Research Article

International Journal of Advanced

Robotic Systems
November-December 2019: 1–19
Design of compliant mechanism-based ª The Author(s) 2019
DOI: 10.1177/1729881419886740
variable camber morphing wing with journals.sagepub.com/home/arx

nonlinear large deformation

Yaqing Zhang , Wenjie Ge, Ziang Zhang, Xiaojuan Mo

and Yonghong Zhang

Abstract
The morphing wing with large deformation can benefit its flight performance a lot in different conditions. In this study, a
variable camber morphing wing with compliant leading and trailing edges is designed by large-displacement compliant
mechanisms. The compliant mechanisms are carried out by a hyperelastic structure topology optimization, based on a
nonlinear meshless method. A laminated leading-edge skin is designed to fit the curvature changing phenomenon of the
leading edge during deformation. A morphing wing demonstrator was manufactured to testify its deformation capability.
Comparing to other variable camber morphing wings, the proposal can realize larger deflection of leading and trailing
edges. The designed morphing wing shows great improvement in aerodynamic performance and enough strength to resist
aerodynamic and structural loadings.

Keywords
Morphing wing, variable camber, compliant mechanisms, topology optimization, hyperelastic structure, meshless method

Date received: 22 May 2019; accepted: 12 October 2019

Topic: Field Robotics

Topic Editor: Yangquan Chen
Associate Editor: Ning Sun

Introduction potentials in different missions than the fixed wing aircraft.

Besides, hinged mechanisms are used to change the flight
The conventional wings are usually designed for a certain
performance in the traditional aircraft control system, such
flight condition of aircraft and thus are suboptimal under
as the flaps and slats. These control systems make the wing
other requirements. While the morphing wings can reduce
surface discontinuous, and thus are effective in changing
the design compromises in the traditional aircraft design.
airflow but inefficient in the aerodynamics.2 Also, the noise
There is neither an exact definition nor an agreement about
of fixed wings can be suppressed largely if continuous
the title of “morphing wing.” The general perception of
skins are covered on the slats and flaps.3
morphing is that of smooth and continuous shape change,
flexibility, and the “art” of mimicking birds.1 Through the
function of morphing, the morphing wings are expected to School of Mechanical Engineering, Northwestern Polytechnical University,
obtain different aerodynamic performances efficiently Xi’an Shaanxi, People’s Republic of China
according to their mission roles.
The comparison between the effectiveness of morphing Corresponding author:
Wenjie Ge, School of Mechanical Engineering, Northwestern
wing aircraft and conventional fixed wing aircraft in dif- Polytechnical University, 127 West Youyi Road, Xi’an Shaanxi, 710072,
ferent flight conditions is shown in Figure 1, based on the People’s Republic of China.
NextGen design. The morphing wing aircraft shows more Email: gwj@nwpu.edu.cn

Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License
(http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/
open-access-at-sage).
2 International Journal of Advanced Robotic Systems

FlexSys Inc (2205 Commonwealth Blvd, Ann Arbor, MI,

Takeoff: Sea Level
USA).25 developed a “Mission Adaptive Compliant Wing”
Climb: 20K Climb: Sea Level
(MACWing), owning a compliant mechanism-based trailing
edge which is designed by continuum material topology opti-
Instant Turn: Cruise: 40K mization method. Its trailing edge can realize +10 flap
20K
deflection with a 3 twist. Taking the whole airfoil as the
design area, Santer and Pellegrino27 designed a compliant
Dash: 20K Climb: 40K mechanism-based morphing wing by load path-based topol-
ogy optimization method. Similarly, Maute and Reich21
designed the morphing wing in the whole airfoil by
Loiter: 60K Sustained Turn: 40K material-based compliant mechanism topology optimization,
Loiter: 40K within a coupled fluid–structure analysis framework to
Fixed Wing Morphing Ideal directly assess aerodynamic performance criteria. Xin
Aircraft Wing Aircraft Performance et al.28 proposed a material-based nonlinear topology optimi-
zation method for the leading edge, and Xinxing et al.9 intro-
Figure 1. Spider plot of comparison of fixed and morphing wing duced the composite materials into the leading-edge
aircraft.4 compliant mechanism design. German Aerospace Center
presented a droop-nose adaptive morphing wingtip by
Many research studies have been carried out to analyze compliant mechanism for the European FP7 project “Novel
the aerodynamic characteristics of morphing wings. Air Vehicle Configurations: From Fluttering Wings to
Burnazzi and Radespiel evaluated the aerodynamics of Morphing Flight.”24 In the same European project, Politec-
flexible droop noses, which can reduce the Coanda flap nico di Milano (POLIMI)29 designed both leading and
blowing power and increase the attack angle of maximum trailing edges by load path-based compliant mechanism
lift coefficient (Cl).5 Jo et al. implemented an aerody- design method. The US Air Force Research Laboratory
namic testing of a variable camber compliant wing by (AFRL)6 also designed a “Variable Camber Compliant
Wing” (VCCW) with both compliant leading and trailing
wind tunnel test, showing 4% lift improvement in takeoff
edges with only one single actuation, and its design can
condition and 15% induced drag reduction in cruise con-
actively recontour the airfoil camber up to 6%. Diaconu
dition.6 Wang et al. investigated the aerodynamic charac-
et al.30 developed a bistable compliant mechanism-based
teristics of leading-edge droop nose devices of A380 and
trailing edge by bistable laminated composite structures
A350 and found that the devices can reduce takeoff drag
design method. However, only two stable shapes can be
as well as noise emission.7 The improvement in aerody-
realized. Ricci et al.31 investigated the configuration of
namic characteristics provides much of the impulsion for
skin and internal compliant mechanisms by a load path-
the design of morphing wings.
based multi-objective optimization simultaneously.
The morphing wing can be designed from many aspects,
However, the deflection of compliant mechanism-based
such as variable camber8–10 and variable thickness11 of leading and trailing edges are usually small, due to the fact
aerofoil-level morphing, variable sweep12,13 of spanwise that the adopted compliant mechanisms were designed by
morphing, twist14,15 and folding16,17 of wing-level morph- linear design methods and manufactured by linear-elastic
ing. In the early stage, morphing wings were designed by materials with high stiffness. That means the morphing
complex rigid-body mechanisms and actuators.18–20 How- wings will profit less from its morphing. Hence, the design
ever, rigid-body solutions always accompanied penalties in of nonlinear compliant mechanisms with large displace-
terms of cost, complexity, or weight. Although the penal- ment is necessary.
ties may be overcome by benefits of morphing wings in One of the widely used compliant mechanism design
some certain circumstances, the penalties prevent their methods is topology optimization,32,33 and nonlinear topol-
practical application, 21 especially for the low-speed/ ogy optimizations of large deformation problem have been
small-shape aircraft.22 Besides, the using of smart materials studied by a lot of researchers.34,35 Besides, meshless meth-
is also quite prevalent in the morphing wing design.14,16 ods, which are quite suitable for large displacement prob-
However, they offer limited potential at the current state, lem, are introduced to replace the finite element (FE)
due to limited scalability, force-deflection characteristics, response analysis in topology optimization. Cho and Kwak
and required power.1 Compliant mechanisms, transferring introduced the geometric nonlinearity into the meshless-
force and displacement through elastic body deformation, based topology optimization. 36 And then compliant
own many benefits, such as ease of manufacture and assem- mechanisms were designed by meshless-based topology
bling, no backlash, lightweight, and so on.23 Thus, many optimization in many research studies.37–39 Considering
researchers introduced the compliant mechanisms in the the high computational cost, Zhang et al. replaced the
morphing wing design, especially in the variable camber costly meshless method by FE meshless coupled method
wing with deflectable leading and trailing edges.24–26 in topology optimization.40,41 In addition, the hyperelastic
Zhang et al. 3

a laminated leading-edge skin is designed, as introduced

Original airfoil
Trailing edge
in Appendix 1. The laminated skin, composed of spring
Middle wing steel and glass fiber-reinforced polymer composites, shows
great advantages in dealing with the curvature changing
phenomenon.
Leading edge
Target morphed airfoil Secondly, optimize the driving forces of the skins.
The deformation of the skin is actuated by the driving
Figure 2. Original airfoil curve and target morphed airfoil curve. forces of the skin, namely, the output of compliant
mechanisms. The driving forces need to conform to the
material, which is more suitable for the large deformation skin deformation and meanwhile resist the aerodynamic
compliant mechanism design, is also introduced in the loads. Hence, more driving forces work better. However,
meshless-based topology optimization.42 too many outputs may lead to the failure of the optimi-
Hence, in this article, a meshless-based topology opti- zation of compliant mechanisms. Hence, two outputs of
mization method for hyperelastic structures is adopted to compliant mechanisms are adopted to actuate the skin
design compliant mechanisms of the morphing wing for a deformation here. And then a driving forces optimization
micro air vehicle (MAV). The MAVs usually fly slow and is needed to find the optimal parameters for the skins,
need high maneuverability, which means they can benefit a including the position, value, direction, and displacement
lot from the morphing wing. Compared to supersonic air- of the driving forces. The optimization is carried out by a
craft, the small or low-speed planes require more dramatic genetic algorithm (GA) with the response analysis calcu-
wing variations for a noticeable and practical change in lated by ANSYS, shown in the flowchart of Figure 3A in
their aerodynamic properties.22 And considering the low Appendix 3.
bearing capacity of compliant mechanisms, the wing only Thirdly, design the compliant mechanisms to actuate the
needs to bear small aerodynamic load just as well. The leading and trailing skins. By the parameters of the driving
hyperelastic structure-based compliant mechanisms are forces from the second step, the compliant mechanisms are
designed for both leading and trailing edges. A laminated optimized by a meshless-based topology optimization
skin is introduced to meet the leading-edge curvature method for hyperelastic structure. The flowchart of the
changing requirement during the leading-edge deflection. method is given in Figure 3B in Appendix 3, and the math-
A preliminary demonstrator is manufactured and its defor- ematical model is introduced in the next section.
mation capability is tested. At last, the aerodynamic per-
formance is evaluated by simulation as well as the
aerodynamic loads bearing capability. Meshless-based topology optimization method
for hyperelastic structure
Variable camber designs of leading Topology optimization is one of the most flexible structure
and trailing edges optimization methods by optimizing the material layout in
The airfoil curves are shown in Figure 2, in which the a specified design region. In this study, the compliant
original airfoil curve before deformation shows as the solid mechanism is optimized by the topology optimization
line and the target morphed airfoil curve after deformation based on the solid isotropic material with penalization
shows as the dashed line. Note that, although the target (SIMP) method. To satisfy the requirement of large defor-
morphed airfoil is the only and final design object, there mation, the hyperelastic structure is adopted in the topology
are various airfoils that can be achieved during the optimization, using nonlinear meshless response analysis.
deformation and that riches the flight performance of In the SIMP method, the configuration is presented by the
the morphing wing. material density rðxÞ 2 ½0; 1 for location x, and thus the
The morphing wing mainly includes two parts: skin and hyperelastic material model W ðxÞ is defined as
compliant mechanism. The skin is the reflection of the W ðxÞ ¼ C ði ðxÞÞ ð1Þ
morphing, while the compliant mechanism is used to
actuate the skin (transmitting the force from actuator to i ðxÞ ¼ ðrðxÞÞp i0 ð2Þ
skin). Therefore, the design process of the variable camber
where i is the material constant of hyperelastic material,
morphing wing includes three steps:
i0 is the solid material constant, and p is the penalty factor.
Firstly, design the skin of the leading and trailing edges.
The curvature of leading edge changes during the wing In this study, the independent pointwise density interpola-
deforming, as shown in Figure 2. If the wing skin is pro- tion method33 is taken to discretize the density field
X
cessed to the original shape beforehand, the skin which has rðxÞ ¼ ’i ðxÞri ð3Þ
maximum curvature can be hardly reshaped to another dif-
ferent curve with only external loads applying to the skin, where ri is thePnodal density and ’i is a Shepard function
such as the design in the literature. 24,43,44 Hence, satisfying that ’i ðxÞ ¼ 1 and 0  ’i  1.
4 International Journal of Advanced Robotic Systems

The objective function is to maximize the weighted @R

summation of the two output displacements. Considering KT ¼  ð10Þ
@U
the nodal density as the design variable and material vol-
T
ume constraint, the optimization problem is stated as Letting the L T ¼ λ KT , the sensitivity can be
expressed as
Find : ri ði ¼ 1; 2; . . . ; nÞ
X m h i2 duj T @R
Max : C ¼ ðxj þ uj Þ  xj ¼λ ð11Þ
dri @ri
j¼1
8 ð4Þ The @R=@ri can be found by differentiation of residual
>
> Rðr; uðrÞÞ ¼ 0
< in the structural equilibrium R. For more details, please
s:t: : V ¼ f VV0 refer to Buhl et al.45
>
>
: The derivative of material volume with respect to the
0 < r min  ri  1
design variables reads
where ri is the nodal density variable of the ith node; n is ð
@V @rðxÞ
the total number of the both FE and element-free Galerkin ¼ dO ð12Þ
@ri O @ri
nodes; C is the objective function; uj is the displacement of
the jth output port; xj and xj are the original and deformed In this study, the topology optimization of the compliant
positions of the jth output port; m is the total number of the mechanism design is solved by the method of moving
output ports; R is the residual in obtaining the structural asymptotes (MMA).46
equilibrium by the meshless response analysis of nonlinear
hyperelastic structures, which are the key to the large defor- Meshless method for hyperelastic structure. In the nonlinear
mation topology optimization; fV is the prescribed fraction hyperelastic meshless method, the governing equations in
of the volume constraint; V0 is the volume of the design terms of the principle of virtual work state that
domain; V is the material volume of the design domain, ð ð ð
calculated by dLT  τ dV 0  dv T  b dV 0  dvT  t  dG
ð R0 R0 Gt
ð ð ð13Þ
V¼ ð5Þ T
rðxÞdO  dλ  ðu  u ÞdG  dv T  λ dG ¼ 0
O Gu Gu

rmin is a lower bound of density which is set rmin ¼ 0.001 where u is the trial function, b is the body force distribution
in here. acting on the unloaded condition R0, t* is the traction on
boundary G t after deformation, τ is the Kirchhoff stress, dv
Sensitivity analysis. There are 2m degrees of freedom of the is the virtual velocity fields, and dL is the virtual velocity
output ports, hence the sensitivity of the objective function gradients. Noted that, the trial functions do not satisfy Kro-
can be given as necker d criterion, hence the essential boundary conditions
are imposed by Lagrange multipliers, which is shown as the
@C X 2m
@uk last two terms in equation (4).
¼ 2ðxk þ uk  xk Þ ð6Þ Equation (4) can be reducible to a linear equations sys-
@ri k¼1
@ri
tem by the incremental Newton–Raphson method
in which the displacement uk of a specified degree of free- " #( ) ( )
ðkÞ
dom k, and it can be written as KT G duðkÞ RðkÞ
¼ ð14Þ
uk ¼ L T U ð7Þ G 0 λ q

where L is a vector with value of one for position j and uðkþ1Þ ¼ uðkÞ þ duðkÞ ð15Þ
zeros for other positions. Because Rðr; uðrÞÞ ¼ 0, the uj ðkÞ
where R(k), K T , and u(k) are the residual force, the tangent
can be rewritten by the adjoint method
stiffness matrix, and the updated displacement in the kth
T
uj ¼ L T U þ λ R ð8Þ iteration of Newton–Raphson prediction, respectively. G,
λ, and q are the by-products of the Lagrange multipliers,
where λ is a vector of Lagrangian multipliers. Then, the the same as the components in the linear meshless
sensitivity of uj can be given by method47 In this article, neo-Hookean material model is
  adopted.
duj T dU T @R dU @R
¼L þλ þ ð9Þ When the structure is undergoing large deformation, the
dri dri @U dri @ri
low-density areas may excessively distorted, and their tan-
In the nonlinear analysis, the structural equilibrium is gent stiffness may lose positive definiteness in the incre-
solved by Newton–Raphson method in this article, in which mental nonlinear analysis. These numerical instabilities
requires the determination of the tangent stiffness matrix KT would stop the density-based topology optimization
Zhang et al. 5

(a) (b)

kout
2
kin Fin
F2 uout
2

Middle Design Middle

wing domain wing
kout
1
F1

uout
1

Figure 3. Schematic diagram of leading-edge design: (a) forces for driving leading-edge skin and (b) design conditions of compliance
mechanism of leading edge.

Table 1. Parameters of topology optimization for leading-edge compliant mechanism design.

Force (N) Displacement (mm) Spring stiffness (N/mm)

Location (mm) X-axis Y-axis X-axis Y-axis X-axis Y-axis

Output F1 (49.74, 18.79) 7.10 17.35 9.48 45.86 0.19 0.48
Output F2 (81.56, 103.39) 2.15 10.41 13.84 33.81 0.05 0.22
Input (174.18, 80.00) 40 0 23 0 0.58 0

iteration. Replacing the nonlinear analysis of the low- wing, shown as the displacement boundary fixed to the
density area by linear analysis is a simple and efficient lower right of the design region. The input actuation force
method,48 due to the fact that the sensitivity in low- Fin is applied horizontally to the left. The larger the dis-
density area would be suppressed largely by low-density, tance between the actuation force and the displacement
especially when considering the penalization (by a power boundary is, the smaller the input actuation force would
function, usually taking power as 3) in SIMP method. be needed. However, when setting the actuation force far
Furthermore, the meshless method can be replaced by the away from the displacement boundary, the drive mechan-
linear FE method in the linear region, which also can save ism may cause interference with the leading-edge skin dur-
the computational cost of the expensive meshless method. ing the deformation. Hence, the input actuation is set at the
Thus, an adaptive directly coupled FE and meshless right edge, 68.52% of the right edge length from the dis-
method is introduced to reduce the computational cost placement boundary.
of meshless methods, and an adaptive interpolation The springs at the output ports, which are used to simu-
scheme is adopted to arrange FE and meshless domains late the resistance from the skin, are actually the stiffness of
during the optimization. For more details, please refer to the workpiece in Sigmund,32 and it can be calculated by
Zhang et al.42
k iout ¼ F i =uiout i ¼ 1; 2 ð16Þ
where Fi is the compliant mechanism output force (also the
Compliant mechanism for leading edge driving force of the skin) and uiout is the output displace-
The compliant mechanism is used to actuate the skin, trans- ment of the compliant mechanism (also the input displace-
mitting the force of actuator to the skin. For the leading ment of the skin actuation). The output force F i is
edge, considering that the compliant mechanism needs to optimized by GA with the response analysis calculated
conform to the skin deformation and meanwhile resist the by ANSYS, which is shown in Figure 2A in Appendix 2.
aerodynamic loads, two outputs are adopted in the compli- The topology optimization parameters from equation (16)
ant mechanism design, shown in Figure 3(a). are presented in Table 1.
The compliant mechanism topology optimization model The deformation of the laminated leading-edge skin is
of the leading edge is shown in Figure 3(b). Note that, there given in Figure 4(a), and the error distribution is shown in
is no need to consider the resistance of skin in the compli- Figure 4(b). The maximum error (3.32 mm) is at the fore-
ant mechanism design, as the skin is actually simulated by part of the leading edge. Besides, the error at the upper end
the springs at the output ports. Hence, the skin area is of the skin is also remarkable, as the leading-edge skin was
excluded from the design domain (the gray area). The com- deviated from the ideal initial curve after bending the lami-
pliant mechanism is fastened to the bottom of the middle nated skin to the leading-edge curve.
6 International Journal of Advanced Robotic Systems

Figure 4. Deformation of leading-edge skin: (a) deformed laminated skin driving by compliant mechanism output forces
and (b) displacement error of the deformed leading-edge skin.

Figure 5. Compliant mechanism designed for leading edge: (a) topology configuration of compliant mechanism and (b) deformation and
von Mises stress of the deformed compliant mechanism.

Considering a two-dimensional design area with thick- rebuilt and its deformation is shown in Figure 5(b), with
ness of 20 mm and the prescribed fraction of the volume von Mises stress displayed in the same figure.
constraint fV ¼ 0.3, the topology optimization of the
leading-edge compliant mechanism design is implemented
according to the input and output parameters in Table 1. Compliant mechanism for trailing edge
The optimal configuration of the leading-edge compliant For the trailing edge, the skin is assembled by two straight
mechanism is shown in Figure 4(a). The topology config- spring-steel plates simply. To realize the variable camber
uration is expressed by a density field, which includes a design, the length of the lower skin should be reduced
great deal of intermediate density. Hence, a post-processing during the deformation, and the length of the upper skin
program is needed. The density below 0.8 is deleted, and should remain. Hence, the lower skin should be detached
the minimum width is limited to 2 mm. A compliant from the wing rib so that it can retreat to the middle wing in
pseudo-hinge is added to the bottom of the vertical bar the deforming process, while the upper skin is fixed on
which is connected to the upper skin, aiming for larger the wing. Three forces are adopted here: two forces at the
deformation. A computer-aided design (CAD) model is upper skin and one force at the lower skin, as shown
Zhang et al. 7

(a) (b)

Middle wing
Middle wing kout
1

kout
2
Design
F2 domain
F3 uout
1
uout
2

Fin
F1 kin

Figure 6. Schematic diagram of trailing-edge design: (a) forces for driving trailing-edge skin and (b) design conditions of compliance
mechanism for trailing edge.

Table 2. Parameters of topology optimization for leading-edge compliant mechanism design.

Force (N) Displacement (mm) Spring stiffness (N/mm)

Location (mm) X-axis Y-axis X-axis Y-axis X-axis Y-axis

Output F1 (76.75, 52.62) 9.45 29.79 3.03 9.55 0.94 2.97
Output F2 (153.50, 40.41) 5.71 20.47 11.73 42.05 0.13 0.47
Input (0, 10.55) 35 0 36 0 0.97 0

in Figure 6(a). The two forces loaded at the upper skin are Manufacturing and testing
provided by the compliant mechanism. And the third force
is located at the end of the lower skin, drawing the lower CAD conceptual model
skin into the middle wing. A morphing wing with 130-mm span is designed, with two
The same topology optimization method is adopted in same compliant mechanisms (10 mm thickness for each
the trailing-edge compliant mechanism design. The com- one) for the leading edge, and so does the trailing edge.
pliant mechanism is fastened to the top of the middle wing, The CAD conceptual model is displayed in Figure 9,
shown as setting the displacement boundary to the upper including the deformed leading and trailing edges which
left of the design region in Figure 6(b). To reduce the are shown translucently.
actuation load, the input actuation force is applied horizon- In the variable camber leading edge, the skin and com-
tally at the bottom on the left design region boundary. The pliant mechanisms are connected by two stringers, which
output forces of the compliant mechanism are also opti- help to distribute the force from the mechanisms to the
mized by the same method introduced in the “Compliant skin, displayed in Figure 10. Two compliant mechanisms
mechanism for leading-edge” section. The parameters of are installed on both sides, with their input forces supplied
topology optimization are presented in Table 2. by a shaft. The shaft is attached to a linear guide rail,
The deformation of the trailing-edge skin is given in ensuring that the input forces can be applied horizontally.
Figure 7, and the error distribution is also shown in the The linear motion actuation is provided by a linear stepper
same figure. The deformed skin coincides with the target motor through a leadscrew.
curve well, with the maximum error (6.07 mm) located at For the variable camber trailing edge, there are also
the tip of the trailing-edge skin.
two compliant mechanisms installed on both sides and
Considering the design area with thickness of 20 mm
connected with the skin by two stringers, as shown in
and the prescribed fraction of the volume constraint
Figure 11. A C-shape connector is designed to drive the
fV ¼ 0.25, the topology optimization of the leading-edge
two compliant mechanisms. The connector is attached to
compliant mechanism design is executed according to
a linear guide rail and actuated by a stepper motor
the parameters in Table 2. The optimal configuration of the
through a leadscrew.
leading-edge compliant mechanism is shown in
For the other input actuation force, which is applied
Figure 8(a). The thin bar in the left side impedes the hor-
to the lower skin, another drive mechanism is adopted.
izontal movement of the input actuation force. Besides, it is
The skin is actuated by a leadscrew directly. Also, a
too far from the lower limit of the bar width (2 mm) men-
linear guide rail is connected to the skin, making sure
tioned above. Thus, it is deleted in the CAD model. The
deformation and von Mises stress of the modified model is that the skin can move straightly along the lower skin of
shown in Figure 8(b). the middle wing.
8 International Journal of Advanced Robotic Systems

Figure 7. Deformation of leading-edge skin driving by compliant mechanism output forces and its displacement error.

Figure 8. Compliant mechanism designed for trailing edge: (a) topology configuration of compliant mechanism and (b) deformation and
von Mises stress of the deformed compliant mechanism.

Figure 9. CAD conceptual model of designed morphing wing.

Fabrication, actuation, and control

The monolithic compliant mechanisms were fabricated by
three-dimentional (3-D) printing with flexible polylactic
acid (PLA) material. The spring steel plate was fabricated
by electrical discharge machining, and the glass fiber-
reinforced polymer composites in laminated skin were fab-
ricated using laser cutting. The compliant mechanisms and
wing skin were fixed to a metallic perforated plate by some
Figure 10. CAD conceptual model of the variable camber leading aluminum-fabricated connectors and 3-D printing pieces.
edge. The demonstrator is shown in Figure 12.
Zhang et al. 9

Figure 11. CAD conceptual model of the variable camber trailing edge.

Figure 12. Demonstrator of the variable camber morphing wing and the displacement measurement system.

Three sets of actuation system are needed: one for the The control system includes an STM32 microcontroller
leading edge and two for the trailing edge. Each actuation (STMicroelectronics, Avenue Mohamed Jazouli Madinat
system includes three parts: a motor to provide rotation AI Irfane, RABAT, Morocco) and three HST-8325B
movement, a leadscrew to convert the rotation movement microstep drivers (HUISITONG No. 28, Chang’an
to straight movement, and a linear guide rail to limit the Changba Road, Huishan District, Wuxi, Jiangsu, China) for
motion path. In the trailing edge, the lower skin and the each stepper motor, which is mounted inside the middle
compliant mechanisms need to be actuated separately. And fuselage, between the two ribs. The input pulses of step
there is a relative sliding between the lower skin and the drivers are provided by the pulse width modulation which
compliant mechanisms. Thus, two actuators are needed is outputted by the STM32 microcontroller. The position is
here: one for the two compliant mechanisms and one for commanded to move and hold at one of these pulses (an
the actuation of lower skin. Besides, two actuation systems open-loop controller). Hence, there is no position sensor for
can realize more precise control and more kinds of morph- feedback in our actuation system, except four limit
ing shapes. switches which are used to limit the travel of the motors.
Note that, the morphing wing designed in this study is Besides, in the demonstrator manufacturing, many stan-
mainly for the deformation testing purpose. Hence, the dard parts are adopted for convenience, hence the weight of
actuation is provided by stop motor as its ease of operation. the demonstrator is overweight for the micro air vehicle. In
The static moment of the stepper motor is 0.12 NM, and the study of a compliant mechanism-based leading edge by
the lead of screw is chosen as 1 mm for larger transmission FlexSys Inc. and AFRL,25 its compliant mechanisms offer
ratio. As the largest input force is 40 N (the leading edge), 22% decrease in weight over the current flap structure.
the motor can work at 0.08 NM torque with 2200 pulse per Although its total weight increased 7%, mainly from the
second, according to the speed–torque characteristics of the rotary actuation scheme, the wing morphing system can
step motor. overcome the weight penalty well. The same benefits are
10 International Journal of Advanced Robotic Systems

Figure 13. Deformation process of the leading edge. deflection angle of (a) 0 , (b) 18 , and (c) 27 .

Considering the leading-edge skin is actually deviating

from the initial curve, the deformation is almost identical
with the target curve, with only a small difference around
the leading-edge bottom. The partial enlarged drawing of
leading-edge bottom part is also shown in Figure 14.
For the trailing edge, several typical deflections in the
deformation process are displayed in Figure 15, including
upward deflection in Figure 15(a) and downward deflection
in Figure 15((c) and (d)). Noted that, though no upward
deflection for trailing edge is considered in the design, the
trailing edge can deflect upward by driving the compliant
mechanism reversely.
The measured deformation curves for the trailing edge
with different deflection angles are shown in Figure 16.
The deflection angle of trailing edge is defined by the line
which is connected to the tip point of the trailing edge and
Figure 14. Measured deformation of leading edge under different the middle point of the interface between the trailing edge
deflection angles (0 –27 ).
and the middle wing. The deflected curve coincides with
the target curve quite well except the tip of the trailing
expected in our design, due to that the wing is manufac- edge, and the error coincides with the error distribution
tured mainly by light composite materials and flexible PLA of the trailing-edge skin, as shown in Figure 7.
material, with only the outer skin of steel with 0.15 mm Although the target is set to 24 downward deflection
thickness. And the weight of the rotary actuation scheme
angle, the designed leading edge can deflect to a maximum
can decrease by adopting more advanced actuation sys-
of 40 , in which the input displacement of the compliant
tems, such as the linear motors or smart materials. The
mechanism actuation is 38.73 mm and the lower skin actua-
actuation system upgrade and the weight reduction are
tion is 48.15 mm. For the upward deflection, the largest
important parts in our next work.
deflection angle is 8 , with the compliant mechanisms
input displacement of 5.38 mm and the lower skin actua-
Displacements testing tion input displacement of 9.74 mm.
The deformation of the morphing wing was measured by a Table 3 shows the comparison of the proposed wing
displacement measurement system, consisting of two with some typical variable camber morphing wings, which
orthogonal grating rulers and a laser for positioning, shown are presented here due to their large deformation properties.
in Figure 12. The morphing leading edge with different Note that, the deflection angles in other research are not all
deflection angles is displayed in Figure 13. presented in their study, some of them are measured by the
For the leading edge, when the largest displacement pictures in their article. The morphing wings in columns
actuation of 48 mm is actuated by the screw, the largest 1–9 are all designed by compliant mechanisms, in which
deflection angle can be up to 27 . The angle is defined by the difference is the compliant mechanism design method
the deflection of a line, which line is connecting the min- they take. The wings in the last three columns are designed
imum curvature point of the leading edge and the middle by other approaches, such as smart materials and rigid-
point of the interface between the leading edge and the body mechanisms. The proposed design owns the largest
middle wing. The deformation curves of four deflection downward deflection in both leading- and trailing-edge
angles (0 , 9 , 18 , and 27 ) are shown in Figure 14 and design, and usually, the large deflection means the large
plotted by the spline interpolation of the measured points. change of the aerodynamic performance.
Zhang et al. 11

Figure 15. Deformation process of the trailing edge: (a) upward deflection angle of 8 , (b) deflection angle of 0 , (c) downward
deflection angle of 24 , and (d) downward deflection angle of 40 .

the aerodynamic load bearing capacity quite well in the

design condition, shown in the “Aerodynamic load testing
simulation” section. Hence, all the deformed airfoils are
assumed to be rigid ones in the simulation, without consid-
ering the extra deformation due to the aerodynamic load.
The meshes of the camber deflection wing were gener-
ated by ANSYS ICEM (Ansys Inc./ANSYS 17.0), and the
aerodynamic characteristics were evaluated by ANSYS
Fluent (Ansys Inc./ANSYS 17.0). The standard k–e model
and SIMPLEC scheme (with standard pressure, least
squares cell-based gradient, second-order upwind, first-
order upwind turbulent kinetic energy, and first-order
upwind turbulent dissipation rate) are used for the turbu-
lence model and the solution method. Freestream velocity
80 m/s, air density 1.225 kg/m3 , viscosity 1.7894 
Figure 16. Measured deformation of trailing edge under different
105 Pa  s, and Reynolds number 5.52  106 are taken in
deflection angles (8 to 40 ). the simulation. Besides, no-slip condition of airfoil wall
and free stream flow of far-field boundary are adopted.
Note that the deflection angles in Table 3 are not all In the simulation, different attack angles are analyzed,
presented in their research articles, thus some of them are from 20 to 20 . The deflection angle of leading edge
measured by the provided pictures. The regions of leading changes from 0 to 27 , and the deflection angle of trailing
and trailing edge are defined as the region where the defor- edge changes from 8 to 40 . The simulation needs to
mation occurs. And their deflection angles are defined as repeat many times when analyzing the situations with dif-
the same as the definition mentioned above. ferent attack angles. Hence, codes by MATLAB are taken
to generate the journal file of Fluent, then call the Fluent to
execute the journal and save the results.
Aerodynamic performance analysis Figure 17 shows the development of pressure coefficient
The aerodynamic performance of the morphing wing with (Cp) of 12 attack angle versus leading-edge deflection angle
deformed leading and trailing edges is analyzed by simula- and trailing-edge deflection angle. The deflection of leading
tion in airfoil level here. Note that, although the morphing and trailing edges can change the Cp distribution signifi-
wing is designed by the compliant mechanisms, it can meet cantly. The deflection of leading edge mainly changes the
12 International Journal of Advanced Robotic Systems

Table 3. Deflection comparison of the proposed wing over some typical variable camber morphing wings

Deflection angle

Reference Approach LE TE Experiment

  
The proposed wing Hyperelastic material topology optimization 27 8 /þ40 Wing level
MAC-Wing8,25 Continuum material topology optimization — +10 Wing level
Diaconu et al.30 Bistable laminated composite structures — 9.4 Simulation
Kota et al.49 Continuum material topology optimization 10 — Wing level
Ricci et al.31 Load-path topology optimization method 20.5 — Simulation
VCCW50 Material-based topology optimization 8.6 4.5 Wing level
Maute and Reich21 Material-based topology optimization 12.3 2.1 Simulation
Vasista et al.29 Load-path topology optimization method 10.1 6.8 /þ8.0 Wing level
Smart Droop Nose26 Traditional rigid-body mechanisms 20 — Wing level
DARPA51 Eccentuator and piezomotor-based actuation — +20 Wing level
Yokozeki et al.52 Designed by flexible corrugated structures — 35 Wing level

Figure 17. Pressure coefficient (Cp) of the morphing wing. The Cp under different deflection angles of (a) leading edge and
(b) trailing edge.

load distribution around the leading edge, while the deflection impact on the Cd than the Cl. In the positive attack angle part,
of trailing edge changes the load distribution almost every- the Cd decreases along with the deflection angle, while in the
where. And the conclusion is valid for other attack angle negative part, the Cd increases along with the deflection angle.
situations. Some fluctuations of Cp exist around Length/ The lift–drag ratio (Cl/Cd) is shown in Figure 18(c). The
Chord ¼ 0.2 in Figure 17(a) and Length/Chord ¼ 0.9 in lift–drag ratios of each attack angles decrease at first, then
Figure 17(b), where the deflection starts. increase in the middle and decrease at last, just like the sine
curve. In the attack angle between 4 and 6 , the larger the
Aerodynamic analysis of leading-edge deflection deflection angle of the leading edge is, the lower the lift–
The Cl and drag coefficient (Cd) of the wing with variable drag ratio, and contrary in other angles. The non-deflecting
leading-edge deflection under different attack angles are leading edge owns the largest lift–drag ratio around zero
shown in Figure 18(a) and (b). The deflection of the leading attack angle, and the lowest lift–drag ratio is also at the
edge has little effect on the Cl, especially when the attack non-deflecting situation at the attack angle of 10 .
angle is greater than 4 and less than 10 . When the attack Thus, for the takeoff phase in large attack angle, 27
angle is less than 4 or greater than 10 , the lift increases deflection angle owns relative high lift and low drag. For
along with the deflection angle slightly, and the tendency is high lift in the climbing phase, large attack angle and large
more distinct in the negative attack angle part. deflection of leading edge should be taken. Non-deflecting
The Cd decreases along with the attack angle in the nega- leading-edge and zero attack angle are more suitable for the
tive attack angle part and increases in the positive part like a U- cruising phase, which owns the highest lift–drag ratio. In
shaped curve. The leading-edge deflection shows higher the plane descending phase in negative attack angle part,
Zhang et al. 13

Figure 18. Aerodynamic performances of morphing airfoil under Figure 19. Aerodynamic performances of morphing airfoil under
different leading-edge deflection angles. (a) Lift coefficient (Cl), different trailing-edge deflection angle. (a) Lift coefficient (Cl),
(b) drag coefficient (Cd), and (c) lift to drag ratio (Cl/Cd) of (b) drag coefficient (Cd), (a) lift to drag ratio (Cl/Cd) of different
different attack angles. attack angle.

large deflection of leading edge can reduce lift and increase

drag, and in the landing phase, large positive attack angle Aerodynamic analysis of trailing-edge deflection
and large downward deflection of leading edge are more The Cls of the wing with variable trailing-edge deflection
appropriate. Besides, non-deflecting of leading edge at under different attack angles are shown in Figure 19(a).
10 attack angle is more beneficial for the nosedive. The Cl increases along with the attack angle when the
14 International Journal of Advanced Robotic Systems

deflection angle is under 16 . When the deflection angle of

trailing edge is larger than 16 , the increase of Cl may be
stagnated along with the increase of attack angle, some
even decreases. For example, the Cl of 40 deflection angle
decreases in 4 –10 attack angle and then keeps steady
roughly. The Cds of different camber trailing edges are
U-shaped curves, with the attack angle of the lowest Cd
decreasing gradually with deflection angle, as displayed in
Figure 19(b).
The lift–drag ratios of different trailing-edge deflection
angles are shown in Figure 19(c). For the deflection angle
of 8 to 8 , there are three phases of the lift–drag ratios: Figure 20. Deformation and von Mises Stress of fluid–structure
decreasing at first, then increasing in the middle, and interaction for leading edge.
decreasing at last. For other deflection angles, the lift–drag
ratios own only two phases: increasing in the first half and values are 206 GPa for the spring steel and 300 MPa for
decreasing in the last half. the flexible compliant mechanism material.
Thus, the trailing edge with 40 downward deflection in When taking the aerodynamic load of the wing in non-
attack angle around 4 provides large lift and relative deflecting and zero attack angle situation, the deformations
small drag, which is suitable for the takeoff phase. For of leading and trailing edges are very small (the largest
the climbing phase, larger trailing-edge deflections are displacement is only 0.35 mm in leading edge and
more applicable to the situation when attack angles are 0.18 mm in trailing edge), and so does the stress (largest
small (0 –8 ), and smaller deflections are more appropri- von Mises stress is only 6.23 MPa in leading edge and
ate when attack angles are large (8 –20 ). Non-deflecting 0.42MPa in trailing edge).
trailing-edge and zero attack angle are suitable for the When considering the wing deflection, the largest varia-
cruising phase. In the plane descending phase, upward tion of the aerodynamic load is close to five times during
deflection of trailing edge can reduce lift and increase the deformation. Thus, five times aerodynamic load of the
drag, and in the landing phase, large attack angle and large no deformation and zero attack angle situation is applied to
downward deflection of trailing edge owns the largest
the wing in simulation. The deformation and von Mises
drag and relatively large lift.
stress of fluid–structure interaction for leading edge are
shown in Figure 20. In this situation, the largest displace-
ment is 1.76 mm, and the von Mises stress is only 31.44
Aerodynamic load testing simulation MPa in the leading edge. Besides, if considering the actual
Compared to the traditional wing, the designed compliant situation (several glass fiber plates sticking inside and pres-
morphing wing is more flexible, and its aerodynamic load tress in the skin, and larger prestress after deformation), the
bearing capability is relative lower. Thus, a one-way fluid– deformation would be lower. Thus, comparing to the
structure interaction coupling method is used to investigate 175.20-mm chord length of leading edge and the yield
the aerodynamic load performance. This simulation is exe- strength of over 100 MPa of flexible material and 500 MPa
cuted by ANSYS Workbench. Through APDL commands of spring steel, the designed compliant leading edge can
of ANSYS, the aerodynamic load is applied to the skin by meet the aerodynamic load bearing capacity quite well in
the pressure load of Fluent. the design condition.
In this modal, the span of the leading and trailing edges For the trailing edge, actual geometric dimensions of the
is 65 mm (half-width of the demonstrator in the skin are set in the simulation. When five times aerodynamic
“Manufacturing and testing” section), and the thickness load is applied, the largest deformation of trailing edge is
of the compliant mechanism is 10 mm. The skin and com- only 0.91 mm (304-mm chord length of trailing edge) and
pliant mechanism are connected by bonded contacts in the largest von Mises stress is only 2.09MPa, which is
Workbench (Ansys Inc./ANSYS 17.0). The fixed and actu- wholly satisfactory in the design. The deformation and von
ated parts of the compliant mechanism are all fixed in the Mises stress of fluid–structure interaction for trailing edge
load bearing test, as well as the two ends of the up and are shown in Figure 21.
down skins which are connected to the middle wing. Besides, many other aeroelastic phenomena are also
Noted that, the leading-edge skin is composed of a very important for the practical application of the
spring-steel plate (0.15 mm thickness) and several glass morphing wing, such as the control reversal, flutter,
fiber plates. To simplify the simulation model, a buffeting, and vortex-induced vibration. The detailed
0.15-mm thick spring-steel plate is adopted to replace the aeroelasticity part will be our next work, by cooperating
leading-edge skin which is originally built by bent lami- with our project partners of AVIC Aerodynaiviics
nated skin with prestress inside. The Young’s modulus Research Institute.
Zhang et al. 15

Figure 21. Deformation and von Mises Stress of fluid–structure interaction for trailing edge.

Conclusions References
A variable camber morphing wing is designed in this study, 1. Vasista S, Tong LY, and Wong KC. Realization of morphing
which shows excellent capability in realizing large deflec- wings: a multidisciplinary challenge. J Aircraft 2012; 49:
tion of leading and trailing edges. To fit the curvature 11–28.
changing phenomenon in the morphing leading edge, a 2. Barbarino S, Bilgen O, Ajaj RM, et al. A review of morphing
laminated skin is designed by a bending-shape design aircraft. J Intell Mater Syst Struct 2011; 22: 823–877.
method. The skin-driving compliant mechanism is 3. Li DC, Zhao SW, Da Ronch A, et al. A review of modelling and
designed by a meshless-based hyperelastic structure topol- analysis of morphing wings. Prog Aerosp Sci 2018; 100: 46–62.
ogy optimization. Demonstrators of the leading and trailing 4. Bowman J, Sanders B, Cannon B, et al. Development of next
edges were fabricated and testified, and their deformation generation morphing aircraft structures. In: Proceedings of
shows satisfying agreement with the target deformed air- 48th AIAA/ASME/ASCE/AHS/ASC structures, structural
foil. Comparing with other variable camber morphing dynamics, and materials conference, Honolulu, Hawaii,
wings, the design morphing wing can realize larger deflec- 23–26 April 2007. DOI:10.2514/6.2007-1730.
tion—up to 27 of leading edge and 40 of trailing edge. 5. Burnazzi M and Radespiel R. Assessment of leading-edge
The aerodynamic analysis was implemented by Fluent, and devices for stall delay on an airfoil with active circulation
the simulation shows that the airfoil benefits a lot from its control. CEAS Aeronaut J 2014; 5: 359–385.
deformation. Besides, a fluid–structure interaction simula- 6. Jo Y, Choi S, Zientarski L, et al. Aerodynamic characteristics
tion was carried out, and the design exhibits high perfor- and shape optimization of a variable camber compliant wing.
mance in resisting aerodynamic and structural loadings in In: Proceedings of 34th AIAA applied aerodynamics confer-
the simulation. The presented aerodynamic analysis should ence, Washington, DC, 13–17 June 2016. DOI: 10.2514/6.
be confirmed by experimental results, and a scheduled 2016-3416.
experiment of a 2-D airfoil section model is prepared in a 7. Wang WH, Liu PQ, Tian Y, et al. Numerical study of the
low-turbulence wind tunnel (40 m/s). aerodynamic characteristics of high-lift droop nose with the
deflection of fowler flap and spoiler. Aerosp Sci Technol
2016; 48: 75–85.
Declaration of conflicting interests 8. Hetrick J, Osborn R, Kota S, et al. Flight testing of mission
The author(s) declared no potential conflicts of interest with adaptive compliant wing. In Proceedings of 48th AIAA/
respect to the research, authorship, and/or publication of this ASME/ASCE/AHS/ASC structures, structural dynamics, and
article. materials conference, Honolulu, Hawaii, 23–26 April 2007.
DOI:10.2514/6.2007-1709.
9. Xinxing T, Wenjie G, Chao S, et al. Topology optimization of
Funding
compliant adaptive wing leading edge with composite mate-
The author(s) disclosed receipt of the following financial support
rials. Chinese J Aeronaut 2014; 27: 1488–1494.
for the research, authorship, and/or publication of this article: This
work was supported by National Key Research and Development 10. Gabor OS, Koreanschi A, Botez RM, et al. Numerical simu-
Program of China (2017YFB1300101) and National Natural Sci- lation and wind tunnel tests investigation and validation of a
ence Foundation of China under grant no. 51375383. morphing wing-tip demonstrator aerodynamic performance.
Aerosp Sci Technol 2016; 53: 136–153.
11. Coutu D, Brailovski V, Terriault P, et al. Aerostructural
ORCID iD model for morphing laminar wing optimization in a wind
Yaqing Zhang https://orcid.org/0000-0001-8546-5795 tunnel. J Aircraft 2011; 48: 66–76.
16 International Journal of Advanced Robotic Systems

12. Bowman J, Sanders B, Cannon B, et al. Development of next materials conference, Palm Springs, California, 4–7 May
generation morphing aircraft structures. In: 48th AIAA/ 2009. DOI: 10.2514/6.2009–2128.
ASME/ASCE/AHS/ASC structures, structural dynamics, and 27. Santer M and Pellegrino S. Topological optimization of
materials conference, Honolulu, Hawaii, 23–26 April 2007. compliant adaptive wing structure. AIAA J 2009; 47:
DOI: 10.2514/6.2007-1730. 523–534.
13. Siddall R, Ancel AO, and Kovac M. Wind and water tunnel 28. Xin K, Ge WJ, and Zhang YH. Topology optimization of
testing of a morphing aquatic micro air vehicle. Interface compliant adaptive leading edge with geometrically nonli-
Focus 2017; 7: 15. nearity. Adv Sci Lett 2011; 4: 2306–2310.
14. Vos R, Gurdal Z, Abdalla M, et al. Mechanism for warp- 29. Vasista S, De Gaspari A, Ricci S, et al. Compliant structures-
controlled twist of a morphing wing. J Aircraft 2010; 47: based wing and wingtip morphing devices. Aircr Eng Aerosp
450–457. Technol 2016; 88: 311–330.
15. Rodrigue H, Cho S, Han MW, et al. Effect of twist morphing 30. Diaconu CG, Weaver PM, and Mattioni F. Concepts for
wing segment on aerodynamic performance of UAV. J Mech morphing airfoil sections using bi-stable laminated compo-
Sci Technol 2016; 30: 229–236. site structures. Thin Walled Struct 2008; 46: 689–701.
16. Love M, Zink S, Stroud R, et al. Impact of actuation concepts 31. Ricci S, Gaspari AD, and Gilardelli A. Design of a leading
on morphing aircraft structures. In: Proceedings of 45th AIAA/ edge morphing based on compliant structures for a twin-prop
ASME/ASCE/AHS/ASC structures, structural dynamics, and regional aircraft. In: 2018 AIAA/AHS adaptive structures con-
materials conference, Palm Springs, California, 19–22 April ference, Kissimmee, Florida, USA, 8–12 January 2018. DOI:
2004, DOI: 10.2514/6.2004-1724. 10.2514/6.2018-1063.
17. Di ML, Mintchev S, Heitz G, et al. Bioinspired morphing 32. Sigmund O. On the design of compliant mechanisms using topol-
wings for extended flight envelope and roll control of small ogy optimization. Mech Based Des Struct 1997; 25: 493–524.
drones. Interface Focus 2017; 7: 20160092. 33. Wang YQ and Luo Z, Zhang XP, et al. Topological design of
18. Monner HP, Hanselka H, and Breitbach EJ. Development and
compliant smart structures with embedded movable actua-
design of flexible Fowler flaps for an adaptive wing. In:
tors. Smart Mater Struct 2014; 23: 15.
Proceedings of fifth annual international symposium on
34. Bruns TE and Tortorelli DA. Topology optimization of non-
smart structures and materials, San Diego, CA, USA, 1–5
linear elastic structures and compliant mechanisms. Comput
March 1998, pp. 60–70, DOI: 10.1117/12.310673.
Method Appl Mech Eng 2001; 190: 3443–3459.
19. Zimmer H. Quertriebskörper mit veränderbarer profillierung,
35. Luo Y, Wang MY, and Kang Z. Topology optimization of
insbesondere flugzeugtragflügel. DE2907912A1, Germany,
geometrically nonlinear structures based on an additive
1979.
hyperelasticity technique. Comput Method Appl Mech Eng
20. Li B and Li G. Analysis and optimization of a camber morph-
2015; 286: 422–441.
ing wing model. Int J Adv Robot Syst 2016; 13: 1–17.
36. Cho SH and Kwak J. Topology design optimization of geo-
21. Maute K and Reich GW. Integrated multidisciplinary topol-
metrically non-linear structures using meshfree method.
ogy optimization approach to adaptive wing design. J Air-
Comput Method Appl Mech Eng 2006; 195: 5909–5925.
craft 2006; 43: 253–263.
37. Du YX, Chen LP, and Luo Z. Topology synthesis of geome-
22. Sofla AYN, Meguid SA, Tan KT, et al. Shape morphing of
aircraft wing: status and challenges. Mater Des 2010; 31: trically nonlinear compliant mechanisms using meshless
1284–1292. methods. Acta Mech Solida Sin 2008; 21: 51–61.
23. Hao GB, Yu JJ, and Li HY. A brief review on nonlinear 38. He QZ, Kang Z, and Wang YQ. A topology optimization
modeling methods and applications of compliant mechan- method for geometrically nonlinear structures with meshless
isms. Front Mech Eng 2016; 11: 119–128. analysis and independent density field interpolation. Comput
24. Vasista S, Riemenschneider J, and Monner HP. Design and Mech 2014; 54: 629–644.
testing of a compliant mechanism-based demonstrator for a 39. Wang Y, Luo Z, Wu JL, et al. Topology optimization of
droop-nose morphing device. In: Proceedings of 23rd AIAA/ compliant mechanisms using element-free Galerkin method.
AHS adaptive structures conference, Kissimmee, Florida, Adv Eng Softw 2015; 85: 61–72.
5–9 January 2015. DOI: 10.2514/6.2015-1049. 40. Zhang YQ, Ge WJ, Tong XX, et al. Topology optimization of
25. Kota S, Hetrick J, Osborn R, et al. Design and application of structures with coupled finite element–element-free Galerkin
compliant mechanisms for morphing aircraft structures. In: method. Proc Inst Mech Eng C-J Mech Eng Sci 2018; 232:
Anderson EH (ed.) Smart structures and materials 2003: 731–745.
industrial and commercial applications of smart structures 41. Zhang Y, Ge W, Zhang Y, et al. Topology optimization
technologies. Bellingham: SPIE-The International Society method with direct coupled finite element–element-free
for Optical Engineering, 2003, Vol. 5054, pp. 24–33–. Galerkin method. Adv Eng Softw 2018; 115: 217–229.
26. Monner H, Kintscher M, Lorkowski T, et al. Design of a 42. Zhang YQ, Ge WJ, Zhang YH, et al. Topology optimization
smart droop nose as leading edge high lift system for trans- of hyperelastic structure based on a directly coupled finite
portation aircrafts. In: Proceedings of 50th AIAA/ASME/ element and element-free Galerkin method. Adv Eng Softw
ASCE/AHS/ASC structures, structural dynamics, and 2018; 123: 25–37.
Zhang et al. 17

43. Kintscher M, Wiedemann M, Monner HP, et al. Design of a optimization of finite strain elastic problems. Comput Method
smart leading edge device for low speed wind tunnel tests in Appl Mech Eng 2014; 276: 453–472.
the European project SADE. Int. J Struct Integr 2011; 2: 49. Kota S, Ervin G, Osborn R, et al. Design and fabrication of an
383–405. adaptive leading edge rotor blade. In: Proceedings of Annual
44. Vasista S, Riemenschneider J, Kamp BV, et al. Evaluation of Forum Proceedings-American Helicopter Society, Montreal,
a compliant droop-nose morphing wing tip via experimental Canada, April 29–May 1, p. 2178.
tests. J Aircraft 2017; 54: 519–534. 50. Joo JJ, Marks CR, Zientarski L, et al. Variable camber com-
45. Buhl T, Pedersen C, and Sigmund O. Stiffness design of pliant wing-design. In: Proceedings of 23rd AIAA/AHS adap-
geometrically nonlinear structures using topology optimiza- tive structures conference, Kissimmee, Florida, 5–9 January
tion. Struct Multidiscip Optim 2000; 19: 93–104. 2015. Kissimmee, Florida, 5–9 January 2015. p. 1050.
46. Svanberg K. The method of moving asymptotes: a new 51. Kudva JN. Overview of the DARPA smart wing project.
method for structural optimization. Int J Numer Meth Eng J Intell Mater Syst Struct 2004; 15: 261–267.
1987; 24: 359–373. 52. Yokozeki T, Sugiura A, and Hirano Y. Development and
47. Belytschko T, Lu YY, and Gu L. Element-free Galerkin wind tunnel test of variable camber morphing wing. In: Pro-
methods. Int J Numer Meth Eng 1994; 37: 229–256. ceedings of 22nd AIAA/ASME/AHS adaptive structures con-
48. Wang FW, Lazarov BS, Sigmund O, et al. Interpolation ference, National Harbor, Maryland, 13–17 January 2014. p.
scheme for fictitious domain techniques and topology 1261, DOI: 10.2514/6.2014-1261.

Appendix 1

Notation

AMWT Adaptive morphing wingtip nTotal number of the both FE and EFG nodes
AFRL US Air Force Research Laboratory pPenalty factor of material density
Cp Pressure coefficient q By-product of the Lagrange multipliers
Cl Lift coefficient R Residual of structural equilibrium
Cd Drag coefficient R0Unloaded condition
Cl/Cd Lift–drag ratio t*Traction on the boundary after deformation
DARPA US Defense Advanced Research Projects Agency uiout
Output displacement of the compliant
mechanism
GA Genetic algorithm u Trial function
MAV Micro air vehicle uj Displacement of the jth output port
MMA Method of moving asymptotes V Material volume of the design domain
PLA Polylactic acid V0 Volume of the design domain
Pps Pulse per second W Hyperelastic material model
POLIMI Politecnico di Milano xj Original positions of the jth output port
PWM Pulse width modulation xj Deformed positions of the jth output port
SIMP Solid isotropic material with penalization λ Vector of Lagrangian multipliers
VCCW Variable camber compliant wing r Material density of topology optimization
b Body force distribution ri Nodal density
C Objective function of topology optimization rmin Lower bound of density
E Young’s modulus τ Kirchhoff stress
fV Prescribed fraction of the volume constraint G t Traction boundary
Fi Compliant mechanism output force ’i Shepard function
G By-product of the Lagrange multipliers i Material constant of hyperelastic material
kiout Stiffness of the skin i0 Solid material constant of hyperelastic material
KT Tangent stiffness matrix dv Virtual velocity fields
L Vector with value of one for position j and zeros for other positions dL Virtual velocity gradients
m Total number of the output ports of the compliant mechanism

Appendix 2 the elasticity of the laminated skin and meanwhile reduce

The skin of leading edge is designed by a bent laminated the weight, spring steel and glass fiber-reinforced polymer
skin, aiming to suit the curvature changing during the composites are adopted in this study: the spring steel would
leading-edge deflection. In this design, sheets of different be laid on the outermost layer; glass fiber-reinforced poly-
materials are sticking together, and each layer has the same mer composites with different length and thickness would
width, variable length, and variable thickness. To ensure be laid inside.
18 International Journal of Advanced Robotic Systems

Table 2A. Design parameters of laminated leading-edge skin.

Thickness of each layer (mm)

E (GPa) B1 B2 B3 B4 B5
1st layer 197.55 0.15
2nd layer 53.15 0.8 0.5
3rd layer 53.15 0.8
4th layer 53.15 1.0
Actual — 0.95 0.15 0.65 1.45 2.45
Converted 0.1 4.91 2.02 3.82 6.71 10.33

Figure 2A. Schematic diagram of the design process for Table 2B. Results of laminated leading-edge skin optimization.
laminated leading-edge skin.
Parameter Values
Length of B1–B5 (mm) 128.11, 84.32, 55.31, 68.00, 17.43
Moment (N/m) 20.00

based model by converting the mechanical properties of

materials in different layers into corresponding thicknesses,
showing in Figure 2B.
The layer distribution of the laminate skin is predicted
by the curvature along the leading edge, and a rough guide-
Figure 2B. Simplified model for the response analysis of line is laying the thicker composites on the smaller curva-
laminated skin.
ture part. Considering the dimension of the ready-made
There are three states of the laminated skin: without any materials and their deformation scales, spring steel with
load, there is only a straight laminated plate; with external 0.15 mm thickness and glass fiber-reinforced polymer com-
moment applying to the ends of the straight plate, the lami- posites of 0.5 mm, 0.8 mm, and 1.0 mm is adopted. Differ-
nated plate will be bent rightly to the shape of the leading ent combinations of the layered compositions are tested in
edge; with external moment applying to the ends of the skin advance and the predicted thickness of each layer is pre-
and appropriate external forces applying to appropriate posi- sented in Table 2A. Taking Young’s modulus (E) of the
tions of the leading edge, the laminated skin will be deformed uniform material as 100 MPa, the thicknesses of beam
rightly to the shape of the target deforming leading edge. elements in the converted model is also listed in this table.
Figure 2A gives the diagrammatic sketch of this design. The optimization is implemented by GA in MATLAB,
The laminated skin is designed by a bending-shape and the variable cross-section beam response analysis is
design method, in which the design variable is the layer provided by ANSYS. The optimal parameter of the lami-
length of the laminated skin and the objective function is nated leading-edge skin is presented in Table 2B, including
the residual sum of squares of the bending curve and the the length of each beam element and the bending moment.
leading-edge curve. In the bending-shape design method, a The deformation of the variable cross-section beam is
uniform material is adopted to simplify the beam element- shown in Figure 2C(a), as well as the von Mises stress. The

Figure 2C. Deformation of bent leading-edge skin: (a) deformation and von Mises stress and (b) displacement error.
Zhang et al. 19

curve of the bent beam matches quite well with the target laminated skin is shown in Figure 2C(b), with the
curve, and the displacement error distribution of the bended maximum error of 8.13 mm.

Appendix 3

Start Start

Input the aerodynamic load, original airfoil

and target airfoil Material and thickness of compliant
mechanism;Support style and position;
Assign the initial property values of
Modal and discretize skin by beam elements Input the geometric modal of design variables; and so on
in ANSYS; Implement displacement trailing-edge skin and the
boundary; Assign initial values and ANSYS modal of leading-
constraints of design variables edge skin(from Appendix A) Discretize the design domain by meshless Input the position, value,
method for the given boundary and direction, and displacement
of the driving forces of
Implement the position, value, direction of loading conditions
compliant mechanism
driving force into loading conditions of FE
modal
Perform structural analysis by non-linear
meshless method of hyperelastic structure
Geometrically nonlinear analysis of beam Update the design variables
modal by ANSYS
Update the design variables
Calculate the objective function,
Calculate the objective function material volume and their sensitivity
Optimization by GA
Optimization by MMA
No No
Convergence Convergence

Yes
Yes
Output the position, value, direction, and
displacement of the driving forces of Output the optimal topology
compliant mechanism configuration

End
End

Figure 3A. Flowchart of the driving forces optimization for the

Figure 3B. Flowchart of the topology optimization for the
leading-edge and trailing-edge skin.
compliant mechanisms.