Professional Documents
Culture Documents
Energy: Zied Driss, Olfa Mlayeh, Dorra Driss, Makram Maaloul, Mohamed Salah Abid
Energy: Zied Driss, Olfa Mlayeh, Dorra Driss, Makram Maaloul, Mohamed Salah Abid
Energy
journal homepage: www.elsevier.com/locate/energy
a r t i c l e i n f o a b s t r a c t
Article history: In this paper, numerical simulation and experimental validation were carried out to study the turbulent
Received 26 September 2013 flow around a small incurved Savonius wind rotor. The software “SolidWorks Flow Simulation” has been
Received in revised form used to present the local characteristics in different transverse and longitudinal planes. The numerical
29 June 2014
model considered is based on the resolution of the NaviereStokes equations in conjunction with the
Accepted 6 July 2014
standard k-ε turbulence model. These equations were solved by a finite volume discretization method.
Available online 6 August 2014
Experimental results are conducted on an open wind tunnel equipped by a small incurved Savonius wind
rotor to validate the numerical method. Use of this knowledge will assist the design of packaged in-
Keywords:
Incurved Savonius rotor
stallations of incurved Savonius wind rotor.
Wind tunnel © 2014 Elsevier Ltd. All rights reserved.
Turbulent flow
Aerodynamic structure
CFD
http://dx.doi.org/10.1016/j.energy.2014.07.016
0360-5442/© 2014 Elsevier Ltd. All rights reserved.
Z. Driss et al. / Energy 74 (2014) 506e517 507
behind the turbine to provide additional rotation of the down- their research was to gain an insight into the complex flow field
stream turbine. The other is obtained from the periodic coupling developed around a Savonius wind rotor and to evaluate its per-
of local flow between the two turbines, which is associated with formance. They validated the model by comparing it with data
vortex shedding and cyclic pressure fluctuations. Akwa et al. [12] obtained at EWT (Environmental Wind Tunnel) laboratory. Irabu
discussed the influence of the buckets overlap ratio of a Savo- and Roy [15] improved and adjusted the output power of Savonius
nius wind rotor on the averaged moment and power coefficients, rotor under various wind power and suggests the method of
over complete cycles of operation. The influence of the buckets prevention the rotor from strong wind disaster. In this study, as
overlap ratio on the moment and power coefficients is checked by the appropriate device to achieve the purpose of it, a guide-box
changing the geometry of the rotor. The values of the moment and tunnel was employed. The guide-box tunnel is like a rectangular
power coefficients obtained as a function of tip speed ratio and the box as wind passage in which a test rotor is included. At first, the
buckets overlap of the rotor indicates that the maximum device experiment was conducted to find the adequate configuration
performance occurs for buckets overlap ratios with values close to which would provide the best relative performance. Mohamed
0.15. Kamoji et al. [13] compared the helical Savonius rotor with et al. [16] employed an optimization process in order to increase
the conventional Savonius rotor. The results indicate that the he- the tangential force induced by a monoplane Wells turbine using
lical Savonius rotors have positive coefficient of static torque. symmetric airfoil blades. The automatic optimization procedure is
Helical rotor without shaft at an overlap ratio of 0.0 and an aspect carried out by coupling an in-house optimization library (OPAL
ratio of 0.88 is found to have almost the same coefficient of power (OPtimization ALgorithms)) with an industrial CFD (Computa-
when compared with the conventional Savonius rotor. Indeed, tional Fluid Dynamics) code (ANSYS-Fluent). This multi-objective
conventional Savonius rotors have high coefficient of static torque optimization relying on Evolutionary Algorithms takes into ac-
at certain rotor angles and a negative coefficient of static torque count both tangential force coefficient and turbine efficiency.
from 135 to 165 and from 315 to 345 in one cycle of 360 . Detailed comparisons are presented between the optimal design
D'Alessandro et al. [14] developed a mathematical model of the and the classical Wells turbine using symmetric airfoils, demon-
interaction between the flow field and the rotor blades. The aim of strating the superiority of the proposed solution. Castelli et al. [17]
508 Z. Driss et al. / Energy 74 (2014) 506e517
presented a CFD model for the evaluation of energy performance types have been investigated numerically using CFD (Computa-
and aerodynamic forces acting on a straight-bladed vertical-axis tional Fluid Dynamics) technique and experimentally. A consid-
Darrieus wind turbine. The basic principles which are currently erable improvement of the H-rotor Darrieus turbine self-starting
applied to BE-M theory for rotor performance prediction are capability can be obtained by these techniques. Davila-Vilchis and
transferred to the CFD code, allowing the correlation between flow Mishra [21] evaluated the performance of a horizontal hydroki-
geometric characteristics (such as blade angles of attack) and netic energy system with variable-pitch blades using an axial-flux
dynamic quantities (such as rotor torque and blade tangential and generator. Particularly, very simple sheet blades have been used to
normal forces). The model is proposed as a powerful design and keep system cost down. The evaluation is based on maximum
optimization tool for the development of new rotor architectures power extraction and energy conversion efficiency normalized by
for which test data is not available. Chen et al. [18] developed a system cost through a simpler electro-mechanical design for the
novel vertical-axis water turbine for hydropower harness inside hydrokinetic system. Experimental results have demonstrated that
water pipelines when extra water head can be consumed. The the proposed prototype possesses higher efficiency with reduced
small hydro power generated can be used for power supply to the energy losses and manufacturing costs. It represents a cost-
data collection systems in underground and congested locations or competitive alternative energy for power supply for civilian ap-
to maintenance work at remote locations. The device was devel- plications in remote areas or an option for expeditionary appli-
oped through CFD (Computational Fluid Dynamic) simulation and cations. Mohamed [22] studied numerically and aerodynamically
lab tests. The results indicate that the simulation outcomes can Darrieus rotor to obtain the generated noise from blades. Time-
offer a good guidance for the rotor design even though the dif- accurate solutions can be obtained from URANS (unsteady Rey-
ference between the simulation and experimental results is fairly nolds averaged NaviereStokes) equations. Blade shape, tip speed
large. The results also show that the rotor with a hollow structure ratio and solidity effects have been studied in this work. The re-
combined with an eye-shaped slanted block in a pipeline could sults indicated that the higher solidity and higher tip speed ratio
generate the maximum power as compared with other types of rotors are more noisy than the normal turbines. Akwa et al. [23]
vertical-axis rotors. Mohamed and Shaaban [19] proposed to presented a review on the performance of Savonius wind tur-
improve the aerodynamic efficiency of Wells turbines by opti- bines. Particularly, numerous adaptations for this device were
mizing the blade pitch angle. Two different airfoil geometries are proposed. The variety of possible configurations of the rotor is
numerically investigated. Optimization results show that self pitch another advantage in using such machine. Each different
control of Wells turbines can substantially improve the turbine arrangement of Savonius rotor affects its performance. Savonius
efficiency while slightly delaying the turbine starting point. The rotor performance is affected by operational conditions, geometric
optimum blade pitch angle depends on airfoil geometry and tur- and air flow parameters. The range of reported values for
bine solidity. Mohamed [20] introduced the main problem of the maximum averaged power coefficient includes values around
self-stating capability of Darrieus turbine and investigates some 0.05e0.30 for most settings. Performance gains of up to 50% for tip
techniques to improve this drawback. The effect of the turbine speed ratio of maximum averaged power coefficient are also re-
solidity and the usage of hybrid system between drag and lift ported with the use of stators.
Z. Driss et al. / Energy 74 (2014) 506e517 509
2. Experimental device
of low-velocity air and reduces it into a small volume of high- the control volume [28e31]. The original fluid-flow simulation
velocity air without creating turbulence. The diffuser slows the tool developed exclusively for SolidWorks users, “Solidworks Flow
speed of airflow in the wind tunnel. The drive section provides the simulation” gives insight into designs related to fluid flow, heat
force that causes the air to move through the wind tunnel. In an transfer, and forces on immersed or surrounding components. In
open tunnel, the air flows in one end of the tunnel and out the this paper, the software “Solidworks Flow simulation” has been
other. The overall design creates high-speed, low-turbulence used to study the turbulent flow around a small incurved Savonius
airflow through the test vein [24e27]. Fig. 2 illustrates the wind rotor.
geometrical parameters of the incurved Savonius wind rotor. This
rotor is constituted by two half incurved buckets characterized by 3.1. Mathematical formulation
the height H ¼ 300 mm, the diameter c ¼ 2q ¼ 100 mm and the
bucket arc angle j ¼ 120 . These buckets are collected on a com- This paragraph deals with the mathematical equations which
mon axis, with a shaft diameter equal to ds ¼ 10 mm, and they are are based on this modeling. The mathematical formulation based
fixed within screws to make an angle equal to 180 . Therefore, the on the equations binding the various flow parameters. The vast
position between the two buckets is defined by the lengths majority of fluid flows encountered in aerodynamics are turbulent.
designed by s ¼ 9 mm and p ¼ 15 mm. The equations that govern the movement of fluids were written
for the first time by Claude Navier in 1823. They are often called
3. Numerical model the NaviereStokes equations. The equations governing the flow of
air with the necessary simplifications are obtained from the con-
Like other CFD (computational fluid dynamic) codes, “Solid- tinuity equation, the equation of momentum, the transport
works Flow simulation” combines a high level of functionality and equation of turbulent kinetic energy k and the transport equation
accuracy with ease-of-use. This code is based on solving of dissipation rate of turbulent kinetic energy ε. Several methods
NaviereStokes equations with a finite volume discretization are available for including turbulence in the NaviereStokes
method. The technique consists in dividing the computational equations. Most of these involve a process of time-averaging the
domain into elementary volumes around each node in the grid; it conservation equations. When turbulence is included, the trans-
ensures continuity of flow between nodes. The spatial discretiza- ported quantity is assumed to be the sum of an equilibrium and a
tion is obtained by following a procedure for tetrahedral inter- fluctuating component. The only term that remains positive defi-
polation scheme. As for the temporal discretization, the implicit nite is one containing the product of two fluctuating terms
formulation is adopted. The transport equation is integrated over [28e31].
Z. Driss et al. / Energy 74 (2014) 506e517 511
means that at this opening the fluid exits the model to an area of
static atmospheric pressure. The inlet velocity value (V ¼ 3 m s1)
corresponds to the air velocity measured in the settling chamber
within a hot wire anemometry AM-4204 model. In these condi-
tions, the frequency of the speed variator is equal to f ¼ 50 Hz.
Without wind rotor, a uniform distribution of the velocity fields
(V ¼ 12.7 m s1) has been observed in the test section [26]. How-
ever, this distribution is affected by the wind rotor and the uniform
distribution has been observed only on the test section inlet.
4. Meshing
k2
mt ¼ rCm (6)
ε
Initial since it is the mesh the calculation starts from and it could be time and the velocity value measured in the test section for the
further refined during the calculation if the solution-adaptive treated cases. According to these results, it's clear that the velocity
meshing is enabled. The initial mesh is constructed from the value obtained for the fifth case is the closest to the experimentally
basic mesh by refining the basic mesh cells in accordance with the measured value for the point referred by x ¼ 0 mm, y ¼ 100 mm
specified mesh settings. The Basic mesh is formed by dividing the and z ¼ 150 mm. Also, it has been noted that the resolution time
computational domain into slices by parallel planes which are increases with the decrease of the size of mesh cells.
orthogonal to the global coordinate system's axes. The basic mesh Fig. 7 shows different profiles of the average velocity for the
in many respects governs the generated computational mesh. The different cells size. It presents the superposition of the numerical
proper basic mesh is necessary for the most optimal mesh. So, we results gathered from the CFD code and the experimental results
can control the basic mesh in several ways by changing the number taken by the anemometer. The considered direction is situated in
of the basic mesh cells along the x, y and z axes, by shifting or the downstream of the rotor in the intersection of planes defined by
inserting basic mesh planes and stretching or contracting the basic y ¼ 100 mm and z ¼ 150 mm.
mesh cells locally by changing the relative distance between the The different velocity profiles seem to have the same appear-
basic mesh planes. ance but the velocity values depends on the cell size. Indeed the
greater the cell size gets the more the gap between numerical and
4.2. Meshing optimization and experimental validation experimental results is large. The best result regarding precision
and time is found to be a cell of 5 mm size. This choice leads to a
This application has been focused on the mesh resolution's in- better result with regards to the precision and the resolution time.
fluence on flow simulation results. In fact, the size of the mesh has
been changed then the obtained results have been compared to the 5. Numerical results
experimental velocity values collected from the test section of the
wind tunnel. In particular, six meshes have been studied. The first As presented in Fig. 8, two longitudinal planes, defined by
case to be treated corresponds to a cell of 40 cm (Fig. 4a). The second x ¼ 0 mm and y ¼ 0 mm, are considered to visualize the velocity
case corresponds to a cell of 10 cm (Fig. 4b). The third case corre- field, the total pressure, the dynamic pressure, the turbulent kinetic
sponds to a cell of 5 cm (Fig. 4c). The fourth case corresponds to a cell energy, the turbulent dissipation rate, the turbulent viscosity and
of 2 cm (Fig. 4d). The fifth case corresponds to a cell of 0.5 cm the vorticity.
(Fig. 4e). The latter one corresponds to a cell of 0.1 cm (Fig. 4f). In all
these cases, the number of cells is respectively equal to 1348, 3358, 5.1. Velocity field
5635, 10310, 37075 and 193701; which corresponds to a coarse mesh
in the first case and a refined mesh in the sixth case. Fig. 9 presents the distribution of the velocity field on the lon-
Figs. 5 and 6 show the meshing on two transverse planes gitudinal planes defined by x ¼ 0 mm and y ¼ 0 mm. According to
defined by x ¼ 0 mm and y ¼ 0 mm. Table 1 presents the resolution these results, it has been noted that the velocity is weak in the inlet
Fig. 12. Distribution of the turbulent kinetic energy in the longitudinal planes.
of the collector. It is indeed governed by the boundary condition y ¼ 0 mm, show a symmetric distribution. Also, a deceleration of
value of the inlet velocity which is equal to V ¼ 3 m s1. In this the velocity field around the shaft and in the concave surface of the
region, the velocity field is found to be uniform and increases buckets has been observed. Another zone is located in the top and
progressively downstream of the collector. At the test vein, an the left wall of the test vein which corresponds to zone around the
important increase of the velocity value has been noted due to the advancing bucket. This explains the acceleration of the velocity in
reduction of the tunnel section that causes the throttling of the the buckets exterior surface.
flow. While the upstream of the rotor is characterized by the high
velocity, a brutal drop is located in the concave surface of the two 5.2. Total pressure
buckets. Downstream of the rotor, the velocity value keeps
increasing till the out of the test section. Then, a sharp decrease has Fig. 10 presents the distribution of the total pressure on the
been noted through the diffuser where the minimum velocity longitudinal planes defined by x ¼ 0 mm and y ¼ 0 mm. While
values are recorded in the lateral walls of the diffuser. The examining these results, it can easily be noted that the total pres-
maximum velocity values are located in the convex surface of the sure is on its maximum in the intake and is globally uniform in the
two buckets according to the distribution shown on the y ¼ 0 mm collector and the upstream of the rotor in the test vein. A brutal
plane. The velocity fields on the longitudinal plane, defined by drop of the total pressure has been noted in the concave surface of
Fig. 13. Distribution of the dissipation rate of the turbulent kinetic energy in the longitudinal planes.
Z. Driss et al. / Energy 74 (2014) 506e517 515
the rotor, downstream of the advancing bucket and around the axe 5.4. Turbulent kinetic energy
of the rotor. The distribution of the total pressure in the longitu-
dinal planes defined by x ¼ 0 mm and y ¼ 0 mm shows that the Fig. 12 presents the distribution of the turbulent kinetic energy
minimum pressure values are located in the downstream of the on the longitudinal planes defined by x ¼ 0 mm and y ¼ 0 mm.
concave surface of the returning bucket, the convex surface of the From these results, it is clear that the turbulent kinetic energy is
advancing bucket and the extremities of the rotor axe. Downstream found to be very weak in the wind tunnel except in the
of the rotor, the total pressure starts to increase gradually in the area surrounding the rotor. The distribution of the turbulent kinetic
way out of the test vein and keeps increasing through the diffuser. energy shows the increase of the energy in the interior zone of the
The distribution in the y ¼ 0 mm shows a wake formed in the rotor. The maximum value of the turbulent kinetic energy is
downstream of the rotor. The transverse planes show the formation recorded along the endplates especially from the interior surface as
of the minimum pressure value zone around the rotor axe, in the shown in the distribution in the longitudinal plane defined
concave surface of the buckets and in the cotes of the test section. by x ¼ 0 mm. The distribution in the second longitudinal
Around the rotor, the total pressure is found to be relatively high plane defined by y ¼ 0 mm shows another wake characterized
and is increasing in the way out of the test vein. by the maximum value in the convex surface of the advancing
bucket.
5.3. Dynamic pressure
Fig. 11 presents the distribution of the dynamic pressure on the 5.5. Dissipation rate of the turbulent kinetic energy
longitudinal planes defined by x ¼ 0 mm and y ¼ 0 mm. The dy-
namic pressure is found to be weak in the collector inlet and in- Fig. 13 presents the distribution of the dissipation rate of the
creases gradually through the collector as long as the tunnel section turbulent kinetic energy on the longitudinal planes defined by
gets smaller. When it gets to the test section, the dynamic pressure x ¼ 0 mm and y ¼ 0 mm. From these results a first view shows that
keeps increasing in the upstream of the rotor and around it. It the dissipation rate of the turbulent kinetic energy is rather weak in
reaches its maximum in the convex surface of both the advancing the whole wind tunnel. It shows an important increase around the
and returning buckets. A zone of minimum dynamic pressure is rotor's buckets and axe. The wake defined by the maximum dissi-
recorded in the concave surface of the rotor buckets and around the pation rate of the turbulent kinetic energy is found in the leading
shaft. Downstream of the rotor, the dynamic pressure remains surface of the rotor and around the rotor's axe while it decreases in
relatively low in the test section and through the diffuser. The the middle of the bucket. The distribution of the turbulent dissi-
distribution of the dynamic pressure in the transverse planes pation rate in the transverse planes shows that a brutal increase of
shows a minimum zone in the concave surface of the buckets and the dissipation rate is more important in the superior and inferior
around the rotor axe. parts of the buckets.
[18] Chen J, Yang HX, Liu CP, Lau CH, Lo M. A novel vertical axis water turbine for [26] Driss Z, Abid MS. Numerical and experimental study of an open circuit tunnel:
power generation from water pipelines. Energy 2013;54:184e93. aerodynamic characteristics. Sci Acad Trans Renew Energy Syst Eng Technol
[19] Mohamed MH, Shaaban S. Optimization of blade pitch angle of an axial tur- 2012;2:116e23.
bine used for wave energy conversion. Energy 2013;56:229e39. [27] Driss Z, Damak A, Karray S, Abid MS. Experimental study of the internal re-
[20] Mohamed MH. Impacts of solidity and hybrid system in small wind turbines covery effect on the performance of a Savonius wind rotor. Res Rev J Eng
performance. Energy 2013;57:495e504. Technol 2012;1:15e21.
[21] Davila-Vilchis JM, Mishra RS. Performance of a hydrokinetic energy system [28] Driss Z, Bouzgarrou G, Chtourou W, Kchaou H, Abid MS. Computational
using an axial-flux permanent magnet generator. Energy 2014;65:631e8. studies of the pitched blade turbines design effect on the stirred tank flow
[22] Mohamed MH. Aero-acoustics noise evaluation of H-rotor darrieus wind characteristics. Eur J Mech B Fluids 2010;29:236e45.
turbines. Energy 2014;65:596e604. [29] Ammar M, Chtourou W, Driss Z, Abid MS. Numerical investigation of turbulent
[23] Akwa JV, Vielmo HA, Petry AP. A review on the performance of Savonius wind flow generated in baffled stirred vessels equipped with three different tur-
turbines. Renew Sustain Energy Rev 2012;16:3054e64. bines in one and two-stage system. Energy 2011;36:5081e93.
[24] Damak A, Driss Z, Kchaou H, Abid MS. Conception et re alisation d’une [30] Driss Z, Ammar M, Chtourou W, Abid MS. CFD modelling of stirred tanks. Eng
aspiration. 4eme Congre
soufflerie a s International Conception et Mode lisation Appl Comput Fluid Dyn 2011;5:145e258.
mes Me
des Syste caniques. Tunisia 2011:1e8. [31] Driss Z, Abid MS. Use of the NaviereStokes equations to study of the flow
[25] Damak A, Driss Z, Abid MS. Experimental investigation of helical Savonius generated by turbines impellers. In: Navier-Stokes equations: properties,
rotor with a twist of 180 . Renew Energy 2013;52:136e42. description and applications, vol. 3; 2012. p. 51e138.