EPA-650/2-74-023
March 1974
Environmental Protection Technology Series
FLAME CHARACTERIZATION PROBES
4
\ s
-------
ABSTRACT
Experimental methods for use in characterizing the "dirty flame"
environments that result from the burning of coal and/or residual
oil were sought. A search of the combustion literature was made
to determine the applicability of the existing techniques that
have been more commonly applied to the characterization of less
formidable flame environments.
Techniques were selected for the measurement of temperature, chem-
ical species (stable and unstable) and velocity (magnitude and
direction) in high temperature "dirty flames": (1) the pneumatic
venturi pyrometer for temperature measurement, (2) the 5-hole
pitot probe for velocity measurement, (3) a quick-quench species
probe for measurement of stable species, and (4) molecular-beam
mass spectrometry for determination of unstable species.
Probe designs, based on the selected techniques, were developed
that would allow: (1) long probe life, (2) low response time,
(3) high spatial resolution, and (4) minimal flowfield disturbance.
After fabrication of the three probes, and after calibration of
the velocity probe, the operational aspects of the probes were
evaluated in a pre-mixed gas flame (at Rocketdyne). Satisfactory
operation was indicated for all probes.
Subsequently, the temperature and species probes were removed to
UCLA for comparative tests with UCLA's molecular-beam mass spectrom-
eter technique in a pre-mixed gas flame. Again, satisfactory agree-
ment was indicated. The molecular-beam mass-spectrometric technique
was also employed to obtain unstable species concentrations in the
same flame.
Evaluation of the probes suitability for use in a residual oil
flame was initially intended, but was not accomplished within the
resources available for the program.
This report was submitted in fulfillment of Contract No. 68-02-
0628 by Rocketdyne Division, Rockwell International, Canoga Park,
California, under the sponsorship of the Environmental Protection
Agency. Work was completed as of 1 October 1973.
111
-------
CONTENTS
Abstract iii
List of Figures vi
Acknowledgements viii
Introduction and Summary 1
Evaluation of Available Experimental Techniques 6
Temperature Measurement 7
Species Measurement 12
Velocity Measurement 16
Selection of Measurement Techniques , 19
Probe Design and Fabrication 26
Heat Transfer and Pressure Drop Requirements 26
Velocity Probe 33
Species Probe 33
Temperature Probe ' 45
Velocity Probe Calibration 56
Calibration Measurements 56
Construction of Calibration Charts 64
Use of Calibration Charts 68
Probe Checkout and Evaluation at Rocketdyne 69
Temperature Probe 69
Velocity Probe 74
Species Probe 76
IV
-------
CONTENTS (Continued)
Molecular-Beam Mass Spectroscopy at UCLA 78
Experimental Apparatus 78
Experimental Measurements and Data Reduction 83
Experimental Results 88
Verification Testing of Rocketdyne Probes 99
Conclusions ' ^ 103
References 104
Appendix A
Flame Measurement Techniques Appearing in the Literature 106
Appendix B
Review of Selected Literature on Flame Characterization by
Molecular-Beam Sampling 137
Appendix C
Tabulation of Data Obtained With UCLA Molecular-Beam
Mass-Spectrometer Sampling System 159
Appendix D
Theoretical Maximum Flame Temperature and Species Composition
of Methane - Air Mixtures at Various Equivalence Ratios 175
Appendix E
Factors for the Conversion of Units to the Metric System 176
-------
FIGURES
No. Page
1 Program Flow Diagram 3
2 Schematic Diagram of Molecular-Beam Sampling System 14
3 Mass Flowrate of Combustion Gas Through a Probe
as a Function of Probe ID and Gas Velocity 28
4 Mass Flowrate of Probe Cooling Water as a Function
of Combustion Gas Flowrate Through Probe, Probe OD
and Combustion Gas AT 29
5 Probe Length Required as a Function of Combustion Gas
Flowrate Through Probe, Probe ID, and Final (Cooled)
Combustion Gas Temperature 32
6 Velocity Probe - Detail Design Drawing 35
7 Five-Hole Pitot-Type Velocity Probe 37
8 Conceptual Design of Quick Quench Sampling Probe 39
9 Hot-Gas Sampling Probe Assembly 43
10 Rework of Hot-Gas Sampling Probe 46
11 Quick Quench Species Probe 47
12 Pneumatic Venturi Pyrometer - Initial
Conceptual Layout Drawing 48
13 Temperature Probe (Pneumatic Venturi Pyrometer)--
Final Detail Design Drawing 51
14 Pneumatic Venturi Temperature Probe 54
15 Temperature Probe - End View of Upstream Venturi 55
16 Typical Velocity Calibration Curves 57
17 Velocity Probe Calibration Rig 58
18 Velocity Probe Calibration Orientation 59
19 Velocity Probe Calibration Chart for Yaw (a)
and Pitch (3) 60
20 Velocity Probe Calibration Chart for
Velocity Magnitude 63
21 Relation Between Angles for Velocity Determination
in Five-Hole Pitot Probe 64
vi
-------
FIGURES (Continued)
No.
22 Velocity Probe Calibration Chart for Determination
of Angle of Flow 66
23 Velocity Probe Calibration Chart for Determination
of Velocity Magnitude 67
24 Gas Burner Design 70
25 Temperature Probe/Gas Burner Schematic 71
26 Evaluation of Temperature Probe in
Premixed Gas Flame 73
27 Velocity Probe/Gas Burner Schematic 75
28 Species Probe/Gas Burner Schematic 77
29 The Mass Spectrometer is Used to Analyze the
Chemical Composition 80
30 Flame Sampling System 82
31 Dual-Disk TOF Chopper 82
32 Average Molecular Weight of Combustion Gases
vs Equivalence Ratio for Methane-Air Flames 89
33 Beam Signals for H, OH, and 0 Radicals vs
Equivalence Ratio at D = 2.00 Inches 91
34 Mole Fraction of Direct-Sampled CO and CO,
Recycled CO,, and CO, vs Equivalence Ratio
at D = 2.00 Inches 92
35 Mole Fraction of Direct-Sampled NO and H~ and
Recycled NO vs Equivalence Ratio for D = 2.00 Inches 93
36 Beam Signals for Direct-Sampled 0_ and HO and
Recycled 0_ and HO vs Equivalence Ratio for
^ ^
D = 2.00 Inches 94
37 TOF and Thermocouple Measured Flame Temperatures
vs Equivalence Ratio 98
vn
-------
ACKNOWLEDGEMENTS
A portion of the research was subcontracted to the University
of California at Los Angeles under Rocketdyne Purchase Order
No. R24PXZ3156678.
The Rocketdyne Program Manager was L. P. Combs and Dr. R. C.
Kesselring was Rocketdyne's principal investigator. Professor
E. L. Knuth was responsible for the unstable species effort
and the overall molecular-beam mass spectrometry work per-
formed at UCLA.
W. E. Rodgers was responsible for the modifications made to
the UCLA molecular-beam mass spectrometer system and K. M.
Gorji and Dr. W. S. Young conducted the molecular-beam tests
and performed the data analysis.
This report has been assigned Rocketdyne Report No. R-9411.
Vlll
-------
INTRODUCTION AND SUMMARY
The objective of this program was the design, development, demonstration,
and delivery of probes that may be used to measure temperatures, chemical
species, and velocities in flames with emphasis on developing operational
capability of the probes in the "dirty flame" environments characteristic
of burning coal and residual oil.
The major design problems that were necessary to consider relate to the
environments in which the probes must operate. The media are potentially
quite dirty, consisting of as many as three phases (solid, liquid, and
gas) of material that is capable of condensing and forming dirty, tena-
cious coatings that can harden to change probe geometry or to catalyze un-
wanted chemical reactions. The combustion media are quite hot, requiring
cooling for most materials, and the combustion chambers are designed such
that access to them with probes is physically difficult. Also, the com-
bustion media are chemically reactive; a factor that can change the compo-
sition of samples gases if not quickly quenched or alter temperature read-
ings by physical sensors if allowed to react catalytically on surfaces.
A prime concern with any probe is that the presence of the probe itself
does not disturb the medium sufficiently to affect the measurement signifi-
cantly. Potential disturbing effects include distortion of pressure or
velocity profiles, chemical effects such as flameholding by the probe it-
self, excessive heat losses to or from the probe, or excessive removal of
combustion gas. Most such effects are minimized by using probes that are
physically small. Therefore, an important objective of this program was
to minimize probe size while avoiding the plugging and contamination
characteristic of dirty combustion environment.
-------
Another factor of concern for temperature, velocity, or species concentra-
tion probes is the time response of the probe. The combustors of interest
typically exhibit fluctuations with time, so that all of the quantities of
interest vary with time. Resolution of the time variance of these quanti-
ties requires fast-response probes, and fast-response recording. Generally,
fast response is associated with small probe and/or system size, although
other factors such as convective heat transfer to thermocouples can enter
into consideration.
To accomplish the objectives, the program was divided into two tasks as
shown in Fig. 1. During Task I, the literature concerning flame measure-
ment techniques was reviewed thoroughly, including flame characterization
by molecular-beam sampling. Based on results from this review, promising
measurements techniques were selected. The selected techniques were used
as the basis for design of individual temperature, velocity, and stable
species probes, which were developed subsequently in Task II.
A 5-hole pitot probe technique was used in the velocity probe. A quick-
quench concept was used in the species probe in which cooled sample gas
is pumped back to the probe entrance and mixed with the hot gas sample
to cool the sample and rapidly quench the reactions, preventing further
NO reaction. A pneumatic venturi pyrometer was used as the temperature
probe. An alternate method, suction pyrometry, was considered carefully
but it was concluded to possess little chance of immediate success in
the measurement of temperatures in excess of 3200 F. The pneumatic ven-
turi pyrometer, if provided with adequate cooling, has no upper tempera-
ture limit.
All probes fabricated during this contract were designed to meet length,
cooling water flow, and pressure drop requirements that permit them to be
adapted easily for future use in EPA experimental installations, if so
desired.
2
-------
Task I
Llteroture Review
Computer Searches
Literature Evaluations
Probe Concept Selections
Temperature, Stable Species, Unstable Species, Velocity
£?A Approval^)
1
-D
city
n and
cation
t Tests
Flames
II-A
Temperature
Design and
Fabrication
Checkout Tests
in Gas Flames
Task II
Probe Development
Molecular Beam
Stable Species and
Temperature Standards
Development
II-B
Stable Species
II
Unstable
Design and
Fabrication
Desig
Fabri
Checkout Tests
in Gas Flames
Testir.
Gas Fl
Verification Tests
Relative to Molecular Beam
Verification Tests
Relative to Molecular Beam
Feasibility
Tests in
Residual Oil
Flames
I Modifications J
Combination Probe
Analysis
Figure 1. Program flow diagram
-------
For determination of unstable species in a pre-mixed gas flame, a UCLA
molecular-beam mass-spectrometer system was modified and upgraded. A
liquid-nitrogen cryo-surface was added to the detection chamber to reduce
the background noise in the mass-spectrometer output. To facilitate
measurements of low-concentration species, a logarithmic amplifier was
designed and constructed. Handling of the signal component owing to meta-
stable molecules (not filtered out by the mass spectrometer) was facili-
tated through the use of a base-line compensator. For the determination
of the effects of vibrational-temperature changes on mass-spectrometer
fragmentation patterns, a heated effusive source was constructed, and the
mass spectrometer was calibrated for beam-gas vibrational temperatures up
to 900 C. For flame temperature measurements, a dual-disk time-of-flight
chopper was installed in the collimating chamber.
In addition, a methane-air burner was designed and built at UCLA in which
flat vertical-flow flames were produced over a porous alumina disk used
as a flame holder. The inlet to the molecular-beam mass-spectrometer sys-
tem and the burner shroud were adapted for sampling from this burner.
The UCLA molecular-beam mass-spectrometer system was used to directly
sample pre-mixed methane/air flame. With this technique, quantitative
measurements of CO , CO, H , and NO concentrations, qualitative measure-
ments of 09, H_0, OH, 0, and H concentrations, and quantitative measure-
ments of flame temperatures were shown to be feasible.
The individual probes fabricated during the program were checked out and
calibrated in a series of tests. Calibration testing of the 5-hole
velocity probe was performed at Rocketdyne in a cold-flow apparatus.
Calibration curves were constructed for the probe to allow determination
of velocity direction and magnitude. A second gas burner, equivalent to
-------
the UCLA gas burner, was assembled at Rocketdyne and used to preliminar-
ily evaluate the temperature, velocity, and species probes in a pre-mixed
methane-air flame. Satisfactory operation of the probes was demonstrated.
After preliminary evaluation in a gas flame at Rocketdyne, the species
and temperature probes were removed to UCLA for calibration by compari-
son of the temperature and composition of gases drawn through the probe
with those obtained with the molecular beam/mass spectrometer (MBMS)
technique. Simultaneous measurements with the Rocketdyne temperature
probe and the time-of-flight MBMS technique were accomplished. Good
agreement between the two temperature techniques was found at stoichio-
metric conditions but not quite as good as off-stoichiometric conditions;
somewhat higher temperatures being indicated by the time-of-flight MBMS
technique than by the probe. Comparison tests at UCLA of the Rocketdyne
species probe with the MBMS technique were accomplished consecutively,
not simultaneously, because the UCLA mass spectrometer was used for the
analysis of the composition of the samples obtained with both techniques.
The species probe was found to eliminate the radicals H, OH, and 0
(anticipated), increase the measured concentrations of CO, CO, and 0
(not anticipated), decrease the measured concentrations of NO (not sur-
prising) , and sharply decrease the measured concentrations of H_0
(anticipated). It is not known whether the elimination of cooled combus-
tion gas recycle in the Rocketdyne species probe would improve, worsen,
or leave unchanged, the comparative results between the species probe
and the MBMS technique. The results of the species comparison tests thus
indicate that further investigations may be merited dependent upon the
accuracy desired in the measurement of the particular species.
Feasibility testing of the probes in a residual oil flame, as well as
the analysis of a combination probe design, was originally planned after
completion of the probe verification tests with the MBMS technique (shown
by the dotted boxes in Fig. 1), but was not performed.
-------
EVALUATION OF AVAILABLE EXPERIMENTAL TECHNIQUES
The most significant factors that must be considered in the development
of probes for characterization of oil or coal flames are related to the
flame environment. Depending on the particular fuel selected, the type
of burner used and the position within an oil or coal flame, the flame
conditions may be difficult to probe and sample, because they present a
two- or even three-phase flow of very hot, turbulent, and chemically re-
active gases. At any particular location, the local gases can, by turbu-
lent fluctuations, change rapidly and frequently from reductive to oxi-
dative and vice versa. Solids, liquids, and condensables in the combus-
tion gases can combine to form very tenacious, unpredictable coatings on
unprotected surfaces. The formation of such coatings on probes inserted
in a flame for the purpose of measuring flame characteristics can easily
alter probe geometries and, thereby, functions. In addition, these coat-
ings may serve to catalyze unpredictable and undesirable chemical reac-
tions. Because the combustion media are so chemically reactive, the com-
position of sampled gases may change appreciably if the sample is not
quenched quickly. Temperature readings may also change appreciably if
catalytic reactions are allowed to occur on the surfaces of physical
sensors.
Initial effort on this program was a study and assessment of existing
techniques for flame characterization. Literature searches were made to
obtain available information and aid selection of suitable techniques.
Measurement techniques reported in the literature, including those con-
cerning molecular beam mass spectrometric techniques are described in
-------
Appendixes A and B. An evaluation of those techniques that might be con-
sidered applicable to measurements in the dirty flame environments char-
acteristic of coal or residual oil flames are further discussed below.
TEMPERATURE MEASUREMENT
Suction Pyrometry
The suction pyrometer technique involves the aspiration of sample gas at
high speed past a noble-metal thermocouple enclosed in multiple shields
in order to maximize convective heat transfer, while minimizing radiative
heat transfer, to the thermocouple.
The aspirated thermocouple probe (suction pyrometer) is a reasonable tech-
nique for measurements of flame temperatures. High aspiration rates can
effectively reduce errors caused by radiation and surface catalysis. How-
ever, a temperature limit exists above which the radiation shield mater-
ials fail and preclude probe use. Also, impingement of liquid (oil)
droplets on the thermocouple represents a problem because of the possible
cooling effects resulting from vaporization and endothermic decomposition
of the oil. The suction pyrometer may not be acceptable in the presence
of large particles because of possible sensor damage by repeated high-
velocity particle collisions caused by high-aspiration rates.
To minimize temperature errors caused by droplet impingement, it is de-
sirable to orient suction pyrometers so that the likelihood of droplet
impingement is minimal. As the gas is aspirated past the thermocouple,
it is possible to cause the gas to turn a corner (or even to swirl) to
help separate droplets from the gas, which ultimately contacts the
temperature-sensing element. In addition to decreasing the possibility
of droplet impingement on the couple, such action also decreases the
rate at which the flow passages in the radiation shields become plugged
(Ref. 1 ).
-------
Pneumatic Venturi Pyrometer
Another attractive temperature measurement technique is the pneumatic
venturi pyrometer. This instrument involves the aspiration of sample
gas through two venturi nozzles in series with cooling of the gas be-
tween nozzles. Application of the continuity equation to the continu-
ously flowing gas stream allows analytical determination of the hot sam-
ple gas temperature from measured values of:
1. The cold sample gas temperature (at downstream nozzle)
2. The pressure drops across the upstream and downstream nozzles
3. An instrument constant (see Appendix A).
The temperature limit of the venturi pyrometer is only a function of the
materials of construction used, probe design and the probe cooling pro-
vided. This technique is particularly attractive for high-temperature
application and involves shorter response times (i.e., seconds) than ex-
perienced with suction pyrometers (i.e., minutes) (see Appendix A).
However, under the dirty flow conditions, characteristic of oil and coal
combustors, solid deposits in the venturi nozzles can cause rapid change
in the value of the instrument constant, K, because of changes in nozzle
throat areas and discharge coefficients. For application in a "dirty"
flame, it is advisable to calibrate such a device in position before and
after each measurement. Pressure drops across the throats of each of the
two nozzles serve to define the area and discharge coefficients of the
contaminated nozzles. Such calibration would also serve as an indication
of when the probe should be removed from the flame for thorough cleaning.
An additional factor that may influence the operation of a pneumatic
probe is the probability of two-phase flow in the probe nozzles (espe-
cially for positions in a flame close to an oil burner's spray nozzle
or a pulverized coal injection and ignition site). The influence of
small amounts of liquid or solid particles may influence the gas flow
through a venturi nozzle in several ways, dependent on the size and
8
-------
concentration of the particles. Possibly the largest effect is the mo-
mentum transfer and gas pressure change resulting from droplet disinte-
gration. Aerodynamic forces to which the liquid droplets are exposed in
an accelerating gas stream, e.g., in a probe nozzle, tend to cause break-
ing of large droplets into many small droplets, the sizes of which are
difficult to predict even if the sizes of the large droplets entering
the nozzle were known. The net result is that it is difficult to correct
the nozzle pressure drops for the effects of two-phase flow in the pre-
sence of liquid droplets. Near the injection site in an oil burning com-
bustor, where oil droplets constitute only about 6 percent of the total
mass flow, inaccuracy in two-phase flow effects could lead to temperature
errors of as much as a few percent for pneumatic probes. However, near
the injection site, temperatures are usually low, and methods other than
pneumatic probes should be satisfactory for measuring those temperatures.
In relatively clean zones (not close to oil or coal injection sites),
where condensed or solid phases amount to less than 1 or 2 percent of
the mass flow, errors in temperature measurements caused by two-phase
flow effects should be less than 1 percent.
Velocity of Sound Probes
Temperature determination from velocity of sound measurements is a tech-
nique usable in the presence of a dirty flame. Measurement of sound
velocity in combustion gases results in the determination of a value
for (YT/M)1/2. The ratio Y/M is nearly constant, and its estimation
permits determination of the combustion gas temperature. Velocity of
sound measurements can be made either by noting the time of passage of
a sound wave past two detectors, or by observation of the physical size
of specially induced standing sound wave patterns. In both of these
cases, on the order of 5 to 10 cm of sound wave path length is required
to obtain a precise sound wave determination. This path length require-
ment infers very poor spatial resolution on the part of the sound velo-
city temperature measurement technique. For 109 Btu/hr combustors, the
-------
5- to 10-cm spatial resolution may be adequate, but for ICr Btu/hr com-
bustors, it is definitely not adequate. Sound velocity techniques thus
appear unacceptable for use in "dirty" flame combustor environments on
the basis of spatial resolution.
Optical Pyrometry
This technique involves the measurement of the brightness of a flame
(whose temperature is to be determined) by comparison with the bright-
ness of an object of known temperature. The measured brightness of the
flame is then related to gas temperature. If the object used for com-
parison is a sodium ion and the brightness of the D-line is used as the
standard, the technique is known as the sodium line reversal method.
The two major drawbacks to optical pyrometry are poor spatial resolution
and a relatively high lower temperature limit (see Appendix A). To mea-
sure local temperatures optically with acceptable spatial resolution, it
is necessary for the combustion gas to be optically dense at the wave-
length(s) of interest, and to use a probe that localizes the source of
light to the spatial point of interest. Such a probe can consist of a
rod of high-quality glass that easily transmits the Na-D line, coupled
with appropriate Na-D line narrow band filters. The glass rod should be
continuously cooled, possibly by containment inside a tube having an
inert gas purge. To make a temperature measurement, the purge must be
interrupted, or preferably changed instantaneously to aspiration of com-
bustion gas, to bring hot gas very close to the end of the glass rod
probe. To avoid overheating the glass rod, the period of combustion gas
aspiration and hence the period of measurement must be severely limited.
The end of the glass rod should be cleaned once every few measurements.
If the Na-D line is used, together with appropriate narrow-band filters,
it may be possible to use a commonly available sensing device, such as a
CdS cell. Disadvantages of the suggested optical technique include the
requirement for development and calibration. Also, this or any related
optical technique is not capable of better than 1 or 2 percent accuracy
10
-------
at most in a high-response system. Additionally, depending upon the fuel
being used, it may be necessary to seed the combustion system with small
amounts of sodium. An obvious disadvantage of this optical probe tech-
nique is its lower temperature limit of 2500 F (caused by weak emission
below this point).
Molecular Beam
Translational temperatures in gas flames can be determined directly from
measurement of molecular speeds. This is accomplished by means of a
molecular-beam sampling apparatus in conjunction with a time-of-flight
apparatus. The molecular-beam sampling apparatus causes gas withdrawn
from a combustion process to undergo a very rapid expansion to supersonic
velocities. The molecules that remain on or near the axis of this ex-
panding flow are naturally those molecules whose velocity was directed
in the axial direction at the end of transition from continuum to free
molecular flow. By intermittently chopping the molecular beam and mea-
suring the time required for a chopped portion of the beam to arrive at
a detector site, the velocity of the axially-directed molecules in the
beam can be determined. This hydrodynamic velocity (V) is a direct
indication of the stagnation temperature (T) of the source gas and the
mainstream gas temperature can be calculated from the expression
T _ T-l mV2*
YR
The molecular beam method offers the advantage of temperature measure-
ments without the need of blackbody or any other sort of temperature
calibration. The calibration required is one of time (used in molecular
velocity determination) that can be performed very accurately. The mo-
lecular beam technique also offers excellent response, well under one
millisecond. However, the molecular beam probe is generally associated
with relatively large probes (90-degree cone typical) that might disturb
*See page 87 for a more detailed description of the calculation
procedure,
11
-------
flow patterns. Unfortunately, the molecular beam technique is not rec-
ommended for use in "dirty flame" environments because of almost certain
plugging of the tiny sampling orifice by oil droplets or solid particles.
With the exception of the plugging and probe-size aspects, the molecular
beam technique is a superior method and this technique can be used to
calibrate any other technique selected for use in the flame field.
Evaluation
From this survey of possible methods for measuring flame temperatures,
both the suction pyrometer technique and the pneumatic venturi pyrometer
technique were selected for further consideration. The final selection
between these two techniques is discussed in a later section of this
report.
SPECIES MEASUREMENT
Sampling Probes
Continuous operation of a species sampling probe in the thermal environ-
ments characteristic of heating units or utility boilers is not difficult
because of the relatively low pressures and velocities. Low-pressure
water cooling is usually adequate for probe survival. However, it is
necessary in designing sampling probes to allow for thermal expansion
of exterior probe surfaces, to ensure that the probe remains intact
without any cracks, breaks, or distortions caused by repetitive heating
and cooling. The various features of species sampling probes are dis-
cussed in Appendix A.
Conventional water-cooled (convective-quench) species probes have been
used quite successfully in obtaining samples from distillate oil flames
(e.g., Ref. 2). However, to ensure that no further reaction occurs in
the probe itself, a "quick quench" of the sampled gas is desired. In
12
-------
addition, operation of a species sampling probe in a dirty, condensation-
and-deposition-prone medium invites plugging and formation of coatings
on both the interior and the exterior surfaces. For a conventional
(i.e., small sized, near-atmospheric pressure) water-cooled probe, the
formation of some exterior and some interior coatings is almost unavoid-
able. Formation of deposits on the exterior of the probe does not in-
hibit use of the probe, and it is of little concern. Plugging and for-
mation of deposits on the interior of a cooled probe can be minimized
by providing a boundary layer of clean gas near the interior probe sur-
faces, particularly at the narrowest restriction where plugging is most
apt to occur. Not only would this clean gas injection serve to quench
the sample gas rapidly, but it would also provide a clean gas internal
boundary layer to inhibit condensation of oils and sticking of small
carbonaceous particles near the probe inlet. This design would not
totally inhibit deposition within the probe, but it would minimize the
deposition and also keep any deposits that do form relatively cool by
direct contact with the cold, clean gas from the injection sites. Keep-
ing the deposits cool (which cannot be well accomplished by heat trans-
fer through the probe wall because of thickness, poor contact with the
wall, and low-thermal conductivity of the depositions themselves) is im-
portant to prevent hot deposits from catalyzing further chemical reac-
tion of the sampled gases. The quench gas could be either an inert gas
or cooled, recycled sample gas. Use of the sample gas itself for quench-
ing would ostensibly simplify chemical analysis procedures, but response
times could be quite long. On the other hand, inert quench gas could be
provided at high flowrates, limited only by total volumetric sample rate
and probe pressure limitations. Using externally supplied inert quench
gas, quenching times would be limited only by the time required for mix-
ing the inert gas with the sample gas and could conceivably be as low as
a few microseconds.
13
-------
Molecular Beam/Mass Spectrometer
The molecular beam technique involves the extremely rapid expansion of a
small sample of combustion gas from the combustion zone through a small
pinhole orifice and into a high-vacuum-source chamber (see Fig, 2 and
Appendix A).
SOURCE
SOURCE
CHAMBER
COLLIMATION
CHAMBER
DETECTION
CHAMBER
SKIMMER
COLLIMATING
ORIFICE
DETECTOR
MOLECULAR
BEAM
g
00
OCL
OS
z
O
w
0 a.
Figure 2. Schematic diagram of molecular-beam sampling system
The aerodynamic expansion is continued until free molecular flow is
achieved, and a portion of the expanded jet is skimmed and collimated to
form a molecular beam that may be analyzed using a mass spectrometer as
the detector.
14
-------
Through the rapid quench times of the molecular beam are highly desirable
and perhaps necessary for quenching fast reactions such as those involved
in NO formation, there are also disadvantages. The pinhole sample port
for a molecular beam is generally quite small, on the order of a few
tenths of a millimeter. The pinhole is made small to minimize the quench
time, but it cannot be so small that it samples only gases from boundary
layers on the end of the sampler. Plugging of the pinhole by liquid or
solid particulates in an oil or coal flame is almost a certainty. An
additional disadvantage of the molecular-beam sample apparatus is the
angle of the Prandtl-Meyer expansion and the high-vacuum requirements
that make the apparatus somewhat bulky for sampling the interior por-
tions of an easily disturbed combustion process.
Spectroscopy
Spectroscopy, the science of measuring and correlating spectral light
emission and absorption characteristics, can be used to identify various
chemical elements within a flame. The techniques employed are discussed
in Appendix A.
While Spectroscopy is a superior method for measuring chemical species in
uniform flowfields, its application to non-uniform flowfields requires
the use of a very complex system that enables the attainment of large
matrices of data. For particle-containing flow streams (such as the com-
bustion zones of oil and coal burners) spectroscopic techniques do not
appear appropriate because the particulates could provide an overwhelming
continuum background signal and could also interfere with absorption mea-
surements by scattering light from the external excitation source. For
clear exhausts much effort would be required to identify useful lines and
to determine their absorption coefficients.
15
-------
Evaluation
Use of a quick-quench probe for determination of stable chemical species
was concluded to be the technique best suited for application in the
"dirty flame" environments of interest. The even faster quenching
molecular-beam apparatus was considered to be the best method for deter-
mining unstable species, and for certain unstable species it is the only
known practical method at present. Additionally, since the molecular-
beam technique can also be used to obtain stable species concentrations
it can be used as a check against the stable species sampling probe in
other than "dirty flame" environments. Similarly, since the molecular-
beam technique can be used to determine temperature, it can also be used
as a reference from which to calibrate and/or evaluate the temperature
measurement technique selected for use.
VELOCITY MEASUREMENT
Magnitude
Pitot Probes-These commonly-used devices permit determination of a flow-
ing stream's dynamic pressure, which is a direct measure of its velocity,
by use of a differential pressure measurement between a total-pressure
(impact) tap and a static-pressure tap. Pitot probes have been success-
fully used in distillate oil flames (see Appendix A) and plugging of the
impact probe tap can be avoided through the use of a purge system. Plug-
ging of pitot probes is less a problem than plugging of certain types of
temperature and species probes, because there is no aspiration of sample
gas through the pitot probe. Modification of pitot probes for applica-
tion in dense two-phase (liquid-gas) flowfields has also been accomplished
(see Appendix A).
Hot-Wire Anemometry-A hot-wire anemometer consists essentially of a fine,
electrically-heated wire inserted in a fluid stream. A change in veloc-
ity of the fluid stream affects the rate of heat flow from wire to fluid
and thus, tends to heat or cool the wire. This, in turn, alters the
16
-------
resistance of the wire and produces a measurable effect in the electri-
cal heating circuit. Hot-wire techniques are not applicable for use in
flowfields where droplet ignition is occurring or where droplets or solid
particles impacting on the hot-wire could grossly alter the wire's ther-
mal balance or even damage it.
Mechanical Anemometry-Mechanical anemometry involves determination of
velocities by the capability of the gas flow to cause movement of small
windmill-type structures. Their use in combustion zones presents many
problems (see Appendix A), not the least of which are excessive probe
size cooling and selection of materials of construction.
Pulsed Tracer-Pulsed tracer techniques (see Appendix A) possess the ad-
vantage of permitting velocity to be measured without a knowledge of gas
density. Such techniques are also feasible in gas flows containing
solids or liquid droplets. A major disadvantage is that velocity magni-
tude cannot be determined without knowledge of velocity direction. This
disadvantage may be overcome if tracers are followed photographically,
but for "dirty flame" environments this presents perhaps even a greater
disadvantage.
Direction
Tracers-Tracer techniques may also be used to determine velocity direc-
tion. However, as stated above, the use of this technique in a "dirty
flame" combustion environment adds complexities to the technique, which
preclude its selection.
Multiple-Port Impact Probe-Multiple-port impact (pitot-type) probes are
reliable methods for the determination of flow direction. One type of
device employs pairs of individual probes that are gimbaled in a three-
dimensional field in order to balance the pressure readings and deter-
mine the flow direction (see Appendix A). Another type of multi-port
probe is the hemispherically-shaped pitot probe having multiple taps
17
-------
on the single probe head. This latter probe avoids the use of a gimbal-
ing apparatus. Flow direction is determined by measurement of the pres-
sure distribution over the probe head and locating the stagnation point,
and thus velocity direction. Careful calibration of each particular
probe is required. Such calibration can easily be accomplished in a
non-reactive flowfield.
Evaluation
A 5-hole hemispherical pitot probe appears to offer a relatively easy
means of determining both velocity magnitude and direction without the
problem of gimbaling the probe in the flame. This technique has been
employed at the International Flame Research Foundation (Ref. 1 ) in
dirty flame environments with apparent success.
18
-------
SELECTION OF MEASUREMENT TECHNIQUES
The preponderance of information concerning measurement techniques in
"dirty flame" environments originates from work done at the International
Flame Research Foundation (IFRF) in IJmuiden, Holland. Only two refer-
ences based on IJmuiden work were available at the program outset (Ref.
1 and 3 ) and they were found to be most informative. However, the
technical details concerning various measurement techniques were not
adequately discussed in all cases.
Based on the evaluation of the literature references obtainable at the
program outset, a tentative selection was made of the measurement tech-
niques to be employed in this program (the selections to remain tenta-
tive until the conclusion of verbal communication with EPA and IFRF per-
sonnel) . The tentatively selected measurement techniques were:
Flame Temperature - Pneumatic Venturi Pyrometry or Suction Pyrometry
Stable Chemical Species - Gas Quench Probe
Unstable Chemical Species - Molecular-Beam Spectroscopy
Velocity - Multi-port Impact Probe
Emphasis was placed on the selection of a temperature measurement tech-
nique for use in residual oil flames. Two methods were considered, the
suction pyrometer technique and the pneumatic venturi pyrometer tech-
nique (Ref. 1 ). The suction pyrometer technique involves the aspira-
tion of the sample gas at high speed past a noble metal thermocouple en-
closed in multiple shields in order to maximize convective heat transfer,
while minimizing radiative heat transfer, to the thermocouple. The pneu-
matic venturi pyrometer involves drawing the sample gas through two
19
-------
nozzles with cooling of the gas between nozzles. The gas temperature at
the downstream (cold) nozzle is easily monitored by a conventional ther-
mocouple or other device. From the continuity equation, the upstream
(hot) gas temperature is calculated from the measured downstream (cold)
gas temperature, the respective pressure drops across the two nozzles,
and an instrument constant, K, as follows:
T - K not T
hot gas ~ AP ,, cold gas
Before communication with the IFRF, a very recent book was received
(Ref. 4) that presents "in readily accessible form" the results of 24
years of cooperative research by the IFRF. The book supplied answers
to a number of questions but, for the most part, appeared too general
and lacked presentation of data comparing measurements obtained with
both the suction pyrometer and the pneumatic venturi pyrometer. With-
out actually stating so, Ref. 1 implied that the suction pyrometer has
been selected as the standard temperature measurement technique at IFRF
for all types of flames. The pneumatic pyrometer was regarded as less
accurate than the suction pyrometer and "only of interest for measure-
ments in gases where the temperature is very high or heavily laden with
dust."
Both the suction pyrometer technique and the pneumatic venturi technique
were discussed with Dr. Michael Heap of the IFRF (Ref. 5 ). Dr. Heap,
who has considerable personal experience with suction pyrometry, stated
emphatically that he preferred the suction pyrometer over the pneumatic
venturi, which he personally has never used, in residual oil or pulver-
ized coal flames. He gave three main reasons for this preference:
1. The suction pyrometer, unlike the pneumatic venturi, requires
no instrument calibration.
20
-------
2. Greater ease of operation is experienced with the suction pyro-
meter because sets of radiation shields may be quickly replaced
if plugged while, with the pneumatic venturi, chemical cleaning
may be required to remove sticky tar-like oil deposits from the
hot throat.
3. The suction pyrometer can give accurate measurements in any
flowfield while work of about 5 years ago at the IFRF indicated
that measurements with the pneumatic venturi are velocity sensi-
tive and that there are circumstances in swirling flows where
the pneumatic venturi is useless.
Work at the IFRF had centered predominantly around gas, residual oil and
pulverized coal flames. Heap (Ref. 5) stated that when blockage is ex-
perienced with a suction pyrometer in a pulverized coal flame, a total
of three sheaths (the outer radiation shield, the inside radiation shield
and the alumina sheath over the thermocouple) quite often fuse together
necessitating replacement of the entire set. With oil flames, soot (if
produced) often blocks the suction lines. In the oily part of the flame,
oil may block the suction lines as well. Heap further stated that, with
the suction pyrometer, measurements in oil flames are easier than in coal
flames. The suction pyrometer has a temperature limit of about 1700 C
(3550 R) imposed by problems with thermal shock on the shield at higher
temperatures. The only advantages of the pneumatic venturi are, in Heap's
opinion, the higher permissible operating temperature, and perhaps, its
ease of prolonged use in dry, dusty atmospheres (such as fly ash).
Concerning the thermal mapping of an individual flame with a suction py-
rometer, Heap stated that, where the flame occupies 1/3 or more of the
volume of a furnace, approximately 90 percent of the temperature mapping
could be accomplished with one or two sets of shields. However, to map
the entire flame, 40 sets of shields may be required for coal flames
(perhaps less for residual oil). Even so, Heap considers himself fortu-
nate to obtain valid measurements in coal and oil flames closer than
21
-------
20 to 30 cm (8 to 12 inches) from the burners and believes some regions
are beyond measurement under certain circumstances.
The pneumatic venturi technique appears to have been used only for sonic
gas flows in the U.S.A. (Ref. 6 and 7), but its use in industrial fur-
naces has been reported (Ref. 8 and 9) by the British Coal Utilization
Research Association (BCURA). Reference 9 describes its use in water-
tube boilers with gas, oil, and pulverized fuel flames and shows favor-
able comparisons in certain uses with the suction pyrometer. The advan-
tages of the pneumatic venturi pyrometer, especially when compared to the
suction pyrometer, are reported as being:
1. Equally accurate at all temperatures
2. Rapid response
3. Slower blockage tendency
4. More easily cleaned
5. Higher operating temperature.
The measurement sensitivity of the pneumatic venturi with respect to the
external velocity flowfield appeared to have been solved (Ref. 8 and 6)
by the substitution of a piezometer ring, with six equally spaced pres-
sure taps, for the single pressure tap.
Thus, the suction pyrometer technique is apparently preferred by re-
searchers at the IFRF (Ref. 5) while the pneumatic venturi pyrometer is
apparently preferred by researchers at BCURA (Ref. 8 and 9 ). In an
attempt to resolve the somewhat contradictory advantages claimed for the
two instruments, the United States representative of the Land Instrument
Co. (which manufactures, in England, both a suction pyrometer and a pneu-
matic venturi pyrometer) was contacted (Ref. 10). While no current use
of the venturi pyrometer in this country was known to him, a recent re-
port of experimental studies on the venturi pyrometer was forwarded to
Rocketdyne. This 1969 report (Ref. 11), which was the basis for Heap's
(Ref. 5) statements, points out the sacrifice in accuracy occasioned by
22
-------
a lack of knowledge of the discharge coefficient of the upstream venturi
nozzle under actual "hot flow" conditions as compared to "cold flow" con-
ditions. The need to correct temperature measurements obtained in flow
of considerable approach velocity is also emphasized*. These elaborate
corrections are a function of the suction velocity, approach velocity
and the direction of flow approach. It is concluded in Ref. 11 that
"practical advantages claimed for the VPP-probe (venturi pyrometer) for
use in dirty gases (Ref. 7 ), namely blockage and soot deposits easy to
remove and growing only slowly, are not attractive when time-consuming
and delicate correction procedures are to be adopted in order to get re-
liable results." Also, "from the present experience, a VPP-probe cannot
compete with a suction pyrometer with regard to measuring accuracy, re-
liability and ease of operation for temperature measurements in furnaces.
Its usefulness is limited to applications where very high gas tempera-
tures have to be measured, or where short response times are required."
While no United States use of the pneumatic venturi in oil or coal com-
bustion atmospheres was located, a suction pyrometer has been used by
Howard and Essenhigh to study the pyrolysis of pulverized coal particles
in pulverized fuel flames (Ref. 12). Suction pyrometer temperatures as
high as 1793 K were recorded. The suction pyrometer temperatures were
taken as particle temperatures. Although such a measurement should in-
dicate a temperature somewhere between that of the particles and that of
the gas, Ref. 12 reports calculations that indicate a negligible temper-
ature difference between the two phases for particles smaller than about
200 microns. No difficulty in pyrometer use was reported (Ref. 12).
Based upon the evaluation of the literature, it was concluded that, for
residual oil flames, the venturi pyrometer is a higher risk technique
*Interestingly, it is mentioned that "the influence of approach flow can
be reduced by using a suitable design of the hot-nozzle profile, plac-
ing the upstream pressure tappings somewhat inside the nozzle." The
placement of the upstream tap is the only apparently significant dif-
ference in design between the pneumatic venturi pyrometer types used
by researchers at the IFRF and at BCURA.
23
-------
than the suction pyrometer. However, the suction pyrometer, as it cur-
rently exists, is limited to operation below 1700 C (3090 F) because of
thermal stresses in the radiation shields. Conversely, the temperature
limit of the venturi pyrometer is only a function of the cooling pro-
vided. Since operation in a temperature range near 3600 F is desired
with the probe, two alternatives were considered: (1) increasing the
operating temperature limit of the suction pyrometer through use of new
shield materials or attachment methods, and (2) designing the "best pos-
sible" venturi pyrometer.
At a meeting at Rocketdyne on 30 January 1974, Dr. Heap related his re-
cent experiences at the IFRF, IJmuiden, in which an attempt was made to
upgrade the suction pyrometer for use at higher temperatures (hopefully
approaching 2000 C). The previously used radiation shields made of
Sillimanite (60-80% Al?0, + S.O,.,) were replaced with more expensive
shields made of recrystallized alumina for these tests. Although recrys-
tallized alumina is reported to be stable to 1900 C (3452 F), sagging of
the sheath was encountered repeatedly during operation at indicated tem-
peratures of about 1780 C (3200 F). In addition, rapid oxidation of the
thermocouple material itself was a serious problem. Attempts were made
to provide a protective layer of inert gas between the thermocouple and
the thermocouple sheath. It was concluded that a serious materials pro-
blem exists when attempting to operate a suction pyrometer above 1800 C.
A number of European investigators are attempting to overcome this
problem.
If a suction pyrometer capable of operation at no more than 1700 C
(3060 F) is desired, the most logical course of action would be to use
IJmuiden's design. However, a temperature probe capable of operation at
3600 F or more was desired. Consequently, the pneumatic venturi pyrom-
eter concept was selected for use in this program.
24
-------
The tentative selections of techniques for measuring flame species and
velocity were also discussed with Dr. Heap. Upon completion of these
discussions, the tentative choices were agreed upon as the techniques
to be applied in the Probe Development stage of the program.
25
-------
PROBE DESIGN AND FABRICATION
HEAT TRANSFER AND PRESSURE DROP REQUIREMENTS
In order to design a probe for a particular flame application, the ex-
pected heat load to the probe must be known or estimated. The probe may
then be designed to avoid its thermal destruction in the intended appli-
cation. In addition, probe cooling requirements must be such that unreal-
istic coolant pressure drops are not required.
All probes fabricated during this program were designed to operate in a
3600 F residual oil flame. The expected local heat flux and total heat
load to the probe caused by radiation (both from the hot combustion gases
and refractory furnace walls) and convection from the combustion gases
were calculated from conventional correlations. Assuming probe operation
in a cylindrical, refractory-lined, 30-in. diameter stainless-steel com-
bustor, expected temperatures of the refractory combustor liner and of
the combustor inner wall were calculated using correlations for gaseous
radiation, developed by H. C. Hottel, together with the separated flow
convection relations of Sparrow and of Zemanick. The resultant heat load
to the outer surface of a probe inserted in the 3600 F flame was calcu-
lated to be 210,000 Btu/hr ft^. This value was assumed to be character-
istic of the environment of any probe developed during this program and
was used to achieve a realistic combination of probe temperature, cool-
ing water AT, and cooling water pressure drop in the final probe design.
To aid in heat transfer analyses, probe requirements for cooling water
flow and cooling length were calculated and plotted. The curves shown
26
-------
in Fig. 3 } which were calculated from the continuity equation, were
used to obtain the mass flowrate of combustion gas through a probe as a
function of probe ID and gas velocity through the probe. Figure 4 was
developed for use in estimating the cooling water requirement for an in-
dividual probe. The curves in this figure were calculated from a heat
balance on the cooling water, i.e.,
(W C AT)U . = q , + (W C AT) (1)
* p JH 0 nrad * p ^gas ^ J
probe OD
A value of AT^ Q = 120 F was assumed* in the development of Fig. 4 as
was a probe immersion length in the flame of 36 inches. The curves shown
in Fig. 4 are nearly independent of combustion gas temperature drop over
a range of 2600 to 3100 F.
The use of Fig. 4 is recommended for the purpose of estimating the mini-
mum amount of water (W]-uo) required to cool a probe once its inside and
outside diameters, and the suction rate of gas through the probe (W~as),
are known. By using greater than this minimum amount of cooling water,
the temperature rise of the water could be decreased from the value of
120 F assumed in Fig. 4 . Since the radiation heat transfer term on the
right hand side of Eq. 1 is much greater than the capacitance term
(W Cp AT) , the W Cp AT term may be assumed negligible in order to ob-
tain a quick estimate of the cooling water temperature rise (AT) for any
*It is desirable to avoid picking up sufficient heat in the cooling water
to result in two-phase flow or steam at the cooling water exit. By
holding ATp^o to a maximum of 120 F, and assuming an inlet water tem-
perature of 70 F, the maximum outlet water temperature of 190 F would
preclude two-phase flow.
27
-------
to
CQ
O
to
<
0
0.25
Figure 3
0.50
ID, INCH
0.75
Mass flowrate of combustion gas through a probe
as a function of probe ID and gas velocity
28
-------
26
25
20
o
LU
to
-a-
o
X
to
CJJ
15
10
vO
o
o
o
o
II
PROBE IMMERSION LENGTH
= 36 IN.
2600 < 4TQAS < 3100 F
80
100
120
UtO
160
w..
180
GPH
200
220
2^0
Figure 4 . Mass flowrate of probe cooling water as a function of
combustion gas flowrate through probe, probe OD and
combustion gas AT
29
-------
outside probe diameter, immersion length, and cooling water flowrate as
follows:
qrad
From Eq. 1:
AT - probe OP
iilji o = .
M2U (W C )
P H20
but qrad = 210,000 B/hr ft2 = 210,000 ir D^ B/hr
CL = 1 B/lb F
Therefore
ATH,0
210,000 Tf D L.
* o i
(W CJ
P H20
or
D L.
T = 549 JL-i.
where
AT^ Q = cooling water temperature drop, F
D = probe outside diameter, inches
L. = probe immersion length, inches
% 0 = cooling water flowrate, gph
The probe length required to cool the 3600 F combustion gas to a temper-
ature, Tcg, by conduction/convection heat exchange with the cooling
water, i.e.,
(W C AT) = hA ATT..
v p ''gas LM
30
-------
can be estimated through the use of Fig. 5 . Figure 5 shows the re-
quired probe length as a function of cool gas temperature, gas flowrate,
and inside probe diameter.
The cooling water pressure drop resulting from flow through the probe was
estimated using the expression
AP,T = 1.965 x 10"7
where
AP,, .. = cooling water pressure drop, psi
"2
L = total length of probe, inches
Wu = cooling water flowrate, gph
2
p = wetted perimeter, inches
A = cross-sec area, in.
f = dimensionless friction factor = (0.00140 + 0.125/Re0'32)
2 .
Re = Reynold's number = 3.515 x 10 Wu ../p
2
All probes fabricated during this program satisfy length and cooling
water flow and pressure drop requirements that should permit them to be
adapted easily for use in EPA experimental installations if so desired.
It is necessary, nonetheless, to use Fig. 3 through 5 in conjunction
with the intended actual suction velocity and immersion depth in order
to ensure that:
1. An adequate quantity of cooling water is used
2. The probe length provided is sufficient to cool the combus-
tion gas to an acceptable value.
31
-------
25
20
CJ
LU
to
\
CO
-3-
o
X
t/J
C3
15
10
0.25
10
20
30
50
60
70 80
L, IN.
Figure B , Probe length required as a function of combustion gas
flowrate through probe, probe ID, and final (cooled)
combustion gas temperature
32
-------
The use of Eq.4 to calculate coolant water pressure drop enables one to
determine if sufficient source pressure exists to allow the desired flow-
rate of cooling water.
VELOCITY PROBE
The 5-hole pitot probe concept was selected for the velocity measurement
probe. Each probe must be calibrated so that both the magnitude and di-
rection of the velocity can be determined from measurement of three dif-
ferential pressures without rotation or gimbaling of the probe in a
flame. A hemispherically-shaped, stainless-steel head of approximately
12 mm diameter that contains five drilled holes of approximately 1.5 mm
diameter was chosen for the design, shown in Fig. 6 . The four holes
surrounding the central hole are oriented at a 45-degree angle to the
probe axis. Except for the stainless-steel probe head, the probe is
water cooled. The purpose of the uncooled stainless-steel probe head is
to minimize deposition of oil and tar within the pressure measurement
passages. The uncooled probe head was designed to function at tempera-
tures sufficiently high to burn those particles that deposit on the head,
thereby eliminating blockage in oil jets in the region between the flame
front and the burner exit (Ref. 4 ). Nevertheless, the probe head was
designed to limit the probe tip temperature to 1800 F. For the heat
transfer calculations, the probe length immersed in the 3600 F flame was
assumed to be 36 inches. Assuming a 120 F rise in cooling water temper-
ature, a cooling water pressure drop of less than 30 psi was calculated
for the concentric tube probe design of Fig. 6 , where water flows toward
the tip through the inner tube and returns through the annulus. Figure
7 is a photograph of the velocity probe after fabrication.
SPECIES PROBE
A quick-quench probe concept was chosen for extraction of gases to be
analyzed for stable species. The design, shown in Fig. 8 , features a
channel design concept in which the basic element of the probe is an
33/34
-------
-------
03
I
to
t^
\o
o
i-,
fx
X
M
H
O
O
r-H
CD
a,
x
(U
1 I
o
X
I
H
IX
0)
M
DO
H
37
-------
externally-splined tube. A cover is electroformed over the channels, and
a second, larger, coaxial tube is used to provide an annular return pas-
sage for cooling water.
The channel construction of the species sampling probe design allows sev-
eral different fluids (e.g., sample gas, quench gas, and probe coolant)
to be transported to and from the probe tip without introducing the com-
plicated fabrication requirements associated with multiple coaxial tubes.
Also, by using thin walls between the channels, a relatively large pro-
portion of the probe cross-sectional area is available for flow. The
sample probe includes a venturi head (see Fig. 8 ), which has the poten-
tial capability of self-induced aspiration of cooled sample gas for
quenching of the hot gas being withdrawn from the combustion environment.
The venturi shape induces the aspiration by reducing the local static
pressure at the probe tip.
The temperature at which the NO reaction is effectively nullified
(quenched) was discussed with EPA and Dr. Heap. The quick quench of com-
bustion gas from 3600 F to between 1110 to 1830 F (600 to 1000 C)* was be-
lieved to be sufficient to preclude further NO reaction. A quartz liner
could be used to decrease surface reactions, but it was felt that these
could also be minimized by rapid cooling with a metal wall, whereas the
quartz would inhibit the cooling. The concensus was that sufficient cold
recycle gas should be introduced at the upstream venturi throat so as to
cool the sample gas to a temperature of at least 1830 F (1000 C).
Assuming a recycle gas temperature of 250 F, the species probe was cal-
culated to require a cooled combustion gas recycle flowrate equal to 56-
percent of the total mass flowrate of combustion gas (sample plus cooled
"The value of 1000 C (1832 F) assumes no carbon deposition on the in-
side probe walls. If carbon deposition is a problem, the quench tem-
perature of 600 C (1112 F) is recommended.
38
-------
-------
recycle for quench) to quick quench 3600 F sample (combustion) gas to a
temperature of 1830 F. The use of such a recycle rate of cooled combus-
tion gas will increase the response time of the probe by approximately
0.2 second. In spite of the relatively large amount of required recycle,
and its adverse effect on response time, cooled combustion gas was chosen
for quenching rather than an inert gas such as helium. The use of am-
bient temperature helium for quench purposes would require (for quench-
ing to 1830 F) an amount equal to 66 percent of the resultant sample gas.
This amount should serve to dilute the sample gas to such a degree that
identification of trace amounts of certain species would be severely ham-
pered. The use of cooled combustion gas having the same species concen-
tration as the hot combustion gas does not result in such a dilution
problem.
An evaluation of the species probe heat transfer indicated that the sam-
ple gas/quench gas (recycle) mixture can easily be cooled from 1830 F at
the venturi throat to ~250 F at a position 3-feet downstream, thus en-
suring adequate quenching of the sample gas with a minimum amount of
cool gas recycle. An examination of the potential for heat transfer to
the various streams (shown schematically below) indicated that heat
41
-------
transfer between the sample combustion gas and the recycle gas streams
(q ), as well as heat transfer between the coolant channels and the
recycle channels (q ), is small compared to heat transfer between
other streams, and therefore, may be ignored. On the other hand, it was
revealed that, because of heat transfer between the coolant in the an-
nulus and the recycle gas stream, further substantial cooling of the re-
cycle gas should be expected. The temperature of the recycle gas stream,
before its injection at the venturi throat, was calculated to approach
closely that of the coolant in the annulus. Therefore, water vapor
should be removed from the recycled gas before it is reintroduced into
the probe in order to avoid:
1. Condensation of water in the pump and possible damage to the
pump
2. Excessive pressure drop in the recycle gas channel.
The ability of the quick-quench probe to self-aspirate the necessary
amount of cooled combustion gas was not pursued. The presence of the
venturi was believed to be of possible value in enhancing the mixing of
the recycle gas with the hot sample gas and, therefore, the venturi
throat was retained.
The preliminary probe design shown in Fig. 8 was reviewed with the EPA
technical monitor and his consultant, Dr. M. Heap. Based on this review,
the probe was redesigned. The overall probe OD of 0.56 inch was left un-
changed but the convergent entrance of the probe was truncated from 0.5
inch (see Fig. 8 ) to 0.25 inch (followed by convergence to an 0.125-inch
throat), a value commensurate with the probe inside diameter of 0.25 inch.
The design of the gas and coolant channels was revised (from that indi-
cated in Fig. 8 ) to achieve a minimum cooling water pressure drop. The
resultant detail design drawing of the species probe is shown in Fig. 9 .
A total cooling water pressure drop of about 25 psi and a recycle gas
pressure drop of about 1 psi were estimated for this design.
42
-------
-------
Fabrication
Initially, an attempt was made to fabricate the tubular nickel body of
the species probe by gun-drilling a nickel bar. However, an x-ray exam-
ination of the bar revealed an irreparable nonconcentricity of the center
(gun-drilled) hole. Consequently, another fabrication technique was
adopted. This time the nickel probe body was built up by electroforming
over a tubular aluminum base. Upon completion of the electroforming pro-
cess, examination of the completed probe body indicated uniform nickel
deposition. However, some slight curvature of the tube was caused by the
build-up weight of the long tube, which was supported only at the ends
while in the electroforming solution. The tube was straightened success-
fully and the aluminum mandrel was etched out to obtain a thick-walled
nickel tube. The external, axial channels were machined into the tubular
nickel probe body and, with these channels temporarily filled with wax,
the outer expansion sleeve was electroformed onto its outer diameter.
After brazing of the manifold block to the probe body, a leak check of
the assembly revealed leaks between the coolant and gas recycle inlet
and exit passages. A second (repair) braze cycle was then attempted but
subsequently halted when indications were found that some coolant passages
were plugged. The block was then machined from the probe and replaced by
a modified configuration (see Fig. 10), which allowed sequential assembly
with hand brazing in order to achieve a leak-free assembly. This approach
was successful. A photograph of the completed species probe is shown in
Fig. 11.
TEMPERATURE PROBE
The conceptual design for a pneumatic venturi pyrometer was based on re-
sults from both the literature evaluation and a heat transfer analysis.
This design is shown in Fig. 12 . To eliminate or minimize the reported
effect of the external flowfield on the temperature measurement (Ref. 11 ),
the upstream "hot" venturi pressure taps were positioned in the convergent
section of this venturi as was done by the BCURA investigators (Ref. 8
and 9). Additionally, a piezometer ring with six taps was also employed
45
-------
d
10
1 «.
a>
,n
o
fH
fr.
bO
C
H
i I
PL,
OT
VI
rt
60
I
n
* J
o
§
^ m
I I
u) fl
3-
«* o
tf Q
5 ?!
bO
H
tu
46
-------
5AD43-8/14/73-C1*
Figure 11. Quick quench species probe
47
-------
oo
PL,
(B
O
O
O
H
c
§
f-l
X
4J
c
CD
d>
CU
(1)
JH
48
-------
to eliminate sensitivity to external-flow direction (Ref. 8 and 9 ).
Provisions were made to allow ready access to the individual taps for
physically cleaning or unplugging the pressure taps, through use of a
copper cap (see Fig. 12).
A venturi diameter of 3/8-inch was selected. This size is intermediate
between the 1/2-inch size "successfully" used in BCURA work (Ref. 8 and
9 ) and the approximately 1/4-inch size used without much success at
IFRF, Umuiden (Ref. 11). The probe shown in Fig.12 was designed for
insertion perpendicular to the main flow and in this respect is similar
to the BCURA probe and unlike the 1/4-inch diameter venturi probe used
by Pengally at IFRF, Umuiden. A cooling length requirement of 60 inches
between the hot and cold venturi tubes was calculated for cooling the
combustion gases from 3600 to 500 F in the 3/4-inch inside diameter sec-
tion over a reasonable range of suction velocities. A maximum insertion
length of 36 inches in the 3600 F flame was assumed for design purposes.
The design of this venturi pyrometer in Fig. 12 was reviewed with the
EPA technical monitor and his consultant, Dr. M. Heap. The design was
generally satisfactory, but the 1.84-inch OD was considered too large to
permit accurate resolution of local temperature differences within a
flame combustion flowfield. Therefore, this design was revised to reduce
the OD of the probe to a maximum value of 1.0-inch while maintaining a
minimum venturi diameter of 0.20 inch and retaining the design features
shown in Fig. 12 . Both a splined-channel design (instead of concentric
tubes) and a higher temperature cold venturi were considered as methods
of reducing the OD while maintaining a reasonable cooling water pressure
drop.
An extensive analysis of the heat transfer and cooling water pressure
drop in the venturi pyrometer as functions of OD, ID, gas flowrate, water
flowrate, probe length, and water flow area indicated that construction
of a venturi pyrometer with a venturi throat diameter of 0.25-inch and
49
-------
an OD of 1 inch or less was feasible. However, the use of a splined-
channel fabrication design was expected to allow only about a 10-percent
reduction in probe OD for equal water AP, so the concentric tube design
concept was retained.
A detail design drawing of the selected configuration for the venturi
pyrometer is shown in Fig.13 . This design features a 1.0-inch OD by
0.50-inch ID probe with 0.25-inch diameter venturi throat. Assuming suc-
tion velocities of sample gas through the probe of 50 and 150 fps, and a
cold (downstream) venturi temperature of 500 F, the following coolant
water flowrate, length between upstream and downstream venturi tubes,
and total coolant pressure drop were calculated:
Suction velocity (fps) 50 150
Coolant water flowrate (gph) 167 171
Length between venturi tubes 37 46
(inch) for 500 F cold
venturi temperature
Total coolant AP (psi) 2 2
These results indicate the design length of 42 inches is adequate for
cooling the gas sample to a temperature of approximately 500 F before it
enters the downstream venturi. Although the predicted coolant pressure
drop is low enough to permit further size reductions, the outside probe
diameter could not be further reduced because of the spatial requirements
of the pressure tap passages. Six pressure taps were used in the conver-
gent section upstream of the hot venturi. Three pressure taps were used
in the hot venturi throat and two taps each at the upstream and throat
locations of the cold venturi. The hot venturi head of the probe was
made of nickel to aid cooling while the remainder of the probe was made
of stainless steel.
In order to reduce the probe OD from the 1.84 inches shown in Fig. 12 to
the 1.0 inch shown in Fig. 13, the ability to easily clean the hot
50
-------
-------
venturi pressure taps was sacrificed to some degree. The original con-
ceptual design (Fig.12 ) included a copper cap for the probe head that
could be removed to permit insertion of a drill in the hot venturi pres-
sure taps from the outside. The final design, Fig. 13 , included only the
option of incorporating small setscrews on the outside probe diameter.
These setscrews could be removed to allow insertion of a drill in the ven-
turi taps. However, a special tool would be needed to clean the venturi
taps from the inside of the probe. Photographs of the completed temper-
ature probe are shown in Fig.14 and IS.
53
-------
O
in
(U
^
3
+->
rt
Ki
cu
i1
4)
0)
O
CD
CXi
54
-------
50A66-6/13/73-S1B
Figure 15. Temperature probe - end view of upstream venturi
55
-------
VELOCITY PROBE CALIBRATION
CALIBRATION MEASUREMENTS
A 5-hole pitot probe can be used to determine both the magnitude and di-
rection of the velocity from measurement of three differential pressures,
once the probe is properly calibrated. Typical calibration curves are in
Ref. 4 and Fig. 16 .
The newly-fabricated velocity probe shown in Fig. 6 was calibrated using
a system consisting of a rotatable lathe table, on which the probe could
be mounted, and a 2-in. diameter steel tube through which a regulated
supply of gaseous nitrogen was flowed (Fig. 16). The probe was mounted
on the lathe table so that both pitch (6) and yaw (a) angles could be
varied separately (Fig. 17). Calibrations were done at a GN2 flow veloc-
ity of about 100 ft/sec. The location of the probe tip was varied between
0.5 and 3.0 inch above the exit of the 2-in. diameter tube. Horizontal
traverses were made with the probe to ascertain that the measurements
were made in a region with a flat velocity profile.
Employing the notation shown in Fig. 18 , variations in the pressure dif-
ferences P^-P-, ^0~^2' P3~P1' P ~P1' ant* P ~P3 were measured (in addition
to P0-P i t) with probe rotation in either the pitch or yaw directions.
These data were plotted as delta pressure ratios, as shown in Fig. 19, for
comparison with typical calibration data obtained at the IFRF (Ref. 4).
The calibration data exhibited very good agreement between corresponding
delta pressure ratios, (P4~P2)/(P -P.) and (P.-P )/(P -P2), "hen the re-
spective pitch or yaw angle was varied an equal amount, confirming symmetry
of the probe tip.
56
-------
Ratio of measured pressure
differentials
Five-hole probe; measurement of angles
40-3
e^4 3-4 Ratios of measuied pressure
b-i 0-3 differentials
Five-hole probe; measurement of velocity magnitude
Figure 16. Typical velocity calibration curves (Ref. 4 )
57
-------
co
i
I
i-H
CM
C
o
CT)
0)
O
0)
J3
o
X
p
H
o
o
r-H
0)
f^
iI
0>
B-,
58
-------
I
LU
CO
O O.
a:
Q- I-
c
o
H
+->
rt
4->
c
0)
H
fn
O
o
H
4->
Oj
O
(U
XI
o
5-i
O<
X
o
o
0)
f-l
3
bo
H
U.
59
-------
Vd
MO
zd-°d
o:
o
cc
o
/.
CQ
o
4->
H
PL,
rt
rt
0
O
oi
H
iI
OS
o
0)
o
X
o
o
II
CD
txO
H
60
-------
Comparison of the Rocketdyne calibration data with predictions based on
potential flow theory showed acceptable agreement with deviations ex-
pected for "real" flow situations. Nevertheless, the difference in the
shape of the Rocketdyne curve (Pd-PJ / (.po-p^ vs & (FiS- 19 J and the simi-
lar curve for the data in Ref. 4 , prompted more detailed calibration test-
ing. This testing involved the experimental measurement of the locations
on the probe tip at which Cpi-Pamb) equals zero. According to potential
flow theory, the stagnation point would have to be rotated an angle of
41.8 degree to record (P.-P , ) =0.* Calibration testing of the probe
1 cifflD
indicated an actual angle of 44.85 degree for attainment of zero and the
geometrical layout of the probe tip as shown schematically below.
P.-P , = 0
i amb
P.-P = 0
i amb
(TYPICAL FOR TWO CROSS-SECTIONS)
In addition the surrounding holes were found to be located at an angle
of 39.55 degree relative to the central hole (instead of the design value,
45 degree) and, also, the tip of the 0.50 inch OD probe was not a perfect
hemisphere (it had a height of ~0.17 inch, instead of 0.250 inch, corres-
ponding to a radius of curvature of 0.268 inch). These fabrication
*P.-P ./, /0 .,2 = 1- 9/4 sin2 8 = 0 yields sin 6 = 2/3,
i amb' l/2pV
or 8 = 41.8 degrees.
61
-------
deviations from the detail design drawing (Fig. 6) are the probable ex-
planation for the differences between the Rocketdyne calibration data
and that of Ref. 4 . These deviations, however, do not preclude in any
way the use of the velocity probe in its intended capacity because the
probe has been calibrated.
In addition to using the calibration data to calculate differential pres-
sure ratio versus yaw and pitch angle (Fig. 19), the data were also used
to calculate velocity recovery factors* versus yaw or pitch angle as
shown in Fig. 20 + along with IFRF results. The velocity recovery factor,
K -K., is defined by the relation
01 '
(Ko-K.) = (Po-P.)/l/2pV2 (1)
where position o is the central probe hole as shown in Fag. 18 and posi-
tion i is that of any surrounding hole.
In addition, the velocity magnitude may also be expressed as
V =
where P . = P when yaw and pitch angles are zero
stag o r
P , = ambient pressure
*The measurement of differential pressure ratios permits determination
of flow direction while the additional measurement of the velocity
recovery factor permits determination of velocity magnitude as well.
+The data shown in Fig. 19 represent only a change in pitch (3) angle.
Owing to probe symmetry, Fig. 19 would also represent a change in yaw
(a) angle, if the x-axis was labeled "a, degrees", the left-hand y-axis
labeled, "K0-K2", and the right-hand y-axis, "K0-K3".
62
-------
o
CM
LTV
o
o
vD
o
LT\
o
-T
CO
LLl
U-l
CC.
~ 0
O uj
M
«H
c
M
rt
O
o
1I
0)
?H
o
M-i
rt
X
o
c
o
rt
rt
O
O
O
0)
o
(N
0)
J-l
bO
H
VL,
O
CM
LA
O
A1I0013A
63
-------
CONSTRUCTION OF CALIBRATION CHARTS
By equating Eq. 1 and 2 , the velocity recovery factor K -K. was calcu-
lated from the calibration data as a function of yaw or pitch angle
(Fig. 20).
The results shown in Fig.19 and 20 are referred to the yaw (a) and pitch
(3) angles. However, in Ref. 4 the angles and 6 as shown in Fig. 21
were also used. The angle tj> is the angle between the velocity vector
and the x axis. The angle 6 is the angle between the projection of the
velocity vector on the zy plane and the y axis. The angle a (yaw angle)
is the angle between the projection of the velocity vector on the xz
plane and the x axis. The angle 3 is the angle between the projection
of the velocity vector on the xy plane and the x axis. The angles are
interrelated through the equations.
tan
tan a
, .
and tan
tan a
Velocity
Probe
Figure 21. Relation between angles for velocity determination
in five-hole pitot probe
64
-------
A calibration chart for use with the Rocketdyne Velocity probe was con-
structed in terms of the angles cj> and 6 for definition of the flow angle
(as was done in Ref. 4 ). This chart is shown in Fig. 22 , which was cal-
culated from the experimental calibration data obtained at yaw and pitch
angles of zero and 90 degrees. Two procedures for calculating this set
of calibration curves were explored; both involved use of the data at the
two pitch and yaw angles to estimate the pressures for any angles. Initi-
ally, an attempt was made to use the potential flow expression for the
pressure distribution over a sphere with the introduction of two empiri-
cal coefficients, D. and Q,
P -P
i amb n f, 9 .2 ,_.-.,.
= D. 1 - T sin (Q6)
V2PV2 'I I' 4
Secondly, the data were used explicitly with interpolation equations, and
assuming the flow was symmetrical around the axis of the probe, to calcu-
late the pressures. The second approach was the most successful in des-
cribing the available data. Therefore, it was used to calculate the
curves shown in Fig. 22 .
A similar calibration chart for determination of velocity magnitude with
the Rocketdyne velocity probe was also constructed. This was accomplished
through use of the relation (KQ-^) = (P0-Pi)/l/2pV2, where the value of
PO and P^ were determined as described above for any flow direction. The
resulting velocity magnitude calibration chart is shown in Fig. 23.
The calibration charts developed for the velocity probe fabricated during
this program, Fig. 21 and 22, are incomplete in as much as only a single
quadrant of the total calibration chart is actually shown. As seen in
Fig. 16, the total calibration chart consists of four quadrants, which
in the"case of Fig. 22 and 23 will be mirror images of each other at the
x and y axes.
65
-------
0123456
(P,-P3)/(P0-P3)
Figure 22. Velocity probe calibration chart for determination
of angle of flow
66
-------
fN
O.
-q-
a.
Figure 23. Velocity probe calibration chart for determination
of velocity magnitude
67
-------
The total calibration chart can thus be obtained by reference to Fig. 22
and 23, which represent the upper right-hand quadrant. For example,
referring to Fig. 16 , the upper left-hand quadrant is a mirror image of
the upper right-hand quadrant about the y axis. The lower left-hand
quadrant is a mirror image of the upper left-hand quadrant about the x
axis and the lower right-hand quadrant is a mirror image of the upper
right-hand quadrant about its x axis.
The pressure ratios representing the x and y axes in the various quadrant
are not constant and vary as shown in Fig.16 . Similarly, the angle 6
has a different range in each of the four quadrant as shown in Fig. 16.
Lastly, the velocity recovery factor also varies from quadrant to quadrant.
A velocity recovery factor of (KQ-KJ) applies in the upper and lower right-
hand quadrants while a recovery factor of (KO-K]J applies in the upper
and lower left-hand quadrants.
USE OF CALIBRATION CHARTS
The use of the velocity probe calibration charts is most simple. Measure-
ment of three differential pressures (P4~P2> P0~P3> anc* Pi-p3) are suffi-
cient to determine the flow direction as defined by the angles and 6
using Fig. 22 . The same three differential pressures also allow deter-
mination of the velocity recovery factor using Fig. 23. Velocity magni-
tude can be calculated from the definition of the recovery factor, i.e.,
V a1
It is again mentioned that the calibration charts presented in Fig. 22
and 23 are applicable only to the particular velocity probe fabricated
under this contract. A different instrument would require its own separ-
ate calibration.
68
-------
PROBE CHECKOUT AND EVALUATION AT ROCKETDYNE
A special gas burner (Fig. 24) was designed and built at UCLA for use in
the molecular beam mass spectroscopy portion of this program conducted
at UCLA. A second gas burner identical in most respects to the UCLA
burner was also fabricated at Rocketdyne for use in the preliminary eval-
uation of the newly fabricated probes. A 2.66-inch diameter porous ce-
ramic disk was used in the burner head to create and hold a flat pre-
mixed methane/air flame within a 3-inch diameter, 4-foot-long quartz
tube. To promote mixing of the air and fuel, five layers of brass
screens were placed below the flame holder, as shown in Fig. 24 .
TEMPERATURE PROBE
Initial tests of the temperature probe were made in a premixed gas flame
with the setup shown schematically in Fig. 25 . The hot-gas temperature
is calculated from the venturi pyrometer measurements by
AP
T C T
!H ' h AP 'c
c
where TH = gas temperature at upstream (hot) venturi, R
TC = gas temperature at downstream (cold) venturi, R
APjj = pressure drop across hot venturi
APC = pressure drop across cold venturi
F = instrument constant
69
-------
LU >~ 5
Q Q CO
£ o
CC
o
CO
o
o
o
< u_
X o
O LU <
O < > LU
LU uj o£
I/I I U. O Z
3 oc
at I- o o
to uj ac i ti-
cs z o
Z OC O- 01 Z LU
ID Q. Z Z5 OC
QC CD =5 LU =>
CO LU LU H
X LU OC OC X
31 O O O
co i i co z: z
CO
^~
OC
o.
0.
LU
^_
OC
0
Q.
Q-
^^
CO
o
(J
z
CD
^3
t-
-
LU
tn
2
ISI
|
^
S
o
z
o
_J
_
-3-
CM
CL,
T.
CJ
oc
LU
Q-
0-
-
a
LU
OC
CU
Z
ac
^
OQ
*
t
UJ
z
LU
^^
U-
LA
LU
CQ
1-
1
Z
^
O
o
o
SO
a.
z:
CJ
OC
LU
3
O
-
LU
O
o
z
K
LU
Z
t
z
1
o
o
o
o
U-
LU
LU
^~
00
LU
aa
i
ac
O
a.
a.
^3
m
0
0)
0)
e
3
(N
(U
^
3
70
-------
o
H
4->
OS
O
0)
1/1
03
00
0)
,£>
o
0)
^H
+-)
rt
PU
LO
(N
3
M
71
-------
The instrument constant, F, was obtained in preliminary cold-flow tests.
The maximum suction flowrate of room temperature air was drawn through
the probe by means of a series of three vacuum pumps installed in con-
junction with the analysis equipment. Using the measured values of APj^
and APC (obtained with an inclined manometer), and noting that Tc = T^ =
ambient, a value of F = 1.225 was calculated.
The pneumatic venturi pyrometer was then used (Fig. 26) to obtain the
gas temperature in a fuel-lean methane-air flame with an equivalence
ratio* of 0.75. The measurement was made in the center of the 3-inch
tube diameter at a distance of 2-inches downstream of the burner head.
A gas flame temperature of 2930 R (1626 K) was calculated from the mea-
sured AP's and the instrument constant"1".
The theoretical maximum flame temperature of a methane-air flame was cal-
culated as a function of equivalence ratio using the Rocketdyne n-element
propellant performance program. This computer program accounts for the
establishment of definite equilibrium conditions between products and
reactants. Results are summarized in Appendix D. The theoretical tem-
perature of a methane-air flame having an ER = 0.75 is 1922 K. The rela-
tively large 296 K difference between the 1922 K theoretical flame tem-
perature and the measured value of 1626 K is believed due to:
1. Loss of heat from the flame
2. Inherent inaccuracy involved in the method of pneumatic
pyrometry.
The accuracy of the pneumatic pyrometry technique was investigated fur-
ther by comparison with the molecular-beam mass spectrometer technique.
These results are reported in a later section of this report.
*ER = (CH4/air)/(CH4/air)stoichiometric
+TU = F APU/AP^ T = 1.225(0.925 inch H-0/0.320 inch FLO)(366 + 460) =
n n c c ^ ^
2930 R
72
-------
O
LT>
0)
OS
0)
X
H
0>
OH
C
H
(U
O
to
O
H
4->
OJ
rI
rt
>
PJ
00
H
U-.
73
-------
It is interesting to note, however, that Perry (Ref. 17) reports that,
for a stoichioraetric methane-air mixture, a theoretical maximum flame
temperature of 2340 K is calculated although the experimentally deter-
mined flame temperature (sodium line reversal) is only 2148 K, a differ-
ence of 192 K.
The response time of the temperature probe was calculated to be on the
order of tens of seconds (Ref. 4 , p. 42). The use of a low-response
device such as the manometer for pressure recording, of course, adds to
this value. A posttest calibration with room temperature air yielded an
instrument constant within 1-percent of the pretest value.
During operation of the temperature probe in the gas flame, it was ob-
served that care was required to avoid overcooling the probe. Overcool-
ing results in condensation of water vapor (present in the methane/air
combustion products) within the probe that ultimately leads to the pre-
sence of water in the annular piezometric rings used to measure pressure
drop across the venturi tubes and adversely affects the pressure drop
measurement.
VELOCITY PROBE
The velocity probe was also initially tested in a premixed methane/air
gas flame (Fig. 27). The primary purpose of this test was to verify the
adequacy of the probe cooling capability. Meaningful velocity measure-
ment was not expected because the estimated gas velocity in the 3-inch OD
quartz tube was on the order of only 6 fps. This value is equivalent to
a manometer reading of only 0.008 inch of waterless than the smallest
scale division on the manometer (0.01 inch of water). No significant
velocity was indicated by the manometer.
74
-------
C3
Z
OL
«-J UJ
0 (-
O < Z
03
r
O
Z
QC
_l LU
0 1- 1-
o < ^
030
^
_
LU
CO
O
CCL
O.
^ /
/
/
/
t
1 ,'
' /
&''
LLJ
o:
o
trt C/5
1 1 1 n
OC <
a- 1-
cd
0)
X
o
l/l
t/1
rt
Q>
JZ
O
X
P
(J
O
-------
SPECIES PROBE
The species probe was also inserted into a premixed methane/air gas flame
(2 inches downstream of the burner head) for initial checkout (Fig. 28).
With a fixed water coolant flowrate, concentrations of CO, C02, and NOX
in the combustion gas were measured as functions of the fraction of the
combustion gas flow being drawn into the probe that was recycled after
cooling. Data analysis showed the following variations with percentage
of recycle gas.
ER
0.75
0.75
0.70
Percent
Recycle
48
43
18
CO,
ppm
1391
1452
1300
C02,
percent
8.59
8.59
8.13
02.
percent
5.90
5.90
7.00
NOX,
ppm
10
8
10
The effect of percent recycle on composition shown by these results is
not considered significant. Percentages of C02 and 02 were quite close
to theoretical values* obtained using the Rocketdyne n-element propellant
performance model (Appendix D). No abnormalities were evident during
testing and operation of the species probe in the flame appeared
satisfactory.
*Theoretical calculations yield (on a water-free basis)
iER C02, percent 02, percent
0.75
0.70
8.51
7.92
5.50
6.65
76
-------
O
f-l
0)
cS
bO
0)
O
f-i
-------
MOLECULAR-BEAM MASS SPECTROSCOPY AT UCLA
A molecular-beam mass-spectrometer system, in existence at UCLA at the
beginning of the program, was substantially modified and improved to
quantitatively determine the molecules and radicals present in a pre-
mixed methane/air flame. Determination of the stable, as well as the
unstable species existing in the flame was planned as well as the flame
temperature itself.
EXPERIMENTAL APPARATUS
The experimental apparatus employed in the molecular-beam mass-
spectrometer system consisted of the following:
A. Gas Burner
A special gas burner designed and built for these studies was described
earlier and is shown in Fig. 24. The flame holder consists of a 2.66-
inch-diameter porous ceramic disk to produce vertical-flow flat flames.
The flame holder was movable in the vertical direction to facilitate
sampling from different locations within the flames. It was water-
cooled by means of two concentric stainless-steel tubes, which also
serve as the burner mount. Air and fuel flowrates were measured with
flowmeters, which in turn, were calibrated with a wet testmeter (Preci-
sion Scientific No. 63125).
The flame-holder assembly was inserted in a 4-foot long, 3-inch diameter
quartz tube. A 2.5-inch diameter hole in the wall of the tube admitted
78
-------
a conical, quartz sampling probe. The probe had an apex angle of 90 de-
grees and a sampling orifice diameter of 0.059 cm. Three layers of as-
bestos sheet sealed the probe-tube junction. The integrity of the seal
was checked by:
1. Blowing air on it and looking for a change in the flame shape
2. Analyzing for 0 in fuel-rich flames.
No air leaks were detected in the experiments reported herein.
Since the primary reaction zone was always below (upstream of) the sam-
pling probe, the presence of the probe did not influence the flatness of
the flame. The flame flatness was influenced, however, by the fuel-air
ratio. A relatively flat flame was realized at and near = 1,* whereas,
a less stable flame, containing small conical regions, was realized at
4> = 0.75 and 1.30.
All measurements were believed to be made outside the boundary layer
formed as a result of the flow through the quartz tube. The (displace-
ment) boundary-layer thickness was calculated to be 0.12 inch; the sam-
pling cone penetrated 0.8 inch into the gas stream.
B. Vacuum System
As shown in Fig. 29, the vacuum system consisted of three separate cham-
bers. The sampling gas was accelerated aerodynamically through the sam-
pling orifice and into the source chamber. This chamber was maintained
near 10~2 torr by a 16-inch Stokes booster diffusion pump in series with
a Stokes mechanical pump.
The core of the supersonic jet enters the collimating chamber through a
skimmer. The skimmer has an internal half angle of 16 degrees, an
= equivalence ratio = (CH4/air)/(CH4/air)stoichiometric
79
-------
4-1
COO)
O 6
H (D (4
P f-i rH
H 3 4-1
in p
o rt H X
g IB P
o
o
o
X
3
rt
o o> in
H x
B P O
o
3
o in
KI O in
nl -O C
C
in c x
^-5 h
fH 0> 0>
0) P
p
(!) p
I
!H
P
O
bO
e
3
O
in o
I 0>
-------
external half angle of 20 degrees, and an orifice diameter of 0.043 inch
(0.11 centimeter). The collimating chamber is evacuated by a 10-inch
CVC diffusion pump in series with a Hereaus mechanical pump. The pres-
sure in this chamber is typically about 2 x 10"5 torr during the sampling.
The molecular beam passes through a slit in the rear wall of the colli-
mating chamber into the detection chamber. The detection chamber is
evacuated by a 6-inch NRC diffusion pump and a liquid-nitrogen-cooled
cryosurface. The detection chamber pressure is maintained at about
3 x 10"' torr during sampling from the flame.
C. Sampling System
During this program, the existing molecular-beam mass-spectrometer sys-
tem was adapted for sampling from the burner (Fig. 30) as follows:
1. A logarithmic amplifier was designed and constructed for mea-
surement of a low concentration species in flames. This am-
plifier uses a baseline compensator to compensate for back-
ground signals produced by metastable molecules.
2. A heated effusive source was built to facilitate the determin-
ation of temperature effects on mass-spectrometer fragmentation
patterns.
3. A liquid-nitrogen shroud was added to the detection chamber to
reduce the background noise.
4. A new time-of-flight chopper (Fig. 31) was constructed and in-
stalled for flame-temperature measurements. The time-of-flight
chopper has been described in detail by Young (Ref. 13); data-
reduction procedures are discussed in a later section of this
report (page 87).
81
-------
Figure 30. Flame sampling system
Figure 31. Dual-Disk TOF Chopper
82
-------
D. Mass Spectrometer
The EA1 quadrupole mass spectrometer, shown schematically in Fig. 29,
consists of three primary components: the ionizer, the quadrupole mass
filter, and the electron multiplier. The gas to be analyzed enters the
ionizer, where a small percentage of the gas is ionized by electron bom-
bardment. These ions are then accelerated and focused in the quadrupole
section. The quadrupole assembly is composed of four stainless-steel
rods, each approximately 5 inches long and 1/4-inch in diameter. A DC
potential and a superimposed RF potential are applied to the rods. An
electrostatic field is thereby generated in the region between the rods.
Most ions entering this region oscillate, collide with one of the poles,
and are removed from the ion beam. Only ions in a very narrow range of
charge-to-mass ratio can pass through the quadrupole. Those ions that
do pass through the quadrupole strike the EAI beryllium-copper electron
multiplier. The multiplier output current can be conveniently displayed
on many types of instruments (such as oscilloscopes and X-Y recorders)
to show the quantitative abundance as a function of atomic mass.
E. Signal Amplifying and Averaging System
For time-of-flight measurements of flame temperature, the output of the
electron multiplier was amplified. The amplified signals were then av-
eraged by a Northern Scientific Model 513 Digital Signal Averager. The
averaging time depends upon the signal-to-noise ratio.
EXPERIMENTAL MEASUREMENTS AND DATA REDUCTION
The experimental measurements and data reduction methods employed are
summarized as follows:
Composition Measurements
Two different data sets were taken in the composition measurements.
First, a primary data set was taken in which the flame compositions
83
-------
were measured at equivalence ratios of 0.75, 0.80, 0.90, 1.00, 1.10,
1.25, and 1.30 and at sampling-orifice to flame-holder distances of
0.75, 1.00, 1.25, 1.50, 1.75, 2.00, 2.25, 2.50, 2.75, and 3.00 inches.
The second data set was taken for measurement comparison with the
Rocketdyne probe (discussed in a later section) and was taken for equiv-
alence ratios of 0.90, 1.00, and 1.10 at a distance of 2.00 inches.
Signals for mass-to-charge (m/e) ratios ranging from 1 to 44 were re-
corded. In order to minimize fragmentation, the ionizing electron
energy was set at 20 ev.
Temperature Measurements
Flame temperatures were measured using the dual-disk time-of-flight (TOP)
chopper with the signal from m/e =28. In molecular beam sampling at
high Reynolds numbers, based on the source conditions, the hydrodynamic
speed V is the same for all species. Therefore, one may use any species
or m/e present in the beam to determine V. The most convenient choice
is m/e = 28, which provides the strongest signal. These measurements
were made at equivalence ratios of 0.75, 0.80, 0.90, 1.00, 1.10, 1.25,
and 1.30 and for sampling to flame-holder distances of 1.00, 2.00, and
3.00 inches. The TOP signals were punched on computer cards and then
analyzed with a computer program generated for that purpose.
Flame temperatures were also measured using a platinum/platinum-10-per-
cent rhodium thermocouple for the aforementioned equivalence ratios and
for a sampling to flame-holder distance of 2.00 inches.
Mass-Spectrometer Calibration
The mass spectrometer was calibrated for gases at elevated temperatures.
to determine:
1. Temperature effects on mass-spectrometer fragmentation patterns
2. lonization efficiencies for CH4, NO, H2, and CC^.
84
-------
Using a heated effusive source, these calibrations were carried out for
known CH.-N0, NO-N9, H,-N , and CO-N- mixtures at room temperature, 250,
T" L. £. £, Lf £*
500, 750, and 900 C. During the flame composition measurements, the mass-
spectrometer was additionally calibrated with a supersonic room-tempera-
ture air beam.
Data Reduction for Composition Measurements
To determine the compositions of stable species and radicals in flames,
the mass spectrometer signals for mass-to-charge (m/e) ratios of 1, 2,
14, 15, 16, 17, 18, 28, 29, 30, 32, 40, and 44 were analyzed. The
supersonic-air-beam calibration data were used to determine the contri-
butions of m/e = 32 to 16, 18 to 17, and 18 to 1 because of fragmenta-
tions. No appreciable changes in fragmentation patterns of CH^, C0_, H_,
and NO were observed for temperatures from room temperature to 900 C.
For those cases in which the analysis is qualitative (caused by lack of
ionization cross-section values), the species signal was multiplied by
the mole fraction of N_ and divided by the N2 signal.*
For those cases in which the effusive-source calibration data were avail-
able, these data were used to determine the relative ionization effi-
ciencies. For example, for an N,,-H2 mixture, the relative ionization
efficiency is determined from
R.I.E. =
a(N9)
where aCH^) = ionization probability for H- molecules
a(N2) = ionization probability for N2 molecules
AH2 = H2 signal level in the calibration data
AN- = N« signal level in the calibration data
^Qualitative data so derived are employed in some of the data plots
presented later with ordinates labeled "Beam Signal, Arbitrary Scale."
85
-------
x^ = H mole fraction in the mixture
H2 2
XN = N_ mole fraction in the mixture
2
mN = molecular weight of N~
2
m.. = molecular weight of H-
The R.I.E. and the flame sampling signals were used to calculate the
mole fraction of H_ from
[H2]
.R.I.E.
where [H?] = H2 mole fraction in the flame
[N_] = N- mole fraction in the flame
2 2.
SH = H? signal level from the flame
2
SXT = N_ signal level from the flame
2
Since the m/e = 28-signal corresponds to both N2 and CO molecules, it
was not possible to determine the CO concentration directly. However,
it is believed that, at = 0.75 to obtain the ionization cross-section ratio
a(N2)/a(Ar) for use at other values of . More specifically,
S28"Sco S28
Data taken with the Rocketdyne probe indicate that S /S2g is less than
0.002 and may be neglected in this determination of cross-section ratio.
Therefore, N? and Ar signals at $ = 0.75 were used in conjunction with
the argon signal at other equivalence ratios to calculate the N2 contri-
bution to the m/e = 28 signals at these equivalence ratios. The
86
-------
difference between the measured m/e = 28 signal and the calculated N_
contribution is the CO signal. A literature value for the relative ion-
ization efficiency then determines quantitatively the CO concentration.
Data Reduction for Temperature Measurements
The time-of-flight technique may be used to determine the stagnation tem-
perature of the source by determining the velocities of the molecules in
a beam extracted from the source.
Alcalay and Knuth (Ref. 14) developed a moment method to extract the beam
density, temperature, and energy from the measured TOP signal. Algebraic
relations between the moments of the measured TOP signal, the speed dis-
tribution function, the modulator gate function, and the dynamic function
of the detector and its electronics were derived. This technique was ap-
plied here to N« time-of-flight measurements for methane/air flames.
The stagnation temperature of the flame is given by:
TQ = T (1 + 1:1 M2) = T [l + M (V/ar
where T = stagnation temperature of the source
T = static temperature in the free jet
Y = specific heat ratio
M = Mach number
V = hydrodynamic velocity
a = speed of sound
Since the static temperature is much less than the stagnation temperature,
substitution for the speed of sound, a, yields
To S
₯ IV,.,*].
87
-------
where m is the mass (molecular weight) of the beam molecules and R is
the universal gas constant. The velocity, V, can be calculated from
CL -L )2 l
-
2S2
where LT-LTT = distance between the two chopper disks
S = hydrodynamic speed ratio
r\ { } = first moment operator
I"*"(t) = instantaneous detected signal
t = time
<£ = phase angle of the dual-disk chopper
a) = angular speed of the chopper
For a diatomic gas with a molecular weight of 28, the value of ((Y-1)/2J
(m/YR) equals 4.81 x 10 K sec^/m . The variation of m with , shown
in Fig. 32, was taken into account in the calculations.
As an example, TQ for $ = 1.10 and D = 2.00 inches may be calculated.
For these conditions, V was determined experimentally to have a value
of 2096 meter/sec. For <}> = 1.10, the mean molecular weight of the com-
bustion gases in thermodynamic equilibrium (Ref. 15) is calculated to be
27. Using the relationship TQ = ((Y-1)/2YRJ (mV2) TQ = 2036 K is obtained.
EXPERIMENTAL RESULTS
The methane-air flames were sampled directly for the stable products H2,
02, H20, CO, NO, and C02 and for the radicals H, 0, and OH. Flame-
temperature measurements were made with both the TOP and a thermocouple.
88
-------
O
CM
co
O
r-
*
o
0)
o
0)
cr
rt
bfl
C
o
H
4->
t/1
1
o
o
<+-!
f-l 0>
rt s
r-H {«
3 X
O 4J
rt
< n
(U
^
M
H
P-.
LTV
LTl
04
O
CM
LTV
S3SV9 Noiisnawoo jo
89
-------
Direct-Sampling Composition Measurements
The results of measurements (qualitative or quantitative, as the case may
be) on methane-air flames with D = 2.00 inches are presented versus equiv-
alence ratios in Fig. 33 through 36. The qualitative results have not
been converted to mole fractions in those cases where values for the ion-
ization cross section were not available. Tabulations of the data (for
several sampling-orifice to flame holder distances and for several equiv-
alence ratios) are provided in Appendix C.
The term "recycled" used in Fig. 34, and subsequently, refers to the gas
that was drawn through the Rocketdyne species probe (which employed
cooled combustion gas recycle) and then returned to the MBMS sampling
system for analysis. These measurements are described in a subsequent
section.
The beam signal levels for the radicals H, 0, and OH are shown in Fig.
33.* The m/e = 1 signal includes, in general, contributions from H20
and H2 fragments as well as from the H radical. In order to isolate the
H-radical contribution, the fragmentation of H20 and H2 to H was estab-
lished through a heated-effusive-source calibration. H2 fragmentation
to H at an ionizing electron energy of 20 ev was found to be negligible.
H20 fragmentation to H and OH was small; about 4 percent of the m/e = 1
signal and taken into account. However, the primary source of H atoms
in lean methane-air flames is believed to be
OH
0 «- CH20 + H
*Equivalent readings on the arbitrary "y axis" scale (shown in Fig. 33 )
for the three radicals H, OH, and 0 d_£ not correspond to equivalent
concentrations. They would so correspond only if the three species
had equal ionization cross sections (values are not known for the un-
stable species).
90
-------
o-
o
<
o <
I/I
0>
o
o
r-i
II
Q
O
H
03
0)
O
c
0>
3
cr
cu
01
3
01
03
O
H
TJ
rt
rt
O
rt
C
0)
oa
a;
Vi
M
H
U-
91
-------
'00 JO NOI1DVYIJ 310W
LA
o
Lf\
O
O
o
LA
O
O
O
LA
03 JO NOIiOVyd 310W
92
O
u
CT)
T3
QJ
r-l
O
X
o
0)
oj
(N
i n
+->
o a
CD
H +->
H rt
T3
O
t^ -H
O 4->
rt
C M
o
H O
O C
aj
-------
0>
.I
o
X
o
CS (D
x:
(N O
X C
H
T3
C O
rt o
O (N
II
T3
H .H
+J 3
o cr
rt a)
CD V)
LO
tO
CD
CxO
H
X
00
X
X
-a-
x
CNl
310W
93
-------
03
(N
O
O
o
0)
13
rt
O A
rg O
X C
H
T3
C O
rt O
CsltN
O
II
T3
0) Q
i (
& fn
1.8
w
I O
P -H
O +J
T3
t-t -H
cd 3
C cr
bo
H
t/1
0) 0>
CQ >
VO
to
(1)
W)
Oo
3ivos Aivyiiayv 'IVNDIS wvaa
94
-------
As the flame becomes fuel rich, the hydrogen atoms are consumed by
H + CH4 J \\2 + CII3
Figure 33 shows that the H-atom signal increases as increases to 1 . 1
and then decreases as the flame becomes more fuel rich.
The presence of 0 atoms in the flame was deduced from the m/e = 16 sig-
nal. This signal represents, in general, both CH. and 0 atoms. Fristom
(Ref. 16) concludes that CH. disappears within the first few millimeters
of the reaction zone. The possible chemical reactions for this disap-
pearance are:
CH4 + OH ? CH3 + H20
CH4 + H £ CH3 + H2
CH4 + 0 J CH3 + OH
CH t CH + H
In the present studies, the distance from the reaction zone to the samp-
ling orifice was always equal to or greater than about 2 cm. Further-
more, no CH or CH3 (CH.- fragment) signals were observed. (In heated-
effusive-source calibrations for CH.-N at an ionizing electron energy
of 20 ev, CH3 and CH fragments of CH had been observed.) Therefore,
the contribution of CH4 to the m/e = 16 signal can be completely ignored
and the beam signal in Fig. 33 for m/e = 16 represents only 0 atoms.
95
-------
When CH disappears in the flame, CH radicals are formed. However, CH.
^\ J *.
was not observed in the samples. Perhaps the CH_ was consumed by the
reactions
CH + 0 ~t H2CO + H
The OH radical is very important in the methane flame. In order to iso-
late the OH signal from the m/e = 17 signal, the contribution caused by
fragmentation of H-0 was established through a heated-effusive-source
calibration. The resulting OH radical signal is shown also in Fig. 33.
The OH is produced by means of
H + 02 £ OH + 0
0 + H2CO t OH + HCO
The measured mole fractions of CO and C0~ versus equivalence ratio are
displayed in Fig. 34. As expected, the mole fraction of CO increases,
because of the incomplete combustion of the fuel, as (j) increases. In
methane-air flames, it is believed that the main sources of CO are
H CO -t- M t CO + H. + M
^ ^
H2CO -H OH J HCO + H20
HCO + OH + CO + H20
where M is an inert species. The mole fraction of C0? peaks in the
neighborhood of = 1. CO is formed, and CO disappears, according to
CO + OH * C02 + H
96
-------
H- and NO mole fractions versus equivalence ratio are displayed in Fig.
35. H^ appears, in fuel-rich flames, for <}> > 1.10. The concentration
of NO increased from 80 ppm at tj) = 0.75 to 500 ppm at = 1.30. NO is
formed by the mechanism
N2 + 0 t NO + N
N + 02 ~t NO + 0
Qualitative 0~ and PLO signals versus equivalence ratio are given in
Fig. 36. It is seen that the concentration of 02 decreases as the flame
becomes more fuel rich. The concentration of H_0 also peaks in the neigh-
borhood of <}> = 1. The main source of H20 is CH4 + OH J CH3 + H20.
Temperature Measurements
Results from time-of-flight and thermocouple measurements of temperatures
in methane-air flames are compared in Fig. 37. (See Appendix C for tab-
ulated data.) TOF measurements were made at sampling to flame-holder
distances of 1.00, 2.00, and 3.00 inches and thermocouple measurements
were made at a sampling to flame-holder distance of 2.00 inches. As
shown in Fig. 37, the flame temperature decreased as the flame holder to
sampling orifice distance increased. The maximum temperature (2209 K)
occurred at = 1.00 and the minimum distance for which measurements were
made, namely 1.00 inch. For the distance at which both TOF and thermo-
couple measurements were made (namely, D = 2.00 inches), the thermocouple
measurements gave values about 400 to 500 K lower than the TOF measure-
ments. This difference was attributed to the fact that the thermocouple
had not been shielded against radiation losses to the surroundings.
97
-------
2000
1600
LU
OC
1200
800
400
0.50
D - DISTANCE FROM FLAMEHOLDER TO SAMPLING PORT
O TOF TEMPERATURE AT D - 1.00"
A - 2.00"
D - 3.00"
A THERMOCOUPLE TEMPERATURE AT D - 2.00"
(NOT CORRECTED FOR RADIATION
LOSS)
THEO. MAXIMUM FLAME TEMPERATURE
* ROCKETDYNE PROBE AT D - 2.00"
I
I
I
1.00
EQUIVALENCE RATIO,
Figure 37. TOF and thermocouple measured flame
temperatures versus equivalence ratio
1.50
98
-------
VERIFICATION TESTING OF ROCKETDYNE PROBES
Both the temperature and species probes were taken to UCLA for compara-
tive testing with the UCLA molecular beam/time-of-flight technique.
A short series of simultaneous flame temperature measurements at various
equivalence ratios was made at UCLA with the Rocketdyne temperature probe
and the UCLA technique. For these tests, a 4-foot-long quartz tube was
used. The tube had a small hole on one side to permit insertion of the
temperature probe (perpendicular to the flame and tube axis) and a larger
hole to permit insertion of the conical quartz probe of the molecular-
beam system directly opposite the probe. Measurements were made simul-
taneously at an axial distance of 2-inches with both measurement techni-
ques. Data obtained using the molecular beam TOP technique resulted in
the flame temperatures shown in Fig. 37. Data taken with the Rocketdyne
temperature probe at equivalence ratios of 0.75, 1.0, and 1.3 resulted
in flame temperatures denoted by the starred symbols in Fig. 37. Good
agreement was obtained between the two techniques at = 1.0, values of
2125 K and 2043 K being obtained with the Rocketdyne probe and the mole-
cular-beam time-of-flight technique, respectively.* The agreement was
not as good at off-stoichiometric conditions. The Rocketdyne probe re-
sulted in flame temperatures of 1630 K and 1790 K at equivalence ratio of
*A theoretical maximum flame temperature of 2340 K is reported by Perry
(Ref.17) who also reports an experimental determination of 2148 K ob-
tained with the sodium line reversal technique. The theoretical flame
temperature calculated as a function of equivalence ratio using the
Rocketdyne n-element propellant performance program (See Appendix D)
is also indicated in Fig. 37.
99
-------
0.75* and 1.30, respectively, while the TOP technique resulted in temper-
atures of 1861 K and 1918 K at the corresponding equivalence ratios.
In addition, comparison tests were made with the species probe and the
UCLA MBMS technique. These measurements were conducted consecutively,
not simultaneously, because the UCLA mass spectrometer was used to analyze
the composition of the samples obtained with both techniques. The UCLA
MBMS technique was first used to obtain flame composition as a function
of equivalence ratio. These direct sampling results are shown in Fig. 33
through 36 described previously. The methane/air gas burner was then
repositioned to allow a gas sample to be withdrawn from the flame through
the Rocketdyne species probe, cooled by passage through a cold trap (placed
in an ice bath), and then passed directly into the pinhole at the end of
the conical quartz probe of the MBMS system. The sample gas bypassing
the MBMS system flowed into either a single pump for recycling or three
other pumps for exhausting to the atmosphere (Fig. 28). Approximately
25-percent recycle was, therefore, used in this test series.
Species measurements, made earlier at Rocketdyne, involved the use of
more sophisticated devices for removal of water from the combustion gas
sample. Approximately 100 percent of the water contained in the combus-
tion gas was removed in the previous tests at Rocketdyne. The less
sophisticated water removal device installed temporarily at UCLA for
verification testing of the species probe was apparently successful in
removing (by condensation of the ice bath) only about 60 percent of the
water from the combustion gas (Fig. 36).
Results from the spectrometer analysis of the gases withdrawn through the
Rocketdyne species probe with internal recycle are shown in Fig. 34
through 36 for comparison with previous direct-sampling results. Caution
*It is recalled that previous preliminary testing of the temperature
probe at Rocketdyne had yielded a temperature of 1626 K at ER = 0.75.
100
-------
should be used, however, in making direct comparisons between the
"direct sampled" and "recycled" curves shown in Fig. 34 through 36.
This is because, according to the H20 curves in Fig. 36, the "recycled"
curves represent combustion gas data in which 60 percent of the water
has been removed. The "direct sampled" curces, on the other hand, rep-
resent data in which no water removal has taken place.
In Fig. 34, the C02 "direct-sampled" data agrees excellently with the
theoretical results tabulated in Appendix D. The CO "direct-sampled"
results, however, are significantly less than the theoretical results of
Appendix D. This, perhaps, is indicative of the unattainment of thermal
equilibrium in the flame at the sample distance (2 inches). If the
"direct-sampled" measurements were reported on a water-free basis, the
"direct-sampled" mole-fractions would be increased by approximately 20-
percent. This would yield excellent agreement with the C02 "recycled"
curve shown in Fig. 34, but a wide difference with the CO "recycled"
curve would still exist. Calculations made on a 60-percent water-free
basis would, of course, yield results in between the "direct-sampled"
and 100-percent water-free results.
In Fig. 35, the "direct-sampled" data for NO and H2 were approximately
an order of magnitude and two orders of magnitude, respectively, below
the theoretical concentrations found in Appendix D. The "recycled" data
fell below the "direct-sampled" data, as shown in Fig. 35, on either an
uncorrected or a water-free basis.
Comparison of the 02 "direct-sampled" and the 02 "recycled" data shown in
Fig. 36 revealed that, if the direct-sampled data was converted to a 60-
percent water-free basis, the amount of 02 measured using the Rocketdyne
probe was greater than that measured employing the MBMS technique over
almost all of the 0.90 to 1.10 range of equivalence ratio.
101
-------
Thus, in summation, passage of the sample gas through the Rocketdyne
probe eliminated the radicals H, OH, 0, and increased the measured con-
centrations of CC>2, CO, and 02. The measured increases in the latter
three species were not anticipated and merit further investigations.
Less nitric oxide was observed in the "recycled" gases than in the
direct-sampling gases. Possibly some of the NO was converted to 02 and
N2 inside the probe. No H2 was detected because comparisons were re-
stricted to <}) = 0.90, 1.00 and 1.10.
102
-------
CONCLUSIONS
Velocity, species and temperature probes for application in "dirty flame"
environments have been designed and fabricated. These probes were de-
signed based upon a review of the current literature and possess those
characteristics believed to be necessary for successful application in
dirty flames. This does not preclude, however, the likely need for
periodic probe cleaning and/or maintenance.
The operation of these probes was evaluated in a premixed gas flame, and
successful operation was indicated. The temperature and species probes
were also evaluated relative to a known "standard" of the molecular beam
technique, in a premixed gas flame. Again, successful operation was in-
dicated, although some differences were observed.
The evaluation of the developed probes in a residual oil flame (or other
"dirty flame" environment) is needed to verify their suitability. The
results of such tests should be obtained before more advanced probe de-
signs (such as combination probes) are undertaken.
Included in this report are relations necessary to estimate requirements,
such as cooling water flow and allowable cooling water AP, which are
needed for testing of the developed probes in other experimental
situations.
103
-------
REFERENCES
1. Beer, J. M., N. A. Chigier, G. Koopmans, and K. B. Lee. Measuring
Instruments for the Study of Flame at Ijmuiden. International Flame
Research Foundation Doc. nr. 72/a/9, Ijmuiden, Holland, May 1966.
2. Dickerson, R. A., A. S. Okuda, and C. L. Oberg. Design of an Optimum
Oil Burner for Control of Pollutant Emissions. Report No. R-9465.
Rocketdyne Division, Rockwell International, Canoga Park, California,
February 1974.
3. Vizioz, J. P., and W. Leuckel. Methodes et instruments utilise's a la
Station Experimentale de la F.R.I.F. pour des mesures dans les flammes
de diffusion. (Instrumentation Methods Used at the F.R.I.F. Experi-
mental Station for Diffusion Flame Measurements). Communication
presentee a la Conference "I Problem della Combustione - Bruciatori"
de 1'Associazone Termoteenica Italiana, Milan, 18 through 19 June
1969.
4. Chedaille, J., and Y. Braud. industrial Flames, Vol. 1, Measurements
in Flames. Crane, Russak § Co., Inc., New York, New York, 1972.
5. Heap, M. P. Private communication, December, 1972.
6. Temperature - Its Measurement and Control in Science and Industry.
Vol. Ill, Part 2, Section IV, Articles 53-57, Reinhold Publ. Co.,
1962.
7. Clark, J. A., and W. M. Rohsenow. A New Method for Determining the
Static Temperature of High-Velocity Gas Streams. Trans. ASME. Feb.
1952, pp 219-228.
8. Godridge, A. M., R. Jackson and G. G. Thurlow. The Venturi Pneumatic
Pyrometer. Journal of Scientific Instruments. Vol. 35, March 1958,
pp. 81-88.
9. Holland, R. E., R. Jackson and G. G. Thurlow. The Behavior of the
Venturi Pneumatic Pyrometer in Industrial Furnaces. Jour, of Insti-
tute of Fuel. April 1960, pp. 180 through 187.
104
-------
10. Peacock, G. R. Land Instrument Co., Private Telephone Communication,
December 11, 1972.
11. Milani, A., and W. Leuckel. Experimental Studies on the Venturi
Pneumatic Pyrometer. International Flame Research Foundation Doc.
nr. F72/a/14, Ijmuiden, Holland, March 1969.
12. Howard, J. B., and R. H. Essenhigh. Pyrolysis of Coal Particles in
Pulverized Fuel Flames. IEC Process Design and Development, Vol. 6,
p. 74, January 1967.
13. Young, W. S. An Arc-Heated Ar-He Binary Supersonic Molecular Beam
with Energies Up to 21 ev. Report No. 69-39. Los Angeles: Depart-
ment of Engineering, University of California, 1969.
14. Alcalay, J. A. and E. L. Knuth. Molecular-Beam Time-of-Flight
Spectroscopy. Rev. Sci. Instr. 40:438, 1969.
15. England, C. Quantitative Evaluation of Reduction Techniques for
Oxides of Nitrogen. Report No. 1200-37. Pasadena: Jet Propulsion
Laboratory, California Institute of Technology, 1972.
16. Fristrom, R. M. and A. A. Westenberg. Flame Structure. New York:
McGraw-Hill, 1965.
17. Chemical Engineers' Handbook. John H. Perry, Editor, Third Edition,
McGraw Hill Book Co., 1950, p. 1589.
105
-------
APPENDIX A.
FLAME MEASUREMENT TECHNIQUES APPEARING
IN THE LITERATURE
LITERATURE SEARCH
In order to determine the measurement techniques most applicable to the
characterization of "dirty" flame environments, a computer search of the
combustion technique was made available at the program outset. The search
included Rockwell International library holdings, as well as NASA and DDC
searches. Among the available techniques for flame characterization re-
ported in the literature are those summarized below.
TEMPERATURE MEASUREMENTS
Thermoelectric Thermometry
Potentially, the most simple temperature-sensing device is a thermocouple.
A Pt/Pt-13 percent Rh thermocouple has been used to measure temperatures
of 500 to 2100 K in swirling propane-air and butane-air flames (Ref. A-l).
The thermoelectric sensor, however, must itself be in thermal equilibrium
with the media being measured. This presents several problems. The
thermocouple material must adequately withstand operation at the media
temperature and cannot participate either directly or as a catalytic agent
in chemical reactions with the surrounding high-temperature oxidizing or
reducing gases. Thermocouple materials such as Ir/Ir-Rh or W/W-Re can
participate directly in reduction or oxidation processes, respectively,
while combinations such as Pt-4 percent Rh/Pt-20 percent Rh do not parti-
cipate directly, but may act in a catalytic manner. Catalytic effects
can have a notable influence on indicated temperature, particularly when
106
-------
there are nonequilibrium chemical species caused by nonequilibrium of
reactions such as N2 + §2 "*" ^NO that could be catalyzed by metallic sen-
sor surfaces. For example, at 20-percent excess air with No. 2 fuel oil,
catalytically forcing completion of the NO formation reaction fendo-
thermic) can reduce the flame temperature by almost 150 F. If a thermo-
couple surface were catalytic to the NO formation reaction, this reduc-
tion in flame temperature would be a localized phenomenon at the surface
of the thermocouple, affecting the temperature indication, but not the
bulk temperature of the combustor. Errors caused by catalytic effects can
be minimized by coating the thermocouple with noncatalytic material and/or
by increasing the rate of forced convection over the sensor. Coating with
materials such as Si02 or A1203 reduces the likelihood of catalytic action.
Conversely, very high rates of forced convection can saturate the catalytic-
reaction-rate capability of the metallic surface, while enhancing heat
transfer from the surface so that the heat transferred to the surface far
overwhelms the endothermic effects of chemical reaction. This results in
the sensor temperature closely approaching the bulk gas temperature in
spite of endothermic reaction catalysis. A combination of coatings and
high-convection rate taken together provides the best assurance that
catalytic effects will be minimized. For the high temperatures found in
flames, the usual Si02 coatings are not usable because of melting and
higher temperature materials, such as A^Oj or NBS ceramic coating A418,
should be used.
An additional problem that arises when the sensor must reach thermal
equilibrium with the high-temperature bulk gas is the loss of heat from
the sensor by radiation to cool walls of the combustor. At temperatures
above 2000 F, unshielded thermocouples can give temperature indications
up to several hundred degrees too low, depending on the absolute tempera-
ture, geometry, temperature of the surrounding walls, opacity of the com-
bustion gas, and local flow conditions. The usual method for reducing
radiation losses from thermocouples is to surround the device with multiple
radiation shields and to attempt to equilibrate the radiation shield and
107
-------
thermocouple temperatures to the bulk gas temperature by forced convec-
tion (Ref. A-2 and A-3). Corrections for radiation losses from the sen-
sor head may also be made based on literature recommendation (Ref. A-4
and A-5).
In the noble metal suction pyrometer combustion gas is aspirated at high
speed through a tube having a thermocouple on its axis, and surrounded
by multiple concentric, ceramic radiation shields. Aspiration of gas
through such a device tends to disturb the normal fluid dynamics of the
combustion process, depending on the aspiration rate, however, it is
helpful in maximizing heat transfer from the gas to the couple and mini-
mizing catalytic reaction errors. Radiation shields are necessary to
minimize radiative heat flow between the thermocouple and the walls of
the combustion chamber or surroundings. According to Ref. A-6, a water-
cooled suction pyrometer with a Pt/Pt-13 percent RH couple should be good
for continuous use up to 1800 C (3280 F). An Ir-Ir/40 percent RH couple
is more brittle, but could operate up to 2000 C (3630 F) , while a W-W/26
percent Re couple could go up to 2800 C (5070 F) but only in a neutral or
reducing atmosphere, all according to Ref. A-6 . Units of the Pt-Pt/13
percent Rh type are available commercially. The design of suction pyrom-
eters is discussed in Ref. A-4- The suction pyrometer is used extensively
in "dirty" flame environments at the International Flame Research Founda-
tion in Ijmuiden, Holland (Ref. A-7).
Reference A-8 reports the application of a platinum resistance thermometer
in an acetylene-oxygen flame at 2200 C (3990 F). It was necessary to make
independent measurement of the emissivity of the platinum surface and take
into account heat transfer and catalytic effects. Commercial resistance
thermometers are said to be available in the range from 600 C (1110 F) up
to 1400 C (2600 F), Ref. A-6 .
The compensated hot-wire method of flame temperature measurement is ap-
plicable to transparent flames in a region where burning is complete and
108
-------
involves compensation by electrical resistance heating of an immersed
wire for radiation and conduction losses. The wire temperature is read
by an optical pyrometer, corrected for wire emissivity, and a vacuum
calibration is needed. This method has been in use for several decades
(e.g., to measure flames at 1800 C (3270 F), Ref. A-9, and is applicable
to transparent flames in a region where burning is complete, so that no
chemical reactions take place on the wire.
The time response of thermocouple or resistance thermometer probes is
dependent upon the geometrical size of the probe and the convective heat-
transfer rate to the probe. Smaller probe size provides more heat-
transfer surface area per unit thermal capacitance (mass), thus allowing
the probe sensor to change temperature more rapidly in response to changes
in flowfield temperature. Greater convection also increases the time
response of the sensor to changes in flowfield temperature. In the case
of radiation shielded temperature sensors, it is not only the mass and con-
vective heat transfer to the sensor that is important, but also the size
and convection heat transfer to the radiation shields. If the turbulent
temperature fluctuations are large, it is necessary that the temperatures
of the radiation shields also follow the fluctuating temperature of the
flowfield, or else a fluctuating radiation error will be introduced into
the temperature measurement. If the flowfield temperature fluctuates
rapidly about some mean value- and if the temperature sensor is small
enough to follow the fluctuations, then the average indicated temperature
will be nearly equal to the true average temperature, but the fluctuations
will be reduced in amplitude because when the thermocouple is hotter than
the sluggish radiation shields it will lose heat to the shields by radia-
tion, and when the thermocouple is colder than the shields, it will be
heated above the local gas temperature by radiation received from the
shields. To obtain best response, it is therefore necessary to minimize
size and maximize convection to both the temperature sensor and the as-
sociated radiation shields.
109
-------
Other, less direct methods of flame temperature measurements such as
pneumatic (choked flow), velocity of sound, optical probes, and molecular-
beam mass spectrometry are available, which allow temperature determina-
tion without the necessity of having a sensor equilibrate to the same
temperature as the gas being measured. Therefore, these methods avoid
the materials problems and radiation corrections characteristic of the
equilibrated devices.
Pneumatic Temperature Probes
A pneumatic probe depends on the application of the continuity equation
to a continuously flowing sample of flame gas and provides an indication
of temperature based on the maximum, choked-flow mass flux through a nozzle.
Usually, two nozzles are used in series (Ref. A-10 and A-ll). The flow
entering the first nozzle is at the local flame temperature, and is not
allowed to undergo any significant cooling during passage through the
first nozzle. The gas flow is cooled in a heat exchanger between the
two nozzle throats. The function of the second nozzle is to determine,
from the easily measured cool stagnation temperature in the second throat,
the product of flowrate and a function of gas molecular weight and specific
heat ratio. The equation by which temperature is determined from pneu-
matic probe measurement is:
2 2
AIMI
v1
01
02
110
-------
or
T
2 2
With
w
W
M.
6
V1
where T. = flame temperature
T_ = cooled gas temperature, entering the second nozzle
P = stagnation pressure entering the nozzle
GW = nozzle discharge coefficient
A = nozzle area
M = gas molecular weight
Y = gas specific heat area
1,2 = subscripts relating the specific parameter to either
the first or the second nozzle
K = a collection of parameters frequently assumed to be
constant
111
-------
The calibration parameter, K, includes the influences of specific heat
ratio, Y» and gas molecular weight, M. In the use of pneumatic probes
for temperature measurement, it is usually assumed that the molecular
weight does not change during the cooling process between the nozzles
and, also, specific heat ratio is assumed constant. Under these assump-
tions, it is not practical to allow partial condensation of species such
as water because this would have strong effect on the molecular weight
(gas density). Similarly, for accurate temperature determination, all
chemical reaction must stop on entry of the combustion gases into the
first nozzle. If suspended solids are present, they must be considered.
Generally, they add mass, but little volume to the gas in the continuity
calculations.
The parameter K is frequently assumed to be, though not necessarily, a
constant. The parameter K is usually determined by calibration in a flow
of a known temperature, though it is possible to calculate K from know-
ledge of nozzle throat areas, discharge coefficients, and gas properties.
The calibration is performed in position using room temperature air. When
the pneumatic probe is calibrated in an inert gas that differs in tempera-
ture from the combustion gas to be ultimately measured, it becomes neces-
sary, for small nozzles, to correct for Reynolds number effects on the
nozzle discharge coefficients, Cw.
The pneumatic venturi pyrometer has been employed in "dirty" flame en-
vironments at the IFRF in IJmuiden, Holland (Ref. A-7).
The temperature-measuring time response of the pneumatic probe is de-
pendent on the volume of the heat exchanger between the two nozzles of
the probe, because it is the volume of this chamber that must be pres-
surized and depressurized to follow temperature fluctuations. The
112
-------
or
T
2 12
With
K
w
]2
M,
r A
w^ Al M^
where T^ = flame temperature
1 = cooled gas temperature, entering the second nozzle
P = stagnation pressure entering the nozzle
C... = nozzle discharge coefficient
w
A = nozzle area
M = gas molecular weight
Y = gas specific heat area
1,2 = subscripts relating the specific parameter to either
the first or the second nozzle
K = a collection of parameters frequently assumed to be
constant
111
-------
The calibration parameter, K, includes the influences of specific heat
ratio, Yf and gas molecular weight, M. In the use of pneumatic probes
for temperature measurement, it is usually assumed that the molecular
weight does not change during the cooling process between the nozzles
and, also, specific heat ratio is assumed constant. Under these assump-
tions, it is not practical to allow partial condensation of species such
as water because this would have strong effect on the molecular weight
(gas density). Similarly, for accurate temperature determination, all
chemical reaction must stop on entry of the combustion gases into the
first nozzle. If suspended solids are present, they must be considered.
Generally, they add mass, but little volume to the gas in the continuity
calculations.
The parameter K is frequently assumed to be, though not necessarily, a
constant. The parameter K is usually determined by calibration in a flow
of a known temperature, though it is possible to calculate K from know-
ledge of nozzle throat areas, discharge coefficients, and gas properties.
The calibration is performed in position using room temperature air. When
the pneumatic probe is calibrated in an inert gas that differs in tempera-
ture from the combustion gas to be ultimately measured, it becomes neces-
sary, for small nozzles, to correct for Reynolds number effects on the
nozzle discharge coefficients, Cw.
The pneumatic venturi pyrometer has been employed in "dirty" flame en-
vironments at the IFRF in IJmuiden, Holland (Ref. A-7).
The temperature-measuring time response of the pneumatic probe is de-
pendent on the volume of the heat exchanger between the two nozzles of
the probe, because it is the volume of this chamber that must be pres-
surized and depressurized to follow temperature fluctuations. The
112
-------
response time of a pneumatic probe is given approximately by the
relationship:
where
y
R
T
g
M
t =
'/Atl
(yRTg)
1/2
a time characterizing the response of the pneumatic
probe
throat area of the probe entry nozzle
specific heat ratio of combustion gas
universal gas constant
combustion gas temperature
gravitational constant
gas molecular weight
cooling chamber volume
For a 1-millisecond response time and typical combustion gas, the above
equation requires the ratio of v/A^, to be approximately equal to 3 feet.
If the cooling chamber were twice the diameter of the throat, the maximum
allowable cooling chamber length would be approximately 8 inches to achieve
the 1-millisecond response time. This does not include any effects of
pressure transducer response time.
Velocity-of-Sound Method
It was suggested 100 years ago (Ref. A-12) that measurement of the velocity
sound in a hot gas would be a way of finding the temperature of the gas
if its thermocynamic properties are known. Measurement of sound velocity
in combustion gases results in the determination of a value of Cyl/M)*/^
where the ideal gas assumption is usually valid because of high tempera-
ture and 1 atmosphere pressure.
113
-------
The ratio y/M is nearly constant, and its estimation should result in
less than 0.5- to 1.0-percent error on the final temperature determina-
tion. Velocity of sound measurements can be made either by noting the
time of passage of a sound wave past two detectors, or by observation of
the physical size of specially induced standing sound wave patterns. In
both of these cases, on the order of 5 to 10 cm of sound wave path length
is required to obtain a precise sound wave determination. This path
length requirement infers very poor spatial resolution on the part of the
sound velocity temperature measurement technique.
The velocity-of-sound method of temperature measurement has been applied
to electric arc flames at 5500 K (Ref. A-13). Also, a method in which
ultrasonic frequencies are used is described in Ref. A-14. It is, of
course, essential to supply thermal protection to the sound source, typi-
cally an oscillating piezoelectric crystal.
Calorimetric Probes
These operate on the principle that the flame gas is aspirated through a
sample tube located on the probe axis and in the process loses heat energy
to the probe. Measurement of the transferred heat flux and of the exit
temperature of the gas sample then gives the inlet temperature (really
enthalpy) of the gas. In the steady-state version (Ref. A-15) all the
heat transferred from the gas supply is absorbed by the coolant flowing
through the probe body. A tare coolant temperature rise is measured with
the probe immersed in the stream, but not aspirating a gas sample. This
tare must be subtracted from the coolant temperature rise when a gas
sample is normally aspirated. A "split-flow" probe (Ref. A-16) reduces
the magnitude of the tare and claims higher accuracy. Another design
(Ref. A-17) introduces an air gap between inner and outer coolant jackets
to reduce probe heat losses. In the transient version (Ref. A-18) the
walls of the sampling tube are insulated from the probe body, and the
heat flux is obtained from transient temperature measurements along the
sampling tube.
114
-------
Optical Probes
Several optical techniques provide well-understood means for local temper-
ature determinations without disturbing combustor physical processes.
Methods of interferometry, pyrometry, radiometry, and absorption are all
applicable to flame temperature measurement.
The simplest optical technique is pyrometry, in which essentially the flame
brightness is measured by comparison with an object of known temperature,
and related thereby to gas temperature. The absolute radiation pyrometer
essentially finds the emissivity of the flame (and hence the temperature
from the black-body radiation law) by viewing the flame against a mirror
and also against a black background. Alternately, the apparatus may be
calibrated against a standard tungsten lamp. References A-19 and A-20
are examples of this method, with photoelectric detection devices. This
pyrometer is suitable for temperatures up to 2500 C (4570 F), according
to Ref. A-6 . In the two-color pyrometer, the flame is observed in turn
through two optical filters, each yielding a monochromatic color tempera-
ture. From the two observations, the flame temperature and absorptivity
are determined.
A most obvious problem with pyrometry is the optical transparency of the
flames at the visible wavelengths. For nearly transparent flames, pyro-
metric measurements must be made on volumes of gas large enough not to be
"seen" through by the pyrometer to ensure that the temperature measurement
turns out to be a gas temperature rather than a wall temperature for the
opposite combustion chamber wall. The requirement for long optical paths
essentially means that optical pyrometer techniques have poor spatial
resolution of temperature. This poor spatial resolution can be improved
by seeding the flame with material that reaches thermal equilibrium with
the gas and emits strongly in the visible wavelengths. If solid particu-
late matter such as talcum powder (which emits as a gray or blackbody)
is added, the flame becomes much less transparent and the usual flame
pyrometer can be used directly with only a small correction for particulate
115
-------
size. If other materials (which vaporize and emit in spectra consisting
of discrete lines) are added, and the brightness is measured only at a
discrete line, then this pyrometric technique becomes known as the "line
reversal" technique. In particular, if the additive is sodium ion, and
the D line is selected, the method is a Na-D line reversal, the most com-
monly used line reversal method.
The sodium line reversal method has been in use for over 50 years (Ref.
A-2l)- This method basically consists of introducing sodium onto the
flame (e.g., by addition of small amounts of a sodium compound to the
fuel) and shining a tungsten or xenon lamp through the flame and further
onto the slit of a spectroscope. The electrical current to the lamp is
adjusted until the sodium D doublet line seen through the spectroscope
matches the brightness of the lamp background. Then the flame tempera-
ture is that of the lamp filament as read by an optical pyrometer. Numer-
ous refinements have been devised, in some of which an interferometer
takes the place of the spectrometer. This method is feasible for very
hot flames, well above the 4500 F level. A temperature profile across a
flame may be obtained by varying the optical depth.
Where additives must be used, line reversal is more desirable than addi-
tion of particulates because wideband or continuum emission from particu-
lates can easily lead to significant cooling of the gas, which would not
occur in the absence of the particles. Line reversal techniques, where
energy is emitted only at a few specific wavelengths, do not drain as
much thermal energy from the gas and, therefore, radiantly caused changes
in temperature caused by the additive are not significant. Conversely,
most line reversal method additives do not emit significantly below 2540 F
and so the line reversal technique is not useful below this temperature.
Molecular Beam
Translational temperatures in gas flames can be determined directly from
measurement of molecular speeds. This is accomplished by means of a
116
-------
molecular-beam sampling apparatus in conjunction with a time-of-flight
mass spectrometer (Ref. A-22), The molecular-beam sampling apparatus
causes gas withdrawn from a combustion process to undergo a very rapid
expansion to supersonic velocities. The molecules that remain on or
near the axis of this expanding flow are naturally those molecules whose
velocity was directed in the axial direction at the end of transition
from continuum to free molecular flow. The velocity of these molecules
is statistically the same as the randomly directed molecular velocities
of the gas when it was in the combustor, except that in the molecular
beam, the velocity has been transformed so that it is essentially all in
the axial direction. By intermittently chopping the molecular beam and
measuring the time required for a chopped portion of the beam to arrive
at the detector site in a time-of-flight mass spectrometer, it is possible
to determine the average axial velocity of particular molecular species
in the beam, and because this velocity is numerically equal to the ran-
domly directed thermal velocities in the main combustion gas stream, the
measured velocity is a direct indication of the mainstream temperature.
The molecular-beam method offers the advantage of temperature measurements
without the need of blackbody or any other sort of temperature calibra-
tion. The calibration required is one of time (used in molecular-velocity
determination), which can be performed very accurately. The molecular-
beam technique also offers excellent response, well under one millisecond.
However, the molecular-beam probe is generally associated with relatively
large probes (90-degree cone typical), which might disturb flow patterns.
STABLE CHEMICAL SPECIES
The determination of stable-chemical species within a flame is generally
conducted by either probe sampling or spectroscopic techniques.
117
-------
Sampling Probes
Probe sampling techniques must make allowances for:
1. Continuous operation in a high-temperature environment
2. Minimizing physical disturbances to the combustor fluid
dynamics and physical processes
3. Rapidly quenching the chemical reactions so that the composition
of the gas analyzed is not influenced by reactions occurring
within the probe.
Water-cooled sampling probes to measure flame composition are available
as shelf and as specialty items from a number of suppliers. To approach
sampling at a point, miniaturized probes are desirable. Miniaturization
is expensive and also fraught with constructional difficulties in that
it is hard to avoid coolant and gas leaks through joints of thin-walled
metals exposed to relatively high temperature and often corrosive gases.
This is particularly true of the probe tip, which is the least coolable
part of the probe and the one that must assume special shapes.
Figure A-1 shows three important probe tip design features. To get true
samples at a point in a flame, the velocity of the sample through the
probe should equal the stream velocity. This is especially important
when solids are suspended in the sample. Such isokinetic sampling is en-
sured by matching probe inlet static pressure with stream static pressure
by proper adjustment of the probe aspiration. In Fig. A-l, (a) shows that
when this is done, the captured sample stream tube then equals the probe
inlet size. When the sample must be quenched rapidly to freeze reactive
consituents, a rapid-expansion inlet (b) and/or a diluent-quench inlet
(c) can be used (Fig. A-l). In both cases, the temperature of the sample
is reduced in microseconds to a level at which components of interest are
fixed. The features discussed have been fabricated as part of specialty
sampling probes.
118
-------
Static
Pressure
B i '7\~ *-
/ '
C^. Adjustable Sample Aspiration
(a) Isoklnetlc Sampling
(B Is correct; A, C are incorrect)
Sample
(t>) Rapid. Expansion Inlet
COLD INERT DII.tfBlT GAS
-C*--
S ample
COLO IHERT~DIUJENT GA§"
(c) Diluent Quench Inlet
Figure A-l. Features of composition sampling probes
119
-------
An ordinary water-cooled probe (United Sensors) has been used success-
fully at Rocketdyne (Ref. A-23) to analyze the composition of typical
distillate oil-burner flames. The results obtained are considered only
near-quantitative in nature since the water-cooled probe does not offer
the very short quench times required to absolutely ensure that the gas
analysis corresponds exactly to the same concentrations as existed in the
original flame.
Molecular Beam/Mass Spectrometer
Another method of rapidly quenching chemical reactions is to use a mole-
cular-beam gas-sampling apparatus. As a sample removal technique, the
molecular beam offers the most rapid possible quenching of chemical reac-
tions. For a molecular-beam sample apparatus, gas is expanded from the
zone of combustion through a small pinhole, as shown in Fig. A-2 into a
high vacuum. This results in a Prandtl-Meyer expansion to supersonic
speeds and low temperatures over a period of time as short as 0.1 micro-
second. This aerodynamic expansion is continued until free molecular flow
is achieved, and a portion of the expanded jet is skimmed and collimated
to form a molecular beam for analysis by mass spectrometer. As described
previously, this technique also is applicable for measurement of tempera-
ture, and it also is applicable to the determination of unstable species
as described later.
Spectroscopy
Spectroscopic techniques offer additional methods for determination of
chemical species in flame structures. Both emission and absorption
characteristics can be used to identify various chemical elements within
the flame. The infrared spectrum of the molecules in an air-oil flame
are well known (Ref. A-24) at ambient conditions and recent work has made
important contribution to the spectra at high temperatures. The spectrum
of water will be dominant and carbon dioxide will also have some very
strong bands. The band centers of the molecules of interest are listed
in Table A-l.
120
-------
H
"O
0)
rt
i <
M-i
O
P
CD
eu
erf
in
X
CO
bO
C
00
rt
0)
OQ
i
f-i
rt
it
3
O
0)
0)
^H
M
H
UH
121
-------
The species N2, 02, H2 anc* ^r are not gi-ven because they are transparent.
Table A-l. SPECTRA OF SOME SPECIES IN AN OIL FLAME
H20
co2
CO
S02
N02
NO
Vl
Vl
Vl
Vl
3657 A°
a
2150
1151
1320
1890
V2
V2
V2
V2
1593
667
519
648
V3
V3
V3
V3
3756
2349
1361
1621
Forbidden
The vibration-rotation bands of atmospheric pressure have a half-width of
about 0.1 cm"-'-. In low resolution spectra, the water and carbon dioxide
would appear opaque throughout the middle infrared region (Ref. A-25) at
temperatures over 2000 F. At high resolution, however, there would be
very many small windows except near the water and carbon dioxide band
centers (Ref. A-26) Use of high resolution is essential then to resolve
the lines in emission. It is essential in absorption spectroscopy since
the apparent intensity decreases rapidly when the resolution is less than
line width. Achieving this resolution requires a monochromator with a
focal length of at least one meter and an echelle grating at least 15 cm
wide. Qualitative arguments indicate that, for instance, NO requires a
pressure path length product of about 1 cm torr to be visible in absorp-
tion with a signal/noise ratio of 100. The limiting sensitivity then is
better than 1 ppm for a burner 30 cm long.
Spectroscopic techniques offer the advantage of essentially zero inter-
ference with the physical and chemical processes occurring within the
122
-------
flame. However, to avoid such interference two alternatives are
available:
1. Line of sight average path measurements with external optics
2. "Local" or "point" measurements with external optics
Average path measurements are not acceptable for nonuniform flowfields.
For "local" or "point" measurements with external optics, a complex sys-
tem is required and an entire matrix of data must be obtained.
UNSTABLE CHEMICAL SPECIES
Unstable species are, by their very nature, more difficult to determine
than the stable species. The unstable species are very difficult to
sample because their high reactivity leads to extremely short lifetime.
Only two practical techniques for determination of the unstable species
exist. These are molecular-beam sampling and spectroscopic methods.
Molecular Beam
Molecular-beam sampling offers very rapid transition of sampled material
from the combustion zone to free molecular flow and is applicable to
sampling of all of the unstable species of interest to the proposed pro-
gram. Once the sample enters the free molecular flow regime, the mole-
cules and radicals undergo no further reaction (other than perhaps spon-
taneous disintegration), because there are effectively no further colli-
sions with other species before completion of mass spectrometric analysis
of the sample.
Spectroscopy
Application of emission spectroscopy is difficult because emission depends
not only on specie identity, but also on the mode and degree of excitation
(temperature) of that specie. Since emission spectroscopy can be used
123
-------
only for vibrationally excited states, absorption techniques must also be
employed to measure ground-state species that may dominate the distribu-
tion. However, because of limited available wavelengths for study of
various species, the only radical specie that has been quantitatively
studied in detail by this method is OH. For other radical species, there
are no known convenient wavelengths at which detailed quantitative de-
termination of species concentration are possible in the presence of the
carbon continuum. Another problem arises from optical thickness and
temperature gradients in thick-flame structures that tend to complicate
the interpretation. The disadvantages of painstaking apparatus setup
requirements and interpretation of experimental results are applicable to
unstable species as well as stable species.
VELOCITY MAGNITUDE
Gas velocities can be measured by several techniques, depending on the
particular environment of interest. Methods include pitot tube, hot-wire
anemometry, mechanical anemometry, and pulsed tracer.
Modified Pitot Probes
Commercially available pitot probes can be used to obtain accurate velocity
measurement in high-temperature gas streams. The pitot probe has an im-
pact tap at the probe head and one or more taps located along the side of
the probe. A measure of the local static pressure is obtained through the
side tap. The dynamic pressure, which is a direct measure of the gas
velocity, is then determined by use of a differential pressure measurement
between the static pressure side tap and the impact probe tap.
Rocketdyne (Ref. A-23) has employed small pitot probes in dirty, but rela-
tively low oil droplet density, combustion flames from domestic oil burn-
ers (distillate fuel oil). It was found that an 0.085-inch-ID, water-
cooled pitot probe used in these dirty flames seldom plugs (mostly be-
cause no net flow passes through the probe), but occasionally suffers
124
-------
from the formation of oil meniscuses across the small diameter passages
of the probe. To avoid errors caused by this effect, a purge system is
installed to purge both the impact and the static pressure ports of the
pitot probe before and after each measurement. This purging ensures that
spurious pressures are not obtained as a result of pressure differences
across the thin miniscuses.
In dense, two-phase (liquid-gas) flowfields the momentum of the liquid
spray introduces errors in dynamic-pressure measurements that are used to
infer gas velocities. Conventional gas-phase stagnation probes (pitot,
etc.) are thus not applicable in dense two-phase sprayfields because of the
interaction of the gas and liquid droplets near and within a conventional
stagnation probe. A two-phase probe was developed at Rocketdyne (Ref.
A-27) for measurement of local values of both liquid and gas mass fluxes
in nonburning media. This probe is patterned after the Dussourd-Shapiro
two-phase flow impact probe developed at MIT (Ref. A-28J.
A schematic of the Rocketdyne probe, is presented in Fig. A-3. The probe
was constructed of two concentric tubes (A and B) with a specially de-
signed tip attached to tube B. The tip was designed to prevent the pass-
age of liquid into the annulus formed by tubes A and B when the probe is
used in high-mass-flow-ratio flowfields.
The operating principle for the determination of the gas-phase stagnation
pressure by the concentric tube two-phase impact probe is illustrated in
Fig. A-4. Basically, the intent is to decelerate the gas and measure the
gas-phase stagnation pressure in a manner that minimizes momentum exchange
from the condensed phase upstream of the measurement location. Droplets
and gas (each at their own velocity) encounter the probe tip but the gas
phase is stagnated at the probe tip where the pressure is approximately
equal to the gas-phase stagnation pressure. Deviation from true gas-phase
stagnation pressure is due to momentum exchange between the droplets and
125
-------
PROBE TIP
o.gas
TUBE B (3/8-INCH 00)
TUBE A (1A-INCH 00)
o,gas
STAGNATION CHAMBER
DRAIN VALVE
Figure A-3- Schematic of concentric tube two-phase impact probe
126
-------
O
f-l
P-,
(U
10
nj
I
O
c
O
H
+-1
cj
^
0)
d,
O
fl,
H
O
c
H
!H
O.
-------
the gas in the near flowfield of the probe tip (termed overpressure error).
A droplet passes through the probe tip and is decelerated to zero velocity
in the stagnation chamber formed by tube A. However, because of momentum
exchange between the particles and the stagnated gas, the particles de-
celerate in the probe tip to some extent over the distance X (Fig. A-4).
The gas-phase stagnation pressure, P , as measured in the probe annulus
is greater than the gas-phase stagnation pressure, P . The difference
o, tip
between the two aforementioned pressures can be made small if the distance
X is minimized. However, the total over-pressure error (caused by parti-
cle/gas momentum exchange both near and within the probe tip) can be de-
termined by proper calibration of the probe in known two-phase flowfields.
Hot-Wire Anemometry
Hot-wire anemometry is another method for gas-velocity determination. For
application to combustion processes, hot-wire anemometry is subject to
considerable error because of flame temperatures and possible impingement
of liquid droplets on the wire. The hot-wire techniques, therefore, are
applicable only for nonspray cold-flow studies, or for use in zones of
the combustion where ignition has not yet taken place and there are no
liquid droplets or large, potentially wire-damaging particles in flight.
The hot-wire technique offers excellent time resolution of velocities, so
that it is readily feasible to study turbulence characteristics of the
flow.
Mechanical Anemometry
Mechanical anemometry involves determination of velocities by the capabil-
ity of the gas flow to cause movement of small, windmill-like structures.
The most common application is in air-conditioning and heating-duct ad-
justment. For use with air, very sensitive mechanical anemometers are
available at low cost from, for example, Anor Instrument Company^ and are
capable of measuring velocities as low as 1/3 ft/sec. Mechanical anemom-
eters can be used with probes similar in design to an ordinary impact
128
-------
tube with air flow through a tube from the probe to the windmill indicat-
ing device. For use with combustion zones, mechanical anemometers
require:
1. Calibration as a function of gas density
2. Parallel gas temperature and composition measurements
3. Accounting for gas temperature losses in the probe line to the
windmill
4. Construction from corrosion-resistant materials
5. Capability for rapid and frequent cleaning to remove internal
hydrocarbon and soot deposits.
Disadvantages of the mechanical anemometer are poor time resolution of
velocity, typically a large probe size, and susceptibility of mechanical
windmill bearings to the effects of dirt.
Pulsed Tracer
Pulsed-tracer techniques can be used to measure magnitude and/or direction
of combustion gas velocity. An advantage of pulsed-tracer techniques is
that velocity magnitude can be determined without knowledge of gas density.
Therefore, it is not necessary to determine gas temperature and composi-
tion to obtain velocity. Also, pulsed-tracer techniques can easily be
used in the dirty, sooty flames without unknown sacrifice of accuracy,
because it is generally obvious when the apparatus is dirty enough to af-
fect the velocity data.
Depending on the particular pulsed-tracer technique selected, care may be
necessary to avoid unduly disturbing the flow patterns in the combustion
process by the system for injecting the tracer, by the presence in the
combustion process of the tracer itself, or by the downstream detector
129
-------
for the tracer. The pulsed-tracer technique involves injecting some
easily detected disturbance into the combustion flow, and determining
the time required for that disturbance to propagate to a position of
known distance downstream. The pulsed tracer can consist of material
such as a gas, which can be detected by thermal conductivity or density,
or by thermal emission. The tracer can also consist of particulate ma-
terial of small particle size selected so that the particle velocities
closely follow the gas velocity. More commonly, the tracer is a non-
material quantity such as thermal energy or electrical energy. For ex-
ample, electrical current can be pulsed through a thin wire to generate
heat in the wire, which in turn is passed on to the gas. This thermal
energy can then be detected downstream by a thermocouple or by a. resist-
ance thermometer, providing the amount of thermal energy is sufficient to
allow detection over and above the random thermal fluctuations caused by
the normal combustion processes. For combustion processes, a more appro-
priate nonmaterial tracer is ionization energy, supplied to the combustion
gas in the form of an electrical spark. The high energy of an electrical
spark ionizes many molecular species in the gas, and these ionized species
flow with the gas past the spark source on downstream to a detection de-
vice. For detecting the pulse of ionized gas, simple detectors based on
electrical capacitance or electrical conductance are applicable. A pri-
mary disadvantage of pulsed-tracer techniques is that it is necessary to
know in advance, or to determine by trial and error, the direction of the
gas flow before the tracer technique can be used to determine velocity
magnitude. This is not true, of course, if the tracers are followed photo-
graphically. In which case, the optical problems become almost as diffi-
cult as the trial and error determination of gas flow direction. In gen-
eral, the tracer techniques are not affected by the presence or absence
of solid or liquid particulates in the gas flow.
VELOCITY DIRECTION
There are several techniques applicable to the determination of velocity
direction in a gas flow. These techniques include tracer methods, multiple-
port impact probes, and multiple crossed hot-wire anemometers.
130
-------
Tracers
The use of tracer techniques in combination processes has already been
discussed with respect to velocity magnitude determinations, but this
method also is applicable to velocity direction determination. Gas
ionization, added gas, added particles, or thermal energy as the tracer
are all applicable to velocity direction determination. The trial and
error process of direction determination by tracers can, in some cases,
be simplified by visual observations of the dirty-gas flow or by visual
observation of injected tracer. For velocity direction, the added tracer
does not need to be added in a pulsed manner and, therefore, with con-
tinuous tracer addition at one point, it is possible to trace or follow
the streamline for a significant distance before random turbulence totally
disintegrates the streamer of tracer. Disadvantages noted previously,
such as flow disturbances caused by the tracer addition mechanism, the
tracer detector, and the tracer itself are also of concern for both
velocity-magnitude and velocity-direction measurements.
Multiple-Port Impact
Multiple-port impact probes are one of the more generally applicable and
reliable methods for determination of flow direction. These probes made
use of the fact that the reading obtained with an impact probe changes
most rapidly with angle when the probe axis is at 45 degrees to the di-
rection of gas flow. For velocity known to be in a single plane, the di-
rection can be obtained by balancing the impact-pressure readings from
two probes having their axes intersect at 90 degrees to each other in the
plane of the velocity. The balance is obtained by maneuvering the orien-
tation of the combined probes. For velocities not known to be in the same
plane, two such pairs of impact tubes can be used. For this four-tube
probe, the pressure readings are balanced for one pair at a time by maneu-
vering the orientation of the probe in the plane of the pair being balanced,
Consecutive balancing of the two pairs of probes will quickly result in
131
-------
obtaining the compound angle of the flow velocity in a three-dimensional
field. Because of the high sensitivity of the impact probe when at 45
degrees to the flow direction and because of the difficulty in making
individual probes of a pair exactly identical, multiple-impact probes
require calibration by a gas velocity stream of known direction. At
velocities in excess of about 50 feet per second, the multiple-impact
tube probe is capable of indicating gas velocity to within less than a
1-degree angle, but at velocities of the order of 5 ft/sec, when the flow
is relatively turbulent, the resultant lower impact pressures makes it
difficult to resolve velocity direction to much better than about ±2- or
±3-degree angle. To achieve velocity direction measurements with the
multiple-impact tube probe requires that the probe be mounted on a gimbal-
ing apparatus capable of rotating the probe through compound angles in
each of the planes corresponding to the two pairs of impact probes. The
gimbaling apparatus must be such that the tip of the multitube probe re-
mains stationary while the orientation angles are changed. An advantage
of this multiple-impact tube probe is that it requires only the indication
of a null differenct between the impact pressures indicated by opposing
tubes and, therefore, it is necessary to have only a highly sensitive
pressure indicator and not a highly accurate, highly sensitive indicator.
The use of a gimbaling apparatus can be avoided if a spherically or hemi-
spherically-shaped multiple-port pitot probe is employed (Ref. A-29) . Con-
sider, for example, the case of a spherical body suspended in a gas flow,
in a fixed orientation, for which the distribution of pressure over the
sphere surface as a function of distance from the forward impact stagnation
point is well known. By measuring the pressure at four locations on the
spherical surface, it is possible to locate the stagnation point and,
therefore, the direction from which the gas flow is coming at the sphere
as well as its magnitude. The same principle applies to multiple-port
impact probes of any tip geometry, but it is generally necessary to use
empirical calibration curves previously obtained for the particular probe
to account for imperfections in symmetry and theoretically difficult-to-
handle geometries. Use of this nongimbaled technique does have the
132
-------
advantage of not requiring the somewhat difficult-to-construct gimbaling
apparatus, but it has the disadvantage of requiring quantitative impact-
pressure measurements at several ports. Obtaining these impact-pressure
measurements with sufficient accuracy to interrelate the multiple read-
ings, and thereby determine velocity magnitude and direction, is generally
not difficult with high-velocity flows (~50 ft/sec), but it becomes dif-
ficult at lower velocities (~10 ft/sec), particularly if the flow is some-
what turbulent, as is usually the case in combustion devices.
A five-hole hemispherical pitot probe is used by the IFRF for the deter-
mination of velocity magnitude and direction in "dirty" flame environ-
ments (Ref. A-7).
Multiple Crossed Hot-Wire Anemometers
Multiple crossed hot-wire anemometers offer an additional method for the
determinations of velocity directions. Determination of direction with
hot-wire anemometers is based on the different magnitude of heat conduc-
tion from the wire to the gas depending on whether the wire is located
perpendicular or parallel to the direction of flow. Heat transfer away
from the wire tends to be maximized when the wire is perpendicular to the
flow direction. Therefore, maximizing the heat transfer from two non-
parallel wires will define the plane that is perpendicular to the gas-flow
direction. Such multiple hot-wire anemometers work well in noncombustion
environments; however, for application in hot, turbulent, dirty combustion
environments, they are generally not acceptable.
133
-------
REFERENCES
A-l. Chervinsky, A., et al. TAE-56 Experimental Investigation of
Turbulent Swirling Flames. September 1966.
A-2. Landenburg, R. W., B. Lewis, R. N. Pease, and H. S. Taylor (ed.).
Physical Measurements in Gas Dynamics and Combustion, Vol. IX.
Princeton Series on High Speed Aerodynamics and Jet Propulsion.
Princeton University Press, 1954.
A-3. International Combustion Symposia (III (1948) through XIII (1970)),
organized by the Combustion Institute. Pittsburgh, Pennsylvania.
A-4. Laud, T., and R. Barber. The Design of Suction Pyrometers. Trans.
Soc. Tust. Tech. (London). 6_, No. 3, 112-130, September 1954.
A-5. Fristrcm, R. M. Experimental Techniques for Study of Flame Struc-
ture. Report No. 300. Applied Physics Lab, Johns Hopkins Univer-
sity. January 1963.
A-6. Barber, R. Review of High Temperature Measurement Methods. The
Chemical Engineer (London). April 1967.
A-7. Chedaille, J., and Y. Braud. Industrial Flames. Vol. 1: Measure-
ments in Flames. New York, Crane, Russak £ Co., Inc., 1972.
A-8. Gilbert, M., and J. H. Lobdell. Resistance-Thermometer Measure-
ments in Low-Pressure-Flame. Presented at the Fourth Symposium of
the International Combustion Institute, IV, 285 (1952).
A-9. Griffiths, E., and J. H. Awbery. Proc. Roy, Soc. 123, 401 (1929).
A-10. Holland, R. E., R. Jackson, and G. G. Thurlow. The Behavior of
the Venturi Pneumatic Pyrometer in Industrial Furnaces. J. Inst.
Fuel. 180-7, April 1960.
A-ll. Grey, J, Thermodynamic Methods of High-Temperature Measurements.
Trans. Inst, Soc. of America. 4_, 102-15, 1965.
A-12. Mayer, A. M. Phil. Map. 45_, 18, 1873.
134
-------
A-13. Suits, C. G. J. App. Phys. 6_, 190, 315, 1935.
A-14. Marlow, D. G., C, R. Nisewanger, and W. M. Cady. J. App. Phys.
_20, 771, 1949.
A-15. Grey, J., P. F. Jacobs, and M. P. Silverman. Calorimetric Probe
for Measurement of Extremely High Temperature. Rev. Sci. Inst.
JS3_, 738-41, July 1962.
A-16. O'Connor , T. J. A Split-Flow Enthalpy Probe for Measurement of
Enthalpy in Highly Heated Subsonic Streams. Inst. Soc. of America
Papd. 68-539, Annual Meeting, New York, October 1968.
A-17. Grossman, R. D., and Associates. Cooled Aspirating Calorimetric
Probes. Canoga Park, California, 1971.
A-18. Vassiallo, F. A. Miniature Enthalpy Probes for High Temperature
Gas Streams. Report No. ARL 66-0115. USAF, June 1966.
A-19. Hett, J. H., and J. B. Gilstein. J. Opt. Soc. Am. 39_, 909, 1949.
A-20. Bundy, F. P., and H. M. Strong. Phys. Review, 76_, 457, 1949.
A-21. Kohn, H. Annelen der Physik. 44, 749, 1914.
A-22. Alcalay, J. A., and E. L. Knuth. Molecular-Beam Time-of-Flight
Spectroscopy. Rev. Sci. Instr. 40:438, 1969.
A-23. Dickerson, R. A., A. S. Okuda, and C. L. Oberg. Design of an
Optimum Oil Burner for Control of Pollutant Emissions. Report
No. R-9465. Rocketdyne Division, Rockwell International, Canoga
Park, California, February, 1974.
A-24. Herzberg, G. N. Molecular Spectra and Molecular Structure, II:
Infrared and Raman Spectra of Polyatomic Molecules. D. Van
Nostrand Co., Inc., New York, 1945, I. Spectra of Diatomic Mole-
cule, 2nd Ed. 1950.
A-25. Study on Exhaust Plume Radiation Predictions, Final Report, under
Contract NAS8-11363, General Dynamics. GDC-DBE66-017. December
1966.
A-26. Herget, W. F., J. S. Muirhead, and S. A. Golden. Band Model Param-
eters for H20. Rocketdyne Division, Rockwell International, under
Contract NAS8-20397, R-6916. Canoga Park, California. January
1967.
135
-------
A-27. Burick, R. J., C. H. Scheuerman, and A. Y. Falk. Determination
of Local Values of Gas and Liquid Mass Flux in Highly Loaded Two-
Phase Flow. Symposium on FlowIts Measurement and Control in
Science and Industry, Pennsylvania (to be published in symposium
proceedings). Paper No. 1-5-21.
A-28. Dussourd, F. L., and A. H. Shapiro. A Deceleration Probe for
Measuring Stagnation Pressure and Velocity of a Particle-Laden Gas
Stream. Jet Propulsion. January 1958. p. 24-34.
A-29. Lee, J. E., and J. E. Ash. A Three-Dimensional Spherical Pitot
Probe. Trans. ASME. Vol. 78: April 1956. p. 603-608.
136
-------
APPENDIX B
REVIEW OF
SELECTED LITERATURE ON FLAME CHARACTERIZATION
BY MOLECULAR-BEAM SAMPLING
Prepared by K. Gorji
Molecular-Beam Laboratory
Energy and Kinetics Department
School of Engineering and Applied Science
University of California
Los Angeles, California 90024
Prepared for Environmental Protection Agency Under Contract No. 68-02-0628
137
-------
I. INTRODUCTION
The desire to observe reactive species in flames has been one of the
chief motivations for developing molecular-beam-sampling techniques. It
is the objective of this paper to review the existing technology for mea-
surements of reactive-species concentrations in flames using the
molecular-beam sampling technique, and for measurements of flame temper-
ature employing the time-of-flight technique. Composition measurements
are reviewed first, and temperature measurements last.
II. COMPOSITION MEASUREMENTS
MECHANISM OF BEAM FORMATION
For the past few years, molecular beams for sampling from steady sources
have been studied extensively, both analytically and experimentally. As
shown in Fig. B-l, a typical molecular-beam mass-spectrometer sampling
system consists mainly of a source, source orifice, source chamber, skim-
mer, collimating chamber, and a detection chamber which contains the beam
detector. The sampling gas is accelerated from the source, at pressures
up to several atmospheres and temperatures up to several thousand K, via
the sampling orifice into a highly evacuated source chamber at pressures
from 10~1 to 10~3 torr. The core of the supersonic jet is then trans-
ferred, via the skimmer and collimating orifices, to the detector.
In theoretical treatments of the molecular beam obtained from a nozzle
source, it is usually assumed that the flow between the source orifice
and the skimmer orifice is continuum flow with isentropic expansion of
the gas through the source orifice. It is also assumed that the specific-
heat ratio Y is constant during the expansion, the flow into the skimmer
orifice is supersonic, the flow is undisturbed by the presence of the
skimmer, molecular collisions occurring after the skimmer orifice are
negligible, and the gas composition is the same in the detector and the
138
-------
SOURCE
SOURCE
CHAMBER
COLLIMATION
CHAMBER
DETECTION
CHAMBER
SKIMMER
COLLIMATING
ORIFICE
MOLECULAR
BEAM
g
CO
g
CO
Figure B-l. Schematic Diagram of Molecular-Beam Sampling System
139
-------
source. When designing and using sampling systems for high-pressure and
high-temperature sources, various departures from the above assumptions
must be considered. The principal causes of these departures are:
1. Nonequilibrium processes and shock waves
2. Skimmer interference
3. Mass separations
4. Nucleation
5. Background scattering
6. Molecule fragmentations in the detector
Nonequilibrium Processes and Shock Waves
When a gas sample expands through the nozzle, simplifications are real-
ized if the expansion time is short in comparison with the reaction times
for the unstable or reactive species. Hence, information on the tempera-
ture and pressure histories for the expansion is required. This informa-
tion can be obtained from subsonic and supersonic calculations. In study-
ing 1-atmosphere flames, Greene (Ref. B-l) found that, for a monatomic
gas at 2000 K and at 1 atmosphere expanding through a 0.125 mm orifice,
the pressure drops to 10~-> atmosphere and the temperature to less than
100 K within 1.5 microseconds after the first significant pressure and
temperature changes.
In the expansion process, the gas obtains high Mach numbers, usually above
10. The expansion continues until its density becomes less than the sur-
rounding gas; a "barrel shock" around the jet and a "Mach disk" shock per-
pendicular to the direction of flow is generated. If the sampling gas is
allowed to pass through this shock, its composition will be altered due
to the change of its density and temperature. Therefore, it is necessary
to place the skimmer upstream of the Mach disk.
It is generally accepted that the expansion downstream of the sampling
orifice within the "barrel shock" is isentropic. Sherman (Ref. B-2) has
140
-------
calculated values of the Mach number as a function of location in the
jet for three different values of Y.
In a typical free-jet expansion of a gas from 1 atmosphere, an individ-
ual molecule undergoes several hundred collisions. If the vibrational,
rotational, and translational degrees of freedom are all active initially,
the number of collisions required for relaxation is greatest for vibra-
tion, least for translation. Therefore, as the gas expands in the jet,
the vibrational degrees of freedom freeze first, the translational de-
grees of freedom last. Possible effects of chemical relaxations in the
free jet have been studied by Knuth (Ref. B-3).
Skimmer Interference
The skimmer may interfere with the properties of a molecular beam formed
from a sampling kit. This interference may result in
1. A decreased beam density
2. An increased velocity distribution width
3. A decreased mean velocity
4. Distortion of the beam composition.
The first three effects have been studied by several investigators, e.g.,
Anderson et al. (Ref. B-4). Composition distortions due to skimmer in-
terference have been reported by Reis and Fenn (Ref. B-5) and Young et al.
(Ref. B-6). The skimmer interference may arise from the Mach disk, shock
detachment, the stand-off shock, small internal skimmer angles, a blunt
leading edge, and a large skimmer orifice. Knuth (Ref. B-3) has reviewed
criteria for minimizing this skimmer interference.
Mass Separations
Several investigators have observed that when a mixture of gases is ex-
panded to form a supersonic jet, the composition of the gas taken from
the core of the jet is depleted in the lighter component. This mass
141
-------
separation has been investigated by several investigators, e.g., Becker
et al. (Ref. B-7), Waterman et al. (Ref. B-8 and B-9), Abuaf et al. (Ref.
B-10), Milne and Greene (Ref. B-ll). Young et al. (Ref. B-12) have
studied the most important possible separations, namely mass separations
due to pressure diffusion, skimmer induced separation, Mach-number focus-
ing, and background invasion by varying (1) the source pressure, (2) the
orifice-skimmer distance and (3) the background pressure. Criteria for
either avoiding or minimizing distortion of gas composition during the
sampling process are available.
Nucleation
When the temperature and pressure drop during a free jet expansion, con-
densation of one or more species may occur. Hence studies of nucleation
in supersonic molecular beams are highly motivated.
Studies in free-jet nucleation have mostly involved observations of the
final extent of nucleation, or cluster growth, after the collisions have
ceased. It would be better if one could probe through the continuum
flowfield with the skimmer., having demonstrated that collisions effec-
tively cease at the skimmer, and thus follow the growth of clusters as a
function of time.
Greene and Milne (Ref. B-13) noticed that, when argon at atmospheric
pressure was passed over water at 4 C and expanded through a 0.004-inch-
diameter sampling orifice, the majority of the water in the free jet was
nucleated to form dimeric and higher species. They suggest that the cor-
rection for nucleation could be made by extrapolating a plot of intensity
versus orifice diameter to zero orifice diameter. (The rate of expansion
is inversely proportional to the orifice diameter.) Other investigators
have studied nucleation of Ne, Ar, C02, 02, N20, N2, NH3, H20 (Ref. B-14),
Ar, C02, N2, 02, H20 (Ref. B-15), Hg, CsCl, CH3OH (Ref. B-16), and NO, H ,
02, Ar, C02 (Ref. B-17).
142
-------
Background Scattering
In sampling studies, the scattering of beam molecules by background mole-
cules can become a problem. Anderson and Fenn (Ref. B-18) investigated
the interaction of background gases and supersonic free jets from sonic
nozzles and found out that, at sufficiently low jet densities, the back-
ground gases freely invade the jet flowfield. Attenuation of the molec-
ular beam results. Knuth (Ref. B-3) states that the scattering of beam
molecules is a problem if the background molecules either attenuate the
beam density appreciably or alter greatly the relative densities of the
several beam species. He presents criteria for avoiding both problems.
Molecule Fragmentation in the Detector
The problem of molecular-beam sampling of unstable and reactive species
is complicated by possible confusions of these species with fragment ions
of other molecules. Furthermore, in sampling of gases at different tem-
peratures, the variations of their fragment patterns with temperature
must be taken into account. The fragment pattern is dependent on the vi-
brational energy of the molecule. For example, the measurements of
Greene and Milne (Ref. B-l) on flames show that the fragmentation of CCL
to CO varies at least from 12 percent at room temperature to 17 percent
at 2000 K. On the other hand, Homann et al. (Ref. B-18) could not find
any change in the mass spectrum of acetylene sampled from burned gases at
1700 K compared with the spectrum obtained at room temperature. The prob-
lems of vibrational relaxation in free jets, and the dependence on the vi-
brational state of fragmentation patterns, must be studied for a variety
of molecules of interest in high-temperature sampling situations.
Applications
The development of the molecular-beam mass-spectrometer sampling technique
has been motivated partly by the desire to observe reactive and unstable
species in flames. In 1953, Foner and Hudson (Ref. B-19) employed a
molecular-beam mass-spectrometer system to sample and detect atoms and
143
-------
radicals of chemical reactions and flames. In their sampling system,
which was especially designed for the study of reactive free radicals,
the gas being sampled entered the system through a small circular aper-
ture in( a glass or quartz cone and was collimated by two additional
slits. The three sections of the molecular-beam sampling system were
separately evacuated by high speed diffusion pumps, the pressure typi-
cally being 10~3 torr in the source chamber, 10~5 torr in the collima-
tion chamber, and 10"^ torr in the ion source chamber. The molecular
beam traversed 10 cm from the entrance aperture to the center of the ion
source. This distance corresponded to a transit time of about 230 usec
for an oxygen molecule at room temperature. The molecular beam was me-
chanically interrupted at 170 cps by a vibrating reed beam chopper in
the source chamber to discriminate against background signals. The sev-
eral species were identified and monitored by a 90-degree sector magnetic
mass analyzer. A movable burner assembly was used to position the flame
at various distances from the sampling pinhole. Hydrogen-oxygen and
methane-oxygen flames were studied. The hydrogen-oxygen flame clearly
showed the presence of the stable species H , 0 , and HO and the unsta-
ble species H, 0, and OH in sufficient abundances to permit mapping of
intensities as functions of burner distance from the pinhole. The atom
and radical intensity measurements were made at sufficiently low ionizing-
electron energies to eliminate contributions from dissociative ionizations
of the stable components. In the absence of absolute calibrations for the
radicals, reasonable estimates were made of their ionization cross sec-
tions. Taking into account that low electron energies were used in the
radical measurements, they suggest multiplying the adjusted ion intensi-
ties in the lower half of their figure (Ref. B-10) by a factor of about
10 to put the concentrations of the stable components and radicals on a
common basis. The maximum radical concentrations were thus estimated to
be of the order of 1 percent of the composition profiles. Although the
measurements were reproducible, their interpretations were complicated
by diffusion effects, turbulent mixing, and changes in the flame config-
uration as the burner was moved with respect to the diaphragm.
144
-------
The mass spectra obtained for the methane-oxygen flame were of unantici-
pated complexity. The stable products that were readily identified were
H2, H20, C2H , CO, CCL, and C.H2< Other stable components were obviously
present, but only tentative identifications could be made for some of
these, such as C^\ at mass 28, C^ or HCHO at mass 30, and CHjOH at
mass 31. Evidence suggested also the presence of C,H., C,H,, and CJ1 .
The only radical that was positively identified was CH .
In these experiments, the detection and positive identification of HO^
radicals as an intermediate was the principal objective. Careful exper-
iments were performed varying the burner position and the hydrogen-
oxygen ratio without obtaining conclusive evidence for existence of the
HO- radical. Due to (1) the complexity of the methane-oxygen reaction,
(2) the inability to identify all the stable components present, and (3)
the possibility that some of the molecules could have been in excited
states, and therefore had different fragmentation patterns than unexcited
molecules, it was felt that identifications of radicals other than methyl
would be somewhat speculative. Several years later, Foner and Hudson
(Ref. B-20) used the same system with a number of modifications to study
the production, identification, and determination of the thermochemical
energy of H0_. The modifications were made to improve the sensitivity
and precision of measurement. Phase-sensitive detection was replaced by
ion counting. The output of the electron multiplier detector consisted
of about 10 Coulomb per ion. The pulses were amplified and sent through
a gated amplifier and an electronic switch which was synchronized with
the beam chopper so that one of the ion counters recorded ions only when
the beam chopper was open, the other only when the beam chopper was
closed. The difference between the two ion counts represented the ion
intensity contributed by the molecular beam, while the square root of the
sum of the two ion counts was approximately equal to the standard devia-
tion of the measurement and served as a useful indicator of the quality
of the data.
145
-------
In this experiment, the reactions found to produce HO radicals, and
examined in some detail, were (1) reaction of H atoms with CL, (2) re-
action of H atoms with H_07, (3) reaction of 0 atoms with PLO , (4) re-
action of OH radicals with H20 , (5) photolysis of H_02, and (6) low-
power electrical discharge in H^CL. Of the various reactions investi-
gated, the low-power electrodeless electrical discharge in a rapidly
flowing system of HO was found to be a fairly intense and convenient
source of H02 radicals. A maximum concentration of about 0.4 percent
HO was obtained at 15 percent H~0_ decomposition in a typical experi-
ment. The production of H0_ in the various reactions involving H^O^
was predominantly by OH radicals, generated in assorted primary steps,
reacting in the fast reaction
OH + H202 -> H20 + H02
From 1964 to 1966, other investigators attempted to study flame reac-
tions using similar sampling techniques. For example, Vriens et al
(Ref. B-21) employed a molecular beam in sampling 1-torr to 1-atmosphere
gases (Ar or H2) for three sampling-orifice diameters. Their apparatus
was somewhat similar to the previously mentioned apparatus. The sampling
orifice was a pinhole in a gold foil of 0.02 mm thickness. The diameters
chosen for the experiments were about 0.20, 0.035, and 0.07 mm. The dis-
tance between the sampling orifice and the second orifice was 10 mm, and
the distance between the second and the third orifices was 240 mm. The
diameters of the second and third orifices were 0.18 and 3.5 mm, respec-
tively. At normal operating conditions, the background pressures in the
first and second chambers were 5 x 10"^ to 3 x 10"^ torr (dependent on
sampling orifice diameter) and 10"^ torr, respectively. The molecular
beam was modulated using a mechanical 50 cps chopper. This modulation
facilitated the measurement of the scattering of the molecular beam.
The 1-atmosphere flame was replaced by a gas container filled with argon
or hydrogen of variable pressure, and for three diameters of the sampling
146
-------
orifice the dependence of the mass spectrometer ion current on gas pres-
sure was measured. The dependence of the background pressure in the
first chamber on gas pressure and sampling-orifice diameter was also
determined.
From the measurements, it was concluded that for sampling-orifice diam-
eters of 0.035 and 0.07 mm, and for a gas pressure of 1-atmosphere, the
density in the beam between the sampling orifice and the second orifice
was large. The numerous collisions between beam molecules in this re-
gion result in conversion of much of the translational energy, and in-
ternal energy in the case of molecules, into directed motion of the beam.
Therefore, the static temperature decreases and a shift of composition
of the flame gas is possible during expansion. Predicting the amount of
this composition change is difficult. It is simpler perhaps to use a
smaller sampling orifice diameter, e.g., 0.02 mm, in order to decrease
the number of collisions appreciably. Fortunately, since the flame has
a higher temperature, the gas density in the flame is less than the gas
density of room-temperature Ar or H~. Therefore, the number of colli-
sions between beam particles is less for the flame gas.
Quantitative studies of mass spectrometric sampling of sources at and
above 1 atmosphere and temperatures up to 3000 K have been carried out
by Greene and Milne (Ref. B-22 § B-l). The sampling system, similar to
previously mentioned systems, included three differentially pumped stages
between the sampling orifice and the ion source of a Bendix time-of-
flight mass spectrometer. The beam was modulated at 10 cps, which facil-
itated the detection of beam signals, even when the ratio of background
density to beam density was greater than 100. In subsequent studies, the
beam chopper and the phase-sensitive detection system were replaced by a
magnetically driven vibrating reed, and lock-in amplifiers. Modulation
at 10 and 50 cps gave identical results, indicating that effects of modu-
lation of the background gas were negligible.
147
-------
Several flames (H2-02-N , CH.-02-Ar, C0-02) were analyzed, both for sta-
ble reaction products and reactive species. In the CH.-O -Ar flame, the
sampling of the stable products H , H?0, CO, 0 , and CO was very simple.
Equilibrium concentrations of these species were maintained by fast bi-
molecular free-radical reactions.
The temperature dependence of the fragmentation of the molecules in the
ionizer complicated the stable-product measurements. In order to deter-
mine the magnitude of this temperature effect, C0_ was added to a lean
H202 flame which burned at 1950 K. Fragmentation of C02 to C0+ increased
from 12 percent at room temperature to 17 percent at 2000 K. The effect
of temperature on fragmentation of HC1 was studied by sampling HC1 from a
series of rich H^-O^-M- flames.
Although spatial resolution for sampling from the main reaction zone was
only fair (the sampling orifice diameter was 0.24 mm), the authors did
measure quantitatively Cl, H, 0, and OH concentration profiles in the re-
combination zones of the flames. Experimental values agreed within the
limits of experimental error with those calculated for chemical equilib-
ria between stable species and radicals at the adiabatic flame tempera-
ture. Other species such as S, SH, SO, HBO., and F were readily detected
in the reaction zones of 1-atmosphere flames; excess radical concentra-
tions were observed. It was predicted that the extremely rapid cooling
achieved in free-jet direct sampling would allow almost any non-condensi-
ble chemical species to be quantitatively measured from systems at pres-
sures of several atmospheres and temperatures as high as 3000 to 5000 K.
Molecular beam sampling also has been used in flame studies by Wagner
et al. (Ref. B-18, B-23, § B-25). The apparatus used in these experiments
was similar to previously mentioned apparatuses. A quartz cone, used
as the sampling probe, was attached to an intermediate vacuum chamber,
which was maintained at pressures from 10"^ to 10~4 torr. The apex an-
gle of the skimmer was so chosen that the particles inside the cone did
148
-------
not undergo collisions with the inner wall of the cone. The objective
of the investigation was to study the formation of solid carbon and to
observe the intermediates formed in this process. Various rich hydro-
carbon oxygen flames were examined for such hydrocarbon radicals as CH ,
C H, C H , acetylenic polymers C.H to C-2H , and for very reactive
cyclic hydrocarbon species in the 60 to 100 amu range. The profiles of
some highly reactive hydrocarbon intermediates were quantitatively re-
lated to the increase in solid carbon concentration. In hydrogen-oxygen
flames concentration profiles of 0, H, OH, and H70~ were measured. Evi-
dence for the presence of the HCL radical in rich ^-0- flames was ob-
tained. In the same manner, hydrazine decomposition flames were inves-
tigated for NH2 and N2H2 intermediates (Ref. B-25).
The same apparatus was used (Ref. B-26) to sample reacting gases in an
isothermal flow reactor. Rapid flow around the probe tip favored the re-
moval of the gas that had hit the wall around the orifice at the probe
tip, and diminished the thermal influence of the probe on the reacting
gases. (This influence cannot be totally excluded when sampling the re-
action zone of flames.) In this way, the mechanism of the oxidation of
CS- by 0 , strongly diluted with argon, was investigated at temperatures
up to 1400 K. CS- was injected into the hot argon stream already con-
taining oxygen. The concentration profiles of all stable species and
radicals were measured by changing the distance between the mixing point
and the sampling probe in the stationary reaction system. SCL and CO
were the main reaction products, with little CO- being formed after the
CS2 had been consumed completely. In addition to the comparatively sta-
ble intermediate COS, radicals such as SO, S, 0, CS, and S20 were mea-
sured. In subsequent experiments, the rates of individual steps in this
complex reaction were studies in a low pressure fast flowing system (Ref.
B-27, B-28, B-29. A molecular-beam system combined with a Bendix time-of-
flight mass spectrometer was used. A suitable electron energy for ion-
izing was found to be 25 eV. At this energy, dissociation of the ions
into sub-particles were relatively small.
149
-------
The elementary step 0 + COS -» CO + SO, by which the COS is oxidized, was
distinguished completely from consecutive reactions of SO to SO,,. The
reaction 0 + CS~ > CS + SO also was followed separately. The reaction
rates for 0 + H , 0 + C_H , 0 + NH and other reactions were measured.
By following quantitatively the concentrations of each participant in
the reactions, primary reaction steps were distinguished from combined
reaction steps.
The reactions of oxygen atoms with nitric oxide and nitrogen dioxide
have been studied by Klein and Herron (Ref. B-30). The calibration for
oxygen atoms at an electron energy of 20 eV was obtained by determining
the atom concentration independently by titration with nitrogen dioxide.
Young et al. (Ref. B-31) sampled reactive gases from engine cylinders
using molecular beam techniques and determined densities of N~, 02, C,HR,
H20, C02, HO, and CO.
Molecular beams have been employed also by several investigators to study
ions in high pressure flames. DeJaegere et al. (Ref. B-32) have identi-
fied the most abundant ions in methane-oxygen, acetylene-oxygen, and
CH -NH_-02-N2 flames. Ions produced in halogen-containing flames and
the most abundant ions in nitrogen-oxide-producing flames were also
studied. Hayhurst and Telfdrd (Ref. B-33) studied the charge exchange re-
action of the H_0+ ion with metal in atmospheric flames of H^-O^-N at
temperatures from 1815 to 2445 K. The metals CS, Rb, K, Na, Li, Tl, Cr,
Pb, Mn, Cu, Fe, and Cd were used.
Shock tubes combined with molecular-beam sampling systems also can be
used to study reactive gases. Skinner (Ref. B-34) pioneered in sampling
from shock tubes; his objective was a neutral beam with energy above 1
eV. Penn and Liquornik (Ref. B-35) used an 0-He mixture in a shock tube
to generate an 0 beam with energy of 3 eV. Jones and Byrne (Ref. B-36)
also developed such a system to study the production of a metastable ar-
gon beam and its interaction with a copper surface.
150
-------
More recently, Solomon et al (Ref. B-37) used a nozzle-beam mass spectrom-
eter system to study 1 atmosphere flames. An ionization energy of 20 eV
was used in the mass spectrometer in order to reduce fragmentation of
the molecules. Output signals were recorded on a high-speed oscillograph.
The scan rate for the mass range m/e from 1 to 60 was 250 milliseconds.
Use of 2000 Hz galvanometers in the recording oscillograph permitted re-
cording at this scan rate without peak clipping. The premixed-gas flame
was slowly traversed toward the sampling orifice, starting approximately
10 mm above the reducing portion. Probing through the reaction zone re-
sulted only in qualitative profiles since the reaction zone was very thin.
Measurements from 1 to 5 mm above the inner reducing cone tip were most
suitable for calculations of equilibrium constants. Beyond 5 mm, heat
loss shifted the reaction equilibrium. No corrections for heat loss were
made. Flames of CH.-02-Ar with four different mole fractions and four
different temperatures from 2444 to 2877 K were samples. The concentra-
tions of H , 0 , OH, H20, CO, Ar, and C02 were determined. These flames
showed similar behavior to previously studied flames, such as ^-0-, and
CH.-air flames. The investigators concluded that their molecular beam
system was capable of quantitatively sampling high-pressure, high-
temperature flames, and that the data agreed well with the equilibrium
properties of methane-oxygen flames. No particular problems were encoun-
tered in sampling atoms and radicals.
III. TEMPERATURE MEASUREMENTS
High temperature sources have been studied using the time-of-flight tech-
nique. Applications of this basic technique can be divided into two gen-
eral types: those in which the beam intensity as a function of time is
measured directly, and those in which the flight time is determined as a
phase shift by means of a phase-sensitive detector. The direct measure-
ment technique is older and provides more detailed information. When a
molecular beam is pulsed on for a short time interval, at time between t
and t + dt the number of molecules n(t) arriving at the detector will have
151
-------
velocities between L/t and L/t + dt, where L is the distance from the
beam pulser to the detector. If the total number of molecules is N,
then n(t)/N will be the fraction of molecules with velocity between v,
and v + dv add the velocity distribution can be determined by measuring
n(t) as a function of t.
In order to use the time-of-flight technique to determine the stagnation
temperature of the source, it is necessary to determine the velocities
of the molecules in a beam extracted from the source. This method was
used, e.g., by Anderson and Fenn (Ref. B-38) in their studies of super-
sonic jets of hydrogen or helium containing small concentrations of
heavier molecules. The beam was chopped into short segments by a rotat-
ing shutter and the flight times to a downstream detector were measured.
Velocities were obtained for 1 mole percent of solute in hydrogen or
helium. If a pure gas expands isentropically in a free jet, then the
average kinetic energy of the molecules is
2 /*T
1/2 mV = / ° C dT
J P
T
where V = hydrodynamic velocity
m = molecular mass
C = molecular specific heat at constant pressure
T = stagnation or source pressure
T = static temperature in the jet
At Mach numbers above 5, the static temperature becomes negligible. Hence,
in expansions to high Mach numbers, the kinetic energy of the heavy species
is given by:
o m, /-T
1/2V =iT j cpdT
m o m
where subscript m indicates the mean value for the mixture and h indicates
the heavy species.
* 152
-------
Alcalay and Knuth (Ref . B-39) developed a moment method to extract from
the measured TOP signal the beam density, temperature and energy. Alge-
braic relations between the moments of the measured time-of-flight sig-
nal, the speed distribution function, the modulator gate function, and
the dynamic function of the detector and its electronics were derived.
The simplest gate functions are the triangular, trapezoidal, step, rec-
tangular, and impulse gate functions. This technique was applied to
time-of-flight measurements of an arc-heated supersonic molecular beam;
values of the temperature and mean beam energy were calculated using
-7 2
T = mYV2k and E = mY n {f (s)}/2, respectively. Here Y is the most
probable random speed of the beam, m is the molecular mass, k is Boltz-
man's constant, and n_{f (s) } is the second moment of the speed distribu-
tion function. Values of the stagnation temperature were calculated re-
lating the energy of the beam to the enthalpy at the source conditions.
Young (Ref . B-40) has overcome the problems in determining the dynamic
function of the detector and its electronics, the gate function, and the
zero-time reference by using a dual-disk TOP chopper. Each disk con-
sists of four small slits and four large slits. The depth of the small
slit (angular width = a) was greater than the large slit (angular width =
0) in order to facilitate use of a photocell in conjunction with the
small slits to trigger the oscilloscope. The two disks were mounted on
a shaft with angular orientation such that the centers of a on one disk
were offset relative to the centers of 0 on the other disk, with a phase
angle of
0 - 2Tr _L
" 2Na " Na
where N was the number of a slits in one disk.
General moment relations for the dual-disk TOP signals were developed.
These relations included expressions for extracting values of the static
temperature and energy of the beam.
153
-------
IV. CONCLUSIONS
The feasibility of using a molecular-beam mass-spectrometer system to
measure the concentrations of most stable species and free radicals in
high-temperature 1-atmosphere flames has been demonstrated. Possible
adverse effects which must be either avoided or handled quantitatively
include certain nonequilibrium processes, shock waves, skimmer interfer-
ence, mass separations, nucleation, background scattering, and molecule
fragmentations. Of these several effects, nucleation is perhaps the
least well understood. Fortunately, available experimental results in-
dicate that all of these effects, excepting mass separations and mole-
cule fragmentations, can be avoided, and that these two exceptions can
be handled quantitatively. Additional data are required in order to ex-
tend criteria and methods obtained for given systems to other systems.
The measurement of flame temperatures by the time-of-flight technique
has unique advantages over other methods. The use of a dual-disk chop-
per simplifies this measurement significantly by avoiding the uncertain-
ties in the zero-time reference point, the gate function, and the dynamic
function of the detector and its electronics.
154
-------
REFERENCES
B-l. Milne, T. A. and F. T. Greene, "Mass Spectrometric Studies of Re-
actions in Flames," J. Chem. Phys., 44:2444, 1966.
B-2. Sherman, F.S., "Hydrodynamical Theory of Diffusive Separation of
Mixtures in a Free Jet," Physics of Fluids, 8:773, 1965.
B-3. Knuth, E. L., "Direct-Sampling Studies of Combustion Processes,"
to be published in Engine Emissions (G. S. Springer and
D. Patterson, eds.), Plenum Publishing Co.
B-4. Anderson, J. B., R. P. Andres, J. B. Fenn, and G. Maise,
"Studies of Low Density Supersonic Jets," Rarefied Gas Dynamics,
11:106, 1966.
B-5. Reis, V. H., and J. B. Fenn, "Separation of Gas Mixtures in Super-
sonic Jets," J. Chem. Phys., 39:3240, 1963.
B-6. Young, W. S., W. E. Rodgers, C. A. Cullian, and E. L. Knuth,
"Molecular Beam Sampling of Gas Mixtures in Cycling-Pressure
Sources," Proceedings of the Seventh International Symposium on
Rarefied Gas Dynamics held at Pisa, Italy, June 29-July 3, 1970.
B-7. Becker, E. W., K. Bier, and H. Burghoff, "Die Trennduse," Z. fur
Naturforschung, 10a:565, 1955.
B-8. Waterman, P. C. and S. A. Stern, "Separation of Gas Mixtures in a
Supersonic Jet," J. Chem. Phys., 31:405, 1959.
B-9. Stern, S. A., P. C. Waterman, and T. F. Sinclair, "Separation of
Gas Mixtures in a Supersonic Jet. II. Behavior of Helium-Argon
Mixtures and Evidence of Shock Separation," J. Chem. Phys.,
33:805, 1960.
B-10. Abuaf, N., J. B. Anderson, R. P. Andres, J. B. Fenn, and
D. R. Miller, "Studies of Low Density Supersonic Jets," Rarefied
Gas Dynamics, 11:1317, Academic Press, New York, 1967.
155
-------
B-ll. Greene, F. T., J. Brewer, and T. A. Milne, "Mass Spectrometric
Studies of Reactions in Flames," J. Chem. Phys., 40:1488, 1964.
B-12. Young, W. S., Y. G. Wang, P. K. Sharma, W. E. Rodgers, E. L.
Knuth, "Molecular-Beam Sampling of Continuum Gas Mixtures," pre-
sented at the Eighth International Symposium on Rarefied Gas Dy-
namics held at Palo Alto, California, 10-14 July 1972.
B-13. Greene, F. T. and T. A. Milne, "Molecular Beam Sampling of High
Temperature Systems," AIAA Paper No. 67-37, 1967.
B-14. Greene, F. T. and T. A. Milne, "Mass Spectrometric Detection of
Polymers in Supersonic Molecular Beams," J. Chem. Phys., 39:3150,
1963.
B-15. Milne, T. A. and F. T. Greene, "Mass Spectrometric Observations
of Argon Clusters in Nozzle Beams. I. General Behavior and
Equilibrium Dimer Concentrations," J. Chem. Phys., 47:4094, 1967.
B-16. Milne, T. A. and F. T. Greene, "Mass Spectrometric Study of Metal-
Containing Flames," Final Tech. Summ. Rept., 1961-67, Contract
NONR-3599100. Kansas City: Midwest Research Institute, 30
January 1968.
B-17. Milne, T. A. and F. T. Greene, "Mass Spectrometric Detection of
Dimers of Nitric Oxide and Other Polyatomic Molecules," J. Chem.
Phys., 47:3668, 1967.
B-18. Homann, K. H. and H. G. Wagner, Ber. Bunsengesellschaft Physik.
Chem., 69:20, 1965.
B-19. Foner, S. N. and R. L. Hudson, "The Detection of Atoms and Free
Radicals in Flames by Mass Spectrometric Techniques," J. Chem.
Phys., 21:1374, 1953.
B-20. Foner, S. N. and R. L. Hudson, "Mass Spectrometry of the H02 Free
Radical," J. Chero. Phys., 36:2681, 1962.
B-21. Vriens, L., A. L. Boers, and J. A. Smit, "Measurement on a Molec-
cular Beam Apparatus for Mass Spectrometric Study of Reactions in
Flames," App. Sci. Research (Sec. B), 12:65, 1965-66.
156
-------
B-22. Greene, F. T., J. Brewer, and T. A. Milne, "Mass Spectrometric
Sampling of 1 Atm. Flame," Tenth Int. Symp. on Combustion, p. 153.
Pittsburgh: The Combustion Inst., 1965.
B-23. Bonne, U., K. H. Homann, and H. G. Wagner, "Carbon Formation in
Premixed Flames," Tenth Int. Symp. on Combustion, p. 503. Pitts-
burgh: The Combustion Inst., 1965.
B-24. Homann, K. H. and H. G. Wagner, "Some New Aspects of the Mechanism
of Carbon Formation in Premixed Flames," Eleventh Int. Symp. on
Combustion, p. 371. Pittsburgh: The Combustion Inst., 1966.
B-25. Homann, K. H., D. I. Maclean, and H. G. Wagner, Naturwissenschaften
52:12, 1965.
B-26. Krome, G., "Massenspektrometrische Untersuchungen des
Reaktionsablaufes der Schwefelkohlenstoff-Oxidation bei Hohen
Temperaturen," Diplomarbeit. Gottingen: Universitat Gottingen,
April 1965.
B-27. Wagner, H. G. and J. Wolfrum, "Bestimmung der Geschwindigkeit der
Reaktion 0 + COS -> CO + SO," Ber. Bunsengesellschaft Phys. Chem.
71: 603: 1967.
B-28. Homann, K. H., K. H. Hoyermann, and J. Wolfrum, "Versuche sur
Vollstandigen Bestimmung der Geschwindigkeiten Bimoleklarer Reak-
tionen," Kurznachrichten Nr. 2, Akademie der Wissenschaften, II.
Math. Phys. Klasse. Gottingen, 1967.
B-29. Homann, K. H., G. Krome, and H. G. Wagner, "Schwefelkohlenstoff-
Oxydation Geschwindigkeit von Elementarreaktionen," Ber.
Bunsengesellschaft Physik. Chem. 72:998, 1968.
B-30. Klein, F. S. and J. T. Herron, "Mass Spectrometric Study of the
Reactions of 0 Atoms with NO and N02," J. Chem. Phys. 41:1285:
1964.
157
-------
B-31. Young, W. S., Y. G. Wang, W. E. Rodgers, and E. L. Knuth, "Molecu-
lar Beam Sampling of Gases in Engine Cylinders," Technology Utili-
zation Ideas for the 70's and Beyond, 281-289. Tarzana: American
Astronautical Society, 1971.
B-32. De Jaegere, S., J. Deckers, and A. Tiggelen, "Identity of the Most
Abundant Ions in Some Flanfes," Eighth Int. Symp. on Combustion, p.
155. Baltimore: Williams and Wilkins, 1962.
B-33. Hayhurst, A. N. and N. R. Telford, "Charge Exchange Reactions of
H30+ with Metals in Flames," Trans. Faraday Soc. 66: 2784, 1970.
B-34. Skinner, G. T., "Scattering at a Solid Surface and Observations
of Radiation Accompanying the Beam," Rarefied Gas Dynamics, II:
1325, Academic Press, New York, 1969.
B-35. Peng, T. C. and D. L. Liquornik, "Shock Tube Molecular Beam for 0
to 3 eV," Rev. Sci. Instr. 38:989, 1967.
B-36. Jones, T. V. and M. A. Byrne, "The Production of a Metastable Ar-
gon Beam with Kinetic Energy up to 3 eV and its Interaction with a
Copper Surface," Rarefied Gas Dynamics 11:1311, Academic Press,
New York, 1969.
B-37. Solomon, W. C. and B. B. Goshgarian, "Nozzle Beam-Mass Spectrometer
Systems for Studying One-Atmosphere Flames," Tech. Rept.
AFRPL-TR-27-30. Edwards: Air Force Rocket Propulsion Laboratory,
April 1972.
B-38. Abuaf, N., J. B. Anderson, R. P. Andres, J. B. Fenn, and
D.G.H. Marsden, "Molecular Beams with Energies Above One Electron
Volt," Science 155:997, 1967.
B-39. Alcalay, J. A. and E. L. Knuth, "Molecular-Beam Time-of-Flight
Spectroscopy," Rev. Sci. Instr. 40:438, 1969.
B-40. Young, W. S., "An Arc-Heated Ar-He Binary Supersonic Molecular
Beam with Energies up to 21 eV," Report No. 69-39, Los Angeles:
Department of Engineering, University of California, 1969.
158
-------
APPENDIX C
TABULATION OF DATA OBTAINED WITH UCLA MOLECULAR-BEAM
MASS-SPECTROMETER SAMPLING SYSTEM
159
-------
Di
PJ
Q
U
0)
II
a
H
CM
-« tti
a:
X
CM
O es
U 2
t/5 W
II
CM
O
U
O
O
M
in
CM
0
in
CM
m
CM
CM
o
o
CM
in
i-H
o
in
14
CM
1 1
o
o
< i
JO
o
stance,
Xinches
\
%f
m o i- CM i^ m o»
vO ^ OO OO vO \O *^t"
o o o o o o o
o o o o o o o
in *- * in in ^ \o 01
\o r^ oo oo vo \o ^f
o o o o o o o
o o o o o o o
r^ o to ^ in oo * *
vo t^ oo oo \o vo in
o o o o o o o
o o o o o o o
in -<
-------
a:
UJ
a
*j
o
a:
i
o
H
i
UJ
u
a,
rH CO
as o
O rH
rH PU
LU
rJ
o
8 Q
a, to
O w
u
§ 5
u,
CM
I
U
(D
oo
CM
0
X
UJ
CO
o
o
to
CM
0
LO
CM
LO
CM
O
O
CM
LO
0
LO
LO
CM
0
o
rH
LO
*
w4
0) 0
0 rC
C 0
d C t
4-> -H /
4/i
k
CTl OO
CM rH
rH rH
VO rH
CM rH
rH rH
*3- LO
CM rH
i 1 rH
r~ LO
CM CM
CTl CTl
rH rH
"3" CM
CM CM
vO ^J"
CM CM
* CM
tO CM
\O i 1
tO CM
rH i 1
"^ rH
tO CM
*
LO O
f^ OO
0 0
II II
-e- -e-
\D
CM
rH
rH
CM
i 1
CM
CM
rH
O
CM
CM
CM
rH
CM
CM
CM
CM
to
to
to
rH
to
to
*
0
CTl
O
II
-e-
00
CM
rH
LO
CM
LO
CM
to
CM
O
CM
rH
to
CM
CM
rH
00
rH
LO
rH
O
0
1 1
II
-e-
to
. i
CTl
to
rH
00
to
to
CTl
to
rH
O
rH
00
to
to
rH
CM
to
O
rH
rH
II
-e-
CTi
CM
OO
CM
rH
CM
rH
O
to
00
CM
rH
rH
CM
r~-
CM
CM
LO
CM
rH
CM
LO
CM
f-H
II
-e-
o
rH
rH
00
O
i I
00
o
t 1
00
o
o
rH
VO
O
LO
o
00
o
r--
0
rH
to
rH
0
to
rH
II
-e-
161
-------
CJ
Z
D-
CO
O
r-H
H
CO
O
w
z
w
_]
ex
ia
%
I-H CO
U
o
tt
ai
u
CO
co
CN Q
O
CO
to
I
U
CD
iI
a
H
o
X
o
o
to
in
CN
o
m
CN
CN
CM
o
0
fN
in
i-H
O
in
i-H
in
CN
r-H
O
o
1 1
in
o
gl
s/
7*
&
f Or
" >
O i-H
rH O>
to CN
Tt O
tO CN
r-- CM
to \O
tO CN
Tt OO
tO (ft
tO rH
^D ^D
CN 00
tO r-H
rH i-H
Tf i-H
to CN
rj- to
0 CN
It
-------
UJ
Q
s
p-
1
O
H
i
UJ
u
1 1
tu
HH
cf.
O
u
z
hH
a.
^
VARIOUS S/
to
O
H
U
(U
i i
cy
to
B
0
-H
X
CM
I-H
1 W
01
X
o <
Z 2
to co
II
"o
I
u
a
ofi
g a
FT^ ^?
o
co
Z OJ
O U
H
to
0
o
to
LO
CM
0
LO
CM
LO
CM
CM
O
0
CM
LO
i {
O
LO
I 1
LO
CM
t 1
O
0
i-H
LO
0
istance,
Shiches
^ ***
\*
?&*
CM
cn
0
CM
cn
o
cn
00
o
i-H
cn
0
00
0
OO
00
0
LO
00
o
en
0
LO
00
o
OO
o
LO
*
o
II
-Q-
CM
LO
-1
vO
LO
I-H
vD
LO
I-H
cn
LO
i-H
LO
^
LO
i-H
LO
LO
i-H
O
LO
t 1
CM
LO
i-H
O
LO
i-H
0
00
0
II
e-
£
CM
LO
CM
5
CM
O
LO
CM
LO
CM
to
CM
£
CM
^
CM
lo
CM
CM
CM
O
cn
d
u
-e-
OO
CM
to
LO
CM
to
o
cn
CM
cn
00
CM
0
to
to
i-H
to
00
cn
CM
i-H
cn
CM
LO
00
CM
vD
O
to
O
O
i-H
II
-e-
cn
to
to
cn
to
to
CM
LO
to
r-
LO
to
LO
to
«*
to
0
to
00
\o
to
cn
CM
LO
CM
to
O
it
II
-e-
CM
LO
"*"
O
*
o
LO
"*
cn
LO
^
\D
\O
^
to
^
CM
*
CM
^
LO
^
>-H
^
LO
CM
i 1
II
-e-
cn
^
cn
^
o
LO
o
LO
00
00
^
0
LO
\o
cn
^
CM
**
cn
^
vO
cn
"*
o
to
i-H
II
-e-
163
-------
in
a
o
ii
H
< m
> u
H uj
§ 3
t-H cx
E- W
UH
U4
H-I
O
s.
CM
in
CTi
CM
O
s
Q
UJ
U
Z Q
S OS
5 g
CM J
* §
U,
o
LO
u
0)
i-H
s
o
H
0-1
U
OS
o
to
o
X
w S
*
CO CO"
CM
0
o
to
LO
CM
0
LO
CM
LO
CM
CM
O
O
CM
LO
i-H
O
LO
r-H
LO
CM
i-H
O
O
i-H
LO
0
LO
i-H
vO
vO
i-H
vO
\O
O
vO
to
vO
to
CM
VO
\O
O
vO
LO
LO
Ol
00
LO
oo
00
LO
0
o>
o
II
-e-
£
LO
-H
LO
r-H
LO
LO
LO
o>
LO
vO
LO
00
t--
LO
oo
LO
CM
vO
LO
vO
o
vO
o
0
i-H
II
-e-
i-H
VO
1 1
VO
oo
o
vO
CM
i-H
vO
i-H
vO
0
to
VO
to
CM
\0
to
CM
VO
LO
i-H
vO
oo
LO
o
r-4
i-H
II
-e-
vO
VO
to
i-H
vO
LO
1 1
vO
to
vO
o>
LO
to
t-H
vO
LO
o
vO
0
i-H
vO
1 t
vO
LO
CM
i-H
It
-e-
s
LO
CM
00
LO
oo
LO
00
LO
I--
LO
oo
LO
to
LO
LO
LO
LO
to
00
LO
oo
LO
o
to
1 1
II
-e-
164
-------
z
o
H
U
§
3
tu to
5 2
z
(-1 x
0 f i
O CM
o z
CM h- 1 '
E-
" g ~
Q__J in
^^ U-J
H U OS
< Z
LU X
O >-J
u <: o (N
> U Z
tt, l-H tO tO
O 3
O* H
MO, O
HO U
CJ
<^
OH
PL.
C/3
tu tu
9 5
2 z
HH
o
\O O
1
U CM
^j-
CM
vO
. 1
0
165
-------
UJ
u
o
I
co
co «
D
c
r-
o
<
w
V
5
H
^
H
>
flj
U
w
E- >
<£, n
r
2
-\
b
r-
r
H
r^
U
u-
u
r
J
J
9
^
CN
2
/ N
K-J
^C
5
CO
o
r
S
\^
g
u
J
3
^l
5
CO
CM
2
^^^X
U4 CM
O
CO *-'
0 X
^""^ _
%
> 2.
Q
§ o
"* CM CM
rn X 2
£q co CO
U
H
co
Q
a:
UJ
a
p-J
0
i
tu
0 S
co i-J
u
J
IX,
3
J
O
i
u
3
t-
0
o
to
in
CM
o
in
CM
in
CM
CM
o
o
CM
in
rH
O
in
rH
CM
rH
0
o
rH
£
rH
» |
rt o1
f f!
U) -H/ £*". O
A
O> CM
1^ OO
* Tf
O CM
^o oo
* "*
tO CM
vO OO
* "*
r- oo
^ *
oo in
Tf OO
* -t
in t--
to o>
* *
CM . -^f
t"**- 00
** *
vo m
* *
vO vo
1^- OO
00
0 0
u u
-e- -e-
£
*
00
"*
vO
OO
Tt
vO
^
OO
*
r--
*
o
in
CM
m
in
in
to
CM
m
o
o>
o
u
e-
to
o
in
o
rH
m
CM
in
o
in
in
O)
*
rH
rH
in
rH
in
R
=*
rH
in
0
to
in
o
o
rH
II
-e-
s
*"
s
^
s
rt-
-
^
in
*
in
C71
^
vO
*
00
*
oo
in
"*
o
rH
rH
II
e-
m
^
m
"*
CM
^
to
"*
m
^
rH
rH
m
vo
^
o
m
*
00
m
'
to
in
"*
m
CM
i-H
II
-e-
(N
m
"*
^
"*
to
"*
en
"*
vO
to
**
rH
*
CM
CM
*
to
*
rH
^
rH
"*
O
to
rH
II
-e-
166
-------
u
u
II
u.
3
o
I
u
CO
co
o
8
rH
H
U4
U
UJ
,-J
<~v C/
2 W
O
rH CO
E- 3
rj o
p-
UJ
CM CO
2 W
H
co
co
to
u,
o
co
PJ
oo
i
u
8
E-
«
W
CM Q
'" ?
o
°
X
CN
z
g
CO
to
o
O
to
LO
CM
0
LO
CM
LO
CM
CM
0
O
CM
LO
rH
O
LO
i 1
LO
CM
rH
O
o
*
rH
0>
o
§
w
Qy
LO
O
it
/?
f %t*
yt-P*?1
-------
PLING
to
e
h-l CO
tL, O
II II
«S H
9 2
CO
o
I-H
ai
<
LENCE
H I-H
O co
H O
tL, >
U Q
CO
CM W
to
co w
a
CM J
u,
o
co
PJ
en
u
0
o
0
to
LO
r--
.
CM
O
LO
C**l
LO
CM
CM
0
o
CM
LO
.
O
LO
LO
CM
i (
O
o
LO
*
o
ryj
y/'
v/
?"vO
^^
vo
o
vD
1 1
vO
en
,
LO
.-H
0
to
*
LO
CM
LO
,
LO
CM
f-
LO
i 1
LO
i 1
VD
LO
O
vo
o
vo
LO
i-t
O
^»
LO
t--
LO
rH
LO
O
II
-e-
o
LO
LO
r~ t
0
CM
*
LO
I 1
o
o
LO
^
^i"
,
LO
LO
0
LO
tt
^
O
LO
^
^^
LO
^
^-
LO
i-H
0
I""^.
*
LO
to
LO'
r 1
O
00
0
II
-e-
^
(N
LO
""^
00
^f
LO
1-1
LO
00
*
LO
to
LO
^
T^-
^J-
. 1
(N
\O
^'
OO
1 1
LO
CM
CM
LO
r-{
CM
en
*
^
^O
^t
J^
t*^.
.
2
^-
"*
r-
to
^
^H
O
O
f-H
II
-e-
en
CM
to
i l
to
CM
.
to
1 1
CM
TJ-
to
^
O
to
to
CM
to
1-1
o
CM
to
OO
to
to
vQ
LO
e
CM
r-4
LO
r-l
to
OO
\O
CM
i-H
O
i 1
i-l
II
-e-
cn
LO
o
i i
og
r 1
o
1 1
en
CM
o
LO
to
0
OO
CM
o
i-H
^J.
to
en
LO
00
en
en
0
o
r-H
o
00
*
o
en
o
1-1
LO
(N
r-l
II
-e-
vO
LO
o
i-H
vO
r^
en
^
^j-
0>
r^
^j*
en
vO
LO
en
vO
t^
en
rf
00
.
en
vO
OO
.
en
^
vO
en
\o
LO
en
0
to
u
e-
168
-------
CJ3
co
O
I-H
oi
>
H
U
tu
w
Qi
LU
o-
LU
CO
o
a;
LO
-H
II
U
UH
0£
O
2
LO CM
rH Z
ICO
o
o
to
LO
r--
CM
0
LO
CM
LO
CM
CM
O
O
CM
LO
r 1
0
LO
i 1
LO
CM
r-l
O
0
i-H
LO
0
t
to
o x:
o o
c c A
rt l~1/
/
^r ^
O i-H CM fO
O O O i 1 r-l i-H > 1
II II II II II II II
-e- -e- -e- -e- -e- -e- -e-
169
-------
to co
O
tO i-l
3 H
O <
u
2:
H m
u
tL.
UJ
ex
UJ
o
a:
tN
to
II
Q
co ae
UJ
r*3
HJ
g
CJ
IX
s
o
u
o
a)
H
05
W
o
o
to
LO
^
fN
O
LO
(N
LO
(N
O
0
LO
1 1
0
LO
I 1
LO
cs
I-l
0
o
i-H
LO
o
.t
fl>
(J 4)
C X
at u
/
/ e.
./ o
' »»
^ C3 f *'*₯*
tn T^ .'S' v?
"H / "-v-^ KJ
Q,/ \S £&
o o o o o o o
O O 0 0 0 0 0
o o o o o o o
o o o o o o o
o o o o o o o
o o o o o o o
o o o o o o o
o o o o o o o
o o o o o o o
o o o o o o o
LO O O O O IO O
I"- OO O» O i-l «NJ to
O O O i-H i-l r-t i-H
II II II II II II II
e- -e- -e- -e- -e- -e- -e-
170
-------
o
H
w
u
PU
H- i
OS
o
1
CJ
2
hH
»_j
0,
"SZ
CO
en
o
1 1
OS
<3^
£>
H
CM
X
tt,
O
J2-
O
1 1
H
y
d
UH
tu
1 ]
CO
o
I 1
H
3
w
u
^
«J
< "it
> 0
HH »-H
CX X
w , ,
to >£*
3 ^
O
h- 1
OS ,_,
> rH [ij
1 ^
<
CO X
tq
U
"^ CM CM
H X O
co co co
HH II
Q "
« ""Tvj
I
u
d)
r-H
9
H
o
o
to
LO
CM
0
LO
CM
LO
CM
CM
O
O
CM
LO
f~~
1
O
LO
1 1
LO
CM
t-H
O
0
> 1
LO
o
* f
e xl
rt o
^r Ct
^r _W
*
Po> O
A*£*
^ A!?*
~
LO
O O O CM
0
CTl
LO
O O O to
o
o o o o
O}
0 O 0 CM
o
^
CM
o o o to
o
CTi
o o o CM
0
0
o o o oo
to
o
o o o o
CTi
o o o CM
0
to
O O O to
o
LO O O O
r- oo CTI o
O O O i-t
II II II II
-e- -e- -e- -e-
^
CM
LO
0
^
CM
LO
0
o
r 1
\O
O
O
\^5
LO
°"
1 1
to
^
0
vO
LO
vD
O
00
to
vO
O
^-(
vO
LO
o
CM
^
^
i I
OO
IS>
o
to
f-H
II
-e-
171
-------
UJ
u
rH
PH
hH
Oi
a
CD
O
OJ
UJ
CO
o
u,
UJ
o
CM
o
O CT>
CM
0
o oo
CM
o
o to
to
o
o oo
to
CTi
o to
to
to
o o
to
LO O
f^ 00
9
0 0
II II
-e- -e-
CM
to
'*
to
"*
en
^
CM
*
CM
LO
"*
vo
to
"*
CM
LO
to
00
to
\o
rH
^
O
o
TJ-
o
o
II
-e-
CM
LO
5o
LO
rH
CM
LO
O
LO
o
o
LO
rH
LO
LO
O
0
LO
CM
to
LO
rH
LO
O
LO
O
0
rH
II
-e-
^^
'-O
0
£
VO
vO
rH
^
OO
LO
f-
0
to
^
oo
1 1
^
00
rH
VO
rH
vO
LO
vO
^
s
f^
O
rH
rH
II
-e-
« i
to
o
o
to
00
o
to
0
o
to
00
to
to
o
to
LO
to
CM
to
to
CM
CM
to
rH
LO
to
LO
CM
rH
II
e-
o
o
CM
LO
CM
CM
CM
CM
CM
CM
S
CM
00
TJ-
CM
0
CM
rH
O
CM
LO
O
CM
rH
CM
O
to
rH
II
-e-
172
-------
0(5
O
Pu
UJ
u
o
o
CM
II
a
O
t(
H
Q
I
a
u
u
UJ
oi
i
u
0)
r-l
§
E-
cx
UJ
H
UJ
c£
UJ
u«
UJ
UJ
[ZN] ^-f-^ * 03^ " l001
1 ZODS
I'Nl *** -5* - [-1
[ZN'%
[ZN]
z
I'3'tf v ...S CAM!
T nw LONJ
S
[JN'4
°'HS
3 \
[ Nl \
[ZNI^
''«^
[ZNI%
[ZN]
i^ x !^ = [ZH]
T \
[3N] TT
S
u
c
« 0
cd *->
> 0)
H Ctf
&
U
o
o
oo
O
« <
o
un
to
f-H
OO
r-H
to
o
i i
«^
LO
UO
LO
, 1
CM
CM
t
O
i «
X
CM
1
O
-H
X
"J"
o
*-H
X
to
1
o
1 1
X
CM
1
o
1 t
X
O
O
o
o
o
o
o
o>
0
II
-e-
o
o
to
0
t 4
o
CM
1
OO O
CM i 1
r-H X
CM
1
** 0
LO -!
-i X
1
r-1 O
CM rH
CM X
to
1
1-1 O
tO .-H
in x
CM
i
f- O
cj> i*
<-t X
o
o
o
o
o
o
o
o
,_,
II
-e-
rt
O
O
OO
CJi
o
o
CM
1
CM O
to ,-<
-i X
CM
to 1
-H O
0 rH
0 X
"f
00 O
CM r-t
tO X
to
1
tt O
tO r-H
iO X
CM
1
CM O
Ol r-l
rH X
0
o
o
o
o
0
0
r-l
rH
II
-e-
173
-------
Table C-15. TIME-OF-FLIGHT TEMPERATURE MEASUREMENTS, K
^xDistance,
^X^nches ^
f^\
4> = 0.75
= 0.90
<(> = 1.00
* = 1.10
* = 1.25
= 1.30
1.00
1896
1968
2163
2209
2192
2148
2109
2.00
1861
1934
2011
2043
2033
1950
1918
3.00
1785
1805
1933
2000
1985
1920
1907
Table C-16. THERMOCOUPLE-TEMPERATURE MEASUREMENTS, K
^xpistance,
finches ^
f^X
<(> = 0.75
4> = 0.80
4 = 0.90
= 1.00
4> = 1.10
t = 1.25
ij) * 1.30
1.00
1506
1552
1649
1711
1627
1571
1613
2.00
1453
1465
1621
1659
1602
1556
1598
3.00
1282
1299
1383
1498
1453
1391
1413
NOTE: The temperature increase as $ increases from 1.25
to 1.30 might be related to the relatively
unstable flame observed at 41 * 1.30.
174
-------
APPENDIX D
THEORETICAL*1 MAXIMUM FLAME TEMPERATURE AND SPECIES
COMPOSITION OF METHANE - AIR MIXTURES13
AT VARIOUS EQUIVALENCE RATIOS0
COMPOSITION - MOLE PERCENT
"^s. Equivalence
1 Species'1-.^
H
0
Ar
°H
H2
H20
CO
co2
NO
N02
N2
°2
0.70
0.00033
0.00420
0.83817
0.07220
0.00443
13.64600
0.00872
6.83462
0.25436
0.00030
72.60706
5.72961
0.75
0.00095
0.00773
0.83389
0.11165
0.01047
14.52675
0.02224
7.27452
0.29556
0.00027
72.21524
4.70070
0.80
0.00245
0.01273
0.82954
0.16075
0.02301
15.38109
0.05204
7.69081
0.32562
0.00023
71.82288
3.69883
0.90
0.01203
0.02419
0.82022
0.26714
0.09557
16.9879.3
0.23536
8.37618
0.32458
0.00013
71.01418
1.84246
1 00
0.03990
0.02190
0.80839
0.28650
0.36763
18.33098
0.90434
8.52657
0.19930
0.00000
70.05022
0.46421
1 10
0.06781
0.00555
0.79052
0.14471
1.25004
18.93414
2.60837
7.53686
0.05299
0.00000
68.57321
0.03576
1 .23
0.05838
0.00043
0.75992
0.0360"
3.53288
18.58275
5.32788
5.75355
0.00659
0.00000
65.94055
0.00098
1 30
0 050-8
0 00019
0 "300)
0.02305
4.419"9
18.29323
6.06929
5.30569
0.00353
0.00000
65.08405
0.00035
Equivalence Ratio
0.70
0.75
0.80
0.90
1.00
1.10
1.25
1.30
Flame Temperature,
1840
1922
2000
2137
2229
2214
2100
2061
K
3
Based on Rocketdyne n-element propellant performance computer program
Accounts for establishment of definite equilibrium conditions between
products and reactants
"Equivalence Ratio = (CH./Air)/(CH./Air)_ . ,.
4 '/ 4 JStoichiometnc
175
-------
APPENDIX E
FACTORS FOR THE CONVERSION OF UNITS
TO THE METRIC SYSTEM
Physical
Quantity
Length
Mass
Pressure
Speed
Temperature
Time
Volume
To Convert
From
inch
foot
pound
atmosphere
inch of water
psi
torr 2
newton/meter
foot/ second
Celsius (C)
Fahrenheit (F)
Fahrenheit (F)
Rankine (R)
hour
Multiply
by
meter
meter
kilogram
pascal
pascal
pascal
pascal
pascal
gallon (U.S. liquid) meter
Kelvin (K)
Kelvin (K)
Celsius (C)
Kelvin (K)
second
3
0.0254
0.3048
0.45359
101325.
248.84
6894.8
133.32
1.0000
meter/second 0.3048
K = C + 273
K = (F + 460)/I.8
C = (F - 32)/1.8
K = R/1.8
3600.
0.0037854
176
-------
TECHNICAL REPORT DATA
/7V< <;\< read /ya//m timf>!
1 HI I'GH F NO
EPA-650/2-74-023
.1 FITLb AND SUBTITLE
Flame Characterization Probes
6 PERFORMING ORGANIZATION CODE
IS) R c Kesselring (Rocketdyne), and K. M
..orjiW.S. Young, W.E.Rodgers, and E. L. Knuth
LJ V> J-J J\ / J.,.^_.. -_-_.-._.
8 PERFORMING ORGANIZATION RtPORT NO
a
PERFORMING OR3ANIZATION NAMt AND ADDRESS
Rocketdyne Division, Rockwell International Corp.
6633 Canoga Avenue
Canoga Park , California 91304
J ML CII'IL N f 5 ACCLOSION»NO
REPORT DATE
March 1S74
70 PROGRAM ELEMENT NO.
1AB014; ROAP 21ADG-51
11 CONTRACT/GRANT NO
68-02-0628
12 SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
NERC-RTP, Control Systems Laboratory
Research Triangle Park, North Carolina 27711
13 TYPE OF REPORT AND PERIOD COVERED
Final
14 SPONSORING AGENCY CODE
15 SUPPLEMENTARY NOTES
16 ABSTRACT
gives results of work dealing with the problem of extracting
temperature, species concentration, and velocity data from flames. A literature
review was conducted to determine the state-of-the-art for making the following
measurements in a particulate -laden flame environment: flame temperature--1100 to
2500°C; stable chemical species --NO, H2 , O2 , CO, SO2 , CO2 , N2 , NO2 , and Ar:
unstable chemical species--O, N, OH. H, and other flame intermediates; and
velocity- -magnitude and direction. Based on results of the literature review, three
separate probes were designed and fabricated to make these measurements. In order
to measure the unstable species and also to provide a calibration reference for the
stables species and temperature, a molecular beam mass spectrometer equipped
with a time-of-flight chopper was used.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b IDENTIFIERS/OPEN ENDEDTERMS
c. COSATI I Icld/Group
Air Pollution
Flames
Measurement
Probes
Mass Spectrometers
Molecular Beams
13B
21B
14 B
jTniUUTION STATfcMENT
Unlimited
19. SECURITY CLASS / flill Ki'/wrlJ
Unclassified
21 NO. OF PAGES
186
20 SECURITY CLASS flIns page)
Unclassified
EPA Form 2220-1 (9-73)
177/178
------- |