Professional Documents
Culture Documents
Chen_2003_An Experimental Investigation of a Solar Chimney Model With Uniform Wall Heat Flux
Chen_2003_An Experimental Investigation of a Solar Chimney Model With Uniform Wall Heat Flux
com
Received 16 October 2002; received in revised form 1 February 2003; accepted 11 March 2003
Abstract
Experiments were carried out using an experimental solar chimney model with uniform heat %ux on one chimney wall with a variable
chimney gap-to-height ratio between 1:15 and 2:5 and di9erent heat %ux and inclination angles. Results showed that a maximum air%ow
rate was achieved at an inclination angle around 45◦ for a 200 mm gap and 1:5 m high chimney, and the air%ow rate is about 45% higher
than that for a vertical chimney at otherwise identical conditions. It was found that the prediction method available in the literature can
substantially overpredict the air%ow rate for the chimney geometry investigated in this work, especially for vertical chimneys with large
gaps. The main reason for the overprediction of air%ow rate was shown due to the underestimation of the pressure losses at the chimney
outlet by using loss coe?cients obtained for normal forced %ows.
? 2003 Elsevier Science Ltd. All rights reserved.
Keywords: Natural ventilation; Solar chimney; Experimental study; Temperature measurement; Velocity measurement
0360-1323/03/$ - see front matter ? 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0360-1323(03)00057-X
894 Z.D. Chen et al. / Building and Environment 38 (2003) 893 – 906
Nomenclature
A channel cross-sectional area (m2 ) Tave average air temperature (◦ C)
Ain chimney inlet area (m2 ) Taverage average air temperature inside the chimney
Aout chimney outlet area (m2 ) (◦ C)
B buoyancy %ux (m4 =s−3 ) Tw wall surface temperature (◦ C)
cin inlet pressure loss coe?cient u local airspeed (m/s)
cout outlet pressure loss coe?cient u1 characteristic air velocity
Cp air speciNc heat capacity at ambient temperature w chimney width (m)
(J=kg ·◦ C) y distance from the heated wall (m)
Dh hydraulic diameter of the chimney channel (m)
Greek symbols
f friction factor for the channel wall
g gravitational acceleration rate (m=s2 ) chimney inclination angle
Gr ∗ modiNed Grashof number air expansion coe?cient (1= ◦ C)
H chimney height (m) momentum boundary layer thickness (m)
h height along the chimney (m) t thermal boundary layer thickness (m)
Pr Prandtl number of air QPL total pressure loss (Pa)
Q air%ow rate in the chimney (m3 =s) QPS stack pressure (Pa)
q heat %ux (W=m2 ) air kinetic viscosity (m2 =s)
T local temperature (◦ C) air density at ambient temperature (kg=m3 )
Tamb ambient temperature (◦ C)
room or by evaporative cooling of the incoming air, thermal chimneys were installed and tested by Khedari et al. [20]
comfort in buildings may be achieved using solar chimneys in the same building. Although the resulting air change rate
in buildings. per hour in the test room was high (8–15), the air movement
It was also shown by Bouchair et al. [1] and Bouchair [8] induced by these solar chimneys was still too low (average
that there is an optimum chimney width (about one-tenth of 0:04 m=s) for thermal comfort of occupants with an indoor
the chimney height) at which a maximum ventilation %ow temperature of about 35 –37◦ C. This conclusion appears to
rate can be achieved. Further increase in the chimney gap be in agreement with the observations of Barozzi et al. [3]
results in a decrease in the air%ow rate due to the occurrence who investigated a solar-chimney-based ventilation system
of back %ow at the outlet of the chimney. Numerical mod- for buildings using a 1:12 small-scale model in which the
eling by Gan and Ri9at [13] conNrmed the existence of the roof performs as a solar chimney.
optimum chimney gap-to-height ratio. Recently, small-scale Moshfegh and Sandberg [15] investigated buoyancy-
modeling by Spencer et al. [30] and numerical simulations driven air movement behind photovoltaic panels in order
by Chen and Li [31] for solar chimneys with uniform wall to cool down the photovoltaic cells. The system has the
heat %ux also demonstrated the existence of an optimum same principle as a solar chimney with uniform heat on
chimney gap-to-height ratio. It is also shown that the op- the sun-facing wall, while the other walls remain unheated.
timum gap-to-height ratio is dependent on the chimney Experimental results revealed that for input heat %ux equal
inlet design. A large inlet size can result in a large optimum to or greater than 200 W=m2 , dependent on the surround-
gap-to-height ratio due to the delay of the occurrence of ing wall emissivity, up to 30% of the heat input may be
reverse %ow. transferred to the otherwise unheated wall via radiation. It
Di9erent designs of roof solar collectors were studied by was also found that an increase in the surrounding surface
Khedari et al. [10]. The roof solar collectors acted as solar emissivity increases the air%ow rate due to the increased
chimneys to induce natural ventilation into houses. It was proportion of heat transferred to the unheated wall. By vary-
shown that to achieve the maximum ventilation %ow rate, ing the inclination angle of the chimney channel, Sandberg
the optimum dimension of the collector is 1 m long, tilted and Moshfegh [16] showed that inclination of the chimney
at 30◦ and with a collector gap of 140 mm. Khedari et al. decreases ventilation %ow rate with the same heat %ux.
[19] and Hirunlabh et al. [23] implemented di9erent roof Kumar et al. [14] studied indoor air quality in a prototype
solar collector designs and found that when using roof so- house with a solar chimney system, and showed that pas-
lar collectors alone, there is little potential to induce su?- sive outdoor air ventilation is e9ective in reducing indoor
cient airspeed to satisfy occupant comfort in hot climates. air contaminants. By comparing the performance of a con-
In order to increase airspeed in a test room, two roof so- ventional brick solar chimney and a solar chimney with the
lar collectors together with three di9erent types of solar sun-facing wall replaced by glazing, Afonso and Oliveira
Z.D. Chen et al. / Building and Environment 38 (2003) 893 – 906 895
[21] showed that the glazed solar chimney drew about %ow rate and the experimental results reported by Bouchair
10 –20% more air through the chimney. [8]. Bansal et al. [5,7], AboulNaga [11], and AboulNaga
and Abdrabboth [22] used an essentially similar approach
for %ow rate predictions induced by a solar chimney with
1.2. Numerical modeling
uniform wall temperature.
Considering heat balance, Sandberg [17] obtained an ex-
Barozzi et al. [3] used a two-dimensional computational
pression for the air%ow rate through a rectangular channel
%uid dynamics (CFD) method in the investigation of the
with uniform wall heat %ux. It was shown that the predicted
air movement inside their 1:12 building model. Although
air%ow rate was in good agreement with the experimental
the CFD model used was considered to be primitive (it did
results for a rectangular channel with one wall heated at uni-
not account for turbulence and three-dimensional e9ects),
form heat %ux and a channel gap-to-height ratio of 1:28. A
reasonable agreement was reported between experimental
similar approach was also used by Fath [6] for the design
temperature and velocity measurements and numerical pre-
of a natural draft solar fan with uniform wall heat %ux.
dictions.
It is noted that the above %ow rate prediction methods
A three-dimensional CFD program with a standard k–
are based on the assumption of uniform temperature distri-
turbulence model was used by Awbi and Gan [4] for mod-
butions across the same vertical height. As proved by Awbi
eling air movement in solar chimneys. Good agreement was
[9] and Sandberg [17], these prediction methods are appli-
achieved for the air%ow rate between numerical predictions
cable to chimneys with a small gap-to-height ratio of less
and the experimental results obtained by Bouchair [8]. A
than or close to 1:10. For wide chimneys, the assumption
similar approach has been used by Gan [12] and Gan and
of uniform temperature distribution may no longer be valid.
Ri9at [13] for numerical simulation of air movement in so-
Furthermore, the occurrence of reverse %ows near the chim-
lar chimneys.
ney outlets for wide chimneys may also make these theoret-
Moshfegh and Sandberg [15] investigated air movement
ical predictions inadequate. Another embedded assumption
behind photovoltaic panels. Their two-dimensional numeri-
of these prediction methods is that the pressure loss coef-
cal model adopted a standard k– turbulence model and wall
Ncients at the chimney inlet, outlet and along the chimney
function. In addition, radiation e9ects were also considered
channel can be evaluated with the corresponding data for
in their numerical model. Predicted air velocity and tem-
forced %ows. To date, no study has been undertaken to ver-
perature distributions were shown to be in good agreement
ify this assumption. Consequently, further investigations are
with their experimental results. Rodrigues et al. [18] inves-
necessary to clarify the conditions under which these pre-
tigated a two-dimensional solar chimney numerically with a
diction methods can be conNdently applied.
standard k– turbulence model and wall function. Details of
In this work, experiments were carried out using a sim-
the velocity and temperature Neld inside the solar chimney
ple solar chimney experimental model with a uniform heat
were reported.
%ux on one chimney wall. Air temperature and air%ow rates
In these previous numerical investigations, simulations
for di9erent chimney gaps, heat %uxes and di9erent chim-
were carried out with a computational domain deNned by
ney inclination angles were measured to provide further
the two solar chimney walls and the chimney inlet (or the
understanding of the ventilation performance of solar
room inlet) and outlet in order to minimize the computa-
chimneys. Experimental results are compared with the pre-
tion e9orts. In reality, contraction and expansion and thus
dictions based on heat balance analysis, and discrepancies
pressure losses occur at the chimney inlet and outlet, respec-
between the theoretical assumptions and the experimental
tively. In order to account for chimney inlet pressure loss,
results are discussed.
Awbi and Gan [4] applied an e9ective discharge coe?cient
at the chimney inlet. The e9ect of this departure from real-
ity on the overall air%ow rate prediction is still not clear and
further investigation is needed. 2. Analysis
where h is the height along the chimney, w is the chimney correct evaluations of the average air temperature and the
width, q is the heat %ux, Q is the air%ow rate in the chim- pressure loss coe?cients are essential for an accurate pre-
ney, and Cp are air density and speciNc heat capacity at diction of the air%ow rate induced by solar chimneys.
ambient temperature, respectively, Taverage is the average air
temperature inside the chimney at the height of h, and Tamb
is the ambient temperature. 3. Experiment
Then, the stack pressure, QPS , can be obtained by the
following integration: Fig. 1 shows the schematic view of the experimental sys-
H tem. The experimental solar chimney has internal dimen-
(Taverage − Tamb )g cos
QPS = dh sions of 1:5 m high, 0:62 m wide and a variable chimney
0 Tamb
gap from 100 to 600 mm. The chimney was heated with a
H
qhwg cos qwgH 2 cos uniform heat %ux on one wall only. The experiments were
= dh = carried out inside a 5 m wide ×7 m long ×3:9 m high exper-
0 QCp Tamb 2QCp Tamb
imental bay which is part of a large air-conditioned space.
BH cos Two sides of the bay were the solid brick walls of the large
= ; (2)
2Q space and the experimental bay was separated from the re-
where is the chimney inclination angle from vertical, H maining by heavy curtains to avoid the in%uence of any air
is the chimney height, and B is the buoyancy %ux: movement caused by ventilation openings outside the bay.
gqwH There were no ventilation openings arranged inside the ex-
B= : perimental bay. The inlet of the chimney was 1 m above the
Cp Tamb
%oor.
The pressure loss along the air path, QPL , may be expressed As shown in Figs. 2 and 3, the heated surface of the so-
as lar chimney was composed of two 0:12 mm thick, 305 mm
(Q=Ain )2 (Q=Aout )2 wide and 1:46 m long stainless-steel shims connected in se-
QPL = cin + cout ries at the bottom of the two shims by a 20 × 20 × 615 mm3
2 2
copper bar. A Hewlett Packard 6671A System DC power
H (Q=A)2 supplier was used to apply a constant low voltage (¡ 8 V)
+f ; (3)
Dh 2 on the shims to simulate uniform solar radiation on the
where A is the channel cross-sectional area, Ain and Aout heated wall. In order to avoid direct contact, a 5 mm gap was
are the inlet and outlet areas, f is the friction factor for the left between the two shims. The heated surface was insu-
channel wall, cin and cout are the inlet and outlet pressure lated with 100 mm Nberglass to reduce the heat loss through
loss coe?cients, and Dh is the hydraulic diameter of the the heated wall to the ambient air.
chimney channel. The two side walls of the chimney were made of 1500 mm
By balancing the stack pressure (Eq. (2)) and the pres- high, 850 mm wide and 6 mm thick plexiglas. The surface
sure losses along the air path (Eq. (3)), the ventilation %ow opposite the heated shims (the front wall) was 1500 mm
rate, Q, for a chimney with a uniform wall heat %ux can be high, 620 mm wide and 3 mm thick plexiglas. This front
obtained as follows: wall facing the heated surface is movable to achieve di9erent
1=3 chimney gaps. The two side walls as well as the front wall
B cos
Q=A ; (4) were insulated with 50 mm expanded polystyrene to reduce
2
heat losses. As shown in Fig. 2, the chimney channel can
where be tilted anticlockwise, which is similar to a solar chimney
2
A H 1 A
2
A with the sun-facing wall replaced by glazing.
= f + cin + cout : (5) As shown in Fig. 3, 13 T-type thermocouples were di-
H 2Dh 2 Ain Aout
rectly soldered onto the back of the stainless-steel shims to
For the pressure loss coe?cients, a general approach is to measure the local temperatures of the heated surface. The air
resort to the available data for normal forced %ows. For the temperature inside the chimney channel was measured by
case of a rectangular channel with both ends open and heated 7 T-type thermocouples arranged on a stretched bicycle
on a single wall, Sandberg [17] used cin = 1:5, cout = 1:0 brake wire which can be placed at any position inside the
and f = 0:056, respectively. Eq. (4) has been validated by chimney. Temperatures of the insulation materials behind
Sandberg [17] using a rectangular channel with one wall the heated shims and the unheated front wall were also mea-
heated at a uniform heat %ux and a chimney gap-to-height sured to estimate the heat losses through the chimney walls,
ratio of 1:28. as well as the radiation heat transferred from the heated
From the above analysis, it is seen that the air%ow rate shims to the unheated front wall.
is mainly determined by two aspects: (a) the stack pressure Thermocouples were calibrated to within ±0:2◦ C at
built up in the chimney and (b) the pressure losses at the 0 C, 40◦ C and 80◦ C before being soldered onto the
◦
inlet, outlet and along the chimney channel. Consequently, shims, while all the other thermocouples were calibrated
Z.D. Chen et al. / Building and Environment 38 (2003) 893 – 906 897
Computer
controlled rail TSI8455 air
velocimeter
Heavy
curtains
Thermocouples
Computer
Solar chimney + –
Computer controlled HP 6671A DC
data logger power supplier
rate may be higher than the real value. Fortunately, the re-
verse %ow velocity at the probe position was very small and
1203
with uniform heat %ux at 200, 300, 400, 500 and 600 W=m2 .
2
70 143cm, 600W/m , 20cm Gap
2
With a 400 W=m2 uniform heat %ux, experiments were 60 143cm, 600W/m , 40cm Gap
(T - Tamb )
also performed with chimney gaps of 100, 200, 300, 400 50
the outlet of the chimney, are not uniform, with higher air 0
temperatures near the heated surface. 0 5 10 15 20 25
It was also observed that the temperature rose when ap- Distance from heated wall (cm)
proaching the opposite surface of the heated shims. This Fig. 6. Temperature distributions across chimney gaps at 297, 750 and
temperature rise is caused by the radiation from the heated 1430 mm above the chimney inlet for a 45◦ inclined chimney with
shims. From the measured surface temperature of the front 200 mm gap and 400 W=m2 heat input.
wall opposite the heated shims, the radiation is estimated to
be less than 10% of the heat input on the heated shims and
decreases with a decrease in the heat input.
4.2. Temperature distributions across the chimney width
50
side the chimney as well as on the unheated front wall are
40
reasonably uniform across the chimney width. During the
30
experiments, up to 4:4◦ C di9erence was observed for the
20
10
temperature measurements on the heated shims at the same
0
height. However, compared with the temperature di9erence
0 10 20 30 40 50 between the heated shims and the ambient air, which is be-
Distance from heated wall (cm) tween 55◦ C and 75◦ C for the chimney with a 400 mm gap
Fig. 4. Temperature distributions across chimney gaps at 523 mm above
and 400 W=m2 heat input, the variation in temperatures on
the chimney inlet for vertical chimney at di9erent heat inputs and chimney the heated wall across the chimney width is still relatively
gaps. small.
Z.D. Chen et al. / Building and Environment 38 (2003) 893 – 906 899
80 1500
75cm, 400W/m , 40cm Gap, 50 mm from the side wall 2cm from the heated surface
40 75cm, 400W/m2, 40cm Gap, 155 mm from the side wall 6cm from the heated surface
143cm, 400W/m2, 40cm Gap, 50 mm from the symmetrical plane 600 16cm from the heated surface
30 2 38cm from the heated surface
143cm, 400W/m , 40cm Gap, 50 mm from the side wall
2
On the wall opposite the heated surface
20 143cm, 400W/m , 40cm Gap, 155 mm from the side wall
300
10
0
0
0 10 20 30 40 50 60 70 80
0 10 20 30 40 50
(T - Tam b)
Distance from heated wall (cm)
Fig. 7. Temperature distributions across chimney gaps at 297, 750 and Fig. 9. Vertical temperature distributions at 1, 2, 6, 16, 38 cm from the
1430 mm above the chimney inlet, 50 mm away from the chimney heated surface and wall temperatures for vertical chimney at 400 mm
symmetrical plane, 50 mm away from one side wall and 155 mm away gap and 400 W=m2 heat input.
from the other side wall, respectively, for vertical chimney at 400 mm
gap and 400 W=m2 heat input.
1500
2
1200 400W/m , 20cm Gap
2
900 400W/m , 40cm Gap
On the heated surface 2
600W/m , 20cm Gap
1cm from the heated surface 2
900 600W/m , 40cm Gap
2cm from the heated surface 600
6cm from the heated surface
12cm from the heated surface
600
16cm from the heated surface 300
On the wall opposite the heated surface
300
0
0 5 10 15 20
0
(Tave - Tam b)
0 10 20 30 40 50 60 70 80
(T - Tam b)
Fig. 10. Average air temperature along the chimney height for di9erent
heat inputs and chimney gaps.
Fig. 8. Vertical temperature distributions at 1, 2, 6, 12, 16 cm from the
heated surface and wall temperatures for vertical chimney at 200 mm
gap and 400 W=m2 heat input.
higher than that away from the walls. This increase in the
airspeed near the heated wall along the chimney height may
4.3. Temperature distributions along the chimney height enhance the heat transfer near the top of the chimney.
Another possible reason for the increase in the local heat
Figs. 8 and 9 show the temperature distributions along transfer coe?cient along the heated surface may be due
a vertical chimney with a 200 mm chimney gap and to the transition between laminar %ow and turbulent %ow.
400 W=m2 uniform heat input, and a 400 mm chimney For a vertical plate heated by a uniform heat %ux, this
gap and 400 W=m2 uniform heat input, respectively. It is transition occurs at 2 × 1013 ¡ Gr ∗ Pr ¡ 1 × 1014 , where
seen that the air temperatures inside the chimney generally Gr ∗ = gqh4 =k2 . For the heat %ux between 200 and 600 W,
increase along the chimney height. However, the tempera- the transition starts at the height of around 2 m. For a solar
tures on the shims and on the surface opposite the shims do chimney channel investigated in this work, this transition is
not increase linearly and even drop above the middle of the likely to be earlier due to the air disturbances in the chimney
chimney height. Similar vertical temperature distributions inlet. It is possible that both facts may exist, which results
have also been reported by Sandberg and Moshfegh [32]. in the high heat transfer coe?cient at the top of the chimney
Considering the gradual increase in the temperature of the channel.
air inside the channel along the chimney height, this de- Fig. 10 shows the average air temperature along the chim-
crease in the heated surface temperature can only be caused ney height for four di9erent chimney gaps and heat inputs.
by an increase in the local heat transfer coe?cient along The average is taken according to Eq. (6) across the same
the heated surface. chimney height:
Smoke visualizations showed that, near the top of the
chimney, the air at the center moves towards the heated (T − Tamb ) ds
Tave = A + Tamb : (6)
walls, and the air velocity near the heated walls is much A
900 Z.D. Chen et al. / Building and Environment 38 (2003) 893 – 906
1 1
0.9 2 0.9 2
52.3cm, 400W/m , 20cm Gap 143cm, 400W/m , 20cm Gap
0.8 52.3cm, 400W/m2, 40cm Gap 0.8 2
143cm, 400W/m , 40cm Gap
2
0.7 0.7 143cm, 600W/m , 20cm Gap
2
52.3cm, 600W/m , 40cm Gap 2
0.6 0.6 143cm, 600W/m , 40cm Gap
2
0.5 0.5 Boundary Theory, Eq. (8), 400W/m
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
0 10 20 30 40 50 0 10 20 30 40 50 60
Distance from heated wall (cm) Distance from heated wall (cm)
Fig. 11. Dimensionless temperature distributions across the chimney gap Fig. 12. Dimensionless temperature distributions across the chimney gap
at 523 mm above the chimney inlet for vertical chimneys at di9erent heat at 1430 mm above the chimney inlet for vertical chimneys at di9erent
inputs and chimney gaps. heat inputs and chimney gaps.
1500
The average air temperature, Tave , obtained by Eq. (6) is
Distance From Chimney Inlet (mm)
responsible for the buildup of stack pressure inside a so- 1200
lar chimney. It should be noted that the average tempera-
ture deNned by Eq. (6) is di9erent from that obtained from 900
2
2cm, 400W/m , 20cm Gap
2
heat balance analysis, i.e. Taverage in Eq. (1), which can be 2cm, 400W/m , 40cm Gap
2
2cm, 600W/m , 20cm Gap
demonstrated by the following equation: 600 2
2cm, 600W/m , 40cm Gap
qhw (T − Tamb )u ds
Taverage = + Tamb = A + Tamb ; (7) 300
QCp Q
where u is the local airspeed parallel to the chimney wall. 0
From Eqs. (6) and (7), it is seen that when either the 0 0.1 0.2 0.3 0.4
airspeed or the temperature is uniform across the chimney (T - Tam b)/(Tw all - Tam b)
gap, the average air temperature obtained from heat balance Fig. 13. Dimensionless air temperature distributions along the chimney
analysis is the same as the average air temperature across height at 20 mm away from the heated surface for chimneys with di9erent
the chimney height obtained from Eq. (6). Consequently, heat inputs and chimney gaps.
although the assumption of uniform temperature across the
chimney gap may not be valid for certain chimneys, the 1500
However, if both the temperature and velocity distributions 900 6cm, 600W/m2, 20cm Gap
2
6cm, 600W/m , 40cm Gap
are highly non-uniform across the chimney gap, the heat
balance analysis may not properly predict the stack pressure 600
Velocity (m/s)
110cm, 600W/m , 20cm Gap
0.5
essentially identical at the same position inside the chim-
ney, especially within 100 mm from the heated surface. For 0.4
%ux, Eckert and Jackson [33] showed that for fully devel- 0.2
oped turbulent %ow, the dimensionless temperature and 0.1
velocity distributions depend on the thermal boundary
layer thickness only. Based on their analysis, the following 0.0
0 5 10 15 20
equations can be obtained [34]: Distance from heated wall (cm)
1=7
T − Tw y Fig. 15. Velocity distributions across the chimney gap for vertical chim-
= 1− ; (8)
Tw − Tamb t neys with 200 mm gap and 200, 300 and 600 W=m2 heat input.
u y 1=7 y 4
= 1− (9)
u1
0.7
and
0.6 110cm, 400W/m2, 10cm Gap
t = F(Pr); (10)
2
110cm, 400W/m , 20cm Gap
2
0.5 110cm, 400W/m , 30cm Gap
2
Velocity (m/s)
110cm, 400W/m , 40cm Gap
∗−1=14
2
−1=2 2=3 1=14
= 0:505Gr Pr (1 + 0:445Pr ) h; (11) 0.4 110cm, 400W/m , 60cm Gap
Boundary Theory, Eq. (9)
0.3
5=14
u1 = 4:35 Gr ∗ Pr −1=6 (1 + 0:445Pr 2=3 )−5=14 ; (12)
h 0.2
gqh4 0.1
Gr ∗ = ; (13)
k2 0.0
0.8 0.8
Velocity (m/s)
0.5 60 Degree 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0.0 0.0
0 5 10 15 20 0 10 20 30 40 50 60 70
Distance from heated wall (cm) Distance from the sidewall (cm)
Fig. 17. Velocity distributions across the chimney gap for vertical chim- Fig. 19. Velocity distributions across the chimney width at 15, 25, 75,
ney for a chimney with 200 mm gap and 400 W=m2 heat input at the 135 and 175 mm away from the heat surface for a vertical chimney with
inclination angles of 0◦ , 15◦ , 30◦ , 45◦ and 60◦ , respectively. 200 mm gap and 500 W=m2 heat input.
0.8 0.07
Experimental
0.7 0.06
Air Flow Rate (m3/s)
Predicted by Eq. (4)
0.6 0.05
Velocity (m/s)
0.5
0.04
0.4
0.03
0.3
0.02
0.2
0.01
0.1 1cm 1.5cm 2.5cm 7.5cm 9.5cm
0.00
0.0
0 10 20 30 40 50 60 70 100 200 300 400 500 600 700
2
Distance from the sidewall (cm) Heat Flux (W/m )
Fig. 18. Velocity distributions across the chimney width for a vertical Fig. 20. Comparison of predicted and experimental air%ow rate through
chimney with 100 mm gap and 400 W=m2 heat input. chimneys with 200 mm gap and 200, 300 400, 500 and 600 W=m2 heat
input.
0.10 1500
0.08 1200
Air Flow Rate (m /s)
0.06 900
2
400W/m , 20cm Gap, vertical chimney
2
0.04 600 400W/m , 20cm Gap, 45 degree
Predicted, vertical chimney
Predicted, 45 Degree inclined chimney
0.02 300
0.00 0
0 20 40 60 80 0 5 10 15 20
Fig. 21. Comparison of predicted and experimental air%ow rate through Fig. 23. Comparison of the measured average air temperature along the
chimneys with 100, 200, 300, 400, 600 mm gaps and 400 W=m2 heat chimney height and the predicted average air temperature using heat
input. balance analysis for the vertical chimney and for the 45◦ inclination
chimney, both with 200 mm gap and 400 W=m2 heat input.
0.06
Fig. 23 compares the average air temperature along the
0.05
chimney height for the vertical chimney and the 45◦ in-
Air Flow Rate (m /s)
clined chimney with the same gap of 200 mm and heat input
3
0.04
of 400 W=m2 . It is seen that the average air temperatures
0.03 inside the vertical chimney and the inclined chimney are
0.02
essentially the same. Considering the reduced stack height
Experimental
Predicted by Eq. (4)
of the inclined chimney, it is clear that the stack pressure for
0.01 the 45◦ inclined chimney should be much smaller than that
of the vertical chimney. Thus, the reduction of pressure loss
0.00
must be the main cause for %ow rate increase with inclined
0 20 40 60 80
chimneys. Compared with vertical chimneys, the velocity
Inclination Angle (Degree)
distributions across the chimney gap in inclined chimneys
Fig. 22. Comparison of predicted and experimental air%ow rate through are much more uniform and this may signiNcantly reduce
chimneys with 200 mm gap and 400 W=m2 heat input at the inclination the inlet and outlet pressure losses.
angles of 0◦ , 15◦ , 30◦ , 45◦ and 60◦ , respectively. Experiments on the chimney tilted clockwise could not be
carried out in this work due to the limitation of the chimney
support frame. When a chimney is inclined clockwise, the
the chimney gap, the reverse %ow enhanced and was further boundary layer may be more closely attached to the heated
down into the chimney channel. Due to the simultaneous in- surface. Combined with the reduction in the stack height,
crease in the chimney inlet size and thus less pressure loss a reduction in the air%ow rate is expected with clockwise
at the chimney inlet, an increase in the chimney gap does inclinations, as shown by Sandberg and Moshfegh [32].
not result in a reduction in the air%ow rate, although reverse Using Eq. (4), predictions were made for the air%ow rate
%ow does increase. However, if the inlet size remains the in chimneys investigated in this work. As shown in Fig. 21,
same when the chimney gap is increased, the air%ow rate with a decrease in the chimney gap, the predicted air%ow rate
may be reduced and an optimum chimney gap may exist, as becomes close to the experimental result, and the predicted
reported by Bouchair [1] and other authors [13,30,31]. %ow rate for the smallest chimney gap of 100 mm and heat
Fig. 22 shows the air%ow rate through the chimney with input of 400 W=m2 is quite acceptable. This conNrms that
di9erent inclination angles. It can be seen that the air%ow the prediction method can be applied to narrow chimneys.
rate peaks at around 45◦ which is about 45% higher than that As shown in Fig. 22, for the 45◦ and 60◦ inclined chimneys
for the vertical chimney. Even at 60◦ , the air%ow rate is still with a 200 mm gap, the predicted air%ow rates were also
about 30% higher than that for the corresponding vertical found to be in reasonable agreement with the experimental
chimney. results.
Considering that the stack height is decreased for inclined From the temperature and airspeed measurements, it has
chimneys, the signiNcant increase in the air%ow rate for large been shown that the temperature is reasonably uniform
inclination angles is interesting. Flow rate increase can be across the chimney width, but is strongly non-uniform
due by two reasons, i.e. increased stack pressure and reduced across the chimney gap. For chimneys with a small gap
pressure losses. of 100 mm and for 45◦ and 60◦ inclined chimneys with a
904 Z.D. Chen et al. / Building and Environment 38 (2003) 893 – 906
200 mm gap, the airspeed distributions in the chimney are to be around 31 compared to 2.7 for normal forced %ow
relatively uniform, as shown in Figs. 16 and 17. Conse- through a sharp opening [34]. Consequently, it is believed
quently, the average temperature across the chimney height that the large pressure loss coe?cient compared to normal
obtained from heat balance analysis (Eq. (1)) is close to forced %ow at the chimney outlet can be the main reason for
that obtained by Eq. (6), which is responsible for the stack the overprediction of the air%ow rate by using Eq. (4) for
pressure inside the solar chimney. Further, the relatively wide chimney designs. It is noted that no speciNc investiga-
uniform air%ow inside the chimney may still approximately tions have been reported on the pressure loss coe?cient for
resemble normal forced %ow. Consequently, reasonable %ows induced by natural convection along a heated channel
%ow rate predictions were obtained by Eq. (4) for the narrow and further research is needed.
chimney and the chimneys with large inclination angles. The present study suggests that the air%ow rate prediction
However, air%ow rates are substantially overpredicted by methods based on heat balance analysis and normal forced
Eq. (4) for vertical chimneys with large chimney gaps. Ex- %ow pressure loss coe?cients can be applied to chimneys
perimental results for vertical chimneys with large gaps with the following two conditions:
show that both air temperature and speed inside chimneys
are highly non-uniform and the velocity proNle has a maxi- • The airspeeds or the air temperature distributions in the
mum very close to the heated wall, as shown in Fig. 16. In chimney are reasonably uniform, which ensures a correct
this situation, the average temperature predicted by the heat evaluation of the stack pressure by heat balance analysis.
balance analysis (Eq. (1)) is di9erent from that obtained by • The airspeed distributions at the chimney inlet and
Eq. (6). Further, highly non-uniform air velocity distribu- outlet resemble normal forced %ows, which ensures a
tions inside the chimney channel suggest that the air%ow in proper evaluation of pressure losses from normal forced
a wide chimney is essentially conNned to a narrow channel %ows.
close to the heated wall and thus results in high-pressure
losses through the air path compared to normal forced %ow These two conditions are normally fulNlled for narrow
at the same air%ow rate inside an identical channel. chimneys and the air%ow rate can be reasonably predicted
In Fig. 23, the predicted average air temperature by heat using heat balance analysis and normal forced %ow pressure
balance analysis (with the measured air%ow rate and heat loss coe?cients, as proved by Awbi [9] and Sandberg [17].
input) for the vertical chimney and the 45◦ inclined chim- However, for wide chimneys, especially with small inlets
ney, both with a 200 mm gap and 400 W=m2 heat input, and large outlets, the above two conditions are no longer
were included for comparison with the measured average air fulNlled. At extreme situations, reverse %ow can occur, and
temperature along the chimney height. It was found that for the velocity and temperature distribution inside the chimney
the vertical chimney, the stack pressure is overpredicted by can be strongly non-uniform. Consequently, the %ow rate
around 30%. Due to the relatively uniform airspeed distri- prediction method based on heat balance analysis and nor-
butions, the predicted average air temperatures for the 45◦ mal pressure loss coe?cients can signiNcantly overpredict
inclined chimney are much closer to the measured average the air%ow rate as demonstrated in this work. Further inves-
air temperatures compared to the predictions for the vertical tigations are needed to develop a simple prediction method
chimney. for the air%ow rate induced by a solar chimney for which the
Considering that the air%ow rate is proportional to the prediction method based on the heat balance analysis and
square root of the stack pressure (refer to Eqs. (2) and(3)), normal forced %ow pressure loss coe?cients are inadequate.
overpredicting the stack pressure by 30% for the vertical It should be noted that there are no clear lines between
chimney may overpredict the air%ow rate by 14%. Since the narrow/wide chimneys and small/large chimney open-
%ow rate for the vertical chimney is actually overpredicted ings. From Eq. (11), the boundary layer thickness for a
by around 86% by Eq. (4), as shown in Fig. 20, pressure vertical plate of 2 m high with a uniform heat %ux of
losses through the air path may have been signiNcantly un- 400 W=m2 is around 1/15th of the plate height. As an indi-
derestimated. Since pressure loss along the chimney wall cation, for single-wall-heated solar chimneys, those with a
is usually much smaller than those at the inlet and outlet, gap-to-height ratio greater than 1:10 may be considered to
underestimation of pressure losses should occur at the inlet be wide, and for two wall-heated solar chimneys, those with
and outlet. a gap-to-height ratio greater than 1:5 may be considered to
Considering that at the chimney outlet, hot air is rising be wide.
through cold surrounding ambient air, a closely related phe-
nomenon might be the air%ow through a sharp horizontal
opening dividing between two chambers with warm air in 5. Conclusions
the bottom chamber and cold air in the top one. For a sharp
horizontal opening dividing two air layers with di9erent air Solar chimneys with a uniform heat %ux on a single
temperatures, it has been shown that reverse %ow occurs wall were investigated experimentally for di9erent chimney
when the %ow rate reduces to a certain low level. The pres- gaps, heat %ux inputs and di9erent chimney inclinations. The
sure loss coe?cient when reverse %ow occurs was found chimneys investigated in this work cover a gap-to-height
Z.D. Chen et al. / Building and Environment 38 (2003) 893 – 906 905
ratio from 1:15 to 2:5. It is shown that by changing the [5] Bansal NK, Mathur R, Bhandari MS. Solar chimney for enhanced
chimney gap while maintaining all the other conditions, the stack ventilation. Building and Environment 1993;28(3):373–7.
air%ow rate increases continuously with increasing chimney [6] Fath HES. Development of a natural draft solar fan for ventilation
of greenhouses in hot climates. International Journal of Solar Energy
gap, even up to the gap-to-height ratio of 2:5—no optimum 1993;13:237–48.
gap has been found. This might be caused by a decrease in [7] Bansal NK, Mathur R, Bhandari MS. A study of solar chimney
the inlet pressure loss due to the simultaneous increase of assisted wind tower system for natural ventilation in buildings.
the chimney inlet by extending the chimney gap. Building and Environment 1994;29(4):495–500.
Results also showed that the air%ow rate reached a max- [8] Bouchair A. Solar chimney for promoting cooling ventilation
in southern Algeria. Building Service Engineering, Research and
imum at a chimney inclination angle of around 45◦ for a Technology 1994;15(2):81–93.
200 mm gap and 1.5 m high chimney, which is about 45% [9] Awbi HB. Design consideration for natural ventilated buildings.
higher than that for a vertical chimney under otherwise iden- Renewable Energy 1994;5:1081–90.
tical conditions. The reason for this %ow rate increase is [10] Khedari J, Hirunlabh J, Bunnag T. Experimental study of a roof
shown to be due to the relatively even airspeed inside the solar collector towards the natural ventilation of new houses. Energy
and Buildings 1997;26:159–64.
chimney, which signiNcantly reduces the pressure loss at [11] AboulNaga MM. A roof solar chimney assisted by cooling cavity
the chimney inlet and outlet compared to the corresponding for natural ventilation in buildings in hot arid climates: an
vertical chimney. energy conservation approach in Al-Ain city. Renewable Energy
With the chimney geometry investigated in this work, 1998;14:357–63.
temperature and velocity distributions are found to be es- [12] Gan G. A parametic study of Trombe walls for passive cooling of
buildings. Energy and Buildings 1998;27:37–43.
sentially uniform across the chimney width. However, the [13] Gan G, Ri9at SB. A numerical study of solar chimney for
temperature distributions, as well as the airspeed inside a natural ventilation of buildings with heat recovery. Applied Thermal
chimney, are highly non-uniform across the chimney gaps, Engineering 1998;18:1171–87.
especially for vertical chimneys with large chimney gaps. [14] Kumar S, Sinha S, Kumar N. Experimental investigation of solar
The air%ow rate prediction method available in the lit- chimney assisted bioclimatic architecture. Energy Conversion and
Management 1998;39(5/6):441–4.
erature is based on two basic assumptions, i.e. heat bal- [15] Moshfegh B, Sandberg M. Flow and heat transfer in the air gap
ance analysis gives the correct stack pressure and normal behind photovoltaic panels. Renewable and Sustainable Energy
forced %ow pressure loss coe?cients can be applied to solar Reviews 1998;2:287–301.
chimneys. It is shown that this prediction method can sub- [16] Sandberg M, Moshfegh B. Ventilated solar roof air%ow and heat
stantially overpredict the air%ow rate for certain chimney transfer investigation. Renewable Energy 1998;15:287–92.
[17] Sandberg M. Cooling of building integrated photovoltaics by
geometries, especially vertical chimneys with wide chim- ventilation air. In: Proceedings of HybVent Forum ’99, First
ney gaps. The validity of the two basic assumptions was International One-day Forum on Natural and Hybrid Ventilation, The
checked against the experimental velocity and temperature University of Sydney, Darlington, New South Wales, Australia, 28
measurements. It was found that for large chimney gaps, the September 1999. p. 10 –8.
non-uniform temperature and airspeed caused certain devia- [18] Rodrigues AM, Canha da Piedade A, Lahellec A, Grandpeix JY.
Modelling natural convection in a heated vertical channel for room
tion between the predicted average air temperature, and thus ventilation. Building and Environment 2000;35:455–69.
the stack pressure based on heat balance analysis and those [19] Khedari J, Boonsri B, Hirunlabh J. Ventilation impact of a solar
actually inside a solar chimney. However, the main reason chimney on indoor temperature %uctuation and air change in a school
for the overprediction of the air%ow rate is the underestima- building. Energy and Buildings 2000;32:89–93.
tion of the pressure losses at the chimney outlet by using [20] Khedari J, Mansirisub W, Chaima S, Pratinthong N, Hirunlabh J.
Field measurements of performance of roof solar collector. Energy
loss coe?cients obtained for normal forced %ows. Further and Buildings 2000;31:171–8.
investigations are needed for the better prediction of air%ow [21] Afonso C, Oliveira A. Solar chimneys: simulation and experiment.
rates in solar chimneys, for which the prediction method Energy and Buildings 2000;32:71–9.
based on heat balance analysis and forced %ow pressure loss [22] AboulNaga MM, Abdrabboth SN. Improving night ventilation
coe?cients are not adequate. into low-rise buildings in hot-arid climates exploring a combined
wall-roof solar chimney. Renewable Energy 2000;19:47–54.
[23] Hirunlabh J, Wachirapuwadon S, Pratinthong N, Khedari J.
New conNgurations of a roof solar collector maximizing natural
References ventilation. Building and Environment 2001;36:383–91.
[24] Das SK, Kumar Y. Design and performance of a solar dryer with
[1] Bouchair A, Fitzgerald D, Tinker JA. Moving air using stored vertical collector chimney suitable for rural application. Energy
solar energy. In: Proceedings of the 13th National Passive Solar Conversion and Management 1989;29(2):129–35.
Conference. Cambridge, MA: ASES, 1988. p. 33–8. [25] Ekechukwu OV, Norton B. Design and measured performance of a
[2] Cristofalo SD, Orioli S, Silvestrini G, Alessandro S. Thermal solar chimney for natural-circulation solar-energy dryers. Renewable
behavior of “Scirocco rooms” in ancient Sicilian villas. Tunneling Energy 1997;10(1):81–90.
and Underground Space Technology 1989;4(4):471–3. [26] Pasumarthi N, Sherif SA. Experimental and theoretical performance
[3] Barozzi GS, Imbabi MSE, Nobile E, Sousa ACM. Physical and of a demonstration solar chimney model: Part I—Mathematical model
numerical modelling of a solar chimney-based ventilation system for development. International Journal of Energy Research 1998;22:
buildings. Building and Environment 1992;27(4):433–45. 277–88.
[4] Awbi HB, Gan G. Simulation of solar-induced ventilation. Renewable [27] Pasumarthi N, Sherif SA. Experimental and theoretical performance
Energy Technology and the Environment 1992;4:2016–30. of a demonstration solar chimney model: Part II—Experimental and
906 Z.D. Chen et al. / Building and Environment 38 (2003) 893 – 906
theoretical results and economic analysis. International Journal of [31] Chen ZD, Li Y. A numerical study of a solar chimney with uniform
Energy Research 1998;22:443–61. wall heat %ux. In: Proceedings of the Fourth International Conference
[28] Trombe F, Robert JF, Cabanat M, Sesolis B. Some performance on Indoor Air Quality, Ventilation and Energy Conservation in
characteristics of C.N.R.S. solar house collectors. In: Proceedings of Buildings, Hunan, China, October 2001. p. 1447–54.
the Passive Solar Heating and Cooling Conference and Workshop, [32] Sandberg M, Moshfegh B. The investigation of %uid %ow and
Alberquerque, New Mexico, 18–19 May 1976. p. 201–2. heat transfer in a vertical channel heated from one side by PV
[29] Janssen JE. The history of ventilation and temperature control. elements: Part II—Experimental study. In: Proceedings of the Fourth
ASHRAE Journal 1999;41(10):48–70. Renewable Energy Congress, Denver, CO, 1996. p. 254 –8.
[30] Spencer S, Chen ZD, Li Y, Haghighat F. Experimental [33] Eckert E, Jackson T. Analysis of turbulent free-convection boundary
investigation of a solar chimney natural ventilation system. In: layer. NACA Technical Note No. 2207, 1950.
Proceedings of RoomVent 2000, Seventh International Conference [34] Cooper LY. Combined buoyancy and pressure-driven %ow through
on Air Distribution in Rooms, Reading, UK, 9 –12 July 2000. a shallow, horizontal, circular vent. Journal of Heat Transfer
p. 813–8. 1995;117:659–67.