Engine And Installation Configurations For A Silent Aircraft

Transcription

1 1 Engine And Installation Conigurations For A Silent Aircrat Cesare A. Hall * and aniel Crichton Cambridge University Engineering eartment Trumington Street, Cambridge CB2 1PZ, UK cah1003@cam.ac.uk ISABE ABSTRACT The Silent Aircrat Initiative is a research roect unded by the Cambridge-MIT Institute aimed at reducing aircrat noise to the oint where it is imercetible in the urban environments around airorts. The aircrat that ulils this obective must also be economically cometitive with conventional aircrat o the uture and thereore uel consumtion is a key consideration or the design. This aer identiies some key eatures o a roulsion system that can achieve the Silent Aircrat noise target and exlores the relationshis between the actors that aect uel consumtion. It also considers the dierent demands made o an ine at dierent oerating conditions in the light enveloe. These studies are used to roose viable ine and installation conigurations that could meet the Silent Aircrat noise requirements. The indings oint towards a multile turboan system with a variable geometry exhaust and a novel, embedded installation. Symbols a A c g l L m n P Q a Re s T V NOMENCLATURE Seed o sound Area Seciic heat caacity at constant ressure iameter, rag Fan ti diameter Acceleration due to gravity Lth Lit Mass low rate Number o ine units in the roulsion system Pressure Power Fan low caacity, Q Reynolds number istance, aircrat range Temerature Flow velocity (relative to moving aircrat) * Research Associate, Ph. Ph Candidate. Other Contributing Authors: T. Hynes, T. Law and M. Sargeant a = m c T A V 0 W X N η,η th ρ Flight seed o aircrat Weight Net thrust Engine roulsive, thermal eiciencies ensity Subscrits 0 Total, stagnation value 1 Conditions at ine inlet entry 2 Conditions at the ine ace Engine arameter Fan arameter Conditions in the exhaust et Far-ield value Abbreviations BPR Engine byass ratio FPR HTR LCV PR sc Fan total ressure ratio Hub-to-ti radius ratio or the an rotor Lower caloriic value o uel Pressure recovery Thrust seciic uel consumtion INTROUCTION Since the dawn o the et age the noise generated by civil aircrat has decreased by more than 20dB at a given thrust level. To the listener this is heard as a ourold reduction in noise and it reresents a all o a actor o more than 100 in terms o the acoustic ower generated [1]. The maority o this reduction has come rom the introduction o high byass ratio ines. Relative to earlier low byass ratio ines these roduce a slower et that is much quieter as well as more eicient at generating thrust. Over the last twenty years, byass ratios have continued to increase, but the resulting reductions in ine noise emission have been more incremental as internal noise sources have started to become dominant. There have also been many other technological imrovements that have been made to urther reduce noise, including advanced comonent design to minimise source noise and imroved acoustic absorbers. However, the task o making signiicant urther noise reductions or turboans has become increasingly diicult. Published by the American Institute o Aeronautics and Astronautics, Inc., with ermission.

2 2 For current conigurations o aircrat and ines we have reached a oint where design imrovements or urther noise reduction are oten at the exense o uel consumtion. Increasing byass ratio and thus an diameter leads to greater drag on ine installations that is not necessarily oset by the simultaneous imrovement in roulsive eiciency. A good examle o this conlict between noise emission and uel consumtion is in the recent ine design or the Airbus A380 [1]. The an diameter o this ine was increased so that the aircrat would incur ewer noise quota oints when oerating in and out o Heathrow airort [2]. This incurred a enalty in overall uel burn. The demand or aircrat to be both quieter and more uel eicient is greater than ever. The increase in air traic means the number o aircrat oerations are continuously leading to both greater noise and greater emissions o ollutants. The ACARE 2020 vision, develoed with industry, has the ambitious target o cutting both noise emission and uel consumtion o aircrat to one hal o the levels rom aircrat built in 2000 by the year 2020 [3]. This level o reduction is exected to require maor technological breakthroughs in both ine and airrame design. The Silent Aircrat Initiative is designing a concet aircrat with noise emission as the rimary design variable. The aircrat is aimed at entry into service in about 20 years and the ambitious obective is to reduce the noise generated to the oint where it would be imercetible above background noise in a tyical urban environment outside an airort. Such an aircrat could be deemed as silent and this would reresent a reduction in aircrat noise greater than that achieved over the last ity years. Figure 1 illustrates the scale o this challe. Figure 1: Reduction in thrust corrected aircrat noise level over time In order to reach the Silent Aircrat noise target large reductions, relative to current technology, are required or all comonents o ine and airrame noise. In addition to this aggressive noise target, the new aircrat must be economic relative to other aircrat o the uture. This will require a roulsion system that has cometitive uel burn as well as accetable develoment, acquisition and maintenance costs. There have been several studies o new ine conigurations aimed at signiicant imrovements in noise and uel consumtion. The NASA study o advanced ines or high eiciency [4] looks at several conigurations, including geared ans and contra-an designs, aimed at weight and uel burn reductions. Another system study o ine concets carried out by NASA [5] investigates the otimum ine arameters or low noise with accetable oerating costs. The work concludes that a low ressure ratio turboan, in a conventional odded installation, can achieve a 10dB noise reduction without incurring unaccetable cost enalties. Reerence [6] gives a good overview o the technology required to urther reduce noise rom conventional aircrat ines and rooses the use o geared turboans to give a large imrovement in noise emission. However, the noise reductions exected still all ar short o the Silent Aircrat target. A study o low-noise concets and the technological barriers to a unctionally-silent aircrat has been resented in [7]. This aer rooses some novel concets, including distributed roulsion integrated with a blended-wing-body tye aircrat. The indings rom the literature suggest that the aims o the Silent Aircrat Initiative can only be achieved with some radical change in the coniguration o aircrat and roulsion system. The current aer aims to answer the question o what eatures o a roulsion system would be needed to achieve the Silent Aircrat noise target and what design hilosohy could be adoted to make such an ine credible. The aroach is as ollows: 1. The basic requirements or the design o a viable, quiet roulsion system are develoed (a large exhaust area at take-o, otimised oerational rocedures or aircrat dearture and large surace areas or acoustic absorbers and noise shielding). 2. Installation design actors are investigated using some simle quantitative analyses and ossible otions are comared in terms o their imact on noise, uel consumtion and weight. 3. The need or a variable cycle ine is shown and the otential o an ine to be low noise at low altitude and uel eicient at cruise is investigated. 4. Several candidate ine conigurations are roosed and the relative merits o the dierent otions are comared in rearation or uture, more detailed, design studies. Published by the American Institute o Aeronautics and Astronautics, Inc., with ermission.

3 3 BASIC PRINCIPLES The exhaust et noise rom a et ine is highly deendent on the basic thermodynamic cycle o the ine and thereore we will consider this irst. Lighthill s well known acoustic analogy [8] relates acoustic ower to tyical velocity and diameter here taken as absolute et velocity and et diameter: ( V V ) A P ρ (1) a Whilst rogress has been made in reducing et noise through nozzle shaing esecially the use o lobes and/or chevrons ([9, 10] or examle) the reductions achieved are modest when comared to that required or silence. The only sure way to obtain signiicant reductions in exhaust et noise is to reduce the et seed. However, it is the dierence in velocity between the et and the surrounding air that generates thrust: X N = ρ A V ( V V (2) 0) To quieten the et signiicantly whilst maintaining thrust thereore requires a very large exhaust area combined with a low et velocity. Figure 2 shows the variation in required exhaust area with et noise target or a 250 seat, 4000 nautical mile range aircrat. Figure 3 shows our take-o roiles in which the et noise outside o the airort boundary remains below the limit imosed or silence. The baseline roile ust meets all required oerational and regulatory requirements and thereore can be considered otimum or a Silent Aircrat. The simle roile also meets all requirements but requires a 33% increase in et area in order to meet the noise target. The remaining roiles are otimised but or areas 10% above and below baseline. For the smaller area, oerational requirements cannot be met whilst simultaneously meeting the noise requirements. The larger area meets all requirements but will result in an unnecessarily large ine. airort boundary current & next gen Figure 2: Relationshi between take-o exhaust area and et noise outside airort or a 250-seat aircrat The simle dearture roile is or a constant angle o climb ater gear-u whilst the otimised roile is thrustmanaged to maximize rate o climb whilst ensuring the noise limit is not exceeded. The baseline et area is that required or silent take-o when using an otimised dearture roile. This et area is 2 to 3 times as large as that o today s conventional et ines, as indicated on the igure. The calculations use the Stone et noise model [11] and urther details on oerational requirements and take-o conditions considered can be ound in Re [12]. Figure 3: earture roiles or a silent take-o. There are imlications o a large exhaust area or the size, weight and erormance o the ine. For a conventional turboan, i the exhaust area increases the byass ratio and the an diameter must also increase. This increases the installation drag otentially leading to uel burn enalties. The size increase imlies an increase in ine weight because the size o the an comonents and the ine ducting rises. These issues are discussed in the next section. Once the thermodynamic cycle is set or low et noise, the reduction o turbomachinery noise can be addressed. Source noise reduction is tackled initially through various simle design rules: blade seeds can be reduced to remove suersonic sources, numbers o or each blade row can be modiied to revent sources that are cut-on, and the ga/chord geometric ratios can be maximized to reduce interaction noise. Unortunately, imroving these arameters or noise will oten comromise the aerodynamic erormance and the noise reductions available will be limited. More advanced methods, such as 3- design otimization, can enable additional reductions. However, source noise reductions alone will not be suicient to reduce turbomachinery noise to the Silent Aircrat noise target, and or these comonents it will be necessary to develo shielding by the airrame to Published by the American Institute o Aeronautics and Astronautics, Inc., with ermission.

4 4 signiicantly reduce orward roagating an noise and to maximize the otential o acoustic liners, articularly to attenuate rearward an and turbine noise. The otential noise reductions available through shielding and acoustic liners are highly deendent on the roulsion system ackaging with the airrame which is discussed below. The oerating costs associated with the develoment, acquisition and maintenance o a quiet roulsion system are diicult to quantiy. Fuel burn will be a maor cost actor, but its recise imact will deend on uture trends in aviation uel rices and the regulation o aviation emissions. The basic eects o the roulsion system on uel consumtion can be exlored using the Breguet range equation: where, W sw 1 W = 1 + e H s W ( ex( s ) 1) uel (3) V0 L LCV H = = g. sc g η ηth Equation 3 tells us that or an aircrat with a ixed range and ayload, to minimise the uel burn er asser-kilometre we need to minimise the ratio o total aircrat emty weight to ayload weight, W e /W, minimise the seciic uel consumtion o the ine, and maximise the lit-to-drag ratio o the aircrat. The ine cycle directly determines the seciic uel consumtion, but the design o ine and installation will also aect the total weight and, or a highly integrated design, can signiicantly change the aircrat lit and drag. L Figure 4: An all-liting body with embedded ines. The roulsion system lacement on this style o aircrat is limited by structural constraints, the locations o asser bays, emergency exits, uel tanks and the osition o essential aircrat systems such as control suraces and the undercarriage. The most easible location is above (or within) the centre-body o the aircrat behind the asser cabins. The lying wing has a greater volume to surace area ratio than a conventional aircrat and there is a lot o available sace at the back o the airrame that can be used to accommodate embedded ines [13]. Figure 5 summarises the main otions that are being considered or ackaging the ines with the Silent Aircrat airrame. INSTALLATION ESIGN FACTORS The Silent Aircrat airrame design is exected to be a coniguration in which the wing and uselage are merged together, as illustrated in igure 4. Several studies have shown that this shae o aircrat has a higher lit to drag ratio and signiicantly lower emty weight than a tubeand-wing aircrat carrying the same ayload, [13] or examle. These actors enable signiicant savings in uel consumtion (equation 3), but there are also advantages o the airrame or noise reduction. The large, continuous surace o the airrame maximises the otential to shield the orward roagating ine noise rom the ground i the inlets are laced on or above the airrame. In addition, the aerodynamically smooth suraces o an all-liting body also reduce airrame noise sources signiicantly [7]. a) Podded roulsion system b) Embedded system with boundary layer diversion c) Embedded system with boundary layer ingestion Figure 5: Otions or owerlant integration. In order to determine a reerred integration otion, the dierences in uel consumtion and otential noise attenuation can be exlored with the qualitative considerations and the simle quantitative analyses resented in the ollowing sub-sections. Published by the American Institute o Aeronautics and Astronautics, Inc., with ermission.

5 5 The trade-o between roulsive eiciency and drag For an aircrat in cruise, using the eiciency deinitions in [14], the overall eiciency and rate o uel consumtion o the roulsion system can be exressed as ollows: useul _ ower X NVo ηo = ηthη = = thermal _ energy m. LCV m uel uel X NVo = η η. LCV I we consider increases in an diameter, roulsive eiciency will imrove because the exhaust et velocity reduces. At the same time, however, the installation size and drag increase leading to a greater thrust requirement. This trade-o can be exlored quantitatively by considering the ine design or a ixed aircrat at cruise. With net thrust equal to drag, the drag comonents can be exressed in terms o the et area and velocity, using the thrust equation (2): X N th (4) = + = ρ A V ( V V ) (5) 0 airrame Thus, the variation in overall uel burn with an diameter can be determined using equation 4. This aroach was alied to roduce igure 6, which shows how the cruise uel burn is exected to vary or odded and embedded conigurations with overall an diameter and the number o ine units. Note that the reerence drag arameters and an size used to roduce igure 6 were derived rom data or a conventional 250-seat asser aircrat with ines that entered service in the 1980s. Since this time an diameters have already increased by u to 30% and on igure 6 a an diameter aroriate to a current ine design is marked. Note that rom igure 2 the overall an diameter o a turboan that would satisy the Silent Aircrat noise target is exected to be 75% larger than a 1980s turboan. The actual an diameter required is examined in greater detail in the next section. odded I the thrust and drag arameters are known or a reerence design oint, equation 5 can be rewritten to show how the net thrust required varies with the drag contribution rom the ine installations: X + N (6) ( 1) X N, re airrame = 1 + k, re airrame +, re where k, re = and is assumed to be small. airrame Changes in the ine installation drag,, can be aroximated to be roortional to changes in the installation wetted area. For the current study, the ollowing simlistic relationshis were assumed or relating the installation drag to the overall an diameter: For odded installations, π l n = π 2 ( n ) ( l ) embedded n For embedded installations, 2 l = 2 2 ( n ) ( l ) n The overall an diameter, n, is used because or a given aircrat thrust requirement, it is indeendent o the number o ines (see equation 14). I the exressions above are substituted into equations 5 and 6, the variation in et velocity with overall an diameter can be aroximated. The Froude equation gives roulsive eiciency as a unction o et velocity: 2V0 η = V + V 0 (7) Published by the American Institute o Aeronautics and Astronautics, Inc., with ermission. n Figure 6: Variation o uel burn with an diameter or a) odded and b) embedded systems. The above analysis is admittedly quite crude, but the results show some useul trends or selecting a roulsion

6 6 system installation. Firstly, they show that with a odded roulsion system, current levels o an diameter (or 2 or 4 ine systems) are close to the otimum level or minimum cruise uel burn. With a odded system, the increase in installation drag with an diameter is greater as the number o ines increases because the total wetted area is increased. For an embedded system, as the number o ines increases, the roulsion system can be better integrated into the airrame leading to a drag reduction. The lots suggest that an embedded roulsion system with a large number o ine units could enable a higher overall an diameter to be achieved with signiicantly lower uel consumtion than a modern odded coniguration. However, the analysis assumes that the thermal eiciency is unaected by the choice o installation and it does not consider whether or not the airrame boundary layer is ingested by the embedded ines. To maintain net thrust at a ixed overall an diameter, the design ressure ratio o the an must be increased rom the ressure ratio with zero inlet ressure loss according to the ollowing equation: ( FPR) PR FPR = (9) ideal Figure 7 uses equations (8) and (9) to show the variation in thermal eiciency with inlet ressure recovery or three values o ideal an ressure ratio. This analysis assumes that the eect o ressure recovery on the core and byass lows will be similar. The imact o inlet loss on ine erormance With embedded ines the intakes can emloy boundary layer ingestion (BLI) or boundary layer diversion (BL). With boundary layer diversion, the nonuniorm airrame boundary layer air ustream o the intake is revented rom entering the ine by some geometrical eature or device. With boundary layer ingestion, the airrame boundary layer is intentionally drawn into the intake in order to reduce the uel consumtion required. In both cases, embedding the ines leads to extra rictional losses aroaching the ine because the shae o the inlet is more comlex, tyically an S-shaed duct (see igure 5). The erormance o an intake is quantiied by its ressure recovery, which is the ratio o the total ressure at the an-ace to that at entry to the inlet, 02 / 01. A value o would be tyical or a odded ine intake at cruise, whereas the value or an S-duct tye inlet will be closer to 0.95, see [15]. A reliminary comutational study o S-shaed inlets or the Silent Aircrat estimated that the ressure recovery would be about 0.96 with BL and 0.94 i the eects o BLI were included. The imact o the ressure recovery on the thermal eiciency o a an can be determined by considering the work inut with inlet ressure losses, relative to the work required or ully isentroic comression: η th ( ) ( γ 1 FPR. PR ) = ( γ 1 ) γ η FPR γ 1 1 This shows that with a ressure recovery o unity, the thermal eiciency equals the an isentroic eiciency, η. However, as the ressure recovery reduces the imact on eiciency is severe, and this imact increases as the an ressure ratio is reduced. Also, as ressure recovery reduces, the thrust rom an ine will decrease unless the an ressure ratio rises or the an diameter increases. (8) Figure 7: Thermal eiciency eect o inlet ressure recovery or three an ressure ratios. Figure 7 demonstrates the imortance o careully designing the ine inlets to maximise the ressure recovery. It also shows that or an embedded system with a ressure recovery o 0.95, the imact o inlet losses on a low ressure ratio an will be a uel burn enalty in excess o 10%. This could cancel out the uel burn beneit gained rom the drag reduction due to embedding the ines (igure 6). However, boundary layer ingestion oers an alternative oortunity or reducing uel consumtion. Several studies have examined the imact o BLI on ine erormance, see [16] or examle, but it is an area that requires urther research and this is being undertaken as art o the Silent Aircrat Initiative. The ollowing descrition is an overview o the eects o BLI on each o the terms in equation 4 that determine uel consumtion: For a BLI system the overall thermal eiciency is even lower than a BL system because the kinetic energy o the low entering the ine intake is reduced. This eect can be considered as an additional inlet ressure recovery actor relative to the ree stream conditions, 01 / 0, which can be estimated rom the airrame boundary layer arameters at entry to the ine. The net thrust (equation 5) is reduced with BLI because some o the low that goes through the ine and generates thrust would have otherwise contributed to airrame drag. The drag removed by the ines can be estimated as being Published by the American Institute o Aeronautics and Astronautics, Inc., with ermission.

7 7 roortional to the airrame surace area ustream o the ines [16]. The roulsive eiciency imroves with BLI because the et velocity relative to the light seed is reduced, as indicated in igure 5c. The change can be estimated using equation 7, once the new thrust requirement has been determined. Based on the reasoning described above, estimates o the uel burn beneits o BLI are included in Table 1 on the ollowing age. Eect o the installation on core size The core size o a quiet ine will be smaller than that o an equivalent current turboan because the byass ratio is higher. Thus the core comonents oerate in lower Reynolds number regimes and the ti clearance gas relative to the turbomachinery blade sizes are larger. The imact o this on comonent olytroic eiciencies can be aroximated rom the ollowing ormula based on inormation rom [17] and [18]: η oly Re 0.1 n (1 + BPR) n 0.1 (10) Equation 10 shows that, or a ixed thrust roulsion system, as the byass ratio or the number o ines increases, the size o the core reduces leading to an eiciency enalty. This eect becomes signiicant or a quiet roulsion system with more than 4 ine units. The beneits o a greater number o ines could be maintained without a reduction in core size, i a single core was used to drive multile ans in searate ducts. This solution may also be more ractical rom an economic oint o view in that there would still be ew ine cores to maintain. This coniguration is also roosed in [7]. Noise considerations The exhaust ducting o an embedded ine can be longer than that o an equivalent odded ine. However, the lth o the airrame centre body available or the ine installations is limited by structural requirements and the need or the intakes to be ositioned in low o an accetable Mach number. As a irst estimate this lth can be assumed to be constant or a given aircrat size. In this case the maximum lth-to-diameter ratio o the exhaust ducts will rise as the square root o the number o ines. Simle ray theory would argue that the number o relections, and thus the attenuation o the liners, should be roortional to the lth-todiameter ratio: duct Attenuatio n = l l duct n n (12) In ractice, liners are tuned to articular requencies, so the attenuation will only aly to a ortion o the total noise in a duct. Figure 9 comares acoustic redictions comleted or an exhaust duct with varying lth and a basic, single layer acoustic liner. Three lines are shown: the redicted attenuation increase or the comlete an noise sectrum, the redicted attenuation o the requency that the liner was designed to attenuate and the attenuation exected using equation 12. The lot shows that while the eectiveness o the acoustic treatment will imrove with the number o ines, the imrovement does not kee increasing linearly with lth-to-diameter ratio. However, it is exected that imroved increases in attenuation will be ossible with more advanced liner designs. Engine weight eects As ine an diameter increases the ine weight rises. The mass does not rise as the cube o an diameter due to the hollowness o arts, and comonents, such as an containment, which are more deendent on the an ti seed rather than diameter. Equation (11) shows a very simle estimate or ine weight variation that its reasonably well to available ine data, see also [19]. W ( n ) 2. 4 n = n (11) This suggests that or a ixed overall an diameter, the overall weight will reduce as the number o ines is increased. The reduction exected is relatively small or ultra high byass ratio ines and a larger change is exected to come rom the reduced airrame structure needed to suort an embedded ine. Embedded ines do not require a ylon and this can account or as much as 20% o the total roulsion system weight [19]. Figure 8: Variation in attenuation o rearward an noise with exhaust duct geometry. Increasing the number o ines will also increase the blade assing requency. This is a result o the shat seed increasing, which at a ixed thrust should also rise as the square root o the number o ines. A higher blade assing requency will shit the entire sectrum o Published by the American Institute o Aeronautics and Astronautics, Inc., with ermission.

8 8 an noise uwards in requency. This will make the noise roduced by the an easier to attenuate by acoustic liners, increase the eectiveness o shielding by the airrame and increase the atmosheric attenuation. Comarison o installation otions Using the simle qualitative analyses above, several otions or the Silent Aircrat roulsion system installation were comared in terms o their exected imact on uel burn, weight and rearward turbomachinery noise. The results are summarised in table 1, which shows changes relative to an advanced next generation turboan (2005 design). No allowances are made or any imrovements in ine comonent technology over time. Tye n PR. sc X n m uel W Noise PO 2 1-6% +9% +3% +12% - BL % +8% +11% -7% -4 db BL % +2% +9% -19% -8 db BL % -6% +4% -35% -13 db BLI % -9% -2% -19% -8 db BLI % -20% -11% -35% -13 db Table 1: The eect o ine installation on cruise uel burn, ine weight and rearward noise. The results in the table are very aroximate, but they demonstrate some o the trade-os that are imortant in the choice o ine integration coniguration. The odded system is signiicantly larger than a current turboan, which leads to a uel burn enalty and greater weight. Embedding only two ines with boundary layer diversion gives a large thermal eiciency enalty due to the inlet ressure losses and minimal imrovement in the drag contribution. As the number o embedded ines increases, the installation drag reduces, but there is also a decrease in the core comonent olytroic eiciencies leading to uel consumtion levels that are still higher than current turboans. With boundary layer ingestion, the otential or signiicant thrust reductions are increased leading to overall beneits in uel consumtion. Table 1 indicates that the reerred installation or the Silent Aircrat should be embedded with boundary layer ingestion and multile ine ducts. Unortunately, this coniguration also carries the highest design risk, and it is exected to have high develoment and maintenance costs. The greatest risk to embedded ines is the imact o non-uniormity o the low at the ine ace. The distortion coeicient, C60, is the standard measure o total ressure non-uniormity or ine intakes and tyical C60 values measured or an embedded S-duct are in the range 0.10 to 0.30 [15]. This is a signiicant amount o low distortion, esecially or a low ressure ratio turboan, and it will aly at all oints in the light enveloe. The distortion imacts the ine erormance, stability, reliability and noise. BLI introduces additional total ressure distortion to the low entering the inlet. This distortion is non-uniorm, in both the radial and circumerential directions. It is in addition to that generated by the inlet duct and it is resent at all light conditions. Thus, the design risk osed by inlet distortion will be greatest or a boundary layer ingesting system. The challe is to realise the uel burn beneits o BLI without them being outweighed by the negative eects o the inlet distortion. VARIABLE CYCLE REQUIREMENT The roulsion system or an aircrat has very dierent requirements at dierent oints in the light mission. Cruise is the most thermodynamically demanding condition, because this is where most uel is consumed and there is the greatest need or high eiciencies. To-o-climb is the most aerodynamically demanding condition. At this oint, the ine has to maintain suicient thrust to kee the aircrat climbing at an altitude where the air is very thin. Take-o is the most mechanically demanding condition: The temeratures within the ine are highest and the risk o damaging the ine is greatest because the an is closer to instability and there are transient eects rom cross-winds and maneuvers that can initiate vibration. Current turboan designs manage to satisy all these requirements with a ixed cycle and ixed geometry coniguration. For the Silent Aircrat ines, take-o is also the most acoustically challing condition, because suicient thrust must be rovided without exceeding the noise target. The current section shows that this additional requirement means that the roulsion system needs an additional degree o reedom in its oeration, which can be rovided by a variable cycle. The to-o-climb oint is the key aerodynamic design condition because it determines the overall size o the ine. Consider the design o a turboan or ixed thrust requirements and ambient low conditions. From conservation o energy the an ressure ratio (neglecting inlet and exhaust losses) determines the et velocity: 1 γ 2 ( FPR ) = c T γ 2 V (13) Published by the American Institute o Aeronautics and Astronautics, Inc., with ermission. V This et velocity sets the an-ace area required through continuity, and this can be geometrically related to the overall diameter: n A X = V c T N 02 V0 Qa 02 0

9 9 thus, n 4X = π N 2 ( 1 HTR ) 1 V V 0 c T Q a (14) In equation 14, once the to-o-climb an ressure ratio has been chosen, all o the terms on the right hand side are ixed by the aircrat mission requirements, excet the an caacity at to-o-climb, Qa, and the design hubto-ti radius ratio, HTR. The an low caacity, Qa, is highest at to-o-climb, because the low through the ine must be maximised to achieve the thrust requirement. The choice o this is crucial to the design: a high low an will reduce an diameter and lower the low diusion in the intake, however, reducing the an low will reduce the an seed and tend to imrove an eiciency and stability. It is exected that a uture quiet ine will have a similar maximum an caacity to today s turboan designs. The hub-to-ti radius ratio is minimised subect to stress level limits in the an root and disc system. A uture quiet ine is exected to have a HTR slightly lower than current turboans. With the total an size ixed by the to-o-climb condition, now consider the ine at take-o. The thrust equation (2) can be rearranged to show how the et velocity at take-o deends on the light seed, ambient conditions, thrust requirement and ine low caacity: V c T 02 = V0 c T 02 + n X A N 02 1 Q a, T / O (15) In this equation, the caacity, Q,a,T/O, is based on the total mass low through the ine exhaust during takeo. For a ixed take-o condition, all the other terms on the right hand side o the equation are already determined. Equation 1 shows how et noise deends mainly on the et velocity. Thus, to design a turboan or minimum et noise, equation 15 shows that we need a coniguration that roduces the maximum ine low caacity at take-o with the an sized at to-o-climb. For a ixed geometry, ixed cycle turboan, the ine low caacity at take-o is determined by the exhaust nozzle area. However, by using some orm o variable exhaust geometry, the ine caacity at take-o can be increased signiicantly. This idea is demonstrated by igure 9 which shows the working lines or a an with a design ressure ratio o This indicates the wide searation between the to-o climb and take-o working lines or a ixed geometry system. This arises because the nozzle is choked at high altitude and light Mach number, but un-chokes at low seed and asses less low. Pressure ratio or quiet take-o Fixed nozzle take-o oint To o climb oerating oint << maximum an caacity Variable nozzle take-o oint Figure 9: Fan working lines showing oerating oints with and without variable exhaust. The an ressure ratio required to meet the Silent Aircrat et noise target at take-o is below 1.2. Thus, with a ixed exhaust, the take-o oerating oint is close to the surge line with low low caacity. A variable exhaust nozzle allows this oerating oint to be moved away rom instability and the an caacity can be increased to the value at to-o-climb. Using equations it is straightorward to consider the oerating oints at to-o-climb and take-o or a series o ine designs each with dierent design an ressure ratios. Figure 10 lots the resulting variation o et noise, as calculated with the Stone Jet noise model [11], against the overall an diameter or a ixed aircrat mission and varying design an ressure ratio. The lot shows that a variable exhaust gives a signiicant et noise reduction at a given an diameter (~10dBA), or or the ixed noise target o the Silent Aircrat, it enables a reduction in an diameter o about 20%. Similar reductions in ine size can be obtained with a 2:1 area ratio eector deloyed at take-o. This coniguration is discussed urther under candidate ine conigurations. Figure 10: Variation o et noise with an diameter using dierent exhaust systems or a 250-seat aircrat Published by the American Institute o Aeronautics and Astronautics, Inc., with ermission.

10 10 This lower overall an diameter reduces the drag o the nacelle during cruise (igure 6) and imlies a higher an ressure ratio, which will also reduce the imact o inlet ressure losses on the overall eiciency (igure 7). In addition, a variable exhaust can be adusted to enable the an to oerate at eak eiciency or a given cruise thrust requirement. Overall this is exected to lead to signiicant uel consumtion savings. The an system will be more stable, because the exhaust area can be adusted to avoid an conditions rone to aeromechanical vibration. All o the beneits o a variable exhaust system can be alied equally to any o the installation otions considered in table 1, however, variable geometry will be easier to accommodate within an embedded coniguration. For an embedded system with multile ine units, the ines could have a common, twodimensional variable geometry exhaust system, which may oer additional beneits in terms o lower weight and reduced comlexity. CANIATE ENGINE CONFIGURATIONS The datum candidate ine or the Silent Aircrat is a conventional turboan ine with the an driven via a reduction gearbox and a large variable exhaust nozzle (igure 11). The gearbox allows a low-seed, quiet an to be driven by a high-seed, low-weight and low-noise turbine. The variable geometry nozzle is oened u during take-o and aroach and closed down at cruise, giving imroved oerating oints, as shown in igure 9. Large, geared an Variable exhaust nozzle side view As the eector area ratio increases, the exhaust low caacity in equation 15 increases, although their otential is limited by the mixing eiciency that they can achieve. Eectors can also be used in combination with a variable nozzle to enhance their eect. The two-stage at contrarotating an arrangement shown in igure 12 has been examined in several studies, which have shown that it could rovide weight and cost beneits. However, the eector could also be used with the conventional ine architecture shown in igure 11. Contra-rotating at an Eector duct side view Eector closes or cruise Figure 12: Embedded at contra-an ine with exhaust eector ducts A large, low-seed et area can also be roduced by having extra ans that are only oerating at take-o (igure 13). These ans can be driven by the main ine but only exosed at low altitude. Thus, their design can be otimised solely to minimize take-o and aroach noise whilst the main ine could be designed to have the best ossible cruise erormance. Take-o, lownoise an view rom above Small ine core Cruise ine Figure 11: ucted high diameter ine with geared an and variable exhaust nozzle A second otion, shown in igure 12, is to have a conventional et ine that uses devices called eectors or take-o and landing. Eectors are ducts outside o the ine exhaust that entrain additional air into the exhaust low, thus increasing the mass low and reducing the mean et velocity. The eectors can be stowed at cruise to remove their drag eect and they do not have to be circular, which makes them more amenable to an embedded system. Figure 10 includes a lot o the variation in et noise with an diameter or an eection system that doubles the eective exhaust area at take-o. Published by the American Institute o Aeronautics and Astronautics, Inc., with ermission. Gearbox and transmission Figure 13: Otimised cruise ine with auxiliary ans or take-o and aroach

11 11 CONCLUSIONS AN FURTHER WORK The main indings o the studies in this aer can be summarised with the ollowing oints: 1. For low noise, a large exhaust et area at take-o is required. The area needed or the Silent Aircrat ines can be reduced signiicantly by using ower managed dearture rocedures, but it will still be 2-3 times larger than that o existing turboans. 2. Embedding the ines into an all liting body aircrat can reduce the installation drag, increase noise attenuation and enable boundary layer ingestion. However, total ressure losses ustream o the ine have a signiicant detrimental imact on erormance and there is a large, uncertain risk to the design rom the eects o inlet low distortion. 3. A greater number o embedded ines are exected to give lower weight, reduced drag contribution and more eective noise attenuation. However, smaller ine cores will have lower thermal eiciency. 4. A variable geometry exhaust system allows a smaller, low-weight ine that can be quiet at low altitude and eicient at cruise. These beneits have been demonstrated with a variable area exhaust nozzle, and other conigurations have been roosed. Work is underway to develo roulsion system designs or the Silent Aircrat based on the ine and installation conigurations roosed in this aer. The dierent designs will be assessed in detail to determine their noise emission, weight and erormance. The datum design is an embedded 4-ine system with a variable exhaust nozzle and boundary layer diversion. This will be ollowed by designs with boundary layer ingestion, a greater number o ine ducts and alternative variable exhaust systems. The studies in this aer have identiied several key challes to reaching the technical obectives o the Silent Aircrat roulsion system. These are being addressed in the ollowing research activities that are already underway as art o the Silent Aircrat Initiative: 1. The design o embedded intakes that deliver low to a an with minimum losses and minimum nonuniormity throughout the light enveloe. 2. The develoment o robust, low-weight ine architectures that generate low turbomachinery source noise. 3. The design o extended exhaust ducts that minimise rearward roagating noise using advanced acoustic liner technology. 4. The develoment o eicient, low-noise, an systems that are tolerant o inlet distortion and comatible with a variable exhaust. 5. The study o the imact o boundary layer ingestion on roulsion erormance, noise and reliability. This is needed in order to show that the uel burn beneits o boundary layer ingestion can be realised desite the inherent ractical roblems. ACKNOWLEGEMENTS The authors would like to acknowledge the Cambridge-MIT Institute or the inancial suort o this research through the Silent Aircrat Initiative roect. They would also like to thank Rolls-Royce lc or access to their reliminary design methods and or technical advice given during the course o this research. Several other members o the Silent Aircrat team have also contributed to the work that has made this aer ossible. Particular thanks to Anurag Agawal, Mark rela, Chris Freeman, Patrick Freuler, Ed Greitzer, Geo Hodges, Tom Hynes, Tom Law, Vahid Madani, Matthew Sargeant, Zoltan Sakovszky and Liing Xu. REFERENCES [1] "Air Travel - Greener by esign: The Technology Challe," Greener by esign, [2] "Review o the quota count (QC) system," eartment or Transort, UK Government, [3] ACARE, "Euroean Aeronautics: A vision or 2020," Advisory Council or Aeronautics Research in Euroe, [4]. L. aggett, S. T. Brown, and R. T. Kawai, "Ultra-Eicient Engine iameter Study," NASA CR , [5] P. R. Gliebe and B. A. Janardan, "Ultra-High Byass Engine Aeroacoustic Study," NASA Oct 2003, [6]. E. Crow, "A comrehensive aroach to ine noise reduction technology," resented at ISABE- 2001, Bangalore, India, [7] A. Manneville,. Pilczer, and Z. S. Sakovszky, "Noise reduction assessments and reliminary design imlications or a unctionally-silent aircrat.," resented at 10th AIAA/CEAS Aeroacoustics Conerence, Manchester, UK, [8] M. J. Lighthill, "On sound generated aerodynamically (Part 1: General theory)," Proceedings o the Royal Society o London. Series A: Mathmatical and Physical Sciences, vol. 211, , [9] C. K. W. Tam and K. B. M. Q. Zaman, "Subsonic Jet Noise rom Nonaxisymetric and Tabbed Nozzles," AIAA, vol. 38, , Published by the American Institute o Aeronautics and Astronautics, Inc., with ermission.

12 12 [10] N. H. Saiyed, K. L. Mikkelsen, and J. E. Bridges, "Acoustics and Thrust o Searate-Flow Exhaust Nozzles With Mixing evices or High-Byass- Ratio Engines," resented at the 6th Aeroacoustics Conerence, Lahaina, Hawaii, [11] J. R. Stone and F. J. Montegani, "An Imroved Prediction Method or the Noise Generated in Flight by Circular Jets.," NASA TM-81470, [12]. Crichton,. Tan, and C. Hall, "Required Jet Area or a silent aircrat at take-o," resented at the 8th ASC-CEAS Worksho, Budaest University o Technology and Economics, Hungary, [13] R. Liebeck, "esign o the Blended Wing Body Subsonic Transort," Journal o Aircrat, vol. 41, , [14] N. A. Cumsty, Jet Proulsion: A simle guide to the aerodynamic and thermodynamic design and erormance o et ines. Cambridge, UK: CUP, [15] A. J. Anabtawi, R. F. Blackwelder, P. B. S. Lissaman, and R. H. Liebeck, "An Exerimental Study o Vortex Generators in Boundary Layer Ingesting iusers with a Centerline Oset," University o Southern Caliornia, Los Angeles, CA, USA, [16]. L. Rodriguez, "A Multidiscilinary Otimization Method or esigning Boundary Layer Ingesting Inlets," Stanord University, Stanord, CA, USA, [17] C. Freeman, "Personal communication on ine design odded versus embedded," [18] J. Kurzke, GasTurb 10, [19] J. Protz, "Engine Models," Massachusetts Institute o Technology, Cambridge, MA, USA, Published by the American Institute o Aeronautics and Astronautics, Inc., with ermission.