GARTEUR Open
by
FM(AG08)
GARTEUR Open
by
FM(AG08)
List of authors
Chapter 1: Jan Terlouw1, Stefano Scala6
Chapter 2: Ewan Muir2, Dieter Moormann3
Chapter 3: Dieter Moormann
Chapter 4: Ewan Muir, Paul Lambrechts1
Chapter 5: Rick Hyde5
Chapter 6: Paul Lambrechts, Samir Bennani4
Appendix A: Dieter Moormann, Ewan Muir
Appendix B: Paul Lambrechts
Appendix C: Jan Schuring, Paul Lambrechts
Appendix D: Ewan Muir
Appendix E: Ewan Muir
Appendix F: Chris Fielding7, Robert Luckner8
Summary
This document denes one of the two Robust Flight Control design challenges prepared
by GARTEUR Action Group FM(AG08). This design challenge consists of the imple-
mentation of a wide envelope
ight control law for the High Incidence Research Model
(HIRM). A non-linear aircraft model is supplied for design and simulation. A realistic set
of design specications for performance and robustness is formulated. Although the set of
specications is limited, the problem formulation is representative of a real world control
law design.
A fully automated evaluation procedure is developed to be able to compare the perfor-
mance of the resulting controllers, irrespective of the methods used to design them. Design
challenge participants are expected to supply insight into the usefulness of their methods,
not only in the sense of controller performance, but also in the sense of controller com-
plexity and design time. The latter is considered to be of particular importance for the
aircraft industry.
Version: 3
iii Date: April 4, 1997
GARTEUR/TP-088-4
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 (IT) INTA 1 copy
Rodlo R (DE) DLR 1 copy
Verbrugge R (FR) ONERA 1 copy
Version: 3
Date: April 4, 1997 iv
GARTEUR/TP-088-4
Contents
List of gures
2.1 HIRM conguration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Simulink block diagram of HIRM aircraft system . . . . . . . . . . . . . 6
2.3 Dynamics objects of HIRM aircraft model inside block hirmex of Fig. 2.2 7
2.4 Coordinate transformation body xed , vehicle carried . . . . . . . . . 12
2.5 Engine model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.6 Sensor hardware assumptions. . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1 Object diagram of HIRM . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.2 Aircraft model library . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.3 Structure of aircraft physical model . . . . . . . . . . . . . . . . . . . . . 28
3.4 HIRM Engine Dynamics model . . . . . . . . . . . . . . . . . . . . . . . 35
3.5 Control block library . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.6 Aircraft Gravity Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.7 From system conguration to symbolic model or simulation model . . . 39
4.1 Closed loop system showing point for analysis. . . . . . . . . . . . . . . 43
4.2 Gain and phase osets excluding tolerances . . . . . . . . . . . . . . . . 46
4.3 Gain and phase osets including tolerances . . . . . . . . . . . . . . . . 46
4.4 Gain and phase exclusion zones on Nichols plot for single loop analysis
frequency response excluding tolerances. . . . . . . . . . . . . . . . . . . 47
4.5 Gain and phase exclusion zones on Nichols plot for single loop analysis
frequency response including tolerances. . . . . . . . . . . . . . . . . . . 47
4.6 Pitch/roll stick force to pitch attitude/roll angle frequency response crite-
rion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.7 Phase rate criterion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.8 Pitch stick force to pitch attitude frequency response criterion. . . . . . 49
4.9 Roll stick force to roll angle frequency response criterion. . . . . . . . . 49
4.10 Denition of dropback. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.11 Sideslip requirements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
A.1 Simulink design model hirm des.m . . . . . . . . . . . . . . . . . . . . . 69
D.1 Response to step in symmetric tailplane. . . . . . . . . . . . . . . . . . . 78
D.2 Response to step in dierential tailplane. . . . . . . . . . . . . . . . . . . 79
D.3 Response to step in symmetric canard. . . . . . . . . . . . . . . . . . . . 79
D.4 Response to step in dierential canard. . . . . . . . . . . . . . . . . . . . 80
D.5 Response to step in rudder. . . . . . . . . . . . . . . . . . . . . . . . . . 80
D.6 Response to step in throttle 1. . . . . . . . . . . . . . . . . . . . . . . . 81
E.1 Lift coecient versus angle of attack . . . . . . . . . . . . . . . . . . . . 82
E.2 Pitching moment coecient versus angle of attack . . . . . . . . . . . . 83
E.3 Yawing moment coecient versus sideslip . . . . . . . . . . . . . . . . . 83
Version: 3
ix Date: April 4, 1997
GARTEUR/TP-088-4
List of tables
2.1 Aircraft model inputs denition . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Aircraft model states denition . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Aircraft model outputs denition . . . . . . . . . . . . . . . . . . . . . . 9
2.4 Aircraft model parameter denition . . . . . . . . . . . . . . . . . . . . 10
2.5 Parameter uncertainty denition . . . . . . . . . . . . . . . . . . . . . . 10
2.6 Relation between throttle and thrust . . . . . . . . . . . . . . . . . . . . 16
2.7 Actuator inputs denition . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.8 Actuator states denition . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.9 Stick characteristics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.1 Matrix element corresponding to modelling error . . . . . . . . . . . . . 42
D.1 Amplitude of step and gure number for each response . . . . . . . . . . 77
D.2 Legends for graphs in gures D.1 to D.6 . . . . . . . . . . . . . . . . . . 78
Version: 3
Date: April 4, 1997 x
GARTEUR/TP-088-4
Abbreviations
ALN Alenia Aeronautica
AVRO Avro International Aerospace
BAeDD MATRA - British Aerospace Defence Dynamics
BAe-MA British Aerospace Military Aircraft
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
ES Spain
FFA The Aeronautical Research Institute of Sweden
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
INTA Instituto Nacional de Tecnica Aerospacial
IT Italy
LAAS Laboratoire d'Analyse et d'Architecture des Systemes
LiTH Linkoping University
LUT Loughborough University
NL The Netherlands
NLR National Aerospace Laboratory
ONERA Oce National d'Etudes et de Recherches Aerospatiales
SE Sweden
Version: 3
xi Date: April 4, 1997
GARTEUR/TP-088-4
1 Introduction
1.1 Objectives of GARTEUR Action Group FM(AG08)
In this document a Robust Flight Control design benchmark problem is proposed. It has
been prepared by GARTEUR Action Group FM(AG08) on \Robust Flight Control (RFC)
in a Computational Aircraft Control Engineering Environment (CACEE)". The objectives
and activities of this Action Group will be discussed in the following. More details can
be found in the FM(AG08) Terms of Reference [2] and in the FM(AG08) Project Docu-
ment [3].
Achievement of this long term goal requires close collaboration between the major aero-
nautical disciplines: Aerodynamics; Structures; Propulsion; Guidance, Navigation and
Control. Bearing this in mind, the
ight control engineering discipline should utilise and
elaborate controller analysis and design methodology suitable for multi-disciplinary con-
siderations. Robust control methodology has this potential and is therefore the main focus
of FM(AG08).
A major problem facing designers of Flight Control Systems (FCS) is uncertainty in char-
acterising not only the vehicle itself, but also the environment in which it must operate.
Gain scheduling is often necessary because of the variation of characteristics for which the
control laws must guarantee stability and performance. The design of gain scheduling is
time consuming for two reasons: the control laws must be designed at each design point,
and a great deal of assessment is required to ensure adequate stability and performance
at o-design points.
Recent advances in control theory research has given rise to a number of novel robust
control techniques [5, 6] specically developed for dealing with model uncertainties and
parameter variations. These new techniques oer potential benets to a control law de-
signer for modern aircraft in the following ways:
Version: 3
Date: April 4, 1997 2
GARTEUR/TP-088-4
However, robust control techniques are seldom used by European aircraft manufactur-
ers for the design of FCS. There are three main reasons for this:
The application of robust control theory to the aircraft control law design problem
has not been demonstrated. The techniques and algorithms associated with robust
control theory are clearly expressed but do not, in their current form, lend themselves
to direct FCS application.
There are a limited number of dedicated robust control design tools, while most
manufacturers have an extensive suite of classical control design tools that they
have developed over a period of several years.
There is no specic bibliographic source available on robust control techniques. Con-
sequently, a lot of time has to be spent in searching for appropriate references in a
variety of widely distributed libraries, journals and general purpose data-bases.
Subproject FM(AG08)-3 aims at removing these drawbacks and at demonstrating to Eu-
ropean aircraft manufacturers that a signicant improvement in the overall design process
is possible by using robust control techniques. In a greater detail, the aim is:
To identify and apply existing and new controller design methods to robust control
problems that are representative of operational industrial needs [10].
Version: 3
3 Date: April 4, 1997
GARTEUR/TP-088-4
To introduce robust controller design and analysis methods into the control law
design cycle, in order to cope more eciently with uncertainty in the models used
and with operational changes that may occur.
To identify tools which can be used in conjunction with multi-disciplinary design
optimization to improve overall dynamic system performance.
To develop robust controller design procedures that interface with industrial require-
ments.
To achieve these objectives, FM(AG08) has opted for the following research approach.
Two robust
ight control benchmark problems have been dened, which will be solved
by design teams from the European aeronautical industry, research establishments and
universities. A wide variety of modern and classical design methods will be applied. The
controllers that are designed in response to these problems will be compared and eval-
uated. It is proposed that a nal conference will be held in Toulouse (France) by the
end of 1996, where the controllers and the results of the comparisons will be presented.
However, it must be stated that the aim of these benchmark problems is not to produce an
optimal control law, but to demonstrate how robust
ight control theory can be applied
to realistic problems and also to demonstrate the limitations of such techniques. It is also
intended that these benchmarks will raise the awareness and condence of the European
aeronautical industry in the use of robust control techniques.
The two benchmarks cover respectively an automatic landing control problem and a high
angle of attack enhanced manual control problem. This document is the manual for
the second problem, which will be referred to as the HIRM (High Incidence Research
Model) benchmark. The main objective of the HIRM design study is the design of a
control augmentation system for HIRM. The aim of the
ight control system is to give
good handling qualities across the specied
ight envelope and also provide robustness
to unmodelled plant dynamics, modelling uncertainties and variations in operating point
within the
ight envelope. Acceptable noise and disturbance rejection must also be demon-
strated. The HIRM mathematical model represents an aircraft typical of modern combat
aircraft. The problem is wirtten to provide an aerospace control law benchmark, which
is both concise and bounded, with a set of representative design aims for both robustness
and handling qualities. At the same time, it contains most of the elements aecting the
implementation of control laws in real systems and which designers need to allow for.
1.3 Contents of this document
The structure of the document is as follows:
In chapter 2, a description of the HIRM model is given, in which analytical expres-
sions for all the variables of interest, states, inputs and outputs of the system, are
Version: 3
Date: April 4, 1997 4
GARTEUR/TP-088-4
The HIRM mathematical model used for this design study was derived by scaling up the
data obtained from the drop model to create an aircraft of F-18 proportions. Engine and
actuator models have been added to create a representative, non-linear simulation of a twin
engined, modern ghter aircraft. The aircraft is basically stable both longitudinally and
laterally. There are however combinations of angle of attack and control surface de
ection
which cause the aircraft to be unstable longitudinally and/or laterally.
Version: 3
Date: April 4, 1997 6
GARTEUR/TP-088-4
Before starting a
simulation
symmetrical taileron (dtsc) load ini****.mat
0 of desired flight condition time
U0 in the workspace
differential taileron (dtdc) Clock Time
trim_inputs Double click here
0 to actuator for more information
symmetrical canard (dcsc) hirmsim
outputs for simulation
0
differential canard (dcdc) Mux +
+ hirmexa hirmexs hirmy
− Mux hirmex Demux
0 measured variables
Sum_inputs1 Actuator Mux Demux Sensor
rudder (drc) model HIRM model model
(16 states)
0 wind
nose suction (suctionc) wind_input
atmosphereConst
( as a function of height )
control
aerodynamic
inputs
calcairspeed
(calculate airspeed)
outputs
engine_1
for system
body
analysis
Equations of motion
engine_2
for control
BodyFixed BodyFixed
gravitationConst wind
( g = 9.81 m/s 2 )
wind
inputs
Fig.2.3 Dynamics objects of HIRM aircraft model inside block hirmex of Fig. 2.2
engine1 and engine2 describe the relevant engine behaviour (see x2.3.6).
atmosphere describes the atmosphere model (see x2.3.7).
gravity describes the gravitational in
uence (see x2.3.8).
2.3.1 Nomenclature: Inputs, States, Outputs, Parameters
The following tables summarizes the adopted nomenclature used both for the formulation
of the problem and the naming of variables in the software. Additional information can
be found in Appendix A of this document and in Ref. [1].
The inputs are given in Table 2.1.
The state variables are given in Table 2.2.
The outputs are given in Table 2.3.
The parameters are detailed in Table 2.4.
Uncertainty is assumed in the aerodynamic parameters as detailed in Table 2.5. Note that
the magnitudes of these uncertainties can only be changed by editing the HIRM software
Version: 3
Date: April 4, 1997 8
GARTEUR/TP-088-4
The equations for the translational movement in bodyxed coordinates are derived from
the force equation:
F = m ( aB + ! x VB )
F is the sum of forces due to the engine, the aerodynamics and gravity, m is the mass of
the aircraft, VB is the airspeed and ! is the rotation rate in body-xed coordinates. The
acceleration aB (in body-xed system) is the time derivative of velocity:
2 3
6 u7
d V d
aB = dt = dt 4 v 75
B 6
w
and the velocity is the time derivative of the position vector expressed in the vehicle-carried
(earth) vertical frame:
2 3
6 x7
d X d
VV = dt = dt 4 y 75
V 6
z
Furthermore the following aircraft specic quantities are dened. The normal load factor
nz is dened by:
nz = agnz
where anz is the z body-axis accelerometer output at the Centre of Gravity.
The height h is the negative z coordinate in the vehicle-carried system:
h= z
The
ight path angle
is calculated from:
tan
= p 2wV 2
uV + vV
given as a function of the speed components in vehicle-carried coordinates.
The track angle is given by:
tan = uvV :
V
2.3.2.2 Rotational motion
The equations for the rotational movement in body-xed coordinates are derived from the
moments equation:
M = I !_ + ! x I !
where M is the sum of moments w.r.t. the Centre of Gravity due to the engine and
the aerodynamics, ! is the rotational velocity, and !_ is the rotational acceleration in
body-xed system:
2 3 2 3
6 p_ 7 6 p7
6 7 d
!_ = 4 q_ 5 = dt 4 q 75
6
r_ r
Version: 3
Date: April 4, 1997 12
GARTEUR/TP-088-4
The relation between the rotational velocities and the Euler Angles is:
2 _3 2 32 3
1 sin tan cos tan 77 66 p 77
d = 66 _ 77 = 66 0 cos sin 5 4 q 5
dt 4 _ 5 4
0 sin = cos cos = cos r
The aircraft inertia tensor I dened in the body frame is:
2 3
6 I x 0 Ixz 7
I = 64 0 Iy 0 75
Ixz 0 Iz
2.3.3 Coordinate transformation (Body-Fixed , Vehicle-Carried)
The rotations between the body-xed and the vehicle-carried coordinate systems are de-
picted in Fig. 2.4.
To describe the orientation of the aircraft, a transformation using the three Euler angles ,
, and becomes necessary. For the transformation the vehicle-carried system is rotated
about the z -axis by the heading angle . The next rotation is done by the pitch angles
about k2 and nally by the roll angle about the xf -axis.
The transformation matrix between body-xed and vehicle-carried axis system results in:
2 3 2 3 2 3
6 1 0 0 7 6 cos 0 sin 7 6 cos sin 0 7
RBV = RVT B 64 0 cos sin 75 64 0 1 0 75 64 sin cos 0 75
0 sin cos sin 0 cos 0 0 1
Multiplying the three matrices yields:
2 3
6 cos cos sin cos sin 7
RBV = 64 cos sin sin sin cos sin sin sin + cos cos cos sin 75
cos sin cos + sin sin sin sin cos cos sin cos cos
Version: 3
13 Date: April 4, 1997
GARTEUR/TP-088-4
For example, the transformation of velocities between vehicle-carried (index V) and body-
xed (index B) coordinates is:
VB = RBV VV
with
2 3 2 3
6 u7 6 uV 7
VB = 64 v 75 and VV = 64 vV 75
w wV
Similarly, the accelerations, rotational velocities, position, forces and moments can be
transformed between the coordinate systems.
The vector airspeed Va (expressed in body axes) is the dierence between the inertial
velocity of the aircraft VB , and the wind velocity WB , expressed in body-xed coordinates:
Va = VB WB
with
2 3
6 ua 7
Va = 4 va 75 ;
6
wa
V is the total velocity:
q
V = (ua 2 + va 2 + wa 2 ) :
The angle of attack and the angle of sideslip are dened as:
tan = wu a
a
sin = vVa
2.3.5 Aerodynamics
The aerodynamic forces FxA , FyA , FzA are given in body-xed coordinates as functions of
the aerodynamic factor qS and the aerodynamic coecients CX , CY , CZ .
FxA = +qS CX
FyA = qS CY
FzA = +qS CZ
The aerodynamic moments LA , MA , NA are given in body-xed coordinates as functions
of the aerodynamic factor qS and the span b or the mean aerodynamic chord c and the
aerodynamic coecients Cl , Cm , Cn .
LA = qS b Cl
MA = qS c Cm
NA = qS b Cn
The dynamic pressure q is:
q = 21 V 2
with V the total airspeed.
The Mach-number M is the quotient of total airspeed V and local speed of sound a:
M = Va
In the following equations the aerodynamic force and moment coecients are given. Each
component has the form Cab (c; d). Its value is computed by linearly interpolating between
the values given in a look-up table as a function of the variables c and d.
CX = CXTS (; TS ) + CXCS (; TS ) CS
CY = CYR () R + CYCS (; ) (CS + ) + CYi (; ) + CYTS (; ) )(TS + 10 + )
20
180 20
+CYr () 2r Vb + CYCD (; CS ) CD + CYp () 2p Vb + CYTD (; TS ) TD + CYsu (; suction)
CZ = CZTS (; TS ) + CZCS (; TS ) CS + CZq (; CS ) 2q Vc
Cm = Cmh (; TS ) + CmCS (; TS ) CS + Cmq (; CS ) 2q Vc + Cmsu (; suction)
Cl = ClR () R + ClCS (; ) (CS + ) + ClTD (; TS ) TD + Cli (; )
) + C ()( + 10 ) r b
20
(
= 010 20
180 > 20
20
Because of the step change in some coecients at 20 angle of attack, designers should
not evaluate their control laws at this value of .
2.3.6 Engine model
Each engine is modelled as shown in Fig. 2.5.
The throttle position is bounded between 0 and 2 in the actuator model. The demanded
thrust FE is a function of the throttle position as specied below:
(
FE = IDLE + (MAXDRY IDLE ) TH TH 2 [0; 1]
MAXDRY + (MAXREHEAT MAXDRY ) (TH 1) TH 2 [1; 2]
The values of the IDLE, MAXDRY, and MAXREHEAT thrusts for each engine are shown
in Table 2.6. Note that the throttle setting is continuously variable between 0 and 2.
Version: 3
Date: April 4, 1997 16
GARTEUR/TP-088-4
If the rate limit is ignored, the engine dynamics can be represented by a second order
transfer function with a damping of 1 and a natural frequency of 2.
The rate limit varies depending on whether the engine is in dry thrust or in reheat.
If the engine is in dry thrust ( 47 kN), the rate limit equals 12000 N/s.
If the engine is in reheat (> 47 kN), the rate limit equals 25000 N/s.
The sea level engine thrust FE 0 is scaled with density as follows:
FE = FE0 :
0
The engine setting angles are zero and so the thrust acts parallel to the aircraft x-body
axis. Therefore:
FEx = FE ; FEy = 0; FEz = 0
The position of the point of application of thrust of each engine w.r.t. center of gravity is
given by the parameters XATP , YATP , and ZATP . Their values are shown also in Table
2.4.
The moments about the aircraft centre of gravity produced by each engine are calculated
from:
LT = YATP FEz ZATP FEy
MT = ZATP FEx XATP FEz
NT = XATP FEy YATP FEx
2.3.7 Atmosphere
The atmospheric conditions are considered to be a function of height h:
= (h)
a = a(h)
and are calculated from the following equations:
= RpT
p
a = RT
Version: 3
17 Date: April 4, 1997
GARTEUR/TP-088-4
where
T = T80 Tgrad h
< p0 TT R T
> g
h < 11000 m
grad
p = > 0
gh
: p0 exp R T (
h 11000 m
11000)
0
(
Tgrad = 0:0065 K=m h < 11000 m
0:0 K=m h 11000 m
(
T0 = 288:15 K h < 11000 m
216:65 K h 11000 m
(
p0 = 101325:0 Pa h < 11000 m
22632:0 Pa h 11000 m :
In the equations above, is the density of air, a the speed of sound in air, p the
static pressure, T the absolute temperature, Tgrad its rate of change w.r.t. height.
R = 287:05287m2 =s2 =K is the universal gas constant, g = 9:80665m=s2 the vertical
gravity acceleration at h=0, and = 1:4 is the specic heat ratio. The subscript 0 for
temperature and pressure represents the starting point for the interpolation, for heights
of either h = 0 m or h = 11000 m.
2.3.8 Gravity
symmetric taileron
engine throttle y.
2.4.2 Nomenclature: Inputs, States, Outputs, Parameters
The actuator inputs and states are listed in Tables 2.7 and 2.8 respectively. The actuators
outputs are the same as the rst eight model inputs shown in Table 2.1.
Name Unit
x(1) = rst state of starboard tailplane
actuator rad
x(2) = second state of starboard tailplane
actuator rad/sec
x(3) = third state of starboard tailplane
actuator rad/sec2
x(4) = rst state of port tailplane actuator
rad
x(5) = second state of port tailplane actuator
rad/sec
x(6) = third state of port tailplane actuator
rad/sec2
x(7) = rst state of starboard canard
actuator rad
x(8) = second state of starboard canard
actuator rad/sec
x(9) = rst state of port canard actuator
rad
x(10) = second state of port canard actuator
rad/sec
x(11) = rst state of rudder rad
x(12) = second state of rudder rad/sec
x(13) = state of nose suction -
The rudder actuator is represented by a second order transfer function with a 80 =s rate
limit:
1
(1 + 0:0191401s + 0:000192367s2 ) :
De
ection limits are 30 . A positive de
ection is dened as trailing edge to port.
The canard consists out of a starboard canard and a port canard, which can be controlled
independently. The following transfer function is valid for both parts of the canard.
The canard actuator is represented by a second order transfer function with a 80 =s rate
limit:
1
(1 + 0:0157333s + 0:00017778s2 ) :
De
ection limits for starboard and port canard are +10 to 20 . A positive de
ection is
dened as trailing edge down.
Version: 3
Date: April 4, 1997 20
GARTEUR/TP-088-4
The inputs of the canard actuator model are the dierential canard demand CDd and
symmetrical canard demand CS d .
The relation between the dierential and symmetrical canard de
ections and the canard
port Cp and starboard Cs de
ections is given by the following equations:
Cp = CS + CD
Cs = CS CD
The above given transfer function is applied to the starboard and port canard.
The outputs are the limited dierential and symmetrical canard de
ections, which are
calculated from the limited starboard and port de
ections.
2.5 Sensor models
2.5.1 Available sensors
The following sensor information can be used:
Body axis angular rates, p, q and r.
Body axis attitudes, , and .
Body axis accelerations, ax , ay and az .
Airspeed.
Mach number.
Altitude.
Angles of attack and sideslip, and .
2.5.2 Sensor dynamics and signal conditioning
The linearized models of sensor dynamics and signal conditioning devices are given below.
Additional dynamics on the air data and attitude information should be modelled as a
combination of:
anti-aliasing lter:
1=(1 + 0:00398s + 0:0000158s2 );
rate and acceleration sensors, combined sensor and notch lter:
(1 0:005346s + 0:0001903s2 )=(1 + 0:03082s + 0:0004942s2 );
compute delay:
(1 0:0062s)=(1 + 0:0062s);
D/A converter:
(1 0:00208s)=(1 + 0:00417s):
Version: 3
Date: April 4, 1997 22
GARTEUR/TP-088-4
Control law designers will need to select an appropriate gain between stick de
ection and
the demand in =s to their control law.
2.7 Atmospheric turbulence models
Since the design envelope in x 4.3.1 includes altitudes from 30 to 7000 m, atmospheric
turbulence models for both low and medium to high altitudes will be considered [11].
The use of Dryden spectral models is suggested because of their relatively simple applica-
tion. The spectra for the three turbulence components read:
ug (
) = u2 2L u (1 + (L1
)2 )
u
vg (
) = v2 2L v (11 + 12(Lv
)2
+ 4(L
)2 )2
v
wg (
) = w2 2Lw (11 + 12(Lw
)2
+ 4(L
)2 )2
w
where
is the spatial frequency in rad/m. The parameters are the scale lengths Lu , Lv ,
Lw and the standard deviations u , v , w . For low altitudes the scale lengths are dened
by:
9
Lu = 2 Lv = (0:177+0:h00274h) : =
; 3 < h < 300 m
12
Lw = h2
For medium to high altitudes the scale lengths are dened by:
Lu = 2 Lv = 2 Lw = 530 m h > 600 m
For the missing part from 300 to 600 m altitude linear interpolation is suggested:
Lu = 2 Lv = 2 Lw = 70 + 0:766 h 300 < h < 600 m
The turbulence standard deviations are generally dened in statistical terms. Common
indications with probability of exceedance are:
Version: 3
Date: April 4, 1997 24
GARTEUR/TP-088-4
light 10 2
moderate 10 3
severe 10 5
For low altitude the standard deviations for the given statistics are dened by:
9
u
w
1
= wv = (0:177+0:00274 >
h) : >
w = 0:8 m=s light
04
>
=
> h < 300 m
w = 1:6 m=s moderate >
w = 2:3 m=s severe >
;
For medium to high altitudes the standard deviations for the three components are equal
to each other. The dependency from altitude for the given statistics is dened by:
light
= 1:55 m=s 600 h 2800 m
= 2:32 0:000274 h m=s 2800 < h < 5100 m
= 0:92 m=s h 5100 m
moderate
= 3:05 m=s 600 h 3400 m
= 3:84 0:000234 h m=s h > 3400 m
severe
= 3:04 + 0:00244 h m=s 600 < h < 1400 m
= 6:45 m=s 1400 h 5800 m
= 8:40 0:000336 h m=s h > 5800 m
For the missing part from 300 to 600 m altitude linear interpolation is suggested:
= 0:05 + 0:0025 h light
= 0:15 + 0:00483 h moderate
= 0:1 + 0:00733 h severe
Version: 3
25 Date: April 4, 1997
GARTEUR/TP-088-4
atmosphere
body6DOF
COG
u atmos
HIRM
engine1
aerodynamics
HIRM
g
engine2
Earth
HIRM gravity gust
Each of these phenomena is most conveniently described as one physical object. All objects
are connected according to Fig. 3.1 to represent the interactions within an aircraft.
In order to make the understanding of the objects easy, each component is described in
RCAM data provided from CERT-ONERA/Aerospatiale (FR)
y HIRM data provided from DRA (UK)
Version: 3
Date: April 4, 1997 26
GARTEUR/TP-088-4
its own coordinate system. Gravity, wind, and atmosphere are conveniently described
in an earth related coordinate system, aerodynamics in an wind coordinate system, and
engines in a system which is related to the body-xed coordinate system. Hence coordinate
transformations and an object to describe the relationship between velocity, wind, and
airspeed are needed in between all of these subsystems when they are connected. Therefore,
in addition to the basic aircraft components, coordinate transformations are also detailed
and handled as objects in the aircraft library, Fig. 3.2.
Earth Earth
Engine wind gust atmosphere atmos1D
In the physical aircraft library, dierent representations of one component can be found.
There is a class Body with six degrees of freedom (Body6DOF) and a class with three
degrees of freedom (BodyLong), which can be used to generate a nonlinear simulation
model for the longitudinal axis only. There are also engine, atmospheric and gravity
models of dierent complexity.
The interconnection structure of an aircraft can be most easily understood in a graphical
view (Fig. 3.1). If a more complex gravity model acts on the aircraft, this object can
simply be taken from the aircraft library to replace the simple gravity object. In the same
way one or more engines can be added or removed from the aircraft or can be modied.
This is the most transparent user layer with no need to think about the structure of any
specic simulation code.
The objects which form the physical model contain equations (and not assignments as
common in programming or simulation languages). This makes the understanding and
the reuse much easier than looking at low level code, whose purpose is to be understood
by a computer. Once the objects are available in computer readable form the object
equations can be sorted automatically by a symbolic equation handler. This is a main
feature of Dymola.
Objects, formulated in that way do not necessarily have to represent causalities. This
allows one object to fulll dierent tasks. For example, the object which does the trans-
formations between the body-xed and the aerodynamic coordinate system, is used for the
transformation of the velocities from the body-xed system to the aerodynamic system,
Version: 3
27 Date: April 4, 1997
GARTEUR/TP-088-4
as they are required within the aerodynamics. The same object is used for the transfor-
mation of the forces and moments from the aerodynamic to the body-xed system. When
connecting components as objects, only the relation between them is dened and not the
order, in which those equations are nally solved.
In Dymola, graphical syntax components are coupled by drawing a line between the dened
'coupling' points of the objects, which are called 'cuts'. These couplings represent either
energy or signal
ow. For example, the cut bsystem (body-xed system) has the following
structure:
terminal v T b [3; 3]; r[3]; vb [3]; wb [3]; ab [3]; zb [3]; Fb [3]; Mb [3]
cut bsystem (v T b ; r; vb ; wb ; ab ; zb =Fb ; Mb )
The matrix v T b denes the orientation of the body-xed system with respect to the vehicle-
carried (Earth) system, the vector r is the aircraft's inertial position in the vehicle-carried
frame; the vectors vb and ab are the velocity and acceleration in the body-xed frame
and the vectors Fb and Mb are the forces and moments, also formulated in the body-xed
frame. In the same way there are cuts dened for the vehicle-carried system (vsystem)
and for the aerodynamic system (asystem).
This cut structure represents physical connections. When objects are connected, Dymola
adds equations for the cut variables. All quantities of the cut before the slash operator
(Across variables) are set equal when connected, as it is reasonable for positions, velocities
and accelerations, quantities after the slash operator (Through variables) are summed up
to zero, as it is reasonable for forces and moments. This principle is used for connecting
engines to the aircraft body for example. The engines have the same position, airspeed,
and accelerations than the aircrafts body, their forces and moments sum up with all the
other forces and moments acting on the aircraft. Because of that formulation it is easy
to add more engines to the aircraft just by adding another engine object to Fig. 3.1 and
connecting it to the aircraft's body.
This object-oriented equation-based form of describing physical systems helps to under-
stand the physical system and enables the user to modify the model most conveniently.
3.2 Hierarchical object structure
An important aspect in object oriented modelling of physical systems is the encapsulation
of objects. The internal implementation of details, e.g. of the aerodynamics, are not
visible, when viewing the HIRM object model as depicted in Fig. 3.1. By encapsulation,
the implementation of an object can be changed without aecting the functionality of the
whole model.
Fig. 3.3 demonstrates, how the HIRM model is structured. Here only the aerodynamics
model is extracted. In the same way details of the engine, gravity, wind, and atmospheric
models can be displayed.
Version: 3
Date: April 4, 1997 28
GARTEUR/TP-088-4
atmosphere
body6DOF cut atmosphere
HIRM aero
cut
actuator
COG
equations
u atmos airspeed
cut
HIRM body kinetic
air
engine1 wind
aerodynamics
HIRM
g
engine2
Earth
HIRM gravity gust
Using the graphical interface, 'double clicking' on aerodynamics displays the parameter
window of this object. This window allows the parameters to be modied. In the same
way, all of the other objects (body, engines) can be instantiated with their parameters.
The objects of HIRM model will be detailed in the following sections. Boxes in the
following subsections contain Dymola code in the form of equations.
Version: 3
29 Date: April 4, 1997
GARTEUR/TP-088-4
ab = der(vb)
vv = der(r)
where Fb is the sum of forces due to the engine, the aerodynamics and gravity, formulated
in the body-xed coordinates, vb is the inertial velocity and wb is the rotation rate in
the same coordinate system. The acceleration ab is the time derivative of velocity vb; the
velocity vv is the time derivative of the position vector r expressed in the vehicle carried
(earth) frame.
Furthermore some aircraft specic quantities are dened:
f height g
h = -r(3)
Rotational motion
The equations of motion for the rotational movement of a rigid body in the body-xed
axis system form the moment equation for Mb:
f rotational equations of motion g
Mb = I * zb + cross(wb,(I*wb))
f Tensor of inertia
I = [ Ix , -Ixy, Ixz ;
-Ixy, Iy , -Iyz ;
Ixz, -Iyz, Iz ]
where Mb is the sum of moments about the centre of gravity due to the engine and
aerodynamics, wb is the inertial rotational velocity in body-xed coordinates, and zb is
the inertial rotational acceleration in the body-axis system. Using the standard notation
[14, 15], the equations listed above are obtained.
The translational dierential equations of motion for an aircraft body are typically ex-
pressed partly in body axes (dynamics) and partly in the vehicle carried (Earth) frame
(kinematics). The same velocity is given in both coordinate systems as body-xed vb and
vehicle-carried vv. Therefore the object TrafoBV of Fig. 3.2, which includes all transfor-
mation equations between these coordinate systems, is inherited to the object body.
3.2.2 Object: TrafoBV (Coordinate transformation: body-xed , vehicle carried)
The rotations between the body-xed and the vehicle-carried coordinate system are dened
in the object TrafoBV. The equations are from x 2.3.3.
This object contains two cuts. The cut bsystem contains all of the kinematic quantities
of the body-xed system. bTv is a [3,3]-transformation matrix between the body-xed and
the vehicle-carried coordinate systems. The cut also contains the inertial position vector r,
the translational and rotational velocity, vb and wb respectively, and the translational and
rotational acceleration vector, ab and zb respectively (all given in the bodyxed coordinate
system), which are set equal when connected. Behind the slash, the cut contains the body-
xed forces and moments, Fb and Mb respectively, which are summed up when connected.
The cut for the vehicle-carried system cut vsystem is dened in the same way.
Version: 3
31 Date: April 4, 1997
GARTEUR/TP-088-4
f Transformation equations g
vb = bTv * vv f velocity vector g
wb = bTv * wv f rate vector g
ab = bTv * av f acceleration vector g
zb = bTv * zv f rotational acceleration vector g
Fb = bTv * Fv f force vector g
Mb = bTv * Mv f moment vector g
where bTv is the transformation matrix between body-xed and vehicle carried coordinate
system. It transforms all translational and rotational quantities between these two refer-
ence frames. Since this object contains equations (and not assignments), it can be used
to transform quantities from vehicle carried to body-xed coordinates and also to trans-
form from body-xed to vehicle carried coordinates. In the latter case, the transformation
matrix is inverted automatically.
3.2.3 Object: Aerodynamics
The object Aerodynamics describes the HIRM aerodynamics. As depicted in Fig. 3.3, it
consists of an object which denes the relation between inertial speed, airspeed and wind
(object Airspeed) and the aerodynamic equations (object Aero equations).
Version: 3
Date: April 4, 1997 32
GARTEUR/TP-088-4
The calculation of airspeed in Dymola -syntax is given by the following equation. The
equation for this object is from x 2.3.4.
vab = vb - vwb
The airspeed vab, solved in body-xed coordinates, contains the x-, y-, and z-components
of the aircraft's velocity relative to the surrounding air. The airspeed is the dierence
between the inertial velocity vb and the gust vwb, which are both also formulated in
body-xed coordinates.
The aerodynamics contain the following model parameters, which are detailed in Table 2.4.
The parameters cmerr, clerr, and cnerr are uncertainty parameters with the bounds given
in Table 2.5.
f lift coecient g
cZ = czh(alpha,dts) ->
+ czsp(alpha,dts)*dcs ->
+ czq(alpha,dcs)*qb*cbar/2/Va
The aerodynamic forces are given in body-xed coordinates Fb as a function of aerody-
namic factors qS and the aerodynamic coecients cX , cY , cZ .
f Aerodynamic Forces g
f Forces in body-xed system g
Fb = qS * [ cX ;
-cY ;
cZ ]
Aerodynamic moments
The aerodynamic moments are determined by means of aerodynamic coecients for roll,
pitch, and yaw (cl, cm, cn), which are functions of the airspeed V a, the angle of attack
alpha, the sideslip angle beta, the roll, pitch, and yaw rates (pb, qb, rb), and the control
surface de
ections (dts, dtd, dcs, dcd, dr, suction). The aerodynamic coecients contain
functions of the form b (c; d). Their values are computed by linearly interpolating between
the values given in look-up tables of the variables c and d.
Version: 3
Date: April 4, 1997 34
GARTEUR/TP-088-4
f pitch coecient g
cm = cmh(alpha,dts) + cmsp(alpha,dts)*dcs ->
+ cmq(alpha,dcs)*qb*cbar/2/Va ->
+ cmsu(alpha,suction) ->
+ cmerr
f yaw coecient g
cn = cnv(alpha)*dr + cnsp(alpha,beta)*(dcs+dalpha20) ->
+ cndh(alpha,dts)*dtd + cni(alpha,beta) ->
+ ( cnr(alpha,dts) + cnrnc(alpha)*abs(dts) ) ->
* rb*b/2/Va ->
+ cnp(alpha,abs(pb)*b/2/Va) * sign(p) ->
+ ( cnpnt(alpha)*abs(dts+20*pi/180) ->
+ cnpnc(alpha)*(dcs+10*pi/180) ) * pb*b/2/Va ->
+ cnh(alpha,beta)*(dts+10*pi/180+dalpha20) ->
+ cnnc(alpha,dcs)*dcd + cnsu(alpha,suction) ->
+ cnerr
The aerodynamic moments Mb are given in body-xed coordinates as a function of the
dynamic pressure times wing area factor qS , the aerodynamic coecients cl, cm, cn, and
the generalized length cbar and span b.
f Aerodynamic moments g
Mb = qS * [ b * cl ;
cbar * cm ;
b * cn ]
The HIRM Engine contains the following model parameters, which are detailed in Ta-
ble 2.4.
Version: 3
35 Date: April 4, 1997
GARTEUR/TP-088-4
For zero thrust angles xi and epsilon, the x-component of force Fb is equal to the thrust.
The engine dynamics are detailed in Fig. 3.4 as a control block diagram:
For describing block diagrams a control block library is available, Fig. 3.5:
1 1
1
*M S
lim
S
s
b(s)
PT1 PT2 PI PID
a(s)
y=Cx+Du - +1
+1
+1
*
bound deadZone bSwitch fixDelay varDelay
1 0 variable
-1
0
1
-1
1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
The throttle nonlinearity gives the relation between the demanded throttlee and the engine
thrust dFe:
dFe = if throttle < 1 then idle + ( maxdry - idle ) * throttle ->
else maxdry + (maxreheat - maxdry) * (throttle - 1)
If the rate limits on the engine thrusts are ignored, the engine dynamics are represented
by a second order transfer function with a damping of 1 and a natural frequency of 2.
The rate limits are also dependent on the engine thrust.
maxval = if derFe0c > 47000 then 25000 else 12000
minval = if derFe0c > 47000 then -25000 else -12000
Due to the geometric location of the engines (given by xcge, ycge, zcge) the engine thrust
also contributes to the moments Mb acting on the aircraft.
The equations for the HIRM Engine are from x 2.3.6.
f thrust calculation g
f Altitude correctiong
F = Fe * rho/1.2250
Note that the engine model also contains the parameters xi and epsilon, which dene a
longitudinal and lateral thrust angle respectively. Since the HIRM engine is aligned with
the x-body axis, these parameters are zero. Since zero parameters are removed in the
resulting code they do not in
uence its eciency. However, their formal inclusion allows
a greater
exibility in using the engine model also for non-zero parameter values as well.
Version: 3
37 Date: April 4, 1997
GARTEUR/TP-088-4
with rho as the density of air, p the static pressure, T the absolute temperature, and a as
the local speed of sound.
3.2.6 Object: Gravity
Gravity is most conveniently described in an earth related coordinate system. The forces
acting on an aircraft are best understood when given in the body-xed coordinate system.
Therefore the coordinate transformation between body-xed and vehicle carried coordinate
system as given in x 3.2.2 is needed to transform the general gravity model to an aircraft
gravity model.
Here gravity is considered not to be a function of altitude. The equations of the gravity
model are given in the vehicle carried reference system.
The gravitational constant near the surface of the earth is taken as g = 9.81 m/s2 . For
constant gravity, there is only a component in the z-direction of the geodetic system.
Version: 3
Date: April 4, 1997 38
GARTEUR/TP-088-4
body frame
bodyfixed
Trafo
veh.carried
1dim gravity
Earth
Mv = [0,0,0]
Fv = [0,0,g*mass]
shown that aircraft dynamics code for Simulink is generated for common use in the Robust
Control Design Challenge. This is achieved by using a generic Dymola aircraft object
library.
mathematical (symbolic)
system model
Matlab / Simulink
automatically generated S-function (cmex)
from physical system
objects aggregation &
class libraries
This approach to automatic code generation has the further advantage that not only can
ecient parameterized simulation code be obtained for dierent simulation and analysis
environments but a parameterized symbolic code can be produced as well. This can be
used as input for symbolic analysis tools such as PUM[18] (Matlab Toolbox for Parametric
Uncertainty Modelling) or PARADISE[19] (PArametric Robustness Analysis and Design
Interactive Software Environment).
Version: 3
Date: April 4, 1997 40
GARTEUR/TP-088-4
For linear assessments, the control laws need to be robust to the following errors in the
aerodynamic moment derivatives:
Cmw 0:001
Clv 0:01
Cnv 0:002
Cmq ; Clp ; Clr ; Cnp ; Cnr 10%
CmTS ; CmCS ; ClTD ; ClCD ; ClRUDDER ; CnTD ; CnCD ; CnRUDDER 10%
The 16 states in the model ordered as in Table 2.2, and the 8 inputs are the rst 8 inputs
listed in Table 2.1, ordered as in the table. Therefore the modelling errors listed above
should be added to the elements of the A and B matrices as described in Table 4.1.
For non{linear assessments, the control laws need to be robust to the following errors in
the total moment coecients:
Cm 0:030
Cl 0:008
Cn 0:008
The engine, actuator and sensor models can be assumed to be correct and have zero
tolerances.
Version: 3
Date: April 4, 1997 42
GARTEUR/TP-088-4
Sensed values of angle of attack and sideslip may not be accurate. The control system
must be robust to the following measurement errors:
and 2
These errors should be considered as steady state osets in the measured signal and are
valid for all values of and . The other measurements listed in x 2.5.1 above can be
assumed to be accurate.
The control laws must be designed to operate at 80Hz, the iteration rate of the
ight
control computer (FCC). They can be designed by neglecting hardware implementation
issues, and should be robust to the dynamics of structural lters, D/A converters and
computational delay, which are given in x 2.5.2 above.
with Ki and i taking values in the regions shown in Fig. 4.2 (no tolerances) and Fig. 4.3
(tolerances applied).
Pilot - Command
Path -
e ?
X
u
- Actuators - Plant -
demands Filtering +
- 6
Controller Sensors
u = actuator demands
e = error signal
1. For single loop analysis, the open loop Nichols plot of the frequency response between
each actuator demand u and the corresponding error signal e, obtained by breaking
the loop at the point shown in Fig. 4.1 while leaving the other loops closed, should
avoid the regions shown in Fig. 4.4 (no tolerances applied) and Fig. 4.5 (tolerances
applied). The tolerances are listed in x 4.3.2 and should be applied to the linearised
aircraft models. Note that when performing the frequency response, a gain of -1
needs to be included on the input or output to obtain the correct phase response.
These characteristics should be valid for each loop.
2. For tests 2 and 3, the results from non-linear frequency responses should be used.
These responses are obtained by supplying a sinusoidal input of increasing frequency
to the system. The frequency should increase logarithmically from 0.5 to 20 rad/s
over a 20 s period. Stick forces of 3.24 and 9.71 pounds (14.4 N, 12 mm and 43.2
Version: 3
Date: April 4, 1997 44
GARTEUR/TP-088-4
N, 36 mm) should be used in pitch and 0.9 and 2.7 pounds (4 N, 8mm and 12 N,
24 mm) in roll. The time response of the pitch or roll attitude should be analysed
using a Fast Fourier Transform to obtain the gain and phase characteristics of the
response. The gain and phase can then be plotted in Figs. 4.6, 4.8, 4.9. To ensure
PIO resistance, the characteristics of the frequency response between stick force in
pounds (1 pound force = 4:448N) and pitch or bank attitude in degrees should meet
the following requirements.
The pitch and bank attitude absolute amplitude gains at 180 phase should be
< 16dB.
The magnitude of the average phase rate, _ average , dened as:
j_ average j = 2fc f fc ;
c
should lie either within the level 1 or level 1 boundaries shown in Fig. 4.7. The
example plot shown in Fig. 4.6 meets these criteria. The gain at -180 degrees of
phase is -21 dB thus meeting the rst requirement and the average phase rate
is:
j_ average j = 204 1( 180) = 12 =Hz
(Assuming that fc is 1 Hz and that the phase at 2 Hz is -204 in the example
shown).
Plotting these values of fc and _ average on Fig. 4.7 shows that this system is
level 1 .
3. This criterion complements the handling qualities metrics illustrated in Figs. 4.6
and 4.7 [20]. The non-linear frequency response from longitudinal stick force to
pitch attitude should lie within the level 1 boundary of Fig. 4.8. When plotting
the pitch frequency response, the response should be displaced vertically until the
curve crosses through the 0dB, 110 point. The non-linear frequency response from
lateral stick force to roll angle should lie within the region labelled as providing a
good response in Fig. 4.9. The frequency responses are non-linear and are between
the pilot demand and the aircraft pitch attitude or roll angle (i.e. the integral of the
velocity vector roll rate). For these criteria, the term open loop means that there is
no pilot closing the loop between aircraft response and pilot demand. The responses
should be obtained for diering stick forces: pitch stick forces of 3:24 and 9:71
pounds (14.4 N, 12 mm and 43.2 N, 36 mm) and roll stick forces of 0:9 and 2:7
pounds (4 N, 8 mm and 12 N, 24 mm) should be used.
4. The ratio of dropback db to steady state pitch rate qss should be:
0 < db=qss < 0:25s
for precision tracking. Dropback is dened in Fig. 4.10 [20].
Version: 3
45 Date: April 4, 1997
GARTEUR/TP-088-4
-4.5 6
@
Gain oset (dB)
@
@
@@
-2.5
Phase
0 30 oset
- ()
+2.5
+4.5
-3.0 @
@@
@ @
-1.0 Phase
0 30 oset
- ()
+1.0
+3.0
6 6 Gain (dB)
@@
@
@@
3
Phase ( )
180 145 -
-3
-6
Fig.4.4 Gain and phase exclusion zones on Nichols plot for single loop analysis frequency
response excluding tolerances.
6 Gain (dB)
4.5 @@
@@
1.5 @
Phase ( )
180 145 -
-1.5
4.5
Fig.4.5 Gain and phase exclusion zones on Nichols plot for single loop analysis frequency
response including tolerances.
Version: 3
Date: April 4, 1997 48
GARTEUR/TP-088-4
10
5
Gain: degrees attitude per pound stick force (dB)
f
-5 bw
-10
-15
-20
fc
-25
2f c
-30
-220 -200 -180 -160 -140 -120 -100
Phase (degrees)
Fig.4.6 Pitch/roll stick force to pitch attitude/roll angle frequency response criterion.
150
Level 2
Average phase rate (Degs/Hz)
100
Level 1
50
Level 1*
0
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Frequency, fc, at -180 degrees phase (Hz)
20
(−100,18)
10 (−75,10)
(−100,6)
5
L1 (−75,4)
(−85,2)
0
(−150,−3) (−110,0)
−5
L1
−10
(−140,−12)
−180 −160 −140 −120 −100 −80 −60 −40
Open loop phase (degrees)
0.25 db
25 0.5 db
20 1 db
−1 db
15
10 3 db
−3 db
6 db
Magnitude (dB)
5
Sluggish response
0 PIO activity −6 db
−5
−10 −12 db
Oscillation ratcheting
−15
Good response Quick jerkey response
−20 db
−20
−25
−350 −300 −250 −200 −150 −100 −50 0
Phase (degrees)
q (rad/s)
pitch attitude
qstat
qmax
dropback = db
1.8
1.6
1.4
Normalised sideslip angle
Upper boundary
1.2
0.8
Lower boundary
0.6
0.4
0.2
0
0 1 2 3 4 5 6 7 8 9 10
Time (seconds)
5 Evaluation criteria
The nal control law design should meet the requirements specied in Chapter 4 across
the
ight envelope. For the purposes of verifying this, this chapter sets out a set of specic
evaluation criteria against which all entries to the design challenge should be measured.
The criteria have been divided into four sub-classes which are:
Robustness
Performance
Physical considerations
Control activity
These four groups are dealt with in the following four sections x 5.1-5.4. By the nature
of the requirements given which are based on MIL-Specication documents, most of the
evaluation criteria result in a response of pass or fail rather than some numerical value.
Most of the evaluation criteria are based on two particular
ight conditions which are:
Flight condition 1: Mach 0:4, 10000 feet altitude, 8:67 degrees incidence and zero
sideslip.
Flight condition 2: Mach 0:24, 20000 feet altitude, 28:9 degrees incidence and zero
sideslip.
The rst
ight condition can be thought of as a nominal
ight condition, and the second
as an edge of the envelope condition likely to cause more stability and actuator limiting
problems. The robustness assessments in x 5.1 are based on linear analysis. The remaining
evaluations in x 5.2, x 5.3 and x 5.4 are based on non{linear responses and analysis. The
control laws should not be evaluated at 20 degrees angle of attack due to the step change
in aerodynamic coecients as detailed in x 2.3.5.
5.1 Robustness criteria
5.1.1 Nominal
ight Nichols plot
For
ight condition 1, plot the 6 Nichols plots obtained from breaking the closed-loop at
the 6 plant actuator demands. Given that nose suction is not used, and the engines are
only used in unison, the 6 actuators are symmetric and diential taileron, symmetric and
dierential canard, rudder, and symmetric engine. These correspond to the robustness
test described in Fig. 4.4. They should be plotted all on the same plot, and the exclusion
zone in Fig. 4.4 should be superimposed.
Report output: 1 gure, and a Pass/Fail result (Pass if the Nichols plots do not pass
through the exclusion region).
Version: 3
Date: April 4, 1997 52
GARTEUR/TP-088-4
For the Dryden turbulence model given in x 2.7, the RMS variation of pitch rate, roll rate,
yaw rate and normal acceleration should be recorded for
ight condition 1. Units should
be =s for the rates, and g for the acceleration. All RMS values should be relative to
the steady state-values corresponding when there is no turbulence. Moderate turbulence
levels should be used for this assessment.
Report output: four RMS values.
in practice a function of stick force. Hence the plots should be constructed using sinusoids
of 3:24 and 9:71 pounds force (14.4 N, 12 mm and 43.2 N, 36 mm) in pitch and 0:9
and 2:7 pounds force (4 N, 8 mm and 12 N, 24 mm) in roll. Two plots should be produced
corresponding to Figs. 4.8 and 4.9. All cases should be superimposed on these two plots.
Report output: two plots, 8 Pass/Fail responses (2
ight conditions 2 amplitudes for
both pitch are roll responses).
The nonlinear response of the aircraft should be tested by applying the following step
inputs.
Responses showing the pitch rate, velocity vector roll rate, sideslip angle, angle of attack
and air speed should be plotted for all of the above responses. Normal acceleration should
be plotted if this reaches its positive or negative limiting value as dened in x 4.2.
Steps 5a to 5d follow on from each other, therefore only the
ight condition at the start
of the manoeuvre is dened. The above series of manoeuvres should be repeated with the
Version: 3
Date: April 4, 1997 54
GARTEUR/TP-088-4
Cm 0:030
Cl 0:008
Cn 0:008
in combination with a measurement error of 2 degrees on incidence and +2 degrees on
sideslip.
It should be recorded whether the following nine criteria (see Chapter 4) are satised:
Max normal acceleration < 7g
Max normal acceleration overshoot = 0:5g
Min normal acceleration > 3g
Max negative normal acceleration overshoot = 0:5g
Max incidence < 30
Max incidence overshoot < 5
Min incidence > 10
Max negative incidence overshoot < 5
Settling time < 2s following the manoeuvre
Report output: nine Pass/Fail responses corresponding to the nine above requirements.
5.3.2 Structural coupling
To avoid structural coupling the following limitations (as stated in x 4.5) should be ob-
served:
The maximum high frequency (frequencies above 4 Hz) gain from pitch rate (rads/s)
to control surface de
ection (rads) should be < 3:0.
The maximum high frequency (frequencies above 4 Hz) gain from normal acceleration
(`g') to control surface de
ection (rads) should be < 0:09 rads/g.
Report output: two Pass/Fail responses.
Version: 3
55 Date: April 4, 1997
GARTEUR/TP-088-4
{ the eort related to the setting up of the design cycle (modelling, controller
architecture, weight functions),
{ the eort related to the execution of the design cycle (numerical eort, degree
of automation),
{ the eort of performing a redesign after a major aircraft design change,
the complexity of the control solution:
{ the controller architecture (required measurement signals, reference signals,
modes, actuators and lters),
{ non-linearity of the controller (adaptive, gain scheduling),
{ (linear) order of the controller,
{ ease of implementation,
the behaviour of the controller as found by the automated evaluation procedure of
Chapter 5.
This implies for instance that information on duration of each design iteration and the
motivation for each relevant design action have to be reported. Furthermore, specic
problems should be pointed out and discussed.
The structure proposed for the standard document to be prepared by each design chal-
lenge contestant is aimed at accommodating all these aspects. The next section will give
a short overview of this structure, after which each element will be discussed in more
detail. Framework documents that accommodate this structure are available in LATEX and
WordPerfect. If necessary, the correct use of these documents is indicated.
6.2 Standard presentation format layout
In short, the standard presentation format will consist of a document with the following
structure:
Title, table of contents, list of gures, list of tables, list of symbols and abbreviations,
Summary,
Chapter 1: introduction,
Chapter 2: a tutorial review of the applied control design methodology:
{ introduction,
{ typical applications,
{ plant model requirements,
{ controller structure,
Version: 3
Date: April 4, 1997 58
GARTEUR/TP-088-4
Summary
The summary should provide a short description of the applied methodology, the ob-
tained controller and some general comments on the achieved results.
Introduction (Chapter 1)
The introduction is mostly standard for all design challenge entries, it describes the frame-
work in which the design challenge was set up (i.e. GARTEUR action group FM-AG08)
as well as its overall objectives. The problem formulation should be adjusted to match the
presented design methodology; furthermore,it may be necessary to adjust the description
of the document's contents.
specications, does it need gain scheduling and can it decouple interaction in loops and,
nally, can it handle feedforward or do you require to consider regulation and a feedfor-
ward loop separately. Is the controller robust in a linear sense, or in a non-linear sense.
Can you guarantee stability, think of non-linear controllers or adaptive controllers.
tomate the procedure and to what extent expert knowledge of the designer is required for
intermediate decisions. This implies an extensive description of weight function selection
criteria, design parameters and search strategies as well as a discussion on the convergence
of the iteration procedure.
Conclusions (Chapter 8)
This chapter should comment on overall aspects of the design methodology and its features
with respect the HIRM design problem. Specic strong and weak points of the method
should be discussed. Possible future extensions or improvements of the presented method
should be indicated. Some comments should be given on the performance of the method
related to the four main objectives mentioned in x 6.1. Any comments on the set up of
the design challenge or the considered design problem can also be considered here.
References
[1] GARTEUR Action Group FM(AG08), Communication Handbook,
GARTEUR/TP-088-5, June 1995.
[2] GARTEUR Exploratory Group FM-EG12, Terms of Reference for GARTEUR
Flight Mechanics Action Group (FM-AG08): Robust Flight Control in a Com-
putational Aircraft Control Egineering Environment, National Aerospace Labo-
ratory (NLR), Amsterdam, Jan. 1995.
[3] GARTEUR Action Group FM(AG08), Project Document,
GARTEUR/M-088-1, Version 2, January 1996.
[4] McRuer D., \Interdisciplinary interactions and dynamics systems integration",
Int. J. Control, Special Issue on Aircraft Flight Control, vol.59, No.1, Jan. 1994.
[5] Dorato P. (ed.), Robust Control, IEEE Press, 1987.
[6] Dorato P. and Yedavalli R.K. (ed.), Recent Advances in Robust Control, IEEE
Press, 1990.
[7] Matlab User's Guide, The MathWorks Inc., 24 Prime Park Way, Natick, Mass.
01760, USA, July 1993.
[8] Simulink User's Guide, The MathWorks Inc., 24 Prime Park Way, Natick, Mass.
01760, USA, April 1993.
[9] MATRIXX /SystemBuild V.2.4 User's Guide, Integrated Systems Inc., Santa
Clara, CA, USA, 1991.
[10] Favre C., `Fly-by-wire for commercial aircraft: the Airbus experience', Int. J.
Control, Special Issue on Aircraft Flight Control, Vol.59, No.1, 1994.
[11] Flying qualities of piloted vehicles, MIL-STD-1797A, Department of Defence,
Washington, USA, 1987.
[12] Elmqvist H., Dymola { Dynamic Modeling Language User's Manual, Dynasim
AB, S{22370 Lund, Sweden, 1993.
[13] Elmqvist H., Object-Oriented Modeling and Automatic Formula Manipulation
in Dymola, Scandinavian Simulation Society SIMS'93, Kongsberg, Norway, June
1993.
[14] Brockhaus R., Flugregelung, Springer Verlag, Berlin, Heidelberg, 1994.
[15] Etkin B., Dynamics of Flight - Stability and Control, John Wiley & Sons, USA,
1958.
Version: 3
Date: April 4, 1997 64
GARTEUR/TP-088-4
[16] Otter M., DSblock: A neutral description of dynamic systems. Version 3.2, Tech-
nical Report TR R81-92, DLR, Institute for Robotics and Systemdynamics,
Oberpfaenhofen, Germany, May 1992.
[17] Grubel G., Joos H-D., Otter M. and Finsterwalder R., The ANDECS Design
Environment for Control Engineering, 12th IFAC World Congress, Sydney, Aus-
tralia, July 19-23, pp 447-454, 1993.
[18] Lambrechts P. and Terlouw J., A Matlab Toolbox for Parametric Uncertainty
Modeling, Philips Research Eindhoven and NLR Amsterdam, 1992.
[19] Sienel W., Bunte T. and Ackermann J., PARADISE { A Matlab-Based Robust
Control Toolbox, Proc. IEEE Symposium on Computer-Aided Control System
Design, Dearborn, MI, 1996.
[20] Handling qualities of unstable highly augmented aircraft, AGARD AR{279, May
1991.
[21] Otter M., Modeling of Multibody Systems with the Object-Oriented Modeling Lan-
guage Dymola, Nonlinear Dynamics 9, 1996 Kluwer Academic Publishers, The
Netherlands, pp 91-112, 1996.
[22] Otter M., Objectorientierte Modellierung mechatronischer Systeme am Beispiel
geregelter Roboter, VDI Fortschrittsberichte, Reihe 20: Rechnerunterstutzte Ver-
fahren, 1995.
[23] Cellier F., Continuous System Modelling, Springer-Verlag, New York, 1991.
Version: 3
65 Date: April 4, 1997
GARTEUR/TP-088-4
start ftp; check whether your current local directory is still your intended work
directory,
The custodians of the HIRM software have yet to be determined. Once decided, the ftp site which
stores HIRM will be made known.
Version: 3
Date: April 4, 1997 66
GARTEUR/TP-088-4
from the ftp> prompt: enter open <ftp site name>, enter username anonymous
and supply your e-mail address as password,
change remote directory by entering the command: cd garteur/hirm,
make sure that the le transfer mode is set to binary by entering the command: bin,
now get the le hirm0202.uue by entering the command: get hirm0202.uue,
leave ftp and check whether your current directory is still your intended work direc-
tory,
decode, uncompress and untar the le hirm0202.uue:
uudecode hirm0202.uue
uncompress -f hirm.tar.Z
tar xvfo hirm.tar,
go to the directory:
./hirm/des/cmex
compile the cmex-le of the aircraft dynamics model with the interpolation routines
hirmex.c and interpol.c by entering the command:
cmex hirmex.c interpol.c
compile the cmex-le of the actuator model hirmexa.c by entering the command:
cmex hirmexa.c
compile the cmex-le of the sensor model hirmexs.c by entering the command:
cmex hirmexs.c
compile the cmex-le of the simplied sensor model hirmexss.c by entering the
command:
cmex hirmexss.c
compile the cmex-le of the simplied sensor model hirmexsss.c for use in the
evaluation procedure by entering the command:
cmex hirmexsss.c
control.m,
aero.tab,
ini2001.mat, ini2001p.mat, ini2420.mat, ini2420p.mat,
ini3005.mat, ini3005p.mat, ini4010.mat, ini4010p.mat,
ini5015.mat, ini5015p.mat
README
hirm des.m is the Simulink Design model including the S-functions of the HIRM dynamics
(hirmex), the actuator dynamics (hirmexa), and the three sensor dynamics (hirmexs,
hirmexss, hirmexsss), which are given as cmex-les: hirmex.c, hirmexa.c, hirmexs.c,
hirmexss.c and hirmexsss.c. Their corresponding help les are hirmex.m, hirmexa.m,
hirmexs.m, hirmexss.m and hirmexsss.m.
Three sensor models are included in the model. hirmexs is the full sensor representation
as described in x 2.5.2 of the manual. hirmexss is a representation of the simplied sensor
dynaamics described in x 2.5.3 of the manual. hirmexsss gives the simplest sensor dynam-
ics which only includes the sensor dynamics blocks shown in Fig. 2.6 (Sensor dynamics +
notch lters in the case of the rates and accelerations). The complex sensor model is the
default used in hirm des. The simpler models can be substituted by editing the name of
the sensor block. The appropriate initialisation vector also needs to be specied: either
X0s, X0ss or X0sss.
The aerodynamics requires interpolation interpol.c from a table aero.tab, which con-
tains all aerodynamic data for linear interpolation in one tabley .
simulink.h, matrix.h, and mex.h are necessary to compile the cmex-les.
A dummy controller with zero gains is included in the S-function control.m.
Finally, the les ini****.mat contain all of the variables (controller, parameters, etc.)
necessary for simulation at a certain
ight condition. The contents of the ini****.mat
and ini****p.mat les are fully described in the le README.
All les, which are required for the design of the controller should be arranged in a single
directory, for instance:
....\garteur\hirm\des
start ftp; check whether your current local directory is still your intended work
directory,
from the ftp> prompt: enter open <ftp site name>, enter username anonymous
and supply you e-mail address as password,
change remote directory by entering the command: cd garteur/hirm
make sure that the le transfer mode is set to binary by entering the command: bin,
now get the le hirm0202.uue by entering the command: get hirm0202.uue,
leave ftp and check whether your current directory is still your intended work direc-
tory,
decode, uncompress and untar the le hirm0202.uue:
uudecode hirm0202.uue
uncompress -f hirm.tar.Z
tar xvfo hirm.tar,
compile the fmex-le of the actuator model hirmexa.f by entering the command:
fmex hirmexa.f
compile the fmex-le of the sensor model hirmexs.f by entering the command:
fmex hirmexs.f
compile the fmex-le of the simplied sensor model hirmexss.f by entering the
command:
fmex hirmexss.f
compile the fmex-le of the simplied sensor model hirmexsss.f for use in the
evaluation procedure by entering the command:
fmex hirmexsss.f
interpol.f,
aero.tab,
control.m,
ini2001.mat, ini2001p.mat, ini2420.mat, ini2420p.mat,
ini3005.mat, ini3005p.mat, ini4010.mat, ini4010p.mat,
ini5015.mat, ini5015p.mat,
README
hirm des.m is the Simulink Design model including the S-functions of the HIRM dynamics
(hirmex), the actuator dynamics (hirmexa), and the three sensor dynamics (hirmexs,
hirmexss, hirmexsss), which are given as fmex-les: hirmex.f, hirmexa.f, hirmexs.f,
hirmexss.f, and hirmexsss.f. Their corresponding help les are hirmex.m, hirmexa.m,
hirmexs.m, hirmexss.m and hirmexsss.m.
The aerodynamics requires interpolation interpol.f from a table aero.tab, which con-
tains all aerodynamic data for linear interpolation in one tablez .
A dummy controller with zero gains is included in the S-function control.m.
Finally, the les ini****.mat contain all of the variables (controller, parameters, etc.)
necessary for simulation at a certain
ight condition. The contents of the ini****.mat
and ini****p.mat les are fully described in the le README.
A.2.3 Use
Before starting a
simulation
symmetrical taileron (dtsc) load ini****.mat
0 of desired flight condition time
U0 in the workspace
differential taileron (dtdc) Clock Time
trim_inputs Double click here
0 to actuator for more information
symmetrical canard (dcsc) hirmsim
outputs for simulation
0
differential canard (dcdc) Mux +
+ hirmexa hirmexs hirmy
− Mux hirmex Demux
0 measured variables
Sum_inputs1 Actuator Mux Demux Sensor
rudder (drc) model HIRM model model
(16 states)
0 wind
nose suction (suctionc) wind_input
The user is free to modify the wind input vector and will need to replace the dummy
controller supplied in the le control.m. The CONTROLLER block format in Fig. A.1 is not
rigidly xed and the user is free to include additional controller blocks such as feedforwards
z For plotting purposes the aerodynamic data and their Matlab plotting routines can be found in the
directory ./hirm/aer.
Version: 3
Date: April 4, 1997 70
GARTEUR/TP-088-4
as required. The format for the inputs shown in hirm/eva/h clo pi.m is suggested as this
is what is required for the evaluation software.
The number and order of the dened reference signals, measurement signals and con-
trol inputs in the Simulink-le hirm/eva/h clo pi.m should not be changed to prevent
problems with the evaluation procedure described in Appendix C.
Version: 3
71 Date: April 4, 1997
GARTEUR/TP-088-4
B.1 Installation
We assume you have a correctly installed version of LATEX on a workstation or PC (for
instance EMTEX) and the possibility to make use of a Postscript printer (or Ghostscript).
All les, which are required for the creation of your design challenge entry document can
be arranged into a single directory, for which we suggest:
....\garteur\hirm\fra
Similar to the HIRM model and design environment software as described in Appendix A,
these les can be obtained from anonymous ftp.
After the installation of the design environment the framework document is in directory
.nhirmnfra.
GARTEUR.STY and FMAG.STY. GARTEUR.STY replaces standard style les like BOOK.STY and
ARTICLE.STY; FMAG.STY is used for some additional denitions. See HIRM-FRA.TEX for
more information on the use of these style les.
The two Encapsulated Postscript les GARTEUR.EPS and GARTHEAD.EPS must remain in
the same directory as HIRM-FRA.TEX.
Version: 3
73 Date: April 4, 1997
GARTEUR/TP-088-4
You should now have at least 38 .m-les and one .mat-le in .nhirmneva. In the evaluation
environment the denition of the controller to be evaluated should be included.
C.2 Use of the evaluation software
The evaluation script le evalhirm.m is in the directory hirm/eva. It should be executed
from the Matlab command window with the working directory set to either hirm/des/fmex
or hirm/des/cmex. Directory hirm/eva should be in the Matlab-path.
A number of questions will be put during the evaluation, requiring entry of a character
followed by pressing the Enter key. The possible answers are mentioned in the questions,
surrounded by parentheses, like (y), for optional answers, or surrounded by brackets, like
[n], for the default answer.
It is possible to execute all four parts of the evaluation successively, or to select any part
separately.
During the evaluation, a number of linear models will be derived from the applicable non-
linear congurations, and saved subsequently for later use. In all these congurations the
Version: 3
Date: April 4, 1997 74
GARTEUR/TP-088-4
control system, developed by the designer, acts as a parameter. When the control system
has been changed, any existing linear models should be removed. A question is put to
deal with such a possibility.
The evaluation proceeds parallel to Chapter 5. The evaluator should inspect each g-
ure produced visually, in order to establish the Pass or Fail condition for every criterion.
Questions will be put at appropriate instants.
Hardcopies of the generated gures may be ordered, for each gure separately, or a
PostScript plotle may be created, for later use. File prques.m in hirm/eva should be
adapted to include the local printer designation.
The Gibson pitch rate criterion in the Performance section requires a special action of
the evaluator: the plot position should be aligned with the crossing of the Level bound-
aries. The plot may be shifted vertically in an unlimited fashion by a specied incremental
number of dB's. Pressing the Enter key only terminates the shifting operation.
At the conclusion of the evaluation a report is generated, containing the Pass/Fail answers
given and the calculated numerical results. The report may contain all possible items in
case of a complete evaluation, or a part of it in case of a partial evaluation.
C.3 Incorporation of results into the framework document
The results are now available in your work directory. To incorporate them into the frame-
work document you should copy the following les to the directory in which you have your
framework document (see Appendix B):
all .EPS les,
HIRM-TBL.TEX.
Note that you will overwrite some of the les in the target directory.
Next, go to the directory with your framework document and use LATEX to compile
HIRM-FRA.TEX: the new results are automatically incorporated. The le HIRM-TBL.TXT is
created for use with other wordprocessors.
C.4 Use with your own controller
In directory hirm/eva, an example controller is supplied to be able to test the software
on the current platform. This controller is given as an S-function named control2.m.
To use your own controller, simply replace this S-function with the one you created with
the design environment. If you use the design environment as indicated, your controller
should be an S-function with the correct number and order of inputs and outputs for use
with the evaluation environment. The procedure for the evaluation is then as follows:
Version: 3
75 Date: April 4, 1997
GARTEUR/TP-088-4
save the controller S-function you created with the design environment as
control2.m,
The following appendix fulls a dual role. First it validates the linearisation process by
comparing small amplitude responses on the non-linear model with responses from the
linear model. Second, it allows designers to validate their model against the check cases
shown. The
ight condition used for the validation is at 5000 ft and Mach 0.3.
The A, B, C and D matrices for the linear model are given below. Note that the full
matrices have not been listed due to their size. The matrices shown are for states [u, v, w,
p, q, r, , , engine1F , engine1F 1 ], for inputs [DTS, DTD, DCS, DCD, DR, THROTTL1]
and for outputs [V , , ].
D = B @0 0 0 0 0 0C A
0 0 0 0 0 0
Table D.1 Amplitude of step and gure number for each response
Responses from the non-linear and linear models were overplotted and the resulting graphs
are shown in the gures below. The same line types are used in each gure and these are
described in Table D.2.
Version: 3
Date: April 4, 1997 78
GARTEUR/TP-088-4
−8
x 10
2 0.005
0 0
roll and yaw rates (r/s)
−2 −0.005
pitch rate (r/s)
−4 −0.01
−6 −0.015
−8 −0.02
−10 −0.025
−12 −0.03
0 1 2 3 4 5 0 1 2 3 4 5
102 0.21
0.205
101.5
0.2
air speed (m/s)
alpha (rad)
0.195
101
0.19
0.185
100.5
0.18
100 0.175
0 1 2 3 4 5 0 1 2 3 4 5
time (secs)
−3
x 10
0.01 1
0
0.8
0.2
−0.04
−0.05 0
0 1 2 3 4 5 0 1 2 3 4 5
−4
x 10
100.3205 5
100.32
0
air speed (m/s)
100.3195
beta (rad)
−5
100.319
−10
100.3185
−15
100.318
100.3175 −20
0 1 2 3 4 5 0 1 2 3 4 5
time (secs)
−8
x 10
2 0.02
0
roll and yaw rates (r/s)
0.015
−2
pitch rate (r/s)
−4
0.01
−6
−8
0.005
−10
−12 0
0 1 2 3 4 5 0 1 2 3 4 5
100.4 0.23
100.2
0.225
air speed (m/s)
100
alpha (rad)
0.22
99.8
0.215
99.6
0.21
99.4
99.2 0.205
0 1 2 3 4 5 0 1 2 3 4 5
time (secs)
−3 −5
x 10 x 10
12 4
10
3
roll and yaw rates (r/s)
4
1
2
0
0
−2 −1
0 1 2 3 4 5 0 1 2 3 4 5
−3
x 10
100.3186 3.5
3
100.3185
2.5
air speed (m/s)
100.3184
2
beta (rad)
100.3183 1.5
1
100.3182
0.5
100.3181
0
100.318 −0.5
0 1 2 3 4 5 0 1 2 3 4 5
time (secs)
−4
x 10
0.01 2.5
0 2
roll and yaw rates (r/s)
−0.01
1.5
pitch rate (r/s)
−0.02
1
−0.03
0.5
−0.04
−0.05 0
−0.06 −0.5
0 1 2 3 4 5 0 1 2 3 4 5
−3
x 10
100.32 12
100.318 10
100.316 8
air speed (m/s)
beta (rad)
100.314 6
100.312 4
100.31 2
100.308 0
100.306 −2
0 1 2 3 4 5 0 1 2 3 4 5
time (secs)
−3
x 10
0.035 7
0.03 6
roll and yaw rates (r/s)
0.025 5
0.015 3
0.01 2
0.005 1
0 0
−0.005 −1
0 1 2 3 4 5 0 1 2 3 4 5
100.9 0.216
100.8 0.215
0.214
air speed (m/s)
100.7
alpha (rad)
0.213
100.6
0.212
100.5
0.211
100.4 0.21
100.3 0.209
0 1 2 3 4 5 0 1 2 3 4 5
time (secs)
1.5
1
Lift coefficient
0.5
−0.5
−10 −5 0 5 10 15 20 25 30 35 40
Angle of attack (degs)
−0.05
Pitching moment coefficient
−0.1
−0.15
−0.2
−0.25
−0.3
−10 −5 0 5 10 15 20 25 30 35 40
Angle of attack (degs)
0.2
Yawing moment coefficient
0.1
−0.1
−0.2
−0.3
−10 −5 0 5 10 15 20 25 30 35 40
Angle of attack (degs)
F Evaluation Questionnaire
GARTEUR ROBUST FLIGHT CONTROL DESIGN CHALLENGE - EVALUATION QUESTIONNAIRE
-------------------------------------------------------------------------
VERSION: 3
DESIGN ENTRY:
EVALUATOR: DATE:
The first four questions follow the four assessment criteria given in the
benchmark definitions, GARTEUR/TP-088-3 (section 4.1) and GARTEUR/TP-088-4
(section 5.1) respectively.
One of the aims of Action Group on Robust Flight Control FM(AG08) was that of
"transferring the knowledge and experience gained through the Action Group
into industry". The design entries support this aim and are here being
assessed in terms of their value in meeting this aim. It is therefore the
educational content and suitability for publication of the entry that is
being evaluated:-
COMMENTS:
Version: 3
85 Date: April 4, 1997
GARTEUR/TP-088-4
1 -> Very high, needs detailed knowledge of the principles, the mathematical
background and software implementation.
2 -> High, needs detailed knowledge of principles and the mathematical
background.
3 -> A lot, needs detailed knowledge of the principles.
4 -> Some understanding of the basic principles.
5 -> Nothing, i.e. 'black box' method.
COMMENTS:
COMMENTS:
Version: 3
Date: April 4, 1997 86
GARTEUR/TP-088-4
2(c) RE-DESIGN
--------------
COMMENTS:
1 -> None.
2 -> Some.
3 -> A lot.
4 -> Most.
5 -> All.
COMMENTS:
Version: 3
87 Date: April 4, 1997
GARTEUR/TP-088-4
1 -> None.
2 -> Some.
3 -> A lot.
4 -> Most.
5 -> All.
COMMENTS:
3(a) VISIBILITY
---------------
It must be recognised that there are a wide range of specialists and managers
who need to work with the control laws at a later stage of the total design
process. These might be piloted simulation engineers, flight control computer
implementers and testers, or engineers and managers responsible for flight
clearance. The level of visibility of the functionality is important to these
people, to help them to carry out their tasks:-
COMMENTS:
Version: 3
Date: April 4, 1997 88
GARTEUR/TP-088-4
3(b) COMPLEXITY
---------------
The controller algorithms need to be executed in real time in the aircraft's
flight control computer. Since this capability is always limited, it is
important that control algorithms are efficient and do not lead to real-time
processing problems. High order controllers, multi-dimensional look-up tables
and complicated nonlinear fuctions all add to this potential problem.
COMMENTS:
3(c) IMPLEMENTATION
-------------------
Execution time was addressed by the previous question, but there are other
aspects which could cause implementation difficulties. For example, the
control algorithm's numerical integrity, timing requirements and potential
for gain scheduling are all important implementation aspects.
COMMENTS:
Version: 3
89 Date: April 4, 1997
GARTEUR/TP-088-4
If the aircraft and its flight control system are to be accepted by the
customer it must be 'qualified' against its specification. Before it flies,
it must be 'certified' against airworthiness criteria. The method and the
resulting design specification need to be compatible with existing
regulations for aircraft response and handling qualities.
COMMENTS:
4(a) ROBUSTNESS
---------------
Do you have any comments regarding the ROBUSTNESS of the design that has
been achieved ?
COMMENTS:
Version: 3
Date: April 4, 1997 90
GARTEUR/TP-088-4
4(b) PERFORMANCE
----------------
Do you have any comments regarding the PERFORMANCE of the design that has
been achieved ?
COMMENTS:
Do you have any comments ragarding the CONTROL SURFACE ACTIVITY associated
with the design ?
COMMENTS:
How much KNOWLEDGE OF THE DESIGN METHOD did you (as an evaluator) 1 2 3 4 5
have before reading the design entry report ?
1. No knowledge.
2. Some knowledge.
3. Theoretical knowledge.
4. Theoretical knowledge and practical experience.
5. As 4; and having used the design method for a similar problem.
COMMENTS:
Version: 3
91 Date: April 4, 1997
GARTEUR/TP-088-4
COMMENTS:
Taking into account the results achieved and any background knowledge that
you may have on the approach considered, please note down any significant
additional thoughts or observations that you have made, which have not been
covered by the above questions (one page maximum please):-
{ End {