Aeroelastic Test, Ground Vibration Test & Correlation

Why Test Matters for Aeroelasticity

An aeroelastic prediction is built on a structural dynamic model — its mass, stiffness, natural frequencies, mode shapes and damping. If that model is wrong, the aeroelastic prediction built on it is wrong, no matter how sophisticated the aerodynamic model or the coupling method. Ground vibration testing (GVT) is the primary physical test that establishes whether the structural dynamic model represents the real structure. It measures the natural frequencies, mode shapes and damping of the structure in a known configuration, under known boundary conditions, with known mass distribution. The correlation of the analytical model with GVT data is the gate that the structural dynamic model must pass before it is used as the basis for aeroelastic prediction. Skipping or shortcutting this gate exports uncertainty silently into every downstream aeroelastic result.

IF THE STRUCTURAL MODES ARE WRONG, THE AEROELASTIC PREDICTION BUILT ON THOSE MODES IS ALSO SUSPECT.

Ground Vibration Testing

A ground vibration test excites the structure with shakers or impact hammers and measures the response at an array of accelerometers distributed over the surface. From the measured frequency response functions, the modal parameters are extracted: natural frequencies, mode shapes, damping ratios, and (where the test is designed for it) modal mass. The test is conducted in a defined mass configuration — typically close to the flight configuration, with representative masses for fuel, stores, and equipment — and under defined boundary conditions, often free-free (suspended on soft supports) to approximate the unconstrained flight condition. The quality of the GVT depends on the shaker placement (must excite all modes of interest), the accelerometer density (must resolve the mode shapes spatially), the frequency resolution (must separate closely spaced modes), and the care taken with boundary conditions and mass configuration.

Left: aircraft on soft supports in a GVT rig, shakers attached at selected locations, accelerometer array over wing and fuselage surfaces. Right: corresponding FE mode shapes (first bending, first torsion, second bending) aligned beside measured mode shapes — visual and MAC correlation. Caption: "GVT establishes the measured modal basis; FE modes are correlated against it before aeroelastic use."

GVT Measurement Objectives

A GVT is designed to measure a set of structural dynamic quantities, each of which validates a different aspect of the analytical model. The table below lists the principal measurement objectives and what each validates.

The Aeroelastic Correlation Hierarchy

Correlation of an aeroelastic model with test is not a single step; it is a hierarchy. Each level is a prerequisite for the credibility of the next. Attempting to fix an aeroelastic mismatch at the top of the hierarchy without first verifying the levels beneath it is a common and frustrating error — the real problem is usually lower down. The hierarchy runs from the most fundamental (mass) through the structural dynamic properties to the coupled aeroelastic response.

1. LEVEL 1 — MASS: Does the model reproduce total mass and CG? If the total mass or centre of gravity is wrong, every modal and aeroelastic result that depends on inertia is wrong.
2. LEVEL 2 — STATIC STIFFNESS: Does measured load–deflection behaviour correlate? If the static stiffness is wrong, the modal frequencies and the static aeroelastic deformation are wrong.
3. LEVEL 3 — MODAL FREQUENCIES: Are frequencies credible? Frequencies depend on mass and stiffness together; matching them is necessary but not sufficient.
4. LEVEL 4 — MODE SHAPES: Do measured and analytical shapes correlate? A model can match frequencies with wrong shapes; shapes validate the spatial distribution of mass and stiffness.
5. LEVEL 5 — DAMPING: Is damping reasonably represented? Damping is the least certain input; it must be understood, even if it is often assumed rather than predicted.
6. LEVEL 6 — AERODYNAMIC LOAD: Does aerodynamic prediction correlate with relevant test/data? Wind-tunnel or CFD-validated aerodynamic loads must be credible before they are coupled to the structure.
7. LEVEL 7 — COUPLED RESPONSE: Does the aeroelastic response match experiment or flight evidence where available? This is the final test of the coupled model, and it depends on all preceding levels.

Do Not Fix the Top Before Verifying the Bottom

When an aeroelastic prediction does not match test or flight evidence, the instinct is often to adjust the aerodynamic model or the coupling — the parts of the hierarchy closest to the aeroelastic result. This is frequently the wrong place to look first. The most common root cause of aeroelastic mismatch is an error in the structural model beneath it: a mass that is wrong, a joint that is too stiff or too soft, a mode that is missing or incorrectly shaped. Before touching the aerodynamics or the coupling, verify the mass, the static stiffness, the frequencies, the mode shapes, and the damping. Only when the structural model is credible should the investigation move to the aerodynamic and coupled levels. Inverting this order wastes effort and can mask the real defect.

DO NOT ATTEMPT TO FIX AN AEROELASTIC MISMATCH BEFORE VERIFYING THE STRUCTURAL MODEL BENEATH IT.

The Danger of Frequency-Only Correlation

Natural frequencies are the most commonly reported correlation metric, and they are the least diagnostic. Two models with quite different mass and stiffness distributions can produce similar frequencies — the frequency depends on the ratio of stiffness to mass, so a model that is too stiff and too heavy can match the frequency of a model that is too soft and too light. The mode shape is what distinguishes them. A model that matches measured frequencies but has the wrong mode shapes will produce wrong generalised aerodynamic forces (which depend on the shape) and therefore wrong aeroelastic predictions, despite the encouraging frequency comparison. Mode shape correlation — via the modal assurance criterion (MAC) and visual comparison — is essential, not optional. A frequency match without a shape match is not a validated model.

CORRELATING ONLY NATURAL FREQUENCIES AND IGNORING MODE SHAPES CAN HIDE FUNDAMENTAL MODEL ERRORS. A model can match frequencies with incorrect mode shapes and still produce erroneous aeroelastic predictions.

Wind-Tunnel Aeroelastic Testing

Beyond GVT, wind-tunnel testing provides aeroelastic data that GVT cannot: the response of the (scaled) structure to aerodynamic load. An aeroelastic wind-tunnel model is a dynamically scaled replica designed so that its modal frequencies, mode shapes and mass distribution represent the full-scale article at the tunnel's speed and dynamic pressure. The model can measure static deflection under load, dynamic response to gust or control input, and — if driven to the condition — flutter. Wind-tunnel aeroelastic testing is valuable because it tests the coupled physics, but it carries the burden of dynamic scaling: the model must reproduce the right non-dimensional parameters (mass ratio, reduced frequency, stiffness ratio), and scaling distortions (Reynolds number, structural damping, material properties) must be understood and accounted for. A wind-tunnel aeroelastic result is only as representative as the scaling is faithful.

Stiffness Tests and Scaled-Model Challenges

Static stiffness tests measure the load–deflection behaviour of the structure (or a component) and validate the analytical stiffness independently of the dynamic modal test. This is Level 2 in the correlation hierarchy and is a direct check on the stiffness that governs both static aeroelastic deformation and modal frequency. Scaled aeroelastic models — whether for wind tunnel or for component test — face specific challenges: scaling structural damping is notoriously difficult (the full-scale damping mechanism may not scale geometrically), material properties at model scale may differ from full-scale, and the mass distribution must be reproduced with added mass that does not alter the stiffness. These challenges mean that a scaled-model result, however well executed, is not a direct substitute for full-scale evidence; it is a corroborating data point whose scaling distortions are documented and understood.

• Static stiffness tests validate the load–deflection behaviour independently of the dynamic modal test.
• Scaled aeroelastic models must reproduce mass ratio, reduced frequency and stiffness ratio — not just geometry.
• Structural damping is hard to scale; the full-scale damping mechanism may not scale geometrically.
• Material properties at model scale may differ from full-scale; document and account for the difference.
• A scaled-model result is corroborating evidence, not a direct substitute for full-scale evidence.

Cross-Reference: Test & Analysis Correlation

The broader discipline of test–analysis correlation — modal assurance criterion, correlation acceptance, model updating without overfitting, strain-gauge correlation, displacement and full-field correlation — is covered in detail in the Test & Analysis Correlation material. For aeroelasticity, the key cross-references are: modal correlation with MAC (for mode shape credibility), model updating without overfitting (do not tune the model to match one mode at the expense of others), and the recognition that correlation is evidence of adequacy for a purpose, not proof of absolute correctness. An aeroelastic model correlated at one configuration is not automatically correlated at another; re-correlate when the mass, stiffness or configuration changes.

Correlation is evidence of adequacy for a purpose, not proof of absolute correctness. Re-correlate when the configuration, mass or stiffness changes.

Key Takeaways

• GVT establishes the measured modal basis that the aeroelastic structural model must match.
• The correlation hierarchy runs from mass → static stiffness → frequencies → mode shapes → damping → aerodynamic load → coupled response.
• Verify the structural model before fixing aeroelastic mismatches; the root cause is usually structural.
• Frequency-only correlation is insufficient — mode shapes (MAC) are essential to validate the spatial mass and stiffness distribution.
• Wind-tunnel aeroelastic tests the coupled physics but carries dynamic-scaling burdens; scaled results are corroborating, not definitive.