See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/315768362

Design and Analysis of a H-Rotor Darrieus

Turbine for Water Pumping Operation

Conference Paper · February 2017

CITATIONS READS

0 150

5 authors, including:

Payam Zarafshan shahriar kouravand

University of Tehran University of Tehran
41 PUBLICATIONS 137 CITATIONS 21 PUBLICATIONS 33 CITATIONS

SEE PROFILE SEE PROFILE

Ali Mashaallah Kermani

University of Tehran
10 PUBLICATIONS 4 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Object Manipulation View project

Cable Robotics View project

All content following this page was uploaded by Payam Zarafshan on 04 April 2017.

The user has requested enhancement of the downloaded file.

Design and Analysis of a H-Rotor Darrieus Turbine
for Water Pumping Operation
M. Shahmari1, P. Zarafshan2, *, Sh. Kouravand2, Ali M. Kermani2, M. Khashehchi2
Department of Agro-Technology
College of Aburaihan, University of Tehran
Pakdasht, Tehran, Iran
*
Corresponding Email: p.zarafshan@ut.ac.ir
Abstract—Renewable energies as a clean replacement
resource of fossil fuels have many advantages. One of these
energies is the wind which already is very interesting source in
the world. To use this energy, two kinds of turbines have been
developed namely; the Vertical Axis Wind Turbine (VAWT) and
Horizontal Axis Wind Turbine (HAWT). In small scale
applications, using a VAWT has some advantages such as the
lower cost and noise, simple mechanism, lower sensitivity to the
wind direction. In this paper, the design and analysis of an H-
Rotor (a VAWT type) is studied which can be used for water
pumping operation. To this end, the Computational Fluid
Dynamics (CFD) method is used to predict the output power. For
simulations, ANSYS fluent and Q-Blade software programs are
used. To simplify the rotor analysis, NACA symmetric airfoils
are considered. The results show the deviation between ANSYS
and Q-Blade predictions. Finally, the output power is calculated
and also, the water flow rate for the pump is obtained.
Keywords—Darrieus Turbine; Design; CFD; ANSYS; Q-Blade
I. INTRODUCTION
Nowadays, finding an efficient way to use renewable energies
is rapidly growing in all societies. This subject also includes
countries which have the fossil fuels like Iran. Fossil fuels
resources are limited and so, using them can be resulted in
environmental pollution. So, using renewable energies could
be feasible. One of the renewable energy resources is wind
which its technology is rapidly developing in many countries.
Wind turbine can produce the electricity or mechanical energy
directly from the kinetic energy of wind. Wind turbines are
classified in two categories: Vertical Axis Wind Turbine
(VAWT) and Horizontal Axis Wind Turbine (HAWT). In
small scale usage, the VAWT is the most common selection,
[1]. In comparison with the HAWT, the VAWT has the simple
structure, lower noise, and the lower maintenance, [1]. Wind
turbines were performed for irrigation and rustic mills since 7th
century AD in the ancient Persia, [2]. So, using wind turbines
for water pumping operation would be a common purpose on
energy consumption. In this work, the main objective is to
design and analysis of a small scale wind turbine with the aim
of water pumping. Many studies have been conducted in the
case of VAWT design, among which, Eriksson et al. [1] made
a comparison between Darrius and H-Rotor (two most famous
1- MSc Student
2- Assistant Professor
VAWT types) wind turbines. They found that the most
advantageous turbine is H-Rotor.
El-Samanoudy et al. [3] focused on the influence of some
design parameters of VAWTs; such as, the number of blades,
pitch angle and the airfoil type. They realized that on NACA
0021 airfoil, the maximum power coefficient is about 25%
under 10 degree pitch angle. Castelli et al. [4] converted the
basic BEM theory to a CFD code and analyzed a three-bladed
wind turbine (NACA 0021 type) performance. Durrani et al.
[5] compared a set of three NACA airfoils (0012, 0015, 0018)
and a thicker airfoil size NACA 0022, resulting that the
NACA 0015 has more stable performance at the various Tip
Speed Ratios (TSRs) than the others. Eleni et al. [6]
investigated the behavior of NACA 0012 in various angles of
attack. They observed that the k method is the best solving
method which was fitted with the experimental data.
Mohammad [7] studied the common symmetrical and
unsymmetrical airfoils using CFD method with the aim of
optimizing the power output of H-rotor turbine. Sabaeifard et
al. [8] compared the computational and the experimental
results of H-rotor VAWT. They showed that parameters like
the airfoil type, blade quantity and the solidity represent the
solving method. Also, the power coefficient is 0.35 in the CFD
model and 0.32 in the wind tunnel experiment. Lanzafame et
al. [9] studied the comparison between the classical turbulent
model and the Shear-Stress Transport (SST) turbulent model.
They obtained a good agreement between the experimental
and the CFD results. Giorgetti et al. [10] investigated the H-
rotor with different layouts of low TSR. They tested the layout
with a single, two and four rotors. Results show 10% increase
in comparison with the isolated rotor. Jouilel et al. [11]
compared different NACA airfoil shapes, and found that the
NACA 00XX airfoil’s manufacturing cost is less than the
NACA 44XX. Mohammad et al. [12] focused on the H-rotor
performance with different airfoils in low Reynolds number.
They compared two meshing models, Fluent and Gambit
software programs, with different solving methods. They
achieved that the SST k-ω method is suitable on ANSYS
meshing and the k-ϵ method is realizable on Gambit meshing.
Nguyen et al. [13] investigated the effect of the NACA 00XX
thickness on the starting capability. They compared four
NACA symmetric airfoils and found that the best starting
capability is related to the thicker airfoil. Bhushan et al. [14]
considered the CFD analysis of NACA 0012 at different
Reynolds numbers and different pitch angles, and results
compared with the experimental data of Sandia National
Laboratory, whereas the lift and the drag coefficient increased
in the higher Reynolds number. Sahin and Acir [15] studied
the NACA 0015 performance in different Reynolds numbers
and different pitch angles. Extracted results of CFD compared
with the experimental data which obtained from the wind
tunnel test. They found that the best performance of the rotor
occurs at pitch angle 8°.
In this paper, an investigation is conducted to design a wind
turbine with the purpose of using the output power for water
pumping operation. So, a VAWT is selected for better
performance at low wind speed. To this end, H-rotor Darrieus
turbine is considered. To select the blades, the NACA 0015
symmetrical airfoil has been proposed. Also, Tehran is the
pilot of investigation and so, the wind data is obtained and
analyzed to find the best station with the most probability of
wind velocity. In this regard, rotor is designed with 3 blades
and 1.5 m height. Next, simulations are performed on ANSYS,
Fluent and Q-Blade software programs with the wind speed
5.6 m/s. In fact, these two common software programs are
selected to increase the results accuracy. Finally, obtained
results are discussed.
height 1.5 m and chord length 0.18 m, it is necessary to
consider a stronger rotor which in this study is the thicker one
(either NACA 0015 or NACA 0018). Due to the similar
behavior of NACA 0015 and NACA 0018 as shown in Fig. 3,
and the fact that the thicker airfoil has more weight than the
other one, NACA 0015 airfoil is selected as the best case. To
validate the obtained results, the performance of NACA 0015
is compared with the existing literature, [6, 13]. In CFD
analysis, ANSYS Fluent software was used. The initial data
which were obtained from Q-Blade software was set as the
input data at ANSYS Fluent software. In the next step, a rotor
solving domain needs to be defined. For the number of blades
in the rotor, the best operational configuration is related to the
rotor with 3 blades, [4]. Fig. 4 shows the rotor domain with 3
blades which is modeled in ANSYS software. The height and
the chord length of blades were selected as 1.5 m and 0.18 m,
respectively. To prevent the boundaries influence on the study,
the domain is defined as the square with length 26D, where D
defined as the rotor diameter.

II. INVESTIGATIN METHOD

A. Simulation Setup
To model the turbine, an H-rotor Darrieus, shown in Fig. 1,
has been chosen, due to its simple structure. This is the
advantage of H-rotors in comparison with the other Darrieus
turbines, [1]. For the wind velocity, collected data since 2009
were taken from Iran Methodological Organization. Also,
Tehran was selected for investigation pilot. So, data were
analyzed with Weibull distribution function, and results
showed that the average wind velocity is 5.6 m/s. To this end, Fig. 1. CAD model of H-Rotor turbine
a graphical method has been chosen for Weibull distribution
Fig. 5 shows the boundary conditions and the boundary type.
function. Data of 12 stations within Tehran are available
The left and the right borders are defined as the velocity inlet
which were studied to achieve the best situation of the wind
(5.6 m/s) and the pressure outlet (Gauge Pressure),
flow. The analysis showed that the best station is located in
respectively. Also, two wall symmetrical boundaries are
the northern part of Tehran at Firozkoh Global Atmospheric
defined in the top and the bottom borders. In the middle of
Watch (GAW), where the air properties are summarized in
domain, a circle with 1 m diameter is defined as the rotating
Table 1. After data collection, airfoil type needs to be
zone. Between this rotating zone and the square domain, an
considered, so symmetric NACA 00XX airfoils were selected.
circular interface is defined to achieve the velocity continuity.
The most common airfoils among H-rotor turbines, as studied
Table 2 shows the main charachteristic of the designed rotor.
in the literature, are 0012, 0015, 0018, [6]. After airfoil
All measuring areas are meshed with triangular method. For
selection, simulations were performed by open source Q-
the area near the blades, the meshing resolution is increased.
Blade software to ensure the airfoils compatibility. So, all
The meshing structure is shown in Fig. 6. For accurate
airfoil models (as shown in Fig. 2) were analyzed with Q-
prediction of the flow near the blades’ wall, the structural
Blade software. All simulations were conducted with the
mesh with rectangle shape is used as shown in Fig. 7. At the
chord length 0.18 m and the radius 0.5 m. Fig. 3 shows the
first, simulation is started in Fluent software with steady state
power coefficient (Cp) for the mentioned three airfoil types,
condition. It should be noted that in the simulation, a moving
whereas the airfoils NACA 0015 and NACA 0018 show a
frame has been chosen. After 1000 iterations, the simulation is
similar pattern in all TSR values. Also, the best obtained TSR
coupled with the other one. In the second simulation, the
is 3.5. Moreover, the NACA 0012 at TSR 3.5 has the same Cp
sliding mesh method is used. Also, TSR of rotor has been set
as the two other models. But, based on the input parameters;
to 3.5. So, the angular velocity is found to be 40 rad/s. To
achieve the better quality results, the time step size is 0.00044 TABLE II. DESIGN FEATURES OF SIMULATED ROTOR
sec, which is related to 1 degree rotating of the rotor. In Features
Denom
addition, 1800 time steps are selected which is equal to 5 turns ination Chord Blade
Height Diameter No. of blade
length section
of the rotor. For each time step, 50 iterations are also selected,
and residuals are adjusted to 1e-5. After these 1800 time steps, Rotor 1.5 m 1m 3 0.18 m NACA0015
the momentum coefficient, Cm, is captured for one more
rotation (360 time steps). Therefore, the total time step is
2160. In addition, the realizable k   solving method is
selected. For soulotion method, the pressure-velocity coupled
is chosen. Pristo solving method is selected for the pressure
and the momentum. Also, the turbulent kinetic energy and
turbulent dissipation rate are adjusted to the second order
upwind. So, the accuracy and the total solving time of
simulation are increased. This method is the most popular for
the prediction of rotating turbines, [13]. Also, the realizable
k   solving method can predict the flow near the blades in
the strong pressure gradient areas better than the standard
k   method, [13].
TABLE I. AIR STANDARD PROPERTIES
Fluid Fluid properties
type Density (kg/m3) Viscosity (kg/m-s) Pressure (Pa)
Fig. 5. Domain bundary with the length of 26D to prevent the boundaries
Air 1.225 1.7894e-05 101325 influence

0.40
0.35
Power Ceofficient

0.30
0.25
0.20
0.15
0.10
0.05
0.00
1.00 1.50 2.00 2.50 3.00 3.50 4.00 Fig. 6. The meshing structure

TSR

NACA 0012 NACA 0015 NACA 0018

Fig. 3. Cp vs TSR simulation result from Q-Blade software

Fig. 7. Structure mesh near the blades with aim of increasing accuracy of
flow prediction

B. Mathemticall Equations
Weibull distribution equation can be stated as:
 k  .  
k 1
   k 
F()     .exp      (1)
Fig. 4. Overall modeled domain in ANSYS software
C C  C 
where C and k are defined as the Weibull scale and shape 2
parameters respectively. For the graphical method, Eq. (1) is τ ij = 2μ t Sij - ρkδij (9)
converted to the logarithmic form as: 3
ln[ ln(1  F())]  k ln()  k ln(C) (2) where μ t or the eddy viscosity is a scalar property and so:
By plotting ln() and [ ln(1  F())] , the Weibull scale and u i u j 2 u k
shape parameters are achievable, [16]. To find the power -ρu i u j  μ t (  ) (ρk  μ t )δij (10)
x j x i 3 x k
coefficient of rotor, Fluent software could give Cm. Total
available power which could achieve from the wind is related Being of Reynolds stress tensor as the proportion of strain
tensor is the main disadvantage of Boussinesq theory. Future
to the velocity of wind flow ( V ), the total area of passing model like the k-ε developed model covers this shortcoming.
flow from the rotor (A) and the flow density ( ρ ) as: The transport equations are given as:
1    μ k
Pmax  (ρk)  (ρku j )  [(μ  t ) ]  pk  pb
3
ρV A (3)
2 t x j x j σ k x j (11)
Power coefficient is described as the power ratio which the
rotor could gain from the flow to the total available power as: ρε  YM  Sk
P and:
Cp = (4)    μ ε
Pmax (ρε)  (ρεu j )  [(μ  t ) ]  ρC1Sε
t x j x j σ ε x j
For the non-dimensional lift, drag and moment coefficient are, (12)
[17]: 2
ε ε
L D M
ρC 2 + C1ε C3ε Pb + Sε
CL = ,C = ,C = (5) k+ νε k
D M
q S q S q  Sl where:
η k
where L is the lift force, S is the reference area, q  is the S= 2SijSij , C1 = max[0.43, ],η = S (13)
η+5 ε
dynamic pressure of flow, D and M are the drag force and
pitching moment respectively and l is the reference length. and the model constants are:
The TSR ratio is also defined as: C1ε = 1.44,C2 = 1.9,σ k = 1.0,σε = 1.2 (14)
rω It is important to note that this model is known as the most
λ= (6) expensive computationally model, [19].
V
where r is the rotor radius and ω is the angular velocity. Total
Cp is given by Cm and the relation between Cm and Cp can be III. RESULTS AND DISCUSSION
defined as, [18]: Weibull distributions of 12 stations are investigated and it
found that the best station with the most probability of wind
CP = CM λ (7)
flow at year is Firozkoh GAW. Scale and shape coefficient is
6.25 and 1.513 respectively. This Weibull distribution with the
B.1: Realizable k - ε Solving Method:
graphical method is shown in Fig. 8. Average velocity of wind
Reynolds-averaged Navier-Stokes equation can be given as:
found as 5.6 m/s and the maximum velocity is 10 m/s. TSR is
  p  u i u j 2 u 3.5 for the NACA 0015 blades which are used for the
(ρu i )  (ρu i ρu j ) =   [μ(   δ ij l )]
t x j x i x i x j x i 3 x l simulation to achieve the best prediction of Cp. Output average
(8) value of Cm from ANSYS and Fluent software programs is
 0.04. To find the Cp coefficient, Eq. 7 has been used and it is
 (-ρu i u j )
x j calculated as 0.14. To approve the obtained value, a
simulation is done in Q-Blade software and also, all data are
The non-linear term of equation or (-ρu i u j ) is described that compared with the studied literatures. Q-blade software
the fluctuation still appear in RANS. This influence is called predicts the value of Cp in TSR 3.5 for the NACA 0015 about
the Closure problem. To solve this problem, Boussinesq is 0.34 as shown in Fig. 9. Also, the total achievable energy from
introduced the concept of eddy viscosity. This theory is the wind has been calculated as 164.4 w. According to the
showed that the momentum transfer is caused by the turbulent Betz limit, the maximum value of Cp is 0.59. Thus, the total
that could be modeled with the eddy viscosity. The relation available power from the wind is 97 w. Fig. 10 shows the
between the Reynold stress tensor and strain tensor can be relative speed of blade. Considering the incompressible flow
defined as: theory, its acceptable value is Ma<0.15. With the obtained data
from Q-Blade software, the considered H-rotor could produce
52.8 w. Also, the output value from Fluent software is
compared with this data. With the exported Cp from ANSYS
and Fluent software programs, the total power is 23 w.
According to [13] for the NACA 0015, it is expected to find
the value about 0.15 for ANSYS and Fluent simulation. This
value has the good agreement with the value which is found in
the stated study. It founds that with Q-Blade output value, the
water could be pumped to the 10 m height with the flow rate
1.9 m3/h. Also, with ANSYS and Fluent obtained result, the
flow rate is 0.84 m3/h. Fig. 11 shows the average Cp of rotor
over 5 revolutions in Q-Blade software. Total turbulent
vectors are shown in Fig. 12. Cm output from ANSYS and Q-
blade software programs is shown in Figs. 13 and 14
respectively.
3 with mechanical water pump to produce the needed torque.
2 y = 1.513x - 2.7739
R² = 0.981
1
0
ln(v)
assumption of study. To this end, simulations were performed
using ANSYS Fluent and Q-Blade software to increase the
results accuracy. According to the results, the best station with
the most probability of wind along one year was Firozkoh
GAW. In this station, the maximum and mean velocity was 10
m/s and 5.6 m/s, respectively. Using the simulation results, the
best performance of blades for H-rotor turbine was related to
the NACA 0015. Also, the best value of TSR for the NACA
0015 was 3.5. Moreover, the simulation was performed with
the k   method. Finally, the obtained results were shown
the good accuracy whereas this CFD method needed more
time and cost. In Fact, for the small scale water pumping
proposes, the investigated H-rotor could be directly coupled

-1 0 0.5 1 1.5 2 2.5 3

-2
-3
-4
ln(1-F(v))

Weibull distribution Linear (Weibull distribution)

Fig. 8. ln(v) vs ln(1-F(v)) scheme for scale and shape parameter of Weibull
distribution Fig. 10. Velocity magntitude for maximum limitation of 45 m/s (Ma <0.15)

0.5 1.0
0.4
0.9
0.3
0.2 0.8
Powe Ceofficient

0.1 0.7
Power Ceofficient

0.0
0.6
-0.1 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
-0.2 0.5
-0.3 0.4
-0.4
-0.5 0.3
-0.6 0.2
-0.7 0.1
-0.8
-0.9 0.0
-1.0 0.0 0.5 1.0 1.5
TSR Time (sec)

NACA 0012 NACA 0015 NACA 0018

Fig. 11. Average Cp vs time for simulation of rotor in Q-Blade after 5
revolutions
Fig. 9. Cp vs TSR from Q-Blade software for 3 NACA symmetric airfoils

IV. CONCLUSIONS
[1]
In this paper, the design and analysis of H-rotor wind turbine
with the aim of water pumping operation was proposed.
Considering the availability of low speed wind, Tehran was [2]
chosen as the investigation pilot. The mentioned wind velocity
was related to the lower Reynolds number which was the main
REFERENCES
Eriksson, S., Bernhoff, H., and Leijon, M., Evaluation of different
turbine concepts for wind power. Renewable and Sustainable Energy
Reviews, 2008. 12(5): p. 1419-1434.
Prasad, S. and Auradi, V., Optimized Design of Rotor Blade for a Wind
Pump. International Journal of Renewable Energy Research (IJRER),
2012. 2(4): p. 746-749.
 Averaged CP: 0.36513
[3] El-Samanoudy, M., Ghorab, A., and Youssef, S. Z., Effect of some V_in @ hub: 5.6 m/s
design parameters on the performance of a Giromill vertical axis wind
turbine. Ain Shams Engineering Journal, 2010. 1(1): p. 85-95.
[4] Castelli, M. R., Englaro, A., and Benini, E., The Darrieus wind turbine:
Proposal for a new performance prediction model based on CFD.
Energy, 2011. 36(8): p. 4919-4934.
[5] Durrani, N., A detailed aerodynamic design and analysis of a 2d vertical
axis wind turbine using sliding mesh in cfd. in 49th AIAA Aerospaces
Sciences Meeting and Exhibit, Orlando, FL, Jan. 2011.
[6] Eleni, D. C., Athanasios, T. I., and Dionissios, M. P., Evaluation of the
turbulence models for the simulation of the flow over a National
Advisory Committee for Aeronautics (NACA) 0012 airfoil. Journal of
Mechanical Engineering Research, 2012. 4(3): p. 100-111.
[7] Mohamed, M., Performance investigation of H-rotor Darrieus turbine
with new airfoil shapes. Energy, 2012. 47(1): p. 522-530.
[8] Sabaeifard, P., Razzaghi, H., and Forouzandeh, A., Determination of
vertical axis wind turbines optimal configuration through CFD Vortex Elements: 4988
simulations. in International Conference on Future Environment and Fig. 12. Velocity magnetiude vectors after 5 revolutions
Energy. 2012.
[9] Lanzafame, R., Mauro, S., and Messina, M., 2D CFD modeling of H-
Darrieus wind turbines using a transition turbulence model. Energy 0.3
Procedia, 2014. 45: p. 131-140. 0.2
[10] Giorgetti, S., Pellegrini, G., and Zanforlin, S., CFD investigation on the
aerodynamic interferences between medium-solidity Darrieus Vertical 0.2
Axis Wind Turbines. Energy Procedia, 2015. 81: p. 227-239. 0.1
[11] Jouilel, N., Radouani, M., and El Fahime, B, Geometric Modeling and

Cp
Contribution to the choice of a Blade profile. in Xème Conférence 0.1
Internationale: Conception et Production Intégrées. 2015. 0.0
[12] Mohamed, M., Ali, A., and Hafiz, A., CFD analysis for H-rotor
Darrieus turbine as a low speed wind energy converter. Engineering
-0.1 0 100 200 300 400
Science and Technology, an International Journal, 2015. 18(1): p. 1-13. -0.1
[13] Nguyen, C., Le, T., and Tran, P., A numerical study of thickness effect of
-0.2
the symmetric NACA 4-digit airfoils on self starting capability of a 1kW Time (sec)
H-type vertical axis wind turbine. Int. J. Mech. Eng. Appl, 2015. 3(3-1):
p. 7-16.
[14] Patil, B. S. and Thakare, H. R., Computational Fluid Dynamics Analysis Fig. 13. Cp against time for Fluent simulation
of Wind Turbine Blade at Various Angles of Attack and Different
Reynolds Number. Procedia Engineering, 2015. 127: p. 1363-1369.
[15] Şahin, İ. and Acir, A., Numerical and Experimental Investigations of Lift
0.6
and Drag Performances of NACA 0015 Wind Turbine Airfoil.
International Journal of Materials, Mechanics and Manufacturing, 2015. 0.4
3(1): p. 22-25.
[16] Kidmo, D., Statistical analysis of wind speed distribution based on six 0.2
Cm

Weibull Methods for wind power evaluation in Garoua, Cameroon.

Revue des Energies Renouvelables, 2015. 18(1): p. 105-125. 0.0
[17] Anderson Jr, J. D., Fundamentals of aerodynamics. 2010: Tata 1.2 1.3 1.4 1.5 1.6
McGraw-Hill Education. -0.2
[18] Patel, M. B. and Kevat, V., Performance prediction of straight bladed
Darrieus wind turbine by single streamtube model. International Journal
-0.4
of Advanced Engineering Technology, 2013. 14: p. 2. Time (sec)
[19] Marzouk, O. A. and Huckaby, E. D., Simulation of a swirling gas-
particle flow using different k-epsilon models and particle-parcel
relationships. Engineering Letters, 2010. 18(1): p. 5 Fig. 14. Cm against time for Q-Blade simulation

X-Scale = 1.0
Y-Scale = 1.0
x = 1.0609
y = 0.1357

Fig 2. Inputed blades shape to Q-Blade software

View publication stats