Aeroelasticity Oral Exam questions and

answers
(2023-2024) He asks 6 questions randomly which are from a batch of questions. 3 of them are
10p, 2 are 20p and 1 is 30p, together it takes 20-25 min. He asks you ‘to answer as short as
possible, but as complete as possible’. In my opinion you can just start with the basics of the
topic/question and go more and more complex/in depth. He has specific words he wants to hear
for each question (for divergence for example: Static, instability, deflection and k to infinity,
driven by K_theta sweep and EA-AC distance, …). These questions are good practice as well
as the learning goals of each lecture.

1. Typical section- representative airfoil at about 65%-75% (⅔ to ¾ of span)- chosen to

represent frequency of bending and torsion
2. What is aeroelasticity (10 marker)
- Aeroelasticity deals with the behaviour of an elastic structure in an airflow
where there is significant interaction between the two.
3. Collar’s triangle - 3 parts: aerodynamic forces, inertia forces and elastic forces. Static
aeroelasticity for interaction between elastic and aerodynamic forces. Dynamic
aeroelasticity for interaction between all three. Flight dynamics connect aerodynamic and
inertial forces. Mechanical vibrations connect elastic and inertial forces.
4. What is a generalized coordinate?

5. What is the aerodynamic stiffness matrix? and why is the first column zero?/What does
the aerodynamic stiffness matrix look like?
The aerodynamic stiffness matrix is a matrix that is multiplied by a set of DOF
giving a force- similar to the structural stiffness matrix.
First column is zero as we take moments around the hinge point. Thus there is
no moment due to heave, only torsion (theta).

1
6. What is divergence, why is it bad?
-Divergence is a static aeroelastic phenomena in which the displacements of a wing go
to infinity in a static manner. (divergence is a stability problem- no dynamic terms
needed)
7. show why divergence is an eigenvalue problem (30 marker)
-(need to be able to argue with the equations) Fs=Fa -> KsX=KaX-> take to one side
and find eigenvalues. If matrix is: positive definite(all eigenvalues positive)-stable, semi
positive definite(all eigenvalues >=0)- boundary of stability, negative definite-unstable
-Deformation of a wing causes a change in aerodynamic loads, and in structural loads
counteracting the aerodynamic loads. At a certain speed the structural loads are no
longer large enough to balance the aerodynamic loads, this speed is known as
divergence speed. Because balancing is no longer possible, the aeroelastic system
becomes unstable. Therefore, divergence is called an aeroelastic stability problem. This
means that we can regard divergence as an eigenvalue problem, which brings many
advantages from a computational point of view.
8. How do we reduce/prevent divergence?
-Change torsional stiffness (weight penalty)
-Move location of elastic axis towards aerodynamic centre (-ve on our coordinate
system)
-Add backward sweep (increased washout, cop moves inboard, reducing torsion- opp for
swept fwd )
9. What are the important parameters for divergence?
- Torsional stiffness of the wing, distance between elastic axis and aerodynamic
centre, sweep angle
10. Why do we call it torsional divergence?
- The mode associated with heave is always stable as the heave displacement
doesn’t affect aerodynamic loads. Those are governed solely by the
pitch/torsional of the typical section. Divergence depends only on torsion

11. What is control reversal? How do you explain control reversal? (10 marker)
- A speed will be reached, q reversal, where a positive control deflection decreases the
angle of attack sufficiently to have no change in lift force. At speeds higher than this

2
 speed the control will act in the sense opposite to normal. I.e it is the point at which the
control effectiveness becomes negative for increasing control surface deflection
12. How are control reversal and divergence related?
- In general we can state that divergence sensitivity and control reversal are
opposite. A bending induced wash in is beneficial for control effectiveness but
detrimental for divergence
- For a swept back wing, CS effectiveness is more of a problem than divergence
due to wash out moment when bending. Vice versa for swept forward wings.

13. Explain the quasi steady aerodynamic model/ what are the limitations of this model?
- This is the case where the fluctuations are not steady, but they are so slow that
we assume all the disturbances we create in the flow are propagated instantly i.e
no lag b/w lift and change in circulation.
- Additionally, mass terms are ignored because the acceleration (frequency,
omega ^2) is still small compared to the (angular) velocity and displacement.
14. What is the quasi steady damping matrix (refer lec 6) (Ca matrix)/⁠What is in the
aerodynamic damping matrix?
- The aerodynamic damping matrix is a matrix that arises when we consider
quasi-steady aero. It is proportional to q/v_inf i.e proportional to v -> it changes
with increase in speed

3
15. What is the physical meaning of flutter k
- Flutter is a dynamic instability problem in which there is an average energy
transfer over a time period from the flow to the structure.(still uses eigenvalues to
assess- if the eigenvalue has a frequency component, we talk about flutter.
Divergence is a special case of flutter-> non-dynamic case in which there is only
negative damping and displacement goes to infinity without any frequency
component in the eigenvalue).
- Flutter is a word used to describe self-sustained oscillations arising from
fluid-structure interactions. It is normally implied that under certain conditions, the
oscillations can grow without bounds. (work done by lift>0) Constant pitching
constantly propagates vorticity in the wake which is constantly having an effect
on the airfoil pressure distribution although this effect will decay with distance
from the airfoil.

The flutter point is defined as the point where a damped oscillation transitions to
one growing in amplitude. A group of flutter points therefore forms a boundary
between a region where one can expect to operate a vehicle “normally” from one
where structural failure is likely. The region within the boundary is referred to as a
flutter envelope
16. What is the condition of aeroelastic eigenvalues to generate flutter?
–They need to be imaginary and sigma(imaginary part) should be moving from negative
to positive
17. Do we need aerodynamics damping to get flutter?
-No
18. Flutter diagrams both frequency and damping.⁠Draw frequency and damping diagrams
for flutter (need to know where flutter occurs in both of these)
-refer lec 6 (rec lec 7)
19. Why does an airfoil with only heave never flutter?

4
 - Airfoil heaving upwards- reduced AoA-lower (or negative) lift, Airfoil heaving
downwards- increased AoA-higher (or positive)lift. So the motion of the airfoil isn't
made worse by the addition of lift or we can say that it is damped.
20. On frequency speed diagram where do you have flutter
Where the bending and torsion frequencies have grown towards each other, as
they start to resonate, for quasi-steady models, they don’t need to coalesce
because we now have damping which can stabilise or destabilise the system.
21. How to prevent flutter by design?
-More fwd position of CG towards LE/ lesser dist b/w CG and EA or even CG ahead of
EA- higher flutter speeds

Assumptions of Theodorsen:
- Element Gamma (wake Vortex) moving to the right relative to the airfoil with velocity v
(free stream)- normally we would say wake is slowed down
- U is harmonic, replacing time with Reduced frequency, thus moving from time domain to
frequency domain.
22. Describe (draw) the Theorodson function in state space/⁠draw the Theodorsen function in
imaginary plane, explain why we need it.

5
- Theodorsen function, c(k) is real and=1 at k=0, and =0.5 when k=inf (if we oscillate with
inf frequency, all TE vortices cancel each other out apart from the starting vortex).
- We can use this to get circulatory lift and then directly find circulatory moment by
multiplying with (0.5+a)b
23. Is Theodorsen formulation in time or Frequency domain? Why Theodorsen written in
terms of frequency and not time?(10 marker)
-Frequency domain, real part is scaling, imaginary part is phase shift
The imaginary part represents the phase shift b/w input and output (eg: lift for an
oscillating wing)- tells us whether to use steady,quasi steady or unsteady models
This phase shit also affects flutter as it affects when lift and heave rate are
in the same direction.
24. Which terms in Theodorsen are dependent on k
-(everything except the mass matrix)
25. 2 questions on Theodorsen on unsteadiness and the general theory of circulatory and
non circulatory parts
- Circulatory part- governed by downwash at 3c/4 pt due to the lumped vortex
system
- Non circulatory part- consists of apparent mass terms
26. Theodorsen-explain to me where does the circulatory part comes from and then he
wanted to know a bit of how he imposed the vortexes on the airfoil when deriving the
theory(20 marker)
- The whole circulatory part comes from a downwash at the 3c/4 pt
- Vortex at AC and vortex in the wake
27. What is the reduced frequency k?
- k=⍵*b/v, substituted in U to go from time to frequency domain

28. reduced a frequency and relation to unsteady and quasi steady aeroelasticity
- Reduced frequency is present in Theodorsen function, C(k) which is multiplied by
the circulatory terms [which are the same between unsteady and quasi-steady
(M_theta = L_circ + (0.5+a)b)].
- We put C(k)=1 for steady and quasi steady. I.e k= values up to 0.02, past that, it
is unsteady
- We can’t do this for the non-circulatory part- written in the time domain.
29. What matrices are dependent on k in Theodorsen
-Damping and stiffness matrices

6
 30. in the convolution integral, what do we multiply the wagner function with?
- Rate of change of angle of attack

31. Explain the rational approximation and the lag terms involved (20 marker)
- A method to convert frequency domain to time domain.
- Roger’s approximation formula and use laplace domain to get the Fa
- Coefficients of matrices are determined by curve fitting methods (eg:least
squares)
Write down the Rogers Approximation and explain the lag terms that cause delay in lift buildup
due to unsteady pressure distributions - link it to the Theodorsen expression and the Jones
approximation of C(k).

32. Why can’t the Lagrangian be used for calculating the displacements in
unsteady/quasi-steady aerodynamic models? (10 marker)
- The Lagrangian assumes that the work done on the system is purely
conservative in nature, while in unsteady/quasi-steady aerodynamics, the
aerodynamic forces involve a damping term, and are therefore, non-conservative
in nature.This requires a different approach, requiring the kinetic energy to be
differentiated twice - first wrt velocity and then wrt time, which is then summed to
the derivative of the internal energy wrt position and equated to the
non-conservative loads.
33. What do we use modes for? (10 marker)/How are modes used in aeroelasticity? (10
marker)

7
 1. They give us a physical representation of the dynamic properties of the
model
2. Modes of a linear structural model are orthogonal because mass and
stiffness matrices are symmetric
3. We can use modes to reduce the size of the aeroelastic system by 1 or 2
orders of magnitude.
34. Why do we use eigenmodes?
35. Complex eigen frequencies
36. Explain complex eigenvalues in aeroelasticity? How can they be plotted? (10 marker)

Due to unsteady aerodynamics (inclusion of the mass and damping terms), the aerodynamic
stiffness matrix is complex, therefore, the eigenvalues of the stiffness matrix (Ks - Ka) are also
complex in nature.
Real eigenmodes can be interpreted spatially as standing waves and are visualised quite easily.
However, complex eigenmodes need to be plotted in the time-domain and cannot be fully
interpreted spatially.

37. K method
- The k method is a method to calculate the flutter speed of a wing which is
subjected to a fully unsteady aerodynamic model.
- The assumption: aeroelastic response of the wing is purely harmonic, so only
dependent on the reduced frequency k i.e p=ik (k=omega*b/v)
- we know the semi chord b, and we chose a value for rho, so we fly at a certain
altitude->we chose k-> calc B(k)->calc eigenvalue of B(k)-> use to get freq,
omega and artificial damping-> use omega and k to get vel,v-> repeat until g=0

38. Pk method:

8
39. Describe p method (30 marker)
- Use Roger’s approximation for the unsteady aerodynamics.
- Assume ik to be a complex number p.
- Solve the aeroelastic equations directly for p, no iterations necessary.
Pros and cons of the p method
• Aerodynamic forces are also function of the decay factor, so also valid for
larger damping values.
• The p method equations can be transformed into time domain.
• The flutter point is not exact as the aerodynamic forces are approximated.

40. How do you go from second order to first order in state space (double variables by
adding lag terms)/state-space explain how we can reduce the order of the system
(identity matrix, lag terms, introduction of new variable, equation)

9
 -

State these red equations. You basically relate the aero-elastic equations to x doubledot
so you can put it in a matrix. You need lag terms since you approximate the aerodynamic
matrix in time, not in frequency
- In state space we have time domain term and an input allowing us to do- time
response, gust response, landing impact

41. ⁠what is a continuous gust?- Non deterministic/stochastic load (10 marker)

-Vertical/horizontal load which is prescribed in the frequency domain

42. What is a discrete gust? (10 marker)

10
 - Vertical/horizontal load which is prescribed in the time domain (need to
remember 1-cos as the shape of the gust)

43. draw & explain the Küssner function

44. What are clustered loads?(10/20 marker)

Video lecture 13, so basically we have many load cases combined in one figure
(envelope of correlated loads). This generates a “potato” where the outer contour
describes the extreme load combinations to be countered.
45. Explain correlated loads (20 marker) and how it is used for component sizing (wing,
HT/VT, etc.)
- Combination of loads between multiple DOF (eg: bending, torsion), we can't just
size for one dof,
46. How do we make sure we can solve for all the lag states?

11