A computational method has been developed toinvestigate the aerodynamic interaction between ahelicopter rotor and the fuselage in steady-statelevel forward flight. In order to model a compoundhelicopter the fuselage is fitted with wings. Themodel uses deformable vortex rings for the rotorwake, a source panel representation of the fuselageand a lifting surface method for the wings which canhandle arbitrary geometries. The wake representationis based on a free wake analysis, relaxing the nodalpoints on the vortex rings and wing wake elementsaccording to the induced velocities in the flowfield after every time step. In this way, neitherwake is restricted to planar shape. At each timestep the fuselage and wing forces were calculated.The trim attitude of the fuselage was determined bythe pressure field surrounding it using a potentialflow model for the external flow and a fullydeveloped turbulent boundary layer model on thefuselage surface. Parametric studies were carriedout to determine the influence of wing position,wing span, wing angle of incidence and advance ratioon the results. The computer program incorporates agraphical postprocessing output routine allowing thevisualisation of the wake shape, the inducedvelocity and the angles of incidence on the rotordisc to be made, together with the velocities atvarious wing stations and the lift coefficientsalong the span.

This paper deals with the problem of determining anoptimal controller which minimises thenorm of a given aircraftwhile guaranteeing that an upper limit of its norm is not exceeded. Theproblem is tackled by means of a linear matrixinequality formulation, which allows one toconstrain the eigenvalues of the model to within afixed region of the complex plane. Key features ofthe method are the possibility of simultaneously

• 1. keeping the maximum value of the inputdemand under control

• 2. taking flying quality requirements intoaccount

• 3. assuring a minimum level of systemrobustness against uncertainties of the model.

The approach is simple to handle and is well suited tothe design of aircraft stability augmentationsystems. A discussion of two case studiesdemonstrates the effectiveness of the procedure.

This paper describes a comparative evaluation of twodata smoothing algorithms for use in a two stepestimation-beforemodelling procedure for aircraftparameter identification. A simple fixed lagsmoother is compared with the usual, and morecomplex, modified-Bryson-Frazier smoother in thefirst, state estimation, step of the aircraftparameter identification procedure. The comparisonis illustrated by application to the analysis of theDutch Roll motion of the Embraer EMB-312 Tucano.Both algorithms were found to give results ofcomparable accuracy although the fixed lag smootheris computationally more efficient. It was concludedthat the fixed lag smoother algorithm is anacceptable alternative to themodified-Bryson-Frazier algorithm in aircraftparameter identification applications.

This paper considers the convection of the dynamicstall vortex. A brief discussion of existing data isgiven. A comparison of two sets of data fromdifferent experimental facilities is then presented,and it is indicated that an important anomaly existsconcerning the behaviour of the dynamic stallvortex. It emerges that freestream Mach number isthe only parameter that can account for the observeddifferences. The importance of this parameter isthen discussed in the context of vorticity flux fromthe aerofoil surface.

An experimental study of the effects on the low-speedaerodynamic characteristics of a strake-like filletis described, modelled on one used on an Airbus A320variant fitted at the leading edge of a sweptwing-plate junction. The wing, swept back at 20°,was of NACA 0015 section and chord 500 mm, bothnormal to its leading edge. A turbulent boundarylayer had developed on the plate well ahead of thejunction. The tests were conducted at a unitReynolds number of 1.56 x 106m-1.

Surface pressure distributions were measured on theplate in the neighbourhood of the leading edgejunction and also on the aerofoil and fillet at wingincidences of 0°, 3°, 6°, 9° and 12°. These weresupplemented by surface oil-flow studies.

The mean velocity and turbulence intensity fieldsaround the leading edge were examined for incidencesof 0° and 9°, using both a single tube yaw meterdeveloped for the purpose and an X-wire anemometer.The X-wire anemometer was also used downstream ofthe trailing edge of the swept wing; five of theReynolds stresses and the mean velocity field weremeasured.

The sectional lift coefficients on the aerofoil werefound to diminish as the junction was approached,slightly more so with the fillet than without it.The sectional drag coefficients due to pressureincreased as the junction was approached, the filletmoderating this increase to only a small extent.

However, the addition of the drooped fillet modifiedthe flow considerably. The horseshoe-like vortex wasless well defined than without it. At zeroincidence, the peak in the turbulence intensitylevels was virtually eliminated on what becameeffectively the compression side of the wing due tothe local camber introduced by the asymmetricfillet. The turbulence levels were also reduced bythe addition of the fillet at an incidence of 9°.However, the turbulent activity was spread through alarger proportion of the viscous region. Thesecondary flows and the turbulence activity in thewake are associated with unrecoverable kineticenergy and will be manifest as drag on the surfacesforming the junction.

It is concluded that a carefully designed fillet,optimised for the cruise incidence of an aircraft,can reduce the peak turbulence levels in thejunction. It remains unclear whether the total dragassociated with the junction flow can be reducedsignificantly.

However because of its effects on the turbulence, theremay be other benefits, for example on the efficiencyof downstream elements, such as fuselage-mountedengine intakes or the following stages of an axialflow machine. Junction fillets might also be used tocontrol the scouring of river beds around bridgepiers.

The classic problem of stability and transition toturbulence of an incompressible flow over a flatplate subjected to uniform suction through thesurface is considered. This is an important “model”problem for laminar flow control and the resultshave direct application to the design of low dragaerofoils. Detailed mean boundary layer calculationsare presented and the flow stability is determinedby obtaining numerical solutions of theOrr-Somerfield equation. Transition conditions areestimated by using the semi-empirical en method.Accurate values have been obtained for the levels ofsuction necessary to delay instability andtransition indefinitely. Consideration has also beengiven to the effect of surface suction upon theplate drag.

The class of aircraft known as gyroplanes (orautogyros) helped to pave the way for thedevelopment of the helicopter. However, they havefound no application in contemporary commercial ormilitary aviation. It is in recreational or sportflying that the gyroplane has proved popular. Mostif not all designs are however homebuilts, and as aconsequence little analysis of any significance hasbeen conducted on the flight mechanics of theseaircraft. This Paper presents an analysis of thelongitudinal stability of gyroplanes. Simpleconsideration of the basic forces and moments thatmight influence stability, in the context ofelementary rotor behaviour, is used to assess thefundamental nature of gyroplane longitudinalstability. This is quantified by the application ofa sophisticated generic rotorcraft mathematicalmodel to the gyroplane problem. It is concluded thatthe basic configuration can be longitudinally stablewith appropriate design, and that the rotorspeeddegree of freedom must also be included whenmodelling the aircraft.

The effects of flap deflection, configuration incidenceand leading edge bluntness on boundary layerseparation and transition have been studiedexperimentally. A quasi two-dimensional flat plateequipped with a full span trailing edge control flaphas been employed for these tests.

The studies have been carried out in a hypersonic guntunnel at M∞ = 8·2 and Re∞/cm= 9·0 × 104. The flap deflection anglesstudied were in the range 0° ≤ β ≤ 30°. Theincidence range was from zero to α = 10° (positive αis nose down). Leading edge bluntness effects weresimulated by the introduction of a hemi-cylindricalleading edge.

The flow structure was studied using high speedSchlieren photography as well as surface pressureand heat transfer measurements. Liquid crystals wereemployed to study the threedimensionality of theflow structure for selected configurations.Analytical theories have been developed to estimatethe flap pressure and heat transfer levels for thesharp and blunt configurations. These are comparedagainst experimental measurements.

Inverse simulation is becoming a more widely usedtechnique in flight mechanics studies. The abilityto predict control and response time histories for apredefined manoeuvre or task has obvious benefits incertain applications (handling qualities studiesrelated to the ADS-33 requirements for helicopters,for example). The main criticisms of the techniquehave always been concerned with the numericalproblems usually encountered when solving theequations of motion in the inverse manner. In thispaper these problems are highlighted and a simpletechnique which overcomes them is presented. Thistechnique provides a robust mechanism wherebyreliable and consistent inverse simulations arepossible.

Flow transition in the developing region of a planewall jet is studied experimentally by means ofhot-wire measurements. The Reynolds number, based onthe jet exit velocity and nozzle exit height, variesfrom 3 x 102 to 3 x 104 in theinvestigations. The results indicate that the mostimportant factor causing wall jet flow to changefrom the initial exit state to the final turbulentstate is the rapid increase of turbulent intensityfrom the formation of vortical structures and theirinteractions with the wall. Data also shows that thetransition process of the wall jet depends on theReynolds number in the operating range. When theReynolds number is greater than 2000, the meanvelocity distribution of the wall jet changesdirectly from a top-hat profile at the nozzle exitto a turbulent profile in the turbulent developedregion. When the Reynolds number is smaller than2000, Glauert's laminar velocity profile is foundbetween the nozzle exit and the turbulent region. Itis also found that the transition will be delayedand the transition region prolonged with thedecrease of the Reynolds number.

The paper presents a method for the prediction of meanflow data (i.e. skin friction, velocity profile andshape parameter) for adiabatic two-dimensionalcompressible turbulent boundary layers at zeropressure gradient. The transformed law of the wall,law of the wake, the van Driest model for thecomplete inner region and a correlation between theReynolds number based on the boundary layer integrallength scale (ReΔ*.) and the Reynoldsnumber based on the boundary layer momentumthickness (Reθw) were used to predict themean flow quantities. The results for skin frictioncoefficient show good agreement with a number ofexisting theories including the van Driest II andthe Huang et al. Comparison with alarge number of experimental data suggests that atleast for transonic and supersonic flows, thevelocity profile as described by van Driest andColes is Reynolds number dependent and should not bepresumed universal. Extra information or perhaps abetter physical approach to the formulation of themean structure of compressible turbulent boundarylayers even in zero pressure gradient and adiabaticcondition, is required in order to achieve complete(physical and mathematical) convergence when it isapplied in any prediction methods.

This paper develops the theory of turning flight in amoving air mass to explain some of the problemsencountered in making downwind turns, particularlyin light aircraft. Although the basic mechanics offlight and aerodynamics of the aircraft are wellunderstood, the methods used by the pilot to controlthe aircraft in turning flight are less wellestablished and can be critical in maintaining asafe flight path. In this paper the effect of thecontrol strategy likely to be employed by a pilotflying close to the ground is used to show how theenergy loss during a turn can lead to an unsafeflight path.

Experiments are performed on the effects of a turbulentboundary layer on surface pressure and heattransfer, using a cylindrical model fitted with avariety of flares at Mach 5. The flow overcompression corners of up to 20° (below the expectedincipient separation value) is examined. Liquidcrystal thermography indicates an unexpected peak inheat flux close to the corner. It is suggested thatthis is due to a small scale separation occurringearlier, i.e. at smaller corner angles, thanpreviously observed. An estimate of the size of sucha separation bubble is made and it is suggested thatthis is a predominantly laminar phenomenon.

It is widely accepted that the effectiveness of cannonin the air-to-air combat scenarío depends on therate of fire. Present firing rates for 20-30 mmaircraft cannon range from 600 to 1500 rounds perminute. New developments are aimed towards evenhigher rates of fire. With these higher rates offire more energy is likely to be transferred to theaircraft fuselage, resulting in possible structuralfatigue damage. To absorb the recoil forces severaltypes of recoil system have been developed and arepresently in use. Conventional recoil systems arenot, however, ideal for very high rates of fire andhence alternatives must be investigated.

With this work the use of an anti-resonant device toreduce the transmission of forces to the aircraftfuselage for a high rate of fire is investigated.Antiresonant isolators operate on the principle thata condition of little or no motion may be enforcedat specific points of interest on a system atparticular frequencies, through suitable tuning ofthe system. This is essentially achieved bydesigning the isolator system in such a way thatinertial forces are used to react spring forcesunder stationary conditions.

A novel automatic grid generation method for complexCFD applications has been developed, driven by theneed of the industrial CFD practitioner to reducethe turnaround time from an initial CAD geometry toobtaining a flow simulation. The method uses asubdivision strategy, allowing hanging nodes in aCartesian unstructured mesh environment and iscombined with a non-overlapping, semi-structuredsurface grid. Unusually, the surface grid is the endrather than the starting point of the presentapproach. This reversal of gridding strategy allowsusability to be enhanced, in that the gridgeneration is reduced to one automatic step. Thecapabilities of the method are demonstrated byreference to selected inviscid and viscous flowsolutions using the Fluent code Rampant v4.0 as aflow solver.

Progress in aeroelasticity of fixed and rotary wingaircraft is compared by means of a selective review.Discussion of the role of concurrent engineeringleads into modelling and validation, these beingdealt with under the headings of aerodynamics,structural and integrated models. Formulation andsolution of equations of motion are considered,which relate mainly to rotorcraft, and thenaeroelastic performance improvement is covered. Thisincludes optimisation, aeroservoelasticity andpassive devices. Basic insight providing models arereviewed and the paper concludes with, furthercomments on insight, and also future progress inaeroelasticity.

This paper examines the changing role aerodynamictechnology plays in the design and development ofmodern combat aircraft. It reviews several aspectsof aerodynamics which contribute to aircraftperformance and handling characteristics. Itconsiders the importance of a range of technologies,the contribution each makes to the final integratedsolution and comments on the compromises necessarythrough the design cycle to optimise the overallweapon system, sometimes to the cost of one of thecomponent technologies. The technologies discussedare in no way exhaustive but attempt to encapsulatethe breadth of the subject, to illustrate itsdiversity and to point the way for the futuredevelopment if aerodynamic technology is to continueto make an important contribution to the design ofcombat aircraft.

Each of the component technologies are discussed interms of the contribution it makes, the tools andtechniques used to predict, analyse and interpretthe technology contribution, and the requirementsfor the direction of the development of thetechnology for the future.

Before considering the technologies themselves it isimportant to understand the environment in whichthey will be used, which increasingly conditions themanner in which they are employed and suggests thedirection for their development.

This paper begins, therefore, by briefly reviewing thechanges to the environment in which the fighterpilot operates and the manner by which hisrequirements, and therefore the specification of theaircraft, are defined before going on to look at theoverall contribution each technology makes in thedesign process.

A rapid quasi-analytical method — low orderperturbation panel method — is presented forcalculating aerodynamic sensitivity derivatives forsubsonic wings. The method is based on the low orderpanel method with the internal Dirichlet problemformulation and analytical differentiations cascadedand inverted. In terms of doublet strengthsensitivity to configuration geometry, the computingcost of the present method for the partialderivative matrix calculation is less than one orderof magnitude than the existing high orderperturbation panel method, but the accuracy iscomparable. Furthermore, by applying a physicalinterpolation instead of a weighted geometricalinterpolation, the sensitivity derivatives ofpressure, lift and pitching moment coefficient withrespect to configuration geometry are derived andcalculated in the present paper. Such sensitivityderivatives are lacking in high order perturbationpanel method. A few calculative examples aregiven.

The start condition of a second-throat ejector-diffuser(STED) system has been studied by solving theaxisymmetric Navier-Stokes equations with a highorder conservative supra-characteristics method(CSCM). According to the conventional concept, whena STED is started, the flow passing through thesecond-throat contraction should be supersonic sothat the normal shock swallowing condition applyingfor a supersonic windtunnel holds. The presentnumerical results, however, suggest that a STED canbe started with a normal shock forming ahead of thesecond-throat contraction because the high speedflow in the region close to the wall of the diffuserserves to release the mass blocked by the normalshock, leading to a decrease in chamber pressure aswell as the start of the system. This not onlycontradicts the conventional concept of the start ofa STED, but also provides an explanation for theexperimental discrepancy between the startconditions of a STED and a supersonicwindtunnel.