Engineering Science and Technology, an International Journal 19 (2016) 904–910

Contents lists available at ScienceDirect

Engineering Science and Technology,

an International Journal Press: Karabuk University, Press Unit
ISSN (Printed) : 1302-0056
ISSN (Online) : 2215-0986
ISSN (E-Mail) : 1308-2043

j o u r n a l h o m e p a g e : h t t p : / / w w w. e l s e v i e r. c o m / l o c a t e / j e s t c h
H O S T E D BY

Available online at www.sciencedirect.com

ScienceDirect

Full Length Article

Numerical study on flow separation in 90° pipe bend under high

Reynolds number by k-ε modelling
Prasun Dutta *, Sumit Kumar Saha, Nityananda Nandi, Nairit Pal
Department of Aerospace Engineering and Applied Mechanics, Indian Institute of Engineering Science and Technology, Shibpur, Howrah 711103, India

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

Article history: The present paper makes an effort to find the flow separation characteristics under high Reynolds number
Received 7 October 2015 in pipe bends. Single phase turbulent flow through pipe bends is investigated using k-ε turbulence model.
Received in revised form After the validation of present model against existing experimental results, a detailed study has been
15 December 2015
performed to study the influence of Reynolds number on flow separation and reattachment. The sepa-
Accepted 15 December 2015
Available online 25 January 2016
ration region and the velocity field of the primary and the secondary flows in different sections have been
illustrated. Numerical results show that flow separation can be clearly visualized for bend with low cur-
vature ratio. Distributions of the velocity vector show the secondary motion clearly induced by the
Keywords:
90° pipe bend movement of fluid from inner to outer wall of the bend leading to flow separation. This paper provides
k-ε turbulence model numerical results to understand the flow characteristics of fluid flow in 90° bend pipe.
Turbulent flow © 2016, Karabuk University. Publishing services by Elsevier B.V.
Flow separation

1. Introduction testimony to this fact. A number of researchers have investigated

turbulent flows in pipe bends by means of theoretical, experimen-
Pipe bends are the most important part of any pipeline network tal and numerical methods [4–9]. To perform numerical simulation
system as these provide flexibility in routing. Investigations of the of fluid flow in curved pipes, on the other hand, the Navier–Stokes
flow through bends are of great significance in understanding and equation has to be expressed in curvilinear or body fitted coordi-
improving their performance and minimizing the losses. It is already nate system. A very useful database for direct numerical simulation
well known that the flow of incompressible viscous fluids through (DNS) and large eddy simulation (LES) on pipe bend is provided by
pipe bends is characterized by flow separation, secondary flow and two studies [10,11]. Recently, in the nuclear sector due to the fatigue
unsteadiness, which are dependent on Reynolds number as well as by the unsteady motion of the vortices, this has also attracted the
the radius of curvature of the bend. Whenever a fluid flows through interest of the researchers [2,12,13]. Hence, it is interesting to see
a bend, there is a radial pressure gradient developed by the cen- the flow separation and reattachment under high Reynolds number.
trifugal force acting on the fluid. Because of this, a double spiral flow Micro and nano size particle erosion in 90° pipe bend and over the
field and a pair of counter-rotating vortices can also be observed backward-facing steps were studied numerically [14–18]. Very re-
inside the bend i.e. because of the presence of pressure gradient, cently, studies on turbulent mixed convection heat transfer [19–25]
fluid at the centre of pipe moves towards the outer side and comes attracted the interest of investigators; many researchers used Lattice
back along the wall towards the inner side. Now if the bend cur- Boltzmann methods to solve natural convection heat transfer problem
vature ratio is very small (Rc/D ≤ 1.5), the adverse pressure gradient [26,27]. Magneto hydrodynamic flow (MHD) has attracted much in-
near the inner wall and immediately downstream of the bend may terest of researchers in recent years due to the effect of magnetic
lead to flow separation, giving rise to a large increase in pressure field on the boundary layer flow control [28–33]. Different exact and
losses [1–3]. However, the flow characteristics of incompressible flows approximate techniques have also been used to solve the different
in pipe bends are not fully clarified yet. Accurate estimation of mass problems in fluid mechanics [34–38]. In this paper, the flow sep-
flow rate and losses is critical for most incompressible flow systems. aration in the most common 90° pipe bend is studied by numerical
The applications of water-flows through pipe bends are found in methods based on computational fluid dynamics. The paper is struc-
many engineering applications. Some of the excellent reviews bear tured in the following fashion. Section 1 gives a brief idea on the
previous research works and motivation for the present work. Section
2 contains the necessary theoretical background. Problem defini-
tion with validation is provided in section 3. Section 4 contains the
* Corresponding author. Tel.: +91 33 2668–4561 to 63 (Extn.: 277).
E-mail address: pd.iiest@gmail.com (P. Dutta). study on various parameters affecting the flow pattern and is fol-
Peer review under responsibility of Karabuk University. lowed by summary bulletin of the study under section 5.

http://dx.doi.org/10.1016/j.jestch.2015.12.005
2215-0986/© 2016, Karabuk University. Publishing services by Elsevier B.V.
 P. Dutta et al. / Engineering Science and Technology, an International Journal 19 (2016) 904–910 905

2. Governing equations and numerical methodology (a)

Three dimensional Reynolds Averaged Navier–Stokes (RANS)
equations are solved using the segregated implicit solver. The right
choice of a turbulence model is critical when an industrial turbu-
lent flow problem is faced, especially when this problem involves
three dimensional flow phenomena, which needs an accurate mod-
elling. The second order scheme is used for the U-RANS equations
calculations, with a pressure velocity coupling achieved using SIMPLE
algorithm. The time step size used in the present study is 0.001 s
with 1000 time steps. The default under relaxation factors were used
to aid convergence for all models.
The governing equations for incompressible fluid flow with con-
stant properties are

∂u i
=0 (1)
∂x i

∂u i ∂u i 1 ∂p ∂ 2ui
+uj = fi − +υ (2)
∂t ∂x j ρ ∂x i ∂x j ∂x j

Equations (1) and (2) are conservations of mass and momen- (b)
tum, respectively; fi is a vector representing external forces, υ is the
kinematic viscosity.

2.1. Turbulence model

It is well known that turbulent flows are basically designated by

the fluctuations of the velocity fields. Different transported quan-
tities such as momentum, energy, etc. also fluctuate for this
fluctuation of velocity field and these fluctuations can be of very
high frequency and small scale; they are very difficult and
computationally crucial to analyse directly in industrial engineer-
ing calculations. The turbulence model needs to be selected based
on some considerations, e.g., the physics of the flow, the insight into Fig. 1. Schematic diagram of the bend geometry and present model with compu-
tational grid.
the capabilities and limitations of turbulence models, the attempt
for the specific problem by other researchers, the accuracy needed,
the available computational resources, and time.
set up 50D for all cases to save computational time. The fluid medium
The k-ε turbulence model is adopted for the present study as k-ε
was air having density (ρ) of 1.2647 kg/m3 and dynamic viscosity
turbulence model performs better for both single-phase and two-
(μ) of 1.983 × 10−5 kg/m-s for validation purpose and water having
phase flows in pipe bend [3,39–43]. In this model, the turbulence
density (ρ) of 990.2 kg/m3 and dynamic viscosity (μ) of 0.0006 kg/m-s
kinetic energy (k) and the turbulence dissipation rate (ε) are solved
for the present study with working temperature of 300 K in both
to determine the coefficient of turbulent viscosity (μt).
cases. Three dimensional structured mesh was used containing hexa-
Transport equation for k-epsilon
hedron elements, which was optimized via a grid-independence
∂ ( pk ) ∂ ( pkui ) ∂ ⎡ μt ∂k ⎤ study. The bend geometry and mesh are shown in Fig. 1a and b re-
+ = + 2μt E ij E ij − ρε (3)
∂t ∂x i ∂x j ⎢⎣ σ k ∂x j ⎥⎦ spectively. It is defined that the axial direction downstream the bend
is x-coordinate, the direction from inner core to outer core of the
bend is y-coordinate and the perpendicular direction to x and y is
∂ ( p ε ) ∂ ( p εui ) ∂ ⎡ μt ∂k ⎤ ε ε2
+ = ⎢ ⎥ + C 1ε 2μt E ij E ij − C 2ε ρ (4) z-coordinate.
∂t ∂x i ∂x j ⎣ σ ε ∂x j ⎦ k k
3.1. Validation
where ui represents velocity component in corresponding direc-
tion, E ij represents component of rate of deformation, and μt
At the very beginning of our study, our model and simulation
represents eddy viscosity.
setup are first validated against the existing experimental and nu-
Equations (3) and (4) also consist of some adjustable constants
merical data in References [3,7,45]. For that intension, same
[44]; these are as follows
geometrical configuration is adopted. In their experiment, the authors
C μ = 0.09 σ κ = 1.00 σ κ = 1.00 σ ε = 1.30 C 1ε = 1.44 C 2ε = 1.92 of the previously mentioned studies used a circular cross sec-
tioned 90° bend with a curvature ratio (R c /D) of 2 and the
measurements of velocities were performed at a Reynolds number
3. Problem definition of 6 × 104. For the validation of our present model, the simulation
is performed on a computational mesh containing total 2.85 million
The problem that is considered here is the fluid flow through hexahedron elements, which was optimized via a grid-independence
90° pipe bends having inner diameter of 0.01 m with curvature ratio study, see Fig. 1a. The value of non-dimensional distance from wall
(Rc/D) = 1 for different Reynolds numbers ranging from 1 × 105 to (Y+) is strictly controlled using standard wall treatment function
10 × 105. The inlet length of straight pipe in the calculations was (30 < Y+ < 90 for a near wall cell used for present study). The mean
906 P. Dutta et al. / Engineering Science and Technology, an International Journal 19 (2016) 904–910

(a) (b)

Fig. 2. Comparison of normalized axial velocity profile of present analysis with published experimental and numerical results.

axial velocity profile normalized with inlet velocity along symme- RP, MP represent positions of separation point, reattachment point,
try line at bend outlet (α = 90°) shows very good agreement with middle point between separation and reattachment points respec-
both experimental and numerical results [3,7,45], see Fig. 2b. In the tively. Velocity acceleration is depicted at the outer core of the bend
inner core region of the bend (negative X-axis in Fig.), some dis- as expected. It is found that with the increase of Reynolds number,
crepancies between experimental and numerical data are observed. the velocity profile at the inner core of the pipe bend tries to recover
This discrepancies may arise due to the adverse pressure gradient its fully developed shape by decelerating and accelerating effects
near the inner wall of the bend and this is also consistent with a at the outer and the inner parts of the pipe bend respectively. Hence,
recent study [3]. From the validation part, it has been seen that the it may be speculated for higher values of the Reynolds number that
model is in close approximation with the published results; hence, pipe curvature effects are reducing.
this procedure of mesh generation and simulation set up has been Fig. 4 shows the normalized flow velocity vector fields at the
used for further analysis. central plane and at the bend outlet (α = 90°) for Re = 1 × 105,
Rc/D = 1. The mean velocity is shifted upward due to the faster
4. Results and discussions moving fluid near the outer core as expressed in Reference [7]. Sec-
ondary flow pattern can clearly be observed in this section. At the
The main objective of the present study is to characterize the inner core of the bend, a region having very low velocity was found
effect of Reynolds number on flow separation of single phase tur- and the flow was highly unsteady and complex.
bulent flow in a 90° pipe bend through numerical simulation. The Fig. 5 shows the normalized velocity fields for two different Reyn-
results of the mean velocity profile, velocity fluctuation and sepa- olds numbers (Re = 1 × 105 & 10 × 105) at the bend outlet. The top
ration reattachment points for different Reynolds numbers along of the each figure is the outer core of the bend and the bottom is
the different positions of the bend in central symmetry plane are the inner core of the bend. The low velocity region near the inner
presented in this section. core of the bend was seen clearly. This region is concluded to be
Fig. 3 shows the mean velocity profile normalized with inlet ve- the separated region [46]. Because of the relation between the cen-
locity (Uin) for different Reynolds numbers (Re = 1 × 105 to 10 × 105). trifugal force with high velocity and pressure gradient on the flow,
The negative r values represent the inner core of the bend and SP, two identical counter rotating Dean vortices were also found for both
Reynolds numbers.
As mentioned above, a flow separation region was found at the
inner core of bend outlet; another study has been made to under-
stand the dependency of separation and reattachment points on
Reynolds number. Fig. 6 shows various starting points of flow sep-
aration from the bend inlet and corresponding reattachment points
from the bend outlet. It was found that the starting point of flow sep-
aration moves upstream in the bend as Reynolds number increases,
while the reattachment point moves downstream. This trend corre-
sponds to the dependency of Reynolds number on the total pressure
loss [45] and consistent with experimental results in References [2,13].
The rms velocity fluctuations at bend outlet in the x and y di-
rections are depicted in Fig. 7 for bend with Rc/D = 1 at different
Reynolds numbers. Non-dimensional fluctuating velocities Urms and
Vrms, normalized by the mean flow velocity in x and y directions re-
spectively. Negative values of r represent the inner core of the bend.
It is found that velocity fluctuation is maximum in both x and y di-
rections at the separation region; however, the intensity of
fluctuation is higher in the x direction. The Urms profiles do not show
much Reynolds number dependency, the peak position is almost
Fig. 3. Normalized velocity profiles at different positions in the bend for different Re. same for all measurements but the peak value is high for higher
 P. Dutta et al. / Engineering Science and Technology, an International Journal 19 (2016) 904–910 907

(a) (b)

Fig. 4. Normalized velocity vector field for Re = 1 × 105 and Rc/D = 1. (a) Symmetry plane (centre cross section). (b) Cross section at bend outlet (∝ = 90°).

(a) (b)

Fig. 5. Normalized velocity vector field. (a) Re = 1 × 105. (b) Re = 10 × 105.

(a) (b)

Fig. 6. Dependency of separation and reattachment points on Reynolds number. (a) Separation points. (b) Reattachment points.
908 P. Dutta et al. / Engineering Science and Technology, an International Journal 19 (2016) 904–910

(a) (b)

Fig. 7. rms profiles of velocity fluctuations at the bend outlet. (a) rms in x direction. (b) rms in y direction.

Reynolds number. For the Vrms profiles, a clear Reynolds number de- As mentioned earlier, a very complicated fluid structure is formed
pendency can be observed. The peak value becomes larger and at the downstream of bend outlet due to the flow fluctuation induced
shifted from the inner core towards the centre of the bend with in- by the movement separation region and the acceleration of flow ve-
creasing Reynolds number. locity. Fig. 8a–d shows the complex flow motions in the bend. Fig. 8a

(a) (b)

(c)

(d)

Fig. 8. Complex flow structure in the pipe bend. (a) Cross sectional planes for flow visualization. (b) Flow at symmetry plane. (c) Flow in separated region. (d) Circumfer-
ential flow from outer core to inner core.
 P. Dutta et al. / Engineering Science and Technology, an International Journal 19 (2016) 904–910 909

describes the three cross sectional planes for the flow visualiza- [8] A. Kalpakli, R. Örlü, Turbulent pipe flow downstream a 90 pipe bend with and
without superimposed swirl, Int. J. Heat Fluid Flow 41 (2013) 103–111.
tions. Three types of flow motions are observed. Takamura et al. [2]
[9] P. Dutta, N. Nandi, Study on pressure drop characteristics of single phase
studied these three motions by statistical analysis (frequency anal- turbulent flow in pipe bend for high Reynolds number, ARPN J. Eng. Appl. Sci.
ysis) in terms of Strouhal number (St) and explained as vortices 10 (5) (2015) 2221–2226.
motion with St ≈ 1.0 (Fig. 8b), separated vortices shedding with [10] T.J. Hüttl, R. Friedrich, Direct numerical simulation of turbulent flows in curved
and helically coiled pipes, Comput. Fluids 30 (5) (2001) 591–605.
St ≈ 0.5 (Fig. 8c) and circumferential motion with St ≈ 0.5 (Fig. 8c). [11] A. Noorani, G.K. El Khoury, P. Schlatter, Evolution of turbulence characteristics
from straight to curved pipes, Int. J. Heat Fluid Flow 41 (2013) 16–26.
[12] J. Sakakibara, N. Machida, Measurement of turbulent flow upstream and
5. Conclusions downstream of a circular pipe bend, Phys. Fluids 24 (4) (2012) 041702.
[13] A. Ono, N. Kimura, H. Kamide, A. Tobita, Influence of elbow curvature on flow
structure at elbow outlet under high Reynolds number condition, Nucl. Eng.
Turbulent flow of single phase incompressible fluid through 90°
Des. 241 (11) (2011) 4409–4419.
pipe bend has been simulated numerically using k-ε turbulent model [14] M.R. Safaei, O. Mahian, F. Garoosi, K. Hooman, A. Karimipour, S.N. Kazi, et al.,
in the present study. The validation of 3D models used for the present Investigation of micro-and nanosized particle erosion in a 90° pipe bend using
study with experiments reported in Reference [7] and numerical a two-phase discrete phase model, ScientificWorldJournal 2014 (2014) Article
ID 740578.
works in References [3,45] indicates a good agreement. The fol- [15] S. Shamshirband, A. Malvandi, A. Karimipour, M. Goodarzi, M. Afrand, D.
lowing conclusions can be made from the present study: Petković, et al., Performance investigation of micro-and nano-sized particle
erosion in a 90 elbow using an ANFIS model, Powder Technol. 284 (2015)
336–343.
• It is found that with the increase of Reynolds number, the ve- [16] M.R. Safaie, B. Rahmanian, M. Goodarzi, Investigation of the coal diameter effect
locity profile at the inner core of the pipe bend tries to recover on pulverized coal combustion for pollutant reduction, J. Math. Comput. Sci.
its fully developed shape by decelerating and accelerating effects 12 (2014) 143–151.
[17] H. Togun, M.R. Safaei, R. Sadri, S.N. Kazi, A. Badarudin, K. Hooman, et al.,
at the outer and the inner parts of the pipe bend respectively. Numerical simulation of laminar to turbulent nanofluid flow and heat transfer
Hence, it may be speculated for higher values of the Reynolds over a backward-facing step, Appl. Math. Comput. 239 (2014) 153–170.
number that pipe curvature effects are reducing. The separa- [18] H. Togun, G. Ahmadi, T. Abdulrazzaq, A.J. Shkarah, S.N. Kazi, A. Badarudin, et al.,
Thermal performance of nanofluid in ducts with double forward-facing steps,
tion region for all bends was generated and a clear Reynolds
J. Taiwan Inst. Chem. Eng. 47 (2015) 28–42.
number dependency on flow separation was found. As Reyn- [19] R. Davarnejad, M. Jamshidzadeh, CFD modeling of heat transfer performance
olds number increased, the separation point moved towards the of MgO-water nanofluid under turbulent flow, Eng. Sci. Technol. Int. J. 18 (4)
(2015) 536–542.
upstream, while the reattachment point moved towards down-
[20] H. Yarmand, S. Gharehkhani, S.N. Kazi, E. Sadeghinezhad, M.R. Safaei, Numerical
stream of the bend. A low velocity region was found near the investigation of heat transfer enhancement in a rectangular heated pipe for
inner core of the bend leading to the flow separation. Two counter turbulent nanofluid, ScientificWorldJournal 2014 (2014) 369593.
rotating Dean vortices were formed at the bend outlet. [21] M. Safaei, M. Goodarzi, M. Mohammadi, Numerical modeling of turbulence
mixed convection heat transfer in air filled enclosures by finite volume method,
• The rms of the velocity fluctuation in x direction of flow has no Int. J. Multiphys. 5 (4) (2011) 307–324.
significant Reynolds number dependence, the peak value and the [22] M.R. Safaei, H. Togun, K. Vafai, S.N. Kazi, A. Badarudin, Investigation of heat
position of the peak value are nearly the same for different Reyn- transfer enhancement in a forward-facing contracting channel using FMWCNT
nanofluids, Numer. Heat Transfer Part A Appl. 66 (12) (2014) 1321–1340.
olds numbers; however, the peak value was observed for the [23] S. Eiamsa-ard, K. Kiatkittipong, W. Jedsadaratanachai, Heat transfer
highest Reynolds number. For the rms profile in y direction, a enhancement of TiO2/water nanofluid in a heat exchanger tube equipped
little Reynolds number dependence can be observed. With in- with overlapped dual twisted-tapes, Eng. Sci. Technol. Int. J. 18 (3) (2015)
336–350.
creasing Reynolds number, the peak value becomes larger and [24] A. Karimipour, M. Afrand, M. Akbari, M.R. Safaei, Simulation of fluid flow
situated further from the wall. Overall velocity fluctuation became and heat transfer in the inclined enclosure, Int. J. Mech. Aerosp. Eng. 6 (2012)
larger in the separation region near the inner core of the bend; 86–91.
[25] M. Goodarzi, M.R. Safaei, H.F. Oztop, A. Karimipour, E. Sadeghinezhad, M. Dahari,
however, the intensity of fluctuation in x direction is lower than
et al., Numerical study of entropy generation due to coupled laminar and
that of y direction. The flow became very complex, unsteady and turbulent mixed convection and thermal radiation in an enclosure filled with
coherent at the downstream of the bend due to the flow sepa- a semitransparent medium, ScientificWorldJournal 2014 (2014) Article ID
761745.
ration showing three flow motions.
[26] B. Mliki, M.A. Abbassi, K. Guedri, A. Omri, Lattice Boltzmann simulation of
natural convection in an L-shaped enclosure in the presence of nanofluid, Eng.
The above conclusions refer especially for present study range. Sci. Technol. Int. J. 18 (3) (2015) 503–511.
To provide a correlation between the separation region, curvature [27] M. Goodarzi, M.R. Safaei, A. Karimipour, K. Hooman, M. Dahari, S.N. Kazi, et al.,
Comparison of the finite volume and Lattice Boltzmann methods for solving
ratio and the Reynolds number, additional studies is required with natural convection heat transfer problems inside cavities and enclosures,
different Reynolds numbers and curvature ratio and will remain for Abstract and Applied Analysis 2014 (2014) Hindawi Publishing Corporation.
future work. http://dx.doi.org/10.1155/2014/762184. <http://www.hindawi.com/journals/
aaa/2014/762184/abs/>.
[28] S. Mukhopadhyay, I.C. Mandal, Magnetohydrodynamic (MHD) mixed convection
References slip flow and heat transfer over a vertical porous plate, Eng. Sci. Technol. Int.
J. 18 (1) (2015) 98–105.
[29] S. Das, R.N. Jana, O.D. Makinde, Mixed convective magnetohydrodynamic flow
[1] L.H. Hellström, A. Sinha, A.J. Smits, Visualizing the very-large-scale motions in in a vertical channel filled with nanofluids, Eng. Sci. Technol. Int. J. 18 (2) (2015)
turbulent pipe flow, Phys. Fluids 23 (1) (2011) 011703. 244–255.
[2] H. Takamura, S. Ebara, H. Hashizume, K. Aizawa, H. Yamano, Flow visualization [30] A. Mahdy, S.E. Ahmed, Thermosolutal Marangoni boundary layer
and frequency characteristics of velocity fluctuations of complex turbulent flow magnetohydrodynamic flow with the Soret and Dufour effects past a vertical
in a short elbow piping under high Reynolds number condition, J. Fluids Eng. flat plate, Eng. Sci. Technol. Int. J. 18 (1) (2015) 24–31.
134 (10) (2012) 101201. [31] K. Javaherdeh, M.M. Nejad, M. Moslemi, Natural convection heat and mass
[3] J. Kim, M. Yadav, S. Kim, Characteristics of secondary flow induced by 90-degree transfer in MHD fluid flow past a moving vertical plate with variable surface
elbow in turbulent pipe flow, Eng. Appl. Comput. Fluid Mech. 8 (2) (2014) temperature and concentration in a porous medium, Eng. Sci. Technol. Int. J.
229–239. 18 (3) (2015) 423–431.
[4] P.L. Spedding, E. Benard, G.M. McNally, Fluid flow through 90 degree bends, [32] A. Khalid, I. Khan, A. Khan, S. Shafie, Unsteady MHD free convection flow of
Dev. Chem. Eng. Miner. Process. 12 (1–2) (2004) 107–128. Casson fluid past over an oscillating vertical plate embedded in a porous
[5] P. Naphon, S. Wongwises, A review of flow and heat transfer characteristics in medium, Eng. Sci. Technol. Int. J. 18 (3) (2015) 309–317.
curved tubes, Renew. Sustain. Energy Rev. 10 (5) (2006) 463–490. [33] M. Hameed, A.A. Khan, R. Ellahi, M. Raza, Study of magnetic and heat transfer
[6] N. Crawford, S. Spence, A. Simpson, G. Cunningham, A numerical investigation on the peristaltic transport of a fractional second grade fluid in a vertical tube,
of the flow structures and losses for turbulent flow in 90 elbow Eng. Sci. Technol. Int. J. 18 (3) (2015) 496–502.
bends. Proc. Inst. Mech. Eng, Part E: J. Process Mech. Eng. 223 (1) (2009) 27–44. [34] I. Ebtehaj, H. Bonakdari, A.H. Zaji, H. Azimi, F. Khoshbin, GMDH-type neural
[7] K. Sudo, M. Sumida, H. Hibara, Experimental investigation on turbulent flow network approach for modeling the discharge coefficient of rectangular
in a circular-sectioned 90-degree bend, Exp. Fluids 25 (1) (1998) 42–49. sharp-crested side weirs, Eng. Sci. Technol. Int. J. 18 (4) (2015) 746–757.
910 P. Dutta et al. / Engineering Science and Technology, an International Journal 19 (2016) 904–910

[35] F. Selimefendigil, H.F. Öztop, Mixed convection of ferrofluids in a lid driven cavity [41] P. Dutta, N. Nandi, Effect of Reynolds number and curvature ratio on single phase
with two rotating cylinders, Eng. Sci. Technol. Int. J. 18 (3) (2015) 439–451. turbulent flow in pipe bends, Mech. Mech. Eng. 19 (1) (2015) 5–16.
[36] P. Valinataj-Bahnemiri, A. Ramiar, S.A. Manavi, A. Mozaffari, Heat transfer [42] M. Goodarzi, M.R. Safaei, K. Vafai, G. Ahmadi, M. Dahari, S.N. Kazi, et al.,
optimization of two phase modeling of nanofluid in a sinusoidal wavy channel Investigation of nanofluid mixed convection in a shallow cavity using a
using Artificial Bee Colony technique, Eng. Sci. Technol. Int. J. 18 (4) (2015) two-phase mixture model, Int. J. Therm. Sci. 75 (2014) 204–220.
727–737. [43] M.R. Safaei, H.R. Goshayeshi, B.S. Razavi, M. Goodarzi, Numerical investigation
[37] M. Azimi, A. Mozaffari, Heat transfer analysis of unsteady graphene oxide of laminar and turbulent mixed convection in a shallow water-filled enclosure
nanofluid flow using a fuzzy identifier evolved by genetically encoded mutable by various turbulence methods, Sci. Res. Essays 6 (22) (2011) 4826–4838.
smart bee algorithm, Eng. Sci. Technol. Int. J. 18 (1) (2015) 106–123. [44] J. Tu, G.H. Yeoh, C. Liu, Computational Fluid Dynamics: A Practical Approach,
[38] M. Shahrbanozadeh, G.A. Barani, S. Shojaee, Simulation of flow through dam Butterworth-Heinemann, Oxford, UK, 2012.
foundation by isogeometric method, Eng. Sci. Technol. Int. J. 18 (2) (2015) [45] M.A. Tanaka, H. Ohshima, H. Monji, Numerical investigation of flow structure
185–193. in pipe elbow with large eddy simulation approach, in: ASME 2009 Pressure
[39] G.F. Homicz, Computational Fluid Dynamic Simulations of Pipe Elbow Flow, Vessels and Piping Conference, American Society of Mechanical Engineers, 2009,
Department of Energy, USA, 2004. pp. 449–458.
[40] H. Rahimzadeh, R. Maghsoodi, H. Sarkardeh, S. Tavakkol, Simulating flow over [46] T. Shiraishi, H. Watakabe, H. Sago, H. Yamano, Pressure fluctuation characteristics
circular spillways by using different turbulence models, Eng. Appl. Comput. Fluid of the short-radius elbow pipe for FBR in the postcritical Reynolds regime,
Mech. 6 (1) (2012) 100–109. J. Fluid Sci. Technol. 4 (2) (2009) 430–441.