structural sizing of a rotorcraft fuselage using … sizing of a...an essential part of the design...

STRUCTURAL SIZING OF A ROTORCRAFT FUSELAGEUSING AN INTEGRATED DESIGN APPROACH

Dominik B. Schwinn∗, Peter Weiand∗∗, Matthias Schmid∗∗∗ and Michel Buchwald∗∗German Aerospace Center (DLR)∗Institute of Structures and Design,∗∗Institute of Flight Systems,

∗∗∗Institute of Aerodynamics and Flow Technology

Keywords: rotorcraft integrated design; multidisciplinary design optimization; airframe structuralsizing; weight estimation

Abstract

Many years the primary design objective of newhelicopters was the design of the main rotor(s).Within the last couple of years, this approachhas changed into an assessment of all helicoptercomponents as an overall system, thus turningrotorcraft design into a highly interdisciplinaryprocess. Aerodynamics, flight mechanics, struc-tural evaluation, etc. strongly affect each otherand these mutual influences are taken into ac-count from the very first design stages. However,weight prediction in early stages still representsan essential part of the design process as it deter-mines the basic properties of the rotorcraft. Dueto its function to carry crew and payload as wellas to serve as the central mounting for all compo-nents, the fuselage represents a major part of therotorcraft, therefore the structural design of thefuselage airframe constitutes a significant factorof the rotorcraft design at preliminary level.

1 Introduction

Rotorcraft design is a highly challenging disci-pline within the aeronautical sciences. Like air-craft design it is, in general, classified into threeconsecutive phases: The conceptual, the prelimi-nary, and the detailed design phase.Conceptual design is mainly concerned with theouter configuration of the helicopter, i.e. the

aerodynamic shape (loft), rotor characteristics,flight performance analysis, etc. using fast an-alytical simplified methods or estimations basedon statistics, if available.The subsequent preliminary design phase useshigher fidelity tools to increase the detail level.During this stage a basic internal arrangement iselaborated, i.e. the distribution of primary struc-ture within the previously determined loft. Flightand ground load cases are evaluated, thus en-abling the designers to derive the main load pathsand to obtain major loads and stresses within thestructure.Ultimately, the detailed design phase is con-cerned with detailed local solutions, such asjoints and fittings. This phase is highly influ-enced by producibility and maintainability as-pects. However, despite this stage being con-ducted immediately before initializing the man-ufacturing process, most of the costs are alreadydetermined during the conceptual and prelimi-nary design stage [17], thus highlighting the im-portance of early design phases.For many years the design of a new helicopterwas mainly synonymous to the design of the mainrotor(s). However, in recent years the design ob-jective has shifted into an overall examination ofall helicopter components as a complete, globalsystem [24]. As an example, the aerodynamics ofthe fuselage has received considerable attentionlately and has become an important part of the de-

1

D.B. SCHWINN, P. WEIAND, M. SCHMID, M. BUCHWALD

sign process [10]. Since aerodynamics, flight me-chanics, structures, etc. work together, helicopterdesign has become a highly interdisciplinary pro-cess.In order to assess novel configurations addressingtypical rotorcraft limitations, e.g. cruise speed,the German Aerospace Center (DLR) started in2010 with the set-up of an automated, inte-grated, parametric, and multidisciplinary processchain for early helicopter design. During theprojects RIDE (Rotorcraft Integrated Design andEvaluation) and EDEN (Evaluation and Designof Novel Rotorcraft Concepts) the data formatCPACS (Common Parametric Aircraft Configu-ration Schema, [12]) was adapted to match para-metric rotorcraft description. The network basedsimulation environment RCE (Remote Compo-nent Environment, [22]) was used to generate andset up workflows to design generic rotorcraft con-cepts according to user specified top level air-craft requirements (TLARs), typically consist-ing of payload, cabin volume, range and cruisespeed.During the conceptual branch of the aforemen-tioned process chain the geometry of the de-sired rotorcraft is generated to fulfill the specifiedTLARs. At this stage component masses are esti-mated using statistical methods eventually result-ing in a converging maximum take-off mass. Inthe subsequent preliminary design stage, the pri-mary structure is distributed within the fuselageloft using knowledge based design criteria and astructural finite element (FE) model of the fuse-lage airframe is automatically generated. Spec-ified load cases are evaluated and a sizing pro-cess using FSD (fully stressed design) principlesis initiated to find a minimum material thicknessdistribution considering strength and stability cri-teria, thus resulting in an updated maximum take-off mass.This paper introduces the overall process chainas well as the data format and simulation frame-work. The presented work focuses on the struc-tural aspect during the preliminary design stageof the DLR rotorcraft design process. The toolsand approaches for the generation of the FEmodel, the calculation of the external loads, as

well as the sizing methods are introduced and ex-plained in detail. Considering novel compoundconfigurations an outlook on structural modelingaspects is given. Finally, the paper discusses de-velopment steps for future applications and en-hancements.

2 Data Format and Process Handling

An important aspect in the set-up of an inte-grated, automated tool chain is the flawless con-nection and communication of all contributingcomputational tools. The integrated tools need tointeract on two levels, namely data transfer andsoftware processes which are introduced subse-quently.

2.1 CPACS

CPACS is used for DLR rotorcraft design activ-ities as a common data model. It is a key com-ponent for the communication and data exchangebetween the individual computational tools andusers. Its benefits are its hierarchical structure,easy access and readability. Since it is actingas the central component in the design chain itserves as an interface for all integrated tools nt ,thus significantly reducing the required interfacesni from

ni = nt(nt−1) (1)

of a traditional approach to

ni = 2nt (2)

as can be seen in Fig. 1 (for nt ≥ 3), where theblue curve shows the number of interfaces usinga traditional approach (as depicted on the left)while the red curve shows the centralized CPACSapproach (as illustrated in the centered figure).

tools

CPACS

traditional

tools

interfaces

interfaces

Fig. 1 Interface reduction due to CPACS

2

STRUCTURAL SIZING OF A ROTORCRAFT FUSELAGE USING AN INTEGRATED DESIGNAPPROACH

2.2 RCE

The design tools do not only exchange data,they also need to interact within a design pro-cess. Since different institutions at different siteswith different expertise are integrated in the pre-sented design process, a distributed computationapproach was chosen. For this purpose, the in-house developed software RCE is used. The cor-responding specialists develop and maintain theirsoftware which is installed on locally separatedservers. The design tools are published to autho-rized partners who can execute the programs butcannot access the source code of the design tools,i.e. the knowledge of each discipline stays at thedeveloping institute. Data is transferred via an in-ternal network, i.e. a CPACS file is sent from thelocal user to the server where the desired tool isstored. The tool reads its input from this CPACSfile and stores its computed output in an updatedCPACS file which is then transferred back to theuser.

3 Rotorcraft Design Process

The presented design process starts from scratchand comprises the conceptual and parts of thepreliminary stage. Typically, the minimumTLARs cover payload, range, cruise speed andthe rotor configuration. Currently, three types ofrotor configurations are supported for automatedinitialization: standard, coaxial and tandem (seeFig. 2, from left to right).

Fig. 2 Supported rotorcraft configurations

Progress in the design process leads to an in-crease in detail level since more input infor-mation becomes available by increasing output.Therefore, design tools can generally be classi-fied in four levels, ranging from a coarse level 0(L0) to a high fidelity level 3 (L3):

• L0 tools mostly use empirical methodswith very simple physical assumptions.They provide much output with only lim-ited input. The objective is the generationof a first data set.

• L1 tools have a better physical modelingbut are still fast enough to perform iterativeprocedures. Primary sizing is performedon this level. The objective is to completethe data set in order to create the first flightmechanics model.

• L2 tools feature a very good physical mod-eling but as a resulting disadvantage theyrequire much computational investment,i.e. hardware and time. These tools typi-cally comprise the preliminary design andexpand the data set.

• L3 tools are the most complex design tools.They have the highest time demand andtheir pre- and post-processing cannot beperformed automatically.

Figure 3 illustrates the global design process asapplied in the presented work. Evaluating theTLARs leads to the generation of an initial modelby using L0 tools which rely on a statistical database of about 160 existing rotorcraft.

This initial model comprises a first estimationof the mass fractions resulting in a first estimationof the maximum take-off mass mmto according to

mmto = moem +mpay +m f uel (3)

where

moem operating empty mass (see Eq. 6)mpay payloadm f uel fuel mass

Payload is generally specified within the TLARswhile the fuel mass depends on the chosen en-gine(s), the drag caused by the rotorcraft sur-faces, and the desired range. This approachshows that for this initial estimation existing con-figurations are required. Novel configurationscan only be roughly estimated by comparison toalready existing similar rotorcraft.

3


Top Level Aircraft Requirements

Initial Sizing(Level 0)

Final Output

Sizing Loop (Level 1)

Sizing /Resizing?

Yes

Assessment By UserHigher Fidelity Analysis &

Design(Level 2)

No

Phases:conceptualpreliminary

Fig. 3 Flowchart of the virtual design approach [26]

3.1 Conceptual Design Stage

After an initial configuration has been created,the sizing loop of Fig. 3 is initialized at the con-ceptual stage using L1 tools, further increasingthe level of detail. Figure 4 overviews this pro-cess which is iterated until convergence of thetake-off mass mmto is achieved.The calculation of the rotor characteristics is sub-stantial for the design. A knowledge based proce-dure [7] is used to optimize radius, chord length,angular velocity respectively tip speed using a se-ries of characteristic rotor parameters (aspect ra-tio, rotor solidity, blade loading, advance ratio,energy ratio, Lock number).Subsequently, an outer fuselage surface is gener-ated using a CATIA based approach [8]. This ap-proach splits the fuselage into several segmentswhich can individually be scaled, thus assem-bling the complete fuselage. Different generictemplates for each fuselage segment are de-posited to allow fuselage generation for each sup-plied configuration as specified in Fig. 2.After the outer shape has been defined, the aero-dynamic properties are calculated by a mod-ule based on the commercial software VSAERO

Sizing Loop(level 1) MMTO

convergence?

Design Rotor blade

planform optimization

Analysis

Estimation of operating empty

mass

Flight-performance & estimation of fuel

mass

Initial dataset (level 0) & updates (level 2)

Yes

Fuselage aerodynamics

Twist optimization

3D-surface generation

No

Fig. 4 Flowchart of the sizing loop [26]

[11] which uses the potential flow theory withan incompressible and inviscid approach. Com-pressibility corrections can be assessed by theintegrated Prandtl-Glauert or Karman-Tsien ap-proaches. Viscous effects as well as boundarylayer transition and separation can be evaluateddue to integration of an integral boundary layerformulation. Pressure drag is estimated based onthe calculated separation line. Currently, drag in-fluence from rotor hub, landing gears or attach-ments are not implemented in the drag calcula-tion since the integrated panel methods cannotresolve it in a reliable manner. The force polarsobtained by this step must be corrected to coverthe influence of skids and additional attachments.The polars of the stabilizers are calculated sepa-rately.An optimization of the blade twist can be per-formed in order to minimize the required powerfor the design flight condition. The trim calcula-tion for this procedure is conducted by the soft-ware HOST (Helicopter Overall Simulation Tool,[2]). A linear twist distribution is iterated in orderto reduce the required power for the consideredflight condition.Fuel mass is estimated in an iterative process,

4


again using HOST for the trim calculation. Forall flight segments, trim calculations are con-ducted at the beginning and at the end of everysegment obtaining the required power and con-sequently predicting the mean fuel flow and ac-tual range. Comparing actual and required rangeleads to a correction of the actual fuel until fuelmass converges and the requested range is met.Detailed information on the aforementionedmethods is given in [7, 26].At this point of the design process sufficient pa-rameters have been calculated to allow a first esti-mation of component masses. Statistical and em-pirical methods are used to break down mmto intogroup masses, e.g. the structural, power plant,systems and furnishings group. The implementedmethods are described in detail in ch. 4.

3.2 Preliminary Design Stage

The external configuration developed during theconceptual phase serves as starting point for thepreliminary design which mainly deals with thedetermination of the internal configuration andmore sophisticated aerodynamics, such as inter-actions and local flow problems. At this stage,specialists of the different disciplines design andanalyze their portion of the aircraft. Sophisti-cated methods are applied, typically comprisingL2 tools.Currently an FE based module for structural siz-ing of metallic fuselages according to specifiedstatic and quasi-static flight and ground loadcases is implemented in the presented designchain. This approach allows an enhanced eval-uation of the fuselage weight, especially whennovel designs are investigated which cannot becompared to already existing designs. A detaileddescription of the structural assessment is givenin ch. 5.

4 Mass estimation

Mass estimation constitutes an essential part ofthe design process since the mass determines andinfluences many design and performance param-eters. For instance does the weight determine

the required lift generated by the rotor for ver-tical flight. The required lift then determines thedimensions of the rotor blade and the rotationalvelocity which in turn determine respectively in-fluence the engine power, fuel consumption andthe gear box(es).For the mass estimation in the presented work,several estimation methods have been imple-mented, such as Beltramo [1], Layton [9], Palasis[14], Prouty [16], and Johnson [5].In order to profit from a high coupling grade ofgeometric and performance characteristics it wasdecided to use the methods presented by John-son for the component mass estimation duringconceptual design. These methods are based onthe U.S. Army Aeroflightdynamics Directorate(AFDD) models and feature technology factorsχi which allow individual scaling according todifferent technology standards. For each compo-nent mass mcomp consisting of n subcomponents(e.g. the fuselage component consists of addi-tional individual elements to account for crash-worthiness, alighting gear integration, tail and/orwing folding, marinization, etc.) it is

mcomp =n

∑i=1

χimi (4)

Validating the reference model (a standard con-figuration loosely based on an EC135) it was de-cided to follow the approach of Russell and Bas-set [18] to use one general technology factor forall components. Applying a technology factor ofχ = 0.7 resulted in a deviation of ∆mmto ≈ 3%between the calculated mmto and the referencedmmto,re f .By summation of the individual componentmasses the group masses can be obtained. Theempty mass mem is calculated by summation ofthe group masses

mem = mstruct +mprop +msys +m f e (5)

with the group masses

mstruct structural massmprop propulsion (power plant) system massmsys systems massm f e mass of furnishings and equipment

5


Adding the mass of the operators items moi tothe empty mass mem leads to the operating emptymass

moem = mem +moi (6)

Adding the payload and fuel mass to moem resultsin an updated take-off mass mmto, thus allowinga comparison to the initial mmto,0 which is usedas convergence criterion of the conceptual sizingloop.Figure 5 shows an overview of the implementedmass estimation methods applied to a genericmedium-sized utility rotorcraft in standard con-figuration, as displayed in the graph. Since aspecific engine was considered in the presentedmass estimation process, the engine mass is un-affected. The methods provided by Palasis arepartly those provided by Beltramo and Layton,therefore the estimated masses show the same re-sults for certain components. Prouty does notprovide an estimation method for the load han-dling system while Beltramo’s and respectivelyPalasis’ regression formula for this mass returnsa negative value, therefore it was automaticallyset to zero as no reliable value was available forthis system. In general, Johnson’s methods ap-pear to estimate the highest masses of all imple-mented methods (assuming χ = 1.0), for instancethe furnishings mass is heavier than calculatedwith the other approaches. An explanation isthat for certain components, Johnson offers massranges for medium- to heavy-weight rotorcraft,which are linearly interpolated by the presentedmodule. Since the integration of certain systemsis highly mission dependent, the user can pro-vide specific input masses (in case the componentmass is known or prescribed) for a more reliablemass estimation.Comprising the operators items mass moi = 180kg, a fuel mass m f uel = 500 kg and a payloadmpay = 800 kg leads to the resulting take-offmasses mmto as presented in Tab. 1. The pre-sented reference value mmto,re f ∗ is the take-offmass calculated by applying Johnson’s methodswith an overall technology factor χ= 0.7 for eachindividual component. The maximum deviationof the other methods is ∆m ≈ 5% indicating that

fuse

lage

main

roto

rs

engi

nes

avioni

cs

elec

trics

furn

ishin

gs

load

han

dlin

g sy

stem

alig

htin

g ge

ar

drive

syst

em

fuel sys

tem

fligh

t con

trols

inst

rum

ents

auxilia

ry p

ower

sys

tem

nace

lle

hydr

aulic

s

empe

nnag

e

tail ro

tor

airc

ondi

tion

& deicing

0

50

100

150

200

250

300

350

400

mass

[kg

]

Johnson (TF = 1.0)

Johnson (TF = 0.7)

Layton

Beltramo

Palasis

Prouty

Fig. 5 Comparison of the implemented mass es-timation methods

the chosen technology factor derived from thevalidated model marks a reliable value for thisrotorcraft class.

Method mem mmtommto

mmto,re f∗

Johnson (χ = 0.7) 1,612 3,092 1.0Johnson (χ = 1.0) 2,199 3,679 1.190Beltramo 1,746 3,226 1.043Palasis 1,764 3,244 1.049Layton 1,752 3,232 1.045Prouty 1,700 3,180 1.028

Table 1 Take-off masses [kg] (rounded values)

Figure 6 shows the composition of the emptymass calculated with the methods provided byJohnson featuring a technology factor of χ = 0.7for all relevant components. It can be seen thatthe fuselage mass constitutes the biggest share ofthe empty mass, thus highlighting its importancein the design process.

5 Airframe Structural Analysis

In order to obtain a more realistic fuselage massestimation, L2 tools are integrated in the pre-sented design process, designating the prelimi-nary design stage. At the end of the concep-tual design a consistent external configurationhas been derived (as shown in Fig, 7), i.e. thedesign space for the structural members of the

6


fuselage14.3%

engines

12.3%furnishings

11.8%

main rotors10.1%

avionics 7.9%

electrics7.9%

alighting gear

7.3%

drive system

7.1%

load handling system

3.9%fuel system

3.4% flight controls

3.2% instruments3.0% auxiliary power system

2.6%nacelle2.1% aircondition & deicinghydraulicsempennagetail rotor

Fig. 6 Composition of mem according to John-son’s methods (χi = 0.7)

fuselage, such as frames or stringers, is avail-able. In a first step at preliminary level, an al-gorithm distributes the primary structure accord-ing to knowledge based design rules, e.g. maxi-mum spacing between individual structural mem-bers or reinforcements around non-loadbearingcut-outs such as doors or cargo ramps [20]. Ad-ditionally, material parameters and (initial) geo-metric properties, such as panel thicknesses andprofile dimensions, are assigned to the airframe.This information has to be specified by the userin advance to the design process. Subsequently,selected steady (nontransient) flight and groundmaneuvers are evaluated and the resulting forcesand moments are stored as external loads in theCPACS file. Figure 7 schematically overviewsthe model preparation process. It shall be notedat this point, that the presented tool is currentlylimited to isotropic materials, e.g. metals.

5.1 Loads Calculation

To calculate the external forces acting on thehelicopter in static load conditions, the externalforces Fi and moments Mi around the three mu-tually perpendicular axes must be in equilibrium:

∑i

Fi = 0 (7a)

∑i

Mi = 0 (7b)

modelgeneration

structural members,materials,loads,boundary conditions

conceptual stage

Fig. 7 Model preparation

For level flight the resultant thrust load T is re-acted by the weight of the helicopter, the iner-tial loading due to horizontal acceleration and thedrag load caused by airspeed, as shown in Fig. 8.

T

nW F = TR

D + mx

Fig. 8 Helicopter scheme for load evaluation atlevel flight

Thrust can be split into a propulsive force Tx andthe lift L which in turn can be written due toEq. 7a as

Tx = D+mx (8a)

andL = nW (8b)

with

W weightn load factorD aerodynamic dragx longitudinal acceleration

7


Load factor limits are specified by EASA-CS1

27.337 for small rotorcraft (mmax ≤ 3175 kg(7000 lb) or less than nine passenger seats)and EASA-CS 29.337 for large rotorcraft. Forbanked turns (see Fig. 9) at constant altitude theload factor n is a function of the bank angle φ

n =1

cos(φ)(9)

as indicated in Fig. 10.

T

W

Φ

ΦFcf acp

Fig. 9 Banked turn at level flight

0 10 20 30 40 50 60 70

bank angle φ [°]

0.5

1.0

1.5

2.0

2.5

3.0

3.5

load f

act

or

n [

-]

Fig. 10 Banked turn: load factor vs. angle

The centripetal acceleration acp required to per-form a turn with the flight path at radius Rturn is

acp =v2

Rturn(10)

1Certification Specifications (CS) of the European Avi-ation and Space Administration (EASA)

where v denotes the airspeed. With the relationbetween weight W , mass m and acceleration gunder gravity

W = mg (11)

and Newton’s second axiom

F = ma (12)

the centrifugal force Fc f can thus be written as

Fc f =Wg

v2

Rturn(13)

Helicopter parasite drag is an important aspect ofperformance calculation since it establishes thepropulsive force and power requirement at highspeed [4]. It is commonly expressed in terms ofthe dynamic pressure q and the parasite drag areaf

D = q f (14)

withq =

12

ρv2 (15)

where ρ denotes the air density. Johnson [6] pro-vides two possibilities to estimate the parasitedrag area, either based on the maximum take-offmass m∗mto or alternatively based on the projectedarea of the rotors Arot . For the loads calculationthe weight based approach has been chosen:

f =Dq= k

(m∗mto1000

)2/3

(16)

with k varying from 9 (for old helicopters) to 2.5for current low-drag helicopters, based on histori-cal helicopters with m∗mto specified in pounds [lb]and f given in square feet [ft2]. Torque Q in-troduced from the main rotor into the fuselage iscalculated by the relation

∑i

Pi = ωQ (17)

where ω denotes the angular velocity of the mainrotor and ∑Pi is the total power of the main ro-tor which consists of several shares, e.g. inducedpower Pi for thrust, P0 to overcome blade pro-file drag, Pp to overcome parasite drag and climb

8


power Pc. The tail rotor force Ftr can then be cal-culated by

Q = Ftrlr (18)

with lr being the distance between the main rotorshaft and the tail rotor shaft.For certification all load cases that can be expe-rienced by the rotorcraft must be considered, i.e.a changing center of gravity (COG), for instancedue to fuel consumption, must be taken into ac-count.The rotorcraft reaction to gusts is less severe thancompared to a fixed-wing aircraft. However, loadfactors on rotorcraft due to gusts are not insignif-icant. Due to the trend towards higher flight ve-locities gust criteria become more important asload factors generally increase with airspeed.Due to the preliminary nature of the structuralanalysis, the aforementioned effects have not(yet) been taken into account.

5.2 Model generation

Subsequently, an FE model in GFEM (globalfinite element) quality is generated, as defaultone skin panel element comprises one stringer-frame-bay. Modeling is conducted in ANSYSusing APDL (ANSYS parametric design lan-guage). Engineering constants of the reinforcingprofiles, such as COG, section area, and momentsof inertia, are calculated by evaluating the pro-file dimensions and assigned to the correspond-ing beam elements. Frame webs and skin pan-els are discretized using elastic shell elements(ANSYS Shell 181) while stringers and frameflanges are modeled using beam elements (AN-SYS Beam 188). Component masses estimatedduring the conceptual design phase are modeledas single nodal masses being coupled to the struc-ture over a user-specified influencing region byRBE3 elements. The rotors are also discretizedas nodal masses where the external forces andmoments are applied. Figure 11 shows an ex-emplary FE model (for visual reasons only oneload/mass constraint is shown) with skin panelsand the reinforcing structure respectively. Cut-outs for non-loadbearing elements, e.g. doors,windshield, lookouts have been applied to the air-

frame using the stage modeling approach [20]which also allows a realistic termination of thestringers in the tail boom.

main rotor constraint

tb

xloc

yloc

Fig. 11 FE airframe model

5.3 Static analyses and sizing process

Static computations are conducted using thelinear-elastic ANSYS solver. Figure 12 shows astatic analysis of the hovering load case, using thegeneric rotorcraft model as introduced in Fig. 11.The fuselage structure is made of aluminum 2024(as specified in Tab. 2), with flat frames of dif-ferent heights and thicknesses of 1.4 ≤ ti ≤ 2.0[mm]. The stringers feature hat profiles (as dis-played in Fig. 11, corner) with tb = 1.5 [mm] forall sheets while the skin panels feature a thick-ness of t = 1.6 [mm].

van Mises stress [MPa]

0.2 140 280

Fig. 12 Static hover analysis

It is observable that in the hovering load case

9


the highest stresses arise around the cabin wherethe main rotor is mounted and the highest nodalmasses are located, and in the first frame of tailboom due to the reduction of stringers and the lat-eral force introduced by the tail rotor.

Young’s modulus E 67.7 [GPa]Density ρ 2,800 [kg/m3]Poisson’s ratio ν 0.248 [-]Yield strength 320 [MPa]

Table 2 Aluminum 2024 - material properties

The sizing tool is an APDL based module, orig-inally designed for aircraft wings [13], thenadapted to aircraft fuselages [19] and finally tohelicopter applications [21]. Strength evalua-tion is based on fully stressed design (FSD) prin-ciples. Local compressive and shear bucklingmethods as proposed by Bruhn [3] are imple-mented to guarantee sufficient safety against sta-bility failure.Equivalent stress σeq is computed for all ele-ments. The thickness of each shell element isthen reduced by the factor

r f =σeq,max.a

σeq(19)

where σeq,max.a denotes the maximum allowableequivalent stress, as specified by the material orstability limits. Therefore, the thickness can bereduced as long as the allowable stress limits arenot exceeded.For the stringers the sheet thicknesses tb,i ofall beam elements of one individual stringer arescaled equally by a common scale factor r (seeFig. 11), i.e.

t ′b,i = r · tb,i (20)

while the basic cross section is maintained.This process is repeated for each specified loadcase and for each element the maximum requiredskin respectively sheet thickness is stored andapplied to the corresponding panel respectivelyframe or stringer element. The resulting up-dated stiffness distribution is then recalculateduntil convergence is achieved. The final thick-nesses and section areas are stored back to the

Turn no. v r acp α n[km/h] [m] [g] [◦] [-]

01 200 500 0.63 32.2 1.1802 300 500 1.42 54.8 1.7303 400 750 1.68 59.2 1.9504 400 500 2.52 68.3 2.71

Table 3 Flight data for banked turns

CPACS file as well as the updated masses mstruct ,mem, moem and mmto.Exemplary sizing processes comprising the hov-ering state, cruise flight with a maximum velocityof vmax = 400 km/h and several banked turns, asspecified in Tab. 3, are shown in Fig. 13. Sizingscenario no. 7 considers a +2.5g pull-up maneu-ver instead of the banked turn no. 4.

It-01 It-02 It-03 It-04 It-05

Iterations [-]

0

50

100

150

200

250

300

350

400

Mass

[kg

]

Johnson(χ=0.7)

01 : hovering

02 : 01 + max. cruise

03 : 02 + turn01

04 : 02 + turn01-02

05 : 02 + turn01-03

06 : 02 + turn01-04

07 : 02 + turn01-03 + pullup

07': 07 + joints

07 : shells

07 : beams

Fig. 13 Sizing process with different load casescenarios

It can be observed that the weight grows with anincrease in considered load cases: The pure hov-ering state results in a fuselage mass of m f us ≈111 kg while the addition of the maximum cruiseflight load case slightly increases the fuselagemass to m f us ≈ 114 kg. The first banked turn in-creases the fuselage mass to m f us ≈ 115 kg. Thecentrifugal load caused by this turn is compara-bly small since the chosen turn radius is ratherlarge and the velocity rather slow. In contrast, thesecond turn scenario at high velocity increasesthe fuselage weight to m f us ≈ 123 kg due tothe higher speed which increases the load fac-tor. This effect can also be observed at turn no.

10


3: Even though the radius has been increased by∆r = 50% to rturn = 750 m the load factor in-creases due to the higher speed thus increasingthe fuselage weight. Turn no. 4 represents arather steep turn causing a load factor of aboutn ≈ 2.7. In load scenario no. 6 it can be seenthat steep turns increase fuselage weight morethan the addition of a +2.5g pull-up maneuver,as shown in scenario no. 7. However, it shallbe noted that this kind of maneuver is not rep-resentative for a transport, especially an emer-gency helicopter. Maneuver issues are of particu-lar importance for combat helicopters which maybe required to perform maneuvers consisting ofhigh load factors and pull-ups, steep turns androllovers.The bars represent the share of the beam elements(representing the stringers) and the shell elements(representing the frames and the skin panels) ofthe sizing process comprising load case scenariono. 7 (hovering, cruise, three banked turn scenar-ios and a pull-up maneuver). The initial thick-nesses of the shell elements were set to t = 5 mmto allow for a reliable convergence of the com-putations. The red triangle represents the sizedfuselage mass of scenario no. 7 with an addi-tional weight penalty due to joints, ranging from20 - 40 % of the fuselage mass.Figure 14 shows the results of the sizing scenariono. 7, the required thicknesses and the corre-sponding load cases responsible for the calcu-lated thicknesses. Note that the empennage el-ements were not included in the sizing process.In general, the tail boom is mainly sized due tothe required turn rates about the yaw-axis, in thepresented example it is the hovering load casewhich introduces a lateral force in the rear. Thedominant load case, however, is the +2.5g pull-up maneuver due to its high load factor. Mostof the load is introduced in the cabin area due tothe position of the main rotor and highest nodalmasses. Majority of the elements is sized accord-ing to the shell buckling criterion caused by the+2.5g pull-up maneuver. The areas around thecut-outs require higher thicknesses since theseregions distribute the loads around the fuselageopenings (assumed that the doors must not carry

any loads to prevent door frame deformation incase of emergency landings).

1.0 2.5 4.0

thickness [mm]

hove

ring

max

. cru

ise

turn

03

2.5g

pul

l-up

sizing load case

Fig. 14 Sizing results (scenario no. 7)

Comparing the FE results with the estimationsbased on statistics a comparably big deviationcan be observed. Banked turns and pull-up ma-neuvers can be considered as a roughly sufficientload scenario for a first structural static sizingat early design stages. Weight penalties mustbe considered since the fuselage model repre-sents preliminary detail level which lacks sec-ondary structure, i.e. clips to connect stringersand frames, fasteners, joints, windshield and itsmounting, doors and their opening mechanisms,cabin floor, etc. are not modeled. Masses ofjoining elements may be estimated, depending ontheir amount and design, to account for a weightpenalty of m joints ≈ 20 - 40% of the fuselagemass [23]. Another weight increase is seen byan extension of the integrated load cases, e.g.the touch down during landing or asymmetricflight maneuvers (combination of maneuvers likerolling pull-ups, sideslips and yaw, etc.) whichresult in a combination of lateral, vertical, longi-tudinal load factors and simultaneously rotationalaccelerations.Figure 13 shows that a realistic fuselage mass es-

11


timation depends on properly predicted maneu-vers which highly depend on the mission profileof the rotorcraft, thus marking a frontier betweenstatistical or analytical L1 and computational L2methods. However, a proper prediction of heli-copter maneuvers is considered an illusive goalthat still challenges helicopter analysts for yearsto come [10].

5.4 Lifting surfaces

To overcome the physical limitations at highspeed cruise (e.g. compression effects) of ro-torcraft, novel concepts may comprise axialthrusters and (additional) lifting surfaces. Thesewings generate lift in horizontal flight so that themain rotor can be unloaded and slowed down.The resulting lack of thrust caused by the un-loaded main rotor is then compensated by addi-tional propulsive force generated with auxiliarypropellers, thus increasing the cruise speed. Inthat state, wings have to carry about 75% of therotorcraft mass.Simple wing models have been integrated in themodel generation. The wings modeled consist oftheir primary structure, i.e. webs, spars and thepanels. Therefore, the leading and trailing edgesare not taken into account for the model genera-tion.Wings can be integrated into the airframe eitherby the integration of a wingbox or by main frameattachments. However, a suitable arrangement ofthe gear box, the drive system, and the wing in-tegration structure is required. Currently, wingsare mounted to the fuselage structure with rigidbeam elements, showing a simplified approach.Figure 15 shows this first approach to integratewings into the rotorcraft airframe, featuring anaperture in the center for the integration of themain rotor shaft and main gear box.

6 Conclusion and Outlook

Within the projects RIDE and EDEN, DLR cre-ated an automated and integrated tool chain forrotorcraft design at the conceptual and prelimi-nary design stages. Integrated tool levels range

Fig. 15 Generic compound rotorcraft airframewith lifting surface

from statistics based to higher fidelity tools, e.g.based on FEM. In this paper the L1 mass es-timation module and the L2 structural analysismodule have been presented. For the presentedmedium weight utility rotorcraft, a reduction ofthe fuselage mass when shifting from L1 to L2tools was observed and explained. However,novel configurations require an extension of thepresented tool chain:

• The integration of wings is essential sincewings are seen as a key driver in the aimto overcome physical limitations, such ascruise speed. A suitable arrangement anda detailed modeling of the wing integrationstructure is necessary for sufficient assess-ment.

• A key technology is the integration of com-posite materials, thus allowing to tailor thematerial to the load paths. Early estima-tions during the 1980’s consider additionaltechnology factors of about χ≈ 0.8 for thefuselage weight [25] indicating a signifi-cant weight reduction compared to metallicairframes.

• The integration of crashworthiness toolsmay be beneficial since helicopters are -statistically seen - more often exposed todangerous situations and therefore moreoften involved in emergency landings com-pared to fixed-wing aircraft. The contradic-

12


tional objectives of static sizing and energyabsorption (linear elastic behavior versusnon-linear compliance) is considered as achallenge.

License costs for commercial tools, especiallyfor CAD and FE software, may be reduced byusing open source solutions. Additionally, therenunciation of the dependence from one codeincreases flexibility and allows for comparativeanalysis without a major change of the program-ming code.Currently the aforementioned tools for weight es-timation and structural sizing are under recon-struction for integration into the new softwareenvironment PANDORA (Parametric NumericalDesign and Optimization Routines for Aircraft,[15]), which is at present under development atthe DLR Institute of Structures and Design.Within this redesign a more detailed mass repre-sentation is intended, thus allowing a more real-istic mass distribution of components, e.g. drivetrain, gearboxes, etc. which in turn shall result ina more realistic sizing output.In 2018, DLR initiated the project TRIAD (Tech-nologies for Rotorcraft in Integrated and Ad-vanced Design, until 2020) in which the loadsanalysis will be transferred to the more power-ful tool HOST. Another objective in this projectis the extension of the aerodynamic calculationfor L2 tools by calculating the interaction of themain rotor with other helicopter components andtransfer this information back into the L1 sizing,similar to the structural mass.With the aim of creating a design for a novelemergency helicopter, it is planned to combinethese tasks with the design of a new, enhancedmedical compartment.

Acknowledgements

The work presented in this manuscript wasachieved within the DLR projects EDEN (Eval-uation and Design of Novel Rotorcraft Con-cepts), FAST-Rescue (Fast and Silent Rescue He-licopter) and TRIAD (Technologies for Rotor-craft in Integrated and Advanced Design). The

authors would like to gratefully acknowledge andappreciate the financial support.

References

[1] M.N. Beltramo and M.A. Morris. Parametricstudy of helicopter aircraft systems costs andweights. Technical report, NASA - NationalAeronautics and Space Administration, NASA-CR-152315, 1980.

[2] B. Benoit, A.-M. Dequin, K. Kampa, W. vonGrünhagen, P.-M. Basset, and B. Gimonet.HOST, a General Helicopter Simulation Toolfor Germany and France. In American He-licopter Society 56th Annual Forum, VirginiaBeach, VA, USA, 2000.

[3] E.F. Bruhn. Analysis and Design of Flight Vehi-cle Structures. Tri-State Offset Company, 1973.

[4] W. Johnson. Helicopter Theory. Dover Publica-tions, Inc., 1994.

[5] W. Johnson. NDARC - NASA Design and Anal-ysis of Rotorcraft. Technical report, NASA- National Aeronautics and Space Administra-tion, NASA/TP-2009-215402, 2009.

[6] W. Johnson. Rotorcraft Aeromechanics. Cam-bridge University Press, 2013.

[7] A. Krenik and P. Weiand. Aspects on Con-ceptual and Preliminary Helicopter Design.In Deutscher Luft- und Raumfahrtkongress(DLRK), Braunschweig, Germany, 2016.

[8] P. Kunze. Parametric Fuselage Geometry Gen-eration and Aerodynamic Performance Predic-tion in Preliminary Rotorcraft Design. In 39th

European Rotorcraft Forum, Moscow. Russia,2013.

[9] D.M. Layton. Introduction to Helicopter Con-ceptual Design, 1992.

[10] J.G. Leishman. Principles of Helicopter Aero-dynamics. Cambridge University Press, 2nd edi-tion, 2006.

[11] B. Maskew. Program VSAERO Theory Doc-ument: A Computer Program for Calculat-ing Nonlinear Aerodynamic Characteristics ofArbitrary Configurations. Technical report,NASA, NASA-CR-4023, 1987.

[12] B. Nagel, D. Böhnke, V. Gollnick, P. Schmoll-gruber, A. Rizzi, G. La Rocca, and J. J. Alonso.Communication in Aircraft Design: Can We Es-

13


tablish a Common Language? In 28th Inter-national Congress of the Aeronautical Sciences,Brisbane, Australia, 2012.

[13] B. Nagel, M. Kintscher, and T. Streit. Activeand Passive Structural Measures For AeroelasticWinglet Design. In 26th International Congressof the Aeronautical Sciences, Anchorage, AK,USA, 2008.

[14] D. Palasis. Erstellung eines Vorentwurfsver-fahrens für Hubschrauber mit einer Er-weiterung für das Kipprotorflugzeug. Phdthesis, Universität der Bundeswehr, München,1992.

[15] Petsch, M. and Kohlgrüber, D. and Walther,J.-N. Development of a fully automatedtransport aircraft fuselage modelling and siz-ing tool using Python. In Deutscher Luft-und Raumfahrtkongress (DLRK), Munich, Ger-many, 2017.

[16] R.W. Prouty. Helicopter Performance, Stabil-ity, and Control. Krieger Publishing Company,2002.

[17] J. Roskam. Airplane Design Series. DARcor-poration, 1989.

[18] C. Russell and P.-M. Basset. Conceptual De-sign of Environmentally Friendly Rotorcraft -A Comparison of NASA and ONERA Ap-proaches. In American Helicopter Society 71th

Annual Forum, Virginia Beach, VA, USA, 2015.[19] J. Scherer, D. Kohlgrüber, F. Dorbath, and

M. Sorour. A Finite Element Based Tool Chainfor Structural Sizing of Transport Aircraft inPreliminary Aircraft Design. In Deutscher Luft-und Raumfahrtkongress (DLRK), Stuttgart, Ger-many, 2013.

[20] D.B. Schwinn. Applied parametrized and au-tomated airframe modeling methods in the pre-liminary design phase. International Journal ofModeling, Simulation, and Scientific Comput-ing, 6(4):1550037, 2015.

[21] D.B. Schwinn, P. Weiand, and M. Schmid.Structural Analysis of a Rotorcraft Fuselage ina Multidisciplinary Environment. In NAFEMSWorld Congress, Stockholm, SWE, 2017.

[22] D. Seider, M. Litz, A. Schreiber, and A. Gerndt.Open Source Software Framework for Appli-cations in Aeronautics and Space. In IEEEAerospace Conference, Big Sky, MT, USA,

2012.[23] F.R. Shanley. Weight-Strength Analysis Of Air-

craft Structures. Dover Publications, Inc., 1960.[24] B.G. van der Wall. Grundlagen der

Hubschrauber-Aerodynamik. Springer, 2015.[25] E. Vega. Advanced Technology Impacts on Ro-

torcraft Weight. Journal of the American Heli-copter Society, 31(4):21–28, 1986.

[26] P. Weiand, D.B. Schwinn, M. Schmid, andM. Buchwald. A Multidisciplinary Process forIntegrated Rotorcraft Design. In 43rd EuropeanRotorcraft Forum, Milan, Italy, 2017.

Contact Author Email Address

To contact the corresponding author please mailto: [email protected]

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.

14

structural sizing of a rotorcraft fuselage using … sizing of a...an essential part of the design...

Documents