Вы находитесь на странице: 1из 27

Active Flutter Suppression by the Method of Receptances: Experimental Results

Evangelos Papatheou, John E Mottershead and Jonathan Cooper Centre for Engineering Dynamics University of Liverpool
DiPaRT Loads and Aeroelastics Workshop Bristol, 13th of December 2012
1

Outline
Background & receptance method Experimental rig Testing & simulation results Closed-loop experiments Future work

1. Background

Flutter suppression in aeroelasticity


Flutter is a self-feeding, unstable vibration, potentially destructive In general, flutter will be caused by the coupling of the modes of the system and/or at least one mode will have zero damping Flutter suppression may be treated with eigenvalue assignment by increasing damping or the frequency spacing between modes

4. Closed-loop experiments

Flutter Margin
Zimmerman N.H. and Weissenburger J.T. (1964)

Quadratic fit flutter speed prediction

1. Background

Receptance method
Active vibration control by pole assignment Known as Receptance Method, can be used for any input-output transfer function Based on vibration measurements, not on physics based models (M,K,C matrices) No need for model order reduction or observers Single-input method which can be extended to multiinput
5

1. Background

Receptance method
General system with feedback

&(t ) + Cx & (t ) + Kx(t ) = r(t) + f (t) M& x


Input:

(1)

r(t)
f (t ) = bu (t )

Single input control force:

For velocity and displacement & gT x (2) u ( t ) = f T x Feedback typically


6

1. Background

Combining eq. (1) & (2) and Laplace transform:

(s M + sC + K + b(g + sf ) ) x (s ) = r (s )
2 T

(3)

Closed-loop system transfer function

x(s) T 2 = s M + sC + K + b (g + sf ) r( s)

(4)

only modification from open loop

x( s) = s 2 M + sC + K r( s)

rank 1 modification

Open loop receptance


7

1. Background

Sherman-Morrison Formula
Determine the inverse of a matrix after a rank-1 modification if the original inverse matrix and the modification are known:
Z and Z = (Z + uv
-1 1 T 1

1 T 1 ( ) ( ) ( ) (s ) Z s u s v s Z Z 1 ( s) = Z 1 (s ) 1 + vT (s ) Z 1 (s )u(s )

(4)

b( sf + g )
rank 1 modification

H(s )b (sf + g ) H(s ) H( s ) = H(s ) T 1 + (sf + g ) H(s )b


Closed-loop receptance
8

(5)

1. Background

Pole placement by receptance method


In practice
H(s )b (sf + g ) H(s ) H( s ) = H(s ) T 1 + (sf + g ) H(s )b
T

H(s) can be measured through H(j) and by curve fitting like rational fraction polynomials e.g. PolyMAX Given: Find:
H( s ), b, and a complex set
g Rn f Rn

{1

2 L 2 n }

closed under conjugation such that

(g + k f )T H ( k ) b = 1
g and f real if system is controllable
9

2. Experimental rig

Experimental rig

NACA 0018 airfoil Chord 0.35 m Span 1.2 m

2 DOF system pitch / heave

10

2. Experimental rig

Experimental rig
Aerofoil Torsion Bar

Flap Torsional Stiffness

V-stack piezo actuators

Vertical Stiffness

11

3. Testing & simulation results

Modal analysis
Pitch mode 3.9 Hz

12

3. Testing & simulation results

Modal analysis
Heave mode 6.7 Hz

13

3. Testing & simulation results

Modal analysis
Bending mode 41 Hz

14

3. Testing & simulation results

Modal analysis
Torsion mode 47 Hz

15

3. Testing & simulation results

16

3. Testing & simulation results

Receptance method - pole assignment


Single input single output feedback Acceleration velocity (derived) feedback Curve fitting of open-loop FRFs Assign the second mode (heave) by increasing the damping by 1 % Use the measured data to calculate the feedback gains g and f Sherman Morrison formula for closed-loop system

17

3. Testing & simulation results

Curve fitting

Fit pole-residue model (rational fraction polynomial) to obtain H(j) Pole assigned from -0.48 41.63j to -0.9 41.63j g = 77 f = -1756
18

3. Testing & simulation results

Pole assignment - simulation

Damping increased in the plunge mode from 1 % to 2 %, pitch mode is also affected
19

4. Closed-loop experiments

Real time control - pole assignment


Single input single output feedback (first approach) dSPACE for real-time control Curve fitting of open-loop dSPACE FRFs get g and f Velocity (laser) displacement (laser) SISO approach SIMO (two outputs - two sensors) Displacement and velocity (derivative)

20

4. Closed-loop experiments

Pole placement Experiment both poles assigned to increase damping and frequency spacing

Heave placed from 6.83 Hz to 7 Hz and 0.5 % damping increase Pitch from 3.89 Hz to 3.5 Hz and damping 1.5 % increase
21

4. Closed-loop experiments

Damping increase in heave mode

22

4. Closed-loop experiments

Damping increase in heave mode

23

4. Closed-loop experiments

Pole placement Experiment frequency separation

Heave placed at same frequency and 0.5 % damping increase Pitch from 5.65 Hz to 6.04 Hz and damping 0.5 % increase
24

4. Closed-loop experiments

Frequency damping trends root locus

25

4. Closed-loop experiments

Flutter Margin

Predicted flutter speed increases from 17 m/s to 20 m/s

26

Conclusions & future work


It is possible to assign damping and frequency spacing in modes of a 2-DOF airfoil with the receptance method By assigning poles in one wind speed and using the same gains g, f we can increase the flutter speed by 15 % FM assignment Introduction of non-linearity on the structure with the ultimate goal of non-linear active control

27

Вам также может понравиться