www.elsevier.com/locate/renene
Technical note
Heat transfer dynamics in an inatable-tunnel
solar air heater
A. Flores-Irigollen
a,
, J.L. Fernandez
b
, E. Rubio-Cerda
c
,
F.T. Poujol
a
a
Departamento de Ingeniera en Pesqueras, Universidad Auto noma de Baja California Sur, La Paz, BCS,
23080, Mexico
b
Instituto de Ingeniera, Universidad Nacional Auto noma de Me xico, Mexico City 04510, Mexico
c
CIBNOR, Apdo. Postal 128, La Paz, BCS 23000, Mexico
Received 1 November 2003; accepted 15 November 2003
Abstract
A mathematical model that describes the dynamics of the heat transfer in an inatable-
tunnel solar collector for air heating is proposed and validated. The model is distributed-
parameters, one-dimensional and unsteady-state. It considers the thermal inertia of a pebble
bed acting as the absorber surface and is constituted by three equations that describe the
temperature distributions of the three system components: polyethylene cover, transfer uid
(air) and absorber surface.
To solve the governing equations, a novel numerical scheme that diers from the standard
method of nite dierences in the form of generating the discretization equations is pro-
posed. In this scheme, the dimensionless versions of the equations are reduced to linear
canonical forms of rst order and then are solved analytically in small spatial domains to
produce discretization equations in an explicit form.
To validate the quality of the present model, some experimental tests in a 50 m long
inatable-tunnel solar collector were carried out. Results of the model compare favorably
with experimental results.
# 2003 Elsevier Ltd. All rights reserved.
Keywords: Inatable-tunnel; Mathematical model; Simulation; Numerical scheme; Validation; Solar air
heater
p;i
q
1
p
1
T
p;i
q
1
p
1
_ _
Expp
1
Dh 22
An identical equation for the uid temperature is reached if Eq. (16) is solved ana-
lytically over the Dx interval, assuming the cover and absorber temperatures are
A. Flores-Irigollen et al. / Renewable Energy 29 (2004) 13671382 1376
constant. Thus:
T
f ;i1
q
2
p
2
T
f ;i
q
2
p
2
_ _
Expp
2
DX 23
The outlet uid temperature of the ith segment becomes the inlet temperature for
the i 1th segment. This explicit equation allows calculation of the outlet uid
temperature in each segment. A complete algorithm that includes the use of
Eqs. (3), (22) and (23) was implemented in a computer program.
5. Comparison with experiments
To validate the mathematical model, the implemented program that solves the
governing equations was run under realistic initial and boundary conditions. The
predictions of the outlet uid temperatures were compared with the measured tem-
peratures.
For the numerical simulations in the present model, the physical properties of
the uid and pebbles were evaluated at the average temperature of the collector.
The delity with which the model reproduces the experimental temperatures was
evaluated by obtaining the average absolute deviation, dened by:
d
1
Dt
_
Dt
0
T
calc
T
exp
dt 24
where the absolute dierence between calculated and experimental temperatures is
averaged over the duration of the experiment [13].
Fig. 4. Discretization of the solution domain.
1377 A. Flores-Irigollen et al. / Renewable Energy 29 (2004) 13671382
The accuracy of this model was also judged with a statistical criterion that con-
siders the instantaneous relative dierences of temperatures, that is:
d
r
1
Dt
_
Dt
0
T
calc
T
exp
T
exp
_ _
2
dt
: 25
The results of the numerical calculation are based on the convective heat transfer
coecient, h
2
, regarded as an adjustable parameter to t the data. This is the para-
meter of major uncertainty. It is nonetheless worth mentioning that the multiplier
constant of Eq. (10) was modied only once to calculate the tting value of h
2
for
an experimental data set, and kept without change for the simulation of the other
case.
Some experimental tests were carried out in the climatic conditions of La Paz
(24.15
v
north latitude and 110.36
v
west longitude), Mexico, in September 2001.
Because the data demonstrated repeatability, only results from the rst two tests
are presented here.
All the tests began approximately at the same time, at 7:30 a.m. ZT, and these
had a duration of 24 h. Since the tests were carried out during the same month, the
solar day length did not vary signicantly, as can be seen in the radiation curves
(dashed line) shown in Figs. 5 and 7. The ow regime during the tests was highly
turbulent (Re > 10; 000), indicated in Table 1, which additionally presents the aver-
age ow velocity, mass ow rate and the model deviations.
Tests 1 and 2 were carried out on 22 and 25 September 2001, respectively.
Fig. 5. Comparison of the experimental and theoretical outlet uid temperatures for the rst test.
A. Flores-Irigollen et al. / Renewable Energy 29 (2004) 13671382 1378
The results of tests 1 and 2 are plotted in Figs. 5 and 7. These graphs present the
evolution of the experimental inlet and outlet uid temperatures. The simulations
results are shown with a continuous line. These reproduce the experimental ther-
mal behavior appropriately as it is appreciated in the graphs. This agreement is
Fig. 6. Correlation between experimental and calculated outlet uid temperature (test 1).
Fig. 7. Comparison of the experimental and theoretical outlet uid temperatures for the second test.
1379 A. Flores-Irigollen et al. / Renewable Energy 29 (2004) 13671382
conrmed with the estimation of the temporary average of the temperatures absol-
ute dierences (OFT
calc
OFT
exp
) and also by Eq. (25) (average relative devi-
ation). Thus for our model the estimated deviation was barely 0.7
v
C, which
corresponds to a relative deviation of 2.5% for the rst test, and 0.9
v
C (2.8%) for
the second one.
Even though the mean absolute deviation d for the present model is 0.7
v
C,
errors up to 1.7
v
C were mainly observed 1.5 h after noon. One possible cause for
this deviation is the underestimation of the convective heat transfer coecient of
the pebbles to the uid. Further work on this topic will be reported at a future
time. Another possible cause for this deviation is an overestimation of the absorber
thermal capacitance that produces a slight phase shift of the theoretical tempera-
ture curve relative to the experimental one (see Fig. 5).
The dashed line corresponds to the solar radiation expressed in W/m
2
and
whose value is read on the secondary scale. The maximum of radiation occurs 6.5
h after the experiment was started, at approximately 14:00 h, and it reaches a value
of 860 W/m
2
.
Table 1
Model deviations
Test m (m/s) m (kg/s) Re d (
v
C) dr (%)
1 1.29 4.36 88,600 0.7 2.5
2 2.14 7.24 147,000 0.9 2.8
Fig. 8. Correlation between experimental and calculated outlet uid temperature (test 2).
A. Flores-Irigollen et al. / Renewable Energy 29 (2004) 13671382 1380
The eect of the pebble bed thermal inertia is another relevant aspect observed
in Figs. 5 and 7. This eect is evident after sunset, when the temperature dierence
between outlet and inlet is 3.5
v
for the rst test and 3.0
v
for the second one. Parti-
cularly for the rst test, this dierence falls gradually and it reaches 1
v
over the last
4 h. According to the theoretical results (continuous line), the model represents this
eect suciently well. However, in the last 5 h of the rst experiment, the outlet
uid temperature is slightly overestimated by the model.
Another way to demonstrate the model tness is by means of the correlation
graphs between measured and calculated temperatures. These correlations are plot-
ted in Figs. 6 and 8. The values of the slopes of the regression straight lines and the
correlation coecients are very close to unity, which conrms the model tness.
6. Conclusions
On the basis of the previous discussions, the following conclusions can be drawn:
A numerical model to simulate the dynamic behavior of an inatable-tunnel
solar air heater is formulated and validated. The model includes the thermal
capacitance of a pebble layer acting as absorber surface and is constituted by three
equations that describe the temperature distributions of the three system compo-
nents.
A new strategy to derive discretization equations is proposed. This is used to
resolve numerically the governing equations and it consists of analytically solving
the dimensionless equations in small spatial and temporal domains to produce
explicit discretization equations.
The average absolute deviation in all the tests was less than 1.0
v
C, which con-
rms that the model reproduces the thermal behavior of the collector acceptably
well. Relative deviations from experimental temperatures (less than 2.8%) conrm
the tness of the model.
Acknowledgements
The authors acknowledge the nancial support for the project from the Mexican
National Council for Science and Technology, (CONACYT, 31310-U) and also
express their gratitude to Miguel Angel Aldana, Mario Salas, Oscar Resendiz and
Juan Carlos Nava for their important contributions to this project.
References
[1] Prasad BN, Saini JS. Thermo-hydraulic optimization of articially roughened solar air heaters.
Proc. NSEC, Solar Energy Society of India, Hyderabad. 1988.
[2] Cortes A, Piacenti R. Improvement of the eciency of a bare solar collector by means of turbu-
lence promoters. Appl Energy 1990;36:2536.
[3] Gupta D, Solanki SC, Saini JS. Thermohydraulic performance of solar air heaters with roughened
absorber plates. Solar Energy 1997;61(1):3342.
1381 A. Flores-Irigollen et al. / Renewable Energy 29 (2004) 13671382
[4] Due JA, Beckman WA. Solar engineering of thermal processes, 2nd ed. New York: John Wiley
and Sons; 1991 Chapter 6.
[5] Yadav YP, Kumar A, Sharan LB, Srivastava VP. Parametric analysis of a suspended at plate
solar air heater. Energy Convers Mgmt 1995;36(5):32535.
[6] Aboul-Enein S, El-Sebaii AA, Ramadan MRI, El-Gohary HG. Parametric study of a solar air hea-
ter with and without thermal storage for solar drying applications. Renew Energy 2000;21:50522.
[7] Medved S, Oman J, Novak P. Numerical model and parametric analyses of an inatable solar hea-
ter. Solar Energy 1999;65(4):26370.
[8] Hegazy AA. Optimum channel geometry for solar air heaters of conventional design and constant
ow operation. Energy Convers Mgmt 1999;40:75774.
[9] Ahmad NT. Agricultural solar air collector made from low-cost plastic packing lm. Renew Energy
2001;23:66371.
[10] Coutier JP, Farber EA. Two applications of a numerical approach of heat transfer process within
rock beds. Solar Energy 1982;26(6):45162.
[11] Niles PN, Carnegie EJ, Pohl JG, Cherne JM. Design and performance of an air collector for indus-
trial crop dehydration. Solar Energy 1978;20:1923.
[12] Bejan A. Heat transfer, 1st ed. New York: John Wiley and Sons; 1993 Chapter 9.
[13] Saez AE, McCoy BJ. Dynamic response of a packed bed thermal storage system. A model for solar
air heating. Solar Energy 1982;29(9):2016.
A. Flores-Irigollen et al. / Renewable Energy 29 (2004) 13671382 1382