Available online at www.sciencedirect.

com

ScienceDirect
Solar Energy 131 (2016) 47–60
www.elsevier.com/locate/solener

Numerical studies on the effect of delta-shaped obstacles’ spacing

on the heat transfer and pressure drop in v-corrugated channel of solar
air heater
Ekadewi A. Handoyo a,⇑, Djatmiko Ichsani b, Prabowo b, Sutardi b
a
Mechanical Engineering Dept, Petra Christian University, Surabaya, Indonesia
b
Mechanical Engineering Dept, Institut Teknologi Sepuluh Nopember, Surabaya, Indonesia

Received 11 October 2015; received in revised form 9 February 2016; accepted 16 February 2016
Available online 1 March 2016

Communicated by: Associate Editor Bibek Bandyopadhyay

Abstract

Solar air heater (SAH) is simple in construction compared to solar water heater. Yet, it is very useful for drying or space heating.
Unfortunately, the convective heat transfer between the absorber plate and the air inside the solar air heater is rather low. Some research-
ers reported that obstacles are able to enhance the heat transfer in a flat plate solar air collector and others found that a v-corrugated
absorber plate gives better heat transfer than a flat plate. Only a few research combines these two in a SAH. This paper will describe the
combination from other point of view, i.e. the spacing between obstacles. Its spacing possibly will effect the heat transfer and pressure
drop of the air flowing across the channel.
The first step in numerical study is generating mesh or grid of the air flow inside a v-corrugated channel which was blocked by some
delta-shaped obstacles. The mesh was designed three-dimension and not uniform. The mesh are made finer for area near obstacles and
walls both for upper and bottom, and then gradually coarser. Grid independency is the next step to be conducted. When the mesh is
already independent, the numerical study begins. To validate the numerical model, an indoor experiment was conducted. Turbulent
model used was Shear Stress Transport K-x (SSTK-x) standard. Having a valid numerical model, the spacing between obstacles was
studied numerically. Ratio spacing to height, S/H of obstacles investigated were 0.5; 1; 1.5; and 2.
From numerical studies in a v-corrugated duct, it is found that backflow between obstacles and high velocity in the gap between
obstacles and absorber plate causes the flow became more turbulent and enhanced the convection heat transfer between the air and
the absorber plate. Obstacles placed in a small spacing will increase Nusselt number (convection heat transfer) and friction factor (pres-
sure drop). The Nusselt number enhanced from 27.2 when no obstacle used to 94.2 when obstacles inserted with S/H = 0.5. The Nusselt
enhanced 3.46 times. The friction factor will increase from 0.0316 at no obstacle to 0.628 at ratio S/H = 0.5. The friction factor increased
19.9 times. Efficiency, Nusselt number, and friction factor are decreasing as ratio S/H is increasing. When ratio S/H used is 1 instead of
0.5, Nusselt number enhancement decreased only 1.13%, but friction factor decreased 15.1%. So, sacrificing a small amount of Nusselt
number but reducing a significant friction factor is advantageous. The optimal spacing ratio S/H of delta-shaped obstacles inserted in a
v-corrugated SAH is one. In other words, the optimal spacing of obstacle equals to its height.
Ó 2016 Elsevier Ltd. All rights reserved.

Keywords: Delta-shaped obstacle; v-corrugated channel; Solar air heater; Numerical study

⇑ Corresponding author.
E-mail address: [email protected] (E.A. Handoyo).

http://dx.doi.org/10.1016/j.solener.2016.02.031
0038-092X/Ó 2016 Elsevier Ltd. All rights reserved.
48 E.A. Handoyo et al. / Solar Energy 131 (2016) 47–60

Nomenclature

W wide of the obstacle (mm) Tabs absorber temperature (K)

H height of the obstacle (mm) Dh hydraulic diameter (m)
S spacing between the obstacles (mm) NuD Nusselt number
x, y, z coordinates h convection heat transfer coefficient (W/m2 K)
Re Reynolds number f friction factor
Cp specific heat of air (J/kg K) # air kinematic viscosity (m2/s)
m_ f mass flow rate of the fluid–air (kg/s) q air density (kg/m3)
Q_ u useful heat transfer rate (W) k thermal conductivity (W/m K)
I radiation intensity (W/m2) a air thermal diffusion coefficient (m2/s)
Ac collector aperture area (m2) g thermal efficiency of SAH
To collector outlet temperature (K)
Ti collector air inlet temperature (K)

1. Introduction Choudhury dan and Garg (1991) made a detailed analysis

of corrugated and flat plate solar air heaters of five different
Solar energy can be converted into thermal energy in a configurations. For the same length, mass flow rate, and air
solar collector. Solar collector basically is device used to velocity, it was found out that the corrugated and double
trap solar energy to heat a plate and transfer the heat to cover glass collector gave the highest efficiency. According
a fluid flowing under or above the plate. When sun light to Naphon (2007) the corrugated surfaces give a significant
falls onto a plate, solar radiation reaches the plate at lower effect on the enhancement of heat transfer and pressure
wavelength and heat it up. Then, the heat is carried away drop. The Nusselt number of flow in a v-corrugated channel
by either water or air that flows under or above the plate. can be 3.2–5.0 times higher than in a plane surfaces while
Solar collector used to heat up air is called solar air heater the pressure drop 1.96 times higher than on the correspond-
(SAH) and solar water heater for water. Generally, the ing plane surface. Islamoglu dan and Parmaksizoglu (2003)
solar air heater is less efficient than the solar water heater, reported that the corrugated channel gave the higher Nus-
because air has less thermal capacity and less convection selt number than the straight channel and the higher chan-
heat transfer coefficient. Yet, air is much lighter and less nel height gave higher Nusselt number for the flow with the
corrosive than water. The other benefit is that heated air same Reynolds number.
can be used for moderate-temperature drying, such as har- Besides changing the cross section area of the channel,
vested grains or fish. Since the solar air heater has less con- some also give effort to increase turbulence inside the chan-
vective heat transfer coefficient than solar water heater, nel with fins or obstacles. The result of experimental study
some researchers tried to increase this convective heat done by Promvonge (2010) in turbulent flow regime (Rey-
transfer coefficient. nolds number of 5000–25,000) showed that multiple 60° V-
A popular type of solar air heaters is the flat plate SAH, baffle turbulator fitted on a channel provides the drastic
which has a cover glass on the top, insulation on the sides increase in Nusselt number, friction factor, and the thermal
and bottom to prevent heat transferred to the surrounding, enhancement factor values over the smooth wall channel.
a flat absorber plate that makes a passage for the air flowing Promvonge found that Nusselt number was five times
with sides and bottom plate. Usually, the passage or chan- higher and friction factor was 30 times than without baffle
nel has a rectangular cross-section. The absorber plate will at Reynolds number = 10000, e/H = 0.3, and PR = 1.
transfer the heat to the air via convection. Unfortunately, Kurtbas dan and Turgut (2006) investigated the effect of
the convection coefficient is very low. To increase the fins located on the absorber surface in free and fixed man-
convection coefficient from the absorber plate, a ners. They used two kinds of rectangular fins which dimen-
v-corrugated plate is used instead of a flat plate. Tao et al. sion is different but total area is the same. The first type fin
(2007) stated that a solar air heater with a v-grooved absor- (I) has dimension 810
60 mm and the second (II) type
ber plate could reach efficiency 18% higher than the flat 200
60 mm. To have the same total area, there are 8 fins
plate on the same operation condition and dimension or and 32 fins for the first and second, respectively. The fins
configuration. Karim dan and Hawlader (2006) found that type II, both free and fixed, were more effective than type
a solar collector with a v-absorber plate gave the highest I and flat-plate collector. The fixed fin collector was more
efficiency and the flat plate gave the least. The results effective than free fin collector. Kurtbas and Turgut
showed that the v-corrugated collector is 10–15% and 5– reported that the maximum Instantaneous efficiency of
11% more efficient in single pass and double pass modes, SAH was 0.78 at air mass flow rate 0.08 kg/s when the
respectively, compared to the flat plate collectors. fix fins type II were used. While the efficiency of flat plate
 E.A. Handoyo et al. / Solar Energy 131 (2016) 47–60 49

SAH was only 0.42. So, its efficiency increased 1.86 times. heat transfer got higher if the obstacle height was taller
Romdhane (2007) created turbulence in the air channel and its longitudinal pitch was smaller. Nusselt number
using obstacles or baffles. The efficiency of collector and increased 3.5 times when obstacles inserted with ratio of
the air temperature was found increasing with the use of longitudinal pitch Pl/e = 3/2 and height e/H = 0.75 at Rey-
baffles. Baffles should be used to guide the flow toward nolds number was 10,000. While its friction factor
the absorber plate. Romdhane reported that the efficiency increased 4.7 times than without obstacle. SAH can be
reached 80% for the best type of baffle (chicane) for an modeled through least-squares support vector machines
air flow rate of 50 m3/h/m2. This efficiency was about 1.6 (LS-SVM) method as done by Esen et al. (2009a). The pre-
times efficiency without baffle. Ho et al. (2011) inserted fins dicted results from LS-SVM model had been compared to
attached by baffles and external recycling to a solar air hea- experimental results. LS-SVM could be used to predict the
ter. The experiment and theoretical investigations gave efficiency of SAH. They ashowed that efficiency of type I
result that heat transfer was improved by employing baffled SAH was the highest. Esen et al. (2009b) also proposed
double-pass with external recycling and fin attached over Artificial Neural Network and Wavelet Neural Network
and under absorber plate. According to Ho et al., the high- (ANN and WNN) to predict the efficiency of SAH. They
est collector efficiency was 61% in a double-pass flat-plate succeeded to show that the method could predict the
SAH with recycle when radiation intensity was 1100 SAH efficiency.
W/m2, 24 fins attached, and mass flow rate was 0.02 kg/s. The combine of those two findings, i.e. obstacles and v-
It was about 1.35 times its efficiency when there is no recy- corrugated absorber plate are able to improve SAH are
cle. Abene et al. (2004) used obstacles on the flat plate of a important to be studied. Handoyo et al. (2014) reported
solar air collector for drying grape. The obstacles ensure a that the obstacles are able to enhance heat transfer in a v-
good air flow over the absorber plate, create the turbulence corrugated SAH but increase the air pressure drop. To
and reduce the dead zones in the collector. The highest effi- reduce the air pressure drop, the obstacles were bent verti-
ciency got was 70% when the air flow rate was 50 m3/h/m2. cally. The optimal bending angle is 30°. The SAH’s effi-
It was about 1.9 times its efficiency when no obstacle used. ciency was 5.3% lower when the obstacles bent 30° instead
The air pressure drop increased about twice with obstacles. of straight (0°), but the pressure drop was 17.2% lower.
Esen (2008) used a double-flow solar air heater to investi- According to Incropera dan and DeWitt (2002), in
gate three different type obstacles placed on absorber plates forced convection heat transfer, there are two quantities
compare with the flat plate. The collector has three absor- that are important to be determined. One is friction coeffi-
ber plates to make three passages for the air flowing cient, Cf, and the other is Nusselt number, Nu. Friction
through. From the research done, it was found that all col- coefficient, Cf is used to calculate the shear stress at the wall
lectors with obstacles gave higher efficiency than the flat and then to calculate pressure drop. While Nusselt number,
plate and type III obstacles with flow in middle passage Nu is used to calculate convection heat transfer rate. The
gave the highest efficiency, i.e. 52.9%, for air flow relation that correlate them is known as Reynolds Anal-
0.02 kg/s. It was about 1.13 times higher than its efficiency ogy: C f ;x Re2L ¼ Nux Pr1=3
with no obstacle. Ozgen et al. (2009) used aluminum cans When Nusselt number increases, then friction coefficient
as obstacles installed over and under absorber plate. They will increase, too. In a SAH, it is expected that convection
found that the SAH using double flow passage (over and heat transfer increases but pressure drop does not. So,
under absorber plate) gave higher efficiency than just one many research were conducted to look for increasing in
flow passage (either over or under absorber plate). They heat transfer at low pressure drop.
also found that SAH with staggered cans (type I) had the The spacing between obstacles mounted on absorber
highest efficiency compared to aligned cans (type II) and plate can be varied. A different spacing will give different
flat plat (type III). The efficiency of SAH type I was 0.73 air pressure drop as the air flows across the collector.
at air flow rate 0.05 kg/s and it was about 1.4 times of type Obstacles increase the air pressure drop. When the spacing
III. Akpinar dan and Koçyiğit (2010) had experimental is large, the obstacles used is less and the air pressure drop
investigation on solar air heater with several obstacles will be reduced. What about the heat transfer from the
(Types I–III) and without obstacles. The optimal value of absorber plate to the flowing air? How will the spacing
efficiency was obtained for the solar air heater with Type between obstacles effect the heat transfer? To the knowl-
II obstacles on absorber plate in flow channel duct for all edge of the authors, there is no research investigating the
operating conditions and the collector with obstacles effect of delta-shaped obstacles’ spacing on a v-
appears significantly better than that without obstacles. corrugated SAH performance, yet.
The efficiency of Type II was 0.6 at air flow rate This paper describes the result of the numerical studies
0.0052 kg/s and it was about twice of SAH without obsta- of obstacles’ spacing inserted in a v-corrugated channel
cle. Bekele et al. (2011) investigated experimentally the of a SAH. The heat transfer from the v-corrugated absor-
effect of delta shaped obstacles mounted on the absorber ber plate and the air pressure drop flowing the v-corrugated
surface of an air heater duct. They found that the obstacle channel is to be discussed. The obstacles are delta-shaped
mounted duct enhances the heat transfer to the air. The and installed on bottom plate of the channel.
50 E.A. Handoyo et al. / Solar Energy 131 (2016) 47–60

125
mm

Cover glass

V-corrugated
absorber
plate
Thermal
insulaon

90 mm
obstacles

Flow H=
direcon 51 mm

obstacles
S W=18
mm
(a) Schemac of the solar collector model

Flow
direcon blowe
Thermo
douple
Pressure
gage

Go
to

(b) Photograph of experimental set-up.

Fig. 1. The SAH model used in experiment.

2. Experimental set-up true. The validation is accomplished via experiment. An

indoor experiment model was constructed to validate
The studies began with choosing a solar collector model numerical result. It is a model of SAH using v-corrugated
to be studied. The study will be conducted numerically. absorber plate. Its schematic view and photograph are
Thus, a validation is required to prove that the result is shown in Fig. 1(a) and (b), respectively. The experiment
 E.A. Handoyo et al. / Solar Energy 131 (2016) 47–60 51

was conducted in a laboratory of Mechanical Engineering spacing, a ratio of spacing to height S/H is used. In this
Dept of Petra Christian University, Surabaya, Indonesia. study, the ratio S/H used were ø, 1, 1 ø, and 2.
The collector model’s dimension was 900 mm long, The numerical model for this problem was developed
90 mm width, and 125 mm height. A single 3-mm under some assumptions, i.e. as steady three-dimensional
transparent-tempered glass was used as the collector cover. turbulent, incompressible flow and constant fluid proper-
The v-corrugated absorber plate was made of 0.8-mm-thick ties. The space of obstacles was taken equal to height of
aluminum and painted black. The apex angle of the v- obstacles, thus ratio is 1. Then, there are 17 obstacles used
corrugated plate was 20°. The v-corrugated channel’s cross in this flow. Body force and viscous dissipation are ignored
section was 30 mm width and 85 mm height. To prevent and it was assumed no radiation heat transfer. With these
heat loss, the left and right walls of collector are insulated assumptions, (Incropera dan and DeWitt, 2002) give the
with a 25-mm Styrofoam each and a 35-mm Styrofoam for related governing equations as follow:
the bottom. The delta-shaped obstacles were made congru- @u @v @w
ent to the channel i.e. triangular and its dimension was þ þ ¼0 ð1Þ
@x @y @z
18 mm wide, 51 mm height.  2 
The experiment was conducted indoor to maintain the @u @u @u 1 @P @ u @2u @2u
u þv þw ¼ þ# þ þ ð2Þ
radiation intensity, wind’s velocity and temperature. So, a @x @y @z q @x @x2 @y 2 @z2
bias result caused by different outdoor condition could be  2 
@v @v @v 1 @P @ v @2v @2v
avoided. The sunlight is replaced with four 500-W halogen u þv þw ¼ þ# þ þ ð3Þ
lamps. The radiation intensity received on the collector was @x @y @z q @y @x2 @y 2 @z2
 2 
measured using a pyranometer (Kipp & Zonen, type SP @w @w @w 1 @P @ w @2w @2w
Lite2) placed on top of the cover glass. To ensure the u þv þw ¼ þ# þ þ 2 ð4Þ
@x @y @z q @z @x2 @y 2 @z
homogenous intensity and to generate a certain absorber  2 
plate’s temperature, these lamps were equipped with adjus- @T @T @T @ T @2T @2T
u þv þw ¼a þ þ ð5Þ
table turner individually. The surrounding air condition @x @y @z @x2 @y 2 @z2
was controlled by an air conditioner installed in the room. In the above equations, q is the density of the air, p is the
Its temperature, humidity, and wind velocity are well con- pressure, # is the kinematic viscosity, a is the thermal diffu-
trolled. The collector was equipped with T-type thermo- sivity, and T is the temperature of the fluid. The boundary
couple which accuracy is 0.1 °C for measuring the air conditions of this problem were:
temperature at inlet and outlet of the collector, tempera-
ture of the absorber plate (at four different locations), At the inlet: u = 5.0 m/s, v = w = 0, T = 297.46 K
and ambient temperature. The pressure drop between inlet At the upper wall which is the absorber plate:
and outlet of the flowing air across the collector is also u=v=w=0
measured with a Magnehelic differential pressure At the bottom wall: u = v = w = 0
gage which accuracy is ±2 Pa. A centrifugal blower On the 17 obstacles: u = v = w = 0
(1000 m3/h, 580 Pa, 0.2 kW, 380 V input) was used to
induce the air flowing through the collector. The air flow
Inlet Reynolds number was calculated with : Re
was adjusted by means of a variable-frequency drive
(VFD) to be 5.0 m/s (or Reynolds number = 10,000). The uDh
¼ ð6Þ
air flow is measured using digital anemometer which accu- #
racy is ±0.1 m/s. All of the measurement equipments such In Eq. (6), Dh is hydraulic diameter and calculated with:
as pressure gages and thermocouples are installed accord-
4A
ing to ASHRAE requirement (ASHRAE, 1986). Dh ¼ ð7Þ
P
The area, A, in Eq. (7) is the channel cross section area,
3. Numerical set up and P, is its perimeter.
The governing equations above with the boundary con-
The discussion in this paper is on the heat transferred to ditions were solved using a commercial CFD package,
the flowing air and its pressure drop, then the domain of FLUENT 6.3.26 (FLUENT, 2003). The grids of the
numerical study is focused on the air flow inside the v- domain were generated using Gambit 2.4.6. The domain
corrugated channel. The apex angle of the v-corrugated and meshing used in this numerical study were shown in
channel is 20°. Its dimension was 900 mm long, 30 mm Fig. 2. Since the geometry is a triangular channel, the
width, and 85 mm height. The delta-shaped obstacles were numerical study should be conducted in three dimensions
made congruent to the channel i.e. triangular and its and all of the meshes were designed to have quality less
dimension was 18 mm wide, 51 mm height. The obstacles than 0.7.
were installed in one line only, because there is no much After having a good mesh, the next step is doing grid
space in the channel. Thus, the spacing to be discussed is independency. The grid or mesh is made smaller to get
also called longitudinal pitch. To specify the obstacles’ more accurate and detailed result. But, when the mesh is
52 E.A. Handoyo et al. / Solar Energy 131 (2016) 47–60

obstacles

Flow direction

Fig. 2. The grid used in numerical studies.

too small, calculation which is done by iteration needs a the residual could not meet the criteria. So, there was no
very long time and might end up with diverge result. There- result of Mesh B in Fig. 4.
fore, in this study the mesh or grids were designed not uni- The experiment result which will be discussed in Sec-
form. The finer mesh are used for area near walls both for tion 4 gives pressure drop as much as 397.7 Pa, air temper-
upper and bottom walls and then gradually the mesh are ature outlet 40.7 °C when the air inlet velocity is 5.0 m/s
made coarser as shown in Fig. 2. When there are obstacles and inlet air temperature 24.5 °C.
in the flow, the mesh are designed to be finer not only near Fig. 4 shows that numerical study with Mesh C and D
the walls but also around the obstacles in Fig. 2. There are give better result of air temperature compare to experiment
four meshes designs to be checked for independency as than Mesh A, but Mesh D gives too high pressure drop
shown in Fig. 3(a)–(d). The number of cells, faces, and result. Numerical studies using Mesh C give the most clo-
nodes used in each design are shown in Table 1. sely result to experiments. Thus, Mesh C was chosen for
To check the grid independency, the result from numer- the numerical studies.
ical study will be compared to experiment result. The same Using Mesh C pattern, some numerical studies were
boundary conditions and setting are used in each numerical conducted to know the effect of obstacles’ spacing. The
study of the four grids design as in Fig. 3. The boundary ratio of spacing to height, S/H, studied numerically were
conditions are: inlet air velocity = 5.0 m/s and outflow ø, 1, 1 ø, and 2. The Mesh used in this study are in
was employed for the outlet; inlet air tempera- Fig. 5a–d. When the ratio S/H is bigger, then the number
ture = 297.46 K; and absorber plate temperature as of obstacle is less. Number of obstacles are 8, 11, 17, and
obtained during experiments = 320 K. The settings in soft- 35 for ratio S/H = 2, 1 ø, 1, and ø, respectively.
ware used are: three dimension, double precision. Shear To give more comprehensive result, numerical study
Stress Transport K-x (SSTK-x) standard turbulent model was also conducted for air flow in a v-corrugated channel
was chosen to simulate the flow. Zhang and Liu conducted without obstacle. The domain and mesh used is shown in
a numerical simulation of the flow inside a diffusing S-duct Fig. 6.
inlet. Full three-dimensional Navier–Stokes equations are
solved and SST turbulence model is employed. Numerical 4. Results and discussion
results, include surface static pressure, total pressure recov-
ery at exit, are compared with experiment. A fairly good Since the experiment is conducted to validate the
agreement is apparent (Zhangdan and Liu, 2009). Other numerical result, then it is not necessary to conduct exper-
researcher, Kirkgoz et al. reported that numerical modeling iments for all ratio S/H. The ratio S/H used in experiment
using K-x and SSTK-x used on the cylinder surface gave equals to 1. Thus, there are 17 obstacles inserted and the
reasonably success result (Kirkgoz et al., 2009). The SIM- percentage of air flow blockage in the channel is 36%.
PLEC algorithm was employed to deal with the problem The experiment followed the numerical setting. The exper-
of velocity and pressure coupling. Second-order upwind iments were conducted at certain inlet air velocity and cer-
scheme were used to discretize the main governing equa- tain radiation intensity or absorber temperature. A VFD
tions. Material of the absorber, bottom plate and the was used to get an inlet air velocity equals to 5.0 m/s.
obstacles is aluminum. Fluid was air with inlet pressure An adjustable turner was used to adjust the radiation
1 atm and its properties, such as density, viscosity, thermal intensity that make the absorber’s temperature as high as
conductivity were function of temperature. 320 K or 47 °C.
Numerical studies were conducted for each of the four When the numerical result is close to the experimental
Mesh: A, B, C, and D. Fig. 4 shows global properties result, then the numerical result is valid. Table 2 shows
resulted from the numerical studies. They were the inlet the comparison between numerical and experimental
and outlet temperature of air flowing through the channel results at 5.0-m/s-inlet air velocity and 320 K absorber tem-
and its pressure drop. Only Mesh B could not converge and perature. The air temperature difference and pressure drop
 E.A. Handoyo et al. / Solar Energy 131 (2016) 47–60 53

500

Pressure drop (Pa)

400

300

200

100

0
Mesh A Mesh C Mesh D Exprmt

(a) The air pressure drop from numerical studies

compared to experiment
(a) Mesh A (most coarse)
315 17

Air temperature (K)

Temperature diff (K)

16.5
310
16
305 15.5
15
300
14.5
295 14
Mesh A Mesh C Mesh D Exprmt

delta_T Tin Tout

(b) The air temperature from numerical studies
compared to experiment
(b) Mesh B (slight coarse)
Fig. 4. The numerical studies of Mesh A, C, and D compared to the result
of experiment for S/H = 1.

To ensure that the model used in numerical study is

valid, a comparison between Nusselt number from numer-
ical result and Gnielinski equation for triangular channel
without obstacle or smooth duct was conducted. Accord-
ing to Incropera dan and DeWitt (2002), Gnielinski gives
more accurate relation to calculate Nusselt number at
(c) Mesh C (slight fine) lower Reynolds number than Dittus–Boelter equation.
The relation of Nusselt number is in Eq. (8):
ðf =8ÞðReD  1000ÞPr
NuD ¼ 1=2
ð8Þ
1 þ 12:7ðf =8Þ ðPr2=3  1Þ

The friction factor in Eq. (8), f, is obtained from Moody

diagram. The comparison in Fig. 7 shows that numerical
simulation gives pretty accurate results.
Having a valid grids, setting, and viscous model, the
numerical study was continued to investigate the spacing
(d) Mesh D (most fine) effect of obstacles on the bottom plate of air duct. Dia-
grams of velocity vector of air flow around the obstacles
Fig. 3. Mesh design checked for independency.
inserted at ratio S/H equals to 2, 1 ø, 1, and ø are shown
in Fig. 8(a)–(d), respectively. All of the velocity vectors
Table 1 were taken at height y = 10 mm from the bottom plate.
Number of cells, faces, and nodes of the four mesh design. When the ratio S/H = 2, the spacing between obsta-
Mesh A Mesh B Mesh C Mesh D cles = 100 mm. For ratio S/H = 1 ø, the spacing = 75 mm,
Cell 55,296 338,688 1,092,000 1,980,000 and for ratio S/H = ø, the spacing = 25 mm.
Face 173,936 1,043,224 3,335,300 6,026,900 Fig. 8(a)–(d) shows how the air flow encounters separa-
Node 63,172 365,480 1,150,798 2,066,248 tion when flow through obstacles. The obstacles cause
backflow between two consecutive obstacles in down-
of numeric is very close to experiment. Thus, the numerical stream. There are more backflows when the obstacles’
study using several setting discussed above is acceptable spacing are closer and they cause a higher pressure drop.
and valid. The air velocity looks higher near absorber plate when
54 E.A. Handoyo et al. / Solar Energy 131 (2016) 47–60

Table 2
Comparison between numerical and experimental results.
Numeric Experiment
Tin (K) 297.9 297.5
Tout (K) 314.4 313.7
delta T (K) 16.5 16.2
delta P (Pa) 462.3 450.3
Air low

(a) Rao S/H = 2 35

30

Nusselt number
25
20
15
10
5
0
0 2000 4000 6000 8000 10000 12000
Reynolds number
Air low
Nu, simulasi Nu, Gnielinski
(b)Rao S/H = 1 ½
Fig. 7. Comparison of Nusselt number for smooth duct (no obstacle)
from numerical study and Gnielinski equation.

there is backflow or swirl in the flow. This high air velocity

makes flow become more turbulent and more heat trans-
ferred from the absorber plate to the air. This convection
Air low heat transfer to the air makes outlet air temperature higher.
Thus, the closer the obstacles’ spacing makes the higher
pressure drop and higher convection heat transfer from
(c) Rao S/H = 1
the absorber plate to the air flow.
When the spacing between obstacles is twice its height
(ratio S/H = 2), small amount of air flow is blocked by
obstacles and causes backflow behind the obstacles. The
backflow caused by separation gradually diminish between
two obstacles in downstream as shown in Fig. 8(a). Since
the air is blocked, most of air is forced to flow in the gap
between obstacles and absorber plate. It is shown by longer
Air low and higher vector of velocity in the gap. More air is heated
by absorber plate. Furthermore, higher velocity increases
(d) Rao S/H = ½ Reynolds number and make the flow more turbulent. More
turbulent flow gives higher Nusselt number and definitely
Fig. 5. The Mesh used for numerical studies of each spacing to height
increase the convection heat transfer. Thus, obstacles
ratio, S/H.
makes outlet air temperature higher than without any

Fig. 6. The Mesh used for numerical studies of air flow without obstacle.
 E.A. Handoyo et al. / Solar Energy 131 (2016) 47–60 55

(a) Velocity vector of air flow around obstacle with rao S/H = 2

(b) Velocity vector of air flow around obstacle with rao S/H = 1 ½

(c) Velocity vector of air flow around obstacle with rao S/H = 1

(d) Velocity vector of air flow around obstacle with rao S/H = ½

Fig. 8. Velocity vector of air flow around obstacle with some ratio S/H.
56 E.A. Handoyo et al. / Solar Energy 131 (2016) 47–60

(Handoyo et al., 2014). Flow in spacing between obstacles Besides efficiency, Nusselt number is interesting to be
experiences backflow after hitting obstacles and then it analyzed. It could be calculated using Eq. (11).
reattaches quickly when the ratio S/H = 2. When the ratio NuD k
S/H or spacing between obstacles is smaller, as shown in h¼ ð11Þ
Dh
Fig. 8(b)–(d), backflow is more dominant than reattached
flow in area between the obstacles. When backflow occurs, While the convection heat transfer rate and log-mean
more air will flow in the gap and contact with absorber temperature difference could be calculated using Eqs. (12)
plate which is the heat source. It is the reason that smaller and (13).
obstacles’ spacing makes higher air outlet temperature and q ¼ hADT lm ð12Þ
higher SAH’s efficiency. The backflow in space between
ðT abs  T o Þ  ðT abs  T i Þ
obstacles contributes pressure drop in the flow. More back- DT lm ¼ ð13Þ
flow causes higher pressure drop in air acrossed the v- ln ðT abs T o Þ
ðT abs T i Þ
corrugated absorber plate SAH.
The ratio S/H of obstacles effect the Nusselt number of
Numerical studies also provide some global properties
air flow as shown in Fig. 10. Nusselt number of air flow
of the flow, such as the air inlet and outlet temperature,
with obstacles are much higher than without obstacle.
and pressure drop. These data were used to acquire the effi-
The highest Nu number and friction factor are 94.2 and
ciency and friction factor of the air flow in a v-corrugated
0.628 at ratio S/H = 0.5 and is only 27.2 and 0.0316 at
SAH with delta-shaped obstacles. The efficiency and fric-
no obstacle. Fig. 11 shows the enhancement of Nusselt
tion factor are calculated using Eqs. (9) and (10) according
number when obstacles inserted in the flow compare to
to Duffie dan and Beckman (1991).
air flow with no obstacle. Nusselt number enhancement
qu m_ f cp ðT fo  T fi Þ was the Nusselt number with obstacles compare to Nusselt
g¼ ¼ ð9Þ
Ac I Ac I number without obstacle. When obstacles inserted with
DP ratio S/H = 0.5, Nusselt number increases 3.46 times and
f ¼ L v2 ð10Þ friction factor increases 19.9 times compare to those with-
Dh
q2
out obstacle.
Comparison of efficiency and friction factor for some Figs. 9–11 showed the same trend of SAH’s efficiency,
ratio S/H from numerical studies is shown in Fig. 9. Ratio Nusselt number, and its enhancement. From efficiency
S/H = 0 in Fig. 9 means no obstacle in the air flow. Obsta- and Nusselt number perspective, point A is the best, but
cles arranged with ratio S/H = ø gives the highest air effi- from friction factor perspective, point D is the best. The
ciency, i.e. 68% and also highest friction factor, i.e. 0.628 gradient of friction factor in segment A–B is the sharpest
for air flowing in a v-corrugated duct, as shown in Fig. 9. among other segments. While the gradient of efficiency
For comparison, efficiency when no obstacle is only 49% and Nusselt number enhancement in segment A–B is the
and friction factor is only 0.0316. This findings are match- least. The efficiency and Nusselt number in point B is only
ing with the flow structure discussed above. The obstacles slightly less than point A. The Nu enhancement in point A
block the air flow and force it to have more contact with is 3.46 and reduce to be 3.42 in point B. Thus, the reduction
absorber plate. Not only forcing the air to the absorber is only 1.13%. While the reduction of friction factor from
plate, obstacles also causing more backflow. More obsta- point A to point B is 15.1% that is from 19.9 to 16.9. Sacri-
cles produce higher efficiency and higher friction factor. fice a little efficiency or Nusselt number (convection heat
SAH requires high efficiency or high outlet air temperature, transfer) but save a lot friction factor (pressure drop) shall
but low friction or pressure drop. Increasing friction factor be a wise choice. So, the optimal spacing ratio, S/H of
means increasing pumping power required in SAH. delta-shaped obstacles attached to a v-corrugated SAH is
one. Since the thermal efficiency, Nusselt number, and fric-
tion factor are non-dimensional parameters, the result

70 0.7
110 0.2
66 0.6
A B
C
Efficiency (%)

D 0.5 90
fricon factor, f

62
Fricon coefficient

0.15
Nusselt number

0.4
58 70
0.3 0.1
54
0.2 50
Eff, %
50 Nu
f 0.1 0.05
30 f
46 0
0.5 1 1.5 2 no obs
10 0
spacing rao, S/H 0.5 1 1.5 2 no obs

Fig. 9. The efficiency of SAH and friction factor of flow with some ratio Fig. 10. Nusselt number and friction factor of air flow with obstacles
S/H of obstacles. inserted at some spacing ratio S/H.
 E.A. Handoyo et al. / Solar Energy 131 (2016) 47–60 57

3.5 21 lader’s finding for v-corrugated absorber plate solar

3.45 A B
Nu/Nu_o 20 collector.
3.4 f/f_o 19
3.35 18
Akpinar dan and Koçyiğit (2010) conducted research on
flat plate solar collector with two air mass flow rates, i.e.
Nu/Nu_o

3.3 17

f/f_o
B
3.25 C 16 0.0074 kg/s and 0.0052 kg/s. There were four collectors
3.2 15 used as shown in Fig. 14. Three collectors were given obsta-
3.15 14
cles on absorber flat plate. The comparison of Akpinar &
3.1 D 13
3.05 12 Koçyiğit’s collector (type II) to investigated collector is
3 11 shown in Fig. 15. The efficiency of investigated collector is
0 0.5 1 1.5 2 2.5
for air mass flow rate 0.00729 kg/s. Parameter (To  Ta)/I
spacing rao S/H
of the investigated collector is slightly lower, because the
Fig. 11. Enhancement Nusselt number and increasing friction factor for air temperature across the investigated collector is lower
some spacing ratio S/H. than Akpinar & Koçyiğit collector. It is because the
dimension of investigated collector is smaller than Akpinar
& Koçyiğit’s. The area of investigated collector was only
900 mm
87 mm compared to Akpinar & Koçyiğit’s that
was 1.2 m
0.7 m. Yet, the difference shown in Fig. 15 is
not too big. It seems a good conformity between the inves-
tigated collector and Akpinar & Koçyiğit’s.
Bekele et al. (2011) conducted experimental investiga-
tion on flat plate SAH with delta-shaped obstacles
Fig. 12. A v-corrugated solar collector used by Karim & Hawlader. mounted on the absorber surface as shown in Fig. 16.
Some parameter to be investigated by Bekele et al. were
Reynolds number (from 3400 to 27,600), longitudinal pitch
could be applied to other SAH that has different geometry of the obstacles Pl/e (from 3/2 to 11/2), and relative obsta-
dimension but the same configuration. cle height, e/H (from 0.25 to 0.75). The smaller Pl/e and the
higher e/H, the efficiency is higher. The term ‘‘e” used by
5. Comparison to others’ research Bekele et al., means obstacle’s height and ‘‘H” means the
height of the channel. Bekele et al. used Pl/e while the
Karim dan and Hawlader (2006) worked on a flat plate investigated collector used S/H term. The ‘‘S” that indi-
solar collector with and without fin and v-corrugated cates spacing of obstacles is the same with ‘‘Pl” that indi-
absorber plate solar collector which schema is shown in cates longitudinal pitch (spacing). The comparison of
Fig. 12. Dimension of the collector used is 1.8 m
0.7 m. Bekele et al.’s collector and collector being investigated
The air flow rate generated by the blower in experiment (height of obstacle to channel’s height, e/H = 0.6 and spac-
of this paper that is nearest to Karim & Hawlader’s SAH is ing or longitudinal pitch, Pl/e = 1 and 3/2) is shown in
0.01164 kg/m2 s. The comparison of the efficiency of two Fig. 17.
collectors is shown in Fig. 13. For smooth duct or no obstacle installed, the Nusselt
The efficiency of SAH investigated is 0.49 and of Karim number of investigated collector is higher than Bekele
& Hawlader is 0.4, as in Fig. 13. There is a slight difference et al.’s. It is because the channel of investigated collector
because the air flow rate is not exactly the same. Therefore, is triangular. This is consistent with the finding of Tao
the finding in this paper is look alike with Karim & Haw- et al. (2007), Karim dan and Hawlader (2006),

0.6

0.5

0.4
efisiensi

0.3

0.2

0.1

0
0.00233 0.00466 0.00699 0.00931 0.01164
flow rate, kg/m2.s

(a) Collector invesgated (b) Karim & Hawlader’s collector

Fig. 13. Comparison of the efficiency of collector investigated to Karim & Hawlader’s.
58 E.A. Handoyo et al. / Solar Energy 131 (2016) 47–60

Fig. 14. Schematic views of absorber plates: (a) with triangular type obstacles, (b) with leaf type obstacles, (c) with rectangular type obstacles, and (d)
without obstacles (Akpinar dan and Koçyiğit, 2010).

0.9
0.8
0.7
0.6
Efisiensi

0.5
0.4
0.3
0.2
0.1
0
0.03 0.04 0.05 0.06 0.07
(To - Ta) / I
(a) collector invesgated (b) Akpinar & Koçyiğit’s collector

Fig. 15. Comparison of efficiency of Akpinar & Koçyiğit’s collector and investigated collector.

Fig. 16. Photographic view of plate with delta-shaped obstacles mounted.

Choudhury dan and Garg (1991), Naphon (2007), same trend, i.e. the smaller Pl/e gives higher Nusselt num-
Islamoglu dan and Parmaksizoglu (2003). Yet, the result ber. It means more obstacles increase heat transfer to the
of Bekele et al.’s and the investigated collector show the air. The relative obstacle height used in Bekele et al.’s
 E.A. Handoyo et al. / Solar Energy 131 (2016) 47–60 59

250
e/H = 0.6, Pl/e = 1

Nusselt number
200

150

100

50

0
0 2000 4000 6000 8000 10000 12000
Reynolds number
obs, Pl/e = 1 No obs

200
e/H = 0.6, Pl/e = 3/2
Nusselt number

150

100

50

0
0 2000 4000 6000 8000 10000 12000
Reynolds number
obs, Pl/e = 3/2 No obs

Fig. 17. Comparison of Nusselt number of Bekele et al.’s collector and investigated collector.

300 in the gap between obstacles and absorber plate causes

250 the flow became more turbulent and enhanced the convec-
tion heat transfer between the air and the absorber plate.
Nusselt number

200
Thus, when it is applied to SAH, it will improve its effi-
150 ciency but increase the air pressure drop.
100 Obstacles placed in a small spacing will increase Nusselt
number (convection heat transfer) and friction factor (pres-
50
sure drop). The Nusselt number enhanced from 27.2 when
0 no obstacle used to 94.2 when obstacles inserted with S/
0 5000 10000 15000 20000 25000 30000
Reynolds number H = 0.5. The Nusselt enhanced 3.46 times. The friction fac-
tor will increase from 0.0316 at no obstacle to 0.628 at ratio
e/H = 0.5, Bekele e/H = 0.75, Bekele e/H = 0.6, invesgated
S/H = 0.5. The friction factor increased 19.9 times.
Fig. 18. Comparison of Nusselt number of Bekele et al.’s and investigated Efficiency, Nusselt number, and friction factor are
collector for Pl/e = 3/2. decreasing as ratio S/H is increasing. When ratio S/H used
is 1 instead of 0.5, Nusselt number enhancement decreased
collector was 0.5 and 0.75, while in the investigated collec- only 1.13%, but friction factor decreased 15.1%. So, sacri-
tor was 0.6. The comparison of the results is shown in ficing a small amount of Nusselt number but reducing a
Fig. 18. When Reynolds number around 10,000 and ratio significant friction factor is advantageous. The optimal
Pl/e = 3/2, Nusselt number of investigated collector was spacing ratio S/H of delta-shaped obstacles inserted in a
178 for e/H = 0.6 and Nusselt number of Bekele et al. v-corrugated SAH is one. In other words, the optimal spac-
was 98 for e/H = 0.5 and 125 for e/H = 0.75. Like in col- ing of obstacle equals to its height.
lector without obstacle, the Nusselt number of investigated
collector is also slightly higher than from Bekele et al. Yet,
the result from this investigated collector is showing the
same trend with the finding of Bekele et al. Acknowledgements

6. Conclusion Here, I am very grateful for the support from Kopertis

Wilayah VII Jawa Timur, Kementerian Pendidikan dan
From numerical studies in a v-corrugated duct, it is Kebudayaan by providing Research Grant under contract
found that backflow between obstacles and high velocity No: 0004/SP2H/PP/K7/KL/II/2012.
60 E.A. Handoyo et al. / Solar Energy 131 (2016) 47–60

References Ozgen, F., Esen dan, M., Esen, H., 2009. Experimental investigation of
thermal performance of a double-flow solar air heater having
Tao, L., Wen, X.L., Wen dan, F.G., Chan, X.L., 2007. A Parametric study aluminium cans. Renew. Energy 34, 2391–2398.
on the termal performance of a solar air collector with a V-groove Akpinar dan, E.K., Koçyiğit, F., 2010. Experimental investigation of
absorber. Int. J. Green Energy 4, 601–622. thermal performance of solar air heater having different obstacles on
Karim dan, M.A., Hawlader, M.N.A., 2006. Performance investigation of absorber plates. Int. Commun. Heat Mass Transfer 37, 416–421.
flat plate, V-corrugated and finned air collector. Energy 31, 452–470. Bekele, A., Mishra dan, M., Dutta, S., 2011. Effects of delta-shaped
Choudhury dan, C., Garg, H.P., 1991. Design analiysis of corrugated and obstacles on the thermal performance of solar air heater. Hindawi
flat plate solar air heaters. Renew. Energy I (5/6), 595–607. Publishing Corporation: Advances in Mechanical Engineering 2011, 10
Naphon, P., 2007. Heat transfer characteristics and pressure drop in pages.
channel with V corrugated upper and lower plates. Energy Convers. Esen, H., Ozgen, F., Esen dan, M., Sengur, A., 2009a. Modelling of a new
Manage. 48, 1516–1524. solar air heater through least-squares support vector machines. Expert
Islamoglu dan, Y., Parmaksizoglu, C., 2003. The effect of channel height Syst. Appl. 36, 10673–10682.
on the enhanced heat transfer characteristics in a corrugated heat Esen, H., Ozgen, F., Esen dan, M., Sengur, A., 2009b. Artificial neural
exchanger channel. Appl. Therm. Eng. 23, 979–987. network and wavelet neural network approaches for modelling of a
Promvonge, P., 2010. Heat transfer and pressure drop in a channel with solar air heater. Expert Syst. Appl. 36, 11240–11248.
multiple 60° V-baffles. Int. Commun. Heat Mass Transfer 37, 835–840. Handoyo, E.A., Ichsani, D., Sutardi, Prabowo dan, 2014. Experimental
Kurtbas dan, I., Turgut, E., 2006. Experimental investigation of solar air studies on a solar air heater having V-corrugated. Appl. Mech. Mater.
heater with free and fixed fins: efficiency and exergy loss. Int. J. Sci. 493, 86–92.
Technol. 1 (1), 75–82. Incropera dan, F.P., DeWitt, D.P., 2002. Fundamentals of Heat and Mass
Romdhane, B.S., 2007. The air solar collectors: comparative study, Transfer, fifth ed. John Wiley & Sons, s.l..
introduction of baffles to favor the heat transfer. Sol. Energy 81, 139– ASHRAE, 1986. Method of Testing to Determine the Thermal Perfor-
149. mance of Solar Collectors. ASHRAE, Atlanta.
Ho, C.-D., Yeh dan, H.-M., Chen, T.-C., 2011. Collector efficiency of I. FLUENT, Fluent User’s Guide, 2003.
upward-type double-pass solar air heaters with fins attached. Int. Zhangdan, L., Liu, Z., 2009. A numerical simulation of the flow in a
Commun. Heat Mass Transfer 38, 49–56. diffusing S-duct inlet. Mod. Appl. Sci. 3 (4), 111–116.
Abene, A., Dubois, V., Le Ray dan, M., Oagued, A., 2004. Study of a Kirkgoz, M.S., Onerdan, A.A., Akoz, M.S., 2009. Numerical modeling of
solar air flat plate collector: use of obstacle and application for the interaction of a current with a circular cylinder near a rigid bed. Adv.
drying of grape. J. Food Eng. 65, 15–22. Eng. Softw. 40, 1191–1199.
Esen, H., 2008. Experimental energy and exergy analysis of a double-flow Duffie dan, J.A., Beckman, W.A., 1991. Solar Engineering of Thermal
solar air heater having different obstacles on absorber plates. Build. Processes, second ed. John Wiley & Sons Inc, penyunt.
Environ. 43, 1046–1054.