Proceedings of PACAM XI
Copyright © 2009 by ABCM
11th Pan-American Congress of Applied Mechanics
January 04-08, 2010, Foz do Iguaçu, PR, Brazil
DEVELOPMENT OF BIO-MIMETIC MORPHING SKINS FOR
MICRO-UAVS
Laila Asheghian, lasheghian@nextgenaero.com
Jeff Street, jstreet@nextgenaero.com
NextGen Aeronautics, 2780 Skypark Drive, Suite 400, Torrance, CA 90505
Dr. Jay Kudva, jkudva@nextgenaero.com
NextGen Aeronautics, 2780 Skypark Drive, Suite 400, Torrance, CA 90505
K. Raymond Olympio, kro125@psu.edu
Dr. Farhan Gandhi, fgandhi@engr.psu.edu
Department of Aerospace Engineering, The Pennsylvania State University, University Park, PA 16802
Abstract. This paper presents work on the optimization of NextGen Aeronautics’ successful shear-morphing wing skin
design, consisting of high-strain silicone “facesheets” supported by an intricate aluminum under-structure made of
thin, closely-spaced aluminum ribbons or “strands” glued to the face-sheets. The strand design was optimized through
ANSYS/non-linear finite element simulation and experimentation, resulting in a wing structure with greatly reduced
actuation forces and much simpler manufacturing process, while still maintaining out-of-plane airfoil shape and
limiting overall weight. The design variables included the pre-strains in the face-sheet and the geometric parameters
defining a panel and its subcomponents such as spacing, thickness, depth, angle, etc. The key to optimizing the design
was to reduce high levels of strain in the strand by incorporating a central strain-relieving feature in each strand. With
these low-strain strands, the facesheet became the largest contributor to the actuation force, so correspondingly,
ANSYS simulations were run with strand and facesheet together to optimize the design for minimum energy while
maintaining airfoil profile (not violating out-of-plane displacement criteria or causing the facesheet to wrinkle). The
final optimal flexible skin strand design for a shear-morphing wing uses a Gaussian shape. Though originally designed
for use with a larger UAV, this design was further modified for smaller micro-UAV (MAV) scales, specifically for
morphing/perching bio-mimetic flight. For the MAV application, the vision is to develop a single-DOF morphing wing
with two stable states – one for minimum drag cruise flight and the other for high L/D landing/perching.
Keywords: morphing aircraft, bio-mimetic structures, non-linear finite element analysis, large-strain morphing skins,
micro-UAVs.
1. INTRODUCTION
Over the past decade, morphing aircraft, i.e., aircraft whose shape, in particular wing geometry, can be dramatically
changed with a goal to optimize system level performance for varying flight conditions, have been under intense and
focused study. In this regard, under the DARPA sponsored Next Generation Morphing Aircraft Structures (N-MAS)
program (Andersen et al., 2007; Bowman et al., 2007; Flanagan et al., 2007; Herbert et al., 2007; Joshi et al., 2007;
Strutzenberg et al., 2007), NextGen Aeronautics developed a 16ft wing span wind tunnel model exhibiting over 200%
change in wing aspect ratio, 70% change in wing area and 40% change in span; this model was successfully tested at
the NASA LaRC TDT tunnel at Mach No. >0.9. A key innovation in this work was the development of “shearmorphing” wing skins (Fig. 1) capable of withstanding up to 400 psf air-loads while simultaneously undergoing shear
strains in excess of 60 degrees (Fig. 2). The initial concept consisted of high-strain silicone “facesheets” supported by
an intricate aluminum under-structure made of thin, closely-spaced aluminum ribbons or “strands” glued to the facesheets.
Figure 1. NextGen composite skin
While the design was successful, subsequent ongoing R&D projects, including the subject of this paper, are
addressing improvements in the morphing skin designs to reduce actuation loads and weight as well as simplify the
Proceedings of PACAM XI
Copyright © 2009 by ABCM
11th Pan-American Congress of Applied Mechanics
January 04-08, 2010, Foz do Iguaçu, PR, Brazil
manufacturing process. The basic unit of study in this research was a panel (Fig. 2), defined as a 508mm × 381mm (20”
× 15”) parallelogram that fits conveniently into NextGen’s previous design. This panel was correspondingly made up of
3 identical strips. The skin panel deformed between a shear angle of 15deg and a shear angle of 60deg. The undeformed position corresponded to a shear angle of 37.5deg. This represented a change in area of approximately 100%.
(a) 15 deg
(b) 60deg
Figure 2. Panel details and morphing extreme positions
2. STRAND DESIGN
Research began with basic cellular designs like chessboards, and used cellular material theory (CMT) (Gibson et al.,
1997) and non-linear finite element analysis (FEA) using ANSYS. Optimization returned a solution in which the cells
had very long and slender walls along one side. The slender walls could bend very easily as the skin panel was
subjected to shear morphing while the horizontal members provided additional stiffness to meet the out-of-plane
displacement requirement. Interestingly, this initial design was very similar to NextGen’s original N-MAS flex-skin
concept with strip support under the face-sheet. This implied that the problem of nonlinear geometric stiffening and the
correspondingly high-energy requirement encountered by the N-MAS skin for the full range of shear morphing would
be experienced here as well.
An FEA on the optimal design showed a stiffening of the cellular structure as the skin deformed to extreme shear
angles due to high curvature at the ends of the slender walls. To reduce this stiffening, a central softening element was
introduced at the center of the strands. This idea was partly based on intuition and partly based on previous work with
honeycomb structures, which have low extensional stiffness due to the bending of the hexagonal cells’ walls. The
central element allows shear deformation in the central region of the strand while providing sufficient extensional
stiffness along the strand direction. The first central element considered was a hexagon as shown in Fig.3.
Figure 3. Strand with a central hexagon, also shown in full panel configuration
2.1. Strand Evolution
The focus of the remainder of the research was on designing the central feature of the strand while also taking into
account the effect of the facesheet. As stated before, ANSYS was used extensively to evaluate the designs and account
for all geometric non-linearities. In the analyses, a unit was defined as the periodic unit needed to build a strip by
Proceedings of PACAM XI
Copyright © 2009 by ABCM
11th Pan-American Congress of Applied Mechanics
January 04-08, 2010, Foz do Iguaçu, PR, Brazil
repetition along the x-direction (Fig. 3). Thus a unit was composed of a strand and a bonded face-sheet. Periodic
boundary conditions were used so that the properties calculated were those of the strand in an infinite strip. Strands with
a hexagon as the central element were the first choice for the strand, and were parameterized as shown in Fig. 3 using
the length l of the inclined wall of the cell, cell angle θ , non-dimensional parameters α , δ and ε (wall lengths),
β , η (wall thicknesses), and depth. Optimization was according to Eq. (1) with an additional geometric constraint
given by (2) where
L is the total strand length, related to the strip width.
Minimize Work per unit width
min(σ 2 ) > 0
⎧
, i = 1...N
⎪
subject to ⎨ Fcritical > 19152 Pa
⎪U z (19152 Pa ) ≤ 2.54mm
⎩
(2δ + α + 2 sin θ )l = L
(1)
(2)
FEA revealed multiple issues with this design, and other strands considered (Fig. 4). Often, high deformation
gradients were found in regions of sharp transition between portions of strands. Additionally, pinch points were created
when large central sections came too close to each other, increasing the likelihood of wrinkling. Further details on the
specific progression of strand design can be found in Olympio et al. (2009).
Figure 4. Other strands considered included: ellipse, half ellipse, Gaussian, sine and cosine
2.2. Final Optimal Strand Design
Key features of the best designs included smooth contours and even spacing between strands. Parametric studies
were conducted using the three final types of strands shown in Fig. 4: Gaussian, sine and cosine. It was found that the
optimal strand was the Gaussian-shape, which proved to be 30% better (less work) than the straight strand which was
similar to the N-MAS design. The Von Mises strain in the strand was also reduced to 1.2% from 3.3%.
By taking into account the face-sheet, the straight strand and curved strand could be very competitive based on work
and out-of-plane displacement. However, the major advantage of the curved strands over the straight strand is the lower
local Von Mises stresses or strains in the strands. In effect, these lower local stresses widen the range of materials that
can be used for the substructure material. So, if material selection is an issue, it is preferable to have curved strands
(δ≠0). Since many traditional materials like aluminum cannot sustain the large strains seen in straight strands, this is a
critical design consideration for the N-MAS application.
Proceedings of PACAM XI
Copyright © 2009 by ABCM
11th Pan-American Congress of Applied Mechanics
January 04-08, 2010, Foz do Iguaçu, PR, Brazil
3. PANEL DESIGN
The next step of the design process was to integrate the optimized or best strand in a full strip (or panel). Contrary to
the previous study, in the analysis of a full strip the boundary effects due to the left and right edges were considered. As
a result, even if the same strands were used all along the strip, quantities like minor principal strain and out-of-plane
deflection changed along the length of the strip/panel.
Rather than doing a full optimization, which would have been computationally expensive, we considered a strip
made of only the best unit found in the previous section. Multiple variations on the basic strand design were considered.
3.1. Variable Amplitude
Analysis with the optimal strand used in the full strip/panel without modifications revealed negative minor principal
strains at the boundaries, particularly at the 60 deg morph position. This is thought to be due to the curved portion of the
extreme right strand coming very close to the boundary compared to other parts of the strand. Thus, it may be beneficial
to have a straighter strand at the right edge to improve the design with respect to wrinkling. Amplitude of the optimal
Gaussian strand was allowed to vary along the length of a strip, with the aim of finding the best strand’s amplitude (δi as
shown in Fig. 5) to reduce boundary effects.
Figure 5. Parameterization of strip and its strands.
We considered three possible variations in the strand amplitude (δ):
(1) δ decreased linearly from left to right starting at δ1=0.05 and ending at δΝ=0
(2) δ increased linearly from left to right starting at δ1=0 and ending at δΝ=0.05 or
(3) δ was δ1=δΝ=0 at both the left and right edges and maximum (δ10=0.05) in the middle of the strip
For the selected prestrain values, having a straighter strand on the right edge (variations 1 and 3) appeared beneficial
in reducing wrinkling as expected. Also for this particular case, it did not seem necessary to also have a straight strand
on the left side (variations 2 and 3) since wrinkling was not a problem there initially. However, if the pre-strains were
smaller, wrinkling could also occur on the left edge. In that scenario, we might revisit the other two cases. In all three
cases shown, the likelihood of wrinkling was considerably reduced on the edges with straight strands. This showed that,
due to the boundary effects, the optimal strip could be made of different strands at the boundaries and at its center. For
our particular application, we would select the straighter strand on just the right edge (variation 1, Fig. 6).
Figure 6. Minor principal strain distribution in a strip with linearly decreasing strand amplitude from left to right
One issue with this modification is the requirement for straighter strands, therefore increasing the Von-Mises strain
in some of the strands and narrowing the available materials that can meet those demands. For a lower-strain application
however, this could still be an attractive solution.
3.2. Variable End Spacing
Another option for modification of the base panel/strip design was the spacing between strands and straight end
walls on each side of the panel. Since wrinkling was most likely when morphing caused the strand to come close to the
Proceedings of PACAM XI
Copyright © 2009 by ABCM
11th Pan-American Congress of Applied Mechanics
January 04-08, 2010, Foz do Iguaçu, PR, Brazil
right hand side, additional space could alleviate this issue (Fig. 7). This also has the advantage of keeping the initial
geometry, and therefore not increasing strain in any of the individual strands by decreasing their curvature.
Figure 7. Reduction in number of strands leads to increased spacing, potential fix is ledge shape
One disadvantage of this method is that it can cause increased out of plane displacement in the larger unsupported
area. This can be fixed with the addition of a “ledge” tied to the end of the strip for support that does not need to bend
during morph (Fig. 7). The final optimal panel design will likely depend on the particular application, but combinations
of the above modifications should prove useful in detailed design.
4. MAV MODIFICATIONS
Though originally designed for use with a larger UAV, this design was further modified for smaller micro-UAV
(MAV) scales, specifically for morphing/perching bio-mimetic flight. In a basic perching maneuver, birds alter the
angle of attack of their wings, which helps to decrease forward velocity and even potentially create stall conditions. It is
thought that additional wing area could be beneficial in this application as well.
The vision is to develop a single-DOF morphing wing with two stable states – one for minimum drag cruise flight
and other for high L/D landing/perching. The initial concept involved direct attachment of a shear-morphing strip to the
body, with simple rotation at the forward to achieve altered angle of attack (Fig. 8).
Figure 8. Initial perching concept
However, it was thought that a combined approach, where the shearing action caused out of plane area change as
well, would be a more interesting option. With that goal, a strip was designed with variable length strands (Fig. 9),
increasing in length they move out in span. Although flat when in the sheared cruise flight condition, actuation to
remove the shear would create variable amounts of additional length, and this additional length/area can be constrained
to translate out of plane. Figure 9 also shows the flat sheared configuration in red, and the perching “unsheared”
configuration in blue.
Figure 9. Variable incline angles (top left) and lengths (bottom left) of MAV panel, representative sheared (red) and
perching (blue) wing areas, direction of actuation for perching wing
Essentially, this is a “variable periodic unit” approach, where each strand can be thought of as having a unique
length and incline angle. The shearing behavior of each strand should be similar to that of the UAV strands, and could
Proceedings of PACAM XI
Copyright © 2009 by ABCM
11th Pan-American Congress of Applied Mechanics
January 04-08, 2010, Foz do Iguaçu, PR, Brazil
be analyzed in a unit fashion. The results for each distinct strand length could then be summed to determine the
approximate global panel behavior.
Initial models of the MAV wing are under construction to demonstrate the motion seen above. Since airloads at the
MAV scale are significantly lower, many additional materials can be used to construct the wing. By using rapid
prototyping materials, complicated geometry and integrated joints can be easily created. The current models use various
axes of revolution to allow for out of plane displacement during shearing while keeping the strands perpendicular to the
XZ plane throughout the morph - no twist is induced, allowing for simpler comparison to the UAV results.
5. CONCLUSIONS AND FUTURE WORK
Through optimization with ANSYS, parameters that define morphing skin panels with strands and facesheets are
well understood at the UAV level. Strand- and panel-level modifications can be made to tailor the skin to various
loading and size constraints. Additionally, by examining the periodic unit with variability of strand parameters like
incline angle and length, the shear-skin concept can be applied at the MAV scale as well for unique applications such as
perching maneuvers.
In the future, analysis, build and test are planned for these MAV scale wings. While initial models make use of
multiple joints to achieve out of plane morphing, further study should allow for removal of the majority of these joints
by designing the strand to take both in plane and out of plane shearing loads and displacements. Modification of the
ANSYS code to include these additional system-level degrees of freedom could also allow for optimization of the entire
structure, without having to consider it as a sum of periodic units. MAV- scale perching wings align with AFOSR UAV
goals for near-term technology, with applications in various areas including urban surveillance and “hide-in-plain-site”
vehicles.
6. ACKNOWLEDGEMENTS
The authors wish to thank Dr. Victor Giurgiutiu, AFOSR program manager, and Dr. Gregory Reich, AFRL senior
aerospace engineer, for their technical guidance and sponsorship.
7. REFERENCES
Andersen, G., Cowan, D., and Piatak, D., “Aeroelastic Modeling, Analysis, and Testing of a Morphing Wing
Structure,” AIAA SDM Conference, AIAA-2007-1734, AIAA, Honolulu, HI, 2007.
Bowman, J., Sanders, B., Kudva, J., Joshi, S., and Weisshaar, T., “Development of Next-Generation Morphing Aircraft
Structures,” AIAA SDM Conference, AIAA-2007-1730, AIAA, Honolulu, HI, 2007.
Flanagan, J., “Development and Flight Testing of a Morphing Aircraft, the NextGen MFX-1,” AIAA SDM Conference,
AIAA-2007-1707, AIAA, Honolulu, HI, 2007.
Gibson, L. J. and M. Ashby (1997) Cellular Solids: structures & properties, 2nd ed., Cambridge University Press.
Hebert, C., West, M., and Cannon, B., “Actuation System Design, Fabrication, and Testing for a Morphing Wing
Structure,” AIAA SDM Conference, AIAA-2007-1732, AIAA, Honolulu, HI, 2007.
Joshi, S., Jha, A., Rodrian, J., Alphenaar, R., and Szema, K., “Design of the NextGen Morphing Wing Wind Tunnel
Model,” AIAA SDM Conference, AIAA-2007-1731, AIAA, Honolulu, HI, 2007.
Olympio, K. R., Asheghian, L., Gandhi, F. and Kudva,J., “Design of a Flexible Skin for a Shear Morphing Wing,”
Journal of IntelligentMaterial Systems and Structures, pending (summer 2009).
Strutzenberg, R., Scott, M., Wieseman, C., and Piatak, D., “Wind Tunnel Test and Results of NextGen Morphing Wind
Tunnel Model,” AIAA SDM Conference, AIAA, Honolulu, HI, 2007.
8. RESPONSIBILITY NOTICE
The authors are the only ones responsible for the printed material included in this paper.