Available online at www.sciencedirect.

com

ScienceDirect
Solar Energy 112 (2015) 310318
www.elsevier.com/locate/solener

On the natural convection heat transfer in a rectangular passage

solar air heater
Hussain H. Al-Kayiem a,, Tadahmun A. Yassen b
a

Mech. Eng. Dept., Universiti Teknologi PETRONAS, 31750 Tronoh, Perak, Malaysia
b
Mech. Eng. Dept., Tikrit University, Tikrit, Iraq

Received 3 April 2014; received in revised form 2 October 2014; accepted 30 November 2014

Communicated by: Associate Editor Claudio Alejandro Estrada-Gasca

Abstract
This paper presents and discusses the experimental measurements versus the results from the available correlations in the literature
that are commonly used to predict the free convection heat transfer coecient between the surfaces and the owing air. The experimental
investigations were accomplished using a rectangular duct comprising a at plate-glass cover as the solar air heater with the following
dimensions: 0.48-m width  0.07-m depth  2-m length. The absorber plate was made of 1-mm gauge of aluminium. The measurements
were performed at inclination angles of 30, 50, and 70 to determine the optimum angle of the absorption-free convection mechanism in
a solar air heater. The comparison is presented and discussed in terms of the Nusselt number. The optimum inclination angle to achieve
the best collector performance was found to be 50. The results of the present work can assist in resolving the issue of predicting the free
convection in thermal passages.
2014 Elsevier Ltd. All rights reserved.

Keywords: Natural convection; Rectangular conduit; Solar air heater; Solar absorber; Solar collector

1. Introduction
Solar air heaters that make use of at-plate collectors
have been widely utilised in the technology of low-temperature solar energy. The benets of the use of solar air heaters include the rather simple design, the ease of operation,
and the low capital costs. Solar air heaters may be used for
space heating, drying and paint spraying operations
(Kalogirou, 2004; Due and Beckman, 2006). The solar
air heater occupies an important place among solar heating
systems because of the minimal use of materials and the
direct use of air as a working substance, which reduces
Corresponding author.

E-mail
addresses:
hussain_kayiem@petronas.com.my
Al-Kayiem), tadamun_77@yahoo.com (T.A. Yassen).
http://dx.doi.org/10.1016/j.solener.2014.11.031
0038-092X/ 2014 Elsevier Ltd. All rights reserved.

(H.H.

the number of required system components (Yeh and Ho,

2009). The eciency of these collectors in air heating is
low due to the low convection heat transfer coecients
between the absorber and the owing air that increases
the absorber plate temperature, resulting in higher heat
losses to the ambient atmosphere.
However, selecting a suitable equation to predict the free
convection in a rectangular solar air heater remains
unclear.
Maiti et al. (2011) designed and developed an indirect,
natural convection batch-type solar dryer tted with
northsouth reectors. These reectors enabled an
enhancement of the collector eciency. This enhancement
ranged from 40% to 58.5%. This system was in India in the
month of January under conditions of peak solar irradiation. The equation proposed by Tiwari (2002) was used

H.H. Al-Kayiem, T.A. Yassen / Solar Energy 112 (2015) 310318

to perform the calculation of the Nu between the absorber

plate and the air passing inside the collector.
A mathematical model of a single-pass, double-duct
solar air heater was described by Forson et al. (2003).
The model is able to predict the incident solar radiation,
the coecients of the heat transfer, the mean air temperature, the mean air ow rates and the relative humidity at
the exit. The authors also presented the results of a parametric study. The factor that was dominant in the determination of the overall heater eciency was shown to be the
air mass ow rate. They used the equation suggested by
Hollands et al. (1976) to predict the heat transfer coecient
between the absorber plate and the air ow inside the collector and the convective heat transfer coecient between
the absorber plate and the glass cover.
Jain (2007) studied the new concept of a solar crop dryer
having a reversed absorber plate-type collector and thermal
storage with natural airow. A reversed absorber plate
(1 m in length by 1 m in width) with a 0.15-m packed bed
was able to dry 95 kg of onions. In a drying period of
24 h, the moisture content decreased from 6.14 to 0.27 kg
water/kg of dry matter. The author used the equation suggested by Hollands et al. (1976) to calculate the Nusselt
number between the absorber plate and the uid ow.
A transient mathematical model was presented for a single-pass at plate solar air heater by El-Sebaii and Al-Snani
(2010). This model described a solar air heater with a single-pass at plate. An analytical solution of the energy balance equations for dierent components of the heater was
the basis of the model. The temperature of the owing
air was assumed to vary in the ow direction only. A computer simulation using the climatic conditions of Jeddah
(lat. 21420 N, long. 39110 E), Saudi Arabia, was used to
study the performance of the heater. The measurements
were collected for the heater possessing a black painted
absorber plate under the weather conditions of Tanta,
lat. 30470 N (Egypt), and were compared with the simulated results to validate the proposed mathematical model.
The proposed model was found to be able to produce an
accurate prediction of the outlet ow air temperature.
The authors used the equation suggested by Hollands
et al. (1976) to predict the heat transfer coecient between
absorber plate and air ow inside the collector, the heat
transfer coecient between glass cover and air ow inside
collector and the heat transfer coecient between the rst
glass cover and the second glass cover.
A solar air collector was studied experimentally by Zhai
et al. (2005a). The air channel had a width of 0.5 m and a
length of 1.5 m, while the variable air channel gap ranged
from 0.1 m to 0.5 m. During the experimentation, three
electric heating plates were used to represent the solar radiation, which aected the uniform heat ux along the air
channel. The temperature distribution of the air and the
rate of the induced natural air ow depended highly on
the inclination angle, heat input and channel gap. The
experimental results indicated that the optimum inclination
angle was 45. At this inclination, the maximum rate of the

311

natural ventilation could be achieved. A suitable channel

length of approximately 1 m was found, beyond which
the obtained heat and the natural ventilation rate cannot
be augmented markedly. The authors used the equation
suggested by Hollands et al. (1976) to calculate the Nusselt
number between the absorber plat and uid ow.
Pakdaman et al. (2011) presented an experimental investigation to evaluate the dierent thermal features of a natural-convection at-plate solar air heater with a
longitudinal rectangular ns array. A Nusselt number correlation was introduced for such nned duct devices after
the thermal performance of the system was determined.
The solar radiation and ambient temperature were considered to be independent parameters in the empirical model.
Other features of the scheme, such as the dierent dimensionless variables, the plates and outow temperatures,
the eciency, and the mass ow rate, were empirically
modelled based on these variables. This model provided
recommendations of design concepts and rules of thumb
and validated the calculations of the design parameters.
Solar radiation was determined to be the key parameter
that characterised the thermal behaviour of the system.
This nding was based on an order-of-magnitude analysis.
Furthermore, the optimum conditions to achieve the highest performance of the system were determined through an
exergy analysis. To evaluate the Nusselt number between
the absorber plate of the solar collector and the air passing
through it, the authors used the equation proposed by BarCohen and Rohsenow (1984).
Bracamonte and Baritto (2013) studied the optimum
aspect ratio for at plate solar collector for air heating.
They calculated the heat transfer coecient using Kays
correlation, as suggested by Yeh and Lin (1996) in the form
of Nu = 0.0158 Re0.8. The characteristic length is the
hydraulic diameter of the cross section.
For application of solar air heater in humidication
dehumidication desalination, Summers et al. (2012) mathematically modelled a solar air heater integrated with PCM
TES. For the thermal analysis in the air ow zone, they
assumed xed value for the convective heat transfer
between the absorber plate and air, and between the cover
glass and air. Then they assumed this thermal resistance as
sheet of metal. They adopted this approach, as they
reported, to avoid the complexity in computational power.
The argument her that the convection heat transfer is inuenced by the air velocity, the surfaces temperatures which
are varying according to the solar intensity. However, they
have not mentioned the correlations used to evaluate the
constant heat transfer coecients.
One realized issue with solar air heaters is that the inclination angle of free convection in a rectangular duct solar
air heater is very important to the heat transfer coecient
between the absorber plate and the air ow inside the
channel.
The eect of the inclination of the absorber on the
airow rate was investigated by Mathur et al. (2006) in
a solar-induced ventilation system using a Roof Solar

312

H.H. Al-Kayiem, T.A. Yassen / Solar Energy 112 (2015) 310318

Chimney concept. Due to the higher altitude of the sun

during the summer months, more solar radiation was captured by the absorber at a smaller inclination relative to the
horizontal, but the system suers from a reduction in the
stack height. The results of the developed solution indicate
that the best absorber inclination ranged from 40 to 60
depending on the latitude of the location. An inclination
angle of 45 was found to be optimal for maximizing the
ventilation rate in Jaipur, India. The rate of ventilation at
this inclination was approximately 10% higher in comparison with the rates at inclinations of 60 and 30.
The reported results of the prediction of the natural convection using the suggested correlations for the channel
ow are inconsistent. During the course of a solar air heater evaluation, the use of the correlations can lead to the
introduction of error. Additionally, some of the correlations are incorrectly used to evaluate the heat transfer coefcient between the absorber and the air ow.
This paper presents an eort to address the issue of the
selection of the proper relationship for the convective heat
transfer in the absorptionconvection heat transfer process, which is typically used in the solar air heater. The
work includes experimental investigation and manipulation
of the measurement data in terms of the Nu  Ra format.
The experimental results were then compared with the predicted results from the most commonly used correlations in
the topic. In addition, the work was extended to investigate
the optimum inclination angle for the absorptionconvection heat transfer mechanism in a solar air heater.
2. Test rig conguration
The experiment investigated the at plate absorber
inclined rectangular duct solar air heater shown in Fig. 1.
The absorber plate was made from an aluminium plate
with a gauge of 1 mm. The internal dimensions of the rectangular duct of the test rig were 0.48-m width  0.07-m
depth  2-m length, with a hydraulic diameter,
Dh = 0.1222 m. The bottom and side walls of the duct were
made of 20-mm-thick wood. To minimise the top losses, a
double glass cover was used to cover the rectangular duct.
Each cover was 3 mm thick, and the gap in between the
components was 5 mm. The supporters used to support
the rectangular passage had an adjustable lock to ensure
that the rectangular passage could be oriented or tilted to
any desired angle with respect to the horizontal base, in this
case: 30, 50, and 70.
2.1. Instrumentation and experimentation
The experiment was performed in the Solar Site
Research (SSR) at the University Technology Petronas,
Perak, Malaysia. The collection of the data for each tilt
angle of the collector was repeated for 4 days to improve
the reliability of the result by taking the average reading.
The data collected every two hours began at 8 am and
ended at 6 pm. The data collection was performed from

(a)

Ambient Temp.
and wind speed
Air out Temp
and velocity

Solar meter
Velocity sensor
Temperature sensor

Collector

Air in

Data logger

(b)
Air out

Perspex
cover

Play wood frame

Air in

insulated at the
back and sides
Absorber plate
painted black

Fig. 1. Experimental test rig of a rectangular duct solar air heater. (a)
Schematics and (b) setup.

early November until the end of December, 2012. A

calibrated thermocouple type K (Chromal Alumal) capable
of measuring temperatures in the range of (50 to
1000 C) with an accuracy of 0.1 C was used to measure
the relevant temperatures. Thermocouple wires, which
were connected to a data logger type GL820-UM-851, were
placed at three points along the length of the surfaces of the
inner and outer glazing covers as well as the absorber plates
to measure their respective mean wall temperatures. The
thermocouple wires were also placed in the ambient atmosphere and at the outlet of the collector to measure the outlet air temperature. A SCHMIDT Flow Sensor SS 20.260
with a range of readings from 0 to 50 m/s and an accuracy
5% of the measured value +0.4% of the nal value with a
minimum reading of 0.02 m/s was used to measure the
velocity of the air at the collector outlet. A solarimeter type
SL200 with an accuracy of 5% of the measurement was
used to measure the angular solar insolation at the solar
site for each interval.

H.H. Al-Kayiem, T.A. Yassen / Solar Energy 112 (2015) 310318

313

T2

2.2. Experimentally measured Nusselt number

The coecient of natural convection heat transfer is
appropriate for characterising the experimental measurement; hexp is obtainable from Nuexp value.
Nuexp

T1

hexp  L
QL



k
Ab k T b  T fm

The experimental heat gain from the collector, Q was

calculated as:
_ p T o  T i
Q mC

where m_ qVA: mass ow rate of air (kg/s), V: outlet air

velocity (m/s), A: channel cross-sectional area (m2) at the
outlet, Cp: specic heat of air (kJ/kg C), To: outlet air temperature, and Ti: inlet air temperature.
And
Ab: base absorber plate area (m2).
L: gap between the absorber plate and the transparent
cover (m).
o
T fm T i T
, and T b : absorber temperature.
2
3. Predicted Nusselt number by correlation
The thermal mechanisms in the solar collectors are complicated and sensitive to the inclination of the absorber, the
gap geometry between the absorber and cover, and the
thermo-physical conditions of the working uid. The thermal energy balance in the natural solar air involves convection heat transfer from the absorber to working uid,
radiation heat transfer from the absorber to the cover,
and losses by convection and radiation from the cover to
the ambient, as well as losses from the sides and the back
to the surrounding, which is commonly negligible by using
insulation. The main issue facing proper prediction of the
thermal process is the calculation of the convective heat
transfer coecient from the absorber to the working uid.
The experimental and mathematical evaluation of the convective heat transfer in terms of Nu number are investigated and analysed three recommended correlations.
The coecient of convective heat transfer in the inclined
conduit conguration, similar to that shown in Fig. 2,
could be obtained from predicted Nusselt number, the
Nuc value. The conguration is an open ow solar air heater. The three commonly used correlations for such conguration are considered in this work and discussed below.

T1>T2

Fig. 2. Inclined rectangular cavity.

0 < h < 70. The results were expected to be indistinguishable from the results for the horizontal layer if one replaces
Ra by Ra  cos h. Signicant dierences from the expected
behaviour were observed, particularly in the range of
1708 < Ra  cos h < 104 and 30 < h < 60. The Nusselt
number, given as function of Ra cos h and h, in the form
of Eq. (2), was the recommended relationship for air as
the medium between the plate and the cover.
This equation is based on Fig. 2 for the inclined rectangular cavity of 0 < h < 70 and H/L P 12:
"

!!#

1708
sin 1:8h1:6
 1  1708
Nuc 1 1:44 1 
Ra  cos h
Ra  cos h
"
#
13
Ra  cos h
1
2

5830

The formula suggested by Hollands et al. (1976), which

is referred to here as the Holland correlation, is used by
many researchers, e.g., Forson et al. (2003), Zhai et al.
(2005b), Jain (2007), and El-Sebaii and Al-Snani (2010),
to evaluate the heat transfer coecient between the absorber plate and the draft of air owing in the collector. The
application of this correlation is thought to be incorrect
because the correlation was established for a closed cavity
to evaluate the heat transfer from the hot surface, i.e., the
absorber plate to the cold surface, i.e., the transparent
cover of the collector.
However, Ong (1995) and Tiwari (2002) used the correlation of Hollands et al. to determine the convective heat
transfer between the absorber and the cover to calculate
the top heat loss from the collector.

3.1. Hollands et al. correlation

3.2. Bar-Cohen and Rohsenow correlation
An experiment was performed by Hollands et al. (1976)
to investigate the free convective heat transfer rate via the
air layers inside an inclined cavity with a high aspect ratio
when the heat originated from below. The Rayleigh number ranged from subcritical to 105; the range of the angles
of inclination, h, measured from the horizontal, was

Bar-Cohen and Rohsenow (1984) proposed an equation

for parallel plates in a vertical channel. These authors
claimed that the correlation can be used to calculate Nu
for an inclined channel by replacing g by g  cos h for the
calculation of the Grashof number, Gr.

314

H.H. Al-Kayiem, T.A. Yassen / Solar Energy 112 (2015) 310318

"
Nuc

C1
Ra  L=H

C2

#1=2

Ra  L=H 1=2

4.1. Analysis of natural convection

3

where C1 = 144 for isothermal plates and C2 = 2.87 for

adiabatic plates, H: length of the rectangular channel.
The Bar-Cohen and Rohsenow (1984) formula was used
by Pakdaman et al. (2011) and DeBlois et al. (2013) to calculate the heat transfer coecient between the absorber
plate and the air ow inside the collector channel.
3.3. Tiwari correlation
For an inclined plate with the heated surface facing
upward, the empirical correlation becomes complicated
for angles between 15 and 75 from the horizontal.


Nuc 0:14 Ra1=3  Rac1=3
1=4

0:56Ra  cos h

105 < Ra  cos h < 1011

where
Rac is the critical Rayleigh number, which is equal to
1708 inside a rectangular cavity (Incropera et al., 2005).


gb T b  T fm L3
Ra
 Pr
5
m2
Maiti et al. (2011) used the equation suggested by Tiwari
(2002), referred to here as the Tiwari correlation, to calculate the heat transfer coecient between the absorber plate
and the air ow inside the collector.
4. Results and discussions
The three correlations are named without the reporting
year to simplify the referencing during the discussion as
below:
Two goals are targeted in this work. The rst goal is to
determine the most suitable empirical correlation to predict
the natural convection in a rectangular solar air heater, and
the second goal is to determine the optimum inclination
angle of the collector for solar air heating. The results are
presented in terms of the Nu values. The experimental
measurements were performed to evaluate the Nu number
over the range of 60,000 < Ra < 280,000 and the inclination angle range of 30 < h < 70. The experiments were
performed to evaluate the Nu number using the three
correlations discussed above. The experimental and the
theoretical Nu values are presented and compared in the
next section.
To estimate the dierences between the predicted and
measured Nu values, the percentage of relative error
(PRE) is calculated as:
RPE %

Nuc  Nuexp
 100
Nuc

Figs. 35 show the values of the predicted and the measured results of Nu  Ra for inclination angles 30, 50 and
70, respectively. Each gure consists of the measured and
the predicted Nu values using the Holland correlation, the
Tiwari correlation, and the Bar-Cohen correlation.
Fig. 3 indicates that the calculated Nuc using the Tiwari
correlation is closer to the experimental Nu compared to
the results predicted by the Hollands et al. and the
Bar-Cohen correlations. The predicted results of Nuc by
the Tiwari correlation follow the trend of the experimental
Nuex with almost the same dierence over the entire range
of 3  104 < Ra < 26  104. The Tiwari correlation overestimates the Nu value by approximately 20% over the tested
range of Ra. The predicted values of Nuc by the Hollands
and the Bar-Cohen correlations are slightly changing with
the change of Ra, while the predicted Nuc values by the
Tiwari correlation are changing considerably as Ra
changes, in the same manner as the experimental results.
As the Ra value increases, the gap between the
experimental results of Nu and the calculated Nuc using
the Hollands and Bar-Cohen correlations increases, and
the error of the underestimation becomes signicant.
Table 1 presents the percentage of relative error (PRE) in
the Nu number at low, medium and high Ra numbers
of the prediction by the three correlations and the
experimental results.
For the solar air heater set at a 50 inclination angle, the
experimental and predicted results of Nu as a function of
Ra are shown in Fig. 4. As in the case of the 30 inclination, the Tiwari correlation exhibits a closer agreement to
the experimental results compared to the other correlations. At approximately 6  104 of Ra number, the values
of the predicted Nuc by the Bar-Cohen and the Hollands
correlations are close to the experimental measurement
results with lower estimations of the Nu values, while the

Fig. 3. Comparison of the calculated Nusselt number by the dierent

correlations with the experimental Nusselt number, determined at a 30
inclination angle.

H.H. Al-Kayiem, T.A. Yassen / Solar Energy 112 (2015) 310318

315

Table 1
Percentages of error in Nuc values at an inclination angles 30, 50, and
70 for the results of the dierent correlations compared to the
experimental measurement results.
Ra

Fig. 4. Comparison of the Nusselt number from the experimental results

with the theoretical Nusselt numbers from dierent correlations, determined at the 50 inclination angle.

Fig. 5. Comparison of the experimental and the theoretical Nusselt

numbers at inclination angle of the collector of 70.

Tiwari correlation predicted Nuc values higher than the

experimental data. As the Ra number increases, the prediction by the Tiwari correlation becomes closer to the measurement results, while the prediction of Nuc by the
Hollands and Bar-Cohen correlations deviate considerably
from the experimental results. For Ra numbers exceeding
2  105, the predicted Nuc values using the Hollands and
Bar-Cohen correlations become signicantly lower than
the experimental results.
At Ra of approximately 2.75  105, the value of RPE is
5.7%, 282%, and 230% when Nuc is predicted by the
Tiwari, Holland et al., and Bar-Cohen correlations, respectively. Further calculations of the deviation between the
predicted and measured results of Nu at the 50 inclination
are presented in Table 1.

PRE (%)
Tiwari

Hollands

Bar-Cohen

At 30
104
1.67  105
2.15  105

+21.4
+15.5
+11.9

189
243
252

155
183
199

At 50
6.5  104
1.75  105
2.75  105

41.7
24.0
5.7

71.4
200.0
282

55.7
155.3
230

At 70
5.1  104
1.98  105
2.4  105

43.7
14.8
2.5

133
266
334

82.5
215
259

Fig. 5 displays the results of measured and predicted

results using the Tiwari, Bar-Cohen, and Hollands correlations when the solar air heater is tilted at a 70 inclination
angle. The trend of the results is similar to the cases of the
30 and 50 inclinations, where the Nuc values determined
by the Tiwari correlation overestimate the measurement
results, while the prediction results by the Hollands and
Bar-Cohen correlations underestimate the measurement
results. Additionally, as the Ra number increases, the difference between the Tiwari correlation results and the measurement results becomes smaller, while the predicted
results by the Hollands and Bar-Cohen correlations exhibit
high error compared to the measurement results. For a Ra
number of approximately 2.5  105, the prediction results
by the Tiwari correlation are in high agreement with the
measurement results. In contrast, the predicted values of
Nuc are far below the measurement results for the other
two correlations.
Table 1 presents the RPEs for the results at the inclination angle of the collector of 70. Over the entire range of
Ra numbers, the predicted results by Tiwari exhibit the
minimum error. At the high range of Ra numbers, at
approximately 2.4  105, the RPE values for the Nu number are 2.5%, 334%, and 259% when using the Tiwari,
Hollands et al. and Bar-Cohen correlations, respectively.
The negative sign in the RPE indicates that the predicted
values of Nuc are less than the measurement results.
Within the tested range of the Ra number, the general
trend of the results demonstrate that the Bar-Cohen and
Hollands correlations are not sensitive to the change in
Ra number; in contrast, the Tiwari correlation predicts
Nu number values with higher accuracy as the Ra number
increases. Additionally, it is clear for the entire range of
collector inclinations that the Bar-Cohen and Hollands
correlations are underestimating the experimental Nu number values, while the prediction by the Tiwari correlation
overestimates the experimental results, with a reduction
in the error as the Ra increases from the lower tested range
to the higher tested range.

316

H.H. Al-Kayiem, T.A. Yassen / Solar Energy 112 (2015) 310318

The closer evaluation of Nu by Tiwari compared to the

experimental measurement can be justied by the contribution of inclination term, h in his correlation. Also, the closer agreement as Ra is increasing is also due to presence of
the absolute Ra without multiplication with geometry factor, as in the case of Hollands et al. and Bar-Cohen
correlations.

and is higher in the up drafting. In addition, the heat transfer rate is even lower because of the lower solar insolation.
Fig. 7 shows the comparison between the eciency of
the collector with the inclination angles of 30, 50, and
70 over the solar day time. The eciency was calculated
as,
g

4.2. Analysis of the inclination angle

Q
I rr  Ac

To achieve the second objective of the paper, further

analyses were performed to investigate the optimum inclination angle for the combined absorptionconvection heat
transfer mechanism. Fig. 6 illustrates the transient behaviour of the solar air heater in terms of the Nu  time at
30, 50, and 70 inclination angles. The displayed results
are the mean of repeated measurements over ve days to
reduce the uncertainty. From the gure, the highest Nu
number is observed at an inclination angle of 50, followed
by 30 and 70. In some cases, the wind has an impact at
low inclination angle congurations, such as 30. This
impact of wind is due to the air being able to freely travel
inside the channel when the air passage is more parallel to
the direction of the wind itself; as a result, the wind
displaces more hot air with the ambient temperature, which
results in a lower heat transfer rate. Another reason is that
the low inclination angle results in a low up-drafting force,
which leads to a low heat transfer rate. A lower angle conguration may be good in terms of absorbing solar radiation, but poor in transferring heat. In contrast, for higher
inclination angle congurations, such as 70, because the
absorber plate is not facing the solar radiation source,
the solar insolation is lower compared to that of the other
congurations. However, because it is approximately perpendicular to the direction of the wind, the hot air remains
inside the channel or is less displaced by the ambient air

where Irr is the solar irradiation (W/m2), and Ac is the area

of the collector (m2). The results indicate that the eciency
of collector at the tilt angle of 50 is higher than the eciency of the collector at the tilt angles of 30 and 70.
The eciency of the collector at the tilt angle of 50 at
noon is 2.6% and 3.6% higher than the eciency of the collector at tilt angles 30 and 70, respectively.
Therefore, the best conguration in this case, as presented by the graphs below, is the 50 inclination angle
conguration because it has the best position for higher
solar radiation absorption and while providing lower displacement of hot air by the incoming wind and the high
up drafting force. These results are in agreement with those
reported by Mathur et al. (2006), who claimed that the
optimum absorber inclination at any place varies from
40 to 60 depending upon the latitude.
The justication of getting best performance for natural
convection in closed conduit, like the one studied in this
work, at around 50 inclination is the interrelation of three
mechanisms. The phenomena consist of radiation conversion, convection heat transfer from the absorber to the particles of the working uid, and the path of motion of the
particles. If the angle is lower than 40 then the particles
moves up and hit the cover, which reduces the kinetic
energy of the working uid and reduce the resulted mass
ow rate in the system. When the inclination angle is higher
than 60, then the particles are moving up with low

Fig. 6. Comparison of the experimental Nusselt number at inclination

angles 30, 50, and 70 over the measurement period of the day.

Fig. 7. Comparison of the eciency at dierent inclination angles 30,

50, and 70.

H.H. Al-Kayiem, T.A. Yassen / Solar Energy 112 (2015) 310318

resistance and the residence time of the particles in the

conduit is less. This will reduce the heat transfer from the
absorber to the working uid. Combination of the radiation absorption, the convection heat transfer from absorber to the uid, and the updraft nature of the particle
motion is proved to be optimum around 50 inclination.
4.3. Experimental uncertainty analysis
An uncertainty analysis is a vital part of any experimental measurement program to support the condence level in
the results. Uncertainty may originate from causes such as
an inaccuracy in the measurement equipment, random variation in the measured quantities, and approximations in
the data-reduction relations.
The method suggested by Wheeler and Ganji (2004) is
adopted in this investigation. In a general case, consider
R to be a function of n measured variables; x1, x2, . . .,
xn; that is:
R f x1 ; x2 ; . . . ; xn

A small change dR in R is due to small changes dxi in the

xi that is expanded through the dierential equation
n
@R
@R
@R X
@R
dR dx1
dx2
   dxn

dxi
8
@x1
@x2
@xn
@xi
i1
In statistical form, this became:
n
X
@R
wR
wxi
@xi
i1

maximum percentages of error in the three sets are 2.7%,

5.1% and 4.1% for data sets of 30o, 50o, and 70o,
respectively.
5. Conclusions
Three dierent correlations of natural convection available in the literature and widely used in the prediction of
the Nu number by many researchers are the correlations
of Bar-Cohen and Rohsenow, Hollands et al. and Tiwari.
The three correlations are compared with the experimentally obtained results to justify their use in the open solar
air heaters. Because the inclination angle is a contributor
factor in the absorptionconvection mechanism, it is used
in the present investigation. The following can be concluded from the analysis of the results:
 Over the tested range of the Ra number, the Tiwari correlation overestimates the Nu number, while the BarCohen and Hollands correlations underestimate the
Nu number.
 As Ra number increases, the error in the results by the
Tiwari correlation is reduced, while the errors become
considerably large in the results obtained by the BarCohen and Hollands correlations.
 The optimum inclination angle of the collector is
approximately 50 to obtain a high Nusselt number
and the highest collector eciency.

In the present experiment

Nu f T i ; T o ; T b ; V

317

10

The percentage of error in each value of the measured

Nu number is shown in Fig. 8. The data is segregated into
three groups corresponding to three angles. The margin of
error is acceptable, where the maximum error is 5.1%. The
mean percentage of error in the entire data is 4.2%. The

Future studies should consider the eect of the gap

between the absorber and the cover. Additionally, the generation of a new or a modied correlation that is more
accurate in predicting the absorption-natural convection
phenomena should be pursued.
Acknowledgements
The research is carried out under the Malaysian Ministry
of Education (MOE) research fund under PRGS scheme,
and for that the authors are expressing their gratitude.
The authors, also, acknowledge Universiti Teknologi
PETRONAS for providing the technical and nancial
assistance in conducting this research work mainly the
Tuition Fees Assistance for the second author.
References

Fig. 8. The uncertainties in the measured Nu number values.

Bar-Cohen, A., Rohsenow, W.M., 1984. Thermally optimum spacing of

vertical natural convection cooled parallel plate. J. Heat Transfer 106,
116123.
Bracamonte, J., Baritto, M., 2013. Optimal aspect ratio for nonisothermal at plate solar collectors for air heating. Sol. Energy 97,
605613.
DeBlois, J.C., Bilec, M.M., Schaefer, L.A., 2013. Design and zonal
building energy modelling of a roof integrated solar chimney. Renew.
Energy 52, 241250.
Due, J.A., Beckman, W.A., 2006. Solar Engineering of Thermal
Processes, third ed. Wiley, New York, USA.

318

H.H. Al-Kayiem, T.A. Yassen / Solar Energy 112 (2015) 310318

El-Sebaii, A.A., Al-Snani, H., 2010. Eect of selective coating on thermal

performance of at plate solar air heaters. Energy 35, 18201828.
Forson, F.K., Nazha, M.A.A., Rajakaruna, H., 2003. Experimental and
simulation studies on a single pass, double duct solar air heater.
Energy Convers. Manage. 44, 12091227.
Hollands, K.G.T., Unny, T.E., Raithby, G.D., Konicek, L., 1976. Free
convective heat transfer across inclined air layers. Trans. ASME, J.
Heat Transfer 98, 189193.
Incropera, F.P., Dewitt, D.P., Bergman, T.L., Lavine, A.S., 2005.
Introduction to Heat Transfer, fth ed. John Wiley & Sons, New
York.
Jain, D., 2007. Modeling the performance of the reversed absorber with
packed bed thermal storage natural convection solar crop dryer. J.
Food Eng. 78, 637647.
Kalogirou, S.A., 2004. Solar thermal collectors and applications. Prog.
Energy Combust. Sci. 30, 231295.
Maiti, S., Patel, P., Vyas, K., Eswaran, K., Ghosh, P.K., 2011.
Performance evaluation of a small scale indirect solar dryer with
static reectors during non-summer months in the Saurashtra region of
western India. Sol. Energy 85, 26862696.
Mathur, J., Mathur, S., Anpuma, 2006. Summer-performance of inclined
roof solar chimney for natural ventilation. Energy Build. 38, 1156
1163.
Ong, K.S., 1995. Thermal performance of solar air heaters: mathematical
model and solution procedure. Sol. Energy 55 (2), 93109. http://
dx.doi.org/10.1016/0038-092X(95)00021-I.

Pakdaman, M.F., Lashkari, A., Tabrizi, H.B., Hosseini, R., 2011.

Performance evaluation of a natural convection solar air-heater with
a rectangular-nned absorber plate. Energy Convers. Manage. 52,
12151225.
Summers, E.K., Antar, M.A., Lienhard, J.H.V., 2012. Design and
optimization of an air heating solar collector with integrated phase
change material energy storage for use in humidication-dehumidication desalination. Sol. Energy 86, 34173429.
Tiwari, G.N., 2002. Solar Energy, Fundamentals, Design, Modelling and
Applications. Narosa Publishing House, New Delhi, India.
Wheeler, A.J., Ganji, A.R., 2004. Introduction to Engineering Experimentation, Second ed. United State of America.
Yeh, H.M., Ho, C.D., 2009. Solar air heaters with external recycle. Appl.
Therm. Eng. 29, 16941701.
Yeh, H., Lin, C., 1996. The eect of collector aspect ratio on the collector
eciency of upward-type at-plate solar air heater. Energy 21 (10),
843850.
Zhai, X.Q., Dai, Y.J., Wang, R.Z., 2005a. Experimental investigation on
air heating and natural ventilation of a solar air collector. Energy
Build. 37, 373381.
Zhai, X.Q., Dai, Y.J., Wang, R.Z., 2005b. Comparison of heating and
natural ventilation in a solar house induced by two roof solar
collectors. Appl. Therm. Eng. 25, 741757.