EPA-R2-73-291
July 1973
Environmental Protection Technology Series
-------
EPA-R2-73-291
NITRIC OXIDE FORMATION
IN COMBUSTION PROCESSES
WITH
STRONG RECIRCULATION
by
C.T. Bowman, L.S. Cohen, and M.N. Director
United Aircraft Research Laboratories
East Hartford, Connecticut 06108
Contract No. 68-02-0252
Program Element No. 1A2014
EPA Project Officer: G.B. Martin
Control Systems Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND MONITORING
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
July 1973
Chicago, lij.i^.
-------
This report has been reviewed by the Environmental Protection Agency and
approved for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the Agency, nor does
mention of trade names or commercial products constitute endorsement
or recommendation for use.
11
-------
ABSTRACT
An investigation of NO formation in a premixed, turbulent flame stabilized
on the recirculation zone downstream from a bluff-body has been carried
out. The objectives of this investigation were to investigate the factors
affecting NO formation in the recirculation zone and to assess the rela-
tive importance of NO production in the zone to overall NO production.
The dependence of NO formation in the recirculation zone on the properties
of the zone was determined. WO production in the recirculation zone was
strongly influenced by non-equilibrium chemical effects and by turbulent
exchange processes. Information on turbulent exchange was obtained in
a complementary investigation of recirculation zone fluid dynamics in
non-reacting flows. Comparison of NO production in the recirculation
zone with overall NO production indicates that, for the experimental
configuration, the recirculation zone is not a major factor in overall
NO production.
An analytical model for NO production in the burner was developed. Results
from this model suggest that the recirculation zone can be a major factor
in NO production in practical combustion devices.
This report was submitted in fulfillment of Contract 68-02-0252, by United
Aircraft Research Laboratories, under the sponsorship of the Environmental
Protection Agency. Work was completed as of March 1973-
iii
-------
COWENTS
Page
Abstract ill
List of Figures vi
List of Tables x
Acknowledgments xi
Sections
I Introduction 1
II Background Information 3
III Combustion Experiments 12
TV Cold Flow Experiments 51
V Analytical Investigation 78
VI Conclusions 1°3
VII References 105
VIII Nomenclature 109
IX Appendices 113
v
-------
FIGURES
No. Page
1 Wake Flow Behind Circular Cylinder 6
2 Schematic Diagram of the Two-Dimensional Burner Facility 13
3 Typical Velocity Profile at the Test Section Entrance 14
4 Two-Dimensional Test Section with Injecting Flameholder 15
5 Stability Limits for Methane-Air Flames in Two-Dimensional
Burners with 0.32-cm Diameter Cylindrical Flameholders 16
6 Spark Schlieren Photograph of a Methane-Air Flame 20
Ta Typical Dependence of Recirculation Zone Length/Flameholder
Diameter on Inlet Velocity 21
7b Typical Dependence of Maximum Recirculation Zone Width/
Flameholder Diameter on Inlet Velocity 22
8 Dependence of Recirculation Zone Volume on Inlet Velocity
and Equivalence Ratio for Methane-Air Mixtures 23
9 Time-Exposure (1/500 Sec) Schlieren Photograph of a
Methane-Air Flame 2.6
lOa Typical Dependence of the Maximum Flame Angle on Inlet
Velocity for a Stoichiometric Methane-Air Mixture 27
lOb Typical Dependence of the Maximum Flame Angle on Equiva-
lence Ratio for Methane-Air Mixtures 28
11 Typical Temperature Distribution in the Recirculation Zone 31
12 Typical Dependence of Recirculation Zone Temperature on
Equivalence Ratio 32
13 Typical Dependence of Recirculation Zone Temperature on
Inlet Velocity 33
VI
-------
No. Page
14 Dependence of Mean Residence Time on Inlet Velocity,,
Equivalence Ratio and Measurement Location 37
15 Hydroxyl Radical Concentration in the Recirculation
Zone as a Function of Equivalence Ratio and Inlet Velocity 4l
l6a Typical Dependence of Nitric Oxide Concentration in the
Recirculation Zone on Equivalence Ratio and Measurement
Location 44
l6b Typical Dependence of Nitric Oxide Concentration in the
Recirculation Zone on Inlet Velocity 45
ITa- Typical Species Concentration Profiles at the Exhaust of
the Combustion Test Section 46
17"b Typical Nitric Oxide Concentration Profiles at the Exhaust
of the Combustion Test Section 47
18 Cold Flow Fluid Dynamics Test Facility 52
19 Schematic Daigram of Cold Flow Fluid Dynamics Test Section 53
20 Fiber Optic Probe 56
21 Typical Velocity Profiles in Cold Flow Test Section 57
22 Cold Flow Test Section with Hot Wire Probe Installed 58
23 Turbulence Level Downstream of Screens 60
24 Schlieren Photograph of Near-Wake Region 63
25 Vortex Shedding Sequence 64
26 Recirculation Zone Geometrical Characteristics 65
27 Residence Time Determination 67
28 Residence Time Correlating Parameter for Cold Flow 69
29 Oscilloscope Trace of Typical Tracer Concentration
Distribution 71
VII
-------
— Page
30 Axial Variation of Temperature in Recirculation Zone 72
31 Typical Tracer Centerline Concentration Distribution 73
32 Tracer Concentration Distribution on Recirculation Zone
Centerline 7^
33 Tracer Concentration Contour Map 75
3^ Analytical Model 85
35 Methane-Air Ignition Delay Time 91
36 Calculated Flame Boundary 98
37 Calculated Wake Velocity Profile 99
38 NO Concentration Distribution 100
39 Calculated Species Distributions 101
A-l Schematic Diagram of the Schlieren Optical System 115
A-2 Schematic Diagram of the Sodium D-Line Photometer 115
A-3 Typical Experimental Traces Obtained from the Temperature
Measurement Experiments 117
A-4 Typical Oscilloscope Trace from Residence Time Experiments 119
A-5 Typical Semi-Logarithmic Plot of Absorption Decay Curve 120
A-6 Schematic Diagram of the UV Photometer 122-
A-7 Typical Nitric Oxide Absorption Spectrum-Nitric Oxide Mole
Fraction = 2 x 10"3 121).
A-8 Schematic Diagram of Probe and Sampling System Used for
Exhaust Concentration Measurements 126
A-9 Sampling Probe in Combustor Exhaust 128
Vlll
-------
No. Page
A-lOa Typical Observed Mass Spectra of the Combustor Exhaust
in the Range m/e = 28-30 129
A-lOb Typical Deconvolved Mass Spectra 129
B-l Schematic Diagram of Fiber Optic Instrumentation 132
B-2 Component and System Performance of Fiber Optic Probe
for W0-N0 Tracer 132
B-3 Properties of NO -N?0^ Tracer 133
ix
-------
TABLES
No. Page
I Residence Times for Three-Dimensional Bluff Bodies 11
II Combustion Test Matrix 18
III Recirculation Zone Volume and Flame Spreading Rates 2k
IV Recirculation Zone Temperatures 3^-
V Mean Residence Times 36
VI OH Concentration 40
VII Nitric Oxide Concentrations in Recirculation Zone k2
VIII Nitric Oxide Concentrations in Exhaust Gas 48
IX Ratio of Nitric Oxide Production in the Recirculation
Zone to Total Nitric Oxide Production 50
X Screens for Turbulence Production 59
XI Cold Flow Test Matrix 62
XII Residence Times and Correlation Parameter 68
XIII Tracer Concentration Distribution Data 76
XIV Summary of Analytical Results 96
-------
ACKNOWLEDGMENTS
The time-of-flight mass spectrometer used in the combustion experiments
was designed by Dr. M. F. Zabielski. His assistance with the exhaust
sampling portion of the combustion experiments is gratefully acknowledged.
The authors thank Dr. L. J. Coulter, Mr. Richard Roback, Mr. David B.
Smith and Mrs. L. Rufleth for their help in the analytical investigation.
In addition, the authors thank Mr. J. E. Wright and Mr. R. Smus for
valuable assistance in the experimental program.
This research program was carried out under the sponsorship of the
Environmental Protection Agency, Durham, North Carolina with Dr. G.
Blair Martin as Project Officer.
XI
-------
SECTION I
INTRODUCTION
Recent experimental investigations of factors affecting pollutant emissions
from various continuous combustion systems (Refs. 1-3) have shown that
changes in operating conditions, which alter the flow patterns in the
combustion chamber, can have a substantial effect on nitrogen oxide emis-
sions. This observation suggests that coupling between fluid dynamic
and chemical processes in the combustion chamber is a major factor gov-
erning nitrogen oxide emissions. In many practical continuous combustion
systems, such as furnaces and gas turbines, the flame is stabilized by a
primary combustion zone, characterized by strong recirculation patterns.
Because of the high temperatures and relatively long residence times
associated with these regions, it appears likely that a substantial
amount of nitric oxide formation occurs in the recirculation zones.
Hence, changes in operating conditions which alter recirculation zone
characteristics -- notably recirculation zone size and the temperature
and residence time of gas in the zone — can be expected to produce
changes in nitrogen oxide emissions. To gain an initial understanding
of how coupling between fluid dynamic and chemical processes in a
combustion device governs nitrogen oxide emissions, one can first
examine the nitric oxide formation process in recirculation zones.
The present report documents the results of an investigation, sponsored
by EPA Contract 68-02-0252, of nitric oxide formation in a combustion
process with strong recirculation. The objectives of this investigation
were to (a) investigate the factors affecting nitric oxide formation in
the recirculation zone and (b) assess the relative importance of nitric
oxide production in the recirculation zone to overall nitric oxide
production.
In this investigation, a simplified experimental configuration was used
to permit detailed examination of the nitric oxide formation process in
the combustor as burner input parameters (fuel/air ratio and inlet velo-
city) were varied. To assist in interpretation of the experimental
results and to permit extrapolation of these results to conditions out-
side the range of the present study, an analytical model for nitric oxide
formation in the model combustor was developed. To obtain necessary
background information on the fluid dynamics of recirculation zones,
a detailed experimental investigation of recirculation zones downstream
-------
from bluff-bodies in non-reacting flows was carried out. The various
aspects of the investigation, outlined above, are discussed in Sections
III, IV, and V.
-------
SECTION II
BACKGROUND INFORMATION
II-A. COMBUSTION EXPERIMENTS
The principal objective of the combustion experiments was to obtain
information on the nitric oxide formation process in a continuous flow
combustion system in which the flame is stabilized by recirculation.
Specifically, the factors affecting nitric oxide production in the
recirculation zone were to be identified and the importance of nitric
oxide production in the recirculation zone to overall nitric oxide
production in the combustor was to be determined.
A simplified experimental configuration was used to permit detailed
examination of the nitric oxide formation process in the combustor.
In this configuration, a flame was stabilized in a two-dimensional
turbulent, premixed, gaseous fuel-air stream by introducing a 0.318-
cm diameter cylinder into the flow. The mixture of burned and unburned
gases, entrained into the two-dimensional recirculation zone located
downstream from the cylinder, served to stabilize the flame in the
combustor. A detailed description of the combustor is given in
Section III-A.
The amount of nitric oxide produced in the recirculation zone is known
to depend on four principal factors -- the recirculation zone volume,
the temperature and residence time of the gas in the zone and the 0-
atom concentration in the zone. To characterize the nitric oxide
formation process in the recirculation zone, the dependence of the
nitric oxide concentration in the zone on these four factors must be
determined. Each of the four factors depends, to a certain extent,
on the burner input parameters -- fuel/air ratio, inlet velocity and
turbulence level. Hence, to obtain the dependence of nitric oxide
production in the recirculation zone on the zone characteristics,
the burner input parameters were varied systematically, and the sub-
sequent variations in recirculation zone volume, temperature and
residence time of the gas in the zone and 0-atom and nitric oxide
concentrations in the zone were determined. These data then were
used to correlate nitric oxide production in the recirculation zone
with the important zone characteristics. To assess the relative
importance of the recirculation zone to the overall nitric oxide
-------
formation process, the nitric oxide produced in the recirculation zone
was compared to the total nitric oxide produced in the combustor, as meas-
ured in the exhaust gas.
Results from the combustion experiments are discussed in Section III, and
a detailed description of the experimental techniques is given in Appendix
A.
II-B. COLD FLOW EXPERIMENTS
Development of a model for nitric oxide production in the recirculation
zone requires knowledge of the turbulent transport processes within the
recirculation zone and between the zone and the outer flow. This informa-
tion was not readily obtained from the combustion experiments due to the
high temperature and small size of the stabilizing recirculation zone.
To provide some understanding of the turbulent transport processes, the
two-dimensional recirculation zone downstream from a 1.59-cm diameter
cylinder placed with its axis normal to a cold, nonreacting flow was
investigated experimentally. In these tests, a tracer material was
injected through holes in the cylinder wall into the recirculation
zone where tracer residence times and tracer concentration distribu-
tions were determined using a fiber optic probe. Recirculation zone
geometry was determined from spark schlieren photographs taken using
injected helium tracer. High speed schlieren movies also were obtained
to aid in the understanding of the vortex shedding phenomenon encountered
with two-dimensional bluff-bodies. Measurements were taken at several
approach velocities, free-stream densities and several initial levels
of turbulent intensity. Results from the cold flow experiments are
discussed in Section IV, and a detailed description of the fiber optic
measurement techniques is given in Appendix B.
II-C. FLOW FIELD DOWNSTREAM FROM 2-DIMENSIONAL BLUFF-BODIES
For the Reynolds number range covered in the present experiments the
flow in the two-dimensional near-wake downstream from the cylindrical
body, which includes the recirculation zone, is categorized as sub-
critical. In non-reacting flow, the subcritical near-wake regime,which
extends from Reynolds numbers of about 300 to 2 x 10 , manifests a
r f
nearly constant Strouhal number,corresponding to a dominant Karman
vortex double-street instability (Ref. U). The Strouhal number, or
dimensional vortex shedding frequency, is given by
sh *
-------
Figure la is a sketch of the incompressible near-wake flow for a circular
cylinder, revealing its nonsteady vortex shedding character. The sketch
in Fig. lb is the equivalent mean flow, which is obtained by averaging
over times which are large compared to the significant periods of the
nonsteady flows, eg., the vortex-shedding period. It has been pointed
out in Ref. 5, that the closure point "r" in Fig. Ib occurs in the same
region where the vortices complete their growth and then break away.
In the Reynolds number range from 1500 to 2 x KK, the predominant
phenomenon in the near wake region is the upstream movement of the
transition point along the free shear layers, from near the closure
point to near the separation point, "s" (Ref. 6). However, at a Rey-
nolds number of about 10 , the transition point has approached quite
close to the cylinder separation point and the major portion of the
free shear layer in the near wake is turbulent. For Reynolds numbers
in excess of 2 x KK but less than 3 x 10°, comprising the so-called
critical flow regime, the vortex shedding is less definite and
regular than in the subcritical regime (Ref. 4), and there is strong
sensitivity to free stream turbulence and surface roughness. The
near-wake characteristics enumerated here for a circular cylinder
also are displayed with other two-dimensional bodies such as wedges,
although the Reynolds number ranges do not necessarily correspond.
The vortex shedding phenomenon which is characteristic of two-dimen-
sional bluff-body flows under nonreacting conditions, has not been
observed in combusting flows (cf., Ref. 7 and results of the present
burner tests). It may be anticipated, therefore, that the nature of
the transport processes in a reacting two-dimensional flow field will
differ from those in cold flow. Nevertheless, detailed cold flow
measurements can be useful in the interpretation and modeling of
hot flow data, and in providing a basis of comparison for reacting
flows.
II-D. TRANSPORT PROCESSES IN THE NEAR-WAKE REGION
In the analytical treatment of transitional/turbulent shear flows,
the local shear stress may be expressed as the product of an eddy
viscosity and the local velocity gradient by analogy with the
laminar flow representation. However, while the molecular viscosity
for laminar flow depends only on the fluid properties, the eddy vis-
cosity is related to the length and time scales which characterize the
transition/turbulence structure of the shear flow. At present, transi-
tional/turbulent flow phenomena are not well-understood so that empirical
hypothyses are used to create a mathematical basis for the investigation
5
-------
a) INSTANTANEOUS, NONSTEADY FLOW (WITH VORTEX SHEDDING)
b) MEAN FLOW
f-
FREE SHEAR LAYER
CLOSURE POINT
SEPARATION'
POINT s
i VORTICAL (TURBULENT) FAR-WAKE
FIG 1
WAKE FLOW BEHIND CIRCULAR CYLINDER.
-------
of such flows. These phenomenological theories lead to a formulation
of the eddy viscosity which may be used with the equations of motion
and a suitable equation of state to determine the local time-average
conditions throughout a flow field.
In any general formulation of the eddy viscosity for two-dimensional
bluff-body near-wake flows, it is necessary to consider the body-
related turbulence associated with vortex shedding, the turbulence
initially present in the flow ("preturbulence"), which may be augmented
with the use of mixing aids such as screens, and the turbulence pro-
duced as a result of the interactions between the outer flow and the
wake. In the Reynolds number range of the present cold flow experiments,
it may be expected that the body-related turbulence plays an important
role since the well-established vortex shedding produces significant
lateral agitation through periodic distortions of the near-wake region.
At higher Reynolds numbers, in the critical flow regime, the Strouhal
number increases rapidly and the distinctive vortex shedding pattern
is lost. Thus, in the critical flow regime, it is anticipated that
body-related turbulence is less important than free stream-wake inter-
actions which produce shearing stresses of large magnitude which, in
turn, induce high levels of turbulent intensity. The presence of
screens or other mixing aids at any Reynolds number will be effective
in influencing transport only if the levels of turbulence generated
exceed those produced by other competing mechanisms. For any given
set of flow conditions, mixing aid geometry and location, and bluff-
body geometry and blockage, it is likely that only one of the three
turbulence producing mechanisms dominates the transport.
II-D.l. Body-Related Turbulent Transport
For the purpose of developing a formulation of the eddy viscosity for
body-related transport and the associated recirculation zone residence
time, it is convenient to begin with the basic expression,
, A
€//> = -V'JI1 (2)
(see Ref.8 ). Equation 2 relates the eddy viscosity, 6, to a parameter
associated with the characteristic disturbance size within the turbulence
field, &', and the transverse fluctuating velocity or transport velocity,
v'. The density p is some appropriate reference value for the system
-------
under consideration. In the spirit of Prandtl's mixing length hypo-
thesis (cf., Ref. 9), it is assumed that the mean of the product of
fluctuating quantities is proportional to the product of the means of
the absolute values of these quantities, ie.,
with the constant c being less than or equal to unity. Since vortex
shedding is the phenomenon which characterizes the body-related trans-
port process, it is natural to express the transport velocity as,
|v'|~ D/TV ~ (Sh)U00
where TV is the vortex shedding period and Sh is the Strouhal number.
Furthermore, since the laminar separation process on the bluff body,
the onset of transition in the shear layer and the formation of the
vortex-street must be intimately related (cf., Ref. 5 ), then
where the Reynolds number is based on the body diameter and approach
flow conditions. Thus,
A relationship between the eddy viscosity and the re circulation zone
residence time is established by balancing the quantity of the tracer
material within the zone at steady-state conditions with that which
is transported out of the zone following suspension of tracer flow
into the zone. Thus,
D
(5)
* (Sh)UooD
/P~ (6)
where rt is the residence time and Vw and Aw are the recirculation zone
volume and surface area, respectively. The tracer concentration, y^,
and the zone density, pw, are expected to vary throughout the recircula-
tion zone.
-------
If Eq.s. 6 and 7 are combined, there results
yt/>w
Sh) (8)
Thus, for the present cold flow experiments in which Sh is approximately
constant and only a single body diameter is considered, it is predicted
that the residence time varies inversely as the square root of the
velocity or since Uoo ~ (Tv) > the residence time varies directly as
the square root of the vortex shedding period. The influence of
density level on TJ is not immediately obvious due to the unknown
variation of A in terms of y+P^ and px .
II-D.2. Free Stream - Wake Interaction Transport
Turbulence developed due to interactions between streams having differ-
ent velocities and/or densities typically is characterized in terms of
the velocity difference between streams and the transverse extent of
the mixing (shear) layer (Ref.10), ie.,
€/p ~ bUooll-Uw/Uoo)
where Uw is some appropriate (mean) wake velocity. Noting that the
slope of the mixing layer may be taken as constant (Ref.9 ), ie., b~L,
it follows that
(10)
where it has been assumed that U «Uoo .
The free stream-wake interaction model can be tested by referring to
the data of Winterfeld, Ref .11, for three-dimensional bodies in iso-
thermal flow, since no vortex shedding occurs in this case. It was
shown in the cited work that the zone volume and surface area are
proportional to the product of the zone length and maximum width, ie.,
Vw = (0.54^0.58)|_BM
Aw = (2.70— 2.90) LBM
independent of velocity in the range 20 to 80 m/sec, for a circular disc,
-------
and ^5 d.eg and 90 3eg cones. Thus, at any given velocity, it follows
from Eq.,10 that the residence time ratio for any two bluff-bodies
varies as,
t, (L/D)2
(L/D),
Winterfelds data for the range of velocity 20-80 m/sec is given in
Table I. Also indicated in Table I are the residence times computed
from Eq.ll using the 90 deg cones as a reference, and the error between
observed and calculated values. The degree of agreement attained is
considered satisfactory.
10
-------
TABLE I
RESIDENCE TIMES FOR THREE-DIMENSIONAL BLUFF BODIES
BODY
CIRCULAR DISC
90 DEC CONE
45 DEC CONE
(L/D) - REF 11
208
1 92
1 67
Tt/Uoc~ REF 11
0925
1 20
1 425
CALCULATED rt/Uric
1 11
1 20 (REFERENCE)
1 38
ERROR
18%
-
3%
11
-------
SECTION III
COMBUSTION EXPERIMENTS
III-A. TURBULENT FLAME BURNER
The combustion experiments were carried out in a two-dimensional burner
facility, Fig. 2 . Methane (Technical Grade) and air (-60°C dew point),
metered by calibrated critical flow orifices, enter a mixing chamber
through an impinging-type injector. Complete mixing of the methane and
air is obtained by flowing the gases through a tightly-packed bed of stain-
less steel balls. The methane-air mixture flows through stainless-steel
calming screens, a 25 to 1 contraction-ratio nozzle and a turbulence-
generating screen and enters the test section with essentially a uniform
velocity profile, Fig. 3• Four different turbulence-generating screens,
20-, 60-, 100- and 200-mesh stainless-steel screens, with blockages ranging
from 33% to 5^$>»'were available for use in the present investigation. The
uncooled stainless-steel test section, Fig. 4, is 15 cm long and has a
rectangular cross-section (1.27 cm x 3«8l cm). A wall static pressure
tap, located near the entrance, is used to measure the test-section pres-
sure drop. The test section is fitted with two fused silica windows to
permit visual observation of the flame in the vicinity of the flameholder.
A cylindrical flameholder (0-32-cm diameter) is located in the center of
the test-section cross-section approximately 1.0 cm downstream from the
turbulence generating screen. The flameholder blockage ratio (BR = flame-
holder diameter/test section height) was 0.25. Two different types of
flameholders were used in the investigation -- a solid, stainless steel
rod and a water-cooled, hollow, stainless-steel injecting flameholder
(shown in Fig.4 ). Most of the experiments were carried out using the
solid-rod flameholder. However, when it was necessary to inject tracer
material into the recirculation zone (as, for example, in the experiments
designed to measure the residence time of gas in the recirculation zone),
the injecting flameholder was used. The injecting flameholder is described
in greater detail in Appendix A.
The range of stable operating conditions for the two-dimensional burner,
using methane-air, is shown in Fig. 5 • The solid curve indicates the
range of stable operating conditions (inlet velocity and equivalence
ratio*) for the uncooled rod flameholder. For inlet velocities and
equivalence ratios lying within the bounds of this curve, a flame can
^Equivalence ratio, 0 = m~-, /m .
XU.SJ- on
air
(' I'
fuel air)stoichiometric
12
-------
-EXHAUST DUCT
FLAMEHOLDER-
NOZZLE SECTION-
EXPANSION SECTION
-TEST SECTION
-FUSED SILICA WINDOW
-TURBULENCE-GENERATING
SCREEN
200-MESH
CALMING SCREENS
MIXING CHAMBER
METHANE AIR
FIG. 2 SCHEMATIC DIAGRAM OF THE TWO-DIMENSIONAL BURNER FACILITY.
13
-------
GO,
I
50
40
30
CJ
o
20
10
T T T I I
I / I
CALCULATED
I
I
I
I
I
z
06 04
TEST SECTION WALL
02 (^ 02
POSITION - CM
04
0.6
FIG 3 TYPICAL VELOCITY PROFILE AT THE TEST SECTION ENTRANCE.
-------
TEST SECTION
FROM WALL STATIC PRESSURE TAP
FIG. 4 TWO-DIMENSIONAL TEST SECTION WITH INJECTING FLAMEHOLDER
15
-------
160
120
ZUKOSKI & MARBLE .
(REF 1 2) BR - 1/32
\
\
\
NO COMBUSTION
u 80
O
40
UARL BURNER
0 7
04
06
08 10
EQUIVALENCE RATIO
1 2
1 4
FIG 5 STABILITY LIMITS FOR METHANE-AIR FLAMES IN TWO-DIMENSIONAL
BURNERS WITH 0.32-CM DIAMETER CYLINDRICAL FLAMEHOLDERS
16
-------
be stabilized in the test section. For inlet velocities and equivalence
ratios lying outside this curve, a flame cannot be stabilized in the test
section. The dotted curve indicates the range of stable operating condi-
tions for the cooled, injecting flameholder (with tracer injection).
Tracer injection results in a significant reduction in the range of stable
operating conditions. No variation in stability limits with the mesh size
of the turbulence-generating screens was observed. The dashed curve shows
the range of stable operating conditions reported by Zukoski and Marble
(Ref.12) for methane-air in a burner with a flameholder blockage ratio
of 1/32.
Based on the experimental stability limits, a test matrix was established
for combustion experiments. This test matrix, TableH, indicates the
nominal test conditions (inlet velocity and equivalence ratio) for which
the various experimental measurements were made. The flow visualization
experiments, discussed in Section III-B, were carried out to obtain in-
formation on recirculation zone geometry and flame spreading rates. The
temperature measurement experiments, discussed in Section III-C, provided
information on the temperature distribution in the recirculation zone.
The residence time measurements, discussed in Section III-D, provided
information on the turbulent exchange process between the recirculation
zone and the main flow. The OH concentration measurements, discussed
in Section III-E, were carried out to obtain information on radical
concentrations (OH, 0) in the recirculation zone. The NO concentration
measurements, discussed in Section III-E, provided information on the
nitric oxide concentration in the recirculation zone and in the exhaust
gas.
III-B. FLOW VISUALIZATION EXPERIMENTAL RESULTS
The volume of the recirculation zone is an important factor in deter-
mining the amount of nitric oxide produced in the zone. For a given
volumetric nitric oxide production rate, the total nitric oxide pro-
duction rate in the recirculation zone is proportional to the volume
of the zone. In the present investigation, a spark-source schlieren
optical system was used to determine the volume of the recirculation
zone located downstream from the flameholder. A schematic diagram
and detailed description of the schlieren optical system are given in
Appendix A.
The sharp concentration and density gradients associated with the flame
surrounding the recirculation zone result in a fairly well-defined
IT
-------
TABLE II
COMBUSTION TEST MATRIX
^\ 0
UoXT
(M/SEClX^
20
25
30
40
50
60
0.8
F, T, R
OH, NO
F, T, R
F, T
0.9
F, T, R
OH, NO
F, T, R
OH
F, T, R
F, T, R
NO
F, T
1.0
F, T, R
OH, NO
F, T, R
OH
F, T, R
F, T, R
NO
F, T
F
NO
1.1
F, T, R
OH, NO
F, T, R
F, T, R
F,T, R
NO
F, T
F
NO
F = FLOW VISUALIZATION
T = TEMPERATURE MEASUREMENT
R = RESIDENCE TIME MEASUREMENT
OH = OH CONCENTRATION MEASUREMENT
NO = NO MEASUREMENT
18
-------
re circulation zone boundary, Fig. 6. From spark schlieren photographs,
such as Fig. 6, the length, width and "cross-sectional" area of the
re circulation zone can be determined.
A typical variation of recirculation zone length/flameholder diameter with
inlet velocity is shown in Fig.Ta. Vertical bars through the present
experimental results indicate the uncertainty in recirculation zone length,
resulting from the lack of definition in the recirculation zone boundary.
Results from two other investigations of two-dimensional bluff-body
stabilized flames are shown in Fig. 7s* Wright (Ref.13) measured recircu-
lation zone lengths for two-dimensional hydrocarbon-air flames stabilized
on flat plates and cylinders.^located transverse to the flow direction.
Pein, Peschel and Fetting (Ref.l^) reported recirculation zone lengths
for a two-dimensional propane-air flames stabilized on a triangular rod.
Results from these two studies are consistent with results from the
present investigation and indicate that the recirculation zone length
increases slightly with increasing inlet velocity for the range of
experimental conditions investigated. The recirculation zone length
was found to be independent of the mesh size of the turbulence-generating
screens.
A typical variation of maximum recirculation zone width/flameholder diameter
with inlet velocity is shown in Fig.7k. Vertical bars through the present
experimental results indicate the uncertainty in recirculation zone width,
resulting from the lack of definition in the recirculation zone boundary.
A result from another investigation of recirculation zones in two-dimen-
sional propane-air flames also is shown on Fig-7b. The experimental data
indicate that the recirculation zone width is independent of velocity and
the mesh size of the turbulence-generating screen for the range of experi-
mental conditions investigated.
The dependence of recirculation zone volume on inlet velocity and equiva-
lence ratio is given in Table III and Fig. 8 . The recirculation zone volume
was determined using the recirculation zone "cross-sectional area", obtained
from the spark schlieren photographs, and the test section width, corrected
for boundary layer growth on the test section walls. The vertical bar in
Fig.8 indicates the uncertainty in recirculation zone volume, resulting
from the lack of definition in the recirculation zone boundary. For the
range of experimental conditions investigated, the recirculation zone
volume was independent of equivalence ratio and mesh-size of the turbu-
lence-generating screen and increased slightly with increasing inlet
velocity.
A careful examination of the spark-schlieren photograph, Fig. 6 , reveals
19
-------
, • 26.4 M/SEC
6 - 1 oo
FLAMEHOLDER
SPARK SCHLIEREN PHOTOGRAPH OF A METHANE-AIR FLAME.
20
-------
12
I I
— X
PEIN, ET AL
(REF 14|
I
I
I
20
40
60
80
FIG 7 a
INLET-VELOCITY - M/SEC
TYPICAL DEPENDENCE OF RECIRCULATION ZONE LENGTH/FLAMEHOLDER DIAMETER ON INLET VELOCITY
21
-------
0
X
REIN ET AL
IREF 1 4)
T. i J
i
j_
20 40 60
INLET VELOCITY - M/SEC
FIG 7b
TYPICAL DEPENDENCE OF MAXIMUM RECIRCULATION ZONE WIDTH/FLAMEHOLDER DIAMETER ON INLET VELOCITY
22
-------
0 ' 0 8
0 - 0 9
0 1 0
0-11
o
5
D
O
>
z
o
o
IX
40
INLET VELOCITY - M/SEC
60
FIG 8
DEPENDENCE OF RECIRCULATION ZONE VOLUME ON INLET VELOCITY AND
EQUIVALENCE RATIO FOR METHANE-AIR MIXTURES.
23
-------
TABLE III
RECIRCULATION ZONE VOLUME AND FLAME SPREADING RATES
U^IM/SEC)
21.9
21.4
21.8
22.6
22.0 *
26.4
26.2
26.4
26.7
31.8
32.5
32.1
31.4
39.5
40.1
40.1
40.2 *
51.2
52.2
52.3
62.0
61.2
61.1 *
0.83
0.92
1.01
1.10
1.11
0.80
0.89
1.00
1.12
0.81
0.92
1.03
1.10
0.89
1.00
1.10
1.10
0.93
1.01
1.11
1.00
1.09
1.10
Vw (CM3)
2.54 + 0.30
2.45
2.40
2.49
2.52
2.52
2.52
2.44
2.61
2.44
2.44
2.40
2.48
2.59
2.55
2.63
2.60
2.68
2.68
2.76
2.71
2.79
2.80
TAN "MAX
0.066 + 0.07
0.082
0.082
0.086
0.084
0.066
0.070
0.076
0.074
0.060
0.070
0.068
0.072
0.066
0.063
0.068
0070
0.060
0.059
0.064
0.058
0.066
0.062
100-MESH TURBULENCE SCREEN USED FOR ALL RUNS EXCEPT THOSE
MARKED * WHICH USED A 20-MESH TURBULENCE SCREEN
-------
that, in the vicinity of the flameholder, the flame zone is well-defined
and appears to be laminar. At a distance downstream of approximately
two flameholder diameters, the flame begins to undergo a transition from
laminar to turbulent. The transition is essentially complete at a dis-
tance downstream of approximately six flameholder diameters, and the
flame structure is characterized by random turbulent fluctuations (eddys).
At the smallest Reynolds number investigated (Re = 0.5 x ICr, based on
inlet conditions and flameholder diameter), the turbulent transition
point occurred very near the end of the recirculation zone. As the
Reynolds number was increased, the transition point moved upstream
until, at the largest Reynolds number investigated (Re = 1.4 x ICn"),
the transition point appeared to be less than one diameter downstream
from the flameholder. Zukoski and Marble (Ref.12) have reported similar
wake transition phenomena in bluff-body-stabilized hydrocarbon-air flames.
The gradual change in the flame zone bounding the recirculation region
from laminar to turbulent, with increasing Reynolds number, likely will
affect nitric oxide production in the recirculation region due to changes
in the recirculation zone characteristics.
To assist in the development of an analytical model for nitric oxide forma-
tion in the burner, flame spreading rates in the burner test section were
determined as functions of inlet velocity and equivalence ratio. These
spreading rates were measured on time-exposure schlieren photographs,
taken using the schlieren system with the spark source replaced by a
Hg-Xe dc arc lamp. These time-exposure schlieren photographs, Fig. 9,
give a time-averaged picture of the reacting flow field. Examination
of Fig. 9 reveals that in the vicinity of the recirculation zone the
flame spreads very slowly. Downstream from the recirculation zone the
flame spreads more rapidly until the flame approaches the test section
wall, where the spreading rate decreases. For present purposes, flame
spreading in the burner is characterized by the maximum flame spreading
rate. This maximum spreading rate is expressed in terms of the tangent
of the maximum angle between the leading edge of the flame front and
the direction of the approach flow of unburned gas. While determination
of the maximum angle is somewhat subject to personal interpretation,
several different observers did agree on the angle to within ±10%.
Typical experimental results are presented in Fig.ID, and all measured
flame angles are tabulated in Table III. The vertical bars through the
experimental results represent the range of flame angles measured by
several different observers. These experimental results will be discussed
in detail in the section on analytical modeling, Section V.
25
-------
n = 264M/SEC
=1.00
FLAMEHOLDER
FIG. 9 TIME-EXPOSURE (1/500 SEC) SCHLIEREN PHOTOGRAPH OF A METHANE-AIR FLAME
26
-------
012
0 10
<
K
008
006
0.04
002
MEASURED
Q CALCULATED
FIG 10a
20
40
INLET VELOCITY - M/SEC
60
80
TYPICAL DEPENDENCE OF THE MAXIMUM FLAME ANGLE ON INLET VELOCITY
FOR A STOICHIOMETRIC METHANE-AIR MIXTURE.
-------
0 10
008
o -1-
LL 22 M/SEC
004
002 •
MEASURED
CALCULATED
I
I
0 fa 08 10 12
EQUIVALENCE RATIO
FIG 10b TYPICAL DEPENDENCE OF THE MAXIMUM FLAME ANGLE ON EQUIVALENCE RATIO
FOR METHANE-AIR MIXTURES.
28
-------
III-C. TEMPERATURE MEASUREMENT EXPERIMENTAL RESULTS
The rate of formation of nitric oxide from atmospheric (molecular)
nitrogen is strongly dependent on temperature. Hence, to characterize
the nitric oxide formation process in the recirculation zone, the
temperature distribution in the zone must be determined. Two different
experimental approaches can be used to measure the temperature of the
hot gas in the recirculation zone -- probe thermometry (eg. thermocouples)
and optical techniques (eg. sodium-line reversal). Probe thermometry can
provide a measure of the local time-mean temperature; however, probe
techniques have the major disadvantage of disturbing the flow field.
This disturbance can be particularly severe for the relatively small-
scale recirculation zones in the present investigation. Optical tech-
niques can be used to measure temperature without disturbing the flow
field; however, these techniques can provide only an average-value of
the temperature along the optical path. In the two-dimensional flows,
used in the present investigation, flow properties are essentially
constant across the flow (ie. in the direction parallel to the flame-
holder axis), so that the space-averaging of the optical techniques is
not a serious drawback.
In the present experiments, a modified sodium-line reversal technique
was used to measure the gas temperature in the recirculation zone. The
basic principal underlying the sodium-line reversal technique is that of
matching the spectral brightness of a light source to the spectral bright-
ness of sodium emission from the flame (Ref.15). At the matching point,
the translational temperature of the flame is equal to the brightness
temperature of the light source (at the wavelength of the sodium emission).
A detailed discussion of the experimental technique used in the present
investigation and a schematic diagram of the optical system are given
in Appendix A.
In the present experiments, two different techniques were used to inject
sodium tracer into the recirculation zone. For most of the measurements,
the solid, uncooled flameholder was used, and a small amount of crystalline
sodium chloride was placed on the downstream side of the flameholder.
During combustion, the salt crystals vaporized, and sodium entered the
recirculation zone by convection. In a limited number of experiments,
the injecting flameholder was used to inject a water/sodium chloride
solution into the recirculation zone.
Temperature measurements were made at four locations in the recirculation
29
-------
zone for conditions presented in the combustion test matrix, Table II.
The four locations are illustrated in Fig. 6. At a distance of 0.5 cm
downstream from the trailing edge of the flameholder;, the temperature
was measured at three different transverse positions -- positions 1
(centerline), 2 and 3- The temperature also was measured at a second
axial position on the recirculation zone centerline, approximately
0.75 cm downstream from the flameholder -- position k. The measured
temperatures are, in fact, volume-averaged values, the volume, over
which the average is taken, being determined by the :'.mage of the mono-
chromator entrance slit in the test section. The approximate cross-
section of this volume, shown in Fig. 6} is 0.01 cm x 0.20 cm.
Some typical experimental observations are illustrated in Figs. 11-13,
and all of the temperature data are tabulated in Table IV. The data
plotted in Fig.H suggest that, within experimental uncertainty, the
temperature is uniform throughout most of the recirculation zone. The
data in Fig. 12 show that the recirculation zone temperature is approxi-
mately 10-15$ lower than the adiabatic combustion temperature for the
freestream equivalence ratio. Heat losses to the flameholder and test
section walls cannot account for the low recirculation zone temperatures.
It appears that the low zone temperatures are due in part to un-
reacted fuel in the zone. In Fig. 13, the recirculation zone tempera-
ture increases with increasing inlet velocity, possibly reflecting an
increase in transport rates due to movement of the turbulent transition
point upstream with increasing velocity. In an investigation of tempera-
ture distributions in the recirculation zone of bluff-body-stabilized
propane_air flames (Ref.l4), observations similar to those of Figs. 11 and 13
were reported. Limited experiments with the injecting flameholder have
shown that tracer injection reduces the recirculation zone temperature by
approximately 75°K. This decrease in temperature can be attributed to
the heat capacity of the water/salt solution injected into the recircula-
tion zone.
III-D. RESIDENCE TIME MEASUREMENT EXPERIMENTAL RESULTS
If the recirculation zone is considered to be a chemical reactor in which
nitric oxide is produced, then the nitric oxide concentration in the zone>
as well as the net nitric oxide production rate, will depend on the rate
of mass exchange between the recirculation zone and the outer flow. Each
gas element entering the recirculation zone will remain there for different
periods of time. The net mass exchange rate may be calculated con-
sidering an ensemble of gas elements with a distribution of residence times.
30
-------
U^ = 22 5 M/SEC
0- 1 10
O
Q- C
UJ
on
cc
RECIRCULATION ZONE BOUNDARY
1700 1800 1900 2000
TEMPERATURE - °K
FIG 11 TYPICAL TEMPERATURE DISTRIBUTION IN THE RECIRCULATION ZONE.
-------
LU - 261 M/SEC
POSITION 1
2300
2200
2100
2000
1900
1800
1700
I
06
08
1 0
EQUIVALENCE RATIO
1 2
FIG 12 TYPICAL DEPENDENCE OF RECIRCULATION ZONE TEMPERATURE ON EQUIVALENCE RATIO.
-------
2400
2200
2000
1800
ADIABATIC COMBUSTION TEMPERATURE
0= 1 0
POSITION 1
I
0
20
40
60
INLET VELOCITY - M/SEC
FIG 13 TYPICAL DEPENDENCE OF RECIRCULATION ZONE TEMPERATURE ON INLET VELOCITY.
80
33
-------
TABLE IV
RECIRCULATION ZONE TEMPERATURES
Uw (M/SEC)
22.0
21.8
22.9
23.2
23.2
22.5
23.3
22.5
23.2
22.6
23.2
22.5
23.1
22.7
26.2
26.6
26.4
26.6
32.2
32.1
33.0
30.8
41.1
40.4
41.0
40.8
40.5
40.7
51.1
53.3
52.2
0
0.80
0.91
1.01
1.00
1.00
1.10
1.01
1.10
1.00
1.10
1.00
1.10
1.00
1.09
0.83
0.92
1.00
1.11
0.82
0.90
1.02
1.12
0.91
1.00
1.10
0.90
1.00
1.11
0.92
1.00
1.10
POSITION
1
1
1
1
1
1
2
2
3
3
4
4
4
4
1
1
1
1
1
1
1
1
1
1
1
4
4
4
1
1
1
T(°K)
1690
1805
1870
1865
1855
1910
1850
1885
1845
1865
1880
1885
1810 *
1805 *
1725
1820
1875
1920
1735
1830
1890
1925
1835
1890
1935
1850
1900
1950
1i840
1910
1940
100^MESH TURBULENCE SCREENS USED FOR ALL RUNS
* MEASURED WITH TRACER INJECTION
-------
Alternatively, it is possible to define a mean residence time for the gas
elements, which is related to the mass exchange rate by Eq.. 7 . For
reasons of simplicity, this latter approach was adopted in the present
experiments • The mean residence time of gas particles in the recircula-
tion zone can be determined by'•measuring the time decay of an injected
tracer, following rapid shut-off of the tracer supply (Refs. 11 and 16).
If the tracer is well-distributed throughout the recirculation zone and
if the mass exchange rate/unit area is essentially constant over the
boundary of the recirculation zone, then the tracer concentration decreases
exponentially with time, with the time constant of the tracer decay identi-
fied as the mean residence time (cf., Appendix B).
The tracer material used in the present combustion experiments was sodium.
In these experiments, a water/sodium chloride solution was injected into
the recirculation zone through the rear of the flameholder. The injecting
flameholder is discussed in detail in Appendix A.
The optical system used to make the residence time measurements was
essentially the same as was used for the temperature measurements, (see
Appendix A). However, in the residence time experiments, the photomulti-
plier output was displayed on an oscilloscope. The oscilloscope sweep
was triggered on the termination of tracer flow. The experimental pro-
cedure followed in making the residence time measurements is outlined
in Appendix A.
The residence time data obtained in the present investigation are tabulated
in Table V and are plotted in Fig. 1^. The experimental data show that
the mean residence time is independent of the freestream equivalence ratio,
the mesh size of the turbulence-generating screen and the location in the
recirculation zone where the measurement was made. For inlet velocities
greater than 25 m/sec, the mean residence time varies inversely with the
inlet velocity Uoo, according to
UQQ .= 90 (12)
D
where D = flameholder diameter. A similar correlation has been reported
in two previous investigations of exchange processes behind bluff -body
f landholders (Refs. 11 and 16) . Results from these two investigations are
plotted on Fig. 1^. The data of Bovina (Ref . 16) for two-dimensional
benzene-air flames stabilized on v-gutters are in approximate agreement
35
-------
TABLE V
MEAN RESIDENCE TIMES
UoolM/SEC)
21.8
21.8
22.2
22.1
21.6
22.0
22.2
26.1
26.0
26.4
26.1
31.1
30.9
32.0
40.2
41.0
40.1
41.0
41.1
40.9
44.8
22.0
22.2
22.1
4>
0.80
0.91
1.01
1.11
1.01
1.00
0.99
0.81
0.91
1.00
1.12
0.92
1.00
1.09
0.91
1.02
1.08
0.99
1.00
1.01
0.99
0.90
1.00
1.09
POSITION
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
4
4
4
T, (MSEC)
16.0
14.5
14.2
15.5
14.5*
14.0**
14.1 +
11.5
11.0
12.1
11.0
10.0
9.5
8.9
7.2
7.0
7.0
6.9
7.1*
7.0**
6.5 +
15.1
13.7
16.4
MEASUREMENTS MADE WITH 100-MESH SCREEN EXCEPT
AS INDICATED:
* 20-MESH SCREEN
** 60-MESH SCREEN
+ 200-MESH SCREEN
-------
1-1
o
UJ
UJ
o
CO
LU
tr
UJ
10
-2
10
-3
10
POSITION 1 POSITION 4
: 0.8 V V
= 0.9 • D
:i.o • O
U A A
. BOVINA (REF.16)
(THIS STUDY)
WINTERFELD
(REF 11)
INLET VELOCITY, UM - M/SEC
100
FIG. 14 DEPENDENCE OF MEAN RESIDENCE TIME ON INLET VELOCITY,
EQUIVALENCE RATIO AND MEASUREMENT LOCATION .
37
-------
with results from the present study. Winterfeld's data (Ref. 11) for
propane-air flames stabilized on axisymmetric flameholders are signifi-
cantly different from the data obtained in the two studies in two-
dimensional flows. This deviation may reflect differences in the flow
fields behind flameholders of differing geometry. For the lowest inlet
velocity investigated, the measured residence times are somewhat longer
than would be predicted by Eq. (12). These longer residence times may
be due to the fact that the exchange process at the low inlet velocity
is dominated by laminar transport while at higher velocities turbulent
phenomena are more important.
The observation that the measured residence times are independent of
measurement location tend to support the assumption of an exponential
tracer decay. The observation that the measured residence times are
independent of the mesh-size of the turbulence-generating screens indicates
that the freestream turbulence produced by these screens does not affect
the mass exchange rate between the recirculation zone and the outer flow.
A likely explanation for this result is that the turbulence produced by
the screens decays with downstream distance to the extent that, at the
flameholder position, the screen-induced turbulence is small compared to
the turbulence produced by the shear-layer downstream from the flame-
holder. The burner configuration was such that it was not possible to
closely couple the turbulence-generating screens and the flameholder.
Hence, independent variation of freestream turbulence could not be
obtained in the combustion experiments.
III-E. CONCENTRATION MEASUREMENT EXPERIMENTAL RESULTS
Three different sets of concentration data were obtained in the combustion
experiments. In the recirculation zone, the nitric oxide formation rate
depends on the concentration of radical species, specifically 0, OH and
N, in the zone. To obtain an estimate of radical concentrations in the
zone, the time-average OH concentration was measured at a single loca-
tion in the zone using an ultraviolet absorption technique. The nitric
oxide concentration at two locations in the recirculation zone also was
measured using an ultraviolet absorption technique. The ultraviolet
absorption technique and the experimental procedures used in making the
concentration measurements in the recirculation zone are discussed in
Appendix A. The overall nitric oxide production in the combustor test
section was determined by measuring the nitric oxide concentration in
the exhaust gas using a sampling probe coupled to a. time-of-flight mass
spectrometer. The sampling system and experimental techniques used for
38
-------
the exhaust concentration measurements are discussed in Appendix A.
III-E.l OH Concentration Measurements
The mean OH concentration in the optical path was determined at a single
location in the recirculation zone (position 1) by measuring absorption
of ultraviolet radiation by the ZH(0,0) band of the molecule. The
measured OH concentrations are tabulated in Table VI, and are plotted
as a function of equivalence ratio and inlet velocity in Fig. 15- The
measured OH concentration is weakly dependent on equivalence ratio, and
attains a maximum value for stoichiometric mixtures. For the two velocities
investigated, the measured OH concentration was essentially independent of
inlet velocity. Three curves showing OH concentration for several differ-
ent sets of assumptions and conditions also are plotted on Fig. 15- The
OH-concentration curve labeled "equilibrium-nonadiabatic" was calculated
for the freestream equivalence ratios assuming chemical equilibration at
the measured recirculation zone temperature. The curve labeled "equilibrium-
adiabatic" was calculated for the freestream equivalence ratios assuming
chemical equilibration at the adiabatic combustion temperature. The curve
labeled "equilibrium-stirred reactor" was calculated for a perfectly-stirred
reactor assuming chemical equilibration at the measured recirculation zone
temperature. This stirred-reactor calculation is discussed in more detail
in Section V-A. Comparison of the three calculated equilibrium OH-concen-
tration curves with the experimental data shows that the OH concentrations,
measured in the recirculation zone, exceed OH concentrations corresponding
to any possible equilibrium state. It follows from this observation that
the OH concentrations (and the concentrations of other radical species
such as 0 and N) in the recirculation zone are kinetically controlled and
hence, an assumption of equilibrium hydrocarbon chemistry cannot be used
to model the nitric oxide formation process in the recirculation zone.
III-E.2 Nitric Oxide Concentration Measurements - Recirculation Zone
The mean nitric oxide concentration in the optical path was determined at
two locations in the recirculation zone (positions 1 and h) by measuring
the absorption of ultraviolet radiation by the ^(OjOj-band of the mole-
cule (see Appendix A). A similar optical technique has been used to
measure nitric oxide concentrations in several recent investigations of
pollutant formation in combustion processes (Refs. 17-20).
The nitric oxide concentrations measured in the recirculation zone are
tabulated in Table VII, and are plotted as a function of equivalence ratio
39
-------
TABLE VI
OH CONCENTRATION
U^IM/SEC)
21.6
21.6
21.7
21.6
21.9
21.8
21.7
26.4
26.7
26.7
4>
0.81
0.87
0.91
0.91
1.01
1.01
1.18
0.88
0.92
1.02
COH (10~8 MOLE/CM3)
2.0 ±0.3
2.3
2.4
2.8
2.8
3.2
2.8
2.4
2.4
3.0
ALL MEASUREMENTS MADE AT POSITION 1 (FIG. 6),
USING 100-MESH TURBULENCE GENERATING SCREEN
-------
10
,-7
10-
I
X
0°
10-
10
-10
0.6
FIG.15
EQUILIBRIUM
STIRRED REACTOR
EQUILIBRIUM
AOIABATIC
EQUILIBRIUM
NONADIABATIC-
I
J UCO=26'
M/SEC
I
0.8 1.0
EQUIVALENCE RATIO
1.2
HYDROXYL RADICAL CONCENTRATION IN THE RECIRCULATION ZONE
AS A FUNCTION OF EQUIVALENCE RATIO AND INLET VELOCITY.
-------
TABLE VII
NITRIC OXIDE CONCENTRATIONS IN RECIRCULATION ZONE
Um (M/SECI
21 4
21 7
21 6
21.4
21 4
220
41.3
41.1
41 0
58.8
21.8
21.6
21.6
0
0.81
0.90
1 01
1 01
1.00
1.10
091
1.00
.11
.12
02
.00
.11
POSITION
1
1
1
1
1
1
1
1
1
1
1
1
1
NO MOLE FRACTION - PPM (V)
MEASURED"
NOT MEASURABLE
40 ± 10
65 + 5
75 i 5
62 ±5
105+ 5
30 + 15
44 ± 10
60 + 5
40 + 10
55 + 5
56 + 5
90 + 5
CALC-EQ
STIRRED REACTOR
0.05
0.50
1 7
—
—
4.1
0.36
1.4
26
2.0
—
—
CALC- NON-EQ
STIRRED REACTOR
4
19
42
-
-
75
14
35
47
42
—
~
1 NO MOLE FRACTION! CALCULATED FROM
NO MOLE FRACTION - CNO/
P(MEASURED)
RT(MEASURED)
-------
and measurement location in Fig. l6a and as a function of inlet velocity in
Fig. l6b. The vertical bars through the symbols indicate the uncertainty
in the concentration measurement resulting from uncertainties in the cali-
bration, in determination of the mean transmissivity (see Appendix A) in
the combustion experiment and in the measured temperature used to calculate
total nitric oxide concentration from the measured ground-state concentra-
tion. The experimental data in Fig. l6a indicate that the nitric oxide con-
centration is nearly constant throughout the recirculation zone. However,
the nitric oxide concentrations measured at the downstream axial position
appear somewhat lower than the concentrations measured at the upstream
position. In an investigation of concentration distributions in the recir-
culation zone of bluff-body-stabillzed propane-air flames (Ref. 14), COg,
HO, and 0 concentrations were found to be uniform throughout the zone.
Tne nitric oxide concentration in the recirculation zone increases as the
equivalence ratio increases from 0.8 to 1.1. The experimental data in
Fig. l6b show that the nitric oxide concentration in the recirculation zone
decreases as the inlet velocity increases from 20 to 60 m/sec. The solid
lines in Figs. l6a and l6b are calculated nitric oxide concentrations obtained
from various analytical models for the nitric oxide formation process in
the recirculation zone. These models are discussed in detail in Section
V-A.
III-E.3 Exhaust Concentration Measurements
The concentration distributions of nitric oxide, methane, oxygen, carbon
dioxide and carbon monoxide in the exhaust gas from the combustion test
section were determined using a traversing sampling probe coupled to an
on-line time-of-flight mass spectrometer. A schematic diagram of the
sampling system and associated data processing equipment is shown in
Appendix A.
Typical species concentration profiles in the exhaust, measured at the
exit plane of the combustor test section, are shown in Figs. IJa and IJb.
Fig. ITa shows the concentration profiles of methane, oxygen and carbon
dioxide. Fig. l?b shows the nitric oxide concentration profiles. These
data show that there are substantial concentration variations in the
exhaust and that there is a significant amount of unreacted fuel at the
test section exit. An average exhaust nitric oxide concentration can
be determined from the concentration profile in Fig. IJb. Some uncertainty
is introduced in the determination of the average nitric oxide concentra-
tion by the lack of concentration data near the test section walls.
Average exhaust nitric oxide concentrations are tabulated in Table VIII for
a range of inlet velocities and equivalence ratios.
-------
M/SEC
120
D.
Q.
g
o
cc.
u.
til
O
LLJ
Q
X
o
o
cc
100
80
60
40
20
I POSITION 1
POSITION 4
NON-EQUILIBRIUM
STIRRED REACTOR
EQUILIBRIUM STIRRED REACTOR
1.1
1.2
EQUIVALENCE RATIO
FIG. 16a TYPICAL DEPENDENCE OF NITRIC OXIDE CONCENTRATION IN THE
RECIRCULATION ZONE ON EQUIVALENCE RATIO AND
MEASUREMENT LOCATION.
-------
' = 1.10
120
Q_
CL
o
<
DC
LLJ
Q
X
o
o
cc
100
80
60
40
20
FIG.16b
NON-EQUILIBRIUM
STIRRED REACTOR
EQUILIBRIUM-STIRRED
REACTOR-
20
40
60
INLET VELOCITY - M/SEC
TYPICAL DEPENDENCE OF NITRIC OXIDE CONCENTRATION
IN THE RECIRCULATION ZONE ON INLET VELOCITY.
-------
U00=21.7 M/SEC 0=0.88
0.16
0.14
0.12
z 0.10
g
o
QC
LL
J 0.08
to
LU
O
0.06
0.04
0.02
0.6 0.4
0.2 Q. 0.2
POSITION -CM
0.4 0.6
FIG.17a
TYPICAL SPECIES CONCENTRATION PROFILES AT THE EXHAUST OF THE
COMBUSTOR TEST SECTION.
-------
160
140
120
to
<
CO
cc
Q
Q_
Q.
I
z
o
EC
Li.
LU
Q
X
O
o
tr
100
80
60
40
20
0.6
1)^=21.7 M/SEC
0=0.88
J
MASS SPECTROMETER DATA
UV ABSORPTION DATA
I
0.4
0.2 Q. 0.2
POSITION -CM
0.4
0.6
FIG. 17 b TYPICAL NITRIC OXIDE CONCENTRATION PROFILES AT THE EXHAUST
OF THE COMBUSTOR TEST SECTION.
-------
TABLE VIM
NITRIC OXIDE CONCENTRATIONS IN EXHAUST GAS
DOO (m/sec)
21.4
21.2
21.5
21.6
41.0
41.1
0.81
0.88
1.00
1.10
0.88
1.01
AV. NO MOLE FRACTION - PPM (DRY)
60+15
90+10
128 + 10
49+15
48+15
62+ 15
-------
The experimental results, discussed above, can be used to assess the
relative importance of nitric oxide production in the recirculation zone
to overall nitric oxide production in the burner. The mean volumetric
nitric oxide production rate in the recirculation zone is given by
where (m /V)w = mean volumetric nitric oxide production rate in
the recirculation zone (Sm/cm3_sec), CTOO = mean nitric oxide concen-
tration in the zone (mole/cm3); M = molecular weight of nitric oxide
(gm/mole) and T, = mean residence time of the gas in the zone (sec).
A mean volumetric nitric oxide production rate in the burner may be
expressed
where yWQ = mean nitric mole fraction at the burner exhaust, M =
mean molecular weight of the exhaust gas (gm/mole), m = total
volumetric nitric oxide production rates in the zone are compared
with overall nitric oxide production rates in Table IX. For the
present experimental configuration, nitric oxide production in the
recirculation zone is not a major factor in overall nitric oxide
formation. This situation is a direct result of the relatively low
temperature associated with the recirculation zone downstream from
the cylindrical f lameholder .
-------
TABLE IX
RATIO OF NITRIC OXIDE PRODUCTION IN THE RECIRCULATION ZONE
TO TOTAL NITRIC OXIDE PRODUCTION
21.4
21.4
21.4
41.0
41.0
-------
SECTION IV
COLD FLOW EXPERIMENTS
IV-A. TEST EQUIPMENT
IV-A.I. Test Facility
The investigation of the characteristics of the two-dimensional recir-
culation zone downstream of a cylindrical bluff-body was conducted at the
UARL fluid dynamics test facility (Fig. IS). The facility consists of
a nominal 10 x 10 cm cross-section test chamber in which a two-dimen-
sional cylinder, 1.59 cm in diameter, is installed with its axis normal
to the air flow. Variation of velocity in the range 20-75 m/sec is
effected by throttling the vacuum exhauster system connected to the
facility. Density levels from 1 atm (1.2 kg/m3) to 1/3 atm (0.4l kg/m3)
are established by installing suitable throttling plates at the facility
inlet to restrict air weight flow.
The cylindrical bluff-body is composed of an outer shell which contains
39 injector ports distributed in three rows at angles of 135, 180 and
225 deg measured from the leading edge, to facilitate injection of a
suitable tracer gas. Each injector port is 0.132 cm in diameter.
Internal to the (primary) cylindrical shell is a second concentric
cylindrical shell containing matching injector ports. The internal
cylindrical shell is closely machined to fit the internal diameter
of the primary cylindrical shell. In the normal open position injector
ports in both cylindrical shells are aligned, thereby allowing unim-
peded tracer flow through the shell ports into the recirculation zone.
Rotation of the inner cylinder 20 deg seals off all ports in the outer
shell, thereby stopping all tracer flow. Closure is accomplished
through the use of a high speed pneumatic cylinder which has been
calibrated for a closure time of approximately 1 msec. The cylinder
body is traversable as a unit in the transverse (z) direction to allow
the acquisition of data at locations removed from the test section
center-plane (Fig. 19)-
An assembly for installation of screens is incorporated in the design
of the test facility at a location 1.25 cm upstream of the cylinder
centerline, as shown in Fig. 19- Screens with various mesh spacing
to wire diameter ratios are employed to vary the level of initial free
51
-------
INJECTOR BODY
TRAVERSING MECHANISM _
-at - - -
IMPACT/HOT WlR E PROBE
TRAVERSING MECHANISM
ro
EXHAUSTERS
FIG. 18
COLD FLOW FLUID DYNAMICS TEST FACILITY
-------
a) TOP VIEW
-CYLINDER
-FIBER OPTIC PROBE
5
AIR FLOW 4
41
4
t
t
1
/ /FIBER OPTIC PROBE AT / M ' "™'u"" "•=«""«"" — ' "-
/ EXTREME FORWARD POSITION^
1
1
0.32 cm 1
-1— 1
' 1
1
cm
1 I
b) SIDE VIEW
INTERCHANGEABLE
SCREEN ASSEMBLY
CYLINDERICAL INJECTOR BODY
AT MAXIMUM TRANSVERSE POSITION
CYLINDRICAL INJECTOR BODY
AT TUNNEL C_ (NEUTRAL POSITION)
UPPER WALL
AIR FLOW
RECIRCULATION
ZONE '
10cm IMPACT/HOTWIRE
PROBE (TRAVERSING)
1 25 cm
1-0.25 cm
-9 4 cm-
LOWER WALL
-FIBER OPTIC PROBE AT <_FIBER OPTIC PROBE AT
EXTREME FORWARD POSITION MAXIMUM REARWARD POSITION
FIG. 19 SCHEMATIC DIAGRAM OF COLD FLOW FLUID DYNAMICS TEST SECTION
53
-------
stream turbulence in the flow. The mesh spacings and wire diameters
of the screens selected for use in this investigation, and the associated
turbulent intensities measured at two locations downstream of each screen
are given in Table X . Note that intensities vary from a high value
of about 11 percent to a low of 3 percent at location "1". This loca-
tion directly downstream of the screen is not significantly influenced
by either the body or the confining walls and, therefore, measured
turbulence levels can be characterized in terms of screen geometrical
parameters. Measurements at a location near the end of the near-wake
region will be discussed later.
Tracer concentration distributions and residence times are determined
using fiber-optic probes (Fig. 20) which consist of a pair of sheathed
bundles of glass fibers. One bundle of fibers serves as a transmitter
and the other as a detector of visible light. The probe is integrated
into a photometer system containing appropriate optics and electronic
readout devices. Spacing between the probes is maintained at a value
of 0.32 cm. The fiber optic probes are installed in a traversing
mechanism actuated by a variable-speed electric motor to allow measure-
ments to be taken throughout the near-wake region.
The tracer material used for the optical investigation consisted of
an equilibrium mixture of N02 and MjsO^ (see Appendix B). Initially,
NH>Cl particles were investigated as a possible tracer but the tendency
of this material to coat the surfaces of the fiber optic probe eliminated
it from further consideration.
IV-A.2. Facility Calibration and Operating Characteristics
Flow field uniformity in both the transverse (z) and axial (x) directions
were ascertained from measurements of velocity profiles at several test
chamber axial locations. Velocity profile information acquired at three
approach velocities without screens for poo = 1.2 kg/m3 (Fig. 21) are
typical of these data. Note that the outer, inviscid flow is quite
uniform throughout the near-wake region. Velocity profiles taken with
the turbulence-producing screens installed displayed a comparable degree
of uniformity.
Turbulent intensities were determined using a constant temperature hot-
wire anemometer (Fig.22). Hot-wire measurements were gathered at two
locations: (1) x = 0.2 cm, z = 2.5 cm; and (2) x = 3.k cm, z = 2.5 cm.
-------
TABLE X
SCREENS FOR TURBULENCE PRODUCTION
SCREEN
1
2
3
4
S (CM)
1 27
1 27
1 27
1 27
h (CM)
0 203
0 160
0 124
0 104
S'h
625
7 93
10 2
12 2
U^IM/SEC)
19 5
575
195
57 E
195
57.5
195
57 5
TURBULENT INTENSITIES (%)
POSITION "1"
u'/Uj-
9
11
7
5
5
3
v'/ucr
11
10
6
6
6
4
"2"
u'/U,,,,
12
9
11
10
10
»'/UCX)
11
10
12
11
11
-------
TRANSMITTER
FLEXIBLE
LIGHT GUIDE
\
.ADAPTER FOR
LIGHT SOURCE OUTPUT
FIBER OPTIC PROBE
FLEXIBLE
LIGHT GUIDE
ADAPTER FOR '
PHOTOMULTIPLIER TUBE INPUT
FIG 20
FIBER OPTIC PROBE
56
-------
Poo= 1.2 kg/m3rNO SCREENS
a) Uoo= 18.5 m/sec
4
3
2
1
(
- -it--- —
-
y — CYLINDRICAL
,^X BLUFF BODY
li
*
—
1 U
x = 3.4 cm
)
^
1 1 1
3 0 20 40 " 0 20 40 60 8C
E VELOCITY, u - m/sec
I b) Doc = 39 m/sec
N
LJ 4
2
CENTERL
CO
0 2
U
CO
fc 1
Ul
I-
1 °
LJ_
A
*^-
-
••ffXi\ 1
x = 0 V
**-
•*
1 1 A
0 0 20 40 V
x = 3 4 cm
J
^^^
1 1
0 20 40 60 8t
^ VELOCITY, u - m/sec
CO
D
4
3
2
1
0
c) U.,0 = 57.5 m/sec
/
*^-
^
x = 0,V
*— -
1 1 IA
x = 3 4 cm
1 1 1
SO 0 20 40 60 "O 20 40 60 8
VELOCITY, u - m/sec
•it REGION WHERE PITOT PRESSURE IS LESS THAN STATIC PRESSURE
FIG. 2 1 TYPICAL VELOCITY PROFILES IN COLD FLOW TEST SECTION
57
-------
w • - :- '.-a*- «-,-'• . •• ^
FIG. 22 COLD FLOW TEST SECTION WITH HOT WIRE PROBE INSTALLED
58
-------
Tests were conducted at the two approach flow velocities 19-5 and 57-5
m/sec for each of four screens, which characteristics are given in Table
X .
Mean wire voltages and velocity-component autocorrelations are recorded
from tungsten wire sensors. The 0.0087 cm diameter, 0.25 cm long
wires are mounted in an x-array for operation in a constant temperature
mode using two DISA 55D01 constant temperature anemometers. It was
assumed in the data reduction that the flow was incompressible with
negligible temperature fluctuations.
Turbulence measurements at the two test locations were influenced to
varying degrees by the vortex shedding phenomenon. Data obtained at
position "1" are in good agreement with measurements reported in Ref.
21, as indicated in Fig. 23. Since the Ref. 21 data were obtained in
the absence of outside influences, it may be concluded that the posi-
tion "1" data are not affected significantly by the presence of the
cylindrical bluff-body. At location "2", the time variant nature of
the flow is clearly dominated by the periodic vortex shedding. The
turbulence intensity at position "2" is found to be at the high levels
of 9-12 percent independent of which screen is installed. Furthermore,
the measured frequencies of the fluctuations at position "2" are in
agreement with those computed from the expected Strouhal number (Ref.
5), using the body diameter and the velocity at the bluff-body loca-
tion, ie., free stream velocity adjusted by the blockage ratio.
IV-A.3. Instrumentation
IV-A.Si. Fiber Optic System - Quantitative data acquisition in the
cold flow investigation was accomplished using a fiber optic probe
system (Appendix B) consisting of: (l) a fiber optic probe, (2)
an optical photometer system, and (3) data acquisition and display
electronics. The two identical units comprising the fiber optic probe,
(Fig. 20) designated the transmitter and the detector unit, con-
tain approximately 50 drawn glass fibers in a 0.25 cm diameter bundle
encased within a stainless steel sheath. The probe transmits visible
light above a wavelength of approximately ^OOOA.
IV-A.3JJ. Flow Visualization - Flow field characteristics were observed
using the laboratory schlieren system. In this effort, helium was
utilized as the tracer medium to establish large density gradients.
Recirculation zone geometrical parameters were determined from spark
schlieren photographs. In addition, high speed (3,000 - 5,°00 frames
59
-------
2.3cm
FLOW
•s" ©
—- T* x
•
•r\
© 1
x L_
2.5 cm
—I K- 1
1.25 cm
SYMBOL
A
D
0
t>
S, cm
b, cm
1.25 0.203
1.25 0.160
1.25 0.125
1.25 0.104
U00= 19.5 rn/sec-OPEN
DOJ= 57.5 m/sec - CLOSED
01 Z
CJ UJ
Z O
UJ CC
_l LU
D Q-
m |
LU
OC
t-
C/3
LU
LU
QC
10 20
DISTANCE DOWNSTREAM OF
100
200
FIG. 23
TURBULENCE LEVEL DOWNSTREAM OF SCREENS.
60
-------
per second) strobographic movies were taken at selected test conditions
to facilitate study of the vortex shedding phenomenon.
IV-B. DISCUSSION OF RESULTS
The test matrix for the cold flow experiments, summarizing measurements
made at each combination of free stream velocity, free stream density
and screen, is contained in Table XI. Four types of test measurements
were made including free stream pitot pressure distributions, recircula-
tion zone geometrical parameters, and tracer concentration distribu-
tions and residence times in the recirculation zone. The matrix was
evolved to insure that sufficient measurements were made to clearly
define the variations of these major dependent variables.
Figure 2U is typical of the spark schlieren photographs obtained in
the near-wake region using a helium tracer gas. The photograph indi-
cates two clearly defined regions. Immediately downstream of the body,
extending for a length of approximately two and one-half bluff-body
diameters, is a region which encompasses the recirculation zone. The
intersection of this region with the center plane of the flow can for
all practical purposes be taken as the rear stagnation (closure) point,
"r". A vortex which is being shed from the near-wake region also is
distinguishable in the photograph of Fig. 2^. The vortex shedding process
results in significant local agitation of the flow and a marked distor-
tion of that half of the recirculation zone from which the vortex issues.
These vortex shedding effects are more discernable in the sequence of
photographs shown in Fig. 25. Wo such vortex shedding effects were
observed in the combustion experiments which were extracted from a
high-speed schlieren motion picture of the flow. The recirculation
zone length measured from the trailing edge of the cylinder to the
closure point, and the maximum zone width, determined by doubling the
maximum half-width of the undisturbed portion of the recirculation zone,
are given in Fig. 2o. The independent variable in this figure is the
velocity at the bluff-body location computed from the approach velocity
and the blockage ratio. As can be seen, within the scatter of the
photographic data, no important changes in recirculation zone geometry
occur for the conditions of the present study.
The increase in photometer system light transmission with time following
the cessation of tracer material injection is shown for a typical case
in the Fig. 2Ja oscilloscope trace. Information extracted from such
photographs is used to determine relative tracer concentration varia-
61
-------
TABLE XI
COLD FLOW TEST MATRIX
SCREE. N
(SfcE 1AB1 b VliU
NONE
'
2
1
4
Pec
UM
Rp
1 ? kg/rn3
24 0 m/ser
?45x 104
TCFR
TCFR
TCFR
TCFR
TCFR
48 0 m/vc
4 76 x 104
TCFR
TCFR
TCFR
TCFR
TCFR
72 0 m/sec
7 03 x 104
TCFR
TCFR
TCFR
TCFR
TCFR
067 kq/m3
24 0 m/sec
1 36 x 1 O4
FR
R
R
48 Om/SPC
2 6b x 1 04
CFR
Cfi
CR
CR
CR
72 0 m/sec
3 86 x 1 O4
FR
R
R
0 41 kq/m3
24 Om'sec
031 x 104
P
R
480 m/sec
1 63 x 1C4
CFR
CR
CR
CR
CR
72 Om-stc
242x 104
FR
R
p
T PITOT PROBE TRAVERSES
F FLOW VISUALIZATION
C CONCENTRATION DISTRIBUTION
R RESIDENCE TIME
62
-------
U» = 24 m/sec, NO SCREENS
AIR FLOW
234
AXIAL DISTANCE FROM CYLINDER, x/D
FIG. 24
SCHLIEREN PHOTOGRAPH OF NEAR-WAKE REGION
63
-------
3000 FRAMES/SEC
T=0
FIG.25
VORTEX SHEDDING SEQUENCE.
-------
O
N
Z
O
C/3
= 1.2 kg/m3
x
Q
LU
z
o
D
5
X
5
Q
Z
Q
L/D
2.3
O
_L
I
20
40 60
VELOCITY, U0 -m/sec
80
FIG. 26
RECIRCULATION ZONE GEOMETRICAL CHARACTERISTICS
-------
tions with time by applying the Beer-Lambert Law (Ref. 22), as given
in Appendix B, ie.,
(13)
where the tented parameters are reference quantities. In agreement with
the results from Refs. 11 and 16, the relative tracer concentration is
found to decrease in an exponential fashion with time until most of the
tracer material has been transported out of the recirculation zone
(Fig. 27b). The time required to reduce the tracer concentration to
1/e of its original concentration (the e-fold point) is taken as the
tracer residence time, T .
u
All residence time data acquired during the test program is provided in
Table XIL Also shown is a residence time parameter, (T-j-Uo/DjTy , which
according to the analytical development of Section II, should provide
a correlation of the data. Note that the residence time correlation
parameter varies within very narrow limits. This high degree of correla-
tion is indicated more clearly in the Fig.28 presentation of typical
results. No effect of free stream density is apparent indicating that
the correct representation of p involves a direct dependence on v'llco .
Furthermore, no trend is found with free stream turbulence level as
augmented by screens placed immediately upstream of the bluff-body.
In fact, a value of the correlating parameter of 2 .k ± 0.3 is appro-
priate for more than 90 percent of the data gathered, independent of
the conditions of the free stream. This successful residence time
correlation supports the view that body-related turbulent transport
is the dominant mechanism for the near-wake region in the sub-critical
Reynolds number regime.
It should be remarked that this finding is somewhat at variance with
the work of Bovina (Ref. 16). In the cited work, T^U /D was found
to be constant and independent of shedding period, in both cold and
reacting flow experiments with two-dimensional V-shaped flameholders;
these flameholders were 2 to 6 cm in size, with a 30 deg apex angle.
Evidently, in these experiments, the transport mechanism was inde-
pendent of whether or not a flame was stabilized on the bluff-body,
suggesting that body-related turbulence for V-gutters (assuming vortex
shedding occurs in cold flow) at Reynolds numbers of approximately
*•, is not significant by comparison with turbulence produced from
66
-------
= 72 m/sec , pm= 1 2 kg/m3, SCREEN 2
al OSCILLOSCOPE TRACE
O
Z
O
TIME r- 10msec/DIV
bl SEMI-LOG PLOT OF DATA
05
04
LLAG ASSOCIATED WITH CLOSURE AND
MEASUREMENT STATION LOCATION
20 30
TIME T-msec
RESIDENCE TIME DETERMINATION
67
-------
TABLE XII
RESIDENCE TIMES AND CORRELATION PARAMETER
SCREEN
(SEE TABLE VIM}
'
'
3
4
Pj. KCi/M3
IV M'SEC
Rp
7, MSEC
(I^05SEC05
r,
(Il^K05
\ D /
T
L ''
l'^\. Ob
ID y '.
7
(£K"
7,
(^)'.»
24 n
2 45* 104
24
23
23
23
25
? 4
28
2 7
29
2 8
1 2
480
4 76xl04
17
24
18
25
17
2 4
1 7
„
r ~
18
25
72 n
7 03x1 04
16
27
15
25
14
2 4
15
26
,.
2 7
?4 0
1 36xio4
22
22
26
25
21
2 1
1 -
0 67
480
2 66x1 04
16
22
15
20
15
20
15
20
1 7
2 4
72 0
386x104
14
24
14
2 4
13
22
240
OSlxlO4
26
25
26
25
-
24
2 2
0 41
48 0
163*104
16
2 2
15
2 0
15
2 1
1 4
1 9
720
242*104
15
26
13
? 1
12
20
NOTE U U J 1 BRI
-------
in
o^
o
o
cr
LU
cc
<
Q_
LU
DC
CC
O
O
LU
LU
O
z
LU
Q
C/3
O Pa,- 1.2 kg/m3
n^co = 0.67 kg/m3
A p_ = 0.41 kg/m3
a) NO SCREENS
3
2
1
0
(
b) SCF
3
2
1
0
c) SCF
3
2
1
0
o 1 r ° °
DD "a " J
i I I
) 2.5 5 7.5 1C
1EEN#1
o a n o o ^
i i i
3 2.5 5 7.5 1
}EEN#3
O
O
0 rn 0 "
_ a
-------
the interaction between the outer flow and the wake .
Tracer concentration distributions in the two-dimensional recirculation
zone were obtained by taking axial traverses with the fiber optic probe
at each of five transverse positions of the bluff-body. At each trans-
verse bluff -body position, axial traverses were made both with and
without tracer injection, see Fig. 29. Thus, at any given axial posi-
tion, the output signal with tracer injection, I, can be normalized
with the output signal obtained with no tracer, Io.
Since the absolute magnitude of the tracer concentration in the recir-
culation zone is not particularly meaningful, it is convenient to pro-
vide a normalization with a reference concentration, y^.. For the pur-
poses of the present effort, the reference concentration was taken to
be that concentration found at the test section center plane (z = 0),
1.2 cm downstream of the trailing edge of the bluff- body. This loca-
tion was selected since the recirculation zone and the outer flow
were found to be at the same temperature (to within 1°K - the expected
experimental accuracy) downstream of this location, see Fig. 30.
All axial tracer concentration distributions measured displayed the
features indicated in Fig. 31. Initially, there is a gradual reduc-
tion in concentration to a point approximately one diameter in length
downstream of the bluff-body.
Thereafter, the concentration decreases rapidly with distance according
to
du)
where 8, the decay exponent, is found to vary between -1.0 and -1.5
with most data indicating a value of approximately -1-3 (Fig. ). An
obvious change in the decay rate occurs at distances from the bluff-
body of 2 to 3 diameters (Fig. 32) which correspond approximately to
the longitudinal extent of the recirculation zone. Tracer concentra-
tion data in terms of the decay slope, 0, and the location at which
the slope abruptly changes, x / are given in Table Mil.
The distribution of tracer material throughout the recirculation zone
is displayed in the contour map of Fig. 33- It is significant that the
transverse distributions of tracer material are quite flat. Evidently,
TO
-------
Uoo= 72 m/sec,
kg/m3, SCREEN #4
I0 = 0.809 V
UPPER CURVE: NO TRACER
AXIAL TRAVERSE
LOWER CURVE: TRAVERSE
WITH TRACER INJECTION
XPHOTO
AXIAL DISTANCE - 10 cm/DIV
FIG. 29
OSCILLOSCOPE TRACE OF TYPICAL TRACER CONCENTRATION DISTRIBUTION
-------
= 310°K
8
u
z
LU
cc
UJ
u_
u_
O
LU
cc
cc
UJ
Qu
2
UJ
H
LU
z
O
N
CENTERLINE
CENTERLINE
O
OFF AXIS
-x = 1.2 cm-
FIG. 30
1.0 2.0
AXIAL DISTANCE FROM CYLINDER, x/D
AXIAL VARIATION OF TEMPERATURE IN RECIRCULATION ZONE
72
-------
Uoo= 72 m/sec, Poo= 1.2 kg/m3, SCREEN #4
i.o
g
H
H 0.5
0.2
01
°-1 02 0.5 1.0 2
DISTANCE FROM CYCLINDER, x/D
FIG 31 TYPICAL TRACER CENTERLINE CONCENTRATION DISTRIBUTION
10
73
-------
Vt
i xL)
O P00= 1.2 kg/m3
D A* = 0.67 kg/m3
0 P^ = 0.41 kg/m3
OPEN SYMBOLS: XL/D
CLOSED SYMBOLS: -Q
3.0i
CD
I
2.0
1.0
L/D
RANGE FROM
SCHLIEREN
PHOTOGRAPHS]
O
0
I
I
•
!
I
20 40 60
FREE-STREAM VELOCITY AT x = 0, Uo-m/sec
80
FIG. 32
TRACER CONCENTRATION DISTRIBUTION ON RECIRCULATION ZONE CENTERLINE
-------
= 24.0m/sec, pJ== 1 2 kg/m3f SCREEN #1
. APPROXIMATE BOUNDARY
OF RECIRCULATION ZONE
V,/V, - 1 0
« 05 1.0 1.5
AXIAL DISTANCE FROM CYLINDER, x/O
3.5
TRACER CONCENTRATION CONTOUR MAP
75
-------
TABLE XIII
TRACER CONCENTRATION DISTRIBUTION DATA
t>t RFEN
(SEE TABLE X)
NONE
1
2
3
4
TRANSVERSE
POSITION
p.f - kq/m3
Uco M/SEC
R(-OT
fl
*L/D
0
*L'D
0
XL,D
0
"L/D
1)
KL/D
CENTERLINE
12
24 0
2 4S*104
1 29
? 1
1 24
3 0
1 29
2 2
-
1 3b
2 4
48 0
4 76xl04
1 3
20
1 0
3 0
- 1 05
2 1
1 36
2 2
1 J2
? 3
720
7 03* 1O4
- 1 1 7
30
-1 39
2 4
-
-
1 13
2 8
067
48 0
26Gx104
-1 09
25
-1 23
~
1 37
1 37
3 0
0 41
480
1 63x104
- 1 52
2 7
-
-- 1 32
30
-
Z
-------
the motions within the recirculation zones operate to minimize trans-
verse gradients and, therefore, a conceptually useful model of a one-
dimensional recirculation zone might be appropriate.
It should be noted that the nature of the tracer concentration distri-
butions found in the cold flow study reflect the absence of tracer in
the outer flow. When a flame is anchored to the bluff-body, by con-
trast, chemical species including NO, and heat can move from the flame
into the recirculation zone. Caution should be exercised, therefore,
in applying the cold flow concentration data to the treatment of
reacting flows.
77
-------
SECTION V
ANALYTICAL INVESTIGATION
To assist in the interpretation of the results obtained from the com-
bustion experiments (Section III) and to permit extrapolation of these
results to conditions outside the range of the experimental investiga-
tion, an analytical model for nitric oxide formation in the combustor
was developed. This analytical model is comprised of two separate
parts: (l) a model for nitric oxide formation in the recirculation
zone downstream from the flameholder and (2) a model for nitric oxide
formation in the downstream flame as it spreads into the flow. In
the following sections, the details of the two analytical models are
presented and a comparison of the theoretical predictions with experi-
mental results is made. Based on this comparison, certain limitations
of the analytical models are apparent. These limitations are discussed
and several recommendations for improved combustor modeling techniques
are made.
V-A. RECIRCULATION ZONE MODEL FOR NITRIC OXIDE FORMATION
The experimental results obtained in the combustion experiments indicate
that the gas temperature and species concentrations are essentially
uniform throughout the recirculation zone. In the cold flow experiments,
the variation of tracer concentration in the transverse direction in
the recirculation zone was small, indicating that turbulent transport
within the zone was moderately rapid. However, the observed axial decay
of tracer concentration in the zone suggests that internal mixing rates
are not sufficiently rapid to offset transport of tracer material out
of the recirculation zone. In the combustion experiments, the observed
recirculation zone temperature was found to be approximately lOfo lower
than the adiabatic combustion temperature for the free stream equiva-
lence ratio. Furthermore, in both the combustion experiments and the
cold flow experiments, the mass exchange rate between the recircula-
tion zone and the outer flow could be characterized by a mean residence
time. In the combustion experiments, the wake downstream from the
flameholder was steady and did not exhibit vortex-shedding, as was
observed in the cold flow experiments. All of the experimental observa-
tions outlined above are consistent with a model in which the recircula-
tion zone, in the combustion experiments, is assumed to be a steady-
state well-stirred reactor. The well-stirred reactor concept has been
78
-------
applied to model the recirculation zone In an investigation of flame
stabilization on bluff-body flameholders (Ref. 23). In the following
section, a stirred reactor model for the recirculation zone is developed
based on the analytical approach of Jones and Prothero (Ref. 2k).
V-A.l. Stirred Reactor Model for the Recirculation Zone
It is assumed that the recirculation zone is an adiabatic, well-stirred
Q '
reactor of volume V(cnr), containing combustion gas at pressure P(atm),
density p (gm/cm3) and uniform temperature l(°K). The overall flow rate
of gas through the reactor is m(gm/sec) and the concentration of species
Sj_ is denoted by (7-°(mole/gm) on input and a^(mole/gm) on output.
m
The reaction mechanism for the stirred reactor may be written in the
general form
where v. ., v.". are integers representing the stoichiometric coefficients
of species Sj_ in reaction j.
In steady-state operation, the governing equations for the stirred
reactor may be written as follows:
Conservation of Mass
The mass conservation equation for each species Sj_ is given by
V . M + J +
where R.. and R_ ^ denote the overall forward and reverse reaction rates
of reaction j. Eq. 16 states that the difference between the input
and output flow rates of species S^ is equal to the net rate of produc-
tion of species Sj_ by chemical reaction.
The overall reaction rates in Eq. 16 may be written:
79
-------
where k., k_ * are the forward and reverse rate constants for reaction
j. The general functional form of the rate constant k is:
k = A T1^ exp (E/RT)
Conservation of Energy
Conservation of energy for an adiabatic stirred reactor may be
expressed by
where H.(T°) and Ej_(l) are temperature -de pendent total molar enthalpies
(cal/mole) (sensible enthalpy + chemical enthalpy) of species S^ in the
input and output streams, respectively.
Equation of State
The gas in the stirred reactor is assumed to obey the perfect
gas law,
P = ^RT°-m (18)
where R is the universal gas constant (cm3 • atm/mole • °K) and
°m =
Eqs . l6 - 18 , together with appropriate values for the reactor input
conditions, can be used to determine the composition, temperature and
density of the gas in the reactor. These equations were programmed for
computer solution using a Wewton-Raphson iteration technique (Ref. 2.k ) .
Use of reactor temperature as a dependent variable led to convergence
difficulties; hence, it was necessary to carry out the calculations for
a specified reactor temperature. In the computations discussed below,
reactor gas compositions and densities were calculated for reactor tempera-
tures encompassing the range of measured recirculation zone temperatures.
80
-------
V-A.2 Calculation of Mitric Oxide Concentration in the Re circula-
tion Zone
The stirred reactor model outlined in Section V-A.l has been used to
calculate the nitric oxide concentration in the recirculation zone for
the conditions of the combustion experiments. In making these calcula
tions, it is necessary to specify: (a) a reaction mechanism, (b) the
reactor input conditions, (c) the reactor temperature and (d) the mean
residence time of the gas in the reactor.
Reaction Mechanism
To calculate the nitric oxide concentration in the recirculation
zone, two reaction mechanisms must be specified: (1) a nitric oxide
formation mechanism and (2) a methane combustion mechanism. In the
present investigation, it is assumed that nitric oxide is formed via
an extended Zeldovich mechanism (Refs. 25 and 26),
0 + W2 jj. NO + N (19)
k-19
/ x
N + 0p j± NO + 0 (20)
k-20
N + OH _ NO + H (2l)
The forward and reverse rate constants for Reactions 19 and 20 were
taken from Ref. 27, and the forward rate constant for Reaction 21 was
taken from Ref. 19 . These rate constants are tabulated below
k]_ = 1.36 x 10 ^ exp(-75AOO/RT) cm3/mole • sec
k_-L = 3.10 x lO1^ exp(-334/RT) cm3/mole • sec
kg = 6A3 x 109 T exp(-6250/RT) cm3/mole • sec
81
-------
k 2 = 1.55 x 1C)9 T exp(-38,61j-0/RT) cm3/mole • sec
ko = it-.O x 10 ^ cmS/mole • sec
where R = universal gas constant = 1.986 cal/mole°K and T is temperature
The nitric oxide formation process in the recirculation zone is coupled
to the hydrocarbon chemistry. Two different degrees of coupling were
considered in the present investigation. Initially, it was assumed
that the nitric oxide formation process was decoupled from the hydro-
carbon chemistry, and that the C-H-0 chemistry was equilibrated prior
to the onset of nitric oxide formation. In the context of this mechanism,
the concentrations of C-H-0 species were assumed to be in equilibrium
at the measured recirculation temperature. As noted in Section III,
the reduced recirculation zone temperature could not be explained ia
terms of heat transfer to the flameholder and test section walls.
Longwell e_t al. (Ref . 23) have measured unreacted 0^ in the recircula-
tion zone of bluff-body stabilized flames. In the present experiments,
substantial departures from equilibrium were observed for radical con-
centrations in the recirculation zone. Hence, it seems reasonable to
attribute the reduced recirculation zone temperature to unreacted
fuel in the zone. The equilibrium calculations were carried out assuming
adiabatic conditions with an amount of fuel unreacted so that the equilib-
rium temperature was equal to the measured recirculation zone tempera-
ture. The concentrations of 0, OH and 0%. obtained from this equilibrium
calculation were used, together with Eqs . 19 - 21 , to calculate the
concentration of N and NO in the recirculation zone.
In the combustion experiments, measured OH-radical concentrations in
the recirculation zone were significantly in excess of the concentra-
tions calculated for a stirred reactor, assuming C-H-0 equilibrium,
Fig. 15. This observation indicates that the nitric oxide formation
process in the recirculation zone may be closely coupled to the hydro-
carbon chemistry. To model the case of coupled nitric oxide formation
and hydrocarbon chemistry, the stirred reactor calculations were carried
out using Eqs. 19 21 , together with the measured OH concentration
and inferred 0-atom concentration. Several investigators (Refs. 2.Q, 29 )
have shown that various rapid bimolecular reactions involving radical
species equilibrate early in the combustion process, prior to attain-
ment of total equilibrium. The partial equilibration of these bimolecular
reactions result in the concentrations of various radical species being
82
-------
inter-related. In the present study, the 0-atom concentration was cal-
culated from the measured OH concentration by invoking the partial
equilibrium approximation for the rapid bimolecular reaction
OH + OH ^ 0 + I£0 (22)
The 0-atom concentration is given by
CQ = 2
where K (l) is the equilibrium constant for Reaction 22 at the measured
recirculation zone temperature. For the conditions of the present
investigation, no significant error in the calculation of the 0-atom
concentration is introduced by using the equilibrium water concentra-
tion.
Reactor Input Conditions
The input conditions which must be specified for the stirred reactor
calculations are the temperature and species concentrations in the input
stream. The composition of the input stream is some mixture of burned
and unburned gas, containing some of the radicals and nitric oxide formed
in the flame zone surrounding the recirculation zone. The composition
of the input stream cannot be well-defined since the state of the gas
transported into the zone from the outer flow is not known. Hence, in
the present calculation, the input concentration of all species except
OH, 0, N and WO were set equal to the equilibrium values associated
with the recirculation zone temperature. The input concentrations of
OH and 0 were set equal to their equilibrium values in the uncoupled
stirred reactor model and to their measured (or inferred) non-equilibrium
values in the coupled model. For all of the calculations, the input
concentrations of N and NO were set equal to zero.
Reactor Temperature
As noted earlier, use of the reactor temperature as a dependent
variable led to convergence difficulties in the stirred reactor numerical
calculations; hence, in these calculations the reactor temperature was
specified equal to the measured recirculation zone temperature. To
determine the sensitivity of the predicted nitric oxide concentrations
in the recirculation zone, the reactor temperature was varied ± 100°K
83
-------
from the recirculation zone temperature.
Mean Residence Time
To obtain a solution to the governing equations for the stirred
reactor, a value for the parameter m/V must be specified. For a well-
stirred reactor, the parameter m/v may be expressed as, (Ref. 30)
m rt
~ = P
where r is the mean residence time of gas elements Ln the reactor and
pis an appropriate gas density. In the present computations, the mean
residence time was taken equal to the measured residence time, (see
Section III-D).
V-B. DESCRIPTION OF THE MIXING-COMBUSTION ANALYSIS
A flow diagram of the two-dimensional (x,z) bluff-body flow field to
be modeled is shown in Fig. 3^- Distribution of the velocity, tempera-
ture, and species concentrations in the premixed methane-air stream and
the high temperature near-wake region at the bluff-body location are
assumed to be known. The mixing and chemical reaction between the outer,
cold flow and the inner high temperature wake region is treated by div-
iding the combustor into a large number of stations a distance Ax apart,
and solving the conservation equations in the manner of an initial value
problem with the conditions at the end of one Ax segment becoming the
initial conditions for the following segment. For a typical problem,
the number of segments usually exceeds 5,000. Within each segment of
the combustor there are a number of streamlines (20-30 for the present
calculations) along which mixing is assumed to proceed independently
of any chemical reaction effects. Therefore, when equations describing
the conservation of mass, momentum, species concentration, and energy
are written for the segment, terms involving the generation of heat
and the production or loss of chemical species are omitted. After the
fluid within the segment has mixed, chemical reaction effects are intro-
duced on the basis of average conditions in Ax. This explicit coupling
procedure has been adopted in order to avoid excess computation time.
In carrying out the analysis, the pressure is assumed to be a function
only of the axial coordinate, that is, P = P (x). The turbulent shear
8k
-------
a) BURNER FLOW SYSTEM TO BE MODELED
LLL L1111111111111111111111111111111111111111111
\\\\\\\\\\\ TTTTT\\\\
b) INITIAL CONDITIONS (x = 0)
INITIAL VELOCITY = U& U/(1-BR))
INITIAL PRESSURE: AS MEASURED AT x = o
h/2
1
D/2
I
^TEMPERATURE
"" 1
f~NO CONCENTRATION
1
1
1
1
1
O2 CONCENTRATION~
1
I~CH4 CONCENTRATION
1
I 1
1
1
1
1
I
re
FIG. 34
ANALYTICAL MODEL
85
-------
stress in the conservation of momentum equation is teiken as the product
of an eddy viscosity and the local velocity gradient in the manner of
Boussinesq (cf., Ref. 10). Lewis and Prandtl numbers are taken equal
to unity, eliminating distinction between the eddy coefficients of momen-
tum^ mass and heat. The conservation equations are transformed by the
method of von Mises (cf., Ref. 31) from the physical (x} z) plane into
a coordinate system in which streamlines are parallel, thereby yielding
expressions of the "heat conduction" type. These are subsequently put
in a form suitable for numerical solution by substituting finite difference
approximations for all derivatives.
The solution of the resulting finite difference equations for each seg-
ment, Ax, is obtained for an incremental change in pressure which is
impressed on the flow between the beginning of the segment, where all
conditions are specified, and the end of the segment. The initial con-
ditions typically are given as step profiles of temperature, velocity
and species concentrations (Fig. 3^b)> within the Ax segment, the gas
streams mix at a rate determined by the eddy viscosity which is a
function of flow conditions. Conditions at the end of the segment
are checked for compatibility by employing the conservation of mass.
If inconsistencies are found, the pressure change across the segment
is adjusted and the entire calculation is repeated until mass conser-
vation is obtained.
Chemical reaction effects are coupled into the analysis at this point.
Average values of temperature, pressure, concentrations, and velocity
for all streamlines between the two stations are calculated and a test
is made to determine if methane-air ignition has occurred locally
(see later discussion of Eq. 39). In the event that ignition has
not occurred along a particular streamline, no further chemistry
computations are carried out for that streamline in the segment under
consideration. For those streamlines on which ignition has occurred,
the flow conditions determined for the segment under consideration are
applied to the calculation of the equilibrium composition and tempera-
ture of the methane-air reacting gas mixture. In this calculation,
a successive approximation procedure is used to find the simultaneous
solution of the standard equations of chemical equilibria, conserva-
tion of (atomic) mass, and conservation of energy. Equilibrium CH^-
air reaction concentrations and temperature are subsequently applied
to the calculation of WO concentration from auxiliary kinetics expres-
sions. Finally, conditions at the downstream station are adjusted to
reflect all chemical reaction effects to complete the analytical proce-
-------
dure. The entire procedure is then repeated between the current station
and the next station further downstream, and so on, until the entire
reacting flow field is mapped out.
Other assumptions used in the analytical development include: (l) per-
fect gases, (2) adiabatic wall conditions, (3) turbulent eddy viscosity
is much larger than the molecular viscosity.
V-B.l. Governing Equations
The system of (n + 6) equations which describes the mixing of variable
density, turbulent streams within the framework of the present model
has been derived by Vasiliu (Ref. 32). The well-known turbulent boun-
dary layer assumptions were applied in this derivation. These equations
contain the (n + 6) time-averaged unknowns m. (i = 1, 2, . . ., n) p,
u, v, P, H, and T, and the eddy viscosity, e . Thus, once the eddy
viscosity is specified, all parameters may be determined at every point
throughout the flow field for selected boundary conditions. The govern-
ing equations include the perfect gas law, a relation for the specific
enthalpy, and the following:
Global Continuity:
dz
Species Continuity:
dm. 3m.. e t
<3(/>vzS) . .
r = o (23)
_— = |Z~—H (2^)
dz z° dz dz
Conservation of Momentum:
du du dP e d [ a du
z
(25)
Conservation of Energy:
dH dH e d \ » dH
(26)
87
-------
Area Relation:
dz= CONSTANT (27)
In the above equations, 6 is zero or unity depending on whether a two
dimensional or axi symmetric flow is being considered..
The transformation of Eqs. 23 through 26 by the method of von Mises
requires the introduction of the stream function, i// . In order to
limit ;// to the region 0< 4/ < 1 the following defining expressions
are utilized for the stream function:
dz Gh
o
(29)
where G is the total mass flux. Transforming Eqs. 221 through 26 by
the application of Eqs. 28 and 29 with the nondimensional independent
variables
(30)
*i- (31)
yields:
Species Continuity:
(32)
-------
Conservation of Momentum:
AI\ i dP
(33)
Conservation of Energy:
.28
w20 •£- <»>
with (3 = D/(Gh)2.
The parabolic Eqs. 32 - 3^ were put in finite difference form to allow
their solution by numerical methods on a Univac 1108 computer. This is
equivalent to replacing the continuous flow system by a rectangular
finite grid network having an axial grid size ofA£ and a transverse
grid size ofA0. In general, neitherA^ norA^need be constant. Both
| and i// are in non-dimensional form and vary from zero to unity.
V-B.2. Eddy Viscosity Model
Before the system of finite difference equations can be solved, it is
necessary to specify the variation of the transport coefficient or eddy
viscosity for the reacting turbulent flow downstream of the re circula-
tion zone. From Eq. 3 and again taking the mixing layer width to grow
linearly in the flow direction, Ref. 9, it follows that
*""~OU
-------
V-B.3- Chemistry
V-B.3J- Ignition Delay Time for CHi -Air - Once mixing has occurred
between two adjacent stations separated by a distance Ax, linear averages
of the dependent variables may be computed on every streamline in Ax and
applied to the determination of the ignition delay time, Typ, from data
correlations. The dwell time, AT, defined as
(37)
also may be calculated. This quantity is a measure of the average time
fluid spends in the region Ax. The ignition delay time corresponds to
the length of the preignition period during which little change in tempera-
ture or species concentration occur. Heat generation and conversion of
reactive species occurs at equilibrium following the ignition delay time.
Ignition delay times were found from the high temperature portion of the
CH,-air data (Ref. 3^) presented in Fig. 35 . That is, for T =±1800°K,
(38)
where C^ denote species concentrations in moles/cc.
The ignition criteria used presumes that ignition occurs on any stream
line at the axial location where the following condition is satisfied:
2 -¥- * l.o ( 39)
X 'ID
Once ignition has occurred on a given streamline, subsequent reaction
between methane and oxygen is assumed to proceed to equilibrium instantan-
eously.
V-B.3JJ. Equilibrium Chemistry Calculation for Fuel-Air Combustion - To
calculate the equilibrium composition, expressed as a mole fraction, y.,
for each of n chemical species, for a given stoichiometry, temperature,
90
-------
2000
0.50
TEMPERATURE - K,
1600 1400
0.60
0.70
0.80
1200
103AT(°K)
FIG. 35
METHANE-AIR IGNITION DELAY TIMES
-------
and pressure at a station, the following equations must be satisfied:
Material balance: for each of q different atoms present in the n differ-
ent chemical species,
"ij Vi = bj
j >0 (i = 1,2,...,q)
where N is the total (variable) number of moles of species present in
the mixture, y.. is the stoichiometric coefficient for a given atom j
in a given chemical species i (for example, for the hydrogen atom count
in methane, CH, , v = ^), and b, equals the total nunber of atoms of j
present in the system as specified by the input stoichiometry.
Free-energy minimization: the chemical potential (ie., the partial molar
Gibbs free-energy) of the ith chemical species, p.^, is given as
/AJ -^.j = RTIn(fi/fi°) (
where f^_ is the fugacity of the ith species, and superscript ° refers to
the standard state of pure i taken as pure gas at unit pressure. Intro-
ducing the assumption of an ideal gas and a perfect mixture, theie
results
H-; = Mi" + RT In p.
i • | r i
in which p- is the partial pressure of i in the mixture in atmospheres
Thus, the total free energy, F, of the system is
or, since pi = y^P, where P is the total pressure in atmospheres,
n
F = N Z y( ( /A.° + RT In P + RT In y.)
92
-------
The free energy per mole of pure i, ^° , can be obtained from the litera-
ture where It is tabulated as a function of temperature up to 6000°K,
Ref. 35- This is the only species property required for solution of
the equilibrium problem. The values of P and T are specified conditions.
Chemical equilibrium is attained when the total free energy of the system
is a minimum. The problem, then, is to determine the set of values,
y^, which will minimize Eq. kk , subject to the q constraints of Eq.
40. A computer program developed to solve this system of equations
was incorporated into the present analytical procedure.
Nitric Oxide Formation Kinetics - For the present analytical
model, all chemical species with the exception of nitric oxide (NO) and
atomic nitrogen (N) are assumed to be in local chemical equilibrium on
any streamline for which ignition has occurred, ie., where Eq. 39 has
been satisfied. The nitric oxide Concentration is determined following
the equilibrium chemistry calculation using equilibrium conditions in
an auxiliary chemical kinetics model.
The kinetics of formation of nitric oxide is modeled using a two-reaction
mechanism (Ref. 25):
0 + N _' NO + N (l9)
*~
k20 . .
N + 0 _,. NO + 0 (20)
k*~
-20
Invoking a steady-state approximation for N, the nitric oxide formation
rate may be expressed as,
2kl9C0CNg [I- CNo/KCNgC0gJ / ^
= | + k_19CNO/k20C02
where Cj_ = concentration of species i (moles/cnP), kj = forward rate con-
stant of reaction j (cm3 /mole-sec), k_ =, = reverse rate constant of reaction
j ( cm3 /mole- sec) and K = k, ^koQ/k , qk ? . Values for the rate constants
'in Eq. 1+5 were taken from Ref. 27, and tabulated earlier in Section V-A.
93
-------
In the present calculations, Eg.. ^5 was used to calculate the nitric
oxide formation rate assuming that 0, OQ and N^ are in local chemical
equilibrium. The quantity of NO formed on any streamline in Ax subse-
quently is added to the existing NO produced upstream of Ax, and all
remaining species are adjusted to reflect chemical reaction effects
before proceeding to the next downstream computational station.
V-C. COMPARISON OF EXPERIMENTAL AND ANALYTICAL RESULTS
V-C.l Measured and Predicted Nitric Oxide Concentrations in the Recir-
culation Zone
The stirred reactor model, discusse d in Section V-A ,. was used to compute
nitric oxide concentrations in the recirculation zone for the conditions
of the combustion experiments. The predicted nitric oxide concentra-
tions are tabulated in Table VIII for a range of inlet velocities and
equivalence ratios. In addition, typical comparisons of the predicted
nitric oxide concentrations with measured values are illustrated in
Figs.l6a and l6b. The equilibrium stirred reactor model predicts nitric
oxide concentrations in the recirculation zone which are substantially
less than were observed in the experiment. The non-equilibrium stirred
reactor model, using measured radical concentrations, predicts nitric
oxide concentrations which are relatively close to, although somewhat
less than, the measured concentrations. In the stirred reactor model,
it was assumed that the nitric oxide concentration in the input stream
was zero. The difference between nitric oxide concentrations predicted
by the non-equilibrium stirred reactor model and mee-sured concentrations
may be attributed to nitric oxide formed in the flame zone surrounding
the recirculation zone and transported into the zone.
The non-equilibrium stirred reactor model correctly predicts trends in
the variation of recirculation zone nitric oxide concentration with equiva-
lence ratio and inlet velocity. The difference between the predicted and
measured nitric oxide concentrations decreases as the inlet velocity
increases. This trend may reflect an improvement in mixing rates in
the recirculation zone as the flame bounding the zone undergoes a trans-
ition from laminar to turbulent with increasing inlet velocity (see
Section III-B).
In two recent investigations of nitric production in stirred reactors
(Refs. 30, 36), observed nitric oxide concentrations were substantially
larger than the concentrations predicted by analytical models which
-------
decouple the combustion chemistry from the nitric oxide chemistry. Pre-
dicted nitric oxide concentrations obtained from models which couple
the nitric oxide chemistry to a detailed combustion mechanism for hydro-
carbon fuels also were lower than observed concentrations. However, in
both of these investigations, no attempt was made to measure radical con-
centrations in the stirred reactor. Without a measure of the concentra-
tions of radical species, it is difficult to assess the validity of the
analytical models used in these two studies.
Based on the foregoing observations, it is reasonable to conclude that
a well-stirred reactor is a satisfactory model for the recirculation zone
for the conditions investigated in the combustion experiments, providing
some aspects of non- equilibrium hydrocarbon chemistry are included in
the model.
V-C.2 Measured and Predicted Mitric Oxide Concentrations in the Exhaust
Gas
The turbulent mixing/reaction model discussed in Section V-B, was applied
to the computation of flame angles and temperature, velocity, and species
concentrations in a two-dimensional combustor containing a cylindrical
bluff-body flameholder (Fig. 3^a) • The four sets of flow conditions listed
in Table XIV were investigated for the geometry of the burner rig. In
the absence of definitive information, the inner hot region was assumed
to have the same initial velocity as the outer pre-mixed methane-air
cold flow. The initial temperature and nitric oxide concentration in
the inner hot zone were taken from the burner experimental data discussed
in Section III. Similarly, the inital pressure used in the calculation
was taken from measurements at the bluff-body location (Fig.
Gradients of velocity, temperature, and species concentration were taken
equal to zero at the combustor center-plane and at the combustor walls.
As a result, wall boundary layer processes were not properly simulated
in the analytical model.
The first task in the analytical study was the determination of the
constant c^ in Eq. 36- This was done by varying c, in calculations
with test conditions number 2, Table XIV, until predicted and measured
flame spreading angles were in agreement. The constant, c,, was estab-
lished at the value 0.121, which is consistent with the constant den-
sity, two-dimensional jet mixing value of 0.0135 (Ref. 37)- Hence, cj_
95
-------
TABLE XIV
SUMMARY OF ANALYTICAL RESULTS
TEST
CASE
1
2
3
4
U^ (M/SEC)
20
20
20
40
0
0.778
0.875
0.972
0972
TAN « max
CALCULATED
0.057
0075
0.079
0.055
MEASURED
0.075
0.082
0.063
AV. NO CONC. (PPM-DRY BASIS BY WT.)
CALCULATED
X/D = 20
0.6
4 0
10.0
11.0
44
9
19
MEASURED
X/D =44
65+15
96± 10
140± 10
67 t 15
96
-------
was fixed at 0.0121 in the remaining analytical effort involving test
cases 1, 3 and 4.
Analytical results presented in Figs.36-39 for test case 2 are typical
of all four test cases. The ignition envelope, which is a curve drawn
through those points in the flow field where the ignition criterion,
Eq. 39, is first satisfied, is shown in Fig.36. For purposes of this
study, the ignition envelope and the flame boundary are coincident.
The calculated flame boundary, in agreement with photographic observa-
tions (Fig. 9}, initially necks down. Approximately 6 to 8 body dia-
meters downstream from the bluff-body location, the flame begins to
spread into the pre-mixed outer flow, ultimately displaying a growth
which is linear with axial distance. The linear spread of the flame
is conveniently characterized by the maximum flame angle a . At
niQtX
a distance downstream from the bluff-body of approximately 20 body-
diameters, the analysis predicts that the flame reaches the burner wall.
This behavior, which is at variance with the experimental observation
that the flame bends away from the burner wall, results from in-
correct boundary conditions. It is expected, however, that the de-
tails of the flow field are realistic throughout the burner except
in the immediate vicinity of the burner wall.
Predicted flame angles for the four test cases are given in Table XIV
and Figs. lOa and lOb. The correct trends of flame angle variation with
approach flow velocity and equivalence ratio are reproduced.
Computed velocity profiles and nitric oxide concentration distribu-
tions throughout the flow field are shown in Figs.37 and 38 • As
expected, the entire flow accelerates in the constant area burner.
The hot, low density inner flow is accelerated more rapidly than
the outer cold flow, however, thereby producing the characteristic
reacting wake flow velocity distribtion. The spread of the flame
into the cold flow is apparent in both Figs. 37 and38- Note that at
an axial distance of x/D = 24, essentially all of the flow has ignited
and the methane-air chemistry is completed. However, nitric oxide
production continues until the end of the burner at station x/D =
44 (Fig. 39). As a result, the HO concentration doubles in the last half
of the burner test section. The buildup of CO, CO^, and NO, in the
reacting methane-air flow can be followed by referring to Fig. 39. It
is predicted, for test case 2, that approximately nine ppm (dry basis
by weight) of NO will be produced in the burner. Nitric oxide production
97
-------
LL= 20 m/sec, x= 0.875
FIG 36
CALCULATED FLAME BOUNDARY
-------
20 m/sec, &. a 0.875
2.0
1.8-
1.6
1.4
1.2
u
I
Q 1.0
LLJ
in
cc
0.8
0.6
0.4
0.2
2 4.5
I
16
24
44 :x/D
0
1.0 2.0 3.0 4.0 5.0 6.0
VELOCITY, u/LL
FIG. 37 CALCULATED WAKE VELOCITY PROFILES
-------
= 20 m/sec, (f>x = 0.875
5X10-4
o
z
E
o
<
en
CO
CO
3X10
0
FIG. 38
0.8 1.2
TRANSVERSE DISTANCE, z/D
NO CONCENTRATION DISTRIBUTION
1.6
2.0
100
-------
= 20 m/sec, $ = 0.875
10
2.5
5.0
7.5 10.0
AXIAL DISTANCE, x/D
35.0
40.0
450
FIG. 39
CALCULATED SPECIES DISTRIBUTIONS
101
-------
levels for the other test cases are listed in Table XIV. in general,
the predicted nitric oxide production levels are significantly lower
than the levels measured at the burner exit using a time-of-flight
mass spectrometer.
The analytical model correctly predicts the essential features of the
flame spreading process in the burner. Hence, it is likely that the
predicted distributions in the mean flow properties in the burner,
ie., temperature, density and velocity are reasonably close to the
actual distributions. It is suspected, therefore, that the discre-
pancy between predicted and measured nitric oxide concentrations
can be attributed to two of the assumptions made in the analytical
model. First, in the modeling of the methane-air chemistry, it was
assumed that methane and air react to equilibrium products following
an induction period. Hence, the analytical model does not consider
the possible effect on non-equilibrium radical concentrations on the
nitric oxide formation process. Measurements in the recirculati on
zone have shown that radical concentrations are significantly greater
than equilibrium values. In studies in premixed, turbulent, methane-
air flames (Ref.38), radical concentrations in the flame zone were found
to exceed equilibrium values. If radical concentrations in the flame
are greater than equilibrium, then the actual nitric oxide formation rate
will exceed that calculated using equilibrium radical concentrations.
To bring predicted and measured exhaust nitric oxide concentrations
into agreement would require 0-atom concentrations more than an order
of magnitude larger than equilibrium values. Such non-equilibrium
0-atom concentrations are possible, but not likely, over the entire
range of experimental conditions.
A second assumption incorporated in the analysis may contribute to
underprediction of nitric oxide production in the turbulent flame.
In the analysis, time-dependence of the flow properties was neglected.
From the flow visualization studies, Section III-B, it is known that
the downstream flame is characterized by turbulent fluctuations and
eddies. The fluctuations in temperature, density, velocity and species
concentrations about the time-mean values, particularly the temperature
fluctuations, can result in an increased nitric oxide production rate.
For the conditions of the present experiments, a temperature fluctua-
tion of ±5% can increase the nitric oxide formation rate by approximately
a factor of two.
102
-------
SECTION VI
CONCLUSIONS
The experimental results obtained from the present investigation
indicate that nitri c oxide production in the recirculation zone down-
stream from the bluff-body flameholder depends on transport rates be-
tween the zone and the outer flow. These transport rates are strongly
influenced by fluid dynamic phenomena, such as wake transition and
vortex shedding (cold flow). It is essential that these fluid dynamic
phenomena be considered in analytical models of nitric oxide formation
in reacting flows with recirculation.
A stirred reactor model can adequately represent nitric oxide produc-
tion in the recirculation zone downstream from the flameholder, providing
non-equilibrium hydrocarbon chemistry is included in the model. A
conventional eddy viscosity model, based on shear-generated turbulence,
predicts flame angles which are in agreement with experimental observa-
tions. However, predicted nitric oxide concentrations in the burner
exhaust, based on average local temperature, are substantially less
than the measured values. The discrepany between predicted and measured
nitric oxide production may be attributed to the neglect of non-equili-
brium chemistry and turbulent fluctuations in the analytical model.
Volumetric nitric oxide production rates in the recirculation zone have
been compared with overall volumetric nitric oxide production rates.
For the present experimental configuration, nitric oxide production in
the recirculation zone is not a major factor in the overall nitric oxide
formation. This situation is the direct result of the relatively low
temperature associated with the recirculation zone downstream from bluff-
body flame stabilizers. It is expected that nitric oxide production
in recirculation zones in practical combustion devices, such as furnaces
and gas turbines, will be significant since the recirculation zone
temperatures in these devices more nearly approach the adiabatic combus-
tion temperature. The stirred reactor model for the recirculation zone
indicates that an increase in temperature of approximately 100°K would
increase the nitric oxide production rate by approximately a factor of
five. Based on present results, the bluff-body flameholder appears
to be an attractive method of flame stabilization from the standpoint
of nitric oxide emissions.
Results from the present investigation point to the recirculation zone
103
-------
as a major factor in nitric oxide production in practical combustion
devices. The results indicate that combustor fluid dynamics strongly
influence nitric oxide production rates in recirculation zones.
-------
SECTION VII
REFERENCES
1. Niedzwiecki, R. W., and Jones, R. E., "Pollution Measurements of a
Swirl-Can Combustor," NASA Technical Memorandum, TMX-68l60, (1972).
2. Heap, M. P., Lowes, T. M., and Walmsley, R., "The Emissions of Nitric
Oxide from Large Turbulent Diffusion Flames," Fourteenth Symposium
(international) on Combustion (in press).
Heap, M. P., Lowes, T. M., and Walmsley, R., "The Effect of Burner
Parameters on Nitric Oxide Formation in Natural Gas and Pulverized
Fuel Flames," paper presented at First American Flame Days, Chicago,
Illinois, (1972).
3. Shoffstall, D. R., and Larson, D. H., "Aerodynamic Influences on
Combustion and Pollution Emissions," paper presented at the Spring
Meeting of the Combustion Institute, Central States Section,
Champaign, Illinois, (1973).
4. Markovin, M. V., "Flow Around Circular Cylinder - A Kaleidoscope
of Challenging Fluid Phenomena," Proceedings of the Symposium on
Fully Separated Flows, AMSE, New York, p. 102 (1964).
5. Roshko A. and Fiszdon, W., "On the Persistance of Transition in the
Near-Wake," Problems of Hydrodynamics and Continuum Mechanics,
Soc. Ind. Appl. Math, Philadelphia , p. 606 (1967)-
6. Bloor, M. S., "The Transition to Turbulence in the Wake of a Cir-
cular Cylinder," J. Fluid Mechanics, 19, p. 290 (1964).
7. Williams, G. C. Hottel, H. C. and Scurlock, A. C., "Flame Stabiliza-
tion and Propagation in High Velocity Gas Streams," Third Symposium
on Combustion, Flame and Explosion Phenomena, The Williams and
Wilkins Co., Baltimore, p. 21 (1949).
8. Shigemitsu, Y., "Statistical Theory of Turbulence," Aerophysics
Department, Mississippi State University, Research Report No. 64,
(1966).
9. Abramovich, G. N., "The Theory of Turbulent Jets," M.I.T. Press,
Cambridge, Mass. (1963).
105
-------
10. Schlichting, H., "Boundary Layer Theory," McGraw-Hill Book Company,
Inc., Fourth Edition, (1960).
11. Winterfeld, G., "On Processes of Turbulent Exchange Behind Flame
Holders, "Tenth Symposium (international) on Combustion., The Combustion
Institute, Pittsburgh, p. 1265 (1965).
12. Zukoski, E. E. and Marble F. E., "The Role of Wake Transition in
the Process of Flame Stabilization on Bluff Bodies," Combustion
Researches and Reviews (Butterworth, London) p. 167 (1955).
13. Wright, F. H., "Bluff-Body Flame Stabilization: Blockage Effects,"
Combustion and Flame 3, p. 319 (1959).
14. Pein, R., Peschel, H., and Fetting F., "Recirculation Zone Con-
centrations and Temperatures of Bluff-Body Stabilized Turbulent
Flames. Combustion Science and Technology 1, p. 327 (1970).
15. Gaydon, A. G. and Wolfhard, H. G., "Flames: Their Structure, Radia-
tion and Temperature," Chapman and Ball, London, p. 234 (1960).
16. Bovina, T. A., "Studies of Exchange Between Re-circulation Zone Behind
the Flameholder and Outer Flow," Seventh Symposium (international) on
Combustion Butterworth, London, p. 692 (1959).
17. Bowman, C. T., "investigation of Nitric Oxide Formation Kinetics in
Combustion Processes: The Hydrogen-Oxygen-Mtrog;en Reaction, " Combus-
tion Science and Technology 3, p. 37 (1971).
18. Newhall, H. K., and Shahed, S. M., "Kinetics ,of Nitric Oxide Formation
in High-Pressure Flames,"Thirteenth Symposium (international) on
Combustion The Combustion Institute, Pittsburgh, p. 381 (1971).
19. Bowman, C. T., and Seery D. J., "investigation of Nitric Oxide Forma-
tion Kinetics in Combustion Processes: The Metha.ne-Oxygen-Nitrogen
Reaction," Emissions from Continuous Combustion Systems, Plenum, New
York p. 123 (1972).
20. Meinel, H., and Just, Th., "A New Analytical Technique for Continuous
NO Detection in the Range from 0.1 to 5000 ppm," paper presented at
the 4lst Meeting of the AGARD Propulsion and Energetics Panel, London,
(1973).
21. Baines, W. D. and Peterson, E. G., "An Investigation of Flow Through
Screens." Transactions of the ASME. p. ^67 (195!).
106
-------
22. Glasstone, S.,"Textbook of Physical Chemistry," Second Edition, D. Van
Nostrand Company, Inc., Princeton, New Jersey, (1946).
23. Longwell, J. P., Frost, E. E. and Weiss, M. A., "Flame Stability
in Bluff Body Recirculation Zones," Ind. Eng. Chem. 45, p. 1629 (1953)-
2k. Jones, A., and Prothero, A., "The Solution of the Steady-State Equa-
tions for an Adiabatic Stirred Reactor," Combustion and Flame, 12,
p. 457 (1968).
25. Zeldovich, Ya. B., Sadovinkov, P. Ya., and Frank-Kamenetskii, D. A.,
"Oxidation of Nitrogen in Combustion," Academy of Sciences of
USSR, Institute of Chemical Physics, Moscow-Leningrad (trans, by
M. Shelef), (1947).
26. Heywood, J. B., "Gas Turbine Combustor Modeling for Calculating
Nitric Oxide Emissions," paper presented at AIAA/SAE 7th Propulsion
Joint Specialist Conference, Salt Lake City, Utah, (1971).
27. Baulch, D. L., Drysdale, D. D., and Lloyd, A. C., "Critical Evalua-
tion of Rate Data for Homogeneous Gas-Phase Reactions of Interest
in High-Temperature Systems," Department of Physical Chemistry,
Report No. 4, Leeds, England, (1969).
28. Schott, G. L., "Kinetics Studies of Hydroxyl Radicals in Shock
Waves. III. The OH Concentration Maximum in the Hydrogen-Oxygen
Reaction," J. of Chemical Physics, 32, p. 710 (1960).
29. Milne, T. A. and Greene, F. T., "Mass-Spectrometric Studies of
Reactions in Flames. II. Quantitative Sampling of Free Radicals
from One-Atmosphere Flames," J. of Chemical Physics, 44, p. 2444
(1966).
30. Engleman, V. S., Edelman, R. B., Bartok, W., and Longwell, J. P.,
"Experimental and Theoretical Studies of NOV Formation in a Jet-
J\
Stirred Combustor, Fourteenth Symposium (international) on Combus-
tion (in press).
31. Cohen, L. S., "An Analytical Study of the Mixing and Nonequilibrium
Chemical Reaction of Ducted Compressible Streams," Paper No. 66-617,
AIAA Second Propulsion Joint Specialist Conference, Colorado Springs,
Colo. (1966).
107
-------
32. Vasiliu, J., "Turbulent Mixing of a Rocket Exhaust with a Supersonic
Stream Including Chemical Reaction,"J. Aerospace Sciences, 29, p. 19
(1962).
33. Cohen, L. S., "A Kinematic Eddy Viscosity Model Including the
Influence of Density Variations and Preturbulence," Langley Confer-
ence on Free Turbulent Shear Flows, Langley, Va. (1972).
34. Seery, D. J., and Bowman, C. T., "An Experimental and Analytical
Study of Methane Oxydation Behind Shock Waves," Combustion and
Flame, l4, p. 37 (1970).
35- JANAF Tables of Thermochemical Data, Dow Chemical Co., Midland,
Mich. (1964).
36. Pratt, D. T., and Malte, P. C., "Formation of Thermal and Prompt
NOX in a Jet-Stirred Combustor," paper 34b presented at the 75th
National A.I.Ch.E. Meeting, Detroit, Michigan, (1973)•
37. Forthmann, E., "Uber Turbulent Strahlansbreitung," Ingen.-Arch.,
5, P. 42 (1934).
38. Thompson, D., Brown, T. D., and Bee'r, J. M., "NOX Formation in
Combustion," Combustion and Flame, 19, p. 69 (1972).
39. Zabielski, M. F., and McHugh, T.M., "Resolution Enhancement by
Iterative and Fourier Techniques," paper presented at the 21st
Annual Conference on Mass Spectrometry and Allied Topics, San
Francisco, Calif., (1973).
40. Penner, S. S., "Quantitative Molecular Spectroscopy and Gas
Emissivities," Addison-Wesley, Reading, Mass., p. 69 (1959).
4l. Kiser, R. W., "introduction to Mass Spectrometry and its Appli-
cations," Prentice Hall, Englewood Cliffs, New Jersey, p. 218
(1965).
42. Roboz, J., "introduction to Mass Spectrometry: Instrumentation
and Techniques" Interscience, New York, p. 3^4 (1968).
108
-------
SECTION VIII
NOMENCLATURE
A Pre-exponential factor
A^, Recirculation zone surface area, cnr
a Absorption coefficient
B Recirculation zone maximum width
BD Blockage ratio, D/h
b Screen wire diameter, cm
bj Total number of atoms of species j present
J
Cj_ Concentration of species i, moles/cc
c, c-. Constants
D Cylindrical bluff-body diameter, cm
d Spacing between fiber optic probe units, cm
E Activation energy, cal
F Total free energy, kcal
f^ Fugacity of species i, atm
G Mass flux, gm/cm^-sec
H Specific enthalpy, kcal/mole
h Test section height, cm
I Intensity
K. Equilibrium constant of reaction j
J
109
-------
k. Rate constant of reaction j
j
L Re circulation zone length, cm
M. Molecular weight of species i, gm/mole
m Mass flow rate, gin/sec
m. Mass fraction of species i, gm i/gm mixtures
N Total number of moles of species present, mole
n^ Molar flow rate of species i, mole i/sec
n Total number of species
P Pressure, atm
p Partial pressure, atm
q_ Number of different atoms present
R Universal gas constant
R. Rate of reaction j
J
Re Reynolds number
r Denotes closure point
S Screen mesh spacing, cm
g. Species i
Sh Strouhal number
s Denotes separation point
T Temperature, °K
U Specific velocity designation, m/sec
u Axial velocity, m/sec
110
-------
u' Fluctuating transverse velocity, m/sec
V Volume, cc
v Transverse velocity, m/sec
v' Fluctuating transverse velocity, m/sec
w Width of test section, cm
x Axial distance from bluff-body, cm
y^ Mole fraction of species i, mole i/mole mixture
z Transverse distance, cm
Greek
a Flame angle, deg
(3 Defined as D/(Gh)2
6 Exponent for two-dimensional (6=0) or axisymmetric
(6=1) flow
e Eddy viscosity, gm/cm-sec
^1 Wondimensional transverse coordinate
Q
Slope of tracer concentration distribution
o
^ Wavelength, A
|j£ Partial molal Gibbs free energy of species i, kcal/mole
v. . Stoichiometric coefficient for species i in chemical reac-
tion j
p Density, kg/rrP
i Specific molar concentration of species i, mole i/gm-i
Time, sec
111
-------
AT
0
CD
Subscripts
a
f
ID
m
max
HO
o
t
tot
w
Other
Dwell time, sec
Residence time, sec
Equivalence ratio
Stream function
/(m/m)
sto:Lchiometric
Nondimensional axial coordinate
Exponent in rate equation or wavenumber, cm'
Air
Fuel
Ignition delay
Mean value
Maximum value
Nitric Oxide
Initial
Tracer
Total
Vortex shedding
Recirculation zone
Free stream
Reference quantity
Average quantity
112
-1
-------
SECTION IX
APPENDICES
Page
A. Experimental Techniques in the Combustion Experiments 113
B. Fiber Optic Measurement Techniques in the Cold Flow
Experiments 129
113
-------
APPENDIX A
EXPERIMENTAL TECHNIQUES IN THE COMBUSTION EXPERIMENTS
A.I Flow Visualization Techniques
The flow visualization measurements, discussed in Section III-B, were made
using the schlieren optical system shown in Fig. A-l« The light source
used for the spark schlieren measurements was an air-gap spark with a 3-^sef
duration. The knife-edge, used in conventional schlieren systems, was
replaced in the present experiments by a O.l6-cm diameter circular aperture
(pinhole). Use of. the circular aperture permits measurement of refrac-
tive index gradients in both transverse and axial directions and facili-
tates determination of the recirculation zone boundary. The light source
used for the time-exposure schlieren measurements was: a 200 watt Hg-Xe
dc arc lamp (Hanovia 901-Bl). The schlieren photographs were recorded on
V x 5" Polaroid 3000 film.
A.2 Modified Sodium Line Reversal Technique
The optical system used to make measurements of temperature and residence
time in the recirculation zone is shown in Fig. A-2 . Light from a tung-
sten filament lamp passes through the burner test section and enters a
monochromator (McPherson Model 235) set on the 5890 A sodium D-line. The
intensity.of the lamp and sodium emission from the flame at 5890A is moni-
tored by a photomultiplier (EMI 9558 BQ). The signal from the photo-
multiplier is fed to the y-axis amplifier of an x-y plotter (Houston Series
2000). The lamp current, and hence the brightness of the lamp, can be
automatically ramped from zero to a maximum value, corresponding to a
lamp brightness temperature of approximately 2^00°K. A signal, propor-
tional to lamp current, is fed to the x-axis amplifier of the x-y plotter.
The optical system is calibrated to give the lamp brightness temperature
as a function of x-axis voltage by manually ramping the lamp current and
measuring the brightness temperature of the filament using an optical
pyrometer. To minimize errors resulting from losses in the focusing
optics and test section windows, the calibration is carried out using
essentially the same optics as are used to make temperature measurements,
Fig. A-2. The brightness temperature of the lamp at 5890A is determined
from the calibration carried out at 6550A, using the correction factors in
Ref. 15.
-------
FIG A 1 SCHEMATIC DIAGRAM OF THE SCHLIEREN OPTICAL SYSTEM.
PHOTOMULTiPLIE
RATING OPTICAL PYROMETER
LAMP VOLTAGE SIGNAL
/
LAMP POWER
SUPPLY
&
HAMP
CONTROL
PHOTOMUL TIPLIER SIGNAL
X Y PLOTTER
FIG A 2 SCHEMATIC DIAGRAM OF THE SODIUM D-LINE PHOTOMETER
115
-------
A.2.1 Experimental Procedure for Temperature Measurements - With no flow
in the test section, the lamp current was manually se~ to its maximum
value, and the brightness temperature of the filament was measured using
the optical pyrometer. This measurement served to confirm the validity
of the temperature-lamp current calibration. The lamp was ramped auto-
matically to zero current, and the corresponding intensity (y-axis) --
brightness temperature (x-axis) curve was traced out by the x-y plotter,
Fig. A-3. The lamp current was reset manually to its maximum value, and
combustion was initiated in the burner. After steady-state conditions
were established in the burner, sodium tracer was continuously injected
into the recirculation zone. With the lamp set at its maximum brightness
temperature (~2400°K), injection of sodium tracer results in a decrease
in the lamp intensity, as observed by the photomultiplier, due to absorp-
tion of lamp radiation by sodium in the flame. The lamp was ramped auto-
matically to zero current, and the corresponding intensity-brightness
temperature curve was traced out by the x-y plotter, Fig. A-3- At the
z,ero lamp current, the observed intensity, with sodium tracer, is the
sodium D-line emission. The point where the curve with combustion inter-
sects the curve with no combustion is the reversal point, and the flame
temperature may be determined directly from the x-coordinate of the inter-
section. Following completion of the combustion run, the intensity-brightness
temperature curve was determined again, with no flow, to verify the lamp
calibration.
To assess the validity of the optical temperature measurement technique,
temperatures measured near the flame zone in premixed methane-air flames
on a Meker-type burner using the sodium-line reversal technique, were com-
pared with temperatures measured using small, coated F't/Pt-Rh thermocouples.
The temperatures measured using the two techniques, after a suitable radia-
tion correction was applied to the thermocouple measurements, were the same
within experimental uncertainty (t25°c).
A.2.2 Experimental Procedure for Residence Time Measurements - Residence
times of gas in the recirculation zone were determined by measuring the
decay of injected tracer, following rapid shut-off of the tracer supply.
The tracer material used in the present combustion experiments was sodium.
In these experiments, a water/sodium chloride solution was injected into
the recirculation zone through the rear of the flameholder. This injecting
flameholder consisted of two concentric stainless-steel tubes, each having
a series of small holes drilled through the wall facing in the downstream
direction. A water/sodium chloride solution flowed through the inner tube
116
-------
>-
z
c
• FLAME TEMPERATURE
LAMP BRIGHTNESS TEMPERATURE *
FIG. A-3
TYPICAL EXPERIMENTAL TRACES OBTAINED FROM THE
TEMPERATURE MEASUREMENT EXPERIMENTS.
117
-------
at all times. If the holes in the two tabes were aligned, by sliding the
inner tube along its axis, a small amount of the water/sodium chloride
solution could flow into the recirculation zone. If the inner tube was
again moved along its axis, so that the holes were no longer aligned, the
tracer solution flow into the recirculation zone stopped. The injecting
flameholder was set to the open position by a solenoid-actuator, Fig. h ,
pulling against a pre-loaded spring. When the solenoid was de-energized,
the pre-loaded spring rapidly moved the inner tube, shutting off the
tracer supply. The calculated closing time of the injecting flameholder
was 1.5 msec.
The optical system used to make the residence time measurements was essen-
tially the same as was used for the temperature measurements. However, in
the residence time experiments, the photomultiplier output was displayed
on an oscilloscope (Tektronix 5^5B). The oscilloscope sweep was triggered
on the closing of the solenoid-actuator. The 10-90$ risetime of the photo-
multiplier/oscilloscope combination was approximately 10^ sec.
Prior to an experiment, the tungsten filament lamp current was set to its
maximum value. After steady-state combustion was established in the
burner, the solenoid-actuator was energized, thereby turning on the tracer
flow. After steady tracer flow was established, as evidenced by a rela-
tively constant absorption of the tungsten lamp emission by the sodium
in the recirculation zone, the solenoid-actuator was de-energized shutting
off the tracer flow and triggering the oscilloscope. The resulting
absorption decay curve was displayed on the oscilloscope and recorded on
Polaroid film. A typical experimental trace is shown in Fig. A-^. As the
tracer concentration in the recirculation zone decreased, the observed
absorption of the tungsten lamp emission decreased. The absorption data,
Fig. A-J+, can be plotted on semi-logarithmic paper, Fig.A-5. Following an
initial period (~1 msec) associated with the shutting off of tracer supply,
the observed absorption decays nearly expontentially in time, with a
characteristic time constant of 8.9 msec. In the residence time experi-
ments, the maximum abosrption by the tracer material (atr=0) was approxi-
mately 5$ of the incident lamp intensity. For these low levels of absorp-
tion, the absorptivity is directly proportional to the tracer concentra-
tion, and the characteristic time constant of the absorption decay curve,
Fig. A-5, may be equated to the mean residence time. In the temperature
measurement experiments, injection of the tracer solution into the recir-
culation zone reduced the temperature by approximately 75°K- This rela-
tively small reduction in recirculation zone temperature (less than
118
-------
U^= 32 M/SEC
0 = 1.09
POSITION 1
z
g
-
oc
o
H h-
5 MSEC
FIG A 4 TYPICAL OSCILLOSCOPE TRACE FROM RESIDENCE TIME EXPERIMENTS
119
-------
1 0
0 1
TIME - MSEC
FIG A 5 TYPICAL SEMI-LOGARITHMIC PLOT OF ABSORPTION DECAY CURVE
120
-------
suggests that the tracer flow is sufficiently small so as not to signifi-
cantly perturb the flow pattern in the recirculation zone. Hence, it is
reasonable to assume that tracer injection does not have a significant
effect on the mass exchange rate.
A.3 Ultraviolet Absorption Technique
The concentration of OH and NO in the recirculation zone was measured usjng
an ultraviolet absorption technique. A schematic diagram of the optical
system used to make these concentration measurements is shown in Fig.A-6.
Radiation from a 1000-watt, Hg-Xe dc arc lamp (Hanovia 528 B-l) passes
through a beam splitter, a 13.8.eps chopping disc and is focused in the
center of the combustor test section. A spherical mirror, located on the
opposite side of the test section, reflects the beam back through the test
section. The return beam passes through the chopper and is reflected by
the beam splitter into a lens which focuses the beam on the entrance slit
of a grating monochromator (McPherson Model 235)- The light intensity at
a specified wavelength is monitored by a photomultiplier (EMI 9558 BQ). The
output signal from the photomultiplier passes through a phase-sensitive
amplifier, synchronized to the chopping frequency. The dc output of the
phase-sensitive amplifier, which is proportional to light intensity, is
displayed on a chart recorder (Hewlett-Packard 7100B). Use of the two-
path optical system, coupled with phase-sensitive detection, greatly
enhances the ability to measure small absorption signals, and hence small
concentrations of absorbing species.
A.3.1 OH Concentration Measurements - The mean OH concentration at a
single location in the recirculation zone (position l) was determined by
measuring the absorption of ultraviolet radiation by the 2 - TT(0,0) band
of the molecule at 3080A. In a previous experimental investigation
(Refs. 17 and 19) the optical system was calibrated to permit measurement
of OH concentration.
A.3-2 Nitric Oxide Concentration Measurements-- Recirculation Zone - The
mean nitric oxide concentration at several locations in the recirculation
zone was determined by measuring the absorption of ultraviolet radiation
by the 7(0,0)_band of the molecule. A similar optical technique has been
used to measure nitric oxide concentrations in several recent investiga-
tions of pollutant formation in combustion processes (Refs. If-20). In the
present investigation, the optical system was calibrated by flowing known
121
-------
CONDENSER LENS
-APERTURE
.CHOPPER 113 8 CPSI
PHASE SENSITIVE
AMPLIFIER
/ -RETURN
MIRROR
^ TOCHAHT HFCOHDFR
FIG A 6
SCHEMATIC DIAGRAM OF THE UV PHOTOMETER.
122
-------
mixtures of NO (CP Grade) and Np through the test section, scanning the
monochromator over the wavelength range 2200-2300A and measuring the
absorption by the 7(0,0) band near 2260A. In these calibration runs the
nitric oxide mole fraction in the gas stream was varied from 5-0 x 10~3
to 8.0 x 10~5. A typical absorption spectrum, obtained from the cali-
bration experiments, is shown in Fig. A-7- In this spectrum, the rota-
tional structure and the sharply-defined band heads have been smeared
by the slit function of the monochromator. The observed absorption spec-
trum was processed using a recently-developed analytical technique (Ref.
39), to remove some of the effects of slit-broadening. Using the processed
(de-convolved) absorption spectrum, shown in Fig.A-7, an average absorption
coefficient for the T(o,0) band was defined by
a = : n_ f " du
U-i'
where a = average absorption coefficient (cm /mole), CJTQ = nitric oxide
concentration in the ground vibrational state (mole/cm3), I = 2 x test
section width (cm), Au = width of the absorption band (cm~^-) and Iu/1^ =
transmissivity at wave number u• There are two major uncertainties in
the optical system calibration. The first uncertainty is introduced by
imprecise determination of the extent of the absorption band,Aw, (Ref.
At low nitric oxide concentrations, the short-wavelength tail of the absorp-
tion band, Fig.A-7, cannot be accurately distinguished from system noise.
To reduce uncertainties associated with imprecise determination of Au, the
average transmissivity, Eq.. A-l, was calculated for a bandwidth corresponding
to lO/o of the maximum absorptivity. Using this arbitrary definition forAu,
room temperature values of the average absorption coefficient were obtained
which were essentially constant for the range of nitric oxide concentrations
in the calibration experiments. The second major uncertainty in the opti-
cal system calibration is associated with extrapolation of the room tempera-
ture values of average absorption coefficient to the temperature range of
the combustion experiments -- 1600 to 2000°K. The high temperatures and
low densities associated with the recirculation zone in the combustion
experiments produce changes in collisional and Doppler broadening of the
rotational lines in the 7(0,0) band. However, for the present experimental
conditions, collisional and Doppler broadening of the T(0,0)-band are not
significant compared with instrument broadening. Hence, variations in
the average absorption coefficient with temperature are neglected. At
123
-------
2
O
h-
Q.
£E
O
in
CO
OBSERVED SPECTRUM
PROCESSED SPECTRUM
WAVELENGTH - A
FIG. A-7 TYPICAL NITRIC OXIDE ABSORPTION SPECTRUM
NITRIC OXIDE MOLE FRACTION = 2 x 10~3
12 h
-------
elevated temperatures only a fraction of the nitric oxide molecules are
in the ground vibrational state. To calculate the total nitric oxide con-
centration from the measured ground-state concentration, the nitric oxide
was assumed to be in thermal equilibrium at the measured recirculation
zone temperature.
A.k Exhaust Concentration Measurement Technique
The concentration distributions of nitric oxide, methane, oxygen, carbon
dioxide and carbon monoxide in the exhaust gas from the combustion test
section were determined using a traversing sampling probe coupled to an
on-line time-of-flight mass spectrometer. A schematic diagram of the
sampling system and associated data processing equipment is shown in
Fig. A-8. The water-cooled, stainless-steel sampling probe rapidly
quenches the gas sample by an aerodynamic expansion across a choked
0.015-cm diameter orifice. A photograph of the sampling probe in the
combustor exhaust is shown in Fig.A-9- The quenched gas sample passes
through a heated, stainless-steel sample line to the inlet valve of the
mass spectrometer. The sample line wall was held at a constant tempera-
ture of 125°C, and the sample line pressure was maintained at 3-00 tO.05
torr. The time-of-flight mass spectrometer, used in the present investi-
gation, is a portable, high-resolution instrument (United Aircraft Research
Laboratories TOF-^0). The output signal of the mass spectrometer is pro-
cessed by a waveform eductor (Princeton Applied Research TDH-9) to enhance
signal-to-noise ratio. The waveform eductor stores the processed mass
spectra, which may then be displayed on a chart recorder (Hewlett-Packard
7100B) following completion of the experiment.
Prior to making exhaust concentration measurements, the probe sampling
system and mass spectrometer were calibrated. This calibration was
carried out by flowing known mixtures of methane, carbon dioxide, carbon
monoxide, oxygen, nitric oxide and nitrogen through the test section and
measuring the mass spectra of the gas at the test section exit. From the
known mixture composition and the observed mass spectra, mass discrimina-
tion effects in the sampling system and the fragmentation patterns for
various species of interest were determined. Mass discrimination effects
and fragmentation patterns are required if a unique determination of
exhaust gas composition is to be obtained from the measured mass spectra.
Although the mass spectrometer, used in the experimental study, is a high-
resolution instrument, nitric oxide concentrations in the exhaust gas were
125
-------
®
0-20TORR
PRESSURE GAGE
TRAVERSING
SAMPLING PROBE
TEST SECTION
FIG. A 8 SCHEMATIC DIAGRAM OF PROBE AND SAMPLING SYSTEM USED FOR EXHAUST CONCENTRATION
MEASUREMENTS.
126
-------
FIG. A-9 SAMPLING PROBE IN COMBUSTOR EXHAUST
127
-------
sufficiently small so that the primary nitric oxide mass peak (m/e = 30)
was overlapped to a certain extent by the m/e = 29 isotope of molecular nitrogen,
Fig. A-lOa. To improve the effective resolution of the mass spectrometer for
nitric oxide, the mass spectra in the range m/e = 28-30 were processed
using the analytical deconvolution technique previously discussed. The
processed (de-convolved) spectra are shown in Fig. A-10b.
In the combustion experiments, the mass spectra of the exhaust gas was
determined in the range m/e = 12-44. Many of the observed mass peaks in
this range are not unique to a single species -- for example, m/e = 16
has components due to methane, oxygen, carbon dioxide, carbon monoxide and
water. Hence to determine a unique exhaust gas composition, the observed
mass spectra must be processed to account for overlapping mass peaks. To
facilitate data processing a computer program was developed which uses
standard analytical techniques (Refs. 4l and 42) to extract a unique com-
position from the observed mass spectra.
The principal mass peak for nitric oxide is m/e =30. Of the major species
present in the exhaust, only nitrogen and carbon monoxide make significant
contributions to the observed m/e = 30 peak (due to the N^ and C^Q!" iso-
topes). Minor species which would contribute to the m/e = 30 peak, if
present in the exhaust, include formaldehyde (HoCO) and ethane (C^Ks}. The
observed mass spectra show that the concentrations of ethane and formal-
dehyde in the exhaust gas are below the detection threshold (-=20 ppm (v)).
Hence only contribution of nitrogen and carbon monoxide to the m/e = 30
peak must be accounted for. The carbon monoxide concentration in the
exhaust was estimated from the measured methane and carbon dioxide con-
centrations through a carbon balance. The carbon monoxide mole fractions,
inferred in this manner, ranged from 0 to 0.02. The correction on the
m/e = 30 peak for contributions from isotopic nitrogen and carbon monoxide
were in the range of 10-25 ppm (v).
To ascertain if nitric oxide is reduced to E^ in the stainless-steel
exhaust probe and sampling line in the presence of fuel, known mixtures
of nitric oxide and methane were sampled. Negligible depletion of nitric
oxide in these samples was observed.
128
-------
FIG. A-10a
TYPICAL OBSERVED MASS SPECTRA OF THE COMBUSTOR
EXHAUST IN THE RANGE m/e = 28-30.
FIG. A-10b
m/e 28 29 30
TYPICAL DECONVOLVED MASS SPECTRA.
129
-------
APPENDIX B
FIBER OPTIC MEASUREMENT TECHNIQUES IN THE COLD FLOW EXPERIMENTS
The fiber optic probe system (Fig.B-l) consists of: (1) a fiber optic
probe, (2) an optical photometer system, and (3) data acquisition and dis-
play electronics. High intensity light for the fiber optic probe system
is made available from a tungsten-halogen projection lamp powered by a
regulated dc power supply (Kepco model JQE-15-12(M)). The light output
from the lamp is directed to a narrow band pass opticeil filter (Baird-
Atomic 460 B9/450 B5) which passes light in the range 4550-4690 A, a wave-
length range near optimum for absorption by the W0o-No04 tracer material.
The filtered light is channeled into the test chamber via the transmitter
unit of the fiber optic probe. Subsequently, a fraction of the light is
absorbed by the tracer material and the remainder is channeled through the
detector unit of the probe to a photomultiplier tube (RCA 6l99) powered by
a high voltage dc power supply (Fluke model 4l2B); photomultiplier tube
operation typically is at 750 volts.
Photomultiplier system output is displayed simultaneously on a Hewlett
Packard digital volt meter and a cathode oscilloscope (Tektronix 531A)-
Oscilloscope traces are recorded photographically using a Polaroid camera.
The optical efficiency of each component in the optical photometer system
as a function of wavelength is given in Fig. B-2. The total system response
of approximately 10 percent of the input light intensity was obtained by
multiplying all component efficiencies at each wavelength.
The light absorption by the tracer material is a function of the absorption
coefficient, a, the tracer concentration, y^, and the spacing between the
transmitter and detector units of the fiber optic probe, d, according to
the Beer-Lambert law (Ref. 22):
I/Io = exp(-q.d-yt) /B-1\
An increase in any of the three parameters contained in Eq. (B-l) will
result in increased absorption; however, only the absorption coefficient
varies with wavelength.
130
-------
The tracer material used for the optical investigations consisted of an
equilibrium mixture of WC^ and NgOij.. According to Ref. ^3, absorption
above k-000 A is due exclusively to the concentration level of NC^. Hence,
increasing the amount of WOp in the tracer mixture by raising the tempera-
ture to shift the equilibrium to favor NC>2 production results in increased
absorption. The equilibrium composition of WC^-l^Olj. mixtures is given in
Pig.B-3aas a function of temperature. In order to establish satisfactory
absorption levels for the current experiments, it was necessary to operate
at elevated temperatures, see Fig.B-3b.
131
-------
TRACER
FLOW
TRANSMITTER-
FIBER OPTIC PROBE (~
L
r
1_
f
BAIRD ATOMIC
460 B9/4SO B5 FILTER
t
SYLVANIA FAV
TUNGSTEN HALOGEN
LAMP
i
KEPCO MODEL JOE
15-12IM) REGULATED
D C POWER SUPPLY
u
T
,
RCA 6199
IS-11CATHODE)
*
FLUKE MODEL 412B HIGH
VOLTAGE DC POWER
SUPPLY
1
1
1
I
1
1
1
1
1
1
1
1
J
1
M DIGITAL VOLTMETER
1 VOLT DC BUCKING
"" VOLTAGE & FILIfcH
*
ITECHTRONIX MODEL 513A
CATHORE RAY
OSCILLOSCOPE
i
1
CAMERA
ACQUISITION
SYSTEM
PHOTOMETER SYSTEM
I
FIG B 1
SCHEMATIC DIAGRAM OF FIBER OPTIC INSTRUMENTATION.
FRACTION TRANSMISSION
OF FIBER OPTIC BUNDLE
3 4 METERS LONG
390
FIG B 2
470 490
WAVELENGTH, X - nm
COMPONENT AND SYSTEM PERFORMANCE OF FIBER OPTIC PROBE FOR NO2-N2O4 TRACER.
570
132
-------
(a)EQUILIBRIUM MOLE FRACTION OF N02 IN N02-N204 MIXTURE
0.35
o
QC
0.30
O
CM
z
CN
O
•z.
u.
O
2
O
O
<
cc
0.25
0.20
0.15
TUNNEL OPERATING
TEMPERATURE
ROOM
TEMPERATURE
(b) TRANSMISSION OF N02-N204 MIXTURE AT 460 nm
0.80
UJ
DC
O 0.75
CM
z
I
CM
O
2 0.70
CO
to
2
cc
h-
£ 0.65
o
5
LU
a:
D.
ROOM
TEMPERATURE
TUNNEL OPERATING
TEMPERATURE
0.601
290
I
295 300 305
TEMPERATURE, °K
310
315
FIG. B-3
PROPERTIES OF NO2-N2O4 TRACER
133
-------
BIBLIOGRAPHIC DATA
SHEET
1- Report No.
EPA-R2-73-291
3. Recipient's Accession .S'.i.
4. I itle and Subtitle
Nitric Oxide Formation in Combustion Processes with
Strong Re circulation
5. Report Date
July 1973
6.
7. Author(s)
C.T. Bowman, L.S.Cohen, and M. N. Director
8. Performing Organization Kept.
No.
9. Performing Organization Name and Address
United Aircraft Research Laboratories
East Hartford, Connecticut 06108
10. Project/Task/Work Unit No.
11. Contract/Grant No.
68-02-0252
12. Sponsoring Organisation Name and Address
EPA, Office of Research and Development
NERC-RTP, Control Systems Laboratory
Research Triangle Park, North Carolina 27711
13. Type of Report & Period
Coveted
Final
14.
15. Supplementary Notes
16. Abstracts Thig rep()rts the resuits of an investigation of NO formation in a premixed
turbulent flame stabilized on the recirculation zone (RZ) downstream from a bluff-
body. Objectives were to investigate factors affecting NO formation in the RZ and to
assess the relative importance of NO production in the RZ to overall NO production.
To characterize NO formation in the RZ, dependence of NO formation on the RZ
volume, temperature and residence time of the gas in the RZ, and O-atom concen-
tration in the RZ were determined. NO production in the RZ was strongly influenced
by non-equilibrium chemical effects and by the turbulent exchange processes. Data
on turbulent exchange was obtained in a complementary investigation of RZ fluid
dynamics in non-reacting flows. Comparison of volumetric NO production rates in
the RZ with overall volumetric NO production rates indicates that, for the experi-
mental configuration, NO production in the RZ is not a major factor in overall NO
formation. An analytical model
17. Key Words and Document Analysis. 17a. Descriptors
Air Pollution
Nitrogen Oxide (NO)
Circulation
Combustion
Turbulent Flow
Mathematical Models
17b. Identifiers/'Open-Ended Terms
Air Pollution Control
Stationary Sources
Recirculating Flow
17c. COSATI Field/Group 21B , 13B
Action Agency
60606
for NO production in the burner
was developed. Results from this
model suggest that the RZ can be
a major factor in NO production
in practical combustion devices.
18. Availability Statement
Unlimited
19. Security Class (This
Report I
UNCLASSIFIED
20. Security Class (This
Page
UNCLASSIFIED
21. No. of Pages
22. Price
FORM NT1S-35 (REV. 3-72)
USCOMM-OC 14952-P72
------- |