special issue paper 175 how industry concepts of ......special issue paper 175 how industry concepts...

SPECIAL ISSUE PAPER 175

How industry concepts of concurrent engineeringenhance aircraft design educationT Melin1†, A T Isikveren1‡, A Rizzi2§∗, C Stamblewski2§, and H V Anders2§1University of Bristol, Bristol, UK2Royal Institute of Technology (KTH), Stockholm, Sweden

The manuscript was received on 16 June 2006 and was accepted after revision for publication on 1 December 2006.

DOI: 10.1243/09544100JAERO127

Abstract: Two student projects are described including the intended goals, the approaches taken,the tools used, and what was learned from the exercises. An international collaborative teachingprotocol between Ecole Polytechnique de Montreal and the Royal Institute of Technology (KTH)was exercised in aircraft design education. Poignantly, a novel instructive design process usingthe analogue of contemporary concurrent engineering practices in industry was implemented.The idea was to strategically assign multi-disciplinary design tasks to each Partner University inaccordance with their respective competencies. The university–industry coupling was initiatedby request for proposals and corresponding marketing requirements and objectives producedby Bombardier Aerospace in Montreal, Canada. Two MATLABTM-based tools were prominentin facilitating the capstone aircraft design projects. They included: Quick Conceptual AircraftResearch and Design, a computer-aided conceptual design engineering system; and TORNADO,a Vortex-Lattice code for computing aerodynamic characteristics. The result of the two exerciseswas found to benefit the participating industry, the educational establishments involved, and thestudents carrying out the projects.

Keywords: aircraft, conceptual, design, education, aerodynamics, stability, control, flight,handling, simulation

1 INTRODUCTION

Today, the standard in many of the capstone aircraftdesign projects at universities and technical insti-tutions is to focus on fostering teamwork in amulti-disciplinary design setting. Many educationalinstitutions count on the involvement of industryin not only defining the nature of each project, butalso in committing to some degree of participation,

∗Corresponding author: Department of Aeronautical Engineering,KTH Royal Institute of Technology, Stockholm 100 44, Sweden.

email: [email protected]†Now at: Department of Aerospace Engineering, KTH Department

of Aeronautical and Vehicle Engineering. email: [email protected]‡Now at: Engineering Design, Aerospace Vehicle Architecture and

Design Integration (AVADI), Department of Aerospace Engineer-

ing, Aircraft Design, Ecole Polytechnique de Montreal, Canada.§Now at: Department of Aeronautical and Vehicle Engineering.

i.e. authorship of course notes, lectures, tutorials,attendance at meetings, marking, etc. One evolution-ary direction in design education that has recentlyemerged is analogous to the modern era of air-craft product development where integrated productdevelopment teams (IPDT) of an aircraft integratorcollaborate with suppliers or risk-sharing partners.This education programme can be best classified asa multi-party approach by which industry–universityand university–university relationships are fostered,subsequently strengthened, even helping to gen-erate previously non-existent third-party industry–university ties.

Present trends in aircraft design towards augmen-ted-stability and expanded flight envelopes call fora more accurate description of the flight-dynamicbehaviour of the aircraft in order to properly designthe flight control system (FCS). Hence the need toincrease the knowledge about stability and control(S&C) as early as possible in the aircraft develop-ment process in order to be ‘First-time-right’ with the

JAERO127 © IMechE 2007 Proc. IMechE Vol. 221 Part G: J. Aerospace Engineering

176 T Melin, A T Isikveren, A Rizzi, C Stamblewski, and H V Anders

FCS design architecture. Up to 80 per cent of the life-cycle cost of an aircraft is a direct result of decisionsmade in the conceptual design phase, and so mis-takes must be avoided. The European air-transportsector has set the agenda in their second AdvisoryCouncil for Aeronautic Research in Europe ‘Vision2020’ Report [1] SRA-2, where a number of goalsfor the environmental impact of air transportationhave been itemized. An emerging view is that suchambitious targets will not be met without some ele-ment of innovation in the education of the futureaerospace engineers who will be entering the work-force. A framework that supports state-of-the-artcomputer-aided concept designs suitable for procur-ing economically amenable and ecologically friendlydesigns is seen to be the priority. In the opinion of theauthors, one aspect that would serve to deliver suchimportant goals is the (architectural) discipline and(technical) subspace of S&C and thus subsequentlyemphasized in this treatise. To this end, it is essentialto integrate appropriate software tools when it con-cerns the instruction of aircraft design and systemsintegration.

A relatively recent tool, Quick Conceptual AircraftResearch and Design QCARD [2], introduced in 2002and currently undergoing an extensive enhancementphase, is the one the authors believe that combinesthe right mix of functionality when it concerns infor-mation technology, because it reflects a measure ofcontemporary industry practice. In addition, one for-tuitous aspect is the fact that it is geared to caterfor the student-level of ability. Traditionally, commer-cial aircraft conceptual design systems like ProjectInteractive Analysis and Optimization [3], AdvancedAircraft Analysis [4], and Raymee Design Software[5] make extensive use of the so-called handbookmethods on the basis of parametric correlations andsemi-empirical analytical constructs. More sophis-ticated and powerful versions of such conceptualdesign software platforms can be found in industry.These proprietary, in-house-developed systems arethe result of experience accumulated from a series ofprevious design projects, which usually infuse a seriesof existing handbook methods.

A growing consensus of opinion, however, findsthat these handbook methods and such comprehen-sive databases are not reliable enough for treatingnovel and unconventional designs, and thus, there isa trend towards replacing the methods with compu-tational procedures that calculate the required infor-mation from first principles. As an example, QCARDincludes, as a subsystem, the Vortex-Lattice Method(VLM) code TORNADO, a numerical aerodynamicssoftware that employs the Vortex method Lattice [6],in order to compute the aerodynamic characteristicsof a given design morphology. TORNADO, written

in MATLABTM, allows the user to define most con-ceivable types of aircraft morphologies, with multiplewings, both cranked and twisted with multiple con-trol surfaces. It currently generates the static andconstant-rate stability derivatives; in an unsteadyaerodynamics version to be released soon, even thedynamic derivatives [7] will be available for compu-tation as well. The code has found many uses in theaeronautical design and educational community, andits continuous development has spawned new edu-cational tools as well as new research possibilities[8–10].

It would be a misleading message if the instructionin design gives students the impression that systemsarchitecture and integration are done with just con-cepts and software, reinforcing the notion that testingis a thing of the past. The aerospace community ingeneral, therefore, has the obligation to embrace therich synergy that comes through close coupling ofsimulation and numerical-physical testing. From theonset of controlled, manned flight up until the 1950s,testing was the predominant approach: the so-called‘cut-and-fly’ approach to design. Today, it co-existswith simulation – both in terms of assessing conceptsand confirming predicted performance. The authorsdo not dispute the idea that physical testing willalways be the ultimate milestone in verifying a designconcept – the only open question that remains is theextent at which economic and man-power outlay is tobe rationalized.

This treatise describes two different types of studentaircraft design projects that have been completedin aerospace systems and design integration coursesoffered at Ecole Polytechnique de Montreal (EPdM) inCanada, and the Royal Institute of Technology (KTH)in Sweden. The authors will provide examples thatreflect issues discussed above in relation to the teach-ing of aircraft design, and the paper will conclude byoffering some reflections on what was learned in theprocess.

2 THE QCARD COMPUTER-AIDED CONCEPTUALDESIGN TOOL

In a broad sense, the entire aircraft design process canbe categorized into three distinct phases:

(a) conceptual definition;(b) preliminary definition;(c) detailed definition.

Depending on the requirements of time and resourcesdeemed appropriate by the airframe manufacturer,the conceptual definition phase itself cannot bebranded as adhering to one type of mindset. In fact,as exemplified by Fig. 1, there exist two tiers under

Proc. IMechE Vol. 221 Part G: J. Aerospace Engineering JAERO127 © IMechE 2007

Concepts of concurrent engineering 177

Fig. 1 Illustration of the conceptual design process segmented into two tiers: the initial or‘predesign’ and refined baseline configuration definitions

this phase; one aimed at establishing a very quick(timescale can be from one to several weeks) yettechnically consistent feasibility study, some callpredesign and the other would be a protractedand labour intensive effort involving more advancedfirst-order or even higher-fidelity trade studies toproduce a refinement in defining the minimumgoals of a candidate project. During the preliminarydefinition, product design is still undergoing a some-what fluid process and indeed warrants some ele-ment of generalist-type thinking. It can be thought ofessentially as a constrained exercise because the min-imum goals of the project have already been estab-lished during the conceptual definition phase andthe aim is to meet these targets using methods that

do not necessarily reflect the conventional wisdomestablished during the conceptual definition phase.Furthermore, the participants in this working groupare mostly genuine specialists in each respective dis-cipline. As the status of a project is well within thepreliminary design phase, assuming the manufac-turer has confidence in the potential for a new productline and has established a development cost it is will-ing to absorb, the detailed definition phase wouldbegin after the project is formally launched.

QCARD [2] is an interactive MATLABTM-based con-ceptual design package, which allows the designof gas-turbine-powered transport aircraft (Fig. 2).The system functionality is specifically tailor-madeto predict, visualize, and assist in optimizing

Fig. 2 Screen shots of QCARD computer-aided engineering system for aircraft conceptual design



conceptual aircraft designs with emphasis placedon user interactivity. Critical development objectivesincluded acceleration of design response time witha significant increase in design freedom, conformity,and consistency of the results. Hidden and explicitDecision Support System functionality provides theuser with a wide-ranging freedom in defining as manyor as few independent variable quantities. A fea-ture allowing preset or in-house-defined parametricassociations, historical data, trend information, andanalytical assessments are also available.

Through total user control, the designer can create,calculate, and analyse 15 configurationally and para-metrically distinct designs concurrently. The pack-age affords a systematic and transparent process tonot only conduct analyses with respect to geometryconstruction, weights and balance, propulsion, aero-dynamics, S&C (using the Mitchell [11] code), andintegrated operational performance, but also facili-tates coupling of the en route performance subspaceto that of economic criteria as defined by directoperating cost and profit/return on investment. Anoption is also available to conduct hypo-dimensionalconstrained multi-objective optimization using evo-lution methods, the Nelder-Mead Simplex search or acocktail of both.

Near-term plans include the coupling of threeadditional subspaces: structural topology, systemsarchitecture definition, and community noise andemission prediction modules. Further developmentwill involve the implementation of formalized trade-studies, uncertainty analysis concepts, and riskassessment toolkit for the purpose of assistingdecision-making and promoting design robustnessduring the initial technical assessment (ITA) phase. Inaddition, to complement the aforementioned func-tionality, the economics module will be extendedto permit quantification of life cycle cost and netpresent value. A final enhancement will be the devel-opment of an aircraft facsimile generation inter-face. This routine via manual-synchronization orauto-synchronization will generate calibration mod-els of known aircraft using methods commensu-rate with the detail of dataset information availableto the user.

2.1 Predicting high-speed and low-speedaerodynamics

Today, early indications are emerging that theaerospace sector is undergoing some changes withrespect to how vehicles are designed, built, and oper-ated. It appears that the analytical tools currentlyavailable for conceptual design engineers to conductfeasibility studies that ‘push the envelope’ in termsof minimizing development costs and creating shifts

in operational paradigms are not suitable due to thepredominant philosophy of simply utilizing and cod-ing existing, sometimes outdated, handbook meth-ods. Many new methodologies that approach theconceptual design problem from a different perspec-tive are to be reviewed in this body of work. Togetherwith the main focus of generating theories more com-patible in applicability and scope for requirementsstipulated by contemporary design offices, they arealso devised expressly for the purpose of being utilizedto investigate the more seriously contemplated con-cepts currently gathering momentum, such as highlysynergized progenitor or family concepts, and hightransonic and/or supersonic commercial flight.

The importance of predicting low-speed and high-speed aerodynamic qualities of aircraft cannot beunderstated. The implication to vehicular definitionrelates to an initial appreciation of how the flight enve-lope will look as well as being one of the integralcomponents in formulating the aeroplane’s opera-tional performance attributes. The main aim is todevelop methodologies, in which the designer hasan ability to examine the design space in a moresophisticated manner, not only in terms of depart-ing from the usual more simplified approach premisebut on account of the impact a technological decisionmakes to the end result. These two primary goals mustalso be tempered by an appreciation for reductionin the analysis complexity. This is surmised as beingachievable by first of all soliciting the designer’s philo-sophical requirements and translating this notion intosingle all-encompassing algorithms that provide vis-ibility to the designer. Secondly, the methodologiesmust be robust with respect to stoppage when keyinformation required on the part of the designer isfound to be lacking.

TheTORNADO [6] code is aVLM, programmed to beused in conceptual aircraft design and in aerodynam-ics education. Work on the code began in 1999 at theDepartment of Aeronautics, KTH, Stockholm, Swe-den. The first version was released in 2001 togetherwith the users manual and code description. Thiswork began as part of the Masters thesis of Melin [12],the code developer. A sample of the scope of resultsgenerated by the software is presented in Fig. 3.

The aircraft geometry in TORNADO is fully three-dimensional with a flexible, freestream followingwake. TORNADO allows a user to define most typesof contemporary aircraft designs with multiple wings,both cranked and twisted with multiple control sur-faces located at the trailing edge. Each wing is per-mitted to have unique definitions of both camber andchord. The TORNADO solver, which computes forcesand moments, provides the basis of quantifying asso-ciated aerodynamic coefficients. The aerodynamicderivatives can be calculated with respect to: angle ofattack, angle of sideslip, roll-pitch-yaw rotations, and



Fig. 3 Screen shots of the TORNADO numerical aerodynamics prediction software

discrete control surface deflection. If necessary, all ofthese conditions may be applied simultaneously. Anyuser may edit the program and design add-on tools asthe program is coded in MATLABTM, and the sourcecode is provided under the General Public Licence(http://www.gnu.org/copyleft/gpl.html).

The core method stems from Moran [13], but hasbeen modified in order to accommodate a three-dimensional solution and discrete trailing-edge con-trol surfaces. The most notable change is the exten-sion of the theory of the horseshoe vortex into thevortex-sling concept. The vortex sling is essentiallya seven-segment vortex line, which for each panel,starts at infinity behind the aircraft, reaches the trail-ing edge, moves upstream to the hinge line of thetrailing-edge control surface, then forward to thequarter chord line of the panel in question, goingacross the panel, and then back downstream in ananalogous way. The issue of the wake passing throughthe geometry at certain flight conditions is resolved bya collocation point proximity detection routine thatautomatically removes the influence from a vortexthread passing too close to a collocation point.

The code is generally distributed in the hope thatit will be useful, but without any warranty; without

even the implied warranty of merchantability or fit-ness for a particular purpose. However, validationcomparisons have been conducted in which thecode output is compared with actual data. The testcase for the original steady aerodynamics versionwas the Cessna 172 – validation was conducted usingCessna Aircraft Company released flight test data[14]. The steady-TORNADO version was also bench-marked against computational results produced byAthena Vortex-Lattice (AVL) [15] and Panel MethodAmes Research numerical aerodynamics softwares(Pmarc) [16], aVLM, and a panel method, respectively.A work-in-progress unsteady aerodynamics versionof TORNADO has recently been validated using staticand dynamic stability derivatives data for the Saab105 (Fig. 4), which is an advanced air force trainerwith reconnaissance and ground attack capabilities.

2.2 Stability and control analysis

Although all the other core disciplines are addressedwith some element of detail during the conceptualdesign phase, one area of investigation that has tra-ditionally lacked any form of sophisticated depthis S&C. Historical trend sizing of the empennage and

Fig. 4 Representation of Saab 105/SK-60 aircraft in TORNADO geometric domain



the control surfaces through a tail-volume approach[16–18] is usually deemed sufficient. This circum-stance is quite puzzling considering the unavoidablefact as Cook [19] states while referring to the WrightBrothers: ‘. . .stability and control comprised the sin-gle most critical requirement for flight. . .’. Withoutquestion, most aircraft projects have experiencedproblems associated with vehicular flying qualitiesand pilot-in-the-loop response attributes, particu-larly those aircraft configured as powered (boosted),or otherwise, manual primary FCS. The difficultiesin achieving satisfactory qualities become even morepronounced when the powered FCS needs to becomplemented by some form of manual reversionas well.

When discussing the S&C attributes of an aircraft,two fundamental handling characteristics are citedas controllability and manoeuvrability. Controllabil-ity pertains to an attribute that enables the pilotto initiate and subsequently maintain a manoeuvre.The primary concern of controllability is establish-ing a vehicular reaction one would normally expectto a given stick command, and then extending thatattribute to an assessment of the ease or difficulty inmaintaining such an initiated manoeuvre. Manoeu-vrability, alludes to the aptitude of a pilot–aircraftsystem to effect changes in the flight path, angularrates, and speed of the aircraft. Other concerns suchas time lag, overshoots, and necessary compensationby the pilot during entry into a manoeuvre and main-tenance of a steady-state acceleration, and return tonormal flight come into consideration as well.

3 COLLABORATIVE STUDENT DESIGNPROJECT – HORIZON 1100

An international collaborative teaching approach toaircraft design education, begun in late-2003, com-prised an industrial partner, Bombardier Aerospacelocated in Montreal, Canada and two educational

institutions: EPdM based in Montreal, Canada andKTH in Stockholm, Sweden. The idea here was toassign multi-disciplinary design tasks to each partnerstrategically chosen in accordance with their respec-tive competencies. The strengths of EPdM includedworld-class facilities for conducting virtual productintegration and a curriculum that imbues a solidunderstanding of systems integration. KTH offersapplication of specialized knowledge and utilizationof internationally recognized in-house-developednumerical tools giving an ability to address aspectsrelating to aircraft morphology, aerodynamic design,and refined sizing for S&C. The approach taken wasinspired by the modem era of aircraft product devel-opment where IPDTs of an aircraft integrator collab-orate with suppliers or risk-sharing partners (Fig. 5).

3.1 Teaching model based on concurrentengineering concept

Marketing requirements and objectives (MR&O) isa set of design specifications in industry, and inthe undergraduates’ design course, it is issued asa request for proposal (RFP) to the collaboratinguniversities. The joint-technical-assessment phase inindustry involves pooling together a select groupof specialists and conceptual designers for a moredetailed assessment. In the design courses, the Prime(EPdM) University’s ITA is handed-off to the Collab-orating (KTH) University for a subsequent, deeperstudy, hence, assuming the role as ‘specialist’. Thejoint-conceptual-definition phase JCDP in industryinvolves potential suppliers and risk-sharing part-ners, along with the aircraft manufacturer’s systemintegrators who further develop the basic systemarchitecture and functionality. Once the milestone ofclosing out the collaborative aircraft design educationis completed, lessons learned are cycled back to eachof the universities involved.

Fig. 5 An international collaborative teaching approach to aircraft design education using theIPDT-Integrator-Supplier analogy



The educational process begin with, for example,Bombardier Aerospace’s Advanced Design Depart-ment producing an RFP document providing a briefoverview of the target market – an exemplar cited forthis paper is a next generation regional transport,and itemization of the MR&O, found in Table 1. Inshort, it called for a 50–70 passenger (PAX) familyof regional transports focused on turboprop renewaland organic growth within the turbofan-engine fleetdomain. A set of high-level specifications for thebaseline aircraft are:

(a) 70 PAX at 31 in. (0.79 m) seat pitch;(b) limited-scale operations such as thin routes,

actual PAX loads fluctuate;(c) combined exceptional field performance with

high productivity;(d) must fly in and out of London city or Toronto city

centre (steep approach);(e) Columbus–Denver city pair;(f) must operate autonomously;(g) emphasis placed on marketability (PAX prefer-

ence).

The family member was stipulated to be a 50-PAXregional, and additional members could be mission-ized aircraft for freighter, corporate, and govern-ment/military usage.

Thereafter, EPdM with one inter-disciplinary teamcomprising 20 undergraduates proceeded with a 14-week conceptual design effort. Paralleling this, threeKTH undergraduates over the course of 14 weeksinvestigated in more detail (although still at a con-ceptual design level) specialized aspects of the initialdesign generated by the EPdM team. A staggering ofposted results and final documentation from eachteam occurred due to each institution’s differenttimetables; KTH generally has additional time, andhence, conducted further refinements. As it can bediscerned in Fig. 5, this process mimics the fashionin which technical task sequencing usually occurs inproduct development programme schedules – in thesense after the preconceptual design phase is com-pleted, an increasingly more detailed (conceptual)ITA phase would be undertaken.

3.2 Market niche identification for nextgeneration regional aircraft

The regional aircraft market comprising 70 PAX equip-ment types is quite attractive because regional air-craft are increasing in capacity with passing time.On the heels of the spectacular expansion witnessedfor 50 PAX regional aircraft, airlines are now look-ing ahead to acquiring 70 PAX or larger equipmenttypes in an effort to maintain a reasonable level of

Table 1 Bombardier aerospace’s request for proposal and itemization of the marketing requirementsand objectives for a next generation regional transport

New regional transport

Accommodation 70 PAXSeat pitch At least 31 in. (0.79 m)Weight per PAX 225 lb (102 kg) includes baggageBaggage volume At least 7.0 cu.ft (0.20 m3) per PAXOverhead bin volume At least 2.0 cu.ft (0.06 m3) per PAXDesign range, IFR, 100 nm alt. 1500 nm (2780 km) at M 0.70Out-and-return maximum range At least 800 nm (1485 km) at M 0.70Number of 200 nm sectors without refuelling At least 4Long range cruise At least M 0.65Normal cruise At least M 0.70Maximum cruise At least M 0.75Takeoff field length, ISA, s.l.


revenue growth and protection of margin. Moreover,one entire generation of old regional turboprop air-craft will be replaced soon. These aircraft are knownby airliners for their superior low-speed climb capa-bilities and exceptional field performance. Althougha healthy strong migration to faster turbofans hasoccurred for the last several years, prohibitively high-fuel consumption of these aircraft now mitigates theireconomical efficiency.

The market served by the AeroX Horizon 1100expects a low trip cost and equivalent low-speed per-formance of turboprops (field performance, initialclimb capabilities) with an increase in cruise speedand range, characteristics that cause the airlines tomigrate from turboprops to turbofans. These advan-tages are posited to open new routes and increase theefficiency of feeding major hubs. The objective of thedesign was also to offer a standard comfort to passen-gers with an acquisition price that falls in the middleof turboprops and turbofans of the same category.

3.3 Ecole polytechnique de Montreal initial design

The AeroX Horizon 1100 (Fig. 6) was EPdM under-graduates’ initial baseline design [20] – a regional,commercial, low operational cost aircraft that offersvery competitive attributes in terms of field perfor-mance, fuel burn, range, and speed. It representsthe best compromise for replacement of soon-to-be-obsolete turboprop aircraft and high cost to operateturbofan aircraft.

Horizon 1100 has a low-wing morphology, with twoaft fuselage-mounted GE-38 unducted-fan enginesand comprises double-wheeled, retractable landinggears in a tricycle layout. Its double-bubble fuselageis specifically designed to store the baggage under thefloor and equipment in the tail cone. It has wing slats

and single slotted Fowler flaps with engine pylonsacting also as stabilons, and a T-tail to provide thebest control particularly during low-speed trim andmanoeuvres.

3.4 The royal institute of technology technicalassessment

The KTH undergraduates were petitioned to providean aerodynamic and flight mechanics analysis [21] ofthe EPdM initial baseline. Their analysis looked at thesuitability of the chosen aerofoil for the cruise speedsoutlined in the MR&O, and ascertained compatibilityfor meeting additional requirements of good short-field performance so that the aeroplane can addition-ally perform the traditional roles of turboprop, i.e.servicing smaller secondary airports.

Having received the final layout of the aeroplanefrom the project partners at the EPdM, the KTHundergraduates performed an S&C analysis of theproposal using QCARD. In turn, changes were made tothe dimensions of the aerodynamic surfaces and totheir position with respect to the fuselage in orderto conform to minimum requirements dictated byhandling qualities.

3.4.1 High-speed and low-speed aerodynamicsassessment

The Horizon 1100 is planned to cruise at an altitude of35 000 ft (10 670 m) at Mach numbers between MLRC =0.65 (long-range cruise) and MMCRZ = 0.75 (maximumcruise speed); the normal or standard operationalcruise Mach number being MSTD = 0.70. At such low-to mid-transonic speeds, the compressibility effectscannot be neglected, and thus, a high-speed studyof the aerodynamics of the wing was deemed nec-essary. Indeed, it was appreciated upfront that a

Fig. 6 AeroX Horizon 1100 is a regional low operational cost aircraft with two aft fuselage-mountedGE-38 unducted-fan engines



better notion of the penalizing aspects imparted bycompressibility effects or the wave drag componentneeds to be gained. At the low- to mid-transonic speedregime, wave drag is considered as being attributableto shock waves appearing over the wing surface dueto localized super-velocities.

For the purpose of examining high-speed aero-dynamic performance, the MSES (numerical aero-dynamics software that employs a coupled viscous/inviscfd Euler method) [22] code was used. There areseveral ways to reduce and delay the drag rise dueto shock formation on the wing. The most dramaticapproach is achieved by selecting an appropriateaerofoil profile, followed by sweeping the wing, andfinally by tuning the wing geometry by adjusting thetwist distribution. The EPdM baseline design usedthe supercritical airfoil MS(1)-0313 [23], and the mainwing had a quarter-chord sweep of 17.6 ◦.

3.4.1.1 Two-dimensional analysis. Initially, a two-dimensional analysis of the selected wing profileMS(1)-0313 had been done using simple sweep the-ory [24] combined with results produced by the MSEScode. The variation of the drag coefficient (Cd fortwo-dimensional and CD for three-dimensional) withincreasing Mach number for different lift coefficientsis presented in Fig. 7. This figure shows a sudden risein the Cd when the wing enters the transonic regime.Up to approximately M = 0.60, the Cd appears some-what constant since it increases very slowly. As Mbecomes larger, it increases at a geometric rate due tothe wave drag effects. When it comes to aerodynamictailoring, there exist two relevant speeds: the so-called

Fig. 7 Predicted drag coefficient versus Mach num-ber for the Horizon 1100 regional transportcomputed with MSES for different constant liftcoefficients

critical Mach number (Mcr for two-dimensional andMCRIT for three-dimensional) and the drag divergenceMach number (Mdd for two-dimensional and MDD forthree-dimensional). Mcr is the M when supersonicflow is first encountered over the wing profile. Mddis the M at which a pronounced divergence of dragtakes place – it is generally defined to be the M whenthe Cd value is 0.0020 (20 counts) higher than thedrag coefficient at Mcr. Figure 7, displays the valuesfor both Mach numbers and corresponding operat-ing lift coefficient (Cl for two-dimensional and CLfor three-dimensional). Upon inspection of Fig. 7,one can conclude there are preliminary indicationsthat the wing of Horizon 1100 will fulfil the high-speed design requirements. The Mdd is always higherthan the MMCRZ and is always at least 0.15 higherthan the Mcr, i.e. in a three-dimensional sense it isMDD − MCR � 0.15.

3.4.1.2 Three-dimensional study: TORNADO code.The aim of this part of the analysis was to determinethe speed at which the aeroplane achieves the req-uisite lift for take-off. The maximum take-off weight(MTOW) of the aeroplane is 28 350 kg (62 500 lb). Dur-ing the take-off roll phase TORNADO calculationsindicate that enough lift will not be attained until anairspeed of 164 Knots, Calibrated Air Speed (KCAS)(84.0 m/s) is reached. At first glance, this speed wasconsidered to be quite high, but of course it doesnot take into account the pilot, will rotate at a lowerspeed, thereby significantly increasing the operat-ing CL. Rotating to an angle of 13.5 ◦ is close to thestall angle of 14 ◦ but shows that take-off is possiblefrom a minimum of about 105 KCAS (54.0 m/s).

One of the requirements for this aeroplane isthe ability to perform steep approach operations asrequired at certain airports, e.g. London City Airport.To achieve the lift required to maintain a constantrate of descent (with a constant airspeed of 115 KCASor 59.0 m/s), an angle of attack of 7.6 ◦ is requiredaccording to TORNADO results.

3.4.2 Stability and control assessment

One important task for the KTH students was toinvestigate the S&C attributes of the Horizon 1100aeroplane. Such an analysis is most relevant, espe-cially for a new design such as the one discussedin this treatise. An aeroplane in a steady flight con-dition constantly faces many disturbances, and topredict how the aeroplane will behave and how itscontrollability is affected is absolutely necessary. Withthis in mind, the KTH undergraduates used QCARDto predict moments of inertia, aerodynamic coeffi-cients, and the stability modes of motion. QCARDis constructed to predict and visualize the stability



modes, and together with TORNADO, is able to cal-culate other mechanical and aerodynamic propertiesof an aeroplane.

Using QCARD, the stability modes of the initialbaseline design generated by the EPdM undergrad-uates were examined. It is highlighted that EPdMundergraduates employed the traditional, so-calledvolume coefficient method for sizing the empen-nage; examples of such an approach can be foundin references [18] and [25] to [27]. By studying theseresults they concluded that the aeroplane had poorS&C and flight handling qualities (FHQ). In partic-ular, the short-period and Dutch-roll modes provedto be quite unsatisfactory. As a result of this circum-stance, significant improvements had to be madein order to ameliorate these problems and to keepwithin the frame of acceptable manoeuvrability andcontrollability.

3.4.2.1 Improved configuration design. As it will bemotivated in the next sections, the initial EPdM-generated geometry was found to be unsatisfactoryand the KTH undergraduates took it upon themselvesto do an array of geometrical modifications. Initially,the main wing was moved forward; its reference area(SW) was increased; and its aspect ratio (AR) and dihe-dral (�) were reduced. In addition, the horizontal tailhad its area, SHT, and AR, ARHT, increased. Finally, thearea of the vertical tail (SVT) was reduced, due to theforward shift of the wing that tended to increasethe tail moment arm. Figure 8 illustrates the differ-ences between original and refined sizing for Horizon1100.

On the basis of the aforementioned resizing ofthe wing and empennage, the undergraduates thenproceeded to calculate the static aerodynamic deriva-tives, and subsequent to this, predicted all five modesof motion (Table 2 for results of Horizon 1100 origi-nal and refined sizing). They found that the refinedconfiguration produced better drag properties andslightly less available lift – poignantly though, a more

Fig. 8 Comparison of sizing for stability and controlconducted by Ecole Polytechnique de Mon-treal (left) using volume coefficients and KTH(right) using semi-empirical analysis for staticderivatives

benign longitudinal static stability derivative, CMα,was predicted. Their trade thus gained improved lon-gitudinal static stability characteristics at the expenseof some loss in available lift.

3.4.2.2 Short-period stability mode. Most of theS&C problems of the initial geometry were associ-ated with the Short-period longitudinal mode. Thequalities of this mode when the aeroplane is in a stan-dard cruise condition are presented in Table 2. Theimproved geometry has a shorter period and, moreimportantly, a shorter time-to-half, which is one keycharacteristic when attempting to fulfil the FHQ cri-terion. The collective impact of the improvementscan be clearly understood upon inspection of Fig. 9,where the results concerning the Short-period modehave been plotted in an Engineering Science dataunit (ESDU) opinion-contour graph [28] and usinga Cooper–Harper rating scale for all cruise conditions(M = 0.65, 0.70, 0.75 at 35 000 ft).

Figure 9 shows that the initial geometry had avery poor rating – bordering on unacceptable FHQ.Although the FHQ improve with increasing M, theyare still categorized as poor. In fact, it appears thatthe aircraft has a sufficient level of damping ratio, butthe natural frequencies are too small, a direct resultof the period- being too long. As a consequence, theresponse of the aeroplane would prove to be verysluggish and an excessive amount of pilot invokedcompensation would be required. The conclusion onewould draw from the opinion plot is that a largecontrol-fed motion is needed to handle the aeroplanesatisfactorily, thus inferring the vehicle would be quitedifficult to trim.

Fig. 9 ESDU [28] evaluation of the short-period char-acteristics for the Ecole Polytechnique de Mon-treal baseline configuration and the KTH refinedconfiguration



To remedy this problem, the main wing was shiftedforward while SW was increased, and AR decreased.In consort, the horizontal tail SHT and ARHT wereincreased. As it is evident in Fig. 9, the improvedgeometry still has quite a good damping ratio of 0.52,however, the natural frequencies have now becomehigh enough (the period has been reduced, Table 2).Thus, the aeroplane still has a relatively slow responsebut its FHQ can now be considered to be acceptable –even rated as satisfactory for the highest M . Noticethat, as for the initial geometry, the FHQ improveswith increasing M .

3.4.2.3 Dutch-roll stability mode. An irritable levelof Dutch-roll motion is a common problemencountered by many swept-wing aeroplanes. It isparticularly exacerbated when the aeroplane hasan over-prescribed level of lateral stability. Table 2gives the characteristics of the Dutch-roll and Fig. 10compares them to military specifications (MIL-spec)[29, 30] requirements. As can be observed in Fig. 10,the Dutch-roll characteristics of the initial geometryare not satisfactory since the damping ratio is too low.This is due to the excessive lateral stability of the ini-tial design that leads the aeroplane to overshoot itsequilibrium position by a large margin during theDutch-roll oscillatory motion. To reduce the level oflateral stability, and thus cure the Dutch-roll prob-lem, the main wing � and the vertical tail SVT werereduced – it resulted in an improved geometry witha much higher damping ratio. Most noteworthy, thearray of geometrical modifications did not affect thevalue of the Dutch-roll period much; Fig. 10 indicatesthat the improved geometry meets minimum FHQrequirements.

3.4.2.4 Limiting speeds. To conclude the S&C anal-ysis of the initial and improved designs, prediction ofthe limiting speeds for dynamic motion of the Hori-zon 1100 was deemed important. The limiting speedis the speed at which one of the longitudinal or lateralmodes of the aeroplane first becomes unstable. Forboth initial and improved geometry, it was observedthat the spiral mode first becomes unstable and itoccurs at speeds:

Fig. 10 Dutch-roll characteristics for the Ecole Poly-technique de Montreal baseline configurationand the KTH refined configuration superposedwith MIL-spec [27, 28] evaluation levels

(a) 136.5 m/s (M = 0.440) for the initial design;(b) 141.5 m/s (M = 0.455) for the improved design.The initial design remains stable at slightly lowerspeeds than the improved one but the difference isnegligible. Moreover, such low speeds should neverbe reached in cruise conditions (35 000 ft); and even ifthe spiral mode becomes unstable, it has a quite largetime constant (period), which gives the pilot timeto counteract any trajectory deviance. In fact, thoseresults bow to the evidence that both geometries ofthe Horizon 1100 have a very large safety marginbefore becoming unstable in the cruise condition.

3.4.3 Lessons learned

During this project, two cycles of conceptual designwere completed. First, starting from an initial geome-try suggested by the undergraduates of EPdM in theiraircraft design project, an enhanced study of the aero-dynamics of the wing, followed by an S&C analysis ofthe design of the full aeroplane was performed. TheS&C analysis highlighted that some aspects of the lon-gitudinal and lateral FHQ of the Horizon 1100 initial

Table 2 Characteristics of the longitudinal and the lateral modes during standard cruise(M = 0.70, 35 000 ft); computed with QCARD

Period (s) Time-to-half (s) Cycles-to-half

Type of motion Name Initial Imprd Initial Imprd Initial Imprd

Longitudinal Phugoid >120 >120 94.98 95.9 �1 �1Short-period 4.35 2.92 0.82 0.58 0.19 0.20

Lateral Dutch-roll 5.11 5.40 12.94 6.84 2.52 1.26Spiral 128.4 117.5 NA NA NA NARolling convergenc 0.86 0.89 NA NA NA NA



sizing were not acceptable. On the basis of a greaterlevel of detail, although still with simplified analy-sis techniques, an improved geometry was identifiedwith satisfactory FHQ.

Although the alternative wing suggested by the KTHteam provided satisfactory FHQ, a trade-off in termsof higher fuel consumption due to increased drag (notconsidering the more benign wave drag) did result.Insofar as landing approach, the low wing AR alterna-tive proved to be advantageous with its significantlyhigher drag. Notwithstanding this outcome, neitherwing alone produces enough drag for steep approachoperations and hence necessitates the incorpora-tion of additional spoilers or dedicated airbrakes.Concerning the notion of high-speed aerodynam-ics, the choice of the supercritical aerofoil sectionMS(1)-0313 was found to be a satisfactory one sinceit fulfils good design practice of MDD − MCR � 0.15.Combined with the fact that MDD is always higherthan MMCRZ means, the drag rise behaviour was there-fore considered to be sufficiently benign in any cruisecondition.

During the S&C analysis, it appeared that the longi-tudinal short-period and the lateral Dutch-roll modesof motion of the initial geometry were sources ofconcern. Indeed, the inherent characteristics of theaeroplane for these two modes meant that therewas a likelihood that FHQ would be below mini-mum acceptable levels. This circumstance warrantedsome modifications to the geometry, leading to animprovement to the characteristics of the aeroplanefor these two modes (short-period and Dutch-roll)whilet preserving the desirable characteristics of thelongitudinal Phugoid and lateral spiral and rollingconvergence modes.

In conclusion, starting from a geometry that hadsatisfactory aerodynamic characteristics but unac-ceptable S&C features, an improved geometry hasbeen found that satisfies qualities for both aerody-namics and S&C.

4 INSTRUCTION IN FLIGHT-CONTROL: WRIGHTGLIDER REVISITED

The second student project carried out at KTH wasquite different from the Horizon 1100 project in anumber of ways: it was not conducted as a col-laborative exercise outside of KTH, and was not anaircraft design project per se. Instead the aim was toaid in teaching the various elements of S&C of air-craft. In order for the undergraduates to carry out theFHQ analysis of the Horizon 1100, and to improveupon it, they needed a sufficient understanding offlight dynamics and acquire a new proficiency inmaking an application of it. In teaching and guid-

ing undergraduates to reach this level of proficiency,several difficulties were observed – all having to dowith the degree of abstraction of the subject matter.For example, given some numerical values for theaerodynamic static derivatives, they were asked tosurmise how these values, once inserted into theequations governing dynamic flight would translateinto influence on certain flight trajectories and asso-ciated FHQ. If this process could be made moreconcrete, the supposition was that the teaching, andthus the understanding of S&C would be markedlyimproved.

The approach that was finally employed drew offthe synergy of combining MATLABTM with an alreadydeveloped SIMULINK simulation tool together withsome simple wind tunnel testing. The intentionwas to retrace the steps of the Wright brothersbecause after all they were the ones who first solvedthe problem of pragmatic controlled flight with-out the benefit of the theory of flight dynamics,and in essence, were the progenitors of modernflight control. Others have taken a similar approach,for example Padfield and Lawrence [31] extensivelyinvestigated the Wright Gliders and correspondingcontrol (Fig. 11). The reader is referred to Culick [32]for an interesting overall account of what the Wrightslearned.

In 1903, Orville and Wilbur Wright performed thefirst controlled flight of a powered aircraft with the1903 Wright Flyer. The progress to obtain the flyerwas over a long timeline and the Wright Brothers builtseveral kites and gliders before they understood thenecessary aerodynamics to build their powered flyer.Prior to the 1903 Wright Flyer, the Wright brothers

Fig. 11 The Wright brothers flight tested their gliderwith the unconventional forward elevator as atethered kite in Kitty Hawk in 1900. It was herethey discovered the rudimentary elements ofstability and control



constructed the 1902 Wright Glider (Fig. 12), whichwas the first vehicle with a three-dimensional controlcapability and one that addressed the problem offlight control.

One is immediately drawn to the question: howdid they reach this accomplishment? Already in 1899,they focused on solving the problem of lateral con-trol by the pilot ‘warping’ the wing, and they testedit on a 5.0 ft (1.52 m) biplane kite. Confident theirdesign was sound, in 1900 the Wrights built a 17.0 ft(5.18 m) glider with an unusual forward elevator (orcanard). They went to Kitty Hawk hoping to gainflying experience, but the wings generated less liftthan expected, so they flew the glider mostly as akite, working the control surfaces from the ground.Figure 11 is a photograph of the glider configuredfor tethered flight in 1900. The results of these testsconvinced the Wrights that they had achieved a suffi-cient level of longitudinal as well as lateral control.Over the course of 1901, they improved the lift oftheir glider and demonstrated that it was possibleto control gliders in two spatial dimensions, namely,pitch and roll. Irrespective of this achievement, themachine still proved to be somewhat unpredictable:when the pilot raised the left wing to initiate anexpected right turn, the vehicle instead tended to slipto the left (adverse yaw). Drawing on all their research,theWright brothers altered the 1902 vehicle with morelift-efficient 32.0 ft (9.75 m) wings, incorporated a ver-tical tail including a movable rudder linked to thewarping mechanism, and as a consequence, foundit was also possible to control the yaw of the gliderso that it could be turned and stabilized smoothly.Six hundred glides that year satisfied them that theyhad the first working airplane (Fig. 12). In a classicexample of the ‘cut-and-fly’ approach during those600 glides, they determined the requisite size of thevertical tail.

Fig. 12 The 1902Wright Glider, the machine that solvedthe problem of flight control

4.1 Student assignment definition

The assignment given to the students then is, forstick-fixed conditions only, use wind tunnel testingtogether with simulation to determine the size ofthe vertical tail that gives acceptable yaw stabilityto the tethered 1902 Glider. In particular, begin withthe initial (given) tail size and carry out the followingactions.

1. Try first to obtain a stable and trimmed conditionfor the Glider with initial vertical tail tethered inthe wind tunnel. Confirm this motion with KiteSim[33]. You will observe in both cases that the Glideris unstable in yaw.

2. Use KiteSim to determine how much additionalauthority is needed from the vertical tail to achievestability. Use TORNADO to determine the size of anew tail that gives this authority.

3. Verify in wind tunnel that Glider with new tail is infact stable. Compare the measured and computedtrajectories.

4. Draw conclusions.

4.2 Wind tunnel model

A wind tunnel model of the 1902 Wright Glider wasbuilt for low-speed wind tunnel tests conducted in atunnel with the cross-sectional size of 300 × 600 mmat the test section. This is a tunnel for student use andis much smaller than the one used in, for example,reference [34]. The model of the glider was built witha wingspan of 150 mm in order to avoid any inter-ference from the wind tunnel walls, and therefore,ensuring a suitable account of finite wing effects. Asthe real glider has a wingspan of 9.75 m, the wind tun-nel model was therefore built to a 1/64 length scale.The wind tunnel model was a simplification of theglider as only the main components were consideredin the construction. The model consisted of the wings,the struts, and the skids only. Figure 13 shows thecomputer-aided design (CAD) description of the windtunnel model.

4.2.1 Measurement system

The wind tunnel test results were analysed with theMATLABTM program DoTrack [35]. The tests wererecorded with a camera that was mounted on top ofthe wind tunnel. The video clips obtained from thewind tunnel tests were analysed with DoTrack in orderto deduce the motion of the model. DoTrack analy-ses each frame captured by the camera and detectsbright spots painted on the model, positioned andnumbered as shown in Fig. 14.

If the relative distance between the bright spotsare known and defined in the program, the relativemotion of the bright spots can be calculated from one



Fig. 13 Solid-edge CAD description of the wind tunnelmodel of the 1902 Glider used to calculate itsmoments of inertia

frame to another. Hence, the motion and attitude ofthe model can be determined. To highlight the brightspots on each frame, the wind tunnel backgroundwas darkened with a black sheet and the model waspainted matte black.

4.3 Flight-dynamic model: KiteSim

A dynamic simulation, using a SIMULINK-based pro-gram called KiteSim, was carried out in order toexamine the flying qualities of the tethered 1902Glider. The flight dynamics model was set up inMATLABTM SIMULINK with AEROSIM, an aeronau-tical simulation BLOCKSET v1.1 [36]. This informa-tion was useful in understanding how the WrightBrothers’ subsequent designs evolved as a result oftheir test program. For the simulations in KiteSim,an aerodynamic database of the 1902 Wright Gliderwas required. The database included the followingconstituent information:

Fig. 14 Placement of eight dots on the wind-tunnelmodel that the optical measurement systemDoTrack [35] uses to determine the model’sposition and orientation in space as if movingin time

(a) aerodynamic static derivatives computed withTORNADO;

(b) drag data, estimated using suitably calibrated,standard conceptual design semi-empirical algo-rithms;

(c) moments of inertia calculated about the centre ofgravity from the Solid-Edge CAD model.

KiteSim employs a linearized aerodynamics modelsimilar to LaRCsim [37], where for instance the lin-earized CL is expressed in equation (1)

CL = CLo + CLαα + CLββ + CLpp + CLqq + CLrr (1)

where CLo is the initial operating lift coefficient, CLαthe aeroplane lift-curve slope, a the angle of attack,CLβ the rolling moment due to side-slip (the so-calleddihedral effect), β the angle of vehicular side-slip, CLpthe so-called dumping-in-roll derivative (resistanceof vehicle to rolling motion) with roll rate, p, CLq therolling moment due to pitching motion (q), CLr therolling moment due to yawing motion, and r the yawrate. The SIMULINK model implements a full set ofequations of motion for a rigid body aircraft.

The total drag, CD, was simplified to a standardizedpolar behaviour, according to equation (2)

CD = CDo + KC 2L = CDo +C 2L

πeAR(2)

where CDo is the zero-lift drag and e is the Oswaldefficiency parameter.

The tether line was modelled as a simply dampedspring system. Although this model would acceptcompression loads only, tension loads were observedin the simulations to follow. According to equation (3),the magnitude of the force, F̄t, was linearly propor-tional to the line elongation, ε, and to the deformationspeed, ε̇, of the tether

F̄t = k1ε + k2ε̇ (3)

k1 and k2 are constants of proportionality.The force transmitted by the line connected to the

Glider was taken to be analogous to the propulsionhub of the kite. This implies that the line force is con-sidered as a propulsion force, acting for a standardaircraft in what would be the propeller hub, and in thedirection of the kite tether. The aerodynamic loads onthe tether were neglected as the total drag force on theline was about four orders of magnitude lower thanthe loads on the kite.

4.3.1 Aerodynamic derivatives, drag, and inertias

A model of the 1902 Wright Glider was created inTORNADO, from which the aerodynamic coefficientsand static derivatives could be calculated. The wings



Fig. 15 Panelling in the TORNADO geometric domainof the 1902 Glider used for the computation ofthe aerodynamic derivatives

were taken as thin plates, and the struts, skids, andwires were neglected. Thus, only the main wings,the stabilizer, and the vertical tail are represented,although for reasons of numerical stability, the tailhad to be simplified to a quadrilateral plate. Figure 15presents the TORNADO representation of the 1902Glider.

The aerodynamic coefficient dependency on α, β, p,q, and r were computed by TORNADO, and from theseresults, it was possible to estimate the aerodynamicderivatives required for the KiteSim simulations.

The drag estimate required values for the mini-mum drag condition, the lift at minimum drag, ande. In order to simplify the drag data calculations,some simplifications of the glider were made. To beginwith, the glider was divided into these main com-ponents: the main wings, the stabilizer, the tail, andthe struts between the main wings. A componentbuild-up approach was subsequently performed. Thissimplification was done in order to avoid any interac-tion between the components, which would undulycomplicate the analysis. The lifting surfaces were sim-plified as flat rectangular plates congruous with the

original surfaces in terms of planform lengths andarea. Furthermore, in the analysis, the struts weresimplified as infinitely long circular cylinders. As thewings of the glider were assumed to be flat symmet-ric plates, the minimum drag could be assumed tooccur at α = 0. For the wings, the friction drag coeffi-cient on each side of the plate was calculated with theBlasius relation [38]. Hence, the total drag at a veloc-ity of V = 10 m/s was estimated to be D = 50.6 N,which corresponds to the total minimum drag ofCD,min = 0.059 as the area of the main wing was chosenas SW . Table 3 presents the data used in the KiteSimsimulation.

4.4 Comparison results: sway response

The Glider with the initial-size tail was seen to behighly unstable laterally in both the wind tunnel testand in the simulation. It is highlighted, however, thatthere are discrepancies between the physical model ofthe Glider and the model invoked in KiteSim and TOR-NADO. Basically, the discrepancies arose by virtue oftolerances in the fabrication of the model.

In order to achieve lateral stability, values for theforces and moments on the vertical tail could be arbi-trarily increased in KiteSim until stable flight wasreached. Then, by adjusting the geometrical size ofthe tail in TORNADO, the values for the forces andmoments on the vertical tail determined in KiteSimthereby facilitating sizing. Theoretically, the investi-gated configuration should exhibit stable qualities inyaw. To verify this posit, the wind tunnel model wasgiven this sized vertical tail and tested. The model wasfirst positioned in a static and trimmed state; and ata given instant, a disturbing force was applied to thewing in a spanwise direction. From the analysis pro-duced by DoTrack, the response to this force in thelateral coordinate, X (t) of the glider was determined.Similarly, KiteSim computes the same response, andthese two sets of data as shown in Fig. 16 were com-pared. Both data sets show a damped behaviour that

Table 3 Data required for the KiteSim simulation

Angle of attack Side-slip Pitch rate Roll rate Yaw ratedependency dependency dependency dependency dependency

(per rad) (s/rad) (s/rad) (s/rad)

(A) Aerodynamic and static derivatives computed with TORNADOCLo = 0.00 CYβ = −0.218 CLq = 0.556 CYp = −0.0120 CYr = −0.0440CLα = 7.35 per rad CLβ = −0.006 40 CMq = −0.257 CLp = −0.248 CLr = 0.001 90CMo = 0.00 CNβ = −0.0610 CNp = 0.000 800 CNr = −0.0120CMα = −0.920 per rad(B) Minimum drag estimationCD,min CL,minD eo0.0590 0.00 1.24

(C) Moments of inertia computed from solid edge CADIxx (kg m2) Iyy (kg m2) Izz (kg m2) Ixz (kg m2)358 51.4 363 3.00



Fig. 16 Comparison between simulated and exper-imental sway response, the sway positionX (t) versus time. Both cases show a dampedbehaviour built up of two frequencies

is built up of two frequencies. One oscillation in side-slip angle produced a frequency of 1 Hz, whereas theone pendulum mode about the tether anchor pointproduced 0.8 Hz. With this evidence in place, thewind tunnel test proved to confirm the theoreticalprediction of KiteSim.

5 CONCLUSION FOR BOTH PROJECTS ANDLESSONS LEARNED

The collaborative teaching approach to aircraft designeducation was found to produce three groups of ben-eficiaries. The first was industry, due to the fact thatthe students who have experienced some of the chal-lenges of managing the design of a high-value man-ufacturing product can be offered this approach. Thesecond, through quality of education in a direct fash-ion, is the focal institutions, namely, EPdM and KTH.On a more subtle level, the third, indirect throughquality of education, is the institutions resulting fromever-increasing numbers of international studentsthat participate in such design projects. Undergrad-uates from Bristol, UK; Milan Italy; and Toulouse,France have found an opportunity to participate inconjunction with the native Canadian and Swedishundergraduates. Collectively, the collaborative teach-ing approach has also served as a catalyst for openingup new research opportunities as well as assistingwith the introduction of new initiatives like industryinternship programmes.

A synopsis of the lessons learned from experiencingthe collaborative teaching approach to aircraft designeducation is as follows:

(a) allows opportunity to gain a greater insight into aspecific technical discipline and comprehend thecomplex interactions;

(b) foster skills on how to interact in a goal-orientedtechnical team;

(c) experience some of the challenges of managingthe design of a high-value product;

(d) effective communication ensures project gover-nance;

(e) appointing a project management coordinatoris key to success; achieved via structured andfrequent technical group meetings;

(f) requirement for a final review to senior academicstaff and industry professionals fosters a highcalibre of presentation skills.

Among the conclusions of experience from flightcontrol learning, are the following benefits:

(a) a deeper insight into the various factors affectingaircraft stability and control;

(b) appreciation of the simulation–experiment syn-ergy;

(c) possibility to involve multi-disciplinary contentmatter and skills, ranging from programming inSIMULINK to building models in balsa wood totechniques in computer-based visualization.

It is worth mentioning the advantages of usingMATLABTM and SIMULINK as the development toolsfor both TORNADO and KiteSim. Because MATLABTM

and SIMULINK are nowadays widely used both inuniversity and industry, undergraduates come suit-ably prepared when undertaking the aforementionedcourses, and they have an ability to continue workingwith the same tools when they go to industry. Fur-thermore, when students ‘fly’ their own designs (evenif this is just in KiteSim), the design process becomesmuch more interesting and captivating.

REFERENCES

1 Argüelles, P., Bischoff, M., Busquin, P., Droste, B. A. C.,Evans, R., Kröll, W., Lagardère, J., Lina, A., Lumsden, J.,Ranque, D., Rasmussen, S., Reutlinger, P., Robins, R.,Terho, H., and Wittlöv, A. European aeronautics: avision for 2020, 2001 (Advisory Council for AeronauticsResearch in Europe (ACARE)).

2 Isikveren, A. T. Quasi-analytical modeling and opti-mization techniques for transport aircraft design. Doc-toral Dissertation, Department of Aeronautics, RoyalInstitute of Technology (KTH), Sweden, 2002, Report2002-13.

3 Simos, D. Project interactive analysis and optimization –PIANO version 3.8., 1990–2001 (Lissys Limited, Wood-house Eaves, UK).

4 Advanced aircraft analysis. User’s manual, version 3.1,2006 (DARcoporation).



5 Raymer, D. RDS aircraft design software, available fromhttp://www.aircraftdesign.com/rds.html, 2007.

6 Melin, T. TORNADO a vortex-lattice MATLAB imple-mentation for linear aerodynamic wing applications.Masters Thesis, Royal Institute of Technology (KTH),Department of Aeronautics, Sweden, 2000.

7 Berard, A. Development and implementation of aircraftconceptual design methods. Masters Thesis, Royal Insti-tute of Technology (KTH), Department of Aeronautics,Sweden, 2006.

8 Ricci, S. and Terraneo, M. Conceptual design of anadaptive wing for a three-surfaces airplane. 46thAIAA/ASME/ASCE/AHS/ASC Structures, StructuralDynamics, and Materials Conference, AIAA-2005-1959,Austin, Texas, 18–21 April 2005.

9 Moschetta, J. and Thipyopas, C. Optimization ofa biplane micro air vehicle. 23rd AIAA AppliedAerodynamics Conference, AIAA-2005-4613, Toronto,2005.

10 Ricci, S. and Terraneo, M. Conceptual design ofan adaptive wing for a three-surfaces airplane.46th AIAA/ASME/ASCE/AHS/ASC Structures, StructuralDynamics, and Materials Conference, Austin, Texas,2005.

11 Mitchell, C. G. B. A computer programme to predict thestability and control characteristics of subsonic aircraft.TR 73079, Royal Aircraft Establishment, ProcurementExecutive, Ministry of Defence, 1973.

12 Melin, T. Using internet interactions in developing vor-tex lattice software for conceptual design. LicentiateThesis, Department Aeronautical and Vehicle Engineer-ing, Royal Institute of Technology (KTH), Sweden, 2003,Report TRITA AVE 2003:12.

13 Moran, J. An introduction to theoretical and computa-tional aerodynamics, 1984 (Wiley and Sons, New York,USA).

14 Leisher, L. L. and Walter, H. L. Stability derivatives ofcessna aircraft. Company report 1356, Cessna AircraftCompany, Wichita, 1957.

15 Drela, M. and Youngren, H. AVL 3.26 user primer, 2006(MIT Aeronautics and Astronautics, Massachusetts).

16 Ashby, D. L. Potential theory and operations guide forthe Panel Code PMARC 14. Technical report TM1999-209582, NASA, 1999.

17 Bil, C. Development and application of a computer-based System for conceptual aircraft design, 1988 (DelftUniversity Press, The Netherlands).

18 Raymer, D. P. Aircraft design: a conceptual approach, 3rdedition, 2002 (AIAA, New York, USA).

19 Cook, M. V. The new age of flight control, aerogram, 1999,vol. 9, no. 4, pp. 9–13 (Cranfield University).

20 Isikveren, A. T., Moisan, A., Houle, N., Postic, R.,Benga, V. A., Tremblay, F., Lesauvage, G., Waffo-Mala, P. C., Moussallem, S., Bodnariuck, P., Vu,K. C., Arsac, B., Longchamps-Belzil, J., BourgeoisCollin, L., Bernier, S., Sadegh, A., Simard Johnson,M. A. S., Brun-Khoobeelass, A., Ivity, M., Palobde-Zagre, G., Corinne Waffo-Mala, P. C., Moussallem, S.,and Ruhlmann, P. AeroX Horizon 1100 PreliminaryEngineering Document, Group submission for AE4950 –2005 Aircraft Design Course, Ecole Polytechnique deMontreal, Canada, 2005.

21 Alwan, H., Berard, A., and McKechnie, G. Aerodynamicreport on the Horizon 1100 concept. Group submissionfor 4E1233 –2005 Project Course in Aerospace. Depart-ment of Aeronautical and Vehicle Engineering, RoyalInstitute of Technology (KTH), Sweden, 2005.

22 Drela, M. A users guide to MSES 2.97, 1996 (MIT Com-putational Aerospace Laboratory, Boston).

23 McGhee, R. J. and Beasley, W. D. Low speed aerody-namic characteristics of a 13 percent thick medium-speedairfoil designed for general aviation applications. NASAtechnical paper-1498, 1979.

24 Jones, R. T. Wing plan forms for high-speed flight. NACATN No.-1033, 1946.

25 Roskam, J. Airplane design, Part I through VIII,1990 (Roskam Aviation and Engineering Corporation,Kansas).

26 Torenbeek, E. Synthesis of subsonic airplane design, 1982(Delft University Press, The Netherlands).

27 Obert, E. Some aspects of aircraft design and air-craft operation, 1996 (Linkoping Institute of Technology(LiTH), Aircraft Design Lecture Series, Sweden).

28 A background to the handling qualities of aircraft, 1992(Engineering Science Data Units, (ESDU)), 92006.

29 Military specification, flying qualities of piloted air-planes, MIL-F-8785C (USAF), 1980.

30 Military standard, flying qualities of piloted airplanes,MIL-STD-1797 (USAF), 1987.

31 Padfield, G. D. and Lawrence, B. The birth of flight con-trol: an engineering analysis of theWright brothers’ 1902glider. Aerosp. J., December 2003, 697–718.

32 Culick, F. E. C. What the Wright brothers did anddid not understand about flight mechanics – in modernterms. AIAA paper Number 2001–3385, 2001.

33 Gonzalo, S. A. and Marguez, C. P. Kite flight simulator.1st AIAA-Pegasus Student Conference, Toulouse, 19–20May 2005.

34 Kochersberger, K., Sandusky, R., Ash, R., Britcher, C.,Landman, D., and Hyde, K. An evaluation of the Wright1901 glider using full scale wind tunnel data. AIAA paperNumber 2002–1134, 2002.

35 Melin, T. Multi-disciplinary design in aeronautics,enhanced by simulation-experiment synergy. DoctoralDissertation, Royal Institute of Technology (KTH),Department of Aeronautical and Vehicle Engineering,(ISBN 91.7178-373-3). Sweden, 2006.

36 AEROSIM BLOCKSET version 1.1 user’s guide, 2003(Unmanned Dynamics LCC, Oregon, USA).

37 Jackson, E. B. Manual for a workstation-based genericflight simulation program (LaRCsim). NASA technicalmemorandum 1064, 1995.

38 Abbot, I. and Doenhoff, A. Theory of wing sections, 1959(Dover Publications, New York).

APPENDIX

Notation

C coefficient for non-dimensional aerody-namic parameters; coefficient for staticstability derivatives



e Oswald efficiency parameterF̄t magnitude of force within tether line (N)k1 coefficient of proportionality in quanti-

fying tether line force (kg/s2)k2 coefficient of proportionality in quanti-

fying tether line force (kg/s)M Mach numberp vehicle rate of roll (deg/s or rad/s)q vehicle rate of pitch (deg/s or rad/s)r vehicle rate of yaw (deg/s or rad/s)S reference planform area (m2)t timeV velocity (m/s)X sway position

α angle of attack (deg or rad)β vehicle side-slip angle (deg or rad)� wing dihedral (deg)ε tether line elongation (m)ε̇ tether line deformation rate (m/s)

Subscripts

cr critical point for drag creep, two-dimensional

CR critical point for drag creep, three-dimensional

d drag, two-dimensionalD drag, three-dimensionaldd critical point for drag divergence, two-

dimensionalDD critical point for drag divergence, three-

dimensionalDo denotes zero-lift dragHT horizontal tailI moments of inertia (kg m2)l lift, two-dimensionalL lift, three-dimensionalLp denotes the vehicle dumping-in-roll or

resistance of vehicle to rolling motion(s/deg or s/rad)

Lq denotes the vehicle rolling moment dueto pitching motion (s/deg or s/rad)

Lr denotes the vehicle rolling momentdue to yawing motion (s/deg or s/rad)

LRC Long-range CruiseLα denotes the vehicle lift-curve slope

(per degree or per radian)Lβ denotes the vehicle rolling moment

due to side-slip or dihedral effect (perdeg or per rad)

min minimum conditionminD minimum total drag force conditionM momentMCRZ maximum cruiseMq denotes the vehicle pitching moment

due to pitching motion (s/deg ors/rad)

Mα denotes the vehicle moment due toangle of attack, static longitudinalstability parameter (per deg or perrad)

Np denotes the vehicle yawing momentdue to rolling motion (s/deg or s/rad)

Nr denotes the vehicle dumping-in-yawor resistance of vehicle to yawingmotion (s/deg or s/rad)

Nβ denotes the vehicle moment dueto side-slip, static yawing stabilityparameter (per deg or per rad)

o initial conditionSTD Standard or normal cruiseVT vertical tailW wingxx denotes about the longitudinal or

x-axisxz denotes a product of inertia between

x and z axesyy denotes about the lateral or y-axisYp denotes the vehicle side force due to

rolling motion (s/deg or s/rad)Yr denotes the vehicle side force due to

yawing motion (s/deg or s/rad)Yβ denotes the vehicle side force due to

side-slip (per deg or per rad)zz denotes about the vertical or z-axis


special issue paper 175 how industry concepts of ......special issue paper 175 how industry concepts...

Documents