United Stats*
Environmental Protection
Agency
Environmental Sciences Research
Laboratory
Research Triangle Park NC 27711
EPA-600 2-79-192
October 1979
Research and Development
Cross-Stack  Optical
Convolution
Velocimeter

Development and
Evaluation of a
Breadboard Design

-------
                 RESEARCH REPORTING SERIES


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

    1. Environmental Health Effects Research

    2. Environmental Protection Technology

    3. Ecological Research

    4. Environmental Monitoring

    5. Socioeconomic Environmental Studies

    6. Scientific  and Technical Assessment Reports (STAR)

    7. Interagency Energy-Environment Research and Development

    8. "Special"  Reports

    9. Miscellaneous Reports

This report has been assigned to the  ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental 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.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                                                 EPA-600/2-79-192
                                                 October 1979
    CROSS-STACK OPTICAL CONVOLUTION VELOCIMETER
 Development and Evaluation of a Breadboard Design
                    M.  J.  Rudd

           Bolt Beranek and Newman,  Inc.
          Cambridge, Massachusetts 02138
                  Project Officer

                   John S. Nader
Emissions Measurement and Characterization Division
    Environmental Sciences Research Laboratory
        Research Triangle Park, N.C. 27711
    ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
        OFFICE OF RESEARCH AND DEVELOPMENT
       U.S.  ENVIRONMENTAL PROTECTION AGENCY
        RESEARCH TRIANGLE PARK, N.C. 27711

-------
                                 DISCLAIMER
     This report has been reviewed by the Environmental Sciences Research
Laboratory, U.S. Environmental Protection Agency and approved for publication.
Approval does not signify that the contents necessarily reflect the views and
policies of the U.S. Environmental Protection Agency, nor does mention of
trade names or commercial products constitute endorsement or recommendations
for use.
                                      ii

-------
                                  ABSTRACT
     A new type of instrument has been designed and evaluated for the meas-
urement of a line average of a stack gas velocity.  The light output from a
lamp is collimated and projected across the stack.  A shadowgraph image of
the turbulence in the stack is produced on the far side and this image is
convected by the stack gas flow.  A grating is placed over the image and the
light transmitted falls on a photodetector.  The frequency output from the
photodetector is the rate at which the shadowgraph image crosses the grating.

     A breadboard design of this cross stack optical convolution velocimeter
(OCV), as it is called, has been built.  It was tested in the EPA Stationary
Source Simulation Facility (SSSF) over a wide range of environmental
conditions.  Agreement was obtained with a corrected pitot tube with a root
mean square error of 1.3%.

     Position sensitivity of the OCV has been extensively studied and a con-
figuration has been found which is completely insensitive to position.

     The cross stack OCV has demonstrated itself as an effective, accurate
velocity monitoring instrument which is simple to build and operate.

     This report was submitted in fulfillment of Contract 68-02-2786 by Bolt
Beranek and Newman, Inc., Contractor, under the sponsorship of the U.S.
Environmental Protection Agency.  This report covers a period from October
1977 to March 1978, and work was completed as of March 1978.
                                     iii

-------
                                  CONTENTS


ABSTRACT	iii

FIGURES	   vi

TABLES	vii

ACKNOWLEDGMENT  	 viii

     1.   INTRODUCTION  	    1

     2.   CONCLUSIONS   	    2

     3.   RECOMMENDATIONS   	    3

     4.   SENSITIVITY OF THE OPTICAL CONVOLUTION VELOCIMETER  	    4

     5.   POSITION SENSITIVITY OF OPTICAL CONVOLUTION VEEOCIMETER   ...    7

     6.   TESTS OF THE  CROSS-STACK OCV IN THE STACK SIMULATOR	   14

APPENDICES

     A.   CALIBRATION OF POIN T OCV AGAINST A LASER DOPPLER
            VELOCIMETER	   23

     B.   DRAWINGS OF PROPOSED DESIGN FOR A CROSS-STACK OPTICAL
            CONVOLUTION VELOCIMETER  	   28

-------
                                  FIGURES
Number
l
2
3
4
5
6
7
8
9
10
Range of Sens
Retro-Reflect
Position Sens
Velocity Prof
Schematic of
Comparison of
Comparison
Comparison
(SMS Error
Comparison
Comparison
Products (I
of
of

of
of
IMS
itivi
ing 0
itivi
ile i
Cross
'ocv
OCV
OCV
1.8%)
OCV
ty c
CV ,
ty :
n S1
Sti
and
and
and
and
OCV and
Error =
jf OCV
Ln a 2
tack -. ,
ick OC\
Pitot
Pitot
Pitot
Pitot
Pitot
1.4%)


m Stack for Three G
j





Tube at 93C (RMS Error - 1.0%) 	
Tube at 150C (RMS
Tube at 150C and 5
Tube at 200C (RMS
Error = 0.9%) ...
kg/hr of Dust
Error = 1.6%) ...
Tube in the Presence of Combustion
Page
5
8
10
12
15
17
18
19
20
21
11     Comparison of OCV and Pitot Tube With a  Dust Loading of 2.5 kg/
       hr (RMS Error = 0.9%) 	   22
                                    vi

-------
                                   TABLES
Number                                                                  Page




  1     Errors with Nonoptimal Gratings 	 ...   9




  2     Weighting Functions for Cross-Stack OCV	"...  13




  3     Error Compared with Line Average  	  13




  4     Test Conditions for Cross Stack OCV	16
                                    vii

-------
                                ACKNOWLEDGMENT


     The author wishes to gratefully acknowledge the help and suggestions
from G. Dubro and D. Rim of the Air Force Flight Dynamics Laboratory of the
Wright Aeronautical Laboratories since they were the original inventors of
the Optical Convolution Velocimeter.  The author also wishes to acknowledge
the aid of J. Nader and T. Ward of the Stationary Source Emissions Research
Branch, EPA, for their support during the tests in the Stationary Source
Simulation Facility.
                                    viii

-------
                                   SECTION 1

                                 INTRODUCTION


THE OPTICAL CONVOLUTION VELOCIMETER (OCV)

     The optical convolution velocimeter was conceived by DuBro and Kim* (U.S.
Patent No. 3,953,126) as a noninvasive method for measuring aircraft speed.
Its principle has been extended here for stack flow measurement.

     The output of a light source is collimated by a lens, and projected
through the turbulence onto a grating.  The turbulence is generated naturally
in the stack.  As the light passes through the turbulence, it is refracted,
and a "shadowgraph" pattern of bright and dark bands is formed on the grating.
As the turbulence is convected with the mean flow, the shadowgraph pattern is
convected over the grating.  We can describe the light transmitted by the
grating as

                        / l(x-y)G(x)dx = F(y)

where I(x-y) is the shadowgraph pattern that is convected in time by distance
y, and G(x) is the grating transfer function.  The function F(y) is the con-
volution of the shadowgraph and the grating.  By Parseval's theorem, the spec-
trum of this convolution is equivalent to the product of the spectra of I(x)
and G(x).  If the spectrum of G(x) is narrow, the spectrum of the convolution
function F(y) is narrow, and it will be sinusoidal with a frequency equal to
that at which the turbulence crosses the grating.  Hence, the velocity can be
found by measuring this frequency.

     The purpose of this report is to describe the development of the OCV for
making in-stack velocity measurements.  The OCV has a number of inherent
advances over the pitot tubes which are currently employed.  First, it is an
absolute instrument and never needs recalibration once it has been set up.
Secondly, it is unaffected by ambient conditions such as pressure and tempera-
ture.  Thirdly, it just measures one component of the velocity.  Fourthly, it
can be given a digital readout very inexpensively:  The OCV promises to be a
much more accurate and convenient to use instrument than the pitot tube.
*D. Kim and G. DuBro, 1974, "The Optical Convolution Velocimeter" presented at
 the second Project Squid Workshop, Purdue University, Lafayette, IN, Mar. 26-27,

-------
                                   SECTION 2

                                  CONCLUSIONS

     A breadboard version of a cross stack OCV has been successfully  designed,
 fabricated and tested in the EPA  Stationary  Source Simulation Facility.  Good
 agreement was obtained with a pitot tube  over a wide range of environmental
 conditions; temperatures up to 204C and  dust loadings up to 0.25 gm/in.3;
 the  root mean square discrepancy  was 1.3% of reading.

     A configuration for the cross stack OCV has been devised which should
give perfectly uniform sensitivity across the stack.   The light beam is retro-
reflected across the stack to a grating of pitch 1.9 (stack width)^ milli-
meters.  Drawings for the suggested cross stack OCV design are attached in
Appendix B.

     The cross stack OCV has shown itself to be a simple and accurate instru-
ment for the measurement of the line average velocity in a stack.

-------
                                   SECTION 3

                                RECOMMENDATIONS

     Now that the basic performance of the cross stack OCV has been demon-
strated, it is ready for the next stage in its development.  This will con-
sist of:

     1.  Fabricating a properly packaged version which can be installed in the
         field.

     2.  Incorporating the double pass feature to give uniform sensitivity.

     3.  Incorporating air curtains over the lenses.

     4.  Investigating the sources of electronic noise and improving the
         signal-to-nbise ratio.

     5.  -Investigate the possibility of using a low power Helium-Neon Laser as
         a light source instead of a tungsten lamp.  This has a much smaller
         source size which may well eliminate much of the low frequency noise
         due to the flow.  Further, the Laser has a very low power consumption
         CIS watts) and, therefore, does not require convection cooling.
         Finally, gas Lasers have a very long life, many thousands of hours.

Such an improved instrument would be suitable for field testing.

-------
                                   SECTION 4

               SENSITIVITY OF  THE OPTICAL CONVOLUTION VELOCIMETER
      The optical convolution velocimeter measures the convection velocity of
 a shadowgraph pattern across a one-dimensional grating.   From Townsend (1965)
 the the spectral intensity of the shadowgraph image due to a thin sheet of
 turbulence and transmitted by the grating is

                101,0) = 8TrN2sin2(2z/2N)FOl,0,0,)Sz-"                     (1)

 where         = wavenumber of the grating
              N = wavenumber of light employed
              z = thickness of turbulence
       F(,m,n)  three-dimensional spectral intensity of turbulence
                  refractive index variations
             Sz = thickness of turbulence.

 It has also been shown by Batchelor (1959) that for fully developed flow of
 large Reynolds number, the theory of local isotropy gives


                     F(Jl,0,0) = 0.lEN ~^ "^ -                            (2)

                                              k -I
 where    mean scale of fluctuations  0.2e v
        e = rate of energy dissipation per unit mass by viscosity
        V = kinematic viscosity
        = rate of destruction of mean-square refractive index fluctuations
            by molecular diffusion.

 Substituting Eq.  2 into Eq.  1,  we obtain


           101,0)  - 0.8TTN2eNe~^ sin2(2z/2N)jT^ Sz  .                     (3)


      In  our  arrangement,  i,  the wavenumber of the grating,  is held constant,
and z, the position in the stack,  varies.  Therefore, if  and e are constant
                                                           pj
across the stack,
A.A. Townsend  (1965), "The Interpretation of  Stellar Shadow-Bands as a Conse-
     quence of Turbulent Mixing,"  Quart.  J. Roy, Met.  Soo.,  91,  pp.  1-9.
G.K. Batchelor (1959), "Small-Scale Variation of Convected Quantities Like
     Temperature in Turbulent Fluid,  Part 1,  J. Fluid Mech*,  5,  pp.  113-133.

-------
                              )  sin2(2z/2N)  .

This dependence of I(,0) on 2 is shown in Figure 1.   For small z,
                                                                           (4)
                                       z2 .
For z approximately between 7TN/3&2 and 2irN/3
                                 ,0   z
and for z around irN/
                              I(,0) = independent of z .
This relationship will apply if a heater wire across the stack is used to
generate the signal.  In this case, e and e,, are independent of position
across the stack.

     In typical turbulent processes, Batchelor (1953) gives the following
values for e and e ,
UJ
en
                                                                          27T
                               RANGE* Z/2/N
                    Figure  1.   Range of sensitivity of OCV
G.K. Batchelor  (1953), "Homogeneous Turbulence," Cambridge University Press,
     London.

-------
                                       L~l

and

                          N = I ("2)^ L^/dz>2

where     u2  =  root mean square  fluctuating velocity
      	  L  =  integral  scale of the  turbulent motion     	
and   dy/dz  =  mean refractive index  gradient = p/T  dy/dp(dT/dz)
where 	  p  =  density of flow,  T = absolute temperature
and   dT/dz  =  mean temperature  gradient.
 Therefore,
                          V* - 1 1* f $ fll/M                          (5)
which is only a function of the  temperature gradient and  scale  of  turbulent
motion for a particular  fluid at a particular temperature.  Typically,  the
temperature gradient  is  greatest at the wall of  a  stack and the mixing  length
is greatest in the center.  Generally, for fully developed turbulent  flow in
a stack, the Prandtl  hypothesis  (Goldstein, 1965)  is that L ffi z1,  the distance
from the wall.   Similarly by the Reynolds analogy  (Goldstein, 1965) between
temperature gradient  and velocity gradient dT/dz <* 1/z',  the reciprocal of the
distance from the wall.   These relations apply to  the bulk of the  turbulent
flow in the center of the stack, but not to the  region  very close  to  the wall,
the so-called viscous sublayer.  Thus, over most of the region  of  the stack,

                          L dT/dz = constant .                              (6).
               \
Hence,  e e <* L3,  which is a very weak relationship to mixing length.  There-

fore,  over most of the flow in a stack where the turbulence is  fully  developed,
the OCV has a very weak  dependence on the structure of  the turbulence.   How-
ever,  it is not clear that this  applies to all flows such as highly skewed
velocity distributions after a bend.

      The most important  parameter affecting the  sensitivity of  the OCV  is the
distance between  the  turbulence  and grating.  If we are to place the  grating
one  stack diameter beyond the grating and choose the grating spacing  accord-
ingly,  (2/z/N), then  there will  be less than  a  25% variation in sensitivity
across  the whole  stack or 10% variation over the central
S. Goldstein (ed.)  (1965), Modern Developments -In Fluid Dynamics, Dover
     Publications (New York), p. 208.
S. Goldstein (ed.)  (1965), Ibid, p. 649.

-------
                                   SECTION 5

            POSITION SENSITIVITY OF OPTICAL CONVOLUTION VELOCIMETER
     In the previous section the signal strength I() of the OCV with a grat-
ing of wavenumber & was given by

               I() = O.STTN^e"^ sin2(2z/2N)5,~^ 6z                        (7)
                             N
where  N = wavenumber of radiation
       e = rate of energy dissipation per unit mass by viscosity
      ew = rate of destruction of mean-square refractive index fluctua-
           tions by molecular diffusion
       z = distance of turbulence of thickness 6z from grating.

Previously, a single pass OCV system was considered with the transmitter and
receiver on opposite sides of the stack.  It was shown that if the receiver
was placed one stack diameter beyond the stack, then the sensitivity of the
OCV did not vary more than 25% across the stack.  However, placing the re-
ceiver so far beyond the stack is rather inconvenient.  Accordingly, another
scheme is considered here whereby the light is reflected back across the
stack, by means of a retro-reflector and the transmitter and receiver are side
by side (see Figure 2).  We now have one signal from the outward beam and an-
other signal from the return beam and we can add the two together.  Thus, we
obtain

          I (A) = A{sin2(2z/2N) -1- sin2[2(2z   -z)/2N]}6z                   (8)
                                            H13.X

where    A = 0.8irN2e.Te^S,~^
                    N
      z    = OCV to retro-reflector distance.
       max
Of these two terms, the second represents the outward beam and the first the
return beam,  z is always less than z
                       J             max
     Now, if we choose the grating wavenumber  such that

                              2&2z   /2N = it/2 .                            (9)
                                  max

Then the above expression reduces to

                    = A[sin2(2z/2N) + cos2(A2z/2N)]6z                     (10)

                    = A5z                                                  (11)

-------
 LAMP
               BEAMSPLITTER
           DETECTOR
STACK
                                                            RETRO-
                                                            REFLECTOR
     SENSITIVITY
     IU)
                         STACK POSITION (Z)

                             ZmoxtrN/2l2
          max
                    Figure 2.   Retro-reflecting OCV

which is independent of z!   The contributions  from the outward and return
beams are shown in Figure 2.

     We are operating a light wavelength X,  and the grating pitch, p, is given
by 3 as
                               2/Az
                                  max

-------
Thus, for a wavelength of 0.9 ym and a stack width of 2 m

                                   p = 2.7 ram  .

A grating of this pitch will then give a  uniform weighting to the velocity
across the stack, independent of velocity profile.  This analysis does assume,
however, that the turbulence intensity and scale are uniform across the stack.

SENSITIVITY ANALYSIS

     It is important to know not only the performance of the optimum grating
described above but also how coarser gratings may perform.  This is because,
in order to maximize signal strength from the OCV, it may be necessary to use
a less fine grating than 2.7 mm.  Accordingly the velocity weighting function,
I(&)/A, has been computed for two other grating pitches, 4.9 mm and 7.6 mm
when used in a 2 m stack.  The results are shown in Figure 3.

     These gratings tend to weight the flow next to the transmitter/receiver
more heavily than that next to the retro-reflector.  In the case of the 7.6 mm
grating, the velocity next to the transmitter/receiver is given a 40% higher
weighting than the average.

     However, because we are averaging across the whole stack, the error on
the  average velocity measured will be much less than this.  We have computed
this error for a highly skewed velocity profile with a 1:1% variation.  The
flow was 80% of the average at one wall and 120% of the average at the other
wall.  Table 1 shows the errors for the average velocity measured with the
three  different gratings for the two cases when the maximum velocity is next
to the transmitter/receiver and when it is next to the retro-reflector.  It
will be seen that in no case is the error greater than 3%.  Therefore, the OCV
is not very sensitive to nonoptimum gratings or nonuniform velocity profile.

                  TABLE 1.  ERRORS WITH NONOPTIMAL GRATINGS.
Velocity Profile = Linear  0.8 to 1.2 x average across stack
Maximum Velocity
2.7 mm grating
4.9 mm grating
7.6 mm grating
At Transmitter/Receiver
.0
.03
.026
At Retro-Reflector
.0
-.0204
.0043
 MIRROR REFLECTIVITY

      In the  foregoing analysis, we have assumed a mirror reflectivity of 100%.
 In  practice  a value  of  80  90% is more typical.  This will change Eq, 8 and
 mean that perfect  independence of z  in Eq.  10 is no longer possible.  The
 error is then proportional  to
                              0.15  sin2a2z/2N)
(12)

-------
    1.50
I-
o
yj
U
o
itl
    1.00
    0.75
    0.501-
    0.23 -
                                                        2.7mm
** -,_   4.9 mm
                                                                 GRATING
                  _L
       0         0.2
        TRANSMITTER/
        RECEIVER
                1.0         1.2
                 RETRO/
                 REFLECTOR
                            O.4        0.6        0.8

                                 STACK POSITION

      Figure 3.  Position sensitivity in a 2 m stack for three gratings

which has a maximum error of 7.5%  at z = z   .  However, in the light of the
                                        max
limited sensitivity of the OCV to  grating spacing, this factor will have a
much smaller effect on the overall velocity reading.

     For a perfectly uniform velocity weighting across the stack,  a grating
pitch p should be chosen such that
                                        max
                                    10

-------
where    X = wavelength of  light  employed (approx.  0.9  pm)
      z    = transmitter  -  retro-reflector distance.

However, gratings which are 2  to  3  times  this  pitch do  not  show a large error,
even with_ quite a nonuniform velocity  profile.

CALCULATED PERFORMANCE OF OCV  IN  AN ACTUAL STACK VELOCITY PROFILE

     The accuracy of  the  cross stack OCV  has been computed  for  a real  stack
velocity profile.   This profile was taken from the  Flow and Gas Sampling
Manual by E.F. Brooks and R.L.  Williams,  U.S.  Environmental Protection Agency
Report EPA-600/2-76-203 of  July 1976.   The profile  was  taken  from Figure 6,
page 31, and was measured in a 8.27 x  3.18 meter duct.  It  was  supposed that
the OCV would  look  across the  duct  from the center  of its sides,  looking
either parallel to  the long sides or the  short  sides.   The  velocity profiles
along these lines of  sight  are shown in Figure  4.

     If we take a linear  average  of the velocity on a line  across the  stack,
we obtain a value of  101.4% of the  mean on a line parallel  to the long  side
and only 77.7% of the mean  parallel to the short side.  This discrepancy high-
lights the main source of error,  that  is  not with the OCV itself  but with the
dangers of taking a line  average.   Since  the line only  samples  part of  the
stack, serious errors can develop if a large part of the flow deviates  sub-
stantially from the average.  Accordingly line  averages should  only be  taken
where the flow is reasonably uniform.

     The performance  of different grating sizes have been compared for  the
profile parallel to the long side (errors associated with the profile parallel
to the  short side are not considered to be meaningful).  The weighting  func-
tions for three grating pitches are given in Table  2.   The  errors for  the three
gratings are compared in  Table  3 with  that  of a  linear  average made parallel
to the  long  side of the duct.

     It will be noted that  the errors  are not  very  large, even  for the  most
coarse grating and  are smaller  than the difference between  a line and true
average.

SITING  OF THE  CROSS-STACK OCV

     The main  source  of error  which is likely  to arise  with the cross-stack
OCV  in  actual  use will arise from its  siting,  rather than any internal  instru-
ment error.  The errors associated  with the instrument  are  of the order of  1%
or  less, but the difference between a  line average  and  a true average may be
20 - 30%.  The flow in the  duct where  the cross-stack OCV is mounted should
be reasonably  uniform.  This means  that it should not be mounted downstream
of a sharp bend or  blockage, or even too  close upstream.  Whenever the cross-
stack OCV is mounted  on a stack,  a  complete velocity survey should be  performed
before  installation and the discrepancy between the line average and true
average computed.   This will at least  make the operator aware of any discrep-
ancies which may arise.
                                      11

-------
     ISO
     140
     120
 2  100
 u_
 O
 O
 3
     60
     40
     20
LONG SIDE

SHORT SIDE
                 I
                 I
I
                20       40       60       80

                 POSITION(% OF DUCT WIDTH)
                                  too
Figure 4.  Velocity profile in stack (Brooks and Williams, 1976)
                           12

-------
TABLE 2.  WEIGHTING FUNCTIONS FOR CROSS-STACK OCV
X = Mean wavelength of radiation employed (0.9 pm),
d = Distance between transmitter and reflector lenses
Position Across Stack
(Distance From Lamp)
0
.05
.10
.15
.20
.25
.30
.35
.40
.45
.50
.55
.60
.65
.70
.75
.80
.85
.90
.95
1.00
Grating Pitch
2Ad
i.oo
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
4Acf
1.364
1.312
1.260
1.213
1.167
1.125
1.083
1.045
1.008
.976
.945
.919
.893
.872
.852
.837
.823
.813
.804
.801
.799
6A3"
1.430
1.368
1.307
1.250
1.194
1.143
1.093
1.050
1.007
.969
.932
.903
.875
.851
.826
.780
.734
.755
.774
.771
.767
    TABLE 3.   ERROR COMPARED WITH LINE AVERAGE
Grating Pitch
5.5 mm
11.0 mm
16.5 mm

0.0%
-0.3%
-1.1%
                        13

-------
                                   SECTION 6

              TESTS OF THE CROSS-STACK OCV IN THE STACK SIMULATOR
 DESCRIPTION OF CROSS-STACK OCV

     The arrangement of the cross-stack OCV is shown in Figure 5.  The light
 source was a  100 watt Quartz-Halogen Lamp with integral dichroic reflector.
 The  lamp was  driven from a 12-Volt regulated supply.  The light output was
 focused by the reflector into a 1 cm diameter spot.  An aperture (initially a
 280  micron slit and later a 2 mm pinhole) was placed at this focus.  A frosted
 glass plate was placed over the slit in order to achieve uniform illumination.
 The  transmitted light was collimated by a 10 cm diameter, 30 cm focal length
 lens.  The collimated light then passes across the stack to a receiver.  This
 receiver consists of a 10 cm diameter grating and  10 cm diameter by 16 cm
 focal length  condenser lens.  This focused the light onto a 1 cm diameter
 silicon photo-diode which generated the signal.  The signal passed through a
 preamplifier  to the correlation discriminator signal processor and an EMR
 real-time spectrum analyzer.

 PRELIMINARY TESTS

     The cross-stack OCV was set up in the Stationary Source Simulation Facil-
 ity  at EPA Research Triangle Park.  The first aperture used for the source was
 a slit 12 mm  long by 0.28 mm wide.  Plenty of signal amplitude was obtained
 but  it had a  very noisy character.  In fact, the signal-to-noise ratio im-
 proved when the slit was misaligned by about 10 with the grating.  Accord-
 ingly a 2 mm  pinhole was substituted and found to give much better results.
 Three gratings were employed with 2.5, 5, and 7.5 mm pitch.  The 5 mm worked
 somewhat better than the 7.5 mm grating.  The 2.5 mm grating worked almost as
well when a 1 mm pinhole was substituted for the 2 mm pinhole.  No large
 changes where noted with grating pitch, although signal frequency did vary
 directly.  The 5 mm grating was used for all subsequent tests.

 EVALUATION TESTS

     The stack OCV was operated under the seven test conditions detailed in
 Table 4 and its performance compared with a pitot tube.  Readings were taken
over a speed  range of 6 - 22 m/sec.  No data could be obtained in clean air at
room temperature since the OCV did not generate any signal.  However, a signal
was obtained  from dusty air at room temperature.  No reading could be obtained
either from Run 2 (10% by volume of water) since condensation occurred on  the
surface of the cooler lenses.  Data from the six successful runs are shown in
Figures 6  11.


                                     14

-------
          10cm DIA. x 30cm F.L.
                 LENS
2mm PINHOLE
 100W QUARTZ
 HALOGEN LAMP
 STACK
SIMULATOR
                            PORT
                                     FLOW
              PORT
10cm DIA. x 16cm FL.
     iLENS
                                                                        PHOTO-
                                                                        DIODE
                         'GRATING
                      Figure 5.  Schematic of cross stack OCV

-------
                TABLE 4.  TEST CONDITIONS FOR CROSS STACK OCV
Run No.
1
2
3
4
5
6
7
Temperature
93C
120C
150C
150C
204C
120C
27C
% of Water
0
10
0
0
0
6
Ambient
Dust Loading
0
0
0
5 kg/hr
0
0
2.5 kg/hr
     For the signal analysis both the correlation discriminator and a spectrum
analyzer were used.  However, due to the characteristics of the signal (noise
spectrum was not white) sigificant discrepancies developed between the two
processors.  Accordingly the spectrum analyzer data has been used in these
figures.  No substantial or systematic discrepancies have been found between
the OCV and the pitot tube.  The root mean square error for all data points is
1.3% of reading.

     Good agreement has been found between the cross-stack OCV and a pitot
tube over a wide range of flow and environmental conditions.  However some
"cleaning up" of the OCV signal is required before the correlation discrimi-
nator can be used as a reliable processor.
                                     16

-------
                               OCV
Figure 6.   Comparison of OCV and pitot tube at  93C  (RMS error = 1.0%)
                                17

-------
                                   12
                                OCV m/sec
Figure 7.  Comparison of  OCV and pitot tube at 150C (RMS error = 0.9%)
                                18

-------
                                OCV m/sec

Figure 8.    Comparison of OCV and pitot tube at 150C and 5 kg/hr of dust
            (RMS error - 1.8%).
                                 19

-------
                                   12
                                OCV m/sec
Figure 9.    Comparison of OCV and pitot tube at 204C (EMS error = 1.
                                20

-------
                      RUNS
                  COMBUSTION
                    PRODUCTS
                               OCV m/sec
Figure  10-
Comparison of OCV and pitot tube in the presence of combustion
products (RMS error = 1.4%).
                                 21

-------
                                                           ,04gm/m
              RUNT
              27C
           2.5kg/hr DUSt
                     '0.045
                                            055
                                  '0.065
                          ).08
                     8
   12
OCV m/sec
16
20
22
Figure  11.  Comparison of OCV and pitot tube with a dust loading of
           2.5 kg/hr (EMS error = 0.9%).
                             22

-------
                                   APPENDIX A

         CALIBRATION OF  POINT OCV AGAINST A LASER  DOPPLER VELOCIMETER
INTRODUCTION

     The point OCV has  previously been calibrated against a pitot tube with
reasonable success*  It was  now desired to  check the OCV against a more abso-
lute type of velocimeter.  Such an instrument  is the Laser Doppler Velocimeter,
This projects a fringe  pattern,  generated by a laser and of well defined size,
into the flow and "watches"  small dust particles cross this fringe pattern.
The light scattered by  these particles is collected by a photomultiplier,
amplified and its frequency  measured.   This frequency is uniquely related to
the flow velocity.  The particular instrument  which was used employed a Helium
Cadmium Laser and had a factor  of 39.5 kHz/meter per second.  A high speed
version of the correlation discriminator was used to measure the signal fre-
quency.

CALIBRATION TESTS

     The OCV was compared with the LDV over a wide range of environmental
conditions.  These are  detailed  in Table A-l.   The data are displayed in
Figures A-l  A-3.

            TABLE A.I.  TEST CONDITIONS FOR POINT OCV CALIBRATION
Run No.
8
9
10
11
12
Temperature
27C
27C
120C
120C
170C
% of Water
0
0
0
10
0
Dust Loading
Light
Moderate
Light
Light
Light
RMS Error
5.8%
0.9%
4.7%
6.4%
8.5%
*M.J. Rudd, "Development  of  an Optical  Convolution Velocimeter for Measuring
 Stack Flow," Submitted to Emissions Measurement and  Characterization Div.?
 Research Triangle Park,  NC.
                                     23

-------
   20
U
0)
o
o
    15
J   10
                                                            .075 mg/m3-

                                                                 .08
                       27 C
                    LIGHT DUST
     27 C
MODERATE DUST

 I	I
                       10       15       0       5       10

                       LASER DOPPLER VELOCIMETER (m/sec)
          15
20
                       Figure A-l.  Point OCV calibration

-------
           20
            15
S3
Ul
            10
U
O
                       5
                             125C
                               1
                                                         I
                      10       15       0        5        10
                      LASER DOPPLER VELOCIMETER(m/3ec)
                                                            125 C
                                                            10 % H0
15
20
                                  Figure A-2.  Point OCV calibration

-------
            20
             15
         u
         o>
         O
         o
             10
                                            170C
                          I
I
I
                          5          10          15         20


                   LASER DOPPLER  VELOCIMETER (m/sec)



                    Figure A 3.   Point OCV calibration
CONCLUSIONS
     The calibration of the OCV probe against the Laser Doppler Velocimeter

system was unsatisfactory and no definite conclusions may be drawn.
                                   26

-------
     The RMS error in this data  is much  larger than when the OCV was compared
with a pitot tube previously.  This  can  only be attributed to the unsatisfac-
tory operation of the LDV system.  No  prefiltering was employed before the
LDV signal was fed into  the  correlation  discriminator.  This led to large
amounts of low frequency noise being admitted and may be the cause of the LDV
reading low, particularly at higher  speeds.
                                       27

-------
                                 APPENDIX B

                DRAWINGS OF PROPOSED DESIGN FOR A CROSS-STACK
                       OPTICAL CONVOLUTION VELOCIMETER
     Attached on the following pages are drawings for a design for a proto-
type cross-stack OCV.  This consists of two housings - a transmitter/receiver
(Figure B-l) and a retroreflector (Figure B-2).   The light source is a tung-
sten halogen lamp powered by a 100 watt regulated supply.  The light is
collimated by a 82 mm dia by 300 mm focal length lens, and projected across
the stack.  The return beam is separated by a beam splitter, passes through
a grating and falls on a photodiode.  The signal is amplified and its fre-
quency detected by two circuit boards.  The retroreflector is of a cats-eye
design, an achromat lens with a mirror at its focus.  This minimizes the
sensitivity to angular alignment.

     The exposed sides of the lenses are protected by air curtains.  This
consists of air blown into a sleeve around the lens and then ejected around
its perimeter.  This prevents dust and condensation from settling on the lens
surfaces.
                                     28

-------
ro
10
                                      ,	LAMP ORIEL: 6397
                                                                                                                                     tO.1l
                                                                                                                                   -14 in I AMERICAN STANDARD
                                                                                                                                          PIPE FLANGE
                                                                                                         49.8]
                                                                                                        " I19.MI '
                                                                                               COATED GLASS WINDOW (40% REFLECTING
                                                                                                   LOW LOSS. 10.18 X 10.16 14.0 X 4.011
                                                                                                   ORIEL = A  46 - 3040 WITH MOUNT
                                                                                                            (ORIEL 12591
                                                                                                                                                         SUPPORT RING
                                                                                                                                                                      SEAVIESSSTEEL TUBI\G
                                                                                                                                                                           B9fj;50)OD
                                                                                                                                                                           8 56( 3.3;) I O
                                                                                                                                                           10 10 <4 ml (IK PIPE
                                                                                                                                                           11 4114 501 DC
                                                                                                                                                           to a 14 om i o
                                                                                                                                         SEAMLESS STEEL TUBING
                                                                                                                                              C.MI200IOD
                                                                                                                                              4.7511 97)1.0.

                                                                                                                                                                  i 72 X S 72
                                                                                                                                      ICVCLESton. 90% THANSPAHANCV 12 26  Z 25'
                                                                                                                                          ORATING WITH MOUNT (ORIEL 17H>
                                                                                   - K>X HOFFMAN > A-ltUSC


                                                                                   ICTHHIAA
                                       Figure  B-l.    Transmitter/receiver
TOLERANCES: UNLEti OTHEXWIH
SPECIFIED              ,
                                                                                                                                     DECIMAL:
                                                                                                                                     FRACTIONS

                                                                                                                                     ANGLES
                    ENGLISH  M9TRIC

              .XX    (JOOtOI    O.OM

              .XXX  ( 0.005)    0.013
                                                                                                                                                         t 1

-------
                           SEAMLESS STEEL
                           TUBING 12.0010.D.  60S
                                  11.87) I.D.  4.75
                                                                                      30.4B
                                                                                     "(12.00)"
                                              10.16
                                              (4 in.) AMERICAN STANDARD
                                                    PIPE FLANGE
                                                     SEAMLESS STEEL TUBING
                                                     8.86 (3.50) O.D.
                                                     (.96 (3.37) I.D.
                                                                -LENS MOUNT
                                                                                                -BOX HOFFMAN * A 1212SC
                                                                                     28.9S
                                                                                   -(11.401-
                                                                                                                                5.08        O.S4
                                                                                                                                (2.00) DIA. BY (0.25) THICK
                                                                                                                                MIRROR AND MOUNT
                                                                                                                              ' ORIEL = A-45 102 1.
                                                                                                                              30.41
                                                                                                                              (12.001
                                                LENSMELLESGRIOT
                                                  * 01 LAO 2*7
NOTES:

1 AIR SUPPLY
  PRESSURE
150 r.lm) 28.3 Mm/mm.
STACK  16" H?Ot
      t S z cm HzO
Figure B-2.    Retroreflector

-------
                                  TECHNICAL REPORT DATA
                           (Please read Instructions on the reverse before completing)
REPORT NO.

 EPA-600/2-79-192
TITLE AND SUBTITLE
 CROSS-STACK OPTICAL CONVOLUTION  VELOCIMETER
 Development and  Evaluation of a  Breadboard Design
AUTHOR
-------