Sie sind auf Seite 1von 12

Journal of Intelligent Material Systems and

Structures
http://jim.sagepub.com/

A morphing trailing edge device for a wind turbine


Stephen Daynes and Paul M Weaver
Journal of Intelligent Material Systems and Structures 2012 23: 691 originally published online 6 March 2012
DOI: 10.1177/1045389X12438622

The online version of this article can be found at:


http://jim.sagepub.com/content/23/6/691

Published by:

http://www.sagepublications.com

Additional services and information for Journal of Intelligent Material Systems and Structures can be found at:

Email Alerts: http://jim.sagepub.com/cgi/alerts

Subscriptions: http://jim.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://jim.sagepub.com/content/23/6/691.refs.html

>> Version of Record - Apr 9, 2012

OnlineFirst Version of Record - Mar 6, 2012

What is This?

Downloaded from jim.sagepub.com at Monash University on December 5, 2014


Article

Journal of Intelligent Material Systems


and Structures
23(6) 691–701
A morphing trailing edge device for Ó The Author(s) 2012
Reprints and permissions:
a wind turbine sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1045389X12438622
jim.sagepub.com

Stephen Daynes and Paul M Weaver

Abstract
The loads on wind turbine components are primarily from the blades. It is important to control these loads in order to
avoid damaging the wind turbine. Rotor control technology is currently limited to controlling the rotor speed and the
blade pitch. As blades increase in length, it becomes less desirable to pitch the entire blade as a single rigid body, but
there is a requirement to control loads more precisely along the length of the blade. This can be achieved with aerody-
namic control devices such as flaps. Morphing structures are good candidates for wind turbine flaps because they have
the potential to create structures that have the conflicting abilities of being load carrying, lightweight and shape adaptive.
A morphing flap design with an anisotropic cellular structure is presented, which is able to undergo large deflections and
high strains. An aeroelastic analysis couples the work done by aerodynamic loads on the flap, the flap strain energy and
the required actuation work to change shape. The morphing flap model is manufactured, and its stiffness is measured.

Keywords
morphing trailing edge, load control, anisotropy, aeroelasticity, composites, wind turbine, variable camber

Introduction considerable interest from the rotorcraft community as


a means of controlling blade loads (Chopra, 2000;
Wind turbine blade load alleviation Giurgiutiu et al., 1995; Straub et al., 2004). Being able
The aerodynamic loads on the horizontal-axis wind to control the camber of wind turbine blades has the
turbine blades are highly variable due to a combination potential to reduce structural fatigue at higher wind
of factors including wind shear, tower shadow, yawed speeds and gusts through load alleviation while captur-
operation and turbulence (Hau, 2006). Precisely con- ing more energy at lower wind speeds.
trolling the variability in blade loading would lead to The standard approach to manipulate air loads on
cost benefits to the wind turbine industry if the system an aerofoil section is to use hinged control surfaces
used for load alleviation did not itself add significant such as those used as aircraft ailerons. However, the
cost (Berg et al., 2009b). Individual or collective wind ability of such devices to alter the geometry of an aero-
turbine blade pitch control systems have been investi- foil for load control is a compromise between aerody-
gated in recent years to reduce the blade stresses caused namic performance and the weight and complexity of
by fluctuating loads (Bossanyi, 2003; Larsen et al., the mechanical design. This often results in a control
2005). However, it is unlikely that blade pitch control surface, which is articulated only about a single point
alone can totally eliminate these loads since they are a to reduce the weight penalty effects of the added struc-
function of blade length. In addition, the blade pitch tural mass (Campanile, 2005). Figure 1 contrasts a con-
response to rapidly varying loads is restricted by blade ventional 20% chord hinged plain flap and a morphing
inertia. With the continuously increasing length of wind trailing edge of the same chord length and the same
turbine blades, there will be even more variability in NACA 63-418 blade section. Early attempts at design-
loading along the length of the blade since the rotor ing wind turbine blade control surfaces focussed on
size will increase with respect to the typical size of tur-
bulent eddies.
As an alternative, or in addition, to individual blade Department of Aerospace Engineering, University of Bristol, Bristol, UK
pitch control, blade section camber change shows
Corresponding author:
promise as a means of load alleviation (Barlas and Van S Daynes, Department of Aerospace Engineering, University of Bristol,
Kuik, 2010; Berg et al., 2009a; Buhl et al., 2007).The Bristol BS8 1TR, UK.
application of control surfaces has also received Email: Stephen.Daynes@bristol.ac.uk

Downloaded from jim.sagepub.com at Monash University on December 5, 2014


692 Journal of Intelligent Material Systems and Structures 23(6)

these three main conflicting requirements of load carry-


ing, low mass and deformability into a single design.
A static aeroelastic model of a novel morphing blade
flap design is presented along with a manufactured and
tested proof-of-concept demonstrator. The work in this
article focuses on the innovative implementation of
Figure 1. Conventional hinged flap and morphing flap what could be described as ‘traditional’ technologies to
geometries; NACA 63-418 blade section with 20% chord flap at achieve structural morphing. There is significant scope
615° deflections. for continued research into the use of smart materials,
such as piezoelectric and shape-memory alloy as a
means of shape change. However, the performance of
using mechanical flap devices similar to those used in materials such as these (Huber et al., 1997) is not at a
the aerospace industry. However, the use of such stage where they can be readily used in full-scale
devices has raised concerns due to the potential increase demonstrators, which require both fast response times
in complexity as well as the potential to increase both and the ability to withstand significant aerodynamic
aeroacoustic noise and aerodynamic drag caused by loading.
discontinuities on the external blade profile (Johnson et
al., 2008; Wilson et al., 2009).
Alternatively, by using structural deformation The morphing trailing edge concept
instead of conventional mechanisms to enable shape
The blade section considered in this study has an
change, ‘morphing’ can result in designs that have no
NACA 63-418 profile (Abbott and Von Doenhoff,
surface discontinuities (Daynes et al., 2010; Wildschek
1959) with a 1.3-m chord length and a 20% chord (260
et al., 2008, 2010). Numerous morphing concepts have
mm) trailing edge flap. It is a requirement of this study
been investigated over recent years, and there are sev-
that the morphing trailing edge deflection can cause the
eral good review articles that describe recent develop-
same change in lift as a similarly sized conventional
ments. There are review articles discussing morphing
hinged trailing edge flap, which can deflect between
structures for aerospace applications (Barbarino et al.,
215° and + 15° with positive flap angles in the direction
2011; Sofla et al., 2010; Thill et al., 2008) and for wind
of the pressure surface. An actuator for a compliant
turbine applications (Barlas and Van Kuik, 2010;
flap must do work not only against aerodynamic loads,
Lachenal et al., in press).
like a conventional flap, but also against the stiffness of
the flap itself. It is important that the morphing flap
The challenge of structural morphing design has controlled deformation under both actuator
and aerodynamic loading in order for the blade profile
Of fundamental importance in the design of morphing to achieve its designed aerodynamic profiles. The
structure concepts is an awareness of the conflicting morphing flap demonstrator investigated in this work is
requirements of compliance, load carrying capability shown in Figure 2. There are three main components to
and low mass. Finding engineering solutions that the flap construction: a compliant carbon fibre rein-
address all three of these requirements simultaneously forced plastic (CFRP) upper skin, a flexible honeycomb
is a key research challenge (Campanile, 2007). The core and a flexible silicone lower skin. The 0.4-mm-
addition of an actuator to a morphing aerofoil can
have an important influence on these requirements,
particularly with regard to stiffness as well as any
dynamic effects.
Wind turbine blades are traditionally designed for
high stiffness and low mass. This results in structures
that are fit for their intended purpose but are far too
stiff to be of use as morphing structures. As already
stated, the conventional mechanical approach of using
hinged flaps leads to highly deformable, load carrying,
structures but with a significant mass penalty.
Alternatively, compliant structures can be used, which
have both low mass and are highly deformable, but
they are often hindered by their low-load-carrying
capability and vulnerability to stress concentrations. It
can also be difficult to control the shape of compliant
structures if they have a large number of degrees of Figure 2. Morphing trailing edge design showing silicone
freedom. It is the objective of this work to integrate pressure surface.

Downloaded from jim.sagepub.com at Monash University on December 5, 2014


Daynes and Weaver 693

thick CFRP upper skin ensures that a smooth aerody- enabling the aerofoil section to retain its shape under
namic surface is achieved. This is particularly important aerodynamic loading.
on the upper surface where aerodynamic pressure loads The highly anisotropic nature of the aramid hexago-
are typically of a larger magnitude, and surface imper- nal honeycomb core in this design also ensures a high
fections can lead to premature flow separation. out-of-plane skin stiffness, which is achieved while cre-
One actuator push–pull rod is attached to each 250- ating a flap with a low flexural stiffness. This anisotropy
mm spanwise flap segment at the trailing edge. Seventeen enables the flap to deform without a large actuation
equally spaced PTFE U-shaped hooks are embedded penalty. Discontinuities are placed in the honeycomb
into the lower surface to transfer the loads from the core, in the plane of the blade section, every 20 mm in
actuator rods to the honeycomb core. The PTFE hooks order to create a one-dimensional camber change by
are bonded to the honeycomb core to allow the actuator suppressing any anticlastic curvature generated by large
rod to freely move in and out of the flap but to prevent it deformations of the hexagonal cellular structure. These
from exerting load on the silicone skin whose main func- 20-mm spaced cuts equate to one row in every six rows
tion is to provide a smooth, weather-tight, surface. The of cells being removed. These core discontinuities enable
flap interfaces with the rest of the blade structure via the sufficient spanwise contraction and expansion of the
rear spar where the actuator rod passes through. This honeycomb segments, so that the global response of the
rear spar separates the morphing function of the flap core material approximates a zero-Poisson’s ratio cellu-
with the structural function of the remaining 80% of the lar structure. This zero-Poisson’s ratio solution was
blade chord. The total mass of one 250-mm flap segment chosen as a matter of convenience since it uses ‘off-the-
is 1.4 kg. Excluding the rear spar and actuator, the total shelf’ hexagonal honeycomb. Other solutions to zero-
mass of the morphing flap model is 0.43 kg per 250-mm Poisson’s ratio (or near zero- Poisson’s ratio) cellular
flap segment. Actuation is provided by a 15-V 11-N m structures exist including the use of overexpanded hon-
servo motor with a maximum current draw of 2.5 A, eycomb where the hexagonal unit cell is manufactured
which is controlled from a PC using a USB to a pulse- in a ‘stretch’ form, ‘Flex-core’ from the Hexcel
width modulation microcontroller. It is not envisaged Corporation (Hexcel, 2010), ‘semi re-entrant’ geome-
that this actuation solution would be robust enough to tries (Grima et al., 2010) and ‘hybrid’ or ‘accordion’
withstand the required in-service fatigue life the flap geometries (Olympio and Gandhi, 2007).
would experience, but it does provide sufficient torque to
demonstrate the flap model’s operation in a laboratory
environment. Aeroelastic analysis
Compressed air is another potential actuation solu-
tion. Hollow cored flaps have been used as actuators in The aeroelastic analysis in this section attempts to
other research programmes where changing the internal describe the operation of the flap as a single structural–
air pressure enables flap deflections to be controlled mechanical–aerodynamic system. This is done by com-
(Madsen et al., 2010; Vos and Barrett, 2011; Vos et al., bining the work contributions of the compliant flap
2011). The ‘pressure-adaptive’ designs have the benefit structure, an actuator, centrifugal forces and the aero-
of being very lightweight since there is no separate dynamic loads. The coupling of the flap’s structural
actuator. However, in this project, an electrical solution deformation with the variation in aerodynamic and
was thought to be a more feasible proposition for wind centrifugal loads is important in order to quantify the
turbine applications. work supplied by the actuator. Being able to reduce the
An important consideration related to morphing amount of work that needs to be supplied by an actua-
structures is how to design structures that are rigid tor by tailoring the stiffness and deformed geometry of
enough to carry aerodynamic loads while being suffi- the structure can have a significant influence on the
ciently compliant for feasible actuation requirements. overall size and mass of the morphing flap system. The
This can lead to conflicting requirements for the desir- aeroelastic model presented in this section is able to cal-
able attributes of a compliant skin, which needs to be culate these work contributions for any given flap geo-
capable of withstanding high in-plane strains with low metry, whether a simple hinged flap or a cambered
in-plane stiffness while having sufficient out-of-plane morphing flap. The model assumes the deformation of
stiffness to withstand localised skin buckling and aero- the flap behaves as a one-degree-of-freedom system
dynamic pressure loads (Gandhi and Anusonti-Inthra, and is solely a function of flap angle. This one-degree-
2008; Joo et al., 2009; Thill et al., 2008). One successful of-freedom assumption is supported by wind tunnel
morphing trailing edge solution to this challenge was testing where a NACA 63-418 blade section fitted with
investigated in the Smart Wing program where a thin this morphing flap design was tested at airspeeds up to
flexible silicone skin was used supported by a honey- 57 m/s (Daynes and Weaver, 2011). During wind tun-
comb core with a composite laminate at the midplane nel testing, the blade section lift coefficient was not
(Bartley-Cho et al., 2004; Kudva, 2004). The honey- found to change with increasing dynamic pressure, and
comb core acted as a support to the silicone skin no flap deformation due to aerodynamic loading was

Downloaded from jim.sagepub.com at Monash University on December 5, 2014


694 Journal of Intelligent Material Systems and Structures 23(6)

observed. In the aeroelastic analysis that follows the The flap root and tip have the same definitions as
dynamic pressure, angle of attack and pitch angle of the morphing flap. The formulation of flap deflection
the blade section are modelled as fixed values. in terms of deformation mode enables both morphing
flaps and conventional hinged flaps to be described
using the same aeroelastic model simply by modifying
Flap geometry definition the deformation mode u. For both equations (1) and
The first step in the aeroelastic analysis is to find an (2), the displacement of the lower aerodynamic surfaces
expression that describes the deflected geometry of the are also approximated by the change in the vertical dis-
morphing flap. Using the approximation of small flap placement w in the aeroelastic analysis that follows
angle deflections (sin(b) ’ b), the vertical change in since any discrepancy between the upper and lower sur-
camber of the upper surface of the morphing flap can faces will be comparatively small. The geometry defini-
be defined using the following third-order polynomial tion presented here also omits any chordwise
displacement since this is much smaller in comparison

0, 0<x\c  b to out-of-plane displacement.
w = uð xÞb, u ð xÞ = ðcxbÞ3 ð1Þ
b2 , c  b<x\c

where b is the flap angle in radians defined using the Work done by aerodynamic loads
point on the upper surface at the flap root and the flap
Once the flap geometry has been defined an expression
tip deflection. The flap deformation mode (Campanile
can then be found for the work done by aerodynamic
and Anders, 2005; Daynes et al., 2009) is given by u,
loads.Here a modal formulation is adopted to describe
where c is the aerofoil chord and b is the chordwise flap
the loads on the flap as a function of a flap angle. The
length. The chordwise x-axis has its origin at the leading
work done on the flap by the aerodynamic loads per
edge, the y-axis is defined as positive upwards and
unit span, Wf, between two flap positions is found by
downwards flap deflection angles are defined as posi-
integrating the modal aerodynamic moment, Mf,
tive. At the flap root at 80% chord there is no flap
exerted on the flap with the change in the flap angle.
deflection, whereas at the flap tip the deflection is given
The modal aerodynamic moment is defined as
by 2bb. The deflected shape for the morphing flap
described by the cubic polynomial in equation (1) was ðc
selected based on the experimentally measured deflected  
Mf = q cpu (x)  cpl (x) u(x)dx ð3Þ
shapes of the upper flap surface shown in Figure 3. The
0
shape of the morphing flap model, for several values of
b, was measured by tracing the shape of the flap’s upper where cpu(x) and cpl(x) are the pressure coefficients for
surface onto a piece of graph paper and then finding a the upper and lower surfaces of the aerofoil, respec-
polynomial curve fit to describe the shape. It was found tively, and q is the dynamic pressure of the fluid flow.
that equation (1) could accurately describe the shape of The aerodynamic pressure coefficients are calculated
the morphing flap in the range of 210° < b < 10°. using XFOIL in this work (Drela and Youngren,
The vertical deflection of the upper surface of a con- 2001). Simplifying assumptions are made to the flap
ventional hinged flap on the other hand can be geometry, and it is assumed that the change in geome-
described using try of both the upper and lower surfaces can be
 expressed using the same deformation mode u. For the
0, 0<x\c  b case of a conventional hinged flap, the deformation
w = uð xÞb, u ð xÞ = ð2Þ
c  x  b, c  b<x\c mode is given by equation (2). In this case, Mf is equiv-
alent to the flap hinge moment. This is not the case for
a morphing flap, which does not pivot about a discrete
point. Instead, the variation of the modal aerodynamic
moment Mf with aerofoil deformation can be repre-
sented by linear aerodynamic operators (Campanile
and Anders, 2005; Daynes et al., 2009). These linear
operators are determined from the viscid aerodynamic
pressure distributions around the NACA 63-418 aero-
foil section calculated in XFOIL at varying flap angles.
Linear interpolations can be done between the data
sets to determine the relationships between pressure
distributions and flap angle b. The change in pressure
Figure 3. Deflected geometries of the morphing flap model in coefficients along the chord for the upper and lower
b = 5° increments from 210° to + 10°. sides of the aerofoil can therefore be expressed as a

Downloaded from jim.sagepub.com at Monash University on December 5, 2014


Daynes and Weaver 695

function of b using the following linear aerodynamic the morphing flap has a flexural stiffness, which must
operators be accounted for. This flexural stiffness is described
using the modal torsion modulus ks, and the associated
dcpu ð xÞ
Dcpu (x) = Db ð4Þ modal bending moment is given by
db
dcpl ð xÞ Ms = ks ðb  b0 Þ ð12Þ
Dcpl (x) = Db ð5Þ
db where b0 is the flap angle the morphing flap is manu-
where Dcpu and Dcpl are the changes in pressure coeffi- factured with. The morphing flap is manufactured with
cients for the upper and lower surfaces, respectively. a triangular cross section for simplicity of manufacture.
The use of these linear aerodynamic operators is only This results in the neutral flap angle position at b0 =
valid for non-stalled aerofoil flow regimes where the 24° in this case. The change in flap strain energy per
variation in loads can be approximated as being line- unit span, Ws, between two flap positions is given by
arly proportional to b. Using equations (4) and (5), the
pressure coefficients for the upper and lower surfaces ð2
b  b2
b2
of the aerofoil are Ws = Ms db = ks  b0 b ð13Þ
2 b1
b1
dcpu ð xÞ
cpu (x) = cp0u (x) + b ð6Þ The formulation for strain energy in equation (13)
db
assumes that the morphing flap structure behaves as a
dcpl ð xÞ linear torsion spring. The stiffness characteristics of the
cpl (x) = cp0l (x) + b ð7Þ
db manufactured morphing flap model are presented in the
where cp0u(x) and cp0l(x) are the pressure coefficients for ‘Results and discussion’ section.
a given angle of attack a when b = 0. Similar to equa-
tions (6) and (7), the modal aerodynamic moment Mf Work done by centrifugal forces
can be written as
In addition to the aeroelastic effects described, there is
M f = M f 0 + kf b ð8Þ an additional stiffening effect on the flap caused by
centrifugal forces. When the blade is rotating at a rota-
where kf is the modal aerodynamic torsion modulus tional speed O and the flap is offset from the blade
and is defined as dMf/db. This modal aerodynamic tor- pitch axis, the centrifugal force on each mass element,
sion modulus is now expressed in terms of the aerofoil dm, exerts a restoring force. This has the effect of stif-
deformation fening the rotation of the flap. First, consider the case
of the hinged flap shown in Figure 4. There is a centri-
ðc  
fugal force on the flap, which acts on a line through the
dcpu ð xÞ dcpl ð xÞ
kf = q  u(x)dx ð9Þ centre of rotation. The blade pitch axis and plane of
db db
0 rotation are approximated as passing through the
chord at x = c/4. For a mass element at a distance
And so the modal aerodynamic moment for the
x 2 c/4 behind the blade pitch axis, there is a chord-
undeformed aerofoil at zero flap deflection at a given
wise component of centrifugal force equal to (x 2 c/
angle of attack is
4)O2dm (Wayne, 1980). This chordwise force can be
ðc approximated as acting on a line at a distance (x 2 c + b)
  (b 2 g) from the flap hinge, where g is the blade pitch
Mf 0 = q cp0u (x)  cp0l (x) u(x)dx ð10Þ
angle.
0

Considering equations (8) to (10), the aerodynamic


work done on the structure per unit span, Wf, between
two flap positions, 1 and 2, can therefore be expressed as

ð2
b  b2
kf
Wf = Mf db = Mf 0 b + b2 ð11Þ
2 b1
b1

Change in flap strain energy


With conventional hinged flaps, there is no stiffness Figure 4. Blade element velocities and centrifugal moment on
associated with the deformation of the flap, whereas plain hinged flap.

Downloaded from jim.sagepub.com at Monash University on December 5, 2014


696 Journal of Intelligent Material Systems and Structures 23(6)

The total mass of the morphing flap model aft of the per unit span, Wa, can now be expressed as the sum of
rear spar, mt, is 1.72 kg per metre span. As an initial the work done by the aerodynamic loads on the struc-
comparison, it is assumed that the hinged flap in this ture Wf, the work done by centrifugal loads on the
study has the same mass distribution. The chordwise structure Wi and the change in strain energy in the flap
distribution of this mass per unit span is approximated Ws between two known flap positions
as a linear distribution from the flap root to the trailing    b
edge   kf + ks + ki 2 2
Wa = Wf +Ws + Wi = Mf 0  ks b0 + Mi0 b + b
2 b1
2mt ðc  xÞ
dm = dx ð14Þ ð18Þ
b2
Therefore, the modal centrifugal moment, Mi, can The actuator force (load) is related to actuator work
be found by integrating over the flap section: by
ð ð2
b
c
e
Mi = ðb  g ÞO2 x  ð x  c + bÞ Wa = e F db = ðF1 + F2 Þðb2  b1 Þ ð19Þ
4 2
flap b1
ðc
2mt ðb  g ÞO 2
c
where e is a constant that relates actuator stroke with
dm = x  ðc  xÞðx  c + bÞdx
b2 4 flap angle and F is the actuator force per unit span.
cb
The load–stroke ratio, for example, can be tailored
ð15Þ between an actuator and the flap using mechanisms
This expression can be generalised for both the such as bell cranks to change the mechanical advan-
hinged and morphing flaps with the same chordwise tage. Such mechanical advantage can be included in
mass distribution by adopting the modal formulation this aeroelastic formulation through the modification
of the constant e.
ðc
2mt ðb  gÞO2 c
If the flap is left free to float, with the actuator
Mi = x  ðx  cÞu(x)dx = Mi0 + ki b switched off (F1 = 0), it will seek an equilibrium deflec-
b2 4
cb tion angle such that the sum of the aerodynamic, cen-
ð16Þ trifugal and structural bending moments are zero. The
equilibrium deflection angle of the free flap, denoted by
where Mi0 is the modal centrifugal moment, which is b1, can be found from
constant for a given g, and ki is the modal centrifugal
torsion modulus associated with flap deflection. It Mf + Ms + Mi = Mf 0 + kf b1 + ks ðb1  b0 Þ + Mi0 + ki b1 = 0
should be noted that in following analysis the calcula- ks b0  Mf 0  Mi0
tion of the work done on the flap by centrifugal forces ) b1 = ð20Þ
kf + ks + ki
is only valid for a particular angle of attack a and blade
pitch angle g. Additional centrifugal forces exerted on The required actuator force F2 to achieve a given
the flap due to the pitch axis moving out of the plane of flap deflection b2 can now be found using equations
rotation are assumed negligible. There will be also addi- (18) to (20). For a hinged flap, there is no associated
tional forces acting on the flap due to dynamic effects; structural bending moment since the flap is modelled as
however, such dynamic analysis is beyond the scope of a zero stiffness structure, so the stable flap position
this current work. Cyclic variations in gravitational occurs when the modal aerodynamic moment is in equi-
loads exerted on the flap with blade rotation are also librium with the modal centrifugal moment.
assumed to be negligible.
The corresponding work done by centrifugal loads Manufactured demonstrator
on the flap per unit span, Wi, between two flap posi-
tions, 1 and 2, can now be found by integrating equa- The objective of the manufactured flap model is to
tion (16) with respect to b investigate the manufacturing feasibility and stiffness
characteristics of the anisotropic flap concept. An
ð2
b   b2 experimental set-up was devised to perform initial
ki
Wi = Mi db = Mi0 b + b2 ð17Þ investigations into the actuator force–displacement per-
2 b1
b1 formance required to deflect the flap model in a ‘bench
top’ environment.
Calculation of actuator work and forces The morphing flap test rig with the 250-mm span
The task of the actuator is to achieve flap deflection model flap mounted vertically is shown in Figure 5. The
through a prescribed range of angles against external test rig consists of a rectangular aluminium frame with
aerodynamic loading, structural stiffness and centrifu- the rear spar of the flap model clamped to the upper hor-
gal loading. The energy input required by the actuator izontal member. For simplicity of manufacture, the

Downloaded from jim.sagepub.com at Monash University on December 5, 2014


Daynes and Weaver 697

example. The numerical example considered is typical


of a generic 90 m diameter wind turbine operating at
higher wind speeds and higher rotational speeds. It is
assumed that the forces on a blade section can be calcu-
lated using two-dimensional aerofoil characteristics.
This is a common assumption in a wind turbine blade
element theory. The velocity components at a radial
position r = 34 m along the blade can be expressed in
terms of wind speed, flow factors and rotational speed
(see Figure 4). Consider a blade section at pitch angle
g = 8.4° relative to the plane of rotation. The blade
section rotates at an angular velocity O = 2 rad/s, the
wind speed is UN = 25 m/s and a is the inflow factor.
The blade section has a tangential velocity Or with the
tangential velocity of the wake given by a’Or. The val-
ues of a and a’, which provide maximum possible effi-
ciency to the wind turbine, are (Burton et al., 2001)
 2
1 U‘
a= and a9 = að1  aÞ = 0:03 ð21Þ
3 Or

Resolving the velocity components in Figure 4


results in a relative velocity V = 72 m/s and an angle
of attack a = 5°. This corresponds to a freestream
dynamic pressure of q = 3.18 kPa and a Reynolds
number of Re = 6.44 3 106 at sea level in standard
atmospheric conditions.
It is a requirement of this study that the change in lift
produced by deflecting the morphing flap must equal
the change in lift of a conventional hinged flap with flap
angles in the range of 215° < b < 15°. The lift coeffi-
cient of an aerofoil under non-stalled conditions can be
approximated with a single linear relationship
Figure 5. Morphing flap model installed in the test rig.
dcl
cl = cl0 + b ð22Þ
db
geometry of the aft 20% of the NACA 63-418 aerofoil
section is approximated as a triangular cross section with where cl is the section lift coefficient and cl0 is the sec-
the direction of the honeycomb cells normal to the tion lift coefficient at zero degrees flap angle at a given
suction surface and at an angle relative to the pressure angle of attack. The change in camber at the trailing
surface. The actuator rod passes from the flap model edge is determined by equation (1) for the morphing
through this horizontal aluminium member and is con- flap, whereas the geometry for the hinged flap is deter-
nected to the moving cross-head of an Instron 1-kN mined by equation (2). In this study, only plain hinged
load–displacement test machine instead of the servo flaps are compared with the morphing flap design.
motor shown in Figure 2. Mounting the flap model verti- Overhanging hinge flaps are frequently used in aero-
cally aids simple and direct installation of the flap model space applications since they typically have much lower
onto the test machine. The lower horizontal member is aerodynamic hinge moments compared to plain flaps
securely bolted to the test machine in order to avoid the since they can be more easily aerodynamically and
test rig moving when an actuator load is applied to the mass balanced (Phillips, 1949; Purser, 1944). However,
flap model. The flap was cycled 610° at up to 68°/s (0.4 the hinge line on such flap designs is difficult to seal
Hz) – this was the maximum speed of the test machine. and the protrusion of the flap overhang into the airflow
is likely to contribute to increased drag and aeroacous-
Results and discussion tic noise. For these reasons, only simple plain flaps are
considered here as acceptable flap geometries for wind
Aerodynamic performance turbine applications.
The lift characteristics of both the hinged and morph- For any non-zero flap angle, a plain hinged flap may
ing flaps are presented in this section using a numerical produce a different amount of lift than this morphing

Downloaded from jim.sagepub.com at Monash University on December 5, 2014


698 Journal of Intelligent Material Systems and Structures 23(6)

Table 1. Lift coefficient components calculated using XFOIL Two cyclic speeds were tested: 0.025 Hz representing a
(Re = 6.44 3 106). quasi-static test and the maximum speed of the test
machine at 0.4 Hz. The tests were also performed with
Hinged Morphing
the morphing flap model detached to verify that the
cl0 (a = 5°) (–) 0.374 inertia of the test machine’s grip was not influencing
dcl/db (/°) 0.076 0.051 the results. Little difference was observed between the
two test speeds. From the gradient of the results in
Figure 7, it is possible to approximate the modal tor-
sion modulus per unit span ks as 2.38 Nm/°/m. Both
flap due to their differing geometries (see Table 1). To test cases demonstrate a hysteretic response in the
make a fair comparison between the hinged and morph- force–displacement results. This hysteresis is due to
ing flaps the morphing flap angles are scaled by 0.051/ friction between the carbon actuator rod and the PTFE
0.076 = 0.7, so the same change of lift is produced by restraints, which are bonded into the core. PTFE was
each flap type. This means the morphing flap tip can used for the restraints due to its extremely low coeffi-
deflect out-of-plane 30% less compared to a hinged flap cient of friction to minimise any hysteresis.
and still result in the same change in lift coefficient. The Once the stiffness of the morphing flap is deter-
main reason for the greater control effectiveness of the mined, the flap actuator requirements can be quantified
morphing flap is due to the increased trailing edge angle and compared to the requirements of the plain hinged
(see equations (1) and (2)). The hinged flap has a con- flap. The angle of attack, a, is set at 5° in the numerical
stant angle b between the flap root and the trailing example that follows representing a typical operating
edge. However, by differentiating equation (1), it can be condition. The various moments and stiffnesses are
seen that the angle of the morphing flap at the trailing given in Table 2. The aerodynamic contributions, kf
edge is 3b.
The changes in the lift coefficients with the angle of
attack for both flaps are shown in Figure 6. The morph-
ing flap angles are scaled by 0.7 compared to the equiv-
alent hinged flap angles. The results show that the
linear analysis implemented in this work is valid for
angles of attack in the range of 210° < a < 10° and for
the hinged flap angle in the range of 210° < b < 10°
where the aerofoils are not stalled.

Actuator requirements
The experimental force–displacement characteristics of
the flap are shown in Figure 7. The constant e, which
relates actuator stroke with flap angle in equation (19),
was experimentally found to be 1.0 mm/°. Force–dis-
placement tests are conducted with the test machine
programmed to operate cyclically with b ranging 610°.
Figure 7. Experimental actuator force–displacement
characteristics for morphing flap; the direction of the loops is
clockwise.

Table 2. Moment contributions and free flap angles for both


the hinged and morphing flap
(q = 3.18 kPa, a = 5°).

Flap type Hinged Morphing

Mf0 (Nm/m) 23.81 8.91


kf (Nm/°/m) 1.56 2.72
Mi0 (Nm/m) 20.07 20.02
ki (Nm/°/m) 0.50 0.16
ks (Nm/°/m) 0.00 2.38
Figure 6. Lift polars of blade sections with hinged and b0 (°) N/A 24.0
morphing flaps (Re = 6.44 3 106). b1 (°) 211.5 23.5

Downloaded from jim.sagepub.com at Monash University on December 5, 2014


Daynes and Weaver 699

Figure 8. Pressure distributions and deformation modes of


hinged and morphing flaps (a = 5°, Re = 6.44 3 106).

and Mf0, are calculated using equations (9) and (10),


respectively. The modal bending moment, Ms, is calcu-
lated using equation (12), and Mi is calculated using
Figure 9. Actuator work requirements for hinged and
equation (16). Table 2 shows that the magnitude of the morphing flaps (a = 5° and q = 3.18 kPa).
aerodynamic terms is more dominant than the centrifu-
gal terms, which could be neglected as the first
approximation. (equation (20)). The large modal aerodynamic moment
Table 2 also shows that the term Mf0 is found to be Mf0 of the plain hinged flap results in a more negative
much larger for the plain hinged flap. The reason for free flap deflection. Even though kf is larger and ks is
this can be seen in Figure 8 where the flap deformation non-zero for the morphing flap, less actuator work is
modes and the pressure distribution along the two flaps required to achieve a full positive flap deflection:
are shown. The deflection of the morphing flap is b2 = 10° for the case of the hinged flap and b2 = 7°
biased towards the trailing edge where the pressure dif- for the morphing flap. At full positive flap deflection,
ference between the upper and lower aerodynamic sur- Wa = 8.3 J/m for the hinged flap and 5.1 J/m for the
faces diminishes to zero. The deflection of the hinged morphing flap. This result shows that the morphing
flap, on the other hand, linearly varies from 80% chord flap can potentially have a smaller actuator compared
to the trailing edge. This combination of pressure dis- to a simple plain flap of similar aerodynamic perfor-
tribution (solid black line) and flap deformation mode mance. However, the aeroelastic analysis presented
leads to smaller modal aerodynamic moments on the here has the limitation of being linear and not model-
morphing flap. However, the modal aerodynamic tor- ling energy losses in the system such as friction.
sion modulus, kf, is larger for the morphing flap. The It should be noted that the comparison between the
reason for this can also be seen in Figure 8 where there two flap designs presented here is only valid for plain
is a large pressure difference, between the upper and hinged flaps. Plain flaps can often have hinge moments,
lower surfaces, near the trailing edge when the morph- which are too large to give acceptable control forces.
ing flap is deflected (solid green lines). The larger Hinged flap designs that include features such as ‘hinge
changes in pressure distributions upon flap deflection overhang’ and mass balancing can have a large decrease
for the hinged flap are closer to 80% chord where they in their flap hinge moments, which can in fact be close
do not result in a large moment (solid red lines). The to zero under certain operating conditions (Sears,
modal torsion modulus, ks, is an additional factor for 1943). The hinge moment characteristics, and subse-
the morphing flap, which will also contribute to actua- quent actuator requirements, of such mechanical
tor work requirements. ‘balanced’ flap designs will also surpass the morphing
As with Mf0, the magnitude of the modal centrifugal flap design presented here. However, as mentioned pre-
moment Mi0 is also larger for the hinged flap although viously, the degraded aerodynamic performance, likely
Mi0 is of opposite sign. The modal centrifugal torsion increase in aeroacoustic noise and the surface disconti-
modulus, ki, is also larger for the hinged flap. These nuities associated with these flap designs mean that
results are as expected since both flaps have the same they were not considered further in this work.
mass distribution (equation (14)) with the deflection of
the morphing flap biased towards the trailing edge
Conclusions and future work
where the mass of the flap reduces to zero.
The actuator work results for the same numerical Research into morphing structures is receiving interest
example are given in Figure 9. The points of minimum in order to further improve the control surface perfor-
energy correspond to the free flap deflections b1 mance over that achieved by mechanical designs. Some

Downloaded from jim.sagepub.com at Monash University on December 5, 2014


700 Journal of Intelligent Material Systems and Structures 23(6)

of the potential advantages include lighter, simpler Barbarino S, Bilgen O, Ajaj RM, et al. (2011) A review of
structures with reduced drag due to less aerodynamic morphing aircraft. Journal of Intelligent Material Systems
surface steps and gaps. The anisotropic core flap struc- and Structures 22(9): 823–877.
ture presented in this work and the associated analysis Barlas TK and Van Kuik GAM (2010) Review of state of the
art in smart rotor control research for wind turbines. Prog-
represent a promising concept in the long-term devel-
ress in Aerospace Sciences 46(1): 1–27.
opment of adaptive control surfaces for both wind tur-
Bartley-Cho JD, Wang DP, Martin CA, et al. (2004) Devel-
bine and wider aerospace applications. opment of high-rate, adaptive trailing edge control surface
A full-sized model of the morphing trailing edge for the smart wing phase 2 wind tunnel model. Journal
device was manufactured and actuated in a laboratory of Intelligent Material Systems and Structures 15(4): 279–
environment. The flap consists of an aramid honey- 291.
comb core with a CFRP skin on one surface and a sili- Berg DE, Wilson DG, Barone MF, et al. (2009a) The impact
cone skin on the other. The flap has integrated CFRP of active aerodynamic load control on fatigue and energy
rods, which are connected to servo motor actuators. capture at low wind speed sites. In: European wind energy
Many design variables were considered including skin conference, Marseille, France. 16–19 March
Berg DE, Wilson DG, Resor BR, et al. (2009b) Active aero-
and core materials, actuator integration and actuator
dynamic blade load control impacts on utility-scale wind
spacing. The final design consisted of actuators being
turbines. In: AWEA windpower conference & exhibition,
spaced every 250 mm in the spanwise direction in order Chicago, IL. 4–7 May
to distribute loading in the flexible structure. Bossanyi EA (2003) Individual blade pitch control for load
A morphing flap model has been actuated from reduction. Wind Energy 6(2): 119–128.
210° to 10° at up to 0.4 Hz, requiring actuator Buhl T, Bak DC, Gaunaa M, et al. (2007) Load alleviation
loads ranging from 285 to 150 N, respectively, for a through adaptive trailing edge control surfaces: ADAPW-
250 mm wide flap section. ING overview. In: European wind energy conference,
An aeroelastic model was presented, which combines Milan, Italy. 7–10 May
the aerodynamic loads on the flap, the flap’s own stiff- Burton T, Sharpe D, Jenkins N, et al. (2001) Chapter 3: aero-
dynamics of horizontal-axis wind turbines. In: Tony B,
ness, centrifugal loads and the required actuator work.
David S, Nick J, Ervin B (eds.) Wind Energy Handbook.
The aeroelastic model developed is generic and can eas-
Chichester: John Wiley & Sons, Ltd, pp.46–61.
ily be adapted for different flap designs. For the NACA Campanile LF (2005) Initial thoughts on weight penalty
63-418 blade section studied, the model shows that the effects in shape-adaptable systems. Journal of Intelligent
morphing flap can have a tip deflection 30% less and Material Systems and Structures 16(1): 47–56.
still produce the same change in lift compared to a Campanile LF (2007) Chapter 4: lightweight shape-adaptable
hinged plain flap of equal chord length. airfoils: a new challenge for an old dream. In: David W,
However, many challenges are still ahead in terms of Ian B, Paul W, Michael F (eds.) Adaptive Structures: Engi-
coupled fluid–structure interaction, design, optimisa- neering Applications. 1st ed. Chichester: John Wiley & Sons
tion and manufacture of shape adaptive concepts as Ltd, p.97.
Campanile LF and Anders S (2005) Aerodynamic and aeroe-
well as showing that there are cost benefits using such
lastic amplification in adaptive belt-rib airfoils. Aerospace
technologies in relation to the overall cost of wind
Science and Technology 9(1): 55–63.
energy. Future work includes the following: Chopra I. (2000) Status of application of smart structures
technology to rotorcraft systems. Journal of the American
 Further developing shape adaptive flap designs Helicopter Society 45(4): 228–252.
suitable for wind turbine applications. Daynes S and Weaver PM (2011) Adaptive trailing edge for
 In-service considerations such as maintenance load alleviation on wind turbine blades. In: 22nd interna-
requirements, operating environment, life-span, tional conference on adaptive structures and technologies
(ICAST), Corfu, Greece. 10–12 October
cost, failsafe modes and so on.
Daynes S, Nall SJ, Weaver PM, et al. (2010) Bistable
 Further wind tunnel model development of composite flap for an airfoil. Journal of Aircraft 47(1):
morphing flap concepts and non-linear aeroelas- 334–338.
tic models. Daynes S, Weaver PM and Potter KD (2009) Aeroelastic
study of bistable composite airfoils. Journal of Aircraft
46(6): 2169–2173.
Funding Drela M and Youngren H (2001) XFOIL 6.94 User Guide.
This work as part of the ‘Morphing composites for wind tur- Cambridge, MA: Massachusetts Institute of Technology.
bine flap applications’ project was funded by Vestas Wind Gandhi F and Anusonti-Inthra P (2008) Skin design studies
Systems. for variable camber morphing airfoils. Smart Materials
and Structures 17(1): 015025.
Giurgiutiu V, Chaudhry Z and Rogers CA (1995) Engineering
References feasibility of induced strain actuators for rotor blade active
Abbott IH and Von Doenhoff AE (1959) Theory of Wing Sec- vibration control. Journal of Intelligent Material Systems
tions. 2nd ed. New York: Dover Publications, pp.536–537. and Structures 6(5): 583–597.

Downloaded from jim.sagepub.com at Monash University on December 5, 2014


Daynes and Weaver 701

Grima JN, Oliveri L, Attard D, et al. (2010) Hexagonal hon- Phillips WH (1949) Appreciation and prediction of flying qua-
eycombs with zero Poisson’s ratios and enhanced stiffness. lities. National Advisory Council for Aeronautics 927: 1–45.
Advanced Engineering Materials 12(9): 855–861. Purser PE (1944) Analysis of available data on control sur-
Hau E (2006) Wind Turbines: Fundamentals, Technologies, faces having plain-overhand and frise balances. National
Application, Economics. 2nd ed. Berlin: Springer, ch. 6, Advisory Council for Aeronautics L4E13: 1–48.
pp.163. Sears RI (1943) Wind-tunnel data on the aerodynamic char-
Hexcel (2010) HexWebÒ Nonmetallic Flex-Core Honeycomb acteristics of airplane control surfaces. National Advisory
Product Data. Stamford, CT: Hexcel Corporation, pp.1–4. Committee for Aeronautics WTRL-663: 1–219.
Available at: http://www.hexcel.com Sofla AYN, Meguid SA, Tan KT, et al. (2010) Shape morph-
Huber JE, Fleck NA and Ashby MF (1997) Selection of ing of aircraft wing: status and challenges. Materials and
mechanical actuators based on performance indices. Pro- Design 31(3): 1284–1292.
ceedings of the Royal Society of London, Series A: Mathe- Straub FK, Kennedy DK, Stemple AD, et al. (2004) Develop-
matical, Physical and Engineering Sciences 453(1965): ment and Whirl Tower Test of the SMART Active Flap
2185–2205. Rotor. San Diego, CA: SPIE Industrial and Commercial
Johnson SJ, Van Dam CP and Berg DE (2008) Active load con- Applications of Smart Structures Technologies.
trol techniques for wind turbines. SAND2008–4809, Sandia Thill C, Etches JA, Bond IP, et al. (2008) Morphing skins.
National Laboratories, Albuquerque, NM, pp.1–124. The Aeronautical Journal 112(1129): 117–139.
Joo JJ, Reich GW and Westfall JT (2009) Flexible skin devel- Vos R and Barrett R (2011) Mechanics of pressure-adaptive
opment for morphing aircraft applications via topology honeycomb and its application to wing morphing. Smart
optimization. Journal of Intelligent Material Systems and Materials and Structures 20(9): 094010.
Structures 20(16): 1969. Vos R, Barrett R and Romkes A (2011) Mechanics of
Kudva JN (2004) Overview of the DARPA smart wing proj- pressure-adaptive honeycomb. Journal of Intelligent Mate-
ect. Journal of Intelligent Material Systems and Structures rial Systems and Structures 22(10): 1041–1055.
15(4): 261–267. Wayne J (1980) Helicopter Theory. Princeton, NJ: Princeton
Lachenal X, Daynes S and Weaver PM (in press) Review of University Press, pp.405–406.
morphing concepts and materials for wind turbine blade Wildschek A, Grünewald M, Maier R, et al. (2008) Multi-
applications. Wind Energy. functional morphing trailing edge device for control of all-
Larsen TJ, Madsen HA and Thomsen K (2005) Active load composite, all-electric flying wing aircraft. In: 26th con-
reduction using individual pitch, based on local blade flow gress of international council of the aeronautical sciences
measurements. Wind Energy 8(1): 67–80. (ICAS), Anchorage, AK. 14–19 September
Madsen HA, Andersen PB, Andersen TL, et al. (2010) The Wildschek A, Havar T and Plötner K (2010) An all-compo-
potentials of the controllable rubber trailing edge flap site, all-electric, morphing trailing edge device for flight
(CRTEF). In: European wind energy conference, Warsaw, control on a blended-wing-body airliner. Proceedings of
Poland. 20–23 April the Institution of Mechanical Engineers, Part G: Journal of
Olympio KR and Gandhi F (2007) Zero-n cellular honey- Aerospace Engineering 224(1): 1–9.
comb flexible skins for one-dimensional wing morphing. Wilson DG, Berg DE, Barone MF, et al. (2009) Active aero-
In: 48th AIAA/ASME/ASCE/AHS/ASC structures, dynamic blade control design for load reduction on large
structural dynamics and materials conference, Waikiki, HI. wind turbines. In: European Wind Energy Conference,
23–26 April Marseille, France. 16–19 March

Downloaded from jim.sagepub.com at Monash University on December 5, 2014

Das könnte Ihnen auch gefallen