1 Evaluation of a static and a dynamic model to simulate the

2 operation of flat-plate solar air collectors driven by natural

3 convection
4 Fernando Bolaoa, Gabriel Merinoa, Christian Correaa, Jernimo Paredesa.
a
5 Department of Power and Machinery, Faculty of Agricultural Engineering,
6 University of Concepcin, Chilln Campus.
7ABSTRACT

8Two mathematical models were evaluated to simulate the operation of a flat-

9plate solar air collector: a static and a dynamic model. The static model does
10not include the effects of heat storage or conductive heat flow on the system
11elements, while the dynamic model includes such effects. To implement the
12static model, nonlinear equations were solved numerically by Newtons method,
13while the dynamic model was solved explicitly.

14A flat-plate solar air collector and a data acquisition system were used to
15validate the models. To this end, operating parameters, such as airflow pattern
16inside the collector, temperature at the bottom part of the collector and local
17environmental data, including ambient temperature, solar radiation and wind
18speed were measured on a summer day. The measured and recorded collector
19outlet temperatures were used to compare and evaluate the models, between
20then, and against experimental data.

21The results showed that the model that best simulated the outlet temperature of
22the fluid was the dynamic model. Even though this model had a slight tendency
23to overestimate the outlet temperature of the fluid, the root mean square error
24(RMSE) was 3.5 C. In contrast, the static model tended to underestimate fluid
25temperature and resulted in an RMSE value of 5.3 C. The average collector
26efficiency was obtained experimentally and reached a value of 2.6 %. The
27dynamic model estimated a collector efficiency of 14.9 %, with an RMSE value
28of 3.8 %, while the static model estimated a collector efficiency of 10.8 %, with
29an RMSE of 4.5 %.

30Keywords: solar air collector, mathematical modeling, numerical methods.

31

32NOMENCLATURE
number
Nu Nusselt
 Ra Rayleigh number
Tilt angle ()
dx Control volume length (m)
L Collector length (m)
W Collector width (m)
M Mass per unit area (kgm-2)
C Specific heat (kJkg-1K-1)
K Thermal conductivity (Wm-1K-1)
Thickness (m)
eP Insulating layer thickness (m)
h Separation between plates (m)
Stefan Boltzman constant (5.67 108 Wm-K-4)
G Mass flux (kgs-1)
t Time (s)
g Gravity acceleration (9.8 ms-2)
f Volumetric expansion coefficient (K-1)
f Air density (kgm-3)
Vf Kinematic viscosity
f Thermal diffusivity
T Temperature (K)
hr,p-c Radiative coefficient between absorber and cover (Wm-2K-1)
hr,cs Radiative coefficient between cover and sky (Wm-2K-1)
hw Convective coefficient between cover and ambient temperature
(Wm-2K-1)
hc,p-c Convective coefficient between cover and absorber (Wm-2K-1)
Collector efficiency
E Model error
n Number of spatial discretization points

33Greek letters
bsorber plate
Absorbance of the a
c bsorbance of the cover
A
c ransmittance of the cover
T
p mittance of the plate
E
c Emittance of the cover

34 Subscript
c Cover
s Sky
a Absorber
f Fluid
e Ambient
p Bottom part of the collector
i Spatial position of the control volume
35
36

1 2
2
371. INTRODUCTION

38One of the most common, cost-effective and reliable applications of solar

39energy is domestic water heating. However, solar food drying is one of the
40oldest techniques related to food preservation in the world (Fudholi et al., 2011).

41Solar collectors absorb solar radiation and convert it into thermal energy. The
42heat is later on transferred to a fluid, which flows through the system (Saxena et
43al., 2015). Outlet fluid temperatures for solar air collectors do not exceed 100 C
44(Altfreld et al., 1988). Because of this, these collectors have a good potential for
45drying, which has prompted studies to generate simulation models to determine
46the efficiency of such integrated systems (Ayadi, 2014).

47Flat-plate solar air collectors are the most commonly used systems to perform
48this transformation due to their structural simplicity, low manufacturing costs and
49maintenance (Peuser et al., 2005; Kalogirou, 2004). Several authors have
50focused on the analysis and operation of this type of solar collectors in order to
51improve their design and performance (Tian and Zhao, 2013; Al-Khaffajy and
52Mossad, 2013; Bracamonte and Baritto, 2013). Some researchers have also
53focused on improving fluid flow rate, validating such improvements through
54experimental trials (Luminosu and Fara, 2005). In addition, other studies have
55focused on generating models to determine the optimum channel geometry for
56flow operation (Hegazy, 2000) in order to assess the effects of channel
57geometry (Ondieki et al., 2014) or determine the effects of selective coating on
58the thermal collector performance (El-Sebaii and Al-Snani, 2010). A recent study
59has even assessed the impact of dust deposition on the transparent cover
60(Deng et al., 2015).

61Most of the studies related to flat-plate solar air collectors use static models to
62simulate their operating parameters, applying finite difference techniques to
63solve the model equations. The results obtained in some of these studies are in
64agreement with experimental data, for flow and temperature distribution inside
65the collector (Alghoul and Ben Nagi, 2014). However, there are no studies in
66which the quality of simulations is assessed, or comparisons are made between

3 3
4
67the results obtained with both the static and dynamic models, and with other
68models capable of storing temporary variables.

69The trend in recent studies is to use tools of computational fluid dynamics

70(CFD) to generate three-dimensional simulations of fluid flow inside the
71collector. These techniques have also been employed to compare theoretical
72and experimental data of flow rate and temperature distribution inside the solar
73air collector. They have also been used to improve collector efficiency through
74modifications of plate shape, number of channels and the use of baffles within
75these systems (Amraoui and Aliane, 2014). CFD is a powerful modeling tool,
76but there are limitations to its applications. CFD requires an adequate turbulent
77flow for the model, and this parameter is difficult to be determined (SinghYadav
78and Bhagoria, 2013).

79Despite numerous studies and advances in mathematical modeling, validation

80is required. This represents the main difficulty, either by the effect of radiation on
81the sensors, when parameters such as temperature of the cover are measured,
82or due to the operating range of the fluid flow rate in natural convection
83systems.

84The present study assessed the performance of a static and a dynamic model
85for flat-plate solar air collectors under the same operating conditions. A data
86acquisition and data storage system was implemented to compare the
87simulation results with actual experimental data.

88The dynamic model included some variables, which corresponded to

89parameters related to heat losses, mass flow and specific heat of the
90components of the system. These variables are considered as constants in the
91original dynamic model developed by Garg (1984).

92

5 4
6
 932. MATERIALS AND METHODS

942.1 Simulation models for flat-plate solar air collectors

95A flat-plate solar air collector was used to analyze the simulation models. The
96collector consisted of a transparent cover, a plate, which acts as a black body to
97absorb sunlight, and a thermally insulated support frame (Figure 1).

98

99 Figure 1. Schematic cross-section of a flat-plate solar air and simple

100 flow collector.

101Solar radiation passes through the transparent cover and hits the absorber
102plate, resulting in an increase of temperature. The plate then transfers this
103energy to the fluid (air) flowing inside the collector driven by natural convection.

1042.2 Energy balances

105Physical simulation models for flat-plate solar air collectors are based on energy
106balances, which can be represented as electrical circuits composed of thermal
107resistance (Figure 2).

108
109 Figure 2. Diagram of a heat transfer model in a section of a flat-plate solar
110 air collector.

7 5
8
111In this model: temperatures are equivalent to electric potentials, heat flows due
112to solar radiation are equivalent to electrical currents, and terms related to
113convection and radiation coefficients are equal to electrical resistance.

114This energy balance allows developing two solution strategies for these
115systems: a static model and a dynamic model.

1162.3 Static simulation model

117 An energy balance for each collector component, results in a mathematical
118model consisting of equations (1), (2) and (3), as proposed by Ong (1995), cited
119by Naphaton and Kongtragool (2003), and by Bracamonte and Baritto (2013),
120among others.

Ic hr,c-a (Ta -Tc ) hc,c-f (T f - Tc ) = hr,c-s(Tc - Ts )+hc,c-e (Tc -Te )

1212.3.1 Energy balance
122 on the transparent cover:

123 (1)
124
125where, Ic represents the solar radiation absorbed by the transparent cover, hr,c-
126a (Ta - Tc) represents the radiative heat transfer between the absorber plate and
127the cover, hc,c-f (Tf - Tc) represents the convective heat transfer between the fluid
128and the cover, hr,c-s (Tc - Ts) represents the radiative heat transfer between the
129cover and the sky, and hc,c-e (Tc - Te) represents the convective heat transfer
130between the cover and the ambient.

1312.3.2 Energy balance on the absorber plate:

I a c = hc,f-a(Ta -T f ) hr,c-a(Ta -Tc )+U b(Ta Te )

132 (2)

133where, Iac represents the solar radiation absorbed by the absorber plate, hc,f-a
134(Ta Tf ) represents the convective heat transfer between the absorber plate
135and the fluid, hr,c-a (Ta - Tc) represents the heat transfer by radiation between the
136cover and the absorber plate, and Ub (Ta Te) represents the conductive heat
137transfer between the absorber plate and the ambient through the collector back
138side.

9 6
10
139

dT f
hc,f-a(Ta - T f ) = hc ,c-f (T f - Tc )+GC f
1402.3.3 dx Energy balance on the fluid:

141 (3)

dT f T f ,i T f ,i 1

142 dx dx where, hc,f-a (Ta Tf) represents the convective heat transfer
143between the absorber plate and the fluid, hc,c-f (Tf - Tc) represents the convective
144heat transfer between the fluid and the cover; and GCf (dTa / dx) represents the
145heat stored in the fluid. In the last equation term, the temperature spatial
146derivative was approximated by finite differences (Backward):

147 (4)
148
149where, Tf,i represents fluid temperature for an element of length dx inside the
150collector and Tf,i-1 represents fluid temperature in the previous component of the
151collector (Figure 3).

1522.4 Dynamic simulation model

153The dynamic model is based on the energy balances included in the static
154model, but adding heat capacity and heat conduction for each component.
155These parameters allow to calculate energy storage and determine the
156temporal evolution of the system variables (Garg, et al., 1984).

Tc 2Tc
I c M c C c K c c hc,c-e (Tc -Te ) hr,c-s (Tc -Ts )
t x 2
hc,c-f (Tc - T f ) hr,c-a (Tc - Ta )
1572.4.1 Energy balance on
158 the cover:

159 (5)

160

161where, Ic represents the net solar radiation absorbed by the glass cover, McCc
162(Tc/t) represents the rate of heat storage at the cover, Kcc (2Tc/x2) represents

11 7
12
163the conductive heat flow along the cover in the direction of flow, hc,c-e (Tc Te)
164represents convective heat transfer between the ambient and the glass cover,
165hr,c-s (Tc Ts) corresponds to heat transfer by radiation between the sky and the
166top of the cover, hc,c-f (Tc Tf) represents the convective heat transfer to the fluid
167from the glass cover, and hr,c-a (Tc Ta) represents heat exchange by radiation
168between the glass cover and the top of the absorber plate.

1692.4.2 Energy balance on the absorber plate:

Ta 2Ta
Ia c = M a Ca K a a hr,c-a(Ta -Tc )
t x 2
hc,f-a(Ta -T f ) U b(Ta Te )
170
171 (6)

172where, Iac represents the net solar radiation absorbed by the absorber plate,
173MaCa (Ta/t) represents the rate of heat storage in the absorber plate, Kaa
174(2Ta/x2) represents the conductive heat flow in the absorber plate in the
175direction of flow, hr,c-a (Ta Tc) represents the radiative heat transfer between the
176ambient and the glass cover, hc,f-a (Ta Tf) represents the convective heat
177transfer between the absorber plate and the fluid, and Ub (Ta Tf) represents the
178conductive heat transfer between the absorber plate and the ambient through
179the collector back side.

T f G f C f T f
M fCf hc ,c- f (Tc - T f ) hc, f -a (Ta - T f )
1802.4.3 t W x Energy balance on
181 the fluid:

182 (7)

183where, Mf,Cf (Tf /t) indicates the rate of heat storage according to time, Gf Cf /W
184(Tf /x) represents the rate of heat storage in the fluid for a control volume, hc,c-f
185(Tc Tf) represents the convective heat transfer between the cover and the fluid,
186and hc,f-a (Ta Tf) represents the convective heat transfer between the absorber
187plate and the fluid inside the collector.

13 8
14
188The dynamic simulation model for flat-plate solar air collectors represented by
189equations (5), (6) and (7) considers spatial derivatives determined by central
190differences and time derivatives solved by Forward differences, using short
191time increments (Garg et al., 1984). The solution originally proposed by the
192same author for this model, considers the following as constants: thermal
193parameters of materials, fluid flow rate inside the collector and conductive and
194convective heat transfer coefficients. These parameters were considered as
195variables in this study since the temperature changes with time in each collector
196component.

1972.5 Initial and boundary conditions in the simulations

198In both models, for each collector component (transparent cover, fluid and
199absorber plate) initial temperatures were set equal to the ambient temperature
200(Te):

201 Tc (i, 0) = Tf (i, 0) =Ta (i, 0) = Te (0)

202For the following time intervals, the dynamic model considered that inlet and
203outlet temperatures, for each component, would equal temperatures in the
204same section in the previous time interval. For the static model, the initial
205condition considered that the boundary temperature at the collector inlet would
206be equal to the ambient temperature at any time.

2072.6 Solution of the simulation models

208For the numerical approximation to both models, a spatial discretization was

209used, considering sections of the same length (dx) along the length of the
210collector (L). For each section, a control volume was set (Figure 3). Each
211control volume considered the effect of temperature change on the properties of
212air and collector materials.

15 9
16
213
214 Figure 3. Diagram of a section of the flat-plate solar air collector with
215 the corresponding control volume.

216The use of finite difference to solve the energy balances in the case of the static
217model allowed developing a system of algebraic equations and to obtain the
218solutions by using Newton-Raphsons method for each control volume.

219In contrast, the equations of the dynamic model, which are obtained from
220energy balances, are independent between them, since there is no relationship
221between temporal and spatial variables. Therefore, those do not constitute a
222system of equations.

223A MATLAB code was implemented for each model. The simulations results
224generated curves for outlet air temperature, cover temperatures and absorber
225plate temperatures. The results of each model were compared with the
226measured data, which allowed assess the collector performance under the
227environmental conditions in which the experiment was conducted.

2282.7 Heat transfer coefficients

229Radiative, convective and conductive heat transfer coefficients between

230collector components were calculated as proposed by Duffie and Beckman
231(1991).

2322.7.1 Radiative coefficient:

233To determine the radiative heat transfer coefficient between the absorption
234surface and the transparent cover (hr,a-c), both plates (opaque and gray for long-
235wavelength radiation) were considered:

17 10
18
 2 2
(T p Tc )(T p Tc )
hr,a c
1 1
1
p c
236

237 (8)

238

2 2
hr,c s c (Tc Ts )(Tc Ts )
239 To determine the radiative heat transfer coefficient
240between the sky and the transparent cover (hr,cs), the sky was considered as a
241black body, while the transparent surface was considered as a gray body for
242long-wavelength radiation:

243 (9)

2442.7.2 Convective coefficient:

245The convective heat transfer coefficient between the cover and the external
246ambient (hw) was determined from the expression given by McAdams et al.
247(1954), which depends on wind speed (V).

hw 5.7 3.8 V (10)

248

Nu K f
hc , p c
249 h To calculate the convective heat transfer coefficient inside the
250collector, between the absorber plate and the cover (hc,p-c), the Nusselt number
251(Nu) was determined by the following equation:

252 (11)

253To calculate the fluid heat transfer coefficient between inclined plates, losses
254generated by forced-air flow were not considered, while the Nusselt number for
255natural convection was calculated as proposed by Hollands (1819):
*
* 1

1708(sin 1,8 )
1 ,6
1708 Ra cos( ) 3
Nu 1 1,44 1 1 1
Ra cos( ) Ra cos( ) 5830
256

19 11
20
257 (12)

258Where, is the collector tilt angle and air is considered as an ideal gas. The
259second and third terms (*) are used to calculate the Nusselt number only if they
260are greater than zero, otherwise they are eliminated from the equation (12).

g f Th 3
Ra
f f
261 Fluid properties, such as volumetric expansion coefficient,
262kinematic viscosity and thermal diffusivity, are required to determine the
263Rayleigh number (Ra).

264 (13)

265The properties of the air inside the collector were calculated as proposed by
266Tiwari et al. (2002), depending on fluid temperature (Tf) of each control volume
267in Celsius degrees.

268

G f C f (T f -T e )

269 I Ac Efficiency of the flat-plate solar air collector, both in the
270simulations and experimentally, was calculated using the following equation:

271 (14)

272Root mean square error (RMSE) was used to calculate the error associated
273with each model:

E
Dm Ds 2

274 n (15)

275
276Where, Ds represents the value of the simulated data for each model, Dm is the
277value of the measured data and n represents the number of data entries
278between 8:00 am and 10 pm.

21 12
22
279Root mean square error (RMSE) was used to compare the results for outlet fluid
280temperature obtained with both models and experimental data.

2812.8 Implementation of the experimental system

282To evaluate the performance of the static and dynamic models, a flat-plate solar
283air collector was built with the following dimensions: 2 m long (L), 1 m wide (W),
28413 cm high (C), 4 cm space between the cover and the absorber plate (H), an
285absorber plate and cover of 0.94 m wide (B) and an 9 cm thick insulating layer
286of (Ep), (Figure 4).

287
288 Figure 4. Diagram of the flat-plate solar air collector built.

289This collector is provided with a 5 mm thick glass cover, a 1 mm thick aluminum

290absorber plate, a 9 cm thick polyester insulating layer, a structural frame with
291steel galvanized profiles (0.85 mm thick) and a back side made of smooth
292galvanized metal (0.35 mm thick). Physical and optical properties of the
293materials are depicted in Table 1.

294 Table 1. Physical and optical properties of the materials used for the
295 flate-plate solar air collector.

Material
Unit Aluminum Glass Polyester
insulating
Property layer
Density kgm-3 2700 2500 6.0
W(mK)
Thermal conductivity -1
209 0.81 0.068
J(kgK)-
Specific heat 1
909 670 -
Absorbance 0.80 0.08 -
Transmittance - 0.80 -

23 13
24
 Emissivity 0.96 0.89 -
296

297Las propiedades opticas del vidrio se obtubieron de xxxxxx,nnnn.

298Only thermal conductivity and thickness of the insulating layer were used to
299calculate subsequent losses, leaving aside the conductive effects in the
300collector back side which is made of smooth galvanized metal.

301Measurements for collector operation and ambient temperatures were

302conducted at the College of Agricultural Engineering of the University of
303Concepcin. To this end, the solar air collector was installed facing true north
304with a 23 tilt angle. (Figure 5).

305A data acquisition system was implemented to measure and storage the
306collector operating variables. This system included temperature measuring
307sensors (NTC thermistors 640-10K) located at three different points in the
308absorber plate, with a sampling frequency of one minute. In addition, an air
309temperature sensor (DTH22 / AM2302) and an air flow sensor (GM8903) were
310placed at the collector outlet, with sampling frequencies of one minute and one
311second, respectively. Additional features are presented in Table 2.

312

25 14
26
313 Figure 5. Location diagram for temperature sensors (A) in the
314 absorber plate and air flow sensor (B) at the collector
315 outlet.

316 Table 2. Characteristics of sensors used for data measurement in the

317 solar air collector.

Sensor
Characteristics 640-10K DTH22 GM8903
Range -40 a 150 C -40 a 80 C 0 a 30 ms-1
Precision 0.5 % 0.5 % 3%
Resolution 0.1 C 0.1 C 0.001 ms-1

318Environmental data (solar radiation, air temperature and wind speed) were
319recorded at a weather station (Campbell Scientific, General Weather Station) of
320the College of Agricultural Engineering, University of Concepcin, Chilln
321Campus 3635'44.0" S 7204'46.3" W) on Saturday January 30, 2016 from 8:00
322am to 10:00 pm.

323Measurements started at 8:00 am with an ambient temperature (Te) close to 15

324C. The temperature gradually increased, reaching 35 C at 5:00 pm, and then
325descended to 30 C at 10:00 pm, while wind speed varied between 1.8 and 3.6
326ms-1 during the day as shown in Figure 6.
327

40 4

35 3.5

30 3

25 2.5
Temperature [C] Wind speed [m s-1]
20 2
Te V
15 1.5

10 1
9 111315171921
8 10121416182022
Time of day
328
329 Figure 6. Ambient temperature (Te) and wind speed (V), during
330 model validation tests.

331Moreover, the solar radiation incident on the 23 tilt collector is shown in Figure
3327.

27 15
28
 1200

1000

800

600
Solar radiation [w m-2]
400

200

0
8 9 10111213141516171819202122
Time of day
333
334 Figure 7. Solar radiation incident on the solar air collector, facing
335 true north with a tilt of 23 respect to the horizontal line.

336Se registr adems la velocidad del aire a la salida del colector (Figura 7), y
337con estos se calcul el flujo msico del fluido en el interior del colector.

3.5

2.5

2
airspeed [m/s]
1.5

0.5
0
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Time of day
338

339 Figure 8. Velocidad del aire a la salida del colector.

340

3413. RESULTS

3423.1 Experimental performance of the solar air collector

343The following operating parameters were recorded during the tests: absorber
344plate temperature, measured at a point close to the collector outlet, air
345temperature at the collector inlet, temperature at the central part of the collector
346back side, ambient temperature, which corresponds to the fluid temperature at

29 16
30
347the collector inlet (Figure 8), incident solar radiation on the collector and air
348mass flow in the collector.

110

90

70
Temperature [C]
50
Tp Tf Ta Te

30

10
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Time of day
349
350 Figure 8. Absorber plate temperature (Ta) measured at a point near
351 the collector output, collector back side temperature (Tp),
352 fluid temperature at the collector outlet (Tf) and ambient
353 temperature (Te).

354The temporal evolution of the absorber plate temperature at the closest point to
355the air outlet, generated a Gaussian curve, with temperatures around 100 C at
356about 3:00 pm. Fluid temperature at the collector outlet, showed a gradual
357increase throughout the day, reaching its highest level (around 60 C) a few
358minutes before 5:00 pm, and then decreased until reaching the ambient
359temperature.

360Rate of the fluid flow was measured at collector outlet. Values ranged from 0.6
361and 1.3 m s-1 during the experimental tests and were used to calculate the
362mass flow of the air through the collector. Based on these data, collector
363efficiency was calculated.

3643.2 Simulation of a plane- plate solar air collector using the static and
365 dynamic models

366Collector performance was simulated using both models under the same
367environmental conditions as those during the experimental test, i.e. ambient
368temperature, solar radiation and wind speed. Temperatures of the cover,
369absorber plate and fluid at the collector outlet were obtained through

31 17
32
370simulations. The results are provided in Figures 9 and 10 and they correspond
371to the static and dynamic models results, respectively.

110

90

70

Temperature [C]
50
Tc Tf Ta Te
30

10
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Time of day
372

373 Figure 9. Ambient temperature (Te), cover temperature (Tc), fluid

374 temperature (Tf) and absorber plate temperature (Ta),
375 obtained at the collector outlet with the static model.

376Figure 9 shows the evolution of the temperature obtained in the simulation with
377the static model for each main component of the collector.

378The plate starts heating up about half an hour earlier than observed
379experimentally, reaching a maximum value of 105 C at 4:00 pm, while the fluid
380temperature does not exceed 50 C when the increase in fluid temperature is
381close to 12 C respect to the ambient temperature. In this case, the cover
382temperature is greater than the fluid temperature, which represents a
383phenomenon with no physical basis, since the fluid supplies heat to the cover
384by convection. Furthermore, as the cover is in direct contact with the
385environment, most heat loss occurs by convection. This phenomenon is
386observed in the static model because this does not consider the contribution of
387heat storage in the components of the system or the conductive heat flow on
388them.

33 18
34
 110

90

70

Temperature [C]
50
Tc Tf Ta Te
30

10
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Time of day
389
390 Figure 10. Ambient temperature (Te), cover temperature (Tc), fluid
391 temperature (Tf) and absorber plate temperature (Ta),
392 obtained at the collector outlet with the dynamic
393 model.

394Figure 10 shows the evolution of temperatures obtained with the dynamic

395model, simulated under the same operating conditions as in the static model.
396Values obtained are similar to those obtained with the static model, with a
397maximum absorber plate temperature of 110 C at about 4:00 pm, when fluid
398temperature reaches 58 C, 26 C above ambient temperature. These results
399are the best fit to the temperatures measured experimentally. Unlike the static
400model, the dynamic model behaves as expected in terms of cover temperature
401and a fluid temperature, i.e., the temperature of the cover is always lower than
402the fluid temperature.

4033.3 Evaluation of model performance

404The results of the simulation with the dynamic model provided in Figure 11
405show the evolution of fluid temperature at the collector outlet, and temperature
406measured experimentally.

35 19
36
 60

50

40

Temperature [C]
30
EX DM
20

10
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Time of the day
407

408 Figure 11. Measured (EX) and simulated outlet fluid temperature
409 using the dynamic model (DM).
410
411When comparing the results of the simulated curve obtained with the dynamic
412model with those of the experimental curve, an RMSE of 3.5 C is observed,
413with an average percentage error of 11.2 % and a standard error of 2.1 C. As
414observed in Figure 11, the dynamic model tends to slightly overestimate outlet
415temperatures of the fluid, especially when temperature reaches its maximum
416value.

60

50

40

Temperature [C]
30
EX EM
20

10
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Time of the day
417
418

419 Figure 12. Measured (EX) and simulated outlet fluid

420 temperature using the static model (EM).

421Figure 12, shows the results of the simulation with the static model. Fluid
422temperature at the collector outlet showed an RMSE of 5.3 C, with an average

37 20
38
423error of 11.7 %, and a typical error of 4.3 C compared to the value obtained in
424experimental measurements. As observed in Figure 12, the static model tends
425to underestimate fluid temperatures at the collector outlet.

426In addition, collector operational efficiency obtained with both models was
427compared with data obtained experimentally (Figure 13).

25

20

15

Eficiency [%]
10
EM DM EX
5

0
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Time of the day
428

429 Figure 13. Efficiency calculated with the static (EM) and
430 dynamic (DM) models in relation to efficiency
431 obtained experimentally (EX).
432
433In addition, operational collector efficiency was obtained experimentally and
434compared to the efficiency obtained with the use of both models. During
435operation, the maximum efficiency was recorded at 3:00 pm and reached a
436value of 18.2%, which is well below the maximum efficiency reported in the
437literature for this type of collector and that, in some cases, reaches values close
438to 50% (Lammardo and Baritto, 2010). On the day of measurement, an average
439collector efficiency of 12.6% was obtained experimentally. The collector
440efficiency obtained with the static model was 10.8% and an RMSE of 4.5%,
441while the dynamic model resulted in a value of 14.9% and an RMSE of 3.8%.
442Although the average efficiency in the static model is similar to that recorded
443experimentally, it does not necessarily represent the tendency of the
444experimental data, and therefore, it is not a good indicator by itself since the
445error associated with this simulation is greater.

39 21
40
446As can be observed in Figure 13, the dynamic model behaves similarly to the
447experimental data in terms of efficiency. Moreover, the static model results in
448very high values at the start of the day because the error associated with the
449fluid temperature directly affects the calculation of efficiency, especially during
450the time periods with less radiation. This model also tends to underestimate
451efficiency during the hours with higher radiation.

452

453CONCLUSIONS

454Two models (static and dynamic) were developed and used for simulating the
455operation of flat-plate solar air collectors. The results of each model were
456compared with experimental data.

457For the dynamic model, data of environmental variables were included, as well
458as the temporal evolution of parameters associated with heat loss of the
459system, mass flow and specific heat of the components of the system. This
460differs from previous studies since these parameters are not included.

461The error of each model was quantified, comparing results with the
462experimental data, which provided information related to the ability of models to
463simulate solar air collectors.

464When comparing data of air temperature at the collector outlet with those
465obtained using the dynamic model, it can be stated that this model generated
466the best results with an RMSE of 3.5 C, while the model static had an RSME
467of 5.1 C. The dynamic model had the best performance with an error of 3.8%
468compared to the static model that recorded an RMSE of 4.5%.

469The collector recorded a maximum efficiency of 18.2% at 3:00 pm, which

470corresponds to the time period of highest incident solar radiation. The average
471collector efficiency was 12.6% during the day though the literature reports
472values close to 50% for this type of collectors.

473Although the static model presents a poorer performance in predicting

474temperature and collector efficiency, it is the most widely used simulation model
475for this type of systems. This is because it requires lower data entry and it is

41 22
42
476simpler to implement than a dynamic model. However, this work proves
477experimentally that the dynamic model has a better fit to the experimental data
478than the static model.

479Based on the prediction of performance parameters such as outlet fluid

480temperature and collector efficiency, it can be concluded that the dynamic
481model fits experimental data better due to the incorporation of temporary
482variables of heat storage and heat flow by conduction. However, this model
483requires data of higher quality; as the model is solved explicitly, stability
484conditions have to be satisfied, which are not required by the static model.

43 23
44
4854. REFERENCES

486Alghoul, S. & Nagi, M. Ben, 2014. Flow and Temperature Analysis inside Flat
487 Plate Air Heating Solar Collectors. International Journal of Recent
488 Development in Engineering and Technology, 3(3):6368.

489 Al-Khaffajy, M. & Mossad, L., 2013. Optimization of the heat exchanger in a flat
490 plate indirect heating integrated collector storage solar water heating
491 system. Renewable Energy, 57(1):413-421

492Altfeld,K., Leiner, W. & Fiebig, M., 1988. Second law optimization of flat-plate
493 solar air heaters. Solar Energy, 41(2):127-132.

494Amraoui, M.A. & Aliane, K., 2014. Numerical analysis of a three dimensional
495 fluid flow in a flat plate solar collector. International Journal of Renewable
496 and Sustainable Energy, 3(3):6875.

497Ayadi, M., Mabrouk, S. B., Zouari, I. & Bellagi, A., 2014. Simulation and
498 performance of a solar air collector and a storage system for a drying unit.
499 Solar Energy, 107(1):292304.

500Baritto, M. & Bracamonte, J., 2013. En Colectores Solares Planos No

501 Isotrmicos Para Calentamiento De Aire. Revista Iberoamericana de
502 Ingeniera Mecnica, 17(1):149162.

503Deng, J., Yang, X. & Wang, P., 2015. Study on the second-order transfer
504 function models for dynamic tests of flat-plate solar collectors Part I: A
505 proposed new model and a fitting methodology. Solar Energy, 114(1):418
506 426.

507Duffie,J.A. & Beckman, W.A., 1991. Energy of Thermal Processes. John Wiley,
508 New York.

509El-Sebaii, A. & Al-Snani, H., 2010. Effect of selective coating on thermal

510 performance of flat plate solar air heaters. Energy, 35(4):18201828.

511Fudholi, A., Ruslan, M.H. & Othman, M. Y., 2014. Mathematical Model of
512 Double-Pass Solar Air Collector. Solar Energy Research Institute, 279
513 283.

514Garg, H.P., Chandra, R. & Usha, R., 1981. Transient analysis of solar air
515 heaters using a finite difference technique. Energy Research, 5(1):243-
516 252.

45 24
46
517Hegazy, A.A., 2000. Performance of flat plate solar air heaters with optimum
518 channel geometry for constant/variable flow operation. Energy Conversion
519 Management, (41):401-17.

520Hollands, K.G.T., Unny, T. E., Konicek, L., 1976. Free Convective heat transfer
521 across inclined air layers. ASME, J. Heat Transfer, 98:189-193.

522Kalogirou, S.A., 2004. Solar thermal collectors and applications. Progress in

523 Energy and Combustion Science, 30(1):231295.

524Lammardo, A. & Baritto, M., 2010. Modelo matemtico del comportamiento

525 trmico de un colector solar de placas planas inclinadas para
526 calentamiento de aire Mathematical, Revista Ingeniera UC, 17(3):1927.

527Luminosu, I. & Fara, L., 2005. Thermodynamic analysis of an air solar collector.
528 International Journal of Energy, 2(4):385-408.

529McAdams, W.H., 1954. Heat Transmission, third ed. McGraw-Hill, New York.

530Naphon, P. & Kongtragool, B., 2003. Theoretical study on heat transfer

531 characteristics and performance of the flat-plate solar air heaters.
532 International Communications in Heat and Mass Transfer, 30(8):112536.

533Ondieki, H.O., Koech R.K., Tonui J.K. & Rotich S.K., 2014. Mathematical
534 Modeling Of Solar Air Collector With A Trapezoidal Corrugated Absorber
535 Plate. International Journal Of Scientific & Technology Research, 3(8):51
536 56.

537Ong K., 1995. Thermal performance of solar air heaters: mathematical model
538 and solution procedure. Solar Energy, 55(2):93109.

539Peuser, F.A., Remmers, K.H. & Schnauss, M., 2005. Sistemas Solares
540 Trmicos, Diseo e Instalacin. Censolar, Espaa.

541Saxena, A., Srivastava, G. & Tirth, V., 2015. Design and thermal performance
542 evaluation of a novel solar air heater. Renewable Energy, 77(1):501511.

543Tian, Y. & Zhao, C., 2013. A review of solar collectors and thermal energy
544 storage in solar thermal applications. Applied Energy, 104(1):538-553.

545Tiwari, G.N., 2002. Solar Energy Fundamentals, Design, Modeling and

546 Application. Narosa Publishing House, No: 247.

547Yadav, A.S. & Bhagoria, J.L., 2013. A CFD (computational fluid dynamics)
548 based heat transfer and fluid flow analysis of a solar air heater provided
549 with circular transverse wire rib roughness on the absorber plate. Energy,
550 55(1):11271142.

47 25
48