You are on page 1of 79

Lecture notes for EG3401

Introduction to Flight Dynamics


Emmanuel Prempain

2018-2019
Contents
1 Lecture 1: Introduction 5
1.1 Flight dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Airframe Components . . . . . . . . . . . . . . . . . . . . . . 6
1.3 Basic Aerodynamics Definitions . . . . . . . . . . . . . . . . . 8

2 Lecture 2: Simplifying Assumptions 10


2.1 Aircraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2 The Earth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3 Properties of the Atmosphere . . . . . . . . . . . . . . . . . . 11

3 Lecture 3: Frames of Reference 15


3.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Inertial Reference Frame . . . . . . . . . . . . . . . . . . . . . 15
3.3 Vehicle Axis Systems . . . . . . . . . . . . . . . . . . . . . . . 16
3.3.1 Body Axes . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3.2 Stability Axes . . . . . . . . . . . . . . . . . . . . . . . 16
3.3.3 Wind Axes . . . . . . . . . . . . . . . . . . . . . . . . 17
3.4 Linear Velocities . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.5 Some Terminology . . . . . . . . . . . . . . . . . . . . . . . . 18
3.6 Angular Velocity . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.7 Aerodynamic Angles . . . . . . . . . . . . . . . . . . . . . . . 20
3.8 Velocity Components in terms of Incidences . . . . . . . . . . 21

4 Lecture 4: Euler Angles 22


4.1 Euler Angles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.2 Some Examples . . . . . . . . . . . . . . . . . . . . . . . . . . 24

5 Lecture 5: Forces and Moments 26


5.1 Forces and Moments . . . . . . . . . . . . . . . . . . . . . . . 26
5.2 Aerodynamic Forces and Moments . . . . . . . . . . . . . . . 26
5.2.1 Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.2.2 Moments . . . . . . . . . . . . . . . . . . . . . . . . . . 27
5.3 Coefficients Forms . . . . . . . . . . . . . . . . . . . . . . . . . 27
5.4 Drag Coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . 28
5.5 Lift Coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . 29
5.6 Side-Force Coefficient . . . . . . . . . . . . . . . . . . . . . . . 29
5.7 Rolling Moment Coefficient . . . . . . . . . . . . . . . . . . . 29
5.8 Pitching Moment Coefficient . . . . . . . . . . . . . . . . . . . 31
5.9 Yawing Moment Coefficient . . . . . . . . . . . . . . . . . . . 31

2
5.10 Propulsion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

6 Lecture: Static and Dynamic Aircraft Stability 33


6.1 Static Stability . . . . . . . . . . . . . . . . . . . . . . . . . . 33
6.2 Longitudinal Stability Analysis . . . . . . . . . . . . . . . . . 35
6.2.1 Static stability . . . . . . . . . . . . . . . . . . . . . . 35
6.2.2 Pitch Control and Trim . . . . . . . . . . . . . . . . . 35
6.2.3 Pitching Moment Equation . . . . . . . . . . . . . . . . 36
6.3 Dynamic Stability . . . . . . . . . . . . . . . . . . . . . . . . . 38
6.3.1 Longitudinal modes . . . . . . . . . . . . . . . . . . . . 39

7 Lecture 7: The Non-linear Aircraft Model 41


7.1 Motion of a Rigid Body in Three Dimensions . . . . . . . . . . 41
7.2 Relations between derivatives of a vector in different reference
frames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
7.3 Body Axis Equations . . . . . . . . . . . . . . . . . . . . . . . 42
7.3.1 Force Equations . . . . . . . . . . . . . . . . . . . . . . 42
7.3.2 Moment Equations . . . . . . . . . . . . . . . . . . . . 43
7.3.3 Kinematic Equations . . . . . . . . . . . . . . . . . . . 44
7.3.4 Navigation Equations . . . . . . . . . . . . . . . . . . 44
7.4 Linearised Equations of Motion . . . . . . . . . . . . . . . . . 45
7.5 Longitudinal Equations . . . . . . . . . . . . . . . . . . . . . . 48
7.6 Lateral-directional Equations . . . . . . . . . . . . . . . . . . 50

8 Lecture 8: Modal Analysis 52


8.1 Eigenvectors and Eigenvalues . . . . . . . . . . . . . . . . . . 52
8.1.1 Eigenvectors and Eigenvalues Properties . . . . . . . . 52
8.2 Modal Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 53
8.3 Linear Time Invariant (LTI) Systems . . . . . . . . . . . . . . 54
8.4 Transfer Matrix and Frequency Response . . . . . . . . . . . . 58

9 Lecture 9: Longitudinal Dynamics 59


9.1 Stability Axes Equations . . . . . . . . . . . . . . . . . . . . . 59
9.2 Short-Period Approximation . . . . . . . . . . . . . . . . . . . 59
9.3 Long-Period (Phugoid) Approximation . . . . . . . . . . . . . 61
9.4 Control Response . . . . . . . . . . . . . . . . . . . . . . . . . 62
9.5 More on Longitudinal Stability Derivatives . . . . . . . . . . . 65
9.5.1 Derivatives Due to Velocity . . . . . . . . . . . . . . . 66
9.5.2 Derivatives Due to α . . . . . . . . . . . . . . . . . . . 67
9.6 Longitudinal Equations in Different Body Frames . . . . . . . 68
9.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

3
10 Lecture 10: Lateral Dynamics 71
10.1 Lateral-directional Linearised Equations of Motion . . . . . . . 71
10.1.1 Roll Mode . . . . . . . . . . . . . . . . . . . . . . . . . 72
10.1.2 Spiral Mode . . . . . . . . . . . . . . . . . . . . . . . . 73
10.1.3 Dutch-Roll Mode . . . . . . . . . . . . . . . . . . . . . 73
10.2 Lateral Stability Derivatives . . . . . . . . . . . . . . . . . . . 74
10.3 Example: DC-8, Mach=0.84, alt=33000 ft, V=825 ft/s . . . . 75

A Aircraft Mass Moment of Inertia Matrix 78

4
1 Lecture 1: Introduction
1.1 Flight dynamics
Flight dynamics is primarily concerned with stability and dynamic charac-
teristics of aircraft. Flight dynamics is about the study of short term motion
in response to control and atmospheric disturbances where motion varies
from small amplitudes (almost linear behaviour) to large amplitudes (very
nonlinear behaviour). Flight dynamics is required in:

• design of aircraft and autonomous vehicles (e.g. UAVs)

• analysis of existing aircraft (stability and flying qualities)

• design of stability augmentation systems (for better performance)

• autopilot design

• simulation (for pilot training etc.)

In order to be able to characterise aircraft stability and performance one


needs to establish suitable mathematical models. Mathematical models re-
quired to

• describe the aircraft, control (control surfaces, propulsion) and subsys-


tems integration (e.g. autopilot)

• describe the properties of the atmosphere (aerodynamic forces are highly


dependent on air density)

• to predict aircraft behaviour

Aircraft motion is usually described by classical mechanics, that is by the


application of Newton’s laws. Mechanics deals with the motion of objects
that possess mass. Objects may be modelled as individual particles (point
masses) or assemblages of particles called bodies. Unlike point masses bodies
are 3-D. Position and the velocity of a body refer to the position and velocity
of its centre of mass (c.m. or c.g.). Orientation and the angular velocity of
a body are of importance and are given with respect to some appropriate
frames of reference.

5
Figure 1: Airframe Components

1.2 Airframe Components


• The wings provide the aircraft largest aerodynamic force (to support
aircraft weight). Wings are shaped to provide the largest amount of
lift with as little drag as possible.

• The fuselage is the aircraft principal structure containing payload and


systems.

• The vertical tail provides directional stability about the aircraft vertical
axis. Gives the aircraft the tendency to nose into the relative wind that
results from forward motion.

• The horizontal tail plane provides angular stability with respect to an


axis parallel to wing span.

• Tail surfaces are called empennage.

Conventional control effectors are: ailerons, elevator, rudder, flaps and thrust
setting:

• The ailerons are movable surfaces located near the wing tips that pro-
duce large rolling torques. The two surfaces are linked so that the
trailing edge of one moves up when the trailing edge of the other moves
down.

6
• The elevator is a movable surface that extends across the trailing edge
of the horizontal tail for angular control. The elevator controls the
angle of attack of the aircraft and the pitch angle.

• The rudder is a movable surface mounted on the trailing edge of the


vertical tail. It controls yawing motion.

• Flaps are movable surfaces mounted on the trailing edge of the wings.
They control lift and drag during take-off and landing. Left and right
surfaces work together. Flaps are normally adjusted to discrete settings
that depend on the flight phase.

• The engine thrust setting is adjusted occasionally to achieve take-off


acceleration, desired cruising conditions, descent etc.

7
1.3 Basic Aerodynamics Definitions
Aerodynamics forces acting on an airfoil are highly non-linear functions of air-
foil geometry, angle-of-attack, free stream velocity (relative wind), air density
etc. The key concept is that the motion of the air around an airfoil produces
pressure variations which, in turn, result in aerodynamic forces.

Figure 2: Airfoil geometry

• Chord Line: straight line from leading to trailing edges

• Mean Line (also known as camber line): Mid-line between upper and
lower airfoil surfaces

• Angle-of-Attack : angle that chord line makes with free stream velocity

Figure 3: Lift and Drag components

• Lift: component of total aerodynamic force perpendicular to free stream


velocity

• Drag: component of the aerodynamic force parallel to free stream ve-


locity

8
• Centre of Pressure (cp): application point of the total aerodynamic
force acting on the airfoil (so moment of total aerodynamic force about
cp is zero). The centre of pressure varies with the angle-of-attack

9
2 Lecture 2: Simplifying Assumptions
2.1 Aircraft
The following assumptions are made:

1. Constant mass: for a conventional aircraft, the mass variation rate per
minute, dm
dt
, is between 0.02% and 0.2%. So, we can assume dm/dt ≈ 0.

2. Rigid body: a rigid body is an idealised system of particles which


does not undergo any change in size or shape. A rigid body can be
treated as a particle whose mass is that of the body and is concen-
trated at the body centre-of-mass. It allows the airframe motion to be
described completely in terms of a translation of the centre-of-mass and
a rotation about it. With this assumption aero-elastic effects are not
included in the equations. The rigid body assumption is well verified
for small/medium size aircraft.

3. Aircraft symmetry: very well verified for the aerodynamics, propulsion


and mass distribution. The aircraft centre-of-mass position does not
vary significantly and the aircraft mass moments of inertia are almost
constant in a frame attached to the body.

Figure 4: Airbus A380, 600 seats, a ‘relatively flexible’ aircraft

10
2.2 The Earth
1. Flat Earth: the Earth regarded as flat. This assumption is based on
the fact that the ratio between centrifugal acceleration and Earth grav-
itational acceleration (g) given by:

V2
RE g
is small, where RE = 6400 km is Earth radius and V is aircraft velocity.
This ratio is of about 0.6% at Mach = 2.

2. Uniform gravitational field: the Earth gravitational acceleration, g, is


a function of altitude, h, given as:
g0
g(h) =
(1 + h/RE )2

where g0 = 9.81 m/s2 at sea level and where RE = 6400 km is the


Earth radius. To a 1 % decrease in gravity

g(h) − g0 1 1
= − 1 = −
g0 (1 + h/RE )2 100

corresponds an altitude increase of about 32 km. For a passenger air-


craft the ceiling is of about 11km and a fighter can fly up to an altitude
of 26km. So assuming g ≈ g0 is a well verified assumption in practice.

2.3 Properties of the Atmosphere


Each point of the atmosphere is characterised by:

• pressure p [Pa]

• density ρ [kg/m3 ]

• temperature T [K]

The standard atmospheric model assumes a steady atmosphere (i.e. time


invariant) and that air pressure, density and temperature only depend on
height (h). A reasonably good atmospheric model is obtained by assuming
that the atmosphere follows the ideal gas law:

p = ρRT

11
Figure 5: Lockheed SR-71 Blackbird, service ceiling: 85,000 ft (25,900 m)

altitude temperature gradient


h [km] Th [K/m]
0 ≤ h ≤ 11 −6.5 × 10−3
11 ≤ h ≤ 20 0
20 ≤ h ≤ 32 10−3
32 ≤ h ≤ 47 2.8 × 10−3

Table 1: Temperature gradient

where R = 287 [J/kg.K] and p and ρ are both functions of height. Because
the atmosphere has a weight it is subject to the law of Laplace:

dp = −ρgdh

Fact: the temperature gradient, Th , is piecewise constant as shown in Table


1. Thus, the temperature of the atmosphere can be described as an affine
function of altitude:
T ≈ T0 + Th h
dT
where Th is temperature gradient defined as Th = dh
.
At sea level (i.e. h = 0), the standard model assumes that ρ0 = 1.225
kg/m3 , T0 = 288.16 K (15 o C) and po = 101, 325 [Pa].

12
By combining ideal gas and Laplace laws one can explicitly express pres-
sure and air density as functions of altitude:
dp ρg
= − dh (Laplace)
p p
g
= − dh (ideal gas law)
RT
g dT dT
= − (because dh = )
RTh T Th
Integrating both sides, from 0 to h, gives
g
ln(p) − ln(p0 ) = − (ln(T ) − ln(T0 ))
RTh
 − RTg  − RTg
p T h T h
= or p = p0
p0 T0 T0

Air temperature and air pressure as defined are shown in Figure 6. Air den-
sity and pressure always decrease with altitude. We can see that air temper-
ature and speed of sound are piecewise linear functions of altitude. Between
0 and 11km, air temperature decreases by -6.5 degrees every kilometre and
does not vary between 11km and 20km.

13
Figure 6: 1962 U.S. Standard Atmosphere

14
3 Lecture 3: Frames of Reference
3.1 Background
• navigation concerned with position and velocity w.r.t. the Earth

• aircraft performance requires position/velocity w.r.t. the atmosphere

• physical phenomenons are more naturally expressed in coordinate sys-


tems attached to the aircraft (e.g. direction of propulsive force can be
considered as fixed w.r.t. the aircraft body)

Coordinate systems must be right-handed. Orthogonal axes are labelled


x, y, z with appropriate subscript (e.g. ‘W‘ for wind axis). A coordinate
system is defined in terms of:

• location of its origin

• two of its axes (the 3rd axis being deduced by completing the right-
handed system)

3.2 Inertial Reference Frame


An inertial coordinate system is a system in which Newton’s second law is
valid. The equations of motion must, therefore, be determined in an inertial
coordinate system. A reference frame is inertial if

• origin is not accelerated

• axes fixed w.r.t. inertial space

Earth fixed coordinate systems can be considered as inertial. An Earth frame


of reference, FE , can be defined as follows:

• origin can be any arbitrary point at the surface of the Earth

• xE points North

• yE points East

• zE points toward the centre of the Earth

This coordinate system is also known as north-east-down (NED).

15
3.3 Vehicle Axis Systems
These coordinate systems have their origins fixed to the vehicle. Three are
commonly used for describing aircraft motion: the body axis system, the
stability axis system and the wind axis system.

3.3.1 Body Axes


This coordinate system is fixed to the airframe and so aircraft mass moments
and products of inertia are constant. The body axis system is referred to as
the xyz system and it is defined as follows:
• origin located at the vehicle centre of mass (c.m. or c.g)
• x is forward
• y is lateral (toward the right wing tip)
• z is downward (normal to the x-y plane)
• axes are fixed (i.e. attached to the airframe)
For convenience axes x and z may be rotated about the y axis (e.g. x
aligned with velocity vector in airframe stability axes). The axes x and z
define the airframe plane of symmetry. (i, j, k) are the unit vectors along
the respective aircraft body axes. The body axis system is also the natural
frame of reference for most on-board measurements of the vehicle motion.

3.3.2 Stability Axes


These are specialised body axes (i.e. attached to the body) determined by
a given equilibrium flight condition. The xs -axis is aligned with the relative
wind at the start of the motion. The stability axis system is defined as
follows:
• origin located at the vehicle centre of mass (c.m.)
• xs is along projection of velocity vector into the plane of symmetry
• ys perpendicular to both xs and zs and off to the right (same as body
y-axis)
• zs perpendicular to xs , pointed down
The plane xs − zs remains in the aircraft plane of symmetry. In this sys-
tem, mass moments and products of inertia are constant in the equations of
motion. However, these will depend on each equilibrium flight condition.

16
y

c.g.
+
i +

+
k
z

Figure 7: Body axes

3.3.3 Wind Axes


The wind axis coordinate system uses the aircraft velocity (relative to the
air mass) as the reference. It is defined as follows:

• origin located at the vehicle centre of mass (c.m.)

• xW is along velocity vector relative to the air mass

• yW is perpendicular to both xW and zW and off to the right

• zW perpendicular to xW , pointed down into the plane of symmetry

Aerodynamic forces and moments can conveniently be expressed in the


wind axis system. Especially, the total aerodynamic force can be expressed
in terms of total drag along −xW and total lift along −zW . However, because
the orientation of the wind axis system varies with wind orientation (and so
do mass moments and products of inertia) wind axes are not normally used
in the analysis of the motion.

17
y
j
xw
V

c.g.
+
i +

+
k
zw z

Figure 8: Wind and body axes

3.4 Linear Velocities


The velocity of the airframe centre of mass w.r.t. a frame attached to the
Earth, FE is the vector quantity VE . It can be considered as absolute (i.e.
inertial) velocity of the airframe centre of mass. The velocity of the airframe
centre of mass w.r.t. the atmosphere is a vector quantity V with components
u, v, and w in the body axis system:

V = ui + vj + wk

If W denotes the wind velocity (relative to FE ) then

VE = V + W

often we shall assume W = 0 making airspeed the same as inertial velocity

3.5 Some Terminology


• VE is called ground-speed (absolute velocity w.r.t. the Earth)

• V is called airspeed (aircraft velocity with respect to air mass)

• u is called forward (or axial) velocity and along vehicle centreline

• v is called lateral velocity

18
• w is called normal velocity (perpendicular to vehicle centreline)

• V = |V| = u2 + v 2 + w2 represents the magnitude of the airspeed

• q = 21 ρV 2 is dynamic pressure
V
• a
is Mach number where a is speed of sound

Airspeed magnitude can be estimated based on the difference between total


and static pressures (sensed by pitot tubes and flush sensors).

v
w v j
w
u c.g.
i
k
p+ +
q
+
r

Figure 9: Airframe velocities

3.6 Angular Velocity


The angular velocity of the airframe is the vector quantity ω with components
p, q, and r in the body axis system

ω = pi + qj + rk

• p is roll rate

• q is pitch rate

• r is yaw rate
The angular rates can be sensed by rate gyros. Linear and rotational veloci-
ties constitute the 6 degree-of-freedom of the aircraft.

19
3.7 Aerodynamic Angles
The orientation of the velocity vector V w.r.t. body-axes yields the definition
of angle-of-attack and side-slip angle (also known as aerodynamic angles or
incidences). Side-slip and angle-of-attack are central to the definition of
aerodynamic forces and moments. The angle-of-attack α is defined as

• The angle between velocity vector projected into the plane of symmetry
and the x-axis

• Positive if relative wind comes from below the x-axis

Figure 10: Angle-of-Attack (AoA)

From Figure 10 we can see that tan α = w/u and so

α = tan−1 (w/u)

The side-slip angle β is defined as:

• Angle between vehicle velocity and plane of symmetry

• Positive if relative wind comes from the right of the plane of symmetry

From Figure 11 we have sin β = v/V and so

β = sin−1 (v/V )

20
Figure 11: Side-slip angle (AoS)

3.8 Velocity Components in terms of Incidences

u = V cos β cos α
v = V sin β
w = V cos β sin α

Figure 12: Aircraft axes and angles

21
4 Lecture 4: Euler Angles
4.1 Euler Angles
One way to describe the rotation (the change in orientation) of a rigid body
is by means of Euler angles. Any arbitrarily oriented reference frame may
be placed in alignment with any other reference frame by three successive
rotations whose order is important. The order of selection of the axes in these
rotations is arbitrary, but the same axis may not be used twice in succession.
The rotation sequences are usually denoted by three numbers, 1 for x, 2 for
y, and 3 for z. The are twelve valid sequences which are 123, 121, 131, 132,
213, 212, 231, 232, 312, 313, 321, and 323.
The orientation (attitude) of an aircraft with respect to fixed axes is given
in terms of the rotation sequence 321, or z-y-x as shown in Figure 13.

Figure 13: Euler angles

The sequence starts by rotating the body fixed coordinate system (x,y,z),
which is initially aligned with the space fixed coordinate system (X,Y ,Z), by
an angle ψ about the Z-axis. In a second step the intermediate coordinate
system is rotated about the new y axis by an angle θ to produce yet another
intermediate coordinate system. Finally this coordinate system is rotated
about the new x axis by an angle φ to produce the body fixed coordinate
system. The entire procedure is summarized as follows:

1. A rotation ψ about Z gives x0 , y 0 , z 0 = Z

22
2. A rotation θ about y 0 gives x00 , y 00 = y 0 , z 00

3. A rotation φ about x00 gives x, y, z


These rotations can be written in terms of matrices as
 
cos ψ sin ψ 0
Tz (ψ) =  − sin ψ cos ψ 0 
0 0 1
 
cos θ 0 − sin θ
Ty (θ) =  0 1 0 
sin θ 0 cos θ
 
1 0 0
Tx (φ) =  0 cos φ sin φ 
0 − sin φ cos φ

Hence,

TBE = Tx (φ)Ty (θ)Tz (ψ)


is the transformation matrix from Earth (xE -yE -zE ) to body-axis coordinates
(x-y-z). Straightforward matrix multiplications yield
 
cos θ cos ψ cos θ sin ψ − sin θ
TBE =  sin φ sin θ cos ψ − cos φ sin ψ sin φ sin θ sin ψ + cos φ cos ψ sin φ cos θ 
cos φ sin θ cos ψ + sin φ sin ψ cos φ sin θ sin ψ − sin φ cos ψ cos φ cos θ

The inverse transformation, TEB , from body to Earth axes, is just obtained
T
by transposing TBE , that is TEB = TBE where T denote matrix transposition.
Actually, it corresponds to the reverse of the sequence of rotations given
above, that is
TEB = Tz (−ψ)Ty (−θ)Tx (−φ)
and it is given by:
 
cos θ cos ψ sin φ sin θ cos ψ − cos φ sin ψ cos φ sin θ cos ψ + sin φ sin ψ
TEB =  cos θ sin ψ sin φ sin θ sin ψ + cos φ cos ψ cos φ sin θ sin ψ − sin φ cos ψ 
− sin θ sin φ cos θ cos φ cos θ

To avoid ambiguities and singularities Euler angles are limited to:

0 ≤ ψ ≤ 2π
−π/2 < θ < π/2
0 ≤ φ ≤ 2π

23
Matrices Tx (φ), Ty (θ), Tz (ψ), TEB , TBE are orthogonal. A matrix U is or-
thogonal if it is square and satisfy U U T = U T U = I, that is, U −1 = U T .
Orthogonal transformations preserve norm of vectors and angles between
vectors.

4.2 Some Examples


Example: Weight in Body Axes: Let us find the weight components of an
aircraft in its body-axis reference frame. The aircraft weight in the Earth
reference coordinate system is given by W = mgK where K is the unit vector
along Z (directed towards the centre of the Earth). Thus, the components
of the aircraft weight in body axes are:
   
0 − sin θ
TBE  0  = mg  cos θ sin φ 
mg cos θ cos φ

Example: Angular Velocity: Let us express the aircraft angular velocity in


terms of Euler angular rates. The aircraft angular velocity is given by

Ω = ψ̇K + θ̇j0 + φ̇i00

where K is unit vector along Earth-fixed axis zE , j0 is unit vector along


intermediate axis y 0 and i00 = i is unit vector along the body x-axis.
Now, the components of unit vector K, expressed in body axes, are (i.e.
the last column of TBE ):  
− sin θ
 sin φ cos θ 
cos φ cos θ
The components of unit vector j0 in body axes are given by:
   
0 0
Tφ  1  =  cos φ 
0 − sin φ

Finally, because Ω = pi + qj + rk, we get:

p = φ̇ − ψ̇ sin θ
q = θ̇ cos φ + ψ̇ cos θ sin φ
r = −θ̇ sin φ + ψ̇ cos θ cos φ

24
or in matrix-vector form:
   
p φ̇
 q  = R  θ̇ 
r ψ̇

where
 
1 0 − sin θ
R =  0 cos φ sin φ cos θ 
0 − sin φ cos φ cos θ

Inverting R we get the Euler angle rates as:


   
φ̇ p
 θ̇  = R−1  q 
ψ̇ r

where
 
1 sin φ tan θ cos φ tan θ
R−1 = 0 cos φ − sin φ 
0 − sin φ sec θ cos φ sec θ

Note det R = cos θ and so R−1 exists if −π/2 < θ < π/2. It should be noted
that R is not orthogonal.

25
5 Lecture 5: Forces and Moments
5.1 Forces and Moments
• A cause capable of modifying the velocity of a body is called a force
• The effect of the force on the rigid body depends on its point of ap-
plication A. The position of the application point with respect to an
−→
arbitrary origin O is given by the position vector r = OA. The moment
of force F about O is then defined as the cross product of r and F
M=r×F
Thus, the moment of a force is a vector perpendicular to the plane
containing r and F
For an aircraft, forces are due to aerodynamic effects, engine thrust and
weight. The three most important variables determining the aerodynamic
forces (and their moments about c.m) are the angle of attack, the angle of
side-slip and the true airspeed.

5.2 Aerodynamic Forces and Moments


5.2.1 Forces
The aerodynamic forces acting on an aircraft are produced by the relative
motion of the air and thus depend on the orientation of the aircraft with
respect to the airflow.
The total aerodynamic force acting on an aircraft is denoted Faero . Its
components are denoted by the upper-case letters X, Y and Z in the body
axes system. The aerodynamic forces are usually defined in terms of dimen-
sionless coefficients. Aerodynamic force components are made non-dimensional
by dividing them by dynamic pressure q̄ = 21 ρV 2 and a characteristic area S
(usually the wing area) so that total aerodynamic force (in body axes) is:
   
X CX
Faero =  Y  = q̄S  CY 
Z CZ
The dimensionless force coefficients can also be specified for wind axes as:
drag, D = q̄SCD
lift, L = q̄SCL
side-force, Y = q̄SCy

26
5.2.2 Moments
Moments are computed with respect to the aircraft centre of mass. The total
aerodynamic moment components are denoted by the upper-case letters L,
M and N . L is called rolling moment, M is pitching moment and N is yawing
moment. Aerodynamic moment components are made non-dimensional by
dividing them by dynamic pressure q̄ = 21 ρV 2 , the reference area, S, and the
wing span, b, for rolling and yawing moments and the wing mean geometric
cord, c̄, for the pitching moment:
   
L bCl
Maero =  M  = q̄S  c̄CM 
N bCN

The moment coefficients have been specified for body axes. However, the
same letters are used if the are specified in the wind axes.

Figure 14: Sketch of wing planform

5.3 Coefficients Forms


The dimensionless coefficients CD , CL etc. depend primarily on α, β and
airspeed V . They also dependent on the control surface deflections. Each
coefficient is modelled as a sum of functions of fewer variables. The total
aerodynamic coefficients are expressed as a baseline component plus incre-
mental correction terms indicated by the symbol ∆. The baseline component
is primarily a function of alpha, beta and Mach number (M ). Data for the

27
coefficients are derived from wind tunnel, flight tests, computer programs
and are compiled in tabular form. For conventional aircraft, are typically of
the form shown in Table 2.

Table 2: Aerodynamic Coefficient Components

CD = CD (CL ) + ∆CD (el) + ∆CD (β) + ∆CD (M ) + ...


CL = CL (α, Tc ) + ∆CL (el) + ∆CL (M ) + ...
CY = CY (β) + ∆CY (rdr)+...
Cl = Cl (β) + ∆Cl (ail) + ∆Cl (rdr)+...
CM = CM (CL , TC ) + ∆CM (el) + δCM (M )

5.4 Drag Coefficient


In the equation for total drag, CD is usually expressed as a function of lift
coefficient and it is referred to as drag polar. The drag polar can be fitted
with a parabolic equation of the form

CD (CL ) = k(CL − CLDM )2 + CDM

where CDM is minimum drag which can occur at a non-zero value of lift.
Drag tends to be independent of Mach number at low Mach.

Figure 15: Drag polar versus lift coefficient for 3 values of Mach number.

28
5.5 Lift Coefficient
The total lift coefficient has a baseline component CL (α, TC ) where TC is
normalize engine thrust:
thrust
TC =
q̄SD
where SD is characteristic area (disc swept out by a propeller blade). Typical
lift curves are shown in Figure 16.

Figure 16: Lift curves versus angle of attack for three normalized values of
thrust.

5.6 Side-Force Coefficient


A typical variation of side force as a function of side slip is shown in Figure
17. Side-force coefficient is negative when side-slip is positive.

5.7 Rolling Moment Coefficient


The baseline term in the rolling moment equation is Cl (β). A typical varia-
tion of Cl (β) is shown in Figure 18. The other terms in the rolling moment
equation represent the components due to aileron deflection and rudder de-
flection. A positive side-slip creates a negative rolling moment. This effect
is responsible for giving an aircraft a tendency to fly with wings level.

29
Figure 17: Side-force versus sideslip.

Figure 18: Rolling moment versus sideslip.

30
5.8 Pitching Moment Coefficient
The pitching moment coefficient depends on angle-of-attack because the net
aerodynamic forces in the plane of symmetry do not generally act directly
through the aircraft centre of gravity. A change in angle-of-attack on the hor-
izontal tail provides moment that tends to counteract changes in the aircraft
angle-of-attack (stability in pitch). Thus, in a typical pitching moment plot
(versus α) the slop is negative indicating that positive changes in α result in
negative changes (nose down) in CM (as seen in Figure 19). The elevator is
the primary control for the pitching moment.

Figure 19: Pitching moment versus angle-of-attack.

5.9 Yawing Moment Coefficient


A typical plot of yawing moment is shown in Figure 20. Vertical tail causes
the aircraft to weathercock into the wind, that is, positive sideslip produces
positive yawing moment. The rudder part of the trailing edge of the vertical
tail is the primary control for yawing moment.

31
Figure 20: Typical Yawing moment versus angle of sideslip.

5.10 Propulsion
A simple thrust model is
T = kT ρV λT δT
where

• T is engine thrust

• V is airspeed

• δT is thrust lever control

• kT and λT are parameters that depend on the type of engine.

Typically,

• λT ≈ −1 for propeller

• λT ∈ [0, 1] for jet engine

32
6 Lecture: Static and Dynamic Aircraft Sta-
bility
6.1 Static Stability
• Equilibrium: equilibrium is realised if the vector sum of all the external
forces is zero together with the vector sum of the moments about the
aircraft’s cg is zero. E.g. straight and level flight is realised if lift equals
weight, thrust equals drag and net moment is zero.

• Statically stable aircraft: assume that, because of a disturbance (e.g.


wind gust), α is increased then if the aircraft tends to come back to
its initial equilibrium condition then the aircraft is said to be statically
stable.

• Statically unstable aircraft: assume again that α is increased then if


the aircraft tendency is to nose up even further then it is said statically
unstable (i.e. the aircraft motion diverges from equilibrium).

• Neutral static stability: if the aircraft stays within the disturbed posi-
tion then the aircraft is said to have neutral static stability.

Figure 21: Static stability

Example: Longitudinal equilibrium

Consider the case of Figure 22 (a). The forces acting on the aircraft are:

33
• Aircraft weight (W = mgK) acting through cg.
• Lift, L, and drag, D, acting at the aircraft aerodynamic centre (ac)
which is normally very close to the aerodynamic centre of the wing.
• Thrust, T, along thrust line (which can lie above or below cg)
• Horizontal tail lift and drag.
• From Figure 22 (a) we can see that L and T produce nose-down mo-
ments while D produces a nose-up moment.

Figure 22: Pitch equilibrium

To achieve equilibrium, the net moment must be zero. This is achieved with
the horizontal tail.
• Horizontal tail (a small wing) generates lift which can be adjusted with
elevator control.
• Because of the relatively large distance from aircraft’s cg to the hori-
zontal tail’s ac only a small tail force is required to supply the balancing
moment, Figure 22 (b).
• Elevator is ‘trimmed’ to a particular angle to fly in a particular equilib-
rium condition. In other words, equilibrium corresponds to particular
angle of attack α = αtrim for which net pitching moment about the
aircraft’s cg is zero CMcg = 0.

34
Figure 23: Horizontal tail and elevator

6.2 Longitudinal Stability Analysis


6.2.1 Static stability
The conditions for equilibrium and static stability are:

CMcg = 0 (at α = αtrim )


∂CMcg
≡ CM α < 0
∂α
If α is not too large one can assume that lift is a linear function of α, i.e.
CL = CLα (α − α0 ) with CLα > 0, then an equivalent condition for static
stability is
∂CMcg
<0
∂CL
For a statically stable aircraft:

• Positive moment values produce nose up motions for angles of attack


lower than αtrim .

• Negative values of the pitching moment produce nose down motions


for angles of attack higher than αtrim .

A statically stable aircraft is said to have positive pitch stiffness.

6.2.2 Pitch Control and Trim


Lift can approximately be written as a linear function of angle of attack and
elevator deflection δe :

CL = CLα (α − α0 ) + CLδe δe

Similarly, pitching moment can be written as:

CMcg = CM0 + CMα α + CMδe δe

35
Figure 24: Pitching moment values vs α

For trim (α = αtrim ) we have:

CMcg = 0
CL = CLtrim

so
    
CMα CMδ e αtrim −CM0
=
CLα CLδe δetrim CLtrim + CLα α0

thus
CLα CM0 + CMα (CLtrim + CLα α0 )
δetrim =
CMα CLδe − CMδe CLα
For level flight CLtrim = W/(1/2ρV 2 S) where V is trim velocity and W = mg
is aircraft weight. So elevator trim δetrim is expected to change with speed.

6.2.3 Pitching Moment Equation


Consider Figure 25 where

• R(xR , 0, zR ) is reference position used to measure the aerodynamic air-


craft data

• M is pitching moment about reference point

• L is net aircraft lift

36
L D

M 
x cg
R

VT
z

Figure 25: Pitch stiffness

• D is net aircraft drag


Inspection of Figure 25 yields

Mcg = M + (L cos α + D sin α)xR + (L sin α − D cos α)zR

where Mcg is the pitching moment about cg. Divide the equation by 0.5ρV 2 Sc̄
to obtain the non-dimensional pitching moment equation
xR zR
CMcg = CM + (CL cos α + CD sin α) + (CL sin α − CD cos α)
c̄ c̄
Under steady level flight condition we can assume that angle of attack is small
and also that CD  CL . Also if we assume zR small we get the approximate
pitching moment equation:
xR
CMcg = CM + CL

Now to determine whether the aircraft has positive pitch stiffness we differ-
entiate the approximate moment equation w.r.t. α:
dCMcg dCM xR dCL
= +
dα dα c̄ dα
dC Mcg
If dα < 0 then the aircraft has positive pitch stiffness. The first term
on the right-hand side is the slope of the pitching moment of the complete
aircraft measured at the aerodynamic data reference point (it can be designed

37
to be negative in order to produce positive pitch stiffness). The second term
dCL

is known to be positive for small values of angle of attack. Therefore,
depending on xR /c̄ (i.e. the aircraft’s cg location) the aircraft may have
positive or negative pitch stiffness. We conclude that:

• If cg is moved rearward (that is if xR is increased) then there is a po-


sition (xnp ) where pitch stiffness becomes zero. This position is known
as neutral point (actually the aerodynamic centre of the entire aircraft,
point C in Figure 26) where the moment curve becomes horizontal.

• The distance of neutral point behind cg divided by c̄ (wing mean chord)


is called static margin, ms := xcg −x

np
. Conventional balanced aircrafts
are designed to have a minimum static margin of about 5%.

• If the centre of gravity is moved further back (point D) the moment


curve has positive slope and the aircraft is statically longitudinally
unstable.

Figure 26: Static stability vs cg location

6.3 Dynamic Stability


Consider a statically stable aircraft and its response to a disturbance in angle
of attack.

38
• If aircraft returns to its equilibrium condition in a decaying and oscil-
latory fashion then it is said dynamically stable.

• If aircraft noses up and down at constant amplitude then it is said to


possess neutral dynamic stability.

• If aircraft noses up and down with increasing magnitude then it is


dynamically unstable. Statically and a dynamically stable aircraft can
be flown hands off.

Figure 27: Dynamic stability

• Static stability necessary but not sufficient for dynamic stability.

6.3.1 Longitudinal modes


There are two forms of oscillatory longitudinal modes:

• The “phugoid” mode which is a long period, poorly damped oscillation


of the aircraft flight path.

• The “short-period” is fast oscillation in α as shown in Figure 29. This


oscillation usually vanishes quickly with no pilot effort. Short-period
oscillations also occur if the elevators are left free.

39
Figure 28: Stable phugoid

Figure 29: Stable short-period

40
7 Lecture 7: The Non-linear Aircraft Model
7.1 Motion of a Rigid Body in Three Dimensions
The three dimensional motion of a rigid body is governed by the fundamental
equations

ΣF = ma
dH
ΣM =
dt
where:
• m is the mass of the body

• a is the absolute acceleration of the body centre of mass

• ΣF represents the sum of all external forces acting on the body

• H is the body angular momentum about its centre of mass and defined
as Z
H= r × vdm
body

where r is position vector of element of mass dm w.r.t. centre of mass


and v is absolute velocity of element of mass dm.

• ΣM is the sum of the moments of the external forces about centre of


mass
These fundamental equations will preferably be written in state-space
form:
d(mV)
ΣF =
dt
dH
ΣM =
dt
where V is inertial (absolute) velocity of the body centre of mass.

7.2 Relations between derivatives of a vector in differ-


ent reference frames
We recall a fundamental result which will be used in the derivation of Euler’s
equations of motion for a rigid body. Consider the vector r in the coordinate
system xyz. The coordinate system xyz is rotating with angular velocity ω

41
with respect to the fixed (inertial) coordinate system XY Z. In terms of the
rotating coordinate system xyz we have that
r = rx i + ry j + rz k
Therefore
dr d(rx i) d(ry j) d(rz k)
= + +
dt dt dt dt
di dj dk
= ṙx i + rx + ṙy j + ry + ṙz k + rz
dt dt dt
Now, define  
dr
= ṙx i + ṙy j + ṙz k
dt xyz
as the rate of change of r as observed in the rotating reference frame xyz.
Then recalling that
di
= ω×i
dt
dj
= ω×j
dt
dk
= ω×k
dt
so we get
 
dr dr
= + rx (ω × i) + ry (ω × j) + rz (ω × k)
dt dt xyz
 
dr
= + ω × (rx i + ry j + rz k)
dt xyz
 
dr
= +ω×r
dt xyz

The above equation states that the time rate of change of a vector in a
rotating frame of reference, measured from another fixed reference frame, is
equal to the rate of change of the vector within the rotating frame plus the
rate of change of the vector due to the rotation of the rotating frame.

7.3 Body Axis Equations


7.3.1 Force Equations

Fx
ΣF =  Fy  = Faero + W + T
Fz

42
where Faero are the aerodynamic forces, W is aircraft weight and T is thrust.
In body axes, these forces are:
     
X − sin θ TX
ΣF =  Y  + mg  cos θ sin φ  +  0 
Z cos θ cos φ TZ
The absolute velocity of the aircraft centre of mass, expressed in body
axes, is V = ui + vj + wk. Thus, using the result given in Section 7.2, the
absolute acceleration of the aircraft’s cg is
 
dV
+ω×V
dt xyz
where ω = pi + qj + rk. Thus, the absolute acceleration of the aircraft centre
of mass, expressed in body axes, is
     
ax u̇ qw − rv
 ay  =  v̇  +  ru − pw 
az ẇ pv − qu
and so the body axis force equations can be written as:
1
u̇ = (X + TX ) − g sin θ − qw + rv
m
1
v̇ = Y + g cos θ sin φ − ru + pw
m
1
ẇ = (Z + TZ ) + g cos θ cos φ − pv + qu
m

7.3.2 Moment Equations


 
Mx
ΣM =  My  = Maero + MT hrust
Mz
or  
L
ΣM =  M + MT 
N
Now, it can be shown that H = IB ω where IB is the aircraft inertia matrix
in body axes. IB is constant in body axes and thus, using the result of the
previous section (section 7.2), we get
 
d(IB ω) d(IB ω)
Ḣ = = + ω × (IB ω)
dt dt xyz

43
Therefore,  
ω̇ = IB−1 Ḣ − ω × (IB ω)

Also, because the aircraft has a plane of symmetry (i.e. the xz plane) then
the product of inertia involving y are zero and thus
 
Ixx 0 −Ixz
IB =  0 Iyy 0 
−Izx 0 Izz

with inverse
 
Izz 0 Ixz
1
IB−1 = 2
2
 0 Iyy (Ixx Izz − Ixz ) 0 
Ixx Izz − Ixz
Izx 0 Ixx

Thus
1
ṗ = 2
(Izz [L + Ixz pq − (Izz − Iyy )qr] + Izx [N − Izx qr − (Iyy − Ixx )pq])
Ixx Izz − Ixz
1
M + MT − (Ixx − Izz )pr − Ixz (p2 − r2 )

q̇ =
Iyy
1
ṙ = 2
(Ixz [L + Ixz pq − (Izz − Iyy )qr] + Ixx [N − Izx qr − (Iyy − Ixx )pq])
Ixx Izz − Ixz

7.3.3 Kinematic Equations


Kinematic equations relate angular and Euler rates. They were derived in
Section 4.2:

φ̇ = p + tan(θ)(q sin φ + r cos φ)


θ̇ = q cos φ − r sin φ
q sin φ + r cos φ
ψ̇ =
cos θ

7.3.4 Navigation Equations


The position of the aircraft relative to the Earth is found by integrating the
aircraft velocity in Earth-fixed coordinates:
   
ẋE u
 ẏE  = TEB  v 
żE w

44
where TEB is defined in Section 4.1 as
 
cos θ cos ψ sin φ sin θ cos ψ − cos φ sin ψ cos φ sin θ cos ψ + sin φ sin ψ
TEB =  cos θ sin ψ sin φ sin θ sin ψ + cos φ cos ψ cos φ sin θ sin ψ − sin φ cos ψ 
− sin θ sin φ cos θ cos φ cos θ
so

ẋE = u cos θ cos ψ + v(sin φ sin θ cos ψ − cos φ sin ψ) +


w(cos φ sin θ cos ψ + sin φ sin ψ)
ẏE = u cos θ sin ψ + v(sin φ sin θ sin ψ + cos φ cos ψ) +
w(cos φ sin θ sin ψ − sin φ cos ψ)
żE = −u sin θ + v sin φ cos θ + w cos φ cos θ

The set of forces, moments, kinematics and navigation equations form a


system of 12 coupled, first order, non-linear differential equations. These
equations are very complex and solutions to these equations require the use
of numerical integration techniques (Matlab/Simulink). However, simpler
equations can be derived using linearisation techniques. The key idea is to
obtain linear approximate models at a given flight condition.

7.4 Linearised Equations of Motion


Linearised models are obtained w.r.t equilibrium (i.e. steady flight condi-
tions). Linearised models permit the analysis of the small amplitude airframe
motion about equilibrium (i.e. trim or steady flight conditions). Consider
the force equations. The forces acting on the airframe can be written as

Fx = X0 + ∆X − mg sin(θ)
Fy = Y0 + ∆Y + mg cos(θ) sin(φ)
Fz = Z0 + ∆Z + mg cos(θ) cos(φ)

where

• X0 is the equilibrium value of the axial forces

• ∆X is the term due to airframe attitude deviation from trim

• mg terms are the body-weight components

• θ, φ airframe attitudes

45
Similarly, the moments acting on the airframe and computed w.r.t. its centre
of mass can be written as
Mx = L0 + ∆L
My = M0 + ∆M
Mz = N0 + ∆N
• L0 is the nominal rolling moment value
• ∆L is the rolling moment term due to airframe attitude deviation from
trim
Thus, the 6 coupled force and moment equations can be written as
X0 + ∆X − mg sin(θ) = m(u̇ + qw − rv)
Y0 + ∆Y + mg cos(θ) sin(φ) = m(v̇ + ru − pw
Z0 + ∆Z + mg cos(θ) cos(φ) = m(ẇ + pv − qu)
L0 + ∆L = Ix ṗ − Ixz ṙ + (Iz − Iy )qr − Ixz pq
M0 + ∆M = Iy q̇ + (Ix − Iz)pr + Ixz (p2 − r2 )
N0 + ∆N = Iz ṙ − Ixz ṗ + (Iy − Ix )qp + Ixz qr
where
u = u0 + ū, p = p0 + p̄
v = v0 + v̄, q = q0 + q̄
w = w0 + w̄, r = r0 + r̄
θ = θ0 + θ̄, φ = φ0 + φ̄
with 0 subscript refers to the given equilibrium point and the bar over the
variables (such as q̄) refers to a small deviation about equilibrium. Now
assume that the fixed body axes are aligned (before disturbance) with the
airframe velocity vector then wings-level flight condition implies that v0 =
w0 = p0 = q0 = r0 = 0, u0 = V0 and φ0 = 0. Thus static equilibrium
conditions are
X0 − mg sin(θ0 ) = 0
Y0 = 0
Z0 + mg cos(θ0 ) = 0
L0 = M0 = N0 = 0
Further approximations can be made with the trigonometric terms:
sin θ = sin(θ0 + θ̄) = sin θ0 cos θ̄ + cos θ0 sin θ̄

46
and because perturbations are assumed to be small i.e. |θ̄| << 1 then sin θ̄ =
θ̄ and cos θ̄ = 1 and so

sin θ = sin(θ0 + θ̄) = sin θ0 cos θ̄ + cos θ0 sin θ̄


= sin θ0 + θ̄ cos θ0

Similarly, because |φ̄| << 1 and φ0 = 0 we get

cos θ ≈ cos θ0 − θ̄ sin θ0


sin φ ≈ φ, cos φ ≈ 1

So the linearised X force equation is

X0 + ∆X − mg sin(θ) = m(u̇ + qw − rv)


=⇒ ∆X − mg θ̄ cos θ0 = m(u̇ + qw − rv)

Neglecting product terms such as qw, rv and pv (because they are small)
yields the three linearised force equations:

∆X − θ̄(mg cos θ0 ) = mu̇ = mV0 (u̇/V0 )


∆Y + mg cos(θ0 )φ̄ = m(v̇ + r̄V0 ) = mV0 (β̇ + r̄)
∆Z − mg θ̄ sin θ0 = m(ẇ − q̄V0 ) = mV0 (α̇ − q̄)

where α(t) ≈ w̄(t)/V0 is perturbed angle of attack and β(t) ≈ v̄(t)/V0 is


perturbed side slip angle.
Finally, the linearised airframe equations of motion are:

∆X − θ(mg cos θ0 ) = mu̇ = mV0 (u̇/V0 )


∆Y + mgφ cos(θ0 ) = m(v̇ + rV0 ) = mV0 (β̇ + r)
∆Z − mgθ sin θ0 = m(ẇ − qV0 ) = mV0 (α̇ − q)
∆L = Ix ṗ − Ixz ṙ
∆M = Iy q̇
∆N = Iz ṙ − Ixz ṗ
θ̇ = q
φ̇ = p
ψ̇ = r

47
where the bar notation has been dropped for convenience (however it must be
understood that, in the equations given above, u now represents a variation
about the nominal velocity u0 , v a variation about v0 etc.) It should be noted
that the above set of equations form a model linear in the state variables:

u/V0 , α, β, p, q, r, φ, θ, ψ

The aerodynamic force and moment terms (e.g. ∆X etc.) will have to be
linearised using Taylor series expansion.

7.5 Longitudinal Equations


The longitudinal motion is obtained when all the forces acting on the air-
craft belong to the aircraft plane of symmetry and when the moments are
perpendicular to it. Longitudinal motion takes place in the vertical plane
and is thus given by the axial force X, the normal force Z and the pitching
moment M equations only. In addition we have:

• zero side-slip (β = 0) and zero roll (φ = 0)

• p=r=0

Therefore, the longitudinal states are: u/V, α, q and θ. In the sequel, we as-
sume that the control is elevator deflection δe w.r.t. to elevator trim position.
A first order Taylor expansion of the the axial force increment, ∆X, can be
expressed in terms of the state variables, their derivatives, and control input
as:
∆X/m = ∆Xu (u/V ) + ∆Xα α + ∆Xq q + ∆Xδe δe
where    
∂∆X ∂∆X
∆Xu = /m, ∆Xq = /m etc.
∂u 0 ∂q 0
are the partial derivatives of the axial force increment ∆X evaluated at
trim (called stability derivatives). Similarly, we can express normal force
increment as

∆Z/m = ∆Zu (u/V ) + ∆Zα α + ∆Zα̇ α̇ + ∆Zq q + ∆Zδe δe

Finally, the Taylor expansion of the pitching moment is

∆M/Iy = ∆Mu (u/V ) + ∆Mα α + ∆Mα̇ α̇ + ∆Mq q + ∆Mδe δe

48
Using the previous first order Taylor expansions, the linearised longitu-
dinal equations are:

V (u̇/V ) = V Xu (u/V ) + Xα α − g cos(θ0 )θ + Xδe δe


(V − Zα̇ )α̇ = V Zu (u/V ) + Zα α + (V + Zq )q − g sin(θ0 )θ + Zδe δe
−Mα̇ α̇ + q̇ = V Mu (u/V ) + Mα α + Mq q + Mδe δe
θ̇ = q

with state vector xlong = [u/V, α, q, θ]T and control input δe . This system
can be written in descriptor form as:

Elong ẋlong = AElong xlong + BElong δe

with
 
V 0 0 0
 0 V − Zα̇ 0 0
Elong = 
 0 −Mα̇ 1 0 ,

0 0 0 1
 
V Xu Xα 0 −g cos θ0
 V Zu Zα V + Zq −g sin θ0 
AElong = 
V Mu Mα
,
Mq 0 
0 0 1 0
 T
BElong = Xδe Zδe Mδe 0

Because Elong is invertible, we get the standard LTI state space model

ẋlong = Along xlong + Blong δe


−1 −1
with Along = Elong AElong and Blong = Elong BElong
where

 
Xu Xα /V 0 −(g/V ) cos θ0
 V Zu /(∗) Zα /(∗) (V + Zq )/(∗) −g sin θ0 /(∗) 
Along = 
V Mu + Mα̇ V Zu /(∗) Mα + Mα̇ Zα /(∗) Mq + Mα̇ (V + Zq )/(∗)
,
0 
0 0 1 0
 T
Blong = Xδe /V Zδe /(∗) Mδe + (Mα̇ Zδ )/(∗) 0

where (∗) = V − Zα̇

49
7.6 Lateral-directional Equations
The partial derivatives used in Taylor series expansions of ∆Y , ∆L and ∆N
will include
v dφ dψ
β= , p= , r=
V dt dt
The lateral-directional states are β, p, r and φ
In the sequel, we shall assume that lateral control input is the vector
δ = [δa , δr ] whereδa is aileron and δr is rudder control w.r.t. trim positions. As
done with the longitudinal equations, we can express the side force increment,
∆Y , as a first order Taylor expansion in terms of the lateral-directional state
variables, their derivatives, and the control inputs as:
∆Y /m = ∆Yβ β + ∆Yp p + ∆Yr r + ∆Yδ δ
where  
∂∆Y
∆Yr = /m etc.
∂r 0
is the partial derivative of the side force increment ∆Y evaluated at trim
(called stability derivative). Using the previous first order Taylor expansions,
the linearised lateral-directional equations are:
V β̇ = Yβ β + Yp p + g cos θ0 φ + (Yr − V )r + Yδ δ
ṗ − (Izx /Ix )ṙ = Lβ β + Lp p + Lr r + Lδ δ
ṙ − (Ixz /Iz )ṗ = Nβ β + Np p + Nr r + Nδ δ
φ̇ = p
with state vector xlat = [β, p, φ, r]T and control input δ. The lateral-directional
equations can be written in descriptor form as:
Elat ẋlat = AElat xlat + BElat δ
with
 
V 0 0 0
0 1 0 −Izx /Ix 
Elat = 
0
,
0 1 0 
0 −Izx /Iz 0 1
 
Yβ Yp g cos θ0 Yr − V
 Lβ Lp 0 Lr 
AElat = 
0
,
1 0 0 
Nβ Np 0 Nr
 T
BElat = Yδ Lδ 0 Nδ

50
Because Elat is invertible, we get the standard LTI state space model for the
lateral-directional dynamics:

ẋlat = Alat xlat + Blat δ


−1 −1
with Alat = Elat AElat and Blat = Elat BElat

51
8 Lecture 8: Modal Analysis
In the previous lecture we have used a small perturbation technique to obtain
the linearised aircraft equations of motion. In order to predict the dynamic
responses of an aircraft we shall now focus on the analysis of the autonomous,
linear time-invariant system:

ẋ = Ax, x(0) = x0

where A is a n-by-n real matrix, x(t) a n-dimensional, time-varying, vector


(known as the system state vector ) and where x0 is the initial condition.
We will see that the state responses (a.k.a. state trajectories) are entirely
characterised by the eigenvalues and the eigenvectors of the system evolution
matrix A.

integrator

Figure 30: Block Diagram of ẋ = Ax,

8.1 Eigenvectors and Eigenvalues


λ ∈ C is an eigenvalue of A ∈ Rn×n if

Pc (λ) = det(λI − A) = 0

Equivalent to say that there exists a non-zero v ∈ Cn such that (λI −A)v = 0.
Such a v is called eigenvector associated with eigenvalue λ.

8.1.1 Eigenvectors and Eigenvalues Properties


• Eigenvalue λ and eigenvector v of A can be complex even if A is real

• If A and λ are real then v is real

• If A is real and v ∈ Cn is an eigenvector associated with λ ∈ C then


the conjugate vector v̄ is the eigenvector associated with the conjugate
eigenvalue λ̄ (conjugate symmetry)

52
• If A is n-by-n then typically A has n eigenvalues and n eigenvectors

• Pc is called the characteristic polynomial of A

8.2 Modal Analysis


Suppose Av = λv, v 6= 0 then if ẋ = Ax and x(0) = v then

x(t) = eλt v

• Solution x(t) = eλt v is called mode of system ẋ = Ax associated with


eigenvalue λ

• If initial state is eigenvector v then corresponding motion is simple (it


is along v)
Suppose Av = λv, v 6= 0 with λ ∈ C then
• If a ∈ C then complex trajectory aeλt v satisfies ẋ = Ax

• The real part of aeλt v satisfies ẋ = Ax as well. In addition, we have

x(t) = <(aeλt v)
  
σt
  cos ωt sin ωt α
= e vre vim
− sin ωt cos ωt −β

where v = vre + jvim , λ = σ + jω and a = α + jβ. Trajectory stays in


the plane {vre , vim }

• σ is logarithmic growth/decay

• ω is pulsation or angular frequency



• ωn = |λ| = σ 2 + ω 2 is undamped natural frequency associated with
complex eigenvalue λ as shown in Figure 31

• ζ = − ωσn is damping ratio associated with λ (Figure 31)

• If initial state is the complex eigenvector v = [v1 , v2 , . . . , vn ]T associated


with the complex eigenvalue λ = σ + jω then the relative amplitudes
and phase differences between states remain constant (Figure 32). Each
state rotates in the complex plane with frequency ω and decays or
growths exponentially at rate σ. Within a mode, the phase angles
between state components indicate whether a component leads or lags
another component.

53
Im
2n

s2 2s  n 2n x 

 sin=
n = 2 2
Re


Figure 31: Complex Conjugate Eigenvalues

• In general, the state solution is a linear combination of the system


modes n
X
tA
x(t) = e x(0) = βi (eλi t vi )
i=1
(tA)2
where etA = I + tA + 2!
+ . . . is matrix exponential.

• System is stable if state x(t) converges to 0 as t tends to infinity or


equivalently if all trajectories of ẋ = Ax converge to 0 as t → ∞.

• ẋ = Ax is stable if and only if all eigenvalues of A have negative real


part:
<λi < 0, i = 1, ..., n

8.3 Linear Time Invariant (LTI) Systems


A continuous linear time-invariant system has the form

ẋ = Ax + Bu
y = Cx + Du

54
Figure 32: Phasor Representation of a Complex Eigenvector v =
[v1 , v2 , v3 , v4 ]T

where x ∈ Rn is the system state vector, u ∈ Rm is the system input and


y ∈ Rp is the system output. When m = p = 1 the system is called sin-
gle input single output (SISO) system. LTI systems can be represented as
interconnected blocks as shown in Figure 33.
It can be verified that the state response to input u(t) is given by
Z t
tA
x(t) = e x(0) + e(t−τ )A Bu(τ )dτ
0
and therefore
Z t
tA
y(t) = Ce x(0) + Ce(t−τ )A Bu(τ )dτ + Du(t)
0

where x(t) = etA x(0) is called unforced response.


If A is invertible and the initial condition is zero then it can be shown,
using the previous result, that the system response to a unit step is:
s(t) = CA−1 (etA − I)B + D
When x and u are both constant then
ẋ = 0 = Ax + Bu
y = Cx + Du

55
B ++
1/s
integrator

+ +
D C

Figure 33: Block Diagram Representation of ẋ = Ax + Bu and y = Cx + Du

So, eliminating x we get y = (−CA−1 B + D)u. In the previous expression,


the matrix in parentheses is the system DC gain:
G(0) = −CA−1 B + D
which represents the system gain at steady state.

Example

Consider the following dynamic system:


ẋ1 = −x1 − 10x2 − 10x3
ẋ2 = x1
ẋ3 = x2
The system evolution matrix A is
 
−1 −10 −10
A= 1 0 0 
0 1 0
It can be verified that the characteristic polynomial of A is:
Pc (s) = det(λI − A) = λ3 + λ2 + 10λ + 10 = (λ + 1)(λ2 + 10)

56
and thus the eigenvalues of A are:
n √ √ o
−1, j 10, −j 10

The eigenvector associated with the first eigenvalue, λ1 = −1, is


 
−1
v1 =  1 
−1

Figure 34 shows the system response with x(0) = v1 . This first mode has
no oscillations at all; the states decay at the same rate and are proportional
to v1 at all time. The eigenvector associated with the second eigenvalue,

Figure 34: Mode associated with v1 ,



λ2 = j 10, is  
−10

v2 =  j 10 
1
For this mode, because
√ σ = 0 there is no decay, and the system states oscillate
at frequency ω = 10 ≈ 3.16 rad/s (period T = 2π ω
=≈ 1.98 s). The phasor
representation of v2 enables us to conclude that x1 and x3 will oscillate in
opposite directions with an amplitude ratio of 1:10 while x2 will lead over x3
with a phase advance of 90 degrees as seen in Figure 35.

57
Figure 35: Mode associated with v2

8.4 Transfer Matrix and Frequency Response


The transfer matrix of a state-space system with state-space data (A, B, C, D)
is given by
G(s) = C(sI − A)−1 B + D
where s is the Laplace variable. The system frequency response is obtained
by substituting s = jω into the transfer matrix and so is given by

G(jω) = C(jωI − A)−1 B + D

Note that when ω = 0 we recover the system DC gain matrix.

58
9 Lecture 9: Longitudinal Dynamics
9.1 Stability Axes Equations
As seen in Lecture 7, the linearised longitudinal equations of motion are

V (u̇/V ) = V Xu (u/V ) + Xα α − g cos(θ0 )θ + Xδe δe


(V − Zα̇ )α̇ = V Zu (u/V ) + Zα α + (V + Zq )q − g sin(θ0 )θ + Zδe δe
−Mα̇ α̇ + q̇ = V Mu (u/V ) + Mα α + Mq q + Mδe δe
θ̇ = q

with state vector xlong = [u/V, α, q, θ]T and control input δe (elevator). From
linear analysis, two modes have been identified (long and short-period modes).
We shall see that approximated low order models can be found for each mode.

9.2 Short-Period Approximation


For the short-period approximation, the following assumptions are made:

• Forward velocity almost unchanged (u ≈ 0 and thus dynamics related


to Fx can be dropped)

• Aircraft can freely rotate about its c.g. (so pitching moment equation
must be kept)

• Aircraft vertical c.g. position unconstrained (so dynamics due to Fz


must be kept)

• θ̇ = q can be dropped because θ0 , in steady level flight, is zero

These assumptions yield the simplified set of equations:

(V − Zα̇ )α̇ = Zα α + (V − Zq )q + Zδ δe
−Mα̇ α̇ + q̇ = Mα α + Mq q + Mδ δe

Both Zα̇ and Zq are negligible when compared to the free stream velocity V
and thus safely can be dropped. Thus, the approximation becomes
Zα Zδ
α̇ = α + q + δe
V V
Mα̇ Zα Mα̇ Zδ
q̇ = (Mα + )α + (Mq + Mα̇ )q + (Mδ + )δe
V V

59
or, in state space form,
   Zα     Zδ 
α̇ V
1 α
= 0 0 + V 0 δe
q̇ Mα Mq q Mδ

where
0 Mα̇ Zα
Mα = Mα +
V
0
Mq = Mq + Mα̇
0 Mα̇ Zδ
Mδ = Mδ +
V
which can be written in compact form as:

ẋsp = Asp xsp + Bsp δe

where  T
xsp = α q
From modal analysis a further approximation can be made by remarking that
α(t) ≈ θ(t) which in turn implies that α̇ ≈ θ̇ = q and thus q̇ ≈ α̈.

Mα̇ Zα Mα̇ Zδ
q̇ = α̈ = (Mα + )α + (Mq + Mα̇ )α̇ + (Mδ + )δe
V V
Zα Zδ
Note that α̇ ≈ θ̇ = q implies that V
α + V
δ ≈ 0 and thus the previous
expression reduces to

α̈ − (Mq + Mα̇ )α̇ − Mα α = Mδ δ

which is a simple second-order differential equation with natural frequency


and damping ratio:
p
ωn = −Mα , ζ = −(Mq + Mα̇ )/(2ωn )

Example (U.S. Navy A-4D Skyhawk)

Consider an A-4D Skyhawk flying at Mach=0.6 (634ft/s) and at altitude h =


15, 000f t. At this flight condition, the aircraft data are: Mα = −12.97s−2 ,
Mα̇ = −0.353 s−1 and Mq = −1.071s−1 , Zα = −518.9 f t/s2 . With the full (4
states) longitudinal model we had:

ωnsp = 3.72 rad/s and ζsp = 0.30

60
Let us compare these values with the highly simplified short-period approx-
imation given above:

ωn = 12.97 = 3.60 rad/s and ζ = (1.071 + 0.353)/(2 × 3.601) = 0.198
We can see that the natural frequency is very well approximated, however the
damping ratio approximation is not that good. Now, consider the 2-state,
short-period approximation model with the given data:

   
V
1 −0.8185 1
Asp = =
Mα + Mα̇VZα Mq + Mα̇ −12.68 −1.424
The complex conjugate pair of eigenvalues of Asp is λ1,2 = −1.118 ± 3.547j
and thus
ωn = |λ1,2 | = 3.719 rad/s and ζ = 1.118/3.719 = 0.301
These values are very close to the values found with the full order longi-
tudinal model. More generally, when an airframe possesses sufficient static
stability margin, the approximation to the short-period mode obtained by
truncating (removing) the slow state variables (i.e. u and θ) is usually very
good. However, this approximation deteriorates when the airframe c.g. gets
closer to the airframe neutral point (i.e. static stability margin reduced) be-
cause such a shift places the short-period eigenvalues closer to the phugoid
eigenvalues.

9.3 Long-Period (Phugoid) Approximation


Inspection of the long-period eigenvector pair shows that α remains small
in this mode. Thus, phugoid occurs at an almost constant angle of attack
(i.e. constant CL ). To approximate the long-period mode, the following
assumptions are made:
• Angle of attack constant =⇒ α̇ = 0 and α = 0 (since no deviation from
trim)
• q̇ = 0 since q is a fast variable (steady state value achieved almost
instantly)
• Phugoid starts in level flight =⇒ θ0 = 0
These assumptions yield the simplified set of equations:
u̇/V = Xu (u/V ) − (g/V )θ + Xδe δe
q = −Zu (u/V ) − Zδe δe
θ̇ = q

61
which are rearranged as
      
u̇/V Xu −g/V u/V Xδe /V
= + δ
θ̇ −Zu 0 θ −Zδe /V e
or
ẋph = Aph xph + Bph δe
 T
where xph = u/V θ and with characteristic equation:
λ2 − Xu λ − (gZu /V ) = 0
Example (U.S. Navy A-4D Skyhawk)
Again consider the A-4D Skyhawk flying at Mach=0.6 (V=634ft/s) and at
altitude h = 15, 000f t. At this flight condition, the aircraft data are: Mα =
−12.97s−2 , Mα̇ = −0.353 s−1 , Mq = −1.071s−1 , Zα = −518.9 f t/s2 , Xu =
−0.0129 s−1 , Zu = −0.104 s−1 , Mu = −0.0129 s−1 , g = 32.17 f t/s2 . With
the full (4 states) longitudinal model we had:
ωnph = 0.0754 rad/s and ζph = 0.00867
With the 2-state approximation model and the given data, the A-matrix of
the long-period approximation is
 
−0.0129 −0.0507
Aph =
0.104 0
with complex conjugate eigenvalues λ1,2 = −0.0065 ± 0.0732j. Thus,
ωn = |λ1,2 | = 0.0726 rad/s and ζ = 0.0888
These values are very close to the values found with the full order state space
model. The state space approximation to the phugoid mode is usually more
general and more accurate than the approximation (Lanchester
√ 1908) based
on conservation of total mechanical energy (ωn = (g/V ) 2 and ζ = 0). Note
that, in this case, the natural frequency of the phugoid given by Lanchester’s
approximation is quite reasonable (ωn = 0.071 rad/s)

9.4 Control Response


We investigate the influence of elevator on the short-period mode. Again we
assume that Zα̇ and Zq are negligible in comparison to V and consider the
control matrix Bsp in the approximation.
   Zα     Zδ 
α̇ V0
1 α
= 0 + V 0 δe
q̇ Mα Mq q Mδ

62
with
0 Mα̇ Zα
Mα = Mα +
V
0
Mq = Mq + Mα̇
0 Mα̇ Zδ
Mδ = Mδ +
V
Example (U.S. Navy A-4D Skyhawk)

Consider again the A-4D Skyhawk flying at Mach=0.6 (V=634ft/s) at an


altitude h = 15, 000f t. At this condition, the data are: Mα = −12.97s−2 ,
Mα̇ = −0.353 s−1 , Mq = −1.071s−1 , Zα = −518.9 f t/s2 , Zu = −0.104 s−1 ,
Mu = −0.0129s−1 , Zδ = −57.02f ts−2 , Mδ = −19.46s−2 and g = 32.17f t/s2 .
With these data, the state space model of the short-period approximation is
   
−0.8185 1 −0.0899
Asp = , Bsp =
−12.68 −1.424 −19.4283

If the output is chosen as y = xsp then C = I2 and D = 02×1 and so the


system DC gain is given by

G(0) = −A−1
sp Bsp

which, in this case, is  


−1.412
G(0) =
−1.066
Note that the negative values of DC gain are due to the fact that a positive
elevator deflection (trailing edge moving down ⇒ nose down) results in a
negative value for q and a negative variation in α when the aircraft is stable.
The frequency responses due to elevator control are shown in Figure 37.
Note that, at ω = 0, gains (in dB) and phase values agree with the DC gain
values previously obtained. Peak gain values occur near to the short-period
resonance frequency of 3.55 rad/s.

63
Figure 36: Short-period response to a unit step demand in elevator

Figure 37: Short-period frequency response to elevator input

64
9.5 More on Longitudinal Stability Derivatives
The dimensional stability derivatives are the coefficients of the matrices of
a linearised airframe. They can be related to the dimensionless stability
derivatives (those which are normally found in aerodynamic data bases).
The longitudinal dimensional derivatives have been previously defined as:

S ∂(q̄Cx )
Xu =
mV ∂(u/V )
S ∂(q̄Cz )
Zu =
mV ∂(u/V )
Sc ∂(q̄Cm )
Mu =
Iy V ∂(u/V )
q̄S ∂Cx
Xα =
m ∂α
q̄S ∂Cz
Zα =
m ∂α
q̄Sc ∂Cm
Xα =
Iy ∂α
etc.

where

S is reference wing area

c is reference wing chord

m is airframe mass

Iy is pitch axis mass moment of inertia

u is axial velocity

V is freestream velocity

q̄ is dynamic pressure

Cx is dimensionless axial force coefficient

Cz is dimensionless normal force coefficient

Cm is dimensionless pitching moment coefficient

65
9.5.1 Derivatives Due to Velocity
We are interested in expressing
∂X
∂V
in terms of the derivatives of the dimensionless coefficient CD . The axial
force (X) can be written as:
X = q̄SCx
The variation of X due to a change in V is
∂X ∂ q̄ ∂Cx
= S Cx + S q̄
∂V ∂V ∂V
∂Cx
= S q̄(2V /V 2 )Cx + S q̄
∂V
∂Cx
= S q̄(2/V )Cx + S q̄
∂V
q̄S ∂Cx
= (2Cx + V )
V ∂V
q̄S
= (2Cx + V CxV )
V

Because Cx = −CD then


∂X q̄S
=− (2CD + CDV )
∂V V
1 ∂X q̄S
⇒ XV ≡ =− (2CD + CDV )
m ∂V mV
where
∂CD
CDV ≡ V
∂V
Or, as a function of Mach number, M ,
q̄S ∂CD
XV = − (2CD + M )
mV ∂M
Similarly, we can show that
q̄S ∂CL
ZV = − (2CL + M )
mV ∂M
Example

66
The characteristic equation associated with the phugoid approximation, as
given in the previous section, is

λ2 − Xu λ − (gZu /V ) = 0

which has the form


λ2 + 2ζωn λ + ωn2 = 0
In level flight, lift balances weight (L = W ), thus CL = W/(S q̄).
Fact: the influence of Mach number at low to medium subsonic flight can
be neglected:
q̄S
Zu = − 2CL = −2g/V
mV
Now because ωn2 = −gZu /V we get ωn2 = 2(g/V )2 or

ωn = 2(g/V )

which is exactly the same as Lanchester’s formula! For the damping, we


have:
ρV S
2ζωn = −Xu = CD
m
Dividing by the approximation for ωn and with the expression of CL given
above, we get an estimate for the phugoid mode damping:

ζ = CD /( 2CL )

Hight performance gliders have lift-drag ratio of about 1:40. Using this damp-
ing ratio approximation,
√ the phugoid is expected to be extremely lightly
damped (ζ ≈ 40 × 2 ≈ 0.017)

9.5.2 Derivatives Due to α


Assume:

• angle-of-attack is disturbed from trim by an amount ∆α(t). That is


α = αT + ∆α(t) where αT is angle-of-attack at trim and where ∆α(t)
is the angle-of-attack disturbance

• velocity vector initially aligned with body axes

Figure 38 shows

Cx = −CD cos ∆α(t) + CL sin ∆α(t)

67
Figure 38: Aircraft in AoA perturbation

Taking partial derivative of Cx w.r.t. ∆α(t) gives


∂Cx ∂CD ∂CL
= CD sin ∆α − cos ∆α + CL cos ∆α + sin ∆α
∂α ∂α ∂α
Using the fact that ∆α is very small (⇒ sin ∆α ≈ 0 and cos ∆α ≈ 1)
yields
∂Cx ∂CD
= CL −
∂α ∂α
thus  
q̄S ∂CD
Xα = CL −
m ∂α
Following the lines of the examples given above all the other dimen-
sional stability derivatives can be expressed in terms of dimensionless sta-
bility derivatives.

9.6 Longitudinal Equations in Different Body Frames


Other sets of body axes can be used to define the longitudinal equations of
motion. For instance, in body axes the longitudinal equations becomes
• Force equations:
X
u̇ = − g sin θ − qw
m
Z
ẇ = + g cos θ + qu
m

68
• Moment equation:
M
q̇ =
Iy

• Kinematic equation:
θ̇ = q

We can also mix stability and inertial frames if altitude is introduced (Figure
39). In this case, if one introduces thrust, the longitudinal equations of
motion take the form:

• Force equations:

mV̇ = T cos α − D − mg sin γ


−mV γ̇ = −T sin α + L + mg cos γ

• Moment equation:
1
q̇ = (M + MT )
Iyy

• Kinematic equation:
α̇ + γ̇ = q

• Heave velocity equation:


ḣ = V sin γ

where γ is the flight-path angle defined as γ = θ − α, h is altitude, D is drag,


L is lift, V is air-speed and T is thrust.

9.7 Summary
• Two longitudinal modes have been identified (phugoid and short-period)

• Low order approximation models have been developed

69
L xbody


xaero
T V
 
D xearth

zaero

Figure 39: Longitudinal Forces

70
10 Lecture 10: Lateral Dynamics
Lateral dynamics are extremely coupled (i.e. yaw and roll motions are not
independent)

10.1 Lateral-directional Linearised Equations of Mo-


tion
The lateral-directional states are β, p, φ and r and we will assume that
v dφ dψ
β= , p= , r=
V dt dt
T
Lateral control is δ = [δa , δr ] where δa is aileron and δr is rudder w.r.t.
trim positions. As done with the longitudinal equations, we can express
force and moment increments as a first order Taylor expansions. With these
first order approximations, the linearised lateral-directional equations of mo-
tion, describing small perturbations about an equilibrium condition, can be
written as:
ẋ = Ax + Bu
 T
with state vector x = β p φ r and control input u = [δr , δa ]T where
the lateral-directional evolution matrix is
 Yβ Yp g cos θ0 Yr −V 
V V0 V V
 L0 Lp 0
0
Lr 
A =  β 
0 1 0 0 
0 0 0
Nβ Np 0 Nr
and the control matrix
Y Yδ a

δr

 LV0 V0
Lδa 
 δr 
B = 
 0
V V 
 0 0 
0

Nδr Nδa
V V
with
0 0
Lβ = G[Lβ + Nβ (Ixz /Ix )], Lp = G[Lp + Np (Ixz /Ix )]
0
Lr = G[Lr + Nr (Ixz /Ix )]
0 0
Nβ = G[Nβ + Lβ (Ixz /Iz )], Np = G[Np + Lp (Ixz /Iz )]
0
Nr = G[Nr + Lr (Ixz /Iz )]
0 0
Lδ = G[Lδ + Nδ (Ixz /Ix )], Nδ = G[Nδ + Lδ (Ixz /Iz )]
2
and with G = 1/[1 − Ixz /Ix Iz ]

71
Note that
X Z
Ix = (yi + zi )mi = (y 2 + z 2 )dm (moment of inertia about x axis)
2 2

X Z
Iy = (zi + xi )mi = (z 2 + x2 )dm (moment of inertia about y axis)
2 2

X Z
Iz = (xi + yi )mi = (x2 + y 2 )dm (moment of inertia about z axis)
2 2

X Z
Iyx = Ixy = xi yi mi = xy dm (xy product of inertia)
X Z
Izx = Ixz = xi zi mi = xz dm (xz product of inertia)
X Z
Izy = Iyz = zi yi mi = zy dm (zy product of inertia)

Mass moments and products of inertia are airframe dependent and we will
assume that they are known (manufacturer data). For the Airbus A300, the
mass moments/products of inertia (in body axes) are:

Ix = 5.55 106 kg m2
Iy = 9.72 106 kg m2
Iz = 14.51 106 kg m2
Izx = 3.3 104 kg m2
Iyx = Iyz = 0

For a conventional aircraft configuration, the lateral-directional dynamics


consist of three modes (2 real and 1 oscillatory) which are now detailed.

10.1.1 Roll Mode


• roll mode (a.k.a roll subsidence) is a real (→ non-oscillatory) lateral
mode

• fastest of all the lateral modes and consists of almost pure rolling mo-
tion about aircraft x-axis (roll rate). Can be approximated as

ṗ = Lp p + Lδa δa

where δa is lateral control (e.g. aileron)

• lateral control produces roll rate (angular velocity about x-axis)

72
• roll mode response to aileron step demand (δ0 ) applied at t = 0 is given
by
p(t) = pss (1 − eLp t ), t > 0
where pss = −Lδa δ0 /Lp

• roll damping (Lp ) is normally negative. Can define roll time constant
as
τ = −1/Lp
and so response to a step is

p(t) = pss (1 − e−t/τ ), t>0

10.1.2 Spiral Mode


• the other real mode (can be unstable)

• very slow mode

• involves couplings in roll, yaw and side-slip

• unlike roll mode roll rate is small

• roll angle φ is the dominant


R state, slight participation by the yaw rate
r, however heading ψ = rdt is large

• stable spiral yields aircraft slowly rolling back towards the wings level
position

• unstable spiral yields slowly diverging bank angle as the aircraft starts
into a gradual spiral turn (tends to a spiral into the ground)

10.1.3 Dutch-Roll Mode


• Dutch-roll is the oscillatory (associated with the complex conjugate
eigenvector/eigenvalue pair) damped lateral mode

• similar to the longitudinal short-period mode (but not as well damped)

• involves interaction between all the lateral-directional variables

The origin of the name Dutch roll is unclear. Describe lateral asymmetric
motion of an aircraft similar to ice skating motion.

73
10.2 Lateral Stability Derivatives
For conventional aircraft we have:

• Rolling moment Lp due to a positive disturbance in roll rate p (Figure


40) is negative, or equivalently if

q̄Sb ∂Cl
Lp = <0
Ix ∂p

Figure 40: Rolling moment due to a disturbance in roll rate

• A positive yaw rate (r > 0) makes left wing advancing and right wing
retreating. Thus, horizontal velocity distribution over wing is U =
U0 − ry. Thus higher lift on left wing than on right wing. Thus positive
yaw rate creates a positive rolling moment, that is
q̄Sb ∂Cl
Lr = >0
Ix ∂r

• The ability of an aircraft to yaw into wind is realised thanks to verti-


cal fin. Directional static stability is achieved if the yawing moment,
due to a positive disturbance in yaw rate r, is negative (Figure 41) or
equivalently if
q̄Sb ∂Cn
Nr = <0
Iz ∂r

74
Figure 41: Yawing moment due to disturbance in yaw rate

10.3 Example: DC-8, Mach=0.84, alt=33000 ft, V=825


ft/s
At this flight condition, the lateral-directional system matrix is
 Yβ Yp g cos θ0 Yr −V   
V0 V0 V V0
−0.0869 0 0.039 −1
 =  −4.424 −1.184 0 0.335 
L L 0 Lr  
A =  β p 
0 1 0 0   0 1 0 0 
0 0 0
Nβ Np 0 Nr 2.148 −0.021 0 −0.228
x = [β, p, φ, r]T

1. The eigenvalues of A are:

λ1 = −1.258: roll mode


λ2 = −0.0040: spiral mode
λ3,4 = −0.1184 ± 1.4932j: Dutch-roll mode

The lateral-directional dynamics are stable. The roll response is the


fastest and thus λ1 is the eigenvalue that characterises the roll mode.
The other (very slow) real mode is the spiral mode which is charac-
terised by λ2 . Finally, the complex eigenvalue pair corresponds to the
Dutch-roll mode.
2. Dutch-roll mode. The information contained in λ3,4 is

ωDR = 1.4979 rad/s and ζDR = 0.0791

75
and the Dutch-roll period is

TDR = = 4.208 s
1.4932
The Dutch-roll mode eigenvector given in polar form (magnitude, phase
in degrees) is

State Magnitude Phase (degrees)


β 0.29 -131.83
p 0.71 0
φ 0.47 -94.5
r 0.43 141.3

Table 3: Dutch-roll mode eigenvector

Figure 42: Dutch-roll mode phasor representation

76
The table shows that the four lateral-directional states are involved (→
no sensible approximation can be made). A phasor diagram is given in
Figure 42.

3. Roll mode. The roll time constant is


1
τroll = − = 0.7949 s
λ1
and the roll eigenvector is
   
β −0.0124
 p  −0.7827
xroll =
φ =  0.6222 
  

r 0.0098

xroll shows that the roll mode is dominated by p and φ whereas β and
r do not contribute much to the mode.

4. To the spiral mode has a time constant of


1
τspiral = − = 250.34 s
λ2
and the spiral eigenvector
   
β 0.004
 p  −0.0040
xspiral =
φ =  0.9992 
  

r 0.0386

confirms that the spiral mode is essentially dominated by φ.

77
A Aircraft Mass Moment of Inertia Matrix
• Because the xz plane is the aircraft plane of symmetry it can be verify
that
Ixy = Iyz = 0
and the only off-diagonal term remaining in the inertia matrix is Izx .

• Very often mass moments/products of inertia are computed w.r.t. the


aircraft centre of mass and within body axes. In such a case, the mass
moment of inertia is a constant matrix.

• An aircraft is particular body with dimensions along the z axis small


compared to the dimensions along the y and x axes. Therefore, in the
above integrals, the z 2 terms are small compared to the x2 and y 2 terms
and so Izz ≈ Ixx + Iyy .

For instance, for the Airbus A300, the mass moments/products of inertia (in
body axes) are:

Ixx = 5.55 106 kgm2


Iyy = 9.72 106 kgm2
Izz = 14.51 106 kgm2
Izx = 3.3 104 kgm2
Iyx = Iyz = 0

Figure 43: A Lufthansa A300

78
References
[1] L. V. Schmidt, Introduction to Aircraft Flight Dynamics, AIAA Educa-
tion Series, (1998)

[2] B. L. Stevens & F. L. Lewis, Aircraft Control and Simulation, Wiley,


(1994)

[3] R. F. Stengel, Flight Dynamics, Princeton University Press, (2004)

[4] M. V. Cook, Flight Dynamics Principles, Elsevier Aerospace Engineer-


ing Series, (2008)

[5] B. Etkin & L. D. Reid, Dynamics of Flight, Wiley, (1995)

[6] http://www.history.nasa.gov/SP-367/contents.htm

[7] http://ocw.mit.edu/courses/aeronautics-and-astronautics.
../16-333-aircraft-stability-and-control-fall-2004/

[8] http://www.princeton.edu/~stengel/MAE331Lectures.html

79

You might also like