EXPERIMENTAL INVESTIGATION INTO THE BUOYANCY DRIVEN CONVECTION IN PASSIVE SOLAR HEATING FACADES

Transcription

1 EXPERIMENTAL INVESTIGATION INTO THE BUOYANCY DRIVEN CONVECTION IN PASSIVE SOLAR HEATING FACADES Dylan A. Ryan 1*, Dr. S. Burek 1 and Dr. P. Baker 2 1 School of the Built and Natural Environment Glasgow Caledonian University 2 Centre for Research in Climate and Health, Glasgow Caledonian University rdy2@gcal.ac.uk Abstract Emphasis has recently been focused on the reduction of energy use in the heating and cooling of building interiors. This has resulted in increased interest into passive solar air heating systems (such as Trombe walls) and passive ventilation techniques (such as solar chimneys). Such Solar Façade elements have now been in use for several decades and all rely on a solar-induced buoyancy-driven convection mechanism to produce the required airflows. The heat and mass transfer mechanics of both of these systems is not completely understood, which has made predicting the performance of these systems difficult to forecast. This paper will detail the latest experimental research into heat and mass transfer in buoyancy driven airflows into the influence of system height and aspect ratio. Two test rigs, resembling the essential features of a passive solar air heater are utilised. The test rigs have a system height 1m and 0.5m allowing the influence of system height on the airflow to be determined. Measurements are made of the air, plate and cover temperatures, the air velocities and the heat fluxes. The results from these two systems are compared with previous work from other researchers and the effect of the height of the channel (and the height to depth aspect ratio) is discussed. Keywords: Passive Solar, Natural Convection, Heating and Ventilation 1. INTRODUCTION Passive solar systems are playing an increasing role in reducing the mechanical cooling and heating loads of modern buildings. Solar façades rely on buoyancy driven (natural) convection to induce airflow. A Trombe Wall (figure 1) comprises a massive thermal wall (for example, made of concrete) with a black surface, which absorbs the incident solar radiation. As the surface (the absorber plate ) heats up, this in turn heats the air in the channel between it and the cover, inducing the heated air to circulate into the room behind it. Some sources have reported up to 65% of a building heating requirements being met by a Trombe Wall (Wigginton and Harris, 2002; Luminosu, 2003). By opening vents at the top of the Trombe wall, the system can operate in ventilation mode, creating a solar chimney. The Double-Skin Façade is essentially a pair of glass skins separated by an air corridor. Hirunlabh et al (1998) achieved an air extract air rate of up to 0.02kg/s, equivalent to 2.2 air changes per hour, for a small house in a tropical climate. Afonso et al (1999) reported achieving a 22% level of ventilation requirements in a Portuguese climate using solar chimneys. Solar chimneys were also extensively utilised in the BRE Environmental building in Garston. This building also used night-time cooling, computer controlled louvre to optimise light levels, ground source heat pumps and a PV array (Wigginton and Harris, 683

2 Ryan, Burek and Baker 2002). The combined effect of these systems was a reduction in total building energy use to 117 kwh/m 2, a 53% improvement on a normal office building and 30% better than best practice. However, overall the performance of the BRE building was a little disappointing as other buildings have achieved far greater reductions. Limerick County Hall, for example, used solar chimney effects, night time cooling and a double skinned façade glazed system is expected to achieve a 87% reduction in CO 2 emissions with a total energy use level of 76.4kWh/m 2. Figure 1: Trombe Wall Figure 2: Solar Chimney Increasing interest in photovoltaic (PV) facades suggests another application. The photovoltaic thermal façade (PVT) consists of a solar chimney type of arrangement which uses a PV array itself as an absorber plate. The convection of heat away from the PV array can be utilised to heat or ventilate the building, and the efficiency of array is also improved by the reduction in temperature. Eicker et al (1999) reported a 22% improvement in the performance of the PV arrays on the Mataro Public Library using this type of arrangement. Further, between 40% and 160% of ventilation needs and 12% of winter heating needs were provided by facade, in addition to 47,000 kwh per annum of electricity from the PV arrays. Although much of the basic science of buoyancy-driven convection has been extensively studied, it is a very complex subject. Specific applications, such as solar façades, are not completely understood, and so their exact performance can seldom be predicted accurately. This lack of knowledge contributes to the poor take-up of this type of technology. In particular, the relationship between airflow, heat transfer, heat input, the precise geometry and other characteristics of such systems are not clearly understood, and therefore many passive solar air collector systems have been designed on an ad-hoc basis. This paper presents some results from a test rig designed to simulate the performance of a passive solar heating panel under controlled conditions, as part of an on-going series of experiments designed to define the characteristics of these systems 2. PREVIOUS RESEARCH Several researchers have specifically investigated the issue of the heat transfer rates in thermosyphoning systems. For example, Bouchair (1994) and Chen et al (2003), studied vertical channels as solar chimneys. Both also considered the influence of 684

3 Passive solar heating facades aspect ratio (channel depth to height), concluding that these influence heat transfer, system efficiency and air mass flow rates. Chen reported that the temperature profile along the channel height was influenced by the aspect ratio, and that airflow rates continued to increase as the channel depth increased. He did not find an optimum channel depth, despite observing a backflow for channel depths greater than 300mm. He also noted a decline in temperatures above a certain vertical height in the channel, depending on the aspect ratio and channel depth. This suggests that there is a link between these factors and maximum temperature, and therefore thermal efficiency. Bouchair reported an increase in the airflow rate up to a channel depth of about 300mm (corresponding to an aspect ratio of 0.1 for his test rig), after which backflow can occur, resulting in a decline in performance. During a numerical analysis of buoyancy driven flow Fedorov and Viskanta (1997) noted that the velocity profile was continually developing up the channel height. These results closely match those of experimental work performed by Miyamoto et al (1986). Other researchers suggest that height may be important, independent of aspect ratio. For example, Afonso and Oliveira (2000) determined that efficiency is a function of height for solar chimneys, and Ong (2003) suggested that mass flow rate is a function of both channel height and depth. Although Balacco (2002) considered different channel heights analytically, most experimental investigations to date have been based on a single channel height, and therefore little experimental evidence exists for the influence of channel height (as opposed to aspect ratio) on the system performance. The present study follows very closely the experimental procedure used by Habeb (2003) and Habeb et al (2004) so that results can be directly compared. Habeb had however only considered a single system height (of 1.025m s) and only aspect ratios up to The present study would expand this to include a 0.51m test rig, cover a greater range of aspect ratio s and advance the study of transient behaviour as well as applying more enhanced dimensionless analysis to the results from both sets of tests. This paper reports some preliminary results, comparing the performance of these two test rigs. La Pica et al (1993) also used a similar test rig, 2.6m high, However, significantly his test rig had a silvered cover specifically to reduce radiation heat losses, so direct comparisons are not necessarily valid. Most investigations focus on steady-state conditions, and only a few experimental studies have referred to transient conditions. The controlled nature of the experimental work reported here allowed transient data to be captured reliably, and this is also presented in this paper. 3. EXPERIMENTAL TEST RIG AND PROCEDURE The experimental test rig was designed to resemble the essential features of a passive solar air heater (see figure 3). It comprised a vertical channel, open top and bottom (but with no inlet or outlet ducting) and closed at the sides. A controlled heat input was via an electrical heating mat behind the aluminium absorber plate, which was painted matt black. The cover was a sheet of transparent Perspex. Heat losses from the back of the channel were minimised by 100mm of rockwool insulation behind the heating mat. 685

4 Ryan, Burek and Baker Temperatures were measured using fine (0.2mm) thermocouples, and airflow with a temperature-compensated hot-bead anemometer, calibrated down to 0.15 m/s. All data was recorded using a multi-channel data logger. Experiments were conducted on the 0.5m test rig at channel depths of 20, 40, 60, 80, 100 and 150mm (corresponding to aspect ratios of 0.04, 0.08, 0.12, 0.16, 0.20, and 0.29 respectively) with heat inputs of 200, 400, 600, 800 and 915 W/m 2 (a total of 30 tests). The test procedure involved setting the channel to the required depth and switching on the power. A uniform heat flux (UHF) was thus applied, and the transient temperatures through the test rig and air velocity through the channel were recorded at 2-minute intervals. When steady-state conditions were established, after about 4-5 hours of heating, smoke tests were performed for a visual check on the airflow patterns, and air velocity profiles across the channel were measured. The power was then switched off and the transient cooling behaviour recorded, until the air temperature returned to ambient conditions. 3.1 Temperature and Velocity Profiles Absorber Plate Electric Heating Mat Plasterboard Rigid Box Insulation Hot Bead Anemometer (shielded) Warm air out Cover frame Perspex cover NOT TO SCALE Thermocouple positions indicated by x Cold air in The horizontal and vertical steady-state Figure 3: Test Rig schematic temperature and velocity profiles were of a similar form to those reported by Habeb (2003). Temperatures increased along the height of the test rig, as can be expected, but started to decrease from a point about twothirds of the height. This has been observed by several previous researchers, but the cause of this effect is open to interpretation. Temperatures across the channel depth showed a fairly flat profile in the airstream, with the temperatures of the (heated) plate and the (unheated) cover both higher than the air temperature. Therefore the heat transfer by radiation to the cover is greater than the heat transfer by convection to the air. The velocity profile at the bottom of the channel was found to be flat, showing that the flow was essentially undeveloped. At the top, the velocity close to the heated plate was higher than in the middle of the channel. A second peak close to the cover is also observed. This is noticeable particularly at high heat inputs (1000 W/m 2 ) and deeper channels (100mm, aspect ratio 0.1). This was also observed in the 1.0m experiments. The main difference between observations from the 1.0m test rig and the current investigation was that, with a shorter test rig but similar channel depths (i.e., greater depth-to-height aspect ratios), flows tended to be more unstable. Although the test rig was shielded as far as possible from the rest of the laboratory, the air flows were more prone to disturbance by external influences, such as draughts and other stray air currents in the room. Flow visualisation tests using smoke indicated that some 686

5 Passive solar heating facades recirculation took place within the channel at high aspect ratios, but no backflow occurred (i.e., no air exited at the bottom of the channel). 3.2 Mass Flow Rate and Thermal Efficiency Air flow rate is a key performance indicator for the system working in ventilation (solar chimney) mode. The mass flow rate, m in kg/s per metre width of the collector, is given by: m = ρvs (1) Figure 4 shows air mass flow rate against channel depth s for different heat inputs Where ρ is the air density, and v is the air bulk velocity. In general, the air flow rate increases with channel depth (could be expected, due to reduced friction forces) and with increasing heat input (also expected, due to higher temperatures and buoyancy effects). However, at low heat inputs, the airflow appears to decrease as the channel depth increases. Habeb s results also suggest a similar trend. This lends some support to the suggestion by Bouchair (1994) that there is an optimum channel depth for air flow through a solar chimney, and further, that the optimum depth increases as the heat input increases. The thermal efficiency of the system is defined as: η = Q gain /Q in (2) Where η is the thermal efficiency, Q gain is the heat gain and Q in the heat input. The heat gain, Q gain, can be determined by: Q gain = m c (T out T in )/H (3) Where c is the specific heat capacity of air, H is the height, Tout & Tin the outlet and inlet temperatures. Figure 5 shows efficiency as a function of heat input. There is no clear dependence of efficiency on channel depth, but efficiency is clearly a function of heat input. The figure shows an obvious distinction between the 1.0m test data and those from the 0.5m tests, hence implying that channel height has an influence on the thermal efficiency. 3.3 Heat Transfer Coefficient An overall heat transfer coefficient h can be defined as: Mass flow rate (kg/s) W/m2 400 W/m2 600 W/m2 800 W/m2 925 W/m Channel depth (m) Figure 4: Flow rate against channel depth h = Q gain / (T mp - T mc ) (4) Note that this is an overall heat transfer coefficient, based on the temperature of the plate and the air in the channel, but by implication it also includes heat gained by the air 687

6 Ryan, Burek and Baker from the cover. Figure 6 shows the heat transfer coefficient h as a function of heat input. This shows trends similar to those of efficiency against heat input (figure 5). Thermal efficiency 60% 50% 40% 30% 20% 10% Ryan (current data) Habeb (2003) 0% Power input (W/m 2 ) Figure 5: Thermal efficiency against power input Overall heat transfer coefficient (W/m 2 C) Power input (W/m 2 ) Ryan (current data) Habeb (2003) Figure 6: Heat transfer coefficient against power input 4. DIMENSIONLESS CORRELATIONS Dimensionless correlations were sought to establish relationships between the independent variables and the performance indicators. The following dimensionless numbers were used, each representing a characteristic of the system: 4.1 Rayleigh number Ra** The Rayleigh number Ra is often used when assessing the performance of buoyancydriven convection, and many workers use a modified Rayleigh number Ra*. This is the product of Rayleigh and Nusselt numbers, and is based on the heat convected to the air stream from the hot surface. In the current investigation, the effect of the heat input to the system is sought, and therefore a Rayleigh number based on the heat input is used, termed Ra** to distinguish it from other variations: Effectively: 2 4 ρ gβqinh Ra** =. Pr 2 kµ (5) Ra** = Ra*/η (6) 4.2 Non-dimensional height H/H0 This represents the height independently of the aspect ratio. H 0 is a nominal standard height, taken conveniently to be 1m. A set of correlations was performed, combining the experimental data from the current investigation with those of Habeb, to include the dimensionless height relationship. La Pica s data was excluded from this set of correlations, because the cover material was silvered during his experiments, which effectively adds another variable to the equation. The following relationships were obtained: 688

7 Passive solar heating facades 4.3 Thermal efficiency η = 3.39 x 10-5 (Ra**) 0.32 (s/h) (H/H 0 ) (7) correlation coefficient R 2 = 0.68 This implies that: 0.32 η Q in (8) η s (9) η H (10) Efficiency is a strong function of Q in and a weak function of H 4.4 Nusselt number and heat transfer coefficient Nu(H) = 1.29 x 10-4 (Ra**) (s/h) (H/H 0 ) (11) correlation coefficient R 2 = 0.92 This implies that: h Q in (12) h s (13) h H (14) The heat transfer co-efficient is a function of heat input 4.5 Reynolds number and mass flow rate Re(s) = 2.95 x 10-2 (Ra**) (s/h) (H/H 0 ) (15) correlation coefficient R 2 = 0.94 This implies that: m Q in m s m H Note that the exponents for H in equations (10), (14) and (18) can be deduced from the exponents for Ra**, s/h and H/H 0 in equations (10), (14) and (18). Mass flowrate is a function of all three, with s being an apparently stronger function than Q in. Nu(H)/(H/H 0 ) Nu(H)/(H/H 0 ) = Ra** R 2 = E E E E+13 Ra** Ryan (current data) Habeb (2003) Figure 7: Graph for correlation equation (14) current data (Ryan) and Habeb s data only. 689

8 Ryan, Burek and Baker 5. TRANSIENT EFFECTS For a passive solar collector the transient performance of the system is important. As the rate of heat input may not be continuous (clouds block out sunshine, sun blocked for part of the day by a neighbouring structure), the rate at which the system reaches steady state or drops out of steady state is important. Trombe Wall systems are designed to gradually built up a stock of residual heat during the day, within the rear thermal mass, which is then slowly released during the night. The current test rigs possess only a modest thermal mass, so they can perform only a limited degree of thermal storage. 5.1 Warm up Results For the 1.0m test rig the time constants for the overall system are in the order of minutes. For the channel air the typical time constants range from 130 to 80 minutes, and for the plate the time constants are shorter, minutes. This means the plate reaches the steady state before the airflow. The cover is the slowest to reach steady state, taking up to to 180 minutes to reach steady state. The air velocity time constant (the time taken for the air velocity to settle at its steady state is much slower than any of the others, with values ranging from 150 to 120 minutes. Figure 8: Transient performance during Heating phase for 20mm channel width The results from the 0.5m test rig were processed using Sigmaplot in a similar manner. The time constants during the heating phase heating for the 0.5m tests are generally noted at being 15% to 8% lower than the time constants for the 1.0m tests, indicating that the 0.5m rig heats up faster than the 1.0 test rig. This may be related to its lower thermal mass. Even though the test rig lacks the normal dense thermal mass of a Trombe wall. A Heat flux meter mounted on the rear wall revealed a heat flow level, of the same intensity as observed during steady state, for 10 to 20 minutes after the power is switched off. This heat flow then declines rapidly over 30 minutes. This indicates that even a relatively minor thermal mass, will still provide some residual heat gain. 5.2 Cooling Results Past experiments by authors such as Habeb did not make any record of the cooling rates of thermo-syhononing panels. Hence experiments were undertaken with the 1.0m high test rig to determine if these trends noted for the heating cycle carried over to the cooling cycle. It is found that the plate cools quickest, returning to steady state value (which is usually a degree or two above ambient) within a period of minutes from power off. No clear trend can be observed linking these time constants to either 690

9 Passive solar heating facades the original rate of heat input or the channel width. This is with the exception of a trend of declining time constants for the air velocity with increasing heat input. However, the T ratio s for many of these readings were low (3-5), raising question about the accuracy of these calculations. 6. DISCUSSION The current investigation was designed to assess the effect of system height on heat transfer and air flow, by comparing data from two test rigs which differed principally by their heights. For a system used as an air heater, it appears that height makes a small difference to the thermal efficiency. The results from the current tests suggest that, for a given height (i.e., 0.5m in this case), the thermal efficiency is a weak inverse function of channel depth, i.e., efficiency decreases slightly as channel depth increases. Habeb s data suggest that channel depth had no appreciable effect on efficiency. Both sets of data agree that the heat input has a positive effect of the system efficiency. This implies that a passive solar air heater would be more efficient in sunnier climates the reverse of what would normally be required! However, for a system used as a solar chimney, the system height does have a significant effect, as do the heat input and the channel depth. It should be noted that these correlations exclude the data from test with channel depths above 80mm, both for the 0.5m test rig and the 1.0m test rig. Above this figure, the flow rate showed signs of decreasing as the channel depth increased, which supports the observation by Bouchair (1994) and the suggestion that there is an optimal channel depth for a solar chimney. However, at present there are insufficient data to take this analysis any further. Although some of La Pica s data are useful for some comparisons, he used a silvered cover, specifically to reduce heat losses. Therefore his data are not directly comparable with those from the rest of the data used in these calculations. The correlations presented in this paper were developed using the regression tools from the Analysis Tool Pack in Microsoft Excel. The calculations were also performed using other curve-fitting software (for example, SigmaPlot). The correlations did not match exactly, but were, nevertheless, close enough to give confidence in the overall correlations and conclusions. However, it does suggest that, even though the results are presented in this paper to 3 significant figures, perhaps, in general, they should be presented to only 2 significant figures. There is a clear link between the rate of heat input and the time taken by the different elements of a collector to reach steady state. Time constants decrease with increasing heat input. A reduction in system height from 1.0m to 0.5m s results in an decrease in time constants by 15%, when compared via channel depths and up to 35% when contrasted using aspect ratios. There is no correlation between heat input nor the maximum temperature reached and the rate at which the system cools down. 7. CONCLUSIONS Tests were performed on a vertical heated channel, which represents the essential workings of a passive solar air heater. In this way, the effect of system height could be 691

10 Ryan, Burek and Baker determined. Using only data from tests with channel depths up to 80mm, the principal results are as follows: For a system working as a solar air heater, the efficiency of the system depends principally on the heat input: η Q in 0.32 (11) Efficiency is a weak function of both channel depth and channel height. For a system operating as a solar chimney, where mass flow rate is the principal requirement, heat input, channel depth and height are important: m Q in (16) m s (17) m H (18) However, it is noted that other workers suggest that there is an optimum channel depth, above which flow rate starts to decrease with increasing channel depth. There are indications from the current investigation that support this suggestion, but there is insufficient data to take this analysis further. In terms of future activities, this study is being continued, and a test rig 2.0m high is being constructed. This will give more data on the effects of height on the key performance parameters of this type of system. 8. REFERENCES Afonso C. and Oliveira A. (2000), Solar Chimneys: Simulation and Experiment, Energy and Buildings 32, pp Balacco C. (2002) A Simple Model to Study Ventilated Facades Energy Performance, Energy and Buildings 34, pp Bouchair A. (1994) Solar Chimney for Promoting Cooling Ventilation in Southern Algeria, Building Serv. Eng. Technol 15 (2) pp Chen Z.D., Bandopadhayay P., Halldorsson J., Byrjalsen C., Heiselberg P. and Li Y. (2003) An Experimental Investigation of a Solar Chimney Model with Uniform Wall Heat Flux, Building and Environment Journal 38 (7) pp Eicker U., Fux V., Infield D., Mei L., Vollmer K. (1999), Thermal performance of building integrated ventilated PV facades, Proc. of International Solar Energy Conference, Jerusalem. Fedorov A.G. and Viskanta R. (1997) Turbulent Natural Convection Heat Transfer in an Asymmetrically Heated, Vertical Parallel-Plate Channel, Int. J. of Heat and Mass Transfer 40 (16) pp Gan G. (1998) A Parametric Study of Trombe Walls for Passive Cooling of Buildings, Energy and Buildings 27, pp Habeb A. (2003) Airflow and Heat Transfer in Passive Solar Collectors. MPhil thesis, Glasgow Caledonian University. Habeb A., Ryan D. and Burek S. (2004) Experimental Investigation into Heating and Airflow in Passive Solar Building Facades, Proc. SET 2004 conference, Nottingham, England Kempe L. (1999) Heat Transfer Modelling, MSc Thesis, Glasgow Caledonian University. Khedari.J, Rachapradit N. and Hirunlabh J. (2003) Field Study of Performance of Solar Chimney with Air-Conditioned Building, Energy 28 (11) pp

11 Passive solar heating facades La Pica A., Rodonò G. and Volpes R. (1993) An Experimental Investigation on Natural Convection of Air in a Vertical Channel, Int. J. Heat and Mass Transfer 36 (3) pp Luminosu I. (2003), Experimental Studies and economic considerations on a living spaced heated through passive solar gain and through electric power, Thermal Science, Vol. 7, No. 2, pp Miyamoto M., Katoh Y., Kurima J., and Sasaki H., (1986) Turbulent Free Convection Heat Transfer from Vertical Parallel Plates, In Heat Transfer, Vol. 4, pp , Hemisphere, Washington, DC. Ong K. S. (2003) A Mathematical Model of a Solar Chimney, Renewable Energy 28, pp Wigginton M. and Harris J. (2002) The Characteristics of an Intelligent Skin, Architectural Press, ISBN Nomenclature a, b & c regression coefficient s c specific heat capacity of air (J/kg C) H height of the channel (m) H 0 reference height (taken as 1m) g acceleration due to gravity (= 9.81 m/s 2 ) h overall heat transfer coefficient (W/m 2 C) k thermal conductivity of air (W/m C) m air mass flow rate (kg/s) Nu(H) Nusselt number based on height = h.h/k Q in heat input (W/m 2 ) Q gain heat gain by the air in the channel (W/m 2 ) R 2 correlation coefficient 2 ρ gβqin H Ra* modified Rayleigh Number = 2 k Ra** modified Rayleigh number based on heat input = Ra*/η (see equation 6) Re Reynolds number = ρv.s/µ Pr Prandtl number s channel depth (m) T in inlet air temperature ( C) T mc mean temperature of the air in the channel ( C) T mp mean temperature of the heated plate ( C) T out outlet air temperature ( C) v air bulk velocity (m/s) β temperature coefficient (1/K) η thermal efficiency µ dynamic viscosity (kg/m s) ρ density (kg/m 3 ) µ 4 693