At a glance
Toronto Metropolitan University researchers mapped 25 configurations of a modular morphing wing, showing sweep controls low vibration modes and spanwise expansion controls high ones.
The number that decides whether a shape-changing wing is safe isn't its lift coefficient. It's the frequency at which the structure wants to shake.
A research group at Toronto Metropolitan University has now mapped that frequency across 25 different configurations of a morphing wing built in its own lab, and found the vibration behaviour splits along a clean dividing line. Sweep — the motion that rakes the wing backwards or forwards — controls the lowest modes of the structure. Spanwise expansion, the wing stretching or contracting along its length, takes over from the sixth mode upward. The findings appear in Aerospace, in a paper by Faisal Mahmood, Seyed M. Hashemi and Hekmat Alighanbari.
The timing matters. Aviation is under sustained pressure to cut fuel burn and carbon dioxide, and a wing that reshapes itself for each phase of flight is one of the few structural ideas with real promise on that front. The authors cite an estimate from earlier work that even a one percent reduction in drag from morphing would save about USD 140 million a year across the US fleet of wide-body transports. NASA and Boeing have been running wind-tunnel campaigns at Langley Research Center on a high-aspect-ratio wing model with ten independently movable trailing-edge panels, work aimed squarely at taming the flutter that comes with flexible wings. Before any of that hardware can be certified, someone has to know how the structure behaves when it is no longer the shape it was analysed at.
A wing built like a robot, not a plank
The wing under study is not a smooth composite shell. It is a truss-like frame descended from parallel-robot manipulator design, an approach pioneered for morphing wings by Finistauri and Moosavian. Two modules make up the test article, 26 beams in total. Each module carries eight load-bearing members: four active beams set diagonally and driven by actuators, four passive. They meet each other and the two wing ribs through spherical eyebolt joints.
Those joints do something specific. They pass axial force into the ribs but cannot transmit bending moment, so each beam is left to flex on its own. In the un-morphed reference configuration the beams are low-carbon steel, 25.4 millimetres in diameter, with the wing ribs spaced 254 millimetres apart. It is a bench-scale structure, not a flight article, and the paper is explicit about that.
Think of the assembly as a set of guitar strings welded into an adjustable frame. Re-angle the frame and every string's pitch shifts — but not by the same amount, and not in the same direction.
Twenty-five runs, one frequency sweep at a time
Each beam was modelled as a Timoshenko beam, a formulation that keeps shear deformation and rotary inertia in the governing equations. That choice matters here: the members are short and stiff enough that the simpler Euler-Bernoulli assumption would misstate how they bend. A convergence study settled on 30 elements per load-bearing beam with six degrees of freedom per node, and four elements for the beams standing in for each wing rib. The finite element results had been checked against a dynamic stiffness method in the group's earlier publication, so the paper does not repeat that validation.
The un-morphed wing's first ten natural frequencies run from 243.33 hertz to 455.56 hertz. The analysis is free vibration — the structure ringing on its own, with no aerodynamic forces pushing on it.
Two morphing parameters were then varied, five levels each: sweep at minus 30, minus 15, zero, 15 and 30 degrees, and spanwise expansion at zero, 5, 10, 15 and 20 percent. That grid is 25 combinations, and the Taguchi L25 orthogonal array covers exactly those 25. With two factors at five levels the array is the full factorial — the statistical method earns its place not by cutting runs but by what it does with the results. Each frequency was converted to a signal-to-noise ratio under the larger-is-better rule, which ranks how strongly each parameter shifts each mode. The full study is published in Aerospace.
| Factor | Level 1 | Level 2 | Level 3 | Level 4 | Level 5 |
| Sweep (degrees) | -30 | -15 | 0 | 15 | 30 |
| Spanwise expansion (percent) | 0.0 | 5 | 10 | 15 | 20 |
What the 25 configurations showed
The spread is large. At a sweep of minus 30 degrees with no expansion, the first natural frequency sat at 150.78 hertz. Stretch that same swept configuration out by 20 percent and it fell to 116.33 hertz — roughly 23 percent lower, computed from the two runs in the paper's results table. Both figures sit well below the 243.33 hertz of the un-morphed wing.
Across the morphed cases a pattern held: the first five natural frequencies dropped relative to the benchmark, while modes six through ten climbed above it. The authors attribute the rise in the upper modes to the fact that a morphed wing is structurally a different object, with mass and stiffness redistributed along a changed geometry, and it therefore carries different mode shapes above the fifth.
Mode shapes did shift in specific, traceable ways. In the un-morphed wing the first mode is dominated by flexing of the upper diagonal beam in module one. Once the wing is morphed, that role passes visibly to the lower diagonal beam of module two. During spanwise expansion, the upper and lower diagonal members flex distinctly for modes five and below; during sweep morphing, it is the upstream and downstream diagonals that dominate for modes six and higher.
Three structural findings stand out for anyone who models wings for a living. The first: despite the truss-type layout, the members genuinely bend, which makes a pin-jointed truss analysis non-conservative — it would understate the real deformation. The second: the bending is lopsided. Vertical bending, the flapping motion, runs more than 90 percent higher in amplitude than in-plane lead-lag bending. The third: across the first ten modes there is no obvious bending-twist coupling, the interaction that classically drives wings toward flutter.
From that, the authors draw a design-relevant conclusion. Sweep contributes less to the flapping motion, so a swept configuration offers more stiffness against aeroelastic instability. They also warn that their own ten-mode window is too narrow — more than ten modes must be carried into any aeroelastic calculation to be reliable — and that collapsing this wing into a single equivalent beam or plate loses the behaviour entirely.
Where the analysis stops
The paper is a structural dynamics study, and its authors are clear about the boundary. No aerodynamic loads are applied. Free vibration tells you the frequencies and mode shapes; it does not tell you the flutter speed, which depends on how unsteady airloads feed energy into those modes. The next step in the literature is a full aeroelastic treatment.
The test article itself sets limits. Two modules, 26 steel beams, bolted joints and a 254-millimetre rib spacing describe a laboratory demonstrator, not a wing that will fly. Only sweep and spanwise expansion were varied; the architecture can also achieve twist and dihedral morphing, and those remain untested in this framework. During expansion the mass of each beam is held constant because the actuation concept is telescopic, so stiffness and effective diameter are updated while mass is not — a reasonable assumption for that mechanism, and a specific one.
A cautious reader would raise one more point. The wing studied here has no flexible skin, and the paper itself notes that this is an unsolved problem across the field: aluminium and polymer skins are too stiff to deform easily during morphing, and candidate elastomeric and sandwich solutions have so far been demonstrated on camber changes rather than full in-plane and out-of-plane morphing. A skin adds stiffness, mass and damping that this model does not contain.
Why a vibration number decides whether a wing flies
Because flutter is a resonance problem. When airflow over a wing excites the structure at its own natural frequency, the oscillation can grow instead of dying out, and the failure that follows is catastrophic. Shift a natural frequency — by sweeping the wing, by extending it — and you shift the airspeed at which flutter begins. The authors state the risk plainly: the change could delay flutter beneficially, or it could bring flutter onset below the designed flutter speed.
That cuts two ways for aircraft health monitoring. If a wing's vibration signature moves in a way the flight control system did not command, something has failed. The same group has already worked that angle: a companion paper in Aerospace treats individual beam failures as hazards, one and two at a time, and tracks how each combination alters the frequencies of the un-morphed and morphed wing. Vibration data becomes a diagnostic channel, not just a design input.
There are also simpler consequences. Flexural motion of individual members shapes the wing's deformed geometry in flight, which feeds back into pressure drag — the very efficiency the morphing was meant to buy. And members that flex undesirably are members that can fail or transmit vibration into the cabin.
Photo by Pathum Danthanarayana on Unsplash
What the next decade of morphing wing work needs
The immediate to-do list is written into the paper's own conclusion. Push past ten modes. Build the aeroelastic model on top of these frequencies. Extend the design of experiments to the other morphing parameters, then to combinations of three or four at once, where the 25-run grid no longer suffices and the Taguchi method starts earning its keep as a way of cutting experiments rather than just organising them.
Beyond that, the field needs hardware. A modular wing whose joints can be instrumented, whose beams can be deliberately damaged, and whose modal response can be measured on a shaker table is the obvious bridge between a finite element model and a flight clearance. The Toronto Metropolitan group's structure is built for exactly that kind of testing, which may be its most useful property. Certification of any morphing transport will eventually rest on an argument that the structure's dynamics are understood at every shape it can take, not only at the one drawn on the type certificate.
Funding for the work came from a Natural Sciences and Engineering Research Council of Canada Discovery Grant, with support acknowledged from the Ontario Graduate Scholarship program, the Queen Elizabeth II Graduate Scholarship in Science and Technology and Toronto Metropolitan University. The authors declare no conflicts of interest.
- aerospace-engineering
- toronto-metropolitan-university
- finite-element-analysis
- structural-dynamics
- morphing-wing

