FINAL REPORT
LOW EMISSION BURNER
FOR RANKINE CYCLE ENGINES
FOR AUTOMOBILES
U.S. Environmental Protection Agency
-------
Low Emission Burner for
Rankine Cycle Engines
for Automobiles
FINAL REPORT
By
T. E. Duffy
J. R. Shekleton
R. T. LeCren
W. A. Compton
Prepared for
Department of Motor Vehicles
Research and Development
Air Pollution Control Office
< Environmental Protection Agency
a
H SOLAR
OmSIOK OF INTERNATIONAL HARVESTER COMPANY
2200 PACIFIC HIGHWAY -SAN DIEGO. CALIFORNIA 92112
KDR 1695
-------
FOREWORD
This Final Report covers the work performed under EPA/APCO during the
work period, July 1, 1970 to March 31, 1971. It is published for information only,
and does not necessarily represent the recommendations, conclusion, or approval of
the Environmental Protection Agency, National Air Pollution Control Administration.
The contract with the Research Laboratories of Solar Division of International
Harvester Company, San Diego, California, was initiated by the EPA/APCO under
contract number EHS 70-106, monitored by Mr. F. Peter Hutchins, Project Officer.
The program was under the general direction of Mr. W. A. Compton, .
Assistant Director-Research, who served as Program Director. Mr. R. T. LeCren,
Group Engineer was the principal investigator at Solar. Mr. Thomas E. Duffy,
Research Staff Engineer was the specialist on controls and systems. Mr. Jack R.
Shekleton, Engineering Specialist was the leader in combustion system development.
This report is identified by Solar as RDR 1695.
11
-------
CONTENTS
Section Page
1 INTRODUCTION 1
2 SUMMARY 3
3 CONCLUSIONS 7
4 RECOMMENDATIONS 9
5 SYSTEM REQUIREMENT DISCUSSION 11
5.1 System Performance Goals 11
5.2 Design Approach 12
6 CONTROL SYSTEM 13
6.1 Air and Fuel Control System Design 13
6.1.1 System Analysis 13
6.1.2 Air Metering Valve Analysis 16
6.1.3 Fan Selection 20
6.1.4 By-Pass Valve Analysis 20
6.1.5 Fuel Pressure Regulator Analysis 23
6.1.6 Fuel Metering Valve 30
6.2 Control System Component Tests 31
6.2.1 Fuel Metering Valve Calibrations 33
6.2.2 Air Metering Valve Development Tests 36
6. 2. 3 Delta-P Fuel Pressure Regulator Calibrations 51
6.3 Transient Response Emissions Test 53
6.3.1 Description of Demonstration Combustor System 53
6.3.2 Startup and Shutdown Transients 58
6.3.3 Power Level Transient Emissions 59
111
-------
CONTENTS (Cont)
Section Page
7 COMBUSTOR DISCUSSION 63
7.1 Reaction Kinetic Study By Computer Modelling 63
7.1.1 Summary 63
7.1.2 Emissions Analysis 63
7.1.3 Initial Calculation 64
7.1.4 Combustor Design by Computer Modelling 70
7.2 Combustor and Test Rig.Design 89
7.2.1 Combustor Design 89
7.2.2 Combustor Test Rig Design 93
7.2.3 Instrumentation 96
7.3 Combustor Development 100
7.3.1 Summary 100
7. 3. 2 Preliminary Combustor Rig Tests 100
7.3.3 Combustor Pressure Loss Reduction 101
7.3.4 Simulated Vaporizer Tests 105
7.3.5 Combustor Tests With Fan 107
7.3.6 Final Combustor Tests 116
7.3.7 Combustor Noise 121
7.3.8 Ignition Tests 125
7.3.9 Aldehydes and Smoke
8 OPTIMUM DESIGN APPROACH FOR RANKINE CYCLE
COMBUSTION SYSTEM 127
APPENDIX A - Emission Monitoring Equipment and
Procedures 131
APPENDIX B - Test Fuel Specifications 141
APPENDIX C - Fan Noise Reduction Methods 147
APPENDIX D - Emission Data Reduction 159
IV
-------
ILLUSTRATIONS
Figure Page
1 Control System Schematic 14
2 Air Metering Valve 17
3 Air Metering Valve Port Shown in the 60 Percent Power Position 17
4 Fan Characteristics (All Curves at 27V Input Except as Noted) -
Joy Model Number P/N 702-93 21
5 Combustor and By-Pass Valve Pressure Drop Versus Power
Demand • 22
6 Dynamic Head in Annulus 22
7 Fan and Combustor Flow Versus Power Lever Position 24
8 By-Pass Flow as a Function of Power Required 25
9 By-Pass Valve Area Versus Power Required 26
10 Fuel Metering Valve Concept for 100:1 Turndown 31
11 Reynolds Number as a Function of Temperature and Orifice
Size at 1 Ib/hr for Kerosene Through a Square Orifice 32
12 Discharge Coefficient as a Function of Orifice Size for Kerosene 32
13 Fuel Metering Valve Assembly 34
14 Fuel Metering Valve Stroke Measurement Arrangement 36
15 Fuel Metering Valve Final Weight Flow Calibration 37
16 Fuel Metering Valve Performance From 0.5 to 10 Pounds
Per Hour (Final Weight Flow Calibration) 38
17 Air Valve and Fan Flow Test Mock-Up Schematic 39
18 Air Valve Flow Test Mock-Up 40
19 Air Valve Flow Test Schematic 40
20 Air Flow Test Bench 41
21 Flow Control Performance of Air Metering Valve 44
-------
ILLUSTRATIONS (Cont)
Figure Page
22 Air Valve AP Test Results With Mock-Up Air Valve 45
23 Demonstration System Air Valve 47
24 Sound Pressure Level Versus Frequency for Blower Without
Inlet Flow Straightener 48
25 Noise Measurement Room Geometry and Materials 49
26 Fuel Pressure Regulator Installed on P^ - Pg Actuator 51
27 Fuel Pressure Regulator \Vith Molded 0.012 Inch Thick PVC
Diaphragm 52
28 Fuel Pressure Regulator Performance With 0.008 Inch Flat
Rubber Diaphragm 54
29 Demonstration System 56
30 Demonstration System With Long Mixing Duct 57
31 Integrated System Demonstration Test Stand Arrangement 57
32 Startup and Shutdown Emissions 59
33 Power Level Transient Emissions (Combustor Configuration "D" 62
34 Flowchart, Generalized Kinetics Program 65
35 Fuel Droplet Lifetime at Maximum Heat Release Rate (2 x 106
BTU/hr) 66
36 Fuel Droplet Lifetime, 50 Micron Drop Size, at 109, 45.5,
and 1. 09 IbAr of Fuel 66
37 Equilibrium Flame Temperature as a Function of Air-Fuel
Ratio 67
38 Equilibrium Concentrations by Volume of Carbon Monoxide
and Nitric Oxide as a Function of Air-Fuel Ratio 68
39 Primary Equilibrium Composition 69
40 Reaction Set 71
41 Design A - Configuration No. 1 75
42 Design A - Configuration No. 2 77
43 Design A - Configuration No. 3 79
VI
-------
ILLUSTRATIONS (Cont)
Figure Page
44 Design A - Configuration No. 4 81
45 Design B - Configuration No. 1 83
46 Design B - Configuration No. 2 85
47 Emissions at One Percent Heat Release With and Without a
Fice Percent Heat Loss For Design A, Configuration No. 1 87
48 Emissions at One Percent Heat Release With and Without a
Five Percent Heat Loss For Design A, Configuration No. 2 88
49 Side View of Rotating Cup Combustor Assembly 90
50 Front View of Rotating Cup Combustor and Case 91
51 Front View of Rotating Cup Combustor and Case With Rotating
Cup Removed 92
52 Rear View of Rotating Cup Combustor and Case 92
53 Rotating Cup and Motor Assembly 93
54 Schematic of Combustor Test Rig 94
55 Rear View of Fan and Control Valve Assembly Showing Anti-
Swirl Plates Installed 95
56 Combustor Rig Showing Arrangement of Air Metering Orifices 95
57 End View - High Temperature Probe With Triple Radiation
Shield and High Velocity Aspiration System 98
58 Installation of a High Temperature Thermocouple 98
59 Circumferential and Radial Positions of Thermocouple at the
Exit of the Combustor 99
60 Schematic of Emission Pickup Probe 100
61 Air-Fuel Ratio For Minimum Emissions as a Function of
Combustor Air Flow 101
62 Emissions of Carbon Monoxide as a Function of Combustor Air
Flow, at Optimum Air Fuel for Minimum Emissions and Also
With a ±10% Deviation of Air Fuel From Optimum 102
vn
-------
ILLUSTRATIONS (Cont)
Figure Page
63 Emissions of NO as a Function of Combustor Air Flow at
Optimum Air-Fuel For Minimum Emissions and Also With
a ±10% Deviation of Air-Fuel From Optimum 103
64 Effect of Fuel Flow on Emissions at Differing Air-Fuel Ratios
Using Rig Air Supplies 104
65 Combustor Rig With Vaporizer Installed 105
66 Effect of Varying Fuel Flow on Emissions When Air-Fuel is
Maintained at a Constant Value (26/1) and Using a Rig Air
Supply and Boiler 106
67 Air Maldistributions Due to Unstable Diffusion 108
68 Average Radial Profile of Temperature Out of Combustor Using
Rig Air Supply 110
69 Average Radial Profile of Temperature Out of Combustor Using
Fan Air Supply 110
70 Repeatability of Radial Profile of Temperature Out of Combustor
Using Both Fan and Rig Air Supplies 111
71 Radial Profile of Te mperature Out of Combustor at Several
Different Circumferential Locations and Using the Rig Air
Supply 112
72 Radial Profile of Temperature Out of the Combustor at Several
Different Circumferential Locations and Using the Fan Air
Supply 113
73 Circumferential Variation of Combustor Outlet Temperature at
Different Radii and Using Rig Air Supplies 114
74 Circumferential Variation of Combustor Outlet Temperature at
Different Radii Using Fan Air Supply 115
75 Radial Profile at Various Fuel Flows, Using Rig Air Supply 116
76 Radial Profile of Temperature (Final Demonstration Compared
to Preliminary) 117
77 Final Air-Fuel Control System. Discharge Temperature
Variation With Fuel Flow 118
viii
-------
ILLUSTRATIONS (Cont)
Figure Page
78 Emission of NO9 as a Function of Fuel Flow 118
LJ
79 Emissions of CO as a Function of Fuel Flow 119
80 Emissions of HC as a Function of Fuel Flow (Test A) 119
81 Emissions at Various Fuel Flows, With Different Fuels and
at Two Air-Fuel Ratios, With and Without Heat Losses 120
82 Before and After Modification Burner Noise at Location "A" 122
83 Identification of Combustor Noise at Location "B" 123
84 Diagram of Combustor Acoustic Test Location 124
85 Two Fan Sketches - Present and Optimum 130
A-l Beckman Model 315A Infrared Analyzer 134
A-2 Beckman Model 402 Hydrocarbon Analyzer 135
A-3 NO Calibration Results 139
A-4 CO Calibration Results 139
A-5 CO2 Calibration Results 140
B-l Variation of ASTM Distillation Temperatures for the Test
Fuels (Average Values) 144
C-l Effect of Pressure Ratio on Rotor Alone Blade Passing
Frequency Noise 153
C-2 Effect of Number of Blades on Rotor Alone Blade Passing
Frequency Noise 153
C-3 Effect of Pressure Ratio on Interaction Noise Generated at the
Blade Passing Frequency 154
C-4 Effect of Number of Blades on Interaction Noise Generated at the
Blade Passing Frequency 154
C-5 Effect of Vane/Blade Ratio on Interaction Noise Generated at the
Blade Passing Frequency 156
C-6 Effect of Spacing on Interaction Noise Generated at the Blade
Passing Frequency 15G
IX
-------
TABLES
Table Page
I Fuel Metering Valve Weight Flow Calibrations 35
II Air Control Valve Test Results With 2.25 Inch Wide By-
Pass Valve (0.5-Inch Band on Orifice Side of By-Pass) 43
III Pressure Regulator Calibration With 0.008 Inch Flat
Rubber Diaphragm 55
IV Power Level Transient Emissions (Combustor Configuration
"D") 60
V Significant Noise Frequency Peak Levels 124
A-I NDIR Compared to Saltzman 136
A-II Reproducibility Test 137
-------
1
INTRODUCTION
Much of the air pollution in the United States is a result of undesirable
emissions from automotive Otto cycle engines. Each day thousands of tons of toxic
gaseous and particulate pollutants are dumped into the limited volume of the atmos-
phere over the United States. Because of the highly complex and transient nature of
combustion combined with the cold walls of an automotive spark ignition engine,
direct control of emissions in the combustion process is a difficult and possibly
insoluble problem. Control by various post combustion schemes appears to offer con-
siderably more potential. Several methods have been demonstrated and show promise.
However, potential maintenance, service life, and high cost problems make it nec-
essary to also consider alternatives to the internal combustion engine if these methods
fail to achieve the goals. Systems utilizing continuous flow external combustors have
been demonstrated to have low emission levels under steady-state operation. A
Rankine Cycle engine has one of the best potentials of any of the continuous combustion
engines for automotive application. The EPA is presently supporting basic research
studies to optimize the low pollution featui es of burner designs for Rankine engines.
The basic problem resolved by this program was the demonstration that a
Rankine cycle combustor system designed to be integrated into a "family car" could
meet the 1980 Advanced Automotive Power Systems (AAPS) goals. A major portion
of the effort for such a system has been devoted to the development of a full scale
(2,000, 000 BTU/hr) prototype combustor system and controls with the desired low
emission characteristics. Automotive requirements for duty cycle, compactness,
cost and efficiency are the constraints that challenge the present state-of-the-art for
combustors. Rapid starts, high response, frequent shutdown, with large and frequent
transients of power level demand, are required automotive performance factors that
cause existing combustor designs to fall short of emission goals.
This program addressed itself to these problems by applying a new and novel
fuel atomization and precise air-fuel ratio control concept to the automotive Rankine
combustion system problem. Rotating cup atomization with a full range air-fuel ratio
control is ideally suited to the wide fuel flow variations necessary for automotive duty
cycles requiring low emissions because of all the various methods of fuel injection,
it has the widest possible range of operation with a single fuel atomizer. Experience
has indicated a major portion of the emissions result from load transients in conven-
-------
Lional combustors. The problem of making an on-off system emission-free is tech-
nically difficult, if not impossible. A combined on-off and modulated system share
some of the problems of both and has greater complexity. Thus, this program
developed a fully modulated system incorporating a rotating cup fuel atomization
system and all necessary controls to supply and regulate both fuel and air at high
response rates while maintaining the optimum air-fuel ratio for lowest emissions.
The purpose of this program was to apply modern analytical and experimental
tools to the design, fabrication, development and demonstration of a low emission
combustor system for an automotive Rankine cycle engine. The result of this study
has been two-fold. First, the capability of meeting the emission goals has been
demonstrated. Second, areas, which require further development in order to make
the system usable in an automobile, have been identified.
-------
2
SUMMARY
The primary objective of this program was to apply modern analytical and
experimental techniques to the design and demonstration of a low emission combustor
for an automotive Rankine engine. Results would show if the 1980 AAPS emission goals
could be achieved and determine technology requiring advancement to equip a "family
car" with a low emission combustor. Excellent progress had previously been made
with combustors to obtain low unburned hydrocarbons (HC) and carbon monoxide (CO)
at rated steady state operation but they could not meet the nitric oxide (NO) emission
levels nor could they operate over the wide fuel flow range and rapid transients required
for the automotive Rankine cycle engine.
Two approaches are open for emission control from automobile engines —
limit NO formation during combustion or eliminate NO after it is formed. Spark
ignition engines can generate reducing exhaust gases and therefore eliminate NO by
catalytic reduction with CO. Rankine cycles, gas turbines, and diesels cannot run
fuel rich because of smoke production, inefficiency and overtemperature, and there-
fore cannot generate the necessary reducing exhaust gases. Thus the limit NO for-
mation approach must be employed in the combustion zone.
The Solar approach has been to obtain low NO emissions through air-fuel
ratio-control in combustion zones. A schematic shows the air-fuel ratio control
and residence time for the primary, secondary, and tertiary zones of the combustor
which were demonstrated to perform well within the established 1980 AAPS emission
goals.
-------
872 LB/HR AIR
436 LB/HR AIR
SOLAR-LOW EMISSION-CONTROLLED ZONE-COMBUSTOR
654 LB/HR 872 LB/HR
AIR AIR
^
FUEL
DROPLET
PATTERN
109 LB/HR
FUEL
V V
SECONDARY
ZONE
18 A/F ^
TOTAL W
1962 LB AIR
109 LB FUEL
V. J
If
TERTIARY
ZONE
26 A/F ^L
TOTAL V
2834 LB AIR
109 LB FUEL
> , i 1
0.0 0.002 0.004 0.006 0.008
RESIDENCE TIME (SEC) - 109 LB/HR FUEL
0.01
The rotating cup atomization system yields precision fuel droplet size and pattern
independent of fuel flow. Precision air and fuel metering values allows scheduling
of the desired air-fuel ratios over the 100 to 1 heat release estimated as a maximum
range necessary to cover all engine systems.
The volume of the demonstrated combustor would be 1.1 cubic feet for the
optimum design which included an optimum fan, demonstration controls and combus-
tor. This corresponds with an initial goal of 1.3 cubic feet. The parasitic power is
computed at 1.25 horsepower without the vaporizer and can be substantially reduced
at part load with minor modifications of the demonstrated control system. Fan and
combustor noise was above acceptable limits during many test conditions. An
analysis of the noise indicated that noise could be reduced to acceptable limits by use
of special design approaches for both the fan and combustor.
A fully modulated air-fuel control system has been designed and was demon-
strated in integrated combustor emission tests. All major components of the system
except the fan were specifically designed to obtain accurate control of air-fuel ratios.
Design emphasis has also been placed upon high response capabilities necessary for
low emission city traffic operations of automotive vehicles. Stop and start duty cycles
require an almost continually fluctuating heat release rate from the combustor system.
On-off control modes, although simple, were rejected because of their potential for
higher emissions during the frequent start-up and shutdown transients required in
automotive Rankine engines. The system developed basically consists of a variable
area air valve mechanically linked to a variable area fuel valve. Integrated with the
-------
basic area control is the air supply (electric fan) and the pressure control system (air
by-pass valve). A fuel supply (electric pump) and fuel pressure regulator subsystem
are incorporated in a manner that allows for compensation of fan motor voltage re-
ductions or other changes that tend to vary the desired air-fuel ratio. High response
to power level demands is provided, while a prescheduled air-fuel ratio is maintained
to minimize emissions.
In both the air and fuel metering systems, a constant (or proportionally
controlled) pressure differential is maintained across the variable area valves. Thus
the 100 to 1 flow range is obtained by providing a 100 to 1 area ratio in each valve.
A low fuel supply pressure, 10 psig, has been selected to minimize the system's
sensitivity to contamination and allow the use of inexpensive pumps. A fuel pressure
regulator is integrated into the system to allow the air-fuel ratio to be essentially
independent of fan motor voltage (or fan efficiency, leakage, brush wear, etc.). An
important feature of incorporation of this fuel pressure control scheme is that part
load parasitic power demands of the fan motor can be reduced by as much as 65 per-
cent by scheduled voltage reductions while still maintaining the air-fuel ratio indep-
endent of the inertial lags of the fan motor.
A series of demonstration tests at various steady state and transient power
levels were performed on the integrated combustion system. Emissions monitored
during transients and indicated that the peak levels of emissions would remain below
the limits at a transient rate of 50 percent power level change per second (54 Ibs/hr
per sec). In most transient ranges the rate of change of power level could go as high
as 150 percent per second without significant emission peaks. However, decreasing
power transients below 30 Ibs/hr at rates of slightly above 50 percent per second
were observed to produce emission peaks of CO above the limits for three or four
seconds. Startup from cold to maximum firing rate in three seconds or less appeared
to present no significant emission peaks. Steady state emission levels were in
general significantly below the 1980 AAPS goals. Measured values across the entire
heat release range were on an average 61 percent below the goal for CO, 90 percent
below the goal for HC and 37 percent lower in the case of NO. The only condition that
caused an emission level above the goal was at 1 pound per hour fuel flow. At this
low flow condition CO levels were above the limit. It should be noted that 1 pound per
hour fuel flow is an extreme boundary normally outside the range of typical engine
systems.
Basically the component and system demonstration tests have indicated the
system can provide the necessary degree of air and fuel regulation for a low emission
combustion system that has imposed on it a wide heat release range and frequent
high response power level changes. Construction, size, weight, and reliability
features of the system have the potential to be incorporated into inexpensive mass
production automotive engines.
-------
3
CONCLUSIONS
This program has demonstrated that a Rankine cycle combustor with a two
million BTU per hour heat release in a 1.33 cubic foot volume operating on JP-5 fuel
can meet the 1980 AAPS emission level goals over a 100 to 1 heat release range at
steady state or during rapid transients. The transients include startup to full power
in three seconds and a 50 percent power rate change per second.
The program has further demonstrated that the package size and power
requirements of combustors need not be excessive. The components used are not
complex and are capable of being mass produced at low cost. All of the components
employed proven concepts that appear capable of a high degree of reliability.
The novel fuel atomization system assures that the fuels can be rapidly
ignited in even the coldest weather and does not need warm up to maintain low emission
as do the current spark ignition engines. Further the fuels used need no special
additives for combustion control and a wide variety of currently available fuels can be
used (however, leaded gasoline is not acceptable).
The novel rotating cup has been shown to be essential for low emission control
over the 100 to 1 heat release range. The precision air and fuel metering valves and
controls have also been shown to be essential to maintain the low emissions over the
100 to 1 heat release range.
Fan and combustor noise was shown to be a problem, thus special attention
must be given to its elimination in future development programs.
-------
4
RE COMME NDATIONS
The demonstrations were conducted to show that given certain boundary
conditions, such as heat release range, fuel and emission levels, a Rankine cycle
combustor could perform satisfactorily and provide incentive for further Rankine
cycle engine development. All major components except fan and cup drive motor
were specially designed and developed to conduct this investigation. Further develop-
ment is necessary, especially in regard to integration with other engine components
because they effect both the shape and final performance of the combustor. The
principal component having the greatest effect on emission levels will be the vaporizer.
The temperature boundary conditions on the combustor will be altered by the addition
of the vaporizer and thus need special development. Further the emission levels will
be altered because of boundary conditions, heat up times required, response time, and
stability of the vaporizer. All of these factors must be considered when coupling the
vaporizer with the low emission combustor. The work conducted on the program has
demonstrated that these problems are solvable and that a combustor with emissions
well under the goals can be expected.
Minimum fan power must also be demonstrated for the optimum design
Rankine cycle combustor. Fan power, system volume and uniform air distribution
(air-fuel ratio control) require the design of a fan that is aerodynamically and geo-
metrically integrated with both the air valve and combustor. The fan motor repres-
ents one of the largest inherent cost factors in the Rankine cycle combustor control
system. Minimum fan power becomes a prominent factor in a continued development
program. Elimination of a separate cup motor by using an extension shaft on the fan
motor will also be necessary to minimize cost and volume. Additional development
testing is necessary to match the cup and fan speeds to obtain satisfactory atomization
and fan aerodynamic design at the same shaft speed. The combustor pressure drop
must be held at a minimum. However, if the pressure drop is too low emissions and
temperature distribution into the vaporizer and exhaust system will become a problem
and can cause hot spots and low reliability of the vaporizer. Further test analysis,
using a vaporizer and optimum fan should be conducted to establish engineering require-
ments necessary for minimum emissions with a reliable automotive vapor generator.
Vaporizer air side pressure drop must be held at minimum to conserve fan
power. However, the weight and response time for the vaporizer becomes excessive
-------
if the pressure drop is too low. An optimum balance will require an integrated
development program.
Transient performance of the combustor-vaporizer system must be further
demonstrated. One of the critical items is the start and warm up period to roadload
power of the complete Rankine cycle engine. Time must be maintained at a minimum
and therefore will require optimum performance of each component during this period,
while still maintaining minimum parasitic power. Further development and integrated
system tests are essential to demonstrate transient and warm up characteristics with
a vaporizer.
A final but important consideration is the development and demonstration of
an integrated vapor generator and fuel control system. Response, temperature
distribution, heat flux, wall temperature, flame radiation, thermal inertia, feed
control, fluid hot spots, and stability must be considered in a high response control
system. It is necessary to demonstrate that fuel and air flows can be made to respond
rapidly and accurately enough to regulate the vapor pressure and temperature within
acceptable limits as vapor flow rates respond to automotive duty cycles. A low cost
control system must be developed without sacrificing the inherent low emission
characteristics of the combustion system.
10
-------
5
SYSTEM REQUIREMENT DISCUSSION
The purpose of this program was to design, fabricate and test a full scale
combustion system suitable, with minor modifications, for installation in a Rankine
cycle engine powered vehicle. Included in the system was the combustor, fan,
controls, ignition system and fuel pump.
5.1 SYSTEM PERFORMANCE GOALS
The performance goals of the combustion system are as follows:
•Maximum heat release of 2,000, 000 BTU/hr (which corresponds to
109 pounds per hour fuel flow) and a minimum heat release of 20, 000
BTU/hr which results in a 100 to 1 turndown ratio. The maximum
heat release was stated as a time averaged rate, thus permitting use
of an on-off or modulating system or a combination of both.
•Steady state emissions - 1980 AAPS goals
Corresponding gram
gm Pollutant pollutant per mile
Pollutant Kilogram of Fuel (assuming 10 mpg)
Carbon monoxide 16.25 4.7
Unburned hydrocarbons
reported as CI^ 85 0.48 0.14
Oxides of nitrogen
reported as NO2 1.38 0.4
Participates 0.10 0.03
In addition no visible smoke is allowed at any operating condition. Meeting
these goals results in high combustion efficiency, which was also listed as a
goal.
11
-------
• The steady state emission goals are to be met within 2 seconds after
any change in fuel flow. There is to be no severe degradation of
emission levels during a transient.
•Startups and shutdowns are to be as clean as possible since they are
included in the latest automotive emission test procedure.
The combustion system is to have the following constraints:
•Maximum combustor volume - 1.33 cubic ft.
• Maximum parasitic power - 2. 0 HP (without the vaporizer)
•Test fuel - Kerosene, #1 Diesel, Jet A
• Test conditions - ambient pressure and temperature
5. 2 DESIGN APPROACH
Reviewing the requirements the major problem was identified as maintaining
low emission levels over a wide range of heat release rates or fuel flows. To obtain
low emission levels of CO, HC and NO simultaneously involves a careful trade-off of
combustor parameters such as local and overall air-fuel ratios, peak temperatures,
residence time and velocity. In general me factors which contribute to low NO promote
high CO and HC and vice versa. This will be discussed in greater detail in Section 7. One
conclusion established was that me air-fuel ratio must be closely controlled over the
entire operating range. The air control valve and fuel control valve are linked to-
gether in such a way as to provide the desired overall air-fuel ratio at each fuel flow.
The local air-fuel ratios are determined by location, size and number of the combus-
tor air entry holes.
Based on published data and our own experience that startups and shutdowns,
i.e., turning the combustor on and off, would result in substantial emissions, it was
concluded that the system must be completely modulating over the entire range. In
order to accomplish full modulation a fuel atomizing device which would atomize the
fuel adequately at flow rates from 1 to 109 pounds per hour was required. Not only is
the wide turndown ratio a problem but so is adequate atomization of a fuel flow as low
as 1 pound per hour. Our investigation indicated mat the rotating cup fuel atomizer
was the most likely device to fulfill the requirement.
The major components of the demonstration system are the air supply system,
air control valve, fuel control valve, fuel pressure regulator valve, and the combustor
with its rotating cup fuel atomizer. The selection and/or development of these com-
ponents is discussed in the following sections.
12
-------
6
CONTROL SYSTEM
6.1 AIR AND FUEL CONTROL SYSTEM DESIGN
Low emissions were the main design criteria used to synthesize the control
system. Analysis and experience determined that a fully modulated control of both
air and fuel across the full range of heat release rates would offer the optimum low
emissions approach. Inherent with this design concept was the necessity to maintain
the ideal air fuel ratio at any heat release rate from one percent to 100 percent. In
addition to the wide range of heat release rates necessary in automotive systems, the
very frequent and high response demands for power level changes dictates that air-
fuel ratio control must be free of fan motor inertia time lags. A fully modulated
control system independent of inertial lags of fan motor was designed with simple
hydromechanical control components that have the potential of being incorporated
into low cost automotive type engine systems.
6.1.1 System Analysis
Experience and analysis has shown that the optimum air flow control and
combustion air delivery system requires a high degree of symetry to provide uniform
flow (and consequent correct air-fuel ratios) into the combustor. Figure 1 schematic-
ally describes the overall air and fuel control system. Air mass flow is controlled
by varing the flow area of 12 orifice ports leading directly into the combustor outer
casing. To ensure that fuel flow is directly proportional to air flow, a mechanical
coupling syncronizes the control of fuel flow area to air flow area. Once the area
ratios have been fixed, it is only necessary to regulate pressure drops to maintain
the desired weight flow ratios.
For low pressure rise, air compression can be neglected and the weight
flow relationships can be written as:
13
-------
BYPASS VALVE
. J
ilX.' -V., lii^CiiU. BAFFLE
POWER ,
LEVER V
±1.0"
FUEL METERING
VALVE 1.0 to 109 LB/HR
^ -g_.
.\\\\ A\\\\\\\\\*A\\Wp]
FIGURE 1. CONTROL SYSTEM SCHEMATIC
-------
(2)
where W = mass air flow into combustor
a
A™ = air metering valve area
PI = upstream pressure (fan discharge)
P9 = downstream pressure
u
T = air temperature at metering valve
Wf = fuel mass flow
APf = pressure drop across fuel valve
AT = fuel metering valve area
If the inlet pressure (altitude) and temperature are constant, a simple relationship
can be written for the air-fuel ratio:
(3)
To ensure the air fuel ratio remains a function of flow area only, the pressure
drop across the fuel valve is controlled proportional to the pressure drop across the
air metering ports. A APr regulator valve performs this function by a force balance
across diaphragms. Operation of this component is illustrated in Figure 1. P^
pressure is connected (through a density compensator) to the bottom side of diaphragm
A^. Downstream pressure (?2) is connected to the opposite side of the diaphragm.
Fuel system pressure is regulated by a flapper valve that by-passes excess fuel to the
tank. A small diaphragm on the fuel side balances the pressure difference across the
air metering ports Aj^ against the fuel pressure. If the air pressure drop increases
(causing Wa to increase), the force across the system becomes unbalanced and the
fuel pressure regulator moves up; reducing the by-pass flow, and thus increasing the
fuel pressure and its mass flow across the fuel metering valve. Since the force
balance must be maintained across the 4Pf regulator, we have:
ZForces = Ag (4Pf - PZ> = \(P±-'?2)
where P.J = P, adjusted for inlet air density and where dP* is large compared to P2
we have AP = - (P ' - P ) (4)
I A 1 2
15
-------
thus equation (3) can be rewritten as
- P2> A2
" = constant
Fan speed changes due to voltage or load variations, fan efficiency reductions
due to fouling or wear, and by-pass valve leakage variations are automatically com-
pensated by maintaining the APf as a function of the air valves pressure differential.
It can be seen from equation (1) that the air flow is a function of
T!
Thus, an altitude and inlet air temperature correction can be incorporated by adding a
PI/TJ compensator (Fig. 1). Fan discharge pressure (P,) is admitted to the AP
regulator by a contoured needle valve. This valve is positioned by a force balance
across a diaphragm having P , T^ on one side and cavity sealed with air at standard
conditions. Thus, as the air temperature increases, the sealed air will expand and
move the valve upwards. This action will increase the pressure drop across the con-
tour valve and thus reduce P, to a corrected value of PI on the bottom side of
diaphragm A. As P^ decreases, the fuel pressure will drop producing the desired
fuel flow reduction as the ambient air temperature increases.
6.1.2 Air Metering Valve Analysis
Air to the combustor is metered across a series of twelve ports cut into two
plates (Fig. 2). An input lever rotates one plate with respect to the stationary backup
plate. The metering port areas are caused to open or close as the input power lever
rotates matched ports in each plate. Figure 3 shows one of the 12 port configurations.
The area is established by maximum mass air flow requirements of the system:
Maximum Power = 2 x 106 BTU/hr.
Let LHV of fuel = 18, 350 BTU/lb for JP-5
.'. fuel flow = 109 Ibs/hr or 1.82 Ibs/min.
Air fuel ratio at maximum rated heat release = 26 (See Sec. 7)
.'. Maximum Air Flow = (26) (1.82) = 47. 3 Ibs/min = 630 SCFM
16
-------
FIGURE 2. AIR METERING VALVE
OPENING IN
METERING PLATE
ROTATION
TO OPEN
OPENING IN
BACKUP PLATE
FIGURE 3. AIR METERING VALVE PORT SHOWN IN THE
60 PERCENT POWER POSITION
17
-------
The area to flow this quantity of air depends upon the selection of a design pressure
drop across the metering valve. It is desirable to maintain this loss as low as possible
to reduce parasitic power requirements. However, very small pressure differences
require larger and heavier valves. A tradeoff analysis indicated that a AP of 2 inches
of H2O results in a valve size that is within the dimensional envelope of the combustor:
Flow relationship =
where
Allow
Assume
For
AM = in. (area of orifice)
y = lb/ft3 (density at standard conditions)
Wa = Ibs/min.
P = in. ofH2O
AP = 2 in. of H2O
K = 0. 6 discharge coefficient
y = 0. 075 and a maximum flow rate condition of 47.3 Ibs/min.
47.3 = 7.617(0.6) AM ^(0.075) 2
AM = 26.8 in.
Since a constant AP will be maintained, the area will be varied proportional to the
power lever position, X, (1 to 100%) modified to allow leaning of air-fuel ratio at low
heat release rates (Fig. 3). (Combustor tests had indicated leaning out was necessary
for minimum emissions.)
Delta-P, Voltage Compensation System
The pressure differential across the fuel metering valve can be regulated to
maintain the correct air-fuel mass flow ratio regardless of inlet air temperature or
absolute pressure (altitude). Initially the system was designed to respond to relatively
small variations of voltage ±10 percent (equivalent to ±10% flow variation from the fan).
A regulator capable of compensating for voltage changes of ±10 percent was considered
adequate for a simple constant voltage motor concept. As component testing progressed
it became evident that modifications to the regulator could extend its range sufficiently
to accommodate a two speed fan. By incorporating this into the system, a 65 percent
reduction of parasitic power at normal driving ranges can be achieved along with a
substantial noise level reduction. This modification was incorporated and tested at a
18
-------
component level in the delta-P regulator. However, since a two speed fan system
requires a redesigned by-pass valve, program schedules required that the emission
demonstration tests be conducted with the single speed system.
An analytical estimate of the part load power reduction presently possible can
be made from component test results. Development tests on the regulator valve have
shown it capable of accurately regulating the fuel pressure directly proportional to
pressure differential at the air metering valve across the range of 1. 0 to 2. 5 inches of
H2O. This wide range control allows the fan motor voltage to be reduced as a function
of power lever position. If the air by-pass valve is maintained closed for the power
lever position range from 100 to 70 percent, the air flow into the combustor can be con-
trolled by proportionately lowering the voltage to the fan while the regulator compensates
for the reduced pressure differential across the air metering valve. Since a mechanical
linkage maintains a known ratio of air valve area to fuel valve area, the regulator will
automatically maintain the air-fuel ratio as the power is reduced. A limitation on this
system is the lowest pressure differential signal that can be utilized as an actuator input
to the simple diaphragm regulator. Analysis and test results have shown that 1.0 inch
of H2O is a reasonable lower limit. Using this value, the part load power demands of
the combustor system can be established. Although the combustor requirements are
moderate (1.25 HP for the optimum configuration and 2.3 HP for the demonstration
system), addition of a boiler can make parasitic power losses at part loads unacceptable.
If a boiler had a high delta-P (as recommended by some designers*), an additional 2. 4
horsepower would be required. Total parasitic power levels would then be as high as
3. 65 horsepower. If the system were to require this high power level from full power
down to idle condition, a highly undesirable condition would exist. Lowering the input
voltage across a voltage range that allows accurate AP compensation by the regulator
will eliminate the motor inertial speed lags since the air-fuel ratio changes are con-
tinually maintained by the regulator. Power reduction by operating at 1. 0 inch of H9O
^
can be estimated by simple calculations based on fan laws for a series wound motor.
For voltage changes of approximately 2 to 1, the following relationships give reasonably
accurate results.
For a given voltage change V^ to V2
flowQ: Qj/Qg = V1/V2
pressure p:
fan BHP is: BHP,/BHP9 = (V,/V9)3
1 £• ±6
By maintaining the by-pass valve closed and reducing the voltage to the motor,
the flow through the metering valve will drop as a function of both the motor voltage
and the area change of the metering valve. Pressure will drop as a function of the square
* "Condensers and Boilers for Steam-Powered Cars: A Parametric Analysis of Their
Size, Weight, and Required Fan Power", by William C. Strack, NASA TN D-5813,
May 1970.
19
-------
of flow. By using 1.0-inch as the lower limit of adequate control, the by-pass valve
can be maintained in a closed position until the voltage is reduced to Jl/2 or O.TOTV^.
Since power is approximately a cubic function, we will obtain (0. 707) (BHP), or a
65 percent reduction in power. With the optimum system and a relatively high air
side boiler pressure drop, a parasitic power loss reduction of approximately 2.4 horse-
power occurs. Total power with this high pressure drop boiler would be approximately
1.25 horsepower at vehicle power demands below 35 percent. One problem with this
approach is that it is more difficult to make the system linear. However, since this
is not a driver input command, linearity is of secondary importance if the correct
air-fuel ratios can be maintained.
6.1.3 Fan Selection
Fan pressure requirements are established by the selected metering AP of 2
inches of F^O added to the combustor maximum rated pressure drop of 8 inches of
H^O. A number of fan designs were investigated but program schedules required the
use of an off-the-shelf unit. A Joy P/N X702-93 was selected based upon its immediate
availability and performance. To match the fan's output characteristics to the com-
bustor flow and pressure requirements, the fan was operated above its design voltage
level. In Figure 4 the characteristics at both design (27V) and the increased voltage
(33V) used to match the fan indicates a significant electrical input power increase.
Final electrical power inputs were approximately 2.4 HP at the less efficient 33V
overload operation point. This power includes all losses associated with distribution,
air turning, air valve friction and combustor. This power can be reduced by a large
percentage by use of a fan specially designed and matched to the aerodynamics of the
air valve and combustor. It should be noted that in an integrated system, the
pressure drop across the vapor generator and its associated ducting would also have
to be added to these pressure rise requirements. A general characteristic of fans
is to increase flow as the output pressure decreases. In order to meter the air from
100 to 1 percent at a high response rate independent of motor inertia! effects, a by-
pass valve has been incorporated. It allows excess fan air at low power settings to
be returned to the atmosphere thereby establishing the correct pressure character-
istics across the air metering valve and combustor.
6.1.4 By-Pass Valve Analysis
The by-pass valve is designed to regulate the delta-P across the air
metering valve at a constant 2 inches of HgO regardless of power lever settings (1 to
100% range). As the power is reduced from 100 percent, the required combustion
air flow must be reduced proportionately. The corresponding pressure drop across
the combustor varies as a function of flow squared as shown in Figure 5. Since the
by-pass valve is designed to maintain a constant air delta-P of 2 inches of ^O
across A^j, the pressure drop across the by-pass valve is equal to the combustor
20
-------
ELECTRIC INPUT (HP)
1.4
200
400
600 800
VOLUME IN CFM
1000
1200
1400
FIGURE 4. FAN CHARACTERISTICS (ALL CURVES AT 27V INPUT
EXCEPT AS NOTED) - JOY, MODEL NUMBER
P/N 702-93
pressure drop plus 2 inches of H-O (shown in Fig. 5). As the back pressure on the
fan is reduced, its output flow increases (Fig. 4) and must be accommodated in the
design of the by-pass valve ports. This increasing flow characteristic causes a
significant increase in the dynamic head in the fan annular flow passage (Fig. 6).
A baffle mounted immediately downstream of the fan discharge is designed to reduce
the effects of the dynamic head by turning and directing the flow into the larger
plenum-like space containing the valve elements.
By means of this baffle arrangement, it was possible to establish the by-pass
valve size as a function of static pressure rise.
21
-------
10 Inches
4P - (P1 - Pg) - 2 Inches
AP
COMBUSTOR
20
FIGURE 5.
30 40 50 60 70 80 90 100
X POWER LEVER POSITION,%
COMBUSTOR AND BY-PASS VALVE PRESSURE DROP
VERSUS POWER DEMAND
Q
u
X
U
i
z
B
600
800
1100
1200
FIGURE 6.
800 1000
VOLUME FLOWRATE, CFM
DYNAMIC HEAD INANNULUS
1300
22
-------
Air flow across the by-pass valve is equal to the difference between fan flow
and the combustor flow (Fig. 7). A plot of air flow through the by-pass valve (Fig. 8)
is obtained by subtracting the two curves plotted in Figure 7. By-pass valve area
versus power lever position can now be calculated from the known pressure drop and
required weight flow. Figure 9 shows the resultant area requirements. It can be
seen that the area deviates only slightly (~5 percent) from a linear valve flow area
across the entire 100 to 1 flow range. At the maximum power conditions (X = 100%),
the valve is fully closed and progressively opens to its full area of 52 square inches at
the 1 percent position. This action is exactly opposite the air metering valve sequence,
thus both are mechanically coupled to ensure correct high response matching.
6.1.5 Fuel Pressure Regulator Analysis
The fuel pressure regulator serves the basic function of maintaining the
correct air fuel ratio in conjunction with the fuel and air flow area control valves.
Its incorporation into the design allows the system to operate across the 100 to 1 fuel
flow range with sufficient accuracy to minimize emissions. Inertial effects of fan
motor leakage, blade fouling, motor performance degradation, air cleaner blockage,
and pressure gradients across the fan inlet can be compensated as can scheduled
voltage reductions to lower parasitic power demands at reduced loads. The demon-
stration system incorporates the basic delta-P regulation system that allows the valve
to regulate the fuel pressure in proportion to the volumetric air flow. This feature
allows compensation for the factors described above. Although a secondary to the
primary goal of the program, a design has been completed to provide an additional
compensation for altitude and inlet air temperature. This added compensation has
not been incorporated into the demonstration system, but analysis indicates a simple
mechanical system will perform the task.
Delta-P Compensation System
Delta-P compensation can be obtained with a single diaphragm actuated
regulator valve. Static pressure upstream (P,) and downstream of the air metering
valve is sensed and transmitted to either side of the delta-P diaphragm actuator
(Fig. 1). Neglecting the small correction between P^ and P^' (discussed in the next
section) we can analyze the action of the delta-P regulator by letting P^' = P,.
The differential pressure P^ - P£ across the large diaphragm produces a force
which is opposed by the differential pressure across a smaller diaphragm, one side
of which is subjected to fuel pressure upstream of the fuel metering valve, and the
other, to P2. Neglecting the effect of the springs which function only to provide
stability and centering action of the movable diaphragm assembly, the forces will be
in equilibrium if:
(Pfuel - P2> A2 ' Al '
23
-------
100
90 h
:iV/
if
-------
90 -
10 20 30 40 50 60 70 80 90 100
X POWER LEVER POSITION. %
FIGURE 8. BY-PASS FLOW AS A FUNCTION OF POWER REQUIRED
25
-------
100
90
80
70
ge
<60
w
«
< 50
£
J 40
UH
30
20
10
100% = 52 In.
= CONSTANT 2 INCHES OF HO
0 10 20 30 40 50 60 70 80
X POWER LEVER POSITION %
90 100
FIGURE 9. BY-PASS VALVE AREA VERSUS POWER REQUIRED
The system is so designed that fuel delta-P across the metering valve will always
remain relatively high (approximately 10 psid). On the other hand, fuel pressure
downstream of the metering valve enters the burner through the rotating cup at a
pressure substantially equal to combustor pressure, Pc. Thus, the pressure drop
across the metering valve equals:
APt = Pc . - (P. + P )
f fuel loss c
where
loss
= fractional flow loss between fuel valve and combustor.
Since P, varies from 0 to 0. 5 psi and PC differs from P by approximately 0. 3 psi
(for a low loss boiler), the total error is 0. 8 psi or ±0. 4 psi. Since pressure
variation about 10 psig is ±0.4 psi if bias springs are incorporated flow error is thus
equal to:
= ±2% flow accuracy
10
26
-------
Thus, for purposes of analysis, we can set the equilibrium to:
4pf = 7T •
Lf
Fuel entering the small diaphragm cavity leaves through a flapper-type valve
which opens or closes as the diaphragm moves in response to changes in the sensed
pressures, and by passes the fixed-displacement pump output back to the tank,
thereby controlling the fuel pressure so that force equilibrium is always maintained.
To a high degree of accuracy, therefore, the fuel delta-P is controlled proportional to
the applied air delta-P, according to the above equation. The large diaphragm (Aj)
is 7.15 inches in diameter and has an effective area of 40 square inches. Small
diaphragm (A2) is 0.6 inch diameter and has an area which is only about 0.28 square
inch. Thus fuel delta-P will be 140 times as great as the air delta-P. The air throttle
valve delta-P, P^ - P2, is nominally 2 inches HgO. With voltage reductions of as
great as 30 percent, the normal range of operation will be 1 to 2 inches of HgO with
corresponding fuel pressures of 5 to 10 psig.
Caution has been taken in the design to eliminate sources of friction whenever
possible, even when this is at slight sacrifice in theoretical accuracy. A balanced
design of the by-pass flapper valve was considered and was rejected because it would
have introduced possible hysteresis. The design shown i8 not balanced, but the forces
due to the imbalance contribute a flow error of less than 1/2 percent.
Air Density Compensator
Although not incorporated into the demonstration system, a design analysis
has been completed that indicates a simple mechanical system could be used to obtain
inlet air density compensation. The variable air pressure, P1 applied to the delta-P
control, is modulated by the air density compensator in such a way that the following
relationship is obtained:
pi' - "2 ' K' r
A small amount of air at pressure, P-p sensed upstream of the air throttle valve, flows
through a variable orifice, controlled by movement of a density- responsive diaphragm,
and into the delta-P control valve diaphragm cavity at the pressure P', finally leaving
through a fixed orifice, where it connects to the P2 pressure sense line downstream of
the air metering valve. This small amount of air which effectively by-passes the AM
is a negligible fraction of the burner flow, even at minimum power levels.
27
-------
For low air delta-P's, the air flow equations for the first and second orifices
(areas An and AO, see Fig. 1) and steady state continuity of mass required that
-T (P -P') = K A -- ,/(P ' -PJ
This can be solved for P ' - P in terms of P - P as follows:
P! - P
1 2
1
/A ^
. -^
£ P '
f =1 + iJ
A / P
n' 1
On the basis of absolute pressures, P..' is never more than one part in 200 different
from unity, thus we can simplify to:
P ' - P —
Pl P2
1
/A N2
L / o \ . J
(P1 ' P2>
+ 1
Therefore, the necessary density correction can be achieved simply by contouring
the variable orifice needle so that
= K'
+ 1
Now, the density compensator consists, essentially, of a sealed cavity, closed at one
side with a flexible air-tight member, within which is trapped a predetermined weight
of air (neglecting the effect of the spring which serves only to apply a gently loading
action to eliminate diaphragm slack), the pressure inside the sealed cavity will always
be substantially equal to P., applied to the opposite of the diaphragm. Similarly, for
steady state conditions, considering the variations in temperature are due almost
entirely to changes in day temperature, the temperature of air inside the sealed cavity
will be substantially the same as the temperature, T^, of the air flowing across the
opposite side of the diaphragm and surrounding the control. It will be required that the
temperature of the air flowing across the diaphragm be representative of the actual air
28
-------
temperature at the metering valve. Although the air is being bled directly from up-
stream of the valve (P, cavity) it may not be of sufficient mass flow to rapidly compen-
sate for hot soak conditions under the hood. By insulating the sealed cavity and an
additional bleed of air from P^ around the bottom of this cavity, a much faster adjust-
ment to inlet temperature variations could be made without undue error resulting from
underhood temperature gradients. The volume of the expandable cavity will be
where R is the gas constant, and U)is the weight of entrapped air. Since the volume is
proportional to depth of the cavity, Y, which varies with diaphragm position, we have
v -
The desired contour for the variable needle is, therefore, given by
A
_ n
A~
o
Since density varies only by about 20 percent over the complete range of operation,
the variable orifice configuration can be selected to operate in a fairly linear range.
Once the appropriate contour has been established, any required periodic readjust-
ment of the density compensator is easily achieved by removing the compensator
assembly so as to expose the end of the contoured needle protruding through the
variable orifice. Then, while the plug in the sealed cavity is opened, the needle is
held in a prescribed position, depending on day temperature and altitude, and the
cavity is then resealed. With the correct adjustment, the net differential pressure
applied to the delta-P control valve will be properly compensated for air density so
that fuel delta-P becomes:
Ai pi
4p( "T KT2
then the fuel-air ratio to the burner will be constant from equations (1) and (2)
29
-------
w
a
Wf
A K
a a A
A(Kf \
/P1P1-P2
' P
1 Pf
AaKa
AfKf
- P2>K'
Thus, the optimum air-fuel ratio will be maintained as a function of the area ratios
selected for the air and fuel valves.
6.1.6 Fuel Metering Valve
Fuel must be metered from 109 to 1. 0 pounds per hour. At low power
settings (1 pound per hour), the flow is approximately two drops per second. Standard
valves do not have sufficient range to accommodate these severe requirements with
sufficient accuracy for a low emission combustor. Additionally, the valve should not
be sensitive to temperature induced fuel viscosity changes. A new approach to this
problem has been taken by the application of a dual slotted shear valve (Fig. 10). Two
flat (ground and lapped) plates with matched contour slots 90 degrees to each other.
At the intersection of the two slots, a square orifice is formed whose area is a
function of the relative position of the top movable plate. The square shape (and thus
the discharge coefficient) can be maintained constant throughout the entire 100 to 1
area ratio. Since the plates are in contact, fuel will flow only through the slot in each
plate and not between the plates , thereby reducing the clearance leakage path to the
microfinish of the contacting surfaces. A drain groove is provided between the up-
stream pressure and the metered outlet fuel passage, thereby reducing the leakage
pressure potential to the level of the frictional flow loss to the rotating cup. Since
this is normally less than 0. 5 psid, resulting leakage of metered fuel into the drain
system will be negligible.
The size of the orifice slots used is a trade-off between four factors.
•Fabrication capabilities - requires large dimensions
•Contamination - requires large dimensions
• Backpressure sensitivity - requires high pressure and thus, small
sizes
•Temperature sensitivity - it is desired that changes in fuel tempera-
ture have little effect on the coefficient of discharge. This factor
requires a high Reynolds number and thereby small orifice dimension.
30
-------
DRAIN TO TANK
MOVABLE PLATE
FLOW AREA SHADED
DJLET FROM FUE1
PRESSURE REGULATOR
SPRING
POWER LEVER
POSITION INPUT
HARDENED AND LAPPED
SHEAR PLATE SURFACE!
~0 PSIG
TO ROTATING CUP
FIGURE 10. FUEL METERING VALVE CONCEPT FOR 100:1 TURNDOWN
Figures 11 and 12 show the effect of orifice size on Reynolds number and
its consequent effect upon the discharge coefficient. These curves indicate the
dimension of the orifice should be approximately 0. 005 inches on a side at the one
pound per hour flow rate to minimize viscosity effects.
A pressure difference of 10 psi will produce the desired minimum flow in a
0.0057 inch square orifice and has been selected as the design point. Each slot varies
in width from 0.0057 to 0.057 inch according to a square root contour for flow linear-
ization with power lever position.
6.2 CONTROL SYSTEM COMPONENT TESTS
Each of the control component designs was evaluated in a series of tests.
Development of the units and final calibration was accomplished at a component level
prior to integration into the complete combustor system for low emission demon-
strations. Final fuel metering valve performance at 10 psid is 61 pounds per hour
per inch stroke with a linearity of ±2. 5 percent across a range of 3 to 115 pounds per
31
-------
0.020
0.018
0.016
3 °.°14
X
g 0.012
g °'010
oo
U 0.008
5 0.006
o
0.004
0.002
0
4000 8000 12.000 16.000 20.000 24,000 28,000
REYNOLDS NUMBER
FIGURE 11. REYNOLDS NUMBER AS A FUNCTION OF TEMPERATURE AND
ORIFICE SIZE AT 1 LB/HR FOR KEROSENE THROUGH A SQUARE
ORIFICE
0.66 r
0.64
H
u
0.62
8
u
Ct
X
U
J3
0
0.60
0.58
0.56
T • 130
T • 50* F
- O'F
0.004 0.008 0.012 0.016
ORIFICE SIZE, INCHES
0.020
FIGURE 12. DISCHARGE COEFFICIENT AS A FUNCTION OF
ORIFICE SIZE FOR KEROSENE
32
-------
hour. Good repeatability and flow control to as low as 0. 5 pounds per hour has
demonstrated the valve has a dynamic range greater than 200 to 1. Fuel pressure
regulator test results show a useful input actuator range from 1. 0 to 2.5 inches of
H2O while regulating fuel flow within ±4 percent. Air metering valve performance
calibrations have shown a problem with aerodynamic matching to the fan discharge
air. Swirl, dynamic head and turning losses required the incorporation of baffles
and flow straightners. Although these devices have given reasonably linear control
down to 2 percent of flow, they produced an undesirably high pressure loss. For the
final demonstration tests, aerodynamic matching between the fan and air valve was
simplified by the use of a plenum created by extending the ducting 36 inches. Although
this provided good control of air flow down to 1 percent, it also indicated the import-
ance of correct aerodynamic matching that would be necessary for a minimum volume
system. An optimum configuration of fan to air valve design has been established
from this experience and is discussed in Section 8.
6.2.1 Fuel Metering Valve Calibrations
In order to obtain linearity and reproducibility over the unprecedented 100
to 1 dynamic range requirements, a new approach to fuel metering was taken. To
minimize the effects of air valve to fuel valve linkage distortions (which directly
change air fuel ratio) due to actuation load, backlash, or thermal expansion, the
valve gain was made as low as practical by making the stroke long. Rigidity was
also emphasized for both its mounting flange and actuation rod. These features
can be seen in Figures 2 and 13. The rigid mounting flange with five bolt holes and
the large diameter actuator are shown in this photograph. Rigidity and freedom from
distortion are essential with a valve that must regulate from 1 percent to 100 percent
flow, since a linkage blacklash or distortion of 1 percent at the low flow end can
change the air fuel ratio by 100 percent. It requires a change of 0. 018 inch for a 1
percent flow variation. The air valve lever is connected directly to the fuel valve
actuator rod, permitting overall allowances in backlash to be kept below ±0. 002-inch
(±0. 001 at each pinned joint) or approximately ±10 percent flow at the 1 percent
valve position. Design analysis indicated that standard metering spool, needle, or
flapper configurations could not be expected to provide repeatable flow regulation as
a function of stroke from 109 down to one pound per hour.
Calibrations were made by a total weight versus elapsed time to obtain a
true weight flow measurement independent of viscosity and accuracy errors associated
with flow meters. A precision laboratory balance determined the mass of fuel metered
by the test valve with 10 psi pressure differential across the valve housing. Tests were
33
-------
FIGURE 13. FUEL METERING VALVE ASSEMBLY
conducted with all major components of the fuel system connected as on the combustor
system. Fuel was pumped by the system's electric driven gear pump to 12 psig.
Using shop air supply a constant 2.15 inches of H^O was maintained across the delta-
P regulator simulating the pressure drop across the air control valves metering
plates. Differential air pressure across the actuator controls the fuel pressure by-
pass valves position to maintain 10 psid across the fuel valve at all flows. A slight
increase in air pressure ( 0.15 inches of I^O above the nominal 2) was necessary to
maintain the fuel pressure at a constant 10 psid from 1 to 115 pounds per hour. All
fuel not being metered by the test valve was bypassed by the regulator back to the
fuel tank. From the discharge side of the valve the fuel passed either through the
flow meter or directly into a container on the scale.
Data obtained as a final calibration of me value is listed in Table I. At 21
different valve stroke positions the weight of fuel that flowed into the container was
determined to the nearest thousandth of a pound. A digital timer recorded the
elapsed time within 1 second with the lowest time span being 300 seconds. The
greatest source of error was associated with the valve position measurement method.
34
-------
TABL.K 1
FUEL METERING VALVE WEIGHT FLOW CALIBRATIONS
Test Point
1
1A
2
3
4
5
5A
6
7
8
9
10
10A
11
12
13A
13
14
15
15A
16A
16
Weight
(pounds)
0.267
0.222
1.032
1.695
2.450
2.634
1.426
3.832
6.853
10.688
7.182
8.882
10.518
11.132
6.114
6.934
6.980
7.775
8.556
8.600
9.458
9.488
Elapsed
Time
(sec)
1800
1800
1800
1500
1500
1200
600
1500
1200
1200
600
600
600
600
300
300
300
300
300
300
300
300
Calculated
Flow Rate
(pph)
0.53
0.44
2.06
4.06
5.88
7.90
8.55
9.20
20.56
32.06
43.09
53.29
61.30
66.79
72.10
83.10
83.80
93.30
102.80
103.2
113.6
113.6
Upstream
Pressure
(psig)
10.8
10.8
10.8
10.8
10.8
10.8
10.8
10.8
10.8
10.8
10.7
10.62
10.60
10.62
10.65
10.75
10.70
10.70
10.70
10.75
10.80
10.85
Downstream
Pressure
(psig)
0.80
0.80
0.80
0.79
0.80
0.62
0.80
0.80
0.80
0.80
0.72
0.68
0.62
0.62
0.62
0.75
0.75
0.70
0.70
0.75
0.80
0.85
Temp.
Fuel
oF
76
75
75
78
78
78
77
77
78
77
77
77
78
78
78
78
Valve
Stroke
(inch)
0.006
0.0065
0.031
0.063
0.0915
0.120
0.131
0.141
0.324
0.521
0.704
0.862
1.021
1.096
1.210
1.382
1.397
1.538
1.729
1.746
1.898
1.900
Date
12/15/70
12/11/70
12/15/70
12/15/70
12/15/70
12/15/70
12/11/70
12/15/70
12/15/70
12/15/70
12/15/70
12/15/70
12/11/70
12/15/70
12/15/70
12/11/70
12/15/70
12/15/70
12/15/70
12/11/70
12/11/70
12/11/70
A two inch stroke dial indicator was rigidly clamped to the valve body with its probe
zeroed against a rigid metal plate bolted to the input rod on the valve (see Fig. 14).
A simple screw jack arrangement locked the valve stem into each test position.
Accuracy of the dial indicator should be well within ±0. 0005 inch but a total hysteresis
of 0. 0015 inch was observed. Backlash within the valve actuator shaft and in the dial
indicator caused the hysteresis. This small amount of backlash has no significant
effect on valve performance until flow values of less than two pounds per hour are
being controlled.
Fuel backpressure was approximately 0. 8 psig throughout the test due to
the height (approximately 25 inches) that the fuel was required to be raised into the
measurement container. Inlet pressure was compensated to account for this head
effect. Additionally both the inlet and pressure gages were located at the same height as
the fuel valve to eliminate pressure head errors.
35
-------
FIGURE 14. FUEL METERING VALVE STROKE MEASUREMENT
ARRANGEMENT
Data from Table I is plotted in Figure 15 for a range of flows from 0.5 to
115 pounds per hour. A good degree of linearity is exhibited over this very wide
range. At high flow the deviation is approximately 2.5 percent below a true linear
operating line. An expanded scale plot of the data from 0.5 to 10 pounds per hour
(Fig. 16) shows an excellent degree of linearity across this low flow range. These
data points determine a linear operation with a slope of 64.2 pounds per hour per
inch of stroke. The overall slope across the entire operating range is 61 pph/in. as
indicated by the dashed line in Figure 16. By establishing 0. Oil inch stoke equal
to one pound per hour flow, we can define the valves performance as linear within
±2.5 percent from 3 to 115 pounds per hour with a gain of 61.0 pounds per hour per
inch. Below 3 pounds per hour the absolute deviation is a maximum of 0.2 pounds
per hour at the lower range limit (1 pph). Although small in absolute terms, it is a
large percentage of the relative flow. However, since it was repeatable, slight
modifications in the flow area of the air valve allowed correct air-fuel ratios to be
adjusted at low fuel rates.
6.2.2 Air Metering Valve Development Tests
The air metering valve required more extensive development testing than any
of the other control components. Figure 17 functionally describes the configuration
of the integrated air valve and fan used for development tests. Air is pumped axially
36
-------
110 -
CO
PRESSURE DIFFERENTIAL: 10 PS1D
FUEL: JP-S
TEMPERATURE. 76'F
ODATA RECORDED 12/11/70
• DATA RECORDED 12/15/70
0.1 U. 2
0.4 0.5
0.7 0.0 u.a 1.0
1.4 Li 1.6 I.T
FUEL VALVE POSITION (INCHES)
FIGURE 15. FUEL METERING VALVE FINAL WEIGHT FLOW CALIBRATION
-------
10.0 -
9.0 -
SLOPE • 61 PPH/IN.
PRESSURE DIFFERENTIAL: 10 PSIO
FUEL: JP-5
TEMPERATURE: 76'F
O DATA RECORDED 12/11/70
• DATA RECORDED 12/15/70
LINEARIZATION REFERENCE POINT (1.0 PPH AT 0.011 INCHES)
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
FUEL VALVE POSITION (INCHES)
FIGURE 16. FUEL METERING VALVE PERFORMANCE FROM 0.5
TO 10 POUNDS PER HOUR (FINAL WEIGHT FLOW
CALIBRATION)
across the fan motor case by a single stage four blade fan operating at 13,000 rpm.
Stationary vanes were incorporated immediately downstream of the fan blades to
recover the rotational energy in the fan discharge. A cavity formed by the fan on one
side and the air valve chamber on the other provides a plenum for air distribution.
As testing progressed, data indicated that special attention to uniform distribution
of static pressure and velocity gradients would be necessary to obtain uniform air
flow circumferentially around the combustor. Flow straightners and baffles were
incorporated at the component test level to reduce swirl and provide uniform static
and dynamic pressure fields across each of the 12 symetrical parts.
38
-------
rANHOTM
coourarM
•ICT
FIGURE 17. Am VALVE AND FAN FLOW TEST MOCK-UP SCHEMATIC
A variable geometry sheet metal mock-up of the air flow configuration (Fig.
18) of the valve was tested to establish performance characteristics. The sheet
metal test valve had all of the functional flow features that were incorporated when the
final design unit was fabricated. It was fabricated of light gage sheet metal to facilitate
rapid rework for the comparison of different port, baffle, and instrumentation config-
urations. From the photograph, it can be seen that the valve is in approximately the
10 percent flow metering position since the by-pass ports are approximately 90 percent
open. Metering is accomplished by the plates immediately downstream of the by-pass
ports. The 12-inch diameter section in the center represents the combustor interface
with the air valve. Figure 19 is the schematic arrangement of the flow test rig.
Actual test component arrangement and instrumentation is shown in Fig. 20. Air
is pumped by the fan into the air valve cavity. Because of wide variations in the
dynamic .head, three static pressure probes located on the face of the air metering
valve proved to give the most consistent results. The average of these three mano-
meter readings was reported at P^. In a similar manner, P2 readings were picked
up by three flush mounted static probes located in the wall of the 12-inch diameter
cavity immediately on the discharge side of the air metering orifice ports. P3 and
PC measurements were made by a total of 5 additional static probes (see Fig. 19),
designed to determine flow pressure loss in flow duct and across the calibrated
metering orifice plate. The metering orifice plate is designed to have 60 (1/2-inch
diameter) holes. This arrangment allows rapid matching of the calibration orifice
pressure drops to the characteristic pressure drop of the combustor. Initial design
point at 100 per cent flow requires the back pressure on the metering valve to be 8
inches of H2O (design combustor pressure drop). The area of the calibration orifice
39
-------
FIGURE 18. AIR VALVE FLOW TEST MOCK-UP
uun
I FOOT KCTMHDM PflB
••on WLoerrr C
FIGURE 19. AIR VALVE FLOW TEST SCHEMATIC
40
-------
FIGURE 20. Am FLOW TEST BENCH
plate was adjusted by covering ports in conjunction with cross calibrations against a
venturi meter until the desired 8 inches of H2O was obtained at a flow of 47 pounds per
minute.
Procedures for all tests were similar. The valve would be positioned to the
100 percent power lever position and the fan motor operated at a voltage necessary to
obtain 10 inches static pressure at P,. The power lever position would be then adjusted
to lower power setting to determine the flow characteristics from 100 to 10 percent
flow. At each lever position setting, the pressure distribution and me flow in the test
rig were measured to determine the characteristic performance of me valve. When
indicated by test results, design changes were incorporated to adjust the performance
toward the established goals. Results from concurrent emission monitoring tests
being performed on the combustor rig revised air fuel ratios during these tests. These
tests indicated that the valve should be designed to produce a lean condition at lower
power settings rather than the rich air-fuel ratios incorporated in the original port
configuration. As a consequence, the nonlinear cutout in the orifice ports were re-
moved and the valve was designed to produce a constant air-fuel ratio of 25.5 across
the range from 100 to 10 percent.
A rich condition (small flow areas) was intentionally maintained to facilitate
optimization of the valve air-fuel ratio during combined combustor air yalve emission
tests. Leaning of the ratio required only metal removal from the orifice plates by
means of simple hand file operations in the range of 10 to 100 percent. Below 10 per-
cent, the design allows the backup plate to be rotated with respect to the housing to
provide a wide adjustment of air-fuel at the low heat release rates.
41
-------
Initial tests were performed with a single conical baffle on the outlet of the
fan. Correct flow, pressure levels, and pressure distributions were obtained down to
a 50 percent power lever position with electrical input power at 2. 02 HP (29.2 volts
and 51. 5 amps). Unsatisfactory results were obtained below the 50 percent flow position.
Investigation isolated the problem to swirl resulting from the air valve-fan combination.
Static pressure readings in the system were severely affected and biased toward the high
side by the swirl conditions. A venturi meter initially used for flow measurement was
particularly sensitivie to swirl induced errors. Since swirl of any significant magni-
tude can cause serious maldistribution of air in the combustor (in addition to the instru-
mentation problem), a series of configuration modifications were made and tested to
reduce this aerodynamic problem. The edge configuration of the rotating and stationary
plates were found to be a prime factor in the swirl condition as was the apparent high
and unstable dynamic pressures in the air valve cavity. Metering edges were made as
sharp as possible and a number of baffle configurations were tested. Although not
optimum, the present configuration has appeared to reduce swirl to acceptable levels.
Baffle arrangements are shown in Figure 17. Three radial baffles were found to be
necessary to reduce the high dynamic head (4 inches of HgO pressure) to allow linear
valve control. Although these baffles permitted linear operation of the air valve, the
pressure difference across the metering ports was not a constant 2 inches of HgO as is
required for accurate operation of the compensation system. Swirl was reduced by
use of 12 axial baffles that compartmented the air valve cavity into 12 separate flow
passages supplying air to each of the variable orifice ports. Although the valve func-
tioned well in flow bench tests with this configuration, an additional pressure loss
(approximately 3 inches of HgO) has been required. As a consequence, the fan must be
matched to the system by operating at 33 volts (14,200 rpm and 2. 3 HP).
Flow Metering Test Results
Component level test results with a calibrated orifice are listed in Table n.
Power lever position has been used as the main parameter for these tests and has been
established by dividing the total stroke between 10 and 100 percent into 9 equal divisions.
An 8-foot long, 8.5-inch diameter, extension to the calibration orifice duct was added
to obtain an accurate flow calibration below 10 percent by means of a smoke velocity
calibration. A puff of smoke was introduced at a port 12 feet from the end of the duct.
From the elapsed time it takes to traverse the duct length (37 seconds for 1%), accurate
flow measurements were obtained. This was necessary since at 10 percent the back
pressure on the air valve must be maintained at 8(0.1)2 or 0. 08 inch of H^O. At the 1
percent flow, the back pressure must be maintained at 0.0008 inch of H^O. Thus, the
smoke velocity method appears to be the best approach for accurate flow calibrations
down to the 1 percent level. Below the 10 percent position of the power lever, smoke
velocity calibrations were made to determine the 7, 5, and 2 percent flows. These
positions were recorded and compared with the 10 to 100 percent valve gain on a linear
basis to determine accuracy.
42
-------
TABLE II
AIR CONTROL VALVE TEST RESULTS WITH 2. 25 INCH WIDE BY-PASS
VALVE (0.5-INCH BAND ON ORIFICE SIDE OF BY-PASS)
Power Lever Position
PI (In. of H90)
P0 (In. of H00)
- -
P - P., (In. of H^O)
L/P1 ~P2
V 2 " 1
L
P.J (In. of H.,0)
p (In. of I100)
'' (Main Man.)
p (In. of H00)
'' (6° Man.)
Measured Flow
/ Posit ion \
\J Flow )l°
1007r *
'•
,_ Deviation From
I)" , .
Linear
Overall Air Fuel Ratio, dP •
Metering Accuracy
100
9.60
8.27
1.33
-18.0%
7.91
8.02
100.0
0
30.0
90
8.37
6.75
1.62
-10.0
6.66
6.73
91.5
•1.8
28.4
80
6.85
5.25
1.60
-11.0
5.08
5.11
79.5
-0.5
28.2
70
5.70
4.02
1.G8
-8.0
3.84
3.85
69.2
-1.1
27.3
60
4.70
2.97
1.73
-7.0
2.78
2.77
58.6
-2.4
26.7
50
3.92
2.09
1.83
-4.0
1.90
1.80
1.76
46.8
-6.5
24.8
40
3.43
1.45
1.9S
-5.0
1.24
1.14
1 . 145
37 . 8
-5.8
24.2
30
3. 1.3
.90
2.23
.5.0
0.74
0.7
O.B7
28.8
-4.2
23.3
20
2.95
. 52
2.43
•11.0
0.40
0. .15 •
0.30
19.3
-.1.5
22 3
* dP Compensation
Effect on Fuel
A plot of Table II results and smoke velocity calibrations is presented in
Figure 21. The 100 percent position was made to be correct by adjusting motor
voltage and valve by-pass port configurations to simulate combustor pressure drop
at P2. It can be seen that the valve has a high degree of linearity across a wide flow
range. Cross plotted with valve position is the main parameter which is to be con-
trolled, air fuel ratio as affected by the air metering valve performance. From 100
percent flow down to 6.5 percent, the air-fuel ratio remains within the initial design
established limits of ±10 percent. Below this flow setting, the air flow is below the
limits, indicating larger flow areas are necessary. This final flow adjustment has
been accomplished in combined valve and combustor emission tests. A comparison
of preliminary emission test results shows that to meet the optimum air-fuel ratio in
the low power setting, the valve required an increased flow area to lean it down to
34:1 air fuel at 1 percent. Since the total area open at 1 percent is 0.2 inch2, increas-
ing flow area at low power settings was accomplished by rotation of the backup plate by
approximately 0. 010 inch. A pressure drop across metering valve ports proportional
to flow is important for proper functioning of the fuel pressure compensation system.
If fan voltage inadvertently changes, or if it is intentionally reduced at part loads to
43
-------
OOMI ONET AIR V*L VI
PERF< RMAf CE
BE ADJXJi TED 1 4 CO* BINEI
/AI.V* CQ1T lUSTQ J EMI SON
CALC JLATBD
ACTUIL
TljBT WTjH 2.2J WIDE
BY-PA88 VALVE
(BAND Oil ORfftCE
4 L. ,
CALIF RATIO N8
10 20 30 40 SO 60 70 80 90 100
POWER LEVEL POSITION (PLP) %
FIGURE 21. FLOW CONTROL PERFORMANCE OF AIR METERING VALVE
lower power losses, the fuel pressure regulator is designed to automatically reduce
flow proportional to air flow. To achieve accurate air-fuel ratio control, it is necessary
that a pressure difference across the metering ports be maintained at 2 inches of H9O.
Figure 22 shows flie pressure differential as a function of valve position. At high
flows, the PI - P2 is less than calculated and at low flows, it is approximately the same
amount (~0.5 inch of H2O) above the 2 inches of H2O design point. Assuming perfect
fuel pressure regulator and fuel metering valve, the effect on fuel flow and air-fuel
ratio are also plotted in Figure 22. It is seen that the air-fuel ratio exceeds the 10
44
-------
hi
30
28
26
24
20
18
B' '-PAS VAL'
SID
or
LEA*
'10%
-10%
RICH
g
u *•
|S •«
If'
J i
| oT-io
*
-9n
N
X
^-,
'^s
' H
^~^
\
1
•1
LI
i
i
k
»
IH
r
e.
i
H
P.
10
71
u
X
H
X
/
10 20 30 40 SO 60 70 80 90 100 %
POWER LEVER POSITION. %
FIGURE 22. AIR VALVE AP TEST RESULTS WITH MOCK-UP AIR VALVE
45
-------
percent lean band at high flows and the 10 per cent rich band at low flows. It appears
that the cause of this large difference between calculated and actual pressure differences
at the metering port is associated with the high dynamic head in the small diameter fan
discharge annulus. High velocity jets must change their direction as a function of geo-
metry and the amount of flow required. Several approaches to the problem exist. The
optimum method would be to specially design a fan to match the large diameter geometry
of the combustor and air valve and thereby have lower speeds and dynamic turning
losses. Schedule and scope limitations of the program have not allowed the design and
incorporation of an optimum fan. An extension duct to allow a larger plenum has been
used in most of the final emission demonstration tests as a simulation of the aero-
dynamics expected in a matched fan-combustor system.
An even greater effect upon the pressure distribution across the metering
port can be obtained by baffle and flow straightener configuration changes. Another
significant factor is the location of the static pressure probes for P^. An apparent
pressure gradient across the face of the valve plate has been observed and found to
vary with power lever position. By optimizing by-pass valve geometry, baffle
configurations and static probe location, significant reductions in variations of P, - P2
with power lever setting were obtained. A major portion of the valve development
tests were directed toward this, since a stable P^ - P£ not only can be used to compen-
sate for operating variations in fan voltage but can form the basis for major flow
changes accompanying voltage reductions to save on parasitic losses.
From these test results the final configuration of the valve was fabricated.
Figures 2 and 23 show the final design prior to installation into the demonstration
system. Mechanical functioning is very simple as the air metering port moveable parts
are mechanically fastened to the by-passport fingers. Air pressure and springs force
the sliding valve elements against the valve housing and backup plate. Capability to
change the low flow rate air-fuel ratio is provided by allowing the backup plate to be
rotationally indexed with respect to the power lever position and by an adjustable
linkage between the power lever and the fuel valve (See Fig. 2 and 23). Adjustments
to the final air-fuel ratio were made in integrated systems tests to minimize emissions
from the combustor (See Section 7).
Fan Noise
Due to the high dynamic turning and straightening losses, the fan has been
operated at 33 volts to match it to the system requirements. As a consequence, the
noise level was above the 80 db at fan rated conditions. An analytical and experimental
investigation of the problem and its potential solutions has been initiated.
Sound pressure levels of the Rankine burner fan section operating separately
from combustor were taken with and without inlet guide vanes. The results were 102 db
46
-------
FIGURE 23. DEMONSTRATION SYSTEM AIR VALVE
(ref. 20ptN/m^) overall sound pressure level with inlet guide vanes and 94 db overall
sound pressure with the inlet guide vanes removed. A 1/10 octave frequency analysis
of noise with guide vanes removed, is given in Figure 24. Test details are given
below:
• Blower Power Setting: 31 Volts, 53 amps
• Blower Speed: 14,000 rpm
• Number of Blades: 4
• Microphone Location and Orientation: Five feet off blower axis from
blade location. Set 45 degrees from vertical toward blower and perpen-
dicular to blower axis (see Fig. 25).
•Surroundings: Semi-reverberant (see Fig. 25 for details)
• Equipment General Radio Type 1564A Sound and Vibration Analyzer
General Radio Type 1521 B Graphic Level Recorder
General Radio Type 1560-P4 Preamplifier
General Radio Type 1560-2131 Microphone
General Radio Type 1562 Sound-Level Calibrator
•17
-------
00
M MM
FREQUENCY (Hi)
FIGURE 24. SOUND PRESSURE LEVEL VERSUS FREQUENCY FOR BLOWER
WITHOUT INLET FLOW STRAIGHTENER
-------
POURED CONCRETE
K
U
ff\
2
&.-»
SMOOTH
i
J
WOOD CEILING
CONCRETE FLOOR
- f
5'
t
(X
U
H
OT
3
*A
SMOOTH
x> 45°
-j-*7!
^c
WOOD PANELING
FIGURE 25. NOISE MEASUREMENT ROOM GEOMETRY AND MATERIALS
•Background Noise: Background noise was observed to be more than
10 db below peak levels occurring at 250 cps, 980 cps, 2000 cps,
and 3000 cps.
•Background levels are at least 5 db below blower levels at all other
frequencies except in the ranges 25-220 cps and 350-400 cps where
blower and background levels are equivalent.
An approximation of overall noise level of the blower in a free acoustic field
can be made using the following relation:
a2
Reduction in Noise Level (db) = 10 log „ —
10 a
where
™ is the new sound absorption coefficient which is one for this case by
definition of a free field
a is the room sound-absorption coefficient determined by relative room
wall areas multiplied by their respective material absorption coefficients.
If S is the total room wall surface area, then for a predominant level of 980 cps
2S
a = — (Absorption coefficient of concrete)
1 «3
2S
+ — (Absorption coefficient of wood paneling)
5
49
-------
s
+ — (Absorption coefficient of smooth plaster)
= s|- (0.0175) + - (0.06) + i (0.035)
5 55
= 0.034S
Then Reduction in Noise Level is = 10 Iog10 1. 00/0. 034 = 14.7 db.
This implies that the blower operating in a free field without inlet guide vanes
would register approximately 80 db overall sound pressure level assuming the same
relative microphone position.
Figure 24 shows four distinct frequency peaks of
1. 92 db at 980 cps (this was also noted as the predominant noise level
frequency with vanes.)
2. 90 db at 2000 cps
3. 83 db at 2900 cps
4. 81 db at 245 cps
Blade passage frequency is
14,000 Rev/Min Cycles
60 Sec/Min 4 isT =
The first three measured peaks have frequencies corresponding to the first,
second and third order of blade passage frequencies, respectively. A fourth peak is
coincident with the blade support rotational frequency. It can be deduced that bearing
noise and or single blade passage frequency is the cause of this peak. The exact cause
is not of immediate interest because it is considerably lower than the fundamental and
first harmonic peaks caused by blade passage excitations. Appendix A discusses
several design approaches that could be incorporated into an optimum fan design that
can reduce blade passing frequency tones in an optimum configuration.
50
-------
6.2. 3 Delta-P Fuel Pressure Regulator Calibrations
In order to maintain the desired air-fuel ratio independent of fan voltage, fan
or motor efficiency reductions, or blade fouling a delta-P regulator was developed to
control fuel pressure directly proportional to the pressure differential across the air
metering valve. Figure 1 functionally describes the operation of the valve. In order to
obtain as low an operating range as practical a large diameter actuator was used (Fig. 26) to
convert the pressure drop (P^ - P2) across the air metering valve into an effective
control force. Development tests on the regulator valve were performed with the fuel
metering valve as the flow load component. Supply pressure was provided by the
Rankine combustor's system fuel pump. Air for the actuator input was regulated from
the shop air system as the main test parameter variable. At a differential actuator
input of 1.0, 2.0 and 2.5 inches of H2O differential, the metering valve was placed In
the 100, 50 and 1 percent flow positions. The fuel bypass adjustment was then set to
give the best regulation across the range of fuel flows and input air pressures. Through-
out the remainder of the test it was locked in position while both air pressure and fuel
flow were varied. Regulated fuel pressure was recorded at each of the three flow
conditions at every 0.1 inch from 1.0 to 2. 5 inches of HgO input pressure.
Several different configurations of springs and diaphragms were tested. A
PVC molded diaphragm and a thin flat rubber diaphragm proved to be the best. A
plot (Fig. 27) of the regulated pressure versus actuator input signal for the PVC
^
FIGURE 26. FUEL PRESSURE REGULATOR INSTALLED
ON P - P ACTUATOR
51
-------
01
to
• 100% FLOW
A 50% FLOW
. 1% FLOW
1.0
1.2
1.3
2.2
2.3
2.4
1.5 1.6 1.7 1.8 1.9 2.0 2.1
CONTROL ACTUATOR INPUT PRESSURE DIFFERENTIAL (INCHES OF WATER)
FIGURE 27. FUEL PRESSURE REGULATOR WITH MOLDED 0.012 INCH THICK PVC DIAPHRAGM
2. 5
-------
diaphragm shows that it has reasonably good performance across the extended oper-
ating range. The dashed line plotted in the figure represents exact (zero error) com-
pensation. An upper and lower 5 percent error in fuel flow is also included. Only
at the low end of the input pressure range does the regulated pressure output cause a
flow error of more than 5 percent (7.5% rich). This diaphragm configuration used
0.012 PVC with a molded convolution between the outside ring seal and the center
plate support. It is a rugged diaphragm but has corresponding high stiffness.
Irregularities in the convolution and the high stiffness caused the nonlinearities in the
performance curve.
A flat diaphragm made of thin resiliant rubber (0. 008 inch thick) had virtually
zero stiffness. Results of test with this material showed that it regulated pressure
accurately enough to hold the flow of fuel within ±4 percent of the ideal across the entire
range of input pressures (Fig. 28). Final calibration tests results are given in
Table III. Normal operation of the system would be in the range of 1. 0 to 2. 0 inches
(if voltage to the motor is reduced to reduce part load parasitic losses). Within this
range the valve regulates pressure accurately enough to hold the flow within ±3 per-
cent of the ideal.
6. 3 TRANSIENT RESPONSE EMISSIONS TESTS
6.3.1 Description of Demonstration Combustor System
A fully integrated combustion system was tested for emissions. Figure 29
shows a cross section of the complete system. Air is drawn through the fan and
delivered into the air valve chamber. From this chamber the air either is metered
into the combustor section through the air metering ports or it is bypassed through
the ports located on the outer diameter of the valve housing. Figure 29 is only one
of several configurations tested in the fully integrated system emission tests. Most
of the tests were performed with a 36-inch long extension between the air valve and
fan to provide a better aerodynamic matching between these components.
Figure 30 illustrates the system with the fan mounted inside of the extension.
By mounting the fan within the duct it was possible to vary the axial distance between
the fan discharge and air valve. Instrumentation, control and a self containpd fuel
system including tanks were mounted in a mobile test stand for maximum test
flexibility (Fig. 30 and 31). All steady state power levels and transient emission
tests were performed with the use of the power lever position as the only variable.
Regulation of air and fuel flows was automatically maintained by the previously
described components integrated with the combustor. Fuel flow rates during a
particular test run were determined by prior calibrations of the fuel metering valve
and measurement of the pressure differential across the metering ports. Two variable
area flow meters were connected through manual valves in series with the flow meter-
ing valve. These were used to verify the calibration of the flow metering valve
53
-------
• 100% FLOW
A 504 FLOW
• 1% FLOW
1.0 1.1
1.4 1.5 1.6 1.7 1.8 l.B 2.0 2.1
CONTROL ACTUATOR INPUT PRESSURE DIFFERENTIAL (INCHES OF WATER)
2.3
2. 5
FIGURE 28. FUEL PRESSURE REGULATOR PERFORMANCE WITH 0.008 INCH FLAT RUBBER DIAPHRAGM
-------
TABLE III
PRESSURE REGULATOR CALIBRATION WITH
0.008 INCH FLAT RUBBER DIAPHRAGM
Test
Points
1A
2A
3 A
4A
5A
fiA
TA
-A
!iA
HIA
11A
12A
13A
14A
ISA
IfiA
50*
IB
2B
3B
40
.">B
6B
7B
SB
9B
10B
11B
12B
13B
14B
15B
1KB
1C
2C
3C
4C
5C
6C
7C
8C
9C
IOC
11C
12C
13C
14C
15C
16C
Flow Motor
Readings (pph)
(Uncorrected)
.0
.0
.05
. 1
. 1
.2
O.'J
0.8
0.7
0.7
0.7
0.7
0.7
O.fi
0.6
0.6
41. 1
nil n
.SI. 5
62.5
53. 5
:"). 0
48.0
47.0
45.5
44.0
43.0
41.8
40. 10
39.0
37.8
36.0
96.0
98.0
99.5
100.2
104.0
106.2
93.5
91.0
88.0
85.5
83.2
81.0
78.0
76.0
73.5
70.2
Regulated
Fuel Valve
Inlet Pressure
(psig)
9.9
10.5
10,8
11.4
11.9
12.5
3.6
y.os
8. 5fi
8.0
7.S
7.25
6.70
6.30
5.85
5.40
9.!l
10.30
10.90
11.30
11.70
12.20
9.55
9. 10
8.50
8. 15
7.70
7.20
6.75
6.30
5.90
5.35
9.60
9.90
10.35
10.80
11.20
11.60
9.20
8.70
8.20
7.75
7.30
7.00
6.45
6. 15
5. 70
5.25
Pump
Discharge
Press
(psig)
10.8
11.3
11.8
12.2
12.8
13.3
10.4
10.0
9.4
9.0
S.8
8.1
7.5
7. 1
6.8
6.2
10.8
11. 1
11.8
12.2
12.6
13.1
10.4
10.0
9.4
9.0
8.6
8.1
7.7
7. 1
6.8
6.3
10.5
10.8
11.2
11.6
12.1
12.5
10.0
9.6
9.2
8.8
8.2
7.9
7.5
7.0
6.5
6.2
Actuator
Sensor Input
(in. of HjO)
2.0
2. 1
2.2
2.3
2.4
2.5
1.9
1.8
1.7
1.6
1.5
1.4
1.3
1.2
1. 1
1.0
2.0
2. 1
2.2
2.3
2.4
2.5
.9
.8
.7
.6
.5
.4
.3
.2
.1
.0
0.95
1.00
1.05
1.10
1. 15
1.20
0.92
0.88
0.80
0.75
0.70
0.65
0.60
0.55
0.50
0.45
Regulator
Back Pressure
(psig)
0.32
0.35
0.35
0.37
0.38
0.38
0.38
0.38
0.38
0.38
0.38
0.38
0.38
0.38
0.38
0.38
0.05
0. 10
0. 10
0.12
1.4
1.6
0.08
0.05
0.02
0.01
0
0
0
0
0
0
2.0
2.1
2.2
2.3
2.4
2.5
.9
.8
.7
.6
.5
1.4
1.3
1.2
1.1
1.0
Valve
St roke
(Inches)
0.006
0.866
0.8915
55
-------
01
r i r
FIGURE 29. DEMONSTRATION SYSTEM
-------
FIGURE 30. DEMONSTRATION SYSTEM WITH LONG MIXING DUCT
FIGURE 31. INTEGRATED SYSTEM DEMONSTRATION TEST STAND
ARRANGEMENT
57
-------
periodically throughout the test phase. They could not be incorporated in the system
continuously since they have a flow sensitive pressure restriction that would modify
the fuel pressure regulator's control of air-fuel ratio. Two flow meters are necessary
since a 10 to 1 flow ratio is the normal range of these instruments. Fuel delta-P
across the metering valve was recorded on both a precision laboratory gauge with
divisions of 0.1 psi and an electric transducer displayed on a strip recorder during
transient tests. An electric position transducer connected directly to the power lever
position recorded stroke or power level directly in percent of full power on a strip
recorder. Its basic function was to establish power rate input changes during trans-
ients. Air-fuel ratio was obtained by an averaging probe CC>2 measurements in the
exhaust and cross checked against combustor air flow calibrations made with standard
orifice flow meters.
6. 3. 2 Startup and Shutdown Transients
The design goal for this demonstration was to be able to start the combustor
and to be up to full power in three seconds. Experimental analysis indicated that
startup was optimum at the 30 percent power lever position. Figure 32 shows the
results of a startup from 30 percent and followed by an immediate power increase to
100 percent in three seconds. In this start the blower was at operating speed prior to
ignition excitation and opening of the fuel valve solenoid. Fuel was circulating through
the regulator valve and being bypassed to the tank at the time the solenoid valve opens
to connect fuel to the rotating cup feed tube. Transient gas emissions were obtained by
recording the output of each of the four gas analyzer systems on a strip chart. Response
delays of each of the four measurement systems is from 2 to 5 seconds with the relatively
short exhaust gas sampling lines used in this test (less than 20 feet). It is not possible
to directly plot the emissions in gm per Kgm of fuel since the fuel flow and air flow
cannot be exactly established during a transient. Additionally the transient response
of present state-of-the-art gas analyzers is not sufficiently fast to provide for exact
analysis of a transient load change. However, a reasonably good quantative estimate
of transient performance can be made by an analysis of the direct volumetric emission
data obtained as a function of time on the Beckman strip chart recorder. Design goal
limits are indicated on the exhaust gas concentration scales. These limits are based
on calculations that relate the mass emission goals to the volumetric flowrate and the
air-fuel ratio as obtained from the CC>2 concentrations. At the startup transient none
of the emission levels exceeded the volumetric design goals. Hydrocarbon (measured
as C,) came relatively close to the limit, 20 ppm compared to a limit of 35 ppm.
After one minute of operation the system was subjected to decrease power transient
from 100 percent to 5 percent in two seconds and then an increase to 100 percent in
2. 7 seconds. After another twenty seconds at 100 percent power the combustor was
shutdown by closing the systems fuel cutoff solenoid valve. A transient peak of 700
ppm in HC emissions was recorded. Cause of this high peak appears to be a small
amount of fuel leaking into the combustor after the initial closing of the solenoid.
58
-------
1000-
900-
800-
700-
E 600-
o.
Q.
¥• 500-
400-
300-
200-
100-
0-
16-
14-
12-
10-
o
(J Q_
6-
4-
2-
0-
1000
900
800
700
O. '
ex
Ssoo-
400-
300J
200-
100-
0-
150-
135
120
105-
E 90-
OL
O.
O
Z75-
60-
45-
30>
15-
0-
•HC
40
60 80 100
TIME (SECONDS)
120
140
160 ISO
FIGURE 32. STARTUP AND SHUTDOWN EMISSIONS
In the demonstration configuration the solenoid valve was located approximately 12
inches from the rotating cup. In a specially designed installation the shutoff valve could
be located directly in the fuel supply tube to eliminate any possibility of after drip.
An alternative would be to utilize a solenoid valve with sufficient "negative" displace-
ment to extract a small fuel volume from the fuel supply tube as the valve closes.
6. 3. 3 Power Level Transient Emissions
A major performance goal of the program was to be able to rapidly modulate
power from any one power level to another without severe increases in emissions. A
response rate of 50 percent of full power per second was established as a preliminary
goal. Table TV lists emission level transients in ppm obtained from strip chart
records for power changes. For each transient test the demonstration system's
power lever was manually positioned from a high heat release (A) down to a lower
heat release (B). At level (B) the fuel flow was stabilized with a dwell period of
approximately 0. 5 second. Immediately after the dwell the lever was moved back to
the initial heat release position of (A). Movement of the lever was controlled in each
test to obtain a rate of change of greater than 50 percent per second. Fuel flow
corresponding to the lower lever position is listed in the first two columns of Table IV.
59
-------
O5
o
TABLE IV
POWER LEVEL TRANSIENT EMISSIONS (COMBUSTOR CONFIGURATION "D")
Transient
From A and B
to A
A(pph) B(pph)
109 5
109 10
109 30
109 50
109 70
50 5
50 20
80 30
Transient Time
(sec)
4 12 13
1.70 0.6 1.85
1.40 O.G 1.65
1.70 0.4 0.9
0.7 0.7 0.6
0.3G 0.46 0.30
0.65 0.7 0.75
0.26 0.58 0.23
0.79 0.58 0.6
Emissions Measured
CO (ppm)
Limit
534
534
534
591
534
591
591
591
Steady
State
45
45
40
15
45
15
15
15
Peak
8
285
40
223
45
940
30
60
NO (ppm)
Limit
28
28
28
31
28
31
31
31
Steady
State
42
4G
49
43
47
46
46
52
Peak
45
48
50
39
47
46
46
52
HC (ppm)
Limit
32
32
32
35
32
35
35
35
Steady
State
14
13
15
13.5
15
12.5
18
16
Peak
14
15
15
25
15
23
18
16
C02 (7o)
Steady
State
G.5
6.5
6.65
7.2
6.5
7.2
7.2
7.2
Peak
8.3
9.7
8.05
G.3
6. 65
9.5
7.7
8.05
* Combustor Configuration "D"
FUEL FLOW TRANSIENT TEST SEQUENCE
(MANUAL OPERATION)
STEADY STATE A
DWELL
EMISSION TRANSIENT
CHARACTERISTIC
PEAK VALUE
STATE
STATE
3-10
sec
:*i
-------
Transient time periods between lever positions is tabulated in the next columns. A
comparison between the steady state and peak value of emissions is used to determine
the transient performance characteristic of the system. From the results it can be
seen that only CO has a characteristic transient peak significantly higher than the
steady state emission levels. In two transient tests the HC also exhibited a significant
peaking tendency.
NO emission levels showed essentially no deviations from steady state levels
during these transient tests. Combustion configuration "D" (see Section 7) was used
for transient tests and thus the steady state NO levels were higher than the design
goals. However, the basic measured response characteristic of the NO to power
level transients is not expected to vary significantly with the lower NO configuration
"A". The configuration differences were that "D" had 25 percent more air in the
primary zone than "A". "D" also had less heat loss from the flame due to the use of
radiation shielding. An analysis of the HC and CO peaks against the design goal
limits indicated that only in one test was the limit exceeded. A detailed analysis
of this test (50 to 5 pounds per hour) was made by comparing the results strip chart
records of the time dependent parameters. Figure 33 compares the high peak
emission results with a typical low peak emission tests. In the top graph a high CO
peak is recorded well above the limit (indicated by an arrowhead on the scale). A
peak in the HC level is relatively high but well within the limit. For comparison a
typical response characteristic (Fig. 33B) indicates virtually no HC peak and a CO
peak well below its limit. A detail analysis of the lever position versus time (dotted
lines) indicates that one major difference between the two tests is rate of change of
power. From this data, and other similar traces it appears that the emission peaks
on CO and HC increase significantly as the rate of change in lever position (fuel flow)
increases. It also appears that the sensitivity to rate change is only during power
decrease transients below 30 pounds per hour fuel flow. It appears transient at rates
several times the goal of 50 percent per second can be achieved without significant
emission peaking at all conditions except decreasing from 30 pounds per hour. The
rate of decrease that appears to have caused the high CO peaking was 140 percent
per second.
Although well below limits even a rate of 68 percent per second causes
significant peaking in CO. It should be noted the HC emissions have virtually no peaks
at the same 68 percent per second rate.
In summary, it can be stated that the system has good transient emission
characteristics and can meet the emission levels at the transient rates established
as goals for interfacing with the vapor generator. However, since some vapor
generators may require very high combustor system response (depending upon margin
of safety between fluid operating and decomposition temperatures) additional consid-
eration should be given to improvement of the air-fuel ratio control during transients
at the low power levels.
61
-------
150
133
118
102
t 71
42
27
14
0
1000
865
740
620
510
a-410- >Q50
0320
235
150
75
0
DEMISSION COAL LMT (IN PPM
AT TEST CONDITION A/F RATIO)
0L
2 34 56 7 8 9 10 11 12 13 14 18 16 17
TIME FROM START OF TRANSIENT (SEC)
A. HIGH CO AND HC PEAK RESPONSE (POWER LEVER RATE
APPROXIMATELY 150%/SEC)
150
133
118
1O2
87
I*
S56
42
27
14
0
C 1000
865
740
• 620
_ 510
0. 410
O
235
* 150
75
0
1UU
90
^ 80
UJ
r
z60
^
*50
S?
a 40
UJ
UJ
IJ3O
ct
10
: I POWER
: ".
I i t :— t!B%«
l"1 '"" -; • -INDICATES SLOPE
; ; PERCENT / secoo
': - : ^EMISSION GOAL LIMIT (IN PPM
. ; .: X^ ^x Al TEST CONDITION A/F RATIO)
'-: : / ^^^
\ _^f ^^^^_ CO2
1 ,^57*;^
\ / "NO
: ' / ^^
Ueev.* • / \
'•• • / \ — -
f- \. HC
\j y ^-^_^L_
1 2 34 56 7 8 9 10 11 12 13 14 18 16 17 1t
TIME FROM START OF TRANSIENT (SEC)
1O.U 1
13.5 •
11.2 •
90
fe
72^.
UJ
55*
tf
40 •
O
O
2B •
4 Q
1.8
8 -
i
i
DU
45
40
35
305
i
25"
I
20"
15
5
B. TYPICAL CHARACTERISTICS (POWER LEVER RATE
APPROXIMATELY 50%/SEC)
FIGURE 33. POWER LEVEL TRANSIENT EMISSIONS (COMBUSTOR
CONFIGURATION "D", SEE FIGURE 81)
62
-------
7
COMBUSTOR DISCUSSION
7.1 REACTION KINETIC STUDY BY COMPUTER MODELLING
7.1.1 Summary
The computer model showed that, given adequate combustion volume, CO
and NO emissions can be kept at a low level at high fuel rates. At low fuel flows NO
emissions would be unacceptable but could be reduced by introducing heat losses to
the vaporizer from the flame.
7.1.2 Emissions Analysis
This section describes the methods used to calculate the emissions from the
combustor.
The method used are incorporated into three computer programs:
1. Steady-State Combustion Program (SCP)
2. Chemical Equilibrium Program (ODE)
3. Generalized Kinetics Program (GKP) - Dynamic Science Report
No. TR-C70-227-I.
The first calculates the liquid phase heat release profile in terms of the
vaporization rate of the liquid droplets. Liquid properties and injection parameters
are required input, and the output consists of the vaporization rate of the liquid spray
in the combustion field.
The Chemical Equilibrium Program (ODE) is used to calculate initial equilib-
rium concentrations in species in the primary zone to initiate the kinetics calculations.
The Generalized Kinetics Program is used for calculation of emissions, and
solves the one-dimensional nonequilibrium reacting gas flow equations for a pressure
or area defined stream tube for any input defined gaseous chemical reactions. Reaction
rates are input as an Arrhenius reaction expression and all competing reaction for a
63
-------
particular species are included. The ability of the program to consider arbitrary mass,
momentum, and energy addition, coupled with the steady state combustion program,
gives GKP the added capability of handling two phase flow problems (liquid droplet
combustion).
The chemical species and the kinetic reaction system are defined by the input
of the symbolic species names (N, NO, NO2, O, etc.) and the reaction set in symbolic
form (NO+ M = N+ O + M), where M is an arbitrary third body (all other species).
Program output consists of the fluid dynamic variables (residence time, velocity,
temperature, area ratio, and density) and species concentrations as a function of
normalized distance. Plotting capability is available for fluid dynamic properties,
chemical species concentration, total derivative with respect to distance, partial
derivatives of these total derivatives with -respect to any other variable and the net
production rate of any reaction.
The fluid dynamic and chemical relaxation equations are integrated numerically
using an implicit procedure developed by Dynamic Science personnel. The advantage
of the implicit technique is that it allows chemical systems near equilibrium to be
analyzed in a practical manner. This is important in emission analysis as the thermo-
dynamic properties of the reacting mixture are controlled by the near equilibrium
concentrations of the major species.
The program was developed for the kinetic analysis of rocket engines, and
was selected by the Interagency Chemical Rocket Propulsion Group as the reference
program for the Aerospace industry. It has been modified to analyze generalized
chemical flow problems rather than the more limited rocket nozzle analysis.
The flow chart in Figure 34 gives a brief description of the input and output
of GKP.
7.1.3 Initial Calculation
The initial effort involved setting up the Steady-State Spray Combustion
Program to provide the liquid heat release rate profile. Figure 35 shows the
resulting calculation of the evaporation rate for Jet A fuel injected into a 200 feet per
second air stream at low velocity (5 feet per second velocity component along the air
stream). Initial mass median drop size was estimated as 50 micron and a relatively
low standard deviation of the spray (nearly uniform size distribution) was used as
these are the characteristics of the rotating cup atomizer being employed. (200 Feet
per second being me air stream velocity that will occur at full heat release in the
preliminary combustor design.)
64
-------
PROGRAM INPUT
1
FLUID DYNAMIC MODEL INPUT
a) pressure or area profile
b) initial conditions (P,V,T)
c) mass addition profile
d) momentum addition profile
e) energy addition profile
CHEMICAL MODEL DATA
a) chemical reaction set e.g.
NO-t-M = N-KHM
b) initial concentrations of the
species in the reaction set
1
THERMO-CHEMICAL DATA
A library of thermal chemical
data on over 400 species Is
available.
GKP
The computer program
solves the 1-D non-
equilibriom reaction gas
flow equations considering
mass, momentum and energy
transfer using an implicit
numerical integration scheme.
PROGRAM OUTPUT
FLUID DYNAMIC PROPERTIES
Pressure, temperature, density,
velocity, specific heat, and enthalpy
are printed out for specific print
stations
CHEMICAL SYSTEM PROPERTIES
Species concentrations, net pro-
duction rates, and influence co-
efficients are printed out for
specific print stations ,
FIGURE 34. FLOWCHART, GENERALIZED KINETICS PROGRAM
From Figure 35, it is seen that vaporization is nearly completed two inches
from the injection point. Also shown are vaporization histories for three of the typical
drop groups comprising the spray. This short lifetime of the liquid at the full heat
release condition of the combustor provides the logical basis to initiate the emission
calculations based on gas-gas equilibrium. Droplet lifetimes at lower heat release
rates are shown in Figure 36.
To provide a comparison base between the non-equilibrium emission estimates
calculated using the kinetics program and those expected at chemical equilibrium, a
matrix of air-fuel ratios was run. Figure 37 shows the calculated flame temperatures
and Figure 38 shows equilibrium concentrations of carbon monoxide and nitric oxide
(defined as CO and NO, respectively). A second series of runs on the equilibrium
program was run with NO suppressed (program did not consider NO species) to provide
start conditions for the non-equilibrium kinetics calculations. Kinetics runs were
65
-------
30 MICRON 50 MICRON
INITIAL DROP SIZE INITIAL DROP SIZE
TOTAL SPRAY
70 MICRON
INFTIAL DROP SIZE
.1 .3 .5 .7 .9 1.1 1.3 1.5
DISTANCE ALONG COMBUSTOR FROM POINT OF INJECTION - INCHES
1.7 1.9 2.1
JET A FUEL AT 80'F
FIGURE 35. FUEL DROPLET LIFETIME AT MAXIMUM HEAT RELEASE
RATE (2 x 106 BTU/HR)
1.09
54.5
109
AIR FUEL OF 26 TO 1
0.2 0.4 0.6 0.8 0.9 1.0
DISTANCE ALONG COMBUSTOR
FROM POINT OF INJECTION, INCHES
FIGURE 36. FUEL DROPLET LIFETIME, 50 MICRON DROP SIZE, AT 109, 45.5,
AND 1.09 LB/HR OF FUEL
66
-------
4000
iiOOO
2000
FUEL IS JET A AT 80° F
LOWER HEAT VALUE 18,400 BTU/LB.
AIR AT 80° F
AIR/FUEL RATIO
10 20 30
FIGURE 37. EQUILIBRIUM FLAME TEMPERATURE AS A FUNCTION
OF AIR/FUEL RATIO
67
-------
10
2CT
AIR/FUEL RATIO
FIGURE 38.
EQUILIBRIUM CONCENTRATIONS BY VOLUME OF CARBON
MONOXIDE AND NITRIC OXIDE AS A FUNCTION OF AIR/
FUEL RATIO
68
-------
A/F = 4 . 0
Mole Fraction
,«6]i957
l.ir-k-2«
C02
CO
t
\?0
H?C
:w 4. ?."<£-14
c ,
.240527
0.
o.a
.000021
A/F
N2
02
H20
H2
OM
0
H
ARGON-
NO
N
N02
C02
CO
C
N20
\M3
= 6.5
Mole Fraction
.53»73l
2.46E-13
.032206
,165783
5.S7E-8
1.8AE-12
,000007
,006907
0.
3,V?E-14
6.51E-1B
0,021520
•1V4641
0,
5.1.1E-14
,m)C004
A/F
N2
02
H20
H2
OM
0
H
A3GON
NO
\
N02
C02
CO
C
N?0
NM3
= 9.2
Mole Fraction
0.671527
5.45E-8
.114005
,079833
0.01)0034
8.7F-B
.000274
.007968
0.
2,3<5t-10
2.8F-12
.044272
.13?084
0.
1.35E-1C
4.1F-7
FIGURE 39. PRIMARY EQUILIBRIUM COMPOSITION
initiated with zero initial NO as the time to vaporize liquid was calculated as small
compared with NO formation time. Also, the primary flame zone is fuel-rich and
little NO is formed even at equilibrium (infinite dwell time) for a fuel-rich system.
Figure 39 shows the equilibrium start conditions used for primary zones at air-fuel
of 4.0, 6.5, and 9.2. The equilibrium results are for a kerosene type fuel (C-^^22
molecular weight used) with heating value of 18,400 BTU/lb and thus are not general
results. Other fuels must be examined on an individual basis.
The third area of initial calculations involved gathering input data for the
kinetics program. The program starts a kinetic chemistry calculation at a prescribed
input chemistry (calculated from equilibrium) representative of the primary zone.
Mass, momentum, and energy are then inputs to the primary zone at prescribed rates
based on mixing processes occurring within the combustor. The basic input then con-
69
-------
sists of tables of these quantities as a function of distance through the combustor. At
constant pressure the gas dynamic relations within the program (hen calculate the
residence time (velocity) as a function of these quantities, i.e., residence time is not
a constant throughout the combustor. Addition rate tables used for analysis of the
combustor configurations will be presented in the following section.
The kinetic reaction mechanism used for emission estimates is shown in
Figure 40. The low molecular weight hydrocarbon reactions were necessitated by
the addition of the methane species at an air-fuel of 4 (see Fig. 39).
7.1.4 Combustor Design by Computer Modelling
The preliminary combustor design used is based on typical criteria used in
the design of gas turbine combustors modified to limit the peak temperatures in the
flame (Fig. 41). This has been shown by the reaction studies to reduce the formation
of nitric oxide. The control of flame temperature is achieved by control of the amount,
position, mode and speed of air injection into the combustor.
The combustor is divided into three zones as follows:
1. A primary flame zone consisting of about one-third of the combustor
volume and lying immediately downstream of the point of fuel injection.
This is where ignition and combustion commences.
2. A secondary flame zone consisting of about one-third of the combustor
volume and lying immediately downstream of the primary flame zone.
In this region, most of the combustion reactions approach completion.
3. A tertiary flame zone consisting of the remaining combustor volume and
lying immediately downstream of the secondary flame zone. In this
region, air is added and mixed with the combustion products to bring
them to a safe level of temperature for use in the engine boiler.
Air is injected into these three regions either as high velocity jets or as
low velocity films flowing along the combustor walls. The later films
serve in the office of a control on wall temperature by minimizing the
amount of hot combustion productions that can impinge on the walls of
the combustor.
To calculate emissions, a knowledge of the method of air injection and mixing
in the combustor is required. Two different methods of air injection were devised
and designated as designs A and B (ref. Fig. 41 and 45, respectively). Small
modifications in the methods of air injection were done with both designs and the
resultant emissions estimated.
70
-------
REACTIONS
N2
02
H2
OH
HJ>0
M02
N20
C02
CO
CH4
CH3
CH?
CH
END
CO?
CO
H2
OH
W20
OH
W20
MQ
N!2
NQ
N2
CO
CO
CO
CO
CO
CO
C
C
C
CO
CH3
C
C
CH4
CH3
CH4
CH3
H2
M
M20
H20
OH
0
OH
CH3
LAST
• N * N , A«l , OElfl ,
=0*0, Aoi . 9Elft ,
= H f H , AS7.5E1*.
s H * (J, Ae.J,e>£*l8. N
= H * OH, Asi.~l7fc.17,
- 0 * NO, Atl.6El5,
s 0 * N2. As i. OE18.
= CO * 0, As 5.1E-H5, N»0
s C * 0 i A = 3,nEl6, N=!],5» 8*0
= fH3 + H , is 3.QE16. ^'= 0
- CH? * H , AS J.CF16, \= C
s CH * H , 4= 3. OF. ift, \s o
' C * H , AS 3.0E16, >.'* 0
TUK REAX
+ H = CO + CH , AS 5.6E.11,
*
*
*
»
*
*
+
+
CO
02
h =
H s
0 s
0 =
= CO?
s OH
H2 + 0
OH * H2
H * 0?
OH + OH
0 s 02 * N ,
0 - k
0 * !>' ,
«• C?s N02* 0
*•
*
*
*
+
*
*
+
*
*
*
*
*
*
*
*
•f
*
*
*
+
+
*
*
*
*
02 =
02 *
H =
.'M :
0 = C
H? '
N20 * 0 ,
C02 «• 0,
C * OH,
C * MO,
* 02,
CH +OH ,
H s CH +0 »
H2
H2Q
OH
OH
H2 s
CH3
CH2
C
CH
CH
CH3
CH
CH3
CH
CH2
CH
CH3
CH3 •
H20
s CH
= CH
= CH
s CH
CH4 * H,
s P.H
* CH
s CH3
: CH?
s CH3
* CH2
= CH?
• CH2
= CH2
= CH3
» CH2
• CH2
CH4 * 0.
* C
* OH
, Ac
A*
As
, As
Asl
A»l
A=l
'As 3
A=1,9E
, A = 2
, A»2
1 , 7 4 F. 1 3 ,
2 , 1 9 F 1 3 ,
2 . Ofc!14 .
•5.75^12,
.8E8, N
.3E13. -N
Nil , 0. B*0,
Uo 0 1 5 1 8s 0 1
=liQi B« Oi
• J » 0 . 0 1 B * 0 ,
.'N8l,.n. fa«0,0
,« R«3.58.
,0>
,5, B=0,0,
,5, 8=0,0.
,5, B*0,0.
,5, 8=0,0,
fJiC.O, B»l,
.11E16,
.70E16,
N s o , n ,
'I»C , 0»
N«0,0i
NaO.Oi
•-l.S.Bs
•0,0. 8=
,8El3, N«0,0. Bs
.OE*l3i
*13, NB0
ASJ..2E + J4, N«0
A= 1 . 2E*
16, ;i=i,
N« 0,0,
, B>54,
, B« 25
0 B«0,,
N =
N =
P« 9
B= 5
R»17
BCQ ,
5,94
0.0,
1,05
B>
15.
,83,
0,
0,
n,
0,
,79,
,
080,
1,0,
1,0,
,45,
,15,
,0,
78,
1
1
26.8,
As ?,4E!+l3» NE0, b°l,9')7,
t* 3 ,
Ao 3.
* M
* OH
* C
* 0?
f>13, n*
F*13. N=
A«5
A« 1.5E*
* CH2
* Cw
+ CH
* CH2
* CH2
+ CH<
* H
* H2
* OH
+ OH
* 0
* OH
AS 1
A«5
Acl
L4. Nt
AB1
A«l
A* 1
A'l
A»l
A»l
A'l
A'l
Ael
AE1
A'l
A'l
Ai 1,E*13, N«0
• CH4 * OH. *• 5
C , B»0.
0 « B"0.
•30E11,
.05E11,
,3EH,
•75E10,
0 , . Bil4
.05E11.
.05E11,
,05E11.
.05E11.
.05E11.
.05E11.
.05E11.
.05L11,
.05F.11,
,05Ell.
.05E11.
,05E11.
,. B» 7,
,E*14, N«0,, B«9
,
,
N =
N«
N«
N>
i
N»
Nc
N«
N>
N«
Nc
N«
No
N«
N«
N>
N*
3,
,9.
-,•5,
-,5,
-,5,
-,5,
-.5,
-,5,
-.5,
-,5,
',5,
-,5,
-,5,
-,5.
-,5,
-,5,
-,5,
-.5,
DDK PACE 7-6
DDK PACE 7.6
ODK PARE 7-6
14
LEEDS 2 NOV 6«, P-31
TR^A69-103 P-17, JAN 69
16
ODK. P-7-6
1D-2P PROG, P 7-49
1D-2P PROG, P 7-49
1D-2P PROC, P 7-49
1D-2P PROC, P 7-49
LEEDS 1 MAY 6«, P-4
B»0,0i 1D-2P PROC 7-bCi.
8»53,, 1D-2P PROC 7-51,
LEEDS ? NOV 68, P-l
LEEDS ? NOV 6B, P-V
TS-AA9-103 P-21, JAN 69
LEEDS 2 NOV 6fl, P-20
TR-A69-103 P-18, JAN 69
TR»A69«103 P-18. JAN (SV
TR.A69-103 P-18. JAN 6V
28 50
163
103
ioa
107
42
43
B«2,24,1D-2P PROC 7-5l,
B«2,24,iD-2P PROG 7-5i,
B»2,24,1D.2P PROC 7-51.
8-28,2ilD-2P PROC 7-5li
174
B«2,24,10.2P PROC 7-52.
B«2,24,1D-2P PROC 7-52.
B«2,24,1D«2P PROC 7-b2.
B«2,74,io.2P PROC 7-52.
B«2,74,1D«2P PROC 7«52i
B-2.82.1D-2P PROC 7-52.
B«2,74,1D-2P PROC 7-52.
8«2.8«,1D.2P PRCC 7-52.
B«2,74,iD«2P PROC 7-53.
B«3,19,1D-2P PROC 7-53.
8"2,74ilD.2P PROC 7-53.
B«2,83,10-2P PROC 7-33.
172
176
33
43
49
50
54
59
63
64
66
6'
70
71
80
81
86
8*
100
101
CARD
FIGURE 40. REACTION SET
71
-------
Design A, Configuration No. 1 (Fig. 41b)
Twenty-five percent of the total air required for combustion is injected over
the rotating cup and mixes instantaneously with the fuel ejected from the lip of the
cup. Another twenty-five percent of the total air is injected as film cooling along the
combustor wall. The remaining air is injected at the juncture of the primary and
secondary flame zones. The assumed rates of mixing of the various quantities of air
is shown in Figure 41a, the resultant air-fuel ratios, gas temperatures and emissions
are shown in Figures 41d, e and f, respectively, for the maximum heat release rate
of 2 x 106 BTU/hr and an exhaust temperature of 2500° F.
The results show carbon monoxide on the high side but low values of nitric
oxide.
Design A, Configuration No. 2 (Fig. 42b)
Identical arrangment of air admission as in Configuration No. 1, but with
modifications to the method of the air injection at the juncture of the primary and
secondary flame zones so that half this air recirculates into the primary zone rather
than, as before, into the secondary zone.
The rates of mixing of the air, resultant air-fuel ratios, gas temperatures
and emissions are shown in Figures 42 c, d, e and f, respectively.
The results again show carbon monoxide on the high side and also a consid-
erable increase in nitric oxide. The large increase in nitric oxide is caused by the
higher primary zone flame temperature which results from the recirculation into it
of additional air.
Therefore, for minimum nitric oxide, the recirculation of air must be mini-
mized and the primary flame zone maintained with minimum air. Some increase in
secondary flame zone volume must be provided to minimize carbon monoxide.
Design A, Configuration No. 3 (Fig. 43b)
The large quantity of wall cooling used in Configurations 1 and 2, while pro-
viding low wall temperatures and hence long combustor life, will tend to reduce effec-
tive combustor volume due to quenching of the flame reaction at the cool wall. The
quantity of film cooling was therefore drastically reduced as it appears that combustor
volume was low.
The results show that emissions are very similar to, but slightly higher than,
the results obtained with Configuration No. 1, and have low nitric oxide with high
carbon monoxide.
72
-------
Design A, Configuration No. 4 (Fig. 44b)
Identical arrangement as in Configuration No. 3, but with modifications to the
method of air injection at the juncture of the primary and secondary flame /ones so that
half this air recirculates into the primary zone rather than into the secondary zone, as
was in Configuration No. 3.
Again, as with Configuration No. 2, a large increase in nitrix oxide but small
change in carbon monoxide is achieved.
The analysis gives results which would not, in practice, he expected; i.e.,
the increased wall temperatures would provide a larger effective volume in Configura-
tions 3 and 4 and reduce carbon monoxide. Therefore, testing of these configurations
is needed for better understanding of the processes at work.
Design B, Configuration No. 1 (Fig. 45b)
Modifications were made to the design of the combustor as in Figure 45b, and
air admissions as in Figure 45c used. The essential difference between Design B and
A is that a much slower rate of mixing of the air admitted at the juncture of the primary
and secondary zones is used in Design B.
Calculations at full heat release were made of the flame temperature and
resultant emissions of carbon monoxide and nitric oxide are as shown in Figures 45c
and f.
In comparison with Design A, Configuration No. 1, a large increase in nitric
oxide occurs, though within required values. Some increase in carbon monoxide also
occurs.
The prime reason for the increased nitric oxide lies in the reduction of the
rate of admission of the secondary air. This permits combustion to proceed at higher
temperatures for a longer period of time than compared to previous configurations.
Increased admission of secondary air and earlier injection of tertiary air would serve
to reduce both NO and CO considerably.
Because the combustor has to operate at any heat release rate from 100 per-
cent to 1 percent, calculations of temperature and emissions were made at heat releases
of 50 percent and 1 percent also. These results are shown superimposed on the results
for 100 percent heat release in Figures 45e and f.
A very large increase in NO emissions occurs as heat release; is reduced;
such values lying outside specification limits. The reason for this is the large increase
73
-------
in time spent at high temperature when combustor heat output is reduced as likewise,
more time is available for complete reaction to carbon dioxide to occur.
Design B, Configuration No. 2 (Fig. 46b)
Identical arrangement of air admission as in Configuration No. 1 but with
modifications to the method of the air injection at the juncture of the primary and
secondary flame zones so that 100 percent of this air recirculates into the primary
zone rather than, as above, into the secondary zone.
As before, the rates of mixing of the air, resultant air-fuel ratios, gas tem-
peratures, and emissions were computed and are shown in Figure 46c, d, e, and f,
respectively, for 100 percent heat release and also for 50 percent and 1 percent of
full heat release.
The results are similar to those obtained with Design B, Configuration No. 1,
save that some increase in nitric oxide occurs. This is due to the increased time at
high temperature caused by introduction of additional air into the primary zone by
recirculation from the secondary zone. A similar result was obtained with Design A,
Configuration No. 2, and for the same reason.
Heat Losses
At low levels of heat release, it is expected that substantial reductions in
flame temperature could occur due to the large masses of cool metals surrounding
the flame. This heat loss would depend considerably on the design and location of ihe
boiler. Calculations were done of the effects of small heat losses on the emissions
from Design B in both Configuration No. 1 and 2.
A substantial reduction in volume of nitric oxide emissions is obtained with
only small heat losses, reference Figures 47 and 48 with both configurations. Effects
on carbon monoxide formation are small.
As nitric oxide emissions are still excessive at low heat release rates, it is
necessary for considerable reductions in flame temperature to be made.
Conclusions
Emissions of nitric oxide are, at low heat releases, considerably outside
required limits and well within limits at high heat releases. Carbon monoxide is well
within limits at low heat releases, and marginally in limit at high heat releases.
Vaporizer (boiler) design could have a major effect on emissions as heat losses from
the flame can be high.
74
-------
10,000_
1,000
a
a
i
E-
K
H
Z
W
O
z
o
o
100
10
CO
NO
5% HEAT
LOSS
NO HEAT
LOSS
6 8
DISTANCE - INCHES
10
12
FIGURE 47.
EMISSIONS AT ONE PERCENT HEAT RELEASE WITH
AND WITHOUT A FIVE PERCENT HEAT LOSS FOR
DESIGN A, CONFIGURATION NO. 1
87
-------
10,000.-
1,000 -
I
o
H
U
u
5% HEAT
LOSS
468
DISTANCE - INCHES
12
FIGURE 48. EMISSION AT ONE PERCENT HEAT RELEASE WITH
AND WITHOUT A FIVE PERCENT HEAT LOSS FOR
DESIGN A, CONFIGURATION NO. 2
88
-------
Discussion
Results show that for minimum emission of nitric oxide, the rate of admission
of air must be carefully controlled. In the primary zone, the flame must be kept rich,
with minimum air; and in the secondary zone excess air must be added as rapidly as
possible. If this can be done over the entire operating range of the combustor from
1 percent to 100 percent heat release, then the formation of nitric oxide can be elimi-
nated almost completely, especially if heat losses to the vaporizer can be used to limit
flame temperature. Carbon monoxide should not present a problem with such an
arrangement of air admission, save perhaps at or near maximum heat release. If
this is so, an increase in combustor volume may be required to hold emissions within
limits. By providing sufficient volume, the carbon monoxide formation would be
reduced to an extremely low level. Heat losses to the vaporizer would, however,
tend to raise CO emissions.
In essence, the distriction of CO and the formation of NO is a process that
is controlled by the mixing residence times and heat losses of the reaction gas within
the combustor. The mixing process can be controlled by hardware configuration and
it is possible, therefore, to design a combustor that will minimize both NO and CO
emissions over a wide range of turndown and operating conditions.
7. 2 COMBUSTOR AND TEST RIG DESIGN
7.2.1 Combustor Design
A combustor was designed and fabricated. These parts are shown in
Figures 29, 49, 50, 51, 52, and 53. The distribution of the air admission into the
combustor was set up initially per Design B, Configuration No. 1.
The heart of the system is the rotating cup fuel injector which makes the
100 to 1 turndown ratio possible. The rotating cup has the following advantages:
•Essentially no fuel pressure. This allows the use of a small low
pressure fuel pump and avoids fuel contamination because there
are no small fuel metering orifices.
•Degree of atomization is not dependent on fuel flow or viscosity.
•Minimum cost. The cup and its drive motor are inexpensive as is
its auxiliary equipment (fuel pump and ignition).
•Is a proven design. It has been used for many years in combustors.
89
-------
COMBUSTOR CASE
EXHAUST STACK & LOCATION
FOR BOILER
I
.
'
'
•
.--'
FIGURE 49. SIDE VIEW OF ROTATING CUP COMBUSTOR ASSEMBLY
•Controlled spray - the angle is predictable and hence ignition reliability
is high.
•No power requirements. For test flexibility the cup is separately
driven. In practice it would be mounted onto the fan motor shaft and
the additional power required would be negligible.
A weight breakdown for the experimental and production versions of the
combustor assembly, together with the materials used is shown below.
Current Weight
Pounds
28.78
38.00
0.69
3.00
4.28
1.25
76.00
Combustor
Combustor Case
Fuel Injector Cup
Injector Drive Motor
Combustor Swirler and
Dome Assembly
Support Pins and Ignitor
Total Weight
Production Version Weight
Pounds
6.0
4.75
0.23
*
1.25
0.75
12.78
* Integrated with fan motor.
90
-------
COMBUSTOR CASE
••flHB
FIGURE 50. FRONT VIEW OF ROTATING CUP COMBUSTOR AND CASE
Combustor materials are:
Combustor Case and Flanges
Combustor Liner
Combustor Swirler and Dome
Assembly
Fuel Injector Cup
Support Pins
Mild Steel (0.062 in. gauge)
Hastelloy X (0.062 in. gauge)
321 Stainless Steel
Mild Steel
Mild Steel
In practice, tests indicate that all materials used could be low cost mild steel
or cast iron save for the liner which would be a 300 series stainless steel.
91
-------
FIGURE 51.
FRONT VIEW OF ROTATING CUP
COMBUSTOR AND CASE WITH ROTATING
CUP REMOVED
COMBl'STOn DOME
COMDUSTOrf DOME
PRIMARY DILUTION TUBES
FIGURE 52.
REAR VIEW OF ROTATING CUP
COMBUSTOR AND CASE
92
-------
ROTATING
CUP MOTOR
FIGURE 53. ROTATING CUP AND MOTOR ASSEMBLY
7.2.2 Combustor Test Rig Design
Air Supplies
Prior to tests with a fan and air metering valve, tests were done with a
slave air supplied by a remote air compressor. The very wide range of fuel flows
to be investigated required a corresponding wide range of air flows. For
tests to have significance, the air flow must be measured accurately. This requires
metering orifices having pressure losses far in excess of the capability of the chosen
fan and therefore air to the combustor had to be provided from a higher pressure
source. This air was supplied from a centrifugal compressor that supplied air, via
a system of metering orifices, through a ten inch diameter pipe, directly mounted
onto the combustor case, Figure 54 and 56. This air was manually controlled to any
required flow by means of a throttling valve. Throughout the test, the air supply was
93
-------
MANOMETERS
EMISSION
EXHAUST PICKUP
PROBE
FLOW STRAIGHTENER
TC
AIR INLET
3.0 ORIFICE
/ SHARP EDGED \
AIR FLOW
\ MEASURING ORIFICES/
10.0 DIA.
FLOW STKAIGHTNER
BOILER COMBUSTOR
- «4»- —*f
SECTION SECTION
FLOW STRAIGHTENER
AIR MEASUREMENT
SECTION
FIGURE 54. SCHEMATIC OF COMBUSTOR TEST RIG
-------
FIGURE 55. REAR VIEW OF FAN AND CONTROL VALVE ASSEMBLY
SHOWING ANTI-SWIRL PLATES INSTALLED
FIGURE 56. COMBUSTOR RIG SHOWING ARRANGEMENT OF AIR
METERING ORIFICES
95
-------
maintained at levels of flow, temperature, and pressure appropriate to the fan
capabilities (2830 Ib/hr, 80° F, and 12 inches water), but is capable of a much larger
range, with up to 7000 pounds per hour of flow, at temperatures from -40° F to +1200° F
and pressures to 250 psig.
There is an aerodynamic influence on the combustor caused by the particular
method of control of the air from the fan to the combustor. Air is passed through a
metering valve consisting of twelve variable area openings into the combustor at a
velocity of 95.5 feet per second as dictated by the pressure drop across this area
(Fig. 55). Because of the interaction of the aerodynamics of the fan and the air by-
passed overboard, and because of the turning losses in the metering chamber, this
air has a profile of velocity in both circumferential and radial directions, and a tan-
gential velocity component that varies considerably over the operating range of the
combustor.
At maximum fuel flow, the air exiting from the metering valve into the com-
bustor annulus, is accelerated and passes into the combustor at a velocity of 195 feet
per second. At this flow, the effects of radial and swirl velocity profiles on combustor
air distribution are assumed negligible. However, at minimum fuel flow the air
velocities into the combustor are small at 1.95 feet per second, and as the same 95. 5
feet per second velocity is maintained at the control area, the effects of velocity pro-
files could be considerable. It is not possible to simulate such aerodynamic influences
except by use of the fan and this, of course, destroys the object of the test rig. There-
fore, final combustor development must be carried out with the fan. In order to
minimize differences between fan and rig aerodynamics, the combustor was designed
such as to minimize the effects of entry velocity variations. On the rig, any velocity
profiles from the upstream metering section are eliminated by means of a flow
straightener.
7.2.3 Instrumentation
Flow Measurements - Air
Air flow is measured by means of any of three orifice runs that are mounted
in parallel upstream of the combustor (Fig. 56 and 54). Because of the wide range
of air flows, three different orifices of 3/4, 1-1/2, and 3 inches diameter are used.
Each orifice is sized for optimum Reynolds number at air flows corresponding to
respective fuel flows midway in the ranges of 109 to 22.8, 22.8 to 4.77, and 4. 77 to
1 pounds per hour. The orifices are sharp edged, with corner pressure taps, and
have upstream flow straighteners to minimize aerodynamic influence on the metering
section from the upstream on-off valves. These valves are manually operated so that
air flow can be diverted to the metering orifice appropriate to the air flow being used.
Because of the cumbersome nature of this instrumentation, it is not possible to simulate,
except approximately, the transients that will in practice occur as fuel flow is varied.
96
-------
Flow Measurements - Fuel
Fuel flow is measured by three Fisher-Porter flow meters mounted in-series
and covering a range of flows from 0. 2 to 200 pounds per hour. The flow meters are
calibrated against a master flow meter which is also used to calibrate the flow meters
used in tests on the fuel metering valve. This minimizes the possibility of cumulative
errors.
Temperature - Measurement
A Solar design and fabricated Pt-Rh thermocouple (Fig. 57 and 58) was used
to measure combustor outlet temperatures. Triple radiation shielding is provided to
a 0. 04 inch diameter bare bead Pt-10%Rh versus Pt thermocouple. Three concentric
Pt-Rh radiation tubes are arranged to reduce radiation losses to low levels in this
unit. High convective heat inputs (to offset radiation heat losses) are provided by a
high velocity aspiration system that draws the combustor discharge across the thermo-
couple bead. Good results were obtained at temperatures up to 3030° F and velocities
of 0. 6 Mach number. The thermocouple could be mounted in forty different locations
to provide an accurate description of both circumferential and radial temperature
profiles at the exit of the combustor as shown in Figure 59.
Pressure Measurement
Pressure measurements are taken by means of Bourdon gauges, water and
mercury manometers. All critical pressures are duplicated to minimize errors.
Because at fuel flows much below 20 pounds per hour, the air flow is so small as to
cause negligible combustor pressure loss, it is not possible nor is it very meaningful
to accurately record the actual loss at such flows. Fuel pressure is not measured, it
can be considered with negligible error, as being at the prevailing combustor pressure
and which is, without a vaporizer, ambient pressure.
Fuel Supplies
Fuel is supplied to the combustor via the flow meters at a pressure slightly
in excess of the combustor pressure and is throttled to the required level by means of
a needle valve. The standard fuel used in the tests is JP-5. Such a fuel has a more
precisely controlled specification than the fuels required to be used (#1 Diesel, Jet
A, and kerosene), and has combustion characteristics typical but slightly worse than
these fuels (i.e., is likely to have slightly higher CO and H/C emissions), Appendix
B. Other fuels available and briefly used are JP-4 and #2 Diesel, and which represent
fuels having combustor characteristics outside the best and worst likely limits expected
of the specific fuels.
97
-------
THERMOCOUPLE LEADS
TO ASPIRATOR
JUNCTION
-1ST RADIATION SHIELD
2ND RADIATION SHIELD
3RD RADIATION SHIELD
FIGURE 57. END VIEW - HIGH TEMPERATURE PROBE WITH TRIPLE RADIATION
SHIELD AND HIGH VELOCITY ASPIRATION SYSTEM
'
FIGURE 58. INSTALLATION OF A HIGH TEMPERATURE THERMOCOUPLE
98
-------
FIGURE 59. CIRCUMFERENTIAL AND RADIAL POSITIONS OF THERMOCOUPLE
AT THE EXIT OF THE COMBUSTOR
Emissions are powerfully influenced by air-fuel ratio which is influenced by,
among other things, fuel volatility. Therefore, a range of volatilities was deemed
necessary in the tests in order to obtain as much understanding as possible of the
processes involved in emissions (especially emissions of NO2)
Emission Sampling Probes
Various types of sampling probes were investigated and the final type used is
as shown in Figure 60. The probes consisted of 29 pickup points located at points of
equal area across the combustor to provide a representative average sample.
In the case of CO, NO and CO2 the probe was cooled to room temperature by
means of an air cooling jacket so as to ensure no further reactions inside the probes
and to allow removal of water vapors.
A separate probe, of identical design, was used to sample for unburnt hydro-
carbons. To ensure no reactions inside the probe, but also to prevent condensation,
the probe was cooled to a temperature of 350 to 375° F. The entire HC sampling line
connecting the probe to the FID was electrically heated to 350 to 375° F for the same
reason.
99
-------
PROBI TEMPERATURE
CONTROL AIR FLOW
• WELDED TOT BMP. CONTROL TUBE
21 PORT
EQUAL
ARIA
SAMPLING
PROBBS (I)
ONI roR HC
ONC FOR NO * CO
(10- TO EACH
OTHER)
EXHAUST
12 INCHC8
SAMPLE TO FID FOR HC PROBI
SAMPLE TO NMR FOR NO 4 CO PROBI
COOLINC
AB
/ { ELICTRIC HIATIR
FOR HC PROBE
"JAMBIDTT AIR TIMPEHATURI
FOR CO It NO PROBI
FIGURE 60. SCHEMATIC OF EMISSION PICKUP PROBE
7.3 COMBUSTOR DEVELOPMENT
7.3.1 Summary
Initial combustor development was done on a rig without use of a fan and
metering valve and emissions were at reasonable levels. Tests with a fan and
metering valve required more development to get acceptable emissions. When a
vaporizer was installed at the combustor exit emissions levels increased to unaccept-
able levels and indicates some radiation shielding may be necessary to get acceptable
emissions.
7. 3.2 Preliminary Combustor Rig Tests
The combustor was run over a range of fuel flows from 5 to 109 Ib/hr using
JP-5 fuel, and with the air-fuel ratio maintained at various values between 22 and 30.
From these results it was possible to find the optimum air-fuel at any fuel flow for
minimum emissions. This optimum air-fuel is shown in Figure 61. It can be seen
that only at maximum fuel flow is the design value of 26 air-fuels achieved. With
reducing fuel flows a small reduction in air-fuel is required. However, at values of
fuel flow less than about 50 Ib/hr a considerable excess of air is required to obtain an
acceptable flame.
100
-------
a
100
90
80
70
GO
o .r>0
40
10
I
I
I
24 2(i 28 30
AIR-FUEL RATIO
34
FIGURE 61. Affi-FUEL RATIO FOR MINIMUM EMISSIONS AS A FUNCTION
OF COMBUSTOR AIR FLOW
The actual emissions resulting from maintenance of this optimum air-fuel
relationship with fuel flow are shown in Figure 62 and 63 together with the increased
emissions resulting from a 10 percent error in air-fuel (it is not to be expected that
control of air-fuel can, in practice, be maintained much closer than this).
7. 3. 3 Combustor Pressure Loss Reduction
Up to this date, combustor pressure loss had been 12. 7 inches HgO. Ic was
necessary to reduce it to 8 inches I^O in order to keep the power consumption of the
fan within reasonable limits. This was done and there was a considerable increase in
flame length and emissions, notably hydrocarbons and carbon monoxide. Changes to
the hole pattern of the combustor were made though the original staging of air-fuel
ratios within the primary, secondary, and tertiary flame zones was maintained.
101
-------
17
Wa
= 2835 LB/HR AIR
±10% FROM OPTIMUM
AIR-FUEL
OPTIMUM AIR-FUEL
100
'100%
FIGURE 62. EMISSIONS OF CARBON MONOXIDE AS A FUNCTION OF COMBUSTOR
AIR FLOW, AT OPTIMUM AIR FUEL FOR MINIMUM EMISSIONS AND
ALSO WITH A ±10% DEVIATION OF AIR FUEL FROM OPTIMUM
The effect of reduced pressure drop was that penetration of air into the flame
was reduced. Consequently, the poor mixing of air and fuel resulted in lean and rich
pockets that caused lengthening flame due to, in both instances, a reduction in the
rate of combustion.
Modifications were made to the hole size and pattern to improve mixing with
beneficial results, though it was not possible to reduce flame length to the values
102
-------
1.6 -
1.4 ^
u
3
^ 1.2
S
§
2
to
S
1.0
0.8
O 0.6
€
0.4
0.2
NOX LIMIT
10% FROM OPTIMUM
AIR-FUEL
OPTIMUM
AIR-FUEL
Wa --- 2835 LB/HR OF AIR
10
20
30
40
50
AIR FLOW,
60 70
Wa
80
90 100
Wa
100%
FIGURE 63. EMISSIONS OF NO2 AS A FUNCTION OF COMBUSTOR AIR FLOW,
AT OPTIMUM AIR-FUEL FOR MINIMUM EMISSIONS AND ALSO
WITH A ±10% DEVIATION OF AIR-FUEL FROM OPTIMUM
recorded with the high loss combustor. Results are shown in Figures 64. By
maintenance of an air-fuel of 22, it was possible to achieve the required goals in
emissions. Small errors in air-fuel would, however, result in emissions outside
limits.
In general, the results are in agreement with theory. Figure 64a shows the
time dependence of NO formation as with decreasing fuel flow (inversely proportional
to combustion time), emissions increase.
Figure 64b shows the time dependence of CO formation in that increasing
fuel flows result in increased emissions of CO. Minimum emissions are achieved
with reduced air-fuels as a result of increased reaction rate of the higher temperatures.
It would appear that an optimum combustor, operating at the design point of 26 air-
fuels, would have NO emissions considerably below limits but would require increased
combustor length to reduce CO.
Figure 64c is more difficult to interpret as large variations in hydrocarbon
emissions are seen. This might be explained by the high background emissions and
below which it is possible to maintain the flame when air-fuel is optimum.
103
-------
LOOT 1.38
0.9
0.8
0.7
0.6
0.5
H
2o.4
£ 0.3
g*0.2
0.1
0
1.1
1.0
ff
H0.9
K °'8
o
Si 0.7
j
u
£ 0.6
0.3
0 0.2
g
o.i
22.1 A/F
JPS FUEL
8 INCH COMBUSTOR LOSS
10 20 30 40 SO 60 70 80 90 100 110
FUEL FLOW LB/HR
B. CARBON MONOXIDE
JPS FUEL
8 INCH COMBUSTOR LOBS
28.5 A/F
A. NO2
REDUCED PRESSURE LOSS
LIMIT 1«.2B
JPS FUEL
INCH COMBUSTOR LOSS
10 20 30 40 SO 60 70 80 90 100 ISO
FUEL FLOW LB/HB
C. HYDROCARBONS
10 20 30 40 SO 60 70 80 90 100 ISO
FUEL FLOW LB/HR
FIGURE 64. EFFECT OF FUEL FLOW ON EMISSIONS AT DIFFERING AIR-FUEL
RATIOS USING RIG AIR SUPPLIES
104
-------
FIGURE 65. COMBUSTOR RIG WITH VAPORIZER INSTALLED
7. 3.4 Simulated Vaporizer Tests
A vaporizer was designed to simulate an actual installation and assembled
onto the end of the combustor (Fig. 65). The vaporizer consisted of 250 feet of 0.7
inch inside diameter stainless tube. In order to obtain sufficient heat transfer it was
necessary to use a gas side pressure drop of 1.75 inch of water. It was not therefore
possible to use with the fan because of the increased pressure losses.
Emission measurements, Figure 66, showed a considerably reduction in
flame performance. Unburned hydrocarbon emissions increased by up to fifty times,
carbon monoxide increased slightly and NG>2 emissions decreased about fifty
percent.
This flame performance is not acceptable and is a direct result of the heat
losses from the flame to the vaporizer caused, principally, by flame radiation.
Therefore, because of the major influence of the vaporizer on flame perfor-
mance, it is essential to develop the combustor as an integrated unit with the vaporizer.
It is noteworthy that nitrogen oxide emissions were reduced due to heat
losses, as is to be expected. This would allow a control on nitrogen oxide emissions
somewhat independent of air-fuel ratio provide sufficient volume was provided as to
allow reaction of hydrocarbons and carbon monoxide. By shielding the vaporizer
from the flame, the emissions as previously reported could be maintained.
105
-------
Q
"0.5
U
u
OS 0.4
§0.2
U.
o
u^
40
5" 35
U
g 30
K
K
O
o
U 20
U.
O 15
O 10
O
" r,
0
S
£25
O
U
U
£10
LIMIT- 1.38
N02 EMISSIONS
\ 1
I I
10 20 30 40 50 60 70 80 90
LIMIT - 21.3
CARBON MONOXIDE EMISSIONS
j i i i i
10 20 30 40 50 60 70 80 90
HYDROCARBON EMISSIONS
LIMIT- 0.48
10 20 30 40 50 60 70
FUEL FLOW LB/HR
80 90
FIGURE 66. EFFECT OF VARYING FUEL FLOW ON EMISSIONS WHEN AIR-FUEL
IS MAINTAINED AT A CONSTANT VALUE (26/1) AND USING A RIG
AIR SUPPLY AND BOILER
106
-------
The low levels of nitrogen oxides that have been reported in domestic furnaces
are probably as a result of radiation heat losses. Heat losses can be readily controlled
in such an installation because of the single fueling rate used. However, modulation
of heat rate requires on-off control of fuel which results in unacceptable warm-up
emissions of carbon monoxide and hydrocarbons.
7. 3. 5 Combustor Tests With Fan
The object of reducing combustor pressure loss was to permit testing with a
fan whose power requirement was within the limits set as a goal for parasitic power.
The combustor and fan were therefore assembled together, with the air metering
valve, for further combustor tests (Fig. 29). The fuel and air metering valves were,
however, manually operated.
Tests were done over a wide range of operating conditions varying from 109
to 5 Ib/hr. It was obvious that considerable differences in flame performance from
that seen using the test rig air supplies. Two effects were seen.
1. At high fuel flows considerable circumferential variations in air flow
occurred. Consequently, large differences, circumferentially, in flame
length resulted. Emissions were high, especially of hydrocarbons and
temperature distribution was unacceptable.
2. At low fuel flows the flame appeared uniform but air-fuels in the
various zones were different from those occurring on the combustor
test rig.
The circumferential air maldistributions were investigated by flow tests.
The basic check was to coat the flow surfaces with oil and then inject talcum into the
fan. These checks indicated that the flow was unstable, resulting in separations in
air at different circumferential locations. The nature of such flow differences are
shown schematically in Figure 67a and b.
The only method of effectively controlling such air maldistributions without
use of excessive pressure loss or combustor volume is to design the fan and combustor
as an integral unit with low flow velocities and the minimum of diffusion, as in
Figure 67c. As the location of the vaporizer can also have an effect on aerodynamics,
it also must be integrated into the aerodynamic design.
The axial air maldistributions are explained by the presence of swirl in the
fan. The swirl angle was too low to measure, but the tangential velocity component
was high in relationship to the combustion velocities. The resultant swirl caused
large axial and radial variations in the hole discharge coefficients.
107
-------
LATS SEPARATION OF AIR RESULTING IN HIGH VELOCITY A IB
ENTERING COMBUSTOR CAUSING A COLO SPOT.
EARLY SEPARATION OF AIR RESULTING IN LOW VELOCITY AIR
ENTERING COMBUSTOR CAUSING A HOT SPOT.
J
PREVENT SEPARATION BY USING A LARGER FAN OF LOWER VELOCITY
THAT DOES NOT REQUIRE DIFFUSION.
CUBaCNT SOLUTION IS TO IMTKKFOU ALONG M1XJNO DUCT BtTWCEN FAN AND
COMBUSTOR TO ALLOW WAKtt TO DIMPATI TO A LOWEB VILOC1TT.
FIGURE 67. AIR MALDISTRIBUTIONS DUE TO UNSTABLE DIFFUSION
108
-------
To control such a phenomena means design of a fan with low air velocities
and with a minimum of swirl. It is an advantage in this respect to run the fan at low
speeds when fuel flows are low, and to have the highest possible combustor pressure
loss. Metering plate pressure loss should be kept low.
An optimum fan design was not feasible at this stage involving, as it did,
unknowns in vaporizer configuration, cost, and time outside the program scope. There-
fore, as a temporary expedient, a large mixing duct, 36 inches in length, was placed
between the fan and combustor, Figure 67d. An immediate and radical improvement in
flame performance was noted though it was still apparent at the low fuel flows that
all air swirl had not been eliminated.
Temperature Traverses
It is essential that rig and fan tests duplicate each other. Otherwise, any
demonstration of a working package by use of a rig becomes worthless. Experience
indicates that at least half the problems of emission control lies in proper design of
a fan and its associated fuel and air control system. Such a system of control,
covering as it does an extreme range of flow variations, is considerably beyond current
state-of-the-art. A means of checking simulation precisely and in a simple manner is
by temperature traverses of both rig and fan package. Ordinarily, at the flame tem-
peratures involved (up to 3200° F), thermocouples are inaccurate and have a very
short life. The alternate approach of emission traverses is time consuming, laborious,
and expensive. Solar has developed a high temperature thermocouple that has a long
life and is accurate (to 1%) at the temperatures involved. Thus a simple tool is avail-
able that enables precise definition of aerodynamic simulation and hence combustion
reactions by comparing temperature profiles on the rig and actual hardware.
A thermocouple was therefore installed for such a traverse. Traversing
was done by locating the thermocouple in forty different locations as shown in
Figure 59 and the results and conclusions noted as follows.
Figures 68 and 69 show the average radial profile at combustor exit at a
fuel flow of 88 IbAr when using rig air and fan air supplies, respectively. A com-
parison of the radial profiles is shown in Figure 70.
It is apparent that good repeatability in radial profile is achieved, indicative
of good simulation, aerodynamically (and hence of combustion reactions), between
rig and fan. Allowing for the small difference in air flow between the two tests, the
difference in temperatures recorded is one percent, and so the traverses can be
assumed accurate and representative. The large difference in temperature that exists
between the center and outside of the combustor is as a result of inadequate mixing of
fuel and air. Consequently, there exists large radial variations in air-fuel.
109
-------
HOT SPOT
14(111 Minn IMOO 2000 2200 2400 2fiOO
COMBUSTOR OUTLET TEMPERATURE.
2*00
AVERAGE TEMPERATURE 2384" F
HOT SPOT TEMPERATURE 2920°F
COLD SPOT TEMPERATURE 1458' F
TEMPERATURE SPREAD
FUEL FLOW
COMBUSTOR PRESSURE LOSS
3000
1462"F
88 LB/HR
5.7 INCHES WATER
FIGUBE 68. AVERAGE RADIAL PROFILE OF TEMPERATURE OUT OF COMBUSTOR
USING RIG AIR SUPPLY
5. O
4.0
O
_
en"
s
a
3.0
2.0
u
j
1.0
0.0
COLD SPOT
HOT SPOT
1300 1400 1600 1800 2000 2200 2400 2600
COMBUSTOR OUTLET TEMPERATURE, *F
AVERAGE TEMPERATURE 2335'F
HOT SPOT TEMPERATURE 2380' F
COLD SPOT TEMPERATURE 1315" F
TEMPERATURE SPREAD
FUEL FLOW
COMBUSTOR PRESSURE LOSS
2800
1565' F
88 LB/HR
5.8 INCHES WATER
FIGURE 69. AVERAGE RADIAL PROFILE OF TEMPERATURE OUT OF COMBUSTOR
USING FAN AIR SUPPLY
110
-------
5.0
0.0
1400 KiOO 1800 2000 2200 2400 2600 2800
COMBUSTOR OUTLET TEMPERATURE, °F
FIGURE 70. REPEATABILITY OF RADIAL PROFILE OF TEMPERATURE OUT OF
COMBUSTOR USING BOTH FAN AND RIG AIR SUPPLIES
Theoretically, either a lean or rich mixture will cause an increase in emissions and
so there's some room for considerable improvements to be made.
In addition to the radial variations in temperature (and air-fuel) noted, it is
seen from Figures 71 and 72 that circumferential variations in radial profile exist.
This is shown more clearly in Figures 73 and 74 which show the circumferential
temperature distributions at various radial position for rig and fan air systems,
respectively.
The variations in temperature distribution between rig and fan are significant,
indicating reasonable, but not exact, aerodynamic simulation. The major source of
temperature maldistributions lies, however, in inadequate mixing that, as shown
above, is simulated exactly on rig and fan systems.
Radial traverses were then done at differing fuel flows, holding air-fuel
constant, Figure 75. Changes in profile occurred, indicative of changes in mixing as
a function of fuel flow. In general, however, the basic characteristics of cold center,
hot wall profile was maintained over a wide range of conditions.
Ill
-------
RADIAL POSITION #1
j i I l
2900
2400
1900
1400
900
RADIAL POSITION #2
RADIAL POSITION #3
j 1 l i
RADIAL POSITION #5
2900
2400
£ 1900
U
g 1400
H
U 900
0.
U
^ 2900
H
U
2 2400
U
5 1900
CO
1400
900
RADIAL POSITION #4
RADIAL POSITION #6
2900
2400
1900
1400
900
RADIAL POSITION #8
I
345 012345
RADIUS (INCHES)
FUEL FLOW 88 LB/HR COMBUSTOR PRESSURE DROP 5.7 INCH WATER
FIGURE 71. RADIAL PROFILE OF TEMPERATURE OUT OF COMBUSTOR AT
SEVERAL DIFFERENT CIRCUMFERENTIAL LOCATIONS AND USING
THE RIG AIR SUPPLY
112
-------
RADIAL POSITION #1
2900
2400
1900
1400
900
RADIAL POSITION #2
I I | I
RADIAL POSITION #3
i i i I
RADIAL POSITION #5
2900
2400
1900
3 1400
H
U 900
s
u
H 2900
H
w
J
2 2400
U
S 1900
1400
900
RADIAL POSITION #4
I i i i
RADIAL POSITION #6
RADIAL POSITION #7
2900
2400
1900
1400
900
RADIAL POSITION #8
012345 012345
RADIUS (INCHES)
FUEL FLOW 88 LB/HR COMBUSTOR PRESSURE DROP 5.8 INCH.WATER
FIGURE 72. RADIAL PROFILE OF TEMPERATURE OUT OF THE COMBUSTOR AT
SEVERAL DIFFERENT CIRCUMFERENTIAL LOCATIONS AND USING
THE FAN AIR SUPPLY
113
-------
3000
2800
2600
2400
AVERAGE
TEMPERATURE
°F
2662
(4.75 R)
3000 r
2743
2283
1857
1600
2000 r
(2.00 R)
1800
1000
1400
" ^^ ^^ T
^*» ,-— •— * ^
i i i i i i _j
12 3456 78
CIRCUMFERENTIAL POSITION (1.25 R)
1704
FUEL FLOW. 88 LB/HR
COMBUSTOR PRESSURE LOSS, 5. 7 INCHES WATER
AVERAGE TEMPERATURE. 2384 "F
HOT SPOT TEMPERATURE, 2920° F
COLD SPOT TEMPERATURE, 1458'F
FIGURE 73. CIRCUMFERENTIAL VARIATION OF COMBUSTOR OUTLET TEMPER-
ATURE AT DIFFERENT RADII USING RIG AIR SUPPLIES
114
-------
3000
2800
2600
2400
3000
2800
^ 2600
o
« 2400
< 2400
c
u
a.
S 2200
u
f-
£ 2000
_)
g 1800
2000
uj
o
u
1800
1600
1400
1800
1600
1400
1200
12 34 56
CIRCUMFERENTIAL POSITION
FUEL FLOW, 88 LB/HR
COMBUSTOR PRESSURE LOSS, 5.8 INCH WATER
AVERAGE TEMPERATURE, 2335'F
(4.75 R)
(4.00 R)
(3.00 R)
(2.00 R)
AVERAGE
TEMPERATURE
OF
2700
2725
2190
1750
1555
(1.25 R)
HOT SPOT TEMPERATURE, 2880° F
COLD SPOT TEMPERATURE, 1315°F
FIGURE 74. CIRCUMFERENTIAL VARIATION OF COMBUSTOR OUTLET TEMPER-
ATURE AT DIFFERENT RADII USING FAN AIR SUPPLY
115
-------
X
z
D
5
4.0
2.0
f
CQ
O 1.0
u
o
27 LB/HR
50 LB/HR
109 LB/HR
88 LB/HR
-V:
j
0 1400 1GOO 1800 2000 2200 2400
COMBUSTOR OUTLET
2600
2800 3000
Fuel Flow
Lb/Hr
27
50
88
109
Combustor Loss
Inch Water
0.515
2.05
5.7
8.4
Temperature Spread
8 F
1130
1510
1210
1060
FIGURE 75. RADIAL PROFILE AT VARIOUS FUEL FLOWS, USING RIG
AIR SUPPLY
7.3.6 Final Combustor Tests
Because of the large radial temperature profile adjustments were made in
the position and number of dilution holes. Considerable modifications to the shape of
the radial profile could be made and the final result is shown in Figure 76. A reduction
in radial profile spread from about 1300° F to 850° F was made. In practice, by more
extensive experimentation, it should be possible to obtain a flat temperature profile,
especially if vaporizer inlet temperature retirements can be kept low.
116
-------
5.0
4.0
u
3.0
K 2.0
O
w
pa
S 1.0
o
o
FIGUBE 76.
\
HEF. FIGURE 68
PRELIMINARY DESIGN
FINAL
DEMONSTRATION
ARRANGEMENT
2(JOO
2800
1400 1600 1800 2000 2200 2400
COMBUSTOR OUTLET TEMPERATURE (° F)
RADIAL PROFILE OF TEMPERATURE (FINAL DEMONSTRATION COM-
PARED TO PRELIMINARY)
Based on previous emission tests the fuel and air metering valves were
adjusted to provide for as optimum as possible air-fuel ratio. The resultant combus-
tor outlet temperature as a function of fuel flow is shown in Figure 77. Because the
dilution air represents about 20 percent of the total air flow through the combustor this
temperature could be increased by a similar amount (to about 3000° F maximum) by its
deletion from the combustor. Conversely the temperature could be reduced by any
amount by addition of more dilution air. Theoretically either addition or subtraction
of dilution should not affect emissions. However, if temperatures of 3000° F are re-
quired to minimize vaporizer size, temperature distributions will be dependent on
mixing in the flame zone and there would be no excess dilution air available for final
trim of the radial profile.
Emission tests were then run and the results are shown in Figures 78, 79,
and 80, and also in Figure 81.
All emissions were substantially below the required limits except for carbon
monoxide which, at the one pound an hour fuel flow, was above limits. This was
caused by excessive air addition (Fig. 77) and, by small modifications to the fuel and
air controls it should be possible to achieve the required emission goals.
117
-------
2300
2100
1*00
1700
1SOO
1300
900
700
20
40 50 60
FUEL FLOW (LB/MRI
90
100 109
FIGURE 77. FINAL AIR-FUEL CONTROL SYSTEM. DISCHARGE TEMPERATURE
VARIATION WITH FUEL FLOW
UJ
o>
1.20
1.00
0.80
0.60
0.40
0.20
138 NOg LIMIT
0 10 20 30 40 50 60 70 80 90 100 109
POUNDS PER HOUR FUEL FLOW
FIGURE 78. EMISSION OF NO2 AS A FUNCTION OF FUEL FLOW
118
-------
271
16
14
12
10
e8
|e
i
8
16.25 CO LIMIT
0 10 20 30 40 50 60 70 80 90 100 109
POUNDS PER HOUR FUEL FLOW
FIGURE 79. EMISSIONS OF CO AS A FUNCTION OF FUEL FLOW (TEST A)
0.40
0.30
Q20
0,10
0.48 HC LIMIT
0 10 20 30 40 50 60 70 80 90 100 109
POUNDS PER HOUR FUEL FLOW
FIGURE 80. EMISSIONS OF HC AS A FUNCTION OF FUEL FLOW (TEST A)
119
-------
Fuel Flow pph .
CO
Test A HC
•NOz
CO
Test B HC
NOj
CO
Ted C HC
NOj
(HC not adjusted
for background)
CO
Test D HC
NO,
109
10.4
0.053
0.616
6.17
0.400
1.818
1.41
0,168
1.452
0.322
2 iS
0 053
1.702
EMISSION AS A FUNCTION OF FUEL FLO V IN GM/KGM
100
4.74
0.044
0.754
5.94
0.364
2.100
1.13
0.089
1.542
0.239
2 60
0 048
1.530
90
2.74
0.039
0.978
7.29
0.250
1.921
1.07
0.129
1.528
0.272
0.69
0.048
1.930
80
2.69
0.036
0.872
3.82
0.241
1.840
0.55
0.056
1.5CO
0.197
0.13
0.055
1.702
70
2.80
0.055
0.851
2.54
0.241
1.928
0.14
0.044
0.430
0.191
0 00
0 060
1.708
60
2.75
0.054
0.835
1.79
0.234
1.912
0.07
0.029
0.330
0.172
0 00
0 055
1.640
50
1.34
0.042
0.854
1.21
0.234
1.913
0.07
0.014
1.196
0.156
0.00
0.059
1.540
40
0.99
0.036
0.874
1.02
0.233
1.910
0.07
0.014
1.065
0.158
0.00
0.062
1.511
30
0.91
0.032
0.795
0.09
0.248
1.840
0.00
0.014
1.060
0.157
0.00
0.055
1.388
20
0.86
0.029
0.672
0.10
0.263
2.010
0.08
0.000
1.093
0.123
0 00
0.061
1.351
15
0.79
0.025
0.734
0.09
0.253
1.836
0.09
0.000
1.158
0.112
0 00
0.063
1.270
10
1.07
0.019
0.661
0.10
0.261
1.808
0.10
0.000
1.118
0.116
0 00
0 070
1.172
5
4.33
0.026
1.105
0.14
0.308
2.350
0.20
0.000
1.208
0.184
0 93
0 110
1.831
2
7.80
0.030
1.140
0.21
0.362
2.910
0.27
0.000
1.468
0.218
5 81
0.202
1.108
1
27.10
0.191
1.300
3.04
0.283
2.650
0.48
0.000
1.970
0.330
3 66
0.2S2
1.430
Fuel
JV 5
JP 5
JP 4
JP 5
Test A - Standard configuration, JP 5 fuel maximum heat losses to ambient. • Measured as NO and Reported NOj
Test B - A« with Test A but with minimum possible heat losses to ambient. (Appendix D)
Test C - Primary flame zone leaned out 25%, JP 4 fuel, minimum possible heat losses.
Test D - As with Test C but with JP S fuel.
FIGURE 81. EMISSIONS AT VARIOUS FUEL FLOWS, WITH DIFFERENT FUELS
AND AT TWO AIR-FUEL RATIOS, WITH AND WITHOUT HEAT LOSSES
These tests were done by taking emission samples immediately aft of the
combustor exit. Heat losses were substantial, especially at low fuel flows, and would
approximate to the heat losses that would occur if a vaporizer had been installed at
this emission sampling point.
A long mixing duct was installed aft of the combustor, of ten inches length.
This hot duct reduced the view and hence radient losses of the flame to ambient. The
resultant emissions are shown in Figure 81 Test B. There was a substantial increase
(approximately double) in NOg emissions and a reduction in CO emissions as would be
expected. The HC emissions increased and this latter phenomena was unexpected
and unexplainable. Further investigations are warranted.
Keeping the same overall air-fuel as previously, modifications were made to
the primary flame zone such as to increase its air-fuel by 25 percent. The radiation
shielding was retained and the results shown in Figure 81, Test D. The results
were a very substantial reduction in CO and HC emissions such that, for most of the
operational range that would ordinarily be used in city driving, these emissions are
negligibly small. Emissions of NO2 reduced slightly, and thus indicates that a more
optimum arrangement of air-fuels in the various zones of combustion exists which
could further reduce emissions.
Maintaining the same lean primary flame zone and radiation shielding the test
was repeated using JP-4 and the results are in Figure 81, Test C. Changes in
emissions were noted, of a small amount. At some fuel flows emissions were higher,
at other flows, less than when JP-5 fuel was used, but in general, emissions with
120
-------
jp_4 were slightly less than with JP-5. Also shown are the HC emissions not corrected
for background. (At the test site background HC emissions were high due to the
proximity of the municipal airport.) The HC emission levels recorded indicate that the
combustion process serves to lower these emissions to less than background.
The results indicate that emissions can be kept below the goals set, but that
the design of the vaporizer must be carefully integrated with the combustor in order to
avoid excessive heat loss and resultant high emissions.
7.3.7 Combustor Noise
Certain operating conditions of combined Rankine Cycle fan and combustor
result in an acoustic resonance. A combustor modification solved the resonance
problem, but unfortunately had adverse effects on emission levels. Sound level
frequency analysis of "before-and-after-fix" runs are shown in Figure 82. The
modification reduced maximum combustor frequency peak from 107 db (sound
pressure level, Ref. 20nN/nO to 94 db at 109 Ib/hr fuel flow (full power). Figure 83
distinguishes noise components of the combustor at an off-exit location.
The test conditions were:
•Fan inlet guide vanes: installed
• Microphone position and orientation: (1) on fan axis three feet in front
of inlet at 45 degrees to axis for runs shown in Figure 82. (2) Data in
Figure 83 is taken 12.5 feet radially from combustor exit, five feet from
ground, 45 degrees orientation. (See Fig. 84).
•Surroundings: Outside, ten feet from brick wall (See Fig. 84).
• Equipment Same as described for fan noise tests (Section 6).
•Background noise: Figure 82 indicates noise levels with only fan power
generator operating. Figure 83 includes operational fan as part of
background noise so that combustor components can be separated.
Table V summarizes the significant frequency peaks of the test series.
Summary of Results
1. Before modification, combustor resonance shows a frequency shift
from 240 cps to 310 cps with a change of fuel flow from 20 Ib/hr to
109.5 Ib/hr. The sound level remains approximately the same at 106
to 107 db.
121
-------
110
100
J 90
u
s
o
8 80
BURNER RIG OPERATIONAL
BEFORE MODIFICATION
SO
BACKGBOUND (INCLUDES MOTOR GENERATOR SET)
_L
J L
_L
J 1 1 1 1 1
80 TO 80 90 100
nor
100
ISO
200 250 300
400 SOO 600 TOO 800 900 1000
FREQUENCY (opt)
BURNER RIG OPERATIONAL
AFTER MODIFICATION
60
sol 1 L
BACKGROUND
(INCLUDES MOTOR GENERATOR SET)
60 TO 80 90 100
ISO 200 ISO 300
FREQUENCY (cp*)
400 500 600 TOO 800 900 1000
FIGURE 82. BEFORE AND AFTER MODIFICATION BURNER NOISE AT LOCATION
"A"
122
-------
COMBUCTOR. FAN AND BACKGROUND
to
OS
\ BLOWER AND BACKGROUND
SOt
250 300 400 500
600 700 800 BOO 1000 1500
FREQUENCY (cpa)
2000 1500 3000 4000 5000 6000 TOM 8000900010.000
7 8 9 10
150 ZOO 150
FIGURE 83. IDENTIFICATION OF COMBUSTOR NOISE AT LOCATION "B"
-------
ONE STORY BRICK STRUCTURE
ASPHALT GROUND COVER
FAN LOCATION BEFORE
COMBUSTOH MODIFICATION
EXHAUST
RANKINE CYCLE
BURNER
FAN LOCATION AFTER V
COMBUSTOR MODIFICATION
LOCATION "A"
MOTOR GENERATOR
SET
12.5'
OUTSIDE LOCATION
LOCATION "B"
WOOD CABINET
FIGURE 84. DIAGRAM OF COMBUSTOR ACOUSTIC TEST LOCATION
TABLE V
SIGNIFICANT NOISE FREQUENCY PEAK LEVELS
Frequency
(cps)
90
240
310
350
900
1750
2700
Before Combustor Modification
After Modification
Inlet Axis Location "A"
20 Ib/hr Fuel Flow
Off Exit Location "B"
109. 5 Ib/hr Fuel Flow
Fan and Combustor
None
106 db
No Peak
94 db
100 db
90 db
None
None
95.5 db
107 db
94 db
101 db
90 db
98 db
94 db
93 db
91 db
94 db
104 db
88 db
94 db
89 db
None
92 db
93 db
88 db
89 db
None
Fan Only
None
None
None
None
88 db
87 db
78 db
Source
Combustor
Fan
and
Harmonics
124
-------
2. The combustor modification eliminates the combustor as the dominant
noise source as measured at location "A". A new 94 db peak at 90 cps
occurs after the change and the 107 db peak at 310 cps is significantly
reduced to 91 db.
3. After modification with fan and combustor operating a measurement
to be 99 db at location "B". (Measured in the all frequency pass mode,
flat frequency response mode-weighting contours not used.)
(It should be noted that the change in the fundamental fan peak level
at 900 cps from 101 db to 104 db before and after combustor modification
is due to a relocation of fan position. See Figure 84 for details.)
7.3.8 Ignition Tests
A prime source of hydrocarbon emissions is cold light off. Ignition must
be instantaneous otherwise resultant emissions are unacceptable. Throughout all
the tests ignition procedures were monitored and, at all times, good ignition occurred
and no failures to light were seen. Tests were done, simulating cold day starts at
-40° F with kerosene,by using a heavy, less volatile fuel of 17 centistokes viscosity
(the highest viscosity likely). Good ignition could be obtained and the flame perfor-
mance at maximum fueling rate appeared similar to that obtained with JP-5 (no
emission data was taken). However, ignition reliability was only 50 percent and it is
suspected that a reduction in fuel flow at ignition (nominally 10 Ib per hour of fuel) due
to viscous effects on the fuel control system was the cause of this. At high fuel flows
this viscous effect is reduced (it is a function of Reynolds number) and fuel flow was
satisfactory. Some further investigations are indicated.
7. 3. 9 Aldehydes and Smoke
No instrumentation was available to measure either aldehydes or smoke.
Throughout the tests reliance had therefore to be placed on sight and smell to assess
these emissions. In steady state operation the flame was odorless and smoke free.
Indicating that aldehydes and smoke were at a very low level or perhaps nonexistant.
During transient decelerative operations, at low levels of fuel flows it was seen that
some smoke was emitted though emissions of CO, NC>2 and HC were maintained low.
The level of smoke was a direct function of deceleration and it was therefore a likely
result of differing response characteristics of the fuel and air flow. The problem was
minimized by controlling the deceleration rate in transient tests as a function of fuel
flow. At high flow rates deceleration was maintained at a higher rate than at low flows.
This fits in well with the response requirements of the system in that vaporizer response
at high fuel rates must be quicker than at low fuel rates. Future work must involve
complete package testing if minimum smoke is to be obtained and synchronization of
the air and fuel metering systems must be improved.
125
-------
8
OPTIMUM DESIGN APPROACH FOR RANKINE CYCLE
COMBUSTION SYSTEM
Based upon the results of the analysis, design and test program conducted on
the demonstration system a number of specific conclusions may be formulated:
• The 1980 AAPS emission levels for automotive vehicles are feasible.
• A relatively small combustion system incorporating hydromechanical
control components can operate at high response across a fully mod-
ulated heat release rate of 100 to 1 (2 x 106 to 2 x 104 BTU/hr).
•Emissions during the required high frequency automotive startup,
power level transients and shutdown cycles can be maintained below
emission level goals (start up in 3 seconds, combustor only, transient
response 50%/sec).
•Air flow changes resulting from voltage, leakage and efficiency changes
can be compensated to maintain correct air-fuel ratios for low emissions.
•Heavy fuels such as Diesel No. 1, Jet A, Kerosene or JP-5 can be
burned across the full flow range.
•Parasitic power can be minimized at part loads by the fuel delta-P
compensator.
A set of design factors determined to be of major importance in the develop-
ment of a low emission combustion system have also been identified. Listed below
are some of the major design guides determined by analysis and test results.
•Precise control of combustion in three axial zones with rapid time
transitions between primary and secondary zones is necessary.
•Mixing must be rapid and uniform.
127
-------
• Overall air-fuel ratio must be precisely controlled across the entire
heat release range.
• Uniform air distribution into the air valve and combustor is essential
for low emission combustion.
• Flame radiation losses are important factors in emission levels.
•High response of both air and fuel flow control is necessary throughout
transients. Fan speed control is probably inadequate due to inertia!
lags.
.Air valve appears to require flow symetry to obtain 100 to 1 flow control.
• Fan design and configuration'has a major effect upon the air valve design
and combustor performance.
.-Vaporizer effects on aerodynamics and radiation flame radiation heat
loss may also have significant effects upon the design and emission
performances of the combustion system.
An optimum system can be synthesized based upon results of the demonstration
system tests. A major factor in determining the package configuration of the demon-
stration system was component availability and the need for design flexibility in this
highly developmental program. Although the best fan found to be readily available
has low volume (0.2 cubic feet) and low weight (13 pounds), it is not optimum. Initially,
it was designed for aircraft applications. Weight and volume are emphasized in these
designs with noise only as a secondary consideration. Thus, the unit is a small diameter,
high speed fan having a high dynamic head. In order to obtain accurate air regulation
across the 100 to 1 range, the adverse aerodynamic effects of the dynamic head must
be eliminated. To do this, a relatively large volume cavity at the discharge side of
the fan 10 needed to turn the flow and reduce its velocity. Another design feature that
would also be eliminated in a production type unit would be a separate drive motor for
1he rotating cup atomization system. Its use in the demonstration unit is necessary
due to rotational speed requirements and need for initial independent optimization of
the atomization system. It also added to the length (~4 inches) and overall volume of
the complete system.
The package design depends critically on the overall pressure drop through
the system. The use of a low boiler pressure drop results in a different system than
with a high boiler pressure drop. The effect on the combustor and control design
philosophy is slight and it appears likely that the best system with either high or low
loss boiler will incorporate a fan of larger diameter than currently used and which
128
-------
will result in considerable improvements in performance and substantial reductions
in length. The presence of a vapor generator could modify flame performance
depending upon its design. By proper integration of the vapor generator, flame per-
formance may not be effected and may even be improved.
Several significant areas for improvement can be identified by matching fan
size and speed to the geometry and flow requirements of the air valve and combustion
system. Optimizing the system will reduce volume, power and noise significantly.
An optimum configuration of this type is compared to the demonstration system in
Figure 85.
The optimum system considered is based on slight modifications to the
demonstration system design. Basically a better aerodynamic interface is the key
element in the optimum configuration. Lower turning and metering losses are
expected together with a corresponding reduction of parasitic power. A small
reduction in combustor pressure drop also appears quite feasible by increasing the
outside diameter of the basic combustor. As a result of use of a large diameter fan
and a combined fan motor rotating cup drive, the overall length and volume can be
reduced to as low as 14 inches. Parasitic power loss at full load could be as low as
1. 25 HP with part load parasitic power drain well under 0. 5 HP if voltage is reduced
at lower power demands.
129
-------
f
.1
\
/
l.ll! FT'1
COMBUSTOR
PLl'S CASE
V CUP
' MOTOR
12.
[*- T ^
1 \
FAN
MOTOR
». . Ili FT'1
AIR VALVE
1 /
1
Jr
1.17 F
1
FAN
• FAN T1LADE
FAN STATOH
-CONTROL SECTION
PANCAKE
DC MOTOR
FUEL INLET
TOTAL VOLUME 1. R9 FT3
DEMONSTRATION SYSTEM
Cl'T>
13.11- DIA
14.2.1
OPTIMUM SYSTEM
3
Total Volume (ft )
Length (Inch)
Diameter (Inch)
3
Combustor Volume (ft )
Combustor Diameter (Inch)
Horsepower
Combustor Loss (Inch Water)
Diffuser Loss (Inch Water)
Metering Loss (Inch Water)
Overall Pressure Loss (Inch Water)
Fan Efficiency (%)
Motor Efficiency (%)
Overall Efficiency (%)
Motor Speed (rpm)
Demonstration
System
Design
1.89
31.0
13.00
0.687
11.00
2.30
8.00
3.00
2.00
13.00
75.0
75.0
56.2
14000
Optimum
System
Design
1.09
14.25
13. 00
0.687
12.00
1.25
6.0
0.5
1.5
8.0
85.0
75.0
68.0
7000
Spec
1.33
2.00
Note: Ignition, Fuel and Air Regulators not included.
FIGURE 85. TWO FAN SKETCHES - PRESENT AND OPTIMUM
130
-------
APPENDIX A
-------
APPENDIX A
EMMISSION MONITORING EQUIPMENT AND PROCEDURES
Emission measurements were taken using three Beckman Model 315A infrared
analyzers, Figure A-l, and a Beckman Model 402 hydrocarbon analyzer, Figure A-2.
These instruments provided continuous and automatic determination of the exhaust
components.
The infrared analysis system is based on a differential measurement of the
absorption of infrared energy. An infrared radiation source is transmitted through two
long cells one containing the exhaust sample and the other a reference gas. In oper-
ation, the presence of the infrared absorbing component of interest in the sample
stream causes a difference in the radiation absorption levels between the sample
and reference sides of the system. Due to this difference the gas in the reference cell
is heated more, thus raising the pressure and causing a metal diaphragm to distend.
This metal diaphragm is part of a capacitor circuit, as the IR source beams are
alternately blocked and unblocked, it pulses, thus causing a cyclic change in the
detector capacitance (Luft Principle). The resultant signal is then routed to the am-
plifier control section, and finally to the recorder.
The hydrocarbon sensor is a burner where a regulated flow of sample gas
passes through a flame sustained by regulated flow of a fuel gas and air. Within the
flame, the hydrocarbon components of the sample stream undergo a complex ionization
that produces electrons and positive ions. Polarized electrodes collect these ions,
causing current to flow through measuring circuitry located in the electronics unit.
The ionization current is proportional to the rate at which carbon atoms enter the
burner and is therefore a measure of the concentration of hydrocarbons in the
original sample.
The following conditions were employed for all analysis:
Gas Detection Range
NO IR-41" cell 0-150 ppm
CO IR-10" cell 0-1000 ppm
CO2 IR-0.25" cell 0-16%
Hydrocarbons FID 0-500 ppm, 0-10 ppm
133
-------
* • ' J « > t i t » »""
trie Oxid
Inch Cell
Electronic Units
Carbon MonoxideBCarbon Dioxide
2.5 Inch CeUl»).25 Inch Cell
10.0 Inch Cell
FIGURE A-l. BECKMAN MODEL 315A INFRARED ANALYZER
134
-------
cc
01
FIGURE A-2. BECKMAN MODEL 402 HYDROCARBON ANALYZER
-------
Each instrument was calibrated and maintained per manufacturer's recomm-
endations. Calibration curves supplied by Beckman were cross checked each day with
known gases to determine the validity of these curves. The NDIR unit was kept on at
all times in order to permit maximum stabilization. No difficulties or large variances
were encountered in maintaining day-to-day gain settings, and zero settings.
Certified calibration gases as received from the vendor were within ±0.5
percent of the stated values with the exception of the nitric oxide standards. Nitric
oxide calibration gases have an analytical accuracy of ±5.0 percent. Each NO stand-
ard gas was analyzed by the modified Saltzman* method employing the evacuated
bottle sampling technique.
Table A-I shows the nitric oxide values obtained by the Saltzman method as
compared to the vendor's analysis. (Average of a minimum of 5 determinations.)
TABLE A-I
NDIR COMPARED TO SALTZMAN
Stated Value
140
105
53
Infrared
100 ppm
45 ppm
Saltzman
144 ppm
118 ppm
47 ppm
The flame ionization hydrocarbon analyzer was calibrated immediately prior
to each run and set up according to the manufacturer's specifications. Eight ranges
are available on the instrument varying from 0-10 ppm to 0-50, 000 ppm. All deter-
minations required the lower two ranges. The heated sample line was maintained at
350° F and the detector at 400° F to avoid condensation problems.
Intensive investigations were performed at the start of the program to
evaluate the sampling technique. The following facts were observed:
•Each analyzer unit required its own probe. Attempts to tee off from
one probe showed adverse flow effects.
* Cornelius, W., and Wade, W. R., "The Formation and Control of NO in a Regen-
erative Gas Turbine Burner", SAE Report No. 700708.
136
-------
• A filter (glass-wool)-dryer (Drierite-nonindicating) was used between
the NDIR and the probe.
• Sample line length (20-200 ft) had little if any effects on the measure-
ments, other than flowrate.
• The use of the pre-drier, plus the refrigerator and additional NO drier
were more than adequate to remove all the water. (Aquasorb was used
just prior to the NO cell).
• An air-quenched averaging probe was determined suitable for the
combustor tail pipe.
The accuracy of the NDffi is dependent upon the accuracy of the calibration
gases used. Each of the calibration gases were analyzed by the supplier using "gold"
standards. These primary standards were prepared using National Bureau of Stand-
ards weights.
Reproducibility of the instrument is within one percent. Reproducibility
tests were performed on separate days and under varying sampling conditions. The
results showed that the measurements were reproducible.
TABLE A-II
REPRODUCIBILITY TEST
Power
Setting
10-28-29
50%
75%
1
<
2|
f
>%
10-29-70
50%
'1
>%
25%
ppm NO
56
61
60
60
56
59
64
55
47
47
59
47
62
56
49
47
ppm CO
128
120
120
128
120
120
282
305
53
53
120
143
295
335
45
38
% C02
7.9
7.9
7.8
7.9
7.7
7.7
7.7
7.9
6.8
6.5
7.5
7.9
7.8
7.9
6.0
6.5
3
Flow ft /m.
4
2
1
4
2
3
3
3
3
3
3
3
3
3
3
3
Remarks
Probe Preheated
~200' Sample Line
Probe Not Heated
Probe Heated
*200' Sample Line
Probe Heated
Probe Unheated
Probe Heated
Probe Unheated
Probe Heated
Probe Unheated
Probe Heated
Probe Unheated
137
-------
Because of the very low emission levels obtained with this combustor (com-
pared to Otto cycle engines), special attention was required to calibrate the measure-
ment system at the low ranges. Calibration gases in the actual ranges being recorded
were obtained to establish the validity of the instrument deflection to volumetric
concentration supplied by Beckman. Gases used for calibration were certified by
the span gas suppliers to be within the following concentrations:
Nitric oxide
Concentration Certified Accuracy
10.5 ± 0.5 ppm ± 5. 0% of component
103.0±2.1ppm ± 2. 0% of component
140.0 ±2.8 ppm . ± 2.0% of component
Carbon dioxide
Concentration Certified Accuracy
14.31 ± 0.29% ± 2% of component
4.31 ± 0. 09% ± 2% of component
2.10 ± 0. 04% ± 2% of component
1.12 ± 0. 02% ± 2% of component
Carbon monoxide
Concentration Certified Accuracy
115 ppm ± 2 ppm ± 2% of component
225 ppm ± 4 ppm ± 2% of component
850 ppm ± 17 ppm ± 2% of component
(Recorder deflection error is ± 0. 5% full scale.)
Results of calibration tests are shown plotted in Figures A-3, A-4, and
A-5. The basic curve was supplied by Beckman. Data points used by Beckman to
verify the basic curve slope are shown as open circles. Solar calibration points
including the zero span gas point are shown with the closed circles. Good correlation
was obtained throughout the range required to measure emissions with the Rankine
cycle combustor.
138
-------
10(1 r
so
z
O
p GO
u
u
u.
u
c
• SOLAR CALIBRATION POINTS
O IJKCKMAN SUPPLIED CALIBRA-
TION POINTS
Q ZERO GAS POINT
4(1
.10 100
PPM NO IN N BY VOLUME
FIGURE A-3. NO CALffiRATION RESULTS
100
80
• SOLAR CALIBRATION POINTS
O BECKMAN SUPPLIED CALIBRATION POINTS
D 7.ERO GAS POINT
60
U
U
J
u.
u
D
C
U
c
s.
40
50
100 150 200
PPM CO IN N0 BY VOLUME
250
FIGURE A-4. CO CALIBRATION RESULTS
139
-------
mo r
• SOLAR CALIBRATION POINTS
O BECKMAN SUPPLIED CALIBRATION POINTS
g ZERO CAS POINT
4.0 8.0 12.0 l(j.O
PERCENT CO, IN NO BY VOLUME
FIGURE A-5. CO2 CALIBRATION RESULTS
140
-------
APPENDIX B
-------
APPENDIX B
TEST FUEL SPECIFICATIONS
The fuels specified to be used for combustion tests are #1 Diesel, kerosene
and Jet A. These fuels can vary widely in their combustion characteristics because
their specifications are loose. Refineries in differing locations can produce different
fuels, meeting the same spec, because their production is guided by economics dictated
by local conditions of demand and the type of crude that is available.
Solar is using JP-5, which is a more rigorously controlled fuel than those
specified by EPA. It is a fuel that typically will be more difficult to burn than those
specified. In addition, JP-4 and #2 diesel will be used to provide information with
fuels that are respectively considerably more and less difficult to burn than those
specified (Fig. B-l).
The Government has recognized this problem of fuel variability in regard to
the testing of diesel engines* and has proposed a fuel spec for #1 diesel as follows:
Specification
Fuel Property
Cetane
Distillation ° F
Initial Boiling Point
10%
50%
90%
End Point
Gravity °API
Total Sulfur %
Aromatics %
Paraffins, Napthenes, Olefins
Flash Point °F (min.)
Viscosity (Centistokes), at 100° F
ASTM D-975
Grade 1-D
40 min.
550 max.
0. 5 max.
100 or legal
1.4-2.5
P roposed
Government Spec
48-54
330-390
370-430
410-480
460-520
500-560
40-44
.05-. 2
8-15
Remainder
120
1.6-2.0
MILrT-5624
Grade JP-5
400 min.
550 max.
140
* Federal Register, Volume 35, Number 136, Part II.
143
-------
700
HOO
.100
H
PS
D
H
U
Pk
w
H
400
200
100
OF
SPl
SPECIFIEI
JET A Al
BAD IN DISTILLATION
FUELS (KEIOSENE,
#1 DIESEL)
20 40 60
PERCENT EVAPORATED
80
100
FIGURE B-l.
VARIATION OF ASTM DISTILLATION TEMPERATURES
FOR THE TEST FUELS (AVERAGE VALUES)
144
-------
The proposed fuel is more rigorously controlled than current ASTM specs, agrees
closely to the JP-5 fuel used by Solar and is representative of a typical grade of kero-
sene that could ordinarily be available. As the combustion characteristics of a diesel
engine are different from those of an atmospheric combustor, it is likely that the
specification above should be modified so as to highlight any combustion difficulties
inherent in an atmospheric flame. Items that warrant more specific control are noted
below with the reasons for such control
•Flash Point, °F, 140°F minimum, 150°F maximum
A high flash point makes ignition more difficult and therefore should
be controlled to the high end of typical practice. Current specs for
fuels used have a maximum of 150° F but no bottom limit.
•End Point, °F, 550° F minimum, 600° F maximum
A high end point tends to carbon build and smoke in exhaust. Current
specs allow 575 ° F maximum but a bottom limit is not specified.
•Aromatics, 20 to 25%
High aromatics tend to cause smoke and carbon build. No specs
currently apply and typically are lower than shown.
•Viscosity, 2.0 to 2.5 centistokes
High viscosity makes control of fuel more difficult and makes for
problems of fuel atomization with certain types of combustion systems.
Typical values are less than shown and the range is 1.4 to 2.5.
•Olefins and Diolefins, 5 to 10%
High diolefins tend to cause deterioration of fuel with resultant gum
formation. This causes unreliability in fuel metering and fuel injection.
No specs exist and typical values are below 5%.
Tests have indicated that most of the problems seen with burning a heavy
grade of fuel such as JP-5 will tend to diminish with the lighter grades of kerosene.
The current specifications of the fuels to be used and of JP-5 are as
follows:
145
-------
Flash Point
10% Point
50% Point
End Point
Jet A
(ASTM D1655)
110-150°F
400°F Max.
450°F Max.
550° F Max.
Kerosene
(VV-K-211)
115° F Min.
572°F Max.
Kerosene
(VV-K-220)
125°F Max.
510°F Max.
JP5
(MIL-T-5624)
140° F Min.
400°F Min.
550°F Min.
Of these fuels, Jet A and JP5 are the most tightly controlled. JP5 will inevitably be
heavier than Jet A as flash point and 10 per cent point tend to be higher.
The specifications for the fuels to be used to monitor the limits of typical
fuels are as follows:
JP4
#2 Diesel (ASTM D975) (MIL-T-5624)
Flash Point 125°F or Legal 20% Point 290°F Max.
90% Point 540°F min., 640°F max. 50% Point 370°F Max.
90% Point 470°F Max.
146
-------
APPENDIX C
-------
APPENDIX C
FAN NOISE REDUCTION METHODS
The National Air Pollution Control Administration Division of Motor Vehicle
Research and Development has recently established a goal of 77 dbA as the maximum
noise generated by the vehicle. Either the combustor or compressor fan are likely to
be the critical noise source in the Rankine cycle vehicle. Thus, a detail analysis of
the fan and its noise generation mechanisms is necessary. Results of noise measure-
ments on the fan currently being used (because of availability) are analyzed in Sections
6 and 7. These results in general indicate that significant noise reduction would be
necessary in a production system. An optimum low noise fan could be designed also
to provide better aerodynamic and mechanical matching to the air valve and combus-
tor. A design analysis indicates that this could be accomplished by incorporating the
following changes:
• Increased diameter (up to 13 inches O. D.)
• Decreased rotational speed
• Decreased pressure rise
• Larger number of blades
• High vane to rotor blade ratios
• Increased spacing between blades and rotors
An investigation into the mechanism and relative noise improvements that
can be expected has been initiated. Recent advances in aerospace acoustic analysis
technology have made the optimum design approach method relatively clear. Up to
the recent work in acoustics on turbo fan engines, the best empirical method to pre-
dict noise of fans was:
PWL = 90+10 log HP + 10 log Ps (1)
or PWL - 55 + 10 log Q + 20 log Ps (2)
or PWL = 125 + 20 log HP - 10 log Q (3)
where PWL - db - overall sound power level re 10
HP - hp - rated motor horsepower
Q - cfm - fan discharge flow
149
-------
P -in. HO - fan total to static pressure rise *
S £
These equations indicate the importance of maintaining as low a pressure rise
as possible. A more comprehensive approach to axial vane fans has recently been
published by M. J. Benzakein and S.B. Kazin**. The following important design infor-
mation has been abstracted from this work.
A study of various fan/compressor noise reduction methods is presented.
The analytical treatment of the basic mechanisms of fan/compressor noise generation
is described. The results are presented in parametric form and indicate the effects
of fan/compressor design, number of blades, vane/blade ratio, aerodynamic para-
meters, and blade row spacing on pure tone noise reduction. These results are based
on non-steady aerodynamic treatment of wake and potential interaction effects and
theoretical extensions of spinning mode theories.
A listener in the vicinity of an axial-flow fan may distinguish two distinct
sound components known as "discrete-frequency noise" and "broadband noise". The
first component consists of a number of pure, or nearly pure, tones which combine to
form a high frequency noise, best described as a whine. The second component is a
background hissing noise, caused by a superposition of sounds over a continuous band
of frequencies from the lower audible range to the higher, and without pronounced
peaks at any particular frequencies.
The relative importance of the two noise components depends on the type of
fan or compressor. Noise from a many-bladed fan, working at subsonic tip speed in
an unobstructed airflow, has broadband characteristics. Noise from a high speed
propeller has predominantly discrete-frequency characteristics. From a compressor
in which the rotor interacts with stators, or a fan with bearing support struts or other
obstacles near the rotor face, the noise is a mixture of the two components. General-
ly, however, the discrete-frequency noise dominates the frequency spectrum. The
object of this paper is to present different methods by which the blade passing freq-
uency tones can be reduced.
BASIC NOISE GENERATION MECHANISMS
The complexity of the fan/compressor noise generation phenomena lead many
researchers to a largely empirical approach to the problem. Some broad understand-
ing of the various sound sources has been derived from experimental data in the last
* Beranek, L. L., "Noise Reduction", McGraw Hill, New York, 1960.
** ASME 69-GT-9 Fan Compressor Noise Reduction, March 1969, M. J. Benzakein and
S. B. Kazin.
150
-------
ten years. It is felt, however, that a basic knowledge of the different noise sources
is indispensable if fan/compressor noise is to be reduced at the source.
Different mechanisms are involved in the pure tone generation in fans and
compressors. These different mechanisms vary in importance from configuration to
configuration, and in a design, from speed to speed. Each particular mechanism can
become the main noise contributor for a particular fan design at a particular speed.
The major mechanisms are defined later.
Rotor Alone Noise
"Rotor alone noise" arises from the pressure field that surrounds each blade
as a consequence of its motion. In a moving blade, the pressure distribution on each
section along the blade span produces force fluctuations on the surrounding air. The
force produced on the air by each blade is equal and opposite to the force produced on
each blade by the air, and the latter force can be resolved into lift and drag com-
ponents of the force on the blade along its aerodynamic axis.
The rotor alone noise is similar in nature to the propeller noise which has
been extensively studied by Gutin, Garrick, and Watkins, and other investigators.
The propeller noise theories are based, primarily, on three mechanisms: (a) thick-
ness noise, (b) lift noise, and (c) vortex noise. The experimental results on fan/
compressor noise showed however, that the lift (blade loading) portion is the primary
contributor of the rotor alone noise. Attention has therefore been directed towards a
prediction of rotor noise due to steady aerodynamic loading. The analysis consists
of an extension of Gutin's work that includes the effect of a many-bladed rotor and the
presence of the duct. It was assumed in this work that the blades Jo not interact.
That is, each blade carries its own discrete pressure profile, and this periodic dis-
turbance (in an absolute frame of reference) is mathematically described as an im-
pulse occurring at the blade passing frequency.
Wake Interaction Noise
Turbomachinery aerodynamicists have long recognized that the airfoils
arranged in rotating and stationary cascades of axial flow machinery, do not operate
in steady flow. A distinction should be made, however, between the unsteadiness in
the flow field created by the presence of an adjacent blade row which is characterized
by the generation of pure tones at the blade passing frequency, and the unsteadiness
due to the flow turbulence along the airfoils and in their wakes which represents the
primary source of broadband noise in the machine. The broadband noise generated,
is in general, of a lower order of intensity. A brief look will first be taken at the
noise created by the interaction of the wakes shed by a stationary or rotating cascade
with the following blade row. The wakes leaving a rotor or stator row necessarily
impinge on the adjacent blade row as they pass downstream. Within the wake there
is a reduction in velocity, and the primary effect of this velocity defect is to cause a
fluctuating incidence at the downstream blade. This fluctuating incidence gives rise
151
-------
to a fluctuating force which results in sound radiation.
PARAMETRIC STUDY OF ROTOR ALONE NOISE
Sound pressure levels generated by a rotor alone are a direct function of the
circulation around the blade row. The larger the circulation the higher the pure tone
levels. This is consistent with aircraft propeller experience. A study has been made
to investigate these effects, parametrically, from a turbomachinery standpoint. Some
of the results are shown in Figures C-l and C-2.
• Figure C-l shows that if the tip speed is held constant and the pressure
ratio is increased, the fundamental frequency sound power levels
increase.
• Figure C-2 shows that as the number of rotor blades is increased,
the sound power levels generated decrease. This seems to indicate
that a high blade design is favorable.
PARAMETRIC STUDY OF INTERACTION NOISE
A parametric study was carried out to determine the functional relationship
between the pure tone noise generated in the fan by the wake interactions and the pri-
mary turbomachinery aerodynamic and geometric parameters. This study was re-
stricted to fans and compressors without inlet guide vanes which, at the present time,
tend to produce lower noise levels than comparable machines with inlet guide vanes.
The study was also directed towards designs incorporating large blade row spacings,
where the wake and not the potential interaction is the major source of noise.
Pressure Ratio Effect
In an initial study, the tip speed of the machine and the fan geometry (num-
ber of blades and vanes, spacing/chord, and so on) were kept constant. Different
designs with pressure ratios varying from 1.2 to 1.4 were investigated. When the
pressure ratio is increased at a particular speed, more turning has to be done in the
blade row and the loading goes up. The terms V1/V2 abd sin 8 in the expression for
the coefficient of unsteady upwash subsequently increase; this is translated into an
increase of pure tone levels. It can be seen from C-3 that when the fan design pressure
ratio is increased from 1.2 to 1.4, the fundamental blade passing frequency power
levels increase about 8 db for a fan of constant size, and 4 db for a fan of constant
thrust. These results indicate that pressure ratio is an important parameter that
cannot be neglected. The use of simple correlation formulas that do not take this
effect into account may lead to erroneous results.
152
-------
14
12
a
•o
J
u
u
o:
£
s
Q
2
u
10
1.2
CONSTANT SPEED
CONSTANT NUMBER OF VANES
i.3
PRESSURE RATIO
1.4
FIGURE C-l. EFFECT OF PRESSURE RATIO ON ROTOR ALONE
BLADE PASSING FREQUENCY NOISE
tn
•a
u
i
Q
3
O
in
I
u
K
2
-2
-8
CONSTANT PRESSURE RATIO
CONSTANT TIP SPEED
20
30 40
NUMBER OF BLADES
50
60
FIGURE C-2.
EFFECT OF NUMBER OF BLADES ON ROTOR
ALONE BLADE PASSING FREQUENCY NOISE
153
-------
CQ
•O
»
J
u
>
u
cc
u
I
Q
2
I
U
U
K
ROTOR/OGV INTERACTION
CONSTANT SIZE
CONSTANT THRUST
CONSTANT PRESSURE RATIO
CONSTANT SIZE
CONSTANT TIP SPEED
CONSTANT SPACING/CHORD
1.20
1.25
1.30
PRESSURE RATIO
1.35
1.40
FIGURE C-3.
EFFECT OF PRESSURE RATIO ON INTERACTION
NOISE GENERATED AT THE BLADE PASSING
FREQUENCY
n
•o
J
u
>
u
j
K
U
I
Q
O
C/3
U
H
U
K
-5
-10 —
CONSTAN
IT PRESSUR
^
E RATIO
CONSTANT VANE/BLADES
CONSTANT TIP SPEED
CONSTANT SIZE
CONSTANT SPACING/CHORD
i
\
ROTOR/
^>
'OGV INTER
X
ACTION
^
10
20
30
40
50
60
70
80
NUMBER OF BI.ADES
FIGURE C-4.
EFFECT OF NUMBER OF BLADES ON INTERACTION
NOISE GENERATED AT THE BLADE PASSING
FREQUENCY
154
-------
Number of Rotor Blades Effect
The number of rotor blades is an important parameter in the noise generation
and can easily be modified (within certain vibration and aerodynamic bounds) to suit
the acoustic designer. Should fan designs, therefore, be oriented toward a high or a
low number of blades? The final answer depends upon the fan size. The number of
blades and the fan rpm will determine the pure tone frequency, which should be kept
out of the critical area of the NOY curve. Therefore, the size, as well as the blade
tip speed of the fan, must be known before a final selection of the number of blades
can be made. An investigation can be done, however, of the effect of the number of
blades on the sound power levels generated at the blade passing frequency. In the
following study, the pressure ratio, tip speed, blade row spacing, and the vane/blade
ratio were kept constant and the number of blades were varied from 20 to 80. The
results are shown in Figure C-4. It can be seen that a design incorporating a high
number of blades will reduce the fundamental frequency tones. The fan size and rpm
will determine the optimum configuration from a PNdb viewpoint.
Vane/Blade Ratio Effect
The analysis of the sound generation shows that the number of interaction
diametral modes (or in other terms, the vane/blade combination) has a major effect
on the blade passing frequency tones. Figure C-5 shows that a high vane/blade ratio
can be beneficial, not only from a sound transmission, but from a sound generation
viewpoint, as well.
Blade Row Spacing Effect
It is well known that increasing the spacing between blade rows will decrease
the interaction noise. Several experimental investigations have determined the reduc-
tion in pure tone levels that can be obtained when the blade row spacing is increased.
Some researchers show a 2 db reduction in sound pressure levels per doubling of the
axial separation, while others prescribe 4 or even 6 db SPL. This inconsistency can
be explained first by the fact that several other parameters besides axial spacing are
involved in the interaction process, and second, because the concept of "per doubling"
is not strictly correct.
In the present study, a particular rotor/OGV configuration was chosen and
the spacing varied from 0.1 to 2. 0 spacing/chord ratio. The results are shown in
Figure C-6. It can be seen that an appreciable reduction in pure tone power levels
can be obtained with large blade row spacings.
155
-------
a
•a
J -2
u
u
K
U
Q
§
U
-8
CONSTANT PRESSURE RATIO
CONSTANT SIZE
CONSTANT TIP SPEED
CONSTANT SPACING/CHORD
1.0
1.4 1.6
VANE/BLADE RATIO
1.8
2.0
FIGURE C-5. EFFECT OF VANE/BLADE RATIO ON INTER-
ACTION NOISE GENERATED AT THE BLADE
PASSING FREQUENCY
u
•D
J
U
D POWER :
1
CJ
z
3
U '10
H
J
W
CC
-15
V
\v
\v
>v
^
CONSTANT TIP SI
\
^
3EED
CONSTANT PRESSURE RATIO
CONSTANT NUMBER OF VANES
CONSTANT NUMBER OF BLADES
CONSTANT SIZE
i
1
ROTOR/OGV
^^
^"•^^
^^"^
INTERACTION
"^^
""^^^^
.5 1.0 1.5
SPACING AS A % OF ROTOR TIP CHORD
2.0
FIGURE C-6.
EFFECT OF SPACING ON INTERACTION NOISE
GENERATED AT THE BLADE PASSING
FREQUENCY
156
-------
SUMMARY AND CONCLUSIONS
Some theoretical methods have been developed to predict fan and compressor
sound power levels generated at the blade passing frequency. These prediction tech-
niques have been used to study the effects of different aerodynamic and geometric
parameters on fan and compressor noise. The results indicate that low sound power
levels will be obtained for:
• Low pressure ratios
• Low rotor blade loadings
• Low rotor diffusion factors
• High number of rotor blades
• High vane/blade ratios
• Large blade row spacings
An appreciable reduction in pure tone levels can be obtained by judicious
selection of design parameters. This reduction of noise at the source can be quite
attractive and should be incorporated in low noise fan and compressor designs.
By incorporation of this latest state-of-the-art design data, an optimum fan
could be designed to interface with the Rankine combustor.
157
-------
APPENDIX D
-------
APPENDIX D
EMISSION DATA REDUCTION
The raw emissions data are read from the strip chart on the Beckman
instrumentation cart and converted to observed values in parts per million (ppm) by
volume or percent by volume from the previously obtained calibration curves for the
instruments.
These observed values for carbon monoxide (CO), nitric oxide (NO), carbon
dioxide (CO2), and unburned hydrocarbons (H/C) together with the other pertinent
rig operational parameters are utilized as input data for the data reduction program
carried out on an IBM 360 Model 40 computer.
The program performs several correctional operations to the volumetric
concentrations before conversion to mass concentrations.
• The observed CO and NO readings are corrected for other gaseous
component interferences which produce higher readings. The inter-
ference correction factors are obtained from the emissions instrumen-
tation manufacturer.
• A correction is made to the nitric oxide value for ambient humidity.
This factor is contained in Section 1201.86 of the February 27, 1971
Federal Register (Volume 35, Number 40), and is normalized at a
humidity corresponding to a water content of 75 grains water per pound
of dry air.
• An exhaust water vapor correction is applied to put the volumetric
concentrations on a wet rather than a dry basis. The wet volume con-
centrations are used for conversion to mass concentrations and flow
rates of the individual species. The correction factor is calculated as
a function of test point air-fuel ratio and fuel ultimate analysis.
• The volume concentrations are converted to unity equivalence ration, i.e.,
are quoted at a stoichiometric air-fuel ratio. This correction is thus a
ratio of the actual test air-fuel ratio to the stoichiometric air-fuel ratio,
161
-------
the latter being calculated from the fuel ultimate analysis. The volume
concentrations are normally quoted as ppm or percent dry, corrected to
stoichiometric (and to standard humidity conditions in the case of NO).
As noted, the wet volume concentrations are used to convert to the mass
concentration using the test air-fuel ratio and the appropriate specie molecular
weight. The molecular weight of the exhaust gas is assumed to be that for air.
A calculated value of CO^ is obtained from a knowledge of the test air-fuel
ratio and the fuel compositions assuming complete combustion. This calculated CO2
value is compared to the measured value and in this manner any spurious results due
to irregularities in the air, fuel, or emissions measurements can be eliminated.
It can be seen from the attached sample sheet of program output that the
measured nitric oxide volumetric concentration is expressed on a weight basis in
terms of the oxidized product, nitrogen dioxide (NC^). This assumes that all the
nitric oxide ultimately reacts in the atmosphere to nitrogen dioxide.
The unburned hydrocarbons are quoted on a weight basis as an "average"
hydrocarbon, CH.. „_, as contained in the Federal Register requirements for light
duty vehicles. This pseudo hydrocarbon was proposed for gasoline fueled power
plants and is unlikely to represent the "average" unburned hydrocarbons while
operating on JP-5 but its use does form a basis for comparison.
The horsepower figure used in the input data and in the brake specific
emissions output is actually the test point fuel flow for convenience and has no other
significance.
162
-------
DATA POINT
FNGINE SPEED
0.0
FUFL % CARBON
T HYDROGEN
LHV 3TU/LB
HUMITITY GN/LB
85.09999
14.90000
18400.0
75.0000
CO 2 %
HORSEPOWER
HA LB/SEC
HF LR/HR
6.80000
109.00300
0.941 10
109.00000
-*-*_ «-*-*
WATER VAPOUR
EQUIVALENCE
HJMIDITY FED
HUMIDITY CAL
CORRECTION
0.93305
2.08027
I.00000
0.99981
NO
FACTORS
PPM 3RSERVED
PPM 30RR INT
PPM <
------- |