Engineering Structures 100 (2015) 564–576

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

A simplified dynamic model for mast design of H-Darrieus vertical axis

wind turbines (VAWTs)
E. Verkinderen a,1, B. Imam b,⇑
a
Price & Myers, 30 Newman Street, London W1T 1LT, United Kingdom
b
Department of Civil and Environmental Engineering, Faculty of Engineering and Physical Sciences, Guildford, Surrey GU2 7XH, United Kingdom

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

Article history: The use of small scale vertical axis turbines (VAWTs) in the urban environment is becoming increasingly
Received 15 October 2013 popular. VAWTs are usually supported on steel masts to increase their efficiency. The resonance between
Revised 18 June 2015 the supporting masts and the VAWTs can disturb the power generation and trigger potential fatigue
Accepted 19 June 2015
issues in the mast structure. As the turbine itself influences greatly the free vibration modes of the mast,
Available online 3 July 2015
VAWTs should not be modelled as a single lumped mass during a dynamic analysis. A simplified analyt-
ical model of an H-Darrieus VAWT is developed to allow the dynamic analysis of the supporting masts
Keywords:
taking into account the coupling of the mast and turbine. The VAWT has a height of 5 m, a diameter of
Mast–turbine system
Frequencies
3 m and a power output of 6 kW whereas the height of the masts ranges between 3 and 15 m. The cou-
FE analysis pled structure is analysed both analytically, through a multi-degree of freedom system, as well as numer-
Multi-degree-of-freedom model ically through the finite element (FE) method. The correlation of the estimated natural frequencies of the
Dynamic analysis system through the FE method with site measurements was found to be within 10% and can allow
designing adequately the mast. Harmonic forces representative of the turbine out of balance and cyclic
aerodynamic forces are proposed and used to predict peak acceleration at resonance.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction
There has been an increased interest in vertical axis wind tur-
bines (VAWTs) in the last decade or so. The use of VAWTs is better
suited to the urban environment because their design allows a
more turbulent wind profile. Most of the VAWTs installed recently
are of small size. For example, since 2007, there have been more
than 500 small VAWTs, having power output ranging between
1.5 and 50 kW, installed across the UK [1]. The largest vertical axis
wind turbine, located in Cap Chat, Quebec, Canada, is 110 m tall
and has a 4 MW capacity. Tall buildings are suitable to locate wind
turbines which have a low visual impact and reduced acoustic
emission. The electricity can be used directly by the building users
without any loss arising from its transportation. Wind turbines can
be used successfully to reduce the CO2 emission rate of buildings in
the same manner as other sources of sustainable energy such as
solar panels or heat pumps. The energy efficiency is required by
Building Control for new build projects and large extensions.
⇑ Corresponding author at: Department of Civil and Environmental Engineering,
Faculty of Engineering and Physical Sciences (C5), University of Surrey, Guildford,
Surrey GU2 7XH, United Kingdom. Tel.: +44 1483 689679; fax: +44 1483 450984.
E-mail address: b.imam@surrey.ac.uk (B. Imam).
1
Present address: AKT II Ltd, 100 St Johns St., London EC1 M 4EH, United Kingdom.
Wind turbines are an effective answer to buildings with low energy
design.
VAWTs are usually supported on steel masts to increase their
efficiency. Previous research efforts have concentrated either on
the behaviour of the supporting steel masts alone or on the aerody-
namics of the wind turbines. There is hardly any research or stan-
dards to provide guidance on how to assess the vibrations induced
by VAWTs on their supporting masts. This paper presents a study
of the vibrations induced by VAWTs onto their supporting mast
or towers where both analytical and numerical models and tools
which can be used by VAWTs manufacturers are developed.
Resonance between the mast and the turbine must be avoided so
that the power generation is optimum and not disturbed. The
important dynamic aspects during the design of masts are the fati-
gue of welded elements and potential excessive displacements at
the top of the mast. These are not associated with higher modes,
which can be technically challenging to predict and require
accounting for the rotation of the rotor and potential centrifugal
forces. The model presented in this paper accounts for the lower
modes and aims at being used by mast fabricators to quickly assess
potential fatigue issues and vertical turbine fabricators to assess
serviceability issues if the mast stiffness is not adequate. The
potential of tower vibrations and resonance due to the dynamic
behaviour of the turbine is mentioned in [2] without, however,

http://dx.doi.org/10.1016/j.engstruct.2015.06.041
0141-0296/Ó 2015 Elsevier Ltd. All rights reserved.
 E. Verkinderen, B. Imam / Engineering Structures 100 (2015) 564–576
any detailed quantitative prescriptions and this paper partly
addresses this gap.
2. Background studies
The majority of the research available on vertical and horizontal
axis wind turbines is focused on the wind turbine itself; even if the
issue of the vibration induced by the out of balance forces is iden-
tified, it is customarily assumed that very stiff supporting structure
and cables are used [3]. However, a VAWT is usually required to be
supported on a mast or a grillage so that it can reach undisturbed
high wind levels and ensure safety for the public.
Wind turbines are flexible structures with aerodynamic forces
acting on the blades at harmonics of the turbine rotational fre-
quency. The prediction and measurement of wind turbine modal
frequencies is very important in the design of wind turbines to
avoid resonance and fatigue and for their successful operation
[4,5]. Usually, the vibration modes of the turbines are analysed
independently of the mast structure. The main criterion to avoid
excitation issues is to have the natural frequencies of the mast
not too close from the natural frequency of the rotor [6].
Typical finite element (FE) analyses for horizontal axis wind tur-
bine (HAWT) towers have been carried out in [7], as shown in
Fig. 1. The simplified mass model (d) is sufficient to calculate the
forces and natural frequencies as long as the natural frequencies
of the tower and blades are far apart [8]. In this case, the tower
has been modelled as a lumped mass multi degree of freedom
(MDOF) flexible entity which includes one lumped mass at the
top of the tower which allows for the mass of the blades and
nacelles. Model (c) takes into account the soil–structure interac-
tion but the soil effects were calculated to be less than 2%.
Refined plate elements model (b) is used to calculate stress con-
centrations. The loading was simplified as being a set of forces
and moments associated with a number of occurrences applied
at the top of the mast. Because of high wind load at the top of
the tower, vortex shedding is usually not the critical factor for this
tower type.
The majority of the research carried out recently on the
dynamic modelling of wind turbine systems focuses on the blades
and nacelle components. The tower is usually less complex and, in
most of the cases, it was seen as sufficient to ensure, by modifying
the stiffness of the tower, that the tower natural frequency did not
match a multiple of the rotational frequency of the blades.
However, with the increasing size of HAWTs, it is becoming neces-
sary to allow for the dynamic coupling of the tower and the rotat-
ing blades. An analytical Lagrangian model taking into account the
kinetic energy and centrifugal stiffening can be developed to estab-
lish a representative time history of the tower tip displacement [9].
In the 1980s, the Natural Excitation Technique (NexT) was used
for acquiring the modal parameters using output only model test-
ing [10]. The method allows structures to be tested in their ambient
environments and the resulting modal frequencies and damping
ratios are then extracted from measured response data through a
time-domain modal identification scheme employing
cross-correlations functions. In the 1990s, Operational Modal
Analysis (OMA) was developed to overcome the challenge of testing
large structures excited with natural environmental inputs [11].
OMA can employ either time history-based and/or frequency
domain-based techniques. Most of the research associated with
OMA focused on the VAWTs of Cap Chat in Canada [5]. One poten-
tial complication for the wind turbines is the fact that their modal
frequencies depend on the rotation speed of the turbine. The modal
frequencies vary with the turbine rotation speed because of the
effects of tension stiffening and centrifugal and Coriolis speed.
However, these effects are more severe at very high speed, above
565
400 rpm for VAWT. In the example of the 2 m Sandia VAWT
described [5], the modal frequencies of the turbine alone vary up
to approximately 15% at 200 rpm and then dramatically at 800 rpm.
An OMA has been used for a rotating VAWT and the different
frequencies and associated modes were successfully extracted
[5]. The analytical solution of the structural dynamic characteris-
tics of VAWTs has been developed to accurately predict the start-
ing torque characteristics and blade vibratory stresses [12].
Recently, an article describing the aerodynamic load measure-
ments on the airfoils of VAWTs has been published [13]. High
amplitude vibration of the turbine was observed, primarily due
to the resonance of the whirling mode of the turbine. The goal of
the study was to validate a VAWT numerical aerodynamic model.
However, because the level of turbine vibration was in some cases
dominating completely the underlying aerodynamic loading, a way
to minimise or eliminate the vibration response was developed by
restraining the top of the turbine. Vibration should be minimised
to reduce the potential noise impact and potential fatigue failure.
Adding a guy wire system at the top of the turbine increases the
primary structural mode because of the increased energy required
to excite the primary mode of the fixed-pinned shaft [13].
The computational fluid dynamics (CFD) model developed in
[13] can be used to determine the thrust and radial loading charac-
teristics for one blade depending on the angle of rotation and for
different blade speeds. It was shown that the resultant force is har-
monic in both the stream-wise and cross-stream direction. The pri-
mary vibration response of the turbine was found to be coincident
with the blade pass frequency.
A number of recent studies have concentrated on the investiga-
tion of the performance of VAWTs under low blade speeds and
have shown that it can be feasible to efficiently use them under
such operating conditions [14–16]. On the other hand, other stud-
ies have focussed on the analysis of the aerodynamics and perfor-
mance of VAWTs through (CFD) simulations [17–19], FE structural
analysis [20] as well as experimental testing [19,21] and on the
optimisation of the design of the turbine itself [22,23].
3. The Quiet Revolution qr5 vertical axis wind turbine
Quiet Revolution has recently successfully developed a 5 m tall
VAWT [24]. The profile of the three helical blades has been opti-
mised for performance and quite operation. The low tip speed fur-
ther reduces noise levels. The rotor size is 5 m tall and of 3.1 m in
diameter having a total mass of approximately 450 kg. The blades
are made of carbon and glass fibre. The 6 kW turbine can generate
up to 7500 kW h per year depending on the average site wind
speed. The turbines are mounted on tilt down masts. The mast
standard heights are 3 m, 6 m, 9 m and 15 m. A standard arrange-
ment on a 6 m mast is shown in Fig. 2. The masts are made of sev-
eral conical sections rammed into each other.
The turbine applies some maximum forces to the mast which
have been determined by wind tunnel testing. Under a maximum
wind speed of 55 m/s = 200 km/h, the lateral wind force (in the
x-direction, see Fig. 2) generated in the middle of the rotor is
5 kN and the vertical load (in the z-direction, see Fig. 2) is 4.4 kN.
The lateral force produces an overturning moment on the mast
which is the critical loading driving its static design. The helically
swept blades of the Quiet Revolution wind turbine reduce the vari-
ation in lateral loading which was a typical issue with the Darrieus
Turbine as discussed in [13].
4. Analytical modelling of the mast and turbine system
The derivation of the stiffness of the mast and the coupled mast
and turbine system is required for analytical calculation of the
566 E. Verkinderen, B. Imam / Engineering Structures 100 (2015) 564–576

Fig. 1. Typical HAWT mast analysis models [7].

3.1m

Fig. 2. qr5 setting out on a 6 m mast.

vibration response. Once the stiffness of the system is determined,
the equation of motion can be formulated and the free vibration of
the system derived. In the following sections, different analytical
models of the system are developed and their suitability in captur-
ing the fundamental dynamic characteristics of the system is
investigated by comparison of the analytical results with field
measurements carried out by Quiet Revolution.
4.1. Stiffness of a cantilever beam
The mast is tapering down in diameter, as shown in Fig. 3 and
therefore the basic expression to calculate the stiffness, deflection
and natural frequencies of a cantilever beam are not applicable.
The tapered beam deflection under a point load F at the tip can
 E. Verkinderen, B. Imam / Engineering Structures 100 (2015) 564–576 567

be found by the integration of the second order ordinary differen-

tial equation of the moment curvature [25]
E  IðxÞ  u00 ðxÞ ¼ MðxÞ ð1Þ
where E is the Young’s Modulus, I is the inertia, u is the displace-
ment and M is the bending moment. The inertia is

IðxÞ ¼ p  RðxÞ3  t ð2Þ

where R is the radius of the mast and t is the thickness of the mast
tube. By defining a = (R1  R0)/(R0L) and K = (R1  R0)/R0, the
expression for the radius can be written as
 x
RðxÞ ¼ R0  ð1 þ axÞ ¼ R0  1 þ K ð3Þ
L
By integrating Eq. (1) and using the boundary conditions Fig. 3. Tapered mast deflection.
u0 ðLÞ ¼ 0 and uðLÞ ¼ 0, the solution for the deflection becomes
2  
  3
1þK
ln 1þKx=L ð1  x=LÞ 1 þ 3
K þ Kx 1
þ K Table 1
uðxÞ ¼ :34   2 L 2 5 Geometrical properties of the mast.
K3 ð1 þ Kx=LÞ K 2 ð1 þ KÞ2
Variable L t D1 D0 K R0 Iro (mm4)
3 (m) (mm) (mm) (mm) (mm)
FL
 ð4Þ
3EtR30 Value 15 5 440 256 0.72 128 3.29  107

By applying Eq. (4) to a simplified 15 m mast with a nominal

force of 5 kN applied at the top and using the geometric properties
of the mast shown in Table 1, the displacement of the mast is esti-
mated and shown in Fig. 4. The tip displacement calculated from
Eq. (4) is 242 mm.
Eq. (4) can be used to justify the design of VAWTs’ supporting
mast under static equivalent loads to ensure that the displacement
is not excessive at the top of the mast. The limitation of this
expression is that the mast design needs to be simple enough to
be idealised as a uniform tapering.
The displacement estimated in Fig. 4 can be used to approxi-
mate the stiffness of a cantilever beam of uniform cross-section
with a concentrated mass at the end as shown in Fig. 5, for which
an analytical solution for predicting its natural frequency is avail-
able as follows [26]
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
k 3EI
xn ¼ with k¼ ð5Þ
mend þ 0:23mmast L3
Using the displacement calculated earlier, the stiffness can be
Fig. 4. Displacement of a tapered 15 m mast under a 5 kN force.
approximated as k = 5000 N/0.242 m = 20,660 N/m.
By using this value in Eq. (5) together with mend = 430 kg and
mmast = 630 kg, the natural frequency of the system is estimated
as 0.96 Hz. The natural frequency of the mast–turbine system mea-
sured by Quiet Revolution in the field was found to be equal to
0.8 Hz which is 20% lower. This shows that the approximation pre-
sented in Fig. 5 should not be used as the result is not consistent
with the site measurements; it would also not be possible to obtain Fig. 5. Cantilever beam of uniform section with end mass.
the second mode through such a model.

4.2. Natural frequency of a tapered cantilever beam ~z¼0

~  z00 þ k
m ð6Þ

A cantilever mast can be analysed as a single degree of freedom where

(SDOF) system with a distributed mass and elasticity. This system Z L
is simplified as a generalised SDOF system with m(x) being the ~ ¼
m mðxÞ½wðxÞ2 dx ð7Þ
mass per unit length. The deflections can be related to a single gen- 0
Z L
eralised displacement z(t) through a shape function w(x) and can ~¼ 2
k EIðxÞ½w00 ðxÞ dx ð8Þ
be expressed as u(x, t) = w(x)z(t). This approach can be used for 0
masts with distributed mass or several lumped masses when a 3 x2 1 x3
shape function can be found that satisfies the displacement bound- and wðxÞ ¼  ð9Þ
2 L 2 L
ary conditions [27].
The equation of motion of the free vibration without damping where the axis convention of x this time is so that x = 0 at the base
and external excitation is of the mast where R(x) = R1. The mass can be expressed as
mðxÞ ¼ m1 ð1 þ Kx=LÞ with K = (R1  R0)/R0, as previously. By
568 E. Verkinderen, B. Imam / Engineering Structures 100 (2015) 564–576

which results in a value equal to 12.8 rad/s which corresponds to a

frequency equal to 2.06 Hz.
This model with a single degree of freedom is useful to calculate
the natural frequency of a tapered mast alone but cannot include
the effect of the coupling of the turbine which has a high concen-
trated mass and different stiffness, as is presented in the following
sections.

4.3. Simplified model of the turbine

The qr5 VAWT of Quiet Revolution has rotating and fixed parts.
As shown in Fig. 6, the rotating parts include carbon fibre blades
attached with pinned connections to top and bottom spokes.
There is an aluminium outer tube which spins around a steel stator
tube. The electrical generator is located at the base of the turbine.
Because of its size and orientation the turbine modifies largely the
first modes of the coupled mast and turbine structure. By compar-
ison, a HAWT has an orientation which affects far less the first
mode of their supporting structure. Therefore, the masts support-
Fig. 6. Quiet Revolution qr5 and simplified model.
ing HAWTs can be analysed with a lumped mass at the top which
concentrates all the mass of the turbine as a first approximation.
substituting the shape and mass functions into Eq. (7) and for a The turbine is a complex mechanical system with many parts
mast tapering from a 440 mm diameter to a 256 mm diameter assembled together. In order to study the effects of the turbine sys-
(K = 0.72): tem on the mast, the turbine needs to be idealised. A model with a
low number of degrees of freedom (DOF) sufficiently enough to
~ ¼ 0:374  m1  L
m ð10Þ capture the lower modes of vibration of the overall structure is
The inertia is calculated assuming that the thickness (t = 5 mm) investigated. The model should allow for an efficient analysis while
is small in relation to the diameter of the mast: predicting the main design effect onto the mast. One important
 feature of the design of the qr5 turbine is that the rotating ele-
x3 ments are assembled onto the static structure using bearings.
IðxÞ ¼ p  t  RðxÞ3 ¼ I1 1 þ K ð11Þ
L The bearings between the rotating parts such as the rotor and
Substituting Eqs. (9) and (11) into (8) and carrying out the integra- the blades as well as the stator are pinned at the bottom and slid-
tion, we get ing pinned at the top and therefore do not participate in the lateral
bending stiffness of the assemblage. Therefore, it is proposed that
~ ¼ 5:13  EI1
k ð12Þ the turbine structure is simplified in the analysis of the mast by
L3 ignoring the stiffness of the rotating elements while still taking
The natural frequency of the specific tapered mast is given by into account their mass. This allows analysis of the coupling effects
sffiffiffiffiffi sffiffiffiffiffiffiffi through a simple static model. The simplified model can be as well
~ 3:71 EI
k reproduced easily for different configurations of turbines with a
1
xn ¼ ~ ¼ 2 ð13Þ different stiffness and mass. Fig. 6 shows the simplified model of
m L m1
the qr5 turbine proposed for the dynamic analysis.

Table 2
Mass distribution of idealised turbine for dynamic analysis.

Length (mm) Level (mm) Mass (kg)

m1 Anemometer 5700 1
Beam 1 Tube anemometer 1200 4
m2 Spoke + 50% blades 4500 64
Beam 2 Tube 4000 137
m3 Spoke + 50% blades + 50% generator 500 117
Beam 3 Tube 500 17
m4 Base pole tube + 50% generator 0 97
5700 1647 437

Fig. 7. Mast discretised with lumped masses.

 E. Verkinderen, B. Imam / Engineering Structures 100 (2015) 564–576 569

The mass analysis of the turbine elements is summarised in Table 3

Table 2 in the form of a simplified mass distribution. The weight Properties of the MDOF model for the 15 m mast.

of the top spokes and half the weight of the blades have been Segment D t E (N/ Density L A m
aggregated at the top node (m2). Similarly, the weight of the bot- (mm) (mm) mm2) (kg/m3) (mm) (mm2) (kg)
tom spokes, half the weight of the blades and half the weight of 1 139 6.35 205,000 12,900 2000 2644.9 65
the generator have been assigned to the node at the base of the 2 139 6.35 205,000 12,900 2500 2644.9 0
rotor (m3). The other half of the generator mass is lumped at the 3 271 4 205,000 7800 2160 3353.5 230
4 298 4 205,000 7800 2160 3692.6 0
base of the turbine (m4). 5 326 4 205,000 7800 2160 4044.3 0
The central part of the turbine is made of an aluminium tube 6 354 4 205,000 7800 2160 4396 0
rotating around a fixed stator made of a steel tube. Only the stiff- 7 381 5 205,000 7800 1830 5903.2 0
ness of the steel tube has been used. The centre of gravity of this 8 404 5 205,000 7800 1830 6364.3 0
9 428 5 205,000 7800 1830 6641.1 0
model is located 1647 mm above the base of the turbine which
10 440 5 205,000 7800 1000 6829.5 0
accounts for only 2% error compared to the actual position mea-
sured by weighing the turbine.

4.4. Multi-degree of freedom system
The tapered mast can be discretised as a series of smaller beams
and therefore the dynamic problem of the system can be discre-
tised using a multi-degree of freedom (MDOF) system with a finite
number of degrees of freedom. The MDOF model is shown in Fig. 7.
The mast and turbines are discretised using a total of 10 nodes and
20 DOFs and the properties of the MDOF system are shown in
Table 3. The mass is distributed throughout the length of the beam
but is idealised as concentrated at the nodes. The turbine part has
been modelled by using the simplified model presented in the pre-
vious section (Fig. 6). At each node n, there are two degrees of free-
dom i.e. the lateral displacement un and the rotation hn. The
stiffness matrix can be established by applying unit displacements
and rotations to the degrees of freedom. The general form of the
stiffness matrix for a cantilever beam with a number of nodes
can expressed as
where Un are the natural mode shapes of the system. The eigen-
value problem can be solved in order to determine natural frequen-
cies and corresponding mode shapes of the system. Static
condensation is used to eliminate from the dynamic analysis the
DOFs of the structure with no mass associated [27].
Calculating the eigenvalues of a structural system can be com-
putationally demanding, especially in the case of a high number
of DOFs, and often requires an iterative process to solve. The
inverse vector iteration method with shifts using Rayleigh’s quo-
tient iteration is preferable for quickly solving this type of struc-
tural problem [27]. This method has been employed to calculate
the natural frequencies of the coupled 15 m mast with the turbine.
The first three natural frequencies calculated analytically were
found equal to 0.80 Hz, 3.33 Hz and 7.49 Hz which were in good
agreement with the field measurements carried out by Quiet
Revolution (0.80, 3.7 and 8.7 Hz, respectively).

22 EI 3 2 E1 I 1 33
1 1
 EL13I1 0 0 0  L2  EL12I1 0 0 0
6 L31
6 6  1
 7 6 1
 1
 77
3 6 6 7 6 77 2 3
2
F1 6 6  E13I1 E1 I 1
þ EL23I2  EL23I2 0 0 7 6 E12I1 E1 I 1
 EL22I2  EL22I2 0 0 77 u1
6 6 L1 L31 7 6 L1 L21 77
6
6 F 2 7 6 12  6
2 2
7 6 2
 2
 77
E2 I2 E2 I 2 E3 I3 E3 I 3 7 6
0 7 66 0 E I E I E I 77 6 u2 7
6
6
7 6
7 6 6 0  L3 ð L3 þ L3 Þ  L3 2 2 2 2
 L323  EL32I3 0 77 6 7
6 F3 7 6 6 2 2 3 3 7 6 L22 L22 77 6 7
  77 6u 7
3 3
6 7 6 6 7 6   7 6 37
6F 7 6 6 0 0  EL33I3 E3 I3
3 þ
E4 I 4
. . . 7 6 0 0 E3 I3 E3 I3
 E4 I 4 7
. . . 5 7 6 u4 7
6 4 7 6 4 3 L3 3
L4 5 4 L32
L32
L24 7 6 7
6 7 6 7 6 7
6 ... 7 6 0 0 0 ... ... 0 0 0 ... ... 7 6...7
6 7 6 37 6 7
6M 7 ¼ 6 2 E1 I1 E1 I 1 3 2 76h 7
6 17 6  L2 L21
0 0 0 E1 I 1 1 E1 I 1
0 0 0 7 6 7 1 7
6 7 6 6 1
7 6 L1 2 L1
77 6 7
6 M2 7 6
6 7 6 6 E1 I1 E1 I1 E2 I2  7 6 1 E1 I1 E1 I1 E2 I2  7 6 h
6 27
7
6M 7 6 6 2  L2 E2 I 2
0 0 7 62 L þ L2 1 E2 I2
0 0 77
6 37 6 6 L1 L21 L22 7 6 1 L1 2 L2 7
7 7 6 h3 7
6
6 7 6 6 2
  7 6   77 6 7 7
4 M4 5 6 6  6
6 0  EL22I2 E2 I 2
 EL32I3 E3 I3 7 6
0 7 46 0 1 E2 I 2 E2 I 2
þ EL33I3 1 E3 I3
0 7 77 4 h4 5
6 6 L22 L23 7 6 2 L2 L2 2 L3
77
... 6 6
2 3
  7 6   77 ...
6 6 0 7
...7 6 0 ...7
E I E I
6 4 0 E
 L323 I E I
2 
3 3 E I
4 4
5 4 0 1 3 3 3 3
þ EL44I4 57
4 3 L3 L 2
4
2 L3 L3 5
0 0 0 ... ... 0 0 0 ... ...

The general equation of motion for a MDOF system is 5. Numerical analysis of the mast and turbine system
€ ðtÞ þ c  uðtÞ
mu _ þ k  uðtÞ ¼ f ðtÞ ð14Þ
5.1. Finite element analysis
where m, c and k are the mass, damping and stiffness matrices of
the system, u is the displacement matrix and f is the applied force A finite element (FE) model of the mast and turbine system is
matrix. For a MDOF system, the free vibration of the undamped sys- also developed and analysed to verify the analytical results
tem can be described by its natural vibration modes and described obtained from the MDOF analysis. The FE method is also based
mathematically by on the theory described in Section 4.4 but the computational
uðtÞ ¼ ðAn cos xn t þ Bn cos xn tÞUn ð15Þ power allows to keep the six DOF at each node and possibly use
the exact mass matrix. The GSA FE software developed by Oasys
570 E. Verkinderen, B. Imam / Engineering Structures 100 (2015) 564–576

Table 4
Estimated natural frequencies through different methods.

Mode Natural frequency (Hz) FE modal mass

(kg)
FE Measured Analytical
model
15 m 1 0.8 0.8 0.8 212
mast 2 3.4 3.7 3.3 79
3 8.5 8.7 7.5 183
9 m mast 1 1.6 1.6 119
2 4.7 4.9 81
3 17.1 16.0 4
6 m mast 1 1.5 1.4 100
2 5.3 5.4 78
3 16.4 16.0 3
3 m mast 1 2.3 2.0 69
2 8.6 8.0 79
3 17.7 16.0 4

The mast and turbine is modelled by using a simple beam and

Fig. 8. FE model of 15 m mast and first three mode shapes.
[28] is used to model the coupled mast–turbine structure; the soft-
ware allows lumping of the mass at nodes or calculates mass from
the element shape functions. The dynamic analysis of a consistent
mass system requires far more computational effort than a lumped
mass idealisation. Even if it leads to greater accuracy in the result,
in practice the improvement is often only slight [27]. The number
of nodes used can as well be greater and as such allow for a more
accurate result or models of more complex structures.
mass FE model. Each beam of the mast has been assigned a differ-
ent diameter to allow for the tapered shape and pedestal. The tur-
bine is modelled by a number of beams and masses as presented in
Section 4.1. Fig. 8 shows the FE model and the first three modes
obtained from the FE analysis for the 15 m mast with the corre-
sponding natural frequencies. Models for 3 m, 6 m and 9 m masts
have also been analysed and their associated first two mode shapes
are shown in Fig. 9.
In Fig. 8, it can be noticed that the first mode of the 15 m mast
corresponds to a typical cantilever beam whereas the second mode
is highly influenced by the turbine. The second mode can be con-
sidered as a turbine mode for the higher masts. For reference, the
same mast modes for the 15 m mast analysed with the turbine
modelled as a simple lumped mass at the top of the mast, as
opposed to using the simplified model of Fig. 6, have been found
to have frequencies equal to 1.0 Hz, 7.7 Hz and 22.4 Hz. These are
significantly different when compared to the frequencies obtained

Fig. 9. Mode shapes of the mast and turbine system for different mast lengths obtained from FE analysis.
 E. Verkinderen, B. Imam / Engineering Structures 100 (2015) 564–576 571

Fig. 10. Normal force coefficient normalised by free stream velocity for one blade.

Fig. 11. Cyclic streamwise force coefficient.

using the FE coupled mast–turbine model (0.8 Hz, 3.4 Hz and
8.5 Hz) which shows that the turbine cannot be assumed as a
lumped mass at the top of the mast as it considerably overesti-
mates the frequencies. The calculated first three natural frequen-
cies of the turbine itself are 3.0 Hz, 17.2 Hz and 25.9 Hz.
Table 4 summarises the FE results for all mast models (analyti-
cal and numerical) with different lengths analysed and compares
them with field measurements. The natural frequencies of installed
masts and turbines have been recorded by Quiet Revolution using
the accelerometer located at the top of the anemometer [29]. The
values shown in Table 4 are average measured natural frequencies.
For comparison, the three natural frequencies obtained from the
analytical model for the 15 m mast are also shown in the table. It
can be seen that the difference between the measured and FE cal-
culated results are generally less than 10%, the highest one being
for the shorter 3 m mast. The analytical results obtained from the
analysis of the MDOF system can be seen to have a maximum
difference of 16% with the measured values. This shows that such
a model offers a quick, easy and relatively accurate way of estimat-
ing the fundamental dynamic properties of mast and turbine struc-
tures. However, the FE method can be used for the analysis of more
complex geometries, which may be difficult to discretise analyti-
cally, and was shown to offer very accurate estimation of the
dynamic characteristics of this type of structures.
5.2. Forced excitation of the mast and turbine system
VAWTs spin on their vertical axis. The main resultant of the
wind action on the turbine is a drag in the direction of the wind.
The resultant forces can be either calculated or measured in a wind
tunnel. The maximum drag on the Quiet Revolution qr5 is 5 kN.
There are, however, some small cyclic radial loads generated by
either the aerodynamic loads on each of the three blades or by
out of balance.
572 E. Verkinderen, B. Imam / Engineering Structures 100 (2015) 564–576

Fig. 12. Cyclic side force coefficient.

Table 5 5.2.1. Aerodynamic forces

Out of balance and aerodynamic cyclic forces. The numerical prediction of the aerodynamic forces generated
Wind speed RPM Out of balance Aerodynamic forces by the VAWT onto the mast can be carried out using double
(m/s)
Frequency Force Frequency Force
(upwind & downwind) – multiple stream tube models [30]. The
(Hz) (N) (Hz) (N) thrust forces on a high solidity VAWT (k = 0.5) and at low velocity
0 0 0.0 0 0.0 0
demonstrate that the passing upstream blade has a significant
1 22 0.4 0 1.1 0 effect on the downstream blade and can result in dynamic stall
2 44 0.7 1 2.2 2 and formation of large scale vortices [31]. However, for low solidity
3 67 1.1 2 3.3 3 VAWTs operating at high rotation speeds, the load application is
4 89 1.5 3 4.4 6
nearly sinusoidal and the dynamic stall is small [30].
5 111 1.9 4 5.5 9
6 133 2.2 6 6.7 14 The exact value of the forces depends on many of the VAWT
7 155 2.6 8 7.8 18 intrinsic parameters such as the geometry of the blades, the tip
8 177 3.0 11 8.9 24 speed ratio, the solidity and the blade profile. The qr5 VAWT uses
9 200 3.3 14 10.0 30 a NACA-0015 Aerofoil and has a low solidity (k = 0.2) with a pro-
10 222 3.7 17 11.1 38
11 244 4.1 20 12.2 45
jected blade area of around 0.9 m2 and the coefficient force varia-
12 260 4.3 23 13.0 54 tion less than 0.7. The rotor area is approximately 0.6 m2. The
13 260 4.3 23 13.0 63 maximum turbine speed is 260 rpm which is reached with wind
14 260 4.3 23 13.0 74 gust of 12 m/s with an associated tip speed ratio (TSR) of around
15 260 4.3 23 13.0 84
3.0. The turbine is stopped at higher winds to prevent damaging it.
16 260 4.3 23 13.0 96
As an initial simplified model without the secondary effects, for
a straight bladed VAWT, the aerodynamic force on each blade is
defined as [32]

Fig. 13. Cyclic forces versus frequency.

 E. Verkinderen, B. Imam / Engineering Structures 100 (2015) 564–576 573

Fig. 14. Harmonic force applied on the FE model.

 
1 component. A similar summation for the lateral force coefficient on
F x ¼ C Fn cos h  qV 2wind  A ð16Þ
2 each blade leads to a total lateral force coefficient shown in Fig. 12.
The average lateral force is a harmonic force with a nil mean value.
where V is the wind speed, q the air density, A the blade area and C The aerodynamic cyclic forces are tabulated in Table 5 in rela-
the normal force coefficient. The normal force coefficient nor- tion to the turbine speed and excitation frequencies. These cyclic
malised by free stream velocity for one blade has been calculated forces are exciting the turbine at the blade pass frequency (BPF).
by Quiet Revolution using a multiple-streamtube and is shown in For example, with 3 blades at 260 rpm, the BPF is
Fig. 10. As each blade is out of phase by 120°, the streamwise force 13 Hz = 260 rpm  3/60 s.
coefficient for each blades noted N1, N2, N3 can be summed and the
total streamwise force coefficient for the three helical blades can be 5.2.2. Out of balance forces
calculated and is shown in Fig. 11. This represents the drag of the The balancing of the rotating wind turbine is important to min-
turbine and it can be seen that force has a small harmonic variable imise vibrations. Quiet Revolution always balances their turbines

Fig. 15. Acceleration under a harmonic nominal 50 N force (15 m mast).

574 E. Verkinderen, B. Imam / Engineering Structures 100 (2015) 564–576

Fig. 16. Peak acceleration response to cyclic forces (15 m mast).

Fig. 17. Peak displacement response to cyclic forces (15 m mast).

5.2.3. Excitation spectra

Table 6
Measured natural frequency and acceleration of the 15 m mast. Above 12 m/s of wind, the wind turbine speed does not increase
as it has reached its optimum working limit. However, the drag and
Mode 1 Mode 2 Mode 3
aerodynamic forces still increase. Above this range the turbine is
Measured frequency 0.8 Hz 3.7 Hz 8.7 Hz stopped by using the integrated brake.
Min. acceleration 0.2 g 0.9 g 0.6 g
The out of balance forces given in Table 5 are plotted against the
Max. acceleration 0.8 g 1.6 g 1.3 g
frequency of the cyclic force in Fig. 13. These cyclic forces are used
to define a number of harmonic excitation force spectra on the cou-
to a maximum out of balance weight of 10 g on each spoke. The pled mast and VAWT system. The harmonic force is applied at the
centrifugal force created by the out of balance mass is top of the turbine in the FE model as shown in Fig. 14.
The structural damping is usually measured or needs to be
F x ¼ m  R  x2 ð17Þ
assessed by reference to measurements on existing structures hav-
where m is the mass, R is the radius to the centre of rotation and ing size, form of construction and foundations similar to the mast
x = rpm  (2p/60) is the angular speed in rad/s. On the Quiet under consideration. The structural damping for a steel mast is
Revolution qr5, the out of balance is estimated at 10 g on a 1.5 m between 0.3% and 1% depending on the type of connection and
radius. Total out of balance force is twice the centrifugal force. stiffness of the foundations [33,34]. The damping used in the anal-
Out of balance forces are given in Table 5. ysis is 0.8% based on Quiet Revolution previous site measurements
The exact position of the out of balance forces is not known and [35].
can vary over time. For the analysis of the mast, this approximation The cyclic forces vary depending on the excitation frequencies.
is therefore conservative especially when applying the exciting Therefore, the response to a nominal harmonic force of 50 N at the
force at the top of the turbine. top of the mast is first extracted. The response for the 15 m mast is
 E. Verkinderen, B. Imam / Engineering Structures 100 (2015) 564–576
Fig. 18. Mast, turbine and coupled system natural frequencies for different mast
lengths.
given in Fig. 15. The exact excitation is simulated by applying a
correction factor related to the amplitude of the cyclic forces; aero-
dynamic and out of balance. The peak predicted acceleration
response is shown in Fig. 16 and the peak predicted displacement
in Fig. 17. The excitation harmonic forces are also shown on the
secondary axis. Resonances are clearly visible at the natural fre-
quencies of the system; 0.8 Hz, 3.4 Hz and 8.5 Hz. Higher modes
are not properly predicted as the model of the turbine itself would
need to be refined.
Table 6 shows the predicted accelerations obtained from the FE
analysis for the 15 m mast. These are found to be slightly lower
than the measurements of the already installed turbines. It is likely
that the out of balance forces and aerodynamic cyclic forces
increase during the service life of the turbine because of wear of
the mechanical elements and potential damage on the blades. It
is proposed that a used turbine be tested to investigate this further.
The displacements associated with the first mode at 0.8 Hz are
larger than for the second mode even if the acceleration is less. The
calculations of the peak acceleration and peak displacement are
key to be able to predict the acceptable stiffness of the mast and
potential serviceability or fatigue issues.
A full fatigue analysis of the mast can be carried out using the
results of the dynamic analysis and local wind distribution. It is
worth noting that the small displacement at the top of the mast
at the third mode of 8.5 Hz has an associated stress range, esti-
mated from the finite element model, of less than 5 MPa in the
steel tube of the mast. This stress is below the ‘‘cut-off’’ stress
range level of 30 MPa for a typical category ‘‘71’’ weld detail for
which cracks do not propagate according to Eurocode [36].
Fig. 18 shows the natural frequencies of the coupled mast and
turbine system plotted along with the natural frequencies of the
mast alone and turbine alone. The associated modal masses are
shown in Table 4. The first mode of the coupled mast and turbine
system corresponds to the degraded first mode of the mast with
the addition of the turbine at its top. It can be noted that the first
575
mode of the 9 m coupled system is higher than the 6 m system
because of the stiffer supporting mast used by Quiet Revolution.
The second mode of the mast is due to the coupling of the mast
and turbine. This mode corresponds to the turbine first mode cou-
pled with the mast shape with no nodal rotation at its tip. For the
mast, this would be equivalent to the first mode of a mast with an
imaginary symmetric boundary condition at its top fixing the rota-
tion. The third mode is the lowest between the mast second mode
and the anemometer natural frequency which is 17 Hz. The trend
for the second mode is to converge to the turbine alone first mode,
and of the third mode to converge to the mast alone second mode.
For all the common heights, up to 15 m, investigated in this paper
for this type of VAWT, the effect of the coupling should be consid-
ered as the modes of the mast and turbines are significantly
modified.
6. Conclusions
This paper has demonstrated analytical and numerical models
which can be used by mast and VAWT fabricators to predict and
mitigate the main adverse vibrations induced by the turbine onto
its supporting mast. A simplified analytical model of the turbine
itself, which consists of a number of beams and lumped masses,
is proposed for the dynamic analysis of the mast allowing for the
coupling of the mast–turbine system. A MDOF analytical model
consisting of the mast and turbine is developed and solved using
the vector iteration method to identify the natural frequencies of
the system. Furthermore, FE models of varying length mast and
turbine systems are also developed and analysed to study their
coupling. Lastly, an excitation force spectrum of the principal aero-
dynamic and out of balances turbine forces has also been proposed
to allow for the dynamic analysis of the first modes of the mast at
different turbine speeds.
The correlation of the natural frequencies predicted by the FE
method with site measurements has been found to be within
10% and validates the analytical turbine model developed. The ana-
lytical MDOF system was also found to give a maximum difference
of 16%, as compared to the average site measurements. The results
demonstrate that the proposed models are adequate, as an initial
approach for the structural analysis of the mast, for modes below
10 Hz without undertaking highly complex analysis.
There are numerous turbine and anemometer modes that can-
not be calculated using the current relatively coarse model. The
prediction of the higher modes is technically challenging as it
involves taking into account the rotation of the rotor and potential
centrifugal forces. The two main dynamic concerns in the design of
masts are the fatigue of welded elements and excessive displace-
ment at the top of the mast. Therefore it is usually not required
to determine accurately the higher modes of the turbine in the
design of the mast as they have been proved to be associated with
extremely small displacements and stresses under which fatigue
cracks do not propagate.
Acknowledgements
The authors would like to acknowledge Stuart Padget from
Quiet Revolution for the insightful discussions on preparing this
paper.
References
[1] Renewable UK. Small and medium wind UK market report 2013. London, UK;
2013.
[2] BS EN 61400-2:2006. Wind turbines – Part2: design requirements for small
wind turbines. British Standards, London.
576 E. Verkinderen, B. Imam / Engineering Structures 100 (2015) 564–576
[3] Balduzzi F, Bianchini A, Carnevale EA, Ferrari L, Magnani S. Feasibility analysis
of a Darrieus vertical-axis wind turbine installation in the rooftop of a building.
Appl Energy 2012;97:921–9.
[4] Bhutta MMA, Hayat N, Farooq AU, Ali Z, Jamil SR, Hussain Z. Vertical axis wind
turbine – a review of various configurations and design techniques. Renew
Sustain Energy Rev 2012;16(4):1926–39.
[5] Carne TG, James III GH. The inception of OMA in the development of modal
testing technology for wind turbines. Mech Syst Signal Process
2010;24(5):1213–26.
[6] Uys PE, Farkas J, Jarmai K, van Tonder F. Optimisation of a steel tower for a
wind turbine structure. Eng Struct 2007;29(7):1337–42.
[7] Lavassas I, Nikolaidis P, Zervas P, Efthimiou E, Doudoumis IN, Baniotopoulos
CC. Analysis and design of the prototype of a steel 1-MW wind turbine tower.
Eng Struct 2003;25(8):1097–106.
[8] Tong W. Wind power generation and wind turbine design. Southampton,
Boston: WIT Press; 2010.
[9] Quilligan A, O’Connor A, Pakrashi V. Fragility analysis of steel and concrete
wind turbine towers. Eng Struct 2012;36:270–82.
[10] James III GH, Carne TG, Lauffer JP. The Natural Excitation Technique (NExT)
for modal parameter extraction from operating wind turbines. In:
Sandia Report SAND92-1666. Albuquerque, USA: Sandia National
Laboratories; 1993.
[11] Rainieri C, Fabbrocino G. Operational modal analysis for civil engineering
structures. Springer; 2014.
[12] Biswas S, Sreedhard BN, Singh YP. Dynamic analysis of a vertical axis wind
turbine using a new windload estimation technique. Comput Struct
1997;65(6):903–16.
[13] McLaren K, Tullis S, Ziada S. Measurement of high solidity vertical axis wind
turbine aerodynamic loads under high vibration response conditions. J Fluids
Struct 2012;32:12–26.
[14] Yen J, Ahmed NA. Enhancing vertical axis wind turbine by dynamic stall
control using synthetic jets. J Wind Eng Ind Aerodyn 2013;114:12–7.
[15] Chong WT, Fazlizan A, Poh SC, Pan KC, Hew WP, Hsiao FB. The design,
simulation and testing of an urban vertical axis wind turbine with the omni-
direction-guide-vane. Appl Energy 2013;112:601–9.
[16] Yen J, Ahmed N. Improving safety and performance of small-scale vertical axis
wind turbines. Procedia Eng 2012;49:99–106.
[17] Rolland S, Newton W, Williams AJ, Croft TN, Gethin DT, Cross M. Simulations
technique for the design of a vertical axis wind turbine device with
experimental validation. Appl Energy 2013;111:1195–203.
[18] Cao L, Wang H, Ji Y, Wang Z, Yuan W. Analysis on the influence of rotational
speed to aerodynamic performance of vertical axis wind turbine. Procedia Eng
2012;31:245–50.
[19] Howell R, Qin N, Edwards J, Durrani N. Wind tunnel and numerical study of a
small vertical axis wind turbine. Renewable Energy 2010;35:412–22.
[20] Bianchini A, Cangioli F, Papini S, Rindi A, Carnevale EA, Ferrari L. Structural
analysis of a small H-Darrieus wind turbine using beam models: development
and assessment. J Turbomach 2015;137:1–11.
[21] Rolland S, Thatcher M, Newton W, Williams AJ, Croft TN, Gethin DT, Cross M.
Benchmark experiments for simulations of a vertical axis wind turbine. Appl
Energy 2013;111:1183–94.
[22] Lim YC, Chong WT, Hsiao FB. Performance investigation and optimization of a
vertical axis wind turbine with the omni-direction-guide-vane. Procedia Eng
2013;67:59–69.
[23] Saeidi D, Sedaghat A, Alamdari P, Alemrajabi AA. Aerodynamic design and
economical evaluation of site specific small vertical axis wind turbines. Appl
Energy 2013;101:765–75.
[24] Quiet Revolution. <http://www.quietrevolution.com/qr5/>; 2011.
[25] Craig RR. Mechanics of materials. 3rd ed. New York: Wiley; 2011.
[26] Stockey WF. Vibration of systems having distributed mass and elasticity. In:
Harris CM, Piersol AG, editors. Harris’ shock and vibration handbook. McGraw
Hill; 2002.
[27] Chopra AK. Dynamics of structures: theory and applications to earthquake
engineering. 3rd ed. Upper Saddle River, N.J: Pearson/Prentice Hall; 2007.
[28] Oasys Software. Oasys GSA suite. <http://www.oasys-software.com/gsa>;
2011.
[29] Quiet Revolution. Personal communication; 2012.
[30] Paraschivoiu I. Wind turbine design: with emphasis on Darrieus
concept. Montréal: Polytechnic International Press; 2012.
[31] McLaren K, Tullis S, Ziada S. Computational fluid dynamics simulation of the
aerodynamics of a high solidity, small-scale vertical axis wind turbine. Wind
Energy 2012;15:349–61.
[32] Anderson JD. Fundamentals of aerodynamics. 5th ed. McGraw-Hill; 2011.
[33] BS 8100. BS 8100-1:1986. Lattice towers and masts – Code of practice for
loading. London: British Standards; 1986.
[34] ESDU 96030. Response of structures to vortex shedding. Structures of circular
or polygonal cross section; 1998.
[35] Quiet Revolution. Site measurements on qr5 wind turbine, unpublished; 2010.
[36] BS EN 1993-1-9. Eurocode 3: design of steel structures – Part 1. 9: Fatigue.
London: British Standards Institute; 2005.