© BAE SYSTEMS (Operations) Limited, 2000 1

THE DESIGN OF FLY-BY-WIRE FLIGHT CONTROL SYSTEMS

Chris FieldingFlight Control Systems Technologist

BAE SYSTEMS, Aerodynamics (W427D) Warton Aerodrome, Preston PR4 1AX

Abstract

The design of an advanced Flight Control System (FCS) is a technically challenging task for which a range of engineeringdisciplines have to align their skills and efforts in order to achieve a successful system design. This paper presents anoverview of some of the factors, which need to be considered, and is intended to serve as an introduction to this stimulatingsubject. Specific aspects covered are: flight dynamics and handling qualities, mechanical and fly-by-wire systems, controllaws and air data systems, stores carriage, actuation systems, flight control computer implementation and flexible airframedynamics. A comprehensive set of references are provided for further reading.

1. Introduction to Flight Control

When studying the mechanics of flight [1,2,3] it iscommon practice to assume that the aircraft can berepresented as a rigid body, defined by a set of bodyaxis co-ordinates as shown in Figure 1. The rigidbody dynamics have six degrees-of-freedom, givenby three translations along, and three rotationsabout, the axes. All forces and moments acting onthe vehicle can be modelled within this framework.

Figure 1 Body axis aircraft co-ordinate system

To achieve flight control we require the capability tocontrol the forces and moments acting on thevehicle; if we can control these, then we havecontrol of accelerations and hence velocities,translations and rotations. The FCS aims to achievethis via the aircraft’s flight control surfaces, shownfor the example in Figure 1: foreplane, trailing

edge flaps and rudder. The thrust provided by theengines must also be taken into account, since thisalso produces forces and moments acting on thevehicle.

2. Mechanical and Fly-by-wire FCS

The early generations of flight control systems weremechanically-based, an example of which is shownin Figure 2 (single-seat Hawk aircraft).

Figure 2 Mechanical flight control system

Direct mechanical linkages were used between thepilot’s cockpit controls (pitch/roll stick and rudderpedals) and the control surfaces that manoeuvre theaircraft, which are for this example: tailplane,ailerons and rudder. This arrangement is inherentlyof high integrity, in terms of probability of loss of

2

aircraft control, and provides us with a very visiblebaseline for explaining FCS developments.

Figure 3 Digital fly-by-wire flight control system

Subsequent generations of FCS have beendeveloped on programmes such as Tornado, JaguarFly-by-wire [4] and the Experimental AircraftProgramme [5], towards the current quadruplexdigital fly-by-wire type, schematically shown inFigure 3 and used, for example, on EurofighterTyphoon [6]. The main emphasis is now on digitalcomputing with the use of inertial motion and airstream sensor units; the direct mechanical linkagesbetween the cockpit controls and the controlsurfaces have been removed and replaced withelectrical signalling with direct motion commands,hence the term ‘fly-by-wire’. This arrangementprovides a significant reduction in mechanicalcomplexity.

In order to achieve the same level of integrity asthat achieved with the earlier mechanical systems,multiple signal sources and several lanes ofcomputing are necessary to provide redundancy,these being cross-monitored in order to isolate anyfailed equipment and to ensure safe operation. Acomprehensive built-in-test capability is alsoincluded, to ensure that the system is ‘safe to fly’prior to each flight and to identify and locatefailures. The current military aircraft trend istowards triplex redundant architectures withreliance on both cross-lane and in-lane monitoring toachieve the required level of integrity, and hence theassociated safety of system operation.

3. The Benefits of Fly-by-wire Technology

The major benefit of fly-by-wire is the ability totailor the system’s characteristics at each point inthe aircraft’s flight envelope. This is achieved by

using ‘control laws', which can be scheduled withflight condition. The introduction of digital computingfor aircraft flight control has allowed complexalgorithms to be implemented. These functions allowthe performance benefits offered by Active ControlTechnology to be fully realised and include:

• ‘Carefree Handling’ by: (i) providing angle ofattack control and angle of sideslipsuppression, which lead to automaticprotection against stall and departure; (ii) bythe automatic limiting of normal accelerationand roll rate to avoid over-stressing of theairframe.

• Handling qualities optimised across the flightenvelope, and for a wide range of aircraftstores.

• Aircraft agility, thereby providing a capabilityfor rapid changes in fuselage aiming and / orvelocity vector, to enhance both targetcapture and evasive manoeuvring.

• Aircraft performance benefits associatedwith controlling an unstable airframe, that is,improved lift / drag ratio and an increase inmaximum lift capability, both leading toincreased aircraft turning capability.

• The use of thrust vectoring to augment orreplace aerodynamic control powers, in orderto extend an aircraft’s conventional flightenvelope.

• Reduced drag due to optimised trim setting ofcontrols, including thrust vectoring.

3

• Reconfiguration to allow mission continuationor safe recovery following system failures orbattle damage.

• Advanced autopilots, providing significantreductions in pilot workload and weaponsystem performance benefits.

• Reduced maintenance costs, resulting fromthe reduction in mechanical complexity andthe introduction of built-in-test.

In order to realise these benefits it is essential toestablish an appropriate control law architecture.This is fundamental to the success of the systemand will require good knowledge of systemsequipment engineering and safety, flightdynamics and flight control. There is however, asignificant cost associated with suchperformance benefits, in terms of systemcomplexity, but usually, the performance andsafety benefits that can be achieved, easilyjustify the necessary investment.

4. Flight Envelope and Gain Scheduling

An aircraft’s flight envelope will usually bedescribed in terms of Mach number, coveringvelocity and air compressibility effects, andaltitude to cover air temperature and densityeffects. An example is shown in Figure 4 for asupersonic aircraft.

Figure 4 Supersonic aircraft’s flight envelope

The boundaries of the flight envelope areassociated with physical limits: the stall limit, athigh incidence and low dynamic pressure, wherethe aircraft’s wing lift is not sufficient to support

the aircraft’s weight; the performance limit,where the rarefaction of the atmosphereprevents a jet engine from sustaining itsoperation; the temperature limit due to the kineticheating of the airframe by the viscous friction ofthe air; and the loading limit at high dynamicpressure, to provide a safe margin againstexcessive aerodynamic loads acting on theairframe.

In order to design control laws to cover such anenvelope, it is necessary to select a grid of‘operating points’ for which the design is to becarried out. This results in a set of localisedcontrollers for the operating points. The numberof design points can always be minimised bytaking physical effects (such as dynamicpressure) into account, within the structure of theflight control laws.

As described so far, the design task is over atwo-dimensional envelope, however, a thirddimension covering an aircraft’s angle of attackneeds to be considered, in order to address theeffects of aerodynamic non-linearity and controlsurface trimming capability. In addition, theeffects of changes in mass, inertia and centre ofgravity need to be considered. The localisedcontroller designs need to be integrated togetherto cover the flight envelope. This can usually besatisfactorily achieved by using gain schedulingto produce a set of control laws. The informationneeded to schedule the flight control law gains isusually derived from the air data system, anexample of which is shown in Figure 5. Thisincludes a set of suitably located external probesfor providing pitot and static pressures and localairflow measurements, in terms of speed anddirection [7].

Figure 5 Distributed air data system

4

The locally derived probe measurements areused within the flight control computing in orderto compute the true velocity vector of theaircraft, that is, its magnitude and direction, thelatter being defined by the angles of attack andsideslip. These can then be used for gainscheduling and to provide feedback signals forstabilisation and flight envelope limiting purposes.The air data system is designed to provide highintegrity information; for example, thearrangement in Figure 5 might provide triplexangles of attack and sideslip and quadruplexairspeed information. In practice, the quality andintegrity of the air data will depend on thecapabilities and locations of the individualsensors. For the arrangement shown in Figure 5,a is a pitot probe, and b, c and d would be multi-hole probes used to resolve local flow anglesfrom pressure data. The air data information iscomplemented with information from theaircraft’s inertial sensors.

5. Aerodynamics and Control

In terms of the aerodynamic design, there are arange of specialist activities, which need to beintegrated and balanced for the satisfactorydesign and control of a combat aircraft [8]. Aspart of the overall design, the flight controlsystem design, qualification and certificationprocesses have to cover many aircraftconfigurations including the carriage of a widerange of aircraft stores [9,10].

Figure 6 Stores carriage

It is usual to design the control system for abaseline configuration, such as the aircraft fittedwith light stores. This involves using a nominalset of aerodynamic data, plus a set of parametrictolerances based on past project experience anduncertainties in the available wind tunnel data. Ifa range of significantly different stores are to befitted to the aircraft, such as heavy under-wingor under-fuselage tanks, then it may benecessary to design control laws for each ‘storegroup’ to account for their differing inertial andaerodynamic properties.

Figure 6 shows a schematic of a Tornadoaircraft carrying a heavy store load. Thepotential variation in aircraft mass, inertia andcentre-of-gravity, due to the carriage and releaseof such stores is obvious. The aircraft and itsflight control system have to be designed forcarriage of a large range of such stores,including a very large number of possiblesymmetric and asymmetric combinations. Othersignificant factors that need to be taken intoaccount in the design are: fuel state, high liftdevices, airbrakes, wing-sweep (for Tornado),performance schedules, powerplant interface (orintegration), reversionary modes, undercarriageoperation and ground handling. All of these canhave a significant effect on the design in termsof stability, handling and airframe loading. For allcombinations of stores, the FCS can offerprotection against over-stressing of the airframeand provide automatic stall and spin prevention.

Figure 7 Aerodynamic non-linearities

Flight to high angle of attack leads to non-linearaerodynamic behaviour as flow separationoccurs, wing and tail fin effectiveness are

5

reduced and control surface power varies, oftenbecoming very low. Such aerodynamic non-linearities are typified by Figure 7 where the lefthand graph indicates how, for an unstableaircraft, pitch instability might vary with aircraftangle of attack, and the right hand graphillustrates how control surface effectivenessmight reduce with increasing angle of attack. Inaddition, similar types of non-linearities areexperienced in the lateral / directional axes,significantly affecting stability and controlpowers. The flight control system has to bedesigned to accommodate such effects. If thelevel of instability is too high or if there isinsufficient control power available, then asatisfactory design will not be possible and flightenvelope limitations will need to be applied, eithermanually observed by the pilot or automaticallycontrolled by the system.

Significant aerodynamic non-linearities are alsoexperienced as a function of Mach number, asan aircraft passes through the ‘transonic region’from subsonic to supersonic flight. This is due toshock-induced flow separations and aircompressibility effects causing the aircraft’saerodynamic centre to move aftwards.

6. System Implementation

For the FCS implementation, there are furtherspecialist areas and inter-disciplinary activities,which are also essential for a satisfactory FCSdesign. Equipment specifications need to beestablished to unambiguously and completelydefine the required levels of functionality,performance and reliability, for the environmentin which the equipment is required to operate.The equipment has to be designed andmanufactured, and as part of the systemqualification process, adequately tested to showcompliance with its specification, as well as forvalidating the models assumed for the controllaws design and clearance processes.

The FCS has to be designed to guarantee thenecessary levels of reliability and integrity, byhaving a system architecture with theappropriate level of multiplexing and associatedredundancy management, as well ascomprehensive built-in-test capabilities. Thesystem design is underpinned by a

comprehensive safety analysis, covering bothnormal operation and failure modes.

The hardware necessary for the functioning ofthe FCS includes advanced sensors andactuation systems [11] such as that shown inFigure 8 (from the Experimental AircraftProgramme), and digital computing with itsinterfaces. All of these hardware componentsintroduce lags and delays into the closed-loopsystem, which tend to reduce the aircraft’sstability margins and impose physical limits onthe aircraft performance that can be achieved.Additional lags are also usually present due tothe structural dynamics filtering required toattenuate the flexible airframe response withinthe control loops. The FCS sensors measure (i)inertial data such as translational accelerations,angular rates and attitudes, (ii) air data, such asangles of attack and sideslip, and airspeed (aspreviously described).

Figure 8 Typical aircraft actuation system(Courtesy of Dowty Aerospace, Wolverhampton)

With current technology, it is usual to implementcontrol laws within a digital flight controlcomputer, an approach which offers greatflexibility and which allows highly complexfunctions to be implemented. The drawbacksare the inherent time delays, with theirassociated effect on closed-loop stability, and theclearance issues associated with safety-criticalsoftware. For digital control laws, the modelsused for the design and simulation must accountfor the digital processing effects, in order to berepresentative of the implementation, to avoidany unexpected results during ground or flight

6

testing of the system. Anti-aliasing filters will beneeded to limit the bandwidth of the input signals,in order to remove higher frequencycomponents. A formal method of control lawspecification is required in order to capture thefunctionality and implementation requirements,including the ordering and timing of the controllaw elements.

7. Handling Qualities and Pilot Interface

Flight control laws are designed to provide goodaircraft handling qualities [12], a low pilotworkload and a high degree of resistance to‘pilot-induced oscillations’ (PIO). To establish asatisfactory design, appropriate design criteriaare needed, firstly to establish a robust feedbackdesign with good disturbance rejection, andsecondly, to provide the desired handlingcharacteristics. The PIO phenomenon, wherebythe pilot’s commands are (involuntarily) in anti-

phase with the aircraft’s response, has attractedmuch attention in the past decade and hasrecently been re-named ‘Aircraft-Pilot Coupling’[13] or APC, to remove any suggestion that thepilot is to blame for the oscillation. The aircraft’shandling qualities should be verified prior to flight,by a thorough programme combining theoreticalanalysis, off-line simulation and pilot-in-the-loopground-based and/or in-flight simulation.Handling qualities and APC are the subject ofongoing research for both civil and militaryaircraft [14].

Finally, the control law algorithms and controlstrategy used must be realisable in terms of theaircraft’s cockpit interface, including theinceptors, switches and displays; these must alsobe taken into account as part of the design andharmonised with the piloting control strategyused by the control laws.

8. Flexible Airframe Aspects

The FCS motionsensors for

detecting the rigid body motion of the aircraft,also detect the higher frequency structuraloscillations due to the flexible modes of theairframe, as indicated in Figure 9, which showsthe first wing bending mode of the EAP aircraft.

Figure 9 Flexible airframe modes

The high frequency components of the sensoroutputs usually require attenuation to preventdriving the aircraft’s flying control surfaces atthese frequencies and further exciting theflexible modes [15]. This is achieved byintroducing analogue or digital filters, for examplenotch (band-stop) filters, into the feedback paths.The major constraints on filter design are theneed to meet specified stability requirements forthe flexible modes, and the need to minimise theadditional phase lag introduced by the filters at‘rigid aircraft’ control frequencies, in order tominimise the impact on achievable aircrafthandling. The effects of stores carriage, fuelstate and flight condition on the flexible modes ofthe airframe, results in changes to the modalfrequencies and response amplitudes; thestructural mode filtering needs to be designed toaccommodate such variations.

Initial structural mode filter designs are based onfinite element modelling of the airframe. This islater updated, following a comprehensive groundtest phase in which aircraft ground vibration andservo-elastic ‘structural coupling’ tests arecarried out to identify the ‘zero speed’characteristics of the airframe. This test data isthen extended to include the theoretical effectsof airspeed variation. Where verification of the

7

aerodynamic effects on the aircraft’s flexiblemodes is necessary, an ‘In-Flight StructuralMode Excitation System’ [16] is used, as shownin Figure 10.

Figure 10 In-flight structural modeexcitation system

This system allows the pilot to input deterministicsignals, such as swept frequency sine wavesgenerated from the flight control computer, inorder to stimulate the flying control surfaces andthus excite the airframe’s flexible modes.Analysis can be carried out on-line andcompared with predictions, as indicated. Similartechniques are used to identify the aircraft’s‘rigid body’ aerodynamics and for validation ofcontrol system stability margins.Such advanced facilities allow flight envelopeexpansion to be carried out in a safe, efficient,and progressive manner.

9. Future developments in the Technology

The world’s first fly-by-wire Advanced ShortTake-off and Vertical Landing aircraft, that isintended for production, is being developed aspart of the US Joint Strike Fighter Programme,with the Concept Demonstrator Aircraft beingdue to fly this year. For this class of aircraft,Active Control Technology has great potential interms of pilot handling and accurate aircraftcontrol. The UK’s ‘Vectored thrust AircraftAdvanced flight Control’ (VAAC) programme[17, 18] is investigating and demonstratingadvanced control strategies with low pilotworkload, based on flight experiments in amodified Harrier. Complementary research isbeing carried out by BAE SYSTEMS toinvestigate aircraft handling qualities for jet-borne flight [19], in terms of evaluation tasks anddesirable aircraft response characteristics.Under the UK’s ‘Integrated Flight andPowerplant Control Systems’ (IFPCS)programme [20], the integration of the flight andpowerplant controls is part of a widerdevelopment aimed at risk reduction of advancedtechnologies for application to future aircraft.

Whilst current applications have tended tointegrate a limited number of systems, forexample, flight control system and powerplantcontrol system, the implementation of a total

vehicle management system is seen to be asignificant further development. Such a systemmight integrate the functionality of traditionallyseparate airframe systems, potentially providingsystems performance improvements associatedwith efficient energy management, and areduction in equipment space and massrequirements. In addition, such systems willmake use of reconfiguration and advanceddiagnostics/prognostics to improve reliability andmaintainability, and to reduce the cost ofownership.

For future stealthy aircraft, advanced air datasystems will be required, since externalmeasurement devices need to be minimised andoptical (laser-based) devices are beingconsidered. The unusual shaping of such aircraft,for example due to faceting, and the need toreduce the number and size of control surfacesfor low observability, the possible reliance onthrust vectoring, and the development of novelcontrol methods such as nose suction / blowing,are likely to lead to highly non-linearaerodynamic characteristics. It is probable thatfor some missions, unmanned air vehicles willbecome the preferred weapons platform. Theintroduction of such technologies will presentcombat aircraft designers with interesting designchallenges.

In terms of the overall technology, it is believedthat most of the new developments will bedominated by the powerful computing facilitiesthat are now readily available to both the systemdesigners and the implementers. It is expectedthat greater emphasis will be placed on modellingthe systems that interface with the flight controlsystem, with an associated reduction in groundand flight testing. This has already started and islargely being driven by the need to reduce costs.The use of on-board aircraft and equipmentmodels and ‘articicial intelligence’ will increase,with the models progressively increasing incomplexity. Such models might be used forequipment performance monitoring, failuredetection and for providing commands or data tothe flight control system's inner control loops (anexample of which is terrain-referencednavigation). Finally, many of the modelling,design and analysis techniques that have becomemature for active flight control technology, willbe increasingly applied to improve the

8

performance of other flight systems, wherepassive control has already reached itslimitations.

References

1. Etkin B, ‘Dynamics of Atmospheric Flight’,John Wiley & Son, 1972.

2. Stevens BL and Lewis FL, ‘Aircraft Controland Simulation’, John Wiley & Son, 1992.

3. Cook MV, ‘Flight Dynamics Principles’, JohnWiley & Son , 1997.

4. Nelson JR and Smith TD, ‘Improved CombatPerformance Using Relaxed Static Stability anda Spin Prevention System’, AGARD CP-409,1984.

5. McCuish A and Caldwell BD, ‘Developmentand Flight Experience of the Control Laws andthe Aeroservoelastic solution in the ExperimentalAircraft Programme’, Advances in AircraftFlight Control, Taylor and Francis, 1996.

6. Kaul HJ, ‘EF2000 - The Flight ControlSystem’, IMechE Conference: EF2000 -Technology for the 21st Century, 1996.

7. Gracey W, ‘Measurement of Aircraft Speedand Altitude’, NASA RP-1046, 1980.

8. McMichael T et al., ‘The role of AerodynamicTechnology in the Design and Development ofModern Combat Aircraft’, RAeS Journal Vol.100 No 1000, 1996.

9. Store Carriage, Integration and Release, RoyalAeronautical Society Proceedings, 1990.

10. Aerodynamics of Store Integration andSeparation, AGARD Conference ProceedingsCP-570, 1996.

11. Raymond ET and Chenoweth CC, ‘AircraftFlight Control Actuation System Design’, SAEBook, 1993.

12. Gibson JC, ‘Development of a methodologyfor excellence in handling qualities design for flyby wire aircraft’, Delft University Press, 1999.

13. ‘Flight Vehicle Integration Panel Workshopon Pilot Induced Oscillations’, AGARD AR-335,1995.

14. ‘Aviation Safety and Pilot Control -Understanding and Preventing UnfavourablePilot-Vehicle Interactions’, National AcademyPress, 1997.

15. Caldwell BD, ‘The FCS / Structural CouplingProblem and its Solution’, AGARD CP-560,1994.

16. Ramsay RB, ‘Flight Flutter Testing ofCombat Aircraft’, AGARD CP-556, 1995.

17. Shanks GT, Gale SL, Fielding C and GriffithDV, ‘Flight Control and Handling Research withthe VAAC Harrier Aircraft’, Advances inAircraft Flight Control, Taylor and Francis, 1996.

18. Lodge PM and Runham SI, ‘The Design andFlight Demonstration of an Advanced FlightControl Law in the VAAC Harrier’, UKPowered Lift Conference, 1998.

19. Gibson JC, Lodge PM and Fielding C,‘Development of Handling Qualities Criteria forSTOVL Aircraft’, International Powered LiftConference, Arlington, USA (in preparation).

20. Fielding C, ‘Design of Integrated Flight andPowerplant Control Systems’, AGARD CP-548,1993.

Bibliography

1. McRuer D and Graham D, ‘Eighty Years ofFlight Control: Triumphs and Pitfalls of theSystems Approach’, AIAA Journal of Guidanceand Control, Vol.4, No. 4, July-August 1981.

2. Abzug MJ and Larrabee EE, ‘AirplaneStability and Control A History of theTechnologies That Made Aviation Possible’,Cambridge Aerospace Series, 1997.

3. Lloyd E and Tye W, ‘Systematic Safety(Safety Assessment of Aircraft Systems)’, CivilAviation Authority, London, 1982.

9

4. Gibson JC and Hess RA, ‘Stick and FeelSystem Design’, AGARD-AG-332, 1997.

5. Graham D and McRuer D, ‘RetrospectiveEssay on Nonlinearities in Aircraft FlightControl’, AIAA Journal of Guidance, Nov-Dec1991.

6. Droste CS and Walker JE, ‘The GeneralDynamics Case Study on the F-16 Fly-By-Wire Flight Control System’, AIAAProfessional Study Series.

7. Middleton DH, ‘Avionic Systems’,Longman Scientific and Technical, 1989.

8. Collinson RPG, ‘Introduction to Avionics’,Chapman & Hall, 1996.

9. Pratt RW, ‘Flight Control Systems’, IEEControl Engineering Series 57, TJ International,2000.

10. ‘Flight Control Design - Best Practices’,RTO Technical Report 29, April 2000 (not yetpublished).