IONS FROM AND WITHIN AN
ALLISON J-33 COMBUSTOR
Environmental Protection Agency Contract
No. 68-04-0001 - Final Report
A. M. Mellor, R. D^Anderson,
R. A. Altenkirch, and J. H. Tuttle
Report No. CL -72-1
THE COMBUSTION LABORATORY
SCHOOL OF MECHANICAL ENGINEERING
PURDUE UNIVERSITY
LAFAYETTE, INDIANA
June 1972
-------
EPA-R2-72-097
Emissions from and within an
Allison J-33 Combustor
Environmental Protection Agency Contract
No. 68-04-0001 - Final Report
by
A. M. Mellor, R. D. Anderson, R. A. Altenklrch. and J. H. Tuttle
Report No. CL-72-1
The Combustion Laboratory
School of Mechanical Engineering
Purdue University
Lafayette, Indiana
June 1972
-------
ACKNOWLEDGEMENTS
Special thanks are extended to Mr. Thomas Miller for his assistance
in the construction and maintenance of the experimental facility and also
for his assistance in obtaining the data contained in this report.
Messrs. P. Leonard, M. Franzen, S. Jochem, R. Shisler, S. Jones,
G. McFarron, and K. Gross greatly aided in both the construction and
maintenance of the experimental facility and data gathering and analysis.
The initial phase of the facility development was conducted under
Contract DAAE07-69-C-0756 with the Army Tank and Automotive Command;
subsequent facility development and the research described herein were
supported by the Environmental Protection Agency under Contract 68-04-
0001. G. Kittredge, B. McNutt, and K. Hellman of the Office of Air
Programs provided valuable assistance.
-------
TABLE OF CONTENTS
Page
LIST OF TABLES v
LIST OF FIGURES vi
ABSTRACT ix
SECTION I. INTRODUCTION AND SUMMARY 1
SECTION II. DESIGN AND OPERATION OF THE
EXPERIMENTAL GAS TURBINE COMBUSTION FACILITY 4
A. Test Cell Hardware 5
1. A1r System Description 5
2. Fuel System Description 7
3. Combustor Description 7
4. Probe Addition Section Description 9
5. Probe Description 9
6. Probe Positioner Description 20
7. Back-Pressure System Description 23
8. Water System Description 28
B. Control Room Instrumentation 28
1. Temperature Monitoring Instrumentation 30
2. Emission Monitoring Instrumentation 31
2.1 The Carbon Monoxide Analyzer System 31
2.2 The Nitric Oxide Analyzer System 33
2.3 Total Hydrocarbon Analyzer System 40
C. Summary 40
SECTION III. RESULTS AND DISCUSSION 42
A. Combustor Operating Points 47
B. Internal Measurements 50
1. Gas Temperatures 50
2. Nitric Oxide 54
3. Carbon Monoxide 59
4. Summary 64
C. Combustor Exhaust Plane Measurements 64
1. Gas Temperatures 66
2. Exhaust Plane Emissions of Nitric
Oxide and Carbon Monoxide 66
2.1 The Influence of Air Flow Rate 69
2.2 The Influence of Overall Equivalence
Ratio 74
2.3 The Influence of Combustor Pressure 75
D. Other Parameters 84
E. Summary 84
-------
IV
Page
SECTION IV. FUTURE EFFORTS 86
LIST OF REFERENCES 93
APPENDIX A: EFFECT OF N02 CONVERTER ON NO 95
-------
LIST OF TABLES
Table Page
2-1. 0-33 Design Resume 11
2-2. NO Interference Data (after Fontljn
et al., 1969, 1970) 39
3-1. Selected Combustor Operating Points 49
4-1. Operating Point and Emissions Obtained
from Two J-33 Combustors in Series 89
-------
LIST OF FIGURES
Figure Page
2-1. Combustion facility air system 6
2-2. Combustion facility fuel system 8
2-3. Combustor arrangement 10
2-4. Combustor and probe addition section 12
2-5. Combustion facility schematic 13
2-6. Gas sampling probe body 16
2-7. Gas sampling probe block 17
2-8. Gas sampling probe tip 18
2-9. Gas sampling probe 21
2-10. Gas sampling probe and
positioning systems 22
2-11. Axial and radial (black box)
probe positioners 24
2-12. Back pressure valve schematic 26
2-13. Back pressure valve guidance
system and engine 27
2-14. Combustion facility water system 29
2-15. Gas handling system 32
2-16. CO analyzer system 34
2-17. Nitric oxide detector schematic 36
2-18. Hydrocarbon sample handling and
analyzer system 41
3-1. Typical gas turbine combustor
flow schematic 43
3-2. J-33 combustor configuration, gas
sampling planes (4 of 6), and
probe trace position 45
-------
vn
Figure Page
3-3. Relationship of probe trace and gas
sampling points to J-33 combustor
configuration 46
3-4. Combustor operating point matrix 48
3-5. Temperature versus axial position
(ma = 6.0 Ibs/sec, $ = .217,
p = 5 atm) 51
3-6. NO concentration versus axial posi-
tion for 0° (ma = 6.0 Ibs/sec,
$ = .217, p = 5 atm) 55
3-7. NO concentration versus axial po-
sition for 90° ccw (ma = 6.0 Ibs/sec,
= .217, p = 5 atm) 58
3-10. CO concentration versus axial po-
sition for 0° (ma = 6.0 Ibs/sec,
<{> = .217, p = 5 atm) 60
3-11. CO concentration versus axial po-
sition for 90° ccw (ma = 6.0 Ibs/sec,
4> = .217, p = 5 atm) 61
3-12. CO concentration versus axial po-
sition for 90° cw (ma = 6.0 Ibs/sec,
4> = .217, p = 5 atm) 62
3-13. CO concentration versus, axial po-
sition for 180° cw (ma = 6.0 Ibs/sec,
4> = .217, p = 5 atm) 63
3-14. Schematic of postulated primary and
secondary zone flow pattern for the
J-33 combustor 65
3-15. Combustor exit plane temperature
traverse (probes in locations
numbered one in Fig. 3-3) 67
-------
vi i i
Figure Page
3-16. Combustor exit plane temperature
traverse (probes in locations
numbered two in Fig. 3-3) 68
3-17. Radial NO concentrations at combustor
exit plane versus air flow rate
(average) 70
3-18. Radial NO concentrations at combustor
exit plane versus air flow rate 71
3-19. Radial CO concentrations at combustor
exit plane versus air flow rate
(average) 72
3-20. Radial CO concentrations at combustor
exit plane versus air flow rate 73
3-21. Radial NO concentrations at the combustor
exit plane versus overall equivalence
ratio (average) 76
3-22. Radial NO concentrations at the combustor
exit plane versus overall equivalence
ratio 77
3-23. Radial CO concentrations at the combustor
exit plane versus overall equivalence
ratio (average) 78
3-24. Radial CO concentrations at the combustor
exit plane versus overall equivalence
ratio 79
3-25. Radial NO concentrations at combustor
exit plane versus pressure (average) .... 80
3-26. Radial NO concentrations at combustor
exit plane versus pressure 81
3-27. Radial CO concentrations at combustor
exit plane versus pressure (average) .... 82
3-28. Radial CO concentrations at combustor
exit plane versus pressure 83
4-1. Complete combustion facility schematic .... 87
A-l. Normalized concentration versus inverse
temperature 97
-------
ABSTRACT
Because of the possible future widespread use of the gas turbine
engine in automobiles, and the imminence of federal emission standards
for aircraft, gas turbine combustion and emission characteristics are
currently being investigated in earnest. Much information concerning
specific pollutant concentrations as measured at the engine exhaust is
presently available. Several analytical combustor modeling programs
have also been developed (Mellor, 1971). The basic processes of pol-
lutant formation and destruction occurring within the combustor have,
however, received little attention to date. In an effort, therefore,
to obtain this fundamental type of experimental information, a gas
turbine combustion facility has been designed and constructed.
After a review and description of the combustion facility, re-
sults are presented which include gas temperature, carbon monoxide,
and nitric oxide concentration profiles as a function of axial and radial
position inside an Allison J-33 combustor. In addition some combustor
exit plane measurements are reported. Specifically, the isolated effects
of combustor pressure, overall equivalence ratio, and air flow rate on
CO and NO concentration at various radial positions are investigated.
These results are qualitatively explained in terms of basic combustor
processes.
Unheated combustor inlet air was used for the above studies; a few
preliminary experiments using heated air are also described.
-------
I. INTRODUCTION AND SUMMARY
Emissions from various types of combustion systems have been in-
vestigated in some detail in recent years. Government legislation has
attempted to force the advancement of emission control technology through
the establishment of federal emission standards for many mobile and sta-
tionary pollution sources. As effort has been expended to meet the fed-
eral emission standards, continuous flow combustion systems have received
increasing attention. A continuous flow combustion device of wide appli-
cation is the gas turbine.
Because reasonably precise control of the combustion process is
possible in a gas turbine combustor, it is felt to possess inherent emis-
sion control advantages over other types of combustion systems. In a
gas turbine combustor, for example, it is theoretically possible and
practically realistic to tailor the combustion zones to those conditions
of local equivalence ratio best suited to total emission reduction. In
general, this tailoring of the combustion zones is made possible because
of the relative freedom of the combustor designer to specify the combus-
tion chamber configuration. This configuration, of course, controls
such things as fuel and air flow distributions, recirculation patterns,
residence times, and other parameters of importance in emission control.
In contrast, the combustion event in the spark or compression igni-
tion engine combustion chamber is difficult to control. Those techniques
which have been utilized to reduce spark ignition engine emissions, such
as the alteration of ignition timing and exhaust gas recirculation, have
in general resulted in the deterioration of that engine's performance.
-------
Therefore to couple good vehicle performance with a minimum of pollu-
tant emission is one reason that the gas turbine engine is being con-
sidered as a possible replacement for spark and compression ignition
engines in many vehicular applications.
Because of the possible future widespread use of the gas turbine
engine in automobiles, and the imminence of federal emission standards
for aircraft (which are in large part powered by gas turbine engines), a
definite need exists for fundamental information concerning gas turbine
emissions. It is most logical to begin a basic study of gas turbine
pollution by investigating the effect of combustor operating parameters
on pollutant emissions. It is, however, imperative that the operating
parameters be varied individually and that the effect of each single
operating parameter change upon pollutant emission levels be noted. To
this end, an experimental program, to be discussed in detail in the fol-
lowing pages, has been started.
Briefly, the experimental setup consists of a J-33 combustor burning
liquid propane. Variable engine parameters are combustor pressure, equiv-
alence ratio, and air flow rate. Later experiments will also include
combustor inlet air temperature. The condition of the engine is monitored
remotely during any experimental investigation. A continuous gas-sampling
system for the determination of CO, NO and HC concentrations is employed.
Samples can be extracted at the exit plane of the combustor as well as at
various internal locations. Temperatures are monitored both at the exit
plane and internally, using a probe-mounted Pt/PtRh thermocouple.
Exit plane and internal combustor CO and NO concentrations and gas
temperatures have been collected. Temperature measurements indicate
that strong radial gradients exist, a result of the influence of penetration
-------
jets. CO and NO concentration measurements at various axial positions
in the combustor yield the same trends reported in the literature. These
internal measurements were for a single operating point.
Exit plane CO and NO concentrations at various radial positions were
studied as functions of combustor pressure, equivalence ratio, and air
flow rate. Combustor centerline CO concentrations increased with pressure;
other radial positions exhibited more erratic behavior. NO concentrations
generally increased with pressure. CO and NO exit plane concentrations
increased with increasing equivalence ratio for almost all radial positions;
this trend is consistent with the observations of other investigators.
Increasing air flow rate caused the NO concentrations to decrease only
slightly and CO to increase.
-------
II. DESIGN AND OPERATION OF THE
EXPERIMENTAL GAS TURBINE COMBUSTION FACILITY
Once the determination was made to pursue in earnest gas turbine
pollutant emissions research, the design and construction of an exper-
imental facility capable of realistically simulating gas turbine com-
bustion was necessary. The design, construction, and testing of such
a facility occupied the major portion of the preceding year's experi-
mental program. The facility was to be capable of simulating both auto-
motive and aircraft type gas turbine combustion realistically so that
combustor operating parameters (pressure, air flow, and overall equiv-
alence ratio) could be varied to determine fundamental trends concerning
gas turbine emissions. With the exception only of inlet air temperature
(which was not varied for most of the experiments conducted to date), it
is felt that the facility to be described is in fact such a realistic
simulation and that the data obtained and reported in the following sec-
tion verify the facility's ability to provide the desired fundamental gas
turbine emission information.
This chapter essentially is divided into two sections. The first
section provides a description of the test cell hardware and those aux-
iliary systems necessary for combustor parameter control or component
survivability. The second section of this chapter entails a description
of control room instrumentation. Included are descriptions of those
systems required for monitoring the combustor condition, various temper-
atures associated with the combustion process, and most significantly,
the combustor emissions.
-------
A. TEST CELL HARDWARE
1. Air System Description
Because of the large mass flow rates and high compressor pressure
ratios used in many modern aircraft gas turbine engines, a large air
compressor network is essential to any research laboratory attempting to
study realistic gas turbine combustion. The air compressor system which
supplied the test facility had a capacity of approximately 3000 cu ft
and could provide air system pressures of up to 2400 psig. Sustained
air flow rates of 6 Ibs/sec were possible for run durations of approx-
imately 40 minutes under normal combustor operating conditions. In ad-
dition, the basic combustion system could pass air flow rates of up to
10 Ibs/sec for short periods. Although the main air system compressors
were in service throughout any run situation, the system was basically
of the blowdown type. To date the air system has proven satisfactory for
the needs of the program.
The basic test cell air system is shown schematically in Fig. 2-1.
The three inch main air supply ducting was of heavy wall carbon steel,
and all high pressure control lines were stainless steel. The main air
supply system included two 3000 psig inlet capacity remotely controlled
air regulators mounted in parallel, a safety burst disc, a 2 1/8 inch
orifice type flow meter, a pneumatically operated air throttle valve,
and finally a diffuser section mounted immediately upstream of the com-
bustor. Orifice plate differential pressure was monitored continuously
on a differential pressure gage in the control room and recorded perma-
nently and continuously through a differential pressure transducer/Ellis
bridge amplifier/Honeywell Model 1508 Visicorder system.
-------
Main Air Line
a
X-D
Y
Instrument
Air Line
Building Air
Burst
Disc
ir Throttle
Valve
Diffuser
Z-
I
To Water System
Figure 2-1. Combustion facilitv air svstem
-------
2. Fuel System Description
The fuel used in the combustion study was commercial purity liquid
propane since it is inexpensive and readily available. The fuel system
is shown schematically in Fig. 2-2. When the 500 gallon storage tank
was pressurized with nitrogen, the five 10 in x 10 ft cylindrical steel
fuel delivery tanks were filled from the storage tank. This tank was
then isolated, and the fuel delivery tanks were pressurized to the desired
supply pressure using a ten bottle nitrogen supply manifold. All fuel
flow was passed through a suitable fuel filter and then underground to
the test cell area. Fuel flow rate was controlled through a pneumati-
cally operated fuel throttle valve and measured using a Potter turbine
flow meter. Frequency converters were used to convert the flow meter
output to a calibrated mi Hi amp output signal. Like the air flow, fuel
flow was also permanently and continuously recorded using a six channel
Honeywell Visicorder.
In the interests of safety the main fuel system tanks were positioned
away from the test cell area, stainless steel lines and fittings were
extensively used, both manual and solenoid type vent valves and automatic
pressure relief valves were judiciously placed at various system locations,
and all fuel system pressures were continuously monitored. The fuel sys-
tem as shown in Fig. 2-2 has performed satisfactorily to date.
3. Comb us tor Description
In the initial facility configuration all combustion occurred in an
Allison J-33 turbojet combustor mounted in a stainless steel converging
housing. The combustor housing, in turn, was secured to a thrust table.
The standard J-33 fuel nozzle and conventional magneto-energized igniter
-------
Fuel
Delivery
Tanks
VP"
fuel Filter
in
-$—Kh-&
Fuel
Storage
Tank
Nitrogen
Manifold
Fuel Tlirottle Valve
Turbine Flow Meter
J-33 Combustor
Figure 2-2. Combustion facility fuel system
-------
were employed. The basic combustor arrangement is shown in Fig. 2-3.
In order to provide the necessary experimental data for a simultaneous
analytical heat transfer study, the J-33 flame tube was outfitted with
several chromel-alumel thermocouples; the results of this study have
been reported by Owens (1972). Through the courtesy of the Detroit
Diesel Allison Division of General Motors, the J-33 design resume shown
in Table 2-1 was obtained.
4. Probe Addition Section Description
The J-33 combustor casing exit was flanged to accept the probe ad-
dition section as shown in Fig. 2-4. The purpose of this stainless steel
section was to change the direction of the exhaust gas flow such that a
probe could be directly inserted into the combustor. The surface of this
section was wrapped with copper water cooling lines, and the probe window
was cooled with injected nitrogen. Fig. 2-5 schematically shows the over-
all facility configuration.
5. Probe Description
The combustion chamber of a gas turbine engine is a most severe
environment. It is characterized by relatively high mass flows, exces-
sive turbulence, large temperature gradients, high pressure in many
cases, recirculating flow patterns, and extremely high local temperatures
which in some locations approach the adiabatic flame temperature. The
task of designing a probe capable of surviving this type of environment
for long periods of time is very challenging. Some probe failures have
in fact been mentioned in the open literature for probes attempting to
sample gas turbine engine exhaust at the engine exit station (Hare et al.,
-------
Igniter
Fuel Line
o
o
vl
D.
/
\
Combustor
/>
Combustor Casing
Fuel Nozzle
Figure 2-3. Combustor arrangement
-------
Table 2-1. J-33 Design Resume
Military Air Flow
Normal Air Flow
Military Inlet Temperature
Normal Inlet Temperature
Military Inlet Pressure
Normal Inlet Pressure
Military Volume Flow
Normal Volume Flow
Average Pressure Loss
Liner Cooling Air Flow
Primary Combustion Air Flow
Secondary Combustion Air Flow
7.64 Ibs/sec
7.17 Ibs/sec
435° F
390° F
141 in Hg abs
126 in Hg abs
40.7 cu ft/sec
40.9 cu ft/sec
4.5%
8.0%
23.4%
68.3%
-------
purge
J-33 Corabustor
Figure 2-4. Combustor and probe addition section
Sampling
Probe
-------
Air System
Thrust Stand
Fuel System
Transition Section
Back-Pressure
Valve
Probe Addition Section
Exhaust
Probe Positioner
Figure 2-5. Combustion facility schematic
-------
14
1971). It is apparent that the engine exit station, although itself a
severe environment, is orders of magnitude less severe than the combustion
chanter. The probe design which is to be reported below was found able
to withstand the combustor environment. One probe has sampled combustor
exit plane gases for approximately eleven hours with only the develop-
ment of a relatively minor probe tip leak during the eleventh hour and
has also sampled within the combustor for approximately 85 minutes (in-
cluding a large fraction of that time sampling from the primary zone).
A second probe of similar design has withstood over two hours inside the
combustor.
The development of the final successful probe design reflected the
difficulties mentioned above. Two early probe designs failed quickly
but did provide through their failure the information needed to design
an acceptable probe.
The intent of the probe was to sample gases and monitor gas temper-
ature at various axial and radial positions within the combustor or at
the combustor exit. For purposes of gas temperature measurement a plat-
inum/10% platinum rhodium thermocouple (MgO insulation, sheathed in in-
conel) was mounted through small "half-moon" shaped sections directly to
the side of a gas sampling probe; the resulting bead position was about
1/16 in ahead of the gas sampling tip. The thermocouple was of the ex-
posed junction type and was intended to provide only gross estimates
of the local combustor temperatures. A 1/16 in diameter thermocouple
with .01 in wires failed, however, to provide the desired temperature
estimates. Typical behavior was as follows: upon insertion of the probe
into the combustor the thermocouple provided reasonable temperature es-
timates. After the probe was rotated through approximately 90°, however,
-------
15
the thermocouple signal became very erratic possibly indicating an inter-
mittent breaking and making or short-circuiting of the thermocouple lead.
The same behavior was noted with two other Pt/10%PtRh thermocouples of
similar dimensions. Consideration of the very brittle nature of the plat-
inum leg of this thermocouple indicated that the failure of the two ther-
mocouples was caused, most probably, by stresses in the platinum leg
induced through the mechanical rotation of the probe. New 1/8 in diameter
thermocouples with .025 in wire diameter and with the platinum leg re-
placed by "Fibro Platinum" (which is better able to withstand turbulent
high temperature conditions) have been mounted to the probe for subsequent
temperature measurements and proved satisfactory for the experiments con-
ducted to date.
The probe was inserted into the J-33 combustor through the probe
addition section (see Fig. 2-4} and supported both at the probe window
and at the cradle of the probe positioner (to be described later). The
gas sampling portion of the probe body consisted of three concentric
316 stainless steel tubes as shown in Fig. 2-6. The relative position
of the three concentric tubes was maintained through the use of several
small welded beads at various locations along the length of the probe,
and three distinct passages inside the probe body were thus assured. Pres-
surized cooling water entered the probe at the probe block (Fig. 2-7) and
flowed to the probe tip in the outer annul us. At the probe tip the cooling
water flow was reversed (see Fig. 2-8) and proceeded through the center
annul us back to the probe block where it exited. The center probe body
tube carried sample gases to the gas sampling system delivery lines.
Cooling water flow pressure was variable to approximately 250 psig.
-------
3" ,.
TT dia.
H2
-------
Plug
Scale: Full
A-A
Figure 2-7. Gas sampling probe block
-------
L.016" H2° -
Scale 6:1
B-B
Figure 2-8. Gas sampling probe tip
00
-------
19
The design of the gas sampling tip reflects consideration of the
requirements inherent in any suitable sampling probe design as well as
consideration of those factors peculiar to the particular combustor
environment encountered. In order to obtain accurate information con-
cerning comfaustor gas composition, the composition of the gases at the
entrance of the gas sampling probe must be essentially identical to the
gas composition at the entrance of the emission monitoring instrumenta-
tion. In other words, all gas reactions must be quenched in the short-
est possible distance in the sampling probe. One ideal way in which
to accomplish this immediate quenching entails the use of a converging-
diverging probe tip nozzle for rapid sample flow acceleration and ex-
pansion. As the sampled gases then flow through the probe body, additional
reduction in gas static temperature may be realized through the added
quenching effect provided by the probe cooling water flow. These two com-
bined quenching methods have been used frequently in the past in elemen-
tary flame studies and other assorted combustion experiments.
However, one of the operating characteristics of the J-33 cornbustor-
fuel nozzle combination used in the subject research was the rapid increase
in carbon formation associated with operation at superatmospheric pressures.
Consequently, the final probe tip design shown in Fig. 2-8 represents a
compromise between the ideal sampling probe and a probe able to withstand
the combustor environment. For example, the ideal sampling probe would be
of infinitesimal diameter to prevent flow disturbance whereas a working
probe must be of sufficient diameter to allow a suitable cooling water
flow, as well as provide sufficient structural rigidity to withstand the
combustor environment. Similarly, an ideal sampling probe tip would
-------
20
consist of a converging-diverging nozzle; however, such a tip could not
be constructed so that mechanical integrity could be maintained and
blockage by soot prevented. Thus the final tip design and dimensions
shown in Fig. 2-8 were finalized. As an added safeguard against the
possibility of carbon blockage, the probe gas sampling tube was purged
with nitrogen during all combustor starts, all purposely induced com-
bustor transients, all probe rotations, and all combustor shutdowns.
Whereas probe blockage had consistently occurred with the earlier probe
designs, this major inconvenience was successfully avoided through the
use of the above mentioned operating procedures and probe design.
In order to obtain a circular probe trace upon probe rotation,
the tip section of the probe was offset from the probe centerline by a
distance of 1 9/32 in following Sawyer et al. (1969). This offset, as
well as probe body, block, mounting plate, and drive gear are shown in
Fig. 2-9.
6. Probe Positioner Description
As previously mentioned, gas sampling and temperature measurements
were made inside the J-33 combustor as a function of axial and radial
position. For the purpose of providing accurate, reproducible, and re-
motely controllable positioning of the probe tip within the combustor,
the probe positioning system as shown pictorially in Fig. 2-10 was con-
structed. The circular probe block was positioned inside a roller bear-
ing pressed into the probe mounting plate. The probe mounting plate in
turn was attached adjustably to the positioner carriage. Remotely con-
trollable longitudinal movement of the positioner carriage was accomplished
through a worm gear/electric motor combination. A series of micro-switches
-------
Figure 2-9. Gas sampling probe
-------
Figure 2-10. Gas sampling probe and
positioning systems
r )
I j
-------
23
(see Fig. 2-11) aligned parallel to the carriage travel and actuated by
a cam extending from the positioner carriage provided accurate control
room information concerning the probe tip axial position by lighting ap-
propriate control panel lamps. To achieve remotely controllable probe
rotation an electric motor/drive gear combination was mounted on the
thrust stand near the probe addition section window. This drive gear
then meshed with the probe gear (seen in Fig. 2-9) which was welded to
a collar mounted on the probe body. A small, rounded pin extended
through the collar and traveled in a smooth longitudinal slot which had
been machined in the probe body. In this way the probe body was allowed
to slide axially through the collar but was prevented from rotating rel-
ative to the collar. A helipot driven by the probe gear and constituting
part of a battery powered electrical circuit impressed a variable voltage
on the Honeywell Visicorder. After calibration of this assembly, the
probe radial position was continuously and permanently recorded via the
oscillograph trace. Both axial and radial remote probe positioning
schemes performed satisfactorily through the extent of the research ef-
fort.
70 Back-Pressure System Description
In order to simulate modern gas turbine combustion, certain require-
ments are imposed upon combustor pressure. In general aircraft gas tur-
bine applications, combustor pressures on the order of 15 atmospheres
are common. However, in automotive gas turbine applications combustor
pressures rarely exceed six atmospheres. Therefore, in an attempt to
construct a facility acceptable for general gas turbine combustion re-
search, it was necessary to provide the capability for continuous combustor
-------
Figure 2-11. Axial and radial (black box)
probe positioners
i j
-i-
-------
25
pressure regulation from 1-15 atmospheres. The device designed for
this purpose and shown in Fig. 2-12 and 2-13 consists of a translating,
conical center-body, a beveled engine exit plane surface, and two con-
centric cylinders.
To insure the survival of the center-body in the hot exhaust gas
stream, pressurized cooling water was forced through a large number of
small holes drilled through the front center-body cone. This method of
cone cooling essentially placed a protective sheet of water over the
leading cone surface and to date has successfully prevented any notice-
able erosion of the center body cone.
As can be seen in Fig. 2-13 the translating section of the back-
pressure valve assembly was mounted on a dolly and accurately guided
in movement through the use of linear bearings and precisely positioned
hardened and ground steel guide shafts. Adjusting mechanisms were pro-
vided on the dolly to insure the accurate positioning of the center-body
cone and concentric cylinders.
Control of combustor pressure was effected through the action of
three air cylinders mounted on the engine transition section (Fig. 2-13)
and connected to the translating section of the back-pressure valve. By
increasing air cylinder pressure the back pressure valve cone was forced
forward effectively reducing the engine exit area and thereby increasing
combustor pressure. The positioning of heavy springs against the piston
head on the unpressurized side of each air cylinder eliminated the jerking
effects of starting friction on the translating section motion and also
insured that atmospheric combustor pressure would be realized in the event
of an air system failure. To date relatively precise and remote control
-------
Air Cylinder
H20 Inlet
Translating
Cylinder
V ,\ _ V \ \\V\\\\ \ \ \
Center Body
Transition Section
Cooling Holes
Figure 2-12. Back pressure valve schematic
-------
27
Figure 2-13. Back pressure valve guidance
system and engine
-------
28
of combustor pressure to a value of 10 atmospheres has been routinely
achieved.
8. Hater System Description
In order to provide adequate cooling for the back-pressure valve
cone and the gas sampling probe, a high pressure water supply system
was designed and installed and is schematically shown in Fig. 2-14.
The total system capacity was approximately 1000 gallons and could sup-
ply water as necessary at pressures up to 400 psig. For the purposes
of insuring satisfactory water flow control, a remotely actuated main
water throttle valve was installed and found to function well. All
system pressures were monitored continuously from the control room.
B. CONTROL ROOM INSTRUMENTATION
Prompted by the magnitude and complexity of the experimental re-
search effort and in the general interest of safety, all test cell
activity was monitored from a separate control room. The instrumenta-
tion assembled within the control room can be conveniently divided into
three broad categories. The first consists of that instrumentation
necessary for the safe and accurate control of all systems relating to
the actual operation of the main combustion system. These systems have
been discussed previously and, therefore, need no further comment. The
second class of instrumentation to be briefly discussed includes all
temperature monitoring systems assembled for the research. The final
instrumentation category, and one of utmost importance to the objectives
of the research, includes all continuous flow emission monitoring
-------
9
Main
Air
Line
1000 Gallon
Water Storage
Tank
Vents
-S—
i-z
Water
Supply
Water
Softener
->- Back Pressure Valve Cooling Water
>- Probe Cooling Water
Figure 2-14. Combustion facility water system
-------
30
equipment. These latter systems will be described fully in the follow-
ing section. A six channel Honeywell Model 1508 Visicorder was used to
record the output of the CO, NO, and HC detectors, as well as probe po-
sition and air and fuel flow rates, as mentioned previously.
1. Temperature Monitoring Instrumentation
In order to monitor the condition of the combustion facility
continuously, it was necessary to measure certain important system
temperatures. Included in this list were local combustor temperature
as measured through the use of a probe-mounted platinum/10% platinum
rhodium exposed junction thermocouple. At the combustor exit sampling
plane, combustor exit temperatures were measured with two United Sensor
Corporation aspirated and radiation shielded chromel-alumel thermo-
couples. These thermocouple probes were positioned with compression
fittings through the combustor casing and could be moved manually to
various, accurately known radial depths into the combustor exhaust plane.
Specially designed thermocouple positioners were used for this purpose.
All thermocouple outputs were recorded continuously on Honeywell class
15 self-balancing potentiometers. The data to be reported were all
referenced to a temperature of 0° C.
In addition, the back-pressure valve front cone temperature was
monitored during the initial testing of the back-pressure system. Once
it was determined that the cone water cooling provisions were more than
adequate to insure the cone's survival, this thermocouple was taken out
of service.
The final thermocouple employment was in the collection of flame
tube wall temperature data to be used in conjunction with an analytical
-------
31
gas turbine heat transfer study (Owens and Mellor, 1971, Owens, 1972).
For this purpose the J-33 flame tube was outfitted with a series of ten
chromel-alumel thermocouples. The results of this investigation as well
as information concerning the relative positions of the thermocouples
and flame tube cooling slots and penetration air holes are presented in
Owens (1972).
2. Emission Monitoring Instrumentation
The gas concentrations measured during the present study were pre-
dominantly CO and NO, as discussed in Section I; the overall gas sampling
system as well as the specific CO, NO, and hydrocarbon detection systems
are fully described below.
The sample gas handling system is shown schematically in Fig. 2-15.
Unless otherwise noted, all lines, fittings, and other components were
stainless steel and/or teflon coated to minimize surface reactions or
adsorption causing distortion of the true gas composition as the sample
gas was transported to the various analyzers; essentially the techniques
recommended by Chase and Hum (1970) were used.
All instrumentation received span and zero checks before and after
each run, as well as frequent zero checks during each run. In addition,
the entire sampling system was inspected weekly and cleaned as required.
2.1 The Carbon Monoxide Analyzer System
A Beckman Model 315A (short path) NDIR (non-dispersive infrared)
analyzer was used for the continuous monitoring of sample gas CO con-
centration. The analyzer section of this instrument employed a stacked
sample cell configuration with both a 13 1/2 inch cell for a useful
range of 0-250 ppm (without back pressure), and a 1/8 inch cell for a
-------
Gas Sampling Probe
Relief
Valve
Purge System
rurgu oybtum i
M2 I f
Converter
NO
Detector
(Chem.)
Bypass
Flow
CO Analyzer
(NDIR)
Heated
Line
Bellows Pump
Vacuum
Pump
I
Vacuum
Pump
Total Hydrocarbon
Analyzer
(flame ionization
technique)
•"igure 2-15. Cas handling system
ts)
-------
33
range of 0-20%. In addition, the CO detector flow system incorporated
a back-pressure regulator in the 0-250 ppm flow line, which allowed the
cell pressure to be increased to 75 psig (limited by the thickness of
the sapphire windows in the cell). This modification allowed the lower
range sensitivity to be increased to 0-100 ppm full scale. Repeatability
for the NDIR measurements was guaranteed to be 1% of full scale reading
for both ranges. The specific CO analyzer flow system is shown schemati-
cally in Fig. 2-16. Normal CO concentrations encountered in the J-33
combustor dictated the constant use of the 0-20% range.
Condensate traps positioned in the sample line and capable of with-
standing 10 atmospheres pressure were reported by Beckman to eliminate
the problem of water vapor interference on sampling measurement accuracy.
However, it should be mentioned that the presence of C02 in concentrations
on the order of 300 ppm can cause approximately a 10 ppm CO measurement
error.
The normal CO detector sample flow rate was 1600 cc/min. Since for
future run conditions low or atmospheric combustor pressure (not conducive
to natural sample flow) might be desired, a vacuum pump was installed to
insure adequate sample flow to the CO analyzer. For the experiments de-
scribed in Section III (with combustor pressures of three atm and above),
the CO analyzer response time was on the order of one second, indicating
rapid passage of the gas sample from the combustor to the instrumentation.
2.2 The Nitric Oxide Analyzer System
The second gas constituent monitored was nitric oxide, and was meas-
ured continuously using a chemiluminescent technique similar to that
recently developed by Fontijn et al. (1969, 1970). This method involves
-------
.._ Water Traps
3-0-
Sample
Inlet
Calibrated Gas Manifold
(
I -rr!
« &
I
H '
Span
Zero
Filter
jl/8"
Cell
r
0-20%
13 1/2" Analyzer Cell
0-100 ppm
Filter
(
»J7 p<
^•/ '
-------
35
the measurement of the intensity emitted by the chemiluminescent reaction
between nitric oxide and ozone. The NO detector designed and constructed
for the present research is shown schematically in Fig. 2-17.
After the completion of the initial construction of the NO detector,
the linearity of the instrument and its dark current behavior were thor-
oughly investigated. For the linearity study a gas mixture of known NO
concentration was prepared by a partial pressure technique. A known vol-
ume of this prepared NO/N? mixture was then injected into the system's
exponential dilution flask while N~ pressure upstream of the dilution
flask was maintained at a constant pressure of 100 torr. Total reaction
flask pressure was maintained at 3 torr through a microvalve throttling
control. The dilution flask nitric oxide concentration will vary ac-
cording to the equation (Fontijn et a!., 1969, 1970):
C = CQ exp (-Qt/V)
where
C = initial NO concentration in the injected sample
Q = volume flow rate
V = dilution flask volume
t = time
Therefore, if the electrometrometer net signal (i.e., total signal less
dark current), when plotted on a semi-log scale versus time, is a straight
line, the linearity of the instrument has been shown. After every NO de-
tector system modification, a linearity check was accomplished. The system
linearity was always realized.
-------
Gas
Sample
Shutoff
Valve
Micro-Valves
Pressure
Regulator
ressure
Gauge
Flow
Meter
J^ Three-way
Valve
Exponential!
Dilution
Flask
\
Calibration Gas
Magnetic
Manometer
Photomultiplier
Tube
High Voltage
Power Supply
Manganese Dioxide Filter
Vacuum Pump
Figure 2-17. Nitric oxide detector schematic
ON
-------
37
With the assurance of linearity, meaningful calibration of the NO
detector was then accomplished with the use of only one span gas of
known NO concentration (usually 202 ppm in Nj. This gas was passed
through the NO detector and the electrometer signal noted. The dark
current electrometer signal was also recorded. From the linearity check
the calibration curve was known to be a straight line. The two points
then determining the calibration graph of NO concentration versus net
electrometer signal were the known concentration and net signal of the
single span gas and the zero concentration point characterized by a net
signal of zero. The NO detector was calibrated using the above single
span gas technique before and after every experimental run.
The method of operation of the nitric oxide detector is determined
by the chemi luminescent reaction between NO and 0,. It has been well
verified that the chemi luminescence is due to the following reaction
scheme (Fontijn et al., 1969, 1970):
NO + 03 •*• N02*
hv
To prevent collisional deactivation of the excited N02* molecule and
thus to insure that a chemi luminescent type reaction does occur, total
reaction flask pressure was maintained at a level of 3 torr. It was
then necessary, of course, to establish a vacuum tight flow system
through the NO detector. A Welsh duo seal vacuum pump having a capac-
ity of 500 Jl/min at STP was used to draw sample gas through the instru-
ment. Sample and ozone flows were monitored through observation of
reaction flask pressure and reactant partial pressures.
-------
38
Normal operation of the NO detector commenced with the establish-
ment of a "relatively absolute" vacuum throughout the nitric oxide de-
tector flow system and reaction flask. An ozone flow sufficient to
maintain a steady 2 torr reaction flask pressure was then established.
This ozone flow magnitude assured that 0., would be present in sufficient
excess to insure that the overall reaction be first order in NO. The
ozone flow rate was then held constant through the duration of the run.
A micro-metering valve on the sample line was subsequently adjusted to
allow sufficient sample flow to establish a 3 torr total pressure in the
reaction flask. The electrometer net signal was noted and the NO con-
centration of the sample gas could then be deduced from the previously
discussed calibration curve.
One difficulty was encountered initially with the NO detector total
response time, here defined as that time necessary for the complete re-
movable of an unwanted, "old" sample gas from the flow system and the
establishment of the correct reaction flask pressure and electrometer
output signal for the new sample. This total response time was on the
order of 5 minutes and considerably reduced the amount of NO data which
could be compiled during any one run. Testing of the total system
showed that the MnO^ filter shown in Fig. 2-17 was responsible, as the
filter was packed too tightly. Modification of the filter design has
reduced the response time to about one minute.
In large part responsible for the great accuracy inherent in the
chemiluminescent technique of NO concentration measurement is the rel-
atively complete lack of interference of other sample gas constituents
upon the accuracy of the NO measurement. Table 2-2 presents data
-------
39
Table 2-2. NO Interference Data
(after,Fontijn et al., 1969,1970)
Constituent
Max Concentration
Generally Encountered
in Air Quality
Monitoring, PPM
Concentration
Used at whi ch NO
Interference Was
Detected at
NO 1 10 PPB, PPM
N02
C00
2
CO
C2H4
NH,
3
SO,
2
H 0
3
500
100
1
3
3
100% Saturation
9
650
300
5
9
25
75% Saturation
-------
40
taken by Fontijn et al. (1969, 1970) which illustrate this most desirable
quality.
The nitric oxide detector system also included a N0? converter as
shown in Fig. 2-15. A detailed discussion of the converter and limited
experimental results obtained from it are contained in Appendix A.
2.3 Total Hydrocarbon Analyzer System
Although a Beckman Model 402 Total Hydrocarbon Analyzer utilizing
flame ionization detection was purchased for use in the subject research
effort, instrument malfunctions delayed its use until late in the contract
period. For this reason only a few unburned hydrocarbon measurements
are reported. In Fig. 2-18 is shown the flow schematic for the hydro-
carbon analyzer. Note that a portion (51/2 feet) of the sample line
between the probe block and the relief valve was unheated; the use of a
flexible teflon-coated high pressure line between the block and the
purge system prohibited the use of heating tapes.
C. SUMMARY
Data have been obtained in the above-described facility and will
be described and discussed in the following section. To repeat, the
only aspect of this work which limits its application to practical en-
gines is that unheated air was used at the combustor inlet for most of
the studies. However, as will be shown below, much insight into the
combustion process in a J-33 combustor has been gained.
-------
Relief Valve
Gas Sampling Probe
Purge System
Unheated
Sample
Line
To Remainder —
of Gas
Handling System
Control Room
Electrometer
Test Cell
Heated
Sample
Line
Bellows Pump
FID
Hydrocarbon
Analyzer
Outside
Vent
Span Zero Burner
Gas Gas
Ai r
Figure 2-18. Hydrocarbon sample handling and analyzer system
-------
42
III. RESULTS AND DISCUSSION
In order to Intelligently present the data contained in this section,
it is first necessary to understand generally the basic gas turbine com-
bustion scheme. Typical gas turbine combustors may be divided into three
characteristic regions as shov/n in Fig. 3-1: the primary zone, the sec-
ondary zone, and the dilution zone. The primary zone extends normally
from the fuel nozzle face to the first row of air addition holes and is
the region of most intense combustion. Typical primary zone equivalence
ratios vary to either side of unity. The secondary zone serves primarily
to complete the combustion initiated in the primary zone and for this
purpose receives additional air from penetration air jets. The final
downstream combustor volume is termed the dilution zone. Its purpose is
to reduce the combustor exit plane bulk temperature through air addition
to an acceptable turbine inlet temperature.
For the Allison J-33 combustor used in the present investigation, a
NASA air flow program (Tacina and Grobman, 1969) has been used to estimate
that approximately eight percent of the total combustor air flow enters the
primary zone through the air swirler located around the fuel nozzle in the
combustor dome. The swirled inlet primary air is characterized by a low
pressure region in the center of the swirl necessary to induce the primary
zone flow recirculation (see Fig. 3-1) required for flameholding. That
remaining air not entering through the combustor dome is directed into the
combustor either through film cooling slots or through secondary and dilu-
tion air holes.
-------
primary combustion jsecondary combustion
secondary
combustion
i dilution zone
zone
vaporization
zone wall
recirculation
zone
central
recirculation
zone
jet impingement
recirculation zone
Figure 5- 1 . Typical gas turbine combustor flow schematic
-------
44
The Allison J-33 combustor is shown in Fig. 3-2. Plane "c" in the
figure cuts through the first rov; of secondary air holes and corresponds
approximately to the end of the primary combustion zone. The volume
contained between planes "c" and "e" can be considered the secondary com-
bution zone, while that combustor volume downstream of plane "e", the
dilution zone. The three small holes aligned upstream of plane "c" are
primary zone air addition holes and the intermittently spaced wedges are
the film cooling slots. Finally, the dome cutaway shows the combustor
fuel nozzle and air swirler assembly.
The four planes shown in Fig. 3-2 constitute four of the axial loca-
tions from which internal gas composition samples were taken. Planes "a"
and "f", both not shown in Fig. 3-2, are located respectively 3 inches
downstream of the fuel nozzle face (i.e. 2 1/2 inches upstream of plane
"b") and 1 3/8 inches downstream of the actual combustor exit plane.
Henceforth plane "f" will be designated the combustor exit plane. Total
combustor length (from fuel nozzle face to actual combustor exit) is
approximately 18 5/8 inches.
As previously mentioned, the gas sampling probe tip was offset to
provide a circular trace upon probe rotation. Fig. 3-3 shows the location
of the probe trace in the combustor exit plane sampling station as viewed
from the combustor dose. The relative positions of the combustor cooling
slots, penetration air holes, and gas sampling probe positions should be
noted. (It should be mentioned again that the designated combustor exit
plane is in reality a plane 1 3/8 inches downstream of the actual combustor
exit.) The 0°, 90° cw (clockwise), 180°, and 90° ccw (counter-clockwise)
-------
Figure 3- 2 . J-33 combustor configuration, gas sampling
planes (4 of 6), and probe trace position
Ul
-------
46
Temperature Probe No. 1,
Location No. 1
Location No. 2 *
Location No. 1
Temper-
ature
Probe
No. 2 •X-"
Location No. 2
View from Combustor Dome
Downstream
5.83'
Jet 5
Jet 4
2 9/16" Dia.
Film Cooling Slot
Secondary Air Hole
Figure 3-3. Relationship of probe trace
and gas sampling points to
J-33 combustor configuration
-------
47
designations will be used consistently to refer to the different radial
probe positions as shown in the figure.
A. COMBUSTOR OPERATING POINTS
Having now established the particular J-33 combustor configuration,
it is appropriate to detail the selected J-33 combustor operating points.
For the subject research combustor air flows of 3.75, 4.57, and 6.0 Ibs/sec,
combustor absolute pressures of 3, 5, and 10 atmospheres, and combustor
overall equivalence ratios of 0.217, 0.283, and 0.345 were used, with liquid
propane as fuel. Unheated inlet air was used for the experiments described
in this section. Those combinations of the above parameters that were
investigated are shown graphically in Fig. 3-4 and in tabular form in
Table 3-1. The base combustor operating point corresponded to an airflow
of 6.0 Ibs/sec, an overall equivalence ratio of 0.217, and a combustor
pressure of 5 atmospheres. It is of importance to emphasize that during
any run only one parameter was varied from the base condition. As can be
inferred from Fig. 3-4, and unlike the technique of Sawyer et al. (1969),
equivalence ratio variation was accomplished through fuel flow rate alter-
ation at a constant air flow rate, thus eliminating major concomitant
primary zone and overall residence time changes.
Because of time limitations and the unrealistic use of unheated
combustor inlet air for these experiments, combustor traverses were made
only at the base operating point, B in Table 3-1. Measurements at the
other six operating points were limited to the exit plane. Due to the
unavailablity of functioning hydrocarbon analyzing instrumentation during
-------
48
m (Ibs/sec)
a
Figure 3-4. Combustor operating point matrix
-------
49
Table 3-1. Selected Combustor
Operating Points1^
Fig. 3-4
Point
Designation
A
8**®
C
D
E
F
G
Overall
Equivalence
Ratio
0.217
0.217
0.217
0.283
0.217
0.345
0.217
Air flow
Rate,
Ibs/sec
6.0
6.0
6.0
6.0
4.57
6.0
3.75
Combustor
Pressure,
atm abs
3
5
10
5
5
5
5
Normal Design Operation:
0.232f 7.17 4.2
± For each condition only a single parameter variation
from the base point was allowed.
A The acquisition of a complete data set for each
point required more than one run.
* Point B is designated the combustor base operating
point.
® Combustor traverses were made at this operating
condition.
t Calculated for C,Hg as fuel, following Zucrow and
Warner (1956).
-------
50
this portion of the study, only gas temperatures and carbon monoxide
and nitric oxide concentrations will be reported.
B. INTERNAL MEASUREMENTS
The long time required for accurate gas sampling measurements and
the finite capacity of the air system prevented the completion of a
combustor traverse during any one experiment; usually three to five
runs were needed. Generally it was found that reproducibility within
any given experiment was good, but somewhat poorer from run to run
at a given operating point. Thus all of the data points obtained inside
the combustor will be reported below. The difficulties with gas tempera-
ture measurement detailed in Section II prevented our obtaining as many
local temperatures as concentrations.
For these internal measurements the probe was inserted into the
combustor through the probe addition section window and stationed at one
of the planes shown in Fig. 3-2. After steady combustion at the base
operating point had been established, measurements were taken at the
initial probe position. The probe could then be rotated to any of the
three other radial positions shown in Fig. 3-2 and 3-3, or translated to
another axial station.
1. Gas Temperatures
The internal temperature measurements shown in Fig. 3-5 provide much
insight into the nature of the flow pattern and combustion process in the
J-33 combustor. For convenience, the axial profiles at each radial position
are connected by the heavy lines.
-------
1700
1500
1300
-M
(O
O)
1100
A 0°
O 90° cw
D 180°
x 90°
ccw
900
700
I
I
I
J_
6 8 10 12
Distance from Injector, Inches
14
16
18
20
Figure 3-5. Temperature versus axial position
(n = 6.0 Ibs/sec, g = .217, P = 5 atm)
-------
52
In the first six inches of the combustor, the maximum gas temperatures
are encountered near the wall of the combustor (180°) while the recir-
culation zone nearer to the centerline is relatively cold; this last
observation suggests that a large portion of the air injected through the
first row of large penetration holes at 8 inches recirculates back toward
the injector. The presence of these holes is most clearly seen in the
180° profile at about 8 inches, in the 90° profiles at about 6 inches, and
possibly in the centerline profile between 4 and 6 inches. Because of
the asymmetry of the probe position at both 90° cw and ccw with respect
to the penetration holes (as shown in Fig. 3-3) these two temperatures are
never equal; the 90° ccw values are generally the lower, due to jet 5.
The flame closes to the centerline between 8 and 10 inches, where the
maximum temperatures are observed. Downstream of this station the tempera-
tures decay for the most part as the remainder of the air is added
through the holes and slots. It thus appears that reaction continues
well into the secondary zone of this combustor.
One surprising result of the temperature measurements is their low
values in the first 10 inches of the combustor: the maximum value of
about 1600° K (on the centerline at 10 inches) is at least 400° K lower
than one would expect theoretically. However, since the thermocouple
bead did not melt during the experiments, temperatures on this order could
not have been experienced in any position through which the probe travelled.
There are several sources of error in temperature measurements with
unshielded, uncoated junctions: one results from conduction along the
thermocouple, but is negligible here due to the depth of immersion of the
probe into the combustor. Catalyzed recombination reactions on the
-------
53
uncoated Pt leg would increase the measured temperature over the gas
temperature, which is the wrong trend. Radiation heat transfer losses
from the bead to the relatively cold combustor walls must also be
considered. However, assuming the bead to have an emissivity of one and
to be surrounded by walls at 0° K results in a temperature measurement
low by only 100° K for an indicated value of 1600° K (the Nusselt number
for heat transfer to the bead is estimated as 43 in the primary zone).
Thus radiative corrections have not been applied to the data of Fig. 3-5
and do not appear responsible for the unexpectedly low temperatures
which were measured.
Most likely direct impingement and subsequent evaporation of liquid
fuel on the thermocouple bead is the source of the error, particularly
near the walls where the hollow fuel spray cone would be expected to
persist (using rLO injected into ambient air, the spray cone angle was
measured to be about 80°). However, since even with this difficulty the
highest temperatures at 4 inches were observed at 90° cw and 180°, the
following qualitative picture of the combustion process can be assembled:
the flame zone follows the fuel spray and is hollow in the primary, with
a relatively cold embedded recirculation zone fed in part by the first
large air holes at 8 inches. The zone ends at about the 8 inch position
where the flame returns to the centerline.
Thus the schematic of Fig. 3-1 is a reasonable representation of the
J-33 combustion process if the central and jet impingement recirculation
zones are merged and substantial combustion is allowed in the secondary
zone. It is also concluded that the particular 90° radial probe positions
chosen for these experiments will not exhibit equal temperatures or
-------
54
concentrations since the 90° ccw points are more directly aligned with
penetration jets; as will be shown below, slightly lower NO and CO
measurements were also obtained at 90° ccw.
2. Nitric Oxide
Axial NO profiles for the four radial positions are presented in
Fig. 3-6 through 3-9; due to the large number of data points and
considerable scatter four separate figures are shown. The general order
of decreasing NO concentrations is 0°, 90° cw, 90° ccw, and 180°,
respectively.
High NO concentrations are found on the center!ine in the region
thought to be within the recirculation zone (the first 6 or 8 inches
in Fig. 3-6), but the scatter in the data is also the worst here. At
10 inches at 0°, where the maximum temperature was measured, NO also
reaches its maximum value, falling off downstream probably as a result
of dilution. The 90° ccw values (Fig. 3-7) are generally lower than
those at 90° cw (Fig. 3-8) due to the closer proximity to penetration
jets, and as in the temperature profiles, dilution from the first large
jet (at 8 inches) can be seen at 8 and at 6 inches, respectively. In
addition, the wall concentrations (Fig. 3-9) clearly show the influence
of this jet at 8 inches. Only slight radial variations are seen at 3
inches from comparison of the four figures; more substantial variations
are exhibited downstream of this point. The scatter in all of the data
decreases in moving downstream.
All of the temperatures reported in Fig. 3-5 are too low for signifi-
cant NO formation via homogeneous reactions. However, the general increase
-------
25
20
A= 0°
A= Average Exit
Plane Value
Q-
O.
C
o
ea
i.
c
QJ
O
c
o
o
10
AA
8
10
12
14
16
18
&
_L_
20
22
Distance from Injector, Inches
Figure 3-6. NO concentration versus axial position for Oc
(m = 6.0 Ibs/sec, £ = .217, P = 5 atm)
a
en
en
-------
25
20
X = 90° ccw
N = Average Exit
Plane Value
o_
D_
c
o
15
£
C
Ol
O
O
10
XX
1
1
I
1
1
1
6 8 10 12 14 16
Distance from Injector, Inches
18
20
22
Figure 3-7. NO concentration versus axial position for 90° ccw
(m = 6.0 Ibs/sec, 0 = .217, P = 5 atm)
a
en
cn
-------
25
20
o = 90* cw
• = Average Exit
Plane Value
Q_
Q-
o
+J
c
o
o
10
o
O
-L
_L
o
o
JL
_L
-L
6 8 10 12 14 16
Distance from Injector, Inches
J_
_L
18 20
22
Figure 3-8. NO concentration versus axial position for 90° cw
(m = 6.0 Ibs/sec, = .217, P = 5 atm)
9
tn
-------
c. *s
20
51
D-
g 15
4J
fO
t-
c
QJ
1 1°
O
0
5
f
D = 180°
• = Average Exit
Plane Value
D
D
a
D
D D D
n ° ° EL
O n r^
D D ™
D
1 1 1 1 1 1 1 1 1 1
2 4 6 8 10 12 14 16 18 20 22
Distance from Injector, Inches
Figure 3-9. NO concentration versus axial position for 180'
(n = 6.0 Ibs/sec, cj> = .217, P = 5 atm)
a
en
00
-------
59
in NO concentration which follows the zone of maximum temperature (i.e.,
3 to 6 inches at 180°, Fig. 3-9; about 7 inches at 90° cw, Fig. 3-8;
and 8 to 10 inches at 0°, Fig. 3-6} suggests firstly that NO is still
forming in this zone and secondly that the measured temperatures are
low, as was concluded in the previous section. The high values of NO
observed in the recirculation zone at the 4 through 7 inch positions are
most likely a result of entrainment into this zone: although these values
are suspect because of the large scatter in the data, it will be shown
subsequently that NO is not formed in this region, when the results of
varying the air flow rate are reported.
3. Carbon Monoxide
Semi logarithmic plots of axial CO concentrations are shown as Fig.
3-10 through 3-13, again grouped by radial probe position. The high
concentrations observed at those stations closest to the injector
suggest that the region in which CO is formed was not accessible to the
probe, a finding which is consistent with the proposed fuel rich zone
very near the injector face. In addition, the lower values at 180°
suggest that CO oxidation is still occurring in the primary zone along the
walls; the reaction has been quenched in the recirculation zone by the
cold air flowing upstream from the large penetration jets at 8 inches.
Because of the logarithmic concentration scale, the pattern of this jet
is not as easily followed as in the temperature and NO concentration
profiles.
-------
60
100,000
10,000
D-
O-
O
c
CJ
u
c
o
CJJ
o
1000
100
A
A
A
A
A
A
A= 0°
A * Average Exit
Plane Value
J I
J_
I
J I
I
J I
4 8 12 16
Distance from Injector, Inches
20
Figure 3-10. CO concentration versus axial position for 0'
(m = 6.0 Ibs/sec, = .217, P = 5 atm)
a
-------
61
10,000
-r '
Q.
0.
*
c
o
I
J_>
^•*
OJ
o
c
o
CJ
o
1000
100
(
x v
: $
X
X
X
- X
•»
^
X
X
H>
X
••
- XM
x = 90° ccw
H = Average Exit
Plane Value
1 1 1 1 ! 1 1 1 1 1
) 4 8 12 16 20
Distance from Injector, Inches
Figure 3-11. CO concentration versus axial position for 90° ccv/
(m, = 6.0 Ibs/sec, = .217, P = 5 atm)
a
-------
62
100,000
10,000
Q.
Q.
-M
re
i.
4->
c
HI
u
o
1000
100
o
o
o
0
o
o
o
0
o
o
-
o = 90° a/
• = Average Exit
Plane Value
1 I I I I I t
0
o
1 1 1
4 8 12 16
Distance from Injector, Inches
20
Figure 3-12. CO concentration versus axial position for 90° cw
(m, = 6.0 Ihs/sec, 4> = .217, P = 5 atm)
a
-------
63
100,000
10,000
£
c
Ol
o
o
1000
100
= 180°
= Average Exit
Plane -Value
I I I I I I I I I
4 8 12 16
Distance from Injector, Inches
20
Figure 3-13. CO concentration versus axial position for 180C
(m = 6.0 Ibs/sec, <> = .217, P = 5 atm)
a
-------
64
4. Summary
The flow pattern in the first 10 inches of the J-33 combustor
postulated on the basis of the temperature and concentration measure-
ments is shown schematically in Fig. 3-14. The most important features
are the fuel rich zone along the dome and liner walls where CO (region
1) and NO (region 1-3-5) are formed and the strong backflow of air
(region 4) which quenches CO oxidation in region 2. It is the lack of
substantial air addition into the fuel rich region and too much air
flow into the recirculation zone which is responsible for the high
emissions of CO from this combustor. However, since the highest tempera-
ture zones are probably fuel rich, the NO emissions are low. For this
particular liner, the results suggest that the placement of the first
row of large penetration holes will largely determine the emissions.
C. COMBUSTOR EXHAUST PLANE MEASUREMENTS
Through internal measurements of temperature and composition the
influence of liner design on CO and NO emissions has been discussed in
the previous section; combustor exit plane measurements are not as
useful for this purpose because they obscure the important chemical and
physical phenomena occurring within the liner. However, studies of this
latter type can help reveal the dependence of emissions on cycle design
(as opposed to liner design) parameters. Most of the data to be reported
in this section pertain to NO and CO emissions as functions of overall
equivalence ratio, combustor pressure, and air flow rate, but the
limited temperature measurements obtained with the exit plane chrome!-alumel
thermocouples will be presented first.
-------
1. Very fuel rich; much CO
formation; fuel drops
continue downstream
2. CO quenched by cold
recirculating air; NO
entrained from wall flow
3. Measured temperatures low
due to drop impingement;
some CO oxidation and NO
formation
Penetration air
feeds recirculation
zone
NO formation
throughout this
high temperature
region 3-5
8
Distance from injector, inches
Figure 3-14. Schematic of postulated primary and secondary
zone flow pattern for the J-33 combustor
01
in
-------
66
1. Gas Temperatures
Only the influence of air flow rate on radial temperature profiles
was determined, for the two sets of probe positions shown in Fig. 3-3.
The data for location number one are shown in Fig. 3-15, and those for
location number two in Fig. 3-16. Note that in the former position probe
one is aligned with a penetration jet while probe two is midway betv/een
a jet and a film cooling slot; in location number two probe one was
directly aligned with a slot.
As expected, the mean gas temperature is higher when measured above
a film cooling slot than above a penetration jet: probe one exhibits
higher values at both air flow rates in Fig. 3-16 than in Fig. 3-15.
Probe two at location two reveals somewhat lower temperatures since in
this position it is closer to a penetration jet (see Fig. 3-3). The
minima in all of the curves represent the cores of the last row of pene-
tration jets near the end of the liner.
These radial profiles clearly indicate the difficulty in calculating
a meaningful mass averaged temperature at the combustor exit; in fact,
Fig. 3-15 and 3-16 suggest that the exhaust temperature decreases with
decreasing air flow rate, which is inconsistent with thermodynamics.
2. Exhaust Plane Emissions of Nitric Oxide and Carbon Monoxide
Exhaust plane concentration data as functions of the cycle design
parameters air flow rate, overall equivalence ratio, and combustor
pressure will be presented in two ways: firstly, numerical averages at
each radial position will be shown, and secondly, in order to demonstrate
the reproducibility of the measurements, all data will be shown. Recall
-------
67
1123
1073
57 Ibs/sec
75 Ibs/sec
75 Ibs/sec
Temp. Probe One
Temp. Probe Two
• Temp. Probe One
X Temp. Probe Two
623
.5 1.0 1.5 2.0 2.5
Radial Distance from Combustor Wall (Inches)
Figure 3-15 • Combustor exit plane temperature traverse
(probes in locations numbered one in Fig. 3-3)
-------
68
1073
1023
973
923
cu
E
o
873
823
773
723
673
623
573
m =3.75 Ibs/sec
a
m =3.75 Ibs/sec
3-
Temp. Probe One
Temp. Probe Two
• Temp. Probe One
.5 1.0 1.5 2.0
Radial Distance from Combustor Wall (Inches)
Figure 3-16. Combustor exit plane temperature traverse
(probes in locations numbered two in Fig. 3-3)
2.5
-------
69
that only one cycle parameter was varied at a time, as shown in Fig. 3-4,
in order to obtain these data.
2.1 The Influence of Air Flow Rate
Since the air flow rate was varied at constant overall equivalence
ratio (and pressure), it was necessary to vary the fuel flow rate
accordingly. This was accomplished by changing the fuel differential
injection pressure. Thus in interpreting the effects of air flow rate
on NO and CO emissions it may be necessary to consider changes in
atomization and spray penetration as well as in primary zone and over-
all residence times.
As shown in Fig. 3-17 and 3-18, only a slight decrease in NO concen-
tration occurs with a 60% decrease in air flow rate from the base
condition; this result indicates that either the residence time in those
zones responsible for NO formation does not change greatly, or liquid
droplet combustion is the predominant source of NO and remains relatively
unchanged throughout this set of experiments.
CO emissions demonstrate a more interesting behavior and are displayed
in Fig. 3-19 and 3-20. A general decrease is observed in passing to the
lowest air flow rate tested, particularly on the centerline and at 90° cw.
At first these results seem inconsistent with the NO results, but both
can be explained in terms of Fig. 3-14. It has been hypothesized that
zone 1 is the predominant region of CO formation and zones 1 and 3 of NO
formation. Since NO formation does not change appreciably with air flow
rate, the flow pattern must remain essentially the same in zones 1 and 3.
The decreased CO emissions at 3.75 Ib/sec can then be attributed to poorer
-------
70
32
28
24
s: 20
Q.
O
a
s-
4J
C
O
O
c:
o
CJ
16
12
= .217
P = 5 atm
A 0°
o 90° cw
a 180°
x 90° ccw
m , Ibs/sec
a
Figure 3-17. Radial NO concentrations at combustor
exit plane versus air flow rate (average)
-------
71
Q.
Q_
O
c
Ol
u
o
32
28
24 L.
20
16
12
cj> = .217
P = 5 atm
Average Values
Connected
o
a
0°
90° cw
180°
90° ccw
m , Ibs/sec
a
Figure 3-18. Radial NO concentrations at combustor
exit plane versus air flow rate
-------
72
10,000
OH
a.
o
•H
(3
-------
73
10,000
a.
a.
c
o
c
o
CJ
o
u
P , = 5 Atm
comb
4o = .217
1,000
100
A o°
O 90°cw
a 180°
90°ccw
Average Values
Connected
m , Ibs/sec
3.
Figure 3-20,
Radial CO concentrations at corabustor exit
plane versus air flow rate
-------
quenching of the CO oxidation in zone 2 and results from less penetration
air from zone 4 getting to zone 2 at the low air flow rate. At higher
flow rates, approaching the design point of 7.17 Ib/sec, the particular
design of the J-33 can prevent substantial variations in primary zone
residence time. These conclusions are also consistent with the postulate
that the high NO concentrations in zone 2 (see Fig. 3-6) result from
entrainment from zones 1 and 3 rather than NO formation in zone 2, since
higher NO emissions v/ere not observed at the lowest flow rate.
If this behavior were characteristic of combustors other than the
J-33 liner, then it could be used to reduce CO (and probably HC) emissions
at aircraft low speed ground idle and at automotive idle. While main-
taining an overall equivalence ratio close to the design value, compressor
bleed air flow could be increased to prevent quenching of the CO oxidation
reaction in the combustor. The penalty for this method of emissions
control would be an increased fuel consumption at idle.
2.2 The Influence of Overall Equivalence Ratio
The primary effect of an increase in overall equivalence ratio would
be a change in the temperature in zones 1 and 3 of Fig. 3-14: if these
zones were predominantly lean, then an increase in temperature and NO, and
little change in CO would result; if rich, then temperature and HO would
decrease while CO increased. On the other hand, if droplet combustion
were important then NO (and possibly CO) should increase simply as a
result of the presence of more droplets. Of minor significance would be
a slight decrease in primary zone residence time due to the increased
fuel flow rate.
-------
75
The results of a change in equivalence ratio are shown in Fig. 3-21
through 3-24, and it is observed that both NO and CO increase strongly
with overall equivalence ratio. Most likely this results from the droplet
effect noted above; the CO increases as well because zones 1 and 3 are
predominantly rich. Both the presence of droplets and the lack of
sufficient air for complete combustion in these regions are entirely
consistent with the other observations which led to the model depicted
in Fig. 3-14.
2.3 The Influence of Combustor Pressure
The final cycle parameter which was varied was combustor pressure,
at constant air flow rate and overall equivalence ratio; the results
of this study are shown in Fig. 3-25 through 3-28. An increase in
pressure will increase the homogeneous rate of NO formation and both
CO formation and oxidation. In addition, small decreases in fuel spray
cone angle and atomization may result.
In Fig. 3-25 and 3-26 it is seen that a substantial increase in NO
occurs with increasing pressure at all radial probe positions: the
mass-averaged NO concentration increases from 3 ppm at 5 atm to 11 ppm
at 10 atm, a factor of 3.67. If one assumes that the Zeldovich mechanism,
with 0/02 equilibrium and N atoms in steady state, describes the gas-phase
NO formation kinetics, then it can be shown that the rate of formation
of NO increases with pressure to the 1.5 power; in other words, on the
basis of homogeneous kinetics and under the assumption that the tempera-
ture and residence time in the NO forming region are relatively
unaffected by total combustor pressure, the NO concentration at 10 atm
-------
76
32
28
24
£ 20
D-
C
o
4J
<0
4J
c
o
u
o
o
16
12
8 -
4 -
0
.20
m, = 6.0 Ibs/sec
a
P = 5 atm
A 0°
o 90° cw
D 180°
x 90° ccw
.35
.25 .30
Overall Equivalence Ratio
Figure 3-21. Radial NO concentrations at combustor
exit plane versus overall equivalence ratio (average)
-------
77
Q.
Q.
ra
i.
4J
o
o
32
28
24
20
16
12
0
.20
rn = 6.0 Ibs/sec
a
P = 5 atm
Average Values
Connected
o
D
0°
90° cw
180°
90° con
.35
. .25 .30
Overall Equivalence Ratio
Figure 3-22. Radial NO concentrations at combustor
exit plane versus overall equivalence ratio
-------
78
c
o
15,000
14,000
13,000
12,000
11,000
10,000
9,000
£ 8,000
o 7,000
o
o 6,000
u
5,000
4,000
3,000
2,000 —
1,000 —
0
m = 6.0 Ibs/sec
a
— P = 5 atm
A
O
n
x
0°
90°cw
180°
90°ccw
.217
.283
.345
Overall Equivalence Ratio
Figure 3-23. Radial CO concentrations at the
combustor exit plane versus overall
equivalence ratio (average)
-------
79
15,000
14,000
13,000
12,000
11,000
| 10,000
o 9,000
£ 8,000
c
g 7,000
u
o
u
6,000
5,000
4,000
3,000
2,000 —
1,000 —
(H
A 0°
O 90°cw
n 180°
X 90°ccw
m = 6.0 Ibs/sec
a
P = 5 atm
A
.217 .283 .345
Overall Equivalence Ratio
Figure 3-24. Radial CO concentrations at the
combustor exit plane versus
overall equivalence ratio
-------
80
Q.
D_
CJ
O
O
O
32
28
24
20
O
£ 16
12
8
4 -
= .217
rci = 6.0 Ibs/sec
a
A 0°
o 90° cw
a 180°
x 90° ccv/
4 6 8 10
Combustor Pressure, Atm
12
Figure 3-25. Radial NO concentrations at combustor
exit plane versus pressure (average)
-------
81
O
ra
O)
o
o
o
32
28
24
20
16
12 .
8 -
4 u
0
= ,217
ITL = 6.0 Ibs/sec
a
Average Values
Connected
A 0°
O 90° cw
a 180°
x 90° ccw
10
12
2 46 8
Combustor Pressure, Atm
Figure 3-26. Radial NO concentrations at combustor
exit plane versus pressure
-------
10,000-
4>o = .217
m = 6.0 Ibs/sec
3.
82
cu
ex,
o
•H
O
O
c
o
u
o
u
100L
90°ccw
Average Values
Connected
2 4 6 8 10
Combustor Pressure, Atm
Figure 3-27. Radial CO concentrations at combustor
exit plane versus pressure (average)
-------
83
10,000
c
o
•H
rt
o
o
1,000
'
0o = -217
„, m =6.0 Ibs/sec
100 '-•-•
•;•.•}--•
\
A
O
A o°
O 90°cw
n 180°
X 90°ccw
Average Values
Connected
] 3 3. f
4 6 8 10
Combustor Pressure, Atra
Figure 3-28. Radial CO concentrations at combustor
exit plane versus pressure
-------
84
should be only about 2.8 times that at 5 atm. Thus the observed increase
cannot be explained on the basis of homogeneous kinetics.
In holding the fuel flow rate constant as the pressure was increased,
it was generally found necessary to decrease the pressure drop across
the injector. Consequently, larger fuel droplets capable of forming in-
creased concentrations of NO were probably present at the higher pressures.
The CO exhaust plane data are shown in Fig. 3-27 and 3-28 and do not
show a single trend with increasing pressure: the center!ine concentration
increases as the wall value decreases. These observations are attributed
to a decrease in injector spray cone angle with increased pressure and are
consistent with the combustion model shown in Fig. 3-14.
D. OTHER PARAMETERS
The heat release rate and combustion intensity are not parameters
that can be varied independently of others such as combustor pressure,
overall equivalence ratio, and air flow rate once the combustor geometry
and fuel are defined. Consequently, since the experimental investigation
discussed here employed a J-33 combustor burning liquid propane, the
isolated effects of heat release rate and combustion intensity on
pollutant emissions were not determined.
E. SUMMARY
In summary, the temperature and concentrations of CO and NO at
various axial and radial positions within the J-33 combustor have been
-------
85
presented and discussed. These profiles are similar to those reported
by Sawyer et al. (1969) but definitely show the effects of the particular
J-33 combustor configuration.
Having acquired some fundamental information concerning the
internal combustor mechanisms responsible for the net CO and NO emissions
(at the base operating point), the effects of combustor pressure, overall
equivalence ratio, and air flow rate were isolated insofar as possible
and measured at the combustor exit plane. General support was obtained
from these data for the model of the combustion process in the 0-33
combustor which had been postulated on the basis of the internal measurements
-------
IV. FUTURE EFFORTS
Emissions studies of the J-33 combustor using unheated inlet air were
reported in Section III; however, as noted previously these results
constituted only a preliminary effort due to this unrealistic combustor
inlet temperature. Gas turbines under consideration as alternatives to the
internal combustion engine in automobiles are of the regenerative type, and
the trend 1n aircraft engines is to higher compression ratios. Thus it
is of interest to add air inlet temperature to those cycle design
parameters which can be varied in the experimental facility.
To this end the double combustor facility shown in Fig. 4-1 has been
constructed. The first J-33 combustor is used as an air heater for a
second J-33 mounted downstream. Although the heated air will contain
pollutants, at levels of contamination similar to those reported in the
previous section, since the primary purpose of the facility is to indicate
trends in emissions as functions of cycle and combustor design parameters
(as opposed to absolute or baseline values) the use of vitiated air is
considered acceptable.
For assistance in evaluating the performance of the double combustor
facility in meeting the objectives mentioned above, a few experiments
have been conducted to date. Because these runs were of an exploratory
nature and the data are not complete, the successful operation of the
facility is of more interest than the actual data to be reported here.
-------
A1r System
Fuel System
J-33 Heating
Contustor
Fuel System
J-33 Test
Combus tor
Probe Addition Section
Back Pressure
Valve
Vobe
Posi tioner
Figure 4-1. Complete combustion facility schematic
CD
-------
88
The probe addition section and back pressure valve were mounted
at the exhaust flange of the second J-33 combustor, and the gas sampling
probe was placed at the exit plane of the test combustor. Although the
thermocouple portion of the probe was destroyed during one overly long
and hot Ignition transient, the gas sampling probe has withstood over
one-half hour at this station to date. We are presently replacing the
Pt/PtRh couple.
A gas sampling rake and chrome!/alumel thermocouple rake for
mounting in the diffuser section just upstream of the test combustor
are under construction but were not available for the emissions measured
to date; thus the data to be reported are composites for both combustors.
In later studies the rakes will provide for null measurements to obtain
the true emissions from the test combustor.
In Table 4-1 are presented those data obtained to date for the follow-
ing run conditions for the test combustor: air flow rate 6 Ib/sec,
combustor pressure 3.4 atm abs, overall equivalence ratio 0.217, and
Inlet air temperature 630° K. Without the null measurement to be provided
by the rakes, no attempt will be made to discuss these preliminary
emissions data since the equivalence ratio and pressure for the first or
heating combustor do not correspond to any of the operating points listed
in Section III. To repeat, the purpose of this experiment was simply to
gain experience with the complete facility and associated instrumentation
(including the FID hydrocarbon analyzer).
After the gas sampling and temperature rakes are Installed upstream
of the J-33 test combustor, detailed gas sampling (including CO, NO and HC)
-------
89
Table 4-1. Operating Point and Emissions Obtained
from Two J-33 Combustors in Series*
Air Flow Rate 6.0 Ib/sec
Combustor Pressure 3.4 atm abs
Overall Equivalence Ratios
First or Heating Combustor 0.130
Second or Test Combustor 0.217
Inlet Temperature to Test Combustor1 630° K (680° F)
Pollutant Radial Probe Position at Test Combustor
Exhaust Plane
0° 90° cw 90° ccw 180°
NO, ppm
CO, ppm
HC, ppm CH4
1.89
2250
950
2.61
1800
400
1.61
4500
2600
1.43
2020
1000
* Emissions are composites for both combustors.
± Calculated following Zucrow and Warner (1956).
-------
90
and temperature measurements are to be made at various stations within
the test combustor as functions of four overall cycle design parameters:
inlet air temperature, combustor pressure, equivalence ratio, and air
flow rate. Each parameter will be varied independently from a standard
run condition.
The possibility of N(L formation in the combustor is a point of
current discussion. Significant amounts of N(L have recently been found
1n the exhausts of commercial and military jet engines, particularly at
idle (Anon., 1971, Hare et al., 1971 and Vaught et a!., 1971). These data
were obtained primarily by use of the NCL-^O converter in conjunction with
a chemiluminescent NO analyzer. In view of the possibility of catalytic
decomposition of NO by these converters (see Appendix A), the observation
of N02 by Hare et al. (1971) is questionable since only the converter and
a chemHuminescent detector were used in the study to obtain NO concen-
A
tratlons.
More confidence can be placed in the results of Airesearch (Anon.,
1971) because they also used a NDUV analyzer at selected points to measure
N02 directly. They noted that "a check of several test power settings
using a NDUV analyzer ... confirmed the magnitude of NO,, readings received
on the chemiluminescent analyzer," but apparently no quantitative
comparisons are presented in their report (Anon., 1971).
Detroit Diesel Allison (Vaught et al., 1971) compared NO as obtained
on a NDIR analyzer with that from a chemiluminescent analyzer and generally
found agreement within 10%. However, total NO from a converter-chemilu-
A
minescent analyzer was considerably less than that obtained with the
-------
91
Saltzman method (27% less at low speed ground idle and 6% less at take-off
for the T-56 engine). Note that if the converter is destroying NO, as
well as converting N02 to NO, a lower NOX would be obtained with the
chemiluminescent analyzer.
The present investigation after completion of the J-33 study will
turn to measurements at the exhaust plane and within an Allison T-56
combustor, particularly with respect to independent verification of the
reported N02 levels (Vaught et al., 1971), as well as the Identification
of those zones within the combustor which are responsible for the
production of N02. Instrumentation will consist of the converter-chemilu-
minescent analyzer for NO and NO , and it is hoped that these data can be
substantiated via the NDUV method for N02. CO, HC, and temperature
measurements will be made simultaneously with the N0x surveys.
Although it will probably not be possible to simulate T-56 low speed
ground Idle 1n the facility due to flameholding requirements in the
heating combustor, since substantial N02 was observed at all of the
engine operating points (Vaught et al., 1971), it is felt that a reasonable
simulation of the T-56-A-15 operation and investigation of N02 formation
in the combustor can be accomplished.
Another study which 1s planned 1s to use a gaseous rather than liquid
fuel; an example of the success of this technique for emissions reductions
is the widespread use of natural gas in stationary combustion systems as
a short range method of reducing pollution. For the aircraft or automotive
turbine engine, however, natural gas is not a practical fuel, and thus
prevaporlzing burners represent one compromise design (see for example
-------
92
Zwlck et a!., 1971, for a Rankine cycle application).
In spite of this concensus, there has been no demonstration of the
importante of heterogeneous effects with regard to emissions presented
in the open literature for a current combustor can (Hellor, 1971). The
Berkeley group (Sawyer et a!., 1969, Starkman et al., 1970, and Parikh
et al., 1971) has shown that gaseous methane produces less HC, CO, and
NO than liquid heptane, when burned in a model laboratory combustor.
However, Inlet conditions to the combustor were not realistic (room
temperature air and pressure only slightly in excess of atmospheric),
and not only the physical state, but also the chemical nature of the fuel
was changed as well.
It is apparent that a controlled investigation of the effect of
using gaseous rather than liquid fuel is needed. It is proposed here
that the experiments carried out with the T-56 combustor be repeated
using gaseous propane fuel. Internal and exhaust plane surveys of tempera-
ture, HC, CO, NO, N02, and NOX will again be made.
In sunuiary, the experimental setup using a single J-33 combustor or
two combustors in series has proven to be operable. Preliminary data have
been obtained, but interesting and important investigations are left to
be carried out. Future efforts are to include the study of a high inlet
temperature combustor, the investigation of NOg formation in combustors
with emphasis on the T-56 can, and a determination of the effect on
pollutant emissions when gaseous fuel rather than liquid fuel is used.
-------
93
LIST OF REFERENCES
Anon., 1971, "Exhaust emissions test-Airesearch aircraft propulsion
and auxiliary power gas turbine engines," Report GT-8747-R, Airesearch
Manufacturing Co. of Ariz.
Chase9 J. 0. and Hurn, R. W., 1970, "Measuring gaseous emissions from
an aircraft turbine engine," SAE Paper 700249.
Fontijn, A., Sabadell, A. J., and Ronco, R. J., 1969, "Feasibility
study for the development of a multifunctional emission detector for
air pollutants based on homogeneous chemiluminescent gas phase reac-
tions," TR-217, AeroChem Research Laboratories.
Fontijn, A. Sabadell, A. J., and Ronco, R. J., 1970, "Homogeneous chem-
iluminescent measurement of nitric oxide with ozone," Anal. Chem. 42,
575.
Fontijn9 A., 19719 AeroChem Research Laboratories, Personal communica-
tion to A. M. Mel lor.
Hare, C. T.s Dietzmann, H. E., and Springer, K. J., 1971, "Gaseous emis-
sions from a limited sample of military and commercial aircraft turbine
engines," Report AR-816, Southwest Research Inst.
Hodgesone J. A. „ Bell, J. P., Rehme, K. A., Krost, K. J., and Stevens,
R. K., 1971t "Application of a chemiluminescence detector for the meas-
urement of total oxides of nitrogen and ammonia in the atmosphere,"
AIAA Paper 71-1057.
Mellor, A0 M.9 19719 "Current kinetic modeling techniques for continuous
flow combustors," Emissions from Continuous Combustion Systems Symposium,
General Motors Research Laboratories.
Nelson, A. W., 1972, "Collection and assessment of aircraft emissions
baseline data - turbine engines," Report PWA-4339, Pratt and Whitney
Aircrafts United Aircraft Corporation.
Owens, C. W. and Mellor, A. H., 1971, "An investigation of gas turbine
combustors with hiqh inlet air temperatures. Second Annual Report:
Part II: Heat Transfer," TM-71-2, Jet Propulsion Center, Purdue Univer-
sity, USA TACOM Report No. 11328.
Owens, C. W., 1972, "Heat transfer in high temperature gas turbine
combustion chambers," M.S.M.E. Thesis, Purdue University.
Parikh, P. G., Sawyer, R. F., and London, A. L., 1971, "Pollutants from
methane fueled gas turbine combustion," College of Engineering Report
No. TS-70-15, University of California, Berkeley.
-------
94
Sawyer, R. F., Teixeira, D. P., and Starkman, E. S., 1969, "Air pol-
lution characteristics of gas turbine engines," ASME Trans., J. Eng.
Power, 91_, 290.
Starkman, E. S., Mizutani, T., Sawyer, R. F., and Teixeira, D. P.,
1970, "The role of chemistry in gas turbine emissions," ASME Paper
70-GT-81.
Tacina, R. R. and Grobman, J., 1969, "An analysis of total pressure
loss and airflow distribution for annular gas turbine comb us tors,"
NASA TN D-5385.
Vaught, J. M., Parks, W. M., Johnsen, S. E. J., and Johnson, R. L.,
1971, "Final technical report collection and assessment of aircraft
emissions base-line data-turboprop engines (Allison T56-A-15)," Report
EDR 7200, Detroit Diesel Allison Division, General Motors.
Zucrow, M. J. and Warner, C. F., 1956, "Constant pressure combustion
charts for gas turbines and turbojet engines," Purdue University
Engineering Experiment Station Bulletin No. 127.
Zwick, E. B., Mills, T. R., and Fio Rito, R., 1971, "Evaluation of a
low NO burner," Report USG-1, Paxve, Inc.
-------
95
APPENDIX A
EFFECT OF NOg CONVERTER ON NO
A NOp converter was added to the nitric oxide detector system so
that the existing chenriluminescent NO detector could be used to meas-
ure the NOp concentration of the sample gas. In principle, the con-
verter operates by heating the sample gas to a sufficient temperature
(approximately 600° C) such that the N02 will dissociate to NO. The
NOp converter-NO system is shown schematically in Figure 2-15. Paral-
lel flow paths to the NO detector are provided for the sample: one
goes directly to the detector and the second passes through the con-
verter. By intermittently changing the flow path taken by the sample
gas, the concentrations of NO and NO can alternately be recorded.
/\
The total NOp concentration is then the difference between the alter-
nating concentration levels.
The converter was constructed according to specifications fur-
nished by the Environmental Protection Agency, which called for six
feet of one-eighth inch by .028 inch wall 316 type stainless steel
tubing. The tubing was resistance heated by flowing an electric cur-
rent directly through it. A thermocouple was attached to the tube
so that the tube temperature could be continuously monitored.
No NOp concentration measurements were recorded, as preliminary
testing showed that the NO concentration in a 215 ppm sample gas (bal-
ance Np) was greatly reduced when the gas was flowed through the con-
verter. It was observed that an increase in the converter temperature
-------
96
from 23° C to 600° C was accompanied by a decrease in the NO concentra-
tion level recorded by the chemiluminescent detector. Similarly, the
recorded concentration level rose with decreasing converter temperature.
The recorded results for four individual runs are shown in Fig. A-l.
The slopes of the curves of Fig. A-l depend upon the rate of temperature
increase, with run one having the highest rate. The lowest concentra-
tions recorded are within an order of magnitude of the dark current of
the detector.
In experiments using a commerical chemiluminescent detector with
a similar converter, reported quite recently by Nelson (1972), similar
irreproducible effects were observed with the "unconditioned" converter
when either NO in Np or N02 in air was flowed through the system. Con-
verter conditioning was accomplished by repeated exposures to NO with
A
the converter at high temperature.
It is presently thought that the observed results can be attributed
to a surface catalytic reaction along the walls of the stainless steel
tubing (Fontijn, 1971). Such a reaction is consistent with the lack of
reproducibility shown in Fig. A-l, the dependence on applied heating
rate, and the failure of Hodgeson et al. (1971) to find a similar effect
for samples containing less than 1 ppm NO.
-------
1.0
c
o
•iH
•M
rt
(H
U
3
tN
OS
O
4)
O
C
O
u
o
o
u
o.i —
Inverse Temperature (°K~ x 10 )
Figure A-l. Normalized concentration versus inverse temperature
VO
------- |