Typically the reliability of the flight control system is analysed by a fault dependence diagram (FDD),

which
assesses the probability of a failure condition to given fault, identifying the critical failure modes. FDD
are
generated by using the FCS architecture to generate a fault tree diagram, from which probability of
failure
can be computed and critical faults identified. For more information on generation of fault tree analysis
data
see Moir and Seabridge1 .
Assessment of these failure cases are not only for regulatory and certification purposes but also for
maintenance scheduling, computing parameters such as the mean time between failure (MTBF) or
mean
cycles between failure (MCBF). These are examples of system reliability assessment parameters5 .
To
achieve certification there is a minimum requirement value for various types of failure which are
summarised
in Table 1.
Failure Condition Classification Probability of Failure Failure Description
Catastrophic P _ 10−9 Extremely Improbable
Hazardous/Severe 10−9 _ P _ 10−7 Extremely Remote
Major 10−7 _ P _ 10−5 Remote
Minor 10−5 _ P _ 10−3 Reasonably Probable
No Safety Effect P _ 10−3 Frequent
II. Flight Control System Designer Toolkit
Recognising that higher fidelity methods and introducing FCS design from a conceptual design phase
is
demonstrated by the European Union (EU) funding a Framework 6 programme (FP6), Simulating
aircraft
Stability And Control (SimSAC). The vision of the programme is to introduce multi-disciplinary design
optimisation (MDO) into the conceptual design phase. It is envisaged that the effect of this will reduce
the
life-cycle time and cost, in addition to an increase in knowledge, performance and safety of the
design, all of
which is outlined in the programmes Technical Annex4 .
Figure 1 summarises the overall goal of EU FP6 SimSAC which is a (software) framework from which
controllability and manoeuvrability requirements can be analysed and assessed from the conceptual
design
phase. This relies on utilising more advanced techniques to model the aerodynamics, structural and
weight
and balance subspaces to construct a flight mechanics model, from which stability, control and
manoeuvra-
bility can be assessed with a reasonable level of confidence. Due to the higher fidelity methods
employed,
References
1Moir I. & Seabridge A., “Aircraft Systems: Mechanical, Electrical & Avionics subsystems integration”, John Wiley &
Sons Ltd., England, Third Edition, 2008.
2Fabrycky W.J. & Blachard B.S. , “Life-Cycle Cost and Economic Analysis”, Prentice Hall, 1991.
3Chudoba B., “Stability & Control of Conventional & Unconventional Aircraft Configurations: A Generic Approach”,
Collage of Aeronautics, Cranfield University, England, April 2001.
4Rizzi A. et al, “Annex 1: Description of Work”, July 2006.
5Jackson S., “Systems Engineering for Commertial Aircraft”, Ashgate Publishing Limited, Hants, England, 1997.
6US Air Force, “Military Standard, Flying Qualities of Piloted Aircraft”, MIL-STD-1797, 1987.
7Engineering Science Data Units, “A Background to the Handling Qualities of Aircraft”, ESDU 92006, July 2006.
8International Civil Aviation Organization, “ICAO Airworthiness Technical Manual”, 1974.
9Stinton D., “Flying Qualities & Flight Testing of the Aeroplane”, Blackwell Science Ltd., England, 1996.
10Thomas H.H.B.M. et al, “A background to the handling qualities of aircraft”, ESDU International Plc., Royal Aeronautical
Socioty, Issue with Amendment A, May 2006.
11US Air Force, “Military Specification, Flying Qualities of Piloted Aircraft”, MIL-F-8785C, 1980.
12Tewari A., “Atmospheric & Space Flight Dynamics”, Birkhauser, Boston, 2007.
13Etkin B., Reid L., “Dynamics of Flight: Stability & Control”, 3nd Edition, John Wiley and Sons Inc., New Jersey, 1996.
14Stevens, B., Lewis, F., “Aircraft Control and Simulation”, 2nd Edition, John Wiley and Sons Inc., New Jersey, 2003.
15Boeing Commertial Airplane Company, “747 Aiplane Characteristics Airport Planning”
16Lanchester F.W., “Aerodonetics”, A. Constable & Co. Ltd., London, 1908.
17Kamesh, S., Pradeep S., “Refined Phugoid Approximations for Conventional Aircraft”, AIAA-98-4269, 1998.
19 of 19
American Institute of Aeronautics and Astronautics
View publication stats 
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/269062389

Methods for Conceptual Flight Control System Design

Conference Paper · January 2009
DOI: 10.2514/6.2009-1620
CITATIONS
4
READS
4,145
5 authors, including:
Some of the authors of this publication are also working on these related projects:
Methods to Improve UAV Performance Using Hybrid-electric Architectures View project
CHANGE FP7 Project View project
Christopher Simon Beaverstock
Swansea University
21 PUBLICATIONS 169 CITATIONS
SEE PROFILE
Alireza Maheri
University of Aberdeen
99 PUBLICATIONS 1,002 CITATIONS
SEE PROFILE
Thomas S Richardson
University of Bristol
47 PUBLICATIONS 373 CITATIONS
SEE PROFILE
Mark Lowenberg
University of Bristol
171 PUBLICATIONS 1,099 CITATIONS
SEE PROFILE
All content following this page was uploaded by Askin T. Isikveren on 10 April 2015.
The user has requested enhancement of the downloaded file.

Methods for Conceptual Flight Control System Design

Christopher S. Beaverstock_ , Alireza Maheri † , Thomas S. Richardson ‡ ,
Mark H. Lowenberg § and Askin T. Isikveren ¶
Department of Aerospace Engineering University of Bristol,
Queens Building, University Walk, Bristol, BS8 1TR, UK
The traditional approach in aircraft conceptual design sizing for stability and control
employs the so called “Tail Volume” method, which basically establishes static stability of
the design via empirical handbook methods. The methodology dispenses with any formal
definition of the Flight Control System architecture and topology, and, does not afford
visibility of critical sizing scenarios to the designer. This situation creates a measure of
uncertainty when attempts are made to model the flight physics problem, thus thwarting
opportunities in performing an advanced assessment of flight handling qualities. This paper
reviews the work-in-progress status of an innovative software package aimed at the con-
ceptual design phase called Flight Control System Designer Toolkit (FCSDT) that permits
Flight Control Systems architecture definition for primary and failure modes, facilitates
generation of control laws, assists the designer in apportioning control allocation sched-
ules, and finally, analyse the stability and control of aircraft models. Results regarding
flight control system architecture design are based on a control surface layout obtained
from the Boeing 747 technical manual. Stability and control assessments were based on
aerodynamic data generated by the aerodynamic model builder interface to Digital DAT-
COM provided by the European funded Framework 6 Program based on the Boeing 747
geometry.
Nomenclature
Acronyms:
CG Centre of Gravity
DoF Degree of Freedom
EASA and European Aviation Safety Agency
ESDU Engineering Science Data Units
EU European Union
FAA Federal Aviation Authority
FBW Fly-By-Wire
FCS Flight Control System
FCSD Flight Control System Design
FCSDT Flight Control System Designer Toolkit
FCSA Flight Control System Architecture
FDD Fault Dependence Diagram
FP6 Framework 6 Programme
ICAO International Civil Aviation Organization
LTIS+ Linear Time Invariant System Control Software
MCBF Mean Cycles Between Failure
_Postgraduate Student
†Research Associate
‡Lecturer of Aerospace Engineering
§Senior Lecturer of Aerospace Engineering & Head of Department
¶Senior Lecturer of Aerospace Engineering & Director of Engineering Design
1 of 19
American Institute of Aeronautics and Astronautics
MDO Multi-disciplinary Design Optimisation
MoD Ministry of Defence
MTBF Mean Time Between Failure
SCAA Stability & Control Analyser Assessor
SimSAC Simulating aircraft Stability And Control
SMJ Small-Medium range regional Jet
Symbols:
¯c Wing Mean Aerodynamic Chord
CZ_ Non-dimensional Transverse Force Derivative due to Angle of Incidence
CZ¯˙_ Non-dimensional Transverse Force Derivative due to Angle of Incidence Temporal Derivative
CZ¯q Non-dimensional Transverse Force Derivative due to Pitch Rate
CM_ Non-dimensional Pitch Moment Derivative due to Angle of Incidence
CM¯˙_ Non-dimensional Pitch Moment Derivative due to Angle of Incidence Temporal Derivative
CM¯q Non-dimensional Pitch Moment Derivative due to Pitch Rate
Ixx Rolling Moment of Inertia
Iyy Pitching Moment of Inertia
Izz Yawing Moment of Inertia
m Mass
Q Dynamic Pressure (= 1
2_V2)
S Reference Area
U0 Cruise speed
!sp Short period frequency (rads/s)
_sp Short period damping ratio
I. Introduction
The design of aircraft is an extremely inter-disciplinary activity produced by simultaneous
consideration
of complex, tightly coupled systems and functions. The design task is to achieve an optimal
integration of all
components into an efficient, robust and reliable aircraft with high performance that can be
manufactured
with low technical and financial risks at an affordable cost over the whole lifetime of the aircraft.
Aircraft design is a part of a multiphase lifecycle process, which can be summarised by the following
lifecycle model presented by Moir & Seabridge1 :
• Conceptual Phase
• Definition Phase
• Design Phase
• Build Phase
• Test Phase
• Operational Phase
• Refurbish or Disposal Phase
The model closely resembles the Downey Cycle used by the UK Ministry Of Defence (MoD), although
many other models exist in both research institutes and industry. The conceptual design phase is
responsible
for understanding the emerging needs of the customer1 , as well as developing a technical solution to
the
requirements specification. The success of a project can largely be attributed by the quality of work
conducted
at the conceptual design phase, which should consume up to 10% of the resource budget, and up to
80%
of the project total budget relies on the work conducted at this stage in the design process2 . During
this
phase, preliminary studies are performed to establish a ‘paper’ design aircraft that meets both
regulatory
constraints and the required technical specifications.
2 of 19
American Institute of Aeronautics and Astronautics
Currently, designs are developed using semi-empirical or empirical ‘handbook’ methods, which are
typ-
ically parametric relationships or a series of multi-layered data-sheets. Databases used to generate
these
‘handbook’ methods are typically based on the conventional tail-plane aft type aircraft. This limits the
conceptual designer to evolving tail-plane aft configuration aircraft, because of the restricted
applicability
of the methods available. Concurrently, no regulation or certification framework exists for atypical
design
concepts, which can account for part of the reason why no resource is invested into developing novel
concepts.
According to Chudoba3 , for the current design environment, it is becoming ever more difficult to
certify
aircraft, with an ever increasing emphasis on environmental factors presents a huge challenge for
aircraft
designers. Chudoba comments that a large level of resource investment and time is required to
evolve the
already well established tail aft aircraft, for insignificant gains in performance. This implies that it may
be necessary to investigated novel concepts for there enhanced level of performance and the
potential to
further develop the concept for future generations in a new evolutionary cycle. Before this can be a
realistic
solution, a framework suitable to develop such designs is required. This would begin with a major
assessment
of current design philosophies and methodology.
The contemporary philosophy is to begin investing significant effort on Flight Control System (FCS)
design towards the end of the conceptual design phase, or early in the preliminary design phase
when the
configuration has been tentatively frozen. This circumstance arises due to a reliance on empirical/
handbook
data, or, if available, experimental data for predicted aerodynamic characteristics. Mistakes at this
point in
the design process must be avoided, however, they are invariably made. All aircraft integrators have
been
subjected to examples of pre-flight-test aerodynamic prediction errors and unidentified problems
related to
stability and control, which lead to an unacceptable increase in programme cost and extensive
developmental
delay, or, even catastrophic failure. Examples of actual cases include:
• 50 Passenger Regional: Wheel force characteristics caused delay in certification leading to a costly
redesign of the control system
• Narrow Body: Unexpected sensitivity to wing rigging resulted in unacceptable number of aircraft not
passing acceptance flights
• Long-Range Wide Body:
– Stalls more rapidly than expected with raked tips, vortilon pattern had to be developed
– Handling and flight control characteristics do not give appropriate cues to flight crew in avoiding
limit loads, e.g. loss of Flight AA587
• Ultra Long Haul Wide Body:
– Under-predicted horizontal tail effectiveness led to larger than needed horizontal tail
– Sudden loss of lateral control during test-flight stimulated engine failure in takeoff initial climb,
resulting in the loss of aircraft and crew
Regulatory constraints are key factors in certifying an aircraft; and along with the aircraft technical
specification provide the designer with a framework to develop a concept. These constraints are
typically
relate to performance, noise, emissions, and safety factors, the final point of which is related to
systems design
and architecture, and the stability and control characteristics of the aircraft. Because of the ‘handbook’
methods used at the conceptual design phase, confidence in the development of a flight mechanics
model of
which a prototype flight control system can be designed is low. Due to the limited consideration of the
FCS
design, leads to a sub-optimal solution with respect to performance as the aircraft system is not
modeled in
its entirety.
This paper demonstrates the advantages of introducing advanced methods into the conceptual design
phase. A sophisticated flight mechanics model is developed using data generated by more advanced
tech-
niques, not typically applied at the conceptual design phase. A holistic approach to the FCS design is
presented for the Boeing 747 aircraft, which is a medium-long haul civil transport airliner. Detailed
stability
and control analysis an aircraft model using the digital DATCOM interface produced by an
aerodynamic
model builder developed in the Framework 6 programme SimSAC (Simulation aircraft Stability And
Con-
trol).
3 of 19
American Institute of Aeronautics and Astronautics
results in improved quality of data, and therefore information and knowledge of a design concept,
owing to
the more accurate modelling of the flight physics.
Figure 1. Considerations when considering controllability and manoeuvrability
The methodology involves utilising tools within the software framework to generate the flight
mechanics
model for analysis and assessment. A mixed or interlaced fidelity approach is employed, where both
low and
high fidelity methods are used to generate the model, depending on there respective regions of
applicability.
The flight control systems (FCS) architecture and design primary intent depends on flight handling
quality criteria coupled to a given cost function. To fully integrate FCS considerations onto a
conceptual
aircraft design, which would facilitate designing a variety of FCS types including manual, boosted and
fly-by-
wire (FBW) systems, a software framework is required. This would require the development of
appropriate
design, analysis and assessment tools, allowing the designer to evaluate the physical FCS
architecture, design
a suitable prototype controller, perform open and closed loop analysis of the aircraft system, analyse
critical
failure scenarios and assessment of the aircraft stability and control characteristics. Figure 2 provides
a
detailed vision to develop a comprehensive Flight Control System Designer Toolkit (FCSDT),
encapsulating
the required elements in the FCSDT framework.
III. Flight Control System Analysis and Assessment
The safety of an aircraft largely depends on the reliability of the flight control system in the event
of failure, the inherent stability and control characteristics and the aircraft structure. The main concern
regarding the aircraft structure is primarily associated with the fatigue life of the aircraft, which
depending
4 of 19
American Institute of Aeronautics and Astronautics
Figure 2. Flight Control System Designer Toolkit comprehensive overview
on the structural design philosophy, determines the method in which to certify the aircraft. Structural
contributions to safety are not the focal point of this paper and so will not be discussed any further.
A. Flight Control System Design
The flight control system includes all elements which are used to control the aircraft dynamics. Control
surfaces and engines are two examples of effectors which are used to control the aircraft’ in-flight
dynamics.
Table 1. Failure Condition Failure Descriptions
A software tool has been developed at Bristol University to perform flight control system architecture
design, along with the associated fault tree analysis. The software interface allows the user to load a
pre-
existing FCS topology from another project, edited accordingly to the current aircraft project, or to
begin
a clean sheet flight control architecture.
Figure 3(a) presents the interface from which the user can edit the FCS architecture, which includes
adding or removing control elements and editing components systems architecture to a control
element.
Once a systems architecture is developed and component links established, a failure mode analysis
can be
5 of 19
American Institute of Aeronautics and Astronautics
performed to investigate the failure of various control element combinations. This can indicate the
failure
rate of a variety flight setups for the control system, for example take-off cruise and landing
configurations,
which may employ the use of a different combinations of control effectors. The event failure rate for a
given
failure condition can be compared with the appropriate failure condition probability, as is stated in
table 1.
The example presented in figure 3(a) is a generic small-medium range regional jet (SMJ) aircraft, with
a conventional arrangement of control effectors including spoilerons, ailerons, elevators and rudders.
The
systems architecture is based on a fly-by-wire (FBW) system with mechanical backup. This generic
structure
along with the SMJ example provides the tool with sufficient enough capability to analyse most
conventional
design concepts, other concepts can be programmed in a similar manner to generate a flight control
system
architecture.
Figure 3(b) provides the interface to allocate the systems architecture for a given component. This
involves connecting the various system components required to actuate a control effector, i.e. flight
control
column, computers and sensors to power sources and actuator components. This is performed by the
use
of a Boolean expression as is highlighted in figure 3(b), using either a predefined architecture or a
manual
input to generate the expression.
(a) FCSA architecture definition interface (b) FCSA Boolean expression editor
Figure 3. Flight Control System Architecture (FCSA) software screenshots
Additionally, a presentation of the FDD is given in figure 4, where each logic gate can be investigate
for
its associated failure rate displayed in a text box above the logic gate. This presents the failure rate
for the
components leading up to that particular logic gate, along with the boolean expression used to obtain
it.
This functionality can be used by a the flight control system designer to investigate the local failure
rates,
so that alterations can be made to the system to decrease the failure rate or to decrease cost by
reducing
for example redundancy, typically leading to increased failure rate. A number of default component
failure
rate values are used, which can be altered to more appropriate values were available.
Figure 4. Fault Dependence Diagram Generated by FCSA
6 of 19
American Institute of Aeronautics and Astronautics
B. Stability & Control
Stability and control characteristics are generated by analysing and assessing both the open and
closed loop
properties of an aircraft. Open loop assessment includes analysing linearised data of the aircraft over
critical
locations in the flight envelope. For the aircraft to be certified, it must satisfy the requirements of the
code
which the aircraft has been designed. Regulatory bodies such as the Federal Aviation Authority (FAA)
and
European Aviation Safety Agency (EASA) use MIL-SPEC6 , Engineering Science Data Units7 (ESDU)
and
International Civil Aviation Organization8 (ICAO) handling quality assessment criteria as a guide to
form
the FAR and CS regulations respectively. A more comprehensive overview is provided by Chudoba.
Before an aircraft is certified, the design must be rigorously tested against airworthiness requirements
for flight operations defined by the relevant regulatory body or code. According to Stinton9 ,
airworthiness
requirements can be summarised into the following categories:
• Flight crew workload
• Flight handling characteristics
• Performance within the flight envelope
• Safety Margins
• Welfare of occupant
• Dispatch Reliability
• Economics
Stability and control assessment is mainly concerned with flight handling qualities and safety margins,
performance requirements are more closely associated with the aerodynamic and propulsion
subspaces, and
crew workloads with the flight control system and aerodynamic loads, some of which for civil aircraft
are
summarised in Table 2. Methods for assessment include figure of merit (FoM) charts, which assess
standard
modal characteristics including the short period, phugoid, dutch roll, roll subsidence and spiral modes.
Scenario CG Location Requirement
Trim at take-off FWD minimum manoeuvrability
Trim at landing FWD manoeuvrability
VS1g demonstration FWD alpha max
Minimum rotation rate FWD pitch acceleration
Push-over FWD minimum man. full stick forward - no control loss
Steady turn at take-off FWD _ > 30◦
Steady turn at landing FWD _ > 40◦
Trim at take-off AFT minimum manoeuvrability
Trim at landing AFT manoeuvrability
CEV Manoeuvre AFT balance the pitch moment of the engines
Stability AFT margin w.r.t. manoeuvre point
Dutch Roll LATERAL damping
VMC LATERAL not greater than specified speed (sideslip constrained)
Steady sideslip LATERAL demonstrate specified minimum cross wind speed
Roll man. take-off LATERAL +30◦ to −30◦ in less than 11 s, OEI
Roll man. landing LATERAL +5◦ to +25◦ in less than 7 s, VMCL
Table 2. Generic rules for commertial transports - used for SMJ Technical Specification Summary
The short period is typically characterised by a well damped, high frequency oscillations of angle of
attack
and pitch rate. An example of a figure of merit is the ESDU 9200610 chart, to asses the short period
damping
7 of 19
American Institute of Aeronautics and Astronautics
and frequency. Other standards of assessment exist for various aircraft specifications, such as the
phugoid
ICAO8 assessment criteria or MIL-F-8785C11 for the dutch roll. These provide a framework from which
aircraft design and flight handling quality assessments can take place. However, the applicability of
these
assessment codes are only suitable for current designs, or configurations which display similar
characteristics.
Therefore a new assessment criteria is required, established by developing a knowledge database
about the
stability characteristics on a spectrum of design configurations. Chudoba reiterates two definitions that
are
mandates for all aircraft, which in essence state that the appropriate authority should declare if an
aircraft
design is ‘fit to fly’.
Closed loop control typically augments dynamic derivatives in order to achieve the desired flying/
handling
qualities, to meet the fundamental stability and control requirement of ‘fit to fly’. A number of methods
exist
to compute a suitable controller such as eigen structure assignment or H-Infinity (H∞) synthesis.
Classically,
these methods compute suitable gains for a linear controller composed of proportional, differential
and
integral components using the pre-defined control laws, derived from the control philosophy as is
illustrated
in figure 5. The final component to complete the flight controller is the control allocation, which
allocates
the required control effector perturbations to meet a desired reference signal.
Figure 5. Control law philosophy
Bristol has developed an interface tool, Stability and Control Analyser Assessor (SCAA), which allows
rapid generation of trim and linear analysis data of an aircraft model across the flight envelope. These
results
can be analysed and assessed using classical analysis techniques or pre-existing figures of merit to
assess the
aircraft stability characteristics. Figure 6(a) is an example of an ESDU 92006 short period FoM, figure
6(b)
is an example of the ICAO phugoid assessment criteria and figure 6(b) the MIL-SPEC dutch roll
assessment
criteria.
10−1 100
1
2
3
4
5
6
Poor
Poor
Unacceptable
Acceptable
Satisfactory
ESDU Short Period Opinion Contours, ESDU 92006
Damping Ratio (−)
Undamped Natural Frequency (rad/s)
Excessive
overshoot
difficult to
manoeuvre
Response too
sluggish
Excessive compensation
required − difficult to trim
Too rapid an initial response −
over sensitive tendency to PIO

(a) ESDU 92006 Short Period Figure of

Merit
−0.04 −0.02 0 0.02 0.04 0.06 0.08
0
20
40
60
80
100
120
UNACCEPTABLE
ACCEPTABLE
for emergency
conditions
SATISFACTORY for
normal operation
ICAO Recommended Phugoid Characteristics
2 x zeta x omega (rad/s)
Phugoid Period (s)

(b) ICAO Phugoid Figure of Merit

0 0.5 1 1.5 2 2.5 3
−0.1
−0.05
0
0.05
0.1
0.15
Min. Frequency
Min. Damping
T/O, Appr and Land
zeta x omega = 0.10
Clb, Crz and Des
zeta x omega = 0.15
Minimum Dutch Roll MIL−F−8785C Level 1 − Cat. B and C
Undamped Natural Frequency (rad/s)
Damping Ratio (−)

(c) MIL-F-8785C Dutch Roll Figure of

Merit
Figure 6. Stability and Control Analiser Assessor (SCAA) Figure of Merit Screenshot
8 of 19
American Institute of Aeronautics and Astronautics
IV. Conceptual FCS Design Methodology
Traditionally, FCS design considerations are only concerned with flight control system architecture
and
basic stability and control assessments using techniques such as volume coefficient, with little
consideration
of developing a prototype controller. The prototype controller requires computation of the control laws
and
allocation i.e. the governing software algorythms that control the aircraft response characteristics,
which is
not typically considered from the conceptual design phase.
Figure 7 summarises the classical conceptual design approach. To provide an overview of the
process,
the initial stage involves concept inception, combining the emerging customer needs generated from
market
research, in addition to the current design environment regulation and certification requirements. The
requirements specification captures these design constraints, providing a framework from which initial
sizing
can be performed. The baseline design is progressively modified to meet the requirements specified,
whilst
optimising performance parameters.
Figure 7. Summary of classic conceptual design approach
This paper aims to integrate FCS design into the aircraft conceptual phase. This includes a holistic
approach of not only the systems architecture, but also the control law formulation and allocation.
Figure 8
describes the integration of FCS into the conceptual design cycle; the modified cycle integrates
control design,
improving the aircraft flying qualities to acceptable levels over the flight envelope. During this process
if
a feasible controller can not be synthesised, a designer decision is made to assess the design
feasibility. If
the design is considered feasible, design modifications are performed, if the design is considered
unfeasible
re-evaluation of the design concept is required.
This framework provides an environment to integrate FCS design into the optimisation of the geome-
try. Flight Control System Designer Toolkit (FCSDT) is a EU FP6 SimSAC software development
which
represents the software implementation required to demonstrate the advantages of integrated FCS
design.
The tool includes of the afore mentioned FCSA and SCAA software tools, also protocols that generate
the
controller gains and control allocation algorithms required to produce a flight controller are included.
V. Results
Results generated by the software interfaces FCSA and SCAA are presented, these are based on a
Boeing
747-100 aircraft model generated from the software framework developed by SimSAC. The model is a
rigid
body representation of the Boeing 747 flight mechanics, combined with the relevant rigid body, flat
earth
equations of motion which includes the governing 6DoF rigid body equations. Euler angles are used
to
describe the orientation, and Cartesian co-ordinates to describe the aircraft position of the aircraft.
The flat earth assumption is used due to the relatively short time span that the aircraft is to be
analysed12 .
Euler angles are used as the aircraft is not expected to be analysed or operate in conditions deemed
necessary
to use the quaternion orientation representation.
9 of 19
American Institute of Aeronautics and Astronautics
Figure 8. Integrated FCS into conceptual design
A. Flight Control System Architecture
The Boeing 747 technical report15 contains the necessary information to estimate the control effector
layout.
However, lack of any information regarding the systems architecture to each effector prevents a
comprehensive
investigation of the complete control systems architecture. As a demonstration, the control effector
layout
shall be combined with the default generic SMJ control system architecture available within the
software.
Effectively, the additional components to augment the flight dynamics are 1 pair of additional spoilers,
2
pairs of elevators and a split rudder. A demonstration of the algorithmic protocol developed to handle
the
flight control system architecture failure rate shall be presented.
The control effector layout can be seen in image presented in figure 9(a), which includes split rudder,
ailerons, elevator, all moving tail, flaps, slats etc. Only control effectors associated to roll, pitch or yaw
control shall be considered, thus slats and flaps will be omitted from the current investigation. Figure
9(b)
shows the reduced control system architecture representation in the FCSA software interface.
(a) Reduced Flight Control System Architecture of Boeing
747
(b) Software Representation of Boeing 747 Flight Control
System Architecture
Figure 9. Boeing 747 Flight Control System Architecture
10 of 19
American Institute of Aeronautics and Astronautics
Table 3 summarises the results obtained from the investigation using FCSA. The systems architecture
is
a modified version of the generic SMJ aircraft design. Component probabilities used are default
values built
into the software, which can be readily changed to values in where the component failure rate is
available.
Failure Condition Probability of Failure Control Effectors
Pitch Option 1 4.0001 × 10−12 Inner and Outer Elevators
Pitch Option 2 1.2212 × 10−17 Inner and Outer Elevators, All Moving Tail
Roll Option 1 3.0000 × 10−08 Ailerons
Roll Option 2 3.0012 × 10−16 Ailerons, 4 Outer Spoilers
Yaw Option 1 1.0006 × 10−08 Upper or Lower Rudder
Yaw Option 2 1.0004 × 10−08 Upper and Lower Rudder
Table 3. Failure Rates Generated by FCSA
According to the results generated by this architecture and comparing with table 1, the pitch and roll
control options 2 are both fail catastrophic, however using only the primary control surfaces (options
1),
reduces the redundancy leading to a significantly reduced failure rate with the roll control reduced to
fail
severe. The yaw control option 2 is only improves on option 1 by 2 × 10−12, suggesting that to improve
the failure rate the systems architecture of the upper or lower rudder is required to change. Both lower
and
upper rudders use the same computer architecture, altering this may significantly reduce the failure
rate.
B. Stability & Control Assessments
Aircraft models generated by CEASIOM produce force and moment coefficient tables, based on the
rigid
body equations of motion states. These are typically defined in the wind axes frame, which uses the
state
vector [_, MACH NO, _, P, Q, R] to generate the aerodynamic tables. Additional tables associated to
control surface deflections for example elevators, ailerons and rudders are also generated. Currently
only
trailing edge devices are modelled, as such does not encompass the entire flight control system
available to
the aircraft, which may also includes flaps, spoilers and split ailerons for example.
These components are omitted as methods used to generate the data would not sufficiently model the
relatively non-linear behavior of these components, for example, boundary layer separation is a
phenomenon
that lower fidelity inviscid aerodynamic solvers such as vortex lattice are unable to predict. As a
conse-
quence only cruise flight conditions shall be considered for this investigation. However, issues
regarding a
methodology to size flight control effectors is highlighted, that to effectively design the flight control
system,
methods developed must attempt to encompass sizing of control elements which are normally
excluded due
to the difficulties in modelling.
The model structure as seen in figure 10 represents the upper most level of the model hierarchy of
the
Simulink representation of the aircraft model. Each model is integral to a more complete investigation
of
the aircraft flight dynamics over the flight envelope for a fully augmented, closed loop aircraft model.
The
model can be reduced to an open loop model by considering the Aircraft Model, Equations of Motion
and
Actuator Model. The open loop model can be used to assess the unaugmented dynamics, and apply
the
appropriate control gains to correct the dynamics, according to a predefined control design
philosophy.
Figure 10. Upper Level Aircraft Model Structure
The model includes a rigid body aircraft model and simple thrust model. The rigid body aircraft model
input is the previously stated state vector, along with a control vector of the aircraft control surface
deflec-
11 of 19
American Institute of Aeronautics and Astronautics
tions. These are used to interpolate 2 and 3 dimensional tables, relating the states and control inputs
to the
force and moment coefficients. The model is summarised by equation 1.
C_ = C_0(_,Mach) + C__ (_,Mach, _) + C_Q(_,Mach,Q) + C_P (_,Mach, P) + C_R(_,Mach,R)
+C__elv
(_,Mach, _elv) + C__rudd
(_,Mach, _rudd) + C__ail
(_,Mach, _ail) − 7C_0(_,Mach) (1)
_ = X, Y,Z,L,M,N (2)
To demonstrate the software interface, model inputs shall include perturbations over a CG, mass and
inertia range. Results shall be compared with the cruise trim point of the Boeing 747-100 presented in
Etkin.13 Table 4 summarises the modes about the cruise point of 40,000ft altitude and a Mach number
of
0.8.
Scenario Eigen Value Period (s) thalf (s) Nhalf (cycles)
Short Period -0.3719 ± 0.8875i 7.08 1.86 0.26
Phugoid -0.003289 ± 0.06723i 93.4 211 22.5
Dutch Roll -0.033011 ± 0.94655i 6.64 21 3.16
Roll Subsidence -0.56248 - 1.23 -
Spiral -0.072973 - 95 -
Table 4. Generic rules for commertial transports - used for SMJ Technical Specification Summary
By using the model generated by CEASIOM, and varying both the vertical and horizontal CG position,
the behavior of the various poles can be observed in figure 11(a) and 11(b). Using the undamped
natural
frequency and short period damping, the ESDU 92006 plot can be generated yielding the results
observed
in figure 12(a). Comparing the results with equation 3 and 4, the relationship between frequency and
CG
position is augmented through the pitch damping term CM¯q and pitch stiffness term CM_(assuming
CZ¯q is
negligible). The effect of CG is dependent on the relative effect between these two terms. As CG
moves
forward towards the nose of the aircraft, CM_ decreases and CM¯q increases. If the overall effect is to
increase
the numerator of equation 3 then the frequency increases.
−0.4 −0.35 −0.3 −0.25 −0.2 −0.15 −0.1 −0.05 0 0.05
−2.5
−2
−1.5
−1
−0.5
0
0.5
1
1.5
2
2.5
Real
Imag
Decreasing
Z position
Increasing
Static Margin
(a) Variation of longitudinal pole position due to CG per-
turbations
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Real
Imag
(b) Variation of lateral pole position due to CG perturba-
tions
Figure 11. Variations in aircraft modal pole positions due to CG perturbations
The case where CM¯q is the dominant term, decreasing the static margin or de-stabalising the system
by
moving the CG aft will increase this term and increase the undamped natural frequency. The damping
is
also observed to increase due to the proportional relationship in equation 3 and 4.
12 of 19
American Institute of Aeronautics and Astronautics
10−1 100
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
5.5
6
Poor
Poor
Unacceptable
Acceptable
Satisfactory
Damping Ratio (−)
Undamped Natural Frequency (rad/s)
Excessive
overshoot
difficult to
manoeuvre
Response too
sluggish
Excessive compensation
required − difficult to trim
Too rapid an initial response −
over sensitive tendency to PIO
(a) Variation in ESDU short period assessment due to vari-
ations in CG position
−0.04 −0.02 0 0.02 0.04 0.06 0.08
0
20
40
60
80
100
120
UNACCEPTABLE
ACCEPTABLE
for emergency
conditions
SATISFACTORY for
normal operation
2 x zeta x omega (rad/s)
Phugoid Period (s)
(b) Variation in ICAO phugoid assessment due to varia-
tions in CG position
Figure 12. Aircraft assessment criteria variations with CG perturbations
!2
sp =
CZ_CM¯q − CM_(( 2U2
0

QS¯c )m + CZ¯q)
( Iyy
QS¯c )((
2U2
0

QS¯c )m + CZ¯˙_)
(3)
2_sp!sp =
QS¯c
Iyy
¯c
2U0
CM¯q(( 2U2
0

QS¯c )m − CZ_) + 2U0Iyy

QS¯c2
2U0
¯c CZ_ + (( 2U2
0

QS¯c )m + CZ¯q)CM¯˙_
((
2U2
0

QS¯c )m + CZ¯˙_)
(4)
However, figure 12(b) indicates that the phugoid damping decreases with decreasing static margin, al-
though the period (and frequency) remains relatively insensitive to variations in CG position. There is
some
amount of dependence as is indicated by equation 5, being primarily dependent on the M_, Mq and Mu
derivatives which appear in both the numerator and denominator. Therefore the effect on the
frequency is
dependent on the relative weighting of each term.
The phugoid damping becomes remarkably more difficult to analyse due mainly to the lack of a simple
and reliable approximation to the phugoid characteristics. Lanchester’s equation16 is one of the
earliest ap-
proximations to the phugoid characteristics, which provides reasonable accuracy for predicting the
frequency,
but is very poor with respect to computing the damping. Pradeep & Kamesh17 present a more
accurate
method for predicting the phugoid damping, which is derived from the equations of motion, making
few
assumptions to arrive at a 4th order longitudinal characteristic equation, substituting an appropriate
short
period approximation, the phugoid can be approximated.
!ph =
s
g(MuZ_ −M_Zu)
U1M_ − Z_Mq
(5)
Comparing the modal results generated by the CEASIOM generated aerodynamic data, the short
period
positions suggests a CG aft configuration but a large difference in the damping is observed. The CG
aft configuration is also supported when comparing the dutch roll pole positions with those presented
in
table 4. The results obtained for the roll subsidence, spiral and the phugoid are sensitive to both
vertical
and horizontal CG position and can be fixed to fit in the positions provided by Etkin. Reviewing the
approximate equations along with the aerodynamic data obtained from both Etkin and the model
generated
by CEASIOM, large errors were observed with respect to control coefficients, leading to errors in the
trim
point obtained or no solution existing. The elevator control coefficient predicted by digital DATCOM is
approximately 5 times greater in the aerodynamic data presented in Etkin.
To obtain a solution at this trim point, artificially increasing the CM_elv derivative was required. Errors
were also observed in the zero angle of attack lift coefficient and lift-curve slope, the former
compounding the
13 of 19
American Institute of Aeronautics and Astronautics
error in elevator angle, to predict an incorrect trim point. According to the equations presented
equations 3
and 4, this would not affect the prediction of the short period, as the results obtained for the
derivatives
involved in prediction of the short period are within the linear region and are unlikely to change with
trim
angle of attack, but will however be affected by the error in the lift coefficient, and relative error in CG
position, CM_ and CM¯q as a result of poor aerodynamic force prediction.
Although the error in trim point may not affect derivatives involved in the short period, all
approximations
indicate that the phugoid is highly dependent on the calculated trim point. This is because derivatives
such
as CX_ are typically non-linear, due to the parabolic relationship between drag and angle of attack
through
the lift coefficient (CD = CD0 + kC2
L). Therefore deviations away from the trim point will certainly lead to
errors in phugoid prediction, as the linear derivative is only valid for a small perturbation in the trim
point
parameters.
Figure 13- 17 investigates the effects of mass and inertia on the aircraft modes of motion. Mass is a
scaler
quantity, unlike the inertial properties which are vector quantities, and as such affects all modal
properties.
Variations in mass are shown in figure 13 to 14, which displays the effect in varying mass ±20% on
the
aircraft modes of motion. The short period damping is observed to decrease, or where as the
frequency
appears to be relatively insensitive to variations in mass.
−0.5 −0.4 −0.3 −0.2 −0.1 0 0.1 0.2
−2.5
−2
−1.5
−1
−0.5
0
0.5
1
1.5
2
2.5
Real
Imag
Roll Subsidence
Short Period
Dutch Roll
(a) Mass effects on short period, dutch roll and roll subsi-
dence modes
−0.014 −0.012 −0.01 −0.008 −0.006 −0.004 −0.002 0
−0.08
−0.06
−0.04
−0.02
0
0.02
0.04
0.06
0.08
Real
Imag
Phugoid
Spiral
(b) Mass effects on phugoid and spiral modes
Figure 13. Effect of mass on the aircraft modes of motion
Figure 13(b) and figure 14(b) indicates that a growth in mass increases the phugoid damping whilst
the
frequency increases. The effect on the dutch roll characteristics however decreases the dutch roll
damping,
whilst increasing the frequency, which can be derived from figure 13(a) and figure 14(c). Unfortunately
dutch roll approximations rely on underlying assumptions for there accuracy, which in the
approximation
presented by Etkin assumes that sideslip and yaw rate are the dominant modes where roll rate is
assumed
negligible. As can be observed by investigation of the associated Eigenvector, this is certainly not the
case
across the flight envelope. Closed form approximations may seem attractive for design purposes, but
can
become misleading if the assumptions are not treated carefully, implying that there use in design
across the
envelope may be quite limited, thus supporting the notion of increased fidelity methods to optimise the
FCS
design according to flight dynamic requirements.
The inertial properties affect the mode which is most closely associated to the plane in which the
mode
acts as will be demonstrated. The results in figure 15 indicate that by increasing the pitching moment
of inertia, both the frequency and damping decrease. The result of the decreasing frequency comply
with
the analytical result obtained from equation 3. Equation 4 indicates that the variation of the short
period
damping is dependent on the proportion of the 2U0Iyy
QS¯c2
2U0
¯c CZ_ term.The results would indicate that the
contribution of this term on the numerator with variations in Iyy is minimal, and that the damping is
dominated by the inverse relationship with the Iyy term, with little contribution from the short period
frequency.
The phugoid frequency is insensitive to variations in Iyy, but decreases with aft migration of the CG,
which is also suggested by the phugoid frequency approximation by Pradeep & Kamesh17 or equation
5.
Unfortunately as highlighted previously the analytical solution leads to a complex interrelationship
between
14 of 19
American Institute of Aeronautics and Astronautics
10−1 100
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
5.5
6
Poor
Poor
Unacceptable
Acceptable
Satisfactory
Damping Ratio (−)
Undamped Natural Frequency (rad/s)
Excessive
overshoot
difficult to
manoeuvre
Response too
sluggish
Excessive compensation
required − difficult to trim
Too rapid an initial response −
over sensitive tendency to PIO
(a) Effect of mass on the ESDU short period assessment
criteria
−0.04 −0.02 0 0.02 0.04 0.06 0.08
0
20
40
60
80
100
120
UNACCEPTABLE
ACCEPTABLE
for emergency
conditions
SATISFACTORY for
normal operation
2 x zeta x omega (rad/s)
Phugoid Period (s)
(b) Effect of mass on the ICAO phugoid assessment criteria
0 0.5 1 1.5 2 2.5 3
−0.1
−0.05
0
0.05
0.1
0.15
0.2
Min. Frequency
Min. Damping
T/O, Appr and Land
zeta x omega = 0.10
Clb, Crz and Des
zeta x omega = 0.15
Undamped Natural Frequency (rad/s)
Damping Ratio (−)
(c) Effect of mass on the MIL-F-8785C Dutch Roll assess-
ment criteria
Figure 14. Effect of mass on the aircraft modes assessment criteria
15 of 19
American Institute of Aeronautics and Astronautics
−0.5 −0.4 −0.3 −0.2 −0.1 0 0.1 0.2
−2.5
−2
−1.5
−1
−0.5
0
0.5
1
1.5
2
2.5
Real
Imag
Short Period
(a) Effect of I
yy on the aircraft pole positions
10−1 100
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
5.5
6
Poor
Poor
Unacceptable
Acceptable
Satisfactory
Damping Ratio (−)
Undamped Natural Frequency (rad/s)
Excessive
overshoot
difficult to
manoeuvre
Response too
sluggish
Excessive compensation
required − difficult to trim
Too rapid an initial response −
over sensitive tendency to PIO
(b) Effect of I
yy on the ESDU assessment criteria
−2.21 −2.2 −2.19 −2.18 −2.17 −2.16 −2.15 −2.14 −2.13
x 10−3
−0.08
−0.06
−0.04
−0.02
0
0.02
0.04
0.06
0.08
Real
Imag
Phugoid
(c) Effect of I
yy on the phugoid pole positions
Figure 15. Effect of I
yy on the aircraft modes of motion
16 of 19
American Institute of Aeronautics and Astronautics
the damping and Iyy, but the reader can refer to Pardeep & Kamesh for further information.
Variations in the lateral mode frequency and damping properties are typically associated to changes
in
Ixx and Izz, which is demonstrated by figures 16 and 17. The results regarding the dutch roll indicate
that
for this particular mode, the modal characteristics are more sensitive to variations in Ixx relative to Izz.
This
suggests that the mode may be predominantly a rolling motion, which is further supported by
investigating
the eigenvector magnitude, where the ratio is 5:1 when comparing roll and yaw rate normalised
eigenvectors
respectively. This is further supported by the results presented in Etkin, although the error of the roll
rate
seems to be far greater than the other normalised states relative to one another. This may suggest
that the
method prediction has large deficiencies in the modelling of roll damping and stiffness terms. This
would
also invalidate the assumption of negligible rolling motion in the dutch roll approximation.
−0.01 −0.005 0 0.005 0.01 0.015 0.02 0.025 0.03
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Real
Imag
Increasing Ixx
Increasing Izz
(a) Effect of I
xx and I
zz on the aircraft Dutch Roll pole
positions
0 0.5 1 1.5 2 2.5 3
−0.1
−0.05
0
0.05
0.1
0.15
0.2
Min. Frequency
Min. Damping
T/O, Appr and Land
zeta x omega = 0.10
Clb, Crz and Des
zeta x omega = 0.15
Undamped Natural Frequency (rad/s)
Damping Ratio (−)
(b) Effect of I
xx and I
zz on the aircraft MIL-F-8785C
Dutch Roll assessment
Figure 16. Effect of I
xx and I
zz on the aircraft Dutch Roll mode
The assessment criteria presented in figure 16(a) is based on assessment criteria for landing
configuration,
and so is inappropriate to draw a conclusion for this particular flight condition. However the results
demon-
strate the effect of varying mass properties which can be altered in the materials used to produce the
aircraft,
or simply due to shifting fuel loads in the wing and tail. Figure 17 shows the effect of varying Ixx and Izz
on the remaining lateral modes of motion. The results suggest that the roll subsidence mode
characteristics
are more sensitive to variations in Ixx, or the rolling moment of inertia. Also, the figure 17(a) implies
that
increasing both rolling and yawing moment of inertia increases the roll subsidence time constant.
The effect of rolling and yawing moment of inertia on spiral mode characteristics can be seen in fig-
ure 17(b), which indicates the mode is more sensitive to variations in yawing moment of inertia. Again,
the
effect of yawing moment of inertia is observed to increase the modes time constant, however
increasing the
rolling moment of inertia decreases the modes time constant.
17 of 19
American Institute of Aeronautics and Astronautics
−0.56 −0.54 −0.52 −0.5 −0.48 −0.46 −0.44 −0.42
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Real
Imag
Increasing Ixx
Increasing Izz
(a) Effect of I
xx and I
zz on the roll subsidence mode
−0.012 −0.0118 −0.0116 −0.0114 −0.0112 −0.011 −0.0108
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Real
Imag
Increasing Ixx
Increasing Izz
(b) Effect of I
xx and I
zz on the spiral mode
Figure 17. Effect of I
xx and I
zz on aperiodic lateral aircraft modes
VI. Conclusion
Presented was a methodology to address the inclusion of flight control system design, along with a
software
framework capable of analising the flight control system design. Models and systems design were
based on a
Boeing 747, due to the availability of stability and control data to compare to the aircraft model
generated
by the CEASIOM software. Investigation of the flight control architecture was based on the control
effector
layout found in the Boeing 747 technical manual, although the results are derived from a arbitrary
control
system architecture.
The FCSA software interface demonstrated that the flight control system architecture problem could
be
solved by adopting a novel algorithmic protocol approach to represent the flight control system. This
can be
used to combine the control effector topology, with the systems architecture of each effector to
compute the
failure rates, and generate the fault dependence diagram of all control effector combinations. This
provides
the required framework to develop and analyse all control system configuration setups across the
flight
envelope, such as take-off, landing and cruise configurations.
Results for stability and control analysis to demonstrate the SCAA software interface were based on
the
trim point of a Boeing 747-100 from Etkin. The software results using the model presented in Etkin
produces
the same results, suggesting that any variation in the stability and control results are due to the
methods
used to produce the model. Results for the model generated by CEASIOM appeared produce ‘ball
park’
results with respect to eigenvalues when compared to the results presented by Etkin, although some
amount
of latitude was required to obtain the results.
It is clear that the CG position can be used to fix the eigenvalue position to that of the results
displayed
in Etkin, although these may be at different CG positions for other modes. However, the variation is
far more
significant in the eigenvectors, which defines the mode shape, leading to errors in simulation time
histories
with respect to relative amplitude and phase of the states. Digital DATCOM was the method used to
produce
the aerodynamic data, and is proven to yield reasonable results, although closer investigation of the
data
shows gross underestimated lift coefficients and control derivatives. This may lead to the
discrepancies shown
in the aircraft modal characteristics. Digital DATCOM is based on HTP aircraft and therefore this is not
a
feasible option to integrate into a conceptual aircraft design for unconventional aircraft morphologies.
The
integration of a vortex lattice code to generate the aircraft model has also been considered due to the
quick
time to a solution, and more analytical approach, increasing the applicability of the model to other
design
concepts. This has been part integrated into CEASIOM, allowing for generation of the inviscid
aerodynamic
data, although Mach corrections have not been applied.
The SCAA interface provides the means to analyse trim points within the limits of applicability of the
model. Linear results are generated at each trim point, which can be used to analyse and assess the
design
according to a set of assessment criteria. This data can then be used to generate controllers across
the flight
envelope, to augment the aircraft dynamics and improve the flight handling qualities. Furthermore,
each
trim point can then be used to simulate a set of inputs to generate time histories, for investigation of
the
18 of 19
American Institute of Aeronautics and Astronautics
performance of the open and closed loop responses.
The software framework presented is a work in progress representation of the methodology to
integrate
development of a prototype FCS into the conceptual design phase. Further work is required to fully
integrate
control design into the software framework, although work on a methodology and tools are required
for sizing
of the control effectors.