Original Article

Journal of Intelligent Material Systems

and Structures
1–15
Design and experimental Ó The Author(s) 2015
Reprints and permissions:
characterisation of a morphing wing sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1045389X15595296

with enhanced corrugated skin jim.sagepub.com

Francesco Previtali1, Giulio Molinari1, Andres F Arrieta1,

Michel Guillaume2 and Paolo Ermanni1

Abstract
In this article, a compliant morphing wing featuring an innovative load-carrying, highly anisotropic, doubly corrugated
morphing skin is introduced. A multi-disciplinary design methodology is used to optimally generate the compliant struc-
ture with the aim of maximising the produced rolling moment, while minimising mass and drag. The design tool consid-
ers the three-dimensional, aeroelastic behaviour and structural constraints. In particular, a parametric metamodel is used
to identify the best morphing skin design. The results show that the wing can achieve high levels of control authority and
has a lower or equivalent weight compared to conventional wings. A wing demonstrator is manufactured and its aeroe-
lastic performance is tested. The measurements of the displacement field show an appreciable deformation without
shape discontinuities. Low-speed wind tunnel tests indicate that the designed wing can produce roll moments that are
sufficient for replacing conventional ailerons. Moreover, the obtained changes in shape have a negligible effect on the
zero-lift drag, thus demonstrating the aerodynamic efficiency of profile changes achieved through morphing. An effective
solution for covering the used corrugation while allowing for shape changes is also introduced and tested.

Keywords
Morphing, corrugated, wind tunnel, weight

Introduction elements where this conflict is most critical is the wing

The possibility of conformably modifying the shape of
a wing potentially offers several advantages compared
with conventional discrete and discontinuous control
surfaces. For example, it can increase the aerodynamic
efficiency, improve the manoeuvrability, alleviate air-
loads and enhance the flight envelope. Previous works
(Barbarino et al., 2011; Gomez and Garcia, 2011;
Rodriguez, 2007; Sofla et al., 2010; Weisshaar, 2013)
offer several examples of morphing wing concepts
which have proved, often in wind tunnels and less fre-
quently in flight, to offer a substantial increase in per-
formance. On the other hand, penalties in terms of
cost, complexity or weight (Barbarino et al., 2011) usu-
ally restrict the use of morphing solutions to research
applications.
In order to efficiently obtain conformal shape
changes, the wing structure needs to deform. As a
consequence, morphing wings need to fulfil a difficult
compromise between stiffness (which influences the
load-carrying capability and the aeroelastic behaviour)
and compliance (which influences the deformability
and the actuation requirements). One of the structural
skin, whose deformable or adaptive part is generally
referred to as morphing skin (Thill et al., 2008b).
Morphing skins can obtain their deformation
through passive, active or semi-active mechanisms.
Solutions range from elastomers (simple, reinforced or
active) to cellular structures (auxetic, corrugated) to
shape memory materials (Barbarino et al., 2014; Liu et
al., 2014). Many of the morphing concepts described in
the literature feature non-structural morphing skins
(i.e. elastomeric or open skins). This results in consider-
able weight penalties due to the pre-defined separation
between the structural and the morphing part of the
wing and to the need for an underlying structure for
1
Laboratory of Composite Materials and Adaptive Structures, ETH
Zurich, Zurich, Switzerland
2
Centre for Aviation, ZHAW Zurich University of Applied Sciences,
Winterthur, Switzerland
Corresponding author:
Francesco Previtali, Laboratory of Composite Materials and Adaptive
Structures, ETH Zurich, Leonhardstr. 21, Zurich 8092, Switzerland.
Email: f.previtali@mavt.ethz.ch

Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016

2 Journal of Intelligent Material Systems and Structures
supporting the skin. Moreover, problems might arise at
the interface between the flexible skin and the remain-
ing, much stiffer, parts of the wing.
Corrugated panels represent an interesting morphing
skin solution because of their highly anisotropic
mechanical properties and low weight (Yokozeki et al.,
2006). A new type of corrugated morphing skin, called
double corrugation (DCo), has been recently presented
by Previtali et al. (2015a). Compared to other corru-
gated skins (Fournier et al., 2013; Golzar and Ghabezi,
2014; Thill et al., 2008a; Winkler and Kress, 2012), the
DCo offers a particularly high ratio between the bend-
ing and the axial stiffness. The highly anisotropic prop-
erties exhibited by this type of morphing skin concept
allow for retaining a significant portion of the original
skin load-carrying capability; this results in efficient
wing structures and therefore in a good weight perfor-
mance. Furthermore, the DCo mechanical performance
matches closely the ideal morphing skin properties
identified by Previtali et al. (in press).
One drawback of corrugated skins is the potential
deterioration of the aerodynamic performance. The
aerodynamic effect of a non-smooth skin was investi-
gated, both numerically and experimentally, by Thill et
al. (2010a) and Xia et al. (2012). The results indicate
that the corrugations can introduce an aerodynamic
penalty and, in particular, an increase in the drag. As
the effect strongly depends on the corrugation size, the
Reynolds numbers and the corrugation position, the
impact on the specific morphing concept needs to be
investigated.
In this article, a compliant wing capable of morph-
ing with no (spanwise and chordwise) discontinuities is
introduced. The morphing concept features a load-car-
rying, doubly corrugated morphing skin, which allows
for a flexible but efficient structure; as a consequence,
the morphing wing can achieve a controllability equiva-
lent to conventional ailerons while resulting in a lower
or comparable system weight. The wing design is
obtained with the help of numerical tools optimising
the roll and weight performance; the calculation con-
siders the three-dimensional (3D), aeroelastic beha-
viour, and structural constraints are enforced to obtain
a realistic wing design. The best suited DCo design is
determined using a new parametric metamodel.
A demonstrator of the designed wing is manufac-
tured and tested to provide experimental proof of the
achieved performance. The displacement field under
actuation and aerodynamic loads is measured using a
digital image correlation (DIC) system, showing good
correlations with numerical predictions. Low-speed
wind tunnel tests are conducted to demonstrate the
good aerodynamic performance and, in particular, the
appreciable control authority of the morphing wing.
The effect of the corrugation on the aerodynamic per-
formance is assessed and effectively compensated for
with a covering skin system.
Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016
Stringers
Compliant
ribs
Actuaon
Spar
Corrugaon
Skin
Wing root
Figure 1. Morphing wing concept.
Concept description
Morphing wing concept
The considered morphing wing concept (Figure 1) is
based on the compliant wing from Previtali et al.
(2014). The wing is mainly constituted of the wing skin
(part of which is replaced by a corrugated panel), com-
pliant ribs, actuators, stringers and a spar.
In this work, the skin is made of unidirectional carbon
fibre–reinforced polymer (CFRP) layers. The skin is
divided in an array of 15 sectors with different thicknesses
(five chordwise and three spanwise). The skin layup con-
siders four plies with 6q orientation, whereas the remain-
ing layers are oriented in the spanwise direction. This
layup allows for a sufficient torsional and bending stiffness
of the wing while maintaining a satisfactory chordwise
compliance. The area where the wing is clamped is rein-
forced with three additional CFRP fabric layers.
The morphing concept relies on compliant ribs and
the DCo (section ‘DCo concept’), which provides the
required compliance to the skin. The axial compliance
of the skin allows for lower actuation forces, more regu-
lar camber deformations, and prevents disadvantageous
3D coupling effect between the section morphing and
the 3D behaviour of the wing (Previtali and Ermanni,
2012). The corrugated skin is located on the lower wing
surface to minimise its aerodynamic penalty; based on
previous investigations (Previtali et al., in press), the
corrugation is positioned aft of the spar.
The wings have 10 compliant ribs, distributed pair-
wise along the span. The ribs, which are free of hinges
and mechanisms, have a truss-like structure and are
used for controlling the shape changes in the wing pro-
file. Because of the intended manufacturing method
(3D printing), the ribs are made of polymeric material
(acrylonitrile butadiene styrene, ABS).
The actuators are located between each rib pair. The
actuation deforms the wing causing a change in the
camber of the aerofoil. The actuators are assumed to
Previtali et al.
Shoulders
Figure 2. Axial deformation of a double corrugation. The
structures are loaded only with forces acting horizontally.
apply an imposed displacement; thus, they have a stif-
fening effect on the structure, and therefore contribute
to the buckling strength of the wing.
The spar is made of CFRP fabric layers and is
attached to the skin with compliant joints. For greater
design freedom, the spar is not constrained to be ortho-
gonal to the wing midplane. The wing has five strin-
gers, three on the upper surface and two on the lower
one. The spar, the stringers and the DCo are inter-
rupted at the ribs in order to avoid interferences with
the rib structure and the actuators.
The root of the wing is clamped so that the obtained
deformed shape has no discontinuities along the span-
wise direction. In this work, the wing has a half-span of
1.5 m and a chord of 0.3 m, resulting in an aspect ratio
of 10. The wing has no taper, sweep or twist; the aero-
dynamic profile (NACA 0012) is constant along the
span.
DCo concept
Recently introduced by Previtali et al. (2015a), the DCo
is an extension of conventional corrugations and it can
be considered the combination of two ‘hat-type’ corru-
gations with stiffened horizontal elements. When axially
elongated (Figure 2), the vertical elements (shoulders)
of the DCo are loaded with transversal shear-bending,
thus displaying an S-shaped deformation. If sufficiently
stiff, the horizontal elements remain essentially unde-
formed and the DCo shows little or no flexure also
when the axial load is not applied on the neutral axis.
The bending moments are transformed into two
axial forces (one tensile and one compressive) in the
shoulders, thus allowing for a high effective bending
stiffness. As a result, the DCo is considerably more
compliant in the axial direction than conventional cor-
rugations having the same bending stiffness and weight.
Interestingly, as opposed to elastomeric and conven-
tionally corrugated skins, the DCo retains its bending
stiffness also when compressed, thus removing the need
of a pre-stretching (Previtali et al., 2015a).
Since axial elongations and bending moments gener-
ate different types of loading on the shoulders, the
structural response in the two cases can be tuned more
independently than for conventional corrugations.
Thus, the DCo offers a greater ratio between bending
and axial stiffness, a higher structural performance and
a greater design flexibility (Previtali et al., 2015a). On
the other hand, as the bending stiffness depends on the
Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016
3
axial elongation, the design of DCos needs to take into
account the structural non-linear behaviour.
Wing design
Wing optimisation method
The considered wing has no ailerons; thus, the morph-
ing deformation must be able to provide a sufficient roll
controllability and manoeuvrability. Additionally, as
for all aerospace structures, the wing should be light-
weight. These two performance criteria are considered
essential in order to propose a viable morphing concept.
Therefore, the presented wing is designed with the help
of a multi-objective optimisation tool, aiming at simul-
taneously maximising the rolling moment coefficient CL
and minimising the wing weight.
The (non-dimensional) rolling moment coefficient
CL is defined as
L
CL = 1 2
ð1Þ
2 rU Sb
where L is the rolling moment, 1=2rU 2 the dynamic
pressure, S the wing area and b the wingspan. The rolling
moment L is achieved by deflecting the two half-wings in
opposite directions up to a maximum allowed actuation
level. For simplicity, the same force is applied to all the
actuators on the same half-wing. To obtain a more gen-
eral performance index, the rolling moment coefficient
CL is calculated for lift coefficients CL from 0.0 to 0.8
and the minimum CL value is used as fitness function.
The wing weight is obtained as the sum of the structural
and the actuator weights, estimated from the actuation
forces and strokes using the approach from Previtali et al.
(2014). The estimated weight is expected to be lower than
the real one: due to the simplified modelling used for the
optimisation, some elements like glue and local reinforce-
ments are not included in the estimation. This is, however,
not considered critical as the obtained value adequately
represents the weight performance of the wing.
To attain a realistic solution, strength and feasibility con-
straints are imposed. As dimensioning load cases, two
steady flight conditions (equivalent to CL = 1:5 and
CL =  0:38) and a roll manoeuvre (at CL = 1:5) are
used; a safety factor of 1.50 (1.25 for buckling) is applied.
The calculations are performed at a design speed of 35 m/s
(dynamic pressure of 750 Pa). Feasibility constraints pre-
vent interference between the different structural elements,
both in the deformed and the undeformed states.
The optimisation is performed with a customised
version of MTALAB’s multi-objective controlled
elitist genetic algorithm from Previtali et al. (2014).
The optimisation considers 146 design variables
(DVs), most of which are devoted to the rib structure
parameterisation. The design parameters are here
summarised:
4 Journal of Intelligent Material Systems and Structures

 Properties, position and length of the corruga-

Begin
tion (three DVs). The corrugation properties are
parameterised as described in section ‘Result’.
 Cross-sectional position of the actuators and Feasible?
intensity of the force (three DVs).
 Skin thickness: the thickness of five cross-sec- 2D (Xfoil) + 3D (ELLT) undeformed
tional, piecewise-constant thickness segments, aerodynamic calculaon
the orientation 6q of the skin layers, the (two)
spanwise positions at which the thickness is uni- Structural calculaon

Aeroelasc iteraon
formly reduced and the related reduction factors
(10 DVs). Strength check and feasibility scaling
 Rib topology, represented by the position of the
nodes and the thickness of the beams connecting

Skip calculang
2D (Xfoil) + 3D (ELLT) deformed
them (115 DVs). aerodynamic calculaon
 Spanwise position of each rib pair (five DVs).
 Cross-sectional position and thickness of the
Converged?
spar and size (length and thickness) of its com-
pliant joints (five DVs).
 Cross-sectional position of each stringer (five End
DVs).
Figure 3. Individual evaluation algorithm.

The algorithm from Previtali et al. (2014) is used for

calculating the performance of the morphing wing. The
calculation steps are illustrated in Figure 3 and sum-
marised in the following. First, the calculation is initialised.
The wing structural finite element (FE) model is created1
and the initial feasibility of the solution is checked. The
aerodynamic properties (coefficients of lift cL , drag cD and
pressure cp ) of the undeformed two-dimensional (2D) pro-
file are calculated using Xfoil (Drela, 1989). The pressure
load distribution on the wing is calculated using the non-
linear extended lifting line (ELLT) method (Anderson,
2001) based on the 2D (viscous) simulations. The pressure
coefficients on the wing surface are transformed by numer-
ical integration into concentrated forces acting on the FE
nodes. The deformations, the stresses and the buckling
loads of the wing under flight loads are calculated with
MSC.Nastran. The structural strength, both considering
yielding of the rib elements and buckling, is verified; if the
structure cannot withstand the previously described flight
dimensioning loads, the solution is rejected.
After the initialisation, the aeroelastic loop is started.
The response of the wing under manoeuvring loads,
both for positive and negative deflections, is calculated.
The maximum actuation level for each half-wing
(asymmetric morphing) is then reduced until, under
manoeuvre loads, both no structural failure and no
interference between the different structural members
occur. Similar to the case of the undeformed profile,
the 2D aerodynamic properties of five deformed sec-
tions along the semi-span are calculated for both the
positive and negative deflections. The aerodynamic
loads and the roll performance are calculated using the
previously introduced ELLT. The calculation is iterated
until convergence is achieved. A validation of the used
aeroelastic tool is reported in Previtali et al. (2014).
w1
h
w2
L
Figure 4. Layout and parameters of one module of the
corrugation considered in the optimisation.
Corrugation metamodel
Corrugation description. As explained in section ‘Wing
optimisation method’, the corrugation is parameterised
using a metamodel. Particularly, a composite DCo
design is considered. As the achievable performance
severely depends on parameters like the layer thickness
and the manufacturing process, the corrugation con-
struction and material are selected before the creation
of the metamodel.
Because of their particularly low thickness (31 mm),
unidirectional prepreg plies from North Thin Ply
Technology (120EP-513/CF, FAW 30 g/m2, FVC
53%) are selected for the shoulders. As the corrugation
is modelled with only one element in the width, the 3D
buckling behaviour cannot be considered. This is com-
pensated for by imposing layups with sufficient reinfor-
cements in both directions.2
The corrugation is assumed to be constructed as
shown in Figure 4. Apart from the mentioned layup for

Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016

Previtali et al. 5

the shoulders, the corrugation is described by five DVs: Axial stiffness [N/mm]
the length of the module3 (L), the height (h), the gap 22
width (w2), the width of the top element (w1) and the 20

5.5
thickness of the reinforcement (in red in Figure 4). 18 6.5
The maximum height of the corrugation has a signif- 16 4.5

Mb max /w [N]
icant impact on the achievable axial compliance. On 14 5.
5

6.5
3.
5
the other hand, because of the interference with the 12

2.5
upper wing skin, the height also limits the most rear- 10

4.5
ward chordwise position of the corrugation on the 8
1.5

3.5
wing. This is expected to have an impact on the perfor- 6

1
0.75

5.5
mance as compliant skins should be optimally placed 4

2.5

6.5
close to the trailing edge (TE) (Previtali et al., 2014). 2

Taking manufacturing complexity also into account, a 10

1
10
2
10
3

compromise value of 12.5 mm is selected for the maxi- Kb [Nmm]

mum corrugation height.
Figure 5. Axial stiffness as function of the bending stiffness Kb
and bending moment Mb max .
Calculation approach. The corrugation metamodel is con-
structed using several multi-objective optimisations in
which the axial stiffness (in the compliant direction) is Area weight [kg/m2]
minimised against the weight (per unit area) of the cor- 22 4
rugation. Constraints are used to assure the required 20
structural strength and bending stiffness. The con-

3.5
18

3
straints are imposed using the stress box function fs box 16
Mb max /w [N]

method by Previtali et al. (2015a). The method starts 14

2.5

4
from the definition of a range of operating conditions 12

3
3.5
(i.e. axial elongation ea and bending moment Mb ), called 10 2
the calculation arena. The fs box imposes that, over the 8

1.5
whole calculation arena, no material failure occurs and 6

2.5
the bending stiffness of a module of the corrugation is 4 1
greater than a given threshold Kb . The function fs box 2

3.5
2

3
requires the definition of three main parameters: the 10
1
10
2
10
3

maximum axial elongation ea max , the maximum bending Kb [Nmm]

moment Mb max and the minimum bending stiffness Kb .
Although the maximum axial elongation ea max could
also be chosen as parameter, this would considerably
complicate the metamodel and require an excessive cal-
culation time; thus, based on the results of preliminary
investigations, ea max is imposed to be 25%. The perfor-
mance is defined as function of the bending stiffness Kb
and the maximum transferrable bending moment
Mb max . A total of 28 optimisations4 for different values
of Kb and Mb max are performed to create the metamo-
del; for each optimisation, one solution with the opti-
mal trade-off between the axial stiffness and the weight
is selected.5 The performance obtained for several
points is expanded through interpolation to the whole
range of Kb and Mb max .
Result. The resulting axial stiffness, Ka , as a function of
the bending stiffness and the moment, is reported in
Figure 5 and the weight is reported in Figure 6. The
axial stiffness Ka is defined as
Fa 1
Ka =  ð2Þ
w ea
Figure 6. Area weight as function of the bending stiffness Kb
and bending moment Mb max .
where Fa is the axial force, w the corrugation width
and ea the axial deformation. The bending stiffness Kb ,
obtained under the assumption that the corrugation is
simply supported and moments are applied to its edges,
is defined as
L ∂Mb EI
Kb =  = ð3Þ
2w ∂q w
where Mb is the bending moment applied to each edge,
w the corrugation width, L the corrugation length and
q the edge rotation. The stiffness Kb is equivalent to the
flexural rigidity EI divided by the corrugation width.
Please note that because of the definition of the stress
box function fs box , Kb represents a lower bound for the
stiffness over the calculation arena; the maximum bend-
ing stiffness is generally higher (Previtali et al., 2015a).
Additionally to the axial stiffness Ka and the weight
r, the axial and bending stiffnesses transversal to the

Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016

6 Journal of Intelligent Material Systems and Structures
2.6
2.4
Half-wing weight [kg]
2.2
2 12
1.8
1.6
1.4
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
[]
Figure 7. Results of the optimisation.
corrugation, Ka tr and Kb tr , and the shear stiffness Ks
can be calculated for each obtained point. Although
these parameters are not part of the design objectives,
the estimated values offer a reasonable guess and are
therefore included in the metamodel.
The corrugation is represented in the wing model
described in section ‘Wing optimisation method’ with
an equivalent plate. The bending stiffness in the com-
pliant direction Kb is selected by the optimiser. A value
of bending moment Mb max is initially assumed and the
remaining properties are obtained from the metamodel
½Ka , Ka tr , Kb tr , Ks , r = f (Kb , Mb max )
At each iteration of the aeroelastic calculation (section
‘Wing optimisation method’), the previous estimation
of Mb max is corrected with the calculated aerodynamic
and structural loads and the corrugation properties are
updated. Constraints impose that the corrugation axial
elongation ea is below the maximum value considered
for the metamodel (25%).
Wing optimisation with corrugation metamodel
Using the approach described in section ‘Wing optimi-
sation method’ and the metamodel from section
‘Corrugation metamodel’, the wing design is optimised.
The optimisation considers 600 individuals and evolves
for 150 generations; the optimisation requires around
260 h on a high-performance workstation.6 The
obtained Pareto front is shown in Figure 7. The points
of the front can be assumed as lying on the line repre-
senting the minimal wing weight for a given required
rolling moment coefficient CL .
Inspection of the DVs (Figure 8) indicates that the
optimiser prefers a corrugation having a low bending
stiffness Kb : the obtained value is equivalent to a 0.12-
mm-thick aluminium sheet. As reported in Previtali et
al. (in press), the axial stiffness in the chordwise direc-
tion has a strong impact on the wing performance;
Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016
22
Moment Mb/w
20
Stiffness K b 18
3
10 16
14
Kb [Nmm]
M /w [N]
10
b
2
10
8
6
4
10
1
2
0 0.05 0.1
[]
Figure 8. Optimal bending stiffness Kb of the corrugated skin
and maximum (absolute value) bending moment applied on the
corrugation. The vertical axes’ limits correspond to the bounds
used in the optimisation.
based on the developed metamodel (Figure 5), a low Kb
allows achieving a high axial compliance thereby reduc-
ing the actuation work and thus weight. Moreover, the
low bending stiffness Kb limits the maximum bending
moment in the corrugation Mb max and results in a fur-
ther decrease in the corrugation axial stiffness and
weight. Thus, the resulting optimal corrugation layout,
which is located in the lower left part of the plots in
Figure 5 and Figure 6, corresponds to the design offer-
ing low axial stiffness and weight.
ð4Þ The optimal corrugation location is shown in Figure
9. The position is referred to the normalised curvilineal
coordinate running along the profile; the coordinate
starts from the TE (0), runs on the upper surface, on
the lower surface and ends at the TE (1). The optimal
corrugation location converges to its upper bound,
which corresponds to the rearmost allowed location; as
explained in section ‘Corrugation metamodel’, the cor-
rugation position is limited by the corrugation height.
Figure 9 also indicates that the optimal corrugation
length is equal to approximately 55 mm independently
of the rolling moment.
Wing design refinement
Based on the desired corrugation performance identi-
fied in the previous section, a corresponding corruga-
tion is designed. To simplify the manufacturing process,
extra design constraints are introduced. Particularly,
the distance between the top elements of two modules is
imposed equal to the gap width: this results in a sym-
metric design and reduces the tooling required for the
manufacturing. Moreover, the same ply preform7 used
for the shoulders is used for the reinforcement. To
obtain well-shaped corners and avoiding delamination
between the layers, two preform layers are used for the
reinforcement, one above and one below the main foil.
Previtali et al.
0.9
0.85
Position (end) [ ]
0.75
0.65
Length
Position
0.55
0 0.05 0.1
[]
Figure 9. Optimal length and position of the corrugated skin.
The position is referred to the normalised curvilineal coordinate
running on the profile; the coordinate have origin at the trailing
edge. The vertical axes’ limits correspond to the bounds used in
the optimisation.
19
12.5
8 optimisation considers the previously obtained solu-
27
Figure 10. Obtained corrugation design. Different ply
preforms are indicated in different colours; the thickness is not
in scale.
Although a corrugation with low bending stiffness
results in lower bending loads and structural weight, it
might also result in vibrations (Thill et al., 2010b). As
only the static aeroelastic behaviour is taken into
account in the optimisation, the bending stiffness is
conservatively increased.8
The obtained design is shown in Figure 10; the result-
ing corrugation has an axial stiffness Ka of 1.0 N/mm, a
weight area of 1.0 kg/m2 and a bending stiffness Kb of
355 N mm/mm (in the undeformed and unloaded state).
To account for the modification in the corrugation
properties, the wing is re-optimised. The corrugation is
represented with its equivalent model. Since the bending
stiffness is fixed, the corrugation is described only by
two parameters: the chordwise position and the length.
As the length of the corrugation modules is known, the
number of modules is used as a DV. Apart from the
corrugation, the wing is described by the same DVs.
As in the previous case, the optimisation considers
600 individuals and evolves for 150 generations. The
Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016
7
90 2.5
Total weight
80
Structural weight
70 Conventional wing
Half-wing weight [kg]
60 2
Length [mm]
50
40
30 1.5
20
10
0
1
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
[]
Figure 11. Results of the wing optimisation. The results for
the conventional wing are from Previtali et al. (2014).
results indicate that the optimal corrugation length is
two modules (54 mm), which is compatible with the
result obtained in the previous case.
Although the (linear) equivalent model reproduces
the corrugation behaviour with a reasonable level of
accuracy, it unavoidably represents an approximation.
In particular, the shear behaviour might have signifi-
cant differences. Thus, the wing model is further refined
by physically modelling the corrugation. Based on the
obtained results, the corrugation length is fixed to two
modules whereas the position is still optimised. The
tions as initial population, and it therefore evolves for
50 generations only. The obtained results are reported
in Figure 11.
The results for an equivalent conventional wing from
Previtali et al. (2014) are compared to those of the
morphing wing (Figure 11). These results are obtained
using an equivalent modelling approach and the same
optimisation algorithm. It can be seen that the morph-
ing solution offers a considerably lower weight for low
values of rolling moment coefficient CL . As CL value of
0.05 O 0.06 is considered sufficient for providing ade-
quate controllability (Previtali and Ermanni, 2012), this
morphing solution is expected to be lighter than a con-
ventional solution.
Wing demonstrator
To demonstrate the obtained results, one solution is
selected and a wing prototype is manufactured. The
multi-objective optimisation provides a set of Pareto-
optimal solutions with respect to the rolling moment
and the weight. The optimal solution needs to be
selected according to the intended application, for
example, depending on the minimum required man-
oeuvrability or the maximum allowable weight.
For demonstration purposes, a solution with a CL
value of 0.12 and a total weight of 1.85 kg (1.2 kg
8 Journal of Intelligent Material Systems and Structures
Figure 12. Example of the optimised rib layout. The skin is indicated in blue, the rib in red, the spar in green, the corrugation in
magenta and the actuator in black. For visualisation purposes, the skin thickness is increased by a factor of 2.
Figure 13. Morphing wing demonstrator.
structural weight, 0.65 kg actuation weight) is selected;
the solution is indicated in Figure 11 by a black aster-
isk. The chosen design can provide a high rolling
moment and is over-sized for common applications.
However, the structural weight (green dots in Figure
11) is negligibly higher (3%) than for solutions having a
CL value between 0.05 and 0.06: the difference in weight
is due to the higher actuation weight. Thus, it is
expected that by selecting different (under-dimen-
sioned) actuators, another point on the front can be
obtained and the performance (CL and weight) of the
chosen solution can be tuned by selecting different
actuators.
The selected wing layout is illustrated in Figure 12.
The corrugation (in magenta) is located on the rear
part of the profile. The rib structure (in red) is inter-
rupted at the corrugation positions and the structural
continuity is reintroduced by the actuation rod (in
black). Interestingly, due to its appreciable bending
stiffness, the corrugation does not need to be supported
by an underlying structure, which would hinder the
axial compliance of the corrugation.
The shape change is obtained by flexing the rear part
of the profile whereas the front part is mainly deputed
to the load-carrying function. The deformation recalls
the one of the conventional solutions based on ailerons
and is thus consistent with the experience of decades of
modern aviation. However, due to the absence of hinges
and discontinuities in the skin and to the distributed
nature of the system, the morphing solution can offer a
lower structural weight (Figure 11) and a higher aero-
dynamic efficiency (Woods et al., 2014) than conven-
tional aileron-like systems.
Although the wing model used in the optimisation
represents a realistic structure, some modifications are
introduced before the wing can be manufactured.
Although the optimisations consider linear actuators,
rotary servos are selected because of the more compact
Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016
size, the lower cost and the higher bandwidth. The ser-
vos are attached to the upper wing skin with custom-
made support and a crankshaft-like mechanism is used
to transform the rotation in linear displacement.9
Moreover, the wing is recalculated using a finer,
detailed FE model including the spar flanges, the rib-
skin connection elements, the thickness of the glue for
bonded parts, the local reinforcements, the servos with
their attachments, the crankshaft-like mechanism and
the openings for accessing the actuators.10 Based on
the detailed model, the skin thickness is modified to
have a maximum difference of two layers between adja-
cent areas; when not possible, a transition area is intro-
duced. The final layout is illustrated in Figure 27.
The wing is manufactured as described in Appendix 1;
the completed demonstrator is shown in Figure 13. The
wing has a weight of 1700 g, of which 310 g is repre-
sented by the actuators.
Actuation tests
The deformation under actuation loads is tested to ver-
ify the wing capability of undergoing the desired geome-
trical shape changes. As for the numerical calculations,
the actuators are assumed to apply the same force at all
rib positions. Since the selected servos are position con-
trolled, a direct feedback on the applied force is not
available; instead, the displacement produced by the
servos is scaled according to the calculated strokes. The
servos apply 14%, 29%, 58%, 77% and 100% of the
displacement applied by the actuator at the wingtip. In
this work, the applied actuation is indicated in levels:
one unit corresponds to a displacement imposed by the
servos equal to 0.09, 0.18, 0.35, 0.45 and 0.57 mm.
The displacements achieved by the actuation are
measured using a stereoscopic DIC system from
Correlated Solutions. The speckle patterns are printed
on adhesive paper and glued to the wing.
Previtali et al. 9

LE

TOP SURFACE
WING ROOT
BOTTOM SURFACE
WING ROOT

LE

-0.85 -0.17 0.51 1.2 1.9 2.5 3.2 3.9 4.6 5.3 5.9 6.6 7.3 8.0 8.6 9.3 10
Displacement [mm]
Figure 14. Measured out-of-plane displacements for actuation level 25. The displacements close to the wing contour cannot be
obtained due to the DIC evaluation algorithm.

20

0 Undeformed Positive Negative

-20
0 50 100 150 200 250 300

Figure 15. Deformed shape of the wing at the tip for positive ( + 4) and negative (25) actuations. The dotted line represents the
estimated shape (obtained by interpolation) across the corrugation gaps. Dimensions are in millimetres.

The vertical (out-of-plane) deformation for the 25 10

actuation level, both for the top and the bottom wing 8
surfaces, is shown in Figure 14. The results confirm that 6
TE displacement [mm]

the deformations are concentrated in the rear part of 4

the wing. The deformations increase without disconti- 2
nuities from the root (which remains undeformed) to 0
the wingtip. -2 Rib 1
The shape change for a section located at the tip of -4 Rib 2
the wing is shown in Figure 15. Although the deforma- -6 Rib 3
tion mainly takes place in the part aft of the spar, the -8 Rib 4
Tip
wing achieves a smooth deformed shape. -10
Numerical
The vertical displacement close to the TE (measured -12
-5 -4 -3 -2 -1 0 1 2 3 4
on the lower surface) is shown in Figure 16 for different Actuation level [ ]
spanwise positions. The wing is deformed with actua-
tion levels ranging between 25 and + 4. Higher levels Figure 16. Deformation of the trailing edge for different
cannot be reached due to limitations in the current pro- positions and actuation levels. Please note that the total
duced by the used power supply. The curves show a deflection is higher as the rest of the wing moves in the
slow, non-linear increase in the displacement at low opposite direction.
actuation levels, followed by a mostly linear regime.
The non-linear area is due to backlash in the actuation increased (absolute value) by 0.65 to account for the
links and in the servo attachment. Figure 16 reports previously described backlash. The numerical values
also the displacement calculated (under non-linear agree well with the experimentally measured values for
assumptions) for the tip section; the actuation level is all the considered actuation levels.

Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016

10 Journal of Intelligent Material Systems and Structures

8 12.5 m/s
vertical displacement [mm]

6 10 22.0 m/s
26.5 m/s

Displacement [mm]
4
Numerical
2 5
0
-2 0
-4
TE
-6 -5
LE
-8 Intermediate
-10 Numerical -10
0 500 1000 1500 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
spanwise position [mm] CL [ ]

Figure 17. Displacement of the leading edge (LE), trailing edge
(TE) and an intermediate position as a function of the spanwise
position. The full lines are for a positive actuation (level + 4)
and the dashed lines for a negative actuation (level 25).
6 rugation in the compliant direction, the shear loads
4
axial deformation [%]
Figure 19. Measured and calculated wingtip deflection for
different test speeds.
introduced in the skin and, in particular, in the corru-
gated skin. Because of the low axial stiffness of the cor-
cause an in-plane rotation of the corrugation: this is
indicated by the difference between the elongation of

2 the first and the second corrugation modules (Figure

0 length of the corrugated panel (Previtali et al., in press),
-2
-4
average
-6 1st module
2nd module
-8
0 500
spanwise position [mm]
Figure 18. Axial deformation of the corrugation as a function
of the spanwise position. The full lines refer to the positive
actuation (level + 4) and the dashed to the negative (level 25).
Figure 17 shows that the TE displacement increases
(absolute value) smoothly along the span, both for pos-
itive (full line) and negative (dashed line) actuation lev-
els. Interestingly, due to the wing being fixed at the
root, the forward part of the profile leading edge (LE)
moves in the opposite direction compared to the TE.
The displacement of an intermediate position located
at 68 mm from the TE is also shown. As in the previ-
ous case, the measured and the calculated values are in
good agreement.
Figure 18 shows the total corrugation elongation
along the span (in blue), and how this is distributed
between the first and the second modules of the corru-
gation. Globally, the corrugation undergoes relatively
small axial deformations; however, the maximum local
deformation of the individual modules is considerably
larger than the average value.
As the wing root is fixed, the wing sections warp
when morphed: as a consequence, shear loads are
18). As the shear stiffness increases with the spanwise
interrupting the corrugation at the rib locations allows
for a more compliant wing and thus for lower actuation
forces.
Wind tunnel tests
1000 1500 The wing demonstrator is tested in the ETH low-speed,
close test section wind tunnel to provide an experimen-
tal indication of the achieved performance and its func-
tionality under both actuation and aerodynamic loads.
The wind tunnel has a cross section of 2 3 3 m and
a test length of 4 m. Tests are conducted by performing
a sweep of the angle of attack (AoA) at the rate of
0.5°/s: the AoA is decreased from 0° to its minimum
value, then increased to the maximum value and finally
brought back to the initial condition. During the test,
forces and moments are measured with a six-compo-
nent, piezoelectric balance (Internal Report, 1973) and
the displacements of the wing are measured using the
previously introduced DIC system. The test speed is
conservatively limited to 26.5 m/s due to the possible
occurrence of critical vibrations at higher speeds.
The wing is tested at three different speeds: 12.5 m/s
(Re 240,000), 22 m/s (Re 420,000) and 26.5 m/s (Re
510,000). The wingtip deflection measured at the differ-
ent speeds is shown in Figure 19. The rigid rotation of
the wing due to the AoA change is removed from the
measured displacements and the deflection is obtained
as the average of at least four points along the chord.
Please note that the irregularity of the curves is due to
the presence of vibrations and measurement noise
(amplified by the removal of the AoA rigid rotation).

Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016

Previtali et al. 11

1 1
0 12.5m/s (Re=240'000)
0.8 ±2 0.8 22.0m/s (Re=400'000)
0.6 ±3 0.6 26.5m/s (Re=500'000)
±4
0.4 0.4
±5
0.2 0.2
CL [ ]

CL [ ]
-3
0 0 0
+3
-0.2 -0.2
-0.4 -0.4
-0.6 -0.6
-0.8 -0.8
-10 -5 0 5 10 -10 -5 0 5 10
α [°] α [°]

Figure 20. Lift curves for different actuation levels at 22 m/s. Figure 21. Lift curves for different test speeds at actuation
levels 23, 0 and + 3.
The experimental values are compared with the cal-
culated deflections (Figure 19). The calculation method
relies on the same fluid–structure interaction described 0.06
in section ‘Wing optimisation method’ but considers a
non-linear structural solution and a fixed actuation 0.05
level. The results show good agreement, thus further
demonstrating the validity of the used simulation tool. 0.04
The lift coefficient measured at 22 m/s for different
[]

actuation levels is reported in Figure 20. The actuation 0.03

shifts the curve vertically: negative actuations increase
0.02
the maximum CL whereas positive actuations decrease -2/+2
the minimum CL . Depending on the AoA, the morph- -3/+3
0.01
ing deformations can result in changes in CL of 0.35 O -4/+4
-5.5/+4.5
0.40 between the maximum and the minimum tested 0
actuation levels. -10 -5 0 5 10 15
α [°]
Tests show that the lift is unchanged when the actua-
tion is unpowered. This is due to the fact that the aero- Figure 22. Rolling moment curves for different actuation levels
dynamic loads are mainly carried by the structure and at 22 m/s.
therefore cause small forces on the actuators; these
small forces can be reacted by friction in the actuation
linkages and by the internal resistance of the unpow- 0.05
ered servo. Thus, for the investigated conditions, the 0.045 -3
0
failure of the servos is not critical for the wing struc- 0.04 +3
tural integrity. 0.035
Figure 21 shows that the speed has a small effect on
0.03
the lift coefficient CL . The lift coefficient at 22 and
CD [ ]

0.025
26.5 m/s is practically identical, whereas the curves at
12.5 m/s display small differences at low AoA and an 0.02
earlier stall at positive AoA. These differences are 0.015
caused by the change in the Reynolds number rather 0.01
than by differences in the aeroelastic behaviour.11 0.005
The rolling moment L is estimated as the difference 0
between the root bending moments when the wing is -4 -2 0 2 4 6 8
α [°]
deformed in the positive and the negative directions12
  Figure 23. Drag curves for different actuation levels at
L’+ M   M  ð5Þ 26.5 m/s.

The moment coefficient for different actuation levels

(Figure 22) shows that the wing is able to provide a suf- Figure 23 shows that the aerodynamic drag curves
ficient roll control authority over a wide range of move horizontally with the actuation level. As the lift is
AoAs.13 similarly affected by morphing, the drag polar

Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016

12 Journal of Intelligent Material Systems and Structures

25
20
15
10
5
E[]

0
-5
-10
-15 -3
0
-20
+3
-25
-5 0 5 10
α [°]

Figure 24. Aerodynamic efficiency (lift-to-drag ratio) for

different actuation levels at 26.5 m/s.

undergoes no significant change with the actuation

level. Also, the zero-lift drag of the actuated wing is,
for the reported actuation levels, almost equal to the
undeformed case: this shows that the shape change is Figure 25. Wing with metal foil covering the corrugation.
obtained with minimal aerodynamic losses. This is con-
firmed by the fact that the wing has approximately the
same maximum lift-to-drag ratio in the undeformed 0.014
and in the morphed state, although the condition of 0.012
maximum efficiency is achieved at different AoAs
0.01
(Figure 24). It is important to highlight that these
CD0 [ ]

results are obtained, although the drag minimisation is 0.008

not part of the optimisation objectives. The good aero-
0.006
dynamic efficiency is due to the conformal changes in
the aerodynamic profile shape which are characteristic 0.004
of morphing wing solutions. 0.002 Open
Closed
0
-5 -4 -3 -2 -1 0 1 2 3 4
Aerodynamic effect of the corrugation Actuation level
The utilised corrugation allows for realising great ani- Figure 26. Zero-lift drag coefficient as a function of the
sotropy necessary for structural purposes. On the other actuation level. The full curves are at 22 m/s and the dashed
hand, this results in a not-smooth wing profile. As dis- ones at 26.5 m/s.
cussed in section ‘Introduction’, corrugations can result
in aerodynamic penalties and, in particular, in an advantage of being tuneable and to have a distributed
increase in the drag. nature (Previtali et al., 2015b). This solution was, how-
To evaluate the effect of the corrugations considered ever, not applied because of technological limitations,
in this work, the aerodynamic performance of the wing that is, the difficulty of producing electrically insulated
with corrugations is compared with the one of a wing foils of the required size and with a sufficient quality to
with a smooth profile. The smooth profile is obtained avoid the occurrence of breakdown phenomena.
by covering the corrugations with a 0.05-mm-thick The influence of the corrugation on the drag is evalu-
DC01 steel foil from MetallJobst. The forward part of ated by considering the analytic polar
the foil is attached to the wing using adhesive and fur-
ther secured with adhesive tape (in blue in Figure 25). CD = k(CL  CL0 )2 + CD0 ð6Þ
In the remaining part, the foil is free to slide on the
wing and on the corrugation and the wing can still be The drag polar is obtained by interpolating the mea-
actuated. sured drag with a second-degree polynomial; to exclude
The foil is maintained attached to the wing using the data close to the stall, only the points with
magnets glued to the inside of the wing. The same effect CD \0:03 are considered in the interpolation. The
can be achieved with electro-bonding (Bergamini et al., results show that the zero-lift (parasitic) drag CD0 is
2007; Di Lillo et al., 2011, 2013), which has the lower when the corrugation is covered (Figure 26), thus

Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016

Previtali et al.
resulting in an increase in the aerodynamic efficiency
by around two points.14 On the other hand, no differ-
ence is measured on the lift-induced drag (coefficient
k). The tests also indicate that the corrugation has no
measurable effect on the generated lift in any tested
condition (speed and actuation level).
The tests demonstrate that the foil remains attached
to the wing during normal operations15 and during
changes in the applied morphing. Thus, the proposed
solution can be implemented on morphing wings with
corrugated skins (not only DCo) for mitigating or elim-
inating the aerodynamic penalties associated with the
non-smooth skin. Please note that compared to the seg-
mented skin from Thill et al. (2010b), the suggested
solution results in a smoother skin and is therefore
expected to offer a lower drag.
Conclusion
In this article, a compliant morphing wing featuring a
novel, load-carrying, doubly corrugated morphing skin
is introduced. The wing relies uniquely on morphing to
produce rolling moments, thus allowing for replacing
conventional ailerons. The wing design is obtained with
a multi-disciplinary methodology considering the 3D
aeroelastic wing behaviour and utilising numerical
optimisations.
The results show that the wing can achieve high lev-
els of control authority. Moreover, thanks to the use of
a highly anisotropic, structural morphing skin, the pre-
sented morphing concept does not show weight penal-
ties (for medium and low CL ) compared to a
conventional wing with ailerons.
A demonstrator is manufactured and its perfor-
mance tested. The displacement field is measured with
a DIC system showing the capability of the wing to
deform without discontinuities, both in the chord- and
spanwise directions, and providing a confirmation of
the numerical predictions. Furthermore, considerable
TE displacements are obtained. Low-speed wind tunnel
tests indicate that the designed morphing wing can pro-
duce roll moments that makes conventional ailerons
unnecessary. Moreover, the obtained changes in shape
have a negligible effect on the zero-lift drag, thus
demonstrating the aerodynamic efficiency of conformal
shape changes.
A solution for covering corrugated (not only DCo)
skins is presented. The wind tunnel tests show that the
covering system operates correctly under the considered
flight conditions and it allows for a significant reduc-
tion of the zero-lift drag.
Further works will concentrate on the implementa-
tion of a semi-active adaptive corrugation from
Previtali et al. (2015b). Moreover, the use of the
selected concept for the optimisation of the wing aero-
dynamic efficiency can be investigated.
Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016
13
Acknowledgements
The authors would like to thank Michael Ammann and
David Borer for their help with the wind tunnel tests and G.
Molinari, T. Delpero and B. Schläpfer for providing their
OP2-reader program.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with
respect to the research, authorship and/or publication of this
article.
Funding
This work was supported by ETH Research Grant CH1-04
08-3. The DIC system was acquired, thanks to SNF R’Equip
grant no. 206021_150729 and the complementary ETH
Scientific Equipment Program. The authors would like to
thank the support of the ETH Research Commission and the
Marie Curie Actions Cofund Program; A.F.A. was partly
funded through an ETH Postdoctoral Fellowship.
Notes
1. The FE model is described in detail in the supplementary
material.
2. Four layouts are considered: [0,90,0], [0,90,90,0],
[0,90,0,90,0] and [0,90,0,0,90,0]. The 90° direction is per-
pendicular to the corrugation.
3. Please note that at this stage of the work, the length of
the corrugation module is disconnected from the length
of the corrugated panel used in section ‘Wing optimisa-
tion method’ as DV for the wing.
4. The optimisations require a total of 340 calculation
hours on a high-performance workstation (Intel i7-3930
K 3.20 GHz, 32 GB RAM, SSD drive).
5. As the thickness of the shoulder is bound to the previ-
ously introduced layups, the obtained Pareto fronts have
a very limited trade-off area and, in some case, converge
to one point. As a consequence, the identification of the
optimal trade-off is generally simple.
6. Intel i7-3930 K 3.20 GHz, 32 GB RAM, SSD drive.
7. Because of the difficulty of handling the selected pre-
preg, the thin UD layers are ordered in preforms made
of three plies, pre-stacked with orientation [0,90,0] to
match the layup of the shoulders.
8. As an alternative solution for the mitigation of vibra-
tions, a semi-active augmentation of the corrugation’s
damping through a reversible electric bonding (Previtali
et al., 2015b) could be used.
9. The rear attachment point is slightly repositioned to
avoid interferences with the skin.
10. The detailed FE model is described in detail in the sup-
plementary material.
11. This is verified by repeating the calculations changing
only the dynamic pressure or only the Re number; in the
first case, no visible change in the curves is obtained.
Please note that the critical Re is around 240,000.
12. As only one half-wing is tested in the wind tunnel, the
results refer to a symmetric lift distribution. However,
during a roll manoeuvre, the lift distribution is not
14
symmetric. Numerical calculations (reported in the sup-
plementary material) indicate that the different distribu-
tion results in approximately 5% overestimation of the
rolling moment.
13. Please consider that as mentioned in section ‘Actuation
tests’, the deformation (and therefore the roll moment)
achievable during the tests is limited by the used power
supply and not by the wing design or the installed
actuators.
14. The results are compatible with the measurements from
Thill et al. (2010b) and Molinari et al. (2015).
15. During one test, the foil separates after stall at negative
AoA (215°) due to the recirculating nature of the flow.
The foil re-attaches to the wing when the profile recovers
from the stall; no damage is detected.
References
Anderson JD (2001) Fundamentals of Aerodynamics. New
York: McGraw-Hill.
Barbarino S, Bilgen O, Ajaj RM, et al. (2011) A review of
morphing aircraft. Journal of Intelligent Material Systems
and Structures 22: 823–877.
Barbarino S, Flores ES, Ajaj R, et al. (2014) A review on
shape memory alloys with applications to morphing air-
craft. Smart Materials and Structures 23: 063001.
Bergamini A, Christen R and Motavalli M (2007) Electrosta-
tically tunable bending stiffness in a GFRP–CFRP compo-
site beam. Smart Materials and Structures 16: 575.
Di Lillo L, Carnelli DA, Bergamini A, et al. (2011) Quasi-sta-
tic electric properties of insulating polymers at a high vol-
tage for electro-bonded laminates. Smart Materials and
Structures 20: 057002.
Di Lillo L, Raither W, Bergamini A, et al. (2013) Tuning the
mechanical behaviour of structural elements by electric
fields. Applied Physics Letters 102: 224106.
Drela M (1989) XFOIL: an analysis and design system for
low Reynolds number airfoils. In: Mueller T (ed.) Low
Reynolds Number Aerodynamics. Berlin, Heidelberg:
Springer, pp. 1–12.
Fournier S, Airoldi A, Borlandelli E, et al. (2013) Flexible
composite supports for morphing skins. In: XXII congress
of Italian association of aeronautics and astronautics,
Naples, 9–12 September.
Golzar M and Ghabezi P (2014) Corrugated composite skins.
Mechanics of Composite Materials 50: 137–148.
Gomez JC and Garcia E (2011) Morphing unmanned aerial
vehicles. Smart Materials and Structures 20: 103001.
Internal Report (1973) Die Entwicklung Der Sechs-Komponen-
ten-Waage Für Den Unterschall-Windkanal Der Ethz. Zur-
ich: ETH Zurich.
Liu Y, Du H, Liu L, et al. (2014) Shape memory polymers
and their composites in aerospace applications: a review.
Smart Materials and Structures 23: 023001.
Molinari G, Arrieta AF and Ermanni P (2015) Numerical and
experimental investigation of aero-structural characteris-
tics and performance of distributed compliance morphing
wings. In: 23rd AIAA/ASME/AHS adaptive structures con-
ference, Kissimmee, FL, 5–9 January.
Previtali F and Ermanni P (2012) Performance of a non-
tapered 3D morphing wing with integrated compliant ribs.
Smart Materials and Structures 21: 055008.
Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016
Journal of Intelligent Material Systems and Structures
Previtali F, Arrieta AF and Ermanni P (2014) Performance of
a three-dimensional morphing wing and comparison with
a conventional wing. AIAA Journal 52: 2101–2113.
Previtali F, Arrieta AF and Ermanni P (2015a) Double-walled
corrugated structure for bending-stiff anisotropic morph-
ing skins. Journal of Intelligent Material Systems and Struc-
tures 26: 599–613.
Previtali F, Arrieta AF and Ermanni P (in press) Investiga-
tion of the requirements for passive morphing skins. Smart
Materials and Structures.
Previtali F, Delpero T, Bergamini A, et al. (2015b) Multi-
functional extremely anisotropic structural element.
Extreme Mechanics Letters 3: 82–88.
Rodriguez AR (2007) Morphing aircraft technology survey.
In: 45th AIAA aerospace sciences meeting and exhibit,
Reno, NV, 8–11 January. Reston, VA: American Institute
of Aeronautics and Astronautics.
Sofla AYN, Meguid SA, Tan KT, et al. (2010) Shape morph-
ing of aircraft wing: status and challenges. Materials &
Design 31: 1284–1292.
Thill C, Downsborough J, Lai S, et al. (2010a) Aerodynamic
study of corrugated skins for morphing wing applications.
Aeronautical Journal 114: 237–244.
Thill C, Etches JA, Bond IP, et al. (2008a) Experimental and
parametric analysis of corrugated composite structures for
morphing skin applications. In: 19th international confer-
ence on adaptive structures technology, Ascona, 6–9
October.
Thill C, Etches JA, Bond IP, et al. (2008b) Morphing skins.
Aeronautical Journal 112: 117–139.
Thill C, Etches JA, Bond IP, et al. (2010b) Composite corru-
gated structures for morphing wing skin applications.
Smart Materials and Structures 19: 124009.
Weisshaar TA (2013) Morphing aircraft systems: historical
perspectives and future challenges. Journal of Aircraft 50:
337–353.
Winkler M and Kress G (2012) Influence of corrugation geo-
metry on the substitute stiffness matrix of corrugated lami-
nates. Composite Structures 94: 2827–2833.
Woods BK, Bilgen O and Friswell MI (2014) Wind tunnel
testing of the Fish Bone Active Camber morphing con-
cept. Journal of Intelligent Material Systems and Structures
25: 772–785.
Xia Y, Bilgen O and Friswell M (2012) The effect of corru-
gated skins on aerodynamic performance. In: 23rd interna-
tional conference on adaptive structures and technologies,
Nanjing, China, 11–13 October.
Yokozeki T, Takeda S, Ogasawara T, et al. (2006) Mechanical
properties of corrugated composites for candidate materi-
als of flexible wing structures. Composites Part A: Applied
Science and Manufacturing 37: 1578–1586.
Appendix 1
Wing manufacturing
The wing is manufactured in the laboratory. The wing
skin is laminated on a positive aluminium mould using
unidirectional CFRP prepreg from c-m-p (CM-Preg T-
C-120/625 CP002 35). A glass fibre–reinforced polymer
Previtali et al. 15

Rib 4 plies 7 plies Corrugation

Spar 5 plies 9 plies Reinforcement
Stringer 6 plies Openings Act. attachements

6 pl. 4 plies
Wing root

7 pl. Lower surface

5 plies

Leading edge
7 5
Wing root

6 plies

5
9 pl. 7 plies Upper surface

Figure 27. Sketch of the final structural layout.

(GFRP) counter-mould is used to obtain a good surface
finish and a controllable thickness.
The DCo is manufactured from two single corruga-
tions. The corrugations are laminated on an aluminium
mould, covered with a solid silicon counter-mould,
vacuum-bagged and cured in the autoclave at 120° and
5.5 bar. The two corrugations are glued together with a
prepreg epoxy adhesive film from Newport (NB-102).
Unfortunately, because of excess glue creating fillets at
the sides of the bonded areas, the resulting axial stiff-
ness of the manufactured corrugation is higher than the
numerically predicted value.
Stringers are laminated using woven CFRP prepreg
from SGL Group (CE 8201-245-45S); the spar is wet-
laminated on a positive mould using a CFRP fabric.
Compliant ribs and actuator attachments are produced
in ABS using a Dimension Elite fused deposition mod-
elling (FDM) machine.
The manufactured parts are then cut to size and
assembled. Shims are used to compensate for manufac-
turing tolerances between the spar and the skin. The
corrugation is bonded to the skin with a single strap
joint. Two reinforced access panels are built in the skin
for the installation of the actuators.
The actuation is provided by S.Bus S9074SB digital
servos from Futaba (maximum torque 20 kgf cm). The
servos are attached to an aluminium shaft installed in
the ribs; loads are transferred to a pultruded CFRP rod
using an aluminium arm. The servos are powered with
a Voltcraft PS 2403 Pro power supply and are con-
trolled using an Arduino Mega 2560 microcontroller
board and a VB6 user interface.
Finally, the wing is mounted on a milled aluminium
support and fixed with screws; the skin at the wing root
is locally strengthened with wet-laminated woven
CFRP reinforcements to compensate for the screw
holes.

Downloaded from jim.sagepub.com at PENNSYLVANIA STATE UNIV on March 6, 2016