Está en la página 1de 9

Building and Environment 35 (2000) 297±305

www.elsevier.com/locate/buildenv

Radiative cooling of buildings with ¯at-plate solar collectors


Evyatar Erell*, Yair Etzion
The Center for Desert Architecture and Urban Planning, J. Blaustein Institute for Desert Research, Ben-Gurion University of the NegevÐ
Sede-Boqer Campus, Israel 84990
Received 20 October 1998; received in revised form 5 February 1999; accepted 25 February 1999

Abstract

Roof ponds cooled by nocturnal long wave radiation have often been proposed as a cheap and e€ective means of providing
thermal comfort in buildings in hot-arid locations. Many of the schemes incorporate ¯at-plate radiators through which the water
is circulated at night to be cooled. An analytical model originally developed for heating applications was adapted to the
description of such a nocturnal, long wave radiative cooling apparatus. The accuracy of the model was veri®ed on the basis of
experimental data from three types of radiators tested at the experimental facilities of the Center for Desert Architecture and
Urban Planning at Sede-Boqer, Israel, under a variety of operating regimes. The model allows accurate prediction of outlet
temperatures, taking into account the design characteristics of the speci®c radiator, the environmental conditions and the pattern
of operation of the system. Analysis of ¯at-plate solar collectors converted to radiative cooling of buildings suggests that under
typical environmental conditions, they may be less ecient for this type of application than radiators having no ®ns. # 2000
Elsevier Science Ltd. All rights reserved.

1. The thermal performance of roof-mounted cooling the model was not used to provide recommendations
radiators as to the preferred characteristics of a system designed
speci®cally for cooling buildings.
Roof ponds have often been proposed as a means of The radiators used in most previous experiments
providing thermal comfort in buildings in hot-arid were based on ¯at-plate solar collectors adapted to
regions [1±6]. In a building cooled by means of a roof cooling, generally by removal of the glazed cover
pond, the water in the roof pond absorbs excess heat plate. While this may have been a convenient means of
through the ceiling, which should have a high thermal testing the concept, closer analysis shows that the
conductivity. This energy is then dissipated by means characteristics required of a cooling radiator di€er in
of convection, radiation or evaporation. There are sev- several small but signi®cant ways from those of a ¯at-
eral approaches to the problems associated with the plate collector designed for solar heating of water.
con¯icting needs of cooling the water in the pond
. First, the magnitude of the radiative heat exchange
(usually at night), yet preventing unwanted solar heat
and the properties of roof pond systems in general
gains during the day. One of the solutions proposed
preclude the possibility of initiating thermo-siphonic
has been to insulate the roof pond by means of ¯oat-
circulation of water. For space cooling to be e€ec-
ing insulation, and to circulate the water through cool-
tive, the water in the roof pond should not be war-
ing radiators placed above the pond, when
mer than about 258C. Under typical environmental
environmental conditions are favorable [4]. The per-
conditions in the desert locations where such a sys-
formance of such radiators has been modeled [7], but
tem might be practical, the temperature at the outlet
of the radiator cannot be expected to be lower than
* Corresponding author. Tel.: +972-7-6596875; fax: +972-7-
158C, and is usually no colder than 208C. The tem-
6596881. perature di€erence between the radiator inlet and
E-mail address: erell@bgumail.bgu.ac.il (E. Erell). outlet is thus not large enough to generate a thermal

0360-1323/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 3 6 0 - 1 3 2 3 ( 9 9 ) 0 0 0 1 9 - 0
298 E. Erell, Y. Etzion / Building and Environment 35 (2000) 297±305

force suciently strong to overcome internal friction F ' is the ratio of the actual useful energy gain to the
and to initiate water circulation. A cooling radiator useful gain that would result if the collector absorbing
must therefore rely on a pump to circulate the cool- surface had been at the local ¯uid temperature. It is
ing ¯uid through it. The rate at which this ¯uid is given by
forced through the system a€ects cooling perform-
1
ance, so it becomes a key operating factor, rather
than being simply the result of the combined e€ects 0  UL 
F ˆ …2†
of the incoming solar radiation and the hydraulic 1 1 1
W ‡ ‡
design of the collector. UL ‰D ‡ …W ÿ D†F Š Cb pDi hfi
. Second, the e€ects of convection on the thermal
where Cb is the bond conductance between pipes and
exchange at the surface of the radiator are not
absorber plate [W/m 8C], h® is the heat transfer
always detrimental, as in a heating collector.
coecient between the ¯uid and the tube interior
Whereas in a heating application, the temperature
[W/m2 8C], D is the external diameter of the pipes [m]
of the ¯uid is assumed to be greater than that of the
and Di is the internal diameter of the tubes [m]. F, the
surroundings, in a cooling radiator this may not
®n eciency factor, is a convenient means of
always be the case. The extent to which convection
expressing the overall geometric characteristics of the
impedes or improves the performance of the cooling
radiator, and is given by
radiator depends on its operating characteristics
with regard to the environmental conditions and to tanh‰m…W ÿ D†=2Š
the temperature of the water in the radiator. Fˆ …3†
m…W ÿ D†=2
. Finally, the outgoing radiative ¯ux on which radia-
tive cooling systems are based is relatively small in where F is the ®n eciency, W is the riser spacing [m],
comparison with the solar gain in heating appli- D the pipe diameter [m], and m is given by
cations. The typical ®n-and-pipe design of ¯at plate r
collectors may not be the most ecient heat exchan- UL
mˆ …4†
ger for radiative cooling. kd
While the di€erences between ¯at plate collectors and where UL is the overall heat loss coecient of the
cooling radiators warrant a separate analytical study radiator [W/m2 8C], k the thermal conductivity of the
of the performance of the latter, the similarities are ®n [W/m 8C] and d the ®n thickness [m].
such that existing models may be adapted for this pur- The adaptation of the Due and Beckman [8] ex-
pose. The following discussion presents such a pro- pression to describe a cooling radiator requires the
cedure. replacement of the value for solar heat gain, S, with
an expression for radiative heat loss. The following lin-
earized form of the Stefan±Boltzmann Law [9], is a
2. Modelling a ¯at-plate cooling radiator convenient means of relating the environmental con-
ditions to the net radiative heat transfer at the surface
The temperature distribution in the ¯ow direction is of the radiator:
a key parameter in the design of a cooling radiator.
The calculation of the ¯uid temperature at any given Rnet ˆ 4EsT 3air …Trad ÿ Tsky † …5†
point along the cooling pipe may be derived from the
where Rnet [W/m2] is the net radiative heat loss, E is
following expression, proposed by Due and
the emissivity of the radiator surface, s is the Stefan±
Beckman [8] for a solar heating collector:
Boltzmann constant [5.67  10ÿ8 W/m2 K4], Tair [K] is
 
Tf ÿ Ta ÿ S=UL UL nWF 0 y the ambient air temperature, Trad [K] the radiator tem-
ˆ exp ÿ …1† perature and Tsky [K] the equivalent sky temperature.
Tfi ÿ Ta ÿ S=UL _ p
mC

where Tf is the ¯uid temperature [8C] at a distance y


[m] from the collector inlet, Ta is the ambient air tem-
perature [8C], T® is the ¯uid temperature at the collec-
tor inlet [8C], S is the absorbed solar energy [W/m2], 3. Application of the model
UL is the overall heat loss coecient of the collector
[W/m28C], n is the number of pipes, W the distance Analysis of the components of the above model,
between the pipes [m], F ' the collector eciency factor, which is presented in a general form, allows con-
.
m the mass ¯ow rate through the collector [kg/s] and clusions to be drawn regarding the design and oper-
Cp the speci®c heat of the ¯uid [J/kg 8C]. ation of a radiator speci®cally designed for the
In a heating collector, the collector eciency factor radiative cooling of buildings.
E. Erell, Y. Etzion / Building and Environment 35 (2000) 297±305 299

3.1. Design characteristics of the radiator the wind velocity. Clark and Berdahl [10] di€erentiated
between three situations (units converted to SI):
3.1.1. Radiator heat loss coecient 1. The radiator is colder than ambient air, and wind
The overall heat loss coecient of the radiator (UL) speed is low enough (v < 0.076 m/s) to allow free con-
is due to the e€ects of radiation and of convection. vection, i.e. the ¯ow is laminar:
The contribution of the components of a typical radia-
torÐtop (Ut), back (Ub) and edge (Uc), are calculated hc ˆ 0:8 ‰W=m2 8CŠ …7†
separately, to account for the di€erences in the heat
2. The radiator is warmer than ambient air, and there
exchange processes occurring at each of them, then
is free convection (v < 0.45 m/s):
combined.
Most of the heat loss from a radiator occurs at the hc ˆ 3:5 ‰W=m2 8CŠ …8†
top surface. For the sake of convenience, the radiative
component of the heat loss may be expressed as the 3. Turbulent ¯ow, regardless of the radiator tempera-
product of the temperature di€erence between the ture (v is greater than the values noted above for the
radiator surface and the sky, with the linearized radia- relevant radiator temperatures):
tive heat transfer coecient hr, which is derived from
Eq. (5): k…0:054Re0:8 Pr0:33 †
hc ˆ …9†
L
hr ˆ 4Erad sT 3air …6†
where k is the thermal conductivity of the air
where hr [W/m2 8C] is the radiative heat exchange [W/m2 8C], Re the Reynolds number and Pr the
coecient, s is the Stefan±Boltzmann constant Prandtl number. L, the characteristic length of the ra-
(5.67  10ÿ8 W/m2 8C), Erad is the emissivity of the diator, is taken to be about 6.1 m. Clark and Berdahl
radiator, and Tair is the ambient air temperature [K]. [10] note that the coecient applied to the Reynolds
In applications involving the radiative cooling of number, 0.054, was increased above the smooth sur-
buildings, the temperature of an exposed, non-glazed face value (0.036), to account for surface roughness.
radiator may be expected to be between 15 and 308C They also suggest that the following simpli®cation
(278±293 K). Therefore, the range of values which this of the relationship for turbulent ¯ow is appropriate for
coecient can have in practice is quite limitedÐabout wind speeds (v ) of 1.35±4.5 m/s:
4.4±5.1 W/m2 8C (Fig. 1).
hc ˆ 1:8 ‡ 3:8n ‰W=m2 8CŠ …10†
If the radiator is not glazed, a signi®cant proportion
of the heat dissipated may be due to convection. The Where wind speeds are lower than 1.35 m/s, this ex-
e€ects of convection are accounted for in the model by pression overestimates the e€ects of convection.
the convective heat exchange coecient (hc). If the back of the radiator is not exposed, the back
Numerous experiments have been carried out to evalu- losses are due to conduction through the insulation.
ate this quantity, which is related to the relative tem- The back loss coecient is approximately
perature of the radiator and the ambient air, and to
k
Ub ˆ …11†
L
where k and L are the thermal conductivity [W/m 8C]
and thickness [m] of the insulation, respectively. If the
radiator is not insulated, so as to increase convective
heat exchange, the back losses due to conduction are
calculated in the same manner as the top losses, allow-
ing for the di€erences in the convective heat exchange
coecient (hc) due to the e€ects of thermal forces and
to possible di€erences in the exposure to the air¯ow.
The edge losses in a non-glazed radiator are negli-
gible (less than 5%), unless the radiator is very small.
In summary, the overall heat loss coecient of an
exposed, non-glazed radiator may vary from about
6 W/m2 8C, if the air is still, to 30 W/m2 8C or more if
there is a signi®cant breeze. While the e€ects of con-
Fig. 1. The e€ect of ambient air temperature on hr, the radiative vection on the overall performance of a particular
heat loss coecient of an exposed (non-glazed) radiator with an radiator may be signi®cant, their calculation requires a
emissivity of 0.9. knowledge of the wind conditions typical of the instal-
300 E. Erell, Y. Etzion / Building and Environment 35 (2000) 297±305

lation site. The empirical relationships set out above maximizing radiative heat loss, there is no practical
allow a fairly accurate estimate of the range of values means of doing so on the scale required to cool a
the heat loss coecient of a radiator may have. building. The most common approachÐusing infra-
However, unlike glazed collectors used for heating, it red transparent glazing such as very thin polyethylene
is not possible to assign an unglazed cooling radiator a sheets [11]Ðdoes not deal with the problems caused
single overall heat loss coecient valid for all (or even by the deposition of dust and the accumulation of con-
most) wind conditions. densation on the cold surface of a wind screen. Erell
and Etzion [4] have pointed out the futility of this
3.1.2. Fin eciency factor approach from a practical point of view, even assum-
The eciency of a typical ¯at plate collector with ing that a relatively cheap, durable infra-red transpar-
pipes and absorber plate is determined by its geometry, ent glazing were available. Radiators exposed to the
the materials it is constructed of and the quality of the atmosphere on building roof tops are therefore not
glazing cover, which a€ects both the solar heat gain glazed, and the e€ects of convection at the surface of
and the heat loss from the absorber plate. The relative the absorber (or rather, emitter) plate become domi-
importance of some of these factors is somewhat chan- nant.
ged in a cooling application. The e€ects of convection on the ®n eciency factor
The primary geometric parameters describing ¯at are expressed in Eqs. 3 and 4. Fig. 2, derived from
plate collectors are the diameter of the risers and their these equations, illustrates the e€ect of increasing con-
spacing. In heating collectors, the use of an absorber vective heat exchange, and thus the overall heat loss
plate allows a concentration of solar energy resulting coecient (UL), on the ®n eciency of a typical rec-
in higher water temperatures. This is because su- tangular ®n. (The calculations were carried out for a
ciently thick copper plates (up to 1 mm thick) in good typical collector design, with a steel absorber plate
thermal contact with the pipes lose little energy to the 0.5 mm thick, and risers 10 mm in diameter in good
surroundings even if the collector cover consists of a thermal contact with it, for several values of W, the
single glazing. spacing of the risers.)
The design of radiators for cooling must address It is evident from Fig. 2 that an increase in the con-
other considerations. Windscreens, in the form of vective heat loss coecient of the radiator (which may
glazed covers or otherwise (honeycomb structures are result, for instance, from increased wind speed), causes
one alternative), are required if it is the intention to a reduction of the ®n eciency (F ). This e€ect is
cool the radiator surface well below the ambient air noticeable for the whole range of ®n widths, but
temperature. This may not always be desirable, since becomes dominant when the tube spacing is large.
convection may increase heat dissipation if the ambient When the tube spacing is smallÐ5 cm or less in this
air temperature is low enough. However, even assum- caseÐthe reduction in ®n eciency is less than 10%
ing that environmental conditions are such that it is even for fairly high heat loss coecients. (A heat loss
desirable to reduce convective heat exchange while coecient of 20 W/m2 8C, for example, may be
measured if the radiator surface is exposed, i.e.
unglazed, and if the wind speed is about 5 m/s).
The e€ect of changes in the ®n eciency on the
overall performance of a cooling radiator are complex,
and depend on the relationship between the ambient
air temperature and the temperature of the radiator
surface:
. If Ta < Tr, convection increases the rate of heat dis-
sipation, reinforcing the e€ects of radiation. In this
case, a higher ®n eciency (F ), which in turn leads
to a higher collector eciency (F '), results in
improved radiator performance both in radiation
and in convection.
. If Ta > Tr, convection leads to heat transfer from
the air to the radiator, counter-acting the e€ects of
radiation. In this case, a higher ®n eciency still
leads to improved heat dissipation through radi-
Fig. 2. The e€ect of the overall heat loss coecient (UL) of an ation, but also results in increased convective gains.
exposed, non-glazed tube-and-®n radiator on ®n eciency (F ), for The net balance of these opposite forces depends on
several widths of ®n. the properties of the speci®c radiator, on the en-
E. Erell, Y. Etzion / Building and Environment 35 (2000) 297±305 301

vironmental conditions and on the ¯uid inlet tem- sky temperature, which determines the potential for
perature and ¯ow rate. radiative cooling, Eq. (5). The e€ect of wind speed is
Fig. 3 shows the e€ect of the ®n eciency factor (F ') accounted for in the value assigned to the collector
on changes in the ¯uid temperature of a hypothetical heat loss coecient, UL, as described previously.
While the ambient air temperature is obtained easily,
radiator exposed to di€erent ambient air temperatures.
the so-called `sky temperature', de®ned as the tempera-
The radiator is 1 m wide by 2.5 m long; ¯uid inlet tem-
ture of a black body radiator emitting the same
perature is 238C; mass ¯ow rate is 0.03 kg/s; UL=15
amount of radiative power as the sky, is dicult to
W/m2 8C; and the sky temperature depression is 158C.
measure. As data are not available for most locations,
Under these conditions, there is a net cooling e€ect
it may be more convenient to calculate approximate
(indicated by a negative change in the ¯uid tempera-
ture) until the ambient air temperature rises to 38C values of the sky temperature, based on ambient air
above the radiator temperature. At this point, radia- temperature and relative humidity data. Sky tempera-
ture is given by the following expression [10]:
tors with di€erent ®n eciencies (but identical in other
respects) would have a similar, neutral, energy balance. Tsky ˆ E1=4
sky Tair …12†
The optimal design of a cooling radiator therefore
depends on an understanding of the environmental where Tsky and Tair are given in K. The sky emissivity
conditions it is likely to operate in. In general, if air may be calculated as a function of the dew point tem-
temperatures are very high, the net cooling output of a perature, using the following empiric relationship:
radiator would probably be too low to justify installa-
tion, assuming that convection cannot be suppressed Esky ˆ 0:006Tdp ‡ 0:74 …13†
entirely. If ambient air temperatures are equal to or
lower than the ¯uid inlet temperature for most of the where the dew point temperature is given in 8C.
operating period, then a high ®n eciency is preferred.
3.3. Operating characteristics
3.2. Environmental conditions
The proposed model allows evaluation of the e€ects
The proposed model, Eq. (1), incorporates two of two factors associated with the operation of the sys-
.
measures of the environmental conditions which have tem. The mass ¯ow rate (m ) is the primary means of
a direct e€ect on the cooling output of a radiator controlling the output temperature of the radiator.
exposed to the night skyÐambient air temperature The inlet temperature ( fi) is not under the direct con-
(Ta), which a€ects convective heat exchange, and the trol of the designer, since it is equal to the temperature
of the roof pond. However, the overall design of the
system should be such that the temperature of the
water in the roof pond ¯uctuates within a fairly
narrow, predetermined range, if it is to be useful as a
cooling system.
The e€ect of altering the operating parametersÐthe

Fig. 3. The e€ect of the ®n eciency factor (F ') on changes in the


¯uid outlet temperature of a hypothetical radiator exposed to di€er-
ent ambient air temperatures. (The radiator is 1 m wide by 2.5 m
long; ¯uid inlet temperature is 238C; mass ¯ow rate is 0.03 kg/s; Fig. 4. The e€ect of changes in the mass ¯ow rate on the longitudi-
UL=15 W/m2 8C; and the sky temperature depression is 158C. nal temperature pro®le of a hypothetical cooling radiator.
302 E. Erell, Y. Etzion / Building and Environment 35 (2000) 297±305

inlet temperature and the mass ¯ow rateÐis illustrated 4. Experimental setup
by applying the proposed model to a hypothetical
radiator 1 m wide, with an overall eciency (F ') of The accuracy of the model described above was
0.8. The environmental conditions were assumed to be: assessed with regard to experimental data assembled in
ambient air temperature of 248C, sky temperature previous years [12], in which three types of ¯at-plate
depression of 15 K, and little wind. Under these solar heating panels were adapted to nocturnal radia-
conditions, an unglazed, exposed radiator would tive cooling.
have a convective heat loss coecient (Uc) of about
10 W/m2 K. 4.1. Properties of the radiators
Fig. 4 shows the e€ect of altering the ¯ow rate on
the longitudinal temperature pro®le of a hypothetical Three types of radiators were used in the exper-
radiator. The trade-o€ between ¯ow rate and radiator iment:
length is self evident: for a speci®c radiator length, any
1. (A) A Lordan LSC-F collector, measuring 2.18 m
desired outlet temperature may be attained, within the
long by 1.27 m wide, and consisting of 12 copper
constraints of the environmental conditions, by manip-
risers 58 00 in diameter, 10 cm on center and con-
ulating the ¯ow rate.
nected to each other by means of a copper manifold
Fig. 5 shows the e€ect of altering the ¯ow rate on
of 1 18 00 in diameter at each end. Each riser is
the cooling output at several inlet temperatures.
installed at the center of a thin stainless steel absor-
Increasing the mass ¯ow rate reduces the temperature ber plate running along the long axis of the collec-
di€erence between radiator inlet and outlet, resulting tor (Fig. 6). The lower surface of the collector was
in an increase in the mean surface temperature of the
encased in 25 mm polyurethane foam insulation
radiator. This in turn results in increased radiative
coated with a re¯ective aluminum foil. The upper
cooling, under all environmental conditions. When the
surface, which is usually glazed in this type of col-
radiator is warmer than the ambient air, increasing the
lector, had the glass removed and was exposed to
¯ow rate also increases the cooling output due to con-
the ambient air.
vection; if the radiator is colder than ambient dry bulb
2. (B) A Lordan LSC-F collector like the ®rst one, dif-
temperature (DBT), then convective heat exchange fering only in that the polyurethane insulation was
tends to counteract the e€ects of radiation. removed from the bottom side. This collector was
Increasing the ¯uid ¯ow rate does, however, have
thus exposed to convective and radiative heat
diminishing returns. As the ¯ow rate increases, the sur-
exchange on both its upper and lower surfaces.
face temperature of the radiator approaches that of
3. (C) A Chromagen CR-12 collector similar to the
the ¯uid at the inlet. The practical limit is a much
LSC-F collector, except that instead of having thin
lower ¯ow rate, which takes into account the power
metal absorber plates running parallel to the risers,
required to operate the pump as well as the heat trans-
it had specially contoured aluminum ®ns perpen-
fer processes occurring at the radiator surface. dicular to them (like an automobile radiator). This
too had both the lower insulation and the upper
glazing removed to maximize convective heat
exchange.

4.2. The cooling system

The system consisted of a shallow roof pond, 10±


15 cm deep, on a reinforced concrete roof. The water
was contained in a waterproof PVC membrane lining,

Fig. 5. The e€ect of changes in the mass ¯ow rate on the net cooling
power of a hypothetical radiator at di€erent temperatures of the
¯uid at the radiator inlet. Fig. 6. Axonometric section of radiator type `A'.
E. Erell, Y. Etzion / Building and Environment 35 (2000) 297±305 303

and was covered by ¯oating polystyrene insulation model under di€erent environmental conditions, and
10 cm thick. The purpose of the insulation was to taking into consideration the e€ect of various mass
reduce unwanted heat gain from the ambient air, par- ¯ow rates.
ticularly during the daytime, when both air tempera- UL, the overall heat loss coecient of the collector,
ture is high and the internal cooling load greatest. was calculated separately for each of the radiators, using
Water contained in this roof pond was circulated at actual wind data on a typical night. The mean wind
night through radiators ®xed to a frame above the velocity for the simulation period was approximately
roof, and was cooled by long wave radiation and by 1.5 m/s. The radiative component of the top loss coe-
convection. cient was taken to be 5 W/m2 8C, while the convective
heat exchange coecient was calculated using Eq. (10).
4.3. Data acquisition and performance monitoring The overall heat loss coecients thus derived for
radiator types `A', `B' were approximately 13 and 18
Fluid temperature was measured using PT-100 sen- W/m2 8C, respectively. The e€ect of the vertical alumi-
sors (in waterproof casings), calibrated individually so num ®ns in radiator type `C' could not be calculated:
that the di€erential error of measurement was less the increased aerodynamic roughness and the much
than 0.18C. Each temperature record was the average greater surface area were estimated to double the con-
of two sensors placed adjacent to each other and mon- vective heat transfer coecient, resulting in an overall
itored separately. Water ¯ow rate was recorded elec- heat loss coecient for this radiator of 32 W/m2 8C.
tronically using electro-magnetic pulse inducers with a The sky temperature, required for the calculation of
resolution of 1 liter. the radiative heat loss, was derived from the net radio-
The environmental conditions were monitored on- meter measurements on site. The value used for the
site. The properties measured included ambient air dry radiator temperature in these calculations was the
bulb temperature, relative humidity, wind speed and average between radiator inlet and outlet. Since the
direction, global radiation on a horizontal surface and di€erence between the inlet and outlet temperature of
net radiation in the range 0.25±60 mm. The data collec- the ¯uid was in the range of 0.5±2.58C, depending on
tion system installed was based on a Campbell 21 the ¯ow rate, the error involved in using such a rela-
data logger. The data were retrieved for processing on tively crude measure was in fact quite small. The other
a PC computer. required inputsÐambient air temperature, mass ¯ow
rate and ¯uid inlet temperature were measured
directly.
5. Veri®cation of analytical model The model calculations were compared with exper-
imental results for all three radiator types, operating
The radiative cooling experiment had been con- under a variety of ¯ow regimes. The calculated outlet
ducted in several phases, allowing an evaluation of the temperature is in very close agreement with the exper-

Fig. 7. Outlet temperature of radiator type `A', on a typical night at Sede-Boqer, 28 July 1991.
304 E. Erell, Y. Etzion / Building and Environment 35 (2000) 297±305

Fig. 8. Outlet temperature of radiator type `B' on a typical night at Sede-Boqer, 28 July 1991.

imental value, under all conditions, and is demon- radiators was slightly di€erent, re¯ecting their ex-
strated for the night of 28 July 1991 (Figs. 7±9). posure to convective heat exchange. Radiator type `A'
An increase in the outlet temperature measured (with back insulation) was the least a€ected, as the
during this particular night between 01:00 and 01:30 h, ¯uid outlet temperature increased by only 0.058C. In
apparent in all three radiators, gives an added insight radiator `B', with no back insulation, the temperature
as to their properties. This temperature increase, fol- increase was 0.278C, and in radiator type `C', with the
lowing a monotonic decline from 22:30 h, was due to highest exposure to convection it was 0.398C. In each
an increase in ambient air temperature of 1.58C in this case, the model allowed accurate prediction of the
interval. This change was accompanied by, and was ¯uid outlet temperature.
probably the result of, an increase in wind speed from
about 2 m/s to over 5 m/s. The sky temperature de-
pression during this interval increased by about 0.68C, 6. Conclusion
re¯ecting (in part) the increase in ambient air tempera-
ture. After 01:30 h, the environmental conditions The application of nocturnal radiative cooling to the
returned to normal. The e€ect on each of the three provision of thermal comfort in buildings requires

Fig. 9. Outlet temperature of radiator type `C' on a typical night at Sede-Boqer, 28 July 1991.
E. Erell, Y. Etzion / Building and Environment 35 (2000) 297±305 305

improved radiator designs. An analytical model was roof pond. In: Proceedings of the Fourth National Passive
adapted for the study of such radiators, and veri®ed Solar Conference American Section of ISES, 1979. p. 499±503.
[4] Erell E, Etzion Y. A radiative cooling system using water as a
against experimental data for three types of cooling
heat exchange medium. Architectural Science Review
radiators. Analysis of the desired properties for cooling 1992;35(2):39±49.
radiators suggests that under some environmental con- [5] Givoni B. Passive and low energy cooling of buildings. New
ditions, the ®ns used in typical ¯at plate solar collec- York: Van Nostrand Reinhold, 1994.
tors are less ecient for cooling applications, and a [6] Santamouris M, Asimakopolous D. Passive cooling of build-
radiator consisting entirely of pipes may be more ings. London: James and James, 1996.
appropriate for this purpose. The model presented [7] Klitisikas N, Geros V, Santamouris M, Dascalaki E,
Kontoyiannidis S, Argiriou A. SummerÐa tool for passive
here may also be used to optimize the length of such a
cooling of buildings Version 2. CIENE, University of Athens,
radiator or the hydraulic speci®cations of the pump, to 1996.
allow for maximum utilization of the fairly limited po- [8] Due JA, Beckman WA. Solar engineering of thermal pro-
tential of radiative cooling. cesses. New York: John Wiley and Sons, 1991.
[9] Martin M. Radiative cooling. In: Cook J, editor. Passive cool-
ing. Cambridge, MA: The MIT Press, 1989.
References [10] Clark E, Berdahl P. Radiative cooling: resource and appli-
cations. In: Miller H, editor. Proceedings of the Passive Cooling
Workshop Amherst, MA 20 October, Center for Energy
[1] Yellot J. Solar roof ponds, early tests of the `skytherm' system.
In: Passive solar heating and cooling, Conference and Ecient Design, Berkeley, CA, 1980. p. 177±212.
Workshop Proceedings 18±19 May, Albuquerque, New Mexico, [11] Granqvist CG, Hjortsberg A. Radiative cooling to low tempera-
1976. p. 54±62. tures: general considerations and application to selectively emit-
[2] Hay H. Skytherm natural thermal controlÐdesign, production ting SiO ®lms. Journal of Applied Physics 1981;52(6):4205±20.
and commercialization. In: Proceedings of US Department of [12] Etzion Y, Erell E Radiative cooling of buildingsЮnal scienti®c
Energy Passive and Hybrid Solar Energy Program Update report presented to the Bundesministerium fuer Forschung und
Meeting, Washington, D.C., 21±24 September, 1980 3-80±3-83. Technologie, Germany, and the Israel Ministry of Science and
[3] Crowther K. Cooling from an evaporating, thermosiphoning Technology. Ben-Gurion University 1993.

También podría gustarte