Journal of Physics: Conference Series

PAPER • OPEN ACCESS

Comparison of simulations and wind tunnel measurements for the

improvement of design tools for Vertical Axis Wind Turbines
To cite this article: Sarah Barber and Henrik Nordborg 2018 J. Phys.: Conf. Ser. 1102 012002

View the article online for updates and enhancements.

This content was downloaded from IP address 139.81.190.197 on 11/10/2018 at 02:20

Global Wind Summit 2018 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1102 (2018)
1234567890 ‘’“” 012002 doi:10.1088/1742-6596/1102/1/012002

Comparison of simulations and wind tunnel

measurements for the improvement of design tools
for Vertical Axis Wind Turbines
Sarah Barber and Henrik Nordborg
University of Applied Sciences Rapperswil, Oberseestrasse 10, 8640 Rapperswil, Switzerland
E-mail: sarah.barber@hsr.ch

Abstract. As wind turbines are getting larger and cost reductions through up-scaling are
reaching a limit, there is increasing pressure to reduce their installation and operating costs,
making the business case for Vertical Axis Wind Turbines (VAWTs) more interesting again.
However, in order to properly assess the economic potential of large-scale VAWTs, improved
design tools are required. As the research budget for VAWT projects is generally too small to
carry out expensive large-scale wind tunnel tests and field measurements, alternative methods
are required. In this work, the application of small-scale, lower budget methods for improving
VAWT design tools is assessed. It is shown that (a) currently available tools for VAWT design
have not been sufficiently validated, (b) lower budget, small-scale wind tunnel tests can be
effective for examining VAWT performance in terms of average power coefficient vs. tip speed
ratio as well as the dynamics of the torque, and (c) Detached Eddy Simulations on small-scale
VAWTs can be effective for examining their performance in terms of average power coefficient
vs. tip speed ratio as well as the forces on the blades. Based on these results, a process
for the transfer of the results of lower budget, small-scale measurements and simulations into
recommendations for improved design tools is being developed.

1. Introduction
Vertical Axis Wind Turbine (VAWT) technology has not yet managed to establish itself as
a reliable technology for wind energy projects. This is mainly because VAWTs have several
disadvantages over Horizontal Axis Wind Turbines (HAWTs), including a lower efficiency, higher
bearing loads and lower average height above the ground. However, advantages include the
positioning of the drivetrain at the base, leading to reduced installation and service costs, and
the independency of operation to wind direction, resulting in less fatigue. As wind turbines are
getting larger and cost reductions through up-scaling are reaching a limit, there is increasing
pressure to reduce their installation and operating costs. This is making the business case for
Vertical Axis Wind Turbines (VAWTs) more interesting again.
This increased interest is reflected in the recent research activities aimed at better
understanding the dynamics and aerodynamics of VAWTs - for example from the DeepWind
project ([1], [2] and [3]) and from Sandia National Laboratories ([4], [5] and [6]) as well as the
development of new offshore floating concepts such as [7], [8] and [9].
However, in order to properly assess the economic potential of large-scale VAWTs, improved
design tools are required. Usually, design tools are improved by comparing them to costly wind

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
Global Wind Summit 2018 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1102 (2018)
1234567890 ‘’“” 012002 doi:10.1088/1742-6596/1102/1/012002

tunnel and full-scale field tests in which the shaft torque and blade loads are measured at a
range of flow conditions, in order to better understand the physics. As the research budget for
VAWT development projects is generally small and because very few large-scale VAWTs are
actually in operation, it is currently not feasible to carry out such measurement campaigns.
In this work, the application of small-scale, lower budget methods for improving VAWT
design tools is assessed. This is done by firstly reviewing the currently available design tools,
then carrying out small-scale wind tunnel measurements and Detached Eddy Simulations (DES)
and comparing the results to simulations using an open-source design tool, QBlade. Finally, the
feasibility of these methods for improving design tools is assessed.

2. Review of currently available tools

Tools such as DNV GL’s Bladed [10], DTU’s HAWC2 [11] and NREL’s Aerodyn [12] are typically
used in the wind energy industry for HAWT design. These tools consist of various different
modules in which, for example, the aerodynamics, structural dynamics and control systems can
be modelled. The aerodynamics are typically modelled with a simple Blade Element Momentum
code [13] including corrections for blade tip and root vortex losses [13], for high tip speed ratios
[14] and for dynamic stall [15] as well as additions for dynamic wake modelling [16]. The models
have been shown to be accurate enough for design purposes and are well-established in the
industry.
However, far less research and development work has been done on VAWT design. As the
aerodynamic behaviour of VAWTs is significantly different to HAWTs, these codes cannot be
directly used for VAWTs. Some of the first in-depth studies on the aerodynamics of VAWTs
were focused on understanding the dynamic stall effects and vortex shedding in the wakes ([17],
[18], [19], [20]). More recent studies have focused on the aerodynamics and structural dynamics
of large-scale floating VAWTs by developing and applying design tools ([3], [6], [21] and [22]),
CFD ([23], [24], [25], [26] and [27]) and wind tunnel tests ([28] and [1]).
The validation status in terms of power coefficient (CP ) or average shaft torque of some
aerodynamic codes is summarised in Table 1 on the next page. It can be seen that most of the
tools have not been properly validated in terms of their shaft torque production (or CP ). It is
concluded that more VAWT validation measurements are required in the community order to
improve the design tools.

3. Assessment of lower budget methods for design tool improvement

3.1. Small-scale wind tunnel tests
Small-scale wind tunnel tests were carried out in an open-section wind tunnel consisting of a
fan connected to a 3 m long pipe of diameter 1.4 m with a honeycomb grid to smooth the flow.
A three-bladed H-rotor VAWT model with a diameter of 0.7 m and a height of 0.5 m was built
out of an aluminium frame and 3D-printed NACA 0024 blades with a chord length of 0.07 m
as shown in Figure 1. An in-house Labview programme was developed in order to hold the
rotational speed of the wind turbine constant at values ranging from 50 to 600 rpm using a
Maxon RE 50 motor with a 4.3:1 gearbox and ESCON 50/5 servo controller.
In order to check the quality of the wind inflow conditions, the wind tunnel was first calibrated
at the centre of the inflow plane with a PeakTech 5060 anemometer mounted on a positioning
frame. Stable conditions were found for wind speeds up to 7 m/s, giving a maximum Reynolds
number of 130,000 at the available rotational speeds and the measured air conditions of 14.4 o C
and 96 kPa. The wind field of the inflow plane was then measured for a range of wind speeds
with the anemometer positioned at 35 locations in a grid of 7 × 5 points as shown in Figure 2
for 2.85 m/s at the centre (the rotor area is shown by the black rectangle). This results in an
average wind speed across the rotor plane of 25% more than the wind speed in the centre of the
channel.

2
Global Wind Summit 2018 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1102 (2018)
1234567890 ‘’“” 012002 doi:10.1088/1742-6596/1102/1/012002

Table 1. Validation status of main VAWT aerodynamic tools.

Tool name Institute Model Validation status

Double multiple Shown to be of

- TU Delft
streamtube model [18] poor quality
Modified linear No published comparison
HAWC2 DTU
actuator cylinder [3] to CP measurements
3D vortex code with
Over-prediction of CP
CACTUS Sandia lifting line
above max. CP
approximation [6]
Mediocre performance
Lifting-line free-vortex
QBlade TU Berlin of CP compared to
wake code [29]
RANS CFD
2D unsteady multibody No published comparison
U2DIVA TU Delft
free-wake panel code [28] to CP measurements
Poor performance of CP
University of Free-vortex
ARDEMA2D compared to wind tunnel
Uppsala / Areva particle wake [30]
tests
Average blade forces
3D vortex
PHARWEN Nenuphar compare well to
panel [8]
field measurements
Blade forces and wake
University of Vorticity transport profiles compare
-
Glasgow model [21] fairly well to
wind tunnel experiments

Figure 1. Wind tunnel set-up.

Figure 2. Measured inflow wind
conditions at 2.85 m/s in the centre.

3
Global Wind Summit 2018 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1102 (2018)
1234567890 ‘’“” 012002 doi:10.1088/1742-6596/1102/1/012002

The key design parameter of a wind turbine is the shaft torque, allowing the rotor power
coefficient (CP ) to be calculated as shown in Equation 1, where τ = rotor torque (Nm), ω =
rotational speed (rad/s), ρ = air density (kg/m3 ), V0 = inflow wind speed (m/s) and A = rotor
cross-sectional area (m2 ).
τω
CP = 1 3
(1)
2 ρV0 A
A Lorenz Messtechnik in-line torque sensor type DR-2112R was mounted between the motor
and the rotor and measurements were carried out at a frequency of 2 kHz for at least 30 seconds
per measurement point. Blade loads are also key to assessing design tools; however, they were
not measured due to the expense and complexity of installing strain gauges in the rotating
system.
The results in terms of power coefficient vs. tip speed ratio are shown in Figure 3 for wind
speeds of 6.0 m/s, 6.5 m/s and 7.0 m/s, where the tip speed ratio (λ) is given by Equation
2 and R = rotor radius (m). They are compared to measurements on a similar-sized VAWT
model from the University of New Hampshire [6], showing that the approximate magnitude is
reasonable (the dimensions, aspect ratio, solidity and airfoil types are different). This shows that
the chosen set-up could be used successfully to measure the full range of the power coefficient
vs. tip speed ratio curve of the VAWT model.

Figure 3. Wind tunnel measurement results.

Rω
λ= (2)
V0
Examination of the 2 kHz torque signal allows the periodical behaviour of the rotor to be
analysed. Figure 4 shows the frequency content of three typical measurements carried out for
60 seconds each at tip speed ratios of 0.82, 1.65 and 2.47 (200, 400 and 600 rpm), obtained by
carrying out Fast Fourier Transforms on the measurement data. As expected, clear peaks can
be seen at the rotational speed (1P ), at the blade passing frequency (three times the rotational
speed = 3P ), at twice the blade passing frequency (6P ) and at three times the blade passing
frequency (9P ). Large peaks at 2P can also be seen for every rotational speed, indicating a rotor

4
Global Wind Summit 2018 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1102 (2018)
1234567890 ‘’“” 012002 doi:10.1088/1742-6596/1102/1/012002

imbalance from an error in manufacture or pitch setting of one of the blades. Additionally, large
peaks at 4.3P can be seen. These are thought to be caused by the dynamics of the motor, which
turns 4.3 times faster than the rotor. A small peak can be seen on each plot at a frequency
of approximately 23 Hz, circled on the figure, indicating a dynamic that is independent of the
rotor speed. This could be the first torsional natural frequency of the structure.

Figure 4. Frequency content of a three typical torque signals at tip speed ratios of 0.82, 1.65
and 2.47 (200, 400 and 600 rpm).

As the frequency content at 4.3P is due to the motor and not to the wind turbine rotor, it
needs to be removed in order to examine the dynamic behaviour of the rotor. On the left-hand
graph of Figure 5, a typical torque signal at a tip speed ratio = 2 is shown with a low-pass
filter of 100 Hz applied. The periodic nature of the torque can be seen; however, the 4.3P
content is dominating. Applying a 30 Hz low pass filter removes the 4.3P frequency, but also
the interesting 6P and 9P frequencies (middle graph). Here only the blade passing frequency
can be seen, as well as the asymmetry due to the misalignment of one blade as discussed above.
The 4.3P peak was therefore removed manually from the Fourier transformed-data and then an
Inverse Fast Fourier Transform was performed on the resulting dataset, giving the graph shown
on the right-hand side of Figure 5.
This shows a distinctive periodical behaviour; the torque does not smoothly transition from
the maximum to the minimum value, but it increases again slightly in the middle. This is
thought to be caused by the interaction of the blades with the wake, and is examined further in
the next section.

Figure 5. Typical torque signal at tip speed ratio = 2 (rotational speed = 435 rpm = 7.25 Hz).

In summary, it has been found that lower budget small-scale wind tunnel tests can be used to
compare the performance in terms of power coefficient vs. tip speed ratio as well as to investigate

5
Global Wind Summit 2018 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1102 (2018)
1234567890 ‘’“” 012002 doi:10.1088/1742-6596/1102/1/012002

the natural frequencies and the dynamic behaviour of the total torque on the shaft within a rotor
revolution. This has the potential for helping to understand the physics of VAWTs and thus
improving design tools.

3.2. Detached Eddy Simulations on small-scale VAWTs

Computational Fluid Dynamics (CFD) can be used in order to gain detailed information of
turbulent flows around objects. In this case, its application can potentially help to better
understand the behaviour of VAWTs and therefore to improve design tools, either in combination
with or instead of small-scale wind tunnel tests.
The basis of conventional CFD simulations are the Navier-Stokes equations, which define
single-phase fluid flows and can only be solved for turbulent flows using turbulence modelling,
as the resolution required to resolve all scales involved in turbulence is beyond what is
computationally possible. The most common - and least computationally expensive - approach
to turbulence modelling is to derive the Reynolds-Averaged Navier Stokes (RANS) equations
from the instantaneous Navier-Stokes equations via a Reynolds decomposition. A more accurate
- and more computationally expensive - approach is Large Eddy Simulations (LES), involving
removal of the smallest scales of the through a filtering operation and modelling of their effect
using subgrid scale models. This allows the largest and most important scales of the turbulence
to be resolved, while greatly reducing the computational cost incurred by the smallest scales. A
compromise between the two methods is Detached Eddy Simulations (DES), which uses RANS
to resolve boundary layers and LES to resolve the larger eddies in the bulk flow. Therefore, the
grid resolution for DES is not as demanding as pure LES, considerably reducing the cost of the
computation.
As this work focuses on the application of small-scale, lower budget methods for improving
VAWT design tools, it was decided to carry out Detached Eddy Simulations (DES) on the
VAWT model in the small-scale wind tunnel. This was found to be the best compromise between
computational expense and model accuracy, as the small scale of the simulations reduces the
computational costs but retains the better accuracy of DES compared to RANS.
First of all, Detached Eddy Simulations were carried out using CFX ANSYS version 19.0
on one blade of the wind turbine for a Reynolds number of 88,000 as shown in Figure 6(a).
The results in terms of lift coefficient (cL ) vs. angle of attack (α) are compared to the results
of other turbulence models (SST k-omega, SST k-omega coarse, k-epsilon and Scale-Adaptive
Simulation [31]) as well as to XFOIL simulations [32] with different values of Ncrit in Figure
6(b). Ncrit is the log of the amplification factor of the most-amplified frequency which triggers
transition, and should be set depending on the ambient disturbance level in which the airfoil
operates (= 4-8 for a dirty wind tunnel, 10-12 for a clean wind tunnel). First of all, it can be
seen that the XFOIL results are highly dependent on the value of Ncrit. Also, it can be seen
that DES is capable of reasonably predicting the flow over the blade.
Next, simulations were carried out on one half of the wind turbine and outlet pipe of the
wind tunnel as shown in Figure 7. Two different meshes were used for the simulations, having
15 million and 44 million cells. The mesh around the structure and the blades was refined
using inflation layers to reach a y+ of less than 30. Incompressible transient simulations were
performed using the Detached Eddy Model for turbulence. The wall function is used to model
the boundary layer very close to the blades. The time steps ranged from 0.5 to 1.0 ms, depending
on the rotational speed of the turbine. The simulation results were checked for mesh and time-
step independence. The simulations were run at the computing cluster at the University of
Applied Sciences Rapperswil, using 80-120 cores per simulation. The input wind speed was set
to a constant value of 6 m/s.

6
Global Wind Summit 2018 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1102 (2018)
1234567890 ‘’“” 012002 doi:10.1088/1742-6596/1102/1/012002

Figure 6. (a) Set-up of the Detached Eddy Simulations for one blade; (b) Results in terms of
lift coefficient vs. angle of attack.

Figure 7. Set-up of the Detached Eddy Simulations.

The results in terms of power coefficient vs. tip speed ratio are shown in in Figure 8 compared
to the wind tunnel measurements. At and above the maximum CP , the results match fairly well.
At lower tip speed ratios, the simulations match less well, due to the increasing average angle
of attack (leading to more flow separation, which is difficult to simulate) and the decreasing
Reynolds Number (reducing the likelihood of a turbulent flow, which is easier to simulate).
Further differences are thought to be caused by the non-uniform wind input field in the wind
tunnel, which is not accounted for in the simulations.
In Figure 9, the variation of torque with time for a tip speed ratio = 2 is examined. In
part (a), the torque exerted on the shaft by one blade is shown. One large peak is visible per
revolution, when the blade passes the front of its swept area and is operating at its optimal angle
of attack and producing lift. A second, smaller peak is seen as the blade passes the back of its
swept area, producing a small amount of lift. The torque even becomes negative at two azimuth
locations, because here the drag dominates. The torque produced by the other two blades shows
a similar pattern, offset by one-third of a revolution. The total torque, shown in part (b), is

7
Global Wind Summit 2018 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1102 (2018)
1234567890 ‘’“” 012002 doi:10.1088/1742-6596/1102/1/012002

then given by the sum of these three torques at each azimuth position. Like the measurements
shown in the previous section, this signal does not smoothly transition from the maximum to the
minimum value, but it increases again slightly in the middle. This occurs due to the presence
of the second small peak as each blade passes behind the tower. Additionally, it can be seen
in part (b) that the magnitude of the torque fluctuations matches well with the measurements
(both approximately 0.15 Nm), confirming the ability of the simulations to predict the physics
correctly. Part (b) also shows a comparison between the positive torque produced by the blades
and the resistance caused by the structure of the wind turbine, comprising approximately 10%
of the total torque.

Figure 8. Comparison of measurements to DES.

Figure 9. Variation of torque with time from DES: (a) For one blade; (b) For all blades (solid
line) and for the structure (dashed line).

In summary, it has been found that lower budget Detached Eddy Simulations on small-scale
VAWTs can be used to compare the performance in terms of power coefficient vs. tip speed

8
Global Wind Summit 2018 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1102 (2018)
1234567890 ‘’“” 012002 doi:10.1088/1742-6596/1102/1/012002

ratio as well as the dynamic behaviour of the wind turbine. This has the potential for helping
to understand the physics of VAWTs and thus improving design tools. Further analysis of these
results is on-going.

3.3. Open-source design tool

The design tool that is applied in this work is the open-source software QBlade [29] from TU
Berlin, which is capable of Double Multiple Streamtube Model (DMSM) and Lifting Line Free
Vortex Wake (LLFVW) simulations. The DMSM [33] applies the actuator disc and blade element
theory to VAWTs by splitting up the stream tube flowing through the VAWT into a set of smaller
stream tubes that are passed through by each blade twice per revolution. The LLFVW method
[34] models the blades with a single line of vortices located on the quarter chord of the blade
and updates the position of the wake end nodes based on the local velocity.
Simulations were carried out in the DMSM and LLFVW mode using the geometry and air
conditions of the small-scale wind tunnel. The lift and drag coefficients were obtained from the
results of the one-bladed DES from the previous section and extrapolated using the Montgomery
method in QBlade [29]. A constant input wind speed was used. For the DMSM simulations,
the blade was discretised into 40 elements and a maximum epsilon for convergence of 0.001 was
used. Tip losses and variable induction factors were neglected. For the LLFVW simulations, 10
rotor revolutions were simulated with 720 time steps. Tower effects and drag of the structure
were neglected.
The average power coefficient vs. tip speed ratio results from the QBlade simulations are
shown in comparison to the small-scale measurements and DES in Figure 10(a). It can be seen
that the DMSM and the LLFVW simulations overestimate the power coefficient compared to
the measurements and the DES at tip speed ratios above the peak power coefficient. This is
expected from previous work showing that DMSM generally overestimates at high tip speed
ratios due to the fact that backflow effects are not taken into account [33] and that LLFVW
generally overestimates the torque [29]. In Figure 10(b), the familiar variation of torque with
time can be seen as discussed in the previous sections.

Figure 10. (a) Comparison of results from measurements, DES and QBlade in terms of power
coefficient vs. tip speed ratio; (b) Power coefficient vs. time for the total torque from the QBlade
LLFVW simulations for a tip speed ratio = 2.

3.4. Summary of lower budget methods

VAWT design codes were found to not match well with simulations and wind tunnel tests, both
in the literature and in the present work. However, lower budget, small-scale wind tunnel tests

9
Global Wind Summit 2018 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1102 (2018)
1234567890 ‘’“” 012002 doi:10.1088/1742-6596/1102/1/012002

and Detached Eddy Simulations were successfully used to examine the average torque as well
as its variation with time in order to examine the physics of VAWTs. This has the potential
for improving VAWT design tools and ultimately improving VAWT design and reducing wind
farm operating costs. Based on these results, a process for the transfer of the results of lower
budget, small-scale measurements and simulations into recommendations for improved design
tools is currently being developed.

4. Conclusions
This work has shown the following:
• Currently available aerodynamic tools for VAWT design have not been sufficiently validated
in terms of the shaft torque and blade loads, the most important design parameters.
• Lower budget, small-scale wind tunnel tests can be effectively used for examining VAWT
performance in terms of average power coefficient vs. tip speed ratio as well as the dynamics
of the torque.
• Lower budget Detached Eddy Simulations on small-scale VAWTs can be effectively used
for examining their performance in terms of average power coefficient vs. tip speed ratio as
well as the forces on the blades.
• A process for the transfer of the results of lower budget, small-scale measurements and
simulations into recommendations for improved design tools is currently being developed.

Acknowledgments
Thanks to Lukas Schnider, Simon Boller and Roland Peterer for help with the wind tunnel
measurements.

References
[1] Najafi N and Paulsen U S 2017 Energy 405–416
[2] Galinos C, Larsen T J, Madsen H A and Paulsen U S 2016 Energy Procedia 94 319–328
[3] Madsen H A, Larsen T J, Paulsen U S and Vita L 2013 Proceedings of 51st AIAA Aerospace Sciences Meeting
including the New Horizons Forum and Aerospace Exposition (Chapter AIAA 2013-0913) (AIAA)
[4] Owens B C and Griffiths D T 2014 Journal of Physics: Conferrence Series vol 012092 p 524
[5] Owens B C and Griffith T D 2014 32nd ASME Wind Energy Symposium, AIAA SciTech Forum vol 0518
(AIAA)
[6] Owens B C, Hurtado J E, Paquette J A, Griffith D T and Barone M 2013 54th
AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference vol
1552 (AIAA)
[7] Borg M, Shires A and Collu M 2013 Renewable and Sustainable Energy Reviews 39 1214–1225
[8] Dupont M, Pitance D, Kluczewska-Bordier J, Immas A and Silvert F 2017 WindEurope Conference and
Exhibition (Poster P.O. 262)
[9] Pedersen T F, Paulsen U S, Madesn H A, Nielsen P H, Enevoldsen K, Tesauro A, Kragh K A, Vita L, Ritchie
E, Leban K, Wedell-Heinen J and Larsen K H 2013 European Wind Energy Conference & Exhibition 2013
(European Wind Energy Association (EWEA))
[10] 2018 Dnv gl bladed URL https://www.dnvgl.com/energy/generation/software/bladed/index.html
[11] 2018 Hawc2 URL www.hawc.de
[12] 2018 Nwtc information portal (aerodyn) URL https://nwtc.nrel.gov/AeroDyn
[13] Glauert H 1935 Airplane propellers vol 4 (Springer)
[14] Glauert H 1926 A General Theory of the Autogyro (ARCR R&M No. 1111)
[15] Leishman J G and Beddoes T S 1989 Journal of the American Helicopter Society 34(3) 3–17
[16] Peters D A and He C J 1992 Journal of American Helicopter Society
[17] Ferreira C J S, van Zuijlen A, Bijl H, van Bussel G and van Kuik G 2010 Wind Energy 13 1–17
[18] Ferreira C S, van Kuik G, van Bussel G and Scarano F 2009 Experiments in Fluids 46 97–108
[19] Frounie P, Beguier C, Paraschivoiu I and Brochier G 1986 Journal of Propulsion and Power 5 445–449
[20] Almohammadi K M, Ingham D B, Ma L and Pourkashanian M 2015 Journal of Fluids and Structures 57
144–158

10
Global Wind Summit 2018 IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1102 (2018)
1234567890 ‘’“” 012002 doi:10.1088/1742-6596/1102/1/012002

[21] Scheurich F, Fletcher T M and Brown R E 2011 Wind Energy 14 159–177

[22] Mendoza V, Bachant P, Wosnik M and Goude A 2016 Journal of Physics: Conference Series (IOP Publishing)
p 753
[23] Mendoza V and Goude A 2017 IOP Conference Series: Journal of Physics vol 012031 (IOP Publishing)
[24] Shamsoddin S and Port-Agel F 2014 Energies 7 890–912
[25] Abkar M and Dabiri J O 2017 Journal of Turbulence
[26] Hezaveh S H, Bou-Zeid E, Lohry M W and Martinelli L 2017 Wind Energy 20 713–730
[27] Bangga G, Lutz T, Dessoky A and Krmer E 2017 Journal of Renewable and Sustainable Energy 9(5)
[28] Tescione G, Ragni D, Ferreira C J S and van Bussel G J W 2014 Renewable Energy 70 47–61
[29] Marten D 2016 Qblade v0.95 - guidelines for lifting line free vortex wake simulations
[30] Eriksson S, Kjellin J and Bernhoff H 2013 Energy Science Engineering 1 135–143
[31] Menter F and Egorov Y 2005 AIAA Paper 2005-1095
[32] Drela M 2000 Xfoil URL http://web.mit.edu/drela/Public/web/xfoil/
[33] Marten D and Wendler J 2013 Qblade guidelines v0.6
[34] Marten D, Pechlivanoglou G, Nayeri C and Paschereit C O 2016 International Symposium on Transport
Phenomena and Dynamics of Rotating Machinery

11