comparative evaluation of cooling tower drift
TRANSCRIPT
COMPARATIVE EVALUATION OF COOLING TOWER
DRIFT ELIMINATOR PERFORMANCE
by
Joseph K. Chan
and
Michael W. Golay
Energy Laboratory
and
Department of Nuclear Engineering
Massachusetts Institute of Technology
Cambridge, Massachusetts 02139
Final Report for Task #3 of theWaste Heat Management Research Program
Sponsored by
New England Electric SystemNortheast Utilities Service Co.
under the
MIT Energy Laboratory Electric Power Program
Energy Laboratory Report No. MIT-EL 77-004
June 1977
COMPARATIVE EVALUATION OF COOLINGTOWER DRIFT ELIMINATOR PERFORMANCE
byJoseph K. Chan andMichael W. Golay
Energy Laboratory Report No. MIT-EL 77-004
June 1977
ABSTRACT
The performance of standard industrial evaporative
cooling tower drift eliminators is analyzed using experiments
and numerical simulations. The experiments measure the
droplet size spectra at the inlet and outlet of the elimi-
nator with a laser light scattering technique. From these
measured spectra, the collection efficiency is deduced as
a function of droplet size. The numerical simulations use
the computer code SOLASUR as a subroutine of the computer
code DRIFT to calculate the two-dimensional laminar flow
velocity field and pressure drop in a drift eliminator. The
SOLASUR subroutine sets up either no-slip or free-slip
boundary conditions at the rigid eliminator boundaries. This
flow field is used by the main program to calculate the elimi-
nator collection efficiency by performing trajectory calcu-
lations for droplets of a given size with a fourth-order
Runge-Kutta Numerical method.
The experimental results are in good agreement with the
collection efficiencies calculated with no-slip boundary
conditions. The pressure drop data for the eliminators is
measured with an electronic manometer. There is good agree-
ment between the measured and calculated pressure losses.
The results show that both particle collection efficiency
and pressure loss increase as the eliminator geometry becomes
more complex, and as the flowrate through the eliminator
increases.
Acknowledgements
This work was begun in recognition of the growing
concern regarding the possible environmental effects of
cooling tower drift in central power station cooling. The
literature in this area has been very sparse, in regard to
the aerodynamic performance and basic physics of drift elimi-
nators, as well as in the related areas of field measurement
of drift transport and field data regarding the environmental
effects of salt exposures. The goal of this work has been to
improve this situation by providing a basic experimental and
theoretical understanding of drift eliminator performance,
spanning the range of designs in current industrial use.
This work has been conducted since 1974 with the
generous support of the New England Electric System and of
Northeast Utilities, through the MIT Energy Laboratory's
Waste Heat Management Program.
The work has been carried out by Joseph K. Chan and
myself, with the doctoral thesis research of Joseph being
derived from the project. The success of this project has
been greatly aided by the generous cooperation of the Ceramic
Cooling Tower Company, Ecodyne Cooling Products, and the
Marley Cooling Tower Company in donating eliminators for
testing and in providing critical reviews as the work has
progressed. In addition, special thanks are due to Thermo-
Systems Inc. for the loan of a Berglund-Liu Monodisperse
3
Droplet Generator, and to Spray Engineering Co. for supplying
SPRACO spray nozzles for use in the Drift Elimination
Experimental Facility.
At MIT, Professors Warren M. Rohsenow and S. H.
Chen have contributed valuably to the work in consultative
discussions regarding the design of the experiments and in
the interpretation of the results. Graduate students Ralph
Bennett and Yi Bin Chen have also provided valuable assistance
to the work in a similar fashion.
In conclusion, the competent typing of this report
by Ms. Marsha Myles also deserves grateful recognition.
Michael W. Golay
Associate Professor
of Nuclear Engineering
W
4
TABLE OF CONTENTS
Abstract 1
Acknowledgments 2
List of Figures 8
List of Tables 14
Chapter 1. Introduction 15
1.1 Background 15
1.2 Previous Theoretical Studies of DriftEliminator Performance 23
1.2.1 Roffman's Analytical Formulation 24
1.2.2 Foster's Model 24
1.2.3 Yao and Schrock's Model 25
1.3 Survey of Drift Measurement Techniques 25
1.3.1 Droplet Size DistributionMeasurement Techniques 26
1.3.1.1 Sensitive Paper 26
1.3.1.2 Coated Slide or Film 27
1.3.1.3 Laser Light Scattering 28
1.3.1.4 Laser Light Imaging 29
1.3.1.5 Holography 30
1.3.1.6 Photography 31
1.3.2 Total Drift Mass Measurement Techniques 31
1.3.2.1 Isokinetic Systems 31
1.3.2.2 High Volume Sampler 33
1.3.2.3 Airborne Particulate Sampler
1.3.2.4 Deposition Pans
1.3.2.5 Chemical Balance
1.3.2.6 The Calorimetric Technique
1.4 Industrial Efforts in Drift EliminatorEvaluation
1.5 Present Approach
1.6 Organization of this Report
Chapter
2.1
2.2
2.3
2.4
2.5
Chapter
3.1
3.2
3.3
3.4
3.5
Chapter
4.1
4.2
2. Theoretical Evaluation of DriftEliminator Performance
Introduction
Assumptions
Calculation of Air Flow Distributions
Pressure Loss Calculations
Droplet Trajectory and CollectionEfficiency Calculations
3. Results of Theoretical Calculations
Introduction
Air Velocity Distributions
Droplet Trajectories
Collection Efficiencies
Pressure Drops
4. Experimental Techniques
Introduction
Drift Elimination Facility
5
33
34
34
34
34
40
41
43
43
43
45
51
52
59
59
59
75
95
99
110
110
111
4.3 Drift Measurement Techniques
4.4 Calibration of the Drift MeasurementInstrumentation
4.5 Data Acquisition and Analysis Techniques
4.6 Pressure Loss and Air Speed MeasurementTechniques
4.7 Sources of Experimental Error
Chapter
5.1
5.2
5.3
5.4
Chapter
6.1
6.2
5. Comparison of Experimental Resultswith Theoretical Calculations
Introduction
Pressure Drop Across Eliminators
Collection Efficiency Results
Estimation of Experimental Error
6. Conclusions and Recommendations
Discussion of Results
Recommendat ions
References
Appendix A DATANA Program
A.1 Introduction
A.2 Description of the Programs
A.3 Description of Input Parameters
A.4 Listing of the DATANA Code
A.5 Sample Problem
6
114
119
125
132
134
139
139
141
144
150
153
153
172
179
186
186
186
191
195
223
Appendix B DAMIE Program 239
B.1 Introduction 239
B.2 Description of the Program 240
B.3 Description of the Input Parameters 241
B.4 Listing of the DAMIE Code 243
B.5 Sample Problem 251
LIST OF FIGURES
No.
1.1.1 Installation Schemes of Drift Eliminatorsin Cooling Towers 19
1.1.2 Some Common Drift Eliminator Geometries 21
1.1.3 Some Modern Industrial Drift Eliminators 22
2.3.1 General Mesh Arrangement in SOLASUR.Fictitious Boundary Cells are Shaded 47
2.3.2 Arrangement of Finite Difference Variablesin a Typical Cell 48
2.3.3 Flow Chart of the SOLASUR Code 50
2.5.1 Flow Chart of the DRIFT Code 58
3.2.1 Velocity Distribution of Air Flow in Single-Layer Louver Eliminator Using Free-SlipConditions at Upper and Lower Boundaries 62
3.2.2 Velocity Distribution of Air Flow in Single-Layer Louver Eliminator Using No-SlipConditions at Upper and Lower Boundaries 63
3.2.3 Velocity Distribution of Air Flow in Double-Layer Louver Eliminator Using Free-SlipConditions at Upper and Lower Boundaries 64
3.2.4 Velocity Distribution of Air Flow in Double-Layer Louver Eliminator Using No-SlipConditions at Upper and Lower Boundaries 65
3.2.5 Velocity Distribution of Air Flow in Sinus-Shaped Eliminator. (A) Free-Slip Conditionsat Upper and Lower Boundaries. (B) No-SlipConditions at Upper and Lower Boundaries 67
3.2.6 Velocity Distribution of Air Flow in Hi-VEliminator. (A) Free-Slip Conditions at Upperand Lower Boundaries. (B) No-Slip Conditionsat Upper and Lower Boundaries 68
9
No.
3.2.7 Velocity Distribution of Air Flow in Zig-Zag Eliminator Using Free-Slip Conditionsat Upper and Lower Boundaries 69
3.2.8 Velocity Distribution of Air Flow in Zig-Zag Eliminator Using No-Slip Conditions atUpper and Lower Boundaries 70
3.2.9 Velocity Distribution of Air Flow in E-EEliminator Using Free-Slip Conditions atUpper and Lower Boundaries 73
3.2.10 Velocity Distribution of Air Flow in E-EEliminator Using No-Slip Conditions atUpper and Lower Boundaries 74
3.3.1 Droplet Trajectory Plot for Single-LayerLouver Eliminator with Droplets Enteringthe Eliminator at Left of Figure . DropletSize is 40 pm 77
3.3.2 Droplet Trajectory Plot for Single-LayerLouver Eliminator with Droplets Enteringthe Eliminator at Left of Figure . DropletSize is 100 m 78
3.3.3 Droplet TrajectOry Plot for Double-LayerLouver Eliminator with Droplets Enteringthe Eliminator at Left of Figure . DropletSize is 40 um 80
3.3.4 Droplet Trajectory Plot for Double-LayerLouver Eliminator with Droplets Enteringthe Eliminator at Left of Figure . DropletSize is 100 m 81
3.3.5 Droplet Trajectory Plot for Sinus ShapedEliminator with Droplets Entering theEliminator at the Left of Figure . DropletSize is 40 m 82
3.3.6 Droplet Trajectory Plot for Sinus ShapedEliminator with Droplets Entering theEliminator at the Left of Figure . DropletSize is 100 m 8383
10
No.
3.3.7 Droplet Trajectory Plot for Asbestos-Cement Eliminator with DropletsEntering the Eliminator at Left of Figure .Droplet Size is 40 m 85
3.3.8 Droplet Trajectory Plot for Asbestos-Cement Eliminator with DropletsEntering the Eliminator at Left of FigureDroplet Size is 100 pm 86
3.3.9 Droplet Trajectory Plot for Hi-V Elim-inator with Droplets Entering the Elimi-nator at Left of Figures.
(A) 40 pm Droplet Size(B) 100 m Droplet Size 87
3.3.10 Droplet Trajectory Plot for Zig-ZagEliminator with Droplets Entering theEliminator at Left of Figure . DropletSize is 30 m 89
3.3.11 Droplet Trajectory Plot for Zig-ZagEliminator with Droplets Entering theEliminator at Left of Figure . Droplet 90Size is 60 m 90
3.3.12 Droplet Trajectory Plot for Two-LayerZig-Zag Eliminator with Droplets Enteringthe Eliminator at Left of Figure . DropletSize is 30 m. 91
3.3.13 Droplet Trajectory Plot for Two-LayerZig-Zag Eliminator with Droplets Enteringthe Eliminator at Left of Figure . DropletSize is 60 m 92
3.3.14 Droplet Trajectory Plot for E-E Elimi-nator with Droplets Entering the Elimina-tor at Left of Figure . Droplet Size is30 m 93
3.3.15 Droplet Trajectory Plot for E-E Elimi-nator with Droplets Entering the Elimi-nator at Left of Figure . Droplet Sizeis 50 m 94
11
No.
3.4.1 Collection Efficiency of Droplets as aFunction of Droplet Size
(a) DRIFT Calculation(b) Roffman Calculation 96
3.4.2 Collection Efficiency of Droplets as aFunction of Droplet Size
(a) Sinus-Shaped Eliminator(b) Double-Layer Louver
Eliminator of Same Dimensions.(Air Inlet Velocity=l m/s) 98
3.5.1 Pressure Drop Distribution Along theLength of Sinus-Shaped Eliminator 102
3.5.2 Pressure Drop Distribution Along theLength of Asbestos-Cement Eliminator 103
3.5.3 Pressure Drop Distribution Along theLength of E-E Eliminator 104
3.5.4 Pressure Drop Distribution Along theLength of Double-Layer Louver Eliminator 106
3.5.5 Pressure Drop Distribution Along theLength of Hi-V Eliminator 107
4.2.1 Schematic Diagram of Drift EliminationFacility 112
4.3.1 Schematic Diagram of Light ScatteringDrift Measurement Instrumentation 116
4.3.2 Scattered Light Intensity Versus DropletSize Calculated by DAMIE 117
4.4.1 Schematic Diagram of the Model 3050Vibrating Orifice Monodisperse AerosolGenerator 121
4.4.2 Schematic Diagram of the Droplet GeneratingSystem 122
4.5.1 Intensity Distribution Across the LaserBeam Cross Section 126
12
No.
4.5.2 Measured Pulse Height Distribution forMonodispersed Water Droplets of 80 mDiameter 127
4.5.3 Calibration Curve - the Peak Voltage ofPulse Height Distribution Versus DropletSize 128
5.1.1 Drift Eliminator Geometries. (A) Belgian-Wave Eliminator, (B) Hi-V Eliminator, (C)Zig-Zag Eliminator 140
5.3.1 Predicted and Measured Droplet CollectionEfficiency Functions for Belgian-WaveDrift Eliminator at 1.5 m/s Air Speed 145
5.3.2 Predicted and Measured Droplet CollectionEfficiency Functions for Hi-V Eliminatorat 1.5 m/s Air Speed 147
5.3.3 Predicted and Measured Droplet CollectionEfficiency Functions for Zig-Zag Eliminator at1.5 m/s Air Speed 148
5.3.4 Predicted and Measured Droplet CollectionEfficiency Functions for Commercial DriftEliminators at 2.5 m/s Air Speed
(A) Belgian Wave Eliminator(B) Hi-V Eliminator(C) Zig-Zag Eliminator 149
6.1.1 Terminal Velocities of Water Droplets 157
6.1.2 Flow Visualization Photograph of theBelgian-Wave Eliminator. (Dye BeingInjected at the Right Side of the Picture) 160
6.1.3 Flow Visualization Photograph of theBelgian-Wave Eliminator (Paper ChipTrajectories) 161
6.1.4 Flow Visualization Photograph of the Hi-VEliminator. (Dye Being Injected at theRight Side of the Picture) 162
13
No.
6.1.5 Flow Visualization Photograph of the Hi-VEliminator (Paper Chip Trajectories) 163
6.1.6 Flow Visualization Photograph of theZig-Zag Eliminator. (Dye Being Injectedat the Right Side of the Picture) 164
6.1.7 Flow Visualization Photograph of the Zig-Zag Eliminator (Paper Chip Trajectories) 165
6.2.1 Schematic Diagram of the Proposed Experi-mental Setup for Studying Droplet Traject-ory and Air Velocity Distribution in DriftEliminators 173
A.2.1 Flow Chart of the DATANA Code 188
Approximation of the Calibration CurveA,3.1 193
14
LIST OF TABLES
No.
1.4.1 Field Work of the Environmental SystemsCorporation 39
3.1.1 Physical Dimensions of the EliminatorsUnder Study 60
3.4.1 Collection Efficiency Calculated by DRIFTat 1.5 m/s Air Velocity for Double-LayerLouver Eliminator and E-E Eliminator 100
3.5.1 Calculated Pressure Loss Across SomeCommon Drift Eliminators 108
4.4.1 Droplet Diameter as a Function of TypicalDroplet Generator Parameters 124
4.5.1 Sensitivity Analysis of the CollectionEfficiency Results 131
5.2.1 Pressure Drop Across Eliminator at LowFan Speed 142
5.2.2 Pressure Drop Across Eliminator at HighFan Speed 143
5.4.1 Measured Collection Efficiencies of theZig-Zag Eliminator at 1.5 m/s Air Speed 151
6.1.1 Pressure Drop and Calculated CollectionEfficiency Results of Some DriftEliminators at an Air Speed of 1.5 m/s 171
A.5.1 Input Data for DATANA Sample Problem 224
Output for DATANA Sample ProblemA.5.2 228
15
CHAPTER 1
INTRODUCTION
1.1 Background
Current practice in the design and operation of new electric
power stations selects a single method of waste heat disposal
and then designs the cooling apparatus to meet the worst sta-
tion heat load throughout the year (Di). This is an outgrowth
of past trends, in which once-through cooling was virtually the
universal method of power station waste heat disposal in the
United States. In the late 1960's waste heat disposal suddenly
became a controversial topic with the introduction of unprece-
dentedly large (>800MWe) and thermally inefficient nuclear
power stations. In 1973 the Environmental Protection Agency
(EPA) added impetus to the use of cooling towers when it took
under advisement a Burns & Roe study indicating that evaporative
cooling towers may well be the only closed circuit cooling option
available in the near future. Based on this study, the EPA
recommended the evaporative cooling tower as the best practical
technology under the Water Pollution Control Amendments. Sub-
sequent concern for protection of the aquatic environment, and
a desire to avoid costly licensing delays has motivated many
utilities to design their new, large power stations using cool-
ing towers rather than once-through cooling. As recently as
October, 1973, a complete listing of all operating or committed
nuclear generating units revealed that 48% of the generating
capacity was to be served by cooling towers. The participation
by fossil-fueled plants is not as great as this, and projections
16
indicates that about 50% of the newly added power generating
installations at early 1980's will be using cooling towers.
The major cooling tower vendors in the United States are
Ecodyne, Inc., Santa Rosa, Calif.; The Marley Co., Mission,
Kansas; Research Cottrell, Inc., Bond Brook, N.J.; Ceramic Co.,
Fort Worth, Texas, and Zurn Industries, Erie, Pa. Other large
corporations which are either entering the field or considering
doing so are Westinghouse Electric Corp., General Electric
Corp., and the Babcock and Wilcox Co.
To meet the increasing demand for electricity in the United
States, the utilities are planning to build a large quantity of
new, large power stations with more emphasis on nuclear power
plants. With the prospect of rapidly increasing cooling require-
ments due to these plants, special attention has been paid to
the environmental effects of cooling methods. The major areas
of concern related to the environmental effects of cooling
towers are fog, icing, and drift deposition.
Drift consists of the water droplets that are mechanically
entrained in the cooling tower's exhaust air stream from the
station's cooling water. Drift particles contribute very little
to the visibility of cooling tower plumes because the quantity
of drift is very small compared to the other forms of water
present. The following order of magnitude numbers for the mass
concentration of typical cooling tower effluents illustrates
this point (S3):
X(vapor) 20 g/m3
X(fog) " 1 g/m3
X(drift)\ 0. 01 g/m3
17
Drift has several important deleterious effects on the
local environment. When the mixture of water vapor and drift
particles in the cooling tower plume, mixed with the ambient
cold air,is carried away, the drift particles may form nuclea-
tion sites for condensation. Also, the mixing of the cooling
tower plume with the stack plume may form acids through chemical
reactions.
In order to meet future electric power requirements and
because of the scarcity of cooling water, it will be necessary
for many of the new power generating plants to utilize cooling
water that contains various concentrations of salt, e.g.,
brackish inland waters, estuarine water, or sea water. There-
fore the drift will contain salt as well as chemicals from the
coolant water chemistry. The main concern about drift is its
potential for damage to nearby facilities, transmission lines
and biota. In some instances, drift has caused serious prob-
lems in electric distribution systems; the drift deposits
being responsible for equipment failures. Cases involving
corrosion and fouling of nearby structures have been reported
from both fresh and sea water cooling towers (L3). Drift can
also be a considerable nuisance when it spots cars, windows,
and buildings.
Estimates of drift from cooling towers range from 0.001%
of the circulating water to more than 3%. The industry practice,
until early 1970, was for cooling tower vendors to guarantee
drift release to be less than 0.2%. At the American Power
Conference in Chicago (April, 1970) a new performance standard
18
of 0. 03% was introduced,and in November, 1970, a further
reduction was proposed, leading to the estimate that future
cooling towers may be certified for drift release less than
0.002%.
Drift from cooling towers is traditionally reduced by
passing the exhaust flow through drift eliminators installed
in the cooling towers. These eliminators operate by passing
the two-phase flow stream through a curved duct, with the heavy
water droplets becoming trapped on the duct walls due to
centrifugal acceleration. The accumulated water on the walls
flows back into the cooling tower.
There are many different ways to install the drift elim-
inators in a cooling tower, depending upon the type and geometry
of the cooling tower. All cooling towers are either crossflow
or counterflow types, which is determined by the flow direction
of he cooling air relative to the downward travel of the water
to be cooled. In general, eliminators are installed either
horizontally or vertically. The horizontal scheme is commonly
used in crossflow type cooling towers and the vertical scheme
in counterflow type cooling towers, as shown in Fig. 1.1.1.
The horizontal installation scheme is easier and more sturdy
in construction. It can also be used to adjust the air flow
pattern within the tower. The main problem with the horizontal
installation scheme is the inefficient drainage of water from
the eliminator walls: a thick water film forms on the elimi-
nator walls and reduces the drift collection effectiveness.
The vertical installation scheme has little water drainage
19
I AIR I OUTLET I
FAN
WATER
INLET
N i _ ___ ___ _ _AIR x = AIR
INLET -- _ INLET
i i i rli tw
Countcr-Flow tower
WATEROUTLET
Cross-Flow tower
Fig. 1.1.1 Installation Schemes of Drift
Eliminators in Cooling Towers
DRIFT'-ELIMINATORS
DISTRIBUTION- SYSTEM
r�I
- - - -"1
=--_--.--
I------------------- ,
20
problem due to the enhanced film flow by gravity.
There are many different types of drift eliminators sold
by cooling tower vendors. The common ones are shown in Fig.
1.1.2. The single and double-layer louvre eliminators are
generally made with wood.The sinus-shaped eliminator is made
from asbestos cement. The Hi-V eliminator is made of polyvinyl
chloride (PVC) plastic. The zig-zag eliminator is made from
fiber. Some other industrial eliminators are also shown in
Fig. 1.1.3.
The performance of drift eliminators can be quantified by
two factors: the droplet collection efficiency and the pressure
drop across the eliminator. The collection efficiency is gen-
erally defined as the ratio of drift mass collected by the
eliminator to the total drift mass entering the eliminator.
For environmental protection, this factor should be high. The
pressure drop across the eliminator represents the resistance
of the eliminator to the exhaust air flow. The presence of an
eliminator will reduce the air flow within the cooling tower,
thus decreasing the tower's cooling capacity. This particular
effect can be very detrimental in natural draft cooling towers,
since they pass only the small draft caused by the air density
difference at the entrance and exit. For mechanical draft
cooling towers, a high pressure drop will cause a high horse-
power requirement in the fans. Therefore, for inexpensive
cooling tower performance, the pressure drop across the elimi-
nators should be as low as possible.
Eliminators operate on the principle of centrifugal
separation caused by turning of the flow in the duct. In
5,0 cm-l 5.0 cm ... q .-
cm
(A)
cm
(B)
-- X
(CH(E)
(F)
Some Common Drift Eliminator Geometries
21
4,3 cm -I rcm
(C)
(D)
Fig. 1. 1. 2
Herringbone
Eliminat or
AIR
FLOW SD
FILL SIDE
1ArDuplex Eliminator
DISCH. SIDE.
PVC Chevron Type
Eliminator
Fig. 1.1.3 Some Modern Industrial Drift Eliminators
22
AIR 3
FLOW V
FILL SIDE DISCH. SIDE
11-1
-11 a;-O-
-- r
Ii
.-
-. _we
II.
*%ON! " *`
040-
23
general, more turning results in a higher collection efficiency,
but a higher pressure drop. In order to achieve a high
collection efficiency and a low pressure drop, the design of
drift eliminators calls for an optimization between these two
factors. In current industrial practice, there is no standard
design procedure for doing this. That is, all existing drift
eliminators are generated through random innovation, experience,
and experiments. This thesis develops a numerical technique
to study the cooling tower drift eliminator performance, which
can eventually be used to evaluate and design drift eliminators.
1.2 Previous Theoretical Studies of Drift EliminatorPerformance
Studies of eliminator performance have been carried out
mainly with experiments. However, the experiments suffer from
the difficulties encountered in-measuring the drift quantity
and distribution. None of the drift measurement techniques
has yet been proven to be generally satisfactory to the point
of their being adapted for general use (Al). Theoretical
studies are rarely performed because it is feared that such
studies would be unreliable due to a number of uncertainties.
These include the possibility of flow turbulence within the
eliminator, the droplets rebounding from or being generated
in the water film on the eliminator walls, and the water film
drainage system design. Despite this, a theoretical model is
still a very useful tool in evaluating the relative performances
of different drift eliminators, and in designing improved drift
eliminators. Recently a few attempts have been made in this
24
direction; the approaches are briefly described next.
1.2.1 Roffman's analytical formulation
An analytical formulation for the estimation of drift
eliminator collection efficiency has been developed by Roffman
et al. (R4). In this model it is assumed that the drift drop-
lets flow longitudinally at the assumed-constant vertical air
velocity within the eliminator, and that it experiences trans-
verse viscous drag due to the transverse air velocity component.
This component is obtained by assuming that the air velocity at
any point in the eliminator is locally parallel to the eliminator
wall. For complex geometries the model uses a Fourier series
expansion of the transverse velocity component in terms of the
duct contour. By using these assumptions an explicit form of
the equation describing the droplet transverse displacement
can be obtained as a function of longitudinal location of the
droplet. From the displacement information it can be determined
which of the entering droplets will hit the eliminator walls.
The collection efficiency of the eliminator can be determined
as a function of droplet size, The results are claimed to be
satisfactory when overall collection efficiencies are compared
with the experimental data obtained by Chilton (C4).
1.2.2 Foster's Model
Foster, et al. (F3) have developed a potential flow
numerical simulation model for theoretical investigations of
drift eliminators. The model defines the effective eliminator
boundaries with experimental flow visualization photography,
25
and it is assumed that all droplets entering this region are
eliminated. The main stream flow fields are obtained by solving
the Laplace equation for the velocity potential within an
experimentally defined laminar flow region. Using this inform-
ation the collection efficiency for any droplet size is esti-
mated from numerically computed droplet trajectories by solving
the droplet equation of motion using a Runge-Kutta-Gill pro-
cedure. However, the estimated efficiencies are much greater
than those observed experimentally. This is thought to be due
to the improper treatment of the turbulent wake region. It has
been found that results obtained from direct calculation of the
flow field without definition of the turbulent wake region
provide better agreement with experiments (F2).
1.2.3 Yao and Schrock's model
Yao and Schrock (Y2) also developed a numerical model
for evaluating the eliminator collection efficiency. The
flow field is calculated by a relaxation method for iterative
solution of the Laplace equation for the stream function.
The droplet trajectories are calculated step by step in space,
with the droplet drag-induced acceleration assumed constant
within a given mesh interval. In this model the pressure drop
across the eliminator is also calculated by using a boundary
layer analysis.
1.3 Survey of Experimental Evaluation of Drift Eliminator
Performance
Experimental evaluations- of drift eliminator performance
are performed by measuring the drift at the exhaust side of the
26
eliminator in a particular cooling tower or a simulated cooling
tower facility. In most cases only the drift rate (defined as
the drift mass flowrate escaping the tower divided by the re-
circulating water flowrate in the tower) is measured. The drop-
let size-dependent collection efficiency of the eliminator is
generally never measured. Many methods exist for measuring
drift in these two ways. Most of them stem from droplet
measurement techniques in cloud physics. Those that are widely
used are summarized below.
1.3.1 Droplet Size Distribution Measurement Techniques
The following methods measure the drift droplet size
distribution. The total drift rate can be determined by inte-
grating the distribution over the droplet size.
1.3.1.1 Sensitive Paper
This method has been used extensively to measure the
liquid water content and size distribution in clouds and fog.
Recently this method was adapted for cooling tower drift measure-
ments (F1,R3,S3,S4,W2). In this method filter paper is sensi-
tized by soaking it with a 1% solution of potassium ferricyanide.
The paper is dried thoroughly and dusted with finely ground
ferrous ammonium sulfate. The treated paper is pale yellow
in color. When a water droplet falls on the paper, it dissolves
both chemicals and forms an insoluble blue precipitate known as
Turnbull's blue which is easily identifiable against the pale
yellow background. The area of the stain is related to the
droplet diameter. Adjustments must be made for the speed of
27
impingement and porosity of the paper. The best method of
obtaining calibration factors for these variables and various
droplet sizes is to use a monodisperse droplet generator to
form stains from a known droplet size, speed of impingement,
and porosity of the paper. The calibration is independent of
sensitizing agent (C4).
There are two types of sensitive paper sampling methods.
The most common method exposes the paper briefly in the air
stream with the paper normal to the air flow. However, in this
method, the impingement speeds are different for different
droplet sizes. A second method (S3) moves the sensitive paper
through the air by a rotating head machine with the axis of
rotation parallel to the air flow. The head velocity is perpen-
dicular to the average air flow and droplet trajectory, there-
fore the droplet impingement speed is always equal to the
rotational speed of the heads.
The collection efficiency of sensitive paper depends on
the droplet sizes and velocities. Calibration of this method
should include consideration of the dynamics of particle motion
and impingement: particles can impinge at an angle, producing
elongated stains, and at higher velocities droplets will produce
larger stains. The collection efficiency decreases for smaller
droplets. For these reasons calibration and data reduction are
time-consuming in the sensitive paper technique.
1.3.1.2 Coated Slide or Film
The measurement technology for this method was also
established by cloud physics investigators. This technique is
28
easily adaptable to field measurements of cooling tower drift
droplet size distribution (R3,S4,W2). In this method a glass
slide or photographic film is coated with a material that
preserves the shape of impinging droplets against coalescence
and evaporation. Of all the slide coatings evaluated, a liquid
plastic coating called FORMVAR gives the clearest and most
distinct representation of the drift droplets. When a water
droplet impacts the coating, it is encapsulated as the plastic
solvent evaporates. The water in the droplet eventually evap-
orates through the thin FORMVAR skin,but the exact shape of the
impacting droplet is preserved by the plastic film for future
size analysis with a microscope. Calibration involves correct-
ions for the flattening of droplets on the slide, and for
evaporation, which is a function of time and droplet mineral
concentration.
This technique has an upper droplet size limitation in
the range of 200 to 300 microns. When droplets larger than
this impinge on the slides, the droplets tend to shatter, making
a size determination impossible. As with the sensitive paper
method, data reduction is lengthy and tedious.
1.3.1.3 Laser Light Scattering
In the laser light scattering technique for drift
measurement (S2,S4,S5,S7), droplets are illuminated by coherent,
monochromatic laser light. Light scattered by a particle within
the sampling volume (defined by the intersection of the laser
beam and the detector acceptance cone) is detected by a photo-
detector, producing a current pulse which is related uniquely
29
to the droplet size. The current pulses are analyzed and
stored in a pulse height analyzer and the data can be processed
by a minicomputer. The size of the sampling volume should
be small, so that lengthy sampling times can be avoided, and
so that the probability of having more than one particle pres-
ent in the volume is small.
The system is calibrated by noting the response of the
instrument to droplets of known size that are generated by
a monodisperse droplet generator. However, this method is
complicated by the variation of the laser light intensity
across the laser beam and by an edge effect.
The main advantage of the laser light scattering system
is that it can operate on-line, providing fast results.
1.3.1.4 Laser Light Imaging
This method has not been used in cooling tower drift
measurement but appears in principle to have some advantages
over the laser light scattering system (K1). In this method
a linear array of photodetectors spaced equally measures the
droplet shadow diameter. The droplet passes between a He-Ne
laser and the detector array of fiberoptics. An optical
system focusses the laser beam to cast the droplet's shadow
at the desired magnification on the detector array. A volt-
age drop across a given detector in the array due to shadowing
is compared to the quiescent voltage of the unshadowed detectors.
Since the ambient light level is always used as a reference,
this method has an increased sensitivity to soiling of its
optics. The size of particle is determined by the number of
30
occulted fibers. Only shadows lying fully within the array
are used, which eliminates the unavoidable edge effects of
scattering or extinction methods.
The device operates on-line and samples particles in situ.
However, it is expected that considerable experimentation and
possibly modification would be required before an imaging
instrument was developed to the point of practical applications
for drift measurement.
1.3.1.5 Holography
The principle of this method is that light from coherent
laser light source scattered by the droplet interferes at
the film plane with light which proceeds unscattered and forms
the hologram interference pattern. The photographic film 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 a focusing of light to produce
a real image of the hologram of the droplet. This can be
viewed with a closed circuit television system. If the record-
ing 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 droplet under
reconstruction will be the same as the cross-section of the
original scattering droplet. In this way, one may therefore
map out a dynamic droplet field with respect to both position
and size distribution.
The method has been used in measuring fog droplets in
the size range of 5 to 35 microns (T3). The system has the
31
disadvantage that the reconstruction necessitates a two-step
process and is therefore lengthy. This method is expensive
and is shown to be inferior to the light scattering method (S4).
1.3.1.6 Photography
Droplets can be filmed using a high-speed cine camera,
with the droplets being diffusely illuminated from the opposite
direction. Droplets down to a diameter of 50 microns have
been measured. The films are studied frame by frame using an
analyzing projector, and the diameter, velocity, and trajectory
of the.droplets that are clearly in focus can be analyzed. This
method has been used in studying drift eliminator collection
efficiency (F3). However, the data reduction is lengthy.
1.3.2 Total Drift Mass Measurement Techniques
In most experimental work drift eliminators are evaluated
by measuring the total drift mass flux escaping cooling towers.
Some of these methods are described below.
1.3.2.1 Isokinetic Systems
In isokinetic systems air is drawn into the collector
with a kinetic energy identical to that of a fluid element at
that position, had the collector not been there. If the
density and temperature of the air do not change as the air
is drawn into the collector, isokinetic sampling requires
only that the velocity of the air flow into the collector
being equal to that in the absence of the collector at the
point of measurement. In an isokinetic system, the mean
air flow within the collector is adjusted by a blower to be
32
equal to the mean air flow outside the collector. There are
many different isokinetic systems which use various collectors.
One of them is cyclone collector (R1,W2), where droplets
entering the cyclone collector are separated from the air
stream by centrifugal force and are collected in a container.
The collection efficiency of the collector is determined in a
fog chamber. Drift droplets collected are analyzed by atomic
absorption spectroscopy for dissolved mineral concentration.
Since the collected water contains not only drift water, but
also condensed water, the drift mass flux cannot be determined
simply from the quantity of the collected water. Rather, the
drift mass flux is determined from the dissolved mineral con-
centration by assuming that the mineral concentration in the
drift is the same as in the makeup water source. This consti-
tutes. the greatest uncertainty in this method.
Another kind of collector is the isokinetic sampler
tube (H4,M1,S3,S4) in which a heated glass tube filled with
glass beads is used to collect drift mineral residue. The
heating element evaporates all of the liquid water sampled.
Only the mineral residues are retained for subsequent chemical
analysis. This method also suffers from the uncertainty in
assuming an equality of mineral concentration in the drift and
the makeup water source.
The mineral background in a real cooling tower is generally
high, and this introduces even more error into either of these
methods.
33
1.3.2.2 High Volume Sampler
The high volume sampler method measures the drift mineral
concentration per unit volume of air (L1,R2). Air is pumped
through a filter and particles in the air are trapped. The air
flow rate through the filter is recorded continuously to give
the total volume of air sampled. The filter is heated to keep
it dry, Data reduction of the drift mineral concentration is
performed with atomic absorption spectroscopy and by comparing
the results to a clean filter background count, This method is
affected by ambient humidity, wind, and background airborne
particulate concentration.
1.3.2.3 Airborne Particulate Sampler
The airborne particulate sampler (APS) was originally
developed for monitoring atmospheric salt loading at coastal
locations. It operates on the principle of collection by
impaction. Two woven polyester meshes mounted on rotating arms
sweep out a known volume of air per revolution. By counting
the number of revolutions, the total volume of air sampled
can be determined. A fan maintains the air flow past the
meshes and keeps it parallel to their plane. A wind vane
rotates the entire system about the vertical axis so that it
always faces into the wind. Calibration can be done with a
monodisperse droplet generator. Data reduction is performed
by a spectroscopic analysis of the meshes for salt content.
The main advantage of the APS over the high volume sampler is
that the APS system does not require as much power, and can be
34
run on a car battery at remote locations,
1.3.2.4 Deposition Pans
In this method petri dishes or polyethylene ars are
put at various locations in the horizontal plane surrounding
the cooling tower to measure the quantity of drift residue that
settles on the ground. Residue is collected for a known length
of time and is analyzed by atomic absorption spectrophotometry.
1.3.2.5 Chemical Balance
This method measures the rate of decrease in concentration
of a chemical such as sulfate or other tracer chemicals added to
the circulating water (C2). The drift rate is calculated from
the amount of change in the concentration of the tracer with
time. The disadvantages of this method are that a long test
period is required and that circulating water systems invariably
have other leaks that deplete the chemical tracer.
1.3.2.6 The Calorimetric Technique
The calorimetric technique incorporates special thermo-
dynamic and hydrodynamic principles by utilizing a calorimeter
with a throttling nozzle (R3). The droplets passing through
the throttle point evaporate because of a pressure drop, and
in doing so, they remove heat from the surrounding air. This
in turn causes a detectable air temperature drop which is used
to determine the drift rate.
1.4 Industrial Efforts in Drift Eliminator Evaluation
The first extensive investigation of drift eliminator
35
performance was done by Chilton (C4) in the late 1940's and
early 1950's. The test apparatus included a closed loop
experimental tower which simulated a natural draught cooling
tower. The drift droplets were collected by a Calder Fox
Scrubber at the tower exit. By measuring the water collected
for a certain period of operating time at different velocities,
the collection efficiencies of various eliminators for several
ranges of droplet size were determined. The pressure drop was
measured by pitot static tubes leading to a Chattock Fry tilting
micro-manometer. Many different eliminator geometries were
tested, and a double-layer louvre eliminator was recommended,
which was subsequently adopted on many cooling towers in England.
Measurements of precipitation from the cooling towers after
installation of the recommended eliminator were then performed
using the sensitive paper technique. The sensitive paper used
was Whatman No. 1 filter paper.
The experiment was considered to be a great success.
Since then, not much work on eliminator performance evaluation
has been reported until recently. In 1969, drizzle from two
modern 2000 MW stations was detected by the Central Electricity
Generating Board Regional Scientific Service Staff. Research
work on drift eliminators was subsequently rekindled by the
Central Electricity Research Board. Tests similar to those by
Chilton were performed on some eliminator geometries (Gl),with
a recommendation for a closer pitched (1.75 in.) asbestos-
cement eliminator. Droplet size measurements were made on water
sensitive papers exposed inside cooling towers at various levels
36
including both under and over the eliminators (M3). Droplet
removal efficiencies were found for conventional louvre elim-
inators. The sensitive paper technique described in this work
is the same as the one reported in Chilton's paper except that
the calibration was extended to smaller droplet sizes (25-400pm).
Theoretical evaluation was also carried out to calculate the
collection efficiency as a function of droplet size (F3). The
theoretical efficiencies were found to be much greater than
the observed efficiencies from their experiments which was done
with a photographic method.
In 1971, Fish and Duncan at Oak Ridge National Laboratory
developed an isokinetic sampling sensitive paper technique
using Whatman No. 41 filter paper (F1). The technique was
used to measure the drift size distribution above drift elimi-
nators of a counterflow hyperbolic cooling tower. The drift
rate was found to be 0.002-0.006%.
The Marley Company has established a strong program in
drift measurement and drift eliminator development since late
1960's. In 1968, a chemical balance method was used in the
Marley Laboratory to check drift levels with and without drift
eliminators in the testing tower. The technique was also used
in drift determinations on an operating mechanical draft in-
dustrial crossflow tower at a Municipal Power Plant. n 1970,
the Marley Co. was interested in operating a cooling tower on
salt water makeup, which required an accurate knowledge of
drift rate. Since that time they have sponsored and cooperated
with the Environmental Systems Corporation (ESC) to develop
37
reliable drift measurement instruments that include the
Particulate Instrumentation by Laser Light Scattering (PILLS)
system, the Isokinetic Sampling (IK) system, and sensitive
paper techniques. Later the Marley Co. added a special drift
test cell to the Marley Laboratory exclusively for drift elim-
inator development. Drift measurements were mostly done with
the isokinetic sampling system developed by ESC. Many different
eliminators have been tested. Some of the important conclusions
are listed here (H4):
(1) Numerous observations have shown that the
circulating water rate has little effect on
the drift level. Specific tests on the Duplex
eliminator revealed, within the limits of test
accuracy, that there was no change in drift
rate with circulating water rates ranging from
12 GPM/ft 2 to 22 GPM/ft 2.
(2) Theoretically, drift eliminator collection
efficiency increases with air velocity. However,
the water load on the eliminator also increases
with the air velocity, but at a greater rate than
the increase in efficiency. Altogether it was
found that drift increases with air velocity.
The rate of this increase can be drastic with
an inefficient eliminator, with the failure to
control the pattern of the water on the fill
side of the eliminator, or with inadequate
provision for draining the eliminator.
38
(3) The effect of efficient air handling in the
tower by the eliminators seriously changes the
tower performance and drift release rate.
The Environmental Systems Corporation was first sponsored
by the Marley Co., but later established itself as an inde-
pendent organization providing services to parties of every
interest. In 1971 ESC received grants from Environmental
Protection Agency to further develop drift measurement tech-
niques, particularly on the PILLS system. Other techniques
to be evaluated were isokinetic sampling using filter papers,
cyclone collector and glass wool fill material, sensitive
paper using milli-pore membrane filter paper, and on-line
holography. APS was developed later for airborne particulate
measurement. Numerous drift measurements at operating cooling
towers by ESC using these techniques have been performed.
Some of them are listed in Table 1.4.1.
In the early 1970's, Ecodyne developed several drift
measurement techniques for field testing. These include the
isokinetic sampling system using a cyclone separator, assembled
and calibrated by Meteorology Research, Inc., the impaction
method using FORMVAR coated slides, and sensitive paper tech-
niques. As of 1973, more than twenty types of drift tests had
been conducted on industrial towers. The tests included towers
equipped with both the standard two pass drift eliminator config-
urations typical of the industry for the past twenty years, and
a new drift eliminator developed by Ecodyne, the Hi-V eliminator.
Test results showed that drift rates for the standard two pass
39
Oc' L~ -- 4 - 4 -4-,
0 0 0 0 0 0 0 0 04 o o H o oC o O ov: O 0 0 0 0 C O I I o O O O ; O; O OO
AQa)
0-i-P
0
00CO o~~~~~~
U ) L
4-'N
) f0
r- H 0 -i* Cd 4-'
-4 4-p bO Cd* U C
H U) HCr= F-4 U)a) CH-4 0 m
H > a)
wI H
HA"0
ml C,:) co coV~~ I-
H H H H H H H H H 4 H
I. ad)
OH - %
v 3 C ) 44 0 :.. 00 "
;.4 =x O ct~ 4-3 0 Vr 1- x r4 111 -
C.~~C EA _c - ^ - 0- > M -r
:3 I0
r-i 4o0)
C)
U)a)
CHa-410.r-i
. b m 0 'H 0 -P
> ) a 0 0 ch t) 0 0 0
-P HbO H-P (1) ~~~~~~~ ~~~ ~ ~~ ~ ~~~~C -)- 4-' Z*Z 4-'H- H .0H - C-4 .'co % i
CqE H O .C 0 .* H > Q
0 z 4-'H QH 4 0"OW 0 H'd CCiCi;4 0~ *HH 4-' rl C 0)
Cd'. ac octti · d 4' oCa ' 4 HU)'H t-40 C/) Pi' C*4 0 ) CH U) .Ph1-4"C (C C< z a) 4 h^
Cd ",1 F- cdi c cd ro =r 4 ^ - M M hH ̂~ - ^: H
C) 1 C) CO 3 ho m V) C o 3 H a ) 4-) :FH S O tH~~~~~~~~~~~- H -, M Z 0,- a, cfj
CCd 43 ) (Cid ~ - 4 cr4 C)0 -4 .14 a) cr3 S:: Cio .C o H 4, · o 4
SO) :s C) C) o0 O Cr c 0 3- 3) 4-' c0 ) c 50 o H :-0 ' > * ) H o
o ¢ C) ¢ F 4 Z4-' . = : P C OZ ,- 4 i4
40
designs varied from 0.02% to 0.12% with a typical value of
0.05%. The Hi-V drift eliminator drift rates varied from 0.001%
to 0.008% with a typical value of 0.004%.
Other companies that are involved in cooling tower drift
measurements are Research Cottrell, Inc., who uses the High
Volume Air Sampling Method (L1), and the Balcke Co., who uses
the cyclone separator (R3), etc.
Although much progress has been made recently in drift
measurement techniques, disagreements and unreconciled differ-
ences frequently show up, which often involve factors of two or
three in the value of certain results. Reliable methods should
be developed soon in order to accurately access the environmental
effects of cooling towers.
1.5 Present Approach
In the present study, an analysis of the performance of
standard industrial drift eliminator devices using both theo-
retical and experimental techniques is carried out. The
theoretical approach makes use of the code SOLASUR (H2) to cal-
culate the air velocity distribution within a drift eliminator
and the pressure loss through the eliminator, using both free-
slip and no-slip boundary conditions at the eliminator walls.
This information is used to perform trajectory calculations
with a fourth order Runge-Kutta numerical technique for droplets
of a given size injected into the eliminator in a uniform
transverse distribution.
In the experimental approach, the laser light scattering
technique is used to measure the droplet size spectra both at
41
the inlet section and outlet section of the eliminator. This
drift measurement technique is selected because of its on-line
data acquisition and reduction capacity, and because of its
successful application in the PILLS system by the Environmental
Systems Corporation. The main differences between the present
technique and the PILLS system are that the laser presently
used is a steady-state laser instead of a pulsed laser as in
the PILLS system (S4), and that there is no fog problem in this
laboratory scale work. The pressure drop across the eliminator
is measured with a differential electronic manometer.
Comparison of the calculated results and the experimental
data for several drift eliminators is presented.
1.6 Organization of this Report
Chapter 2 describes the numerical model for theoretical
evaluation of drift eliminator performance. The assumptions
made in the theoretical model are also listed in Chapter 2.
The results of this calculation for some common cooling tower
drift eliminators are presented in Chapter 3.
In Chapter 4, the details of the experimental measurement
techniques in this work are described. The experimental data
is displayed in Chapter 5, where it is compared with the calcu-
lated results. The sources of experimental error, and efforts
to quantify this error are included in both Chapters 4 and 5.
Chapter 6 discusses the discrepancies between the measured
and calculated results, the validations of the assumptions
42
made in the theoretical calculations, and the usefulness
of the theoretical model. The overall performances of many
drift eliminators are compared, and future improvements are
recommended for drift eliminator design.
43
CHAPTER 2
THEORETICAL EVALUATION OF
DRIFT ELIMINATOR PERFORMANCE
2.1 Introduction
Despite the fact that many of the important parameters
which affect the performance of drift eliminators cannot be
easily accounted for, a theoretical model remains very useful
in evaluating the relative performances of different drift
eliminators, and in designing improved devices.
In order to do this, a computer program, DRIFT, has
been written to numerically simulate the performance of
drift eliminators. This chapter describes the theory of
the calculations performed by the code and the assumptions
that are made in the analysis. A detailed discussion of
the use of the code can be found in Ref. C5.
2.2 Assumptions
There are many parameters that affect eliminator per-
formance that cannot be easily included in a theoretical
model. Therefore, the following assumptions have to be
made in the numerical analysis:
(1) The air flow within the eliminator is laminar.
It was demonstrated (F3) experimentally that the
flow in a typical drift eliminator is laminar
throughout most of the eliminator volume, with
44
Reynolds numbers lying in the range 2000 to 4000.
(2) The exhaust flow field is not affected by the
presence of the drift since the drift density is
low.
(3) The flow is two dimensional.
(4) The flow is incompressible since the flow Mach
number Is low.
(5) Any water film effects on the air flow are
neglected.
(6) The initial velocity of the droplet at the inlet
of the eliminator is the vector sum of the exhaust
flow velocity at the inlet and the vertical droplet
terminal velocity.
(7) The probability of a droplet of a given size
entering the eliminator inlet at any location is
uniform.
(8) There is no droplet mass loss due to either
evaporation or friction.
(9) Interactions among the droplets can be neglected
since the drift density is low.
(10) The drift is eliminated if it impinges on the
eliminator walls, i.e., the "bounce" effect and
any water film effects are neglected. Re-entrain-
ment of water droplets from the water film on wall
into the exhaust flow can be neglected if the
drainage is properly designed and if the film
45
thickness is sufficiently small so that film
surface instabilities do not develop over the
anticipated range of exhaust speeds.
Further discussion of the validity of some of these
assumptions is presented in Chapter 6.
2.3 Calculation of Air Flow Distributions
In order to calculate the droplet trajectory within an
eliminator, it is necessary to know the air velocity distri-
bution within the eliminator. In all previous studies
either a uniform flow distribution (R3), or potential flow
(F3,Y2) is assumed. In this study, the flow distribution is
calculated by the SOLASUR code (H2) which is included in the
DRIFT code as a subroutine. In the original SOLASUR code a
free-slip boundary condition is used at the rigid boundaries
of the eliminators. In this work the option of a no-slip
boundary condition at the rigid boundaries has been added to
the code so that the mass-averaged total pressure drop between
the inlet phase and the outlet phase of the eliminator can
be evaluated. It is found that the flow distributions calcu-
lated with no-slip boundary conditions look more realistic
than those with free-slip boundary conditions. Also, the
collection efficiencies calculated with these more realistic
flow distributions agree better with measured values. All of
these results are shown in later chapters.
The SOLASUR code is a modified version of the SOLA
code for calculating confined fluid flows having curved
rigid or free surfaces as boundaries. It solves the two-
dimensional, transient Navier-Stokes equations for an incom-
pressible fluid using an implicit finite difference tech-
nique. This technique is based on the Marker-and-Cell (MAC)
method (Hl, W). The description of a flow transient pro-
ceeds step by step from an assumed initial velocity field to
an asymptotically steady final exhaust flow distribution.
The time step size is determined from numerical stability
considerations (H2). The fluid region is made up of uniform
rectangular cells, and is surrounded by a single layer of
fictitious cells as shown in Fig. 2.3.1. Fluid velocities
and pressures are located at cell positions as shown in Fig.
2.3.2; horizontal velocities at the middle of the vertical
sides of a cell, vertical velocities at the middle of the
horizontal sides, and pressure at the cell center.
The procedures involved in one calculational cycle
(one time step) consist of:
(1) Computing guesses for the new velocities for the
entire mesh from the difference form of the Navier-
Stokes equations, which involve only the previous
values of contributing pressures and velocities
in the various flux contributions. The velocities
at boundary cells are adjusted so that the boundary
conditions are satisfied.
47
co
C)I,:
Fm
t)0U)bO
Nd4-
U)0Cd
C
a)
arr:
P(rd
-l0,cor.
a)
Cd
v0'0Cd
ul
a)0Cd
0r-0f
rn
CM
FX4
48
a)r-l
0
Q)
a)
rtC
a)
.,i
-e CO
b0 E
c
Ckl
rr,
l-'j
49
(2) Adjusting these velocities iteratively to
satisfy the continuity equation by making
appropriate changes in the cell pressures. In
the iteration, each cell is considered successively
and is given a pressure change that drives its
instantaneous velocity divergence to zero, thus
satisfying the continuity equation.
(3) When convergence has been achieved, the velocity
and pressure fields are at the advanced time level
and are used as starting values in the next calcu-
lational cycle.
The above procedures are repeated in each time step
until an asymptotic distribution is reached. The results
are then used for droplet trajectory calculations in the
main program. The flow chart of the SOLASUR subroutine is
shown in Fig. 2.3.3.
In the original code, free-slip boundary conditions
are used at the rigid boundaries (the top and bottom bound-
aries), where-in each top surface cell the u-velocity in the
top fictitious cell (the cell above the surface cell) is set
equal to the u-velocity in the top surface cell, and for
each bottom surface cell the u-velocity in the bottom
fictitious cell (the cell below the bottom cell) is set
equal to the u-velocity in the bottom surface cell. In the
DRIFT code, no-slip boundary conditions were added as an
50
SET INITIALCONDITIONS FOR
PROBLEM
.COMPUTE INITIALGUESS VELOCITIES
. SET BOUNDARY1"r T 'T T - fnN I
I .IJ1lJJ I
/HAVE PRESSRRESCONVERGED. _ F
1-. -A -
NO
UVL)A It t LLPRESSURE ANDI VELOCITIES i
COnNDTTTONS I
UPDATE FREEURFACE POSITION
ADVANCE TI ME]AND CYCLE
IPRINT & PLOTI
IHtbUKt ROPCCUL TION!
(TIME TO STOP?>
IYES
STOP
Fig. 2.3.3 Flow Chart of the SOLASUR Code
v.
I i II 11
rs _ _ -
I V 114 I? I JSET SPECIA
~v·u ., . .ll .........
_ I... i I
II I II
II
. Ir-_^ ^. .,I;C ~A~I
NO -
I-
A . .I
I
option, where the u-velocity in the fictitious cells at the
top and bottom boundaries are set equal to the negative u-
velocity in the top and bottom surface cells.
A detailed discussion of the SOLASUR code is given in
Ref. H2. Results of air velocity distribution calculations
are presented in Chapter 3.
2.4 Pressure Loss Calculations
The pressure loss of the air stream flowing through an
eliminator is an important factor in designing a drift elim-
inator. A large pressure drop will reduce the tower cooling
capacity and will thus either increase the capital cost or the
operating cost of the tower. An estima.te (Gl) reveals that
a flow resistance of three velocity heads (=AP/½ pV2 ) will
increase the final temperature of the condensate by approxi-
mately 0.20C. This seems to be a very small increase, yet it
is significant in terms of overall station economics, bearing
in mind that 10C is valued at about $3M over the life of a
2000MW station. The flow resistances of current industrial
drift eliminators range from two to ten velocity heads.
Prediction of the pressure drop across an eliminator is
complicated by the fact that flow separation occurs in most
eliminator geometries, and this induces a large pressure drop.
Yao and Schrock (Y1, Y2) calculated the pressure drop across
drift eliminators using the method of Lieblein and Roudebush
(L2), in which the total pressure loss is expressed as a
52
function of boundary layer thickness, provided that no flow
separation occurs in the eliminator. The hydrodynamic boundary
layer thickness is determined by an approximation method
proposed by Thwaites (T5).
In the SOLASUR code values of pressure are calculated
at all cells. Using no-slip boundary conditions at the
rigid walls, the pressure drop can be calculated. Assuming
equal air density at the inlet and outlet regions of the
eliminators, the mass averaged pressure loss is defined as
JT JT
U2j P2,j E ;IBAR,j , PIBAR,jAP = j=JB - JJB (2.4.1.)
JT JT
EJ Buj E UIBAR,j
where the summation is from the bottom boundary cell (JB) to
the top boundary cell (JT). i=IBAR is the outlet region,
and i=2 is the inlet region. uij and Pi j are horizontal
velocity component and pressure at cell (i,j), respectively.
This pressure loss calculation is performed at each time step
until a steady state value is reached. Results of this calcu-
lation are presented in Chapter 3.
2.5 Droplet Trajectory and Collection Efficiency Calculations
The droplet collection efficiency of an eliminator is
generally defined as the ratio of drift mass collected by the
eliminator to the drift mass entering the eliminator. It is
customary to study eliminator efficiency only in terms of its
effect on the total mass of droplets leaving the cooling
53
tower. However, it is currently known that the droplet size
distribution also plays an important part in determining the
nature of drift deposition. Therefore it is necessary in
evaluating an eliminator to investigate the variation of
eliminator efficiency as a function of droplet size. The
collection efficiency is defined as
N (d)n(d) = , (2.5.1)
where NC(d) represents the number of droplets of diameter d
being-captured by the eliminator, and Ni(d) represents the
number of droplets of diameter d entering the eliminator.
In the numerical simulation process, a certain number
of droplets of a given size are injected into the eliminator
with a uniform transverse distribution. By calculating their
trajectories within the eliminator, the number of droplets
that encounter the eliminator boundaries and are then assumed
to be captured can be found. The collection efficiency of
the eliminator for this droplet size is then determined from
Eq. 2.5.1.
The drift trajectory is calculated by solving the droplet
equation of motion. For a sphere moving in a flow field, the
general solution is governed by the momentum equation (M4)
dVd CdRedV
md = 6 aR( Va -Vd 24 + md , (2.5.2)
dt
where
= 1 + 0.197Re0 ' 6 3 + 2.6xl0-4Re 38 (2.5.3)
2 V-Vd lRpV-Vd RPa (2.5.4)
pa
For a spherical water droplet, Eq. 2,5.2 can be
dVd = 9 "a
dt 2 pwR2
Cd Re
21 (V a-Vd ) (2.5.5)
The symbols appearing in the above equations are:
md = droplet mass
Vd = droplet velocity
t = time
air viscosity
R = droplet radius
air velocity
Cd = drag coefficient
Re = Reynolds number
g = gravitational acceleration
Pa = air density
Pw= water density
Equation 2.5.5 is a nonlinear differential equation.
A fourth-order Runge-Kutta numerical analysis is applied to
54
C dRe
24
and
Re =
fied;
simpli-
1a
Va =
55
determine the droplet trajectory. At any time step, the
position of the droplet and its velocity are found. At each
location the air velocity is interpolated from the cell values
calculated by the SOLASUR code. The air velocity at the
beginning of each time step is used throughout that time step,
and the local drag coefficient and droplet acceleration are
calculated from these velocities and from the local values
of the remaining parameters.
A variable time step size is used in the calculation.
The step size is determined from a consideration of the
propagation of errors in the following manner: For a differ-
ential equation of the form
dVd = f(t,Vd), (2.5.6)dt
the error at time step i + 1 in the fourth order Runge-Kutta
method is (C1)
af hi+1-l d' (.5.7)ti ,c
where a is a velocity value somewhere in the interval between
ti and ti+l, E is a time value somewhere in the interval be-
tween ti and ti+l, and h is the time step size.
The first term on the right hand side of Eq. 2.5.7
represents the propagation error, and the second term is the
local truncation error, which is generally small for small
values of h. Then, if av is negative, a value of hvalues o h. Thend tia
56
afcan be found which will make (l+hav )<1 , and thedterror will tend to diminish or die away, so the solution will
be stable. For the cases considered, aV is always found to
be negative, so by specifying a proper value for the step
factor, h , a stable solution can be obtained. A large
value for this step factor will yield a smaller propagation
error but a larger truncation error. A small step factor will
result in too small a step-size, thus prolonging the compu-
tation. A step factor of 0.1 has been found to be satis-
factory for the cases under study. In the present model,
af/aVd is determined at the beginning of each time step
using the local values of droplet velocity and air velocity.
The step size of this time step is then the constant step
factor divided by af/ Vd.
If the eliminator is installed in a vertical scheme, the
droplets are assumed to enter the eliminator at a velocity
which is the difference between the air velocity and their
terminal velocities. The terminal velocity of a droplet of
radius R is determined from Eq. 2.5.5 by requiring dVd/dt
to be zero. Thus
2 Pw gR
Vt V- - 2 pgRif (2.5.8)Vt Va-Vd -9 1IaCdRe
24
57
This nonlinear algebraic equation is solved by Newton's
method of tangents with a calculational accuracy of 0.1%.
If a droplet enters the eliminator other than vertically
(as in a horizontal scheme), then the initial velocity of the
droplet will have a vertical component which equals the
difference between the vertical component of the air velocity
and the droplet terminal velocity, and a horizontal component
which equals the horizontal component of the air velocity.
Droplets of a certain size are introduced uniformly
across the inlet of the eliminator. The trajectory of each
droplet is calculated until it either hits the eliminator walls
or passes through the eliminator. The collection efficiency
for this droplet size is then the ratio of the number of
captured droplets to the number introduced at the entrance.
The number of droplets introduced at the entrance determines
the accuracy of the collection efficiency calculation. If Nd
droplets are introduced uniformly at the entrance, then the
error in the collection efficiency calculation will be
proportional to 1/Nd. In the DRIFT code, a provision is
made for testing a finer distribution of droplets at the
locations where the condition of trap and escape changes
between two adjacent droplets. This method greatly improves
the accuracy but does not demand too much computation time.
A flow chart of the DRIFT code is presented in Fig. 2.5.1.
Trajectory plots and calculated collection efficiencies
are presented in Chapter 3.
58READ INPUTAR
READ INPUT PARAMETERS
CALCULATE THEAIR VELOCITYDISTRIBUTION BYSOLASUR SUBROUTINE
/TO CALCULATECOLLECTION
NO \FICIENCY? / YES
NO OES USER SUPPLY\THE AIR VELOCITY _1 I;RIBUTION? /
YES
INPUT AIR VELOCITYI DISTRI BUTION I
J CALCULATE DROPLET i1 TERMINATE VELOCITYI
FOR ALL GIVENDROPLET SIZES IFIND INITIAL CONDITIONS-
CALCULATE THE- IS THE ENTRANCE POINTCOLLECTION YES OF THE ELIMINATOR?EFFICIENCY
NO
FIND CONDITIONS AT BE GINNING OF TIME STEP
CALCULATE TIMEL - I DECREASE ENTRANCE ISTEP SIZE DIVISION SIZE FOR -
IMORE ACCURATE RESULTI
YESiCALCULATE CONDITIONSAT END OF TIME STEP I
DETERMINE THE ELIMINATOR BOUNDARY
IS DROPLET OUT OF \ELIMINATOR BOUNDARY?/ NO
lT C ThDDI CT AT DD\ITfIIC \NO
I.0 . L C~ u 1 Ml F I v iv uo I O %ENTRANCE POINT TRAPPED?--
YE
IS DROPLET OUT OF NOTOP OF ELIMINATOR?
EIS DROPLET AT PREVIOUS N I "I-"". "'l tNTRANCE POINT TRAPPED? MORE ACCURATE RESULT
[RECORD TRAPPED DROPLE
RECORD TRAPPED DROPLET-
I nFrFDAF F:NKITDAAIIc I
Fig. 2.5.1 Flow Chart of the DRIFT Code
YES
I
I B I
- i
Jl
I
,i II
B i I II L
s, i , i
. .
.. 1
I
!
I ...... I m ! i ~ .............. I
I1
:s _._ _._ _. - - ---- --;-.1
1
59
CHAPTER 3
RESULTS OF THEORETICAL CALCULATIONS
3.1 Introduction
This chapter presents the results of the calculations
performed with the DRIFT code. The air velocity distributions
in some common cooling tower drift eliminators are calculated
by the SOLASUR subroutine. The calculations employ both
free-slip and no-slip boundary conditions at the eliminator
walls, and results are compared and discussed in Section 3.2.
The calculated droplet trajectories within these eliminators
are presented in Section 3.3. The collection efficiencies
calculated from these trajectories are compared with those
obtained from other sources in Section 3.4. The last section
of this chapter presents the calculated pressure loss across
some common industrial drift eliminators.
Table 3.1.1 tabulates the physical dimensions of the
eliminators under study. The case numbers in the table will
be referred to throughout this thesis.
3.2 Air Velocity Distributions
In this section, ir velocity distribution plots for
some drift eliminators are presented and discussed. In the
plots, the length of the line segments are proportional to
the magnitudes of the velocities at the mesh points, and the
directions of the lines represent the directions of the
flow at the mesh points. In all of the cases presented here
II-% 0 0)--T 0 o
cN 0
r- coH r
C(m
01-.I
IO-4
I Iin i n
I JI ~ x
O0 co l r0 0 0rH - - r- H rH
C\j 'IO 0 0 0 1O t-- Ln U Lf Ln
a) at) )
0 0 0
a) a) a a
Cd Cd Cd 4-)
I II,) a) 0 4)
H H HCdbD a a Ql]
0)
d
a) a I
ar-co CdC) C) boI I H
CO COO)0
4)
a,)c)
4)U)
CO¢,
bOCd
II bO
Lnb--
Hm m0
-T CY)
Wrw
r-I ,J r- r-IH~5 H H H H
0o
Cd
HH0 UIH'z
60
E4)
,1)
.I
o-
E0.r-r
HC
'I)
CCO
r-4 O01)04-Cd
-HE-Hr-A
0CO
tq)d
C)
r-IC(
I
Cl C)QI a
61
the air flow direction at the inlet of the eliminator is
normal to the flow channel cross section. The gravity
effect on the air flow is negligible; therefore the air
velocity fields will be assumed to be the same whether the
eliminators are installed horizontally or vertically.
Figures 3.2.1 and 3.2.2 display the air velocity distri-
butions for a single-layer louver eliminator as calculated
by the SOLASUR subroutine using free-slip and no-slip
boundary conditions at the eliminator walls, respectively.
With the free-slip condition, the calculated velocity is
quite uniform ( see Fig. 3.2.1). With the no-slip boundary
condition, the calculated velocity field,shown in Fig. 3.2.2,
is more realistic. Near the lower boundary a wake region
can clearly be seen. Such a wake region is expected in the
real flow. Note that the velocity in both cases is mainly
parallel to the duct boundary, thus, the collection efficiency
can be expected to be low for this type of eliminator.
Figures 3.2.3 and 3.2.4 show similar air velocity fields
for a two-layer louver eliminator. The free-slip prediction,
Fig. 3.2.3, shows a nearly uniform distribution except at
the turn in the eliminator where the velocity decreases as
the radius of curvature increases. With the no-slip condition,
the velocity distribution plot, Fig. 3.2.4, shows very small
velocities at the lower boundary in the first half of the
eliminator, and at the upper boundary in the second half of
the eliminator. In fact, these are the regions where a wake
is expected.
62
-_\An\\\~\\\ \\\ \\\ \\\\\
~-`\\\\\\\\\Ad\\\\ \\\\\\
\\\\\\\ \\\ \e\\\\\
\\\\
Fig. 3.2.1 Velocity Distribution of Air Flow in Single-
Layer Louver Eliminator Using Free-Slip
Conditions at Upper and Lower Boundaries
63
\
NNN-\\\`% '\\
% \\% %
Fig. 3.2.2 Velocity Distribution of Air Flow in Single-
Layer Louver Eliminator Using No-Slip Conditionsat Upper and Lower Boundariesat Upper and Lower Boundaries
~ \ \ \ \ \ \ \ \\\\\\\\\\
\\\\\\\\\\\\\\\\\\\\ E\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ ~~~~~~~~\\\\\\\\\\\\\\\\\\\\ \ \\\\\\\\\\\\\\\\\\\\ 0)\\\\\\\\\\ \\\\ \\\\\\\
\\\\\\\\\\\\\\\\\\\\\ 0
\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ 0 c
\\\\\\\\\\\\\\\\\\\\
\ll I rlI 0)
/ I I I I I I Ir I I Il r lr I I I j ,,, o r
//////11 1 t! / t/ / / /////!/!!Illlrii c .¢
/111/11II~~~~~~ // / / /// /. ¢ 0///11I!!!t~//////// / = o-P0I~~//11/!/ / / /I / / / / / / .a
Iii1111./ / / / ~~/ ., ./.,/., / / //¢ -P riHI0~ )///////////// /., /., h /
IIII!!!!11I //.I / / r / °
I!11!!!!!1/~/I/.,,/ o bC
0l d//111/1 1 / //./ / // / , / I / /1i1II!I!!I / / / / / / / / / / /
C-'J///1///lt ! / //// ////I!/////ll/ c,, ,,11 ////1111! ! // /////11//111!!1111//
111111111111111 11111 //Il1111111I I J I I I ! I I ! Z / o s
.. ,,rrrrir\\\\..\\\\, ' \\\\\\\\\\\\\\..\\\\\\\\\\\\\\\\ %\\\\\\\\\\\\\\\\\ E
ii/IIIIIIItI//
I/ //I//I ///////////////// ////
I//I//I//I,,/ /1///f//I/f//I///// / / / 114
It11////f//1//f/Il/f////1/ /i / ,.1/f/I/f//f,,/ I/I/u, f//f//Il/Il///////,,./I \/\ \/\ \/\\\\\/11111 \ //\\ \\\\\/I/1111111////////,.
/1111ll~ II I I I I !I 11/
11I/111l111111111111
IIIII1111111111111 I.,.\\ \ \ \\\\ \ \ \ \\\\
.\\\ \ \ \ \\ \I \ \ \ \ \,
, \ \ \ 1 18 \ 1 / I \I I II ~I II
I I t I /I If I II
///////////// / /
///////////// / /,,
////////////,X, . .
/////// / / / // /,. . .
//// / / / / / / / ,/,...
J//// / / / // / / ,/,. ..
/ //// // IIIIl/ / / . ..
///// /IIII {III/f/,,
1 | | 1 | 1 1 1 1 | |1 1
65
\\'
\\\\
III1%
111
IIIItIll,
IIIII
II,,,ffI
I,
I .a ..
f .
0o
r4-
-CH
: O0-i¢)
O 0H d
O C
0
o 0
I
...-- t
0c a --H
'ItI I
It
66
Figure 3.2.5 compares the air velocity fields for a sinus-
shaped eliminator calculated with free-slip and no-slip
boundary conditions. In the free-slip case, A,the velocity
distribution is approximately uniform. At the mid-length of
the eliminator the velocity decreases as the radius of
curvature increases. Also, at the high pressure sides of the
eliminator the velocity is slightly greater than that at the
low pressure sides. It can also be observed at each trans-
verse cross section that the maximum velocity always occurs
at the wall. This is not true in the no-slip case, B, where
the maximum velocities occur at short distances away from the
high pressure walls, and approach a value of zero at the walls.
At the mid-length of the eliminator the maximum velocity
occurs close to the center of the cross section. Note that
the velocities shown at the upper and lower boundaries of
the eliminator do not represent the velocities exactly at
the walls, but rather at short distances away from the walls.
It can also be seen from these plots that the no-slip results
predict a more realistic flow because they show the wake
regions. This will be illustrated with the help of flow
visualization photographs in Chapter 6.
Similar observations can be made regarding the velocity
fields of the Hi-V and Zig-Zag type eliminators shown in
Figs. 3.2.6 through 3.2,8. For these more complicated
geometries it is anticipated that significant recirculating
eddies and turbulent wake regions exist at and near the bends
67
... \\\\\\., \\\\\\, / \ \ \ \ \ \\\./\\/ \\\\\ \\
I.\\\\\\\ I
i\ \ \\\\\\\.
C li
////////,, , . H oH
Kr , 0 _d 0 II I I I I I I
/ II I I I C O
H 4 -/1////,/,/. U) 0
l////l/ //, , o ///////I,,
'Cl
0 CO Z
//I//I"" ' t 4-
/ HlI!1~ I I4 U \ I I
\I I \\ \I I , * . . H I O0\ I\ \ \ \ I. I \ ,, 4 Z;
a: ( l I I 1 * . . .
it II>/ I I I I , .I//I/, " ",
/ I / I I I I t , . .: ,
f//// / -.///l I / I /-
68
... \\..\\O 0.. .. ,), .~ ~ \\\\.. . \\\\\.. ~ \\\\\\. \\\\\
aIItI ,0 E ./l/I/cl ·0 4i ,..////, , ... C //////, . ! 0
* Billies. ,
///0 *IoH O 0. , * , I I, , I 4 )
\\\ \\\\ I U)...x
Cdo C IoI It I I C iI Q O
//1 / / / .//I I/ / / //I I @ I I *
' ///'/'/I ,
I gI1ll1 |~~ ~~~ /
/I/ / II // / / / * , , .
/ / / / / / / / ///// / / / / /
/ / / / / / / / /
/ / / / / / / / /
// z / / //-/ I, 1,11, , I I I I I /
, ,. , , , I ~ \ \
% 4' ' 4' 4 '\ \\ \ \ \ \ \ \ \
\ \ \ \ \ \ \ \ \
\ \ \ \ \ \ \ \ \\ \ \ \ \ \ \ N N\ \ \ \ \ \ , \\\ \ \I \\\ \ '4' 9,
\ I I ,
/ / I I I , , ,
// / / i /// / / / / ,
// / // / I I I I I,,
/ I / I I I I///////I///f/ // /
I I I I I I I I I
69
bO
0
C03 C i)
haS
o0t0 Cb- 00
ro
4 00 z)
0i i
r-trk4
/ /
/ / /
''1/ / / /
.. iI / / / // / I I / /
* * *
* I I I I /
* · . , , I \ \ \* . X 8 \ \\* S S 8 \ \ \\\* . % \\ \ · ~~ . . \ \ \ \.~~ ., \ \ \ \
.~~ . \ \ \\· \ \ '.,\
~~ fI//lI,I////////I//
//////,///////,
///////, / / .~~~~~~~- -
/ / / / I, I / I I
/ I t , . .
, . ./ I, o
I
I
II
II
I
I
II
IU
U0
a
I* SS *. .
70
0
a)r d
CI
o 0oH C0
0
0 D
D CO4H4) CO
.0
CD
H a
o H0 c.
q) 0
CO
-H
0 71
in the flows, From the distributions shown it is concluded
that the no-slip results predict more realistic distributions
than the free-slip results, which will also be discussed in
Chapter 6.
For the Zig-Zag eliminator with no-slip boundary condi-
tions, the calculation fails to achieve a steady-state value.
Or, if a small mesh size is used for a more accurate determi-
nation of the actual flow, the calculation fails, These
effects are due to the fact that in the actual flow, turbu-
lence is very significant for this eliminator. This will be
shown in Chapter 6. The effects of turbulence are not taken
into account in the current calculation. The recirculating
eddies in interior corners are not fully resolved in the
solution. This could be done by using a finer calculational
mesh. However, this was not done since the mesh size already
is in a range for which the calculated droplet capture
efficiencies are relatively insensitive to the choice of mesh
size. In addition, the droplet capture dynamics are relatively
insensitive to whether the calculation of an eddying region-;is
exact or if the region is treated as being approximately stagnant-
which is what occurs with an inadequate spatial resolution of
the calculated flow field.
The flow structure of any turbulent wake flow cannot be
resolved by simply using a finer mesh, but it could be treated
explicitly with a "turbulence model" calculation (for which
several different computer programs are available). However,
72
in view of the success of DRIFT in predicting the experiment-
ally observed behavior of the drift eliminators, it was
decided that a turbulence model calculation would not be
required. Effectively, the error introduced into the capture
efficiency prediction by a failure to describe turbulent
eddy regions is relatively small. This is mainly true because
the devices examined have droplet trajectories that result in
captures lying far from these wake regions. This will be
discussed further in the following section and in Chapter 6.
Figures 3.2.9 and 3.2.10 show the results for the E-E
eliminator designed by Yao and Schrock (Yl, Y2). The criterion
for the design is that in order to minimize the air stream
total pressure loss, any flow separation of the hydrodynamic
boundary layer from the walls is to be avoided. Separation
can be avoided if the air velocity increases monotonically
along the flow direction. This is done by making the cross
section of the flow channel decrease monotonically. After
using this criterion and examining several geometries, it
was found the E-E eliminator performs satisfactorily. Looking
at the velocity distribution calculated by the SOLASUR sub-
routine using a free-slip boundary condition (Fig. 3.2.9), it
is found that the velocity increases monotonically along the
flow direction except near the outlet of the eliminator.
However, the result predicted with a no-slip calculation
(Fig. 3.2.10) indicates that separation does occur and that
a wake region exists at the upper boundary just after the
73
/s
//
////
////////////////-- /t///1 /
Fig. 3.2.9 Velocity Distribution of Air Flow in E-EEliminator Using Free-Slip Conditions atUpper and Lower Boundaries
* a
Ill//f/I//, / ////I// ////
v - *% -e *~ * , / //////
4 4 4 4 4 4 i s 4 4 A- - - - - - - - . - - //////// / -
__________________________ofcrrrr~)l/// / / I
/4 / //111111
Fig. 3.2.10 Velocity Distribution of Air Flow in E-E
Eliminator Using No-Slip Conditions at
Upper and Lower Boundaries
75
sharp turn. As a result of this, the pressure loss calculated
by the DRIFT code is much higher than that predicted by Yao
and Schrock (Y1 and Y2). The collection efficiency calcu-
lated by the DRIFT code using the free-slip condition has a
velocity distribution that is very close to that calculated
by Yao and Schrock. Using a no-slip condition, the collection
efficiency calculated by the DRIFT code is not significantly
different. These results will be developed in the following
sections.
3.3 Droplet Trajectories
The collection efficiency of a drift eliminator at a
certain droplet size is determined theoretically by uniformly
injecting droplets of that size into the inlet of the elim-
inators and observing their trajectories inside the elimina-
tors. If the trajectory of a droplet ends at an eliminator
wall, then that droplet is assumed to be captured. If the
trajectory exits the eliminator without touching the walls,
then the droplet is assumed to have escaped the eliminator.
By comparing the numbers of escaped and captured droplets,
the collection efficiency can be determined.
Droplet trajectories are calculated by solving numeric-
ally the droplet equation of motion within the eliminator
flow field as described in Chapter 2. Figs. 33.1 through
3.3.13 illustrate the droplet trajectory plots for seven
different eliminator geometries, with all eliminators assumed
76
to be in vertical scheme, and droplets entering the eliminators
at the left of the figures. For each geometry the trajectories
for two droplet sizes are shown. It can be seen in all of the
eliminator geometries that larger droplets are more easily
captured. In fact, it will be shown in a later section that
for any eliminator and any air speed, the calculated droplet
collection efficiency increases monotonically with droplet
size. This occurs because the net acceleration on a drop varies
approximately as l/R, where R is the droplet radius. To see
this, note that at a given air speed, the aerodynamic drag
force increases approximately as R2 (see Fig. 2.3.1), while
the particle mass varies in proportion to R3, and the time
during which the particle is affected by drag is approximately
constant. Thus, the net acceleration on a particle, and the
resulting displacement vary approximately as R, resulting in
easier capture for larger droplets.
Figures 3.3.1 and 3.3.2 show the droplet trajectories
within a single-layer louver eliminator determined by using
a free-slip air velocity distribution. It is observed that
the paths of the droplets are essentially parallel to the duct
boundary except at the entrance where the inertial motion of
the droplets carries them towards the duct boundary. It is
because of this inertial effect that the droplets are trapped,
and it can be expected that they will be trapped at the
boundary towards which they are initially directed, in this
case being the upper boundary. Thus, it is concluded that as
77
DRBOPLET D IA= 40 MICRON
Fig. 3.3.1 Droplet Trajectory Plot for Single-Layer
Louver Eliminator with Droplets Entering
the Eliminator at Left of Figure. Droplet
Size is 40 pm
78
DROPLET DI R= 100 MICRON
Fig. 3.3.2 Droplet Trajectory Plot for Single-Layer
Louver Eliminator with Droplets Entering the
Eliminator at Left of Figure. Droplet Size
is 100 m
-'
ji
I:
0
k"
S<
I
79
the duct boundary steepness increases the collection efficiency
will also increase since the area of the boundary towards
which the droplets are initially directed is increased.
Figures.3.3.3 through 3.3.6 display similar sets of data
for the two-layer louver and sinus-shaped geometries. In
these two cases the length, pitch, and entrance conditions
are the same in each of the two geometries in order to compare
their drift collection efficiencies. The trajectories are
calculated using free-slip boundary conditions in the air
velocity determination. Their collection efficiencies are
tabulated in Table 3.3.1 as a function of droplet size. The
fact that the sinus-shaped geometry has a higher collection
efficiency can be explained with the fluid velocity vector
and droplet trajectory plots. Comparing these figures, it is
seen that the sinus-shaped geometry has a steeper slope at the
entrance than does the two-layer louver eliminator, and this
is where most of the smaller droplets are trapped (see Figs.
3.3.3 and 3.3.5). For larger droplets, the number trapped
near the turn in the duct becomes significant. However, the
number is about the same for both geometries, as shown in
Figs. 3.3.4 and 3.3.6. Therefore, in order to collect small
droplets a geometry that has a steep slope at the entrance
should be used, while for large droplets both slopes are
important. Thus, if the drift size distribution is known, an
optimal drift eliminator for that distribution can be indicated
in this manner.
4
zIEiT"O
80
Q4
o
o 0
O
0
cti
H .0QdO O
.4-
a) vo v
4-)
04 IO S
a~ r.~·r
.i
b Bc
II
CC
LLJ
tL-
EDCr
ci
4.9'
;L
I:
81
4)
0
r. O
O H
H 4-
. oa)aH 00
H O
0O CO 4
0 -Pa
bO
a) -H O
a)
.aIe
0CE
t-LIr-
E
a
ZID
DI
82
o o
oC''
o 4.r4 N
EdH a
0
coa)
a bo
I C
O
0 0
r 40) 0
E
o -
o aH 4-3
O ·,r\ Ca, I0)
0.U-
IICE
-JED'mrC)
83
o.)H OO r
OC, 0-
I Hr 0
4 0co
0 40
O O0 0
CE -
O p,4-)0
,.-
0 Hk- 4-,Howe
4) ~U)
H
04)
\9z
(m~
Z:
I'C)
O
II
CD
r-D
CDOS
84
Figures 3.3.7 and 3.3.8 show the droplet trajectory plots
for the asbestos-cement eliminator using a free-slip air
velocity distribution. Trajectory plots for those using no-
slip air velocity distributions in a Hi-V eliminator are
shown in Fig. 3.3.9. By observing the results of all of the
essentially two-layer type eliminator geometries (two-layer
louver, sinus-shaped, asbestos-cement, Hi-V), it is concluded
that in the first half of those eliminators, the droplets
are trapped on the upper duct boundaries, while for the
second half of the eliminators, the droplets are trapped on
the lower duct boundaries. It can therefore be expected
that most of the water loading on these eliminators will occur
at these two regions. In designing an effective drainage
technique for this water loading, special attention must
be paid to these two regions. One drainage technique is
to put small grooves on the eliminator surfaces at these two
regions to direct the water away.
Another point can be made by observing the traject-
ories in these two-layer eliminators. The droplet trajector-
ies shown in Fig. 3.3.9 for the Hi-V eliminator are a good
example. It is seen that few water droplets enter the two
regions where turbulent wakes are expected to occur as
discussed in the previous section. These two regions are
near the lower boundary for the first layer of the eliminators
.85
r- rao E
r-I4 i
I0HU a
4 COH) o
r Oco 0
q)
O a
O O
0C
E-4
4, C
0.,
zccI6-
IIFI
oLL
_J
zED
C:)
,,--
IICE
F-I
_J
-oC
86
4O3a)
o04
r9-
-
4-Is:
l)
-oQ)
UI
04-,
c,
4,
04
EO-
o0o
U)r-
N
co4-,
a)
or4
a,0
eL
,-I4D
,r
4-0o
4--
coCO
bo"-4
.i
N
Cl
a)H
O ~ O0 r-t0
2 NI E 4 3
O a
O bOH S
>owo> ) o0)
4-) CO F-
o h
O 4a
_\
.
:OO~
ZC
c:OwE
i-
ll
a:
ZID
cc
I-o=r
LI
0.CDE3
88
and near the upper boundary for the second layer of the elim-
inators. This means that the turbulent wakes will not have
a great effect on the droplet trajectories. Therefore, even
if these wake regions are not described exactly, the calcu-
lated collection efficiency results may approximate the results
of an exact flow field, as long as the velocity distribution
is calculated using no-slip boundary conditions which will
show the general pattern of the flow field. Further discussion
will be made in Chapters 5 and 6.
Figures 3.3.10 and 3.3.11 show the droplet trajectories
in the Zig-Zag eliminator, using no-slip boundary conditions
for the air velocity calculation. It is seen that all droplets
that are captured were trapped in the first two layers of the
eliminator. Therefore it is doubtful that a third layer is
necessary. Figs. 3.3.12 and 3.3.13 display similar data,
but with the third layer of the Zig-Zag eliminator removed.
It is observed that the capture efficiencies are about the
same for the two droplet sizes shown. This is in fact true
for all droplet sizes, though the Zig-Zag eliminator has a
slightly higher efficiency at smaller droplet sizes. It is
expected that the drift collection efficiency for the two
cases will not differ significantly,and therefore it is
suggested that two layers of the Zig-Zag eliminator will
probably suffice. If this is done, then the pressure loss
will be significantly reduced.
Figures 3.3.14 and 3.3.15 display the trajectory plots
89
4;3
Cdo4o
o )h N
Cd
4-3 -p
0 0a
cr-
CCd 0
h 4 OH ob O
0 0
0 .So
EbO4o
C4 o
r.
Z
_
IICE
I
JD
o0Dr
90
E
o
o tae0 a)
O N
2 -HH a)
CO
I4 0
oN
O ObO
C)
o IO 0
h C
E-4bO
a)
HC
-H
zC)
Z
(D
cE:
LJI
-
cD
91
DROPLET DIR= 30 MICRON
Fig. 3.3.12 Droplet Trajectory Plot for Two-Layer
Zig-Zag Eliminator with Droplets Entering
the Eliminator at Left of Figure. Droplet
Size is 30 m
92
DROPLET DIR= 60 MICRON
Fig. 3.3.13 Droplet Trajectory Plot for Two-Layer
Zig-Zag Eliminator with Droplets Entering
the Eliminator at Left of Figure. Droplet
Size is 60 pm
93
DROPLET DI R= 30 MICRON
Fig. 3.3.14 Droplet Trajectory Plot for E-E Eliminator
with Droplets Entering the Eliminator at Left
of Figure. Droplet Size is 30 m
94
DROPLET D I AR= 50 MICRON
Fig. 3.3.15 Droplet Trajectory Plot for E-E Eliminator
with Droplets Entering the Eliminator at Left
of Figure. Droplet Size is 50 m
95
for the E-E eliminator. It is seen that the capture efficiency
of this eliminator is indeed very high, as predicted by Yao
and Schrock (Y1, Y2). However, it will be shown in Sec. 3.5
that the pressure drop across this eliminator calculated by
the DRIFT code is much higher than that predicted by Yao and
Shrock.
3.4 Collection Efficiencies
The droplet collection efficiencies of drift eliminators
are calculated from the droplet trajectories through the
eliminators. These trajectories are obtained by using either
no-slip or free-slip predictions of the air velocity field.
Figure 3.4.1 compares the collection efficiency results
calculated by the DRIFT code using a free-slip air velocity
field with those calculated by Roffman et al. Roffman et al°.
used an analytical formulation for the estimation of drift
eliminator efficiency by assuming that the drift flows longi-
tudinally at the assumed-constant vertical air velocity within
the eliminator, and that it experiences transverse viscous
drag due to the transverse air velocity component which is
obtained by assuming that the air velocity at any point in
the eliminator is locally parallel to the eliminator wall.
For complex geometries the model uses a Fourier series expansion
of the transverse velocity component in terms of the duct
contour. Results obtained by the DRIFT code using free-slip
boundary conditions for air flow field calculations are shown
100
DROPLET DIAMETER (MICRON)
Fig. 3 .4.1 Collection Efficiency of Droplets as
a Function of Droplet Size
(a) DRIFT Calculation
(b) Roffman Calculation
1.0
96
L)
L)-4-
U-U-LL
.I
CD
-C)_J-J
.0110 1000
97
in Fig. 3.4.1 as the curves A and Roffman's results are the
curves B. The solid curves are the results for the double-
layer louver eliminator, Case D1 as identified in Table 3.1.1.
The broken curves are for the sinus-shaped eliminator, Case
N2 as identified in Table 3.1.1 For both geometries there is
a fair agreement between the two calculations over the range
of droplet sizes considered. However, at small droplet sizes,
the results deviate from each other. The reason for this is
that for small droplets, the trajectory depends strongly on
the air velocity distribution. Therefore, for an air
velocity distribution that is mainly parallel to the duct
walls, even near the entrances and turns as assumed by Roff-
man's model, the smaller droplets will follow the air stream
and escape the eliminator. In this way, a smaller collection
efficiency than the real value is predicted.
It is found from these comparisons that the calculated
results are quite sensitive to the assumptions regarding the
air-stream velocity distribution. It is noted that the
numerical simulation model in this work is physically more
realistic than others which are available.
Figure 3.4.2 shows the drift collection efficiencies for
the two-layer louver and sinus-shaped type geometries of the
same length, pitch, and entrance conditions. The comparisons
of their air velocity distributions and trajectory plots were
given in previous sections. The fact that the sinus-shaped
geometry has a higher collection efficiency can be explained
)
%W
DROPLET n, r.-_ V 100ICRON)
P19- 3.4.2Ri *.3 CO1lection Efficienc
Untlon of Droplet roplets a(a ) Snus-Shaped ze(b) DUble-Layer Eliminato
Of SamrD Louver Eliminator(Air e DimensiOns
Inlet Velocty_ l/s)
98
U.
(
LU
(-
C3
Lj
CD
0.0
_
- " t FER (f,
99
by the fluid velocity vector and droplet trajectory plots as
explained in the previous section.
The comparison of collection efficiency results will be
discussed in Chapter 5. Generally, for simple geometries like
the one or two-layer louver eliminators, and sinus-shaped
eliminator, the difference in the calculated collection
efficiency using a no-slip or free-slip velocity distribution
is not significant. This is demonstrated by the results
shown in Table 3 .4.1 for the two-layer louver eliminator at
a 1.5 m/s air velocity. For more complicated geometries,
this difference becomes significant, and will be shown in
Chapter 5.
Also shown in Table 3.4.1 is the capture efficiency
comparison of the DRIFT code calculation with Yao's calcu-
lation for the E-E eliminator. It is seen that with free-
slip conditions in the DRIFT calculation, the results are
very close to Yao's results. This is expected because po-
tential flow is assumed in Yao's calculation. Using a no-
slip condition in the DRIFT calculation, the results are
different, but not significantly. However, the difference
in the pressure drop results is very great, and is demon-
strated in the net section.
3.5 Pressure Drops
The SOLASUR subroutine determines the pressure at each
nodal point. From this information the total mass-averaged
Ln LCn
O c' ')
L\ Lf
C) C
OC)
OC)
C)i
L' in CH zz t- C
C C O H
O LEn nm -3 Lnr- r-H H
Ln O O O Ln00 C'j \Q H HH Cj CO in
J 0 0 0 C)
O O O n 0 0 Ln 0I Cm z t- O Ln H CO 0
H H - H r \l M M U
C\j 0 ,- n t ) t CO O C)H
100
il
OHU)II01
ZI
0)
a)aE4
OdO"-I4-)C)
a)
co
OU)0
04-,JO
Cd
Iw
OCdr-
fo
J04-)
I
0~EH
W.q):O0
U)H*0
0
0
C)OH
U)Ci)LN
E
4.
aH
E O
CHC)a)
q4-
04-)
H
0HrlU)
I0z4
r-iHr r
) .
oa::s0 c
a *H
(1) p\a -ri
L) C\j
U);I Ce0, cOCi)
I--,0
-riHU)Q
Ca)a)
E:1
4oa) 4-H a)
.a E0 4
w
101
pressure drop across the eliminator can be determined if a
no-slip boundary condition is assumed at the eliminator walls.
The results of such calculations will be compared with experi-
mental data in later chapters. In this section the pressure
drop distribution along some of the eliminators will be
presented to get some insight into the effect of the geometry
upon pressure loss across eliminators. The pressure drop
distributions presented here are normalized as the ratio of
the pressure drop at any location in the eliminator to the
total pressure drop across the eliminator.
Figure 3.5.1 shows the pressure drop distribution for
the sinus-shaped eliminator. It can be seen that most of the
pressure loss occurs near the inlet and outlet regions of the
eliminator. This is due to the fact that the slope of the
duct boundaries is steeper at these regions. Similar results
are obtained for the asbestos-cement eliminator (Fig.3.5.2)
which has a shape similar to the sinus-shaped eliminator except
that the slope at the entrance and exit regions is not as
steep while it is steeper at other regions. These regions of steeper
slope extends farther along the eliminator length than in
the sinus-shaped eliminator. It will be shown later that the
total pressure loss across the asbestos-cement eliminator is
larger than that of the sinus-shaped eliminator.
Figure 3.5.3 shows the results for another smooth geometry,
the E-E eliminator designed by Yao and Schrock (Y1,Y2). The
flow channel cross-sectional area decreases along the eliminator
102
O ~ ~ ~ O
rl ~ ~ ~ r
4Q 0) ~~o i
E~~~~~~~~~~0 .-~~~~~~~~~~~~~~~~~~.c,~~~~~~~~~~~~~~~~~~~~~~~~
TJ· EL X.
0 .dP Q,~~~~~~~~~~~
rZ~~~~~~~~~~~.C~~~~~~~~~cd j~~~~~~~~~~~~~.
o 1';o
v '
4-4
m olhOo) 4a) C;_ a
r-nLr\
b.D
OCDm-
'dV/ (1)d-XCDd
dV/((I)d-(X)d)
FI
103
z o4U
.C 0o4 4.
rl
E
0H
0 U)
i3 0 ·
4-)Uo bh¢
-bOLCh vD
C) C)
IdV/((I)d-(X)d)
104
CD
O
0
0
4-'CH
01
r-l~3 ·rf
Ln C
I d((O)d-(X)d)idV/( (q)d- (X)d)
105
length, which causes the pressure loss to increase steadily
along the eliminator length. The pressure loss increases
sharply near the outlet region of the eliminator. This is
probably due to the occurrence of flow separation as discussed
in Sec. 3.2.
Figures 3.5.4 and 3.5.5 show the results of the two
eliminators with sharp corners, the double-layer eliminator
and the Hi-V eliminator. For both of these eliminators, the
pressure distribution curves exhibit a complicated behavior.
For the double-layer louver eliminator in Fig. 3.5.4, the first
discontinuity occurs near the inlet region, and is probably
due to the occurrence of flow separation. The maximum
pressure loss occurs at the turn in the eliminator, where it
decreases sharply in a very short length. It then increases
again to the outlet of the eliminator. Similar results are
obtained for the Hi-V eliminator in Fig. 3.5.5. These
distributions are not physically reasonable. Therefore, the
pressure drop results for geometries with sharp corners are
not reasonable, and should be used with care.
Table 3.5.1 lists the calculated pressure losses across
some common drift eliminators. They are expressed in terms
of the velocity head, which is AP/pV2. It can be seen
from this table that the pressure loss increases as the
eliminator geometry becomes more complex. Since the collection
efficiency also increases in this manner, it is necessary to
compromise between these two parameters in the design of a
106
0a,
bo0
0H,-
0Ca)Xla)4
*-o
Fl4 D
_ r--C CVdV/((1)d-(X)d)1
107
0,~~~S 4
04-)
.rC
4U)Ert
0H4-:O
.C
C
U)hOfr-t5-4,C
X.oo3
LfonC
oC- CD
'dV/( (I)d-(X)d)
CI--
108
Table 3.5.1
Calculated Pressure Loss Across Some CommonDrift Eliminators
EliminatorGeometry
Case No. inTable 3.1.1
Single-layerLouver
Double-LayerLouver
Sinus
Asbestos-Cement
Hi-V
Zig-Zag
E-E
E-E(Yao's result)
S1
AP4 2fpV
2.35
D2 0.97
N2 2.82
Al
H1
Z1
El
3.91
3.43
2.54
14.61
El 1.20
109
drift eliminator with low pressure loss and high collection
efficiency,
The pressure loss across the E-E eliminator calculated
by Yao and Schrock (Y1,Y2) is also found in the table. Its
value is much lower than the one calculated by the DRIFT code.
This is because Yao and Schrock assumed that there was no
flow separation. This was, however, shown not to be true in
Sec. 3.2. If this is true, then this eliminator will be nsuit-.
able for use in cooling towers despite its high capture efficiency.
As mentioned earlier, for eliminators with sharp turns,
the pressure loss calculation will not yield reasonable
results. This is illustrated in Table 3.5.1; for the double-
layer louver, Hi-V, and Zig-Zag eliminators, the results are
unreasonably low. The experimental values of the pressure
loss are higher, and are reported in Chapter 5.
110
CHAPTER 4
EXPERIMENTAL TECHNIQUES
4.1 I' ntroduction
Experimental evaluations of drift eliminator performance
are usually performed by measuring the drift distribution at
the exhaust side of an eliminator which is installed in a
particular cooling tower or in a simulated cooling tower
facility. In most cases only the drift rate (defined as the
drift mass current divided by the recirculating water flowrate
in the tower) was measured. Very limited experimental work
has been done on the droplet size-dependent collection
·efficiencies of drift eliminators.
Recently it has been realized that the droplet size
distribution plays an important part in determining the nature
of any drizzle which may arise from the drift (M3). Investi-
gations have therefore been made into the droplet capture
efficiency of drift eliminators as a function of droplet
size. This data can also validate the calculations performed
by the DRIFT code.
This chapter describes the drift elimination facility,
the drift measurement techniques, the methods for analyzing
the measured data, and the technique for pressure loss measure-
ments. Results obtained from the experiments will be presented
and discussed in Chapter 5.
I
i
111
4.2 Drift Elimination Facility
In order to perform comparative performance studies of
cooling tower drift eliminators experimentally, a Drift
Elimination Facility has been constructed. The facility
simulates a cooling tower fill-outlet environment in which
drift eliminators can be installed for testing. A schematic
diagram of the facility is presented in Fig. 4.2.1.
This facility is a low-speed wind tunnel, 0.8 m by 0.8 m
(2.5' x 2.5') with plexiglass walls, supported by a Dexion
angle skeleton. Chrome felt gaskets are placed between the
plexiglass and Dexion angles to prevent leakage. In order to
have access to various regions, most of the plexiglass plates
can be removed. Air speeds are adjusted by means of a two-
speed exhaust fan at either 1.5 m/s or 2.5 m/s, which simulates
natural-draft or mechanical-draft cooling tower conditions,
respectively. The fan is placed at the inlet of the tunnel
and forces air into the tunnel. To dampen the flow turbulence,
some soda straws and honeycomb sheets have been installed
downstream from the fan. The turning vanes are made of plastic
sheets, and are spaced in such a way that the air flow will
remain uniform after turning out of the horizontal section of
the facility into the vertical section of the facility. The
air is recirculated continuously through a 0.46 m (18") diameter
flexible air duct to insure that water vapor saturation is
maintained. Recirculating droplets are thought to contribute
only an insignificant amount to the drift generated in the
1 I2
F--
ILLt;E
H
Q
CCd
0
C-H
m4.)cri
0r-q
w
Cd
bO
C)4-)
Cd
S0C)H
CMQ
bC.1-f
113
tunnel and are neglected.
Water droplets are injected into the flow by a spray
head which consists of 20 full-cone center-jet nozzles (SPRACO
Model 3B), each delivering about 0.3 gallons per minute of
water (S8). The drift quantity and droplet size spectrum
can be controlled by means of the spray flowrate valve located
above the pump. The valve is a PVC ball valve and the pump
is rated at two horsepower. The droplets produced by the nozzles
lie mainly in the 5 to 200pm diameter range. A Fulflo
water filter with a cellulose acetate honeycomb cartridge
that has a removal rating of 20 m filters out any solid
particles in the circulating water that might plug the spray
nozzles. The facility is slightly tilted so that water will
run down to a drain and be collected in the water tank. This
recirculating water is changed to fresh, clear water before
an experiment is performed.
The vertical test section in which the eliminators are
installed is shown in Fig. 4.2.1. Eliminators could also
be installed horizontally in the horizontal test section of
the facility. However, in the work reported in this thesis
the eliminators have all been installed in the vertical test
section. This is done so that no significant water film can
accumulate on the eliminator walls; also the droplet entrance
conditions are simpler and are consistent with the theoretical
calculations.
Three types of industrial drift eliminators have been
114
received from cooling tower vendors. They are the following:
(A) Belgian-Wave (sinus-shaped) eliminator
(B) Hi-V eliminator
(C) Zig-Zag eliminator
The eliminators are cut into suitable lengths so that
they can be fitted into the test section of the facility.
These eliminators are secured by special eliminator holders,
which can hold eliminators of different lengths. The holders
can adjust the pitch of the eliminators and the angle inclined
to the air flow direction. The holders are made of aluminum to
resist corrosion in the humid environment of the facility.
4.3 Drift Measurement Techniques
There are many methods for measuring the water droplet
size distribution of water entrained in an air flow stream.
In the work reported here, the laser light scattering technique
is used because of its capability to measure very small water
droplets online. This technique is' similar to the PILLS
system developed by the Environmental Systems Corporation
(S2,S4).
Soon after the laser was developed, it was recognized
to be extraordinarily useful for light scattering studies
because of its monochromaticity, high power density, spatial
coherence in the TEMoO mode, temporal coherence, and the small
divergence of a laser beam. The laser used in the present
study is a steady-state, Helium-Neon gas laser (Spectra-
115
Physics Model 125A) which provides 50 milliwatts of single
transverse mode optical power at 632.8 nm. The schematic
diagram of the droplet measuring instrument shown in Fig. 4.3.1
illustrates the general principle of this light scattering
technique for sensing flowing droplets. The laser light
source illuminates a narrow beam in the medium where the water
droplets are flowing. The illuminated water droplets scatter
light in all directions. The scattered light intensity is
related to the parameters of the scattering medium and to
the geometry of the apparatus through the familiar Mie theory
(M5, V1). It has been found that the scattered light intensity
Is, at any angle can be related to the size of the spherical
water droplet by the relation
Is = Kd 2 (4.3.1)
where K is the proportionality constant and d is the droplet
diameter. Fig. 4.3.2 shows a plot of scattered light
intensity versus droplet size calculated by the DAMIE code.
A description of the code is given in Appendix B. It can be
observed from this figure that Eq. 4.3.1 holds. A scattering
volume, designated V in Fig. 4.3.1, is defined by the inter-
section of the laser beam and a collimated photodetector
acceptance cone, The acceptance cone is defined by the two
1000 m diameter apertures in front of the detector. The
photodetector is an RCA Model 7265 photomultiplier tube. This
is a 14-stage, head-on type detector having an S-20 spectral
response. It is placed at an angle of approximately 30° to
116
cn
SL4
cin0a)
-4 cW ; > ¢ ( C 4 ;
bO
-' 0d 4H-
CO0 O
e: Eb :j
'H SI)H
r.,-tQ a)
bC
CL
I
117
e_ _ _ __
105
wL-4
z-4
103
103
101 102 10 3
DROPLET DIAMETER (m)
Fig. 4.3.2 Scattered Light Intensity Versus Droplet
Size Calculated by DAMIE
118
the laser beam, which was found to be optimum. When a droplet
passes through the scattering volume, light is scattered and
detected by the photomultiplier tubes,producing a voltage
pulse. The height of the pulse represents the scattered intens-
ity. The pulse is amplified and recorded in a multichannel
analyzer. The multichannel analyzer determines the pulse
height and records each pulse in an appropriate channel. The
analyzer used in this work is an NS 900 pulse height analyzer
with 1024 channels. It can record pulse heights of zero to
eight volts. In the current work, only 256 channels are used.
It has been demonstrated,by using a standard pulse generator,
that this analyzer is capable of analyzing the typical pulse
shapes encountered in the experiments. The spectrum being
recorded represents the droplet size spectrum of the droplets
passing through the scattering volume.
This measuring technique was developed from the PILLS
system (S2). Limitations of the PILLS system due to fog-
induced background signals havebeen reported in field measurements
of drift in cooling towers, but are nonexistent in this lab-
oratory work since cold water is used in the experiments.
Multi-particle scattering can be avoided by controlling the
size of the scattering volume and by adjusting the drift density
appropriately. A single droplet is never counted more than
once because a steady-state laser is used rather than a pulsed
laser as in the PILLS system, where a slow moving particle may
remain in the scattering volume during more than one laser pulse.
119
Other background signals are derived from the high-voltage
power supply, photomultiplier tube, amplifier, and from any
surrounding light sources. These contribute very little to
the measured signal and are neglected.
4.4 Calibration of the Drift Measurement Instrumentation
A calibration is necessary in order to find the voltage
response of the drift measurement instrumentation versus the
droplet size. The instrument is calibrated by introducing
monodisperse water droplets of certain sizes into the scatter-
ing volume and noting the output signals. Monodisperse means
that the droplets all have the same size. Monodisperse drop-
lets are generated by a Berglund-Liu Monodisperse Aerosol
Generator, on loan from Thermo Systems, Inc. This generator
produces water droplets of a certain uniform size by utilizing
a vibrating orifice. Its operation is based on the instability
and uniform breakup of a cylindrical water jet under mechanical
disturbances (B1). When these mechanical disturbances are
generated at a constant frequency and with sufficient amplitude
in a liquid jet of constant velocity, the jet will break up into
equally sized droplets. To form a source of monodisperse drop-
lets, these uniform droplets must be dispersed and diluted
before they recombine. The generator is unique in that it can
produce droplets of a known size, the droplet size being calcu-
lable from the generator operating conditions to an accuracy
of 2%. The droplets generated are exceedingly uniform in size-
the standard deviation is approximately 1% of the mean droplet
diameter (B1).
120
A schematic diagram of this generator is shown in
Fig. 4.4.1 . It consists of four major parts :the liquid
feed system, the droplet generator, the droplet dispersion
system, and the wave generator. The liquid feed system is a
syringe pump which forces water through a membrane filter at
a constant rate into the droplet generator. The rate is
determined by noting the time (using a stop watch) in which
a known volume of liquid is forced into the droplet generator.
The droplet generator pictured in Fig. 4.4.2 consists of a
stainless steel cup with a 1.15 in. diameter flange and a
hole in the bottom. A 0.375 in. O.D. orifice disc is placed
in a groove inside the bottom of the cup, a Teflon O-ring is
placed on top of the orifice disc, and a stainless steel cap
is tightened onto the O-ring holding the orifice disc in
place. A ring-shaped piezoelectric ceramic with two silvered
faces is epoxied to the flange on the cup with conductive
epoxy. The liquid from the liquid feed system is fed through
the cap into the cup and is then sprayed through the orifice.
An A.C. voltage from the wave generator is applied to the
piezoelectric ceramic which vibrates the cup and disturbs the
liquid et at a constant adjustable frequency. Because the
syringe pump delivers the liquid at a constant rate, the liquid
Jet breaks up into uniform droplets at the frequency of the A.C.
voltage. The uniform droplet stream then enters the dispersion
system. The droplet dispersion system consists of a stainless
steel holder and cover for the droplet generator, a pressure
13~ ZO O0 0
onw ww~~~zZ a-
< CD _) ,,~:~~ <.
0J <c: Z .(, Z.-.~ L.
Z.cr
w: w
W .
121
w
I(n Zr ,--- _
CLIJ
w
a
Q,
O
0
a)
,-IoObO
-Ha0C)o
:
a)C)
-0OU
IL
-J wC) -,
< LL.
0O4-,
Cd
r-O0
S~4
-iFr4
CnLn
a.00-U <
122
DISPERSEDDROPLETS
DISPERSIONCOVER .. .. DISPERSION
Fig. 4.4.2 Schematic Diagram of the Droplet Generating
System
.
123
regulator, a flow meter, and an absolute filter. The cover
has a dispersion orifice through which both the droplet stream
and a turbulent air Jet pass. When the droplet stream mixes
with this air Jet, it is dispersed into a conical shape. The
dispersed droplets are then ready to enter the wind tunnel flow
stream. The droplet generator is placed under the laser light
beam at an appropriate distance so that the droplets are well
enough dispersed to avoid multiple scattering, yet it is close
enough to avoid significant evaporation of the droplets. The
droplets are carried up to the scattering volume by the
normal air flow in the Drift Elimination Facility described in
Sec. 4.2. The scattered light from these droplets is detected
and recorded by the drift measuring instrument in the manner
described in Sec. 4.3.
The water droplet size produced by the vibrating orifice
monodisperse aerosol generator is deduced from a knowledge of
the orifice size, the liquid feed system flowrate, and the wave
generator frequency by
6Q 1/3d (-) , (4.4.1)
f
where Dd is the droplet diameter, Q is the liquid feed flowrate
and f is the disturbance frequency. Table 4.4.1 tabulates some
droplet sizes under typical operating parameters. The droplet
sizes generated in this work range from 50 m to 100 m
diameter. More detailed information on this generator is
contained in Refs. B and T1.
Pc a,a EH COcdh w$1 H
'PC,04
_o N
zh ~ m
04 ~ ~ a0C) p CE-q za $ a)O: a)
E-4 0 u
0 $k . PC.,
a)
0 C
* 4o C
r 0 aIT
H c::C
H h C)ld a) E C, ~ zo~~q z ~ 0C, C, r-D t ·tH1 : aE O' 4-)E.i 4 wC) dcd
C0 -r in no t N Hf-i
H 0 0 0\0 1 Lf
rH m
3\ o
H- H o0
o o o o
'P040fa0kq
a,
0d EC4,
EcO H:4 C
0 O O LnULf OC r-H
124
125
4.5 Data Acquisition and Analysis Techniques
Because the laser light intensity is not uniform across
the beam cross section, monodisperse water droplets passing
through the beam will emit different scattered intensities.
The laser light has the Gaussian intensity distribution along
its diameter shown in Fig. 4.5.1, where the laser light
intensity was measured across the beam cross section using a
Spectra-Physics Model 404 laser power meter. Because of this,
the pulse heights recorded in the multichannel analyzer will
not yield a single sharp peak as would be expected from a beam
of uniform intensity if the edge effect is negligible (which is
true when the beam size is much greater than the droplet size).
Figure 4.5.2 shows a typical measured monodisperse
droplet pulse height distribution using 80 m diameter drop-
lets. The location of the peak in the distribution is unique
to this droplet size. This measurement is repeated for several
monodisperse droplet sizes. Fig. 4.5.3 plots the heights of
the peaks versus the droplet sizes. The slope of the curve
in this logarithm plot is very close to two, which verifies
the correlation of Eq. 4.3.1.
Since the measured relationship between an output voltage
pulse and an input droplet size is not unique, a complicated
matrix operation must be used to analyze the measured voltage
distribution. To transform the measured voltage distribution
data into a droplet size distribution the following transform-
ation procedure is used.
126
-2,0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2r0
DISTANCE (MM)
Fig. 4,5.1 Intensity Distribution Across the Laser Beam
Cross Section
20
15
>-10H-
LU
5
0
127
I)4-
toC:
a
'0LL 0
4n
,HZ4 03 CD
a EI .
o to co
o
0
X,--a
~1 o
J4
009 09h 09C OhZl OZ 0
SINOO3 JO 39GWlN
1.2
1,0
0.8
0.6
0.4
z
0.2
0.0
-0,2
-n 4.0 4.2 4.4 4.6 4.0 5.0
LN(D)
Fig. 4.5.3 Calibration Curve - the Peak Voltage of PulseHeight Distribution Versus Droplet Size
128
129
If the matrix R (v;d) represents the voltage response
function for droplets of diameter d, and if vector A(d) is
the actual size spectrum of the drift, then the measured
voltage distribution vector, M(v), is given as
M(v) = R (v;d) A(d) (4.5.1)
So the actual size spectrum can be determined as
A(d) = R-l(v;d) M(v), (4.5.2)
where R and M are found by calibration and field measurements,
respectively
For 256 channels, the R matrix is a 256 by 256 square
matrix, This matrix is determined from calibration measurement
with the relationship in Eq. 4.3.1. The matrix inversion for Eq.
.4.5.2 is done by the LEQT1F subroutine of the IMSL Library
(I2). A description of the subroutine is given in Appendix A.
To determine the collection efficiency of the eliminators
as a function of droplet size, the droplet size spectra should
be measured at the inlet and outlet of the eliminators. In
the present experiment, the scattering volume is fixed in
space while the eliminators are either placed below the volume
or above it. By placing the eliminators above the scattering
volume the spectrum measured represents the inlet droplet
spectrum, Pin(d), and similarly the spectrum measured when the
eliminators are below the scattering volume is the outlet
spectrum, Pout(d). From these two spectra, the collection
efficiency as a function of droplet size, n(d), can be
130
calculated by
P (d)n(d) = 1 (4.5.3)
Pin(d)
The measured spectra are recorded on paper tape and analyzed
by a computer program called DATANA which performs the response
matrix multiplication and calculates the collection efficiency
if a pair of measured spectra are provided. A description of
this code can be found in Appendix A.
A sensitivity analysis of the effect of the response
function R(v;d) on the collection efficiency calculation was
performed. This was done by using a typical set of measured
voltage distributions at the inlet and outlet of the eliminator,
and the calibration curve. By changing the parameters in the
calibration curve, the sensitivity of the collection efficiency
results are recorded. The parameters include the size of the
monodisperse droplets used for calibration, DC, and NC1, NC2,
and CC1, which are three parameters that specify the cali-
bration curve (as explained in Appendix A). Table 4.5.1 lists
the maximum changes in the calculated collection efficiency
from its mean value as the parameters are individually changed
by 10% from their mean values. It is observed that the only
parameter that yields a significant change in the collection
efficiency is DC, and this change is only significant at small
droplet sizes. It is concluded that the collection efficiency
results are not very sensitive to the response function, and
131
Table 4.5.1
Sensitivity Analysis of the Collection
Efficiency Results
Parameter whosevalue is changedby 10%
Maximum Changein Collection Efficiency
17% at 40 pm
8% at 50 m
4% at 60 m
8% at 40 m
3% at 50 m
No significant change
No significant change
DC
NC2
NC1
CC1
132
the error introduced by the calibration will not amplify the
error in the collection efficiency results.
4.6 Pressure Loss and Air Speed Measurement Techniques
The pressure loss across the eliminators is measured with
a differential electronic manometer (D4). Two static pressure
pitot tubes monitor the static pressure at the inlet and
outlet of the eliminators. The tubes are made of 1/16" ID,
18" long, stainless steel. One pitot tube is located 6"
downstream of the eliminators and the other is located 10"
upstream. They are 21" apart overall. The pressures are
measured at these two points, and averaged over a certain time
period. The differential pressure is first set to zero with
the air flowing, but without the eliminators. In this way
the pressure loss due to other structures will not be included
in the pressure loss measurement of the eliminators.
The sensor is a Barocell differential pressure trans-
ducer (Datametrics Model 570D). The pressure range that can
be measured by this sensor is from zero to 10 torr(i.e., zero
to 0.1934 psi). The pressure-sensing element in this unit is
a high-precision stable capacitive potentiometer; its
variable element is a thin, highly prestressed metal diaphragm
positioned between two gas-tight enclosures which are connected
to the external pressure ports. A difference in pressure
between the two enclosures produces a deflection of the dia-
phragm which varies the capacitance of the diaphragm and the
'133
fixed capacitor plates. The Barocell is wired into a 10 KHz
carrier-excited bridge so that the variable capacitance un-
balances the bridge and produces a 10 KHz signal whose ampli-
tude is proportional to the applied pressure. This A.C. volt-
age is measured by a high-precision electronic manometer
(Datametrics Model 1173), which gives a zero to 1.0 volt D.C.
output. This D.C. voltage is read by a digital voltmeter.
The accuracy of the system is about 0.5% of the reading.
-6The sensitivity of the system is 3 x 10 torr.
The pressure difference can be accurately measured, but
there are other errors in the interpretation due to pressure
fluctuations in the flow turbulence and pressure loss contri-
butions by the eliminator holder structure, which have been
found to be significant. The measured results will be
presented in Chapter 5.
The air velocity is measured with a hot wire anemometer
(Datametrics Series 800-VTP Flowmeter- (D5)). It measures the
average and instantaneous velocities in the flow of air by
considering the cooling effect of the stream on a very thin
electrically-heated wire filament. The probe that holds the
flow-sensing wire filaments is a 3/8" diameter stainless
steel wand. It is inserted into the drift measurement facility
at any point to measure the local, time-averaged air velocity.
The velocity range metered by this system is from zero to
6000 ft/min (0-30 m ). It has a two volt D.C. output connection
for a digital voltmeter. The voltage reading can be converted
.,, I
134
to a velocity with a calibration curve. The accuracy of this
unit is about 2% of the reading. However, due to flow turbulence
and a non-uniform flow distribution, an error of about 10% is
introduced.
The results of these measurements will be presented and
compared with calculations in Chapter 5.
4.7 Sources of Experimental Error
It is difficult to analyze the contributions of the
experimental uncertainties in the final drift measurement
results. In this work, the approach is to repeat the measure-
ments several times, so that the experimental accuracy is
indicated by the variations in the final results. These
measurements include the calibration measurement and the
measurements of voltage distributions at the inlet and outlet
of the eliminators. The results of this test are presented
in Chapter 5. In this section, all possible sources of
experimental error are identified.
(1) The position of the laser beam moves in the first
few hours after it is turned on. Therefore,
a warm up period of at least one hour should be
allowed before any measurement. Movement of
other drift measuring components is not significant
throughout the experiment, which takes about
ten to fifteen hours.
135
(2) Calibration is the most difficult measurement
in the experiment. The droplet generator
orifice is frequently plugged, and sometimes
may be partially plugged, which decreases the
pumping flow rate and thus changes the droplet
size and the uniformity of generated droplets.
Therefore, the flow rate should be constantly
checked throughout a measurement.
One important uncertainty in the calibration
is that the chance of more than one droplet
appearing in the scattering volume is signifi-
cant, although efforts have been made to reduce
this. However, the final results are found not
to be very sensitive to the calibration curve
except at small droplet sizes, as discussed before.
Since the monodisperse droplets are carried by
the flow up to the scattering volume a few inches
away from the droplet generator, the droplet
size will decrease due to evaporation. It was
estimated that this decrease can be as large as
5% for 100 m droplets. This error could intro-
duce an error of 10% in the final results for
small droplet sizes, as was shown previously.
(3) Statistical counting errors are difficult to
estimate because they are not uniform for all
136
droplet sizes, but become larger for bigger
droplets. In the data analysis, the distribu-
tion is smoothed, and this will eliminate some
of this error. Dead time in the pulse height
analyzer contributes another uncertainty to the
results. Since the drift rate is much higher
at the eliminator inlet than at the outlet, the
dead time is very different for the two distri-
butions, and it is found that the analyzer used
in this work cannot account for this very well.
(4) There is a possibility that more than one drop-
let may appear in the scattering volume and cause
the analyzer to record a wrong signal. This
possibility cannot be avoided but can be reduced
by decreasing the size of scattering volume and
by reducing the quantity of drift. This error is
thought to be small.
(5) The quantity of drift fluctuates with time due
to the pumping power fluctuations and changing
conditions within the Drift Elimination Facility.
However, if the data acquisition time is long
(three to five hours in the present experiment),
this fluctuation will average out in the measured
results.
137
The pressure drop across the filter in the
circulating water system increases with time
because of the gradual plugging of the filter
by foreign particles in the circulating water.
This affects the quantity of the drift generated
by the facility. But the effect is small during
one test.
(6) Signal noises that may contribute to the measure-
ment uncertainty consists of electronic noise
in the amplifier and high voltage supply, the
dark current of the photomultiplier tube, and
scattered light from other sources. It is found
that these noises contribute about 10% to the
lowest recordable signal and become insignificant
for larger signals.
(7) The laser light intensity at the measuring
point depends on the drift concentration which
shadows some of the laser light. However, by
measuring the intensity at the scattering volume
with a power meter for different drift concen-
trations, this effect was found to be almost
undetectable.
138
The above discussion gives possible sources of
experimental uncertainties. In Chapter 6, a quantitative
analysis of experimental errors is developed.
139
CHAPTER 5
COMPARISON OF EXPERIMENTAL RESULTS
WITH THEORETICAL CALCULATIONS
5.1 Introduction
In this chapter the experimental results for the drift
collection efficiency and pressure loss across some industrial
cooling tower drift eliminators are presented. These results
are compared with the calculations from the DRIFT code. The
experimental measurement techniques are discussed in Chapter 4,
and the numerical simulation techniques in the DRIFT code are
described in Chapter 2.
Three industrial drift eliminators were donated for this
study by cooling tower vendors. These are the Belgian-wave elim-
inator, the Hi-V eliminator, and the Zig-Zag eliminator. Their
geometries and dimensions are shown in Fig. 5.1.1. The Belgian-
wave eliminator is made of sinusoidally shaped asbestos cement
board, and it has a uniform flow channel cross section. The Hi-
V eliminator is made of polyvinyl chloride. Its flow channel
cross section is not uniform, as shown in Fig. 5.1.1. The
Zig~Zag eliminator is made of fiberglass. Its flow channel
cross section is uniform, except at the corners.
The measured pressure losses across these eliminators are
compared with calculated pressure losses in Section 5.2, and the
collection efficiency comparisons are presented in Section
5.3, The data is developed at two air speeds, 1.5 m/s and
2.5 m/s.
140
I I
LI ;
-JLi X--,0
U CO
LU LUL
E CoO ~~~~~o-
UUE ) t
141
5.2 Pressure Drop Across Eliminators
Table 5.2.1 is a comparison of the theoretically
calculated pressure losses across the three types of drift
eliminators with the experimentally measured values at an air
speed of 1.5 m/s. Table 5.2.2 presents the same comparison
at an air speed of 2.5 m/s. The true air speeds are slightly
different for each eliminator, since their flow resistances
are different.
The calculated and measured results are in good agreement,
and are within the bounds of experimental error. It is clear
from the data that as the geometry of the drift eliminator
becomes more complex, the pressure loss increases. Since the
drift collection efficiency also increases with increasing
.geometrical complexity (shown in Section 5.3), it is necessary
to strike a compromise in designing an eliminator that will
achieve an acceptable pressure loss and an acceptable collec-
tion efficiency. The values in brackets are the resistances
of the eliminators to the air flow expressed in terms of
velocity heads corresponding to the nominal air speed. This
is an advantageous way to report the data since its numerical
value for a particular eliminator is independent of the air
speed and the working fluid if the flow is fully turbulent.
The theoretical pressure loss calculations for the Zig-
Zag eliminator have been unsuccessful because of the complex
eliminator geometry and because of the limitations of the
calculational method in describing turbulence (see Chapter 3).
O_
-J
C)
'$4r)
O
Co
4-)
-H
10
· a. a)
D 04
H a* Cdo a)
a)
O Cd
Ca0- a)4-) r
*
4- 4o1o z
Cd a)
¢.: x:
142
cOH
cn oMo ril0 mO
00 *
0 CJC\i co
V..c'J
Irk
M COH
co
O m*
O ·
0'-
0 ^O o. -
4-) a)0 bo
00o d C)
1-11'dO
a) 4
cd n ada) I: Cd
4-)0
COd
3
'0a)a) -a cU) \
h.r-
'0aD(a04
CO
4-
OCd0H-- 4)
* Cd
a) H
C)
::::
a
Coto
CD
c
Lf
4
a)
Cd
r-lbO
a)
04-:Cd
*1-I
04 -Hr-
E- W
boCd
IhbO.-I
I-H
HWk
r-iH
Lc m H 00CU C r H O0
. O0 OCQ 0Cj
0 0
\C t- Cl-0· 0
*C'J *CJ
CJ CM
U)
Cd:
Cd-HboQ)a)CQ
boCd
~~> I~
-Hi -H:1: t)L~::: L'q
OCa) c U
C: XE
143
a) HtCd
0a)U)r:
UOCd
r)MboCd
C[- 0
CQ0U)$Ma,-::
0c
O0
a)
r
C>
'0
(1)U)4-,rdoH
r..)HCdi0
t- Oo L--
O *0
0 01rv
C Y)
Clu
Cd
aU
C)o
0a,)
0
4-)
10
CUa)
0 :
O
O oUa4)
H
* *
II
bO40:>H
0*-Cd
O0C
a E
E W
144
However, the agreement of the calculated and measured results
for the other two, more simple geometries, demonstrates that
the calculations are valuable in reasonably smooth geometries
for predicting pressure loss.
Comparing Table 5.2.1 and Table 5.2.2, it is found that
the pressure loss increases approximately as the square of the
air speed for all three eliminators. The observations made
about Table 5.2.1 also apply to Table 5.2.2.
The pressure losses for these eliminators quoted by their
vendors are higher than the values presented here (H6, S9).
The reason is that in their measurements, the pressure differ-
ences with no eliminators installed were not set to zero.
This means that their measured data include the pressure loss
due to other structures, which are considerable in comparison
with just the pressure loss across the eliminators.
5.3 Collection Efficiency Results
In this section, the calculated and measured collection
efficiency results are presented. Fig. 5.3.1 shows the drop-
let collection efficiency as a function of droplet size for
the Belgian wave eliminator at an air speed of 1.5 m/s. The
measured data are presented as a broken line in the figure.
The experimental technique is described in Chapter 4. The
calculated results using both no-slip and free-slip boundary
conditions are compared with the experimental results. It
is found that the calculated and measured droplet capture
145
CC
04-
C
o ECC
)
0 0 CC) C d
O rH
,-. a)
> >
r-
Cd
Ln cCd
00)
C,)
OCDb< X~~~~~*-
A9N3131JJ3 NOI1331103
146
efficiences agree well for this smooth eliminator geometry,
with the no-slip boundary results providing more accurate
predictions. It is also found that droplets smaller than 60
um in diameter can easily escape from the eliminator.
The calculated and measured collection efficiency data
for the Hi-V eliminator are shown in Fig. 5.3.2. It is found
that this eliminator is more efficient in capturing droplets
of any size than the Belgian wave eliminator. Also, the agree-
ment between the no-slip prediction and the measurement is
reasonably better than the agreement between the free-slip pre-
diction and the measurement. The no-slip result predicts a high-
er collection efficiency than the measured data, while the free-
slip prediction is generally lower than the measured data.
Figure 5.3.3 displays the data for the Zig-Zag eliminator.
The calculation of the air velocity distribution for this elim-
inator failed to achieve asymtoptic values using no-slip
boundary conditions, as discussed in Chapter 3. By using the
calculated distribution of the nonasymtoptic solution, approx-
imate results of droplet collection efficiency for this elim-
inator were obtained. This is compared with the free-slip
prediction and the measured data in Fig. 5.3.3.. This approx-
imate result predicts a higher collection efficiency than the
free,slip prediction, and agrees with the measured data better
than the free-slip prediction in general.
Figure 5.3.4 shows the data at high fan speed,
where the calculated results are no-slip prediction. Similar
conclusions can be made from this figure.
147
CDr-I
0
C
C)
CD H DCoC) CD
o vo0 42 ®uo 10 3
a) N-L.LI 40
LLD 4 E
cd
OrH
U) Q~C) -.CD m O
Lf
U-N CD bO
1CD3132 j tJ113110ArtJIIJIJJ3 N011931109
A3tJ3l3lJ3 N01I133~03
148
C'OCO
0oU
CD:C
00
0o
C)O O
ctdb
C)o .r 1tq
Z C,u F c
C)I,.-%
UF CD bo
gADO313IJ3 M 0 I1331103
149
ell-20
r-Il.-
CDC)
09
.- 00 >q~ ac) h1 L
0= C EO
- C)
O
0O ' Cq U
< a cCD
-I O 4H- C)Q aFc
U' CD bO
AJN3I 3 NO133103
-D
150
5.4 Estimation of Experimental Error
A qualitative discussion of the sources of experimental
errors is given in Chapter 4. It was mentioned that the quanti-
tative contributions of those uncertainties in the data are
not easily accounted for. In this work only the repeatability
of the measured data is established. This is done for each
eliminator by repeating the measurements several times under
similar conditions.
Table 5,4.1 shows the results of four measurements on the
Zig-Zag eliminator at an air speed of 1.5 m/s . The measure-
ments included the calibration run and the droplet size
distribution measurements at the inlet and the outlet of the
eliminator. The data was taken on different days. From
these results it can be seen that the data is repeatable with
a maximum standard deviation of about 10% at the smallest
droplet size.
The results for the Zig-Zag eliminator are the most
consistent among the three eliminators tested in this work.
The reason is that this eliminator is constructed in one block
so that the pitch and the inclination of the louvres always
remains the same even though it was taken out after each
measurement, The other two eliminators are furnished in
pieces which are installed by using the eliminator holders
described in Chapter 4. When taking these eliminators in and
out of the test section, it is difficult to repeat the exact
pitch and angle of inclination. Therefore, for the Belgian
LCn LC 0'D - H
o 0 0g~ g
o o oo o o
gl g
CNj c'j m _i .
cvi ~10 0- 0N 0N ON 0E o o o o o o
o.-0
Lr\ ( o CY)0o oh L \0t- oo oa\
o o o o
YN o oo' 0 0o o o
o0
n m
o o
o O O 0o o 0
o
r-t
r-4H0l
Lf 0c 0 0 . H \OH r-H O L lO0'o t- co Oa a C
o o o o 0 o
o o 0 0 0 0(M' t Lr' 0 t-o cO
151
o
;d -HCd >
c C
0
C
-Iw
)1I
a)
0H
.0))
.0 rl
0.4-)
0
:Sr
0
Cd
0)
.0()a
H
E
Lf
r
0
-0
0.5r-r-0
0a)HH
a)
CaQ)2:
iBCi
Hr.:
:3
4-'00) -,
C Q
152
wave and the Hi-V eliminators the results are not very
repeatable; the maximum standard deviation is as high as 20%.
The discrepancy between the calculated and measured
results is discussed further in Chapter 6.
153
CHAPTER 6
CONCLUSIONS AND RECOMMENDATIONS
6,1 Discussion of Results
It was demonstrated in previous chapters that predictions
of collection efficiency and pressure drop for some drift
eliminators by the DRIFT code agreed fairly well with measured
results. The discrepancies are due to the experimental un-
certainties and the calculational and experimental assumptions.
The results are discussed in this section.
It was indicated earlier that both the calculated and
measured pressure drop results of this work are significantly
lower than the values quoted by cooling tower vendors. This
is due to a difference in the definition of pressure drop.
The vendors include in their definition the pressure loss due
to structures other than the eliminators themselves, and this
contributes a significant amount to the measured pressure loss.
The collection efficiency results reported in this study
are higher than expected because these results generally
show 100% efficiency for droplets larger than 100 m, yet
droplets much larger than this have been found to escape
actual cooling towers. This is probably due to the assumpt-
ions made in the calculations and the simplified conditions
in the experiments. Droplet growth effects,water film
effects, and flow turbulence are thought to be the main
factors. These effects are discussed here.
154
(1) Water film effects
The droplet collection efficiency calculations have
ignored the possible effects of the water film on the elimi-
nator walls. Two effects arise from the drag of the exhaust
flow on the water film, and another effect arises from the
impacting of captured water droplets on the film. The
presence of the water film modifies the exhaust flow boundary
conditions from simple no-slip conditions to those of matching
the air-water velocities at the gas-liquid interface. Except
in the case of a thick film, the effect of this complicated
boundary condition on the velocity distribution within the
eliminator should be small. Also, drag on the liquid film
can lead to droplet generation because water can be drawn off
of the trailing edge of the eliminator, or droplets can be
stripped from the film surface through the formation of Helm-
holtz instability waves.
The work of Yao and Schrock (Y1) indicates that at an air
speed of 2.0 m/s in a smooth drift eliminator geometry the
minimum film thickness required for droplet stripping is 0.2
mm, and that substantially higher air speeds are required for
droplet generation via pickup from the peaks of Helmholtz
instability waves for a film of this thickness. In the present
work, the water film thickness on the eliminator walls was
not measured, however, visual observations were unable to
detect either droplet stripping at the trailing edge of the
eliminator or droplet generation on the interior walls. The
155
absence of droplet stripping at the outlet is implicit
evidence for the absence of wave generated droplets in the
interior (Y1).
With sufficiently thick films and with droplets impacting
at sufficiently acute angles and high velocities, it has been
observed that a droplet can rebound from the film, or "bounce"
back into the exhaust flow (J2). This droplet bouncing
problem is thought to be significant for drift eliminator
performance in actual cooling towers where the water loading
on the eliminator is high and the water film is thick.
Foster et al. (F3) studied this problem and observed that this
effect is indeed significant, however, they found that many
droplets meeting the impact conditions established by Jayar-
atne and Mason (J2) did not bounce. It was concluded that
it is not possible to estimate the true importance of droplet
bouncing in a real tower environment due to the lack of
information about the surface water coverage. Further work
on this effect is being carried on by Foster in the Central
Electricity Research Laboratories (F5).
(2) Droplet growth effect
It was mentioned that droplets as large as several
hundred microns could escape commercial cooling towers out-
fitted with the drift eliminators being tested in this work.
According to the collection efficiency results for these
eliminators, these droplets should have been trapped. It is
156
suggested that the droplets might bypass the eliminators
through the openings around the tower structure. It might
also be due to the stripping and rebounding effects in the
water film as discussed earlier. Another possibility that
has so far been overlooked in the literature is droplet
growth in cooling towers. It has been suggested that droplet
growth is not significant in a cooling tower because a small
drift droplet, moving at a velocity approaching that of the
air stream, might leave the top of the tallest natural draft
tower in less than one minute from the time it passes through
the eliminators. If that is true, then even though the air
in the surrounding air stream is generally saturated, the time
span is too short for any significant droplet growth. However,
this is not true for some droplets in cooling towers. Large
droplets (>100 pm) do not travel at the air stream speed
because their terminal velocities approach that of the air
stream speed as shown in Fig. 6.1.1. Therefore they will
stay in the tower for a long time. Another factor that causes
some water droplets to reside in a cooling tower for a long
time is the air flow pattern inside the cooling tower. The
turbulent effect, the quiescent region, and the variation of
air velocity at the throat of the shell (in the case of a
hyperbolic cooling tower) can prolong the residence time of
water droplets inside the tower, thus making the growth effect
significant. This effect will not only change the size dis-
tribution of the escaping drift, it will also change the
10 100 1000
DROPLET SIZEj MICRON
Fig. 6.1.1 Terminal Velocities of Water Droplets
157
1.0
-i
F-
o 0.10-Jw
LL
0.01
158
drift chemical concentrations.
(3) Turbulence effects
In the theoretical calculation of the performance of a
drift eliminator, it is assumed that the air flow is laminar,
and that the turbulent wake region is neglected. However,
for some eliminator geometries, this wake region extends
over a large portion of the eliminator cross section. In this
work, flow visualization is performed for three eliminators:
the Belgian wave eliminator, the Hi-V eliminator, and the Zig-
Zag eliminator. The flow is established in a long, two foot
wide free surface flume. The working fluid is water, instead
of air. The water depth is about three inches and the water
flow velocity was adjusted to be about 0.1 m/s so that the
Reynolds number matches that of a flow of air whose velocity
is 1.5 m/s. Blue dye was injected at the eliminator entrance
and at the middle of the water deptjh to get away from the
free-surface and boundary layer regions, each of which is
about half an inch thick. The dye was injected at the elim-
inator boundaries so that wake regions could be observed.
Photographs were taken of these dye traces for the three
eliminators. To observe the general flow pattern within these
eliminators, tiny paper chips were sprinkled on to the water
surface and time-exposure pictures recorded the trajectories
of these chips. The results of these simple experiments were
159
quite satisfactory.
Figure 6.1.2 displays the wake regions in the Belgian
wave eliminator by injecting dye into the flow. The water
enters at the right side of the picture where the dye inject-
ion tubes are shown. It can be seen that a significant
wake region exists at the lower boundary and another wake
region exists in the second half of the upper boundary. Fig.
6.1.3 shows the flow pattern by using paper chips. The water
enters at the right side of the picture where the dye inject-
ion tubes are shown. This picture also shows the wake region
at the central portion of the lower boundary where recircu-
lation occurs. The calculated velocity distribution using
no-slip boundary conditions shown in Fig. 3.2.5b simulates
these regions with almost stagnant regions. These regions
are poorly represented if free-slip boundary conditions are
used (see Fig. 3.2.5a).
Similar observations can be made for the Hi-V elimina-
tor. Figs. 6.1.4 and 6.1.5 show the wake regions and flow
pattern respectively. Again,the wake regions exist at the
lower boundary and the second half of the upper boundary.
These wake regions are larger than those in the Belgian wave
eliminator. In the calculation these regions are represented
by stagnant regions if no-slip boundary conditions are used,
as shown in Fig. 3.2.6b. Again, the free-slip result cannot
account for these regions very well.
Figures 6.1.6 and 6.1.7 show the same set of pictures
I 1J~
INi \ I:F ~*ol\.o ,?
A, ·. ,
N<
7Y
/i
t
Us
Isis
W
I./At
Of
t. t
4 t
A/A{
#x
...B
A
·t j
,~ ·' ." -; 1
·. ·
f
.4·b i i-a1%' �
C. 'f
·a4c.
�
i,
PI-1,40";-z , 14
., -,. "I ,'
'., "I . _ -, f-;·r. · id.z9dE 4-. " .
~~~~ ~~~~~~a s~~~~ I _ -
-issF V~"?,ai s
·S + - ^ :
: F
.,
I 1
I.I.
: !I
.,
SdA
A ..
t5
i
t. .'
'
'OrAX
' ,
.g .
.
. ,,. I'
* Sg
e <
I'C .
·. gr.,I I,
11 1 - - '
-A "'I.-I
P, i
a .I , ! j
--,
_A,-1-"
II
I'4
1
,V·
{;
I
* J
s i. , t.,. ,. .V 1 *
I;,1 j, I F
t�
9. ;·.;··i:
'e·Zt
·'` E i·' r
t` ;-.·
' I,
:1I l I
- As
S.
x ie
.
. \S,
:X t:I e : ~ I
'
I>
Vi
1'
I ,
Vt"~~~~~,aM. ~ ), 8 P,
:'.
It
I
p~~~~~~~~r
'4.1
I~
'I~~,
241.7/ -4$;
:14;i
' ~ ~ .- '/. ... ., . . . - ..... ~.~ ~--,:.; .... -.~~ - ' f g _ wslt~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ie
,
r
t-
fly~~* : W.' ~4~'¾; '" ri:
· · ~ ~ ~ ~ ~ ~ -
Is I
I
II
.I'-.1Ast
*r -I 'V'j, I,,*4
I;I
r.
b-
!
t' '
.Fwt r
i
4 *
I
I .
g,
I.,,6
4,, 4
NI,
I 1N, Nt9'*
N
N
'
.k
'so
1
. . .A -.-; .. I. .. . .l ii
4
*1i
I~ ~~;...~ -.~. -.. ·~~a r ~W~.~&~L.Lt~a2A&..LX&4h A.. A..AAhyr·arra ·-E�rlE�iPm;auU·sur�-�EIIJ·�-iYIYPLi·IP �*r��U�L--�-I-- -·L-L�f�lY- ��i�l*i�·-L. · ..
I:
i I
nu; a 'it~rL S is-
166
for the Zig-Zag eliminator. It is seen that the wake regions
are very large for this eliminator. In fact, the third layer
of this eliminator is so turbulent that the dye is well-mixed.
Since the flow is so turbulent in this layer, the pressure
drop in this layer will be greater than the pressure drop in
either of the other two layers. As discussed in Chapter 3
the third layer has little effect on the capture efficiency
of this eliminator, and by taking off this layer great
savings can be realized in cooling tower operation without
producing additional drift.
The calculated velocity distributions using either no-
slip or free-slip boundary conditions, as shown in Figs.
3.2.7 and 3.2.8, cannot account for these turbulent wake
regions. However, the no-slip prediction gives a better
approximation than the free-slip prediction, and the no-slip
results predict the flow pattern quite well in the first two
layers where most droplets are captured. Therefore, the
collection efficiency calculated using these results should
be close to the actual result.
For this eliminator it can also be seen that at the
entrance of the upper boundary there is flow diversion. This
is caused by a pressure difference above and below the elimi-
nator boundary. It is expected that this would not affect
the-collection efficiency of the eliminator but would certainly
increase the pressure drop, and thus it is not desirable. In
order to avoid this flow diversion, the eliminator walls at
167
the entrance should be more parallel to the inlet flow.
From these flow visualization photographs it is con-
cluded that turbulent wake and eddy regions occur in all the
eliminators investigated. These effects are especially sig-
nificant in complex geometries. In the calculations, the
turbulence effects are not accounted for and the wakes are
not completely resolved. In order to achieve an exact
solution, the mesh size should be reduced enough so that re-
circulating eddies can be fully resolved, and a turbulence
model must be included in the equations. However, in view
of the success of the present calculation for predicting
the experimentally observed behavior of the drift eliminators,
it was decided that this more expensive and complicated
approach would not be required. The reason for the success
of present approximation is that by using no-slip boundary
conditions, the calculated velocity distributions represent
the actual flow patterns quite well as observed by comparing
the flow pattern photographs in this section with the velocity
distribution plots in Chapter 3. Although the eddy and turb-
ulent wake regions are not exactly described in the calculated
velocity distribution, these regions are approximated by
stagnant regions. Moreover, the water droplets seldom travel
into these regions as seen in the droplet trajectory plots
displayed in Section 3.3. Therefore, using this approach,
the collection efficiency can be fairly accurately calculated.
However, the pressure drop depends greatly on the existence
168
of turbulence, and thus it cannot be accurately predicted
by these calculations if significant turbulence occurs in the
eliminator. For smooth geometries the predictions agree quite
well with measurements, but for complicated geometries (such
as the Zig-Zag eliminator), the calculation predicts a much
smaller pressure drop than the measurement.
Another factor that the calculation does not take into
account is the air turbulence inside cooling towers. In order
to solve this problem, it is necessary to have a turbulence
code to calculate the air velocity distribution inside elimi-
nators, and also to have quantitative information about the
nature of air turbulence inside cooling towers. However, this
quantitative information is not well known. Martin and Barbar
(M3) have indicated the possible variations in the velocity
of air approaching an eliminator, but it is also important
to know the frequency with which this variation occurs. Some
idea of the variation in flow direction was obtained using
an ammonium chloride smoke generator (F3). It was suggested
that the variation was sensitive to the ambient wind conditions,
fluctuating approximately 100 about the vertical with a one
second period when the wind was gusting strongly, and remaining
steady when the ambient conditions were calm. However, these
observations were made close to the tower center, and larger
variations would be expected towards its perimeter. The
positioning of towers relative to other constructions might
also be an important factor in this respect.
169
The following conclusions are made in this study:
(1) The calculational method can accurately predict the
collection efficiencies of cooling tower drift
eliminators
(2) The pressure drop calculations are reasonably good
for smooth eliminator geometries. For eliminator
geometries with sharp corners the calculation pre-
dicts much smaller pressure drops than the actual
values.
(3) The design of the Zig-Zag eliminator is economically
unsound, The third layer of this eliminator should
be removed. By doing this, the drift collection
efficiency will not be significantly affected, yet an
appreciable savings from the reduced pressure drop
will be obtained.
(4) The E-E eliminator designed by Yao and Schrock
collects droplets very efficiently as predicted by
the designers, yet the pressure drop is much higher
due to the occurrence of flow separation. Therefore,
this eliminator does not appear as promising as its
collection efficiency shows.
(5) In order to achieve a high drift collection efficiency
the drift eliminator geometry should be as complex
as possible. However, the pressure drop will also
increase as the geometry becomes more complex. It
is found that an eliminator yielding a higher collection
170
efficiency will always have a higher pressure
drop. Table 6,1,1 demonstrates this point. The
table shows the calculated collection efficiency
(using no-slip boundary conditions) for some
common drift eliminators at an air speed of 1.5 m/s.
The values of pressure drop presented in the table
are either experimental results, where available,
or calculated results. The calculated pressure
drop results for complicated geometries with sharp
corners might not be reliable. This table shows
that more complex geometries have better collection
efficiencies yet higher pressure drops.
Up to now, there is no available technique for designing
drift eliminators that optimizes between the collection
efficiency and the pressure drop. It is therefore suggested
that in designing a drift eliminator for a particular cooling
tower, a pressure drop limit across the eliminator that can
be tolerated should be set first. Then an eliminator geometry
should be chosen that has a pressure drop lower than the set
limit, yet has the best collection efficiency. This can be
done theoretically with the DRIFT code, After selecting the
eliminator geometry in this way, it should be constructed and
tested in an experimental facility to check that the overall
drift emission is lower than the environmental standard set
by the Environmental Protection Agency. This approach will
171
o 0
o L
o o
O Ln LC H t-
o cOj kO
o o oO Lf'.o 03
o o
o HO -I
o oC
o LC,m LinH H
o o
o nlm oHl r
o o
i t-
o 0.-
O
oC
o 00
O.
O~ -T-
Ln L\oD 0o
--o o
O 0O
kc CD
o 0
L~ OC)
0C)
O O
C) n Lr OH L -J
DO OD
O If L\0 .- CO 0n C C)!.0 '- 0 0.* H
N(\ \-0 H Hr-O O O I.,. k.O ,-- ,''
O O Ln LnO -C) C;
O O O O
O O OC) O O ) O O OC(' ' f -3 i' kO c- CO O\ O
b.-
H
l
Hm,-.I
H
H
b-,'*kOf
,--
--
r-
0h r
r-
r-t
- -
o o) Q
d
CLn a)
a)aaaIc Ca0 c~
N1
*C
co4-'lr-I::I
Ca
r(-a)
JLU)04x
**
.,bo
rH :3CEN
C)
4-4
a,r-t4o0r-U)O
0U)
CeIL)
bO
Cd, H
I -I
a( EQU) S_
ZIr
H,
0,-0
co
10 0
4 CdC)U O
C)
C)O O H
Hl O
a)
.0U() ('41
C) *1l
-1)bO U) HIa a ~, cqH c--
;:V
4~
H a)
Oct, -CT a
i
172
certainly save a lot of unnecessary effort and money while a
better eliminator is designed.
6,2 Recommendations
The results from the DRIFT code in this paper suggest
that this code is very useful in the evaluation and design of
cooling tower drift eliminators. The following future work
on this numerical technique is recommended:
(1) It has been shown that the air velocity distribution
inside an eliminator calculated by the SOLASUR sub-
routine using no-slip boundary conditions approximates
the actual flow pattern quite well. In order to
further validate this calculation, a spatial
measurement of the air velocity distribution inside
the eliminator would be appropriate. There are
many such measurement techniques; a suitable one
would be the Laser Doppler Anemometer (LDA) tech-
nique which is commercially available. Fig. 6.2.1
shows a recommended test section for this purpose
and for the droplet dynamics experiments which
are described later. The LDA technique will not
only measure the velocity values but also the turb-
ulent flow parameters. These will be very useful
for investigating the eliminator performance.
(2) The droplet trajectory calculation performed by the
173
AIR
MONODISPER'DROPLETGENERATOR
t
INATOR
T WIRE
EMOMETER
AIR
Fig. 6.2.1 Schematic Diagram of the Proposed ExperimentalSetup for Studying Droplet Trajectory and AirVelocity Distribution in Drift Eliminators
174
DRIFT code can be checked by a simple experiment.
The suggested experimental setup is shown in Fig.
6.2,1. A small exhaust fan will induce the required
air flow. Drift eliminators are installed in a
test section that consists of plexiglass walls
and a flow channel that is several inches thick.
The nominal air speed can be measured by hot wire
anemometer at the inlet region of the eliminator.
Colored water droplets generated by a monodisperse
droplet generator are introduced below the elim-
inator . Their trajectories can be observed with
a high-speed cine camera, These trajectories can
be compared with calculated results. This experi-
ment can also study the droplet rebounding effect
at the walls. Water film effects can also be
studied by introducing a water film at the elimi-
nator walls. Since the test section is small, it
can be placed either horizontally or vertically.
As mentioned in (1), the air velocity distribution
inside the eliminator can also be measured by an
LDA technique with the eliminators installed in
this test section. This basic experiment could
validate the present calculation as well as provide
improvements to the code concerning the droplet
rebounding and water film effects,
175
For the design of drift eliminators, the following
recommendations are made:
(1) It is found from the present calculation that the
third layer of the Zig-Zag eliminator serves no
purpose in capturing drift droplets, and thus should
be removed to reduce the pressure drop across the
eliminator. Also, to avoid flow diversion at the
eliminator entrance, the boundaries should be reshaped
so that they are parallel to the flow. An experimental
test of the performance of the Zig-Zag eliminator
with only two layers and a smoothed entrance is
recommended.
(2) The performances of many common drift eliminator geom-
etries have been evaluated in this work either by
numerical simulation or experiments. However, there
are still many other eliminator designs that have been
used commercially, and some potential designs that are
worth investigating. These include the Chevron-type
eliminator used by French cooling tower vendors, a
helical flow channel design with smooth inlet and outlet
nozzles, a "polisher" eliminator design, and many others.
(3) Total drift emission is the most important parameter
of the environmental acceptance of cooling tower
drift. Therefore, after selecting a particular
eliminator design, it is necessary to test the
176
design for its total drift emission. However,
none of the available drift measurement techniques
has been proven to be generally satisfactory-,to
the point of being adopted for general use (Al).
A reliable technique for drift emission measurement
is needed.
It is proposed that future work should be
performed to demonstrate the feasibility of using
a radioactive tracer for absolute drift rate measure-
ments, The suggested candidate tracer is Na24
(T1 = 15 hr.; Ey's = 2.75 Mev and 1.37 Mev) which
23can be produced from irradiating stable Na (in
NaOH form) in a nuclear reactor. After irradiation,
the activated solution will be neutralized with
HC1, and injected into the recirculating water flow
in the experimental drift facility. Because the
air flow in the wind tunnel is recirculated and
rapidly becomes saturated, it is expected that drop-
let evaporation will be negligible. Thus, the salt
concentration in the entrained droplets should be
the same as in the recirculating water, making
possible a direct comparison of drift rates measured
by drop-size spectra methods, and by deducing the
total salt current using the radioactive tracer.
This point has been a problem in the past in inter-
comparisons between various drift measurement methods -
177
those methods which observe total salt flow rate
(e.g., isokinetic samplers, cyclone samplers, etc.)
do not provide a measure of the droplet spectra,
and those methods which observe the droplet spectra
(e.g., the PILLS systems sensitive paper sampling)
do not provide information regarding the total
salt current without some assumption being made
regarding the salt concentration in the droplets
due to evaporation or growth. This is difficult to
provide since evaporation and growth rates vary
with droplet sizes.
In the test the salt current will be sampled
with an array of NaI detectors that scans a trans-
verse section of the experimental drift facility
so that the total amount of radiation observed will
be directly proportional to the amount of salt that
flows past the measurement station.
For a reasonably accurate (2%) experiment
it is estimated that the laboratory demonstration
will require approximately 10 mCi of activity
injected into the recirculating water. For a field
test roughly 10 Ci of activity would be required,
depending upon the desired accuracy and the extent
of the NaI detector array, It is envisioned that
this measurement technique would be used sparingly
for absolute drift measurements,and that less accurate
178
methods would be sufficient for routine drift
monitoring. In field applications the total amounts
of released activity would be no greater than those
currently encountered on a chronic basis with
boiling water reactors.
In conclusion, the DRIFT code is quite capable of
predicting the performance of cooling tower drift eliminators,
although some care should be taken in the pressure drop
calculations for complex (sharp-cornered) eliminator geom-
etries. This code should be very useful in evaluating drift
eliminator performance to aid in the design of drift elimi-
nators for any cooling tower. Provisions for the effect of
water film on the eliminator walls, of flow turbulence within
the eliminator, and of droplets rebounding from the elimina-
tor walls should be investigated and developed to supplement
the DRIFT code.
179
REFERENCES
Al "Cooling Tower Plume Modeling and Drift Measurement -A Review of the State-of-the-Art", Sponsored by theResearch Committee on Atmospheric Emissions and PlumBehavior from Cooling Towers, ASME, New York, 1975.
B1 Berglund, R.N., and Liu, B.T.H., "Generation of Mono-disperse Aerosol Standards," Env. Sci. & Tech., 7,1973, pp. 147-153.
B2 Bevington, P.R., Data Reduction and Error Analysis inthe Physical Sciences, McGraw-Hill Book Company,New York, 1969.
C1 Carnahan, B., Luther, H.A., and Wilkes, J.O., AppliedNumerical Method, John Wiley & Sons, Inc., 1969, pp.363-365.
C2 Campbell, J.C., "A Review of CTI Work on the Measurementof Cooling Tower Drift Loss," TP68A, Cooling TowerInst., Houston, Texas, 1969.
C3 Chan, J. and Golay, M.W., "Numerical Simulation ofCooling Tower Drift Eliminator Performance," Numerical/Laboratory Computer Methods in Fluid Mechanics, presentedat the ASME Annual Winter Meeting, edited by A.A.Pouring and U.L. Shah, ASME, New York, Dec., 1976,pp. 229.
C4 Chilton, H., "Elimination of Carryover from PackedTowers with Special Reference of Natural Draught WaterCooling Towers," Trans. Am. Inst. Chem. Engr., Vol. 30,1952, pp. 235-250.
C5 Chan, J., and Golay, M.W., "DRIFT-A Numerical SimulationSolution of Cooling Tower Drift Eliminator Perform-ance," MIT Energy Laboratory Report No. MITEL, April,1977.
D1 Dickey, J.B., and Cate, R.E., "Managing Waste Heat withthe Water Cooling Tower," The Marley Company, Mission,Kansas, 1970.
D2 Dave, J.V., "Subroutines for Computing the Parametersof the Electromagnetic Radiation Scattered by a Sphere,"Report No. 320-3237, IBM Scientific Center, Palo Alto,Calif., May, 1968.
180
D3 Dave, J.V., "Scattering of Visible Light by LargeWater Spheres," Applied Optics, Vol. 8, No.1, January1969, pp. 155-164.
D4 "Instruction Manual. 1173/1174 Electronic Manometer,"Datametrics, Wilmington, Mass.
F1 Fish, B.R., and Duncan, J.R., "Measurement of LiquidDroplet Emissions from Cooling Towers and ProcessStacks," Paper No. 72-35 presented at the 65th AnnualMeeting of the Air Pollution Control Association,Miami Beach, Florida, June 18-22, 1972.
F2 Foster, P.M., Personal Correspondence, CentralElectricity Research Laboratories, Surrey, England,March 17, 1976.
F3 Foster, P.M., Williams, M.I. and Winter, F.T., "DropletBehavior and Collection by Counterflow Cooling TowerEliminators," Atmospheric Environment, Vol. 8, No. 4,April 1974, pp. 349-360.
F4 Forsythe, G., and Moler, C.B., Computer Solution ofLinear Algebraic Systems, Prentice-Hall, Inc.,Englewood Cliffs, N.J., 1967, Chapter 9.
F5 Foster, P.M., Personal Correspondence, Central Elec-tricity Research Laboratories, Surrey, England,February 22, 1977.
G1 Gardner, B.R., and Lowe, H.J., "The Research and Develop-ment Background to the Environmental Problems ofNatural Draught Cooling Towers," Atmospheric Environ-ment, Vol. 8, No. 4, April 1974, pp. 313-320.
G2 Golay, M.W., "Progress Report - Drift EliminationProject," Jan. 1, 1976, Waste Heat Management Program,MIT Energy Lab., Cambridge, Mass.
G3 Golay, M.W., "Progress Report - Drift EliminationProject," July 1, 1976, Waste Heat Management Program,MIT Energy Lab., Cambridge, Mass.
G4 Golay, M.W., "Progress Report - Drift EliminationProJect," Jan. 1, 1977, Waste Heat Management Program,MIT Energy Lab., Cambridge, Mass.
G5 Grolme, M.A., Lambert, G.A., and Fauske, H.K., "Flood-ing Correlation for Sodium and Cladding Motion inSubassembly Voiding," Trans. Am. Nucl. Soc., Vol. 18,1974, pp. 209.
181
H1 Harlow, F.H., and Welch, J.E., "Numerical Calculationof Time-Dependent Viscous Incompressible Flow," Phys.Fluids, Vol. 8, 1965, pp. 2182.
H2 Hirt, C.W., Nichols, B.D., and Romero, N.C., "SOLA -A Numerical Solution Algorithm for Transient FluidFlows," LA-5832, 1974, Los Alamos Scientific Lab.
H3 Holmberg, J.D., "Drift Management in the Chalk PointCooling Tower," The Marley Company, Mission, Kansas.
H4 Holmberg, J.D., and Kinney, O.L., "Drift Technologyfor Cooling Towers," 1973, The Marley Company, Mission,Kansas.
H5 Hosler, C.L., Pena, J., and Pena, R., "Determinationof Salt Deposition Rates from Drift from EvaporativeCooling Towers," Journal of Engineering for Power,July, 1974, pp. 283-291.
H6 Holmberg, J.D., Personal Communication, The MarleyCompany, Mission, Kansas, Jan., 1977.
I1 Ishii, M. and Grolmes, M.A., "Inception Criteria forDroplet Entrainment in Two-Phase Con-current FilmFlow," AIChE Journal, Vol. 21, No. 2, 1975, pp. 308-318.
12 "IMSL Library 1, Edition 4 (FORTRAN IV), S/370-360",International Mathematics and Statistical Libraries,Inc., Houston, Texas, 1975, Vol. II.
J1 Jallouk, P.A., Kidd, G.J., Jr., and Shapiro, T.,"Environmental Aspects of Cooling Tower Operation:Survey of Emission, Transport, and Deposition of Driftfrom the K-31 and K-33 Cooling Towers at ORGDP,"K-1859, Oak Ridge Gaseous Diffusion Plant, Oak Ridge,Tenn., 1974.
J2 Jayaratne, O.W., and Mason, B.J., "The Coalescenceand Bouncing of Water Drops at an Air/Water Interface,"Proc. R. Soc. A , 280, 1964, pp. 545-565.
K1 Knollenberg, R.G., "The Optical Array: An Alternativeto Scattering or Extinction for Airborne ParticleSize Determination," J. Appl. Meteor.,9, 1970, pp. 86-103.
L1 LeFevre, M., Personal Correspondence, Research Cottrell,Inc.
182
L2 Lieblein, S., and Roudebush, W., "Theoretical LossRelations for Low-Speed Two-Dimensional-Cascade Flow,"NACA, TN3662, 1956.
L3 Lyerly, R.L., et al., "An Evaluation of the Feasibilityof Salt Water Cooling Towers for Turkey Point," SouthernNuclear Engineering, Inc., Report SNE-54 for FloridaPower and Light Co., Miami, Feb., 1970.
L4 Levenberg, K., "A Method for the Solution of CertainNon-Linear Problems in Least Squares," Quarterly ofApplied Mathematics, Vol. 2, No. 2, 1944, pp. 164-168.
M1 Margetts, M.J., and Shofner, F.M., "Characterization ofthe Drift Emissions of a Natural Draft Cooling Towerand Examination of Sensitivity to Operational ParameterVariations," presented at the Joint Power GenerationConference, ASME, IEEE, ASCE, New Orleans, Louisiana,Sept. 16-19, 1973.
M2 "Cooling Tower Fundamentals and Application Principles,"The Marley Company, Mission, Kansas.
M3 Martin, A., and Barber, F.R., "Some Water DropletMeasurements Inside Cooling Towers," Atmospheric Environ-ment, Vol. 8, No. 4, April 1974, pp. 325-336.
M4 Mason, B.J., The Physics of Clouds, Oxford UniversityPress, Oxford, 1957.
M5 Mie, G., Ann. der Physik., Vol. 25, 1908, pp. 377-445.
M6 Marguardt, D.W., "An Algorithm for Least-Squares Esti-mation of Nonlinear Parameters," Journal of the Societyfor Industrial and Applied Mathematics, Vol. 11, No. 2,June 1963, pp. 431-441.
R1 Reisman, J.I., and Ovard, J.C., "Cooling Towers andthe Environment - An Overview," presented at theAmerican Power Conference 35th Annual Meeting, Chicago,Illinois, May 1973.
R2 Roffman, A., "Drift Loss from Saltwater Cooling Towers -Some Atmospheric Considerations," Amer. GeophysicalUnion, 54th Annual Meeting, Washington, D.C., April 16-20, 1973.
R3 Roffman, A., et al., "The State of the Art of SaltwaterCooling Tower for Steam Electric Generating Plants,"WASH-1244, Feb, 1973, U.S. Atomic Energy Commission,pp. E.1-E.21.
183
R4 Roffman, A., and VanVleck, L.D., "The State-of-theArt of Measuring and Predicting Cooling Tower Driftand Its Deposition," Paper No. 73-140, presentedat the 66th Air Pollution Control Assoc. Annual Meeting,Chicago, Illinois, 1973.
S1 Schrecker, G.O., Wilber, K.R., Shofner, F.M., "Pre-diction and Measurement of Airborne ParticulateConcentrations from Cooling Device Sources and in theAmbient Atmosphere," presented at the 'Cooling TowerEnvironment-1974' Symposium, March 4-6, 1974.
S2 Shofner, F.M., "Explicit Calibration of the PILLS IISystem," EPA Project No. 16130GNK, EnvironmentalProtection Agency, Sept. 1973.
S3 Shofner, F.M., Schrecker, G.O., Carlson, T.B., andWebb, R.O., "Measurement and Interpretation of DriftParticle Characteristics," presented at the 'CoolingTower Environment-1974' Symposium, March 4-6, 1974.
S4 Shofner, F.M., and Thomas, C.O., "Development andDemonstration of Low-Level Drift Instrumentation,"Environmental Protection Agency Report No. 16131GNK,Oct. 1971.
S5 Shofner, F.M., and Thomas, C.O., "Drift Measurementin Cooling Towers," Cooling Tower, AIChE, New York,N.Y., 1971.
S6 Shofner, F.M., Thomas, C.O., Schrecker, G.O., Lyon,C.R., Carlson, T.B., and Bishop, A.O., "Measurementof the Drift and Plume Characteristics of PEPCO'sChalk Point Unit #3 Cooling Tower, Phase One:Experimental Design," Final Report submitted to Mary-land Department of Natural Resources, Baltimore,Maryland, August, 1973.
S7 Shofner, F.M., Watanabe, Y., and Carlson, T.B.,"Design Considerations for Particulate Instrumentationby Laser Light Scattering (PILLS) Systems," ISATransactions 12, 1973, pp. 56-61.
S8 "SPRACO Spray Nozzles and Industrial Finishing Equip-ment," Catalog 73, Spray Engineering Company, Burling-ton, Mass.
S9 Stewart, Paul B., Personal Communication, CeramicCooling Tower Company, Fort Worth, Texas, Jan. 20, 1977.
T1 Thermo-Systems, Inc., The Mlel 3050 Berglund-LiuMonodisperse Aerosol Generator, Minneapolis, Minnesota,1974.
T2 Thomas, C.O., and Shofner, .M., "Cooling Tower Monitor-ing Systems for Electrical Power Generating Plants,"presented at the Midwest Power Symposium in Cincinnati,Ohio, Oct. 22-23, 1973.
T3 Thompson, B.J., and Ward, J.H., "Particle Sizing- TheFirst Direct Use of Holography," Sci. Res., 1, 1966,pp. 37-40.
T4 Thompson, B.J., Parrent, G.B., Ward, J.H., and Justi,B., "A Readout Technique for the Laser Fog Disdrometer,"J. of Appl. Meteor., , 1966, pp. 343-348.
T5 Thwaites, B., "Approximate Calculation of the LaminarBoundary Layer," Aeronautical Quarterly, Vol. 1,1949, pp. 245-280.
V1 Van de Hulst, H.C., Light Scattering by Small Particles,Wiley, New York, 1957.
W1 Welch, J.E., Harlow, F.H., Shannon, J.P., and Daly, B.J.,"The MAC Method. A Computing Technique for SolvingViscous , Incompressible, Transient Fluid Flow ProblemsInvolving Free Surfaces," LA-3425, March 1966, LosAlamos Scientific Lab.
W2 Wistrom, G.K., and Ovard, J.C., "Cooling Tower Drift,Its Measurement, Control and Environmental Effects,"presented at the Cooling Tower Institute Annual Meeting,Houston, Texas, Jan. 1973.
Y1 Yao, S.C., "Investigations on Falling Drop Heat-MassTransfer and Drift Elimination in Wet Cooling Towers,"PhD thesis, University of California, Berkeley, 1974.
Y2 Yao, S.C. and Schrock, V.E., "Aerodynamic Design ofCooling Tower Drift Eliminators," Amer. Soc. Mech.Engrs., Paper No. 75-WA/Pwr-5, Aug. 1975.
185
APPENDICES
186
APPENDIX A
DATANA PROGRAM
A.1 Introduction
In the drift measurements by laser light scattering the
voltage distribution, M_(v), recorded by the multichannel analy-
zer is not the true droplet size distribution, P(d), as
mentioned in Chapter 4. However, these two distributions can
be related by
M(v) = R(v;d) P(d) , (4.4.1)
where R(v;d) is the voltage response matrix to droplets of
diameter d.
In order to recover P(d) from the measured distribution
M(v) in Eq. 4.4.1, a system of linear algebraic equations must
be solved. The DATANA program transforms the measured spectra
at the inlet and outlet sections of the eliminator to the true
droplet size spectra so that the collection efficiency as a
function of droplet size can be determined.
This appendix describes this program in detail. The
necessary input parameters are described in Section B.3. A
listing of the program and a sample problem are also given.
A.2 Description of the Program
The DATANA computer code is written in FORTRAN IV and
analyzes the measured voltage pulse height distribution. This
187
distribution represents the scattered light intensity distri-
bution produced when a spectrum of droplets passes through
the scattering volume.
The procedures of the code are listed in the flow chart
in Fig,* A.2.1. The calibration parameters determined from a
calibration check of the drift measuring instrument are used
by the program to determine the calibration factors and
response function. To transform a voltage pulse height dis-
tribution into a droplet size distribution, the measured
distribution is first smoothed by a least-squares fitting in
the LSMARQ subroutine. This subroutine can use any approp-
riate fitting function that is supplied by the user through
the external function YFCN. It is found that a polynomial of
the form8 a
f(x) = i- (A.2.1)i=l x
provides the best fit of the typical measured distributions.
Guessed values of the parameters ai are input by user, and
best fit.values of ai are calculated by the subroutine. The
transformation of the measured distribution to the droplet
size distribution is performed either by backward substitu-
tion method, if the response function is an upper triangular
matrix, or otherwise by the LEQT1F subroutine, which obtains
a solution of a system of linear equations. The backward
substitution method is the simpler approach and takes less
computation time. Collection efficiencies as a function of
188
Ip
(READ CALIBRATION
PARAMETERS
COMPUTE CALIBRATION
FACTORS
READ MEASURED
VOLTAGE PULSE
HEIGHT DISTRIBUTION
/CALCULATE COLLECTION \
EFFICIENCY?
NO
YES
DO LOOP
I=1, IEFF
DETERMINE
COLLECTION
EFFICIENCY
l
Flow 3hart of the DATANA Code
LEAST-SQUARE
FITTING OF
THE DISTRIBUTION
COMPUTE THE
RESPONSE FUNCTION
1 ~~~
, .
L L ~ ~ -
ii
1---~~~~~~
--- ·I . I
w .
--
l_ . .
I
Fig. A.2.1
189
droplet size can be determined when two measured distributions,
one at the inlet and one at the outlet of the eliminator, are
provided,
The DATANA code uses several subroutines. They are
described below.
(A) LSMARQ
The subprogram LSMARQ computes the solutions of non-
linear least-squares curve and surface-fitting problems. That
is, LSMARQ finds values of bl,b2,...,,b p , which minimizes
nE Wi(Y-f(x i,lsXIsx im xi ;bllb2...lbp)), (A.2.2)
i-1
where the fitting function f depends on m> 1 independent varia-
bles ( lxi ,...,xim)' and on the p unknown parameters, bj.i~l' l i,2i .. x,m j
The ith dependent and independent variables, i and xi,l,...Xim'
are known values corresponding to the ith data point or obser-
vation. The number of data points is n, and the wi are para-
meters that weight the errors at each data point.
LSMARQ uses the Levenberg-Marquardt algorithm described
in Refs. B2, L4, and M6. It computes the coefficients of a
partial Taylor series for the fitting function and then uses
the steepest-descent method (or gradient method) to find a
"neighborhood" in parameter-space where the series provides
an adequate approximation to the data.
This subroutine is called by the main program.
190
(B) RSIMQ
This is a subroutine called by LSMARQ. It performs
forward elimination with partial pivoting.
(C) YFCN
This is a user-supplied FUNCTION subprogram, which
computes the function f(x,b) of Eq. A.2.2. This external
function is called from the main program, the LSMARQ, and
RSIMQ subroutines.
(D) LEQTF
This subprogram solves a set of linear equations, AX=B,
for X, given the NxN matrix A in full storage mode.
LEQT1F performs Gaussian elimination (Crout algorithm)
with equilibration and partial pivoting (F4).
This subroutine is taken from the IMSL library (I2), and
is called by the main program.
(E) LUDATF
This subroutine decomposes the N by matrix A into the
matrices U, where L is lower triangular with one's on the
diagonal, and U is upper triangular.
LUDATF is called by LEQT1F. Its algorithm is described
in Ref. I2.
191
(F) LUELMF
LUELMF performs the elimination part of the solution of
a set of simultaneous equations (I2). It is called by LEQT1F.
(G) UERTST
This subprogram prints a message reflecting any error
detected by an IMSL subroutine (2).
A.3 Description of the Input Parameters
Card No 1
I.EFF, IRT
FORMAT (215)
IEFF= 1 for data transformation.only, no collection
efficiency calculation will be performed.
IEFF= 2 for a collection efficiency calculation. In
this.case two spectra, one at the inlet and
one at the outlet of the eliminator, should be
provided.
IRT = 0 if the response function is an upper triangular
matrix where the simpler transformation method
can be used.
IRT 0 if the response function is not an upper triangu-
lar matrix so that the subroutine LEQT1F must be
used.
Card No.2
A,B
FORMAT (2E20.6)
A and B convert the channel number to a voltage pulse
192
height with
voltage = A + B x Channel number.
Card No.3
NC1,NC2,NC3,DC,CC1,CC2
FORMAT (3I5,3F10.3)
These are all calibration parameters and are obtained
from a calibration check of the drift measurement instrument-
ation. DC is the monodisperse droplet size used in the cali-
bration, which is about 80 pm in this work. The meanings of
other parameters are found in Fig.A.3.1, which is a calibra-
tion curve recorded by the multichannel analyzer for mono-
disperse droplets.
Card No.4
NC
FORMAT (I5)
NC is the maximum channel number of the data.
Card No. 5 and Card No. 6
(PARAM(I), I = 1, NPARAM)
FORMAT(4E20.6)
PARAM(I) are the guessed values of the least-square
fitting parameters. They can be set to be 1.0, however,
values closer to the true values will save a lot of computation
time.
NPARAM is the number of parameters. It is 8 in the
present work.
CI-
0c43
Cd
J)CdU
xO
<
4D
H 'I * C. C)CI) C-)
SImNo3 :10 diguntFii
193
194
Card No. 7 to Card No. (6+NC/5)
(PV(I), I=1, NC)
FORMAT (4X,5F7.0)
PV(I) is the voltage pulse height distribution measured
at the outlet of the eliminator if a collection efficiency
calculation is to be performed.
If IEFF equals 2 in Card No. 1, the following insertions
are made in the input deck:
Card No. 4A (If IEFF=2)
NC
FORMAT (I5)
NC = the maximum channel number for the distribution
measured at the inlet of the eliminator
Card No, 5A and Card No. 6A (If IEFF=2)
(PARAM(I), I = 1, NPARAM)
FORMAT(4E20.6)
PARAM(I) are the guessed values of the least-square
fitting parameters for the distribution measured at the
inlet of the eliminator.
Card No, 7A to Card No. (6+NC/5)A (If IEFF=2)
(PV(I), I=1,NC)
FORMAT (4X,5F7.0)
PV(I) is the distribution measured at the inlet of the
eliminator.
195
A.4 Listing of the DATANA Code
196
C OC
O Ox ZH- H
: :
m' 3 u' % r' 0 o r t W. ., r cc ' c -- 0N ::y In r- o o r - 01 o y n OO O O O O C - r r- - - - q. r'r. ( .N N C4 .j N j " \' oJ N 'f Cq. f ) m ' 1 r
O O O O O O O C O O C O C O C Oc, C C ~ $O O O o C O C O O O O O) O O O O O O C O C C O O C O C' O O O O O C O O O O O O O CO OZ Z Z; Z Z EZ Z XZ E: i E X Z Z E Z E: ;ZZ E: C z Z IZ Z Z Z Z E : E ZE
CU
gr;
E-
J;
CL
-
_. _
r ).m "
z-
C
,- .a,:FQ \U. Z
C; \
3C .--
1- FQ
2 ZP O _ E .
m - t;
.- \L C v,\ LU z
Q r C _CP-2 W·
.
c 'C O
. or
_ U
C
E-'
I-C:
H _
Oc U W Cu
O- W 3 11 XF E r
.e -II x' 'C
u-r lr. u7 L. PY .
OC >Or C C
11
014
rz.
,i
Ln-
r-F -X7a'
Crv
4e mm W
E-113Aac WC W: Ix
M o:1,4 C1r-e LI)
m crO -
ONC) -
C('q"
U') Lr''c 01
N%V _Dr
%
O -
.- cor
O3l O'; b' z,,-
.r 'O CO
) ,
PC
-n h4*O` C-CI a Pcca a X
00 0 cz cFl )-r H rH 11
Cd a Wa p; z(
g;
Cn;F
H _ -a; '2 HZz WH z M - OH z; Cz H
ImEI U e H t-j t D _
- E, : ' O E -tM ¢en z a E-,rH U
a a D L P;
W: En .o ' Pri
C P U E4 U)E CO W' I,·
V- m E-, ' V.t --rI 3H: P ' -4-a: i a
*C _L l >l b 1 4
F--:; 13 I: O' 13 r-.
O * DCC *'*O_ __ * ' I- ' ' " N,r f r - 4 '*
_1C *: oc ow v u4 oZ om~ - , At ox W c- *S _ X:] X C C C IFCr4 C s Crl CL r C O rr.o P' L 9 =; \ \ \ \ \ \· wj m uT r \ \ \ \ \
r < Cr~a~ c a KL' cr 3: <
O C O C C C C CL O~ OL C
,, ZC: Z 3
S~~~~~~ z EnW Lr. DI< a, * S) W 1·\ :Z. :
PI ~ C O Zc? i} ' ' c 5Vc C E
F: * qr"*O a:: . ' -: tn Z - H c
· .' *Iz Ub ~ :- O' -; F. I# U~. C HO -' ,,L: * l IftWs C o< F S N H " U E ) ;-, * E,; Z E'. P- 'X 'Q * In
' ' · J-'.) W : U
I'-- - - ~- ~ C. .. U ~ Ia
JI ~ II I 3-.IE...~'~I U _c > .I * 11 V U U)(L)i~ ~ z t: 'U U C= mgr: -Uv L -U ,~~ Fi Li X L 1 :fZ
, _ s \ t H L P f Z
. *r I z = cX ><f P- W ~- FFZF C 3-L r L b )
\ u` rF. P. 3:: CStiII
u ' '· SL C X Z <
t· -~ P Z C' _ rli
I I I Fqe H >' rr t , ; C:
2: L2: U P P H a: t- r bF C. S a _- t ,_
re, - L; E; Er < U OA L F p O ; tr- " r U r- m C - c "
.---
rl zz� U -! %'�' r-�r- T-' T- 9- ir-
Ut tU L' UU U
E Mc a N M :-t Wo r- cr a N MM Mn 3 n r * :t ;r S _ e : urtn n nn u
00OOOOOOOOO00000Z ZZZZZ ZZZ ZZ ZZZH H H H H H H H -H H H H 4
oceC oc E : E E I¢ I C 0¢ ¢ : E ac Er : M : M M v M W D M M W: M M
I U, %O rLn u) L)
o000H IH H HooooI: Ie ZC ;:
OCD ON o o o o o C
zO O Oz-H -4 H
a1 oEE wc r
M uD W 0 %, 0 cr ! i - -N000000a0a000oooooooo000000ZzZZZz zzz
H H H H H H HAz a Pi s c a a A:C M M M vM M M
NL)U
9r
,U
L)
cr
.U
,4 , h -j 0--a.-
-U O-
P o * Cz C
P. LD P; - (N :
_. p .Ca ~IC) C r L 1 ~9- V r ++
_I
10 Z
E-
II I C II H*U U & Q Z 0-aC
L o I*
u I U
SL C pw - C 0f O
L a G IN C
Pd0
S-. Cv U ')
HS %-II
Hk H , _,. -I-4 "-
1 S u .
l , -: II
0 'C . 0 i110 or A C-
0,0 W 0Cl 1X M AZ ,
ct
=1
H E-e
-C
LT
U
II
II H
* zC
F4 F- F- C 11 E-Z Z ZN -Z
-H H F H H
sa. CL ~C(N
197
C0
C
z2 M
C.O
UUO
' H
,: rL a. D.
zCH
C:,r,-,
.)H00
z :
S [-
C
H h
C-- V,z IH r
1k.O-Z
000 u
' '0 r N m I uE %O r"
0000000000oo00000o000000 pC NC o: o: 0 o: ad .
C x w r w xE E C x
0 0C a0
o oC E
o0 - (N r- ) D r- oa o% o s o- @ a \ 0 y oOocooooocCOCO o oO OOCC, 0 0 0000 C- H H H H H H H
E E ~'EcxE]
oc '- 0 0 I. - O
000000COOO-ocoooooo-4 H H H H HHH: 0C m0 : C x C : C =EExE : EErEEE
_ _
13. <a: 0 a
1 z C O:' ' ' I *!P G H N"
Z Z O M
r c -- tc* 0 (
0:Z Z E-4. H _- _- O
C>~1 0 1 H4 H t
Cv:
HO
C F O -- Cu " * :::, C
-4 Z -' ^ ·
U _O L r v- * III I .. ~> . - -
+ - t- ,
.C 0 _ - : X 0 0; 0 . -0. H U
II
c
r -
Y- U _
C _ C
_ 0 " O
H F F- F. I C II -'Fz- 2 2 2-i _ 2C 0C 0 L O 0-U, 0- 0 J z C .
C Ct C
2CH,-'
._
r1.
U Utnz 2
I C
0 U- I
P Z
C . t!Z
CCt· PE-1t
\ .P
I
* *U U* *
L I-ZE-. ~. H
a n Z H H
L; t; U
00 2 z
H HocPCI-I I,.
U) \o 3,. r- ri
I o o o
198
o o0 -O OH H
C
V C4- r-
0 0O oHH0; 014
f :t u0 !'- w 0r~
T 0- - 0 - T-
OOOOO22H H H W4H H
: x: = ; w: x v
uN .
u rI.,
* II '
I % %.2
(N C4 '
C .
- - rt r H - .0 C' C C ~ ~ P m C'~~~,u l
OH
H
U)I-4d~HmE-40i1
0¢3pDpi
.,
0
2
0
z
.,
CE-0PHE-4:DCDHW
a:
En
ac
E-
r- -o o aZ Z
IC 3: r-
Ma 0 % D r- 0 o -- V C I V- r
2OOOOOO22HHH W H H H H I-H
: c:EEEEEE a : EE ax = w w m
mC~ ro OZ z
H
E:E
H
b
0V.
:-0
4 O
lb
O CU
E -O Ln
C ;
_- E
HS.J2a C & a. .- H'- U QItL; C * t _
E., Z- z 1. <: 4
)-i U Cfi
UJ2zm
u3
u - IIO C- I.; r- ;
um r...)
o'-r
C
1-44, II r . + I
- F E- ',-C" U '- + a a I- HO II 11C II
o 11 II E C It : 0
o IIt " :: . .':P :r
= Un)%r- 0 0 -" ct" m m m --T -- -- -- --t
00000000000: Tr H H H ! : 1; I: HHH H
IC-
1-Z *. C.
I I .IN " I
ur-. - 0CL O r
H ,-
H"
,.. O
C CO CU'3 %C
U2
HHr~PI~
2i04
HUL)H
C
[..,
U
cHIL,
0IE
U U u
199
0
CIs-(N)0z
H
P-4
U' U
0O00000ZZZZZZ'_
O OI II ,oI r- i.
O O"" _ H-E,,- H -1IP1
I~ H !::.* .-.
. -H HI- '
x H (N > H t. I
T U o0 a r.I IH Z a H a 1-i Cfl C
CN - .'u) oD r- 0 a, oLU Ln Ln A u) u' Ln ul) UD
C OO CHHHHH4 -. H $ H 14 o.4 McH
x m x : r: r mZ x
IIC
II H
-,
C~
- - U .. ." ::. N, ~., - -. --
r-,
II
t-I
P-HH
H
b 0
,O
E- C HZ ce Z I
oa COz L
U
200
201
0000000000000000Co000UULUUU
a:
a:
r-'
H I
U < a:= u
C? H Lr * I
202
O000000000 00
VI Ir, 0 u-j .j .J -J-J J
- ..
UJ J r
£1.-j
AU' OLU 3
LUu>0 V cW 3> r
0 O>
zJ (a ZZL 0
X D
0. zS ML CU
0 : z0 (A D S.W In I
LU:IC
0 0 000 C 0 0 0 00000 C 0 000 0000 O0. . .,,I4 __ . C .I N . -. ' ., N Y ,- ,_ _ .,. c00000000000000 00000000000
JI J Jo(A V) V(A U, A (A J) Le VI J Jn J) *A LA JA (A JJ
czz 4z 4
--
a uL:
tZ
z_ z
C 0.z
LC cc-
u O< LU -
S Z X
z-4 z- --4 p. -
,0
:zZ
A
_ N
-',.r N
z Cz
ZZ -
4Or
0 DM-- .j
I,.aJ N
z e
Z 3
:t W
0
woC
LAZ c-
- 0I _03 u.
Z
.OWOZc:.- Z U-4 LL
<0
Z L v)
LiA' ,J rt) V M-I- ' 0.0J( I. 0
14( A ll-I C
U)
Lr
C.
I W
,
JI0Ir
.=
,J
CL,~IZzt
V)* I
* Z
OZ z 0
C '0
.- * _UJ 'L
4IZZS JUi :* *
0 0t _>:x:
-i UJ. I.
W I
-J0 rO
LcCl -
Z4
CDI-Q
2-LL'
UJ
U
bL
-Z
0
-=
UJ0
0z
CD
,o_Cz0I--
x04
-JLLI
I.-
0L C
O wocx>
C)
1- (A>0 .D
-I -,,-
_-. I,-0 Z'c U.J - aLC D WO J Z I
UJ -. -> 4 r
C3 W ¢x~ZU I-Z Or C tn
LUJ
O0 0C) 0
I 11rI z
V U U0
000000000000
0OQO
C
i- .
ZW
Uw
zt3:
U.CLo 3X Cz 0
w Z O
0 <co tY
a 4
Zu-
U L U UU
O OG 0 OOO O OO OOO
0000000000000o WE oE 3E -:E E or, 3& Uc %E i: :if.
-J .=1.. -J J .J. t . J .'t
00 C OCO000000o,-4 N pj , UO f ¢ C O -4
I, U 0 ; rU c0 u3 V3 V30 In (- JJJJJJ000000000000JJ.. _ _j _ j . j - J - J -
0I.U
S-
X
V)I/-
z,.
LL
IfOa
OI,/0
,:
z
Z
0.
zI z
z U
W -
CtII
0
0:
(N
0-0'-"
* w
-4 LU
II I
U". * __! _
* C:)
0000.0 IW W40 W I
U C -4 jF *- Z 110 V S W -
3, I '-r -
Sw 3 -LL UJ -- LL -UU·. U w L3L
1-3
:-4 U
0
_ 4 -* 1-4
_ 4
O LLZ 0:3- C 03 L 4
II U. .I L
OZ <4< <,~ ,.., ~ :3 LL
Z Z: <V0 44 00 E 0U 0
uI0O C0 ULer %J - fM P.
0U0
54w
m 0o O
0W0O
CC0.II
'3I-.
0~
O-4
00000000000M rfr %t Us-C t - cr C, o _ Oj10 %0 %O %O 1.0 %C 0O 0 - I-00000000000: :E :5 c : 3 X ; : :EC 0 0 J 0 v CA 0 _. V1.-J i J J J i J J J J J
-J
4
.
I-
zU.
3
Z0Qw
UL Z
I-
UIJ0. W
_
- 40W . 0 .
Z O 3:0W _ C L OWUO O-1- 0 CI 3 0
O O
o-I Io
2:
40.
II
- UJ
-4
cr03z
5.
V
203
0
UJwO-I
-
OW · W
z
OEO0. W 0.4LU
4: W
3ej L I a
U.
00000000000000000O0 0 o o o o o oo o C'o o Co C).L V,(In V) Ln = 0 0 _.. =.J 0 l I.J , i -J- J J iJ J J>> *I - i i J JJ J J
i..) O C-) O O 0 O O 0 0 0 00 0 O 0O-W r" - LT a r- a, C O -4 N o s ULt 'O P- C(c Ca c C' Oy " a oa a a o o o0 o 0o o o
0 0 :0 U 0 0 00... 0 04 4 O'E~ 0 0 -4JJ J J JJJ JJ 0J J J 0
V-4
I-ac
3*
a-I-.
0I6-
U3
0-4
I-
C)
It
9-
00
z4
Uj
-J
a:LU
a:
L#-I
0 _ 0
_ a C C -* cr
0 Z 3W
00 0 0 :
*:z Of
Z 0
co a, _ 0
ZI- LC-Z-Z
CLU. tL'
4
c04a: o0D0
Wa:
L) t
2: -Lr I0
J
-J
Z 0
>-
-JLUJ 0.U -11
awO> 0
a- a:
ui -.--.J ,* C 0a *Z
¢ >
II <> 11
>.-w 2:
L. UJ>O-F
EUJ
N
C
4 11
_ _
- Y
U aLA3:
I.-ZLL'
0a0L-
a:
z
-J4.
N
N-i
O.
esi
ceC3
.
0zt
X:cc0zCYZC
0rriN-40
CZ
ul I0-d '-4
I
-
-4LL I-- 4
-4
If
C3000
U. UU UU
204
U.
01
.J a-C)z4 Z
0aw03
LU
J ULiJ LL
a LL*
II <
_ 0
t U. Y
0. -
Z r LCK-:
0.a:
030.9-
~- aN
0
LUa-
9.-NH
z0w-
LU3.4-JIIc0 0C 04
X
a:
Od-.
U--
4
:L
0U
L.) 0 U
205
0c 0 -4 N M r LM 'r IC C P- N C^ -- U- C I-- CW a 0 - N - U C i" -C c0 C 0- o-' 4o _ -_ 4 w _ _ _ _ N N J "N e N N M N ,e- Mf ,f M n M M 4 t e.E P4 P"4 e..I r- v-4 w - -4 4 -40-4, p.4,44 - -4.. 4 - -4 - -p4 --4 p.4-~4pp4 p. p 4 ,p4 p.4
0 Z 0 V 0 0 0 0 0 I 0 0 I 0 0 0 0 LI I 1 V I c n 0 0 L 0 0 I 0 0 I In 0J JJJJJJ JJcJj J JJJ i >JJi v j - -j .j v u v j J .l v, v , J -v vJ J J J J
*
I-
J
Z 4N a
c ccc I J0 I -W3 40
>. 0
ZW F _w o
CY L) x
Oz C0 4
6 0
W O
I--
U -
U..O b"C~"
Ice-C3 n
U p
I II t
11 %t- '--4
- -
0.00
W
U
Z_q 4zW ,0
Z <
WI V
I- - Z0 Uz C
0· LAz 4 D
OZ ZL
0I-- : ' ZZ Z WZ ,' , I.-U ,- U. --,' ZZ .= I,- -:0 w 4LL = 00 > .- c
00 0 00%r t UJ
.J
UV)-W *
0-1e ,x 9-
<O o of -00
. I I. :
L-,JYo
oLLD> - U
N0'-r
OU3L* *z4 _
U
4+
II I
_t C
4cc
C.z
0'p.4
0
00V"4
U3Z- U
Z Z
0,
0 0'-'"-4
I
0.
O
I 000,-,4 N -43x3E
1:390&n .)0
U
UJ-a
0I
L0
zZ
C C O O O O O O O O Op O O O O O O O %O 1 O O , O O O
4 -9.4--4O -4 -4 -- - 4 9-4 -0-4-0- -4949-4 9-4 9 -4 -4 -4 4 C -4 _ -4~O O0000000 c30000000000000.) 0 .0 l0 .0 0 i 0 ) 0 =,_ 0 L.0 .i -) i = , 0 0 Lr 0 0 0 0
-j -i 1 - j i - J i J J j > -j >J j > J i -j -j -j .j J -
oooooo~.-- 0J -o,%% Ie co Oll I- f- r- P- , ca Y a
JJJJJJLn 0 0 ) ) )-J -i -i -j -
WUI-w
L
0o
Wo4
-4Z00
-J4
cc
4ao40.'/4n*
0.
-L-
.-w0
0.0w O
* ,. II11 - 0,- 4 V) -4
( ii - 0OX X II - i/ OZW - Co -4 ' - -4 w*4 o aLL -CL - CL I- -,0 ) O L c
.. , -4
U. < 0 04 > 9e-
Z
aOzIU
Ln0.
I.-IA
4
CL
-4.
0I-
03
U
Ir ZI
Li< 1
eCm0Z i* I-4
:O 0 4N .
C000_. N
0
00
Wo0OI0
L
uj 03
**
_
C -
4L3
O -
Z- CI Z FI0 * II
Z ,4 -* II Li ~3II Y
0 -. :NO --.NO: IU.
0
0-4
N
-
o 0
I'l%ow
0 0 .-- 0 I-N 0 Z- 0 U
0 A. U 0ze OW4./ Z *- U OLU U.
W C V -
z _ 0 00 p
- a_ - ZOZ 4 U0 U .O IL I-
O U- U.N
U uUU
206
W-J4
Ill0Z
UII
WZ4U
Z
I=-0CyZ>
V)a
IAZ
0
0w0
AJ-0.
z
0IU0.0,,
4)
207
0oCD eOCO ct 0 oD 00 CO Co ' O C% C O CT C ONP-4 I ,-4 P.- 4 ,-4 . -4 . =4 ,-4 , .4 -4 ,-4 ,.4 -4 -4 .
Q u) V O In u) cn ) 00000 n 00000 u
W
Za -z4
UJ4 ZfJ LU3 0.W Z .V) Z* _< oo N Z
z > Z 0 . -W zS Z
OU' U D Z
4 :, > CL I tY )Z -
' U r {
U Z WO .z,' D ccw z w a c2!-0 lL Z O ,-- *'" Y' oc :: OU 4 ZI- 4'CI UJ cD I WU C O V)
wc~ ~ ~ z z -4 xc -40 O Z Z Z 4 0 ui- - - CO f
0 1Z40 W2 0 L 0 0 No OZ
0 U O..JO d? cO W3 V UJ :-c u. .W LUWQ UWJ OO -UUU .
208
000000000ccN r7! W % e F C 000000000oooo000
0 0 ) 0 0 0 awwwawww'wawaQ Cf
000003000 - N m - L 0 I--4 .- -- - - -4 .-400000000
8 a ;E a: : a: aCYO O Oe TOOOOOOOO0u v v r
oo
W 0
aaOf
Z 0 iJ o uZ :o
a w0t W0 C
O Z O
0 .z cw ZO lL Z uu- Z -W) US * O0 1.W 0
U 0 ZZ O
W
U
o c 0 - N U'0 O 0 0M 0 - I N 4 U 0 ,- - N N N N N N M M M M e. , mMCr. Q C) oW W cc O o O o o CC W c Ou W oaNNNNNNN CCO N N s A WCX e 4 3 00;: (:0000: 0 o00 o::E Eo: Eo o w0ooooc U 0 ooc ur0 V oc oci uy~~~~ Y. CKC YC !YYYt hzYMC
Z
Z o O
.J
0 Wa I--
0 W a0 0Z JI -4
* W O- 43 LUc O C_O O t
0o
a:C
O .0
Io
I!
C)Oc-0
0N
OI-
O0
-
U.U.W0U
OU.o w
u It4- t) -
W.LUI-
-
0
LL
z
Oi- 0
UJ0uJJO_ J
z4 I-
: 0
WL -
, J w
o W LL
0 U-JN
I1
l.U
6-
Oo
F- -
3L C
a
o
UJc4Z
3-
CDUWz0
O3E
zU-Z-WWU.
-0
- .
U UU UU
a: za.
.
IL e
O --UJ 4w C
W <
0 W3Ca > n0 4
-4
0Z-4
o
z-4
O0oCA
0.
zu Cw
0LLZI0A:O w
0Wz
J W
U U UU
0000 00000 000 000 C 0 0 0 0 C) 0 0 0 0 00 0 00 0oo , N r 4 0 ' 0 a a -a N _ . L O - * a a-N ' a ur ao * a C c Na a a a , a a a a a UF a G UE S a UE o Un a a a s0 a a 0 vD 0 ̂ 0 a 9 a00) 0000 0(00 U L U 00 0 O U: cf ul 0w w w w w cf c c 0h a c c w C w w c cl cc w w w cc w w w Q
Ui
C
o
UJ3C0ZZ
Uw'4
C Ia0LU.
C3
_ V)ofI: C UJ ._ 0 0Z"~ II LL
A AX I _
l-- U.
,,"3 : I.I ci
, Z1--C
(.J
O
0I--
0
U
00
tz
0U
U.*4
C:Y
Z
.4.4
, 11
n 4 tO
0c-C
CL
-ICU.
LU0 III
CC
o-_, O
0L 0
W 3W
U-.
_ _
_ L
. LUU-
U >
: I l.
0 U 0Le 0u.I, (A 1+
lu. -. 11 o I 4.
. - - Z i cc:D O- W -
) Z UJ 00 0 0 0 U . - Z Z Z
Z _, _, _4 _O . L- 0 L tI Z Z :
3: a0 00 0)0U% % I_ z V ao
U L)-0 L
209
o o o000cr C m000) V VI
C c wr
.0U C.
_) _.
'U-X >
oc 11 4
" 0UO 't I (A
Z LL L..
0WLL -C
a
U
C
I:I
3W
UJ
IoW
.1
I-OO
l
z*
lLLL
--
w
III
II '1"
-3'n 0
0 .0n3 YC
U
I--zZ:Q
wz6--
ZCL)00 r
U )
210
000000m ¢ r r of,- t,. . I~. p,. I,-0000002: x x: 7- :E
I Q
z
w-ZO0Iw
ZD o
O
I V)
Z cc
D xN
00( J
0000000000O N un vo % 0 cr 000000000--0000000000U. L . U. L U LU - L U. LL.- -. , ., 4 -- - -4 -
WJ WWW -j uI J w w
OOO Q00000000000000 O 00000_-4 4 -_ -4 -_ 4 - 4 N N N N N N N N M fn M} M M Mn M00000000000000000000000000L L L L U. L. LL LL L L- LU. L- L L-U L LL ULA L L L L LL U.L LL L-_ p.4 _4 - -4 -- - - - - ---- - -_ -4 4 _ -4 -4 - -4
WUJ W U UJ UJ UW UJ W W LL' J UJ LU UJ UI UJ W W WUJ W J J J JJJ .JI J --J .J -JJ -J .J -JJ--J J -J J i J J
L.Q
--z C) 0.WZ 3Z
- ~CJ : I.- U
10. Z w 2 U,>-0 ZCc ) cc uj
W U)) W U.IT CL z 3U W I Wc UJ :w a X
0 a 0 ~J4 Ot UUJa Z tD W
LA-A 4LflZ I- Q =0co Z 3
- 0Z3D U·· W cc *l-C) U Laz D u.
110 ce0a Op a D V wCeWx o UJ . a LU a
O ZOZ
Z-zI 216.4Z- -X
40cc U400 xI _
J ZN J O
I Z0.)'-Wz z.i 0-U tAz
C
L:-X
'CcZ E0 Zc
_4-
LL0.z
I - 4LL0t
*) OC. V) Ln aW :- W
IBO U ceI '- 0 CY
X Ji z LL U0 Z. L 0. W
W w W wUJ w 0 W-ZY WUQ <5X UJ U U :
I Z 4<z L O c CeDO =- W w' -U L O U WI ~- WO0 U. 4
D Coce < U.IL *I C CK
IC I WW W 0L A 0 0
I CI LJ C: ^ c0 -WO0 P-I: Z 0 VIc e D m 7
LU I VI L W W
(A WJ W ( 0I * cew L >- co-zoce-u4 Lz 44 D * Zs r _ VI MI c aI ;:) C 'S 9 03 W
c a t ICy c 0 a
.C a 2 FC w Q t3 Q CI C V Z-<O Dcom2-O<u cmcc < CD W
I O Z --30 ." O - 3II I I I
4JL Z
4IU
Oa46
SC3
CD 0
0 0 0 L ) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 00UOU U
211
J o.4Waw Dx JI Wi-Ce6- 0.cr c
0O W*) CL
< U)_, a
I40
0zc Z Z.z HO
c <_ W
U Icc LUat lJ
I-F
'o ZoU-zWWC OZc 0-
a. z
-W e40-*-. 0 -zL-
' Z
U.I-W(J0- WC LL U_ Y -
U
M J
4zI- W;Q CL Z
W04ce
0 O'-ZQ
O Zv, .41-3 OjJ. ZU)Iu-0(AzW
0 0v, U
W zOW
I. U46
< CL4D00W
c W.40W Z -O.-
I
I
I
I
W I
o III
W Ic I4 I
.I3z IO B
.- B
_ I
CY I
14 -J
- I
o BU 0
Z IZ Io I~ I
: IO -C. I
W
I
I-W
*CUcc0UUw
ZI
)W-11 ~ < O
) - 1j
I 2 <.4
I U.zz+1ZII 9-
, If
zCZzZ:t3
W
: U.I-W
CK
-(A0ceZ-z(A
Z m+4a3 I
V)
11 LUce
0 U0LU CC _x cc C:< AL-
* . 4 Z Z
CC- 11O O x6- cc ez
UUJ F
I I I I I
z0I-UCD
LL
W
(A
ccW
W xc
V) <= CL
Wcc >-0 .J
ZLL ZUO
Z.IZ o 0C) Q 0 0u' U --
O 1
C JZ Z
::: -9-
W 44ZUIJ UJ II > U: :
0000Q00C 00000000000o0000000- CN1 4-t 0 z- I 0 G O- N 4 L ' - ' 0 (N ' 4 LOP 'O - MO O 4 C O O O O O O O O O O CU O O O O O O O c O O O
IL. U- UtU. L LL U. U U J it .LL U U L U U L LL UL LL LL_ ----- -- p ._-4 d, _ - p4 -4 _ - 4 -. 4, -4
L t W ' LU UJ W WL U U W U W LU W U W WL U W U ULI
I I CW I Z-,--:: I <
: c I
Z: I a:.z 1 ,U- I O
ZO I 3 U 0 I * JI -< F I W * Z- u I W W
- O-13 I - Y -4 - 4 --
W. I 1:[ 3 Z U E O <t4 - I' I W W: 4'0 ZUJ * V) "yU CI CC V aU c t. Lb9 L In I _ ' otx 3 O- _ 0 Z I C..) * -_ - LN .ao
O U.: I 3 C -zaY 0 | 1 Ln* ex' * 0 cr
/ 3 I l I <{ w - < - ° WI < < - _. JI I I I I '- ~,C9 :9 U)Z0eI0 * CY , _ U. 1I Z 0.- O 30
LJl I 4 Z < o 0 W- I 4 O4 -0 1 - l
W I '-4 Z Z I C- 0 . l o" I X 4- U- O _"I- IZ U v' .
IU.), .O I , U) U. t- 3W C I-V I W w -V *U L LJ'
I - Z Zh < c J · -.J I > c 00~ 0 ii W W
Z vI I W - D -' -- JZ D ua 4 0. uJ I U .J W -. Z L Z : Zd '""t 0 Z.J 0 < - -.* CZ I O ZJ O - Oi t -- _n : I 1 n - 1.J II .J C - - - -- J ·: I cO Ex D Cc : Z Z -l- C
Uo o I UJ : -- O W <t t O C LL O < W ZW Z L- A' C: Q _ U O U - U
e I 0 I 0 C 0"i . s . . . . v . . . . CtUuU U UUU U U C) u
212
00000 0000.ULL L. U
4U W -W LU
>-crc
U-O."
OOOOOQQOOOOQO O a O VO C) O O O O w %~O - w 0% 0 a N m It VI d0000 0 __ - -4 00000000000044444444J
col J c ~ cc-JJ - cIJ-J
0z-4z
> 4- 9-
_ 0
Z
-4 LuN ZaW * Z9-h-
) Q Zc< UJ
V.. *31 aLiZ *-
J Jr 3 IL4 * C:I0ZQ UJO4 X
L a C
0 ):=: F
J a04 Z4:~~
4 z
UJwV)
0
4c, XU. UZ I.- 9-LL 4CO) u. U.0 0Z J-Z 0 i cc
Owwa wtZw FZ a C:0 -' 0 0
ui 4A 0 uj UV) Z 0. I: L
0. x0 O C WO 00 LL CU C U. Z
| wZ crWU. O ZI-4zz -
CICwi i'z0I h. 0
)X OW ac D_ LU * JI-Z W u) <J4Qw -V)
L: 1) c" a a V)
093 1w W ; 3 at a cI -J O X Dt 4 U L U CL
UwCc
I I I I
ox Ji
Z-4
O O O C 0 N S L'· O O O O O C - 4OW Ooo 0J U o Ooooooooooooooo_ _ _ N N N N N 4 C-4 N N M M M M ,
000000000000000000004 a r a Q rI I Q a sr Q 5 ct a a a a a a <a ^ a C C C O C: O C a a Q O C ) a C: O O-J J J J -J - J J J -J J j jJ J X J J J
Ui V5Z -
LL 00 0a 0 N0 UJ Z Z2 *
za C- < C u u ° O I 0. . > - WU NW
a0 Z w V £ % I- 00 0 C 0Zcc z CcccL V)L V.. gF *UJ 00C ow a u * 0oz U L I C W W -- Z_ c 0 t W I. x m U J I 2 Z
C Z w W UJ CL .O w L u 4
0-iQ 0L- 4Dcc Oz < : J * L LIU) C} * <L t lLw aO J1 ) O U _m O C O Z c O : O cc< C -Y h LIlL << - <WZce < w ~- C· U- - LU < , c w zI. W 4W ZZ WZ OWL UO
U N CO Z U 0 - Z Z SIw W W::- <:< OXWtYOW DI. zX jxxL- - a at( a h-I J ccu O izI < Cc W L cO ow W-o <0 IV WU UJ L Wc Z b Ul Z W . IZZ I-V ) *h - U UW L I J- 1 1
- 4C.WUWuOZ ZOcZ- 0 Z Cc W - Y c- Z- U WV I G W U w UJZ O U J 0 I -J OW U ) OU 1 I.J h -c Z4 0 v t OA V z -V- 4S a J C· O c Z Z )ZOZ ce < h- Z 4 U w0t U z- x0
LL ,X C Z LJ Z 0 Czy V) X;- C- I W V-) O < Z Z
-- a Oow c X J OC S OCcc r - ) c c 0 < Z F < O Z W)a WJ C C F W U C a V _ UJ CQ UJ CS w: Wx o : c a G v: Uz ) X : =
- -w O C W - c U 0 0 0 V CK at aZ ct r IL 1 0 Vll E (3 _ C 1L 0 )L; Ct s a:oba: uJ C3 o o aL a I E on 3 c 3z Z L D
O O O a 4
I I I
I,-
z -
I I I
- NU O
I-
-4
UJ
0 4C I Iu.I ]
UUU UU UUUUUU L ) L ) Q L ) L )LUU U U )
213
0000'-4 N
0000.I 44 . .0300
w-4
W
J0430.-
0
b0.
N
0cW
Ctc
xI.-,-
00L)4
9-0J6ccIVCL
UU:r
Ia-Z0-4
w9-
M-400.
QCW0-J
ft
0.,
9-L30-4
0.4
z >-* Cc
J 4* w4 JU. I
4 Io I0
J IZ I-4 I9 Ia I0t Ia I
C-)
I
Z0-4
I-
UL
0 0 0 0 0 0 0 0 0 0 0 C) 0 0 000 o 0 0 Q 0 0 0 0 0rff Mf M+ I 4 b N N 4 Vt W 4n n 4Ln W ' L- U1 ' 0 ' D C c 0 I000000000000000000000000000000O O O O O O O O ) O O O O O O O O O O O , C; O O O O O O O
LU IO~ X u
U. S L t >- o Z · I - O .JW I
UI LI UL U u ZtL ,-Z ' I
t ' ) * C LLJ I ' O: Z D IL-- Z LUL.J J- C[ -- : LLI*, IZ _ : Z Z U U Z - U) IU Z - O 0 0 4 >- 4 : Z C Ir ,,, tu a cr c a a o o I-) - - i ) < C ' U O |I
:2 4 > : 2 I- a 4 4 U C LL IC Z w u4 - U. x u Z. IU0 0-> - · -O 4 F- ZO IU) cDZ Z . UJJ w W Iw .. iW CU 0C0 - W L n u. c .J IwJ Z ; W 4 .- VU I W UJV) VI) z a - Z Z : I
r co .i a D o < >4t > - a t-, Z ZI- = JI- J - . ) V .)-W IU O U - <1-4 Z 4 J UJ < Z -.. IWU ZU) < I1: - Z I - C. I O Z .W C4 I. I
J W =- 0 *- 3 - Z 3 - - t : W Z I0 Z t Q V J ' J + _ I ) a C Z C"- IZ ULL iL. < UJ C U NU W a . .- I
0O - . O CC 4 t WI- - C -- 2 W u I < V) I n* * :rL ~ '-tL<-w 11 <'C , C O z - I fC
Z ^ -3 Z- Ul U G V i- C' U ; . ) O V D U L I -- f , U. ~ IJ .~ E n~ Q .3 C: a - W cu J-- W .. D I
:: 0 - uJ0 . [ -W . N -ZIZ OW I- W4 1 Z Z W_9- $ -
0 _1 0 CC C w C W I1 - LL J 0 QL 1 0W -, . W I Z D WJ O Z - <[ I -.C -FY - _ 1 tYU 0V d Z czLZ Z a ct I oI *,U UJ Oh- I Iu D
4 W o. 3 v )U- I ::
I I I I I I I tL U ) I
Z I- 4 U- I
a '- IC 1 0 I-_: I )"00 I t. /I
Z I ( cc U,
WUOZ I I. .J I zV, 4I 4
. l -0 x Lu cc
0000000Z 1 C C cN,C C O C r I--
XOr-_ Z
'- -.JILJ--r * 0
J -LL O
O :D W-U. cVZ3c·. -L~.) :2
N u0: 0- ,
_- - W..at
.a. -.J C3Z cQ U J:
u * *
ct zS 2 3L
-LU
Z
C
0Ucc
z0
L)LU
Z r0 0.,-,C Wz J
a
0 0-J
Uu[3
*
* .
214
I.
,
tii
i
z
O 0
o 044c .J J
o a 0 0 0 0 0 0 0 0 0 0 0 0 0 C 0 0 0 0 0 0 0 0 ' 0 4 0 0 0 0 0 0 0-j -i - 0' C' '- Nj -i 4 U.' j r ew- -a Ci j 6 6 a a a a a I N a aO OO OQ COa OOO .o cO O % 0 000000 o ~1 Q :QQ~QQ~Q C4C4QQ~QCQdQQ*
LI
.00
,0.- 4
.
$
O *0 O0 O
X)< X
) IV)--z Z ,- ,--..
CU C _
' Z
0-4
I
Slbtl
-4IIIi,.
"0C V-4
CL -4II
II 04 -0 _OCC) C-0L~L
W Lu O
I., -,4N V1 d CC " ,, LI) cr aoC, LUIJ
_ cc0 Z
O -0 * V 'n 11 W 11"") II O..-I- · · 0
- UW OCl - 11 it D U.w - tD J UJ a ZL wU _c uz~ 0 J e...-'" <X n · ·-1 0- O Z ,' [ O..II , I I "" tl0 0 II Z' ~ I0 Z Lt L - J
I II W Z .cY. NO O .O4 4<lI~ tl~ l 4 .- daJ I 11 03 t U "- A-4 L k- I,- a w z
Oi-4
* - Cuu U
uJ
I--
03
cr
O
0
6
I
-
-4
H-)5..
-
0 l
.-
-
Ii.=drCf ) "
C
I-
I-
U
U
I.-
0
00
C 0uJa
6-
-
b-i
0N0
0
-4
-2 )J
-O 0
It LL< 3t '.
U U)
215
0
C300a II
U.. L.)
216
O0 - O ° t O O O O % 0 O O C O C O 0 O O O O C C) 0 -c;M 0 _, 4 . _, _ , ,, N N N N N N N N N r mrC f n m ¢ ¢ ¢ ¢%
. 4 4 _-I 4 - . -I _-4 -4
< < I C C z << d <
. _ J J .J .J _J J J
0-
C:
_ I
YL +
J -; I ^ O.* .. [ V ^ + 4. D U ..C, U 110 a3 oWII J II LI + +.
I;C-~.' ~~1 ..
c44 c - c- - 4 - 4 - - 4 - -a -< e- -c- -c- -4 -4 c- - -4
3 ) 0 0 0 -0 _0 0 0 00 _ _ j = > J1 > J J J1 J J > X J J1 J J J ~J J ~J IX J _ J ~J J
u
l-
tYuw
" 0I-
uuU.J U. --_t-U It
I '-WQF O -
9.
-J
-JI--i
II
V)
S
'-4
4.
'.
-j0z4
U,
0U
V)
!1II
Z'-o,
z.
= cc .
Z IIz l0 0U I a
I-
0
Lr-
0
al-0tc0
- 0
I w- _
D * 0 (12II'D U
0
O C- 0
_ _.4 -~ ^ D .
IA _' _. I I t!
Q V) 0 C 1I<cr r uj it
II 11 _ I t I 3
< <3
0c4
-J
zk-
z00
0 & CIOm M n +
'-I
L U
006
UI 00
v4N.-4
L''4.
000000C'00000000ti0C)00 - N M t UN NO r 0% O0 V" N M 4 UM z r l- W Q UNo0 M n CA ; o o I to O o o o o o %C X - - 1 4 a aD a a o 4 a -4 a a a u fi 4 a -4
< c < c < < < < < < < < < < < < <c-c c a a o a a cj o 2 a c c a a c m G C 11 - , ,, D, ,44444444444444444444<
n MW
co ' 3UU
I I .) -wK _--a W
II 0r - ;S- 11 LU
n I v)I L- -J LV) s -V
~ J33.c -k0
-10
xM
CU
I-
.'
J *
II* M_
le o
U a
) ~= eV), M ':3 v Ii,. 0 '
0 UI Z * *
II ~JJ1W
bn t b
Zr _ 1 IOa r LI Qa ac
.J
0z
IJU
IL,W
c 4
0Ui
' X.4
0z4
U,
C
oI-
c:
I- O ,
0
OC3
e-Z -cI.
a-
0
z
U.-c -
S.
Z 12 -
o beI 11-It-i0 0o o
0000000000- I, -41 U %; - Ol c..F r . t- r P l- rl. p I - W,-4 -4 -4 -4 -- - 4 --- -
.ooccoc co
-I Ji j J-.J----J-J
-
3-i
S.telIII
:X -l Y
, -. 1
w9r
LU x11
, D IIuw-UCl'I I -_
C3
0I,-
0
Z
UJ
,m._- a43 0
* V)
04 I
00%a0
LUzi Z0
. II IIF C0I4
.4 U.L .4 N0 - a 0
UN 0 U uIin %O ' 0o
0-U U U
217
00000CZ O - co O
am " , 1,J .1 .J J
00U% p- co c 0 0069
coU
000000000- N m % - C ' 7
130 G 0 (~ co 0O aO O0 a-4 - -. ,-. -4 4-4 -_ << _ < <
. J c. r c a c . .3 D = = n = n)-J -i j 1 j -J J
O 0 0 0 0 C O0 C. 0 oo 0 o o o o o o a o o o , O- N ' t' ,O C : O O~, O 4 u O 0 0 -4 -
_- -4 -4 -i - -4 -4 - -4 C N N N N V N f ,,, N N N,
a a a a a a a a a a a a a a C.j i a a a a a
Ut( V0
0I-.0C:
I- x t-X
A ·O W* (42
- 1 U.
0 o -o <I - -
0 0O0 41
zLLUJ
LU
a.I1-
a.
,,l
Ln0
-40
,-0
Z- -z t a-"0 U. .
Z-4
J C"..- 1
II
aC3
UO
a.
-3
-)
-
-.
UL.C-
0 OU%. 00a.
to-3oL a
-- ICS U Z0
c 0
Lu W UJ 0OO OZ*++ZZ C*Q.<<A- . I _: 0*0.*<*
Z0Lo
Otut0 U
U U . U
0
II0I`-
ura1 C
0 0 o-4 0 0- 0 0cF% av
UU
218
.4-J
0co
3:
I-
r4
oc0cr-
LU
7'.4
0.
0
-JI'0
a:
U z
J:L Z
U CC j
· C i UJ_ UJ i 3
II -I! II -
a' ZU -4 N 0-LICC;..
P,"In"
00 000 00 o oo oo o o o %· L O fI C 0 - N M' .t U % Or- a, 0 N M 400 000 00 00 0000000 00 0L L IL. I. l. .L U. L UI. U.L IL U L L. UA U i LL. LLI L. L I J . U J L LI W i W U U J LW W Li UJ L ' lJ
33333333333~33 4A
4U.LU 0-c LuJ 3 Cz O.I4 *4< I -J J 4O V3 Z Wr O:
9D Y 2 0 Z oII 014004!
0 O
4c _ uj . LU L -o I a: z w-F W cC ZZ X 3~ Q 3w 4wJ W1
J 4mJ W.- -.JzW t4o .Im-n 0{/) i_ 3 _c Z ~J Z
)Q O cta 0 4 C
Cr W < I o)- C O Ci z i
0* V- 3t= I Z Y- obuj :)
Z C LL -
0- J j 4W4 ~~ I CI V) D I x w4 m v <-W JZ-J-JWO~W0O
ZJ < L UUI I I
0z4a
::I0
IO-2
0I-z0
LUzI-0
cL OCv0
Uct9- i..OwVI uJZ >
.J O.-
Ill
II-
zZ LU W Z
z= uj
0 ZZ UJix
Uz 0Q 0ZcoW l 0
l-C C w
oW- 2:4OI U.=1 vLL - W d a ) LL
CY D z _
w iO cc U. UD WIW ;- 2! 3
IC 0C4Z at
CUJ C ZO co U Q ie u-W W U o
Wz OI ( 0- 0'/Z
-C-co b
4Z -
cr
o000.
0z-J-J
U4s
I..
0
0ILL,
-1
II
III
I
oU.Ll
.. I
w I x
Z I I
uj uj < I UJ-.. 0 I * ZZ Q I _ *Z I S
C;) 1.1.o I)~
Y - JtS,. J C.L Wz I I-
I -I I . _Lu
I 4
I LI Z
Z L 9-
-V9 0
O4 t Jo~~~~~~~~~-J-
QUUu U U UUwUUUUoUUUo U
0000000000O O) O O C O O O O 0000000000U ILi U. I UI L iL LL JiD Z - D ) : D :-JX -.-J .. J J J .J
a-
-4.1-4I-b-
Le
bIAJ
-J.- 0
a0D_ 0- O
_ L
m )
c
):0
Uco
IL
Z .
II
IIAO-
0 xUN
0*
II
0 0
II 40
.OVu
219
0 o_ N0 0U. .w Ul
DD -
x4-r
Ou.00Li
- 4* 4
4P 9l
co II
=C IW I
ILU I
. I9 I
. I,IuJ ILU IM I
LU
-JULIW
ZC
U.Li
IuOf
U0 <:D w
CL
I _
Q U
220
000000000000000Q000000are O a r- N r o A O c a, 0 0 c rf m LAM o 0, CD7c e '4%tt t r % 4 S I ¢ N L U > u n U* lr -N00000000o00000000o00 Ooow U IL U LU U. U. w U. L U. I L L U- U. U L U. L U.IWJ LU LJ LL' LL ,L W UJ . LU W l W , LLI W U lW J , U- L LI J W
X
0UL
II
D
0
n 3 _-4
0I-
0
- a -C
II , .I 3
D IL. P" ) 0 w
_ 3
-.1...
I
0 X
0 0I C
Lu
0
LUz0
Co
Z D,O ) V)z _DOuLL
x0 UlO-4 -4 C
Z I.- - _.4 ".-
I + _ 11- - X -1 + cZ 11 i11 U_t -
O . LLO0
O
0 -,0-z X
ZZ I
* 0
I11
I:ID,
-
II Z
-w Z- 1- w- IJ ZX C I~
Li
z
z0U
LAOUN
x
t^
221
000000000000I U b 0 \ C cr o 1 N n + u90 0 0 0 0 0 -- ,d --4-000000000000w c C c Q! CL L o eC C W LUW LU W LU UW LU LU LU LW W U1 :33 3 3 3 5 3 3
Z0-4
,iWLU4
UJ *
W0-
(I VI
a:
3acO JCZ <u U
I I
wZ
cc J
OUJ0WI-
X _ c
cc LL -J0~0 < Z
- -b_ I ---EZZ> V)
WLUL I ) UJ
0. c zI CV>-. t (I 3 .- 44
W U LU LU <i -3: Z IZ Z Z I- x=: Z Z> 0
LUC ,X OZ WWW'- U U 4 UJ
L O U CL t It I , D - I o
I z M U C ):L ch ZL U
LUwii
z
C
L.I!
UJ
z
Z UJO Wz wu-4 LU
U LI <Z4 a
LU
044
Z.J
G m ) r OOOOOOO f C OOO OOOOO N 0 _c _ -4 N N N N N N N N N , e n fCI ,M Mo 00 00000 00 0 0 00 0 o 0 0 00C C C - )- - - C - C t I C - - - - -
LL LU W UJ L UU LU LU LL LU L LD W LU W L U U UJ
W-
63
-4
.. * -
4- -,3 -x
X U.w
0
z4a0t
ZLi
U
I.-V)cc
LUI1-4-j
U)LUb-
ZI--CCIz
c3CD
t-
CD
w
1-4>-2c-O
Nz -
Z -_ c
£ -ct Z
3
Z < -C= -
LUU
ZNu Z*LUi
IA LU W4
UJWL U-3: c3 C3 :)
0WWw=r t 0a O
I * * * -.
1 l3:
L- C 0
X. .- Z ·_ -o QN* 0b - _ZC Z O N C33 3 Z fsO C, - -- 0
C 3
_ uJC 3- C3 *-0 WN
0 ' c a
0LU7.
U-
LI0
%4
II O
1c4--40
0
0
00
I-CK
uJ
w4
U.II
UI
U U L
000000000~- I-
I
IIIII~
IIIIIIIiIIII
U -
4 >.z cc
LU
I#-- Ic IIUJ I
LU iv IIw I
Io IZ Icc IO I
U) la:3-
I
I
222
O o o o O o o o o 0 ) 00o" aC 0 o o- N 4 + .r F'- CO (7 O N e
00000000000000000f- C a, N en P, M a t-')-
W UJ L W UJ UJ U LL'W U LUL
W N .r
tl 0 I 0 ._x _ 0 *Z Z Z 3: - -
Z Z Z Z-
Q 3 3 UL tS
0 4, u w -Vf U
4e _4 O-$
m 3 N N_
* C _
* I *- *
l0'- 11 0 11 L U W A
: CC or cc CC a: cU OL W O W u 0 U Z
LI a " 3 U. el _ U
O U% 0 UJ N
223
A.5 Sample Problem
Table A.5.1 lists the input' data for a sample problem,
and Table A.5.2 displays the output from the DATANA code.
o U)
o 0
O 0 U( _ _V
C- 4 L O f r
. M 4 N ¢%lor-
CO) oC o0 Cv
0 a O C) " ur 0 CN ) o 0) o0 o C
o00000
00O
O 0OO
o o 0D X. ,C P 3 N r V' r r o (C4 _ M Ln %.C _ *- %C - D V- o D U m
r u N O O 0000000000000000
C O , O 0 0 C C> OO C' 0O CO)
(N r-
- coO O0O
0000 00000000C1 ;) f- Nun r -Ln m U') 00000000
00 ccT 0o (Xt 0¢ O
O O )OC OOO "hO 1 ) o , O' 'ODo r' m 0crO m r Ln
0 C ( N - r- 00 0 00 00 C 0 C)I r rL, Ln U.) U'O O O O
! C ! iV m C r- ,O O.r ON C ON Vr r 'IC N m%C t -T N _ -o-
000000000o~~~~~)a0 >00C
C) 0 OO O00 C00 C
ooo0000000000000000000000000000o fio w %D N 4 n u n N- ru r C " D C r- %o f- n%D V)3 :t m N r O o o O00000000000000000000000000000 00 000 0000O r
0 00 oo oO O
o oo om " - " n .VI Lf) U: U) ) U.OOOOC>COO>
r Ln C N D N O In, Ol CO O4 _- O O
O~ 1- :: r u' ,o uO O ULn1- eC4 L> O' 0 0C
000000000000000000o ^ o a N £ N 'C>
%D en I
000O C> O0 00OCO Oo o
0 00OOCOOo o 0
'" NO OO OO OO O
C N a, C7
000 C
00000o o oOoo O ooo" ( "N r - " (N - ( N f r-0 00D 0- t0 0 0 0 O o o o o "
Un Ln L U' )n n n Ln o %D %O %C %O %C0 >ooDo o oo oao o ooo
rI
*O
O-00
IC
a:. no
0fr4
EaA0
C4uVC
fic
(L Ir- O
0
C( n(n N0 -4"4
S0V-V1Y(N
0I04"
40'rY
U)co
Ln.-
(N
0-Icrtr.C9
225
o 0 "o
0
ILnV-
--r 0 000000000000000000000oo000000
* C4 T- W 9- f0 000000000 0 Co oO O oOtOOOOOO
O 000000000O00000000000000000000000000O000OCaOOC 00000
0a*t
O fn Ln nm
I n OD 0 0
o o o o0000000000,O t' O0 0O OO OO O
0 oN
O O 0oOO00O000000000000000-- v- -00000000000000000000000000000000000000
' -t Ln ,"
0000000000000 O O a
0000 C00000000O OOO
O- OOO000
-f-O OO OO OO O0 0
' - O O O0000000000000000000
0 00000O C O
o00000000000000lf) 1
0 0
oo o
0O0
O O0O O0O C
O0N Ln00 0 Oo 0O 0O 0
O OO OO OO O
O O0 OO O0O O00O0
-O OO OO OO o0 O O
O OO O OO O
000000 O - C00000O00O O OO OO O O oO OO
,'- ,I (Ni
OO00 0000
00000 00O00
AOmO OO OO OO OO O0 CO O
O O00o
O -O OO
OO o
O OO OO OO OO O
o oo oo oo o0 o
r- q0,- OO000000000000
000000O O a m OO O O OC000000000000000000000000
.- N ".- N- o O "maOOOO C a Oaoo000000000000000000O C, O O O O a O000000000COO>0a " )o oOO
Co 0* U)
t0
I
) 0OLA 0 0rq r C 5 " r " l f " - " - " CIO'," as ~ ' 0 C> ~ ~N I~. " q M' ~ t,% 1' ¢Li !~ ,CI !"
%0%OOO OO O r O r V C = o ooD > a 0 _l a QJ a a C> 3 0 a aDO Ct O O O O O O O O O O O3 O O O
m 1 M _TJ- aGc %0 D N.1 a U c 01
O aO C O
O O cr OLr, 4 o 0CN
O co fl- D n
I O oO0000000003 UZ rr O
o o OO O r MO O " C 4O C OO O
00000
- - -000000000000U) T-
O O OO OO O
O 00 o0O0O
¢M m
O OU)
O O
" r- r e r " IC-3 _-r n n o r-' O D o o %U %D %0 %00000C)ao0
,- I~ O, I , r
rn %C L M n 0_m r F N l- D
ooooooOOOOO
tOO Ln - O %O
r - n W- - -o 00 a 0000 a 000000co - r OCOO0000OCCo
rv- M Vo "N t L W Y m %Q (": 0 0 D O 0 LN in S J w - ( fYn N ( N0000 000 00000000000000 0 0 a O O 0 a 00 aO
%O r- uO v .~
00000000Oo oo o
U)( r- -
0O >000
0rr
O 0
I A MMw Nt- 0o - - a%O 0 n r- N
0000
00000o4 o o o o~o O m t r~t : t)@ o Z _
o cooo
O\ cIr 08 F In Q uno o
0o \o D _t NV m ur : t r(NO¢ --- ~f-~v0N o 0 q - m N o - v- 000000000000cO o.,' O O O" Or¢Z O ,C m,m ¢2 O ~d O, O ai ¢D O q," O00000000000O X s u :tO O O O O m O
NG a 0\ ej~C 0 QCV ~ r0rQr0t-C000000000000000000000
0 r-0 Coa 0 0
0 0 o0 O0
r P a > % - a C m t~ t n kVo :-r q- C :O'3n 1Ln U) -- :T m n (N N N "N (No0000000000000000000000000000000000000
COl "T 0 m ar m 0I- T r- iur a % M w0 r %O n Ln Un t n m n 3 V 4 N N OCO OOOOOOOoOOOOO00000 C O O0 C0000000 O O O O OO O OOO OO O OOOO
0o ; (4 (N 7-0 t - ICa 0) oY r- N m ) OC 0 -a ) en w L 0w 0 i M - C - rj w (N - N _v- OD I C V-m cr a C Cn N 0 >O 0 C = CP0 Co r o 0 ut a : > 0) 0 0
00000000 0000 000 O O 0000Cx)O O °O O OO O O OC O O O OO OO°O O O
C I 000000000O
000000000 C000 000
ON (n O Oc m -00000000 a 000000
a M D n 0 t" rl a O .' 0 04 (N O r r J tN _0000000000o00000
m - u' r :- 0 ~D O O0 " Co 0ov- M r- W v c r Lr N %C :? n Ln t ra 0 r- I O 0 -I - m m n-OO C- 0000 00 00OO O OOOOOOOO000
Caa C O0000O00000O> OO C) O O a O~ O a O c:: 0 Q O
(NN (N CN co L m " CN
OC1 O O 00000_ i %D * O V- _D V- W V- \0 %D V % C - to V-D o ) V- rCc - %D - D r 0 a cON CS a a o o r_ N fn M -t -f > Ln % D %O r r 0 W M a> a Co o- r- " " M :r -r un un
N " " " m m m x m m m m m) m rn m )n m m m t m m :m m :t t* j r -- J - t E T0OO 00 (OOOOOOOOO0COO O O O aOOn(Nen OOO
226
) c0 eO . i
co m oO CD :S
000OCO O
'-000 00
227
C- t _C C C O CD 0D
0000000000
0000000000O O O o O O o O O O
-c- o o o V O o0 O
0000000000a) do QD f) UW (N t4 ao fa: 000000000000000000000000000000
..-I - O ,,--,," 0O 0'~ 0
00000000000e,. uO q,.-. ,,O r, %- v' tz --
000l a00 0 00a000
228
* l
Ln
¢Oco
ca-o
0
0^ 0 0n 00
a* " I"n -
Mh io UULQUUVMM( -IO mr~uU
*;wP N 'IPar C 00 0 oU
C o _Cr r -_ o ra j C J ¢ O _ C, _ * C O W i O C', o O
OOC0000000000000000000000
00000 0n000000000o0000000000>00000 00000
4_ * * .4 .4 - 4 * 4- 4 4 4 4-4r 4 o 4- 4 4 4 4
O I C I A '0 . ~ J ' I . N .c ' C 00 0 0 0 0 0 o0 Co0000000000 000O0OOc 000000000
00000000000C0 o0o0c0 0 0000omr i'03 P 1 P. ~ ~0~r 3OIA 0 000 ~0 0
o o o o o o N o o o o o o o o o o o N a o N C,. W , 1 o 0
00000000000000000000000 0000000
" _ O P P P - _ N N N N N _-N N 0 0 Co a cc .a _ _ _- fn 44 %0 O -
0 00 0000 000 000 000 06000 0000 0000 000
F K" " M I M a M 01 K K U -K K K - - - - - K K K K Uov-o0 oe0'00IAN0'o *0 N 0 00 00 00 0000000
O M I- . a 0 N 1- IA _ I A N O O 2, C0 0 0 t 0000000000000000000000000000000000*p1 NN N N NNN _ _ _ _ _ _0_ 00_ 0000000000000C000000000000000 00
0000 00 00 000000 000 0 00 000 00 00 000 0
. @ @ . @ @ @ . . @ e e * . e @ . @ . . . .
W K14 K W K DI W 14 U W: K Wi 52 hi13 K i K h W O S. W K hI W K U KP. D N ' 0 N . 0 0 Q 0 W a 0 0 O O O O 0 00
: e Mn 0 4 1A C 0 n rP'- WI NA.IA 0 ' o . N 0 03 .N C 01 4 P 1 P 1 0 ' .0 P 1 N P 1 1 Io o. 0 o o o
00000000000000000000000000000000
O 00000000000000000000000000000 0U+**~~~**~~***~~~ooo~oo§~ o
C P or ·o·ww~r ro rnwl ~~NwaDw~I~wu ~ p· IC l ,h ~ J3O~ O) L'C ' I >OOE
=10§14
C)aA,4
4
o140ri-4r0
QC C) __. Coc _ CO. C_
C, 1 a C; C1 cv CP P c c P - P1P1P10oCooooooooooooooo
.- - - - - - - - s 0 COO 000000000000044444444444444444
crl-CU);McYM. n MOK K K ' K ' I K K1 - i ' M 'a M-mO~h O _0s S~r Lr~otJ C,
W- - -_ -- 1^t.: tO ZOOtOOOOO~voOtO~o.. * * * * 0 ". 0 +
2; W ea W; W: Di: W . ad U 3 U 043.0 0 N 3 o 0 a IA.0 0 W o 1 68
la _ 10 'I o _: la M C, IC 0 n O W f __ In _ ' NN N - PI P13 *O r.
*~~~~1 . . * .000 0 00 ; ;C 0 a 0 00 00 C 0 00 0
00C 00,000000000000444444444444444*4P. C . - P . . .P o Q. N IA P 0 P: ! A 0 O
fD- _ -N Nn P s Om v% W V-0e
00000000000000000,, . - " C', .- 1 P1 P P -
O: . _ _ - _ _~ 1 ~__ 000OOO000000000044444444444444444
U U K U Si S i K U M U K U U UOP. 4. 0 P. P1 P . P1OP P1^O0 P. P10 P.C' _0 _1. _O _i N N1 .0 P10.0 P1w0'
*0 .1. Oc S .CO N . P. N .d 00 [email protected]
0000000 C 0000000044444444444444444
U i U U b" hi K I U U K K; K K K M MO z N1 -: On * <- w@ ed eF
P . O . .1 * P. P 1 P. P 1 . P 1 . P .00000000000000000
0 0 - -0 0 0 0 0 _ _0 0 0 0 0
K U K K X U K K a U - * K U : K U U h
P P P1 _ -_ N N N N 1 1 P1 P1 O ne IA*00000000000000000 °0°OrC~C)OOOCOOOCC00000000000000000
44444444444444444U K i K K 1 h KU K iU K U M hO P . U P .~ C ' 0 P 1 0 P 1 0 P 0 P
0 00 000000 00000 00 0
O P 1 1 P P 1 1 P P 11M aPrPlc cC'. O- ______- 0000000000000000044444444444444444P. P 1 . P W 0 P. 1 P. P. P 1 P P1tM u W 0 0 %1 0 0 P1 0 .0 P 0 PUW0WWY~W T roI IoI
.0 P1 N 3 P O t I 00 P.00 _ @100b'.00"-c c-NNN P1P1P PI3@3
00000000000000000
0000000000000000844444444444444444
Si ml K4 hI : K i K U i hi h K K US
0 .1 .' . 1 P. P1 P *. P1 P. P10e6o .'o o o1 - o o o o ocP1 1
00000000000000000
229
000000000000000
I I I. I M C-- - O C 00 C 0 0l 0 0 t 0 0O M
C-0 -0 - P0 - P0 - 0 C-0
InOIA ID 0 C- OC- C- 0 0
M1C) -0 P1 C- P ~C C P 0 M -C 0 IL 00 0 P10 I
0 0 0 0 0 0 0 0 0 0 0 0 0 0000000000000000
CC- P10 C P 0 - P 0 C- P0 OC000000000000000
000000000000000
n V. 0 . W U- Io D 0' a - 0,
I I In 0 - C C- C- 00 a'00000000000000000000000000000
CC - - CC- P C P1 C- P1 -
N In - P1 In 0 '- P 1 t C -N a O000000000000000N C- O C*o 0 N n 0 v0 a OOO000000000000000000000000000000
000000000000000
* ' ' ' uUUKUK2Ub2§ o
_ __ __ __ __ 0O o CfO Oo O O o O C + * - + - . .-
0 U0 P 0 000 -S XS "-
la
UI-IDI-
M
0
N vMl
UP1 P P-. PIN N N ,N N C N -.IC ,-0N
0000 -0000000000 0000 * 0 W s0 el McU 0 c 1 0 N Pl . U .- C C
. d u _ r N l: . ~ In 0-7 a a o' i v 0 1 w
C O 0 0 0 0 *0 0 0 0 0 0 .0 0 0 0 0
0000000000000000000000
C, Ca o CO'0 '000' 0- OP_1 0 O N 0 a * C CW [ N U a C-D a *_ Qi w a l0000000000000000000000
*.* . *- 44 . . * 4 . e * . 4 v% sr.O' N C O O 0O a 0' , 0 WI 9 O C' O O In In N N* U P . u1 P 1 P . P1 P 1 N : N N N N N N £C v 0 0 CO_ N a a a a In 0 . l oCy 0 0 0 .-'O C ,0.
* 2 oP P P . -- 1 .N N N 00-X C -o
o00000 Cc0000000000000
0000000000000000000000
444*444*.*444***444**4
O 0OOO OOOIn 10 C. NP 'C-O0
00
W1000
1i0
MP10
1.a4A
W
I0P1
14
aN
0iF<;
0
C-
,IM.4
0
.371DI3L3aU
.4Sa.3UaU.3U
I-
vI
Ur
00"I0
In.3aVI
00
I"1..
0 - .a - '- o * El 9 . . . .. 0 N 0 00.. .
N' OD e0' oI P1 N' : 0.0 a P1 - 'Nt r It a P1 N'-W. W
0000000000000000000000
C 0 10 C -. . a * N C C 0 a C- N .0 0
0 000 000 0 00 0 00 0 _ OO~~~~ M. M hC
230
a
0000o 00000000 000 C o o
N - 0 P F "I F . U r r N P 0N U U I 1 3 V 3 o D I O ' C 3 D 3D F - F - F Fr - F
00080OOO00000000000880
f t N 4 N N N N N N N NN N N Nt N N N ft N000000C0000000000000 C
WP all . N: Wo 0 W o o W la W W W W Noo W Na0 W-f n - iN 2 F- N n F- a, M In bl U U F-
000000000000000000000
N N N N N N N N N N N N N N N N N N N N N N0000000000000000800000444+4*-+*-+*4++*44**
0 000C 1 00 a 000 a a @
aE 2 P2_ -UWU2In C. tC3 - - - -
oo*VIoo W.oo a ,;:.000000000000000000000
0080000000000000080000
" WI 213 U .3 1 U N P 20 S P 2 2 1 2 0 U LIM IO P. e O C _ F 0 N- U ' O '- O2 ' 0 ON U 1t F_
000000000 00000 000000 00
o oCoooooooooo
N ' J 0 V P. C" VI U N IC. FN- C_ 0 N-D W O20 F < -3D , F r- O U Dn o0 - U 3D 0 0 N O 3D 00 N _ 0 13 F-
0000000000000000000000
OOOO0000800008O000OOO8
0 0 8 8 0 8 0 0 0 0 0 0 0 0 0
1.4 N NNNN NNNN NNNNNNNNN N Na 8.000oo00o0o00000o000
1 W 2 :0 P .0 0 0 F - U O WI 3 C N 2 3 D U P . 0 . 1 3 P .24 wn~o~000 ~ rn08880one8 00n e eam033unnenovo^o^D brra . << <* * @ ................. | -
OCOOO COOOC000000000004 4 4 - - - 4444* - C4 44 a * I a - C - W-
- ¢ P _3 U1 C C N _ F-0 O a- C P. W 0 C 0atonrrb" r~oo §~oo aOr oo. N . * . . . . . . P. P * *
0080000000000000000000
N N N N N _--_-_-O 0 O O
800000 0000C00000000000
21, o - - - - O - 2 0 3D P2 N - 0 P N'- _ '-F- C C N - ' - 0 3800000000000000000080
121 1 .2 2 1 0 1 a, lo 1 m 121.2 b 1 m O 1 UU
NC N 0 . N 0 N P2 ' 2m 3 F -FO D A ' N _ 0 0 - 0 02 P2 C 0e ' lo N di0000000000000000000800NeN NN N'- - -D 000O o O --
0 2 1 3 3 1 3 3 3D 0 U P0 3D 0r o N 213
~NNNNNN_ __OP P O 0OOOO-W
NNNNNNPNNP.PP-P._OOO0OOOOO--
000000000CCOO>OOOOOOO80*4444444444444444444g
1.20U
2210.
2312
NifNtNNNNt P.P.P.P P ;000 088000 '-
F N 5 N _----O O O Om O
0 8 0 0 0*0 0 0 0 . -.JJ~JJ~~jJ~JjJojjjJ~j~!~1~~3N0~0~0
231
O J O - - - _ P- P 1 P 1 P*. P 1 PP- 1 - ' P1 P N N N4 N NP0000000000000C., CC C00000000000
hi h ha hi hi hi ha hI hi h o hi hi h a h a : h hC W -e i - a C _ a C, C hU C CI - fU eV U. a N C C -h NO IC hi C0 1 U , I C' C. GY i 0 0
V P gV N *- _ v- _ c. hi IC, aD I U P rM N _ N _ , . . N P .0 I
C00000000 000000 00000000000000
000000000000COOOO o000 000000 0000
'r a~. C r o ", :r ft: W h.W , WCOOO C Ci N C0 D N 01 aD C r- r N O 10 U P1 N ON --D NO N , . N
P 1 v - v - U h i U ~ P 1 v N O A N N M O a 0 0a , r - a 1 0 a
00000000000000000000000000000000
AW A NWm A Oa .3 N N N 00 N 0 N 0 N 0 0 0 0W oooMN §v hi U uP;1 N ov- v -U a ' r - N
oobpjooooooooo~n~~o~ooooooooo
00000000000000000000000000 0 OO
* N . 0 0 p 1 N h + C U U h
_ MN A W 57 I N N ..U hi _ aD 0 -W D C C 0 W Wi W W 0 C W W Oi W hi
· - -....h3 U. W I C N N i 1 W N 0 0 IU W U hi W 0 0 C. W W C IC 0 0 0 W W W
Co a. vir , P1 0 D C, o o U G I t h N OC P a N ' I 1 3 N C h N O 0M fNv - -Wa N N __rrr 0Nqab
b o J _ o I W n n r cr, n ru e* cu N
o at a U a a, a D o1 M P' o 3 P 1 P14 P1 N N N N Nh e a h . h h h . h . h I h h
00000000000000000o000000000o0o. .. ~ . .~+~~*~~~~~*~*yU~UUWUUUWUY~WWU ~~UUQQ~~~r~0~S~OC~Nr~0=0
a a3 3 : .3 ; PM M mM P M M 1 M P- MM PMM N4 N. . N. N
00000000000000000000000000
hO 0i h h ha ha ha i a h C D h h h a C a a a h a hi ha o h . h h
v - N P _ 1 'n 10 0 I N O N N I N C , ' IA OC Do N _- r N N U N hi N o a0 1 .O N N"0I _ a.-_-v a Mr fN CN NN_ ___0.D 000000 OOa000000000000000000OOOon
0I hihih alIP ihiho h h: hh2 # _ C , h i CD, _ e h .N
u 000%oNN1000C,40 WCa-4NC),l00a0vNOOOOO
0 v-C; tN 0 NOD 0 InO 1 0 1 . W 'N 1-a -O N IL ) -Ov -P 10 a,10
.O a C o O O a o o C) O o o o O O o O C O hi OOO~NO~tOtbOOOOOOQOOOOOOOOOOOOOOhi~r~O~n~f~r~~~o ~~
=0.4I-I=
el54f..4
0
hia
0lo
oo000COOOOCOOOOOO
o, Q M I - - - Mo P1* PIr n OP o l e r F c L M. O Cc o rP1 0 v- IC U NI, N O N I f, 'I 0 P110 W 0i _ -_-v-N N , MM M o Dao
00000000000000000
ov-oP I v U 0 No o 0 P o o
. .v--. - - - - . .- -, -M a 4W "
00000000000000000
0-N -NM f, MO IONMOn
o- - - --o0 c 10 hi o- 1 0 " I NM O N IA 0 0 0 M a v- a1 I M v - N- 0 N IA N O
01O N IA NO P1 *I h v-P1_ 0 v-o e 0000000000000000000000000000 ' 00000
. . * @ .r . e .
o o o o O ooO CF o oo o
o oes o o csc oo oo oo . . - * . * * - -
X b.1 -- K- -W ' N " -
232
000C o0000 C 000
ON ai U b O N P ' ON a- ON P0 'C Pb 0 ' -a 0 'C a-. C r e o rO a @ OC Cs . 1 O t
a , Oh N a N ONs ul 0 0 van an an *G IC 'C N N N Na 0 00 o000000000000000
___ew_________ev- a v-a- a a- --- va-v a-a-W
000000000000000* **4++ * 9* ***** * 4 N Pb ON P : N b ON Pb O Pb 0tW to~aofOrfl
C, re z. r *s**
O c, t. 5 " " ""M n P an an an'C' N000O
1 0 C D > C, 0 0 0a a . . v a- a- a- a- a- 000000000000000
' 0000000000 000-orovoa - a--
In W 0 .%O W' ' S1 tM a000000000000000
0 ' Pb0 Pb0 ' Pb0 'C Pb 0 'C P
94 II * am 4, W 1***
---------------Pb 'C 0- P 'C Oh 0- J N O N a N O00000000000000
---------------Nb _C ___-, ON O b
a- a- a- a- a- a -a -a------ a-000000000000000000000000000000
Pb 0 'C Pb0' b0' b0 'C P 0 '000000000000000- **- a-----a a---v-v-v+-000000000000000hi hi hi hi hi hi hi W hi i hi i hi hiomcmogonor rmornoO Pb O N Pb ON Pb ON Pb ON Pb0 ac Pb 0 'C r Pb C ' 0 '
N a NON an 00 Pb 'O a- a 'C Oan n 4''C C C C NN N0000000000O000000000
a -v--aa--aa-- a - a- a- - a-
hihihihihihih hi o ob h h h h hN Pb C N Pb 0 N Pb 0 N Pb 0 ND rt 0'C Pb 0 'C Pb 0 'C Pb ' b 0 'C Pb 0
* N Oh N an NO Pb 'C a-P 'C Oh 0000000000 U 00
a0
hI
N
aI
Ift0
0
av-
.9
i
0
0
a
I
i
aNOh
0aGAa
09
Air.a0
I0I
Ma
W
a1*
a
0 4
OSnacov)
I-a04.4a,Ul r
o 0 o 0 0 C 0 0 Ia W , t. 0 . C' 00 0 a 0 0 W a
N O C OO O O C' C *' OC O O OC O OC O 0 G, aC O O O O**~~* *+*4-+-*-****4*,*
b Oh a- V O C Pb Q- I 00 N 'C 0 I * a 'C a a Oh0
* a O a . = . IC P . . . a-i .b .b P P- . P0000000000000a000000000000
- aO a a 0 N N D a C N - 0 P ' a P N ' ' Sa C
O a a N I an - o O O O O C O N O O O O OC, N n a rP Om N , W l b oC a 3 0 a
a a aa aarIn P a b Pben Pb P Pb Pb Pb P P P m pb0000000 c000 o000o000000000Vo 'C O nCN NO NaOO ONObo
0 O00000000 CO0000000000000
0000000000000000000000000
N co P Oh Oh h0 O r N C ' D W - a Oh P 0 ra O C - N tN , P v-N o a o 1-a an Oh*Pb 0v- o N O o oa a' ' N a N b o a- N. -v- aC n O O e bv-00000000000000000000000000
~*+ * + ~*~* * + + * * * * + * * * * -U : MW S sh; bW S ; ;d N W
an0o
hian
Pb
0
m0
hi
IY01'CN
Cb
r-C4.nt0
N0e
hi
ND
!
an
!h
N0Pb
ahi
In
C0
Ca.ahi
uaa-an
adV,
O O N O Oh O Oh O - NO N N- Oh . a a N ' N a OPbaN r anN a a ' C o N W a C -v a N N N a-I O : rh oPb
F Iiri > _ _ e
N 0N O C 0 0 NC N P 0 0 0 OalV N00* an ^E' -v-v-v-ON ANNJ v-a-a-a-
- 00000000000000000000000000
233
0
N N N N N N N N Nw 9-
a N N N r r N
U) N aC S ao oM N N C - U C Sr
11 I. . . .
N N N N N N N N 9 - 90000000000044444444444
Sr Sr 'C C O oSr C o O
- a C, C, U -v w w 6o w v
U ) N 3 ' C Sr W a S r N 9 - 0 S r W00000000000
N, N N N Nn N N- N w-000000000000
4 4 4 44 4 4 44M"""r*n"Nr"rs
9- a Sr N MO aO U) 'C 9Sr40. N U) N C U .U9- - U) N e w ~ o v ia a r O ' O rU) N 'C S a Sr N N _ ¢ a S000000000008
N N N N N N N N N N - C),000000000000
0 UD .0 _N 00 C' U) SrO9 - 8 0 0 U) C C r 0r OU) rO N . 'C ) 0 M .0 O N 080000000000
N t'4 N N N N1 N N N N -0OO~rOOOOOOOOO0000000080004444444444..
hihi*i*ihi*i*ihihih*hih-;"S"S""MU""S0 4 U) N Sr 9 - 0 . 0N ' C 0of ~~~rc-a*o0 U) N 1 N N _w C N s 9 0O N ~ ~ ~ ~ i O S r S r N 9 9 S
800000000000
N N N N NW N N N N N _ _-000000000000444444444444
rOOOOOCOOOQOOO
U. hi hi i hi i hi i hi i hi hi
CON Co C ieO N r Sr oNN O U)0 .0 0 N r U N' S
008000000000O|:^D0 X t__¢* ¢ @ .@ @@ @ @
Ooo~chooooooo
N NM N N " N N MN- 9-000000000000
0 :f e, W- P, e. - - 0,04.444444444*C W)NNNN0aN9..
0 ) SrS C.' MN , -- N
W - To 888000000000N N iNNNN N N Nw-000000000000
0 . - N S . .0 N Srao0OO0ooCCOo(Yl(hNCYrNNSrr808080088000~~+ + *+ + ~+ UW ULb~U -K U o ov r N n v;o W ~D r w o0a --
N -C-" F o o ~O~bset tr*N_ 0* e| @ · .@ §@
000000040000
O C O C 0000 C 000CCC00 00 00 O 4*4*44**44-*4+444*-.*4*-4*44444*+-*4
I. b 0. 0. W WI 0 u b. f. L W r: Sr b C * NC . W 0 . W f' I ; U W W 1 wW- r W0 v. ) r -*Ia Co9-r C .er c a A co C*-o r Irr C C Vr. r oN Sr Sr W. a : V O Lf' r- e0 r r- N IC I a e u . o C U 1 . ) IV N 0 i
0 o O C O O O CO 0 00 a C O C 0 000
M, U) _- - U) 'W N IC " r _r Sr a Sr NO a C,4 0 N Sr M C '7v o w 0 " , in
0 C C, C, 0 0 C a c C0000 0 0 r _ a 0 -0 a 00 00 00 00 0 0 0
eJ N N N N N N V4N N N " N N # N N M N N N N N .N p, 6s N N N N
000000000~000C-0000000C00,0A00000O U 0 N O O C O af CU 0 O CJ N 0 NO N OC 0 O a N O O 0 O O O O N* 'C Sr N11) N a C C U) 0 N N N ON 7 N U) *" -1 N N " .Ut 0 N : N SP UM U ) - -
VN S rN O a N M U' N C C O C N C N S N ) C 0M C N r C G N %fJ~r,7rM _ v 'C~ne~z~sN wCU)Ur)UFfr U)U)O 000000s 5IC e cr C
N N N N N N N N N N N N N N N N N N Nt Nl NS N N N N N N N N Ns N1 N N N N N N t
O C a N a a a) C O O 0 0 a a 0 N N OOO C O O C 0 NO N O
NON i S ' N ON U a ;: N 08' N SC C '7 N U 0N r a a a. a v.o r N C. U) U)m U) t- U) U) 00000000
N N N N N N N N N N N N N N N N N N N N' N N N N N N N NN N N.N N N N
W la hi W hi i W W h. h h7 W U h li W I i hi . W ha U r WU
C N 'll "I00 a C 'C U O C W U ON r. if, N 0 0 - a * N 00 Sr N Sr N O 0NM Sr SCC C Sr SC :>.00 0 N N c o) oU- UUn)0 0 0
0000000000000000000000D 0000000000 00. . * @ . * * . * . @ @ @ * . * @ @ * * * . @ % @ @
1o0E.
an
4
hiEo
N N N N N NN NN N N NN o "
N""
N N N N N N N N N N N N N N N N NOO 0000C 0000 0000000000000 00 0 C 000000
. - h. hh, hh hi h h hi C h h 4- h ih h , i hi N " ' a WN I ' 'aNC'CN .C 0r CN W o9 -, W oU _a,i . .k . 4, . . . . . . . . 4 . . . . . . . . . . . . .. + . ., * N N C C C O O S S O O 0.0C0.v.0.0 N N N N N O U O a C O O O) 0 00 0 00
(' N N N N N N N N N N N N NrN N N- N N N N N N N N " NN N N N N N"0 O C 0000000 0000000 O 00000 000000000 0 0 0 0 0
N: NM Sr Ma C, S Sr r W - ' . . - ' N_ N N N N ) ) v U U U 0 0 0
234
W01el
I.U
Go
as
hi0~
235
ieelNNNNsiNNe4Ns _______-_ __-_000000. . . .000000000000 0000 000000000000000
No s W i U 0 u W to W 0 1 0 t0 K M W I"C Kto C 'U0 W c W W U M W W Ub u C7 r- Intra'-'0 c t*oo ILI r.:0, -' 00 0L'0a. 0 -L CIaUrn , Ii''0U
0000000000000oo000ooooooooooooo000000000000000000000o
0N~NNNNNN rc 0000000O C00000 C0000 0 00 C0 C 00 000 0 000 00 00 0 0 0
e ON0 'NO f 0'0'0J N NN 0 Nz :tU NC D; 0
'00Mtu NN D9 r Su 7a ( > o- 00NU0 ;tO N t C4- a0_ oqooo jzteoo>SeOct~.O_5o0000000000000000000000 000000000000000
aO 0QQCCOC0NQNN ;00r0000
0 N '0 5- v N - O 0 O e _ N c r0 t N 0 aC 0N r - a - n 0 0 t0 N N 0 o
OOOOOoOOOOroOOOOoOOOOOOooOOOOOOOOOO000000000000000000000000000000000000
'00 n~a,0'0t'i90 0'r-Oz Ooa)ao O<o.r 0rIaOooNeoNS=_0 N NIineeI-_or vuN Nw Ntfes __*-- -_e '0
0000000000000000000000000000000000000.**.**.
00N~~~N ~~~NNNNNNNNrrr -r'r- 00000~00
a a o t_0 0c 0 U0 Nf), Nv00 i ) 0''- No N c io_ 'a vN
No0 oootNoooooooooooooood_ oooo ao
O t - I NN C't OD M C
~o 0 n _ _ o ot ~ ~ ~ ON~~~r _ F N c7^ gt qD 2r Xt Ch _ O U) Wt O ¢t t _ q _ _ _ _ r_L, e vf *- 0 ' N 0 DON 0 0 N 0 r- N -0 0 0
:Z:0 Z V :&&0 &:,&&&&&:N &&N &.-&&:&*&:*:
000 0000 00 0 00 0 O O0 0 0 0 0 00t OOO OO OO O OOOOOOO COO
WIONNNNNNNNN N9Thz z h 00a OOO0000000000000000000000000000
0 0 C r C J N N N 0 N 00 Oia) C N 0 C 00 _ Na _ a. N 00 0 0
a m) 0 0 , OD N 03 N 0 -0O0 Q 0 - 0 0 i Q 0 0' a0a1 0 Na C 0 0 0 0 0aa
IOO O O OO OO OOOOO OOOO ONO Or ONI O ON O OO
l ~ ~ ~ ~ o o o oo o o o o oo
00000000000000000
'O C N C O a 0
000000000000
0 0 NO N '00O N r
000000000000
0
I-04hia.
W
uILhi
Uhhik.b.to
236
00000000000000 000d00000 0 006000000000000000 0 00QC, OC:07
U' £ N '~ aIn WI a0 t 1 'A N r- r- -O 0 a a,0'C O t U f f f V m I i I I A I r e. I D ' 0 .. . r' C - N N e N I N L. I .I. LI - . . i L .
N N O C a C), C, C.) N C , N C `3 Nn C, N N N, N 'N N N N N N N N "J N N N N N N N rI N N N N N N N . QI N N N N a OO iOOOOC.OOOC 000C00C'0000.)00C0C0C~c00C 0CC..0~U,.. w0I., C.,0000000000000000000000000000000000 Ci0000000000000000000
,co a, r- CC· C r Ch r- .OUI .r· r- C.C- ~ ce~ t O ~h 3 o " ~ ~ t rcl i~ J ~~: r· ,- n~ c ~ r N r r- c~ \r r J; 1 N ~ c ~ r ;.`C ~ ~ ri u C ~.0~~~~~~~~~~~~~~~~
0 a C C, 0 0 C, Q3 CD 0 0 0 CQ cl N 0 0 L C) C. a 0 J C D , C>' C, 0 m O C C C., a c> C, 0 c' 0 T o o v o Q o C, C, o 0 a 0 a ~ r Y. r · C O V)O O e~~~~~~~ r~cr fN I'" t'
t' ~
' " t' ~c -~f gC 1r~·C tr C r N hi m m ~ ~ ;ru· rr· \r, ~ ~ r- ~~ uJ rr o~ ~ o o c r N ru N rr m t J ~r v m u· ~ ,r- rer r~;- ~- j~ J c ~ ~ ~ r r r cuc~-"t ,~~~~~~O~~~~r ~ ~ m~~~~~m~ . ...................d~;~ddddddd~d;~ddddd;,~j ;~; ;~...............
237
00000000000000000~00000000 0000000 000000000000000000 00000 00 00 0 0 00 00000 0000000000 00000000000000000000*+,+++*v+**++++~*+~*~+)+*+44**hh***-4*
U Y W W OS W aY rdN ra l us W U f W W rW r rdW a W. W a - w3 W W 3 W go W W . to W dwW W W
P* el f0 a, 0' I V V I C, E'I a ft 0o0 f, C, Uo 0 (' _N 0' fo I C> _ J' 0 n fV P1 O a' 1W - o C o G 'D f 0 PCS "" I', 00 'fW , 0r~- V'-t'-'oa0o' ooa' 0me oo000000000000 OO--.- 'C-''CCO o oo
* a. a a'a' ' a a a ' w a a, a m 0 O' 000 0" o 0 I I0000 000 00
000000 C - 0In. 0 n 0 0 0 0 00 000 00 0 0 0 0*4*.444*..4...4e4*4.~0~~~~~~~~6~~6~~~6~ * · · · , · I··· · · · I· · · · · h h3 h hhhh h h h hh
00 0 ft.000,00ft0V00,-t- 0.-a'0~000g'-*O00 '0 V i1 00 LI ' '0 a ti 0I. -0 C0U t 0 I 0 ' a0 f0 = -
238
ooo000oooo oo0000000000000000occcoooooooooo000000 00 0 000000000000000000 o00000000000000
* *-+ 4+* * +0* +++*+ 444-4.444*4+ *mg to h- he W la W W P. a.. *a W W h Ca hi W , la U. W U. W W U. W . U. h : - C W Ca W. Wi W W W W
C_- m cP' !n i P0 s c S o a N ' S 00 Cr Mi C v c 0 N a' 0 OM
0000000000000000000000 ' 000000000000000
0000000000 a Ca 0 00 00000000000000
P a 0N '0 0 P ll P. a' N t W N 0 '0 O 1 h Ii p 0 . P M - 1 O N r' .0 ' O 0 ' ON Ii .1 a' P S " P. 0%
c 0 M C- M - , a, a C> . In p I p. p p. p. p p. p , p . p. p C, p. p p. p. . p p
239
APPENDIX B
DAMIE PROGRAM
B.1 Introduction
DAMIE (D2) is a FORTRAN subroutine which computes the
so-called "Efficiency Factors" and Stokes parameters for
electromagnetic radiation scattered by a sphere. The form-
ulas calculated in this subroutine were first derived by
G. Mie (M5), and thus this scattering process is referred
to as Mie scattering.
Mie's expressions for the radiation scattered by a
sphere are valid when the radius of the sphere is comparable
to or greater than the wavelength of the incident radiation.
The index of refraction of the material of the sphere is
assumed to have the form nl-in2. In the DAMIE subroutine,
all functions are computed with an upward recurrence pro-
cedure. This procedure is stable for non-absorbing (n2=0),
moderate or large-sized spheres. For partially-absorbing
(n2>0) spheres, the DBMIE subroutine (D2), which uses a
downward recurrence procedure, gives more reliable results.
In the present study, the scattered intensity of water
droplets is found as a function of droplet size. Since water
is essentially a non-absorbing medium, and the droplet sizes
under consideration are large, the DAMIE subroutine is used,
requiring much less computer storage.
240
This appendix describes the part of the DAMIE code that
finds the relationship between the scattered light intensity
and droplet size. Other features of the subroutine will not
be presented here. They are described in Ref. D2. A listing
of the program and a sample problem are included at the end
of this appendix.
B.2 Description of the Program
The expressions for Mie scattering can be written as
I F' I. (B.2.1)s -i
where I and I respectively represent the Stokes parameters
of the incident and scattered radiation, and F' is a
four-by-four matrix referred to as a "transformation matrix".
It has the following form:
F' =
M2 0 0 0
0 M 1 0 0
0 0 S21 -D21
o 0 D2 1 S21_L
(B.2.2)
The DAMIE subroutine calculates the elements in this
transformation matrix at any scattering angle. The scattered
intensity, Is, at any scattering angle, can simply be expressed
asas1 /2(Ml ( e ) + M2 (e ))
Is 2 2 ik ·
where
-01
I
241
k = 2 (B.2.4)
where X is the wavelength of the incident radiation, and L is
the distance from the scattering location. In the program out-
put, the value of (1/2)(Ml+M2) is also given under the heading
"INTENSITY" for each scattering angle, as shown in Section
B.5. The only purpose of using this subroutine in the present
study is to find this value as a function of droplet size.
All of the necessary information for this computation
are input through the main program. These include the
refractive index of water, the wavelength of the incident
radiation, the scattering angle, and the droplet size. The
subroutine is called to compute the transformation matrix
elements, which are then used to compute the scattered
intensity factor.
B.3 Description of the Input Parameters
Card No. 1
RFR, RFI, ALAM
FORMAT (4D15.5)
RFR is nl, the real part of the refractive index of
the material of the sphere. RFR equals 1.341 for water.
RFI is n2, the imaginary part of the refractive index of
the material of the sphere. RFI equals 0.0 for water.
ALAM is the wavelength of the incident light source
expressed in microns.
242
Card No. 2
THETD(1), AJX, JX
FORMAT(2D15.5, I5)
THETD(1) is 01, the smallest angle between the
direction of the scattered light and the direction of the
incident light in the calculations. It is expressed in
degrees, and its value should not exceed 900° .
AJX is AO, the interval between successive O's for calcu-
lations.
JX is the total number of O's for calculations of a
scattered intensity. Its value should not exceed 100, unless
the dimensions in all related statements are appropriately
changed. It must be greater than or equal to 1.
Card No. 3 and Onward
X
FORMAT (D15.5)
X is the water droplet radius expressed in microns, for
the calculations. Execution will be terminated when there
are no more data cards.
243
B.4 Listing of the DAMIE Code
- N n O e C O N n t U'% r o o - N U un Q I-C, O o0 0 0 O O . c 0 0 4- _-_ N N N-. ,. N N \ j CNj NO C O O O O o o O O c O O O O O O o O C O O o C, O C) C,000000000 000o 0 00 00 o0000000Z Z / . z z' Z' J Z;.' ; Z Z ''7 ;' - Z - . 7 z
4 4 4__444_44_4444 44< 44444
M C0 -4 N er -4. '*sI .0N N <! F. n re, re000O0000
11 -1Z ;71z z
- -- -- -4 - -
xxxxxxxxx
zO
4 t- -J 4
_. 0-.J - 0.
U 0
> 0 =J C C
LA 0, LOLL < I IO .
UJ z O
u. - 4:lJ 0 ZO ce Lu V
ZLI C- c ,r =.,U_ 4 I L C>Y CZ- U. - -
LU. U cZ0 rC 0- lQL W UJ F-II I iI= U. xL- -. OZUWcc W -3
I I OI9 -as
Io-I w 11 YJ - o I- > - Lo o C i
UL 4W L3-U. U. J:Ia ct 41- :
z0;-
D
U-r: QI0L)
IzoLu
I.)
Io
4
0UJU-0
L
uJ
IIX
0.-
0O
0 W
: FILiJ CLLU
I-i--
"C C%.C30.0IIx
LI I
I :O 4
0 Z
c st.x UM
-I 0
UtZ L *
C -- 0
U-C w Iz w
CL
- C -0. L
HrU 1.C.- I.
-- _ -,_'
-4 Pv- ". l,
LU -
(A -_)
00 00C C30 --O CC C-' O ULL A. L U. . C0
0 Z 0 0 0000 -0 0- - - -.. N" e4i
IA
- z"C
N-4
I,.L -
I
7-'
%D
-4
U0-
LL
,UL0-U
LA.
-n .-4 I.
0.--4
U.
LU
0.
0- O*L _
c - 4 4t - ci 0
0 000 00oO'O o-. o
0OZ O--n 0
l- 00 0
I. . L .OZ I--ZO .
, < _ L '
-0 0 _UL. . L -
- .U I, N U ,LL O O - IA.L L W U. J L WI- 0ar
WW4X .NIfU.'' -- X a, Li r\
U . U. .4 w U w
4 4 Q * *
IsII .,J. H ,. t 0. 4 4 Ct Z 4OOwLii UJ CU
000000P F co
244
70
I-z4
UV)-IUz
0
LJ
DfL
1
U. J
0z
400
4LA.) 0
Lu
40
fl 0 O rr - N , ,, 1- t ,rOOOOOOO O OG O00000000000Z Z: . Z Z Z Z Z Z Z Z
d d 4 i 4 b I.- - .4 4 . i
4:4IE : 4 r4 4 x
0 0 o N - 0 1.0 fl C O -4 N n 000000O000 000000Z Z Z Z i Z Z ; Z ; Z z Z Z /.. Z
ZE ' ~ E E : E E : 3E 3E 3E X :E 3E E
X
UJcc-iw0.acc~
x V)2IC
_ X
_j nC -
O**, 4U. WW _
I- 0CV* <ZO
C ) - Or eW w I
0 _ a X --o< ui If 12*O*%' * 4.11)
Z- _J L v)| _ J CD CL X U O 4 Nc 04WII4 - (V ) CZY 1144>0c
le43
Y 4
NX
-J
U .U -i
- uJ_ *
LL) YrC n
X -X cy- tY J 1*- LU l)
I - I
LL0.
X l:
O4" y ^' "UJ O O
"3 - -" Z N " -n *d I--N I- Z ZZJ Z Z ,-'
< < U V CL-C(
00OI-z
* NV
e 0-i 0.o
*N
zz -b. 44 **
OF r O- II-I *C1 N
- ** *
- -
cx c
1- I.W W
I' 0 -~cW - I
I WL-4I* 00.00
o 0000x
Z Oi,==-4
c: '' II
II-V)U
0.
0O0C O UJa C =
z- h4 -C- -
O0 ~ b- O.0. O L ) W0w c w 0c
009-4
245
x r4
0.0'(I-< xI
4 1
0 0C 0IaII--LU 0= c
If
ON
00 -4WWuI
uj :
X
0
z.-4IIO
I.z0, -
Ocr
Q CJ o o C . O C -C) C) ) 0000000000000000OG O O O O O O O O
L- t L. U IJ ' I L tl J -, L
ooooie , 1
-·N , t U o ' Cc O ¢ _,(-_.. .-4 _4 .- _9 - 4 _ p - .0 0 C) 0 0 0 0 0 ()() O C) C C, C fJ C O C) O C) C) C) 0 0 C) LI U, J L -' Li ! I L' L U._ Q4 - - - ' _ -- C-
C' N r ur rc, aN N o-N Ne e r n tr') o C. O :_) rD C) O f ) ) 0 C0 3 OC)
L. i L ' L J L. JL ' 11 L-, I U L LL LL .J
C00C)HYUY C 01 C)00000
0C)..
IN- _
, N
U~ 0 0N C )_ 4. UYV CN
Li - s
*4- I-.* I N_
N I) ZQ U
.. C) NO O0 C
-
U
ZwL
_-' * * I -
< r
_ ,Z 0 Z
0. Cc U -. -I- L.. *
· ,, 3L r c · o * 4 .
c C) . Z C
WO - o LL _
OX s.--O
0. L ac U.X CW L , UL _
^ -O ww a:IJ ->>· Y w _ CL l
<_ M u
X- W C U- 3:r eJO-1 != -<
U-
x ox
--J OwaoU -'"-J II U.
U-C
LU
I--0C)-"'--
II CO-0 0W 0
* 0o O
x - '
U -, C L
au IY .L
0o 0i '--l0, tI -I I -cU), O
UA
u-iN0
C00
UJ
IT
0a0O
-
a,
LU
I
0,
'N
*0
co6~Cy
N .-CO
* O
- 0t- 0
-O ·_'O I.-aF0 ab-C b -0 00
O * 0V C I.-
II --- O C -n - f<r I- -
F _ _ L WU -I 1- l O OI-_ -- N _ I.---A b- C IL (A
.. - Qi V V-4
N UiN
246
.-I
W L
- Cr_
-̂ w
* -
C <
U.
O. rO-
ZI -LU ·C-0 O
cr c4-
U4
WOXA D 0
LLJ
0 L. La..Z C0UJ x
I --a LO LL
3 0 CLi. .
L _.. _-
IO r CO
'4.
a=40rn
I=
I-
wci
5D0a0 .
ujLU
* UJcc
)- UJe-0< 0-
010 r
U-"-4611
247
0O00000000 00 0 000 00000000 000 00000 s -O O Oa C) OO O C O O O O O O O G CL. O O O O C 'O O O O O O C - O 0000000000000000000000000000000000000LL tUJ Li ' U L-' L. LU U-L Li L LJ UI LJ L L L' U J i 'LL L' L W L. U LU 6L; ii L ULJ L,' I' UL L .-,, ,4 -., Ira,,,-4 - Imbl I,,1* ,,= f , b ,, !, , , , b ' ,, ,, " " . b -- 4 b , ~' -- " =-
ooX X x 7 o Z o 7oo o ooo E o E o x c c , 0 Q c 0 0c0 r x 2 Xo o o o o c no o o Po o oo oon acoo o o o o o o p aooc
O
_CF
0 _o OiC W 01-
0 I X 0001
30 1 I --11 11
7 0 W - 0- W 11 i ic _ _
N
0u
-4 -.
F -3N*_ - *.-
z _
XZ
Z J .I X *Z LV r7>.I I -..._ 11 l 4I _ 1
Z-=.P-Z Z -O -=- ,,,. Li. -
VU.
0 = -
OX- -0 0 - I--. 0 + I
"_ " " t 1 It
LL- - "- __e o\ W F
N__*
- -- - - -
N
** * -' -+ *
t,.- I., I.. 0~ /'" ~ I,. ..
II II I! II II gI - I
- --- -- - -
00O
_ o
_ O
0 r.L.) IL c II 0 4 0- 0
aO0'4
248
m 9 ul '. I- w 0 0 "- N r4 it O r- C2 % a0 c' r'n t U. r -I aoC 0 - N 01 ' o- Z .- O- F rf- I'- rI , t c a x: ao a- a ac ' a- a a C " 6 C > c F D C) 0 (O C O
o 0 O O O C C 0 C C 0 0 O C O 0 (-I 0 U' C) O O O ° O 0- °, c ° - '-4 .- c4 '-400000000000000000000 "0000000000000000
I LI L ' ' ' L J LJUJ L- LL U I.I '.'' ;! il .J L L. L , U. t I.U 6 L. L. L - -' j U. L [J tJ
c00c ) o o a o o c) co o0oo o oo o
NNNNNNNN
I-- - F- I- - -++++ I I I I
N fq N N N eQ C~ N
.- 4 4 .'.N-
. r 5 . -
II 99 II II II
- 4 - I," -* S. S. 9S. . . S S
- - -N -
l- Cn9 II 91
ti It 11
od or cXI nt
wwuwwwwww
00
+o0
0iN
U
U, 9-+_ N
, _ C* A c:
UOOU
'-
4+
*
9--
N-INw
0.--~9.,-- I
i ,-,I.-* __ V)
IIt -...-, I_ II3 -4
Z C-N _. _97_ .I cI -I
* -_) _
' _
_ _4
Q _* tx I
I + _1 _ *1Z9 z1 1-_* I * 7^A 7
i II 11 0 ^ cC
11 - - - G Q <-
_ _
Z ZLL U.: 3
I
Z:
_N3U
LL L) U
+ I I C- IL LL
-4
UlIx
U.
OCO X
* M-4 +- LL
+ ci:x Ci
U *C-I -
U UL I--
0 0O U
0
c- _
0-
C:) .. 4O 7. ·.- I-:I II
LL - -1- - "
UU-- 1- x
I V II
II II. _ -0
UUO-I-C.
-
II --- Z3:
I-Z0- U0
249
0 o -4 N n 0 ,-4 -4 ,-4 , ,.4
.4 -4 - 4 -- -4U CU LU UJ CU,
v4 -4
- ' Z-- ·' XLL U,0. 33 U·X I IL - _
N NI _ _
U. U. LL
- Z g..)-1 IIIII
I O :-t O X --
- Cl X -
3: ++Z1--* X U.+ I I
*CL C * *1
< <
U. UUQZZ
0 O 4 w % N r -A %O P0 O c N M --t L O F wo 0 c N -4 _ 4 _ N N N N N N N N M M M n r , M r + 4, -- % ' t- -_ -4 -4 -4 _-4 - - - -4 - 4 -4 - .4 -- - 4 , - -- 4-- _4 -OO O O O O O OO OO 00000000000000000000000 0000000
L.J LL L.! LL. L ll i J Li I J L ! t L U J '-' L- U L L LiJ -L' LL LJ0- I- 04 I-- O.. d -. - b-4 - . & 4. - b" -b- -61 4 0 4 b -- -O- 4 -4 -.
: : 0: : ZE: X 0 X : 0 7 0 £ X 0 o 20 0 0 X X X c Q c Po POsoa:c 6 o o Qo o o D o c3o Qo
-0No
--
-
9 -
A N,-4-
II II IIZ f- -I, 1, t,
+L)
-*-4w
uI--+,#.-_ _
_ (V
az -
I LU
* A_ N
-
*.--,+-_ +
u -.D -
* .-
tO-
- ,
9-LUt * II t
1--
c~,,- .._ _. -
N**NU-
**
_ +9_t Yt ++ N.- #
I-- -
X..-IAx--)
09- II- 6D -+ Y
+ A 44
ill co Oa - IIII 1II-
x -WY -a1 C o
4 ~ 4F4~4 4 +4~ E. ...._ _ _ _ _ _ , n
- 4 . II I ....N ..*~ -. .. ,~ ....~ .= , N C,WZC4Y~ St~t 4-
*4.** *-*** ****4 N _ N _ cc C C ce __
-- - I - - - ..- - -- -; I
.J C) -.0 --I -. 2 --. -.* , + + + -- + + + W W Il tl t11l11 11 i 1 II Il t-"7, -. * . * . . . . .
Z"N'-4 * * Z * * * * e. . .. __.=" t I II . I. ,
Zlj t , N Xx- Xf Xx xZ
- W-, U WU-.W WLUJ I W., L,, U WL- - - - L- - - - - - -
~~c 43c
00
0I-
0043%t
V-4
000F-
0
..1- +uI, +9 Z
250
UN ". a:- - O -4 f,n ., UC '-.I ( o 0" 0 - % _ N m .- U'\ '.0 I- OM 04 4 4 - 4 .t , ,-' LT' LT.- i L' V, I' . ' s C ,, C ,ca ; . ,o , O
,4 _-4 - -_ 4 -- 4 -- 4 i -- - -4 - .4 , -- - .4- -4 -4 , -4 -_4
O 000000000000 C)O O 00 G O 000L- LJ i; L 1 L LJ LL LL L . L L; UJ L L -J U U; L . '.L L L- LJ
X X E X c X E E E E E 3 a a a 3E o 3E c E ~ a ~coc3c3c3 cc::C oC3C3Doo sc:3oc3~oo:3c cn
- II II
" *. Z iL_..4 N - II
- =I ,D Z <
00'
w _
N N * ***l-
_ I I I tI
C: b' 3 ' y" - UJ wN - ,
II C N N _-4 - I 1
Z C a o a _ _ J_ CIO Va X 1 1 1
00 %n
LI
X - 0* -X I .
Z Co t WO X. 0
Z-X 4 C a
Nr4
3-
x -
II II
C .
251
B.5 Sample Problem
A sample problem is given here to demonstrate the
use of the DAMIE code. The material of the scattering
medium is pure water which has a refractive index of 1.34-i0.
The wavelength of the incident light source is 0.6328 pm. The
scattering angle is set to be 26.50. The radii of the
water droplets for this computation are listed in the input
section of the sample problem.
The output of the code gives the input information
as well as the elements of the transformation matrix for
the input scattering angle and its complementary angle. In
the present study only the values of the intensity are used.
All of this information is repeated for each input droplet
size.
252
N t 0, p so w 07 0 o r N M 4t
00OO C)00000 o O C, 00000000000000;;) 3 3 ) C ) ) 8 C) 0
. C L CL m C,* a- . M C. G. aQ aZ;ZZZZZZZZZ ZZ_ _ _4 M _ - - _ _- b
C
IJ
U
co
-J2:x O
0 VIOo
IL)-
zC-
0 ·*
00N
m0-4
000N
0000000000OoooooocoO00000000000O * . . . . 0 0 0 0 *000 O00 O O O 00000U- ri J t U. 1' . r- % am 0 O
.-
253
0 ,o0 Z 00
0O a i
O II . +++ . - UJ
O LU Z ,,N
V0 Ur O Z L,O f , OO* 0 O * Y
0.O
UN §II rf N
O 0 > + +0+ c c 00o : W
- -Z ' 3 I"" I+
a0 w r 30 NOCO* -0 .O W 0C
C. .N· + +V) r' 00 00/-0 O C C)CJ ++ UE U .Q C33 e n CD OO O O
U. O O N N O+ 00 0 0 O X C3 ,- - O O OO C N O0
I 4U U I< 4 Z 3 Zr _- O Z O O
O O I + + W) LU U C+ Z 00 Z U - L <C
O .~ A.D CO -4 O N0 4 OD U L i L >-N L. ?' X U 0 aU L II ) W W L *
0 Z L 00 U. . U. >
- - O O O 09 UJ W 00 U U U O I: C N ++ <4 < U+ - Z CO I. U. UIL 0 "" a a U.0 U. ~ ~.. >. >. 0O O UJ % LO U U U -o , u C O W W Wc
0 Z Z * *
0 IA - Z LA. LAC U U U U £f; QI LLWI. U. U. -J tJW WA W tW Y 00
O 0O O
J eU.Z N, U4 '.4
254
o , 0
C, . ._NN 0C 00' 0
0 CO
+ , UJ
O UJ Z ._ ,,O - . LL' eC~OU. W O rI.-C m O0 Z " " -N 4 * -i C+ \* 0 O.
Q.
N II ,4 t'IO O 00O U- 4' +4.
N 3 0 0 L ^0 - C LUD co)
I O O 0
0 0Q.. ' + +f - Cf O Ov
,-" NC3 .
U.0 O 0 o0 Nr 0 0 0
O 0 0 O 0 cr N O0
4 i Z 0 ZA O- Z O 0
u Z 0o o - cr: +0 C , + 4. 0 Lu L. C+ t 0 UO Z U .0 I O I N > r
0 , 4 N ** O I C ) OO Z ILI O U. UL LL
+ 3. I,. U Z .L. i .>0 : C)N LU W 0 U.. U. LL
O IL LU O0 Li o o 0 N +4 4 4 4 Lio Z CO Z Z Z O ' - -r -0 w w uJ* Z L * * L L :
LU L. U . U. V.'
w OO 0
Z N u.¢. "4~
z-4
4
-0
255
f,- u'
00O 0
>-- O0,- +U.
IW uj
0 Z (n0
O O
_4 + +0N 0
O
00O 0
I
,-" 00N + +
CCO N V) 0N
0 IO O
.- -4a cO
U; 0
0 0N N
* 0
0r 0Z
-4 ++ u0 0 ZV) N X
I1 U, N W:~ ao ,-
U ** UL)
Li* 0
W 00 U
, N +4. 4Z 0 IL
< O _.
Uat U.LL LLW W00
Z N U4
0
(0 Zz OCie Cw a.
0 0U. LIL IL
00
.-4
cO0t-
0
0 a0
1-- I-. 0U U 9--
U U *
U U
U. LL IL lL U)
UJ W 4
%
ao
0'
II
atwUJ
N
Z.--3
00
N0
00
0
a0'0
II
040-4
00N4oI
O LIc.
UL
x0.
* U.O )
- -zO :+ -40
0 N I:
0 Z* 0
O zT
IA-
N Wc\ uj
O x
+ I.0 I0 I
* zo U
wJw
25627
-4 U4o
* z ICJ
- 0 0 C 4C\1
+ w z, WL
o wU F O Qt
0 < GO
O - 0 0
0 - Z cC OC0c0 _ ,_ = OC ~
+ 3 00 vr V\
0 UJ O0 ,ocI I . 03I % 0 VWJ 4 _
· o O o atv + + +
U, a ~D A CD m a
0 X VI .- CO 0 0 o Ct U c o
- 0 Z 0+4 0
a OL7 CC Q e C0 05 0 0 0 F t
·' Z O - . Q* - 0 0 0 C Ox. ,- , s0 - - Z G4 4 I. O - ,40 - x . at (7'O O'if UJ% U 0 I 'I
O1 W V\ Z ' Z 00 _ : t .O WOU- -4J e C
t LLC 0 0 00 Z LJ 0 U. LL U.
-- 0 D0 0 0 N u W 00 U 0 .U-
+ - Z 00 U. L. LU. 40 U. D (In >- >- >-0 La. >. V) IS rn U U0 , cI Z Z
/ E .- -W W-4 J-* Z 0 · · t 0 0 00 WA 4Z 00 - - - Z:
2: U. U. U.LU UL. LL U. U. L).. tAJ Lu LU UJ
000 0
Z cu I.'
257z
-.4 4- Q-I
* .
P- .JQ
O ( .
0g 7N ' 00O -i I +++ - wU
O W NO' 0 0 · ·* 0 O*O 4 00
Q.
0 ~- 00O ' 0 +++ Nf OCO -- Z CC "- OOC ,- ,- D , NN 3 tO F
0 IAJ -0'0 UU 0
-. 0 0W 00
0 or0 0. C + c&O) - O D0
O c'N 0 O OIL O N .- N N O
+ ) mt * * Uo X 0 N 0 0 O OO a! N O 0
4 y Z U ZQ > O Z O O
O O - ·+ U+ LJ 0 ; O+ - 0D Z I a !Q t .4 O .JIO Z % O0 U. U U .
_ U. J N n l U. a- )-* 4 so. O O O OO Z l 00 u. ·. L
a! a! a! a!I * O O O c
N W U 00 U U U WO I 0 N 4+ 4 4 4 U+ t- Z 00 U- U. LL L-O _ (D r. · L
O U- D > 0 , UNO 0 U N U U
o 4 u r o o - - -. W
I a U UL LLW U. LL L UL -
U W W UJ 4
O O
Z N4, '-
258z
N 0 C)
sCY OC)
o -j
O O %
I'.'
0 I O 00 "') Oo wo i - io UJ
! : ;) * CJC) W Z U eO 0 Lo r· ~ k, 0 · *,a0 E O Z ~ ZC
CC O * 00 C 4 + +
a' Nc .+ -4 0 .- C0 I -
x -_ _ D , CDN 3 0 i t OC l O IA,_r) U .-
* . * , , , >-0O W 00
L,4 U.I O O0. ".0 t + +L
C) I ,.A . ,A00 U. U.at O C N %;
L 0 O rN N N 0
> Y) * c
0 O C N 0 O
4 4 Z-0 Z: - O - 0 Z O O
0 %O l -4 O
- Z O OC) C uOr O - + O - c U.
0 0 ' 4 W 0 C' 0O C : u C
<~~$4 0
i * O 4 O 0X a0 l t a C
N W Lu 00 0 0 0 O O N 4+ 4 4 4 0+ - Z O L LL L 4
O O LUJ i 0 0 0 u Vo > aa t z L z CO C) _ : -03 LU LU LI W
* Z 0 *. * 0 V 2:O W 4 00 - 'I a:UJ L WL Lu L L W
U at 000 C)O O
Z Nu )4 -4
'-
z 259_N I,, , C-4
4D CO -4o , 00cCO CO CO0s C
· 4 II
0C
O ~~~It~0 -N >- 00C I, +++ U WUJc3 tce a r
O LU Z O oO 0 lJ 00
F LU c CO
0.
UJN II It, ,tLrtO 00O0 + _ ++
+ ce N ' 0 Z CO n -
t I i
o 00
O OCJ 0 U 0N + + -
O N 4 4y 0 .0U0 _ CD 4 0 0 -L0 0 NN N 0
+ tn N' *O X 0 UIi'. 0 0 0e _ O
O C · + U U O 0
0 LA. :D % 0 -z - 0,4 Z Z 0 00 I O 0 + + U W J 0. + 00 a z c '.
O 4 CD I. I- . 0 .N Ct v L Nc C 0 J
CIAJ If C U VI ONI c 4 o¢ NLL .. e Ce c c c7* 40 0 0 0 O
O Z WJ O LA LL U..
ce CY w ciI ' O O 0X 0 rn C' t- O
N L w OW O 0 0 0 0 N + + 4 + I Z 0 U. UL IL U
0 U. D U.0.o 0 U 0 U, 0 V 00 > u C) Z Z Z xo - _ : .0 w wL w '0 N - U* Z V 0 0 U0 0 O W 4 00 - - L
W U. LL L. LA J uJ LL W LLJ <
I LO c
Z NLA4 -
z
-4
I-440
0CL
260
,0 co
0 0
C)I I
NO N
- O W-Z 0 P
0 Z CO 0 2 0 0
O0
0 0"4 + +N 00 -
"' C;)
CO O
0 a0 C
a
D re
%O -
* 0
-4 -40 04 +
0 0r 0O
O o
0 a0 00
m3 z (..3 z, , Z ;. Z
O 00 !-- am 4 4+ U W .
C0 Z - a:
11 m' Qi - V, C
· Q. r0 a: a· a0W 00 U. U. IL
.O ICl0 O0
a 0 0
0 *O OO 0
00C)0o0
O o. · 0
Z 1 r4 " 4r
0C
N
r0co
cw cc ci0 C 0 ca- - - 0
- < < U
z Z Z aCtUJ UJ - W -
U U U
LL L LU >-U. LL U. leUJ wu w 4
NN.
0
0In0
00
00
U)000
II
alLJw
CL
UJ40
3:
0000aN
%00Vqd
'I
zC,-Z
0
000,I-t
0 WIL*r0.V)
a:0U.
0 X0 -4
0 a:I-
0 0+ r-00 4
v 0
a:
o U-o 0· z
0 ww
Ulwj
wX
zw
LU
n-
wa:
I
I
I
261Z0_- AO(
0 t t o 0
o oc
o r- wC ,-, O ++
L+ WW
o W Z 0N 4 -4 U
* , -0 '-o - 0O 00 X _ +++ c N 0 0O I LA % 0- 1 Z C: o
+1 ,,.- t Z _. C. 0 ~ D 4 -C)
O W C) :;:cO OV) 0 0 .N ++ o O00 0 o ODo 0 c u 0u
c ) o N N NO -< O OL 0 0 N N- 0
+ 0 -0 0 0 O X 0V r' 0 0 0
* _4 0O * U UO0 I- k - ~U~ ,- - -0 < UJ < Z ZO .- L 0 U U O
tO c . % . U .
N 40 U0 ¢ CU X u _ L J D < * 4 * , O O O
O Z U 0J U. LL UL
UJ * O O O c
cN W tWL 0L0 V V U -O r O N + 4 4 4r r v+ I Z 00 U. U. .L O L .- > >. 0 r F 0Z Z -_
* Z 0 *0 * 0 0 V :0 W 4 00 _ _ _ a,
LU U. L U. U. (.J U w W LU 4
O 6
0Z N L4
262L;
U' Ci
N0 4 O 1
N - * *
Cnt~-0 C)
*,. 0 c.+ - U UJ0: Crz
O W Z W jO W W _ -
'.0 1 C Z 0 FN 4 * _, ' ..* C OO0 4 00
.
LU
0 I 1 C cr
N C3 0 - C
IO C Z c: r
0 C.~ _ O-
0 0 O N0 W 00 UJ - '-X O O0 - + +C
O 00 - 0 0
< O O + 0 0 U * Or~x xI o - I -o o
0 co 0 %O I 0
0 U. 0 (~ C - 0* L O Z O 0 m N - z ON0
· Z Z · > ' 'Z z 0
0 -L l U. U U. -4O O -u ++ U L C O
N . vC ) U.U CU 0S ^ 0 n C 0 )n a 4 s C 0 0
_ U J b AJ U-)
* 9 '-*.* C O O Oz+ I 00a U. U. U. 4o * O U.
O 0 W A 'ON U U U0 " ZN Z C L I0 c(A I O) LU U L
* Z U *' U U 0 W 4 00 - w -L U. U. LL U. (A
W LU w L LU UJ llJ s 00
0 C'
Z4 '-
263
" - co,4
co N 0 C 0.
0 "4 IIN - J0
N 00-~ 0O II ,.-. ++
0 , 0 - W 0
0.
LU
+0 004 W 0N -
0 4 -O Z -e0'- < * ~ e
N 3 I O0
0 0O LU 000. 11h-t + + O _ 00 - -
No + ,- %" N
c c 0 0 ·O. 0 0 0
0 N 00+ +Y <. , O C3
I S C. _O 4 Z Z 04 Z· l : _ D Z 0 0
- Z 0 00 a. .0 04 +. d 0 + 00 z 0 W 0C 0 X I 4I L 40w O Or
-40 0· O 0 C Q+ Z W 00 U. U. U. 't I 0 0 O 0 0
0 0~ NA +N . U, .= O .I-,4 O " 7 0.,. = , . UJI U0 U. U. U. 4I
0U U U.. L V It w i OO U 0 00 _ 4 O Z O OL U U U. U. U. I ., J - O _
+ Z 0 0 X L Q·I( C c rl LCh r Y C1 0
O U4 O OW ~ t L L Uw t X ~ ~ Wu V L J l.. UJ z o oQ