Solar Energy 157 (2017) 1095–1107

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

A parametric analysis of rectangular rib roughened triangular duct solar air MARK
heater using computational fluid dynamics
⁎
Rajneesh Kumar , Anoop Kumar, Varun Goel
Department of Mechanical Engineering, NIT Hamirpur, 177005, India

A R T I C L E I N F O A B S T R A C T

Keywords: Solar air heater (SAH) absorbs thermal radiations from Sun and utilizes it to heat air. The thermal performance
Computational fluid dynamics (CFD) of SAH can be improved by using ribs on the absorber plate. In this paper, rectangular geometry of ribs is
Heat transfer enhancement considered over the absorber plate. The fluid flow characteristics and heat transfer in ribbed triangular duct
Rectangular ribs (with an apex angle of 60°) SAH is analyzed using computational fluid dynamics (CFD). The roughened side of
Triangular duct
duct is subjected to a constant heat flux of 1000 W m−2, whereas, roughness elements are adiabatic in nature.
Solar air heater
The three-dimensional model of SAH is developed and numerical simulations are carried out by developing CFD
code with the help of finite volume method. The numerical simulations are performed on commercial ANSYS
Fluent 12.1 software. A new roughness parameter called rib aspect ratio (e/w) is introduced in this study and its
effect on friction factor (f) and Nusselt number (Nu) is investigated along with relative roughness height (e/D)
and relative roughness pitch (P/e) for the Reynolds number ranges from of 4000 to 18,000. The values of rib
aspect ratio (e/w), roughness height (e/D) and relative roughness pitch (P/e) range from 0.25 to 4.0, 0.02 to 0.04
and 5 to 15, respectively. A considerable increase in friction factor (f) and Nusselt number (Nu) is observed due
to rectangular rib in comparison to smooth one. The numerically predicted results are compared with the
available results and a good agreement between them is observed with maximum error of ± 4.04%. A significant
variation in friction factor (f) and Nusselt number (Nu) is observed by varying rib aspect ratio (e/w) values from
0.25 to 4.0. The maximum value of thermohydraulic performance parameter (TPP) is found 1.89 in case of
relative roughness pitch (P/e), rib aspect ratio (e/w) and relative roughness height (e/D) value of 10, 4.0 and
0.04, respectively, at the Reynolds number (Re) of 15,000.

1. Introduction
Solar air heater (SAH) is a device which stores solar radiations for
heating applications. The air acts as a working medium in SAH. But,
due to its low coefficient of heat transfer value, SAH has poor perfor-
mance (thermal). The different heat transfer augmentation techniques
such as surface, fluid and combined were discussed by Balaras (1990)
and Bergles (1988). Among them the heat transfer augmentation using
surface methods is most effective and economic technique (Yadav and
Thapak, 2014). In this technique, the artificial roughness is placed over
the hot surface and because of these roughness elements, generation of
local turbulence takes place which increases heat transfer along with
pressure drop in the SAH (Iacovides and Raisee, 2001; Varun et al.,
2007; Kumar et al., 2017a). For getting optimum thermal performance,
the transition, sub-layer thickness should be smaller than the height of
roughness element (Prasad and Saini, 1991).
The orientation and shape of duct are the two parameters which
affects the flow characteristics and heat transfer through the duct
(Kumar et al., 2016). The triangular cross-sectional duct has minimum
friction factor value in comparison to other cross-sectional duct (square,
circular, rectangular and semi-circular) and therefore, triangular
shaped duct required low pumping power for fluid flow (Cebeci and
Bradshaw, 1988). Due to advantage of low pumping power, the trian-
gular duct is used in many engineering applications such as compact
heat transfer (Zhang, 2005; Sunden, 2010; Nascimento and Gracia,
2016; Chennu, 2016), SAH (Bharadwaj et al., 2013, 2017), etc.
A lot of experimental and numerical studied are available on
roughened rectangular duct in comparison to triangular duct SAH (Tsai
et al., 2000; Gao and Sunden, 2001; Unalan et al., 2007; Liou et al.,
2007; Ma et al., 2015; Raisee and Rokhzadi, 2009; Sethi et al., 2012;
Singh et al., 2014, 2015; Yadav et al., 2014; Scholl et al., 2016; Jiang
et al., 2016; Pandey et al., 2016; Kumar et al., 2017b). In case of tri-
angular duct SAH only few studies are available in the area of heat
transfer augmentation using artificial roughness. Chegini and
Chaturvedi (1986) studied the effect of fin location on friction factor in
duct and observed that placing of fins on same side of the duct results

⁎
Corresponding author.
E-mail address: rajneesh127.nith@gmail.com (R. Kumar).

http://dx.doi.org/10.1016/j.solener.2017.08.071
Received 6 February 2017; Received in revised form 19 July 2017; Accepted 25 August 2017
0038-092X/ © 2017 Elsevier Ltd. All rights reserved.
R. Kumar et al. Solar Energy 157 (2017) 1095–1107

Nomenclature ρ density (kg m−3)

ε dissipation rate
a side length of the duct (m) k turbulent kinetic energy (m2 s−3)
D hydraulic diameter (m) α thermal conductively (W m−1 K−1)
k thermal conductivity (W m−1 K−1) Γ molecular thermal diffusivity (=μ/Pr)
h coefficient of heat transfer (air) (W m−2 K−1) Γt turbulent thermal diffusivity (=μt /Pr)
H height of duct (m)
L length (m) Subscripts
n number of ribs
P pressure (Pascal) avg average
Pwetted wetted perimeter (m) eff effective
T temperature (K) enh enhancement
u axial velocity (m s−1) L0 start of test section
W width of duct (m) penalty penalty
w width of rectangular rib roughness element (m) p plate
x distance in the flow direction (m) r ribbed
s smooth
Dimensionless test test
test,0 start of test section
e/D relative roughness height test,l end of test section
f friction factor t total
I turbulence intensity
Nu Nusselt number Abbreviations
P/e relative roughness pitch
Pr Prandtl number ASHRAE American Society of Heating, Refrigeration Air
Re Reynolds number Conditioning Engineers
e/w rib aspect ratio CFD Computational Fluid Dynamics
y+ dimensionless wall distance ICEM Integrated Computer-aided Engineering and
Manufacturing
Greek symbols RNG Renormalized group method.
SAH Solar Air Heater
µ viscosity (N s m−2) TPP Thermohydraulic performance parameter
µt thermal viscosity

low friction factor in comparison to three side finned duct, under si-
milar conditions. It is observed that fins can result in heat transfer
improvement up to 200% under particular conditions but it increases
pressure drop along the flow considerably. Similarly, Obot et al. (1987)
analyzed the effect of roughness on pressure drop in round corner tri-
angular duct. They found that pressure drop is higher in case of three
side roughened duct in comparison to two sides. Due to presence of
laminar as well as turbulent flow at the corner region, the turbulence is
suppressed by flow distribution around corners (Daschiel et al., 2013;
Karabulut et al., 2016) which results in low heat transfer in triangular
duct (Sparrow and Haji-Sheikh, 1965). Even though, micro size
roughness element can results a better heat transfer in the duct (Kang
et al., 1998). Kang et al. (1998) has used rolling and shaping process to
create micro sized roughness inside the duct. The maximum roughness
size was of the order of 11.5 μm and they had observed maximum in-
crease of Nusselt number (Nu) in the range of 15–30% in comparison to
smooth duct. The roughness can be created by producing groove in the
fluid flow region. Leung et al. (2000) had examined the effect of v-
groove roughness. They had considered constant value of roughness
pitch and groove depth which is 34 mm and 1 mm, respectively. But the
groove angle varied from 0° to 150°. The maximum value of Nu is ob-
tained in particular range of groove angle which is 15° to 18°. The effect
of square ribs on performance of duct has been studied by Leung et al.
(2001), Ahn and Son (2002) and Luo et al. (2004). Leung et al. (2001)
had analyzed the effect of rib height (range from 6.35 to 12.7 mm) by
considered fixed vale of roughness pitch i.e. 57.15 mm. The Nu in-
creased by amount of 170% in roughened duct as compared to smooth.
Ahn and Son (2002) had analyzed the effect of relative roughness pitch
(range from 4 to 16) for constant rib height value i.e. 2 mm. They had
concluded that the friction factor increases with decrease in gap
between two successive ribs and best thermal performance is achieved
for relative roughness pitch (P/e) value of 10. The combined effect of
relative roughness (P/e) and relative rib height (H/D) has been studied
by Luo et al. (2004). The P/e and e/D value range from 3.41 to 13.93
and 0.11 to 0.21, respectively. The maximum increase in Nu is observed
for H/D and P/e value of 0.18 and 7.22. The flow and heat transfer
through the duct also get affected by the rib inclination angle. The ef-
fect of rib inclined angle on flow and heat transfer in SAH is analyzed
by Bharadwaj et al. (2013). The rib inclination angle, e/D and P/e was
varied from 30° to 60°, 0.021 to 0.043 and 8 to 16, respectively. The
maximum increase in Nu and friction factor (f) is observed for rib in-
clination angle of 60°. A non-dimensional correlation has been also
developed for inclined ribbed SAH (Bharadwaj et al., 2017). The
maximum increase in Nu and f is found for rib inclination angle, P/e
and e/D value of 60°, 12 and 0.043, respectively. The heat transfer in
semi-circular rib roughened triangular duct has been investigated by
Kumar et al. (2017a). The commercial ANSYS software has been used
for the numerical analysis. They found that k-ε model calculate Nu
accurately as compared to other available turbulence models in Fluent.
The best thermohydraulic performance is observed of the order of 1.7
for e/D and P/e value of 0.067 and 7.5. From discussed literature, it is
concluded that rib roughness has significant effect on Nu and f. Dif-
ferent shaped rib elements (such as circular, square, V-grooved, semi-
circular has been analyzed by different researchers and Nu and f value
changes remarkable by varying the shape of rib elements. There is no
such study which analyse the effect of rectangular ribs on thermal
performance of triangular duct SAH. Therefore, in this article numerical
study has been performed on rectangular ribbed SAH with the help of
computational fluid dynamic (CFD) based software.
Computational fluid dynamics (CFD) is a powerful tool to analyze

1096
R. Kumar et al.
heat transfer through ducts by solving mass, momentum and energy
conservation equations with the help of finite volume method (Versteeg
and Malalasekera, 2007; Date, 2005). With the help of commercial
ANSYS software, an attempt has been made to analyze flow char-
acteristics and heat transfer in rectangular rib roughened triangular
duct. A new roughness parameter is introduced in this article called rib
aspect ratio (e/w) and its effect is studied on friction factor (f) and
Nusselt number (Nu) along with relative roughness pitch (P/e) and
relative roughness height (e/D) for Reynolds number (Re) range of 4000
to 18,000. The objective of the study is to analyze the effect of different
roughness parameters on fluid flow characteristics and heat transfer
augmentation in SAH. The results are generalized by developing cor-
relation for Nusselt number (Nu) and friction factor (f). The effective-
ness of the roughness elements is studied by calculating thermo-
hydraulic performance parameter (TPP) for different combinations of
roughness parameters. It is observed that rectangular rib element sig-
nificant improve heat transfer augmentation along with friction pen-
alty. But in studied range of roughness parameters, TPP observed higher
than the flat absorber plate SAH.
2. Description of problem
The schematic diagram of ribbed triangular duct is shown in Fig. 1a.
The triangular duct has side length of 52 mm, height 45 mm and apex
angle of 60°. The entrance portion is made long enough (10 times the
hydraulic diameter) for fully developed flow conditions before the test
section (Incropera et al., 2013). On the basis of hydraulic diameter of
duct, the Reynolds number (Re) varies from 4000 to 18,000. The base
side of triangular duct is heated with 1000 W m−2 heat flux and ribs are
placed over it. The schematic diagram of 3-D model of triangular duct
SAH with different sections and roughened absorber plate is shown in
Fig. 1b. The problem is simulated numerically with following assump-
tion;
i. Flow is incompressible, steady and turbulent.
ii. The fluid has constant properties throughout the length of duct.
iii. Air is taken as working medium and it enters the computational
domain at ambient conditions (300 K). The value of properties at
ambient condition is given in Table 1.
iv. No-Slip conditions at wall.
Fig. 1a. Schematic diagram of ribbed triangular duct.
1097
Solar Energy 157 (2017) 1095–1107
2.1. Numerical modeling of roughened duct
The roughened triangular duct is modeled using commercial ANSYS
software. The different models of duct are developed depending on
roughness parameters. The computational model is developed in Design
Modeler, mesh is generated in ICEM (Integrated Computer-aided
Engineering and Manufacturing) CFD and flow governing equations are
solved using ANSYS Fluent to investigate thermohydraulic perfor-
mance.
2.1.1. Computational domain and meshing
The three dimensional model of triangular duct is developed using
Design Modeler available in commercial ANSYS software as shown in
Fig. 2. The computational domain is a horizontal triangular duct with
length of entrance and length of exit portion greater than 5√(Width3-
Height) and 2.5√(Width3Height), respectively as per ASHRAE Standard
(2003). The ribs are placed over the hot surface in such as way that flow
strikes perpendicular to the ribs. The geometric and roughness para-
meters are given in Table 2.
The meshing of computational domain is carried out using ICEM
CFD which is a meshing module of commercial ANSYS software and
meshed model of SAH is shown in Fig. 3. The structured non-uniform
quadrilateral grids are generated in computational domain and highly
dense grids are placed near the hot surface where ribs are placed in
order to resolve laminar sub-layers accurately. The quadrilateral grids
are spaced by y+ ≈ 2 from roughened hot surface to predict effect of
viscous sub-layer (Yadav and Bhagoria, 2014).
2.1.2. Grid independent test
The grid independence test is done to study the influence of grid size
on Nu and f. The simulation was started from coarse grid size with
394,713 elements and refined up to 915,785 elements for predicting Nu
and f value. In the present work, the grid independence test is per-
formed for e/D, P/e value 0.03 and 10, respectively, at Reynolds
number of 12,000. The absorber plate temperature (Tp) and pressure
drop (ΔP) through the SAH varied significantly with grid refinement
from 394,713 to 915,785. The detail of grid independence test is given
in Table 3. The percentage deviation in Tp and ΔP value is less than 1%
after 821,667 elements. Therefore, analysis is done with 821,667 ele-
ments for all configurations of different roughness parameters.
R. Kumar et al. Solar Energy 157 (2017) 1095–1107

Fig. 1b. 3-D model of computational domain.

Table 1 Table 2
Properties of working fluid i.e., air. Details of geometric and roughness parameters.

S. no. Properties Value S. no. Parameters Range/value

−1 −1
1. Thermal conductivity (W m K ), k 0.024 Flow parameters
2. Density (kg m−3), ρ 1.225 1. Reynolds number (Re) 4000–18,000
3. Dynamic viscosity (N m−2), μ 1.783 10−5
Roughness parameters
4. Specific heat (J kg−1 K−1), Cp 1006
2. Relative roughness pitch (P/e) 5–15
3. Relative roughness height (e/D) 0.02–0.4
4. Relative roughness width (e/w) 0.25–4.0
2.1.3. Governing equations
Geometric parameters
The motion of fluid is modeled with the help of continuity, mo-
5. Length of triangular duct (Ltotal) 0.830 m
mentum and energy equations. These can be expressed as, 6. Length of inlet section (Linlet) 0.350 m
Continuity equation; 7. Length of test section (Ltest) 0.300 m
8. Length of outlet section (Loutlet) 0.180 m
∂ (ρuj ) 9. Apex angle of duct 60°
=0 10. Hydraulic diameter (D) 0.030 m
∂x j (1) 11. Roughness shape Rectangular shaped

Momentum equation;
∂ ∂ ⎡ ∂T ⎤
∂ ∂P ∂ ⎡ ⎛ ∂ui ∂uj ⎞ ⎤ ∂ ⎡ ⎛ ∂ui ∂uj ⎞ ⎤ (ρuj T )− (Γ + Γ)
t =0
(ρui uj ) + = μ⎜ + ⎟ + μ ⎜ + ⎟ ∂x i ∂x j ⎢ ∂x j ⎥ (3)
∂x j ∂x i ∂x j ⎢ ⎝ ∂x j ∂x i ⎠ ⎥ ∂x j ⎢ t ⎝ ∂x j ∂x i ⎠ ⎥ (2)
⎣ ⎦
⎣ ⎦ ⎣ ⎦
Turbulence is mathematically modeled by solving Navier-Stokes
Energy equation;

Fig. 2. Meshed computational domain of duct.

1098
R. Kumar et al. Solar Energy 157 (2017) 1095–1107

equations. It is quite important to use appropriate turbulence model for

modeling turbulent behavior of the fluid in order to predict accurate
results (ANSYS Fluent, 2003). In the present work turbulence is mod-
eled by using k - ε model which is based on renormalization group
(RNG) theory. The k and ε shows turbulent kinetic energy and dis-
sipation rate respectively. The equations used to present transport
phenomena are given below;

∂ (ρkui ) ∂ ⎛ ∂k ⎞
− ⎜αk μeff ⎟−Gk + ρε = 0
∂x i ∂x j ⎝ ∂x j ⎠ (4)

∂ (ρεui ) ∂ ⎛ ∂ε ⎞ ε ε2
− ⎜α ε μ ⎟−C1ε Gk + C2ε ρ = Rε
∂x i ∂x j ⎝ eff ∂x j ⎠ k k (5)

where αk , αε , C1ε and C2ε are constant with value of 1.39, 1.39, 1.42,
Fig. 3. Comparison of Nuavg and favg with the available results of Luo et al. (2004). 1.68. In Eqs. (4) and (5), Gk shows the generation of turbulence kinetic
energy.
Table 3
Grid independence test at e/D = 0.03 and P/e = 10. 2.1.4. Boundary conditions and solution strategy
Reynolds No of Hot wall Pressure Percentage Percentage
The computational domain is designed in such a way that air enters
number elements temperature; drop in deviation in deviation in the duct with certain velocity depending on Reynolds number (Re).
Tp (K) test Tp (K) ΔPtest Thus, the inlet of duct is considered as velocity inlet boundary with
section; (Pascal) velocity variation from 1.9 to 11.7 m s−1. The outlet is assumed as
ΔPtest
pressure outlet with zero gauge pressure value. The one side of the duct
(Pascal)
is roughened with non-conducting ribs and is heated by maintaining
12,000 394,713 345.721 25.145 0 0 constant heat flux of 1000 W m−2. Whereas, other sides of duct are
12,000 510,593 332.847 24.169 3.87 4.04 insulated. The intensity of turbulence for fully developed flow is esti-
12,000 655,405 328.046 23.485 1.46 2.91
mated by following empirical formula (ANSYS Fluent, 2003).
12,000 762,060 324.469 23.158 1.10 1.41
12,000 821,667 322.487 23.089 0.61 0.29 0.16
12,000 915,785 322.265 23.074 0.07 0.08 I=
(Re )1/8 (6)

It is assumed that with increase in air temperature, the thermal

Fig. 4. Variation of Nuavg with Re for different P/e values.
conductivity of working fluid (i.e. air), duct and roughness material
remains constant. At the interface of solid and fluid bodies, no-slip
conditions were considered because in this region fluid has relatively
zero velocity to the solid body (Chaube et al., 2006). The computational
(solution) domain is sub-divided into small grid of finite number and
governing equations such as continuity, momentum and energy are
solved at all grid points to predict the heat transfer. The numerical
equations are solved to a precision of four decimal places and condition
for convergence is 10−4. The governing equations are discretized using
segregated grid method and solved using upwind scheme of second
order. The pressure and viscosity linked equation are solved using semi-
implicit method for pressure linked equation (SIMPLE) algorithm de-
veloped by Patankar (1980).
The turbulence is an important parameter and need to be model
accurately in order to simulate heat transfer in the duct. The selection
of turbulence model depends on flow type and available computational
facility (ANSYS Fluent, 2003). The Reynolds-averaged Navier-Stokes
models are mostly used instead for large eddy simulation models for
modeling turbulence heat transfer in duct (Rokni and Gatski, 2001).
The flow turbulence is modeled with the help of RNG k-ε turbulence
model and accuracy of solution is ensured by using enhanced wall
function for viscous heating and thermal effects. It is concluded from
the literature that the k-ε model predicts results more accurately than
other available models in commercial ANSYS software for flow through
duct (Ramsak and Skerget, 2008; Karmare and Tikekar, 2010; Kumar
et al., 2017a). Thus, the present analysis is carried out using the k-ε
turbulent model.

2.2. Model validation

The numerical three dimensional model of ribbed triangular duct is

Fig. 5. Variation of favg with Re for different P/e values. developed using CFD code in commercial ANSYS (12.1) software to
predict Nu and f values for Re range of 4000–18,000. The numerically
predicted results are validated by comparing Nu and f with the

1099
R. Kumar et al. Solar Energy 157 (2017) 1095–1107

Fig. 6. Formation of reattachement points due to rectangulatr ribbed SAH because of different P/e values; (a) P/e = 5, (b) P/e = 10, and (c) P/e = 15.

Fig. 8. Variation of local Nu with axial distance for different relative roughness pitch
Fig. 7. Variation of local turbulent intensity with the flow. values.

experimental findings of Luo et al. (2004). For validation, the roughness observed that numerically predicted Nu and f has good match with the
parameters such as P/e, e/D and e/w is fixed at 10, 0.11 and 1.0, re- experimental results within the percentage error of ± 3.87 and ± 4.04
spectively. The comparison of results is shown in Fig. 3 and it is in Nu and f values, respectively. Thus, it is considered that the estimated

1100
R. Kumar et al. Solar Energy 157 (2017) 1095–1107

Fig. 12. Variation of favg with Re for different e/D values.

Fig. 9. Variation of Nuenh with Re for different P/e values.

Fig. 13. Variation of Nuenh with Re for different e/D values.

Fig. 10. Variation of fpenalty with Re for different P/e values.

Fig. 14. Variation of fpenalty with Re for different e/D values.

Fig. 11. Variation of Nuavg with Re for different e/D values.
Aeff
D=4×
numerical results methodology used is reliable and can be used for Pwetted (7)
further analysis. where Aeff and Pwetted are area and wetted perimeter of the duct,
respectively.

3. Data reduction Aeff = Aduct −Aribs (8)

where Aduct and Aribs is the area of triangular duct and ribs, re-
The hydraulic diameter of the triangular duct is calculated as,
spectively. But, it is seen that the value of Aribs is negligible as compared

1101
R. Kumar et al. Solar Energy 157 (2017) 1095–1107

number (Nu) in roughened triangular duct SAH.

4.1. Effect of roughness parameters

The variation of Nuavg and favg with Re for different relative rough-
ness pitch (P/e) values from 5 to 15 are shown in Figs. 4 and 5, re-
spectively, keeping value of relative roughness height (e/D) and rib
aspect ratio (e/w) constant at 0.067 and 2.0, respectively. The Nu in-
creased with increase in Re and attained maximum value at Re value of
18,000. However, f decreased with increase in Re. Moreover, the heat
transfer increased by providing ribs on the heated surface as compared
to smooth duct but Nu decreased with increase or decrease in P/e value
beyond 10. The flow reattachment over the hot surface plays an im-
portant role in heat transfer (Kumar et al., 2017a,b). The formation of
reattachment points due to rectangular ribs is shown in Fig. 6(a)–(c). It
Fig. 15. Variation of Nuavg with Re for different e/w values. is seen that the pattern of streamlines between two successive ribs re-
markably varies by changing P/e value from 5 to 15. The reattachment
to Aduct. Therefore, Aeff is assumed to be equal to the Aduct. The ex- or breaking of laminar sub-layers due to ribs resulted in better thermal
pression for calculating Aeff is, performance. It is observed from the Fig. 6 that there is no reattachment
point formation between ribs in case of P/e value of 5. But, for P/e value
Aeff = Aduct (9) of 10, there is formation of reattachment point which occurs just before
Similarly, the perimeter of the duct is expressed as, the adjacent rib. Opposite to this, in case of P/e value of 15, the re-
attachment point occurs in between two ribs and after that the flow
and Pwetted = 3 × a (10) attached (laminar sublayers formation) to the hot surface which re-
where a is side length of the duct. sulted in reduction of heat transfer and therefore, for the constant test
The air velocity through the duct is calculated in terms of Reynolds section length, the number of reattachment points reduced due to in-
number (Re) and it is calculated as, crease in the spacing between the ribs. For getting a better heat transfer
from the heated surface, the ribs should be spaced sufficiently apart
ρuD from each other, so that, the reattachment point occurs just before the
Re =
μ (11) adjacent rib (as it is seen in case of P/e value of 10). Because of more
where ρ, u and μ is density, velocity and viscosity of air. reattachment points, the local turbulence promotes better interaction of
Nu is calculated by using equation given below; main stream with the flow adjacent to the hot surface. Therefore, the
maximum Nu is observed for P/e value of 10. The variation of local
Nuavg =
1
∫L
Ltest ,l − nw
⎛ havg D⎞ dx
⎜ ⎟
turbulent intensity with the flow for starting few ribs in the test section
x test ,o
⎝ k ⎠ (12) has been shown in Fig. 7 for different P/e values. It is seen that the
maximum increase in local turbulent intensity is found for P/e value of
The f is also a non-dimensional number and it is calculated using the
5, whereas, minimum increase value is observed for P/e value of 15.
amount of pressure drop in flow direction. Mathematically it is com-
Due to the less spacing between the adjacent ribs, formation of eddies
puted as,
due to the ribs is higher (because the number of ribs are more and every
1 ΔP rib results in formation of eddies) and that can lead to pressure loss in
favg = ×D×
2u2ρ Ltest (13) the flow. Therefore, it can be concluded that the small value of P/e
lends to higher pressure drop which results higher friction factor values
The enhancement in Nusselt number (Nuenh) is calculated using the
as it seen in Fig. 5. The variation of local Nu is also studied to analyze
following expression,
the effect of P/e in more detail. The variation of local Nu with axial
Nuavg,r distance for different P/e values has been shown in Fig. 8. The distance
Nuenh =
Nuavg,s (14) between two successive ribs gets changed by varying P/e values. In
present study, the distance between two successive ribs for P/e value of
Similarly, increase in friction factor penalty (fpenalty) is numerically 5, 10 and 15 is 10, 20, 30, respectively. It is observed form the figure
calculated as: that when flow passed the rib than it resulted in higher local Nu value
favg,r and with the increase of axial distance it start decrease. A minimum
fpenalty = value of local Nu is achieved at the point of flow reattachment. But after
favg,s (15) occurrence of flow reattachment, Nu value again starts increasing and
Thermohydraulic performance parameter is evaluated by com- achieve maximum value when flow just about to reach the adjacent rib.
paring Nu and f values of roughened and smooth duct as given by Webb Therefore, Nu and f values get changed due to variation of P/e.
and Eckert (1972). A significant enhancement in Nu and f is observed by using rec-
0.33
tangular rib roughness and it is shown in Figs. 9 and 10, respectively.
Nuavg,r ⎛ favg,r ⎞ The maximum enhancement in Nusselt number (Nuenh) is seen in case of
Thermohydraulic performance parameter (TPP) = ⎜f ⎟
Nuavg,s P/e value of 10 in comparison with smooth duct SAH. With increase of
⎝ avg,s ⎠
P/e ratio from 5 to 10, Nuenh value increased. But with increase in P/e
(16)
value beyond 10 to 15, Nuenh started decreasing. Nu increased by 2.70
times the smooth duct SAH in case of P/e value of 10, whereas,
4. Results and discussion minimum enhancement of 2.02 is observed for P/e value of 15. Al-
though, Nu increased with increase in Re but maximum enhancement in
This study is performed to analyze the effect of roughness para- Nu is observed at Re of 15,000 instead of 18,000. Fig. 8 shows the fpenalty
meters such as rib aspect ratio (e/w), relative roughness pitch (P/e) and in roughened absorber plate SAH as compared to smooth one. The
relative roughness height (e/D) on fluid characteristics and Nusselt maximum fpenalty value is observed in case of P/e value of 5 and it is

1102
R. Kumar et al.
Fig. 16. Variation of streamlines pattern for different e/w values; (a). e/w = 4.0; (b) e/w = 2.0; (c) e/w = 1.0; (d) e/w = 0.67; (e) e/w = 0.25.
approximately 4.93 times the smooth absorber plate SAH at Re of 4000.
Moreover, the fpenalty value increased with decrease in Re from 18,000
to 4000.
Figs. 11 and 12, show variation of Nuavg and f with different Re
values, respectively. The value of e/D in this case, ranges from 0.02 to
0.04 and P/e and e/w value is taken to be constant at 10 and 2.0, re-
spectively. It is observed that Nu increased with increase in Re. More-
over, Nu is also a function of e/D and with increase of e/D value, Nu
increased. The maximum value of Nu is found for e/D value of 0.04 and
1103
Solar Energy 157 (2017) 1095–1107
it decreased by decreasing e/D value from 0.04 to 0.02. The laminar
sub-layer formation takes place over the hot surface which reduces the
heat transfer with main stream (Incropera et al., 2013; Varun et al.,
2007) but the presence of roughness breaks this laminar sub-layer and
creates local turbulence which improves heat transfer through duct
(Prasad and Saini, 1991).
Fig. 12 shows variation of f with Re for different e/D values. For a
given e/D value, f decreased with increase in Re. The maximum value of
f is observed in case of higher e/D value of 0.04. f increases due to the
R. Kumar et al.
Fig. 17. Variation of favg with Re for different e/w values.
Fig. 18. Variation of Nuenh with Re for different e/w values.
Fig. 19. Variation of fpenalty with Re for different e/w values.
presence of roughness in flow direction which results in high pressure
drop as compared to smooth duct. The maximum increase in f is ob-
served to be 4.61 times more than that for smooth duct.
Due to roughness both Nu and f values get increased in comparison
to smooth duct. Figs. 13 and 14 show enhancement in Nu (Nuenh) and
penalty in f (fpenalty), respectively. It is observed that Nuenh value in-
creased with increase in both Re and e/D values. The maximum Nuenh is
found for e/D value of 0.04. More importantly, Nuenh is maximum at Re
value of 15,000 as compared to other studied values (i.e. 4000, 8000,
12,000 and 18,000). Moreover, Nu increased with increase in Re but
Nuenh increased with the increase in Re up to 15,000 and further in-
crease in Re resulted in decrease of Nuenh. Fig. 14 shows variation of
fpenalty with e/D. In case of low e/D (=0.02) value, fpenalty is
1104
Solar Energy 157 (2017) 1095–1107
comparatively very low as compared to high e/D (=0.04) value. The
fpenalty increased with increase in e/D and achieved maximum value at
low Re (4000).
The effect of varying rib aspect ratio (e/w) is also analyzed on f and
Nu. During study of e/w, the value of e/D and P/e is kept constant at
0.04 and 10, respectively. The variation of Nu with Re for different e/w
values which ranges from 0.25 to 4.0 is shown in Fig. 15. It is seen that
variation of e/w value significantly affects Nu with increase in Re. For a
given value of Re, maximum Nu value is observed for e/w value of 4.
This is due to non-conducting rectangular ribs which were provided
over the absorber plate of equilateral triangular duct SAH. The area of
absorber plate which is under the ribs is not able to take part in heat
transfer. This can be evidenced from the Fig. 16(a)–(e) and it can be
seen from the streamline pattern that the variation of e/w value
strongly influenced the flow behavior inside the duct. Moreover, with
the increase of e/w ratio, for constant e, w decreased and more area is
available for heat transfer is available for heat transfer which results in
higher Nu value. There should be one limiting situation at which the
width of the roughness is long enough that it covers the entire absorber
plate and therefore, no heat transfer takes place through the absorber
plate.
Fig. 17 shows the variation of f with Re for different e/w values. It is
observed that for a given value of Re, maximum f is observed for e/w
value of 4. A significant increase in f is seen as compared to smooth duct
which is maximum at low Re value of 4000. But with the increase in Re,
f value decreased. With increase in e/w ratio from 0.25 to 4, an ap-
proximate 48.6% increase in f is observed at Re value of 4000.
Figs. 18 and 19 show variation of Nuenh and fpenalty for different
values of e/w. The maximum Nuenh value is observed for e/w of 4.0
which is approximately 44% and 137% higher than the e/w value of
0.25 and smooth duct SAH, respectively. Similarly, a significant in-
crease is observed in fpenalty value which is 4.5 times higher for e/w
value of 4 in comparison to smooth duct SAH.
The thermohydraulic performance parameter (TPP) is evaluated to
analyze the effectiveness of roughness geometry. Generally, the value of
TPP should be greater than one for roughened duct (Webb and Eckert,
1972). In present study, thermohydraulic performance is calculated
separately for different roughness parameters i.e. e/D, P/e and e/w.
With the increase in Re, Nu increases and f decreases. But the TPP is a
function of both Nusselt number and friction factor. Table 4 shows the
variation of TPP with Re for different roughness parameters. It is ob-
served that the maximum TPP value is observed in case of P/e, e/D and
e/w value of 10, 0.04, and 4.0, respectively. TPP value increased with
increase in e/D (from 0.02 to 0.04) and e/w (from 0.25 to 4.0) but
further increase or decrease of P/e beyond 10 results in decrease of TPP
values. Moreover, the value of TPP increased with increase in Re from
4000 to 15,000 and achieves maximum value at Re of 15,000 but fur-
ther increase in Re resulted in decrease of TPP value in case of all
studied values of P/e, e/D and e/w values. The percentage rise in Nu is
more at Re value of 15,000 as compared to 18,000. Therefore, the
maximum TPP value is observed maximum at Re of 15,000. The var-
iation of Nu at Re values of 15,000 and 18,000 is very minimal and
because of this variation in Nu, the Nuenh value is slightly higher in case
of Re of 15,000 in comparison to 18,000 as it can be seen in Table 4.
Therefore, the TPP value is higher at Re of 15,000 in comparison to
18,000.
4.2. Correlation development
The development of correlation is an important process to quantify
the association between the Nu, flow and roughness parameters. It is
observed in Section 4 that the predicted value of Nu and f significantly
varies with the change of both flow and roughness parameters. The
objective of correlation development is to generate a generalized
equation for Nu and f under considered range of parameters by doing
regression analysis. The correlation for Nu and f is developed doing
R. Kumar et al. Solar Energy 157 (2017) 1095–1107

Table 4
Variation of thermohydraulic performance parameter (TPP) in different case of different configuration of roughened absorber plate SAH.

S. no. P/e e/D e/w Re

4000 8000 12,000 15,000 18,000

1 10 0.02 2.00 1.35253 1.44116 1.54204 1.59525 1.60477

2 10 0.03 2.00 1.401786 1.518481 1.647917 1.716364 1.72498
3 10 0.04 2.00 1.425612 1.588043 1.731022 1.776171 1.779529
4 5 0.03 2.00 1.369097 1.486882 1.618498 1.662824 1.66045
5 10 0.03 2.00 1.401786 1.518481 1.647917 1.716364 1.69498
6 15 0.03 2.00 1.342129 1.466622 1.610073 1.677439 1.67105
7 10 0.04 0.25 1.20599 1.47372 1.645274 1.728181 1.702376
8 10 0.04 0.67 1.280719 1.484151 1.648743 1.731977 1.715501
9 10 0.04 1.00 1.405505 1.56434 1.71951 1.767583 1.762327
10 10 0.04 2.00 1.425612 1.588043 1.731022 1.776171 1.779529
11 10 0.04 4.00 1.49779 1.637535 1.775883 1.894407 1.793334

Fig. 20a. Variation of Nuavg with different Re values.

Fig. 20d. Variation of Nuavg/(C5 × ReC(P/e)−C3 × exp(C2(ln(P/e))2) × (e/w)C7with dif-

ferent e/D values.

Fig. 20b. Variation of ln(Nuavg/ReC) with different ln(P/e) values.

Fig. 21a. Variation of favg with different Re values.

regression analysis. It is a statistical process for determining relation

among variables. The relationship is developed between dependent
variable (average Nusselt number; Nuavg) and average friction factor;
favg) and independent variables (roughness and flow parameters). With
the help of correlation, the optimum thermohydraulic performance can
be predicted easily under the influence of specific range of parameters.
In this particular study, Nu and f are two parameters which significantly
varies with the value of Re, P/e, e/w, and e/D. Therefore, Nu and f are
considered as a function of Re, P/e, e/w, and e/D as shown in Eqs. (11)
and (12).
Fig. 20c. Variation of Nuavg/(C5 × ReC(P/e)−C3 × exp(C2(ln(P/e))2 with different e/w
P e e
values. Nu = fxn ⎛Re, , , ⎞
⎝ e w D⎠ (17)

1105
R. Kumar et al. Solar Energy 157 (2017) 1095–1107

Fig. 21b. Variation of ln(favg/ReC′)with different ln(P/e) values.

Fig. 22a. Variation of Nuavg value calculated numerically with Nuavg predicted from
correlation.

Fig. 21c. Variation of favg/(C′5 × ReC′(P/e)C3′×exp(C′2(ln(P/e))2)) with different e/w

values. Fig. 22b. Variation of favg value calculated numerically with favg predicted from corre-
lation.

⎡ P −2.294 e 0.130 e 0.378

favg = 3.394 × 102 ⎢Re−0.56 ⎛ ⎞ ⎛ ⎞ ⎛ ⎞
⎣ ⎝e⎠ ⎝w⎠ ⎝D⎠
2
P ⎤
× exp ⎧0.408 ⎛ln ⎛ ⎞ ⎞ ⎫ ⎥
⎨
⎩ ⎝ ⎝ e ⎠⎠ ⎬⎭⎦ (20)

⎧ 4000 ⩽ Re⩽18,000
⎪5 ⩽ P ⩽ 15
⎪ e
for the range e
⎨ 0.25 ⩽ w ⩽ 4.0
⎪ e
⎪ 0.02 ⩽ D ⩽ 0.04
⎩
The variation of predicted values (using correlation) of Nuavg and
favg with the numerically calculated values of Nuavg and favg is shown in
Figs. 22a–22b. The developed correlation shows a good match with the
Fig. 21d. Variation of favg/(C′5 × ReC′(P/e)C3′×exp(C′2(ln(P/e))2) × (e/w)C′7) with dif-
results which were estimated using CFD. The percentage error between
ferent e/D values.
the correlation and results is in the range of ± 12% and ± 15% for Nu
and f, respectively. Therefore, it is concluded that the correlation is
P e e reliable and acceptable for the considered range of parameters.
f = fxn ⎛Re, , , ⎞
⎝ e w D⎠ (18)
5. Conclusions
Figs. 20a–20d and Figs. 21a–21d shows the different regression
process which are used during development of statistical correlation for The flow characteristics and heat transfer augmentation is studied in
Nuavg and favg, respectively. The final correlation for Nuavg and favg has ribbed triangular duct using computational fluid dynamics (CFD). The
the form of (the detail of correlation development is shown in Appendix analysis is carried out for fixed range of Re which ranges from 4000 to
section); 18,000 to study the effect of e/D, P/e and e/w on Nu and f. Depending
on results following conclusions are drawn:
⎡ P −1.322 e 0.122 e 0.216
Nuavg = 0.365 × ⎢Re 0.83 ⎛ ⎞ ⎛ ⎞ ⎛ ⎞
⎣ ⎝e⎠ ⎝w⎠ ⎝D⎠ (a) The present numerical methodology and available experimental
2 correlation show good agreement with each other. Thus, present
P ⎤
× exp ⎧0.265 ⎛ln ⎛ ⎞ ⎞ ⎫ ⎥ methodology can be employed for roughened duct to predict fluid
⎨
⎩ ⎝ ⎝ e ⎠ ⎠ ⎬
⎭⎦ (19) characteristics and heat transfer.

1106
R. Kumar et al.
(b) The Nusselt number increased with decreases in relative roughness
pitch but increased with increases in relative roughness height and
rib aspect ratio under considered parameters.
(c) Under considered parameters, friction factor value is observed in-
creasing with increase in relative roughness height and rib aspect
ratio but it decreased with the increase in relative roughness pitch
value.
(d) The maximum improvement in Nusselt number is seen at Reynolds
number of 15,000 and with increase of Reynolds number from 4000
to 15,000 but further increase or decrease in Reynolds number re-
sults in decrease in Nusselt number enhancement under considered
value of roughness parameters.
(e) The thermohydraulic performance parameter is found maximum at
Reynolds number of 15,000. Further decrease or increase in
Reynolds number value results in decrease of thermohydraulic
performance parameter, in case of all studied roughness para-
meters.
(f) Under studied parameters, optimum thermohydraulic performance
parameter of the order of 1.89 is observed for relative roughness
pitch of 10, relative roughness height of 0.04 and rib aspect ratio of
4.0 at Reynolds number of 15,000.
Acknowledgement
This work is financially support by the grant received from
Department of Science and Technology, India.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in the
online version, at http://dx.doi.org/10.1016/j.solener.2017.08.071.
References
Ahn, S.W., Son, K.P., 2002. Heat transfer and pressure drop in the roughened equilateral
triangular duct. Int. Commun. Heat Mass Transf. 29 (4), 479–488.
ANSYS FLUENT 12.0, 2003–04. Documentation (Theory Guide), ANSYS Inc.
ASHRAE Standard 93, 2003. Method of Testing to Determine the Thermal Performance of
Solar Collectors. American Society of Heating, Refrigeration Air Conditioning
Engineers, Atlanta, GA 30329.
Balaras, C.A., 1990. A review of augmentation techniques for heat transfers surfaces in
single phase heat exchangers. Energy 15, 899–906.
Bergles, A.E., 1988. Some perspective on enhanced heat transfer, second-generation heat
transfer technology. ASME J. Heat Transf. 110, 1082–1096.
Bharadwaj, G., Kaushal, M., Goel, V., 2013. Heat transfer and friction characteristics of an
equilateral triangular solar air heater duct using inclined continuous ribs as rough-
ness element on the absorber plate. Int. J. Sustain. Energy 32 (6), 515–530.
Bharadwaj, G., Varun, Kumar, R., Sharma, A., 2017. Heat transfer augmentation and flow
characteristics in ribbed triangular duct solar air heater: an experimental analysis.
Int. J. Green Energy. http://dx.doi.org/10.1080/15435075.2017.1307751.
Cebeci, T., Bradshaw, P., 1988. Physical and Computational Aspects of Convective Heat
Transfer. Springer-Verlag, Germany.
Chaube, A., Sahoo, P.K., Solanki, S.C., 2006. Analysis of heat transfer augmentation and
flow characteristics due to rib roughness over absorber plate of a solar air heater.
Renew. Energy 31, 317–331.
Chegini, H., Chaturvedi, S.K., 1986. An experimental and analytical investigation of
friction factor for fully developed flow in internally finned triangular duct. ASME J.
Heat Transf. 108, 507–512.
Chennu, R., 2016. Steady state and transient analysis of compact plate-fin heat exchanger
fins for generation of design data using CFD. Int. J. Numer. Meth. Heat Fluid Flow 26
(2), 440–460.
Daschiel, G., Frohnapfel, B., Jovanovic, J., 2013. Numerical investigation of flow through
a triangular duct: the coexistence of laminar and turbulent flow. Int. J. Heat Fluid
Flow 41, 27–33.
Date, A.W., 2005. Introduction to Computational Fluid Dynamics. Cambridge University
Press.
Gao, X., Sunden, B., 2001. Heat transfer and pressure drop measurements in rib-rough-
ened rectangular ducts. Exp. Therm. Fluid Sci. 24, 25–34.
Iacovides, H., Raisee, M., 2001. Computation of flow and heat transfer in two-dimen-
sional rib-roughened passages, using low-Reynolds-number turbulence model. Int. J.
Numer. Meth. Heat Fluid Flow 11 (2), 138–155.
Incropera, F., Dewitt, P.D., Bergman, T.L., Lavine, A.S., Setharamu, K.N., Setharama, T.R.,
2013. Fundamentals of heat and mass transfe, fifth ed. John Wiley and Sons Inc.,
United States.
Jiang, Z., Xiao, Z., Shi, Y., Chen, S., 2016. Constrained large-eddy simulation of turbulent
flow and heat transfer in a stationary ribbed duct. Int. J. Numer. Meth. Heat Fluid
Flow 26, 1069–1091.
1107
Solar Energy 157 (2017) 1095–1107
Kang, H.J., Wong, T.T., Leung, C.W., 1998. Effects of surface roughness on forced con-
vective and friction in triangular ducts. Exp. Heat Transf. 11, 241–253.
Karabulut, H., Ipci, D., Cinar, C., 2016. Numerical solution of fully developed heat
transfer problem with constant wall temperature and application to isosceles trian-
gular and parabolic ducts. Appl. Therm. Eng. 102, 115–124.
Karmare, S.V., Tikekar, A.N., 2010. Analysis of fluid flow and heat transfer in a rib grit
roughened surface solar air heater using CFD. Sol. Energy 84, 409–417.
Kumar, R., Kumar, A., Varun, 2017a. Computational fluid dynamics based study for
analyzing heat transfer and friction factor in semi-circular rib roughened equilateral
triangular duct. Int. J. Numer. Meth. Heat Fluid Flow 27 (4). http://dx.doi.org/10.
1108/HFF-10-2015-0438.
Kumar, R., Varun, Kumar, A., 2016. Thermal and fluid dynamic characteristics of flow
through triangular cross-sectional duct: a review. Renew. Sustain. Energy Rev. 61,
123–140.
Kumar, R., Varun, Kumar, A., 2017b. A parametric study of the 2D model of solar air
heater with elliptical rib roughness using CFD. J. Mech. Sci. Technol. 31 (2),
959–964.
Leung, C.W., Chen, S., Wong, T.T., Probert, S.D., 2000. Forced convection and pressure
drop in a horizontal triangular sectional duct with V-grooved (i.e. orthogonal to the
mean flow) inner surfaces. Appl. Energy 66, 199–211.
Leung, C.W., Wong, T.T., Probert, S.D., 2001. Enhanced forced – convection from ribbed
or machine – roughened inner surfaces within triangular ducts. Appl. Energy 69,
87–99.
Liou, T.M., Chang, S.W., Hung, J.H., Chiou, S.F., 2007. High rotation number heat
transfer of a 458 rib-roughened rectangular duct with two channel orientations. Int.
J. Heat Mass Transf. 50, 4063–4078.
Luo, D.D., Leung, C.W., Chan, T.L., 2004. Forced convective and flow friction char-
acteristics of air-cooled horizontal equilateral triangular ducts with ribbed internal
surfaces. Int. J. Heat Mass Transf. 47, 5439–5450.
Ma, C., Chen, X., Wang, J., Zhang, S., Ji, S., 2015. An experimental investigation of heat
transfer characteristics for steam cooling and air cooling in a rectangular channel
roughened with parallel ribs. Exp. Therm. Fluid Sci. 64, 142–151.
Nascimento, I.P., Gracia, E.C., 2016. Heat transfer performance enhancement in compact
heat exchangers by using shallow square dimples in flat tubes. Appl. Therm. Eng. 96,
659–670.
Obot, N.T., Esen, E.B., Adu-Wusu, K., 1987. Pressure drop for rib – roughened scalene
triangular duct having two rounded corners. Int. Commun. Heat Mass Transf. 14,
11–20.
Pandey, N.K., Bajpai, V.K., Varun, 2016. Experimental investigation of heat transfer
augmentation using multiple arcs with gap on absorber plate of solar air heaters. Sol.
Energy 134, 314–326.
Patankar, S.V., 1980. Numerical Heat Transfer and Fluid Flow. Hemisphere,
Washington DC.
Prasad, B.N., Saini, J.S., 1991. Optimal thermohydraulic performance of artificially
roughened solar air heaters. Sol. Energy 47 (2), 91–96.
Raisee, M., Rokhzadi, A., 2009. Prediction of turbulent flow and heat transfer in channels
with array of detached and alternative attached-detached ribs. Int. J. Numer. Meth.
Heat Fluid Flow 19 (3/4), 266–284.
Ramsak, M., Skerget, L., 2008. Boundary element method for thermal flow using k-ε
turbulence models. Int. J. Numer. Meth. Heat Fluid Flow 18, 514–532.
Rokni, M., Gatski, T.B., 2001. Predicting turbulent convective heat transfer in fully de-
veloped duct flows. Int. J. Heat Fluid Flow 22 (4), 381–392.
Scholl, S., Verstraete, T., Duchaine, F., Gicquel, L., 2016. Conjugate heat transfer of a rib-
roughened internal turbine blade cooling channel using large eddy simulation. Int. J.
Heat Fluid Flow 61, 650–664.
Sethi, M., Varun, Thakur, N.S., 2012. Correlations for solar air heater duct with dimple
shape roughness elements on absorber plate. Sol. Energy 86, 2852–2861.
Singh, A.P., Varun, Siddhartha, 2014. Effect of artificial roughness on heat transfer and
friction characteristics having multiple arc shaped roughness elements on the ab-
sorber plate. Sol. Energy 105, 479–493.
Singh, S., Singh, B., Hans, V.S., Gill, R.S., 2015. CFD (computational fluid dynamics)
investigation on Nusselt number and friction factor of solar air heater duct roughened
with non-uniform cross-section transverse rib. Energy 84, 509–517.
Sparrow, E.M., Haji-Sheikh, A., 1965. Laminar heat transfer and pressure drop in isosceles
triangular, right triangular and circular sector ducts. ASME J. Heat Transf. 87 (3),
426–427.
Sunden, B., 2010. Simulation of compact heat exchanger performance. Int. J. Numer.
Meth. Heat Fluid Flow 20 (5), 551–569.
Tsai, W.B., Lin, W.W., Chieng, C.C., 2000. Computation of enhanced turbulent heat
transfer in a channel with periodic ribs. Int. J. Numer. Meth. Heat Fluid Flow 10 (1),
47–66.
Unalan, S., Akansu, S.O., Konca, A., 2007. Investigation of heat transfer and pressure drop
for various obstacles in a rectangular-sectioned 90° bend. Int. J. Numer. Meth. Heat
Fluid Flow 17 (5), 494–511.
Varun, Saini, R.P., Singal, S.K., 2007. A review on roughness geometry used in solar air
heaters. Sol. Energy 81, 1340–1350.
Versteeg, H.K., Malalasekera, W., 2007. An Introduction to Computational Fluid
Dynamics, second ed. Pearson Education Limited.
Webb, R.L., Eckert, E.G.R., 1972. Applications of roughness surface to heat exchanger
design. Int. J. Heat Mass Transf. 15 (9), 1647–1658.
Yadav, A.S., Thapak, M.A., 2014. Artificially roughened solar air heater: experimental
investigations. Renew. Sustain. Energy Rev. 36, 370–411.
Yadav, A.S., Bhagoria, J.L., 2014. A CFD based thermo-hydraulic performance analysis of
an artificially roughened solar air heater having equilateral triangular section rib
roughness on absorber plate. Int. J. Heat Mass Transf. 70, 1016–1039.
Yadav, S., Kaushal, M., Varun, Siddhartha, 2014. Exergetic performance evaluation of
solar air heater having arc shape oriented protrusions as roughness element. Sol.
Energy 105, 181–189.
Zhang, L., 2005. Turbulent three-dimensional air flow and heat transfer in a cross-cor-
rugated triangular duct. ASME J. Heat Transf. 127, 1151–1158.