Design Optimization of Vertical Axis Wind Turbine

Faseeh Ahmed Khan, Farooq Akram, Hassan Raza

Dept. of Aerospace Engineering
College of Aeronautical Engineering, National University of Sciences and Technology - NUST
Risalpur, Pakistan
[email protected], [email protected], [email protected]

Abstract— Wind energy is currently one of the fastest growing been pursued in space constrained area due to large
green energy sources in the world because of its importance in urbanization. It has a simplified geometry with no yaw
national and international energy policy in response to climate mechanism and only single moving part or, in case of straight-
change and energy crisis. The most conventional and famous turbine bladed VAWTs, no pitch regulation, and have neither tapered
used to extract wind energy is Horizontal Axis Wind turbine. While
nor twisted blades, making them relatively easy to extrude and
the new concept i.e. Vertical Axis Wind Turbine has advantage over
Horizontal Wind Turbine due of its low cut-in speed, independence
fabricate [1]. Furthermore, they are omni-directional i.e. they
of wind direction and any yawing mechanism. Commercially can take the wind from any direction regardless of orientation
available wind turbines are designed to produce rated power output at [2]. The geometric parameters which define the performance
wind speeds of 10-14 m/s and their performance significantly drop of a turbine and its aerodynamic characteristics are rotor
down at off-design conditions. Keeping in view the low average wind radius, chord, height, number of blades etc [3].
velocity available in many areas of Pakistan, commercially available
high wind speed turbines produce less power as suggested at wind
speeds of 5 m/s to 7m/s. This resulted in an idea to aerodynamically
optimize the wind turbine with an aim to maximize power coefficient
for low wind speeds. Comparison is done between the optimized
design operating at low wind speed with the available design at high
wind speed. The final optimized design showed a better behavior of
Power coefficient with Tip Speed Ratio. This paper aims at design,
analysis and optimization of Straight Bladed Darrieus wind turbine
with Double Multiple Streamtube Theory used for performance
prediction of Darrieus wind turbine. Also, the design space is
generated to define constraints for turbine geometric parameters and
BOX BEHNKEN technique is used to generate DESIGN OF
EXPERIMENT.

Keywords: Wind Energy, Vertical Axis Wind Turbine, Wind

Turbine Design, Aerodynamics, Optimization, Design of
Experiment, Double Multiple Streamtube Model. Fig. 1. New Capacity Installation (MW) in Global Wind Energy Industry

I. INTRODUCTION The constraints on which design of a wind turbine is founded

are size of turbine, rated power output, cost of production and
To explore the alternative energy resources, efforts have wind speed on which it will operate. The performance of a
geared up with the reducing stocks of the fossil and a need to wind turbine is largely dependent on location where it is
lessen the greenhouse gases. The first few steps to move installed which ultimately results in the selection of shape,
towards the wind energy are taken from the individual to aerodynamic design and profile. Not only that, the design
governmental levels all around the globe. In spite of the power output value which is to be extracted largely depends
economic crisis around the globe, investment in new wind on the rotational speed and Reynold Number of the turbine
turbines exceeded by far all preceding years. The market of blade [4]. In this paper, the design, analysis and optimization
global wind energy is expected to continue its slow growth of Giromill (also known as H-Darrieus) VAWT is described
pace, by the end of 2017, world’s capacity and capability of i.e. how to optimize the coefficient of power w.r.t its tip speed
installed wind power is going to reach up to 572 GW as shown ratio [5].
in Fig. 1.
There are two main categories of wind turbine which produce
electrical energy from the wind: this includes vertical axis
wind turbines (VAWTs) and horizontal axis wind turbines
(HAWTs). Being better in performance and efficiency,
Horizontal wind turbines are widely in use but the latter have
 energy. In that case the maximum available energy is given by
1
Pmax = ρ ṁV ∞2.
2
The wind speed is practically not reduced to zero due to a
finite blockage. Therefore, only a fraction and small portion of
maximum extractable power is achieved. This fraction is of
great significance in judging the efficiency of turbine and is
represented by power coefficient Cp. This power coefficient
Cp is defined as the ratio of the extracted power and the
maximum available power in the wind.

Captured mechanical power by blades P

C p= =
Power available∈ wind 1
ρA V ∞3
2

The optimal operating conditions of a wind turbine depend

Fig. 2. Schematic View of H-Darrieus Turbine
upon the rotor solidity ratio and TSR [6].

TABLE I. NOMENCLATURE Nc
Solidity Ratio=σ =
R
HAWT Horizontal Axis Wind Turbine
VAWT Vertical Axis Wind Turbine Rω
Cp Coefficient of Power Tip Speed Ratio=λ=
V∞
P Power
ρ Density In case of VAWTs, theoretical maximum achievable limit of
ṁ Mass Flow Rate the Cp is not given by Betz limit of Cp max = 16/27 which
V∞ Freestream Velocity comes out to be 59.3% rather Paraschivoiu [6] stated that it
A Frontal Area could go upto 64% according to actuator disk momentum
σ Solidity Ratio theory.
λ Tip Speed Ratio (TSR)
N Number of Blades
C Airfoil Chord III. AERODYNAMIC PREDICTION AND CALCULATION OF
R Rotor Radius DARRIEUS WIND TURBINE
ω Rotational Speed Flow analysis of VAWT is a complex thing to understand.
Va Axial Velocity Several mathematical models, based on different theories, are
Vt Tangential Velocity available for the performance prediction and design of
Vn Normal Velocity Darrieus-type VAWTs. The model which is used as an initial
W Relative Velocity design of VAWT is MOMENTUM model because it is one of
Α Angle of Attack the most accurate and famous model used for such purpose.
Vau Upstream Axial Velocity
Vad Downstream Axial Velocity
A. Variation in Relative Velocity and Angle of Attack
U Interference Factor
Re Reynold Number While describing momentum model, Islam [1] explained that
µ Viscosity in the Darrieus wind turbines, flow is considered to occur in
axial direction splitting the flow into tangential (chordal)
II. MAXIMUM AND EXTRACTABLE POWER component and normal component.
A wind turbine starts rotating and producing electric energy
due to kinetic energy transfer from the oncoming air. This V c =V t =Rω+V a cosθ
rotational motion is first of all transformed into the
mechanical energy through the mechanism of gears which is V n=V a sinθ
attached with the rotating shaft and later through the generator
box, is converted into the electrical power. The maximum Here Va is the velocity of the axial flow.
energy is extracted through air if theoretically the wind speed From the Fig. 3, angle of attack α can be calculated as
is reduced to zero i.e. extracting all the available kinetic
 Vn DOWNSTREAM:
α =tan
−1
( )
Vt
V ad =ud ( 2
V au
−1)V ∞ i
According to the actuator disk theory, V ∞i
V ad
V ∞ +V w ud =
V a= Ve
2
If there is negligible variation in the velocities then
W =√V t2 +V n2
2
W =V ∞ √( λ+cos θ ) +sin 2 θ

Fig. 4. Double multiple streamtube model

V. DESIGN AND ANALYSIS OF DARRIEUS WIND TURBINE

First of all, the geometric parameters were set and defined and
an initial design was created using DMS model for 1kW
power output and 7 m/s freestream velocity. As many
variables are involved in designing a turbine so to check their
variation individually i.e. trend analysis doesn’t give a good
Fig. 3. Flow Velocities
approximation, for that Design of Experiment (DOE) was
used as an optimization technique to check out overall effect
of variables in performance of a turbine [10]. Pitch angle is
IV. DOUBLE MULTIPLE STREAMTUBE MODEL one of the variables that affect the turbine performance. From
Out of all the momentum models, the most accurate i.e. -4o to 4o, pitch angle is studied and it is concluded that the best
DOUBLE MULTIPLE STREAMTUBE MODEL is used. In performance for the straight bladed fixed pitch Darrieus wind
this model, calculation is done separate for the upstream and turbine is at zero degree pitch angle [11].
downstream half cycles of the turbine [7]. Induced velocities
are obtained using the principles of two actuator discs in
tandem [1, 8, 9]. These are the most refined model except in
case of large TSR and high solidity ratio [6].
From Fig. 4,
UPSTREAM:
V au
V e =V ∞ i (2 −1)
V ∞i
V au Fig. 5. Cp variation with Pitch Angle
uu =
V ∞i
Box-Behnken design, which is a quick response surface
methodology, gives us an idea of how to achieve an optimal
design while keeping the variables within limit [12]. Total of
50 runs were carried out for the DOE technique in which
output POWER and its COEFFICIENT was calculated using
the procedure described in the following flowchart [13, 14].

Fig. 6. Design procedure

In Fig. 6, a flowchart for designing the wind turbine is
adopted. First of all an airfoil is selected for a turbine.
Traditionally NACA 4-digit series have been used for
Darrieus- type VAWTs [15] and most of the time its
symmetric airfoil (NACA 0018), no doubt the thin symmetric
airfoils have a self-starting problem [16], which according to
some authors, can be solved with variable pitch angle of
turbine [17] or by introducing guide vanes [18]. But the main
advantage of symmetric airfoil is that it gives lift during the
complete rotation of the wind turbine making its coefficient of
power high and stable [19]. After that, lift and drag
coefficients at different angle of attacks are calculated using
the software XFOIL. XFOIL is an collaborating software for
the analysis and design of subsonic airfoils. Given the airfoil
shape coordinates, Mach and Reynold numbers, XFOIL can
calculate the distribution of pressure on the airfoil and
Fig. 7. Power Actual vs Power Predicted
hence lift and drag characteristics [20, 21]. These
characteristics are then extrapolated to 360 o as the wind
turbine is rotated to the complete 360o azimuth angle, for that
a circular foil is generated to help the turbine rotate in a circle.
Reynold number directly affects the performance of the wind
turbine because as the Reynold number increases, so is the lift
coefficient [22, 23] and for its calculation relative velocity is
used.

ρWc
ℜ=
µ
After finalizing the airfoil, blade chord, height, turbine radius,
pitch, number of blades etc. are defined to describe the
complete shape of a turbine. Finally, the wind speed and
rotational speed of the turbine is set to carry out the simulation
using Double Multiple Streamtube Model and to predict the Fig. 8. Power Residual vs Power Predicted
behavior of the turbine.

VI. DESIGN OF EXPERIMENT AND OPTIMIZATION

To analyze the 50 runs of Design of Experiment, the model
and data is fitted using Least Square Method. The least
squares approach is a standard method in regression
analysis to the approximate and estimated solution and result
of systems those are overdetermined [24].  "Least squares"
means that the solution tries to minimize overall the squares of
the errors sum made in the each and every single equation
results. The best fit in the least-square minimizes and reduces
the sum of squared errors and residuals i.e. a difference
between the fitted value and an observed value provided by a
model. To minimize the sum of squared residuals and for a
better fit, combination of variables upto 3 rd degree is used as
the response of the variables for the POWER output and Fig. 9: Cp Actual vs Cp Predicted
POWER COEFFICIENT was nonlinear. This gives us the
responses between predicted power, actual power and residual
power.
 Fig. 12. Optimized Cp curve comparison

In the case of straight bladed Darrieus wind turbines, the best

operating TSR is 3-6 [26] and value of solidity ratio is
between 0.1-0.2 [27]. Fig. 12 shows how the Cp of three
designs vary with TSR and within the operating range,
optimized design shows the best result and performance.
Power relation with velocity of the optimized design is shown
Fig. 10. Cp Residual vs Cp Predicted
in Fig. 13.
The above four figures show how actual, predicted and
residual Power and Cp vary with each other. The more residual
values close to zero, the better the fit is. After analyzing these
responses, surrogate modelling is done to generate an equation
out of these designs with the minute errors as shown in above
figures to get an optimal final design and model. Surrogate
model is an engineering method which is used when the
noticeable outcome cannot be measured easily and directly, so
a model of the outcome is used instead. It is an approximate
response surface model that mimics the performance of the
model of simulation as closely and thoroughly as possible
Fig. 13. Power vs Velocity curve
while being computationally cheap to approximation and
estimate. [25].
This surrogate model helps in generating the equation which is
then exported to MATLAB for optimization. Upper/Lower
bounds are defined, nearly ± 25% of the designed values. The
comparison between initial result (evaluated before final
optimization in MATLAB) and what Brusca [23] showed in
his paper are

Fig. 11. Cp curve comparison

In Fig. 11, the variation in the curve is obvious because two

different momentum models (Multiple Streamtube and Double
Multiple Streamtube) were used for performance prediction.
These results were then compared with the MATLAB results
i.e. the final optimized design generated using FMINCON
mathematical algorithm. The results of initial design and final
optimized design are then compared. Fig. 14. CAD Model
 VII. CONCLUSION [11] M. Mohamed, A. Ali, and A. Hafiz, "CFD analysis for H-rotor
Darrieus turbine as a low speed wind energy converter,"
This work aims at designing a vertical wind turbine to Engineering Science and Technology, an International Journal,
maximize its coefficient of power in the operating range of vol. 18, pp. 1-13, 2015.
TSR. It is concluded that the power coefficient is influenced [12] S. C. Ferreira, R. Bruns, H. Ferreira, G. Matos, J. David, G.
by Reynolds number and rotor solidity. The aerodynamic Brandao, et al., "Box-Behnken design: An alternative for the
design of a commercial vertical axis wind turbine is improved optimization of analytical methods," Analytica chimica acta,
by adopting Design of Experiment technique. In DOE, various vol. 597, pp. 179-186, 2007.
combinations of geometric variables of 1kW turbine design [13] D. Marten, J. Wendler, G. Pechlivanoglou, C. Nayeri, and C.
were arranged and Power and its Coefficient were calculated Paschereit, "QBLADE: an open source tool for design and
simulation of horizontal and vertical axis wind turbines,"
using the Double Multiple streamtube model developed by I. International Journal of Emerging Technologies and Advanced
Paraschivoiu. The trend we get from this technique is a key Engineering, vol. 3, pp. 264-269, 2013.
towards optimization. Due to multiple functions and [14] D. Marten and J. Wendler, "QBlade Guidelines," Version 0.6,
constraints and lesser computational intensity, FMINCON is TUBerlin, Germany, 2013.
used as optimization solver, despite of higher number of [15] M. Islam, D. S. Ting, and A. Fartaj, "Desirable airfoil features
variables involve, which yield better optimization results in for smaller-capacity straight-bladed VAWT," Wind
achieving the global/local optima. Engineering, vol. 31, pp. 165-196, 2007.
[16] R. Bos, "Self-starting of a small urban Darrieus rotor," Delft
University of Technology, 2012.
[17] P. Cooper and O. C. Kennedy, "Development and analysis of a
REFERENCES novel vertical axis wind turbine," 2004.
. [18] Y. Lim, W. Chong, and F. Hsiao, "Performance investigation
[1] M. Islam, D. S.-K. Ting, and A. Fartaj, "Aerodynamic models and optimization of a vertical axis wind turbine with the omni-
for Darrieus-type straight-bladed vertical axis wind turbines," direction-guide-vane," Procedia Engineering, vol. 67, pp. 59-69,
Renewable and Sustainable Energy Reviews, vol. 12, pp. 1087- 2013.
1109, 2008. [19] M. S. Hameed and S. K. Afaq, "Design and analysis of a straight
bladed vertical axis wind turbine blade using analytical and
[2] K. Pope, G. Naterer, I. Dincer, and E. Tsang, "Power correlation
numerical techniques," Ocean Engineering, vol. 57, pp. 248-
for vertical axis wind turbines with varying geometries,"
255, 2013.
International Journal of Energy Research, vol. 35, pp. 423-435,
[20] M. Drela, "XFOIL: An analysis and design system for low
2011.
Reynolds number airfoils," in Low Reynolds number
[3] A. Bianchini, G. Ferrara, and L. Ferrari, "Design guidelines for aerodynamics, ed: Springer, 1989, pp. 1-12.
H-Darrieus wind turbines: Optimization of the annual energy [21] D. Marten, G. Pechlivanoglou, C. Nayeri, and C. Paschereit,
yield," Energy Conversion and Management, vol. 89, pp. 690- "Integration of a WT Blade Design tool in XFOIL/XFLR5," in
707, 2015. 10th German Wind Energy Conference (DEWEK 2010),
[4] M. A. Naqvi, A. Abbas, M. Hamid, H. Tarar, and L. Ahmad, Bremen, Germany, Nov, 2010, pp. 17-18.
"Aerodynamic Design Optimization of Residential Scale Wind [22] R. E. Sheldahl and P. C. Klimas, "Aerodynamic characteristics
Turbine Blades for Lower Wind Speeds," International Journal of seven symmetrical airfoil sections through 180-degree angle
of Renewable Energy Research (IJRER), vol. 5, pp. 373-385, of attack for use in aerodynamic analysis of vertical axis wind
2015. turbines," Sandia National Labs., Albuquerque, NM
[5] M. M. Aslam Bhutta, N. Hayat, A. U. Farooq, Z. Ali, S. R. (USA)1981.
Jamil, and Z. Hussain, "Vertical axis wind turbine – A review of [23] S. Brusca, R. Lanzafame, and M. Messina, "Design of a vertical-
various configurations and design techniques," Renewable and axis wind turbine: how the aspect ratio affects the turbine’s
Sustainable Energy Reviews, vol. 16, pp. 1926-1939, 2012. performance," International Journal of Energy and
[6] I. Paraschivoiu, Wind turbine design: with emphasis on Environmental Engineering, vol. 5, pp. 333-340, 2014.
Darrieus concept: Presses inter Polytechnique, 2002. [24] K. Choi, B. D. Youn, and R.-J. Yang, "Moving least square
method for reliability-based design optimization," in 4th world
[7] H. Beri and Y. Yao, "Double multiple streamtube model and
congress of structural and multidisciplinary optimization, 2001,
numerical analysis of vertical axis wind turbine," Energy and
pp. 4-8.
Power Engineering, vol. 3, p. 262, 2011.
[25] C. Audet, A. J. Booker, J. Dennis Jr, P. D. Frank, and D. W.
[8] I. Paraschivoiu, "Double-multiple streamtube model for
Moore, "A surrogate-model-based method for constrained
Darrieus in turbines," in Wind turbine dynamics, 1981, pp. 19-
optimization," AIAA paper, vol. 4891, 2000.
25.
[26] J. DeCoste, D. McKay, B. Robinson, S. Whitehead, S. Wright,
[9] R. Melício, N. Batista, J. Matias, and J. Catalão, "Vertical axis M. Koksal, et al., "Vertical Axis Wind Turbine," Mechanical
wind turbine performance prediction: a new approach to the Engineering Dalhousie University, 2005.
double multiple streamtube model," 2012. [27] J. DeCoste, D. McKay, B. Robinson, S. Whitehead, S. Wright,
[10] A. Keane, "Wing optimization using design of experiment, M. Koksal, et al., "Self-starting vertical axis wind turbine,"
response surface, and data fusion methods," Journal of Aircraft, Final Year Design Project, Mechanical Engineering, Dalhousie
vol. 40, pp. 741-750, 2003. University, Canada, 2006.