181

A publication of

CHEMICAL ENGINEERING TRANSACTIONS

VOL. 103, 2023 The Italian Association
of Chemical Engineering
Online at www.cetjournal.it

Guest Editors: Petar S. Varbanov, Panos Seferlis, Yee Van Fan, Athanasios Papadopoulos
Copyright © 2023, AIDIC Servizi S.r.l.
DOI: 10.3303/CET23103031
ISBN 979-12-81206-02-1; ISSN 2283-9216

The Performance of a Magnus Vertical Axis Wind Turbine

in Typhoon Wind Speeds
Harriet Elaine Limpot, Alyssa Somido, Angela Shayne Yamsuan, Binoe Abuan,
Louis Angelo Danao*
University of the Philippines Diliman, Quezon City, Philippines 1101
[email protected]

The study involved determining the performance of a Magnus-type Vertical Axis Wind Turbine (VAWT)
subjected to typhoon wind speeds. The investigation is purely numerical, starting with the validation of the
Magnus effect on lift and drag generation of a rotating cylinder in similar flow conditions as the wind turbine
experiences. Validation data was culled from a related experimental study on Flettner rotors as used in ship
propulsion augmentation applications. The validated aerodynamic performance of the cylinders is assumed to
be sufficient to model the wind turbine performance. The scale of the 2-cylinder model rotor is equal to the actual
Magnus VAWT of Challenergy Inc. The published rated capacity of the Magnus VAWT is 10 kW while the
simulated rotor running at the same rated conditions produced 9.42 kW at a Tip Speed Ratio (TSR) of 1.17
resulting in a Power Coefficient (CP) equal to 0.5. Simulating the rotor at a wider range of TSRs reveals a peak
performance point of CP = 0.58 at a tip speed ratio of 2. This does not take into account the input power
consumed by rotating the cylinders and hence is expected to be significantly lower in the actual conditions. A 2-
dimensional model of a 2-cylinder Magnus-type rotor was created with a cylindrical blade diameter of 1 m and
a rotor radius of 3 m. Rated conditions are at 8 m/s wind speed with a rotor rotational speed of 30 rpm and
cylinder rotational speed of 150 rpm. Experimental validation data was adopted from a wind tunnel test of a
Flettner rotor under flow conditions of Reynolds number equal to 1 × 106. Lift and drag curves were compared
between simulation and experiments, revealing an excellent agreement in lift while a significant disagreement
in drag. Simulations underpredict the drag, which resulted in the unusually high predicted performance of the
wind turbine. The VAWT performance agreed well with the published rated capacity of 10 kW with a predicted
power output of 9.42 kW. A tip speed ratio sweep was conducted and revealed a peak performance point at
TSR = 2 with CP equal to 0.58, subsequently dropping to CP = 0.35 at TSR = 5. When the rotor was simulated
at typhoon wind speeds, the results showed increasing power outputs from 30.87 kW at 13.89 m/s, 99.67 kW
at 20.83 m/s, 140.38 kW at 28.61 m/s, up to 176.79 kW at a cut-out wind speed of 40 m/s. The corresponding
CPs are observed to decrease as wind speeds increase from 0.5, 0.31, 0.30, 0.16 down to 0.8.

1. Introduction
As the world population continues to rise, so does the demand for power. In the next decades, more power-
generating devices are expected to be built to convert renewable energy into usable electrical energy. This will
reduce the world's dependence on fossil fuels to heat homes, run vehicles, power industries, and provide
communities with electricity. One of the said devices is the wind turbine which converts wind into electrical
energy (Elliot, 2000). Elliot (2000) published a report containing the results of a wind resource analysis and
mapping study for the Philippine archipelago. He discovered that the country's wind resource relies heavily on
latitude, elevation, and proximity to the coastline — claiming that areas in Batanes, Ilocos Norte, and Samar
have great potential for wind farms. Most of the wind farms already operating in the country have horizontal
wind turbines installed for greater power generation.
The Philippines is frequently subjected to strong winds that result in heavy rains and monsoons, as well as
storms that evolve into typhoons. In 2013, the Philippines was cited by Time Magazine to have been the country
most exposed to typhoons, following being hit by Typhoon Haiyan, which had some of the highest recorded

Paper Received: 10 April 2023; Revised: 09 June 2023; Accepted: 21 July 2023
Please cite this article as: Limpot H.E., Somido A., Yamsuan A.S., Abuan B.E., Danao L.A.M., 2023, The Performance of a Magnus Vertical
Axis Wind Turbine in Typhoon Wind Speeds, Chemical Engineering Transactions, 103, 181-186 DOI:10.3303/CET23103031
182

wind speeds of a tropical cyclone in the modern era. The Philippines has actually borne the brunt of some of
the most powerful tropical cyclones to landfall in this century, with the most powerful being Goni from 2020
which hit Catanduanes. In 2021, the Philippines’ first VAWT was installed on the island of Batanes; it is expected
to withstand and take advantage of the strong winds in the area. It has been shown that VAWTS are more
effective than HAWTs during harsher weather conditions where there is more turbulence. This is because
direction fluctuations have minimal effect on VAWTs’ power generation, and only velocity fluctuations will need
to be accounted for (Kooiman, Tullis, 2010). It also produces lower sound at a lower tip speed ratio (Danao,
2013). For these reasons, VAWTs have increasingly grown popular as a power-generating device over recent
years, especially in areas that regularly experience highly transient weather conditions.
Only a few studies have been conducted to assess the power output of wind turbines during inclement or
extreme weather. This is because the simulation of extreme weather is difficult given the unpredictable
conditions of wind and other factors involved, thus requiring advanced technologies (Lin et al., 2012). With
respect to VAWTs, its flow physics is also quite complex, making predictions difficult to achieve with
mathematical or computational models alone (Edwards et al., 2012). There are especially lesser studies on
Magnus VAWTs since the concept of Magnus VAWTs has only recently grown popular. The most recent was
an unvalidated numerical study published last 2015 by Khadir and Mrad (2015). Magnus VAWT generates
power by utilizing the lift force brought about by the rotation of its cylinders. This lift force acts vertically relative
to the wind direction. Furthermore, the speed of the cylinders can be controlled so it can operate even in typhoon
wind speeds. The objective of this study is to improve knowledge of Magnus Vertical Axis Wind Turbines. This
study also aims to determine how extreme weather conditions, particularly increased wind speeds, affect the
performance of a Magnus VAWT. This study specifically investigated the effect of typhoon wind speeds on the
performance of the Batanes Magnus VAWT in terms of power generation. As such, the last objective of this
study was to determine the potential of the Batanes Magnus VAWT for power generation during extreme
weather conditions. The Batanes Magnus VAWT is the sole model of a Magnus VAWT that was studied. This
limits the available information from what the manufacturer of the Batanes Magnus VAWT (Challenergy Inc.)
can provide. The researchers were also limited to conducting the study through simulations, given the lack of
access to wind tunnel testing. The researchers had no option but to utilize experiments on single cylinders as
reference for validation of the parameters that will be used for the actual simulations.

2. Methodology
Due to limited studies on performance of Magnus VAWTs, the validation was based on experimental data found
for rotating cylinders. The experiment chosen was conducted at a high Reynolds number since actual operating
conditions of the Batanes Magnus VAWT (Figure 1a) occur in such. Specifically, the calculated Reynolds
number for the rated operating conditions of the Batanes Magnus VAWT was at 6.57 × 105 given the rated wind
speed of 8 m/s and the cylinder diameter of the blades at 1 m. A study conducted by Bordogna et al. (2019),
which tested the effect of varied Reynolds number values on the performance of rotating cylinders through a
series of wind-tunnel experiments, was used to validate the CFD model of the Batanes Magnus VAWT. Each
series of wind-tunnel experiments conducted in the study was performed under different Reynolds numbers with
the velocity ratio being varied among tests with the same Re. The experimental tests conducted at Re = 10.01
× 105 were chosen as it was closest to the calculated Reynolds number of the Batanes Magnus VAWT under
rated operating conditions. As such, four simulations were conducted corresponding to four different velocity
ratios under the same Re. The lift coefficient Cl vs velocity ratio k and drag coefficient Cd vs velocity ratio k plots
derived from the simulations were compared to the results obtained by the experiment to determine the accuracy
of the CFD parameters used in the simulations. The turbulence model k-ω shear stress transport (SST) was
adopted for this study as it is said to be the most accurate model for predicting the dynamic behavior of airfoil
with a moving fluid (Danao et al., 2013). It is a combination of the k-ω model at the inner boundary layer and
the k-ε model on the outside of the boundary layer. This model was employed in numerous studies involving
wind turbines, including the investigation of the aerodynamic performance of H-type Darrieus VAWT by Belabes
and Paraschivoiu (2021). Ozturk (2020) also adopted the same model to study the flow past a rotating cylinder
with Reynolds Numbers of 7.33 × 105, 1.01 × 106 and 1.26×106, which are cylindrical diameter based. For k- ω
SST models, the acceptable wall unit y+ for coarse mesh is around 30 to 300 and for finer mesh, less than 5.
This was considered during the verification of the accuracy of the model such that results of simulations that
yielded a y+ value greater than 5 were rejected. One round of simulation ran 20 full rotations of the cylinder,
with the last 3 rotations considered in determining the accuracy of the results. This was to make sure the results
had already converged. As such, the simulation was time-dependent, and the time step was specified for each
round of simulation. In the simulations, each time step was made to correspond to a 5-degree angular
displacement of the rotating cylinder. In terms of convergence, there were two convergence targets per round
of simulation. The residual convergence criteria guaranteed minimal error in the results, while the periodic
 183

convergence criteria guaranteed that the values of torque were already consistent in magnitude despite
fluctuations due to the changing position of the cylinder. Particularly, the convergence criteria were set to a
tolerance of 10 ×10-6.
The computational domain of the simulation played a vital role in ensuring the accuracy of the results such that
an incorrect replication of the experimental set-up implemented in real life would have led to inaccurate results
in the simulation. The computational domain consisted of the rotating infinite cylinder represented by a circle,
and the fluid domain, which is the region that confined the cylinder. An O-type mesh was used in a study by
Padrino and Joseph (2006), as it has been proven to be advantageous for rotating surfaces. As such, an O-type
mesh was employed in the meshing. The shape and size of the fluid domain must provide sufficient clearance
between the geometry of interest and the walls of the fluid domain to avoid blockage effects that may
compromise the quality of the results. In a wind tunnel experiment of Mcintosh (2009), the blockage effects of
the tunnel walls being too close to the turbine led to higher flow velocity passing through the turbine. This
resulted in a power coefficient higher than the actual, where air flow is unconstrained. As such, the experimental
results had to be corrected to eliminate the discrepancy caused by blockage effects. To avoid this scenario, the
dimensions of the fluid domain were initially set as the cylinder being concentric to the circular fluid domain with
a radius twenty times the diameter of the cylinder. These values were increased and decreased depending on
how fully the domain captured the wake and whether the boundary conditions altered the flow passing through
the cylinder in the domain. The implemented node density in the simulation can also affect the quality of the
results. Determination of the appropriate node density was thus performed by starting at 100 nodes and
progressively increasing the density by a factor of 2 twice over. For the boundary conditions, the side from which
fluid entered the domain was set as the velocity inlet and the opposite of it was the pressure outlet. A constant
inlet velocity equal to the wind velocity experienced in the experimental conditions was assigned at the inlet,
while the outlet was set at atmospheric pressure. This domain set-up has been widely used in previous CFD
simulations on wind turbines for its convenient set-up and capability to sufficiently capture the behaviour of fluid
around the studied object. The rotation of the cylinder was specified through the moving wall function in the wall
boundary condition. After validation of the CFD set-up used to implement the experiments of Bordogna et al.
(2019), the parameters used from the validation simulations were carried over to the simulation of the Batanes
Magnus VAWT. The simulations for the Batanes Magnus VAWT were divided into two sets: the first set
corresponded to rated operating conditions, while the second set corresponded to typhoon wind conditions. The
accuracy of the model was limited by the available Batanes Magnus VAWT specifications that Challenergy was
able to provide. Particularly, the rectifier plates attached to the cylinders of the VAWT were not reflected in the
CFD model because of its proprietary design that Challenergy, Inc. could not disclose. The first set of simulations
was conducted at the rated wind speed of 8 m/s in different tip speed ratios TSRs, specifically 1.17, 2, 3, 4, 5,
6, and 8. Since wind speed was fixed for all set-ups, the rotational speed of the rotor was made to vary to meet
the specified TSRs. The second set of simulations was conducted at wind speeds observed in typhoon weather
conditions. The Tropical Cyclone Wind Signal System of PAG-ASA updated as of March 2022, was used to
determine the specific wind speeds. Since the cut-out wind speed of the Batanes Magnus VAWT is at 40 m/s,
the wind signals considered in the simulations were only those below 40 m/s. The median of the range of wind
speeds covered in each wind signal was identified and used in the simulations. Specifically, the second set of
simulations subjected the Batanes Magnus VAWT to the following median wind speeds: 13.89 m/s (Wind Signal
1), 20.83 m/s (Wind Signal 2), 28.61 m/s (Wind Signal 3), and 40 m/s (cut-out wind speed). The TSRs for all
these set-ups were varied as it was impossible to maintain TSR = 1.17 from the rated conditions without having
the turbine rotational speed go beyond the maximum specified by Challenergy.
For the computational domain of the Batanes Magnus VAWT, an inner and outer domain was created. The inner
domain contained the turbine, while the other domain represented the environment immediately outside the
turbine. The inner domain was a circle concentric to the turbine’s rotor shaft (not depicted in the CFD model)
with a diameter of 12 m, twice the rotor diameter. The outer domain was a rectangle with dimensions of 60 m x
120 m. The 60 m sides corresponded to the inlet and outlet. The rotor was situated 30 m orthogonal from the
inlet and 30 m orthogonal from the 120 m sides. Figure 1b shows the illustration of the model. For the set-up of
the simulations in Fluent, most of the parameters implemented from the validation set-up were carried over, with
alterations made mostly on boundary conditions. Firstly, mesh motion was implemented on the inner domain by
adding a rotational velocity in the cell zone conditions. Inlet velocity was adjusted according to wind speeds. For
the boundary conditions of the cylinder walls, aside from the rotational velocity of the cylinder at its own axis,
the rotation axis origin of the cylinder was also specified through an expression to make sure the cylinders
revolve around the rotor.
184

(a) Challenergy’s Magnus VAWT in Basco, (b) Illustration of 2D Domain

Batanes (Challenergy Inc., 2021)

Figure 1: The VaWT illustration and the model diagram

3. Results and discussion

Among the different experimental set-ups from the reference study, the experiments carried out at a Reynolds
number of 10 × 105 were used for validation as this was closest to the calculated Reynolds number of the
Batanes Magnus VAWT at 6.57 × 105. Four simulations were conducted with varying velocity ratios specifically
at k = 0.5, 1, 1.5, and 2.0. These were the only velocity ratios that Bordogna et al. (2019) carried out an
experiment on at Re = 10 × 105. The simulation results are shown in Figure 2. Comparison was based on
estimated values from the Bordogna’s graph of sectional lift coefficient vs. velocity ratio since a table showing
the exact values of the plotted points in the graph was not made available in the study. It can be observed that
the trend of Cl over increasing velocity ratio satisfactorily resembled that of the reference. This indicates an
accurate simulation of Bordogna’s experiment. It must be noted that the peak value of Cl for rotating cylinders
is yet to be conclusively defined by existing studies (Badalamenti, 2010). To date, existing studies have only
proposed that Cl increases indefinitely with increasing velocity ratio. The Kutta-Joukowski theorem explains the
continuous increasing trend of Cl over increasing velocity ratio as observed in both Bordogna’s and the
simulation results. It can therefore be said that the result of the simulation satisfactorily agrees with the prevailing
knowledge that Cl increases with velocity ratio. However, no suggestions can be made regarding the behavior
of Cl over higher values of k since the velocity ratios tested were only up to 2. Values for Cd, on the other hand,
were smaller than the experimental values and did not follow the upward trend over increasing velocity ratio that
was exhibited in Bordogna’s experiments. Possible causes of this error could be the limitations of a 2D
simulation, such as 3D factors that cannot be modelled in the simulations, particularly 2D assumptions that are
not present in real-world observations. The setup of the experiment conducted by Bordogna et al. (2019) had
the ceiling and floor of the wind tunnel significantly close to the top and bottom surfaces of the cylinder to
minimize the vertical clearance.

(a) (b)

Figure 2: Force Coefficient plots for experiments vs simulations, a) lift coefficient, b) drag coefficient

3.1 Batanes magnus VAWT simulations

The first set of simulations for Batanes Magnus VAWT was conducted at a constant rated wind speed with
varied TSRs to produce a power curve. The angular velocity for the cylindrical blades, however, was maintained
at the rated value of 150 RPM as the manufacturer did not disclose details regarding functions that vary the
blades’ angular velocity with respect to the wind velocity. The plot of the power coefficient CP shown in Figure
3 considerably resembles the typical performance of VAWTs as described by Edwards et al. (2012). However,
 185

the initial low performance common to that of VAWTs at low TSRs cannot be observed in Figure 4a as the
simulations only started at TSR = 1.17. The broad peak of the power coefficient curve may be attributed to the
high Reynolds number. Another notable observation from the results is that the highest CP is reached at TSR =
2. With the cylinder’s rotational speed and the wind velocity maintained at constant values, TSR = 2 entails the
rotor’s rotational speed to be at 5.33 rad/s. This falls out of the provided maximum rotational speed of the turbine,
which is at 35 rpm or 3.67 rad/s. This suggests that the current rated operating conditions of the Batanes Magnus
VAWT may be below its optimal performing capacity. The calculated power extracted at rated wind speeds was
found to be 941 W per unit length of the blade, and considering the rotor’s 10 m long blade, it is close to the
rated power output of 10 kW from Challenergy. It can also be observed that the power coefficients obtained are
significantly high, with some being remarkably close to the 59.3 % Betz limit. It must be noted that due to a lack
of information pertaining to the operation of the Batanes Magnus VAWT, the calculated power extracted does
not consider the parasitic load of rotating the cylinders. During actual operation, the rotation of the cylinders
would consume power and would thus reduce the CP of the Magnus VAWT.

Figure 3: Power Coefficient curve of Batanes Magnus VAWT at rated wind speed of 8 m/s

3.2 Typhoon wind speeds

The extracted power and the power coefficient per typhoon wind speed are presented Figure 4a and Figure 4b.
It must be noted that the parasitic load of rotating the cylinders were not taken into account in the calculation of
the VAWT’s extracted power. The CP’s presented are higher than actual. It can be observed that the power
coefficient is decreasing with increased speeds and decreased TSRs. This would be consistent with the trend
shown in Figure 3 where CP decreases as the TSR decreases further from TSR = 2 where CP is at maximum.
Although CP is decreasing as the wind speed increases, the extracted power, on the other hand, is increasing.
The increase in extracted power is not very evident in the plot of CP because the available power is also
increasing at a greater rate.

(a) Extracted Power of Batanes Magnus VAWT under (b) Power Coefficient of Batanes Magnus VAWT at
typhoon wind speeds typhoon wind speeds

Figure 4: Power performance of the turbine

It can be said that the increase in extracted power is significant at typhoon wind speeds. At wind signal 1, the
extracted power is roughly three times greater than the extracted power of the turbine at rated wind speed; while
at wind signal 2, the extracted power is ten times greater. These findings confirm the advantage of Magnus
VAWTs in power generation amidst strong wind speeds during typhoons. Although the power coefficient is
evidently decreasing over increased wind speeds, the extracted power is significantly increasing. It must be
noted that the decreasing CP is accompanied by the rotor and the cylinders’ rotational speeds being held at
constant values.
186

4. Conclusions
The study deals with a series of CFD simulations investigating the aerodynamic performance of the Batanes
Magnus VAWT when varying the TSR at the rated wind speed and varying the typhoon wind conditions. Based
on the results discussed in the previous chapter, the following conclusions can be made:
• The optimum power extracted by the Magnus VAWT under the rated wind speed 8m/s is achieved at TSR =
2. For cylinders rotating at 150 rpm subjected to the said rated wind speed, the optimal rotation of the rotor is
5.33 rad/s. However, the current maximum rotational speed of the Batanes Magnus VAWT’s rotor is at 3.665
rad/s. An additional 146 W can be generated if the optimal rotor rotation is implemented during operation.
• The Batanes Magnus VAWT power coefficient is decreased at typhoon wind speeds because of lowered TSRs
at higher wind speeds but constant rotor angular velocity. However, despite lowered CP the power generated
by the Batanes Magnus VAWT is improved. This means that the Batanes Magnus VAWT is indeed
advantageous in generating power amidst typhoon conditions.
• It can be confirmed that the lift coefficient increases with increasing velocity ratio after exceeding a velocity
ratio of 1. Additionally, this study has also found that increasing velocity ratio of a cylinder blade in a Magnus
VAWT improves the performance of the VAWT such that the Cl/Cd ratio is improved at higher velocity ratios.

Acknowledgments
The authors would like to thank the DOST ERDT Program for providing the funds to have this paper presented
in-person in the conference.

References
Badalamenti C., 2010, On the Application of Rotating Cylinders to Micro Air Vehicles, PhD Thesis, City
University London, London, UK.
Belabes B., Paraschivoiu M., 2021, Numerical study of the effect of turbulence intensity on VAWT performance.
Energy, 233, 121139, DOI: 10.1016/j.energy.2021.121139.
Bordogna G., Muggiasca S., Giappino S., Belloli M., Keuning J.A., Huijsmans R.H.M., van ’t Veer A.P., 2019,
Experiments on a flettner rotor at critical and supercritical Reynolds Numbers. Journal of Wind Engineering
and Industrial Aerodynamics, 188, 19–29, DOI: 10.1016/j.jweia.2019.02.006.
Challenergy Inc., 2021, Press Release - Challenergy’s 1st Magnus VAWT in the Philippines Begins Operations.
<challenergy.com/en-news/_october_12020_we_have_updated_the_maximum_recorded_instantaneous-
2/>, accessed 17.02.2023.
Danao L.A., Eboibi O., Howell R., 2013, An experimental investigation into the influence of unsteady wind on
the performance of a vertical axis wind turbine. Applied Energy, 107, 403-411, DOI:
10.1016/j.apenergy.2013.02.012.
Danao L., 2013, A numerical investigation into the performance and aerodynamics of a vertical axis wind turbine.
Philippine Engineering Journal, 32. 1-22.
Edwards J.M., Angelo Danao L., Howell R.J., 2012, Novel Experimental Power Curve Determination and
Computational Methods for the Performance Analysis of Vertical Axis Wind Turbines. Journal of Solar
Energy Engineering, 134(3), DOI: 10.1115/1.4006196.
Elliott D., 2000, Philippines wind energy resource atlas development (No. NREL/CP-500-28903). National
Renewable Energy Lab., Golden, CO (US), <www.nrel.gov/docs/fy01osti/28903.pdf>, accessed 06.07.2022.
Khadir L., Mrad H., 2015. Numerical investigation of aerodynamic performance of darrieus wind turbine based
on the magnus effect. The International Journal of Multiphysics, 9(4), 383-396, DOI: 10.1260/1750-
9548.9.4.383.
Kooiman S.J., Tullis S.W., 2010, Response of a vertical axis wind turbine to time varying wind conditions found
within the urban environment. Wind Engineering, 34(4), 389-401, DOI: 10.1260/0309-524X.34.4.3.
Lin J., Sun Y.Z., Cheng L., Gao W.Z., 2012, Assessment of the power reduction of wind farms under extreme
wind condition by a high-resolution simulation model. Applied Energy, 96, 21–32, DOI:
10.1016/j.apenergy.2011.10.028.
McIntosh S.C., 2009, Wind energy for the built environment, PhD Thesis, University of Cambridge, Cambridge,
UK.
Ozturk D., 2020, Performance of a Magnus effect-based cylindrical roll stabilizer on a full-scale Motor-yacht.
Ocean Engineering, 218, 108247, DOI: 10.1016/j.oceaneng.2020.108247.
Padrino J.C., Joseph D.D., 2006, Numerical study of the steady-state uniform flow past a rotating cylinder.
Journal of fluid mechanics, 557, 191-223, DOI: 10.1017/S0022112006009682.