EPA-600/2-77-007
JANUARY 1977
Environmental Protection Technology Series
                                                                        xxxxxxxx,
                                                                        xxxxxxxx;
                                                                        xxxxxxxx;
                                                                        xxxxxxxx)

                                                                        xxxxxxxx>


-------
               RESEARCH REPORTING SERIES

Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:

    1.   Environmental Health Effects Research
    2.   Environmental Protection Technology
    3.   Ecological Research
    4.   Environmental Monitoring
    5.   Socioeconomic  Environmental Studies

This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to develop and
demonstrate instrumentation, equipment, and methodology to repair or prevent
environmental degradation  from point and non-point sources of pollution. This
work provides the new or  improved technology required for the control and
treatment of pollution sources to meet environmental quality standards.
                    EPA REVIEW NOTICE

This report has been reviewed by  the U.S.  Environmental
Protection Agency, and approved for publication.  Approval
does not signify that the contents necessarily reflect the
views and policy of the Agency, nor does mention of trade
names or commercial products constitute endorsement or
recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                                   EPA-600/2-77-007

                                   January 1977
         DEVELOPMENT OF A

   LASER  VELOCIMETER  SYSTEM

         FOR  FLAME STUDIES
                     by

H.T.  Bentley, IE and B.W.  Bomar (ARO, Inc.)

               U.S. Air Force
           Arnold Air Force Base
         Tullahoma, Tennessee  37389


    Interagency Agreement No. IAG-189(D)
            ROAPNo. 21ADG-048
         Program Element No. 1AB014


  EPA Project Officer: William B. Kuykendal

 Industrial Environmental Research Laboratory
   Office of Energy, Minerals, and Industry
      Research Triangle Park, NC  27711


                Prepared for

U.S. ENVIRONMENTAL PROTECTION AGENCY
      Office of Research and Development
            Washington, DC 20460

-------
                              PREFACE
    The work reported herein was conducted by the Arnold Engineer-
ing Development Center (AEDC), Air Force Systems  Command (AFSC),
at the request of the Environmental Protection Agency, Research
Triangle Park, North Carolina, under Program Element 1A2014.  The
results of the research were obtained by ARO, Inc. ,  AEDC Division
(a Sverdrup Corporation Company), operating contractor for the  AEDC,
AFSC, Arnold Air Force Station, Tennessee,  under ARO Project
Numbers B32S-11A and BF-406-134A.  The authors of this report were
H. T.  Bentley III and B. W.  Bomar, ARO, Inc.  The manuscript
(ARO Control No.  ARO-OMD-TR-76-78) was submitted for publication
on July 22,  1976.

    The authors are indebted to W. M. Farmer, Spectron Develop-
ment Laboratories, Tullahoma, Tennessee, and  H. R. Bevis,  ARO,
Inc.,  for their work on the initial studies of the "rainbow" furnace,
which contributed greatly to this report; to T.  O.  Bolden,  Sci-Metrics,
Tullahoma,  Tennessee, for his design and fabrication of the  hardware;
to D.  W. Roberds,  ARO, Inc. , for his contributions to the particle-
sizing studies; and to A. E. Lennert, ARO, Inc. , who furnished
guidance throughout the project.

-------
                             CONTENTS
1.0 INTRODUCTION	     7
2.0 DESIGN CRITERIA
    2.1  Turbulence Effects	    10
    2.2  Depth of Field	    10
    2.3  Description of Experimental Apparatus	    14
    2.4  Experimental Results	    16
    2.5  Summary  of Rainbow Furnace Design Criteria  ...    26
3. 0 FLAME ANALYZER SYSTEM DESCRIPTION
    3. 1  Self-Aligning,  Two-Component Laser Vector
         Velocimeter   	    28
    3. 2  LV Signal Separation and Spectrum Translation
         System	    31
    3.3  Optical  System	    33
    3.4  Traverse  Assembly	    38
    3. 5  Doppler Processor System	    39
    3.6  Data Multiplexer	    43
4. 0 EVALUATION  OF THE APPLICABILITY  OF
    INTERFEROMETRIC PARTICLE SIZING	    44
5. 0 THIRD COMPONENT OF VELOCITY	    49
6.0 SUMMARY AND CONCLUSIONS	    51
    REFERENCES	    52
                          ILLUSTRATIONS
Figure
    1.  EPA Rainbow Furnace with Photometer	     8
    2.  Location of Furnace Observation Positions	     9

    3.  Optics for Interference Fringe Generation	    11
    4.  Relative Collection Efficiency versus z	    13

    5.  Probe Volume Image Scanner System	    15
    6.  Laser Velocimeter System	    15

    7.  Spectral Radiance as a Function of Wavelength
       (Fuel Oil at Position I)	    17

-------
Figure                                                        Page
   8.  Spectral Radiance as a Function of Vane Position
       (Fuel Oil at Position I)	    18
   9.  Spectral Radiance as a Function of Wavelength
       (Fuel Oil at Position IV)	    19
  10.  Spectral Radiance versus Wavelength as a Function
       of Furnace Warmup Time	    20
  11.  Spectral Radiance versus Wavelength (Propane Gas
       Flame)	    21
  12.  Signal Output from Probe Volume Image Scan	    22
  13.  Examples of LDV Signal Distortions	    25
  14.  Two-Dimensional Bragg Cell	    29
  15.  Diffracted Beams from Bragg Cell	    31
  16.  Flame Analyzer Spectra for ±100 m/sec
       Velocities	    32
  17.  Signal Separation and Spectrum Translation
       Systems	    33
  18.  Flame Analyzer Optical System Schematic	    34
  19.  Overview of Optics Showing Forwardscatter Unit  ...    36
  20.  Photometer System Schematic	    37
  21.  Traverse Assembly Schematic	    38
  22.  Doppler Data Processor Block Diagram	    40
  23.  EPA Flame Analyzer System	    44
  24.  Scattered Intensity for Three Values of D/6,  z = 0 .  .  .    45
  25.  Theoretical Visibility versus Ratio of Particle
       Diameter to Fringe Spacing	    46
  26.  Theoretical and Experimental Comparison of
       Visibility versus Ratio of Particle Diameter to Fringe
       Spacing for a Single Circular Beam Stop	    47
  27.  Proposed Three-Component Velocimeter
       Schematic	    50

-------
                               TABLES
                                                              Page
1.  Particle Velocities in Fuel Oil Flame	    23
2.  Particle Velocities in Propane Flame (with 60 percent
    Excess Air)	    24
3.  Measured Data Rates	    24
NOMENCLATURE	    54

-------
                          1.0  INTRODUCTION
    The purpose of this report is to outline the basic considerations
and principles of the  design and operation of the EPA flame analyzer
laser velocimeter (LV).  The flame analyzer concept was proposed as
an aid in the development and evaluation of furnace designs.   It also
was intended to supplement the use of conventional instrumentation.
Areas of application  include mixing, burner efficiency, particle
dynamics,  and verification of mathematical models.  The overall
program provided not only  the resultant hardware (laser flame
analyzer) but information concerning the flame environment and the
applicability of laser velocimetry and interferometric particle sizing
for these areas  of research.  Specific details will be minimized, and
references will  be  cited for further detailed study.  The laser velocim-
eter designed for the Environmental Protection Agency (EPA) incorpo-
rates  a number  of compromises.   For instance, the LV was required
to operate under very stringent environmental conditions.  At the same
time,  ease of operation, accuracy, and high data rates were required.
The system which was proposed and built represents a good combination
of the state-of-the-art technology to give accurate velocity flow-field
information within  a  developmental furnace.  The development of a
laser interferometer for use as a velocimeter has been well docu-
mented as to basic principles and applications (Refs. 1-7) and will
not be detailed in this report.   Three-dimensional velocity capability
was included in the design considerations,  but this was not developed.
Provisions were made,  upon completion of the development of the
LV, for simultaneously obtaining particle size information based
upon the processing of fringe visibility  (Refs.  8-10).


                         2.0 DESIGN CRITERIA
     The EPA "rainbow" furnace is shown in Fig.  1.  Basic EPA design
requirements were to develop a velocimeter system to:  (1) furnish
two-dimensional velocity data, (2) incorporate the capability of scanning
across the entire furnace core, (3) provide position readout,  (4) provide
simultaniety of two-component data acquisition, (5) include a provision
for the addition of interferometric particle sizing, and (6) design the
system with sufficient versatility that a third component of velocity
could be added at a later time without major modifications.

     The conditions under which the measurements were to be made
were unfavorable, at best,  as Fig.  2 shows; the velocimeter system

-------
                Figure 1. EPA rainbow furnace with photometer.

was to be mated to the EPA furnace observation ports that were 1-1/4
in. in diameter, with  a minimum viewing distance from the inner core
interface of 7. 25 in.   Traverse was to be made over a distance of 10. 5
in.,  the  inner core diameter.  The furnace environment included the
presence of fuel oil or gas flames with attendant severe turbulence and
density gradients.  The background radiation from the flame provided
a continuum of unknown opacity.  While estimates based on expected
equilibrium temperatures  could give a measure of the background light
level,  a  high degree of opacity and nonequilibrium effects could greatly
modify this finding.  Furthermore, the scale and degree of turbulence
would affect the formation of high quality probe volume fringes.  There-
fore,  an in situ investigation was made to ascertain the magnitude of
these perturbations.

    While distortion of the interference fringes in the probe volume of
a dual-beam velocimeter could affect signal processing, such  distortion
would more dramatically affect the particle-sizing instrumentation, the
operation of which is based on visibility  (Ref.  9).  Considering the  two

-------
basic LV system operating modes,  forward- and backscatter, the latter
type is least disturbed by turbulence effects, if the signal intensity level
is not considered.  In the backscatter mode, radiation must pass back-
wards over the same optical path it took to reach the probe volume,
thereby redrcing the effects of beam wander over the  phototube aperture.
This obviously not the case for forwardscatter systems,  as the effects
are additive and in no way tend to cancel.  For these reasons, a back-
scatter system appeared to be preferable for the EPA application if
the attendant loss in signal intensity level caused by the lower back-
scatter cross section could be resolved by using a high-powered laser.
                                         Variable Geometry
                                         Fuel I njection Region
                               111         IV          LSwirl Vanes
                                            All Dimensions in Inches

               Figure 2. Location of furnace observation positions.
    A second reason for selecting the backscatter mode of operation
is the simplicity of the optical system.  The simultaneous tracking of
the transmitting and collection optics system is inherent with the back-
scatter technique, whereas in the forwardscattering system this is not
easily achieved.  However, for situations where increased signal
strength is required, such as in a gas-fired furnace where the particu-
lates  are sparse,  a forwards catter system was designed and built to
track with the transmitter,  across the probe volume,  in one dimension
only.

-------
 2.1 TURBULENCE EFFECTS

     Experiments were performed in the EPA furnace to determine
 (1) the quality of the probe volume fringes, (2) the quality of the
 scattered signals,  (3) the background radiation levels, and (4) probable
 data acquisition rates.

     Effects of turbulence on beam quality for a dual-beam or dual-
 scatter velocimeter system have been reported previously (Refs.  11-13).
 It has been shown that as long as the separation of the beams at the
 transmitter in the dual-scatter system is,  at most,  one-half the outer
 scale of turbulence or equivalent to the turbulence correlation length,
 relative beam quality  is unaffected by the turbulence.  That is to say,
 high quality fringes will be obtained within the probe volume even
 though gross movement of the probe volume can and will occur because
 of turbulence-induced index-of-refraction changes.  Insofar  as the
system is unaffected by the collection optics in these cases,  neither a
forward- nor a backscatter system is greatly affected by the turbulence.
However, as previously noted,  image wander over an aperture can
 cause variations  in the signal intensity and consequently a distortion
of the signal  profile.   To reduce these effects in a forwards catter
system,  a larger aperture must be employed; consequently,  more
background radiation must be accepted,  with the attendent decrease  in
signal-to-noise ratio.

    In the case of local oscillator  systems (reference beam type), the
scattered wave is perturbed by turbulence as it returns to the collector
prior to the mixing with the unperturbed local oscillator reference
beam.   In this case, the correlation between  the wavefronts and the
phase variation is destroyed.   The net result has been shown to be a
limiting aperture effect, beyond which an increase in the collection
 aperture will not increase the signal-to-noise ratio and in fact may
result in a decrease.  This is one  of the reasons that a reference beam
 system was not selected for this EPA application.
2.2 DEPTH OF FIELD

    To determine the spatial resolution, consideration must be given
to those factors affecting the probe volume.  The probe volume is
normally defined by three criteria:  (1) the geometry of the incident
radiation, (2) the collection optics and aperture system, and (3) the
electronic discrimination.
                                 10

-------
     The first criterion for determining the probe volume is based on
the intensity distribution resulting from the interaction of the two
crossed Gaussian beams.  The one-dimensional velocimeter probe
volume, for Gaussian waveforms, has  been described (Ref. 2).  It
was shown that the ellipsoidal probe volume intensity distribution at
the 1/e2 point (Fig. 3) is given by the following equations:
                           Ax(l/e2) = 2bo                          W


                           Ay(l/e2) = 2bo/cos(a/2)                   (2)


                           Az(l/e2) = 2bo/sin(a/2)                   (3)


where a is the angle between the two beams.  The characteristic
dimension,  2b0,  is given by

                                                                (4)


where f-^ is the focal length of the lens,  2b is the input beam diameter,
and A is the wavelength of light.  (For 125 -jUm fringe spacing,  a 22 -in.
focal length lens, and a 3 -mm beam diameter,  Az is 60 mm. )   Fur-
thermore,  the 1/e2 equal  intensity surface can be shown to be  an
ellipsoid.  The actual volume of this ellipsoid is determined by

                             s    /*TA3      8fL x3
                      V - -^— U-)  - -j±—3                (5)
                           3772 sin a  Xb/    3n-2 D b3
where D is the beam separation at the output lens and sin a as a =
Particles traversing outside of this region contribute little to the signal
intensity.

    The sampled probe volume is further determined from the char-
acteristics of the collection lens  and aperture system.  The effect of
a pinhole aperture  on the collection efficiency for particles on axis is
shown in Fig. 4.  Here the collection efficiency is determined for a
backscatter system with a 45. 7-cm focal length lens.  The effective
limiting aperture is therefore the pinhole image located at the point
z = 91.4 cm.  Relative efficiency is plotted as a function of distance
from the collection lens.  It is assumed that the term "geometric
optics" adequately describes the system and that the on-axis collection
efficiency is also valid in the neighborhood of the aperture image plane.
                                  11

-------
 In practice, this aperture image plane would coincide with the x-y
 plane of Fig. 3, the probe volume center.  The very rapid falloff of
 collection efficiency as one  moves away from the aperture plane is to
 be noted.  A detailed geometric optics evaluation of this character-
 istic, for both  on- and off-axis cases,  is found in Ref.  14.  (For an
 aperture image of 1 mm  and system f number  of 10,  the characteris-
 tic depth of focus  is ± 1 cm.)
                                                  Interference Fringes
                                                  Probe Volume
                                  Enlarged View of Region
                                  of Cross-Focus Point
      Laser
                                                    Aperture (Pinhole)
       Transmitting  x
15MHz  °Ptics
                                                            PM
                                                            Tube
                                             Collection      i
                                             Optics    To Preamplifier
                                                     and Signal
                                                     Processor
                 Figure 3.  Optics for interference fringe generation.

     The analysis of collection efficiency assumes that there is isotropic
scattering along the optical axis.  This is not strictly the  case,  as in
practice the radiation is strongly lobed in the backward  and forward
directions in a manner which is  dependent upon the particle size,  the
index of refraction,  and the geometry of the particle.  Therefore,  the
particles themselves can contribute to the effective depth  of field.   In
the case of forwardscatter where Fraunhofer diffraction applies, for
particles of diameter D, the first minimurr-  occurs at an angle from
the axis given by
                    - ^
                                                                  (6)
This angle therefore determines an effective collection solid angle
for that particular size particle when on-axis collection is being used.
                                   12

-------
Inherent with the laser flame analyzer design, where the central beam
stop (Fig.  3)  is located on axis to block the direct unscattered radiation,
the overall effect is to increase the relative system response to small
particles.   The system response, as  it relates to particle size process-
ing, is dependent upon the collection lens diameter.   The effects of the
beam stop diameter will be discussed at greater length in a later
section.   Electronic discrimination occurs because of low signal to
noise ratio near the limits of the depth  of field or through the coin-
cidence requirement with other electronics such as a visibility
processor. To formulate a mathematical  statement of these effects
is difficult, at best.  The actual effective probe volume is best found
experimentally.
                                    Aperture Image Plane
                10'
                 ,-5
                   0   0.50   1.00   1.50   2.00  2.50   3.00   3.50
                                    z, m

                 Figure 4. Relative collection efficiency versus z.
                                  13

-------
2.3 DESCRIPTION OF EXPERIMENTAL APPARATUS

     Several preliminary experiments were performed.  Radiometric
measurements were made to determine the effects  of background
radiation.   The quality of the probe volume fringes was examined to
determine the effects of turbulence.  Finally,, signal quality and data
rate were  measured.  The apparatus is described in the following
section.

2.3.1 Spectral Radiometer

     The spectral radiometer used in these experiments employed a
reflection-grating monochromator which was capable of scanning the
visible  spectrum from 0. 350 to 0. 800 jum with a resolution of 0. 001
Mm. The photodetector utilized an S-20 type photocathode, and the
signal values were read with a 3-1/2 digit, digital display.  The mono-
chromator was coupled to a  photometric telescope equipped with
selectable limiting apertures of 6 min,  20 min,  1 deg, and 3 deg.  The
telescope was capable of a 1-m to infinity focusing range.  The radiom-
eter was calibrated for 0. 442,  0.448, 0.5145, 0.633,  and 0. 694 jum
against a tungsten ribbon lamp traceable to National Bureau of Standards.
2.3.2 Laser Velocimeter Transmitter

     The laser velocimeter transmitter used in the experiments utilized
a 15-mw helium-neon (He-Ne) laser operating at 0. 633 /urn.  The laser
beam was split into two parallel beams with a glass prism, and the
intensities in the beams matched to within 15 percent of each other with
a neutral density filter.  The probe volume was generated with a 500-
mm focal length lens. Fringe period was 17 jum, and the e~^ modulation
contour probe volume was approximately 2.4 x 10~3 cm  .  The
transmitter was mounted on a heavy duty tripod,  and the probe volume
was positioned in the flame by manual adjustment of the tripod.
2.3.3 Probe Volume Image-Sampling System

     To provide information concerning the quality of the probe volume
fringes, an image-sampling system was used.  Figure 5 is a schematic
of the image-sampling system.   After passage through the flame, the
beams were collected with a lens and brought to focus to form an image
of the probe volume. The probe volume image was scanned with a
50-jLtm tungsten wire.  The scanning  wire was mounted on the solenoid
of an acoustic speaker so that it could be driven across the probe volume
                                  14

-------
image at a rate determined by an input driving frequency.  The light
reflected from the wire edge was collected by a lens and focused onto an
RCA 931A photomultiplier (PM) tube.  The signal current as a function
of time -was  displayed on a storage oscilloscope.
   Signal
   to Storage
   Oscilloscope
                                                              V. Supply
                                                                   II"
                                                               PM Tube
                                                               Detector
                         Flame Region
                Speaker Cone
                Scanner
               Oscillator
                   Figure 5.  Probe volume image scanner system.
 2.3.4 Laser Velocimeter

     The velocimeter transmitter described previously was used with
 the radiometer as a scattered light detector to form a forwardscatter
 velocimeter system which could be used to observe large scattering
 particles in the flame.  A schematic of the arrangement is shown in
 Fig.  6.  After transmission through the flame the illumination beams
 pass to a beam stop in front of the photometric telescope.   A portion
 of the scattered light passes around the stops and through the  radiom-
 eter  onto the PM tube detector. The signal from the particle  was
 displayed on a storage oscilloscope.
                -Beam Splitter
                  Block
                                  HomosiI Windows
Scattered
Light
Collection
System
            H. V. Supply
   Signal      • •  •   .,
    *—II—"I'M—li'1
Collecting M
       [j—PMTube
          -Radiometer
      Pinhole
      Aperture

                        Figure 6.  Laser velocimeter system.
                                     15

-------
2.4 EXPERIMENTAL RESULTS

    A schematic of the furnace on which these measurements were
made is shown in Fig. 2.  Radiometric observations were made on
positions I, III, and IV,  Band C.  Positions III and IV contained ports
approximately 1-1/4  in. in diameter.  The external ports of these
positions comprise Homosil®  quartz windows with less than 1/8 wave-
length surface distortion and with less than 10 arcseconds of wedge
angle between the surface faces.  Viewing port I was approximately 3
in.  in diameter and of unknown optical quality.  Both fuel oil and
propane gas flames were observed in these measurements.
2.4.1 Radiometric Results

    The results of the radiometric measurements are summarized in
Figs. 7 through 11.  Figure 7 contains a plot of the spectral radiance
of a fuel oil flame as a function of wavelength with the position of the
swirl vanes as a parameter, as shown in Fig.  2.   The radiometer  was
focused at position IV-B.  Visual observation of the flame showed that
it was torroidal in cross section and that the size of the torroid center
was a sensitive function of vane position.  Thus, there is a  considerable
variation in the radiance measurements as a function of swirl vane
position.  It is interesting to note that for a constant wavelength the
data,  as shown in Fig. 8, reveal a remarkably consistent trend as the
vane position is varied.  Since spectral radiance is a measure of the
energy output of the source it may be concluded either that more
particles are being burned at swirl position 6 than at  position 5 or  that
the temperature of the flame has increased considerably. Because of
the torroidal shape of the flame, it is suspected that the former
possibility is the cause of the variation.  The measurements presented
in Figs.  7 and  8 were repeated for observation position IV-C,  and  the
results  are presented in Fig. 9.  There appears to be relatively little
radiance dependence on vane position,  in contrast  to the previous  case.
The relatively high radiance values  and insensitivity to vane position
suggest that previous variations were caused by fluctuation  in flame
shape.  Figure 10 illustrates the contribution made to the radiance by
furnace walls,  as  the furnace warms to equilibrium conditions. As
expected, the radiant output of the furnace walls can be significant
when flame/wall equilibrium is reached.  Figure 10 shows-that approx-
mately 50 percent of the radiant power produced for the furnace is from
the walls, as the effects of window degradation are not significant.
Figure 11 summarizes the radiometric observations of a propane  gas
                                  16

-------
flame.   Note that the radiance values for the gas flame were significantly
lower than for the oil flame; these lower values suggest that there is
much less  solid material in the flame,  as should be expected.  Further-
more, visual  examination of the gas flame revealed that it was not nearly
as opaque as the oil  flame.
      102
  o>    m
  ts    10
  9
  a.
  of
  o
  ro
  'S
              Position I - Fuel Oil;
              30-deg Spray Nozzle;
          	14% Excess Air
                               600                   500
                                  Wavelength, nm
              Figure 7. Spectral  radiance as a function of wavelength
                       (fuel oil  at position I).
                                    17

-------
      102
 c

 l_
 o>
CM

 E
 O>
 o
 ra

                     34567

                                     Swirl Vane Position


               Figure 8.  Spectral radiance as a function of vane position

                         (fuel oil at position 1).
                                         18

-------
     103
 c
  I
CM

 E
 u
 ra


 TB
 O£

 "ic5
      10







°-^
fi

























^















Fin
No;
14?l







^x|















3lON;
>.z\e (h
» Exce








rTh*
v/1^













30-de
lollow
ssAir









..
"^^^
^











gSpr
Cone











v
X










rtV
ay
>;














N
>

























\
S


















Swirl Vane Positior
a #2
o 13
A #4
0 #5
v 16
o #7
« #8






ss
N
















X
fi
















No
Ov
'^ "\
_fi_
















\
\
>







1

















"y
\
t
&



























        700
600
500
420
                                       Wavelength,  nm
                 Figure 9.  Spectral radiance as a function of wavelength

                           (fuel oil at position IV).
                                          19

-------
     102
      10
 O)

 to
  i
«VJ
 to
     10"





-0-^
c^

























^s















Po
Fu
02
30






^v
\














sition
el Oil
, 22%
% inj<






Sf-i
^
X.
?Vv
s











1MB
= 16%
C02;
Action







\



s
\









Exces
Cone







N
^y
^



\
S








;s




o


D



<
>v
\


v
\












1

All Vane Settings. Furnace
Immediately after Startup.
All Vane Settings with Furn
Hot for Three Hours.





\
N

y
\
















\
N
\
>,
















X
s

1-^^
X
\
\













X





\w
\











V
N.
>v





X.






ace













y_
Nn
TJ


^^



























        700
600
500
                                      Wavelength, nm
              Figure 10. Spectral radiance versus wavelength as a function

                        of furnace warmup time.
420
                                        20

-------
 E
 «/»
CM
 to
 1
       10
     ID
       '2
         700
                           u
                                               n  24% Excess 02; 39% Excess C02
                                                  Position III
                                               o  60% Excess Air,  Position 111
                                               A  63% Excess Air,  Position IV

                                                 \
600
500
420
                                       Wavelength,  nm

         Figure 11.  Spectral radiance versus wavelength (propane gas flame).
                                         21

-------
2.4.2 Probe Volume Image-Sampling System Results

    The probe volume sampling system was positioned at observation
port IV.   The image was sampled at a rate of 15 scans/sec for a large
number of flame parameter variations.   For this sample rate and a
633-nm wavelength, virtually no changes  in the probe volume image
could be detected.  Figure  12 shows an example of such a scan, taken
while the furnace was not in operation.  The slight distortion in the
image is  caused by the optical imaging and scanning system.  This signal
shape is identical to those obtained when the furnace  was in operation.

    The data should be interpreted only as indicative,  and not as a
definitive statement that turbulent distortion of the probe volume is
insignificant.  The nominal 100-jnsec scan time occupied only about
0. 3 percent of the time over  which a signal distortion could occur.
Particles with longer residence times within the probe volume could be
subject to slower-changing turbulence effects.
              o>
              :«
              o>
              a:
                               Time

             Figure 12. Signal output from probe volume image scan.
2.4.3 Laser Velocimeter Results

    The data obtained from the velocimeter observations are summa-
rized in Tables 1 through 3 and in Fig.  13.  Both fuel oil and propane
gas measurements were made.  In order to obtain a sufficient data rate
for the  observations, it was necessary to utilize approximately 60
percent excess air in the gas flame measurements.   Measurements of
the fuel oil flame were made at positions IV-B and C, and a large
number of scattered light signals were observed.  The vast majority of
                                  22

-------
  the signals observed at position IV-B contained little or no velocity
  information:  the fuel droplets were evidently too large for the observa-
  tional fringe period to produce signals with high visibility.  When the
  observation position was changed to IV-C (i. e. , near the edge of the
  fuel spray), most of the signals again contained no AC information,
  although some measurable signals did appear. However,  at position
  III-B the sizes of the fuel droplets were much reduced,  and a relatively
  large number of information-carry ing signals could be observed.   As
  Tables 1 and 2 show, the velocity of the particles  in the oil and gas
  flames are comparable at positions III-B  and  C, with the fuel oil
  probably having  a slightly higher velocity.  Table  2 shows that the
  velocity of the particles in the gas flame near the  fuel injection region
  is considerably higher than that observed further down the flame path.
  In fact,  it appears that the particle velocity decreased by a factor of
  3 to 4 in a distance of about 15 in.
                  Table 1. Particle Velocities in Fuel Oil Flame
                   Measured  Particle  Velocities,  m/sec
Position IIIB
2.88
3.76
4.24
4.36
4.72
4.95
5.31
5.9
6.14
6.78
7.08
Position IIIC
1.18
1.76
3.82
4.12







                       7.79
                       8.25
           Average:   5.55                        2.72

                  Table 2.  Particle Velocities in Propane Gas Flame
                          (with 60-percent Excess Air)
                   Measured Particle Velocities, m/sec
      Position IIIB    Position IIIC    Position IVB    Position IVC
           2.35             1.18            11.8            11.3
           4.0              1.76            23.6            12.2
           4.7              3.23                             16.5
                             3.82                             17.6
                             4.12
Average:  3.683            2.822           17.7            14.4
                                  23

-------
                        Table 3. Measured Data Rates
                               a.  Fuel Oil

                        Data Rate,  number/sec
                 Position  IIIB             Position  IIIC
                       17                          19
                       32                          36
                                                   20
                                                    8
            Average:  24.5                       20.75
                              b. Propane Gas
                           Data Rate,  number/sec
      Position  IIIB   Position IIIC   Position IVB    Position IVC
           0.22             0.52             0.74            0.66
                             0.32
Average:   0.22             0.42             0.74            0.66
       Data rate measurements were obtained by allowing the oscilloscope
  to sweep over a relatively long time interval and to record all scattered
  light pulses.  Examples of these values are tabulated in Table 3.  Table
  3 clearly shows the expected discrepancy in data rate between the gas
  and oil flames.  It should be noted that the data rates recorded in Table
  3 are for total input signals into the PM tube.  It is estimated that 10
  to 20 percent of these signals would provide velocity information and
  that  of those signals perhaps only 20 to 50 percent could be analyzed
  for particle size.  Furthermore,  it should be  emphasized that the data
  represent relatively long time averages.  Thus, in any specified interval
  of time the data rate can be considerably higher or lower  than the values
  presented in Table 3.   The reason for the large discrepancy in the
  recorded versus usable data rate lies in the fact that the receiving lens
  on the radiometer had a relatively large acceptance angle which observed
                                   24

-------
                   Time
                  Time
Figure 13. Examples of LDV signal distortions.
                   25

-------
not only the probe volume but also light scattered from the individual
beams forming the probe volume.  Signals were also observed,  partic-
ularly in the oil flames, that were so distorted as to be virtually unusable.
Examples of such signals (which were not unique) are shown in Fig. 13.
The mechanism that produces such signals is not presently understood,
but the signals could be caused by (1) multiple particles in the probe
volume,  (2) asymmetric particle shapes, (3) particles changing size and/
or shape while in the probe volume, (4) turbulent distortion of the inter-
ference fringes, or (5) window contamination,  or any combination of the
foregoing.   Considerable work remains to be done to determine the cause
of these phenomena.
2.5  SUMMARY OF RAINBOW FURNACE DESIGN CRITERIA

    Radiometric,  turbulent distortion, and velocity measurements have
been performed on large oil and gas flames.  The radiometric measure-
ments show that much less background light exists for light in the blue-
green laser lines (0. 4880 and 0. 5145 ^m) than in the red lines (0. 6328
Mm).  Since turbulent fluctuations did  not appear to have a significant
effect on the operation of a velocimeter system for a red line, it appears
that a laser line at around 0. 500 Mm would be a judicious choice for a
flame analyzer.  Since the argon ion laser offers such a wavelength at
the higher power levels required for a backscatter device, the data
strongly suggest the use of an argon laser for the EPA application.  A
10-A, half-band width (HBW) laser line filter will be sufficient for the
system when used with a 10- to 15-mw laser velocimeter in a forward-
scatter observation mode or a 500-mw (approximate value) laser
velocimeter in a backscatter mode.

    The velocity data verified that measurements can be obtained in
oil  and gas flames in a relatively  straightforward fashion. For a
reasonably high data rate, however,  it will be necessary to run the
gas flame  with a high value of excess  air, to supply airborne  particles,
or with artificially generated scattering particles.  Furthermore,  in
order to obtain velocity measurements near the injector cone for fuel
oil  flames it will be necessary to  employ a probe volume with a large
fringe period.   The data arrival rate for the oil flames was observed to
be reasonably uniform when compared to gas flames as evidenced by
the low data rates for propane in Table 3.  Observation of the gas flame
showed that the scattering particles tended to cluster in pockets, with
the result that there were long time intervals over which no data were
obtained.  When the cluster of particles passes through the probe volume,
                                 26

-------
the data rate becomes quite high.  Such effects can seriously bias the
statistics used in determining turbulent velocity characteristics of the
air flow in the flame and must be considered in future analyses.
               3.0  FLAME ANALYZER SYSTEM DESCRIPTION
    The environment to which a laser velocimeter was to be matched
and integrated involved regions of recirculation and high degrees of
turbulence.  In addition, variable amounts of swirl were inherently
built into the burner assembly, further  complicating the velocity flow
field.  In cases  such as this, where mixing is occurring, reversals in
the velocity vector in both space and time are expected.  Without a
priori knowledge of the flow field, a conventional laser velocimeter,
which lacks directionality resolution, would be unable to detect the
sign of the velocity vector and would give only the speed.  The flow
field to be  encountered can be assumed  to be asymmetric.  With a
further assumption that the radial velocities are minimal, a two-
dimensional,  Bragg cell laser velocimeter  capable of scanning
across the flow  field could provide both axial and azmuthal velocity
components.  The capability for performing higher order correlation,
other than  the simple velocity mean,  requires the simultaneous
acquisition of two or more velocity components.

    LDV systems can range in complexity from a simple forwardscatter
system consisting of a laser, a beamsplitter,  a lens,  and a collection
optics package using an oscilloscope for data analysis for speed measure-
ments,  to the sophisticated dual beam system with Bragg cell beam-
splitting and frequency separation techniques to provide velocity informa-
tion.  In any system, however, four essential elements are required to
obtain velocity field information.  They are:

         1.  A laser source and transmitting and receiving optics.

         2.  A photodetector, with its associated signal-conditioning
            and interface  electronics.

         3.  Signal processor and data acquisition electronics.

         4.  Data processing.

    For speed,  versatility and accuracy, all phases of the system must
perform in an optimum manner.
                                 27

-------
    The first subsystem provides a well defined probe volume and a
collection optics system which will collect the scattered light containing
the velocity information from the probe volume.  For backscatter
operation a nominal 2-w argon laser operating at 0. 5145 jum would
provide more than adequate power for most applications operating in the
TEMQQ or Gaussian mode.  The net effect of higher order modes is to
introduce extraneous frequencies caused by the slight frequency variation
from mode to mode.  The optics further define the probe volume by
transmitting and intersecting the four output beams of a two-dimensional
Bragg  cell system to a common cross and focal point.   For maximum
contrast in the fringes, path differences between velocity component
beam pairs should be zero.

    The propagation of Gaussian beams through optical systems has  been
discussed in considerable detail (Ref. 15).  The effect of a mismatch
between the crossover points and the focal point of the individual beams
can result not only in a larger probe volume but also in a spreading of the
observed frequency spectrum (Ref.  16).  This is of particular importance
when a turbulent intensity measurement or particle size measurements
are to  be made.  It can be shown that when one is in the far field of the
beam waist the fringes form hyperbolic curves in space.  This causes a
variation in the spatial frequency of the fringes and consequently a
variation in the Doppler signal as a function of particle trajectory. The
problem is usually found to be more  critical when very  small fringe
spacings and probe volumes are desired (that is,  less than or equal to
10 X.   A discussion of the parameters influencing this broadening effect
in the frequency spectrum is given in Ref. 16.
3.1  SELF-ALIGNING, TWO-COMPONENT LASER VECTOR VELOCIMETER
    While most contemporary laser velocimeters utilize the dual beam
concept, the Bragg cell-based velocimeter simplifies the optics usually
required with the more conventional design.  Furthermore, the system
is self-aligning and allows two vector components of velocity to be made
after  initial alignment has been completed.  The two-component velocity
signals  are  detected with a single photodetector.  Conventional techniques
utilizing stationary fringe systems are ambiguous with respect to the flow
direction.  This directional ambiguity severely limits the use of such
instruments in all but the simplest flow field situations  (Refs. 1,  17, and
18).
                                 28

-------
     The Bragg cell,  dual-beam velocimeter used in the EPA flame
analyzer takes advantage of the beamsplitting and frequency-shifting
characteristics of a two-dimensional Bragg cell.  The .beamsplitting
characteristics simplify the optical system.   The frequency-shifting
characteristics separate the two components of velocity into different
frequency ranges for detection and analysis.

     The two-dimensional Bragg cell  (TDBC) consists of two Bragg cells
in a common housing arranged such that the  centerlines for the cells
are  coincident and orthogonal  (Fig. 14).  At the base of each leg is an
x-cut quartz transducer disk (Bragg crystal) designed to oscillate in
the third overtone.  The operating frequencies of the two transducers
for the EPA flame analyzer are 15 and 24 MHz.
                                            Water
                                             Optical Quality Window
                                                Bragg Crystal
                                                        Oscillator
                                           /   L	Amplifier
                                                Bubble Vent

                                              Impedance-Matching
                                              Networks
                 Amplifier
                   Oscillator
                    Figure 14.  Two-dimensional bragg cell.
    Each transducer is driven by a 3-w broadband R-F power amplifier
which in turn is driven by a precision crystal-controlled  oscillator.
Impedance-matching networks are mounted next to the Bragg cell.
Their purpose is to eliminate wave reflection at the crystal,  reducing
                                  29

-------
the standing wave ratio and optimizing transducer power requirements
(Ref.  17).

    The ultrasonic waves propagate through the media, in this case
water, giving rise to index-of-refraction variations which may be
imagined as two superimposed, linearly independent diffraction gratings
moving at a rate  equal to the speed of sound in the media (Ref. 19).
The diffraction characteristics of the  grating are a sensitive function
of the orientation of the grating relative to the input illumination.
Considering the Bragg mode where half the light is diffracted into a
single first order, the angle between the beams,  |3, satisfies the Bragg
equation

                               sin 13 = ^                     (7>
where X is the vacuum wavelength of the light and A is the wavelength
of sound in the diffracting medium.  The sound wavelength is derived
thus:
                              A =    m                     (8)
where C is the speed of sound in the diffracting medium and fm is the
modulation frequency of the ultrasonic wave (15 or 24 MHz, Fig. 15).
The figure indicates the possible frequency combination for several
different Bragg orders.  The interference fringes generated by any two
of the beams are perpendicular to the line joining the centers  of the
beams.  Therefore, taking the four beams in the upper right-hand
corner, for example,  in addition to the desired 15- and 24-MHz fringes,
39- and 9-MHz crosstalk fringes are also generated.   Other diffraction
modes are evidenced when the acousto-optic cell is operated in the so-
called Raman-Nath region.  The deviated beam is shifted in frequency
from the undeviated beam by an amount Nfj-,,  where f^ is the driver
frequency and N is the order of diffraction.  When a lens  is used to
bring the beams to a simultaneous  crossover in the focus region, or  the
probe volume,  the travelling waves are imaged as moving, mutually
orthogonal interference fringes.  A scattering particle moving in the
same direction as the fringes generates a Doppler frequency less than
the moving fringe frequency, f^.  When the particle moves in the oppo-
site direction,  the resulting Doppler frequency is higher than f^.
                                 30

-------
                                  24MHz
                   Selected
                   Beams—
                                              Crosstalk
                                  (0, 0)
      (-1, -1)
               <-0, -1)

Figure 15. Diffracted beams from bragg cell.
a -i)
3.2  LV SIGNAL SEPARATION AND SPECTRUM TRANSLATION SYSTEM

    The signal from the photomultiplier tube contains the velocity in-
formation of both components (Fig. 16) from the information bands II
and 12.   The modulation frequency is  centered in the middle of each
band at 15  and 25 MHz, respectively.  Crosstalk channels 1 and 2,
located at 9 and 39 MHz,  respectively,  correspond to the sum and
difference frequencies of the two information bands.  Separation of the
two information bands prior to processing by the Doppler processor
system is accomplished by means  of the signal separation and spectrum
translation electronics.

    Figure 17 shows that the photomultiplier tube output is fed directly
to a low noise preamplifier.  The gain of the preamplifier is 22 db, with
frequency response extending from below  100 KHz to above 100 MHz
and a typical noise figure of less than 7 db.  The signal  is then fed  in
parallel to each of  the two channels for further processing.  The input
                                  31

-------
is a high impedance amplifier followed by a highpass filter.   Filters
and translation oscillators have been selected to suit the requirements
of the laser flame analyzer.  The separation technique will be discussed
in some detail for the 15-MHz channel.   The crosstalk in channel 1 is
eliminated by the highpass filtering beginning at  13. 8 MHz.  Depending
upon the velocity range expected, either a 12. 7- or a 14. 77-MHz
oscillator is mixed for the ±100 m/sec or ±10 m/sec velocity range,
respectively. The corresponding signals are shifted to 2. 3 or 0. 33 MHz,
respectively, and higher frequency crosstalk information is contained
in the range from 10 to 54 MHz.   A lowpass filter beginning at 5 MHz
removes this undesired information, retaining the  low  frequency Doppler
information.  The signal is then fed to the Doppler data processor,
where the period is  determined.
                          o
                          'is
                          E
                                        CM
                                         c.
                                         o
E
i_
o
                                                       CM
Q>
•o
T5.
E

1
Q)

«
t
K
L
C
(

C1
§
*
1
?
•>
5
)

I
$
a
c
^



5 «
N C
D ^
H i-
C
1

1
3
vj
-<
5

1
U
^
r-
L



•> "
-J r«
S s
C
2-

1
N
^
J
4

2
i/
e>
u
e>
ra
to
to
0
39

CT2
•\ u\ if\
sj Cf» NO
^ o
3 {^ 8
                             Frequency, MHz

           Figure 16. Flame analyzer spectra for ±100 m/sec velocities.
                                  32

-------
                                                   Photomultiplier
                                                    Power Supply
            Photomultiplier Tube
                  Low-Noise
                  Preamplifier

                50-n Line
+24 V
                 To Model 6 DDP No. 1
                 2 1-5 MHz Range
                 1 0.1-0.5 MHz Range
Max. Vel.
Max Vel


r r

High nput _.
Impedance \
Amplifier
12.7MHz
Osc.
1
1
1 J
±100 m/sec 24,
± 10 m/sec 1 »
b- '
1
14.77MHz
Osc.


Ch.

13. 8 MHz
High-Pass
Filter
1
•^Mixe
5-MHz
Low-Pass
Filter
L


i
1
-\
r
_
H
Ir
A

gh 1
nped
nplif
Mi
nput Y
ance \
ier
1 i
Ch. 2

21MHz
High-Pass
Filter
xer (>
^
9 MHz
Low-Pass
Filter

i

\
r
i
'
za5°-fi
Terminator
21MHz
Osc.
, 1
Max Vel. ±100 m/sec
Max Vel. ±10 m/sec
' 1
23.7MHz
Osc.

      To Model 6 DDP No. 2
      2 1-5 MHz Range
      1 0.1-0.5 MHz Range
            Figure 17.  Signal separation and spectrum translation systems.
3.3 OPTICAL SYSTEM

     The EPA laser flame analyzer is designed to operate at a wavelength
of 0. 5145 jum,  which is the  predominant wavelength in the argon ion laser
                                     33

-------
(Fig. 18).  Spatial orientation of the fringes within the probe volume to
determine the specific velocity  components is achieved with a dove prism
used in conjunction with a vernier scale.   The positioning accuracy is
approximately 0. 2 deg.
                     Output Mirror
Beam Deflector
Mirror
           To Probe   N>
           Volume
                                     Traverse Assembly
                      "^ Condensing Lens

                        -Mirror

                               Dove
                               Prism
                               Rotator-
     Collimating
     Lens
                                                       -Beam
                                                        Reduction
                                                        Telescope


                                                       -Window
                                                          Crystal
Mirror-
               Figure 18.  Flame analyzer optical system schematic.
3.3.1  Transmitting Optics


     The transmitting  lens assembly is capable of positioning the geo-
metric center of the probe volume ±254 mm (±10 in.) on either side of
the center focus.  The position is indicated by means of a digital display.
Calibration is accomplished with a zero adjust potentiometer and by
varying the power supply voltage to obtain the correct  setting.  The
digital readout is in inches,  as measured from any position set by  the
                                    34

-------
zero point adjustment.  However, due to restraints imposed by the
data multiplexer (to the number of available digits), the recommended
scanning range is from 0 to 20 in.

    The fringe spacing in the probe volume is given by

                               F   y      V
                                    =  M                    <9>
where Fp and FQ are the primary and coUimator focal lengths, respec-
tively, V is the velocity of sound in water (1, 500 m/sec), and M is the
system magnification.  The available ranges of fringe  spacing are from
114 to 240 urn. and from  71 to 150 /um for the  15- and 24-MHz fringes,
respectively.

    The purpose of the beam reduction telescope is to reduce the laser
beam diameter and to properly focus it into the Bragg  cell.  The size
of the beam when focused into the Bragg  cell directly determines the
beam divergence of an individual beam upon leaving the Bragg cell.
The angular separation between the diffracted and undiffracted beams
is determined by the  modulation frequency [Eqs. (7)  and (8)] and there-
fore affects the probe volume size.  The beam reduction optics  consist
of two lenses spaced  approximately the sum of the focal lengths apart.
When used in this mode, the output diameter is proportional to the
input beam diameter, as is the ratio of their focal lengths as deter-
mined by geometrical optics:
                             
-------
separations, and consequently, upon focusing by the primary lens,
one will obtain both a smaller fringe spacing and a smaller probe
volume.  The collimator,  in conjunction with the primary focusing
lens, controls the fringe spacing.

     The primary focusing lens, or transmitter lens,  focuses  the
beams to form the probe volume.  The primary transmitting lens has
been cored, the core being replaced, and a light baffle tube used to
reduce  scattered radiation from its  surfaces.  This is particularly
important at this point due to its close proximity to the phototube
collecting optics and the phototube assembly.   A 28- and a 36-in.
primary focusing lens  cell were provided for the EPA system.
3.3.2 Collection Optics

    The collection optics in either the backscatter or the forward-
scatter mode consists of a  collection lens (in the backs catter mode
this is also the primary focusing lens) (Figs.  18 and 19).  In the
            Figure 19.  Overview of optics showing forwardscatter unit.
                                  36

-------
forwardscatter mode the lens is not cored.  A condensing lens doublet
is located between two mirrors used to fold the optical path for com-
pactness in the backs catter mode of operation.  The lens doublet
provides a very large aperture and a short focal length condenser.
Its purpose is  to take the light collimated by the collector lens and
to focus it onto the aperture of the phototube assembly.  Adjustments
can be made in either of the two mirror mounts to obtain imaging of
the probe volume into the aperture.   These  optics,  in conjunction with
the phototube assembly (photometer),  are located on a traverse
assembly to give scanning through the region of interest.

     The purpose of the photometer is to  provide a means of measuring,
imaging, viewing,  and filtering the scattered radiation from the probe
volume in a single unit (Fig. 20).  Spatial filtering is provided by

                                             5X Eyepiece
                                               Eyepiece Barrel
           Collimating Lens
                                                    5145 Line Filter
                                               Mounting Rod
                   Figure 20.  Photometer system schematic.

means of the adjustable entrance iris.  The iris can be envisioned as
being imaged by the collection optics into the probe volume and acts
as the limiting aperture.   A collimating lens  provides parallel light to
the line filter at normal incidence.  The 0. 5145-jum line filter is
0. 0030 /urn wide and and is critically dependent on angle.  A prism
can be inserted to deflect the light onto a mylar viewing screen.   The
eyepiece assembly also can be used to block unwanted radiation from
the photodetector.
                                  37

-------
    The photodetector is  an RCA model 931 A,  a rugged phototube
with an S-4 response; this peaks at 0. 4 jum.  The dynodes are connected
to a bridge divider consisting of 100-K^ thin film carbon resistors.
The first dynodes utilize two 100-v zenier diodes to maintain an
optimum voltage.  At 1,000 v the resistor divider will pass 1 ma.  A
430-ohm load resistor is  provided.


3.4 TRAVERSE ASSEMBLY

    The carriage assembly consists of two subassemblies.  The base
is used primarily as an extender to provide flexibility for use in the
various  applications (Fig. 21).  The base may  be separated from the
main assembly,  as wheels were provided for both.  Vertical adjust-
ment was achieved by means of four lead screws driven by sprocket
chain and worm gear drive.  Since the design requirements were for
essentially fixed vertical  positions, the assembly was not motorized.
To insure stability of the  main assembly, the four lead screws are
augmented by four angular slide brackets located on each side of the
table.   Horizontal motion of the table is controlled by another lead
screw which can be used to control overall motion by  up to ±3 in.
               Forwardscatter Receiving
               Optics Table (z)
                                      Overhead
                                      Cable
                                      Assembly
 Transmitting
 Optics Table (z)
                                            Table Traverse (x)
                                      -Vertical Lead Screws (y)

                                Extender Base (y)

                    Figure 21.  Traverse assembly schematic.
                                   38

-------
    The traverse of the probe volume is obtained by means of a
carriage and lens traverse assembly.  The traverse assembly consists
of a movable carriage, a position readout device,  and a small drive
motor.  The carriage may be moved manually or by means of a small
globe-type motor which is directly coupled to the lead screw by means
of an O-ring drive belt. The carriage  slides onto precisely machined
rods straight within one thousandth inch per foot.  Contact is maintained
by a three-point ball bearing roller assembly.  Directional control is
determined by two control levers located at the front  of the flame ana-
lyzer.

    The position readout consists of a  constant tension spring which is
attached to the traverse assembly by means of a cable.  The cable
makes multiple passes over  a  drum and drives  a 500-ohm potentiometer.
Offset voltage is  provided through a 100-ohm potentiometer adjustable
on the readout device.  Power is supplied from a regulated Hewlett-
Packard 200 power supply.
3.5  DOPPLER PROCESSOR SYSTEM

    The AEDC Model 6 Doppler Data Processor is a sampling device
uniquely suited to the analysis of discontinuous signal data.  These
"frequency first" data arise when the scatter particle density  is such
that only one particle at a time is traversing the LDV probe volume.
This signal type typically occurs when there is no artificial seeding in
the flow.

    Figure 22 illustrates the  Doppler processor in block diagram form.
The circuit shown is usable for capture of Doppler burst data  over the
frequency range from 1 to 500 kHz.  A method for extended coverage
to frequencies above 500 kHz  will be subsequently discussed.

    In Fig.  22 the burst data  (1)  are shown entering the multiband
analog filter  (2)  with simultaneous display on an oscilloscope (3).  The
filter (2) removes the pedestal (4) and attenuates noise outside the
filter pass band.  Nine bands,  covering 2-1/2  octaves per band, pro-
vide an instrument frequency  coverage from 1 kHz to 50 MHz.  An
oscilloscope-triggered gate is used to synchronize the  processor
sampling interval with the burst signal event displayed on the  scope.
The scope offers real-time examination of burst signal quality and
selective rejection of baseline noise.  The processor employs high-
speed, emitter-coupled logic,  integrated circuit flip-flops (5)  and (6)
to convert the first eight pulses of Doppler frequency data, after the
scope gate (4) opens,  into a time interval pulse, the D pulse (39).
                                 39

-------
40

-------
    Simultaneously, a second time interval pulse, called the  A pulse
(40), is generated by the A binary (8).  The time duration of the A
pulse is equal to the time period of the first four pulses in the burst
signal.  The inhibit binary (7) prevents further pulse activity  on the
part of the D and A binaries after the eight- and four-pulse sample
interval.

    The Doppler processor applies to the two time interval pulses the
criterion that if an ideally periodic burst signal were being  sampled
the time interval of the D pulse would be twice that of the A pulse.
The processor can make this time interval comparison to better than
0. 05 percent of the D pulse interval on those ranges  offering adequate
time resolution.

    The field-operated LDV can produce a variety of signals  in
addition to the desired Doppler frequency burst signal  with acceptable
signal-to-noise ratio.  Noise bursts,  short Doppler  signal bursts
caused by particle passage through too few fringe lines for accurate
transient time resolution,  and large amplitude pedestal signals  with
small Doppler frequency signal amplitudes are typical "problem"
signals to be avoided.  As an aid in the rejection of such undesirable
signal information and in placing limits on the deviations permitted in
the signal period averaging,  the processor is  equipped with a time
interval comparator.  The comparator functions to compare twice the
time interval of the A  pulse  (40) to the D pulse (39).  A preselected
percentage of the D pulse is  used as a limit detector against which the
double-A  pulse is compared. Only double-A  pulses that lie within
the accepted percentage of the D pulse time interval  are accepted  as
period data.  The D pulse interval percentage value forms a time
interval "data window. "  Signal quality versus signal quantity trade-
offs have arrived at data window values of 1-1/2 or 3 percent of the
D interval.

    Doppler signal input is accomplished by means of  two BNC  jacks,
one covering the burst signal range from 1 KHz to 20 MHz with  a
threshold sensitivity (minimum Doppler pulse amplitude acceptable) of
2 mv RMS and input impedance of 100 Kfi, input signal amplitude
limited to 1 v peak-to-peak.  The second jack covers the range  from
10 to 50 MHz with a threshold sensitivity of 3 mv RMS  and an input
impedance of 50 fi.  Two  50-fi BNC jacks permit monitoring the
sampled signal at the filter output.
                                 41

-------
    The BNC connector requires a positive edge (minimum 0- to 5-v
signal,  maximum 30-v signal) in synchronism with the initiation of an
oscilloscope sweep.  The scope is set to trigger on the leading edge of
signal bursts with minimum triggering on the signal source baseline
noise.  Scope bandwidth and gate trigger-rise time must be correlated
to the burst signal frequency to  be sampled.   For well characterized,
known areas of LDV application where routine data taking is being
performed, the scope may be replaced by a voltage comparator having
a suitable rise time.   The scope provides a valuable real-time display
of the signal quality (signal-to-noise ratio approximation), relative
signal amplitude to the processor minimum threshold signal require-
ments,  and the assurance that the processor is sampling frequency
bursts a large portion of the time, not random noise sources.  An
approximation of the burst signal frequency is available from the scope
display,  which is useful in the selection of the appropriate processor
frequency band.

    A solid-state LED produces a nominal 60-msec flash each time
the processor completes a sample interval.  The lamp functions
whether data transfer is effected during the sample or not.  The lamp
is a visual check that the two  necessary conditions for processor sam-
pling  have  been met,  namely: (1) that the burst signal amplitude is
above the minimum threshold value  and (2) that the proper +A gate
signal has  been furnished.

    A five-digit Nixie array (27) displays the period of the Doppler
frequency directly in units of  microseconds for the five lower frequency
ranges, covering from 1 to 500  KHz, and in nanoseconds for the upper
0.4- to 50-MHz  ranges, the pulse-stretcned range.  Data are available
in either BCD (26) or binary form (29) to an external data acquisition
system (42 and 30,  respectively).

    Print command generation  (41)  by the print command flip-flop  (45)
is under the control of the recycle logic (50).   The logic also generates
the strobe  to output the data from the data registers (26 and 29).  The
^max/^min c*rcuit $8) inhibits the recycle logic if the data exceed the
preselected frequency range.  The sample recycle rate is  controlled by
(32).  The  processor can operate independently of  a data acquisition
system by  routing the print command back as a hold-off pulse (46).
                                 42

-------
3.6  DATA MULTIPLEXER

    The purpose of the data multiplexer is to coordinate operation of
the various flame analyzer  data sources and arrange the data in a
format acceptable to a data acquisition system.  Coordination of the
data sources (two AEDC Model 6 Doppler Data Processors, a visibility
processor, and a traverse position readout) is accomplished by gating
the multiplexer with the scope +A gate and  allowing the multiplexer to
gate all data  sources simultaneously via its gate output.

    The data sources required for a given set of measurements are
selected by programming appropriate data  select switches on the front
panel of the multiplexer to an up position.   At least one data source is
required to initiate the  data validation and data transfer cycle.

    Following  the gating of all data sources,  the multiplexer waits an
interval of time,  nominally 900  msec,  for all units to complete a
measurement.  A logic network within the multiplexer then determines
if validated record  commands have been received from the required
data sources.  Validated data therefore were measured and processed
within the multiplexer settle time.

    If all required  sources have provided  record commands, the multi-
plexer proceeds to  output data from those  sources along with any other
data which may exist from nonrequired sources.  Nonrequired data
sources which  failed to obtain readings at  a given scope gating will
have zeroes  output for their readings (zero would never be a valid
reading from any required source).

    Traverse position is the only data source which will always have
a measurement ready to be output following each scope gating.  The
three other data sources make a determination of whether or not to
provide data  from a given burst waveform based on such parameters
as signal-to-noise ratio and particle acceleration.

    If any of the  required data sources selected on the multiplexer fails
to provide a  reading after 900 msec from a given gating, all data sources
are reset for another sample attempt.  This resetting is accomplished
by applying and terminating a hold-off pulse to each data source.  The
EPA flame analyzer system is shown in Fig.  23 exclusive of the
traversing forwardscatter unit previously  shown (Fig. 19).
                                 43

-------
4.0
                Figure 23. EPA flame analyzer system.

EVALUATION OF THE APPLICABILITY OF INTERFEROMETRIC
PARTICLE SIZING
     Including an interferometric particle-sizing capability with the
velocimeter system will offer the capability of measuring  not only the
velocity but also the size of the particle whose velocity is  being
measured.  The interferometric  particle-sizing technique has a distinct
advantage in that it is independent of the absolute magnitudes of scattered
light; furthermore,  it uses the same optical system as that of the
velocimeter (Ref. 20).  This affords the opportunity of making simul-
taneous velocity and particle size measurements.   A prototype electronic
                                  44

-------
signal processor has been designed and built and is currently being
calibrated in the laboratory.  With the successful development of the
technique, it -will be possible to provide rapid,  online data acquisition
of particle sizes.

     The operation of the technique can be briefly summarized by
referring to Fig.  3.   When a particle crosses a set of interference
fringes formed at the crossover point or focal volume of the two laser
beams, light is scattered by the particle and collected by the receiving
optics,  where  it is focused onto a photomultiplier tube.  The character-
istic waveforms evidenced by the  particle traversing the focal region
is sketched in  Fig. 24,  In this particular figure, Doppler bursts of
particles moving in the y-direction and through  the z = 0 plane are
depicted.  It can be seen from the figure that the waveforms consist of
a Gaussian modulated cosine component, called the "ac",  in addition to
a lower frequency Gaussian waveform, or "dc", called the "pedestal. "
                            a.  Visibility = 1.0    Time
                            	*' \S\-

                            .  Visibility = 0.47  I
             Time
                            c. Visibility = 0.17
                                               Time
                            Visibility •
"max-'min"2
(lmax+lmin"2
           Figure 24. Scattered intensity for three values of D/5, z = 0.
                                 45

-------
    The so-called visibility of the point on the waveform is defined as
the ratio of the magnitude of the ac component (one half the peak-to-
peak value) to the value of the dc pedestal at that point.  The value of
the visibility  is a function of the ratio of the particle size to the fringe
spacing.  The characteristic signal is also a function of the particle
location within the probe volume and the geometry of the collecting
optics. In Figure 25 the theoretical relationship between the visibility
and the ratio  of particle-to-fringe  spacing (d/<5) is plotted.   In this
particular figure the forwardscattered light is collected for a spherical
particle in either the y = 0 or z - 0 plane.  Figure 26 shows experimen-
tal and theoretical comparison for a single,  circular, 1. 55-cm beam
stop radius with a 5. 5-cm collection aperture radius.  Fraunhofer
diffraction theory was used to predict the on-axis visibility values.
       1.0
       0.9
       0.8
       0.7

    ^0.6

    1 °'5
    > 0.4
       0.3
       0.2
       0.1
         0
"SI
   \

                  \
          0  0.2    0.6     1.0     1.4    1.8    2.2     2.6     3.0
                                    D/6
    Figure 25.  Theoretical visibility versus ratio of particle diameter to fringe spacing.
    The following considerations must be taken into account in order
to apply this technique of particle sizing to a given environment.

    1.  In the one-dimensional case,  spherical particles are assumed
        in the event the actual particle shapes  are unknown.  In this
                                  46

-------
   manner,  it is possible to obtain a first order approximation to
   the size of the particle being measured.   By including a second
   visibility measurement using an orthogonal fringe set (such as
   is used for the two-component  velocimeter),  a check can be
   made on the accuracy of the particle size, including a confir-
   mation of whether or not the particles are indeed spherical.
   In the event the visibility differs for each of the two so-called
   visibility components,  nonspherical shape must be assumed.
                                     Fringe Spacing = 24.4 urn
                                     Distance from Scatterer to
                                       Collection Lens = 64 cm
                                     Collection Aperture Radius = 5.5 cm
                                     Beam Stop Radius • 1.55cm
                                           v
                                             D
  0     0.2    0.4     0.6    0.8    1.0    1.2    1.4     1.6    1.8    2.0
                                 D/6

 Figure 26. Theoretical and experimental comparison of visibility versus ratio of
          particle diameter to fringe spacing for a single circular beam stop.

I.  The minimum size which may be measured with this technique
   has yet to be determined.  One  limitation is set by the slope
   of the curve relating to visibility and the ratio of particle  size
   to fringe spacing.  For particle diameters smaller than approx-
   imately 2/10 of the fringe spacing,  the slope of the visibility
   curve is not sufficiently great to accurately relate the visibility
   to the particle  size (see Fig. 25).  In addition,  the minimum
   obtainable fringe spacing is limited by the maximum beam
   separation angle which can be obtained in a practical measuring
                               47

-------
    situation.  It is believed that the maximum angle is approxi-
    mately 12 deg,  giving a fringe spacing of approximately five
    times the  optical wavelength being used for the measurement.
    This limiting condition fixes the minimum measurable particle
    diameter at  approximately one wavelength of the illuminating
    radiation.

3.  Currently the visibility is measured when the particle is in
    either the y  = 0 or the z  = 0 plane. If the particles are all
    generally  moving in the same direction, this may be accom-
    plished electronically by orienting the optical system so that
    the particles pass through the y = 0 plane.  An alternative
    approach, where the particles are not all moving in the same
    direction, is to consider the possibility of aperturing the
    system so that measurements are made only for particles
    moving in or near the z = 0 plane.

4.  Background  light collected by the optical  system, including
    light scattered from the probe volume by small aerosols,
    which are always present, should be minimal (to avoid gener-
    ation of shot noise in the detector) and relatively constant.
    Determining visibility requires measuring the height of the
    pedestal with respect to the background light.  The present
    signal processor requires that the background be constant so
    that it may be measured and subtracted out.

5.  The method  assumes there is only one particle large enough
    to be measured in the probe volume at a time.  If the density
    of particles  in the system is high, creating a high probability
    of the occurrence of multiple particles in the probe volume,
    then  some method must be used either to reject those wave-
    forms caused by multiple particles or to investigate means
    for considering their influence on variations in the visibility
    from a single particle visibility function.

6.  The EPA flame analyzer was designed to be interfaced with a
    four digit, particle size visibility processor.  The data multi-
    plexer can be programmed for simultaneous particle  size and
    velocity measurements if desired.
                             48

-------
                 5.0 THIRD COMPONENT OF VELOCITY
    The purpose of this section is to outline conceptually the design
criteria and systems considerations required for the inclusion of a
third  component velocity measurement.  Several possible methods can
be used to make this third component measurement. The various
alternatives can take advantage of color separation, frequency separa-
tion,  and off -axis or backs catter systems.  Furthermore,  either refer-
ence beam or dual-scatter techniques can be employed (Ref.  21).  The
argon-ion laser supplied with the  EPA flame analyzer is capable of
multiwavelength operation. In addition, an intr a- cavity etalon has
been supplied with the laser system.   The purpose of this etalon is to
increase the temporal coherence of the laser to  allow for large mis-
matches in optical path.  This is of particular importance with the
reference beam systems.  The principal argon laser wavelengths are
0.4880 and 0. 5145
    The choice of a measurement system for the third component for
use in the rainbow furnace or similar environments is best filled by
means of a backs catter reference beam system.  Under more suitable
viewing conditions, for example, when a  large viewing port is available,
an off-axis, dual-scatter system could be used.  With the latter tech-
nique, distortions and tracking are primary considerations in their
design.  Figure 27 shows a single-beam system operating in the back-
scatter mode.  If directionality were required,  the Bragg cell would
necessarily be employed.  The argon laser is assumed to be running
at the 0.4880- and 0. 5145-jum wavelengths.  Two sets of beams will
therefore be generated and frequency shifted as previously described.
Color separation techniques utilizing dielectric interference beam-
splitters or careful positioning of a partially silvered mirror are used
to remove the reference beam for the 0.4880-Mm wavelength. A
frequency -shifted  beam with respect to the reference beam is allowed
to pass out through the optical system and to be imaged in the probe
volume by means of the output lens.  Light is scattered from the parti-
cles within the probe volume, as before.   The backs catter radiation is
related to the velocity vector and the propagation vectors by the follow-
ing equation:

                       f  = f0 + 1/2* (K8 -  K0) •  V                 (il)

where fQ is the fundamental optical frequency,  K^  and K^ are the^ wave
vectors of the scattered and  incident beams, respectively,  and V is thf
                                 49

-------
particle velocity vector.   For operation in the backscatter mode,  the
Doppler shift becomes

                               fd » K •  Vz                           (12)

where Vz is the component of velocity  injusec along the  optical axis
and K = 4 MHz /m/ sec.  This backs catter radiation is mixed with  the
reference beam to produce, in a square-law detector, the sum and
difference frequencies.  The measured current in the photodetector
becomes
                 i = a[l/2Ef0 +
  EloEg cos
Mo
- o>s)t
           1/2
(13)
where u,  /2?r and u  /2?r are reference and signal beam frequencies,
respectively, a is the proportionality constant, and E^0 and Es are the
corresponding electric field strengths.  Since frequency separation
techniques are being used, the local oscillator frequency differs from
the signal frequency under zero velocity conditions by an amount equal
to the Bragg cell frequency.   Furthermore,  the  difference in the color
allows the frequency separation techniques to be applied independently
of the two-component,  dual-scatter signals using 0. 5145-jum.radiation.
                     Dielectric
                     Beam
                     Splitter
Cored
Mirror
            Cored
            Output
            Lens
                Probe
                Volume-
           4
 Reference
 0.4880-um Beam-
   Collimating
   Lens
                	0.4880-pm Beam


             /- 0.4880-um Filter
    Bragg Cell
                               To Third Component
                             -*- Reference Beam
                               Electronics
                                                             To Two-Component
                                                             Frequency
                                                             Electronics
                                            -0.5145-um Filter

               're 27.  Proposed three-component velocimeter schematic.
                                    50

-------
    A detailed analysis of system parameters such as spot size, system
accuracy, bandwidth considerations due to the finite laser  spot size and
to the finite collection angle,  and the effects of signal strength and
signal-to-noise ratios has been reported (Ref. 20).  One can generally
conclude that the system accuracy is reduced with smaller spot sizes
and larger receiver apertures because of the frequency spreading of
the signal and the inability to accurately determine the line center.
Also, a reduction in coherency between signal and the local oscillator
beams can result in a requirement for larger reference or local
oscillator beams to give the same signal-to-noise ratio.  The minimum
conversion gain permitted is limited by the allowable continuous wave
(CW) or pulse current at the anode of the photomultiplier tube.  Finally,
the scattered particle density required for a given signal power varies
inversely with the incident beam power density and the probe volume.
A typical minimum particle density of 10^ particles/cc is calculated
for continuous signals, which are required for most reference beam
processors.  However, through the use of a burst-type counter this
requirement can be reduced to measuring single-particle events,
making the third-component particle density requirements  consistent
with the other velocity component counters and the particle-sizing
electronics.
                   6.0 SUMMARY AND CONCLUSIONS
    Theoretical analysis and confirming experimental data at the EPA
"rainbow" furnace verify that the EPA flame analyzer velocimeter
design under consideration was efficient for the application intended.
Additional effort is required to include a third component so that three-
dimensional turbulent flow characteristics may be measured.   The
viewport space limitations prevent the application of the off-axis, dual-
beam concept.  This  essentially  restricts the technique to the develop-
ment  of a reference beam, two-color  system.  Preliminary laboratory
experiments have shown feasibility, and no major development difficul-
ties are foreseen.

    A prototype particle  size/visibility  processor has been developed
and is currently being evaluated  and calibrated in the laboratory.
Preliminary results verify the application of the technique.  Additional
experiments are being planned in an operational wind tunnel to acquire
the field experience necessary for the development of a second-
generation system.  A two-component visibility processor will result
in further definition of the shapes of the particles being measured.
                                 51

-------
                             REFERENCES
 1.  Lennert, A. E. ,  Brayton,  D. B. , Crosswy, F. L. , et al.
          "Summary Report of the Development of a Laser Velocim-
          eter to be Used in AEDC Wind Tunnels. "  AEDC TR-70-
          101 (AD871321), July 1970.

 2.  Brayton, D. B. and Goethert, W.  H.  "A New Dual-Scatter
          Doppler-Shift Velocity Measuring Technique. "  ISA
          Transactions, Volume 10, No.  1,  1971, pp. 40-50.

 3.  Rudd,  M. J.   "A New Theoretical Model for the Laser
          Dopplermeter. " Journal of Scientific Instruments
          (Journal of Physics E), Series 2, Vol. 2, 1969,  pp.  55-58.

 4.  Penney,  C.  M.  "Differential Doppler Velocity Measurements. "
          IEEE Journal of Quantum Electronics, Vol. QE5, No.  6,
          pp. 318-319.

 5.  Durst, F. and Whitelaw, J.  H.  "Optimization of Optical
          Anemometers. " Proceedings of the Royal Society of
          London, Series A, Vol. 324, 1971, pp. 157-181.

 6.  Mayo,  W. T. , Jr.  "Simplified Laser Doppler Velocimeter
          Optics." Journal of  Physics E: Scientific Instruments,
          Vol. 3,  March 1970, pp. 235-237.

 7.  Durst, F.  "Scattering Phenomena and Their Application in
          Optical Anemometry. "  Journal of Applied Mathematics
          and Physics (ZAMP), Vol.  24,  1973,  pp. 619-643.

 8.  Farmer, W.  M.  "The Interferometric Observation of Dynamic
          Particle Size, Velocity and Number Density. " PhD
          Dissertation, University of Tennessee, Knoxville, 1973.

 9.  Farmer, W.  M.  "Measurement of Particle  Size, Number
          Density, and Velocity Using a Laser Interferometer. "
          Applied Optics, Vol. 11, November 1972, pp. 2603-2612.

10.  Roberds, D.  W.  "Electronic Instrumentation for Interferometric
           Particle Sizing. "  PhD Dissertation,  University of
           Tennessee, Knoxville,  1975.
                                 52

-------
11.  Fried,  D.  L.   "Optical Heterodyne Detection of an Atmospheri-
          cally Distorted Signal Wave Front. "  Proceedings of the
          IEEE, Vol. 55,  No. 1,  January 1967,  pp. 57-67.

12.  Hodara, H.  "Laser Wave Propagation Through the Atmosphere. "
          Proceedings of the IEEE, Vol. 54,  No. 3, March 1966,
          pp. 368-375.

13.  Treacy, E. B.  "An Analysis of Some Factors Affecting the
          Accuracy of Laser Doppler Velocimetry. "  Instrumen-
          tation in the Aerospace Industry, Vol. 17, 1971,
          pp. 165-173.

14.  Goethert, W.  H.  "Laser Doppler Velocimeter Dual-Scatter
          Probe Volume." AEDC-TR-71-85  (AD727005), July 1971.

15.  Kogelnik, H.  "Imaging of Optical Modes  - Resonators  with
          Internal Lenses. "  Bell System Technical Journal,
          Vol. 44, March 1965,  pp.  455-494.

16.  Durst,  F. and Stevenson, W.  H.   "The Influence of Gaussian
          Beam Effects on the Spectral Nature of Laser Doppler
          Signals. "  Proceedings of the Minnesota Symposium on
          Laser Anemometry, October 22-24, '1975.

17.  Lennert, A. E.,  Crosswy, F.  L. , and Kalb, H. T.  "Applica-
          tion of the  Laser Velocimeter for Trailing Vortex
          Measurements." AEDC-TR-74-26  (ADA002151),
          December  1974.

18.  Farmer, W. M.  and Hornkohl, J. O.  "Two-Component, Self
          Aligning Laser Vector Velocimeter.17  Applied Optics,
          Vol. 12, No. 11, November 1973,  pp.  2636-2640.

19.  Chu,  W. P. and Mauldin, L. E.  "Bragg Diffraction of  Light by
          Two Orthogonal Ultrasonic Waves in Water. "  Applied
          Physics Letters, Vol.  22,  No. 11,  June 1973, pp. 557-559.

20.  Roberds,  D. W. ,  Farmer, W. M. ,  and Lennert,  A. E.
          "Interferometric Instrumentation for Particle Size Analysis. "
          AEDC-TR-74-82 (ADA006136), February 1975.

21.  Davis,  Donald T.  "Analysis of a Laser Doppler Velocimeter. "
          ISA Transactions, Vol. 7,  No. 1, 1968,  pp. 43-51.
                                53

-------
                           NOMENCLATURE



2b        Input beam diameter, Fig. 3


2bQ       Characteristic dimension of probe volume


C         Speed of sound in diffracting medium


D         Beam separation at output lens


d^, dQ    Input and output  beam diameters, respectively,  Eq. (10)


Fp,  F£   Primary and collimator focal lengths, respectively


F.,  FO    Input and output  focal lengths, respectively,  Eq. (10)


FL       Focal length of lens, Eq.  (4)


f.         Driver frequency


fm        Modulation frequency of ultrasonic wave


fQ        Fundamental optical frequency

--*-  —*-
Ks, Ko   Wave vectors  of scattered and incident beams, respectively


LED      Light-emitting diode


M        Magnification


N         Order of diffraction


V         Velocity of sound in water


Vz        Component of  velocity along optical axis, jusec


a         Angle between two crossed Gaussian beams


j3         Angle between crossed beams in Gaussian mode


6         Fringe spacing in probe volume
                                 54

-------
A         Wavelength of sound in diffracting medium

X         Wavelength of light

0         Angle from optical axis of first diffraction minimum,
          Eq.  (6)

Uj0/27r    Reference beam frequency

wg/27T     Reference signal frequency
                                 55

-------