Universidad Carlos III de Madrid

Aeroelasticity
Bachelor in Aerospace Engineering

Wing divergence and normal modes analysis

Martín Alejandro Moracho García 100428643

Rodrigo Martín Ortiz 100432541
Wing divergence and normal modes analysis Homework I

Contents
1 Introduction 2

2 Relationship between structural and viscous damping 2

3 Flutter prediction using Flight Vibration Test data 3

3.1 Direct damping evolution method . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.2 Flutter Margin method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

4 Conclusion 7

List of Figures
1 FVT frequency data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 FVT frequency and damping data including data scatter . . . . . . . . . . . . . . 3
3 FVT frequency evolution with flight speed prediction . . . . . . . . . . . . . . . . 4
4 FVT damping evolution with flight speed prediction . . . . . . . . . . . . . . . . 5
5 Flutter speed prediction based on damping evolution criteria . . . . . . . . . . . . 5
6 Flutter Margin evolution with flight-speed prediction . . . . . . . . . . . . . . . . 6

List of Tables

1
Wing divergence and normal modes analysis Homework I

1 Introduction
In aeronautics, "flutter" stands as a critical phenomenon where aerodynamic forces and the
structural flexibility of an aircraft interact, potentially leading to self-excited oscillations. This
aeroelastic instability necessitates a thorough understanding and prediction for the safe design
and operation of aircraft.
Predicting flutter is paramount for maintaining structural integrity. The resonance between
aerodynamic loads and the inherent flexibility of an aircraft’s components can lead to excessive
vibrations, posing risks of structural failure. Safety assurance is a key outcome of accurate flutter
prediction, helping establish operational limits and preventing encounters with aeroelastic issues
during flight.
Two methods commonly employed for predicting flutter speed are the direct damping evolution
with flight speed and the flutter margin expression, such as that proposed by Zimmerman &
Weissenburger.
Direct Damping Evolution with Flight Speed analyzes the evolution of damping (g) with
changes in flight speed using measured Free Vibration Test (FVT) data. Flutter Margin Expression
utilizes an analytical expression, like the one proposed by Zimmerman & Weissenburger, to
estimate the margin between actual flutter speed and the maximum operating speed. In Homework
2, we are going to perform a flutter analysis for the given modes of a new aircraft comparing
these two methods in order to predict the flutter speed.

2 Relationship between structural and viscous damping

When it is assumed that damping force is proportional to velocity and frequency we are
using viscous damping representation. However, practical tests have shown that damping in
structures and materials is independent of the frequency, but acts in quadrature (90º phase)
with the displacement of the system. This is due to the structural damping (g) that acts as an
internal damping mechanism. For this reason it is needed to combine damping and stiffness
properties of a system into a complex stiffness k*:

k ∗ = k(1 + ig), (1)

Then, we can write the one degree of freedom equation of the system as:

mẍ + k(1 + ig)x = f (t). (2)

This equation is not possible to solve in this form, but it can be rewritten in the time domain
form as:
mẍ + ceq ẋ + kx = f (t) (3)

Where ceq = gk/ω. Remember that stiffness k can be written as k = mωn2 and the viscous
damping factor is defined as ζ = c/2mωn . With all of this ζ is:

g · m · ωn2 g  ωn 
ζ= = (4)
2 · m · ωn · ω 2 ω

If the system is vibrating at the natural frequency, ζ can be approximated as:

2ζ = g (5)

2
Wing divergence and normal modes analysis Homework I

3 Flutter prediction using Flight Vibration Test data

In this section, the data collected for a prototype aircraft during a FVT is processed and an
attempt is made to predict flutter speed using several techniques.
The FVT data is presented in the following figures:

Figure 1: FVT frequency data

Figure 2: FVT frequency and damping data including data scatter

Where the reference damping of each mode was taken as:

• G1 = 0.023
• G2 = 0.007
In order to take the FVT measurement scatter into account, upper and lower limits for each

3
Wing divergence and normal modes analysis Homework I

data point are created based on the scatter they present. This fact allowed the team to estimate
upper and lower bounds for the expected flutter speed, reducing the uncertainty while making
the decision to keep testing the aircraft or not. The same procedure was used for both presented
methods in this paper.

3.1 Direct damping evolution method

The first approach to estimate flutter speed is based on the direct damping evolution with
flight speed. In order to do so, an extrapolation of the available FVT data is required to predict
frequency and damping at larger airspeeds, not yet flown due to safety concerns regarding flutter.
The results of this extrapolation will be used to decide whether the flight tests can continue
for increasing speeds, or whether the aircraft’s design needs to be revised in order to postpone
flutter to a larger velocity, desirably 15% greater than the diving velocity of the aircraft.
The results of the processed and predicted FVT data are shown in the following figures, where
the green and yellow lines represent the upper and lower bounds of the evolution estimation:

Figure 3: FVT frequency evolution with flight speed prediction

4
Wing divergence and normal modes analysis Homework I

Figure 4: FVT damping evolution with flight speed prediction

As stated in the paper by Zimmerman, N.H. And Weissenburger, the intersection between
both modes in the frequency evolution plot may indicate approximately the appearance of
flutter. However, the frequency graph cannot be used as an accurate criteria for flutter speed
determination. Instead, one needs to turn to the damping plot for an accurate and physical
understanding of the flutter phenomena. Regarding this criteria, a negative damping value
indicates that the amplitude of the system’s vibration increases over time, which implies that
the aircraft is extracting energy from the flow, i.e., flutter phenomena appears.

Figure 5: Flutter speed prediction based on damping evolution criteria

As it can be observed in figure 5, the damping of Mode 2 is predicted to reach a negative

value for V1 = 241.9kt. Regarding the upper and lower bounds, the error is quite small and
will not influence the decision of continuing the testing or not. The lower bound for Vf 1 is
Vf 1,low = 241.5kt, while the upper one is Vf 1,up = 242.3kt. This small error in the flutter speed

5
Wing divergence and normal modes analysis Homework I

prediction comes from the small absolute errors of the damping measurements of Mode 2.
The nature of the appearing flutter will be an unstable increment in the amplitude of Mode
2, quickly leading to the aircraft’s failure. In addition, one can predict the severity of the flutter
mechanism by observing the flight speed for which the damping becomes g = −0.03. In this case,
this is approximately Vg1 = 308.5kt. This makes the difference between flutter appearance speed
and critical flutter speed equal to ∆V = 66.6kt. The severity of this flutter case is categorized as
"Moderate", since the negative slope of the Mode 2 curve is relatively high, but not too abrupt.

3.2 Flutter Margin method

Another available approach to estimating flutter speed can be found in the paper “Prediction
of Flutter Onset Speed Based on Flight Testing at Subcritical Speeds” by Zimmerman, N.H. And
Weissenburger. This paper provides a formula to calculate a parameter called "Flutter Margin",
where β = ωζ = ωg 2 is called "system decay rate".
2
(ω22 − ω12 ) (β22 − β12 )

Fmargin (ω1 , ω2 , β1 , β2 ) = +
2 2
 !
(ω22 + ω12 ) (β2 + β1 ) 2

+ 4β1 β2 +2 (6)
2 2
 !2
(β2 − β1 ) (ω22 − ω12 ) (β2 + β1 ) 2

− · +2
(β2 + β1 ) 2 2

Using FVT data, one can calculate Flutter Margin as a function of the measured frequency
and damping data of the coupled modes that are of concern regarding flutter. After that, the
paper demonstrates that one can approximate the evolution of the Flutter Margin as a quadratic
function. Using this fact, it is possible to calculate the Flutter Margin using the available FVT
data, after which a quadratic polynomial fitting is performed to estimate the evolution of the
Flutter Margin.

Figure 6: Flutter Margin evolution with flight-speed prediction

6
Wing divergence and normal modes analysis Homework I

As it can be observed in figure 6, flutter is expected to occur at Vf 2 = 250kt. Regarding

the upper and lower bounds, the error is quite small and will again not influence the decision
of proceeding with the testing or not. The lower bound for Vf 2 is Vf 2,low = 248.2kt, while the
upper one is Vf 2,up = 253.1kt.

4 Conclusion
Firstly, note that, even though the flutter speed estimation is similar for both procedures,
some deviation ∆Vf = 8.1 still occurs. This constitutes a relative error of 3.3 %, which can be
considered acceptable taking into account the small range of velocities for which the aircraft was
tested.
The nature of this error is believed to appear from the fact that the aircraft was tested at
only 4 different flight-speeds, so many of the extrapolation methods available do not work for this
exercise. Some degree of error must then have been introduced by the extrapolation algorithm
due to the lack of test points.
The team now faces a decision whether to continue testing the aircraft at higher flight-speeds
or not. The team expected a compromise to be needed while making this choice. However, in
this case, it is clear that higher velocity testing of the aircraft needs to stop immediately. The
last test point was taken at V = 225kt, so flutter is just around 15 kt away from appearing,
which is an incredibly small margin knowing the prediction difficulties flutter implies.
It is clear that the aircraft needs to be re-designed in order to change the dynamic coupling
of the bending and torsion modes, such that flutter occurs at much higher speeds. At most, the
aircraft would be eligible to perform further testing at lower speeds than V = 225kt, so that the
prediction can be made more accurately with more data points. Without any data of the aircraft,
it is tough to suggest any possible changes.