Beruflich Dokumente
Kultur Dokumente
GARTEUR Open
by
GARTEUR Open
by
List of authors
Gertjan Looye1 ,
Samir Bennani2
1 Delft University of Technology, fac. of Aerospace Engineering, sect. Stability & Control
from 1/10/97: DLR Oberpfaenhofen, German Aerospace Research Establishment,
Institute for Robotics and System Dynamics
P.O. Box 1116, D-82230 Wessling, Germany
E-mail: Gertjan.Looye@dlr.de
2 Delft University of Technology, fac. of Aerospace Engineering, sect. Stability & Control
Kluyverweg 1, 2629 HS Delft, The Netherlands
E-mail: S.Bennani@lr.tudelft.nl
Version: 1
Date: April 4, 1997 ii
GARTEUR/TP-088-27
Summary
This report contains a description and an analysis of the Research Civil Aircraft Model
(RCAM). The RCAM is based on a modied civil aircraft during the nal approach. It
is used for controller design and simulation in the civil variant of the two Robust Flight
Control design challenges, prepared by the GARTEUR Action Group FM(AG08). A set
of design requirements for the design challenge is given in the RCAM problem formulation
manual [2].
This document has two main objectives:
documentation of the aircraft modeling and disturbance modeling (windshear, gust),
including implementation in Matlab/Simulink;
open loop analysis of the RCAM as a basis for the design of a
ight controller. This
analysis includes performance limitations, dynamic characteristics and co-ordination
of controls in specic situations, like turns, windshear, and engine failure.
The analysis is carried out using linear system theory, point-mass approximations and
steady state analysis in several
ight situations. Especially energy concepts and Korham-
mer diagrams have proven to be very illuminating.
The longitudinal performance and dynamics of the RCAM are representative for a civil
airliner, while the lateral dynamics have strange properties, like a very fast spiral mode.
In fact, in spite of many simplications, the most interesting
ight situations are covered
by the model, including backside operation and stall.
Version: 1
iii Date: April 4, 1997
GARTEUR/TP-088-27
Distribution list
FM(AG08) Principal Persons
Ahmed S (UK) CCL 1 copy
Ambrosino G (IT) UNAP 1 copy
Bernussou J (FR) LAAS 1 copy
Cruz J de la (ES) UCM 1 copy
Delgado I (ES) INTA 1 copy
Dormido S (ES) UNED 1 copy
Duda H (DE) DLR 1 copy
Escande B (FR) ONERA 1 copy
Faleiro L (UK) LUT 1 copy
Fielding C (UK) BAe-MA 1 copy
Game G (UK) BAeDD 1 copy
Gautrey J (UK) CUN 1 copy
Helmersson A (SE) LiTH 1 copy
Joos H (DE) DLR 1 copy
Laidlaw D (UK) AVRO 1 copy
Luckner R (DE) DASA 1 copy
Maciejowski J (UK) UCAM 1 copy
Magni J (FR) ONERA 1 copy
Muir E (UK) DRA 1 copy
Postlethwaite I (UK) ULES 1 copy
Schram G (NL) DUT-EE 1 copy
Stahl-Gunnarsson K (SE) SMA 1 copy
Terlouw J (NL) NLR 1 copy
Tonon A (IT) ALN 1 copy
Vaart J van der (NL) DUT-AE 1 copy
Verde L (IT) CIRA 1 copy
Members of the Flight Mechanics, Systems and Integration Group of
Responsables
Brannstrom B (SE) FMV 1 copy
Doorn J van (NL) NLR 1 copy
England P (UK) DRA 1 copy
Mu~noz F (ES) INTA 1 copy
Rodlo R (DE) DLR 1 copy
Verbrugge R (FR) ONERA 1 copy
Version: 1
Date: April 4, 1997 iv
GARTEUR/TP-088-27
Contents
List of gures
2.1 The body-xed, vehicle carried-vertical and the earth-xed axes systems 5
2.2 The vehicle-carried vertical and the
ightpath axes systems . . . . . . . 6
2.3 The body-xed and the
ightpath axes systems . . . . . . . . . . . . . . 7
2.4 The wind, stability and body-xed axes systems . . . . . . . . . . . . . 7
2.5 Transformation from the vehicle-carried system to the wind axes system 8
2.6 The denition of wind angles . . . . . . . . . . . . . . . . . . . . . . . . 9
2.7 Denitions of lift and drag . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.8 Curves of CL and CD vs. . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.9 The lift-drag polar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.10 Location of moment points . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.11 Power spectral densities of Dryden lters . . . . . . . . . . . . . . . . . 23
2.12 Location of vortices of windshear model . . . . . . . . . . . . . . . . . . 25
2.13 Windeld due to windshear and downburst . . . . . . . . . . . . . . . . 25
2.14 Encountered wind speeds in windshear/downburst . . . . . . . . . . . . 26
2.15 Simulink block diagram of the RCAM simulation environment . . . . . . 27
3.1 Longitudinal aircraft motion . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.2 Force equilibrium for unpowered
ight in undisturbed atmosphere . . . 32
3.3 Korhammer Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.4 Korhammer Diagram; accelerated
ight . . . . . . . . . . . . . . . . . . 36
3.5 Eect of changing aerodynamic trim point in Korhammer diagram: angle
of attack and
ap de
ection . . . . . . . . . . . . . . . . . . . . . . . . 38
3.6 Korhammer Diagram;
ight below minimum drag speed . . . . . . . . . 39
3.7 Korhammer Diagram; eect of thrust commands . . . . . . . . . . . . . 40
3.8 Performance related to the lift-drag polar . . . . . . . . . . . . . . . . . 44
3.9 Modication of the lift-drag polar . . . . . . . . . . . . . . . . . . . . . . 45
3.10 Performance diagrams for dierent masses (forces) . . . . . . . . . . . . 47
3.11 RCAM performance diagrams (power) . . . . . . . . . . . . . . . . . . . 48
3.12 Korhammer Diagram; maximum
ightpath angle . . . . . . . . . . . . . 49
3.13 Maximum
ightpath angle vs speed . . . . . . . . . . . . . . . . . . . . . 50
3.14 Minimum
ightpath angle vs speed . . . . . . . . . . . . . . . . . . . . . 50
3.15 Aircraft in co-ordinated turn . . . . . . . . . . . . . . . . . . . . . . . . 53
3.16 RCAM performance diagrams for dierent bank angles (force) . . . . . . 55
3.17 RCAM performance diagrams for dierent bank angles (power) . . . . . 56
3.18 Limitations on the turning radius . . . . . . . . . . . . . . . . . . . . . . 57
3.19 Limitations on the turning rate . . . . . . . . . . . . . . . . . . . . . . . 59
3.20 Modied Korhammer diagram for a steady turn . . . . . . . . . . . . . . 60
3.21 Maximum
ightpath angle vs speed in turn . . . . . . . . . . . . . . . . 61
3.22 Denition of wind vectors . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Version: 1
ix Date: April 4, 1997
GARTEUR/TP-088-27
List of tables
2.1 Aircraft geometry and inertial parameters . . . . . . . . . . . . . . . . . 12
2.2 Numerical values of the RCAM aerodynamic model parameters . . . . . 20
2.3 Numerical values control saturations . . . . . . . . . . . . . . . . . . . . 21
2.4 States denitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.5 Inputs denitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.6 Outputs denitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.1 Characteristics of the longitudunal and the lateral modes . . . . . . . . 83
4.2 Initial and steady state gains for step inputs . . . . . . . . . . . . . . . . 108
C.1 Trim settings of the controls . . . . . . . . . . . . . . . . . . . . . . . . . 132
C.2 Flight data for trim condition . . . . . . . . . . . . . . . . . . . . . . . . 132
C.3 RCAM state space model for nominal trim condition . . . . . . . . . . . 133
C.4 Longitudinal linear model, with transformed state vector . . . . . . . . . 134
C.5 Lateral linear model, with transformed state vector . . . . . . . . . . . . 134
E.1 Characteristics of lon. modes, speed variations . . . . . . . . . . . . . . 136
E.2 Characteristics of the lateral modes, speed variations . . . . . . . . . . . 136
E.3 Characteristics of the lon. modes, center of gravity shifts . . . . . . . . . 137
E.4 Characteristics of the lateral modes, center of gravity shifts . . . . . . . 137
E.5 Characteristics of the lon. modes, mass variations . . . . . . . . . . . . . 137
E.6 Characteristics of the lateral modes, mass variations . . . . . . . . . . . 137
Version: 1
Date: April 4, 1997 xii
GARTEUR/TP-088-27
Derivatives of variables
D_ specic energy distribution rate m=s
E_s specic energy rate m=s
E_ tot total specic energy rate m=s
Control surface de
ections
f
ap position rad
Wind related quantities
B b {
2Lg
Lg scale of turbulence m
ug gust speed in x-direction m=s
vg gust speed in y-direction m=s
wg gust speed in z -direction m=s
uwx spacial der. wind speed in x-direction s 1
vwx spacial der. wind speed in y-direction s 1
wwx spacial der. wind speed in z -direction s 1
g angle of attack due to gust rad
w angle of attack due to wind rad
g sideslip angle due to gust rad
w sideslip angle due to wind rad
misc. symbols
H (s) transfer function {
!0 eigen frequency rad=s
Aerodynamic coecients
CLwb dCLwb =d {
CLt dCLt =dt {
CR total aerodynamic force coecient {
CW weight normalized with qS {
Accelerations and load factors
ax measured acceleration in xF -direction m=s2
az measured acceleration in zF -direction m=s2
nx
load factor, nx
= (T D)=W {
nz
load factor, nz
= L=W {
Subscripts
a w.r.t. airmass
ac aerodynamic center
cg center of gravity
f w.r.t.
ightpath
g gust, turbulence
B relative to body axes
Version: 1
Date: April 4, 1997 xiv
GARTEUR/TP-088-27
Abbreviations
ANDECS ANalysis and DEsign of Controlled Systems
ALN Alenia Aeronautica
AVRO Avro International Aerospace
BAeDD MATRA - British Aerospace Defence Dynamics
BAe-MA British Aerospace Military Aircraft
CC Classical Control
CCL Cambridge Control
CERT Centre d'Etudes et de Recherches de Toulouse
CIRA Centro Italiano Ricerche Aerospaziali
CUN Craneld University
DASA Daimler-Benz Aerospace Airbus
DE Germany
DLR Deutsche Forschungsanstalt fur Luft- und Raumfahrt
DRA Defence Research Agency
DUT-AE Delft University of Technology, Department of
Aerospace Engineering
DUT-EE Delft University of Technology, Department of
Electrical Engineering
EA Eigenstructure Assignment
ES Spain
FFA The Aeronautical Research Institute of Sweden
FL Fuzzy Logic control
FMAG Flight Mechanics Action Group
FM-GoR Flight Mechanics, Systems and Integration Group of Responsables
FMV Defense Material Administration
FR France
GARTEUR Group for Aeronautical Research and Technology in EURope
HI H1-Synthesis
INTA Instituto Nacional de Tecnica Aerospacial
IT Italy
LAAS Laboratoire d'Analyse et d'Architecture des Systemes
Version: 1
xv Date: April 4, 1997
GARTEUR/TP-088-27
1 Introduction
This document contains a description and an analysis of the Research Civil Aircraft Model
(RCAM). This nonlinear model is based on a large civil transport aircraft in landing
conguration. It is used for controller design and simulation in one of the two Robust
Flight Control design challenges, prepared by GARTEUR Action Group FM(AG08), see
[3, 4]. The idea behind these design challenges is to invite the European industry, research
establishments and universities involved with
ight control to apply their modern and
classical
ight control techniques to realistic
ight control problems and to demonstrate in
this way the capabilities, applicability and limitations of these methods. It is also intended
that these problems will raise the awareness and condence of the European aeronautical
industry in the use of robust control techniques.
The design challenges are based on a military and a civil
ight control problem. RCAM
is the basis for the civil variant. An extensive set of design objectives and additional
requirements can be found in the manual prepared by GARTEUR, see [2].
This document has two main objectives:
detailed documentation of the aircraft modeling and disturbance modeling (winds-
hear, gust) for the RCAM, including implementation in Matlab/Simulink;
open loop analysis of the RCAM as a basis for the design of a
ight controller. This
analysis includes performance limitations, dynamic characteristics and co-ordination
of controls in specic situations, like turns, windshear, engine failure etc.
The main task of the controller is guidance of the aircraft along trajectories with turns,
climbs and descents, while for example also wind disturbances or an engine failure may
occur. Along these trajectories, physical aircraft performance (e.g. turning performance,
maximum climb angles etc.) is a limiting factor, that of course can not be overcome by the
implementation of a "robust" controller. For this reason aircraft performance is analysed,
mainly using point-mass
ight mechanics.
From the nonlinear model, linearizations can be determined at dierent operating points.
One operating point is dened as nominal, where standard tools from linear system theory
are applied, like bode plots, step responses, and eigenvalues to analyse dynamic charac-
teristics of the RCAM. Furthermore, the eects of airspeed variations, center of gravity
shifts and mass variations are considered.
Aircraft control is basically a multi-input multi-output problem. Therefore an analysis is
carried out to derive some a priori knowledge about proper co-ordination of controls, like
for example elevator/throttle co-ordination in climbing
ight. This study is based on the
equations of motion, point-mass approximations, and steady state analysis.
Although in many modern multivariable design techniques control laws are synthesized
automatically, the results can be used as a reference in the assessment of the controller.
Version: 1
Date: April 4, 1997 2
GARTEUR/TP-088-27
This document may serve as a "database" with RCAM data, concerning the modeling,
dynamic characteristics and performance, as far as these can be derived from the model.
Therefore, most of the results are presented in gures or in tabular form.
The outline is as follows: in the chapter 2 the modeling and implementation of the RCAM
is described. In the third chapter the analysis work based on point-mass and steady state
ight mechanics is described, including Korhammer diagrams, energy principles and steady
state aircraft performance. In chapter 4 the analysis work based on linearized models is
described.
Version: 1
3 Date: April 4, 1997
GARTEUR/TP-088-27
2.1.1 Assumptions
In order to make the nonlinear equations of motion not unnecessarily complicated, some
assumptions for the generic nonlinear model will be made rst. The aerodynamic model
and the engine model, which are specic for the RCAM, are based on more detailed
assumptions.
The earth is in rest and is considered to be
at. This implies that any reference frame
xed with respect to the earth (position and orientation), is an inertial reference
system.
The equations are valid for a rigid body, which means that all elastic degrees of
freedom of any part of the aircraft are neglected. This assumption holds as long
as the eigenfrequencies of the elastic modes are much higher than those of the rigid
body motions.
The mass of the aircraft is constant. This is a reasonable assumption since the time
constant of fuel consumption is much higher than the time constants of the rigid
body modes. Simulation times will be in the order of minutes.
The aircraft is symmetric with respect to its xB -zB {plane. For this reason the Ixy
and Iyz in the inertia tensor disappear.
Interactions between the engines and the other parts of the airframe are not taken
into account. This means that cross coupling induced by the rotation of the engines
Version: 1
Date: April 4, 1997 4
GARTEUR/TP-088-27
The forces and moments acting on the aircraft can be divided into three groups:
aerodynamic forces and moments,
forces and moments induced by the power plant,
gravity forces.
The rst two are complex, multidimensional nonlinear functions and they are usually ap-
proximated by polynomial expansions. These forces and moments arise in dierent axes
systems. Although these systems are described in the Communication Handbook, [1], they
will also be discussed in this section, so that some important
ight mechanical quantities
can be introduced simultaneously.
xB
xV
θ k1
ψ
yV
CoG
ψ
xV, yV φ k2
xE R xB , z v
yB
φ
yE
yB , z B
zB
θ
k3
zV
zE
Fig.2.1 The body-xed, vehicle carried-vertical and the earth-xed axes systems
The second vehicle-carried system, the body-xed axes system, is obtained through succes-
sive rotations of the vertical vehicle-carried system with heading angle , pitch angle and
roll angle , see gure 2.1. These angles form the vector of Euler angles E = [ ; ; ]T ,
which describes the attitude of the aircraft. The xB -axis is directed towards the nose of
the aircraft, the yB -axis towards the right wing and the zB -axis towards the bottom of
the aircraft. The components of the velocity vector along these axes are uB , vB and wB ,
the components of the vector of angular rates around the axes,
B , are respectively: the
roll rate p, the pitch rate q and the yaw rate r.
The x-axis of the
ightpath axes system is in the direction of the velocity vector V. The
system is obtained by rotating the vehicle-carried vertical system about the zV -axis over
the
ight path azimuth angle () and about the yF -axis over the
ightpath (elevation)
angle (
) respectively, see gure 2.2.
Both angles can be calculated using the velocity components in the vehicle-carried vertical
frame from respectively:
tan
= q 2 wV 2 (2.1)
uV + vV
and:
tan = uvV (2.2)
V
Version: 1
Date: April 4, 1997 6
GARTEUR/TP-088-27
xF
xV χ γ
V yV
χ
CoG
yF
zF
zV
xB
αf
βf xF
V
CoG
βf
yF
µf
yB
µf
αf
zB
zF
xB
xS
α xW
β
VA
CoG Y
D
β
yB , yS
yW
zB
zW , z S
Where aax , aay and aaz are the time derivatives of the airspeed components along the body
axes (e.g. aax = dudta ). The aerodynamic forces in the model to be derived are expressed
with respect to the stability, or experimental axes system. The term \experimental" comes
from the fact that this system is also used in wind tunnel experiments. The system should
not be confused with the "quasi-stationary" stability axes, used in many textbooks. In
stationary
ight the systems are identical, but when the air
ow becomes unsteady, the
orientation of the stability axes in these books is considered to be xed, while the stability
axes used here vary with the direction of the airspeed, Va . Although it would be more
convenient to use the term "experimental", the nomenclare from [1] will be adopted here.
The stability axes system is obtained from the body axes system in almost the same way
as the wind axes; the body axes system however, is only rotated about the yB -axis over
the angle of attack (), see gure 2.4.
Parallel to the body axes, the wind reference system can obtained from the vehicle carried
vertical system by successive rotations over the angles a ,
a and a respectively, as can
be seen in gure 2.5. The azimuth angle a can be calculated from:
xW
xV χa γa
VA yV
χa
CoG
µa
yW
µa
γa
zW
zV
Fig.2.5 Transformation from the vehicle-carried system to the wind axes system
va
a = arctan u V (2.8)
aV
The elevation angle is obtained as follows:
0 1
a = arctan @ q 2 waV A (2.9)
uaV + va2V
For the small sideslip angle and small angle of attack the bank angle a is approxi-
Version: 1
9 Date: April 4, 1997
GARTEUR/TP-088-27
V
CoG
yF
yB
β
µw yW
βw
zF
zV
with respect to another system using transformation matrices. The transformation matrix
from axes system M to axes system N is expressed using the notation RNM , where
RNM = RMN T .
Usually one system is obtained from the other via three successive rotations, like the body-
xed system is obtained from the vehicle-carried vertical system via the angles , and
respectively, see gure 2.1. In a more general form, the transformation matrix resulting
from rotations over 1 , 2 and 3 respectively, can be written as:
RNM = R
2 3(3 ) R2(2 ) R1(1 ) 3 2 32 (2.11)
3
6 1 0 0 7 6 cos 2 0 sin 2 7 6 cos 1 sin 1 0 7
= 4 0 cos 3 sin 3 75 64 0 1
6 0 75 64 sin 1 cos 1 0 75
0 sin 3 cos 3 sin 2 0 cos 2 0 0 1
For the vehicle-carried systems, the most important transformation matrices are:
Version: 1
Date: April 4, 1997 10
GARTEUR/TP-088-27
Force equations
The force and moment equations for a rigid body in an axes system that is xed to the
Version: 1
11 Date: April 4, 1997
GARTEUR/TP-088-27
center of gravity of this body, were rst derived by Euler. Using the second assumption
in section 2.1.1, the force equations in the body-xed system become:
2 3 2 3 2 3 2 3
dV 6 u_ 77 = 1 R 66 FxA 77 + 1 66 FxT 77 +R 66 0 77 g
V (2.16)
6
= 4 v_ 5 m BS 4 FyA 5 m 4 FyT 5 BV 4 0 5 B B
dt B
w_ B FzA S FzT
1 V B
(vector subscripts indicate axes system) The term induced by the coupling between velocity
and rotation can be expanded as follows:
2 3 2 3 2 3
6 p7 6u 77 = 66 qw rv 7
B VB = 4 q 75 64 v
6 5 4 ru pw 75 (2.17)
r B w B pv qu B
Expanding the gravity term gives:
2 3 2 3
6 0 7 6 g sin 77
RBV 64 0 75 g = 64 g sin cos 5 (2.18)
1 V g cos cos
where g = 9:81 m=s.
Moment equations
The moment equations can be written in a similar general form as the force equations:
2 3 02 3 2 3 1
d
6 p_ 7 LA 77 + 66 LT
= 64 q_ 75 = I 1B 6
B@64 MA 77
B I
B CCA
dt B 5 4 MT 5 (2.19)
r_ B NA B NT B
The inertia tensor I contains the moments of inertia with respect to the body-xed axes
system. With Ixy = 0 and Iyz = 0 (symmetry assumed) the tensor can be written as:
2 3
6 Ix 0 Ixz 7
I = 64 0 Iy 0 75 (2.20)
Ixz 0 Iz
The values of the principal moments of inertia IX , IY and IZ can be found in table 2.1.
These are the moments of inertia with respect to the principal axes, in which the cross
products are zero. The principal axes are obtained by rotating the body axes about the
yB -axis over the angle p, see also for example [10]. The inertia tensor in body axes is
calculated using standard formulas:
2 (IX +IZ ) 3
6 2 + (IX 2 IZ ) cos( 2p ) 0 (IX IZ ) sin(
2 2p ) 77
I = 64 0 IY 0 5 (2.21)
(IX IZ ) sin( 2 ) 0 (IX +IZ ) + (IX IZ ) cos( 2p )
2 p 2 2
Version: 1
Date: April 4, 1997 12
GARTEUR/TP-088-27
Kinematic relations
The remaining six equations needed to solve the equations of motion for twelve states are
obtained through kinematic relations:
2 3 2 3
dR 6 x_ 7 6 u 77
6 7 6
= 4 y_ 5 = RV B 4 v 5 (2.22)
dt E
z_ E w B
with RV B = T .
RBV
For the Euler angular rates holds:
2 3
dE 6 p7
dt = RB 64 q 75 (2.23)
r B
with the transformation matrix:
2 3
6 1 sin tan cos tan 77
RB = 64 0 cos sin 5 (2.24)
0 sin = cos cos = cos
respect to the body axes ([WXB WY B WZB ]T ) and another set for wind generated in FE
([WXE WY E WZE ]T ). The gust models used in the simulation software generate gusts
w.r.t. the body axes while the constant windeld and windshear are generated in FE .
The airspeed components along the body axes can be calculated from:
2 3 2 3 2 3
6 u 7 6 WXB 77 + R 66 WXE 77
6 7 6
VaB = 4 v 5 + 4 WY B 5 BV 4 WY E 5 (2.26)
w B WZB WZE
CL
wb
xB
CoG CL
CD t
α
Va ε δT
AC
AC t
αt
q=0 lt
CL , CD
1.5
0.5 CD
0
−10 −5 0 5 10 15 20
(deg)
Fig.2.8 Curves of CL and CD vs.
to the lift coecient, CLt can be calculated from:
CLt = CLt t (2.29)
where CLt is the derivative of CLt with respect to t , the angle of attack of the horizontal
tailplane:
t = " + T + kq Vqlt (2.30)
A
with the downwash angle:
d" ( )
" = d (2.31)
0
For pitch control of the RCAM, the whole tailplane is rotated (T ); this de
ection directly
aects t . Although the original aircraft on which the RCAM is based, is certainly con-
trolled using the elevator, no aerodynamic data of this device are availabley.
The last term of equation 2.30 is introduced by the rotation of the aircraft about the yB -
axis, which makes that at the stabilizer an extra vertical airspeed component is felt, equal
to q lt . The factor lt is the distance between the aerodynamic centers of the wing-body
combination and the stabilizer. This is equivalent to a contribution to the angle of attack
of q lt =VA . The constant kq is a correction factor to take into account that during rotation
a similar change in is felt along the whole aircraft, aecting the total lift and the lift
distribution over the aircraft.
The in
uence of the tailplane de
ection on the lift coecient (CLT ) can for the RCAM
model be written as:
CLT = CLt St Vt 2 (2.33)
S V
The contribution to the lift coecient due to a pitch rate is modeled with:
2
C qc = k C St Vt q lt
Lq V q Lt S V VA (2.34)
A
Thus:
2
CLq = kq CLt SStclt VVt (2.35)
2.5
2
CL
1.5
0.5
f = 32:5 deg, gear down
0
0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6
CD
Fig.2.9 The lift-drag polar
The contribution of the stabilizer to the drag coecient is neglected. However, if the
aircraft is trimmed, the stabilizer has a secondary eect on the total aircraft drag. If the
equilibrium state requires a negative tailplane de
ection, the device will have a negative
contribution to the total lift. This has to be compensated by the wing, so that a higher
Version: 1
17 Date: April 4, 1997
GARTEUR/TP-088-27
angle of attack is required. However, this also means a higher drag coecient, as can
easily be seen from equation 2.37.
In practice this extra drag is very important, because over a longer period of time it might
cost a considerable amount of fuel. This is one of the reasons why the application of
relaxed stability is considered for civil aircraft. By moving the location of the center of
gravity to a more aft position, the angle of incidence of the stabilizer will become positive,
or at least less negative, resulting in less drag. On the other hand, it is also possible
to decrease the size of tail plane in the design of an aircraft (see [11]). The decreased
longitudinal stability needs to be compensated by application of active controls.
The lateral component of the aerodynamic force vector, CY , is calculated from:
CY = CY + CYR R (2.38)
zM
xn fix
x cg
x ac wb lt
xB
CoG NP AC t
z cg
Ref. Point AC wb xM
z APT1,2
zB
These moment derivatives are mainly aected by the center of gravity location. In ap-
pendix A the eect of center of gravity shifts on the moment derivatives and static stability
is discussed.
2.1.5 Thrust contribution to forces and moments
The aircraft is powered by two engines. The thrust Ti of each of the engines is calculated
by:
Ti = THi m120 g; i = 1; 2 (2.47)
g = 9:81 m=s2 is the gravity acceleration, m120 = 120000 kg, a mean value for the aircraft
mass. The factor m120 g is only a scale factor in order to get reasonable values for throttle
settings. The thrust is aligned with the xB -axis; the components along the yB - and zB -axis
are zero. The vector of the thrust contributions to the force equations becomes:
2 3 2 3
66 FxT 77 66 T1 + T2 77
4 FyT 5 =4 0 5 (2.48)
FzT B 0 B
The contribution of the thrust of each of the engines to the moment equations can be
calculated from:
2 3 2 3
6 xcg xAPTi 77 66 Ti 77
6
MTi = 4 ycg + yAPTi 5 4 0 5 (2.49)
zcg zAPTi B 0 B
xcg , ycg , zcg , xAPTi, yAPTi and zAPTi (i = 1; 2) can be found in table 2.1.
2.2 The actuator and engine models
Control surface de
ections are generated using actuators that have limited bandwidth,
limited de
ections and limited de
ection rates. The engine dynamics have identical lim-
itations. All control devices are modeled using rst order models. These are described
below:
Aerodynamic model
h = 0 m, f = 32:5 deg
gear down
Non-coecients
k 0.07
kq 1.3
0 -0.21 rad
d" 0.25
d2
Vt 1.0
V2
Force equations
CD0 0.13 CL0 0.45 {
{ CL 6.07 {
{ CLwb 5.5 {
{ CLt 3.1 {
{ CLq 3.73 {
{ CLT 0.76 {
{ { CY -1.6
{ { CYR 0.24
Moment equations
{ Cm0 -0.015 {
{ Cm -2.15 {
Cl -0.205 { Cn 0.147
{ { Cn -0.560
Clp -0.473 { Cnp 0.073
{ Cmq -14.0 {
Clr 0.215 { Cnr -0.495
ClA -0.088 { CnA 0.0
{ CmT -2.87 {
ClR 0.032 { CnR -0.092
Note: all coecients w.r.t. aerodyn. center,
force coe. in FS , moment coe. in FB .
account. The limit of 10 degrees for the throttles means that the maximum T=W -ratio is
approximately 10=57:3 2 = 0:35, since the throttle de
ections are scaled with a nominal
weight, m120 g, see section 2.1.5.
In the case of an engine failure the throttle setting of the failed engine reduces to THi =
0:5 deg (i = 1; 2) with rst order system dynamics given by the transfer function:
Heng:failure(s) = 3:3s1+ 1 (2.54)
This transfer function is chosen such that the maximum rate of the decreasing throttle
setting starting from a trim position (4 5 deg) is somewhat less than 1:6 deg=s. When
a controller is used to compensate automatically by increasing the throttle position of the
working engine, this can be done at a rate just below the maximum allowed throttle rate.
2.3 The wind models
In order to make both open loop and closed loop nonlinear simulations more realistic,
the nonlinear model is equiped with inputs for wind and turbulence. In this section the
models are described that will be used to generate these wind inputs.
2.3.1 Constant wind
The rst group of three wind inputs to the aircraft model enter the equations of motion
in the inertial reference system. The windspeeds are simply subtracted from the inertial
speed, resulting in the airspeed. The constant wind needs no further modeling, only a
realistic windspeed with repect to the earth should be set in a desired direction.
Before a simulation is started where a constant wind eld is applied, it is necessary to trim
the aircraft for this wind. With the trim routine it is possible to nd equilibrium with the
Version: 1
Date: April 4, 1997 22
GARTEUR/TP-088-27
correct crab angle, ground speed etc. For controller evaluation in time simulations, wind
inputs of 10{20 m=s are applied.
2.3.2 Turbulence models
Turbulence is a stochastic process that can be dened by velocity spectra. Commonly
used velocity spectra for turbulence modeling are the Dryden spectra. These spectra are
based on a number of assumptions; the most important are:
Atmospheric turbulence is a random process with Gaussian distribution
Atmospheric turbulence is a stationary process, hence the distribution of wind ve-
locities in a windeld is considered to be independent of time
Atmospheric turbulence is homogeneous along the
ightpath
Atmospheric turbulence is an isentropic process
For an aircraft
ying at a speed V through a `frozen' turbulence eld with a spatial
frequency of
rad/m, the circular frequency of the turbulence can be calculated as:
! = V
rad/s (2.55)
With this, the spectra of the encountered gust velocities can be described as follows:
2L
ug (!) = u2g Vug (1 + (L1 ! )2 )
ug V
L 1 + 3( L ! )2
vg (!) = v2g Vvg (1 + (L vg!V)2 )2 (2.56)
vg V
L 1 + 3(L ! )2
wg (!) = w2 g Vwg (1 + (L wg!V)2 )2
wg V
A Bode-plot of these power spectral densities can be found in gure 2.11. For this plot
is taken: Lug = Lvg = Lwg = Lg = 305 m and ug = vg = wg = = 1:5 m=s. A
turbulence scale length (Lug , Lvg , Lwg ) can be interpreted as a measure of spacial extent
of signicant correlation. The turbulence scale lengths and turbulence standard deviations
ug , vg , wg are dependent on altitude and atmospheric conditions. As an indicator for
the atmospheric conditions it is possible to take the wind speed at 20 ft above the ground
(W20 ). For `moderate' conditions, W20 = 15:4 m/s (30 kts) is selected. The turbulence
standard deviation wg is then given as follows:
wg = 0:1W20 (2.57)
ug and vg are assumed to be functions of wg and the altitude h. For h < 305 m
(1000 ft):
ug = vg = (0:177 +0w:0027
g
h)0:4 (2.58)
Version: 1
23 Date: April 4, 1997
GARTEUR/TP-088-27
10
2 lon (|{), lat/vert (- - - -)
1
10
0
10
ug ; wg
−1
10
−2
10
−3
10
! (rad/s)
Fig.2.11 Power spectral densities of Dryden lters
and for h > 305 m (1000 ft):
ug = vg = wg (2.59)
The turbulence scale lengths Lug , Lvg and Lwg are calculated as a function of altitude: for
3 < h < 305 m:
Lug = Lvg = (0:177 + 0h:0027h)1:2 (2.60)
Lwg = h (2.61)
and for h > 305 m is taken:
Lug = Lvg = Lwg = 305 m (2.62)
With this procedure, the gust velocities ug , vg and wg are dened in the stability reference
frame. However, as an approximation the RCAM inputs WXB , WY B and WZB are used.
To simulate turbulence, white noise is ltered through forming lters. A great advantage
of the Dryden lters is that these lters can easily be derived from equation 2.56. As
an example, the transfer function of the lter for simulating the gust velocity wg will be
considered.
Given white noise w, the spectrum of wg can be obtained as:
wg = jHwg w (!)j2 w (2.63)
Where w = 1, and Hwg w (!) is the frequency response function of the forming lter.
Therefore,
3(Lwg V! )2
w2 g LVwg (11 +
+ (Lwg V! )2 )2 = jHwg w (!)j = Hwg w (!)Hwg w ( !)
2 (2.64)
Version: 1
Date: April 4, 1997 24
GARTEUR/TP-088-27
To obtain a stable and minimum phase lter, the following frequency response function is
selected:
s p Lwg
Hwg w (!) = wg VL wg 1 + 3 V j!
(2.65)
(1 + LVwg j!)2
Replacing the variable j! by s, the following transfer function is obtained:
s p
Hwg w (s) = wg Lwg 1 + 3 LVwg s (2.66)
V (1 + LVwg s)2
The transfer function for generating vg is equivalent. The transfer function for generating
ug can be found as:
s
Hug w (s) = ug 2LVug 1Lug (2.67)
1+ V s
The three forming lters have been implemented in Simulink-blocks. It is important
to note that for correct application of these lters the white noise inputs need to be
independent.
As can be seen from 2.2, the RCAM aerodynamic model is relatively simple. An important
limitation is that no gust penetration eects have been modeled. Furthermore, the gust
windeld is considered to be two dimensional. Due to these simplications the turbulence
windspeeds only aect the airspeed vector Va at the aerodynamic center. In order to
obtain more realistic gust responses of the aircraft model, a more complex aerodynamic
model is needed. With the available data of the aircraft it is possible to implement the
gust penetration eect and to model eects of three dimensional turbulence wind elds in
a future release of the model.
For a more detailed discussion on turbulence modeling, the reader is referred to for example
[6].
2.3.3 Windshear model
For the simulation of windshear, an existing model is used from [16], which is 2-dimensional
and symmetrical with respect to the y{z plane. It is basically a combination of two counter
rotating vortices. In order to obtain zero vertical windpeeds on the ground, there are also
two virtual vortices mirrored under the earth surface, see gure 2.12.
The model is implemented as a look-up table of positions x and z in FE with corresponding
windspeeds wx and wz . The symmetry axis can be placed at a desired position on the
trajectory to be
own. The windspeeds WXE and WZE at a location in the windeld are
calculated by interpolating using the x and z coordinates of the aircraft with respect to
the inertial system in the look-up table. The windeld is plotted in gure 2.13.
When
ying through the wind eld, a registration can be created of the encountered wind
speeds. An example is given in gure 2.14. This registration was made during a simulated
Version: 1
25 Date: April 4, 1997
GARTEUR/TP-088-27
symmetry axis
vortices
ground
imag. vortices
symmetry axis
2000
1800
1600
1400
altitude (m)
1200
1000
800
600
400
200
0
−2000 −1500 −1000 −500 0 500 1000 1500 2000
x-position (m)
Fig.2.13 Windeld due to windshear and downburst
Version: 1
Date: April 4, 1997 26
GARTEUR/TP-088-27
10
WXE
8 WZE
wind velocities WXE and WZE [m/s]
6
−2
−4
−6
−8
−10
−11000−10000−9000 −8000 −7000 −6000 −5000 −4000 −3000 −2000 −1000
x−position (XE) [m]
The RCAM simulation environment consists of the nonlinear state equations, the actuators
and engine dynamics, the wind and gust models and the controller that has to be designed.
This environment has been implemented in Simulink, see gure 2.15. There is one input
for references (REFERENCES-block). Other inputs can be set to either 0 or 1, in order
to command an engine to fail, or to add constant wind or gust.
On the right hand side there are six blocks that write simulation outputs to the Matlab
workspace.
traject
reference
signals
Controls
controls
controller control−
outputs inputs
REFERENCES sim
ouputs for
simulation
CONTROLLER
Engine failure
(1/0)
lon
ACTUATORS
longitudinal
0 measurements
Const. windfield
(1/0)
lat
0 lateral
Turbulence measurements
WIND &
(1/0) TURBULENCE
AIRCRAFT
wind
total wind−
inputs
The contents of the sub-blocks in gure 2.15 are explained in more detail in appendix B.
In section 2.1.3 the nonlinear state equations have been derived. These have been im-
plemented in the AIRCRAFT sub-block in gure 2.15. The twelve states used in the
equations of motion are summarized in table 2.4.
As already mentioned in section 2.1.3, the control inputs for the nonlinear state equations
consist of throttles, elevator, rudder and ailerons. There are six more inputs for wind and
turbulence. The rst group of three wind inputs is in the inertial reference system. The
Version: 1
Date: April 4, 1997 28
GARTEUR/TP-088-27
second group is for turbulence, it is dened with repect to the body axes. The inputs for
the simulation model are given in table 2.5.
For the RCAM simulation model three groups of outputs have been dened, see gure 2.15.
The rst group consists of longitudinal measurements, the second group consists of the
lateral measurements. The longitudinal and lateral measurements may be used as inputs
for the controller. The third group consists of simulation outputs that may only be used
for evaluation purposes. The model outputs are given in table 2.6. All inputs, outputs
and states have SI units.
The nonlinear RCAM model with the dened states, inputs and outputs has been imple-
mented in Dymola, a software package for object-oriented modeling [8, 19, 9]. Dymola
features automatic code-generation for several simulation environments. For RCAM code
was generated for use with Matlab/Simulink, in both standard Matlab-format and C-
format (CMEX). For more information, refer to appendix F in [3].
Trimming
Before linearizing the model or starting a simulation, the aircraft needs to be trimmed. To
nd the trim settings of the controls, the RCAM software-package is equiped with a trim
routine, in which constraints and conditions at the equilibrium point can be specied. For
nding equilibrium the routine makes use of the existing Simulink trim functions, based
on constrained optimization routines.
The aircraft model can be trimmed in symmetric
ight in the undisturbed atmosphere (no
Version: 1
29 Date: April 4, 1997
GARTEUR/TP-088-27
linearization
For analysis and controller design purposes it is necessary to obtain linearizations of the
nonlinear aircraft model at desired operating conditions. In many text books on aircraft
stability and control linearized models are derived analytically, see for example [10, 6,
21]. Describing the analytical linearizations goes beyond the scope of this chapter. The
linear models of the RCAM can be obtained numerically using the linearization routine
provided with Simulink. In this way a linearization of the trimmed nonlinear model can
be calculated for small perturbations around the equilibrium point found using the trim
routine. Some more details about linearization can be found in section B.2.
Version: 1
Date: April 4, 1997 30
GARTEUR/TP-088-27
xB
αw L
xA
RA
q, M
VA
θ αf Vw α
xF
αw
T
γ V γa
xV
D
β = βw = φ = 0
p=r=0 W
zB zA zF zV
CD
CL
γgl = - γ
xv C W= - CR Lift-drag
polar
xa VA CD
zv za
are parallel to the forces L and D induced by dynamic pressure on the reference surface
S . This implies that the axes system of the lift-drag polar and the aerodynamic (or wind)
Version: 1
33 Date: April 4, 1997
GARTEUR/TP-088-27
axes system are the same. The resulting aerodynamic force coecient is CR = 1=2RVAA2 S ,
which is parallel to the aerodynamic force vector RA .
Equilibrium in unpowered
ight is only satised if:
RA + W = 0 (3.5)
or, in terms of coecients, CR must be equal to CW (gure 3.2), where CW is the coecient
for the aircraft weight: CW = 1=2W VA2 S . The weight vector determines the direction of the
zV -axis, so that the angle between the zV axis and the zF -axis (
) is determined. In steady
unpowered
ight without wind the
ightpath angle will thus be equal to the glide angle
gl (always negative), see gure 3.2:
tan
gl = CCD = D L
gl (3.6)
L
Substitution in equation 3.4 for small
ightpath angles and with the assumption that
nz
1, gives:
T =
+
+ + V_ (3.7)
W gl w g
With the aerodynamic
ightpath dened by
a =
+ w , the equation can be written as
follows:
T =
+
+ V_ (3.8)
W gl a g
This expression shows the relation between the thrust input T , the aerodynamic state
gl
and what is available for manoeuvring:
a and V_ . By writing it in terms of pure
ightpath
quantities, this equation has become very simple.
3.1.2 Energy considerations and the Korhammer diagram
In this section the aircraft motions will be considered from an energetical point of view.
If rotational energy can be neglected, the total energy can be written as:
E = Ekin + Epot = m2 V 2 + mgh (3.9)
and dividing by the aircraft weight, the specic energy is obtained:
E = V2 +h =h (3.10)
W 2g E
which is equal to the energy height hE .
hE is the height that the aircraft can achieve with the available energy level, when all
kinetic energy is transferred to potential energy. A dual denition can be made for the
energy speed VE .
The rate of change of the energy height can be written as (it is assumed that the aircraft
mass is constant):
E_ = V_ V + h_ = V ( V_ + sin
)
h_ E = W (3.11)
g g
Version: 1
Date: April 4, 1997 34
GARTEUR/TP-088-27
The right hand side of this equation is also called the specic surplus energy, which is
often related to the climb performance. Therefore, similarly to the
ightpath angle
, the
energy
ightpath angle
E is dened as:
h_ E = V sin
E (3.12)
where
_ E_
sin
E = Vg + sin
= WV (3.13)
This quantity is a measure for the energy surplus with respect to the aircraft weight and it
is the total amount of energy that can be extracted from the system during a manoeuvre.
This term is often referred to as potential energy
ightpath angle. However,
E is a
measure for the total energy and not only for the potential energy.
With equation 3.4 the energy rate can be written as:
T + CD + n = V T V (
)L = P P
E_ = WV W CL w z
w gl a r (3.14)
The right hand side of this equation describes the available power provided by thrust (Pa )
and the power consumed by the aerodynamic force (Pr ). The dierence between these
quantities, the surplus power, is available for manoeuvring.
This energy dierence can be considered with respect to any energy level. Altitude is
mostly dened with respect to the earth surface, so that the potential energy equals mgh.
The aerodynamic state of the aircraft is related to the airspeed VA rather than the inertial
speed V . As a control variable the airspeed is also more important and therefore plays
a key role in the analysis of aircraft motion. For these reasons in literature the kinetic
energy is mostly related to the airspeed: Eakin = 21 mVA2 . In this respect a mixed energy
quantity, the aero-kinetic energy Ea , can be dened as:
Ea = Eakin + Epot = m2 VA2 + mgh (3.15)
However, in this analysis the denition of the total system energy according to equation 3.9
is used.
Both force equilibrium and energy management can be visualized in the so called
Korhammer-diagram. An example of this diagram is shown in gure 3.3. In a Korhammer
diagram all forces and accelerations acting in the aircraft center of gravity can directly be
related to the CL CD polar. In order to improve the visibility of some force components,
the angles are exaggerated. The only quantities that are drawn in the right proportions
with respect to each other, are the components of equation 3.1.
The diagram is self explanatory, but to understand the dynamics involved it is necessary
to proceed in a prescribed fashion. In rst instance the aerodynamic trim point on the
lift-drag polar is chosen, this determines the value of the lift coecient CL and also the
coecient CR . The direction of RA has now been determined. With the given wind angle
w and
ightpath angle
the directions of the
ightpath axes and the vehicle carried
Version: 1
35 Date: April 4, 1997
GARTEUR/TP-088-27
CL
α
xT
xa VA
CR CD
α + iT
Vw αw
xf V γ
xv
RA
W
γa
γgl
αw γ
α f + iT T
xf W sin( γ )
RAsin( α w+ γ gl
)
zv
zb zf
za
Some further system characteristics can be deduced geometrically from this diagram. In
the case of constant thrust, the sum of the glidepath angle and aerodynamic
ightpath
angle remains constant when a new trim point is chosen. This can also directly be seen
from equation 3.8 (where
gl is positive and V_ = 0). The aerodynamic
ightpath angle
a , which characterizes the
ightpath with respect to the airmass, does not depend on
the wind but depends on the dierence between thrust and drag. A stationary air motion
has an eect on forces and energy management only through the wind angle w . Lift is
aected by this angle such that a positive w (down wind) results in a smaller
in order
to keep
a constant. A part of the available power is thus absorbed by the wind. Finally,
thrust compensates the energy loss induced aerodynamically by w
gl and provides
the additional power that is required for a specied
ightpath angle
(specic potential
energy rate).
Aircraft performance is often studied through power analysis. This power analysis is
carried out with respect to the
ightpath axes. The available power provided by thrust is
V T cos (f + iT ). The resulting potential energy increase is V W sin
while loss of power
due to aerodynamics is V RA sin(w
gl ). If the thrust is aligned with the horizontal the
Version: 1
Date: April 4, 1997 36
GARTEUR/TP-088-27
For an accelerated
ight the force equilibrium has to be extended with the acceleration
term ( ddtV )g , as shown in gure 3.4. This term results in a kinetic energy rate of E_ kin =
mV V_ and a potential energy rate of E_ pot = WV sin
. Further, the total energy rate
or power surplus becomes: W sin
E = W sin
+ mV_ . The energy
ightpath angle can
be deduced through rotation of the weight vector until the
ightpath contribution of the
acceleration term is compensated. The energy angle is the maximum
ightpath angle that
can be achieved at the current thrust setting.
In the same diagram it is also possible to show the acceleration contributions to external
forces in the
ightpath axis system. For the horizontal and vertical contributions we have:
maxF = Wsin
E = Wnx
(3.16)
mazF = Wnz
both vectors determine the direction of the sensed acceleration (virtual gravity accelera-
tion) which takes a value with respect to the zF axis:
tan = axF = sin
E
azF nz
(3.17)
All the quantities as mentioned here which characterize the aircraft performance are now
put into one framework describing and providing insight in the physical process of ma-
noeuvred
ight. This means that in the Korhammer diagram all gliding, accelerated or
decelerated
ights with or without wind can be represented, provided that the force and
speed vectors remain within the aircraft symmetry plane (no lateral eects).
xa
xa
ρ
W
γ γ γ
gl αw gl αw γ
γa
RA γa RA
W γE
T m a zf T
= W n tz
m dV αk+ ιT
dt g m dV
dt g
x .
f
mV
W sin γ
sin γ E γ gl)
R Asin( α w W sin γ E
+
x m a xf= W
f
za zf zv za zf
CL CL
δf
α α
CR
CR
xa VA CD xa VA CD
RA RA
W W
γ − γ gl γ − γ gl
T T
za zv zv za
Fig.3.5 Eect of changing aerodynamic trim point in Korhammer diagram: angle of attack
and
ap de
ection
Backside operation
If the angle of attack is further increased until the total aerodynamic force coecient
is tangent to the polar (see gure 3.6), then also the point is reached where the glide
path angle has achieved its minimal value in absolute sense. The trim speed at which this
operating point is achieved is called minimum drag speed. This trim point could also be
called the speed for minimum glide angle and is aerodynamically the most ecient point
for cruising. The slope of the polar in this situation is equal to the ratio CCDL , therefore:
CD = @CD = CD (3.22)
CL @CL CL
In this case the glide angle remains approximately constant for small variations in angle of
attack. This implies that at constant thrust the
ightpath angle also becomes independent
of the angle of attack. An elevator de
ection then implies, through its eect on CR , only
a change in airspeed. If the angle of attack is further increased, the glide angle starts to
increase again. However, variations in angle of attack for this region on the polar imply
CD > CD . Since the total force coecient grows, the speed will decrease further. This
CL CL
speed region is called "below the minimum drag speed", where the eect of angle of attack
on the
ightpath angle is reversed.
The increase of the angle of attack induces an increase in the glide angle, which at a
constant thrust can only be compensated by a decrease in
ightpath angle. This means
that a negative elevator de
ection (up) results in an unexpected decrease of the
ightpath
angle. At the same time CR increases so that the speed decreases, as in the normal
situation.
The origin of this phenomenon can directly be seen from the Korhammer diagram. The
(constant) thrust vector can not compensate the overproportional growth of the drag
Version: 1
39 Date: April 4, 1997
GARTEUR/TP-088-27
CL CD CL
CD
CL
CL
α
CD
α CL
CD
CL
γ gl
γ gl
min
CR
CR
xa VA CD xa VA CD
RA W
RA W
γ max γ
T max T
zv za za zv
Throttle input
To analyse the eect of thrust changes the equation of motion along the
ightpath can be
written in the form:
T = CD + WZE n + sin
+ V_ (3.23)
W C V z
g
L A
which reveals that the required thrust is a function o:
the vertical load factor nz
the aerodynamic cost due to glide angle
gl
the vertical component of the windeld WZE
Version: 1
Date: April 4, 1997 40
GARTEUR/TP-088-27
γ gl
γ gl
xT
xa
α + iT
xa
W
RA
W
RA
∆γ
∆γ
T
T
∆ iT
θ + iT
xv
za zv za zv
returns to the original trim point on the polar. Therefore the glide angle
gl and the
airspeed VA remain constant. Under this condition the steady state response of speed to
thrust is:
VAss = Vss = 0 (3.27)
Thus only the
ightpath angle will be aected:
= T
ss W (3.28)
This means that in steady state all energy input by thrust is completely transformed into
potential energy, while airspeed remains constant.
In the case that T sin(f + iT ) is taken into account, a slight modication has to be made.
In the Korhammer diagram in gure 3.7 the glide angle
gl remains xed, therefore also
the direction of the resulting aerodynamic force R remains constant. Mostly, the thrust
has also a small contribution in the negative zV -direction, thus upwards when + iT > 0.
Since the gravity contribution is constant, a thrust increase not only results in \opening"
the Force triangle, but also results in a shorter vector R, and therefore a smaller airspeed
VA . This system response is opposite to what one may expect. Furthermore, it is also
opposite to the initial response of the system.
The magnitude of the total aerodynamic force vector, and therefore also the airspeed, will
be invariant to small thrust increments only if the thrust vector is perpendicular to the
gravity vector. Only in that case thrust can be changed while RA remains unaltered,
resulting only in opening the force polygone and thus in increasing the
ightpath angle.
Another eect of thrust is when the thrust vector induces a contribution to the pitching
moment equilibrium. In a trim point this moment due to thrust is compensated by a
steady state aerodynamic moment (induced by ):
1 V 2 SC
2 A m0 zAPT T0 = 0 (3.29)
If there is a thrust increment new moment equilibrium is achieved only aerodynamically
by:
1 V 2 SC z T = 0 (3.30)
2 A m APT
which means that
ss = + 1 zAPT T (3.31)
2 VA SCm
2
This means that a positive thrust moment induces an increase in angle of attack , there-
fore an increase in CR and thus a decrease in the airspeed VA . As a consequence, when
ying on the right side of the power curve, the glide angle
gl decreases (in absolute sense).
The steady state airspeed change due to a thrust increment remains small. It will decrease
when the thrust vector induces a pitch up moment. The initial response to a stepwise
thrust increment is:
q_ = zAPTI T (3.32)
y
Version: 1
Date: April 4, 1997 42
GARTEUR/TP-088-27
Suppose that it is possible to change the thrust vector by means of iT . In that case both
force and moment equilibrium are aected. A positive iT results in a decrease of the
total aerodynamic force and thus in an decrease of the airspeed. Furthermore, the weight
vector will rotate to compensate for the acceleration contribution, so that the
ightpath
angle also increases.
Version: 1
43 Date: April 4, 1997
GARTEUR/TP-088-27
At a given altitude (given air density ) and a given aircraft weight, the airspeed, drag and
required power are a function of the angle of attack and thus a function of the equilibrium
point on the lift-drag polar. The rate of climb and
ightpath angle also depend on the
engine control setting.
From the equations minimum values can be calculated for the airspeed (at CLmax ), for
the drag (at (CL =CD )max ) and for the required power (at CL3 =CD2 max ). The minimum
airspeed, or stall speed, is important for safety reasons. In horizontal
ight the longest
possible range is achieved when
ying at the minimum drag speed, while maximum en-
durance
ight requires a minimum fuel
ow and therefore minimum required power.
In gure 3.8 these quantities are easiliy related to the lift-drag polar.
3 3
C CL 3
2.5 2.5
CDL CD 2
minimum Pr
2 2
minimum D
CL
1.5 1.5
1 1
0.5 0.5
CL 3
20 40 60 80CD 2100
0 0
0 0.2
CD
0.4 0 2 4 6 8
CL 10
CD
Fig.3.8 Performance related to the lift-drag polar
For the RCAM model the parameters of the polar are (table 2.2):
CD0 = 0:13
CL0 = 0:45 (3.41)
k = 0:07
The total drag can be calculated from:
D = CD0 12 Vair 2 S + k (C 1 2S=D +D
L CL0 )2 2 Vair 0 i (3.42)
D0 is the zero lift drag and Di is the induced drag of the aircraft. All minima for parameters
to be calculated are valid for steady state. A very important constraint is therefore vertical
equilibrium: L W . Substitution gives:
CL = W (3.43)
1 V 2 S
2 air
The minima for parameters can now be calculated analytically.
The maximum angle of attack for the RCAM is 18 deg. Using the simple parabola ap-
proximation that was in the original model, this corresponds to a maximum lift coecient
of CLmax = 3, which is quite high. For this reason the lift-drag relation was modied, so
that the polar becomes
at for higher angles of attack. The original and modied polars
are compared in gure 3.9. As the maximum lift coecient a value of CLmax = 2:75 has
been chosen.
original
3.5
2.5
modied
CL
1.5
0.5
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
CD
Fig.3.9 Modication of the lift-drag polar
The speed for minimum drag corresponds to the condition dD=dVA = 0. Dierentiating
equation 3.42 with respect to VA gives (with substitution of eq. 3.43):
v
u s
u W 2 k
VDmin =t S CD0 + kCL2 0 (3.45)
For analysis of the RCAM model the parabolic estimation is thus still good enough.
The minimum required power can be calculated by substitutions:
CD = 0:13 + 0:07(2:4 0:45)2 = 0:40 (3.54)
Prmin = 0:40 21 1:225 553 260 = 10:6 103 kW
5
x 10
4.5
Max. Thrust
4
150000 kg
2.5
2 120000 kg
1.5
100000 kg
1
40 50 60 70 80 90 100 110 120 130
Vtrim (m/s)
Fig.3.10 Performance diagrams for dierent masses (forces)
4
x 10
5.5
xcg = 0.23c
5 h = 0m
100000 kg ||- Pa
4.5 120000 kg - - - -
150000 kg .......
4
VRCmax
Pr
Pa , Pr (kW)
3.5
3
VDmin
V
VstallPrmin
2.5
1.5
0.5
40 50 60 70 80 90 100 110 120 130
Vtrim (m/s)
Fig.3.11 RCAM performance diagrams (power)
higher speeds requires more power than available from the power plant.
For the rate of climb holds (equation 3.36):
RC = Pa W Pr (3.55)
For constant weight the rate of climb will be maximum where Pa Pr is maximum, thus
where the Pa and Pr lines in gure 3.11 are parallel. From this gure can be seen that
for a weight of 117:7 104 N (120000 kg) the RCmax occurs at a speed of 89 m=s. The
power surplus is at that point approximately 1:9 104 kW , so that:
1:9 107 = 16 m=s
RCmax = 120000 (3.56)
9:81
The maximum
ightpath angle
max can be deduced from gure 3.10. From equation 3.35
can be seen that:
T D
= arcsin W (3.57)
The term between the brackets is often referred to as the specic excess power. For a
given aircraft weight the maximum
ightpath angle occurs when T D is maximum. In
the case of the RCAM maximum thrust is assumed to be invariant to the airspeed. The
maximum
ightpath angle is therefore at the minimum-drag speed. This is illustrated in
gure 3.12. In equation 3.49 the minimum drag appeared to be 162 103 N . The maximum
available thrust is 411 103 N , so that for a weight of 117:7 104 N the specic excess power
equals:
!
T D = arcsin 411 103 162 103 = 0:21 (3.58)
W 117:7 104
Version: 1
49 Date: April 4, 1997
GARTEUR/TP-088-27
CL CD
CL
CD
CL
γ gl
min
CR
xa VA CD
RA
W
γ max
T max
zv za
13
12
11
10
9
(deg)
5 xcg = 0.23c
h = 0m
4 100000 kg ||-
120000 kg - - - - V
max
3 150000 kg .......
40 50 60 70 80 90 100 110 120 130
Vtrim (m/s)
Fig.3.13 Maximum
ightpath angle vs speed
−6
−7
−8
−9
(deg)
−10
−11
−12
−13
xcg = 0.23c
−14 h = 0m
100000 kg ||-
−15 120000 kg - - - - V
min
150000 kg .......
−16
40 50 60 70 80 90 100 110 120 130
Vtrim (m/s)
Fig.3.14 Minimum
ightpath angle vs speed
Version: 1
51 Date: April 4, 1997
GARTEUR/TP-088-27
enough (no severe transients), this part of the trajectory can be
own without saturating
the engines at their minimum throttle setting.
In this section only static performance has been analysed. However, the thrust excess that
in equation 3.35 is used for a
ightpath angle, can of course also be used for accelerating
the aircraft. This does not seem to change the analysis basically, because it is in fact only
a matter of distribution of excess power. However, speed changes imply a change in the
aircraft state. For example, the equilibrium point on the lift drag polar is going to move
if the aircraft is accelerating. Therefore the analysis should be carried out more carefully
and for limited speed changes.
3.2.4 Analytical expressions for weight variations
In the gures of the previous section the curves were drawn for varying aircraft masses.
From the expressions in section 3.2.1 it is very easy to derive the in
uence of weight on
parameters like the minimum required power with the corresponding airspeeds. From
equation 3.39 the following ratios can be deduced for a condition (1) compared to a
condition (2) with both the same equilibrium point on the lift-drag polar.
s
VA2 = W2 (3.60)
VA1 W1
D2 = W2 (3.61)
D1 W1
s
Pr2 = W2 3 (3.62)
Pr1 W1
For a given point on the lift-drag polar (given angle of attack) the ratio of drag for the
two conditions can be expressed as a function of the corresponding trim speeds:
D2 = VA2 2 (3.63)
D1 VA1
The same function can easily be derived for the required power:
Pr2 = VA2 3 (3.64)
Pr1 VA1
For a given weight and altitude the three parameters only depend on the equilibrium
point on the lift drag polar. Therefore the minima for these quantities, which are also
only determined by the equilibrium point, vary in the same way as in equation 3.62.
If for example an aircraft mass of m = 150000 kg is compared to the nominal weight of
m = 120000 kg, the ratio of stall speeds becomes:
r
VA2 = 150000 = 1:1
VA1 120000 (3.65)
The ratio of minimum drags can be found from:
Dmin2 = 150000 = 1:25 (3.66)
Dmin1 120000
Version: 1
Date: April 4, 1997 52
GARTEUR/TP-088-27
The corresponding ratio of speeds is of course also 1.1. The ratio of minimum required
power levels:
s
Pr2 = 150000 3 = 1:4 (3.67)
Pr1 120000
These ratios can easily be veried in gures 3.10 and 3.11.
Version: 1
53 Date: April 4, 1997
GARTEUR/TP-088-27
.
W
g Vχ
W sin µ
µ
y
b
zb
W
From the simple derivation of the force equilibrium in the previous section immediately
follow three important bounds on turning performance:
The maximum thrust, Tmax
The maximum load factor allowed, jnz
jmax
The maximum lift coecient, CLmax
In the discussion of turning performance these will give the bounds that determine achiev-
able bank angles, turning radii and turning rates.
If there is vertical equilibrium, the expressions from the previous sections for VA , D and
Pr can be easily modied by substitution of equation 3.71 and 3.72:
s
VA = nz
W 2 1 (3.73)
S CL
D = nz
W CCD (3.74)
sL 2
Pr = nz
W nz
SW 2 CCD3 (3.75)
L
For a given weight (W ), bank angle (), wing area (S ) and altitude (), the minimum
values of these performance parameters are only determined by a certain equilibrium
point on the lift drag polar. For the stall speed that point is CLmax , for the minimum drag
(CL =CD )max and for the minimum required power the point corresponds to (CL3 =CD2 )max .
For this reason the optima for dierent bank angles are simply related by the load factors.
Comparison of two turns (1) and (2) gives:
s
Vstall 2 = nz
2 = r cos 1 (3.76)
Vstall 1 nz
1 cos 2
Dmin 2 = nz
2 = cos 1 (3.77)
Dmin 1 nz
1 cos 2
and:
v
u !3 s
Prmin 2 = u
t z
2 = cos 1 3
n (3.78)
Prmin 1 nz
1 cos 2
Version: 1
55 Date: April 4, 1997
GARTEUR/TP-088-27
4.5
Max. Thrust
4
= 60 deg
3.5
Tmax , D (N)
3
= 45 deg
= 25 deg
2.5
2 h=0m
xcg = 0.23c
= 0 deg
1.5
m = 120000 kg
1
50 60 70 80 90 100 110 120
Vtrim (m/s)
Fig.3.16 RCAM performance diagrams for dierent bank angles (force)
4
x 10
6
h = 0m
5
xcg = 0.23c
m = 120000 kg
4
Pa
Pa , Pr (kW)
3
= 60 deg
2 = 45 deg
= 25 deg
1
= 0 deg
0
50 60 70 80 90 100 110 120
Vtrim (m/s)
Fig.3.17 RCAM performance diagrams for dierent bank angles (power)
both functions 3.85 and 3.87 can be plotted in a diagram, see gure 3.18. In this gure the
area of achievable turn radii at given airspeeds is visible for a maximum lift coecient of
CLmax = 2:75 and a maximum load factor of nz
= 1:15. The latter is smaller than the
maximum structural load factor to allow for a margin, since the
ap setting is maximum.
Furthermore, the factor is limited for passenger comfort. The point where both curves
2500
1500
CLmax -limit
rf (m)
nz
-limit
1000
(rfmin )abs
500
0
0 20 40 60 80 100 120
VA (m/s)
Fig.3.18 Limitations on the turning radius
10
7 CLmax -limit
6 (_ max )abs
_ (deg/s)
5
nz
-limit
4
2 Achievable rates
1
0
0 20 40 60 80 100 120
VA (m/s)
Fig.3.19 Limitations on the turning rate
3.3.5 Climb performance in turns
In section 3.1.1 the following equation has been derived: (the acceleration is set to zero,
cos
1 and wind is not considered here)
sin
= WT + CD n T +
n (3.100)
CL z
W gl z
During a turn in the rst place the glide angle (
gl = CD =CL ) and the load factor
(nz
= 1= cos ) change. If the turn is performed in the horizontal plane, thrust has to
compensate for extra drag to maintain zero
ightpath angle.
As long as the throttles do not saturate, extra thrust is available for climbing. If the aircraft
goes from horizontal
ight into a turn, the available power for climbing is decreased by
the increased load factor and the changed glide angle. This can in principle be shown in
a Korhammer diagram, although a little cheating is necessary.
In a Korhammer diagram the aerodynamic forces are directly related to the lift-drag polar.
In horizontal
ight the aerodynamic, thrust and gravity forces are in the same vertical
plane, so that the symmetric force geometry can be simply completed in the diagram,
including other
ightpath parameters (
and V_ ). The problem is that the aerodynamic
axis system is rotated with respect to the
ightpath axes about the xF -axes over the bank
angle . The lift, drag and the thrust are (approximately) in the rotated xA -zA plane,
while gravity and the
ightpath terms remain in the vertical xF -zF plane.
The best way to modify the diagram is by rotating the vertical xF -zF plane into the xA-
zA plane. Equilibrium along the xF -axis can now be constructed directly (for stationary
ight):
T D = W sin
(3.101)
Version: 1
Date: April 4, 1997 60
GARTEUR/TP-088-27
CR CD
= turn
= hor. flight
RA W
γgl
γ
xf
T -n z γW
projection
z a z f’
this way has become somewhat complex and therefore loses most of its functionality. From
equation 3.100 the eect of the glide number on the climb angle is visible. If the aircraft
is
ying on the right side of the power curve, the necessary increased lift coecient will
result in a smaller glide number (in absolute sense). This slightly compensates decreased
ightpath performance due to the increased load factor. This is for example the case in
the diagram of gure 3.20. If the aircraft on the other hand is operating on the backside of
the powercurve, the glide number will increase. This is an extra performance decreasing
term.
Using equation 3.100 and the performance diagram 3.16 maximum
ight path angles have
been calculated for four bank angles as a function of the airspeed, see gure 3.21. Of
course the climb performance degrades considerably at higher bank angles.
In a steady co-ordinated turn expressions for rudder and aileron de
ections required for
equilibrium can be derived as functions of the yaw rate. The equilibrium equations for
Version: 1
61 Date: April 4, 1997
GARTEUR/TP-088-27
15
= 0 deg
10
= 25 deg
= 45 deg
(deg)
= 60 deg
0
h = 0m
xcg = 0.23c
120000 kg
−5
50 60 70 80 90 100 110 120 130
Vtrim (m/s)
Fig.3.21 Maximum
ightpath angle vs speed in turn
the co-ordinated turn can approximately be written as follows:
CY = Cy + CyR R + Cyr 2rbVA = 0
rb
Cn = Cn + CnR R + CnA A + Cnr 2VA = 0 (3.102)
rb
Cl = Cl + ClR R + ClA A + Clr 2VA = 0
In the RCAM model Cyr and CnA are zero. Solving for R and dierentiation gives:
2 3
Rrbss = CCnr 664
d CY
CYR
77
5 (3.103)
d 2VA nR CY CnR Cn
Using the values that can be found in table 2.2 can be calculated:
" #
d 0
ss = 0:092
R : 495 1 : 6 = 5:38 1:24 = 6:7 (3.104)
d 2rbVA 1:6 00::24092 0:118
Cn substituted here also depends on the angle of attack. For a for the RCAM common
value of 3 deg is chosen:
Cn = Cn0 + Cn == 0:147 0:5603=57:3 = 0:118
The expression between brackets in equation 3.103 is usually close to one, so that the
rudder de
ection is mainly used to overcome the yaw rate damping Cnr , which acts in
opposition to the established yaw rate. It is clear that the de
ection is inversely propor-
tional to the rudder control power, CnR . A positive yaw rate demands negative rudder,
so that the pilot will have to apply a rudder pedal de
ection toward the direction of the
turn.
Version: 1
Date: April 4, 1997 62
GARTEUR/TP-088-27
If the term between brackets in equation 3.103 is abbreviated to kY the steady state
rudder de
ection becomes:
Rss = CCnr kY 2rb
VA (3.105)
nR
Substitution in the yawing moment equation gives (CnA = 0):
Cn CnR CCnr kY 2rb
V + Cnr 2rb
VA = 0 (3.106)
nR A
Thus the sideslip angle is:
C (1 k )
= nr C Y 2rb VA (3.107)
n
And with substitution of RCAM aerodynamic derivatives and geometry (see table 2.2
and 2.1):
= 0:495(1 0:118
1:24) rb = 1:0 rb
2VA 2VA (3.108)
Thus for a rate I turn (r = 3 deg =s) at a speed of VA = 80 m=s:
Rss = 6:7 23 80
45 = 5:7 deg (3.109)
and:
= 1:0 23 80
45 = 0:8 deg (3.110)
In the problem formulation for the RCAM
ight control benchmark ([2]) no lateral ac-
celeration measurement is available. Since is small in a co-ordinated turn, one way to
obtain approximate co-ordination is to drive the sideslip angle to zero using an integrator.
This integration is necessary anyway in order to obtain the required zero sideslip in an
engine failure case. However, for the RCAM there is no solution for equation 3.102. From
the rst equation follows that the rudder de
ection is zero, since CYr = 0. On the other
hand, for the RCAM also CnA is zero, so that the yawing moment equilibrium cannot be
fullled. A solution can only be found if the bank angle is changed, so that the turn is no
longer
own co-ordinated. Therefore, a better way to obtain turn co-ordination is driving
the dierence tan V =g _ to zero.
Finally, the required aileron de
ection can be calculated from the roll moment equilibrium:
C + ClR R + Clr 2rbVA
A = l C (3.111)
lA
Thus:
0:205 0:8 + 0:032 5:7 + 0:215 23 80
45
A = 0:088 = 1:9 deg (3.112)
Version: 1
63 Date: April 4, 1997
GARTEUR/TP-088-27
VA at t = 0 and has been exchanged into V in steady state. Meanwhile the aircraft ac-
celeration leads to a loss of altitude at a constant energy level while the
ightpath angle
returns to its equilibrium point, because the force equilibrium returns to its initial state.
3.4.2 Vertical wind
The angle of attack due to wind is given by: w = WVZE A . The attitude angle is given by:
=
+ + w . Since and
can only respond to a stepwise change in WZE with a lag
in time, the initial response is only characterized by a change in angle of attack given by:
t=0 = w = WVZE (3.116)
A
The angle of attack, the airspeed VA and the glide angle return to their equilibrium values
and hence also the total aerodynamic force RA .
If the thrust is constant, the force polygone of gure 3.4 will return to its initial position.
This is only possible if the
ightpath speed vector V is rotated in opposite direction (in
earth co-ordinates) with respect to w . In steady state this results in:
ss = w = WVZE (3.117)
A
so that the total aerodynamic
ightpath angle
a remains unaltered. Since
a and both
remain unchanged, also the attitude angle does not change:
=
+ + w (3.118)
As can be seen from equation 3.117, the
ightpath angle decreases due to the vertical
wind, while
a does not change (steady state). If the aircraft has to be in equilibrium on
a pre-specied inertial
ightpath angle (
), trimming must be done for
a =
+ w . This
is described in section B.2.
3.4.3 Cross wind
A step in cross wind will result in a sideslip angle:
w = WVY E (3.119)
A
For the track angle holds:
= + + w (3.120)
Like in the case of vertical wind, the angles and can only respond with a time delay,
so that the sideslip angle will respond stepwisely:
t=0 = w (3.121)
Due to wind vane stability (Cn ) the aircraft course will change until has become zero,
thus f = w . During the transient motions the aircraft will roll due to Cl , so that the
lift vector is rotated. This will result in some decrease in altitude.
Version: 1
65 Date: April 4, 1997
GARTEUR/TP-088-27
The eect of windshear on the aircraft performance can be expressed by the F -factor.
This factor is derived from the eect of wind on the total energy of the aircraft relative to
the airmass, see for example [17] and [18].
In gure 3.22 some of the parameters used in this section are visible. For symmetric
ight
xB
θ WX E VA
WZ E α
Vw
T
V γa
xV γ
zB zV
with
a = , the elevation angle of the airspeed vector relative to the inertial axes.
Version: 1
67 Date: April 4, 1997
GARTEUR/TP-088-27
In equation 3.15 the expression for the aero-kinetic energy can be found. Dividing this
equation by the weight W , the specic aero-kinetic energy is found:
2
Eas = V2Ag + h (3.124)
Dierentiating this expression:
E_ as = V_A VA + h_
g (3.125)
Substituting equations 3.122 and 3.123 gives:
" #
_Es = VA T cos D W_ XE cos
a + W_ ZE sin
a WZE (3.126)
W g g V A
The rst term between the brackets is the specic excess power, as derived in section 3.2.1
(with cos 1). If there is no wind and all excess power is used to increase the potential
energy, thus V_ A = 0, equation 3.122 can be written as:
sin
= sin
= T cos D
a W (3.127)
The three wind terms in equation 3.126 describe the impact of the windshear on the
aircraft energy state; these may be combined into a single term, the F -factor:
_ _
F = WgXE cos
a WgZE sin
a + WVZE (3.128)
A
The eect of windshear on the aircraft performance can thus be expressed as an eective
reduction in specic excess power due to horizontal and vertical shears and downdrafts.
Regions where F is negative are performance increasing shears and regions where F is
positive are performance decreasing. For the RCAM the climb performance at an airspeed
of VA = 70 m=s and a mass of m = 120000 kg with maximum thrust setting is completely
cancelled by an F -factor of 0.21, as appeared in section 3.2.3, equation 3.58.
The F -factor along a reference
ightpath for the windshear model implemented in the
RCAM simulation environment (see section 2.3.3) is given in gure 3.23. The
ightpath
starts at h = 700 m and goes straight through the core of the wind eld with a
ightpath
angle of 3 deg. The encountered windspeeds can be found in gure 2.14.
As can be seen from this gure, the contribution of the downdraft is very important.
Since it is positive, it has a performance decreasing eect. The eect of the horizontal
shear is also signicant, mainly when the head wind suddenly becomes tail wind. Over a
short distance the the F -factor is larger than 0.21, thus if the aircraft is
own along this
ightpath, it will not be able to maintain both airspeed and
ightpath angle. If priority
is given to airspeed, the aircraft will deviate from its
ightpath and vice versa.
The F -factor gives an indication of how much thrust is required to
y safely through the
wind eld. For the required thrust to compensate for the downdraft can be written:
T = WZE (3.129)
W VA
Version: 1
Date: April 4, 1997 68
GARTEUR/TP-088-27
0.4
F−factor
0.35
WXEdot−contrib
WZEdot−contrib
0.3
WXE−contrib
0.25
0.2
F−factor
0.15
0.1
0.05
−0.05
−0.1
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
XE (m)
In [12] another relation between the aerodynamic
ightpath angle and the wind gradients
is derived.
Starting with equation 3.122 and assuming that the aircraft
ies at a constant airspeed in
a rectilinear
ight through a wind shear eld and assuming that the angles ,
and
a
are small, this expression can be written as follows:
_ _
0 = T W D
a WgXE + WgZE
a (3.130)
Since the airspeed is constant, the specic excess power in constant wind conditions would
be used for an aerodynamic
ightpath angle
a0 . The in
uence of the windshear can now
be characterized through:
a0 W_ XE
a = g (3.131)
1 W_ ZE
g
This equation shows how the aerodynamic
ightpath angle depends on the windgradients if
the airspeed does not vary.
a is constant when the gradients are constant. The
ightpath
angle is given by:
=
a w (3.132)
Version: 1
69 Date: April 4, 1997
GARTEUR/TP-088-27
(the eect of W_ ZE is assumed to be less important compared to the other terms, see also
gure 3.23). Like in equation 3.11 a vertical energy altitude rate, h_ Eaerr , is introduced;
this time with respect to the airmass. This term can be integrated to a specic energy
altitude error:
Z t V A
hEaerr = WZE + WXE d_ (3.138)
t0 g
If it is assumed that all quantities remain constant during the shear then:
hEaerr = WZE t + VgA W_ XE t = WZE t + VgA WXE (3.139)
thus the energy error is proportional to the duration of the vertical wind velocity and
amplitude of the horizontal windshear component. For a typical approach speed of VA =
70 m=s, a vertical wind WZE = 5 m=s and a windshear gradient of W_ XE = 4 m=s2 during
only 3 s, the loss in energy height will be HEA = 5 3 m + 70=9:81 4 3 m 100 m.
The main contribution in this loss is due to the gradient W_ XE .
Now it is necessary to evaluate how much energy reserve the aircraft has for windshear
compensation. A typical approach speed lies at 1:3 Vstall , while the critical airspeed is at
1:05Vstall . For a stall speed of 54 m=s the approach speed is then VA = 70 m=s. The speed
reserve will be equal to: VA = 70 1:05 54 = 13:3 m=s.
Version: 1
Date: April 4, 1997 70
GARTEUR/TP-088-27
800
700
600
500
altitude
400
300
200
100
0 −1000
0
5000 0
10000
1000
y distance
x distance
The requirement on an engine failure is that the aircraft remains controllable in steady
ights at low speeds, where this malfunction is most critical (e.g. during take o). Fur-
thermore, the aircraft should be able to maintain a desired heading with zero sideslip
angle. Based on these requirements it is necessary to study lateral aircraft dynamics in
order to obtain information like the minimum control airspeed Vmc and the desired roll
angle to maintain zero sideslip heading. Furthermore, it is important to have an indication
of the remaining climb performance in an engine failure situation.
An engine failure, as shown in gure 3.25, is characterized by a yawing moment NT =
T YCGE (YCGE in FM ) resulting from asymmetric thrust development T at a distance
YCGE from the center of gravity. NT is positive if the right engine fails. It is important
to know the factors in
uencing the magnitude of NT since the degree of diculty of
operating with an engine failure is directly related to this parameter. It is assumed that
the inoperative engine does not produce any thrust or drag.
Version: 1
Date: April 4, 1997 72
GARTEUR/TP-088-27
proportional to the rudder control eectivity CnR and aircraft mass m. To achieve zero
sideslip some bank angle is needed to provide side force equilibrium. This situation is
sketched in gure (3.25). The bank angle required for zero sideslip angle can be obtained
β=0
y APT
NT T
Inoperative
Engine
W sin Φ
Yδ Nδ
R R
δR
Yδ
R
δR
W sin Φ
W Equilibrium
W sin Φ = Yδ
R
The bank angle has to remain in most of the situations less than 5 degrees. This can be
calculated for the RCAM. Consider the situation that the right engine thrust is reduced
to its minimum value of 0:10 105 N (idle), while
ying at a speed of VA = 80 m=s. The
nominal aircraft mass is 120 000 kg. The required thrust for steady horizontal
ight for this
mass and airspeed appears to be (from trimming the model) T = 1:79 105 N , so that the
operative engine at least will have to deliver a thrust of (1:79 0:10) 105 = 1:69 105 N . The
dierence in delivered thrust between both engines is then: (1:69 0:10) 105 = 1:59 105 N ,
causing the asymmetrical thrust moment.
Version: 1
Date: April 4, 1997 74
GARTEUR/TP-088-27
The lift coecient will approximately be equal to ( = 0 = 1:225 kg=m3 and S = 260 m2 ):
Furthermore, from tables 2.1 and 2.2 in chapter 2 can be obtained: yCGE = yAPT 1 =
7:94 m, b = 45 m and CnR = 0:092. The required rudder de
ection can now be
calculated:
5
Rss = 1:59 10 =(120000 90::81)092
1:16 ( 7:94=45) 57:3 = 17:2 deg (3.148)
The roll angle will be equal to: (with CYR = 0:24)
= 0:24 ( 17:2) = 3:6 deg (3.149)
1:16
This less than the accepted 5 degrees in practice. Also the rudder de
ection is less than
the maximum of 30 degrees. Of course these values calculated here are in steady state,
however, during the transient just after the engine failure these values may be larger.
The required aileron de
ection can be calculated from the roll equilibrium:
C +C
A = l C lR R (3.150)
lA
Thus, with Cl = 0:205, ClA = 0:088 and ClR = 0:032:
The second trim solution can be obtained by setting the bank angle to zero. This situation
is sketched in gure (3.26). To achieve steady heading
ight the required rudder for
equilibrium can be obtained via the solution of the lateral force and the yawing moment
equations in 3.145:
T yCGE
Rss = W CL b Cy (3.152)
CnR Cy CyR Cn
this can be also written as:
T yCGE
Rss = CL Cyb Cn
W (3.153)
CnR 1 CnR Cy
R
From the last equation it can be seen that the required rudder angle for trim is larger than
in the case of zero sideslip angle.
The required sideslip angle for the zero bank angle can be obtained from the side force
equation in 3.145:
CYR R
= Cy (3.154)
Version: 1
75 Date: April 4, 1997
GARTEUR/TP-088-27
T y APT
NT
Inoperative
Engine
Yβ
sin γ E
m a xf= W
δR
δR Yδ
R
Yβ
T sin β
Finally, equilibrium can be obtained via the zero rudder option. The situation is sketched
in gure (3.27). From the yawing moment equation 3.145 it is possible to derive the
required sideslip angle for the steady heading equilibrium condition:
T yCGE
ss = W CL b (3.158)
Cn
For a positive yawing moment, the sideslip requirement is negative. The sideslip angle is
generally very large, particularly when
ying at low airspeeds, when the operative engine
thrust is high or if the static lateral directional stability (Cn ) is poor. The bank angle
Version: 1
Date: April 4, 1997 76
GARTEUR/TP-088-27
T
y APT
NT
Inoperative
Engine
Yβ
W sin Φ
Nβ
Yβ
Φ
W sin Φ
T sin β
required to balance the sideforces for the zero rudder de
ection condition is obtained from
the side force equation:
CY
= CL (3.159)
For a positive yawing moment, the required bank angle for the zero rudder de
ection
condition must be negative. This bank angle is relatively large at low airspeeds. At rst
glance this
ight condition might be desirable since the pilot would not have to perform
lateral directional control via his rudder. However, the drag induced by the large sideslip
angle is high. Furthermore, this situation is dynamically dangerous since the vertical
tail is near stall at these sideslip angles. Finally,
ying with a large roll angle is highly
uncomfortable for passengers due to the large lateral acceleration. In the case of automatic
control, the lateral directional controller will try to minimize the sideslip and the bank
angle, somewhere in between the zero sideslip and zero bank angle situation. For the
RCAM the equilibrium will be found for:
5
ss = 1:59 10 =(120000 09:131
:81) 1:16 ( 7:94=45) 57:3 = 12:1 deg (3.160)
and:
= ( 1:6) ( 12:1) = 16:7 deg
1:16 (3.161)
The aileron de
ection is:
A = 0:205 ( 12 :1) + 0:032 0 = 28 deg
0:088 (3.162)
This equilibrium condition requires more aileron de
ection than allowed: the ailerons sat-
urate at 25 deg. Therefore, at this trim condition the zero-rudder option is not applicable.
It is clear that there is a trade o in lateral directional control during an engine failure.
The control of an airplane may be limited either by rudder or aileron control. The rudder
de
ection is minimised if the aircraft is banked into the side of the operative engine, but a
high bank angle may be very uncomfortable for the passengers. Furthermore, to keep the
vertical velocity constant with increasing bank angle, CL must be increased. This results
in an increase in induced drag and an increase in stall speed. A higher sideslip angle will
also result in higher drag. These performance and control considerations determine in fact
the optimum equilibrium
ight condition.
3.6.2 The minimum control speed
Having now discussed the dierent trim solutions in the case of an engine failure, the
minimum control speed, a key parameter for safe operation, will be introduced.
The most critical condition for the airplane due to engine failure occurs at high thrust
settings and low airspeeds, as is during take-o. Under these conditions, the lateral
directional control cannot be regained and maintained after a sudden engine failure if
Version: 1
Date: April 4, 1997 78
GARTEUR/TP-088-27
the velocity is below a certain airspeed, the minimum control airspeed (Vmc ). It may be
dened for a static or a dynamic case. For certication purposes the minimum control
speed for the static case is determined by failing the engine prior to approaching the
minimum condition. It may also be considered as a sudden dynamic condition in which
the critical engine is failed at various airspeeds approaching the minimum condition. For
both cases, Vmc may be limited by the lateral or directional control de
ections available
to counteract the rolling or yawing moments and/or control forces involved. At any rate,
there will be a dierent static and dynamic minimum control airspeed for each power
setting, conguration and bank angle. Vmc for maximum single engine thrust can be
derived from equation 3.145 with = 0 (the compensating yawing moment may only
come from R ), Timax and Rmax :
Timax yCGE = C (3.163)
qSb nR Rmax
Due to an engine failure, the available power decreases, while the drag, and therefore
the required power increases. In the RCAM model drag components due to sideslip and
control de
ections have not been modelled. The only increase in drag is caused by the roll
angle. Due to this roll angle the lift vector is rotated, so that total lift has to be increased
to compensate for the aircraft weight. This is thus the same situation as in a turn. Also
for this case a performance diagram can be drawn. From this diagram the maximum
Version: 1
79 Date: April 4, 1997
GARTEUR/TP-088-27
5
x 10
2.5
h=0m
2.4
xcg = 0.23c
2.3 m = 120000 kg
2.2
Max. Thrust
Tmax , D (N)
2.1
1.9 = 20 deg
1.8 = 10 deg
1.7 = 0 deg
1.6
Vmc
1.5
50 55 60 65 70 75 80 85 90 95 100
Vtrim (m/s)
Fig.3.28 Performance diagram for engine failure; bank angles
climb angles can be drawn as a function of the airspeed for dierent roll angles. The
performance diagram can be found in gure 3.28. From this diagram the maximum climb
angle as a function of the airspeed can be obtained. In gure 3.29 the curve for normal
operation has also been drawn for comparison. Also curves for nonzero bank angles have
been added. It is clear that the total performance degradation due to an engine failure is
dramatic. During take-o the aircraft will hardly or even not be able to climb out with
one engine operative and fully extended
aps, since the climb performance will degrade
even more due to a considerably higher weight at take-o.
To illustrate the eect of an engine failure, a nonlinear simulation has been carried out
with the RCAM. The aircraft response is plotted in gure 3.30. The right engine fails at
t = 5 sec and thrust is restored at t = 15 s. This gure shows the severe eect of the
large yawing moment when it is not compensated quickly enough by rudder and aileron
de
ections.
The gures F.8 and F.9 in appendix F show time responses of longitudinal and lateral
outputs. The roll angle reaches a value of more than 50 deg in less than 15 seconds. In
the same time period the heading changes more than 40 deg, while the lateral acceleration
shows peaks of up to 0:15g. In stationary
ight the thrust of the engines contribute to
both the forces in xB -direction and the pitching moment equilibrium. At rst instance
the failure of an engine decelerates the aircraft, causing a negative horizontal load factor
and a decreasing speed. The engine failure also introduces a sudden pitching moment,
giving rise to a negative pitch rate (q). Only a few seconds later the coupling between
Version: 1
Date: April 4, 1997 80
GARTEUR/TP-088-27
20
h = 0m
xcg = 0.23c
15
120000 kg
Normal operation
10
(deg)
= 20 deg
Vmc
−5
50 55 60 65 70 75 80 85 90 95 100
Vtrim (m/s)
Fig.3.29 Maximum
ightpath angle vs speed, engine failure
longitudinal and lateral motions takes over. Due to the roll angle the lift vector is rotated
up to 50 deg, so that the component that compensates the weight becomes smaller with a
factor of 1 sin = 0:36. The altitude decreases approximately 450 m in 20 seconds.
The loss of potential energy is partly compensated by an increase in kinetic energy due to
the increasing speed. After the left engine thrust has been restored, the lightly damped
phugoid remains.
Version: 1
81 Date: April 4, 1997
GARTEUR/TP-088-27
1200
1100
1000
900
altitude
800
700
600
500
400
0
2000
2000 1000
4000 0
−1000
y distance x distance
From the linearization a 9-state state space model is obtained. The numerical values of
the state-space matrices of the nominal model can be found in table C.3. In these tables
the states have been rearranged, so that the decoupling between longitudinal and lateral
motions is visible. The interaction term in the A-matrix between and w_ B is of course
due to the smaller gravity component along the zB -axis when the aircraft has a roll angle.
However, only perturbations around symmetrical
ight will be considered, so that will
be small, or even zero if only longitudinal dynamics are excited. Usually separate lateral
and longitudinal controllers are designed. Because of the interactions, the longitudinal
controller needs additional compensation when the aircraft is turning.
As can be seen from the B -matrix, each of the engines aects the lateral motions (p_ and r_ ).
Of course the signs of both engines are opposite, so that the interaction is only important
in the case of asymmetrical thrust (e.g. engine failure). The engines are usually not used
Version: 1
83 Date: April 4, 1997
GARTEUR/TP-088-27
to control lateral dynamics and therefore both engines deliver the same thrust during
normal operation.
The eigenvalues of A-matrix sub-blocks for the longitudinal and lateral dynamics are
summarized in table 4.1.
The step responses to tailplane and throttle input (gures F.2 and F.4) are very common
for transport aircraft. The short period is well damped, as can be seen from the pitch
rate and load factor responses and table 4.1. The phugoid mode is clearly dominating and
badly damped. Both modes have dierent physical interpretations. This will be discussed
in the following sections.
The oscillation in the responses due to the rudder and aileron input (gures F.1 and F.3)
comes from the Dutch roll mode, which is lightly damped (table 4.1). When this mode
is excited, the aircraft is slipping, rolling and yawing, where the roll motion has a certain
phase lag. The characteristics of this mode strongly depend on the aerodynamic derivatives
and moments of inertia, and therefore on the aircraft conguration.
The aperiodic roll mode is dominated by the roll rate. Due to both rudder and aileron
de
ections the aircraft rolls with approximately rst order dynamics to a steady state roll
Version: 1
Date: April 4, 1997 84
GARTEUR/TP-088-27
angle. This motion is continued in the spiral mode, which is slightly unstable for many
aircraft. However, not for the RCAM: the spiral mode is stable and has a remarkably low
time constant of 3:5 s (see table 4.1). This is very likely caused by the unconventional
shape of the aircraft fuselage and the extra vertical tail surfaces. For the RCAM spiral
mode the roll angle reaches a steady state value, while the aircraft is slipping somewhat to
the outer side of the turn since and have the same sign. The altitude is continuously
decreasing, so that the
ightpath looks like a spiral.
y v +
1 1 u
s s
b0
- -
a1
a0
In gure 4.1 the system of eq. 4.1 is represented graphically in a block diagram. Note that
in section B.1 the actuator and engine models have been implemented in Simulink in the
same way; this allowed easy implementation of (rate) limits and logic.
It is very easy to derive a state-space representation or a transfer function from a given
block diagram. Drawing this diagram from right to the left facilitates this even more: the
variables and system blocks appear in both the diagram and the equation in the same
order. This is safer in the case of multivariable systems, where the order of multiplication
of matrices is important.
From the diagram in gure 4.1 can be seen that there are two integrations with two
internal feedback loops in the relationship between u and y. If both y and v are chosen as
Version: 1
85 Date: April 4, 1997
GARTEUR/TP-088-27
the outputs of this system the transfer functions can be calculated from the state-space
representation using:
H (s) = C (sI A) 1 B (4.2)
Substitution of eq. 4.1 gives:
y(s) = a0 +ab10s+s2 u v(s) = a0 +ab01ss+s2 u (4.3)
Using these transfer functions the initial and steady state response of y and v to a step
input of u can be calculated with the initial and nal value theorems for functions in the
Laplace domain:
lim y ( t ) = lim sy ( s ) = lim sb0 1 = b0 (4.4)
t!1 s!0 s!0 a + a s + s s a
0 1 2 0
If the relation between y and v is only an integration (a0 = 0), the response of y to a step
input goes to innity, while its derivative, v reaches a steady state value:
lim y_ (t) = tlim
t!1 !1 v(t) = slim
!0
s(sy(s)) = ab0 (4.5)
1
The initial response is an acceleration:
lim y(t) = slim 2
t!0 !1 s(s y(s)) = b0 (4.6)
which does not depend on the internal feedback loops a0 and a1 :
lim v_ (t) = slim
t!0 !1 s(sv(s)) = b0 (4.7)
The same characteristics can directly be read from the block diagram. Because there are
integrations between u and v and y, the states can not directly change due to a step input,
therefore at t = 0s: y = v = 0 and v_ = b0 u. Between u and v there is one integration
so that the initial rate of v is nite. Between u and y there are two integrations, and
therefore the initial response is an acceleration.
If d = n m is the dierence between the number of poles (n) and the number of zeros
(m), then the dth derivative of the output due to a step input is nonzero. d equals the
number of integrations between the input and the output variable and is therefore called
the relative order. The total gain over the connection lines between the input and the
output in the block diagram equals the gain of the initial system response.
The steady state gain can indirectly be obtained from the block diagram. Provided that
the system is stable, the state derivatives will be zero at t ! 1. Therefore the inputs to
all integrators in the diagram are zero. This means that v = 0 (input to second integrator)
and b0 u a0 y = 0 (input to rst integrator), so that in steady state y = ab00 u.
In many cases a block diagram can make it easier to understand input-output relationships,
interactions in a system, controllability and observability of states, (time-scale) separation
of modes etc.
For controller design block diagrams may give insight in which feedback loops via which
control inputs will be eective or not, since the input-output relationships are more tran-
parant.
Version: 1
Date: April 4, 1997 86
GARTEUR/TP-088-27
Zq 1 (in table C.4 Zq = 0:97). This means that there is only a kinematic relation
between and q: the vertical force due to the pitch rate is neglected (in RCAM CZ
has a q-component via equation 2.30);
Z
0 (in table C.4 Z
= 0:007). This means that the vertical component of the
weight vector remains approximately constant when the
ightpath angle changes;
Xq 0 (in table C.4 Xq = 0:05). This is a reasonable assumption for conventional
civil aircraft. Not only is Xq small, but also the pitch rate q is small compared to
V ,
and ;
X
g (in table C.4 X
= 9:785). For the in
uence of the
ightpath angle only
gravity is thus taken into account.
The inputs of both throttles will be taken together, so that ZTH = 2 ZTH 1 = 2 ZTH 2 .
The linearized longitudinal equations now become:
0 1 0 10 1 0 1
B q_ C BB M q M M u 0 q
CC BB CC BB E M M TH C !
B
B _ C
C B 1 Z Z u 0 C B C B Z Z C
C E
B
B V_ CC = BB 0 X g X g
CC BB V CC + BB X E X TH CC (4.13)
@ A @ u A @ A @ E TH A TH
_ 0 Z Zu 0
ZE ZTH
A block diagram of this equation can be found in gure 4.2.
4.3.2 Approximations for the short period
For an approximation of the short period, the in
uence of VA and
to q_ and _ are
neglected (see gure 4.2). In this way the following reduced model is obtained:
! ! ! !
q_ = Mq M q + ME (4.14)
_ 1 Z ZE E
As can be seen from table 4.1, the eigenfrequency and damping of the short period mode are
considerably higher than those of the phugoid. Therefore the short period oscillations have
almost vanished when the phugoid takes over. There is thus a clear time scale separation
between the modes, so that both can be considered separately for approximations.
The elements of the matrices are mainly determined by the center of gravity location via
the moment derivative Cm . This is discussed in more detail in appendix A.
The characteristic polynomial of the reduced A matrix is:
s2 (Z + Mq )s + (Z Mq M ) = 0 (4.15)
or:
s2 2!0 s + !02 = 0 (4.16)
The eigen frequency therefore equals:
q
!0 = Z Mq M (4.17)
Version: 1
Date: April 4, 1997 88
GARTEUR/TP-088-27
δ TH
αw
Vw
δE
Period
Short
E
Μδ
TH
Mu
Μδ
Z δE
Mα
Mq
1
s
q
Zu
Ζ δ TH
αf
.
1
s
Zα
-
α
TH
Xδ
Xα -g
X δE
1
s
Xu
Xγ
-
VA
-Z u
- Zα
-
Phugoid
1
s
γ
These coecients can easily be calculated. In the trim condition _ = 0 deg =s, CD and
CL can be read from gure 2.8: CL = 1:5 and CD = 0:22. Furthermore, = 1:225 kg=m3 ,
g = 9:81 m=s2 , V0 = 70:0 m=s, S = 260 m2 and m = m120 = 120000 kg.
Xu 0:22 260 1:225
= 120000 70 = 0:041
X = g = 9:81
ZTH = 0
The calculated values are indeed almost equal to the numerically calculated coecients,
compared to table C.4. The derivatives in the
-equation that turned out to be zero (Zu ,
ZTH ), are not equal to zero in table C.4, but these parameters are less important for a
physical interpretation.
For the RCAM nominal model the damping will approximately be equal to:
= p1 01::22 = 0:1 (4.38)
2 5
Also in this case the approximation is somewhat higher than the value in table 4.1: 0.1 vs
0.07. When deriving the phugoid approximation, only derivatives with respect to V and
have been taken into account. CD and CL are constant in that case, because _ is set to
zero. However, assuming the angle of attack to be constant is not realistic. Therefore the
approximation will be expanded by taking into account. This is done by including the
moment equation with the assumption that the short period oscillations vanish quickly
after perturbations. q and q_ and can then be taken zero:
M + MuV + MTH TH + ME E = 0 (4.39)
Thus:
= M1 (Mu V + MTH TH + ME E ) (4.40)
This includes in the model the fact that the phugoid mode is also controllable by the
elevator via the angle of attack and thus via the short period dynamics, as illustrated in
gure 4.2. Including the in
uence of the angle of attack in equation 4.24 gives:
0 1
! !B CC ! !
V_ X g Xu X
B XTH XE TH
= V A+ (4.41)
_ Z Zu 0 @ ZTH ZE E
Substitution of equation 4.40 gives:
! 0
X g X
1 !
V_ = @
X u M u M
A V + (4.42)
_ Zu Mu MZ 0
0 X g
X g
1 !
@ X TH M TH M ME M Z A TH
Z
MTH M ZE ME M E
The phugoid mode can be interpreted from an energetical point of view. In section 3.1.2
was derived for the specic energy rate:
E_ s = E_ = V_ + sin
= sin
(4.43)
V WV g E
In controlling this energy rate (total energy management) there are apparently two degrees
of freedom: the total (specic) energy rate and the distribution between its components:
sin
and Vg_ . This distribution is usually characterized by the specic energy distribution
rate, the dierence between the acceleration term and the
ightpath term:
_
D_ = sin
Vg (4.44)
Version: 1
93 Date: April 4, 1997
GARTEUR/TP-088-27
D_ quanties the change in the distribution between potential energy (altitude) and kinetic
energy (velocity) per unit time.
If there is no external input of energy, for example when the phugoid is initiated by an
elevator input (the short period is assumed to be energy conservative), only potential and
kinetic energy can be exchanged: E_ s = 0 and thus:
V_ = sin
(4.45)
g
And for D_ :
_
D_ = 2 Vg = 2 sin
(4.46)
If D_ = 0, no energy is transferred between kinetic and potential energy.
The energy household during a phugoid oscillation can be illustrated using Korhammer di-
agrams. In chapter 3 these diagrams have been used for quasi-stationary analysis, however,
they can also be used for studying dynamic behaviour. For that reason a MatLAB-routine
has been written (KORHAMMER.M) that plots the diagram at every time step of a pre-
viously recorded nonlinear simulation. Running this on a PC or Workstation gives a nice
animation. In this section however, these diagrams will be drawn for only nine points on
one sine wave, as illustrated in gure 4.3. In this gure a time registration of the phugoid
excited by a tailplane input is drawn. Figure 4.4 shows an animation with the aircraft at
approximately the same time points.
15
10 h_
3
4
V , h_ (m/s)
5
2
V
0 1 5 9
6
8
−5 7
−10
0 50 100 150
time (s)
Fig.4.3 Phugoid simulation
A registration of the specic energy rate and the specic energy distribution rate can
be found in gure 4.5. The reason why E_ s is not equal to zero, is mainly the slightly
Version: 1
Date: April 4, 1997 94
GARTEUR/TP-088-27
1100
1080
1060
1040
1020
altitude
1000
980
960
940
920
900
0 1000 2000 3000 4000 5000 6000 7000
x distance
0.3
0.2 1 2 3 4 5 6 7 8 9
V_
g
_ E_ s =V; V_ =g; sin
0.1
0
E_ s
V
sin
−0.1
D;
−0.2
D_
−0.3
0 50 100 150
time (s)
CL
CD
CL
γ gl
CR
xa VA CD
γ
RA
xv
W
. . begin point γ
thrust vector
E D Xf
dV
. dt
T -m V
W sin γ
za zv
1.25 1.25
0.625 0.625
0.3 0 0.3 0
0 0.125 0.25 0 0.125 0.25
0.2 0.2
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 1 2 0 1 2
1.25 1.25
0.625 0.625
0.3 0 0.3 0
0 0.125 0.25 0 0.125 0.25
0.2 0.2
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 1 2 0 1 2
1.25 1.25
0.625 0.625
0.3 0 0.3 0
0 0.125 0.25 0 0.125 0.25
0.2 0.2
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 1 2 0 1 2
Fig.4.7 Korhammer diagrams at dierent time points during phugoid period (A)
Version: 1
97 Date: April 4, 1997
GARTEUR/TP-088-27
1.25 1.25
0.625 0.625
0.3 0 0.3 0
0 0.125 0.25 0 0.125 0.25
0.2 0.2
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 1 2 0 1 2
1.25
0.625
0.3 0
0 0.125 0.25
0.2
0.1
−0.1
−0.2
0 1 2
Fig.4.8 Korhammer diagrams at dierent time points during phugoid period (B)
Version: 1
Date: April 4, 1997 98
GARTEUR/TP-088-27
the projections of the weight vector and the aerodynamic force vector on the zF -axis are
equal. This corresponds with the fact that the
ightpath angle has reached its maximum
value. If the thrust vector is below the xF -axis, zF -component of the aerodynamic force
is larger than the zF -component of the weight vector, and thus there is an upward accel-
eration (remember that the weight vector always ends on the xF -axis). This occurs in the
second gure. If the thrust vector is above the xF -axis the opposite holds. The vertical
acceleration in the second gure is is caused by the airspeed that is still increasing due to
the positive sign of the mV_ projection on the xF -axis. Therefore the total aerodynamic
force increases, even though the angle of attack is less than in the rst gure (compare the
points on the lift drag polars in the rst and the second gure). The angle of glide
gl does
not change much, so that the larger total aerodynamic force also means a higher drag.
Due to this extra drag some energy is dissipated, which can be seen from the non zero
E_ s-bar. The situation in this gure is half way to the situation in the third gure. There
the xF -projection of the acceleration vector has become zero. If the damping would have
been zero, the
ightpath angle would also have been zero, but this is obviously not the
case. While the velocity is going to decrease, the
ightpath angle is still slightly negative,
resulting in a negative E_ -term. The distribution rate is almost zero and changes sign:
from now on kinetic energy is
ushed back to potential energy again.
The situation sketched in the fourth gure shows the picture half way to the situation in
the fth gure. There the distribution rate is maximum, the acceleration and
ightpath
terms again achieve maximum values in absolute sense and have opposite signs. In this
case however, the
ightpath angle is positive and the acceleration is negative.
In the seventh gure the airspeed has reached a minimum, while the
ightpath angle is
still slightly positive. Due to the lower airspeed, the total aerodynamic force is smaller,
resulting in a downward acceleration. Via the situation in the eighth gure, the picture of
the rst gure returns. However, due to the damping the absolute values of the acceleration
and
ightpath angle are smaller than in the rst gure.
Overlooking all pictures, it can be seen that during a phugoid period the acceleration vector
is rotating about the end point of the weight vector, while its length varies. Unfortunately
the variation of the work point along the lift-drag polar is dicult to see, because the
angle of attack does not change much.
Although many remarks made above are quite trivial, this is an example of how a Ko-
rhammer animation visualizes the changing force pattern during longitudinal manoeuvres.
Korhammer diagrams can be used to analyse the performance of the longitudinal controller
in the sense of throttle/elevator co-ordination.
The total energy principles outlined above, are the key to decoupled co-ordination of
elevator and thrust in energy terms, see for example [13]. From equation 4.25 the force
Version: 1
99 Date: April 4, 1997
GARTEUR/TP-088-27
while the elevator should distribute this power input such that indeed all the energy is
used to increase kinetic energy. The same holds when a
ightpath angle is demanded at
constant speed. Of course injection and distribution of power have to be performed at
the same speed, which means that the altitude and the speed control loops must approx-
imately have the same bandwidth. An example of ideal elevator/throttle co-ordination in
energy and distribution rate terms is given in gure 4.9.
This control concept is based on point-mass dynamics. Therefore, due to for example en-
γc .
Vc
. . .
Dε = E ε = γε Eε
.
Vε /g γε
. .
Dε = - Vε /g
Fig.4.9 Energy rate and distribution rate errors for optimal elevator-throttle co-ordination
ergy dissipation in the innerloops, always more power is necessary so that the co-ordination
will not be ideal. Furthermore, aircraft do not always
y near the minimum drag speed,
so that drag variations may play an important role in the energy management. However,
the total energy concept is a good concept for both analysis and design. Using modern
synthesis methods the co-ordination of the control inputs is preferably left to the opti-
mization algorithms. In this case the concept can be used to analyse the extent of the
achieved decoupling between altitude and speed control. As a way of thinking for design,
the total energy considerations have proven their value in research projects of Boeing and
NASA, see for example [13, 14].
Version: 1
101 Date: April 4, 1997
GARTEUR/TP-088-27
δR
βw
δA
yaw motion
R
Nδ
A
Nδ
Nβ
Np
Nr
1
s
r
Yp
-
R
Yδ
βf
.
Yβ
g/V0
1
s
−
β
φ
p
δR
δA
r
A
Lβ
Lδ
R
Lr
Lδ
Lp
1
s
p
Roll motion
θ0
1
s
φ
In the following sections some approximations for the dierent modes will be discussed.
Since the time scales of the modes are relatively close these approximations are only
suitable to investigate initial behaviour and to obtain approximations for characteristic
numbers (!0 , , ) in relation to the aerodynamic derivatives.
A very rough approximation can be made by separating roll and yaw motion, see the block
diagram in gure 4.10. The Dutch roll mode can then be characterized by the states r
and :
! ! ! !
r_ = Nr N r + NR (4.59)
_ 1 Y 0
The characteristic polynomial is:
s2 s(Nr + Y ) + (N + Nr Y ) = 0 (4.60)
The eigen frequency is therefore equal to:
q
!0dr = N + Nr Y (4.61)
Version: 1
Date: April 4, 1997 104
GARTEUR/TP-088-27
Although it is inaccurate to characterize the spiral mode using a linear model, the lin-
earization can be used to derive a stability condition for this mode.
For studying spiral stability in the rst place an extra constraint is added to the lateral
model: it is assumed that the centrifugal force and the lateral weight component by the
roll angle are in equilibrium (turn co-ordination):
g 0 V0 r (4.65)
This means that the aerodynamic force vector is assumed to be in the symmetry plane of
the aircraft and the component CY is neglected. The homogeneous state equations now
reduce to:
0 1 0 10 1
B r
_ C BB N r N N p 0 CC BB r CC
B
B C
0 C B 1 0 0 g=V0 C B C
B
B C = BB L L L 0 CC BB p CC (4.66)
@ p_ CA @ r p A@ A
_ 0 0 1 0
The characteristic equation of this this matrix is:
" !#
s2 Lp Np g L s + Vg N1 (Nr L N Lr ) = 0 (4.67)
V0 N 0
Version: 1
105 Date: April 4, 1997
GARTEUR/TP-088-27
In this
ight condition, this equation has two real poles: from these roots time constants
for the roll mode and the spiral mode follow. For many aircraft these time constants dier
considerably, but this is not the case for the RCAM, as can be seen in table 4.1. This is
mainly due to a high Cl (probably because of the high fuselage) and a high Cnr (possibly
due to extra vertical surfaces on the tailplanes).
It it possible that in some
ight conditions the real poles become a complex pair. This
gives rise to a "lateral phugoid". Looking at table E.2 and E.6 the time constants rapidly
get closer to each other when the mass increases and the speed decreases. Trimming the
RCAM at a speed of V0 = 60 m=s and with a mass of 150000 kg gives a conjugate pair of
roots: 0:5801 + 0:0923i: !0 = 0:5874 and = 0:9876.
The expression between brackets is usually very small. TR is positive, so that for a stable
spiral mode is required:
!
1 (N L N L ) = N L Lr > 0 (4.71)
N r r Nrr N
Nr and L are usually negative and N and Lr are mostly positive (see table C.5), thus
for stability is required:
!
L Lr < 0 (4.72)
N Nr
or:
(Nr L N Lr ) > 0 (4.73)
For the RCAM model can be substituted:
(Nr = 0:4556, L = 1:4900, N = 0:1791 and Lr = 0:4811)
(Nr L N Lr ) = ( 0:4556 1:4900 0:1791 0:4811) = 0:59 > 0 (4.74)
This number is very high. As already noted in section 4.1, the spiral mode is clearly stable
for this trim condition.
Version: 1
Date: April 4, 1997 106
GARTEUR/TP-088-27
Thrust input
Before calculating the initial and steady state gains, the transfer functions need to be
determined. This can be done easily using the well known Cramer's rule. The transfer
function TH to V is calculated by replacing the u-column of (sI A) with the TH -column
of B and calculating the determinant of the matrix obtained, divided by the characteristic
polynomial of A, which is equal to the determinant of (sI A):
s Mq M MTH 0
s Z ZTH 0
HuTH = jsI 1 Aj 1
0 (X g) XTH g
0 Z ZTH s
3 2
= Xs4TH+sa +s3b+2 sa +s2b+
1 s g (M ZTH MTH Z )
a1 s + g (M Zu Mu Z ) (4.75)
3 2
Some of the terms in this equation have not been expanded, because for the inital and
nal value theorems only the rst and the last terms of the polynomials are needed. Now
the steady state response to a unit step on TH can be calculated:
lim H = MZTH MTH Z (4.76)
s!0 u TH M Zu Mu Z
If ZTH and MTH would be zero, HuTH is equal to zero, and all power provided by the
engines will be transferred to the potential energy rate. In the case of the RCAM model
Version: 1
107 Date: April 4, 1997
GARTEUR/TP-088-27
however, the MTH -term is positive due to the fact that the engines are situated below
the center of gravity. By the eect on the moment equilibrium, the energy distribution
changes: some of the engine power is used to accelerate the aircraft.
The transfer function from TH to
can be calculated in a similar way as HuTH :
s Mq M Mu MTH
1 s Z Zu ZTH
H
TH = jsI 1 Aj 0 (X g) s Xu XTH
=
(4.77)
0 Z Zu ZTH
s3 +c3 s2 +c2 s+M(XTH Zu ZTH Xu )+ x (MTH Zu ZTH Mu )+Z (MTH Xu XTH Mu )
s4 +a3 s3 +a2 s2 +a1 s+g (M Zu Mu Z )
The steady state response to a unit thrust input is:
lims!0 H
TH =
M (XTH Zu ZTH Xu ) (X g)(MTH Zu ZTH Mu )+Z (MTH Xu XTH Mu (4.78)
g(M Zu Mu Z )
and if ZTH and MTH are equal to zero:
lim H
s!0
TH
= XgTH (4.79)
The initial response is mainly determined by the relative orders:
d = 0: direct feed through, TH ! V_
d = 1: one integration, TH ! V; nz
d = 2: two integrations, TH !
;
Thus:
lim _ = XTH TH = T
V (4.80)
t!0 m
The numerical values for the RCAM model can be found in table 4.2.
Elevator
If only the short period approximation is considered rst, the initial elevator response is
determined by the following relative orders:
d=0, direct feed through: E ! q_
d=1, one integration: E ! q (and via ZE )
d=2, two integrations: E ! ;
The relative order is of great signicance for nonlinear inverse controller design. In that
case it is important to know how many integrations there are between control inputs and
the outputs to be controlled in order to nd an invertible relationship between these inputs
Version: 1
Date: April 4, 1997 108
GARTEUR/TP-088-27
Tailplane
outputs: inital gain steady state gain
longitudinal short per.
q_ -1.8667 0.0 0.0
q 0.0 0.0 -0.69
0.0 -2.3 -1.29
0.0 -2.4 {
V 0.0 330.8 {
0.0 -0.1 {
Thrust
outputs: inital gain steady state gain
V 19.5 172.0 {
0.0 2.1 {
Table 4.2 Initial and steady state gains for step inputs
and derivatives of the outputs. This seems quite straight forward, but secondary eects
of controls may cause the relative order to be lower than it is intended to be. An example
is the transfer function HE (short period approximation):
s Mq M ! 1
ME
!
HE = 0 1
1 s Z ZE
= s2 (MZE+(sZ )M
s
q ) + ME
+ (Mq Z M ) (4.81)
q
Due to the ZE factor in the nummerator, the relative order is 1. However, the contribution
of the elevator to the total lift is only a secondary eect, because the device is primarily
intended to control the pitching moment. Thus designing an inverse controller based on
this unintended single integration relationship doesn't make any sense. For this application
ZE should be set to zero so that the relative order is 2. Otherwise a right half plane zero
is cancelled by the controller, giving rise to internal instability. In a nonlinear sense this
is known as unstable zero-dynamics.
The analysis of steady state behaviour from the short period approximation should in
this case be carried out carefully, because some time after an elevator input the phugoid
component of the response takes over. It is possible to calculate a short time steady state
gain, because there is quite a lot of time scale separation between both modes and the
short period is well damped.
q H = ZM
ME M ZE
E tTsp = slim
!0 qE Z Mq
(4.82)
Version: 1
109 Date: April 4, 1997
GARTEUR/TP-088-27
or:
Cm0 = zAPT 1 T
c =2V02 S 0 (4.92)
On the other hand, for steady state condition also the following holds:
T0 =2V02SCD0 (4.93)
and:
Cm0 = CD0 zAPT
c (4.94)
This term is negative if zAPT is positive, like in the case of the RCAM, where the engines
are situated below the center of gravity. From the equations above the following can be
derived:
zAPT =cCL iT Cm
V0S (CL0 Cm + CD0 CL zAPT =c) m120 g (4.95)
The denominator will usually be negative (Cm < 0; zAPT < c), as long as the term
CL0 Cm is dominating. Thus if zAPT and iT are positive, the steady state gain will have
a negative sign, opposite to the sign of the initial response, causing nonminimum phase
behaviour. A consequence of the right half plane zero is that for the design of a SISO
feedback loop V via TH no integrator can be put in the controller, because this new pole
will immediately be attracted by this zero into the right half plane, resulting in instability.
A very common example of nonminimum phase behaviour is the response of the load factor
to an elevator input. This has an easy physical interpretation: for a positive elevator
command the steady state response of
will be negative. The change in elevator position
however, will initially increase the total lift of the aircraft, so that the initial response
of the load factor (which is proportional to
_ ) will be upward. When the aircraft starts
pitching the load factor will become negative. The order of magnitude of the positive zero
is the same as that of the poles of the short period.
The position of the zero depends on the location of the sensor for the vertical acceleration.
Not only nz has a direct feed through from the elevator, but also q_ due to the pitching
moment that arises from the de
ection. When the vertical acceleration is not measured
in the center of gravity, the q_ term will have a contribution of lsensq_, where lsens is the
distance in xM -direction between the sensor and the center of gravity. The total measured
vertical acceleration therefore equals:
zsens = zcg lsensq_ (4.96)
The initial gain from an elevator input to zcg is approximately equal to V0 ZE (see equa-
tion 4.14, with zcg _ V0 ) and the initial gain of the input to q_ is ME . The initial
response of the vertical acceleration measured at a distance lsens from the center of grav-
ity is therefore:
zsens = (V0 ZE lsens=ME ) E (4.97)
Version: 1
111 Date: April 4, 1997
GARTEUR/TP-088-27
Thus in the case of the RCAM model the sensor needs to be moved from the center of
gravity over: (for the RCAM tailplane derivatives are used)
lsens = V0 ZT =MT = 70( 0:0709= 1:8667) = 2:7 m (4.98)
in order to make the measured acceleration zero. This location is called the percussion
point. If the sensor is positioned more than 2:7 m in front of the center of gravity, the
zero has shifted to the left half plane.
This non-minimum phase behaviour can of course also be avoided by relocating the con-
trols. An example is the use of canards. In the Lockheed 1011
ightpath control is
augmented with direct lift devices on the wing during the approach to land, see [15].
Another, slower nonminimum phase response arises when the aircraft is
ying on the
backside of the power curve. This can be seen from the steady state gain of the
ightpath
angle:
ME (X Z Z (X g))
ss = gM (4.99)
Z u u
u
Furthermore:
W =2V02SCL0 (4.100)
Substituting from appendix D:
Zu S (C + C )
m L0 0 D0
Xu S
V0 m (CD0 0 CL0 ) (4.101)
Z =2V0 mS (CL + CD0 + 0 CD )
X =2V02 mS (CD 0 CL )
gives:
=2V02 mS CmE
gCm CL0 (CD0 CL CL0 CD ) (4.102)
The left ratio is normally negative. When
ying on the right side of the power curve, the
right term is positive, so that a positive elevator de
ection results in a negative
ightpath
angle. This term however changes sign when:
CD0 < CD (4.103)
CL0 CL
see also section 3.1.3. For the linearized RCAM models this phenomenon can be illustrated
by plotting the specic zero at dierent linearization points, see gure 4.11. At the airspeed
of VA = 70 m=s the zero is very close to the origin. This is also visible in gure E.1, where
at lower frequencies the frequency response of
to T shows dierentiator-like behaviour.
Indeed, the
ightpath angle has become independent of the elevator position in steady
state.
Version: 1
Date: April 4, 1997 112
GARTEUR/TP-088-27
0.1
0.08
0.06
0.04
90 70 55
imag 0.02
−0.02
−0.04
−0.06
−0.08
−0.1
−0.1 −0.08 −0.06 −0.04 −0.02 0 0.02 0.04 0.06 0.08 0.1
real
Fig.4.11 RHP-zero due to backside operation
This right half plane zero will give troubles when the
ightpath angle is held constant while
the speed is not controlled. If in that case the airspeed may decrease due to for example
sudden head wind, the drag increases as can be seen from the power curve (see gure 3.10)
decelerating the aircraft even more. The
ight condition will therefore not return to its
initial operating point. When an aircraft is in low speed conguration in the approach
to landing, it might enter the backside of the power curve. In that case a conventional
ight controller designed using a SISO, one loop at a time approach (autopilot:
ightpath
via elevator, autothrottle: speed via throttles) will no longer work properly. If one of the
systems is switched o, the other will cause instability of the aircraft. This is usually
solved by interchanging the controls of the
ightpath angle and speed loops: speed is then
controlled using the elevator, while
is controlled via the throttles.
In fact it is not a real stability problem, since the uncontrolled aircraft will nd equilibrium.
When the control problem is attacked as a multi-variable problem, this speed instability
is completely avioded.
As already seen in section 4.4 the lateral motions are more dicult to characterize. After
a lateral control input the aircraft will get a roll angle and the spiral mode will dominate
the response. Since the equations have been linearized for zero roll angle, it does not make
sense to calculate steady state responses using the transfer functions, since the motions
can no longer be considered linear. However, some remarks can be made using the block
diagram in gure 4.10.
Aileron input
For the initial response holds (see gure 4.10):
d = 1, (1 integration): A ! p, (A ! r via NA )
Version: 1
113 Date: April 4, 1997
GARTEUR/TP-088-27
d = 2, (2 integrations): A ! .
For the initial response of the yaw angular acceleration holds:
lim r_ = NA A
t!0
(4.104)
Positive NA may give rise to non-minimum phase behaviour (adverse yaw), since the
initial yaw acceleration is opposite to the intended yaw rate. In the case of the RCAM,
in the nominal trim condition holds NA = 0:0135 < 0 . It is not unlikely that in other
trim congurations the number becomes positive.
Rudder input
The eect of a rudder input can be studied from the complete block diagram, see g-
ure 4.10. The following holds:
d = 1, (1 integration), R ! r, (R ! p, R ! )
d = 2, (2 integrations) R ! , (R ! ).
The response of after only one integration is due to the term YR . This is the same kind
of side eect as ZE : a force is generated by a device that principally controls a moment.
Since LR = 0:2223 > 0, the initial roll acceleration will be in opposite direction compared
to the direction of the steady state roll angle.
Version: 1
Date: April 4, 1997 114
GARTEUR/TP-088-27
This can also be seen from gure F.5, where indeed VA and the other states return to their
initial values, except V which increases with WXE .
In gure E.7 in appendix E bodeplots for the wind inputs can be found. The observations
made above are conrmed by the rst picture: the relative order is of VA is zero, while all
variables in the gure have zero steady state gain .
4.6.2 Vertical wind input
The eect can be seen as a disturbance on the angle of attack: w WZE =V0 , see the
blockdiagram 4.10. From this gure can immediately be seen that , nx and nz respond
with a zero relative order, thus by direct feed through. The nonlinear time simulations
can be found in gure F.6 (section 3.4.4). The steady state responses for all outputs are
zero, except the
ightpath angle.
This can also be seen from the block diagram, where for t ! 1 the integrator inputs are
zero. Therefore:
Mq qss + Mss = 0 (4.110)
qss + Zss = 0
Thus and q need to be zero, which means that the moment equilibrium returns to its
initial situation. For the inputs of the integrators of
and V holds:
ZuVAss Z ss = 0 (4.111)
X ss + Xu VAss gss = 0
Thus VAss = V = 0 and = 0. Furthermore:
ss = fss w ! fss = w (4.112)
= ss ss w !
ss = w
For the initial response the following relations hold:
lim = w
t!0
(4.113)
lim _
t!0 f
= Z w (4.114)
Although the RCAM model has no inputs for wind rates, these are very important in the
case of turbulence. From the time simulations it is clear that the short period mode is
excited via q. As can be seen from the block diagram, there is a feedback loop from to
q_, so that the relative order of a wind input to q is one.
Again these observations can be seen in bode plots as well: see gure E.7.
4.6.3 Lateral wind input
A lateral wind input directly aects the side slip angle: w = WY E =V0 . For the initial
responses can be written (see the blockdiagram in gure 4.10):
limt!0 = w and: limt!0 _f = Y w (4.115)
Version: 1
Date: April 4, 1997 116
GARTEUR/TP-088-27
From the same diagram also steady state behaviour can be analysed.
pss = 0
Lr rss + L ss + Lppss = 0 (4.116)
Nr rss + N ss + Nppss = 0
Thus pss, rss and ss are zero. Furthermore:
ss = fss wss = 0 ! fss = w (4.117)
Time simulations are discussed in section 3.4.4.
Version: 1
117 Date: April 4, 1997
GARTEUR/TP-088-27
5 Conclusions
In this document the Research Civil Aircraft Model has been discussed. In the second
chapter the equations of motion have been described in a structured way, as used in [6].
On the basis of this model a number of performance and dynamic characteristics have
been calculated. Although many of the formulas can be found in standard text books, it
was very interesting to do the whole range of calculations for one nonlinear aircraft model.
Usually aircraft performance (along trajectories) and dynamics around the center of grav-
ity are considered as two separate disciplines. Of course the second is the most important
for controller design, but it is not possible to design a
ight controller without knowing
the performance limits of the aircraft, especially when the controller is designed for a
large part of the
ight envelope, and when the controller has to guide the aircraft along
trajectories with turns, climbs, and descents, under severe disturbances.
Using point-mass mechanics and steady state analysis, control strategies have been de-
rived for engine failure, wind elds, windshear, climbing and accelerated
ight. At this
point energy considerations and Korhammer diagrams are very interesting concepts to
physically understand (longitudinal) aircraft behaviour and control, including phenomena
like backside operation.
The longitudinal performance and dynamics of the RCAM are representative for a civil
airliner, while the lateral dynamics have strange properties, like a very fast spiral mode.
In fact, in spite of many simplications, the most interesting
ight situations are covered
by the model, including backside operation and stall.
Version: 1
Date: April 4, 1997 118
GARTEUR/TP-088-27
References
[1] Flight Mechanics Action Group 08. Communication Handbook. GARTEUR FM-
AG08/TP-088-5, second edition, February 1996.
[2] Flight Mechanics Action Group 08. Robust Flight Control Design Challenge Problem
Formulation and Manual: the Research Civil Aircraft Model (RCAM). GARTEUR
FM(AG08)/TP-088-03, second edition, January 1996.
[3] Flight Mechanics Action Group 08. Robust Flight Control Design Challenge Problem
Formulation and Manual: the Research Civil Aircraft Model (RCAM). GARTEUR
FM(AG08)/TP-088-03, third edition, February 1997.
[4] Flight Mechanics Action Group 08. Robust Flight Control Design Challenge Problem
Formulation and Manual: the High Incidence Research Model (HIRM). GARTEUR
FM(AG08)/TP-088-04, third edition, February 1997.
[5] S. Bennani and G.H.N. Looye. RCAM Design Challenge Presentation Document: the
-Synthesis Approach. Technical report, GARTEUR, April 1997. RCAM entry of
Delft University of Technology.
[6] Rudolf Brockhaus. Flugregelung. Springer-Verlag, Berlin Heidelberg, 1994.
[7] G. Bruning, X. Hafer, and G. Sachs. Flugleistungen. Hochschultext. Springer-Verlag,
Berlin Heidelberg, 1986.
[8] F.E. Cellier. Continuous System Modeling. Springer-Verlag, New York, 1991.
[9] H. Elmqvist. Dymola { Dynamic Modeling Language User's Manual. Dynasim AB,
S-22370 Lund, Sweden, 1993.
[10] O.H. Gerlach. Vliegeigenschappen I. Delft University of Technology, Delft, 1981.
[11] X. Hafer and G. Sachs. Flugmechanik (Moderne Flugzeugentwurfs- und
Steuerungskonzepte). Hochschultext. Springer-Verlag, Berlin Heidelberg, 1980.
[12] K.U. Hahn. Beitrage zur Flugleistungsbestimmung beim Startsteig
ug unter
Berucksichtigung variabler Windein
usse. PhD thesis, TU Braunschweig, 1988.
[13] A A. Lambregts. Operational aspects of the integrated vertical
ight path and speed
control system. AIAA Paper 83-1420, 1983.
[14] A A. Lambregts. Vertical
ight path and speed control autopilot design using total
energy principles. AIAA Paper 83-2239, 1983.
[15] L.O. Lykken and N. Shah. Direct lift control for improved automatic landing safety
and performance for a large transport aircraft. In Proceedings of the AIAA Guidance,
Control and Flight Mechanics Conference, August 1971.
Version: 1
119 Date: April 4, 1997
GARTEUR/TP-088-27
[16] Donald McLean. Automatic Flight Control Systems. Series in Systems and Control
Engineering. Prentice Hall International (UK) Ltd, Hertfordshire, 1990.
[17] Sandeep S. Mulgund and Robert F. Stengel. Aircraft
ight control in wind shear
using partial dynamic inversion. In Proceedings of the American Control Conference,
pages 400{404, June 1993.
[18] Sandeep S. Mulgund and Robert F. Stengel. Target pitch angle for the microburst
escape maneuver. Journal of Aircraft, 30(6):826{832, Nov.{Dec. 1993.
[19] Martin Otter. Objektorientierte Modellierung mechatronischer Systeme am Beispiel
Geregelter Roboter. PhD thesis, Fakultat fur Maschinenbau der Ruhr-Universitat
Bochum, November 1993.
[20] G.J.J. Ruijgrok. Elements of Airplane Performance. Delft University Press, Delft,
1990.
[21] Brian L. Stevens and Frank L. Lewis. Aircraft Control and Simulation. Wiley-
Interscience Publication. John Wiley & Sons, Inc., New York, 1992.
Version: 1
Date: April 4, 1997 120
GARTEUR/TP-088-27
between 0:35 0:19 = 0:16 and 0:35 0:03 = 0:32. The locations of the aerodynamic
centers, the neutral point and the center of gravity can also be found in gure 2.10.
In gure 4.2 the in
uence of Cm via the M -block on the short period dynamics can
be seen. Additional feedback of an angle of attack measurement via the tailplane will
eectively change M and therefore Cm . In this way the neutral point can articially be
moved to a more aft location (equation A.4), increasing the stability margin. This margin
can thus be interpreted as a gain margin in a control theoretical sense. The augmentation
of the longitudinal stability allows for a lower natural stability margin, which is the idea
behind "relaxed stability" mentioned in section 2.1.4.
The required elevator de
ection for trimming can be obtained from:
= 1 (C + C ( ))
E CmE m0 m 0 (A.6)
If now is substituted:
CL CL ( 0 ) = 1=2W
VA2 S (A.7)
the elevator position as a function of the trim speed is obtained:
!
E = C 1 C m W
Cm0 + C 1=2V 2 S
(A.8)
mE L A
Dierentiating this expression with respect to the airspeed gives:
dE = 1 Cm 4W (A.9)
dVA CmE CL VA3 S
Since CmE is always negative and CL is always positive, for a negative Cm is required
that dE =dVA is positive. In gure A.1 the required tailplane position of the RCAM
has been plotted as a function of the trim speed. Static stability is conrmed by the
dT > 0. In equation A.8 also the in
uence of the aircraft mass is
observation that dV A
visible. In gure A.2 trimcurves have been plotted for three masses.
The conclusions derived here for elevator trim, also hold for the tailplane de
ection
Version: 1
Date: April 4, 1997 122
GARTEUR/TP-088-27
20
18 mass: 120000 kg
h=0m
16 0.15 c ||-
0.23 c - - - -
14 0.31 c ........
T (deg)
12
10
0
40 50 60 70 80 90 100 110 120 130
Vtrim (m/s)
Fig.A.1 Trim curves for dierent CoG locations
20
18 xcg : 0.23c
h=0m
16 100000 kg ||-
120000 kg - - - -
14 150000 kg ........
T (deg)
12
10
0
40 50 60 70 80 90 100 110 120 130
Vtrim (m/s)
Fig.A.2 Trim curves for dierent masses
Version: 1
123 Date: April 4, 1997
GARTEUR/TP-088-27
traject
reference
signals
Controls
controls
controller control−
outputs inputs
REFERENCES sim
ouputs for
simulation
CONTROLLER
Engine failure
(1/0)
lon
ACTUATORS
longitudinal
0 measurements
Const. windfield
(1/0)
lat
0 lateral
Turbulence measurements
WIND &
(1/0) TURBULENCE
AIRCRAFT
wind
total wind−
inputs
gure the control and wind inputs and the three groups of outputs are visible. From the
block NONLINEAR AIRCRAFT MODEL the compiled c-code with the nonlinear state
equations is called, see also appendix F in [3]. The meaning of the alpha-numeric variable
names can be found in [1].
The CONTROLLER subblock will contain the control laws that are to be designed. The
contents of this block will be described in detail in [5].
A more detailed view of the ACTUATOR MODELS block can be found in gure B.3.
There is a separate input (ENGINE FAILURE (1/0)) that can be set to 1 if the left
engine is failed. The contents of the TAILPLANE ACTUATOR block can be found in
gure B.4. The RUDDER ACTUATOR and AILERON ACTUATOR blocks are similar.
Version: 1
Date: April 4, 1997 124
GARTEUR/TP-088-27
In this block the limiters and rate limiters have been implemented. The integrator is
limited with the maximum and minimum tailplane de
ections.
A detailed picture of the ENGINE MODEL can be found in gure B.5. In this gure
the trim setting and the control input are added and position and rate limits have been
implemented. In the upper block system a switch has been implemented for engine failure.
As soon as the EF (1/0) input is set to 1, the rst order system is opened while the other
with a time constant of 3:3 s is closed, so that the thrust decays with rst order dynamics
of the lter in equation 2.54.
The contents of the WIND MODELS block can be found in gure B.6. The WINDSHEAR
block can be found in gure B.7. In this gure can be seen that two look-up tabels have
been implemented for the vertical and horizontal wind. The MEMORY block is necessary
to prevent an algebraic loop in Simulink.
In the block DRYDEN the forming lters for Dryden turbulence described in section 2.3.2
have been implemented.
Q (rad/s)
NX (−)
NZ (−)
Floating scope
WV (m/s)
Mux 2
DA ( rad) Z (m) lon
VA (m/s)
DT( rad)
V (m/s)
DR ( rad)
1 Demux BETA (rad)
u THROTTLE1 (rad)
P (rad/s)
THROTTLE2 (rad)
R (rad/s)
ALPHA (rad)
Mux
GAMMA (rad) Mux 1
sim
X (m)
NY (−)
U0(1:5) Demux
Trim settings
controls
DA_C
Aileron actuator
DT_C
Tailplane actuator
DR_C Mux 1
1 Demux
Control inputs
uc Rudder actuator aircraft model
THROTTLE1_C
THROTTLE2_C
2
engine failure Engine model
(1/0)
1
DT_trim
+
2 + 1
6.7 1/s +
DT_c − DT
Tailplane Rate limiter Limited
actuator Integrator
bandwidth
Engine
failure
ENGINE 1 constant
5 −1/3.3
ef (1/0)
1 + − +
1 1
THROTTLE1_trim + +
+ − 0.6 s
Switch Integrator THROTTLE1
2 + Lim Engine dyn.
THROTTLE bandwidth Rate lim
THROTTLE1_c THROTTLE
0.5/57.3 0.5/57.3
THROTTLE_min THROTTLE_min
ENGINE 2
3 + −
+ +
THROTTLE2_trim 1 2
+ − 0.6 +
s
4 + Engine dyn. Integrator THROTTLE2
Lim
THROTTLE bandwidth Rate lim
THROTTLE2_c THROTTLE
CONSTANT WIND
1 +
From traj.
generator
Const. wind input w.r.t. FE
logic and scaling axis system
longitudinal WINDSHEAR
measurements +
2 Mux 1
simulation TOTAL
outputs WIND−
Windshear INPUT
w.r.t FB
Dryden TURBULENCE axis system
Turbulence
*
Product
Dryden
4
Turbulence
(1/0)
1 Demu
lon. −1
meas. WINDSHEAR DATA
positive = 0 Mux 1
downward HORIZONTAL
Constant Memory Windshear
−1 (break alge−
braic loop)
positive =
WINDSHEAR DATA downward
DOWNBURST
2 Demu
sim.
outp.
TRIMRCAM is the trimming and linearization routine supplied with the RCAM software.
When running this program, the following functions are executed:
initialization: the user may enter parameters to dene the trim condition and to set
other parameters for time simulations,
trimming: for this purpose a nine states (uB , vB , wB , p, q, r, , , ) nonlinear
RCAM model is called, for the actual trimming job the Simulink routine TRIM.M
is used,
linearization: a state-space model is obtained for small perturbations around the
equilibrium point, using the Simulink routine LINMOD.M,
initialization for simulations in the Simulink environment: the state vector is mod-
ied to set the specied initial position, track angle, crab angle (wind correction)
etc. These modications do not aect the equilibrium.
simulation: time responses of the linear and the nonlinear model (starting from the
trim point) are compared in plots. This enables the user to check the quality and
the scope of the linearization.
The main parameters that may be set to dene a trim point are:
the aircraft mass, m,
the center of gravity location, [xcg ycg zzg ]T ,
the airspeed, VA,
the
ightpath angle,
,
the inertial wind vector, [WXE ; WY E ; WZE ]T
These parameters may be chosen within their limits, these can be found in table 2.1.
Although the suplied trim algorithm can only be used for nding longitudinal equilibrium,
it is possible to correct for a constant wind vector. In gure B.8 the vectors of the
airspeed (Va ), the inertial velocity (V) and the windspeed (VW ) are drawn. The airspeed,
the
ightpath angle and the track angle have been specied for the trim condition. To
compensate for the vertical wind, the
ightpath angle w.r.t. to the airmass (
a ) has to
Version: 1
Date: April 4, 1997 128
GARTEUR/TP-088-27
xE
VW
VA VW
VAcos γa VW (projection)
γa V VW
χ
γ βw
Vcos γ
βw y
E
-z V
VWZE
VW
VAsin γ a Vsin γ
B.2.3 linearization
0 0 2
Q [deg/s]
0
PHI [deg]
ALPHA [deg]
BETA [deg]
linear
−0.5 linear −0.5 0
nonlinear
nonlinear
−2 −0.5
−1 −1 0 20 40 60 0 20 40 60
0 20 40 60 0 20 40 60
−6
x 10 80 8
1 60
60 6
0.5 40
40 4
0 20
20 2
VA [m/s]
R [deg/s]
PSI [deg]
GAMMA [deg]
−0.5 0 0 0
−1 −20 −20 −2
0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60
131
−6 −6 2 5
x 10 x 10
1 1
0 4
0.5 0.5 −2 3
−4 2
0 0
−6 1
DA [deg]
DR [deg]
in FE
WXE,WXB [m/s]
WYE,WYB [m/s]
−0.5 −0.5 −8 0
in FB
−10 −1
−1 −1 0 20 40 60 0 20 40 60
0 20 40 60 0 20 40 60
2 8 4
0 6 3
−2 4 2
−4 2 1
DT [deg]
WZE,WZB [m/s]
−6 0 0
THROTTLE1,2 [deg]
−8 −2 −1
0 20 40 60 0 20 40 60 0 20 40 60
GARTEUR/TP-088-27
Date: April 4, 1997
Version: 1
z 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
VA 0. 0. 0. 0. 0.0007 0. 0. 0.9962 0.0879 0. 0. 0. 0. 0. 0.0007 0.0007 -1.0000 0.0007 -0.9962 -0.0865
V 0. 0. 0.0000 0.0000 0.0007 0. 0. 0.9962 0.0879 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. 0.0000 0.0000 0.0143 0. 0. 0.0000 0.0000 0. 0. 0. 0. 0. -0.0143 -0.0143 0.0000 0.0000 0.0000 0.0000
p 1.0000 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
r 0. 1.0000 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. 1.0000 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
uV 0. 0. -0.0266 -3.4971 0. 0. -3.4974 0.9962 0.0872 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
vV 0. 0. -6.0926 69.8834 1.0000 0. 0.0000 0.0000 0.0000 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
y 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. -0.0870 1.0000 0.0143 0. 0.0000 0.0000 0.0000 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 1.0000 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 1.0000 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0. -0.0012 0.0142 0. 0. 0. 0. 0. 0. 0. 0. -0.0143 0.0012 -0.0142
0. 0. 0.0043 0.0000 0.0000 0. 1.0000 0.0012 -0.0142 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
x 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.
ny 0.6221 -7.1084 0.9935 0.0000 -0.0152 0. 0.0000 0.0000 0.0000 0. 0.1591 0. 0. 0. 0.0152 0.0152 0.0000 0.0000 0.0000 0.0000
Table C.3 RCAM state space model for nominal trim condition
GARTEUR/TP-088-27
Date: April 4, 1997
Version: 1
Version: 1
Date: April 4, 1997 134
GARTEUR/TP-088-27
Longitudinal model
Lateral model
Y = qmS CY L = qSb= 2 (Iz Cl + Ixz Cn )
Y_ = qmS b=V02 CY_ L_ = qSb= 2 b=2 (Iz Cl + Ixz Cn )
2 b=V02 _ _
Yp = qmS b=V02 CYp Lp = qSb=
V0 (Iz Clp + Ixz Cnp )
Yr = qmS b=V02 CYr Lr = qSb=
2 b=2 (Iz Cl + Ixz Cn )
2 (VI0z Cl + Ixz Cn )
r r
YA = qmS CYA LA = qSb=
2 A A
YR = qmS CYR LR = qSb=
(Iz ClR + Ixz CnR )
N = qSb= 2 (Ix Cn + Ixz Cl ) Nr = qSb= 2 b=2
2 b=2 (Ix Cn + Ixz Cl )
2 V0 (Ix Cnr + Ixz Clr )
N_ = qSb=
2 b=V02 _ _
Sb=
NA = q (Ix CnA + Ixz ClA )
Np = qSb= NR = qSb= 2
V0 ( I x C n p + I xz C l p) (Iz ClR + Ixz CnR )
q = 2 V02 = Ix Iz 2
Ixz
Version: 1
Date: April 4, 1997 136
GARTEUR/TP-088-27
V = 60 m=s - - - VA V = 60 m=s - - -
10
0
V = 70 m=s |{ V = 70 m=s |{
V = 80 m=s .....
2
10
V = 80 m=s .....
−1
10
0
10
q
Gain
Gain
−2
10
−2
10
−3
10
−4
−4 10
10
−5 −6
10 −4 −2 0 2 4
10 −4 −2 0 2 4
10 10 10 10 10 10 10 10 10 10
! (rad/s) ! (rad/s)
10
4
and VA , TH input
V = 60 m=s - - -
VA V = 70 m=s |{
V = 80 m=s .....
2
10
0
10
Gain
−2
10
−4
10
−6
10 −4 −2 0 2 4
10 10 10 10 10
! (rad/s)
Fig.E.1 Bode magnitude plots for varying airspeed
1 and q, T input 10
4
and VA , T input
VA
10
−1
10
0
10
q
Gain
Gain
−2
10
−2
10
−3
10
−4
−4 10
10
−5 −6
10 −4 −2 0 2 4
10 −4 −2 0 2 4
10 10 10 10 10 10 10 10 10 10
! (rad/s) ! (rad/s)
10
4
and VA , TH input
VA xcg = 0:15 c - - -
2 xcg = 0:23 c |{
10
xcg = 0:30 c .....
0
10
Gain
−2
10
−4
10
−6
10 −4 −2 0 2 4
10 10 10 10 10
! (rad/s)
Fig.E.2 Bode magnitude plots for varying center of gravity location
Version: 1
139 Date: April 4, 1997
GARTEUR/TP-088-27
1 and q, T input 10
4
and VA , T input
VA
10
10 104 kg - - - 10 104 kg - - -
0
12 104 kg |{ 12 104 kg |{
15 104 kg .....
10 2
15 104 kg .....
10
−1
10
0
10
q
Gain
Gain
−2
10
−2
10
−3
10
−4
−4 10
10
−5 −6
10 −4 −2 0 2 4
10 −4 −2 0 2 4
10 10 10 10 10 10 10 10 10 10
! (rad/s) ! (rad/s)
10
4
and VA , TH input
VA 10 104 kg - - -
2 12 104 kg |{
10
15 104 kg .....
0
10
Gain
−2
10
−4
10
−6
10 −4 −2 0 2 4
10 10 10 10 10
! (rad/s)
Fig.E.3 Bode magnitude plots for varying aircraft mass
10
1 p and , A input 10
1 p and , R input
10
0
10
0
−1
−1
10 10
p
Gain
Gain
p
−2 −2
10 10
−3
V = 60 m=s - - - −3
V = 60 m=s - - -
V = 70 m=s |{ V = 70 m=s |{
10 10
−4
V = 80 m=s ..... −4
V = 80 m=s .....
10 10
−5 −5
10 −4 −2 0 2
10 −4 −2 0 2
10 10 10 10 10 10 10 10
! (rad/s) ! (rad/s)
10
1 r and , A input 10
1 r and , R input
10
0
10
0
−1 −1
10 10
r r
Gain
Gain
−2 −2
10 10
−3
V = 60 m=s - - - −3
V = 60 m=s - - -
V = 70 m=s |{ V = 70 m=s |{
10 10
−4
V = 80 m=s ..... −4
V = 80 m=s .....
10 10
−5 −5
10 −4 −2 0 2
10 −4 −2 0 2
10 10 10 10 10 10 10 10
! (rad/s) ! (rad/s)
Fig.E.4 Bode magnitude plots for varying airspeed
Version: 1
Date: April 4, 1997 140
GARTEUR/TP-088-27
10
1 p and , A input 10
1 p and , R input
10
0
10
0
−1 −1
10 10
p
Gain
Gain
10
−2
p 10
−2
−3
xcg = 0:15 c - - - −3
xcg = 0:15 c - - -
xcg = 0:23 c |{ xcg = 0:23 c |{
10 10
−4
xcg = 0:30 c ..... −4
xcg = 0:30 c .....
10 10
−5 −5
10 −4 −2 0 2
10 −4 −2 0 2
10 10 10 10 10 10 10 10
! (rad/s) ! (rad/s)
10
1 r and , A input 10
1 r and , R input
10
0 10
0
r
−1 −1
10 10
r
Gain
Gain
−2 −2
10 10
−3
xcg = 0:15 c - - - −3
xcg = 0:15 c - - -
xcg = 0:23 c |{ xcg = 0:23 c |{
10 10
−4
xcg = 0:30 c ..... −4
xcg = 0:30 c .....
10 10
−5 −5
10 −4 −2 0 2
10 −4 −2 0 2
10 10 10 10 10 10 10 10
! (rad/s) ! (rad/s)
Fig.E.5 Bode magnitude plots for varying center of gravity location
10
1 p and , A input 10
1 p and , R input
0 0
10 10
−1 −1
10 10
p
Gain
Gain
10
−2
p 10
−2
−3
m = 10 1044 kg - - - −3
m = 10 1044 kg - - -
m = 12 104 kg |{ m = 12 104 kg |{
10 10
−4
m = 15 10 kg ..... −4
m = 15 10 kg .....
10 10
−5 −5
10 −4 −2 0 2
10 −4 −2 0 2
10 10 10 10 10 10 10 10
! (rad/s) ! (rad/s)
10
1 r and , A input 10
1 r and , R input
10
0 10
0
−1 −1
r
10 10
r
Gain
Gain
−2 −2
10 10
−3
m = 10 1044 kg - - - −3
m = 10 1044 kg - - -
m = 12 104 kg |{ m = 12 104 kg |{
10 10
−4
m = 15 10 kg ..... −4
m = 15 10 kg .....
10 10
−5 −5
10 −4 −2 0 2
10 −4 −2 0 2
10 10 10 10 10 10 10 10
! (rad/s) ! (rad/s)
Fig.E.6 Bode magnitude plots for varying aircraft mass
Version: 1
141 Date: April 4, 1997
GARTEUR/TP-088-27
10
2 and q,
and VA , WXE input 10
2 , q,
and VA , WZE input
10
0
VA 10
0
VA
−2
−2
10 10
Gain
Gain
−4 −4
10 10
10
−6 10
−6
q
q
−8 −8
10 −4 −2 0 2 4
10 −4 −2 0 2 4
10 10 10 10 10 10 10 10 10 10
! (rad/s) ! (rad/s)
10
0 p, r, and , WY E input
−1
10
10
−2
Gain
p
−3
10
r
−4
10
10
−5
−6
10 −4 −2 0 2
10 10 10 10
! (rad/s)
Fig.E.7 Bode magnitude plots for disturbance inputs
Version: 1
Date: April 4, 1997 142
GARTEUR/TP-088-27
F RCAM simulations
In this appendix gures with time reponses of the linear and the nonlinear RCAM model
to control inputs and wind disturbances can be found. The input for each simulation is
given in the rst subgure. All other inputs or de
ections w.r.t. trim values are zero.
All simulations start from the same trim conditions, see table C.1 and table C.2.
−3
x 10
2 15
1.5 10
DA [deg]
NY
1 5
0.5 0
0 −5
0 50 100 0 50 100
time [s] time [s]
0 0
R [deg/s]
P [deg/s]
−0.2 −0.1
−0.4 −0.2
0 50 100 0 50 100
time [s] time [s]
0.5 0.2
0
0
BETA [deg]
−0.5
PHI [deg]
−1 −0.2
−1.5
−0.4
−2
−2.5 −0.6
0 50 100 0 50 100
time [s] time [s]
5 1000
0 999.5
CHI, PSI [deg]
−5
999
H [m]
−10
998.5
−15
−20 998
−25 997.5
0 50 100 0 50 100
time [s] time [s]
−5 0.1
−6
0.05
DT [deg]
−7
NZ
0
−8
−0.05
−9
−10 −0.1
0 50 100 0 50 100
time [s] time [s]
1.5 9
1 ALPHA [deg] 8
Q [deg/s]
0.5
7
0
6
−0.5
−1 5
0 50 100 0 50 100
time [s] time [s]
12 6
10 4
GAMMA [deg]
THETA [deg]
8 2
6 0
4 −2
2 −4
0 50 100 0 50 100
time [s] time [s]
75 1080
1060
70
VA, V [m/s]
1040
H [m]
65
1020
60
1000
55 980
0 50 100 0 50 100
time [s] time [s]
2 0.01
0.005
1.5
DR [deg]
NY
1
−0.005
0.5
−0.01
0 −0.015
0 50 100 0 50 100
time [s] time [s]
0.5 0.2
0
0
R [deg/s]
P [deg/s]
−0.2
−0.5
−0.4
−1 −0.6
0 50 100 0 50 100
time [s] time [s]
0
0.5
BETA [deg]
−1
PHI [deg]
−2
−3 0
−4
−5
0 50 100 0 50 100
time [s] time [s]
20 1001
1000
0
CHI, PSI [deg]
999
H [m]
−20 998
997
−40
996
−60 995
0 50 100 0 50 100
time [s] time [s]
10 0.05
THROTTLE1,2 [deg]
NZ
0
4
0 −0.05
0 50 100 0 50 100
time [s] time [s]
0.4 6.5
0.2
ALPHA [deg]
6
Q [deg/s]
5.5
−0.2
−0.4 5
0 50 100 0 50 100
time [s] time [s]
11 5
10 4
GAMMA [deg]
THETA [deg]
9 3
8 2
7 1
6 0
5 −1
0 50 100 0 50 100
time [s] time [s]
71 1300
70 1250
VA, V [m/s]
69 1200
H [m]
68 1150
67 1100
66 1050
65 1000
0 50 100 0 50 100
time [s] time [s]
2 0.03
0.02
1.5
WXE [m/s]
0.01
NZ
1
0
0.5
−0.01
0 −0.02
0 50 100 0 50 100
time [s] time [s]
5.3
0.1 5.2
ALPHA [deg]
Q [deg/s]
5.1
0
5
−0.1 4.9
4.8
0 50 100 0 50 100
time [s] time [s]
6 1
5.5 0.5
GAMMA [deg]
THETA [deg]
5 0
4.5 −0.5
4 −1
0 50 100 0 50 100
time [s] time [s]
72 1000
71.5
V
VA, V [m/s]
71 995
H [m]
70.5
70 990
69.5 VA
69 985
0 50 100 0 50 100
time [s] time [s]
2 0.02
0.015
1.5
0.01
WYE [m/s]
NY
1 0.005
0
0.5
−0.005
0 −0.01
0 50 100 0 50 100
time [s] time [s]
1 0.1
0.05
0.5
R [deg/s]
P [deg/s]
0
−0.05
−0.5 −0.1
0 50 100 0 50 100
time [s] time [s]
1.5 0.5
1
BETA [deg]
0
PHI [deg]
0.5
−0.5
0
−0.5 −1
0 50 100 0 50 100
time [s] time [s]
1 1000.3
1000.2
CHI, PSI [deg]
0.5
1000.1
H [m]
1000
0
999.9
−0.5 999.8
0 50 100 0 50 100
time [s] time [s]
2 0.06
1.5 0.04
WZE [m/s]
NZ
1 0.02
0.5 0
0 −0.02
0 50 100 0 50 100
time [s] time [s]
0.4
0.3 5
ALPHA [deg]
Q [deg/s]
0.2
0.1 4.5
−0.1 4
0 50 100 0 50 100
time [s] time [s]
6 0.5
0
GAMMA [deg]
THETA [deg]
5.5
−0.5
5
−1
4.5 −1.5
0 50 100 0 50 100
time [s] time [s]
1000
70.5
980
VA, V [m/s]
960
H [m]
70
940
920
69.5
900
0 50 100 0 50 100
time [s] time [s]
thrust setting
6
0.4
load fact. nx (|), nz (- - -)
5 0.3
0.2
TH 1;2 (deg)
4
0.1
nx ; nz (-)
3 0
−0.1
2
−0.2
−0.3
1
−0.4
0 −0.5
0 20 40 60 80 0 20 40 60 80
4
pitch rate 40
(|), (- - -) and
(...)
3 30
2
20
; ;
(deg)
1
q (deg/s)
10
0
−1 0
−2
−10
−3
−20
−4
−5 −30
0 20 40 60 80 0 20 40 60 80
110
VA (|) and V (- - -) size altitude
1000
100 950
900
90
VA ; V (m/s)
850
80
h (m)
800
70 750
700
60
650
50
600
40 550
0 20 40 60 80 0 20 40 60 80
0.2
ny (|), nz (- - -) ang. rates p (|), q (- - -), r (...)
8
0.15
6
0.1
p; q; r (deg/s)
4
ny ; nz (-)
0.05
2
0
0
−0.05
−0.1 −2
−0.15 −4
−0.2 −6
0 20 40 60 80 0 20 40 60 80
80
; ; (deg)
40
60
(deg)
30
40
20
20 10
0 0
−20 −10
0 20 40 60 80 0 20 40 60 80
10 0.1
8
WXE ; WZE (m/s)
0.05
6
nz (-)
0
2
−0.05
0
−2 −0.1
−4
−0.15
−6
−8 −0.2
0 50 100 150 0 50 100 150
15
1.5
1 10
; ;
(deg)
0.5
5
q (deg/s)
0
0
−0.5
−1 −5
−1.5
−10
−2
−15
−2.5
−3 −20
0 50 100 150 0 50 100 150
95
VA (|) and V (- - -) altitude
700
90 650
85 600
VA ; V (m/s)
80 550
h (m)
75 500
70 450
65 400
60 350
55 300
50 250
0 50 100 150 0 50 100 150