Design of modified Magnus wind rotors using computational fluid dynamics simulation

and multi-response optimization

Neil Lopez, Brian Mara, Brian Mercado, Luigi Mercado, Miguel Pascual, and Michael Angelo Promentilla

Citation: Journal of Renewable and Sustainable Energy 7, 063135 (2015); doi: 10.1063/1.4939192
View online: http://dx.doi.org/10.1063/1.4939192
View Table of Contents: http://scitation.aip.org/content/aip/journal/jrse/7/6?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in

Optimal design and verification tests of cycloidal vertical axis wind turbine
J. Renewable Sustainable Energy 7, 063116 (2015); 10.1063/1.4936362

Large eddy simulation of wind turbine wake dynamics in the stable boundary layer using the Weather Research
and Forecasting Model
J. Renewable Sustainable Energy 6, 033137 (2014); 10.1063/1.4885111

Optimal rotor design for reducing the partial demagnetization effect and cogging torque in spoke type PM motor
J. Appl. Phys. 105, 07F123 (2009); 10.1063/1.3068541

The torque ripple reduction of a concentrated winding synchronous reluctance motor according to stator and
rotor structure variations using response surface methodology
J. Appl. Phys. 103, 07F133 (2008); 10.1063/1.2839348

Optimum design criteria for a synchronous reluctance motor with concentrated winding using response surface
methodology
J. Appl. Phys. 99, 08R325 (2006); 10.1063/1.2165601

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
121.54.44.91 On: Sat, 02 Jan 2016 09:34:54
 JOURNAL OF RENEWABLE AND SUSTAINABLE ENERGY 7, 063135 (2015)

Design of modified Magnus wind rotors using computational

fluid dynamics simulation and multi-response optimization
Neil Lopez,1,a) Brian Mara,1 Brian Mercado,1 Luigi Mercado,1
Miguel Pascual,1 and Michael Angelo Promentilla2
1
Mechanical Engineering Department, De La Salle University, Manila 1004, Philippines
2
Chemical Engineering Department, De La Salle University, Manila 1004, Philippines
(Received 1 June 2015; accepted 16 October 2015; published online 30 December 2015)

Concerns on climate change and dwindling fossil fuel supply have renewed interest
on alternative ways of harnessing renewable energy. Using rotating cylinders to
generate lift from a fluid stream, a Magnus rotor can produce up to 10 times more
lift compared to an airfoil. However, it is also producing more drag. Recent studies
have demonstrated improvement on the aerodynamic efficiency of a Magnus rotor
through the application of surface modifications such as grooves, bumps, dimples,
and even changing the shape of the cylinder into a frustum. However, it is
unknown which is most desirable among those modifications; moreover, if some
may be combined for even better performance. This present study seeks to fill the
mentioned research gap with the aid of computer simulation tool ANSYS CFX.
Simulation results showed that modifying cylinder shape into a frustum generates
the most lift force. However, it is also increasing the drag on the cylinder.
Interestingly, a helical groove may be employed around the frustum cylinder to
mitigate the increase in drag, making the two modifications a promising
combination. Multiple response surface analysis using desirability function was
used to investigate the sensitivity of the rotor design to the different modifications.
Furthermore, a new perspective is introduced wherein the rotor may be able to
withstand more drag in exchange for more lift. For lift generation purposes, bumps
are not desirable. Finally, the aerodynamic performances of the modified rotors are
compared against other published results by means of a drag polar plot. V C 2015

AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4939192]

I. INTRODUCTION
With growing concerns on fossil-based fuel supply and climate change, interest on energy
efficient technologies has reawakened in the last decade. One promising technology is the
Magnus wind rotor, which basically replaces a conventional wind turbine’s airfoil-type blades
with rotating cylinders. As an air stream passes across a rotating cylinder, flow is accelerated
on the one side and is decelerated on the opposite. The pressure difference created from this
phenomenon results in the lift force of a Magnus wind rotor. Seifert (2012) made a comprehen-
sive review of the Magnus effect in aeronautics. The mechanism started to gain plenty of atten-
tion in the first half of 20th century when Flettner attempted to replace the sails of a sea vessel
with rotating cylinders. Murakami and Ito (2009) patented a wind turbine generator utilizing
rotating cylinders. In Enercon (2010), as cited in Seifert (2012), a cargo ship equipped with
four Magnus rotors was designed, claiming to reduce fuel cost by 30%–40%.
Luo et al. (2011) analyzed the power performance of a Magnus wind rotor using blade
element momentum theory. A typical airfoil-type wind rotor blade will have lift coefficients
(CL) of 1–2, and drag coefficients (CD) of 103–102 at best. On the other hand, a Magnus-
type wind rotor can generate up to 10 times more lift (CL  12), and also 50 times more drag

a)
Author to whom correspondence should be addressed. Electronic mail: [email protected].

1941-7012/2015/7(6)/063135/10/$30.00 7, 063135-1 C 2015 AIP Publishing LLC

V

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
121.54.44.91 On: Sat, 02 Jan 2016 09:34:54
 063135-2 Lopez et al. J. Renewable Sustainable Energy 7, 063135 (2015)

(CD  1), based on previous studies by Reid (1924), Thom and Sengupta (1932), and
Badalamenti and Prince (2008). Although yet to be supported experimentally, some computa-
tional studies by Tokomaru and Dimotakis (1991) have shown evidence that the maximum lift
coefficient of 4p, as determined by Prandtl, may be exceeded by a Magnus rotor. Several stud-
ies have demonstrated that the aerodynamics of a rotating cylinder may be enhanced through
the use of various modifications. Seifert (2012) observed higher lift coefficient from a sand-
roughened cylinder. Butt and Egbers (2013) achieved a significant drag reduction from employ-
ing hexagonal bumps. This affirmed the findings of Srivastav (2012), which compared drag
reduction between a bump (outward) and a dimple (inward). Drag reduction of 20% was also
reported by Huang (2010) through the use of helical grooves within a subcritical range of
Reynolds number. Finally, Sun et al. (2012) experimented on the performance of various cylin-
der shapes, which included the use of a frustum cylinder. However, it is yet to be known which
particular modification is more worthwhile to pursue and if particular combinations would yield
even better results. This study aims to fill the aforementioned gap with the aid of computational
fluid dynamics (CFD) simulation and statistical design of experiment (DOE). Sun et al. (2012)
and Mowry et al. (2009) similarly used computational fluid dynamics modeling to characterize
a wind turbine operating on similar principles with the Magnus effect.
Selecting the best design is a challenge on its own. The lift force (force component perpen-
dicular to wind direction) is the primary quantity contributing to the rotational torque of the power
generator. However, the drag force (force component parallel to wind direction) must be given
ample consideration as well, as it contributes to wear and tear due to structural deformation and
vibration. It also contributes to losses due to friction in the rotor bearing and the flow separation
could affect upstream flow. Since each improvement in lift corresponds to a particular increase in
drag, the question is how much drag should a design be willing to accept for more lift? Towards
the end, this study conducts a sensitivity analysis of the results obtained from multi-response sur-
face optimization via desirability functions. The design of the Magnus rotor is optimized for dif-
ferent scenarios with varying importance given to lift, drag, and lift-to-drag ratio (LDR).

II. METHODS
A. Simulation design
DOE techniques may be applied to simulations as well, in which case it should be called
Design of Simulations (DOS). Implementing DOS enables engineers to develop functional rela-
tionships between variables. The difference between DOE and DOS was first identified by
Sparrow et al. (2005). A D-Optimal custom experiment design with 12 runs and 3 factors with
R
3 discrete numeric levels each was made using JMP software (JMPV, Version 11. SAS Institute
Inc., Cary, NC, 1989–2007). A separate design of simulation was made for a set with bumps
and a set with dimples. The same set was executed at five different speed ratios, making a total
of 60 runs each, and 120 in total. The different factors with their corresponding levels are sum-
marized in Table I. The ratio of the tip and root diameter (diameter ratio) of a frustum cylinder
was varied from 1 to 2 (i.e., regular cylinder has a diameter ratio of 1). The ratio of depth to
diameter (k/c ratio) of bumps and dimples was varied from 0.11481 to 0.20282 and 0.17051 to
0.33898, respectively. The height of helical grooves was varied from 0 to 0.075 m. The
responses recorded were lift, drag, and lift-to-drag ratio. All models were sanded with a rough-
ness height of 2.5 mm and had an endplate with diameter equal to twice the cylinder (de/d ¼ 2).
According to Sedaghat (2014), the rotational torque of the Magnus rotor only varies with the
speed ratio (ratio of cylinder rotational speed to wind speed), allowing the experiments to be
performed with a constant free stream velocity (U1) of 4 m/s.

B. Simulation model development

All experiments were run on a single, fixed cylinder in a wind tunnel setup simulated using
R
ANSYS CFX (ANSYSV Academic Research, Release 14.5). The same software was used by
Sun (2012) to simulate the aerodynamics of different Magnus rotor shapes: frustum, wavy, and

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
121.54.44.91 On: Sat, 02 Jan 2016 09:34:54
 063135-3 Lopez et al. J. Renewable Sustainable Energy 7, 063135 (2015)

TABLE I. Factors considered in the design of simulations with their corresponding levels.

Factor Cylinder shape Bump Helical grooves

Bump set Levels Diameter ratio ¼ 1 k/c ratio ¼ 0.11481 Height ¼ 0 m

Diameter ratio ¼ 1.5 k/c ratio ¼ 0.15113 Height ¼ 0.0375 m
Diameter ratio ¼ 2 k/c ratio ¼ 0.20282 Height ¼ 0.075 m
Dimple set Levels Diameter ratio ¼ 1 k/c ratio ¼ 0.17051 Height ¼ 0 m
Diameter ratio ¼ 1.5 k/c ratio ¼ 0.25748 Height ¼ 0.0375 m
Diameter ratio ¼ 2 k/c ratio ¼ 0.33898 Height ¼ 0.075 m

regular. In this current study, different modifications were combined as illustrated in Fig. 1.
Boundary conditions are illustrated in Fig. 2. The cylinder is fixed, but allowed to rotate about
its own axis. This rotation is needed for the Magnus effect to manifest. The cylinder cannot be
deflected by the torque it is producing. The fluid domain is bounded by four no-slip walls, and
has one inlet and one outlet 2 m and 8 m from the cylinder, respectively. The inlet air velocity
is constant at 4 m/s. The speed ratio is adjusted by varying the rotating speed of the cylinder
about its own axis.
Different turbulence models and meshing strategies were tested for the study. Among the
turbulence models tested, the Eddy-Viscosity Transport Equation (EVTE) had the best accu-
racy, with prediction errors for lift and drag coefficients equivalent to 2.53% and 7.9%,
respectively. According to the ANSYS CFX Theory Guide, the EVTE turbulence mode is a
simple one equation model developed by Menter (1994), and is derived directly from the k-E
model. Like the k-x model, it is solved with a scalable wall function, allowing the model to
take advantage of a fine near-boundary mesh. Further information on turbulence modeling are
available in Wilcox (1993). With regard to meshing, the model adopted a multi-zone strategy—
using fine mesh around the cylinder, and coarser mesh away from the cylinder. Near the cylinder,
an inflation layer was created with a first node distance of approximately 0.0001 m (yþ  1.5). For
low Reynolds number k-x and k-E simulations, a yþ  2 is needed. A fine tetrahedral mesh fol-
lows the inflation layer until the point when the element size is no longer increasing, after which a
coarser, hexahedral mesh is utilized to save on computation time. Using a multi-zone mesh did
not affect the results by more than 1.7%, which was acceptable for the study. Fig. 3 is a screen-
shot of the final mesh design. The final CFD model was pre-validated against the actual experi-
mental results of Bychkov et al. (2007) with a Magnus rotor. Mesh design specifications are pro-
vided in Table II. Each mesh model consisted of 1.5  106 nodes and 4  106 elements.
More details on the turbulence model selection, mesh design development, and model pre-
validation are provided on Mara et al. (2014a; 2014b).

C. Calculation of lift and drag coefficients

Aerodynamic forces acting on the surface of the cylinder were obtained from the simula-
tions. Since the simulations were performed on a fixed cylinder, the relative wind velocity to

FIG. 1. Different cylinder configurations simulated in the study.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
121.54.44.91 On: Sat, 02 Jan 2016 09:34:54
 063135-4 Lopez et al. J. Renewable Sustainable Energy 7, 063135 (2015)

FIG. 2. Boundary conditions.

FIG. 3. Mesh design.

TABLE II. Summary of mesh settings using ANSYS CFX meshing.

Location Meshing strategy Local settings

Cylinder surface Prism (inflation layers) Method First aspect ratio

First node distance 0.0001 m
First aspect ratio 20
Growth rate 1.2
Maximum layers 1000
Near-cylinder zone Tetrahedral Element size (on zone wall) 0.1 m
Zone diameter 2.25 m
Zones away from cylinder Hexahedral Element size 0.25

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
121.54.44.91 On: Sat, 02 Jan 2016 09:34:54
 063135-5 Lopez et al. J. Renewable Sustainable Energy 7, 063135 (2015)

the cylinder is similar to its inlet or free stream velocity, and the forces acting parallel and per-
pendicular to the wind are already equivalent to drag and lift, respectively. Lift and drag coeffi-
cients (CL and CD) were then estimated using Eqs. (1) and (2) below:

FL
CL ¼ ; (1)
1 2
qAU1
2
FD
CD ¼ ; (2)
1
qAU1 2
2

where q is the air density, A is the frontal cross-sectional area of the cylinder, U1 is the free
stream velocity, while FL and FD are lift and drag forces, respectively. With regard to blockage
due to wind tunnel walls, the blockage ratio was around 15% in the model. Based on literature,
the effect on drag coefficient is significant at this level. Thus, it can be expected that the actual
drag forces would be lower. The readers are directed to external references about blockage
correction (Plourde et al., 2011; Abraham et al., 2011; Takeda and Kato, 1992; and West and
Apelt, 1982).

R
D. Analysis of data using JMPV statistical software
A regression model containing the main effects, two- and three-way interactions and quad-
ratic effects, was fitted on the simulation data. The average R2 obtained was 0.967. Such regres-
sion model provides the predictive equation for multiple responses, namely, lift, drag, and lift-
to-drag ratio as a function of frustum diameter ratio, bump k/c ratio, helical groove height and
speed ratio. Multiple response surface analysis using desirability function was used to optimize
the aerodynamic design of a Magnus wind rotor. This method is one of the widely used statisti-
cal tools to solve multiple response variable problems (Derringer and Suich, 1980 and Myers
and Montgomery, 2002). A desirability function is defined where each response variable (Yi) is
transformed to a desirability value that could range between zero to one. In the case of lift and
lift-to-drag ratio wherein the response variables are to be maximized (larger-the-better type),
the individual desirability function is defined as follows:
8
>
> 0 Yi < Li
>
< Y  L 
i i
d i ðY i Þ  Li  Yi  Ti (3)
>
> T  Li
> i
: 1 Yi > Ti ;

where Yi is the response variable, Li is the lower limit or the smallest measured value of the
response in the experiment, and Ti is the target (or most desirable) value of the response. In
the case of drag wherein the response variable is to be minimized (smaller-the-better type), the
individual desirability function is defined as follows:
8
>
> 0 Yi > Ui
>
< 
Ui  Yi
di ðYi Þ ¼ Ti  Yi  Ui (4)
>
> Ui  Ti
>
:1 Yi < Ti ;

where Yi is the response variable, Ui is the upper limit or the largest measured value of the
response in the experiment, and Ti is the target (or most desirable) value of the response. Fig. 4
describes the linear desirability function for each response. Basically, the higher lift and lift-to-
drag ratio is, the better it is for power generation. The opposite is true for drag. The overall

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
121.54.44.91 On: Sat, 02 Jan 2016 09:34:54
 063135-6 Lopez et al. J. Renewable Sustainable Energy 7, 063135 (2015)

FIG. 4. Desirability functions for lift, drag and lift-to-drag ratio.

desirability, D, is then obtained from the weighted geometric mean of the individual desirabil-
ities as shown below

D ¼ dL wL dD wD dLDR wLDR ; (5)

where dL, dD, and dLDR are the individual linear desirability functions of lift, drag and lift-to-drag
ratio, respectively. The exponent, w, of each desirability function is a fraction between 0 and 1,
and is the weight or importance given to that response. The weights are normalized such that the
sum of wL, wD, and wLDR is always equal to 1. If a particular response is excluded from the cri-
teria, w is set to zero. A sensitivity analysis was done by optimizing the regression model using
desirability functions with varying importance given to lift, drag, and lift-to-drag ratio.

III. RESULTS AND DISCUSSION

A. Characteristics of the regression model
For all responses, the main effects of the modifications were more significant than the
interaction effects. The designs with dimples all generated the lowest lift forces, quickly elimi-
nating them from the study. The general trend of lift, drag, and lift-to-drag ratio as a function
of modification size is shown in Fig. 5. Enlarging the tip diameter of a frustum cylinder would
generate more lift and drag, as its larger cross-sectional area intercepts more wind. The result-
ing trend from modifying bump size is inconclusive. The trend shows that the smaller the
bump size, the better. It decreases both lift and drag. It is hypothesized by the authors that opti-
mal bump size is unique for different speed ratios, making it difficult to employ. Helical
grooves decrease both lift and drag generated, with drag at a faster rate. The best speed ratio
observed is high, from 4.9 to 5.5, which is congruent with the findings by Sun et al. (2012).
Increasing the frustum diameter ratio delivered the most lift enhancement. However, the
increase in drag accompanying it is significant. The most drag reduction comes from imple-
menting bumps, but its negative effect to lift generation is severe.

B. Sensitivity analysis
As mentioned in the methodology, the design of the Magnus wind rotor was optimized for
different scenarios with varying importance given to lift, drag, and lift-to-drag ratio. The sensi-
tivity analysis plotted in Fig. 6 shows how the optimized design changes as a function of lift
and drag importance.
It is important to take note that the analysis is limited only to the size range of modifica-
tions tested (see Table I). In a large region of the design space, the presence of the bump is
almost undesirable for the rotor. For wD < 0.5, when lift is more important than drag, the pre-
ferred bump size is always the smallest. When drag is not important, the preferred diameter ra-
tio increases, giving the cylinder more lift. Also, it is notable that the preferred height of helical
grooves increases as drag importance increases. This implies that helical grooves are very good
drag mitigators. For wD > 0.5, when minimization of drag becomes more important than maxi-
mization of lift, dramatic changes on the preferred design are observed. When maximization of
lift is no longer important (wD  0.7), helical grooves, formerly preferred due to its excellent

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
121.54.44.91 On: Sat, 02 Jan 2016 09:34:54
 063135-7 Lopez et al. J. Renewable Sustainable Energy 7, 063135 (2015)

FIG. 5. General trend of lift, drag and LDR as a function of modification size from regression model. Screenshot from
R
JMPV statistical software.

drag mitigation and minimal lift reduction, become undesirable. The bump, an even better drag
reducer but was previously undesirable due to its extremely negative effect on lift, now
becomes desirable for the rotor. Finally, it is also noticeable that a frustum cylinder is no longer
needed when lift is no longer important. It should be noted that a frustum cylinder generates
plenty of lift but also plenty of drag.
Overall, regression and sensitivity analysis suggests that utilizing frustum cylinders are best
for lift enhancement, while helical grooves are excellent for mitigating the drag accompanying
frustum cylinders with large diameter ratios. This is because while helical grooves are excellent
drag reducers, they are relatively neutral with regard to lift generation compared to the other
modifications. The trade-off between lift and drag for helical grooves is insignificant. The two
seem to be a promising combination. For lift generation purposes, bumps are not desirable.

FIG. 6. Sensitivity analysis of the design. Optimal design is plotted against drag importance (wD). Lift importance wL ¼ 1  wD
(e.g., lift and drag are equally important at wD ¼ 0.5).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
121.54.44.91 On: Sat, 02 Jan 2016 09:34:54
 063135-8 Lopez et al. J. Renewable Sustainable Energy 7, 063135 (2015)

TABLE III. Optimized designs for each scenario (weights in parenthesis).

Cylinder shape Bump Helical groove Overall

Scenario Desirability criterion (diameter ratio) (k/c ratio) (height) Speed ratio desirability

1 Max lift (0.5), Max LDR (0.5) 1.698 0.11481 0.0418 4.91 0.785
2 Max lift (0.5), Min drag (0.5) 1.451 0.11481 0.0657 5.2 0.644
3 Max lift (0.63), Min drag (0.37) 1.582 0.11481 0.0544 5.05 0.662
4 Max lift (0.83), Min drag (0.17) 1.794 0.11481 0.0366 5.5 0.744
5 Max lift (0.0.37), Min drag (0.63) 1 0.20218 0 5.5 0.664
6 Max lift (0.17), Min drag (0.83) 1.05 0.11481 0 1.5 0.756

C. Case studies
Six specific scenarios were selected as case studies (see Table III). With lift-based wind
rotors, it is common to optimize only with lift and LDR. However, in some scenarios, drag
may not be as important as lift. For example, for small rotor diameter applications and where
bending is expected to be minimal, a rotor blade may be able to withstand more thrust in
exchange for more torque. On the other hand, from the perspective of comparing a Magnus’ to
an airfoil rotor’s performance, drag reduction may be prioritized over lift enhancement as the
lift generated by a Magnus rotor is already ten times higher by default. The optimized design
and overall desirability is shown for each scenario in Table III. It is seen that the bump size is
the same for all scenarios (smallest), the diameter ratio increases as drag importance is reduced,
and helical groove height increases as drag importance increases. It is also observed that opti-
mizing with lift and LDR alone is similar to optimizing with lift and drag only with wL ¼ 0.74
and wD  0.26.

D. Aerodynamic performance
Over the whole study, the range of Reynolds number tested was 38 265.3 < Re < 76 530.6.
For comparison, the aerodynamic efficiencies of the optimized designs from each case study
scenario are plotted against other published results in Fig. 7 through a drag polar plot. The lift

FIG. 7. Aerodynamic efficiencies of optimized designs for each case study scenarios in comparison to published results.
Original data from Seifert (2012).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
121.54.44.91 On: Sat, 02 Jan 2016 09:34:54
 063135-9 Lopez et al. J. Renewable Sustainable Energy 7, 063135 (2015)

FIG. 8. Aerodynamic efficiencies of optimized designs for each scenario as a function of speed ratio.

coefficients were plotted with their corresponding drag coefficients at different speed ratios.
The aerodynamic efficiencies (lift to drag ratios) of all optimized cylinders are comparable,
except for the results from scenarios 5 and 6, which had low values for both lift and drag. It
can also be noted that the results obtained in this study are comparable to the results obtained
by Reid (1924) using plain cylinders operating within the same range of Reynolds number
(Re  39 k). However, it is observed that the modified cylinders from the current study achieved
lower lift and drag coefficients than the plain cylinder reported by Reid (1924). Swanson
(1961) and Badalamenti (2008) observed a significant dependency of a rotating cylinder’s lift
and drag to the Reynolds number. An increase in lift was noted as Re decreased. The frustum
cylinder in this study had Re  38 k at the root and about 55 k–65 k at the tip, where the diame-
ter is wider. The results suggest careful consideration of the Reynolds number in future experi-
ments. The lift-to-drag ratios of the optimized cylinders are plotted separately against speed
ratio in Fig. 8. It is seen in scenarios 2–4 that the peak lift-to-drag ratio is delayed in a higher
speed ratio as drag importance is increased. It is also noticeable that scenario 5 had very good
aerodynamic efficiency, though this is misleading. A high aerodynamic efficiency will not nec-
essarily translate to more power generation because if it has a very low lift coefficient.

IV. CONCLUSIONS
The present study demonstrated potential in improving the performance of Magnus wind
rotors through the use of frustum cylinders and helical grooves. Frustum cylinders provide the
most lift enhancement from the modifications considered while helical grooves, as utilized by
Murakami et al. (2009) in their proprietary Magnus wind turbine, offer excellent drag mitiga-
tion. This makes the two a promising combination moving forward with the technology. The
approach utilized to optimize rotor design using desirability functions provides a novel alterna-
tive to current techniques. Its flexible criteria cater realistic scenarios, in which drag reduction
may not be as important as lift enhancement.
Future studies may test the frustum-helical groove cylinder under lower Reynolds numbers,
as the aerodynamics’ dependency on it has been emphasized in literature multiple times, and as
seen on this study, it could have been the possible reason why the performance of our modified
cylinder was not able to surpass that of Reid’s (1924) plain cylinder. It would also be interest-
ing to extend the study to even higher speed ratios (>5.5) and to find the optimum roughness
height. Requiring validation is also the authors’ hypothesis that the optimal bump size is unique
for each speed ratio, making it a challenge to employ on a rotor expected to operate in a range
of varying speed ratios.

Abraham, J. P., Plourde, B. D., Mowry, G. S., Sparrow, E. M., and Minkowycz, W. J., “Numerical simulation of fluid flow
around a vertical-axis turbine,” J. Renewable Sustainable Energy 3, 033109 (2011).
R
ANSYSV, Academic Research, Release 14.5, ANSYS CFX-Solver Theory Guide (ANSYS, Inc., 2011).
R
ANSYSV, Academic Research, Release 14.5.
Badalamenti, C. and Prince, S., “Effects of endplates on a rotating cylinder in crossflow,” in Proceedings of the 26th AIAA
Applied Aerodynamics Conference, AIAA, Honolulu, Hawaii (2008).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
121.54.44.91 On: Sat, 02 Jan 2016 09:34:54
 063135-10 Lopez et al. J. Renewable Sustainable Energy 7, 063135 (2015)

Butt, U. and Egbers, C., “Aerodynamic characteristics of flow over circular cylinders with patterned surface,” Int. J.
Mater., Mech., Manuf. 1(2) (2013).
Bychkov, N. M., Dovgal, A. V., and Kozlov, V. V., “Magnus wind turbines as an alternative to the blade ones,” J. Phys.
75, 012004 (2007), available at http://m.iopscience.iop.org/article/10.1088/1742-6596/75/1/012004/meta;jsessionid=
81E8745E62DA2E3D4414F2F938B436CB.c1.iopscience.cld.iop.org.
Derringer, G. and Suich, R., “Simultaneous optimization of several response variables,” J. Qual. Technol. 12(4), 214–219
(1980), available at http://asq.org/qic/display-item/?item=5341.
Enercon, “Powered by sailing rotors: E-Ship 1 in the testing phase,” in Windblatt (2010), No. 2, pp. 06–07.
Huang, S., “Cylinder drag reduction by the use of helical grooves,” in Proceedings of the HYDRALAB III Joint User
Meeting (2010).
R
JMPV, Version 11.SAS Institute Inc., Cary, NC, 1989–2007.
Luo, D., Huang, D., and Wu, G., “Analytical solution on Magnus wind turbine power performance based on the blade ele-
ment momentum theory,” J. Renewable Sustainable Energy 3, 033104 (2011).
Mara, B. K., Mercado, B. C., Mercado, L. A., Pascual, J. M., and Lopez, N. S., “Development and validation of a CFD
model using ANSYS CFX for aerodynamics simulation of Magnus wind rotor blades,” in Proceedings of the 7th IEEE
HNICEM Conference, Manila, Philippines, 12–16 November (2014a).
Mara, B. K., Mercado, B. C., Mercado, L. A., Pascual, J. M., and Lopez, N. S., “Preliminary turbulence model validation
for flow across rotating cylinders using ANSYS CFX,” in Proceedings of the 7th IEEE HNICEM Conference, Manila,
Philippines, 12–16 November (2014b).
Menter, F. R., Eddy Viscosity Transport Equations and Their Relation to the k-Epsilon Model (National Aeronautics and
Space Administration/Ames Research Center, Moffett Field, CA, 1994).
Mowry, G. S., Erickson, R., and Abraham, J. P., “Computational model of a novel, two-cup horizontal wind-turbine sys-
tem,” Open Mech. Eng. J. 3, 26–34 (2009).
Murakami, N. and Ito, J., “Magnus type wind power generator,” US patent 7504740 B2 (17 March 2009).
Myers, R. H. and Montgomery, D. C., Response Surface Methodology: Process and Product Optimization Using Designed
Experiments, 2nd ed. (Wiley, New York, 2002).
Plourde, B. D., Abraham, J. P., Mowry, G. S., and Minkowycz, W. J., “An experimental investigation of a large, vertical-
axis wind turbine: Effects of venting and capping,” Wind Eng. 35, 213–220 (2011).
Reid, E. R., “Tests of rotating cylinders,” NACA Technical Memorandum No. TM-209 (1924).
Sedaghat, A., “Magnus type wind turbines: Prospects and challenges in design,” Renewable Energy 62, 619–628 (2014).
Seifert, J., “A review of the Magnus effect in aeronautics,” Prog. Aerosp. Sci. 55, 17–45 (2012).
Sparrow, E. M., Abraham, J. P., and Chevalier, P. W., “A DOS-enhanced numerical simulation of heat transfer and fluid
flow through an array of offset fins with conjugate heating in the bounding solid,” J. Heat Transfer 127, 27–33 (2005).
Srivastav, D., “Flow control over airfoils using different shaped dimples,” in International Proceedings of Computer
Science & Information Technology (2012), Vol. 33, p. 92.
Sun, X., Zhuang, Y., Cao, Y., Huang, D., and Wu, G., “A three-dimensional numerical study of the Magnus wind turbine
with different blade shapes,” J. Renewable Sustainable Energy 4, 063139 (2012).
Swanson, W. M., “The Magnus Effect: A Summary of Investigation to Date,” J. Basic Eng. 83(3), 461–470 (1961).
Takeda, K. and Kato, M., “Wind tunnel blockage effects on drag coefficient and wind-induced vibration,” J. Wind Eng.
Ind. Aerodyn. 42, 897–908 (1992).
Thom, A. and Sengupta, S. R., “Effect of discs on the air forces on a rotating cylinder,” Aeronautical Research Committee,
Reports and Memoranda No. 1623 (1932).
Tokomaru, P. and Dimotakis, P., “Rotary oscillation control of a cylinder wake,” J. Fluid Mech. 224, 77–90 (1991).
West, G. S. and Apelt, C., “The effects of tunnel blockage and aspect ratio on the mean flow past a circular cylinder with
Reynolds numbers between 104 and 105,” J Fluid Mech. 114, 361–377 (1982).
Wilcox, D. C., Turbulence Modeling for CFD (DCW Industries, Inc., La Canada, CA, 1993).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
121.54.44.91 On: Sat, 02 Jan 2016 09:34:54