Professional Documents
Culture Documents
Aeroelastic Modeling, Analysis and Testing of A Morphing Wing Structure PDF
Aeroelastic Modeling, Analysis and Testing of A Morphing Wing Structure PDF
David J. Piatak 3
NASA Langley Research Center, Hampton, Virginia, 23681
NextGen Aeronautics, Inc., of Torrance, CA, has been involved in all aspects of the
design, development and testing of morphing aircraft structures, much of this in conjunction
with NASA Langley Research Center. The single most substantial effort in this area has
been the DARPA Next-Generation Morphing Aircraft Structures project, in which NextGen
has been one of two prime contractors tasked with advancing the state of technology of
morphing aircraft. NextGen developed an in-plane morphing geometry concept with
mechanized four-bar linkages, and motion imparted by computer-controlled linear
actuators. Flexible elastomeric skins with out-of-plane stiffeners accommodated the wing
motion while transmitting air pressure loads to the wing substructure. The development
process involved wind tunnel testing of a full-scale wing for a 2400-lb vehicle and flight
testing of a subscale unmanned air vehicle. Among the more critical challenges has been that
of addressing wing design for aeroelasticity, given the unique features of morphing wings.
Specific relevant issues include: the need to address multiple geometries and flight envelopes
to account for morphing shape changes; in-plane flexibility of the wing resulting from its
mechanisms and soft (relative to typical fixed wing structure) restraint by linear actuators;
nonlinearity and equivalent loss of stiffness in the wing’s out-of-plane direction as a result of
multiple revolute joints in the structure. These unfamiliar features required, in some cases,
the development of new analytical or testing approaches. In addition, due to a somewhat
untried approach to mounting the half-span wind tunnel model of its size at the NASA
Transonic Dynamics Tunnel, particular considerations of support structure dynamics had to
be given particular attention prior to wind-on testing there. Ultimately, all of these
challenges were overcome, and successful wind tunnel and flight demonstrations were
completed.
1
Senior Engineer, Design and Analysis Group, 2780 Skypark Drive, Ste. 490, Torrance, CA, AIAA Member
2
Principal Engineer, Design and Analysis Group, 2780 Skypark Drive, Ste. 400, Torrance, CA, AIAA Member
3
Senior Engineer, Design and Analysis Group, 2780 Skypark Drive, Ste. 490, Torrance, CA, AIAA Member
1
American Institute of Aeronautics and Astronautics
Copyright © 2007 by 07 by NextGen Aeronautics, Inc. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
I. Introduction
D URING the past four years, NextGen Aeronautics, Inc. has been one of the prime contractors in the DARPA-
sponsored Morphing Aircraft Structures (MAS) program. This project seeks to enhance and extend the mission
capability of aircraft through the application of reconfigurable “morphing” structures that enable an aircraft to alter
its configuration in flight to most effectively operate in a given flight regime. NextGen’s morphing concept utilized
two morphing degrees of freedom, a variable wing sweep and variable shear angle of an internal 4-bar linkage
assembly, to accommodate wing geometry changes that met the DARPA MAS programs requirements while
addressing NextGen’s targeted mission profiles. Specifically, the variable geometry allowed the wing to achieve and
maintain, in flight, five distinct configurations – one for each of five identified mission segments: high lift, climb,
loiter, cruise/dash, and high-speed maneuver.
The utilization of a reconfigurable wing structure inherently led to unique issues in the aerodynamic and
structural analysis tasks leading to the November 2005 wind tunnel test at the NASA Langley Transonic Dynamics
Tunnel (TDT). In particular, the analytic flutter clearance procedures were complicated by the large number of
configurations to be analyzed. Additionally, the structural design implementation used to achieve the morphing
capabilities led to unique aeroelastic interactions which needed to be considered during the flutter clearance effort.
The combination of a large number of physical configurations and a flexible structure with unique dynamic
characteristics resulted in a complex and unique flutter clearance investigation.
The unsteady aeroelastic analyses in support of the TDT test evolved through two major efforts. The first was the
analytic prediction of the flutter boundaries and margins for each of the configuration, the results of which were
presented at the September 2005 model system review at NASA Langley. The second major effort took place during
the test activities when updated ground vibration test data of the TDT/model interface structure became available. At
this point, real time flutter clearance runs were made, with an update flutter model, on a daily basis.
This paper describes the details of these efforts, including a summary of NextGen’s wing structural design,
identification of the unique aeroelastic phenomenon, analytic aeroelastic model generation, analysis methods, and
the flutter analysis and clearance efforts that resulted in the first successful wind tunnel test and the first ever in-
flight demonstration of a morphing aircraft wing.
2
American Institute of Aeronautics and Astronautics
Fairing Slotted 1 Wing Main Pivot
Rib
Leading Edge Closeout
Fairing Main
Weldment
Spars
Spanwise-running
4-Bar/Slider
Assembly Actuators skin strip
3
Ribs
Underlying support
“ribbons”
Sliders
2 Revolute
Joints
4-Bar Assembly
Figure 1. Morphing Wing Structure Figure 2. Elastomeric Skin and Support Ribbons
The wing morphing degrees of freedom required the use of a unique internal wing structure. Wing sweep was
enabled through a large revolute joint (item 1 in the figure), located at the root of the forward-most wing spar. Wing
area change was enabled by an assembly of 4-bar linkages (item 2) with smaller revolute joints, joining ribs to spars,
allowing shear motion. To accommodate the wing root motion resulting from sweep and 4-bar shear angle changes,
a combination of an additional 4-bar linkage assembly and slider mechanism was employed (item 3).
The internal structure was not the only distinctive aspect of the wing. The external skin, having to accommodate
an exceedingly large shear strain, was made of an elastomeric silicon material. To withstand the airloads, the skin
was supported with an underlying ribbon structure, as shown in Figure 2. This provided out-of-plane stiffness while
allowing the required shearing motion.
3
American Institute of Aeronautics and Astronautics
Figure 3. MAS Finite Element Models
4
American Institute of Aeronautics and Astronautics
Spars
Ribs
Actuators
Leading edge closeout
Leading Edge
Configuration 1
Secondary Rib
Bulkheads
Pylon to wing
main pivot
Pylon to wing
main pivot
Slot to accommodate
4-bar linkage and
wing spar roots
Configuration 2
Main Rib
Secondary Rib
As the TDT test approached and GVT measurements were obtained, it became apparent that a more realistic
representation of the modal behavior of the model would be obtained by creating a fully built-up model, containing
the main and internal weldments, bulkheads, and significant leading edge solid structure of the fairing substructure,
5
American Institute of Aeronautics and Astronautics
as shown in case 2. Note a highly detailed model of the fairing skin was also developed for the built-up model; it is
not however, shown for clarity. The skin model may be seen in Figure 3.
At this point it is necessary to explain the treatment of the fairing stiffness for each of these fairing models.
Configuration 1 did not employ composite fairing skin models. Rather, all fuselage structural stiffness was assumed
to be included in the main weldment, its attachment to the wing main pivot, and the contribution of the 4-bar linkage
since loads were transferred to the main weldment through the 4-bar links. Fuselage masses and moments of inertia
were lumped and added at the appropriate location to give the proper center of gravity (CG) location and system
mass and moments of inertia. This lumped mass was then rigidly connected to the weldment through multi-point
constraints. To bracket the stiffness contribution of the fairing skin, the length of rigid attachment between the CG
mass and the weldment was varied. In one case, the attachment ran from the leading edge of the fuselage to
approximately 60% of the fuselage length. A stiffer connection, attaching the CG lumped mass to the entire length
of the weldment was also employed. These two versions, referred to as case 1 and case 2, of configuration 1 led to
the results shown in Figure 9-13.
The stiffness bracketing employed for configuration 1 was not necessary for configuration 2 since the fairing
skin was accurately represented by a laminate composite built up structure. As with case 1, additional mass and
moments of inertia were added such that the total fuselage and fairing masses and moments of inertia were the
values and at the location as predicted by SolidWorksTM. This high-fidelity configuration was used in the day-to-day
determination of flutter margins during the actual testing in the TDT.
4-Bar Actuator
6
American Institute of Aeronautics and Astronautics
It was assumed that the structural stiffness of the NMAS model, to include the wing, fuselage, and 4-bar
substructures, was entirely contained in relatively rigid structure – spars, ribs, stringers, composite structures,
actuators, etc. Flexible skin contributions to the wing stiffness were not included. This was considered to be
conservative as any additional stiffness from the skin would tend to raise modal frequencies contributing to flutter
mechanisms.
The mass distribution of the NMAS model was based upon mass values obtained from the components as
modeled in SolidWorksTM. For each major structural component, the mass was adjusted such that it matched the
SolidWorksTM value. Components included in this approach were the spars, ribs, leading and trailing edge closeouts,
fairing weldment, ribs, and bulkheads, and 4-bar linkage structures. Running masses such as the flexible skin and
associated fastening hardware, stringers, hydraulic lines, and instrumentation lines were lumped as appropriate by
area or length and added at each of the spar/rib joints in the wing. Concentrated masses not structurally modeled
such as lugs at revolute joints and 4-bar guide blocks, were also lumped according to their distance from a spar/rib
joint and thereby accounted for. Similarly, all mass in the fairing structure not accounted for in the structural FEM
was positioned to give the correct center of gravity location and value for the fuselage structure. Additionally, in the
case of the fuselage, moments of inertia were added as well at the center of gravity to account for those inertias not
preserved by the finite element model.
Loiter
Climb
HSM
High-Lift
Cruise/Dash
C. Aerodynamic/Structure Connections
Standard NASTRAN aeroelastic spline techniques were used to connect the aerodynamic model with the
structural model. In this case, the finite plate spline (FPS) was used the three inboard lifting surface macroelements
rather than the infinite plate spline (IPS). This is because the structural nodes used in the spline did not extend to the
7
American Institute of Aeronautics and Astronautics
edge of the aerodynamic model (only nodes at the intersections of spars and ribs were used, thereby omitting nodes
on the leading and trailing edge composite structures). In these situations it is generally recognized that the FPS
outperforms the IPS. The outboard most wing macroelement was connected to the wing substructure with a beam
spline since the only available structural nodes lay along a singe axis. In the case of the fairing aerodynamic model,
the aerodynamic panels were also connected to the main weldment with a beam spline. In the case of the surface
splines, spanwise smoothness of the separate splines on the wing was improved by using the convention of selecting
structural nodes slightly outside the aerodynamic macroelement used in the spline.
It should be noted that the structural nodes used in the spline generally changed among configurations since the
spanwise and chordwise motion of the panels did not match the shearing motion of the underlying structural nodes.
Thus for each configuration, the splines were checked and manually updated as required to maintain the correct
relationship between the aerodynamic macroelements and the underlying structure.
8
American Institute of Aeronautics and Astronautics
The point at which the curve formed by the collection of these points intersected the desired Mach number was
noted, and from this the matched-point flutter velocity and dynamic pressure were directly obtained.
Cruise/Dash
High-Lift
σ = 1.0
HSM 700
Analysis Case
Key 600
Case 1
1000 Case 2
Dynamic Pressure (psf)
500 σ = 0.5
400 σ = 0.3
σ = 0.2
300
100
σ = 0.1
200
100
10 0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2
Mach Mach
9
American Institute of Aeronautics and Astronautics
Test Point Key
Flutter Boundary and Test Envelopes Climb Flutter Boundary and Test Envelops
on TDT Operating Envelopes Loiter Hi-Lift Configuration
Hi-Lift Configuration Cruise/Dash
High-Lift σ = 8.0 σ = 4.0 σ = 2.0 σ = 1.0
1000 600
HSM
Analysis Case
Key
Case 1 500 σ = 0.5
Case 2
400
σ = 0.3
σ = 0.2
100 300
σ = 0.1
200
σ = 0.05
100 σ = 0.02
10 0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2
Mach Mach
σ = 0.2
100 300
σ = 0.1
200
σ = 0.05
100
10 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4
Mach
Mach
10
American Institute of Aeronautics and Astronautics
Flutter Boundary and Test Envelopes Flutter and Test Envelops
on TDT Operating Envelopes Test Point Key Climb Configuration
Climb Configuration σ = 8.0 σ = 4.0 σ = 2.0
10000 Climb 800
Loiter
Cruise/Dash
High-Lift 700 σ = 1.0
HSM
500 σ = 0.5
400 σ = 0.3
σ = 0.2
300
100
σ = 0.1
200
σ = 0.05
100
10 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4
Mach Mach
High-Lift
HSM 900
Analysis Case
Key 800
Case 1
Case 2 700 σ = 1.0
1000 σ = 3.25
Equivalent Airspeed (keas)
Dynamic Pressure (psf)
600
500 σ = 0.5
400 σ = 0.3
100
σ = 0.2
300
σ = 0.1
200
100
10
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2
Mach
Mach
11
American Institute of Aeronautics and Astronautics
This procedure began with obtaining frequency response information from a suite of accelerometers, shown
Figure 14, for each of the five configurations as mounted in the TDT. The frequency response function (FRF) for
each configuration was obtained from a NASA Langley data acquisition system. As an example, the FRF for the
high lift configuration is shown in Figure 15.
From the obtained FRFs, it was possible to identify the first five fundamental modal frequencies to within
approximately 0.25 Hz. Table 1 shows the frequency results for the high-lift configuration. Each morphed
configuration model was then adjusted, as discussed previously, to match these frequencies. Table 2 shows the GVT
frequencies with the frequencies obtained after model tuning and the corresponding percent differences. A further
qualitative check of the modal response of the structure was obtained by comparing the modeshapes obtained from
the GVT with those obtained analytically. Figures 16 and 17 show such a comparison for the high-lift configuration
for the first wing bending and wing first torsion modes.
12
American Institute of Aeronautics and Astronautics
Table 1: GVT Frequencies for High Lift Configuration
Test Analytic
Figure 16: Mode Shape Correlation – Wing First Bending – High Lift
Figure 17: Mode Shape Correlation – Wing First Torsion – High Lift
13
American Institute of Aeronautics and Astronautics
VII. Opportunities for Further Research and Development
NextGen’s experience during the design, analysis, and wind tunnel testing of the MAS model indicated several
areas in which advances in modeling and analysis techniques would enable more efficient and effective prediction of
steady and unsteady aeroelastic phenomenon. The opportunities for research and development work, both in terms
of general modeling techniques and flutter specific analyses are presented below.
A. Aeroelastic Modeling
A great deal of time was spent developing the aeroelastic model for each of five distinct configurations. Each of
the structural, aerodynamic, and aero/structure connection models was unique for each configuration, resulting in a
very time consuming process to generate all of the required models. In some cases, a certain amount of automation
was possible. For instance, it was possible to automate a fair portion of the structural FEM development. Knowing
the geometry of each configuration, it was possible to use the PATRAN Command LanguageTM (PCL) to correctly
position each wing structural node based up the known wing sweep angle and internal angle of the wing 4-bar
mechanism. Other modeling details, particularly the connections between the wing spar roots and outer channel of
the main 4-bar mechanism had to be changed by hand for each configuration.
Similar issues arose concerning the aerodynamic model development. The placement of each macroelement had
to be changed for each configuration. Fortunately, a check of the panel aspect ratios revealed that the same panel
model within each macroelement could be used for each configuration. This was because the morphing mechanism
was through shear only, resulting only a moderate amount of aspect ratio change. However, in the more general case
of morphing where shear and telescoping mechanisms are employed aspect could change considerably, thereby
requiring unique paneling for each configuration.
Finally, the spline model providing the aero/structural connection had to be inspected for each configuration.
Generally, the shear motion of structural model did not correspond with the spanwise change of each aerodynamic
macroelement. This resulted in the fact that a structural node that lay under a particular macroelement in one
configuration lay under a different panel in another. This required the splines to be reworked for each configuration.
These challenges could be addressed through the implementation of a kinematics solver that, knowing the
appropriate morphing degrees of freedom, could appropriately locate structural nodes and aerodynamic
macroelement corner locations. Beyond this, finite element and aerodynamic mesh generators would have to be
employed to fully automate the aeroelastic model development process.
B. Flutter Analyses
As mentioned previously, there are unique aeroelastic mechanisms presented by the novel nature of the MAS
wing structure. Of particular interest was the in plane motion as modal frequencies much lower than are traditionally
encountered. Even when in plane modes are encountered, they are typically assumed to contribute no out of plane
motion. Since the classical flutter techniques within NASTRAN only consider pitch and plunge motion, in plane
dynamics do not affect the flutter solution. However, the potential for in-plane dynamics to affect the flutter
behavior does exist, particularly when the in-plane modal frequencies are of the same magnitude as the out-of-plane
modal frequencies, as was the case with the MAS model. In addition to any out-of-plane motion that may be induced
through structural asymmetries, the fore-aft motion itself introduces a changing local velocity that would change the
local lift coefficient periodically with at the fore-aft modal frequencies. This would result in an out-of-plane
component induced by purely in-plane motion – a phenomenon not captured within the current frequency domain
NASTRANTM flutter solutions. Additionally, both steady and unsteady aeroelastic effects stemming from the motion
in the morphing degrees of freedom were not considered, and in fact were treated as quasi-static. For rapid motion in
the morphing degrees of freedom, however, the potential exists to generate transient dynamics and aerodynamics
that may be important in an aeroelastic sense. As morphing aircraft structures are pursued in the future, the
analytical means of handling a broader class of motion extending beyond the usual pitch and plunge motions is
desirable.
VIII. Conclusion
An evolutionary process was utilized to analyze and characterize the dynamic aeroelastic characteristics of the
NMAS TDT model. Increasingly accurate structural models were used to further confidence that the model would
encounter no aeroelastic instabilities within the test envelop of interest for each configuration. A purely analytic
process was used to present positive flutter margin results at the Model Systems Review. Beyond the MSR, GVT
and other structural stiffness data from both the NMAS wing model and the TDT interface structure became
14
American Institute of Aeronautics and Astronautics
available. Upon examination, it became clear that the stiffness as originally modeled in the TDT structure was
significantly higher than in reality.
Because of this situation, it became necessary to perform additional flutter margin checks as the actual TDT test
was being conducted. By this time a high-fidelity fuselage structural model had been added to the wing model.
Knowing the fundamental wind-off modal frequencies from GVT data, it was possible to tune each configuration
model to match, with reasonable accuracy, the test frequencies. Qualitative examination of the mode shapes
generated further confidence that the aeroelastic models were sufficient for flutter margin prediction. Positive flutter
margins over all the desired test points were predicted for all configurations, allowing an essentially unrestricted test
matrix throughout the NMAS TDT test.
NextGen’s experience gained during the flutter prediction efforts indicated significant opportunities for
aeroelastic modeling and analysis methods development. Aeroelastic characterization of morphing aircraft structures
would benefit from both advances in automated structural, aerodynamic, and spline modeling. Furthermore, the
ability to handle aeroelastic phenomena unique to morphing structures such as low frequency in-plane motion and
dynamics associated with the morphing degrees of freedom would be advantageous.
Acknowledgement
The research and development effort documented herein was performed under Contract No. F33615-02-C-3257,
“Next-Generation Morphing Aircraft Structures (N-MAS).” This contract was sponsored and funded by DARPA
with Dr. Terrence A. Weisshaar as the program manager. AFRL-Dayton, OH, served as the contracting agency with
Dr. Brian Sanders and Dr. Bryan Cannon as the technical monitors. HyPerComp was the prime contractor for the
program with NextGen Aeronautics as the major subcontractor; NextGen also served as the technical lead
organization.
15
American Institute of Aeronautics and Astronautics