Documente Academic
Documente Profesional
Documente Cultură
COURSEWORK
88486
DECEMBER 7, 2014
UNIVERSITY OF SUSSEX
School of Engineering and Informatics
Summary
In this report, aero-elastic instability (flutter) is analysed by using a mathematical model
to investigate an aero-elastic design considered with special specifications, study was
considered with special regard given to the prevention of flutter.
Two Newtonian equations of motion were derived from an initial free body diagram of a
mass – spring (2DOF) system. These equations were then put into the state space form
𝑥̇ = 𝐴𝑥, and an Eigen-value analysis of a theoretical aircraft wing were carried out using
MATLAB.
A 5m section of a theoretical wing was studied. The wing was modelled from a Boeing
DC – 10 aircraft. The chord length was 13.73m; the distances from the leading edge to the
centre of lift and centre of mass were 4.6m and 5.71m respectively. The mass of the
section was 25270Kg. The study was applied to an aircraft at free stream air velocity of
272.7m/s; air density was 1.2 Kg/m³ and the stiffness values of the springs were assumed
to be in the range of 1x104 Nm - 1x105 Nm for a stable aircraft.
For the K1 (Nm) the values were 100000Nm for Non-Oscillatory Divergent, 1,000,000Nm
for Oscillatory Divergent (Flutter) and 100,000Nm for Stable Oscillation. While for K2
(Nm) the values were 200,000Nm for Non-Oscillatory Divergent, 50,000Nm for
Oscillatory Divergent and 50,000Nm for Stable Oscillation.
The report also discusses the limitations of using classical theory to predict aero-elastic
instability. One of those limitations was that steady flows were used in the calculations
rather than unsteady flow
i
Table of Contents
Summary ............................................................................................................................................................... i
Table of Figures ................................................................................................................................................. 2
Introduction ........................................................................................................................................................ 3
Question 1 ............................................................................................................................................................ 4
Question 2 ............................................................................................................................................................ 4
Question 3 ............................................................................................................................................................ 7
Question 4 ............................................................................................................................................................ 9
Non Oscillatory Divergence: ................................................................................................................. 12
Unstable oscillatory divergence (flutter): ...................................................................................... 12
Stable Oscillation: ..................................................................................................................................... 12
Question 5 ......................................................................................................................................................... 14
Conclusions ...................................................................................................................................................... 14
Appendix A1 MATLAB code For Stable Oscillation ......................................................................... 15
Appendix A2 – MATLAB Code for Non-Oscillatory Divergent .................................................... 16
Appendix A3 – MATLAB Code for Oscillatory Divergent (Flutter)............................................ 17
References ........................................................................................................................................................ 18
1
Table of Figures
Figure 1 - An American Airlines McDonnell Douglas DC-10 lowers its landing gear ........... 3
Figure 2 - Wing of a Boeing DC-10............................................................................................................. 4
Figure 3 - Elastically Mounted Airfoil Section (From Question Sheet) ...................................... 4
Figure 4 - Free Body Diagram of the airfoil section............................................................................ 5
Figure 5 - A DC-10 Aircraft ........................................................................................................................... 9
Figure 6 - CAD Model of DC-10 ................................................................................................................... 9
Figure 7 - Comparison of the dimensions of the CAD model to the Boeing DC-10
technical specification .................................................................................................................................... 9
Figure 8 - CAD Model of the airfoil Section ......................................................................................... 10
Figure 9 - Summary of Values and Results.......................................................................................... 13
2
Introduction
This report analyses the design of an airfoil for the application of a large passenger
aircraft. The aircraft used in this analysis is the Boeing DC- 10 Series 10.
The DC-10 was produced in 3 series – the series 10, 30 and 40. The series 10 were used
for domestic routes to 3500 miles while series 30 and 40 were used for extended range
and intercontinental travel. The DC-10 was designed and built in Long Beach, California,
by the then Douglas Aircraft Company known today as the Long Beach Division of
Boeing Commercial Airplanes. Production of this aircraft started in 1968 and in 1971,
saw its first deliveries. From 1968 to 1989, 386 commercial DC-10s were delivered, plus
60 KC-10 tanker/cargo models built for the US Air Force [1].
http://images.bwbx.io/cms/2013-01-17/0117-aa-dc-10-plane-630x420.jpg
Figure 1 - An American Airlines McDonnell Douglas DC-10 lowers its landing gear
The DC-10s enjoyed success until 1989 when its last model was produced for Nigeria
Airways. This was because it was hampered by tragic accidents described by Jordan
Rane for CNN [3]
The design process is to consider the problem of flutter. Since the invention of the first
aircraft, aero-elastic stability has been a key area of interest and concern not only to
aviation engineers but also to all engineers.
The flutter phenomenon was first discovered in a Handley Page bomber in 1918 [4] and
since then the study of instability and unstable vibration has become even more
important in aviation. More important in the sense that today, engine technology
advances have enabled even much higher velocities.
3
Question 1
The chosen airfoil design is the one used in the wings of a Boeing DC-10 Passenger
Aircraft.
http://airchive.com/blog/wp-content/uploads/2014/02/FINAL-DC-10-FLIGHT-BG008-WING-TAKE-OFF-FEB-2014-4-300x199.jpg
Question 2
4
The model above has 2 degrees of freedom (2DOF), which corresponds to the
displacement variables y (vertical displacement) and θ (angle of attack). The model is
mounted on two springs which represent the stiffness and elastic behaviour of the wing
and its material, they are linear for small values of θ. In order to solve most engineering
problems, a free body diagram is essential. Figure 4 below shows the free body diagram
of the airfoil section.
+ve
𝑭𝑳
y
Mg
𝐕∞
G
𝒅𝟐
𝒅𝟑 θ
3
𝒅𝟏
𝑹𝟏 𝑹𝟐
FL is the lift force which acts at the centre of lift. G is the centre of mass of the wing. 𝑅1
and 𝑅2 are the resultant forces from the springs at the leading and trailing edges, they
have spring stiffness values of 𝑘1 and 𝑘2 . 𝑑2 , 𝑑3 and 𝑑1 are the distances from the leading
edge to; the centre of lift, the centre of mass and the trailing edge respectively. V∞ steady
stream air velocity. b is the span of the wing The sign convention for this analysis will be
positive for vertically upwards and clockwise.
Forces
1
The lift force, 𝐹𝐿 = 2 𝐶𝐿 𝜌𝐴𝑉∞2
5
𝑅2 = 𝑘2 [(𝑦 + 𝜃(𝑑1 − 𝑑3 )] (3)
Moment
𝛴𝑀 = 𝐼𝐺 𝜃̈ = 𝛴𝑀𝑅 + 𝑀𝐹 (9)
Using the sign convention where clockwise motion is positive and anticlockwise
negative, the moment created from the reaction forces is defined as:
𝑀𝐹 = 𝐹𝐿 (𝑑3 − 𝑑2 ) (12)
(8) and (14) are the 2DOF rigid body equations of motion for the model.
6
Question 3
Putting (8) and (14) into state space:
For equation (8), factorising in terms of y and 𝜃 and equating to zero gives:
(𝑘1 +𝑘2) 2)
(𝑘1 𝑑3 +𝑘2 (𝑑1 −𝑑3 )−5𝜌𝑏𝑑1 𝑉∞
𝑦̈ + 𝑦 [ ]+𝜃[ ]=0 (17)
𝑚 𝑚
𝒅𝟐 𝒚
But 𝑦̈ can be written as 𝒅𝒕𝟐 and by making
(𝑘1 +𝑘2 ) 2)
(𝑘1 𝑑3 +𝑘2 (𝑑1 −𝑑3 )−5𝜌𝑏𝑑1 𝑉∞
𝑎11 = [ 𝑚
] and 𝑎12 = [ 𝑚
] (17) can be written as:
𝒅𝟐 𝒚
+ 𝒚[𝒂𝟏𝟏 ] + 𝜽[𝒂𝟏𝟐 ] = 𝟎 (18)
𝒅𝒕𝟐
𝑑2 𝜃
Similarly, dividing (16) by 𝐼𝐺 and making 𝜃̈ = gives:
𝑑𝑡 2
By making,
𝑘1 𝑑3 − 𝑘2 (𝑑1 − 𝑑3 )
𝑎21 = [ ]
𝐼𝐺
𝒅𝟐 𝜽
𝒅𝒕𝟐
+ 𝒚[𝒂𝟐𝟏 ] + 𝜽[𝒂𝟐𝟐 ] = 𝟎 (20)
(21) has a second order term (𝑧̈ ) and so to rewrite the equation with only first order
terms, the following state variables are chosen:
𝑥1 = 𝑧
7
𝑥̇ 1 = 𝑧̇
𝑥2 = 𝑧̇
𝑥̇ 2 = 𝑧̈
But 𝑥̇ 1 = 𝑥2 (22)
𝑥̇ 1 = 𝑥2 (25)
𝒙̇ 𝟏 𝟎 𝑰 𝒙𝟏
[ ]=[ −𝟏 ][ ] (27)
𝒙̇ 𝟐 −[𝒎] [𝒌] 𝟎 𝒙𝟐
8
Question 4
The airfoil design is that of a Boeing DC-10. A 3D CAD model of the aircraft was
downloaded from grabcad.com [6] for the analyses.
http://fl410.files.wordpress.com/2011/11/american-airlines-mcdonnell-douglas-dc-10-30.jpg
https://d2t1xqejof9utc.cloudfront.net/screenshots/pics/b0c7e13b2e7f0a65df5dd5867f48102a/medium.jpg
To see if the downloaded CAD model was to scale, it was measured against the original
specifications for the aircraft from the Boeing DC-10 website [7]
The table below compares the dimensions of both the original aircraft and the
solidworks model.
Wingspan 47.3 50
Figure 7 - Comparison of the dimensions of the CAD model to the Boeing DC-10 technical
specification
9
From Figure 7, it is seen that the dimension comparison of the CAD model to the original
model is very small. Therefore it is assumed that the CAD model was to scale. After
confirming the scale certainty of the CAD model, all the other parts of the aircraft were
suppressed except for one wing. Solidworks was then used to calculate the centre of mass
on the 2D surface. For this analyses, because the centre of lift changes during the flight, it
was assumed that it will be one-third of the overall airfoil distance from the leading edge
(see Figure 8)
For the section of wing used for the analyses, it was assumed that (b = 5m). Reason for
this being that the wing does not change direction until it passes this point.
For the analyses of any engineering product, it is important to note that it is always a good
engineering practice to make calculations using the worst/maximum possible scenario.
For our model therefore, we will analyse using the weight of the plane when it is at its
heaviest (during take-off) and its fastest.
From [7], the take-off weight of DC-10 is 195,045kg and the fastest the plane can travel is
272.7m/s (This is used for the velocity). Thus, for this analyses, it will be assumed that
the weight of the wing 25,270kg – based on an educative guess.
For the values of d1, d2, and d3 for the airfoil see Figure 8 - CAD Model of the airfoil
Section.
10
The only values that have not been given are the spring constant values for k1 and k2.
These variables can be changed to create a stable motion, non-oscillatory divergent
motion or flutter.
The values of k1 and k2 were assumed to be in the range of 1x104 Nm – 1x105 for a stable
oscillation because values in this range will make the airfoil a rigid structure which is
what we want.
A code was written using MATLAB based on the state space equations derived in Question
3. The code can be seen in the Appendix {A1}. The input variables were 𝑑1 , 𝑑2 , 𝑑3 , m, V∞,
b, 𝑘1 and 𝑘2 . IG was calculated in the code. The outputs are A, B and A². A message is also
displayed as to whether the values of the input variables resulted in; non-oscillatory
divergence, stable oscillation or unstable oscillatory divergence (flutter). These were
calculated based upon the conditions for each case with regards to the values of A, B and
A².
Using the same equations for𝑎11 ,𝑎12 , 𝑎21 and 𝑎22 It can be seen that.
𝑎11 𝑎12
−[𝑚]−1 [𝑘] = [𝑎 𝑎22 ] (28)
21
Now we need to decompose the problem to obtain a quadratic in terms of 𝜆. This can be
done by writing (28) in terms of eigenvalues.
𝜆2 + 𝑎11 𝑎12
[ 2 ] (29)
𝑎21 𝜆 + 𝑎22
2
2
(𝑎11 + 𝑎22 ) (𝑎11 + 𝑎22 )
𝜆 =[ ± √( ) − (𝑎11 𝑎22 − 𝑎12 + 𝑎21 )]
2 2
1
2 2
𝑎 +𝑎 𝑎 +𝑎
𝜆= ± [− ( 11 2 22 ) ± √( 11 2 22 ) − (𝑎11 𝑎22 − 𝑎12 𝑎21 )] (31)
11
To simplify, two new variables are defined; A and B;
(𝑎11 +𝑎22 )
Where A = and B = (𝑎11 𝑎22 − 𝑎12 𝑎21 )
2
The three conditions can be determined from the results of A and B as follows:
From the Matlab code produced in Appendix I, after all the input values were calculated,
the outputs A, B and A2 where:
A = -267.2951
B = -3.4648e+03
A2 = 7.1447e+04
From the Matlab code produced in Appendix I, after all the input values were calculated,
the outputs A, B and A2 were:
A = 95.2302
B = 1.3167e+04
A2 = 9.0688e+03
Stable Oscillation:
−𝐴 ± √𝐴2 − 𝐵 is always real but < 0.
12
1/2
(−𝐴 ± √𝐴2 − 𝐵 ) is always imaginary
From the Matlab code produced in Appendix I, after all the input values were calculated,
the outputs A, B and A2 were:
A = -205.7801
B = 91.0556
A2 = 4.2345e+04
The table below summarises the values and result of the analyses
b(m) 5 5 5
13
Question 5
The classical theory of using the mathematical model to predict aero elasticity has various
limitations.
One limitation is that for the case of predicting flutter, results are inaccurate due to
changes in mass which will affect the dynamic response of the wing. Changes in mass
comes as a result of fuel consumption because most fuel tanks of aircrafts are often
located in their wings.
Another limitation is the inability of this analyses to take into account the flow over the
entire span of the wings. The flow over a wing is different at different points along the
span of the wing. Consequently, the dynamic response of the wing will also vary along its
span
For this analyses, steady flows are used. The use of unsteady flows is therefore another
limitation.
Finally, this analyses only considered spring stiffness for small values of pitch (θ). This is
because spring stiffness is only linear for small values of (θ). For higher values of (θ),
spring stiffness becomes non-linear and much harder to predict.
Conclusions
This report demonstrated how classical flutter theory is used to predict the dynamic
response of an airfoil, and what can be done to reduce the likelihood of flutter occurring
at the design stage.
This analysis also showed that flutter becomes more probable as the mass decreases.
During long haul flight, the mass of the wings will change due to the fuel tanks being
located in the wing structure.
This analysis provided a very useful starting point from which to begin understanding the
design problems of flutter. The approach used also has a lot of limitations as discussed,
and may not be advisable to use in practice.
14
Appendix A1 MATLAB code For Stable Oscillation
d1=13.73; %Enter chord length (m)
d2=4.6; % Enter distance from leading edge to centre of lift
(m)
d3=5.71; %Enter distance from leading edge to centre of mass
(m)
b=5; %Enter width of wing (m)
k1=100000; %Enter spring stiffness coefficient 1 (leading) (Nm)
k2=50000; %Enter spring stiffness coefficient 2 (trailing) (Nm)
v=272.7; %Enter free stream air velocity (m/s)
m=25270; %Enter mass of wing (Kg)
I=((m*(d1^2))/12); % Moment of mass (Kg*m^2)
rho=1.2; % Density of air (Kg/m^3)
Area=(b*d1); % Projected Area (m^2)
a11=(k1+k2)/m;
a12=(((k1*d3)+(k2*(d1-d3))-(5*rho*b*d1*v^2))/m);
a21=(((k1*d3)-(k2*(d1-d3)))/I);
a22=(((k1*d3^2)+(k2*(d1-d3)^2)-(5*rho*b*d1*v^2*(d3-d2)))/I);
A=((a11*a22)/2)
B=((a11*a22)-(a12*a21))
Asq=(A^2)
if 0>=B
display ('Non-oscillatory Divergence')
end
if B>0
if (A^2)>=B>0
display ('Stable Oscillation')
end
end
if B>(Asq)
display ('Divergent Oscillation (Flutter)')
end
15
Appendix A2 – MATLAB Code for Non-Oscillatory Divergent
d1=13.73; %Enter chord length (m)
d2=4.6; % Enter distance from leading edge to centre of lift
(m)
d3=5.71; %Enter distance from leading edge to centre of mass
(m)
b=5; %Enter width of wing (m)
k1=100000; %Enter spring stiffness coefficient 1 (leading) (Nm)
k2=200000; %Enter spring stiffness coefficient 2 (trailing) (Nm)
v=272.7; %Enter free stream air velocity (m/s)
m=25270; %Enter mass of wing (Kg)
I=((m*(d1^2))/12); % Moment of mass (Kg*m^2)
rho=1.2; % Density of air (Kg/m^3)
Area=(b*d1); % Projected Area (m^2)
a11=(k1+k2)/m;
a12=(((k1*d3)+(k2*(d1-d3))-(5*rho*b*d1*v^2))/m);
a21=(((k1*d3)-(k2*(d1-d3)))/I);
a22=(((k1*d3^2)+(k2*(d1-d3)^2)-(5*rho*b*d1*v^2*(d3-d2)))/I);
A=((a11*a22)/2)
B=((a11*a22)-(a12*a21))
Asq=(A^2)
if 0>=B
display ('Non-oscillatory Divergence')
end
if B>0
if (A^2)>=B>0
display ('Stable Oscillation')
end
end
if B>(Asq)
display ('Divergent Oscillation (Flutter)')
end
16
Appendix A3 – MATLAB Code for Oscillatory Divergent (Flutter)
d1=13.73; %Enter chord length (m)
d2=4.6; % Enter distance from leading edge to centre of lift
(m)
d3=5.71; %Enter distance from leading edge to centre of mass
(m)
b=5; %Enter width of wing (m)
k1=1000000; %Enter spring stiffness coefficient 1 (leading) (Nm)
k2=50000; %Enter spring stiffness coefficient 2 (trailing) (Nm)
v=272.7; %Enter free stream air velocity (m/s)
m=25270; %Enter mass of wing (Kg)
I=((m*(d1^2))/12); % Moment of mass (Kg*m^2)
rho=1.2; % Density of air (Kg/m^3)
Area=(b*d1); % Projected Area (m^2)
a11=(k1+k2)/m;
a12=(((k1*d3)+(k2*(d1-d3))-(5*rho*b*d1*v^2))/m);
a21=(((k1*d3)-(k2*(d1-d3)))/I);
a22=(((k1*d3^2)+(k2*(d1-d3)^2)-(5*rho*b*d1*v^2*(d3-d2)))/I);
A=((a11*a22)/2)
B=((a11*a22)-(a12*a21))
Asq=(A^2)
if 0>=B
display ('Non-oscillatory Divergence')
end
if B>0
if (A^2)>=B>0
display ('Stable Oscillation')
end
end
if B>(Asq)
display ('Divergent Oscillation (Flutter)')
end
17
References
[2] J. Rane, “Final boarding: McDonnell Douglas DC-10 makes last passenger flight,” 6
December 2013. [Online]. Available:
http://edition.cnn.com/2013/12/05/travel/dc-10-last-flight/. [Accessed 5
December 2014].
18