EPA-660/2-73 011
September 1973
                       Environmental  Protection Technology Series
Explicit Calibration of  The

PILLS  II System

                                Office of Research and Development

                                U.S. Environmental Protection Agency
                                Washington. D.C. 20460


-------
                    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 STUDIES 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 Office of Research and
Development, 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 commerical
products constitute endorsement or recommendation for use.

-------
                                                    EPA-660/2-73-011
                                                    September 1973
                  EXPLICIT  CALIBRATION OF THE

                        PILLS  II SYSTEM
                               By

                     Frederick M.  Shofner
                       Project  16130 GNK
                    Program Element 1BB392
                        Project  Officer

                      Frank H. Rainwater
     Pacific  Northwest Environmental Research  Laboratory
                Environmental Protection Agency
                    Corvallis, Oregon 97330
                         Prepared  for

              OFFICE OF RESEARCH AND DEVELOPMENT
             U.S.  ENVIRONMENTAL  PROTECTION AGENCY
                    WASHINGTON,  D.C. 20460
For sale by the Superintendent of Document*, U.S. Government Printing Office, Washington, D.C. 20102 • Price 65 cents

-------
                                ABSTRACT
The basic characteristics of the PILLS I and II Systems are reviewed
and up-dated from previous publications.  Emphasis is given to the
explicit calibration of PILLS II with water droplets.  The operational
characteristics and accuracy of the monodisperse particle generator
are discussed.  General procedures for deriving the particle density
distribution from the measured voltage density distribution are
described.

This report was submitted in fulfillment of Project Number 16130 GNK,
by the Environmental Systems Corporation under the partial sponsorship
of the Environmental Protection Agency.  Work was completed as of
June 1972.
                                   ii

-------
                                CONTENTS
                                                                 Page
Abstract                                                          ii
List of Figures                                                   iv
Acknowledgments                                                   v
Sections
I      Conclusions                                                1
II     Introduction and Background                                2
III    Review and Updating of PILLS Characteristics               4
IV     Description of Calibration Apparatus                       16
V      Calibration Procedures and Results for PILLS II            21
VI     References                                                 26
VII    Glossary                                                   27
                                   m

-------
                               FIGURES
o^                                                              Page
1   PILLS II System                                               5
2   PILLS Prototype                                               6
3   Calibration Apparatus                                        17
4   Monodisperse Particle Generator                              18
5   300 ym Droplets near the Exit of the Vibrating Jet           20
6   PILLS II Response                                            22
7   PILLS II Average Voltage Output Versus Particle Diameter     24
                                   iv

-------
                            ACKNOWLEDGMENTS
The assistance of Thomas B.  Carlson, Yasuo Watanabe, Lawrence  T.  Hall,
and R. Wesley Johnson in the execution of these experiments  is
gratefully acknowledged as are technical  discussions with Drs. Harold
W. Schmitt and Carl 0.  Thomas.

The interest and patience of Frank H.  Rainwater, Project Monitor, is
appreciated.

-------
                               SECTION I

                              CONCLUSIONS
The basic characteristics of the PILLS I and II Systems are reviewed and
up-dated from previous publications.   Emphasis is given to the explicit
calibration of PILLS II with water droplets.  The operational  character-
istics and accuracy of the monodisperse particle generator are discussed.
General procedures for deriving the particle density distribution from
the measured voltage density distribution are described.
                                                                    p
The essential results of the calibration of PILLS II are (1) AV = Kd
where K is a constant for either of the two PILLS II scattering volumes,
(2) K is independent of droplet mineral concentration up to 50,000 ppm
for NaCl, and (3) errors in determination of particle density distribution
of < + 15%.

-------
                              SECTION II

                      INTRODUCTION AND BACKGROUND


This report is an extension of work initiated under EPA Demonstration Grant
No. 16131 GNK which led to a report entitled "Development and Demonstration
of a Low-Level Drift Instrumentation" (No. 16131 GNK 10/71) which was
published by the Environmental Protection Agency in October 1971 (Ref. 1).
The purpose of that work was to demonstrate the effectiveness of cooling
tower drift instrumentation which Environmental Systems had developed up to
that point in time.  The prime instrumentation discussed was the Particulate
instrumentation by Laser Light Scattering (PILLS I) System, although our
complementary technTques of heated glass bead isokinetic sampling (IK) and
sensitive paper (SP) were described fully and data from them reported.  This
intensive activity was initiated in March 1971 and culminated in a
demonstration and seminar on September 28, 1971.  The meeting was held for
convenience at the Oak Ridge National Laboratory and approximately 80 persons
from across the country intimately concerned with the drift problem were in
attendance.  The equipment demonstration was given on a mechanical draft
cooling tower on air conditioning service at ORNL.  Most of the measurements
reported in the above-mentioned document were made on this tower.

At the time of that report, an explicit calibration of PILLS I using water
droplets had not been performed.  Implicit calibrations comparing the
PILLS I data to complementary data obtained on a cooling tower via IK
sampling and an explicit calibration using small glass spheres had been
performed. Subsequently, their predictions were verified by explicit
calibrations performed during 1972, wherein the PILLS I System was calibrated
with monodisperse water droplets of approximate diameter 150, 300 and 500
microns.  These results are described in "Design Considerations for
Particulate Instrumentation by Laser Light Scattering (PILLS) Systems)"which
was presented at the ISA National Meeting in October 1972 and published
recently in the ISA Transactions.  This paper is designated as Reference 2.
Although an explicit calibration of the PILLS I System was provided, the
calibration was accomplished with only three particle sizes and upon equip-
ment superceded by the PILLS II generation.

Our experience with PILLS I indicated a basic quadratic response (v a d2)
where v is average output voltage and d is particle diameter.  Furthermore,
examination of Figures 17 and 19 in Reference 1 indicate two peaks in the
particle density distribution near 120 and 200 ym.  It is most important to
note that these two peaks are evident in both PILLS I and SP data.  This
proved that the PILLS I system was capable of resolving particle density
distribution structure of Ad/d» +_ 0.152.  We presumed the same for PILLS II
since its basic electro-optical design features were taken from PILLS I.
Performing the present calibration experiments carefully has indicated that
the PILLS II configuration, which has been used extensively in data
acquisition, did not have a "peaking" voltage response for monodisperse

-------
particles, as PILLS I apparently had, but yielded a wider range of voltage
outputs for single particle diameter input.  The physical reasons for this
type response have recently received intensive investigation and the causes
have been identified and corrected.

However, it was found that the average voltage response for PILLS II was
v = Kd^, as expected.  More important, determination of the basic drift
parameter, particle density distribution, was obtained in a straight forward
manner.

This report is an extension of the earlier report (1) and paper (2) and is
limited primarily to reporting a recent explicit calibration of the PILLS II
instrument.  The equipment demonstrated on September 28, 1971 is now called
the PILLS I System and had a single scattering volume and employed a photo-
multiplier detector.  Since that date there have been significant improve-
ments  in the PILLS System concept and calibration apparatus.  This report
describes the PILLS II generation equipment, its calibrations and data
interpretation procedures.  Further improvements in the equipment are being
continually made but the present calibration data accurately describe the
PILLS  II  instrument which has been used extensively for both natural and
mechanical draft cooling tower drift tests in both the U. S. and Europe.  The
describing equations for PILLS are given,  along with the defining equations
relevant  to  the drift properties measured.  These latter equations are of
general utility and  are independent of measurement techniques.

A more complete definition  of the  drift  problem and a discussion of its
importance are contained  in our earlier  report, together with a general
description  of the PILLS,  SP, and  IK  techniques.  We note that that part of
the drift  problem  which the PILLS  System advantageously  addresses is of much
greater importance than envisioned at our origination of this work over
three  years  ago.   Finally, we would  like  to point out that  significant improve-
ments  in  the  IK and SP techniques  have also been realized since our earlier
report.   These developments now enable us to use these techniques reliably in
conjunction  with the PILLS  Systems to obtain more complete  drift information.

-------
                              SECTION III

              REVIEW AND UPDATING OF PILLS  CHARACTERISTICS


REVIEW OF PRINCIPLES AND DEFINING EQUATIONS

It is appropriate to briefly review the principles  of the PILLS  System
concept.  Figure 1  gives the schematic diagram for  the PILLS  II  System
and Figure 2 shows  photographs of a prototype device.  (Refer to the
EPA report, Ref. 1, or to the ISA paper, Ref. 2,  for a more complete
description of the electro-optical elements.)  The  instrumentation
behind the PILLS II sensor head is Environmental  Systems' dual multi-
channel pulse height analyzer which will be described below.

The basic operating principles of the system are  simple:   a single
drift particle present in one of the scattering volumes at the instant
of a pulse of radiation from the laser scatters light into one of the
detectors and produces a voltage signal.  This voltage signal is then
recorded on magnetic tape or electronically analyzed and  a count is
stored in the appropriate channel in our multi-channel analyzer. The
voltage distribution, which corresponds to the size distribution of^
the particles present, is obtained from the number of counts  accumulated
in each channel of  the analyzer.  Combining this  measured voltage
distribution and the known response function of the instrument gives
the particle density distribution, the basic drift parameter measured.

Figure Ib shows a representative PILLS II output typical  of either
of"the scattering channels.  Photons scattered from the scattering
volumes arrive at the off-axis detectors and produce a pulse of  current
which is amplified, converted into a voltage signal, stretched,  and
driven into a coaxial cable at the end of which the voltage monitoring
and signal processing instrumentation are placed.

Because fog constitutes a very dense ensemble of small particles,
having mean diameter of a few micrometers, many of them  are in the
scattering volume at each laser  pulse.  The voltage produced is  often
large  ancfexhibits  random fluctuations  frcm  pulse to pulse: the latter
sets the  practical  lower limit on drift particle measurements for the
system.   Since  the  fog yield and  fluctuations decrease with scattering
volume  size,  smaller scattering  volumes permit measurements  to smaller
drift  particle  sizes.

Also,  the  smaller  the scattering  volume, the smaller  is  the  particle
size which  can  still entertain single  particle scattering.   In mathe-
matical  terms,  it  is required  that  the  probability of more than one
drift  particle  in  the scattering volume be very  low.  Typical particle
density  distributions decrease  rapidly  with  increasing diameter.  The
 lower  limit  set by  the  requirement  of  single particle  scattering is
 typically much  smaller  than  the  lower  limit  imposed  by fog fluctuations

-------
JUNCTION DIODE LASER       —
AND BEAM FORMING OPTICS  5SSS8T0


       (A) PILLS E     CONFIGURATION
             PRI- AND WT AMPUHEK$;
             STRETCHERS; DRIVER AMPLIFIERS
                                                                                        FMONITORING  AND SIGNAL
                                                                                         PROCESSING:
                                                                                          OSCILLOSCOPE.  VM, PULSE
                                                                                          HEIGHT ANALYZER
                                                                                          TAPE RECORDER
                                                                                          (ANALOG  OR DIGITAL)
                                                                                        ^MINICOMPUTER
DETECTOR
IT,-
OR
1T2
L
w
•••
^M*
L
Tl
1
ku
i
rtDRIFT PART
nfiS
ICI
XV
.E)
ER
«•
AGE FOG YIELD
n
n ,
                                                                  AN
                                                                               'V    'V
                                                                               Jz    ivi
       (B)  PILLS  H    OUTPUT
                                                                                                    SATURATION
                                                                                                      LIMIT
                            »50  s»200          *1000'


                        (C) PARTICLE  DISTRIBUTION  DATA
                                               FIGURE
        1
                                     PILLS H      SYSTEM

                       (PARTICULATE   INSTRUMENTATION   BY  LASER  LIGHT  SCATTERING)

-------
            (a) Close-up of sampling volumes
(b)  PILLS II  and the ESC pulse height analyzer
        Figure 2.   PILLS II  prototype
                      6

-------
for a given scattering volume.   Alternatively stated, fog fluctuations
typically pose a more severe limitation on the lower practical  limit
than having more than one particle in the scattering volume.

However, small scattering volumes decrease the data rate for  the
infrequent but highly important large particles.   Consequently, a
wider range of drift particle size is covered by appropriate  combination
of a small volume which extends downward the practical range  of
measurements as limited by fog fluctuations on the lower end  of the
range and a large volume which extends upward the range, due  to higher
data rate, for the large particles.  If the fog yield were constant,
then the AV information corresponding to the presence of a drift
particle would permit subtraction down to the electronics noise
limit which is about 30 ym for V2, the small scattering volume.  Based
on experience with large natural draft cooling tower measurements in
the near vicinity of the drift eliminators, two independent scattering
volumes of V-j = 1.2 cm^ and Vp = 0.4 cm^ have proven adequate and
represent the configuration wnose calibration is given here.

As shown on Figure Ic, we have found that the small scattering  volume
permits measurements from approximately J50 to 200 urn and the  large
scattering volume permits measurements from, typically, about 200 yrn
up to a limit that is either determined by saturation of the  detector
electronics or by physical size of the particle as compared to  the
laser beam.   For PILLS  II, the saturation  limit dominates and is
approximately 1,000 urn.  Thus we are able  to cover the full range of
particle diameters from 50 to  1,000 ytn with good data acquisition
rates and with accuracies optimized for both smaller and larger
particle sizes.

In addition to fog fluctuation considerations,, the pulse repetition
rate and the scattering volume transverse  dimension dt in the direction
of the flow must be so chosen  that a given entrained particle will not
be counted more than once by successive laser pulses.  That is,

                        uAt >  dt.                                     (1)

It is readily seen that this relation is satisfied for the flow velocities
u > 4 m/s and repetition rates of 300 Hz =  I/At.  This is confirmed
experimentally in the field by observing that in very few instances
does one observe multiple pulses.

The basic measured output from the PILLS System is the voltage density
distribution for the voltages AV above the average fog yield which are
produced by single drift particles and which are recorded in the field
with the PHA or produced with minicomputer processing of field-recorded
magnetic tapes as described later.  Formally, if matrix R(v;d)  represents
the PILLS voltage response to particles of diameter d and matrix P(d)
is the actual particle density distribution, then the measured voTtage
distribution ^.(v) is given by

-------
                     Q(v) = J*(v;d) P.(d).                              (2)


Since £ and j* are known by field and calibration measurements, P/d)
is determined according to


                     P.(d) = R'1 (v;d) Q(v).                           (3)


Although the relationship between an output voltage pulse and an input
particle of size d is not unique, or one-to-one unless R(v;d) corresponds
to the mathematically ideal "impulse or delta function" response, the
drift statistical properties, i.e., particle density distribution are
rigorously determined with minor restrictions in the class of allowable
matrix element functions in P(d).  These procedures require, in general,
use of numerical methods.

Equation (2) can also be formulated in terms of an integral equation:


                               P^max
                       q(v) =  \     r(v;d)p(d)dd                     (4)
                               'dmin

where now q, r, and p are normalized probability densities per unit volume.

Thus, the principal objective of calibration is accurate determination of
the instrument response.

Rather than employing matrix or integral equation inversion for high
volume data reduction, we have discovered, developed, and applied
extensively a simpler, approximate data analysis technique which obviates
the use of digital computers.  Procedurally, a predicted particle density
distribution is determined according to


                        p'(d)dd = q(v)dv

                                = q(Kd2} 2Kd dd                       (5)


The interpretation of (5) is that we assume the number of voltage pulses
in the voltage interval (v, v + dv) is exactly equal to the number of
particles in the interval (d, d + dd) where v = Kd2 defines a unique
relationship between the interval end points.  Alternatively stated,
this amounts to assuming that the instrument response is described by


                       r(v;d) = 6 (v-Kd2)                             (6)


                                   8

-------
which is the ideal case.
This technique gives satisfactory engineering results.  As will be seen,
the instrument response function may be approximately modeled as


                         r(vjd) = l/vd e'v/vd                         (7)


where

                         vd = v = Kd2                                 (8)


For this type response and for p(d) «c l/da a closed-form analytical
treatment is possible and it can be shown by direct substitution that
                                      C(a), large d                   (9)


where C(a) is a constant depending on a.


Thus for "power law" particle density distribution,  for example, this
technique is precise for large d since  inspection of the p'(d) or
q(v) data indicates the value of a from which  C(a)  is  then determined.
C(a) =  1, 1.3, 2,  and  3.3  for a = 3, 4, 5,  and 6 respectively.

We  have employed this  technique extensively in the  reduction  of data.
A most  encouraging experimental fact has  been  the good agreement between
PILLS  II  data  reduced  as described and  SP data in the overlapping  range
where both  give  p(d).  This  good experimental  agreement via two
completely  independent measurement techniques  verifies the assumptions
made in Equation  (5).  Precise  results, as  noted, are obtainable for
 typical and mathematically well-behaved particle  density distributions.


 As  with any physical  measurement,  the  accuracy of the PILLS System is
 related to the accuracy  of the  calibration  and maintenance of cali-
 bration conditions during field operations.  Calibration particle  sizes
 are known to within +5% and  field  and  laboratory  voltage measurements
 are readily made to within +5%.   The number of counts in a given
 voltage channel  is very accurately  known.  Assuming +5% errors to  allow
 for voltage drifts, departures  from  linearity and fog fluctuation
 effects, the determination of the  number of particles in a given
 interval (d, d+Ad) is taken  as  4 +15%.   Although  a  thorough error
 analysis has not been made,  we  conclude that +15% represents  the
 operational errors of the PILLS II System in  making drift particle

-------
density distribution measurements over the range 50 < d < 1000 ym
when sufficient counts  are obtained to permit good statistical
definition of the measured voltage density distribution.

We now define the measured drift parameters and discuss the improvements
in the system.  The PILLS II equipment is  operated for a  period T of,
for example, one hour during which a total volume Vs is sampled
according to:
_ 1.2 x 3.6 x 103 - 3.33 x TO"3 = 1.30 m3/hr,  V-j
-
          T/.t
          T//,t
Interestingly, the repetitively-sampled,  non-contiguous  scattering
volumes can be called "PILL-BOXES".

The number of voltage pulses in a given voltage interval  is  converted,
via the instrument response, into the number of particles having
diameter dj in the associated size interval  Ad.,-.   This information,
with knowledge of the total volume sampled Vs,' leads to  the  particle
density distribution:

Note that the i subscript has been omitted where it is redundant; this
practice will be continued in remaining definitions.   Note also that
"density" refers to the number of particles in a given range per unit
volume and that "distribution" refers to the manner in which the total
number of particles per unit volume is distributed over the range of
diameters of interest.  The present uses of these adjectives are to be
clearly distinguished from the conventional statistical usages  wherein,
for example, p(d) is the probability density per unit volume and the

probability distribution per unit volume is J    p(d)dd.  Note that
  rdn                                      yo
      p(d)dd -H as dn—4 - .
 '0
the mass density distribution is easily derived from the particle density
distribution according to:


                         AX-J  A  TT        AN1
                                   10

-------
where  s  is  the specific mass  (gm/cm3) of the particles and j   Adi                      (15)


                   = -rrs/6 j u^- d,-3  ANi   . Adj (gm/n^-sec)


and the total mass emission is:


                              D = EAD. • An- =
                                  J  J    J

        ^  [A.-xz u,, (Tr/6  s  d,3  ANi   ) Ad.1    (gms/sec)         (16)
        0    J i   J           1   Ad-Vc     n
where Aj represents the appropriately-chosen segments of the total emitting
surface area across which the drift particles are transported.  The u^-
velocity components must be representative of the i™ size particle
absolute velocities in the Aj area.

Historically, the total drift emission of a cooling tower has been
characterized as a percentage of the circulating water rate R in grams/sec
and is therefore given by:


                                   11

-------
                        A = D/R x 1003J                               (17)


For completeness, it is appropriate to define a specific drift fraction
« which is applicable for measurements made in the immediate vicinity
of the drift eliminators and is more appropriate for counter-flow than
cross-flow towers.  The specific drift fraction is the mass flux at
a given location near the drift eliminators divided by the water
loading per unit area of drift eliminators associated with that location.
IMPROVEMENTS

The improvements in the PILLS II System over the PILLS I System are
briefly summarized in the following paragraphs.  It is interesting to
examine the recommendations for further work in the October 1971  report
for comparison with the improvements actually made.

Explicit Calibration

As noted above, explicit calibration using water droplets had not been
realized by October 1971.  This is the principle subject of this  report
and is discussed in succeeding sections.

Improvements in Field Reliability

The PILLS I System was a developmental device whose basic optical  con-
figuration could be varied but which lacked the rigidity necessary for
a routinely operational field device.   The various  angles and distances
as determined from the PILLS I instrument were chosen as the basic inputs
for the design of PILLS II which, upon comparison,  can be seen to  be
more rigid and smaller in physical size.  In addition, the troublesome
and noisy photomultiplier detector in  PILLS I was replaced with a  semi-
conductor photodetector in conjunction with a high  current gain,  low
noise amplifier circuit whose overall  performance is superior to  that
of the photomultiplier for the 9040 A  radiation. As a consequence of
these and other improvements, the PILLS II System has logged more  than
1,000 hours with minimal down time in  operational cooling towers.   In a
recent test in a natural draft unit, the system remained inside for
a continuous period of more than 400 hours.

Fog Suppression

As indicated schematically on Figure 1 and discussed above, the high
density of fog particles having mean diameter of a  few microns produces
a scattered component of light each time the laser  fires; this is  termed
the fog yield.  The fog yield voltage  is approximately proportional to


                                   12

-------
the size of the scattering volume and,  as  pointed  out above,  if  it wera
constant over a period embracing several  laser pulses (a  few  milliseconds),
then the single drift particle information extending to smaller  sizes
could be realized by subtracting out the  fog yield.   Alternatively
stated, fog fluctuations set the lower  practical  limit for meaningful
drift particle measurements and, as a consequence, small  scattering
volumes permit smaller practical measurements of drift particle  size.
However, small scattering volumes have  lower data  rate for the
infrequent large particles.  Accordingly,  we have  found that  the two
completely independent scattering volumes  described above permit
measurements from about 50 microns up to  1,000 microns in typical cooling
tower fogs and with sampling intervals  of a few hours. But the  fog yield
fluctuations may also be minimized in signal processing,  as discussed
next.  Indeed, the extraction of meaningful information from the_"fog
plus drift particle" pulses has had a major influence in  the  design
of signal processing techniques and equipment.

Signal Processing

PILLS I data were recorded by the "brute  force" technique of setting
a precision oscillosocpe trigger level  slightly above the maximum
excursion of fog yield and recording, in  the single sweep mode,  on
photographic film the individual fog plus drift particle  pulses.
Although slow and tedious as compared to  techniques presently employed
with PILLS II, this technique allowed accurate and easily interpretable
voltage measurements.

PILLS II data are now routinely acquired with the two following  complementary
recording/processing techniques:   (1) field pulse height  analyzer and
(2)  analog tape recording/minicomputer processing.

The  Environmental Systems' dual 10-channel pulse height analyzer was
specifically designed and fabricated to accomodate the output of the
PILLS scattering channels.  The analyzer records the number of voltage
pulses whose excursion above the average fog  level falls  within  the
adjustable voltage  intervals of the analyzer.  The average fog yield
is established by long time constant, analog  circuitry (termed the "fog
cancellation circuit") which precedes the voltage height analysis section
of the  PHA.

This device permits true  on-line acquisition  of data  and has proven
extremely  valuable  in the  exploratory determination  of test point position
within  a  cooling tower as well  as  sampling  time duration.  The data are
"read out" by means of a  single  light emitting diode  indicator which  is
connected  sequentially with the outputs of  decade counters in each of
the  dual  10 channels.  The PHA  data  are especially valuable for large
particles  which  require  long sampling times  for adequate statistical
counts.   Routinely, we  leave the PILLS II/PHA System in operation overnight
and  unattended,  a most  effective way to collect data!  The analog fog


                                    13

-------
cancellation circuitry does not permit extension to drift particles as
small as does the computer processing of the tapes described next.   The
PHA data are of prime use for the large particles in each of the
scattering volumes as realized with long periods of data collection,
as in an overnight run.

An FM cassette magnetic tape recorder is used to collect the PILLS
output pulse train in an analog fashion.  The outputs of the two
scattering channels plus the reference detector output are recorded.
The tapes are then returned to the computer lab wherein a minicomputer
examines these analog outputs and extracts the drift particle data  in
each of the two scattering channels in the following way.  First, each
scattering volume pulse is normalized to the laser power for that
pulse, thus removing peak to peak variations in laser power down to
the noise limit of the detectors and amplifiers.  Next, each normalized
pulse is defined as a drift pulse ("B" in Figure Ib) and the average
of the pre- and post-pulses ("A" and "C" in Figure Ib) is subtracted
from it.  The value of this voltage difference is then determined and
stored in one of two thousand channels corresponding to voltage height.
According to this algorithm, the minicomputer then produces the net
voltage distribution from which the particle density distribution may
be derived.

The minicomputer software and PILLS II hardware have been so configured
that the intermediate step of analog tape recording can be omitted  and
the minicomputer can operate upon the PILLS II output in a truly on-line
fashion, giving the voltage distribution immediately after (or during)
a test.

We have also developed the capability of digitally recording the voltage
pulses on magnetic tape.  This capability eliminates the usual problems
of variations of gain in analog recording.  Also, computers can more
readily digest digital information than the analog form; indeed, the
analog data must first be digitized for a computer to digest it. This
technique will be routinely employed in the future when a field tape
record is required*

In conclusion, the PILLS II System whose calibration is reported here
represents substantial improvements over the PILLS I device with regard
to field reliability, range of practical measurements, and signal
recording/processing capabilities.  More important, our understanding
of the electro-optic properties of the device have advanced considerably
as has our understanding of the drift data produced by it.

Improvements of equal magnitude in both the sensitive paper technique
and the isokinetic sampling technique have also been realized since the
October 1971 report.  Numerous intensive tests involving all three
techniques.have been performed and the data thereby generated are now
being "fed into the system" for evaluation of the feasibility of,
                                   14

-------
particularly, salt water, natural  draft cooling  towers.   It  is most
important to note that, in addition  to  being  complementary,  the  data
generated by these three techniques  exhibit a high  degree of internal
consistency and good agreement when  compared  to  each  other in common
terms.
                                   15

-------
                              SECTION IV

                 DESCRIPTION OF CALIBRATION APPARATUS


It is well known that a highly monodisperse beam of liquid droplets can be
generated by vibrating orifice techniques which exploit the inherent
instability of a fluid jet interacting with a viscous medium, (References
3 and 4).  Our initial attempts with these techniques were with the
Princeton Fluidics Corporation earphone driven, longitudinally vibrating,
orifices.  Although we were able to calibrate the PILLS I device with three
particle sizes as discussed in Reference 2, this equipment proved unreliable
and particles over the entire range of interest could not be generated.

It proved necessary for us to completely develop the particle generating
technology ourselves.  Two laboratory visits at the University of Minnesota
and the University of Illinois proved very helpful and acknowledgements are
made of the discussions with Drs. Benjamin Liu and Richard Bergland of the
University of Minnesota and with Dr. Charles H. Hendricks and Mr. Chris
Foster of the Charged Particle Research Laboratory at the University of
Illinois.  In addition to learning how to reliably and control 1 ably generate
monodisperse beams from 15 to 700 microns using transversely vibrating
glass orifices, it was necessary to develop separation techniques so that one
particle of the prescribed size could be admitted to the scattering volume
at a time.  It was found that roughly one thousand volts DC applied to a
charging collar near the nozzle exit and through which the particles are  :
projected permitted a charge to be applied to each of the droplets.

Unless the particles were deflected as described below, this voltage would
produce a rapid spatial dispersion of the droplets which were then allowed
to fall through the appropriate scattering volume.  The single particle
occupancy condition was cerified by inspection of the output of that channel.
This technique was employed in the data reported here.

Considerable time was spent in attempts to electrostatically deflect single
particles out of the monodisperse beam by applying several hundred volt
pulses to deflector plates immediately following the charging collar.
Although this is relatively easily done for particles in the range of 70 to
300 microns, it becomes increasingly difficult for both larger and smaller
particles.  Figure 3 shows a schematic diagram of the calibration apparatus
as used for these tests.  Figure 4 is a photograph of the principal components
of the equipment.  The orifices are produced by heating and drawing down to a
small diameter standard Pyrex tubing.  The ends are then polished off.
squarely using an abrasive.  As a rough estimate, the particle size generated
by a particular nozzle exit diameter will be about twice the nozzle diameter.
These nozzles are then mounted upon electrical bimorphs which flex in a
bending mode upon the application of a voltage across the insulated disimilar
materials.  Specific particle sizes are realized by "cut and try" on the
preparation of the nozzles and by minor variations of the frequency and
pressure parameters.  As an example, a 50 micron orifice will generate roughly
a 100 micron droplet when the frequency is 20 KHz and the pressure is 10 psi.


                                   16

-------
         Precision
         Regulator
Distilled
and Filtered
Water or Salt
Solution
                                                          Charging
                                                          Collar
Microscope
                                     Drawn
                                     Glass
                                     Tubing
                                            Timing
                                            Circuits
                                             and
                                            Strobe
                                            Driver
     Sensitive  Paper
                                         FIGURE 3.  CALIBRATION APPARATUS

-------
     (a)   General  view of apparatus
(b)  Close-up of jet and deflection plates
Figure 4.   Monodisperse Particle Generator
                    18

-------
Larger particles require lower frequencies and lower pressures  and are
easier to generate.  Small  particles, in the range of 15 to 50  microns,  are
more difficult to generate and observe.   The importance of having very clean
solution should not be underestimated.   Clogged nozzles can usually be
opened by forcing alcohol backwards into the nozzle with a syringe.

The strobe lighting of the particles is  essential  for optimum adjustment of
the frequency and pressure.   Satisfactory operating conditions  are
represented by uniformly sized and spaced particles with no small satellite
droplets.  A useful test of the nozzle  prior to cementing it on the bimorph
is to observe the distance that the free jet will  project before breakup; the
greater the distance of the unbroken free jet, the easier it is to generate a
monodisperse beam.

Particle size is determined in two ways:  (1) via a calibrated  reticule  in
the microscope and  (2) by collecting a  few milliliters from the jet and
dividing by the known number of particles in the collection; over most of
the particle size range of interest one  particle is produced per cycle of
the oscillator wave form.  This relationship is usually true but must be
carefully checked since it is possible  for two particles per cycle to be
produced.  The independently determined  diameters of the droplets agree
within better than  +5%.

Figure 5 shows a stream of approximately 300 micron particles.   The photo-
graph was made very near to the exit of the nozzle.  No charging was applied.
Mote the oscillatory shape of the droplets which is typical for distances
within a few centimeters of the nozzle.

When the droplets are dispersed by application of high voltage  to the
charging ring, the  angle of dispersion  depends upon the particle size and
the voltage applied.  Typically, 1000 volts are required to disperse a 100
micron beam into an area several inches  in diameter.

This apparatus has  been essential for calibration of our other  drift measure-
ment apparatus as well.  The sensitive  paper machine exhibits a strong
impingement velocity dependence of the  stain size versus particle size.
Further, the trapping efficiency as a function of particle size of our heated
IK tubes has been determined with this  droplet generator.  Finally, the
equipment has also  been used in generating opaque liquid droplets.

In conclusion, although the principles  of vibratory capillary particle
generation are easily understood, operation of such equipment rapidly
becomes more art than science, especially when particles are deflected
by electro-static means from the main beam.  Stated more practically, it
is very easy to underestimate the difficulty and expense of the application
of these straightforward techniques.
                                   19

-------
                                 o
                                  0
                                  o

Figure 5.   300 ym droplets near the exit of the vibrating jet
                             20

-------
                               SECTION V

            CALIBRATION PROCEDURES AND RESULTS FOR PILLS. II
PROCEDURES
Since both the PILLS II System and the calibration apparatus have been
described in the two previous sections, the results can be concisely
presented.  Procedurally, the calibration apparatus was set up to generate
a horizontal beam of particles which was then dispersed with the charging
ring as shown in Figure 3.  One or the other of the scattering volume
outputs was monitored with an oscilloscope/camera system for the bulk of
the measurements.  More recently, calibrations have been effected using
the dual 10-channel pulse height analyzer.

The dispersed particle beam proceeds several centimeters, depending upon
the size of the particles.  The particles approach terminal settling
velocity as they reach the scattering volume under concern.  Single
particle scattering is ensured by requiring the particle pulse rate to be
significantly lower than the repetition rate of the laser.  Uniformity of
the dispersed particle beam is tested by allowing the particles to fall on
discs of sensitive paper placed below the sampling area.  It is required
that each portion of the sampling volume under calibration have equal
probability of particle occupancy.  This is ensured by making the stain
density of the sensitive paper uniform.  Also, the fact that the particles
have the same size as the point at which they are measured is verified
with the sensitive paper.  Due to the inherently "wet" nature of the air
around the dispersed beam, no evaporation effects were observed as expected
for the relatively large particles used in the calibration.

The particle pulses were recorded using the single sweep oscilloscope
technique on Polaroid film.  Approximately 200 pulses for each particle
size were obtained.
RESULTS

Determination of Response Function

Figure 6 shows the output voltage distribution when particles of diameter
d - 256 ym were introduced into the large scattering volume.  This experi-
ment was performed with the PHA and would have been impractically difficult
without it.  It is to be noted that the response for large voltages is
approximated well by a relation r(v;d) /*• e -v/vd.  since for exponential
response v
-------
10
                                                           1   pfffij^



                                                              X -;Expier
                                                              . - Experiment #2


                                                 '

                                                            I



                                                                  -

                                       •
                                                      \      v


                            ;

                                                        5

                                                          \!

                                                              \
                                                             •\



^.OGARITHMIC SCYCLESXIOTOTH^
     5TH LINES ACCENTED
                                                                     AV, VOltS

-------
parameter) may be used directly in numerical evaluations.  We have employed
this numerical technique in the reduction of critically important data for
which maximum accuracy was essential but utilize the approximate technique
discussed in Section II more routinely.

Determination of K] and K£

Figure 7 shows the resultant average output voltage versus particle size for
the PILLS II System.  It should be pointed out that these are the net
stretcher output voltages (as appears at the end of the coaxial single cables)
versus diameter relationship and are routinely used for the reduction of data.
This calibration thus represents the average voltage response over the
particle size range from 60 to 700 microns.  It is seen that both curves fit
well to a Kd2 relationship.  Accordingly, the values of K and the range over
which they apply are:
               K-, = 1.0 x 10'5 V/ym2, 100 < d < 1000 ym

               K2 = 3.6 x 10~5 V/ym2,  50 < d <
                                                 300 ym


Accuracy of particle sizes is better'than +5% and voltage determinations are
better than +5%.  Multiple repetitions of each data point to determine "error
trees" were not performed due to time consumption; each point represents at
least two man days just to acquire the raw data.  Repeatability is good, as
explained later.

Each data point on these curves is produced by simple averaging of approxi-
mately 200 voltages, excluding the values less than 5% of the maximum value
observed.  The principal basis for this approach is that it leads to the
expected Kd2 relationship.  Smaller or larger fractions would also lead to
this relation but would have been impractical^ difficult to measure or
would have omitted a significant fraction of the pulses.

The data point for the small scattering volume at 470 microns is below the
straight line portion defined by the other four data points since the small
scattering volume stretcher departs from linearity in the neighborhood of
3 to 4 volts.  The circuitry for the large scattering volume is linear to
approximately 10 volts, thus making the large scattering volume extrapolatable
to 1,000 micron particle diameter.

The encircled data points shown on the large scattering volume curve were
generated with the pulse height analyzer equipment in a post-test calibration
performed several months after the other data were generated and after the
system had been used in several field tests and operated for a total period
of roughly ten weeks in operating cooling towers. It is to be concluded that
the calibration is repeatable and the PILLS II characteristics did not shift
during field use.
                                   23

-------


10 mV
                                                              24
  Fi-   Logarithmic,  3 x 3 Cycles

-------
Effects of Dissolved Salts

An experiment was performed to determine the effect upon the  calibration
of high concentrations of salt dissolved in the water droplets.   Procedural!",
a beam of pure water droplets, d = 108 pm, with no electrostatic, dispersion
was admitted at a fixed location in the large scattering volume  to produce  a
reference voltage.  Multiple particle scattering was realized for this,
undispersed beam case; nominally, 5 particles were in the scattering volume.
The particle generator and PILLS System were mounted in fixed positions  and
the particle size and resultant voltage observed and recorded.

The distilled water in the particle generator reservoir was then exchanged
for 50,000 ppm NaCl solution.  It was then verified that the salt solution
oarticle size was identical with that of the pure water droplets   The
resultant voltage was also identical, leading to the conclusion  that salt
concentrations up to  50,000 ppm do not  affect the calibration of the PILLS
instrument.  The practical consequence  of this result is that the calibration
can be expected  to hold independent of  mineral concentrations when the PILLS
System is employed in brackish or sea water  applications.
                                     25

-------
                             SECTION  VI

                              REFERENCES
1.   Shofner, F. M., and Thomas, C. 0., "Development and Demonstration of
     Low-Level Drift Instrumentation," Environmental Protection Agency
     Report No.  16131  GNK 10/71.

2.   Shofner, F. M., Watanabe, Y., and Carlson, T. B., "Design Considera-
     tions for Particulate Instrumentation by Laser Light Scattering
     (PILLS) Systems," ISA Transactions 12 (56-61) 1973.

3.   Mason, B. J., Jayaratne, 0. W., and Woods, J. D., "An Improved
     Vibrating Capillary Device for Producing Uniform Water Droplets of
     15 - 200 pm Radius," J.  Sci. Instruments 40, 1967.

4.   Liu, Benjamin, and Berglund, R. N., University of Minnesota,
     Minneapolis, private communications.
                                   26

-------
                              SECTION VII
                                GLOSSARY
Aj   =    portion of total emitted surface area
C(a) =    constant
d    =    particle diameter
Ad   =    incremental diameter
dj   =    mean diameter of the particles
dt   =    scattering volume transverse dimension
D    =    mass emission
IK   =    isokinetic sampling instrument
K    =    constant for either PILLS II scattering volumes
AN.,-  =    number of particles in i™ size element
p(d) =    normalized true particle probability density per unit volume
F^(d) =    actual particle density distribution
p'(d)=    predicted particle density distribution
q(v) =    normalized voltage distribution per unit volume
£(v) =    measured voltage distribution
R(v;d)=   matrix representation of instrument response function
R    =    circulating water rate
s    -    specific mass
SP   =    sensitive paper particulate instrumentation apparatus
T    =    duration of measurement interval
At   =    time between pulses of LASER
u    =    air flow velocity
Uj•  =    absolute particle speed
                                   27

-------
   V-j    =    large scattering volume
   ^2    ~    small scattering volume
   v     =    average voltage response
   Av    =    incremental voltage
   v^    =    average voltage response for diameter d
   Vs    =    volume sampled

   a     =    constant describing  power law function
   A     =    total drift emission
   6     =    specific drift fraction
   X     =    total mass density
   AX.;   =    mass of i"1 size element
*US. GOVERNMENT PRINTING OFFICE: 1973 546-312/125  1-3
                                        28

-------
   SELECTED WATER
   RESOURCES ABSTRACTS
   INPUT TRANSACTION FORM
                      ;. Report No.
               2.
  4. Title

     EXPLICIT CALIBRATION  OF THE PILLS  II  SYSTEM
   7. Author(s)
     Shofner,  Frederick M.
   9. Organization
                Environmental  Systems Corporation
  12. Sponsoring Organization

  IS. Supplementary Notes

     Environmental Protection Agency report number,
     EPA-660/2-73-011, September  1973.	
             3. Accession No.

             w

             5. Report Date
             6.
             8. Performing Organization
               Report No.

            10. Project No.
               16130 GNK
                                           11.  Contract/Grant No.
                                                                     13.  Type of Report and
                                                                         Period Covered
  16. Abstract

   The basic characteristics of the PILLS  I  and II Systems  are reviewed  and up-dated
   from  previous publications.   Emphasis is  given to the  explicit calibration of PILLS
   II with water droplets.   The operational  characteristics and accuracy of the mono-
   disperse particle generator are discussed.   General  procedures for  deriving the
   particle density distribution from the  measured voltage  density distribution are
   described.
  17a. Descriptors

   Cooling towers, drift measurements,  calibration, electro-optic instruments.
  17b. Identifiers
  I7c. COWRR Field & Group
  18. Availability
19. Security Class.
   (Report)

20. Security Class.
           pyederickM.  Shofner
21. No. of
   Pages

22. Price
Send To:
                                                         WATER RESOURCES SCIUNT i^lC INFORMATION CENTER
                                                         US. DEPARTMENT OF THE INTERIOR
                                                         WASHINGTON. D. C 20240
              institution  Environmental  Systems Corporation
WRSIC 102 (REV JUNE 1971)

-------