Copyright 2002 by DLR-SART. Published by the American Institute of Aeronautics and Astronautics, Inc. with permission.

1

AIAA 2002-5220

Progress in the Design of a Reusable Launch Vehicle Stage

Martin Sippel, Josef Klevanski, Holger BurkhardtSpace Launcher Systems Analysis (SART), DLR, Cologne, Germany

Thino Eggers, Ognjan BoiInstitute of Aerodynamics and Flow Technology, DLR, Braunschweig, Germany

Philipp LangholfMAN-Technologie AG, Augsburg, Germany

Andreas RittwegerAstrium GmbH, Bremen, Germany

The German future launcher technology research program ASTRA investigates in its system study two types ofpartially reusable launch vehicles. This paper describes one of those concepts, a reusable first stage designed for a nearterm application with a heavy lift launcher. The attached reference expendable space transportation system is a futureAriane 5 with cryogenic core and upper stage, but skipped solid rocket boosters. The design of the reference liquid fly-back boosters (LFBB) is focused on LOX/LH2 propellant and a future derivative of the Vulcain rocket motor.

After achieving a convergent design in the first iteration loop, a more detailed level of investigation has been started.This includes the ascent control requirements on the booster TVC system, a refinement of the aerodynamic shape, andthe preliminary mechanical lay-out of body and wing structure. All major results are presented, and used in an updateof the mass budget, as well as trajectory simulations and optimizations for ascent and reentry.

Nomenclature

D Drag NM Mach-number -Q heat flux W/m2T Thrust NW weight Nl body length mm mass kgsfc specific fuel consumption g/kNsq dynamic pressure Pav velocity m/s ratio of total pressures Pi/Pi-1 - angle of attack - flight path angle - deflection angle - expansion ratio - control surface defection angle - bypass-ratio - pitch angle -y pitch velocity s-1

Subscripts, Abbreviations

AFRSI Advanced Flexible Reusable Surface InsulationAOA angle of attackCAD computer aided designCFRP Carbon Fiber Reinforced PolymerEAP Etage d'Acclration Poudre (of Ariane 5)EPC Etage Principal Cryotechnique (of Ariane 5)ESC-B Etage Suprieur Cryotechnique (of Ariane 5)FEM finite element methodGLOW Gross Lift-Off MassGTO Geostationary Transfer OrbitH2K Hypersonic Wind Tunnel (at DLR Cologne)

JAVE Jupe AVant Equipe (forward skirt of Ariane 5)LFBB Liquid Fly-Back BoosterLH2 Liquid HydrogenLOX Liquid OxygenMECO Main Engine Cut OffSRM Solid Rocket MotorTMK Trisonic Test Section (at DLR Cologne)TVC Thrust Vector Controlcan canardcog center of gravitysep separations/l sea-level0,0 sea-level, static conditionsxS x-coordinate of the center of gravityxN neutral point position

1 INTRODUCTION

Within the system studies of the German future launchertechnology research program ASTRA two reusable firststage designs are under investigation. The one dedicatedfor near term application with an existing expendable corestage, is called a winged fly-back booster. As one of theresults of the ASTRA-analyses of the years 2000 and2001, it can be stated, that such a reusable booster stage inconnection with the unchanged Ariane 5 expendable corestage is technically feasible, and competitive with otherreusable and advanced expendable launchers.

The basic design philosophy is to choose a robust vehicle,which gives a relatively high degree of confidence toachieve the promised performance and cost estimations.In the second part of the research study 'lessons learned'from the first phase and previous investigations (e.g. ref. 1to 3) are integrated. This notably means, find anacceptable solution for the hypersonic and subsonic cruisetrim requirements by refining the aerodynamic con-

2American Institute of Aeronautics and Astronautics Paper 2002-5220

figuration. A major point is to improve the mechanicallay-out and introduce a structurally based mass estimationto gain a more solid basis for a forthcoming developmentdecision. As far as possible the applicability of existingand already qualified parts should be assessed forintegration in the booster stage.

2 OVERVIEW OF THE INVESTIGATED SEMI-REUSABLE LAUNCH VEHICLE

The regarded partially reusable space transportationsystem consists of dual booster stages, which are attachedto the expendable Ariane 5 core stage (EPC) at anupgraded future technology level. The EPC stage isassumed to be powered by a single advanced derivative ofthe Vulcain engine with increased vacuum thrust, andcontains about 185000 kg of subcooled propellants. A newcryogenic upper stage (ESC-B) is already in thedevelopment phase. It should include a new advancedexpander cycle motor of 180 kN class (VINCI) by 2006.

Two symmetrically attached reusable boosters acceleratethe expendable Ariane 5 core stage (Figure 1). Theyshould replace the solid rocket motors EAP in use today.

Figure 1: Semi-reusable launch vehicle with Ariane 5core stage and two attached reusable fly-back boosters

2.1 LFBB Geometry Data and Lay-OutThe reusable booster stage is based on the same advancedversion of the EPC's Vulcain engine, but employs anadapted nozzle with reduced expansion ratio. The enginesare installed in a group of three in a circular arrangementin the vehicles back. Overall LFBB length is 42 m. Afuselage and outer tank diameter of 5.45 m is selected, toachieve a high commonality with Ariane's main cryogenicstage EPC.

Three air-breathing engines for fly-back are installed inthe vehicle's nose section (see Figure 2), which alsohouses the RCS and the front landing gear. The nose is ofellipsoidal shape with a length of 6.7 m. The positions ofthe turbo-engines is selected with regard to integration

requirements of the core engine, as well as the exhaustduct. The recent design locates the complete RCS in thenose to provide sufficient torque with regard to vehiclecog.

The nose section is followed by an annular attachmentstructure, for which a more detailed description is given inchapter 5.2. The structure for canard mounting andactuation is provided at the center of this attachment ring.The following cylindrical tank is integral and of similarlay-out as for the EPC core stage with same diameter butshortened length. This geometry constraint might reducemanufacturing costs if realized, and enables to bettercompare expendable with reusable structures within thisinvestigation. LOX is stored in the upper position,separated by a common bulkhead from the first LH2 tank.The ascent propellant mass reaches 165000 kg when asubcooling and hence density increase of LOX to 1240kg/m3 and to 76 kg/m for super-cooled LH2 can beachieved. It is assumed to install both the cryo- andthermal insulation externally. This approach is preferred toany arrangement inside, to have better accessibility of thetank walls for inspection between two flights. The integraltank section is followed by the wing and fuselage framesection. A second, non-integral LH2 tank is mountedabove the wing carry-through. This tank is interconnectedwith the main hydrogen tank, and it is currently foreseento perform the engine feed through this second tank.

Figure 2: LFBB projection in the x-z-plane

Figure 3: LFBB projection in the x-y-plane

Figure 4: LFBB projection in the y-z-planeThe main wing lay-out is presently based on an airfoilwith flat lower surface. This type is chosen because theflat lower side is advantageous in hypersonic flow. Thecomplete aerodynamic configuration, including therequired deflection angle of the canards, has beenpreliminarily checked on its trim performance during the

turbo engines

LOX - tank LH2 tank #1

LH2 tank #2

separation motors

RCS - engines

separation motors

3American Institute of Aeronautics and Astronautics Paper 2002-5220

full return flight trajectory. (see chapters 4 and 6) Thewing span reaches more than 21 m and the exposed area isabout 115 m2.

The rocket engines are mounted on a thrustframe. A full2D gimballing of all engines is required to obtainsufficient controllability of the launch vehicle (see chapter3). The engines are protected on the lower side by a bodyflap, also necessary for aerodynamic trimming and control.Two vertical fins are attached to the upper part of thefuselage, inclined by 45 deg. (see Figure 4). The structuralsupport of the complete launch vehicle on the launch tablehas to be provided by the two LFBB.

2.2 Propulsion System DataAn advanced, more puissant version 2 of the Europeancryogenic gas-generator cycle rocket engine Vulcain isunder development, and will be operational in 2002. Aneven more powerful Vulcain 3 is currently in definitionstudies. It might include increased mass flow, higherchamber pressure, and a larger expansion ratio. Althoughno technical data are fixed yet, the presented results of thispaper, and the ASTRA-study is based on assumptionsconcerning the performance of this motor. Engine data ofthe advanced Vulcain variant with reduced expansion ratioto be used in the LFBB configuration are given in Table 1.

Cycle open gas-generatorpropellant combination LOX / LH2nominal thrust (s/l) 1412 kNnominal thrust (vacuum) 1622 kNspecific impulse (s/l) 367.23 sspecific impulse (vacuum) 421.7 schamber pressure 13.9 MPamixture ratio 5.9 -nozzle area ratio 35 -length 2890 mmdiameter 1625 mmdry weight 2370 kgT/W (s/l) 60.7 -T/W (vacuum) 69.8 -Table 1: Proposed Vulcain 3 (= 35) main enginecharacteristics as used in the study

To reduce the mass of fly-back fuel, turbo engines whichuse hydrogen should be implemented. The replacement ofkerosene by hydrogen is not an insurmountable problem,although the special requirements for the existing militaryturbofan EJ-200 will be addressed within the ASTRA-study. The engines will be installed without afterburnerand have a nozzle with fixed throat. Mach and altitudedependent data sets are calculated regarding the Kourouatmospheric conditions. Main technical data at sea-levelare given in Table 2.

The reaction control system (RCS) thrust requirements aredefined with regard to the only flown RLVs: The SpaceShuttle and the Buran orbiter. The sizing of the SpaceShuttle RCS thrusters is based on the yaw acceleration forre-entry attitude control. At maximum vehicle mass about0.5 /s2 has to be achieved 5. Buran had obviously beendesigned in an analogous manner. In case of the LFBBconfiguration these requirement leads to 10 thrusters on

each side of the vehicle with a thrust level of 2 kN perengine. Currently different propellant combinations arelooked upon. Besides the classical but toxic N2O4 / MMH,the environmentally friendly GO2 / Ethanol and GO2 / GH2are regarded. The latter have an operational advantage fora reusable system, but need to be proven concerningreliable ignition capability under all circumstances.Cryogenics have been disregarded due to concern aboutfreezing of feed lines or valves. The low density GH2needs a large and heavy tank, which nevertheless can beintegrated in the nose section. Preliminary RCSdimensioning is done by Astrium with some heritage intesting of advanced propellants for this type of engines. 6

OPR - 26FAN/LPC - 4.35 (3 Stages)HPC - 5.98 (5 Stages)HP-Turbine - 1 StageLP-Turbine - 1 StageBypass ratio - 0.4air mass flow kg/s 77TET K 1800F 0,0 , dry N 54000sfc 0,0 , dry g/kNs 8.1Table 2: EJ-200 technical specification data at Kourousea-level static conditions and hydrogen propellantaccording to abp 4 calculation

The last propulsion system of the LFBB are the solidseparation motors, which provide a safe separation afterbooster MECO. Two of them are located in the attachmentring, and further six are integrated symmetrically on thelower side of the wing. (see Figure 3 and Figure 4) Themotor design is similar to that of today's Ariane 5separation system, but with increased propellant loading.The additional amount is required because of the increasedseparation mass of the reusable stage compared to theexpendable SRM.

3 ASCENT CONTROLA launcher with a dual winged booster configuration issubject to increased wind and gust perturbations comparedwith a fully expendable system without wings. Therefore,the controllability during ascent flight is to be verified. Ina dynamic simulation, control is performed by thrustvector gimballing of all available rocket engines of the twoboosters. A simple 2D flight mechanical model is usedthroughout the investigation of ascent controllability.Conservative assumptions for boundary conditions aretaken, to obtain results with high confidence.

The launcher configuration's orientation in the nominalcase pitches versa the east with the main wingsperpendicular to the flow. This configuration is howeverquite insensitive to the more powerful east-west wind. It istherefore necessary in the 2-D study to turn the launchvehicle by 90 to estimate appropriately the controllabilityduring ascent flight for a worst case scenario of windloads.

3.1 Mass and Inertia ModelThe launcher is modeled using simple geometricalstructures to calculate mass and moments of inertia. The

4American Institute of Aeronautics and Astronautics Paper 2002-5220

propellant center of mass can be rapidly calculated fordifferent filling levels assuming cylindrical tanks. Themoments of inertia are obtained using the formula forcylindrical solids, neglecting fluid dynamic effects.

The center of gravity evolves during ascent flight due tothe consumption of propellant. A linear interpolationbetween the initial and final value is chosen for thepropellant mass of each tank separately, which is a goodapproximation in the case of cylindrical tanks and constantmass flow. The total pitching moment of inertia iscompletely calculated with regard to the launcher center ofgravity position at each instant of the trajectory. Theinertia of the propellants is linearly interpolated in eachtank likewise as the cog movement.

3.2 Aerodynamic ModelThe ascent control analysis requires complete data sets ofthe aerodynamic coefficients CD (Drag), CL (Lift) and CM(Momentum) as function of the angle of attack and Machnumber for the complete ascent configuration. These setscan not be obtained by sophisticated calculations in thisearly phase of preliminary design. Theoretical andempirical methods are used to calculate data for fuselages,wings and wing sections. Aerodynamic coefficients for thewhole launcher are then obtained by linear superposition.The quality of the implied method for pre-analysis studieshas been demonstrated for various launcher andhypersonic configurations 7.

Interference forces are not considered in this simple, butfast and efficient approach. Note, that this is aconservative assumption for the regarded launcherconfiguration. This is especially true at high angle ofattack flight, due to actual partial shading of the secondwing and body not considered in this aerodynamic forcecalculation.

3.3 Wind ModelWind is modeled according to GRAM-95 (see Figure 5).In accordance to 2D flight dynamics, the mean east-westwind for Jan 01 2010 (high solar activity) is simulated.

0

20

40

60

80

100

120

-40 -20 0 20 40 60E-W Wind [m/s]

Altitude

[km]

Figure 5: Mean wind in east-west course at Kourou,projection for 2010Additionally, an instantaneous, angle of attack augmentinggust of 10 m/s is applied during the ascent flightsimulations. The time of the gusts after lift-off isparametrically varied.

3.4 Flight dynamic Model and Control SystemThe simulation is done with a simplified flight dynamicmodel. Flight in 2D over flat earth is assumed. Thedifferential equations are integrated numerically with afixed time step of 0.01 s. The closed loop control systemfollows the optimal ascent flight trajectory alreadycalculated in a 3DOF optimization. The externaldisturbances as e.g. wind have to be compensated. Theprincipal control algorithm complies with the followinglaw and its control system coefficients k:

( ) += dtkk actsetyTVC y All six booster engines are used for control purposes in thenominal case. The nozzle actuators are modeled as aproportional element with a threshold of 0.1. Themaximum deflection of the nozzles is limited to 5.5,maximum gimbal velocity is limited to 10/s. These valuesare well within the operational constraints of the currentAriane 5 EPC.

3.5 Results of dynamic ascent calculationIn a systematical approach, various winged LFBBlauncher configurations with Ariane 5 had been analyzedon their sensitivity to atmospheric perturbations 8. Thesimulations show that the nominal vehicle without gustsperpendicular to the wings is insensitive to changing windconditions. The maximum nozzle deflection during theascent flight stays below 1. This result is in coherencewith expectations as the cross section opposed to windforces is only merely bigger than for a standard Ariane 5configuration.

The most critical instant during the ascent flight isidentified at about 30 s after lift-off. The maximalobtained deflection of the booster engines takes anextreme value for a wind gust at that time 8. This outcomeis in coherence with expectations based on the evolutionof the product of angle of attack times dynamic pressure(q.)max. Another result of this analysis is that both, areduced wing size and a forward position reduce thecontrol effort and augment margins.

The following figures show plots of characteristicparameters during the ascent flight, where it is assumedthat the whole wing surface is exposed to the wind and theaerodynamic moments are therefore bigger. A gust of 10m/s is introduced at 30 s. The transverse nz and axial loadfactors nx are depicted in Figure 6. The impact of theatmospheric perturbation can clearly be seen, even thoughnz never exceeds 0.3 g.

The angular pitch velocity (Figure 7) reaches a peak ofslightly below -1.2 degrees /s immediately after the gust.The other pitching movement is the result of the changingwind profile and the requirement to follow pitch angle ofthe optimized ascent trajectory.

5American Institute of Aeronautics and Astronautics Paper 2002-5220

-0.50

0.51

1.52

2.53

3.54

0 25 50 75 100 125 150time [s]

nx,nz[-

]

nxnz

Figure 6: Load factors nx and nz as function of ascentflight time for an LFBB configuration subject to windand gust

-1.4-1.2-1

-0.8-0.6-0.4-0.2

00.20.4

0 25 50 75 100 125 150time [s]

y[deg/s]

Figure 7: Angular pitch velocity y as function ofascent flight time for an LFBB configuration subject towind and gust

The required nozzle deflection angle of all six boosterengines to counter the perturbation is shown in Figure 8.The simulation shows that the most demanding deflectionvalue of approximately 3 degrees stays well below thelimits of 5.5. The remaining maneuvering margin ishowever reduced. The presented results as well asprevious studies 8 proof the controllability of the examinedlauncher configurations during the mated ascent untilbooster separation, if all booster engines are available fortwo dimensional thrust vector control. The study indicatesthat the ascent flight requirements of control seem to benot the dimensioning factor for wing layout andpositioning.

-3.5-3

-2.5-2

-1.5-1

-0.50

0.5

0 25 50 75 100 125 150time [s]

TVC[deg

]

Figure 8: Thrust vector deflection angle TVC asfunction of ascent flight time for an LFBBconfiguration subject to wind and gust

4 AERODYNAMIC DESIGN AND ANALYSISThe applied aerodynamic and flight dynamic simulation ofthe return flight requires trimmed aerodynamic data setsfor the complete trajectory from separation at M=7 downto the landing phase at M=0.27. The resulting

configuration has to comply with tight margins concerninglongitudinal stability and trim and the behaviour of thebooster has to be robust over the complete Mach numberrange. Another demand is the analysis of the transonicflight regime. These boundary conditions require theapplication of aerodynamic codes which allow to resolveeven small geometric effects.

Therefore, the aerodynamic work is based on unstructuredEuler simulations for M < 2 (DLR TAU code 9) andsurface inclination methods are used for M > 2 (DLRHOTSOSE code 10). One goal of the study is to provide acomplete aerodynamic data set for trim and balance.Additionally, the obtained results are the basis for thedefinition of a wind tunnel model (Figure 9) to beinvestigated in the DLR wind tunnels TMK and H2K. Theforce measurements at Mach numbers between 0.5 < M 2 and in the order of =8 during the cruise flight atM=0.5. The configuration is stable for M > 5.5, butunfortunately a stability problem is detected at the cruiseflight conditions of M=0.5 (Figure 13). A detailedaerodynamic analysis of the problem is given in ref. 11.Although active control (compare chapter 6.2) should notbe completely ruled out, options to find an acceptablesolution in this important flight regime will be analysed in

the near future, regarding an updated vehicle cog andadditional aerodynamic data. The analysis of the windtunnel force measurements will allow a final analysis ofsubsonic cruise flight, and based on these results the wingis going to be adapted. The goal is to stabilise the vehicle,which allows to minimise the required canard deflectionsfor trim during cruise conditions.

The second focus of the aerodynamic investigations is theassessment of the aerodynamic interactions betweenAriane 5 core stage and the LFBBs during mated ascent.With view to the structural layout of wing and bodyperformed (see chapter 5), especially the flight conditionwith maximum dynamic pressure at M=1.6 is of interest.Due to the fact that the configuration is asymmetric forlateral flow conditions the complete configuration has tobe considered (Figure 14).

Figure 14: Flowfield and Mach number contoursaround the Ariane 5 and two attached LFBB at M=1.6, =0, can=0 (Euler calculation)In order to extract the aerodynamic interactions on thestructural loads the isolated LFBB is compared to thecombination of Ariane 5 with two LFBB without andincluding attachment structure between rocket andbooster. As an engineering approach Euler calculationsusing an unstructured mesh are applied (see reference 11).The flowfields and the obtained structural loads are usedas the basis for the mechanical layout of the vehicle.

5 MECHANICAL LAY-OUT OF VEHICLESTRUCTURE

In this phase of the ASTRA-study a preliminarymechanical design of major structural elements isperformed. This work is executed by the German launcherindustry Astrium and MAN. The wing, thrust frame, tanks,and fuselage are dimensioned according to the operationalloads calculated from flight dynamic and aerodynamicanalyses.

The main function of the booster structure is to transferthe thrust into the EPC-stage. The load introduction isforeseen at the forward attachment, in order to keep thesame structural architecture as for the EPC of the presentAriane 5. The booster thrust is submitted from the thrustframe via the rear fuselage, through the LH2 and LOXtank to the attachment ring structure into the EPC.

5.1 Tank, Fuselage and Wing StructureThe general loads for dimensioning of the fuselage takeinto account:

7American Institute of Aeronautics and Astronautics Paper 2002-5220

the axial loads during ascent caused by thrust, inertialoads and aerodynamic drag

the axial flux distribution (warping forces) due to thebooster load introduction

bending moments during ascent due to lateralaerodynamic loads (an effective angle of attack due towind and gusts of 5 at maximum dynamic pressureis assumed) and lateral dynamic loads, which aresupposed to be quasi-static with 1 g

bending moments during descent (the maximumbending moment occurs at M= 5.6 with an angle ofattack of 16 and a dynamic pressure of 27300 Paleading to 3.5 g lateral acceleration)

Axial loads of the descent flight phase are neglected.While for the fuselage the ascent flight is the dimen-sioning case, the descent flight gives the maximum loadfor the wing. All aerodynamic loads (in form of pressuredistributions), coming from 3-D numerical analysis (seechapter 4) are transferred automatically to the 3-Dstructural code (finite element program NASTRAN).

At the LFBB's top the nose cap structure is attached,which is an aerodynamic cover and completely removablefor easy accessibility in case of maintenance.

The load carrying LH2 and LOX tank as part of theforward fuselage as well as the attachment ring structure(see chapter 5.2), are designed similar to the Ariane 5 EPCtank and front skirt JAVE. The minimum and maximuminternal pressures of the integral LOX and LH2 tank aretaken from the Ariane 5 EPC tank.

The axial force fluxes (warping fluxes) due to booster loadintroduction depend on the design and wall stiffness of theattachment ring and the tank structure. Therefore, aniterative sizing approach is necessary. It turns out, that thewarping fluxes due to the load introduction are stilleffective within the cylindrical part of the LOX tank, but"die out" with the beginning of the LH2 tank. Figure 15shows the axial force flux distribution obtained from afinite element analysis.

Figure 15: Axial Force Flux due to Load Introductionfrom attached Core Stage

The cylindrical tank parts are integrally stiffened with thestiffening outside. Since the insulation is foreseenexternally, an internal inspection of the tank skin ispossible.

The tank sizing is made for the two materials Al 2219 (asused in Ariane 5) and the aluminum lithium alloy Al 2195.A mass optimization has been performed (pocket size, skinthickness, stage thickness and height) regarding strength,global buckling, local buckling, and manufacturingconstraints (minimum thickness, number of pockets). Theisogrid stiffened type turns out to be the lightest design.The bare (non-equipped) integral tank mass is calculatedto: Al 2219 : 4580 kg

Al-2195 : 4040 kg

The sizing of the wing and the rear fuselage has also beendone by an iterative approach. In a first step the importantstructural members are pre-designed and dimensionedwith use of local FE-models, manual calculations, andestimated interface forces. As an example, Figure 16shows a design sketch of a ring frame in the rear fuselageloaded by wing interface forces.

Figure 16: Ring Frame Design for the Rear Fuselage

The result of such a frame pre-dimensioning is shown inFigure 17 as a stress plot.

Figure 17: Stress Plot of a Ring Frame at the RearFuselage

8American Institute of Aeronautics and Astronautics Paper 2002-5220

Then a course 3-D finite element model based on thechosen structural concept is made, representing thestiffness of structural members. This model, called staticsystem model (Figure 18), delivers the internal loads(force fluxes in ring frames, shear loads in panels) whenthe external loads are applied. Within parametric inves-tigations the properties of the structural members aremodified in order to reach minimum mass. Strength andstability analyses are done with more detailed finiteelement models.

Figure 18: Static SystemModel for Internal Loads

The rear fuselage is proposed to be made of CFRP, locallyreinforced against buckling.

The structural concept of the wing is shown in Figure 19.The wing box has four spars and it is stiffened with ribs.The shear panels are designed as CFRP sandwich panels,reinforced by T-sections at the lower and upper end.

Figure 19: Wing Box Structural Concept

The finite element model for stress and bucklingverification is shown in Figure 20 and a correspondingstress plot is given in Figure 21.

Figure 20: FE-Mesh of the Wing Box

Figure 21: Stress Plot of the Wing Box

For the thrust frame a trade-off between a truss structure(CFRP struts) and a conical shell has been performed. Itturns out that the shell structure, also made of CFRP, hasmore advantages. A stress plot of the selected thrust framefor the three Vulcain 3 engines is shown in Figure 22.

Figure 22: Stress Plot of the Thrust Frame

The overall structural mass (without thermal protectionand equipment) of the rear fuselage, wings, fins, thrustframe and non-integral second LH2 tank has beendetermined by the structural analysis to arrive at about10370 kg

Stiffness requirements, which can influence the structuralmass, are not defined yet. They have to be derived fromdynamic and aeroelastic investigations, which are foreseenin the next study phase.

5.2 Attachment StructureTo get a more detailed knowledge of the mechanicalarchitecture an in depth analysis of the booster attachmentring is performed. Its basic design should be analogous tothe Ariane 5 EPC forward skirt, but it is especiallyequipped to take care for the requirements of a reusable re-entry vehicle.

This ring is located between the forward end of the oxygentank and the LFBB's nose section. It is one of the mainstructural elements of the booster with very high loads andseveral interfaces like the canard support and the mainattachment fitting, introducing the thrust loads to theexpendable core stage. The length of the ring is 2.5 m withthe booster's external diameter of 5.45 m.

The main task of the attachment ring is to transfer thebooster's thrust loads to the core stage. As with today'sAriane5 all axial thrust is transferred at the forward end ofthe booster into the front skirt JAVE above the EPC. Thehighest load on the main fitting is reached shortly beforethe LFBB's separation, and reaches a peak value of almost3000 kN. The ring further transmits the loads from the air

9American Institute of Aeronautics and Astronautics Paper 2002-5220

breathing engines, from the nose landing gear and fromthe canards. It houses the mechanism of the canardactuation, the two forward separation rocket motors andother secondary structures.

The preliminary structural lay-out is shown in Figure 23.The basic design is similar to the Ariane 5 EPC forwardskirt. But as the booster skirt is unsymmetrically loaded, ithas a strong section around the attachment fitting and aconsiderably thinner and lighter region on the oppositeside. The nose landing gear is located inside the noseassembly close to the ring structure. Therefore, it ispossible to attach the gear's strut support to the samemajor frames of the ring, which already transfer the thrustloads during ascent. (see Figure 24) The multiple use ofstructural elements during different phases of the boostermission enables considerable weight savings.

Figure 23: Preliminary design of LFBB attachmentring showing the stage attachment on the right and thesupport structure for separation motors and canardactuation inside

Figure 24: Preliminary design of LFBB attachmentring showing the internal lay-out including the supportstructure of the nose landing gear

The general layout of this ring comprises: a load carrying outer skin made of integrally stiffened

Al-panels in the highly loaded area and CFRPsandwich panels in the lightly loaded area oppositethe main fitting

two major frames to accommodate the radial loads an inner skin building a closed box with the outer

skin and the frames that takes up the torsion andbending loads from the main fitting.

Because of the booster's reusability all parts have to bedesigned to ease inspection and to allow replacement ofdamaged parts with little effort. Especially limited lifetimeparts (e.g. separation motors) have to be consideredcarefully in their accessibility.

During the re-entry flight the booster is subject toincreased aero-thermal heating. To protect the outer skinagainst this heat flow a thermal protection has to be used.Preliminary estimation of the heat flux indicates that onthe windward side a flexible insulation like AFRSI wouldbe sufficient.

For the analysis and iterative design optimization of thestructure a finite element model has been built. This modelconsists of 7000 elements, six load cases are calculatedand the computing time is approximately 2 minutes each.Due to this quick tool it is possible to perform basicoptimizations very easily. Figure 25 shows the FEM-structure. Adjacent structural items (e.g. the LOX-tank)are also modeled to get realistic boundary conditions.

Figure 25: FEM half model of the attachment ringstructure

Based on this FE-model the mass of the ring structure iscalculated. For all single parts the material thickness istaken from the optimized finite element model and theparts are designed in the CATIA-3D CAD-tool to includethe mass of flanges and stringers.

The mass of the complete attachment ring structure,derived from this preliminary but detailed structuralinvestigation is found at 691 kg.

6 SYSTEM PERFORMANCE

6.1 Ascent flight optimizationThe usual mission of commercial Ariane 5 flights willcontinue to be operated from Kourou to a 180 km x 35786km GTO with an inclination of 7 degrees. This orbit dataand a double satellite launch including the multiple launch

10American Institute of Aeronautics and Astronautics Paper 2002-5220

structure SPELTRA are assumed. The overall ascenttrajectory of Ariane 5 with LFBB is similar to the genericGTO flight path of Ariane 5 with SRM. After vertical lift-off the vehicle turns during a pitch maneuver, and headseastward to its low inclined transfer orbit. This trajectoryhas to respect certain constraints, which are close to thoseof Ariane 5+ ascent. Throttling of the Liquid Fly-backBooster is not performed, since the Ariane 5 accelerationlimit is not reached.

Some characteristic mass data of the investigated LFBBconfiguration as of August 2002 is listed in Table 3. Thedry mass is already partially incorporating the results ofthe structural analyses of chapter 5. The separated satellitepayload mass in double launch configuration exceeds 13Mg. The fully cryogenic launcher (boosters, core, andupper stage) is able to deliver almost 2 % of its gross lift-off mass into GTO.

kgLFBB dry mass: 47500GLOW LFBB mass: 220500GLOW launcher mass: 695775GTO payload mass: 13250Table 3: Characteristic mass data of the Fly-backBooster for GTO mission with Ariane 5 core stage

6.2 Descent and Return FlightInitial re-entry and return flight mass of each LFBB is54460 kg, which is below the MECO-value, because theaft part of the stage attachment is jettisoned, and the solidpropellant of the separation motors is already burned.During the ballistic phase of the trajectory the remainingoxygen in the tank and in the fuel lines will be drained.

Aerodynamic data sets of the booster's return flightconfiguration have been generated in the aerodynamicanalysis of chapter 4. Lift-, drag-, and pitching momentcoefficients with regard to canard and bodyflap deflectionare used in combination with a calculation of center ofgravity movement, to perform a flight dynamics andcontrol simulation. The trimmed hypersonic maximumlift-to-drag ratio reaches about 2.0. In the low subsonicand cruise flight regime trimmed L/D is slightly above 5.0.Hypersonic trimming is performed by the canards andsupported by the RCS. A stable condition is achievable atleast up to angles of attack of 35 degrees. For this type ofLFBB 35 deg. are used as the upper limit during returnflight.

Due to the remaining flight path angle of about 25 degreesat staging the LFBB climbs in a ballistic trajectory above90 km. Falling back the booster still reaches a velocityclose to the separation conditions of 1.95 km/s (aboutMach 6) at 50 km, since the atmospheric drag is low.Although the angle of attack is held at the 35 deg. limit, asteep trajectory is performed due to the restricted dynamicpressure and lift force, with a path angle diving as low as-23.

When entering the denser layers of the atmosphere theaerodynamic forces rapidly increase, finally stabilizing theLFBB altitude, and achieving maximum deceleration at analtitude of around 20 km. The simulation is performedunder a closed control loop, which regulates the trajectorywithin normal load boundaries, as far as flap efficiency is

available. An optimal trajectory is found by parametricvariation of the initial banking maneuver. The return ofthe LFBB should start as early as possible, but is notallowed to violate any restrictions. The banking isautomatically controlled to a flight direction withminimum distance to the launch site. After turning thevehicle, the gliding flight is continued to an altitude ofoptimum cruise condition.

An elaborate method is implemented to calculate therequired fuel mass of the turbojets for the powered returnflight to the launch site. The complete flight is controlledalong an optimized flight profile. Aerodynamic data,vehicle mass, and engine performance (available thrustand sfc) are analyzed in such a way, to determine thestable cruise condition with the lowest possible fuelconsumption per range (g/km). This is not a trivial task,since engine performance is dependent of altitude andMach number, and the equivalence of drag-thrustrespectively lift-weight is usually not exactly found atmaximum L/D. The changing booster mass due to fuelconsuming, and a minimum necessary accelerationperformance have also to be taken into account.

The powered return trajectory is automatically controlledto follow the optimum flight condition, always directlyheading to the launch site. Fuel flow is integrated to get itsexact amount. In case of the most recent LFBBconfiguration, a specific complication has to be addressed.As the aerodynamic calculations show, the vehicle seemsto be subject to an unstable condition during cruise flight.Even though, future improvements of the aerodynamicshape should avoid this difficulty, the flight dynamicssimulation demonstrates that under realistic canardactuator conditions the LFBB is fully controllable byactive means. The main parameters for vehicle control aredepicted in Figure 26 along return flight. Note that theshown Mach number dependency of the canard deflectionis similar but not identical to that in Figure 12 due to theflight dynamics simulation performed here with regard tocog movement.

-10

0

10

20

30

40

0 1 2 3 4 5 6 7M [-]

AngleofA

ttack

[]

-10

-5

0

5

10

15

20Ca

nardsDeflectio

n[]

AoA []Canard defl. []

Figure 26: Control parameters and Canard of theLFBB re-entry and return flight trajectory from flightdynamics simulation

Including 20% fly-back fuel reserves to take into accountadverse conditions like head winds, the booster needsabout 4.1 Mg hydrogen for its more than one hour returnleg.

11American Institute of Aeronautics and Astronautics Paper 2002-5220

7 CONCLUSIONTechnical investigations on a partially reusable spacetransportation system with reusable booster stages,attached to an advanced future derivative of the expen-dable Ariane 5 core stage, demonstrate the feasibility ofseveral promising design features. The fully cryogeniclauncher is able to deliver more than 13000 kg of payloadinto GTO.

The reusable boosters are designed with the same externaldiameter as Ariane5's EPC, the large integral tank is ofsimilar architecture, and the basic lay-out of Ariane 5'sforward skirt JAVE is reused for the LFBB's attachmentring. Therefore, existing manufacturing infrastructuremight be continuously operated for the RLV assembly.The wing and fuselage structure also incorporatesadvanced materials like CFRP.

An aerodynamic vehicle configuration with two largecanards, a clean wing, and an aft-positioned bodyflap isselected. Numerically calculated aerodynamic data sets forthe complete trajectory (M= 0.27 up to M= 7.0) have beenprepared. Canard deflections for trimming remain within amoderate range. However, in the subsonic fly-back cruiseregime a vehicle stability problem is detected. Windtunnel tests of the LFBB scheduled for the near future,will support an adaptation of the wing to eventually reacha naturally stable configuration.

Flight dynamics simulations of the launcher with wingedstages subject to atmospheric perturbations have beenperformed. It can be proofed that the controllability of theexamined configuration during the mated ascent untilseparation is achievable, if all booster engines areavailable for two dimensional thrust vector control. Thestudy indicates that the ascent flight requirements ofcontrol seem to be not the dimensioning factor for winglayout and positioning. A return trajectory flightsimulation demonstrates that under realistic canardactuator conditions the LFBB is fully controllable byactive means despite its stability problem.

Future work on the described Liquid Fly-Back Boosterwill address a further enhancement in system robustnessand cost efficiency. Besides the already mentionedaerodynamic improvement, dynamic and aeroelasticinvestigations of the structure are foreseen. Selection ofmajor subsystems, fly-back turbojet integration issues, aswell as vehicle health monitoring will be covered.

All applied technologies of the LFBB-RLV are wellwithin reach for Europe in the next 10 years. Whensatisfactory operational efficiency is achievable, reusablebooster stages represent an interesting and serious optionin the future European launcher architecture.

8 REFERENCES

1. Sippel, M.; Herbertz, A.; Kauffmann, J.; Schmid, V.:Investigations on Liquid Fly-Back Boosters Based onExisting Rocket Engines, IAF 99-V.3.06, 1999

2. Sippel, M.; Atanassov, U.; Klevanski, J.; Schmid, V.:First Stage Design Variations of Partially Reusable

Launch Vehicles, J. Spacecraft, V.39, No.4, pp. 571-579, July-August 2002

3. Klevanski, I.; Herbertz, A.; Kauffmann, J.; Schmid,V. Sippel, M.: Aspekte der Stabilitt undSteuerbarkeit in der Flug- und Separationsphaseunsymmetrischer Trgerkonfigurationen, DGLR-JT2000-165, 2000

4. Sippel, M.: Programmsystem zur Vorauslegung vonluftatmenden Antrieben fr Raumtransportsysteme,abp 2.0, DLR TB-319-98/07, 1998

5. Edwards, P.R.; Svenson, F.C.; Chandler, F.O.: TheDevelopment and Testing of the Space ShuttleReaction Control Subsystem, ASME-paper 78-WA/AERO-20, 1978

6. Seidel, A.; Pulkert, G.; Wolf, D.: DevelopmentHistory of Small H2/O2- and H2/F2-Engines forAdvanced AOCS-, RVD- and OMS-Systems, AIAAPaper 91-3389, 1991

7. Klevanski, J; Sippel, M. : Beschreibung desProgramms zur aerodynamischen Voranalyse CACVersion 1.0, SART TN-003/2001, DLR-IB 645-2001/02, 2001

8. Klevanski, J.; Burkhardt, H.; Sippel, M.: ParametricalAnalyses of Liquid Fly Back Booster AscentControllability, ASTRA Doc. No. 1000-008, SARTTN 001/2002, DLR-IB 645-2002 / 02, 2002

9. Mack, A; Hannemann, V.: Validation of theUnstructured DLR-TAU-Code for Hypersonic Flows,AIAA Paper 2002-3111, 2002

10. Reisch, U.; Anseaume, Y.: Validation of theApproximate Calculation Procedure HOTSOSE forAerodynamic and Thermal Loads in Hypersonic Flowwith Existing Experimental and Numerical Results,DLR-FB 98-23, 1998

11. Eggers, Th.; Boi, O.: Aerodynamic Design andAnalysis of an Ariane 5 Liquid Fly-Back Booster,AIAA Paper 2002-5197, 2002

Further updated information concerning the SART spacetransportation concepts is available at:http://www.dlr.de/SART