You are on page 1of 16

Applied Energy 199 (2017) 479–494

Contents lists available at ScienceDirect

Applied Energy
journal homepage:

Increasing the operational capability of a horizontal axis wind turbine by

its integration with a vertical axis wind turbine
Bala Govind
Department of Mechanical and Aerospace Engineering, North Carolina State University, Raleigh, North Carolina 27606, United States

h i g h l i g h t s g r a p h i c a l a b s t r a c t

 A new mechanism transfers torque Sha Rotaon Reversal CVT HAWT Generator
from HAWT rotor to an integrated
VAWT drive-train at high wind
 Operational range is improved for
prototype-scale 12 kW HAWT-10 kW
VAWT combination.
 A k-x (SST) turbulence model
suggests safe rotor clearance for the Primary Right Angle Transmission
integrated system.
Shell-VAWT Moon Transfer
 Aerodynamic feasibility reveals
effects of torque ripple on the
VAWT Generator
combined power output.

a r t i c l e i n f o a b s t r a c t

Article history: A major difficulty encountered by a horizontal axis wind turbine is the limit of aerodynamic torque that it
Received 31 January 2017 can withstand at high wind speeds. A novel strategy is proposed to improve the operational capability of a
Received in revised form 8 April 2017 prototype scale system by increasing its rated wind speed for power generation. This is achieved by inte-
Accepted 26 April 2017
grating its drivetrain with that of a vertical axis wind turbine supported on a common tower. Excess torque
Available online 12 May 2017
is transferred from the horizontal axis rotor to the vertical axis rotor’s drivetrain by coupling them using a
continuously variable transmission. In this article, firstly, the concepts of motion transfer that facilitate
this combined operation are discussed. A combination of a 12-kW horizontal axis rotor and a 10-kW ver-
Horizontal axis wind turbine
Vertical axis wind turbine
tical axis wind turbine is studied to estimate the increased benefit of increments in rated wind speed.
Hybrid wind turbine design Performance of this hybrid system is predicted at potential wind sites and is shown to exceed the stan-
Motion transfer dalone mechanical power output of both subsystems under different wind regimes. The critical criterion
Computational fluid dynamics of the system’s aerodynamic feasibility is then investigated. Turbulence modelling is performed for a con-
Turbulence modelling figuration which involves a combination of the NREL Phase VI rotor and a NACA 0021 profiled vertical axis
H-rotor. A 3-D simulation, using a validated k-x (Shear Stress Transport) computational fluid dynamics
model helps confirm the ability of both turbines to operate aerodynamically independent of each other.
Further, by this methodology, a safe clearance between the two rotors is pre-determined. Analysis of tur-
bulent flow scenarios reveals the characteristic effects of aerodynamic torque ripple experienced by the
vertical axis wind turbine and its impact on combined power output. Parameters outlined in this article
will be of assistance in the practical implementation of the integrated axes wind turbine.
Ó 2017 Elsevier Ltd. All rights reserved.

E-mail addresses:,
0306-2619/Ó 2017 Elsevier Ltd. All rights reserved.
480 B. Govind / Applied Energy 199 (2017) 479–494


HAWT horizontal axis wind turbine xHB rated Angular speed of HAWT rotor
VAWT vertical axis wind turbine xV angular speed of VAWT rotor
CFD Computational Fluid Dynamics saH aerodynamic torque on HAWT main shaft
NREL National Renewable Energy Laboratory scH control torque exerted by HAWT generator
TSR Tip Speed Ratio saV aerodynamic torque on VAWT central shaft
CVT Continuously Variable Transmission scV control torque exerted by VAWT generator
RANS Reynolds Averaged Navier-Stokes (Equations) R radius of rotor
NACA National Advisory Committee for Aeronautics q density of air
NASA National Aeronautics and Space Administration k Tip Speed Ratio
Rpm revolutions per minute Cp coefficient of power
Z 1 ,Z 2 ,Z 3 ,Z 4 tooth numbers of VAWT motion transfer gear-train Cq coefficient of torque
r CVT ratio P  P1 gauge pressure
ucH cut in wind speed for the HAWT A projected area of rotor
uHB rated wind speed for the HAWT v1 free stream wind speed
uVB rated wind speed for the VAWT TKE Turbulent Kinetic Energy
uFH furling wind speed for the HAWT
uFV furling wind speed for the VAWT
xH angular speed of HAWT rotor

1. Introduction overhang in such a configuration could be challenging. The key

consideration for all such hybrid designs is that implementation
Horizontal axis wind turbines (HAWTs) are the primary source at a megawatt-scale is governed by economy. This is compounded
of grid connected wind power [1]. A primary issue encountered is by the breakdown of critical components, thus increasing opera-
that at high wind speed nearing its wind rated speed, the HAWT’s tional cost [22].
operation is stall regulated or pitch controlled to limit power gen- To improve the mechanical power output at the hub height of
eration by reducing lift on its blades [2]. This limits the optimal the HAWT, some improvements in the existing motion transfer
torque exerted and causes the wastage of a portion of potential mechanism within the nacelle have been undertaken. These
mechanical power at a site. Therefore, to better facilitate energy include the use of direct-driven PMSGs which avoid the need of
capture from the three-bladed turbine, drivetrain modifications intermediate gearboxes. Also, to facilitate smooth gear ratio transi-
and control schemes are being designed and improved to maxi- tion, the use of Continuously Variable Transmission [23–25] is
mize the power drawn at varying wind speed. More recently, being investigated. Contrary to a manual or a conventional auto-
research is veering towards the incorporation of artificial neural matic transmission, the operation of a CVT involves no torque
networks [3,4] which form a framework of learning algorithms interruption during change in angular speed. While CVTs have
which help correlate on-site wind data with the power coefficient been investigated for active drivetrains in HAWTs to convert gusty
and pitch angles. In particular, the adaptive neuro-fuzzy inference wind power to stable alternating current for synchronous genera-
system (ANFIS) [5–7] is gaining popularity. On a broader scale, it tion [26], practical designs which address optimum torque transfer
can also assist in maximizing profit from empirical wind profiles and ability to raise rated wind speed haven’t been adequately
for wind farms [8–10]. Feathering and pitch actuation models are implemented.
also being improved to optimize the angle of attack of the wind With an apparently disparate purview, to harvest more energy
and the instantaneous Tip Speed Ratio (T.S.R) of the HAWT. These near the ground surface, vertical axis wind turbines (VAWTs) too
are supported by soft computing techniques for power coefficient are being re-designed to up-scale their drivetrains. To save costs,
optimization [11,12] and Maximum Power Point Tracking (MPPT) intermediate gearboxes are avoided and instead, low solidity ratio
algorithms [13,14] which enhance the synchronous power output models which facilitate low torque and high rpm central shaft rota-
from the turbine’s grid-connected Permanent Magnet Synchronous tion to optimize the use of alternators, are used [27]. Consequently,
Generator (PMSG) or Doubly-Fed induction Generator. Here, the more effort has been directed to fabricate VAWTs of a 1–20 kW
use of the sliding mode control strategy [15–17] is a popular scale [28]. By arranging these turbines in arrays so they may mutu-
method which helps optimize the functioning of the turbine’s driv- ally aid each other in rotation [29], it is claimed that they produce a
etrain elements. While these strategies may assist in compensating higher aggregate of mechanical power. Unlike their horizontal axis
for the limitations in the existing capability of the HAWT, attention counterparts, however, VAWTs cannot access the stream of fast
must pivot towards the fundamental design of the three-bladed flowing wind as they are located at a lower elevation, have a smal-
system. ler swept area and suffer from turbulence at ground level.
To increase the mechanical power a HAWT can harness, In an attempt to address these difficulties, the author proposes a
changes in the design of the three-bladed rotor are being new mechanism for a prototype scale wind turbine system. This
attempted. One imaginative method to draw more power is to incorporates the use of a shell-type VAWT installed on the same
access the stream of fast-flowing wind at higher altitudes by the support tower as the HAWT. This constitutes a strategy to optimize
crosswind motion of tethered wings [18,19]. Another bold the power output of the HAWT in a unique way. A novel motion
approach entails radically increasing the swept area of the turbine. transfer mechanism comes into operation when the wind speed
For example, a new design was tested by Vestas Wind Systems [20] at the hub height of the HAWT approaches or exceeds its original
wherein multiple 3-bladed rotors were supported at different rated wind speed. The modified HAWT drivetrain is designed to
eccentricities from a central tower to achieve a 900-kW output. transfer the excess torque exerted on the HAWT’s main shaft to
While it is claimed that this reduces the Levelized Cost of Energy the VAWT’s central shaft. This further utilizes a CVT for a purpose
(LCOE) [21] and construction and transportation costs, the rotor removed of its normal purpose of power transmission from rotor to
B. Govind / Applied Energy 199 (2017) 479–494 481

generator. As it is capable of smooth acceleration with no motion the concepts of motion transfer and more importantly, findings
seizure, the CVT facilitates the seamless integration of the HAWT pertinent to aerodynamic feasibility of the hybrid system. The pre-
rotor drivetrain and the shell-VAWT drivetrain at high wind ferred method of actuation of drivetrain components is explained
speeds. further in the author’s specification (US Pat. Appl. 2016/0201652
In this design, the excess mechanical power of the HAWT acts as A1 [33]).
a source and the VAWT generator acts as a sink. While two-ended
torque transfer can be achieved until both turbines reach their 2. Primary drive-train and methodology to effect torque
rated wind speeds, a VAWT’s rotor may not normally facilitate transfer from the horizontal-axis subsystem to the vertical-axis
the source of extra energy. The HAWT, therefore, possesses the subsystem
higher angular momentum which must be limited. Since electric
power is also independently generated by the VAWT at a lower Under normal operating conditions, a conventional HAWT rotor
height location on the same support tower, this arrangement pro- facilitates a prime mover rotating at an angular speed conducive to
vides a high-density power resource in wind farms which have independent synchronous electricity generation. When its main
space constraints. This progressive technology also aims to provide shaft’s angular speed dips below a reference rpm, auxiliary power
a failsafe option for operators to use a low-capacity aging HAWT at is supplied to make the rotation synchronous. Alternatively, the
higher wind speeds without having to decrease aerodynamic tor- pitch of its blades is varied to alter the Tip Speed Ratio (TSR) which
que by sub-optimally using feathering or pitch mechanisms. By normally lies in range of 5–6 for a 3- bladed turbine. In cases of
reducing the loading on the HAWT’s original drivetrain, excess excessive dynamic load, mechanical or electrical brakes may be
power is derived at the VAWT’s generator while also making opti- applied to limit electricity generation until favourable conditions
mum use of ‘real estate’ of the tower. This considerably increases ensue. The energy supplied by the wind at every instant is propor-
the net swept area of the original turbine. tional to swept area of the rotor and to the cube of the prevailing
The article is divided into two main sections which discuss the wind speed. Therefore, by limiting the force of the wind by sub-
practical considerations of the proposed design’s implementation. optimally altering the Tip Speed Ratio, the full potential of the
The first section expounds on the design of drivetrain elements wind may, at higher wind speed, be unutilized due to limitations
within the nacelle and H-rotor VAWT which benefit from torque of generator loading. Instead, if there were a mechanism by which
transfer at hub-height wind speeds exceeding the standalone the excess mechanical power from this energy source at a higher
HAWT’s rated wind speed. It further offers the preferred mode of altitude could be siphoned to a sink (or a lower energy source),
operation in different regimes. A combination of the NREL Phase then the HAWT rotor may continue operating at its rated angular
VI rotor and a small manufacturer’s 10 kW VAWT and their rele- speed in a free wind stream which exceeds its original rated wind
vant power characteristics is utilized to estimate of the mechanical speed.
power drawn for increments in new rated wind speeds. Power pre- To obtain a profitable combined performance, the coupled sys-
dictions at two potential wind sites in the state of North Carolina tem proposed must fulfil three criteria:
are also made. For instance, it is shown that during integrated
operation, by preventing further reduction of power coefficient of (1) A mechanically robust motion transfer mechanism which
a 12-kW HAWT beyond its original rated wind speed and using ensures complementary rotary action between both
only a 1 m/s rise in rated wind speed, the system can yield an extra subsystems.
2.9 kW at the VAWT’s generator. This is in addition to the 12 kW of (2) Aerodynamic feasibility, i.e. the effect of the rotation of one
independent HAWT power output and the 10–12 kW of indepen- turbine on the flow field should not adversely affect the
dent VAWT output power. other turbine.
The second section critically examines the aerodynamic feasi- (3) An intelligent operation sequence/control algorithm which
bility intended for the prototype scale hybrid system. It is impor- ensures a smooth transition in the hybrid power curve.
tant to minimize the possible mutual effects of turbulence caused
by either turbine during combined operation. This is discussed The mechanism presented here serves to aid the original low-
using a 3-D time transient Computational Fluid Dynamics (CFD) capacity HAWT and accommodate the increment in extra torque
approach to visualize the quasi steady-state performance of the at higher wind speeds. This is done by directing the extra aerody-
integrated system. The hybrid design uses the NREL Phase VI namic torque exerted on the HAWT to the drivetrain of a VAWT
S809 rotor and NACA 0021 profiled H-rotor geometry. The mounted on the same support tower. This VAWT’s generator pro-
HAWT’s independent performance is first validated with NREL’s duces a control torque which may be varied electromechanically
experimental results and the model is extended to study and con- to track the optimal power point according to prevalent wind con-
firm the aerodynamically independent operation of both turbine ditions. While this may avoid having to completely reinforce the
subsystems for cases of different angular speed and wind speed. blades and tower, local strength at certain divisions of the tower
The analysis further reveals the effects of the characteristic tor- will need to be enhanced.
que ripple experienced by a straight-bladed VAWT on the total From Fig. 1, it can be conceived that in the prototype-scale ‘Inte-
mechanical power output. This will help design the H-rotor to grated Axes Wind Turbine’ system, a modified H-rotor VAWT is
operate in a pre-determined profitable range of angular speed. mounted on the support tower of a HAWT. The model for this pro-
The stability of solutions of turbulence modelling also helps totype consists of two rotors whose design is contingent on a few
establish the suitability of the k  x (Shear Stress Transport) trade-offs:
model [30,31] over the k  e (Realizable) model [32] for this
investigation.  Drivetrain analysis [34] shows that the torque ripple effect
An attempt to fabricate the design of the prototype scale inte- common in VAWT designs due to upstream and downstream
grated axes wind turbine will be undertaken shortly. Notably, the loading on VAWT blades is more pronounced for a 2-bladed
strategy of the proposed operational sequence of its drivetrain is design than a 3-bladed design. There would a time periodicity
generic. Only real-time data may facilitate an optimal learning in the net torque acting on the VAWT shell and support tower
framework for the control of the hybrid modes and the associated if a 2-bladed design was used. This supports the selection of a
convertor circuity. This article will constitute an effort to explain giromill-type rotor for a simplified model to avoid complica-
482 B. Govind / Applied Energy 199 (2017) 479–494

Sha Rotaon Reversal CVT HAWT Generator Z 3 are tooth numbers of gear and pinion for the inner drive. How-
ever, installation of gear drives necessitates that the tower be made
locally strong. If this modification is extended to a multi-megawatt
utility, this may require a newer certification.
Another implication of this alteration in a conventional VAWT’s
design is the change in active pitch mechanism design. Its skeletal
representation is shown in Animation sequence 2. Here, straight
blades are mounted on the H-rotor shell by means of revolute
joints at the ends of radial arms. In practice, the angular position
of blades would depend on the elevation of a cylindrical slider
whose position is in turn dependant on the action of a plurality
Primary Right Angle Transmission
of hydraulic cylinders which are housed outside the tower (refer
specification [33]). This enhances serviceability and ease of
Shell-VAWT Moon Transfer replacement.

VAWT Generator 2.2. Continuously Variable Transmission for coupling the two turbine

To resolve the issue of a smooth transition in rotary power

transfer, an intermediate Continuously Variable Transmission unit
is installed in the HAWT nacelle. Although numerous variants in
CVT designs can be considered, the push-belt design is employed
here for explanatory purposes. However, for a higher torque
(100 kW–2 MW) wind turbine utility, the hydro-viscous type
[23,24] is preferable.
Fig. 1. Primary drive train of the prototype-scale hybrid system consisting of Fig. 2 shows an expanded schematic of this setup. In practice,
horizontal axis wind turbine and vertical axis wind turbine sub-systems. Excess
compact actuation is possible. This configuration is governed by
mechanical power is drawn at the VAWT generator by torque transfer.
two sets of movable and fixed pulley sheaves and a segmented
steel belt which transfers power between two parallel shafts akin
a transaxle arrangement. The radii of the CVT’s sheaves depend
tions arising from excessive superimposition of oscillations on on the varying speed ratio of the parallel shafts passing through
the steady mean torque used to estimate the integrated axes them.
turbine’s performance. The CVT’s transient ‘gear ratio’ is continuously varied by chang-
 Exhaustive aerodynamic analysis [35,36], however, shows that ing the width of the sheaves which result in seamless change in
VAWT designs with struts can introduce large amounts of drag angular speed. As shown, the secondary main shaft’s excess torque
and can significantly reduce its performance. acts as the drive. Unlike gear-trains, this ensures smooth coupling
 To alleviate ground turbulence effects the HAWT hub height of the rotors without sudden changes in rpm of power transmitting
may be increased but this necessitates more material for tower elements of both supporting drivetrains. This further makes provi-
installation. sion for a constant speed drive from the HAWT when the VAWT’s
central shaft’s rpm is maintained constant after it reaches its rated
To better visualize the sequence of motion transfer, the action value. This, however, requires that a highly durable steel belt of
of the new drivetrain is explained as a series of its components constant length be used to withstand the increased sheave pres-
which now follows. sure during compression.
Fig. 2 also shows the mechanism of engagement with the CVT
2.1. Rotary motion transfer within the vertical axis wind turbine and and is similar to the mechanism of a dog clutch. Hydraulic linear
its active pitch mechanism actuators engage toothed wheels to the slotted cavities in the
sheaves of the CVT. Retractable telescopic shafts engage with the
The VAWT consists of a thin-walled shell which rotates about CVT via a spring mechanism. They are protected from wobbling
the central axis of the HAWT’s tower. Radial arms are attached effects by support fixtures.
by suitable means to the shell. This provision is made so that mate-
rial is distributed at a greater radius from the axis and subse- 2.3. Main Right-Angled Transmission within the HAWT nacelle
quently yields a greater moment of inertia. This generates
greater torque and, therefore, more mechanical power. The excess torque is transferred from the secondary main shaft
Animation sequence 1 (a) shows a variation in motion transfer to an eccentrically aligned horizontal shaft which is engaged with
arrangement from the tower-concentric rotating shell of the verti- the Main Right-Angled Transmission. This transfer occurs only
cal axis turbine to the central shaft of the VAWT sub-system. As when the two parallel, horizontally aligned shafts are engaged
shown, the configuration consists an external ring bevel gear with the CVT. There is an advantage to using two helically profiled
which transfers rotary motion to an internal central gear via bevel bevel gears to transfer motion from the secondary main shaft the
gear pinions. The pinions rotate faster that the main bevel gear. VAWT central shaft. They ensure that torque interplay between
Noticeably, the apexes of all the bevel gears are coincident. from the horizontal rotor to the central shaft occurs when the
It is possible to improve this arrangement. Animation nacelle simultaneously yaws about the tower. This is shown in
sequence 1 (b) shows a face gear drive which is easier to assemble Animation Sequence 3. Here, the bevel gear which is integral to
than the bevel gear drive. For instance, it is simple to equalize the nacelle rolls along the circumference of the other bevel gear
input and output velocities by selecting appropriate tooth numbers which is integral to the central shaft. The latter’s axis of rotation
to fulfil the condition given by ZZ21  ZZ43 ¼ 1, where Z 1 and Z 2 are corresponds to the axis of rotation of the nacelle and also to the
tooth numbers of gear and pinion for the outer drive and Z 4 and longitudinal axis of the support tower of the HAWT. In the pre-
B. Govind / Applied Energy 199 (2017) 479–494 483

Fig. 2. Expanded schematic of mechanisms incorporated in (a) the HAWT’s nacelle, (b) the shell-VAWT’s rotary motion transfer and (c) simplified VAWT active pitch control.

ferred construction, it is desired that both bevel gears be equally box ensures that the two parallel horizontal shafts engaged with
sized to transmit uniform angular speed without speed reduction. the CVT complement the rotation of the VAWT central shaft.
The integration of the aforementioned components embodies a
2.4. Shaft rotation reversal gearbox to ensure complementary rotary new system that transmits optimal torque of horizontal and verti-
action cal axes rotors to their electricity generating components.
Other support fixtures, fluid couplings and bearings [33]
One disadvantage of the push-belt CVT is that the belt is direc- installed will mitigate problems caused by the fatigue loading
tional. It isn’t designed to transfer equal torque to the transaxle in and vibration in the tower. Arguably, to increase the rated wind-
the reverse direction. If the HAWT and VAWT rotors do not facili- speed of the HAWT rotor, a generator with a higher power rating
tate such complementary rotation by independent operation when could be installed, and thereby avoid the entire CVT/VAWT
engaged, there wouldn’t be profitable power transfer between the arrangement. However, it may only accommodate an increased
coupled drivetrains and as such, they would oppose each other. aerodynamic torque which cannot far exceed the rated loading of
Therefore, for completeness in operating procedure, a gearbox the original drivetrain. The point of the design is to make the inte-
containing a bevel gear clutch is installed prior to the engagement grated components modular. While the idea of using a single gen-
of the HAWT main shaft to the CVT in the nacelle. erator in the nacelle to couple both turbines may be considered, it
The action of this shaft rotation reversal mechanism in the drive isn’t preferable for two reasons:
train is shown in Animation sequence 4. As shown in Fig. 2, the
bevel gear A is the driver. The secondary main shaft rotates in  It is expensive to redesign or install a generator which was orig-
either clockwise or anticlockwise direction depending on which inally electrically rated to match the original HAWT drivetrain’s
of the two gears B or C is engaged by the clutch. B and C rotate capabilities
freely on the secondary main shaft. The clutch consists of a toothed  Directly engaging an extended central shaft to the HAWT gener-
collar and has a splined profile to engage with the output shaft. The ator and without variable speed transmission may be mechan-
gearbox further forms a secondary right angle transmission which ically and materially infeasible when scaled up. Further, during
transfers motion from the eccentrically aligned HAWT primary downtime or component service, operations of both turbines
main shaft to the secondary main shaft. When actuated, this gear- would be halted.
484 B. Govind / Applied Energy 199 (2017) 479–494

3. Generic drivetrain coupling sequence and prediction of At every instant, the collective aerodynamic torque exerted on
additional power derived the HAWT rotor hub and blades is saH ¼ 12 q  pR3  C q ðb; kÞ  uH ðtÞ2 .
xH R
Here, Tip Speed Ratio (T.S.R), k, equals uH
and the torque coeffi-
The flow-of-control sequence leading to torque interplay
Simultaneously, a control torque, scH is exerted
between the horizontal and vertical axes rotors is shown in cient, C q , equals k
Fig. 3. The controller receives input data of wind speed at the by the HAWT generator and is responsible for the horizontal axis
HAWT hub height (uH ), wind speed at the VAWT mid-plane height rotor’s synchronous speed regulation. This control torque cannot
(uV ), aerodynamic torque on the primary main shaft (saH ), aerody- exceed a maximum value scHB which may be pre-determined by
namic torque on VAWT central shaft (saH ), angular speed of the the DFIG/PM generator characteristics.
secondary main shaft (xH ) and angular speed of the VAWT central The most typical scenario of hybrid operation may arise as fol-
shaft (xV ). lows. If by site-specific design, the VAWT central shaft reaches its
The controller then monitors the gradient in angular speeds xH rated angular speed xVB before the HAWT reaches its rated rpm
and xV . It may optimize independent synchronous generation of xHB , xV would be maintained constant at xVB even before cou-
from the HAWT according to the Maximum Power Point Tracking pling the two drivetrains. Also, at the culmination of the mode of
algorithm in the wind-speed range ucH 6 uH 6 uHB at the hub operation corresponding to ucH 6 uH 6 uHB , the HAWT’s angular
height of the HAWT, where ucH is the HAWT cut-in speed and uHB speed would approach xHB and would need to be maintained con-
is the original rated wind-speed of the HAWT. It may further follow stant by a combination of torque transfer and optimal pitch con-
a similar tracking procedure or Hill Climb Search algorithm for the trol. This implies that after coupling the two rotors, the
VAWT, i.e. in the range ucV 6 uV 6 uVB at the mid-plane height of controller would monitor the gradient of the rise in their rpm
the VAWT. and by suitable actuation, vary the lift forces on the turbines’
blades by the pitch mechanism of either turbine to keep xH under

Release HAWT and VAWT

subsystems’ mechanical brakes

Check if starting torque( ) on HAWT is sufficient Check if starting torque ( )on VAWT is sufficient
Yes No No Yes

Vary HAWT blade pitch angle/ Vary VAWT blade pitch angle/
yaw angle/ supply auxiliary power supply auxiliary power

Monitor gradient of rise in secondary main shaft’s Monitor gradient of rise in central shaft’s rpm by
rpm by varying HAWT control torque varying VAWT control torque

Check if secondary main shaft

Yes No

Check if rotational sense of both rotors would be

complementary during engagement with CVT.

Yes No

Check if wind Yes

Actuate Shaft Rotation Reversal Mechanism speed
equals or exceeds No

Engage (or continue engagement of) HAWT secondary

main shaft and VAWT central shaft by CVT

Vary VAWT generator’s to keep integrated Disengage/ (continue disengagement of) HAWT
HAWT’s under original rated main shaft and VAWT central shaft from CVT

Continue HAWT’s independent operation Continue VAWT’s independent operation

Fig. 3. Mode of operation leading to the coupling of HAWT and VAWT drivetrains. It culminates torque transfer from the horizontal rotor to the vertical rotor’s drivetrain
when uH exceeds uHB .
B. Govind / Applied Energy 199 (2017) 479–494 485

xHB . Effectively, even though the wind-speed exceeds that corre- Table 1
sponding to the HAWT’s rated angular speed, it continues to draw Vital characteristics of SAWT Inc. PK-10 AB 10 kW (nominal rating) VAWT.

a larger magnitude of power due to the transfer of a fraction of tor- Rated Power/Maximum Power 10 kW/12 kW
que which exceeds that of its independent rated operation. Rated Wind Speed, uHB 12 m/s
DP HAWT , the incremental power extracted, is a product of incre- Cut-in Wind Speed, ucV /Cut-out Wind 2 m/s/25 m/s
Speed, uFV
mental torque DsH and the constant rated angular speed xHB . To H-rotor diameter 6m
implement this, the CVT must be brought into action to couple Blade height 6.2 m
the HAWT’s main shaft and the central shaft of the VAWT genera- Generator Type/Rated Voltage Permanent Magnet-3 Phase/AC
tor at its desired constant speed of xVB . The additional power is 250 V
Speed Control Automatic Pitch Control and
optimized by varying VAWT control torque and keeping constant
Hydraulic Brake
the angular velocity of both the HAWT main shaft and VAWT cen-
tral shaft after their integration, while maintaining an optimal T.S.
R. Ideally, CVT ratio r would be given by x .
H The two subsystems’ independent power curves are shown in
The mode of hybrid operation is effective in the range Fig. 5. The independently operating HAWT’s uHB equals 12.5 m/s
uHB 6 uH 6 uFH , where uFH is the furling speed. Importantly, during and the independently operating VAWT’s uVB equals 12 m/s. When
this mode, effective torque on the HAWT during operation of the uH exceeds uHB , torque transfer comes into effect. Fig. 5 further
CVT equals saH  scHB . shows the consequent increase in mechanical power drawn from
The excess torque can be expressed as the integrated axes wind turbine. It far exceeds the combined indi-
1 1 vidual capacity of both turbines. Predictions are also shown for
DsH ¼ q  pR3  C qH ðb; kÞ  uH ðtÞ2  q  pR3  C qHB ðb; kÞ  uHB ðtÞ2 increments of 1 m/s, 2 m/s and 3 m/s in uHB of the Phase VI rotor.
2 2
ð1Þ The torque transfer to the VAWT generator in this range of uHB is
immediately apparent. These predictions show that even if rotary
C qH would be maintained constant at C qHB in this regime. If the motion is inefficiently transmitted by the CVT and the intermedi-
transmission efficiency of the CVT and right angle transmission is ate right angle transmission and HAWT’s rotation is maintained
say, gtr ; the equivalent torque transmitted to the VAWT generator is at a C qHB of only 0.524, or a C pH of 0.1, the integrated system would
  harness an additional 2.93 kW of mechanical power at the VAWT
1 1 gtr
DscVH ¼ q  pR3  C qH ðb; kÞ  uH ðtÞ2  q  pR3  C qH ðb; kÞ  uHB ðtÞ2  generator with an increase of only 1 m/s in uHB . This is in addition
2 2 r
to the independent 10–12 kW mechanical power output of the
VAWT. Further, this can be implemented without feathering or
Consequently, the increment in mechanical power derived is sub-optimally varying the HAWT’s blades’ pitch. This formulation
equal to DscVH  xVBH . This is manifested as electromechanical con- will be used to draw parallels with the forthcoming aerodynamic
version at the VAWT’s generator, subject to constraints of effi- feasibility discussion.
ciency of transmission such as influence of slip and friction in To better estimate the performance of the integrated system at
the CVT variator. gtr should normally vary in the range of 70– potential installation sites, wind profile data was collected from
90%, depending on xH [37]. two onshore weather stations in the state of North Carolina. This
If uH subsides to a value less than uHB , the decoupling of the data is archived by the State Climate Office of North Carolina at
main HAWT shaft and the VAWT central shaft will follow. How- North Carolina State University. The data was collected for Hat-
ever, if uH exceeds the furling speed uFH of the integrated system, teras, Dare County and Grandfather Mountain, Avery County and
the HAWT’s further operation is stopped due to mechanical and was sampled at an hourly rate for a 90-h cycle. By estimating the
electrical limitations. Feathering mechanisms and mechanical mechanical power available at the hub height of the HAWT
brakes would then be used for both rotors when uH P uHF and
uV P uVF .

4. An illustrative power curve prediction r = 0.5

Additional Torque at VAWT Generator (in kN-m)

As outlined in the preceding section, in its simplest mode of

operation, the VAWT is designed to independently reach its rated r = 0.625
mechanical power output before the HAWT reaches its indepen-
dent rated mechanical power output. One preferred combination r = 0.75
of a prototype setup entails integration of the two-bladed NREL
4 r = 0.875
Phase VI S809 HAWT rotor [38–40] and SAWT Inc.’s PK-10 AB
r = 1.0
VAWT [41,42]. The former has a blade length of 4.8 m and its r = 1.125
rotor’s design will be discussed further in the forthcoming section 3 r = 1.25
r = 1.375
of turbulence modelling of the hybrid turbine. The VAWT is a small r = 1.5
manufacturer’s commercially available 10 kW H-rotor and is of 2
comparable scale as the HAWT. Its details are listed in Table 1. Fur-
ther, this VAWT’s capabilities are similar to those used in the forth- 1
coming discussion on aerodynamic feasibility.
Using Eq. (2) of the preceding formulation, the maximum excess 0
torque exerted at the VAWT generator can be estimated. This var- 14 14.5 15 15.5 16 16.5 17 17.5 18
ies with CVT ratio. Assuming ideal transmission, this increment in Wind Speed u (in m/s)
sH is shown in Fig. 4. Importantly, this predicts the degree of over-
Fig. 4. Maximum additional design torque, assuming ideal transmission, exerted at
sizing of the VAWT generator. This prediction is site-specific with a the VAWT generator. This varies with CVT ratio r during the coupled operation of
target of utilizing torque transfer at wind speeds exceeding the the NREL S809 rotor HAWT and SAWT Inc. PK-10 VAWT turbines and comes into
Phase VI rotor’s rated wind speed. effect when uH exceeds uHB .
486 B. Govind / Applied Energy 199 (2017) 479–494

40 This far exceeds the Phase VI rotor’s performance even when the
hybrid system is operated at a suboptimal C pH of 0.1. While this
IAWT, u =17.3 m/s
optimistic estimate assumes ideal transmission, future research
into a robust control algorithm will better determine gtr derived
IAWT, u =16.3 m/s
30 HB from a friction model [43] of the CVT and intermediate right-
IAWT, u =15.3 m/s angled transmission elements.
Power (in kW)

5. A note on practical installation


Here, an initial assessment of additional capital costs for com-

NREL S809 HAWT ponents in the 20 kW HAWT – 10 kW VAWT integrated axis wind
turbine prototype system is provided (Table 2), roughly in accor-
dance with current market value.
Realization of a small-scale system is made easier by increased
expenditure on individual components to ensure a robust design.
This may be further aided by aeroelastic analyses done by compu-
0 5 10 15 20 tational researchers for tower loading such as that for the NREL
Phase VI rotor’s tower [44,45] and structural loading of a standard
Wind Speed, u (in m/s)
VAWT [46]. The tower design will be more challenging for scaled
Fig. 5. Power curves showing mechanical power vs wind speed for (a) the versions and models could be extended to predict dynamic
independent operation of the NREL Phase VI rotor rotating at an enforced xH of response of tower modes for a multi-megawatt scale integrated
72 rpm and C pH 0.1 (b) independent operation of the SAWT Inc. PK-10 VAWT axes wind turbine system. The increased breadth of the design
subsystem and (c) combined power output from the integrated system for modes of relating to power electronics and spacing of cables and ladders
operation uH 6 uHB and uH P uHB , assuming ideal transmission for three plausible
increments in uHB .
inside a scaled tower lends to a future discussion.

6. Aerodynamic feasibility
(19 m) and the mid-blade height of the VAWT (5.2 m), the suitabil-
ity of both sites can be gauged. The hourly wind profiles for the It is critical to investigate the effect of turbulence caused by one
sites are shown in Figs. 6a and 7a. turbine on the flow field of the other. Here, a computational fluid
The mechanical power drawn by the VAWT subsystem dynamics approach is used. The simulation justifies a minimum
approaches its independent rated performance at several instances clearance between the rotors for pre-stall conditions.
during the cycle. The leading distinction is the predicted mechan-
ical power output of the Phase VI rotor subsystem at the two loca- 6.1. Validation of the CFD model
tions. As shown in Fig. 6b, at Hatteras, Dare County, the original uHB
is not encountered. The net output is, therefore, a simple sum of First, the accuracy of the commercial solver Fluent 16 is vali-
power of the two independently operating subsystems. However, dated. The NREL Unsteady Aerodynamic Experiment (UAE) Phase
the combined operation has decisive effect when uH exceeds uHB VI provides a standard test case for 3D CFD HAWT rotor analysis.
at the Grandfather Mountain site, Avery County where a wind Experimental results are documented by the National Renewable
speed uH up to 22.3 m/s is recorded. As shown in Fig. 7b, this fur- Energy Laboratory. The test case was studied exhaustively by com-
nishes an opportunity to design the integrated system for three putational researchers [47–49] to validate their software and
increments in uHB . Assuming ideal transmission when the drive- improve prediction of aerodynamic performance.
train is coupled, for a new uHB of 15.3 m/s, the system yields The rotor geometry makes use of the NREL S809 airfoil, the
28.8 kW while for a heightened uHB of 17.3, it may yield 33.2 kW. geometry of which is modelled in GAMBIT. To implement a rotat-

Latitude: 35.23 o u V at VAWT mid-plane height of 5.2 m
Longitude: -75.62 u H at HAWT hub height of 19 m
Elevation: 17 ft. above sea level
Wind Speed (in m/s)



0 10 20 30 40 50 60 70 80 90

12 a.m: 29th March, 2017 Time of record (in hours) 6 p.m: 1st March, 2017

Fig. 6a. Hourly wind speed profile recorded at Hatteras Weather Service Office, Dare County, NC.
B. Govind / Applied Energy 199 (2017) 479–494 487


IAWT uHB=15.3 m/s

NREL Phase VI rotor

Mechanical Power (in kW)





0 10 20 30 40 50 60 70 80 90
12 a.m: 29th March, 2017 Time of record (in hours) 6 p.m: 1st March, 2017

Fig. 6b. Time-transient prediction of mechanical power from independent HAWT and VAWT subsystems and the integrated axes wind turbine (IAWT) for a1 m/s increment
in uHB .


Latitude: 36.10833 o u H at HAWT hub height of 19 m

Longitude:-81.8325 u V at VAWT mid-plane height of 5.2 m
25 Elevation: 5280 feet above sea level

Wind Speed (in m/s)



0 10 20 30 40 50 60 70 80 90
12 a.m: 29th March, 2017 6 p.m: 1st March, 2017
Time of record (in hours)

Fig. 7a. Hourly wind speed profile recorded at Grandfather Mountain (Experiment), Avery County, NC.

IAWT uHB =15.3 m/s
40 IAWT uHB =16.3 m/s
IAWT uHB =17.3 m/s
NREL Phase VI rotor
Mechanical Power (in kW)






0 10 20 30 40 50 60 70 80 90
12 a.m: 29th March, 2017 Time of record (in hours) 6 p.m: 1st March, 2017

Fig. 7b. Time-transient prediction of mechanical power from independent HAWT and VAWT subsystems and the integrated axes wind turbine (IAWT) for three 1 m/s
increments in uHB .
488 B. Govind / Applied Energy 199 (2017) 479–494

Table 2
Estimated additional component costs for integrating a 10 kW VAWT on the common
tower of a 20 kW HAWT (in k$).

10 kW VAWT Permanent Magnet generator and blade kit 11.0

On-grid 3 phase inverter and control unit 9.5
19-meter Mono-pole kit (Support tower) (1700 kg) 8.6
Push-belt Continuously Variable Transmission/Infinitely Variable 0.8
HAWT right-angle transmission and shaft rotation reversal gearboxes 0.6
VAWT shell motion transfer gear train 0.6
Plurality of linear hydraulic actuators and housing unit for VAWT pitch 1.2
Miscellaneous support members, bearings and cables 1.0

ing mesh methodology, sliding interfaces are defined between an

inner rotating volume (containing the rotor) and a stationary outer Fig. 8b. Computational domain and relative mesh resolution for validation of NREL
volume representing the remaining fluid domain. The influence of S809 rotor’s aerodynamic performance.
the hub, nacelle and tower on the rotor aerodynamics is negligible,
Fig. 9 shows parity of CFD predictions with experimental results
which is a fair approximation for an upwind turbine. The rotor
[47,50] after 2 revolutions. A maximum variance of 9% is observed.
radius is 5.029 m and the blades are assumed to be rigid. The the-
Therefore, this modelling is reliable and can be extended to reliably
oretical definition of the S809 profile indicates that it has a very
predict the aerodynamic performance of the integrated axes
sharp trailing edge. Fig. 8a shows the development of the blade
by the specified chord lengths, taper angles and twist angles.
In this upwind configuration, the input wind speed is a constant
6.2. CFD simulation of the integrated axes wind turbines’ operation
7 m/s with the nacelle oriented at a yaw angle of zero degrees. The
rotor rotates at an angular speed of 72 rpm. Blade surfaces are
6.2.1. Modelling the hybrid system of the HAWT and VAWT
defined as no-slip walls. Fig. 8b shows different boundaries and
The horizontal axis rotor chosen is the same 2-bladed NREL
contrast in mesh resolution in the computational domain. The
S809 rotor used in the solver’s validation. The HAWT’s nacelle is
mesh is refined in the inner cylindrical region for better resolution
of 0.7 m  0.7 m in cross section and 1 m in length.
near the rotor. Air density and viscosity are 1.225 kg/m3 and
The selection of a vertical axis rotor for an integrated system is
1:78  105 kg/(m s) respectively. The mesh consists of 2,883,965
specific design and depends on the predicted wind speed (and con-
tetrahedral elements.
sequently, T.S.R). In this model, the rotor, concentric to the tower of
The computations employ compressible Reynolds Averaged
the HAWT, has straight blades which are 4.8 m in height. Its blades
Navier-Stokes (RANS) equations, using the k  x (SST) model
have a uniform NACA 0021 profile with a chord length of 0.7 m. Its
which is recommended for small-scale turbine systems [30,31].
struts, six in number, are simplistically modelled and connect the
The tip pitch angle is set to 3°. Unsteady inflow is neglected, con-
air-foils to the rotating shell integral to the lower portion of the
sidering that the experimental data set was derived from repeated
tower. Choosing an aspect ratio (diameter/height) of 2 yields a
measurements. A time step of 0.01 s is chosen for simulation. The
rotor radius of 4.8 m.
parameter considered for validation is the pressure coefficient at
The clearance lies between the tip of the lower blade of the
varying span-lengths, given by
S809 rotor at initial 6 o’ clock position, and the plane comprising
the higher end surface of the NACA 0021 profiled blade of the ver-
P  P1 tical axis rotor. Again, the rotor zones are embedded in moving
cp ¼ ð3Þ
q½v 21 þ ðxRÞ2  mesh regions and a stationary zone represents the remainder of
the control volume.
P  P1 is the gauge pressure of the point under consideration on the The computational domain is represented in Fig. 10a. A size
surface of the blade, v 1 is free stream velocity and R is the radial function associated with the rotor’s solid surfaces effects a progres-
distance from the hub. sively increasing cell size. Fig. 10b shows the relative mesh
A summary of boundary conditions may be found in Table 3. As
a neutral atmosphere is assumed, the Boussinesq approximation is
not employed to account for buoyancy effects caused by thermal
variation. For all cases, the unsteady, time-transient simulation
was performed with a time step of 0.01 s.
A preliminary choice is made between two turbulence models.
Case 1 (a) utilizes the k-x (Shear Stress Transport) equations while
Case 1 (b) utilizes the k-e (Realizable) equations. Conditions entail-
ing equal angular speeds for rotors of comparable moments of
inertia and altitude-invariant wind profile of 7 m/s may be less
realistic [51]. However, a higher angular speed of 72 rpm for both
rotors and a lessened clearance of 4 m produce greater turbulence
in the rotors’ clearance and this helps determine the degree of
intermixing of streamlines of induced velocity in the near wake
region and further interaction downstream.
Upon establishing a convergent model, performance is analysed
Fig. 8a. Development of NREL S809 airfoil by successively tapered and twisted at two expected operating points. In effect, the same VAWT geom-
profiles. etry is used but different angular speeds/power output are consid-
B. Govind / Applied Energy 199 (2017) 479–494 489

-6 -6
63.3 % Span Length Solver (Fluent 16) 80 % Span Length Solver (Fluent 16)
-5 NREL Experimental Data NREL Experimental data

-4 -4

Coefficient of Pressure
Coefficient of Pressure

-3 -3

-2 -2

-1 -1

0 0

1 1

2 2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Fig. 9. Comparison of chord-wise (x-position/chord-length) pressure coefficients at 2 stations along the span (63.3% and 80%). Shown here is the comparison between cp
obtained numerically using Fluent 16 and that measured by NREL experimental results at after 2 revolutions of the Phase VI rotor.

Fig. 10a. Computational Domain for Cases 1 and 2.

Fig. 10b. Relative grid resolution in the computational domain.

ered. In Case 2 (a), the VAWT rotates at 45 rpm (CVT ratio 6.3. Results and discussion of turbulence modelling of the integrated
r ¼ 0:625) and in Case 2 (b), a VAWT rotates at a lower speed of axes wind turbine
36 rpm (CVT ratio r ¼ 0:5). The HAWT has an increase of 2.4 m in
hub height, while its speed is a steady 72 rpm. The free stream A neat convergence at each time-step is achieved after approx-
wind speed is increased to 9 m/s. As the choice of a vertical axis imately 25 iterations. A qualitative analysis of the simulation of the
rotor is dependent on site location and requirement to integrate rotors’ combined operation in the flow field now follows.
with a micro-scale grid, an independent power rating range of 5– The parameter used for visualization of effects of vorticity is
10 kW for the vertical rotor is targeted. Turbulence kinetic energy (TKE). TKE is the mean kinetic energy
490 B. Govind / Applied Energy 199 (2017) 479–494

Table 3
Summary of Cases 1 and 2.

Case 1 (a) Case 1 (b) Case 2 (a) Case 2 (b)

Turbulent Model used k  x (Shear Stress Transport) k  e (Realizable) k  x (Shear Stress Transport)
Boundary Conditions  v 1 = 7 m/s  v 1 = 9 m/s
 Rotor clearance of 4 m  Rotor clearance of 6.4 m
 xH ; xV = 72 rpm  xH = 72 rpm  xH = 72 rpm
 xV = 45 rpm  xV = 36 rpm
Domain Size (x, y and z dimension) 117 m  49 m  80 m
Mesh strength 2,583,400 polyhedral cells (varying resolution) 2,640,669 polyhedral cells (varying

per unit mass associated with eddies in turbulent flow and is char- either rotor does not appreciably affect the other’s operation for
acterised by measured root-mean-square (RMS) velocity fluctua- a free stream speed of 7 m/s.
tions. TKE can be produced by fluid shear, friction or buoyancy, In Cases 2 (a) and (b) which use the preferred k  x (SST)
or through external forcing at low-frequency Eddie scales [51]. This model, an expanded wake is formed behind both rotors in their
energy is transferred down the turbulence energy cascade, and is region of induced velocity. Again, the initial low-pressure circula-
dissipated by viscous forces at the Kolmogorov scale. tion is not sustained and is carried downstream. Fig. 12 indicates
In the integrated axes wind turbine model, TKE is measured in that after 15 s of flow time, a quasi-steady flow regime is estab-
relation to the centre-lines of three subsystems extending through lished, albeit weak intermixing in the far wake region. With an
the length (x-positions) of the computational domain and which lie increased wind speed of 9 m/s, an increase in 2.4 m in clearance
in the x-y plane: permits the HAWT rotor to operate under little influence of the
VAWT rotor and this may be considered a safe configuration.
(1) The HAWT subsystem whose equatorial centre-line passes
through the origin of the domain and coincidentally, the 6.4. Mechanical power estimation and effects of VAWT torque ripple
centre of the nacelle. on the combined operation
(2) The rotor clearance, the equatorial line of which passes
though the tower at a point 7 m below the nacelle for Cases Having established the hybrid system’s aerodynamic feasibility,
1 (a) and (b) and 8.2 m below the nacelle for Cases 2 (a) and a straightforward prediction of instantaneous mechanical power
(b). from each rotor can now be made. A product of the aerodynamic
(3) The VAWT subsystem whose equatorial centre-line passes torque on each turbine about their relevant axes, as reported by
through a point 11.4 m below the nacelle, for Cases 1 (a) the validated solver, and their respective angular speeds, i.e.
and (b) and 13.8 m below the nacelle for Cases 2 (a) and (b). 7.53 rad/s for the S809 rotor and 4.71 rad/s (Case 2 (a)) and
3.76 rad/s (Case 2 (b)) for the NACA 0021 rotor, is taken. This out-
Fig. 11 shows the variation of Turbulent Kinetic Energy along put, relative to flow time is shown in Fig. 13(a) and (b).
these centre-lines for two instances of flow time for Cases 1 (a) Interestingly, when a quasi-steady flow regime is achieved, the
and (b). The simulation also helps visualize if the streamlines indi- time-response of mechanical power cases reveal that HAWT’s per-
cate a mutual interaction in the near and far wake regions of the formance is relatively unaffected by the VAWT’s rotation when
hybrid system. The wake development is shown by means of veloc- given a rotor clearance of 6.4 m. This is evident from Fig. 13
ity contours plotted on iso-surfaces spaced at consecutive dis- (a) and (b). Further, its independent output in the neighbourhood
tances of 1 m in the computational domain. of 11 kW correlates reasonably with experimental and predicted
An initial interaction of the two near wakes is manifested as an power curves [40] of the Unsteady Aerodynamics Experiment
intermixing and circulation of velocity vectors. This circulation is a Phase VI.
temporary phenomenon which is transported downstream and The violent variation in axial moment exerted on the VAWT
diminishes with progressing time. The near wake behaves akin a typical of this H-rotor type design will also cause its characteristic
body of revolution reducing the effect of the free stream which torque ripple effect [52] on its generator during its practical real-
diverges around it and converges again. The tower shadow effect ization. This is the natural outcome of integrating torque produced
does not bear significant aerodynamic influence. As Fig. 11 shows, by aerodynamic loads due to its rotating blades’ interaction with
both k  x (SST) and k  e (Realizable) models predict similar steady wind.
trends in far wake regions. The k  e (Realizable) model, however, Case 2 (b) is instructive as it shows that enforcing xV at
has stability issues due to numerical stiffness when flow is sepa- 36 rpm is hardly profitable (see Fig. 13(b)). It effects an averaged
rated. The k  e model performs poorly for the complex flow drop of 8 kW (40 percent drop in efficiency) in net output of the
involving steep pressure gradient and change in streamline curva- integrated system’s output from that obtained at xV of 45 rpm in
ture near the tower. On the other hand, the k  x equation (SST) Case 2 (a) (see Fig. 13(a)). The compensated torque per cycle
model more accurately predicts vorticity at small magnitudes shows that energy is not harvested for a portion of its rotation
and time scales and proves convenient near blade surfaces. It is, and for a small period, the VAWT behaves akin a propeller that
therefore, selected for pre-stall conditions in Cases 2 (a) and (b). adds energy to the free stream. Also, after12 s of flow time, net
The development of downstream-wake region shows that as saV drops sharply to produce a near –zero (a mean of
time progresses, there is lessened effect by either turbine on the 0.13 kN m) positive moment.
mean wind speed in the intermediate clearance (IAWT centre-
line). The only major variance in vorticity and TKE is only encoun- 7. Conclusions and scope of future work
tered at the position of the tower which presents an obstacle (x-
position equals zero). The TKE is found to have significantly The design of an integrated axes wind turbine has been pro-
reduced after 10 revolutions of both turbines. Visibly, after a posed and discussed. Its novel drivetrain involves the coupling of
quasi-steady flow regime is reached, the circulation caused by a horizontal axis wind turbine and a vertical axis wind turbine at
B. Govind / Applied Energy 199 (2017) 479–494 491

HAWT center -line
IAWT center-line HAWT center -line
0.5 VAWT center-line IAWT center-line
0.5 VAWT center-line

T.K.E (m /s )

T.K.E (m2/s2)
2 2


0.2 0.2

0.1 0.1

0 0
-40 -20 0 20 40 60 -40 -20 0 20 40 60

X-Position (m) X-Position (m)

Case 1 (a): Time of flow = 6 seconds (7.2 revolutions) Case 1 (a): Time of flow = 10 seconds (12 revolutions)

HAWT center -line HAWT center -line
2 IAWT center-line
IAWT center-line
VAWT center-line VAWT center-line

T.K.E (m 2/s 2)

T.K.E (m 2/s 2)





0 0
-40 -20 0 20 40 60 -40 -20 0 20 40 60
X-Position (m) X-Position (m)
Case 1 (b): Time of flow = 6 seconds (7.2 revolutions) Case 1 (b): Time of flow = 10 seconds (12 revolutions)

Fig. 11. Time-transient wake development for a rotor clearance of 4 m (v 1 = 7 m/s) at instances of 6 s and 10 s of flow time. Above: Velocity contours plotted on iso-surfaces
separated by 1 m in the computational domain. Below: Turbulent Kinetic Energy comparison along centre-lines through the length of the control volume (x-position) for
k  x (SST) model (Case 1 (a)) and k  e (Realizable) model (Case 1 (b)). The latter has stability issues.

high wind speeds. This facility enhances the existing capability of a in rated wind speed was estimated. The change in the hybrid
prototype scale HAWT while simultaneously producing power power curve reveals that the coupled system would yield power
from a tower-mounted VAWT. Further, it provides a source of far exceeding the individual capacity of either turbine subsystem.
high-density wind energy conversion. A generic mode of operation Predictions for the small-scale system show, for example, a poten-
to facilitate the torque transfer from the horizontal-axis rotor to tial nominal power of up to 33.5 kW for a 3 m/s increase in rated
the drivetrain of the vertical–axis rotor has also been presented. wind speed when the S809 rotor HAWT is sustained at even a
To investigate its mechanical power capabilities, two potential low power coefficient of 0.1 after it reaches independent rated
wind sites were selected and the output for plausible increments power output.
492 B. Govind / Applied Energy 199 (2017) 479–494

IAWT center-line
HAWT center -line
HAWT center-line
IAWT center-line
VAWT center-line 0.5
0.5 VAWT center-line

0.4 0.4

T.K.E (m /s )
2 2
T.K.E (m /s )
2 2

0.3 0.3

0.2 0.2

0.1 0.1

0 0
-40 -20 0 20 40 60 -40 -20 0 20 40 60
X-Position (m) X-Position (m)
Case 2 (a): Time of flow = 7.5 seconds Case 2 (a): Time of flow = 15 seconds

0.6 0.6
IAWT center-line
HAWT center-line HAWT center -line
VAWT center-line IAWT center-line
0.5 0.5 VAWT center-line

0.4 0.4
T.K.E (m /s )

T.K.E (m /s )
2 2

2 2

0.3 0.3

0.2 0.2

0.1 0.1

0 0
-40 -20 0 20 40 60 -40 -20 0 20 40 60
X-Position (m) X-Position (m)

Case 2 (b): Time of flow = 7.5 seconds Case 2 (b): Time of flow = 15 seconds

Fig. 12. Time-transient wake development for a rotor clearance of 6.4 m and v 1 = 9 m/s, using a k  x (SST) model. Above: Turbulent Kinetic Energy comparison along
centre-lines through the length of the control volume (x-position) for Case 2 (a). Below: Turbulent Kinetic Energy comparison along centre-lines through the length of the
control volume (x-position) for Case 2 (b) for two instances (7.5 s and 15 s) of flow time.

Turbulence modelling of the combined operation was used to Future work would include examining the cyclical variation in
confirm the aerodynamically independent operation of the NREL vertical axis rotor’s speed to determine the sensitivity of a suit-
Phase VI rotor and a small manufacturer’s 10 kW VAWT used in able controller. If used in a scaled version, strength of the
combination. Permutations of wind speed conditions and angular mechanical linkage to the generator depends on such harmonic
speed of the two rotors assist in establishing safe configurations aerodynamic torque fluctuations. Here, the smoothening of tran-
with varying hub height by using Turbulent Kinetic Energy as a sient response caused by torque ripple by the VAWT is necessary
parameter to find a safe rotor clearance. The model also helps for synchronous power generation from the rotors and their on-
pre-determine the aerodynamic torque on both rotors in a quasi- grid integration. Further, a complete friction model will be for-
steady regime. The k  x (SST) model proved more stable for mulated to accurately determine the losses in transmission in
pre-stall conditions. It allays concerns of possible mutual influence the coupled drivetrain and CVT. A practical fabrication of the pro-
of either rotors’ rotation on the other’s torque production. In prac- totype scale integrated system will be undertaken shortly. This
tice, it would behove the designer to consider an accurate wind- will determine, by extensive testing, the suitable sliding-mode
speed profile from the ground up while implementing the recom- control logic for optimal power point tracking of hybrid modes
mended system. The effect of torque ripple is predominant. As sus- for various Tip Speed Ratios. While simulation implies that
pected, not all angular speed ratios are profitable. In one case, a small-scale, standard rotors and blade profiles for a prototype
decrease of only 12 percent in the H-rotor VAWT’s angular speed system are profitable, the concept should bear fruition in a
causes an overall decrease of 8 kW of the integrated system during multi-MW setting and could re-power aging HAWTs using scaled
the two rotors’ normal, aerodynamically independent operation. Darrieus VAWT designs.
With the current design, it is indicated that the independent oper- Objectively, this hybrid design would greatly increase net
ation of both rotors is relatively unaffected by their superimposi- swept area of a conventional wind turbine and is assisted by
tion on the same tower. This provides a guideline for designing the two-ended torque transfer that helps increase rated wind
the rotating shell’s pitch mechanism to maintain its angular speed speed.
above predetermined thresholds.
B. Govind / Applied Energy 199 (2017) 479–494 493

30 Appendix B. Supplementary material

VAWT (NACA 0021) subsystem-45 rpm
HAWT (NREL S809) subsystem-72 rpm
25 Integrated axes wind turbine Supplementary data associated with this article can be found, in
the online version, at
Mechanical Power (in kW)

15 References

[1] McKenna R, Ostman P, Leye VD, Fichtner W. Key challenges and prospects for
large wind turbines. Renew Sustain Energy Rev 2016;53:1212–21.
[2] Kumar Dipesh, Chatterjee Kalyan. A review of conventional and advanced
MPPT algorithms for wind energy systems. Renew Sustain Energy Rev
5 2016;55:957–70.
[3] Petković Dalibor, Nikolić Vlastimir, Mitić Vojislav V, Kocić Ljubiša. Estimation
of fractal representation of wind speed fluctuation by artificial neural network
0 with different training algorithms. Flow Meas Instrum 2017;54:172–6.
[4] Shamshirband Shahaboddin, Petković Dalibor, Amini Amineh, Anuar Nor
Badrul, Nikolić Vlastimir, Ćojbašič Žarko, et al. Support vector regression
-5 methodology for wind turbine reaction torque prediction with power-split
0 5 10 15
hydrostatic continuous variable transmission”. Energy 2014;67:623–30.
Time (seconds) [5] Petković Dalibor, Ćojbašič Žarko, Nikolić Vlastimir. Adaptive neuro-fuzzy
(a) approach for wind turbine power coefficient estimation. Renew Sustain Energy
Rev 2013;28:191–5.
30 [6] Petković Dalibor, Ćojbašič Žarko, Nikolić Vlastimir, Kiah Miss Laiha Mat, Anuar
Nor Badrul, Wahab Wahid Abdul. Adaptive neuro-fuzzy maximal power
HAWT (NREL S809) subsystem-72 rpm
VAWT (NACA 0021) subsystem-36 rpm
extraction of wind turbine with continuously variable transmission. Energy
25 Integrated axes wind turbine 2014;64:868–74.
[7] Chong WT, Gwani M, Shamshirband S, Muzammil WK, Tan CJ, Fazlizan A, et al.
Application of adaptive neuro-fuzzy methodology for performance
investigation of a power-augmented vertical axis wind turbine. Energy
Mechanical Power (in kW)

[8] Shamshirband Shahaboddin, Ćojbašič Žarko, Nikolić Vlastimir, Anuar Nor

Badrul, Shuib Nor Liyana Mohd, Kiah Miss Laiha Mat, et al. Adaptive neuro-
fuzzy optimization of wind farm project net profit. Energy Convers Manage
[9] Petković Dalibor, Pavlović Nenad T, Ćojbašić Žarko. Wind farm efficiency by
10 adaptive neuro-fuzzy strategy. Int J Electr Power Energy Syst 2016;81:215–21.
[10] Nikolić Vlastimir, Sajjadi Shahin, Petković Dalibor, Shamshirband
Shahaboddin, Ćojbašić Žarko, Por Lip Yee. Design and state of art of
5 innovative wind turbine systems. Renew Sustain Energy Rev 2016;61:258–65.
[11] Petković Dalibor, Shamshirband Shahaboddin, Anuar Nor Badrul, Saboohi
Hadi, Wahab Ainuddin Wahid Abdul, Protić Milan, et al. An appraisal of wind
0 speed distribution prediction by soft computing methodologies: a
comparative study. Energy Convers Manage 2014;84:133–9.
[12] Indragandhi V, Subramaniyaswamy V, Logesh R. Resources, configurations,
-5 and soft computing techniques for power management and control of PV/wind
0 5 10 15 hybrid system. Renew Sustain Energy Rev 2017;69:129–43.
Time (seconds) [13] Song Dongran et al. Wind estimation with a non-standard extended Kalman
filter and its application on maximum power extraction for variable speed
(b) wind turbines. Appl Energy 2017;190:670–85.
[14] Kusiak Andrew, Li Wenyan, Song Zhe. Dynamic control of wind turbines.
Fig. 13. Transient predicted mechanical power for HAWT, VAWT subsystems and Renewable Energy 2010;35:456–63.
[15] Yin Xiu-xing, Lin Yong-gang, Li Wei, Gu Ya-jing, Liu Hong-wei, Lei Peng-fei. A
integrated axes system when xH = 72 rpm. (a) Profitable combined output when
novel fuzzy integral sliding mode current control strategy for maximizing
xV = 45 rpm. (b) Near zero independent VAWT contribution when xV = 36 rpm.
wind power extraction and eliminating voltage harmonics. Energy
[16] Yin Xiu-xing, Lin Yong-gang, Li Wei, Liu Hong-wei, Gu Ya-jing. Adaptive sliding
Acknowledgements mode back-stepping pitch angle control of a variable-displacement pump
controlled pitch system for wind turbines. ISA Trans 2015;58:629–34.
[17] Gu Ya-jing, Yin Xiu-xing, Liu Hong-wei, Li Wei, Lin Yong-gang. Fuzzy terminal
The author would like to acknowledge the discussions and sliding mode control for extracting maximum marine current energy. Energy
provision of computational resources at North Carolina State 2015;90:258–65.
[18] Cherubini Antonello et al. Airborne wind energy systems: a review of the
University, Raleigh and help with explanatory sequences by
technologies. Renew Sustain Energy Rev 2015;51:1461–76.
Er. Thắng Nguyễn, Hanoi. [19] Goldstein Leo. Theoretical analysis of an airborne wind energy conversion
system with a ground generator and fast motion transfer. Energy
Appendix A. Explanatory Animation Sequences [20] David Weston. Vestas four-rotor turbine in action.
Published on: 1 Sep., 2016 [accessed: 22 Dec., 2016].
[21] Wiser Ryan et al. Expert elicitation survey on future wind energy costs. Nat
Sequences 1 (a) and (b) Shell-VAWT’s motion transfer Energy 2016;1(10):16135.
mechanism [22] Yang Zhimin, Chai Yi. A survey of fault diagnosis for onshore grid-connected
converter in wind energy conversion systems. Renew Sustain Energy Rev
Sequence 2 Shell-VAWT’s pitch mechanism 2016;66:345–59.
Sequence 3 Simultaneous yawing and motion [23] Yin Xiu-xing, Lin Yong-gang, Li Wei, Gu Hai-gang. Hydro-viscous transmission
transfer based maximum power extraction control for continuously variable speed
wind turbine with enhanced efficiency. Renewable Energy 2016;8:646–55.
Sequence 4 Simplified shaft rotation reversal [24] Yin Xiu-xing, Lin Yong-gang, Li Wei, Liu Hong-wei, Gu Ya-jing. Output power
mechanism control for hydro-viscous transmission based continuously variable speed
wind turbine. Renewable Energy 2014;72:395–405.
494 B. Govind / Applied Energy 199 (2017) 479–494

[25] Giallanza A et al. Analysis of the maximization of wind turbine energy yield [39] Hand M, Simms D, Fingersh L, Jager D, Cotrell J, Schreck S, et al. Unsteady
using a Continuously Variable Transmission system. Renewable Energy aerodynamics experiment phase VI: wind tunnel test configurations and
2017;102:481–6. available data campaigns. National Renewable Energy Laboratory/TP-500-
[26] Peter Fairley. Testing cheap wind power. MIT’s Technology Review. Published: 29955; 2001.
29 Oct., 2009. L. [accessed: 29 December, 2016]. [40] Giguère P, Selig M. Design of a tapered and twisted blade for the NREL
[27] Li Qing’an et al. Effect of solidity on aerodynamic forces around straight- combined experiment rotor. National Renewable Energy Laboratory/SR-500-
bladed vertical axis wind turbine by wind tunnel experiments (depending on 26173; 1999.
number of blades). Renewable Energy 2016;96:928–39. [41] Qiang Yan. Vertical axis wind turbine and method of installing blades therein.
[28] Sutherland B, Ashwill. A retrospective of VAWT technology. Sandia National US Patent- 8322035 B2; 2012.
Laboratories/2012-0304; 2012. [42] Web resource: Product: pk10 [accessed: 15 September,
[29] Shaheen Mohammed, Abdallah Shaaban. Development of efficient vertical axis 2016].
wind turbine clustered farms. Renew Sustain Energy Rev 2016;63:237–44. [43] Duan Chengwu, Hebbale Kumar, Liu Fengyu, Yao Jian. Physics-based modeling
[30] Rocha P, Costa A, et al. A case study on the calibration of the k  x SST (Shear of a chain continuously variable transmission. Mech Mach Theory
Stress Transport) turbulence model for small scale wind turbines designed 2016;105:397–408.
with cambered and symmetrical airfoils. Energy 2016;97:144–50. [44] Lee Kyoungsoo, Huque Ziaul, Kommalapati Raghava, Han Sang-Eul. Evaluation
[31] Rocha P, Costa A, et al. k  x SST (Shear Stress Transport) turbulence model of equivalent structural properties of NREL phase VI wind turbine blade.
calibration: a case study on a small scale horizontal axis wind turbine. Energy Renewable Energy 2016;86:796–818.
2014;65:412–8. [45] Wang Lin, Liu Xiongwei, Kolios Athanasios. State of the art in the aeroelasticity
[32] Paul van der Laan M, Sørensen Niels N, Réthoré Pierre-Elouan, Mann Jakob, of wind turbine blades: aeroelastic modelling. Renew Sustain Energy Rev
Kelly Mark C, Troldborg Niels, et al. An improved k-e model applied to a wind 2016;64:195–210.
turbine wake in atmospheric turbulence. Wind Energy 2015;18:889–907. [46] Berg D. Structural design of the Sandia 34-meter vertical-axis wind
[33] Govind, Gurumurthy. System and method for integrating a horizontal axis turbine. Sandia National Laboratories/84-1287; 1985.
wind turbine and a vertical axis wind turbine. US Patent appl. publication - [47] Elfarra Monier A, Sezer-Uzol Nilay, Sinan Akmandor I. NREL VI rotor blade:
2016/0201652 A1; 2016. numerical investigation and winglet design and optimization using CFD. Wind
[34] Mabrouk Imen Bel et al. Dynamic vibrations in wind energy systems: Energy 2014;17:605–26.
application to vertical axis wind turbine. Mech Syst Signal Process [48] Huang Juan-Chen, Lin Herng, Hsieh Tsang-Jen, Hsieh Tse-Yang. Parallel
2017;85:396–414. preconditioned WENO scheme for three-dimensional flow simulation of
[35] Li Qing’an et al. Wind tunnel and numerical study of a straight-bladed vertical NREL Phase VI Rotor. Comput Fluids 2011;45:276–82.
axis wind turbine in three-dimensional analysis (Part II: For predicting flow [49] Duque E, Burklund M, Johnson W. Navier-stokes and comprehensive analysis.
field and performance). Energy 2016;104:295–307. Performance predictions of the NREL phase VI experiment. American Institute
[36] Asr Mahdi Torabi, Nezhad Erfan Zal, Mustapha Faizal, Wiriadidjaja Surjatin. of Aeronautics and Astronautics/2003–355.
Study on start-up characteristics of H-Darrieus vertical axis wind turbines [50] Giahi M, Dehkordi A. Investigating the influence of dimensional scaling on
comprising NACA 4-digit series blade airfoils. Energy 2016;112:528–37. aerodynamic characteristics of wind turbine using CFD simulation. Renewable
[37] Giallanza A, Porretto M, Cannizzaro L, Marannano G. Analysis of the Energy 2016;97:162–8.
maximization of wind turbine energy yield using a continuously variable [51] Pope SB. Turbulent flows. Cambridge UP; 2000.
transmission system. Renewable Energy 2017;102:481–6. [52] Reuter Robert O. Torque ripple in Darrieus, vertical axis wind turbine. Sandia
[38] Giguère P, Selig M. Design of a tapered and twisted blade for the NREL National Laboratories/80-0475; 1980.
Combined Experiment Rotor. National Renewable Energy Laboratory/SR-500-
26173; 1999.