A SIMPLIFIED MODEL FOR ESTIMATING DISTILLATE YIELD IN LARGE SOLAR STILLS Arturo Palacio, Alejandro Rodríguez, Italia Millan, Jose L. Fernández Engineering Institute, National University Autonomous of Mexico Apartado Postal 70-472, Coyoacan, 04510, Mexico, D.F. email: vortex@vortex.iingen.unam.mx ABSTRACT A simplified one-phase model based which considers an homogeneous gas mixture with an “apparent diffusion coefficient”, has been developed to represent the flow characteristics inside a double slope solar still. Its application for the calculation of the heat flow and therefore of the distillate yield shows good agreement with experimental data. Preliminary results about the asymmetry of the flow patterns attained for symmetrical boundary conditions is presented and discussed. INTRODUCTION The use of solar energy in engineering applications has gained more importance over the last decades. Several studies, theoretic as well as experimental, have been realized on the heat transfer mechanisms that take place in triangular cavities, which is the typical shape in the X-Y plane geometry of a solar still. Convection has been proven to be the main heat transfer phenomenon in the water distillation process, as concluded by several authors. In 1979, Flack, Konopnicki and Rooke described an experimental study of the heat transfer due to convection in a laminar flow contained in a triangular cavity (isothermal walls, hot and cold, and adiabatic base). The results are similar to those shown in rectangular cavities with the exception of the apex, where conduction is proven to be the predominant heat transfer mechanism due to temperature difference and the proximity of the walls. A second study realized by Flack et al. [1980] concludes that under summer conditions (hot top, cold base) the air flow remains laminar, convection near the walls is minimum and the aspect ratio at a given Gr has very little influence over Nu. As for the winter conditions (cold top, hot base), Flack shows the flow changes from laminar to turbulent as Gr increments and most of the heat transfer through the base is realized via convection. The centreline temperature profile illustrate that the temperature is almost constant in the central area of the still but increases significantly near the base and walls. The indirect relation between the Nu and the walls inclination angle (at a given Gr) is established too. This is confirmed later, in a following experiment made by Poulikakos and Bejan [1982] in larger scale triangular spaces (increased Rayleigh Nos.) Martín del Campo et al carry out in 1988 a more extensive numerical analysis with all possible combinations of cold, hot and adiabatic walls studied for different aspect ratios and Gr numbers. His results embrace Flack's conclusions that bottom heated cavities are more convection and heat transfer effective than top heated ones and show the direct relation between Gr and the convective heat transfer as well as the influence that horizontal gradients have over convective flow, favoring it. This analysis also focuses on the influence that geometry has on the efficiency of the still. Aspect ratios of 1 to 3 are proven as the most effective. Geometry impact on water distillate becomes the principal concern that numerous authors focus on, such as Rubio et al [2000] make a comparison between double and single slope stills finding there is no significant difference in their productions when subjected to same water and cover temperature and therefore demonstrating the distillate production is independent of the cavity's geometry. Palacio et al [6] comment on the fact that in solar stills with larger aspect ratios, the dominating heat transfer mechanism is convection which is not as efficient as diffusion, more proper of shallow cavity solar stills. Despite this, more recent analytical studies done in high inclination solar stills by Palacio and Fernández [1993], suggest that tall vertex stills are viable and their geometry can be enhanced to facilitate the development of convection vortices (which will compensate for the loss in heat and mass transfer by diffusion) and improve the heat transfer process as the increased height was proven not to incur substantially in the loss of distillate yield. This theory is reinforced by the results obtained by Porta et al [1994] on shallow cavity stills although it is suggested, in opposition to what was concluded before, that convection plays an important role in shallow cavity solar stills as well. Kwantra [1996] studies the importance of the water evaporation area in a solar still. His mathematical analysis illustrates the quantitative relationship between it and the distillation yield. An enlarged evaporation area results in a more efficient evaporationcondensation process, increasing the yield. Rubio et al [2002], present a procedure to estimate the glass cover production in double slop solar stills, as a function of still temperature and area fraction, as an extension to the model proposed by Dunkle. The endeavor for a better understanding of the seawater distillation phenomenon and the search for ideal geometries and conditions that will improve the distillate yield are still an ongoing task. The intent of this paper is to report the results obtained through the development of a mathematical model capable of reproducing actual experimental data regarding distillate yield and its application on different cases that will lead the path to a continuous improvement process on this field. PROBLEM DESCRIPTION A typical double slope solar still, consists on an insulated box with channels located along the base of each condenser to collect the distillate yield; the glass covers are attached to the basin at an angle which depends on the particular design, as shown on figure 1. The evaporation area is given by the product of the length and the width of the basin, lw. The height h represents the distance between the water surface and the vertex of the glass covers. Fig. 1 Typical double slope still In this paper, the performance of a relatively small section solar still (w=1.6 m, h=0.46 m) is compared to that of a large section still (w=5 m, h=2.5 m), both constructed and operated in La Paz, B.C.S., Mexico, Rubio et al [2002]. Mathematical Formulation The set of partial differential equations describing the problem examined herein, are those of continuity, momentum and energy, together with an additional couple of turbulence transport equations. This set of equations can be expressed in a general form as follows: ( ) + ( V ) = ( ) + S t (1) where t denotes time, represents any dependent variable, V the velocity vector, the fluid density, the transport coefficient of the dependent variable, S the source of per unit volume and is the vector differential operator. The dependent variables and their corresponding transport equations, are listed below in Table 1, where u, v are the velocity components in the coordinates directions x and y respectively, T is the temperature, which is directly calculated from the energy equation, k is the turbulent kinetic energy and its dissipation rate; l is the laminar dynamic viscosity, t the turbulent viscosity, and e the effective viscosity; k and are the standard turbulence-model coefficients which will be described later. The source term for the momentum equation is given by: Sm = - p + g + e ( U ) 2 ( e U ) I 3 (2) TABLE 1. Transport coefficients for variable Continuity Momentum 1 0 u, v e ( l / l + t / T ) Temperature T Turbulent kinetic energy K ( l + t / k ) Dissipation rate of k ( l + t / ) where p is static pressure, g gravity vector, I the unit tensor, and the superscript T denotes the transpose of the dyadic. The Boussinesq approximation was considered to take into account buoyancy effects. Turbulence model The turbulence model employed is the standard k- model, where the kinetic energy k, and its dissipation rate , represent the velocity and length scale of the turbulent motion respectively. The corresponding source terms are: S k = ( Pk - + G B ) S = C 1 P k - C 2 + C 3 G B (3) k (4) where Pk is the production rate of k, and GB is the production or destruction of k due to buoyancy effects; these two terms may be written as: Pk = t U : ( U + ( U ) ) GB = - t g t (5) (6) The turbulent viscosity is calculated from the local values of k and as follows: t = C k2 / (7) The values assigned to the turbulent constants are the standard ones recommended by Launder and Spalding [1974]. Boundary Conditions Two dimensional calculations were performed neglecting the end effects of the still solar. Assuming steady state operation conditions of the still, glass cover temperatures T g1 and Tg2 as well as water surface temperature Tw were prescribed. The model takes into account the friction effect at the glass covers and water surfaces through the wallfunction approach outlined by Rodi [1993]. A reference pressure value is assigned at one cell in the middle of the domain. The solution algorithm is based upon the well-known iterative “guess-and-correct” procedure of Patankar and Spalding [1972], but modified according to the SIMPLEST algorithm of Spalding [1982]. PRESENTATION AND DISCUSSION OF RESULTS Model validation In order to assess the accuracy of the model to represent the flow of a mass of fluid inside a triangular cavity, the experimental results reported by Flack [1979, 1980] were considered. In his paper dated 1979, Flack et al. [1979] measured local and overall heat transfer data for convection air flow in triangular enclosures with two side walls which were heated and cooled and an adiabatic bottom. The experimental data generated for the local heat flux distributions is compared with the results obtained from the present model in figure 2, showing an acceptable agreement. o 6 Q2=90 y Gr2=5.60x10 1.00E+02 HOT: Flack laminar Nu1 no kebuoyancy kebuoy=0.2 1.00E+01 COLD: Flack laminar no kebuoyancy kebuoy=0.2 1.00E+00 1.00E+03 1.00E+04 1.00E+05 1.00E+06 1.00E+07 Gr1 Fig 2. Local Nusselt numbers for both hot and cold walls as a function of Grashof. The cases where the triangular enclosures are heated or cooled from the bottom are reported by Flack et al. [1980] and the typical local heat flux distributions for both cases are compared with the present model in figures 3 and 4, showing again a good agreement. 6.00E+02 HFLX 5.00E+02 Flack2 4.00E+02 q" 3.00E+02 2.00E+02 1.00E+02 0.00E+00 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Figure 3. Heat flux distribution for hot top. x/L 2.50E+02 HFLX1 HFLX2 2.00E+02 Flack2 1.50E+02 q" 1.00E+02 5.00E+01 0.00E+00 0 0.2 0.4 0.6 0.8 1 x/W Figure 4. Heat flux distribution for cold top. 1.2 1 Model Application Even though the real problem inside a solar still corresponds to that of a multiphase flow of an air-water vapour mixture, the simplicity of the one-phase model has been kept by considering an apparent diffusion coefficient for an homogeneous gas mixture, as described by Palacio et al [1993]. This means that a lower Prandtl number than the one for pure air or pure water vapour needs to be determined for the mixture, in order to obtain an adequate calculation of the heat flow as compared with the experimental measurements of the distillate. Following this approach, the Prandtl number effect on the calculation of the heat flux was investigated employing the average thermal operation conditions reported by Rubio et al [2002] for the small section solar still: T w= 80 0C, Tg1= Tg2= 65 0C, q”=366 W/m2. Given the symmetry of the data, half of the domain was considered for the calculation. Figure 5 shows that for Pr numbers smaller than 0.1, the heat flux increases very rapidly, while for larger values the variation is much slower. The required heat flux value needed to match the experimental data resulted for a Prandtl number of 0.11. 450 Heat flux (W/m2) 400 350 300 250 small 200 large 150 100 50 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Prandtl Fig. 5. Effect of Prandtl number on heat flux calculation The characteristic flow patterns, as well as the temperature distribution inside the solar still are shown in figures 6 and 7 respectively. Maximum flow velocities of the order of 0.36 m/s are present along the vertical centre line of the cavity and down the glass cover, while a single recirculation region is formed near the centre of the domain. Isotherms still show the characteristic shape of a convective dominating flow. Fig. 6 Flow patterns for the small section still Fig. 7 Temperature distribution for the small section still The same exercise was repeated for the large section solar still, with the following parameters: Tw= 60 0C, Tg1= 48 0C, Tg2= 44 0C, q”=220 W/m2. The same behaviour observed for the previous case regarding the effect of the Prandtl number on the calculated heat flux was found, and as figure 5 reveals, the measured heat flux is reproduced again when the Prandtl number approaches the value of 0.11, the difference being less than 5%. It is of interest to notice on the same figure that even though the maximum temperature difference between the water and the glass is practically the same for both cases, the heat flux predicted for the large section still is nearly half of that for the small section still at the low Pr number region. Given the above operation conditions for the big section still, the flow patterns and temperature distribution are very different from the ones obtained for the small section still, the reason being the asymmetry prescribed on the glass covers temperatures. Most of the flow on figure 8 is turning in clock-wise direction, except for a small region located at the left end, where recirculation takes place on the opposite direction due to the temperature distribution generated according to figure 9. This time the maximum velocity reaches 0.465 m/s at the right side of the still, where the largest temperature gradients takes place. For this temperature boundary condition, nearly 54% of the heat transfer takes place on the right glass cover, where the prescribed temperature is 4 0C lower than the other glass. Fig. 8 Flow patterns for the large section still Fig. 9 Temperature distribution for the large section still Employing visualisation techniques developed by Porta et al. [1998], Poujol et al. [1999] have detected that even for symmetrical temperature conditions on both glass covers, the flow patterns do not attain symmetry. In order to make a preliminary exploration of this fact, another numerical simulation was performed for the operating conditions of the small section still, but this time considering the complete domain, with a 50 x 30 grid shown in figure 10. Even though the general flow structure is very similar and the maximum fluid velocity is again about 0.35 m/s, there is a clear tendency of the flow to form two asymmetric regions, figure 11, with a stronger recirculation present on the left side where larger temperature gradients are formed according to figure 12. This result was obtained repeatedly after using several different sets of relaxation parameters. Fig 10. Computational grid employed for the complete domain Fig 11. Flow pattern for the small section still considering the whole domain Fig 12. Temperature distribution for the small section still considering the whole domain The same exercise was repeated for the large section still applying equal temperature values at the glass covers (46 0C), and the results were qualitatively the same as for the small one. CONCLUSIONS AND RECOMMENDATIONS A mathematical model was developed to evaluate the flow characteristics inside a double slope solar still and applied to reproduce the experimental results of two experimental solar stills located in La Paz, B.C.S., Mexico. The results seem to indicate that if the gas mixture inside the apparatus is treated as homogeneous, a Prandtl number value of 0.11 allows to reproduce with a very good approximation the heat flux, and therefore the distillate yield produced, for both experimental devices whose cavity sectional areas differ by a factor larger than 15. Even for symmetrical temperature boundary conditions, the numerical model shows that the temperature distribution inside the solar still, and therefore the corresponding flow patterns are not symmetrical, which confirms the preliminary observations made by the working group of La Paz, Mexico. It would be recommendable to apply the model to try to represent the transient stages of heating and cooling of the stills. REFERENCES Flack R. D., Konopnicki T.T. and Rooke J. H., "The Measurement of Natural Convective Heat Transfer in Triangular Enclosure", ASME Journal of Heat Transfer Vol. 101 No. 4, pp. 648-653, 1979. Flack R. D., "The Experimental Measurement of Natural Convection Heat Transfer in Triangular Enclosures Heated or Cooled from Below", ASME Journal of Heat Transfer Vol. 102, pp.770-772, 1980. Kwatra H., "Performance of a Solar Still: Predicted Effect of Enhanced Evaporation Area on Yield and Evaporation Temperature", Solar Energy Vol. 56 No. 3, pp. 261266, 1996. Launder, B E y D B, Spalding (1974), “The Numerical Computation of Turbulent Flow”, Comp Meth Appl Mech Eng 269-289. Martín del Campo E., Sen M. and Ramos E., "Analysis of Laminar Natural Convection in a Triangular Enclosure", Numerical Heat Transfer Vol. 13, pp. 353-372, 1988. Palacio A. and Fernández J. L., "Numerical Analysis of Greenhouse-Type Solar Stills with High Inclination", Solar Energy Vol. 50 No. 6, pp. 469-476, 1993. Porta M. A., Fernández J. L. and Chargoy N., "Influencia de la Distancia Vidrio-Agua en Destiladores Solares de Caseta", ANES, 1994. Porta, MA, Rubio, E, y Fernández, JL, “Visualization of natural convection inside shallow solar stills”, Journal Experiments in Fluids, Vol. 25, pp 369-370, 1998 Poujol, FT y Fernández, JL “Evolution of flow patterns in a space with triangular top”, Journal Experiments in Fluids. Septiembre, 1999 (review) Poulikakos D. and Bejan A., "Natural Convection Experiments in a Triangular Enclosure", Technical Notes Journal of Heat Transfer Vol. 105, pp. 652-655, 1982. Rodi W (1993), “Turbulence models an their Application in Hydraulics. A state-of-theart review”, 3rd De IAHR Monograph, Balkema, Rotterdam NL. Rubio E., Porta M. A. and Fernández J. L., "Cavity Geometry Influence on Mass Flow Rate for Single and Double Slope Solar Stills", ”, Applied Thermal Enginnering Vol. 20, pp1105-1111, 2000. Rubio E., Porta M., Fernández J.L., “Thermal Performance of the condensing covers in a triangular Solar Still”, Renewable Energy, (2002), in press