620

JOURNAL OF PROPULSION AND POWER

Vol 14 No 5 Septemberndash October 1998

Advanced Rocket Nozzles

Gerald HagemannDLR German Aerospace Research Center Lampoldshausen 74239 Germany

Hans ImmichdaggerDaimler ndash Benz AG Munich 81663 Germany

Thong Van NguyenDaggerGenCorp Aerojet Sacramento California 95813

andGennady E Dumnovsect

Keldysh Research Center Moscow 125438 Russia

Several nozzle concepts that promise a gain in performance over existing conventional nozzles arediscussed in this paper It is shown that signi cant performance gains result from the adaptation of theexhaust ow to the ambient pressure Special attention is then given to altitude-adaptive nozzle conceptswhich have recently received new interest in the space industry Current research results are presentedfor dual-bell nozzles and other nozzles with devices for forced ow separation and for plug nozzles withexternal freestream expansion In addition results of former research on nozzles of dual-mode enginessuch as dual-throat and dual-expander engines and on expansionndash de ection nozzles are shown In general ow adaptation induces shocks and expansion waves which result in exit pro les that are quite differentfrom idealized one-dimensional assumptions Flow phenomena observed in experiments and numericalsimulations during different nozzle operations are highlighted critical design aspects and operation con-ditions are discussed and performance characteristics of selected nozzles are presented The considerationof derived performance characteristics in launcher and trajectory optimization calculations reveal sig-ni cant payload gains at least for some of these advanced nozzle concepts

NomenclatureA = areaF = thrusth = ight altitudeI = impulsel = lengthmCcedil = mass ow ratep = pressurer = mass ratio oxidizer fuel mixturer = radiusx y = coordinateslaquo = nozzle area ratio

Subscriptsamb = ambientc = combustion chambercr = criticale = exit planegeom = geometricalref = referencesp = speci ct = throat

Received Sept 2 1997 revision received Feb 27 1998 acceptedfor publication March 19 1998 Copyright Q 1998 by the AmericanInstitute of Aeronautics and Astronautics Inc All rights reserved

Research Engineer PhD Space Propulsion E-mail geraldhagemanndlrde

daggerResearch Engineer PhD Space Infrastructure E-mailIm24344dbmaildasede Member AIAA

DaggerTechnical Principal PhD E-mail thongnguyenaerojetcomMember AIAA

sectDivision Head PhD Aerothermodynamics E-mail dumnovrelayrinetru

vac = vacuumw = wall

I Introduction

T HE reduction of Earth-to-orbit launch costs in conjunctionwith an increase in launcher reliability and operational

ef ciency are the key demands on future space transportationsystems like single-stage-to-orbit vehicles (SSTO) The realiz-ation of these vehicles strongly depends on the performanceof the engines which should deliver high performance withlow system complexity

Performance data for rocket engines are practically alwayslower than the theoretically attainable values because of im-perfections in the mixing combustion and expansion of thepropellants Figure 1 illustrates the different loss sources inrocket engine nozzles The examination and evaluation ofthese loss effects is and has for some time been the subject ofresearch at scienti c institutes and in industry Table 1 sum-marizes performance losses in the thrust chambers and nozzlesof typical high-performance rocket engines The SSME- andVulcain 1 engine1 (Space Shuttle main engine Rocketdynehydrogenndash oxygen engine and hydrogen ndash oxygen core engineof European Ariane-5 launcher) Among the important losssources in thrust chambers and nozzles are viscous effects be-cause of turbulent boundary layers and the nonuniformity ofthe ow in the exit area whereas chemical nonequilibriumeffects can be neglected in H2ndash O2 rocket engines with chamberpressures above pc = 50 bar1 Furthermore the nonadaptationof the exhaust ow to varying ambient pressures induces asigni cant negative thrust contribution (see Figs 2 ndash 4) Am-bient pressures that are higher than nozzle wall exit pressuresalso increase the danger of ow separation inside the nozzleresulting in the possible generation of side loads A brief de-scription of state-of-the-art prediction methods for both phe-nomena is given

HAGEMANN ET AL 621

Table 1 Performance losses in conventional rocket nozzlesa

LossesVulcain 1

SSME

Chemical nonequilibrium 02 01Friction 11 06Divergence nonuniformity of exit ow 12 10Imperfections in mixing and combustion 10 05Nonadapted nozzle ow 0ndash 15 0ndash 15aOther loss sources also shown in Fig 1

Fig 1 Flow phenomena and loss sources in rocket nozzles

Fig 2 Rocket nozzle ow elds during off-design operation a)overexpanded ow RL10A-5 engine and b) underexpanded owSaturn-1B Apollo-7 (Photographs United Technologies Pratt ampWhitney NASA)

Fig 4 Flow phenomena for a conventional rocket nozzle

Fig 3 Performance data for nozzle of Vulcain 1 engine (designparameters of Vulcain 1 nozzle laquo = 45 pc pamb = 555 r = 589)

The main part of the paper addresses different nozzle conceptswith improvements in performance as compared to conventionalnozzles achieved by altitude adaptation and thus by minimizinglosses caused by over- or underexpansion Several concepts forthe altitude adaptation of rocket nozzles exist in the literature andthese are considered in this paper in more detail2ndash 4 This discus-sion of advanced nozzles includes nozzles with inserts for con-trolled ow separation two-position nozzles dual-bell nozzlesdual-expander and dual-throat nozzles expansionndash de ection noz-zles and plug nozzles In the past practically all of these conceptshave been the subject of analytical and experimental work5ndash7

Although bene ts in performance were indicated in most of theavailable publications none of these nozzle concepts has yet beenused in existing rocket launchers This may change in the nearfuture because a rocket engine with a linear plug nozzle is fore-seen as the propulsion system for the Lockheed Martin liftingbody RLV X-33 concept

II Conventional NozzlesConventional bell-type rocket nozzles which are in use in

practically all of todayrsquos rockets limit the overall engine per-

622 HAGEMANN ET AL

Fig 5 Flow separation in overexpanding rocket nozzles wall pressure pro le and phenomenology

formance during the ascent of the launcher owing to their xedgeometry Signi cant performance losses are induced duringthe off-design operation of the nozzles when the ow is over-expanded during low-altitude operation with ambient pressureshigher than the nozzle exit pressure or underexpanded duringhigh-altitude operation with ambient pressures lower than thenozzle exit pressure Figure 2 shows photographs of nozzleexhaust ows during off-design operation In the case of over-expanded ow oblique shocks emanating into the ow eldadapt the exhaust ow to the ambient pressure Further down-stream a system of shocks and expansion waves leads to thecharacteristic barrel-like form of the exhaust ow In contrastthe underexpansion of the ow results in a further expansionof the exhaust gases behind the rocket

Off-design operations with either overexpanded or under-expanded exhaust ow induce performance losses Figure 3shows calculated performance data for the Vulcain 1 nozzle asfunction of ambient pressure together with performance datafor an ideally adapted nozzle Flow phenomena at differentpressure ratios pc pamb are included in Fig 4 [The sketch with ow phenomena for the lower pressure ratio pcpamb shows anormal shock (Mach disk) Depending on the pressure ratiothis normal shock might not appear see eg Fig 2] TheVulcain 1 nozzle is designed in such a manner that no uncon-trolled ow separation should occur during steady-state oper-ation at low altitude resulting in a wall exit pressure of pwe

rsquo 04 bar which is in accordance with the Summer eld cri-terion8 The nozzle ow is adapted at an ambient pressure ofpamb rsquo 018 bar which corresponds to a ight altitude of h rsquo15000 m and performance losses observed at this ambientpressure are caused by internal loss effects (friction diver-gence mixing) as shown in Fig 1 and Table 1 Losses inperformance during off-design operations with over- or un-derexpansion of the exhaust ow rise up to 15 In principlethe nozzle could be designed for a much higher area ratio toachieve better vacuum performance but the ow would thenseparate inside the nozzle during low-altitude operation withan undesired generation of side-loads

A Flow Separation and Side-Loads

Flow separation in overexpanding nozzles and its theoreticalprediction have been the subject of several studies in thepast348 and different physical models and hypotheses for theprediction of ow separation have been developed In stronglyoverexpanding nozzles the ow separates from the wall at acertain pressure ratio of wall pressure to ambient pressure pwpamb The typical structure of the ow eld near the separationpoint is shown in Fig 5 together with wall pressure dataSeparation and the formation of a recirculation zone at the wall

induce an oblique shock wave near the wall which leads to arecompression of the ow

The physical phenomenon of ow separation can be dividedinto two simple phenomena3 The rst is the turbulent bound-ary-layer separation from the nozzle wall which is character-ized by the ratio of the nozzle wall pressure just after theseparation pp to the nozzle wall pressure just before separa-tion psep This pressure ratio is referred to as the critical pres-sure ratio pcr = pseppp = p1p2 The second phenomenon isconnected with the ow in the separated zone which is char-acterized by a minor pressure gradient along the wall Theanalysis of model experimental data on separated turbulent su-personic ows shows that the pressure ratio pcr is equal forseparated ows before an obstacle and for separated nozzle ows3 For turbulent ows pcr shows a slight dependence onReynolds number but depends strongly on Mach number Fur-thermore investigations of separated ows in rocket enginenozzles showed that pcr is also a function of wall temperaturegas composition and nozzle wall roughness

Various approaches to the prediction of ow separation areused in industry and research institutes For example the Kel-dysh Centerrsquos method for the determination of the ow sepa-ration point in rocket nozzles is based on empirical relation-ships obtained for pcr pppwe and pwepamb Using theserelations the pressure ratio pseppamb the separation point andthe wall pressure distribution are determined Aerojet uses tab-ulated data for the prediction of ow separation from variousexperiments with conical and bell-shaped nozzles4 At a knownpressure ratio of chamber pressure to ambient pressure a lowerand an upper separation pressure ratio follows from the data-base that is different for conical and bell-shaped nozzles Aseparation criterion used at the European industry and scien-ti c institutes describes the pressure ratio pseppamb as functionof the local ow Mach number at the wall near the separationpoint This analytical model was derived from various cold-and hot- ring tests of overexpanding nozzles9

Various numerical simulations of ow problems featuring ow separation have shown that the numerical prediction of ow separation with state-of-the-art turbulence modeling re-sults in good agreement with experimental data310 As an ex-ample Figs 6 and 7 compare experimental and numerical dataon wall pressure ratios in two overexpanding nozzles Fur-thermore former experiments of separated ows indicated thatthere is a random movement of the separation line in over-expanding rocket nozzles91112 resulting in the possible gen-eration of side loads The correct prediction of side-loads withmodels is uncertain and still a subject of on-going researchCommon side-load models described in Refs 4 and 9 assumethat the separation line in the nozzle has a maximum tilt angle

HAGEMANN ET AL 623

Fig 6 Experimental and numerical data on wall pressure ratiosin overexpanding conical nozzles with ow separation Cold-gastest case with subscale conical nozzle

Fig 7 Experimental and numerical data on wall pressure ratiosin overexpanding bell nozzles with ow separation Hot-gas testcase with J-2S engine

with the centerline The minimum and maximum separationpoints are predicted using the upper and lower separation datathat result from the applied separation model In the case ofdistinct separation points as obtained with the Summer eld- orSchmucker-criterion a certain scattering around the predictedseparation point is assumed The side-load is then calculatedby integrating the wall pressure in the asymmetrical ow sep-aration region49 These approaches gave reasonable results forthe J2-S and Vulcain 1 nozzle

At the Keldysh Center in Russia model experiments of sep-arated ow effects on side-loads acting on nozzles began inthe early 1980s11 These investigations were directed mainlyto the determination of the unsteady pressure uctuations inthe separation zone on the whole engine The method is de-scribed in Ref 11 in more detail it is based on a generalizationof obtained empirical data for the spectral density of the pul-sating pressure eld in the separation zone Distribution of thepressure amplitudes is governed by probability density func-tions This method was validated with the experimental rms-val-ues of side-loads acting on the RD-0120 nozzle (CABD hydro-genndash oxygen engine rst-stage engine of Russian Energialauncher) Other hypotheses for the appearance and predictionof side-loads also exist eg based on aeroelastic analyses13

which account for the coupling of ow oscillations with thinnozzle shells However these hypotheses require experimentalveri cation and validation

Side-loads are an undesired phenomenon that may result inthe destruction of the rocket nozzle and should therefore beavoided In Ref 9 the destruction of a J-2D engine (Rocket-dyne hydrogen ndash oxygen engine second- and third-stage engineof Saturn-5 launcher) as a result of side-loads is reported The

transient startup of the rocket engine generates a nozzle owwith uncontrolled ow separation and with possible side-loadgeneration in the nozzle whereas maximum area ratios of all rst-stage nozzles or booster nozzles are chosen to avoid owseparation at the nominal chamber pressure operation As aresult the vacuum performance of rocket engines that operateduring the entire launcher trajectory such as the SSME orVulcain 1 engines is limited

B Potential Performance Improvements

Compared to existing rocket engines a gain in performanceis achieved with advanced engines such as mixed-mode pro-pulsion systems dual-mixture ratio engines or dual-expanderengines Nevertheless the upgrade of existing engines withbetter performing subsystems such as turbines and pumpsalso leads to a gain in overall performance data and is dis-cussed in more detail in Ref 14 Nozzle performance of con-ventional rocket engines is already very high with regard tointernal loss effects (friction nonuniformity) However fornozzles of gas-generator open-cycle engines such as the Vul-cain 1 engine a slight improvement in performance can beachieved with turbine exhaust gas (TEG) injection into themain nozzle as realized in the F-1 (kerosenendash oxygen engine rst stage of Saturn-5 launcher) and J-2S and it is foreseenfor the Vulcain 2 engine (hydrogen ndash oxygen engine upgradeof Vulcain 1 engine) and con rmed by numerical simula-tions131516 and experimental results3 This is mainly achievedthrough lower friction losses in the main nozzle1315 and be-cause the bypass nozzles used for the expansion of the TEGswhich have higher divergence losses are removed

Despite the slight performance gain by TEG injection thelow-pressure near-wall stream of the injected gas favors a re-duction of the critical pressure ratio at which ow separationoccurs and therefore an earlier nozzle ow separation312 Fur-thermore the presence of the secondary exhaust gas injectioncomplicates nozzle-contouring methods with regard to theavoidance of uncontrolled ow separation for rst-stage orbooster nozzles

III Altitude Adaptive NozzlesA critical comparison of performance losses shown in Table

1 reveals that most signi cant improvements in nozzle perfor-mance for rst-stage or SSTO engines can be achieved throughthe adaptation of nozzle exit pressures to the variations in am-bient pressure during the launcherrsquos ascent through the atmo-sphere Various concepts have been previously mentioned andwill be discussed in detail in the following text

A Nozzles with Devices for Controlled Flow Separation

Several nozzle concepts with devices for controlled owseparation have been proposed in the literature with primaryemphasis on the reduction of side-loads during sea level orlow-altitude operation But the application of these conceptsalso results in an improved performance through the avoidanceof signi cant overexpansion of the exhaust ow

1 Dual-Bell Nozzle

This nozzle concept was rst studied at the Jet PropulsionLaboratory17 in 1949 In the late 1960s Rocketdyne patentedthis nozzle concept which has received attention in recentyears in the US and Europe671819 Figure 8 illustrates thedesign of this nozzle concept with its typical inner base nozzlethe wall in ection and the outer nozzle extension This nozzleconcept offers an altitude adaptation achieved only by nozzlewall in ection In low altitudes controlled and symmetrical ow separation occurs at this wall in ection (Fig 9) whichresults in a lower effective area ratio For higher altitudes thenozzle ow is attached to the wall until the exit plane and thefull geometrical area ratio is used Because of the higher arearatio an improved vacuum performance is achieved Howeveradditional performance losses are induced in dual-bell nozzles

624 HAGEMANN ET AL

Fig 8 Sketch of a dual-bell nozzle

Fig 9 Flow eld phenomena in dual-bell nozzles a) sea-levelmode with ow separation at the wall in ection point and b) al-titude mode with a full- owing nozzle

Fig 10 Performance data of a dual-bell nozzle Performance iscompared with two baseline bell-type nozzles as function of ightaltitude (baseline nozzle 1 same area ratio as dual-bell base noz-zle baseline nozzle 2 same area ratio as nozzle extension)

Fig 11 Performance characteristics of a dual-bell nozzle Per-formance is compared with two baseline bell-type nozzles as func-tion of ight altitude (baseline nozzle 1 same area ratio as dual-bell base nozzle baseline nozzle 2 same area ratio as nozzleextension)

as compared with two baseline nozzles having the same arearatio as the dual-bell nozzle at its wall in ection and in its exitplane Figures 10 and 11 illustrate the performance of a dual-bell nozzle as a function of ight altitude in comparison withboth baseline bell-type nozzles (Design parameters of thedual-bell nozzle are taken from a launcher analysis publishedin Ref 18 Propellants hydrogenoxygen r = 6 pc = 200 barrt = 007 m wall in ection at area ratio laquoB = 30 and totalarea ratio laquoE = 100) The pressure within the separated owregion of the dual-bell nozzle extension at sea-level operationis slightly below the ambient pressure inducing a thrust lossreferred to as lsquolsquoaspiration dragrsquorsquo In addition ow transitionoccurs before the optimum crossover point which leads tofurther thrust loss as compared to an ideal switchover Thenonoptimum contour of the full owing dual-bell nozzle re-sults in further losses at high altitudes

To gain insight into the performance and ow behavior ofdual-bell nozzles at different ambient pressures extensive nu-merical simulations with parametrical variations of contour de-sign parameters were performed18 An optimized bell nozzlewith equal total length and area ratio was used as the referencenozzle for comparison As a result the vacuum performance ofthe dual-bell nozzles has a degradation because of the imper-fect contour and this additional loss has the same order ofmagnitude as the divergence loss of the optimized bell nozzle

The simulations of sea-level operation also revealed an ad-ditional performance loss because of aspiration drag which isless than 3 for these dual-bell nozzles This additional lossdepends linearly on the ambient pressure and therefore it isreduced during the ascent of the launcher Furthermore thesesimulations showed that the application of commonly usedseparation criteria derived for conventional nozzles and ap-

plied to dual-bell nozzles with their wall in ection to estimatethe critical pressure ratio yields reasonable results with an ac-curacy of rsquo15 Experimental data obtained at the KeldyshCenter revealed that the critical pressure ratio at the wall in- ection is about 15ndash 20 less than the critical pressure ratioof a conventional nozzle and that usual separation criteriamust be corrected for an accurate prediction of the critical wallpressure ratio

Flow transition behavior in dual-bell nozzles strongly de-pends on the contour type of the nozzle extension3618 A sud-den transition from sea-level to vacuum operation can be atleast theoretically achieved by two different extensions witha zero wall pressure gradient (constant pressure extension) ora positive wall pressure gradient (overturned extension) But acritical analysis of the transition behavior considering decreas-ing ambient pressures during the launcher ascent revealed thata considerable time with uncontrolled ow separation withinthe nozzle extension exists even for these types of extensionsThe duration of this period can be reduced drastically by throt-tling the chamber pressure18

The main advantage of dual-bell nozzles as compared toother means of controlling nozzle ow separation is its sim-plicity because of the absence of any movable parts and there-fore its high reliability It is necessary to note that the external ow over the vehicle in ight reduces the pressure in the ve-hicle base region where engines are installed The ambient

HAGEMANN ET AL 625

Fig 13 Principle performance characteristics of the nozzle withinsert

Fig 12 RD-0120 nozzle hardware with removed nozzle insertand sketch of secondary nozzle mounted inside of the RD-0120nozzle (photographs taken from Ref 23)

pressure triggering the ow transition is the vehicle base pres-sure instead of the atmospheric pressure at the speci c ightaltitude As the base pressure is lower than the atmosphericpressure the nozzle ow transition occurs at a lower altitudethan the one showed in Fig 11 which slightly decreases theef ciency of the dual-bell nozzle operation along the trajectory

Despite the additional losses induced in dual-bell nozzlesthey still provide a signi cant net impulse gain over the entiretrajectory as compared to conventional bell nozzles Indepen-dent combined launcher and trajectory analyses performed byDasa within the European Space Agency Future EuropeanSpace Transportation Investigations Program (ESA FESTIP)study19 and at DLR on SSTO vehicles powered with dual-bellnozzles result in a signi cant payload gain when comparedwith a reference launcher equipped with conventional nozzles

2 Nozzles with Fixed Inserts

A trip ring attached to the inside of a conventional nozzledisturbs the turbulent boundary layer and causes ow separa-tion at higher ambient pressures At higher altitudes with lowerambient pressures the ow reattaches to the wall behind thetrip ring and full owing of the nozzle is achieved The tran-sition from sea level to vacuum mode depends on the wallpressure near the trip-ring location and on the disturbance in-duced by the trip ring The size of the trip ring is a compromisebetween stable ow separation during sea-level operation andthe induced performance loss during vacuum operation In Ref9 it is reported that a trip-ring size of 10 of the local bound-ary-layer thickness is suf cient to ensure stable ow separa-tion

In principle this concept is similar to the dual-bell nozzleconcept with regard to performance characteristics as shownin Fig 11 The sea-level performance of this nozzle concept islower than the performance of a conventional bell nozzle trun-cated at the trip-ring location because of the aspiration dragin the separated ow region of the nozzle Furthermore at sealevel the bell nozzle with trip rings has even higher divergencelosses than a comparable dual-bell nozzle because the nozzlecontour upstream of the obstacle differs from the optimal con-tour for this low-area ratio as a result of the bell nozzle designfor best vacuum performance The additional losses inducedduring vacuum operation is about 1 compared with the per-formance of the bell nozzle without an obstacle Thus theadditional losses are comparable to the additional losses in-duced in dual-bell nozzles As for dual-bell nozzles the tran-sition behavior of this nozzle concept is uncertain but it willbe even more uncertain than for a dual-bell nozzle with a con-stant pressure or overturned nozzle extension

In principle several altitude adaptations can be achievedwith one nozzle by various trip rings mounted one behind theother However this results in increasing vacuum performancelosses The trip rings can also be attached into existing nozzlesand therefore represent a low-cost concept at least for testpurposes with low technological risk Trip rings have beendemonstrated to be effective for side-load reductions duringthe transient startup of rocket engines20 The main problemswith trip-ring nozzles are not only performance losses but alsoring resistance in high-temperature boundary layers the exactcircumferential xing and the uncertainties in the transitionbehavior These uncertainties might be why active interest inthis nozzle concept in the 1970s which is documented in var-ious publications920ndash22 has disappeared in recent years

A further concept with a xed wall discontinuity is a nozzlewith a circumferential groove The aerodynamic behavior of anozzle with a groove and nozzle with trip rings are quite sim-ilar9 but this concept is hard to realize in the case of thinnozzle shells which are used in all of todayrsquos rocket nozzles

3 Nozzles with Temporary Inserts

Nozzle concepts with xed wall discontinuities have the dis-advantage of lower vacuum performance as compared with a

conventional bell nozzle with equal design and operation dataA promising concept for controlled ow separation is thereforetemporary inserts which are removed for vacuum operationThese inserts can be either ablative or ejectible The insertsmay have the form of a complete secondary nozzle23 (Fig 12)or of small steps attached inside the nozzle wall In case ofejectible inserts a reliable mechanism is needed to provide asudden and symmetrical detachment In any case shocks areinduced during the transient ejection because the inserts act asan obstacle in the supersonic exhaust ow These shocks alsointeract with the nozzle walls and increase pressure loads onthe wall and local heat uxes A nonsymmetrical ejectionwould then result in the generation of side-loads Furthermorethe danger of a downstream collision with the nozzle wallarises because the inserts might also experience a transversalmovement toward the walls

Recent hot- ring tests performed in Russia with a modi edRD-0120 engine equipped with a secondary nozzle insert re-vealed a signi cant performance gain of 12 during the sea-level operation at 100 chamber pressure compared with theoriginal RD-0120 performance23 Nominal chamber pressureof this engine is pc = 206 bar with an area ratio of laquo = 857These full-scale hot- ring tests demonstrated the durability ofmaterials sealings and the release mechanism and thus thefeasibility of this concept Figure 12 shows the nozzle hard-ware and a sketch of the secondary nozzle mounted inside theRD-0120 nozzle23

The principle performance characteristics of this RD-0120nozzle with ejectible insert are included in Fig 13 The nozzleoperation with insert results in a slight performance loss com-pared with an ideal bell nozzle with the same reduced area

626 HAGEMANN ET AL

Fig 14 Sketch of a two-position nozzle during a) sea-level andb) vacuum operations

ratio because of aspiration drag and the presumably nonoptim-ized insert contour The performance degradation is compara-ble to the one induced in dual-bell nozzles during sea-leveloperation (Fig 11) In vacuum or high-altitude operations thehigher performance of the baseline nozzle is achieved

Another method for removing the inserts is to use combus-tible or ablative elements924 During the ascent of the launcherthe size of the insert is continuously reduced until it is com-pletely consumed resulting in a full owing bell nozzle witha clean contour for best vacuum performance The principaluncertainties of this nozzle concept are the stability and surfaceregression rates of the inserts Furthermore a homogeneoussymmetrical and temporally de ned consumption must beguaranteed despite possible local pressure and temperature uctuations near the nozzle walls This is currently highly un-certain

4 Nozzles with Active or Passive Secondary Gas Injection

With this nozzle concept the ow in an overexpanding noz-zle is forced to separate at a desired location by injecting asecond uid into the gas stream in the wall normal directionto achieve maximum disturbance of the main exhaust gas andthus to induce ow separation This secondary gas injectionis quite different from the aforementioned TEG injectionwhich should be injected with few or no disturbances to themain ow The injection could be either active that is forcedthrough gas expansion from a higher-pressure reservoir or pas-sive using holes in the wall through which ambient gas issucked in (vented nozzle concept) The latter concept can onlywork in case of gas pressures near the wall inside the nozzlethat are lower than the ambient pressure Experience on forcedsecondary gas injection gained at Aerojet shows that a largeamount of injected uid is required to induce a signi cant owseparation Furthermore no net speci c impulse gain is real-ized when considering the additional gas ow rate

In the vented nozzle concept segments of the bell nozzlewall have various slots or holes opened to the outside ambientpressure During low-altitude operation the slots are opened totrigger ow separation within the nozzle whereas the slots areclosed during high-altitude or vacuum operation so that thegas fully expands inside the entire nozzle Experiments with amodi ed RL10A-3 engine equipped with this nozzle conceptwere performed at Pratt amp Whitney25 Performance resultsshowed that over a small range of low-pressure ratios the per-forated nozzle performed as well as a nozzle with its area ratiotruncated immediately upstream of the vented area Howeverat above some intermediate pressure ratios the thrust ef -ciency suddenly dropped and approached that of the full- ow-ing nozzle The measured performance characteristics werequite similar to those of the dual-bell nozzles shown in Fig11

The altitude range of this nozzle concept is limited by thenumber and position of the holes because the pressure withinthe nozzle must be lower than the ambient pressure Further-more as the rocket aft-body base pressure is lower than thealtitude atmospheric pressure in the surrounding ambient ows the nozzle ow transition occurs at a lower altitude andthus the range of compensation is further reduced

5 Two-Position or Extendible Nozzles

Nozzles of this type with extendible exit cones are currentlyused only for rocket motors of upper stages to reduce the pack-age volume for the nozzle eg at present for solid rocketengines such as the inertial upper stage (IUS) or for the liquidrocket engine RL10 The main idea of the extendible extensionis to use a truncated nozzle with low expansion in low- ightaltitudes and to have a higher nozzle extension at high alti-tudes Figure 14 illustrates this nozzle concept Its capabilityfor altitude compensation is indisputable and the nozzle per-formance is easily predictable The whole nozzle contour in-cluding the extendible extension is contoured for maximum

performance at the high-area ratio with either the method ofcharacteristics or a variational method The area ratio of the rst nozzle section is then determined and the nozzle contouris divided into two parts The xed nozzle part and the ex-tendible extension Investigations conducted at the KeldyshCenter have shown that this nozzle-contouring method is notonly the simplest but also provides a good overall trajectoryperformance3 A minor performance loss is incorporated duringlow-altitude operations because of the truncated inner nozzlewhich has a nonoptimal contour for this interim exit area ratioThe performance characteristics as a function of ight altitudeare similar to those of the nozzle with the ejectible insert asshown in Fig 13 A tradeoff study performed at the KeldyshCenter on nozzles with an ejectible insert and an extendibleextension showed that the thrust characteristics of the nozzlewith an ejectible insert are slightly better than those of theextendible nozzle concept with the same overall dimensionsfor both nozzles

The main drawback of the extendible nozzle concept is thatit requires mechanical devices for the deployment of the ex-tension which reduces engine reliability and increases totalengine mass The necessity for active cooling of the extendibleextension requires exible or movable elements in the coolingsystem which also reduces system reliability The in uencesof the external ow and the generated jet noise on the extend-ible extension in the retracted initial position have not yetbeen fully investigated Former investigations conducted in theKeldysh Center showed that the external ow causes bothsteady and unsteady pressure loads on the retracted nozzle ex-tension whereas the engine jet noise causes strong vibrationsof the nozzle extension

B Plug Nozzles

Various experimental analytical and numerical research onplug nozzles have been performed since the 1950s in theUS4532ndash 35 Europe5710192627 Russia3 and Japan28 In contrastto the previously discussed nozzle concepts plug nozzles pro-vide at least theoretically a continuous altitude adaptation upto their geometrical area ratio Figures 15 and 16 show twodifferent design approaches for circular plug nozzles whichdiffer only in the chamber and primary nozzle layout A con-ical central body is shown here which could also be designedwith more sophisticated contouring methods2930 Different de-sign approaches include plug nozzles with a toroidal chamberand throat (with and without truncation) and plug nozzles witha cluster of circular bell nozzle modules or with clusteredquasirectangular nozzle modules The latter approach seems to

HAGEMANN ET AL 627

Fig 17 Flow phenomena of a plug nozzle with full length (left column) and truncated central body (right column) at different pressureratios pc pamb off-design (top bottom) and design (center) pressure ratio

Fig 16 Principle design of plug nozzles clustered plug 36 mod-ules with truncated plug body

Fig 15 Principle design of plug nozzles toroidal plug fulllength

be advantageous because further losses induced by the gapsbetween individual modules and the ow eld interactionsdownstream of the module exits can be minimized It has beenshown that transition from a round to a square nozzle resultsin a very small performance loss31 In principle the ow elddevelopment of a clustered plug nozzle with rectangular nozzlemodules is similar to that of a toroidal plug nozzle but avoidsthe inherent disadvantages of the toroidal plug design regard-ing 1) the control of a constant throat gap during manufactur-

ing and thermal expansion (side-loads and thrust vector devi-ations) 2) the cooling of toroidal throat with tiny throat gapsand 3) the control of combustion instabilities in the toroidalcombustion chamber Another plug nozzle con guration is thelinear plug nozzle which is foreseen for the propulsion systemof the RLV X-33 concept

1 Circular Plug Nozzles

Figure 17 summarizes the principle ow phenomena of cir-cular plug nozzles with full length and truncated central bodiesat different off-design (top and bottom) and design (center)pressure ratios that were observed in experiments and numer-ical simulations For pressure ratios lower than the design pres-sure ratio of a plug nozzle with a well-contoured central bodythe ow expands near the central plug body without separa-tion and a system of recompression shocks and expansionwaves adapts the exhaust ow to the ambient pressure pambThe characteristic barrel-like form with several in ections ofthe shear-layer results from various interactions of compres-sion and expansion waves with the shear layer and turbulentdiffusion enlarges the shear layer farther downstream of thethroat The existence of the overexpansion and recompressionprocesses is inferred from up- and down-variations of plugwall pressure pro les observed in various cold- ow tests andnumerical simulations and will also be shown later for linearplug nozzles

At the design pressure ratio (see Fig 17 left column cen-ter) the characteristic with the design Mach number should bea straight line emanating to the tip of the central plug bodyand the shear layer is parallel to the centerline However forcircular plug nozzles designed with contouring methods pro-posed in Refs 29 and 30 no exact one-dimensional exit owpro le can be achieved because both methods use thePrandtlndash Meyer relations that are only valid for planar owsFurthermore nonhomogeneous ow in the throat regionwhich is in general not considered within the contour designprocess also in uences the exit ow pro le The wall pressuredistribution remains constant at pressure ratios above the de-sign pressure ratio ie the plug nozzle behaves like a con-ventional nozzle the loss of its capability of further altitude

628 HAGEMANN ET AL

Table 2 Design parameters of toroidal plug nozzles

Design data Full-scale nozzlea Subscale nozzleb

Chamber pressure pc 100 bar 12 barPropellants Hydrogenoxygen Gasndash oilnitric-acidMixture ratio r 60 25Inner chamber diameter dcinner 43 m 019 mOuter chamber diameter dcouter 44 m 034 mGeometric area ratio laquo 55 10aUsed for numerical simulations and performance predictions see Fig 18

bUsed in experiments see Fig 20

Fig 18 Performance of numerically simulated plug nozzle withfull-length central body

Fig 19 Performance of numerically simulated plug nozzle withtruncated central body

adaptation being included Figure 17 (left column bottom) il-lustrates the ow eld at higher pressure ratios Performancedata of a typical plug nozzle are included in Fig 18 and com-pared to a conventional bell nozzle with equal area ratio Thedesign and combustion chamber parameters are included in theleft column of Table 227

The truncation of the central plug body which is an advan-tage because of the huge length and high structural mass ofthe well-contoured central body results in a different ow andperformance behavior as compared to the full-length plug noz-zle At lower pressure ratios an open wake ow establisheswith a pressure level practically equal to the ambient pressure(Fig 17 right column top) At a certain pressure ratio closeto the design pressure ratio of the full-length plug nozzle thebase ow suddenly changes its character and turns over to theclosed form characterized by a constant base pressure that isno longer in uenced by the ambient pressure Analyses indi-cate that shorter plug bodies with higher truncations trigger anearlier change in wake ow at pressure ratios below the designpressure ratio At the transition point the pressure within thewake approaches a value that is below ambient pressure andthe full base area induces a negative thrust (Fig 17 right col-umn center) This thrust loss depends on the percentage oftruncation and the total size of the base area Published ex-perimental data and numerical simulations reveal an increasingthrust loss for shorter plug bodies because the total base areaincreases

Beyond the transition point the pressure within the closedwake remains constant At these lower ambient pressures thebase pressure is then higher than the ambient pressure result-ing in a positive thrust contribution of the total base area Per-formance data of a numerically simulated truncated plug noz-zle are shown in Fig 19 and compared to the same plugnozzle with full-length central body and a conventional bellnozzle Design parameters for this truncated plug nozzle arethe same as for the full length plug (see Fig 18 and Table 2)

Figure 20 shows a typical photograph of a sea-level hot-runtest with a truncated toroidal subscale plug nozzle performedat DLR5 Nozzle design data are included in Table 2 For com-parison of experimental results with numerical computations

the ow eld of the toroidal plug nozzle was calculated with anumerical method10 and Fig 20 shows the calculated Mach-number distribution in the combustion chamber and nozzlePrincipal physical processes like expansion waves shocks andthe recirculating base- ow region are in good agreement Boththe experiment and the numerical simulation show that the ow separates from the conical plug body before reaching itstruncated end Recent results on experiments with plug nozzlespublished in Ref 28 also reveal separation of the ow fromthe central plug body for conical contours In contrast no sep-aration was observed for contoured central plug bodies de-signed with the method proposed in Ref 29 Recent numericalsimulations of contoured plug nozzles performed at DLRwithin the ESA Advanced Rocket Propulsion Technology(ARPT) Program on advanced rocket propulsion also showthat no ow separation occurs from well-contoured full-lengthcentral plug bodies26 Principal ow eld developments pre-dicted by these numerical simulations are again in a goodagreement with experimental data published in Ref 28 Withinthe frame of this ESA ARPT Program performance and owbehavior of clustered plug nozzles at different truncations arebeing examined by European industries [Societe Europeennede Propulsion (SEP) Volvo Dasa] and research institutes(ONERA DLR) with subscale cold- ow plug models26 (Table3) Numerous experimental investigations on subscale modelsof circular and clustered plug nozzles with cold- and hot-gas ow were performed at the Keldysh Center with and withoutexternal ows Thrust characteristics pressure and heat- uxdistributions along the plug and acoustic characteristics of thecircular jet were investigated

2 Linear Plug Nozzles

Performance behavior and ow eld development for linearplug nozzles as a function of ambient pressures are in principlesimilar to those of circular plug nozzles (Fig 17) Howeverspecial attention must be paid to the in uence of both endsides where the surrounding ow disturbs the expanding ow- eld resulting in an expansion of the ow normal to the main ow direction and therefore in an effective performance loss

HAGEMANN ET AL 629

Table 3 Design parameters for linear and toroidal subscale plug nozzlesa

Design data Linear plug (FESTIP) Circular clustered plug (ARPT)

Chamber pressure pc 875 bar 80 barDriving gas Air AirMass ow rate mCcedil 595 kgs 546 kgsDesign pressure ratio pc pamb 200 200Geometric area ratio laquo 1265 1265aUsed in the ESA FESTIP and ESA ARPT studies

Fig 20 Plug nozzle ow eld experiment vs numerical simula-tion a) toroidal plug nozzle experiment with gas-oilnitric-acidpropellant combination side view photograph and b) computa-tional results of toroidal plug nozzle Mach number distributionfull gray and isolines centerline section

Fig 21 Design of linear subscale plug nozzle of the ESA FESTIPTechnology Program (plug size 192 mm high 214 mm wide)

Fig 22 Calculated wall pressure data for linear subscale plugnozzle at different pressure ratios

The change of wake ow behavior may be strongly in uencedby the penetration of ambient pressures through both end sidesparticularly for truncated plug nozzles End plates as foreseenfor the linear plug nozzle of the X-33 demonstrator vehiclecould be used to avoid this ambient pressure penetration

Within the framework of the ESA FESTIP Technology Pro-gram lsquolsquoTechnology Developments on Rocket and Air Breath-ing Propulsion for Reusable Launch Vehiclesrsquorsquo subscale cold-gas tests and numerical simulations of linear plug nozzles arebeing performed by Dasa at the high-speed wind tunnel [Hoch-geschwindigheitswind kanal (HWK) in Bad Salzungen Ger-many] of the Technical University of Dresden719 Design pa-

rameters for the linear plug model are included in Table 3Figure 21 shows the baseline model with a linear throat Afterpassing the linear throat the ow expands directly onto theplug surface A second model will be tested with a bell-shapedlinear nozzle extension following the linear throat for internalexpansion prior to further external expansion onto the plugsurface19 The contour of the plug was designed by the methodof Angelino29 and three different plug lengths as 5 20 and40 of the ideal length were tested Figure 22 shows calcu-lated wall pressures for the full-length plug for different am-bient pressures simulating the operational range of sea-leveloperation up to higher-altitude operation The ow structure inform of the left- and right-running characteristics at two pres-sure ratios is shown in Fig 23 For lower pressure ratios thanthe design pressure ratio of pc pamb = 200 the in uence of the

630 HAGEMANN ET AL

Fig 23 Left- and right-running characteristics in ow eld at twopressure ratios pc pamb = a) 8 and b) 200 (inviscid analysis withmethod of characteristics)

Fig 24 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 20 plug body

Fig 25 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 40 plug body

Fig 27 Flow phenomena of an E ndash D nozzle with a) open and b)closed wakes

Fig 26 Sketch of an E ndash D nozzle

system of recompression shocks and expansion waves on thewall pressure distribution (Figs 17 and 22) can be seen bythe waviness of the pressure distribution on the plug surfaceFirst experimental results are shown in Figs 24 and 25 whichalso show the waviness of the pressure distribution for low-pressure ratios

3 Plug Nozzles Final Aspects

The performance of circular and linear plug nozzles is ad-versely affected by the external airstream ow ie the aspi-ration effect of the external ow slightly reduces performanceand favors an earlier change in wake- ow development for

truncated plug nozzles32ndash 35 Cold- ow tests showed that theeffect of external ow on performance is con ned to a narrowrange of external ow speeds with Mach numbers near unity

The altitude compensation capability of circular and linearplug nozzles for higher ambient pressures is indisputable Be-cause plug nozzles lose this capability for pressure ratios abovethe design pressure ratio the latter should be chosen as highas possible Taking this into account plug nozzles will featurean even better overall performance than those shown in Figs18 and 19 Unfortunately for equal geometrical area ratiosplug nozzles perform worse at high altitude than do conven-tional bell nozzles because of truncation and clustering Forhigh-area-ratio nozzles with relatively short lengths plug noz-zles perform better than conventional bell nozzles In additionto having excellent capabilities for altitude compensation plugnozzles have additional advantages including ease in vehicleand engine integration

C E ndash D Nozzles

An expansionndash de ection (E ndash D) nozzle is shown in Fig 26E ndash D nozzles were at one time thought to have capabilities foraltitude compensation because the gas expansion takes placewith a constant pressure free boundary Thus the aerodynamicbehavior of E ndash D nozzles as a function of altitude is in prin-ciple quite similar to plug nozzles because the expansion pro-cess is controlled by the ambient pressure hence by the alti-tude In contrast to plug nozzles however the expansionprocess is controlled from inside the nozzle for E ndash D nozzlesAt low altitude the higher ambient pressure limits the gasexpansion resulting in a low effective expansion area ratioThe exhaust gas is adapted to the ambient pressure level by

HAGEMANN ET AL 631

Fig 30 Performance losses in dual-throat nozzles as a functionof nozzle design parameters (with r4r1 = 352 r2r1 = 116 and mCcedil 2

= 00)

Fig 29 Flow phenomena during a) sea-level and b) high-altitudeoperations

Fig 28 Sketch of a dual-throat nozzle view of combustion cham-ber and throat region

systems of recompression and expansion waves (Fig 27) Athigher altitude the lower ambient pressure allows more gasexpansion within the nozzle resulting in a higher effectiveexpansion area ratio But in contrast to plug nozzles the pres-sure in the wake of the center plug is always less than theambient pressure because of the aspiration effect This occursat low pressure ratios when the wake is opened and results inan aspiration loss Furthermore because the exhaust ow ex-pands to this base pressure rather than to the ambient pressurelevel wall pressures downstream are overexpanded This re-sults in an additional overexpansion loss As the pressure ratioincreases the wake region closes and is thus totally isolatedfrom the ambient environment (see Fig 27) The behavior dur-ing transition from open wake to closed wake is again equalto plug nozzles and the base pressure in the closed-wake re-gion is essentially independent of the ambient pressure

The E ndash D nozzle concept has also been a subject of numer-ous analytical and experimental studies Results from thesestudies show that E ndash D nozzle capabilities for altitude com-pensation are poor and are in fact worse than those of plugnozzles because of aspiration and overexpansion losses35 Forhigh-area-ratio nozzles with a relatively short length an E ndash Dnozzle performs better than a comparable conventional bellnozzle at the same length because of lower divergence andpro le losses than the bell nozzle

The advantages of the E ndash D nozzle concept include its smallengine envelope and no moving parts However like toroidalplug nozzles E ndash D nozzles have the disadvantage of havinghigher throat heat uxes relative to a conventional bell nozzlewith an equal throat area The higher throat heat ux resultsagain from the relatively thin annular throat gap This problemhowever can be remedied with the modular thrust cell clusterconcept Additional advantages of E ndash D nozzles with clusteredthrust cells are the same as those already discussed for theclustered plug nozzle concept with regard to ease in manu-facturing thrust vector control by throttling or shutting off anindividual or group of thrust cells and lower nozzle throatheating

D Nozzles with Throat Area Varied by a Mechanical Pintle

This nozzle concept utilizes a conventional bell nozzle witha xed exit area and a mechanical pintle in the combustionchamber and throat region to vary the throat area and hencethe expansion area ratio The area of the nozzle throatmdashanannulus between the pintle and the shroudmdashis varied by mov-ing the pintle axially

The pintle concept has been used in solid rocket motors asa mean to provide variable thrust The concept in principleallows a continuous variation of the throat area and thus op-timum expansion area ratios throughout a mission Howeverit requires an actuator and a sophisticated control system Theconcept raises issues of engine weight design complexitycooling of the pintle and nozzle throat and reliability

In Ref 36 the aerodynamic performances of nozzles with ve different pintle geometries were calculated and comparedwith a reference bell nozzle The performance losses of thesepintle nozzles when compared with the bell nozzle are in therange of 1ndash 25 Performances of a xed pintle geometry atthree different locations were also calculated The results showthat the performance loss varies with the pintle location

E Dual-Mode Nozzles

Dual-mode rocket engines using one or two fuels offer atrajectory-adapted dual-mode operation during the ascent of alauncher which may be of signi cant advantage for single-stage Earth-to-orbit vehicles as compared to conventionalrocket engines with bell-type nozzles This engine concept in-volves the use of a dense propellant combination with mod-erate performance during liftoff to provide high thrust duringthe initial ight phase and a better performing propellant com-bination in vacuum which result in higher speci c impulse

The fuels are burned in two different combustion chamberswith one located completely inside the other in the case ofengines with dual-throat nozzles or with a conventional bellthrust chamber surrounded by an annular thrust chamber in thecase of dual-expander engines This type of engine has a built-in accelerationndash reduction capability achieved by shuttingdown one of two thrust chambers The total engine thrust isthen provided by the remaining thrust chamber with the useof the total nozzle exit area leading to an increase in speci cimpulse Apart from the indicated bene ts of dual-mode en-gines which will be discussed in more detail later their de-

632 HAGEMANN ET AL

Table 4 Thrust chamber parameters for simulated dual-expander nozzlesa

Operation mode

Aerojet

First (inout) Second (inout)

DLR

First (inout) Second (inout)

Propellants C3H8ndash O2H2ndash O2 mdashH2ndash O2 H2ndash O2H2ndash O2 mdashH2ndash O2

Chamber pressure pc 414207 bar mdash207 bar 200200 bar mdash200 barMixture ratio r 337 mdash7 77 mdash7Exit area ratio laquo 69 146 58 116aInnerouter chambers

Fig 32 Flow phenomena in a dual-expander nozzle during a)sea-level- and b) high-altitude operation

Fig 31 Sketch of a dual-expander nozzle view of combustionchamber and throat region

velopment and construction require considerable technologicaleffort

1 Dual-Throat Nozzles

A dual-throat nozzle con guration is shown in Fig 28 Itconsists of two conventional bell-thrust chambers with onelocated completely inside the other At low altitude the outerthrust chamber operates with the inner thrust chamber runningin parallel In this operation mode the engine has a larger throatproviding a moderate expansion area ratio During the missionthe outer thrust chamber is shut off and operation continueswith only the inner engine In this con guration ow fromthe inner engine expands and attaches supersonically to theouter engine resulting in a higher expansion area ratio for theremainder of the burn Flow phenomena in both operationmodes are included in Fig 29

Hot- red tests at Aerojet were conducted to provide heattransfer data that were very useful for the thermal analysis anddesign of the dual-throat nozzle con guration37 These testsshowed that ow separation occurred in the inner engine noz-zle at higher ratios of outer to inner chamber pressures duringthe rst operation mode with both chambers burning in par-allel The ow separation resulted in a higher heat load to theinner nozzle Subscale tests performed at the Keldysh Center3

have shown that the additional loss caused by the nozzle con-tour discontinuity during vacuum operation with active innerchamber is in the range of 08ndash 4 depending on geometricaldata (see Fig 30) This high-performance loss results from theinteraction of the inner chamber jet with the outer chambernozzle wall3 The decrease of the jet incidence angle on thewall by means of gas injection through the outer chamber re-duces this performance loss by 04ndash 073

2 Dual-Expander Nozzles

A dual-expander nozzle has two concentric thrust chambersand nozzles It consists of a conventional bell thrust chambersurrounded by an annular thrust chamber Both chambers haveshort primary nozzles which end in a common divergent noz-zle extension Figure 31 shows a typical dual-expander nozzlecon guration At low altitude both thrust chambers operatesharing the same exit area which results in a moderate expan-sion area ratio Part way into the mission one thrust chamberis shut off allowing the other nozzle to use the whole exitarea creating a high-expansion-area ratio for the remainder of

the burn In principle the two operation modes are comparableto those of dual-throat nozzles

Numerical simulations of the ow elds in dual-expandernozzles during all operation modes have been performed atAerojet38 DLR39 and the Keldysh Center Aerojet simulationsare based on an engine design using different propellantshydrocarbonoxygen for the inner chamber and hydrogenox-ygen for the outer chamber whereas DLR simulations arebased on a combined vehicleengine analysis using single-fuelsingle-mixture ratio dual-expander engines with hydrogenoxy-gen Table 4 summarizes combustion chamber parameters Flowphenomena observed in numerical simulations are shown in Fig32a for the mode 1 operation with both thrust chambers burn-ing Compression waves are induced near the inner nozzle lipas a result of the strongly inhomogeneous ow character at thepoint where both exhaust gases of inner and outer combustionchamber merge into the common divergent nozzle part Upand down variations of pressure ratios at the lip did not changethis wave formation even in cases of signi cantly lower exitpressures in the inner nozzle compared to the pressure eld ofthe outer nozzle at the lip Further downstream the compres-sion waves interact with the wall resulting in a re ection ofthe compression waves back into the ow eld

During mode 2 operation a strong expansion of the outernozzle ow is observed when the expansion ratio suddenlyincreases at the end of the nozzle lip The ow is directedtoward the axis of symmetry Near the centerline the ow thenturns over to the axial direction inducing a recompressionshock The static pressure rises signi cantly in this recom-pression region on the centerline A sub- and supersonic recir-culation zone establishes in the inner chamber of the dual-expander nozzle Figure 32b emphasizes the essential owpattern in this operation mode

These analyses have shown that dual-expander nozzles pro-duce high performance in both operation modes3839 Figure 33summarizes performance behavior as a function of ight alti-tude for the dual-expander engine simulated at DLR For com-parison performance data are included for two reference bell

HAGEMANN ET AL 633

Fig 33 Performance data for a dual-expander nozzle

nozzles that have the same area ratios as the dual-expandernozzle during its two operation modes Losses because of thenonhomogeneous exit ow and the induced shocks are com-parable with corresponding values of both conventional noz-zles1

Within the design process of dual-expander engines it is nec-essary to take into account that the nonburning chamber isexposed to high heat uxes and pressure oscillations duringmode 2 operation which might reach a high level and lead tostructural disintegration of the inner nozzle structure There-fore a mode-2 operation with bleed gas generation at a ratherlow thrust chamber pressure level in the inner chamber seemsto be advantageous Because of the bleed gas generation in theprimary chamber the overall mass ow rate during this alter-native mode-2 operation is increased In addition the reductionof the effective exit area ratio results in minor impulse deg-radations during this operation mode when compared with themode-2 operation with no bleed gas generation

Several analytical works on SSTO- and two-stage-to-orbit(TSTO)-vehicles using hydrogenpropane or hydrogenmeth-ane as fuels revealed the lowest vehicle dry masses for dual-mode engines in comparison to other engines1439 Other dual-mode engines using hydrogen as the single fuel but using twomixture ratios also revealed some bene ts over conventionalengines for SSTO- and TSTO applications Even a single-fueloperation with constant mixture ratios in both combustionchambers indicated a gain in launcher performance40

V ConclusionsSeveral nozzle concepts that promise gains in performance

over conventional nozzles were discussed in this paper in-cluding performance enhancements achieved by slight modi- cations of existing nozzles eg through cool gas injectioninto the supersonic nozzle part It is shown that signi cantperformance gains result from the adaptation of the exhaust ow to the ambient pressure and special emphasis was givento altitude adaptive nozzle concepts

A number of nozzle concepts with altitude-compensating ca-pability were identi ed and described To assist the selectionof the best nozzle concept for launch vehicle applications theperformance of the nozzles must be characterized This can bedone using computational uid dynamics (CFD) andor cold- ow tests Existing CFD methods that are in use in the aero-space industry and at research institutes have been veri ed fora wide number of sub- and full-scale experiments and providesuf ciently reliable performance determination for the differ-ent nozzle types

Theoretical evaluations numerical simulations and test re-sults showed that the different concepts have real altitude-com-pensating capabilities However the compensation capabilitiesare limited and there are some drawbacks associated with eachof the concepts Additional performance losses are induced inpractically all of these nozzle concepts when compared to anideal expansion mainly because of the nonisentropic effects

such as shock waves and pressure losses in recirculation zonesHowever these additional performance losses are less than1 ndash 3 depending on the different nozzle concepts

Four nozzle concepts have been selected for further evalu-ation under a NASA Aerojet cooperative agreement They in-clude a plug nozzle an E ndash D nozzle a dual-bell nozzle and adual-expander nozzle The nozzles in addition to a referencebell nozzle have been designed by Aerojet for cold ow testsCFD calculations performed at Aerojet demonstrated that allof these nozzles have high and approximately equal perfor-mances at the design points

In Europe special attention is being paid to plug- and dual-bell nozzles Different plug nozzle concepts are being testedin cold- ow tests within the ESA ARPT (clustered plug noz-zle) and ESA FESTIP (linear plug nozzle) contracts In addi-tion to experiments three-dimensional numerical simulationson these nozzle concepts are being performed Additional ex-periments on different nozzle concepts with forced- ow sep-aration devices eg dual-bell nozzles are currently plannedat DLR

Different advanced nozzle concepts have been examined andare under examination now at the Keldysh Center with bothCFD calculations and experiments on the Nozzle DifferentialFacility which has a nozzle performance determination errorof about 005 This high accuracy is achieved by simulta-neously running the test nozzle and a reference nozzle bothmounted on a common longitudinal axis The performancecharacteristics of the reference nozzle are well known and thethrust difference is measured with strain gauges This facilityis equipped with a wind tunnel to determine nozzle thrust per-formance in the presence of external transonic ows withMach numbers up to 12

In addition to aerodynamic performance other technical is-sues (weight cost design thermal management manufactur-ing system performance and reliability) must be addressedFurthermore before a nal decision can be made as to whichnozzle concept offers the greatest bene ts with regard to aneffective payload mass injection combined launcher and tra-jectory calculations must be performed and compared to a ref-erence launcher concept with conventional nozzles Differentnozzle ef ciencies which account for the additional losses ofadvanced rocket nozzles and are extracted from numerical sim-ulations and experiments must be taken into account

At least for some of these nozzle concepts the plug nozzlethe dual-bell nozzle and the dual-expander nozzle bene tswith regard to lower overall launcher masses have been dem-onstrated in the literature Furthermore the plug nozzle con-cept will be the rst in ight of all of these advanced nozzleconcepts with the X-33 demonstrator vehicle

AcknowledgmentPresented as Paper 18 at the 3rd International Symposium

on Space Propulsion Beijing Peoplersquos Republic of ChinaAug 11ndash 13 1997

References1Manski D and Hagemann G lsquolsquoIn uence of Rocket Design Pa-

rameters on Engine Nozzle Ef cienciesrsquorsquo Journal of Propulsion andPower Vol 12 No 1 1996 pp 41ndash 47

2lsquolsquoLiquid Rocket Engine Nozzlesrsquorsquo NASA Space Vehicle DesignCriteria NASA SP-8120 1976

3Dumnov G E Nikulin G Z and Ponomaryov N B lsquolsquoInves-tigation of Advanced Nozzles for Rocket Enginesrsquorsquo Space Rocket En-gines and Power Plants Vol 4 No 142 NIITP 1993 pp 1012 ndash

1018 (in Russian)4Nguyen T V and Pieper J L lsquolsquoNozzle Flow Separationrsquorsquo Pro-

ceedings of the 5th International Symposium of Propulsion in SpaceTransportation (Paris France) 1996

5Manski D lsquolsquoClustered Plug Nozzles for Future European Reus-able Rocket Launchersrsquorsquo German Aerospace Research Establishment643-817 Lampoldshausen Germany 1981 (in German)

6Horn M and Fisher S lsquolsquoDual-Bell Altitude Compensating Noz-

HAGEMANN ET AL 621

Table 1 Performance losses in conventional rocket nozzlesa

LossesVulcain 1

SSME

Chemical nonequilibrium 02 01Friction 11 06Divergence nonuniformity of exit ow 12 10Imperfections in mixing and combustion 10 05Nonadapted nozzle ow 0ndash 15 0ndash 15aOther loss sources also shown in Fig 1

Fig 1 Flow phenomena and loss sources in rocket nozzles

Fig 2 Rocket nozzle ow elds during off-design operation a)overexpanded ow RL10A-5 engine and b) underexpanded owSaturn-1B Apollo-7 (Photographs United Technologies Pratt ampWhitney NASA)

Fig 4 Flow phenomena for a conventional rocket nozzle

Fig 3 Performance data for nozzle of Vulcain 1 engine (designparameters of Vulcain 1 nozzle laquo = 45 pc pamb = 555 r = 589)

The main part of the paper addresses different nozzle conceptswith improvements in performance as compared to conventionalnozzles achieved by altitude adaptation and thus by minimizinglosses caused by over- or underexpansion Several concepts forthe altitude adaptation of rocket nozzles exist in the literature andthese are considered in this paper in more detail2ndash 4 This discus-sion of advanced nozzles includes nozzles with inserts for con-trolled ow separation two-position nozzles dual-bell nozzlesdual-expander and dual-throat nozzles expansionndash de ection noz-zles and plug nozzles In the past practically all of these conceptshave been the subject of analytical and experimental work5ndash7

Although bene ts in performance were indicated in most of theavailable publications none of these nozzle concepts has yet beenused in existing rocket launchers This may change in the nearfuture because a rocket engine with a linear plug nozzle is fore-seen as the propulsion system for the Lockheed Martin liftingbody RLV X-33 concept

II Conventional NozzlesConventional bell-type rocket nozzles which are in use in

practically all of todayrsquos rockets limit the overall engine per-

622 HAGEMANN ET AL

Fig 5 Flow separation in overexpanding rocket nozzles wall pressure pro le and phenomenology

formance during the ascent of the launcher owing to their xedgeometry Signi cant performance losses are induced duringthe off-design operation of the nozzles when the ow is over-expanded during low-altitude operation with ambient pressureshigher than the nozzle exit pressure or underexpanded duringhigh-altitude operation with ambient pressures lower than thenozzle exit pressure Figure 2 shows photographs of nozzleexhaust ows during off-design operation In the case of over-expanded ow oblique shocks emanating into the ow eldadapt the exhaust ow to the ambient pressure Further down-stream a system of shocks and expansion waves leads to thecharacteristic barrel-like form of the exhaust ow In contrastthe underexpansion of the ow results in a further expansionof the exhaust gases behind the rocket

Off-design operations with either overexpanded or under-expanded exhaust ow induce performance losses Figure 3shows calculated performance data for the Vulcain 1 nozzle asfunction of ambient pressure together with performance datafor an ideally adapted nozzle Flow phenomena at differentpressure ratios pc pamb are included in Fig 4 [The sketch with ow phenomena for the lower pressure ratio pcpamb shows anormal shock (Mach disk) Depending on the pressure ratiothis normal shock might not appear see eg Fig 2] TheVulcain 1 nozzle is designed in such a manner that no uncon-trolled ow separation should occur during steady-state oper-ation at low altitude resulting in a wall exit pressure of pwe

rsquo 04 bar which is in accordance with the Summer eld cri-terion8 The nozzle ow is adapted at an ambient pressure ofpamb rsquo 018 bar which corresponds to a ight altitude of h rsquo15000 m and performance losses observed at this ambientpressure are caused by internal loss effects (friction diver-gence mixing) as shown in Fig 1 and Table 1 Losses inperformance during off-design operations with over- or un-derexpansion of the exhaust ow rise up to 15 In principlethe nozzle could be designed for a much higher area ratio toachieve better vacuum performance but the ow would thenseparate inside the nozzle during low-altitude operation withan undesired generation of side-loads

A Flow Separation and Side-Loads

Flow separation in overexpanding nozzles and its theoreticalprediction have been the subject of several studies in thepast348 and different physical models and hypotheses for theprediction of ow separation have been developed In stronglyoverexpanding nozzles the ow separates from the wall at acertain pressure ratio of wall pressure to ambient pressure pwpamb The typical structure of the ow eld near the separationpoint is shown in Fig 5 together with wall pressure dataSeparation and the formation of a recirculation zone at the wall

induce an oblique shock wave near the wall which leads to arecompression of the ow

The physical phenomenon of ow separation can be dividedinto two simple phenomena3 The rst is the turbulent bound-ary-layer separation from the nozzle wall which is character-ized by the ratio of the nozzle wall pressure just after theseparation pp to the nozzle wall pressure just before separa-tion psep This pressure ratio is referred to as the critical pres-sure ratio pcr = pseppp = p1p2 The second phenomenon isconnected with the ow in the separated zone which is char-acterized by a minor pressure gradient along the wall Theanalysis of model experimental data on separated turbulent su-personic ows shows that the pressure ratio pcr is equal forseparated ows before an obstacle and for separated nozzle ows3 For turbulent ows pcr shows a slight dependence onReynolds number but depends strongly on Mach number Fur-thermore investigations of separated ows in rocket enginenozzles showed that pcr is also a function of wall temperaturegas composition and nozzle wall roughness

Various approaches to the prediction of ow separation areused in industry and research institutes For example the Kel-dysh Centerrsquos method for the determination of the ow sepa-ration point in rocket nozzles is based on empirical relation-ships obtained for pcr pppwe and pwepamb Using theserelations the pressure ratio pseppamb the separation point andthe wall pressure distribution are determined Aerojet uses tab-ulated data for the prediction of ow separation from variousexperiments with conical and bell-shaped nozzles4 At a knownpressure ratio of chamber pressure to ambient pressure a lowerand an upper separation pressure ratio follows from the data-base that is different for conical and bell-shaped nozzles Aseparation criterion used at the European industry and scien-ti c institutes describes the pressure ratio pseppamb as functionof the local ow Mach number at the wall near the separationpoint This analytical model was derived from various cold-and hot- ring tests of overexpanding nozzles9

Various numerical simulations of ow problems featuring ow separation have shown that the numerical prediction of ow separation with state-of-the-art turbulence modeling re-sults in good agreement with experimental data310 As an ex-ample Figs 6 and 7 compare experimental and numerical dataon wall pressure ratios in two overexpanding nozzles Fur-thermore former experiments of separated ows indicated thatthere is a random movement of the separation line in over-expanding rocket nozzles91112 resulting in the possible gen-eration of side loads The correct prediction of side-loads withmodels is uncertain and still a subject of on-going researchCommon side-load models described in Refs 4 and 9 assumethat the separation line in the nozzle has a maximum tilt angle

HAGEMANN ET AL 623

Fig 6 Experimental and numerical data on wall pressure ratiosin overexpanding conical nozzles with ow separation Cold-gastest case with subscale conical nozzle

Fig 7 Experimental and numerical data on wall pressure ratiosin overexpanding bell nozzles with ow separation Hot-gas testcase with J-2S engine

with the centerline The minimum and maximum separationpoints are predicted using the upper and lower separation datathat result from the applied separation model In the case ofdistinct separation points as obtained with the Summer eld- orSchmucker-criterion a certain scattering around the predictedseparation point is assumed The side-load is then calculatedby integrating the wall pressure in the asymmetrical ow sep-aration region49 These approaches gave reasonable results forthe J2-S and Vulcain 1 nozzle

At the Keldysh Center in Russia model experiments of sep-arated ow effects on side-loads acting on nozzles began inthe early 1980s11 These investigations were directed mainlyto the determination of the unsteady pressure uctuations inthe separation zone on the whole engine The method is de-scribed in Ref 11 in more detail it is based on a generalizationof obtained empirical data for the spectral density of the pul-sating pressure eld in the separation zone Distribution of thepressure amplitudes is governed by probability density func-tions This method was validated with the experimental rms-val-ues of side-loads acting on the RD-0120 nozzle (CABD hydro-genndash oxygen engine rst-stage engine of Russian Energialauncher) Other hypotheses for the appearance and predictionof side-loads also exist eg based on aeroelastic analyses13

which account for the coupling of ow oscillations with thinnozzle shells However these hypotheses require experimentalveri cation and validation

Side-loads are an undesired phenomenon that may result inthe destruction of the rocket nozzle and should therefore beavoided In Ref 9 the destruction of a J-2D engine (Rocket-dyne hydrogen ndash oxygen engine second- and third-stage engineof Saturn-5 launcher) as a result of side-loads is reported The

transient startup of the rocket engine generates a nozzle owwith uncontrolled ow separation and with possible side-loadgeneration in the nozzle whereas maximum area ratios of all rst-stage nozzles or booster nozzles are chosen to avoid owseparation at the nominal chamber pressure operation As aresult the vacuum performance of rocket engines that operateduring the entire launcher trajectory such as the SSME orVulcain 1 engines is limited

B Potential Performance Improvements

Compared to existing rocket engines a gain in performanceis achieved with advanced engines such as mixed-mode pro-pulsion systems dual-mixture ratio engines or dual-expanderengines Nevertheless the upgrade of existing engines withbetter performing subsystems such as turbines and pumpsalso leads to a gain in overall performance data and is dis-cussed in more detail in Ref 14 Nozzle performance of con-ventional rocket engines is already very high with regard tointernal loss effects (friction nonuniformity) However fornozzles of gas-generator open-cycle engines such as the Vul-cain 1 engine a slight improvement in performance can beachieved with turbine exhaust gas (TEG) injection into themain nozzle as realized in the F-1 (kerosenendash oxygen engine rst stage of Saturn-5 launcher) and J-2S and it is foreseenfor the Vulcain 2 engine (hydrogen ndash oxygen engine upgradeof Vulcain 1 engine) and con rmed by numerical simula-tions131516 and experimental results3 This is mainly achievedthrough lower friction losses in the main nozzle1315 and be-cause the bypass nozzles used for the expansion of the TEGswhich have higher divergence losses are removed

Despite the slight performance gain by TEG injection thelow-pressure near-wall stream of the injected gas favors a re-duction of the critical pressure ratio at which ow separationoccurs and therefore an earlier nozzle ow separation312 Fur-thermore the presence of the secondary exhaust gas injectioncomplicates nozzle-contouring methods with regard to theavoidance of uncontrolled ow separation for rst-stage orbooster nozzles

III Altitude Adaptive NozzlesA critical comparison of performance losses shown in Table

1 reveals that most signi cant improvements in nozzle perfor-mance for rst-stage or SSTO engines can be achieved throughthe adaptation of nozzle exit pressures to the variations in am-bient pressure during the launcherrsquos ascent through the atmo-sphere Various concepts have been previously mentioned andwill be discussed in detail in the following text

A Nozzles with Devices for Controlled Flow Separation

Several nozzle concepts with devices for controlled owseparation have been proposed in the literature with primaryemphasis on the reduction of side-loads during sea level orlow-altitude operation But the application of these conceptsalso results in an improved performance through the avoidanceof signi cant overexpansion of the exhaust ow

1 Dual-Bell Nozzle

This nozzle concept was rst studied at the Jet PropulsionLaboratory17 in 1949 In the late 1960s Rocketdyne patentedthis nozzle concept which has received attention in recentyears in the US and Europe671819 Figure 8 illustrates thedesign of this nozzle concept with its typical inner base nozzlethe wall in ection and the outer nozzle extension This nozzleconcept offers an altitude adaptation achieved only by nozzlewall in ection In low altitudes controlled and symmetrical ow separation occurs at this wall in ection (Fig 9) whichresults in a lower effective area ratio For higher altitudes thenozzle ow is attached to the wall until the exit plane and thefull geometrical area ratio is used Because of the higher arearatio an improved vacuum performance is achieved Howeveradditional performance losses are induced in dual-bell nozzles

624 HAGEMANN ET AL

Fig 8 Sketch of a dual-bell nozzle

Fig 9 Flow eld phenomena in dual-bell nozzles a) sea-levelmode with ow separation at the wall in ection point and b) al-titude mode with a full- owing nozzle

Fig 10 Performance data of a dual-bell nozzle Performance iscompared with two baseline bell-type nozzles as function of ightaltitude (baseline nozzle 1 same area ratio as dual-bell base noz-zle baseline nozzle 2 same area ratio as nozzle extension)

Fig 11 Performance characteristics of a dual-bell nozzle Per-formance is compared with two baseline bell-type nozzles as func-tion of ight altitude (baseline nozzle 1 same area ratio as dual-bell base nozzle baseline nozzle 2 same area ratio as nozzleextension)

as compared with two baseline nozzles having the same arearatio as the dual-bell nozzle at its wall in ection and in its exitplane Figures 10 and 11 illustrate the performance of a dual-bell nozzle as a function of ight altitude in comparison withboth baseline bell-type nozzles (Design parameters of thedual-bell nozzle are taken from a launcher analysis publishedin Ref 18 Propellants hydrogenoxygen r = 6 pc = 200 barrt = 007 m wall in ection at area ratio laquoB = 30 and totalarea ratio laquoE = 100) The pressure within the separated owregion of the dual-bell nozzle extension at sea-level operationis slightly below the ambient pressure inducing a thrust lossreferred to as lsquolsquoaspiration dragrsquorsquo In addition ow transitionoccurs before the optimum crossover point which leads tofurther thrust loss as compared to an ideal switchover Thenonoptimum contour of the full owing dual-bell nozzle re-sults in further losses at high altitudes

To gain insight into the performance and ow behavior ofdual-bell nozzles at different ambient pressures extensive nu-merical simulations with parametrical variations of contour de-sign parameters were performed18 An optimized bell nozzlewith equal total length and area ratio was used as the referencenozzle for comparison As a result the vacuum performance ofthe dual-bell nozzles has a degradation because of the imper-fect contour and this additional loss has the same order ofmagnitude as the divergence loss of the optimized bell nozzle

The simulations of sea-level operation also revealed an ad-ditional performance loss because of aspiration drag which isless than 3 for these dual-bell nozzles This additional lossdepends linearly on the ambient pressure and therefore it isreduced during the ascent of the launcher Furthermore thesesimulations showed that the application of commonly usedseparation criteria derived for conventional nozzles and ap-

plied to dual-bell nozzles with their wall in ection to estimatethe critical pressure ratio yields reasonable results with an ac-curacy of rsquo15 Experimental data obtained at the KeldyshCenter revealed that the critical pressure ratio at the wall in- ection is about 15ndash 20 less than the critical pressure ratioof a conventional nozzle and that usual separation criteriamust be corrected for an accurate prediction of the critical wallpressure ratio

Flow transition behavior in dual-bell nozzles strongly de-pends on the contour type of the nozzle extension3618 A sud-den transition from sea-level to vacuum operation can be atleast theoretically achieved by two different extensions witha zero wall pressure gradient (constant pressure extension) ora positive wall pressure gradient (overturned extension) But acritical analysis of the transition behavior considering decreas-ing ambient pressures during the launcher ascent revealed thata considerable time with uncontrolled ow separation withinthe nozzle extension exists even for these types of extensionsThe duration of this period can be reduced drastically by throt-tling the chamber pressure18

The main advantage of dual-bell nozzles as compared toother means of controlling nozzle ow separation is its sim-plicity because of the absence of any movable parts and there-fore its high reliability It is necessary to note that the external ow over the vehicle in ight reduces the pressure in the ve-hicle base region where engines are installed The ambient

HAGEMANN ET AL 625

Fig 13 Principle performance characteristics of the nozzle withinsert

Fig 12 RD-0120 nozzle hardware with removed nozzle insertand sketch of secondary nozzle mounted inside of the RD-0120nozzle (photographs taken from Ref 23)

pressure triggering the ow transition is the vehicle base pres-sure instead of the atmospheric pressure at the speci c ightaltitude As the base pressure is lower than the atmosphericpressure the nozzle ow transition occurs at a lower altitudethan the one showed in Fig 11 which slightly decreases theef ciency of the dual-bell nozzle operation along the trajectory

Despite the additional losses induced in dual-bell nozzlesthey still provide a signi cant net impulse gain over the entiretrajectory as compared to conventional bell nozzles Indepen-dent combined launcher and trajectory analyses performed byDasa within the European Space Agency Future EuropeanSpace Transportation Investigations Program (ESA FESTIP)study19 and at DLR on SSTO vehicles powered with dual-bellnozzles result in a signi cant payload gain when comparedwith a reference launcher equipped with conventional nozzles

2 Nozzles with Fixed Inserts

A trip ring attached to the inside of a conventional nozzledisturbs the turbulent boundary layer and causes ow separa-tion at higher ambient pressures At higher altitudes with lowerambient pressures the ow reattaches to the wall behind thetrip ring and full owing of the nozzle is achieved The tran-sition from sea level to vacuum mode depends on the wallpressure near the trip-ring location and on the disturbance in-duced by the trip ring The size of the trip ring is a compromisebetween stable ow separation during sea-level operation andthe induced performance loss during vacuum operation In Ref9 it is reported that a trip-ring size of 10 of the local bound-ary-layer thickness is suf cient to ensure stable ow separa-tion

In principle this concept is similar to the dual-bell nozzleconcept with regard to performance characteristics as shownin Fig 11 The sea-level performance of this nozzle concept islower than the performance of a conventional bell nozzle trun-cated at the trip-ring location because of the aspiration dragin the separated ow region of the nozzle Furthermore at sealevel the bell nozzle with trip rings has even higher divergencelosses than a comparable dual-bell nozzle because the nozzlecontour upstream of the obstacle differs from the optimal con-tour for this low-area ratio as a result of the bell nozzle designfor best vacuum performance The additional losses inducedduring vacuum operation is about 1 compared with the per-formance of the bell nozzle without an obstacle Thus theadditional losses are comparable to the additional losses in-duced in dual-bell nozzles As for dual-bell nozzles the tran-sition behavior of this nozzle concept is uncertain but it willbe even more uncertain than for a dual-bell nozzle with a con-stant pressure or overturned nozzle extension

In principle several altitude adaptations can be achievedwith one nozzle by various trip rings mounted one behind theother However this results in increasing vacuum performancelosses The trip rings can also be attached into existing nozzlesand therefore represent a low-cost concept at least for testpurposes with low technological risk Trip rings have beendemonstrated to be effective for side-load reductions duringthe transient startup of rocket engines20 The main problemswith trip-ring nozzles are not only performance losses but alsoring resistance in high-temperature boundary layers the exactcircumferential xing and the uncertainties in the transitionbehavior These uncertainties might be why active interest inthis nozzle concept in the 1970s which is documented in var-ious publications920ndash22 has disappeared in recent years

A further concept with a xed wall discontinuity is a nozzlewith a circumferential groove The aerodynamic behavior of anozzle with a groove and nozzle with trip rings are quite sim-ilar9 but this concept is hard to realize in the case of thinnozzle shells which are used in all of todayrsquos rocket nozzles

3 Nozzles with Temporary Inserts

Nozzle concepts with xed wall discontinuities have the dis-advantage of lower vacuum performance as compared with a

conventional bell nozzle with equal design and operation dataA promising concept for controlled ow separation is thereforetemporary inserts which are removed for vacuum operationThese inserts can be either ablative or ejectible The insertsmay have the form of a complete secondary nozzle23 (Fig 12)or of small steps attached inside the nozzle wall In case ofejectible inserts a reliable mechanism is needed to provide asudden and symmetrical detachment In any case shocks areinduced during the transient ejection because the inserts act asan obstacle in the supersonic exhaust ow These shocks alsointeract with the nozzle walls and increase pressure loads onthe wall and local heat uxes A nonsymmetrical ejectionwould then result in the generation of side-loads Furthermorethe danger of a downstream collision with the nozzle wallarises because the inserts might also experience a transversalmovement toward the walls

Recent hot- ring tests performed in Russia with a modi edRD-0120 engine equipped with a secondary nozzle insert re-vealed a signi cant performance gain of 12 during the sea-level operation at 100 chamber pressure compared with theoriginal RD-0120 performance23 Nominal chamber pressureof this engine is pc = 206 bar with an area ratio of laquo = 857These full-scale hot- ring tests demonstrated the durability ofmaterials sealings and the release mechanism and thus thefeasibility of this concept Figure 12 shows the nozzle hard-ware and a sketch of the secondary nozzle mounted inside theRD-0120 nozzle23

The principle performance characteristics of this RD-0120nozzle with ejectible insert are included in Fig 13 The nozzleoperation with insert results in a slight performance loss com-pared with an ideal bell nozzle with the same reduced area

626 HAGEMANN ET AL

Fig 14 Sketch of a two-position nozzle during a) sea-level andb) vacuum operations

ratio because of aspiration drag and the presumably nonoptim-ized insert contour The performance degradation is compara-ble to the one induced in dual-bell nozzles during sea-leveloperation (Fig 11) In vacuum or high-altitude operations thehigher performance of the baseline nozzle is achieved

Another method for removing the inserts is to use combus-tible or ablative elements924 During the ascent of the launcherthe size of the insert is continuously reduced until it is com-pletely consumed resulting in a full owing bell nozzle witha clean contour for best vacuum performance The principaluncertainties of this nozzle concept are the stability and surfaceregression rates of the inserts Furthermore a homogeneoussymmetrical and temporally de ned consumption must beguaranteed despite possible local pressure and temperature uctuations near the nozzle walls This is currently highly un-certain

4 Nozzles with Active or Passive Secondary Gas Injection

With this nozzle concept the ow in an overexpanding noz-zle is forced to separate at a desired location by injecting asecond uid into the gas stream in the wall normal directionto achieve maximum disturbance of the main exhaust gas andthus to induce ow separation This secondary gas injectionis quite different from the aforementioned TEG injectionwhich should be injected with few or no disturbances to themain ow The injection could be either active that is forcedthrough gas expansion from a higher-pressure reservoir or pas-sive using holes in the wall through which ambient gas issucked in (vented nozzle concept) The latter concept can onlywork in case of gas pressures near the wall inside the nozzlethat are lower than the ambient pressure Experience on forcedsecondary gas injection gained at Aerojet shows that a largeamount of injected uid is required to induce a signi cant owseparation Furthermore no net speci c impulse gain is real-ized when considering the additional gas ow rate

In the vented nozzle concept segments of the bell nozzlewall have various slots or holes opened to the outside ambientpressure During low-altitude operation the slots are opened totrigger ow separation within the nozzle whereas the slots areclosed during high-altitude or vacuum operation so that thegas fully expands inside the entire nozzle Experiments with amodi ed RL10A-3 engine equipped with this nozzle conceptwere performed at Pratt amp Whitney25 Performance resultsshowed that over a small range of low-pressure ratios the per-forated nozzle performed as well as a nozzle with its area ratiotruncated immediately upstream of the vented area Howeverat above some intermediate pressure ratios the thrust ef -ciency suddenly dropped and approached that of the full- ow-ing nozzle The measured performance characteristics werequite similar to those of the dual-bell nozzles shown in Fig11

The altitude range of this nozzle concept is limited by thenumber and position of the holes because the pressure withinthe nozzle must be lower than the ambient pressure Further-more as the rocket aft-body base pressure is lower than thealtitude atmospheric pressure in the surrounding ambient ows the nozzle ow transition occurs at a lower altitude andthus the range of compensation is further reduced

5 Two-Position or Extendible Nozzles

Nozzles of this type with extendible exit cones are currentlyused only for rocket motors of upper stages to reduce the pack-age volume for the nozzle eg at present for solid rocketengines such as the inertial upper stage (IUS) or for the liquidrocket engine RL10 The main idea of the extendible extensionis to use a truncated nozzle with low expansion in low- ightaltitudes and to have a higher nozzle extension at high alti-tudes Figure 14 illustrates this nozzle concept Its capabilityfor altitude compensation is indisputable and the nozzle per-formance is easily predictable The whole nozzle contour in-cluding the extendible extension is contoured for maximum

performance at the high-area ratio with either the method ofcharacteristics or a variational method The area ratio of the rst nozzle section is then determined and the nozzle contouris divided into two parts The xed nozzle part and the ex-tendible extension Investigations conducted at the KeldyshCenter have shown that this nozzle-contouring method is notonly the simplest but also provides a good overall trajectoryperformance3 A minor performance loss is incorporated duringlow-altitude operations because of the truncated inner nozzlewhich has a nonoptimal contour for this interim exit area ratioThe performance characteristics as a function of ight altitudeare similar to those of the nozzle with the ejectible insert asshown in Fig 13 A tradeoff study performed at the KeldyshCenter on nozzles with an ejectible insert and an extendibleextension showed that the thrust characteristics of the nozzlewith an ejectible insert are slightly better than those of theextendible nozzle concept with the same overall dimensionsfor both nozzles

The main drawback of the extendible nozzle concept is thatit requires mechanical devices for the deployment of the ex-tension which reduces engine reliability and increases totalengine mass The necessity for active cooling of the extendibleextension requires exible or movable elements in the coolingsystem which also reduces system reliability The in uencesof the external ow and the generated jet noise on the extend-ible extension in the retracted initial position have not yetbeen fully investigated Former investigations conducted in theKeldysh Center showed that the external ow causes bothsteady and unsteady pressure loads on the retracted nozzle ex-tension whereas the engine jet noise causes strong vibrationsof the nozzle extension

B Plug Nozzles

Various experimental analytical and numerical research onplug nozzles have been performed since the 1950s in theUS4532ndash 35 Europe5710192627 Russia3 and Japan28 In contrastto the previously discussed nozzle concepts plug nozzles pro-vide at least theoretically a continuous altitude adaptation upto their geometrical area ratio Figures 15 and 16 show twodifferent design approaches for circular plug nozzles whichdiffer only in the chamber and primary nozzle layout A con-ical central body is shown here which could also be designedwith more sophisticated contouring methods2930 Different de-sign approaches include plug nozzles with a toroidal chamberand throat (with and without truncation) and plug nozzles witha cluster of circular bell nozzle modules or with clusteredquasirectangular nozzle modules The latter approach seems to

HAGEMANN ET AL 627

Fig 17 Flow phenomena of a plug nozzle with full length (left column) and truncated central body (right column) at different pressureratios pc pamb off-design (top bottom) and design (center) pressure ratio

Fig 16 Principle design of plug nozzles clustered plug 36 mod-ules with truncated plug body

Fig 15 Principle design of plug nozzles toroidal plug fulllength

be advantageous because further losses induced by the gapsbetween individual modules and the ow eld interactionsdownstream of the module exits can be minimized It has beenshown that transition from a round to a square nozzle resultsin a very small performance loss31 In principle the ow elddevelopment of a clustered plug nozzle with rectangular nozzlemodules is similar to that of a toroidal plug nozzle but avoidsthe inherent disadvantages of the toroidal plug design regard-ing 1) the control of a constant throat gap during manufactur-

ing and thermal expansion (side-loads and thrust vector devi-ations) 2) the cooling of toroidal throat with tiny throat gapsand 3) the control of combustion instabilities in the toroidalcombustion chamber Another plug nozzle con guration is thelinear plug nozzle which is foreseen for the propulsion systemof the RLV X-33 concept

1 Circular Plug Nozzles

Figure 17 summarizes the principle ow phenomena of cir-cular plug nozzles with full length and truncated central bodiesat different off-design (top and bottom) and design (center)pressure ratios that were observed in experiments and numer-ical simulations For pressure ratios lower than the design pres-sure ratio of a plug nozzle with a well-contoured central bodythe ow expands near the central plug body without separa-tion and a system of recompression shocks and expansionwaves adapts the exhaust ow to the ambient pressure pambThe characteristic barrel-like form with several in ections ofthe shear-layer results from various interactions of compres-sion and expansion waves with the shear layer and turbulentdiffusion enlarges the shear layer farther downstream of thethroat The existence of the overexpansion and recompressionprocesses is inferred from up- and down-variations of plugwall pressure pro les observed in various cold- ow tests andnumerical simulations and will also be shown later for linearplug nozzles

At the design pressure ratio (see Fig 17 left column cen-ter) the characteristic with the design Mach number should bea straight line emanating to the tip of the central plug bodyand the shear layer is parallel to the centerline However forcircular plug nozzles designed with contouring methods pro-posed in Refs 29 and 30 no exact one-dimensional exit owpro le can be achieved because both methods use thePrandtlndash Meyer relations that are only valid for planar owsFurthermore nonhomogeneous ow in the throat regionwhich is in general not considered within the contour designprocess also in uences the exit ow pro le The wall pressuredistribution remains constant at pressure ratios above the de-sign pressure ratio ie the plug nozzle behaves like a con-ventional nozzle the loss of its capability of further altitude

628 HAGEMANN ET AL

Table 2 Design parameters of toroidal plug nozzles

Design data Full-scale nozzlea Subscale nozzleb

Chamber pressure pc 100 bar 12 barPropellants Hydrogenoxygen Gasndash oilnitric-acidMixture ratio r 60 25Inner chamber diameter dcinner 43 m 019 mOuter chamber diameter dcouter 44 m 034 mGeometric area ratio laquo 55 10aUsed for numerical simulations and performance predictions see Fig 18

bUsed in experiments see Fig 20

Fig 18 Performance of numerically simulated plug nozzle withfull-length central body

Fig 19 Performance of numerically simulated plug nozzle withtruncated central body

adaptation being included Figure 17 (left column bottom) il-lustrates the ow eld at higher pressure ratios Performancedata of a typical plug nozzle are included in Fig 18 and com-pared to a conventional bell nozzle with equal area ratio Thedesign and combustion chamber parameters are included in theleft column of Table 227

The truncation of the central plug body which is an advan-tage because of the huge length and high structural mass ofthe well-contoured central body results in a different ow andperformance behavior as compared to the full-length plug noz-zle At lower pressure ratios an open wake ow establisheswith a pressure level practically equal to the ambient pressure(Fig 17 right column top) At a certain pressure ratio closeto the design pressure ratio of the full-length plug nozzle thebase ow suddenly changes its character and turns over to theclosed form characterized by a constant base pressure that isno longer in uenced by the ambient pressure Analyses indi-cate that shorter plug bodies with higher truncations trigger anearlier change in wake ow at pressure ratios below the designpressure ratio At the transition point the pressure within thewake approaches a value that is below ambient pressure andthe full base area induces a negative thrust (Fig 17 right col-umn center) This thrust loss depends on the percentage oftruncation and the total size of the base area Published ex-perimental data and numerical simulations reveal an increasingthrust loss for shorter plug bodies because the total base areaincreases

Beyond the transition point the pressure within the closedwake remains constant At these lower ambient pressures thebase pressure is then higher than the ambient pressure result-ing in a positive thrust contribution of the total base area Per-formance data of a numerically simulated truncated plug noz-zle are shown in Fig 19 and compared to the same plugnozzle with full-length central body and a conventional bellnozzle Design parameters for this truncated plug nozzle arethe same as for the full length plug (see Fig 18 and Table 2)

Figure 20 shows a typical photograph of a sea-level hot-runtest with a truncated toroidal subscale plug nozzle performedat DLR5 Nozzle design data are included in Table 2 For com-parison of experimental results with numerical computations

the ow eld of the toroidal plug nozzle was calculated with anumerical method10 and Fig 20 shows the calculated Mach-number distribution in the combustion chamber and nozzlePrincipal physical processes like expansion waves shocks andthe recirculating base- ow region are in good agreement Boththe experiment and the numerical simulation show that the ow separates from the conical plug body before reaching itstruncated end Recent results on experiments with plug nozzlespublished in Ref 28 also reveal separation of the ow fromthe central plug body for conical contours In contrast no sep-aration was observed for contoured central plug bodies de-signed with the method proposed in Ref 29 Recent numericalsimulations of contoured plug nozzles performed at DLRwithin the ESA Advanced Rocket Propulsion Technology(ARPT) Program on advanced rocket propulsion also showthat no ow separation occurs from well-contoured full-lengthcentral plug bodies26 Principal ow eld developments pre-dicted by these numerical simulations are again in a goodagreement with experimental data published in Ref 28 Withinthe frame of this ESA ARPT Program performance and owbehavior of clustered plug nozzles at different truncations arebeing examined by European industries [Societe Europeennede Propulsion (SEP) Volvo Dasa] and research institutes(ONERA DLR) with subscale cold- ow plug models26 (Table3) Numerous experimental investigations on subscale modelsof circular and clustered plug nozzles with cold- and hot-gas ow were performed at the Keldysh Center with and withoutexternal ows Thrust characteristics pressure and heat- uxdistributions along the plug and acoustic characteristics of thecircular jet were investigated

2 Linear Plug Nozzles

Performance behavior and ow eld development for linearplug nozzles as a function of ambient pressures are in principlesimilar to those of circular plug nozzles (Fig 17) Howeverspecial attention must be paid to the in uence of both endsides where the surrounding ow disturbs the expanding ow- eld resulting in an expansion of the ow normal to the main ow direction and therefore in an effective performance loss

HAGEMANN ET AL 629

Table 3 Design parameters for linear and toroidal subscale plug nozzlesa

Design data Linear plug (FESTIP) Circular clustered plug (ARPT)

Chamber pressure pc 875 bar 80 barDriving gas Air AirMass ow rate mCcedil 595 kgs 546 kgsDesign pressure ratio pc pamb 200 200Geometric area ratio laquo 1265 1265aUsed in the ESA FESTIP and ESA ARPT studies

Fig 20 Plug nozzle ow eld experiment vs numerical simula-tion a) toroidal plug nozzle experiment with gas-oilnitric-acidpropellant combination side view photograph and b) computa-tional results of toroidal plug nozzle Mach number distributionfull gray and isolines centerline section

Fig 21 Design of linear subscale plug nozzle of the ESA FESTIPTechnology Program (plug size 192 mm high 214 mm wide)

Fig 22 Calculated wall pressure data for linear subscale plugnozzle at different pressure ratios

The change of wake ow behavior may be strongly in uencedby the penetration of ambient pressures through both end sidesparticularly for truncated plug nozzles End plates as foreseenfor the linear plug nozzle of the X-33 demonstrator vehiclecould be used to avoid this ambient pressure penetration

Within the framework of the ESA FESTIP Technology Pro-gram lsquolsquoTechnology Developments on Rocket and Air Breath-ing Propulsion for Reusable Launch Vehiclesrsquorsquo subscale cold-gas tests and numerical simulations of linear plug nozzles arebeing performed by Dasa at the high-speed wind tunnel [Hoch-geschwindigheitswind kanal (HWK) in Bad Salzungen Ger-many] of the Technical University of Dresden719 Design pa-

rameters for the linear plug model are included in Table 3Figure 21 shows the baseline model with a linear throat Afterpassing the linear throat the ow expands directly onto theplug surface A second model will be tested with a bell-shapedlinear nozzle extension following the linear throat for internalexpansion prior to further external expansion onto the plugsurface19 The contour of the plug was designed by the methodof Angelino29 and three different plug lengths as 5 20 and40 of the ideal length were tested Figure 22 shows calcu-lated wall pressures for the full-length plug for different am-bient pressures simulating the operational range of sea-leveloperation up to higher-altitude operation The ow structure inform of the left- and right-running characteristics at two pres-sure ratios is shown in Fig 23 For lower pressure ratios thanthe design pressure ratio of pc pamb = 200 the in uence of the

630 HAGEMANN ET AL

Fig 23 Left- and right-running characteristics in ow eld at twopressure ratios pc pamb = a) 8 and b) 200 (inviscid analysis withmethod of characteristics)

Fig 24 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 20 plug body

Fig 25 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 40 plug body

Fig 27 Flow phenomena of an E ndash D nozzle with a) open and b)closed wakes

Fig 26 Sketch of an E ndash D nozzle

system of recompression shocks and expansion waves on thewall pressure distribution (Figs 17 and 22) can be seen bythe waviness of the pressure distribution on the plug surfaceFirst experimental results are shown in Figs 24 and 25 whichalso show the waviness of the pressure distribution for low-pressure ratios

3 Plug Nozzles Final Aspects

The performance of circular and linear plug nozzles is ad-versely affected by the external airstream ow ie the aspi-ration effect of the external ow slightly reduces performanceand favors an earlier change in wake- ow development for

truncated plug nozzles32ndash 35 Cold- ow tests showed that theeffect of external ow on performance is con ned to a narrowrange of external ow speeds with Mach numbers near unity

The altitude compensation capability of circular and linearplug nozzles for higher ambient pressures is indisputable Be-cause plug nozzles lose this capability for pressure ratios abovethe design pressure ratio the latter should be chosen as highas possible Taking this into account plug nozzles will featurean even better overall performance than those shown in Figs18 and 19 Unfortunately for equal geometrical area ratiosplug nozzles perform worse at high altitude than do conven-tional bell nozzles because of truncation and clustering Forhigh-area-ratio nozzles with relatively short lengths plug noz-zles perform better than conventional bell nozzles In additionto having excellent capabilities for altitude compensation plugnozzles have additional advantages including ease in vehicleand engine integration

C E ndash D Nozzles

An expansionndash de ection (E ndash D) nozzle is shown in Fig 26E ndash D nozzles were at one time thought to have capabilities foraltitude compensation because the gas expansion takes placewith a constant pressure free boundary Thus the aerodynamicbehavior of E ndash D nozzles as a function of altitude is in prin-ciple quite similar to plug nozzles because the expansion pro-cess is controlled by the ambient pressure hence by the alti-tude In contrast to plug nozzles however the expansionprocess is controlled from inside the nozzle for E ndash D nozzlesAt low altitude the higher ambient pressure limits the gasexpansion resulting in a low effective expansion area ratioThe exhaust gas is adapted to the ambient pressure level by

HAGEMANN ET AL 631

Fig 30 Performance losses in dual-throat nozzles as a functionof nozzle design parameters (with r4r1 = 352 r2r1 = 116 and mCcedil 2

= 00)

Fig 29 Flow phenomena during a) sea-level and b) high-altitudeoperations

Fig 28 Sketch of a dual-throat nozzle view of combustion cham-ber and throat region

systems of recompression and expansion waves (Fig 27) Athigher altitude the lower ambient pressure allows more gasexpansion within the nozzle resulting in a higher effectiveexpansion area ratio But in contrast to plug nozzles the pres-sure in the wake of the center plug is always less than theambient pressure because of the aspiration effect This occursat low pressure ratios when the wake is opened and results inan aspiration loss Furthermore because the exhaust ow ex-pands to this base pressure rather than to the ambient pressurelevel wall pressures downstream are overexpanded This re-sults in an additional overexpansion loss As the pressure ratioincreases the wake region closes and is thus totally isolatedfrom the ambient environment (see Fig 27) The behavior dur-ing transition from open wake to closed wake is again equalto plug nozzles and the base pressure in the closed-wake re-gion is essentially independent of the ambient pressure

The E ndash D nozzle concept has also been a subject of numer-ous analytical and experimental studies Results from thesestudies show that E ndash D nozzle capabilities for altitude com-pensation are poor and are in fact worse than those of plugnozzles because of aspiration and overexpansion losses35 Forhigh-area-ratio nozzles with a relatively short length an E ndash Dnozzle performs better than a comparable conventional bellnozzle at the same length because of lower divergence andpro le losses than the bell nozzle

The advantages of the E ndash D nozzle concept include its smallengine envelope and no moving parts However like toroidalplug nozzles E ndash D nozzles have the disadvantage of havinghigher throat heat uxes relative to a conventional bell nozzlewith an equal throat area The higher throat heat ux resultsagain from the relatively thin annular throat gap This problemhowever can be remedied with the modular thrust cell clusterconcept Additional advantages of E ndash D nozzles with clusteredthrust cells are the same as those already discussed for theclustered plug nozzle concept with regard to ease in manu-facturing thrust vector control by throttling or shutting off anindividual or group of thrust cells and lower nozzle throatheating

D Nozzles with Throat Area Varied by a Mechanical Pintle

This nozzle concept utilizes a conventional bell nozzle witha xed exit area and a mechanical pintle in the combustionchamber and throat region to vary the throat area and hencethe expansion area ratio The area of the nozzle throatmdashanannulus between the pintle and the shroudmdashis varied by mov-ing the pintle axially

The pintle concept has been used in solid rocket motors asa mean to provide variable thrust The concept in principleallows a continuous variation of the throat area and thus op-timum expansion area ratios throughout a mission Howeverit requires an actuator and a sophisticated control system Theconcept raises issues of engine weight design complexitycooling of the pintle and nozzle throat and reliability

In Ref 36 the aerodynamic performances of nozzles with ve different pintle geometries were calculated and comparedwith a reference bell nozzle The performance losses of thesepintle nozzles when compared with the bell nozzle are in therange of 1ndash 25 Performances of a xed pintle geometry atthree different locations were also calculated The results showthat the performance loss varies with the pintle location

E Dual-Mode Nozzles

Dual-mode rocket engines using one or two fuels offer atrajectory-adapted dual-mode operation during the ascent of alauncher which may be of signi cant advantage for single-stage Earth-to-orbit vehicles as compared to conventionalrocket engines with bell-type nozzles This engine concept in-volves the use of a dense propellant combination with mod-erate performance during liftoff to provide high thrust duringthe initial ight phase and a better performing propellant com-bination in vacuum which result in higher speci c impulse

The fuels are burned in two different combustion chamberswith one located completely inside the other in the case ofengines with dual-throat nozzles or with a conventional bellthrust chamber surrounded by an annular thrust chamber in thecase of dual-expander engines This type of engine has a built-in accelerationndash reduction capability achieved by shuttingdown one of two thrust chambers The total engine thrust isthen provided by the remaining thrust chamber with the useof the total nozzle exit area leading to an increase in speci cimpulse Apart from the indicated bene ts of dual-mode en-gines which will be discussed in more detail later their de-

632 HAGEMANN ET AL

Table 4 Thrust chamber parameters for simulated dual-expander nozzlesa

Operation mode

Aerojet

First (inout) Second (inout)

DLR

First (inout) Second (inout)

Propellants C3H8ndash O2H2ndash O2 mdashH2ndash O2 H2ndash O2H2ndash O2 mdashH2ndash O2

Chamber pressure pc 414207 bar mdash207 bar 200200 bar mdash200 barMixture ratio r 337 mdash7 77 mdash7Exit area ratio laquo 69 146 58 116aInnerouter chambers

Fig 32 Flow phenomena in a dual-expander nozzle during a)sea-level- and b) high-altitude operation

Fig 31 Sketch of a dual-expander nozzle view of combustionchamber and throat region

velopment and construction require considerable technologicaleffort

1 Dual-Throat Nozzles

A dual-throat nozzle con guration is shown in Fig 28 Itconsists of two conventional bell-thrust chambers with onelocated completely inside the other At low altitude the outerthrust chamber operates with the inner thrust chamber runningin parallel In this operation mode the engine has a larger throatproviding a moderate expansion area ratio During the missionthe outer thrust chamber is shut off and operation continueswith only the inner engine In this con guration ow fromthe inner engine expands and attaches supersonically to theouter engine resulting in a higher expansion area ratio for theremainder of the burn Flow phenomena in both operationmodes are included in Fig 29

Hot- red tests at Aerojet were conducted to provide heattransfer data that were very useful for the thermal analysis anddesign of the dual-throat nozzle con guration37 These testsshowed that ow separation occurred in the inner engine noz-zle at higher ratios of outer to inner chamber pressures duringthe rst operation mode with both chambers burning in par-allel The ow separation resulted in a higher heat load to theinner nozzle Subscale tests performed at the Keldysh Center3

have shown that the additional loss caused by the nozzle con-tour discontinuity during vacuum operation with active innerchamber is in the range of 08ndash 4 depending on geometricaldata (see Fig 30) This high-performance loss results from theinteraction of the inner chamber jet with the outer chambernozzle wall3 The decrease of the jet incidence angle on thewall by means of gas injection through the outer chamber re-duces this performance loss by 04ndash 073

2 Dual-Expander Nozzles

A dual-expander nozzle has two concentric thrust chambersand nozzles It consists of a conventional bell thrust chambersurrounded by an annular thrust chamber Both chambers haveshort primary nozzles which end in a common divergent noz-zle extension Figure 31 shows a typical dual-expander nozzlecon guration At low altitude both thrust chambers operatesharing the same exit area which results in a moderate expan-sion area ratio Part way into the mission one thrust chamberis shut off allowing the other nozzle to use the whole exitarea creating a high-expansion-area ratio for the remainder of

the burn In principle the two operation modes are comparableto those of dual-throat nozzles

Numerical simulations of the ow elds in dual-expandernozzles during all operation modes have been performed atAerojet38 DLR39 and the Keldysh Center Aerojet simulationsare based on an engine design using different propellantshydrocarbonoxygen for the inner chamber and hydrogenox-ygen for the outer chamber whereas DLR simulations arebased on a combined vehicleengine analysis using single-fuelsingle-mixture ratio dual-expander engines with hydrogenoxy-gen Table 4 summarizes combustion chamber parameters Flowphenomena observed in numerical simulations are shown in Fig32a for the mode 1 operation with both thrust chambers burn-ing Compression waves are induced near the inner nozzle lipas a result of the strongly inhomogeneous ow character at thepoint where both exhaust gases of inner and outer combustionchamber merge into the common divergent nozzle part Upand down variations of pressure ratios at the lip did not changethis wave formation even in cases of signi cantly lower exitpressures in the inner nozzle compared to the pressure eld ofthe outer nozzle at the lip Further downstream the compres-sion waves interact with the wall resulting in a re ection ofthe compression waves back into the ow eld

During mode 2 operation a strong expansion of the outernozzle ow is observed when the expansion ratio suddenlyincreases at the end of the nozzle lip The ow is directedtoward the axis of symmetry Near the centerline the ow thenturns over to the axial direction inducing a recompressionshock The static pressure rises signi cantly in this recom-pression region on the centerline A sub- and supersonic recir-culation zone establishes in the inner chamber of the dual-expander nozzle Figure 32b emphasizes the essential owpattern in this operation mode

These analyses have shown that dual-expander nozzles pro-duce high performance in both operation modes3839 Figure 33summarizes performance behavior as a function of ight alti-tude for the dual-expander engine simulated at DLR For com-parison performance data are included for two reference bell

HAGEMANN ET AL 633

Fig 33 Performance data for a dual-expander nozzle

nozzles that have the same area ratios as the dual-expandernozzle during its two operation modes Losses because of thenonhomogeneous exit ow and the induced shocks are com-parable with corresponding values of both conventional noz-zles1

Within the design process of dual-expander engines it is nec-essary to take into account that the nonburning chamber isexposed to high heat uxes and pressure oscillations duringmode 2 operation which might reach a high level and lead tostructural disintegration of the inner nozzle structure There-fore a mode-2 operation with bleed gas generation at a ratherlow thrust chamber pressure level in the inner chamber seemsto be advantageous Because of the bleed gas generation in theprimary chamber the overall mass ow rate during this alter-native mode-2 operation is increased In addition the reductionof the effective exit area ratio results in minor impulse deg-radations during this operation mode when compared with themode-2 operation with no bleed gas generation

Several analytical works on SSTO- and two-stage-to-orbit(TSTO)-vehicles using hydrogenpropane or hydrogenmeth-ane as fuels revealed the lowest vehicle dry masses for dual-mode engines in comparison to other engines1439 Other dual-mode engines using hydrogen as the single fuel but using twomixture ratios also revealed some bene ts over conventionalengines for SSTO- and TSTO applications Even a single-fueloperation with constant mixture ratios in both combustionchambers indicated a gain in launcher performance40

V ConclusionsSeveral nozzle concepts that promise gains in performance

over conventional nozzles were discussed in this paper in-cluding performance enhancements achieved by slight modi- cations of existing nozzles eg through cool gas injectioninto the supersonic nozzle part It is shown that signi cantperformance gains result from the adaptation of the exhaust ow to the ambient pressure and special emphasis was givento altitude adaptive nozzle concepts

A number of nozzle concepts with altitude-compensating ca-pability were identi ed and described To assist the selectionof the best nozzle concept for launch vehicle applications theperformance of the nozzles must be characterized This can bedone using computational uid dynamics (CFD) andor cold- ow tests Existing CFD methods that are in use in the aero-space industry and at research institutes have been veri ed fora wide number of sub- and full-scale experiments and providesuf ciently reliable performance determination for the differ-ent nozzle types

Theoretical evaluations numerical simulations and test re-sults showed that the different concepts have real altitude-com-pensating capabilities However the compensation capabilitiesare limited and there are some drawbacks associated with eachof the concepts Additional performance losses are induced inpractically all of these nozzle concepts when compared to anideal expansion mainly because of the nonisentropic effects

such as shock waves and pressure losses in recirculation zonesHowever these additional performance losses are less than1 ndash 3 depending on the different nozzle concepts

Four nozzle concepts have been selected for further evalu-ation under a NASA Aerojet cooperative agreement They in-clude a plug nozzle an E ndash D nozzle a dual-bell nozzle and adual-expander nozzle The nozzles in addition to a referencebell nozzle have been designed by Aerojet for cold ow testsCFD calculations performed at Aerojet demonstrated that allof these nozzles have high and approximately equal perfor-mances at the design points

In Europe special attention is being paid to plug- and dual-bell nozzles Different plug nozzle concepts are being testedin cold- ow tests within the ESA ARPT (clustered plug noz-zle) and ESA FESTIP (linear plug nozzle) contracts In addi-tion to experiments three-dimensional numerical simulationson these nozzle concepts are being performed Additional ex-periments on different nozzle concepts with forced- ow sep-aration devices eg dual-bell nozzles are currently plannedat DLR

Different advanced nozzle concepts have been examined andare under examination now at the Keldysh Center with bothCFD calculations and experiments on the Nozzle DifferentialFacility which has a nozzle performance determination errorof about 005 This high accuracy is achieved by simulta-neously running the test nozzle and a reference nozzle bothmounted on a common longitudinal axis The performancecharacteristics of the reference nozzle are well known and thethrust difference is measured with strain gauges This facilityis equipped with a wind tunnel to determine nozzle thrust per-formance in the presence of external transonic ows withMach numbers up to 12

In addition to aerodynamic performance other technical is-sues (weight cost design thermal management manufactur-ing system performance and reliability) must be addressedFurthermore before a nal decision can be made as to whichnozzle concept offers the greatest bene ts with regard to aneffective payload mass injection combined launcher and tra-jectory calculations must be performed and compared to a ref-erence launcher concept with conventional nozzles Differentnozzle ef ciencies which account for the additional losses ofadvanced rocket nozzles and are extracted from numerical sim-ulations and experiments must be taken into account

At least for some of these nozzle concepts the plug nozzlethe dual-bell nozzle and the dual-expander nozzle bene tswith regard to lower overall launcher masses have been dem-onstrated in the literature Furthermore the plug nozzle con-cept will be the rst in ight of all of these advanced nozzleconcepts with the X-33 demonstrator vehicle

AcknowledgmentPresented as Paper 18 at the 3rd International Symposium

on Space Propulsion Beijing Peoplersquos Republic of ChinaAug 11ndash 13 1997

References1Manski D and Hagemann G lsquolsquoIn uence of Rocket Design Pa-

rameters on Engine Nozzle Ef cienciesrsquorsquo Journal of Propulsion andPower Vol 12 No 1 1996 pp 41ndash 47

2lsquolsquoLiquid Rocket Engine Nozzlesrsquorsquo NASA Space Vehicle DesignCriteria NASA SP-8120 1976

3Dumnov G E Nikulin G Z and Ponomaryov N B lsquolsquoInves-tigation of Advanced Nozzles for Rocket Enginesrsquorsquo Space Rocket En-gines and Power Plants Vol 4 No 142 NIITP 1993 pp 1012 ndash

1018 (in Russian)4Nguyen T V and Pieper J L lsquolsquoNozzle Flow Separationrsquorsquo Pro-

ceedings of the 5th International Symposium of Propulsion in SpaceTransportation (Paris France) 1996

5Manski D lsquolsquoClustered Plug Nozzles for Future European Reus-able Rocket Launchersrsquorsquo German Aerospace Research Establishment643-817 Lampoldshausen Germany 1981 (in German)

6Horn M and Fisher S lsquolsquoDual-Bell Altitude Compensating Noz-

622 HAGEMANN ET AL

Fig 5 Flow separation in overexpanding rocket nozzles wall pressure pro le and phenomenology

formance during the ascent of the launcher owing to their xedgeometry Signi cant performance losses are induced duringthe off-design operation of the nozzles when the ow is over-expanded during low-altitude operation with ambient pressureshigher than the nozzle exit pressure or underexpanded duringhigh-altitude operation with ambient pressures lower than thenozzle exit pressure Figure 2 shows photographs of nozzleexhaust ows during off-design operation In the case of over-expanded ow oblique shocks emanating into the ow eldadapt the exhaust ow to the ambient pressure Further down-stream a system of shocks and expansion waves leads to thecharacteristic barrel-like form of the exhaust ow In contrastthe underexpansion of the ow results in a further expansionof the exhaust gases behind the rocket

Off-design operations with either overexpanded or under-expanded exhaust ow induce performance losses Figure 3shows calculated performance data for the Vulcain 1 nozzle asfunction of ambient pressure together with performance datafor an ideally adapted nozzle Flow phenomena at differentpressure ratios pc pamb are included in Fig 4 [The sketch with ow phenomena for the lower pressure ratio pcpamb shows anormal shock (Mach disk) Depending on the pressure ratiothis normal shock might not appear see eg Fig 2] TheVulcain 1 nozzle is designed in such a manner that no uncon-trolled ow separation should occur during steady-state oper-ation at low altitude resulting in a wall exit pressure of pwe

rsquo 04 bar which is in accordance with the Summer eld cri-terion8 The nozzle ow is adapted at an ambient pressure ofpamb rsquo 018 bar which corresponds to a ight altitude of h rsquo15000 m and performance losses observed at this ambientpressure are caused by internal loss effects (friction diver-gence mixing) as shown in Fig 1 and Table 1 Losses inperformance during off-design operations with over- or un-derexpansion of the exhaust ow rise up to 15 In principlethe nozzle could be designed for a much higher area ratio toachieve better vacuum performance but the ow would thenseparate inside the nozzle during low-altitude operation withan undesired generation of side-loads

A Flow Separation and Side-Loads

Flow separation in overexpanding nozzles and its theoreticalprediction have been the subject of several studies in thepast348 and different physical models and hypotheses for theprediction of ow separation have been developed In stronglyoverexpanding nozzles the ow separates from the wall at acertain pressure ratio of wall pressure to ambient pressure pwpamb The typical structure of the ow eld near the separationpoint is shown in Fig 5 together with wall pressure dataSeparation and the formation of a recirculation zone at the wall

induce an oblique shock wave near the wall which leads to arecompression of the ow

The physical phenomenon of ow separation can be dividedinto two simple phenomena3 The rst is the turbulent bound-ary-layer separation from the nozzle wall which is character-ized by the ratio of the nozzle wall pressure just after theseparation pp to the nozzle wall pressure just before separa-tion psep This pressure ratio is referred to as the critical pres-sure ratio pcr = pseppp = p1p2 The second phenomenon isconnected with the ow in the separated zone which is char-acterized by a minor pressure gradient along the wall Theanalysis of model experimental data on separated turbulent su-personic ows shows that the pressure ratio pcr is equal forseparated ows before an obstacle and for separated nozzle ows3 For turbulent ows pcr shows a slight dependence onReynolds number but depends strongly on Mach number Fur-thermore investigations of separated ows in rocket enginenozzles showed that pcr is also a function of wall temperaturegas composition and nozzle wall roughness

Various approaches to the prediction of ow separation areused in industry and research institutes For example the Kel-dysh Centerrsquos method for the determination of the ow sepa-ration point in rocket nozzles is based on empirical relation-ships obtained for pcr pppwe and pwepamb Using theserelations the pressure ratio pseppamb the separation point andthe wall pressure distribution are determined Aerojet uses tab-ulated data for the prediction of ow separation from variousexperiments with conical and bell-shaped nozzles4 At a knownpressure ratio of chamber pressure to ambient pressure a lowerand an upper separation pressure ratio follows from the data-base that is different for conical and bell-shaped nozzles Aseparation criterion used at the European industry and scien-ti c institutes describes the pressure ratio pseppamb as functionof the local ow Mach number at the wall near the separationpoint This analytical model was derived from various cold-and hot- ring tests of overexpanding nozzles9

Various numerical simulations of ow problems featuring ow separation have shown that the numerical prediction of ow separation with state-of-the-art turbulence modeling re-sults in good agreement with experimental data310 As an ex-ample Figs 6 and 7 compare experimental and numerical dataon wall pressure ratios in two overexpanding nozzles Fur-thermore former experiments of separated ows indicated thatthere is a random movement of the separation line in over-expanding rocket nozzles91112 resulting in the possible gen-eration of side loads The correct prediction of side-loads withmodels is uncertain and still a subject of on-going researchCommon side-load models described in Refs 4 and 9 assumethat the separation line in the nozzle has a maximum tilt angle

HAGEMANN ET AL 623

Fig 6 Experimental and numerical data on wall pressure ratiosin overexpanding conical nozzles with ow separation Cold-gastest case with subscale conical nozzle

Fig 7 Experimental and numerical data on wall pressure ratiosin overexpanding bell nozzles with ow separation Hot-gas testcase with J-2S engine

with the centerline The minimum and maximum separationpoints are predicted using the upper and lower separation datathat result from the applied separation model In the case ofdistinct separation points as obtained with the Summer eld- orSchmucker-criterion a certain scattering around the predictedseparation point is assumed The side-load is then calculatedby integrating the wall pressure in the asymmetrical ow sep-aration region49 These approaches gave reasonable results forthe J2-S and Vulcain 1 nozzle

At the Keldysh Center in Russia model experiments of sep-arated ow effects on side-loads acting on nozzles began inthe early 1980s11 These investigations were directed mainlyto the determination of the unsteady pressure uctuations inthe separation zone on the whole engine The method is de-scribed in Ref 11 in more detail it is based on a generalizationof obtained empirical data for the spectral density of the pul-sating pressure eld in the separation zone Distribution of thepressure amplitudes is governed by probability density func-tions This method was validated with the experimental rms-val-ues of side-loads acting on the RD-0120 nozzle (CABD hydro-genndash oxygen engine rst-stage engine of Russian Energialauncher) Other hypotheses for the appearance and predictionof side-loads also exist eg based on aeroelastic analyses13

which account for the coupling of ow oscillations with thinnozzle shells However these hypotheses require experimentalveri cation and validation

Side-loads are an undesired phenomenon that may result inthe destruction of the rocket nozzle and should therefore beavoided In Ref 9 the destruction of a J-2D engine (Rocket-dyne hydrogen ndash oxygen engine second- and third-stage engineof Saturn-5 launcher) as a result of side-loads is reported The

transient startup of the rocket engine generates a nozzle owwith uncontrolled ow separation and with possible side-loadgeneration in the nozzle whereas maximum area ratios of all rst-stage nozzles or booster nozzles are chosen to avoid owseparation at the nominal chamber pressure operation As aresult the vacuum performance of rocket engines that operateduring the entire launcher trajectory such as the SSME orVulcain 1 engines is limited

B Potential Performance Improvements

Compared to existing rocket engines a gain in performanceis achieved with advanced engines such as mixed-mode pro-pulsion systems dual-mixture ratio engines or dual-expanderengines Nevertheless the upgrade of existing engines withbetter performing subsystems such as turbines and pumpsalso leads to a gain in overall performance data and is dis-cussed in more detail in Ref 14 Nozzle performance of con-ventional rocket engines is already very high with regard tointernal loss effects (friction nonuniformity) However fornozzles of gas-generator open-cycle engines such as the Vul-cain 1 engine a slight improvement in performance can beachieved with turbine exhaust gas (TEG) injection into themain nozzle as realized in the F-1 (kerosenendash oxygen engine rst stage of Saturn-5 launcher) and J-2S and it is foreseenfor the Vulcain 2 engine (hydrogen ndash oxygen engine upgradeof Vulcain 1 engine) and con rmed by numerical simula-tions131516 and experimental results3 This is mainly achievedthrough lower friction losses in the main nozzle1315 and be-cause the bypass nozzles used for the expansion of the TEGswhich have higher divergence losses are removed

Despite the slight performance gain by TEG injection thelow-pressure near-wall stream of the injected gas favors a re-duction of the critical pressure ratio at which ow separationoccurs and therefore an earlier nozzle ow separation312 Fur-thermore the presence of the secondary exhaust gas injectioncomplicates nozzle-contouring methods with regard to theavoidance of uncontrolled ow separation for rst-stage orbooster nozzles

III Altitude Adaptive NozzlesA critical comparison of performance losses shown in Table

1 reveals that most signi cant improvements in nozzle perfor-mance for rst-stage or SSTO engines can be achieved throughthe adaptation of nozzle exit pressures to the variations in am-bient pressure during the launcherrsquos ascent through the atmo-sphere Various concepts have been previously mentioned andwill be discussed in detail in the following text

A Nozzles with Devices for Controlled Flow Separation

Several nozzle concepts with devices for controlled owseparation have been proposed in the literature with primaryemphasis on the reduction of side-loads during sea level orlow-altitude operation But the application of these conceptsalso results in an improved performance through the avoidanceof signi cant overexpansion of the exhaust ow

1 Dual-Bell Nozzle

This nozzle concept was rst studied at the Jet PropulsionLaboratory17 in 1949 In the late 1960s Rocketdyne patentedthis nozzle concept which has received attention in recentyears in the US and Europe671819 Figure 8 illustrates thedesign of this nozzle concept with its typical inner base nozzlethe wall in ection and the outer nozzle extension This nozzleconcept offers an altitude adaptation achieved only by nozzlewall in ection In low altitudes controlled and symmetrical ow separation occurs at this wall in ection (Fig 9) whichresults in a lower effective area ratio For higher altitudes thenozzle ow is attached to the wall until the exit plane and thefull geometrical area ratio is used Because of the higher arearatio an improved vacuum performance is achieved Howeveradditional performance losses are induced in dual-bell nozzles

624 HAGEMANN ET AL

Fig 8 Sketch of a dual-bell nozzle

Fig 9 Flow eld phenomena in dual-bell nozzles a) sea-levelmode with ow separation at the wall in ection point and b) al-titude mode with a full- owing nozzle

Fig 10 Performance data of a dual-bell nozzle Performance iscompared with two baseline bell-type nozzles as function of ightaltitude (baseline nozzle 1 same area ratio as dual-bell base noz-zle baseline nozzle 2 same area ratio as nozzle extension)

Fig 11 Performance characteristics of a dual-bell nozzle Per-formance is compared with two baseline bell-type nozzles as func-tion of ight altitude (baseline nozzle 1 same area ratio as dual-bell base nozzle baseline nozzle 2 same area ratio as nozzleextension)

as compared with two baseline nozzles having the same arearatio as the dual-bell nozzle at its wall in ection and in its exitplane Figures 10 and 11 illustrate the performance of a dual-bell nozzle as a function of ight altitude in comparison withboth baseline bell-type nozzles (Design parameters of thedual-bell nozzle are taken from a launcher analysis publishedin Ref 18 Propellants hydrogenoxygen r = 6 pc = 200 barrt = 007 m wall in ection at area ratio laquoB = 30 and totalarea ratio laquoE = 100) The pressure within the separated owregion of the dual-bell nozzle extension at sea-level operationis slightly below the ambient pressure inducing a thrust lossreferred to as lsquolsquoaspiration dragrsquorsquo In addition ow transitionoccurs before the optimum crossover point which leads tofurther thrust loss as compared to an ideal switchover Thenonoptimum contour of the full owing dual-bell nozzle re-sults in further losses at high altitudes

To gain insight into the performance and ow behavior ofdual-bell nozzles at different ambient pressures extensive nu-merical simulations with parametrical variations of contour de-sign parameters were performed18 An optimized bell nozzlewith equal total length and area ratio was used as the referencenozzle for comparison As a result the vacuum performance ofthe dual-bell nozzles has a degradation because of the imper-fect contour and this additional loss has the same order ofmagnitude as the divergence loss of the optimized bell nozzle

The simulations of sea-level operation also revealed an ad-ditional performance loss because of aspiration drag which isless than 3 for these dual-bell nozzles This additional lossdepends linearly on the ambient pressure and therefore it isreduced during the ascent of the launcher Furthermore thesesimulations showed that the application of commonly usedseparation criteria derived for conventional nozzles and ap-

plied to dual-bell nozzles with their wall in ection to estimatethe critical pressure ratio yields reasonable results with an ac-curacy of rsquo15 Experimental data obtained at the KeldyshCenter revealed that the critical pressure ratio at the wall in- ection is about 15ndash 20 less than the critical pressure ratioof a conventional nozzle and that usual separation criteriamust be corrected for an accurate prediction of the critical wallpressure ratio

Flow transition behavior in dual-bell nozzles strongly de-pends on the contour type of the nozzle extension3618 A sud-den transition from sea-level to vacuum operation can be atleast theoretically achieved by two different extensions witha zero wall pressure gradient (constant pressure extension) ora positive wall pressure gradient (overturned extension) But acritical analysis of the transition behavior considering decreas-ing ambient pressures during the launcher ascent revealed thata considerable time with uncontrolled ow separation withinthe nozzle extension exists even for these types of extensionsThe duration of this period can be reduced drastically by throt-tling the chamber pressure18

The main advantage of dual-bell nozzles as compared toother means of controlling nozzle ow separation is its sim-plicity because of the absence of any movable parts and there-fore its high reliability It is necessary to note that the external ow over the vehicle in ight reduces the pressure in the ve-hicle base region where engines are installed The ambient

HAGEMANN ET AL 625

Fig 13 Principle performance characteristics of the nozzle withinsert

Fig 12 RD-0120 nozzle hardware with removed nozzle insertand sketch of secondary nozzle mounted inside of the RD-0120nozzle (photographs taken from Ref 23)

pressure triggering the ow transition is the vehicle base pres-sure instead of the atmospheric pressure at the speci c ightaltitude As the base pressure is lower than the atmosphericpressure the nozzle ow transition occurs at a lower altitudethan the one showed in Fig 11 which slightly decreases theef ciency of the dual-bell nozzle operation along the trajectory

Despite the additional losses induced in dual-bell nozzlesthey still provide a signi cant net impulse gain over the entiretrajectory as compared to conventional bell nozzles Indepen-dent combined launcher and trajectory analyses performed byDasa within the European Space Agency Future EuropeanSpace Transportation Investigations Program (ESA FESTIP)study19 and at DLR on SSTO vehicles powered with dual-bellnozzles result in a signi cant payload gain when comparedwith a reference launcher equipped with conventional nozzles

2 Nozzles with Fixed Inserts

A trip ring attached to the inside of a conventional nozzledisturbs the turbulent boundary layer and causes ow separa-tion at higher ambient pressures At higher altitudes with lowerambient pressures the ow reattaches to the wall behind thetrip ring and full owing of the nozzle is achieved The tran-sition from sea level to vacuum mode depends on the wallpressure near the trip-ring location and on the disturbance in-duced by the trip ring The size of the trip ring is a compromisebetween stable ow separation during sea-level operation andthe induced performance loss during vacuum operation In Ref9 it is reported that a trip-ring size of 10 of the local bound-ary-layer thickness is suf cient to ensure stable ow separa-tion

In principle this concept is similar to the dual-bell nozzleconcept with regard to performance characteristics as shownin Fig 11 The sea-level performance of this nozzle concept islower than the performance of a conventional bell nozzle trun-cated at the trip-ring location because of the aspiration dragin the separated ow region of the nozzle Furthermore at sealevel the bell nozzle with trip rings has even higher divergencelosses than a comparable dual-bell nozzle because the nozzlecontour upstream of the obstacle differs from the optimal con-tour for this low-area ratio as a result of the bell nozzle designfor best vacuum performance The additional losses inducedduring vacuum operation is about 1 compared with the per-formance of the bell nozzle without an obstacle Thus theadditional losses are comparable to the additional losses in-duced in dual-bell nozzles As for dual-bell nozzles the tran-sition behavior of this nozzle concept is uncertain but it willbe even more uncertain than for a dual-bell nozzle with a con-stant pressure or overturned nozzle extension

In principle several altitude adaptations can be achievedwith one nozzle by various trip rings mounted one behind theother However this results in increasing vacuum performancelosses The trip rings can also be attached into existing nozzlesand therefore represent a low-cost concept at least for testpurposes with low technological risk Trip rings have beendemonstrated to be effective for side-load reductions duringthe transient startup of rocket engines20 The main problemswith trip-ring nozzles are not only performance losses but alsoring resistance in high-temperature boundary layers the exactcircumferential xing and the uncertainties in the transitionbehavior These uncertainties might be why active interest inthis nozzle concept in the 1970s which is documented in var-ious publications920ndash22 has disappeared in recent years

A further concept with a xed wall discontinuity is a nozzlewith a circumferential groove The aerodynamic behavior of anozzle with a groove and nozzle with trip rings are quite sim-ilar9 but this concept is hard to realize in the case of thinnozzle shells which are used in all of todayrsquos rocket nozzles

3 Nozzles with Temporary Inserts

Nozzle concepts with xed wall discontinuities have the dis-advantage of lower vacuum performance as compared with a

conventional bell nozzle with equal design and operation dataA promising concept for controlled ow separation is thereforetemporary inserts which are removed for vacuum operationThese inserts can be either ablative or ejectible The insertsmay have the form of a complete secondary nozzle23 (Fig 12)or of small steps attached inside the nozzle wall In case ofejectible inserts a reliable mechanism is needed to provide asudden and symmetrical detachment In any case shocks areinduced during the transient ejection because the inserts act asan obstacle in the supersonic exhaust ow These shocks alsointeract with the nozzle walls and increase pressure loads onthe wall and local heat uxes A nonsymmetrical ejectionwould then result in the generation of side-loads Furthermorethe danger of a downstream collision with the nozzle wallarises because the inserts might also experience a transversalmovement toward the walls

Recent hot- ring tests performed in Russia with a modi edRD-0120 engine equipped with a secondary nozzle insert re-vealed a signi cant performance gain of 12 during the sea-level operation at 100 chamber pressure compared with theoriginal RD-0120 performance23 Nominal chamber pressureof this engine is pc = 206 bar with an area ratio of laquo = 857These full-scale hot- ring tests demonstrated the durability ofmaterials sealings and the release mechanism and thus thefeasibility of this concept Figure 12 shows the nozzle hard-ware and a sketch of the secondary nozzle mounted inside theRD-0120 nozzle23

The principle performance characteristics of this RD-0120nozzle with ejectible insert are included in Fig 13 The nozzleoperation with insert results in a slight performance loss com-pared with an ideal bell nozzle with the same reduced area

626 HAGEMANN ET AL

Fig 14 Sketch of a two-position nozzle during a) sea-level andb) vacuum operations

ratio because of aspiration drag and the presumably nonoptim-ized insert contour The performance degradation is compara-ble to the one induced in dual-bell nozzles during sea-leveloperation (Fig 11) In vacuum or high-altitude operations thehigher performance of the baseline nozzle is achieved

Another method for removing the inserts is to use combus-tible or ablative elements924 During the ascent of the launcherthe size of the insert is continuously reduced until it is com-pletely consumed resulting in a full owing bell nozzle witha clean contour for best vacuum performance The principaluncertainties of this nozzle concept are the stability and surfaceregression rates of the inserts Furthermore a homogeneoussymmetrical and temporally de ned consumption must beguaranteed despite possible local pressure and temperature uctuations near the nozzle walls This is currently highly un-certain

4 Nozzles with Active or Passive Secondary Gas Injection

With this nozzle concept the ow in an overexpanding noz-zle is forced to separate at a desired location by injecting asecond uid into the gas stream in the wall normal directionto achieve maximum disturbance of the main exhaust gas andthus to induce ow separation This secondary gas injectionis quite different from the aforementioned TEG injectionwhich should be injected with few or no disturbances to themain ow The injection could be either active that is forcedthrough gas expansion from a higher-pressure reservoir or pas-sive using holes in the wall through which ambient gas issucked in (vented nozzle concept) The latter concept can onlywork in case of gas pressures near the wall inside the nozzlethat are lower than the ambient pressure Experience on forcedsecondary gas injection gained at Aerojet shows that a largeamount of injected uid is required to induce a signi cant owseparation Furthermore no net speci c impulse gain is real-ized when considering the additional gas ow rate

In the vented nozzle concept segments of the bell nozzlewall have various slots or holes opened to the outside ambientpressure During low-altitude operation the slots are opened totrigger ow separation within the nozzle whereas the slots areclosed during high-altitude or vacuum operation so that thegas fully expands inside the entire nozzle Experiments with amodi ed RL10A-3 engine equipped with this nozzle conceptwere performed at Pratt amp Whitney25 Performance resultsshowed that over a small range of low-pressure ratios the per-forated nozzle performed as well as a nozzle with its area ratiotruncated immediately upstream of the vented area Howeverat above some intermediate pressure ratios the thrust ef -ciency suddenly dropped and approached that of the full- ow-ing nozzle The measured performance characteristics werequite similar to those of the dual-bell nozzles shown in Fig11

The altitude range of this nozzle concept is limited by thenumber and position of the holes because the pressure withinthe nozzle must be lower than the ambient pressure Further-more as the rocket aft-body base pressure is lower than thealtitude atmospheric pressure in the surrounding ambient ows the nozzle ow transition occurs at a lower altitude andthus the range of compensation is further reduced

5 Two-Position or Extendible Nozzles

Nozzles of this type with extendible exit cones are currentlyused only for rocket motors of upper stages to reduce the pack-age volume for the nozzle eg at present for solid rocketengines such as the inertial upper stage (IUS) or for the liquidrocket engine RL10 The main idea of the extendible extensionis to use a truncated nozzle with low expansion in low- ightaltitudes and to have a higher nozzle extension at high alti-tudes Figure 14 illustrates this nozzle concept Its capabilityfor altitude compensation is indisputable and the nozzle per-formance is easily predictable The whole nozzle contour in-cluding the extendible extension is contoured for maximum

performance at the high-area ratio with either the method ofcharacteristics or a variational method The area ratio of the rst nozzle section is then determined and the nozzle contouris divided into two parts The xed nozzle part and the ex-tendible extension Investigations conducted at the KeldyshCenter have shown that this nozzle-contouring method is notonly the simplest but also provides a good overall trajectoryperformance3 A minor performance loss is incorporated duringlow-altitude operations because of the truncated inner nozzlewhich has a nonoptimal contour for this interim exit area ratioThe performance characteristics as a function of ight altitudeare similar to those of the nozzle with the ejectible insert asshown in Fig 13 A tradeoff study performed at the KeldyshCenter on nozzles with an ejectible insert and an extendibleextension showed that the thrust characteristics of the nozzlewith an ejectible insert are slightly better than those of theextendible nozzle concept with the same overall dimensionsfor both nozzles

The main drawback of the extendible nozzle concept is thatit requires mechanical devices for the deployment of the ex-tension which reduces engine reliability and increases totalengine mass The necessity for active cooling of the extendibleextension requires exible or movable elements in the coolingsystem which also reduces system reliability The in uencesof the external ow and the generated jet noise on the extend-ible extension in the retracted initial position have not yetbeen fully investigated Former investigations conducted in theKeldysh Center showed that the external ow causes bothsteady and unsteady pressure loads on the retracted nozzle ex-tension whereas the engine jet noise causes strong vibrationsof the nozzle extension

B Plug Nozzles

Various experimental analytical and numerical research onplug nozzles have been performed since the 1950s in theUS4532ndash 35 Europe5710192627 Russia3 and Japan28 In contrastto the previously discussed nozzle concepts plug nozzles pro-vide at least theoretically a continuous altitude adaptation upto their geometrical area ratio Figures 15 and 16 show twodifferent design approaches for circular plug nozzles whichdiffer only in the chamber and primary nozzle layout A con-ical central body is shown here which could also be designedwith more sophisticated contouring methods2930 Different de-sign approaches include plug nozzles with a toroidal chamberand throat (with and without truncation) and plug nozzles witha cluster of circular bell nozzle modules or with clusteredquasirectangular nozzle modules The latter approach seems to

HAGEMANN ET AL 627

Fig 17 Flow phenomena of a plug nozzle with full length (left column) and truncated central body (right column) at different pressureratios pc pamb off-design (top bottom) and design (center) pressure ratio

Fig 16 Principle design of plug nozzles clustered plug 36 mod-ules with truncated plug body

Fig 15 Principle design of plug nozzles toroidal plug fulllength

be advantageous because further losses induced by the gapsbetween individual modules and the ow eld interactionsdownstream of the module exits can be minimized It has beenshown that transition from a round to a square nozzle resultsin a very small performance loss31 In principle the ow elddevelopment of a clustered plug nozzle with rectangular nozzlemodules is similar to that of a toroidal plug nozzle but avoidsthe inherent disadvantages of the toroidal plug design regard-ing 1) the control of a constant throat gap during manufactur-

ing and thermal expansion (side-loads and thrust vector devi-ations) 2) the cooling of toroidal throat with tiny throat gapsand 3) the control of combustion instabilities in the toroidalcombustion chamber Another plug nozzle con guration is thelinear plug nozzle which is foreseen for the propulsion systemof the RLV X-33 concept

1 Circular Plug Nozzles

Figure 17 summarizes the principle ow phenomena of cir-cular plug nozzles with full length and truncated central bodiesat different off-design (top and bottom) and design (center)pressure ratios that were observed in experiments and numer-ical simulations For pressure ratios lower than the design pres-sure ratio of a plug nozzle with a well-contoured central bodythe ow expands near the central plug body without separa-tion and a system of recompression shocks and expansionwaves adapts the exhaust ow to the ambient pressure pambThe characteristic barrel-like form with several in ections ofthe shear-layer results from various interactions of compres-sion and expansion waves with the shear layer and turbulentdiffusion enlarges the shear layer farther downstream of thethroat The existence of the overexpansion and recompressionprocesses is inferred from up- and down-variations of plugwall pressure pro les observed in various cold- ow tests andnumerical simulations and will also be shown later for linearplug nozzles

At the design pressure ratio (see Fig 17 left column cen-ter) the characteristic with the design Mach number should bea straight line emanating to the tip of the central plug bodyand the shear layer is parallel to the centerline However forcircular plug nozzles designed with contouring methods pro-posed in Refs 29 and 30 no exact one-dimensional exit owpro le can be achieved because both methods use thePrandtlndash Meyer relations that are only valid for planar owsFurthermore nonhomogeneous ow in the throat regionwhich is in general not considered within the contour designprocess also in uences the exit ow pro le The wall pressuredistribution remains constant at pressure ratios above the de-sign pressure ratio ie the plug nozzle behaves like a con-ventional nozzle the loss of its capability of further altitude

628 HAGEMANN ET AL

Table 2 Design parameters of toroidal plug nozzles

Design data Full-scale nozzlea Subscale nozzleb

Chamber pressure pc 100 bar 12 barPropellants Hydrogenoxygen Gasndash oilnitric-acidMixture ratio r 60 25Inner chamber diameter dcinner 43 m 019 mOuter chamber diameter dcouter 44 m 034 mGeometric area ratio laquo 55 10aUsed for numerical simulations and performance predictions see Fig 18

bUsed in experiments see Fig 20

Fig 18 Performance of numerically simulated plug nozzle withfull-length central body

Fig 19 Performance of numerically simulated plug nozzle withtruncated central body

adaptation being included Figure 17 (left column bottom) il-lustrates the ow eld at higher pressure ratios Performancedata of a typical plug nozzle are included in Fig 18 and com-pared to a conventional bell nozzle with equal area ratio Thedesign and combustion chamber parameters are included in theleft column of Table 227

The truncation of the central plug body which is an advan-tage because of the huge length and high structural mass ofthe well-contoured central body results in a different ow andperformance behavior as compared to the full-length plug noz-zle At lower pressure ratios an open wake ow establisheswith a pressure level practically equal to the ambient pressure(Fig 17 right column top) At a certain pressure ratio closeto the design pressure ratio of the full-length plug nozzle thebase ow suddenly changes its character and turns over to theclosed form characterized by a constant base pressure that isno longer in uenced by the ambient pressure Analyses indi-cate that shorter plug bodies with higher truncations trigger anearlier change in wake ow at pressure ratios below the designpressure ratio At the transition point the pressure within thewake approaches a value that is below ambient pressure andthe full base area induces a negative thrust (Fig 17 right col-umn center) This thrust loss depends on the percentage oftruncation and the total size of the base area Published ex-perimental data and numerical simulations reveal an increasingthrust loss for shorter plug bodies because the total base areaincreases

Beyond the transition point the pressure within the closedwake remains constant At these lower ambient pressures thebase pressure is then higher than the ambient pressure result-ing in a positive thrust contribution of the total base area Per-formance data of a numerically simulated truncated plug noz-zle are shown in Fig 19 and compared to the same plugnozzle with full-length central body and a conventional bellnozzle Design parameters for this truncated plug nozzle arethe same as for the full length plug (see Fig 18 and Table 2)

Figure 20 shows a typical photograph of a sea-level hot-runtest with a truncated toroidal subscale plug nozzle performedat DLR5 Nozzle design data are included in Table 2 For com-parison of experimental results with numerical computations

the ow eld of the toroidal plug nozzle was calculated with anumerical method10 and Fig 20 shows the calculated Mach-number distribution in the combustion chamber and nozzlePrincipal physical processes like expansion waves shocks andthe recirculating base- ow region are in good agreement Boththe experiment and the numerical simulation show that the ow separates from the conical plug body before reaching itstruncated end Recent results on experiments with plug nozzlespublished in Ref 28 also reveal separation of the ow fromthe central plug body for conical contours In contrast no sep-aration was observed for contoured central plug bodies de-signed with the method proposed in Ref 29 Recent numericalsimulations of contoured plug nozzles performed at DLRwithin the ESA Advanced Rocket Propulsion Technology(ARPT) Program on advanced rocket propulsion also showthat no ow separation occurs from well-contoured full-lengthcentral plug bodies26 Principal ow eld developments pre-dicted by these numerical simulations are again in a goodagreement with experimental data published in Ref 28 Withinthe frame of this ESA ARPT Program performance and owbehavior of clustered plug nozzles at different truncations arebeing examined by European industries [Societe Europeennede Propulsion (SEP) Volvo Dasa] and research institutes(ONERA DLR) with subscale cold- ow plug models26 (Table3) Numerous experimental investigations on subscale modelsof circular and clustered plug nozzles with cold- and hot-gas ow were performed at the Keldysh Center with and withoutexternal ows Thrust characteristics pressure and heat- uxdistributions along the plug and acoustic characteristics of thecircular jet were investigated

2 Linear Plug Nozzles

Performance behavior and ow eld development for linearplug nozzles as a function of ambient pressures are in principlesimilar to those of circular plug nozzles (Fig 17) Howeverspecial attention must be paid to the in uence of both endsides where the surrounding ow disturbs the expanding ow- eld resulting in an expansion of the ow normal to the main ow direction and therefore in an effective performance loss

HAGEMANN ET AL 629

Table 3 Design parameters for linear and toroidal subscale plug nozzlesa

Design data Linear plug (FESTIP) Circular clustered plug (ARPT)

Chamber pressure pc 875 bar 80 barDriving gas Air AirMass ow rate mCcedil 595 kgs 546 kgsDesign pressure ratio pc pamb 200 200Geometric area ratio laquo 1265 1265aUsed in the ESA FESTIP and ESA ARPT studies

Fig 20 Plug nozzle ow eld experiment vs numerical simula-tion a) toroidal plug nozzle experiment with gas-oilnitric-acidpropellant combination side view photograph and b) computa-tional results of toroidal plug nozzle Mach number distributionfull gray and isolines centerline section

Fig 21 Design of linear subscale plug nozzle of the ESA FESTIPTechnology Program (plug size 192 mm high 214 mm wide)

Fig 22 Calculated wall pressure data for linear subscale plugnozzle at different pressure ratios

The change of wake ow behavior may be strongly in uencedby the penetration of ambient pressures through both end sidesparticularly for truncated plug nozzles End plates as foreseenfor the linear plug nozzle of the X-33 demonstrator vehiclecould be used to avoid this ambient pressure penetration

Within the framework of the ESA FESTIP Technology Pro-gram lsquolsquoTechnology Developments on Rocket and Air Breath-ing Propulsion for Reusable Launch Vehiclesrsquorsquo subscale cold-gas tests and numerical simulations of linear plug nozzles arebeing performed by Dasa at the high-speed wind tunnel [Hoch-geschwindigheitswind kanal (HWK) in Bad Salzungen Ger-many] of the Technical University of Dresden719 Design pa-

rameters for the linear plug model are included in Table 3Figure 21 shows the baseline model with a linear throat Afterpassing the linear throat the ow expands directly onto theplug surface A second model will be tested with a bell-shapedlinear nozzle extension following the linear throat for internalexpansion prior to further external expansion onto the plugsurface19 The contour of the plug was designed by the methodof Angelino29 and three different plug lengths as 5 20 and40 of the ideal length were tested Figure 22 shows calcu-lated wall pressures for the full-length plug for different am-bient pressures simulating the operational range of sea-leveloperation up to higher-altitude operation The ow structure inform of the left- and right-running characteristics at two pres-sure ratios is shown in Fig 23 For lower pressure ratios thanthe design pressure ratio of pc pamb = 200 the in uence of the

630 HAGEMANN ET AL

Fig 23 Left- and right-running characteristics in ow eld at twopressure ratios pc pamb = a) 8 and b) 200 (inviscid analysis withmethod of characteristics)

Fig 24 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 20 plug body

Fig 25 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 40 plug body

Fig 27 Flow phenomena of an E ndash D nozzle with a) open and b)closed wakes

Fig 26 Sketch of an E ndash D nozzle

system of recompression shocks and expansion waves on thewall pressure distribution (Figs 17 and 22) can be seen bythe waviness of the pressure distribution on the plug surfaceFirst experimental results are shown in Figs 24 and 25 whichalso show the waviness of the pressure distribution for low-pressure ratios

3 Plug Nozzles Final Aspects

The performance of circular and linear plug nozzles is ad-versely affected by the external airstream ow ie the aspi-ration effect of the external ow slightly reduces performanceand favors an earlier change in wake- ow development for

truncated plug nozzles32ndash 35 Cold- ow tests showed that theeffect of external ow on performance is con ned to a narrowrange of external ow speeds with Mach numbers near unity

The altitude compensation capability of circular and linearplug nozzles for higher ambient pressures is indisputable Be-cause plug nozzles lose this capability for pressure ratios abovethe design pressure ratio the latter should be chosen as highas possible Taking this into account plug nozzles will featurean even better overall performance than those shown in Figs18 and 19 Unfortunately for equal geometrical area ratiosplug nozzles perform worse at high altitude than do conven-tional bell nozzles because of truncation and clustering Forhigh-area-ratio nozzles with relatively short lengths plug noz-zles perform better than conventional bell nozzles In additionto having excellent capabilities for altitude compensation plugnozzles have additional advantages including ease in vehicleand engine integration

C E ndash D Nozzles

An expansionndash de ection (E ndash D) nozzle is shown in Fig 26E ndash D nozzles were at one time thought to have capabilities foraltitude compensation because the gas expansion takes placewith a constant pressure free boundary Thus the aerodynamicbehavior of E ndash D nozzles as a function of altitude is in prin-ciple quite similar to plug nozzles because the expansion pro-cess is controlled by the ambient pressure hence by the alti-tude In contrast to plug nozzles however the expansionprocess is controlled from inside the nozzle for E ndash D nozzlesAt low altitude the higher ambient pressure limits the gasexpansion resulting in a low effective expansion area ratioThe exhaust gas is adapted to the ambient pressure level by

HAGEMANN ET AL 631

Fig 30 Performance losses in dual-throat nozzles as a functionof nozzle design parameters (with r4r1 = 352 r2r1 = 116 and mCcedil 2

= 00)

Fig 29 Flow phenomena during a) sea-level and b) high-altitudeoperations

Fig 28 Sketch of a dual-throat nozzle view of combustion cham-ber and throat region

systems of recompression and expansion waves (Fig 27) Athigher altitude the lower ambient pressure allows more gasexpansion within the nozzle resulting in a higher effectiveexpansion area ratio But in contrast to plug nozzles the pres-sure in the wake of the center plug is always less than theambient pressure because of the aspiration effect This occursat low pressure ratios when the wake is opened and results inan aspiration loss Furthermore because the exhaust ow ex-pands to this base pressure rather than to the ambient pressurelevel wall pressures downstream are overexpanded This re-sults in an additional overexpansion loss As the pressure ratioincreases the wake region closes and is thus totally isolatedfrom the ambient environment (see Fig 27) The behavior dur-ing transition from open wake to closed wake is again equalto plug nozzles and the base pressure in the closed-wake re-gion is essentially independent of the ambient pressure

The E ndash D nozzle concept has also been a subject of numer-ous analytical and experimental studies Results from thesestudies show that E ndash D nozzle capabilities for altitude com-pensation are poor and are in fact worse than those of plugnozzles because of aspiration and overexpansion losses35 Forhigh-area-ratio nozzles with a relatively short length an E ndash Dnozzle performs better than a comparable conventional bellnozzle at the same length because of lower divergence andpro le losses than the bell nozzle

The advantages of the E ndash D nozzle concept include its smallengine envelope and no moving parts However like toroidalplug nozzles E ndash D nozzles have the disadvantage of havinghigher throat heat uxes relative to a conventional bell nozzlewith an equal throat area The higher throat heat ux resultsagain from the relatively thin annular throat gap This problemhowever can be remedied with the modular thrust cell clusterconcept Additional advantages of E ndash D nozzles with clusteredthrust cells are the same as those already discussed for theclustered plug nozzle concept with regard to ease in manu-facturing thrust vector control by throttling or shutting off anindividual or group of thrust cells and lower nozzle throatheating

D Nozzles with Throat Area Varied by a Mechanical Pintle

This nozzle concept utilizes a conventional bell nozzle witha xed exit area and a mechanical pintle in the combustionchamber and throat region to vary the throat area and hencethe expansion area ratio The area of the nozzle throatmdashanannulus between the pintle and the shroudmdashis varied by mov-ing the pintle axially

The pintle concept has been used in solid rocket motors asa mean to provide variable thrust The concept in principleallows a continuous variation of the throat area and thus op-timum expansion area ratios throughout a mission Howeverit requires an actuator and a sophisticated control system Theconcept raises issues of engine weight design complexitycooling of the pintle and nozzle throat and reliability

In Ref 36 the aerodynamic performances of nozzles with ve different pintle geometries were calculated and comparedwith a reference bell nozzle The performance losses of thesepintle nozzles when compared with the bell nozzle are in therange of 1ndash 25 Performances of a xed pintle geometry atthree different locations were also calculated The results showthat the performance loss varies with the pintle location

E Dual-Mode Nozzles

Dual-mode rocket engines using one or two fuels offer atrajectory-adapted dual-mode operation during the ascent of alauncher which may be of signi cant advantage for single-stage Earth-to-orbit vehicles as compared to conventionalrocket engines with bell-type nozzles This engine concept in-volves the use of a dense propellant combination with mod-erate performance during liftoff to provide high thrust duringthe initial ight phase and a better performing propellant com-bination in vacuum which result in higher speci c impulse

The fuels are burned in two different combustion chamberswith one located completely inside the other in the case ofengines with dual-throat nozzles or with a conventional bellthrust chamber surrounded by an annular thrust chamber in thecase of dual-expander engines This type of engine has a built-in accelerationndash reduction capability achieved by shuttingdown one of two thrust chambers The total engine thrust isthen provided by the remaining thrust chamber with the useof the total nozzle exit area leading to an increase in speci cimpulse Apart from the indicated bene ts of dual-mode en-gines which will be discussed in more detail later their de-

632 HAGEMANN ET AL

Table 4 Thrust chamber parameters for simulated dual-expander nozzlesa

Operation mode

Aerojet

First (inout) Second (inout)

DLR

First (inout) Second (inout)

Propellants C3H8ndash O2H2ndash O2 mdashH2ndash O2 H2ndash O2H2ndash O2 mdashH2ndash O2

Chamber pressure pc 414207 bar mdash207 bar 200200 bar mdash200 barMixture ratio r 337 mdash7 77 mdash7Exit area ratio laquo 69 146 58 116aInnerouter chambers

Fig 32 Flow phenomena in a dual-expander nozzle during a)sea-level- and b) high-altitude operation

Fig 31 Sketch of a dual-expander nozzle view of combustionchamber and throat region

velopment and construction require considerable technologicaleffort

1 Dual-Throat Nozzles

A dual-throat nozzle con guration is shown in Fig 28 Itconsists of two conventional bell-thrust chambers with onelocated completely inside the other At low altitude the outerthrust chamber operates with the inner thrust chamber runningin parallel In this operation mode the engine has a larger throatproviding a moderate expansion area ratio During the missionthe outer thrust chamber is shut off and operation continueswith only the inner engine In this con guration ow fromthe inner engine expands and attaches supersonically to theouter engine resulting in a higher expansion area ratio for theremainder of the burn Flow phenomena in both operationmodes are included in Fig 29

Hot- red tests at Aerojet were conducted to provide heattransfer data that were very useful for the thermal analysis anddesign of the dual-throat nozzle con guration37 These testsshowed that ow separation occurred in the inner engine noz-zle at higher ratios of outer to inner chamber pressures duringthe rst operation mode with both chambers burning in par-allel The ow separation resulted in a higher heat load to theinner nozzle Subscale tests performed at the Keldysh Center3

have shown that the additional loss caused by the nozzle con-tour discontinuity during vacuum operation with active innerchamber is in the range of 08ndash 4 depending on geometricaldata (see Fig 30) This high-performance loss results from theinteraction of the inner chamber jet with the outer chambernozzle wall3 The decrease of the jet incidence angle on thewall by means of gas injection through the outer chamber re-duces this performance loss by 04ndash 073

2 Dual-Expander Nozzles

A dual-expander nozzle has two concentric thrust chambersand nozzles It consists of a conventional bell thrust chambersurrounded by an annular thrust chamber Both chambers haveshort primary nozzles which end in a common divergent noz-zle extension Figure 31 shows a typical dual-expander nozzlecon guration At low altitude both thrust chambers operatesharing the same exit area which results in a moderate expan-sion area ratio Part way into the mission one thrust chamberis shut off allowing the other nozzle to use the whole exitarea creating a high-expansion-area ratio for the remainder of

the burn In principle the two operation modes are comparableto those of dual-throat nozzles

Numerical simulations of the ow elds in dual-expandernozzles during all operation modes have been performed atAerojet38 DLR39 and the Keldysh Center Aerojet simulationsare based on an engine design using different propellantshydrocarbonoxygen for the inner chamber and hydrogenox-ygen for the outer chamber whereas DLR simulations arebased on a combined vehicleengine analysis using single-fuelsingle-mixture ratio dual-expander engines with hydrogenoxy-gen Table 4 summarizes combustion chamber parameters Flowphenomena observed in numerical simulations are shown in Fig32a for the mode 1 operation with both thrust chambers burn-ing Compression waves are induced near the inner nozzle lipas a result of the strongly inhomogeneous ow character at thepoint where both exhaust gases of inner and outer combustionchamber merge into the common divergent nozzle part Upand down variations of pressure ratios at the lip did not changethis wave formation even in cases of signi cantly lower exitpressures in the inner nozzle compared to the pressure eld ofthe outer nozzle at the lip Further downstream the compres-sion waves interact with the wall resulting in a re ection ofthe compression waves back into the ow eld

During mode 2 operation a strong expansion of the outernozzle ow is observed when the expansion ratio suddenlyincreases at the end of the nozzle lip The ow is directedtoward the axis of symmetry Near the centerline the ow thenturns over to the axial direction inducing a recompressionshock The static pressure rises signi cantly in this recom-pression region on the centerline A sub- and supersonic recir-culation zone establishes in the inner chamber of the dual-expander nozzle Figure 32b emphasizes the essential owpattern in this operation mode

These analyses have shown that dual-expander nozzles pro-duce high performance in both operation modes3839 Figure 33summarizes performance behavior as a function of ight alti-tude for the dual-expander engine simulated at DLR For com-parison performance data are included for two reference bell

HAGEMANN ET AL 633

Fig 33 Performance data for a dual-expander nozzle

nozzles that have the same area ratios as the dual-expandernozzle during its two operation modes Losses because of thenonhomogeneous exit ow and the induced shocks are com-parable with corresponding values of both conventional noz-zles1

Within the design process of dual-expander engines it is nec-essary to take into account that the nonburning chamber isexposed to high heat uxes and pressure oscillations duringmode 2 operation which might reach a high level and lead tostructural disintegration of the inner nozzle structure There-fore a mode-2 operation with bleed gas generation at a ratherlow thrust chamber pressure level in the inner chamber seemsto be advantageous Because of the bleed gas generation in theprimary chamber the overall mass ow rate during this alter-native mode-2 operation is increased In addition the reductionof the effective exit area ratio results in minor impulse deg-radations during this operation mode when compared with themode-2 operation with no bleed gas generation

Several analytical works on SSTO- and two-stage-to-orbit(TSTO)-vehicles using hydrogenpropane or hydrogenmeth-ane as fuels revealed the lowest vehicle dry masses for dual-mode engines in comparison to other engines1439 Other dual-mode engines using hydrogen as the single fuel but using twomixture ratios also revealed some bene ts over conventionalengines for SSTO- and TSTO applications Even a single-fueloperation with constant mixture ratios in both combustionchambers indicated a gain in launcher performance40

V ConclusionsSeveral nozzle concepts that promise gains in performance

over conventional nozzles were discussed in this paper in-cluding performance enhancements achieved by slight modi- cations of existing nozzles eg through cool gas injectioninto the supersonic nozzle part It is shown that signi cantperformance gains result from the adaptation of the exhaust ow to the ambient pressure and special emphasis was givento altitude adaptive nozzle concepts

A number of nozzle concepts with altitude-compensating ca-pability were identi ed and described To assist the selectionof the best nozzle concept for launch vehicle applications theperformance of the nozzles must be characterized This can bedone using computational uid dynamics (CFD) andor cold- ow tests Existing CFD methods that are in use in the aero-space industry and at research institutes have been veri ed fora wide number of sub- and full-scale experiments and providesuf ciently reliable performance determination for the differ-ent nozzle types

Theoretical evaluations numerical simulations and test re-sults showed that the different concepts have real altitude-com-pensating capabilities However the compensation capabilitiesare limited and there are some drawbacks associated with eachof the concepts Additional performance losses are induced inpractically all of these nozzle concepts when compared to anideal expansion mainly because of the nonisentropic effects

such as shock waves and pressure losses in recirculation zonesHowever these additional performance losses are less than1 ndash 3 depending on the different nozzle concepts

Four nozzle concepts have been selected for further evalu-ation under a NASA Aerojet cooperative agreement They in-clude a plug nozzle an E ndash D nozzle a dual-bell nozzle and adual-expander nozzle The nozzles in addition to a referencebell nozzle have been designed by Aerojet for cold ow testsCFD calculations performed at Aerojet demonstrated that allof these nozzles have high and approximately equal perfor-mances at the design points

In Europe special attention is being paid to plug- and dual-bell nozzles Different plug nozzle concepts are being testedin cold- ow tests within the ESA ARPT (clustered plug noz-zle) and ESA FESTIP (linear plug nozzle) contracts In addi-tion to experiments three-dimensional numerical simulationson these nozzle concepts are being performed Additional ex-periments on different nozzle concepts with forced- ow sep-aration devices eg dual-bell nozzles are currently plannedat DLR

Different advanced nozzle concepts have been examined andare under examination now at the Keldysh Center with bothCFD calculations and experiments on the Nozzle DifferentialFacility which has a nozzle performance determination errorof about 005 This high accuracy is achieved by simulta-neously running the test nozzle and a reference nozzle bothmounted on a common longitudinal axis The performancecharacteristics of the reference nozzle are well known and thethrust difference is measured with strain gauges This facilityis equipped with a wind tunnel to determine nozzle thrust per-formance in the presence of external transonic ows withMach numbers up to 12

In addition to aerodynamic performance other technical is-sues (weight cost design thermal management manufactur-ing system performance and reliability) must be addressedFurthermore before a nal decision can be made as to whichnozzle concept offers the greatest bene ts with regard to aneffective payload mass injection combined launcher and tra-jectory calculations must be performed and compared to a ref-erence launcher concept with conventional nozzles Differentnozzle ef ciencies which account for the additional losses ofadvanced rocket nozzles and are extracted from numerical sim-ulations and experiments must be taken into account

At least for some of these nozzle concepts the plug nozzlethe dual-bell nozzle and the dual-expander nozzle bene tswith regard to lower overall launcher masses have been dem-onstrated in the literature Furthermore the plug nozzle con-cept will be the rst in ight of all of these advanced nozzleconcepts with the X-33 demonstrator vehicle

AcknowledgmentPresented as Paper 18 at the 3rd International Symposium

on Space Propulsion Beijing Peoplersquos Republic of ChinaAug 11ndash 13 1997

References1Manski D and Hagemann G lsquolsquoIn uence of Rocket Design Pa-

rameters on Engine Nozzle Ef cienciesrsquorsquo Journal of Propulsion andPower Vol 12 No 1 1996 pp 41ndash 47

2lsquolsquoLiquid Rocket Engine Nozzlesrsquorsquo NASA Space Vehicle DesignCriteria NASA SP-8120 1976

3Dumnov G E Nikulin G Z and Ponomaryov N B lsquolsquoInves-tigation of Advanced Nozzles for Rocket Enginesrsquorsquo Space Rocket En-gines and Power Plants Vol 4 No 142 NIITP 1993 pp 1012 ndash

1018 (in Russian)4Nguyen T V and Pieper J L lsquolsquoNozzle Flow Separationrsquorsquo Pro-

ceedings of the 5th International Symposium of Propulsion in SpaceTransportation (Paris France) 1996

5Manski D lsquolsquoClustered Plug Nozzles for Future European Reus-able Rocket Launchersrsquorsquo German Aerospace Research Establishment643-817 Lampoldshausen Germany 1981 (in German)

6Horn M and Fisher S lsquolsquoDual-Bell Altitude Compensating Noz-

HAGEMANN ET AL 623

Fig 6 Experimental and numerical data on wall pressure ratiosin overexpanding conical nozzles with ow separation Cold-gastest case with subscale conical nozzle

Fig 7 Experimental and numerical data on wall pressure ratiosin overexpanding bell nozzles with ow separation Hot-gas testcase with J-2S engine

with the centerline The minimum and maximum separationpoints are predicted using the upper and lower separation datathat result from the applied separation model In the case ofdistinct separation points as obtained with the Summer eld- orSchmucker-criterion a certain scattering around the predictedseparation point is assumed The side-load is then calculatedby integrating the wall pressure in the asymmetrical ow sep-aration region49 These approaches gave reasonable results forthe J2-S and Vulcain 1 nozzle

At the Keldysh Center in Russia model experiments of sep-arated ow effects on side-loads acting on nozzles began inthe early 1980s11 These investigations were directed mainlyto the determination of the unsteady pressure uctuations inthe separation zone on the whole engine The method is de-scribed in Ref 11 in more detail it is based on a generalizationof obtained empirical data for the spectral density of the pul-sating pressure eld in the separation zone Distribution of thepressure amplitudes is governed by probability density func-tions This method was validated with the experimental rms-val-ues of side-loads acting on the RD-0120 nozzle (CABD hydro-genndash oxygen engine rst-stage engine of Russian Energialauncher) Other hypotheses for the appearance and predictionof side-loads also exist eg based on aeroelastic analyses13

which account for the coupling of ow oscillations with thinnozzle shells However these hypotheses require experimentalveri cation and validation

Side-loads are an undesired phenomenon that may result inthe destruction of the rocket nozzle and should therefore beavoided In Ref 9 the destruction of a J-2D engine (Rocket-dyne hydrogen ndash oxygen engine second- and third-stage engineof Saturn-5 launcher) as a result of side-loads is reported The

transient startup of the rocket engine generates a nozzle owwith uncontrolled ow separation and with possible side-loadgeneration in the nozzle whereas maximum area ratios of all rst-stage nozzles or booster nozzles are chosen to avoid owseparation at the nominal chamber pressure operation As aresult the vacuum performance of rocket engines that operateduring the entire launcher trajectory such as the SSME orVulcain 1 engines is limited

B Potential Performance Improvements

Compared to existing rocket engines a gain in performanceis achieved with advanced engines such as mixed-mode pro-pulsion systems dual-mixture ratio engines or dual-expanderengines Nevertheless the upgrade of existing engines withbetter performing subsystems such as turbines and pumpsalso leads to a gain in overall performance data and is dis-cussed in more detail in Ref 14 Nozzle performance of con-ventional rocket engines is already very high with regard tointernal loss effects (friction nonuniformity) However fornozzles of gas-generator open-cycle engines such as the Vul-cain 1 engine a slight improvement in performance can beachieved with turbine exhaust gas (TEG) injection into themain nozzle as realized in the F-1 (kerosenendash oxygen engine rst stage of Saturn-5 launcher) and J-2S and it is foreseenfor the Vulcain 2 engine (hydrogen ndash oxygen engine upgradeof Vulcain 1 engine) and con rmed by numerical simula-tions131516 and experimental results3 This is mainly achievedthrough lower friction losses in the main nozzle1315 and be-cause the bypass nozzles used for the expansion of the TEGswhich have higher divergence losses are removed

Despite the slight performance gain by TEG injection thelow-pressure near-wall stream of the injected gas favors a re-duction of the critical pressure ratio at which ow separationoccurs and therefore an earlier nozzle ow separation312 Fur-thermore the presence of the secondary exhaust gas injectioncomplicates nozzle-contouring methods with regard to theavoidance of uncontrolled ow separation for rst-stage orbooster nozzles

III Altitude Adaptive NozzlesA critical comparison of performance losses shown in Table

1 reveals that most signi cant improvements in nozzle perfor-mance for rst-stage or SSTO engines can be achieved throughthe adaptation of nozzle exit pressures to the variations in am-bient pressure during the launcherrsquos ascent through the atmo-sphere Various concepts have been previously mentioned andwill be discussed in detail in the following text

A Nozzles with Devices for Controlled Flow Separation

Several nozzle concepts with devices for controlled owseparation have been proposed in the literature with primaryemphasis on the reduction of side-loads during sea level orlow-altitude operation But the application of these conceptsalso results in an improved performance through the avoidanceof signi cant overexpansion of the exhaust ow

1 Dual-Bell Nozzle

This nozzle concept was rst studied at the Jet PropulsionLaboratory17 in 1949 In the late 1960s Rocketdyne patentedthis nozzle concept which has received attention in recentyears in the US and Europe671819 Figure 8 illustrates thedesign of this nozzle concept with its typical inner base nozzlethe wall in ection and the outer nozzle extension This nozzleconcept offers an altitude adaptation achieved only by nozzlewall in ection In low altitudes controlled and symmetrical ow separation occurs at this wall in ection (Fig 9) whichresults in a lower effective area ratio For higher altitudes thenozzle ow is attached to the wall until the exit plane and thefull geometrical area ratio is used Because of the higher arearatio an improved vacuum performance is achieved Howeveradditional performance losses are induced in dual-bell nozzles

624 HAGEMANN ET AL

Fig 8 Sketch of a dual-bell nozzle

Fig 9 Flow eld phenomena in dual-bell nozzles a) sea-levelmode with ow separation at the wall in ection point and b) al-titude mode with a full- owing nozzle

Fig 10 Performance data of a dual-bell nozzle Performance iscompared with two baseline bell-type nozzles as function of ightaltitude (baseline nozzle 1 same area ratio as dual-bell base noz-zle baseline nozzle 2 same area ratio as nozzle extension)

Fig 11 Performance characteristics of a dual-bell nozzle Per-formance is compared with two baseline bell-type nozzles as func-tion of ight altitude (baseline nozzle 1 same area ratio as dual-bell base nozzle baseline nozzle 2 same area ratio as nozzleextension)

as compared with two baseline nozzles having the same arearatio as the dual-bell nozzle at its wall in ection and in its exitplane Figures 10 and 11 illustrate the performance of a dual-bell nozzle as a function of ight altitude in comparison withboth baseline bell-type nozzles (Design parameters of thedual-bell nozzle are taken from a launcher analysis publishedin Ref 18 Propellants hydrogenoxygen r = 6 pc = 200 barrt = 007 m wall in ection at area ratio laquoB = 30 and totalarea ratio laquoE = 100) The pressure within the separated owregion of the dual-bell nozzle extension at sea-level operationis slightly below the ambient pressure inducing a thrust lossreferred to as lsquolsquoaspiration dragrsquorsquo In addition ow transitionoccurs before the optimum crossover point which leads tofurther thrust loss as compared to an ideal switchover Thenonoptimum contour of the full owing dual-bell nozzle re-sults in further losses at high altitudes

To gain insight into the performance and ow behavior ofdual-bell nozzles at different ambient pressures extensive nu-merical simulations with parametrical variations of contour de-sign parameters were performed18 An optimized bell nozzlewith equal total length and area ratio was used as the referencenozzle for comparison As a result the vacuum performance ofthe dual-bell nozzles has a degradation because of the imper-fect contour and this additional loss has the same order ofmagnitude as the divergence loss of the optimized bell nozzle

The simulations of sea-level operation also revealed an ad-ditional performance loss because of aspiration drag which isless than 3 for these dual-bell nozzles This additional lossdepends linearly on the ambient pressure and therefore it isreduced during the ascent of the launcher Furthermore thesesimulations showed that the application of commonly usedseparation criteria derived for conventional nozzles and ap-

plied to dual-bell nozzles with their wall in ection to estimatethe critical pressure ratio yields reasonable results with an ac-curacy of rsquo15 Experimental data obtained at the KeldyshCenter revealed that the critical pressure ratio at the wall in- ection is about 15ndash 20 less than the critical pressure ratioof a conventional nozzle and that usual separation criteriamust be corrected for an accurate prediction of the critical wallpressure ratio

Flow transition behavior in dual-bell nozzles strongly de-pends on the contour type of the nozzle extension3618 A sud-den transition from sea-level to vacuum operation can be atleast theoretically achieved by two different extensions witha zero wall pressure gradient (constant pressure extension) ora positive wall pressure gradient (overturned extension) But acritical analysis of the transition behavior considering decreas-ing ambient pressures during the launcher ascent revealed thata considerable time with uncontrolled ow separation withinthe nozzle extension exists even for these types of extensionsThe duration of this period can be reduced drastically by throt-tling the chamber pressure18

The main advantage of dual-bell nozzles as compared toother means of controlling nozzle ow separation is its sim-plicity because of the absence of any movable parts and there-fore its high reliability It is necessary to note that the external ow over the vehicle in ight reduces the pressure in the ve-hicle base region where engines are installed The ambient

HAGEMANN ET AL 625

Fig 13 Principle performance characteristics of the nozzle withinsert

Fig 12 RD-0120 nozzle hardware with removed nozzle insertand sketch of secondary nozzle mounted inside of the RD-0120nozzle (photographs taken from Ref 23)

pressure triggering the ow transition is the vehicle base pres-sure instead of the atmospheric pressure at the speci c ightaltitude As the base pressure is lower than the atmosphericpressure the nozzle ow transition occurs at a lower altitudethan the one showed in Fig 11 which slightly decreases theef ciency of the dual-bell nozzle operation along the trajectory

Despite the additional losses induced in dual-bell nozzlesthey still provide a signi cant net impulse gain over the entiretrajectory as compared to conventional bell nozzles Indepen-dent combined launcher and trajectory analyses performed byDasa within the European Space Agency Future EuropeanSpace Transportation Investigations Program (ESA FESTIP)study19 and at DLR on SSTO vehicles powered with dual-bellnozzles result in a signi cant payload gain when comparedwith a reference launcher equipped with conventional nozzles

2 Nozzles with Fixed Inserts

A trip ring attached to the inside of a conventional nozzledisturbs the turbulent boundary layer and causes ow separa-tion at higher ambient pressures At higher altitudes with lowerambient pressures the ow reattaches to the wall behind thetrip ring and full owing of the nozzle is achieved The tran-sition from sea level to vacuum mode depends on the wallpressure near the trip-ring location and on the disturbance in-duced by the trip ring The size of the trip ring is a compromisebetween stable ow separation during sea-level operation andthe induced performance loss during vacuum operation In Ref9 it is reported that a trip-ring size of 10 of the local bound-ary-layer thickness is suf cient to ensure stable ow separa-tion

In principle this concept is similar to the dual-bell nozzleconcept with regard to performance characteristics as shownin Fig 11 The sea-level performance of this nozzle concept islower than the performance of a conventional bell nozzle trun-cated at the trip-ring location because of the aspiration dragin the separated ow region of the nozzle Furthermore at sealevel the bell nozzle with trip rings has even higher divergencelosses than a comparable dual-bell nozzle because the nozzlecontour upstream of the obstacle differs from the optimal con-tour for this low-area ratio as a result of the bell nozzle designfor best vacuum performance The additional losses inducedduring vacuum operation is about 1 compared with the per-formance of the bell nozzle without an obstacle Thus theadditional losses are comparable to the additional losses in-duced in dual-bell nozzles As for dual-bell nozzles the tran-sition behavior of this nozzle concept is uncertain but it willbe even more uncertain than for a dual-bell nozzle with a con-stant pressure or overturned nozzle extension

In principle several altitude adaptations can be achievedwith one nozzle by various trip rings mounted one behind theother However this results in increasing vacuum performancelosses The trip rings can also be attached into existing nozzlesand therefore represent a low-cost concept at least for testpurposes with low technological risk Trip rings have beendemonstrated to be effective for side-load reductions duringthe transient startup of rocket engines20 The main problemswith trip-ring nozzles are not only performance losses but alsoring resistance in high-temperature boundary layers the exactcircumferential xing and the uncertainties in the transitionbehavior These uncertainties might be why active interest inthis nozzle concept in the 1970s which is documented in var-ious publications920ndash22 has disappeared in recent years

A further concept with a xed wall discontinuity is a nozzlewith a circumferential groove The aerodynamic behavior of anozzle with a groove and nozzle with trip rings are quite sim-ilar9 but this concept is hard to realize in the case of thinnozzle shells which are used in all of todayrsquos rocket nozzles

3 Nozzles with Temporary Inserts

Nozzle concepts with xed wall discontinuities have the dis-advantage of lower vacuum performance as compared with a

conventional bell nozzle with equal design and operation dataA promising concept for controlled ow separation is thereforetemporary inserts which are removed for vacuum operationThese inserts can be either ablative or ejectible The insertsmay have the form of a complete secondary nozzle23 (Fig 12)or of small steps attached inside the nozzle wall In case ofejectible inserts a reliable mechanism is needed to provide asudden and symmetrical detachment In any case shocks areinduced during the transient ejection because the inserts act asan obstacle in the supersonic exhaust ow These shocks alsointeract with the nozzle walls and increase pressure loads onthe wall and local heat uxes A nonsymmetrical ejectionwould then result in the generation of side-loads Furthermorethe danger of a downstream collision with the nozzle wallarises because the inserts might also experience a transversalmovement toward the walls

Recent hot- ring tests performed in Russia with a modi edRD-0120 engine equipped with a secondary nozzle insert re-vealed a signi cant performance gain of 12 during the sea-level operation at 100 chamber pressure compared with theoriginal RD-0120 performance23 Nominal chamber pressureof this engine is pc = 206 bar with an area ratio of laquo = 857These full-scale hot- ring tests demonstrated the durability ofmaterials sealings and the release mechanism and thus thefeasibility of this concept Figure 12 shows the nozzle hard-ware and a sketch of the secondary nozzle mounted inside theRD-0120 nozzle23

The principle performance characteristics of this RD-0120nozzle with ejectible insert are included in Fig 13 The nozzleoperation with insert results in a slight performance loss com-pared with an ideal bell nozzle with the same reduced area

626 HAGEMANN ET AL

Fig 14 Sketch of a two-position nozzle during a) sea-level andb) vacuum operations

ratio because of aspiration drag and the presumably nonoptim-ized insert contour The performance degradation is compara-ble to the one induced in dual-bell nozzles during sea-leveloperation (Fig 11) In vacuum or high-altitude operations thehigher performance of the baseline nozzle is achieved

Another method for removing the inserts is to use combus-tible or ablative elements924 During the ascent of the launcherthe size of the insert is continuously reduced until it is com-pletely consumed resulting in a full owing bell nozzle witha clean contour for best vacuum performance The principaluncertainties of this nozzle concept are the stability and surfaceregression rates of the inserts Furthermore a homogeneoussymmetrical and temporally de ned consumption must beguaranteed despite possible local pressure and temperature uctuations near the nozzle walls This is currently highly un-certain

4 Nozzles with Active or Passive Secondary Gas Injection

With this nozzle concept the ow in an overexpanding noz-zle is forced to separate at a desired location by injecting asecond uid into the gas stream in the wall normal directionto achieve maximum disturbance of the main exhaust gas andthus to induce ow separation This secondary gas injectionis quite different from the aforementioned TEG injectionwhich should be injected with few or no disturbances to themain ow The injection could be either active that is forcedthrough gas expansion from a higher-pressure reservoir or pas-sive using holes in the wall through which ambient gas issucked in (vented nozzle concept) The latter concept can onlywork in case of gas pressures near the wall inside the nozzlethat are lower than the ambient pressure Experience on forcedsecondary gas injection gained at Aerojet shows that a largeamount of injected uid is required to induce a signi cant owseparation Furthermore no net speci c impulse gain is real-ized when considering the additional gas ow rate

In the vented nozzle concept segments of the bell nozzlewall have various slots or holes opened to the outside ambientpressure During low-altitude operation the slots are opened totrigger ow separation within the nozzle whereas the slots areclosed during high-altitude or vacuum operation so that thegas fully expands inside the entire nozzle Experiments with amodi ed RL10A-3 engine equipped with this nozzle conceptwere performed at Pratt amp Whitney25 Performance resultsshowed that over a small range of low-pressure ratios the per-forated nozzle performed as well as a nozzle with its area ratiotruncated immediately upstream of the vented area Howeverat above some intermediate pressure ratios the thrust ef -ciency suddenly dropped and approached that of the full- ow-ing nozzle The measured performance characteristics werequite similar to those of the dual-bell nozzles shown in Fig11

The altitude range of this nozzle concept is limited by thenumber and position of the holes because the pressure withinthe nozzle must be lower than the ambient pressure Further-more as the rocket aft-body base pressure is lower than thealtitude atmospheric pressure in the surrounding ambient ows the nozzle ow transition occurs at a lower altitude andthus the range of compensation is further reduced

5 Two-Position or Extendible Nozzles

Nozzles of this type with extendible exit cones are currentlyused only for rocket motors of upper stages to reduce the pack-age volume for the nozzle eg at present for solid rocketengines such as the inertial upper stage (IUS) or for the liquidrocket engine RL10 The main idea of the extendible extensionis to use a truncated nozzle with low expansion in low- ightaltitudes and to have a higher nozzle extension at high alti-tudes Figure 14 illustrates this nozzle concept Its capabilityfor altitude compensation is indisputable and the nozzle per-formance is easily predictable The whole nozzle contour in-cluding the extendible extension is contoured for maximum

performance at the high-area ratio with either the method ofcharacteristics or a variational method The area ratio of the rst nozzle section is then determined and the nozzle contouris divided into two parts The xed nozzle part and the ex-tendible extension Investigations conducted at the KeldyshCenter have shown that this nozzle-contouring method is notonly the simplest but also provides a good overall trajectoryperformance3 A minor performance loss is incorporated duringlow-altitude operations because of the truncated inner nozzlewhich has a nonoptimal contour for this interim exit area ratioThe performance characteristics as a function of ight altitudeare similar to those of the nozzle with the ejectible insert asshown in Fig 13 A tradeoff study performed at the KeldyshCenter on nozzles with an ejectible insert and an extendibleextension showed that the thrust characteristics of the nozzlewith an ejectible insert are slightly better than those of theextendible nozzle concept with the same overall dimensionsfor both nozzles

The main drawback of the extendible nozzle concept is thatit requires mechanical devices for the deployment of the ex-tension which reduces engine reliability and increases totalengine mass The necessity for active cooling of the extendibleextension requires exible or movable elements in the coolingsystem which also reduces system reliability The in uencesof the external ow and the generated jet noise on the extend-ible extension in the retracted initial position have not yetbeen fully investigated Former investigations conducted in theKeldysh Center showed that the external ow causes bothsteady and unsteady pressure loads on the retracted nozzle ex-tension whereas the engine jet noise causes strong vibrationsof the nozzle extension

B Plug Nozzles

Various experimental analytical and numerical research onplug nozzles have been performed since the 1950s in theUS4532ndash 35 Europe5710192627 Russia3 and Japan28 In contrastto the previously discussed nozzle concepts plug nozzles pro-vide at least theoretically a continuous altitude adaptation upto their geometrical area ratio Figures 15 and 16 show twodifferent design approaches for circular plug nozzles whichdiffer only in the chamber and primary nozzle layout A con-ical central body is shown here which could also be designedwith more sophisticated contouring methods2930 Different de-sign approaches include plug nozzles with a toroidal chamberand throat (with and without truncation) and plug nozzles witha cluster of circular bell nozzle modules or with clusteredquasirectangular nozzle modules The latter approach seems to

HAGEMANN ET AL 627

Fig 17 Flow phenomena of a plug nozzle with full length (left column) and truncated central body (right column) at different pressureratios pc pamb off-design (top bottom) and design (center) pressure ratio

Fig 16 Principle design of plug nozzles clustered plug 36 mod-ules with truncated plug body

Fig 15 Principle design of plug nozzles toroidal plug fulllength

be advantageous because further losses induced by the gapsbetween individual modules and the ow eld interactionsdownstream of the module exits can be minimized It has beenshown that transition from a round to a square nozzle resultsin a very small performance loss31 In principle the ow elddevelopment of a clustered plug nozzle with rectangular nozzlemodules is similar to that of a toroidal plug nozzle but avoidsthe inherent disadvantages of the toroidal plug design regard-ing 1) the control of a constant throat gap during manufactur-

ing and thermal expansion (side-loads and thrust vector devi-ations) 2) the cooling of toroidal throat with tiny throat gapsand 3) the control of combustion instabilities in the toroidalcombustion chamber Another plug nozzle con guration is thelinear plug nozzle which is foreseen for the propulsion systemof the RLV X-33 concept

1 Circular Plug Nozzles

Figure 17 summarizes the principle ow phenomena of cir-cular plug nozzles with full length and truncated central bodiesat different off-design (top and bottom) and design (center)pressure ratios that were observed in experiments and numer-ical simulations For pressure ratios lower than the design pres-sure ratio of a plug nozzle with a well-contoured central bodythe ow expands near the central plug body without separa-tion and a system of recompression shocks and expansionwaves adapts the exhaust ow to the ambient pressure pambThe characteristic barrel-like form with several in ections ofthe shear-layer results from various interactions of compres-sion and expansion waves with the shear layer and turbulentdiffusion enlarges the shear layer farther downstream of thethroat The existence of the overexpansion and recompressionprocesses is inferred from up- and down-variations of plugwall pressure pro les observed in various cold- ow tests andnumerical simulations and will also be shown later for linearplug nozzles

At the design pressure ratio (see Fig 17 left column cen-ter) the characteristic with the design Mach number should bea straight line emanating to the tip of the central plug bodyand the shear layer is parallel to the centerline However forcircular plug nozzles designed with contouring methods pro-posed in Refs 29 and 30 no exact one-dimensional exit owpro le can be achieved because both methods use thePrandtlndash Meyer relations that are only valid for planar owsFurthermore nonhomogeneous ow in the throat regionwhich is in general not considered within the contour designprocess also in uences the exit ow pro le The wall pressuredistribution remains constant at pressure ratios above the de-sign pressure ratio ie the plug nozzle behaves like a con-ventional nozzle the loss of its capability of further altitude

628 HAGEMANN ET AL

Table 2 Design parameters of toroidal plug nozzles

Design data Full-scale nozzlea Subscale nozzleb

Chamber pressure pc 100 bar 12 barPropellants Hydrogenoxygen Gasndash oilnitric-acidMixture ratio r 60 25Inner chamber diameter dcinner 43 m 019 mOuter chamber diameter dcouter 44 m 034 mGeometric area ratio laquo 55 10aUsed for numerical simulations and performance predictions see Fig 18

bUsed in experiments see Fig 20

Fig 18 Performance of numerically simulated plug nozzle withfull-length central body

Fig 19 Performance of numerically simulated plug nozzle withtruncated central body

adaptation being included Figure 17 (left column bottom) il-lustrates the ow eld at higher pressure ratios Performancedata of a typical plug nozzle are included in Fig 18 and com-pared to a conventional bell nozzle with equal area ratio Thedesign and combustion chamber parameters are included in theleft column of Table 227

The truncation of the central plug body which is an advan-tage because of the huge length and high structural mass ofthe well-contoured central body results in a different ow andperformance behavior as compared to the full-length plug noz-zle At lower pressure ratios an open wake ow establisheswith a pressure level practically equal to the ambient pressure(Fig 17 right column top) At a certain pressure ratio closeto the design pressure ratio of the full-length plug nozzle thebase ow suddenly changes its character and turns over to theclosed form characterized by a constant base pressure that isno longer in uenced by the ambient pressure Analyses indi-cate that shorter plug bodies with higher truncations trigger anearlier change in wake ow at pressure ratios below the designpressure ratio At the transition point the pressure within thewake approaches a value that is below ambient pressure andthe full base area induces a negative thrust (Fig 17 right col-umn center) This thrust loss depends on the percentage oftruncation and the total size of the base area Published ex-perimental data and numerical simulations reveal an increasingthrust loss for shorter plug bodies because the total base areaincreases

Beyond the transition point the pressure within the closedwake remains constant At these lower ambient pressures thebase pressure is then higher than the ambient pressure result-ing in a positive thrust contribution of the total base area Per-formance data of a numerically simulated truncated plug noz-zle are shown in Fig 19 and compared to the same plugnozzle with full-length central body and a conventional bellnozzle Design parameters for this truncated plug nozzle arethe same as for the full length plug (see Fig 18 and Table 2)

Figure 20 shows a typical photograph of a sea-level hot-runtest with a truncated toroidal subscale plug nozzle performedat DLR5 Nozzle design data are included in Table 2 For com-parison of experimental results with numerical computations

the ow eld of the toroidal plug nozzle was calculated with anumerical method10 and Fig 20 shows the calculated Mach-number distribution in the combustion chamber and nozzlePrincipal physical processes like expansion waves shocks andthe recirculating base- ow region are in good agreement Boththe experiment and the numerical simulation show that the ow separates from the conical plug body before reaching itstruncated end Recent results on experiments with plug nozzlespublished in Ref 28 also reveal separation of the ow fromthe central plug body for conical contours In contrast no sep-aration was observed for contoured central plug bodies de-signed with the method proposed in Ref 29 Recent numericalsimulations of contoured plug nozzles performed at DLRwithin the ESA Advanced Rocket Propulsion Technology(ARPT) Program on advanced rocket propulsion also showthat no ow separation occurs from well-contoured full-lengthcentral plug bodies26 Principal ow eld developments pre-dicted by these numerical simulations are again in a goodagreement with experimental data published in Ref 28 Withinthe frame of this ESA ARPT Program performance and owbehavior of clustered plug nozzles at different truncations arebeing examined by European industries [Societe Europeennede Propulsion (SEP) Volvo Dasa] and research institutes(ONERA DLR) with subscale cold- ow plug models26 (Table3) Numerous experimental investigations on subscale modelsof circular and clustered plug nozzles with cold- and hot-gas ow were performed at the Keldysh Center with and withoutexternal ows Thrust characteristics pressure and heat- uxdistributions along the plug and acoustic characteristics of thecircular jet were investigated

2 Linear Plug Nozzles

Performance behavior and ow eld development for linearplug nozzles as a function of ambient pressures are in principlesimilar to those of circular plug nozzles (Fig 17) Howeverspecial attention must be paid to the in uence of both endsides where the surrounding ow disturbs the expanding ow- eld resulting in an expansion of the ow normal to the main ow direction and therefore in an effective performance loss

HAGEMANN ET AL 629

Table 3 Design parameters for linear and toroidal subscale plug nozzlesa

Design data Linear plug (FESTIP) Circular clustered plug (ARPT)

Chamber pressure pc 875 bar 80 barDriving gas Air AirMass ow rate mCcedil 595 kgs 546 kgsDesign pressure ratio pc pamb 200 200Geometric area ratio laquo 1265 1265aUsed in the ESA FESTIP and ESA ARPT studies

Fig 20 Plug nozzle ow eld experiment vs numerical simula-tion a) toroidal plug nozzle experiment with gas-oilnitric-acidpropellant combination side view photograph and b) computa-tional results of toroidal plug nozzle Mach number distributionfull gray and isolines centerline section

Fig 21 Design of linear subscale plug nozzle of the ESA FESTIPTechnology Program (plug size 192 mm high 214 mm wide)

Fig 22 Calculated wall pressure data for linear subscale plugnozzle at different pressure ratios

The change of wake ow behavior may be strongly in uencedby the penetration of ambient pressures through both end sidesparticularly for truncated plug nozzles End plates as foreseenfor the linear plug nozzle of the X-33 demonstrator vehiclecould be used to avoid this ambient pressure penetration

Within the framework of the ESA FESTIP Technology Pro-gram lsquolsquoTechnology Developments on Rocket and Air Breath-ing Propulsion for Reusable Launch Vehiclesrsquorsquo subscale cold-gas tests and numerical simulations of linear plug nozzles arebeing performed by Dasa at the high-speed wind tunnel [Hoch-geschwindigheitswind kanal (HWK) in Bad Salzungen Ger-many] of the Technical University of Dresden719 Design pa-

rameters for the linear plug model are included in Table 3Figure 21 shows the baseline model with a linear throat Afterpassing the linear throat the ow expands directly onto theplug surface A second model will be tested with a bell-shapedlinear nozzle extension following the linear throat for internalexpansion prior to further external expansion onto the plugsurface19 The contour of the plug was designed by the methodof Angelino29 and three different plug lengths as 5 20 and40 of the ideal length were tested Figure 22 shows calcu-lated wall pressures for the full-length plug for different am-bient pressures simulating the operational range of sea-leveloperation up to higher-altitude operation The ow structure inform of the left- and right-running characteristics at two pres-sure ratios is shown in Fig 23 For lower pressure ratios thanthe design pressure ratio of pc pamb = 200 the in uence of the

630 HAGEMANN ET AL

Fig 23 Left- and right-running characteristics in ow eld at twopressure ratios pc pamb = a) 8 and b) 200 (inviscid analysis withmethod of characteristics)

Fig 24 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 20 plug body

Fig 25 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 40 plug body

Fig 27 Flow phenomena of an E ndash D nozzle with a) open and b)closed wakes

Fig 26 Sketch of an E ndash D nozzle

system of recompression shocks and expansion waves on thewall pressure distribution (Figs 17 and 22) can be seen bythe waviness of the pressure distribution on the plug surfaceFirst experimental results are shown in Figs 24 and 25 whichalso show the waviness of the pressure distribution for low-pressure ratios

3 Plug Nozzles Final Aspects

The performance of circular and linear plug nozzles is ad-versely affected by the external airstream ow ie the aspi-ration effect of the external ow slightly reduces performanceand favors an earlier change in wake- ow development for

truncated plug nozzles32ndash 35 Cold- ow tests showed that theeffect of external ow on performance is con ned to a narrowrange of external ow speeds with Mach numbers near unity

The altitude compensation capability of circular and linearplug nozzles for higher ambient pressures is indisputable Be-cause plug nozzles lose this capability for pressure ratios abovethe design pressure ratio the latter should be chosen as highas possible Taking this into account plug nozzles will featurean even better overall performance than those shown in Figs18 and 19 Unfortunately for equal geometrical area ratiosplug nozzles perform worse at high altitude than do conven-tional bell nozzles because of truncation and clustering Forhigh-area-ratio nozzles with relatively short lengths plug noz-zles perform better than conventional bell nozzles In additionto having excellent capabilities for altitude compensation plugnozzles have additional advantages including ease in vehicleand engine integration

C E ndash D Nozzles

An expansionndash de ection (E ndash D) nozzle is shown in Fig 26E ndash D nozzles were at one time thought to have capabilities foraltitude compensation because the gas expansion takes placewith a constant pressure free boundary Thus the aerodynamicbehavior of E ndash D nozzles as a function of altitude is in prin-ciple quite similar to plug nozzles because the expansion pro-cess is controlled by the ambient pressure hence by the alti-tude In contrast to plug nozzles however the expansionprocess is controlled from inside the nozzle for E ndash D nozzlesAt low altitude the higher ambient pressure limits the gasexpansion resulting in a low effective expansion area ratioThe exhaust gas is adapted to the ambient pressure level by

HAGEMANN ET AL 631

Fig 30 Performance losses in dual-throat nozzles as a functionof nozzle design parameters (with r4r1 = 352 r2r1 = 116 and mCcedil 2

= 00)

Fig 29 Flow phenomena during a) sea-level and b) high-altitudeoperations

Fig 28 Sketch of a dual-throat nozzle view of combustion cham-ber and throat region

systems of recompression and expansion waves (Fig 27) Athigher altitude the lower ambient pressure allows more gasexpansion within the nozzle resulting in a higher effectiveexpansion area ratio But in contrast to plug nozzles the pres-sure in the wake of the center plug is always less than theambient pressure because of the aspiration effect This occursat low pressure ratios when the wake is opened and results inan aspiration loss Furthermore because the exhaust ow ex-pands to this base pressure rather than to the ambient pressurelevel wall pressures downstream are overexpanded This re-sults in an additional overexpansion loss As the pressure ratioincreases the wake region closes and is thus totally isolatedfrom the ambient environment (see Fig 27) The behavior dur-ing transition from open wake to closed wake is again equalto plug nozzles and the base pressure in the closed-wake re-gion is essentially independent of the ambient pressure

The E ndash D nozzle concept has also been a subject of numer-ous analytical and experimental studies Results from thesestudies show that E ndash D nozzle capabilities for altitude com-pensation are poor and are in fact worse than those of plugnozzles because of aspiration and overexpansion losses35 Forhigh-area-ratio nozzles with a relatively short length an E ndash Dnozzle performs better than a comparable conventional bellnozzle at the same length because of lower divergence andpro le losses than the bell nozzle

The advantages of the E ndash D nozzle concept include its smallengine envelope and no moving parts However like toroidalplug nozzles E ndash D nozzles have the disadvantage of havinghigher throat heat uxes relative to a conventional bell nozzlewith an equal throat area The higher throat heat ux resultsagain from the relatively thin annular throat gap This problemhowever can be remedied with the modular thrust cell clusterconcept Additional advantages of E ndash D nozzles with clusteredthrust cells are the same as those already discussed for theclustered plug nozzle concept with regard to ease in manu-facturing thrust vector control by throttling or shutting off anindividual or group of thrust cells and lower nozzle throatheating

D Nozzles with Throat Area Varied by a Mechanical Pintle

This nozzle concept utilizes a conventional bell nozzle witha xed exit area and a mechanical pintle in the combustionchamber and throat region to vary the throat area and hencethe expansion area ratio The area of the nozzle throatmdashanannulus between the pintle and the shroudmdashis varied by mov-ing the pintle axially

The pintle concept has been used in solid rocket motors asa mean to provide variable thrust The concept in principleallows a continuous variation of the throat area and thus op-timum expansion area ratios throughout a mission Howeverit requires an actuator and a sophisticated control system Theconcept raises issues of engine weight design complexitycooling of the pintle and nozzle throat and reliability

In Ref 36 the aerodynamic performances of nozzles with ve different pintle geometries were calculated and comparedwith a reference bell nozzle The performance losses of thesepintle nozzles when compared with the bell nozzle are in therange of 1ndash 25 Performances of a xed pintle geometry atthree different locations were also calculated The results showthat the performance loss varies with the pintle location

E Dual-Mode Nozzles

Dual-mode rocket engines using one or two fuels offer atrajectory-adapted dual-mode operation during the ascent of alauncher which may be of signi cant advantage for single-stage Earth-to-orbit vehicles as compared to conventionalrocket engines with bell-type nozzles This engine concept in-volves the use of a dense propellant combination with mod-erate performance during liftoff to provide high thrust duringthe initial ight phase and a better performing propellant com-bination in vacuum which result in higher speci c impulse

The fuels are burned in two different combustion chamberswith one located completely inside the other in the case ofengines with dual-throat nozzles or with a conventional bellthrust chamber surrounded by an annular thrust chamber in thecase of dual-expander engines This type of engine has a built-in accelerationndash reduction capability achieved by shuttingdown one of two thrust chambers The total engine thrust isthen provided by the remaining thrust chamber with the useof the total nozzle exit area leading to an increase in speci cimpulse Apart from the indicated bene ts of dual-mode en-gines which will be discussed in more detail later their de-

632 HAGEMANN ET AL

Table 4 Thrust chamber parameters for simulated dual-expander nozzlesa

Operation mode

Aerojet

First (inout) Second (inout)

DLR

First (inout) Second (inout)

Propellants C3H8ndash O2H2ndash O2 mdashH2ndash O2 H2ndash O2H2ndash O2 mdashH2ndash O2

Chamber pressure pc 414207 bar mdash207 bar 200200 bar mdash200 barMixture ratio r 337 mdash7 77 mdash7Exit area ratio laquo 69 146 58 116aInnerouter chambers

Fig 32 Flow phenomena in a dual-expander nozzle during a)sea-level- and b) high-altitude operation

Fig 31 Sketch of a dual-expander nozzle view of combustionchamber and throat region

velopment and construction require considerable technologicaleffort

1 Dual-Throat Nozzles

A dual-throat nozzle con guration is shown in Fig 28 Itconsists of two conventional bell-thrust chambers with onelocated completely inside the other At low altitude the outerthrust chamber operates with the inner thrust chamber runningin parallel In this operation mode the engine has a larger throatproviding a moderate expansion area ratio During the missionthe outer thrust chamber is shut off and operation continueswith only the inner engine In this con guration ow fromthe inner engine expands and attaches supersonically to theouter engine resulting in a higher expansion area ratio for theremainder of the burn Flow phenomena in both operationmodes are included in Fig 29

Hot- red tests at Aerojet were conducted to provide heattransfer data that were very useful for the thermal analysis anddesign of the dual-throat nozzle con guration37 These testsshowed that ow separation occurred in the inner engine noz-zle at higher ratios of outer to inner chamber pressures duringthe rst operation mode with both chambers burning in par-allel The ow separation resulted in a higher heat load to theinner nozzle Subscale tests performed at the Keldysh Center3

have shown that the additional loss caused by the nozzle con-tour discontinuity during vacuum operation with active innerchamber is in the range of 08ndash 4 depending on geometricaldata (see Fig 30) This high-performance loss results from theinteraction of the inner chamber jet with the outer chambernozzle wall3 The decrease of the jet incidence angle on thewall by means of gas injection through the outer chamber re-duces this performance loss by 04ndash 073

2 Dual-Expander Nozzles

A dual-expander nozzle has two concentric thrust chambersand nozzles It consists of a conventional bell thrust chambersurrounded by an annular thrust chamber Both chambers haveshort primary nozzles which end in a common divergent noz-zle extension Figure 31 shows a typical dual-expander nozzlecon guration At low altitude both thrust chambers operatesharing the same exit area which results in a moderate expan-sion area ratio Part way into the mission one thrust chamberis shut off allowing the other nozzle to use the whole exitarea creating a high-expansion-area ratio for the remainder of

the burn In principle the two operation modes are comparableto those of dual-throat nozzles

Numerical simulations of the ow elds in dual-expandernozzles during all operation modes have been performed atAerojet38 DLR39 and the Keldysh Center Aerojet simulationsare based on an engine design using different propellantshydrocarbonoxygen for the inner chamber and hydrogenox-ygen for the outer chamber whereas DLR simulations arebased on a combined vehicleengine analysis using single-fuelsingle-mixture ratio dual-expander engines with hydrogenoxy-gen Table 4 summarizes combustion chamber parameters Flowphenomena observed in numerical simulations are shown in Fig32a for the mode 1 operation with both thrust chambers burn-ing Compression waves are induced near the inner nozzle lipas a result of the strongly inhomogeneous ow character at thepoint where both exhaust gases of inner and outer combustionchamber merge into the common divergent nozzle part Upand down variations of pressure ratios at the lip did not changethis wave formation even in cases of signi cantly lower exitpressures in the inner nozzle compared to the pressure eld ofthe outer nozzle at the lip Further downstream the compres-sion waves interact with the wall resulting in a re ection ofthe compression waves back into the ow eld

During mode 2 operation a strong expansion of the outernozzle ow is observed when the expansion ratio suddenlyincreases at the end of the nozzle lip The ow is directedtoward the axis of symmetry Near the centerline the ow thenturns over to the axial direction inducing a recompressionshock The static pressure rises signi cantly in this recom-pression region on the centerline A sub- and supersonic recir-culation zone establishes in the inner chamber of the dual-expander nozzle Figure 32b emphasizes the essential owpattern in this operation mode

These analyses have shown that dual-expander nozzles pro-duce high performance in both operation modes3839 Figure 33summarizes performance behavior as a function of ight alti-tude for the dual-expander engine simulated at DLR For com-parison performance data are included for two reference bell

HAGEMANN ET AL 633

Fig 33 Performance data for a dual-expander nozzle

nozzles that have the same area ratios as the dual-expandernozzle during its two operation modes Losses because of thenonhomogeneous exit ow and the induced shocks are com-parable with corresponding values of both conventional noz-zles1

Within the design process of dual-expander engines it is nec-essary to take into account that the nonburning chamber isexposed to high heat uxes and pressure oscillations duringmode 2 operation which might reach a high level and lead tostructural disintegration of the inner nozzle structure There-fore a mode-2 operation with bleed gas generation at a ratherlow thrust chamber pressure level in the inner chamber seemsto be advantageous Because of the bleed gas generation in theprimary chamber the overall mass ow rate during this alter-native mode-2 operation is increased In addition the reductionof the effective exit area ratio results in minor impulse deg-radations during this operation mode when compared with themode-2 operation with no bleed gas generation

Several analytical works on SSTO- and two-stage-to-orbit(TSTO)-vehicles using hydrogenpropane or hydrogenmeth-ane as fuels revealed the lowest vehicle dry masses for dual-mode engines in comparison to other engines1439 Other dual-mode engines using hydrogen as the single fuel but using twomixture ratios also revealed some bene ts over conventionalengines for SSTO- and TSTO applications Even a single-fueloperation with constant mixture ratios in both combustionchambers indicated a gain in launcher performance40

V ConclusionsSeveral nozzle concepts that promise gains in performance

over conventional nozzles were discussed in this paper in-cluding performance enhancements achieved by slight modi- cations of existing nozzles eg through cool gas injectioninto the supersonic nozzle part It is shown that signi cantperformance gains result from the adaptation of the exhaust ow to the ambient pressure and special emphasis was givento altitude adaptive nozzle concepts

A number of nozzle concepts with altitude-compensating ca-pability were identi ed and described To assist the selectionof the best nozzle concept for launch vehicle applications theperformance of the nozzles must be characterized This can bedone using computational uid dynamics (CFD) andor cold- ow tests Existing CFD methods that are in use in the aero-space industry and at research institutes have been veri ed fora wide number of sub- and full-scale experiments and providesuf ciently reliable performance determination for the differ-ent nozzle types

Theoretical evaluations numerical simulations and test re-sults showed that the different concepts have real altitude-com-pensating capabilities However the compensation capabilitiesare limited and there are some drawbacks associated with eachof the concepts Additional performance losses are induced inpractically all of these nozzle concepts when compared to anideal expansion mainly because of the nonisentropic effects

such as shock waves and pressure losses in recirculation zonesHowever these additional performance losses are less than1 ndash 3 depending on the different nozzle concepts

Four nozzle concepts have been selected for further evalu-ation under a NASA Aerojet cooperative agreement They in-clude a plug nozzle an E ndash D nozzle a dual-bell nozzle and adual-expander nozzle The nozzles in addition to a referencebell nozzle have been designed by Aerojet for cold ow testsCFD calculations performed at Aerojet demonstrated that allof these nozzles have high and approximately equal perfor-mances at the design points

In Europe special attention is being paid to plug- and dual-bell nozzles Different plug nozzle concepts are being testedin cold- ow tests within the ESA ARPT (clustered plug noz-zle) and ESA FESTIP (linear plug nozzle) contracts In addi-tion to experiments three-dimensional numerical simulationson these nozzle concepts are being performed Additional ex-periments on different nozzle concepts with forced- ow sep-aration devices eg dual-bell nozzles are currently plannedat DLR

Different advanced nozzle concepts have been examined andare under examination now at the Keldysh Center with bothCFD calculations and experiments on the Nozzle DifferentialFacility which has a nozzle performance determination errorof about 005 This high accuracy is achieved by simulta-neously running the test nozzle and a reference nozzle bothmounted on a common longitudinal axis The performancecharacteristics of the reference nozzle are well known and thethrust difference is measured with strain gauges This facilityis equipped with a wind tunnel to determine nozzle thrust per-formance in the presence of external transonic ows withMach numbers up to 12

In addition to aerodynamic performance other technical is-sues (weight cost design thermal management manufactur-ing system performance and reliability) must be addressedFurthermore before a nal decision can be made as to whichnozzle concept offers the greatest bene ts with regard to aneffective payload mass injection combined launcher and tra-jectory calculations must be performed and compared to a ref-erence launcher concept with conventional nozzles Differentnozzle ef ciencies which account for the additional losses ofadvanced rocket nozzles and are extracted from numerical sim-ulations and experiments must be taken into account

At least for some of these nozzle concepts the plug nozzlethe dual-bell nozzle and the dual-expander nozzle bene tswith regard to lower overall launcher masses have been dem-onstrated in the literature Furthermore the plug nozzle con-cept will be the rst in ight of all of these advanced nozzleconcepts with the X-33 demonstrator vehicle

AcknowledgmentPresented as Paper 18 at the 3rd International Symposium

on Space Propulsion Beijing Peoplersquos Republic of ChinaAug 11ndash 13 1997

References1Manski D and Hagemann G lsquolsquoIn uence of Rocket Design Pa-

rameters on Engine Nozzle Ef cienciesrsquorsquo Journal of Propulsion andPower Vol 12 No 1 1996 pp 41ndash 47

2lsquolsquoLiquid Rocket Engine Nozzlesrsquorsquo NASA Space Vehicle DesignCriteria NASA SP-8120 1976

3Dumnov G E Nikulin G Z and Ponomaryov N B lsquolsquoInves-tigation of Advanced Nozzles for Rocket Enginesrsquorsquo Space Rocket En-gines and Power Plants Vol 4 No 142 NIITP 1993 pp 1012 ndash

1018 (in Russian)4Nguyen T V and Pieper J L lsquolsquoNozzle Flow Separationrsquorsquo Pro-

ceedings of the 5th International Symposium of Propulsion in SpaceTransportation (Paris France) 1996

5Manski D lsquolsquoClustered Plug Nozzles for Future European Reus-able Rocket Launchersrsquorsquo German Aerospace Research Establishment643-817 Lampoldshausen Germany 1981 (in German)

6Horn M and Fisher S lsquolsquoDual-Bell Altitude Compensating Noz-

624 HAGEMANN ET AL

Fig 8 Sketch of a dual-bell nozzle

Fig 9 Flow eld phenomena in dual-bell nozzles a) sea-levelmode with ow separation at the wall in ection point and b) al-titude mode with a full- owing nozzle

Fig 10 Performance data of a dual-bell nozzle Performance iscompared with two baseline bell-type nozzles as function of ightaltitude (baseline nozzle 1 same area ratio as dual-bell base noz-zle baseline nozzle 2 same area ratio as nozzle extension)

Fig 11 Performance characteristics of a dual-bell nozzle Per-formance is compared with two baseline bell-type nozzles as func-tion of ight altitude (baseline nozzle 1 same area ratio as dual-bell base nozzle baseline nozzle 2 same area ratio as nozzleextension)

as compared with two baseline nozzles having the same arearatio as the dual-bell nozzle at its wall in ection and in its exitplane Figures 10 and 11 illustrate the performance of a dual-bell nozzle as a function of ight altitude in comparison withboth baseline bell-type nozzles (Design parameters of thedual-bell nozzle are taken from a launcher analysis publishedin Ref 18 Propellants hydrogenoxygen r = 6 pc = 200 barrt = 007 m wall in ection at area ratio laquoB = 30 and totalarea ratio laquoE = 100) The pressure within the separated owregion of the dual-bell nozzle extension at sea-level operationis slightly below the ambient pressure inducing a thrust lossreferred to as lsquolsquoaspiration dragrsquorsquo In addition ow transitionoccurs before the optimum crossover point which leads tofurther thrust loss as compared to an ideal switchover Thenonoptimum contour of the full owing dual-bell nozzle re-sults in further losses at high altitudes

To gain insight into the performance and ow behavior ofdual-bell nozzles at different ambient pressures extensive nu-merical simulations with parametrical variations of contour de-sign parameters were performed18 An optimized bell nozzlewith equal total length and area ratio was used as the referencenozzle for comparison As a result the vacuum performance ofthe dual-bell nozzles has a degradation because of the imper-fect contour and this additional loss has the same order ofmagnitude as the divergence loss of the optimized bell nozzle

The simulations of sea-level operation also revealed an ad-ditional performance loss because of aspiration drag which isless than 3 for these dual-bell nozzles This additional lossdepends linearly on the ambient pressure and therefore it isreduced during the ascent of the launcher Furthermore thesesimulations showed that the application of commonly usedseparation criteria derived for conventional nozzles and ap-

plied to dual-bell nozzles with their wall in ection to estimatethe critical pressure ratio yields reasonable results with an ac-curacy of rsquo15 Experimental data obtained at the KeldyshCenter revealed that the critical pressure ratio at the wall in- ection is about 15ndash 20 less than the critical pressure ratioof a conventional nozzle and that usual separation criteriamust be corrected for an accurate prediction of the critical wallpressure ratio

Flow transition behavior in dual-bell nozzles strongly de-pends on the contour type of the nozzle extension3618 A sud-den transition from sea-level to vacuum operation can be atleast theoretically achieved by two different extensions witha zero wall pressure gradient (constant pressure extension) ora positive wall pressure gradient (overturned extension) But acritical analysis of the transition behavior considering decreas-ing ambient pressures during the launcher ascent revealed thata considerable time with uncontrolled ow separation withinthe nozzle extension exists even for these types of extensionsThe duration of this period can be reduced drastically by throt-tling the chamber pressure18

The main advantage of dual-bell nozzles as compared toother means of controlling nozzle ow separation is its sim-plicity because of the absence of any movable parts and there-fore its high reliability It is necessary to note that the external ow over the vehicle in ight reduces the pressure in the ve-hicle base region where engines are installed The ambient

HAGEMANN ET AL 625

Fig 13 Principle performance characteristics of the nozzle withinsert

Fig 12 RD-0120 nozzle hardware with removed nozzle insertand sketch of secondary nozzle mounted inside of the RD-0120nozzle (photographs taken from Ref 23)

pressure triggering the ow transition is the vehicle base pres-sure instead of the atmospheric pressure at the speci c ightaltitude As the base pressure is lower than the atmosphericpressure the nozzle ow transition occurs at a lower altitudethan the one showed in Fig 11 which slightly decreases theef ciency of the dual-bell nozzle operation along the trajectory

Despite the additional losses induced in dual-bell nozzlesthey still provide a signi cant net impulse gain over the entiretrajectory as compared to conventional bell nozzles Indepen-dent combined launcher and trajectory analyses performed byDasa within the European Space Agency Future EuropeanSpace Transportation Investigations Program (ESA FESTIP)study19 and at DLR on SSTO vehicles powered with dual-bellnozzles result in a signi cant payload gain when comparedwith a reference launcher equipped with conventional nozzles

2 Nozzles with Fixed Inserts

A trip ring attached to the inside of a conventional nozzledisturbs the turbulent boundary layer and causes ow separa-tion at higher ambient pressures At higher altitudes with lowerambient pressures the ow reattaches to the wall behind thetrip ring and full owing of the nozzle is achieved The tran-sition from sea level to vacuum mode depends on the wallpressure near the trip-ring location and on the disturbance in-duced by the trip ring The size of the trip ring is a compromisebetween stable ow separation during sea-level operation andthe induced performance loss during vacuum operation In Ref9 it is reported that a trip-ring size of 10 of the local bound-ary-layer thickness is suf cient to ensure stable ow separa-tion

In principle this concept is similar to the dual-bell nozzleconcept with regard to performance characteristics as shownin Fig 11 The sea-level performance of this nozzle concept islower than the performance of a conventional bell nozzle trun-cated at the trip-ring location because of the aspiration dragin the separated ow region of the nozzle Furthermore at sealevel the bell nozzle with trip rings has even higher divergencelosses than a comparable dual-bell nozzle because the nozzlecontour upstream of the obstacle differs from the optimal con-tour for this low-area ratio as a result of the bell nozzle designfor best vacuum performance The additional losses inducedduring vacuum operation is about 1 compared with the per-formance of the bell nozzle without an obstacle Thus theadditional losses are comparable to the additional losses in-duced in dual-bell nozzles As for dual-bell nozzles the tran-sition behavior of this nozzle concept is uncertain but it willbe even more uncertain than for a dual-bell nozzle with a con-stant pressure or overturned nozzle extension

In principle several altitude adaptations can be achievedwith one nozzle by various trip rings mounted one behind theother However this results in increasing vacuum performancelosses The trip rings can also be attached into existing nozzlesand therefore represent a low-cost concept at least for testpurposes with low technological risk Trip rings have beendemonstrated to be effective for side-load reductions duringthe transient startup of rocket engines20 The main problemswith trip-ring nozzles are not only performance losses but alsoring resistance in high-temperature boundary layers the exactcircumferential xing and the uncertainties in the transitionbehavior These uncertainties might be why active interest inthis nozzle concept in the 1970s which is documented in var-ious publications920ndash22 has disappeared in recent years

A further concept with a xed wall discontinuity is a nozzlewith a circumferential groove The aerodynamic behavior of anozzle with a groove and nozzle with trip rings are quite sim-ilar9 but this concept is hard to realize in the case of thinnozzle shells which are used in all of todayrsquos rocket nozzles

3 Nozzles with Temporary Inserts

Nozzle concepts with xed wall discontinuities have the dis-advantage of lower vacuum performance as compared with a

conventional bell nozzle with equal design and operation dataA promising concept for controlled ow separation is thereforetemporary inserts which are removed for vacuum operationThese inserts can be either ablative or ejectible The insertsmay have the form of a complete secondary nozzle23 (Fig 12)or of small steps attached inside the nozzle wall In case ofejectible inserts a reliable mechanism is needed to provide asudden and symmetrical detachment In any case shocks areinduced during the transient ejection because the inserts act asan obstacle in the supersonic exhaust ow These shocks alsointeract with the nozzle walls and increase pressure loads onthe wall and local heat uxes A nonsymmetrical ejectionwould then result in the generation of side-loads Furthermorethe danger of a downstream collision with the nozzle wallarises because the inserts might also experience a transversalmovement toward the walls

Recent hot- ring tests performed in Russia with a modi edRD-0120 engine equipped with a secondary nozzle insert re-vealed a signi cant performance gain of 12 during the sea-level operation at 100 chamber pressure compared with theoriginal RD-0120 performance23 Nominal chamber pressureof this engine is pc = 206 bar with an area ratio of laquo = 857These full-scale hot- ring tests demonstrated the durability ofmaterials sealings and the release mechanism and thus thefeasibility of this concept Figure 12 shows the nozzle hard-ware and a sketch of the secondary nozzle mounted inside theRD-0120 nozzle23

The principle performance characteristics of this RD-0120nozzle with ejectible insert are included in Fig 13 The nozzleoperation with insert results in a slight performance loss com-pared with an ideal bell nozzle with the same reduced area

626 HAGEMANN ET AL

Fig 14 Sketch of a two-position nozzle during a) sea-level andb) vacuum operations

ratio because of aspiration drag and the presumably nonoptim-ized insert contour The performance degradation is compara-ble to the one induced in dual-bell nozzles during sea-leveloperation (Fig 11) In vacuum or high-altitude operations thehigher performance of the baseline nozzle is achieved

Another method for removing the inserts is to use combus-tible or ablative elements924 During the ascent of the launcherthe size of the insert is continuously reduced until it is com-pletely consumed resulting in a full owing bell nozzle witha clean contour for best vacuum performance The principaluncertainties of this nozzle concept are the stability and surfaceregression rates of the inserts Furthermore a homogeneoussymmetrical and temporally de ned consumption must beguaranteed despite possible local pressure and temperature uctuations near the nozzle walls This is currently highly un-certain

4 Nozzles with Active or Passive Secondary Gas Injection

With this nozzle concept the ow in an overexpanding noz-zle is forced to separate at a desired location by injecting asecond uid into the gas stream in the wall normal directionto achieve maximum disturbance of the main exhaust gas andthus to induce ow separation This secondary gas injectionis quite different from the aforementioned TEG injectionwhich should be injected with few or no disturbances to themain ow The injection could be either active that is forcedthrough gas expansion from a higher-pressure reservoir or pas-sive using holes in the wall through which ambient gas issucked in (vented nozzle concept) The latter concept can onlywork in case of gas pressures near the wall inside the nozzlethat are lower than the ambient pressure Experience on forcedsecondary gas injection gained at Aerojet shows that a largeamount of injected uid is required to induce a signi cant owseparation Furthermore no net speci c impulse gain is real-ized when considering the additional gas ow rate

In the vented nozzle concept segments of the bell nozzlewall have various slots or holes opened to the outside ambientpressure During low-altitude operation the slots are opened totrigger ow separation within the nozzle whereas the slots areclosed during high-altitude or vacuum operation so that thegas fully expands inside the entire nozzle Experiments with amodi ed RL10A-3 engine equipped with this nozzle conceptwere performed at Pratt amp Whitney25 Performance resultsshowed that over a small range of low-pressure ratios the per-forated nozzle performed as well as a nozzle with its area ratiotruncated immediately upstream of the vented area Howeverat above some intermediate pressure ratios the thrust ef -ciency suddenly dropped and approached that of the full- ow-ing nozzle The measured performance characteristics werequite similar to those of the dual-bell nozzles shown in Fig11

The altitude range of this nozzle concept is limited by thenumber and position of the holes because the pressure withinthe nozzle must be lower than the ambient pressure Further-more as the rocket aft-body base pressure is lower than thealtitude atmospheric pressure in the surrounding ambient ows the nozzle ow transition occurs at a lower altitude andthus the range of compensation is further reduced

5 Two-Position or Extendible Nozzles

Nozzles of this type with extendible exit cones are currentlyused only for rocket motors of upper stages to reduce the pack-age volume for the nozzle eg at present for solid rocketengines such as the inertial upper stage (IUS) or for the liquidrocket engine RL10 The main idea of the extendible extensionis to use a truncated nozzle with low expansion in low- ightaltitudes and to have a higher nozzle extension at high alti-tudes Figure 14 illustrates this nozzle concept Its capabilityfor altitude compensation is indisputable and the nozzle per-formance is easily predictable The whole nozzle contour in-cluding the extendible extension is contoured for maximum

performance at the high-area ratio with either the method ofcharacteristics or a variational method The area ratio of the rst nozzle section is then determined and the nozzle contouris divided into two parts The xed nozzle part and the ex-tendible extension Investigations conducted at the KeldyshCenter have shown that this nozzle-contouring method is notonly the simplest but also provides a good overall trajectoryperformance3 A minor performance loss is incorporated duringlow-altitude operations because of the truncated inner nozzlewhich has a nonoptimal contour for this interim exit area ratioThe performance characteristics as a function of ight altitudeare similar to those of the nozzle with the ejectible insert asshown in Fig 13 A tradeoff study performed at the KeldyshCenter on nozzles with an ejectible insert and an extendibleextension showed that the thrust characteristics of the nozzlewith an ejectible insert are slightly better than those of theextendible nozzle concept with the same overall dimensionsfor both nozzles

The main drawback of the extendible nozzle concept is thatit requires mechanical devices for the deployment of the ex-tension which reduces engine reliability and increases totalengine mass The necessity for active cooling of the extendibleextension requires exible or movable elements in the coolingsystem which also reduces system reliability The in uencesof the external ow and the generated jet noise on the extend-ible extension in the retracted initial position have not yetbeen fully investigated Former investigations conducted in theKeldysh Center showed that the external ow causes bothsteady and unsteady pressure loads on the retracted nozzle ex-tension whereas the engine jet noise causes strong vibrationsof the nozzle extension

B Plug Nozzles

Various experimental analytical and numerical research onplug nozzles have been performed since the 1950s in theUS4532ndash 35 Europe5710192627 Russia3 and Japan28 In contrastto the previously discussed nozzle concepts plug nozzles pro-vide at least theoretically a continuous altitude adaptation upto their geometrical area ratio Figures 15 and 16 show twodifferent design approaches for circular plug nozzles whichdiffer only in the chamber and primary nozzle layout A con-ical central body is shown here which could also be designedwith more sophisticated contouring methods2930 Different de-sign approaches include plug nozzles with a toroidal chamberand throat (with and without truncation) and plug nozzles witha cluster of circular bell nozzle modules or with clusteredquasirectangular nozzle modules The latter approach seems to

HAGEMANN ET AL 627

Fig 17 Flow phenomena of a plug nozzle with full length (left column) and truncated central body (right column) at different pressureratios pc pamb off-design (top bottom) and design (center) pressure ratio

Fig 16 Principle design of plug nozzles clustered plug 36 mod-ules with truncated plug body

Fig 15 Principle design of plug nozzles toroidal plug fulllength

be advantageous because further losses induced by the gapsbetween individual modules and the ow eld interactionsdownstream of the module exits can be minimized It has beenshown that transition from a round to a square nozzle resultsin a very small performance loss31 In principle the ow elddevelopment of a clustered plug nozzle with rectangular nozzlemodules is similar to that of a toroidal plug nozzle but avoidsthe inherent disadvantages of the toroidal plug design regard-ing 1) the control of a constant throat gap during manufactur-

ing and thermal expansion (side-loads and thrust vector devi-ations) 2) the cooling of toroidal throat with tiny throat gapsand 3) the control of combustion instabilities in the toroidalcombustion chamber Another plug nozzle con guration is thelinear plug nozzle which is foreseen for the propulsion systemof the RLV X-33 concept

1 Circular Plug Nozzles

Figure 17 summarizes the principle ow phenomena of cir-cular plug nozzles with full length and truncated central bodiesat different off-design (top and bottom) and design (center)pressure ratios that were observed in experiments and numer-ical simulations For pressure ratios lower than the design pres-sure ratio of a plug nozzle with a well-contoured central bodythe ow expands near the central plug body without separa-tion and a system of recompression shocks and expansionwaves adapts the exhaust ow to the ambient pressure pambThe characteristic barrel-like form with several in ections ofthe shear-layer results from various interactions of compres-sion and expansion waves with the shear layer and turbulentdiffusion enlarges the shear layer farther downstream of thethroat The existence of the overexpansion and recompressionprocesses is inferred from up- and down-variations of plugwall pressure pro les observed in various cold- ow tests andnumerical simulations and will also be shown later for linearplug nozzles

At the design pressure ratio (see Fig 17 left column cen-ter) the characteristic with the design Mach number should bea straight line emanating to the tip of the central plug bodyand the shear layer is parallel to the centerline However forcircular plug nozzles designed with contouring methods pro-posed in Refs 29 and 30 no exact one-dimensional exit owpro le can be achieved because both methods use thePrandtlndash Meyer relations that are only valid for planar owsFurthermore nonhomogeneous ow in the throat regionwhich is in general not considered within the contour designprocess also in uences the exit ow pro le The wall pressuredistribution remains constant at pressure ratios above the de-sign pressure ratio ie the plug nozzle behaves like a con-ventional nozzle the loss of its capability of further altitude

628 HAGEMANN ET AL

Table 2 Design parameters of toroidal plug nozzles

Design data Full-scale nozzlea Subscale nozzleb

Chamber pressure pc 100 bar 12 barPropellants Hydrogenoxygen Gasndash oilnitric-acidMixture ratio r 60 25Inner chamber diameter dcinner 43 m 019 mOuter chamber diameter dcouter 44 m 034 mGeometric area ratio laquo 55 10aUsed for numerical simulations and performance predictions see Fig 18

bUsed in experiments see Fig 20

Fig 18 Performance of numerically simulated plug nozzle withfull-length central body

Fig 19 Performance of numerically simulated plug nozzle withtruncated central body

adaptation being included Figure 17 (left column bottom) il-lustrates the ow eld at higher pressure ratios Performancedata of a typical plug nozzle are included in Fig 18 and com-pared to a conventional bell nozzle with equal area ratio Thedesign and combustion chamber parameters are included in theleft column of Table 227

The truncation of the central plug body which is an advan-tage because of the huge length and high structural mass ofthe well-contoured central body results in a different ow andperformance behavior as compared to the full-length plug noz-zle At lower pressure ratios an open wake ow establisheswith a pressure level practically equal to the ambient pressure(Fig 17 right column top) At a certain pressure ratio closeto the design pressure ratio of the full-length plug nozzle thebase ow suddenly changes its character and turns over to theclosed form characterized by a constant base pressure that isno longer in uenced by the ambient pressure Analyses indi-cate that shorter plug bodies with higher truncations trigger anearlier change in wake ow at pressure ratios below the designpressure ratio At the transition point the pressure within thewake approaches a value that is below ambient pressure andthe full base area induces a negative thrust (Fig 17 right col-umn center) This thrust loss depends on the percentage oftruncation and the total size of the base area Published ex-perimental data and numerical simulations reveal an increasingthrust loss for shorter plug bodies because the total base areaincreases

Beyond the transition point the pressure within the closedwake remains constant At these lower ambient pressures thebase pressure is then higher than the ambient pressure result-ing in a positive thrust contribution of the total base area Per-formance data of a numerically simulated truncated plug noz-zle are shown in Fig 19 and compared to the same plugnozzle with full-length central body and a conventional bellnozzle Design parameters for this truncated plug nozzle arethe same as for the full length plug (see Fig 18 and Table 2)

Figure 20 shows a typical photograph of a sea-level hot-runtest with a truncated toroidal subscale plug nozzle performedat DLR5 Nozzle design data are included in Table 2 For com-parison of experimental results with numerical computations

the ow eld of the toroidal plug nozzle was calculated with anumerical method10 and Fig 20 shows the calculated Mach-number distribution in the combustion chamber and nozzlePrincipal physical processes like expansion waves shocks andthe recirculating base- ow region are in good agreement Boththe experiment and the numerical simulation show that the ow separates from the conical plug body before reaching itstruncated end Recent results on experiments with plug nozzlespublished in Ref 28 also reveal separation of the ow fromthe central plug body for conical contours In contrast no sep-aration was observed for contoured central plug bodies de-signed with the method proposed in Ref 29 Recent numericalsimulations of contoured plug nozzles performed at DLRwithin the ESA Advanced Rocket Propulsion Technology(ARPT) Program on advanced rocket propulsion also showthat no ow separation occurs from well-contoured full-lengthcentral plug bodies26 Principal ow eld developments pre-dicted by these numerical simulations are again in a goodagreement with experimental data published in Ref 28 Withinthe frame of this ESA ARPT Program performance and owbehavior of clustered plug nozzles at different truncations arebeing examined by European industries [Societe Europeennede Propulsion (SEP) Volvo Dasa] and research institutes(ONERA DLR) with subscale cold- ow plug models26 (Table3) Numerous experimental investigations on subscale modelsof circular and clustered plug nozzles with cold- and hot-gas ow were performed at the Keldysh Center with and withoutexternal ows Thrust characteristics pressure and heat- uxdistributions along the plug and acoustic characteristics of thecircular jet were investigated

2 Linear Plug Nozzles

Performance behavior and ow eld development for linearplug nozzles as a function of ambient pressures are in principlesimilar to those of circular plug nozzles (Fig 17) Howeverspecial attention must be paid to the in uence of both endsides where the surrounding ow disturbs the expanding ow- eld resulting in an expansion of the ow normal to the main ow direction and therefore in an effective performance loss

HAGEMANN ET AL 629

Table 3 Design parameters for linear and toroidal subscale plug nozzlesa

Design data Linear plug (FESTIP) Circular clustered plug (ARPT)

Chamber pressure pc 875 bar 80 barDriving gas Air AirMass ow rate mCcedil 595 kgs 546 kgsDesign pressure ratio pc pamb 200 200Geometric area ratio laquo 1265 1265aUsed in the ESA FESTIP and ESA ARPT studies

Fig 20 Plug nozzle ow eld experiment vs numerical simula-tion a) toroidal plug nozzle experiment with gas-oilnitric-acidpropellant combination side view photograph and b) computa-tional results of toroidal plug nozzle Mach number distributionfull gray and isolines centerline section

Fig 21 Design of linear subscale plug nozzle of the ESA FESTIPTechnology Program (plug size 192 mm high 214 mm wide)

Fig 22 Calculated wall pressure data for linear subscale plugnozzle at different pressure ratios

The change of wake ow behavior may be strongly in uencedby the penetration of ambient pressures through both end sidesparticularly for truncated plug nozzles End plates as foreseenfor the linear plug nozzle of the X-33 demonstrator vehiclecould be used to avoid this ambient pressure penetration

Within the framework of the ESA FESTIP Technology Pro-gram lsquolsquoTechnology Developments on Rocket and Air Breath-ing Propulsion for Reusable Launch Vehiclesrsquorsquo subscale cold-gas tests and numerical simulations of linear plug nozzles arebeing performed by Dasa at the high-speed wind tunnel [Hoch-geschwindigheitswind kanal (HWK) in Bad Salzungen Ger-many] of the Technical University of Dresden719 Design pa-

rameters for the linear plug model are included in Table 3Figure 21 shows the baseline model with a linear throat Afterpassing the linear throat the ow expands directly onto theplug surface A second model will be tested with a bell-shapedlinear nozzle extension following the linear throat for internalexpansion prior to further external expansion onto the plugsurface19 The contour of the plug was designed by the methodof Angelino29 and three different plug lengths as 5 20 and40 of the ideal length were tested Figure 22 shows calcu-lated wall pressures for the full-length plug for different am-bient pressures simulating the operational range of sea-leveloperation up to higher-altitude operation The ow structure inform of the left- and right-running characteristics at two pres-sure ratios is shown in Fig 23 For lower pressure ratios thanthe design pressure ratio of pc pamb = 200 the in uence of the

630 HAGEMANN ET AL

Fig 23 Left- and right-running characteristics in ow eld at twopressure ratios pc pamb = a) 8 and b) 200 (inviscid analysis withmethod of characteristics)

Fig 24 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 20 plug body

Fig 25 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 40 plug body

Fig 27 Flow phenomena of an E ndash D nozzle with a) open and b)closed wakes

Fig 26 Sketch of an E ndash D nozzle

system of recompression shocks and expansion waves on thewall pressure distribution (Figs 17 and 22) can be seen bythe waviness of the pressure distribution on the plug surfaceFirst experimental results are shown in Figs 24 and 25 whichalso show the waviness of the pressure distribution for low-pressure ratios

3 Plug Nozzles Final Aspects

The performance of circular and linear plug nozzles is ad-versely affected by the external airstream ow ie the aspi-ration effect of the external ow slightly reduces performanceand favors an earlier change in wake- ow development for

truncated plug nozzles32ndash 35 Cold- ow tests showed that theeffect of external ow on performance is con ned to a narrowrange of external ow speeds with Mach numbers near unity

The altitude compensation capability of circular and linearplug nozzles for higher ambient pressures is indisputable Be-cause plug nozzles lose this capability for pressure ratios abovethe design pressure ratio the latter should be chosen as highas possible Taking this into account plug nozzles will featurean even better overall performance than those shown in Figs18 and 19 Unfortunately for equal geometrical area ratiosplug nozzles perform worse at high altitude than do conven-tional bell nozzles because of truncation and clustering Forhigh-area-ratio nozzles with relatively short lengths plug noz-zles perform better than conventional bell nozzles In additionto having excellent capabilities for altitude compensation plugnozzles have additional advantages including ease in vehicleand engine integration

C E ndash D Nozzles

An expansionndash de ection (E ndash D) nozzle is shown in Fig 26E ndash D nozzles were at one time thought to have capabilities foraltitude compensation because the gas expansion takes placewith a constant pressure free boundary Thus the aerodynamicbehavior of E ndash D nozzles as a function of altitude is in prin-ciple quite similar to plug nozzles because the expansion pro-cess is controlled by the ambient pressure hence by the alti-tude In contrast to plug nozzles however the expansionprocess is controlled from inside the nozzle for E ndash D nozzlesAt low altitude the higher ambient pressure limits the gasexpansion resulting in a low effective expansion area ratioThe exhaust gas is adapted to the ambient pressure level by

HAGEMANN ET AL 631

Fig 30 Performance losses in dual-throat nozzles as a functionof nozzle design parameters (with r4r1 = 352 r2r1 = 116 and mCcedil 2

= 00)

Fig 29 Flow phenomena during a) sea-level and b) high-altitudeoperations

Fig 28 Sketch of a dual-throat nozzle view of combustion cham-ber and throat region

systems of recompression and expansion waves (Fig 27) Athigher altitude the lower ambient pressure allows more gasexpansion within the nozzle resulting in a higher effectiveexpansion area ratio But in contrast to plug nozzles the pres-sure in the wake of the center plug is always less than theambient pressure because of the aspiration effect This occursat low pressure ratios when the wake is opened and results inan aspiration loss Furthermore because the exhaust ow ex-pands to this base pressure rather than to the ambient pressurelevel wall pressures downstream are overexpanded This re-sults in an additional overexpansion loss As the pressure ratioincreases the wake region closes and is thus totally isolatedfrom the ambient environment (see Fig 27) The behavior dur-ing transition from open wake to closed wake is again equalto plug nozzles and the base pressure in the closed-wake re-gion is essentially independent of the ambient pressure

The E ndash D nozzle concept has also been a subject of numer-ous analytical and experimental studies Results from thesestudies show that E ndash D nozzle capabilities for altitude com-pensation are poor and are in fact worse than those of plugnozzles because of aspiration and overexpansion losses35 Forhigh-area-ratio nozzles with a relatively short length an E ndash Dnozzle performs better than a comparable conventional bellnozzle at the same length because of lower divergence andpro le losses than the bell nozzle

The advantages of the E ndash D nozzle concept include its smallengine envelope and no moving parts However like toroidalplug nozzles E ndash D nozzles have the disadvantage of havinghigher throat heat uxes relative to a conventional bell nozzlewith an equal throat area The higher throat heat ux resultsagain from the relatively thin annular throat gap This problemhowever can be remedied with the modular thrust cell clusterconcept Additional advantages of E ndash D nozzles with clusteredthrust cells are the same as those already discussed for theclustered plug nozzle concept with regard to ease in manu-facturing thrust vector control by throttling or shutting off anindividual or group of thrust cells and lower nozzle throatheating

D Nozzles with Throat Area Varied by a Mechanical Pintle

This nozzle concept utilizes a conventional bell nozzle witha xed exit area and a mechanical pintle in the combustionchamber and throat region to vary the throat area and hencethe expansion area ratio The area of the nozzle throatmdashanannulus between the pintle and the shroudmdashis varied by mov-ing the pintle axially

The pintle concept has been used in solid rocket motors asa mean to provide variable thrust The concept in principleallows a continuous variation of the throat area and thus op-timum expansion area ratios throughout a mission Howeverit requires an actuator and a sophisticated control system Theconcept raises issues of engine weight design complexitycooling of the pintle and nozzle throat and reliability

In Ref 36 the aerodynamic performances of nozzles with ve different pintle geometries were calculated and comparedwith a reference bell nozzle The performance losses of thesepintle nozzles when compared with the bell nozzle are in therange of 1ndash 25 Performances of a xed pintle geometry atthree different locations were also calculated The results showthat the performance loss varies with the pintle location

E Dual-Mode Nozzles

Dual-mode rocket engines using one or two fuels offer atrajectory-adapted dual-mode operation during the ascent of alauncher which may be of signi cant advantage for single-stage Earth-to-orbit vehicles as compared to conventionalrocket engines with bell-type nozzles This engine concept in-volves the use of a dense propellant combination with mod-erate performance during liftoff to provide high thrust duringthe initial ight phase and a better performing propellant com-bination in vacuum which result in higher speci c impulse

The fuels are burned in two different combustion chamberswith one located completely inside the other in the case ofengines with dual-throat nozzles or with a conventional bellthrust chamber surrounded by an annular thrust chamber in thecase of dual-expander engines This type of engine has a built-in accelerationndash reduction capability achieved by shuttingdown one of two thrust chambers The total engine thrust isthen provided by the remaining thrust chamber with the useof the total nozzle exit area leading to an increase in speci cimpulse Apart from the indicated bene ts of dual-mode en-gines which will be discussed in more detail later their de-

632 HAGEMANN ET AL

Table 4 Thrust chamber parameters for simulated dual-expander nozzlesa

Operation mode

Aerojet

First (inout) Second (inout)

DLR

First (inout) Second (inout)

Propellants C3H8ndash O2H2ndash O2 mdashH2ndash O2 H2ndash O2H2ndash O2 mdashH2ndash O2

Chamber pressure pc 414207 bar mdash207 bar 200200 bar mdash200 barMixture ratio r 337 mdash7 77 mdash7Exit area ratio laquo 69 146 58 116aInnerouter chambers

Fig 32 Flow phenomena in a dual-expander nozzle during a)sea-level- and b) high-altitude operation

Fig 31 Sketch of a dual-expander nozzle view of combustionchamber and throat region

velopment and construction require considerable technologicaleffort

1 Dual-Throat Nozzles

A dual-throat nozzle con guration is shown in Fig 28 Itconsists of two conventional bell-thrust chambers with onelocated completely inside the other At low altitude the outerthrust chamber operates with the inner thrust chamber runningin parallel In this operation mode the engine has a larger throatproviding a moderate expansion area ratio During the missionthe outer thrust chamber is shut off and operation continueswith only the inner engine In this con guration ow fromthe inner engine expands and attaches supersonically to theouter engine resulting in a higher expansion area ratio for theremainder of the burn Flow phenomena in both operationmodes are included in Fig 29

Hot- red tests at Aerojet were conducted to provide heattransfer data that were very useful for the thermal analysis anddesign of the dual-throat nozzle con guration37 These testsshowed that ow separation occurred in the inner engine noz-zle at higher ratios of outer to inner chamber pressures duringthe rst operation mode with both chambers burning in par-allel The ow separation resulted in a higher heat load to theinner nozzle Subscale tests performed at the Keldysh Center3

have shown that the additional loss caused by the nozzle con-tour discontinuity during vacuum operation with active innerchamber is in the range of 08ndash 4 depending on geometricaldata (see Fig 30) This high-performance loss results from theinteraction of the inner chamber jet with the outer chambernozzle wall3 The decrease of the jet incidence angle on thewall by means of gas injection through the outer chamber re-duces this performance loss by 04ndash 073

2 Dual-Expander Nozzles

A dual-expander nozzle has two concentric thrust chambersand nozzles It consists of a conventional bell thrust chambersurrounded by an annular thrust chamber Both chambers haveshort primary nozzles which end in a common divergent noz-zle extension Figure 31 shows a typical dual-expander nozzlecon guration At low altitude both thrust chambers operatesharing the same exit area which results in a moderate expan-sion area ratio Part way into the mission one thrust chamberis shut off allowing the other nozzle to use the whole exitarea creating a high-expansion-area ratio for the remainder of

the burn In principle the two operation modes are comparableto those of dual-throat nozzles

Numerical simulations of the ow elds in dual-expandernozzles during all operation modes have been performed atAerojet38 DLR39 and the Keldysh Center Aerojet simulationsare based on an engine design using different propellantshydrocarbonoxygen for the inner chamber and hydrogenox-ygen for the outer chamber whereas DLR simulations arebased on a combined vehicleengine analysis using single-fuelsingle-mixture ratio dual-expander engines with hydrogenoxy-gen Table 4 summarizes combustion chamber parameters Flowphenomena observed in numerical simulations are shown in Fig32a for the mode 1 operation with both thrust chambers burn-ing Compression waves are induced near the inner nozzle lipas a result of the strongly inhomogeneous ow character at thepoint where both exhaust gases of inner and outer combustionchamber merge into the common divergent nozzle part Upand down variations of pressure ratios at the lip did not changethis wave formation even in cases of signi cantly lower exitpressures in the inner nozzle compared to the pressure eld ofthe outer nozzle at the lip Further downstream the compres-sion waves interact with the wall resulting in a re ection ofthe compression waves back into the ow eld

During mode 2 operation a strong expansion of the outernozzle ow is observed when the expansion ratio suddenlyincreases at the end of the nozzle lip The ow is directedtoward the axis of symmetry Near the centerline the ow thenturns over to the axial direction inducing a recompressionshock The static pressure rises signi cantly in this recom-pression region on the centerline A sub- and supersonic recir-culation zone establishes in the inner chamber of the dual-expander nozzle Figure 32b emphasizes the essential owpattern in this operation mode

These analyses have shown that dual-expander nozzles pro-duce high performance in both operation modes3839 Figure 33summarizes performance behavior as a function of ight alti-tude for the dual-expander engine simulated at DLR For com-parison performance data are included for two reference bell

HAGEMANN ET AL 633

Fig 33 Performance data for a dual-expander nozzle

nozzles that have the same area ratios as the dual-expandernozzle during its two operation modes Losses because of thenonhomogeneous exit ow and the induced shocks are com-parable with corresponding values of both conventional noz-zles1

Within the design process of dual-expander engines it is nec-essary to take into account that the nonburning chamber isexposed to high heat uxes and pressure oscillations duringmode 2 operation which might reach a high level and lead tostructural disintegration of the inner nozzle structure There-fore a mode-2 operation with bleed gas generation at a ratherlow thrust chamber pressure level in the inner chamber seemsto be advantageous Because of the bleed gas generation in theprimary chamber the overall mass ow rate during this alter-native mode-2 operation is increased In addition the reductionof the effective exit area ratio results in minor impulse deg-radations during this operation mode when compared with themode-2 operation with no bleed gas generation

Several analytical works on SSTO- and two-stage-to-orbit(TSTO)-vehicles using hydrogenpropane or hydrogenmeth-ane as fuels revealed the lowest vehicle dry masses for dual-mode engines in comparison to other engines1439 Other dual-mode engines using hydrogen as the single fuel but using twomixture ratios also revealed some bene ts over conventionalengines for SSTO- and TSTO applications Even a single-fueloperation with constant mixture ratios in both combustionchambers indicated a gain in launcher performance40

V ConclusionsSeveral nozzle concepts that promise gains in performance

over conventional nozzles were discussed in this paper in-cluding performance enhancements achieved by slight modi- cations of existing nozzles eg through cool gas injectioninto the supersonic nozzle part It is shown that signi cantperformance gains result from the adaptation of the exhaust ow to the ambient pressure and special emphasis was givento altitude adaptive nozzle concepts

A number of nozzle concepts with altitude-compensating ca-pability were identi ed and described To assist the selectionof the best nozzle concept for launch vehicle applications theperformance of the nozzles must be characterized This can bedone using computational uid dynamics (CFD) andor cold- ow tests Existing CFD methods that are in use in the aero-space industry and at research institutes have been veri ed fora wide number of sub- and full-scale experiments and providesuf ciently reliable performance determination for the differ-ent nozzle types

Theoretical evaluations numerical simulations and test re-sults showed that the different concepts have real altitude-com-pensating capabilities However the compensation capabilitiesare limited and there are some drawbacks associated with eachof the concepts Additional performance losses are induced inpractically all of these nozzle concepts when compared to anideal expansion mainly because of the nonisentropic effects

such as shock waves and pressure losses in recirculation zonesHowever these additional performance losses are less than1 ndash 3 depending on the different nozzle concepts

Four nozzle concepts have been selected for further evalu-ation under a NASA Aerojet cooperative agreement They in-clude a plug nozzle an E ndash D nozzle a dual-bell nozzle and adual-expander nozzle The nozzles in addition to a referencebell nozzle have been designed by Aerojet for cold ow testsCFD calculations performed at Aerojet demonstrated that allof these nozzles have high and approximately equal perfor-mances at the design points

In Europe special attention is being paid to plug- and dual-bell nozzles Different plug nozzle concepts are being testedin cold- ow tests within the ESA ARPT (clustered plug noz-zle) and ESA FESTIP (linear plug nozzle) contracts In addi-tion to experiments three-dimensional numerical simulationson these nozzle concepts are being performed Additional ex-periments on different nozzle concepts with forced- ow sep-aration devices eg dual-bell nozzles are currently plannedat DLR

Different advanced nozzle concepts have been examined andare under examination now at the Keldysh Center with bothCFD calculations and experiments on the Nozzle DifferentialFacility which has a nozzle performance determination errorof about 005 This high accuracy is achieved by simulta-neously running the test nozzle and a reference nozzle bothmounted on a common longitudinal axis The performancecharacteristics of the reference nozzle are well known and thethrust difference is measured with strain gauges This facilityis equipped with a wind tunnel to determine nozzle thrust per-formance in the presence of external transonic ows withMach numbers up to 12

In addition to aerodynamic performance other technical is-sues (weight cost design thermal management manufactur-ing system performance and reliability) must be addressedFurthermore before a nal decision can be made as to whichnozzle concept offers the greatest bene ts with regard to aneffective payload mass injection combined launcher and tra-jectory calculations must be performed and compared to a ref-erence launcher concept with conventional nozzles Differentnozzle ef ciencies which account for the additional losses ofadvanced rocket nozzles and are extracted from numerical sim-ulations and experiments must be taken into account

At least for some of these nozzle concepts the plug nozzlethe dual-bell nozzle and the dual-expander nozzle bene tswith regard to lower overall launcher masses have been dem-onstrated in the literature Furthermore the plug nozzle con-cept will be the rst in ight of all of these advanced nozzleconcepts with the X-33 demonstrator vehicle

AcknowledgmentPresented as Paper 18 at the 3rd International Symposium

on Space Propulsion Beijing Peoplersquos Republic of ChinaAug 11ndash 13 1997

References1Manski D and Hagemann G lsquolsquoIn uence of Rocket Design Pa-

rameters on Engine Nozzle Ef cienciesrsquorsquo Journal of Propulsion andPower Vol 12 No 1 1996 pp 41ndash 47

2lsquolsquoLiquid Rocket Engine Nozzlesrsquorsquo NASA Space Vehicle DesignCriteria NASA SP-8120 1976

3Dumnov G E Nikulin G Z and Ponomaryov N B lsquolsquoInves-tigation of Advanced Nozzles for Rocket Enginesrsquorsquo Space Rocket En-gines and Power Plants Vol 4 No 142 NIITP 1993 pp 1012 ndash

1018 (in Russian)4Nguyen T V and Pieper J L lsquolsquoNozzle Flow Separationrsquorsquo Pro-

ceedings of the 5th International Symposium of Propulsion in SpaceTransportation (Paris France) 1996

5Manski D lsquolsquoClustered Plug Nozzles for Future European Reus-able Rocket Launchersrsquorsquo German Aerospace Research Establishment643-817 Lampoldshausen Germany 1981 (in German)

6Horn M and Fisher S lsquolsquoDual-Bell Altitude Compensating Noz-

HAGEMANN ET AL 625

Fig 13 Principle performance characteristics of the nozzle withinsert

Fig 12 RD-0120 nozzle hardware with removed nozzle insertand sketch of secondary nozzle mounted inside of the RD-0120nozzle (photographs taken from Ref 23)

pressure triggering the ow transition is the vehicle base pres-sure instead of the atmospheric pressure at the speci c ightaltitude As the base pressure is lower than the atmosphericpressure the nozzle ow transition occurs at a lower altitudethan the one showed in Fig 11 which slightly decreases theef ciency of the dual-bell nozzle operation along the trajectory

Despite the additional losses induced in dual-bell nozzlesthey still provide a signi cant net impulse gain over the entiretrajectory as compared to conventional bell nozzles Indepen-dent combined launcher and trajectory analyses performed byDasa within the European Space Agency Future EuropeanSpace Transportation Investigations Program (ESA FESTIP)study19 and at DLR on SSTO vehicles powered with dual-bellnozzles result in a signi cant payload gain when comparedwith a reference launcher equipped with conventional nozzles

2 Nozzles with Fixed Inserts

A trip ring attached to the inside of a conventional nozzledisturbs the turbulent boundary layer and causes ow separa-tion at higher ambient pressures At higher altitudes with lowerambient pressures the ow reattaches to the wall behind thetrip ring and full owing of the nozzle is achieved The tran-sition from sea level to vacuum mode depends on the wallpressure near the trip-ring location and on the disturbance in-duced by the trip ring The size of the trip ring is a compromisebetween stable ow separation during sea-level operation andthe induced performance loss during vacuum operation In Ref9 it is reported that a trip-ring size of 10 of the local bound-ary-layer thickness is suf cient to ensure stable ow separa-tion

In principle this concept is similar to the dual-bell nozzleconcept with regard to performance characteristics as shownin Fig 11 The sea-level performance of this nozzle concept islower than the performance of a conventional bell nozzle trun-cated at the trip-ring location because of the aspiration dragin the separated ow region of the nozzle Furthermore at sealevel the bell nozzle with trip rings has even higher divergencelosses than a comparable dual-bell nozzle because the nozzlecontour upstream of the obstacle differs from the optimal con-tour for this low-area ratio as a result of the bell nozzle designfor best vacuum performance The additional losses inducedduring vacuum operation is about 1 compared with the per-formance of the bell nozzle without an obstacle Thus theadditional losses are comparable to the additional losses in-duced in dual-bell nozzles As for dual-bell nozzles the tran-sition behavior of this nozzle concept is uncertain but it willbe even more uncertain than for a dual-bell nozzle with a con-stant pressure or overturned nozzle extension

In principle several altitude adaptations can be achievedwith one nozzle by various trip rings mounted one behind theother However this results in increasing vacuum performancelosses The trip rings can also be attached into existing nozzlesand therefore represent a low-cost concept at least for testpurposes with low technological risk Trip rings have beendemonstrated to be effective for side-load reductions duringthe transient startup of rocket engines20 The main problemswith trip-ring nozzles are not only performance losses but alsoring resistance in high-temperature boundary layers the exactcircumferential xing and the uncertainties in the transitionbehavior These uncertainties might be why active interest inthis nozzle concept in the 1970s which is documented in var-ious publications920ndash22 has disappeared in recent years

A further concept with a xed wall discontinuity is a nozzlewith a circumferential groove The aerodynamic behavior of anozzle with a groove and nozzle with trip rings are quite sim-ilar9 but this concept is hard to realize in the case of thinnozzle shells which are used in all of todayrsquos rocket nozzles

3 Nozzles with Temporary Inserts

Nozzle concepts with xed wall discontinuities have the dis-advantage of lower vacuum performance as compared with a

conventional bell nozzle with equal design and operation dataA promising concept for controlled ow separation is thereforetemporary inserts which are removed for vacuum operationThese inserts can be either ablative or ejectible The insertsmay have the form of a complete secondary nozzle23 (Fig 12)or of small steps attached inside the nozzle wall In case ofejectible inserts a reliable mechanism is needed to provide asudden and symmetrical detachment In any case shocks areinduced during the transient ejection because the inserts act asan obstacle in the supersonic exhaust ow These shocks alsointeract with the nozzle walls and increase pressure loads onthe wall and local heat uxes A nonsymmetrical ejectionwould then result in the generation of side-loads Furthermorethe danger of a downstream collision with the nozzle wallarises because the inserts might also experience a transversalmovement toward the walls

Recent hot- ring tests performed in Russia with a modi edRD-0120 engine equipped with a secondary nozzle insert re-vealed a signi cant performance gain of 12 during the sea-level operation at 100 chamber pressure compared with theoriginal RD-0120 performance23 Nominal chamber pressureof this engine is pc = 206 bar with an area ratio of laquo = 857These full-scale hot- ring tests demonstrated the durability ofmaterials sealings and the release mechanism and thus thefeasibility of this concept Figure 12 shows the nozzle hard-ware and a sketch of the secondary nozzle mounted inside theRD-0120 nozzle23

The principle performance characteristics of this RD-0120nozzle with ejectible insert are included in Fig 13 The nozzleoperation with insert results in a slight performance loss com-pared with an ideal bell nozzle with the same reduced area

626 HAGEMANN ET AL

Fig 14 Sketch of a two-position nozzle during a) sea-level andb) vacuum operations

ratio because of aspiration drag and the presumably nonoptim-ized insert contour The performance degradation is compara-ble to the one induced in dual-bell nozzles during sea-leveloperation (Fig 11) In vacuum or high-altitude operations thehigher performance of the baseline nozzle is achieved

Another method for removing the inserts is to use combus-tible or ablative elements924 During the ascent of the launcherthe size of the insert is continuously reduced until it is com-pletely consumed resulting in a full owing bell nozzle witha clean contour for best vacuum performance The principaluncertainties of this nozzle concept are the stability and surfaceregression rates of the inserts Furthermore a homogeneoussymmetrical and temporally de ned consumption must beguaranteed despite possible local pressure and temperature uctuations near the nozzle walls This is currently highly un-certain

4 Nozzles with Active or Passive Secondary Gas Injection

With this nozzle concept the ow in an overexpanding noz-zle is forced to separate at a desired location by injecting asecond uid into the gas stream in the wall normal directionto achieve maximum disturbance of the main exhaust gas andthus to induce ow separation This secondary gas injectionis quite different from the aforementioned TEG injectionwhich should be injected with few or no disturbances to themain ow The injection could be either active that is forcedthrough gas expansion from a higher-pressure reservoir or pas-sive using holes in the wall through which ambient gas issucked in (vented nozzle concept) The latter concept can onlywork in case of gas pressures near the wall inside the nozzlethat are lower than the ambient pressure Experience on forcedsecondary gas injection gained at Aerojet shows that a largeamount of injected uid is required to induce a signi cant owseparation Furthermore no net speci c impulse gain is real-ized when considering the additional gas ow rate

In the vented nozzle concept segments of the bell nozzlewall have various slots or holes opened to the outside ambientpressure During low-altitude operation the slots are opened totrigger ow separation within the nozzle whereas the slots areclosed during high-altitude or vacuum operation so that thegas fully expands inside the entire nozzle Experiments with amodi ed RL10A-3 engine equipped with this nozzle conceptwere performed at Pratt amp Whitney25 Performance resultsshowed that over a small range of low-pressure ratios the per-forated nozzle performed as well as a nozzle with its area ratiotruncated immediately upstream of the vented area Howeverat above some intermediate pressure ratios the thrust ef -ciency suddenly dropped and approached that of the full- ow-ing nozzle The measured performance characteristics werequite similar to those of the dual-bell nozzles shown in Fig11

The altitude range of this nozzle concept is limited by thenumber and position of the holes because the pressure withinthe nozzle must be lower than the ambient pressure Further-more as the rocket aft-body base pressure is lower than thealtitude atmospheric pressure in the surrounding ambient ows the nozzle ow transition occurs at a lower altitude andthus the range of compensation is further reduced

5 Two-Position or Extendible Nozzles

Nozzles of this type with extendible exit cones are currentlyused only for rocket motors of upper stages to reduce the pack-age volume for the nozzle eg at present for solid rocketengines such as the inertial upper stage (IUS) or for the liquidrocket engine RL10 The main idea of the extendible extensionis to use a truncated nozzle with low expansion in low- ightaltitudes and to have a higher nozzle extension at high alti-tudes Figure 14 illustrates this nozzle concept Its capabilityfor altitude compensation is indisputable and the nozzle per-formance is easily predictable The whole nozzle contour in-cluding the extendible extension is contoured for maximum

performance at the high-area ratio with either the method ofcharacteristics or a variational method The area ratio of the rst nozzle section is then determined and the nozzle contouris divided into two parts The xed nozzle part and the ex-tendible extension Investigations conducted at the KeldyshCenter have shown that this nozzle-contouring method is notonly the simplest but also provides a good overall trajectoryperformance3 A minor performance loss is incorporated duringlow-altitude operations because of the truncated inner nozzlewhich has a nonoptimal contour for this interim exit area ratioThe performance characteristics as a function of ight altitudeare similar to those of the nozzle with the ejectible insert asshown in Fig 13 A tradeoff study performed at the KeldyshCenter on nozzles with an ejectible insert and an extendibleextension showed that the thrust characteristics of the nozzlewith an ejectible insert are slightly better than those of theextendible nozzle concept with the same overall dimensionsfor both nozzles

The main drawback of the extendible nozzle concept is thatit requires mechanical devices for the deployment of the ex-tension which reduces engine reliability and increases totalengine mass The necessity for active cooling of the extendibleextension requires exible or movable elements in the coolingsystem which also reduces system reliability The in uencesof the external ow and the generated jet noise on the extend-ible extension in the retracted initial position have not yetbeen fully investigated Former investigations conducted in theKeldysh Center showed that the external ow causes bothsteady and unsteady pressure loads on the retracted nozzle ex-tension whereas the engine jet noise causes strong vibrationsof the nozzle extension

B Plug Nozzles

Various experimental analytical and numerical research onplug nozzles have been performed since the 1950s in theUS4532ndash 35 Europe5710192627 Russia3 and Japan28 In contrastto the previously discussed nozzle concepts plug nozzles pro-vide at least theoretically a continuous altitude adaptation upto their geometrical area ratio Figures 15 and 16 show twodifferent design approaches for circular plug nozzles whichdiffer only in the chamber and primary nozzle layout A con-ical central body is shown here which could also be designedwith more sophisticated contouring methods2930 Different de-sign approaches include plug nozzles with a toroidal chamberand throat (with and without truncation) and plug nozzles witha cluster of circular bell nozzle modules or with clusteredquasirectangular nozzle modules The latter approach seems to

HAGEMANN ET AL 627

Fig 17 Flow phenomena of a plug nozzle with full length (left column) and truncated central body (right column) at different pressureratios pc pamb off-design (top bottom) and design (center) pressure ratio

Fig 16 Principle design of plug nozzles clustered plug 36 mod-ules with truncated plug body

Fig 15 Principle design of plug nozzles toroidal plug fulllength

be advantageous because further losses induced by the gapsbetween individual modules and the ow eld interactionsdownstream of the module exits can be minimized It has beenshown that transition from a round to a square nozzle resultsin a very small performance loss31 In principle the ow elddevelopment of a clustered plug nozzle with rectangular nozzlemodules is similar to that of a toroidal plug nozzle but avoidsthe inherent disadvantages of the toroidal plug design regard-ing 1) the control of a constant throat gap during manufactur-

ing and thermal expansion (side-loads and thrust vector devi-ations) 2) the cooling of toroidal throat with tiny throat gapsand 3) the control of combustion instabilities in the toroidalcombustion chamber Another plug nozzle con guration is thelinear plug nozzle which is foreseen for the propulsion systemof the RLV X-33 concept

1 Circular Plug Nozzles

Figure 17 summarizes the principle ow phenomena of cir-cular plug nozzles with full length and truncated central bodiesat different off-design (top and bottom) and design (center)pressure ratios that were observed in experiments and numer-ical simulations For pressure ratios lower than the design pres-sure ratio of a plug nozzle with a well-contoured central bodythe ow expands near the central plug body without separa-tion and a system of recompression shocks and expansionwaves adapts the exhaust ow to the ambient pressure pambThe characteristic barrel-like form with several in ections ofthe shear-layer results from various interactions of compres-sion and expansion waves with the shear layer and turbulentdiffusion enlarges the shear layer farther downstream of thethroat The existence of the overexpansion and recompressionprocesses is inferred from up- and down-variations of plugwall pressure pro les observed in various cold- ow tests andnumerical simulations and will also be shown later for linearplug nozzles

At the design pressure ratio (see Fig 17 left column cen-ter) the characteristic with the design Mach number should bea straight line emanating to the tip of the central plug bodyand the shear layer is parallel to the centerline However forcircular plug nozzles designed with contouring methods pro-posed in Refs 29 and 30 no exact one-dimensional exit owpro le can be achieved because both methods use thePrandtlndash Meyer relations that are only valid for planar owsFurthermore nonhomogeneous ow in the throat regionwhich is in general not considered within the contour designprocess also in uences the exit ow pro le The wall pressuredistribution remains constant at pressure ratios above the de-sign pressure ratio ie the plug nozzle behaves like a con-ventional nozzle the loss of its capability of further altitude

628 HAGEMANN ET AL

Table 2 Design parameters of toroidal plug nozzles

Design data Full-scale nozzlea Subscale nozzleb

Chamber pressure pc 100 bar 12 barPropellants Hydrogenoxygen Gasndash oilnitric-acidMixture ratio r 60 25Inner chamber diameter dcinner 43 m 019 mOuter chamber diameter dcouter 44 m 034 mGeometric area ratio laquo 55 10aUsed for numerical simulations and performance predictions see Fig 18

bUsed in experiments see Fig 20

Fig 18 Performance of numerically simulated plug nozzle withfull-length central body

Fig 19 Performance of numerically simulated plug nozzle withtruncated central body

adaptation being included Figure 17 (left column bottom) il-lustrates the ow eld at higher pressure ratios Performancedata of a typical plug nozzle are included in Fig 18 and com-pared to a conventional bell nozzle with equal area ratio Thedesign and combustion chamber parameters are included in theleft column of Table 227

The truncation of the central plug body which is an advan-tage because of the huge length and high structural mass ofthe well-contoured central body results in a different ow andperformance behavior as compared to the full-length plug noz-zle At lower pressure ratios an open wake ow establisheswith a pressure level practically equal to the ambient pressure(Fig 17 right column top) At a certain pressure ratio closeto the design pressure ratio of the full-length plug nozzle thebase ow suddenly changes its character and turns over to theclosed form characterized by a constant base pressure that isno longer in uenced by the ambient pressure Analyses indi-cate that shorter plug bodies with higher truncations trigger anearlier change in wake ow at pressure ratios below the designpressure ratio At the transition point the pressure within thewake approaches a value that is below ambient pressure andthe full base area induces a negative thrust (Fig 17 right col-umn center) This thrust loss depends on the percentage oftruncation and the total size of the base area Published ex-perimental data and numerical simulations reveal an increasingthrust loss for shorter plug bodies because the total base areaincreases

Beyond the transition point the pressure within the closedwake remains constant At these lower ambient pressures thebase pressure is then higher than the ambient pressure result-ing in a positive thrust contribution of the total base area Per-formance data of a numerically simulated truncated plug noz-zle are shown in Fig 19 and compared to the same plugnozzle with full-length central body and a conventional bellnozzle Design parameters for this truncated plug nozzle arethe same as for the full length plug (see Fig 18 and Table 2)

Figure 20 shows a typical photograph of a sea-level hot-runtest with a truncated toroidal subscale plug nozzle performedat DLR5 Nozzle design data are included in Table 2 For com-parison of experimental results with numerical computations

the ow eld of the toroidal plug nozzle was calculated with anumerical method10 and Fig 20 shows the calculated Mach-number distribution in the combustion chamber and nozzlePrincipal physical processes like expansion waves shocks andthe recirculating base- ow region are in good agreement Boththe experiment and the numerical simulation show that the ow separates from the conical plug body before reaching itstruncated end Recent results on experiments with plug nozzlespublished in Ref 28 also reveal separation of the ow fromthe central plug body for conical contours In contrast no sep-aration was observed for contoured central plug bodies de-signed with the method proposed in Ref 29 Recent numericalsimulations of contoured plug nozzles performed at DLRwithin the ESA Advanced Rocket Propulsion Technology(ARPT) Program on advanced rocket propulsion also showthat no ow separation occurs from well-contoured full-lengthcentral plug bodies26 Principal ow eld developments pre-dicted by these numerical simulations are again in a goodagreement with experimental data published in Ref 28 Withinthe frame of this ESA ARPT Program performance and owbehavior of clustered plug nozzles at different truncations arebeing examined by European industries [Societe Europeennede Propulsion (SEP) Volvo Dasa] and research institutes(ONERA DLR) with subscale cold- ow plug models26 (Table3) Numerous experimental investigations on subscale modelsof circular and clustered plug nozzles with cold- and hot-gas ow were performed at the Keldysh Center with and withoutexternal ows Thrust characteristics pressure and heat- uxdistributions along the plug and acoustic characteristics of thecircular jet were investigated

2 Linear Plug Nozzles

Performance behavior and ow eld development for linearplug nozzles as a function of ambient pressures are in principlesimilar to those of circular plug nozzles (Fig 17) Howeverspecial attention must be paid to the in uence of both endsides where the surrounding ow disturbs the expanding ow- eld resulting in an expansion of the ow normal to the main ow direction and therefore in an effective performance loss

HAGEMANN ET AL 629

Table 3 Design parameters for linear and toroidal subscale plug nozzlesa

Design data Linear plug (FESTIP) Circular clustered plug (ARPT)

Chamber pressure pc 875 bar 80 barDriving gas Air AirMass ow rate mCcedil 595 kgs 546 kgsDesign pressure ratio pc pamb 200 200Geometric area ratio laquo 1265 1265aUsed in the ESA FESTIP and ESA ARPT studies

Fig 20 Plug nozzle ow eld experiment vs numerical simula-tion a) toroidal plug nozzle experiment with gas-oilnitric-acidpropellant combination side view photograph and b) computa-tional results of toroidal plug nozzle Mach number distributionfull gray and isolines centerline section

Fig 21 Design of linear subscale plug nozzle of the ESA FESTIPTechnology Program (plug size 192 mm high 214 mm wide)

Fig 22 Calculated wall pressure data for linear subscale plugnozzle at different pressure ratios

The change of wake ow behavior may be strongly in uencedby the penetration of ambient pressures through both end sidesparticularly for truncated plug nozzles End plates as foreseenfor the linear plug nozzle of the X-33 demonstrator vehiclecould be used to avoid this ambient pressure penetration

Within the framework of the ESA FESTIP Technology Pro-gram lsquolsquoTechnology Developments on Rocket and Air Breath-ing Propulsion for Reusable Launch Vehiclesrsquorsquo subscale cold-gas tests and numerical simulations of linear plug nozzles arebeing performed by Dasa at the high-speed wind tunnel [Hoch-geschwindigheitswind kanal (HWK) in Bad Salzungen Ger-many] of the Technical University of Dresden719 Design pa-

rameters for the linear plug model are included in Table 3Figure 21 shows the baseline model with a linear throat Afterpassing the linear throat the ow expands directly onto theplug surface A second model will be tested with a bell-shapedlinear nozzle extension following the linear throat for internalexpansion prior to further external expansion onto the plugsurface19 The contour of the plug was designed by the methodof Angelino29 and three different plug lengths as 5 20 and40 of the ideal length were tested Figure 22 shows calcu-lated wall pressures for the full-length plug for different am-bient pressures simulating the operational range of sea-leveloperation up to higher-altitude operation The ow structure inform of the left- and right-running characteristics at two pres-sure ratios is shown in Fig 23 For lower pressure ratios thanthe design pressure ratio of pc pamb = 200 the in uence of the

630 HAGEMANN ET AL

Fig 23 Left- and right-running characteristics in ow eld at twopressure ratios pc pamb = a) 8 and b) 200 (inviscid analysis withmethod of characteristics)

Fig 24 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 20 plug body

Fig 25 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 40 plug body

Fig 27 Flow phenomena of an E ndash D nozzle with a) open and b)closed wakes

Fig 26 Sketch of an E ndash D nozzle

system of recompression shocks and expansion waves on thewall pressure distribution (Figs 17 and 22) can be seen bythe waviness of the pressure distribution on the plug surfaceFirst experimental results are shown in Figs 24 and 25 whichalso show the waviness of the pressure distribution for low-pressure ratios

3 Plug Nozzles Final Aspects

The performance of circular and linear plug nozzles is ad-versely affected by the external airstream ow ie the aspi-ration effect of the external ow slightly reduces performanceand favors an earlier change in wake- ow development for

truncated plug nozzles32ndash 35 Cold- ow tests showed that theeffect of external ow on performance is con ned to a narrowrange of external ow speeds with Mach numbers near unity

The altitude compensation capability of circular and linearplug nozzles for higher ambient pressures is indisputable Be-cause plug nozzles lose this capability for pressure ratios abovethe design pressure ratio the latter should be chosen as highas possible Taking this into account plug nozzles will featurean even better overall performance than those shown in Figs18 and 19 Unfortunately for equal geometrical area ratiosplug nozzles perform worse at high altitude than do conven-tional bell nozzles because of truncation and clustering Forhigh-area-ratio nozzles with relatively short lengths plug noz-zles perform better than conventional bell nozzles In additionto having excellent capabilities for altitude compensation plugnozzles have additional advantages including ease in vehicleand engine integration

C E ndash D Nozzles

An expansionndash de ection (E ndash D) nozzle is shown in Fig 26E ndash D nozzles were at one time thought to have capabilities foraltitude compensation because the gas expansion takes placewith a constant pressure free boundary Thus the aerodynamicbehavior of E ndash D nozzles as a function of altitude is in prin-ciple quite similar to plug nozzles because the expansion pro-cess is controlled by the ambient pressure hence by the alti-tude In contrast to plug nozzles however the expansionprocess is controlled from inside the nozzle for E ndash D nozzlesAt low altitude the higher ambient pressure limits the gasexpansion resulting in a low effective expansion area ratioThe exhaust gas is adapted to the ambient pressure level by

HAGEMANN ET AL 631

Fig 30 Performance losses in dual-throat nozzles as a functionof nozzle design parameters (with r4r1 = 352 r2r1 = 116 and mCcedil 2

= 00)

Fig 29 Flow phenomena during a) sea-level and b) high-altitudeoperations

Fig 28 Sketch of a dual-throat nozzle view of combustion cham-ber and throat region

systems of recompression and expansion waves (Fig 27) Athigher altitude the lower ambient pressure allows more gasexpansion within the nozzle resulting in a higher effectiveexpansion area ratio But in contrast to plug nozzles the pres-sure in the wake of the center plug is always less than theambient pressure because of the aspiration effect This occursat low pressure ratios when the wake is opened and results inan aspiration loss Furthermore because the exhaust ow ex-pands to this base pressure rather than to the ambient pressurelevel wall pressures downstream are overexpanded This re-sults in an additional overexpansion loss As the pressure ratioincreases the wake region closes and is thus totally isolatedfrom the ambient environment (see Fig 27) The behavior dur-ing transition from open wake to closed wake is again equalto plug nozzles and the base pressure in the closed-wake re-gion is essentially independent of the ambient pressure

The E ndash D nozzle concept has also been a subject of numer-ous analytical and experimental studies Results from thesestudies show that E ndash D nozzle capabilities for altitude com-pensation are poor and are in fact worse than those of plugnozzles because of aspiration and overexpansion losses35 Forhigh-area-ratio nozzles with a relatively short length an E ndash Dnozzle performs better than a comparable conventional bellnozzle at the same length because of lower divergence andpro le losses than the bell nozzle

The advantages of the E ndash D nozzle concept include its smallengine envelope and no moving parts However like toroidalplug nozzles E ndash D nozzles have the disadvantage of havinghigher throat heat uxes relative to a conventional bell nozzlewith an equal throat area The higher throat heat ux resultsagain from the relatively thin annular throat gap This problemhowever can be remedied with the modular thrust cell clusterconcept Additional advantages of E ndash D nozzles with clusteredthrust cells are the same as those already discussed for theclustered plug nozzle concept with regard to ease in manu-facturing thrust vector control by throttling or shutting off anindividual or group of thrust cells and lower nozzle throatheating

D Nozzles with Throat Area Varied by a Mechanical Pintle

This nozzle concept utilizes a conventional bell nozzle witha xed exit area and a mechanical pintle in the combustionchamber and throat region to vary the throat area and hencethe expansion area ratio The area of the nozzle throatmdashanannulus between the pintle and the shroudmdashis varied by mov-ing the pintle axially

The pintle concept has been used in solid rocket motors asa mean to provide variable thrust The concept in principleallows a continuous variation of the throat area and thus op-timum expansion area ratios throughout a mission Howeverit requires an actuator and a sophisticated control system Theconcept raises issues of engine weight design complexitycooling of the pintle and nozzle throat and reliability

In Ref 36 the aerodynamic performances of nozzles with ve different pintle geometries were calculated and comparedwith a reference bell nozzle The performance losses of thesepintle nozzles when compared with the bell nozzle are in therange of 1ndash 25 Performances of a xed pintle geometry atthree different locations were also calculated The results showthat the performance loss varies with the pintle location

E Dual-Mode Nozzles

Dual-mode rocket engines using one or two fuels offer atrajectory-adapted dual-mode operation during the ascent of alauncher which may be of signi cant advantage for single-stage Earth-to-orbit vehicles as compared to conventionalrocket engines with bell-type nozzles This engine concept in-volves the use of a dense propellant combination with mod-erate performance during liftoff to provide high thrust duringthe initial ight phase and a better performing propellant com-bination in vacuum which result in higher speci c impulse

The fuels are burned in two different combustion chamberswith one located completely inside the other in the case ofengines with dual-throat nozzles or with a conventional bellthrust chamber surrounded by an annular thrust chamber in thecase of dual-expander engines This type of engine has a built-in accelerationndash reduction capability achieved by shuttingdown one of two thrust chambers The total engine thrust isthen provided by the remaining thrust chamber with the useof the total nozzle exit area leading to an increase in speci cimpulse Apart from the indicated bene ts of dual-mode en-gines which will be discussed in more detail later their de-

632 HAGEMANN ET AL

Table 4 Thrust chamber parameters for simulated dual-expander nozzlesa

Operation mode

Aerojet

First (inout) Second (inout)

DLR

First (inout) Second (inout)

Propellants C3H8ndash O2H2ndash O2 mdashH2ndash O2 H2ndash O2H2ndash O2 mdashH2ndash O2

Chamber pressure pc 414207 bar mdash207 bar 200200 bar mdash200 barMixture ratio r 337 mdash7 77 mdash7Exit area ratio laquo 69 146 58 116aInnerouter chambers

Fig 32 Flow phenomena in a dual-expander nozzle during a)sea-level- and b) high-altitude operation

Fig 31 Sketch of a dual-expander nozzle view of combustionchamber and throat region

velopment and construction require considerable technologicaleffort

1 Dual-Throat Nozzles

A dual-throat nozzle con guration is shown in Fig 28 Itconsists of two conventional bell-thrust chambers with onelocated completely inside the other At low altitude the outerthrust chamber operates with the inner thrust chamber runningin parallel In this operation mode the engine has a larger throatproviding a moderate expansion area ratio During the missionthe outer thrust chamber is shut off and operation continueswith only the inner engine In this con guration ow fromthe inner engine expands and attaches supersonically to theouter engine resulting in a higher expansion area ratio for theremainder of the burn Flow phenomena in both operationmodes are included in Fig 29

Hot- red tests at Aerojet were conducted to provide heattransfer data that were very useful for the thermal analysis anddesign of the dual-throat nozzle con guration37 These testsshowed that ow separation occurred in the inner engine noz-zle at higher ratios of outer to inner chamber pressures duringthe rst operation mode with both chambers burning in par-allel The ow separation resulted in a higher heat load to theinner nozzle Subscale tests performed at the Keldysh Center3

have shown that the additional loss caused by the nozzle con-tour discontinuity during vacuum operation with active innerchamber is in the range of 08ndash 4 depending on geometricaldata (see Fig 30) This high-performance loss results from theinteraction of the inner chamber jet with the outer chambernozzle wall3 The decrease of the jet incidence angle on thewall by means of gas injection through the outer chamber re-duces this performance loss by 04ndash 073

2 Dual-Expander Nozzles

A dual-expander nozzle has two concentric thrust chambersand nozzles It consists of a conventional bell thrust chambersurrounded by an annular thrust chamber Both chambers haveshort primary nozzles which end in a common divergent noz-zle extension Figure 31 shows a typical dual-expander nozzlecon guration At low altitude both thrust chambers operatesharing the same exit area which results in a moderate expan-sion area ratio Part way into the mission one thrust chamberis shut off allowing the other nozzle to use the whole exitarea creating a high-expansion-area ratio for the remainder of

the burn In principle the two operation modes are comparableto those of dual-throat nozzles

Numerical simulations of the ow elds in dual-expandernozzles during all operation modes have been performed atAerojet38 DLR39 and the Keldysh Center Aerojet simulationsare based on an engine design using different propellantshydrocarbonoxygen for the inner chamber and hydrogenox-ygen for the outer chamber whereas DLR simulations arebased on a combined vehicleengine analysis using single-fuelsingle-mixture ratio dual-expander engines with hydrogenoxy-gen Table 4 summarizes combustion chamber parameters Flowphenomena observed in numerical simulations are shown in Fig32a for the mode 1 operation with both thrust chambers burn-ing Compression waves are induced near the inner nozzle lipas a result of the strongly inhomogeneous ow character at thepoint where both exhaust gases of inner and outer combustionchamber merge into the common divergent nozzle part Upand down variations of pressure ratios at the lip did not changethis wave formation even in cases of signi cantly lower exitpressures in the inner nozzle compared to the pressure eld ofthe outer nozzle at the lip Further downstream the compres-sion waves interact with the wall resulting in a re ection ofthe compression waves back into the ow eld

During mode 2 operation a strong expansion of the outernozzle ow is observed when the expansion ratio suddenlyincreases at the end of the nozzle lip The ow is directedtoward the axis of symmetry Near the centerline the ow thenturns over to the axial direction inducing a recompressionshock The static pressure rises signi cantly in this recom-pression region on the centerline A sub- and supersonic recir-culation zone establishes in the inner chamber of the dual-expander nozzle Figure 32b emphasizes the essential owpattern in this operation mode

These analyses have shown that dual-expander nozzles pro-duce high performance in both operation modes3839 Figure 33summarizes performance behavior as a function of ight alti-tude for the dual-expander engine simulated at DLR For com-parison performance data are included for two reference bell

HAGEMANN ET AL 633

Fig 33 Performance data for a dual-expander nozzle

nozzles that have the same area ratios as the dual-expandernozzle during its two operation modes Losses because of thenonhomogeneous exit ow and the induced shocks are com-parable with corresponding values of both conventional noz-zles1

Within the design process of dual-expander engines it is nec-essary to take into account that the nonburning chamber isexposed to high heat uxes and pressure oscillations duringmode 2 operation which might reach a high level and lead tostructural disintegration of the inner nozzle structure There-fore a mode-2 operation with bleed gas generation at a ratherlow thrust chamber pressure level in the inner chamber seemsto be advantageous Because of the bleed gas generation in theprimary chamber the overall mass ow rate during this alter-native mode-2 operation is increased In addition the reductionof the effective exit area ratio results in minor impulse deg-radations during this operation mode when compared with themode-2 operation with no bleed gas generation

Several analytical works on SSTO- and two-stage-to-orbit(TSTO)-vehicles using hydrogenpropane or hydrogenmeth-ane as fuels revealed the lowest vehicle dry masses for dual-mode engines in comparison to other engines1439 Other dual-mode engines using hydrogen as the single fuel but using twomixture ratios also revealed some bene ts over conventionalengines for SSTO- and TSTO applications Even a single-fueloperation with constant mixture ratios in both combustionchambers indicated a gain in launcher performance40

V ConclusionsSeveral nozzle concepts that promise gains in performance

over conventional nozzles were discussed in this paper in-cluding performance enhancements achieved by slight modi- cations of existing nozzles eg through cool gas injectioninto the supersonic nozzle part It is shown that signi cantperformance gains result from the adaptation of the exhaust ow to the ambient pressure and special emphasis was givento altitude adaptive nozzle concepts

A number of nozzle concepts with altitude-compensating ca-pability were identi ed and described To assist the selectionof the best nozzle concept for launch vehicle applications theperformance of the nozzles must be characterized This can bedone using computational uid dynamics (CFD) andor cold- ow tests Existing CFD methods that are in use in the aero-space industry and at research institutes have been veri ed fora wide number of sub- and full-scale experiments and providesuf ciently reliable performance determination for the differ-ent nozzle types

Theoretical evaluations numerical simulations and test re-sults showed that the different concepts have real altitude-com-pensating capabilities However the compensation capabilitiesare limited and there are some drawbacks associated with eachof the concepts Additional performance losses are induced inpractically all of these nozzle concepts when compared to anideal expansion mainly because of the nonisentropic effects

such as shock waves and pressure losses in recirculation zonesHowever these additional performance losses are less than1 ndash 3 depending on the different nozzle concepts

Four nozzle concepts have been selected for further evalu-ation under a NASA Aerojet cooperative agreement They in-clude a plug nozzle an E ndash D nozzle a dual-bell nozzle and adual-expander nozzle The nozzles in addition to a referencebell nozzle have been designed by Aerojet for cold ow testsCFD calculations performed at Aerojet demonstrated that allof these nozzles have high and approximately equal perfor-mances at the design points

In Europe special attention is being paid to plug- and dual-bell nozzles Different plug nozzle concepts are being testedin cold- ow tests within the ESA ARPT (clustered plug noz-zle) and ESA FESTIP (linear plug nozzle) contracts In addi-tion to experiments three-dimensional numerical simulationson these nozzle concepts are being performed Additional ex-periments on different nozzle concepts with forced- ow sep-aration devices eg dual-bell nozzles are currently plannedat DLR

Different advanced nozzle concepts have been examined andare under examination now at the Keldysh Center with bothCFD calculations and experiments on the Nozzle DifferentialFacility which has a nozzle performance determination errorof about 005 This high accuracy is achieved by simulta-neously running the test nozzle and a reference nozzle bothmounted on a common longitudinal axis The performancecharacteristics of the reference nozzle are well known and thethrust difference is measured with strain gauges This facilityis equipped with a wind tunnel to determine nozzle thrust per-formance in the presence of external transonic ows withMach numbers up to 12

In addition to aerodynamic performance other technical is-sues (weight cost design thermal management manufactur-ing system performance and reliability) must be addressedFurthermore before a nal decision can be made as to whichnozzle concept offers the greatest bene ts with regard to aneffective payload mass injection combined launcher and tra-jectory calculations must be performed and compared to a ref-erence launcher concept with conventional nozzles Differentnozzle ef ciencies which account for the additional losses ofadvanced rocket nozzles and are extracted from numerical sim-ulations and experiments must be taken into account

At least for some of these nozzle concepts the plug nozzlethe dual-bell nozzle and the dual-expander nozzle bene tswith regard to lower overall launcher masses have been dem-onstrated in the literature Furthermore the plug nozzle con-cept will be the rst in ight of all of these advanced nozzleconcepts with the X-33 demonstrator vehicle

AcknowledgmentPresented as Paper 18 at the 3rd International Symposium

on Space Propulsion Beijing Peoplersquos Republic of ChinaAug 11ndash 13 1997

References1Manski D and Hagemann G lsquolsquoIn uence of Rocket Design Pa-

rameters on Engine Nozzle Ef cienciesrsquorsquo Journal of Propulsion andPower Vol 12 No 1 1996 pp 41ndash 47

2lsquolsquoLiquid Rocket Engine Nozzlesrsquorsquo NASA Space Vehicle DesignCriteria NASA SP-8120 1976

3Dumnov G E Nikulin G Z and Ponomaryov N B lsquolsquoInves-tigation of Advanced Nozzles for Rocket Enginesrsquorsquo Space Rocket En-gines and Power Plants Vol 4 No 142 NIITP 1993 pp 1012 ndash

1018 (in Russian)4Nguyen T V and Pieper J L lsquolsquoNozzle Flow Separationrsquorsquo Pro-

ceedings of the 5th International Symposium of Propulsion in SpaceTransportation (Paris France) 1996

5Manski D lsquolsquoClustered Plug Nozzles for Future European Reus-able Rocket Launchersrsquorsquo German Aerospace Research Establishment643-817 Lampoldshausen Germany 1981 (in German)

6Horn M and Fisher S lsquolsquoDual-Bell Altitude Compensating Noz-

626 HAGEMANN ET AL

Fig 14 Sketch of a two-position nozzle during a) sea-level andb) vacuum operations

ratio because of aspiration drag and the presumably nonoptim-ized insert contour The performance degradation is compara-ble to the one induced in dual-bell nozzles during sea-leveloperation (Fig 11) In vacuum or high-altitude operations thehigher performance of the baseline nozzle is achieved

Another method for removing the inserts is to use combus-tible or ablative elements924 During the ascent of the launcherthe size of the insert is continuously reduced until it is com-pletely consumed resulting in a full owing bell nozzle witha clean contour for best vacuum performance The principaluncertainties of this nozzle concept are the stability and surfaceregression rates of the inserts Furthermore a homogeneoussymmetrical and temporally de ned consumption must beguaranteed despite possible local pressure and temperature uctuations near the nozzle walls This is currently highly un-certain

4 Nozzles with Active or Passive Secondary Gas Injection

With this nozzle concept the ow in an overexpanding noz-zle is forced to separate at a desired location by injecting asecond uid into the gas stream in the wall normal directionto achieve maximum disturbance of the main exhaust gas andthus to induce ow separation This secondary gas injectionis quite different from the aforementioned TEG injectionwhich should be injected with few or no disturbances to themain ow The injection could be either active that is forcedthrough gas expansion from a higher-pressure reservoir or pas-sive using holes in the wall through which ambient gas issucked in (vented nozzle concept) The latter concept can onlywork in case of gas pressures near the wall inside the nozzlethat are lower than the ambient pressure Experience on forcedsecondary gas injection gained at Aerojet shows that a largeamount of injected uid is required to induce a signi cant owseparation Furthermore no net speci c impulse gain is real-ized when considering the additional gas ow rate

In the vented nozzle concept segments of the bell nozzlewall have various slots or holes opened to the outside ambientpressure During low-altitude operation the slots are opened totrigger ow separation within the nozzle whereas the slots areclosed during high-altitude or vacuum operation so that thegas fully expands inside the entire nozzle Experiments with amodi ed RL10A-3 engine equipped with this nozzle conceptwere performed at Pratt amp Whitney25 Performance resultsshowed that over a small range of low-pressure ratios the per-forated nozzle performed as well as a nozzle with its area ratiotruncated immediately upstream of the vented area Howeverat above some intermediate pressure ratios the thrust ef -ciency suddenly dropped and approached that of the full- ow-ing nozzle The measured performance characteristics werequite similar to those of the dual-bell nozzles shown in Fig11

The altitude range of this nozzle concept is limited by thenumber and position of the holes because the pressure withinthe nozzle must be lower than the ambient pressure Further-more as the rocket aft-body base pressure is lower than thealtitude atmospheric pressure in the surrounding ambient ows the nozzle ow transition occurs at a lower altitude andthus the range of compensation is further reduced

5 Two-Position or Extendible Nozzles

Nozzles of this type with extendible exit cones are currentlyused only for rocket motors of upper stages to reduce the pack-age volume for the nozzle eg at present for solid rocketengines such as the inertial upper stage (IUS) or for the liquidrocket engine RL10 The main idea of the extendible extensionis to use a truncated nozzle with low expansion in low- ightaltitudes and to have a higher nozzle extension at high alti-tudes Figure 14 illustrates this nozzle concept Its capabilityfor altitude compensation is indisputable and the nozzle per-formance is easily predictable The whole nozzle contour in-cluding the extendible extension is contoured for maximum

performance at the high-area ratio with either the method ofcharacteristics or a variational method The area ratio of the rst nozzle section is then determined and the nozzle contouris divided into two parts The xed nozzle part and the ex-tendible extension Investigations conducted at the KeldyshCenter have shown that this nozzle-contouring method is notonly the simplest but also provides a good overall trajectoryperformance3 A minor performance loss is incorporated duringlow-altitude operations because of the truncated inner nozzlewhich has a nonoptimal contour for this interim exit area ratioThe performance characteristics as a function of ight altitudeare similar to those of the nozzle with the ejectible insert asshown in Fig 13 A tradeoff study performed at the KeldyshCenter on nozzles with an ejectible insert and an extendibleextension showed that the thrust characteristics of the nozzlewith an ejectible insert are slightly better than those of theextendible nozzle concept with the same overall dimensionsfor both nozzles

The main drawback of the extendible nozzle concept is thatit requires mechanical devices for the deployment of the ex-tension which reduces engine reliability and increases totalengine mass The necessity for active cooling of the extendibleextension requires exible or movable elements in the coolingsystem which also reduces system reliability The in uencesof the external ow and the generated jet noise on the extend-ible extension in the retracted initial position have not yetbeen fully investigated Former investigations conducted in theKeldysh Center showed that the external ow causes bothsteady and unsteady pressure loads on the retracted nozzle ex-tension whereas the engine jet noise causes strong vibrationsof the nozzle extension

B Plug Nozzles

Various experimental analytical and numerical research onplug nozzles have been performed since the 1950s in theUS4532ndash 35 Europe5710192627 Russia3 and Japan28 In contrastto the previously discussed nozzle concepts plug nozzles pro-vide at least theoretically a continuous altitude adaptation upto their geometrical area ratio Figures 15 and 16 show twodifferent design approaches for circular plug nozzles whichdiffer only in the chamber and primary nozzle layout A con-ical central body is shown here which could also be designedwith more sophisticated contouring methods2930 Different de-sign approaches include plug nozzles with a toroidal chamberand throat (with and without truncation) and plug nozzles witha cluster of circular bell nozzle modules or with clusteredquasirectangular nozzle modules The latter approach seems to

HAGEMANN ET AL 627

Fig 17 Flow phenomena of a plug nozzle with full length (left column) and truncated central body (right column) at different pressureratios pc pamb off-design (top bottom) and design (center) pressure ratio

Fig 16 Principle design of plug nozzles clustered plug 36 mod-ules with truncated plug body

Fig 15 Principle design of plug nozzles toroidal plug fulllength

be advantageous because further losses induced by the gapsbetween individual modules and the ow eld interactionsdownstream of the module exits can be minimized It has beenshown that transition from a round to a square nozzle resultsin a very small performance loss31 In principle the ow elddevelopment of a clustered plug nozzle with rectangular nozzlemodules is similar to that of a toroidal plug nozzle but avoidsthe inherent disadvantages of the toroidal plug design regard-ing 1) the control of a constant throat gap during manufactur-

ing and thermal expansion (side-loads and thrust vector devi-ations) 2) the cooling of toroidal throat with tiny throat gapsand 3) the control of combustion instabilities in the toroidalcombustion chamber Another plug nozzle con guration is thelinear plug nozzle which is foreseen for the propulsion systemof the RLV X-33 concept

1 Circular Plug Nozzles

Figure 17 summarizes the principle ow phenomena of cir-cular plug nozzles with full length and truncated central bodiesat different off-design (top and bottom) and design (center)pressure ratios that were observed in experiments and numer-ical simulations For pressure ratios lower than the design pres-sure ratio of a plug nozzle with a well-contoured central bodythe ow expands near the central plug body without separa-tion and a system of recompression shocks and expansionwaves adapts the exhaust ow to the ambient pressure pambThe characteristic barrel-like form with several in ections ofthe shear-layer results from various interactions of compres-sion and expansion waves with the shear layer and turbulentdiffusion enlarges the shear layer farther downstream of thethroat The existence of the overexpansion and recompressionprocesses is inferred from up- and down-variations of plugwall pressure pro les observed in various cold- ow tests andnumerical simulations and will also be shown later for linearplug nozzles

At the design pressure ratio (see Fig 17 left column cen-ter) the characteristic with the design Mach number should bea straight line emanating to the tip of the central plug bodyand the shear layer is parallel to the centerline However forcircular plug nozzles designed with contouring methods pro-posed in Refs 29 and 30 no exact one-dimensional exit owpro le can be achieved because both methods use thePrandtlndash Meyer relations that are only valid for planar owsFurthermore nonhomogeneous ow in the throat regionwhich is in general not considered within the contour designprocess also in uences the exit ow pro le The wall pressuredistribution remains constant at pressure ratios above the de-sign pressure ratio ie the plug nozzle behaves like a con-ventional nozzle the loss of its capability of further altitude

628 HAGEMANN ET AL

Table 2 Design parameters of toroidal plug nozzles

Design data Full-scale nozzlea Subscale nozzleb

Chamber pressure pc 100 bar 12 barPropellants Hydrogenoxygen Gasndash oilnitric-acidMixture ratio r 60 25Inner chamber diameter dcinner 43 m 019 mOuter chamber diameter dcouter 44 m 034 mGeometric area ratio laquo 55 10aUsed for numerical simulations and performance predictions see Fig 18

bUsed in experiments see Fig 20

Fig 18 Performance of numerically simulated plug nozzle withfull-length central body

Fig 19 Performance of numerically simulated plug nozzle withtruncated central body

adaptation being included Figure 17 (left column bottom) il-lustrates the ow eld at higher pressure ratios Performancedata of a typical plug nozzle are included in Fig 18 and com-pared to a conventional bell nozzle with equal area ratio Thedesign and combustion chamber parameters are included in theleft column of Table 227

The truncation of the central plug body which is an advan-tage because of the huge length and high structural mass ofthe well-contoured central body results in a different ow andperformance behavior as compared to the full-length plug noz-zle At lower pressure ratios an open wake ow establisheswith a pressure level practically equal to the ambient pressure(Fig 17 right column top) At a certain pressure ratio closeto the design pressure ratio of the full-length plug nozzle thebase ow suddenly changes its character and turns over to theclosed form characterized by a constant base pressure that isno longer in uenced by the ambient pressure Analyses indi-cate that shorter plug bodies with higher truncations trigger anearlier change in wake ow at pressure ratios below the designpressure ratio At the transition point the pressure within thewake approaches a value that is below ambient pressure andthe full base area induces a negative thrust (Fig 17 right col-umn center) This thrust loss depends on the percentage oftruncation and the total size of the base area Published ex-perimental data and numerical simulations reveal an increasingthrust loss for shorter plug bodies because the total base areaincreases

Beyond the transition point the pressure within the closedwake remains constant At these lower ambient pressures thebase pressure is then higher than the ambient pressure result-ing in a positive thrust contribution of the total base area Per-formance data of a numerically simulated truncated plug noz-zle are shown in Fig 19 and compared to the same plugnozzle with full-length central body and a conventional bellnozzle Design parameters for this truncated plug nozzle arethe same as for the full length plug (see Fig 18 and Table 2)

Figure 20 shows a typical photograph of a sea-level hot-runtest with a truncated toroidal subscale plug nozzle performedat DLR5 Nozzle design data are included in Table 2 For com-parison of experimental results with numerical computations

the ow eld of the toroidal plug nozzle was calculated with anumerical method10 and Fig 20 shows the calculated Mach-number distribution in the combustion chamber and nozzlePrincipal physical processes like expansion waves shocks andthe recirculating base- ow region are in good agreement Boththe experiment and the numerical simulation show that the ow separates from the conical plug body before reaching itstruncated end Recent results on experiments with plug nozzlespublished in Ref 28 also reveal separation of the ow fromthe central plug body for conical contours In contrast no sep-aration was observed for contoured central plug bodies de-signed with the method proposed in Ref 29 Recent numericalsimulations of contoured plug nozzles performed at DLRwithin the ESA Advanced Rocket Propulsion Technology(ARPT) Program on advanced rocket propulsion also showthat no ow separation occurs from well-contoured full-lengthcentral plug bodies26 Principal ow eld developments pre-dicted by these numerical simulations are again in a goodagreement with experimental data published in Ref 28 Withinthe frame of this ESA ARPT Program performance and owbehavior of clustered plug nozzles at different truncations arebeing examined by European industries [Societe Europeennede Propulsion (SEP) Volvo Dasa] and research institutes(ONERA DLR) with subscale cold- ow plug models26 (Table3) Numerous experimental investigations on subscale modelsof circular and clustered plug nozzles with cold- and hot-gas ow were performed at the Keldysh Center with and withoutexternal ows Thrust characteristics pressure and heat- uxdistributions along the plug and acoustic characteristics of thecircular jet were investigated

2 Linear Plug Nozzles

Performance behavior and ow eld development for linearplug nozzles as a function of ambient pressures are in principlesimilar to those of circular plug nozzles (Fig 17) Howeverspecial attention must be paid to the in uence of both endsides where the surrounding ow disturbs the expanding ow- eld resulting in an expansion of the ow normal to the main ow direction and therefore in an effective performance loss

HAGEMANN ET AL 629

Table 3 Design parameters for linear and toroidal subscale plug nozzlesa

Design data Linear plug (FESTIP) Circular clustered plug (ARPT)

Chamber pressure pc 875 bar 80 barDriving gas Air AirMass ow rate mCcedil 595 kgs 546 kgsDesign pressure ratio pc pamb 200 200Geometric area ratio laquo 1265 1265aUsed in the ESA FESTIP and ESA ARPT studies

Fig 20 Plug nozzle ow eld experiment vs numerical simula-tion a) toroidal plug nozzle experiment with gas-oilnitric-acidpropellant combination side view photograph and b) computa-tional results of toroidal plug nozzle Mach number distributionfull gray and isolines centerline section

Fig 21 Design of linear subscale plug nozzle of the ESA FESTIPTechnology Program (plug size 192 mm high 214 mm wide)

Fig 22 Calculated wall pressure data for linear subscale plugnozzle at different pressure ratios

The change of wake ow behavior may be strongly in uencedby the penetration of ambient pressures through both end sidesparticularly for truncated plug nozzles End plates as foreseenfor the linear plug nozzle of the X-33 demonstrator vehiclecould be used to avoid this ambient pressure penetration

Within the framework of the ESA FESTIP Technology Pro-gram lsquolsquoTechnology Developments on Rocket and Air Breath-ing Propulsion for Reusable Launch Vehiclesrsquorsquo subscale cold-gas tests and numerical simulations of linear plug nozzles arebeing performed by Dasa at the high-speed wind tunnel [Hoch-geschwindigheitswind kanal (HWK) in Bad Salzungen Ger-many] of the Technical University of Dresden719 Design pa-

rameters for the linear plug model are included in Table 3Figure 21 shows the baseline model with a linear throat Afterpassing the linear throat the ow expands directly onto theplug surface A second model will be tested with a bell-shapedlinear nozzle extension following the linear throat for internalexpansion prior to further external expansion onto the plugsurface19 The contour of the plug was designed by the methodof Angelino29 and three different plug lengths as 5 20 and40 of the ideal length were tested Figure 22 shows calcu-lated wall pressures for the full-length plug for different am-bient pressures simulating the operational range of sea-leveloperation up to higher-altitude operation The ow structure inform of the left- and right-running characteristics at two pres-sure ratios is shown in Fig 23 For lower pressure ratios thanthe design pressure ratio of pc pamb = 200 the in uence of the

630 HAGEMANN ET AL

Fig 23 Left- and right-running characteristics in ow eld at twopressure ratios pc pamb = a) 8 and b) 200 (inviscid analysis withmethod of characteristics)

Fig 24 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 20 plug body

Fig 25 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 40 plug body

Fig 27 Flow phenomena of an E ndash D nozzle with a) open and b)closed wakes

Fig 26 Sketch of an E ndash D nozzle

system of recompression shocks and expansion waves on thewall pressure distribution (Figs 17 and 22) can be seen bythe waviness of the pressure distribution on the plug surfaceFirst experimental results are shown in Figs 24 and 25 whichalso show the waviness of the pressure distribution for low-pressure ratios

3 Plug Nozzles Final Aspects

The performance of circular and linear plug nozzles is ad-versely affected by the external airstream ow ie the aspi-ration effect of the external ow slightly reduces performanceand favors an earlier change in wake- ow development for

truncated plug nozzles32ndash 35 Cold- ow tests showed that theeffect of external ow on performance is con ned to a narrowrange of external ow speeds with Mach numbers near unity

The altitude compensation capability of circular and linearplug nozzles for higher ambient pressures is indisputable Be-cause plug nozzles lose this capability for pressure ratios abovethe design pressure ratio the latter should be chosen as highas possible Taking this into account plug nozzles will featurean even better overall performance than those shown in Figs18 and 19 Unfortunately for equal geometrical area ratiosplug nozzles perform worse at high altitude than do conven-tional bell nozzles because of truncation and clustering Forhigh-area-ratio nozzles with relatively short lengths plug noz-zles perform better than conventional bell nozzles In additionto having excellent capabilities for altitude compensation plugnozzles have additional advantages including ease in vehicleand engine integration

C E ndash D Nozzles

An expansionndash de ection (E ndash D) nozzle is shown in Fig 26E ndash D nozzles were at one time thought to have capabilities foraltitude compensation because the gas expansion takes placewith a constant pressure free boundary Thus the aerodynamicbehavior of E ndash D nozzles as a function of altitude is in prin-ciple quite similar to plug nozzles because the expansion pro-cess is controlled by the ambient pressure hence by the alti-tude In contrast to plug nozzles however the expansionprocess is controlled from inside the nozzle for E ndash D nozzlesAt low altitude the higher ambient pressure limits the gasexpansion resulting in a low effective expansion area ratioThe exhaust gas is adapted to the ambient pressure level by

HAGEMANN ET AL 631

Fig 30 Performance losses in dual-throat nozzles as a functionof nozzle design parameters (with r4r1 = 352 r2r1 = 116 and mCcedil 2

= 00)

Fig 29 Flow phenomena during a) sea-level and b) high-altitudeoperations

Fig 28 Sketch of a dual-throat nozzle view of combustion cham-ber and throat region

systems of recompression and expansion waves (Fig 27) Athigher altitude the lower ambient pressure allows more gasexpansion within the nozzle resulting in a higher effectiveexpansion area ratio But in contrast to plug nozzles the pres-sure in the wake of the center plug is always less than theambient pressure because of the aspiration effect This occursat low pressure ratios when the wake is opened and results inan aspiration loss Furthermore because the exhaust ow ex-pands to this base pressure rather than to the ambient pressurelevel wall pressures downstream are overexpanded This re-sults in an additional overexpansion loss As the pressure ratioincreases the wake region closes and is thus totally isolatedfrom the ambient environment (see Fig 27) The behavior dur-ing transition from open wake to closed wake is again equalto plug nozzles and the base pressure in the closed-wake re-gion is essentially independent of the ambient pressure

The E ndash D nozzle concept has also been a subject of numer-ous analytical and experimental studies Results from thesestudies show that E ndash D nozzle capabilities for altitude com-pensation are poor and are in fact worse than those of plugnozzles because of aspiration and overexpansion losses35 Forhigh-area-ratio nozzles with a relatively short length an E ndash Dnozzle performs better than a comparable conventional bellnozzle at the same length because of lower divergence andpro le losses than the bell nozzle

The advantages of the E ndash D nozzle concept include its smallengine envelope and no moving parts However like toroidalplug nozzles E ndash D nozzles have the disadvantage of havinghigher throat heat uxes relative to a conventional bell nozzlewith an equal throat area The higher throat heat ux resultsagain from the relatively thin annular throat gap This problemhowever can be remedied with the modular thrust cell clusterconcept Additional advantages of E ndash D nozzles with clusteredthrust cells are the same as those already discussed for theclustered plug nozzle concept with regard to ease in manu-facturing thrust vector control by throttling or shutting off anindividual or group of thrust cells and lower nozzle throatheating

D Nozzles with Throat Area Varied by a Mechanical Pintle

This nozzle concept utilizes a conventional bell nozzle witha xed exit area and a mechanical pintle in the combustionchamber and throat region to vary the throat area and hencethe expansion area ratio The area of the nozzle throatmdashanannulus between the pintle and the shroudmdashis varied by mov-ing the pintle axially

The pintle concept has been used in solid rocket motors asa mean to provide variable thrust The concept in principleallows a continuous variation of the throat area and thus op-timum expansion area ratios throughout a mission Howeverit requires an actuator and a sophisticated control system Theconcept raises issues of engine weight design complexitycooling of the pintle and nozzle throat and reliability

In Ref 36 the aerodynamic performances of nozzles with ve different pintle geometries were calculated and comparedwith a reference bell nozzle The performance losses of thesepintle nozzles when compared with the bell nozzle are in therange of 1ndash 25 Performances of a xed pintle geometry atthree different locations were also calculated The results showthat the performance loss varies with the pintle location

E Dual-Mode Nozzles

Dual-mode rocket engines using one or two fuels offer atrajectory-adapted dual-mode operation during the ascent of alauncher which may be of signi cant advantage for single-stage Earth-to-orbit vehicles as compared to conventionalrocket engines with bell-type nozzles This engine concept in-volves the use of a dense propellant combination with mod-erate performance during liftoff to provide high thrust duringthe initial ight phase and a better performing propellant com-bination in vacuum which result in higher speci c impulse

The fuels are burned in two different combustion chamberswith one located completely inside the other in the case ofengines with dual-throat nozzles or with a conventional bellthrust chamber surrounded by an annular thrust chamber in thecase of dual-expander engines This type of engine has a built-in accelerationndash reduction capability achieved by shuttingdown one of two thrust chambers The total engine thrust isthen provided by the remaining thrust chamber with the useof the total nozzle exit area leading to an increase in speci cimpulse Apart from the indicated bene ts of dual-mode en-gines which will be discussed in more detail later their de-

632 HAGEMANN ET AL

Table 4 Thrust chamber parameters for simulated dual-expander nozzlesa

Operation mode

Aerojet

First (inout) Second (inout)

DLR

First (inout) Second (inout)

Propellants C3H8ndash O2H2ndash O2 mdashH2ndash O2 H2ndash O2H2ndash O2 mdashH2ndash O2

Chamber pressure pc 414207 bar mdash207 bar 200200 bar mdash200 barMixture ratio r 337 mdash7 77 mdash7Exit area ratio laquo 69 146 58 116aInnerouter chambers

Fig 32 Flow phenomena in a dual-expander nozzle during a)sea-level- and b) high-altitude operation

Fig 31 Sketch of a dual-expander nozzle view of combustionchamber and throat region

velopment and construction require considerable technologicaleffort

1 Dual-Throat Nozzles

A dual-throat nozzle con guration is shown in Fig 28 Itconsists of two conventional bell-thrust chambers with onelocated completely inside the other At low altitude the outerthrust chamber operates with the inner thrust chamber runningin parallel In this operation mode the engine has a larger throatproviding a moderate expansion area ratio During the missionthe outer thrust chamber is shut off and operation continueswith only the inner engine In this con guration ow fromthe inner engine expands and attaches supersonically to theouter engine resulting in a higher expansion area ratio for theremainder of the burn Flow phenomena in both operationmodes are included in Fig 29

Hot- red tests at Aerojet were conducted to provide heattransfer data that were very useful for the thermal analysis anddesign of the dual-throat nozzle con guration37 These testsshowed that ow separation occurred in the inner engine noz-zle at higher ratios of outer to inner chamber pressures duringthe rst operation mode with both chambers burning in par-allel The ow separation resulted in a higher heat load to theinner nozzle Subscale tests performed at the Keldysh Center3

have shown that the additional loss caused by the nozzle con-tour discontinuity during vacuum operation with active innerchamber is in the range of 08ndash 4 depending on geometricaldata (see Fig 30) This high-performance loss results from theinteraction of the inner chamber jet with the outer chambernozzle wall3 The decrease of the jet incidence angle on thewall by means of gas injection through the outer chamber re-duces this performance loss by 04ndash 073

2 Dual-Expander Nozzles

A dual-expander nozzle has two concentric thrust chambersand nozzles It consists of a conventional bell thrust chambersurrounded by an annular thrust chamber Both chambers haveshort primary nozzles which end in a common divergent noz-zle extension Figure 31 shows a typical dual-expander nozzlecon guration At low altitude both thrust chambers operatesharing the same exit area which results in a moderate expan-sion area ratio Part way into the mission one thrust chamberis shut off allowing the other nozzle to use the whole exitarea creating a high-expansion-area ratio for the remainder of

the burn In principle the two operation modes are comparableto those of dual-throat nozzles

Numerical simulations of the ow elds in dual-expandernozzles during all operation modes have been performed atAerojet38 DLR39 and the Keldysh Center Aerojet simulationsare based on an engine design using different propellantshydrocarbonoxygen for the inner chamber and hydrogenox-ygen for the outer chamber whereas DLR simulations arebased on a combined vehicleengine analysis using single-fuelsingle-mixture ratio dual-expander engines with hydrogenoxy-gen Table 4 summarizes combustion chamber parameters Flowphenomena observed in numerical simulations are shown in Fig32a for the mode 1 operation with both thrust chambers burn-ing Compression waves are induced near the inner nozzle lipas a result of the strongly inhomogeneous ow character at thepoint where both exhaust gases of inner and outer combustionchamber merge into the common divergent nozzle part Upand down variations of pressure ratios at the lip did not changethis wave formation even in cases of signi cantly lower exitpressures in the inner nozzle compared to the pressure eld ofthe outer nozzle at the lip Further downstream the compres-sion waves interact with the wall resulting in a re ection ofthe compression waves back into the ow eld

During mode 2 operation a strong expansion of the outernozzle ow is observed when the expansion ratio suddenlyincreases at the end of the nozzle lip The ow is directedtoward the axis of symmetry Near the centerline the ow thenturns over to the axial direction inducing a recompressionshock The static pressure rises signi cantly in this recom-pression region on the centerline A sub- and supersonic recir-culation zone establishes in the inner chamber of the dual-expander nozzle Figure 32b emphasizes the essential owpattern in this operation mode

These analyses have shown that dual-expander nozzles pro-duce high performance in both operation modes3839 Figure 33summarizes performance behavior as a function of ight alti-tude for the dual-expander engine simulated at DLR For com-parison performance data are included for two reference bell

HAGEMANN ET AL 633

Fig 33 Performance data for a dual-expander nozzle

nozzles that have the same area ratios as the dual-expandernozzle during its two operation modes Losses because of thenonhomogeneous exit ow and the induced shocks are com-parable with corresponding values of both conventional noz-zles1

Within the design process of dual-expander engines it is nec-essary to take into account that the nonburning chamber isexposed to high heat uxes and pressure oscillations duringmode 2 operation which might reach a high level and lead tostructural disintegration of the inner nozzle structure There-fore a mode-2 operation with bleed gas generation at a ratherlow thrust chamber pressure level in the inner chamber seemsto be advantageous Because of the bleed gas generation in theprimary chamber the overall mass ow rate during this alter-native mode-2 operation is increased In addition the reductionof the effective exit area ratio results in minor impulse deg-radations during this operation mode when compared with themode-2 operation with no bleed gas generation

Several analytical works on SSTO- and two-stage-to-orbit(TSTO)-vehicles using hydrogenpropane or hydrogenmeth-ane as fuels revealed the lowest vehicle dry masses for dual-mode engines in comparison to other engines1439 Other dual-mode engines using hydrogen as the single fuel but using twomixture ratios also revealed some bene ts over conventionalengines for SSTO- and TSTO applications Even a single-fueloperation with constant mixture ratios in both combustionchambers indicated a gain in launcher performance40

V ConclusionsSeveral nozzle concepts that promise gains in performance

over conventional nozzles were discussed in this paper in-cluding performance enhancements achieved by slight modi- cations of existing nozzles eg through cool gas injectioninto the supersonic nozzle part It is shown that signi cantperformance gains result from the adaptation of the exhaust ow to the ambient pressure and special emphasis was givento altitude adaptive nozzle concepts

A number of nozzle concepts with altitude-compensating ca-pability were identi ed and described To assist the selectionof the best nozzle concept for launch vehicle applications theperformance of the nozzles must be characterized This can bedone using computational uid dynamics (CFD) andor cold- ow tests Existing CFD methods that are in use in the aero-space industry and at research institutes have been veri ed fora wide number of sub- and full-scale experiments and providesuf ciently reliable performance determination for the differ-ent nozzle types

Theoretical evaluations numerical simulations and test re-sults showed that the different concepts have real altitude-com-pensating capabilities However the compensation capabilitiesare limited and there are some drawbacks associated with eachof the concepts Additional performance losses are induced inpractically all of these nozzle concepts when compared to anideal expansion mainly because of the nonisentropic effects

such as shock waves and pressure losses in recirculation zonesHowever these additional performance losses are less than1 ndash 3 depending on the different nozzle concepts

Four nozzle concepts have been selected for further evalu-ation under a NASA Aerojet cooperative agreement They in-clude a plug nozzle an E ndash D nozzle a dual-bell nozzle and adual-expander nozzle The nozzles in addition to a referencebell nozzle have been designed by Aerojet for cold ow testsCFD calculations performed at Aerojet demonstrated that allof these nozzles have high and approximately equal perfor-mances at the design points

In Europe special attention is being paid to plug- and dual-bell nozzles Different plug nozzle concepts are being testedin cold- ow tests within the ESA ARPT (clustered plug noz-zle) and ESA FESTIP (linear plug nozzle) contracts In addi-tion to experiments three-dimensional numerical simulationson these nozzle concepts are being performed Additional ex-periments on different nozzle concepts with forced- ow sep-aration devices eg dual-bell nozzles are currently plannedat DLR

Different advanced nozzle concepts have been examined andare under examination now at the Keldysh Center with bothCFD calculations and experiments on the Nozzle DifferentialFacility which has a nozzle performance determination errorof about 005 This high accuracy is achieved by simulta-neously running the test nozzle and a reference nozzle bothmounted on a common longitudinal axis The performancecharacteristics of the reference nozzle are well known and thethrust difference is measured with strain gauges This facilityis equipped with a wind tunnel to determine nozzle thrust per-formance in the presence of external transonic ows withMach numbers up to 12

In addition to aerodynamic performance other technical is-sues (weight cost design thermal management manufactur-ing system performance and reliability) must be addressedFurthermore before a nal decision can be made as to whichnozzle concept offers the greatest bene ts with regard to aneffective payload mass injection combined launcher and tra-jectory calculations must be performed and compared to a ref-erence launcher concept with conventional nozzles Differentnozzle ef ciencies which account for the additional losses ofadvanced rocket nozzles and are extracted from numerical sim-ulations and experiments must be taken into account

At least for some of these nozzle concepts the plug nozzlethe dual-bell nozzle and the dual-expander nozzle bene tswith regard to lower overall launcher masses have been dem-onstrated in the literature Furthermore the plug nozzle con-cept will be the rst in ight of all of these advanced nozzleconcepts with the X-33 demonstrator vehicle

AcknowledgmentPresented as Paper 18 at the 3rd International Symposium

on Space Propulsion Beijing Peoplersquos Republic of ChinaAug 11ndash 13 1997

References1Manski D and Hagemann G lsquolsquoIn uence of Rocket Design Pa-

rameters on Engine Nozzle Ef cienciesrsquorsquo Journal of Propulsion andPower Vol 12 No 1 1996 pp 41ndash 47

2lsquolsquoLiquid Rocket Engine Nozzlesrsquorsquo NASA Space Vehicle DesignCriteria NASA SP-8120 1976

3Dumnov G E Nikulin G Z and Ponomaryov N B lsquolsquoInves-tigation of Advanced Nozzles for Rocket Enginesrsquorsquo Space Rocket En-gines and Power Plants Vol 4 No 142 NIITP 1993 pp 1012 ndash

1018 (in Russian)4Nguyen T V and Pieper J L lsquolsquoNozzle Flow Separationrsquorsquo Pro-

ceedings of the 5th International Symposium of Propulsion in SpaceTransportation (Paris France) 1996

5Manski D lsquolsquoClustered Plug Nozzles for Future European Reus-able Rocket Launchersrsquorsquo German Aerospace Research Establishment643-817 Lampoldshausen Germany 1981 (in German)

6Horn M and Fisher S lsquolsquoDual-Bell Altitude Compensating Noz-

HAGEMANN ET AL 627

Fig 17 Flow phenomena of a plug nozzle with full length (left column) and truncated central body (right column) at different pressureratios pc pamb off-design (top bottom) and design (center) pressure ratio

Fig 16 Principle design of plug nozzles clustered plug 36 mod-ules with truncated plug body

Fig 15 Principle design of plug nozzles toroidal plug fulllength

be advantageous because further losses induced by the gapsbetween individual modules and the ow eld interactionsdownstream of the module exits can be minimized It has beenshown that transition from a round to a square nozzle resultsin a very small performance loss31 In principle the ow elddevelopment of a clustered plug nozzle with rectangular nozzlemodules is similar to that of a toroidal plug nozzle but avoidsthe inherent disadvantages of the toroidal plug design regard-ing 1) the control of a constant throat gap during manufactur-

ing and thermal expansion (side-loads and thrust vector devi-ations) 2) the cooling of toroidal throat with tiny throat gapsand 3) the control of combustion instabilities in the toroidalcombustion chamber Another plug nozzle con guration is thelinear plug nozzle which is foreseen for the propulsion systemof the RLV X-33 concept

1 Circular Plug Nozzles

Figure 17 summarizes the principle ow phenomena of cir-cular plug nozzles with full length and truncated central bodiesat different off-design (top and bottom) and design (center)pressure ratios that were observed in experiments and numer-ical simulations For pressure ratios lower than the design pres-sure ratio of a plug nozzle with a well-contoured central bodythe ow expands near the central plug body without separa-tion and a system of recompression shocks and expansionwaves adapts the exhaust ow to the ambient pressure pambThe characteristic barrel-like form with several in ections ofthe shear-layer results from various interactions of compres-sion and expansion waves with the shear layer and turbulentdiffusion enlarges the shear layer farther downstream of thethroat The existence of the overexpansion and recompressionprocesses is inferred from up- and down-variations of plugwall pressure pro les observed in various cold- ow tests andnumerical simulations and will also be shown later for linearplug nozzles

At the design pressure ratio (see Fig 17 left column cen-ter) the characteristic with the design Mach number should bea straight line emanating to the tip of the central plug bodyand the shear layer is parallel to the centerline However forcircular plug nozzles designed with contouring methods pro-posed in Refs 29 and 30 no exact one-dimensional exit owpro le can be achieved because both methods use thePrandtlndash Meyer relations that are only valid for planar owsFurthermore nonhomogeneous ow in the throat regionwhich is in general not considered within the contour designprocess also in uences the exit ow pro le The wall pressuredistribution remains constant at pressure ratios above the de-sign pressure ratio ie the plug nozzle behaves like a con-ventional nozzle the loss of its capability of further altitude

628 HAGEMANN ET AL

Table 2 Design parameters of toroidal plug nozzles

Design data Full-scale nozzlea Subscale nozzleb

Chamber pressure pc 100 bar 12 barPropellants Hydrogenoxygen Gasndash oilnitric-acidMixture ratio r 60 25Inner chamber diameter dcinner 43 m 019 mOuter chamber diameter dcouter 44 m 034 mGeometric area ratio laquo 55 10aUsed for numerical simulations and performance predictions see Fig 18

bUsed in experiments see Fig 20

Fig 18 Performance of numerically simulated plug nozzle withfull-length central body

Fig 19 Performance of numerically simulated plug nozzle withtruncated central body

adaptation being included Figure 17 (left column bottom) il-lustrates the ow eld at higher pressure ratios Performancedata of a typical plug nozzle are included in Fig 18 and com-pared to a conventional bell nozzle with equal area ratio Thedesign and combustion chamber parameters are included in theleft column of Table 227

The truncation of the central plug body which is an advan-tage because of the huge length and high structural mass ofthe well-contoured central body results in a different ow andperformance behavior as compared to the full-length plug noz-zle At lower pressure ratios an open wake ow establisheswith a pressure level practically equal to the ambient pressure(Fig 17 right column top) At a certain pressure ratio closeto the design pressure ratio of the full-length plug nozzle thebase ow suddenly changes its character and turns over to theclosed form characterized by a constant base pressure that isno longer in uenced by the ambient pressure Analyses indi-cate that shorter plug bodies with higher truncations trigger anearlier change in wake ow at pressure ratios below the designpressure ratio At the transition point the pressure within thewake approaches a value that is below ambient pressure andthe full base area induces a negative thrust (Fig 17 right col-umn center) This thrust loss depends on the percentage oftruncation and the total size of the base area Published ex-perimental data and numerical simulations reveal an increasingthrust loss for shorter plug bodies because the total base areaincreases

Beyond the transition point the pressure within the closedwake remains constant At these lower ambient pressures thebase pressure is then higher than the ambient pressure result-ing in a positive thrust contribution of the total base area Per-formance data of a numerically simulated truncated plug noz-zle are shown in Fig 19 and compared to the same plugnozzle with full-length central body and a conventional bellnozzle Design parameters for this truncated plug nozzle arethe same as for the full length plug (see Fig 18 and Table 2)

Figure 20 shows a typical photograph of a sea-level hot-runtest with a truncated toroidal subscale plug nozzle performedat DLR5 Nozzle design data are included in Table 2 For com-parison of experimental results with numerical computations

the ow eld of the toroidal plug nozzle was calculated with anumerical method10 and Fig 20 shows the calculated Mach-number distribution in the combustion chamber and nozzlePrincipal physical processes like expansion waves shocks andthe recirculating base- ow region are in good agreement Boththe experiment and the numerical simulation show that the ow separates from the conical plug body before reaching itstruncated end Recent results on experiments with plug nozzlespublished in Ref 28 also reveal separation of the ow fromthe central plug body for conical contours In contrast no sep-aration was observed for contoured central plug bodies de-signed with the method proposed in Ref 29 Recent numericalsimulations of contoured plug nozzles performed at DLRwithin the ESA Advanced Rocket Propulsion Technology(ARPT) Program on advanced rocket propulsion also showthat no ow separation occurs from well-contoured full-lengthcentral plug bodies26 Principal ow eld developments pre-dicted by these numerical simulations are again in a goodagreement with experimental data published in Ref 28 Withinthe frame of this ESA ARPT Program performance and owbehavior of clustered plug nozzles at different truncations arebeing examined by European industries [Societe Europeennede Propulsion (SEP) Volvo Dasa] and research institutes(ONERA DLR) with subscale cold- ow plug models26 (Table3) Numerous experimental investigations on subscale modelsof circular and clustered plug nozzles with cold- and hot-gas ow were performed at the Keldysh Center with and withoutexternal ows Thrust characteristics pressure and heat- uxdistributions along the plug and acoustic characteristics of thecircular jet were investigated

2 Linear Plug Nozzles

Performance behavior and ow eld development for linearplug nozzles as a function of ambient pressures are in principlesimilar to those of circular plug nozzles (Fig 17) Howeverspecial attention must be paid to the in uence of both endsides where the surrounding ow disturbs the expanding ow- eld resulting in an expansion of the ow normal to the main ow direction and therefore in an effective performance loss

HAGEMANN ET AL 629

Table 3 Design parameters for linear and toroidal subscale plug nozzlesa

Design data Linear plug (FESTIP) Circular clustered plug (ARPT)

Chamber pressure pc 875 bar 80 barDriving gas Air AirMass ow rate mCcedil 595 kgs 546 kgsDesign pressure ratio pc pamb 200 200Geometric area ratio laquo 1265 1265aUsed in the ESA FESTIP and ESA ARPT studies

Fig 20 Plug nozzle ow eld experiment vs numerical simula-tion a) toroidal plug nozzle experiment with gas-oilnitric-acidpropellant combination side view photograph and b) computa-tional results of toroidal plug nozzle Mach number distributionfull gray and isolines centerline section

Fig 21 Design of linear subscale plug nozzle of the ESA FESTIPTechnology Program (plug size 192 mm high 214 mm wide)

Fig 22 Calculated wall pressure data for linear subscale plugnozzle at different pressure ratios

The change of wake ow behavior may be strongly in uencedby the penetration of ambient pressures through both end sidesparticularly for truncated plug nozzles End plates as foreseenfor the linear plug nozzle of the X-33 demonstrator vehiclecould be used to avoid this ambient pressure penetration

Within the framework of the ESA FESTIP Technology Pro-gram lsquolsquoTechnology Developments on Rocket and Air Breath-ing Propulsion for Reusable Launch Vehiclesrsquorsquo subscale cold-gas tests and numerical simulations of linear plug nozzles arebeing performed by Dasa at the high-speed wind tunnel [Hoch-geschwindigheitswind kanal (HWK) in Bad Salzungen Ger-many] of the Technical University of Dresden719 Design pa-

rameters for the linear plug model are included in Table 3Figure 21 shows the baseline model with a linear throat Afterpassing the linear throat the ow expands directly onto theplug surface A second model will be tested with a bell-shapedlinear nozzle extension following the linear throat for internalexpansion prior to further external expansion onto the plugsurface19 The contour of the plug was designed by the methodof Angelino29 and three different plug lengths as 5 20 and40 of the ideal length were tested Figure 22 shows calcu-lated wall pressures for the full-length plug for different am-bient pressures simulating the operational range of sea-leveloperation up to higher-altitude operation The ow structure inform of the left- and right-running characteristics at two pres-sure ratios is shown in Fig 23 For lower pressure ratios thanthe design pressure ratio of pc pamb = 200 the in uence of the

630 HAGEMANN ET AL

Fig 23 Left- and right-running characteristics in ow eld at twopressure ratios pc pamb = a) 8 and b) 200 (inviscid analysis withmethod of characteristics)

Fig 24 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 20 plug body

Fig 25 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 40 plug body

Fig 27 Flow phenomena of an E ndash D nozzle with a) open and b)closed wakes

Fig 26 Sketch of an E ndash D nozzle

system of recompression shocks and expansion waves on thewall pressure distribution (Figs 17 and 22) can be seen bythe waviness of the pressure distribution on the plug surfaceFirst experimental results are shown in Figs 24 and 25 whichalso show the waviness of the pressure distribution for low-pressure ratios

3 Plug Nozzles Final Aspects

The performance of circular and linear plug nozzles is ad-versely affected by the external airstream ow ie the aspi-ration effect of the external ow slightly reduces performanceand favors an earlier change in wake- ow development for

truncated plug nozzles32ndash 35 Cold- ow tests showed that theeffect of external ow on performance is con ned to a narrowrange of external ow speeds with Mach numbers near unity

The altitude compensation capability of circular and linearplug nozzles for higher ambient pressures is indisputable Be-cause plug nozzles lose this capability for pressure ratios abovethe design pressure ratio the latter should be chosen as highas possible Taking this into account plug nozzles will featurean even better overall performance than those shown in Figs18 and 19 Unfortunately for equal geometrical area ratiosplug nozzles perform worse at high altitude than do conven-tional bell nozzles because of truncation and clustering Forhigh-area-ratio nozzles with relatively short lengths plug noz-zles perform better than conventional bell nozzles In additionto having excellent capabilities for altitude compensation plugnozzles have additional advantages including ease in vehicleand engine integration

C E ndash D Nozzles

An expansionndash de ection (E ndash D) nozzle is shown in Fig 26E ndash D nozzles were at one time thought to have capabilities foraltitude compensation because the gas expansion takes placewith a constant pressure free boundary Thus the aerodynamicbehavior of E ndash D nozzles as a function of altitude is in prin-ciple quite similar to plug nozzles because the expansion pro-cess is controlled by the ambient pressure hence by the alti-tude In contrast to plug nozzles however the expansionprocess is controlled from inside the nozzle for E ndash D nozzlesAt low altitude the higher ambient pressure limits the gasexpansion resulting in a low effective expansion area ratioThe exhaust gas is adapted to the ambient pressure level by

HAGEMANN ET AL 631

Fig 30 Performance losses in dual-throat nozzles as a functionof nozzle design parameters (with r4r1 = 352 r2r1 = 116 and mCcedil 2

= 00)

Fig 29 Flow phenomena during a) sea-level and b) high-altitudeoperations

Fig 28 Sketch of a dual-throat nozzle view of combustion cham-ber and throat region

systems of recompression and expansion waves (Fig 27) Athigher altitude the lower ambient pressure allows more gasexpansion within the nozzle resulting in a higher effectiveexpansion area ratio But in contrast to plug nozzles the pres-sure in the wake of the center plug is always less than theambient pressure because of the aspiration effect This occursat low pressure ratios when the wake is opened and results inan aspiration loss Furthermore because the exhaust ow ex-pands to this base pressure rather than to the ambient pressurelevel wall pressures downstream are overexpanded This re-sults in an additional overexpansion loss As the pressure ratioincreases the wake region closes and is thus totally isolatedfrom the ambient environment (see Fig 27) The behavior dur-ing transition from open wake to closed wake is again equalto plug nozzles and the base pressure in the closed-wake re-gion is essentially independent of the ambient pressure

The E ndash D nozzle concept has also been a subject of numer-ous analytical and experimental studies Results from thesestudies show that E ndash D nozzle capabilities for altitude com-pensation are poor and are in fact worse than those of plugnozzles because of aspiration and overexpansion losses35 Forhigh-area-ratio nozzles with a relatively short length an E ndash Dnozzle performs better than a comparable conventional bellnozzle at the same length because of lower divergence andpro le losses than the bell nozzle

The advantages of the E ndash D nozzle concept include its smallengine envelope and no moving parts However like toroidalplug nozzles E ndash D nozzles have the disadvantage of havinghigher throat heat uxes relative to a conventional bell nozzlewith an equal throat area The higher throat heat ux resultsagain from the relatively thin annular throat gap This problemhowever can be remedied with the modular thrust cell clusterconcept Additional advantages of E ndash D nozzles with clusteredthrust cells are the same as those already discussed for theclustered plug nozzle concept with regard to ease in manu-facturing thrust vector control by throttling or shutting off anindividual or group of thrust cells and lower nozzle throatheating

D Nozzles with Throat Area Varied by a Mechanical Pintle

This nozzle concept utilizes a conventional bell nozzle witha xed exit area and a mechanical pintle in the combustionchamber and throat region to vary the throat area and hencethe expansion area ratio The area of the nozzle throatmdashanannulus between the pintle and the shroudmdashis varied by mov-ing the pintle axially

The pintle concept has been used in solid rocket motors asa mean to provide variable thrust The concept in principleallows a continuous variation of the throat area and thus op-timum expansion area ratios throughout a mission Howeverit requires an actuator and a sophisticated control system Theconcept raises issues of engine weight design complexitycooling of the pintle and nozzle throat and reliability

In Ref 36 the aerodynamic performances of nozzles with ve different pintle geometries were calculated and comparedwith a reference bell nozzle The performance losses of thesepintle nozzles when compared with the bell nozzle are in therange of 1ndash 25 Performances of a xed pintle geometry atthree different locations were also calculated The results showthat the performance loss varies with the pintle location

E Dual-Mode Nozzles

Dual-mode rocket engines using one or two fuels offer atrajectory-adapted dual-mode operation during the ascent of alauncher which may be of signi cant advantage for single-stage Earth-to-orbit vehicles as compared to conventionalrocket engines with bell-type nozzles This engine concept in-volves the use of a dense propellant combination with mod-erate performance during liftoff to provide high thrust duringthe initial ight phase and a better performing propellant com-bination in vacuum which result in higher speci c impulse

The fuels are burned in two different combustion chamberswith one located completely inside the other in the case ofengines with dual-throat nozzles or with a conventional bellthrust chamber surrounded by an annular thrust chamber in thecase of dual-expander engines This type of engine has a built-in accelerationndash reduction capability achieved by shuttingdown one of two thrust chambers The total engine thrust isthen provided by the remaining thrust chamber with the useof the total nozzle exit area leading to an increase in speci cimpulse Apart from the indicated bene ts of dual-mode en-gines which will be discussed in more detail later their de-

632 HAGEMANN ET AL

Table 4 Thrust chamber parameters for simulated dual-expander nozzlesa

Operation mode

Aerojet

First (inout) Second (inout)

DLR

First (inout) Second (inout)

Propellants C3H8ndash O2H2ndash O2 mdashH2ndash O2 H2ndash O2H2ndash O2 mdashH2ndash O2

Chamber pressure pc 414207 bar mdash207 bar 200200 bar mdash200 barMixture ratio r 337 mdash7 77 mdash7Exit area ratio laquo 69 146 58 116aInnerouter chambers

Fig 32 Flow phenomena in a dual-expander nozzle during a)sea-level- and b) high-altitude operation

Fig 31 Sketch of a dual-expander nozzle view of combustionchamber and throat region

velopment and construction require considerable technologicaleffort

1 Dual-Throat Nozzles

A dual-throat nozzle con guration is shown in Fig 28 Itconsists of two conventional bell-thrust chambers with onelocated completely inside the other At low altitude the outerthrust chamber operates with the inner thrust chamber runningin parallel In this operation mode the engine has a larger throatproviding a moderate expansion area ratio During the missionthe outer thrust chamber is shut off and operation continueswith only the inner engine In this con guration ow fromthe inner engine expands and attaches supersonically to theouter engine resulting in a higher expansion area ratio for theremainder of the burn Flow phenomena in both operationmodes are included in Fig 29

Hot- red tests at Aerojet were conducted to provide heattransfer data that were very useful for the thermal analysis anddesign of the dual-throat nozzle con guration37 These testsshowed that ow separation occurred in the inner engine noz-zle at higher ratios of outer to inner chamber pressures duringthe rst operation mode with both chambers burning in par-allel The ow separation resulted in a higher heat load to theinner nozzle Subscale tests performed at the Keldysh Center3

have shown that the additional loss caused by the nozzle con-tour discontinuity during vacuum operation with active innerchamber is in the range of 08ndash 4 depending on geometricaldata (see Fig 30) This high-performance loss results from theinteraction of the inner chamber jet with the outer chambernozzle wall3 The decrease of the jet incidence angle on thewall by means of gas injection through the outer chamber re-duces this performance loss by 04ndash 073

2 Dual-Expander Nozzles

A dual-expander nozzle has two concentric thrust chambersand nozzles It consists of a conventional bell thrust chambersurrounded by an annular thrust chamber Both chambers haveshort primary nozzles which end in a common divergent noz-zle extension Figure 31 shows a typical dual-expander nozzlecon guration At low altitude both thrust chambers operatesharing the same exit area which results in a moderate expan-sion area ratio Part way into the mission one thrust chamberis shut off allowing the other nozzle to use the whole exitarea creating a high-expansion-area ratio for the remainder of

the burn In principle the two operation modes are comparableto those of dual-throat nozzles

Numerical simulations of the ow elds in dual-expandernozzles during all operation modes have been performed atAerojet38 DLR39 and the Keldysh Center Aerojet simulationsare based on an engine design using different propellantshydrocarbonoxygen for the inner chamber and hydrogenox-ygen for the outer chamber whereas DLR simulations arebased on a combined vehicleengine analysis using single-fuelsingle-mixture ratio dual-expander engines with hydrogenoxy-gen Table 4 summarizes combustion chamber parameters Flowphenomena observed in numerical simulations are shown in Fig32a for the mode 1 operation with both thrust chambers burn-ing Compression waves are induced near the inner nozzle lipas a result of the strongly inhomogeneous ow character at thepoint where both exhaust gases of inner and outer combustionchamber merge into the common divergent nozzle part Upand down variations of pressure ratios at the lip did not changethis wave formation even in cases of signi cantly lower exitpressures in the inner nozzle compared to the pressure eld ofthe outer nozzle at the lip Further downstream the compres-sion waves interact with the wall resulting in a re ection ofthe compression waves back into the ow eld

During mode 2 operation a strong expansion of the outernozzle ow is observed when the expansion ratio suddenlyincreases at the end of the nozzle lip The ow is directedtoward the axis of symmetry Near the centerline the ow thenturns over to the axial direction inducing a recompressionshock The static pressure rises signi cantly in this recom-pression region on the centerline A sub- and supersonic recir-culation zone establishes in the inner chamber of the dual-expander nozzle Figure 32b emphasizes the essential owpattern in this operation mode

These analyses have shown that dual-expander nozzles pro-duce high performance in both operation modes3839 Figure 33summarizes performance behavior as a function of ight alti-tude for the dual-expander engine simulated at DLR For com-parison performance data are included for two reference bell

HAGEMANN ET AL 633

Fig 33 Performance data for a dual-expander nozzle

nozzles that have the same area ratios as the dual-expandernozzle during its two operation modes Losses because of thenonhomogeneous exit ow and the induced shocks are com-parable with corresponding values of both conventional noz-zles1

Within the design process of dual-expander engines it is nec-essary to take into account that the nonburning chamber isexposed to high heat uxes and pressure oscillations duringmode 2 operation which might reach a high level and lead tostructural disintegration of the inner nozzle structure There-fore a mode-2 operation with bleed gas generation at a ratherlow thrust chamber pressure level in the inner chamber seemsto be advantageous Because of the bleed gas generation in theprimary chamber the overall mass ow rate during this alter-native mode-2 operation is increased In addition the reductionof the effective exit area ratio results in minor impulse deg-radations during this operation mode when compared with themode-2 operation with no bleed gas generation

Several analytical works on SSTO- and two-stage-to-orbit(TSTO)-vehicles using hydrogenpropane or hydrogenmeth-ane as fuels revealed the lowest vehicle dry masses for dual-mode engines in comparison to other engines1439 Other dual-mode engines using hydrogen as the single fuel but using twomixture ratios also revealed some bene ts over conventionalengines for SSTO- and TSTO applications Even a single-fueloperation with constant mixture ratios in both combustionchambers indicated a gain in launcher performance40

V ConclusionsSeveral nozzle concepts that promise gains in performance

over conventional nozzles were discussed in this paper in-cluding performance enhancements achieved by slight modi- cations of existing nozzles eg through cool gas injectioninto the supersonic nozzle part It is shown that signi cantperformance gains result from the adaptation of the exhaust ow to the ambient pressure and special emphasis was givento altitude adaptive nozzle concepts

A number of nozzle concepts with altitude-compensating ca-pability were identi ed and described To assist the selectionof the best nozzle concept for launch vehicle applications theperformance of the nozzles must be characterized This can bedone using computational uid dynamics (CFD) andor cold- ow tests Existing CFD methods that are in use in the aero-space industry and at research institutes have been veri ed fora wide number of sub- and full-scale experiments and providesuf ciently reliable performance determination for the differ-ent nozzle types

Theoretical evaluations numerical simulations and test re-sults showed that the different concepts have real altitude-com-pensating capabilities However the compensation capabilitiesare limited and there are some drawbacks associated with eachof the concepts Additional performance losses are induced inpractically all of these nozzle concepts when compared to anideal expansion mainly because of the nonisentropic effects

such as shock waves and pressure losses in recirculation zonesHowever these additional performance losses are less than1 ndash 3 depending on the different nozzle concepts

Four nozzle concepts have been selected for further evalu-ation under a NASA Aerojet cooperative agreement They in-clude a plug nozzle an E ndash D nozzle a dual-bell nozzle and adual-expander nozzle The nozzles in addition to a referencebell nozzle have been designed by Aerojet for cold ow testsCFD calculations performed at Aerojet demonstrated that allof these nozzles have high and approximately equal perfor-mances at the design points

In Europe special attention is being paid to plug- and dual-bell nozzles Different plug nozzle concepts are being testedin cold- ow tests within the ESA ARPT (clustered plug noz-zle) and ESA FESTIP (linear plug nozzle) contracts In addi-tion to experiments three-dimensional numerical simulationson these nozzle concepts are being performed Additional ex-periments on different nozzle concepts with forced- ow sep-aration devices eg dual-bell nozzles are currently plannedat DLR

Different advanced nozzle concepts have been examined andare under examination now at the Keldysh Center with bothCFD calculations and experiments on the Nozzle DifferentialFacility which has a nozzle performance determination errorof about 005 This high accuracy is achieved by simulta-neously running the test nozzle and a reference nozzle bothmounted on a common longitudinal axis The performancecharacteristics of the reference nozzle are well known and thethrust difference is measured with strain gauges This facilityis equipped with a wind tunnel to determine nozzle thrust per-formance in the presence of external transonic ows withMach numbers up to 12

In addition to aerodynamic performance other technical is-sues (weight cost design thermal management manufactur-ing system performance and reliability) must be addressedFurthermore before a nal decision can be made as to whichnozzle concept offers the greatest bene ts with regard to aneffective payload mass injection combined launcher and tra-jectory calculations must be performed and compared to a ref-erence launcher concept with conventional nozzles Differentnozzle ef ciencies which account for the additional losses ofadvanced rocket nozzles and are extracted from numerical sim-ulations and experiments must be taken into account

At least for some of these nozzle concepts the plug nozzlethe dual-bell nozzle and the dual-expander nozzle bene tswith regard to lower overall launcher masses have been dem-onstrated in the literature Furthermore the plug nozzle con-cept will be the rst in ight of all of these advanced nozzleconcepts with the X-33 demonstrator vehicle

AcknowledgmentPresented as Paper 18 at the 3rd International Symposium

on Space Propulsion Beijing Peoplersquos Republic of ChinaAug 11ndash 13 1997

References1Manski D and Hagemann G lsquolsquoIn uence of Rocket Design Pa-

rameters on Engine Nozzle Ef cienciesrsquorsquo Journal of Propulsion andPower Vol 12 No 1 1996 pp 41ndash 47

2lsquolsquoLiquid Rocket Engine Nozzlesrsquorsquo NASA Space Vehicle DesignCriteria NASA SP-8120 1976

3Dumnov G E Nikulin G Z and Ponomaryov N B lsquolsquoInves-tigation of Advanced Nozzles for Rocket Enginesrsquorsquo Space Rocket En-gines and Power Plants Vol 4 No 142 NIITP 1993 pp 1012 ndash

1018 (in Russian)4Nguyen T V and Pieper J L lsquolsquoNozzle Flow Separationrsquorsquo Pro-

ceedings of the 5th International Symposium of Propulsion in SpaceTransportation (Paris France) 1996

5Manski D lsquolsquoClustered Plug Nozzles for Future European Reus-able Rocket Launchersrsquorsquo German Aerospace Research Establishment643-817 Lampoldshausen Germany 1981 (in German)

6Horn M and Fisher S lsquolsquoDual-Bell Altitude Compensating Noz-

628 HAGEMANN ET AL

Table 2 Design parameters of toroidal plug nozzles

Design data Full-scale nozzlea Subscale nozzleb

Chamber pressure pc 100 bar 12 barPropellants Hydrogenoxygen Gasndash oilnitric-acidMixture ratio r 60 25Inner chamber diameter dcinner 43 m 019 mOuter chamber diameter dcouter 44 m 034 mGeometric area ratio laquo 55 10aUsed for numerical simulations and performance predictions see Fig 18

bUsed in experiments see Fig 20

Fig 18 Performance of numerically simulated plug nozzle withfull-length central body

Fig 19 Performance of numerically simulated plug nozzle withtruncated central body

adaptation being included Figure 17 (left column bottom) il-lustrates the ow eld at higher pressure ratios Performancedata of a typical plug nozzle are included in Fig 18 and com-pared to a conventional bell nozzle with equal area ratio Thedesign and combustion chamber parameters are included in theleft column of Table 227

The truncation of the central plug body which is an advan-tage because of the huge length and high structural mass ofthe well-contoured central body results in a different ow andperformance behavior as compared to the full-length plug noz-zle At lower pressure ratios an open wake ow establisheswith a pressure level practically equal to the ambient pressure(Fig 17 right column top) At a certain pressure ratio closeto the design pressure ratio of the full-length plug nozzle thebase ow suddenly changes its character and turns over to theclosed form characterized by a constant base pressure that isno longer in uenced by the ambient pressure Analyses indi-cate that shorter plug bodies with higher truncations trigger anearlier change in wake ow at pressure ratios below the designpressure ratio At the transition point the pressure within thewake approaches a value that is below ambient pressure andthe full base area induces a negative thrust (Fig 17 right col-umn center) This thrust loss depends on the percentage oftruncation and the total size of the base area Published ex-perimental data and numerical simulations reveal an increasingthrust loss for shorter plug bodies because the total base areaincreases

Beyond the transition point the pressure within the closedwake remains constant At these lower ambient pressures thebase pressure is then higher than the ambient pressure result-ing in a positive thrust contribution of the total base area Per-formance data of a numerically simulated truncated plug noz-zle are shown in Fig 19 and compared to the same plugnozzle with full-length central body and a conventional bellnozzle Design parameters for this truncated plug nozzle arethe same as for the full length plug (see Fig 18 and Table 2)

Figure 20 shows a typical photograph of a sea-level hot-runtest with a truncated toroidal subscale plug nozzle performedat DLR5 Nozzle design data are included in Table 2 For com-parison of experimental results with numerical computations

the ow eld of the toroidal plug nozzle was calculated with anumerical method10 and Fig 20 shows the calculated Mach-number distribution in the combustion chamber and nozzlePrincipal physical processes like expansion waves shocks andthe recirculating base- ow region are in good agreement Boththe experiment and the numerical simulation show that the ow separates from the conical plug body before reaching itstruncated end Recent results on experiments with plug nozzlespublished in Ref 28 also reveal separation of the ow fromthe central plug body for conical contours In contrast no sep-aration was observed for contoured central plug bodies de-signed with the method proposed in Ref 29 Recent numericalsimulations of contoured plug nozzles performed at DLRwithin the ESA Advanced Rocket Propulsion Technology(ARPT) Program on advanced rocket propulsion also showthat no ow separation occurs from well-contoured full-lengthcentral plug bodies26 Principal ow eld developments pre-dicted by these numerical simulations are again in a goodagreement with experimental data published in Ref 28 Withinthe frame of this ESA ARPT Program performance and owbehavior of clustered plug nozzles at different truncations arebeing examined by European industries [Societe Europeennede Propulsion (SEP) Volvo Dasa] and research institutes(ONERA DLR) with subscale cold- ow plug models26 (Table3) Numerous experimental investigations on subscale modelsof circular and clustered plug nozzles with cold- and hot-gas ow were performed at the Keldysh Center with and withoutexternal ows Thrust characteristics pressure and heat- uxdistributions along the plug and acoustic characteristics of thecircular jet were investigated

2 Linear Plug Nozzles

Performance behavior and ow eld development for linearplug nozzles as a function of ambient pressures are in principlesimilar to those of circular plug nozzles (Fig 17) Howeverspecial attention must be paid to the in uence of both endsides where the surrounding ow disturbs the expanding ow- eld resulting in an expansion of the ow normal to the main ow direction and therefore in an effective performance loss

HAGEMANN ET AL 629

Table 3 Design parameters for linear and toroidal subscale plug nozzlesa

Design data Linear plug (FESTIP) Circular clustered plug (ARPT)

Chamber pressure pc 875 bar 80 barDriving gas Air AirMass ow rate mCcedil 595 kgs 546 kgsDesign pressure ratio pc pamb 200 200Geometric area ratio laquo 1265 1265aUsed in the ESA FESTIP and ESA ARPT studies

Fig 20 Plug nozzle ow eld experiment vs numerical simula-tion a) toroidal plug nozzle experiment with gas-oilnitric-acidpropellant combination side view photograph and b) computa-tional results of toroidal plug nozzle Mach number distributionfull gray and isolines centerline section

Fig 21 Design of linear subscale plug nozzle of the ESA FESTIPTechnology Program (plug size 192 mm high 214 mm wide)

Fig 22 Calculated wall pressure data for linear subscale plugnozzle at different pressure ratios

The change of wake ow behavior may be strongly in uencedby the penetration of ambient pressures through both end sidesparticularly for truncated plug nozzles End plates as foreseenfor the linear plug nozzle of the X-33 demonstrator vehiclecould be used to avoid this ambient pressure penetration

Within the framework of the ESA FESTIP Technology Pro-gram lsquolsquoTechnology Developments on Rocket and Air Breath-ing Propulsion for Reusable Launch Vehiclesrsquorsquo subscale cold-gas tests and numerical simulations of linear plug nozzles arebeing performed by Dasa at the high-speed wind tunnel [Hoch-geschwindigheitswind kanal (HWK) in Bad Salzungen Ger-many] of the Technical University of Dresden719 Design pa-

rameters for the linear plug model are included in Table 3Figure 21 shows the baseline model with a linear throat Afterpassing the linear throat the ow expands directly onto theplug surface A second model will be tested with a bell-shapedlinear nozzle extension following the linear throat for internalexpansion prior to further external expansion onto the plugsurface19 The contour of the plug was designed by the methodof Angelino29 and three different plug lengths as 5 20 and40 of the ideal length were tested Figure 22 shows calcu-lated wall pressures for the full-length plug for different am-bient pressures simulating the operational range of sea-leveloperation up to higher-altitude operation The ow structure inform of the left- and right-running characteristics at two pres-sure ratios is shown in Fig 23 For lower pressure ratios thanthe design pressure ratio of pc pamb = 200 the in uence of the

630 HAGEMANN ET AL

Fig 23 Left- and right-running characteristics in ow eld at twopressure ratios pc pamb = a) 8 and b) 200 (inviscid analysis withmethod of characteristics)

Fig 24 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 20 plug body

Fig 25 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 40 plug body

Fig 27 Flow phenomena of an E ndash D nozzle with a) open and b)closed wakes

Fig 26 Sketch of an E ndash D nozzle

system of recompression shocks and expansion waves on thewall pressure distribution (Figs 17 and 22) can be seen bythe waviness of the pressure distribution on the plug surfaceFirst experimental results are shown in Figs 24 and 25 whichalso show the waviness of the pressure distribution for low-pressure ratios

3 Plug Nozzles Final Aspects

The performance of circular and linear plug nozzles is ad-versely affected by the external airstream ow ie the aspi-ration effect of the external ow slightly reduces performanceand favors an earlier change in wake- ow development for

truncated plug nozzles32ndash 35 Cold- ow tests showed that theeffect of external ow on performance is con ned to a narrowrange of external ow speeds with Mach numbers near unity

The altitude compensation capability of circular and linearplug nozzles for higher ambient pressures is indisputable Be-cause plug nozzles lose this capability for pressure ratios abovethe design pressure ratio the latter should be chosen as highas possible Taking this into account plug nozzles will featurean even better overall performance than those shown in Figs18 and 19 Unfortunately for equal geometrical area ratiosplug nozzles perform worse at high altitude than do conven-tional bell nozzles because of truncation and clustering Forhigh-area-ratio nozzles with relatively short lengths plug noz-zles perform better than conventional bell nozzles In additionto having excellent capabilities for altitude compensation plugnozzles have additional advantages including ease in vehicleand engine integration

C E ndash D Nozzles

An expansionndash de ection (E ndash D) nozzle is shown in Fig 26E ndash D nozzles were at one time thought to have capabilities foraltitude compensation because the gas expansion takes placewith a constant pressure free boundary Thus the aerodynamicbehavior of E ndash D nozzles as a function of altitude is in prin-ciple quite similar to plug nozzles because the expansion pro-cess is controlled by the ambient pressure hence by the alti-tude In contrast to plug nozzles however the expansionprocess is controlled from inside the nozzle for E ndash D nozzlesAt low altitude the higher ambient pressure limits the gasexpansion resulting in a low effective expansion area ratioThe exhaust gas is adapted to the ambient pressure level by

HAGEMANN ET AL 631

Fig 30 Performance losses in dual-throat nozzles as a functionof nozzle design parameters (with r4r1 = 352 r2r1 = 116 and mCcedil 2

= 00)

Fig 29 Flow phenomena during a) sea-level and b) high-altitudeoperations

Fig 28 Sketch of a dual-throat nozzle view of combustion cham-ber and throat region

systems of recompression and expansion waves (Fig 27) Athigher altitude the lower ambient pressure allows more gasexpansion within the nozzle resulting in a higher effectiveexpansion area ratio But in contrast to plug nozzles the pres-sure in the wake of the center plug is always less than theambient pressure because of the aspiration effect This occursat low pressure ratios when the wake is opened and results inan aspiration loss Furthermore because the exhaust ow ex-pands to this base pressure rather than to the ambient pressurelevel wall pressures downstream are overexpanded This re-sults in an additional overexpansion loss As the pressure ratioincreases the wake region closes and is thus totally isolatedfrom the ambient environment (see Fig 27) The behavior dur-ing transition from open wake to closed wake is again equalto plug nozzles and the base pressure in the closed-wake re-gion is essentially independent of the ambient pressure

The E ndash D nozzle concept has also been a subject of numer-ous analytical and experimental studies Results from thesestudies show that E ndash D nozzle capabilities for altitude com-pensation are poor and are in fact worse than those of plugnozzles because of aspiration and overexpansion losses35 Forhigh-area-ratio nozzles with a relatively short length an E ndash Dnozzle performs better than a comparable conventional bellnozzle at the same length because of lower divergence andpro le losses than the bell nozzle

The advantages of the E ndash D nozzle concept include its smallengine envelope and no moving parts However like toroidalplug nozzles E ndash D nozzles have the disadvantage of havinghigher throat heat uxes relative to a conventional bell nozzlewith an equal throat area The higher throat heat ux resultsagain from the relatively thin annular throat gap This problemhowever can be remedied with the modular thrust cell clusterconcept Additional advantages of E ndash D nozzles with clusteredthrust cells are the same as those already discussed for theclustered plug nozzle concept with regard to ease in manu-facturing thrust vector control by throttling or shutting off anindividual or group of thrust cells and lower nozzle throatheating

D Nozzles with Throat Area Varied by a Mechanical Pintle

This nozzle concept utilizes a conventional bell nozzle witha xed exit area and a mechanical pintle in the combustionchamber and throat region to vary the throat area and hencethe expansion area ratio The area of the nozzle throatmdashanannulus between the pintle and the shroudmdashis varied by mov-ing the pintle axially

The pintle concept has been used in solid rocket motors asa mean to provide variable thrust The concept in principleallows a continuous variation of the throat area and thus op-timum expansion area ratios throughout a mission Howeverit requires an actuator and a sophisticated control system Theconcept raises issues of engine weight design complexitycooling of the pintle and nozzle throat and reliability

In Ref 36 the aerodynamic performances of nozzles with ve different pintle geometries were calculated and comparedwith a reference bell nozzle The performance losses of thesepintle nozzles when compared with the bell nozzle are in therange of 1ndash 25 Performances of a xed pintle geometry atthree different locations were also calculated The results showthat the performance loss varies with the pintle location

E Dual-Mode Nozzles

Dual-mode rocket engines using one or two fuels offer atrajectory-adapted dual-mode operation during the ascent of alauncher which may be of signi cant advantage for single-stage Earth-to-orbit vehicles as compared to conventionalrocket engines with bell-type nozzles This engine concept in-volves the use of a dense propellant combination with mod-erate performance during liftoff to provide high thrust duringthe initial ight phase and a better performing propellant com-bination in vacuum which result in higher speci c impulse

The fuels are burned in two different combustion chamberswith one located completely inside the other in the case ofengines with dual-throat nozzles or with a conventional bellthrust chamber surrounded by an annular thrust chamber in thecase of dual-expander engines This type of engine has a built-in accelerationndash reduction capability achieved by shuttingdown one of two thrust chambers The total engine thrust isthen provided by the remaining thrust chamber with the useof the total nozzle exit area leading to an increase in speci cimpulse Apart from the indicated bene ts of dual-mode en-gines which will be discussed in more detail later their de-

632 HAGEMANN ET AL

Table 4 Thrust chamber parameters for simulated dual-expander nozzlesa

Operation mode

Aerojet

First (inout) Second (inout)

DLR

First (inout) Second (inout)

Propellants C3H8ndash O2H2ndash O2 mdashH2ndash O2 H2ndash O2H2ndash O2 mdashH2ndash O2

Chamber pressure pc 414207 bar mdash207 bar 200200 bar mdash200 barMixture ratio r 337 mdash7 77 mdash7Exit area ratio laquo 69 146 58 116aInnerouter chambers

Fig 32 Flow phenomena in a dual-expander nozzle during a)sea-level- and b) high-altitude operation

Fig 31 Sketch of a dual-expander nozzle view of combustionchamber and throat region

velopment and construction require considerable technologicaleffort

1 Dual-Throat Nozzles

A dual-throat nozzle con guration is shown in Fig 28 Itconsists of two conventional bell-thrust chambers with onelocated completely inside the other At low altitude the outerthrust chamber operates with the inner thrust chamber runningin parallel In this operation mode the engine has a larger throatproviding a moderate expansion area ratio During the missionthe outer thrust chamber is shut off and operation continueswith only the inner engine In this con guration ow fromthe inner engine expands and attaches supersonically to theouter engine resulting in a higher expansion area ratio for theremainder of the burn Flow phenomena in both operationmodes are included in Fig 29

Hot- red tests at Aerojet were conducted to provide heattransfer data that were very useful for the thermal analysis anddesign of the dual-throat nozzle con guration37 These testsshowed that ow separation occurred in the inner engine noz-zle at higher ratios of outer to inner chamber pressures duringthe rst operation mode with both chambers burning in par-allel The ow separation resulted in a higher heat load to theinner nozzle Subscale tests performed at the Keldysh Center3

have shown that the additional loss caused by the nozzle con-tour discontinuity during vacuum operation with active innerchamber is in the range of 08ndash 4 depending on geometricaldata (see Fig 30) This high-performance loss results from theinteraction of the inner chamber jet with the outer chambernozzle wall3 The decrease of the jet incidence angle on thewall by means of gas injection through the outer chamber re-duces this performance loss by 04ndash 073

2 Dual-Expander Nozzles

A dual-expander nozzle has two concentric thrust chambersand nozzles It consists of a conventional bell thrust chambersurrounded by an annular thrust chamber Both chambers haveshort primary nozzles which end in a common divergent noz-zle extension Figure 31 shows a typical dual-expander nozzlecon guration At low altitude both thrust chambers operatesharing the same exit area which results in a moderate expan-sion area ratio Part way into the mission one thrust chamberis shut off allowing the other nozzle to use the whole exitarea creating a high-expansion-area ratio for the remainder of

the burn In principle the two operation modes are comparableto those of dual-throat nozzles

Numerical simulations of the ow elds in dual-expandernozzles during all operation modes have been performed atAerojet38 DLR39 and the Keldysh Center Aerojet simulationsare based on an engine design using different propellantshydrocarbonoxygen for the inner chamber and hydrogenox-ygen for the outer chamber whereas DLR simulations arebased on a combined vehicleengine analysis using single-fuelsingle-mixture ratio dual-expander engines with hydrogenoxy-gen Table 4 summarizes combustion chamber parameters Flowphenomena observed in numerical simulations are shown in Fig32a for the mode 1 operation with both thrust chambers burn-ing Compression waves are induced near the inner nozzle lipas a result of the strongly inhomogeneous ow character at thepoint where both exhaust gases of inner and outer combustionchamber merge into the common divergent nozzle part Upand down variations of pressure ratios at the lip did not changethis wave formation even in cases of signi cantly lower exitpressures in the inner nozzle compared to the pressure eld ofthe outer nozzle at the lip Further downstream the compres-sion waves interact with the wall resulting in a re ection ofthe compression waves back into the ow eld

During mode 2 operation a strong expansion of the outernozzle ow is observed when the expansion ratio suddenlyincreases at the end of the nozzle lip The ow is directedtoward the axis of symmetry Near the centerline the ow thenturns over to the axial direction inducing a recompressionshock The static pressure rises signi cantly in this recom-pression region on the centerline A sub- and supersonic recir-culation zone establishes in the inner chamber of the dual-expander nozzle Figure 32b emphasizes the essential owpattern in this operation mode

These analyses have shown that dual-expander nozzles pro-duce high performance in both operation modes3839 Figure 33summarizes performance behavior as a function of ight alti-tude for the dual-expander engine simulated at DLR For com-parison performance data are included for two reference bell

HAGEMANN ET AL 633

Fig 33 Performance data for a dual-expander nozzle

nozzles that have the same area ratios as the dual-expandernozzle during its two operation modes Losses because of thenonhomogeneous exit ow and the induced shocks are com-parable with corresponding values of both conventional noz-zles1

Within the design process of dual-expander engines it is nec-essary to take into account that the nonburning chamber isexposed to high heat uxes and pressure oscillations duringmode 2 operation which might reach a high level and lead tostructural disintegration of the inner nozzle structure There-fore a mode-2 operation with bleed gas generation at a ratherlow thrust chamber pressure level in the inner chamber seemsto be advantageous Because of the bleed gas generation in theprimary chamber the overall mass ow rate during this alter-native mode-2 operation is increased In addition the reductionof the effective exit area ratio results in minor impulse deg-radations during this operation mode when compared with themode-2 operation with no bleed gas generation

Several analytical works on SSTO- and two-stage-to-orbit(TSTO)-vehicles using hydrogenpropane or hydrogenmeth-ane as fuels revealed the lowest vehicle dry masses for dual-mode engines in comparison to other engines1439 Other dual-mode engines using hydrogen as the single fuel but using twomixture ratios also revealed some bene ts over conventionalengines for SSTO- and TSTO applications Even a single-fueloperation with constant mixture ratios in both combustionchambers indicated a gain in launcher performance40

V ConclusionsSeveral nozzle concepts that promise gains in performance

over conventional nozzles were discussed in this paper in-cluding performance enhancements achieved by slight modi- cations of existing nozzles eg through cool gas injectioninto the supersonic nozzle part It is shown that signi cantperformance gains result from the adaptation of the exhaust ow to the ambient pressure and special emphasis was givento altitude adaptive nozzle concepts

A number of nozzle concepts with altitude-compensating ca-pability were identi ed and described To assist the selectionof the best nozzle concept for launch vehicle applications theperformance of the nozzles must be characterized This can bedone using computational uid dynamics (CFD) andor cold- ow tests Existing CFD methods that are in use in the aero-space industry and at research institutes have been veri ed fora wide number of sub- and full-scale experiments and providesuf ciently reliable performance determination for the differ-ent nozzle types

Theoretical evaluations numerical simulations and test re-sults showed that the different concepts have real altitude-com-pensating capabilities However the compensation capabilitiesare limited and there are some drawbacks associated with eachof the concepts Additional performance losses are induced inpractically all of these nozzle concepts when compared to anideal expansion mainly because of the nonisentropic effects

such as shock waves and pressure losses in recirculation zonesHowever these additional performance losses are less than1 ndash 3 depending on the different nozzle concepts

Four nozzle concepts have been selected for further evalu-ation under a NASA Aerojet cooperative agreement They in-clude a plug nozzle an E ndash D nozzle a dual-bell nozzle and adual-expander nozzle The nozzles in addition to a referencebell nozzle have been designed by Aerojet for cold ow testsCFD calculations performed at Aerojet demonstrated that allof these nozzles have high and approximately equal perfor-mances at the design points

In Europe special attention is being paid to plug- and dual-bell nozzles Different plug nozzle concepts are being testedin cold- ow tests within the ESA ARPT (clustered plug noz-zle) and ESA FESTIP (linear plug nozzle) contracts In addi-tion to experiments three-dimensional numerical simulationson these nozzle concepts are being performed Additional ex-periments on different nozzle concepts with forced- ow sep-aration devices eg dual-bell nozzles are currently plannedat DLR

Different advanced nozzle concepts have been examined andare under examination now at the Keldysh Center with bothCFD calculations and experiments on the Nozzle DifferentialFacility which has a nozzle performance determination errorof about 005 This high accuracy is achieved by simulta-neously running the test nozzle and a reference nozzle bothmounted on a common longitudinal axis The performancecharacteristics of the reference nozzle are well known and thethrust difference is measured with strain gauges This facilityis equipped with a wind tunnel to determine nozzle thrust per-formance in the presence of external transonic ows withMach numbers up to 12

In addition to aerodynamic performance other technical is-sues (weight cost design thermal management manufactur-ing system performance and reliability) must be addressedFurthermore before a nal decision can be made as to whichnozzle concept offers the greatest bene ts with regard to aneffective payload mass injection combined launcher and tra-jectory calculations must be performed and compared to a ref-erence launcher concept with conventional nozzles Differentnozzle ef ciencies which account for the additional losses ofadvanced rocket nozzles and are extracted from numerical sim-ulations and experiments must be taken into account

At least for some of these nozzle concepts the plug nozzlethe dual-bell nozzle and the dual-expander nozzle bene tswith regard to lower overall launcher masses have been dem-onstrated in the literature Furthermore the plug nozzle con-cept will be the rst in ight of all of these advanced nozzleconcepts with the X-33 demonstrator vehicle

AcknowledgmentPresented as Paper 18 at the 3rd International Symposium

on Space Propulsion Beijing Peoplersquos Republic of ChinaAug 11ndash 13 1997

References1Manski D and Hagemann G lsquolsquoIn uence of Rocket Design Pa-

rameters on Engine Nozzle Ef cienciesrsquorsquo Journal of Propulsion andPower Vol 12 No 1 1996 pp 41ndash 47

2lsquolsquoLiquid Rocket Engine Nozzlesrsquorsquo NASA Space Vehicle DesignCriteria NASA SP-8120 1976

3Dumnov G E Nikulin G Z and Ponomaryov N B lsquolsquoInves-tigation of Advanced Nozzles for Rocket Enginesrsquorsquo Space Rocket En-gines and Power Plants Vol 4 No 142 NIITP 1993 pp 1012 ndash

1018 (in Russian)4Nguyen T V and Pieper J L lsquolsquoNozzle Flow Separationrsquorsquo Pro-

ceedings of the 5th International Symposium of Propulsion in SpaceTransportation (Paris France) 1996

5Manski D lsquolsquoClustered Plug Nozzles for Future European Reus-able Rocket Launchersrsquorsquo German Aerospace Research Establishment643-817 Lampoldshausen Germany 1981 (in German)

6Horn M and Fisher S lsquolsquoDual-Bell Altitude Compensating Noz-

HAGEMANN ET AL 629

Table 3 Design parameters for linear and toroidal subscale plug nozzlesa

Design data Linear plug (FESTIP) Circular clustered plug (ARPT)

Chamber pressure pc 875 bar 80 barDriving gas Air AirMass ow rate mCcedil 595 kgs 546 kgsDesign pressure ratio pc pamb 200 200Geometric area ratio laquo 1265 1265aUsed in the ESA FESTIP and ESA ARPT studies

Fig 20 Plug nozzle ow eld experiment vs numerical simula-tion a) toroidal plug nozzle experiment with gas-oilnitric-acidpropellant combination side view photograph and b) computa-tional results of toroidal plug nozzle Mach number distributionfull gray and isolines centerline section

Fig 21 Design of linear subscale plug nozzle of the ESA FESTIPTechnology Program (plug size 192 mm high 214 mm wide)

Fig 22 Calculated wall pressure data for linear subscale plugnozzle at different pressure ratios

The change of wake ow behavior may be strongly in uencedby the penetration of ambient pressures through both end sidesparticularly for truncated plug nozzles End plates as foreseenfor the linear plug nozzle of the X-33 demonstrator vehiclecould be used to avoid this ambient pressure penetration

Within the framework of the ESA FESTIP Technology Pro-gram lsquolsquoTechnology Developments on Rocket and Air Breath-ing Propulsion for Reusable Launch Vehiclesrsquorsquo subscale cold-gas tests and numerical simulations of linear plug nozzles arebeing performed by Dasa at the high-speed wind tunnel [Hoch-geschwindigheitswind kanal (HWK) in Bad Salzungen Ger-many] of the Technical University of Dresden719 Design pa-

rameters for the linear plug model are included in Table 3Figure 21 shows the baseline model with a linear throat Afterpassing the linear throat the ow expands directly onto theplug surface A second model will be tested with a bell-shapedlinear nozzle extension following the linear throat for internalexpansion prior to further external expansion onto the plugsurface19 The contour of the plug was designed by the methodof Angelino29 and three different plug lengths as 5 20 and40 of the ideal length were tested Figure 22 shows calcu-lated wall pressures for the full-length plug for different am-bient pressures simulating the operational range of sea-leveloperation up to higher-altitude operation The ow structure inform of the left- and right-running characteristics at two pres-sure ratios is shown in Fig 23 For lower pressure ratios thanthe design pressure ratio of pc pamb = 200 the in uence of the

630 HAGEMANN ET AL

Fig 23 Left- and right-running characteristics in ow eld at twopressure ratios pc pamb = a) 8 and b) 200 (inviscid analysis withmethod of characteristics)

Fig 24 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 20 plug body

Fig 25 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 40 plug body

Fig 27 Flow phenomena of an E ndash D nozzle with a) open and b)closed wakes

Fig 26 Sketch of an E ndash D nozzle

system of recompression shocks and expansion waves on thewall pressure distribution (Figs 17 and 22) can be seen bythe waviness of the pressure distribution on the plug surfaceFirst experimental results are shown in Figs 24 and 25 whichalso show the waviness of the pressure distribution for low-pressure ratios

3 Plug Nozzles Final Aspects

The performance of circular and linear plug nozzles is ad-versely affected by the external airstream ow ie the aspi-ration effect of the external ow slightly reduces performanceand favors an earlier change in wake- ow development for

truncated plug nozzles32ndash 35 Cold- ow tests showed that theeffect of external ow on performance is con ned to a narrowrange of external ow speeds with Mach numbers near unity

The altitude compensation capability of circular and linearplug nozzles for higher ambient pressures is indisputable Be-cause plug nozzles lose this capability for pressure ratios abovethe design pressure ratio the latter should be chosen as highas possible Taking this into account plug nozzles will featurean even better overall performance than those shown in Figs18 and 19 Unfortunately for equal geometrical area ratiosplug nozzles perform worse at high altitude than do conven-tional bell nozzles because of truncation and clustering Forhigh-area-ratio nozzles with relatively short lengths plug noz-zles perform better than conventional bell nozzles In additionto having excellent capabilities for altitude compensation plugnozzles have additional advantages including ease in vehicleand engine integration

C E ndash D Nozzles

An expansionndash de ection (E ndash D) nozzle is shown in Fig 26E ndash D nozzles were at one time thought to have capabilities foraltitude compensation because the gas expansion takes placewith a constant pressure free boundary Thus the aerodynamicbehavior of E ndash D nozzles as a function of altitude is in prin-ciple quite similar to plug nozzles because the expansion pro-cess is controlled by the ambient pressure hence by the alti-tude In contrast to plug nozzles however the expansionprocess is controlled from inside the nozzle for E ndash D nozzlesAt low altitude the higher ambient pressure limits the gasexpansion resulting in a low effective expansion area ratioThe exhaust gas is adapted to the ambient pressure level by

HAGEMANN ET AL 631

Fig 30 Performance losses in dual-throat nozzles as a functionof nozzle design parameters (with r4r1 = 352 r2r1 = 116 and mCcedil 2

= 00)

Fig 29 Flow phenomena during a) sea-level and b) high-altitudeoperations

Fig 28 Sketch of a dual-throat nozzle view of combustion cham-ber and throat region

systems of recompression and expansion waves (Fig 27) Athigher altitude the lower ambient pressure allows more gasexpansion within the nozzle resulting in a higher effectiveexpansion area ratio But in contrast to plug nozzles the pres-sure in the wake of the center plug is always less than theambient pressure because of the aspiration effect This occursat low pressure ratios when the wake is opened and results inan aspiration loss Furthermore because the exhaust ow ex-pands to this base pressure rather than to the ambient pressurelevel wall pressures downstream are overexpanded This re-sults in an additional overexpansion loss As the pressure ratioincreases the wake region closes and is thus totally isolatedfrom the ambient environment (see Fig 27) The behavior dur-ing transition from open wake to closed wake is again equalto plug nozzles and the base pressure in the closed-wake re-gion is essentially independent of the ambient pressure

The E ndash D nozzle concept has also been a subject of numer-ous analytical and experimental studies Results from thesestudies show that E ndash D nozzle capabilities for altitude com-pensation are poor and are in fact worse than those of plugnozzles because of aspiration and overexpansion losses35 Forhigh-area-ratio nozzles with a relatively short length an E ndash Dnozzle performs better than a comparable conventional bellnozzle at the same length because of lower divergence andpro le losses than the bell nozzle

The advantages of the E ndash D nozzle concept include its smallengine envelope and no moving parts However like toroidalplug nozzles E ndash D nozzles have the disadvantage of havinghigher throat heat uxes relative to a conventional bell nozzlewith an equal throat area The higher throat heat ux resultsagain from the relatively thin annular throat gap This problemhowever can be remedied with the modular thrust cell clusterconcept Additional advantages of E ndash D nozzles with clusteredthrust cells are the same as those already discussed for theclustered plug nozzle concept with regard to ease in manu-facturing thrust vector control by throttling or shutting off anindividual or group of thrust cells and lower nozzle throatheating

D Nozzles with Throat Area Varied by a Mechanical Pintle

This nozzle concept utilizes a conventional bell nozzle witha xed exit area and a mechanical pintle in the combustionchamber and throat region to vary the throat area and hencethe expansion area ratio The area of the nozzle throatmdashanannulus between the pintle and the shroudmdashis varied by mov-ing the pintle axially

The pintle concept has been used in solid rocket motors asa mean to provide variable thrust The concept in principleallows a continuous variation of the throat area and thus op-timum expansion area ratios throughout a mission Howeverit requires an actuator and a sophisticated control system Theconcept raises issues of engine weight design complexitycooling of the pintle and nozzle throat and reliability

In Ref 36 the aerodynamic performances of nozzles with ve different pintle geometries were calculated and comparedwith a reference bell nozzle The performance losses of thesepintle nozzles when compared with the bell nozzle are in therange of 1ndash 25 Performances of a xed pintle geometry atthree different locations were also calculated The results showthat the performance loss varies with the pintle location

E Dual-Mode Nozzles

Dual-mode rocket engines using one or two fuels offer atrajectory-adapted dual-mode operation during the ascent of alauncher which may be of signi cant advantage for single-stage Earth-to-orbit vehicles as compared to conventionalrocket engines with bell-type nozzles This engine concept in-volves the use of a dense propellant combination with mod-erate performance during liftoff to provide high thrust duringthe initial ight phase and a better performing propellant com-bination in vacuum which result in higher speci c impulse

The fuels are burned in two different combustion chamberswith one located completely inside the other in the case ofengines with dual-throat nozzles or with a conventional bellthrust chamber surrounded by an annular thrust chamber in thecase of dual-expander engines This type of engine has a built-in accelerationndash reduction capability achieved by shuttingdown one of two thrust chambers The total engine thrust isthen provided by the remaining thrust chamber with the useof the total nozzle exit area leading to an increase in speci cimpulse Apart from the indicated bene ts of dual-mode en-gines which will be discussed in more detail later their de-

632 HAGEMANN ET AL

Table 4 Thrust chamber parameters for simulated dual-expander nozzlesa

Operation mode

Aerojet

First (inout) Second (inout)

DLR

First (inout) Second (inout)

Propellants C3H8ndash O2H2ndash O2 mdashH2ndash O2 H2ndash O2H2ndash O2 mdashH2ndash O2

Chamber pressure pc 414207 bar mdash207 bar 200200 bar mdash200 barMixture ratio r 337 mdash7 77 mdash7Exit area ratio laquo 69 146 58 116aInnerouter chambers

Fig 32 Flow phenomena in a dual-expander nozzle during a)sea-level- and b) high-altitude operation

Fig 31 Sketch of a dual-expander nozzle view of combustionchamber and throat region

velopment and construction require considerable technologicaleffort

1 Dual-Throat Nozzles

A dual-throat nozzle con guration is shown in Fig 28 Itconsists of two conventional bell-thrust chambers with onelocated completely inside the other At low altitude the outerthrust chamber operates with the inner thrust chamber runningin parallel In this operation mode the engine has a larger throatproviding a moderate expansion area ratio During the missionthe outer thrust chamber is shut off and operation continueswith only the inner engine In this con guration ow fromthe inner engine expands and attaches supersonically to theouter engine resulting in a higher expansion area ratio for theremainder of the burn Flow phenomena in both operationmodes are included in Fig 29

Hot- red tests at Aerojet were conducted to provide heattransfer data that were very useful for the thermal analysis anddesign of the dual-throat nozzle con guration37 These testsshowed that ow separation occurred in the inner engine noz-zle at higher ratios of outer to inner chamber pressures duringthe rst operation mode with both chambers burning in par-allel The ow separation resulted in a higher heat load to theinner nozzle Subscale tests performed at the Keldysh Center3

have shown that the additional loss caused by the nozzle con-tour discontinuity during vacuum operation with active innerchamber is in the range of 08ndash 4 depending on geometricaldata (see Fig 30) This high-performance loss results from theinteraction of the inner chamber jet with the outer chambernozzle wall3 The decrease of the jet incidence angle on thewall by means of gas injection through the outer chamber re-duces this performance loss by 04ndash 073

2 Dual-Expander Nozzles

A dual-expander nozzle has two concentric thrust chambersand nozzles It consists of a conventional bell thrust chambersurrounded by an annular thrust chamber Both chambers haveshort primary nozzles which end in a common divergent noz-zle extension Figure 31 shows a typical dual-expander nozzlecon guration At low altitude both thrust chambers operatesharing the same exit area which results in a moderate expan-sion area ratio Part way into the mission one thrust chamberis shut off allowing the other nozzle to use the whole exitarea creating a high-expansion-area ratio for the remainder of

the burn In principle the two operation modes are comparableto those of dual-throat nozzles

Numerical simulations of the ow elds in dual-expandernozzles during all operation modes have been performed atAerojet38 DLR39 and the Keldysh Center Aerojet simulationsare based on an engine design using different propellantshydrocarbonoxygen for the inner chamber and hydrogenox-ygen for the outer chamber whereas DLR simulations arebased on a combined vehicleengine analysis using single-fuelsingle-mixture ratio dual-expander engines with hydrogenoxy-gen Table 4 summarizes combustion chamber parameters Flowphenomena observed in numerical simulations are shown in Fig32a for the mode 1 operation with both thrust chambers burn-ing Compression waves are induced near the inner nozzle lipas a result of the strongly inhomogeneous ow character at thepoint where both exhaust gases of inner and outer combustionchamber merge into the common divergent nozzle part Upand down variations of pressure ratios at the lip did not changethis wave formation even in cases of signi cantly lower exitpressures in the inner nozzle compared to the pressure eld ofthe outer nozzle at the lip Further downstream the compres-sion waves interact with the wall resulting in a re ection ofthe compression waves back into the ow eld

During mode 2 operation a strong expansion of the outernozzle ow is observed when the expansion ratio suddenlyincreases at the end of the nozzle lip The ow is directedtoward the axis of symmetry Near the centerline the ow thenturns over to the axial direction inducing a recompressionshock The static pressure rises signi cantly in this recom-pression region on the centerline A sub- and supersonic recir-culation zone establishes in the inner chamber of the dual-expander nozzle Figure 32b emphasizes the essential owpattern in this operation mode

These analyses have shown that dual-expander nozzles pro-duce high performance in both operation modes3839 Figure 33summarizes performance behavior as a function of ight alti-tude for the dual-expander engine simulated at DLR For com-parison performance data are included for two reference bell

HAGEMANN ET AL 633

Fig 33 Performance data for a dual-expander nozzle

nozzles that have the same area ratios as the dual-expandernozzle during its two operation modes Losses because of thenonhomogeneous exit ow and the induced shocks are com-parable with corresponding values of both conventional noz-zles1

Within the design process of dual-expander engines it is nec-essary to take into account that the nonburning chamber isexposed to high heat uxes and pressure oscillations duringmode 2 operation which might reach a high level and lead tostructural disintegration of the inner nozzle structure There-fore a mode-2 operation with bleed gas generation at a ratherlow thrust chamber pressure level in the inner chamber seemsto be advantageous Because of the bleed gas generation in theprimary chamber the overall mass ow rate during this alter-native mode-2 operation is increased In addition the reductionof the effective exit area ratio results in minor impulse deg-radations during this operation mode when compared with themode-2 operation with no bleed gas generation

Several analytical works on SSTO- and two-stage-to-orbit(TSTO)-vehicles using hydrogenpropane or hydrogenmeth-ane as fuels revealed the lowest vehicle dry masses for dual-mode engines in comparison to other engines1439 Other dual-mode engines using hydrogen as the single fuel but using twomixture ratios also revealed some bene ts over conventionalengines for SSTO- and TSTO applications Even a single-fueloperation with constant mixture ratios in both combustionchambers indicated a gain in launcher performance40

V ConclusionsSeveral nozzle concepts that promise gains in performance

over conventional nozzles were discussed in this paper in-cluding performance enhancements achieved by slight modi- cations of existing nozzles eg through cool gas injectioninto the supersonic nozzle part It is shown that signi cantperformance gains result from the adaptation of the exhaust ow to the ambient pressure and special emphasis was givento altitude adaptive nozzle concepts

A number of nozzle concepts with altitude-compensating ca-pability were identi ed and described To assist the selectionof the best nozzle concept for launch vehicle applications theperformance of the nozzles must be characterized This can bedone using computational uid dynamics (CFD) andor cold- ow tests Existing CFD methods that are in use in the aero-space industry and at research institutes have been veri ed fora wide number of sub- and full-scale experiments and providesuf ciently reliable performance determination for the differ-ent nozzle types

Theoretical evaluations numerical simulations and test re-sults showed that the different concepts have real altitude-com-pensating capabilities However the compensation capabilitiesare limited and there are some drawbacks associated with eachof the concepts Additional performance losses are induced inpractically all of these nozzle concepts when compared to anideal expansion mainly because of the nonisentropic effects

such as shock waves and pressure losses in recirculation zonesHowever these additional performance losses are less than1 ndash 3 depending on the different nozzle concepts

Four nozzle concepts have been selected for further evalu-ation under a NASA Aerojet cooperative agreement They in-clude a plug nozzle an E ndash D nozzle a dual-bell nozzle and adual-expander nozzle The nozzles in addition to a referencebell nozzle have been designed by Aerojet for cold ow testsCFD calculations performed at Aerojet demonstrated that allof these nozzles have high and approximately equal perfor-mances at the design points

In Europe special attention is being paid to plug- and dual-bell nozzles Different plug nozzle concepts are being testedin cold- ow tests within the ESA ARPT (clustered plug noz-zle) and ESA FESTIP (linear plug nozzle) contracts In addi-tion to experiments three-dimensional numerical simulationson these nozzle concepts are being performed Additional ex-periments on different nozzle concepts with forced- ow sep-aration devices eg dual-bell nozzles are currently plannedat DLR

Different advanced nozzle concepts have been examined andare under examination now at the Keldysh Center with bothCFD calculations and experiments on the Nozzle DifferentialFacility which has a nozzle performance determination errorof about 005 This high accuracy is achieved by simulta-neously running the test nozzle and a reference nozzle bothmounted on a common longitudinal axis The performancecharacteristics of the reference nozzle are well known and thethrust difference is measured with strain gauges This facilityis equipped with a wind tunnel to determine nozzle thrust per-formance in the presence of external transonic ows withMach numbers up to 12

In addition to aerodynamic performance other technical is-sues (weight cost design thermal management manufactur-ing system performance and reliability) must be addressedFurthermore before a nal decision can be made as to whichnozzle concept offers the greatest bene ts with regard to aneffective payload mass injection combined launcher and tra-jectory calculations must be performed and compared to a ref-erence launcher concept with conventional nozzles Differentnozzle ef ciencies which account for the additional losses ofadvanced rocket nozzles and are extracted from numerical sim-ulations and experiments must be taken into account

At least for some of these nozzle concepts the plug nozzlethe dual-bell nozzle and the dual-expander nozzle bene tswith regard to lower overall launcher masses have been dem-onstrated in the literature Furthermore the plug nozzle con-cept will be the rst in ight of all of these advanced nozzleconcepts with the X-33 demonstrator vehicle

AcknowledgmentPresented as Paper 18 at the 3rd International Symposium

on Space Propulsion Beijing Peoplersquos Republic of ChinaAug 11ndash 13 1997

References1Manski D and Hagemann G lsquolsquoIn uence of Rocket Design Pa-

rameters on Engine Nozzle Ef cienciesrsquorsquo Journal of Propulsion andPower Vol 12 No 1 1996 pp 41ndash 47

2lsquolsquoLiquid Rocket Engine Nozzlesrsquorsquo NASA Space Vehicle DesignCriteria NASA SP-8120 1976

3Dumnov G E Nikulin G Z and Ponomaryov N B lsquolsquoInves-tigation of Advanced Nozzles for Rocket Enginesrsquorsquo Space Rocket En-gines and Power Plants Vol 4 No 142 NIITP 1993 pp 1012 ndash

1018 (in Russian)4Nguyen T V and Pieper J L lsquolsquoNozzle Flow Separationrsquorsquo Pro-

ceedings of the 5th International Symposium of Propulsion in SpaceTransportation (Paris France) 1996

5Manski D lsquolsquoClustered Plug Nozzles for Future European Reus-able Rocket Launchersrsquorsquo German Aerospace Research Establishment643-817 Lampoldshausen Germany 1981 (in German)

6Horn M and Fisher S lsquolsquoDual-Bell Altitude Compensating Noz-

630 HAGEMANN ET AL

Fig 23 Left- and right-running characteristics in ow eld at twopressure ratios pc pamb = a) 8 and b) 200 (inviscid analysis withmethod of characteristics)

Fig 24 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 20 plug body

Fig 25 Measured wall pressure data of linear subscale plug noz-zle at different pressure ratios and truncations 40 plug body

Fig 27 Flow phenomena of an E ndash D nozzle with a) open and b)closed wakes

Fig 26 Sketch of an E ndash D nozzle

system of recompression shocks and expansion waves on thewall pressure distribution (Figs 17 and 22) can be seen bythe waviness of the pressure distribution on the plug surfaceFirst experimental results are shown in Figs 24 and 25 whichalso show the waviness of the pressure distribution for low-pressure ratios

3 Plug Nozzles Final Aspects

The performance of circular and linear plug nozzles is ad-versely affected by the external airstream ow ie the aspi-ration effect of the external ow slightly reduces performanceand favors an earlier change in wake- ow development for

truncated plug nozzles32ndash 35 Cold- ow tests showed that theeffect of external ow on performance is con ned to a narrowrange of external ow speeds with Mach numbers near unity

The altitude compensation capability of circular and linearplug nozzles for higher ambient pressures is indisputable Be-cause plug nozzles lose this capability for pressure ratios abovethe design pressure ratio the latter should be chosen as highas possible Taking this into account plug nozzles will featurean even better overall performance than those shown in Figs18 and 19 Unfortunately for equal geometrical area ratiosplug nozzles perform worse at high altitude than do conven-tional bell nozzles because of truncation and clustering Forhigh-area-ratio nozzles with relatively short lengths plug noz-zles perform better than conventional bell nozzles In additionto having excellent capabilities for altitude compensation plugnozzles have additional advantages including ease in vehicleand engine integration

C E ndash D Nozzles

An expansionndash de ection (E ndash D) nozzle is shown in Fig 26E ndash D nozzles were at one time thought to have capabilities foraltitude compensation because the gas expansion takes placewith a constant pressure free boundary Thus the aerodynamicbehavior of E ndash D nozzles as a function of altitude is in prin-ciple quite similar to plug nozzles because the expansion pro-cess is controlled by the ambient pressure hence by the alti-tude In contrast to plug nozzles however the expansionprocess is controlled from inside the nozzle for E ndash D nozzlesAt low altitude the higher ambient pressure limits the gasexpansion resulting in a low effective expansion area ratioThe exhaust gas is adapted to the ambient pressure level by

HAGEMANN ET AL 631

Fig 30 Performance losses in dual-throat nozzles as a functionof nozzle design parameters (with r4r1 = 352 r2r1 = 116 and mCcedil 2

= 00)

Fig 29 Flow phenomena during a) sea-level and b) high-altitudeoperations

Fig 28 Sketch of a dual-throat nozzle view of combustion cham-ber and throat region

systems of recompression and expansion waves (Fig 27) Athigher altitude the lower ambient pressure allows more gasexpansion within the nozzle resulting in a higher effectiveexpansion area ratio But in contrast to plug nozzles the pres-sure in the wake of the center plug is always less than theambient pressure because of the aspiration effect This occursat low pressure ratios when the wake is opened and results inan aspiration loss Furthermore because the exhaust ow ex-pands to this base pressure rather than to the ambient pressurelevel wall pressures downstream are overexpanded This re-sults in an additional overexpansion loss As the pressure ratioincreases the wake region closes and is thus totally isolatedfrom the ambient environment (see Fig 27) The behavior dur-ing transition from open wake to closed wake is again equalto plug nozzles and the base pressure in the closed-wake re-gion is essentially independent of the ambient pressure

The E ndash D nozzle concept has also been a subject of numer-ous analytical and experimental studies Results from thesestudies show that E ndash D nozzle capabilities for altitude com-pensation are poor and are in fact worse than those of plugnozzles because of aspiration and overexpansion losses35 Forhigh-area-ratio nozzles with a relatively short length an E ndash Dnozzle performs better than a comparable conventional bellnozzle at the same length because of lower divergence andpro le losses than the bell nozzle

The advantages of the E ndash D nozzle concept include its smallengine envelope and no moving parts However like toroidalplug nozzles E ndash D nozzles have the disadvantage of havinghigher throat heat uxes relative to a conventional bell nozzlewith an equal throat area The higher throat heat ux resultsagain from the relatively thin annular throat gap This problemhowever can be remedied with the modular thrust cell clusterconcept Additional advantages of E ndash D nozzles with clusteredthrust cells are the same as those already discussed for theclustered plug nozzle concept with regard to ease in manu-facturing thrust vector control by throttling or shutting off anindividual or group of thrust cells and lower nozzle throatheating

D Nozzles with Throat Area Varied by a Mechanical Pintle

This nozzle concept utilizes a conventional bell nozzle witha xed exit area and a mechanical pintle in the combustionchamber and throat region to vary the throat area and hencethe expansion area ratio The area of the nozzle throatmdashanannulus between the pintle and the shroudmdashis varied by mov-ing the pintle axially

The pintle concept has been used in solid rocket motors asa mean to provide variable thrust The concept in principleallows a continuous variation of the throat area and thus op-timum expansion area ratios throughout a mission Howeverit requires an actuator and a sophisticated control system Theconcept raises issues of engine weight design complexitycooling of the pintle and nozzle throat and reliability

In Ref 36 the aerodynamic performances of nozzles with ve different pintle geometries were calculated and comparedwith a reference bell nozzle The performance losses of thesepintle nozzles when compared with the bell nozzle are in therange of 1ndash 25 Performances of a xed pintle geometry atthree different locations were also calculated The results showthat the performance loss varies with the pintle location

E Dual-Mode Nozzles

Dual-mode rocket engines using one or two fuels offer atrajectory-adapted dual-mode operation during the ascent of alauncher which may be of signi cant advantage for single-stage Earth-to-orbit vehicles as compared to conventionalrocket engines with bell-type nozzles This engine concept in-volves the use of a dense propellant combination with mod-erate performance during liftoff to provide high thrust duringthe initial ight phase and a better performing propellant com-bination in vacuum which result in higher speci c impulse

The fuels are burned in two different combustion chamberswith one located completely inside the other in the case ofengines with dual-throat nozzles or with a conventional bellthrust chamber surrounded by an annular thrust chamber in thecase of dual-expander engines This type of engine has a built-in accelerationndash reduction capability achieved by shuttingdown one of two thrust chambers The total engine thrust isthen provided by the remaining thrust chamber with the useof the total nozzle exit area leading to an increase in speci cimpulse Apart from the indicated bene ts of dual-mode en-gines which will be discussed in more detail later their de-

632 HAGEMANN ET AL

Table 4 Thrust chamber parameters for simulated dual-expander nozzlesa

Operation mode

Aerojet

First (inout) Second (inout)

DLR

First (inout) Second (inout)

Propellants C3H8ndash O2H2ndash O2 mdashH2ndash O2 H2ndash O2H2ndash O2 mdashH2ndash O2

Chamber pressure pc 414207 bar mdash207 bar 200200 bar mdash200 barMixture ratio r 337 mdash7 77 mdash7Exit area ratio laquo 69 146 58 116aInnerouter chambers

Fig 32 Flow phenomena in a dual-expander nozzle during a)sea-level- and b) high-altitude operation

Fig 31 Sketch of a dual-expander nozzle view of combustionchamber and throat region

velopment and construction require considerable technologicaleffort

1 Dual-Throat Nozzles

A dual-throat nozzle con guration is shown in Fig 28 Itconsists of two conventional bell-thrust chambers with onelocated completely inside the other At low altitude the outerthrust chamber operates with the inner thrust chamber runningin parallel In this operation mode the engine has a larger throatproviding a moderate expansion area ratio During the missionthe outer thrust chamber is shut off and operation continueswith only the inner engine In this con guration ow fromthe inner engine expands and attaches supersonically to theouter engine resulting in a higher expansion area ratio for theremainder of the burn Flow phenomena in both operationmodes are included in Fig 29

Hot- red tests at Aerojet were conducted to provide heattransfer data that were very useful for the thermal analysis anddesign of the dual-throat nozzle con guration37 These testsshowed that ow separation occurred in the inner engine noz-zle at higher ratios of outer to inner chamber pressures duringthe rst operation mode with both chambers burning in par-allel The ow separation resulted in a higher heat load to theinner nozzle Subscale tests performed at the Keldysh Center3

have shown that the additional loss caused by the nozzle con-tour discontinuity during vacuum operation with active innerchamber is in the range of 08ndash 4 depending on geometricaldata (see Fig 30) This high-performance loss results from theinteraction of the inner chamber jet with the outer chambernozzle wall3 The decrease of the jet incidence angle on thewall by means of gas injection through the outer chamber re-duces this performance loss by 04ndash 073

2 Dual-Expander Nozzles

A dual-expander nozzle has two concentric thrust chambersand nozzles It consists of a conventional bell thrust chambersurrounded by an annular thrust chamber Both chambers haveshort primary nozzles which end in a common divergent noz-zle extension Figure 31 shows a typical dual-expander nozzlecon guration At low altitude both thrust chambers operatesharing the same exit area which results in a moderate expan-sion area ratio Part way into the mission one thrust chamberis shut off allowing the other nozzle to use the whole exitarea creating a high-expansion-area ratio for the remainder of

the burn In principle the two operation modes are comparableto those of dual-throat nozzles

Numerical simulations of the ow elds in dual-expandernozzles during all operation modes have been performed atAerojet38 DLR39 and the Keldysh Center Aerojet simulationsare based on an engine design using different propellantshydrocarbonoxygen for the inner chamber and hydrogenox-ygen for the outer chamber whereas DLR simulations arebased on a combined vehicleengine analysis using single-fuelsingle-mixture ratio dual-expander engines with hydrogenoxy-gen Table 4 summarizes combustion chamber parameters Flowphenomena observed in numerical simulations are shown in Fig32a for the mode 1 operation with both thrust chambers burn-ing Compression waves are induced near the inner nozzle lipas a result of the strongly inhomogeneous ow character at thepoint where both exhaust gases of inner and outer combustionchamber merge into the common divergent nozzle part Upand down variations of pressure ratios at the lip did not changethis wave formation even in cases of signi cantly lower exitpressures in the inner nozzle compared to the pressure eld ofthe outer nozzle at the lip Further downstream the compres-sion waves interact with the wall resulting in a re ection ofthe compression waves back into the ow eld

During mode 2 operation a strong expansion of the outernozzle ow is observed when the expansion ratio suddenlyincreases at the end of the nozzle lip The ow is directedtoward the axis of symmetry Near the centerline the ow thenturns over to the axial direction inducing a recompressionshock The static pressure rises signi cantly in this recom-pression region on the centerline A sub- and supersonic recir-culation zone establishes in the inner chamber of the dual-expander nozzle Figure 32b emphasizes the essential owpattern in this operation mode

These analyses have shown that dual-expander nozzles pro-duce high performance in both operation modes3839 Figure 33summarizes performance behavior as a function of ight alti-tude for the dual-expander engine simulated at DLR For com-parison performance data are included for two reference bell

HAGEMANN ET AL 633

Fig 33 Performance data for a dual-expander nozzle

nozzles that have the same area ratios as the dual-expandernozzle during its two operation modes Losses because of thenonhomogeneous exit ow and the induced shocks are com-parable with corresponding values of both conventional noz-zles1

Within the design process of dual-expander engines it is nec-essary to take into account that the nonburning chamber isexposed to high heat uxes and pressure oscillations duringmode 2 operation which might reach a high level and lead tostructural disintegration of the inner nozzle structure There-fore a mode-2 operation with bleed gas generation at a ratherlow thrust chamber pressure level in the inner chamber seemsto be advantageous Because of the bleed gas generation in theprimary chamber the overall mass ow rate during this alter-native mode-2 operation is increased In addition the reductionof the effective exit area ratio results in minor impulse deg-radations during this operation mode when compared with themode-2 operation with no bleed gas generation

Several analytical works on SSTO- and two-stage-to-orbit(TSTO)-vehicles using hydrogenpropane or hydrogenmeth-ane as fuels revealed the lowest vehicle dry masses for dual-mode engines in comparison to other engines1439 Other dual-mode engines using hydrogen as the single fuel but using twomixture ratios also revealed some bene ts over conventionalengines for SSTO- and TSTO applications Even a single-fueloperation with constant mixture ratios in both combustionchambers indicated a gain in launcher performance40

V ConclusionsSeveral nozzle concepts that promise gains in performance

over conventional nozzles were discussed in this paper in-cluding performance enhancements achieved by slight modi- cations of existing nozzles eg through cool gas injectioninto the supersonic nozzle part It is shown that signi cantperformance gains result from the adaptation of the exhaust ow to the ambient pressure and special emphasis was givento altitude adaptive nozzle concepts

A number of nozzle concepts with altitude-compensating ca-pability were identi ed and described To assist the selectionof the best nozzle concept for launch vehicle applications theperformance of the nozzles must be characterized This can bedone using computational uid dynamics (CFD) andor cold- ow tests Existing CFD methods that are in use in the aero-space industry and at research institutes have been veri ed fora wide number of sub- and full-scale experiments and providesuf ciently reliable performance determination for the differ-ent nozzle types

Theoretical evaluations numerical simulations and test re-sults showed that the different concepts have real altitude-com-pensating capabilities However the compensation capabilitiesare limited and there are some drawbacks associated with eachof the concepts Additional performance losses are induced inpractically all of these nozzle concepts when compared to anideal expansion mainly because of the nonisentropic effects

such as shock waves and pressure losses in recirculation zonesHowever these additional performance losses are less than1 ndash 3 depending on the different nozzle concepts

Four nozzle concepts have been selected for further evalu-ation under a NASA Aerojet cooperative agreement They in-clude a plug nozzle an E ndash D nozzle a dual-bell nozzle and adual-expander nozzle The nozzles in addition to a referencebell nozzle have been designed by Aerojet for cold ow testsCFD calculations performed at Aerojet demonstrated that allof these nozzles have high and approximately equal perfor-mances at the design points

In Europe special attention is being paid to plug- and dual-bell nozzles Different plug nozzle concepts are being testedin cold- ow tests within the ESA ARPT (clustered plug noz-zle) and ESA FESTIP (linear plug nozzle) contracts In addi-tion to experiments three-dimensional numerical simulationson these nozzle concepts are being performed Additional ex-periments on different nozzle concepts with forced- ow sep-aration devices eg dual-bell nozzles are currently plannedat DLR

Different advanced nozzle concepts have been examined andare under examination now at the Keldysh Center with bothCFD calculations and experiments on the Nozzle DifferentialFacility which has a nozzle performance determination errorof about 005 This high accuracy is achieved by simulta-neously running the test nozzle and a reference nozzle bothmounted on a common longitudinal axis The performancecharacteristics of the reference nozzle are well known and thethrust difference is measured with strain gauges This facilityis equipped with a wind tunnel to determine nozzle thrust per-formance in the presence of external transonic ows withMach numbers up to 12

In addition to aerodynamic performance other technical is-sues (weight cost design thermal management manufactur-ing system performance and reliability) must be addressedFurthermore before a nal decision can be made as to whichnozzle concept offers the greatest bene ts with regard to aneffective payload mass injection combined launcher and tra-jectory calculations must be performed and compared to a ref-erence launcher concept with conventional nozzles Differentnozzle ef ciencies which account for the additional losses ofadvanced rocket nozzles and are extracted from numerical sim-ulations and experiments must be taken into account

At least for some of these nozzle concepts the plug nozzlethe dual-bell nozzle and the dual-expander nozzle bene tswith regard to lower overall launcher masses have been dem-onstrated in the literature Furthermore the plug nozzle con-cept will be the rst in ight of all of these advanced nozzleconcepts with the X-33 demonstrator vehicle

AcknowledgmentPresented as Paper 18 at the 3rd International Symposium

on Space Propulsion Beijing Peoplersquos Republic of ChinaAug 11ndash 13 1997

References1Manski D and Hagemann G lsquolsquoIn uence of Rocket Design Pa-

rameters on Engine Nozzle Ef cienciesrsquorsquo Journal of Propulsion andPower Vol 12 No 1 1996 pp 41ndash 47

2lsquolsquoLiquid Rocket Engine Nozzlesrsquorsquo NASA Space Vehicle DesignCriteria NASA SP-8120 1976

3Dumnov G E Nikulin G Z and Ponomaryov N B lsquolsquoInves-tigation of Advanced Nozzles for Rocket Enginesrsquorsquo Space Rocket En-gines and Power Plants Vol 4 No 142 NIITP 1993 pp 1012 ndash

1018 (in Russian)4Nguyen T V and Pieper J L lsquolsquoNozzle Flow Separationrsquorsquo Pro-

ceedings of the 5th International Symposium of Propulsion in SpaceTransportation (Paris France) 1996

5Manski D lsquolsquoClustered Plug Nozzles for Future European Reus-able Rocket Launchersrsquorsquo German Aerospace Research Establishment643-817 Lampoldshausen Germany 1981 (in German)

6Horn M and Fisher S lsquolsquoDual-Bell Altitude Compensating Noz-

HAGEMANN ET AL 631

Fig 30 Performance losses in dual-throat nozzles as a functionof nozzle design parameters (with r4r1 = 352 r2r1 = 116 and mCcedil 2

= 00)

Fig 29 Flow phenomena during a) sea-level and b) high-altitudeoperations

Fig 28 Sketch of a dual-throat nozzle view of combustion cham-ber and throat region

systems of recompression and expansion waves (Fig 27) Athigher altitude the lower ambient pressure allows more gasexpansion within the nozzle resulting in a higher effectiveexpansion area ratio But in contrast to plug nozzles the pres-sure in the wake of the center plug is always less than theambient pressure because of the aspiration effect This occursat low pressure ratios when the wake is opened and results inan aspiration loss Furthermore because the exhaust ow ex-pands to this base pressure rather than to the ambient pressurelevel wall pressures downstream are overexpanded This re-sults in an additional overexpansion loss As the pressure ratioincreases the wake region closes and is thus totally isolatedfrom the ambient environment (see Fig 27) The behavior dur-ing transition from open wake to closed wake is again equalto plug nozzles and the base pressure in the closed-wake re-gion is essentially independent of the ambient pressure

The E ndash D nozzle concept has also been a subject of numer-ous analytical and experimental studies Results from thesestudies show that E ndash D nozzle capabilities for altitude com-pensation are poor and are in fact worse than those of plugnozzles because of aspiration and overexpansion losses35 Forhigh-area-ratio nozzles with a relatively short length an E ndash Dnozzle performs better than a comparable conventional bellnozzle at the same length because of lower divergence andpro le losses than the bell nozzle

The advantages of the E ndash D nozzle concept include its smallengine envelope and no moving parts However like toroidalplug nozzles E ndash D nozzles have the disadvantage of havinghigher throat heat uxes relative to a conventional bell nozzlewith an equal throat area The higher throat heat ux resultsagain from the relatively thin annular throat gap This problemhowever can be remedied with the modular thrust cell clusterconcept Additional advantages of E ndash D nozzles with clusteredthrust cells are the same as those already discussed for theclustered plug nozzle concept with regard to ease in manu-facturing thrust vector control by throttling or shutting off anindividual or group of thrust cells and lower nozzle throatheating

D Nozzles with Throat Area Varied by a Mechanical Pintle

This nozzle concept utilizes a conventional bell nozzle witha xed exit area and a mechanical pintle in the combustionchamber and throat region to vary the throat area and hencethe expansion area ratio The area of the nozzle throatmdashanannulus between the pintle and the shroudmdashis varied by mov-ing the pintle axially

The pintle concept has been used in solid rocket motors asa mean to provide variable thrust The concept in principleallows a continuous variation of the throat area and thus op-timum expansion area ratios throughout a mission Howeverit requires an actuator and a sophisticated control system Theconcept raises issues of engine weight design complexitycooling of the pintle and nozzle throat and reliability

In Ref 36 the aerodynamic performances of nozzles with ve different pintle geometries were calculated and comparedwith a reference bell nozzle The performance losses of thesepintle nozzles when compared with the bell nozzle are in therange of 1ndash 25 Performances of a xed pintle geometry atthree different locations were also calculated The results showthat the performance loss varies with the pintle location

E Dual-Mode Nozzles

Dual-mode rocket engines using one or two fuels offer atrajectory-adapted dual-mode operation during the ascent of alauncher which may be of signi cant advantage for single-stage Earth-to-orbit vehicles as compared to conventionalrocket engines with bell-type nozzles This engine concept in-volves the use of a dense propellant combination with mod-erate performance during liftoff to provide high thrust duringthe initial ight phase and a better performing propellant com-bination in vacuum which result in higher speci c impulse

The fuels are burned in two different combustion chamberswith one located completely inside the other in the case ofengines with dual-throat nozzles or with a conventional bellthrust chamber surrounded by an annular thrust chamber in thecase of dual-expander engines This type of engine has a built-in accelerationndash reduction capability achieved by shuttingdown one of two thrust chambers The total engine thrust isthen provided by the remaining thrust chamber with the useof the total nozzle exit area leading to an increase in speci cimpulse Apart from the indicated bene ts of dual-mode en-gines which will be discussed in more detail later their de-

632 HAGEMANN ET AL

Table 4 Thrust chamber parameters for simulated dual-expander nozzlesa

Operation mode

Aerojet

First (inout) Second (inout)

DLR

First (inout) Second (inout)

Propellants C3H8ndash O2H2ndash O2 mdashH2ndash O2 H2ndash O2H2ndash O2 mdashH2ndash O2

Chamber pressure pc 414207 bar mdash207 bar 200200 bar mdash200 barMixture ratio r 337 mdash7 77 mdash7Exit area ratio laquo 69 146 58 116aInnerouter chambers

Fig 32 Flow phenomena in a dual-expander nozzle during a)sea-level- and b) high-altitude operation

Fig 31 Sketch of a dual-expander nozzle view of combustionchamber and throat region

velopment and construction require considerable technologicaleffort

1 Dual-Throat Nozzles

A dual-throat nozzle con guration is shown in Fig 28 Itconsists of two conventional bell-thrust chambers with onelocated completely inside the other At low altitude the outerthrust chamber operates with the inner thrust chamber runningin parallel In this operation mode the engine has a larger throatproviding a moderate expansion area ratio During the missionthe outer thrust chamber is shut off and operation continueswith only the inner engine In this con guration ow fromthe inner engine expands and attaches supersonically to theouter engine resulting in a higher expansion area ratio for theremainder of the burn Flow phenomena in both operationmodes are included in Fig 29

Hot- red tests at Aerojet were conducted to provide heattransfer data that were very useful for the thermal analysis anddesign of the dual-throat nozzle con guration37 These testsshowed that ow separation occurred in the inner engine noz-zle at higher ratios of outer to inner chamber pressures duringthe rst operation mode with both chambers burning in par-allel The ow separation resulted in a higher heat load to theinner nozzle Subscale tests performed at the Keldysh Center3

have shown that the additional loss caused by the nozzle con-tour discontinuity during vacuum operation with active innerchamber is in the range of 08ndash 4 depending on geometricaldata (see Fig 30) This high-performance loss results from theinteraction of the inner chamber jet with the outer chambernozzle wall3 The decrease of the jet incidence angle on thewall by means of gas injection through the outer chamber re-duces this performance loss by 04ndash 073

2 Dual-Expander Nozzles

A dual-expander nozzle has two concentric thrust chambersand nozzles It consists of a conventional bell thrust chambersurrounded by an annular thrust chamber Both chambers haveshort primary nozzles which end in a common divergent noz-zle extension Figure 31 shows a typical dual-expander nozzlecon guration At low altitude both thrust chambers operatesharing the same exit area which results in a moderate expan-sion area ratio Part way into the mission one thrust chamberis shut off allowing the other nozzle to use the whole exitarea creating a high-expansion-area ratio for the remainder of

the burn In principle the two operation modes are comparableto those of dual-throat nozzles

Numerical simulations of the ow elds in dual-expandernozzles during all operation modes have been performed atAerojet38 DLR39 and the Keldysh Center Aerojet simulationsare based on an engine design using different propellantshydrocarbonoxygen for the inner chamber and hydrogenox-ygen for the outer chamber whereas DLR simulations arebased on a combined vehicleengine analysis using single-fuelsingle-mixture ratio dual-expander engines with hydrogenoxy-gen Table 4 summarizes combustion chamber parameters Flowphenomena observed in numerical simulations are shown in Fig32a for the mode 1 operation with both thrust chambers burn-ing Compression waves are induced near the inner nozzle lipas a result of the strongly inhomogeneous ow character at thepoint where both exhaust gases of inner and outer combustionchamber merge into the common divergent nozzle part Upand down variations of pressure ratios at the lip did not changethis wave formation even in cases of signi cantly lower exitpressures in the inner nozzle compared to the pressure eld ofthe outer nozzle at the lip Further downstream the compres-sion waves interact with the wall resulting in a re ection ofthe compression waves back into the ow eld

During mode 2 operation a strong expansion of the outernozzle ow is observed when the expansion ratio suddenlyincreases at the end of the nozzle lip The ow is directedtoward the axis of symmetry Near the centerline the ow thenturns over to the axial direction inducing a recompressionshock The static pressure rises signi cantly in this recom-pression region on the centerline A sub- and supersonic recir-culation zone establishes in the inner chamber of the dual-expander nozzle Figure 32b emphasizes the essential owpattern in this operation mode

These analyses have shown that dual-expander nozzles pro-duce high performance in both operation modes3839 Figure 33summarizes performance behavior as a function of ight alti-tude for the dual-expander engine simulated at DLR For com-parison performance data are included for two reference bell

HAGEMANN ET AL 633

Fig 33 Performance data for a dual-expander nozzle

nozzles that have the same area ratios as the dual-expandernozzle during its two operation modes Losses because of thenonhomogeneous exit ow and the induced shocks are com-parable with corresponding values of both conventional noz-zles1

Within the design process of dual-expander engines it is nec-essary to take into account that the nonburning chamber isexposed to high heat uxes and pressure oscillations duringmode 2 operation which might reach a high level and lead tostructural disintegration of the inner nozzle structure There-fore a mode-2 operation with bleed gas generation at a ratherlow thrust chamber pressure level in the inner chamber seemsto be advantageous Because of the bleed gas generation in theprimary chamber the overall mass ow rate during this alter-native mode-2 operation is increased In addition the reductionof the effective exit area ratio results in minor impulse deg-radations during this operation mode when compared with themode-2 operation with no bleed gas generation

Several analytical works on SSTO- and two-stage-to-orbit(TSTO)-vehicles using hydrogenpropane or hydrogenmeth-ane as fuels revealed the lowest vehicle dry masses for dual-mode engines in comparison to other engines1439 Other dual-mode engines using hydrogen as the single fuel but using twomixture ratios also revealed some bene ts over conventionalengines for SSTO- and TSTO applications Even a single-fueloperation with constant mixture ratios in both combustionchambers indicated a gain in launcher performance40

V ConclusionsSeveral nozzle concepts that promise gains in performance

over conventional nozzles were discussed in this paper in-cluding performance enhancements achieved by slight modi- cations of existing nozzles eg through cool gas injectioninto the supersonic nozzle part It is shown that signi cantperformance gains result from the adaptation of the exhaust ow to the ambient pressure and special emphasis was givento altitude adaptive nozzle concepts

A number of nozzle concepts with altitude-compensating ca-pability were identi ed and described To assist the selectionof the best nozzle concept for launch vehicle applications theperformance of the nozzles must be characterized This can bedone using computational uid dynamics (CFD) andor cold- ow tests Existing CFD methods that are in use in the aero-space industry and at research institutes have been veri ed fora wide number of sub- and full-scale experiments and providesuf ciently reliable performance determination for the differ-ent nozzle types

Theoretical evaluations numerical simulations and test re-sults showed that the different concepts have real altitude-com-pensating capabilities However the compensation capabilitiesare limited and there are some drawbacks associated with eachof the concepts Additional performance losses are induced inpractically all of these nozzle concepts when compared to anideal expansion mainly because of the nonisentropic effects

such as shock waves and pressure losses in recirculation zonesHowever these additional performance losses are less than1 ndash 3 depending on the different nozzle concepts

Four nozzle concepts have been selected for further evalu-ation under a NASA Aerojet cooperative agreement They in-clude a plug nozzle an E ndash D nozzle a dual-bell nozzle and adual-expander nozzle The nozzles in addition to a referencebell nozzle have been designed by Aerojet for cold ow testsCFD calculations performed at Aerojet demonstrated that allof these nozzles have high and approximately equal perfor-mances at the design points

In Europe special attention is being paid to plug- and dual-bell nozzles Different plug nozzle concepts are being testedin cold- ow tests within the ESA ARPT (clustered plug noz-zle) and ESA FESTIP (linear plug nozzle) contracts In addi-tion to experiments three-dimensional numerical simulationson these nozzle concepts are being performed Additional ex-periments on different nozzle concepts with forced- ow sep-aration devices eg dual-bell nozzles are currently plannedat DLR

Different advanced nozzle concepts have been examined andare under examination now at the Keldysh Center with bothCFD calculations and experiments on the Nozzle DifferentialFacility which has a nozzle performance determination errorof about 005 This high accuracy is achieved by simulta-neously running the test nozzle and a reference nozzle bothmounted on a common longitudinal axis The performancecharacteristics of the reference nozzle are well known and thethrust difference is measured with strain gauges This facilityis equipped with a wind tunnel to determine nozzle thrust per-formance in the presence of external transonic ows withMach numbers up to 12

In addition to aerodynamic performance other technical is-sues (weight cost design thermal management manufactur-ing system performance and reliability) must be addressedFurthermore before a nal decision can be made as to whichnozzle concept offers the greatest bene ts with regard to aneffective payload mass injection combined launcher and tra-jectory calculations must be performed and compared to a ref-erence launcher concept with conventional nozzles Differentnozzle ef ciencies which account for the additional losses ofadvanced rocket nozzles and are extracted from numerical sim-ulations and experiments must be taken into account

At least for some of these nozzle concepts the plug nozzlethe dual-bell nozzle and the dual-expander nozzle bene tswith regard to lower overall launcher masses have been dem-onstrated in the literature Furthermore the plug nozzle con-cept will be the rst in ight of all of these advanced nozzleconcepts with the X-33 demonstrator vehicle

AcknowledgmentPresented as Paper 18 at the 3rd International Symposium

on Space Propulsion Beijing Peoplersquos Republic of ChinaAug 11ndash 13 1997

References1Manski D and Hagemann G lsquolsquoIn uence of Rocket Design Pa-

rameters on Engine Nozzle Ef cienciesrsquorsquo Journal of Propulsion andPower Vol 12 No 1 1996 pp 41ndash 47

2lsquolsquoLiquid Rocket Engine Nozzlesrsquorsquo NASA Space Vehicle DesignCriteria NASA SP-8120 1976

3Dumnov G E Nikulin G Z and Ponomaryov N B lsquolsquoInves-tigation of Advanced Nozzles for Rocket Enginesrsquorsquo Space Rocket En-gines and Power Plants Vol 4 No 142 NIITP 1993 pp 1012 ndash

1018 (in Russian)4Nguyen T V and Pieper J L lsquolsquoNozzle Flow Separationrsquorsquo Pro-

ceedings of the 5th International Symposium of Propulsion in SpaceTransportation (Paris France) 1996

5Manski D lsquolsquoClustered Plug Nozzles for Future European Reus-able Rocket Launchersrsquorsquo German Aerospace Research Establishment643-817 Lampoldshausen Germany 1981 (in German)

6Horn M and Fisher S lsquolsquoDual-Bell Altitude Compensating Noz-

632 HAGEMANN ET AL

Table 4 Thrust chamber parameters for simulated dual-expander nozzlesa

Operation mode

Aerojet

First (inout) Second (inout)

DLR

First (inout) Second (inout)

Propellants C3H8ndash O2H2ndash O2 mdashH2ndash O2 H2ndash O2H2ndash O2 mdashH2ndash O2

Chamber pressure pc 414207 bar mdash207 bar 200200 bar mdash200 barMixture ratio r 337 mdash7 77 mdash7Exit area ratio laquo 69 146 58 116aInnerouter chambers

Fig 32 Flow phenomena in a dual-expander nozzle during a)sea-level- and b) high-altitude operation

Fig 31 Sketch of a dual-expander nozzle view of combustionchamber and throat region

velopment and construction require considerable technologicaleffort

1 Dual-Throat Nozzles

A dual-throat nozzle con guration is shown in Fig 28 Itconsists of two conventional bell-thrust chambers with onelocated completely inside the other At low altitude the outerthrust chamber operates with the inner thrust chamber runningin parallel In this operation mode the engine has a larger throatproviding a moderate expansion area ratio During the missionthe outer thrust chamber is shut off and operation continueswith only the inner engine In this con guration ow fromthe inner engine expands and attaches supersonically to theouter engine resulting in a higher expansion area ratio for theremainder of the burn Flow phenomena in both operationmodes are included in Fig 29

Hot- red tests at Aerojet were conducted to provide heattransfer data that were very useful for the thermal analysis anddesign of the dual-throat nozzle con guration37 These testsshowed that ow separation occurred in the inner engine noz-zle at higher ratios of outer to inner chamber pressures duringthe rst operation mode with both chambers burning in par-allel The ow separation resulted in a higher heat load to theinner nozzle Subscale tests performed at the Keldysh Center3

have shown that the additional loss caused by the nozzle con-tour discontinuity during vacuum operation with active innerchamber is in the range of 08ndash 4 depending on geometricaldata (see Fig 30) This high-performance loss results from theinteraction of the inner chamber jet with the outer chambernozzle wall3 The decrease of the jet incidence angle on thewall by means of gas injection through the outer chamber re-duces this performance loss by 04ndash 073

2 Dual-Expander Nozzles

A dual-expander nozzle has two concentric thrust chambersand nozzles It consists of a conventional bell thrust chambersurrounded by an annular thrust chamber Both chambers haveshort primary nozzles which end in a common divergent noz-zle extension Figure 31 shows a typical dual-expander nozzlecon guration At low altitude both thrust chambers operatesharing the same exit area which results in a moderate expan-sion area ratio Part way into the mission one thrust chamberis shut off allowing the other nozzle to use the whole exitarea creating a high-expansion-area ratio for the remainder of

the burn In principle the two operation modes are comparableto those of dual-throat nozzles

Numerical simulations of the ow elds in dual-expandernozzles during all operation modes have been performed atAerojet38 DLR39 and the Keldysh Center Aerojet simulationsare based on an engine design using different propellantshydrocarbonoxygen for the inner chamber and hydrogenox-ygen for the outer chamber whereas DLR simulations arebased on a combined vehicleengine analysis using single-fuelsingle-mixture ratio dual-expander engines with hydrogenoxy-gen Table 4 summarizes combustion chamber parameters Flowphenomena observed in numerical simulations are shown in Fig32a for the mode 1 operation with both thrust chambers burn-ing Compression waves are induced near the inner nozzle lipas a result of the strongly inhomogeneous ow character at thepoint where both exhaust gases of inner and outer combustionchamber merge into the common divergent nozzle part Upand down variations of pressure ratios at the lip did not changethis wave formation even in cases of signi cantly lower exitpressures in the inner nozzle compared to the pressure eld ofthe outer nozzle at the lip Further downstream the compres-sion waves interact with the wall resulting in a re ection ofthe compression waves back into the ow eld

During mode 2 operation a strong expansion of the outernozzle ow is observed when the expansion ratio suddenlyincreases at the end of the nozzle lip The ow is directedtoward the axis of symmetry Near the centerline the ow thenturns over to the axial direction inducing a recompressionshock The static pressure rises signi cantly in this recom-pression region on the centerline A sub- and supersonic recir-culation zone establishes in the inner chamber of the dual-expander nozzle Figure 32b emphasizes the essential owpattern in this operation mode

These analyses have shown that dual-expander nozzles pro-duce high performance in both operation modes3839 Figure 33summarizes performance behavior as a function of ight alti-tude for the dual-expander engine simulated at DLR For com-parison performance data are included for two reference bell

HAGEMANN ET AL 633

Fig 33 Performance data for a dual-expander nozzle

nozzles that have the same area ratios as the dual-expandernozzle during its two operation modes Losses because of thenonhomogeneous exit ow and the induced shocks are com-parable with corresponding values of both conventional noz-zles1

Within the design process of dual-expander engines it is nec-essary to take into account that the nonburning chamber isexposed to high heat uxes and pressure oscillations duringmode 2 operation which might reach a high level and lead tostructural disintegration of the inner nozzle structure There-fore a mode-2 operation with bleed gas generation at a ratherlow thrust chamber pressure level in the inner chamber seemsto be advantageous Because of the bleed gas generation in theprimary chamber the overall mass ow rate during this alter-native mode-2 operation is increased In addition the reductionof the effective exit area ratio results in minor impulse deg-radations during this operation mode when compared with themode-2 operation with no bleed gas generation

Several analytical works on SSTO- and two-stage-to-orbit(TSTO)-vehicles using hydrogenpropane or hydrogenmeth-ane as fuels revealed the lowest vehicle dry masses for dual-mode engines in comparison to other engines1439 Other dual-mode engines using hydrogen as the single fuel but using twomixture ratios also revealed some bene ts over conventionalengines for SSTO- and TSTO applications Even a single-fueloperation with constant mixture ratios in both combustionchambers indicated a gain in launcher performance40

V ConclusionsSeveral nozzle concepts that promise gains in performance

over conventional nozzles were discussed in this paper in-cluding performance enhancements achieved by slight modi- cations of existing nozzles eg through cool gas injectioninto the supersonic nozzle part It is shown that signi cantperformance gains result from the adaptation of the exhaust ow to the ambient pressure and special emphasis was givento altitude adaptive nozzle concepts

A number of nozzle concepts with altitude-compensating ca-pability were identi ed and described To assist the selectionof the best nozzle concept for launch vehicle applications theperformance of the nozzles must be characterized This can bedone using computational uid dynamics (CFD) andor cold- ow tests Existing CFD methods that are in use in the aero-space industry and at research institutes have been veri ed fora wide number of sub- and full-scale experiments and providesuf ciently reliable performance determination for the differ-ent nozzle types

Theoretical evaluations numerical simulations and test re-sults showed that the different concepts have real altitude-com-pensating capabilities However the compensation capabilitiesare limited and there are some drawbacks associated with eachof the concepts Additional performance losses are induced inpractically all of these nozzle concepts when compared to anideal expansion mainly because of the nonisentropic effects

such as shock waves and pressure losses in recirculation zonesHowever these additional performance losses are less than1 ndash 3 depending on the different nozzle concepts

Four nozzle concepts have been selected for further evalu-ation under a NASA Aerojet cooperative agreement They in-clude a plug nozzle an E ndash D nozzle a dual-bell nozzle and adual-expander nozzle The nozzles in addition to a referencebell nozzle have been designed by Aerojet for cold ow testsCFD calculations performed at Aerojet demonstrated that allof these nozzles have high and approximately equal perfor-mances at the design points

In Europe special attention is being paid to plug- and dual-bell nozzles Different plug nozzle concepts are being testedin cold- ow tests within the ESA ARPT (clustered plug noz-zle) and ESA FESTIP (linear plug nozzle) contracts In addi-tion to experiments three-dimensional numerical simulationson these nozzle concepts are being performed Additional ex-periments on different nozzle concepts with forced- ow sep-aration devices eg dual-bell nozzles are currently plannedat DLR

Different advanced nozzle concepts have been examined andare under examination now at the Keldysh Center with bothCFD calculations and experiments on the Nozzle DifferentialFacility which has a nozzle performance determination errorof about 005 This high accuracy is achieved by simulta-neously running the test nozzle and a reference nozzle bothmounted on a common longitudinal axis The performancecharacteristics of the reference nozzle are well known and thethrust difference is measured with strain gauges This facilityis equipped with a wind tunnel to determine nozzle thrust per-formance in the presence of external transonic ows withMach numbers up to 12

In addition to aerodynamic performance other technical is-sues (weight cost design thermal management manufactur-ing system performance and reliability) must be addressedFurthermore before a nal decision can be made as to whichnozzle concept offers the greatest bene ts with regard to aneffective payload mass injection combined launcher and tra-jectory calculations must be performed and compared to a ref-erence launcher concept with conventional nozzles Differentnozzle ef ciencies which account for the additional losses ofadvanced rocket nozzles and are extracted from numerical sim-ulations and experiments must be taken into account

At least for some of these nozzle concepts the plug nozzlethe dual-bell nozzle and the dual-expander nozzle bene tswith regard to lower overall launcher masses have been dem-onstrated in the literature Furthermore the plug nozzle con-cept will be the rst in ight of all of these advanced nozzleconcepts with the X-33 demonstrator vehicle

AcknowledgmentPresented as Paper 18 at the 3rd International Symposium

on Space Propulsion Beijing Peoplersquos Republic of ChinaAug 11ndash 13 1997

References1Manski D and Hagemann G lsquolsquoIn uence of Rocket Design Pa-

rameters on Engine Nozzle Ef cienciesrsquorsquo Journal of Propulsion andPower Vol 12 No 1 1996 pp 41ndash 47

2lsquolsquoLiquid Rocket Engine Nozzlesrsquorsquo NASA Space Vehicle DesignCriteria NASA SP-8120 1976

3Dumnov G E Nikulin G Z and Ponomaryov N B lsquolsquoInves-tigation of Advanced Nozzles for Rocket Enginesrsquorsquo Space Rocket En-gines and Power Plants Vol 4 No 142 NIITP 1993 pp 1012 ndash

1018 (in Russian)4Nguyen T V and Pieper J L lsquolsquoNozzle Flow Separationrsquorsquo Pro-

ceedings of the 5th International Symposium of Propulsion in SpaceTransportation (Paris France) 1996

5Manski D lsquolsquoClustered Plug Nozzles for Future European Reus-able Rocket Launchersrsquorsquo German Aerospace Research Establishment643-817 Lampoldshausen Germany 1981 (in German)

6Horn M and Fisher S lsquolsquoDual-Bell Altitude Compensating Noz-

HAGEMANN ET AL 633

Fig 33 Performance data for a dual-expander nozzle

nozzles that have the same area ratios as the dual-expandernozzle during its two operation modes Losses because of thenonhomogeneous exit ow and the induced shocks are com-parable with corresponding values of both conventional noz-zles1

Within the design process of dual-expander engines it is nec-essary to take into account that the nonburning chamber isexposed to high heat uxes and pressure oscillations duringmode 2 operation which might reach a high level and lead tostructural disintegration of the inner nozzle structure There-fore a mode-2 operation with bleed gas generation at a ratherlow thrust chamber pressure level in the inner chamber seemsto be advantageous Because of the bleed gas generation in theprimary chamber the overall mass ow rate during this alter-native mode-2 operation is increased In addition the reductionof the effective exit area ratio results in minor impulse deg-radations during this operation mode when compared with themode-2 operation with no bleed gas generation

Several analytical works on SSTO- and two-stage-to-orbit(TSTO)-vehicles using hydrogenpropane or hydrogenmeth-ane as fuels revealed the lowest vehicle dry masses for dual-mode engines in comparison to other engines1439 Other dual-mode engines using hydrogen as the single fuel but using twomixture ratios also revealed some bene ts over conventionalengines for SSTO- and TSTO applications Even a single-fueloperation with constant mixture ratios in both combustionchambers indicated a gain in launcher performance40

V ConclusionsSeveral nozzle concepts that promise gains in performance

over conventional nozzles were discussed in this paper in-cluding performance enhancements achieved by slight modi- cations of existing nozzles eg through cool gas injectioninto the supersonic nozzle part It is shown that signi cantperformance gains result from the adaptation of the exhaust ow to the ambient pressure and special emphasis was givento altitude adaptive nozzle concepts

A number of nozzle concepts with altitude-compensating ca-pability were identi ed and described To assist the selectionof the best nozzle concept for launch vehicle applications theperformance of the nozzles must be characterized This can bedone using computational uid dynamics (CFD) andor cold- ow tests Existing CFD methods that are in use in the aero-space industry and at research institutes have been veri ed fora wide number of sub- and full-scale experiments and providesuf ciently reliable performance determination for the differ-ent nozzle types

Theoretical evaluations numerical simulations and test re-sults showed that the different concepts have real altitude-com-pensating capabilities However the compensation capabilitiesare limited and there are some drawbacks associated with eachof the concepts Additional performance losses are induced inpractically all of these nozzle concepts when compared to anideal expansion mainly because of the nonisentropic effects

such as shock waves and pressure losses in recirculation zonesHowever these additional performance losses are less than1 ndash 3 depending on the different nozzle concepts

Four nozzle concepts have been selected for further evalu-ation under a NASA Aerojet cooperative agreement They in-clude a plug nozzle an E ndash D nozzle a dual-bell nozzle and adual-expander nozzle The nozzles in addition to a referencebell nozzle have been designed by Aerojet for cold ow testsCFD calculations performed at Aerojet demonstrated that allof these nozzles have high and approximately equal perfor-mances at the design points

In Europe special attention is being paid to plug- and dual-bell nozzles Different plug nozzle concepts are being testedin cold- ow tests within the ESA ARPT (clustered plug noz-zle) and ESA FESTIP (linear plug nozzle) contracts In addi-tion to experiments three-dimensional numerical simulationson these nozzle concepts are being performed Additional ex-periments on different nozzle concepts with forced- ow sep-aration devices eg dual-bell nozzles are currently plannedat DLR

Different advanced nozzle concepts have been examined andare under examination now at the Keldysh Center with bothCFD calculations and experiments on the Nozzle DifferentialFacility which has a nozzle performance determination errorof about 005 This high accuracy is achieved by simulta-neously running the test nozzle and a reference nozzle bothmounted on a common longitudinal axis The performancecharacteristics of the reference nozzle are well known and thethrust difference is measured with strain gauges This facilityis equipped with a wind tunnel to determine nozzle thrust per-formance in the presence of external transonic ows withMach numbers up to 12

In addition to aerodynamic performance other technical is-sues (weight cost design thermal management manufactur-ing system performance and reliability) must be addressedFurthermore before a nal decision can be made as to whichnozzle concept offers the greatest bene ts with regard to aneffective payload mass injection combined launcher and tra-jectory calculations must be performed and compared to a ref-erence launcher concept with conventional nozzles Differentnozzle ef ciencies which account for the additional losses ofadvanced rocket nozzles and are extracted from numerical sim-ulations and experiments must be taken into account

At least for some of these nozzle concepts the plug nozzlethe dual-bell nozzle and the dual-expander nozzle bene tswith regard to lower overall launcher masses have been dem-onstrated in the literature Furthermore the plug nozzle con-cept will be the rst in ight of all of these advanced nozzleconcepts with the X-33 demonstrator vehicle

AcknowledgmentPresented as Paper 18 at the 3rd International Symposium

on Space Propulsion Beijing Peoplersquos Republic of ChinaAug 11ndash 13 1997

References1Manski D and Hagemann G lsquolsquoIn uence of Rocket Design Pa-

rameters on Engine Nozzle Ef cienciesrsquorsquo Journal of Propulsion andPower Vol 12 No 1 1996 pp 41ndash 47

2lsquolsquoLiquid Rocket Engine Nozzlesrsquorsquo NASA Space Vehicle DesignCriteria NASA SP-8120 1976

3Dumnov G E Nikulin G Z and Ponomaryov N B lsquolsquoInves-tigation of Advanced Nozzles for Rocket Enginesrsquorsquo Space Rocket En-gines and Power Plants Vol 4 No 142 NIITP 1993 pp 1012 ndash

1018 (in Russian)4Nguyen T V and Pieper J L lsquolsquoNozzle Flow Separationrsquorsquo Pro-

ceedings of the 5th International Symposium of Propulsion in SpaceTransportation (Paris France) 1996

5Manski D lsquolsquoClustered Plug Nozzles for Future European Reus-able Rocket Launchersrsquorsquo German Aerospace Research Establishment643-817 Lampoldshausen Germany 1981 (in German)

6Horn M and Fisher S lsquolsquoDual-Bell Altitude Compensating Noz-

634 HAGEMANN ET AL

zlesrsquorsquo NASA CR-194719 19947Immich H and Caporicci M lsquolsquoFESTIP Technology Develop-

ments in Liquid Rocket Propulsion for Reusable Launch VehiclesrsquorsquoAIAA Paper 96-3113 July 1996

8Summer eld M Foster C and Swan W lsquolsquoFlow Separation inOverexpanded Supersonic Exhaust Nozzlesrsquorsquo Jet Propulsion Sept1954 pp 319ndash 321

9Schmucker R lsquolsquoFlow Processes in Overexpanding Nozzles ofChemical Rocket Enginesrsquorsquo Technical Univ TB-7-10-14 MunichGermany 1973 (in German)

10Hagemann G Schley C-A Odintsov E and Sobatchkine AlsquolsquoNozzle Flow eld Analysis with Particular Regard to 3D-Plug-Clus-ter Con gurationsrsquorsquo AIAA Paper 96-2954 July 1996

11Dumnov G E lsquolsquoUnsteady Side-Loads Acting on the Nozzle withDeveloped Separation Zonersquorsquo AIAA Paper 96-3220 July 1996

12Nave L H and Coffey G A lsquolsquoSea-Level Side-Loads in HighArea Ratio Rocket Enginesrsquorsquo AIAA Paper 73-1284 July 1973

13Pekkari L O lsquolsquoAdvanced Nozzlesrsquorsquo Proceedings of the 5th In-ternational Symposium of Propulsion in Space Transportation (ParisFrance) 1996 pp 104 ndash 1011

14Manski D Goertz C Saszlignick H-D Hulka J R Goracke BD and Levack D J H lsquolsquoCycles for Earth-to-Orbit PropulsionrsquorsquoJournal of Propulsion and Power Vol 14 No 5 1998 pp 588 ndash 604

15Hagemann G Krulle G and Hannemann K lsquolsquoNumericalFlow eld Analysis of the Next Generation Vulcain Nozzlersquorsquo Journalof Propulsion and Power Vol 12 No 4 1996 pp 655ndash 661

16Voinow A L and Melnikov D A lsquolsquoPerformance of RocketEngine Nozzles with Slot Injectionrsquorsquo AIAA Paper 96-3218 July 1996

17Forster C and Cowles F lsquolsquoExperimental Study of Gas FlowSeparation in Overexpanded Exhaust Nozzles for Rocket Motorsrsquorsquo JetPropulsion Lab Progress Rept 4-103 California Inst of TechnologyPasadena PA May 1949

18Hagemann G and Frey M lsquolsquoA Critical Assessment of Dual-Bell Nozzlesrsquorsquo AIAA Paper 97-3299 July 1997

19Immich H and Caporicci M lsquolsquoStatus of the FESTIP RocketPropulsion Technology Programrsquorsquo AIAA Paper 97-3311 July 1997

20Luke G lsquolsquoUse of Nozzle Trip Rings to Reduce Nozzle Separa-tion Side Force During Stagingrsquorsquo AIAA Paper 92-3617 July 1992

21Chiou J and Hung R lsquolsquoA Study of Forced Flow Separation inRocket Nozzlersquorsquo Alabama Univ Final Rept Huntsville AL July 1974

22Schmucker R lsquolsquoA Procedure for Calculation of Boundary LayerTrip Protuberances in Overexpanded Rocket Nozzlesrsquorsquo NASA TM X-64843 1973

23Goncharov N Orlov V Rachuk V Shostak A and Starke RlsquolsquoReusable Launch Vehicle Propulsion Based on the RD-0120 En-ginersquorsquo AIAA Paper 95-3003 April 1995

24Clayton R and Back L lsquolsquoThrust Improvement with Ablative

Insert Nozzle Extensionrsquorsquo Jet Propulsion Vol 2 No 1 1986 pp91ndash 93

25Parsley R C and van Stelle K J lsquolsquoAltitude Compensating Noz-zle Evaluationrsquorsquo AIAA Paper 92-3456 July 1992

26lsquolsquoARPTmdashAdvanced Rocket Propulsion Technology Program Fi-nal Reports Phase 1 and 2rsquorsquo European Space AgencymdashEuropeanSpace Research and Technology Centre The Netherlands 1996

27Rommel T Hagemann G Schley C-A Manski D andKrulle G lsquolsquoPlug Nozzle Flow eld Calculations for SSTO Applica-tionsrsquorsquo Journal of Propulsion and Power Vol 13 No 6 1997 pp629ndash 634

28Tomita T Tamura H and Takahashi M lsquolsquoAn ExperimentalEvaluation of Plug Nozzle Flow Fieldrsquorsquo AIAA Paper 96-2632 July1996

29Angelino G lsquolsquoTheoretical and Experimental Investigations of theDesign and Performance of a Plug Type Nozzlersquorsquo NASA TN-12 July1963

30Lee C C lsquolsquoFortran Programs for Plug Nozzle Designrsquorsquo NASATN R-41 March 1963

31Nguyen T V Spencer R G and Siebenhaar A lsquolsquoAerodynamicPerformance of a Round-to-Square Nozzlersquorsquo Proceedings of the 35thHeat Transfer and Fluid Mechanics Institute 1997

32Beheim M A and Boksenbom A S lsquolsquoVariable Geometry Re-quirements in Inlets and Exhaust Nozzles for High Mach NumberApplicationsrsquorsquo NASA TM X-52447 1968

33Valerino A S Zappa R F and Abdalla K L lsquolsquoEffects ofExternal Stream on the Performance of Isentropic Plug-Type Nozzlesat Mach Numbers of 20 18 and 15rsquorsquo NASA 2-17-59E 1969

34Mercer C E and Salters L E Jr lsquolsquoPerformance of a PlugNozzle Having a Concave Central Base with and Without TerminalFairings at Transonic Speedsrsquorsquo NASA TN D-1804 May 1963

35Wasko R A lsquolsquoPerformance of Annular Plug and Expansion-De- ection Nozzles Including External Flow Effects at Transonic MachNumbersrsquorsquo NASA TN D-4462 April 1968

36Smith-Kent R Loh H and Chwalowski P lsquolsquoAnalytical Con-touring of Pintle Nozzle Exit Cone Using Computational Fluid Dy-namicsrsquorsquo AIAA Paper 95-2877 1995

37Ewen R L and OlsquoBrian C J lsquolsquoDual-Throat Thruster ResultsrsquorsquoAIAA Paper 86-1518 1986

38Nguyen T V Hyde J C and Ostrander M J lsquolsquoAerodynamicPerformance Analysis of Dual-FuelDual-Expander Nozzlesrsquorsquo AIAAPaper 88-2818 1988

39Hagemann G Krulle G and Manski D lsquolsquoDual-Expander En-gine Flow eld Simulationsrsquorsquo AIAA Paper 95-3135 July 1995

40Manski D Hagemann G and Saszlignick H D lsquolsquoOptimizationof Dual-Expander Rocket Engines in Single-Stage-to-Orbit VehiclesrsquorsquoActa Astronautica Vol 40 No 2ndash 8 1997 pp 151ndash 163