Professional Documents
Culture Documents
This content has been downloaded from IOPscience. Please scroll down to see the full text.
(http://iopscience.iop.org/0964-1726/17/1/015025)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 128.111.121.42
This content was downloaded on 22/07/2015 at 20:20
E-mail: fgandhi@psu.edu
Abstract
This paper identifies the desirable attributes of a flexible skin of a morphing wing. The study is
conducted using airfoil camber morphing as an example. The ideal flex-skin would be highly
anisotropic, having a low in-plane axial stiffness but a high out-of-plane flexural stiffness.
Reduced skin axial stiffness allows morphing at low actuation cost. However, for some
substructure and actuation designs, a lower limit on the skin’s in-plane axial stiffness may be
required to prevent unacceptable global camber deformation under aerodynamic loads. High
flexural stiffness prevents local deformation of skin sections between supports due to
aerodynamic pressure loads, and avoids buckling of skin sections under compression as the
airfoil cambers under actuation force. For the camber morphing application the strain levels in
the flex-skin are not expected to exceed around 2%. If the axial stiffness of the flex-skin is
reduced significantly, it may be necessary to consider aerodynamic stiffness (negligible vis-à-vis
structural stiffness for classical airfoils) to accurately calculate deformation under loading. The
approach followed in the study can be used to identify specifications for the skin and then
reverse engineer and design highly anisotropic composite skins that meet the specifications.
(Some figures in this article are in colour only in the electronic version)
Figure 1. (a) NACA 0012 airfoil with rigid leading-edge D-spar and
(b) compliant section extending aft of the D-spar to the trailing edge.
2
Smart Mater. Struct. 17 (2008) 015025 F Gandhi and P Anusonti-Inthra
20 elements
3
Smart Mater. Struct. 17 (2008) 015025 F Gandhi and P Anusonti-Inthra
o
with Faero representing, nominally, the loads of the baseline 30% chord Trailing-edge
airfoil, then ∂ Faero /∂w represents an aerodynamic stiffness Undeformed(Baseline) Airfoil flap undergoing same
tip deflection
contribution. If equation (3) were introduced in equation (2),
Trailing-edge
∂ Faero
K s w = Faero
o
+ w or deflection
∂w Morphed (Cambered) Airfoil
(4) Equivalent Flap Angle
∂ Faero
Ks − w = Faero
o
. Figure 7. Calculation of the equivalent flap angle for the cambering
∂w
airfoil.
It is evident, then, that the total system stiffness comprises
structural and aerodynamic components. For conventional
airfoils the structural stiffness, K s , is dominant, and the including vertical tip deflection or equivalent flap angle. In
aerodynamic contribution is insignificant. However, it is seen the present study the latter is used, and figure 7 shows how it is
in section 3 that if the structural stiffness of a morphing airfoil calculated. Essentially the vertical tip deflection produced by
is reduced sufficiently, the aerodynamic contribution may no the camber is equated to the vertical tip deflection of a 30%
longer be negligible. chord trailing-edge flap and the corresponding flap angle is
The deformation due to the actuation loads can be then used as a measure of the camber.
calculated without the effects of aerodynamic loads as follows:
3. Results and discussion
K s u = Fact (5)
where Fact is the global load vector due to the actuation loads, Figure 8 shows the airfoil camber deformation under
and u is the displacement field due to only the actuation loads. aerodynamic loading ( M = 0.6, α = 5◦ ), as a function of skin
The solution to equation (5) is obtained by simply inverting the axial and flexural stiffness. Results are presented both with
structural stiffness matrix. In the calculation of the response and without the inclusion of aerodynamic stiffness (figures 8(a)
to actuation forces the influence of the aerodynamic loads is and (b), respectively). In both figures it is seen that as the
accounted for in an approximate manner by considering only axial stiffness decreases, the camber deformations under the
the aerodynamic stiffness term. Thus, aerodynamic loading can become very large. Further, for
moderate to high values of skin axial stiffness, aerodynamic
∂ Faero stiffness does not have an important influence, but for low
K s v = Fact + v or
∂v values of skin axial stiffness, the results differ. For low
(6) skin axial stiffness the camber deformations predicted in the
∂ Faero
Ks − v = Fact absence of aerodynamic stiffness are greater (these areas have
∂v
been highlighted in figures 8(a) and (b) to emphasize the
where v is the displacement field due to the actuation differences). So the variation in aerodynamic loading with
loads with the influence of the aerodynamic stiffness, and deformation (which manifests itself as aerodynamic stiffness)
the aerodynamic stiffness, ∂ Faero /∂v , is obtained using an should be considered in that range. Figures 9(a) and (b) show
approach similar to that given in appendices A and B. Since the camber deformations of the airfoil under aerodynamic
the aerodynamic stiffness ∂ Faero /∂v is itself a function of the loading. For nominal values of skin axial and flexural stiffness
displacement field, v , the solution to equation (6) must be the airfoil shows virtually no camber deformation (figure 9(a)),
obtained using an iterative procedure similar to that presented but when the skin axial stiffness is decreased substantially
in appendix A. (by two orders of magnitude in figure 9(b)) an airfoil camber
The simulation results provided in the following section deflection of 3.2◦ is observed. The aerodynamic pressure
are based on a substructure made of aluminum (Young’s distributions on the upper and lower surfaces of the airfoil
modulus = 70 GPa) and having a cross-sectional thickness cause the section to camber upward. Figure 10 shows the
of 1.5 mm. The ‘nominal’ or ‘reference’ skin is arbitrarily deformations of the airfoil for the nominal value of skin axial
assumed to be an aluminum skin of 0.5 mm thickness. The stiffness but for a reduction in skin flexural stiffness by a
axial and flexural stiffness values of this skin are considered as factor of 5. For this reduced skin flexural stiffness, it is
a baseline. In this paper, the skin axial and flexural stiffness observed that sections of the skin between support points
are then independently varied from the baseline values, over undergo local flexural deformation due to the aerodynamic
large ranges, and are essentially treated as design variables. surface pressure (compare with figures 9(a) and (b) where no
For the results presented in the following section, the skin axial such local deformations were observed for the nominal value
stiffness is varied from 10−3 to 10+3 of the nominal value, and of skin flexural stiffness). From the results in figures 8–10
the flexural stiffness is varied from 10−1 to 10+5 of the nominal it is evident that lower limits must be imposed on both the
value. While the axial and flexural stiffness of an isotropic axial and the flexural stiffness of the skin—the first to prevent
skin do not vary independently, an anisotropic composite skin excessive airfoil global camber deformation, and the second
could, in principle, be designed to yield a desired combination to prevent local bending or ‘bubbling’ of the skin sections
of axial and flexural stiffness. between support points.
The airfoil camber produced due to actuation or It is to be noted that as long as the skin is intermittently
aerodynamic loading can be quantified in different ways (and not continuously) supported, the flexural stiffness
4
Smart Mater. Struct. 17 (2008) 015025 F Gandhi and P Anusonti-Inthra
Skin Flexural stiffness (relative to nominal)
deg
0.066 Deg.
Piezostack
constraint will have to be considered. However, if the actuator
contracting
substructure were to be designed differently such that it
provided a greater chordwise bending stiffness, the constraint
on the skin axial stiffness may no longer be as critical. Piezostack
actuator
This may appear to be an unattractive proposition as a stiff elongating
substructure would suggest greater actuation force requirement
to camber the airfoil. However, this does not have to be the
case if the actuators are cleverly integrated into the substructure
design and themselves contribute to the chordwise bending
stiffness under external aerodynamic loads. On activation, Figure 11. Substructure and actuation concept of a variable camber
however, they would not provide a high chordwise bending rotor airfoil (representation of concept in [12]).
stiffness and allow easy camber morphing at low actuation
effort. As an example, figure 11 shows the substructure
of a controllable camber airfoil design proposed in [12]. elongate, the junction point between the two translates upward,
Several compliant mechanism units are arranged sequentially causing the passive links to deform as shown. The top
along the chord aft of the spar. The vertical members are skins stretch, the bottom skins contract, and the deformation
active members—antagonistically operated piezoelectric stack of each unit additively produces camber deformation of the
actuators. The inclined members are passive links. As the airfoil. The particularly interesting feature about this design
top stack actuators contract and the bottom ones synchronously is that in the absence of activation, the airfoil has a high
5
Smart Mater. Struct. 17 (2008) 015025 F Gandhi and P Anusonti-Inthra
-26.915 Deg.
6
Smart Mater. Struct. 17 (2008) 015025 F Gandhi and P Anusonti-Inthra
Stain
level
x/c
Figure 15. Strains in the upper and lower skins for different
actuation loading levels.
4. Conclusions
7
Smart Mater. Struct. 17 (2008) 015025 F Gandhi and P Anusonti-Inthra
This section outlines solution procedure for solving a nonlinear w = φζ. (B.3)
equation of the following form:
Then the perturbation matrix can be written as
K s w = Faero (w) or K s w − Faero (w) = 0. (A.1)
∂ Faero ∂ Faero ∂ζ
= . (B.4)
Equation (A.1) is nonlinear and can be solved using an iterative ∂w ∂ζ ∂w
method. First, let us assume that the displacement field,
The vector ∂ Faero /∂w can be obtained using the
w, is
perturbation method, and the row vector ∂ζ /∂w can be
wi+1 = wi + wi (A.2)
calculated from equation (B.3) as follows:
where wi is the initial guess values for the displacement
field and wi is the changes required for wi+1 to satisfy the ∂ζ
= (φ T φ)−1 φ T . (B.5)
nonlinear equation. Substituting equation (A.2) into (A.1) ∂w
yields Thus the perturbation matrix for any displacement field, w,
can be approximated by using perturbation of the trailing edge
K s wi + K s wi − Faero (wi + wi ) = 0 or
displacement, ζ , and a known mode shape, φ , as follows:
∂ Faero (A.3)
K s wi + K s wi − Faero (wi ) − wi = 0 ∂ Faero ∂ Faero T −1 T
∂w wi = (φ φ) φ . (B.6)
∂w ∂ζ
where ∂ Faero /∂w is the perturbation of the aerodynamic force
due to a change in displacement field, and calculation of this
matrix is given in appendix B. Equation (A.3) can be used to References
solve for wi as follows: [1] Hall J 1989 Executive summary AFTI/F-111 mission adaptive
−1 wing Tech. Rep. WRDC-TR-89-2083 Wright Research
∂ Faero Development Center, Wright–Patterson Air Force Base
wi = K s − {Faero (wi ) − K s wi }. (A.4) [2] Pendleton E W, Bessette D, Field P B, Miller G D and
∂w wi Griffin K E 2000 Active aeroelastic wing flight research
program: technical program and model analytical
Then the displacement field is updated using equation (A.2). development J. Aircr. 37 554–61
[3] Kudva J 2004 Overview of the DARPA smart wing project
For the next iteration the aerodynamic force vector, Faero , and
J. Intell. Mater. Syst. Struct. 15 261–7
the perturbation matrix, ∂ Faero /∂w, have to be reevaluated at [4] Bartley-Cho J D, Wang D P, Martin C A, Kudva J N and
the updated displacement field. The whole process repeats West M N 2004 Development of high-rate, adaptive trailing
until wi at the trailing edge is less than a small number close edge control surface for the smart wing phase 2 wind tunnel
model J. Intell. Mater. Syst. Struct. 15 279–91
to zero. [5] Bowman J, Sanders B, Cannon B, Kudva J, Joshi S and
Weisshaar T 2007 Development of next generation aircraft
Appendix B structures Proc. 48th AIAA/ASME/ASCE/AHS/ASC
Structures, Structural Dynamics and Materials Conf.
(Honolulu, HI, April 2007)
The perturbation of the aerodynamic force due to a change in [6] Andersen G R, Cowan D L and Piatak D J 2007 Aeroelastic
displacement field, ∂ Faero /∂w, is calculated in this section. modeling, analysis and testing of a morphing wing structure
The displacement field, w, that is used to calculate the Proc. 48th AIAA/ASME/ASCE/AHS/ASC Structures,
Structural Dynamics and Materials Conf. (Honolulu, HI,
perturbation matrix is defined as displacement due to actuation April 2007)
force, u , without any aerodynamic loads. [7] Love M H, Zink P S, Stroud R L, Bye D R, Rizk S and
White D 2007 Demonstration of morphing technology
K s u = Fact (B.1) through ground and wind tunnel tests Proc. 48th AIAA/
ASME/ASCE/AHS/ASC Structures, Structural Dynamics and
where K s is the structural stiffness matrix and Fact is the Materials Conf. (Honolulu, HI, April 2007)
[8] Bein Th, Hanselka H and Breitbach E 2000 An adaptive spoiler
global load vector due to the actuation loads. Let ζ be the to control the transonic shock Smart Mater. Struct. 9 141–8
vertical deflection of the trailing edge; the mode shape of this [9] Campanile L F and Sachau D 2000 The belt-rib concept: a
displacement field, u , can be represented by structronic approach to variable camber J. Intell. Mater. Syst.
Struct. 11 215–24
u [10] Campanile L F, Keimer R and Breitbach E J 2004 The
φ= . (B.2) ‘fish-mouth’ actuator: design issues and test results J. Intell.
ζ Mater. Syst. Struct. 15 711–9
[11] Katz J and Plotkin A 1991 Low-Speed Aerodynamics: From
This mode shape, φ , will be used to calculate the perturbation Wing Theory to Panel Methods (New York: McGraw-Hill)
matrix by assuming that the airfoil displacement field, w, [12] Gandhi F, Frecker M and Nissly A 2008 Design optimization
can be approximated by the mode shape, φ , and the vertical of a controllable camber rotor airfoil AIAA J. 46 142–53