SCALABLE DESIGN APPROACH TO ANALYZE FLIGHTMECHANICAL PERFORMANCE OF TILT-WING UAVS

Marten Schütt∗, Tobias Islam∗, Philipp Hartmann∗, Dieter Moorman∗∗Institute of Flight System Dynamics, RWTH Aachen University, Germany

Keywords: Tilt-wing, VTOL, UAV, Conceptual Design

Abstract

This paper presents a method to evaluate differ-ent tilt-wing concepts for preliminary design. Aseparation of the aircraft into standardized com-ponents supports easy adaption and fast calcula-tions, as all components have identical structure.Despite the flight mechanical complexity of tilt-wing aircraft, this method allows to keep the levelof detail in the range of analytical or semi empir-ical equations with limited experimental data.

The component setup and implementation arepresented as well as a method for flight mechan-ical analysis. Equations to calculate the forcesand moments of each individual component arederived and discussed, whereby each componentcan consist of a propulsion unit, a wing and aflap. The effect of propeller slipstream on thewing aerodynamics are described in detail.

Results of the method in terms of differentairfoils, flap deflections and thrust settings couldbe validated by wind tunnel measurements.

An exemplary tilt-wing aircraft is evaluated,presenting a possible application of the compo-nent buildup method. The special characteristicsof the design of unconventional configurationsare particularly emphasized regarding flight me-chanics.

1 Introduction

Recent UAV designs try to bridge the gap be-tween vertical take-off and landing (VTOL) andefficient cruise flight. Dependent on designatedmissions, the aircraft design has to set an opti-

mal balance between efficiency during hover andcruise flight. Some aircraft in VTOL configura-tion use the same engines1 for hover and cruiseflight. This approach can be found in variousUAVs like Germandrones Songbird [1], NASAGL-10 [2] but also in manned concepts like theJoby J2 [3]. Some UAVs have additional enginesfor hover, like Quantum Tron [4], AVIGLE tilt-wing [5] and DHL Pacelcopter [6]. A differentapproach is to have separate engines, designatedfor hover and cruise flight. This can also be foundin UAVs, e.g. Airbus Quadcruiser [7] or AltiTransition [8].

For these configurations, the engines are ei-ther not efficient, as they are a trade-off betweenhover and cruise flight, or there must be more en-gines on board than necessary. The presented tilt-wing concept with a tiltable tailplane proposes apromising approach to compensate for this prob-lem. This concept uses only three engines duringhover, while one is designed for efficient and fastcruise flight.

The flight envelope of a convertible aircraftis extensive, combining characteristics and ca-pabilities of rotorcraft with those of fixed-wingairplanes. The angel of attack (AoA) vary from−180◦ to +180◦ and the aerodynamic propertiesare highly nonlinear. Approaches to model con-vertible aircraft found in literature are extensivewind-tunnel tests ([9] and [10]), detailed compu-tational fluid dynamics (CFD) analysis [3] or dif-ferent approaches of vortex lattice methods [7].

1Engine is used in the following as a term for the com-bination of electric motor and propeller.

1

MARTEN SCHÜTT , TOBIAS ISLAM , PHILIPP HARTMANN , DIETER MOORMAN

Apart from being expensive and time consuming,these approaches are very sensitive to changes ofthe configuration, which are likely during prelim-inary design phase. Instead an analytical and em-pirical model can be of high value for flight me-chanical analysis. An analytic, component baseddescription of aircraft was already used by Selig[11] for conventional aircraft. In literature dif-ferent approaches for control design of uncon-ventional UAVs can be found, whereby most ofthem are model-based. The applied models areoften component based, in particular Tobing [12],Francenso [13] and Hartmann [14].

Within this paper a generic componentbuildup method (CBM) is presented that sepa-rates the aircraft into individual components. Theforces and moments acting on each element aredescribed by analytical and empirical calcula-tions. Nevertheless, the basic aerodynamic andflight mechanical interactions are taken into ac-count. The model is easily modifiable to evaluatedifferent configurations and their varying flightmechanics. Within this model, the level of detailis kept appropriate for initial design and evalua-tion, but can be increased for selected sensitiveeffects or interactions.

This paper is structured as follows: In Sect. 2the tilt-wing configuration including flight me-chanics is presented, followed by a general in-troduction to the conceptual design approach(Sect. 3). In Sect. 4 all equations to evaluate thedifferent components are presented, followed bya validation, using wind tunnel measurements(Sect. 5). In Sect. 6 an application of the methodto analyze the flight mechanics of tilt-wing air-craft is presented.

2 Tilt-wing Configuration

The tilt-wing configuration combines advantagesof rotary-wing and fixed-wing configurations.During cruise the wing is rotated to form a fixedwing configuration, producing sufficient lift. Forhovering and VTOL the tilt-wing aircraft is ableto rotate its wing around the lateral axis. As themain engines are installed on the wing, the thrustis used to compensate the weight of the aircraft.

fmain,lσ1

faux

σ2

η fmain,r

Fig. 1 Analyzed tilt-wing configuration with itslongitudinal control devices (upper part) and sep-arated into elements by the CBM (lower part)

The tilt-wing configuration includes all controlsurfaces typical to an fixed-wing aircraft. Theailerons are located in the slipstream of the pro-pellers to gain yaw control in hover. To controlpitching moment during hover an auxiliary en-gine is needed.

2.1 Analyzed Configuration

The presented tilt-wing concept includes atiltable tailplane with an auxiliary enginemounted in front, see Fig. 1. During hover thetwo main engines produce most of the thrustwhile the auxiliary propeller mainly supports forpitch control. During cruise flight the main en-gines are switched off and the propellers foldedaway, while the auxiliary engine produces thethrust to compensate the drag of the aircraft. Asthe propellers of VTOL aircraft act in very dif-ferent inflow conditions, the propeller design ingeneral is a trade-off. For the analyzed config-urations the propellers of the main engines canbe designed to be efficient during hover while thepropeller of auxiliary engine is efficient duringcruise flight. This way the tilt-wing aircraft fea-tures efficient and fast cruise flight capabilities in

2

Scalable Design Approach to Analyze Flight Mechanical Performance of Tilt-wing UAVs

contrast to multirotor systems.

2.2 Flight Mechanics

During the transition from hover to cruise flightand vice versa a distinct assignment of controlsurface effects and the aircraft’s rotational axisis not possible. Because of changing aerodynam-ics at different flight states, the actuators effec-tiveness varies in direction and magnitude. De-pendent on a suitable choice of wing and en-gine, the tilt-wing aircraft can perform stationaryflight within the entire flight envelope from hoverto cruise. Stationary longitudinal flight condi-tions for all airspeeds over the entire flight en-velope can be achieved by suitable trim controldeflections and trim thrust settings. Dependenton the configuration, the number of trim controldevices varies and thereby the dimension of thesolution space of trim control deflections for eachflight state. Considering the longitudinal motion,the presented configuration features the follow-ing control devices. The effect of one controldevice may be dependent on another (nonlinearbehavior), e.g. as the engines are mounted on thewing/tailplane:

• main thrust: fmain = fmain,r + fmain,l• auxiliary thrust: faux• main tilt angle: σ1• tailplane tilt angle: σ2• elevator deflection: η

During conceptual design of tilt-wing aircraft,the design has to ensure that stationary flightwithin the entire flight envelope is possible.

3 Conceptual Design Approach

The CBM described in this paper is part ofa generic simulation environment for analyzingflight mechanics and developing control algo-rithms, see Fig. 2 (adapted from Hartmann [14]).The CBM calculates the forces and moments onthe aircraft based on control deflections, thrustsettings and the current inflow. The forces andmoments are used to evaluate the equations ofmotion, which describe rigid body motion [14].

The lateral, longitudinal and rotational motion ofthe aircraft results in a inflow condition. An ex-isting wind model represents the motion of the at-mosphere with respect to the inertial frame. Bothmotions amount to the total inflow. Controllercommands are used in combination with actua-tor models to calculate control deflections. A de-tailed description of the simulation can be foundin [14].

3.1 Objective

The analyzed tilt-wing configuration presents anapproach to bridge the gap between flight per-formance and flight mechanical stability. Dueto the wing and engine design, this configura-tion is more sensitive to effects of propeller slip-stream on the wing and the wing downwash onthe tailplane, compared to previously analyzedtilt-wing concepts. Smaller areas of the wing areinfluenced by the slipstream, which reduces theReynolds numbers and therefore requires moredetail in the description of the aerodynamics. Thelevel of detail of the existing simulation is there-fore increased for these effects.

The challenges of preliminary design of tilt-wing aircraft are quite different from those offixed wing aircraft. Of course aerodynamic ef-ficiency and lift-to-drag ratio during cruise aredrivers, but the flight mechanical analysis overthe entire flight envelope is also of high impor-tance. This claims an appropriate description ofthe slipstream of the propeller as well as the de-scription of the downwash of the wing affectingthe tailplane. Considering these effects helps toanalyze flight mechanics, especially the equilib-rium of moments, with appropriate precision.

The goal of this CBM is to show the signif-icant flight mechanical effects while being ableto evaluate various different configurations in ashort amount of time. The need of external datais limited to coefficients of the airfoils and thrustmodels of the engines. The adaption of differ-ent wings, propellers and overall geometry canbe performed iteratively. This way it is possibleto find an optimized configuration and analyze itfor longitudinal motion.

3

MARTEN SCHÜTT , TOBIAS ISLAM , PHILIPP HARTMANN , DIETER MOORMAN

Actuator

Models

Component

Method

Rigid Body

Motion

Wind

Model

Gravity

Model

Forces and MomentsControl Input

Inflow

Controller

Commands State of Motion

Fig. 2 The presented method, showing its input and output, embedded into an simulation environment,adapted from Hartmann [14]

3.2 Structure

As mentioned before, the presented CBM is partof a simulation environment. In this paper onlythe longitudinal motion is presented, neglectingany lateral flow. Key property of this method isthe generic decomposition of the aircraft and thepossibility to model all different parts by usingthe same component structure. Each componentconsists of a propeller disc, a wing element anda flap, see Fig. 3. The internal interaction of thedifferent elements inside the component are con-sidered, while all components are calculated in-dividually without external interaction in the firstplace. Inside one component the impact of thepropeller slipstream on the wing and the flap isconsidered by calculating individual inflow con-ditions. The size of one wing element is definedby the contracted propeller slipstream, see 4.1.For each component the aerodynamics are con-sidered stationary. There are no iterative calcula-tions, as for each timestep all elements are calcu-lated unidirectional from nose to tail. The loca-tion and geometry of each component in terms ofposition and orientation as well as geometric pa-rameters can be specified and each element (pro-peller, wing or flap) of a component can be omit-ted.

The calculation of each component is exe-cuted as follows. First the external inflow condi-tion is used to calculate the thrust and slipstreamof the propeller (Sect. 4.1). The slipstream andexternal inflow are used to calculate the lift, dragand moment of the wing element, dependent on

¼ lμ

u∞

uin

du

ced

ddisc

Pro

p. D

isc

Win

g E

lem

ent

Fla

p

l μ

0.5 ddisc

Fig. 3 The elements of the component and theirinflow conditions.

the airfoil and flap deflection (Sect. 4.2). The en-tire lift distribution of the wing is used to calcu-late the downwash and its effect on the tailplane(Sect. 4.3).

To achieve an optimized tilt-wing configu-ration, different configurations can be trimmedusing the CBM, for flight mechanical analysis(Sect. 6).

4 Implementation

The equations of motion refer to and the inflowconditions are calculated with respect to the CoGof the aircraft. The aircraft is divided into com-ponents that have a relative position to this refer-ence system. Linear transformations can be used

4

Scalable Design Approach to Analyze Flight Mechanical Performance of Tilt-wing UAVs

to calculate the component’s local inflow condi-tion and the forces and moments acting on theCoG based on the component’s local forces andmoments [14]. Transformations are dependenton relative position between CoG and componentonly.

4.1 Propeller Disc and Slipstream

The propeller control input for each componentis static thrust, as the effective thrust is depen-dent on the local inflow condition, which is cal-culated for each element individually. Actuatormodels describe the correlation between throttleand static thrust of each engine. The propellerdiameter (ddisc) and the distance of the propellerdisc to the aerodynamic center of the wing arenecessary geometry information. In general thecalculations consists of the following steps:

• Reduce the static thrust due to the local in-flow condition• Calculate the slipstream of the propeller by

momentum theory• Reduce the induced velocity by use of con-

traction and continuity equation

The correlation between static ( f0) and effectivethrust ( feffective), dependent on the current axialinflow (u∞), is different for each propeller. There-fore it is based on experimental measurements.

0 5 10 15 20 25 30u , m/s

0

5

10

15

20

25

30

35

f, N

Measurement APC-12x10EFit APC-12x10EMeasurement APC-14x4WFit APC-14x4W

throttle

Fig. 4 Measurement data and quadratic fit resultsof two different APC-Propellers.

Fig. 4 shows wind tunnel data of the correla-tion between effective thrust and axial inflow for

two different propellers. For the CBM these dataare approximated by a quadratic function (withparameters a11 to a23), see Equ. 1. Thanks to thisrelationship, the effective thrust can be calculateddependent on the current axial inflow:

p1 = a13 · f02 +a12 · f0 +a11

p2 = a23 · f02 +a22 · f0 +a21 (1)

feffective = p1 ·u∞2 + p2 ·u∞ + f0

The average slipstream velocity (ucontr) of thepropeller is calculated by using momentum the-ory, see Equ. 2. The velocity is calculated forthe fully contracted slipstream in the far-field ofthe propeller. Therefore the induced part of thevelocities has to be corrected by transferring itto the aerodynamic center

(xchord =

14 lµ)

of thewing, by using the contraction factor (kd) of aslipstream (see Equ. 3 from [15]). The dynamicpressure of the wing, as sum of induced andfreestream velocity, can be used to calculate liftand drag, see Fig. 3.

ucontr =

√2 · feffective

ρ ·Adisc+u∞

2 (2)

kd = 1+xchord√

xchord2 +(

ddisc2

)2(3)

4.2 Wing Element

The geometry, the airfoil and the inflow condi-tion, in terms of dynamic pressure (q), AoA andReynolds-Number (Re), of the wing is known.These data are used to calculate the aerodynamicforces of the wing element. The polars of the co-efficients (cl,cd,cm) of each airfoil have to be pre-calculated, e.g. by use of xfoil [16]. Usually thepolars calculated by xfoil are valid in a maximumrange of AoA from−10◦ to +20◦. For VTOL air-craft this is insufficient, as the inflow might comefrom any direction. Therefore the polars are ex-trapolated by use of the Montgomerie approach[17]. In Fig. 5 the calculated and extrapolated po-lars for two airfoils are shown.

The calculation of the currently acting forcesand moments of the wing element consists of thefollowing steps:

5

MARTEN SCHÜTT , TOBIAS ISLAM , PHILIPP HARTMANN , DIETER MOORMAN

-150 -100 -50 0 50 100 150, °

-2

0

2

c l

-150 -100 -50 0 50 100 150, °

0

1

2

c d

-150 -100 -50 0 50 100 150,°

-1

0

1

c m

Clark-YNACA0012

Fig. 5 Extrapolated polars of Clark-Y andNACA0012 airfoil for Re = 450,000.

• Identify the current AoA and Re of thewing element• Look up the current corresponding aerody-

namic coefficients• Calculate the change of the aerodynamic

coefficients due to flap deflection• Reduce the lift coefficient for the finite

wing• Calculate the induced drag for the finite

wing• Calculate the pitching moment coefficient

for the finite wing• Calculate the forces and moments

As we only investigate the longitudinal mo-tion, all crossflow and side slip angles are ne-glected. This means the current AoA is sim-ply α = arctan

(w∞

uchord

)and the Reynolds num-

ber Re =lµ·√

w∞2+uchord

2

ν. As these aerodynamic

coefficients are valid for airfoils of infinite spanand no flap deflection, they have to be modifiedto finite wing coefficients (cL,cD,cM). These co-efficients are finally applied to calculate the cur-rent lift, drag and pitching moment. In combina-tion with the thrust these are transfered into theelement coordinate system to obtain the actingforces and moments for each element.

4.2.1 Flap Deflection

The effect of flap deflection is represented by ma-nipulation of the aerodynamic coefficients of theairfoil. Thereby change of lift, drag and momentis taken into account. Equ. (4) and (5) are derivedfrom Datcom [18]. This approach is suitable forplain flaps ([18]), which are commonly used inUAVs as ailerons and elevator.

δcl

δκ= χd1 ·χd2 ·ηd · cos(α) (4)

δcm

δκ=

δcl

δκ·0.25 · (λ−1) · cos(α) (5)

The empirical factors χd1 and χd1 are approxi-mated linear and quadratic dependent on the flapto chord ratio λ.

χd1 =−5.56 ·λ2 +11.39 ·λ+1.54 (6)

χd2 = 0.36 ·λ+0.36 (7)

The non-linear effect of greater deflection angles(> 12◦) is represented by the factor ηd , quadraticdependent on the deflection angle κ in radiant.

ηd =

{1 for κ≤ 12◦

0.822 ·κ2−1.73 ·κ+1.35 for κ > 12◦

(8)The effect of drag due to flap deflection is par-tial represented by increasing the induced drag,as mentioned in the next section. AdditionallyEqu. 9 was derived from empirical analysis usingxfoil, as in literature the impact of flap deflectionon drag is not discussed. Observed nonlinear ef-fects for high AoA were implemented for all co-efficient by trigonometric functions of AoA anddeflection angle.

δcd = 0.33 ·δκ2 +0.35 · sin(α) · tan(δκ) (9)

4.2.2 Lift of the Wing Element

The finite wing causes loss in lift compared to anairfoil of infinite span. Equ. 10, from [19], de-scribes the reduction of the linear slope of the liftcoefficient from an airfoil to a finite wing depen-dent on the aspect ratio (AR). This equation is ap-plied for the entire AoA and not only for the lin-ear region of the polar. The AR, applied for lift re-duction in Equ. (10), represents the entire wing,

6

Scalable Design Approach to Analyze Flight Mechanical Performance of Tilt-wing UAVs

independent of the dimensions of the single com-ponents. The tailplane and fin are treated equallyas wings, as long as it is a fuselage mounted con-figuration. The analyzed aircraft features a T-tailand the geometric AR of the fin is therefore mod-ified by ARcorr = 1.9 ·ARgeo, from [18].

cLα = clα ·AR√

AR2 +4+2(10)

4.2.3 Drag of the Wing Element

The drag of a wing element on the one handconsists of the friction-drag and the form-drag,which is already considered in the drag polarof xfloil. On the other hand, the generation oflift produces a downwash, which induces anotherform of drag, that has to be added. The rela-tionship of lift and induced drag is quadratic, seeEqu. 11 ([19]).

cD = cd +cL

2

π ·AR · e(11)

The Oswald-Factor e is calculated by semi-empiric Equ. 12 ([20]).

e =2

2−AR+√

4+AR2(12)

Additional aircraft drag like fuselage, nacelle orinterference drag is summed up into a drag-plate,an element without flap or engine, see Fig. 1.This element is of a certain size, to obtain a suit-able drag force.

4.2.4 Moment of the Wing Element

The pitching moment coefficient calculated byxfoil is referred to the

(14 lµ)

line. The coeffi-cient describes the shift of the center of pressure(CoP) on the airfoil in combination with the de-or increase of the lift coefficient. The variationof the lever arm and force influences the momentregarding the

(14 lµ)

line for different AoA. As thereference point of the component itself is on the(1

4 lµ)

line, no transformation is necessary. Mont-gomery gives an equation to calculate the CoP inreference to the leading edge (LE) for an airfoil,

-21

-1

0

2

i,h /

i

1

ze,t/s

10.5

ye,t/s

2

0-1

0 -2

Fig. 6 Change of AoA of the tailpane due todownwash as function of position δye,t/s andδze,t/s, extrapolated from [21].

see Equ. 13 ([17]).

arm =cm|( 1

4 lµ)

−cl · cos(α)− cd · sin(α)+0.25 (13)

For any inflow condition, in combination with theairfoil polars, the location of the CoP (arm) canbe calculated. The CoP is assumed identicallyfor the finite wing. By Equ. 14 the pitching mo-ment coefficient (cM) of the wing can be calcu-lated thanks to the previously determined coeffi-cients (cL,cD).

cM|( 14 lµ) = (−cL · cos(α)− cD · sin(α))·(arm−0.25)

(14)

4.3 Wing Downwash

The impact of the wing on the empennage ismainly a change in AoA of the tailplane due todownwash (wi) of the wing [19]. According to[14] this effect is essential, as it influences thepitching moment and changes significantly fordifferent tilt-angles of the wing. The downwashis created by the vortex system of the wing. Themagnitude of the effect is dependent on the lift ofthe wing, the span of the tailplane and the relativeposition between wing and tailplane.

Glauert [21] describes a theoretical down-wash distribution for the three dimensionalfarfield of the wing. Dependent on the location

7

MARTEN SCHÜTT , TOBIAS ISLAM , PHILIPP HARTMANN , DIETER MOORMAN

u∞

zt

xt

trailing edge

α

Fig. 7 Geometric relation of wing, tailplane andvortex system

of the tailplane, the relation(

αi,tailαi

)of the in-

duced AoA of the wing (αi) and the downwashAoA at the loaction of the tailpane

(αi,tail

)can

be calculated from these data. The relative posi-tion is given in terms of the half span s. The dis-tribution was inter- and extrapolated to use it forthe geometric correlations of tilt-wing UAV, seeFig. 6. The downwash distribution was calculatedfor a wing with an elliptical lift distribution, butthey are also applicable for wings with differentlift distributions, in particular trapezoidal wings[19]. In Fig. 7 the geometric relations of wing,vortex system and tailplane are described. Theinduced AoA of the wing

(αi =

cLwingπ·AR

)is derived

from the lift coefficient of the entire wing, includ-ing propeller induced lift,

(cLwing =

2·Lwingρ·u∞·Awing

).

The downwash (winduced =αi,tail ·uwing , for smallαi,tail) is finally added to the undisturbed inflowof the tailplane. This way the inflow condition ofall components located behind the wing can becalculated.

5 Validation

Some of the analytical and semi-empirical equa-tions could be validated by wind tunnel measure-ments. Wings of different geometries and withdifferent airfoils were analyzed for various in-flow conditions. Additionally the effect of flapdeflection and propeller interaction on the aero-dynamic coefficients are presented. All measure-

ments were performed in a Göttingen-type windtunnel using a 6-components-gauge to measureforces and moments. The parameters of the ana-lyzed wings are varying in geometry, airfoil andinflow condition, as the measurements were notperformed exclusively to validate this approach.Wings corresponding to the measurement datawere modeled by the CBM for the given inflowcondition.

5.1 Wind Tunnel Measurements

Two clean rectangular wings with an AR of4.2 are analyzed featuring two different airfoils,Clark-Y and NACA0012. Both were analyzed fora Re of 402,000 including AoA from −3◦ to 15◦.

-5 0 5 10 15, °

0

0.5

1

c L

-5 0 5 10 15, °

0

0.05

0.1

0.15

c D

Clark-Y - CBMClark-Y - Wind tunnelNaca0012 - CBMNaca0012 - Wind tunnel

-5 0 5 10 15, °

-0.15

-0.1

-0.05

0

c M

Fig. 8 Comparison of wind tunnel measure-ments and simulation of a rectangular wing ofAR = 4.2 with Clark-Y and NACA0012 airfoil forRe = 402,000

In Fig. 8 the aerodynamic coefficients of bothwings are presented. The wind tunnel measure-ment shows a linear relation of cL and α for bothwings up to 15◦. The CBM reduces the lift polar

8

Scalable Design Approach to Analyze Flight Mechanical Performance of Tilt-wing UAVs

of the airfoil which results in an overestimationof cl between AoA of 0◦ and 10◦. For higherAoA the slope of cL of the CBM decreases, rep-resenting stall. This behavior can not be foundfor neither of both wings in the wind tunnel mea-surements. As stall is expected to occur for thewind tunnel for slightly higher AoA, it is not as-sumed that the error will increase by much forhigher AoA.

The zero lift drag is underestimated by theCBM for both wings likely. For higher AoA thetotal drag coefficient cD increases more than thewind tunnel measurement shows, which resultsin an maximum error by overestimation for bothwings at AoA of around 9◦.

The wind tunnel measurements show a ap-proximately constant cM for both wings for theentire AoA range. The CBM calculates a con-stant cM up to 10◦, the coefficient of NACA0012airfoils fits the measurement very well. Thewrong stall characteristics of the CBM increasesthe error in moment for both wings for higherAoA.

Another wing of the airfoil NACA0012,which is equipped with a trailing edge flap of35% chord, was analyzed in the wind tunnel forRe of 100,000. The AR is 5.2 and the flap wasdeflected±20◦ and±10◦. In Fig. 9 the change ofthe aerodynamic coefficients due to flap deflec-tion is presented for AoA from 0◦ to 90◦. Theeffect in change of lift is very constant for AoAfrom 0◦ to 20◦ and decreases for higher AoA.This is estimated by the CBM very well, exceptfor the stall region around α = 23◦. The effectin drag shows greater deviations to the measure-ment, again especially for AoA in the stall region.Nevertheless, asymmetric effects between posi-tive and negative flap deflections are representedby the CBM. The constant change in moment(except for the stall region) for the entire rangeof AoA, is estimated by the CBM with small de-viations.

A wing with an AR of 5.2 and a NACA0012airfoil, which was equipped with a propeller infront of it, was analyzed in the wind tunnel. Thepropeller is an APC12x10E, which was mounted6 cm in front of the LE of the wing, producing

0 10 20 30 40 50 60 70 80 90, °

-0.5

0

0.5

cL

0 10 20 30 40 50 60 70 80 90, °

-0.1

0

0.1

0.2

0.3

cD

Wind tunnelCBM

0 10 20 30 40 50 60 70 80 90, °

-0.02

-0.01

0

0.01

0.02

cM

flap deflection

flap deflection

flap deflection

Fig. 9 Effect of flap deflection(−20◦,−10◦,+10◦,+20◦) for a rectangularwing of AR = 5.2 with NACA0012 airfoil forλ = 0.35 for Re = 100,000

5N and 10N static thrust. The freestream Re was67,000, not including the downwash of the pro-peller. In Fig. 10 the change of the aerodynamiccoefficients due to thrust settings is presented.The effect in change of lift is overestimated bythe CBM for both thrust settings for AoA up to≈ 27◦. The change in drag is negative and de-creasing for higher AoA, as the thrust acts in theopposite direction of the drag for an AoA of 0.The trend of the change in cD is estimated by theCBM very well. The expected change in momentdue to thrust is very low, as the engine is mountedcoaxial to the chord of the symmetric wing. TheCBM estimates similar, very low change of pitch-ing moment.

5.2 Evaluation

The CBM estimated the trend for all coefficientsof the entire analyzed AoA range. The most ob-

9

MARTEN SCHÜTT , TOBIAS ISLAM , PHILIPP HARTMANN , DIETER MOORMAN

0 5 10 15 20 25 30 35 40 45, °

-0.5

0

0.5

1

1.5

2

2.5

cL

0 5 10 15 20 25 30 35 40 45, °

-3

-2

-1

0

cD

0 5 10 15 20 25 30 35 40 45, °

-0.05

0

0.05

cM

Wind tunnelCBM

thrust

thrust

Fig. 10 Impact of propeller downwash (en-gine with ddisc = 0.3m producing 5N and 10Nthrust) for a rectangular wing of AR = 5.2 withNACA0012 airfoil for freestream Re = 67,000

vious deviations occur in the stall region, whichhave to be analyzed and modeled in greater de-tail. Also the zero lift drag needs to be increasedfrom airfoil to finite wing. Significant differencesbetween airfoils as well as due to flap deflec-tion could be observed and represent the mea-surements very well. The lift due to thrust showssignificant deviations, again in the area of stall.

Comparing the effects of deviations in pitch-ing moment coefficient to the occurring momentsof the entire aircraft, shows only minor effects inthe equilibrium of moments.

The applied general airfoil polars and equa-tions for finite wings are not direct applicable towings off small UAVs and low Re. Nevertheless,the flight mechanics for initial aircraft and con-troller design are represented adequate.

The effect of wing downwash on the empen-nage could not be validated in detail, but the oc-

curring AoA are physically plausible.

6 Flight Mechanical Analysis

In the following an exemplary application of theCBM is presented. For a tilt-wing aircraft withgiven requirements, the procedure to design asuitable wing is presented. Despite the simpleapproach to calculate drag and lift, significantdifferences between airfoils and wing geometrycan be identified. The aircraft requirements likecruise-speed, mass, engine and empennage aredetermined in the following section, additional toboundaries for the variable wing parameters, likespan, airfoils chord-width. Each set of require-ments and parameters defines an unique aircraftwith different geometry, lever arms and wing. Allaircraft have to be analyzed for their stationarycruise flight but additionally for their flight me-chanics in the entire flight envelope.

6.1 Requirements

The following parameters state aircraft require-ments, which need to be derived from the mis-sion. The maximum takeoff mass and the designcruise speed are most relevant for the design ofthe wing. Next to cruise flight, the location ofCoG and engines are very much dependent onflight mechanics during hover. In this case theempennage, in terms of lever arm, airfoil andsize, is designed based on flight mechanical con-siderations, which are not part of this paper. Thefixed parameters defining the aircraft are:

• ucruise: Design cruise speed, most relevantfor wing design.• Total mass: The entire aircraft is summed

up into a point mass, as the elements areweightless.• CoG: In distance behind wing LE, may be

dependent on the current tilt angle.• Moment of inertia: Based on an estimated

mass distribution in reference to the CoG.• D f uselage: Drag plate of a size to compen-

sate for all interference/parasitic drag.

The parameters characterizing a trapezoidalwing are summed up in Tab. 1. The boundaries

10

Scalable Design Approach to Analyze Flight Mechanical Performance of Tilt-wing UAVs

were set by aerodynamic considerations. The tipchord must not fall below 0.1 m, maintaining aminimum Re, as polars for very low Re are hardto derive. Speed and size of the aircraft lead toairfoils for low Re performance, which were cho-sen from UIUC database [22].

Table 1 Parameters defining the Wing ComponentsParameter min maxspan 1.0 m 2.0 mcroot 0.2 m 0.4 mctip 0.1 m 0.3 m

Göttingen GOE-398, Clark-Yairfoil Selig S-2027, Eppler E434

Wortmann FX 84-W-140

6.2 Flight Mechanics

The flight mechanics of all configurations haveto be analyzed for the flight envelope from hoverto cruise speed. Trimmed flight states should besatisfied by all configurations due to the numberof control variables. The state variables of thelongitudinal motion are:

• position: x,z• airspeed: u,w• attitude angle: θ

• angular velocity: q

In consideration of trimmed flight states, the an-gular velocity q is zero and the positions x andz are irrelevant. Thanks to the number of con-trol variables, the three remaining state variables(speeds u, w, and pitch angle θ) can be chosenfreely within the flight envelope, while remain-ing the equilibrium of forces and moments.

6.3 Application

For each of the possible configurations, a modelof the aircraft was set up using the CBM. Theconfigurations were then trimmed for stationaryflight for the entire flight envelope. A nonlin-ear numerical method2 is used to minimize theremaining horizontal and vertical forces (X and

2The Mathworks Matlab - fminsearch

Z) and the pitch moment (M), by actuator deflec-tion and thrust setting. This states a multidimen-sional, nonlinear problem.

6.3.1 Wing Design

The analysis for an optimal wing focuses on levelcruiseflight (u = ucruise and w = 0) and a horizon-tal fuselage (θ = 0). The chosen actuators to trimthe aircraft during cruise are the tilt angle of thewing and tailplane and the thrust of the tail en-gine. The effect of the main engines is neglectedas the propellers are folded away during cruiseflight. This way the nonlinear problem reveals aunique solution.

Each trimmed configuration creates the sameamount of lift-force, but wing and tailplane in-duce different amount of drag-force. There-fore, the configurations equipped with a differentwing, can be ranked by lift-to-drag ratio. The lift-to-drag ratio (Equ. 15) also takes fixed drag fromthe fuselage etc. and drag due to trim defectionsto achieve trimmed flight state into account.

γac =m ·g

Dwing +Dempennage +Dfuselage(15)

6.3.2 Flight Envelope

For each configuration all combinations of hor-izontal and vertical speed of the flight envelopehave to be analyzed. Again, the fuselage shouldbe horizontal, but the number of actuators to trimthe aircraft during hover and transition to cruiseflight increases to four: Tilt angle of wing (σ1)and tailplane (σ2), thrust of main ( fmain) and tail( faux) engine. This way the multidimensional,nonlinear problem doesn’t reveal a unique solu-tion but requires additional boundaries or weigh-ing functions. A possibility to compare all differ-ent solutions are the following properties:

• Max. Acceleration: Assuming limitedcontrol speed of all control variables, thegradient of corresponding trim control de-flections between two flight states limitsthe flight state transition velocity.• Controllability: Each control variable has

a limited deflection. The trim deflection

11

MARTEN SCHÜTT , TOBIAS ISLAM , PHILIPP HARTMANN , DIETER MOORMAN

reduces the remaining deflection availablefor steering.• Robustness: Each set of trim control de-

flection is differently sensitive to errors intrim-state (velocity, attitude), resulting inunintended accelerations.

A possible weight function, making these proper-ties comparable to analyze the flight mechanicalperformance, is dependent on the mission. Anoptimization concerning flight mechanics mayreduce the cruise flight performance by far. Asthe CBM is designed to mimic the flight mechan-ics of tilt-wing UAVs, it is a powerful tool to findthe optimal compromise.

7 Conclusion and Future Work

This analysis aims to explore the general applica-bility of the CBM for preliminary design of tilt-wing UAVs. A possibility to model each compo-nent of a tilt-wing aircraft in terms of aerodynam-ics and flight mechanics was presented. A levelof detail was proposed, keeping the experimentalor empirical data to a minimum, while the possi-bility to add detailed information is provided.

The setup of the method, especially the de-composition of the aircraft into components ofidentical structure was described. The compo-nents are able to model all essential parts of a tilt-wing aircraft and can consist of a propulsion unit,a wing element and a flap. The equations to cal-culate the forces and moments of all componentsof the aircraft were presented. While deriving therelations, a focus was set on the description of theeffect of the propeller slipstream and flap deflec-tion.

A validation by wind tunnel measurementsshows that effects of different airfoil, flap de-flection and thrust setting are represented by theCBM. Despite deviations in zero-lift-drag andstall characteristics, all significant effects of theflight mechanics are depict for initial design.

An application to analyze the flight mechan-ics of tilt-wing aircraft was presented. Based onan example configuration, the necessary require-ments of the method were discussed. Finally a

procedure to interpret the results of the methodfor wing design and trimmed flight states couldbe described.

While existing deviations of stall characteris-tics are state of current improvements, a valida-tion of the complete model, including the effectof wing downwash on the empennage, is still tobe made.

Contact

M. Schütt: schuett@fsd.rwth-aachen.de

References

[1] https://www.germandrones.com/[2] Fredericks, W. J.; Moore, M. D.; Busan, R.

C.: Benefits of hybrid-electric propulsion toachieve 4x increase in cruise efficiency for aVTOL aircraft. AIAA International Powered LiftConference, AVIATION Forum, American Insti-tute of Aeronautics and Astronautics, (2013).https://doi.org/10.2514/6.2013-4324

[3] Stoll, A.M; et al.: Conceptual Design of theJoby S2 Electric VTOL PAV. 14th AIAA Avi-ation Technology, Integration, and OperationsConference, Atlanta, US-GA, (2014).

[4] https://www.quantum-systems.com[5] Holsten, J.; Ostermann, T.; Moormann, D.: De-

sign and wind tunnel tests of a tiltwing UAV.CEAS Aeronautical Journal Jg. 2, Nr. 1/4, S. 69-79, (2011).

[6] Schütt, M.; Hartmann, P.; Holsten, J.; Moor-mann, D.: Mission Control Concept for ParcelDelivery Operations based on a Tiltwing Air-craft System. 4th CEAS Specialist Conferenceon Guidance, Navigation & Control, Warsaw,Poland, (2017).

[7] Haimerl, M.; Binz, F.; Engels, S.; Moor-mann, D.: Entwicklung einer modifizierten,instationären Wirbelgittermethode für einesenkrechtstartfähige Flugzeugkonfiguration.Deutscher Luft- und Raumfahrtkongress,Deutsche Gesellschaft für Luft- und Raumfahrt- Lilienthal-Oberth e.V, (2017).

[8] https://www.altiuas.com/[9] Rothhaar, P. M. et al.: NASA Langley Dis-

tributed Propulsion VTOL Tilt-Wing Aircraft

12

Scalable Design Approach to Analyze Flight Mechanical Performance of Tilt-wing UAVs

Testing, Modeling, Simulation, Control, andFlight Test Development. 14th AIAA AviationTechnology, Integration, and Operations Con-ference, Atlanta, US-GA, (2014).

[10] Schütt, M.; Hartmann, P.; Moormann, D.:Fullscale Windtunnel Investigation of Actua-tor Effectiveness during Stationary Flight withinthe Entire Flight Envelope of a Tilt-wing MAV.International Micro Air Vehicle Conference andCompetition, Delft University of Technology,(2014).

[11] Selig, M. S.: Modeling Full Envelope Aero-dynamics of Small UAVs in Real Time.AIAA Atmospheric Flight Mechanics Con-ference, AIAA Paper 2010-7635, (2010).https://doi.org/10.2514/6.2010-7635

[12] Tobing, S.; Go, T. H.; Vasilescu R.: ImprovedComponent Buildup Method for Fast Predictionof the Aerodynamic Performances of a VerticalTakeoff and Landing Micro Air Vehicle. Com-putational Fluid Dynamics, pp 209-214, (2008).

[13] Di Francesco, G.; Mattei, M.: Modeling and in-cremental nonlinear dynamic inversion controlof a novel unmanned tiltrotor. Journal of Air-craft, Jg. 53, Nr. 1, S. 73-86, (2015).

[14] Hartmann, P.: Predictive Flight Path Con-trol for Tilt-Wing Aircraft. Dissertation, RWTHAachen, (2017).

[15] McCormick, B. W.: Aerodynamics, Aeronau-tics, and Flight Mechanics. Wiley, New York,(1994).

[16] Drela M.: XFOIL: An Analysis and DesignSystem for Low Reynolds Number Airfoils.Low Reynolds Number Aerodynamics. LectureNotes in Engineering, Springer, Berlin Heidel-berg, (1989).

[17] Montgomerie, B.: Methods for root effects, tipeffects and extending the angle of attack rangeto +/-180, with application to aerodynamicsfor blades on wind turbines and propellers.Swedish Defence Research Agency, (2004).

[18] Hoak, D.E.; Finck, R.D.: USAF Stability andControl Datcom. Flight Control Division, AirForce Flight Dynamics Laboratory, Wright-Patternson Air Force Base, Ohio, (1978).

[19] Schlichting, H.; Truckenbrot, E.: Aerodynamikdes Flugzeuges. Springer, Berlin Heidelberg,(2001).

[20] Brandt, S.A.; et al.: Introduction to Aeronau-tics: A Design Perspective. AIAA EducationSeries, AIAA Inc., Washington, (2004).

[21] Glauert, H.: The Elements of Aerofoil andAirscrew Theory. Cambridge Science Classics,Cambridge University Press, Cambridge(1983).

[22] Williamson, G.A.; et al.: Summary of Low-Speed Airfoil Data Vol. 5. Department ofAerospace Engineering, University of Illinois atUrbana-Champaign, Urbana, (2012).

Copyright Statement

The authors confirm that they, and/or their companyor organization, hold copyright on all of the origi-nal material included in this paper. The authors alsoconfirm that they have obtained permission, from thecopyright holder of any third party material includedin this paper, to publish it as part of their paper. Theauthors confirm that they give permission, or have ob-tained permission from the copyright holder of thispaper, for the publication and distribution of this pa-per as part of the ICAS proceedings or as individualoff-prints from the proceedings.

13