Citation
Analytical and experimental investigation of the effect of dyes on solar distillation

Material Information

Title:
Analytical and experimental investigation of the effect of dyes on solar distillation
Creator:
Rajvanshi, Anil Kumar, 1950-
Publication Date:
Language:
English
Physical Description:
xv, 186 leaves : ill. ; 28 cm.

Subjects

Subjects / Keywords:
Absorption spectra ( jstor )
Ambient temperature ( jstor )
Distillation ( jstor )
Dyes ( jstor )
Evaporation ( jstor )
Solar radiation ( jstor )
Stills ( jstor )
Surface temperature ( jstor )
Surface water ( jstor )
Water temperature ( jstor )
Dissertations, Academic -- Mechanical Engineering -- UF
Mechanical Engineering thesis Ph. D
Saline water conversion -- Distillation process ( lcsh )
Solar stills ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis--University of Florida.
Bibliography:
Bibliography: leaves 182-185.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Anil Kumar Rajvanshi.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Anil Kumar Rajvanshi. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
023062275 ( ALEPH )
05715665 ( OCLC )

Downloads

This item has the following downloads:


Full Text














ANALYTICAL AND EXPERIMENTAL INVESTIGATION OF
THE EFFECT OF DYES ON SOLAR DISTILLATION













By

ANIL KUMAR RAJVANSHI


A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY







UNIVERSITY OF FLORIDA


1979

















ACKNOWLEDGEmENTS


The author wishes to express his sincere appreciation and

gratitude to Dr. E.A. Farber, Chairman of his supervisory committee, for directing this research. He wishes to thank the other members of his committee, Dr. C.K. Hsieh, Dr. C.C. Oliver, Dr. R.K. Irey and Dr. A.K. Varma, for their cooperation in serving on the committee.

The author gratefully acknowledges the financial assistance of the Government of India, which made his stay at the University of Florida possible. He is grateful to Dr. C.K. Hsieh for allowing him to use the Perkin-Elmer Monochromator for measuring the absorption spectrum of the dyes. He would also like to thank Dr. P. Buscemi for allowing him to use the Beckman Spectrophotometer. Special thanks are extended to the Mechanical Engineering workshop for building the solar still. The author wishes to thank his fellow graduate students for numerous helpful suggestions and wishes them luck in their future endeavors. He also wishes to thank Lorraine Thomas for the excellent job she did on typing this manuscript.

Finally, the author extends his thanks to his wife, Nandini, whose active support made this work both possible and worthwhile.



















TABLE OF CONTENTS


Page


ACKNOWLEDGEMENTS .. ........

LIST OF TABLES ... ........

LIST OF FIGURES ....... KEY TO SYMBOLS ... ........

ABSTRACT .... ............

CHAPTER

I INTRODUCTION .

II LITERATURE REVIEW
Historical Introduc
Solar Radiation Brine Depth . .


. . . xiii


1


tion . . . . . . . . .
.
.


Cover Material and Its Shape ....
Ambient Temperature and Wind Velocity Temperature of the Condensing Surface Use of Dyes to Enhance Solar Evaporation Scope of the Present Investigation . . .


THEORETICAL ANALYSIS ............
Physical Model ...... ...............
Incident Solar Radiation .......
Absorption of Radiation .............
Conduction and Convection Heat Transfer Water System .............
Evaporation and Convection Loss from the
Surface ....... .................
Formulation of Equations .. ...........
Layer n ....... .................
Top Layer ...... ................
Bottom Surface ............
Energy Exchanger with the Still Covers During Daytime ............
During Nighttime ...........
Method of Solution ..... .............


in DyeWater


O
O
Q
g
O











IV EXPERIMENTAL SETUP AND PROCEDURES .. ........... . 42
Distillation Units ..... ................. 42
Thermocouples and Their Locations .. .......... . 47
Temperature and Solar Radiation Measurements .... 47 Absorption Spectrum of Dyes ... ............. . 51
Experimental Procedure .... ............... . 53

V RESULTS AND DISCUSSIONS .... ................ . 59
Dye-Absorption Spectrum Results .. ........... . 59
Temperature-Time History of the Stills ....... .. 74 Productivity of the Still .... .............. . 96
Comparison of Different Dyes ... ............ 116
Comparison of Theoretical and Experimental Results 137 Effect of Various Variables on Still Productivity 154
Effect of Ambient Temperature .. .......... . 154
Effect of Wind Velocity .... ............. . 159
Effect of Dye Concentration ... ........... . 161

VI CONCLUSIONS AND RECOMMENDATIONS ... ............ . 167
Conclusions ....... ..................... . 167
Recommendations for Future Investigations ...... . 169

APPENDICES

I PROPERTY VALUES USED IN ANALYTICAL MODEL .. ........ .. 172

II STABILITY CRITERION FOR DIFFERENCE EQUATIONS ...... . 175

III CALCULATION OF ABSORPTION COEFFICIENT OF DYE SOLUTION 179

REFERENCES ........... ........................... 182

BIOGRAPHICAL SKETCH ........ ....................... . 186


















LIST OF TABLES

Table Page

1 Duration of Test ....... .................... . 58

2 Comparison of Dyes (Experimental) .. ........... . 101


















LIST OF FIGURES


Figure

1 2 3 4 5 6 7 8 9

10 11

12 13 14 15 16 17 18 19 20


Schematic diagram of solar still ... ..........

Effect of brine depth on still productivity . . . Single sloping solar still ..... .............

Model of still used in analysis ......... Radiation transfer in the dye-water solution . . . Temperature profile of the layers ........ Flow diagram of the computer program .......... Experimental stills ...............

South view of the still .............

Schematic of the glass-cover frame ... ........

Schematic of the constant head feed system . ... Thermocouple locations on still ......... Attachment of thermocouple on glass cover . ... Location of thermocouples in dye-water solution . Beckman UV-VIS spectrophotometer ... ..........

Perkin-Elmer Monochromator ..... .............

Specimen cell . . . . . . . . . . . . . . . . . .

Cross section of the specimen cell used in PerkinElmer Monochromater ...............

Arrangement of the two stills .......... Absorption spectrum of Red dye (50 ppm) .....


Page

4 9

. . . 13 . . . 20 . . . 26 . . . 40 . . . 41 . . . 43 . . . 43 � . . 44 . . . 46 . . . 48 . . . 49 � . . 50 . . . 52 . . . 54 . . . 54 � � . 55

* . . 56 . . . 61










21 Absorption spectrum of Red dye (100 ppm) ......... . 63 22 Absorption spectrum of Black dye (50 ppm) ....... . 65 23 Absorption spectrum of Black dye (172.5 ppm) ...... . 67 24 Absorption spectrum of Green dye (50 ppm) ....... . 69 25 Absorption spectrum of Green dye (100 ppm) ........ . 71 26 Temperature-time history of Red dye (50 ppm); November
26, 1977 ........ ....................... .. 76

27 Temperature-time history of Black dye (50 ppm);
March 6, 1978 ....... ..................... . 78

28 Temperature-time history of Black dye (172.5 ppm);
March 31, 1978 ....... ..................... . 80

29 Temperature-time history of Green dye (50 ppm);
May 15, 1978 ....... ..................... 82

30 Temperature-time history of Black dye (172.5 ppm);
April 14, 1978 ....... ..................... . 84

31 Temperature-time history of the two stills; March 31,
1978 .......... .......................... . 86

32 Percentage of solar energy absorbed with depth of
solution; April 14, 1978 at 12:30 p.m. (E.S.T.) . . . . 88 33 Temperature profile of dye-water and control still;
March 31, 1978 ....... ..................... . 91

34 Temperature-time history of the two stills; April 14,
1978 .......... .......................... . 93

35 Temperature-time history of the two stills; May 15,
1978 .......... .......................... . 95

36 Distillate output from still with dye and control
still; April 15, 1978 ..... ................. . 98

37 Distillate output from still with dye and from
control still; May 15, 1978 .... .............. 100

38 Evaporation heat transfer-time history of Black dye
(172.5 ppm); April 15, 1978 .... .............. . 104

39 Evaporation heat transfer-time history for control
still; May 15, 1978 ...... .................. 105


vii










40 Distillate output from Black dye (172.5 ppm); April 14,
1978 .......... .......................... . 107

41 Correlation between h and T ...... ............ 110
evap s
42 Correlation of distillate output and T .... ........ ill

43 Analysis of energy transfer mechanisms for Black
dye (172.5 ppm); April 14, 1978 ... ............ . 113

44 Analysis of the energy transfer mechanisms for
water; April 14, 1978 ..... .................. 115

45 Correlation between the performance of the two stills 118 46 Comparison of outputs of Red dye (50 ppm) . ....... . 119 47 Comparison of distillate outputs for Red dye (100 ppm) 120 48 Comparison of distillate outputs for Black dye
(50 ppm) ......... ........................ 122

49 Comparison of distillate output for Black dye
(172.5 ppm) ........ ...................... 124

50 Comparison of distillate output for Green dye (50 ppm) 126 51 Comparison of distillate output for Green dye
(100 ppm) ........ .... ............... 128

52 Correlation of S and X ..................... 131

53 Correlation between slope S and intercept b ...... .. 133 54 Seasonal variation of the distillate from control
still ......... ......................... 135

55 Seasonal variation of the distillate for various
dyes .......... .......................... . 136

56 Distillate output from two stills on completely
cloudy days ........ ...................... 139

57 Absorbance spectrum of distillate from various dyes . . 140 58 Comparison between theoretical and experimental
results for Red dye (50 ppm); November 26, 1977 .... 143 59 Comparison between theoretical and experimental
results for Black dye (50 ppm); March 6, 1978 ..... . 145 60 Comparison between theoretical and experimental
results for Black dye (172.5 ppm); March 31, 1978 . . . 147


viii










61 Comparison between theoretical and experimental
results for Black dye (172.5 ppm); April 14, 1978 . 149

62 Comparison between theoretical and experimental
results for Green dye (50 ppm); May 15, 1978 ...... . 151

63 Comparison between theoretical and experimental
temperature profiles for Black dye (172.5 ppm); March
31, 1978 ......... ........................ . 153

64 Analytical plot of the effect of ambient temperature
on distillate output; March 31, 1978 ... .......... . 156

65 Analytical plot of the temperature-time history for
Black dye (172.5 ppm), March 31, 1978, for two
ambient temperatures ...... .................. . 158

66 Analytical plot of the effect of wind speed on
distillate output; March 6, 1978 .... ............ . 160

67 Analytical plot of the effect of dye concentration
on distillate output; April 14, 1978 ... .......... . 162

68 Analytical plot of the temperature-time history for
Black dye, April 14, 1978, for two different dye
concentrations ........ ..................... . 163

69 Optimum concentrations for different dyes on a
typical spring day at Gainesville, Florida ....... ... 165


















KEY TO SYMBOLS


A Area of still cover, ft2
c

A Area of glass cover, ft2
g

A Area of water surface, ft2
s

b Intercept of regression lines in Figures 46 through 51,
lbs/ft2-day

c Concentration of dye, ppm c Specific heat, Btu/lb0F
p

D Distillate output [equation (43)], lbs/ft 2hr D.. Binary diffusion coefficient, ft 2/hr F Factor given by equation (50), lbs/ft -day Fe, Fa Emissivity and geometrical shape factors g Acceleration due to gravity, ft/s2 Gr Grashoff Number h Convection heat transfer coefficient from water surface
to glass cover, Btu/ft2hroF

h Evaporation heat transfer coefficient, Btu/ft 2hr0F
evap

hfg Enthalpy of vaporization for water, Btu/lb hm Mass transfer coefficient, lbs/ft -hr h Outside heat transfer coefficient, Btu/ft 2hr0F
o

I Percentage increase in evaporation k Equivalent conductivity, Btu/ft-hr-0F
e
kp ~ Thermal conductivity of Plexiglas, Btu/ft-hr-�F










k Thermal conductivity of water, Btu/ft-hr-0F
w

kw/a Thermal conductivity of air-water mixture, Btu/ft-hr-0F Z Thickness of layer for convection heat transfer in dyewater solution, ft.

L Distance between the water surface and glass cover, ft. M. Mass fraction of species i near surface j
122
m Distillate output from control unit, lbs/ft -day
c

md Distillate output from still with dye, lbs/ft -day M Distillate output, lbs/ft -day M. Molecular weight of material i, lbs/mole nI Index of refraction for air-water vapor mixture n 2 Index of refraction for dye-water solution P. Partial pressure of water near surface i, psia P Prandtl Number
r
q abs(t) Solar radiation absorbed in a layer, Btu/ft -hr q c Heat transfer by convection, Btu/hr qX Spectral solar energy flux, Btu/ft -hr-m qX(t) Spectral solar radiation on a horizontal surface, Btu/ft 2-hrPm

qo (t) Spectral solar radiation passing the air-water interface, Xi Btu/ft2-hr-m

qin (t) Spectral solar radiation incident on layer n, Btu/ft -hr-m qs(t) Incident solar radiation on glass cover, Btu/ft -hr q(t) Total solar radiation on a horizontal surface, Btu/ft -hr r Reflectance

Ra. Raleigh Number

R Universal gas constant, ft lb f/lb-mole 'R S Slope of the regression lines in Figures 46 through 51 S Schmidt Number
e










Tamb, Tc, Ambient temperature, OF TH$ T Temperature of hot and cold layer, respectively, OF
H'c

T Temperature of layer n, OF
n

T , T Temperature of the water surface and glass cover,
g respectively, OF
Tsky Apparent sky temperature, OF cThermal diffusivity of water, ft 2/s (1 Solar absorptivity of glass
g

Absorption coefficient of the dye solution, 1/ft

Thermal coefficient of volume expansion, l/R Ax Thickness of layer, ft Ax* Path length of incident solar radiation ray in a layer, ft C(t) Dimensionless parameter in equation (10) X Normalized absorption given by equation (44) XWavelength (pm) A1, X2 Upper and lower limits of wavelengths of solar spectrum,
Pm

iEfficiency of distillation Vw Kinematic viscosity of water, ft 2/s Tg/w Average transmittance of condensate covered glass

0., 0r Angle of incidence and refraction at air-water interface,
degrees


P Density of fluid, lb/ft3

















Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy


ANALYTICAL AND EXPERIMENTAL INVESTIGATION OF
THE EFFECT OF DYES ON SOLAR DISTILLATION By

Anil Kumar Rajvanshi

June 1979

Chairman: Erich A. Farber
Cochairman: C. K. Hsieh
Major Department: Mechanical Engineering

An analytical and experimental study of the effect of dyes on

solar distillation was conducted in this'study. The analytical model developed treated the transient heat transfer inside the dye-water system as one dimensional. The bulk fluid was discretized into layers with conduction, convection and radiation interactions occurring between them. Coupled with this was the energy exchange from the surface of water to the glass cover by convection, radiation and evaporation of water. The bottom surface of the distiller was assumed to be insulated and a black absorber. A finite difference technique was used in solving the nonlinear partial differential equations. In the computer program the measured solar radiation, ambient temperature and the wind speed was used as input; the output from the program was the temperature-time history of the water layers and glass cover and the distiller output.


xiii










The experimental investigation was conducted using two identical solar stills (4 ft. x 4 ft. x 1 ft.). The control still contained water only while the other had dye-water solution. The dyes used were Black Napthylamine, Red Carmoisine and Dark Green, with various concentrations. The Green dye was made by mixing a 33 percent by-weight mixture of Neptune Blue, Tartrazine and Red Carmoisine (all GAF compounds). Test results showed that dye-water solution was able to increase the distillation output by as much as 29 percent (for Black dye with 172.5 ppm concentration). Based on these tests a simple method of calculating the percentage increase in evaporation effected by a specific dye over that from the control unit was developed. The increase in output with the addition of dyes was found to be pronounced for clear days. However, no difference in the still productivities, between the control and with dye, was noticed on a completely cloudy day. An empirical relationship was developed for evaporation heat transfer coefficient and the distillate output as a function of brine top layer temperature.

Among the dyes tested, Black Napthylamine dye was found to be most suitable from two points of view: increased evaporation and lack of noticeable degradation by sunlight during periods of tests. Red Carmoisine dye underwent severe degradation which resulted in the reduction of the absorption coefficient by as much as 95 percent at 0.5 Pm wavelength after a month of exposure. The degradation of Green dye was not as severe as the Red dye, and this was primarily because of the Red dye component in it.

The results from the model were compared with those from the experiment and the agreement between them was found to be excellent. The










effects of ambient temperature, wind velocity and dye concentration on still productivity were subsequently investigated. The model predicted a decrease in output with increase in ambient temperature and an increase in still productivity with wind speed.

An increase in distillate output resulted with increasing dye

concentration, but became independent of concentration after 500 ppm. The change of increase of output with concentration of the dye was different for different dyes (among those tested) with Black having the maximum rate and Red having the lowest. A proposal was made of predicting an optimum dye concentration. The optimum concentration was defined as the concentration where the sum of the change of increase of distillate with concentration times concentration and the distillate with concentration was maximum. For the spring conditions at Gainesville, Florida, the optimum concentrations for Black, Green and Red dyes were 218 ppm, 377 ppm and 408 ppm, respectively. Recommendations for further research were given.

















CHAPTER I

INTRODUCTION



Water is the most abundant natural resource on earth. Essential to most human activities, it can mean life or death, bounty or poverty, war or peace. Abundant it is, but not infinite. Of the 453 014
units (1 unit = 2.65 x 10 gallons) of sea water that evaporates every year from the sea, only 38 units are available for possible uses by life on earth. Of these 38 units only 70 percent can be utilized [1] which is about 25 units. At present the world uses 2.5 units. Thus, theoretically, we have enough supply of fresh water and the present estimates [2] predict that the hydrological cycle can support about 25 billion people with 70 percent utilization of the water resources. However, with the present rise of world population at 2 percent per annum, the available annual water supply will probably be insufficient for the world by the year 2080 A.D. But there is a basic flaw in the above reasoning since a major portion of the fresh water supply is not available where it is needed. This factor, in conjunction with the growth of population, will be the source of water shortage. Already quite a number of communities and nations do feel the pinch of water shortages because of this. However, the problem can be partially solved by transporting fresh water to some of these communities, but the costs involved are of such magnitude that this proposition is not feasible.










The problem is even more compounded by the fact that in some of the arid and semiarid regions the ground water has high salt content or is contaminated and thus not fit for drinking. Even in areas well endowed with ground water supply, because of increased industrialization and population growth, the rate of depletion of these supplies is reaching alarming proportions. With the cost of water transportation already high, some other way of getting fresh water will have to be found.

One of the promising ways to solve this problem of water shortage appears to be desalination. Desalination processes already supply about 7.5 x 108 gallons of water per day (gpd) in about a dozen countries [3]. Among the processes used about 85 percent are Multistage Flash Evaporation (MSF) type, due to the advanced stage of present-day technology in this area. The energy used in these plants is either petroleum or nuclear fuels. Thus, these plants are normally located in areas which are well endowed with fossil fuels. Indeed 40 percent of all the desalination plants in the world are located in Mid-East countries [4].

However, with increasing fossil fuel shortages and price rise, it is but natural to look at some other methods of desalination which use renewable sources of energy like solar, wind and biomass among others. Interestingly enough, the areas which lack potable water supplies have abundant solar and wind energies, which is a strong case for using these renewable energy sources for desalination.

Solar distillation has been tried on a limited scale. Normally,

two approaches have been used in using solar energy. One is the direct absorption of energy in saline water and the other is indirect heating










of water and evaporating it in a centralized facility. Only one big plant (capacity of 5,000 gpd in Saudi Arabia) has been built to date, using the second approach [5]. It is, nevertheless, in an experimental stage and thus no evaluation has been given. However, it has been shown [6] that this above approach is economically unfeasible for plants with capacity of less than 50,000 gpd. Thus for smaller capacity the first approach, namely of direct absorption of radiation in saline water, is more suitable. This method is even more suited for small communities (about 200 people) in developing countries where an average of about 500 gpd of water is used for drinking and cooking purposes. In these communities setting up a large solar desalination plant is economically unfeasible, both because of capital costs and transportation costs. Several research workers have therefore built, and experimented on, medium size solar distillation units (output of less than 10,000 gallons/day).

A solar distillation unit (solar still) is a very simple device in operation but the heat and mass transfer in it is complex. Chapter III deals in detail with the energy transfer processes. Thus, in the present chapter it is sufficient to describe briefly the working of such a still which is schematically shown in Figure 1. Saline or brackish water is supplied continuously or intermittently to the pool having depth ranging from 1/2 inch to approximately 1 to 3 ft. Depending upon the depth of water in basin a still is either called shallow basin (depth - 2 inches) or deep basin. The bottom of the basin is black to absorb solar energy and contains a drain to discard brine, continuously or periodically. Above the basin is a sloping






























Solar radiation
Condensate film




Glass cover






Distillate out Saline Depth of basin Water in * I F


Schematic diagram of solar still.


Figure 1.










transparent cover of glass or plastic sheet, which acts as a condensing surface.

The incident solar radiation after passing through the glass cover is partially absorbed in the saline water, the major portion being absorbed in the basin bottom. Heat is then transferred from the bottom surface into the water, thereby increasing its temperature. Partial vaporization of this heated water then occurs, and convection currents inside the still carry this warm vapor-laden air to the cooler glass cover. Moisture condenses on the underside of this cover, the heat of condensation being conducted through it to the surrounding atmosphere and the partially dehumidified air drifts back to the surface of water for further addition of moisture. The thin condensate film flows down the cover into the collecting troughs and is collected as distilled water. The output of these stills varies with various atmospheric conditions but on an average is between 25 to 35 gallons of fresh water per square foot of the basin area per year. Numerous variations of the above basic design have been experimented upon by research workers all over the world but the principle is the same. Some of these variations are described in the next chapter.

It will be seen, however, that most of these variations in design suffer from certain drawbacks which have made them technically and economically unattractive as distillation units. Thus there is a need to design, build and analyze a better unit. With this in mind, the present work is an attempt to test the feasibility of a novel concept of using water soluble dyes to enhance solar energy absorption near the water surface, thereby increasing the productivity of the still.

















CHAPTER II

LITERATURE REVIEW



Historical Introduction


Solar distillation is not a new subject of study. It has been

extensively studied for the past two hundred years [7, p. 162]. The first such study was conducted by an Italian, Nicols Ghezzi, in 1742, who proposed the following:

Perhaps placing a cast iron vase containing sea water in such a manner that the sun's rays will strike it (and during mild days and seasons, not an insignificant amount of vapor will be formed) and if the spout of the vase is shaded from the
sun, it will result in a more copious and more
extended flow of fresh water.

But the earliest significant solar distillation plant on record

was the one built by Charles Wilson in Las Salinas, Chile, in 1872 [8]. The still was a shallow basin type, with a basin area of 48,000 ft2 and was reported to have produced, when new, 3,900 U.S. gallons of fresh water daily for mules working in the mining operation. Since then till the present day numerous such stills have been constructed and deployed both on a large scale (still in Greece produces about 6,900 gallons/day) and small scale (output of about a gallon/day).

An excellent summary of solar distillation work done around the world from 1872 to 1970 has been presented by Talbert et al. [9].











This survey covered both the large installations (output of about 2-5 x 102 gallons/day) and the small laboratory scale models. He reported following operating and maintenance difficulties:

a. Deposition of salt in the basin for shallow basin stills,

thus requiring frequent flushings and increased maintenance.

b. Breakage of the cover, especially the plastic cover, with

wind.

c. Leakage of brine from the basin and leakage of the vapor

thereby reducing the productivity of the still.

Besides the above, the major drawback of all these systems was

the high initial cost, which made them unattractive for other communities.

In order to overcome the above problems several new designs of

solar distillation units have been experimentally investigated. However, these designs were beset with problems thereby making them unsuitable for large or even small scale usage. Thus, there is a need for a better distillation unit. But before such a unit can be proposed, it is necessary to look at some of the parameters affecting the productivity of solar stills. In addition, such a study will also be useful in analyzing various designs built. The parameters affecting the performance of a still are:

1. Solar radiation

2. Depth of brine in the basin 3. Cover material and its shape

4. Ambient temperature

5. Wind velocity

6. Temperature of the condensing surface.











Solar Radiation

This is the most important atmospheric variable affecting the still output. The amount of radiation absorbed by the brine is directly reflected in the productivity of still. This absorption is governed by various factors, for example, depth of brine, the absorption characteristics of water and angle of glass. These factors are explained below.


Brine Depth

The effect of brine depth on the output of solar stills is plotted in Figure 2. The data in this plot are from a number of stills built all over the world [9]. It can be easily seen that decreasing the brine depth increased the productivity. This was expected since the heat transfer to water in a shallow basin is very rapid, thereby increasing its surface temperature and hence the evaporation. Thus, most of the distillation output of shallow basin takes place during the day; whereas in the case of deep basin still an appreciable amount of solar radiation is absorbed in the water and because of the thermal capacity of the still, the output is over a 24 hour cycle. In addition, the thermal losses from the sides and bottom of the deep basin still tend to be larger than those from a shallow basin still. Several studies, therefore, have been conducted on the shallow basin stills.

Scientists at the University of California have studied the

cascade type still [10]. The depth of the brine was about 1/4 inch. High output rates were obtained (0.13 gal/ft2-day at about 2000 Btu/ft2 solar radiation input) but the still was beset with problems like dry spots in the basin, blistering of black epoxy paint, deposition of salt


































12k




0

.oS



4
4 F"






01 1
0.04 0.06 0.08 0.10 0.12
Distillate (gallons/ft2-day)


Effect of brine depth on still productivity.


Figure 2.











and thus required expensive maintenance, but above all the cost of the unit was high. Maria Telkes [11] has studied extensively wick type solar stills. A porous black material acts like a wick and is soaked by saline water. Since the layer of the water is very thin (1/32 inch), there is rapid evaporation and increase of productivity results (output of 0.16 gal/ft -day at about 2000 Btu/ft -day). However, there are severe problems of salt clogging the pores, dry spots developing resulting in deterioration of the wick material and precise control of brine flow. These problems have resulted in making these stills unattractive for commercial production. Recently ESB Corporation (121 patented a new type of wick material which is claimed to have selfcleaning properties so that it is not clogged by salt. Since no detailed data are available, it is difficult to assess its potential as a commercial unit.

Several investigators have also looked at the effect of floating porous pads on the surface of water so that the brine depth is effectively reduced. Batelle researchers [13] reported an increase in productivity by 10 percent over the similar shallow basin stills, but the system was beset with problems of salt accumulation and dry spots. Recently Szulmayer [14] has reported tests on flotation of black liner with outputs of 0.072 gal/ft -day. No results on salt accumulation or dry spots were reported, however.


Cover Material and its Shape

The cover material has two functions. Besides allowing solar

radiation to pass through it, it acts as a condensing surface for water vapor. The first function requires that the cover should transmit as










much radiation as possible and thus the water film on the underside should be thin. The second function requires that there should be least resistance to the transfer of heat of condensation meaning that the thickness of the cover material should be very small, there should be enough area of condensing surface and the condensation should be dropwise. However, it has been shown [15] that the decrease of solar radiation transmission, because of formation of droplets on the underside of the glass, more than offsets the increase in heat transfer. Thus filmwise condensation rather than dropwise is advantageous.

Normally, two types of cover materials are used: glass or plastic. The advantages of glass are: high transmission for solar radiation, high wettability of glass and relatively high stability of properties over extended periods of time. The disadvantages are its relative poor strength and thus vulnerability to mechanical damage. This requires an increase in its thickness (normally 0.125 inch thick glass is used for cover). Plastic films normally used for solar stills have high solar transmission (an average of 90 percent for 0.004 inch thick Tedlar film at angles of incidence between 0 and 45 degrees, as compared to about 85 percent transmittance for 0.125 inch glass). At the same time these films have high infrared transmittance (80 percent in the wavelength region of 4 pm to 7 pm) and high susceptibility to ultraviolet degradation. It is these properties, together with the fact that they have to be treated so as to make them water wettable, has made use of plastic cover unattractive in solar stills. Thus, till the advent of better plastic material which will overcome the difficulties listed above, it is believed that glass will be mostly used.











The shape of the cover is governed by three factors, namely a) it should provide enough condensing surface for water vapor, b) its angle should be such that it intercepts maximum solar radiation for the whole day, and c) the spacing between the cover and the water surface. Two types of cover designs are normally used. One is the greenhouse type or double sloping cover (Figure 1) while the other has single sloping cover with the south facing side normally having a reflective surface as shown in Figure 3. However, it has been shown, both experimentally [16] and theoretically (next chapter) that if enough condensing area is available then the effect of spacing between cover and water surface is negligible over the productivity. It also appears [17] that the effect of glass inclination on the amount of solar radiation entering the still is small, because the interception area is always horizontal except in the cases of tilled wick. Thus the major factor governing the shape of the cover is the availability of enough condensing area. It is interesting to note that very few investigators appear to have considered this factor in design of their stills. Most have tried to minimize the spacing between the cover and the water surface which ultimately results in low outputs in the absence of enough condensing surface. Ambient Temperature and Wind Velocity

These two variables have been lumped together because they affect

the productivity of the still by their effect on the temperature of the cover. As the ambient temperature increases the amount of heat transfer from the cover to the atmosphere is reduced, thus decreasing the productivity. However, some computer results of Lof et al. [18],
































Solar radiation


Reflective back plate


Glass cover


Black liner


Single sloping solar still.


Figure 3.










Baum [19] and Bloemer et al. [20] have shown that there is an increase in output of the still with increasing ambient temperature. These results are not borne out by actual experiments [16].

Increase of wind velocity increases the outside heat transfer coefficient thereby increasing the heat loss from the cover and the productivity. However, Lof et al. [18], Baum [19] and Bloemer et al. [20] have predicted that increase of wind velocity decreases the output. They found that an increase of external heat transfer coefficient from 1.5 to 10 Btu/hrft 2F decreased the output by 20 percent. Again, this is not borne out by experiments [21], calculations [22] and by computer analysis (present work).


Temperature of the Condensing Surface

The temperature of the condensing surface is a function of both the brine temperature and the ambient conditions. A greater difference between the brine and cover temperature results in increase of the productivity of the still. However, during the daytime, because of higher brine temperature this difference is not maximum. Several investigators have tried to increase this difference by providing an external condenser for water vapor. Grune et al. [23] conducted extensive tests on blowing air through the still and condensing the water vapor on an external water-cooled condenser. Dunkle [24] investigated a multiple effect solar still where the heat of condensation from the first effect is used to provide the heat of evaportioi in the second effect and so on. The condenser of the last effect was cooled by water. Salam and Daniels [25] used two inflated plastic tubes with the inner tube partially filled with salt water over which the











air was passed. This water vapor laden air was then taken to an external condenser where the water was condensed. The reported still efficiencies (efficiency = energy used in evaporating water were total solar energy input to the still' between 20 and 40 percent on bright summer days. Besides low efficiencies this system suffered from another drawback, the deterioration of the plastic tubes.

Although production as high as 0.22 gal/ft -day on a bright summer day was reported by Grune et al. [23], the additional costs required in equipment for air blowing and water circulation more than offset the gain in yield. Similar conclusions were also reached by Dunkle [24], who also reported the problems of corrosion of the metal condenser by saline water. There are other factors also which affect the output of solar stills. They are good still construction so as to have maximum vapor tightness; no or minimal seepage of brine from the basin; good insulation of the sides and bottom for shallow basin stills; and the frequency of flushing and the orientation of the still [26].

From the above considerations of various parameters affecting a

solar still, some of the desirable features of a good distillation unit are:

a. To ensure maximum absorption of solar radiation near the

surface of brine.

b. Cooler condensing cover temperature.

c. Minimum amount of maintenance and flushing.

In the present study a design of a still has been investigated

which incorporates the above features. By mixing water soluble dyes in the saline water, solar radiation can be absorbed in a thin ( 1 inch)










surface layer of brine. At the same time if the basin is made deep, then the still will also work during nighttime, thus using the advantage of cooler condenser surface. The problem of salt buildup (and thus periodic flushing) is also reduced because it settles at the bottom of the basin and does not interfere in the solar radiation absorption near the surface.



Use of Dyes to Enhance Solar Evaporation


The use of water soluble dyes to enhance evaporation of water is not a new concept. Block et al. [27] were the first to study the effect of 2-Napthol Green dye on the evaporation of brine for large scale salt production. They tested this dye in different depth ponds and for different concentrations. It was found that if enough dye was added (20 ppm by weight) the depth of the pond does not have too much effect on evaporation (the difference in evaporation was less than 5 percent for brine layers of 20 cm and 50 cm deep). They did report that with the above concentration of the dye the increase in evaporation of 19 percent over that of uncolored brine was achieved. However, no attempt was made to study the effect of other dyes, the effect of sunlight on the degradation of the dye or to correlate their experiments with analytical models. The use of dyes for salt production subsequently became an accepted practice. However, very little data are available on tile parameters that affect production because most of tile Information is proprietary in nature [28].

Other means of coloring the brinehave also been tested with limited success. The use of red bacteria Halobacteriwn or blue green algae











Spirulina has been tried to enhance the evaporation of brine [29]. The advantages of such a system are obvious. There is no solar degradation of the bacteria and at the same time the culture can be obtained free from the ocean since it exists in nature, thus avoiding the use of costly dyes altogether, which are not anyway recoverable. However, these bacteria are extremely susceptible to changes of salt concentration and temperature and have not been therefore commercially successful [30].

Keyes et al. [31] have extended Block's work to include other dyes and tested their degradation by sunlight. Among the six dyes tested Napthol Green was the most stable in the presence of salt solutions and sunlight for the test period of 6 months. However, conflicting reports were published on the effect of some of these dyes on solar evaporation. It was reported that black Nigrosine and Congo red dyes do not increase the evaporation of brine, while Napthol Green increased it by 7 percent as compared to 19 percent increase achieved by Block et al. [27]. Keyes et al. [31] developed an analytical model of the system. However, the model failed to take into account the spatial and time variations of temperatures of the brine.

From the brief description of the effect of dyes on solar evaporation it is evident that sketchy data are available on their use with no study done on their use for solar distillation. Besides this lack of experimental data no detailed analysis has been conducted on the effect of dye on solar evaporation and on spatial and temporal variations of the dye-brine system. The present work is therefore an attempt to study some of these effects.










Scope of the Present Investigation


The present investigation is an attempt to achieve the following objectives:

1. To develop a simple analytical model for study of the effect

of different parameters on productivity of solar stills.

These parameters are ambient temperature, wind velocity, type

of dye and concentration. The model will be developed for

prediction of temperature-time history of the dye-water

system and the glass cover for different environmental conditions and input solar radiation. The model is developed in

Chapter III.

2. To experimentally evaluate the effect of different dyes on

productivity of deep basin solar still. The productivity

from this still is compared with that from an identical

control unit. Chapter IV presents the experimental setup and

procedures.

3. To obtain experimentally the absorption spectrum of the dyes

used in this study. The dyes used are Napthylamine (Black), Carmoisine (Red) and mixture of Neptune Blue, Carmoisine and

Tartrazine (Dark Green).

4. To compare the results obtained from this experiment with

that from the analytical model. This comparison is presented

in Chapter V.


















CHAPTER III

THEORETICAL ANALYSIS



In this chapter a theoretical analysis for a deep basin distillation unit is presented.



Physical Model


Figure 4 is a sketch of a deep basin distillation unit. It shows the energy attenuation of a ray from the sun as it passes through the condensate covered glass and into the dye-water mixture. In order to analyze the energy transfer process in the basin the dye-water system has been divided into discrete layers of thickness Ax. An energy balance is then given for these layers after making the following assumptions:

1. The dimensions of each layer, as shown in Figure 4, in the

y-z direction are large compared to those in the x direction
,
and B. � 1 , so that the temperature varies only in the x
1

direction.

2. In the temperature-range normally encountered in the solar

distillation units, the changes in physical properties of the




Biot number (Bi) is defined later in the text.



























Reflected from
letdfrom

glass
Convection






~in 1

n-l n z


-rxI n *


Dye-water solution

_ b


Plexiglas basin


Figure 4. Model of still used in analysis.


Aluminum cover









I

x










solution like viscosity, density, specific heat and thermal

conductivity are negligible.

3. The spectral emission and scattering coefficient of the dyewater mixture is assumed to be small and thus neglected.

4. The attenuation of the incoming solar radiation by the glass

with condensate film inside has been taken into account by

averaging out the reflection and absorption losses with

respect to angle of incidence.

5. The bottom surface of the still is assumed to be a black

absorber.

6. There is negligible attenuation of solar radiation as it

passes through the air-water vapor mixture inside the still.
th
With the above assumptions, the energy balance on the n layer of thickness Ax then leads to the following: Rate of increase of 1
aten o incred [Rate of energy in] - [Rate of energy out](la)
L energy stored I

th
where energy in to the n layer comprises of:

ate of energy Rate of solar radiation + Rate of energy in
in absorbed in nth layer by convection from layer n+l



LRate of energy in by
+ conduction from layer n+l or n-i

(lb)


and rate of energy out can similarly be written as:











te of energy Rate of energy out by conduction to





Rate of energy out by Rate of energy out by conduction
+ convection to layer + from the sides to environment
n-i
(2)


At the upper boundary (air-water interface) the energy is transferred to the glass cover by convection, radiation and evaporation. Thus, the instantaneous energy balance on the surface leads to:


0 = [Rate of energy in] - [Rate of energy out] (3a) where

[Rate of energy in] = Rate of energy in by eonductio from layer 2


+ [Rate of energy in by convection from layer 2]
(3b)

and
[Rate of energy out Rate of evaporation of water]
[ R a t e~~~~ o e n r y o t = fr o m t h e s u r f a c eJ


+ [Rate of radiation loss to the glass cover]


+ [Rate of convection loss to the glass cover] (4)



At the lower boundary, namely the base (layer b), the energy balance is: [Rate of energy in] = [Solar radiation absorbed in this layer] (5a)










[Rate of energy out] =



+ [Rate of energy + [Rate of energy + [Rate of energy


Rate of energy loss by convectio to b-i layer



loss by conduction to base] loss by conduction to layer b-1] loss by conduction from the sides]


(5b)


Finally, the energy balance on the glass cover yields:


[Rate of energy in] = [Solar radiation absorbed by glass cover]



+ ate of energy in by convection from the dye-water] + surface

+ ate of energy in by radiation from the dye-water + surface I


+ Rate of energy
surface


in by evaporation from dye-water


[Rate of energy out] Rate of energy loss by convection1 [and radiation to the environment]



Therefore,


VRate of] Rate of Rate of increase of energy energy in [energy out] stored in the glass cover]


(6)



(7a)


(7b)











Incident Solar Radiation

The incoming spectral solar radiation reaching the dye-water

surface is considered as having wavelengths in the range of 0.3 pm to

2.0 pm. The choice of this range has been dictated by the fact that most of the radiation after 2.0 pm is absorbed by the condensate film on the glass surface. Furthermore, to facilitate computation, the solar spectrum [32] has been smoothed by fitting a polynomial in the above range. The radiant flux then can be written as:



Btu (2379X - 713.7; 0.2 pm < A < 0.5 pm
't hr Pm t1261.2 exp(-l.95X); 0.5 pm < X < 2.0 Pm (8)



The spectrum given by equation (8) is the maximum radiation

available at the solar noon on a clear day, and thus will have to be modified for different times of the day. This modification is achieved by assuming that the "shape" of q, remains the same but the area under the curve of q. vs A becomes equal to the total solar radiation falling on a horizontal surface at time t during the day.

If the solar radiation incident on a horizontal surface at time t is given by q(t) then q,(t) becomes: qX(t) =(t) qA (9) where

C(t) = _Q<)_ (1-0)
2.0

0.3 qXdX










Absorption of Radiation

The incident angle of solar radiation on the dye-water surface changes with time, as shown in Figure 5. This change in the angle of incidence changes the absorption thickness for any layer of thickness Ax.

Thus, from Snell's Law,


SinG. n2
Sin r n1(



where nI = index of refraction for air-water vapor mixture and n2 = index of refraction for water-dye mixture.

Since the amount of water vapor, as compared to the air inside the still, is small it is assumed that n1 1. Also it has been assumed that the index of refraction for dye-water mixture is the same as that for pure water and both n2 and n1 remain constant for the range of wavelengths for 0.3 pm to 2 pm.

From Fresnel equation the reflectance can be calculated by knowing

0. and 0 and is given as:
1 r


1 sin2(8i -0) Cos 2( +
r 2 1 1 1+ - (12)
sin( + 0r) cos (0i - Or)




and for 0i 0


2
n - n
2 1
r n2 + n : (13a)






26



















Solar radiation q0
q i(t) Reflected beam ' Water-air interface




r

S Dye-water
solution


4 A

Layer n Ax* A


\ n+l q (t)


Figure 5. Radiation transfer in the dye-water solution.










Thus, the amount of solar radiation passing through the water-air interface has to be multiplied by the transmittance of both the waterair interface and condensate covered glass. For the condensate covered glass an average transmittance value T g/w has been obtained from experimental data. Therefore, the radiation passing the water-air interface is:

0
qX,i(t) = Tg/w(l-r) q,(t) (13b)


where q%(t) is obtained from equation (9), and r is given by equations

(12) and (13) above.

The path length of the incident radiation is:


Ax* = Ax/cosr (14)


where Ax* is the path length of the ray of incident solar radiation for a layer of thickness Ax. As can be seen from equation (14), Ax* is also a function of time.

The absorption of incident radiation in the dye-water medium takes place in accordance with the Bouger's Law [331. Thus, the amount of energy absorbed in a layer of thickness Ax* is:


qab(t) = f 2 n %2 n
q= q ni(t) dX - fI q ni(t)[exp(-a Ax*)Id% (15)


n

where qi (t) is the incident spectral radiation on the layer n and is the spectral absorption coefficient of the medium. The absorption coefficients of different dyes have been obtained experimentally and will be discussed in greater detail in Chapter V.










Conduction and Convection Heat Transfer in Dye-Water System

The heat transfer in a layer of thickness Ax occurs by both conduction and convection from adjacent layers besides the solar radiation absorbed by it. The conduction heat transfer is easily modeled while for convection the concept of equivalent conductivity [34] is used. Thus, the average heat flow per unit time because of convection is characterized by equivalent conductivity k defined as:
e

k
q = (TH - T) (16)



where Z = thickness of layer TH, Tc = temperatures of hot and cold layers, respectively. The data for equivalent conductivity for plane layers from different investigations has been plotted [34] and fall on a straight line, given by:


Fk ]f= 0.3 log10 Ra/- 1; for Ra. > 2xI04
log10 L = 0.25 log10 Ra -0.7; Rat < 2x104 (17)




where k = conductivity of the water
w

Ra = Raleigh number based on thickness � of the layer

= Bg (TH - Tc)Z3
H ct
ww

= thermal expansion coefficient of water

g = acceleration due to gravity

V = kinematic viscosity of water
w
a= thermal diffusivity of water.
w






29



Heat loss from the sides of the layer to the environment is:


qloss,sides = Ax(Tn - T) UP (18a)



where T and T. are temperatures of layer n and the environment,
n

respectively. The factor UxP is the product of overall heat transfer coefficient U between the water-dye solution and the environment, and the perimeter P of the still. Equation (18) is a one-dimensional model; it implies that Biot number (Bi) be much less than 1 where B.
1 1

is

B� = UA (18b) I kP
w

where A is the area of the layers. The value of B. for the present
1

case is 0.19. This makes the one-dimensional assumption reasonable. Evaporation and Convection Loss from the Water Surface

The mass transfer inside the still occurs mainly by diffusion and convection of the air-water mixture. Baum and Bairamor [35] have done interferometric studies on a laboratory still and found that most of the energy transfer by evaporation of water and convection from the water surface takes place in a boundary layer which circulates inside the enclosure. The bulk of air at the center of the enclosure remains essentially stationary. Thus, the evaporation rate is determined by the analogy between heat and mass transfer for free convection.

The heat transfer by convection from the water surface to the glass cover is given by:


qc = hc A (Ts - T )


(19)










where h is the heat transfer coefficient and T is the average glass
c g

cover temperature. There is spatial temperature variation in the glass cover because of the geometry of the still but for the sake of simplicity an average temperature between the top and bottom portion of the glass has been used.

The heat transfer coefficient h can be calculated from the
c
following nondimensional equation [36]:


Nu = C (Cr Pr)p (20)

=h L
where Nu = Nusselt number = c kw/a


Gr = Grashoff number = L3 (Ap/p)
V
m
1-c
and Pr = Prandtl number = w/a


C = Constant

The choice of exponent p and constant C in equation (20) is dictated by the Grashoff number. In the temperature ranges encountered in the solar still the exponent p has the value of 1/3. It should be noted that h from equation (20) is independent of length L (distance between
c

the water surface and the glass cover) for most stills. Experimentally the above fact is not borne out because there is heat and mass transfer to the glass cover on the sides also, which changes the free convection pattern.

The bouyancy factor (Ap/p) in the Grashoff number of equation (20) was evaluated for the conditions of constant composition of the convecting fluid. However, this has to be modified for the solar still






31



since in this case the humidity of air changes along the flow path. This modification is achieved by assuming that air-water vapor mixture is an ideal gas [36].

Thus,

P M P.M. Pw/a Pw + Pa w w + air air (21) RT RT



where pw/a density of water-air mixture

P = partial pressure of water vapor
w

P air = partial pressure of air

and M wMair= molecular weights of water and air, respectively

R = universal gas constant

T = temperature of the mixture

For an enclosed system,


w air Ptotal (22) In the case of a solar still, since it is open to the atmosphere Ptotal = P atm. Therefore, equation (21) is modified to: P M (Pat -P )M. = w w + atm w air w/a R T R T


Thus, the density of water-air mixture near the glass cover and at the water surface can be respectively written as: Pw M (P -P Pg)M. = wg w + atm wg air (23)
w/a 1g R T 9R T











P M (Patm -P ) M Pw/a Is s T + R T
s S


while


Ap= Pa/w g-Pa/w Is


and Pa/w is the arithmetic means of pw/a Is and pw/a Is After algebraic manipulation:



AP = 2 (l-x) Pl+x


where


T x=-- T
T S


(25)


Sws + (Patm - ws (Mair /Mw I Pwg + (Patm - wg) (Mair/Mw)


and h from equation (20) then becomes:
c



S= C k g(Ap/p) p) 1/3
h w/a m k
c m kw/a



The mass transfer coefficient is defined as:


J = hm(iis - i.e )


(26)


(27)


where,


J

h=
m
(mi. - mie) = is i


mass transfer rate per unit area mass transfer coefficient, mass/area - time driving potential for the mass transfer, dimensionless


(24)












m. = Pi = mass fraction of species i; (where subscript s implies
I.
P
near the surface and e, in the environment)


From analogy between heat and mass transfer, an equation similar to equation (20) can be written for mass transfer. Thus,


Num = C (Grm S )P (28)


where

Gr = Grashoff number for mass transfer
m
h L
Nu = mass transfer Nusselt number = m
m
p D..

p = average density of the mixture

Dij = binary diffusion coefficient or mass diffusivity

S = Schmidt number = m
c
D..
ii

V= kinematic viscosity of the mixture
m


The above analogy is based on the following assumptions:


(i) Small rate of mass transfer

(ii) No chemical reactions in the fluid

(iii) Negligible viscous dissipation

(iv) Negligible emission or absorption of radiant energy

In a solar still there is combined heat and mass transfer from the surface of water. For such a case the ratio of Nu aid Nu proves to be functions of Pr, Sc, and the ratio of bouyancy induced by the temperature difference to that induced by the mass concentration difference. This latter ratio is z 1 if Pr - Sc, which indeed is the case










for water evaporating in air. Thus the problem is simplified in the sense that mass and heat transfer can be treated separately. Hence,


Nu = C (Gr Sc)1/3
m


and h is given as:
m


h = p D.. c(g(APlP))/3 m Dij D..
m ij


(29)


(30)


and heat exchange because of mass transfer is:


qm = hm A(m. - mi.) h
is i hfg


(31)


where h
m A=

(mis-m ig

hfg


mass transfer coefficient area of the still mass transfer driving potential enthalpy of vaporization


Thus, from equations (24) and (28) the mass fraction of species is:


P M P M (Pat - P ) M. = Pis ws w ws w + atm ws is R T R T R T P s s s


P M
m = ws w is P m + (Pa ws w atm


(32)


ws air


Similarly,

P M
m.- wg w
iwg P M +(P - P) M
wg w atm wg air


M a air)


(33)











Formulation of Equations


The difference equations have been written for different sections and have been thus identified.


Layer n

Combinations of equations (1) and (2) give:

A A kA
A 2 qn (t)dX - f 2 qn (t) [exp(-a Ax*) I dA} - k e A (rt - Tt)
A q,i 1 ,i A Ax n n+l


kA
e A t
+ Ax [n-1


- T - [Tt - Tt] UP Ax = p A C Ax
n n W p


Tt+At - Tt] n n 34) At


where


k = apparent conductivity given by equation (16)
e

A = area of the layer

In the above equation if there is no convection then k simply becomes
e

equal to k. The boundary conditions can be written as:


Top Layer

Combining equations (3) and (4) give the temperature of the top layer. Thus,


k A (Tt-Tt) e (A2) = h A(m.
(Ax/2) m is


- m. )h + h A(Tt - Tt ) + OF F A[T - T 4
ig fg c s g e(a s g
(35)


where

F = emissivity shape factor between water surface and glass
e
cover










F = geometrical shape factor between water surface and glass
a

cover


Bottom Surface

An instantaneous energy balance on bottom surface yields:



A { f2 q,0 (t) [exp(-a Z*) dA} =



keA(T - T_) k A(T - T t
e b b-_ pg b bout)
(Ax) + Ax (36) Pg

where
0
qi(t) = the solar energy incident on the water-air interface

Z* = transmission length for the whole system

as shown in Figure 4

k = conductivity of Plexiglas
Pg
Ax = thickness of the Plexiglas
pg
Tb,out = temperature of the Plexiglas bottom


Inherent in equation (36) is the assumption that the bottom surface is a black absorber so that all the energy incident on it is absorbed.


Energy Exchanger with the Still Covers

The right-hand side of equation (35) gives the rate of energy into the glass and aluminum cover. This energy must be given up to the environment. At the same time there is a small amount of solar radiation absorbed by the glass cover. Thus, the temperature of the cover can be obtained from the following equation:











OF F A [ T-4 T ] + hmA(m i )h + h A(Tt - Tt) + a A q(t)
e a s g - is ighfg c s g gg


t t 4 4 Tt+At -Tt
-h A(T -T) OF ec a -T T sky c A Ax ( g - g) oc e ac g kyggg g A



+ t+At - Tt
al Cpal al al a Aal (37)



where

A = area of still cover = A + Aal
c g a

A = glass cover area
g

Aal = aluminum back plate area

a, = solar absorptivity of glass

h = outside convection heat transfer coefficient
0

Pg' Pal = density of glass and aluminum, respectively

cpg c pal = specific heat of glass and aluminum cover, respectively

and Ax , AXal = thickness of glass and aluminum cover, respectively

Tsky = effective ambient temperature for radiation

= T - 150F [37]

and other symbols are defined before.

Experimentally it was found that during daytime the temperature of the aluminum back plate was greater than the surface temperature of water. Thus, no condensation took place on the back plate. At night, however, there was condensation at the aluminum plate but its temperature was equal to that of the glass cover. Moreover, an order of magnitude analysis of equation (37) shows that the storage term in










the aluminum is negligible. Thus, all the temperatures Tal in the above equation can be considered as T . This yields
g


During Daytime


OF F A [T - T + h A (i h + h A(Tt - T) + a A q(t)
e a s g m is - mig fg c s g g g


t t 4 4 t+At t
-h Ag(T - T) - OFeg Fag A (T- Tsky) = Pg c A Ax (Tt+ -T ) oggeg gg g kyggg g g -g At

(38)

During Nighttime


aF F A (T4 - T 4) + h A (m. - h + h A(T - Tt) - h
ea s g in is ig fg c c g a

Tt+At T t
A (Tt- Tt)-aF F A (T 4 T4 A Ax ( g
cg ec ac cg sky gpgg g At

(39)

Note the change in cover areas for day and night in the above equations.




Method of Solution


The depth of the still is 10.5 inches and with grid size of Ax =

1 inch there are 11 nodal equations and two boundary conditions for the top and bottom surfaces. Therefore, a computer program was written to solve these 13 nodal equations for the dye-water, and one equation for the glass cover, simultaneously. Explicit method of solution was used in solving them and therefore the time interval At was carefully chosen so as to avoid unstable solutions. Appendix II shows the stability










criterion used in the present analysis. Below are listed the steps used in getting the temperature-time history of various layers.

1. The temperatures of the layers and glass cover are

initialized at sunrise, using their experimental values.
Ths t an t
Thus, T and T are known. Also used in the program is the
n g
actual solar insolation, ambient temperature and wind velocity.
2. Oce t adt
2. Once T and T are known then equation (35) is solved to
g 2
t
obtain T . The solution of equation (35) is by regulafalsi
s

method [36].

3. Equations (34), (36) and (38) or (39) are then solved for

t+At t+At
T and T
n g
4. Once the temperature vs depth profile is generated from step

3, then the apparent conductivity k is calculated. The
e
method of calculation is as follows:

(a) Calculation of Tnmax , as shown in Figure 6.

(b) All the layers above layer n will have convection

transport with AT = T - T and subsequently k is n,max s e calculated using equation (17) and depth e.
th
(c) The layers below the n layer have heat transfer between

them by conduction only.

5. Steps 2 to 4 are then repeated with an increment of time At,

till a 24 hour cycle is completed.

Figure 7 shows the detailed flow diagram of the computer program used to generate temperature-time history for different layers.

The results of the analytical model are presented in Chapter V.






40















T
s /i
S /
2
3



L...___.





T
n




/T I
/7



T nl ,mlax Temperature of layers (arbitrary units)


Figure 6. Temperature profile of the layers.


















Initialize tmons of all layers & glass
Tt=Tw, . 1 113
t
T = T g g


Initialize time TIME = 0


Subroutine to solve
equation (35) by re. ' lvlasi method




Subroutino to crtlCoI later al~liou t of solar radl a I (,I) ahtsorbed in ]aver n


Calculate Tt from equation (35) Subroutine to
s calculate heat
and ma-s t rans f(,r be t' en water surface & glass COVI].

Calculat" the temperatur's Sti'u lieuc to of di fferetnt lavers using (a1cul1 t the c(quiatiou (31) aP11:11:lilt conit+At d tI ivi ( %, k Thus, "" is known based upon Tn, max


daY ___1,- it day or night? ---P Calculate the tesp. of Vglass Tt4At using nFighIIt I,
equ,ation (38)


Calculate T t+At using equation (39)


Inctenlci time h
At
TIME= TjM[ 4 A]

No P I ime ic

to 2,1 hours?

Yes


rinit theSo Er's' J It 1)


Flow diagram of the computer program.


Figure 7.


















CHAPTER IV

EXPERIMENTAL SETUP AND PROCEDURES



Distillation Units


Two identical deep basin solar distillation units were constructed for testing. One of these was used as a control while the other was used for testing the effects of dyes. The basin of these stills was made of 3/8 inch thick Plexiglas which was selected for its strength and corrosion resistance to salt water. Figures 8 and 9 show two views of the stills. Plexiglas plates were bolted together and then glued using epoxy resin to make them water tight. Before experiments the basins were tested for water tightness. The overall dimensions of each basin were 4 ft. x 4 ft. x 1 ft. These basins were painted on the inside using two coats of Suntec Acrylic Black paint which acted as a black liner for the stills. At the top of the basin, extruded aluminum angle of dimensions 2 inches x 1-1/4 inches and 3/16 inch thick was bolted to serve as support for the glass cover frame.

The glass cover frame was made of aluminum extrusions, shown

schematically in Figure 10. To this was bolted aluminum channels of

1.5 inches x 0.5 inch and 0.125 inch thick, which acted as disLillation troughs. This trough also provided support for the glass cover. A thin cotton cloth wick covered the distillate channels in order to




































Experimental stills.


Figure 9. South view of the still.


Figure 8.


















Aluminum frame


2.5'


Schematic of the glass-cover frame.


Figure 10.










enhance the distillate flow. The cover frame was welded together using Argon-arc welding. This removable cover facilitates in changing the dyes and giving access to the interior of the basin. The glass angle was kept 30' to the horizontal and facing south. This angle was decided so that a) there should be sufficient condensing area and b) the still should intercept maximum radiation for a fixed angle throughout the year.

The south facing glass cover was double strength (1/8 inch thick) pane with dimensions of 5 ft. x 2 ft. Two of these panes were used for the cover. The side glass covers were two right angle triangular panes with hypotenuse as 5 ft.

The back plate was made of aluminum sheet (1/45 inch thick) and

was bolted to the glass cover frame. This plate acted as reflector for solar radiation as well as condenser surface.

The glass panes and the back plate were then sealed to the frame by silicone rubber sealant to make the cover vapor tight. The vapor tightness of the cover was then checked by passing water from a rubber hose over it and leaks, if any, were then again sealed. In order that the still be vapor tight after the cover is put over the basin, a
R
ribbon of sealing compound Permagum was run the entire length of the basin top. This compound remains pliable over the temperature range of

-20*F to 200*F and thus acts as a good sealant.

The water level in the stills was maintained by a constant head feed tank, shown schematically in Figure 11.

The basins were insulated on the sides by a polyurethane slab (Thermax R) which was 3/4 inch thick and dimensions of 1 ft. x 4 ft.




















City water supply


To drain


Figure 11. Schematic of the constant head feed system.











The bottom of the basin was insulated by two 1-1/2 inch thick foam glass slabs each having dimensions of 2 ft. x 4 ft.



Thermocouples and Their Locations


Copper-Constantan thermocouples of size 24 gauge (0.02 inches) were used in all temperature measurements. Temperature measurements were made at various locations as shown in Figure 12. These thermocouples were attached on outside of the Plexiglas basin by a paste of Plexiglas chips and Ethylene dichloride, while on the glass cover they were held down firmly by transparent scotch tape. This tape was over an aluminum covered styrofoam piece pressing down on the thermocouple as shown schematically in Figure 13. Besides these locations thermocouples were attached on the outside of the insulation around the basin.

To measure temperatures of dye-water in the still, five thermocouples were attached on a Plexiglas rod at various distances as shown in Figure 14. The thermocouple junctions were coated with a very thin coating of shellac to make them corrosion resistant.

These thermocouples were calibrated, together with the temperature measurement recorders, at four temperatures, namely the ice point, and three other temperatures between 320F and 150�F.


Temperature and Solar Radiation Measurements


There were about 30 thermocouples attached at different locations on both the stills and this necessitated the use of multiple point

































Figure 12. Thermocouple locations on still.


Thermocouple Locations

0 On outside of plexiglass basin A On outside of glass and back plate
































Thin shiny aluminum foil

Thermocouple bead


Scotch tape


Styrofoam piece


Figure 13. Attachment of thermocouple on glass cover.

























































Figure 14.


Location of thermocouples in dye-water solution.











recorders. Two such recorders, one made by Honeywell and the other by Leeds & Northrup, were used. The Honeywell-Brown electronix recorder had 20 points with accuracy of � 0.50F and range of 0-150'F. The Leeds & Northrup speedomax recorder was 24 point with an accuracy of � 1.0'F and range of 0-200'F. These recorders were connected to the mechanical timer so that the temperature readings could be taken after every 1 hour interval.

The solar radiation data was measured by Eppley Phyrheliometer,

Model No. 10, and recorded on a single point Leeds & Northrup speedomax recorder with accuracy of � 1 percent of the total scale. The ambient temperature readings were obtained from the hygrothermograph at the Solar Energy Laboratory. This laboratory was about 10 miles from the site of the present experiments.



Absorption Spectrum of Dyes


The dyes used in the present study were water soluble and supplied by GAF Corporation. They were:

1. Napthylamine 10 BR (Black)

2. Carmoisine BA Ex (Red)

3. Mixture of 33 percent by weight of each of the following

dyes: Neptune B.R (Blue), Carmoisine BA Ex (Red) and

Tartrazine C Extra (Yellow). This mixture resulted in Dark

Green dye.

The absorption spectrum from wavelengths of 0.3 Om to 0.75 pm was obtained using a Beckman UV-VIS Model No. 25 spectrophotometer as shown in Figure 15, where (A) is the spectrophotometer and (B) is the strip

















































Figure 15. Beckman UV-VIS spectrophotometer.










chart recorder for output data. The specimen cell was a standard quartz rectangular cell with optical thickness of 1 cm. The absorption data for wavelengths from 0.75 pm to 2.0 pm were obtained using a Perkins-Elmer Model E-1 Monochromator as shown in Figure 16 with (A) the monochromator and (B) the strip chart recorder with the controls. The specimen cell for this instrument was designed for the present study (see Figure 17). It consisted of an aluminum split ring I inch in diameter and 1/4 inch in length and the thickness of the ring being 1/16 inch. This ring was sandwiched between 2 glass windows each of diameter 1 inch and thickness of 1/8 inch. To make the assembly water tight rubber gaskets (1/32 inch thick) were put between the aluminum ring and the glass windows. This cell was held in place by aluminum supports, which were bolted to each other. Figure 18 shows the crosssectional view of this cell.

The Beckman Spectrophotometer is a differential device wherein the spectrum of the specimen is continuously compared with that of distilled water. Thus, it gives the absorbance spectrum [Absorbance = log (Incident radiation/Transmitted radiation)] directly. However, the Perkin-Elmer Monochromator is not a differential device and thus separate readings of the transmission spectrum of cell only, cell filled with distilled water and cell filled with dye sample, were taken. The absorption coefficient a was then calculated from these readings.



Experimental Procedure


The two stills were set side by side separated by a distance of about 3 feet, as shown in Figure 19. They were filled with tap water





































Perkin-Elmer Monochromator.


Figure 17. Specimen cell.


Figure 16.

























Aluminum supports


Rubber gaskets Glass windows Aluminum split ring



Bolts


Figure 18. Cross section of the specimen cell used in
Perkin-Elmer Monochromater.



















































Figure 19. Arrangement of the two stills.












to the depth of 10.5 inches and the level was maintained by the feed tank. Common salt (composition unknown), 3.5 percent by weight, was added to the two stills. In one of the stills, dye of known concentration was added. The covers were then put over the stills and allowed to stand for a day or two to attain a steady periodic operation. After this period of time the distillate was collected every 24 hours and the weight of the distillate was measured by a cantilever type balance with an accuracy of � 0.1 lbs. The temperature measurements of various locations were recorded on the strip chart at every 1 hour interval.

Three dyes were tested with concentrations of 50 ppm and 100 ppm for each except for the Black dye where the testing was done for 50 ppm
,
and 172.5 ppm concentrations. The duration of testing for each dye is given in Table 1.

On a number of occasions the distillate was collected every hour so as to study its hourly variation. This also helped in calculating the constant C in equation (20), as will be shown in Chapter V.

The absorption spectrum of the dyes before and after the duration of experiment was also obtained so as to study their degradation by sunlight. Also evaluated was the absorption spectrum of the distillate to study carry over of the dyes by water vapor.

The results of the experiment are presented in Chapter V.








,
The maximum concentration of Black dye was supposed to be 150 ppm but the resulting mixture inadvertently resulted in 172.5 ppm concentration.























Table 1
Duration of Test


Concentration
Dye (ppm) Duration of Test Carmoisine (Red) 50.0 11/10/77 - 12/8/77 100.0 12/10/77 - 02/14/78 Napthylamine (Black) 50.0 02/16/78 - 03/28-78 172.5 03/29/78 - 04/24/78 Dark Green 50.0 04/26/78 - 08/13/78 100.0 08/15/78 - 09/21/78

















CHAPTER V

RESULTS AND DISCUSSIONS



Dye-Absorption Spectrum Results


The absorption spectrum of different dyes is shown in Figures 20 through 25 where the absorption coefficient a is plotted against wavelength X. The absorption coefficient is a function of concentration of the dye and the temperature and pressure of the system. In the present analysis the temperature and pressure effects are ignored.

Figure 20 shows the absorption coefficient for 50 ppm Carmoisine (Red) dye* taken both before and after the experiment. This dye shows the dip in the absorption curve between 0.6 and 0.7 pm which is the red part of the spectrum. Similarly in Figures 22 and 23, which show the absorption spectrum of the Black dye, the dip in the value of ct occurs between 0.4 and about 0.475 pm which is the blue region. Thus, this dye, strictly speaking, is a dark blue dye and not black. The Dark Green dye, as has been pointed out before, was made up of a mixture of Neptune Blue, Tartrazine (Yellow) and Carmoisine (Red) dyes. The dip in the absorption curve in the visible region is at 0.5 pm which is the green region of the spectrum (see Figures 24 and 25). Also shown in Figure 22




*Appendix III shows the calculations for obtaining a from experimental data.

























Figure 20. Absorption spectrum of Red dye (50 ppm).


















Before experiment After experiment


I - J
1.9 2.0 a"


0.9 1.1 1.3
Wavelength X(wm)


.LU
4-4




-4

4-4 CH


0

0

"4

0 U)





25


0 L 0.3


0.5




























Figure 21. Absorption spectrum of Red dye (100 ppm).








140


Q Before experiment 100 After experiment



4.a



< 75
4-a



4-4 4-4 C)
50

0
41i
o) ,0

25








0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.0 0

Wavelength X(pm)w





























Figure 22. Absorption spectrum of Black dye (50 ppm).



















Q Before experiments


After experiments


7 Distilled water .0



75


44 C)
0 .C-)

u

0 50-0

-o
.14





25







0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2. 0 Wavelength X(pm)























Figure 23. Absorption spectrum of Black dye (172.5 ppm).













0 Data point


250






200






150




4j
u

100
0

0
4j

0
50






0
0.3


" II I L
1.1 1.3 1.5 1.7 1.9 2.0 Wavelength A(pm)


0.5 0.7

























Figure 24. Absorption spectrum of Green dye (50 ppm).








140 r125 Q Before experiments


5 After experiments 100


41i


z 75 (U


0
,4 (U
0 o 50
42

0



25


� = = - I _____I____0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.0 o

Wavelength X(pm)























Figure 25. Absorption spectrum of Green dye (100 ppm).







140



125 100


,.3 1.5 1.7 1.9 2.0


Wavelength X(pm)


0 Before experiments nj After experiments


0.3










is the absorption coefficient of distilled water. As can be seen, all the dyes tested in the present study have aX very nearly the same as that of water above 0.7 pm. This is to be expected, since above 0.7 Om wavelength region most of the energy of the incoming photon goes to the rotational, vibrational or translational modes of the energy of the dyewater complex. Since the amount of the dye is small (~i00 ppm), the absorption spectrum of the dye is very nearly the same as that of water. However, in the wavelength region of 0.7 pm to 0.9 pm there is still enough energy (1.55 eV) in the photon to produce higher a X by changing the rotational-vibrational states of a suitable dye-water complex. The dyes tested in this study did not show this change, and thus there is a need to develop dyes which will have high a X in this region since it contains about 25 percent of the total solar insulation.

The degradation of the dyes by sunlight was studied by comparing their absorption spectrum before and after the experiment. Figure 20 shows a X for Red dye at the end of the experiment. As can be seen there is severe degradation of this dye resulting in reduction of a by 95 percent at a wavelength of 0.5 vm. This also resulted in the change of the color of the dye from bright red to an almost colorless liquid. The reason for this degradation appears to be photolysis caused by ultraviolet radiation in the solar spectrum since a part of this spectrum (ultraviolet region from 0.3 vm to 0.4 pm) is transmitted by the condensate covered glass cover. This can be understood more clearly by examining the molecule of the Carmoisine (Red) dye shown below,

HO
NaO3 S N = N


NaO 3S










where the color of this dye is because of -N = N- chromophore [391. The bond strength of N=N is 3.45 eV [401 while the energy of the photon in 0.3 urm wavelength is 4.13 eV and thus the degradation of dye is inevitable. However, no one single mechanism explains the fading of dyes and involves three photochemical reactions--oxidation, reduction and photolysis. In the case of Carmoisine it is possible that both photolysis and oxidation reactions are possible for its fading. One possible indication is the change in pH of this dye, in 100 ppm concentration solution, from 5.2 (before experiments) to 6.7 (after experiments).

A similar degradation is found in the Dark Green dye as shown in Figure 24, where the degradation is primarily a result of the fading of the Carmoisine dye. The absorption coefficient of this dye drops by about 85 percent at 0.5 pm. This resulted in the change of the solution color from dark green to transparent green. The other constituents, namely Tartrazine and Neptune Blue, appear to be unaffected.

The most stable dye (among those tested) from the solar degradation point of view appears to be Black Napthylamine. For the duration of the experiment, its a did not change as can be seen from Figure 22. The dye has the following structure: HN OH

O 2N C =_N -N N ==N


Na03S S03Na


It appears that the OH group together with NO2 is responsible for lightfastness of this dye [41]. However, no satisfactory theory exists











yet which correlates the lightfastness with its chemical composition and most of the information available is yet empirical.



Temperature-Time History of the Stills


The measured temperature-time history of the various dye-water systems are plotted in Figures 26-31. As can be seen from these figures, the general trend is similar and thus it would be sufficient to explain one of these figures. Figure 27 shows the temperature-time history of Black dye (50 ppm) on March 6, 1978. The top layer temperature reaches a maximum of about 110*F around 2:30 p.m. (EST) whereas the maximum solar radiation is reached around 12:40 p.m. (EST). This lag in the top layer temperature is because of the mass of the dye-water system, the ambient temperature and the conduction of heat from this layer to the bottom ones. The glass temperature also follows the similar trend.

The back aluminum plate temperature, however, rises very rapidly and reaches a maximum of about 130'F at 1:30 p.m. (EST) (Figure 27). As was mentioned before (Chapter IV), this plate's purpose was to reflect incident solar radiation onto the solar still and at the same time act as a condenser for water vapor. However, because of rapid oxidation of the plate the reflection property deteriorated rapidly thereby increasing its solar absorptance and hence its temperature. Moreover, it can be seen from Figure 27 that the plate temperature is greater than the water surface temperature till about 4:30 p.m. (EST) after which it rapidly approaches the glass cover temperature. Thus




























Figure 26. Temperature-time history of Red dye (50 ppm); November
26, 1977.









Q Top layer temperature

7 Glass cover temperature A Bottom layer temperature nAmbient temperature �Solar radiation


80 F-


40- - V


2 __ _ _ _ __ /__0 1:30 5:30 9:30 1:30 5:30 9:30
A.M. P.M. A.M.

Time, hours (E.S.T.)


300

0
200 100

a
0


100 -
























Figure 27.


Temperature-time history of Black dye (50 ppm); March 6, 1978.




Temperatures


130 120






100





' 80



.4

CU
S60






40






20
5:30


o Back plate 0 Top layer
Glass cover Bottom layer [I Ambient Solar radiation


9:30
A. M.


1 _ l_____A __ [ __ LL 1:30 5: 30 9: 30 1:30 5: 30 9:30
P.M. A.M.
Time, hours (E.S.T.)


400
44

300

o 200 *H
4J


100 P
Ca ,-4 0 0 Ln





























Figure 28. Temperature-time history of Black dye (172.5 ppm);
March 31, 1978.







Temperatures


Top layer
Layer 1 1/2 below air-water interface Layer 4" below air-water interface Layer 7" below air-water interface Bottom layer

See Figure 31 for solar radiation data


- -~ - ________-


5:30 9:30
P .M.
Time, hours (E.S.T.)


1:30


5:30


9: 30 A.M.


140 r-


120 ;-


100


60 1-


5: 30


9 30 A. M.


1:30
























Figure 29. Temperature-time history of Green dye (50 ppm);
May 15, 1978.














Top layer temperature Glass cover temperature Bottom layer temperature Ambient temperature See Figure 35 for solar radiation data


v


601-


5:30 9:30 1:30
A.M.


5:30 9:30
P.M.
Time, hours (E.S.T.)


1:30


5:30 9:30
A. M.


140 -


120 _


o 100 801


M M M M
























Figure 30. Temperature-time history of Black dye (172.5 ppm);
April 14, 1978.








Top layer temperature Glass cover temperature Bottom layer temperature Ambient temperature


:30 - 9:30
P.M.
Time, hours (E.S.T.)


140 120 .100


Q)


Q)
H 80






60 40


9: 30 A.MI.


4-4

300





(0
o 100 ,Co
o


A.1M.
























Figure 31. Temperature-time history of the two stills; March 31,
1978.




Full Text

PAGE 1

REDUCTION OF OVERSELECTIVE RESPONDING AND ESTABLISHMENT OF A GENERALIZED RESPONSE SET ON THE BASIS OF MULTIPLE CUES UNDER MULTIPLE SCHEDULES OF POSITIVE REINFORCEMENT AND TIMEOUT BY MARIA DEL ROSARIO RUIZ A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1982

PAGE 2

ACKNOWLEDGMENTS The completion of a dissertation is a landmark in one's professional development. The road to this juncture has often seemed arduous and circuitous, and the experiences along it have been colored by thrill as well as despair. This, then, is a time to recognize and give thanks to those who have supported, and even nurtured, me along that road, and to whom I am indebted: to my parents, models of relentless courage, who have provided an endless source of intellectual, emotional and spiritual comfort to my teachers. Professors Brackbill, Goldstein, Johnston, Malagodi and Wolking, who have provided a constant source of intellectual stimulation and guidance, as well as personal support and enc ouragement to Dr. H. S. Pennypacker, who has provided the environment in which to develop a true appreciation for discovery, and himself as a model of intellectual honesty. ii

PAGE 3

TABLE OF CONTENTS Page ACKNOWLEDGMENTS ii LIST OF TABLES ., iv LIST OF FIGURES v ABSTRACT o vi INTRODUCTION i METHOD 12 Subjects 12 Apparatus 13 Training Stimuli 15 Procedure 16 RESULTS 32 Responding in Multiple Schedule Components Signaled by Color, Line-Tilt and the Corresponding Color /Line-Tilt Compound . . 41 Responding in Stimulus Control Probes 43 Discrimination Performance Under Novel Color/Line-Tilt Compounds 44 DISCUSSION 47 APPENDIX 59 REFERENCES 77 BIOGRAPHICAL SKETCH 81 iii

PAGE 4

LIST OF TABLES Table Page 1. The Order of Experimental Phases and the Number of Sessions in Each for Sandy 18 2. The Order of Experimental Phases and the Number of Sessions in Each for Dawn 20 3. Sandy's Mean Response Rate Per Session 59 4. Dawn's Mean Response Rate Per Session 68 iv

PAGE 5

LIST OF FIGURES Figure Page 1. Sandy's median rates of responding in the first and second discriminations, plotted across experimental conditions in their order of occurrence as numbered along the x-axis 3^ 2. Sandy's median rates of responding in the third and fourth discriminations, plotted across experimental conditions in their order of occurrence as numbered along the x-axis 36 3. Sandy's median rates of responding in discriminations five through eight, plotted in their order of occurrence as numbered along the x-axis 38 4. Dawn's median rates of responding in the first, second and third discriminations, plotted across experimental conditions in their order of occurrence as numbered along the x-axis 40 V

PAGE 6

Abstract of Dissertation Presented to the Graduate Council of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy REDUCTION OF OVERSELECTIVE RESPONDING AND ESTABLISHMENT OF A GENEPvALIZED RESPONSE SET ON THE BASIS OF MULTIPLE CUES UNDER MULTIPLE SCHEDULES OF POSITIVE REINFORCEMENT AND TIMEOUT By Maria del Rosario Ruiz December, 1982 Chairperson: H. S. Pennj'packer, Jr. Major Department: Psychology Following acquisition of a discrimination involving a compound discriminative stimulus, developmentally delayed children often respond to only one, or to a reduced number, of the elements making up the compound when they are presented separately in stimulus control tests. In the present study, a discrimination training procedure involving multiple schedules was used to decrease the "overselective" responding of two developmentally delayed subjects. Multiple schedule contingencies were arranged for responding in the presence of single vs. multiple cues on either of two levers. Responses on one lever in the presence of a single colored circle and a single line were reinforced on a VI 1 min.:token schedule; responses on the alternate lever were on extinction. Following this training, a compound stimulus composed of the line superimposed on the colored circle was introduced. Under this three-component multiple schedule, the contingencies on the levers

PAGE 7

remained as before in single element components, and were reversed in the compound stimulus components. With the extinction schedules, the subjects continued to respond on the reinforcement lever associated with single elements, and did not switch-over in the compound components. The extinction schedules were then replaced by two-minute timeout periods. The timeout schedule produced reliable switch-overs for both subjects. This discriminated multiple schedule performance was obtained under the punisliment schedule, but not under extinction. Stimulus control probes involving novel color/line-tilt combinations were presented following acquisition, but the subjects' performances were specific to the stimuli associated with the training schedules. However, when a series of discriminations involving additional variations of the color/line-tilt dimensions were presented under the training schedules, the number of timeouts presented following "incorrect" responses decreased for both subjects across successive discriminations. The differential effects of extinction and timeout in decreasing errors and arranging for generalization were significant. These findings suggested that discrimination training techniques which minimize the opportunity for the occurrence of non-reinforced stimulusresponse relations may be advisable in teaching developmentally delayed subjects to respond on the basis of multiple cues. vii

PAGE 8

INTRODUCTION Over the last decade a series of studies has been published describing certain learning deficits which characterize the performance of autistic and other developmentally delayed children in discrimination training tasks. Lovaas and his colleagues, the principal contributors to this thematically integrated research effort, have employed the term "stimulus overselectivity" (Lovaas, Schreibman, Koegel, & Rehm, 1971) to describe what generally occurs when such children are presented with a stimulus composed of multiple cues in discrimination tasks. The basic training and testing procedure followed across the various studies is quite similar to that originally described by Reynolds (1961) in his early studies of selective attention in the pigeon. Following acquisition of a discrimination involving a compound stimulus, these subjects typically respond to only one, or to a reduced number, of the elements making up the compound when they are presented separately in stimulus control tests. Although stimulus overselectivity has been studied most frequently with autistic children (Koegel & Schreibman, 1977; Koegel & Wilhelm, 1973; Lovaas & Schreibman, 1971; Reynolds, Newsom, & Lovaas, 1974; Rincover & Koegel, 1975; Schover & Newsom, 1976; Schreibman, 1975; Schreibman, Koegel, & Craig, 1977; Schreibman & Lovaas, 1971), several reports have pointed out that mentally retarded (Hale & Morgan, 1973; Lovaas et al., 1971; Olson, 1971; Wilhelm & Lovaas, 1976) and children 1

PAGE 9

2 labeled learning disabled (Bailey, 1981) also display this response pattern when performing complex discriminations. Moreover, the degree of overselectivity exhibited by non-autistic retarded subjects appears to be negatively correlated with IQ level (Wilhelra & Lovaas, 1975). In addition, normal subjects have been studied, and similar findings reported on the relation between chronological (Schover & Newsom, 1976) and mental age of both normal and retarded children and overselective responding (Fischer & Zeaman, 1973; Hale & Morgan, 1973). In the first study of this series (Lovaas et al., 1971), three groups of children — autistic, retarded and normal— received reinforcement for responding when presented with a cross-modal compound stimulus involving simultaneous auditory, visual and tactile cues. Following acquisition, the elements of the compound were presented separately to assess the degree of stimulus control acquired by each. They found that the autistic children responded primarily to one of the elements; the normal subjects responded uniformly to all three elements while the retarded group responded at intermediate levels. When nonfunctional elements were presented in isolation, overselective subjects acquired the discriminations quickly, ruling out the possibility that sensory impairments produced the original results. Subsequent to this study, a number of systematic replications have reported similar results when the discriminative stimulus has involved the simultaneous presentation of two cross-modal (i.e., visual and auditory) cues (Lovaas & Schreibraan, 1971); or when two cues either in the same visual (Koegel & Wilhelm, 1973; Schreibmen et al., 1977) or auditory (Reynolds et al., 1974) modality have been presented.

PAGE 10

3 Overselective responding has been discussed as a potential contributor to the language deficits characteristic of many retarded and most autistic iiidividuals, such as the contextiially impoverished verbalizations of echolalic children (Koegel, Rincover, & Egel, 1982; Lovaas, Koegel, & Schreibman, 1979; Reynolds et al., 1974). For example, language training which requires imitation and thus the acquisition of various discriminations based on simultaneously presented visual and auditory cues is frequently ineffective (Lovaas et al., 1971). Similarly, other procedures which are highly productive in the regular classroom may in fact interfere with the learning process when used with children who have difficulties acquiring complex discriminations. Some common teaching techniques which have been ineffective with this population include observational learning procedures (Varni, Lovaas, Koegel, & Everett, 1979) and traditional prompting and fading techniques (Koegel & Rincover, 1976; Schreibmen, 1975). In such cases, control of behavior by contiguous or nearcontiguous cues and transfer of stimulus control (i.e., as in prompting procedures) actually interfere with the acquisition of contextual "meaning," appearing to promote overselective responding. Along the same lines it has been shown that the behavioral gains acquired by these subjects in one learning situation (such as in a particular setting, or with a given stimulus display, or through the mediation of a specific adult), do not readily transfer or generalize to a new one (Rincover & Koegel, 1975; Schreibman & Lovaas, 1973). In many cases, "incidental" stimuli such as a particular piece of furniture or clothing, or a gesture by the adult acquire stimulus control over the target response.

PAGE 11

4 The implications of these findings bear directly upon traditional formulations of maximal learning environments for these youngsters derived from research with normal subjects. More recently, research efforts have been directed at the development of a more suitable technology which adequately accommodates the behavioral characteristics of these children. One technique developed by Schreibman (1975) employs within-stimulus prompts to insure that the designated element of a compound stimulus acquires control over responding. In this procedure, the relevant element of a visual or auditory display was initially exaggerated and gradually faded to match the dimensions of the redundant elements. While this procedure was highly effective in producing the target discriminations, its main drawback is that it sidesteps the problem encountered when a subject is required to "attend" to more than one relevant element in order to master a discrimination. In other words, the subject's responses may persist under the control of a single (relevant) element and still produce reinforcement. A second technique which has also proven effective is overtraining an already acquired discrimination (Schover & Newsom, 1976) . Although this approach represents a somewhat more direct attack on the problem, the results have been equivocal. Overtraining trials alone may not be sufficient, but may require prolonged testing trials interspersed with the reinforced trials to effect the desired results (Schreibman et al., 1977). The final technique reported represents the most straightforward attempt to reduce overselective responding (Koegel & Schreibman, 1977). In this procedure, described by the authors as "conditional discrimination training," four autistic subjects initially obtained

PAGE 12

5 reinforcement for pressing a bar when either a visual or auditory stimulus was presented. After single-cue training, responding was reinforced only when the two cues were presented simultaneously; responding was not reinforced under single cues. All subjects eventually stopped responding under single cues and learned to respond to both elements. The authors reported the discriminations were learned only after many trials, in contrast to normal subjects who required only a few trials to master them. In the same study, one autistic child was taught a series of nine conditional discriminations involving discriminative stimuli with two relevant redundant cues (e.g., green square). In the test sessions, the subject was presented with three color-form combinations, one containing the original color, another containing the original form and the original compound discriminative stimulus. After two conditional discriminations, the subject made no further errors and appeared to have learned to respond to new discriminations on the basis of multiple cues. The study by Koegel and Schreibman (197 7) is perhaps the most 5 significant reported to date in the literature on overselective responding. Its importance lies in the fact that it illustrates the first attempt in this series to exert direct control over stimulusresponse relations by arranging appropriate consequences (Skinner, 1938) for their occurrence. Various authors have pointed out that differential reinforcement in the presence of a compound stimulus is not sufficient to establish predictable control over responding by any or all elements composing the stimulus (Ray & Sidman, 1970; Terrace, 1966). The animal literature, for example, contains many illustrations of "stimulus overselectivity" in various species when responses were

PAGE 13

6 reinforced in the presence of compound stimuli (e.g., Lashley, 1942; LoLordo & Furrow, 1976; Ray, 1967; Reynolds, 1961; Segal & Harrison, 1978; v;arren, 1953). In this sense, the occurrence of overselective responding, per se, may not be altogether surprising, but its generality among the present population is noteworthy and deserving of research efforts to elucidate the environmental manipulations which influence its development and modulation. One variable which has been shown to determine which element of a compound stimulus will exert stimulus control over responding is prior reinforcement history. If a discrimination is established between two stimuli from the same dimension, and a second set of stimuli is subsequently added under the same reinforcement contingencies, typically, the initial stimuli alone control discrimination performance following discrimination training under the compound (Fields, 1978; Johnson & Gumming, 1968; Mackintosh, 1965). This effect has been reported when tlie second element was introduced abruptly (Fields, Bruno, ^ Keller, 1976) and gradually (Fields, 1978). Analysis of reinforcement histories prior to the stimulus compounding operations is critical to the analysis of overselective responding, particularly when one considers that assessing element control following exposure to the compound stimulus may not always reveal the actual stimulus-response relations established in training. For example, different stimulus control tests may yield conflicting information about the actual relations acquired in training (Reynolds, 1961; Wilkie & Mason, 1976). More importantly, the influence of reinforcement history extends to the test situation itself, such that,

PAGE 14

7 the reinforcement contingencies which prevail during the test may influence the acquisition process (Fields, 1979, 1981; Huguenin & Touchette, 1980). The importance of establishing stimulus control baselines prior to the compounding operations was illustrated by Ray (1969) in a study using monkeys as subjects. Two elements with conflicting prior histories and which controlled incompatible responses were combined to form a compound stimulus. For all subjects, the unchanged element which was compatible with the compound training contingencies supported acquisition. However, even though errors dropped to zero level in the course of compound training, the reversed element continued to control the stimulus-response relations established during single element training when the reversed elements were presented alone. Thus the subjects showed no evidence of learning about the reversed reinforcement contingencies, a finding that would have gone undetected in the absence of single element baselines. Huguenin and Touchette (1980) replicated Ray's (1969) procedure with severely retarded subjects and found that in conflict compound training under differential reinforcement, unchanged elements tended to exert control over responding, but in many cases, subjects showed evidence of learning about the reversed element contingencies. On the other hand, when nondif f erential reinforcement was employed in the test trials, there was a greater tendency for the unchanged element alone to exert control. In fact, the unchanged element continued to be the only reliable source of stimulus control even with overtraining when nondifferential reinforcement tests were used. A rather significant aspect of their findings was that when elements with compatible

PAGE 15

8 reinforcement histories were combined, performance in the single element baseline checks showed that both elements controlled responding. Therefore, when prior reinforcement histories were compatible, overselective responding did not occur in the presence of compound stimuli. The significance of arranging appropriate reinforcement histories in the modulation of overselective responding was demonstrated by Koorland and Wolking (1982) in a study with learning disabled children who displayed either auditory or visual modality preferences. In bisensory learning tasks, visual and auditory versions of the same material were simultaneously presented and the subjects were asked to recall the content. One subject consistently reported the auditory version, while the second reported the visual. During training, responses based on the content presented in the non-preferred modality were differentially reinforced, producing a preference shift compatible with the reinforcement contingencies prevailing during training. The study illustrated the direct effect of reinforcement history in modulating what is frequently interpreted as an immutable attentional-behavioral deficit. J While the above is a clear illustration that unidirectional shifts in stimulus control can be reliably generated given appropriate reinforcement contingencies, bidirectional shifts in stimulus-response relations have also been obtained when reinforcement histories were directly controlled. Singh and Beale (1978) employed a variation of Wyckoff's (1952) observing response procedure with retarded subjects to measure the development of stimulus-response relations under conditions in which the relevant dimension (i.e., the dimension associated with reinforcement) of a two-dimensional stimulus was manipulated. The subjects responded on either of two observing response keys to begin a

PAGE 16

9 trial responding on one key produced color elements on the choiceresponse keys, while responding on the alternate key produced letter elements on the choice-response keys. The subjects were then presented with four transfer problems in alternating order in whicli the previously relevant dimension continued to be correlated with reinforcement contingencies (i.e., intradimensional shift) or the previously irrelevant dimension v;as correlated with reinforcement contingencies (i.e., extradimensional shift). With the first extradimensional shift, the subjects' rates of responding on the observing response key associated with the irrelevant dimension increased, as did their error scores on the choice-response keys. Following mastery, however, the rate of observing responses on both keys increased within trials across conditions, and acquisition in the second discrimination involving an extradimensional shift was significantly faster. These data indicated that the subjects' responses came under the control of the prevailing contingencies of reinforcement which required repeated acquisition of novel stimulus-response relations. The foregoing analyses of overselective responding do not support the view that restricted attention is an unmodifiable (Zeaman, 1973), or potentially diagnostic (Lovaas et al., 1971; Wilhelm & Lovaas, 1976) characteristic of developmentally delayed individuals. They do, on the other hand, underscore first, the importance of employing measurement operations which literally make visible the interplay between several controlling relations, and secondly they demonstrate the role of prior history of reinforcement as a determinant of the conditions under which these relations occur.

PAGE 17

10 The present study was concerned with the development of compound stimulus control over responding by developmentally delayed subjects. The strategy elected was to make reinforcement contingent upon the occurrence of specific stimulus-response relations. Ray and Sidman (1970) have pointed out that by delivering reinforcement only in the presence of certain stimuli, the likelihood of reinforcing the desired controlling relation is increased. Thus, these authors suggest that the stimulus-response relation can function as an operant unit in the establisl-unent and maintenance of stimulus control. The approach taken in the present study follows this general scheme, and makes use of a combination of procedures each of which has been shown to facilitate the development and maintenance of stimulus control. Early studies of stimulus control demonstrated the utility of multiple schedules in examining discrimination problems in normal and retarded children, and the sensitivity provided by rate of responding as a measure of learning (e.g.. Bijou, 1961; Bijou & Orlando, 1961). In the current study, multiple schedules were used in a two-response procedure to establish independent stimulus-response relations in the presence of single color and line-tilt elements, and in the corresponding compound stimulus composed of the combination of these single elements. Three questions were of interest: first, whether discriminative responding would develop in the presence of the compound vs. single elements under the multiple schedule arrangement; and second, what schedule contingencies would most efficiently reduce errors and thus increase the cost-effectiveness of the procedure. In the second part of the study novel color/line-tilt combinations were introduced as

PAGE 18

11 tests of stimulus control and to examine the course of acquisition of subsequent stimulus-response relations bowed on multiple and single stimulus cues. S

PAGE 19

METHOD Subjects The two subjects who participated were residents of Threshold, Inc., a day school and residential facility for autistic and other severely developmentally delayed children. Dawn was 16 years 8 months of age, and had been institutionalized for 4.5 years. Her scores on the Stanf ord-Binet showed her to be functioning at a mental age of 3 years 8 months, and the Leiter International Performance Scales indicated a 45 10. Dawn had good receptive verbal skills, but her expressive speech was characteristically echolalic. She exhibited a great deal of self-stimulatory behavior (rocking; hand flapping), and was classified in the Moderate range of retardation. Sandy was also 16 years 8 months of age, and had been institutj^onalized since the age of two. Her scores on the Leiter International Performance Scales indicated a mental age of 4 years 3 months, with a 28 IQ. She frequently engaged in self-stimulatory behaviors such as flicking objects. Sandy had no vocal speech and was classified at the Profoundly-Severe level of retardation. However, during the last three years she had acquired a growing sign-language repertoire and was mainstreamed into a Special Education classroom while this project was being conducted. Prior to the experiment, the subjects were screened for overselective responding (Lovaas et al., 1971) in a standard discrimination 12

PAGE 20

13 training task using compound stimuli. Tv/o different colored circles with vertical and horizontal lines superimposed were presented simultaneously on 3 in. by 5 in. notecards. The subjects were trained to point to one color-line combination arbitrarily designated as the discriminative stimulus. Stimulus control probes followed acquisition of the discrimination. Both subjects responded with 100% accuracy in the presence of the color component, but at chance levels to the line stimuli. The results were reproduced in a second variation of the procedure and indicated that, with these subjects, differential reinforcement in the presence of a compound stimulus was not sufficient to establish stimulus control of responding by both elements of the stimulus. Apparatus The experimental chamber consisted of a table 69 cm. high with a wooden cubicle 61 cm. long, 28 cm. deep, and 61 cm. high, placed on its center, A wooden platform raised 6 cm. was mounted inside the chamber ajid housed two levers set 25 cm. apart. The subject sat inside the chamber facing the stimulus display wall. The ceiling of the chamber extended 61 cm. beyond the table edge over the subject's head, and a curtain suspended around it was drawn closed during sessions. A 40W. lightbulb used as houselight provided the only illumination in the room, except for a flashlight in the Experimenter's work area. The Experimenter sat behind the stimulus display wall and manually dropped plastic tokens through a metal chute mounted along the subject's left wall. Tokens fell into a metallic cup inside the chamber.

PAGE 21

14 A "magic eye" was mounted under the token cluite which the Experimenter frequently used to view the subject during sessions. A double screen retractable panel was built into the stimulus display wall. The front panel was wooden and could be raised and latched into position out of view of the subject, or lowered into the chamber. The back panel, i.e., the stimulus display panel, was lined with white felt, and had a small hook attached to its center at the subject's eye level. This panel was hinged to the outside of the stimulus display wall and could be opened and latched closed from behind the cubicle. During sessions, the Experimenter suspended a stimulus on the hook attached to the display panel, latched it closed and secured the wooden panel in the raised position while the subject responded. To change the stimuli, the Experimenter lowered the wooden panel, opened the display panel, changed stimuli and repeated the previous operation. Recording of responses and timing of stimulus presentations were carried out automatically through BRS Poringer solid state programming J equipment set up on a table behind the cubicle next to the Experimenter. Each stimulus presentation lasted one minute, and the Experimenter initiated this interval by pressing a foot pedal activating the counters which automatically stopped recording after one minute. After this interval, the Experimenter manually changed the stimuli and recorded the counter readings for that period, a process that took 5 seconds to complete. To simplify the process, four counters were used along with a channel switch. Positioning the switch on channel 1 activated the recording of right and left lever responses on counters 1 and 2. Switching to channel 5 activated these recordings

PAGE 22

15 on channels 3 and 4. Thus, when a single element was presented, the channel switch was set manually in position 1. When a compound stimulus was presented, the channel was switched to position 5. The stimulus presentation order was preset on daily data sheets which the Experimenter followed to change stimuli. The interreinf orcement intervals were taped on a cassette with a 5 second delay between schedule components. The Experimenter listened to the recordings through a small earphone attached to the cassette player. Changeover delays and limited hold intervals were timed manually with a stopwatch. The cassette recorder was turned on at the beginning of each session and ran either until a token exchange occurred, the session was completed, or until a timeout interval was initiated. During timeouts, the recorder was turned off and restarted after 2 minutes. Training Stimuli Three different types of stimuli were used: (a) A colored circle ip.8 cm. in diameter was cut out of construction paper and sealed between two sheets of clear adhesive plastic. A 1.3 cm. square of plastic was left attached to a tangent of the circle and a hole was perforated in the center of the square. The figure could thus be suspended from the hook inside the stimulus display panel, (b) A line strip 3.8 cm. long and .6 cm. in diameter was cut out of black construction paper. Two circles 10.8 cm. in diameter were cut out of clear adhesive plastic, and the black line was mounted and sealed along the diameter of the plastic circles. For each line stimulus, the square plastic hanger was positioned on a tangent of the circle such

PAGE 23

16 that the line was suspended on a specific axis (e.g., horizontal, vertical, or 45-degree axis, etc.). (c) A compound stimulus was created by combining the single elements described above. For example, for Dawn, an orange circle and a vertical line were used as single elements. These elements were combined and presented simultaneously as a compound stimulus, that is, a vertical black line superimposed on an orange circle of the same dimensions as above. For Sandy, a green circle and a horizontal line were used. In successive experimental phases, six additional stimulus combinations were used. Procedure Phase 1 — Multiple schedule components signaled by color and linetilt . Sessions were held, on the average, three times per week. Initially, the subjects were allowed to exchange tokens delivered in the experimental chamber for small bites of food and sips of juice and soda. Following this exchange, they were trained to respond for tokens by depressing either of the two levers set on the work panel in the c^iamber. During the first session, the subjects were physically prompted to depress each lever. Wliite plastic poker chips were manually dropped through the metal chute. Release of a poker chip by the Experimenter created a clicking sound lasting .5 seconds travel time before reaching the token cup. Tokens were delivered on a continuous reinforcement schedule, and the exchange ratio advanced to five tokens per exchange. Both subjects had extensive histories with token reinforcement outside of the experimental chamber, so that once level pressing was established, the schedule contingencies were rapidly changed to FR 1 [Mult (Cone EXT VI 5 sec: token) (Cone EXT VI 5 sec: token)]. The

PAGE 24

17 discriminative stimuli used in the multiple component schedules were an orange circle and a vertical line for Dawn, and a green circle and a horizontal line for Sandy. The color and line stimuli were presented in alternate order, each presentation lasting 1 minute. Under this schedule, depressions of the right lever in the presence of either the color or line stimulus were reinforced, on the average, every 5 seconds with a token. An extinction schedule was concurrently arranged on the left lever, and a 5 second changeover delay (COD) timed from the first response on the VI lever was in effect. Tokens were exchanged for edibles following each multiple schedule sub-component. At all tiroes during this and subsequent phases throughout this experiment, extinction was in effect for simultaneous responding on both levers. During this phase, the subjects were instructed to use one hand only. After the second session the subjects were prompted to respond on the right lever only, and to switch over when left lever responses were emitted in the presence of the discriminative stimuli. Both subjects stopped responding on the left lever after nine sessions. 5 Phase 1 conditions remained in effect for six weeks. A total of 18 sessions were held with Sandy and 20 with Dawn. During this time, the schedule contingencies were gradually changed such that by the end of the phase, tokens on the right lever were delivered on a VI 20 sec. schedule with a 5 second limited hold, and token exchanges occurred, on the average, after five multiple schedule sub-components, with two exchange periods per session. The transition exchange schedules used during this phase, the number of exchanges per session, and the average number of tokens per exchange are listed in Table 1 for Sandy and Table 2 for Dawn. Five minutes free time spent either playing with

PAGE 25

18 c w o c •H c o •H cn 0) U-l o u OJ XI E 3 z; 0) c ro CO l/i CO c Oi 6 •H 0) a X w m o -a o H i-H •TD 03 •i T-l > rH ^ cN-a M M > ce: o > o CN ^ PH H H > o > ^ H La Ea U • o H H tJ rH C CN H O H EX X C DS 0 OS CJ •• C •• W o CJ O u > C CN O CN o C C O o cli CJ es o U c 0 o ^ H H u > w > c C o u (J o o u r H H ^ H o H O • • O H •• ^ — ' S N •• CN r CN H — . H P H H r M O M •• rH rH — s H •• >< o > CN > M rH H p H = W CN M rH > IH > o O tH O > O M > O CN H CM H > c C > o u in in = > : O U H O o o H O > > > o o H H U •• O H CN H CN CN fN ^ H H H >< M M w > rH CM OS EX EX EX w > 05 H tX4 _.. > H pS ^ > H > ^ H u u o o o : o a o a c c c c O O c o CJ •C c c C o o o o C CN O CN C <-\ O M o o o U CJ u u O M U M O M CJ M u u o ^— ^ ^ — ' >^ U > w > CJ > w > — y ' u 4J 4-1 ^ — ' 4-1 4J 4-) 4.1 rH r-i 4-1 •u .H .H iH tH 3 3 tH 3 rH 3 3 3 3 a 3 3 S 2 Z O CN O CO r— I CN m rH rH 1^ rH On rH ai OS a; oi > > > > > > > tH CO O H CN CN

PAGE 26

u T3 C o H M > > o c H O '-s H O '• C O t-l U M w > 3 00 O H 0) QJ U x: to •H U to -0 -a 0) c c > to •H le Id J rH 0) •H QJ Qi to cn 0 rH XI 0 a 3 E 01 3 tn QJ 0 4-1 hi Ch C )-( u cn CO 01 1-1 0) XI 0; u M uin ev QJ +J > the 2 a c QJ cn • 0 4J QJ a c C rH 0) 0) e QJ •H 3 DO 4-1 0 U XJ c tn u U • cn QJ to •H 3 u to X! x; iH u 0 c CJ u cn C -H cn X 0) >-i 0 0 T5 •H 0 n. QJ to 0 00 c iH CO 3:; C 0 C Q> rH -rl to 0 x: to x: x: u •H 4-1 QJ CJ •H tn u 4J X to tn •1 0 v< •H w c 0) tn U-l QJ B to cn Qj 4-1 •t-( QJ u i-H 4-1 <4-( t-J x; 4J u 3 U CO 4-) QJ T) Q) 1 M 0 0 rH OJ x: 0 0 CJ ^-^ vO u U CM •H • CO % Q) 0:1 u CO x> to r-\ QJ C tn E r-l to tn -H CO QJ 3 2 QJ XI 01 ^1 UO 4-1 0 4-i .C 0 c. C -U '•j; QJ 0 2; CO to u X! OJ s

PAGE 27

20 to C U QJ O -I u C/2 c • • Q) CJ W W C/2 3 -o 0) to CO XI p-l I in fO CO ^ N p H 0 CN| In 1— 1 l-l M > > > H H X X W W U U u c c c 0 0 0 u 0 u — ^ , — , ^ — . H H 0 In CM M M > > H H H X X W 0 a C C c 0 0 0 u u u ^ — ' ^ — ' u u 4-) t-t .H -H 3 3 S 2: ^ — • ^ — ' in oi a: ft. > > o CM u O H 00 H 0 CM l-H > H 0 C 0 U — ' H 0 (N H M > 0 H CN M > 0 U C c 0 0 u u 3 as oa > 00 o H -J 0 IN ~* 00 r-i 1 CM 1 CM ( C^J 1 cn CM 1 c^ rH .-1 rH ^ (M — ^ CN CM OS — N OS 02 u H > > > ^ X H H ^ — s M 0 •• 0 u 0 u 0 C rH c CM H C H 0 0 02 0 02 0 0 C > U CM U CM 0 ^ — ' OS OS u H H > > • • r ^ / s H = p H H H H ' N •• 0 0 p ^ , 1 CM CM H 0 M 0 M rH rH rH rH X CM > CM > M M M M 0 W M M > > > > eg > 0 > U M H c c a 0 U > 0 0 0 H c H C H 0 H u 0 0 H 0 N CM 0 CM CJ rg rH CM OS 02 v_/ M PS ^ 02 — N > > > H > H / — V /r*s 0 H 0 H c (J = u c C •• 0 c c 0 a 0 0 0 tj 0 0 CM 0 CM u rH U rH CJ U M M M ^ M u > > > 4-1 > rH 4-1 rH rH 3 rH rH 3 3 3 ' — ' N -' in 00 rH ON 00 rH rH OS a: DC 02 > > r> > o H O H

PAGE 28

21 01 3 C •H U C o o I CM CO rH X! in TO C u 0) ra O '-I pi H n) u c • • to w 0) O Cfl tn • C C J3 cfl OJ (J SOW 3 HI o C/3 CO x: o o -a1 0^ iH 00 CN o H OS u u o C H C CJ !N OS > H H > > o a H n •• o CM u Pi > y H c •• o U r-i M 4-1 > r-H 3 >^ 00 tH OS ta o H o Q) 4-1 •r-4 ,x; TO O in •H u TO TD 11 c C > (tJ •i-t 0) 1— ( Id QJ CO o 60 -r-( 4-J C x: TO 0) 00 T3 x: !/) •I-l 0) o TO S-i •u X x: •H u a. II e •H 4-J J 0) 0) c a a QJ d O Cfl cn o QJ c a tn TO B Q o o u H ar '13 TO 4-1 c o en CO nt V-l OJ •H Q) TO tn o XI U D. e QJ 3 M-l x: tn Z O 4-1 hi c >-l 4-1 TO 01 01 XI QJ u IS^ e 3 > 0) QJ 4J 0) z iH l4-( TO al 4-1 U-( Cfl 4-) 01 0) CD o iH r-i • 01 H 3 cn rH II 13 OJ 3 OJ OJ 3 x; 4-1 r; >— 1 0) H C CJ TO x: QJ cn > o C C/l O U -o a C r-4 01 0) e QJ 3 CO 4J o »-i t3 c CO u u • QJ TO •H 3 4J x: x: 1-1 4-1 O d u o cn c •H cn 0) u o o tu TO •H u a 0) 00 c <-t cn c o C QJ iH •H TO •H o x: TO x; x: 4J •H 4-1 4J o •H tn U cn CO O M c QJ cn >4-( QJ TO C/3 OJ 4J QJ 4-J u-t XI er du C 0) TO 4J 0) cu QJ 14-1 cn C x: x: M-4 QJ H CO u OJ i-l 0) cn 3 XI 00 c 'XI OJ 0) c 4-1 •H 0) T3 cn TO C x; 3 TO x: • OJ cn o rH XI CJ C u TO tn o D. .3 § u c 3 M •H C CJ ;3 > •H C4-I 0 c O 0) x: O O o o iH u 4-J u .-I QJ •H 4J TO OJ cn u XI TO e -i r-l 0) C tn TO cn •H QJ 3 Xi OJ Z QJ U 4J O 4-1 x: o O u o C 4-1 OJ S3 TO 0) C TO eff o

PAGE 29

22 preferred toys or sitting in a rocking chair were included in the reinforcer selection list at this time. At the completion of each multiple schedule sub-component, the Experimenter manually released the double screen and changed the component schedule stimuli from behind the experimental chamber while the subjects viewed the wooden panel. In effect, an inter-component interval lasting 5 seconds occurred during multiple stimulus alternations. During session eight. Dawn continued to respond during the inter-component interval. To eliminate responding in the absence of a discriminative stimuli, a two minute timeout (TO) contingency was implemented in which the Experimenter stated loudly "No, fold your hands," and turned off the houselight in the chamber. Responding during the inter-component intervals v/as eliminated by the end of this session and did not recur after this. Phase 2 — Multiple schedule components signaled by color, line t ilt , and the corresponding color/line-tilt compound . During week seven a third discriminative stimulus was introduced in which the single elemjjnts used in Phase 1 with each subject were combined to form a compound stimulus. For Dawn, the compound stimulus consisted of the vertical line superimposed on the orange circle, and for Sandy, the horizontal line superimposed on the green circle. The elements in the compound stimulus were in all respects equivalent to the single elements, except that now they also appeared in combination, A third sub-component concurrent schedule was added to the multiple component schedule with the introduction of the compound stimulus, creating a three-ply multiple component schedule. With this new schedule, contingencies in the presence of the single color and line

PAGE 30

23 stimuli remained as before, that is, responding on the right lever was reinforced with a token, on the average, every 20 seconds with a 5 second limited hold. Extinction was concurrently programmed on the left lever. In the presence of the compound stimulus, however, right and left lever contingencies were reversed . Responses on the left lever in the presence of the compound were reinforced on VI 20 sec.itoken, 5 sec. L.H., while extinction was concurrently arranged on the right lever. A 5 second COD, as described above, was also in effect. Throughout this phase, the three discriminative stimuli were presented in alternate order, each appearing an equal number of times per session. The total number of sessions in this phase was 38 for Dawn and 30 for Sandy. The schedule contingencies in effect were for Dawn, VR 9 [Mult (Cone EXT VI 20 sec. : token, 5 sec. L.H. ) (Cone EXT VI 20 sec: token, 5 sec L.H.) (Cone VI 20 sec .: token, 5 sec. L.H. EXT)], with a 5 second COD in effect within each sub-component concurrent schedule. The exchange schedule for Sandy wasVRlO; all other conditions were the same as Dawn's. The number of exchanges per session and the average number of tokens per exchange for each subject are listed in Tables 1 and 2 for Sandy and Dawn respectively. Phase 3 — Establishment of the color/line-tilt compound as a dis criminative stimulus . The experimental manipulations implemented during this phase were for the purpose of establishing reliable left lever responding and switch-overs from the right lever in the presence of the compound stimulus. A total of 27 sessions were held with Sandy, and 31 with Dawn. During the previous phase, Sandy had earned a total of three reinforcers for responding on the left lever in the presence of the compound, and Uawn had earned one. The subjects, therefore, had

PAGE 31

24 very limited contact with the reinforcement contingencies in the presence of the compound stimulus. An attempt was made to "force" switch-overs on Day 114 with Sandy and 141 with Dawn by increasing the previous ratio of compound stimulus to single element presentations, thus increasing exposure to the extinction schedule on the right lever. Whereas in the previous phase each single element and the compound were presented an equal number of times (1:1 compound-element ratio), in the present session, for every single element presented there were six presentations of the compound stimulus (6:1 compound-element ratio). The length of the sessions was extended until one of two criteria was met: either the subject made at least one left lever response in the presence of the confound stimulus during five consecutive compound subcomponents, or the subject emitted no left lever responses under these conditions. Token exchanges during this manipulation occurred on the average, after 13 components for Dawn, and 12 components for Sandy. In this session. Dawn was exposed to 12 presentations of the compound subcomponent schedule, that is twice as many as in the previous phase, before meeting the no left lever response criterion. The number of compound sub-component schedule presentations for Sandy was five times the number in the previous phase. She v/as presented with 27 subcomponent schedules associated with the compound stimulus before reaching the left lever response criterion. Under the above conditions, left lever response rates in the presence of the compound stimulus were .18 responses per minute for Dawn, and 2.3 responses per minute for Sandy. An additional session was held for Sandy under the old stimulus presentation conditions (i.e., 1:1 compound-element ratio) and left lever response rates in the

PAGE 32

25 presence of the compound stimulus increased sligliCly to 3.3 responses per minute. The rates of responding on the right (extinction) lever, however, did not deviate from previous levels at any time during these sessions for either subject. A second manipulation was thus implemented in an attempt to "force" switch-overs in the presence of the compound stimulus and establish stimulus control of left lever responding while maintaining the same schedule contingencies. On the session held on Day 129 with Sandy and on Day 143 with Dawn, a can was used to cover the right lever in the presence of the compound stimulus. Two token exchanges were scheduled to occur during these sessions, and the number of compound sub-component schedules presented was increased such that 66% of the tokens in the first exchange and 100% of the tokens in the second had to be earned by responding in the presence of the compound stimulus. In meeting these criteria the total number of compound sub-component schedules presented was 20 for Dawn and 36 for Sandy. During the following session, the can was gradually faded out, first by 5 uncovering tlie right lever and placing the can next to it, and eventually by removing it from the chamber altogether. Attempts to fade out the lever cover from the experimental chamber on day 143 disrupted Dawn's performance; in the presence of the cover, responding on the left lever was maintained, but, in its absence, right lever responding returned to its previous levels. Therefore, a punishment contingency was introduced in this session for Dawn to "force" switch-overs. The punishment contingency replaced the extinction contingency in all multiple sub-component schedules. With this new schedule, responses on the right lever in the presence of either the

PAGE 33

26 color or line stimulus continued to be reinforced every 20 seconds, on the average. Responses on the left lever were now followed by a 2 minute tineout period in which the Experimenter stated loudly "No, fold your hands," turned off the houselight and lowered the wooden panel covering the discriminative stimulus. Responses during TO had no effect, except for resetting the TO interval. In the presence of the compound stimulus, responses on the left lever were reinforced, on the average, every 20 seconds, while responses on the right lever were followed by a 2 minute TO interval as above. The cover fading procedure was somewhat more effective with Sandy, and four sessions were held in the absence of the lever cover under the previous schedule contingencies. A nine v/eek period followed in which no sessions were held. The subjects went home for Christmas vacation and were unavailable for sessions during part of this time. Upon their return, Sandy was mainstreamed into a Special Education classroom in the public school system, creating scheduling conflicts. Sessions were resumed during week 32 with Dawn and week 33 with Sandy. The schedule contingencies in effect prior to the break were maintained for Dawn over the next 12 sessions. For Sandy, the extinction contingencies were replaced by a timeout contingency on Cay 226. The timeout operation was the same as described above. Under this new schedule, it was observed that the subjects would often emit the first response after the introduction of the discriminative stimulus in the multiple sub-component schedules, and pause before continuing to respond. This was unusual in that, generally, both subjects responded throughout each sub-component schedule on the VI lever in a steady fashion with infrequent pauses. This new response

PAGE 34

27 pattern suggested that perhaps the FR 1:T0 contingency was functioning as a discriminative stimulus for switch-overs, and its absence after the first response as a discriminative stimulus for responding on a particular lever. Since the objective of the procedure was to establish stimulus control of responding by the multiple component stimuli for three discriminated operants, it was decided to change the TO contingency from FR 1 to VR 2. In this way, the discriminative control over subsequent responding potentially generated by the consequences of the first response was reduced. In effect, this manipulation was made in order to "force" the subject to attend to the multiple schedule stimuli befoi-e responding. The VR 2:T0 contingency was implemented with Dawn on Day 225 and with Sandy on Day 233. On this schedule, responding on the timeout lever produced a timeout period, on the average, after two responses. All other conditions remained as before. The ratio of single element presentations to compound stimulus presentations was changed on Day 247 to a 2:1 compound-element ratio; that is, for each presentation of the color or line stimulus, there S were two presentations of the compound stimulus. This change was made in order to insure equivalent reinforcement rates on both levers each session. The multiple sub-component schedules after this point were alternated in irregular order. Sandy's rates of responding declined to below 50% of the previous level during weeks 35 to 38. On Days 261-263, the 5 second limited hold on the VI schedules was reduced to 3 seconds in an attempt to increase her rates of responding. This manipulation had no effect, and a second manipulation was thus undertaken. Prior to the session on Day 264, Sandy was exposed to a mini re-training session in which

PAGE 35

28 tokens were delivered on the original VI 5 sec schedule. The mini session lasted four minutes, and four multiple sub-component schedules were presented, with two token exchanges. Response rates following the mini session recovered to previous levels. During Phase 3, the token reinforcement schedules were gradually leaned to VI 1 min. By the end of the phase, token exchanges occurred, on the average, after 15 components for Dawn and 9 for Sandy. The transition exchange schedules used during this phase, the number of exchanges per session, and the average number of tokens per exchange for each subject are listed in Table 1 for Sandy and Table 2 for Dawn. Phase 4 — Stimulus control probes . Both subjects had performed without errors over the last 13 sessions in Phase 3; that is, in the presence of single elements, all responses were made on the right lever. Conversely, in the presence of the compound stimulus, all responses were made on the left lever. The elements making up the compound stimulus were in all respects equivalent to the single ele~ n^nts, except that in the compound, they were combined and appeared simultaneously. The subjects' performance indicated that, under the present conditions, both dimensions of the compound were functional; in other words, both exerted stimulus control over left lever responding. Stimulus control probes were held for each subject over two consecutive sessions, using a new set of stimuli. Dawn was presented with Sandy's training stimuli — a green circle, a horizontal line and the compound of these, a horizontal line superimposed on a green circle. Similarly, Sandy was presented with Dawn's training stimuli — an orange circle, a vertical line and a vertical line superimposed on an orange

PAGE 36

29 circle. Each new element and the new compound were introduced twice within probe sessions. The new stimuli were alternated with the subjects' training stimuli, each presentation lasting one minute. Extinction was in effect throughout the new stimulus probes. From this point on, token exchanges occurred after 18 components for both subjects, with one token exchange in each session. Phase 5 — Establishment of novel color/line-tilt compounds as discriminative stimuli . The subjects' performance in the probe sessions indicated that discriminative responding acquired in Phase 3 was specific to the training schedule contingencies. Discriminative responding was not maintained with variations of the original stimulus dimensions. A second training procedure was initiated in which the new stimuli were presented during half of each session, and alternated in irregular order with the training stimuli. Under all four single elements, right lever responses were reinforced with tokens every minute, on the average, and left lever responses were followed by a two n^nute timeout on a VR 2 schedule. Under both compound stimuli, the left and right lever contingencies were reversed. To set up equal reinforcement rates on both levers, half of the sub-component stimuli presented were single elements, and half were compounds. Each presentation lasted one minute. The number of token exchanges per session and the average number of tokens per exchange for Sandy and Dawn in this phase are listed in Tables 1 and 2, respectively. Sandy required a total of seven sessions before performing without errors in three consecutive sessions. Dawn^ on the other hand, failed to acquire the new discrimination after 28 sessions. Moreover, the discrimination

PAGE 37

30 previously acquired under the training stimulus components was disrupted. The novel stimulus components were therefore withdrawn for Da;,m, and as in Phase 3, only the original training stimuli were presented for 12 sessions. Following re-acquisition of the original discrimination, the training stimulus components were withdrawn, and the three novel stimulus components were reintroduced for seven sessions . Sandy was trained on two additional color/line-tilt combinations with the probe-train procedure. A purple circle, a 45-degree righttilt line and the corresponding compound were introduced under probe conditions, but discriminative responding did not occur. The training schedules were then introduced. The new stimuli were presented during half of each session as described above, and alternated in irregular order with the six discriminative stimuli for eight sessions. Following acquisition, a yellow circle, a 45--degree left-Uilt line and the corresponding compound were introduced under probe conditions. Discriminative responding once again did not occur, and the training conditions described above were repeated. The new stimuli were presented during half of each session, and alternated in irregular order with the nine discriminative stimuli for seven sessions. While Sandy's performance in the probes indicated that discriminative responding was specific to the stimuli associated with the training schedules, she nevertheless emitted progressively fewer errors in acquiring each successive discrimination. Differential reinforcement was therefore substituted for the extinction probes as an alternate test of stimulus control. Four novel color/line-tilt combinations were trained separately: a brown circle, a 68 degree

PAGE 38

31 right-tilt line and the corresponding compound; a pink circle, a 68 degree left-tilt line and the correspoixiing compound; a blue circle, a 23 degree right-tilt line and the corresponding compound; and a red circle, a 23 degree left-tilt line and the corresponding compound. Each stimulus set was presented over two consecutive training sessions. Discrimination performance was obtained under each new condition as Sandj' responded without errors across these eight successive training sessions . Dawn's discrimination performance had been disrupted when the training schedules were instituted for two color/line-tilt combinations within the same sessions. However, she had required successively fewer sessions to reacquire the original discrimination and to master the second discrimination. Dawn was subsequently trained to respond under a third color/line-tilt combination — a yellow circle, a 45-degree lefttilt line and the corresponding compound — in isolation using the training schedules previously described. S

PAGE 39

RESULTS Response rates were recorded after the completion of each multiple sub-component schedule, and the average rates of responding under the color, line-tilt and compound components vjere calculated for each session (see Appendix Tables 3 and 4 for a complete listing of average rates of responding under each component over successive sessions). The median response rates (i.e. , responses per minute) for successive experimental conditions, numbered in their order of presentation, are shown in Figures 1, 2 and 3 for Sandy and Figure 4 for Dawn. Sandy's performance in the original training stimulus components are shown in the upper panels of Figure 1; the lower panels show her performance in the second set of color/line-tilt components. Sandy's performances are the third and fourth sets of color/line-tilt components are shown in Figure 2, upper and lower panels, respectively. Figure 3 shows Sandy's performance in the final four sets of color /line-tilt components. Dawn's performances in the original training stimulus components are shovm in the upper panels of Figure 4. The lower panels show her performance in the second and third sets of color/line-tilt components. In each figure, the squares represent median rates on the color components; the triangles represent median rates on the line-tilt components; and the circles represent median rates on the compound components. Open sjmibols indicate responding on the lever correlated with reinforcement, while dark symbols represent responding on the lever 32

PAGE 40

1 CO 1 cu P Q) a CO .g o 1 Ut QJ o •J c (U i-H OJ D CT* •H to O QJ •I rJ \J L, CO 1 1 to 01 CU u r\ \J CD *H Qj u ^> QJ QJ •n \J I— • O , 1 n) V, M > Xf dJ *r-t 60 u (U ^. i-J '^ I-* T— 1 •U 1 .-J Q J-t tXJ o CJ *H 4-) •H I-T3 OJ CL' CD CO ;3 o *H OJ CO ill O c , 1 J_j J> d) w • JIT 4-» rrt CD B CO 0) QJ m >H CT* CU QJ Q •> »> UJ W-4 CO 4-) L-l iH 'H r\ \J to fH CJ to r; o •H CJ 4-1 f 1 \j *H to Q CJ •H t-i c rl-i QJ *H QJ +J 1 CO CJ me 'H r* 'H W CJ CJ CD QJ *r-( Cj *H 1 1 f-* \-* QJ nt tJ •H cy 0 1 CU o ni to ni C 4>J 1 — 1 1^ i-4 to Q to r* c_/ o w r\-t o *H 0) 01 14-1 QJ *H r\ C-* C CU c *H *H 10 T3 4-1 •H CO d ) 1 m 1 1 0) ^ 01 *H to .p^ 4-1 c CU CJ Tl c CU (C o •> Q CO (=4 e •H dJ •rH QJ X) to QJ Q o 'O p! o cn CO 0) CO to V-i *S *H O 4J cC QJ o >^ u UD 4J r-l 4-1 CO o r-H pG) (30 O U) CO to •H OCO co a c to tfl u c O O CO C U 4-1 0) o to ca 0) o a QJ •H c t-i CO !-4 u •H 6 cn c U T3 o CU (0 O C V4 •r-l cn a a C 03 r-i QJ QJ • ^4 •H o TO 0 to QJ QJ QJ CO X Q) u u x: •H C 4-1 U CJ •H )-l Cfl 4-1 • 0 o 13 to CO a 0) Ui o cn (0 rH CO C CO c o CO 4J 14-4 CO CU QJ •H CU M 3 00 fa

PAGE 41

34 o O i 0) c c o w o I 5 c o 2 -D i § I o £ ^ $ o O O o • i-o° i.UA l.UA^Hjt OiZdA^i! HQ 1 t-CH i.UA^Kw OIZUA i.UA^-H OIZUA iUA^i OiZdA l.UAf^ OiSHA-^p i.OZIA^Ki 01 i-Oj9Ae| lO 1 ixe Ii.OZIA l.,OS8A4 i.OSIA >-< — I in CO to CO TNi^M/dsaa ) 3iva NVla3^M 0) o O O! 2 0 O > am <* (13 C CO o I— OH K>— I I.UA KM I.UA w K3 1 I.UA H
PAGE 42

o 1 CO c to H • •H 1 0) 0) 01 * • • QJ 6 a> ^ o iJ ^ — s CN) u to •H u CO o 1 CO CO 0) 3 to U o D. 3 c CO u s > U CO o O 0) O" •H CO 4J to u tn u t-i c/) 0) 4-1 ^ ' c o OJ CJ •H 0) u 4J c rH B CO •v •n . — s r-l 01 CO a. u > -a •H CO 01 X "O jC o o CO (U o O he CO M 3 CJ •H M 'J cu 4J •H 4-1 0) o o i-H 3 o 4J u > ai CO O <4H c -a c CO CI) e CO 0) S-i QJ •rl c rH •H CO X o CO 3 B 01 (0 •H cr 1-1 4J T3 O •H 60 QJ CO 4J u V — ^ (1) •U c t-4 TJ M C c iH c • ^ £ CO to M c C 0) o •H o CD U CO u •rl o (U 6 H 4-1 0) 01 4.1 S-I J= •H o. OJ 4-1 1 CO rH > >^ CJ QJ 4-1 4-1 O u CJ 01 0) CJ 0) XI X > •H u c C! 4J V4 I-H 0 4-J QJ OJ T3 o ••A CO •H •H "0 rH XI c • u r-H CJ +J QJ OJ u o CO c C u 3 4J X o •H 0) OJ C OJ 3 O CO O c X x; — ^ £ 'H rH CO •H CO M •t3 4J &. 4J rH •H QJ CO 1 0) o o O TD 00 ^ X T) 4J C >-i OJ mh C C c c 0) CO o o •H o •H QJ CO CO O •H D. 60 O CO CO 60 V4 •H 3 X O CO c CO CO C r-l CJ 4-1 c 0) o CO CJ QJ .H CO u 4-1 •H 4-1 OJ CO CO CO OO CO o C 3 u CO c * CO ^1 iH 0) 4-1 O CO CO CO 6 •H CO 0 CO T3 c O CO (U •t-t 01 o 4J c o dJ rH O OJ 0 o o CO >-l c C3 tn u U O •H o c c 4H o •H (U a a) a 01 a. CO 4J 'Ji 3 o u 4.1 u CO x: QJ to o a 00 •H CD C CO to C 4J T3 u -a x; o e OJ H c C (U C c / — \ U •H o u X) QJ O c XJ o 0) 4J c o c a a CO 4.1 CO c <1 > O 3 14-1 01 QJ •H 0 CO o to c 0 OJ o 4J O rH o 0) — ' r-l •H X X CO 01 CO 4J 4-1 u 0) O e o CU CO CO 4J X o 0) 14H c C iJ C OJ o c X to C QJ 01 CO 0) CO 0) o )H •H .9 c )-4 U 60 B -a QJ to QJ V. o c (U 0) x: to c X Sj •H 3 u CO u 4-1 4-1 QJ QJ c U CJ c 8 •H >-l CO U 0) 0) OJ DCO u 0) o •H c to X X X X C o CO V — y •4-1 CJ o U 4J H H QJ CM CU u 3 00 •H

PAGE 43

36 o b CC'J <5 <« £ 3 O o • l-O7 l~CM 05 CO o o o ^ 0) •oi.E o 44 $ a o £ o • I — cv^ i.,UA
PAGE 44

to >> to QJ 0) r-l 0) X) t-H 1 4-1 rH to H 0) 00 • c to CO rH a tsQ 4-1 1 c •H 3 c C ,c to o 4-1 to OJ 0) o U QJ o 4-1 •H to c •H > V4 o u o to 0) t-l 0) to CO to o to to 4-< rH TD to to •H u c to tl) OJ CO to B 4-i o tn u 4-) 4-j QJ a • a) r-l CO o &. 3 10 •rj CO o o t/; > CO (U to O •r-l C )4 o C •H u o M 13 QJ CJ 3 o m o 'O o o E O O D4-1 •H 0) — N • 0-. •3 to Ou to 00 c 4J cn 1^ o c o X) 0) c •H g >^ 0) o "C3 »4H c o o 4_l 3 x: c 1—1 •H C 0) 0 OJ u CO 5 o to 4-1 3 x: 4-1 o tu 4J o O 4J U 14-1 t-H rl to to I— 1 c c D. QJ QJ c to S to o OJ •H g c X> •H u •a 4-1 q a 4-1 O o B to 0) c i U 3 >-l S-4 tu o o a) u o OJ to C CO E CO )-l QJ QJ QJ •H •i-t to 4-1 O T5 x: 4-> OJ i-l u u ^ — X <4H QJ 4J to x: QJ H OJ u c C 4-1 M u x: a. ti 4-1 0 •H to C H > 4J to I-l c (U CO lO u 4-1 QJ rj o u > •H 4-1 to C o E to IX QJ CT U rH 4-1 QJ M rH to TD to tH O •H c rH 'O to 1 to QJ to 4J 3 CO QJ 0) i-l QJ to )-i ^ — / CO o C I-i o QJ to CJ o •H ID u •H CO B rH rH 01 4-1 o X) to ^ QJ • •H •H IH .H 3 <4H c (U 4J )-i 4J 4J QJ to O •rl 4-1 to tu 1 -C 4J QJ QJ « c 3 > c 0) 4-1 3 B 00 0) •H V o c o O •H C »-> to iH 'rn •H c QJ U to to c -o 4-1 r-t o QJ e u to (U c U U > •H CO V4 •H 4J Q) a c 0) to to QJ 4-1 QJ 0) X) 4-) CX •H OJ 0) rH !>, x: > o O i 4J 4-1 *J c 4J QJ e r-( X CD o 4-1 O rH O. u 0) C >-< C •H X) QJ CO u o •H 0) U 01 4-1 x: to o X) «\ O B O CO u •H cw
PAGE 45

38 i,HA<~ to-, (CM a a MX 012 UA^. 012 UA^ I 012 UA^ I IT) CO CM CO O w o JC c a3.E CO cog C o ii-O o e O D 0) CC 0) o CO 0* CM O I a X) e Q) O oco CO z o a z o o 0} o O c o CQ H x: 0) CDC % CD C3) if I 2o mo o* 0) o O 0) QQ be -I CM JZ gi iXT) CO 3 CM O I a ^ o CDO o* ! 1 I "" '' ' ' — O O O ^ lO — (Nli^ -dSBti) 31Vb Nviaai^

PAGE 46

I w 1 •H u 1 )-l T3 o c CO •u W3 •H 3 x: T) u W rH O CO P U o CO 4-1 to > ai CO x; u O 4J x: •H XI J-' H u iJ to B •H o T3 O c x: 4J 1 O o CO 01 0) ---(U U •H o c en CV-i'MCCCncflC-t-J-H-HttJ 'O -H to O -H 0) 01 CJ .-( -rl C a+-'-H.-44J.-!SCOtflO otocrC't-' toua)-H<4-(iHC occnoioojvji-ia-i-i oo) rl Q x: r'S >. X> 60 to ft nl 4J Q Qj rH i— < 1 1 QJ CJ 1 1 t4H I — 1 Q C M ^ CO to 4J ] t T-1 U Q o rtl OJ c CO c CH D. OJ 4J e o •H }^ O d to O 1.1 CO o o o •o •H o • •H a; 0) •H CO c IM tfl 0) 4J 4J CO O 14H 4H o t-l •H M to CO tu CU O tfl c o 0) CO X u U 4J to 0) 60 tu x; u to •r) a 60 0) to CO to C e rH C I o 4J c > •rH CO c •H 0) to H o T3 a c C •H •H OJ •H to CO u CO C CO OJ o 01 (-1 x: O (U u 4J to 3 (U 4J CM E 60 OJ u 0 u 3 CO O ».( X C •H C Ou 00 CO o o a. 0) •H IH •H X CO 0) u 0) ID e 14H 0) Oi a 0) C to o CO S n o o CO ).! a c o o •H 3 o X o 03 u 4J o 1.1 CO 0) 0) a CO CO o 0) a q; QJ X X 0) o u c E X X H to u o u 0) o o u e q; CO u o 0) o > •H u 3 x: o CO »-l G 0) 4.1 C c H u 0) a O CO rH U Oi O tu V o XI Xi x: 0) C x; 0) to 0) e CO r-( 4J •H 4J CO X 3 CO 3 OJ C 4.1 0) 4J •X3 tfl T) 0) c c 4.1 c 0) X 3 4H 0) CO OJ u o o to CO e 0) 0 CO o 4.1 0) 4J to tfl 0) H 4J e CO (X CO to u o T) to o o (U u (.1 OJ CO c •H c c 01 G •H u • tu OJ •H t3 o 0) a H O c Q) •H 4.1 o 0) O 4H ^ — ' M CO u OJ e 4J C V-l to c o CO 3 0) CD to 0) to 4.1 14H to CO C o (0 01 • t3 to to 0) 3 •H o 0) 1-4 0) 0) rH !h O QJ u '% 6 o CO 0) 4.1 4J O OJ OJ 00 0) •H & 3 > CO rH D. ^4 > c > CO M CM 01 cr 0) •H •H Q) •H 0) CO OJ Q O O 1-. to O 4.1 U U rH 4J u rH 0) >-< 3 60 •H b

PAGE 47

J) o <3 Q> O) C CO O « 0) c to o a; > KM CO O > C 65 o • 013 HACO OIZHK OlZtiA 012UA i.OZIA<-Kj — I Olldd jxa I ^^ ,1 oJSA05 ' C: Vl — J..02IA< lX)2IA-=c':1 jxa— I L.UA^-c-t i-i-a — I 9 .b CO O s o Uj o X c o • 013UA 012 UA OIZUA )X3 C o HI I' 5| ao o • o o O if> o in Tin z 03 O o o (NI|/M/dS3H) 31Vd NVia3l^

PAGE 48

correlated with extinction/timeout. The specific multiple schedule contingencies associated with each lever are indicated across conditions. The bars extending from symbols represent the range recorded for each measure. Responding in Multiple Schedule Components Signaled by Color, LineTilt and the Corresponding Color/Line-Tilt Compound Equivalent rates of responding were maintained in the color and line-tilt components by both subjects in the two-component multiple schedule signaled by single elements. The first panel in Figures 1 and 4 shows the median response rate over the last nine sessions of the first condition for Sandy (i.e., 30 responses per minute), and for Dawn (i.e., 34 responses per minute), respectively. Both subjects stopped responding on the left (extinction) lever after nine sessions. When the compound components were introduced, Sandy's median response rate in the single element components decreased slightly, but remained well within the range of the previous condition, as shown in the second panel of Figure 1. Throughout the 30 sessions in this condition, Sandy continued to respond on the right lever in the compound components at a median rate of 19 responses per minute, even though extinction was in effect. In the last session, she responded on the VI lever in five of the six compound components, averaging 3.5 responses per minute for the session. Sandy typically responded with her right hand only, and throughout this session she received three tokens for responding on the left lever in the compound components. The increased variability in error rates across single element components was also recorded during this session.

PAGE 49

42 Dawn also continued to respond on the right lever throughout all compound components across 38 sessions in this condition. In fact, her median response rates on the right lever in the compound components were slightly higher than in the single element components, i.e., 51 responses per minute as compared to 42.5 and 45.5 responses per minute as shown in Figure 4, Panel 2. The only left lever responses emitted by Dawn in the compound components occurred in sessions 49 through 51. On these occasions, Dawn responded throughout most of the sessions with both hands simultaneously and only one token was delivered in session 49 for left lever responding. These three sessions accounted for the increased range in error rates across single element components, and in response rates on the reinforcement lever across compound components in this condition. On a number of occasions, a single left lever response was recorded in single element components. In all cases, these responses occurred simultaneously with right lever responses when Dawn positioned her left arm over the lever in a resting posture while responding with her right hand on the right lever. Increasing the number of compound stimulus components did not generate reliable switch-overs for either subject. Even though Sandy made occasional responses on the left lever, right lever responses in the compound stimulus components did not deviate from previous levels at any time during these sessions. As the third panel in Figure 4 shows, attempts to "force" switchovers in the presence of the compound stimulus by covering the right lever were not effective with Dawn. While she responded on the left lever in the presence of the cover. Dawn immediately reverted back to the right lever when the cover was removed. The procedure was more

PAGE 50

43 effective with Sandy, but as shown in the third panel of Figure 1, she continued to respond on the extinction lever throughout these sessions. An interesting lever pressing topography was observed at this time on the reinforcement lever in the compound components, as Sandy responded by pressing the left lever with her head. When the extinction schedules were replaced by timeout (Panels 3 and 4) , right lever responding in the compound components was eliminated after three sessions for Sandy, and after six sessions for Dawn. Thus, both subjects mastered the discrimination in relatively few sessions with the punishment contingency. At this time, the token reinforcement schedules were gradually leaned to VI 1 min. , and the component requirement per exchange increased to FR 18 with one exchange at the completion of each session. During this transition, Sandy's response rates decreased to approximately 50% of the previous levels. Her rates recovered after the retraining period prior to session 69. Sandy's median response rates across the last seven sessions of this phase was 35 responses per minute in all three component stimuli on the S corresponding VI levers. Dawn's rates increased sharply in the final week of this condition, and then returned to base levels. Her median rates over the last seven sessions were 51 and 61 responses per minute in the color and line-tilt components, respectively, and 61 responses per minute on the left lever in the compound stimulus components. Responding in Stimulus Control Probes Lower Panel 6 in Figures 1 and 4 shows each subject's performances in the stimulus control probes that followed acquisition of the discrimination; both subjects responded on the right lever in the

PAGE 51

44 presence of all three novel stimuli. In the novel color and line-tilt components, 100% of the responses were made on the color and line-tilt associated reinforcement lever; but in the novel compound components, all responses were made on the compound associated timeout lever. Discrimination Performance Under Novel Color/Line-Tilt Compounds Sandy . Sandy mastered the new discrimination quickly as shown in lower Panel 7 of Figure 1. She received a total of four timeouts in the novel stimulus components: two in the vertical line and two in the orange-vertical compound components. She maintained nearly perfect accuracy in the training stimulus components as shown in upper Panel 7, receiving only one timeout in the green-horizontal compound component in the first session, and made no further errors after the fourth session. In the second probe condition, shown in Figure 2, upper Panel 8, Sandy responded on the color associated reinforcement lever (i.e., right) in the purple components, and on the compound associated reinfj)rcement lever (i.e., left) in the purple — 45 degree right compound components. However, in the 45 degree right line-tilt components, 100% of her responses were on the line-tilt associated timeout (I.e., left) lever. She made no errors in the first two training sessions, and required only one timeout in the third session in the novel line-tilt component to master the discrimination. She maintained perfect accuracy in the remaining components throughout training, although her rates of responding became more variable across all components (see upper and lower Panel 9, Figure 1).

PAGE 52

45 Lower Panels 10 and 11 of Figure 2 show Sandy's median rates of responding in the yellow, 45 degree left line-tilt and yellow-45 degree left compound components across probe and training conditions, respectively. As in the initial probes, Sandy responded on the right lever in all probe components. In the yellow and 45 degree left line-tilt components, 100% of the responses were made on the single element associated reinforcement lever (i.e., right), but in the corresponding compound components, all responses were made on the compound associated timeout lever. One timeout was delivered in the first training session in the compound component. Following this timeout, Sandy stopped responding altogether in the yellow-45 degree left compound components, but responded at previous levels throughout the remaining components. Twenty-two novel compound components were presented in the first training session. At the beginning of the second training session, 36 compound components were presented successively, but the subject did not respond as sliown by the range bar in lower Panel 11. When the Experimenter manually guided her right hand towards the left lever, she J physically resisted the prompt. She resumed left lever responding after the Experimenter modeled the response and produced a token. Her response rates in all components were subsequently maintained at previous levels, and she made no further errors after this session. Sandy performed without errors in the final four novel stimulus combinations when each set was presented separately over two sessions. These results are shown in Figure 3, Panels 12 through 15. Pawn . Unlike Sandy, Dawn did not acquire the new discrimination readily as sho-Am in Figure 4, lower Panel 7. Dawn made errors

PAGE 53

46 consistently throughout the first 28 training sessions in which the training schedules were in effect in all six multiple schedule components. In fact, the introduction of the novel component schedules disrupted Dawn's discriminative responding in the training stimulus components as shown in upper Panel 7 of Figure 4. On the 29th session of this condition the novel component schedules were eliminated, and only the original training stimulus components were presented (see upper and lower Panel 8). Dawn received two timeouts in the orange components, two in the vertical components, and five in the orangevertical compound components before performing without errors. These conditions remained in effect over 12 sessions. The original stimuli were then withdrawn, and the novel stimuli reintroduced (see Panel 9). Dawn required a total of seven timeouts to master the new discrimination, five of which were delivered in the compound stimulus components. Both of these figures represent a substantial decrease from the number of responses which were followed by timeouts in the original discrimination (i.e., 47 vs. 8 and 7 timeouts). As shown in Panels 8 J and 9, response rates on the reinforcement levers during both training conditions were maintained at levels equivalent to response rates in the previous phase. Dawn acquired the final discrimination rapidly. As Panel 10 shows, errors remained at or near zero levels throughout this training phase.

PAGE 54

DISCUSSION Standard discrimination training techniques which employ differential reinforcement to establish stimulus control of responding by simultaneous multiple cues are often ineffective with developmentally delayed children. Differential reinforcement alone is not sufficient to generate predictable stimulus control over responding by any or all elements of a compound stimulus with many of these subjects. Alternative teaching procedures which reliably reduce overselective responding in complex discrimination training tasks may eventually yield increased optimism in the learning prognosis of these children, especially in such important areas as language acquisition. The purpose of the present study was to examine the applicability of one such alternative procedure in training two developmentally delayed overselective subjects to respond to simultaneous multiple cues, A multiple schedule was used in a two-lever procedure in order to establish independent stimulus-response classes with single and multiple cues. The results showed that when reinforcement and timeout were contingent on the occurrence of specific stimulus-response relations, both subjects learned to respond differentially in the presence of multiple vs. single elements signaling the contingencies in multiple schedule components. Extinction and timeout were examined to assess the relative efficiency of each in decreasing errors and establishing discriminative 47

PAGE 55

48 responding on the right and left levers in the presence of single and multiple elements, respectively. Following training under single elements, a compound stimulus composed of the combined single elements was introduced. Although extinction was in effect for responding on the right lever in the compound components, both subjects continued to respond on this lever for over 180 presentations of the compound at rates equivalent to those maintained on the same lever by the VI 20 sec. schedule in the single element components. Furthermore, with few exceptions, left lever responding remained at or near zero levels. Since the elements in the compound stimulus were in all respects equivalent to the single elements in the alternate components, it is not surprising given the subjects' reinforcement histories under single elements, that either or both elements of the compound continued to function as a discriminative stimulus for right lever responding. This is consistent with previous findings which showed that once a stimulusresponse relation has been established, subjects may continue to respond for hundreds of trials under extinction in the presence of a J single element if reinforcement is concurrently delivered in the presence of a compound containing the same element (Koegel & Schreibraan, 1977), Moreover, in the present case, the reinforcement rates under the more frequently interspersed single elements (i.e., 2:1 single element to compound stimulus component ratio) were probably sufficient to maintain right lever responding in the presence of the compound stimulus signaling extinction. These results are congruent with those reported by Bijou (1961); in his early studies using multiple schedules with retarded children, responding in S-delta periods showed considerable resistance to extinction.

PAGE 56

49 Manipulating the relative number and distribution of compound stimulus components in a single session did not significantly alter switch-over rates. Similar manipulations have been effective in establishing rapid acquisition of multiple schedule discrimination performance with developmentally delayed subjects (Orlando & Bijou, 1961), and it is possible that with further sessions these manipulations would have been effective with the present subjects as well. However, after one session, minimal gains in terms of switch-overs were obtained with Sandy, and none with Davm. Moreover, both subjects' response rates on the extinction lever were maintained at previous levels. It therefore seemed advisable to examine other procedures that might prove more efficient in generating switch-overs, while retaining the same schedule contingencies. Using a lever cover to "force" switch-overs was not effective with Dawn. As might be expected from studies using prompts (Koegel & Rincover, 1976; Schreibman, 1975), employing an added stimulus appeared to interfere with the acquisition of stimulus control by the designated « discriminative stimulus, and attempts to transfer stimulus control from the lever cover to the training stimulus were ineffective. Although the procedure was more effective with Sandy, when the cover was removed, she continued to make alternate responses on the extinction lever across approximately 85 compound component presentations. Sandy's errors decreased quickly when timeout was introduced, and she required a total of nine timeout periods to master the discrimination. Dawn, on the other hand, required 47 timeout periods before performing without errors. For both subjects, two-thirds of the timeout periods were delivered for errors in the compound stimulus

PAGE 57

50 components, indicating that even after the introduction of the punishment contingency, the elements of the compound continued to exert considerable control over right lever responding. Cook and Rincover (1979) liave pointed out that autistic and very young normal children (i.e., matched on "mental age") tend to use rigid response strategies in discrimination tasks involving multiple responses, and often fail to switch to alternative response strategies called for by the task. A tendency towards response perseveration was also evident with these subjects, particularly Dawn. However, with the timeout contingency, the number of stimulus cues and response classes sampled by both subjects reliably increased. The disparity between subjects in the amount of training required to reach mastery level was significant, though not unusual (e.g., Koegel & Schriebman, 1977). More importantly, the different speeds of acquisition should not overshadow the central issue addressed by the procedures and featured in the results — the course and pattern of acquisition was similar for both subjects. Following mastery of the original discrimination, acquisition of subsequent discriminations involving variations of the original color and line-tilt dimensions progressed more rapidly. Sandy performed on the last four without errors in stimulus control test sessions using training schedule contingencies. Unlike Sandy, Dawn's performance was disrupted when two color/line-tilt combinations were presented within the same training sessions. It is interesting that the stimulus control exerted by the original training stimuli remained intact while the novel stimuli were initially introduced as probes. However, when the training schedules were in effect for all six stimuli. Dawn's errors in the presence of the original training stimuli increased by a factor of

PAGE 58

51 1.7 over baseline. She received a total of 115 timeouts over 28 sessions across the six components. However, when each set of stimuli was presented separately, she reacquired the original discrimination after eight timeout periods, and the second discrimination after seven. While Sandy's performance strongly indicated that she had learned to respond on the basis of multiple vs. single cues, the data on Dawn were more equivocal. On the one hand, there was a substantial decrease in errors during reacquisition of the original discrimination, and acquisition of the second (17% and 13% of the original rate of responses followed by timeout, respectively). Moreover, errors remained at or near zero levels during acquisition of the third discrimination, throughout which, she received a maximum of one timeout per session. On the other hand, if she had learned to respond on the basis of multiple vs. single cues, the dramatic increase in errors observed when two sets of tirauli were presented in the same sessions would not have been expected to occur. It is possible, however, that the disruption of stimulus control in the latter case was due to added 5 variability in potential stimulus classes that resulted when two sets of stimuli were interspersed. With a single combination of stimuli, both the color and line-tilt dimensions were constant, and the only variant across each single element and the corresponding compound was the absence/presence of one cue. However, when a second set was introduced, the constant dimensions became variable. In this case, there were two variants across each compound and the single elements from the alternate set; i.e., presence/absence of one cue plus the varying value of the color/line-tilt dimensions. Thus, the number of controlling stimulus-response relations that may have been "inadvertently" acquired

PAGE 59

52 by the subject in the course of training was increased. The disruption in Dawn's performance during this condition need not suggest the failure to acquire the discriminations on the basis of single vs. multiple cues, but may reflect instead the course of acquisition of new controlling stimulus-response relations generated by the increased variability in potential stimulus classes. There were three questions of interest underlying this experiment. The first was whether discrimination performance to simultaneous multiple cues could be established using multiple schedules with otherxtfise overselec tive subjects. In a multiple schedule there are two or more component schedules of reinforcement, each associated with its distinctive discriminative stimulus. The results showed that when stimulus-response relations involving simultaneous multiple cues were directly controlled by arranging appropriate consequences for their occurrence in a multiple schedule, the experimentally designated stimulus-response relations were acquired by these subjects. The second question concerned the identification of schedule conJ tingencies that would most rapidly reduce errors, thus increasing the cost efficiency of the procedure. The results showed that errors were maintained at levels equivalent to reinforced rates when extinction was in effect, but were reduced quickly when two-minute timeout periods were substituted. It is likely that the cost efficiency of the procedure would be further increased by arranging a three-component multiple schedule at the onset of training, and eliminating the reinforcement histories under single cue training. Exposing subjects to the reinforcement and punishment contingencies associated with each stimulus-response relation concurrently may encourage more rapid

PAGE 60

development of stimulus control by the compound stimulus and enhance the efficiency of the procedure. In the present experiment, this procedural variation may in fact have contributed to the increased speed of acquisition of each successive discrimination following the original . The final question of interest was whether the subjects had acquired specific stimulus-response relations, or whether they had learned to discriminate on the basis of single vs. multiple cues given novel element combinations. With one subject the results clearly affirmed the latter. The results were not as clearly supportive with the second subject, although her increased proficiency in acquiring successive discriminations suggested some degree of generalization. The most obvious implication suggested by the present findings is tliat the overselective responding of developmentally delayed children in discrimination tasks involving multiple cues can be modified, and the number of stimulus response relations acquired increased, by J arranging appropriate reinforcement histories. An important, yet perhaps subtle point to note here is that, in order to deliver appropriate consequences following the occurrence of a stimulus-response relation, one must first be able to observe and measure the relation directly. When reinforcement is delivered for a single response only in the presence of a compound stimulus containing two cues, it is impossible to detect the specific stimulus-response relation(s) which is actually reinforced. In other words, it is impossible to track the development of stimulus control. In arranging appropriate consequences

PAGE 61

54 for designated stimulus-response relations, the subjects in the present experiment were required to engage in two distinct response topographies such that, the occurrence of these relations could be observed and measured independently, and could thus be reinforced directly. The use of multiple response topographies seems advisable in training overselective subjects to respond to multiple vs. single cues, since it affords the opportunity to observe and measure the development of correct and incorrect response classes early in training, and to insure their incompatibility by arranging direct consequences for their occurrence. Perhaps the most significant finding reported in this study was the greater effectiveness of timeout as compared to extinction in decelerating errors. In this procedure, responding on the extinction lever in the compound components was always ultimately followed by a stimulus change and reinforcement for responding in the single element components. It is possible tliat the increased intermittency of reinforcement on this lever served to increase resistance to extinction for t responding on the lever in the compound components. In contrast, timeout produced a rapid and striking decrease in errors, and was clearly the more cost-efficient procedure. It appeared that with timeout, each stimulus acquired a dual function: a discriminative function for a specific response class(es), and a punishing function for the alternate response class(es). Under extinction, on the other hand, the designated response classes failed to acquire independent status. Moreover, after the subjects mastered the original discriminations, the rates of responses followed by timeout in each component across successive discriminations remained below .5 responses per

PAGE 62

55 minute. The only exception was Dawn's performance when two stimulus sets were presented together within the same sessions. But, given the variability in potential stimulus classes occasioned by this manipulation, this exception might be attributed to the development of unobserved and unmeasured stimulus-response relations. The subjects' improved accuracy with timeout is particularly interesting when one considers that in all probe conditions which preceded training, error rates in extinction occurred at levels equal to correct rates. The differential sensitivity to extinction and timeout observed in the subjects' responding was illustrated most dramatically by their perform.ance on the final discriminations which were not preceded by probes. Sandy performed on the last four without errors, and in her third and final discrimination Dawn received a maximum of one timeout per session for errors. It is tempting to infer from these results that the efficiency of the training contingencies was increased when the exposure to extinction as a means of testing for stimulus control and the error rates therein obtained, were eliminated, HowJ ever, the confidence of this inference is made questionable by the order in which experimental conditions were introduced, since the increased accuracy in performance in the later phases may also have been effected by cumulative training history. This question could be addressed directly by introducing training contingencies not preceded by probes earlier in the sequence, and conversely, by introducing probes prior to training in the later phases. It is possible that by eliminating extinction components and associated error rates one might increase the efficiency of timeout in decelerating errors during training. This would be doubly significant, and would imply that its

PAGE 63

56 effectiveness in arranging for generalization could be concurrently maximized . The influence of extinction probes on the subsequent course of acquisition of stimulus control seems particularly critical in light of recent findings using errorless discrimination training procedures. Fields (1978, 1979) reported that when probes in extinction were used in conjunction with fading procedures to measure acquisition of stimulus control by the new stimuli, the probes had a positive influence on the process of acquisition. Moreover, the earlier introduction of probes during fading resulted in earlier acquisition, whether the probes were presented under extinction (Fields, 1978, 1979) or reinforcement conditions (Fields, 1981). Wliile these findings may appear to conflict with the suggestions of the present results, they may in fact lend support to our interpretation. In the fading procedures, the same reinforcement contingencies were correlated with the compound stimulus and both single elements, such that the stimulus-response relations in all tliree instances were # topographically and functionally equivalent. Throughout fading, therefore, correct responses during extinction probes were always followed by a training trial, that is, by reinforcement for responding in the presence of the novel element. Thus, differential reinforcement of correct stimulus-response relations probably contributed to the enhancing effects of the probes. In contrast, the stimulus-response relation reinforced during compound stimulus training in the present case was topographically and functionally incompatible with the previously established single element-response relations. In this case, incorrect controlling

PAGE 64

57 relations during extinction probes were differentially reinforced in a similar fashion. Whether extinction probes have a facilitative or retarding effect on acquisition of stimulus control may be determined by the subjects' reinforcement history with respect to errors, and the compatibility of this history v;ith stimulus-response relations correlated with reinforcement during subsequent training.

PAGE 65

APPENDIX

PAGE 66

u-1 0-) O O tH o m o O O CN O O ooooo ooooo 00 r~N 00 i-H CN «-l .H CM m C^T iH CN 00 O fO CN n >H cjN r-^ cTi m CN CN CN m i-H u-1 >^ 00 < CN cn CM CN| m T-t C> r-i CM CO O rH O LO O ooooo ooooo CN cT\ 00
PAGE 67

60 TO T) C 3 O CJ Q. a; )_i o 1-1 o o I — 1 to M c o N •H 4-1 )^ tj o 0) ^ 1 c o Oi c a* u c o o p. g o o u 4J c •H aj g H cn 0 (J «; rH c CO M 4J c o TO M 4J PC •H o 0) o CO c o o u D. M OJ PS C to (U r-i r-H 00 r-H in (N 1 r-l ^ 00 4 rH CM CM rH rH cy>
PAGE 68

61 C 3 •u o (X 0) E )-< O M CJ o u iH c cd M u c o N •H u o a 1 c o 0) t-l a u c u •H q; e c •H u to hJ o o rH C ns M 0) C o N U •H O ^ (U 0) O 0) 25 M c o o a. tn 01 ci C Q a OJ (-1 o o C o 0) M O O H O O CO On cvj CN H in vo m cN CO CO cNi r-i On 00 CTi t-H CNi ON r-00 -H CO CO OOOOO OOOOO CM cNi OO OOOOO OOOOO ON CM CM ON in rH CM CM rH CM 00 o o in <3CM CM CO M CM CO in o o CO CM CO CO H rH NO -i OO in • CO CO OOOOO OOOOO in in CM CNl OOO OOOOO OOOOO 00 O CM vO ON i-l CM
PAGE 69

62 T3 C D u O a a tu S o !-( O a c to M u C o M •H 4-1 a O 01 1 C o in ^ CO cn CO CO d in r-H o o n -. r-^ 00 rH CM CO rH CM fo -din 0^ 0> C3> t7^

PAGE 70

coooo ooooo oooo M in m CM m -vT -( m o -H ^ CO CN -vt lA iH LTi O 0> 0~. CTi o^ O i-i CM r^, O O C o o vT) a.i cTv oooo rH 1-1 rH rH c 3 o a e o o •H > 1 0) w C til u o 01 c •H J CO o •H W 1-1 0) > o •l-t u c CO M O lo m LTi ro i-H o o CO cn o r^. 00 o 1^ i^, CO

PAGE 71

64 c •-4 0 o a 0) e Vi c ^-1 o o o c M o •H 4J V( 4-1 0) u > 01 1 u 0) u o c u CO o u u 01 o u c o 0) u u o a 0) s-< o a c 0) o coooo ooooo ooooo ooooo ooooo o a> c c^ ON 4 ^ O rHCTiLTiincyv ONcrivDioco rOCMCM<}-4 m CO 00 00 n »H iH m r~ cN o CM in CM n CO r-i 1^ CM i in CTi On cyv ON OS ^ r~ 00 cTv o On ON CTv C3N O iH CM m
PAGE 72

o o o o 00 \0 Lfl iH o o o o HI r; D O a 6 O u H I ni o in CO H CM CO 00 o CTN CO 00 O 00 CO 00 00 a^ rH CM CO
PAGE 73

66 -o r-~ 00 a\ OOOO (0 c 3 o a, e o u u H I in in in ro rO iH ( si CN! 1^ O M CN n CM 10 CO 0^ o 0^ (7^ 0> CTi O •H
PAGE 74

73 T) c C 3 o 3 o 01 O D. cx 6 o o B o o o o u u c 1— 1 o H I H 1 Si 1-) o 4.1 0 00 a 00 to a c to

PAGE 75

O H O CN CM o o m o o o o o o o o o o o o cNooOfOm irimcs>^>(N oor--t-^OCN CT>r^ooo •Jfonro (Nrofnnn mrocnron cnron-^j-^r in CO m ro .CMOO • -oo -o ooooo ooooo o-)vDoo
PAGE 76

69 Cm O o CU
  • H o CJ o u c CO o •H ) 1 t-l 4-1 0) CI > OJ 1 u •1> u 60 o C u to o 00 r-l .0 IT) !/> m -vT lO r~CTi C^J in r-^ <}• vo m CNi H r~s ctn -a-am cN CO OOOOO coooo oooco oooco ooooo 0> ro On n CO iH o in f) in vdin o> ON 00 m o in iH
    PAGE 77

    0Oi^r^iHLO00 >£>n oooo iriuoLnmo

    PAGE 78

    71 Q (J B Q M Q C ^ •H U u U > 0) 1 q; o C o to o o u u o u c o >^ o u o o c a u o ooomcM ooooo ooooo ooooo omr-ivOLn 00 CT\ r-t en n fn n 00 vD O in On ro tn 1^ O i-t lO m lo in
    PAGE 79

    72 c 3 •u O u a. 01 B u o u o u 1—1 c M u •H at a > (U 1 u 0) u o § u u o • c u B (1) 1-1 (0 c o •H o OS c l-l (U CO 4-1 o nj •H 4-1 03 U u u (U 0) u m > u c o o Ou w OJ PS (0 -( o y o o u-1 n 00 CO lA 00 lO O r- CT\ CM CM CM t-H r-l CM CM CM m m m <) CO CO CO
    PAGE 80

    OOOOiH r-(f--1000 O . 00 in csj CM cn i-i n 00 00 C O N •H O 32 U •H O C 0) (N O O CM in rH ,D CM cn -
    PAGE 81

    74 T3 C u O o 0) e § i-1 o a M c O N •H *J U O 0) 35 1 S-i C o 01 0) )-l o O O O O csi ooooo ooooo ocoo 1-1 u-i n r-l
    PAGE 82

    CJ 3 4-1 o o a. B o (_) o o rH c to M u C o N •H U o o 0) a: V4 1 c O CJ (U o o 0) P4 4-1 c CJ •H <3^ CN tn CM CM CO CO o o o o o o o vD
    PAGE 83

    c a o a c 3 O a. e o u H I in -< o o u u o u u >-l »-l o u rH O O O O O in 00 i-H O
    PAGE 84

    REFERENCES Bailey, S. L. Stimulus overselectivity in learning disabled children. Journal of Applied Behavior Analysis , 1981, 3^, 239-248. Bijou, S. W. Discrimination performance as a baseline for individual analysis of young children. Child Development , 1961, 32^, 163-170. Bijou, S. W. , & Baer, D. M. Operant methods in child behavior and development. In W. K. Honig (Ed.), Operant behavior: Areas of research and application . New York: Appleton-Century-Crof ts, 1966. Bijou, S. W. , & Orlando, R. Rapid development of multiple schedule performances with retarded children. Journal of the Experimental Analysis of Behavior , 1961, 4_, 7-16. Fields, L. Fading and errorless transfer in successive discriminations. Journal of the Experimental Analysis of Behavior , 1978, 30, 123-128. Fields, L. Acquisition of stimulus control while introducing new stimuli in fading. Journal of the Experimental Analysis of Behavior , 1979, 32, 121-127. Fields, L. Enhanced learning of new discriminations after stimulus fading. Bulletin of the Psychonomic Societ y, 1980, 15, 327-330. i Fields, L. Early and late introduction of probes and stimulus control acquisition in fading. Journal of the Experimental Analysis of Behavior, 1981, 36, 363-370. Fields, L. , Bruno, V., & Keller, K. The stages of acquisition in stimulus fading. Journal of the Experimental Analysis of Behavior , 1976, 26, 295-300. Fischer, M. A., & Zeaman, D. An attention-retention theory of retardate discrimination learning. In N. R. Ellis (Ed.), International review of research in mental retardati on. New York: Academic Press, 1973. Hale, G. A., & Morgan, J. S. Developmental trends in children's component selection. Journal of Ex p erime ntal C hild P sycho logy , 1973, 15, 302-314. ' 77

    PAGE 85

    78 Huguenin, N. H. , & Touchette, P. E. Visual attention in retarded adults: Combining stimuli which control incompatible behavior. Journal of the Experimental Analysis of Behavi or, 1980, 33^, 77-86. Johnson, D. F., & Gumming, W. Some determiners of attention. Journal of the Experimental Analysis of Behavior , 1968, 11^, 157-166. Johnston, J. M. , & Pennypacker, H. S. Strategies and tactics of human behavioral research . Hillsdale, New Jersey: Lawrence Erlbaum Associates, 1980. Koegel, R, L. , & Rincover, A. Some detrimental effects of using extra stimuli to guide responding in autistic and normal children. Journal of Abnormal Child Psychology , 1976, j4, 59-71. Koegel, R. L., Rincover, A., & Egel, A. L. Educating and understanding autistic children . San Diego, California: College-Hill Press, 1982. Koegel, R. L., & Schreibman, L. Teaching autistic children to respond to simultaneous multiple cues. Journal of Experimental Chil d Psychology , 1977, 24, 299-311. — — Koegel, R. L., & Wllhelm, H. Selective responding to the components of multiple visual cues by autistic children. Journal of Experi mental Child Psychology , 1973, 15, 442-453. — — Koorland, M. A., & Wolking, W. D. Effect of reinforcement on modality of stimulus control in learning disabled students. Learning Disabilities , 1982, 5, 264-273. Lashley, K. S. An examination of the "continuity theory" as applied to discrimination learning. Journal of General Psychology, 1942 18 I 241-265. ^ ' _» LoLordo, V. M., & Furrow, D. R. Control by the auditory or the visual element of a compound discriminative stimulus: Effects of feedback. Journal of the Experimental Analysis of Beh avior, 1976, 25, 251-256. Lovaas, 0. I., Koegel, R. L., & Schreibman, L. St imulus overselectivity in autism: A review of research. Psychological Bulletin , 1979, 86, 1236-1254. Lovaas, 0. I., & Schreibman, L. Stimulus overselectivity of autistic children in a two stimulus situation. Behav iour Research and T herapy , 1971, 9^, 305-310. Lovaas, 0. I., Schreibman, L., Koegel, R. L., & Rehm, R. Selective responding by autistic children to multiple sensorv input. Journal o f Abnomal Psycholog y. 1971, 77_, 211-222. ' Mackintosh, N. J. The effect of attention on the slope of generalization gradients. British Journal of Psychology , 1965, 5-6, 87-93.

    PAGE 86

    79 Olson, D. R, Information-processing limitations of mentally retarded children. American Journal of Mental Deficiency , 1971, 75_, 478-486. Ray, B. A. The course of acquisition of a line-tilt discrimination by rhesus monkeys. Journal of the Experimental Analysis of Behavior , 1967, 10, 17-33. Ray, B. A. Selective attention: The effects of combining stimuli which control incompatible behavior. Journal of the Experimental Analysis of Behavior , 1969, 12, 539-550. Ray, B. A., & Sidman, M. Reinforcement schedules and stimulus control. In W. N. Schoenfeld (Ed.), The theory of reinforcement schedules . Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1970. Reynolds, G. S. Attention in the pigeon. Journal of the Experimental Analysis of Behavior , 1961, 4^, 203-208. Reynolds, B. S., Newsora, C. D., & Lovaas, 0. I. Auditory overselec tivity in autistic children. Journal of Abnormal Chil d Psychology, 1974, 2, 253-263. ' Rincover, A., & Koegel, R. L. Setting {generality and stimulus control in autistic children. Journal of Applied Behavior Analysis , 1975, 8, 235-246. ~ ~ ' Schover, L. R. , & Newsom, C. D. Overselectivity, developmental level, and overtraining in autistic and normal children. Journal of Abnormal Child Psychology , 1976, 4_, 289-298. Schreibman, L. Effects of wi thin-stimulus and extra-stimulus prompting on discrimination learning in autistic children. Journal of Applied Behavior Analysis , 1975, 8^, 91-112. I Schreibman, L., Koegel, R. L., & Craig, M. S. Reducing stimulus overselectivity in autistic children. Journal of Abnormal Child Psychology , 1977, 5, 425-436. — — Schreibman, L., & Lovaas, 0. I. Overselective response to social stimuli by autistic children. Journal of Ab normal Child Psychology, 1973, 1, 152-168. Segal, M., & Harrison, J. M. The control of responding by auditory stimuli: Interactions between diffarent dimensions of the stimuli. Journal of the Experimental Analysis of Behavior , 1978, 30, 97-106. Singh, N. N., & Beale, I. L. Attentional changes during discrimination learning by retarded children. Journal of the Experimental Analysis of Behavior , 1978, 29, 527-533. Skinner, B. F. The behavior of organisms: An experimental anal ysis New York: Appleton-Century-Crof ts , 1938.

    PAGE 87

    80 Terrace, H. S. Stimulus control. In W. K. Honig (Ed,), Operant behavior: Areas of research an d application. New York: AppletonCentury-Crof ts, 1966. Varni, J. W. , Lovaas, 0. I., Koegel, R. L., & Everett, N. L. An analysis of observational learning in autistic and normal children. Journal of Abnormal Child Psychology , 1979, 31-43. Warren, J. M. Additivlty of cues in a visual pattern discrimination by monkeys. Journal of Comparative Physiological Psychology , 1953, 46, 484-486. Wiilielm, H., & Lovaas, 0. I. Stimulus overselectivity : A common feature in autism and mental retardation. American Journal of Mental Deficiency , 1976, 81, 227-241. ~~ Wilkie, D. M. , & Mason, M. E. Attention in the pigeon: A reevaluation. Journal of the Experimental Analysis of Behavior , 1976, 26 , 207-212. Wyckoff, L. B., Jr. The role of observing responses in discrimination learning. Psychological Revie w, 1952, 59^, 431-442. Zeaman, D. One programmatic approach to retardation. In D. K. Routh (Ed.), Th e experimental psychology of mental retardation .. Chicago: Aldine, 1973. 1

    PAGE 88

    BIOGRAPHICAL SKETCH Maria del Rosario Ruiz was born in La Habana, Cuba, on July 7, 1950, and is the daughter of Dr. Guillermo Vicente Ruiz and Dr. Maria Casas Ruiz. She left La Habana in August 1961 and entered the United States of America where she has resided over the past 21 years. The author holds the Master of Education and the Master of Arts degrees from the University of Florida. During her tenure as a doctoral student at the same institution, she has worked as special education teacher for learning disabled children, behavioral trainer/ consultant for Sunland, Gainesville, and outpatient drug abuse treatment counselor for a research facility sponsored by the College of Pharmacy at the University of Florida. During the past two years, she has been a doctoral-teaching fellow at Rollins College, where she is cftrrently Assistant Professor of Psychology. 81

    PAGE 89

    I certify that I have read this study and that ±Ti-«y opinion it conforms to acceptable standards of scholarly presentatioa, and is fully adequate, in scope and quality, as a dissertation for the ^egree of Doctor of Philosophy. lenry S. Pennv6acker, Jr., Chairpersoi' Professor of Psychology I certify that I have read this study and that in ray opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. • YvAnne Brackbill Graduate Research Professor of Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. ^ Aark K.' Goldstein Associate Professor o^ Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Psychology I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for^ the degree of Doctor of Philosophy.

    PAGE 90

    I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as Doctor of Philosophy. William D. Wolking Professor of Spec ial(^ucat ion This dissertation was submitted to the Graduate Faculty of the Department of Psychology in the College of Liberal Arts and Sciences and to the Graduate Council, and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. I dissertation tor the degree ot DcccniluT 19fl2



    PAGE 1

    ANALYTICAL AND EXPERIMENTAL INVESTIGATION OF THE EFFECT OF DYES ON SOLAR DISTILLATION By ANIL KUMAR RAJVANSHI A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1979

    PAGE 2

    ACKNOWLEDGMENTS The author wishes to express his sincere appreciation and gratitude to Dr. E.A. Farber, Chairman of his supervisory committee, for directing this research. He wishes to thank the other members of his committee, Dr. C.K. Hsieh, Dr. C.C. Oliver, Dr. R.K. Irey and Dr. A.K. Varma, for their cooperation in serving on the committee. The author gratefully acknowledges the financial assistance of the Government of India, which made his stay at the University of Florida possible. He is grateful to Dr. C.K. Hsieh for allowing him to use the Perkin -Elmer Monochromator for measuring the absorption spectrum of the dyes. He would also like to thank Dr. P. Buscemi for allowing him to use the Beckman Spectrophotometer. Special thanks are extended to the Mechanical Engineering workshop for building the solar still. The author wishes to thank his fellow graduate students for numerous helpful suggestions and wishes them luck in their future endeavors. He also wishes to thank Lorraine Thomas for the excellent job she did on typing this manuscript. Finally, the author extends his thanks to his wife, Nandini, whose active support made this work both possible and worthwhile. ii

    PAGE 3

    TABLE OF CONTENTS Page ACKNOWLEDGEMENTS LIST OF TABLES ^ LIST OF FIGURES KEY TO SYMBOLS ^ ABSTRACT ^^^^ CHAPTER I INTRODUCTION 1 II LITERATURE REVIEW 6 Historical Introduction 6 Solar Radiation ^ Brine Depth ^ Cover Material and Its Shape 10 Ambient Temperature and Wind Velocity 12 Temperature of the Condensing Surface 14 Use of Dyes to Enhance Solar Evaporation 16 Scope of the Present Investigation 18 III TrlEORETICAL ANALYSIS . . • 19 Physical Model 19 Incident Solar Radiation 24 Absorption of Radiation 25 Conduction and Convection Heat Transfer in DyeWater System 28 Evaporation and Convection Loss from the Water Surface 29 Formulation of Equations 35 Layer n 35 Top Layer 35 Bottom Surface 36 Energy Exchanger with the Still Covers 36 During Daytime 38 During Nighttime 38 Method of Solution 38 iii

    PAGE 4

    IV EXPERIMENTAL SETUP AND PROCEDURES 42 Distillation Units '^2 Thermocouples and Their Locations 47 Temperature and Solar Radiation Measurements .... 47 Absorption Spectrum of Dyes 51 Experimental Procedure 53 V RESULTS AND DISCUSSIONS 59 Dye-Absorption Spectrum Results 59 Temperature-Time History of the Stills 74 Productivity of the Still 96 Comparison of Different Dyes 116 Comparison of Theoretical and Experimental Results . 137 Effect of Various Variables on Still Productivity . . 154 Effect of Ambient Temperature 154 Effect of Wind Velocity 159 Effect of Dye Concentration 161 VI CONCLUSIONS AND RECOMMENDATIONS 167 Conclusions 167 Recommendations for Future Investigations 169 APPENDICES I PROPERTY VALUES USED IN ANALYTICAL MODEL 172 II STABILITY CRITERION FOR DIFFERENCE EQUATIONS 175 III CALCULATION OF ABSORPTION COEFFICIENT OF DYE SOLUTION . 179 REFERENCES 182 BIOGRAPHICAL SKETCH 186 iv

    PAGE 5

    LIST OF TABLES Table Page 1 Duration of Test 58 2 Comparison of Dyes (Experimental) 101 V

    PAGE 6

    LIST OF FIGURES Figure Pafie 1 Schematic diagram of solar still 4 2 Effect of brine depth on still productivity 9 3 Single sloping solar still 13 4 Model of still used in analysis 20 5 Radiation transfer in the dye-water solution 26 6 Temperature profile of the layers 40 7 Flow diagram of the computer program 41 8 Experimental stills ^3 9 South view of the still '^3 10 Schematic of the glass-cover frame 44 11 Schematic of the constant head feed system 46 12 Thermocouple locations on still 48 13 Attachment of thermocouple on glass cover 49 14 Location of thermocouples in dye-water solution .... 50 15 Beckman UV-VIS spectrophotometer . 52 16 Perkin-Elmer Monochromator 54 17 Specimen cell 54 18 Cross section of the specimen cell used in PerkinElmer Monochromater 55 19 Arrangement of the two stills 56 20 Absorption spectrum of Red dye (50 ppm) 61 vi

    PAGE 7

    21 Absorption spectrum of Red dye (100 ppm) 63 22 Absorption spectrum of Black dye (50 ppm) 65 23 Absorption spectrum of Black dye (172.5 ppm) 67 24 Absorption spectrum of Green dye (50 ppm) 69 25 Absorption spectrum of Green dye (100 ppm) 71 26 Temperature-time history of Red dye (50 ppm); November 26, 1977 76 27 Temperature-time history of Black dye (50 ppm); March 6, 1978 78 28 Temperature-time history of Black dye (172.5 ppm); March 31, 1978 80 29 Temperature-time history of Green dye (50 ppm); May 15, 1978 82 30 Temperature-time history of Black dye (172.5 ppm); April 14, 1978 84 31 Temperature-time history of the two stills; March 31, 1978 86 32 Percentage of solar energy absorbed with depth of solution; April 14, 1978 at 12:30 p.m. (E.S.T.) .... 88 33 Temperature profile of dye-water and control still; March 31, 1978 91 34 Temperature-time history of the two stills; April 14, 1978 93 35 Temperature-time history of the two stills; May 15, 1978 95 36 Distillate output from still with dye and control still; April 15, 1978 98 37 Distillate output from still with dye and from control still; May 15, 1978 100 38 Evaporation heat transfer-time history of Black dye (172.5 ppm); April 15, 1978 104 39 Evaporation heat transfer-time history for control still; May 15, 1978 105 vii

    PAGE 8

    40 Distillate output from Black dye (172.5 ppm) ; April 14, 1978 107 41 Correlation between h and T 110 evap s 42 Correlation of distillate output and T Ill s 43 Analysis of energy transfer mechanisms for Black dye (172.5 ppm); April 14, 1978 113 44 Analysis of the energy transfer mechanisms for water; April 14, 1978 115 45 Correlation between the performance of the two stills . 118 46 Comparison of outputs of Red dye (50 ppm) 119 47 Comparison of distillate outputs for Red dye (100 ppm) . 120 48 Comparison of distillate outputs for Black dye (50 ppm) 122 49 Comparison of distillate output for Black dye (172.5 ppm) 124 50 Comparison of distillate output for Green dye (50 ppm) . 126 51 Comparison of distillate output for Green dye (100 ppm) 128 52 Correlation of S and X 131 53 Correlation between slope S and intercept b 133 54 Seasonal variation of the distillate from control still 135 55 Seasonal variation of the distillate for various dyes 136 56 Distillate output from two stills on completely cloudy days 139 57 Absorbance spectrum of distillate from various dyes . . 140 58 Comparison between theoretical and experimental results for Red dye (50 ppm); November 26, 1977 .... 143 59 Comparison between theoretical and experimental results for Black dye (50 ppm); March 6, 1978 145 60 Comparison between theoretical and experimental results for Black dye (172.5 ppm); March 31, 1978 ... 147 viii

    PAGE 9

    61 Comparison between theoretical and experimental results for Black dye (172.5 ppm) ; April 14, 1978 ... 149 62 Comparison between theoretical and experimental results for Green dye (50 ppm); May 15, 1978 151 63 Comparison between theoretical and experimental temperature profiles for Slack dye (172.5 ppm); March 31, 1978 153 64 Analytical plot of the effect of ambient temperature on distillate output; March 31, 1978 156 65 Analytical plot of the temperature-time history for Black dye (172.5 ppm), March 31, 1978, for two ambient temperatures 158 66 Analytical plot of the effect of wind speed on distillate output; March 6, 1978 160 67 Analytical plot of the effect of dye concentration on distillate output; April 14, 1978 162 68 Analytical plot of the temperature-time history for Black dye, April 14, 19 78, for two different dye concentrations 163 69 Optimum concentrations for different dyes on a typical spring day at Gainesville, Florida 165 ix

    PAGE 10

    KEY TO SY>ffiOLS 2 Area of still cover, ft 2 Area of glass cover, ft 2 Area of water surface, ft Intercept of regression lines in Figures 46 through 51, Ibs/f t^-day Concentration of dye, ppm Specific heat, Btu/lb°F 2 Distillate output [equation (43)], lbs/ft hr 2 Binary diffusion coefficient, ft /hr 2 Factor given by equation (50), lbs/ft -day Emissivity and geometrical shape factors 2 Acceleration due to gravity, ft/s Grashoff Number Convection heat transfer coefficient from water surface to glass cover, Btu/ft2hr°F 2 Evaporation heat transfer coefficient, Btu/ft hr°F Enthalpy of vaporization for water, Btu/lb 2 Mass transfer coefficient, lbs/ft -hr 2 Outside heat transfer coefficient, Btu/ft hr°F Percentage increase in evaporation Equivalent conductivity, Btu/ft-hr-°F Thermal conductivity of Plexiglas, Btu/ft-hr-°F

    PAGE 11

    Thermal conductivity of water, Btu/ft-hr-°F k / Thermal conductivity of air-water mixture, Btu/ft-hr-°F w/a ' L Thickness of layer for convection heat transfer in dyewater solution, ft. L ' Distance between the water surface and glass cover, ft. m. . Mass fraction of species i near surface i 2 m^ Distillate output from control unit, lbs/ft -day 2 Distillate output from still with dye, lbs/ft -day 2 M Distillate output, lbs/ft -day M_j^ Molecular weight of material i, lbs/mole n^ Index of refraction for air -water vapor mixture Vi^ Index of refraction for dye-water solution Partial pressure of water near surface i, psia P Prandtl Number r 2 q^j^g(t) Solar radiation absorbed in a layer, Btu/ft -hr Heat transfer by convection, Btu/hr 2 Spectral solar energy flux, Btu/ft -hr-ym 2 qii(t) Spectral solar radiation on a horizontal surface, Btu/ft -hro ym q^ ^(t) Spectral solar radiation passing the air-water interface, ' Btu/f t^-hr-ym n 2 q^ ^(t) Spectral solar radiation incident on layer n, Btu/ft -hr-ym 2 q^Ct) Incident solar radiation on glass cover, Btu/ft -hr 2 q(t) Total solar radiation on a horizontal surface, Btu/ft -hr r Reflectance Ra^ Raleigh Number R Universal gas constant, ft Ib^ /lb-mole °R S Slope of the regression lines in Figures 46 through 51 S Schmidt Number c xi

    PAGE 12

    Ambient temperature, °F Temperature of hot and cold layer, respectively, °F Temperature of layer n, °F Temperature of the water surface and glass cover, respectively, °F Apparent sky temperature, °F 2 Thermal diffusivity of water, ft /s Solar absorptivity of glass Absorption coefficient of the dye solution, 1/ft Thermal coefficient of volume expansion, 1/R° Thickness of layer, ft Path length of incident solar radiation ray in a layer, ft Dimensionless parameter in equation (10) Normalized absorption given by equation (44) Wavelength (\im) Upper and lower limits of wavelengths of solar spectrum, ym Efficiency of distillation 2 Kinematic viscosity of water, ft /s Average transmittance of condensate covered glass Angle of incidence and refraction at air-water interface, degrees 3 Density of fluid, lb/ft xii

    PAGE 13

    Abstract of Dissertation Presented to the Graduate Council of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ANALYTICAL AND EXPERIMENTAL INVESTIGATION OF THE EFFECT OF DYES ON SOLAR DISTILLATION By Anil Kumar Rajvanshi June 1979 Chairman: Erich A. Farber Cochairman: C. K. Hsieh Major Department: Mechanical Engineering An analytical and experimental study of the effect of dyes on solar distillation was conducted in this study. The analytical model developed treated the transient heat transfer inside the dye-water system as one dimensional. The bulk fluid was discretized into layers with conduction, convection and radiation interactions occurring between them. Coupled with this was the energy exchange from the surface of water to the glass cover by convection, radiation and evaporation of water. The bottom surface of the distiller was assumed to be insulated and a black absorber. A finite difference technique was used in solving the nonlinear partial differential equations. In the computer program the measured solar radiation, ambient temperature and the wind speed was used as input; the output from the program was the temperature-time history of the water layers and glass cover and the distiller output. xiii

    PAGE 14

    The experimental investigation was conducted using two identical solar stills (4 ft. x 4 ft. x 1 ft.). The control still contained water only while the other had dye-water solution. The dyes used were Black Napthylamine, Red Carmoisine and Dark Green, with various concentrations. The Green dye was made by mixing a 33 percent by-weight mixture of Neptune Blue, Tartrazine and Red Carmoisine (all GAF compounds) . Test results showed that dye-water solution was able to increase the distillation output by as much as 29 percent (for Black dye with 172.5 ppm concentration). Based on these tests a simple method of calculating the percentage increase in evaporation effected by a specific dye over that from the control unit was developed. The increase in output with the addition of dyes was found to be pronounced for clear days. However, no difference in the still productivities, between the control and with dye, was noticed on a completely cloudy day. An empirical relationship was developed for evaporation heat transfer coefficient and the distillate output as a function of brine top layer temperature. Among the dyes tested. Black Napthylamine dye was found to be most suitable from two points of view: increased evaporation and lack of noticeable degradation by sunlight during periods of tests. Red Carmoisine dye underwent severe degradation which resulted in the reduction of the absorption coefficient by as much as 95 percent at 0.5 pm wavelength after a month of exposure. The degradation of Green dye was not as severe as the Red dye, and this was primarily because of the Red dye component in it. The results from the model were compared with those from the experiment and the agreement between them was found to be excellent. The xiv

    PAGE 15

    effects of ambient temperature, wind velocity and dye concentration on still productivity were subsequently investigated. The model predicted a decrease in output with increase in ambient temperature and an increase in still productivity with wind speed. An increase in distillate output resulted with increasing dye concentration, but became independent of concentration after 500 ppm. The change of increase of output with concentration of the dye was different for different dyes (among those tested) with Black having the maximum rate and Red having the lowest. A proposal was made of predicting an optimum dye concentration. The optimum concentration was defined as the concentration where the sum of the change of increase of distillate with concentration times concentration and the distillate with concentration was maximum. For the spring conditions at Gainesville, Florida, the optimum concentrations for Black, Green and Red dyes were 218 ppm, 377 ppm and 408 ppm, respectively. Recommendations for further research were given. XV

    PAGE 16

    CHAPTER I INTRODUCTION • Water is the most abundant natural resource on earth. Essential to most human activities, it can mean life or death, bounty or poverty, war or peace. Abundant it is, but not infinite. Of the 453 14 units (1 unit = 2.65 x 10 gallons) of sea water that evaporates every year from the sea, only 38 units are available for possible uses by life on earth. Of these 38 units only 70 percent can be utilized [1] which is about 25 units. At present the world uses 2.5 units. Thus, theoretically, we have enough supply of fresh water and the present estimates [2] predict that the hydrological cycle can support about 25 billion people with 70 percent utilization of the water resources. However, with the present rise of world population at 2 percent per annum, the available annual water supply will probably be insufficient for the world by the year 2080 A.D. But there is a basic flaw in the above reasoning since a major portion of the fresh water supply is not available where it is needed. This factor, in conjunction with the growth of population, will be the source of water shortage. Already quite a number of communities and nations do feel the pinch of water shortages because of this. However, the problem can be partially solved by transporting fresh water to some of these communities, but the costs involved are of such magnitude that this proposition is not feasible. 1

    PAGE 17

    2 The problem is even more compounded by the fact that in some of the arid and semiarid regions the ground water has high salt content or is contaminated and thus not fit for drinking. Even in areas well endowed with ground water supply, because of increased industrialization and population growth, the rate of depletion of these supplies is reaching alarming proportions. With the cost of water transportation already high, some other way of getting fresh water will have to be found. One of the promising ways to solve this problem of water shortage appears to be desalination. Desalination processes already supply g about 7.5 X 10 gallons of water per day (gpd) in about a dozen countries [3]. Among the processes used about 85 percent are Multistage Flash Evaporation (MSF) type, due to the advanced stage of present-day technology in this area. The energy used in these plants is either petroleum or nuclear fuels. Thus, these plants are normally located in areas which are well endowed with fossil fuels. Indeed 40 percent of all the desalination plants in the world are located in Mid-East countries [4 ] . However, with increasing fossil fuel shortages and price rise, it is but natural to look at some other methods of desalination which use renewable sources of energy like solar, wind and biomass among others. Interestingly enough, the areas which lack potable water supplies have abundant solar and wind energies, which is a strong case for using these renewable energy sources for desalination. Solar distillation has been tried on a limited scale. Normally, two approaches have been used in using solar energy. One is the direct absorption of energy in saline water and the other is indirect heating

    PAGE 18

    of water and evaporating it in a centralized facility. Only one big plant (capacity of 5,000 gpd in Saudi Arabia) has been built to date, using the second approach [5]. It is, nevertheless, in an experimental stage and thus no evaluation has been given. However, it has been shown [6] that this above approach is economically unfeasible for plants with capacity of less than 50,000 gpd. Thus for smaller capacity the first approach, namely of direct absorption of radiation in saline water, is more suitable. This method is even more suited for small communities (about 200 people) in developing countries where an average of about 500 gpd of water is used for drinking and cooking purposes. In these communities setting up a large solar desalination plant is economically unfeasible, both because of capital costs and transportation costs. Several research workers have therefore built, and experimented on, medium size solar distillation units (output of less than 10,000 gallons/day) . A solar distillation unit (solar still) is a very simple device in operation but the heat and mass transfer in it is complex. Chapter III deals in detail with the energy transfer processes. Thus, in the present chapter it is sufficient to describe briefly the working of such a still which is schematically shown in Figure 1. Saline or brackish water is supplied continuously or intermittently to the pool having depth ranging from 1/2 inch to approximately 1 to 3 ft. Depending upon the depth of water in basin a still is either called shallow basin (depth ~ 2 inches) or deep basin. The bottom of the basin is black to absorb solar energy and contains a drain to discard brine, continuously or periodically. Above the basin is a sloping

    PAGE 19

    4 Solar radiation Saline Water in Black liner Figure 1. Schematic diagram of solar still.

    PAGE 20

    5 transparent cover of glass or plastic sheet, which acts as a condensing surface. The incident solar radiation after passing through the glass cover is partially absorbed in the saline water, the major portion being absorbed in the basin bottom. Heat is then transferred from the bottom surface into the water, thereby increasing its temperature. Partial vaporization of this heated water then occurs, and convection currents inside the still carry this warm vapor-laden air to the cooler glass cover. Moisture condenses on the underside of this cover, the heat of condensation being conducted through it to the surrounding atmosphere and the partially dehumidified air drifts back to the surface of water for further addition of moisture. The thin condensate film flows down the cover into the collecting troughs and is collected as distilled water. The output of these stills varies with various atmospheric conditions but on an average is between 25 to 35 gallons of fresh water per square foot of the basin area per year. Numerous variations of the above basic design have been experimented upon by research workers all over the world but the principle is the same. Some of these variations are described in the next chapter. It will be seen, however, that most of these variations in design suffer from certain drawbacks which have made them technically and economically unattractive as distillation units. Thus there is a need to design, build and analyze a better unit. With this in mind, the present work is an attempt to test the feasibility of a novel concept of using water soluble dyes to enhance solar energy absorption near the water surface, thereby increasing the productivity of the still.

    PAGE 21

    CHAPTER II LITERATURE REVIEW Historical Introduction Solar distillation is not a new subject of study. It has been extensively studied for the past two hundred years [7, p. 162] . The first such study was conducted by an Italian, Nicols Ghezzi, in 1742, who proposed the following: Perhaps placing a cast iron vase containing sea water in such a manner that the sun's rays will strike it (and during mild days and seasons, not an insignificant amount of vapor will be formed) and if the spout of the vase is shaded from the sun, it will result in a more copious and more extended flow of fresh water. But the earliest significant solar distillation plant on record was the one built by Charles Wilson in Las Salinas, Chile, in 1872 [8]. 2 The still was a shallow basin type, with a basin area of 48,000 ft and was reported to have produced, when new, 3,900 U.S. gallons of fresh water daily for mules working in the mining operation. Since then till the present day numerous such stills have been constructed and deployed both on a large scale (still in Greece produces about 6,900 gallons/day) and small scale (output of about a gallon/day) . An excellent summary of solar distillation work done around the world from 1872 to 1970 has been presented by Talbert et al. [9]. 6

    PAGE 22

    7 This suirvey covered both the large installations (output of about 2 2-5 X 10 gallons/day) and the small laboratory scale models. He reported following operating and maintenance difficulties: a. Deposition of salt in the basin for shallow basin stills, thus requiring frequent flushings and increased maintenance. b. Breakage of the cover, especially the plastic cover, with wind. c. Leakage of brine from the basin and leakage of the vapor thereby reducing the productivity of the still. Besides the above, the major drawback of all these systems was the high initial cost, which made them unattractive for other communities . In order to overcome the above problems several new designs of solar distillation units have been experimentally investigated. However, these designs were beset with problems thereby making them unsuitable for large or even small scale usage. Thus, there is a need for a better distillation unit. But before such a unit can be proposed, it is necessary to look at some of the parameters affecting the productivity of solar stills. In addition, such a study will also be useful in analyzing various designs built. The parameters affecting the performance of a still are: 1. Solar radiation 2. Depth of brine in the basin 3. Cover material and its shape 4. Ambient temperature 5 . Wind velocity 6. Temperature of the condensing surface.

    PAGE 23

    8 Solar Radiation This is the most important atmospheric variable affecting the still output. The amount of radiation absorbed by the brine is directly reflected in the productivity of still. This absorption is governed by various factors, for example, depth of brine, the absorption characteristics of water and angle of glass. These factors are explained below. Brine Depth The effect of brine depth on the output of solar stills is plotted in Figure 2. The data in this plot are from a number of stills built all over the world [9]. It can be easily seen that decreasing the brine depth increased the productivity. This was expected since the heat transfer to water in a shallow basin is very rapid, thereby increasing its surface temperature and hence the evaporation. Thus, most of the distillation output of shallow basin takes place during the day; whereas in the case of deep basin still an appreciable amount of solar radiation is absorbed in the water and because of the thermal '-^ . capacity of the still, the output is over a 24 hour cycle. In addition, the thermal losses from the sides and bottom of the deep basin still tend to be larger than those from a shallow basin still. Several studies, therefore, have been conducted on the shallow basin stills. Scientists at the University of California have studied the cascade type still [10]. The depth of the brine was about 1/4 incli. 2 2 High output rates were obtained (0.13 gal/ft -day at about 2000 Btu/ft solar radiation input) but the still was beset with problems like dry spots in the basin, blistering of black epoxy paint, deposition of salt

    PAGE 24

    9 w q; a c 12 0) c H u-i o 0) Q 8 0.04 00 5t I o 0.06 0.08 0.10 Distillate (gallons/f t^-day) 0.12 Figure 2. Effect of brine depth on still productivity.

    PAGE 25

    10 and thus required expensive maintenance, but above all the cost of the unit was high. Maria Telkes [11] has studied extensively wick type solar stills. A porous black material acts like a wick and is soaked by saline water. Since the layer of the water is very thin (1/32 inch), there is rapid evaporation and increase of productivity results (out2 2 put of 0.16 gal/ft -day at about 2000 Btu/ft -day). However, there are severe problems of salt clogging the pores, dry spots developing resulting in deterioration of the wick material and precise control of brine flow. These problems have resulted in making these stills unattractive for commercial production. Recently ESB Corporation [12] patented a new type of wick material which is claimed to have selfcleaning properties so that it is not clogged by salt. Since no detailed data are available, it is difficult to assess its potential as a commercial unit. Several investigators have also looked at the effect of floating porous pads on the surface of water so that the brine depth is effectively reduced. Batelle researchers [13] reported an increase in productivity by 10 percent over the similar shallow basin stills, but the system was beset with problems of salt accumulation and dry spots. Recently Szulmayer [14] has reported tests on flotation of black liner 2 with outputs of 0.072 gal/ft -day. No results on salt accumulation or dry spots were reported, however. Cover Material and Its Shape The cover material has two functions. Besides allowing solar radiation to pass through it, it acts as a condensing surface for water vapor. The first function requires that the cover should transmit as

    PAGE 26

    11 much radiation as possible and thus the water film on the underside should be thin. The second function requires that there should be least resistance to the transfer of heat of condensation meaning that the thickness of the cover material should be very small, there should be enough area of condensing surface and the condensation should be dropwise. However, it has been shown [15] that the decrease of solar radiation transmission, because of formation of droplets on the underside of the glass, more than offsets the increase in heat transfer. Thus filmwise condensation rather than dropwise is advantageous. Normally, two types of cover materials are used: glass or plastic. The advantages of glass are: high transmission for solar radiation, high wettability of glass and relatively high stability of properties over extended periods of time. The disadvantages are its relative poor strength and thus vulnerability to mechanical damage. This requires an increase in its thickness (normally 0.125 inch thick glass is used for cover) . Plastic films normally used for solar stills have high solar transmission (an average of 90 percent for 0.004 inch thick Tedlar film at angles of incidence between 0 and 45 degrees, as compared to about 85 percent transmittance for 0.125 inch glass). At the same time these films have high infrared transmittance (80 percent in the wavelength region of 4 ym to 7 ym) and high susceptibility to ultraviolet degradation. It is these properties, together with the fact that they have to be treated so as to make them water wettable, has made use of plastic cover unattractive in solar stills. Thus, till the advent of better plastic material which will overcome the difficulties listed above, it is believed that glass will be mostly used.

    PAGE 27

    12 The shape of the cover is governed by three factors, namely a) it should provide enough condensing surface for water vapor, b) its angle should be such that it intercepts maximum solar radiation for the whole day, and c) the spacing between the cover and the water surface. Two types of cover designs are normally used. One is the greenhouse type or double sloping cover (Figure 1) while the other has single sloping cover with the south facing side normally having a reflective surface as shown in Figure 3, However, it has been shown, both experimentally [16] and theoretically (next chapter) that if enough condensing area is available then the effect of spacing between cover and water surface is negligible over the productivity. It also appears [17] that the effect of glass inclination on the amount of solar radiation entering the still is small, because the interception area is always horizontal except in the cases of tilled wick. Thus the major factor governing the shape of the cover is the availability of enough condensing area. It is interesting to note that very few investigators appear to have considered this factor in design of their stills. Most have tried to minimize the spacing between the cover and the water surface which ultimately results in low outputs in the absence of enough condensing surface. Ambient Temperature and Wind Velocity These two variables have been lumped together because they affect the productivity of the still by their effect on the temperature of the cover. As the ambient temperature increases the amount of heat transfer from the cover to the atmosphere is reduced, thus decreasing the productivity. However, some computer results of Lof et al . [18],

    PAGE 28

    13 Solar radiation Black liner Figure 3. Single sloping solar still.

    PAGE 29

    14 Baum [19] and Bloemer et al. [20] have shown that there is an increase in output of the still with increasing ambient temperature. These results are not borne out by actual experiments [16]. Increase of wind velocity increases the outside heat transfer coefficient thereby increasing the heat loss from the cover and the productivity. However, Lof et al. [18], Baum [19] and Bloemer et al. [20] have predicted that increase of wind velocity decreases the output. They found that an increase of external heat transfer coefficient from 2o 1.5 to 10 Btu/hrft °F decreased the output by 20 percent. Again, this is not borne out by experiments [21], calculations [22] and by computer analysis (present work) . Temperature of the Condensing Surface The temperature of the condensing surface is a function of both the brine temperature and the ambient conditions. A greater difference between the brine and cover temperature results in increase of the productivity of the still. However, during the daytime, because of higher brine temperature this difference is not maximum. Several investigators have tried to increase this difference by providing an external condenser for water vapor. Grune et al. [23] conducted extensive tests on blowing air through the still and condensing the water vapor on an external water-cooled condenser. Dunkle [24] investigated a multiple effect solar still where the heat of condensation from the first effect is used to provide the boat oC evaporation in the second effect and so on. The condenser of the last effect was cooled by water. Salam and Daniels [25] used two inflated plastic tubes with the inner tube partially filled with salt water over which the

    PAGE 30

    15 air was passed. This water vapor laden air was then taken to an external condenser where the water was condensed. The reported still , . energy used in evaporating water s efficiencxes (efficiency = ^ P ^ *H — — Tu total solar energy input to the still between 20 and 40 percent on bright summer days. Besides low efficiencies this system suffered from another drawback, the deterioration of the plastic tubes. 2 Although production as high as 0.22 gal/ft -day on a bright summer day was reported by Grune et al. [23], the additional costs required in equipment for air blowing and water circulation more than offset the gain in yield. Similar conclusions were also reached by Dunkle [24], who also reported the problems of corrosion of the metal condenser by saline water. There are other factors also which affect the output of solar stills. They are good still construction so as to have maximum vapor tightness; no or minimal seepage of brine from the basin; good insulation of the sides and bottom for shallow basin stills; and the frequency of flushing and the orientation of the still [26]. From the above considerations of various parameters affecting a solar still, some of the desirable features of a good distillation unit are: a. To ensure maximum absorption of solar radiation near the surface of brine. b. Cooler condensing cover temperature. c. Minimum amount of maintenance and flushing. In the present study a design of a still has been investigated which incorporates the above features. By mixing water soluble dyes in the saline water, solar radiation can be absorbed in a thin ( ~ 1 inch)

    PAGE 31

    16 surface layer of brine. At the same time if the basin is made deep, then the still will also work during nighttime, thus using the advantage of cooler condenser surface. The problem of salt buildup (and thus periodic flushing) is also reduced because it settles at the bottom of the basin and does not interfere in the solar radiation absorption near the surface. Use of Dyes to Enhance Solar Evaporation The use of water soluble dyes to enhance evaporation of water is not a new concept. Block et al. [27] were the first to study the effect of 2-Napthol Green dye on the evaporation of brine for large scale salt production. They tested this dye in different depth ponds and for different concentrations. It was found that if enough dye was added (20 ppm by weight) the depth of the pond does not have too much effect on evaporation (the difference in evaporation was less than 5 percent for brine layers of 20 cm and 50 cm deep) . They did report that with the above concentration of the dye the increase in evaporation of 19 percent over that of uncolored brine was achieved. However, no attempt was made to study the effect of other dyes, the effect of sunlight on the degradation of the dye or to correlate their experiments with analytical models. The use of dyes for salt production subsequently became an accepted practice. However, very little data are available on the parameters that affect production because most of tlio information is proprietary in nature [28] . Other means of coloring the brine have also been tested with limited success. The use of red bacteria Halobaaterium or blue green algae

    PAGE 32

    17 Spirulina has been tried to enhance the evaporation of brine [29]. The advantages of such a system are obvious. There is no solar degradation of the bacteria and at the same time the culture can be obtained free from the ocean since it exists in nature, thus avoiding the use of costly dyes altogether, which are not anyway recoverable. However, these bacteria are extremely susceptible to changes of salt concentration and temperature and have not been therefore commercially successful [30 ] . Keyes et al. [31] have extended Block's work to include other dyes and tested their degradation by sunlight. Among the six dyes tested Napthol Green was the most stable in the presence of salt solutions and sunlight for the test period of 6 months. However, conflicting reports were published on the effect of some of these dyes on solar evaporation. It was reported that black Nigrosine and Congo red dyes do not increase the evaporation of brine, while Napthol Green increased it by 7 percent as compared to 19 percent increase achieved by Block et al. [27]. Keyes et al. [31] developed an analytical model of the system. However, the model failed to take into account the spatial and time variations of temperatures of the brine. From the brief description of the effect of dyes on solar evaporation it is evident that sketchy data are available on their use with no study done on their use for solar distillation. Besides this lack of experimental data no detailed analysis has been conducted on the effect of dye on solar evaporation and on spatial and temporal variations of the dye-brine system. The present work is therefore an attempt to study some of these effects.

    PAGE 33

    18 Scope of the Present Investigation The present investigation is an attempt to achieve the following obj ectives : 1. To develop a simple analytical model for study of the effect of different parameters on productivity of solar stills. These parameters are ambient temperature, wind velocity, type of dye and concentration. The model will be developed for prediction of temperature-time history of the dye-water system and the glass cover for different environmental conditions and input solar radiation. The model is developed in Chapter III. 2. To experimentally evaluate the effect of different dyes on productivity of deep basin solar still. The productivity from this still is compared with that from an identical control unit. Chapter IV presents the experimental setup and procedures . 3. To obtain experimentally the absorption spectrum of the dyes used in this study. The dyes used are Napthylamine (Black), Carmoisine (Red) and mixture of Neptune Blue, Carmoisine and Tartrazine (Dark Green) . 4. To compare the results obtained from this experiment with that from the analytical model. This comparison is presented in Cliapter V.

    PAGE 34

    CHAPTER III THEORETICAL ANALYSIS In this chapter a theoretical analysis for a deep basin distillation unit is presented. Physical Model Figure 4 is a sketch of a deep basin distillation unit. It shows the energy attenuation of a ray from the sun as it passes through the condensate covered glass and into the dye-water mixture. In order to analyze the energy transfer process in the basin the dye-water system has been divided into discrete layers of thickness Ax. An energy balance is then given for these layers after making the following assumptions : 1. The dimensions of each layer, as shown in Figure 4, in the y-z direction are large compared to those in the x direction and B_j^ « 1 , so that the temperature varies only in the x direction. 2. In the teraperature-range normally encountered in the solar distillation units, the changes in physical properties of the Biot number (B.) is defined later in the text. 19

    PAGE 35

    20 Figure 4. Model of still used in analysis.

    PAGE 36

    21 solution like viscosity, density, specific heat and thermal conductivity are negligible. 3. The spectral emission and scattering coefficient of the dyewater mixture is assumed to be small and thus neglected. 4. The attenuation of the incoming solar radiation by the glass with condensate film inside has been taken into account by averaging out the reflection and absorption losses with respect to angle of incidence. 5. The bottom surface of the still is assumed to be a black absorber . 5. There is negligible attenuation of solar radiation as it passes through the air-water vapor mixture inside the still. With the above assumptions, the energy balance on the n^^ layer of thickness Ax then leads to the following: Rate of increase of energy stored where energy in to the n layer comprises of: [Rate of energy in] [Rate of energy out] (la) th Rate of energy in Rate of solar radiation absorbed in n^h layer + Rate of energy in by convection from layer n+1 Rate of energy in by conduction from layer n+1 or n-1 (lb) and rate of energy out can similarly be written as;

    PAGE 37

    22 Rate of energy out Rate of energy out by conduction to layer n-1 or n+1 Rate of energy out by convection to layer n-1 Rate of energy out by conduction from the sides to environment (2) At the upper boundary (air-water interface) the energy is transferred to the glass cover by convection, radiation and evaporation. Thus, the instantaneous energy balance on the surface leads to: 0 = [Rate of energy in] [Rate of energy out] (3a) where [Rate of energy in] = Rate of energy in by conduction from layer 2 + [Rate of energy in by convection from layer 2] (3b) and [Rate of energy out] = Rate of evaporation of water from the surface + [Rate of radiation loss to the glass cover] + [Rate of convection loss to the glass cover] (4) At the lower boundary, namely the base (layer b) , the energy balance is: [Rate of energy in] = [Solar radiation absorbed in this layer] (5a) and

    PAGE 38

    23 [Rate of energy out] Rate of energy loss by convection to b-1 layer + [Rate of energy loss by conduction to base] + [Rate of energy loss by conduction to layer b-1] + [Rate of energy loss by conduction from the sides] (5b) Finally, the energy balance on the glass cover yields: [Rate of energy in] = [Solar radiation absorbed by glass cover] Rate of energy in by convection from the dye-water surface Rate of energy in by radiation from the dye-water surface Rate of energy in by evaporation from dye-water surface (6) [Rate of energy out] Rate of energy loss by convection and radiation to the environment (7a) Therefore, Rate of energy in Rate of energy out Rate of increase of energy stored in the glass cover (7b)

    PAGE 39

    24 Incident Solar Radiation The incoming spectral solar radiation reaching the dye-water surface is considered as having wavelengths in the range of 0.3 ym to 2.0 ym. The choice of this range has been dictated by the fact that most of the radiation after 2.0 ym is absorbed by the condensate film on the glass surface. Furthermore, to facilitate computation, the solar spectrum [32] has been smoothed by fitting a polynomial in the above range. The radiant flux then can be written as: ^ ^Btu ^ ^ j 2379X 713.7; 0.2 ym ^ A £ 0.5 ym A ft hr m 1 1261.2 exp(-1.95A); 0.5 ym < A < 2.0 ym (8) The spectrum given by equation (8) is the maximum radiation available at the solar noon on a clear day, and thus will have to be modified for different times of the day. This modification is achieved by assuming that the "shape" of remains the same but the area under the curve of q^ vs A becomes equal to the total solar radiation falling on a horizontal surface at time t during the day. If the solar radiation incident on a horizontal surface at time t is given by q(t) then becomes: where q^(t) = at) q^ (9) C(t) = (10) 'o.3 '^A'^^ 4

    PAGE 40

    « 25 Absorption of Radiation The incident angle of solar radiation on the dye-water surface changes with time, as shown in Figure 5. This change in the angle of incidence changes the absorption thickness for any layer of thickness Ax. Thus, from Snell's Law, Sin 9 . n„ 1 _ _2 Sin 0 n. (11) where n^ = index of refraction for air-water vapor mixture and = index of refraction for water-dye mixture. Since the amount of water vapor, as compared to the air inside the still, is small it is assumed that n^^ ~ 1 . Also it has been assumed that the index of refraction for dye-water mixture is the same as that for pure water and both n^ and n^^ remain constant for the range of wavelengths for 0.3 nm to 2 ym. From Fresnel equation the reflectance can be calculated by knowing 0. and 0 and is given as: , sin^(0. 0 ) ± 1 r r „ 2 sin (0. + 9 ) 1 r 1 + cos (0 . + 0 ) 1 r 2 cos (0. 9 ) 1 r (12) and for 0^ = 0 , "2 "l. (13a)

    PAGE 41

    26 Solar radiation Reflected beam Water-air Interface Figure 5, Radiation transfer in the dye-water solution.

    PAGE 42

    27 Thus, the amount of solar radiation passing through the water-air interface has to be multiplied by the transmittance of both the waterair interface and condensate covered glass. For the condensate covered glass an average transmittance value been obtained from experimental data. Therefore, the radiation passing the water-air interface is: where is obtained from equation (9) , and r is given by equations (12) and (13) above. The path length of the incident radiation is: Ax* = Ax/cosG^ (14) where Ax* is the path length of the ray of incident solar radiation for a layer of thickness Ax. As can be seen from equation (14) , Ax* is also a function of time. The absorption of incident radiation in the dye-water medium takes place in accordance with the Bouger's Law [33]. Thus, the amount of energy absorbed in a layer of thickness Ax* is: 'labs^''^ = ql^^U) dX j^^ qJJ^ .(t)[exp(-a^Ax*)]d^ (15) where q^ .(t) is the incident spectral radiation on tlie layer n and a, A , 1 A is the spectral absorption coefficient of the medium. The absorption coefficients of different dyes have been obtained experimentally and will be discussed in greater detail in Chapter V.

    PAGE 43

    28 Conduction and Convection Heat Transfer in Dye-Water System The heat transfer in a layer of thickness Ax occurs by both conduction and convection from adjacent layers besides the solar radiation absorbed by it. The conduction heat transfer is easily modeled while for convection the concept of equivalent conductivity [34] is used. Thus, the average heat flow per unit time because of convection is characterized by equivalent conductivity defined as: conv f ^^H ^c> (16) where t = thickness of layer Tj^, T^ = temperatures of hot and cold layers, respectively. The data for equivalent conductivity for plane layers from different investigations has been plotted [34] and fall on a straight line, given by: log 10 w = 0.3 logj^Q Ra^ 1; for Ra^ > 2x10 = 0.25 logj^Q Ra^ 0.7; Ra^ £ 2x10^ (17) where k = conductivity of the water Ra^ = Raleigh number based on thickness t of the layer = !iiYlVf! V a w w 6 = thermal expansion coefficient of water g = acceleration due to gravity = kinematic viscosity of water a = thermal diffusivity of water, w

    PAGE 44

    29 Heat loss from the sides of the layer to the environment is; q. = Ax(T T ) UP (18a) loss, sides n 00 where and are temperatures of layer n and the environment, respectively. The factor UxP is the product of overall heat transfer coefficient U between the water-dye solution and the environment, and the perimeter P of the still. Equation (18) is a one-dimensional model; it implies that Biot number (B^) be much less than 1 where is B . = ^ (18b) w where A is the area of the layers. The value of B^ for the present case is 0.19. This makes the one-dimensional assumption reasonable. Evaporation and Convection Loss from the Water Surface The mass transfer inside the still occurs mainly by diffusion and convection of the air-water mixture. Baum and Bairamor [35] have done interf erometric studies on a laboratory still and found that most of the energy transfer by evaporation of water and convection from the water surface takes place in a boundary layer which circulates inside the enclosure. The bulk of air at the center of the enclosure remains essentially stationary. Thus, the evaporation rate is determined by the analogy between heat and mass transfer for free convection. The heat transfer by convection from the water surface to tlic glass cover is given by: q = h A (T T ) (19) C CSS g ^ '

    PAGE 45

    30 where is the heat transfer coefficient and is the average glass cover temperature. There is spatial temperature variation in the glass cover because of the geometry of the still but for the sake of simplicity an average temperature between the top and bottom portion of the glass has been used. The heat transfer coefficient h can be calculated from the c following nondimensional equation [36]: Nu = C (Gr Pr)P (20) where Nu = Nusselt number = c w/a 3 Gr = Grashoff number = L g (hp/p) m uc and Pr = Prandtl number = p w/a C = Constant The choice of exponent p and constant C in equation (20) is dictated by the Grashoff number. In the temperature ranges encountered in the solar still the exponent p has the value of 1/3. It should be noted that h^ from equation (20) is independent of length L (distance between the water surface and the glass cover) for most stills. Experimentally the above fact is not borne out because there is heat and mass transfer to the glass cover on the sides also, which changes the free convection pattern. The bouyancy factor (Ap/p) in the Grashoff number of equation (20) was evaluated for the conditions of constant composition of the convecting fluid. However, this has to be modified for the solar still

    PAGE 46

    31 since in this case the humidity of air changes along the flow path. This modification is achieved by assuming that air-water vapor mixture is an ideal gas [36]. Thus, P M P M P„/^ = P„ + = _w w + air air (21) where p , = density of water-air mixture w/a •' P^ = partial pressure of water vapor P . = partial pressure of air air ^ and "^^'^g-^j-" molecular weights of water and air, respectively R = universal gas constant T = temperature of the mixture For an enclosed system, P + P . = P, , , (22) w air total ^ ' In the case of a solar still, since it is open to the atmosphere P^ ^ , = P ^ . Therefore, equation (21) is modified to: total atm ^ P M (P ^ P ) M . _ w w atm w air w/a " R T R T Thus, the density of water-air mixture near the glass cover and at the water surface can be respectively written as: P M (P ^ P ) M w ^ atm^ ^ wg air ^^3) w/a g R T R T g g and

    PAGE 47

    32 w/a P M (P P ) M . ws w atm ws axr R T R T (24) while Ap = p , I p , I a/w |g a/w [s and p , is the arithmetic means of p , i and p a/w w/a s w/a |s' After algebraic manipulation: 4g = 2 (1-x) 1+x (25) where X = T P + (P P ) (M . /M ) ws atm ws air w P + (P P ) (M . /M ) wg atm wg air w and h from equation (20) then becomes: c h = c k , i^m^ f-^) c w/a V k , m w/a (26) The mass transfer coefficient is defined as: J = h (m . m . ) m IS le (27) where , h = m (m. m . ) 1 s le mass transfer rate per unit area mass transfer coefficient, mass/area time driving potential for the mass transfer, dimensionless

    PAGE 48

    33 in. = ^ = mass fraction of species i; (where subscript s implies ^ P near the surface and e, in the environment) From analogy between heat and mass transfer, an equation similar to equation (20) can be written for mass transfer. Thus, Nu = C (Gr S )P (28) m m c ^ where Gr = Grashoff number for mass transfer m Nu = mass transfer Nusselt number = ^^m ^ m p = average density of the mixture D_ = binary diffusion coefficient or mass diffusivlty S = Schmidt number = ^m c D. . V = kinematic viscosity of the mixture m The above analogy is based on the following assumptions: (i) Small rate of mass transfer (ii) No chemical reactions in the fluid (iii) Negligible viscous dissipation (iv) Negligible emission or absorption of radiant energy In a solar still there is combined heat and mass transfer from the surface of water. For such a case the ratio of Nu and Nu proves to be m ' functions of Pr, Sc, and the ratio of bouyancy induced by the temperature difference to that induced by the mass concentration difference. This latter ratio is 5 1 if Pr 5 Sc, which indeed is the case

    PAGE 49

    34 for water evaporating in air. Thus the problem is simplified in the sense that mass and heat transfer can be treated separately. Hence, Nu = C (Gr Sc)-'"''-^ (29) m and h is given as: m m ij and heat exchange because of mass transfer is: q = h A(m. m. ) h, (31) th m ^ xs ig fg ^ where h = mass transfer coefficient m A = area of the still (m. -m. )= mass transfer driving potential IS ig ^ ^ h^^ = enthalpy of vaporization Thus, from equations (24) and (28) the mass fraction of species is: PM PM (P-P)M TT, = Pn Q c w , , , ws w atm ws air . ""is — R T ^'^ R T R~T ' P s s s or, P M m . = ws w P M + (P P ) M (^^^ ws w atm ws air Similarly, P M wg w ™ig P M + (P P ) M (^^^ wg w atm wg air

    PAGE 50

    35 Formulation of Equations The difference equations have been written for different sections and have been thus identified. Layer n Combinations of equations (1) and (2) give: ^2 n ^2 n ^ ^ A { / q (t)dX / q (t) [exp(-a,Ax*) ] dA} (T*" T^J ^"L A , 1 A , 1 A Ax n n+1 k A ^^t+M _ ^tj ^ -t K UP Ax = A Cp Ax " " (34) where = apparent conductivity given by equation (16) A = area of the layer In the above equation if there is no convection then simply becomes equal to k. The boundary conditions can be written as: Top Layer Combining equations (3) and (4) give the temperature of the top layer . Thus , \ A (T^ T^) (Ax/2) = \ ^('"is ™ig)hfg + V^^i Tg) + ^Va^tT^ (35) where = emissivity shape factor between water surface and glass cover

    PAGE 51

    36 = geometrical shape factor between water surface and glass cover Bottom Surface An instantaneous energy balance on bottom surface yields: A { q° ^(t) [exp(-a^£*) dA} = k A(T, J k A(t5 ) b b^ Pg b b.out^ (Ax) + Ax (^^^ Pg where o q, .(t) = the solar energy incident on the water-air interface A, X t* = transmission length for the whole system as shown in Figure 4 k = conductivity of Plexiglas pg J b Ax = thickness of the Plexiglas pg ^ Tj^ out ^ temperature of the Plexiglas bottom Inherent in equation (36) is the assumption that the bottom surface is a black absorber so that all the energy incident on it is absorbed. Energy Exchanger with the Still Covers The right-hand side of equation (35) gives the rate of energy into the glass and aluminum cover. This energy must be given up to the environment. At the same time there is a small amount of solar radiation absorbed by the glass cover. Thus, the temperature of the cover can be obtained from the following equation:

    PAGE 52

    37 4 4 t t aFFA[T T ]+h A(m. m. )h. + h A(T T ) + a A q(t) e a s g m xs xg fg c ' s g' g g ^ t t 4 4 t'^"*'^'' t'^ A (T T ) aF F A (T T , ) = p c A Ax ( g g) o c g 00 ec ac c g sky' ^g pg g g' ^ — Pal '^pal \l ^^al ^ (37) ^ At where A = area of still cover = A + A , c g al A = glass cover area g A^^ = aluminum back plate area a = solar absorptivity of glass O = outside convection heat transfer coefficient P > P n = density of glass and aluminum, respectively Cpg, ^pal ^ specific heat of glass and aluminum cover, respectively and '^'^al ~ thickness of glass and aluminum cover, respectively ^sky ^ effective ambient temperature for radiation = T 15°F [37] and other symbols are defined before. Experimentally it was found that during daytime the temperature of the aluminum back plate was greater than the surface temperature of water. Thus, no condensation took place on the back plate. At night, however, there was condensation at the aluminum plate but its temperature was equal to that of the glass cover. Moreover, an order of magnitude analysis of equation (37) shows that the storage term in

    PAGE 53

    38 the aluminum is negligible. Thus, all the temperatures T in the a Jabove equation can be considered as T . This yields g During Daytime 4 4 t f OF F A [T T ] + h A (m. m. ) h^ + h A(T T ) + a A q(t) e a s g m is ig fg c ^ s g' g g h A (T*" tS OF F A (T^ T^, ) = p c A Ax (T^'^^'^-t'') o g g °° eg ag g g sky g Pg g g _g g. At. (38) During Nighttime aF^F A (T^ T^) + h A (m. m . ) h^ + h ACT*^ T*^) h e a s g m is ig fg c ' c g a ^t+At _ ^t A (T^ T^) -OF F A (T^ T^, ) = p c A Ax S-) eg 0° ec ac c g sky ^g pg g g^ At ^ (39) Note the change in cover areas for day and night in the above equations. Method of Solution The depth of the still is 10.5 inches and with grid size of Ax = 1 inch there are 11 nodal equations and two boundary conditions for the top and bottom surfaces. Therefore, a computer program was written to solve these 13 nodal equations for the dye-water, and one equation for the glass cover, simultaneously. Explicit method of solution was used in solving them and therefore the time interval At was carefully chosen so as to avoid unstable solutions. Appendix II shows the stability

    PAGE 54

    39 criterion used in the present analysis. Below are listed the steps used in getting the temperature-time history of various layers. 1. The temperatures of the layers and glass cover are initialized at sunrise, using their experimental values. Thus, T^ and T^ are known. Also used in the program is the actual solar insolation, ambient temperature and wind velocity. 2. Once T^ and T2 are known then equation (35) is solved to obtain T^ . The solution of equation (35) is by regutafalsi method [36]. 3. Equations (34), (36) and (38) or (39) are then solved for T^^^'^ and T^+^^ n 8 4. Once the temperature vs depth profile is generated from step 3, then the apparent conductivity k is calculated. The e method of calculation is as follows: (a) Calculation of T , as shown in Figure 6. n,raax " (b) All the layers above layer n will have convection transport with AT = T T and subsequently k is n,max s e calculated using equation (17) and depth t . (c) The layers below the n'''^ layer have heat transfer between them by conduction only. 5. Steps 2 to 4 are then repeated with an increment of time At, till a 24 hour cycle is completed. Figure 7 shows the detailed flow diagram of the computer program used to generate temperature-time history for different layers. The results of the analytical model are presented in Chapter V.

    PAGE 55

    T s T n,Tnax Temperature of layers (arbitrary units) Figure 6. Temperature profile of the layers.

    PAGE 56

    41 Ini tializG temns of all layers & glass tJ = T,,. i = 1.13 Initialize TIMK = 0 t imc Subroutine to solve cqua t i on ( 35 ) by regulu falsi method Subroutine to calculate amount of Kol a r rnd i a I i cii absorbed i n 1 aver n Calculate T*^ from equation (35)^ Subrout i no t o calculate heat and mass transfer belueen water su rface & glass cover Calculate' the of di f f eren t 1 equat i on ( 3.1 ) tom|)cra tu res ay(.'rs u.sinj; Ihus, 1 n i s known . Subroutine to ra 1 eu 1 a te the apparant con(lui( i V i 1 y k , based upon n ' max Is it day or ni ght f d a v night C:a leu late the temp, of glass T^"*< using g •quation (38) Calculate 1 t + At " g usi ng equat ion ( 39 ) I Increnieit time by At TIMK = TIMI-: ^ Al No Is T i me eipia I to 2'] hours ? Yes Pri nt the Figure 7. Flow diagram of the computer program.

    PAGE 57

    CHAPTER IV EXPERIMENTAL SETUP AND PROCEDURES Distillation Units Two identical deep basin solar distillation units were constructed for testing. One of these was used as a control while the other was used for testing the effects of dyes. The basin of these stills was made of 3/8 inch thick Plexiglas which was selected for its strength and corrosion resistance to salt water. Figures 8 and 9 show two views of the stills. Plexiglas plates were bolted together and then glued using epoxy resin to make them water tight. Before experiments the basins were tested for water tightness. The overall dimensions of each basin were 4 f t . x 4 f t . x 1 f t . These basins were painted on the inside using two coats of Suntec Acrylic Black paint which acted as a black liner for the stills. At the top of the basin, extruded aluminum angle of dimensions 2 inches x 1-1/4 inches and 3/16 inch thick was bolted to serve as support for the glass cover frame. The glass cover frame was made of aluminum extrusions, shown schematically in Figure 10. To this was bolted aluminum channels of 1.5 inches x 0.5 inch and 0.125 inch thick, which acted as distillation troughs. This trough also provided support for the glass cover. A thin cotton cloth wick covered the distillate channels in order to 42

    PAGE 59

    44

    PAGE 60

    45 enhance the distillate flow. The cover frame was welded together using Argon-arc welding. This removable cover facilitates in changing the dyes and giving access to the interior of the basin. The glass angle was kept 30° to the horizontal and facing south. This angle was decided so that a) there should be sufficient condensing area and b) the still should intercept maximum radiation for a fixed angle throughout the year. The south facing glass cover was double strength (1/8 inch thick) pane with dimensions of 5 ft. x 2 ft. Two of these panes were used for the cover. The side glass covers were two right angle triangular panes with hypotenuse as 5 ft. The back plate was made of aluminum sheet (1/45 inch thick) and was bolted to the glass cover frame. This plate acted as reflector for solar radiation as well as condenser surface. The glass panes and the back plate were then sealed to the frame by silicone rubber sealant to make the cover vapor tight. The vapor tightness of the cover was then checked by passing water from a rubber hose over it and leaks, if any, were then again sealed. In order that the still be vapor tight after the cover is put over the basin, a ribbon of sealing compound Permagum was run the entire length of the basin top. This compound remains pliable over the temperature range of -20°F to 200°F and thus acts as a good sealant. The water level in the stills was maintained by a constant head feed tank, shown schematically in Figure 11. The basins were insulated on the sides by a polyurethane slab (Thermax ) which was 3/4 inch thick and dimensions of 1 ft. x 4 ft.

    PAGE 61

    46 u X) 0) to •H fa •r-l CO

    PAGE 62

    47 The bottom of the basin was insulated by two 1-1/2 inch thick foam glass slabs each having dimensions of 2 ft. x 4 ft. Thermocouples and Their Locations Copper-Constantan thermocouples of size 24 gauge (0.02 inches) were used in all temperature measurements. Temperature measurements were made at various locations as shown in Figure 12. These thermocouples v;ere attached on outside of the Plexiglas basin by a paste of Plexiglas chips and Ethylene dichloride, while on the glass cover they were held down firmly by transparent scotch tape. This tape was over an aluminum covered styrofoam piece pressing down on the thermocouple as shown schematically in Figure 13. Besides these locations thermocouples were attached on the outside of the insulation around the basin. To measure temperatures of dye-water in the still, five thermocouples were attached on a Plexiglas rod at various distances as shown in Figure 14. The thermocouple junctions were coated with a very thin coating of shellac to make them corrosion resistant. These thermocouples were calibrated, together with the temperature measurement recorders, at four temperatures, namely the ice point, and three other temperatures between 32°F and 150°F. Temperature and Solar Radiation Measurements There were about 30 thermocouples attached at different locations on both the stills and this necessitated the use of multiple point

    PAGE 64

    49 Thin shiny aluminum foil le bead Styrofoam piece Figure 13, Attachment of thermocouple on glass cover.

    PAGE 65

    50

    PAGE 66

    51 recorders. Two such recorders, one made by Honeywell and the other by Leeds & Northrup, were used. The Honeywell -Br own electronix recorder had 20 points with accuracy of ± 0.5°F and range of 0-150°F. The Leeds & Northrup speedomax recorder was 24 point with an accuracy of ± 1.0°F and range of 0-200°F. These recorders were connected to the mechanical timer so that the temperature readings could be taken after every 1 hour interval. The solar radiation data was measured by Eppley Phyrheliometer , Model No. 10, and recorded on a single point Leeds & Northrup speedomax recorder with accuracy of ± 1 percent of the total scale. The ambient temperature readings were obtained from the hygrothermograph at the Solar Energy Laboratory. This laboratory was about 10 miles from the site of the present experiments. Absorption Spectrum of Dyes The dyes used in the present study were water soluble and supplied by GAF Corporation. They were: 1. Napthylamine 10 BR (Black) 2. Carmoisine BA Ex (Red) 3. Mixture of 33 percent by weight of each of the following dyes: Neptune B.R (Blue), Carmoisine BA Ex (Red) and Tartrazine C Extra (Yellow) . This mixture resulted in Dark Green dye. The absorption spectrum from wavelengths of 0.3 ym to 0.75 pm was obtained using a Beckman UV-VIS Model No. 25 spectrophotometer as shown in Figure 15, where (A) is the spectrophotometer and (B) is the strip

    PAGE 67

    52 Figure 15. Beckman UV-VIS spectrophotometer.

    PAGE 68

    53 chart recorder for output data. The specimen cell was a standard quartz rectangular cell with optical thickness of 1 cm. The absorption data for wavelengths from 0.75 \im to 2.0 )jm were obtained using a Perkins-Elmer Model E-1 Monochromator as shown in Figure 16 with (A) the monochromator and (B) the strip chart recorder with the controls. The specimen cell for this instrument was designed for the present study (see Figure 17) . It consisted of an aluminum split ring 1 inch in diameter and 1/4 inch in length and the thickness of the ring being 1/16 inch. This ring was sandwiched between 2 glass windows each of diameter 1 inch and thickness of 1/8 inch. To make the assembly water tight rubber gaskets (1/32 inch thick) were put between the aluminum ring and the glass windows. This cell was held in place by aluminum supports, which were bolted to each other. Figure 18 shows the crosssectional view of this cell. The Beckman Spectrophotometer is a differential device wherein the spectrum of the specimen is continuously compared with that of distilled water. Thus, it gives the absorbance spectrum [Absorbance = log (Incident radiation/Transmitted radiation)] directly. However, the Perkin-Elmer Monochromator is not a differential device and thus separate readings of the transmission spectrum of cell only, cell filled with distilled water and cell filled with dye sample, were taken. The absorption coefficient was then calculated from these readings. Experimental Procedure The two stills were set side by side separated by a distance of about 3 feet, as shown in Figure 19. They were filled with tap water

    PAGE 69

    Figure 17. Specimen cell.

    PAGE 70

    55 Figure 18. Cross section of the specimen cell used in Perkin-Elmer Monochromater .

    PAGE 71

    56

    PAGE 72

    57 to the depth of 10.5 inches and the level was maintained by the feed tank. Common salt (composition unknown), 3.5 percent by weight, was added to the two stills. In one of the stills, dye of known concentration was added. The covers were then put over the stills and allowed to stand for a day or two to attain a steady periodic operation. After this period of time the distillate was collected every 24 hours and the weight of the distillate was measured by a cantilever type balance with an accuracy of ± 0.1 lbs. The temperature measurements of various locations were recorded on the strip chart at every 1 hour interval. Three dyes were tested with concentrations of 50 ppm and 100 ppm for each except for the Black dye where the testing was done for 50 ppm and 172.5 ppm concentrations. The duration of testing for each dye is given in Table 1. On a number of occasions the distillate was collected every hour so as to study its hourly variation. This also helped in calculating the constant C in equation (20), as will be shown in Chapter V. The absorption spectrum of the dyes before and after the duration of experiment was also obtained so as to study their degradation by sunlight. Also evaluated was the absorption spectrum of the distillate to study carry over of the dyes by water vapor. The results of the experiment are presented in Chapter V. The maximum concentration of Black dye was supposed to be 150 ppm but the resulting mixture inadvertently resulted in 172.5 ppm concentration.

    PAGE 73

    Table 1 Duration of Test Dye Concentration (ppm) Duration of Test Carmoisine (Red) Napthylamine (Black) Dark Green 50.0 100.0 50.0 172.5 50.0 100.0 11/10/77 12/10/77 02/16/78 03/29/78 04/26/78 08/15/78 12/8/77 02/14/78 03/28-78 04/24/78 08/13/78 09/21/78

    PAGE 74

    CHAPTER V RESULTS AND DISCUSSIONS Dye-Absorption Spectrum Results The absorption spectrum of different dyes is shown in Figures 20 through 25 where the absorption coefficient a-^^ is plotted against wavelength A. The absorption coefficient is a function of concentration of the dye and the temperature and pressure of the system. In the present analysis the temperature and pressure effects are ignored. Figure 20 shows the absorption coefficient for 50 ppm Carmoisine (Red) dye* taken both before and after the experiment. This dye shows the dip in the absorption curve between 0.6 and 0.7 pm which is the red part of the spectrum. Similarly in Figures 22 and 23, which show the absorption spectrum of the Black dye, the dip in the value of occurs between 0.4 and about 0.475 pm which is the blue region. Thus, this dye, strictly speaking, is a dark blue dye and not black. The Dark Green dye, as has been pointed out before, was made up of a mixture of Neptune Blue, Tartrazine (Yellow) and Carmoisine (Red) dyes. The dip in the absorption curve in the visible region is at 0.5 ym which is the green region of the spectrum (see Figures 24 and 25) . Also shown in Figure 22 Appendix III shows the calculations for obtaining from experimental data. 59

    PAGE 75

    e a. a. o 0) o o
    PAGE 79

    0 o 03 -a o nj 0 3 U u o Hi p. to a o •rl O. M O n
    PAGE 80

    65

    PAGE 81

    a p. p. (0 a n) I u 4J U 0) P. CO a o •H 4J O. U O CO 3 00 ri

    PAGE 83

    D. P. O 0) >, X) a (U (U M o 4-1 o M o o. (0 c o •H 4J P. U o m CM (U 3 to •H

    PAGE 85

    73 where the color of this dye is because of -N = Nchroraophore [39]. The bond strength of N=N is 3.45 eV [40] while the energy of the photon in 0.3 ym wavelength is 4.13 eV and thus the degradation of dye is inevitable. However, no one single mechanism explains the fading of dyes and involves three photochemical reactions — oxidation, reduction and photolysis. In the case of Carmoisine it is possible that both photolysis and oxidation reactions are possible for its fading. One possible indication is the change in pH of this dye, in 100 ppm concentration solution, from 5.2 (before experiments) to 6.7 (after experiments) . A similar degradation is found in the Dark Green dye as shown in Figure 24, where the degradation is primarily a result of the fading of the Carmoisine dye. The absorption coefficient of this dye drops by about 85 percent at 0.5 ym. This resulted in the change of the solution color from dark green to transparent green. The other constituents, namely Tartrazine and Neptune Blue, appear to be unaffected. The most stable dye (among those tested) from the solar degradation point of view appears to be Black Napthylamine. For the duration of the experiment, its did not change as can be seen from Figure 22. The dye has the following structure: NaO^S 3" VN/-3 It appears that the OH group together with NO^ is responsible for lightf astness of this dye [41] . However, no satisfactory theory exists SO^Na

    PAGE 86

    74 yet which correlates the lightf astness with its chemical composition and most of the information available is yet empirical. Temperature-Time History of the Stills The measured temperature-time history of the various dye-water systems are plotted in Figures 26-31. As can be seen from these figures, the general trend is similar and thus it would be sufficient to explain one of these figures. Figure 27 shows the temperature-time history of Black dye (50 ppm) on March 6, 1978. The top layer temperature reaches a maximum of about 110 °F around 2:30 p.m. (EST) whereas the maximum solar radiation is reached around 12:40 p.m. (EST). This lag in the top layer temperature is because of the mass of the dye-water system, the ambient temperature and the conduction of heat from this layer to the bottom ones. The glass temperature also follows the similar trend. The back aluminum plate temperature, however, rises very rapidly and reaches a maximum of about 130°F at 1:30 p.m. (EST) (Figure 27). As was mentioned before (Chapter IV), this plate's purpose was to reflect incident solar radiation onto the solar still and at the same time act as a condenser for water vapor. However, because of rapid oxidation of the plate the reflection property deteriorated rapidly thereby increasing its solar absorptance and hence its temperature. Moreover, it can be seen from Figure 27 that the plate temperature is greater than the water surface temperature till about 4:30 p.m. (EST) after which it rapidly approaches the glass cover temperature. Thus

    PAGE 87

    u § > o e o 0) T) <4-l o >. u o •u tn (D B •H I 0) u 3 4-1 n) »^ a> e ' H CN (U •H Fi4

    PAGE 89

    • «>• /--^ B D. O. O in o W U-l o S-l o u w •H x: 0) E •H 4J 00 1 D U rH d •s u D. o B M OJ H s 3

    PAGE 90

    (io) 3-in3Bjaduiai

    PAGE 91

    ^ — N d C cx sx CM 1 — 1 QJ >> TD O CS iH CQ O O 4J cn •H B •H 00 4-1 r-» 1 CTv 0) iH H ct) (-1 0) ft a B M 0) H S 00 CM Q) !-( P 60 •H |i4

    PAGE 92

    80

    PAGE 93

    B o in a a 0) Q) U o o M O cn •H a> 4J • I 00 3 .-I u nj ~ D. 6 >^ 0) cd eg •H

    PAGE 94

    82

    PAGE 95

    s p. fk. CM (U ^ u cd i-i m o >. M O U 0] •H J2
    PAGE 96

    84

    PAGE 97

    u M cd o Oi 4J o >> o 4J CO x: e •H •U I 0) u 3 4-1 (fl M 01 • O.00 0) a\ H ^ u a

    PAGE 98

    86 — ^ e y — N cx e 0o. • >H in iH • •H rH 4-1 i-l i-l >^ O U >^ 4-1 C o o u u i-H ^ — ^ u 3 3 0) 4-1 4-1 u >-l nj cfl 3 3 U 4-1 4-1 Oi 0) 03 )-l !-i E E QJ ^ •H 1-1 u nj 0) 1— ( T— 1 nj >^ n! a e e i-H o o 4-1 4J D4-1 4-1 -H o 0 O O O H H m m CO o o © (aq-^3j/nag) uox ^BipBH J-tsjos o o o o o o o o On CNI r-4 o (Jo) s^in^Baaduisx i

    PAGE 99

    87 till 4:30 p.m. (EST) no condensation takes place on this plate thereby justifying the assumptions for equations (38) and (39) . The temperature of the bottom layer, in the meantime, continues to , rise since there is heat conducted from the top layers during daytime and also a small portion of incident solar radiation is absorbed by it. After about 3:30 p.m. (EST) the top layer temperature falls rapidly, both because of rapid evaporation of water from the surface as well as the cooling of environment. At around 3:30 a.m. (EST) the bottom layer temperature and top layer temperature merge. This is because after this time the convection currents are set up and the whole water-dye system loses heat as a single mass. The intermediate layer also follows the same trend as shown in Figure 28. The bottom layer temperature and the time at which it merges with the top layer temperature is an indication of the optical depth of the dye-water system. As can be seen from Figure 26, for the Red dye (50 ppm) , the bottom layer temperature merges with the top layer temperature at around 9:30 p.m. (EST) whereas the two temperatures for the Black dye (172.5 ppm) meet around 1:30 a.m. (EST) (Figure 28). This point can be further explained by looking at the incident solar energy absorbed in different layers of the dye-water system. Figure 32 shows the percentage of solar energy absorbed in different layers at 12:30 p.m. (EST). The Red dye (100 ppm) absorbs about 75 percent of the incident solar energy in the top 1 inch layer, as compared to about 88 percent energy being absorbed by Black dye (100 ppm) in the same layer thickness. This results in the top layer temperature in the Black dye system being much higher, thereby prolonging the time of merging .

    PAGE 100

    88

    PAGE 101

    89 The above point can be further clarified by comparing the temperature-time history of the dye-water system with that of the control still. This comparison is shown in Figure 31 where the temperature of the top layer of Black dye (172.5 ppm) and water are plotted for the same day, March 31, 1978, The top layer and bottom layer temperatures for control still merge around 8:30 p.m. whereas for dye-water system they merge at around 1:30 a.m. (EST). Again, this can be explained by Figure 32 where it can be seen that about 20 percent of the incident solar radiation reaches the bottom layer of the control still as compared to about 3 percent for dye system. Thus, the convection currents in the control still are set up earlier as compared to those in the still with dye. This also explains the difference in maximum temperature difference between the top layer and bottom layers for water only and dye-water systems. From Figure 31 one can see that the maximum temperature difference for Black dye is about 44°F at 1:30 p.m. as compared to about 10°F for control still. This then leads to more uniform heating of the water in the control still and because of the mass of the system the top layer temperature drops less rapidly as compared to that of the dye-water system. Figure 33 shows the temperature-depth profile for these two systems for different times. Figures 34 and 35 show the plot of temperature-time history of the top layer temperature and glass cover temperature for dye-water solution and water only for two days. The higher temperature of the top layer of water during night (after 7:30 p.m. EST) as compared to that of dye-water solution results in higher temperature for the glass cover. The higher top layer temperature is also an indication of the

    PAGE 102

    O U U a o u u 01 4J n) a> "O (M O
    PAGE 103

    LO r>.

    PAGE 104

    •H M -a! 4J m o (U (4-1 o o u CO •H x: ^ 4-1 I 0) M 3 4J tC M Q) . D.CO e r-. H .H n 0) »^ 3 •H

    PAGE 105

    93

    PAGE 106

    00 •rl (0 o (U J3 O >» o u CO •rl Xi (U .% u I (U 3 «fl (U § m
    PAGE 107

    95 g , — > e a, a, o rH iH o ,—1 •H .-1 4-1 C/2 4J oT i-i >, O a rH d U o 0) 4-1 d 0) C 0) 4J o Q) c o o U o ^ — ^ ' — ' O ^ ^ re re (U 3 >-i 4J n3 (D 4-) 4J n! > T) 01 0) o o n! >, >^ o o Pi m iH OT w M CO w cd ex. a O o o H H a: oo c>o © o O O o ro a H ON CO P 0) d o o in 0) S •H H o • ON <; o n

    PAGE 108

    96 higher output at night from the control unit as compared to that from the still with dye. This point will be further clarified in the next section. Productivity of the Still One consequence of the temperature profiles explained above can be seen in the distillate output. Figures 36 and 37 show the histograms of the distillate output for the two systems for April 15 and May 15, 19 78. Since the top layer temperature in the dye-water system temperature is higher than that of the control unit during daytime, 2 the distillate output is higher (about 0.075 lbs/ft hr) for this 2 system as compared to that from the control still (about 0.048 lbs/ft hr) . The maximum distillate for still with dye is observed around 2:30 p.m. (EST) (Figure 37) while that for control is at 10:30 p.m. Consequently, around 60 percent of the total distillate output for the dye system takes place during daytime (7:30 a.m. to 7:30 p.m.) as compared to that from the control unit where only 30 percent occurs during this time and 70 percent during the night. It is anticipated that because of the higher brine temperature of the control unit at night, greater heat loss occurs from the bottom and the sides of the still, as compared to that from the still with the dye. This results in lower efficiencies for deep basin still (with water only) . Table 2 shows the comparison of efficiencies, for various days, between the control unit and the dye unit. The distillate output, however, depends both on the brine surface temperature and the temperature difference between the brine surface and

    PAGE 109

    CO iH O M c o u to (U >> Xi u •rH rH iH •H 4J (0 6 o u d 4-1 00 o r-l 0) nj m r-l iH rH •H .H 4J i-l •H Q <3 n

    PAGE 110

    o O rsi o <> o • o^ <; 04 o 0-1 o C) <[> Q) <4> O i o O H tn o e H 0 00 O <> o -X0 0) o o • (aq/sqx) a^Hixi^STa

    PAGE 111

    O U 4J a o a e o fr C (U X! 4J •H rH •H u U) B o u 14-1 00 4J 3 rH +J 3 in 0 H (U iJ (13 .H -H •r( tH JJ w •H •H •u Q M ro 0) a 00 •H

    PAGE 112

    100 o o in o CN O B o< o in 0) c OJ 01 !-i 4>({) C)

    (bo <> 0 <> <> () <>(D o ( p < > c c H CO Q 3 O 0) e •H H O CO C • 00 (jq/sqx) a^BXXT^STQ

    PAGE 113

    101 — s H nl J-) C OJ 6 •H J-i 0) C X w CM OJ ^ Q H U-l O C 0 CO •H B o u 0) • a u-i (-1 U-l CM u 0) •H O •H <4-( U-l W ta r-H ^^ o X3 Q o cn O 43 H 4-1 U3 •H Q T3 3 O J-> CO eq o c e O D. Q Oi Q OJ XI B 0) > o 00 as 00 O r-^ in CO \o CM rH i-H CM CM 1—1 rH r-^ 00 <-t CJ> rH in vO -I 00 CO CO CO CM CO CM CM CM CO <3" ro CO iH CO eg 00 CO ^ a. a. ai <: < a o o c OJ cu o 00 CM CM cn 3 00 3 o o 4J u U-l XI cfl rH rH VO rH rH 3 •rl 4J 4-1 PQ tn •H Q >^ O Cfl O 1 -d o 1 rH 4-1 CM 3 'oj X a. IM a d) •M 3 4-1 •u Cfl u a) PQ rH cfl 4J rH rH Cfl •H O -H M 4J CO rH CO •H ^ — s •H rH 4J >^ T3 Cfl 05 Cfl 4-1 •H <4H o P o H 3 II 4J Cfl OJ XI CI rH c c ^ •rl X) CJ Q Cfl C ai (U U •H c u G (U (fl •r o rH l4-( p O U-l (U CO W a, rH CM CO

    PAGE 114

    102 the glass cover (AT) . Figure 38 shows the plot of h ' vs time and evap At vs time, where ^^^^^ is given by the following equation: q = h AAT (40) m evap and q is given by equation (31) . However, in order to calculate q^ the value of coefficient C in equation (30) has to be found experimentally. This is done by analyzing the distillate output m, calculated from the following equation: m = h A(m. m. ) (41) m IS ig where all the terms in this equation have been explained in Chapter IV. This m is plotted against the experimental value for two different days. From this plot the value of C is determined to be 0.08 for all times except from 11:30 a.m. to 7:30 p.m., when it is 0.12. The two values take into account the changing convection pattern inside the still because of condensation on the side glass cover and back plate. These values of C were then subsequently used in the analytical model, whose results will be explained later. Figure 40 shows the experimental and calculated distillate output using the above calculated coefficient C. Once q^ is calculated h^^^^ can be plotted against time (Figure 38) . It can be seen that the product of h and AT has a maximum at around evap 2:30 p.m. Similarly, for water (control still) this maximum takes place around 9:30 p.m. (Figure 39), a fact which is confirmed experimentally (Figure 37) . From Figure 38 it is also evident that h follows the trend of evap surface layer temperature T^ and hence a relationship between them must

    PAGE 115

    13 O to pq u-t O ts (J O 4-1 W •H x: 0) 4J 00 I U a\ 0) H <4-( CO C LO Vj •U r-( •H U M 0) <; J= c ^ o e •H D, cd o • cd > W ^ CO 01 3 M •H

    PAGE 117

    105

    PAGE 118

    00 < 6 a, p. m CM >, T3 U pa I 3 P. 4J 3 O u CO o 0) 3 00 •H

    PAGE 119

    107

    PAGE 120

    108 exist. Figure 41 shows such a relationship. Here the value of h ° evap for various days and different dyes is plotted against the surface layer temperature and the resulting empirical relation is given by: h = 0.284 exp (0.0253 T )-1.25; (Btu/f t^hr°F) (42) evap s where is in °F. However, a more meaningful relationship for a researcher in the area of solar distillation is that of distillate output and T^. Figure 42 shows such a relationship where the experimental distillate output has been plotted against for various dyes and different days. The least square curve fitting the data is given by: D = 0.011 exp (0.0168 T^) 0.31; (Ibs/ft^hr) (43) where is again in °F. There is a theoretical basis for such a plot. It has been shown in equation (31) that distillate output is a function of T^ and temperature of the glass T^ . However, these temperatures are, in turn, functions of input solar radiation and ambient temperature. Finally, the ambient temperature is also a function of solar radiation and thus a relationship exists between distillate output and T^. It should be noted that if the glass cover temperature or T^ is changed artificially then equation (43) is not valid. Thus, this equation exists only for passive type solar distillation units. It is also instructive to analyze the different energy exchange mechanisms from the brine surface to the glass cover and the losses to the environment from the sides and bottom of the still. Figures 43 and 44 show different energy exchange mechanisms for the dye and control still, respectively, for April 14, 1978. Each loss is a percentage of

    PAGE 121

    H

    PAGE 122

    110

    PAGE 123

    Ill o o CO O CM 00 — . CO ^ — N CO ^ — N ON ON ON CO o i-i iH — , o CO rH o CO ^ u U-) a. o iH < < rH <: u ' — •H ^ — u >. U w 0) Cl, at S >^ >. < >. >s 13 V a u cd u ^ u ^ d 13 0) o O 01 u OJ C u n) m •u ca QJ CO n3 fH iH n) .H I-I 4-1 PQ o o 0 0 OOQ 1 o o o o 00 C nJ 4J 3 3 O (U OJ •H -d <4-l O c o H 4J n3 rH
    PAGE 124

    T3 M o « PQ O CO a to •rl C to o (U 6 t-i 00 (U 14-1 ay en <-i C nJ , CN d < ^ — ^ n (U 3 00 •H

    PAGE 125

    113

    PAGE 126

    I u ^>-l •H nj c a. <: -J0) u o M •H

    PAGE 127

    115

    PAGE 128

    all the energy exchange by the four mechanisms accounted. As can be seen from these two figures, greater evaporation takes place from a dye surface than from the control still surface. Also, the losses from the bottom and sides in the control still are greater (about 10 to 15 percent of the total loss) as compared to those from the still with dye (about 5 to 10 percent of the total) . Indeed from the control still the bottom and side losses are about 2.5 times those from the dye-water still. This is the cause of the lower efficiencies exhibited by the deep basin still with water only. Comparison of Different Dyes As was pointed out in Chapter V before different dyes were tested, the two stills were filled with water only and their distillate output was monitored. Figure 45 shows the correlation of distillate output of the two stills, and as can be seen they are identical. The dyes are compared to each other by comparing the distillate output from the dye-water system m^, against that from the control still, m^ . Such comparisons have been made for different dye concentrations (Figures 46 to 51) . The data have been plotted for completely clear and "nearly clear" days. Nearly clear day has been defined arbitrarily as a day in which the sun produces a shadow for at least 70 percent of the time. As can be seen from these figures a linear regression line is used to fit the data points. However, the results from these figures become meaningful only if some way of generalizing them is found so that the effect of a certain dye in increasing the evaporation can be calculated.

    PAGE 130

    118

    PAGE 131

    119

    PAGE 132

    120 ON o II c o •H U n) tH vo 0) ON 1-1 CM U O O . u 0) o

    PAGE 133

    6 o m u i-i o 3 O. 4J 3 O 0) CO •rl TJ «-l o c: o M •H cd go a CO 0) a 60 •H P4

    PAGE 134

    122 00 o u o 1-1 u 0) -a u n) a C (0 CO a o •H 4J 03 00 rH O OJ O 0 1 a c o 0 I-l J-J vo c C O 0) o 0^ •H •H a O •H cn U-l CJ II U-l 01 O O 0) S U e»3 O O in o ^ >^ as -O I 4-1 U-l w O J3 o d O d in o o o CN O (XBp-^3j/sqT) ui

    PAGE 135

    a p. in >, U 0) tH PQ O 3 3 O o c o CO •H (-1 n) D. e o u 3 00

    PAGE 136

    124 CO o o +1 II u o M U a) u rt -o c td 4-1 00 o II o •H 00 n) >H r-H CO QJ .-1 U U o O U + C y-i •^^ a o r-i 0 u d C o (D •H vO •H tfi u m o •H 0) U-4 t-i 11 14-1 Ml 0)
    PAGE 138

    126

    PAGE 140

    128 vD ro O O + 1 i-t O u u 0) T) U nj TD C CO 11 11 r-" Q ,—1 n 1 O )-l M d O O QJ + •H XI O .H S 4J d c o 0) H •H CO O M O •iH Q) l+-t V-J II bO (U 0) a o »S 6 o a^ o 00 d O I (N U o ui o o 00 o d o o d o (ADp-^aj/sqx) "ui

    PAGE 141

    129 A generalization scheme is thus attempted by relating the slope S of the regression line, for different dyes, and the absorption of solar radiation by dye-water system. From Figures 46-51 it can be seen that the slope decreases as the dye is changed from red to black. This is to be expected, since black dye absorbs more radiation at the surface as compared to red dye and hence increases the output. Figure 52 shows the plot of slope S against a nomalized absorption x» which is an indication of how well a dye-water system absorbs solar radiation and is given by the following equation: ^2 [1 exp(-a^x*) ]dA where q^ is given by equation (8) and x* is the path length of the solar radiation in the dye-water system. In Figure 52, x* is arbitrarily taken to be 1/2 inch. Thus, as x increases the slope S decreases and the relationship is given by the following regression line, S = 1.444 1.028 X (45) with a standard error of ± 0.072. Figure 53 presents the slope S versus the intercept b of various regression lines (Figures 46 to 51) . This relationship is given by the following regression line: b = -0.5726 S + 0.5132 (46) with a standard error of ± 0.0325.

    PAGE 142

    a cd o C! O •rl (U U U o 1-1 3 •rl

    PAGE 146

    134 Thus, by using equations (45) and (46) the percentage increase in evaporation can be evaluated in the following manner. If the distillate 2 output from the control unit is m^(lbs/ft day) and that from the dye2 water system is m^(lbs/ft day), then the percentage increase (I) in evaporation is given by: m J m I = — X 100% (47) m c and if the equation of the regression line is written in the form m^ = Sm^ + b (48) then, I = ^1 (1 ^ ) 1} X 100% (49) c Thus, for any dye where x is known (which can be found from its absorption spectrum) the percentage increase in evaporation can be found using equations (45) through (49) . From equation (49) it is evident that this increase is dependent on the distillate output from the control unit, m . c An attempt was also made to plot m^ and m^ for different months of the year (see Figures 54 and 55) . It is obvious that the maximum of m^ occurs in June, since the solar radiation is also maximum at this time. It should be po inted out that the above equations for percent increase are only valid when the outputs from dye-water system are compared from an equivalent control still of depth 10.5 inches. In both Figures 54 and 55 the distillate output is for clear and nearly

    PAGE 147

    135 •H to < o O U u c o o B o u C o m o in -a0) (-1 3

    PAGE 148

    J I o 00 o O o o O O r-i O (y^Ep-^qj/sqx) tn 0) w D O •H ^1 CO > O u n3 4J c o U-l O C o •H > c o CO to (U (U u 3 60 •H

    PAGE 149

    137 clear days. This is because there is no difference in outputs from both the stills on completely cloudy days (days when the sun did not produce any shadow) as shown in Figure 56. The reason for the two outputs to be the same on a cloudy day is not difficult to see, since on such a day the amount of radiation reaching the water surface is very small (about 5 percent of the clear day radiation) and diffuse and thus is absorbed mostly at the surface of the water, thereby making their outputs nearly identical. Finally, the absorbance spectrum of the distillate from dye-water system is plotted in Figure 57. This is done to ascertain whether there is any carry-over of the dye molecules. As can be seen from this figure there appears to be no indication of such a carry-over. This is to be expected since these dyes have boiling points of about 350°F and above [41]; whereas the maximum surface temperature recorded for dyewater systems has been about 130°F and thus only water evaporates. Moreover, because of quiescent evaporation from the water surface, there is no carry-over of the dye by convection currents. Comparison of Theoretical and Experimental Results ^ The computer program for the analytical model was run for various dyes to compare the predicted temperature-time history of the dye-water system with that obtained experimentally. The input to the program included the solar insolation, ambient temperature data, wind velocity, value of coefficient C [in equation (30)] and the value of average transmittance T , for condensate covered glass cover. The solar

    PAGE 150

    O r-l U 0) rH I u o •H 4J CO o S o 3 4J 3 O (U •H • 4J CO 03 Q 3 60

    PAGE 151

    139

    PAGE 152

    140 00 ON 00 o> ,—1 ,—1 U 0) iH g •H QJ 4-i P. < 0) CO pu p. o . c 0) O OJ nj u .-H o pa e e o o (-1 >-l y-j (1) QJ 4-1 u 1-1 ca n3 cu iH iH r-i iH cC •H •H 4-) 4-J cn •H o o H O O [> c 1-1 QJ in > CO HI CO o •rl CO > G o u UH Q) u aouuciaosqv

    PAGE 153

    141 radiation and the ambient temperature data were modeled in the program by polynomial expressions. These polynomials fitted the data well (maximum error was ± 2 percent) . The determination of the coefficient C has already been explained. The value of was experimentally calculated by making a First Law of Thermodynamics analysis on the still, and after making corrections for the reflection losses of incoming solar radiation from the water-air interface. This correction was made using equation (12) and an average reflection loss of 11 percent from the water-air interface was obtained. With this analysis the value of x , was obtained as 0.75, which was subsequently used in the program. It should be pointed out that this is an average value for the whole day and will obviously change during the day, but for simplicity it has been used as such. The thermophysical property values for water-vapor viscosity, density, etc. have been calculated at an average temperature of 100°F (see Appendix I for data compilation) . The comparisons of temperature-time history for dye water systems, between calculated and experimental results, for different dyes, are plotted in Figures 58 to 62. As can be seen from these figures, the agreement between the calculated and experimental results is excellent. However, for Red dye (Figure 58) there was a slight difference (about 4°F) between the calculated and experimental temperatures of the top layer, after 1:30 a.m. This difference can be attributed to lower wind velocity used in the program and approximations used in the value of coefficient C [equation (30) ] . The wind velocities used for computer simulation were average for the whole day. However, on certain days there were considerable fluctuations in the wind speeds during day and

    PAGE 154

    tn u ^ to 0) ^1 to 4-1 0) e a. X O o z x: . N e c a. cx > o T3 tn •H T3 u (U tfl Pi a U o O o U-l t30 LO OJ (-1 tJO •H

    PAGE 155

    143 u C l-l n) QJ cd 4-1 G 4-J C •H C -a (1) U QJ a) e QJ e 4J •H P•H o )-l X S-l •H OJ QJ 01 T3 P. V — ' QJ X X U 0) QJ QJ CI. ^ — ' ur QJ W W U yi 3 >-i H iH u QJ 4-1 CTJ CD
    PAGE 156

    CO u a 0} (U M i-l cd u a eri D. • >i 00 a) 13 C CO *< vO rH ca o o •H u 4-1 CO 0} s u o 0) x; e 4-1 c o 4-1 J iH (« P. s !-i o O u CM LT) (U U 3 00 •H Fl4

    PAGE 158

    t/1 4-1 iH 3 0) M iH n! JJ c (X i-l rH 13 « o (J cd o a •H 4-1 0) (J e O a. 0) a. -C 4-1 in 0) 4-1 X) c o CO o •H V-i rH nj pq a E o o u 4-1 o U D M •H ^4

    PAGE 159

    147 4-J i-H C13 4-J b: C *H 1-H GJ !-i CO OJ 4-1 ' ^ •H G. C X 0) CD d) -i Cu (U D X 1-1 j-i 0) OJ ct) u ^ — ' 3 S-i 0 4-1 CJ i-i CD & n3 1-1 e (J QJ 0) (1) rH .-1 CI4J iH g e (-1 0) II II ye •U OJ T3 4-1 0) CO 4J i-H e O 1-1 o •1-1 •rH u T3 4-J D. 4-1 QJ W C o 1-p •H H m Pm Q o oo o • o CO C7N Oi o m u o e •f-i H O O • o •Jo CM O o o CO o o -3(io) S-in^Baaduiax

    PAGE 160

    CO r-t 3 CO 0) M r-l to C (U CO M (U p. H X
    PAGE 161

    149 — ^ r~H — N iH n3 CO 4-i I — 1 c CT3 CD OJ T3 E E -i tx ' — ' X ^ — ^ Qi ^"^ QJ LJ u ^ i-l J-i 3 y-i iJ m 4-J QJ 0^ TO >-i • e rH
    PAGE 162

    4_( , — 1 CO 1) CD p, C I-) nj .H in nJ 1-1 a •r-l >^ 4-1 0) a >-l o
    PAGE 164

    152 night and this may be one of the reasons for the difference in calculated and experimental temperatures. The maximum calculated temperature difference for convection heat transfer between the bottom and the top layers in the dye-water system was about 0.5°F, which is within the maximum accuracy range of the recorders and thus was not observed experimentally. However, even at this temperature difference the apparent conductivity as calculated from equation (17) is about 30 times the value of thermal conductivity of water. This then resulted in producing nearly the same temperatures for the layers. Figure 62 is the plot of temperature-time history for Green dye. There is about 4°F difference between the predicted and the experimental values of the top layer temperature around 2:30 p.m. This can be attributed to the uncertainty in the value of a, . As was pointed out A earlier this dye undergoes degradation by sunlight and in the absence of the knowledge about the rate of degradation, an average value of a, A (before and after the experiments) was used. Finally, the temperature profile for Black dye (172.5 ppm) for March 31, 1978, is plotted in Figure 63, showing the comparison between the calculated and the experimental results. The calculated top surface temperature is slightly lower (about 2°F) than that of layer 1/2 inch below because of evaporative cooling. There is a difference of about 6°F between the calculated and experimental top layer temperature around 6:30 p.m. This can again be attributed to the approximation of coefficient C and wind velocities. It should be pointed out that the distillate output calculated from the computer program is within ± 10 percent of the experimental value.

    PAGE 165

    153 o o -o <]X1 CX) o w OJ .H •r-l U-( 0 a. OJ V-i 3 4J tfl »-l OJ & E -l E 0 a x; in c OJ ^ JJ 0) 0) >> XI c 0 w •H .-1 pa a e 0 0 0 U-l ro 0) U 3 •H

    PAGE 166

    154 Effect of Various Variables on Still Productivity Once sufficient confidence is achieved in the validity of the analytical model it is appropriate to investigate the effect of various parameters like ambient temperature, wind velocity, and various concentrations of different dyes, on the productivity of solar stills. Effect of Ambient Temperature The effect of various ambient temperatures (expressed as percent of the actual ambient temperature, T^^^^) for March 31, 1978 (Black dye, 172.5 ppm) on distillate output is plotted in Figure 64. As can be seen from this figure a decrease in T^^^^ by 20 percent increases the distillate output by 10.2 percent, while an increase of 1^^^^ by 20 percent decreases the output by 13.6 percent. This is to be expected since condensation of distillate on a glass cover is primarily a function of the temperature difference between glass and ambient, for the same outside heat transfer coefficient. This result is contrary to some of the results of theoretical and experimental studies [18] documented in solar distillation literature, where it is reported that increasing the ambient temperature increases the productivity of the still. The justification for this has been provided by faulty argument that an increase in water surface temperature results with increase in ambient temperature, thereby increasing the distillate output [18]. However, it should be painted out that the water surface temperature does increase with increase in ambient temperature but the temperature difference between the water surface and the glass cover (the driving potential for mass transfer) decreases. Figure 65 shows the plot of

    PAGE 167

    1-4 M O (u M 4-J rt M V Q< el S nl w •H TO CO '"^ o 0^ iH o Q) rH U-l U-l . •H iH W to M •H < XI D •rl fa

    PAGE 169

    u o >1 M o O & 4J •u to •H >-l o 00 •u 1 a\ (U rH u 3 4J rH ca CO 0) x: o. o to s e w a. o d in ta o CM rH ^ •H 4-1 C!
    PAGE 170

    158

    PAGE 171

    159 surface temperature of the water surface and the difference between the temperatures of surface and glass versus time for two ambient temperatures. The difference in maximum surface temperature for these two conditions is about 40°F (about AO percent increase in temperature, for higher ambient temperature condition, over that for lower ambient temperature condition) at about 2:30 p.m. (EST) whereas the AT(T T ) s g for lower ambient temperature condition is 4 times that for higher ambient temperature conditions. As has been pointed out before, evaporation heat transfer is a product of h (function of T ) and AT evap s and thus the lower ambient temperature condition results in higher evaporation (and thus greater productivity) compared to the higher ambient temperature condition. Effect of Wind Velocity The effect of wind velocity on distillate output for the still is plotted in Figure 66. There is an increase of output by about 10 percent with wind velocity increasing from 0 to 25 mph. This is because there is better heat transfer from the glass cover to the ambient. However, with increasing wind velocity (about 25 mph) the output starts leveling off. The reason for this is that at higher wind velocities the limitation on heat transfer is imposed by the conductance of glass and not by the outside heat transfer coefficient. It should be pointed out that there are conflicting results presented in the literature on the effect of wind velocity on productivity [18] . However, some computer simulation results [22] do confirm the increase of productivity with increasing wind velocity.

    PAGE 172

    160

    PAGE 173

    161 Effect of Dye Concentration The effect of various dye concentrations on distillate output for the three dyes is plotted in Figure 67. At zero ppm concentration they have the same output, and as the dye concentration increase, different dye-water systems exhibit different outputs with Black having the maximum distillate while Red having the lowest. This is because of better absorption of solar energy by Black dye as compared to Red. However, after about 500 ppm there is no difference in distillate outputs from various dyes, since around this concentration level about 99.9 percent of the incoming solar radiation is absorbed within the first 1/2 inch layer of dye-water mixture, thus making the evaporation independent of the nature of dye. It is also instructive to examine the surface temperature and bottom layer temperature-time histories for Black dye with 50 ppm and 400 ppm concentrations (Figure 68) . Besides the higher surface temperature for 400 ppm concentration, an interesting point can be seen in the bottom layer temperatures for the two concentrations. The maximum variation in bottom layer temperature for 400 ppm concentration is 2°F while for 50 ppm concentration it is 7°F. These temperature variations are an indication of heat losses from the bottom and sides of the still, which are ultimately reflected in the productivity of the still. Thus, in a way the above point is an indication of the effect of depth of the dye-water system on productivity, since by increasing the concentration of dye the effective depth of the still is increased while with decreasing concentration of the dye the depth is decreased. Thus, it can be concluded from this argument that for the same concentration of dye, increasing the basin

    PAGE 174

    162

    PAGE 175

    163 CO (U u M OJ a E a) u (J iH P. o H E e w a o. 0) o. u o 3 o o 4J m (« M U-l U-i o O a, e c c o o 4J •H •H U S-i m n3 OJ )-i >^ 4-1 4-1 n) C a 0) 0) u u e c c o o o 4-1 o o 4-1 0 0) >, o oo o • r-H en •H in <; a. <: 0) >s T> o u CO r-i m v< fo >. u o o 4J CO (0 •rl On x: • s H •H . 4J to 1 C 0) o u U -H O ' — ^ 3 4J CO 4J n) (0 M in u u u 3 0) c o a 0) 6 y (u c: 4-1 O (U o B 0) •H j2 qj o H 4J >, -o 14-1 H O 4J c 4J OJ o u rH 0) O. «-( U-l rH -iH n) T3 O O CO •H O 4-1 S 4-1 rH U c o <3 MH o • in <; o o o o o 00 o o

    PAGE 176

    164 depth will increase the productivity. In fact, it can be shown that if the depth of the basin is greater than 15 inches, then the bottom layer temperature remains unchanged during a 24 hour cycle. For the researchers in the solar distillation area, the important piece of information from the above study is some estimate of the optimum cost benefit ratio of using a particular dye. The problem is one of seeking the lowest concentration, C, that will achieve a high output, M, and it is therefore especially sensitive to the rate of change of output with concentration, dM, or in symbolic form, dc ijj = tjj (C, dM, M) (50) dc In the present study a particular form of this function is proposed: F E A dMc + M (51) dc Figure 69 shows the plot of equation (51) with A equal to 1. From this plot the concentrations of Black, Green and Red dye (where the value of F is maximum) are 218, 377 and 408 ppm, respectively. The optimum concentration increases from Black to Red dye, an observation to be expected since a lower concentration of Black dye (because of better absorption of solar energy) gives the same distillate output as a higher concentration of Red dye. These optimum concentra2 2 tions correspond to distillate output of 0.823 lbs/ft day; 0.842 lbs/ft 2 day and 0.843 lbs/ft day for Black, Green and Red dyes, respectively. Thus, an increase of about 2.3 percent in distillate output of Green dye over that from Black dye is ach ieved at the cost of increase in

    PAGE 177

    165 (Aep-^aj/sqi) i

    PAGE 178

    166 concentration of Green dye over Black dye by 73 percent, which is directly reflected in the cost of the dye. It should be pointed out, however, that the above numbers for optimum concentrations are only valid for that particular day (April 14, 1978) and will change with seasons, with the largest value of this concentration for a particular dye being in winter and lowest in summer.

    PAGE 179

    CHAPTER VI CONCLUSIONS AND RECOMMENDATIONS Conclusions The following conclusions can be drawn based on the present investigation; points 1 to 7 are based on experimental data, while points 8 to 11 are extrapolation on the analytical model. 1. By adding water-soluble dyes the productivity of deep basin solar still can be increased, by as much as 29 percent (for Black dye with concentration of 172.5 ppm) , over that of a still with no dye. 2. Among the dyes tested. Black Napthylamine dye has been found to be most suitable, both from the point of view of increasing the evaporation and lightf astness . 3. There is rapid fading of the Red Carmoisine dye under sunlight. Thus, in one month of exposure to sunlight its a, at 0.5 Mm A wavelength is reduced by 95 percent. The Green dye also exhibited slight degradation and thus reduction in its a, at A 0.5 im wavelengtli. This can be attributed to its Red Carmoisine dye composition (33 percent by weiglit) . 4. Based upon the experiments, a simple method of calculating the percentage increase in evaporation by using a dye over that 167

    PAGE 180

    168 from the control unit has been developed [equations (44) to (49)]. An empirical relationship has been developed for h and evap distillate output as a function of top layer temperature T^, and is given by equations (42) and (43) , respectively. There is no difference in the distillate output from the control still and the still with dye on a completely cloudy day . An analytical model which predicts the temperature-time history of the dye-water system has been developed and the agreement between the results from it and the experiments is excellent over the range of parameters investigated. The analytical model predicts a decrease in distillate output with increasing ambient temperature. Thus, a 20 percent increase in the ambient temperature results in a 13.7 percent decrease in output. The effect of wind speeds on distillate output, as predicted by the analytical model, is to increase it with increasing speeds. Thus, output increases by 10 percent with the Increase in wind speed from 0 to 25 raph. An increase in distillate output occurs with increasing concentration of dyes till 500 ppm, after which it is independent of the concentration. However, the rate of increase of output with concentration is different for different dyes with Black having the maximum rate and Red dye having the lowest.

    PAGE 181

    169 A method of predicting an optimum concentration of a particular dye, at the point where the sum of the rate of increase of distillate with concentration times concentration and the distillate with concentration is maximum, has been developed. For the spring conditions at Gainesville, Florida, the optimum concentrations for Black, Green and Red dyes were 218, 377 and 408 ppm, respectively. Recommendations for Future Investigations It will be worthwhile to develop water soluble dyes which have higher in the wavelength range of 0.7 to 0.9 ym and have higher x as compared to those tested in the present study. Besides higher x an attempt can also be made to develop dyes with greater lightf astness so that there is no degradation by sunlight. A worthwhile effort would be to study the effects of dyewater surfactant complex on evaporation of water. Addition of surfactants decreases the surface tension of water and with better absorption of solar energy near the water surface with addition of dyes, there is a possibility of enhancing the evaporation of water. Besides the alteration of the surface chemically, an effort should also be made to study the effect of mechanical disturbances on evaporation. An effort should also be made to look, at different ways, both chemical and physical, of extracting the dye from concentrated brine so as to recycle it. Success in recycling the dyes may

    PAGE 182

    170 ultimately prove whether the use of dyes in large scale solar distillation plants is economically feasible or not. 4. Besides the use of these dyes in solar distillation it appears to be advantageous to use them for disposal of effluents, from chemical plants, by evaporation. 5. There also appears the possibility of using these dyes in liquid solar collectors.

    PAGE 183

    APPEITOIX I PROPERTY VALUES USED IN ANALYTICAL MODEL

    PAGE 184

    APPENDIX I PROPERTY VALUES USED IN ANALYTICAL MODEL —6 2 Thermal diffusivity of water = 1.64 x 10 ft /sec [36] "^glass Solar absorptance of 1/8 inch thick glass = 0.07 [42] 3 Thermal expansion coefficient of water = 0.2 x 10 (1/°R) [36] Cp gj^^gg Specific heat of glass = 0.2 Btu/lb°F [43] D. . Binary diffusion coefficient of water vapor in air = 1.08 ft^/hr [44] F F = 0.82 e a h^g Enthalpy of vaporization of water = 1053 (52/90) (T-70) Btu/lb where T is in °F [45] Outside convection heat transfer coefficient = 1 + .304 V, £^-2hr°F ^'^^^^ ^ m.p.h. [46] K, Thermal conductivity of water = 0.364 , ^J^"o^ [36] w hrft F Thermal conductivity of water-air mixture = 0.016 , ^j^^o [36] w/d. hrft F K Thermal conductivity of glass = 0.59 . [43] Thermal conductivity of Plexiglas = 0.072 . ^ [47] K ^poly Thermal conductivity of polyurethane = 0.0133 Btu . . , , h^fpF f"*^^ n^, n^ Indices of refraction for air (1) and water (1.33), respectively 172

    PAGE 185

    173 w w -5 2 Kinematic viscosity of water = 0.74 x 10 ft /sec [36] Kinematic viscosity of air-water mixture = 17.9 X 10~^ ft^/sec [36] Partial vapor pressure of water given by the following equation [44] P = P X 10 w c X ( a' + b'x + c'x"^) T 1 + d'x ; Ib^/in. (I-l) where P = c 218.167 X 14.7 Ib^/in.^ T = t + 273.16; °K X = 647.27 T t = temperature of the water, °C a' = 3.24378 b' = 5.86826 X lO""^ c' = 1.17023 X 10"^ d' = 2.18784 X lO"-^ Pr S c UP w/a Prandtl number of water-air mixture = 0.69 [36] 3 Density of ordinary glass = 162 lb/ft [43] Schmidt number for water-air mixture = 0.61 [36] Product of overall heat transfer coefficient and perimeter of the still = 1.56 Btu hrftT

    PAGE 186

    APPE^^DIx II STABILITY CRITERION FOR DIFFERENCE EQUATIONS

    PAGE 187

    APPE>]DIX II STABILITY CRITERION FOR DIFFERENCE EQUATIONS In the calculation of the stability criterion for the difference equations formulated in Chapter III, the method of Arpaci [4 8] is followed. The nodal equation for n^^ layer is: k ACt'^ T^^ k A(T*^ , T*^) t"^ e n n+1 , e n-1 n , n . %hs A^^ ^ 0.64 = ^t+At _ ^t p C Ax A(-^^— ~) ,,,, p At (34) where 2 n , ,2 n Qabs = ^A,i(^) exp(-a^Ax*) dX} and the steady state solution of equation (34) is: k^ Ad^+f t') k A(T^ ^ T^-^^'^) (T^+^^ t"^) Ax Q _ e n, s n+1 ^ e n-1 n,s _ n,s °o abs Ax Ax 0.64 = 0 (I 1-1) Subtracting equation (II-l) from (34) yields: 175

    PAGE 188

    176 k A(T^+^^ t"^) k ACt"^-^^*^ T^) (T e n, s n_ ^ e n, s n_ _^ __ t+At X Ax Ax 0.64 p C Ax A w p n n__ At (II-2) Since steady state temperature T*'^'^'' will be greater than j^'*'^^ thus, n,s " n ^t+At _ ^t n, 8 n ^t+At _ ^t n n (II-3) Rearranging equation (II-2) then leads to: ,t+At .t+At T T p C Ax A w p At (II-4) 2k A e Ax Ax 0.64 Hence, the stability criterion for equation (34) is: At < p C Ax A w p 2 ^e^ ^ Ax Ax 0.64 (II-5) Thus, with a temperature difference of 0.5°F between the bott om surface of the still and the top surface of water (this gives the 2 maximum value of apparent conductivity k^ of 12 Btu/hrft "F) and for layer thickness Ax of 1 inch, equation (II-5) results in:

    PAGE 189

    177 At < 0.018 hrs. (II-6) In the program the value of 0.01 hrs. for At was subsequently used. A similar sort of analysis on the glass cover yields the value of At = 0.07 hrs. and thus At = 0.01 hrs, the minimum value of the two, was used.

    PAGE 190

    APPENDIX III CALCULATION OF ABSORPTION COEFFICIENT OF DYE SOLUTION

    PAGE 191

    APPENDIX III CALCULATION OF ABSORPTION COEFFICIENT OF DYE SOLUTION The transmittance of the cell-water solution (for Perkins-Elmer Monochromator) can be found using the method outlined below. Figure III-l shows the schematic diagram of the cell and attenuation of incident beam on glass 1-2. p2(l-pj^)ag* t* ag*-^p„ p. (1-pJ (1-P-) ag*(lp^Xl-p^) Figure III-l. Attenuation of incident beam through glass. The spectral transmittance of glass plate 1-2 can be written as. or ^1-2 = "g ^^-Pl^ ^^-P2^ *^ 2 2 1 + p^ P2 + P2 P^^ + • • • 1-2 g ( 1 a^^ g '2 Pl^ (III-l) (III-2) where 179

    PAGE 192

    180 a* = spectral absorption of incident radiation by glass = l-exp(a t) g = spectral reflectance of glass-air interface = spectral reflectance of dye-water-glass interface Similarly the transmittance ^^'^ ^3-4 written as. *2 (1 P2 P3) (III-3) where a, = spectral absorption of incident radiation by dye solution d * l-exp(a^t ) and ^3-4 = «g (I-P3) (1-P4> (III-4) (1 a^*^ P3 p^) Therefore the overall transmittance for the cell-water solution is: "^cell/w "^1-2 "^2-3 "^3-4 (III-5) or cell/w a (1-p^) (I-P2) (1-a*^ Pi P2) a ^ (I-P2) (I-P3) a-a/ P2 P3) a ^ (I-P3) (1-p^) a-a P3 p^) (III-6) since p^ = p^ and P2 = P3, '•^^.g^ then given by the following equation:

    PAGE 193

    181 ^cell/w *2 * 4 2 a a (l-p„) (1-p,) R d 2 1_ *2 *2 2 *2 (l-a^ p^) P2 ) (l-Cg P2 P^) (III-7) thus knowing the values of T^^j^-j^ (empty cell) and ^j^gj^^/jyg (cell with dye solution) the spectral absorption coefficient a can be calculated d from equation (III-7) .

    PAGE 194

    REFERENCES 1. Falkenmank, M. and Linder, G. Water for a Starving World , West View Press, Boulder (1976) , pg. ix. 2. Batisse, M. "Balancing Needs and Resources in the Use of Water," Proceedings of the Study Week, at the Pontifical Academy of Science, April 14-19, 1975 (Ed. Passimo, R.) , Elsevier Scientific Publishing Company, Amsterdam (1976) . 3. Deliyannis, E. and Deliyannis, A. "Water Desalination," Naturwissenchaf ten 65, 462-472 (1978) . 4. Morris, R.M. "Desalination of Sea Water," Chemistry and Industry, 6, August (1977) . 5. Anonymous. "West Germany to Put Solar Distillation Plant in Saudi Arabia," Times of India, 16, March (1978). 6. Howe, E.D. and Tleimat, B.W. "Fundamentals of Water Desalination," in Solar Energy Engineering (Ed. Sayigh, A. A.M.), Academic Press, New York (1977) . 7. Nebbia, G. and Menozzi, G. "A Short History of Water Desalination," Proceedings of the International Symposium, Milano, April 1966, 129-172 (1967) . 8. Harding, J. "Apparatus for Solar Distillation," Proceedings of Institution of Civil Engineers, _73, 284-228 (1883). 9. Talbert, S.G., Eibling, J. A., Lof, G.O.G., Wong, C. and Sieder, E.N. Manual on Solar Distillation of Saline Water , Research and Development Program Report No. 546, United States Department of Interior, April (1970). 10. Howe, E.D. and Tleimat, B.W. "Twenty Years of Work on Solar Distillation at University of California," Solar Energy, 16^, 97105 (1974) . 11. Telkes, M. Research on Methods for Solar Distillation , Research and Development Report No. 13, Office of Saline Water, United States Department of Interior (1956) . 12. Coffey, J. P. "Vertical" Solar Distillation," Solar Energy, 17, 375-378 (1975) . ~ 182

    PAGE 195

    183 13. Bloemer, J.W. and Eibling, J. A. "Final Three Years," Progress on Study and Field Evaluation of Solar Sea-Water Stills , Report No. 19, Office of Saline Water, United States Department of Interior, May (1966) . 14. Szulmayer, W. "Solar Stills with Low Thermal Inertia," Solar Energy, lA, 415-421 (1973). 15. Hsieh, C.K. and Rajvanshi, A.K. "The Effect of Dropwise Condensation on Glass Solar Properties," Solar Energy, 19^(4), 389-393 (1977) . 16. Bloemer, J.W. "Factors Affecting Solar-Still Performance," ASME Paper 65-WA/Sol-l, 8 pages (1965). 17. Martens, CP. "Theoretical Determination of Flux Entering Solar Stills," Solar Energy, 10(2), 77-80 (1966). 18. Lof, G.O.G., Eibling, J. A. and Bloemer, J.W. "Energy Balances in Solar Distillers, AICHE Journal, _7(4) , 641-649 (1961). 19. Baum, V.A. "Solar Distillers," United Nations Conference on New Sources of Energy, Paper 35/S/119, Rome, 43 pages (1961). 20. Bloemer, J.W., Eibling, J. A., Irwin, J.R. and Lof, G.O.G. "Analog Computer Simulation of Solar Still Operation," ASME Paper 63-WA-313, 8 pages (1963). 21. Telkes, M. "Fresh Water From Sea Water by Solar Distillation," Industrial and Engineering Chemistry, 45(5), 1108-1114 (1953). 22. Cooper, P.I. "Digital Simulation of Transient Solar Still Processes," Solar Energy, 12(3), 333-346 (1969). 23. Grune, W.N., Hughes, R.B. and Thompson, T.L. "Operating Experiences with Natural and Forced Convection Solar Stills," Water and Sewage Work, 108, 378-383 (1961) . 24. Dunkle, R.V. "Solar Water Distillation: The Roof Type Still and a Multiple Effect Diffusion Still," International Developments in Heat Transfer, ASME, 895-902 (1961). 25. Salam, E. and Daniels, F. "Solar Distillation of Salt Water in Plastic Tubes Using a Flowing Air Stream," Solar Energy, 3^^1) , 19-22 (1959) . 26. Garett, C.R. "A Method of Producing Frcsli Water by Means of Solar Energy," Master's thesis, College of Engineering, University of Florida (1960) . 27. Block, M.R., Farkes, L. and Spiegler, D.S. "Solar Evaporation of Salt Brines," Industrial and Engineering Chemistry, 43(7), 15441553 (1951) .

    PAGE 196

    184 28. Davis, J.S, Personal communication. 29. Davis, J.S. "Biological Communities of a Nutrient Enriched Saline," Aquatic Botany, h_, 23-42 (1978). 30. Lincholn, E. Personal communication. 31. Keyes, C.G., Gregory, W.S., Gunaji, N.N., Lunsford, J.V., Wong, C, Savage, W.F. and Rinne, W.W. Disposal of Brine by Solar Evapora tion: Design Criteria , Research and Development Progress No. 564, United States Department of Interior (1970) . 32. Threlkeld, J.L. and Jordan, R.D. "Direct Solar Radiation Available on Clear Days," ASHRAE Transactions, 64, 45-53 (1958). 33. Siegel, R. and Kowell, J.R. Thermal Radiation Heat Transfer , McGraw-Hill Book Company, New York (1972). 34. Eckert, E.R.G. and Drake, R.M. Analysis of Heat and Mass Transfer , McGraw-Hill Book Company, New York (1972) . 35. Baum, V.A. and Bairamor, R. "Heat and Mass Transfer Processes in Solar Stills of Hotbox Type," Solar Energy, 8(3), 78-82 (1964). 36. Edwards, D.K., Denny, V.E. and Mills, A.F. Transfer Processes , McGraw-Hill Book Company, New York (1976) . 37. Farber, E.A. Personal communication. 38. Carnaham, B., Luther, H.A. and Wilkes, J.O. App lied Numerical Methods , John Wiley & Sons, Inc., New York (1969). 39. Orna, M.V. "Chemical Origins of Color," Journal of Chemical Education, 55(8), 478-484 (1978). 40. Venkataraman, K. Chemistry of Synthetic Dyes , Vol. II, Academic Press, Inc., New York (1952). 41. Venkataraman, K. Chemistry of Synthetic Dyes , Vol. I, Academic Press, Inc., New York (1952). 42. Duffie, J. A. and Beckraan, W.A. Solar Energy Thermal Processes , John Wiley & Sons, New York (1974). 43. Baumeister, T. Standard Handbook for Mechanical Engineers , Seventh Edition, McGraw-Hill Book Company, New York (1967). 44. Keenan, J.H. and Keyes, F.G. Thermodynamic Properties of Steam , John Wiley & Sons, New York (1936). 45. ASHRAE Handbook of Fundamentals , American Society of Heating, Refrigerating and Air Conditioning Engineers, Inc., New York (1972) .

    PAGE 197

    185 46. Hottel, H.C. and Woertz, B.B. "Performance of Flat-Plate SolarHeat Exchangers," Trans. ASME, 64_(2) , p. 91 (1942). 47. Touloukian, Y.S. (Ed.) Thermophysical Properties of Matter , Vol. 2, IFA/Plenum, New York (1976). 48. Arpaci, V.C. Conduction Heat Transfer , Addison-Wesley Publishing Company, Reading, Massachusetts (1966).

    PAGE 198

    BIOGRAPHICAL SKETCH Anil Kumar Rajvanshi was born in Lucknow, India, on September 1, 1950. After passing his Indian School Certificate exam in 1966, he entered a five-year bachelor's program in mechanical engineering at the Indian Institute of Technology, Kampur. He received his bachelor's degree in mechanical engineering in 1972 and subsequently his master's degree in mechanical engineering from the same institute in 1974. In the same year he was awarded the Government of India National Scholarship for Training Abroad, which made it possible for him to pursue the doctoral work at the University of Florida. 186

    PAGE 199

    I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Erich A. Farber, Chairman Professor of Mechanical Engineering I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Chung K. Hsieh, Cochairman Associate Professor of Mechanical Engineering I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Calvin C. Oliver Professor of Mechanical Engineering I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Cl^ L (. / Richard K. Irey .1 Professor of Mechanical Engineering

    PAGE 200

    I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Arun K. Varma Professor of Mathematics This dissertation was submitted to the Graduate Faculty of the College of Engineering and to the Graduate Council, and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. June 1979 Dean, Graduate School