ENERGY RESEARCH, VOL.

10, 29-38 (1986)

SALT DIFFUSION IN A TEMPERATURE FIELD: APPLICATION

TO SALINITY PROFILES IN SOLAR PONDS
YOJANA KATTI AND P. K. BANSAL
Centre of Energy Studies. Indian Institute of Technology, New Delhi. 110016, India

SUMMARY
In the present investigation, the process of diffusion of salt in a vertical column of liquid, subjected to temperature
variations of the types T ( x )= constant, linear ( = a + bx) and parabolic ( = a + bx - cx2); with a constant concentration
dinerence between the top and the bottom (0and 25 per cent, respectively)is studied. It is seen that a linear temperature
gradient, T ( x )= a + bx, leads to a near convex parabolic salt concentration profile with maximum deviations increasing
from 13.5 per cent (at 40°C) to 14.8 per cent (at 70°C) and eventually to 15.7 per cent (at WOC) with respect to the linear
concentration value of 12.5 percent (by weight)at the midpoint. Conversely, the parabolic temperature profile as well as the
modified profile due to the Soret effect leads to near cubic salt profiles which differ only by 2-3 per cent in the upper half of
the pond. However, they show a point of inflexion at larger depths near the bottom around which the convex profiles
change over and become concave. Subsequently, these studies have been extended to compute the salinity profiles of
thermal configurations of the operational solar pond.

KEY WORDS Solar energy Solar ponds Salinity profiles

INTRODUCTION
Artificial solar ponds are stabilized against convective breakdown by the provision of adequate salt gradients
to compensate for buoyancy effects at the pond bottom, which have to be maintained at high temperatures for
the use of solar energy with optimum efficiency. Since the emergence of solar ponds as potential economic
sources, comprehensive efforts-basic and applied research-on the diffusion of salt in fluids, thermo-
dynamics and hydrodynamics of salt solutions for applications to solar pond mechanics, have been made by
several workers.’ - ’
In general, the salt concentration gradients are maintained by keeping the pond surface at zero
concentration by continuously flushing it with fresh water and the bottom at saturation level by replenishing it
with saturated brine. Under operatingconditions, salt migrates continuously from the bottom towards the top,
by diffusion across the large concentration difference. The nature and magnitude of salt migration across a
constant concentration difference and the establishment of salt concentration and salinity gradient profiles
along the depth of the pond have been investigated by Stern,3Tabor: Chepurniy and Savage*and Akbarzadeh
and Ahmadi.” These investigations provide design data for
(i) choice of the salt
(ii) selection of suitable procedures for the rapid establishment of the necessary gradients
(iii) follow up stability conditions in the pond fluid, and identifying the development of large deviations
leading to convective breakdown
(iv) application of correction in time
to ensure continuous proper operation of the pond.
In the three zone solar pond, the non-convective zone (NCZ), which acts as an insulating cover for the stored
thermal energy at the bottom, is sandwiched between the upper convective zone (UCZ), which is nearly at the
ambient temperature, and the lower convective zone (LCZ) which is held at as high a temperature as is possible
(nearly 95°C). The temperature distribution, T (x), along the depth of the NCZ is in general of a parabolic

0363-907X/86/010029-10$01 .OO Received 25 June I984

0 1986 by John Wiley & Sons, Ltd.
30 YOJANA KATTI A N D P. K. BANSAL

nature (T(x) = a + bx - cx2)in thermal configurations for energy extraction at constant bottom temperature.
Therefore, salt diffusion across a constant concentration difference in a temperature field (along the NCZ)
leads to salinity profiles which are functions of the pond depth.
In the present communication, the relations for salt distribution along the pond depth have been derived
by solving the steady state diffusion equation with diffusion coefficient varying linearly with temperature,
for three types of temperature fields along the depth, namely constant, linear (T (x) = a + bx) and parabolic
(T(x) = a + bx - cx’). The Soret effect arising from the salt diffusion in the temperature gradient has also been
investigated by introducing an additional term (corresponding to the Soret effect) depending upon the local
temperature gradient in the diffusion equation. The salinity profiles have been computed for the most realistic
temperature profiles, which are incidentally parabolic.

RELATIONS FOR CONCENTRATION PROFILES IN SOLAR PONDS

Figure 1 shows the nonconvective zone of a typical thermal configuration of a solar pond, established for heat

- -
extraction at constant bottom temperature. The top and the bottom are maintained at constant concentrations
C , ( 0) and C , ( 25 per cent), respectively. During the operation, the fluid is subjected to a temperature
gradient along the depth. These temperature gradients and temperature profiles can be evaluated from the
thermal analysis of the operational configurations, following Sodha et al.’ and Katti and Bansal,I4 whereas
the salinity profiles C(x) can be evaluated by solving the diffusion equation for the three different cases (from
the procedure described below) with the boundary conditions:
at x = O , C=C,
atx=1, C=C,

Constant temperature over the entire NCZ

The salt diffusion equation can be written in its simplest form as
d2C
K -=O
dx2
The solution of this may be written as

which represents that the salinity profiles are linear ( = ax + b).

Temperature varying along the NCZ

The diffusion equation may be written in the form

a.[K.S]=o
a
(4)

where K, (T) is a function of temperature. Experimental studies on the variation of K, with temperature for
sodium chloride and magnesium chloride have shown that in the range of temperatures involved in solar pond
operations, K, can be taken to be linearly dependent on temperature as
K , = Ko(l + a T )
and hence, the diffusion equation (4) takes the form

A
ax [ K o ( l + a T )d xE ] = 0
 +-
SALT DIFFUSION IN A TEMPERATURE FIELD 31

Figure 1. Schematic diagram of a three zone salt gradient solar pond

Case I . For T ( x ) to be linear, i.e. of the form a + bx, the molecular diffusion coefficient is given by

where K O = 1.39 x m2/s and 1-12 x lo-’ m2/s is the molecular diffusivity of NaCl and MgCI2
respectively at 20°C and the values of a are 0 0 2 9 and 0027 for NaCl and MgClz solar ponds respectively.
Therefore, the diffusion equation ( 5 ) in conjunction with the diffusion coefficient as expressed by equation
(6) can now be written in its general form

”[
ax Ko(l +aa-2Oa+abx)- ax
ac 1 =0 (7)
32 YOJANA KATTI AND P. K. BANSAL

leading to the solution

+ aa - 20a + abx) + C , ln(1 + aa - 20a +abl) -CbIn(1 + aa - 2Oa)
(Cb- C,)ln(l
C(X) =
[ ln(1 + aa - 2001 + abl) - ln(1 +aa - 2Oa)
which can be expressed in the form of a parabolic profile, (A, + B,x - C,xz), within an accuracy of 5 per cent.
Case 11. For a parabolic temperature profile,T (x) = a + bx - cx2,the solution of the diffusion equation ( 5 )
may further be written as

where a , and a2 are the roots of the quadratic equation

(1 +aa-20a)+abx-acx2 =0 (94
Equation (9)can also be expressed in the form of a cubic polynomial, A , + B,x - C,x2 + D,x3, to represent the
above profile within the accuracy of 5 per cent.

Efect of temperature on solubility

Besides the dependence of diffusion coefficient on temperature, one has to take into consideration the
‘Sorer effect’, wherein the salt migration takes place in a medium of homogeneous concentration across a
temperature gradient. In fluid with large temperature differences and gradients, this effect can be quite
appreciable.
For inclusion of this effect, we use the modified diffusion equation, suggested by Rothmeyer” as

” K O { 1 +a(T-20))
ax ax

For a parabolic temperature profile, [T (x) = a + bx - cx’], the resulting diffusion equation can be written
in the form

leading to a solution of the form

Cexp{ jg(x)dx} = jf(x)exp{ jg(x)dx)dx+dz
where

In the range of quantities involved in solar pond operations, exp { jg(x) dx} can be expanded in the form of a
series and the higher order terms [ jg(x) dxI2 onwards be neglected. In this approximation, the ultimate
solution can be written as
C { 1 + j g w dx} = If(.) dx + j[f(x) { j g w d x j l dx + 4 (14)
where the integration constants dl and d2 are determined from the boundary conditions at the surface (C, = 0
at x = 0) and the bottom (C, = C, at x = l).

dT
- has been obtained from Katti and Bansal” and taken in the form of a xcond order polynomial within the required accuracy.
dx
 SALT DIFFUSION IN A TEMPERATURE FIELD 33

: Constant

. Linear

X (m)
Figure 2. Variation of salinity profiles, C(x), with the depth of the non-convective zone (x) for two temperature profiles, namely constant
and linear; keeping the bottom temperatures at (b)W C , (c) 70°C and (d) 70°C
34 YOJANA KATTI A N D P. K. BANSAL

25

2c

15

-
5
r
n

-3
V

10

C
0.5 1.0 1.5 2.0 2.5
X(m)
Figure 3. Plot of salinity profiles,C (x), with x for different temperature profiles:(a) constant, (b) linear, (c)parabolicand (d) parabolic with
Soret effect for constant bottom temperature of 90°C
 SALT DIFFUSION IN A TEMPERATURE FIELD 35

RESULTS AND DISCUSSION

The relations (3), (9) and (14) have been used to illustrate the nature and the magnitude of salt profiles in the
NCZ of the thermal configurations of a solar pond with the following set of parameters:
1 = 2-5 m, C , = 0 at x = 0, Ambient temperature = 25°C
C , = C, at x = I, Bottom temperature = 40,70 and 90°C
The numerical computation corresponding to the case of constant temperature over the entire NCZ, has
indicated that the salt distribution is linear with pond depth x between the extreme values of salinity ofOand 25
per cent (by weight) at the top and bottom, respectively,irrespective of the numerical value of the temperature.
This is shown in curve (a) of Figures 2 and 3.
The variation of salinity profiles for the case of temperature varying linearly with x between the extreme
values of 25°C at the top and bottom temperatures ( = W C , 70°C and WC), is shown in Figure 2. It is
interesting to note that the profiles are nearly convex parabolae, symmetric about the mid-point, and the
maximum deviation from the linear distribution (curve (a)) at constant temperature increases with bottom
temperature.ThevaluesofsalinityC(x)atthemid-pint increase from 13-5(percent by wt.)at 40°C (curve (b)),
to 14.8 at 70°C (curve (c)), to 15.7 at 90°C (curve(d))with respect to the linear distribution value of 12.5 (per
cent by weight, curve (a)).As is shown in the Appendix, the salinity profiles are independent of the thickness of
the NCZ.
For a typical temperature profile T ( x )= 25.0 + 0 . 6 1 2 ~- 1.248 x lo-’ xz,across the NCZ in a typical solar
pond configuration for heat extraction at a constant bottom temperature of 90°C with no reflectivity, the
nature of the salinity profile (Figure 3(c)) is nearly a cubic polynomial, coinciding with the linear profile
(curve (b)) in the upper half of the pond. A comparison of the corresponding values of salinity, C, (in
Figures 3(a), (b)and (c))at different depths x reveals that the major deviations are from curve (a) to curve (b),
i.e. for the constant temperature gradient approximation. The salt distribution described by curve (c)is about 2
per cent more than that in Figure 3(b) in the upper half of the pond fluid. However, in the lower half, the

I 1

1 I 1 1

0.7 1.2 1.7 2.2 2.7

X = [11+12(rnjJ

Figure 4. Variation with pond depth, x, of temperature at different bottom temperatures 40, 70 and 90°C and reflectivities,
-R = 00, ----R = 0 4 and . . . . . . R = 0 7
36 YOJANA KAlTI AND P. K. BANSAL

21 I

Soret effect included, R = 0.0

_....... .. ...... s r e t effect included, R: 0.7

2[

15

-
5
r
n

-s
V

1c

2
I 0.7 1.2 1.7 2.2 2

X (rn)

Figure 5. Variation of salt concentration profile with pond depth at different bottom temperatures (40and 7O"C)andreflectivities(0.0and
0 7 ) for the following temperature profiles; -T(constant),---- Soret effect included ( R = 0 ) a n d . . . . . . Soret effect included
(R = 0.7)

positive deviation decreases more rapidly (Figure 3(c)) forming a point of inflexion around 2.3 m, beyond
which the profile takes a slightly concave form.
Figure 3(d)is plotted, using the relation (14) for the salinity profile, which takes into consideration the Soret
effect. The upper half of this curve is very near to Figure 3(c) and it differs on the positive side by about 2 per
cent. In the lower half, however, one gets an appreciable difference, with the point of inflexion moving away
 SALT DIFFUSION IN A TEMPERATURE FIELD 37

from the bottom to around 2.2 m. This shows that the salt concentration gradient becomes very low, making
the pond highly susceptible to perturbations which will eventually render the pond unstable.

Application to heat and mass transfer analysis of solar ponds.

Following the thermal analyses by Sodha et a1.13 and Katti and Bansal,14 the variation of typical
temperature profiles with the depth of the NCZ is illustrated in Figure 4 for different bottom temperatures
( = 40,70 and W C ) and reflectivities ( = 0.0, 0.4 and 0.7). The corresponding behaviour of salinity profiles
(using relation (14)) established by diffusion, has been studied and is exhibited in Figure 5. In this study, the
concentrations of salt at the surface and the bottom are assumed to be 2 and 25 per cent (by weight),
respectively, for the most realistic case of operational solar ponds. From the curves, it can be seen that for a
non-reflecting pond bottom (R = O.O), the inversion position for bottom temperature W C is at 2.5 m; it shifts
away from the base to 2.2 m for 70°C and still further to 1.6 m for W C . The effect of increasing reflectivity is
also to shift the inversion point away from the base. For W C , the shift at R = 0.7 is from 2.5 m to 2.3 m, for
70°C from 2.2 m to about nearly 2.1 m and for W C from 1.6 m to 1.0 m.

CONCLUSIONS
For solar ponds operating with specified energy extraction under available insolation conditions, the heat and
mass transfer analysis leads to the assessment of thermal performance and stability. I t may be concluded that
around the point of inflexion, the salt gradient continues to have relatively low values over large thicknesses,
rendering these layers prone to instability.
ACKNOWLEDGEMENTS

The authors wish to thank Prof. M. S. Sodha (Energy Specialist, Ministry of Commerce and Industry. Nicosia,
Cyprus) for his invaluable guidance and constant encouragement during the course of this investigation.

NOMENCLATURE
C(x) = Salinity distribution in the pond system at different levels (per cent by weight).
c b = Salinity concentration at the bottom of the pond (per cent by weight).
C, = Maximum value of cb (per cent by weight)
C, = Salt concentration at the top of the pond (per cent by weight).
K, = Coefficient of salt diffusivity (m2/s).
KO = Molecular diffusivity of salt (m2/s).
1 = Depth of the NCZ (m).
S, = Soret coefficient (m/sLC).
T ( x ) = Temperature distribution in the pond system at different levels ('C).
x = Position co-ordinate (vertically downwards, m).
a = Coefficient of linear expansion of salt (-C)-'.

APPENDIX
From Figure 6

Figure 6
38 YOJANA KATTI AND P. K. BANSAL

i.e.

Atx=O
C ( x )= a
1
Atx=-
2
bl c12
C(x)=u+---
2 4
Atx=l
+
C ( X )= a bl -c12
Solving (1) and (2) simultaneously for b and c we obtain
Kt -cb -30
b=
1
2(2ct -cb - a)
C =
l2

where X ( = x / l )is the dimensionless depth and hence in the above expression C ( x )does not depend explicitly
on 1.

REFERENCES

1. Veronis, G. (1968). ‘Effect of a stabilizing gradient of solute on thermal convection’, Journal o/Fluid Mechanics, 34, 315.
2. Stiles, W. (1923). ‘Indicator method for the determination of coefficients of diffusion’, Proc. Roy. Soc., 103, 260-275.
3. Stem, M. E. (1960). The salt-fountain and thermohaline convection’, Tellus, 12, 172.
4. Tabor, H. (1963). ‘Large area solar collectors for power production’, Solar Energy, 7, 189.
5. Nielsen, C. E. and Rabl, A. (1975). ‘Operation of a small salt gradient solar pond’, ISES Meeting. Los Angela, California.
6. Bames, P. G. and Gill, A. E. (1969).‘On thermohaline convection with linear gradients’, Journal ojFluid Mechanics, 37. 289.
7. Kaufmann, D. W. (1960). Sodium Chloride, American Chemical Society, Washington, Chapter 25.
8. Chepurniy. V. and Savage, S. B. (1974). ‘An analytical and experimental investigation of a laboratory solar pond model’, ASME
Publication 74- WAISOL-3.
9. Weinbergr, H. (1964).‘The physics of the solar pond, Solor Energy, 8,45.
10. Zangtando, F. (1979).‘Observations and analysis of a full scale experimental salt gradient solar pond’, Ph.D. n e s i s , Universityof New
Mexico.
1 1. Hull, J. R. (1978).‘Effectsof radiation absorption on convective instability in salt gradient solar ponds’, Meeting ofthe American Section
of the International Energy Society, Denver, Colorado, 1978.
12. Akbarzadeh, A. and Ahmadi, G. (1981). ‘On the development of the salt concentration profile in a solar pond, Energy, 6,36%382.
13. Sodha, M. S., Kaushik, N. D. and Rao, S. K. (1981). ‘Thermal analysis of three zone solar pond’, Int. J . Energy Research, 5,321-340.
14. Katti, Y. and Bansal, P. K. (1983).‘Heat extraction from a diffusely reflectingbottom salt gradient solar pond’, fnf. J . Solar Energy, 1,
(6), 405417.
15. Rothmeyer, M. K. (1979). ‘Saturated solar ponds-modified equations and results of a laboratory experiment’, M . Tech. Thesis,
University of New Mexico.