Experimental Thermal and Fluid Science 26 (2002) 131137 www.elsevier.

com/locate/etfs

Experimental measurement of heat loss in an advanced solar collector

Abstract A novel design for an advanced solar water heater has been proposed. The design incorporates a double-sided at plate absorber mounted on stationary concentrators. An experimental program is being undertaken to investigate the heat loss characteristics of the proposed design. The experimental rig is a full-scale model, located in a temperature-controlled room. Electric heaters have been used to simulate the absorber plate and thermocouples are used to measure the surface temperature on the heater plates as well as the ambient temperature. Low iron, anti-reective glass, of the type used in a standard at plate collector is used for the glass cover and the two concentrating reectors have been modelled using aluminised reective sheet. Heat loss measurements have been carried out over a range of heater temperatures from 30 to 100 C. The overall heat transfer co-ecients obtained from the model are 30 70% lower than from conventional at plate designs. The preliminary results indicate that considerable reductions in heat loss may be obtained with this design compared to conventional designs. 2002 Elsevier Science Inc. All rights reserved.

1. Introduction Flat plate collectors of the type used in most modern domestic solar water heaters in Australia and many other countries have high heat loss and poor performance during winter months when demand for hot water is generally greatest. A novel design for an advanced solar water heater has been proposed that addresses these issues [1]. The proposed design incorporates a double-sided at plat absorber mounted on stationary concentrators as shown in Fig. 1. The absorber plate is a standard 2 m 1 m ooded absorber as used in conventional systems but rotated 90 and oriented along the roof slope rather than up the slope. The shape of the concentrators gives the absorber plate an inclination of 10 to the roof slope resulting in an overall inclination of approximately 30 on a standard pitch roof. This increase gives a strong seasonal bias to the collector during the winter months, when demand for hot water is high and solar contribution is low. It is anticipated that this design will have a very high solar contribution. To assess the performance of the design a comprehensive experimental program has been initiated conCorresponding author. Tel.: +61-2-9385-4253/5127; fax: +61-29663-1222. E-mail address: m.behnia@unsw.edu.au (M. Behnia).
*

sisting of indoor and outdoor tests. The program will investigate heat loss characteristics, optical and thermal performance and benchmark the proposed design against current technology at plate collectors. The data obtained from the experiments will be used to develop design optimisation tools that can be used in the design and manufacture of an advanced domestic solar water heater. These tools will also enable new products to be developed more cheaply by optimisation of design parameters prior to expensive prototypes being built. 2. Experimental procedure The work presented here reports the ndings from an indoors experimental program currently being undertaken to investigate the heat loss characteristics of the proposed design. Fig. 2 shows a schematic diagram of the experimental set-up. The experimental rig is a fullscale model, located in a small temperature controlled room. 2.1. Construction The collector is constructed from timber ply and uses low iron, anti-reective glass, of the type used in a standard at plate collector, for the glass cover. Three 1 kW element type heaters, sandwiched between two

0894-1777/02/$ - see front matter 2002 Elsevier Science Inc. All rights reserved. PII: S 0 8 9 4 - 1 7 7 7 ( 0 2 ) 0 0 1 1 9 - X

132

N.K. Groenhout et al. / Experimental Thermal and Fluid Science 26 (2002) 131137

Nomenclature A At g hg k L Nuexp Nud Pr P Ra Red exposed surface area of individual heater, m2 combined total exposed surface area of the heater plate, m2 acceleration due to gravity (taken to be 9.797 m/s2 in Sydney, Australia) convection heat transfer co-ecient of glass cover, W/m2 K thermal conductivity, W/mK characteristic length, m experimental Nusselt number average Nusselt number in ducted section Prandtl Number time weighted average power of individual heater, W Rayleigh number average Reynolds number based on hydrodynamic diameter of ducted section T T amb T ave Tw Tb Uo a b dT DT lb lw m time weighted average temperature of individual heaters, C time weighted average ambient temperature, C time weighted average of combined heaters, C wall temperature of concentrators, C bulk temperature, C overall heat transfer co-ecient, W/m2 K thermal diusivity, m2 /s temperature co-ecient of expansion, 1/K thermal boundary layer thickness (T ave T amb ), C dynamic viscosity of air evaluated at bulk temperature, kg/ms dynamic viscosity of air evaluated at wall temperature, kg/ms kinematic viscosity, m2 /s

Fig. 1. Schematic of the proposed design showing stationary concentrators and double sided absorber.

sheets of 2 mm thick aluminium plate, are used to model the absorber plate. The aluminium sheet provides a more even temperature prole across the heater plates, which suer from small amounts of localised heating due to their construction. The heaters are connected to variable voltage power supplies (Variacs) through Algodue UPM30 digital power meters that measure the power input to the heaters. The Variacs allow the power to heaters to be adjusted by varying the supply current. The power meters are connected in turn to a desktop computer via RS232 connection. The two concentrating reectors have been modelled using lightweight Laminex sheet covered with aluminised reective foil. The underside and walls of the test rig have been insulated using wool bre insulation with a

rating of R2.0. All openings are sealed with tape to reduce unwanted heat loss and airow.

2.2. Thermocouples Thirty T-type thermocouples are used to measure: (a) the surface temperature on the heater plates, (b) the surface temperatures on the walls of the concentrators, (c) the temperature gradient across the glass cover, and (d) the ambient temperature. Fig. 3 shows the location of the thermocouples within the collector cavity. The thermocouples are connected to a Yokogawa HR2500E 30 channel Hybrid Recorder and this in turn is connected via RS232 serial connection to the computer.

N.K. Groenhout et al. / Experimental Thermal and Fluid Science 26 (2002) 131137

133

Fig. 2. Schematic showing the experimental set up.

A simple data collection and control program is used to maintain the ambient temperature in the test room at 20 C and the heaters at a user dened temperature between ambient and 110 C. The program cycles over a period of approximately 6 s, and is used to store all 30 temperature channels and the three power meter readings. The heater temperatures and power readings are then averaged over a 30 s period and the values written to a data le. Solid state relays are used to turn the heaters and the chiller unit control valve on and o via a 5 VDC signal from the computer. The control program can maintain the heater temperatures with 0.5 C of the required temperature and the ambient at 20 1 C. 2.4. Wind simulation To enable a consistent measurement of the heat transfer coecient of the glass cover and to allow simulation of wind across the top cover, a ducted airow was installed over the inclined glass surface. The duct was constructed to provide a constant velocity of 1.5 ms1 . It consists of a bell mouth entry with a honeycomb insert acting as straightening vanes. The honeycomb has a square cross-section of 5 mm and is approximately 100 mm long. The duct has a small axial fan with a 1/4 hp motor on the outlet. The external heat transfer co-ecient for the glass was determined from Eq. (1) [3];  n f =8Red Pr lb Nud 1 1=2 2=3 1 lw 1:07 12:7f =8 Pr where f 1:82 log10 Red 1:64 n 0:11 for Tw > Tb
2

Fig. 3. Location of thermocouples within the collector cavity. All thermocouples are located within a plane through the mid section of the collector.

The thermocouples were calibrated and found to be accurate to 0.3 C. T type thermocouples are the most suitable for this application [2]. 2.3. Temperature control A condenser unit is installed inside the test room and connected to a small chiller unit located just outside. This is used to control the ambient temperature within the test room. The condenser unit has a small solenoid valve on the cold-water inlet side, which is controlled by a 5 VDC signal from the computer.

134

N.K. Groenhout et al. / Experimental Thermal and Fluid Science 26 (2002) 131137

This gives an average hg of 5.94 W/m2 K for a temperature range of 30100 C with an average ambient temperature of 20 C. 3. Results and discussion A number of experiments have been performed to determine the heat loss characteristics of the proposed design. These have been carried out over temperatures ranging from 30 to 100 C. Typically, 35 h of steady state measurements are recorded and analysed. The data le obtained from the experimental run includes: (a) time since start up in seconds, (b) three power values averaged over the period, (c) average ambient temperature over the period, (d) the average temperature of each heater over the period and (e) the instantaneous temperature on all channels at the end of each period. Each period is approximately 30 s. Fig. 4 shows a typical plot of temperature over time for the three heaters and the ambient temperature. The heat lost is equal to the heat input to the collector at steady state and hence the sum of the losses through the glass cover and the insulated walls is equal to the power input. The power and temperature data were averaged over thirty minute intervals and the timeweighted heat transfer co-ecient, U i , was then calculated by Eq. (3) for each heater: P U A T T amb W=m2 K 3

The overall measured heat transfer co-ecients obtained from the experiments were plotted as a function of the temperature dierence between the average plate temperature and the ambient temperature, and the equation for the line of best t was obtained. Fig. 5 shows the variation of the overall heat transfer co-ecient as a function of DT . The dimensional equation describing this data is given by Eq. (5): 2 U o 1:96 104 T ave T amb 0:0406 T ave T amb 5 2:00 W=m2 K The overall heat transfer co-ecient varies between 2.37 and 4.03 W/m2 K. Flat plate collectors of the type used in most domestic applications typically have overall heat transfer co-ecients in the range 5.5 W/ m2 K for the best selective surfaces to 1011 W/m2 K for plain black absorber surfaces. This signicant reduction in heat loss, particularly at lower temperatures is attributable to a number of factors including the coupling between the radiation and convection within the collector. Convective losses from the absorber surface in typical at collectors generally account for around 50% of the heat loss. Conductive losses, mainly through the insulation at the rear of the collector, account for around 5 10%, whilst the remaining losses are through radiation [4]. The convective and radiative heat loss from the front of the collector has been reduced in the proposed design through the use of the double-sided absorber plate. The reduction in the footprint of absorber plate is 50% with only a 15% reduction in eective absorber area. Conduction losses through the insulation at the rear of the absorber plate are reduced signicantly as the directly insulated area of the absorber is reduced from 4 to 0.61

The average overall heat transfer co-ecient, U o , is given by Eq. (4). P3 U i Ai U o i1 W=m2 K 4 At

Fig. 4. Plot of typical temperature versus time data from experimental data.

N.K. Groenhout et al. / Experimental Thermal and Fluid Science 26 (2002) 131137

135

Fig. 5. Plot of overall heat transfer co-ecient as a function of DT (Tave Tamb ).

m2 . The use of a single absorber also makes a signicant cost saving in manufacturing. The application of selective surfaces to the absorber plate makes the most signicant contribution to any reduction in radiative loss. High absorptance low emittance surfaces of the type proposed for this design have emissivities in the order 0.15 with absorptance as high as 0.95. The interaction between the walls and the absorber plate is complex due in part to the coupling and the geometry of the design. The complex geometry of the collector presents an added diculty in the analysis, as the length scale for the determination of non-dimensional quantities is dicult to dene. For the purposes of this study the characteristic length has been dened as the length of the absorber plate up the slope and is equal to one metre. All uid properties were evaluated at the lm temperature, which is the average of the ambient temperature and the average heater temperature. Fig. 6 shows average Nusselt number as a function of Rayleigh number. Scaling analysis has shown that a relationship of the form Nu cRa
n

duction, is in the range 90140, again indicating that convection is a dominant mode of heat transfer. Nusselt number is calculated using Eq. (8). Nuexp Uoverall L k 8

The data in Fig. 6 is represented by Eq. (9) Nuexp 0:240Ra

where n 0:2 for constant heat ux surfaces, 0.25 for isothermal surfaces, and c constant of proportionality, exists for uids with constant Prandtl number less than one [5]. The Rayleigh number, which is a measure of the strength of the natural convective forces in the cavities, is in the range of 8:8 108 4:6 109 and is calculated using Eq. (7). gbTave Tamb L3 7 am These values suggest the natural convection plays a dominant role in the heat loss and that the ow inside the collector cavity is turbulent. The Nusselt number, which denes the ratio between convection and conRa

To gain a physical understanding of the reduction in the heat loss that this design oers it is important to consider the modes of heat transfer present. Energy is lost from the absorber plate through radiation exchange with the walls and the glass surface and through convection within the cavities and between the plate and the glass cover. The aluminised sheeting on the walls has high reectivity, low absorptance and low emissivity. Therefore radiation exchange with the reectors is low, reducing the overall heat transfer. Conduction through the walls is reduced by two factors, rstly the installation of insulation around the outside of the rig and secondly the low radiation exchange with the heater plates resulting in cooler wall temperatures. Radiation exchange between the external surfaces and the ambient environment is considered negligible. Cooler walls on the reectors and the glass cover induce a strong convective ow from the heater plate inside the larger upper concentrator cavity. This relatively fast moving plume of hot air is clearly visible in the streamlines shown in preliminary numerical results of the collector. Fig. 7 shows a typical streamline plot for an absorber plate temperature of 50 C. This particular ow pattern is a large source of heat loss from the absorber plate and the continuation of this work involves the investigation of optically thin convection suppression

136

N.K. Groenhout et al. / Experimental Thermal and Fluid Science 26 (2002) 131137

Fig. 6. Average Nusselt number as a function of Rayleigh number.

Fig. 7. Plot of streamlines for an average heater temperature of 50 C obtained using Fluent v5.

barriers that may oer further reductions in the overall heat loss. The theoretical result also shows the presence of multicellular ow in the larger cavity and the presence of RayleighBenard cells between the heater plates and the glass. This is consistent with the type of ow observed in high aspect ratio cavities with heating on one wall [6,7]. The experimental value of n 0:285 in Eq. (7) suggests that an isothermal condition is present on the plate surface. The relationship in Eq. (6), however, assumes dT , the thermal boundary layer thickness, is asymptotic. In reality, the presence of the multicellular ow over the absorber plate means the thermal boundary layer is not fully developed and the ow between the cavities becomes choked. The meshing between the counter-rotating cells in this region may account for the higher than expected co-ecient on the Rayleigh term in Eq. (7).

Due to the inclination of the plate and the intersection of the plate with the concentrating reector, the rising hot air in the lower cavity becomes trapped under the heater plate thermally stratifying that space. Conrmation of the ow regime will be possible once ow visualisation has been undertaken.

4. Conclusions A novel design for a collector for domestic solar water heating has been presented. Experiments have been performed to investigate the heat loss characteristics of the design and these have shown that the collector can oer signicant reductions in the overall system heat loss. The use of a double-sided absorber

N.K. Groenhout et al. / Experimental Thermal and Fluid Science 26 (2002) 131137

137

with low emissivity selective surface coupled with high reectance stationary concentrators reduces the radiative and conductive losses through the back of the collector. Top losses are reduced through a reduction in the footprint of the absorber plate; however, the main source of heat loss in this design is through the glass cover. Whilst this particular design may increase the ratio of convective losses to radiation and conduction losses, but the net heat loss appears to be 3070% lower then conventional systems. Further work is required to conrm the ow regimes that appear in the numerical modelling of the collector. This will be achieved using ow visualisation techniques. Suppression of the strong convective plume in the larger cavity of the collector may be possible with the insertion of optically thin barriers, trapping the warm air under the absorber plate and further reducing the heat loss.

References
[1] N.K. Groenhout, G.L. Morrison, M. Behnia, Design of advanced solar water heaters, Proc. ANZSES Solar 2000 Conf., 2000, pp. 295303. [2] J.W. Dally et al., in: Instrumentation for Engineering Measurements, second ed., John Wiley & Sons, New York, 1993, p. 435. [3] J.P. Holman, in: Heat Transfer, eighth ed., McGraw-Hill, New York, 1997, p. 289. [4] M. Madhusudan, G.N. Tiwari, D.S. Hrishikeshan, H.K. Sehgal, Optimization of heat losses in normal and reverse at-plate collector congurations: analysis and performance, Energy Conversion Manag. 21 (1981) 191198. [5] A. Bejan, in: Convect. Heat Transfer, Wiley, New York, 1993, pp. 161179. [6] J.G. Symons, M.K. Peck, Natural convection heat transfer through inclined longitudinal slots, J. Heat Transfer 106 (1984) 824829. [7] M. Behnia, J.A. Reizes, G. de Vahl Davis, Natural convection in a rectangular slot with convective radiative boundaries, National Heat Transfer Conf., Denver, CO, 1985.