WATER POLLUTION CONTROL RESEARCH SERIES • 16130GNK10/71
DEVELOPMENT AND DEMONSTRATION
OF
LOW-LEVEL DRIFT INSTRUMENTATION
U.S. ENVIRONMENTAL PROTECTION AGENCY
-------
WATER POLLUTION CONTROL RESEARCH SERIES
The Water Pollution Control Research Series describes the
results and progress in the control and abatement of pollution
in our Nation's waters. They provide a central source of
information on the research, development and demonstration
activities in the Environmental Protection Agency, through
inhouse research and grants and contracts with Federal,
State, and local agencies, research institutions, and
industrial organizations.
Inquiries pertaining to Water Pollution Control Research
Reports should be directed to. the Chief, Publications Branch
(Water), Research Information Division, R&M, Environmental
Protection Agency, Washington, B.C. 20U60.
-------
DEVELOPMENT AND DEMONSTRATION
OF
LOW-LEVEL DRIFT INSTRUMENTATION
ENVIRONMENTAL SYSTEMS CORPORATION
Post Office Box 2525
Knoxville, Tennessee 37901
for
Environmental Protection Agency
Research Grant 16130 GNK
October 1971
-------
EPA Review Notice
This report has been reviewed by the Environmental Protection
Agency and approved for publication. Approval does not
signify that the contents necessarily reflect the views and
policies of the Environmental Protection Agency nor does
mention of trade names or commercial products constitute
endorsement or recommendation for use.
For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 - Price 65 cents
ii
-------
ABSTRACT
Instrumentation for measurement of low level drift from cooling towers
was developed. Emphasis was placed on the Particulate Instrumentation
by Laser Light Scattering (PILLS) System which is capable of on-line
measurement and, with incorporation of existing pulse height analyzer
and mini-computer equipment, complete on-line data reduction. Com-
plementary techniques of isokinetic sampling and sensitive paper
sampling were developed and field proven. Feasibility was demonstrated
for an infrared, in-line holographic system. The design principles
and engineering trade-offs for the PILLS, IK, and sensitive paper
techniques are described. Drift performance data are given for a small
air conditioning cooling tower unit, two large mechanical draft cooling
towers, and a natural draft tower.
This report was submitted in fulfillment of Contract Number 16130 GNK
under the sponsorship of the Office of Research and Monitoring,
Environmental Protection Agency.
m
-------
CONTENTS
Section Page
I Conclusions 1
II Recommendations 3
III Introduction and Background 5
IV Particle Instrumentation by Laser Light 7
Scattering (PILLS) System
V Complementary Drift Measurement Techniques 23
VI Summary of Drift Measurements Performed by 35
September 1971
VII Acknowledgements 51
VIII References 53
IX Appendix 55
-------
Figures
PAGE
1 PARTICLE INSTRUMENTATION VIA LASER LIGHT SCATTERING 8
2 PN LASER IN OPTICAL MOUNT 9
3 PILLS SYSTEM 11
4 ARTI LIGHT SCATTERING BY APIEZON WAX PARTICLES 15
5 TYPICAL PILLS DATA 18
6 PULSE HEIGHT ANALYZER DATA (CRT PRESENTATION) 21
7 ISOKINETIC SAMPLER SYSTEM 24
8 ISOKINETIC SAMPLER TUBES 27
9 PAPER STAIN 29
10 PAPER STAIN CALIBRATION 30
11 IN-LINE HOLOGRAPHY 31
12 AQUATOWER PARTICLE DISTRIBUTION 36
13 AQUATOWER MASS DISTRIBUTION 37
14 TOWER SEGMENTATION AND DRIFT RESULT 38
15 TOTAL DRIFT VERSUS PARTICLE SIZE 39
16 OAK RIDGE DATA SUMMARY 42
17 OAK RIDGE PARTICLE DISTRIBUTION 44
18 OAK RIDGE MASS DISTRIBUTION 45
19 PAPER STAIN HISTOGRAM 46
-------
TABLES
No. Page
1 Oak Ridge Cooling Tower Data Summary 41
2 Summary: Cooling Tower Observations 48
-------
SECTION I
CONCLUSIONS
Preparation of this report has been difficult because of the breadth
of the problem attacked and the number of solutions evolved. This
most challenging project has required an iterative problem definition
as we came to understand both the drift problem and the instrumental
techniques we have applied to low level drift measurements. Presen-
tation format and level for the developmental activities and results
was chosen to give in each of the individual sections the introductory
material, the data, and the detailed conclusions and recommendations
to be drawn therefrom. This section is confined to general remarks
on the drift problem as it is now understood and the applicability
of the instrumentation available.
(1) The most meaningful parameter for concluding remarks on the PILLS
system is the drift size range. There does not appear to be a lower
limit for the PILLS system with regard to total drift mass emission
level. The major effect of measuring lower levels of drift will be
increased sampling time to acquire sufficient samples for statistical
meaningfulness. The influence of other parameters such as air flow
velocity, humidity, temperature, fog level, mineral concentration of
the circulating water, etc., are discussed.
The present prototype PILLS system is entirely satisfactory for drift
measurements on mechanical draft towers and spray pond coolers with
low fog level. The minor improvements in the instrument which are
presently being implemented will permit measurements from 50pn to
lOOOjum or larger. It is further concluded for the state-of-the-art
drift measurement that measurements can be made immediately beyond
approximately lOO^im for fogging conditions which have been encountered
in our experience.
(2} For situations where fog precludes measurements below, for example,
100/jm, the double scattering volume PILLS system should be developed.
Our preliminary investigations using rather unsatisfactory components
demonstrates feasibility for relatively easy reduction by a factor of
two to three the minimum particle size measurable in the presence of
typical fogs. Thus, even with fog, it is entirely feasible with the
double scattering volume instrument to record the particles from 50yjm
up. It is our present conclusion for hyperbolic cooling towers that
the size range 50 to ISOjum embraces the bulk of the drift emission.
The drift emission for particles less than 50jum is even more negligible
in terms of percentage contribution to the total emission for mechanical
draft towers than for natural draft units.
(3) The independent, complementary drift measurement techniques 6f
isokinetic and sensitive paper sampling deserve mention in conclusion.
For estimates of integrated drift, and direct measurements of a par-
ticular mineral effluent, IK measurements are useful. For estimates
of particle size distribution, with emphasis on the smaller diameter
range, sensitive paper is useful. Both require lengthy data reduction.
1
-------
Simultaneous application of the three complementary techniques was
regarded as particularly important for initial studies even though the
scope of effort was necessarily increased.
(4) Generally, it may be concluded that the responses of cooling tower
manufacturers to environmental impact interests will have the effect of
driving down the drift emission level via improvements in drift eliminator
efficiency. The most effective reductions will be elimination of larger
particles and it may therefore be expected that as drift levels go down
so will the mean particle diameter. If drift emission levels of order
0.005% can be shown to have acceptable environmental impact, it can be
expected that cooling tower utilization will increase. Brackish
installations place severe requirements on drift elimination and
measurement and this application should be studied with priority.
(.5) Finally, our experience with this challenging and many-faceted
problem has defined a clear and increasing need for a laboratory simulation
facility for examination of cooling tower performance. One of the most
severe difficulties encountered in instrumentation development was making
measurements on operational towers although the field experience gained
was valuable. The country needs a government-sponsored facility where
drift eliminator designs and other cooling tower parameters can be
studied with the repeatability and cost effectiveness of a laboratory
environment.
-------
SECTION II
RECOMMENDATIONS
In addition to the minor modifications and explicit calibration now
in progress, the extension of the PILLS technique toward smaller drift
particles (below approximately 100pm) in typical cooling tower fog
backgrounds is the highest priority improvement in instrument
performance per se. A factor of two to three reduction in minimum
measureable diameter will probably permit measurement of all but a
few percent mass emission even for hyperbolics. It is further
recommended that pulse height analyzer-mini-computer electronic data
processing equipment be employed for data acquisition and reduction,
because the smaller, more frequently occurring drift components will
become increasingly important.
However, the present PILLS system is adequate for measuring particles
larger than # lOOjum in typical mechanical draft cooling tower fogs.
Thus an important recommendation is to utilize the PILLS system with
the complementary isokinetic and paper stain techniques for a careful
and complete drift characterization program as soon as possible and
simultaneously with drift outfall measurements, preferably on a
brackish or simulated brackish installation. In addition to
environmental impact assessment, an important part of such a program
would be evaluation of improved-drift eliminator designs.
-------
SECTION III
INTRODUCTION AND BACKGROUND
Wet cycle cooling towers provide an attractive alternative for dispersal
of large quantities of waste heat. These towers operate on the principle
of heat transfer from hot circulating water to an atmospheric air flow.
The heat transfer takes place via the large surface area presented to
the air by a profuse distribution of small particles as well as via
heat transfer fins which are the so-called fill material.
Cooling towers discharge heat into the environment in the primary form
of latent heat as a result of the evaporative cooling process. Visual
observation of cooling towers in operation also, under most conditions,
indicates a substantial fog plume which results from condensation of
the increased water vapor content of the air which went through the
tower. It can be assumed that the vapor as well as the fog constitutes
pure water and therefore does not generate a contamination pollution
problem. However, the environmental impact of the hot, moist plume is
important and this thermal pollution problem is under study.
The environmental problem addressed by this effort relates to the fact
that the air in wet cycle cooling towers is in direct contact with the
circulating water from which the heat is taken and thereby provides the
opportunity for some of the particles generated in the process to be
entrained in the air flow and carried through the tower and into the
environment. It is known from simple Stoke's flow theory as well as
more representative experiments that air flow velocities typically
encountered in cooling towers are capable of carrying particles of
several hundred microns into the environment. These particles, termed
drift, constitute a pollution problem since they will have to first
approximation the concentration of minerals which is found in the
circulating water. For the important application to brackish water
installations with high mineral content, the mineral residue can con-
stitute a severe environmental impact problem if the level of tower
drift emission is high. An example is illustrative.
A number that has become associated with the drift phenomena is 0.2%;
i.e., the drift emission rate is 0.2% of the circulating rate. This
number originates with early developmental studies associated with cooling
towers perhaps even before efficient drift eliminator designs were put
into practice. It is easily shown that a very large cooling tower having
500,000 gallons per minute for its circulation rate would generate a salt
deposition of 7.5 tons per hour if one uses the 0.2% fraction for drift
emission and assumes the normal salinity of 30,000 ppm for sea water.
Clearly, this environmental impact is severe.
It has been presumed for some time that modern cooling towers with efficient
drift eliminators have far lower levels of drift. However, prior to
mid-1970 there existed no proven techniques for measuring low levels of
drift. In early 1970, Environmental Systems Corporation addressed the
problem of measuring low level drift as a result of interests expressed
by The Marley Company. Feasibility was demonstrated on a very small scale
-------
for laser light scattering and holography techniques by the mid-part
of 1970. Shortly after feasibility of these techniques was demonstrated
to a number of private utility and EPA personnel a proposal was sub-
mitted to the Environmental Protection Agency for the purpose of providing
a demonstration grant to the ends of showing feasibility to those
concerned with the drift problem. Funds were approved and work was
formally initiated for the demonstration grant in March 1971. The
development culminated in a demonstration experiment performed on a
mechanical draft cooling tower located at the Oak Ridge National
Laboratories in Oak Ridge, Tennessee on September 28, 1971. Approximately
eighty persons from across the country intimately concerned with the drift
problem were in attendance of this special demonstration.
Before proceeding with the discussion of the instrumentation developed,
it is worth reviewing the requirements which were placed on the drift
measurement system as they were understood in 1970. At that time it
was presumed that the drift particles were several hundred microns in
diameter. It was further estimated that the drift percentage was at
least one order of magnitude lower than 0.2% and that the concentration
of the mineral content in the drift particles was identical with that of
the circulating water.
Numerous instrumental techniques were considered which had potential.
In addition to the obvious considerations of accuracy, reliability and
cost, it was decided to pursue the technique(s) which had the potential
to develop into an on-line monitor and, more important, technique(s)
which would resolve the particle size distribution. Particle size
distribution is of fundamental importance because the dynamical behavior
of the contaminating drift particles is different for different sizes,
both in the tower and when they leave the cooling tower and are placed
in the environment. Large particles will fall out more quickly.
Sufficiently small particles having sufficiently low mineral concentrations
may even evaporate to residues which are airborne essentially indefinitely,
Of course, the total amount of particulate matter can be obtained by
integration over the particle distribution.
With these factors in mind, two electro-optic techniques - that of laser
light scattering and in-line holographic recording - seemed the most
viable. The scattering technique has the most potential for a truly
on-line measuring capability and the most potential for extension to
smaller and/or larger particles than the few hundred microns assumed.
Although holography generates a permanent record of the particle field,
it can be argued that the two-step holographic recording and reconstruction
process is little better than photomicrography using a pulsed light
source for this application. It was concluded after feasibility had
been demonstrated for both to emphasize Particulate Instrumentation by
Laser Light Scattering, or the PILLS system. The next section discloses
the principles of this system. The fifth section discloses the principles
of two complementary techniques which we have employed in the field to
measure drift as well as a disclosure of the holographic technique and
compares all these techniques for relative merits. Section VI summarizes
drift measurements which were performed in preparation for the Oak Ridge
demonstration.
-------
SECTION IV
PARTICLE INSTRUMENTATION BY LASER LIGHT SCATTERING
(PILLS) SYSTEM
BASIC PRINCIPLES
The essential operating principles of the PILLS system can be efficiently
explained with use of Figure 1. A laser emits pulses of radiation with
a wave form having the shape as shown in inset A. Peak powers of 12
watts of infrared radiation at 9040 Angstroms are emitted by a gallium
arsenide junction laser diode. A most important innovation employed in
the fabrication of the PILLS system has been utilization of the gallium
arsenide laser diode, which permits design of a very compact and reliable
device. Figure 2 shows a photograph of the laser diode mounted in the
optical support as well as an additional diode shown on the machinist's
scale for size comparison. The copper tube in front of the diode lens
directs dry N2 which is used to continuously purge both the laser and
detector components.
To our knowledge this is the first utilization of the laser diode in this
application. The peculiar emission properties of the laser diode necessitate
collection of the light to realize a collimated beam. Although in some
of the initial work a rather complicated collimation system of spherical
and cylindrical lenses was used, it proved acceptable and far more con-
venient to.obtain essentially a collimated beam with a single microscope
objective as shown in the mount in Figure 2.
Referring again to Figure 1, a scattering volume Vs is defined by the
intersection of the laser beam (3x4mm) and a detector acceptance cone.
The acceptance cone is defined in the present configuration by two
apertures AT (64mm), the entrance aperture, and A£ (25mm), the0exit
aperture which precedes a narrow band interference filter (100A) and a
photomultiplier detector (S-| photocathode). When a particle of diameter
d is within Vs when a laser pulse occurs, light is scattered through the
collection system to the photodetector which provides a pulse of current
proportional to the scattered light intensity. This current pulse is
amplified by an integrated circuit preamplifier which broadens the pulse
to a width more convenient for processing (tflus). The 1C is followed
by a driver circuit suitable for launching the pulse into coaxial cable.
The essential information is in the height or amplitude of the pulse
generated at the output of the coaxial cable at point B. Beyond point B
the essential information may be processed in a variety of ways depending
upon the circumstances of the particle field being examined - this is
discussed later.
Before examining the characteristic wave forms that appear at B it is
most important to present two essential design features of the PILLS
system relating to the scattering volume Vs. First, the scattering
volume is sufficiently small that particles in the range of interest will
exist in the scattering volume only one at a time. In more mathematical
-------
oo
NARROW BAND FILTER
PHOTODETECTOR
AMPLIFIER
COAX
SCATTERING VOLUME
PL
12W
ft«-At^3ms
t = I30ns
-MtT-11-
T[
BACKGROUND NOISE
-PULSE HEIGHT ANALYZER
MINICOMPUTER; READ-OUT
OSCILLOSCOPE ;
TAPE RECORDER;
DIGITAL COUNTER
AD, gm/sec-jjm
6=0.005 °70
FIGURE 1
PARTICLE INSTRUMENTATION VIA LASER LIGHT SCATTERING
ENVIRONMENTAL SYSTEMS CORPORATION
KNOXVILLE , TENNESSEE
-------
FIGURE 2
PN LASER IN OPTICAL MOUNT
-------
terms, the probability of finding a particle in the range of interest
in Vs at any instant of time is significantly less than unity. This
has the substantial consequence that one is dealing with the phenomena
of single particle scattering. When light reaches the detector from a
collection of particles of different sizes, the sizes of the individual
particles cannot be deduced from a simple amplitude measurement. A
related design consideration is that the scattering volume transverse
dimensions and the pulse repetition rate are so chosen that a given
particle will not be counted twice by two successive laser pulses. The
pulse repetition rate f = I/At and the flow velocity v through the
scattering volume must satisfy the relation
vAt>dt (1)
where d^- is the appropriate transverse dimension of Vs.
The second fundamental point is that the scattering volume is clearly
presented to the flow rather than the alternative design concept of
taking particulates from the flow and presenting them to an internal
scattering volume. The latter concept is viable when the particles are
small and dense. It is not viable when the particles are as large as a
few hundred microns and occur in a given volume of the order of one
cubic centimeter relatively infrequently. Thus, the external scattering
volume presents a minimal aerodynamic disturbance to the dynamic particle
field.
Examining now the wave form that appears at point B shown in inset B of
Figure 1, a typical yield from the scattering volume is sketched. There
are small background pulses which can occur each time the laser emits.
These occur if there are many small particles in Vs. If few small particles
are present, or better, if their density and size is sufficiently small,
then the lower sensitivity of the instrument is determined by detector/
amplifier noise. Turning to the signal of interest, a random distribution
of voltage pulses larger than the background/noise is shown occurring at
a rate that is significantly less than the repetition rate of the laser.
Also, there are more small pulses than large pulses. This train of pulses
represents what has been found to be descriptive of the particle distri-
butions observed in actual cooling towers.
It must be clearly noted that the design considerations of the instrument
relate to the particle distribution and the flow in which these particles
are entrained. It is also important that the mechanical and electrical
design considerations of the system take into account the extraordinarily
hostile-hot, humid and turbulent-environment which can be encountered in
cooling tower flows.
Figure 3 shows the final prototype configuration installed on the traverse
rail above the mechanical draft tower used for the Oak Ridge Demonstration.
In this configuration, the electro-optical properties of the instrument
permit measurements from approximately 80pi to over 1,000/jm in subdued
light and in a light fog background with a typical sampling time of
approximately thirty minutes per station. The instrument was so configured
because of knowledge of the particle distribution being examined in the
10
-------
FIGURE 3 PILLS SYSTEM
-------
mechanical drift tower chosen for the demonstration; this knowledge
was accumulated via "cut-and-try" experience with the PILLS system
itself.
Design variations, trade-offs and desirable improvements w^'ll be presented
below. The remainder of this section covers the analytical aspects of
the PILLS system.
The basic response of the PILLS instrument to particles within Vs is
to produce a voltage v
v = Kd2 (2)
where K is the calibration factor and d is the particle diameter. The
theoretical and experimental evidence for this relation is given in "Cali-
bration." . The nature of the instrument is to sample the volume V§
at a rate F = I/At times per second. Let a successful event be defined
as obtaining a voltage pulse v^ (within a range AV-J ) and the number of
these events per second be f(v-j). The probability of a successful event
is f(v-j)/F. Let the number of particles per unit volume having df
(within a range d^ ) such that v-j = Kd-j2 be n-j . Thus the probability
of the same successful event is n^V.. Equating these probabilities gives
the basic describing equation for the instrument:
"iVs = f(v-j)/F = f(Vi)At (3)
n^ plotted versus d gives the particle histogram, n-j /Ad gives the particle
distribution function. The mass per unit volume associated with the ith
size particles is
where s is the specific weight. The total mass per unit volume is
(5)
When plotted against particle diameter, Am-j gives the mass histogram.
.j versus d. is the mass distribution function.
The flow rate is simply ^
Q = VA |e4 (6)
where V is the flow velocity and A is an area normal to the flow velocity
vector. The quantity of material carried by the flow is
D = mVA (7)
Isec
12
-------
For the PILLS instrument the area substitutedjs At, the projection of
the scattering volume onto a plane normal to V.
A basic data output presentation for the integrated mass flux is to give
for each "j" measuring point the quantity D/At:
The drift emission in gms/sec associated with the "j" area An- is
J
and the total drift emission in gms/sec is
(10)
Equations 8, 9, and 10 are easily converted to give, respectively, the
point, area, and total mineral emissions by multiplying both sides by
the total mineral concentration of the circulating water c or for a
specific mineral such as Na, Ca, Mg, by C. , the concentration of the
mineral component.
Finally, the drift percentage is
= x 100% (11)
R
where R is the circulation rate in gms/sec.
CALIBRATION
In principle, the relation between particle diameter d and PILLS system
output voltage v may be predicted by application of the Mie scattering
theory. The theory is rather complicated, even though considerable
reduction in the mathematical complexity results from the fact that the
drift particles are much larger than the wavelength of the laser radiation
However, there are too many scattering particle and electro-optic system
parameters whose values are too inaccurately measureable to permit an
accurate functional description of the v-d characteristic. It is more
practical and accurate to introduce particles of known size and com-
position into Vs and measure the v-d characteristic explicitly for a
given system configuration.
Unfortunately, time pressure did not permit the evolution of an explicit
calibration apparatus. Calibration was realized for the instrument by
comparing its data with a complementary measurement technique, isokinetic
13
-------
sampling, described below. With isokinetic sampling as with the PILLS
system after integration over the particle distribution, one measures
the drift water mass per unit volume. If the Kd^ relation of Equation
2 is assumed, simultaneous PILLS-IK measurements yield K. The cali-
bration analysis is given in the appendix.
p
Justification of the v«c d relation is based on theoretical and
experimental evidence. Rather than present Mie theory results, a simple
physical model may be employed as a plausibility argument. The trans-
parent, spherical drift particles collect and converge the incident
radiation analogously to a thick spherical lens. The "focal length"
may be shown to be proportional to the particle diameter. Thus the
collected and "focused" light diverges with an angle independent of d.
For water particles having index of refraction of 1.33, the angle Q^M
between the incident light direction and the scattered light direction
is approximately 45°. A simple experiment was performed with a widely
disperse particle distribution of drift water particles which sub-
stantiated this model in that little scattered radiation was detected
beyond £50° .
o
The generality of the v«cd relationship is further supported by the
data of Figure 4. The scattering particles for this experiment were
semi transparent Apiezon wax particles melted onto a glass flat. The
laser used was a pulsed argon ion device filtered to yield «0.5W
of green light at 5145 Angstroms. The detector was a standard silicon
photodiode. (Clearly, a constant K derived from Figure 4 is not
directly applicable or relatable to the PILLS system.)
This evidence supports the v<£d^ assumption necessary for implementation
of the IK-PILLS implicit determination of the calibration factor K.
Considerable confidence in this implicitly-determined value is generated
as a result of the agreement between the three complementary drift
instrumentation techniques when applied to the Oak Ridge Tower (See
Section VI.) Of particular importance is the agreement between the
PILLS and sensitive paper data in the location of distinguishing features
of the particle histogram.
Notwithstanding the general agreement between independent measurements
and the confidence in K implied or the evidence supporting the vocd
relation, it is highly desirable to explicitly calibrate the PILLS system
with particles of known size and composition. This activity is being
pursued as an extension of the present effort and will be reported
separately. These experiments are expected to confirm the v = Kd2
relation as well as provide data on the effects of high mineral
concentrations as encountered in brackish, water applications of cooling
towers .
A monodisperse stream of particles from sizes as small as 15^im through
500/jm can be generated by vibrating capillary techniques (1). Smaller
particles can be produced via vibrating orifice techniques (2). Both
exploit the inherent hydrodynamic instability of a stream of liquid
moving through a viscous medium. These techniques require a substantial
amount of experience and a nominal amount of equipment and are well-
documented in the literature.
14
-------
t/l
OJ
0.01
100
Voo d2
9 = 26.5
i ' I I I I—L
1000
d, )U m
FIGURE 4
AR n LIGHT SCATTERING
BY APIEZON WAX PARTICLES
15
-------
It should be pointed out that it is well known that there are available
mono-disperse glass beads which are in the size range of interest for
the calibration of the instrument. Such particles were used to obtain
a rough estimate of the calibration factor but were not used for the
primary calibration. They may not adequately simulate the water
particles because of the differences in index of refraction and surface
conditions.
SIGNAL PROCESSING
Consider again the wave form presented at point B in Figure 1 in which
the particle's size is represented in the voltage amplitude. The pulses
occurring at random times and with random amplitudes must thus be
analyzed to give the particle distribution and integrated mass of the
drift phenomena under examination. Field experiments with preliminary
PILLS prototypes on mechanical draft towers indicated that drift
particles could be smaller than 50jjm and larger than 500jum. However, _
very few particles greater than BOOjum were observed and a small fraction
of the drift emission mass was less than 50/jm. It was thus assumed
that the most important size range was approximately 50-500^im.
Preliminary experience also indicated that the typical fog background
present under most cooling tower operational conditions yields a quantity
of light that is proportional to the scattering volume and whose
magnitude may be larger than the yield of a single drift particle.
Unfortunately, the fog voltage cannot be completely subtracted out because
of random pulse-to-pulse variations in the laser intensity and,
consequently, the fog effively sets a lower limit on the drift particle
size which can be measured. One way to reduce the fog problem is to
reduce the scattering volume even below the limits wherein single
particle scattering obtains for the drift particle distribution. The
trade-off in this reduction is that one must wait for a longer period
of time to collect enough voltage samples to be statistically meaningful.
Cut and try iterations on the instrument configuration design for the
particular particle distribution and fog level associated with mechanical
draft towers led to a compromise of a minimum measurable particle of,
typically, 100/jm and a data collection time of 30 minutes. In the
absence of fog, the system - in its present configuration - is limited
by electronic noise to 80um. This lower limit can be easily reduced to
a few microns by reconfiguration and optimizion of the laser-detector-
amplifier components.
The rate at which data pulses are collected with the present configuration
is, for drift particles larger than approximately one hundred microns,
one count approximately every second for a mechanical draft tower.
Consequently, it was convenient for data collection to set the trigger
level on a precision oscilliscope to a voltage value corresponding to
a particle diameter of, for example, lOOjum and then measuring the time
required to collect ten pulses corresponding to particles larger than lOOjum.
16
-------
These pulses were recorded in a multiple exposure fashion with the
oscilloscope camera. Usually, several photographs having ten pulses
each are made. Figure 5a shows a typical photograph thus obtained.
(Figure 5b is a real time recording wherein each laser pulse is
identifiable and was used for a particle distribution having roughly
10-3 times larger mass flux.) This recording technique is suitable for
low count rates and has the inherent advantage of permitting measurement
of the pulse height directly from the oscilloscope photographs.
In several situations to which the instrument was applied, the fog
level was low enough such that the counting rate for small particles
contributing significant mass to the drift emission was sufficiently
rapid that stopwatch time measurements proved inadequate. For one of
those situations, it was possible to utilize a digital counter which
counted the number of pulses exceeding a preset level in a given period
of time. Although the more frequently occuring particles were recorded
with this technique, it was still necessary to sample for approximately
thirty minutes to obtain the same statistical accuracy for the large
particles as before.
These considerations reflect the engineering trade-offs made to
accomodate specific particle distributions as well as discovery of
these distributions and dealing with the fog problem. However, the most
important advantage for the PILLS system is that the random pulse train
generated at point B is ideally suitable for analysis by available pulse
height analyzer equipment which has been developed primarily as nuclear
instrumentation. This is described more fully in a later section.
SUMMARY OF PILLS DEVELOPMENT AND RECOMMENDATIONS FOR FURTHER WORK.
The state of development of the PILLS system at the writing of this report
may be summarized as follows. The system is operational and has been
field proven. It has been configured for a field-determined particle
size distribution and is presently capable of measuring under conditions
of minimal fog from approximately 80 to over l,000jum. The size range
may be extended both toward smaller and larger particle sizes giving
due consideration to the particle distribution to be measured and the
background fog. It is estimated that a lower limit of a few microns
can be achieved. Particles larger than 1,000|im do not appear to be of
significant interest with regard to drift instrumentation although they
could be easily measured with the PILLS system.
The usefulness, design adequacy and field worthiness as well as limitations
of the PILLS system are best supported by the data on operational cooling
towers reported in Section VI and no further performance claims are
given here. It is important to state, however, that the present con-
figuration is directly suitable for drift measurements on all cooling
towers or spray ponds for which particles larger than, typically, lOOp
constitute the principle concern. This lower limit is set by instrument
configuration in conjunction with fog background usually encountered with
cooling towers but can be reduced.
17
-------
DOWN-WIND
(SAMPLED)
UP-WIND
(REAL TIME)
FIGURE 5
TYPICAL PILLS DATA
18
-------
However, as should be expected with a developmental activity, there
are several important areas which will profit with further effort.
These are listed in the following discussion in the order of their
importance to the drift measurement problem as it is now understood.
Numerous minor improvements, such as reduction of light background,
cooling of critical components, and design refinements, will yield
greater convenience in data interpretation and in field operation and
are straight-forward in reduction to practice.
(1) Explicit Calibration. The realization of vibratory capillary or
orifice particle generators needs to be accomplished and applied to
generation of the V0 - d characteristic of the device in a specific,
optimized configuration. It is not expected that the relationship will
depart materially from Kd2 but nevertheless needs explicit verification
for water particles closely simulating actual drift. This activity is
in progress.
(2) Fog Suppression. The fog yield can be reduced by reduction of the
scattering volume with the cost of increased sampling time. Thirty
minutes now typical for a sampling period is regarded as long when the
basic sampling rate capability of the instrument and, specifically,
pulse height analyzer processing equipment, are considered. But larger
scattering volumes cannot be used because the fog yield can be as large
as that from a small drift particle. This background cannot be simply
subtracted out for the single Vs because the fog yield fluctuates.
Some preliminary design consideration and experimentation has been
given to a so-called double Vs instrument whose principle is as follows.
The laser beam proceeds as in Figure 1 but illuminates two scattering
volumes. The size of the scattering volumes and the overall system
transfer gain are adjusted such that for a homogeneous fog particle
distribution - but no drift particles - an identical output voltage is
realized. These two voltages are then fed to a differential amplifier.
If no particle is present in either scattering volume a reading of zero
is realized. If a particle is in one of the scattering volumes then the
voltage realized for that particular channel is larger and the difference
is proportional to Kd2. This system has the capability of resolving
smaller particles in the fog background as long as the drift particles
are substantially larger than the fog particles whose mean diameter is
probably <5/um for cooling towers b'ecause the droplet formation time is
much shorter than for meteorological fogs whose mean diameters are
typically 10-20^im (3). The scattering volumes can be designed larger
(but not large enough to violate the single particle occupancy require-
ment) with the important result that the data collection rate can be
increased. Thus, a double Vs instrument permits resolution of smaller
particles and more rapid data collection.
The initial experiments which were performed met in marginal success
because photomultipliers were used which had very poor noise characteristics
The noise in each channel of the detector circuits is uncorrelated and
thus cannot be directly subtracted out whereas the fog yield is obviously
19
-------
correlated and can be subtracted. The preliminary experiments do
indicate feasibility for photomultipliers or other detectors having
significantly lower noise. These - or alternative - design con-
siderations for fog yield suppression must be pursued as the interest
in particle sizes shifts toward smaller particles which are present
in a large background of fog.
(3) Pulse Height Analyzer/Mini-Computer Processing. Pulse height
analyzer instrumentation generates the voltage height distribution by
sub-dividing a voltage range of typically 0-10 volts into, for example,
512 channels or voltage increments and then storing in each channel's
digital memory the number of counts observed in a sampling period.
In some units, 106 counts can be stored in each channel. The number
of counts per channel then can be directly displayed in an analog
presentation on a CRT as the voltage histogram or the voltage distri-
bution function. But because of the digital nature of pulse height
analyzer equipment, the count data are already coded in a digital
form and are suitable for presentation to a plotter or a teletypewriter
system which will plot or print out the number of pulses per channel.
Most important to the automatic processing consideration is that the
output of a pulse height analyzer, being digital, is suitable for direct
insertion into digital computation equipment, especially the new
generation of mini-computers. Thus, the complete integration of the
laser scattering system with a pulse height analyzer appropriately
interfaced with a mini-computer could produce a readout as shown in
inset C of Figure 1 wherein the parameter of interest, grams per second
emitted by the tower in a given micron size range is presented on, for
example, a cathode ray tube or on a digital plotter. Simultaneously
with a presentation of the particle distribution the mini-computer can
be programmed to generate the integrated amount of drift water leaving
the tower per second and/or to produce a direct visual display of the
drift percentage. Clearly, this system could be employed in an operating
power station to provide an automated readout of drift performance of
a given tower. It could be used for rapid data acquisition for tower
certification. Still another application would be as an integral part
of a laboratory cooling tower simulation facility which would permit
convenient testing of drift performance.
It is to be recalled that the capability of providing an on-line readout
of the drift parameters of interest was a prime consideration in choosing
the PILLS system. A pulse height analyzer was incorporated into the
Oak Ridge Demonstration of September 28 and Figure 6 shows a CRT
presentation of the voltage height distribution.
20
-------
LINEAR
LOG
FIGURE 6
PULSE HEIGHT ANALYZER DATA
(CRT PRESENTATION)
21
-------
SECTION V
COMPLEMENTARY DRIFT MEASUREMENT TECHNIQUES
Rather than invest all our effort in the development of the PILLS
system, it was considered advantageous from an engineering viewpoint
to develop three other drift measurement techniques to complement the
PILLS system and to provide thereby a check on its accuracy and applicability
Several field tests were performed during the course of this development
and it was possible for us to develop two of the other techniques beyond
laboratory feasibility to the point of field use. In the order that they
have been utilized and developed they are: (1) isokinetic sampling, (2)
sensitive paper sampling, and (3) in-line holography. The principles of
each of these will now be discussed.
ISOKINETIC SAMPLING
Operational fresh water cooling towers have a substantial mineral content
in the circulating water, generally greater than a few tens of ppm of
elements such as calcium and magnesium. (Of course, the concentration
of sodium and chlorine for brackish water towers would be much higher,
of the order of 30,000 ppm.) Analytical chemical techniques, particularly
atomic absorption spectroscopy, permit the determination of material
concentrations as small as a few parts per billion for magnesium, for
example. Thus, if one isokinetically samples a volume of air moving
through the tower and extracts from it the drift water and from the
drift water the mineral residue and assumes that the concentration in
the drift water is identical with that of the circulating water, then a
measure of the trapped mineral residue yields the amount of drift material
in the sampled volume. From this the total drift emission is realized by
numerical integration via an appropriate area segmentation of the tower
based on the measured air velocity profile.
The terminology "isokinetic sampling" relates to the measurement of air
flow in the following way: air is drawn into a sampler tube having the
kinetic energy of a fluid element identical with that which existed
had the tube not been there. If it is assumed that the density and
temperature of the air do not change upon being drawn into the tube, this
reduces to maintaining equality between the velocity of the air flow into
the tube and the flow velocity in the absence of the tube at the point
of measurement. Procedurally, one measures the mean flow velocity at
the test point with, for example, a vane anemometer, and then adjusts the
volumetric flow rate through the sampler such that the mean velocity in
the sampler tube inlet is equal to the flow velocity. Figure 7 indicates
the elements of an isokinetic sampling system including the vacuum pump,
integrating gas flow meter and sampler head.
Isokinetic sampling of dry particulate matter as, for example, in power
plant stacks, has been realized for several years and is a standard and
23
-------
.a..- o
11/2" so.
TYGON TUBJNG (20' x 3/e" ID)
HEAVY WALL
FLOW METER ,—METERING VALVE
Lllhlilihlll
1
fly
-PUMP
, — MOTO
L-f'
R
AC
FIGURE 7
ISOKINETIC SAMPLER SYSTEM
-------
accepted technique. These sampling heads are generally constituted of
a 1/4" to 1/2" entrance diameter probe tube followed by a filter paper
system which collects with high efficiency particulate matter larger
than a few microns. As a matter of interest, the so-called "collection
or trapping efficiency" of paper type filters is minimum for particles
of approximately O.Sum diameter. Larger particles are trapped by
inertical impact and smaller particles are driven onto the filter fibers
by random Brownian motion. A standard filter paper has been Mhatman No. 1
and^can have trapping efficiency for dry lOOjjm particles approaching
We examined the straight-forward extension of these sampling techniques
to drift measurement and assumed that the mechanism of trapping would
be particle impingement and subsequent evaporation of the water, leaving
the mineral residue on the filter. For analysis, the filters are then
burned in an oxygen atmosphere; the burned residue is dissolved and
finally analyzed by atomic absorption techniques. These measurements
proved inadequate and inaccurate for two basic reasons. First, the
amount of magnesium and calcium in the filter paper is of the order of
a few micrograms and varies substantially from paper to paper and even
within the same paper. Consequently, one must sample for a substantially
long time to acquire a quantity of residue material significantly larger
than the background of the filter.
Second, it was found that when the filters were introduced to the moist
hot atmosphere of the cooling tower that questions arose with regard to
the holding efficiency of the filters. It is useful to note at this
point that if the quantity of drift water per cubic foot is defined as
one unit, then by orders of magnitude the quantity of water in the form
of fog will be 10 units and the quantity of water in the form of vapor
will be more than 100 units. Thus, any minor fractional condensation of
the vapor or collection of the fog could redissolve the trapper material
and permit it to be entrained in the flow, thus being lost from the filter
This mechanism has been proposed because in a number of experiments when
there were two and, in some cases, three filters in series more material
on the second filter than the first was found. This was not the case,
however, in laboratory experiments with drift particles when the filters
were operated without the fog and moist environment where the trapping
efficiency was measured to be approximately 90%.
Still another undesirable factor in the behavior of the paper samplers
was the fact that the quantity of material collected appeared to be
consistently higher than the value determined by the PILLS system even
when it was taken into account the fact that the PILLS system recorded
only that part of the particle distribution function beyond, for example,
lOO^im. The inadequacy of extending this standard technique for dry
particulate sampling to aerosols or large liquid particle sampling such
as drift led us to consider other techniques or refinements. The appli-
cation of standard cyclone techniques which operate by centrifugal
separation was attempted. The results were slightly more encouraging
25
-------
but still led to values larger than those determined by the PILLS system.
One mechanism that would lead to the larger values in both cases would
be field contamination. Another mechanism which would lead to more
collected material is entrainment of liquid collected on the outside of
the tube and drawn in to the inlet. In a cooling tower environment
any metal or solid objects placed in the flow will literally be dripping
with water.
To circumvent these difficulties, a new type isokinetic sampler was
conceived and developed by Environmental Systems which circumvented by
design some of these most important problems. The final design is
shown in Figure 8 along with a newer design using glass wool fill material
which has yet not been used in field situations. Its background levels
for Mg and Ca are roughly three times that of the bead tube but its
trapping efficiency should be higher. No tests on other background
levels have been run.
The principles of the bead-filled tube are as follows. The resistance
wire on the outside of the tube provides a heat input of 60-70 watts.
The heat transfer slug forces the air to flow near the walls for more
effective heat introduction to elevate the temperature of the air. The
hot air in turn heats the bead column and other glass surfaces. The
drift particles entering isokinetically then impinge upon the hot glass
surfaces and evaporate to dryness leaving their material residue.
The advantages of this construction are that the sampling volume rate
can be substantially higher than for the paper filter (I.D. = 25mm)
and, more important, the glass tubes can be cleaned to have a much
lower and more predictable background contribution.
The addition of the heat external to the tube provides two further
improvements. First, it heats the air such that condensation cannot
occur thereby defeating the mechanism for loss by re-dilution and
re-entrainment of condensate within the tube. Second, the heat outside
the tube prevents the condensate from forming on the tube and then being
entrained in the flow. In operation in an actual cooling tower environ-
ment the tubes thus appear dry, both inside and outside, and feel warm
to the touch even at the top of the tube.
A substantial number of performance experiments were run on these tubes
in the laboratory and in the field. The trapping and holding efficiency
of the tubes for the dried residue is typically better than 90%. More
important, the tubes agree reasonably well with the integrated quantity
of material predicted by the PILLS system when they are run simultaneously
and under compatible conditions.
SENSITIVE PAPERS
A technique which has been widely used for measurement of particle size
distributions external to cooling towers has been the sensitive .paper
26
-------
ro
-•4
FIGURE 8
ISOKINETIC SAMPLER TUBES
-------
approach. According to Chilton (5), the standard preparation of the
papers proceeds by soaking them in a 1% solution of potassium ferricyanide,
allowing them to dry, and then dusting them liberally with finely
ground ferrous ammonium sulfate. The papers thus prepared are pale
yellow in color and when impinged by a liquid droplet form a blue stain
which is clearly distinguishable on the yellow background and whose
diameter is related to the diameter of the impinging particle. The
paper itself is important. Considerable improvement resulted by using
milli-pore membrane filter paper rather than the Whatman No. 1 paper
used by Chilton. With the latter paper, the fibrous structure is so
coarse as to obviate measuring particle stains which are smaller than
30 or 40um.
Figure 9a shows a photo-micrograph showing both the sensitized paper and
a teflon sheet placed over a piece of the gridded paper upon which a
spray from an atomizer has been applied. Figure 10 shows the calibration
realized by the following technique. The spray aerosol sprayer produced
a particle distribution on the teflon surface. A piece of the unexposed
sensitized paper was then laid on top of the teflon sheet. Before the
paper was placed over the distribution, a photo-micrograph was taken
of the particle field. After the stains were made, the paper was
turned over and another photo-micrograph was taken. By this means, a
one to one correspondence between the original particle field and its
inverted image on the sensitive papers was generated. These data are
shown on the calibration curve of Figure 10. Figure 9b shows typical
data which were collected at Oak Ridge for comparison with the isokinetic
and PILLS system.
It should be noted that the sensitive paper measurement suffers from the
fog background problem in much the same way that the PILLS system does.
The large quantity of fog and any condensate from the even larger quantity
of vapor produces an exposure of the sensitive paper precluding large
exposure times. It can therefore be appreciated that the sensitive
paper must be exposed for a brief period of time (typically 1-2 seconds)
and consequently will provide less information about the large drift
particles which occur far less frequently than small particles. In
this sense the laser and the sensitive paper are complementary.
IN-LINE HOLOGRAPHY
It was mentioned earlier that two electro-optic techniques were considered
at the outset. For completeness, a discussion of the feasibility
demonstration of in-line holography using the gallium arsenide laser is
included. To our knowledge, this effort produced the first instrumentation
application of in-line infrared holography and the second hologram ever
to be made with infrared laser light (6). Figure 11 shows schematically
the typical in-line holographic recording and reconstruction arrangement.
The principle of holographic recording is that light from the coherent
source scattered by the particle interfers at the film plane with light
which proceeds unscattered and forms the hologram interference pattern.
28
-------
o
73
•o
m
»
Vt
>
Z
OAK RIDGE DATA
CALIBRATION
-------
Ke Incfi
40 60 80 100 20 40 60 80 200 2 40 60 80
d , Droplet Diameter , Microns
300
30
-------
SPECTRAL FILTER
LASER SPATIAL FILTER
O) IN-LINE RECORDING
FILM
SPECTRAL FILTER
SPATIAL FILTER
b) RECONSTRUCTION
HOLOGRAM
MAGNIFYING
OPTICS
VIDICON
FIGURE 11
MONITOR
e
O
IN-LINE HOLOGRAPHY
-------
It is also often callecd a Gabor hologram because this was the first type
of hologram to be invented and was accomplished by Dennis Gabor in 1948.
The photographic film, desirably of high spatial resolution, is then
processed and replaced in the electromagnetic wave. The diffraction by
the interference pattern density variations in the film is such as to
produce, essentially, a focusing of light to produce a real image of
the hologram of the particle as shown in Figure lib. This can be
viewed on a white, diffusely-reflecting card or, with considerable
advantage, with a closed circuit television system. If the recording
and reconstruction light waves have the same properties the reconstructed
image will be at the same distance as the recording distance and the
cross-section of the particles under reconstruction will be the same as
the cross-sections of the original scattering particles. In this way,
one may therefore map out a dynamic particle field with respect to both
position and size distribution. This technique has been extended to
provide the concept of holographic velocemetry as reported in Reference 7.
The holographic system has the advantages that the hologram may be
recorded in the field and reconstructed in the calm, quiet environment
of the laboratory and of providing a permanent record of the particle
distribution. It has the disadvantages that the reconstruction necessitates
a two-step process and is therefore lengthy. It also has the disadvantage
that particles smaller than 50um are difficult to resolve, particularly
in the presence of a large number of fog particles. It therefore suffers
from the fog problem also. Its-principle range would be the larger
particles. But these occur less frequently, necessitating collection
of many holograms.
Although determined not to be the best choice for drift instrumentation,
it should be said in defense of the in-line, infrared holocamera whose
feasibility we did demonstrate that the recording and reconstruction was
realized on standard infrared film using standard processing techniques
and the accuracy and ease of reconstruction of large particles with
background was equivalent to that using visible lasers and associated
techniques which have been practiced widely (6).
COMPARISON OF THE VARIOUS DRIFT INSTRUMENTATION SYSTEMS
It is worth summarizing the important points comparing the PILLS and other
drift instrumentation systems to indicate their respective relative
advantages as well as the complementary nature of the three field-proven
systems.
The principle advantage of the PILLS system is its amenability to auto-
matic data processing permitting it to be an essentially on-line measuring
instrument. The disadvantage of the present single \L system is that it
is limited to measuring particles above a size which is determined by
the quantity of fog present.
32
-------
Isokinetic sampling is not capable of resolving particle size, but gives
an integrated measure of the mass of material contained in unit volume
of air flow. A significant advantage of this integrated measurement is
that it gives, directly, data on the materials which may be regarded as
providing the pollution or environmental impact.
The sensitive paper technique has the advantage of simplicity in data
collection but the disadvantage of lengthy data reduction. It is
sensitive also to fog and condensate which constitutes for it a background
as for the PILLS system. It is important to recognize, however, that
the limitation imposed by the fog on the PILLS system is to make it
incapable of measuring small particles and applicable, therefore, to
the larger particles whereas the limitation on the sensitive paper is
just the converse. Thus for a complete drift measurement program it
is conceivable to utilize all three of these techniques.
Holography is of limited value for recording small drift particles and
requires a lengthy data reduction process.
Several other techniques which would measure small particles of water
have been considered and/or proposed to us in the course of this study.
Such techniques include electron beam scattering and a wide variety of
impingement devices which leave a crater in some material which can be
related through calibration to the diameter of the impinging particle.
It is even conceivable that the isokinetic sampling can be made mass
selective by utilizing centrifugal force effects. As drift continues
to be a problem, it is certain that many other techniques will be
proposed and realized. The presentation here was limited to those
techniques for which we have actually demonstrated feasibility and
used in the field.
33
-------
SECTION VI
SUMMARY OF DRIFT MEASUREMENTS PERFORMED BY SEPTEMBER 1971
The results of several drift measurement experiments is now reported
and include measurements on a small commercial cooling tower, two
large mechanical draft towers, and brief data for a hyperbolic tower.
The most complete data are given for the mechanical draft towers which
were at Oak Ridge. The others are included for completeness and are
with the permission of our clients for whom we ran these tests.
AQUATOWER
This small commercial unit (# 15 tons, circulation rate 45 gpm) was
found to drift approximately .01%. It was from simultaneous isokinetic
and PILLS measurements that the calibration factor for the PILLS system
was determined as shown in the Appendix. Figure 12 gives the particle
size distribution versus particle diameter. Here the data are reported
as the number of particles per cubic foot per micron size range. The
particle density per unit size range extends over two orders of magnitude
in going from 80pm to over SOOjum. Note the humps in the curve around
100pm and around 200pm. Figure 13 shows the mass distribution per
cubic foot per unit size range which has a minimum in the neighborhood
of 175um. It is important to observe that this curve is plotted against
linear coordinates whereas the particle distribution is plotted
logarithmically.
DOUBLE FLOW MECHANICAL DRAFT TOWER
Measurements were made on a commercial tower (fan diameter 28 feet,
circulation rate 12,500 gpm, range 23°F, approach 7°F) when the PILLS
system was in an early state of development. Procedurally, the system
was traversed across the vertical efflux of the tower at the top of
the fan cone and appropriate area segmentations were generated to
realize a numerical integration for the total drift in grams per second
leaving the tower. The area segmentation and the integrated drift
yield associated with a given area are shown in Figure 14. Station
No. 2 was obviously a "hot spot" for drift emission and led to a
substantial portion of the total drift emission. Visual observations
indicated that this hot spot was rather localized and a better
representation would have been realized had several diameters scans
been made.
In this early stage of development the dynamic range of the PILLS
system provided only a factor of about 4:1 in particle size. The upper
and lower points could, however, be adjusted. Figure 15 shows the mass
emission histogram versus particle size and exhibits a distinct minimum
in the neighborhood of 110pm. Without the advantage of having reduced
35
-------
.OS
100
300
36
-------
FIGURE 13
AQUATOWER
MASS DISTRIBUTION
30
O>
20
10
100
150
37
200
250
-------
Station 1 2 3
4 5
+ 30
D(r)(gms/sec )
FIGURE 14
TOWER SEGMENTATION AND DRIFT RESULT
38
-------
O)
CO
IQ
12
11
10
9
8
7
6
5|
4
3
2
1
FIGURE 15
TOTAL DRIFT VERSUS PARTICLE SIZE
©
©
-t 1-
55 63 72 80 87 94
102
no
120
132
148
168
-------
data, a field decision was made which resulted in the choice for the
range of particle diameters shown. Data reduction indicated the surprising
minimum. It is now conjectured that the basis for the minimum - which
occurred for every station - is that the components below lOOjjm
constitute drift which has passed through the drift eliminators and the
components above lOOjum evidence a secondary generation and/or trans-
mission mechanism. A secondary generation mechanism might be described
as "tearing-off" at leading and trailing edges of the drift eliminator
and other structural members. It is interesting to note that the
Aquatower data taken when the instrument was in a more refined state of
development as shown in Figure 13 also show a minimum but in the vicinity
of 175yum. Note that both towers had a horizontal velocity component
in the fill whereas the large mechanical draft tower had a final vertical
component.
OAK RIDGE MECHANICAL DRAFT TOWER
The most comprehensive drift measurements were made at Oak Ridge involving
the PILLS system, the isokinetic sampler system, and the sensitive paper
techniques. Significant design parameters are: fan diameter 18 feet,
circulation rate 6,050 gpm, range 10°F, approach 7°F. (The circulation
rate was 4,000 gpm during the drift measurements.) The PILLS and isokinetic
system data are recorded simultaneously and the isokinetic sampling head
was immediately above the scattering volume Vs as shown in Figure 3
such as to make them simultaneously observe the same particle distri-
bution at the same point in space.
The principles for each of the measurement techniques have been given
earlier and the data in Table 1 are presented for brevity. Figure 16
presents graphically the essential data.
Procedurally, we first determined the velocity profile with a vane
anemometer mounted on the same carriage which was used subsequently for
the PILLS and isokinetic system traverse. The velocity profile is shown
as the dashed curve in Figure 16. An area segmentation as shown in
Figure 16 was realized to accomplish the approximate numerical integration
for drift. Visual observations indicated a high degree of uniformity for
the drift abound the tower. Further, the night on which these measurements
were made had very low wind conditions. A single diameter traverse was
thus adequate. The fourth column in Table 1 indicates the total number
of cubic feet collected at each of the stations. It was not possible
with the pump available to realize isokinetic conditions at Stations 2
and 6 and velocities were utilized which were approximately 30% lower
than isokinetic. The fifth and sixth columns in the table give the amount
of water determined by two analyses on both magnesium and calcium. The
seventh column gives the mean for these values. The eighth, ninth, and
tenth columas give the drift per unit area, that is the grams per second
per square foot, as determined by the three techniques. The isokinetic
results are determined according to the formula shown on the table. It
was assumed that the trapping efficiency of the glass bead tubes was 100%.
40
-------
TABLE 1
OAK RIDGE COOLING TOWER DATA SUMMARY
Station
j
1
2
3
5
6
7
Ctrl M9 =
ctrl Ca =
M= Mtr
ctr
D * V * M
\) — j
ft/sec ft2 «3
20 27.5 356
35 64.4 354
18.3 38.5 252
18.3 38.5 259
34.2 64.4 264
29 27.5 237
24.5 ppm
160 ppm
j
MMg
gms
.52
1.54
.30
.34
.95
.65
dri
MCa
gms
.50
.87
.38
.41
.63
.56
M
gms
.51
1.21
.34
.37
.79
.61
ft percentage
D/A/IK
.029
.119
.025
.026
.10
.074
R =
= '
jms/sec ft
D/ A/PILLS
.012, d>80jjm
.047, d>120
.017, d>100
.0046, d>80
.13, d>140u
.037, d>120
4000 gpm = 25
'19 x 100% _ n
25 x 104 U
13.6 x 100% _
V25 x 104
qtns/sec
D/ A/Paper TJj/IK Dj /PILLS
.79 .33
7.67 3.03
.95 .65
1.01 .18
.16 6.55 8.36
2.04 1.02
D = S D, = 19 13.6 gm/sec
"i
x 10^ gm/sec
.0076%, IK
0 0055% PILLS
A TVt
ENVIRONMENTAL SYSTEMS CORPORATION
KNOXVILLE, TENNESSEE
SEPTEMBER 1971
-------
res to the Inch
20 40
60
80 100 120
42
-------
Note particularly the qualification on particle diameter in colume
nine giving the determination based on the PILLS system. For example,
d > 8Qum means that the mass was measured by the PILLS system for all
particles larger than 80/jm at Station 1. At all stations except 1 and
5, it turned out that the counting rate was too rapid for the visual-CRO
technique described in Section IV. In order to reduce the counting rate
to within the capability of the observers it was necessary to increase
the lower level cutoff. For Station 6 this was 140jum and therefore a
substantial amount of material at this point was not counted. The
availability of either a digital counter or, better, a pulse height
analyzer system would have obviated this difficulty and it would have
been possible to record the drift down to the approximate 80/um limit
under the meteorological conditions under which the drift was made
since fog was not a severe problem. In other measurements made both
earlier and later, the fog yield was higher and provided a higher
cutoff value.
Figure 16 summarizes these data for the individual stations. The
straight line connecting points for the data are for clarity of presentation
and do not indicate the drift ^emission profile. Some comparison of the
PILLS and IK curves is in order. First, in every case except at Station
6 the isokinetic data are larger than the PILLS data, which is as it
should be since the isokinetic data includes all particles and the
PILLS data includes all particles beyond a given diameter. At Station
6 the drift was sufficiently large that the heat input to the hot bead
sampler was inadequate to keep the tube hot which could possibly have
led to a reduced trapping efficiency. At Station 2 the sampling velocity
was lower than isokinetic. Consequently, the actual drift is lower than
that shown. When the sampling velocity is lower than isokinetic there
is a tendency to record more drift.
Columns 11 and 12 give the integrated drift water emission as determined
via the PILLS and IK techniques using the velocity profile and the area
segmentation shown in Figure 16. The total tower emission is shown at
the bottom of the respective columns.
The general agreement in all these data must be regarded at this point
as encouraging.
Figure 17 shows the particle size histogram and again it is to be noted
that it spans more than two orders of magnitude and shows the presence
of minor variations in the vicinity of approximately llCtyjm and
approximately 190/jm. Figure 18 shows on a linear scale the mass histo-
gram or distribution and we again note the minor peaks as also evident
on the particle histogram. Note that there are particles larger than
SOOjjm being emitted.
Figure 19 indicates the results made for paper stain measurements made
at one station on the identical unit immediately adjacent to the unit
on which the measurements using the PILLS and isokinetic systems was
made. The presentation is in the form of a block type 'histogram with
43
-------
. 2
100
44
-------
25
20
-p-
en
15
10
FIGURE 18
OAK RIDGE MASS DISTRIBUTION
100
150
200
250
i m
-------
160
FIGURE 19
PAPER STAIN HISTOGRAM
OAK RIDGE
SEPT. 1971
EXPOSURE TIME - TWO SEC
UJ
et
<
o
OT
140
o:
120
100
-
- 80
OT
UJ
60
cc
LU
m
40
20
^n n
55
110 165 220 275
d, m
33O
385
-------
Hum increments. For the first time we saw the particle distribution
tend toward 0 below lOOjum. This was possible because the fog was light
when these data were taken and therefore did not constitute a severe
background problem. Careful examination of the data permit drawing
the dotted line shown in Figure 19 which again suggests that there are
two minor peaks in the particle distribution. The result is regarded
as highly encouraging giving as it does a check on the calibration of
the PILLS system in demonstrating the existence of these two peaks.
That the mass emission of the small particles is negligible is verified
by a rough estimate of the integrated mass as follows:
0 - lOOjjm , 814 particles with d = 50jj, 53 x 10~6
100 - ZOOjum , 176 particles with d = 150, 311
200 - 300jum , 49 particles with d = 250, 400
300um and up, 12 particles with d = 350, 269
1,033 x ID'6 gms
Thus the 0-100jjm size range contributes only about 5% to the mass
emission.
A final rough but very interesting calculation based on the paper stain
data is to note that the particle identified above were counted on
thirty 1/8 x 1/8" squares and that the sampling time was approximately
two seconds. This leads to an approximate value of
1,033 gms gms
D/A = 30x(l/8)^x(l/12)2ft2x2 sees = °'16 ftSsec
whose agreement with the PILLS and IK values in a similar position on
the other cell is also regarded as encouraging.
NATURAL DRAFT COOLING TOWER MEASUREMENTS
The PILLS and isokinetic systems developed by means of this contract have
been applied to a large natural draft cooling tower. It is possible at
this time to release the following data as shown in Table 2 which also
summarizes the data from the other cooling tower observations. A drift
percentage of 0.005% as determined via IK measurements is regarded as a
representative number for state-of-the-art hyperbolic cooling towers
having efficient and intact drift eliminators.., It is to be noted that
the PILLS system determined a drift percentage of 0.0012% for particles
larger than 145jjm in the vicinity of an inherent but minor void in the
drift eliminator structure. Far greater levels were found in the
vicinity of major voids. (These measurements were made within 6-10 feet
of the drift eliminator surface.)
47
-------
TABLE 2
SUMMARY: COOLING TOWER OBSERVATIONS
I. AQUATOWER
(1)5*.01% IK
(2) PILLS
Two Whatman #1 filters,
no heat load
Significant Mass>250 jjm
II. MECHANICAL DRAFT
(Oak Ridge)
(1) &* .0076% IK
(2) && .0055% PILLS
III. MECHANICAL DRAFT (!)$£ .005% PILLS
IV. HYPERBOLIC
~.005% IK
Z .0012% PILLS
3) -------- RILLS
(4) £
PILLS
Hot Beads ^
Count Rate Too Fast for
Small Particles
55-168 ^im Distinct mini-
mum; secondary mechanism
j.
Substantial Variations
d ? 145 jum, minor void
Few Particles with d>
lOOjum in normal areas
d? 150 urn (ratio to case 2;
e- drift elimination
efficiency)
Visual counting of oscilloscope sweeps.
Both in position and in time.
48
-------
Probably as important as the 0.005% number, the PILLS system yielded
as a result of a scan across a normal and intact portion of the drift
eliminator structure that very few particles larger than a lOOjjm are
present. It is suggested that the major components of drift in such
a hyperbolic are to be found in the particle size range 50-150/jm.
Larger particles will be eliminated by the drift eliminators and
smaller particles are unlikely to be generated unless a greater degree
of atomization is employed.
Item IV also introduces the interesting concept of drift eliminator
efficiency. PILLS measurements were made on the "up-wind" side of the
drift eliminator structure and it was found that the quantity of water
presented to the drift eliminator structure is at least two orders of
magnitude larger than the quantity which goes throughs (See Figure 5b
for typical "up-wind" data.)
It must be stressed that these measurements were made inside the cooling
tower near the drift eliminator face. The environmental conditions for
equipment and personnel must be reported as extraordinarily hostile and
unsafe. Fog was a severe problem. The variations in the significant
parameters of interest, drift, air velocity, and temperature, are
substantial in both space and time and these must be considered when
generating a number representative of the total tower performance.
49
-------
SECTION VII
ACKNOWLEDGEMENTS
Realizing the results herein reported has brought us into contact with
many individuals in widely spaced disciplines. It is important to
cite those who have been most instrumental but it is not possible to
thank all.
The initial interests in the drift problem expressed by The Marley
Company via J.B. Dickey, J.O. Kadel, and J.D. Holmberg, are acknowledged
The support, both financial and technical, of the Environmental
Protection Agency via Project Monitor Frank H. Rainwater is greatly
appreciated.
The cooperation of the Atomic Energy Commission in permitting the
"shake-down" tests and final demonstration at the Oak Ridge National
Laboratory permitted the work to be performed more conveniently and
effectively. The administrative assistance of Dr. J.L. Liverman,
ORNL Associate Director for Biomedical and Environmental Sciences, and
Mr. M.M. Yarosh, Environmental Quality Program was beneficial. Con-
venient access to the cooling tower facilities and efficient and
pleasant working conditions wer provided by Mr. B.B. Smith, Maintenance
Supervisor.
Much of the basic information for this development was gathered by
private conmunication, chiefly in the form of laboratory visits and
extensive telephone conversations. Most helpful were Mr. R. Berglund
and Dr. B. Liu, University of Minnesota, Mr. W. Kochmond, Cornell
Aeronautical Labs, Dr. R. Nutt, ORTEC, and Mr. J. Keathley, ORNL. In
addition to interesting technical discussions, Dr. Nutt and Mr. Keathley
loaned us several specialized instruments for exploratory development
and for the September 28 demonstration. Mr. J. Eddlemon, PULCIR, Inc.,
also provided equipment for these purposes.
We are grateful to Mr. R.A. Burns of General Public Utilities for
permission to release the natural draft cooling tower data and to The
Marley Company for permission to release data on the other large
mechanical draft unit.
Finally, the enthusiastic and creative work of Research Associates
G. Kreikebaum, T. Carlson, and Y. Watanabe is especially worthy of
thanks as is the engineering development on the sensitive papers by
J. Womack, the technical assistance of D. Maples and the secretarial
assistance of Carolyn Wells and Diana Corbett.
Frederick M. Shofner
Technical Director
Carl 0. Thomas
President
51
-------
SECTION VIII
REFERENCES
1. Mason, B.J., Jayaratne, O.W., and Woods, J.D., "An Improved
Vibrating Capillary Device for Producing Uniform Water Droplets
of 15 - 500jjm Radius," J. Sci. Instruments, 40, (1967).
2. Liu, Benjamin, and Berglund, R.N., University of Minnesota,
Minneapolis, private communications.
3. Pilie, Roland J., "Project Fog Drops: An Investigation of
Warm Fog Properties and Fog Modification Concepts," NASA, CR
368 (1966); also Kochmond, Warren, private communication.
4. Whitby, K.T., "Calculation of the Clean Fractional Efficiency
of Low Media Density Fibers," ASHRAE Journal. (1965).
5. Chilton, H., "Elimination of Carryover from Packed Towers with
Special Reference to Natural Draught Water Cooling Towers,"
Trans. Instn. Chem. Engrs.. 30 (1952).
6. Shofner, F.M., Kreikebaum, G., and Thomas, C.O., "Infrared In-Line
Holography," EOSDC Record (1971).
7. Shofner, F.M., "Fundamentals of Holographic Velocimetry," ICIASF
Record (1969).
53
-------
SECTION IX
APPENDIX
MATHEMATICS OF PILLS CALIBRATION DETERMINATION
VIA ISOKINETIC MEASUREMENTS
The calibration constant K of Equation 2 can be determined by recording
a voltage histogram with the PILLS instrument and taking isokinetic
sampling data of the same particle distribution. For accuracy, it is
advantageous to sample a distribution similar to that expected in the
field at the same position and at the same time. Both PILLS and
isokinetic sampling can measure the same basic quantity, the mass of
drift water per unit volume m.
By isokinetic sampling a mass of a certain j^1 mineral Mi( is collected
from a sampling volume V^. Thus the mass of drift water per unit volume
is
(A-l)
where M; is the mass of drift water and fy is the mineral concentration
in the circulating water. It is assumed that thej*r-n mineral concen-
tration in the drift and the circulating water are the same.
With the PILLS instrument one resolves the particle distribution and
obtains a voltage histogram. Let f(v-j) be the count rate of particles
with a size yielding a voltage v-j within an implied incremental range vi •
The number of the same size particles per volume is
ru = f(V1)4t (A-2)
Vs
as has been derived in Equation 3. The water content per unit volume
comprised of particles of this size is written as
•n" 3
Ami = sni^T di
Vs $
3/2
s f(Vl) M. ir v^ (A_2)
Vs 6 K
55
-------
The total water mass is the sum of the masses attributed by the
individual particle sizes:
m =Ss f(Vl) ^_At tr (A-3)
,3/2
Vs 6
Equating (A-l) and (A-3) gives
M-o sfct
m =
6
Since all parameters except k are known or measureable, the calibration
constant is determined as a function of simultaneous isokinetic sampling
and laser data:
Vk sAtfr -^ ,/9 2/3
f —?2jf(Vi)V:.3/2 (A-4)
Vs 5 i 11
The calibration factor K was determined using the Aquatower operating with
no heat load, thereby permitting measurements down the instrument noise
limit («80 jjm for this configuration). Finally, even though the isokinetic
collection includes the drift particles below the noise limit, their
exclusion by the PILLS system makes negligible difference in the value of
K because their mass is a small fraction of the total emission.
56
-------
1
Accession Number
w
5
Organization
2
Subject Field & Group
02 D
SELECTED WATER RESOURCES ABSTRACTS
INPUT TRANSACTION FORM
ENVIRONMENTAL SYSTEMS CORPORATION
KNOXVILLE, TENNESSEE 37901
Title
DEVELOPEMENT AND DEMONSTRATION OF
LOW-LEVEL DRIFT INSTRUMENTATION
•] Q Authors)
Shofner, Frederick M.
Thomas, Carl 0.
16
21
Project Designation
Demonstration Grant # 16130
GNK
Note
22
Citation
23
Descriptors (Starred First)
*Cooling Towers, *Test Procedures, Saline water, Acceptance Testing,
Fallout, Thermal Pollution
25
Identifiers (Starred First)
*Cooling Tower Drift Instrumentation
27
Abstract
Instrumentation for measurement of low level drift from cooling towers was developed.
Emphasis was placed on the Particulate Instrumentation by Laser Light Scattering
(PILLS) System.which is capable of on-line measurement and, with incorporation of
existing pulse height analyzer and mini-computer equipment, complete on-line data
reduction. Complementary techniques of isokinetic sampling and sensitive paper samp-
ling were developed and field proven. Feasibility was demonstrated for an infrared
in-line holocamera system. The design principles and engineering trade-offs for
the PILLS, IK» and sensitive paper techniques are described. Drift performance
data are given for a small air conditioning cooling tower unit, two large mechanical
draft cooling towers, and a natural draft tower.
Abstractor
Shofner, Frederick M.
Institution
ENVIRONMENTAL
SYSTEMS CORPORATION
WR:102 (REV. JULY 1969)
WRSIC
SEND. WITH COPY OF DOCUMENT, TO: WATER RESOURCES SCIENTIFIC INFORMATION CENTER
U.S. DEPARTMENT OF THE INTERIOR
WASHINGTON, D. C. 20240
*U. S. GOVERNMENT PRINTING OFFICE: 1972—Il8l|-lt82/38
* GPO: I 970-389-930
------- |