
ACKNOWLEDGEMENTS
The eventual usefulness of an applied research study hinges on
whether efforts are focussed on questions of real importance to industry
and society. The EPA and the CAPE 20-71 Steering Committee of the Coordi-
nating Research Council, having conceived the original project outline, have
aligned this study to be "on target" through numerous critical reviews and
suggestions. The following individuals have been of service in this regard:
J. L. Bascunana Environmental Protection Agency
T. C. Belian Coordinating Research Council
J. E. Bennethum General Motors Research Laboratory
W. L. Brown Caterpillar Tractor
F. I. Hills Mobil R&D (Committee Chairman)
J. C. Hoelzer International Harvester
G. D. Kittredge Environmental Protection Agency
D. F. Merrion General Motors
P. C. Meurer International Harvester
J. M. Perez Caterpillar Tractor
S. M. Shahed Cummins Engine '
A. V. Wilson Cummins Engine
A. E. Zengel Coordinating Research Council
In addition, the following fluid physics specialists and advisors
from the academic community have shared their experience and technical
expertise during Phase I of the study
G. L. Borman University of Wisconsin
P. S. Myers University of Wisconsin
J. Shipinski John Deere Tractor
F. A. Williams University of California, San Diego
Finally, the support and competent engineering of the Advanced
Products Division of White Motors Corporation, especially the following
persons, is warmly acknowledged
W. F. Dittman President
E. B. Muir Product Development
F. A. Pellicciotti Engineer
J. Salyer Technician
H. LaHomme Design
iii
-------
Description of Project Team
The study is being conducted by the Applied Combustion Research
Group of Ultra systems, Inc. During Phase I, staff members participating
were as follows:
R. P. Wilson, Jr. Program Management, Single Cylinder
Emissions, Photography, Spectroscopy,
Model Evaluation, New Mechanistic Model
C. H. Waldman Model Evaluation, Diffusion Flame
Studies, New Mechanistic Model
I. J. Muzio Spectroscopy, New Mechanistic Model
In addition, T. J. Tyson provided substantial technical guidance in modeling
cylinder fluid dynamics and, as Division Vice President, garnered vital
resources (e.g. instrumentation) for the program. Technical support was
provided by E. Madsen, C. McComis, and C. Bradley. Typing and artwork
were by G. Cresswell and J. Stewart, respectively.
iv
-------
TABLE OF CONTENTS
Section Page
I. SYNOPSIS AND RECOMMENDATIONS 1
II. ORIENTATION 3
A. Practical Reasons for an Attempt to Characterize
NO and Smoke Formation in Diesel Flames 3
B. Characterizing NO and Soot Formation will Demand
a Clear Picture of the Diesel Flame 4
C. Program Strategy to Identify NO and Smoke
Mechanisms ,X 5
III. EVIDENCE FROM DIESEL EMISSIONS BEHAVIOR 7
A. Single-Cylinder Experimental Technique 8
B. Observed Emissions 11
C. Summary of Emissions Evidence 45
IV. AN INTERPRETATION OF DIESEL COMBUSTION AND
POLLUTANT FORMATION . 48
A. Mixing and Combustion Mechanisms 50
B. Outline of an Emissions Model 53
C. Flame Studies Needed 64
D. Droplet Diffusion Flame as a NOx-Source 67
V. ASSESSMENT OF EXISTING MODELS 74
A. The NREC Model 77
B. The CAV Model 84
C. The Cummins Model 89
VI. DIESEL FLAME STUDIES 93
A. Spectroscopic Observations 93
B. Diesel Flame Photography 102
NOMENCLATURE 104
REFERENCES 107
APPENDICES
A. Essentials of NOX and Smoke Formation 111
B. Single-Cylinder Experimental Technique 121
C. Complete Data from Single Cylinder Emission Tests 133
D. Compilation of Published Diesel Emissions Data 143
E. Equilibrium Analysis of Diffusion Flame Structure 149
F. Unsteady Diffusion as a Factor in Droplet Combustion 155
G. Experience in the Use of Windowed Combustion Chambers 169
-------
I. SYNOPSIS AND RECOMMENDATIONS
Emissions reduction fordiesel engines can be approached either
directly by exploratory testing or indirectly by using a mathematical model
to predict low-emission modifications. The empirical approach suffers from
excessive costs and limited extrapolations from given engines; at the same
time, theoretical predictions are too uncertain to use alone because of lack
of understanding of the diesel combustion mechanism. The two approaches
are complementary and should be pursued together. Short term improvements
can be attained with engine testing. However, the premise of the study
reported herein is that it is cost effective in the long term to generate a
mathematical model which embodies the key combustion mechanisms well
enough to guide the development and design of engines. In Phase I we have
established a foundation for a mechanistic model by four activities:
• Emissions data was generated and correlated with
changes in engine parameters.
* Existing models were critically assessed.
• A mechanistic model of heat release was outlined
and an analysis of key questions was begun.
• Diesel flame measurements were conceived which
can resolve modeling issues and thereby insure
that model development will be cost effective.
Exhaust measurements of NO, soot, and hydrocarbons were made on
3
a 2340-cm displacement, single-cylinder diesel engine operated over a range
of speed, fuel-air ratio, and timing. In addition to confirmation of the well
known effects of A/F and timing, the following parameters were found to change
NO emissions by 40% or more (with corresponding soot changes):
Ji
Engine Geometry State of the Intake Mixing Parameters
• Divided chamber • EGR • Fuel orifice size
• Prechamber volume ratio • Water injection • Air swirl
• Compression ratio
-------
The emissions behavior, as well as published movies and apparent-
heat-release studies, can be interpreted by assuming the following mecha-
nisms of diesel combustion: .heat release occurs at flame surfaces at a rate
which is limited by molecular diffusion. Although there is no direct evidence,
these diffusion flames are taken to be envelope flames around droplets which
are entrained by swirling air from each fuel jet. Thus two scales of mixing arise:
• Microscale of molecular diffusion at the flame
surface, governing the pollutant formation,,
• Macroscale of turbulent entrainment, governing
the rate of heat release.
3
The value of the droplet diffusion flame is not in an improved heat-release
prediction, nor do we claim it describes the actual burning process. Rather,
it is a useful artifice to describe in detail the high temperature diffusion flames
which give rise to nitric oxide and soot. The nature of the diffusion flame (wake-
type, ensemble-type, or single-droplet-type) is not known. But regardless of
type, the flame itself is expected to take on a profile universal to any type.
Three existing models developed by Northern Research and Engineer-
ing Company, CAV, Ltd. (Lucas Company), and Cummins were critically
reviewed based on treatment of physical heat release mechanisms, ability to
predict emissions behavior, and the need to:readjust empirical coefficients.
These models do not explicitly treat the fluid-physics of air motion, fuel spray,
ignition delay, or detailed diffusion flames. When these phenomena are omitted
or simulated with arbitrary phenomenological relations, not only is the model's
range of applicability limited (e.g., fuel orifice sensitivity is not predicted),
but coefficients must be laboriously and empirically fit to each engine.
The following recommendations are made:
Recommendation; 1: A mathematical model of NOX and smoke
production in diesel flames should be developed with
mechanistic, semi-geometric treatments,of the macro-
scale mixing (air swirl-=and fuel spray) and the molecular
mixing (diffusion flame profiles). An outline for such a
model is presented in Section IV.B.
Recommendation 2: To resolve key questions about mecha-
nisms, measurements of air motion, fuel dispersion,
temperature, and NO in the diesel combustion environ-
ment should be aggressively pursued. Preliminary
attempts using UV spectroscopy and photography of a
windowed engine are described in Section VI0
-------
II. ORIENTATION
A. PRACTICAL REASONS FOR AN ATTEMPT TO CHARACTERIZE
NO AND SMOKE FORMATION IN DIESEL FLAMES
The modem diesel engine enjoys widespread use in heavy duty
transport, due in large part to its simplicity and relatively low fuel con-
sumption. The high flame temperatures and mixture heterogeneity which
produce the advantages of the compression ignition engine also give rise
to the side effects which concern us in this study—NO and smoke forma-
tion. An investigation was undertaken to attempt to better understand the
diesel flame, so that the formation of these pollutants can be controlled—
without compromising the basic advantages of the diesel engine by complex
add-on gadgetry or higher fuel consumption.
Through a number of past studies, the industry has determined cer-
tain simple measures which will reduce nitric oxide and smoke emissions
from specific engines/ foremost among them retarded timing and aftercooled
turbocharging. Other measures such as EGR, water injection, fuel injection
variations/ and prechamber designs have also shown emissions control
potential but require further research and development. If engines are to be
optimized/ it will be helpful to know why these methods work. In the long
term/ the industry may be considering advanced designs for the diesel engine;
designs which may be so different from current engines that existing emissions
data cannot be extrapolated with reasonable confidence. At this point, a
predictive framework of understanding of diesel-generated NO and soot will
be valuable. In the meantime, model predictions could be used to improve
and guide the "cut-and-try" emissions testing which is currently being
carried out on major production engines. When smoke and fuel consumption
limits are reached on a given engine, the test engineer could use a table of
predicted influence coefficients to assist the intuition in selecting promising
test set-ups. It is the intent of this study to forge such a tool by setting
-------
down in a mathematical model all we know or can learn about the complex,
coupled flame phenomena in which NO and smoke are produced: spray com-
bustion, radiation, and pollutant kinetics.
B. CHARACTERIZING "NO AND SOOT FORMATION WILL DEMAND
A CLEAR PICTURE OF THE DIESEL FLAME
NO and soot formation are activated by combinations of temperature
and species composition which fortunately are not widespread in a diesel flame
(for a complete discussion, see Appendix A). Therefore, in a given diesel
combustion chamber, there are only certain zones which are actively produc-
ing NO or soot during certain crank-angle intervals. An accurate description
of NO and soot formation requires detailed temperature and species profiles
in these local regions. Details such as temperature and O-atom gradients,
supercritical envelope or wake flames, turbulent eddy size, and spray droplet
size distribution may be required.
None of these actual happenings have been measured or described
for the flame of any production diesel engine. To date, it has simply not
been worth the trouble. In order to simulate or predict performance, it
sufficed to describe the heat release either with empirical formulas [after Lyn
(1957) or Shipinski (1968)], or to assume burning of homogeneous regions of
prescribed fuel/air ratio. Such "global" models paid a penalty for whitewash-
ing the physical mechanisms with prescribed empiricism: A new set of
empirical constants for each new engine or fuel system had to be developed
to make the computer simulation work. Only Shipinski et al. (1968) and
Khan and Wang (1971) have attempted more universally applicable treatments
of diesel combustion based on the diffusion flame. These models are evaluated
in Section V. The basic limitation to formulating a more applicable model is
lack of measurements on the diesel flame. The available evidence essentially
consists of high-speed movies, chamber pressure (heat release) recordings,
and heat transfer measurements.
-------
Substantial experimental and theoretical effort is required to go
deeper into the details of diesel spray combustion. If the empiricism is to
be replaced, ultimately this effort must be made. The current program is
oriented toward this goal.
C. PROGRAM STRATEGY TO IDENTIFY NO AND SMOKE MECHANISMS
X
The program is structured to focus on the missing diesel flame
information needed to predict NO and soot formation. Tasks are grouped
into five main activities as summarized below in Table 1, along with the
approximate percent completion as of this writing. Note that the diesel
flame studies and analytical modeling are at a very preliminary stage. The
Phase I report should be read as an interim report; except for the emissions
data (Chapter III), a reasonable number of the interpretive notions about
diesel combustion in this report are likely to be revised or even abandoned
as new data becomes available.
-------
Table 1
FIVE MAJOR PROGRAM ELEMENTS
DIESEL FLAME PHENOMENA
Fuel/Air Mixing
• Fuel spray injection
• Air swirl and turbulence
Ignition and Spray
Combustion
Pollutant Formation
Emissions
EVALUATION AND ASSESSMENT
(100% complete, see Section V)
The current state-of-the-art of
NOx and smoke modeling for
dlesel flames Is defined and
critically compared to Industry
criteria for a reasonably useful
model .
Review treatments of mixing
In current models
i
Review treatments of Ignition
and combustion in current
models
1
Review treatments of NO and
soot In current models
• Observed emissions correlated
against emission levels pre-
dicted by the existing CAV and
NREC models
• Discrepancies identified and
corresponding model weak-
nesses diagnosed
DIESEL FLAME STUDIES
(25% complete, see Section VI)
Extensive flame diagnostic
studies are being conducted In
order to provide conceptual
clues for model development.
Fuel spray studies
Air swirl and turbulence
measurements by anemo-
metry
High speed movies
Measure flame temperature
by IR spectroscopy
High speed movies
Diffusion-flame experiments
Measure chamber pressure
Measure NO and other key
species
(a) spectroscopy
(b) sampling valve
EMISSIONS DATA
(100% complete, see Section III)
Measure emissions of single-
cylinder engine subject to these
test variables: .
• 'Air State (turbocharge, EGR,
water, air temperature)
• Fuel (number, size of orifices,
pilot) and Air Swirl
• Design (CR, prechamber)
• Operation (RPM, load, timing)
Compare with-published data
ANALYSIS AND MODELING
(10% complete, see Section IV)
The following analyses will be
synthesized into cycle thermo-
dynamics for an unproved com-
bustion and emission model.
Analyze spray breakup by
swirling air crossflow and
air Impingement
Analyze ignition and premixed
burning
Analyze droplet diffusion flame
structure
Describe diffusion-controlled.
heat release
Analyze NO and soot kinetics
EMISSIONS TESTS OF
PRODUCTION ENGINES
TO CHECKOUT MODEL
Check out using single-
cylinder data
Multicylinder tests performed
Corresponding model predic-
tions made to Insure applica-
bility to production engines
-------
III. EVIDENCE FROM DIESEL EMISSIONS BEHAVIOR
CHAPTER SUMMARY
Exhaust measurements of NO, soot, and hydrocarbons were made on
3
a 2340-cm displacement, single-cylinder diesel engine operated over a range
of speed, fuel-air ratio, and timing. In addition to confirmation of the well
known effects of A/F and timing, the following parameters were found to
change NO emissions by 40% or more (with corresponding soot changes):
X
Engine Geometry State of the Intake Mixing Parameters
• Divided chamber • EGR • Fuel orifice size
• Prechamber volume ratio • Water injection • Air swirl
• Compression ratio
A preliminary analysis of this emission data suggests that the fol-
lowing phenomena require further study to achieve a better understanding of
diesel-flame generated NO and smoke:
Local Diffusion Flames: mixing-controlled combustion may
take many forms: wake burning as studied by Natarajan
and Brzustowski (1970), envelope flames classically
studied by Godsave (1950), or spray combustion as
investigated by McCreath and Chigier (1972) .
Swirl Effects: angular motion and entrainment of fuel, radial
stratification due to centrifugal effects, and turbulence
levels.
Fuel Spray Details: penetration, drop size distribution,
possible wall impingement, and entrainment of air.
Time-dependent Phenomena: must be precisely character-
ized and overlayed with an exactly (+.5°CA) specified
piston motion and fuel delivery schedule. Key pheno-
mena such as ignition, mixing rates, burning rates,
evaporation rates, and heat transfer rates must be
characterized as precisely as possible.
Prechamber Phenomena: such as fluid transfer between the
two chambers, including phase lags and heat transfer.
-------
III. EVIDENCE FROM DIESEL EMISSIONS BEHAVIOR
A. SINGLE-CYLINDER EXPERIMENTAL TECHNIQUE
3 2
A 2340-cm (143 in ) displacement research engine was constructed
for the experimental program. The engine and test procedures are described
in detail in Appendix B. The selected chamber geometry is a 5-1/2" bore,
6" stroke, with 12" connecting rod to minimize piston slap at TDC. Two
cylinder head configurations were fabricated—a direct injection head and
a prechamber version, hereafter referred to as the DI engine and PC engine
(Figure 1). Piston and bowl geometry are given in Figure 1 for both types of
heads. Changes in piston geometry gave compression ratios of 20:1, 17:1,
and 14:1 for the DI configuration; changes in piston caps and prechamber
gave four combinations of PC ratio and compression ratio, as listed in
Figure 1.
The M.A.N. and Lanova combustion chambers were also tested in
order to examine emissions sensitivity to substantial changes in provisions
for fuel/air mixing. These systems were tested as standard multicylinder
engines of comparable displacement. Sketches of the head design appear in
Figure 1.
The nozzle for the DI head was a Roosa pencil injector, popularly used
in farm tractors, nominally with six .010" orifices at 160 deg. cone angle.
2
For the PC head a single pintle orifice at .37 mm size with 12 deg. cone
angle was standard.
By means of changes in the fuel line length, injectors, cam profiles,
fuel valve opening pressure, and plunger diameter, it was possible to study
the following variations:
No. of orifices: • 4, 6, 8
Orifice size: .008, .010, .012, .014 inches
Rate of fuel injection: 3 to 8 mm /°CA
Pilot injection: 10 to 20% of fuel injected at -40°CA
Cone angle: 120 deg. vs. 160 deg. for DI
8 to 12 deg. for PC
Air was taken from the laboratory compressed air supply and heated
(or cooled) after the filter and flowmeter. A throttle valve in the exhaust line
was used in some tests to build up exhaust pressure to simulate turbocharging.
-------
Figure 1
CYLINDER HEADS
PRECHAMBER
H-2.00 in.—*|
d/D = 0.84
. D = 5.50 in.
_CR . tot h(ln.)
Plug
(Projection)
16°4S'
16°45' 1.113
0.690
17:1 25% 0.395
19.4:1 15% 0.395 .. __
17:1 35% 0.345 14°44' 6.398
19.4:1 25% 0.345 14°94' 0.826
DIRECT INJECTION
H - Hold down
F - Fuel nozzle
d/D = 0.59
D = 5.50 in.
CR P _ hjln.) R(in) Serial No.
17:1 119° 0.877 ^500 TE 02650
14:1 100 1.125 .500 TE 02697
20:1 140 0.715 .500 TE 02696
M.A.N.
LAN OVA
ij^rpSs
rLT"
-P| 3ECTIONA-R
^gp^g
-------
Water mist was available at the intake, and an EGR system was available
to supply measured amounts of exhaust gas to the intake. In this manner,
the state of the intake air could be controlled and varied a s follows:
m
nH20/rnf = 0 to 1.0
m M ^ = 0 to 30%
egr tot
30 to 60" Hg
100 to 200 F
Water injection:
EGR:
Air pressure:
Air temperature:
A masked valve was used to generate swirl (see Figure 2); rotation of the valve
controlled both the sense and degree of swirl. In addition, for each "top-end",
baseline tests were run over a 20-point matrix of speed, load, and timing.
Variables were changed one at a time in order to clarify the NO
and smoke behavior. Runs with two effects which compensate or amplify
one another would be appropriate for a low-emissions development program,
but not for this study of mechanisms. In each case, extreme levels of the
variables (for example, 30% EGR; a range of X5 in swirl, etc.) were selected
in order to bring to the surface whatever NO and smoke changes were occurring.
CW Viewed from
Top of Engine
adjustable
angle of
mask position
Mask subtends 90
SECTION A-A
r.
Figure 2
PROVISION FOR AIR SWIRL
10
-------
B. OBSERVED EMISSIONS
In this section we report how diesel emissions responded as
operating parameters were changed. A substantial bank of data has been
acquired in a closely controlled, highly instrumented manner. Complete -
data records are supplied as Appendix C to this report. This data bank
serves the three functions of (1) providing diverse emissions data against
which model predictions can be compared, (2) stimulating mechanistic
thinking about NO behavior which the model must simulate, and (3) sug-
j£
gesting the most interesting emissions behavior for planned diesel flame
studies.
The form of the graphical data presentation is designed to faci-
litate comparison with model predictions. All values are normalized by
baseline emissions and performance data; what appears on the graph is the
fractional change in NO (lb/1000 Ib fuel), soot (% opacity) or indicated
J\,
specific fuel consumption (Ib/IHP-hr) due to the test variable. This
reflects our predisposition that it is more realistic to expect relative trends
or influence coefficients from a model than absolute predictions. In addi-
tion, values of brake specific NO (g/BHP-hr)* are given to permit direct
.X.
comparison with multicylinder engines.
Results from earlier studies generally corroborate our findings.
A compilation of published emission data may be found in Appendix D.
This data listing includes the studies of Abthoff and Luther (1969), Landen
(1963), Khan and Wang (1971), Schmidt et al. (1966), Marshall and Fleming
(1971), Bascom et al. (1971), Hames et al. (1971), Pischinger and
Cartellieri (1972), Parker and Walker (1972), Shahed, Chiu and Yumlu(1973),
Walder (1973), and McConnell (1963).
Even though the combustion chamber was treated as a "black box",
these measurements of external behavior can provide valuable clues about
*A slight FMEP correction was applied to the BMEP in order to simulate the
lower specific mechanical friction experienced in a multicylinder engine.
This correction is minor (2 to 7 FMEP, depending on engine configuration).
11
-------
the flame processes which produce NO and smoke. Possible ihterpreta-
.X
tions are hypothesized in preparation for synthesizing these interpretations
into a model of diesel-generated pollutants.
1.
Effect of Different Combustion Chambers
The behavior at baseline RPM and air/fuel ratio of the single
cylinder engine is shown in Table 6. Results for the MAN and Lanova
multicylinder engines are listed for comparison. It cannot be over-
stressed that none of the engines, except the MAN and Lanova engines,
were tuned or adjusted into optimum performance; they represent single
cylinder designs which were simply set up and run. In comparing PC engines
to DI engines, it should be noted that the fuel system differed.
Table 6
COMPARATIVE BEHAVIOR OF ENGINES TESTED
(Baseline: 1500 RPM; A/F = 32, N.A.)
Volume Fuel Timing
NO-
'x
Soot
Type Compression Ratio Nozzle (°BTDC) (lb/1000 Ib fuel) (% Opac) BSFC
DI
DI
PC
PC
PC
PC
MAN
Lanova
17:1
14:1
17:1
17:1
19.4:1
19.4:1
17:1
17:1
--
—
25%
35%
25%
15%
--
—
6 -hole
6 -hole
Pintle
Pintle
Pintle
Pintle
Not avail .
Not avail .
-20
-20
- 6
- 6
- 6
- 6
-21
-20
92
46
26
37
17
11
51
51
.1
.7
.4
.0,
.2
.7
.2
.8
5
5
1
1
4
24
1
2
.3
.3
.8
.0
.0
.0
.2
.0
.397
.412
.397
.409
.428
.490
.408
.358
A significant variation in NO can be seen between engine configura-
J\.
tions. The main trends are as follows:
• The prechamber generally gives lower NO with higher
fuel consumption.
• Increased volume ratio produces increased NO but
reduced soot.
• As compression ratio is increased, nitric oxide emissions
increase.
12
-------
a. Effect of Prechamber
Prechamber engines exhibit markedly lower NO than DI engines. A
,/t
popular hypothesis is that premixed burning occurs at a lower temperature
because the overall mixture is richer in the prechamber. The merit of this hypo-
thesis is questionable since there are some indications that the combustion is
diffusion limited. In a diffusion flame, the overall A/F ratio does not affect
the local flame temperature but only affects the final mean temperature of the
product mixture. An alternative hypothesis is the staged-combustion argument:
following partial combustion in the prechamber, the hot gases lose heat as
they are pumped out into the main chamber (for this reason, PC engines have
relatively large radiators). Thus the heat release is staged, and the inter-
vening heat transfer presumably reduces peak temperatures and NO .
J^
b. Effect of Volume Ratio
The effects of prechamber volume ratio are presented in detail in
Figures 3 and 4. It is plausible that the larger prechamber volumes contain
enough air to burn more of the fuel before expulsion occurs, thus reaching
higher flame temperatures and generating more NO . To corroborate this
^Tfc
hypothesis, we have extended the plot to include the DI engine which is,
in effect, 100% prechamber. The MO emissions for DI and PC engines are
J^
related systematically as shown in Figure 40
c. Effect of Compression Ratio
As to the effect of compression ratio shown in Figures 5 and 6,
clearly more than one mechanism is at work since Figure 5 shows the oppo-
site trend of Figure 6. The higher compression temperatures associated
with an increase in the compression ratio might stimulate thermal NO forma-
tion. Even though NO production is presumably confined to flame fronts
X,
and other local zones, the local peak temperatures associated with these
"active" zones will be boosted along with any increase in the average gas
temperature, increasing NO formation. Increased heat transfer and disso-
ciation will of course attenuate such an effect. Compression ratio also
affects the rate of fuel/air mixing by governing the swirl at TDC when fuel
is injected. If fuel mixes and bums earlier in the cycle, before volume
expansion drops the temperature, NO emissions may increase. Furthermore,
compression ratio may affect ignition delay.
13
-------
2. Effect of Operational Parameters (Load, Speed, and Timing)
a. Effect of Load
The effect of equivalence ratio (<£) upon NO formation supports the
notion that overall mixture ratio is not as relevant a parameter as it would be
in a premixed flame. Comparison of DI and PC engines in Figures 7 and 8
show that NO peaks at ^ 0.3 in both cases, whereas for a premixed system
NO would peak at 4> w .95 . Nitric oxide emissions per unit fuel actually
decrease as <£ approaches stoichiometric for <£ > 0.3/ as shown in Figures 7
and 8. There are at least two possible factors which could cause NO to drop
with equivalence ratio:
(i) The start of injection and rate of injection are fixed.
As the period of injection increases with <£, more
burning occurs later in the expansion cycle, where the
gas temperature has dropped due to expansion and heat
losses, so that MO may be produced at a reduced rate.
(ii) Each successive fuel element introduced is likely to
see an oxidizer which is progressively more diluted
with products. At high loads, the available oxygen
concentration for late-burning fuel is lower and local
flame temperature and NO formation rates may be
lowered accordingly.
b. Effect of Timing
Looking at further evidence, we document in Figure 9 the well-
known effect of retarded timing on emissions from DI engines. For each
19 CA that the start of fuel feed is delayed, i.e., "retarded" from normal,
the exhaust NO level drops about 50%. The effect is most pronounced at
J\
high loads, that is, when the duration of fuel injection is largest. Several
flarne alterations may occur when timing is retarded; their relative importance
on MO is not yet known:
x
(i) Aerodynamic Effect: Late fuel is injected into a different
air-flow environment, probably lower angular momentum
but higher turbulence intensity. Also the piston ledge
* Equivalence ratio is taken as (F/A)/(F/A)stoic. Equivalence ratio was
selected over BMEP or IMEP as the "load" variable because the latter are
inherently dependent variables.
14
-------
becomes accessible at late crank angles so that one
might expect more stratification of the heavier unused
air which is thrown outward away from the fuel.
(ii) Thermal Effect: Fuel burning well into the expansion
stroke presumably enjoys an environment of lower
temperature and pressure (corresponding to lower per-
formance and lower NO production).
(iii) Residence Time Effect: With late injection, the time
for NO-formation in the hot products throughout the
cylinder may be cut short.
(iv) Ignition Delay Effect: Evidence shows that the pre-
paration-to-burn time shortens with injection retard.
The effect of timing for the prechamber engines is shown in
Figure 10. Note the marked minimum in the NO curve at start of injection
"
(about 4 BTDC), which seems to be uniformly characteristic of all four PC
engines. Before commenting on this minimum at -4° BTDC/ let us review
the prechamber combustion phenomena. The fuel is injected into the pre-
chamber, which nominally contains 25% of the air by volume. As heat is
released, the prechamber pressure rises, expelling the products and remain-
ing fuel through an orifice into the main chamber. At low loads (<# < .25),
none of the main chamber air is needed to complete combustion—the NO
x
becomes insensitive to timing as shown by Landen (1963).
For normal operation (.5 < < .7), the sensitivity of NO to timing
is greatest for small prechambers (see Figure 10). In fact, Walder (1973)
reports no timing effect for large prechambers with V /V = 50%; again we
PC
suspect enough air in the prechamber to complete combustion. This suggests
that NO formation in the main chamber is causing the NO changes with timing.
Perhaps prechamber activity is complete by the time volume changes occur
at high crank angle (& > +20°).
The minimum in NO vs. timing seems peculiar to PC engines tested
in the present program [previous studies by Landen (1963), McConnell (1963),
and Eyzat (1967) showed no minimum with retarded timing]. The sharp edged,
rather long (L/D « 3) passageway connecting prechamber to main chamber may
15
-------
be responsible: If large AP's exist between the two chambers without com-
bustion (due to the flow restriction), then the timing of fuel injection and
precharnber ignition would couple with the timing of this natural pumping.
The NO could increase at retarded timings because of a phase lag (between
pumping and ignition) more favorable to expulsion and rapid heat release in
the main chamber. Indeed the NO minimum is seen to be more severe at
19:1 compression ratio than at 17:1, and pumping lag might be more marked
at 19:1.
c . Effect of Speed
From Figure 11, the effect on DI engines of increasing engine speed
is to reduce NO emissions. The increase from peak-torque speed to maximum
speed results in about a 15% penalty in BSFC. One interpretation of this NO
J\
trend is based on fixed burning time: recall that residence time between TDC
and 60° ATDC is inversely related to engine speed, ranging from 15 to 5 milli-
sec as speed ranges from 800 to 2400 RPM. So if heat release requires a
fixed 5 millisec, more burning will occur on the expansion stroke at high
speeds (lower temperature and lower NO). Actually, burning time is reported
to decrease somewhat for higher engine speeds [Shipinski (1968)*], but not
enough to make burning duration a constant crank angle interval.
Other effects of speed must be considered. Higher piston veloci-
ties at high speeds will cause greater air turbulence (earlier burning and
higher MO). Engine breathing-is reduced at high RPM, while residual gas is
increased by the higher valve overlap. Both of these effects might reduce
NO emissions. Fuel injection pressure rises with speed and thus might pro-
duce smaller drops. Finally, the compressed air temperature is higher at
high speeds due to lower heat losses and more pumping work. The precham-
ber NO emissions behave oppositely (increase) as engine speed is increased
A.
(Figure 12). The dip in the curve for the baseline case does not appear for
cases 2 to 6, however, intermediate data at 1800 RPM was not taken for
cases 2 to 6.
*This observation is incompatible with droplet-burning models for quiescent
oxidizers of invariant,composition, but may be consistent with diffusion
flames with an "RPM-sensitive" air environment and drop size distribution.
16
-------
3. Effect of Alterations to the Intake Charge
a. Effect of EGR
Figures 13 and 14 display the effect of EGR on NO , smoke, and
•*»•
engine performance. One of the strongest implications that diffusion flames
may dominate diesel combustion and NO formation arises from a close
examination of the EGR emissions behavior. The effect of exhaust gas
recirculation is to replace a portion of the incoming air with exhaust gases.
This has three results:
(i) The new mixture is warmer than the air alone (by
virtue of introduction of heated combustion products).
(ii) The new mixture has a higher mean specific heat
(since the combustion products HO and CO? have
more degrees of freedom than air, which is diatomic).
(iii) The mixture has a reduced oxygen concentration.
All three of these alterations in the intake charge have an effect on the flame
structure (temperature and species profiles) following ignition. It is the pur-
pose of this brief analysis to show that the measured nitric oxide reductions
with EGR can be "explained" on the basis of the predicted decrease of peak
flame temperature in a diffusion flame. The factor dominating this suppression
is neither item (i) nor item (ii), (in fact< these effects are small and tend to
cancel each other), but rather item (iii), the reduced oxygen mass fraction.
The conditions surrounding the diffusion flame were taken arbitrarily to be
the intake charge compressed adiabatically to TDC; thus the calculation is
representative of burning early in the cycle.
The Zeldovich mechanism for NO formation [Zeldovich and Raizer (1966)]
tells us that the rate of formation rises exponentially with temperature:
~exp(-E/RT), E/R= 123,000°R (1)
where the activation energy is derived from the activation energy of the con-
trolling reaction, O + N? -* NO + N, and the O? dissociation energy.
17
-------
We examine the rate of NO formation at the place most likely to
produce significant amounts of nitric oxide, i.e., at the flame front of a
diffusion flame. Here the temperature can be written [Williams (1965)]:
Tfl= Tf
1
•-I
"'p
Y°°
(Q-L) , •+ C (T -
i p °°
/ Y-\
( i i ox )
\1 ' > 1
V
where
fl
Q
I
OX
•• flame temperature
: fuel-droplet surface temperature
: specific heat (at constant pressure) of the gas
heat of combustion per unit mass of fuel
= latent heat of vaporization of the fuel
: ambient oxygen mass fraction
i = stoichiometric oxygen to fuel ratio (by mass)
T^ = gas temperature far from droplet
Although this expression is accurate enough to suit our purposes, it was
derived subject to simplifying assumptions, the most serious of which are
(i) Infinitely fast irreversible hydrocarbon oxidation
(completely diffusion-controlled)
(ii) Dissociation not considered
(iii) Single C common to both sides of the flame
For diesel fuel we take T, = 973°R (boiling point), Q = 19,000 Btu/lb, L =
144 Btu/lb, and i = 3.36. For T^, we take the compression temperature of
a 19:1 engine (appro*. 1200UF), and assume that the EGR is 100°R hotter
than the air in the intake manifold. These relative magnitudes permit the
following approximation:
ox
(2)
18
-------
The remaining quantities, C , Y^f and !» depend on the % EGR
according to the following recipes (derived for a 17:1 compression ratio at
24:1 overall air/fuel ratio):
TOO =
C =
p
ox
1677 + 300 x (% EGR), R
.30 + .02 x (% EGR), Btu/lb-°F
.232 - .143 x (% EGR)
(3)
(4)
(5)
Based on expressions (1) through (5) we can calculate the relative
NO-formation rate at the diffusion flame surface with different amounts of
recirculation:
% EGR
0
10
20
30
Y00
ox
.232
.218
.203
.189
C
p
Btu/lb-°F
.300
.302
.304
.306
Tro,°R
1677
1707
1737
1767
TfI'°R
5610
5360
5157
4947
Calculated
dNO/dt
(relative)
1.00
• 41
.17
.07
It is important to note that the Y" effect dominates; for example, at 30% EGR,
C and Too give rise to offsetting 2% changes in Tp, whereas Y°° causes an 11%
' p ^^ OX
suppression.
Figure 15 compares the observed NO reductions with the calculated
reductions in the rate of NO formation. The shapes of the curves agree rather
well for values of EGR above 10%; the leveling off of measured values at 0%
EGR is not simulated by the present calculations.
b. Effect of Water Injection
The introduction of one part water for every two parts fuel into the
intake air causes NO to drop about 50%, as shown in Figures 16 and 17.
Abthoff and Luther (1969) have shown that the effect occurs in the combustion
chamber and is not due to NO absorption by recondensed water in the exhaust.
19
-------
The interpretation for this effect presumably lies in thermal quenching of
NO-producing reactions. The effect of adding 0.5 gm water/gin fuel on the
temperature of the combustion products at = 1 is about -150°K . It can
be readily shown that, for the Zeldovich mechanism at ^w 1.0, a 100°K
drop would cause a factor of three reduction in NO-production rate. The
measurements bear these facts out: When water is injected into the air
intake and hence throughout the cylinder, some water is wasted; by contrast,
the potential for NO reductions for fuel/water emulsions appear to be
yt
greater (see Appendix D). It might be noted that water injection seems to
have a greater potential for reduction at high speeds and low-to-medium
loads.
c. Effect of Intake Air Temperature
Since the effects of water injection on NO emissions are so sub-
stantial, it is somewhat surprising to find the small effect of air temperature
shown in Figure 18. The compression temperature should be boosted about
300 R by a 100 R rise in intake temperature, according to the adiabatic
approximation:
T
. y=i.4
Offsetting this 300 R rise is a slight increase in heat loss during compression.
Nevertheless the flame temperature for the first fuel elements which bum is
bound to be boosted at least 100 R, even considering dissociation. Whereas
a flame temperature change of this order should increase NO production rate
by 50 to 100%, only 10 to 15% increase in NO emission is observed (Figure
18). One is led to the hypothesis that the early burning matters relatively
little and NO-formation is more sensitive to the conditions at later stages of
*The temperature decrease would be much less if the entire contents of the
cylinder were homogeneous and at a uniform temperature (AT" ~ -75°K).
20
-------
combustion. Due to heat transfer, the memory of the initial compression
temperature is washed out. Another conceivable effect of greater air tempera-
ture is greater fuel evaporation prior to ignition, so that heat release is weighted
more toward the premixed "spike" than the ensuing diffusion flame burning.
Also, ignition delay is smaller with air preheat.
d. Effect of Intake Air Pressure (Turbocharging)
As shown in Figures 19 and 20, the effect of simulated turbocharge
with aftercool on NO is not substantial, provided A/F rather than BMEP is
held constant as in the current tests. As the air density goes up, the fuel
burning rate increases; counteracting this is the fact that the average point
of fuel injection is delayed because the duration of fuel injection is increased
to keep A/F constant. Increased air density also shortens ignition delay and
alters spray penetration. The diffusion flame temperature should not change
significantly, because for a given engine the compression temperature depends
only on the intake temperature and not on the intake density. Because of
these compensating effects, NO remains at about the same level as might
J\,
be expectedo However, the BSFC drops about 20% with turbocharging due to
the improved ratio of BMEP/FMEP (the incremental heat release is not taxed
by friction). It is primarily this BSFC factor which reduces the gm NO,.,/
BHP-hr and has made turbocharging popular as an emission control technique.
4. Effect of Air/Fuel Mixing Parameters
a. Effect of Fuel Orifice Diameter
Variations in those parameters which affect dispersion of fuel and
rate of mixing provide significant clues about the diesel flame. Fuel orifice
size was varied in an attempt to find the effect of fuel dispersion (e.g.,
initial mean droplet diameter). In these tests the total orifice area (and
hence the rate of injection) was maintained near constant by making com-
pensating changes in the number of orifices. The results are shown in
Figure 21. The system with fewer, larger orifices displayed substantially
lower NO emissions. This result is in accord with the data of Hames et al.(1971)
x
as shown in Appendix D.
21
-------
At least three mechanisms may be at work here:
(i) The system with fewer fuel jets will have a smaller
surface-to-volume ratio of the fuel spray, creating
delayed mixing and heat release, and hence lower
NO .
x
(ii) The larger diameter jets will penetrate further, perhaps
splashing off the chamber walls.
(iii) The larger orifices should produce larger droplets (in
the mean). On the one hand, large droplets will take
longer to bum (reducing NO ), however for constant
ambient conditions, Seery and Bowman have suggested
NO ~d .
b. Effect of Injection Rate
The rate of injection was varied by changing orifice size for a
given number of orifices. Although orifice area was doubled in these tests,
the rate of injection increased only a factor of 1.3 (because the fuel line
pressure relaxed for larger orifices). The results are given in Figures 22 and
23 and show opposite trends for 14 and 17 compression ratio. Hames et al.(1971)
and Landen (1963) found lower NO emissions for larger rates of injection
through large orifices. Several factors come into play:
(i) Larger orifices produce larger droplets. The larger
droplets could increase NO (the diffusion-flame
structure argument) or decrease NO (the retarded
burning argument).
(ii) Faster injection permits faster burning (which would
increase NO .
x
c. Effect of Air Swirl
Increasing air swirl apparently promotes NO formation, as docu-
mented in Figure 24. The effect is stronger in our engine than previous
studies have observed (Appendix D) and may be due to faster mixing in the
high swirl case, which would lead to earlier heat release. Thus NO might
be increased because the active subzones would be hotter due to (a) faster
heat release for a given heat loss, and (b) burning before expansion cools
the bulk gases. Another effect which probably occurs with high swirl is an
increased tendency for both fuel and cool air to be thrown outward to the
chamber walls.
22
-------
1.2
Figure 3V';
EFFECT OF VOLUME RATIO
' (Prechamber Engines)
^: 1.1 -
I—I
£1.0
0.9
•T-CU
1.6
1.4
1.2
_
O
S
O
0.6
0.4
0.2
O .76 1500 SID
D .76 1800 STD
O.60 1500 STD
V .45 1500 STD
A .30 1500 STD
bx.45 150.0 -9
CR = 19/1
CR= 17/1
o
CO
o ••-•
3 o
o
CO
30
20
oa
tn
15% 25%
Ratio of Prechamber-to-Total Volume
23
35%
-------
1.2
o
0
O
,1.0
*
0.9
1.6
1.4
1.2
1.0
o
O
SO. 8
O
S3
0.6
0.4
0.2
Figure 4
EFFECT OF VOLUME RATIO
(All Engines.of OR = 17)
RPM NO Soot ISFC
__ O 0.0
O .45 1500 91.0 6.0 .294
A .76 2100 49.1 10.0 .327
Reference values for Figures 3
to 24 are given in Ib NO /1000
J\
Ib fuel, % opacity, and Ib NO /
J\
BHP-hr.
5
4
o
Direct
Injection
30
20
10
1-1
ffi
CQ
I
I
D>
0.2
0.4 .. 0.6 0.8
V(fuel chamber)/V(total)
24
1.0
-------
1.2
o
O)
I—I
&
C/D
0.9
O
O
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
Figure 5
EFFECT OF COMPRESSION RATIO
(Direct Injection Engines)
O
A
C
V
.75 1500
.75 2100
.60 1500
.60 2100
.45 1500
,45 2100
RPM Variation NO Soot ISFC
O O Q_
STD
61.
49.
77.
64.
91.
5 11.5 .341
1 10.0 .327
7 10.0 .311
85.0
45 1500 Turboch. 97.0
45 1500 Low S. 92.9
6
6,
4,
9.
14
15
16
17
18
19
Compression Ratio
25
.304
,294
,283
,294
0.9 .281 _
o
o
CO
o
o
CO
30
20
10 f
a,
30
^
2 &
-------
L.2
Figure 6
EFFECT OF COMPRESSION RATIO
(Prechamber Engines, Volume Ratio = 25%)
O]i
En L' L
cn
Ol.O
tin
CO
I—I
0.9 —
O
A
D
V
RPM NO Soot ISFC
_ O O O
Timing
.75 1500 12,7 12.0
STD
,75 2100 12,
.60 1500 17,
.60 2100 17.
.45 1500 26
.45 2100 32,
O .45 1500 41.5
9.
4.
2,
i.
i.
3.
o
o
O
o
CO
1.6
1.4
1.2
1.0
O
So. 8
O
s
0.6
0.4
0.2
30
O-
I
_L
10 ffi
3
2
14
15
16 17
Compression Ratio
26
18
19
-------
1.
1-1
CO
Figure 7
EFFECT OF LOAD
(Direct Injection Engines)
0.4
l.fi
1.4
1.2
—P
§1.0
o
*'' 0. 8
0.6
0.4
0.2
CR RPM
NO Soot ISFC
o o o
o
D
O
v
14 1500 High swirl 94.4
14 1500 Med.swirl 46.7
14 2100 Med.swirl 40.9
17 1500 Med. swirl 91.0
3.5 .298
5.3 .316
6.0 .305
6.0 .294
A 17 2100 Med.swirl 85.0 4.0.283
3:X
O
O
2<:-
0
0
D
0'
^ox
3 ^
D-.
i
l
.3 .4
Equivalence Ratio,
.5 .6
(Air utilization fraction)
27
.7
-------
Figure 8
EFFECT OF LOAD
(Prechamber Engines)
1500 26.4
2100 32.2
1500 36.8
2100 44.8
1500 11
2100 12
1500 16
.3 .4 ' .5 .6
Equivalence Ratio, 0 (Air utilization fraction)
28
-------
Figure 9
EFFECT OF TIMING
(Direct Injection Engines)
Soot^ ISFCn
1500 6x.012 64.
1500 4x.014 51.8
1500 6x.014 50.3
1500 6x.010(std) 46.7
2100 " 64.3
1500 " 91.
2100 " 85.
1500 " 102.
10
Start of Injection, &. . (°BTDC)
29
-------
1.2
[X.
co
-l
1 '
0.9
1.6
1.4
1.2
O
0.6
0.4
0.2
Figure 10
EFFECT OF TIMING
(Prechamber Engines)
CR VR RPM NO Soot ISFC
o o g
O 17 35 .45 1500 36.8 1.4 .313
D 19 15 .45 1500 11.7 24.0 .357
O17 25 .45 1500 26.4 1.7 .306
A19 25 .45 1500 16.7 5.0 .322
I
I
15
10 5
Start of Injection, 0
30
inj
(°BTDC)
TDC
o
o
CO
CO
30
20
lOi
QD
-------
1.2
^JD
o
CO1'1
i—I
\
O i n
UH i ' U
CO
)—I
0.9
O
o
1.6
1.4
1.2
[l.O
>
'0.8
0.6
0.4
0.2
Figure 11
EFFECT OF ENGINE SPEED
(Direct Injection Engines)
I
OR
Variables
NO Soot ISFC
o o o
n 14
O 14
v 14
A 14
> 14
< 17
-M7
A17
^1 17
£14
.76
.45
.76
.60
.60
.76
.45
.60
.30
,76
Low Swirl
STD
Turbocharged 56.3
STD
High Swirl
14.2 17.0
46.7 5.3
32.0 20.0
36.5 9.5
8.0
61.5 11.5
91.0 6.0
77.7 10.0
102.0 3.0
.346
.316
.370
.346
.338
,341
,294
,311
,311
Low Swirl
i
1200
_J
1500
-
O
o
co
o
o
CO
0
0 u
i
a,
o rc
CQ
O
1800
2100
RPM
31
-------
Figure 12
EFFECT OF ENGINE SPEED
(Prechamber Engines)
o 19 25
n 19 25
19 25
v 19 25
O 19 15
17 35
>19 15
9.0 31.9
14.5 11.0
7 5.0
7 1.5
7.5 46.0
13.9 13.0
11.7 24.0
9.1 44.0
< 19 15 .60
1200
1500
1800
I
2100
o
o
CO
o
o
CO
30
20 £
ex
1 0
* U
EC
X
RPM
32
-------
1.2
Figure 13
EFFECT OF EGR
(Direct Injection Engines)
I
•GO
oo
1.0
0.9
1.6
1.4
1.2
1.0
_p
O
5. 0.8
O
*0.6
0.4
0.2
O
O
O
o
CO
O
D
14 .60
14 .60
O 14 .45
A 14 .45
V 17 .60
< 17 .60
> 17 .45
&. 17 .45
RPM NO Soot ISFC
g g o
1500 36.5 9.5 .346
2100 29.3 14.0 .338
1500 46.7 5.3 .316
2100 40.9 6.0 .305
1500 77.7 10.0 .311
2100 64.3 6.5 .304
1500 91.0 6.0 .294
2100 85.0 4.0 .283
30
20
CD
5\
3
2
10
33
20
30
-------
Figure 14
EFFECT OF EGR
(Prechamber Engines)
RPM CR VR NO Soot ISFC
o o o
O .45 2100 19 25 18.8
D .45 1500 19 25 16.7
O.45 2100 17 25 32,
V.60 2100 17 25 17,
A. 45 1500 17 25 26,
>.. 60 1500 17 25 17,
<.45 2100 17 35 48.8
^.45 1500 17 35 36.8
.2
.6
,4
,7
.1
.0
.7
.7
.7
,5
,7
.297
,322
.301
.347
,306
,340
323
1.4 .313
o
o
CO
3\
o
o
CO
30
20
-------
Top End
RPM
DI-17 .60 1500 -20
DI-17 .60 2100 -20
3000
NO
(ppm)
Calculated relative rate of N
formation at the flame front of
an idealized diffusion flame
2000
1000
% EGR
A-
30%
20%
10%
.16
.17
.18
.19
0%
AF = 24
.20
.21
X (Start of Compression)
L*
•Assume
7% Resid.
Figure 15. Effect of Exhaust Recirculation
35
-------
1.2
o
Figure 16
EFFECT OF WATER INJECTION
(Direct Injection Engines)
CO
0.9
RPM BTDC NO Soot ISFC
o o o
O .31 1500
n .40 2100
O .63 2100
15 68.6 2.4 .283
25 117.0 1.0 .288
25 91.2 2.9 .397
-
o
o
co
3\
j-j
o
o
CO
1.6
1.4
1.2
o
§1.0
O
^'0.8
0.6
0.4
0.2
O
30
20
lOo.
ffi
CO
0.5
1.0
1.5
(Intake)
36
-------
o
(J-,
CO
1—I
"pi.o
(j-i
co
i—i
0.9
1.6
1.4
1.2
1.0
_p
O
SO.8
O
0.6
0.4
0.2
Figure 17
EFFECT OF WATER INJECTION
(Prechamber Engines)
d> RPM NO Soot ISFC
^ g o o
O .45 2100 32.2 1.7 .301
O .45 1500 26.4 1.7 .306
O .76 2100 12.1 9.0 .393
V .76 1500 12.7 12.0 .376
0.5
1.0 1.5
!*„ ^/m. (Intake)
HOU i
L 37
5
14
o
o
co
-M
o
o
co
30
20
110
DQ
-------
Figure 18
EFFECT OF INTAKE AIR TEMPERATURE
T
T
RPM Engine
NO Soot ISFC
o o g
o
n
O
v
A
76 1500
76 2100
45 1500
45 2100
45 21dO
>.45 1500
<.76 2100
^.76 1500
Direct Inj
Direct Inj
Direct Inj
Direct Inj
Precham.
Precham.
Precham.
Precham.
61.5 11.5
49.1 10.0
91.0 6.0
85.0 4.0
32.2 1.7
26.4 1.7
12,1 9.0
341
327
294
283
301
306
393
12.7 12.0 .376
o
o
co
o
o
OD
30
20
.c
100 150 200
Intake Air Temperature, T (°F)
3
38
250
-------
1.2
co
0.9
1.6
1.4
1.2
O
g.1.0
o
^0.8
0.6
0.4
0.2
Figure 19
EFFECT OF AIR PRESSURE (TURBOCHARGING)
(Direct Injection Engines)
RPM NO Soot
o o
ISFC
O 17
a 17
017
v 17
A 17
> 14
< 14
V 14
^ 14
.60 1500 77.7 10.0
.60 2100 64.3 6.5
.45 1500 91.0
.45 2100 85.0
.30 1500 102
.60 1500 36.5
.60 2100 29.3 14
.45 1500 46.7 5
.45 2100 40.9
6,
4.
3,
9,
0
0
0
5
0
3
6.0
.311
.304
.294
.283
.281
.345
.338
.315
.305
o
•5 °
3 co
4-1
O
30
20
S
01
_L
JL
30
38 46 54
Intake Pressure, P ("Hg abs)
d.
39
62
70
-------
co
I-H
hi;1-0
CO
I—I
0.9|—
1.6
1.4
1.2
__
O
S
o
0.8
0.6
0.4
0.2
Figure, 20
EFFECT OF AIR PRESSURE (TURBOCHARGING)
(Prechamber Engines)
VR
RPM NO Soot ISFC
o o o
O
D
O
V
A
25 .76 1500 12.7 12.0
25 .60 2100 17.6 2.7
25 .45 1500 26.4 1
25 .45 2100 32.2 1
35 .76 1500 13.9 13
35 .45 1500 36.8
35 .45 2100 44.8
,376
,347
.306
,301
.385
.313
9.7 .323 —
-L
o
o
££.
4-»
O
o
30
20
.c
10K
2
1
30
50
Intake Pressure, P
40
70
("Hg abs)
-------
Figure 21
EFFECT OF ORIFICE DIAMETER
(Direct Injection Engines)
,.
U
"
CO
I — I
\
h
co
.0
0.9
CR
RPM NO Soot
o o
ISFC
O 17
n 17
o 17
A 17
V 14
< 14
014
.76
.76
.45
.45
.76
.76
.45
.45
.1500 41
2100 33
1500 15
2100 77
1500 33
2100 35
1500 73.1
2100 60.4
20
22
2
3
2 21
6 15
0
0
0
0
0
0
4.0
8.0
.347
.342
.306
.280
.367
.362
.306
.301
1.6
1.4
1.2
_
O
si.o
\
o
0.6
0.4
0.2
.002
.004
.006
.008
.010
.012
.014
d (in), Fuel Rate Held Constant
f 41
-------
Figure 22
EFFECT OF FUEL RATE
(Direct Injection Engines)
co
0.9
0 RPM NOQ Soot0 ISFC0
O 14 .45 2100 40.9 6.0 .305
D 14 .45 1500 46.7 5.3 .315
A 14 .76 1500 32.0 20.0 .370
• 17 .76 2100 49.1 10.0 .327
O 17 .45 2100 85.0 4.0 .283
V 17 .76 1500 61.5 11.5 .341
V 17 .45 1500 91.0 6.0 .294
O
o
1.6
1.4
1.:
o
1.0
0.8
0.6
0.4
0.
o
o
C/D
3\
4-*
O
o
30
i
a,
50
o O^
2
b
dm,
mm / CA
42
-------
1.2
o
CO
to
0.9
1.6
1.4
rlgure 23
EFFECT OF FUEL RATE
(Prechamber Engines)
CR VR 0 RPM NO0 SootQ ISFCC
O 17 25-.76 2100 12.1.9.0 .376
D 17 25 .45 2100 32.2 1.7 .301
O 17 35 .45 2100 44.8 9.7 .323
* 17 35 .76 2100 19.0 3.0 .368
^p
o
o
W.
\
o
1.2
o
O
Sl.O
O
^
0.8
0.6
0.4
0.2
-O-
-O-
30
20
10
i
D-,
3
2
0
2
01
6
dm.
8
43
-------
Figure 24
EFFECT OF SWIRL
(Direct Injection Engines)
CR RPM NO Soot
^ oo
ISFC
o 14
a 14
O14
v 14
A 17
b.17
A 17
> 17
.45
.76
.76
.45
.76
.76
.45
.45
2100 40.9 6.0
2100 25.3 21.0
1500 32.0 20.0
1500 46.7 5.3
1500 61.5 11.5
2100 49.1 10.0
2100 85.0
1500 91.0
4.0
6.0
.305
.378
.370
.316
.341
.327
.283
.294
o
o
CO
o
o
I
30
20
10
2
1
Relative Swirl Magnitude
44
-------
C. SUMMARY OF EMISSIONS EVIDENCE
Seven parameters were found to change NO emissions by at least
40%, as summarized in Table 7 and Figure 25. Corresponding changes in soot
were observed; typically the soot increased when NO was reduced, although
X
some exceptions such as use of divided chamber and water injection were
found. Other measures have increased soot with very little NO change—
these might be avoided if other tests corroborate these trends.
Table 7
LINEAR INFLUENCE COEFFICIENTS
DERIVED FROM SINGLE CYLINDER TESTS
CAUTION
Test engine not optimized, results not strictly representative of production engines.
% changes arc relative to arbitrarily selected baseline—this allows numerical values
of dY/Yto exceed 100%.
Emissions variations were non-linear In most cases; a linear fit was Imposed.
Most coefficients averaged over several cases involving distinct combinations of speed,
load, etc.
€
C
I
§
V
fi
f
e
X
1
*
Variable
X
Compression Ratio
Timing
Load
(Equivalence Ratio)
Speed
EGR
Water Injection
Turbocharglng
Air Temperature
Air Swirl
Fuel Orifice Size
Fuel Rate
Fuel Temperature
Pilot Injection
Compression Ratio
Divided Chamber
Prechamber Volume
Timing
Load
(Equivalence Ratio)
Speed
EGR
Water Injection
Turbocharglng
Air Temperature
Fuel Rate
Range of Variable
4X
3 units
4 0^-17 CA retard
A$« .30
ARPM - 600
lAm^/mtot- 30%
A(m 2 _/m,>- 1.5
2
AP - 20" Hg
ATa - 100°F
"med" to "high"
4df- .004"
-1.5
mm V°CA
100°F
ON/OFF (15% fuel
at -40°CA)
2 units
DI vs . PC
A(V/Vtot) - 20%
ASj - 15°CA
A*- .30
ARPM - 600
Ll&^-^T*30*
A(ift|^ Q/ift J "1.5
APa - 20" Hg
ATa » 100°F
A3"l
Figure
5
9
7
11
13
16
19
18
24
21
22
~
-.
6
Table 6
3
10
8
12
14
17
20
18
23
Observed % Change In
Em
ANO/NO
+ 90%
-160%
+ 5%
- 20%
- 70%
- 70%
+ 12%
+ 20%
+ 60%
- 30%
*
- 12%
+ 40%
- 30%
- 60%
+ 75%
*
- 30%
*
- 64%
- 66%
- 8%
+ 2%
*
Isslons and C
ASoot/Soot
- 30%
+120%
+210%
- 10%
+220%
+ 40%
+ 50%
- 10%
- 50%
- 10%
- 40%
- 30%
+ 30%
+170%
- 50%
- 80%
-150%
+500%
*
+150%
- 80%
+140%
+ 70%
+120%
utput
AISFC/ISFC
- 9%
- 14%
+ 17%
- 2%
+ 6%
+ 1%
+ 7%
- 1%
- 5%
*
- 10%
- 3%-
- 2%
+ 4%
+ 1%
- 12%
+ 13%
+ 28%
*
+ 3%
• + 2% .
+ 15%
+ 1%
*
•Excessive variation among cases or excessive non-linearity
45
-------
Figure 25
EMISSIONS SUMMARY ILLUSTRATING THE SOOT-NO TRADEOFF
CAUTION
+100
450
0
NO
-50
-100
-150
-200
Test engine not optimized, results not strictly representative of production engines.
% changes are relative to arbitrarily selected baseline — this allows numerical values
of dYA to exceed 100%.
Emissions variations were non-linear in most cases; a linear fit was imposed.
Most coefficients averaged over several cases involving distinct combinations of
speed, load. etc.
Ratio
Air Temp*
Fuel Ternp,^
RPM •
Orifice^
BPC
Water
N
Increased
Control of Smoke
and NO
_L
• Direct Injection
• Prechamber
*" N Measures showing high
^ /NO sensitivity
Pilot
Turbochg
Air
T
emp
'Load
•Swirl
Water
Turbochg
• Comp Ratio
N
Load
Timing
/
-100
-1-100 +200
A Soot
Soot
+300
+400
+500
46
-------
In Table 8 we have noted some of the NO mechanisms which were
J\.
hypothesized to explain the data. A qualitative picture of what may be
affecting pollutant formation in the diesel flame emerges from the common
network of arguments in Table 8.
ENGINE
PARAMETERS
Table 8
HYPOTHETICAL MECHANISMS CITED
TO RATIONALIZE NO EMISSIONS BEHAVIOR
POSSIBLE MECHANISMS
Swirl
Speed
Fuel Rate
Fuel Orifice Size
Swirl
Speed
Fuel Rate
Fuel Orifice Size
Turbocharging
Fuel Rate
Timing
Load
Spaed
Water Injection
Swirl
Compression Ratio
Air Temperature
Load
EG a
Speed
Orifice Size
CENTRIFUGAL AND
SPRAY PENETRATION
(A/F DISTRIBUTION)
CRANKANGLE
AT WHICH
BURNING OCCURS
HEAT LOSS RATE —'
COMPRESSION
TEMPERATURE
MEAN OXYGEN
CONTENT
DROPLET SIZE
OR FUEL
ELEMENT SIZE
NO
PRODUCTION RATE
FLAME RESIDENCE
TIME OR
DIFFUSION TIME
47
-------
According to these results, the following phenomena should be sub-
jected to rigorous theoretical analysis and experimental studies in order to
reach a better understanding of diesel-flame generated MO and smoke:
Local Diffusion Flames: The temperature and burning rate
of such flames is independent of the overall fuel/
air ratio. Many forms are conceivable: wake bum-
ing as studied by Natarajan and Brzustowski (1970),
envelope flames classically studied by Godsave (1950),
and spray combustion as investigated by McCreath and
Chigier (1972).
Swirl Effects: Angular motion and entrainment of fuel, radial
stratification due to centrifugal effects , and turbulence
levels.
Fuel Spray Details: Penetration, drop size distribution,
possible wall impingement, and entrainment by air.
Time-Dependent Phenomena: Must be precisely character-
ized and overlayed with an exactly (+ . 5 CA) specified
piston motion and fuel delivery schedule. The phasing
of key phenomena such as ignition, mixing rates, bum-
ing rates, evaporation rates, and heat transfer rates
must be characterized as precisely as possible.
Prechamber Phenomena: Such as fluid transfer between.the
two chambers, including phase lags and heat transfer.
48
-------
IV. AN INTERPRETATION OF DIESEL COMBUSTION AND
POLLUTANT FORMATION
CHAPTER SUMMARY
On the basis of the following evidence, it is suggested that the
mechanism of diesel combustion is diffusion-limited heat release at flame
surfaces:
• NO emissions insensitive to overall fuel/air ratio.
.X
• Emissions and heat release sensitive to swirl and
fuel dispersion.
• EGR data well correlated by diffusion flame theory.
• Shape and duration of measured apparent heat release
indicates mixing control.
• Movies showing heterogeneous burning and yellow-
orange flames.
Although the geometry of these flame surfaces is not known, on the micro-
scale the form of the flame temperature and species profiles needed to
predict NO and smoke formation can be precisely described. These profiles
jC
are unique functions of the relative velocity, oxidizer concentration, and
temperature of the ambient environment in which a given fuel element is
burning.
In order to supply this information about flame environments, one
must shift to a larger "macroscale" and analyze the fuel/air mixing pro-
cesses. We assume the swirling air passing over each fuel jet entrains a
distribution of droplet sizes into the mixture of air and hot combustion
products.
An outline for synthesizing this concept of heat release into a
complete diesel emissions model with heat transfer, ignition delay, and
pollutant formation is presented. Our objective is to incorporate funda-
mental fluid physics, even if some of the actual diesel processes are sim-
plified, in order to avoid (where possible) the empirical coefficients which
have limited the usefulness of existing models.
It is recognized that this modeling attempt will be speculative
without more insight into actual flame behavior than we have at present. A
number of diesel flame measurements are suggested to better characterize
the fuel spray, air swirl, mixing and pollutant-formation processes.
-------
IV. AN INTERPRETATION OF DIESEL COMBUSTION AND
POLLUTANT FORMATION
A. MIXING AND COMBUSTION MECHANISMS
The emissions data present definite arguments about the nature
of the diesel flame. Emissions and performance are quite sensitive to
chamber shape, swirl, and fuel dispersion. The fuel specific NO emissions
X
appear insensitive to A/F ratio. The EGR data is well correlated by a sim-
plified droplet diffusion flame theory. These behavioral patterns suggest
that the combustion is mixing-controlled; by this we mean that the time for
chemical reaction, once fuel and air are brought into molecular contact, is
quite short compared to the mixing time. In the combustion literature [e.g.
Fendell (1967)], the mixing-controlled flame is recognized as the limit of
large Damkohler number:
ry
LJ~L (Characteristic Mixing Time) ^ „
AT = ^ ' ** ' *• OO
D pD (Characteristic Reaction Time)
Such flames have the singular property of locating themselves at
the interface region between the unreacted fuel and the unbumed air. Pro-
duct/reactant mixing occurs on each side of the flame but fuel/air mixing
per se does not occur (the flame intervenes)*.
Mixing occurs in two distinct stages, and it will be useful to
define two scales of mixing:
(1) The scale on the order of the fuel spray or piston bowl
(the term "macromixing" has been suggested by Khan).
The zones and rate of fuel accessibility are defined by
mixing on this scale.
(2) The scale on the order of the fuel elements (droplets or
fuel vapor eddies). The "micromixing" on a molecular
scale, and hence the detailed flame structure itself, is
defined on this scale.
*Fuel/air mixing occurs to the extent that N < oo and the heat releasing
reaction is reversible.
49
-------
The suggestion of a mixing -limited heat release rate is corro-
borated if we look beyond the emissions behavior to chamber-pressure
measurements and high speed movies. Appealing to energy conservation,
the rate of heat release can be deduced from the pressure trace and known
volume changes:
q = f[P».
Figure 26 shows the pressure and derived heat release traces for a typical
DI engine, as recorded by Lyn (1963) . The first feature one might notice is
that characteristically the duration of heat release exceeds the period of
fuel injection. This means fuel is not consumed immediately upon entering
the chamber, and a rate-limiting process is at work. Knowing that hydro-
-4
carbon burning times are short (10 sec) at these temperatures, the flame
is suspected to be mixing-limited.
A curious feature noted by Lyn (1963) is the brief period of intense
heat-release at the start of combustion (often termed the "spike"). Typi-
cally only 10 to 15% of the energy release occurs during the spike, which
implies that after ignition the combustion slows down for some reason. This
would not occur in a premixed flame; thus we infer that sufficient fuel is not
available and the burning is modulated by the mixing processes.
High-speed movies provide additional clues as a qualitative expla-
nation of the spike followed by the plateau. Lyn (1963) observed that the
spike emission was blue (indicative of a combustion wave passing through
premixed fuel vapor and air) , whereas the plateau appeared yellow-orange
(suggesting the carbon-laden diffusion flames — the most familiar of which
is the ordinary candle) . Movies taken by Scott (1969) show the same yellow-
white emission, and reveal that the fuel distribution is non-uniform. Burning
starts locally on the outer edges of the fuel spray and quickly envelops the
spray. Apparently, adjacent burning sprays do not interact, at least early in
51
-------
Figure 26
Overlay of Combustion and Mixing Processes
Ignition
99% Energy Release
en
t-o
Cumulative
Heat Release
Temperature
Rate of Heat Release
Fuel Schedule
i Crank Angle ®
Cycle Thermodynamics
Fuel Spray Delivery
Spray Breakup
Air/Spray Macromix
Premixed Combustion
Diffusion Combustion
ylinder contents
Turbulent Entrainment
Nitric Oxide Kinetics
Soot Kinetics
uniformly mixed
Too cold to form NO
Too cold to bum soot
-------
the heat release period. The air swirl in Scott's engine was sufficient that
wall-impingement of fuel occurred; wall markings of the single-cylinder test
engine indicated little or no wall impingement.
To recapitulate, the clues examined above indicate the following
mechanisms may account for mixing and combustion:
(i) Non-uniform distribution of liquid fuel spray.
(ii) Ignition after a 5 to 15°CA delay.
(iii) Rapid premixed combustion of evaporated "fines",
yielding the blue emission and the "spike" account-
ing for 10 to 15% of the heat release.
(iv) Mixing-controlled burning, accounting for most of
the heat release and the continuous emission from
soot. Mixing occurs simultaneously on two scales:
a. Macromixing, whereby fuel elements (e.g.
droplets) are delivered to an oxidizing zone.
b. Molecular diffusion, whereby fuel and air
molecules approach the interfacial flame zone,
actually collide, and bum.
B. OUTLINE OF AN EMISSIONS MODEL
1. Model Components Identified
The complexity of the diesel flame is widely appreciated; our
approach is to break such a problem into simpler subproblems which are more
amenable to analysis. It is conceptually helpful to disassemble the diesel
model into the following ten recognizable phenomena (rather than by crank
angle regimes or gas vs. liquid phase, say):
1. Cycle Thermodynamics (Volume Change and Heat Transfer)
2. Air Motion
3. Fuel Spray Delivery
4. Air/Spray Entrainment (Macromixing)
5. Delay to Spontaneous Ignition
6. Heat-Release "Spike"
7. Combustion (Diffusion-Controlled Combustion; Micromixing)
53
-------
8. Air/Product Entrainment
9. Nitric Oxide Kinetics
10. Soot Kinetics
A reasonable procedure in developing the computer code will be to
write distinct subroutines for each component; then improvements to any
one component can be introduced with a minimum of alterations to the basic
program.
It is recognized that the model components interact for two reasons:
(1) certain phenomena may be physically coupled if they happen simultaneously
as represented in Figure 26, and (2) late phenomena will be influenced by the
cumulative effect of previous processes. The overall model must account for
physical coupling and "memory" interactions with iteration loops as shown
in Figure 27.
2 . Specifications for the Computer Code
The computer routine has other guidelines and specifications as
\
well. Once developed, the model should help the engineer to focus valuable
experimental test resources on those engine configurations indicated by com-
puter runs to be most promising. The extent to which this can be done cost-
effectively depends upon how well the following criteria are met:
(i) The model must be human engineered, with input/output
in diesel engine jargon, and not requiring extra subpro-
programs for the compression stroke, etc. In short,
ideally the model will be accessible to the engineer.
(ii) The model must be versatile enough to describe a reason-
able variety of engine changes.
(iii) The numerical computation scheme must be efficient;
typical run times on modem computers should be in the
10^ to 10^ second range for a 360 revolution of the
crank. (Computer time/engine time « 10 .)
(iv) The relative accuracy should be about +10% so that predic-
tions of substantial (20 to 80%) NO changes are reliable.
(v) As far as possible, the model should be free of non-
universal empiricism.
54
-------
•Figure 27
Interaction of Model Components
Prescribed
Initial Gas
State
Injection
Characteristics
FUEL SPRAY
DELIVERY
AIR MOTION
CYCLE
THERMODYNAMICS
Compression
AIR/SPRAY INTERACTION
Combustion
Generated
Turbulence
DELAY TO
SPONTANEOUS IGNITION
HEAT-RELEASE
OVERSHOOT
COMBUSTION
Change
in Internal
Energy
AIR/PRODUCT ENTRAINMENT
Flame
Profiles and
Mean Species
and!
SOOT
KINETICS
NO
KINETICS
55
-------
Item (v) is particularly important and bears further comment. If
several flame-related constants must be adjusted for each diesel engine
design or operational change, then the model will not markedly reduce the
test matrix. Thus as a modeling objective, we seek a model which has
been formulated from first principles as much as possible.
Here the choice of simplifying assumptions is critical. Attempts
to use empirical "fits" to describe complicated flame processes
which are little understood will open the door for several adjustable con-
stants. These constants make the model produce answers, however they
are merely a substitute for our ignorance of complex flame couplings. In
this sense the model is not useful except as, a simulation; it performs the
task,"given-A, calculate-A" . In short, we recommend that the model incor-
porate processes which are well understood and can be described by phy-
sical laws. Some phenomena known to occur in the diesel flame may have
to be idealized (or even ignored) to achieve a physical treatment. Other
phenomena about which little is known (turbulence and diffusion flame geo-
metry) will have to be hypothesized.
3. Alternative Treatments of Increasing Completeness
and Physical Basis - -
The ten model components have been examined using a ladder of
alternative assumptions which moves toward the ideal of representing pre-
cisely what is happening in the flame. The modeling program can move up
this hierarchy to more and more accurate plateaus until one of four limits
is reached:
(i) Lack of knowledge of what is happening in the flame
(e.g. nature of diffusion flame).
(ii) Lack of ability to write and solve the governing equations
for a known phenomenon (e.g. turbulent mixing).
(iii) Lack of program resources to complete a solvable compo-
nent analysis (e.g. 3 dimensions, spectral radiation).
(iv) Lack of usefulness of increased component accuracy
relative to predicting NO from all ten components.
56
-------
As shown in Table 9, the most coarse assumption which can be
taken to handle a phenomena is to assume a limiting case which circumvents
any detailed description. Frequent choices are that the process is assumed
to occur instantaneously or that it does not occur at all. For example, non-
premixed flames are frequently assumed to have instantaneous chemical
reaction rates, so that burning is diffusion-limited. Simplified models of
combustion such as the NREC model may not be able to use fuel spray or
air swirl information; in these cases A-l and F-l would be selected.
Moving down to the next row, phenomenological models are postu-
lated wherein the process is assumed to exist (e.g., have a finite rate), but
the mathematical expression describing the process can be arbitrary. This
mathematical form may or may not have a basis in physics; coefficients are
introduced and adjusted to give agreement with the behavior of as large a
class of engine data as possible. Typical of this level of approach are the
arbitrary ignition relay, the relaxation law for transport (flux equals coeffi-
cient times perturbation), and the Arrhenius rate expression for soot forma-
tion with adjustable coefficients. Current state-of-the-art in modeling
dies el-generated NO appears to adopt this "phenomenological" approach
J\.
in six of the ten model components • '
The proposed model for Phase II attempts to reach the next level of
sophistication, which replaces the phenomenological laws with expressions
based on fluid-physics mechanisms. These expressions contain transport
and rate coefficients Just like the coarser models, but now the values must
be measured or derived rather than being freely adjustable. For example,
T-5 derives a value for emissivity from non-negotiable quantities such as
the overall fuel/air ratio. In model NO-4, the highly respected Leeds data
on the Zeldovich mechanism of NO formation is taken in place of arbitrary
coefficients as appearing in NO-3. Once the coefficients are fixed by the
physics, the burden of achieving agreement with engine data is shifted to
other model components. Inadequacies in, say, the entrainmerit will be
forced to the surface which otherwise might be hidden in the NO component.
57
-------
Table 9
ALTERNATE APPROACHES FOR MODEL COMPONENTS
le of This
miponent
ss
•g
g
c
D) -3
11
BZ
sl
'
Cycle
Thermodynamics
Given the initial state,
chamber geometry,
composition, heat
release rate and mass
injection, describe the
state of the gas (P,T)
as a function of crank
angle.
_.
Ad la ba tic compres-
sion and expansion
(no heat loss) .
fr^l
cycle averaged.
Adjustable heat loss,
released during com-
bustion only.
11-41
Convective and radia-
tive heat loss from T
and Tw coefficients
assigned.
Same as T-4 but trans-
fer coefficients derived
or measured.
Eg
Same as T-5 but allow
local hot zones to
radiate Independently.
Air Motion
Given the Inlet angular
momentum and RPM,
describe whatever
swirl, squish, and
turbulence character-
istics are needed for
mixing and combustion
models.
n
fewlrl not considered.
A^2l
flcient .
E3-
Mean gwirl from con-
servation of angular
momentum.
A^31
ujuT solid body or free
vortex .
tu(r) solid body Inner,
free vortex outer
2 parameters.
IA-61 .
Consider squish.
\-7]
Consider wall boundary
layer.
Fuel Spray
Delivery
Given the fuel sched-
ule, number and dia-
meter of orifices,
describe the droplet
dispersion spreading
angle, and injection
velocity.
n
Fuel dispersion not
considered (spray
structure not con-
sidered) .
7=H
persion. coefficient
(no structure) .
Homogeneous jet
nlzed); adjustable
spreading coefficient.
Homogeneous Jet,
width derived from
entralnment.
(Not applicable; spray
trajectory determined
from A/S models.
Monodlsperse spray.
r^g
Drop-size distribu-
tion:
(a) upper limit log
norma 1
• (b) Rosin-Rammler
(c) Nukigama
(d) Bl-dlsperse
Air/Spray
Entrapment
Given the air motion
and fuel spray, des-
cribe the macromixlng
of spray before burning
can be considered.
|ft/>s_1[
Instantaneous entrain-
ment of spray upon
Injection; no velocity
lag.
fA/S-2 |
(e.g. , Lyn's triangu-
lar profiles); adjust-
1A/S-3I
Entralnment based on
cylinder of fuel in
crossflow (without
drop size) .
Entra inment as A/S-3
but gives drop size
and velocity lag.
jA/3-51
Radial Jet In cross-
flow analyzed to give
peel off of droplets .
[A/3-51
High swirl model with
body forces analyzed to
give ballistic droplet
trajectories .
[ATS-?]
Wall impingement
analysis.
Delay to
Spontaneous
Ignition
Given the state and
motion of the air at
st°rt of injection, des-
cribe the delay before
first notlcable net heat
release.
pTfl
Instantaneous ignition
at start of injection
(no delay) .
CJP
adjustable coefficient.
Ignition delay derived
from both chemical rate
able coefficient) .
|5^_
Time delay based on
physical transport
(heat-up, evaporation.
mixing) . Stoichiometry
criteria for ignition.
Rate of buildup of
radicals to critical
level.
Rate of buildup of
temperature to critical
level.
Same as D-3 but
coefficients derived
from physics.
S3)
Describe ignition of
liquid fuel droplets,
Including trajectories
and droplet boundary
layer.
Heat-Release
Overshoot
Account for the "spike'
In the heat release
profiles.
53
Instantaneous heat
release from all
prepared fuel.
io^n
law for accumulated
fuel (e.g., Lyn's
triangular profiles) >
_
Sequence of sponta-
neous Ignition for
premlxed pockets
progressively resis-
tant A/F.
Flame speed model for
premlxed mixture con-
sumption.
FTr?
release due to pre-
mlxed combustion
kinetics.
Combustion
Given the fuel and air
distribution, describe
the heet release rate
and flame T and spe-
cles profiles.
[c-il
Instantaneous burning
upon fuel availability.
Burning specified to
occur at stoichio-
metric A/l? ratio.
Same as C-L but A/F
ratio la distributed.
[C^3j
giving finite burning
rate. Adjustable
coefficients include
size, residence time.
Macro size diffusion
flame based on droplet
clusters. Shape arbi-
trary.
lC-5l
Spherical droplet diffu-
sion flame, quasi-
steady "flame sheet".
[_C;JHJ>]
Droplet diffusion
flame; instead of
flame sheet, use
chemical equili-
brium diffusion
profile. .
Macro-size diffusion
flame with equilibrium
profiles as in C-6.
Ic-fll
Velocity lag effects in
droplet diffusion flame;
possible wake flame.
(Amend C-6 model).
Account for rising T
and falling O~ due to
neighboring droplets.
(Amend C-6) . Droplet
spacing must be
assumed.
Air/Product
Entralnment
Given the air motion
and combustion pro-
ducts, describe how
turbulent transport
will disperse and cool
the products.
A/P-ll
Instantaneous mixing
of products (uniform
gas composition).
A7FT1
mixed air, adjustable
mixing coefficient.
A/P-4 I
Proportional to AT, 4V,
or ff>2, turbulent
.dlffuslvity derived or
measured.
A/P-SI
Local turbulent diffu-
slvity based on local
gradients .
A/P-6 I
Diffusivity derived
from conservation
equation for turbu-
lence
Turbulent cross corre-
lations allowed. '
Nitric Oxide
Kinetics
Given the species and
temperature fields,
compute the net accu-
mulation of nitric
oxide.
Imo-il
Full equilibrium.
iNO^n
of exhaust NO with
key combustion para-
temperoture) .
Arrhenius rate law
with adjustable
coefficients.
1^6-41
Zeldbvich rate mecha-
nism.O/O., equilibrated
i NO-Sj
Same as NO-3 but
extended to Include
OH effect and NO
decomposition .
O^atom concentration
in partial equilibrium
with CO but not with
Oo .
NO-7|
Non-equilibrium O-
atom concentration
(complete C-H-O-N
scheme required).
Soot Kinetics
Given the species and
temperature fields,
compute the net accu-
mulation of soot.
13
Soot formation ignored.
pOt
of exhaust soot with
key combustion para-
A/Fl.
S~3 ]
Rate law involving
cylinder-averaged T,
A/F, P.; adjustable
coefficients.
s:;i
Basis for formation:
.Mechanism for fuel
pyrolysts and carbon
agglomeration.
33]
Local soot formation
rate (e.g. , higher In
fuel-rich regions).
533 :
Include soot combustion
rate.
-------
The description of spatial variations in temperature, species, and
velocity represents a substantial leap in complexity and accuracy. Although
spatial considerations such as the shape of spray flames (C-3) are likely to
be necessary to predict NO to high accuracy, we do not project that a com-
plete spatial model can be achieved under the current program scope. How-
ever, spatial distributions and gradients can be partially included in a few
model elements without requiring all ten elements to be .3-dimensional. For
example, a fraction of the chamber volume can be described as radiating at
a higher temperature than the remainder. Or the fuel spray trajectory as
influenced by the swirling air can be analyzed to give better estimates of
the velocity lags and characteristic dimensions of droplets or macro sized
flame zones.
Finally, at the bottom of Table 9 are listed a number of further
refinements pertaining to the individual model components.
4. Tentative Model
With reference to Table 9, the tentative model can be described as
follows:
Cycle Thermodynamics (T-6)
Convective heat loss from an Annand type expression, with coeffi-
cients measured or derived, not freely adjustable. Radiation will be con-
sidered allowing local hot zones to radiate independent of the bulk mean
temperature to accurately describe NO formation. Internal energy changes
will be converted to species distributions and temperature by means of the
equilibrium program "ODE" developed under NASA auspices. This program
allows for dissociation effects. Residual gas effects will be included, but
subroutines for valve flow are not to be re-invented.
Air Motion (S-5)
Rigorous swirl radial profiles will be derived for the bowl based on
chamber geometry and angular momentum after compression. Both an inner
solid body rotation and an outer free vortex flow will be assumed. Squish
will not be considered, nor will the wall boundary layer. Turbulence levels
will be derived from Reynolds number considerations and scale lengths.
59
-------
Fuel Spray (F-6)
The injection velocity will be computed from known injection schedule
and orifice dimensions. Spreading angle will be measured or derived from jet
theory. A drop size distribution is needed because its decay (shift of mass
mean size upward) will affect NO production, heat release, and entrainment
differently. A single drop size would force inaccuracies in the relative rates
of these three processes.
Air/Spray Entrainment (A/S-4)
Entrainment will be based on the well-studied behavior of a cylin-
drical spray in a crossflow of air. Since swirl and spray structure are known
as a function of chamber radius, this entrainment analysis will be conducted
at a number of discrete radial positions. The analysis will yield the rate of
entrainment of a distribution of drop sizes, each having certain velocity lag
(if required) .
Ignition Delay (D-7)
Ignition delay will be derived from both a chemical delay r , taken
c
from D-5 or D-6, coupled with a physical transport delay T , taken from D-4,
into an expression of the form
r' = (r +CT )/(1 + C)
c p -
Clearly, whenC is large, transport controls; whereas when C— »0, chemistry
controls. The expression for T will be derived from heat-up, evapora-
tion and mixing, with a limiting stoichiometry criterion. The expression for
T will be derived from an Arrhenius expression for local temperature
c
buildup or from a chain-branching expression for radical buildup.
Heat-Release Overshoot (O-3)
The ignition model D-7 will be generalized to permit a succession
of ignitions. If r controls, then premixed pockets of progressively
P
"resistant" A/F ratio will be allowed to ignite. If r controls, the critical
c
kinetic condition will be progressively reached. In this way a sudden but
controlled burst of heat-release will occur. The constant controlling the rate
of successive ignitions will be adjustable if necessary to match observed
ignition delay; essentially this will be an empirical flame speed.
60
-------
Combustion (C-6)
Diesel combustion and emissions behavior strongly suggest that
heat release is not homogeneous and is controlled by the rate of diffusive
transport. Current models represent these main features by assuming instan-
taneous combustion at.a prescribed stoichiometry, occurring with a rate
limited by a prescribed mixing coefficient. We propose to treat a distribu-
tion of droplet sizes burning with quasi-steady spherical diffusion flames.
Broadening of the heat release zone due to dissociation/recombination will
*
be treated by using the local equilibrium diffusion model for the flame.
The temperature and composition of the oxidizer in which the droplets burn
would be derived from macromixing (the air/spray entrainment hypothesis—
see A/S-4). The rate of heat release will be derived from (1) the burning
rate (molecular transport rate) of individual droplets, and (2) the rate of
entrainment of droplets of different sizes into the flame regions.
It should be noted that the value of the droplet diffusion flame is
not in an improved heat-release prediction, nor do we claim it describes
the actual burning process. Rather, it is a useful artifice to describe in
detail the high temperature diffusion flames which give rise to nitric oxide
and soot. The nature of the diffusion flame (wake-type, ensemble-type,
or single-droplet-type) is not known. But regardless of type, the flame
itself is expected to take on a profile universal to any type.
The use of droplet burning laws to describe measured diesel heat
release rates was first attempted by Lyn (1960), who was not able, to obtain
good agreement. However, in Lyn's model, size distribution was not con-
sidered, nor was the change in droplet burning conditions (key offsetting
effects of oxygen depletion and average temperature increase). Shipinski et al,
(1968) made an effort to extend Lyn's model to account for size distribution
and time-dependent boundary conditions. Tariasawa (1960) had started with
2 2
a size distribution f (d), and applied the burning law d = d (t=0) -£t, which
resulted in a fuel mass history of the form
*Obviously, nitric oxide will be exempted from the equilibrium calculations
and treated separately by kinetic overlay.
61
-------
mf(t)
= exp
When Shipinski used this expression directly for diesel flames, he allowed
a distribution f (d) but attempted to empirically modify ft to account for air
dilution with hot products. This procedure has two shortcomings as follows:
(i) The physically-derived expression for the effect of
ambient conditions (?«,, Y „/ Tj , as given by
Williams (1965) for a thin flame, is not used.
(ii) The expression for m (t)/mf(t=0) derived by Tanasawa
is not valid for gradual oxygen dilution and must be
re-derived (numerically) for time-dependent boundary
conditions.
The above assumptions about diesel combustion may be said to set
the pace or calibre of the entire model. The level of sophistication with
which any remaining single component is approached must not be out of
balance with the combustion component, or else data will be required/
generated which has not been generated/required. For example, the fuel
spray model need not compute a distribution of droplet velocity lags if a
quiescent diffusion flame is assumed. Assumptions about the nature of the
main diffusion burning will largely determine not only the requirements of
the air, spray, and entrainment components, but also determine the informa-
tion available for NO formation, soot formation, and local heat transfer.
Air/Product Entrainment (A/P-4)
Entrainment of hot combustion products in cooler air is crucial to
limiting the NO formation process. We will assume turbulent transport pro-
portional to macroscopic (~1 cm) gradients in temperature, oxygen, or velo-
city set up between the flame zone and the environment. The turbulent
diffusivity will be derived from transport theory. Preliminary calculations
indicate that the turbulent microscale may be on the order of the droplet
62
-------
size (10 to 30 fi). If this is true, the progressive turbulent convolutions
may bring fresh cool air into contact with the hot products produced by
single droplets within the time scale of burning (~1 msec). Otherwise turbu-
lent should only be considered for the subsequent breakup of hot gaseous
regions on the macroscale size (1mm to 1cm).
Nitric Oxide Formation (NO-5)
The extended Zeldovich mechanism will be employed/ with oxygen
atoms in equilibrium with O_:
0/02 = Keg(T,P)
O .+ N2 i=J NO + N
N + O2 ** NO + O
N 4 OH >=* NO + H
Soot Kinetics (S-6)
Both soot combustion and soot formation will be included, with
local kinetics described to allow intense soot activity in the fuel-rich
zones. The mechanism for soot combustion appears to be well enough
established to use measured coefficients, whereas the soot formation rate
will involve an adjustable coefficient.
It should be emphasized that this set of assumptions is a current
tentative guess for an improved model which is still realizable in 12 months,
and does not preclude a number of alternative combustion models which are
being considered. For example, it may be important to describe (1) a macro-
size diffusion flame fed by a cluster of droplets, (2) non-equilibrium hydro-
carbon burning at the flame front to permit O-atom overshoots, and (3)
supercritical effects. The achievement of this model should achieve a sub-
stantial broadening of the engine data which can be described without
adjusting coefficients. The model outlined above would have 27 coeffi-
cients, most physico-chemical properties, distributed as follows:
63
-------
Cycle Thermodynamics
Air Motion
Fuel Spray Delivery
Air/Spray Interaction
Ignition Delay
Adjustable
Coefficients
Universal or Derived
Coefficients
a (conv)
«(rad)
K(equilibrium consts)
h(T) for species i
C (fraction of air in
solid body rotation)
N
Re
C9(drop size distribu- c (discharge coeff)
tion parameter)
D (spray/air diffu-
sivity)
(relative impor-
A,E for chem
tance of chem. ft for evaporation
and physical delay) D for vapor diffusivity
Heat-Release "Spike"
Combustion
Air/Product Entrainment
(flame speed)
[Q.v.c (gas)
D (turbulent diffusivity)
Nitric Oxide Kinetics
Soot Kinetics
Total number
C ,N (formation
s s rate coeff)
Keq(0/02)
A (O + N)
E (O + N J
rate
constants
A ,E (combustion
s~ s~ rate coeff)
21
C.
FLAME STUDIES NEEDED
In Sections A and B above, we have offered the skeleton of an
improved diesel flame model (spray/air mixing with local diffusion flames).
Within this framework, certain pivotal questions arise. The remaining
unresolved issues in our understanding of diesel flames can be set forth as
in Table 10, with possible approaches toward resolving these issues.
64
-------
Table 10
UNRESOLVED QUESTIONS ABOUT THE DIESEL FLAME
Issue
Method of Resolution
Is the air swirl best described by
solid body rotation, potential vortex
flow, or a combination?
Can the spray be described by the
same drop-size distribution law for
all crank angles ? Do differences
in droplet velocity transform the
effective size distribution?
Is the spray/air interaction domi-
nated by either the spray or the air
swirl for the research engine under
consideration? Is wall impingement
significant? Do rotational body
forces need to be considered in the
air/fuel mixing process?
Is ignition controlled by fuel avail-
ability (mixing and evaporation), by
chemical induction times, or by
both?
Is there any particular crank angle
interval (e.g., the "spike1) which
produces most of the NO ?
Is NO production widespread or
primarily confined to localized
regions of heat release ?
Heat release is postulated to occur
in diffusion flames. Are these
flames droplet envelope flames,
wake burning flames, or zones on
the scale of a cluster of droplets ?
Are combustion products rapidly
diluted with air or does the mixture
remain stratified during the heat
release period?
What are the details of NO and
smoke production in a diffusion
flame ?
Measure air speeds as a function
of radius and crank angle using
anemometry.
High magnification photography
or holography of "sprays.
High speed photography will
provide qualitative information.
Analysis of the processes will
be revealing.
High speed movies can provide
qualitative clues. Heat release
traces give indirect evidence.
Time resolved (+ 5°CA) measure-
ment of nitric oxide and other
key species.
Spatially resolved (+ 0.5cm)
measurements of nitric oxide
and other key species. On a
microscale, probably cannot be
resolved; only indirect evidence.
High speed photography may
identify large flame zones, but
limited information is expected.
Measure temperature and species
with spatial resolution. Measure
turbulence levels by anemometry.
Out-of-cylinder steady-state
diffusion flame experiments.
65
-------
Engine improvements due to timing shifts or EGR are relatively
easy to understand (in terms of thermodynamics) and optimize. By contrast,
the effects of air swirl, fuel orifice size, and chamber geometry are poorly
comprehended and engine improvements are only made by trial and error. It
is worthwhile to note that the unresolved questions about diesel combustion
(fluid physics of fuel spray interacting with swirling air, and diffusion flame
structure) correspond one-to-one to these gaps in our ability to optimize
diesel efficiency and emissions. There is little sense in embarking on a
major costly model development without more insight as to flame behavior
than we have at present. Anticipating this need, the single-cylinder engine
was designed to accommodate a variety of diagnostic probes, windows, and
inserts. The flame techniques listed in Table 11 and described in Section
VI are selected to begin to remedy this situation,,
Table 11
SUMMARY OF RECOMMENDED TECHNIQUES
Flame Characteristic
Measurement Technique
Local NO level
Local Temperature
Droplet-generated NO
Air Flow Patterns
Pressure
Flame Appearance
Fuel Spray Characteristics
UV absorption at 2260^/direct sampling
Two wavelength infrared measurements/
cooled film anemometry
Porous sphere simulation of droplet
burning; UV detection of NO radial
distribution
Hot-wire anemometry of motored engine
Cooled film anemometry
High speed photograph with tracers
Radial pressure variation indicates swirl
Piezoelectric transducers
High speed photography
Various methods under study
66
-------
D. DROPLET DIFFUSION FLAME AS A NO -SOURCE
J^,
1. Considerations in Calculating NO from Diffusion Flames
j£
The extent to which significant NO is generated in diffusion flames
J\i
is an extremely crucial point. The flame envelope which surrounds an element
of fuel vapor (such as the mantle of a liquid fuel droplet), has somewhat mixed
qualifications for generating nitric oxide. In Figure 28 we display the relevant
theoretical species and temperature profiles*.
Threshold
NO-formation
Temperature
(~2000°K)
NO formation would
occur in this region
where high temperature
and O-content overlap.
droplet
radius
flame
radius
Figure 28
Temperature and Mass Fraction Profiles for a Burning Fuel
Droplet in the Flame Surface Approximation [Williams (1965)]
On one hand, the peak temperature is near adiabatic with considerable disso-
ciation; on the other hand, the oxygen concentration tapers off to extremely
low values at the flame, so that dissociation of the product species might
have to serve as the principle source of O-laden radicals. There are a
number of complications which must be closely examined to evaluate the
diffusion flame as a source of nitric oxide:
(i) Dropping the flame surface approximation in favor
of a realistic treatment of dissociation behavior.
*For diesel engines, the droplets bum at elevated pressure; upon transition to
the supercritical regime, the mathematical description replaces the boundary
condition of equal partial pressure and vapor pressure with a condition of
equal fugacity [see Rosner (1967), Spalding (1958), Natarajan and Brzustowski
(1970), and Tarifa et al.(1971)].
67
-------
(ii) Integration of dNO/dt over radius and droplet lifetime.
(iii) Drop size effects.
(iv) Dropping the qua si-steady approximation in favor of a
rigorous treatment of the flame envelope movement
during combustion.
(v) Supercritical boundary conditions applied at the droplet
surface.
Let us review each of these items in turn. Detailed analysis of points (i) and
(iv) has been initiated and preliminary results are found in Appendices E and F.
2. Detailed Flame Profile Calculations Necessary
It is necessary to execute a careful theoretical analysis superimpos-
ing N-O kinetics on computed species and temperature profiles/ and to check
the results with simple measurements of NO-profiles in the immediate, neigh-
borhood of isolated burning oil droplets (or in some other idealized diffusion
flame geometry). Theoretical predictions of nitric oxide production in diffu-
sion flames have been carried out by Bracco (1973), and
Bowman and Kesten (1971). Since the results on NO are extremely sensitive
X
to the choice of diffusion flame model, it is imperative that a sufficiently
accurate model be used. For example, "flame-surface" models derived from
the original work of Burke-Schumann (1928) are popular because one can
obtain a closed-form solution. However, the flame-surface models are inade-
quate for NO prediction because they overestimate the peak flame temperature
J\.
and correspondingly manhandle the species profiles . For a more sophisticated
approach, one must turn to formulation requiring numerical integration, such
as the Coffin model (1957), which is based on H-C-O chemical equilibrium
throughout the field, or to recent studies which incorporate certain aspects of
non-equilibrium chemical kinetics into the diffusion flame model [Kassoy and
Williams (1968), and Fendell (1967)].
*The merit of the Burke-Schumann approach lies in accurate description of
the burning rate, which proves to be more less insensitive to species and
temperature profiles in the near field of the droplet [Wilson (1970)].
68
-------
As a first simplifying step, we neglect bulk convection and non-
steady effects and consider only the balance between the molecular-
diffusion transport term and chemical source terms. We postulate that this
flame balance is independent of the geometric configuration. That is to
say, it is a universally valid approximation and will apply equally as well
to burning droplets, sprays, jets, or gaseous counter-flow. It is assumed
here that all of the reactions are in equilibrium with the exception of the
NO -formation reactions .
x
If we further assume (1) diffusion pairs with equal diffusion
coefficients and obey Pick's Law, and (2) unit Lewis number, and (3)
spherical symmetry, then the governing equations take the form
j~ = V
= V' (PDVh)
where
Y, = -u
J ij
Y. = element mass fraction
Y. = species mass fraction
H . = number of atoms of element J in molecule of specie i
M = atomic or molecular weight
Once the element mass functions Y and enthalpy h have been
solved as a function of radius (see Appendix E) , the actual species distri-
bution can be found from an equilibrium analysis. Equilibrium composition
through the diffusion flame is accomplished with the aid of the NASA One-
Dimensional Equilibrium (ODE) Program. Results are shown in Figure 29 for
N ,=.51, where the thin flame solution is given for comparison. Note the
effect of dissociation. Clearly the NO production rate would differ markedly
for the two models .
69
-------
T, °K
Burke-Schumann
Equilibrium
I I f t - I t
I JL
. 1
01 g
o
s
PL,
CD
.001
.0001
.78 9 10 11 12 13 14 15 16
Dimsnsionless Distance *7= r/rj
Figure 29
70
-------
3. Integration over Radial Coordinate and Burning Time
Another point concerns the estimation of the contribution of burning
droplets to NO production. Because the lifetime in the flame zone is quite
J^
brief, it is not enough merely to establish the production rate (in ppm/sec)
as a function of radial coordinate for given droplet diameter and ambient
composition and temperature. One must estimate the cumulative flux of NO
outward from the droplet at some radial point far from the flame (say, 10
flame radii out), and combine this with the fuel burning rate (e.g. kg/sec)
to obtain the grams NO/kg fuel produced. This emission level will be valid
only within the framework of the quasi-steady model—i.e., it will only apply
instantaneously for a given droplet size. In reality the droplet size changes,
and one must average the NO -production rate over the lifetime of the droplet.
J\
Then it will be possible to evaluate NO production by burning diesel fuel
JC
droplets, and to optimize the size distribution of fuel spray nozzles to
minimize NO formation.
4. Importance of Droplet Size Distribution
The preliminary calculations of Bowman and Kesten (1971) show that
the droplet emission index (gNO/g fuel) goes up with the square of droplet
diameter. The greater NO emission was due to the broader diffusion flames
J\
associated with larger drops, such that NO-breeding gaseous products take
longer to move out from the hot flame. In any case, droplet sizes must be
established if possible.
5. Asymptotic Analysis of Non-Steady Effects
The quasi-steady approximation is often made to simplify the
analysis of diffusion flames. Under this hypothesis, the burning time is
assumed much longer than other potentially complex relaxation times such
as (1) the time for thermal equilibration of the droplet interior, and (2) the
time required to set up a diffusive mixing field centered about the flame
envelope. The validity of this approximation has been analyzed by intro-
ducing non-steady effect (2) into the equations and observing the predicted
71
-------
flame behavior. An analysis by matched asymptotic expansions was con-
ducted to distinguish the droplet vicinity (quasi-steady) and the unsteady
effects which predominate further out in the field, and which may affect the
flame position. Typical values for the evaporation constant and the diffusion
_o 2 —1 9
coefficient (yS«10 cm /sec, B^wlO cm /sec) suggest taking the expan-
sion parameter
Details of the analysis are presented in Appendix F; here we outline the
major conclusions.
After the analysis given in Appendix F , one obtains the following
results: The flame radius (r ) is given implicitly by the equation
» +Yf ) r. m rfl(t)-rf(t)
2VDt
-m(t)
f (t'-t)erfc dt' - '*" p DCrfi(t) -rf (ta
0
where r (t) and rn/t) are given by the quasi-steady solution in the inner region
and hence are known. Here Y is defined by Y = Y£ f"M/(j/"-i/')M, where v are
stoichiometric coefficients. This implicit equation for flame radius was numer-
ically solved, and the results are given in Figure 30.
Likewise, the flame temperature can be determined as shown in Appen-
dix F and is given by the expression
T Y 4- Voo Y -4- T Voo
T = * £~ OX f" t- *OX
YSx+V
where T = CpT/Q.
This equation shows that flame temperature is constant in spite of
the motion of the flame. This results runs counter to the exact numerical
solutions of Kotake and Okazaki (1969) which exhibit temporally varying flame
temperature. This discrepancy conceivably is attributable to the neglect of
liquid-phase heat conduction in the present case.
72
-------
Figure 30 shows that over this envelope of validity the flame to
droplet radius does not change very much and may perhaps be treated as
constant (this is not examined here). This is not to be interpreted, how-
ever, as an acceptance of the qua si-steady theory which predicts such a
constant ratio somewhat in excess of the mean value obtained by the
unsteady theory.
Figure 30
FLAME ENVELOPE EXPANDS RELATIVE TO DROPLET SURFACE
(BOTH DECREASE DURING BURNING)
Conditions:
CIOH20
Air
I508°K, lOOatm
2
= 0.3cm/sec
/?=2xl
-------
V. ASSESSMENT OF EXISTING MODELS
CHAPTER SUMMARY
The following models were examined closely:
(i) The Northern Research and Engineering Corporation
(NREC, hereafter) Model [Bastress, Chng and Dix (1971)]
(ii) The CAV Model [Khan, Greeves , and Probert (1971),
and Khan and Greeves (1973)]
(iii) The Cummins Model [Shahed, Chiu and Yumlu (1973)]
Each model's potential usefulness and applicability was examined based on
its adherence to known physical mechanisms, ability to predict the emissions
behavior of one single-cylinder engine, and the need to readjust empirical
coefficients. Although each model appeared to leave room for improvement,
the logic and insight displayed in these earlier models have been invaluable
to our own efforts.
The state-of-the-art in diesel modeling is perhaps best seen in
Table 14, which shows alternate ways of treating the key processes. There
are four areas where current models have made simplifying approximations
and where significant advances may be attempted:
Air Motion
Swirl is not considered by the NREC or Cummins models, and
is empirically inserted into the CAV coefficient for entrainment.
An explicit description of the swirling air flow offers improve-
ment.
Fuel Spray and Mixing
Droplets are not considered in any of the models evaluated
in this section. Dropsize distribution, spray dynamics, and
droplet entrainment by crossflow as a function of injection
parameters would appear attractive at this time.
74
-------
Ignition Delay
Ignition delays are prescribed in all models evaluated here.
It is recommended that droplet evaporation, formation of
premixed zones, and spontaneous chemical ignition be con-
sidered.
Combustion (Heat Release)
Coefficients of exchange between uniform pockets of reactants
are fit to heat-release data in all existing models save the
early work of Lyn (1960) as extended by Shipinski (1968) . A
realistic treatment of diffusion flame gradients can obviate the
need to invent assigned fuel/air ratios . The flame structure
is especially critical to pollutant formation.
Existing models have opted for phenomenological treatments of these pheno-
mena in lieu of descriptions of the underlying mechanisms. Two serious
repercussions from this approach are (1) that coefficients of existing models
must be custom fit laboriously to each engine, and (2) that the range of
parameter variation often excludes chamber geometry, fuel dispersion, air
swirl, water injection, EGR, and other emissions-sensitive parameters. .
-------
Table 14
ALTERNATE APPROACHES FOR MODEL COMPONENTS
Cycle
Therm odyna mlc«
Fuel Spray
Delivery
Air/Spray
'En trainmen!
Delay to
Spontaneous
Ignition
Heot-Ralease
Overshoot
Air/Product
Entrelnment
Nitric Oxide
Kinetics
Soot Kinetics
Given the Initial state,
chamber geometry,
composition, heat
release rate and mass
injection, describe the
suite of the gas (P,T)
a* a (unction of crank
Given the Inlet angular
momentum and RPM,
describe whatever
rl, squish, and
turbulence character-
itlcs are needed for
mixing and combustion
models.
an the fuel sched-
ule, number and dia-
meter of orifices,
describe the droplet
dispersion spreading
angle, and Injection
velocity.
Given the air motion
and fuel spray, des-
cribe the mecromixlng
of spray before bun
can be considered.
ng
:n the state and
motion of the air at
start of injection, des-
cribe the delay before
Irst notlcable net heal
release.
Account for the "spike'
In the heat release
profiles.
Given the fuel and air
distribution, describe
tho heat release rote
and (lame T and spe-
cies profiles.
Given the air motion
and combustion pro-
ducts, describe how
turbulent transport
will disperse and cool
the products.
Given the species and I Given the species and
temperature fields,
compute th? net accu-
latlon of nitric
oxide.
tenperarjrc fields,
compute the net «cc»
ilatlon of soot.
Cn
AJubailc conpres-
ftlon and expansion
(no heat loss).
r-il
isidered (spray
structure not con-
sidered) .
ment of spray upon
injection; no velocity
lag.
itantaneous Igniti
at start of injection
(no delay).
EZ3
itantaneous heat
release from all
prepared fuel.
Instantaneous burning
upon (uel availability.
Burning specified to
occur at siolchio-
S-U
Soot formation Ignored.
Instantaneous mixing
of products (uniform
composition).
Empirical correlation
ftrburary ignition delay
Arbitrary heat release
(
tir.pincal correlation
it with
tusuble heat loss,
cycle averaged.
q£3SUbie hcat
released durlno/com-
bustion only
motors (e.g. , peak
terr.r.cratxjre).
meters (e.g., overall
A/T).
from both chemical rati
and mixing rat? (adjust
able coefficient).
Homogeneous let
(droplets not recog-
nized); adjustable
Spreading coolii
.,.,„
ci-Ur.der-averaced T.
JV/T. P.; adj-j stable
coefficients.
flame based on droplet
clusters. Shape arbi-
trary.
with adjustable
coefficients.
live heat loss from T
and Tw coefficients
assigned.
GE3
Same as T-4 but trans-
fer coefficients derived
or measured.
tervation of angular
momentum.
width derived from
momentum and
enffainment,
cylinder of fuel in
cross.'low (without
drop si
physical transport
hcdt-up. evaporotl
. turbule
diffusivity derived d
measured.
neous tcnltl
prernlxed pockctl
progressively rcais-
pyrolysis ar.d carbon
ac.clorr.cration.
Drjplet diffusion
(lane: m = !eJd o!
me shti"t, use
chemical equili-
brium dl!.'u3ton
profile.
['tdtne speed model foi
prefixed mixture co.v
Pmtto rate of energy
release due to pre-
Miicru-sizc diffusion
uilibriu
profiles as In C-6.
Same as 0-3 but
coefficients derived
from physics.
(Not applicable: spra
trajectory determined
from A/S t^bdela.
cal soot formation
rate (e.g. , higher In
fuel-rich recions).
Jet tn cross-
flow analyzed to gl
peel off of droplets
l turbulent diffu-
sivity based on l
gradients.
local hot tones
radlau independently
liquid fuel droplets,
Including trajectories
and droplet boundary
layer.
lgTawlrl model with
body lorcea analyzed to
give bo 111 s tic droplet
In partial equilibrium
with CO tut not with
droplet diffusion flame
possible woke
-5 model).
from conservation
equation for turbu
lence
irop-slze distribu-
it log
pingement
analysis.
(a) upper li
normal
(b) Rosln-ftammler
(c) Nukigama
(d) Bl-dlsperse
Account for
and falling O, due to
neighboring cToplcts.
(Anend C-6J. Droplet
spacing must be
assumed.
lations allowed.
ilibrium O-
otom concentration
(complete C-H-O-N
scheme required) .
Upper curve:
Lower curve:
Current models
Tentative model for Phase II
-------
A. THE NREC MODEL
1. Approach to Modeling the Mixing, Heat Release and
Pollutant Formation
The mixing and heat release process is entirely artificial and pre-
scribed by four parameters, C , C , F, and AT. Two stages of heat release
are represented, following Lyn (1963):
(a) An initial triangular "spike" of heat release having
width specified by C_.
(b) Heat release according to a Gaussian profile with
half width C.:
f inj
-37T ~ * exp
Cl
In both cases, the chemical reactions are instantaneous (equilibrium prevails).
C, and C9 are mixing coefficients prescribing fuel availability. During the
second heat release plateau, heat release occurs at a specified mixture
ratio F, often taken at stoichiometric to simulate a diffusion flame zone.
Excursions from the nominal value F are permitted during the spike; these
excursions are characterized by the parameter AF.
Subsequent dilution of combustion products by air is specified by
a parameter C«. Successive pockets of combustion products are distributed
O
randomly in the chamber and no attempt is made to describe temperature or
concentration gradients. Heat transfer is prescribed in proportion to the
-3
temperature of each pocket by an exchange of coefficient C. (10 ft-lb
R - CA -ft ). The kinetics of NO formation are represented adequately
by the Zeldovich mechanism, although in practice the program is difficult to
use because NO production is a tabular entry based on given C/H ratio,
pressure, F/A ratio, and temperature. Soot is not considered.
77
-------
There are a number of interacting systems and processes which
seem important to incorporate in an improved model for diesel combustion
and emissions. These are outlined in Figure 31(a) . The approach of the
NREC model is outlined for comparison in Figure 31(b), with bracketed
criticisms, omissions, and characterizations.
The main objection to this approach is obvious . The incorporation
of six parameters which are to be specified arbitrarily (C , C , C , C.,
_ J. £ G fc
F , 4F) reduces the method to little more than a data correlation technique.
The physics of fuel spray macromixing, ignition, and diffusion-controlled
burning has been bypassed with these six parameters. It is not clear how
to relate these six parameters to measurable or calculable fluid physics
properties. Such a method will allow the correlation of data from a tightly
defined family of engines and will allow a certain amount of extrapolation
to the boundaries of this family. However, it will tell very little if anything
about how engine design parameters are related to fundamental combustion
and pollution-formation phenomena. The method cannot be used for new
optimum engine design.
To be more specific in our criticism, the program in effect specifies
a priori the local stoichiometry to which the products of combustion will imme-
diately mix and the quench rate to which these products are then subjected.
It further specifies a priori the "delay" (preparation time) which injected fuel
experiences before it bums. Furthermore the program assumes that fuel vapor
combines only with fresh air and that no combustion product dilution of the
air prior to combustion is present. All of these points are factors which must
be dictated by the physics of fuel spray-air interaction and diffusion flame
mechanics.
Such a model cannot be used as a basis for soot formation predic-
tion. Soot formation kinetics depends on the lifetime at high temperature of
rich zones. Clearly these features are built into the program a priori rather
than derived.
78
-------
Figure 31 (a)
Figure 31(b)
BLOCK DIAGRAM OF SYSTEM .INTERACTIONS
FOR DIESEL COMBUSTION
BLOCK DIAGRAM OF NREC MODEL
[ ] = CHARACTERIZATION OR CRITICISM
COMPR
PRESCRIBED SWIRL
EGR
:SSION WATERING
TURBOCHARGE.
TEMPERATURE .
INJECTION
COMPRESSED
SWIRLING AIR
CD
PRESCRIBED DELIVERY
SCHEDULE AND
INJECTOR CHARACTERISTICS
LIQUID FUEL SPRAY
MIXING
ENTRAINED
DROPLETS
TURBULENT
DILUTION,
HEAT TRANSFER
AND
EXPANSION
I
EVAPORATION
AND MIXING
PREMKED
REACTANTS
DIFFUSION-CONTROLLED
BURNING
("MICROMKING")
IGNITION
MIXTURES SUBJECT TO
COMBUSTION WAVE
PASSAGE OF
COMBUSTION WAVE
HOT PRODUCTS OF COMBUSTION
DILUTED AND COOLED
PRODUCTS
[NO INJECTOR
CHARACTERISTICS]
COMPRESSED AIR
[NO SWIRL]
[NO HETEROGENEOUS
COMBUSTION OR
DIFFUSION FLAMES]
LIQUID FUEL
[NO SPRAY DETAILS]
[SPECIFIED EVAPORATION
+ MIXING RATE]
[SPECIFIED MIXTURE
RATIO]
PREMKED
REACTANTS
[HEAT TRANSFER PRESCRIBED]
[DILUTION PRESCRIBED]
[SPECIFIED
IGNITION DELA'
MIXTURES SUBJECT TO
COMBUSTION WAVE
[INSTANTANEOUS COMBUSTION]
[EQUILIBRIUM CHEMISTRY]
HOT PRODUCTS OF COMBUSTION
DILUTED AND COOLED
PRODUCTS
[ZELDOVICH
NO KINETICS]
[SOOT NOT CONSIDERED]
-•^EMISSION-LADEh
EXHAUST
-------
2.
Predictive Capability of the NREC Model
The NREC model has been run under the following conditions which
match the baseline test matrix of the experimental single cylinder tests:
Engine:
Bore
Connecting Rod Length
Crank Radius
Clearance Volume
5.50"
12.00"
3.00" 3
0.00485 ft (CR= 17:1)
Fuel System:
Fuel injection schedule shown in Figure 32
0
inj
/ 20' 25CABTDC
Q = adjusted to give desired A/F (maximum
126 mmVstroke)
Operating Parameters:
Air Temperature
Initial Air Pressure
Fuel Temperature
Engine Speed
100°F
1 atm
100°F
1500, 1800, 2100 RPM
Rate of Inj ection
(slugs/°CA)
4.6 x 10
-7
9. .-3 0. . 0. .+2
inj in] inj
Crank Angle, BTDC
Figure 32
Fuel Injection Schedule
TDC
80
-------
In addition, a number of model parameters were first arbitrarily set as
follows:
Fuel Vaporization Rate (C ) 10°CA
Fuel Burning Rate (C ) 10°CA
£* i
Dilution Rate (C ) .05°CA
Heat Transfer (C4) 10~3 Ibf-ft/ft2-°R-°CA
Fuel Mass Fraction AT .01
Increment
The results are compared with observed emissions in Table 15.
Table 15
$ RPM Timing Runt Observed NO (ppm) Calc. NO (ppm)
.15
.33
.63
.62
.60
1500
1500
1500
2100
2100
-20°
-20°
-20°
-25°
-15°
33-8
33-11
34-2
35-11
39-4
620
2000
2900
2420
1200
611
1091
726
783
554
Generally, while these original NREC predictions agreed at low load, the
predicted NO levels were a factor of three"too low at high load. The rela-
tive trend with increasing load was opposite to that observed, and the
relative trend with timing was correct, but quantitatively less (3% reduction
rather than 5% reduction per °CA retard).
Certain adjustments to the NREC model parameters were made and
the results were much more reasonable. The original and revised sets are
given below. First Second
Original Revision Revision
Fuel Vaporization Rate (Cj 10°CA 17°CA 17°CA
Fuel Gas Burning Rate (Cj 10°CA 5°CA 10°CA
Dilution Rate (C) .05°CA .O^CA'1 .O^CA*1
Heat Transfer (C4) 10~3 2 x 10~3 10~3
Fuel Mass Fraction Increment (4F) .01 .001 .003
81
-------
In order to explain the model's behavior, it is useful to recall that
the original set of parameters gave NO decreasing with load (see Table 15).
J\.
Apparently no air was left for combustion when the last several "packages"
of fuel were ready. Our first remedy was to decrease the air dilution rate
(C-), hoping to save air for later in the cycle. The result was unsatisfactory.
Closer analysis showed all of the air consumed during the ignition "spike",
which in a real engine amounts to only 5 to 10% of the heat release. The root
cause of the spike's appetite for air was the distribution of fuel/air ratios
specified during the ignition spike. With AT = .01, the distribution was so
wide that extremely lean "packages" were created—so lean that all of the
air was promptly assigned. The remedy for this was to reduce the width of
the spike (C2) and to reduce the F/A dispersion (AF). These changes gave
rough agreement with the data; fine tuning was then done by adjusting C, .to
favor delayed burning. The "tuned" results are shown in Table 16. The model
could have been further tuned by assigning an individual set (C, , C9) for each
-L £
fuel package.
The agreement on load effects (Columns 2,8,9 and 10) was better
than 15%. The 50% reduction in NO with 10° CA retard was also well simu-
lated (Columns 11 and 12). The reduction of NO with increased speed was
overpredicted by the model (Column 7). Turbocharging effects were well repre-
sented while increased air effects were overpredicted. Neither compression
ratio changes nor rate of injection effects were very well described.
In short, the model performed reasonably well only on emissions beha-
vior with changes in load, timing, and turbocharging. The predictive capability
of the NREC model for variations in air temperature, compression ratio, RPM, or
rate of injection appeared inadequate for practical use by engineers.
3. Versatility and Adjustment of Parameters
The NREC model is not set up to investigate the effect of EGR, water
injection, air swirl, fuel dispersion, or chamber shape. In fact the residual
exhaust is ignored in the cycle thermodynamics; pure air is taken as the intake
charge. Thus the model is quite limited in its ability to predict a wide variety
of emissions control variations.
Three iterations were necessary to find a set of six parameters which
gave reasonable baseline agreement. However, when extended beyond the
baseline the predictive capability was quite limited, so three iterations is not
unexpectedly few. A more rigorous procedure would be to attempt to fit the
heat release profile by adjustment of the six parameters.
82
-------
Table 16
COMPARISON OF NREC PREDICTIONS TO SINGLE CYLINDER EMISSIONS
Description
. w
>» u
fc* QJ
i c
•2 E
S.
V}
C
o o
C 7;
C
u
a.
Cj Fuel Vaporization Rate (°JCA)
C. Fuel Gas Burning Rate ( CA)
C. Dilution Rate (°CA )
C . Heat Transfer
4 (ft-lbs/ft2-°R-°CA)
4F Fuel Mass Fraction Increment
P Mean Fuel/Air Ratio
Original
• 10
5
.05
.001
.01 .
.01
Pa (Ib/ft2)
Ta (°R)
VC Clearance Volume (ft3)
Fuel Schedule (as represented
Timing
RPM
.NO Predicted
X
N6 Experimental
#1
17
10
.01
.002
.001
.066
1900
560
.00485
4.6
'-20°
.45 -.
1500
2300 .
1800
#2
17
10
.01
.00.1
.003
.066
3800
560
.00485
4.6
-20°
.45
1500
1875
1800
#3
17
10
.01
.001
.003
.066
1900
560
.00485
9.2
-20°
.•45
1500
1936
1910
#4
17
10
.01
.001
..003
.066'
1900
660
.00589
3.9
-20°
.45
1500
2300
2060
#5
17
10
.01
.001
.003
.066
1900
560
.00485
. 4.6
-20°
.45
1500
1700
920
#6a
17
10
.01
.001
.003
.066
1900
56-0
.00485
6.8
-20°
.45
1500
1925
825
#6b
17
10
.00715
.001
.003
.066 .
1900
560
.00485
9.2
-20°
.45
1500
1916
990
#7
23.8
14.0
.01.
000715
.003
.066
1900
560
.00485
4.6
-20°
.45
2100
700
1670
#8
.17
10
.01
.001
.003
.066
1900
560
.00485
4.6
-20°
.15
1500
1500
620
#9
17
10
.01
.001
.003
.066 .
1900
560
.00485
4.6
-20°
.33
1500
1500
2000
#10
.17
10
.01
.001
.003.
.066
1900
560.
.00485
4.6
-20°
.60
1500
1500
2900
#11
17
10
.01
.001
.003
.066
1900
560
.00485
4.6 .
-25°
.60
2100
2100
2420
#12
17
10
.01
.001
.003
.066
1900
560
.00485
4.6
-15°
.60
2100
• 2265
1200
CO
CO
#2: New Baseline
#3: Turbocharge (Double Air Pressure)
#4: Increase Air Temperature by 100°F
#5: Decrease Compression Ratio to 14:1
#6a: High Rate of Fuel Injection
#6b: Very High Rate of Fuel Injection
#7: Increased RPM
#8-10: A/F Variation
#11: Increase RPM
#12: Retard
-------
B. THE CAV MODEL
1. Approach to Modeling the Mixing, Heat Release, and
Pollutant Formation
In contrast to the NREC approach, the CAV model considers the mixing
of a fuel spray with high density swirling air. Khan, Greeves and Probert (1971)
derived their mixing model from Grigg and Syed (1970) . Starting (as all models
must) with a prescribed fuel schedule, a conical plume of half angle Ofis des-
cribed, as shown in Figure 33.
Injector
Orifize size, d
Fuel pressure, P
26.
Advancing Front
Air entrained to
conserve momentum
Figure 33
Conical Plume Schematic
The front advances according to the Schweitzer (1938) expression
d P 1/2
2 aff
t[p(intake)/p(stp)]
where the notation is given in Figure 32. The rate of air entrainment is such
that the plume density is constant. Thus air fills in the widening plume at
a rate which can be derived from geometrical considerations:
dt
2 3
E p Trtan 0, r
r a f
t[p(intake)/p(stp)]
where E is an "entrainment" coefficient, empirically fitted as a function of
speed and swirl to data on soot emission. A similar "macromix" model has
been worked out for wall impingement.
84
-------
Heat release is described as follows: An ignition delay is specified
from measured chamber pressure data, followed by a triangular heat release
"spike" with a base arbitrarily set at 6 CA. The major heat release occurs by
a diffusion-controlled process within the entrainment cone defined above.
The rate of diffusion (and instantaneous heat release) is given by a quasi-
Fick's law proportional to not only the concentration difference but also to
the velocity of the advancing conical plume:
.
a cum T^. d r
dl - = D ^T
/ \ n
- (ma)bum]
where D'(cm ) is a "diffusion coefficient" selected as an empirical function
of swirl and speed to give good agreement with heat release data. A peak
temperature T for NO formation, unique to this zone of hot products, is
P
derived from equilibrium considerations. However, heat loss occurs at the
same rate as for the cylinder-averaged temperature.
Let us turn to the assumed pollutant production mechanisms. Within
this heat release zone, a fuel-rich zone (denoted \$$\ in Figure 34) is defined
nFuel lean and .
unmixed
ty///\ Micromlxed zone
P..-..I Fuel rich and
unmixed
• Realistic profiles
.Approximate 3-zone model
XP
T
plume
cyl
•
S\
/ ' \
/ ^
•r \
/
Section A-A
(Temperature)
,rw.v». i*-. v.\ wi/ mill 11
Section A-A
(Equivalence Ratio)
Figure 34
THREE-ZONE MODEL FOR MICROMKING
AND POLLUTANT FORMATION
85
-------
where smoke is formed at a rate given by
dm V
where the subscript s denotes the fuel rich soot zone, and the remaining
parameters are set by an array of recipes as follows:
C =9.3x10 (Best fit to soot data)
o
E =40 kcal/mole (Best fit to soot data)
S
n =3 (Best fit to soot data)
m
, . , (Equivalence ratio for unmixed
(mJ. unmixed . ._. ,
f bum reactants in entire plume, weighted
to fuel rich in early stages of burning)
T = T j (Ranges from compression temperature of
unburned air up to final temperature of conical
plume after combustion is complete)
V = Volume of fuel rich zone, given by thermodynamics, and
S the definitions of and T .
s s
Soot oxidation is acknowledged to occur but is not included because it did
not seem to heavily influence the emission data, relative to soot formation.
Nitric oxide is formed by the modified Zeldovich mechanism within
the zone of hot products , where the temperature T and equivalence ratio are
given by the micromixed fuel and air taken to equilibrium. The pre- exponential
factor was boosted a factor of 5 over the literature values; otherwise NO
emissions were predicted too low. Peak temperatures and oxygen profiles are
ignored with the details of the diffusion flame, so this correction is not
unexpected.
Figure 35 (b) gives the block diagram of the CAV approach.
*Adopted by Khan and Greeves (1973) as a significant improvement over the
earlier model [Khan et al (1971)].
86
-------
Figure 35 (a)
BLOCK DIAGRAM OF SYSTEM INTERACTIONS
FOR DIESEL COMBUSTION
Figure 35(b)
BLOCK DIAGRAM OF CAV MODEL
' Characterization or Criticism
I FUEL SUPPLY
^
COMPR
r PRESCRIBED SWIRL
ESSION SERING ™IECTION
ITURBOCHARGE
TEMPERATURE
PRESCRIBED DELIVERY
SCHEDULE AND
INJECTOR CHARACTERISTICS
SSS?S LIOUID FUEL SPRAY
00
•>!
TURBUL
DILUTI
HEAT TRA^
AND
EXPANS
'
.MIXING ""*:><• ^"
EVAPORATION
AND MIXING
ENTRAINED 1 PREMKED
DROPLETS | REACTANTS
ENT
ON,
JSFER
ION
DIFFUSION-CONTROLLED
IGNITION
BURNING MIXTURES SUBJECT TO
("MICROMKING") COMBUSTION WAVE
1
PASSAGE OF
COMBUSTION WAVE
HOT PRODUCTS OF COMBUSTION
NO- FORMATION f
^S
N
COMPRESSION
1
INJECTION
[Droplets not
Considered]
[Evaporation
Instantaneous]
COMPRESSED pa^pntTQ PTTPT ^PRAV
SWIRLING AIR MACROMIXING GASEOUS FUEL SPRAK
ENTRfl
FUEL/
DILUTION
plume model]^ '
_^ ^^
^AIR MICROMKING REAC'
[Prescribed
[Mlcromlxlng between Ignition Delay]
homogeneous zones;
IIXED
CANTS
no gradients] MKTURE SUBJECT TO
COMBUSTION WAVE
[Prescribed triangular
heat release
with 6°CA base)
HOT PRODUCTS OF COMBUSTION
'
[Radiative heat loss [Soot format
not local] in unbumed m
S
on
^"*V 1 1
Dii.iiTpn A^O ^OOT.FP , p_p.MisqTnN-TJinpNj nn.iiTFn AMP CO^LFP / >
PRODU
rT» fc\ PYHJ
\UST
1 PRODUCTS "^(EMISS
/
V EX
ION-LADE^
HAUST
1
-------
Our main criticisms are as follows:
(i) The macro-mixing model may work as a mental construct,
but is far from reality as high speed movies reveal (CAV
acknowledges this). For example, the spray cone pene-
tration time is small compared to the heat release time;
some other mechanism of entrainment such as crossflow
stripping of droplets appears more likely from the photo-
graphs . The effect of swirl would be more fundamental
in a crossflow model and E = f(swirl) would be more
mechanistic.
(ii) Droplet details such as the dropsize distribution, evap-
oration, or local diffusion flames are not treated. This
would be a more fundamental approach to the effect of
orifice size, for example.
(iii) The smoothing out of gradients (temperature and conc-
entration) in the heat-release zone will underpredict
nitric oxide. The established Zeldovich mechanism
should not have to be "doctored" as in the 1971 CAV model,
(iv) The NO emissions are quite sensitive to heat losses
from the heat-release zone, yet the CAV approach was
to assign the same radiation as for the entire chamber.
Doubtless this decision was tied to the underprediction
of NO (the need for higher temperature). This again
reflects the need for detailed flame profiles where heat
losses would be allowed along with actual "undipped"
peak temperatures.
(v) The rate of micromixing or molecular diffusion is
assumed proportional to the plume velocity r . It would
be more fundamental to derive a turbulent diffusivity
from Prandtl mixing length considerations using the air
swirl and/or piston speed.
In many respects, the CAV model is the most advanced model currently in
existence, particularly in regard to the fuel/air mixing and soot formation
processes.
2. Predictive Capability of the CAV Model
The CAV model was exercised and predictions compared against
our single cylinder data. Results were not available at this writing.
88
-------
3. Versatility and Adjustment of Parameters
The CAV model is expressly structured to handle a wide variety of
parameter variations including variations in air swirl, engine speed, load,
rate of injection, timing, and compression ratio. According to communica-
tions from the CAV, the program is not general enough to accommodate EGR,
water injection, turbocharging, fuel orifice size, or pilot injection as para-
meter variations. The computer code was not available for direct evaluation.
The diffusivity coefficient, D', must be adjusted using cumulative
heat release diagrams from precise cylinder pressure measurements. Appa-
rently, a unique set of coefficients must be derived for each engine speed
and swirl level.
C. THE CUMMINS MODEL
The approach of the Cummins model to the following physical aspects
which should be included in a model of NO formation in a diesel cylinder will
H
be commented on:
1. Aerodynamics of the macromixing of air/fuel and air/
combustion products
2. Local mechanism of combustion, including diffusion
if necessary
3. Thermodynamic cycle processes
4. Detailed description of the chemical kinetic mechanism
of NO formation
x
The model presented by the authors correctly represents cycle thermo-
dynamics but incorporates empirical treatments of NO kinetics and rate of heat
JC
release. With this simple approach, and with suitable adjustment of para-
meters, the model is able to simulate the NO effects of compression ratio,
J\.
inlet pressure, temperature, composition (EGR), injection timing and schedule,
and engine speed. However, mixing is not treated. Consequently, the NO -
X
influences of swirl, spray parameters, and bowl geometry cannot be explained.
89
-------
The model adopts the assumption of stoichiometric burning, which
is quite remarkable despite its common acceptance. The authors note that
11 stoichiometric combustion represents approximately the mid range of com-
bustible mixtures". An alternate justification, which seems more reasonable
to us, is that nonpremixed combustion manifests.itself in the form of a
diffusion-controlled flame, which will self-regulate itself to be "fed" stoi-
chiometrically.
The next aspect of the model which deserves comment is the rate
expression used to describe NO formation. Based on the Zeldovich mecha-
nism for NO formation, the authors' rate equation contains two parameters,
A and E, which the authors have arbitrarily adjusted to give agreement with
NO emission data for a number of diesel engines. The difference between
the adjusted (A, E) and the values calculated from basic kinetic data (A =
1.3 x 109 mole °K/atm ' Btu °CA, and E = 137 kcal/mole) might be attri-
buted to other effects, such as mixing, which have been inadequately
treated in the model.
Also regarding the empirical parameters, the parameter A includes
the square root of the oxygen concentration. It is consistent with the
authors' assumptions of equilibrium combustion, 0=1, and no mixing to
assign a constant value of O concentration (say 1%). In reality during NO
formation the oxygen concentration may change by a factor of 3 due to dilution
or oxygen consumption. The omission of a ,rigorous description of O-atom
profiles and temperature in the flame is also serious (although common to
all existing models).
We are also concerned with the lack of mixing between a stoichio-
metric package of combustion products and excess air in the model. This can
have two ramifications on predicted nitric oxide formation. First, the cooling
effect of air dilution is neglected and thus NO production may be overesti-
j£
mated. However, without air dilution, the amount of oxygen available for NO
90
-------
formation is also artificially curtailed, reducing the amount of NO predicted,
By holding the oxygen concentration constant throughout the process, the
authors have partially compensated for the latter effect.
Finally, concerning the heat release calculation, the model has
been exercised with two options: (1) empirical burning law following Lyn,
and (2) a spray mixing model which is mentioned but not elaborated. It
would be of interest to have details of the spray mixing model and to learn
whether the two calculations of heat release give comparable results.
In commenting on the preliminary Cummins model, it is useful to
reflect on two key characteristics which a model should incorporate in order
to be useful to an engineer in explaining observed system performance or
predicting performance of untested systems. First, as the authors have
pointed out, assigned constants should be universal to the entire class of
direct injection engines. Second, the model should be able to accurately
predict the performance and emissions behavior over a broad range of com-
bustion system parameters. In order to accomplish these goals, the model
must curtail empiricism in favor of principles describing underlying mecha-
nisms. Unfortunately, many of these mechanisms are not understood and
like all other current models, the preliminary Cummins model must rely on
empirically adjusted constants. A block diagram is given as Figure 36(b).
91
-------
Figure 36(a)
BLOCK DIAGRAM OF SYSTEM INTERACTIONS
FOR DIESEL COMBUSTION
PRESCRIBED SWIRL
EGR
COMPRESSION WATERING
TURBOCHARGE .
TEMPERATURE
INJECTION
COMPRESSED
SWIRLING AIR
ID
to
PRESCRIBED DELIVERY
SCHEDULE AND
INJECTOR CHARACTERISTICS
LIQUID FUEL SPRAY
MIXING
ENTRAINED
DROPLETS
TURBULENT
DILUTION,
HEAT TRANSFER
AND
EXPANSION
I
EVAPORATION
AND MIXING
PREMKED
REACTANTS
DIFFUSION-CONTROLLED
BURNING
("MICROMKING")
IGNITION
MIXTURES SUBJECT TO
COMBUSTION WAVE
HOT PRODUCTS OF COMBUSTION
_JEMISSION-LADEN
EXHAUST
PASSAGE OF
COMBUSTION WAVE
DILUTED AND .COOLED
PRODUCTS
Figure 36(b)
BLOCK DIAGRAM OF CUMMINS MODEL
CHARACTERIZATION
OR CRITICISM
[NO INJECTOR
CHARACTERISTICS]
COMPRESSED AIR
[NO SWIRL]
LIQUID FUEL SPRAY
[NO SPRAY DETAILS]
^rf-— [NO MIXING; •»•
[NO HETEROGENEOUS EMPIRICAL
COMBUSTION OR PREPARE-TO-
DIFFUSION FLAMES] BURN SCHEDULE]
[HEAT TRANSFER PRESCRIBED]
[DILUTION PRESCRIBED"]
PREMLXED
REACTANTS
[ = 1 ARBITRARY]
[SPECIFIED
.IGNITION DELAY]
MIXTURES SUBJECT TO
COMBUSTION WAVE
[INSTANTANEOUS
COMBUSTION]
HOT PRODUCTS OF COMBUSTION
DILUTED AND COOLED
• PRODUCTS
[UNSPECIFIED NOX '
KINETICS CONSTANTS!,
:MISSION-LADEN
EXHAUST
-------
VI. DIESEL FLAME STUDIES
CHAPTER SUMMARY
Ultraviolet emission and absorption measurements were conducted
on the prechamber engine in an attempt to detect NO. Although the technique
has not proven fruitful for NO, it has yielded some interesting results includ-
ing an unidentified emission band. In addition, qualitative information about
diesel mixing, ignition, and combustion has been obtained by high speed
photography through large uncooled windows.
-------
VI. DIESEL FLAME STUDIES
A. SPECTROSCOPIC OBSERVATIONS
1. Past Studies and Description of Method
The light radiated from air-fed flames creates three distinct signa-
tures of nitric oxide:
Spectral
, Region Wavelength Type of Transition Structure _ Source
UV ,2260/u Electronic Line y (0,0) transition to
lowest vibrational
level of upper state
VIS .5- .?// Chemilumine scent Continuum NO + O reaction
IR 5 . 29// Vibrational Band Fundamental
Spectroscopic species determinations are notably rare in diesel flame
studies. In the IR region there appears to be good reason: Lyn (1963) and
others have observed large quantities of smoke in the flame with correspond-
ing continuum emission overshadowing even the strong CO_ and H~O bands
L* L*
in the IR. This behavior is illustrated below in an IR scan taken by Lyn (1957) .
The IR emission from a diesel flame resembles a grey body with very little
structure to analyze for species contribution (Figure 37) . This strong contin-
uum peaks in the near IR (for a black body, X T = 2884 K) and presumably
max
gives rise to the high peak levels of radiant heat transfer (up to 500 watts/
2
cm ) measured by Henein (1971). The 30% uncertainty encountered by
Newhall (1967) for emission determination of NO in spark-ignited engines
would definitely be a lower limit for diesel flames with these lower S/N ratios.
A further difficulty with emission spectroscopy is the dependence of emission
strength on temperature, which in turn depends on crank angle. Thus a decrease
in signal could be attributed to either lower temperature or NO decomposition at
constant temperature .
93
-------
] /-Grey Gas Emission Curve
123456
Spark-Ignition
(no smoke)
12345
X,
Diesel Acetylene/Air
(heavy smoke) (heavy smoke)
Figure 37
IR SPECTRA SHOWING BLACK BODY EMISSION
FROM DIESEL ENGINE [Lyn (1957)]
The near UV was selected for the NO diagnostic technique primarily
because the carbon-smoke continuum at 2500°K was expected to fall off
sharply below .4^, according to blackbody curve shapes. The work of
Shahed and Newhall (1971) had shown that the y(0,0) NO band at 2260 $
could be seen above the shoulder of the O_ Schumann-Runge continuum even
at 350 psi (Figure 38), although this result was for smokeless fuel (propane).
70
t- 60
z
111
o
5 50
o.
I
2 40
tc.
O
S30
I 20
O
UJ
S> 10
(0,0) Gamma Band
Nitric Oxide
Absorption^
2IOO 2200 2300
WAVELENGTH-ANGSTROMS
Figure 38
(0,0) gamma band nitric oxide absorption
superimposed on Schumann-Runge background
[Shahed and Newhall (1971)]
94
-------
Quader (1973) at General Motors Research Laboratories has also attempted
to utilize the NO gamma bands to measure the nitric oxide histories in a
single cylinder spark ignition engine. The 2358 A (0,1) band of NO has
also proved useful, according to Seery and Bowman (1970) and McGregor (1973)
The /? band at 2200 A apparently was obscured by the O absorption. Further-
£*
more we attempted to measure the onset of combustion by monitoring the
growth of the key OH radical through the 3064 A absorption band.
It was hoped that the time-resolved traces of the 2260 A band of
NO and the nearly O -smoke continuum (say 2240 A) would appear as in
w
Figure 39.
Unattenuated Source Intensity (Io)
w
Q)
4->
C
2240 A trace (IBackground)
2260 A trace (I)
'• . Crank angle 3
Figure 39
EXPECTED ABSORPTION TRACES
The expression for transmitted intensity in the absence of self absorption
will obey BeeVs Law:
Back. .
Back.
> WhereI'IBack;'andIo
are defined in Figure 39
95
-------
Combining these, we define a
total'
*Back. l aTotal
1 - a
Total , ,v
- = exp (-kX..T
1 - «„ . * v N
Back.
One only need calibrate or predict k (cm ) to solve for
2. Experimental Technique
Large uncooled windows were developed and used with both open and
divided chambers (the experience with windows is documented in Appendix G).
After establishing optical access to the flame of an operating prechamber engine,
the modulated UV spectroscopic system shown in Figure 40 was set up.
Fugl JnjegtarJfeedleJ.ift_ Signal
_Engine_Crankshaft_tagle Signa^
1 r ~~
L ~™l
prechamber ~l I
Chopper
Engine Pressure
Signal I
Ploto multipliers
Mctnochrcmator
' 1000 W Xe Arc
Cassegranian Optical
Transfer System
Figure 40.
DIESEL SPECTROSCOPIC SYSTEM
96
-------
3.
UV-Absorption Results
Twenty-second intervals of fuel injection were run with motoring in
between these intervals to clean up the windows. Signal strength at the
beginning of each interval was 5 volts over a 5 A spectral window. Measure-
ments taken at 2220 A (background) and 2265 A (NO band) are reproduced in
Figure 41.
«TpC
Case (a) X = 226S8. (NO band max.p
32/1 A/F
10°CA/cm
1
\
Max. Absorption I
i
•
!• ^
/
\
r
l
V
s/
1500 RPM
P.C. 13/16" plug
Max. Absorption
t
Absorption
Needle Lift
Case (b) A. = 2220$ (Background)
Figure 41.
Salient features of the traces are as follows:
(i) Absorption begins upon needle lift (+ 1 CA) and
diminishes about 35°ATDC. Absorption signal
width is only 30 CA (indicating absorption is
combustion related).
97
-------
(ii) Absorption is optically thick, giving only about 5%
transmissivity over the 10 to 20 CA wide peak. This
prohibits detecting the NO bands by the method .shown
in Figure 39.
(iii) Shape of absorption trace includes a "notch". This
"notch" is due to emission, as discussed below. The
Xenon arc lamp was turned off and emission observed.
(iv) The absorption is diffuse, i.e. has no band structure
over the 2200 - 2400 A region. This was demonstrated
by spectroscopic plates taken through a chopper phased
to allow only the TDC-30°CA ATDC interval to reach the
spectrometer.
(v) When the intake air was doctored with pure NO so
that the mixture contained 3000 ppm NO, the peak
transmissivity remained at about 5%; however, the
absorption began earlier (rise time 5 CA instead of
2 CA to peak value).
(vi) Absorption histories differ for different alignments
of the Xenon arc beam through the prechamber. For
certain alignments, a secondary absorption 180 ATDC
can be observed.
Based on the 95% absorption described in items (ii), (iv), and (v),
it can be inferred that nitric oxide cannot be observed by UV absorption in
the present prechamber engine. Because the optical depth is so great (item
ii), the rise and fall of species concentration X cannot be detected. Since
absorption goes as A ~ 1 - exp (-kXi), when the product kJ2 is sufficiently
large, A « 1 practically independent of X . There are no changes in A from
which to deduce changes inX . There are at least three possible absorption
sources which could be responsible for the interference:
(a) Photo dissociation of CO , as suggested by McAlevy
and Cole (1973).
(b) Interference from Schumann-Runge transitions in O?
molecules.
(c) Absorption or scattering interference by soot.
Further experimentation is necessary.
98
-------
NO itself has a large optical depth kj2 at these pressures. Table 17
gives the absorption coefficient k (cm ) at 1 atm for five y bands of NO, along
with the ratio of NO absorption to Schumann-Runge O absorption.
Table 17
ABSORPTION COEFFICIENTS FOR THE r-BANDS OF NO*
Fractional
Absorption for k_ [a /a ]
j2=4cm,P=5atm y Bands of NO cm JNIU U2
97% (2,0) 2047 £ 17 2
98% (1,0) 2153 20 4
86% (0,0) 2262 12 3
50% (0,1) 2365 3.5 3
15% (0,2) 2465 0.6 2
*Absorption strengths k taken from Golden (1967).
Perhaps more serious is the lack of band structure (item iv), without
which it will be impossible to subtract the banded NO absorption from the
background absorption. Diedrichsen and Wolfhard (1956) observed the 7 bands
of NO at high pressure to become quite diffuse even at 20 atm; the pressures
encountered here are two or three times higher. These two difficulties would
plague an investigator who had comparable absorption from NO and other
species, which we call the "background". However, from item (v) it would
appear that NO absorption is quite insignificant compared to the background.
4. UV Emission Results
The remaining items (i), (iii), and (vi) led us to use emission spec-
troscopy. We observed a banded emission, with the wavelength structure
recorded in Figure 42, which begins about 2 to 3 CA prior to the "spike" seen
in the pressure trace, and falls off with pressure (see Figure 43). Some of
the possible emitters are as follows:
99
-------
Figure 42
OBSERVED EMISSION FROM PC-DIESEL ENGINE
ol
2200
2300
2400
x.X
2500
2600
Figure 43
CORRELATION OF DIESEL EMISSION
WITH PRESSURE
Emission
(2420 S)
Conditions:
Engine PC - 17:1
1500 RPM
A/T = 32
Volume Ratio 25%
100
-------
NO itself has a large optical depth kj2 at these pressures. Table 17
gives the absorption coefficient k (cm ) at 1 atm for five y bands of NO, along
with the ratio of NO absorption to Schumann-Runge O absorption.
Table 17
ABSORPTION COEFFICIENTS FOR THE 7-RANDS OF NO*
Fractional
Absorption for \ [a /a ]
j2=4cm,P=5atm y Bands of NO cm ™u "2
97% (2,0) 2047 £ 17 2
98% (1,0) 2153 20 4
86% (0,0) 2262 12 3
50% (0,1) 2365 3.5 3
15% (0,2) 2465 0.6 2
*Absorption strengths k taken from Golden (1967).
Perhaps more serious is the lack of band structure (item iv), without
which it will be impossible to subtract the banded NO absorption from the
background absorption. Diedrichsen and Wolfhard (1956) observed the y bands
t
of NO at high pressure to become quite diffuse even at 20 atm; the pressures
encountered here are two or three times higher. These two difficulties would
plague an investigator who had comparable absorption from NO and other
species, which we call the "background". However, from item (v) it would
appear that NO absorption is quite insignificant compared to the background.
4. UV Emission Results
The remaining items (i), (iii), and (vi) led us to use emission spec-
troscopy. We observed a banded emission, with the wavelength structure
recorded in Figure 42, which begins about 2 to 3 CA prior to the "spike" seen
in the pressure trace, and falls off with pressure (see Figure 43). Some of
the possible emitters are as follows:
99
-------
Figure 42
OBSERVED- EMISSION FROM PC-DIESEL ENGINE
si
2
a
eg
II
0
2200
2300
2400
A.*'
2500
2600
Figure 43
CORRELATION OF DIESEL EMISSION
WITH PRESSURE
Pressure
Emission
(2420 S)
Conditions:
Engine PC - 17:1
1500 RPM
A/F = 32
Volume Ratio 25%
100
-------
C HZ A - X band system (2400-2500 $ at high temperature)
OH (3,0) transition of the A2£+- X2/7band
NO (3,2) transition of the y band (see Figure 43)
3
CO Cameron bands (a 77 - x'X)
h +H 0-H(2S) + OH (X2/7)
£t
h + C.CL - CO + O
chemiluminescence, as reported
by McNesby and Okabe (1964)
Again, NO may be questioned because there seemed to be no effect of adding
NO to the intake air; however, the emission spectrum of NO is strongest at
2480 8 in rough agreement with Figure 44. The difficulty in using emission
is that the observations are heavily weighted by what happens in the path
segment nearest the observer.
NO 7 -Bands
U J,L .jit
I
<- *P
"^Z o
r^ fTJ
1
in
in
00
CN
1
O
00
N
z
1
CM
CM
«
in
u>
(M
Figure 44
BAND SPECTRUM OF. AN AIR-FILLED GEISSLER TUBE
Further spectroscopic tests were run on the precup engine including
(1) runs at a low-NO engine setting to see if the emission or absorption traces
are altered (no effect), and (2) runs with H0O injection and observation of
£t
spectroscopic changes (no effect in UV).
Lyn (1957) observed similarity of diesel and acetylene flame spectra, and
Quader (1969) observed a similar phenomenon in the 2500-2600 X region
for preflame reactions. The Q^i bands have been reported by Kistiakowsky
(1931) and Woo etal.(1938).
101
-------
B. DIESEL FLAME PHOTOGRAPHY
A rotating prism camera has been used to observe fuel spray,
ignition, and combustion in the prechamber. Both panchromatic and color
movies were taken at 5000 fps (approximately 2 CA between frames at
1500 RPM engine speed), under both Xenon backlighting and self-illumination.
Magnification was approximately 1/10; an f/11 setting was adequate for
ASA 800 processing of Kodak EAR 2479 film.
The qualitative information about diesel spray combustion has been
correlated with measurements of chamber pressure, fuel injection, and
ultraviolet emission/absorption as shown schematically in Figure 45.
The following observations may be worth noting:
Injection: The first evidence of fuel injection is a small
cloud of vapor in the vicinity of the orifice. Three
frames (6 CA) are required for the jet to reach the
opposite wall. Absorption at 2264 A begins during
this fuel penetration stage and may be simply physical
obscuration. Needle lift synchronizes well with the .
first appearance of fuel. As the spray crosses the
chamber, the front is blunted and flattened with waves
or ripples on the edges. Conceivably, fuel vapor and
small droplets are shredded from these non-uniformities
along the edge of the spray. As the spray is estab-
lished and fuel line pressure builds up, the jet thick-
ness decreases.
Ignition: Ignition begins immediately, but is quite localized
(so that the net heat release is negative due to fuel
evaporation). Visible emission is observed to emanate
from the edges of the spray at the base. This implies
that the chemical delay time is negligible and the time
required to obtain burning is simply the mixing time for
molecular contact of fuel and air. The flame propagates
from the base of the spray along the edges parallel to
the jet axis. Just as the spray is completely enveloped,
a separate ignition zone is often noticed at the impact
region where the jet has struck the opposing wall.
Emission (2300 - 2500 $) is first noticed at this time.
102
-------
Combustion: The entire chamber is engulfed with a bright
orange flame one frame (2°CA) after the jet is completely
enveloped. This point corresponds precisely to the
sudden rise in chamber pressure, often referred to as the
ignition "spike". (Perhaps hot main-chamber gases are
propelled into the prechamber.) Combustion is optically
opaque; even the jet spray is a vague shadow on those
runs where it can be seen at all. Turbulent structure is
clear with eddy sizes as small as 500// seen (1/50
chamber diameter) . Wall regions are darker, possibly
indicating smoke formation. A cone of ordered-motion
forms about the throat entry to the main chamber; this
cone has an included angle of about 120 deg. As the
gases empty out and orange radiation decreases, the
UV emission decreases and the pressure trace falls off.
Prechamber and direct injected movie strips are available upon
request to the authors.
Figure 45
REPRESENTATION OF HIGH-SPEED PHOTOGRAPHS
OF PRECHAMBER COMBUSTION
-6
TDC
+12
+20
Fuel -
Combustion
rrrr
Emission
•Absorption
Chamber Pressure
Needle Lift
-------
NOMENCLATURE
a ,b Coefficients
A Pre-exponential factor in Arrhenius rate expression
A/F Air/fuel ratio (dimensionless) by mass
C Constant
C Specific heat at constant pressure
P
d Droplet size or piston bowl diameter
d Pxiel orifice diameter
2
D Diffusion coefficient (cm /sec) or cylinder bore (cm)
D1 Diffusion coefficient used by CAV (cm )
DI Direct-injected combustion chamber
E Activation energy in Arrhenius rate expression
E Entrainment coefficient of CAV model
r
f Function, as for size distribution
F Mixture ratio of NREC model
h Specific enthalpy or piston bowl height
i Stoichiornetric oxygen to fuel ratio (by mass)
k Rate coefficient or absorption coefficient (cm )
K Equilibrium constant
L Latent heat of vaporization
L/D Length to diameter ratio
m Mass
m Mass flow (g/sec)
M Molecular mass
n Coefficient
N Damkohler number (see page 50)
N Plclet number
104
-------
<3 Rate of heat release
q. Rate of heat loss to the walls
loss
P Pressure
PC Prechamber combustion geometry
Q Heat of combustion per unit mass of fuel
r Radius
R Gas constant, 1.98 cal/^K mole, or piston bown radius of curvature
S Schvab-Zeldovich variable T~ + Y or T + Y,
ox f
t Time
T Temperature
v Velocity
V(&) Volume of cylinder
X Mole fraction
Y Mass fraction A11 Reactions,
~ ~ v» Mj
Y, Element mass fraction (Y, = /_,u.. 77" Y.) for element j
J j Y ljMi x
ct Absorptivity or piston bowl angle
2
(3 Burning rate coefficient (cm /sec) or piston bowl angle
y Gas property, C /C
P ^
5 Expansion parameter
6 Emissivity
f Adjustable angle of mask position
n r/rf non-dimensional radius
Q Coordinate in (r, 0, z) system
K Thermal conductivity
(£> Crank angle
/i.. Number of atoms of element j in molecule of specie i
y Stoichiometric coefficients
p Density
105
-------
r Critical time (e.g. for ignition)
r Chemical reaction time (sec)
c
r Physical transport time (sec)
d> Equivalence ratio (fuel/air)/(fuel/air) ^ . .
• •- ~ >-; • " stoich.
u) Reaction rate (g/sec-cc)
Subscripts and Superscripts
a Air
co Coolant
e Exhaust
egr Exhaust gas recycled
f Fuel
fl Flame
HO Water
Lt
i Species index
inj Injection
1 Liquid
ox Oxygen
P Products
PC Prechamber
r Reference
s Soot
tot Total intake flow of air plus EGR, or total chamber volume
1 start
2 end
( ) Non-dimensional, transformed/-_or.averaged variable
oo Far from the droplet surface/- surrounding conditions
{ > Averaged quantity
10$:..'
-------
REFERENCES
Abthoff, J. and Luther, H., "The Measurement of Oxide of Nitrogen Emission
from Diesel Engines and its Control by Modes of Engine Operation,"
Auto. Zeits. 71, (4), 1969.
Bascom, R. C., Broering, L. C. and Wulfhorst, D. E., "Design Factors That
Affect Diesel Emissions," SAE Paper 710484, 1971.
Bastress, E. K., Chng, K. M., and Dix, D. M., "Models of Combustion and
Nitrogen Oxide Formation in Direct and Indirect Injection Compression-
Ignition Engines," SAE Paper 719053, 1971.
Bosecker, R. E. and Webster, D. F., "Precombustion Chamber Diesel Engine
Emissions—A Progress Report," SAE Paper 710672, 1971.
Bowman, C. T. and Kesten, A. S., "Kinetic Modeling of Nitric Oxide Formation
in Combustion Processes," presented at the Western States Section/
Combustion Institute, Fall Meeting, 1971.
Bracco, F. V., "Nitric Oxide Formation in Droplet Diffusion Flames," 14th
Symposium (International) on Combustion, p. 831, The Combustion
Institute, Pittsburgh, 1973.
Burke, S. P. and Schumann, T. E. W., "Diffusion Flames," Industrial Eng.
Chem. 20.:998, 1928.
Coffin, K. P. and Brokaw, R. S., "A General System for Calculating Burning
Rates of Particles and Drops and Comparison of Calculated Rates for
Carbon, Boron, Magnesium, and Isx>octane," NACA TN 3929, 1957.
Diederichsen, J. and Wolfhard, H. G., "Spectrographic Examination of
Gaseous Flames at High Pressure," Proc. Royal Soc. London, 1956.
Eyzat, P., "Etude Experimental et Theorique de la Formation des oxydes
d'Azote Dans les motuers," Rev. de L'Inst. Franc de Petrol 22, 1967.
Fendell, F. E., "Combustion in Initially Unmixed Reactants for One-Step
Reversible Chemical Kinetics," Astronautica Acta 13:183, 1967.
Godsave, G.A.E., "The Burning of Single Drops of Fuel. Part I: The Tempera-
ture Distribution and Heat Transfer in the Preflame Region," Report
R-66, Natl. Gas Turbine Estab. of the Minister of Supply, London, 1950.
Golden, S. A., "Approximate Spectral Absorption Coefficients of Electronic
Transitions in Diatomic Molecules," J.Q.S.R.T. 7.:225, 1967.
Grigg, H. C. and Syed, M. H., "The Problem of Predicting Rate of Heat
Release in Diesel Engines," SAE Paper 700503, 1970.
107
-------
Hames, R. J., Merrion, D. F., and Ford, H. S., "Some Effects of Fuel
Injection System Parameters on Diesel Exhaust Emissions," SAE
Paper 710671, 1971.
Henein, N. A., "Combustion and Emission Formation in Fuel Sprays Injected
in Swirling Air," SAE Paper 710220, 1971.
Kassoy, D. R. and Williams, F. A., "Effects of Chemical Kinetics on Near
Equilibrium Combustion in Nonpremixed Systems," Physics of Fluids
jj.:1343, 1968.
Khan, I. M., Greeves, G. and Probert, D. M., "Prediction of Soot and
Nitric Oxide Concentrations in Diesel Engine Exhaust," Inst. Mech.
Engr. Paper C142/71, 1971.
Khan, I. M. and Greeves, G., "A Method of Calculating Emissions of Soot
and Nitric Oxide from Diesel Engines," SAE Paper 730169, 1973.
Khan, I. M. and Wang, C.H.T., "Factors Affecting Emissions of Smoke and
Gaseous Pollutants from Direct Injection Diesel Engines," Inst. Mech.
Engr. Paper C151/71, 1971.
Kistiakowsky, G. B., Phys. Rev. 37, 1931.
Kotake, S. and Okazaki, T., "Evaporation and Combustion of a Fuel Droplet,"
Int. J. Heat Mass Transfer ^2.:595, 1969.
Krause, S. R., Merrion, D. F., and Green, G. L., " Effect of Inlet Air
Humidity and Temperature on Diesel Exhaust Emissions," SAE Paper
730213, 1973.
Landen, E. W., "Nitrogen Oxides and Variables in Precombustion Chamber
Type Diesel Engines," SAE Paper 714B, 1963.
Lyn, W. T., J. Inst. Petr. 43_, 1957.
Lyn, W. T., "Calculations of the Effect of Rate of Heat Release on the Shape
of Cylinder-Pressure Diagrams and Cycle Efficiency," Proc. Auto. Div.
Inst. Mech. Engr. No. 1, p. 34, 1960.
Lyn, W. T., "Study of Burning Rate and Nature of Combustion in Diesel
Engines," 9th SIOC, 1963.
Marshall, W. F. and Fleming, R. D., "Diesel Emissions as Related to Engine
Variables and Fuel Characteristics," SAE Paper 710836, 1971.
McAlevy, R. F. Ill, and Cole, R. B., "Nitric Oxide Measurement in a Spark-
Ignition Engine," Stevens Inst. Tech. TR #ME-RT 73001, 1973.
108
-------
McConnell, G., "Oxides of Nitrogen in Diesel Engine Exhaust Gas: Their
Formation and Control," Proc. Inst. Mech. Engrs. 178, 1963.
McCreath, C. G. and Chigier, N. A. / "Liquid Spray Burning in the Wake
of a Stabilizer Disc/1 14th SIOC, 1972.
McGregor, W. K., Sieber, G. L. and Few, J. D.., "Concentration of OH
and NO in YJ93-GE-3 Engine Exhausts Measured in situ by Narrow-
Line UV Absorption," AIAA Paper 73-506, 1973.
McNesby, J. R. and Okabe, H., "Vacuum Ultraviolet Photochemistry,"
Adv. in Photochem. 3., 1964.
Natarajan, R. and Brzustowski, T. A., "Some New Observations on the
Combustion of Hydrocarbon Droplets at Elevated Pressures," Comb.
Sci. & Tech. 2, 1970.
Newhall, H. K. and Starkman, E. S., Direct Spectroscopic Determination
of Nitric Oxide in Reciprocating Engine Cylinders," SAE Paper
670122, 1967.
Parker, R. F. and Walker, J. W., "Exhaust Emission Control in Medium
Swirl Rate Direct Injection Diesel Engines," Combined FCIM and
Powerplant Meeting, SAE Paper 720755, 1972.
Pischinger, R. and Cartellieri, W., "Combustion System Parameters and
Their Effect Upon Diesel Engine Exhaust Emissions," SAE Paper
720756, 1972.
Quader, A. A., Myers, P. S., and Uyehara, O. A., "UVAbsorbance Histor-
ies and Knock in a Spark Ignited Engine," SAE Paper 690519, 1969 .
Quader, A. A., Studies of Diesel-NO by UV Absorption, private communica-
tion, 1973.
Rosner, D. E., "On Liquid Droplet Combustion at High Pressures," AIAA J.
5.:163, 1967. .
Schmidt, R. C., Carey, A. W., and Kamo, R., "Exhaust Characteristics of
the Automotive Diesel," SAE Paper 660550, 1966.
Scott, W. M., "Looking in on Diesel Combustion," SAE Paper 690002, 1969.
Seery, D. J. and Bowman, C. T., "An Experimental and Analytical Study of
Methane Oxidation Behind Shock Waves," Comb. & Flame 14, 1970.
Shahed, S. M. and Newhall, H. K., "Kinetics of Nitric Oxide Formation in
Propane-Air and Hydrogen-Air Diluent Flames," Comb. & Flame 17, 1971.
109
-------
Shahed, S. M., Chiu, W. S. and Yumlu, V. S., "A Preliminary Model for
the Formation of Nitric Oxide in Direct Injection Diesel Engines and
its Application in Parametric Studies/1 SAE Paper 730083, 1973.
Shipinski, J., Uyehara, O. A., and Myers, P. S., "Experimental Correla-
tion Between Rate-of-Injection and Rate-of-Heat-Release in a Diesel
Engine/1 ASME Paper 68-DGP-ll, 1968. .
Spalding, D. B., ARS Journal 29.:828, 1958.
Tanasawa, Y., "On the Combustion Rate of a Group of Fuel Particles/1
Tech. Rep. Tohoku Univ. .18:61, 1953.
Tarifa, C. S., Crespo, A., Fraga, E., and Munoz, A., "Supercritical Com-
bustion of Fuels and Propellants in Droplets," AD 725 749, 1971.
Valdmanis, E. and Wulfhorst, D. E., "The Effects of Emulsified Fuels and
Water Induction on Diesel Combustion," SAE Paper 700736, 1970.
Walder, C. J., "Reduction of Emissions from Diesel Engines," SAE Paper
730214, 1973.
Williams, F. A., Combustion Theory, Addison-Wesley, Reading, 1965.
Williams, G. C. and Sarofim, A. F., "Models for NO Formation in Combus-
tion Processes," Task Order #3 under HEW Contract CPA 22-69-44,
Final Report, 1970.
Wilson, R. P. Jr., "Combustion of Aluminum Particles in OVAir," Ph.D.
Thesis, U.C. San Diego, 1970.
Wise, H., Lorell, J., and Wood, B. J., "The Effects of Chemical and Physical
Parameters on the Burning Rate of a Liquid Droplet," 5th SIOC, 1955.
Woo, S. C., Badger, R. J., Chu, T. C., and Chih, W., J.C.P. 6., 1938.
Zeldovich, Ya. B. and Raizer, Yu. P., Physics of Shock Waves and High-
Temperature Hydrodynamic Phenomena, Volume 1, Academic Press,
1966.
110
-------
APPENDIX A
ESSENTIALS OF NO AND SMOKE FORMATION IN FLAMES
.X.
The processes of heat release and fuel spray/air mixing are not
altered by slight traces of NO and smoke, therefore these major flame pro-
cesses can be described as uncoupled to NO and smoke mechanisms*. The
reverse is definitely not true; nitric oxide and smoke formation are quite
sensitive to how mixing and heat release occur in certain parts of the flame.
In mathematical terms, smoke and nitric oxide concentration are dependent
variables.
In the following section we wish to review which flame conditions
dominate the manufacture of NO and smoke. The essential mechanisms of
NO and smoke formation are discussed insofar as current state-of-the-art
permits, so that pollutant-active flame conditions can be recognized.
A. NITROGEN OXIDES
Nitric oxide is composed of two atoms, each of which is normally
tightly bound in molecules such as N« and O or in organic-nitrogen compounds
in the fuel. Unless the oxygen and nitrogen bonds are broken, NO cannot form.
Once free nitrogen or oxygen are formed, reactions of the following type occur:
O + N2 —*• NO + fragment (1)
N-radical + O2 —* NO + fragment (2)
N + O2 —» NO + fragment (3)
N + OH —* NO + fragment (4)
Pyrolysis products bearing nitrogen such as CHN and ON are critical since these
fuel fragments are readily broken up to yield "loose" N atoms. However, diesel
fuels contain only enough chemically-bound nitrogen to produce 30 to 50 ppm at
100% conversion.
*An exception is heavy smoke formation which establishes a coupling with the
flame processes through increased emissivity and radiative heat loss.
Ill
-------
The O2 bond rather suddenly becomes breakable as the temperature
rises above about 2000°K0 Oxygen atoms are generated and reaction (1)
occurs, followed immediately by reaction (3) or (4). This is known as the
extended Zeldovich (1946) mechanism and is widely assumed to be the only
NO mechanism of importance in the diesel flame. Once formed, nitric oxide
does not readily decompose; the NO molecule is surprisingly stable as the
flame gases cool. Thus we need only be concerned with NO formation.mecha-
nism.
The conditions for generating oxygen radicals are high temperature
and high oxygen concentration. At low NO concentrations, the rate of NO
formation can be expressed in terms of T and X-. as follows:
d(NO) 3, ,-:-
\. = l.lx 10 (O_) 2 exp (-135,000/RT) ,
QI £
where (O_) is molecular density in cm and is related to mole fraction X
by the expression (O9) = p Xn Na M , where N is Avogadro's number. ^
i (.)„ A A
Hot oxygen rich regions can be found in several distinct subzones in a diesel
engine:
1. Laminar flame front around burning drop
2. Turbulent diffusion flame front
3. Hot burnt gases
These are discussed briefly below.
1. Droplet Diffusion Flame as an NO Source
The extent to which significant NO is generated in diffusion flames
X
relative to the other candidate zones is an extremely crucial point. The flame
envelope which surrounds a burning liquid fuel droplet moving with low velocity
relative to the surrounding air has both the high temperature and high oxygen
content needed for generating nitric oxide, as shown in Figure A-l.
112
-------
Threshold
NO-formation
Temperature
(~2000°K)
NO formation would
occur in this region
where high temperature
and O-content overlap.
0
droplet
radius
flame
radius
Figure A-l
Temperature and Mass Fraction Profiles
for a Burning Fuel Droplet in the
Flame Surface Approximation [Williams (1965)]
There is some question whether the oxygen content is high where the
temperature is large. The peak flame temperature is near adiabatic with
considerable dissociation; on the other hand, the oxygen concentration tapers
off to extremely low values at the flame, so that .dissociation of the product
species might have to serve as the principal source of O-atoms. Further
discussion of the laminar diffusion flame appears in Appendix E.
2.
Turbulent Diffusion Flame Front as an NO -Source
x
In a turbulent unmixed flame, fuel vapor and air mix with hot, pre-
viously burnt gases and bum in near-stoichiometric zones. At the flame
front itself the temperature approaches the adiabatic maximum; a non-
equilibrium distribution of species with a temporary "oversupply" of O-atoms
may occur. Thus, conditions may be quite conducive to NO formation.
113
-------
Here in the flamelets we would require a local fine-scale description of the
turbulent diffusion flame. At present this understanding is only available
for laminar diffusion flames such as the fuel spray flames with low relative
velocity (no "wake burning"). There are a number of speculative theories
[Spalding (1971), Libby (1972) , Bush and Fendell (1973)] and clever measurements
[Batt et al.(l970)] of reaction in a turbulent eddy, but the state of knowledge
does not permit a detailed model. Perhaps the most perplexing aspect is
the interaction of turbulent fluctuations and chemical conversion rates.
The description of NO is tied closely with the description of turbu-
j£.
lent mixing with simultaneous kinetics. This will undoubtedly receive
intensive long-term research but has yet to be modeled in any satisfactory
way. The prospects for including a rigorous treatment of NO for turbulent
X
diffusion flames in the proposed program are admittedly quite dim.
3. Hot Combustion Products as an NO -Source
'""' ~" X "
The zones of intense diffusion-limited burning produce a steady
source of hot gases (mainly CO , HO, and N ). This gas may be as hot
as 2300 to 2600 K initially (depending on the fuel characteristics, air/fuel
ratio, preheat, chamber geometry and emissivity), and this is well above
the 2000°K threshold for generating O-radicals which lead to nitric oxide.
However, the hot flame products are quenched by mixing with cooler gases,
by radiative heat losses to the wall, and by piston expansion, so that gases
leave the cylinder at about 850 to 950°K.
The NO-prediction technique in this situation consists of trying to
describe the quenching process, i.e., the mean thermal histories of discrete
volumes of hot gas originating from given locations in the diffusion flame
zone. To this end, it is useful to know mixing patterns and heat transfer
coefficients. Heat transfer coefficients can be estimated, but for confined
turbulent flames, the product/air mixing should be measured, even if only
in cold flow.
114
-------
B.
SOOT FORMATION AND OXIDATION
The presence of carbon particulates in the exhaust gas can be attributed
to either excessive formation or inadequate oxidation, or usually both of these.
1. Soot Formation
Soot formation is believed to occur in fuel rich zones fed by massive
evaporation such as the apex of the fuel spray, where burning is temporarily
incomplete because of (1) lack of air, or (2) quenching by wall contact or cold
gases. When partially decomposed fuel vapor mixes with hot combustion pro-
ducts, carbon-forming reactions (e.g., C_H- + O, HO + CJ are stimulated.
& £* Z» £
This process may also occur on a smaller scale on the fuel side of a diffusion
flame mantle. The mechanism of soot formation is illustrated in Figure A-2
[from Meier zu Kocker (1972)]:
Condensed
Phase
Soot
Agglomeration
/
Polyacetylene j
Nucleation
Gaseous
Phase
Figure A-2
Mechanism of Soot Formation
Soot formation is a compound process involving both decomposition to
acetylene (believed to occur by gas-phase cracking) and subsequent formation
of carbon agglomerates. The cracking step is probably rate controlling. As
evidence of this, the tendency for soot formation is sensitive to the C/H ratio
and boiling temperature of the fuel, as shown by Meier zu Kocker (1968) in
Figure A-3.
115
-------
Relative Soot 3
Residue
2
1
0
5 7 9 11 13 15 17 19
C/H Ratio by Weight
Figure A-3
Soot Formation of Some Hydrocarbons
in One-Dimensional High Pressure Combustion
The soot formation rate is often expressed in the form
drn n m
m d t s -
s
where rn is the mass of soot. Typical values quoted for propane and methane
s
are n = 1 and E = 32 to 58 kcal/mole. For diesel engines, Khan and Greeves
S S
(1973) found best agreement with emission data by using n = 3, m =1,
S S
E = 40.0 kcal/mole , although the and T profiles in their calculations
S
were necessarily hypothesized.
The prevention of carbon formation seems to rest on mixing considera-
tions—on air penetration into the fuel spray pattern. In gas turbines and burners
this may be accomplished by direct air jets [Bahr et al. (1969), and Faitani (1968)]
or by increasing the atomizing air [Durrant (1969)] (if the burner is so equipped.
In both diesels and gas turbines, better mixing can be achieved by increasing the
spray cone angle [Faitani (1968), Toone (1968), Durrant (1969), and Lefebvre and
116
-------
Durrant (I960)] or by increasing swirl [Bahr et al. (1969)]. It goes without
saying that many of these techniques contradict the principles of delayed
mixing to reduce NO .
.X.
2. Soot Oxidation
Some carbon soot presumably forms in the spray cone no matter what
measures are taken. This is evidence by the yellow/white black-body emissions
from an ordinary candle. Thus soot oxidation is also a primary goal of combus-
tion system design. Methods to promote oxidation include increasing the tem-
perature [Faitani (1968), Toone (1968), and Gross -Gronowski (1967)], and main-
taining uniform air/fuel ratio [0e (.8, 1.2)] in the chamber.
The rate of soot combustion can be taken proportional to the exposed
area of the particles, the oxygen concentration, and an Arrhenius rate term.
Lee, Thring and Beer (1962) found for a laminar diffusion flame an activation
energy A =39.3 kcal/mole in an expression of the form:
c
dm m n q
= AP ' * C -p (-ERT) T C
dA* C
Lee also found n = -1, m = 1, q = 1/2. Tesner and Tsibulevsky (1967) also
c c c
found n = -1, m = 1, q = 1/2, and E = 40 kcal/mole.
c c c c
The surface regression rate (micron/sec) can be calculated from
dm /dAdt by using dr/dt = -104(dm /dAdt)/pa. Radcliffe and Appleton (1971)
s s s
have performed such calculations for soot placed in equilibrium products of
combustion (C H9 /air at 15 atm, preheat 700 K) and obtain the result shown
f\
in Figure A-4. Park and Appleton (1973) determined dm/dAdt at 2500 K as
S
shown in Figure A-5 . Measurements of soot particle sizes show the mass mean
size in the .05 to .5# range [DeCorso (1967) and Faitani (1968)], and the
residence time for diesel engines is about 3 msec. For 0w 0.7 a .1/u particle
will require a burnup time of about 6 msec according to Figure A-4, and about
2 msec according to Figure A-5. These values are marginal since the duration
of diesel heat release is about 4 msec. Fenimore and Jones (1967) indicate
much shorter burning times . However, the conditions giving high soot burnup
rates also result in maximum NO formation.
117
-------
20
15
o
ui
z
o
IT
O
10
O.4 O.6 O.8
EQUIVALENCE RATIO. cf>
I.O
Figure A-4
Estimated Rate of Surface Recession of a Soot Particle
[Radcliffe and Appleton (1971)]
10
rl
3
UJ
CE
10"
O
X
o
V)
o
LL
85 ID'4
NAGLE a STRICKLAND-CONSTABLE
SEMI-EMPIRICAL FORMULA
cjecP02
T=2500±IOOeK
o SOOT RADIUS Rm=
A SOOT RADIUS Rm= I8O A
45 A
0.01
IOO
O.I I IO
OXYGEN PARTIAL PRESSURE, Po2, ATM
Figure A-5
Specific Soot Oxidation Rate at 2500 + 100°K vs. Oxygen Partial Pressure
[Park and Appleton (1973)]
118
-------
REFERENCES FOR APPENDIX A
Bahr, D. W., Smith, J. R. and Kenworthy, M. J., "Development of Low Smoke
Emission Characteristics for Large Aircraft Turbine Engines, " AIAA
Paper 69-493, 1969.
Batt, R. G., Kubota, T. and Laufer, J., "Experimental Investigation of the
Effect of Shear-Flow Turbulence on a Chemical Reaction," AIAA
Paper No. 70-721, 1970.
Bush, W. B. and Fendell, F. E., "On Diffusion Flames in Turbulent Shear
Flows," Project SQUID Technical Report TRW-7-PU, July 1973.
DeCorso, S. M., Hussey, C. E. and Ambrose, M. J., "Smokeless Combustion
in Oil Burning Gas Turbines," ASME Paper 67-PWR-5, 1967.
Durrant, T., "The Reduction of Smoke from Gas Turbine Engines," presented
to 9th International Aeronautical Congress, A.F.I.T.A.E., Paris, 1969.
Faitani, J. J., "Smoke Reduction in Jet Engines through Burner Design," SAE
Paper 680348, 1968.
Fenimore, C. P. and Jones, G. W., "Oxidation of Soot by Hydroxyl Radicals,"
J. Phys. Chem. 71_, 593-597, 1967.
Gross-Gronowski, L., "Smoke in Gas-Turbine Exhaust," ASME Paper 67-WA/
GT-5, 1967.
Khan, I. M. and Greeves, G., "Factors Affecting Diesel Smoke and Emissions
and a Method of Calculation," SAE Paper 730169, 1973.
Lee, K., Thring, M. and Beer, J., "On the Rate of Combustion of Soot in a
Laminar Soot Flame," Combustion and Flame £, 137-145, 1962.
Lefebve, A. H. and Durrant, T., "Design Characteristics Affecting Gas Turbine
Combustion Performance," presented to National Aeronautical Meeting,
SAE, Los Angeles, 1960.
Libby, P. A., "On Turbulent Flows with Fast Chemical Reactions. Part I: The
Closure Problem," Combustion Science and Technology .6, 22-28, 1972.
Meier zu Kocker, H., Brennstoff-Chemie 49, S 193/198, 1968.
Meier zu Kocker, H., "Kinetics of Soot Formation Investigations into the Mecha-
nism of Soot Formation in Hydrocarbon Diffusion Flames," Combustion
Science and Technology 5., 219-224, 1972.
119
-------
Park, C. and Appleton, J. P., "Shock-Tube Measurements of Soot Oxidation
Rates," Combustion and Flame 20., 369-379, 1973.
Radcliffe, S. W. and Appleton, J. P., "Soot Oxidation Rates in Gas Turbine
Engines," Combustion Science and Technology ±, 171-175, 1971.
Spalding, D. B., "Mixing and Chemical Reaction in Steady Confined Turbulent
Flames," Thirteenth Symposium (International) on Combustion, 649-
657, The Combustion Institute, 1971.
Tesner, P. A. and Tsibulevsky, A. M., "Gasification of Dispersed Carbon in
Hydrocarbon Diffusion Flames, III. Flames of Acetylene-Hydrogen
and Acetylene-Water Vapor Mixtures," Combustion, Explosion and
Shock Waves 3., 1963-1967 (Translation from Fizika Gorenya i
Vzryva 3., 261-267), 1967.
Toone, B., "A Review of Aero Engine Smoke Emission," Cranfield International
Symposium Series 10, Combustion in Advanced Gas Turbine Systems
(I. E. Smith ed.), Pergamon Press, 1968.
Zeldovich, Y. B., "The Oxidation of Nitrogen in Combustion and Explosions,"
Acta Physicochim. USSR 2l_, 577, 1946.
120
-------
APPENDIX B ;
SINGLE CYLINDER EXPERIMENTAL TECHNIQUE
A. Design Specifications
A research engine was built specifically to provide the following
features for the experimental program:
(i) Representative of medium output engines (150 to 350 HP)
(ii) Modular interchangeability to permit rapid design
changes (head, prechamber volume, piston, fuel system)
(iii) Flexibility to permit operational changes while engine is
firing (speed, load, timing, EGR, water injection, air
temperature, air density, air swirl)
(iv) Special head design to accommodate flame diagnostics.
Cylinder hold down bolts, intake and exhaust porting,
valve actuating mechanisms, pencil fuel nozzle, and
other hardware necessary to make an engine function
are fitted subordinate to the primary objective of making
combustion measurements through windows and probe
access ports.
Table B-l indicates the diesel engine types currently in use. The
focus of the study is on the open chamber, direct-injection engine because
of its geometric simplicity. Larger marine and locomotive engines rely on
intense fuel dispersion, whereas the smaller generator and automotive
engines operate at high swirl, high speed to promote mixing.
Table B-l
DOMINANT DIESEL CHAMBER TYPES
General
Class
Direct
Injection
Indirect
Injection
Size
(HP)
1000-20000
150-350
5-100
10-400
Main Provision
for Mixing
High fuel
dispersion
Med. swirl;
Med. fuel
dispersion
High swirl;
Wall Impinge-
ment
Combustion-
generated
turbulence
Air
Swirl
Low
Med
High
Operating
•Speed
To 600 RPM
To 2400 RPM
To 4000 RPM
Bowl/Piston
Ratio
0.8
0.6
Fuel
Dispersion
High
Med
Med-Low
Injector
Pressure
20,000
7,000
—
No. of
Holes
8-12
3-8
1-4
Prechamber
< Lanova/energy cell
; •Poker" prechamber
Representative
Manufacturers
GM Electromotive
.Nordberg
Cooper Bessemer
Fairbanks Morse
Cummins
Detroit Diesel
Mack
Allls Chalmers
John Deere
Int. Harvester
Caterpillar
M.A.N.
Deutz
Hercules
Perkins
Caterpillar
Mercedes
European and
Japanese
Obsolete
McCullogh
Application
Marine
Locomotive
Oil drilling
Trucks . .
Buses
Tractors
Construction
Automotive,
Portable
generators.
Mining,
Material
handling
Automotive
Trucks
Buses
Tractors
Construction
121
-------
Single cylinder crankcases are made by Waukesha (largely for fuel test work),
Labeco (for lubricant test work), AVL and BICERA (for combustion development);
but none of these were felt to offer both the size (bore and stroke) and inter-
changeability of parts that was required to accomplish the objectives of this
program.
B. Facility
The selected chamber geometry is a 5-1/2" bore, 6" stroke, with 12"
connecting rod to minimize piston slap at TDC. Two cylinder head configura-
tions were fabricated—a direct injection head and a prechamber version, here-
after referred to as the DI engine and PC engine (see Fig. 1, page 9). Only two
valves are used in order to leave space in the top deck for probe access and
windows. The prechamber design evolved with suggestions by J. Perez and
his colleagues at Caterpillar. Piston and bowl geometry are given in Figure 1
for both types of heads. Changes in piston geometry gave compression ratios
of 20:1, 17:1, and 14:1 for the DI configuration; changes in piston caps and
prechamber gave four combinations of PC ratio and compression ratio, as
listed in Figure 1.
Twenty drums of certified fuel were set aside under a nitrogen blanket;
the laboratory report on the fuel is given in Table B-2. Fuel was transferred to
the nozzle by one cylinder of an American Bosch model APE-6BB pump. Key fuel
system parameters are given in Table B-3.
Table B-2 Table B-3
FUEL PROPERTIES NOMINAL FUEL SYSTEM PARAMETERS
H/C (by mole, approx.) , 1.7 Line pressure 10,000 psi
Gravity, °API 34.5 Valve opening pressure 3,500 psi
Total sulfur, wt % 0.3 Duration at full load 20°
Flash point, °F 170 fiiiH i9=;3/lf
Pour point, °F 20 .
o A/F at full load 22:1 (<£w.7)
Cloud point, °F 24
„ , ... _ Plunger diameter 12mm
Cetane number 46.7
Viscosity,.CS @ 100°F . 2.5 Cam lift 10mm
Aromatics, % 37.6 Cam profile #1 (tangential)
Distillation: Line length 14"
Initial boiling (°F) 386
50% point (°F) 514 Line ID .072"
End point (°F) 658 '
1 Residue 1% ;
Total nitrogen, wt % <0.05
122
-------
Typical histories of fuel pressure, needle lift> and chamber pressure are
given in Figure B-l.Nozzle for the DI head was a Roosa pencil injector,
popularly used in farm tractors, nominally with six .010" orifices at 160
2
cone angle. For the PC head a single pintle orifice at .37 mm size with
12 cone angle was standard.
Figure B- 1
TYPICAL HISTORY OF FUEL PRESSURE,
NEEDLE LIFT AND CHAMBER PRESSURE
(DI engine, 17:1 CR, 6 x .010" fuel orifices)
Fuel Pressure
Chamber Pressure
(200 psi/cm)
• Needle Lift
(Shows period of
fuel injection)
Crank Angle
Marker
TDC
By means of changes in the fuel line length, injectors, cam profiles,
fuel valve opening pressure, and plunger diameter, it was possible to study
the following variations:
No. of orifices:
_3
Orifice size (10 in.):
Rate of fuel injection:
Pilot injection
Cone angle
4, 6, 8
.008, .010, .012, .014
3 to 8 mm//°CA
10 to 20% of fuel injected at -40°CA
120° vs. 160° for DI, 8 to 12° for PC
123
-------
Air was taken from the laboratory compressed air supply and heated
(or cooled) after the filter and flowmeter. A throttle valve in the exhaust line
was used in some tests to build up exhaust pressure to simulate turbocharging.
Water mist was available at the intake, and the EGR system shown in Figure B-2
was available to supply measured amounts of exhaust gas to the intake.
Compressed Air
I
To Stack,
Muffler,
Smokemeter
= Temperature Sensor
= Exhaust Damper Orifice Plate,
= ' Shutoff Valve
Water Trap
High Pressure
Regulator
Filter
, T Strainer ,
'-*-rt;fE \ c?
Flowmeter
Water Water
Out „. . In
Heat Exchanger
Low Pressure
Regulator
Figure B-2
AIR SYSTEM PROVIDING FOR EGR
When introducing EGR, fuel flow was not adjusted to compensate
for the displaced intake airflow. Although this procedure caused the A/F
parameter (equivalence ratio} to vary with % EGR, this method permitted
comparison with past EGR studies, was simple, and also gave a wider
variation in oxygen mole fraction that if air were boosted or fuel reduced.
It was necessary to run the intake at a slight vacuum (0.5" Hg) to insure
positive flow of EGR into the intake. The state of the intake air from the
compressor was not subject to significant changes in humidity during the
course of the testing:
Equivalence ratio is taken as (F/A)/(F/A)
stoic
124
-------
humidity ranged from 17 to 33 grains/lb air (this corresponds to a + 1.5%
variation in NO emissions, according to Krause, Merrion, and Green (1973).
In this manner, the state of the intake air could be controlled and varied as
follows :
Water injection: m ^/m.c = 0 to 1.0
H2° f
EGR: m^/ift.. = ° to
sgr LOT.
Air pressure: 30 to 60" Hg
Air temperature: 100 to 200°F
Engine breathing efficiency ranged from 80% to .a maximum of 87%
as RPM dropped from 2100 to 1500. The slightly undersized intake valve
was offset by having a larger than normal valve lift (.530") .
A masked valve was used to generate swirl; rotation of. the valve
controlled both the sense and degree of swirl as shown in Figure B-3. The
nominal swirl is measured on a standard "paddle wheel" test stand with the
valve fixed in open position (.530" lift). Paddle wheel RPM is measured
by a proximity transducer. The reason that the values of RPM are a factor
of 5 or 10 below those often reported in the literature is that a subnormal
air flow was supplied to the intake (AP = 10" w.c.) . The relative values of
swirl are expected to be still valid since typically paddle wheel RPM rises
nearly linearly with air flow.
A summary of the instrumentation is given in Table B-4. The sampling
train is shown in Figure B-4. Stainless and teflon line was used throughout
the total response time of the system was controlled by that of the oxygen
analyzer (20 sec) .
An automatically controlled cooling tower keeps engine and dyna-
mometer water and oil temperatures within desired limits through a system
of heat exchangers and process piping.
125
-------
Figure B-3
PROVISION FOR AIR SWIRL
Air Flow @ = 120
CW Viewed from
Top of Engine
= adjustable
. angle of
mask position
Mask subtends .90
SECTION A-A
1.6
1.4 U
~'o 1.2
0 I i t I I
65 105 145 185 225-265 305 345
Mask Position
•viewed from Injector
126
-------
Table B-4
mf
m
°
Ta
Tf
RPM
0
Pa
*H O
,tn2^
mH2O/
me
m +r
a
T
egr
T
cool
Trtn
Input
Variable
Fuel flow
Air flow
Air temp.
"Fuel temp.
Speed
Timing
Air pressure
Air humidity
An, Water injection
?r Exhaust
\ Recirculation
egr
Temp, of recirc.
Temp. of coolant
(in)
Temp . of oil
Range
Provided
«-»£
0-150 CFM
100-200°F
100-150°F
0-2800RPM
-30°to-10°
BTDC
15-30 psia
10-100°F
0-1.00
0-30%
200-500°F
0-200°F
0-500°F
Precision
1%
1/4%
1 CFM
2°F
1°F
2 RPM
1/2° CA
.1%
2°F
.05
2%
5°F
1°F
2°F
Output
Instrumentatior Variable •
Rotameter
Fuel scale
LFE
RTD
RTD
Pickup
Bentley-
Nevada
Proximity
sensor
Manometer
Dew point
sensor
Rotameter
Orifices
RTD
RTD
RTD .
XNQ Nitric Oxide
XQ Oxygen
2
X-,..-. Hydrocarbon
Smoke
XJ-.Q Carbon Monoxide;
BMEP
T Exhaust temp.
T , Temp, of coolant
C00i (out)
T ., Temp, of oil (out)
oil
P Exhaust pressure
P Cylinder press. ,
Range
Observed Precision
0-3500 ppm
0-20%
0-1000ppm
0-50% Opacity
0-10 Bosch
0-150ft-lb
0-1500°F
0-200
0-500
15-30 psia
0-7000 psia
2%
-5%
1%
1%
.2
2%
.25%
5°F
1°F
2°F
+ 1 %
10 psia
Instrumentation
NDIR
Polarographic
FID
PHS smoke-
meter
Bosch spot
NDIR
Load cell
RTD
RTD
RTD
Manometer
Cooled AVL
transducer
(in)
-------
Smoke Sample *°
Probe (Bosch Test;
10 ft of 3/8"
teflon
I j Smokemeter
I J (opacity test)
-Knockout Filter
•Temp Gage
Heated SS ,,
LineV
Heated-FID
Analyzer)
Purge and Spai
Inlet to Confir
tegrity of Sami
>
i-Gas
ale Line
Flowmeter
Sample Pump
Condenser
Knockout Filter
O7-Analyzer
(Polarographic)
Beckman 715
NO-Analyzer
(NDIR*)
Beckman 315
x—x ©Flowmeter
Bypass Flow
Sample Pump
*Chemiluminescent analyzer available to check NO : normally NO « NO
Figure B-4
SAMPLING TRAIN
128
-------
C. Procedures and Test Matrix
Reproducibility of emissions and performance behavior has been
methodically tested. As one returns to a given set fuel-air ratio, timing,
and engine speed, the BMEP is reproduced to + 1%, nitric oxide emission
to + 2%, smoke level to + .3 Bosch number. Oscilloscope traces of needle
lift, combustion chamber pressure, and fuel line pressure manifest no dis-
cemable differences.
Response time or equilibration time after adjustments in speed,
load or timing is 2 minutes to 90% of final reading, 5 minutes to 99% of
final reading. Slowest to respond are the O--meter and exhaust tempera-
£
tures. The rate of data acquisition was paced at about 10-15 minutes per
point to allow change in operating parameter, 5 minute equilibration, and
data recording. Naturally some changes in engine variables require longer
times to set up (e.g., fuel injector changes).
Friction tests were performed for all engine configurations. The DI
head fitted with 17:1 compression piston manifested friction losses under
motoring conditions rising linearly with engine speed from 22 MEP at 700 RPM
to 35 MEP at 2100 RPM.
Each engine configuration was run 20 hours break-in before testing
as a standard shakedown procedure. During this time, minor oil leaks were
sealed and counterbalance weights were adjusted to reduce engine vibration
amplitude to below 0.5 mm. Upon first start up, the prechamber engine
showed credible BMEP and specific fuel consumption, low smoke, and base-
line NO levels (ppm) about a factor of three less than DI emission levels.
X
This performance was unexpected for an untried chamber geometry; fortunately
the customary development work was bypassed.
Emissions instrumentation was calibrated before and after testing.
Spot checks of the NDIR nitric oxide analyzer against a chemiluminescent
analyzer gave agreement to + 3%, and the difference between NO and NO
~~ x
was found to be less than 5% of the NO reading.
129
-------
Following the shakedown runs for each "top-end", baseline tests
were run over a 20-point matrix of speed, load, and timing as shown in
Figure B-5.A11 other runs were conducted over a simpler 4-point matrix.
Both test matrices were set up to emphasize peak torque and rated speed
conditions, and closely reflect the 13-Mode Cycle.
Figure B-5
TEST MATRIX
TEST SEQUENCE
SHAKEDOWN TESTS
OPERATING TEST MATRIX
REP
QUESTI
TES
EAT
OMABLE
TS
BASELINE TESTS
(Timing, Speed, and Load)
"
INTAKE AIR TESTS
• EGR
• Water Injection
• Air Temperature
• Supercharging
MIXING PARAMETERS
Rate of Injection
Orifice Size
Cam Shape
Pilot
Air Swirl
•• MATRIX FOR BASELINE RUNS ONT.Yr
•s
Motoring
96
A 48
f 32
24
19
>•« STAN
32
A
F
19
Engine Speed (RPM)
900 1500 1800 2100
X X X X
-20* -20
-20 -20
6 timings (-31 to -14) -20
-20 -20
-20 -20 -20
DARD MATRIX (ALL OTHER RUNS}:
RPM
1500 2100
-20 -20
-20 -20
Note: *tlmlng In °BTDC (start of Injection)
The timing variations were over a range wide enough to uncover the
characteristic nonlinearities in the NO vs. timing curve. Air/fuel ratio was
selected over BMEP or IMEP as the "load" variable because the latter are
inherently dependent variables.
130
-------
When A/F = 19 could not be reached because of excessive smoke
or exhaust temperature (particularly during turbocharging), the minimum A/F
was run instead. Additional baseline runs were repeated after about every
10 runs in order to check the validity of the baseline data (i.e., to detect
any unplanned deviations or systematic drift).
Variables were changed one at a time in order to clarify the NO
and smoke behavior. Runs with two effects which compensate or amplify
one another would be appropriate for a low-emissions development program,
but not for this study of mechanisms. In each case, extreme levels of the
variables (for example, 30% EGR; a range of X5 in swirl, etc.) were selected
in order to bring to the surface whatever NO and smoke changes were occurring,
The current data reduction procedure derives the following three key
measurements:
(i) NO : Expressed in Ib NO2/1000 Ib fuel, which is derived
x from the measured NO (ppm), the measured air/fuel
ratio, known molecular weights, the C/H ratio of the
fuel, and a correction for the dry measurement. No
correction for humidity is applied because measured
humidity variations in the compressed air are only
about + 5 grains I^O/lb air, which corresponds to
1% correction to NOX [Krause et al.(1973)]. Uncer-
tainty of the NDIR measurement is + 2% .
Values of gm NO2/BHP-hr were obtained directly from
the BSFC (Ib fuel/BHP-hr) and the emission value (Ib
NO2/1000 Ib fuel) by applying a conversion factor of
grams to pounds.
(ii) Soot: Expressed in percent opacity, as measured by a PHS
meter with a shielded 3" stack, with estimated accu-
racy ± .2% opacity. Although Bosch units are also
found in the literature and in fact are being recorded
in the present program, the percent opacity manifests
less scatter at low soot levels and correlates against
mass concentration (mg/m^) with less uncertainty than
the Bosch unit.
(iii) ISFC: Expressed in Ib/IHP-hr and derived from measured
BMEP, fuel flow rate, engine speed, and FMEP which
is known at each engine speed based on motoring data.
131
-------
Table B-5
TEST MATRIX
~^_^TOP END
PARAMETER I*"" -^_!^___
Baseline:
Speed (RPM)
Load (A/F)
Timing (°BTDC)
_ JJ1, ,
Standard
Air Swirl (paddle RPM) 850
Air Boost ("Hg abs) 30
Air Temp (°F) 100
Water Injection (W/F) 0
EGR (%) 0
No. x Fuel Orifice dia,(10~3in) 6x10
Fuel Orifice Area (10"4in2) 6
Fuel Rate Shape l~l
Fuel Pilot Injection fl ..
Fuel Temperature (°F) 100
Number of Variations
Number of Variation Runs
Number of Baseline Runs
Number of baseline check runs
Number of runs repeated
Total Runs
Direct Injection
CR = 17/1
Original 27-
point matrix
900 to 2 100
19 to 96
-31 to -14
350, 1600
50
200
.5,1.0,1.5
10,20,30
8x10,6x12,4x14
9, 12(6x12,6x141
^L. I^_ .
200
17
80
27
9
8
124
Direct Injection
CR = 14/1 ,
Revised 20-
point matrix
900 to 2 100
19 to 96
-31 to -14
350. 1600
5U
10,20,30
8x10, 4x14
^L
10
36
20
3
4
63
PC
Standard
30
100
0
0
1 x 27
6
100
Prechamber
CR=17/1
V/Vtot = 25%
20-point
matrix
900 to 2100
19 to 96
-12 to +3
50
200
.5,1.0,1.5
10,20,30
1x30,1x34
^L ±^
12
48
20
5
5.
78
Prechamber
CR = 19/1
V/Vtot=15%
20-point
matrix
900 to 2 IOC
19 to 96
-12 to +3
^L J^_
2
8
20
2
3
33
Prechamber
CR=17/1
V/Vtot=35%
20-point
matrix
900 to 2100
19 to 96
-12 to +3
50
10,20,30
1x30
5
20 :
20
2
3 ...
45
Prechamber
CR = 19/1
V/Vtot = 25%
20-point
matrix
900 to' 2 100
19 to 96
-12 to +3
0
', 0
.20
' 0
0
" 20
M.A.N.
20-point
matrix
900 to 2 100
19 to 96
-31 to -14
0
0
20
0
0
20
La nova
20-point
matrix
900 to 2100:
19 to 96
-31 to -14
0
0
20
0
0
20
co
-------
APPENDIX C
COMPLETE DATA FROM
SINGLE CYLINDER EMISSION TESTS
The following tables are a complete listing of single-cylinder data
gathered for eight combustion chambers as noted in column 2 (e.g. DI-14
indicates direct injection, compression ratio = 14). The remaining table
headings are given below to assist in using the listing.
VR Volume ratio
PLACE Data sheet and line number
PHI Equivalence ratio
RPM Engine speed
TIM Start of injection
SWIRL Swirl level
PA Air intake pressure
TA Air intake temperature
HO Water/fuel ratio
{.A
PCT EGR Exhaust gas recirculation
ORIF DIA Fuel orifice diameter
FUEL RATE Mean rate of injection
PILOT INJ Pilot injection
CONE ANGLE Angle subtended by opposing
fuel jets
dimensionless
dimensionless
dimensionless
_1
min
deg. crank angle
l=low, 2=med/ 3=high
inches Hg
°F
dimensionless
as % of intake
inches
3
mm
/°CA
40.10 indicates 10%
at 40°CA BTDC
deg.
133
-------
EN VR PLACE
AUTH CR PHI
PA TA
H o PCT ORFC FUEL PTLOT CONE PPM OPAC BMEP LB NOX/ OPACITY GNOX/BHP MR
2 EGR OIA. RATE INJ ANGLE NO* (SOOT) IK L6 FUEL tSFC 14
RPw DT-14
RPw DT-14
RPW DT-14
RPw 0T-14
RPw Ol-l4
RPW OT-14
RPw Di-14
RPW Di-14
RPW DI-I*
RPW 01-14
RPw Dt-14
RPw DT-14
RPW OT-14
RPW OT-14
RPw OT-14
RPw 0-1-1*
RPw OT-U
RPw DT-14
RPw DT-14
RPw Oi-14
RPw Dl-14
RPw DT-14
HPW DT-14
RPw DT-14
RPw OT-14
RPw OT-14
RPw DT-U
RPw Di-14
Rpw Oi-U
RPW OT-H
RPW DT-U
RPM DI-U
RPW DI-U
RPW DT-H
•RPW OT-U
RPW OT-U
R°W DT-U
RPW OT-U
RPW OT-U
RPW OT-U
PPW OT-U
RPw DI-U
RPW OT-H
105- 6 ,76 1500 -20
109- 9 .76 1500 -20
109-U ,76 1500 -20
110- 4 .76 1500 -20
113- 6 .76 1500 -20
111- 3 ,76 1500 -20
112- 5 .76 1500 -20
105- 3 .76 IflOO -20
104-13 .76 2100 -20
109- 8 .76 2100 -20
109-13 .76 2100 -20
110- 2 .76 2100 -20
111- 4 .76 2100 -20
112- 3 ,76 2100 -20
105- 7 ,60 1500 -20
108- 4 ,60 1500 -20
108-13 .60 1500 -20
108- 5 ,60 1500 -20
108- 6 ,60 1500 -20
108- 7 .60 1500 -20
104-14 .60 2100 -20
106-11 ,60 2100-20
109- 4 .60 2100 -20
106-H ,60 2100 -20
106-13 .60 2100 -20
106-12 .60 2100 -20
108- 8 .45 1500 -20
105- 8 .45 1500 -20
106- 7 .45 1500 -31
106- 6 .45 1500 -27
106- 5 .45 1500 -23
106- 3 .45 1500 -17
106- 4 .45 1500 -H
109-10 .45 1500 -20
109-15 .45 1500 -20
108-14 .45 1500 -20
108- 9 .45 1500 -20
108-10 .45 1500 -20
108-11 .45 1500 -20
110- 5 ,45 1500 -20
111- 2 .45 1500 -20
113- 2 .45 1500 -20
112- 6 .45 1500 -20
104-15 .45 2100 -20
106-15 .45 2100 -20
109- 7 .45 2100 -20
109-12 .45 2100 -20
108-16 .45 2100 -20
106-16 .45 2100 -20
Z
3
1
£
2
Z
2
2
2
3
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
3
1
2
2
2
2
2
2
2
2
2
2
3
i
2
2
30 100 0,00 0
30 100 0.00 0
30 100 0.00 0
30 100 0,00 0
30 100 0.00 0
30 100 0,00 0
30 100 0.00 0
30 100 0.00 0
30 100 0,00 0
30 100 0.00 0
30 100 0.00 0
30 100 0,00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
50 100 0.00 0
30 100 0,00 30
30 100 0.00 20
30 100 0.00 10
30 100 0.00 0
30 100 0,00 0
50 100 0.00 0
30 100 0,00 10
30 100 0.00 20
30 100 0.00 30
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0"
30 100 0,00 0
30 100 0.00 0
30 100 0.00 0
SO 100 0.00 0
30 100 0.00 30
30 100 0,00 20
30 100 0,00 10
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
SO 100 0.00 0
30 100 0.00- 30
.010
,010
.010
,014
.010
,010
.010
.010
.010
.010
.010
,014
.010
.010
.010
.010
.010
.010
.010
.010
.010
,010
.010
.010
.010
.010
.010
.010
,010
.010
,010
.010
.010
,010
,010
.010
.010
.010
.010
,014
,010
.010
.010
.010
.010
,010
.010
.010
.010
7,0 0,00
7.0 0,00
7.0 0.00
6.9 0,00
5.1 0.00
6,5 0,00
7.0 40.10
6.2 0.00
5,4 0.00
5.4 0.00
5.4 0.00
5.2 0.00
5.2 0.00
5,4 40.10
7,0 0.00
7.0 0.00
7.0 0.00
7.0 0*00
7.0 0.00
7.0 0.00
5.4 0.00
5,4 0.00
5.4 0.00
5,4 0.00
5.4 0.00
.5.4 0.00
7.0 0.00
7.0 0.00
7.0 0.00
7.0 0.00
7.0 0,00
7.0 0.00
7.0 0,00
7,0 0.00
7,0 0.00
7.0 0.00
7.0' 0.00
7,0 0.00
7.0 0.00
6.9 0.00
6.5 0.00
5,1 0.00
.7,0 40,17
5,4 0,00
5.4 0.00
5.4 0.00
5.4 0,00
5.4 0.00
5.4 0.00
160 1060
160 1220
160 470
160 1160
160 1120
160 1100
160 1520
160 1000
160 840
160 980
160 540
160 850
160 1180
160 11SO
160 1020
160 900
160 1480
160 220
160 430
160 890
160 800
160 740
160 1110
160 610
160 390
160 260
160 900
160 940
160 2210
160 1460
160 1060
160 730
160 ,640
160 I860
160 450
160 1350
160 290
160 520
160 890
160 880
160 1440
160 1120
160 1370
160 800
160 ^10
160 1500
160 420
160 1020
160 260
20'.0 105.5
17,0 102,3
17.0 107.6
25.0 108,7
20.0 Q4.9
21.0 106.6
2i»«0 102,3
21.0 99.2
21.0 79.1
30.0 79.1
18.0 86.5
16, 0 86,5
15.0 85.5
24,0 83.3
9.0 65.5
10,0 81,2
B.O 166.7
37.0 74.9
24,0 78.1
15.0 80.2
13.0 68-6
15,0 66.5
10.0 130.8
18.0 64.4
29,0 62.2
3b.O 61,2
4,0 61,2
. 6.5 64.4
5.0 63.3
5.5 65.4
7.0 65.4
8,0 65.4
8.0 63.3
3.5 66.5
5.0 70.7
4.0 127.7
15.0 60.1
9.0 60.1
6.0 60.1
9.0 64,4
4.0 68,6
7.5 63,3
8.0 66.5
6.0 50.6
6.0 49.6
5.0 50.6
5.0 52.7
5,5 104.4
10.0 48.5
31,95
36.77
14.16
3^.96
33.75
33.15
45,81
30.14
25.32
29.54
16.27
25,62
35.56
35,56
38.83
3*, 26
56.34
8.36
lt>,37
33.88
30.46
28,17
42.26
23.22
14,65
9.90
45.68
47.71
112,18
74.11
53.80
37,05
32.49
94,41
22.84
68.52
H.72
. 26,39
45.17
44,67
73.09
56.85
69.54
40.61
41.11
76,14
21,32
51.77
13.20
20.0
17.0
17.0
25,0
20,0
21.0
25,0
21.0
21.0
30.0
18.0
18,0
15.0
24.0
9.0
10.0
8,0
37.0
24,0
15.0
13.0
15.0
10.0
18.0
29,0
35.0
4,0
6.5
5,0
5.5
7,0
8.0
8.0
3.5
5.0
4.0
15.0
9.0
6.0
9.0
4,0
7.5
8,0
6.0
6.0
5,0
5.0
5.5
10.0
.370
,368
.346
.363
.365
.367
.374
.363
.378
.376
.355
.357
.362
.367
.346
.345
.338
.367
,355
,348
.336
,339
.363
.348
.355
.359
.315
.316
,320
.315
.315
,318
.325
.298
.300
.314
,319
.320
.317
.324
.306
.322
.316
,301
.310
.294
.294
.322
.303
6. B6
7,57
2.72
7,03
7.00
6.76
9,60
6.37
6.09
7,08
3.59
5.68
8.02
8.18
7.80
6.94
9.88
1.64
3.44
6,93
6.80
6,40
8.65
5.46
3.61
2.45
9.04
9.36
22.37
14,43
10.48
7.27
6,58
17.29
4.16
11.60
2.96
5.34
9.05
8.98
13.66
11.40
13.51
ft. 99
9,44
16,43
4.54
9.87
2.99
-------
EN VR
AUTH CR
RPW Dl-14
RPW 01-14
RPW Di-14
RPw Of-14
RPw 1>I-14
RPW Dj-14
RPW Dj-14
RPW Df-14
RPW Dj-14
RPW oj-14
PLACE
107- 3
108- 3
110- 3
113- 8
111- 5
112- 4
106- 8
104-16
106- 9
105- 1
RPM
PHI
.45 2100
,45 2100
.45 2100
.45 2100
.45 2100
.45 2100
..30 1500
.30 2100
.15 1500
.15 2100
SfcIRL
TIM PA TA
-20
-20
-20
-20
-20
-20
-20
-20
-20
-20
2
2
2
2
2
2
2
2
2
2
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
HO PCT
2 EGR
0.00 20
0.00 10
0.00 0
0.00 0
0.00 0
0.00 0
0.00 0
0.00 0
0.00 0
0.00 0
ORFC FUEL
DIA RATE
.010
.010
.014
.010
.010
.010
.010
.010
.010
.010
5.4
5.4
5.2
3.8
5,2
5.4
7.0
5.4
7.0
5.4
PILOT CONE
INJ ANGLE
0.00
0.00
0.00
0.00
0.00
40.17
0.00
0.00
0.00
0.00
160
160
160
160
160
160
160
160
160
160
PPM OPAC
NOX (SOOT)
430
790
680
1180
1190
1180
400
750
360
180
7.0
5.0
8.0
6.0
8.0
7.5
5.0
4,0
4.5
4.0
BMtP LB NOX/ OPACITY GNOX/BHP MR
IK LB FUEL ISFC (MIJLTI)
47.5
48.5
51.7
52.7
50.6
50.6
39.0
27.4
-7.4
-3.2
21.83
40.10
3^.52
59.89
60.40
59,89
30,46
57.10
54.83
27.41
7.0
5.0
8,0
6.0
8.0
7.5
5.0
4.0
4.5
4.0
.309
.303
.306
.306
.301
.303
.290
.280
,495
.246
5.07
9.07
7.69
13.27
13.33
13,35
6.40
15.28
-43,44
-82.50
to
01
-------
CT>
EN VR
flUTM CR
RPW 01-17
RPJ OT-17
PPW Oj-i7
RPW OT-17
RPw DT-17
RPw Di-17
RPW DT-17
PPw DI-I?
RPw OT-17
RPw DT-17
RPW OT-17
RPW OT-17
RPw 0^-17
RPW Di-i7
RPW OT-17
. RPw Oj-17
RPW OT-17
HPW DT-17
RPW Oi-i7
RPW DT-17
RPW Dt-17
RPw Oj-17
RPW Di-17
RPw DT-17
RPW DT-17
RPw Oj-l7
RPW Dj-17
RPW DT-17
RPW Ol~l7
RPW DT-17
HPW DT-17
RPw DT-i7
RPW Dj-17
RPw Ot-17
RPW Oj-17
RPW OT-17
RPw Di-17
RPW Di-17
RPw OT-17
RPw Di-17
RPw Ot-17
RPw DT-17
RPw Df-17
RPW Dl-17
RPW OT-17
r» r 1 " i 1 •
RPW OT-17
RPW Ol-)7
RPW OT-17
HPw DT-17
PI ACF RPM SWTRL H o PCT ORFC FUEL PILOT CONE PPM OPAC
PHI TIM PA TA 2 CGR DlA. RATE INJ AMGLE NOX (SOOT)
62-10 .76 1500 -20
73- 7 .76 1500 "20
So-15 .76 1500 -20
48-12 .76 1500 -20
73- 5 .76 1500 -20
48- 5 .76 1500 -20
50- 8 .76 1500 -20
63- 5 .76 1500 -20
02-14 .76 1500 -20
64- 8 .76 1500 -20
64- 1 .76 1500 -20
71- 7 .76 1500 -20
71-13 .76 1500 -20
46-16 .76 1800 -20
47-12 .76 2100 -20
49- 7 .76 2100 -20
48-14 .76 2100 -20
50-12 .76 2100 -20
62-16 .76 2100--20
64-10 .76 2100 -20
63- 6 .76 Z1UO -20
64- 2 ,76 2100 -20
71- 8 .76 2100 -20
71-U .76 2100 -20
68- 2 .60 1500 -20
46- 8 .60 1500 -20
72-14 ,60 1500 -20
67- 3 .60 1500 -20
66- 5 .63 1500 -20
67-13 .60 1500 -20
68- 3 .60 1500 -20
67-12 .60 1500 -20
67-11 ,60 1500 -20
69- 8 .60 2100 -20
47-11 .60 2100 -20
67- 5 ,60 2100 -20
66- 3 .63 2100 -20
41-11 , 3 2100 -25
41-12 .63 2100 -25
41-13 .63 2100 -25
41-14 ,63 2100 -25
69-11 .60 2100 -20
69-10 .60 2100 -20
69- 9 ,60 2100 -20
72-15 .45 1200 -20
47- 1 .45 1500 -20
72-13 .45 1500 -20
73- 1 ,45 1500 -20
68- 5 .45 1500 -20
im •
2
2
2
1
3
3
2
2
2
2
2
2
2
2
2
3
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
JO 100 0.00 0
30 100 0,00 0
JO 100 0.00 0
30 201 0.0° °
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 . 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0,00 0
30 206 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100
30 100
30 100
50 100
60 100
30 100
30 100
30 100
30 100
30 100
30 100
50 100
60 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
0.00 0
0.00 0
0.00 0
0,00 0
0.00 0
0.00 10
0,00 10
0.00 20
0.00 30
0,00 0
o.oo o
0.00 0
o.oo o
0,00 0
,50 0
1.00 0
1.50 0
0.00 10
0.00 20
0,00 30
0.00 0
0.00 0
0.00 0
0,00 0
0.00 0
.010
,010
.010
.010
.010
.010
.010
.012
.014
.014
.010
,010
.010
.010
.010
.010
.010
,010
.014
.014
.012
.010
.010
.010
.010
.010
.010
.010
.010
.010
.010
.010
,010
,010
.010
• 010
.010
.010
.010
.010
,010
.010
.010
.010
.010
.010
.010
.010
,010
7.0
7,0
7.0
7.0
7,0
7.0
7,0
6.3
6.9
7,5
6.5
6.1
5.1
6.2
5.4
5.4
5.4
5.4
5.2
6.0
4,9
5.2
4.8
3.8
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
5.4
5.4
5.4
5.4
5.4"
5.4
5.4
5.4
5.4
5.4
5,4
7.8
7.0
7.0
7,0
7,0
0.00
0,00
0.00
0.00
0,00
0.00
0.00
o.oo
0.00
0.00
0.00
o.oo
0.00
0.00
0.00
0.00
0.00
o.oo
0.00
o.oo
0.00
o.oo
0.00
0.00
0.00
0.00
0.00
o.oo
0.00
0.00
0,00
0.00
0.00
o.oo
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0,00
0.00
0,00
0.00
0,00
0.00
o.oo
0,00
B*EP
IK
160 1950 12,0 104,4
160 2130 11.0 102.3
160 1750 12.0 111.8
160 2090 7.0 106.6
160 2300 8.0 106.6
160 3380 1.5 105,5
UO 2320 11.0 91.8
160 1250 21.0 l'J3.4
160 1370 25.0 100.2
160 1480 41.0 102,3
160 1390 20.0 107,6
160 1430 21.0 103.4
160 1200 20,0 98,1
160 1750 9.0 97.1
160 Ife3o 10.0 88.6
160 2630 6.0 69,7
160 1810 5.5 80.2
160 1910 U.O 77.0
160 860 2H.O 79,1
160 1340 31.0 8.3.3
160 1140 22,0 64.4
160 1110 22.0 85.5
160 1400 1^.0 81.2
160 1110 25,0 79,1
160 1990
160 2UO
160 2050
160 2220
160 2030
160 1870
160 1^00
160 760
160 370
160 1670
160 1710
160 1620
160 1470
160 2500
160 1280
160 900
160 670
160 1460
160 790
160 400
160 1610
160 1710
160 1910
160 1690
160 1?50
12.0
7.0
10.5
11.0
7.5
13.0
13,0
23.0
36.0
9.0
4.0
8.5
7,0
2.9
2,8
3.5
3.0
11.0
21.0
40.0
7.0
1.0
7.5
7.0
8,0
84.4
91. B
87.6
Ift8. 8
177.2
84.4
84.4
81.2
80.2
71.7
69.6
146.6
177,2
71.0
74.0
74,0
72.0
71.7
69.6
67.5
64.4
63,3
64.4
62.2
6554
tB NOX/ OPACITY
LB FUEL
5H.77
64.19
52.74
62.99
69.32
101,67
69.92
37,67
41,29
44.60
41,89
43.10
36,17
52.74
49.12
79.26
54.55
57.56
25.92
40.38
34.36
33.45
42.19
33.45
75.76
80,33
78.04
64.51
74,06
71.19
72.33
28,93
14.09
63,57
65.10
61.67
53,63
91.21
46,70
32.83
24.44
55.58
30.07
15.23
81.72
86.80
90,95
85.78
88,83
12,0
11.0
12-0
7,0
8.0
1.5
11.0
21.0
25.0
41.0
20,0
21.0
28.0
V.O
10,0
6,0
5,5
11.0
28,0
31.0
22.0
22.0
15.0
25.0
12.0
7.0
10.5
11.0
7.5
13,0
13.0
23.0
36.0
9.0
4,0
8.5
7.0
2.9
2.8
3.5
3.0
11.0
21,0
40,0
7.0
1.0
7o5
7.0
8,0
GWOX/9HP HR
I5FC (MULTI)
mmmm •••••••»
.343
.340
.335
.327
.338
,312
.327
.352
,349
.358
.347
,349
,363
.324
.327
.311
,315
.325
,348
,348
.345
.342
.359
.365
.307
.312
.315
.322
.407
.307
.307
,316
.319
.306
.302
.328
.357
,397
.393
,393
.398
,306
.312
.318
. .299
,291
,297
.301
,292
11.20
12.18
9.70
11,40
12.96
17.63
13.01
7.37
8,06
8.89
8.03
8.37
7.38
9.89
9.76
14,96
10,71
11.77
5,65
8.69
7.30
7,03
9.40
7.64
13.47
14.26
14.11
14.09
15.52
12.66
12.66
5,33
2.63
12.48
12.71
11.10
10,19
23.29
11.68
8.21
6.24
10,91
6.07
3.16
14.51
15.62
17.69
15.99
15,89
-------
EN VR
AUTH CR
RPW DI-i7
RPW Dl-17
RPW Dj-17
RPW DT-17
RPw. Ui-i7
RPw DT-I?
RPw DT-17
RPw Oj-17
RPW Di-17
RPW Dj-17
RPW Di-17
RPW DT-17
RPW DT-17
RPW DT-17
RPW DT-17
RPW Di-17
RPW . Dj-17
RPw DT-17
RPW DT-I?
RPW Ui-17
RPw Di-17
RPw OT-17
RPW Ot-17
RPW 0T-17
RPW Di-17
RPW Dl-17
RPW Di-17
RPW Di-17
RPw DT-17
RPW DT-17
RPW Di-17
RPW DT-17
RPW Di-17
RPW DT-17
RPw DT-17
RPW DT-17
RPw DT-17
RPW DT-I?.
RPw OT-17
RPW Oi-17
RPW Di-17
RPw Oi-17
RPW OT-17
RPw Dj-17
RPw DT-17
RPW Dj-i7
RPW Di-17
RPW Dl-17
RPW Di-17
RPW Oj-17
PLACE RPM SwIRL H 0' PCT
PHI TIM PA TA 2 EGP)
65-1= .43 i5oo -20
65- 7 .45 1500 -20
50-13 .45 1500 -20
50-14 .45 1500 -14
62- 9 .45 1500 -31
62- 8 ,45 1500 -28
73- 8 .45 1500 -25
48-11 ,45 1500 -14
48- 3 .45 1500 -20
66- 8 .45 1500 -20
48- 4 .45 1500 -14
48-10 .45 1500 -20
67- 4 .45 1500 -20
49- 4 .45 1500 -20
65- 9 .45 1500 -20
49- 5 .45 1500 -14
50- 7 .45 1500 -14
50- 6 .45 1500 -20
68- 8 .45 1500 -20
68- 7 .45 1500 -20
68- 6 .45 1500 -20
54- 8 .45 1500 -20
54- 9 .45 1500 -14
52-13 .45 1500 -20
52-14 .45 1500 -14
63-11 .45 1500 -28
63-12 .45 1500 -31
62-13 .45 1500 -20
63- 1 .45 1500 -28
63- 2 .45 1500 -31
51- 4 .45 1500 -20
64- 5 .45 1500 -20
64- 6 .45 1500 -28
64- 7 .45 1500 -31
58- 1 .45 1500 -20
71- 6 ,45 1500 -20
63-16 .45 1500 -31
63-15 .45 1500 -28
53-11 .45 1500 -20
53-12 .45 1500 -14
72-16 .45 1800 -20
73- 3 .45 1800 -20
47- 7 .45 2100 -20
69- 6 .45 2100 -20
68-H .45 2100 -20
51- 1 .45 2100 -20
47- 9 .45 2100 -17
47-10 .45 2100 -14
47- 8 ,45 2100 -25
41- 6 ,40 2100 -25
2
2
2
2
2
2
2
1
3
3
3
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
3
2
2
2
2
2
2
2
2
3.0 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
bO 100 0.00 0
60 100 0.00 0
60 100 0.00 0
60 100 0,00 0
30 200 0.00 0
30 202 0.00 0
30 100 0.00 10
30 100 0.00 20
30 100 0.00 30
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100. 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0,00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0,00 0
30 100 0.00 0
30 100 0.00 0
30 100 0,00 0
30 100 0.00 0
30 100 O.OC 0
30 100 0.00 0
30 100 0.00 0
30 100 0,00 0
30 100 0.00 0
30 100 0,00 0
30 100 0.00 0
ORKC FUEL PILOT CONE PPM OPAC BMEP LH NOX/ OPACITY GNOX/BHP HR
OIA. RATE INJ ANGLE NOX (SOOT) IK LB FUEL ISFC (MULTI)
.010
.010
.010
.010
.010
.010
.010
.010
.010
.010
.010
,010
.010
.010
.010
.010
.010
.010
.010
.010
.010
.008
.008
.012
.012
.012
.012
.014
.014
.014
.014
.014
,014
.014
.010
.010
.010
.010
.010
.010
.010
.010
.010
.010
.010
.010
.010
.010
.010
.010
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7,0
7.0
7,0
7.0
7.0
7,0
7.0
7.0
7.0-
7.0
, 7.0
7.0
5.6
5.6
6.3
6.3
6.3
6.3
6.9
' 6.9
6.9
6.9
7.5
7.5
7.5
5.1
6.1
6.5
6.5
6.5
6.5
6.2
6.2
5.4
5,4
5.4
5.4
5.4
5.4
5.4
5.4
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0,00
o.-oo
o.oo
0.00
0.00
o.oo
0.00
o.oo
0.00
0.00
o.oo
0.00
0.00
o.oo
o.oo
0.00
.0,00
0.00
o.oo
0.00
o.oo
o.oo
o.oo
0,00
0.00
0.00
o.oo
0.00
o.oo
0.00
0.00
0.00
0.00
0.00
o.oo
0.00
o.oo
0.00
0.00
160 1«70
160 1800
160 1670
160 1150
160 2520
160 2470
160 3750
160 1405
160 2540
160 3080
160 1710
160 1«30
160 1910
160 1910
160 1-990
160 1350
160 146Q
160 2060
160 1650
160 950
160 580
120 2010
120 1420
160 1270
160 825
160 2320
160 2720
160 950
160 1710
160 2130
160 1020
160 990
160 1950
160 2390
160 1180
160 1250
160 2950
160 2470
160 1905
160 134Q
160 1830
160 2170
160 1670
160 1690
160 1655
160 1460
160 1340
160 970
160 2300
160 2050
0.0 6". s
5.5 65.4
3.7 65.4
5.9 65.4
3.0 43.3
3.0 49.6
4.0 63.3
1.0 63.3
1.2 63.3
2.5 65.4
1.0 63,3
.9 63.3
9,0 137.1
3.0 176.2
6.0 154.0
3.0 173.0
7.0 59.1
6.0 58.0
9,0 63.3
16.0 63,3
20,0 64.4-
1,0 66.5
1.2 67.5
4,0 63.3
7.0 63.3
3.0 62.2
2.5 60.1
13,0 60,1
8,0 59.1
7,0 53,8
11.0 63.3
17.0 62,2
7.0 61.2
6,5 58.0
11.0 63,3
9.0 61.2
2.0 63.3
2.5 66.5
2.0 63.3
3.5 63.3
5.5 59.1
4.5 60.1
1.0 48,5
6.5 52.7
5,5 52.7
3.0 52.7
1,7 48,5
2.0 48.5
.5 46.4
1.0 40.1
*4.92
91.37
84.77
58.37
127.91
125.37
190,34
71.32
128,93
156.34
86.80
92.89
96,95
96.95
101.01
68.52
74.11
104.56
83.75
48.22
29.44
102,02
72.08
64,46
41.88
117.76
138.06
48.22
86.80
108.12
51.77
50.25
98.98
121.31
59.89
63,45
149.74
. 125.37
96.69
68.02
92,89
110,15
84.77
85.78
84.01
74.11
68.02
49.24
116.74
117.06
6,0
5.5
3.7
5.9
3.0
3.0
4.0
1.0
1.2
2.5
1.0
.9
9.0
3.0
6.0
3.0
7.0
6.0
9,0
16.0
20.0
1.0
1.2
4.0
7.0
3.0
2.5
13.0
8.0
7.0
11.0
17.0
7.0
6.5
11.0
9.0
2.0
2.5
2.0
3.5
5.5
4.5
1.0
6.5
5.5
3.0
1,7
2.0
.5
1.0
,29f!
.293
.288
.289
.376
.344
.299
.282
.285
.291
.278
.281
.295
.294
.312
.295
.291
.295
.299
.297
.295
.289
.286
.287
.291
.299
.306
.305
.310
.333
.297
.301
.306
.318
.300
.304
.306
.296
,287
.282
.298
.293
.282
.283
.283
.285
.283
.281
.286
.288
16,89 '
16.42
14.94
10.34
33,00
28.42
35,10
12.41
22.65
27.83
14,90
16.09
15,22
14.66
16.51
10.43
13,56
19.46
15.45
8.85
5.35
18.02
12.55
11.43
7.54
21.84
26.45
9,19
16.90
23.19
9.50
9.37
18.84
24.38
11.10
12.02
28,32
22.65
17.15
11.84
18.11
21.00
17.32
17.13
16.78
14.90
13.95
10.01
24.57
26.18
-------
V EN VR
AUTH CR
RPW 01-17
RPw DT-17
RPW DT-17
RPw UT-17
RPw Oi-l7
RPw DT-17
RPw OT-17
RPW DT-17
RPu D T — 1 7
n r w "^ i 1 *
RPw Oi-17
RPw Di-17
RPw Di-17
RPW Dj-17
RPw DT-17
RPw Dl-17
RPW DT-17
RPW OT-17
RPw Dl-17
RPw OT-17
RPW DT-17
RPw Oi-l7
RPw Dl-17
RPw DT-17
RPW OT-17
w RPW DT-17
0> RPw DT-17
RPW DT-17
RPW DT-17
RPw Dl-17
RPW Dl-17
RPW DT-I?
RPW Df-17
RPW OT-17
PLACE RPM SWIRL
PHI TIM
feU- 3 ,4b 2100 -20
48- 6 .45 ?100 -20
66- 9 .45 2100 -20
67- 6 .45 2100 -20
65-11 .45 2100 -20
49- 8 .45 2100 -20
50-11 .45 2100 -20
41- 7 .40 2100 -25
41- 8 .40 2100 -25
41- 9 .40 2100 -25
69- 5 .45 2100 -20
69- 4 .45 2100 -20
69- 3 .45 2100 -20
68-10 .45 2100 -20
54-11 .45 2100 -20
53- 3 .45 2100 -20
62-15 .45 2100 -20
64- 9 .45 2100 -20
58- 5 .45 2100 -20
71- 9 .45 2100 -20
54- 1 .45 2100 -20
45- 8 .30 1500 -20
65-14 .30 1500 -20
40-12 .32 1500 -25
40-13 .32 1500 -25
40-14 .32 1500 -25
4Q-15 .32 1500 -25
41- 1 .31 1500 -15
41- 2 .31 1500 -15
41- 3 .31 1500 -15
41- 4 ,31 1500 -15
47- 6 .30 2100 -20
65-13 .30 2100 -20
1
3
3
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
H 0 PCT
PA TA 2 EGR
•>0 100 0.00 0
30 100 0.00 0
30 100 0.00 0
50 100 0.00 0
60 100 0.00 0
60 100 0.00 0
30 205 0.00 0
30 100 .50 0
30 100 1.00 0
30 100 1.50 0
30 100 0.00 10
30 100 0.00 20
30 100 0.00 30
30 100 0.00 30
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
30 100 0.00 0
60 100 0.00 0
30 100 0.00 0
30 100 .50 0
30 100 1.00 0
30 100 1.50 0
30 100 0.00 0
30 100 .50 0
30 100 1.00 0
30 100 1.5-0 0
30 100 0,00 0
60 100 0,00 0
ORFC FUEL
DIA. RATE
.0 0 5.4
.010 5,4
.010 5.4
.010 5.4
.010 5.4
,010 5.4
.010 5.4
.010 5.4
.010 5.4
.010 5.4
.010 5.4
.010 5.4
.010 5.4
.0.10 5.4
,008 4.2
.012 4.9
.014 5.2
.014 6.0
.010 3.8
.010 4.fl
.010 5.2
.010 7.0
.010 7.0
.010 7.0
.010 7.0
.010 7.P
.010 7.0
.010 7.0
.010 7.0
.010 7.0
.010 7.0
.010 5.4
.010 5.4
PILOT ''ONE PPM OPAC BMEP LB NOX/ OPACITY GNOX/BHP HP,
INJ ANGLE NOX (SOOT) IK LB FUEL ISFC (MULTI)
0.00
0.00
0.00
o.oo
0.00
0.00
o.oo
0.00
0.00
o.oo
o.oo
0.00
0.00
0.00
0.00
o.oo
0.00
o.oo
0.00
0.00
0,00
0.00
0,00
0.00
0.00
0.00
o.oo
0.00
o.oo
0.00
o.oo
0,00
0.00
160 I'lO
160 2390
160 2860
160 1460
160 1*50
160 1630
160 1930
160 1520
160 750
160 590
160 1560
160 1020
160 660
160 700
120 1650
160 980
160 760
160 860
160 980
160 1340
160 1520
160 1340
160 1600
160 1510
160 1510
160 1040
160 660
160 940
160 600
160 460
160 320
160 830
160 1270
2.1 4B.3
.6 48.5
2.2 48.5
7.0 108.7
7.0 129.8
3.0 139.3
2.2 4b,4
1.2 40.1
1.7 40.1
1.7 40.1
7.0 52.7
15.0 52.7
15.0 51.7
13.0 51.7
1.5 52.7
5.0 47.5
11.0 46.4
13.0 48,5
7.0 49.6
5.0 48.5
3.0 50.6
3.0 36.9
4.0 Io7.6
1.2 40.1
1.0 40.1
1,2 40.1
1.2 40.1
2.4 40.1
2.6 40.1
2.6 40.1
2.5 40.1
.6 24,3
3.5 83.3
86, bo
121.31
145.17
74.11
73.60
82.74
97.96
86,80
42.83
33.69
79.18
51.77
33.50
35.53
83,75
49,74
38.58
43,65
49,74
68.02
77,15
102.02
121.82
107.78
107,78
74.23
47,11
68,59
43,78
33,56
23.35
63,19
96,69
2.1
,6
2.2
7.0
7.0
3.0
2.2
1.2
1.7
1.7
7.0
15,0
15.0
13,0
1.5
5.0
11.0
13.0
7.0
5.0
3,0
3.0
4.0
1.2
1.0
1,2
1.2
2.4
2.6
2.6
2.5
.6
3.B
.278
.273
.285
.310
.322
,305
.280
.287
.288
.290
.283
.283
.287
.287
.282
.285
.299
.296
.292
.299
.280
.281
,298
.295
.294
,294
,295
.283
.286
,287
,286
.262
,307
17.49
24.02
29.97
13.36
13.30
13.96
20,39
19.31
9.58
7.58
15.82
10,34
6.82
7,23
16,62
10.35
8.50
9.35
10.45
14.76
15.41
20.67
20.08
22,34
22,25
15,33
9,76
13.66
8.81
6.78
4.70
15.81
18.30
RPW Dj-17 47- 5 .15 2100 -20 2 30 100 0.00 0 .010 5.* 0.00 160 200 -4 -10.5
30.46
.4 .259 -15,16
-------
Ew VR PLACE RPM SwIRL H 0 PCT ORFC FUEL PILOT CONE PPM OPAC
AUTH CR PHI TIM PA TA 2 EGR DIA. RATE INJ ANGLE NOX 0
3.0
3.0
3.0
1.5
1.5
2.0
3.0
2.5
GNOX/BHP HR
ISFC (MULTI)
.376
.445
.386
.366
.376
.386
.422
.426
.408
.393
,424
.378
.405
.389 .
.418
.412
.340
.340
.337
.340
.376
.347
.347
.410
.343
.351
.363
.320
,311
,302
.308
.314
.318
.322
.340
.316
.314
.315
.319
.311
.310
.313
.319
.317
.321
.301
.301
.341
.295
2.70
2.54
3.09
1.68
1.48
1.33
2.59
2.94
2.43
3.08
3.35
1.97
1.21
1.44
3.16
3.97
3.57
3.50
3.23
2.20
1.12
4.01
4,28
3.76
3.95
2.45
1.72
4.93
5.15
4.94
6,54
8.25
4.88
5.70
4.26
5.52
4.02
2.52
1.74
5.16
2.76
1.50
5.48
5.52
5.16
6.72
7.74
6.39
7.72
-------
EN VR
AUTH CR
RPW PC17253
RPW PC1725
RPW PC1725
PPw PC1725
RPw PC1725
RPW PC1725
RPW PC1725
RPW PC1725
RPW PC1725
RPW PC1725
RPW PC1725
RPW PC1725
PLACE
81- 5
81- 6
81- 7
83- 7
83- 6
B3- S
84- 8
89-U
79- 8
78- 9
79- 9
78-10
RPM SWIRL
PHI TlK PA TA
.*$ *100
.45 2100
.45 2100
.45 2100
.45 2100
.45 2100
.45 2100
.45 2100
.30 1500
.30 2100
.15 1500
.15 2100
.6
-6
-6
-6
-6
-6
-6
-6
-6
-6
-6
-6
0
0
0
0
0
0
0
0
0
0
0
0
20 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
H 0 PCT ORFC FUEL PILOT CONE PPM OPAC
2 EGR OlA. RATE INJ ANGLE NOX (SOOT)
.bo o
1.00 0
1.50 0
0.00 10
0.00 20
0.00 30
0.00 0
0.00 0
0.00 0
o.oo o
o.oo o
0.00 0
0.000
0.000
o.ooo
0.000
0.000
0.000
0.000
0.000
0.000
0*000
0.000
0.000
5.0
5.0
5.0
5.0
5.0
5.0
6.0
5.0
7.0
5.0
7.0
5.0
0.00
0.00
0*00
0.00
0.00
0.00
0.00
o.oo
0.00
o.oo
0.00
0.00
12
12
12
12
12
12
8
30
12
12
12
12
390
180
130
610
36Q
280
670
550
530
460
120
100
0.0
o.o
o.o
2.0
2.0
2.0
1.5
1.5
1.5
1.5
2.5
3.0
8MEP LB NOX/
IK LB FUEL
42.2
41.1
38.0
42.2
42.2
41.1
42.2
41.1
31.6
19.0
0.0
-14,8
19.80
9.14
6.60
30.96
18.27
14.21
34.01
27,92
40,35
35.02
18.27
15.23
OPACITY GNOX/BHP HR
ISFC (MULTI)
0.0
0*0
0*0
2,0
2.0
2.0
1.5
1.5
1.5
1.5
2.5
3.0
.^01
.305
.317
.301
.299
.304
.309
.324
.286
• 282
1.314
.308
4.44
2.09
1.62
6.95
4,08
3.25
7,82
6,80
R.55
9.89
46.74
-7.23
EN VR
AUTH CR
RPW PCi73*>
RPw Pcl735
RPW PC1735
RPW PC1735
RPW PC1735
RPW PC1735
RPW PC1735
RPw PC1735
RPW PC1735
RPW PC1735
RPW PC1735
RPW Pfl735
RPw PC1735
RPw Pcl735
RPw Pfl735
RPW Pfl735
RPw Prl735
RPw Pfl735
HPW Prl735
RPW Prl735
RPW PC1735
RPW PC1735
RPW PC1735
RPw Pr 1735
RPW Prl735
HPW Pfl735
RPw Prl735
RPW PC1735
RPw PC1735
RPW Pcl735
RPw PC1735
RPW PC1735
RPw PC1735
PLACE
9b- 9
97-15
98- 4
95- 7
94- 4
98- 5
95-10
94- 5
96- 3
95-11
97-10
97-16
97- 9
97- 8
97- 7
95-13
95-16
9.5-15
95-12
95-14
98- 3
95- fl
94- 6
97- 6
97-14
97- 5
97- 4
97- 3
98- 6
96- 1
94- 7
96- 2
94- 8
RPM
PHI
.76 i50o
.76 1500
.76 1500
.76 1«00
.76 2100
.76 2100
.60 1500
.60 2100
.45 1200
.45 1500
.45 1500
.45 1500
.45 1500
.45 1500
.45 1500
*4S 1500
,45 1500
.45 1500
.45 1500
.45 1500
.45 1500
.45 1800
.45 2100
.45 2100
.45 2100
.45 2100
.45 2100
.45 2100
.45 2100
.30 1500
.30 2100
.15 1500
.15 2100
SWIRL
TIM PA TA
-6
-6
-6
-6
-6
-6
-6
-6
-6
-6
-6
-6
-6
-6
-6
0
-12
-9
-3
3
-6
-6
-6
-6
-6
-6
-6
-6
-6
-6
-6
-6
-6
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
30 100
50 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
50 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
30 100
*0 100
30 100
JO 100
30 100
30 100
30 100
30 100
30 100
30 100
H 0 PCT ORFC FUEL PILOT CONE
2 EGR OIA RATE INJ ANGLE
0.00 0
0.00 0
0.00 0
0.00 0
0.00 0
o.oo o
0.00 0
0.00 0
o.oo o
0.00 0
o.oo o
0.00 0
0.00 10
0.00 20
0.00 30
0.00 0
0.00 0
0.00 0
0.00 0
0,00 0
0.00 0
0.00 0
0.00 0
0.00 0
0.00 0
0.00 10
0.00 20
0.00 30
o.oo o
0.00 0
0.00 0
o.oo o
0.00 0
o.ooo
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
o.ooo
o.ooo
0.000
0.000
o.ooo
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
o.ooo
0.000
o.ooo
0.000
0,000
7.0
7.P
7.0
6.0
5.0
6.0
7.0
5.0
8.0
7.0
7.0
7,0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
7.0
6.0
5.0
5.0
5.0
5.0
5.0
5.0
6.0
7.0
.5.0
7.0
5.0
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0,00
0.00
0.00
0.00
o.oo
0.00
0.00
0.00
o.oo
0.00
0.00
0.00
0.00
0.00
0.00
0,00
o.oo
o.oo
o.oo
0.00
0,00
12
12
8
12
12
8
12
12
12
12
12
12
12
12
12
12
12
12
12
12
8
12
12
1.2
12
12
12
12
8
12
12
12
12
PPM OPAC
NOX (SOOT)
46Q
630
550
540
630
670
550
730
670
730
720
490
900
390
180
7bO
1180
780
640
790
870
870
870
900
960
870
4bO
230
730
690
380
150
180
13,0
-0.0
18.2
6,0
3.0
7.6
1.0
2.5
,5
1.0
1.8
-0.0
1.8
3.4
3.4
0,0
8,0
2.5
0.0
0.0
2,5
2.0
8,0
11.5
-0.0
12.5
13.7
18.2
17.5
1.0
9.0
3.0
5.0
BMEP LB NOX/ OPACITY GNOX/BHP HR
IK LB FUEL ISFC (MULTI)
?3.9
168.8
90,7
84,4
74.9
73.8
82.3
61.2
64.4
59.1
60.1
116.0
61.2
60.1
60.1
58.0
58.0
60.1
59.1
5b.9
57.0
50.6 .
39.0 .
40.1
U4.4
39,0
40.1
40.1
33.8
31.6
12.7.
.0
-13.7
13,86
18.99
16,58
16.27
18.99
20.19
20.94
27.79
34.01
37,05
36,55
24.87
45.68
19.80
9.14
38.07
59,89
39.59
32,49
40,10
44.16
44.16
44.16
45.68
48.73
44.16
23,35
11.67
37.05
52.53
28,93
22.84
27.41
13.0
-0.0
18.2
6.0
3.0
7.6
1.0
2.5
.5
1.0
1.8
-0.0
1.8
3.4
3,4
0.0
8.0
2.5
0.0
0.0
2.5
2.0
8,0
11.5
-0.0
12.5
13.7
18.2
17.5
1.0
9.0
3.0
5.0
,38b
.405
.383
,382-
.368
.373
.337
,336
.299
.315
.312
.328
,413
.311
.313
.317
.318
.311
.315
.326
.317
.315
.329
.317
.343
.'321
.317
.317
.340
.302
.483
.460
.313
3.C1
3.98
3,60
3.71
4.4Q
4.77
4.08
6.21
5.95
7.25
7,05
4.44
11.65
3.81
1.77
7.53
11..90
7.62
6.35
8.23
8.78
9.47
11.13
10.98
10.24
10.85
5.61
2.81
10.14
11.75
16.41
20.46
-1ft. 16
-------
EN \/R
AUTH CR
RPW PC1925
RPW PC1925
RPW PC1925
RPW PC1925
RPW PC1925
RPW PC1925
RPw PC1925
RPW PC1925
RPw PC1925
RPW Pr-1925
flPw PC1925
Rpw PC1935
RPW PC1925
RPW Prl9?5
RPw Pcl925
HPW Prl925
RPw Pcl025
RPW PC1925
RPW Pcl925
RPW Pcl925
RPW PC1925
RPW Pcl925
HPW PC1925
RPw Pcl925
PLACE
98-10
96- 9
99-U
99- 4
99-16
99- 5
101- 7
101-10
101- 9
101- 8
99- 7
99- 6
99- 6
99-10
99-11
100- 1
101- 6
101- 5
101- 4
101- 3
99-12
100- 2
99-13
100- 3
RPM
PHI
.76 1500
.76 laoi)
.76 2100
.60 1500
.60 2100
.45 1500
.45 1500
.45 IbOO
.45 1500
.45 1500
.45 1500
.45 1500
.45 1500
.45 1500
.45 1500
.45 2100
.45 2100
.45 2100
.45 2100
.45 2100
.30 1500
.30 2100
.15 1500
.15 2100
SwIRL * 0 PCT OHFC FUFL PILOT CONE PPM OPAC
TIM »A TA 2 EGH DIA. HATE INJ AMGlE NOX (SOOT)
-6
-6
-6
-6
-6
-6
-6
-6
-6
-6
-12
-9
-3
0
a
-6
-6
-6
-6
-6
-6
-6
-6
-6
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
30 100 0.00 0 0.000 7.0
30 100 0.00 0 0.000 6.0
30 100 0.00 0 0.000 5.0
30 100 0.00 0 0.000 7.0
30 100 0.00 0 0.000 5.0
30 100 0.00 0 0.000 7.0
30 100 0.00 0 0.000 7.0
30 100 0.00 10 0.000 7.0
30 100 0.00 20 0.000 7.0
JO 100 0.00 30 0.000 7.0
30 100 0.00 0 0.000 7.0
30 100 0.00 0 0.000 7.0
30 100 O.OU 0 0.000 7.0
30 100 0.00 0 0.000 7.0
30 100 0.00 0 0.000 7,0
30 100 0.00 0 O.OUO 5.0
30 100 0.00 0 0.000 5.0
30 100 0.00 10 0.000 5.0
30 100 0.00 20 0.000 5.0
30 100 0,00 30 0.000 5,0
30 100 0.00 0 0.000 7.0
30 100 0,00 0 0.000 5.0
30 100 0.00 0 0.000 7.0
30 100 0.00 0 0.000 5.0
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
n.oo
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
o.oo
0.00
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
300 31.9
280 27.7
290 16.0
360 11.0
320 Y.6
340 4.0
320 t»0
330 6.0
180 b.O
120 13.0
720 10.0
4BQ *>.0
290 4.0
450 2.0
510 1.5
370 2.2
370 4.0
360 4.0
230 4.5
170 6,5
430 1.5
430 1.8
150 3.5
100 3.0
F)MEP LH NOX/ OPACITY GNOX/HHP MR
IK LB FUEL I5FC (MULTI)
87.6
71.7
61.2
77.0
52,7
55.9
52.7
52.7
52.7
51.7
59.1
60.1
54.9
57.0
52.7
43.3
43.3
43.3
41.1
41.1
33.8
22.2
o.o
-7.4
9.04
U. 44
8.74
14.47
12.18
17.26
16.24
lb.75
9.14
6.09
36.55
2*. 36
14.72
22.84
25.89
18.78
18.78
1H.27
11.67
8.63
32.74
32,74
22.84
15.23
31.9
27.7
16.0
11.0
9.6
4.0
6.0
6.0
8.0
13.0
10.0
5.0
4.0
2.0
1.5
2.2
4.0
4.0
4.5
6,5
1.5
1.8
3.5
3.0
.406
.423
.423
.350
.364
.326
.317
.318
.318
.330
.313
.311
.330
.323
.3*0
.303
.291
.300
.324
.306
.282
.265
.819
.248
2.10
?.21
2.46
2.97
3.08
3.54
3.29
3.40
1.66
1.29
7.11
4.69
3.07
4.63
5.63
4.20
4,05
4.05
2.84
1.99
6.70
B.19
36.45
479.26
EN VR
AUTH CR
HPW PC1Q15
RPw Pcl9l5
RPW PC1915
RPW PC1915
RPW PC1915
RPW PC1915
RPw Prl9l5
RPw Pfl9l5
RPW PC1915
RPw Prl915
RPw Prl915
RPW PC1915
RPW PC1915
RP* PC1915
RPW PC1915
RPW PC1915
PLACE
8«- 4
88-15
86-14
88-13
86-15
88- 7
88- 9
68-. U
H8-10
88-11
U8-12
06-16
88- 6
87- 1
88- '5
87- 2
PHI
.'6
.76
.76
.60
.60
.45
.45
.45
.45
.45
.45
.45
.30
.30
.15
.15
RPM
15
1800
2100
1500
2100
1500
1500
1500
1500
1500
1500
2100
1500
2100
1500
2100
SwIRL
TIM
-0
-6
-6
-6
-6
-6
-12
-9
-3
0
3
-6
-6
-6
-6
-6 .
n
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
PA
JO
30
30
30
30
30
30
30
30
30
30
30
30
30
30
30
TA
1 0
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
H 0 PCT OHFC FUEL PILOT CONE PPM OPAC
2 ESH D.I A. RATE INJ ANGLE NOX (SOOT)
0.0
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0. 0
0^000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
o.uoo
0.000
0.000
0.000
0.000
?.o
6.0
5.0
7.0
5.0
7.0
7.0
7.0
7.0
7.0
7.0
5.0
7.0
5.0
7.0
5.0
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
12
2&0
250
240
240
230
230
400
320
210
250
290
250
330
250
170
120
46.0
34.0
30.0
44.0
30.0
24.0
28.0
20.0
20.0
9.0
2.0
12.0
2.0
2.5
2.0
UO
RMEP LB NOX/ OPACITY GNOX/8HP HR
IK LB FUtL ISFC (MULTI)
ft9.t>
60.1
47.5
59.1
42.2
47.5
49.6
50.6
45.4
42.2
36.9
31.6
32.7
17.9
1.1
-10.5
7.53
7.53
7.23
9.14
8.76
11.67
20.30
16.24
10.66
12.6V
14.72
12.69
25.13
19,03
25.89
18.27
46.0
34.0
30.0
44.0
30.0
24.0
28.0
20.0
20.0
9.0 .
2.0
12.0
2.0
2.5
2.0
1.0
,47J
.485
.479
.419
.423
.357
.346
.343
.370
.382
.409
.346
.290
.287
.277
.246
2.1-*
2.37
2.49
2.38
2.76
2.74
4.56
3.60
2.63
3.29
4.25
^ • f- -J
3.64
5.34
J 9 w -»
5.60
1 ? . 6S
i f, . ^* •/
-19.34
-------
to
EN VR PLACE RPM SWIRL M o PCT ORFC FUEL
AUTH CR PHI TIM PA TA 2 EGR D1A. RATE
RPW M.A.N. 91-lfl .76 1500 -21
«PW M.A.N. 9i- e .76 leso -24
HPW M.A.N. 91- 3 .76 2100 -21
RPW M.A.N. 91-11 .60 1500 -21
RPw M.A.N. 91- 4 .60 2100 -21
RPW M.A.N. 92- 2 .45 1500 -21
RPw M.A.N. 91-12 .45 1500 -21
RPW M.A.N. 91-14 .45 1500 -27
RPW M.A.N. 91-13 .45 1500 -24
RPw M.A.N. 92- 4 .45 1500 -18
RPW M.A.N. 92- 5 .45 1500 -15
RPW M.A.N. 91- 9 .45 1B50 -21
PP* M.A.N. 91- 5 .45 2100 -21
RPW M.A.N. 92- 6 .30 1500 -21
RPw M.A.N. 91- 6 .30 2100 -21
HP* M.A.N. 92- 7 .15 1500 -21
RPW M.A.N, 91- 1 .15 2100 -21
0
0
0
0
0
0
0
0
0
0
0
d
0
0
0
0
0
EN VR PLACE RPM SwIRL
AUTH CR PHI TIM
RPW L*NOVA l-b-14 ,/6 15 -20
RPW LANOVA 115-12 .76 I85o -20
RPW LANOVA us- 2 .76 2200 -20
RPW LANOVA 115-15 .60 1500 -20
RPW LANOVA 115- 7 .60 2200 -20
RPW LANOVA ii5-i« .45 isoo -20
RPW LANOVA 116- 1 .45 1500 -23
RPW LANOVA 116- 2 .45 isoo -26
RPw LANOVA 116- 3 ,45 1500 -17
RPW LANOVA 116- 4 .45 1500 -14
RPW LANOVA 115-13 .45 1850 -20
RPW LANOVA 115- a .45 2200 -20
RPW LANOVA 116- s .30 isoo -20
RPW LANOVA 115- 9 .30 2200 -20
,-.
0
0
0
0
0
0
0
0
0
0
0
0
0
JO 100 0.00
30 100 0.00
30 100 0*00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
H 0
PA TA 2
Jo i o o.o
3fl 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0.00
30 100 0*00
30 100 0.00
30 100 0.00
30 100 0.00
0 0*000 o.O
0 0.000 *.0
0 0*000 «,0
0 0,000 «,0
0 0.000 o.O
0 0*000 ».0
0 0.000 *.0
0 0.000 «.0
0 0.000 «.0
0 0.000 *,0
0 0.000 o.O
o o.ooo «.n
0 0.000 «.0
0 0.000 o.O
0 0,000 «,0
0 0.000 ».0
0 0.000 «,0
PCT ORFC FUEL
EGH CIA. PATE
00. 0 «,0
0 0.000 ».0
0 0.000 ».0
0 0.000 o.O
o o.ooo ».o
0 0.000 »,0
0 0.000 *.0
0 0.000 o.O
0 0.000 «,0
0 0.000 «.0
0 0*000 *.0
0 0.000 o.O
0 0.000 «,0
0 0,000 o.O
PILOT CONE PP" OPAC
INJ ANGLE NOX (500T)
0,00
0.00
0*00
0.00
0.00
0.00
0.00
o.oo
0.00
0.00
0.00
0.00
0.00
0,00
o.oo
o.oo
0.00
PILOT
0 1«>10
0 1300
0 1330
0 1270
0 1280
0 94Q
0 1080
0 1750
0 1500
. 0 730
0 560
0 1100
0 850
0 390
0 400
0 50
0 90
CONE PPM
INJ AMGLE NOX
o.oo
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0,00
0.00
0*00
0,00
0.00
0,00
0 .13^0
0 1100
0 910
0 1370
0 870
0 1020
0 1210
0 1140
0 850
0 580
0 920
0 700
0 545
0 305
3,0
6.2
4,5
2.3
2.1
1.2
1.2
3.6
2,0
1.6
1.6
1.*
1.4
1,4
.9
.8
.9
OPAC
(SOOT)
10.0
10.0
7.2
2.0
3.8
1.6
2.0
6.9
1.6
2.6
2.0
3.2
1.4
1.6
0MEP
113.0
106.8
100,1
92.1
85,7
67,1
68.9
68.9
68,9
65.3
64.2
65,5
61.4
34, b
33.5
-10.3
.6.4
BMEP
92. /
95,9
93.8
72.5
70.4
49,6
48.5
4b,8
50.1
51.2
50.1
49,6
?4.5
22,4
UB NOX/ OPACITY GNOX/BHP HR
IK Lb FUt|_ 15FC (MULTI)
39!l8
40,08
4ft. 35
48.73
47.71
54,82
88.33
76,14
37.05
28,42
55,t53
43.14
29.69
30.46
7.61
13.70
LB NOX/
IK LB FUEL
40.69
33, iS
27.43
52.15
33.12
51.77
61.42
57.86
43.14
29.44
46.70
35.53
41,49
23.22
S.'' ,326
6.2 .335
4.5 ,340
2.3
2,1
1.2
1.2
3.6
2,0
1*6
1.6
1.4
1.4
1.4
.9
.8 1
.9 2
OPACITY
10.0
10.0
7.2
2.0
3.8
1.6
2.0
6.9
1.6
2.6
2«0
3.2
1.4
1,6
.310
,316
.310
,297
.297
.296
,309
.314
,293
.304
,316
.292
.149
.482
8,80
7.54
8,18
8.60
9,61
0,16
10.03
16.26
13.86
7.1b
5.61
10,62
9.07
7.28
7,89
-5.49
-fel.60
GNOX/Ry--1 HR
ISFC
.356
.353
.443
.347
.328
.351
,358
.371
,349
,345
.334
.319
.365
.314
(MULTI)
8.31
6.87
7.47
11.01
7.28
12.34
15.03
14,98
10.17
6.83
11.06
8.60
13,78
8.26
-------
APPENDIX D
COMPILATION OF PUBLISHED DIESEL EMISSIONS DATA
Eighteen publications reporting tests of 51 engines have been compiled
in order to determine what emissions behavior is representative of a wide class
of engines—both trends (say, in NO vs. EGR), and variations from those trends
for specific engines. In this way our single-cylinder test data could be placed
in context. Only operational changes were tested for 37 of the 51 engines.
Abbreviated information about the 14 more extensively tested engines (compression
ratio, displacement, etc.) is listed in Table D-l, along with an indication of
which of the parameters were varied.
Table D-l
PUBLISHED EMISSIONS DATA
*
0
c
1j>
First Author(year) £
Pischinger (72) 3
" 4
Parker (72) 5
Ba scorn (71) 7
Shahe-d (73) 8
Hames (71) 9
Marshall (71) 10
Khan-ClSl (71) 12
AbthoS (69) 17
21
Walder (73) 27
Khan-C142 (71) 29
Landen (63) 32
McConnell (63) 34
Bosecker (71) 47
. 48
Valdmanls (70) 51
UNITS
Comp.
Type Ratio Nwl V^
DI . 6 120
DI 16.2 1 83
DI 6 67
DI
DI 12.3 1 130
DI 6 "7IN'
DI 6 112
DI 16 1 62
DI 20 1 33
MAN 17 4 73
DI 6 143
DI 16 63
PC 17.5 1
DI
PC
_.__ In3
Load Speed Timing
50-55 14
A/F
* 18-26 29-9
16-100 18-4
A/F
40-140 15-21 20-28
BMEP
20-200 12-21 10-16
BHP
19-100 12-22 30-22
A/F
.24-T.84 30-+5
9
A/F
17-70 10-26 41-18
VF
30-75 11-27 tS-0
A/F
12-24 S.5-0
.3-1.8 8-20 15-+8
9
BHP
23-38 16-22
A/F
as 100 °CA
shown RPM BTDC
Ta EGR H20 Pa
0-20
50-265 0-15 0-1.5 (TC)
0-10 0-1.0
90-250 . 1-3
0-20 (TC)
80-240 0-25 (TC)
°F . % per fuel atm
Swirl d. Rate
Low (fumig)
High
Mecj
Med-Hi (fumig)
(ho effect)
10-13
High
2.4-4.8
— —
.001" mm3/°CA
*13-mode
143
-------
In order to compare data from separate studies it was necessary
to convert reported data into common units. For example, EGR percentage
was converted to initial oxygen mole fraction by means of the expression:
X = .21 ll -
-------
Engine Code Timing
Symbol Author (Table D-l) <£ (°BTDC)
o
•
*
a
•
-------
Figure D-2
EFFECT OF AIR DENSITY
(Previous Studies)
Engine Code
(Table D-l)
Author
Load
RPM
4 Pischinger = .65 2600
17 Abthoff 3.6kp/cm2 1700
32 Landen = .40 2400
Note: A/F constant
S 2
E
-------
Figure D-4
EFFECT OF WATER INJECTION
(Previous Studies)
§
i-i
4->
E
0)
o
c
o
U
O
Engine Code
(Table D-l)
Author
Load RPM
10
21
27
47
51
Marshall(1971) 220HP 2100
Abthoff (1969) 100% 2000
Walder (1973) 550% Carb
Bosecker(1971) 75% 1600
Valdmanis(1970) ~
51
—— Intake
_ _ _ Emulsion
27
Water/Fuel
Figure D-5
EFFECT OF EGR
(Previous Studies)
1.0
O
i
0.5
^
Author
3
5
7
8
10
27
34
O Pischinger (1972)
D Parker (1972)
O Bascom (1971)
A Shahed (1971)
0 Marshall (1971)
Q Walder (1973)
^ McConnell (1963)
O This study (1500
Q This study
(2100 RPM)
1.0
0.5
Calculated relative rate of
NO formation (see page 17)
15 .16 .17 .18 .19
XQ (Start of Injection)'
2 147
.20
.21
-------
Figure D-6
EFFECT OF TIMING
(Previous Studies - DI Engines)
o
E
O
§
O
(0
•—H
(D
S-
O
Engine Code
(Table D-l)
Pischinger (1972) '= .69
Bascom (1971)
Hames (1971)
Marshall(1971)
Khan (1971)
Abthoff (1969)
= .45
200 HP
A/F = 20
= .72
2600
2100
2100
2000
2000
7.3 kp/cm
McConnell(1963)
Start of Injection ( BTDC)
148
-------
APPENDIX E
EQUILIBRIUM ANALYSIS OF DIFFUSION FLAME STRUCTURE
An equilibrium diffusion flame model was developed in which it is
assumed that all of the reactions are in equilibrium with the exception of the
NO -formation reactions. This assumption may be justifiable since it is the
X.
nature of a diffusion flame that the reaction rates are fast compared to the
transport rates (the reactions relax faster than the diffusion field can change) .
Moreover, and perhaps more importantly, there is some experimental evidence
which supports this notion*.
Governing Equations
The equations governing the equilibrium diffusion are considerably
simpler than those for finite-rate kinetics and are amenable to closed form
solution or solution in terms of known tabulated functions for many geometrical
configurations. The equations are most conveniently solved when expressed
in terms of the enthalpy and the element mass fractions, defined as follows:
~ M,
Y = £ ft _J. Y.
1 ij Ui i
If it is assumed that (1) all diffusion pairs have equal diffusion coeffi-
cients and obey Pick's Law, (2) unit Lewis number, and (3) spherical symmetry,
then the governing equations take the form
= V (PDVY.)
p Jt = V
*A. D. Tuteja and H. K. Newhall, "Nitric Oxide Formation in Laminar
Diffusion Flames," Emission from Continuous Combustors (1972).
149
-------
where .# is the total derivative including convective transport.
These equations are readily solved for many cases. For droplet
combustion (assuming quasi-steady conditions and pD = constant), we
have the boundary conditions
-Y
where Y, and h_ are the element mass fractions and enthalpy of the injected
fuel. The solutions take the form
Y = y + (Y - Y ) e"Pe///77
h = h_ + (ho, - h J e'
where T\ = r/r and Pe' is the usual ratio of convective to diffusive transport.
Once the element mass fractions Y. and enthalpy and h have been
solved as a function of radius, the actual species distribution can be found
from an equilibrium analysis. Equilibrium composition through the diffusion
flame is accomplished with the aid of the NASA One-Dimensional Equilibrium
(ODE) Program. The specified input for this program is the molecular structure
(obtained from the element mass fractions) plus two thermodynamic state
variables (here taken as enthalpy and pressure).
Sample Calculations
Sample calculations shown in Figure El are for methane gas injected
into ambient air at 300°K. The Pe'cle't number was taken as Pe' = .51. The
Burke-Schumann (flame sheet) model calculations were also obtained from the
ODE Program by suppressing all species not present in the global reaction, viz.,
+202
150
-------
Figure El
EQUILIBRIUM DIFFUSION FLAME-SPECIES MOLE FRACTIONS
Mole
Fraction
.001
-, 3000
(Flame Sheet Model)
T
,0001
7 8 9 10 11 12 13 14 15
Dimensionless Distance f} = r/r
151
w
-------
Por the equilibrium calculations the special variation enters as a
parameter only. Thus the computation must be performed at many locations
through the flame zone to determine the profiles.
Notes Toward a Flame-Structure Model with Finite-Rate Chemistry
It may be necessary to go beyond the equilibrium chemistry approach
of the previous section and consider finite rate reactions. If this becomes
necessary, an asymptotic approach can be taken as follows. We neglect the
bulk convection and nonsteady effects and consider only the balance between
. the molecular-diffusion transport term and the chemical source term. This
assumption is justified because the governing equations simplify when atten-
tion is focused on the flame zone. (See Table El which suggests how the
asymptotic theory would be.developed). In general, the general
conservation equations consist of the following four terms: (1) the unsteady
term arising due to temporal changes in the field, (2) a convective term
representing the transport of heat or mass to a point in the fluid by the mean
fluid motion, (3) a diffusion term representing the transport of heat or mass
due to gradients in the flow, and finally (4) a chemical reaction term repre-
senting the creation or destruction of species and heat by chemical kinetics.
Of these four terms, only the last two are of significance if one confines
attention to the flame zone itself. It can be rigorously demonstrated that
in diffusion flames with large DamkShler numbers*, the flame zone is suffi-
ciently thin that the flame responds in a quasi-steady manner to temporal
changes and that convective transport is much less effective than diffusive
transport. Thus the local diffusion flame problem can be treated as if it
consists of a balance between the diffusion term and the chemical reaction
term. Moreover, and perhaps more importantly, is the fact that this flame
balance is independent of the geometric configuration. That is to say, it is
*The Darnko'hler number is the ratio of the characteristic mixing time to a
characteristic chemical reaction time.
152
-------
a universal approximation and will apply equally to burning droplets , sprays,
jets, or gaseous counter-flow. Furthermore, the model will be basically
iterative because the chemical production in a given subzone will sensitively
depend on the flow of species into it from other regions.
Table El
Physical Motivation for Asymptotic Expansions
Flame Viewed from a Distance
(Outer Region)
Droplet
Flame
Range of Influence of Diffusion
In the zones on either side of the
flame one or other of the reactants
is absent. Hence there is no
reaction, and we have an
unsteady-diffusion balance.
The flame zone appears to be a
surface where the reactants
disappear. We cannot satisfy
all the boundary conditions, and
must match with an inner solution.
Flame Viewed Close-up
(Inner Region)
FUEL
OXIDIZER
INTERMEDIATES
AND RADICALS
The region of interest is sufficiently
small that relaxation is rapid, thus
the flame zone is quasi-steady and
we have a diffusion-reaction balance.
The droplet and the range of influence
seem far away. We cannot satisfy
boundary conditions so we must
"match" with the outer solution.
153
-------
As suggested earlier, the diffusion flame is amenable to a universal
treatment. It is intended for our model to satisfy the following criteria,
which are put forward both as a goal and as a guide as to what to include
in the model:
(1) The model should predict temperature and reactant
concentration profiles through the flame zone.
(2) It should predict the O-atom distribution and determine
for once and for all if this is in equilibrium.
(3) It should include dissociation for this lowers the
temperature and thus affects NO production rates.
X
(4) It should include sufficiently complex C-H-O kinetics
to accurately predict the reaction intermediates of
interest.
154
-------
APPENDIX F
UNSTEADY DIFFUSION AS A FACTOR IN DROPLET COMBUSTION
Background
Some recent evidence, both experimental and theoretical, has
suggested the possibility that droplets deviate from the classical quasi-
steady behavior during combustion,, The classical theories assume that the
time for the diffusion field to establish itself is small compared to the droplet
lifetime, so that at any instant of time the diffusion field corresponds to the
steady-state associated with the instantaneous droplet size*. Strictly speaking,
the relaxation time for diffusion only vanishes in the limit of zero gas density.
Kirkaldy (1958) has pointed out the inherent unsteadiness of spherically
symmetric diffusion problems (owing to the finite rate of diffusion).
Nevertheless, the quasi-steady theories persist, probably because of the
2
excellent predictions of burning rate (notably the so-called "d -law"). More
recently, however, investigators have looked not just at overall burning
rate but at flame radius, and observations seem to indicate that under some
conditions burning occurs in a compound-unsteady fashion (flame and drop radii
changing independently). Such observations of unsteadiness have been
made by Nuruzzaman and Beer (1971) and Krier and Wronkiewicz (1972),
although experimental resolution was not precise enough to rule out alterna-
tive explanations for unsteadiness such as droplet heat-up or natural
convection. Theoretical studies of droplet combustion such as the exact
numerical solutions of Kotake and Okayaki (1969) and the approximate
solutions of Chervinsky (1969) have treated the unsteadiness due to droplet
2
heat-up and diffusion. It is interesting to note that the d -law would be
expected theoretically to hold true even under unsteady conditions, thus
*There are other relaxation times of importance in droplet combustion. These
include the delay to spontaneous ignition, chemical relaxation time, and
liquid-phase heat-up time. In the classical theories, as in the present
case, these are assumed to be small compared to both the droplet lifetime
and the diffusion relaxation time0
155
-------
2
d -law measurements should not be taken as evidence for the quasi-steady
theory.
The.purpose of the present study is to determine the conditions
under which the quasi-steady assumption breaks down and how unsteady
diffusion affects flame temperature and NO formation. Thus we seek to
JC
develop a theory for unsteady droplet evaporation and combustion. The
present effort will focus attention on the temporal variations of the flame
position and temperature by adopting the flame sheet approximation. Thus
the present study assumes that finite rate kinetic effects are confined to the
immediate flame zone and do not affect the fuel and oxidizer mantles to any
appreciable degree. A more realistic analysis of the structure of the flame
zone itself is treated in Appendix E.
Governing Equations and Boundary Conditions
The present treatment will be limited to spherically symmetric
evaporating or burning droplets in a constant pressure environment. It is
further assumed that the diffusion velocity is given by Pick's law, Lewis
number is unity, and that we can define a mean specific heat and a mean
molecular weight for the mixture. The chemical rate processes are taken to
be restricted to an interfacial flame surface and thus no rate terms appear in
the field equations. Also, following Williams' (1965) Schvab-Zeldovich
formulation, body forces, Soret and Dufour effects, and radiation are
neglected.
In addition, liquid-phase heat conduction is neglected. While the
influence of this phenomenon on droplet burning is not clear at the present
time its inclusion is beyond the scope of the present effort*. In the present
analysis the droplet has a constant and uniform internal temperature.
Physically this means that all of the heat supplied from the gas-phase is
*Those who are interested in pursuing the problem may examine the numerical
solutions of Wise and Ablow (1957) or the analytical solution of Waldman
(1971), both of which are based on a surface regression rate corresponding
to the d2-law and constant surface temperature. The analytical solutions of
Waldman were recently extended by Sonalkar (1972) to include arbitrary
surface temperature histories.
156
-------
used to evaporate additional fuel to sustain the combustion. Under these
conditions we have the following set of equations:
Continuity:
Transport:
3S (2)
T
subject to the boundary condition at the droplet surface:
+Pv (S-S_) = 0 at ri=Tis(t) . (3)
The initial conditions and boundary conditions at infinity are given by
S (r>r ,0) = S (oo,t) - S^ (4)
S
Let us define S. Since attention will be restricted here to the case of infinite
Damkohler number, i.e., a flame sheet, it is appropriate to introduce a general
Schvab-Zeldovich variable S which represents linear combination of the non-
dimensionalized temperature and species variables:
S = T + Y.
i
These variables then satisfy homogeneous equations. In the simple case of
pure evaporation the quantity S may represent the mass fraction or temperature
itself, of course. T and Y. represent dimensionless temperature and reduced
mass fraction, following Waldman (1968):
T = C T/Q
Y. =Y./a. where a =
i M
157
-------
In expression (3), S_ = Yi_-T_, where
- Q Cp s .
The other nondimensionalized variables are as follows:
TI = r/rr
P = P/P»
(6)
v = P08rrv/PD
T =
where r is an (as yet unspecified) reference radial position. It is also
assumed from here on that the product pD is constant.
Analysis by Matched Asymptotic Expansions
By uncoupling the droplet regression rate from the gaseous diffusion
rate, quasi-steady theories may fail to adequately account for certain unsteady
effects, notably the relative motion of the flame and the droplet surface. We
suggest here that the unsteady effects are likely to be important far from the
droplet, and that the qua si- steady theories confine their attention to the
droplet vicinity, and thus may fail to "see" these effects. Thus it is sug-
gested that the droplet be viewed from afar as well as close up, and so analysis
by matched asymptotic expansions seems apropos. The physical motivation for
asymptotic expansions is demonstrated in Table Fl.
Having proposed this notion, it is incumbent upon us to select an
expansion parameter which will distinguish the two regions . The above notion
suggests that the droplet radius is small compared to the diffusion field of
interest. We can define the characteristic diffusion radius, r from diffusion
158
-------
theory, r «
, where t is the characteristic time of the problem (here
taken as the burning time) . The characteristic burning time can be obtained
2 22
from the d -law (i . e . , r = r - 0 1) by setting r = 0 . It then follows that
s so s
Table Fl
Physical Motivation for Asymptotic Expansions
Droplet Viewed from a Distance
(Outer Region)
Droplet
Flame
(Assumed to
be in outer
region)
Range of Influence of Diffusion
The droplet appears to be a point
source of mass and sink for heat.
We cannot satisfy the boundary
conditions at the surface, so we
must "match" with the inner
solution.
Convective effects are negligible/-
this leads to an unsteady-
diffusion balance.
Droplet Viewed from Close-up
(Inner Region)
Fuel
On the scale of the droplet, the
influence of the diffusion (and
perhaps of the flame) is not felt
because it acts on a larger scale.
We cannot satisfy the boundary
conditions at infinity so we must
"match" with the outer solution.
Relaxation is rapid, the system is
quasi-steady; this leads to a
convection-diffusion balance.
159
-------
Typical values for the evaporation constant and the diffusion coeffi-
****
parameter
-3 2 —1 2
cient (p « 10 cm /sec,'D^ « 10' cm /sec) suggest taking the expansion
With 6 so defined, the surface boundary conditions are to be applied at
77 = (5£(t) [where £(t) = r (t)/r ]. In the limit 6—>0, the boundary conditions
' ' S ' SO
cannot be applied because the droplet literally vanishes. Recourse must be
made to matched asymptotic expansions in order to observe the details of the
inner region. .:
The analysis proceeds formally as follows: The lowest order
solution must be dominated by the outer region, for which we postulate
y = *7 ~ o£ and for which we postulate the velocity to be small
( v = v = O(<5)), so that the primary outer region balance is between
unsteadiness and diffusion. The first-order outer equations lead to the trivial
solutions
So = S- Po= -1' vo = 0 • - • ?
We now turn our attention to the inner region where the following
expansion-scheme is postulated.
Inner Region: 1st Order
z =11/5
S. = S .= S + 0(6)
inner o
p., ' = S = o + O(<5)
inner ' ' o .
1 ". v. = v.= d~l v . +0(1)
inner -1
The inner region equations are determined by substituting the above
in Eqs. (l)-(3) and taking the limit 6 -» 0, from which we obtain the inner
160
-------
region solutions:
z2 DQ v_x = C(t) = Pe f (t)
and S = S_ + (S
o s
, (Pvr2)
where Pe =
(PDr)s
is the Pe'cle't number (ratio of convection to diffusion at the droplet surface
and is determined by matching with the outer solution. The boundary condition
for z -» °° is obtained later by matching with the outer solution.
Notice that the lowest order inner solutions are independent of
time, thus showing that on a scale of the droplet radius the quasi-steady
approximation is valid. Note, however, that in eliminating the time
derivative we cannot apply an arbitrary initial condition. In other words, we
cannot account for the ignition phenomenon; this is consistent with the flame
sheet approximation. Physically, the qua si-steadiness of the inner region
means that the zone of interest is sufficiently small that timewise disturbances
relax quite rapidly. This result agrees with the conclusion previously reached
by Williams (1960) that unsteady effects have only higher order effects on the
burning rate.
Following the formalism suggested by Van Dyke (1964) for matching
and generation of the next order outer functions, we express the 1-term inner
solution in outer variables, expand for small 6 , and truncate terms of order
greater than 6 , to obtain the following 2-term outer expansion:
p' (S-S)ePe'p
-------
Matching the lowest order term of Eq. (8) with Eq. (7) gives
Pe =
: = An(l+B)
or m = 4rr pDr 4n (1 + B)
o
which is the familiar result associated with the quasi-steady theories.
Notice that the above expansion (Eq. (8)) also shows that the inner region
induces additional perturbations in the outer region, and these will be
unsteady as we shall see below. Formally, Eq. (8) suggests the following
expansion in the outer region:
~ ~ C3>
S = Sw + 6Sl +0(6*)
+0(0
as,
162
-------
and is to be determined from the matching as follows:
We expressed the 2-term outer solution in inner variables, expanded
for small 5 and truncated terms of order greater than 6 to obtain the 2-term outer
solution expressed in outer variables:
Matching with Eq. (8) gives
f(T) = (S -S )ePe'pef (12)
s
Returning now to the velocity equation (9) the solution can be
expressed as
Since P-, can be determined from Eqs. (11) and (12) the quadrature
can be completed. It is not in the present interest to do this, however,
as we will not be seeking higher order inner perturbations. It is clear that
such perturbations do exist and will be of increasing importance as e becomes
larger (e.g. , for more rapidly burning droplets) .
In order to examine the temporal history of the flame radius and flame
temperature, it is necessary to construct the composite solution. This is so
because the flame does not necessarily reside in either the inner or outer
regions during its entire lifetime, though it may in some cases. Again
following Van Dyke (1964) we can construct an additive composite solution.
Thus, from Eqs. (7), (10), and (11) we obtain
S =S.+(S.-S
c s
(13)
erfc
163
-------
Equation (13) is readily expressed in terms of the original variables
and the mass burning rate. Noting that
m
so
we obtain
(a _ c \ r—r ft}
At>oo & ) r r r VW
S_ ri J JL. lj.\ JL t ~\ _..j:_ O
~ |J_
4nPD[r-rs(t)] V" w l" vu'
(14)
^^df - 4noD[r-rs(tj[exp (^J - l]} .
0
It is worthwhile to note that this result is independent of the reference
radius r . This is consistent with the notion that diffusion in an infinite
domain is characterized by an infinite relaxation time. Notice also that
Eq. (14) applies both to actual droplets (evaporating and burning) and to
porous spheres. For porous spheres m (t) and r (t) are constants and the
equation simplifies somewhat.
Results and Discussion
Equation (14) can be used to determine the flame radius and flame
temperature. To determine the former, choose S = Yp - Y and set S = 0.
X" O5C
(This assures that the reactants meet and react in stoichiometric proportions.)
Thus, the flame radius, is given implicitly by the equation
(Y* +-Y' ) ;- r,(t)-r (t)
- — — * Ti(t) -m (o) erfc -^ s—
•ox^ 4^oD[rf(t) -rs(t)] I
(15)
P D[rf(t) -
0
where r (t) and m(t) are given by the quasi-steady solution in the inner region
o
and hence are known.
164
-------
Likewise, the flame temperature is determined by letting S = (T +Y ) or
O JC
(f + Y_) and evaluating Eq. (21) at r = rf(t).
f Y + Y°° Y + T Y°°
ff= T"YF-+YoxYF- T-Yox ^
Equation (16) shows that flame temperature is constant in spite of the
motion of the flame. This result is surprising in the light of the exact
numerical solutions of Kotake and Okayaki (1969) which exhibit temporally
varying flame temperature. We do not presently understand this discrepancy,
perhaps it is attributable to the neglect of liquid-phase heat conduction in
the present case.
As pointed out previously, Eq. (14) is the principle result of this analysis,
being the first-order asymptotic solution to the unsteady droplet combustion
problem. Equation (15), which gives an implicit solution for the flame radius,
was programmed for solution on a computer. Figure Fl shows some typical
results of the computation. It is seen that for the assumed initial conditions
[T(r > r ) = T^, Y(r > r ) = YM] the flame radius is always increasing relative
s s
to the droplet radius (by contrast with the quasi-steady theory which predicts
r,/r = constant). It is also seen that the flame radius itself first increases
f s
then decreases in the latter stages of combustion.
Figure F2 confines attention to the flame-to-droplet radius over the enve-
lope of validity. The theory is not valid at very early stages of the burning
because it ignores the ignition process. Moreover, in the very last stages of
the burning most of the mass is already consumed and the exact flame behavior
is inconsequential. Thus, Figure F2 shows that over this envelope of validity
the flame-to-droplet radius increases from 3 to 6. For comparison, the given
conditions for quasi-steady theory predicts a flame-to-droplet radius
*Admittedly not a realistic simulation of the ignition or post-ignition conditions.
165
-------
Figure Fl.
TYPICAL RESULTS FROM THE UNSTEADY DIFFUSION THEORY
0.2
O.I
0.2 0.3 "0.4 0.5 0.6 0.7 ~ 0.8 0.9
1/2 mass -1/2 radius
T (Dimensionless Time)
1.0
Figure F2.
FLAME ENVELOPE EXPANDS RELATIVE TO DROPLET SURFACE
,_ .-(BOTH DECREASE DURING BURNING)
Conditions:
CIOH20
Air
I500°K, lOOatm
= 0.3 cm2/sec
! x I0~3cm2/sec
166
-------
„
somewhat in excess of the mean value obtained by the unsteady theory. Thus/
the unsteady theory seems to be more appropriate than the quasi-steady theory.
It is also interesting to note that the parameter d, which may be rewritten
as
a_\oprf» A oo
6 - 2Pe =
is seen to be proportional to the gas-to-liquid-phase density ratio. Owing to
the high pressure environment found in diesel combustion chambers it may yet
turn out that unsteady effects are minimized in diesel engines. This point
requires further examination before any definite conclusions can be reached .
167
-------
SUPPLEMENTARY REFERENCES FOR APPENDIX F
(See also main list, p. 107).
Chervinsky, A. , "Transient Burning of Spherical Symmetric Fuel Droplets,"
Israel Journal of Technology 7_, 35 (1969).
Kirkaldy, J- S. , -"The Time Dependent Diffusion Theory for Condensation on
Spherical and Plane Surfaces, " Canadian Journal of Physics 36, 446
(1958). ,.
Krier, H. and Wronkiewicz, J. A. , "Combustion of Single Drops of Fuel,"
Combustion and Flame IjL, 159(1972).
Nuruzzaman, A.S.M. and Beer, J. M., "On the Non-Steady State Nature
of Droplet Combustion," Combustion Science and Technology _3, 17
(1971).
Sonalkar, R., "Examination of'the Fuel Droplet Combustion Problem with
Variable Surface Temperature, " M.S. Thesis, Rensselaer Polytechnic
' Institute, Troy, New York (1972). •..'•.
Tuteja, A. D. and H. K. Newha 11, "Nitric Oxide Formation in Laminar Diffusion
Flames," Emission from Continuous Combustors (1972).
Van Dyke, M., Perturbation Methods in Fluid Mechanics, Academic Press,
New York (1964).
Waldman, C. H., "Theoretical Studies of Diffusion Flame Structures," Ph.D.
Thesis, Princeton University (1968).
Waldman, C. H., "Heat Conduction Within a Burning Droplet," unpublished
notes (1971).
Williams, F. A. , "On the Assumptions Underlying Droplet Vaporization and
Combustion Theories," J. Chemical Physics 33, 133 (1960).
Wise, H. andAblow, C. M., "Burning of a Liquid Droplet. III. Conductive
Heat Transfer within the Condensed Phase during Combustion,"
J. Chemical Physics 2_7, 389 (1957).
168
-------
APPENDIX G
EXPERIENCE WITH WINDOWS FOR
DIESEL COMBUSTION CHAMBERS
Optical access to a fired diesel combustion chamber was needed for
high speed photography and spectroscopic species measurements. Windows
were developed and used with combustion chambers of both the direct-
injection and prechamber type. Before outlining the experience of window
development, it will be useful to recall the design objectives or potential
problem areas:
1. Optical Access: A line-of-sight configuration (two
opposing windows) for the prechamber, and a single
window directly above the piston for the direct-
injected head. Windows must pass all light of X >
2100 A. Also, for photography, windows of about
1 to 2" diameter (as large as feasible) to maximize
the field of view.
2. Mechanical Properties: Sufficient tensile strength
to withstand up to 100 atm peak chamber pressure.
3. Provision to prevent rapid soot accumulation.
4. Window installation must be free of leakage yet
allow windows to be readily replaced.
5. Sufficient cooling must be considered.
Design objectives 1 and 2 were met by selecting fused quartz windows, a
pair of 1.5" diameter x 3/8" thick for the prechamber and a single window
2.0" diameter x 1/2" thick for the direct-injection head.
The transmittance and relevant physical properties of type 125 quartz
are given below in Figure G-l and Table G-l, respectively.
169
-------
Figure G-l
TRANSMITTANCE CURVE- 125 QUARTZ
FOR I CM. THICKNESS
(EXCLUDING SURFACE REFLECTION LOSSES*
20 22 .24 .26 .28 .30 1.0 2.0 2.5 3.O 3.5 4.0 4.5
Table G-l
Physical Properties of Type 125 Quartz
Property
English and Metric
System Value
Density
Hardness
Tensile Strength
Compressive Strength
Bulk Modulus
Rigidity Modulus
Young's Modulus
Poisson^s Ratio
Coefficient of Thermal
Expansion
Thermal Conductivity
Specific Heat
Fusion Temperature
Softening Point
Index of Refraction
2.2 gm/cm
4.9 Mohs1 Scale
7,000 psi
>160,000 psi
5.3 x 106 psi
4.5 x 106 osi
10.4 x lO^psi
.16
5.5 x 10~7 cm/cm-C
(20°C-320°C)
3.3x10 gm cal-cm/
cm -sec-°C
. 18 gm cal/gm
1800°C
1670°C
1.4585
170
-------
The following calculations support the notion that the windows
would not break under combustion pressures.
For a disk of radius r(in) and tensile strength S(psi), the
thickness t(in) to withstand pressure P(psi) is given by
the expression
For S = 7000 psi and P = 1500 psi, we derive r/t « 2.2.
With a 40% safety factor (i.e. assume peak pressure
2100 psi), the required diameter-to-thickness ratio is
d/t
= 4
Window designs to allow spectroscopy without smoke observation
were evaluated following very helpful communication from Bill Brown of
Caterpillar and Dr. P. Flynn of Cummins. Options considered include:
1. Conventional window reportedly usable for only five
to ten power strokes. Either inject fuel intermittently
or remove and clean windows intermittently.
2. Continuously flushed window of reduced size (1/4"
to 1/2"); six to twelve small jets provided to flush
across the window surface. Based on a review of the
flow rates and back pressures used by previous inves-
tigators, flushing appears feasible provided the back
pressure exceeds twice the peak cylinder pressure.
This technique has been used successfully by Ebersole
et al. [SAE Paper 701C (1963)], on a smaller window
by Dr. Flynn in prechamber engines, and by Dr. Landen
who used a 1/8" diameter passage, 1/2" long. Pre-
liminary calculations show that four holes of 300)U
diameter can be continuously purged at sonic flow
without adding more than 10% to the cylinder contents.
Mr. Brown suggests that this approach only works when
the unmixed potential core of the jet completely covers
the window. This requires a continuous wall jet com-
pletely around the circumference of the window with a
height of .08" for the 1/4" window. The momentum of
the jet must be large compared to the momentum of the
eddies in the cylinder.
171
-------
3.. Bum off smoke as it accumulates by heating the
window, focused radiation, or other means. At
Caterpillar, a plate with a temperature gradient
was inserted in the gas stream from a combustor.
The hot end of the plate did not collect soot, but
the cold end did. The boundary of the soot was
at a temperature of 700 to 800°F. This data fits
Mr. Brown's experience on diesel piston crowns.
To burn off smoke, there is some experience indi-
cating 1000 to 3000 watt/in2 is necessary.
4. Massive blast of air at BDC on the intake stroke;
only the smoke accumulated in one cycle need be
removed. Although smoke buildup would occur
eventually, it may be considerably postponed in
this manner.
Two configurations were selected based on a naturally heated window
and a window with continuous flushing. Preliminary calculations given below
show that the center of a 2" x 1/2" thick window will rather quickly reach a
steady state temperature which is too hot for soot accumulation (T > 1000 K) .
To simplify the analysis, we assume that a steady heat flux of 13
9 *
watt/cm [taken from Le Feuvre et al. (1969)] is applied to the surface of
a semi-infinite quartz body. The temperature gradient within the glass will
be on the. order of:
AZ ~ k
2 —3 o
Substituting q = 13 watt/cm and k = 3.3 x 10 cal/cm-sec K we obtain
AT/AZ ~ 1000°K/cm. The actual temperature profiles are non-linear; thus
the temperature drop across a 1/2" window would be more than 1000 K. It
can be concluded that the exposed surface of the window will exceed the
soot bum-up temperature. The heat loss in the radial direction to the cyl-
inder head does not alter this conclusion as long as the window diameter-
to-thickness ratio is large (d/t = 4 in this case) .
*Le Feuvre, T., Myers, P. S. and Uyehara, O. A., "Experimental Instanta-
neous Heat Fluxes in a Diesel Engine and Their Correlation," SAE Paper
No. 70 1C, SAE Trans. 78, 1969.
172
-------
Although a flu shed-window (slotted gasket) was also built, the.
uncooled windows worked well enough to obviate the need for the more
complex flushed system. The precup windows become prohibitively opaque
after about 20 to 30 seconds of running at full load and 1500 rpm. How-
ever, the windows clean up to an adequate transparency (approximately 70%)
after 20 to 30 seconds of motoring. Apparently this is due to oxidation of
the freshly deposited soot by hot compressed air. The larger the precup
volume, the cleaner the windows can be maintained.
After the soot problem was solved, the spectroscopic measurements
were delayed by two window problems: cracking and poor sealing. Quartz
window cracking had occurred because of precup deformation when hold^
down bolts were tightened; an extra .010" clearance solved the problem.,
Sealing was accomplished when a trial-and-error series of tests led to the
use of a soft brass and asbestos sandwich-gasket.
The final configurations are shown below in Figure G-2.
173
-------
Prechamber
Bolt
Hole
Injector
Hole
Ccpper and
Assestos-
Caskets
Window—,
\
N
-X
X
PISTON
^, High Pressure
rr Gas
Slotted Gasket
With Reservoir
(a) Window for Prechamber Engine
(b) Window for Direct Injection
Engine
(c) Schematic for Optional Flushed
Window
Figure G-2
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-460/3-74-002a
2.
3. RECIPIENT'S ACCESSION>NO.
4. TITLE AND SUBTITLE
Foundation for Modeling NO and Smoke Formation
in Diesel Flames
5. REPORT DATE
Issued January 1974
6. PERFORMING ORGANIZATION CODE
0286
7. AUTHOR(S)
R. P. Wilson, Jr., C. H. Waldman, L. J. Muzio
8. PERFORMING ORGANIZATION REPORT NO
^PERFORMING ORG ANIZATION NAME AND ADDRESS
Ultra systems, Inc.
2400 Michelson Drive
Irvine, California 92664
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68 02 0222
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
2565 Plymouth Road
Ann Arbor, MI 48105
13. TYPE OF REPORT AND PERIOD COVERED
Phase I (1 July 72-30 June 7
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
Co-sponsor: Coordinating Research Council, 30 Rockefeller Plaza, New York,
New York, 10020 under APRAC Project CAPE 20-17
16. ABSTRACT
A mathematical model of diesel combustion with NOX formation and smoke is
sought to guide the development and design of engines. A foundation for a model was
established in Phase I with the following four activities: (1) Single-cylinder emissioi
data was generated; NOX and soot were affected 40% or more by seven parameters:
divided chamber, prechamber volume ratio, compression ratio, EGR, water injection,
fuel orifice size, and air swirl; (2) Three existing models were critically reviewed
based on treatment of physical heat release mechanisms, ability to predict emissions
behavior, and the need to readjust empirical coefficients; (3) A mechanistic heat
release model was outlined with treatments of the macroscale mixing (air swirl and
fuel spray) and the moledular mixing (diffusion flame profiles); (4) Measurements of
air motion, fuel dispersion, temperature, and NO in the diesel combustion environ-
ment were designed in order to resolve key questions about mechanisms.
(56 references)
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. cos AT I Field/Group
jjiffusion Flames
Air Pollution
Diesel Engines
Combustion
Emission
Nitric Oxide (NO)
Nitrogen Oxides
Smoke
Mathematical Modeling
Photography
Ultraviolet Spectrometry
Soot
Computer Program
Fuel Sprays
Internal Combustion Engine
Fuel Consumption
Mobile Sources
Exhaust Gas Recirculatio:
Divided Chamber Engine
21-07 (Recipro-
L eating Engines)
21-02 (Combus-
tion)
18. DISTRIBUTION STATEMENT
Release Unlimited
19. SECURITY CLASS (This Report)
21, NO. OF PAGES
180
20. SECURITY CLASS (This page)
22. PRICE
EPA Form 2220-1 (9-73)
175
------- |