REVIEW OF PROPELLER-WING AERODYNAMIC INTERFERENCE

Transcription

1 24 TH INTERNATIONA CONGRESS OF THE AERONAUTICA SCIENCES REVIEW OF PROPEER-WING AERODYNAMIC INTERFERENCE..M. Veldhuis Delft University of Tehnology Keywords: roeller-wing interation, numerial modeling, exeriments. Abstrat The main goal of this aer is to review the mehanisms and desribe the henomena that lay a role in the aerodynami interferene between trator roellers and a wing. Moreover, the effet of various arameters like the roeller osition and inlination will be disussed in detail. Besides this means to analyze roeller-wing design will be resented using alulation tehniques of distint omlexity. Rather than an evaluation of the omlete design enveloe of a roeller owered onet, one tyial aset is investigated herein: the aerodynami asets of the roeller integration for the ruise hase of the airraft. Symbols a axial inflow fator ( = v a U ) a ' tangential veloity ratio ( = v t U ) b wing san C drag oeffiient D C lift oeffiient mean aerodynami hord D roeller diameter G roeller geometry influene funtion I Wing interferene influene w funtion J roeller advane ratio ( U nd) P ower oeffiient ( 2 π Q / J ) Q roeller torque oeffiient n roeller seed R roeller radius Re Reynolds number T thrust oeffiient () v a axial veloity inrease v t tangential veloity inrease x, yz, oordinates in flow axis system x, y, z roeller osition (entre of sinner) α angle of attak α roeller inidene angle relative to the wing referene line α eff effetive roeller angle of attak roeller blade angle at 0.75R β0.75r η ω Indies orr tot w roulsive effiieny roeller rotational seed orreted (due to) roeller total (due to) wing 1 Introdution Modern airraft onets, like the Euroean A400M, exhibit high disk loading and a high number of (swet) blades to enable high ruising seed. The strong swirl veloities in the slistream ombined with inreased dynami ressure generate a onsiderable deformation of the lift distribution, whih has an imat on the aerodynami behavior and erformane of the wing. The wing loading in turn indues a disturbed inflow field for the roellers, eseially in the ase where the roeller and 1

2 ..M. VEDHUIS the wing are losely ouled. Hene the aerodynami interferene for tyial trator roeller wing airraft may be summarized as roeller effets on the wing and vie versa. The desrition of the interative flow around the roeller-wing onfiguration requires detailed information about the harateristis of the slistream. Due to the self-indued veloities rodued by the roeller vortex system the slistream tends to deform and roll u whih rodues a so-alled slistream tube with strong gradients in various flow quantities both in streamwise and radial diretion. In ase of an asymmetrial loaded roeller, for examle aused by a non-zero angle of attak of the thrust axis ( α 0 ), a variation of the flow eff quantities in azimuthal diretion exists. Summarizing one may state that the distribution of the axial veloity ratio, denoted with a, the tangential veloity ratio, denoted with a ', and the total ressure distribution, are a funtion of the roeller geometry ( G ), blade setting ( β0.75r ), oerating onditions ( J ), effetive roeller angle of attak (α eff ) and the (interferene) effet of the wing on the flow around the roeller ( I w ): (, ) = a(, β0.75r,, α, ) eff w '(, ) = t(, β0,75r,, α, ) eff w t(, ) = (, β0,75r,, α, ) eff w a x r f G J I a x r f G J I x r f G J I where the oordinates ( x, ) (1) r are taken in the referene system fixed to the thrust axis. The fat that this thrust axis an have any osition in sae means that the relevant flow roerties in the flow fixed referene system, ( x, yzexhibit, ) satial distributions that without any form of symmetry. For a seleted ruise ondition the arameter G, β 0.75 R and J are fixed. This means that the roblem in the sense of the roeller wing interferene is found in the deendeny of α eff and I w on the roeller osition relative to the wing and the airraft state. A simle solution of the roblem is hindered by the fat that the latter arameters in turn are influened by the form and osition of the roeller slistream. Hene the erformane of the omlete roeller-wing ombination will only be attained by aeting a full interation between roeller and wing whih will be denoted as FIM (full interation mode). Nevertheless many researhers have aeted the single interation mode (SIM), in whih the wing effet on the roeller is simly negleted. As will be shown in subsequent setions this SIM aroah obstruts the analysis of the roeller osition effets on the roulsive effiieny of the onfiguration. Although the roeller exhibits a tyial unsteady flow field is has been shown by several authors [1,2,3,4,5] that for most ratial design alulations it is aetable to treat the flow as being steady. This timeaveraged aroah will be adoted during the subsequent analysis of the roeller-wing interferene roblem. 2 Regions of influene The slistream roerties hange throughout the loal flow field resulting in a strong deformation of the wing loading distribution. In this reset mainly the hanges in radial diretion and the streamwise develoment of the roeller slistream must be taken into aount. To desribe the most imortant interferene effet it is benefiial to slit the wing and the roeller ro off ro on (inboard u) W-I W-II W-III W-IV P-IV P-I P-III P-II ro on (inboard u) Fig. 1 Influene areas related to roellerwing interation based on the loading distributions. 2

3 REVIEW OF PROPEER-WING AERODYNAMIC INTERFERENCE in several regions of influene, as skethed in Fig. 1. u going blade α Z 2.1 Wing regions Wing regions, W-II and W-III are diretly influened by the slistream that washes the wing. In W-II the lift effet of the roeller swirl veloity, that hanges the loal wing angle of attak, is enhaned by the inreased dynami ressure. Considering the IU rotation ase, in W-III these two slistream effets ounterat eah other. The result is a smaller differene between the owered and unowered ase in this region. It an be learly seen that the roeller effet is not limited to the wing art (with a san equal to the ontrated slistream diameter) diretly behind the roeller. Due to the hanged wing inflow onditions generated by the roeller the loading in W-I and W-IV hanges as well, both for the inboard and outboard u running roeller. This is the result of the distorted vortiity sheet that leaves the wing. 2.2 Proeller regions To understand the wing effets on the roeller, 4 regions of influene an be defined (see Fig. 1). One should onsider that these regions, loated at azimuthal ositions of θ = 0,90,180 and 270 are in fat not omletely searated. Rather, a gradual hange of the slistream roerties is found going from one region to a neighboring one. The effet of the resene of the naelle is a small axial veloity inrease in all four P- regions. As suh the detailed naelle effet on the roeller (whih is quite small) is left out of the disussion on roeller-wing interferene here. Tyial differenes at P-II and P-IV are found due to the wing indued uwash. As skethed in Fig. 2 the loal blade angle of attak inreases at the downgoing blade side (P-II, for IU-rotating roeller) and dereases on the oosite side (P- IV). The result is a loading asymmetry in the slistream that has to be aounted for in the roeller-wing interation model. α V V x Y V z X The differenes in the indued axial and tangential veloities found for P-I and P-III is attributed to the wing indued axial veloity inrease and derease for the high and low roeller blade osition resetively. 3 Swirl reovery An imortant faet in the alulation of the slistream-indued veloities with simle models is the redution of the rotational veloity in the slistream due to the wing. Both exerimental and numerial studies have shown that there is a signifiant redution in rotation (swirl veloity) due to the resene of the wing. Various windtunnel tests have indiated that the amount of the redution in the rotational veloity deends on numerous fators like the roeller osition relative to the wing, the ower setting, the wing loading and so forth. It should be noted that while there is some redution in rotational veloity due to frition and visous effet, it is more likely that a hange in the slistream helix angle is the main ause for the redution in the rotation in the rotational veloity. From a onetual oint of view the redution in the slistream helix angle an be attributed to the wing indued uwash (in front) and downwash (behind). The wing is assumed to redue the angle of rotation of the slistream within those annuli that wash over it. In subsequent aragrah it will be shown that it is of vital imortane to imlement a Swirl Reovery Fator (SRF) in the simle stati slistream models to arrive at aetable alulation results. ω r down going blade ω r Fig. 2 Blade angle of attak variation due to roeller ith angle. V x V z α V z V x 3

4 ..M. VEDHUIS 4 Analysis methods The data resented in this aer were rodued by CFD tehniques of different omlexity ombined with a number of exeriments. The main goal is to twofold: a) identify the influene of various flow arameters and onfiguration adatations on the roeller-wing harateristis and the roulsive effiieny, b) determine to what level of omlexity the redition odes should be develoed to aquire aetable data that an be used in the reliminary design of roeller airraft. The numerial alulations that are desribed herein were erformed at two levels: an enhaned vortex lattie ode inororating a Blade element Method (BEM)-analysis of the roeller a ommerial Navier-Stokes ode The exeriments were erformed using three different windtunnel models: a straight wing-naelle-roeller model, denoted PROWIM a straight wing model with searate (movable) roeller, denoted APROPOS a full 3D, 1:20 sale model of a tyial turboro airraft (F27) 4.1 Predition odes VM-BEM model Although this method onstitutes a rather rude model of the real flow around the wing its simliity allows some interesting analyses to be erformed. The redued alulation time that is tyial for the method allows a quik survey of various onfiguration layouts. Adequate desritions of the VM tehnique an be found in oen literature [6,7]. The effet of the roeller on the wing loading harateristis is based on the fat that at the ontrol oint osition an additional veloity vetor, V, indued by the roeller slistream is introdued. The differenes between earlier VM-odes are tyially found in the way these additional veloity omonents are alulated without omromising the original VMaroah. In most ases, only a one-way interation is imlemented, e.g. only the roeller effet on the wing (further denoted as single-interation-mode or SIM) is modeled. The method suggested in this reort emloys a full interation allowing the wing effet on the roeller as well (full-interation-mode or FIM). Adatation of the undisturbed veloity ontribution is required both on the lattie elements inside the slistream tube and outside the slistream tube sine the loal flow angles outside the slistream are hanged by the ontration (and deformation) of the slistream. For the VM models that were ublished u till now, however, this adatation outside the slistream tube was negleted. This leads to erroneous results when extreme ositions of the roeller relative to the wing, like an over-thewing arrangement are seleted. The VMresults desribed herein are based on the FIM aroah taking into aount the satial develoment of the slistream harateristis, inluding a swirl reovery fator SRF NS-model The Navier-Stokes ode that was used for the roeller-wing alulations is the ommerial Fluent 6.1 akage [8]. To model the roeller an atuator disk aroah was adoted whih resribes the jum onditions at the roeller lane. Due to the asymmetry in the roeller loading the ressure and veloity are deendent both on the radial osition and the azimuthal osition of the ell faes that onstitute the infinitesimal thin roeller lane [9,10]. 4.2 Windtunnel models PROWIM and APROPOS The first wind tunnel model, denoted PROWIM (roeller wing interferene model), onsists of a straight wing of aset ratio A = 5.33 with no twist, onstant hord and airfoil setion (NACA A015). Its (half) san is 0.64m. For the owered tests, the model is equied with a 4- bladed metal roeller of 36m diameter that is driven by a 5.5kW eletrial 3-hase indution motor ontained inside the naelle. The axis-symmetrial naelle is mounted with its rotation axis on the MAC-line and at 0.3m from the wing root. The dimensionless sanwise 4

5 REVIEW OF PROPEER-WING AERODYNAMIC INTERFERENCE Dimensions Dimensions in in mm mm refletion late naelle roeller rows with ressure tas naelle traversing diretions suort strut =inboard fla 2=outboard fla Fig. 4 Dimensions and layout of the PROWIM model. Fig. 3 Dimensions and layout of the APROPOS model. roeller osition is y / b / 2 = 69. A sketh is resented in Fig. 4. The seond windtunnel model referened as APROPOS (Fig. 3) is idential with PROWIM exet for the fat that the naelle is detahed from the wing. The naelle is suorted by a strut whih an be traversed with the 3- omonent traversing system mentioned earlier. In the APROPOS layout the searate effet of the roeller osition relative to the wing was investigated. 640 Fig. 5 Fokker F27 sale model in the Delft University low seed windtunnel F27 airraft model This model is a 1:20 sale model of the rototye of the Fokker F27 airraft (Fig. 5). The model onsists of several arts that an be detahed for detail tests. The san of the high aset ratio wing ( A = 12) is1.45m. 5 Results The results of some alulations and exeriments will be disussed on the basis of the geometrial asets of the onfiguration as well as the wing and roeller modeling tehniques emloyed in the analysis. Where ossible, the numerial results will be omared diretly with the exerimental results. 5.1 Proeller rotation effet As suggested by Veldhuis [9], Miranda [1] and Kroo [11] the roeller inboard u (IU) rotation is benefiial sine the wing the magnitude of the lift vetor at the naelle inboard side that is tilted forward is higher than the bakward tilted vetor at the naelle outboard side. This effet was onfirmed by our alulations and exeriments as indiated in Fig. 6 to Fig. 10. In Fig. 6a, a small lift inreased an be notied for the PROWIM model due to the roeller inboard u running roeller slistream running against the wing ti vortex. More ronouned however is the effet on the (effetive) drag oeffiient, C resented in Fig. 6b. ' D 5

6 ..M. VEDHUIS inb-u o-right outb-u 0.54 C D ' inboard u outboard u α (deg) (a) inboard u outboar u Fig. 6 Effet of roeller rotation diretion on the lift oeffiient (a) and the drag oeffiient of PROWIM ; J = The ositive ower effet on the lift is found also for the F27 airraft model (Fig. 7) ro off inb-u out-u windmilling α (deg) Fig. 7 Effet of roeller rotation diretion and wind milling on the lift urve of the F27 model; 6 Re = 1 10 ; T = ; β 0.75R = α=5 deg Re=5 million Clean J Fig. 8 Effet of the roeller rotation diretion and the advane ratio on the lift oeffiient of the F27 model ; α = 5. Even if the roeller's distane from the wing tis is large for the F27 model its influene on the effetive aset ratio of the wing aarently is still notieable. The differene of both the ro-on ases with ro-off ase again arises from the ontribution of the inreased dynami ressure in the slistream area and the roeller normal fore. As exeted the wind milling ase (o-right) rodues the lowest lift urve sloe due to the strong dynami ressure loss in the slistream. The effet of the thrust oeffiient for a onstant angle of attak of α = 5 is resented in Fig. 8. Although the effets are relatively small the outboard u (OU) rotating roeller learly shows the lowest in rease in the lift oeffiient. The effet of the rotation diretion on the liftdrag olar is resented in Fig. 9. It should be noted that the drag oeffiient, along the x- axis, inororates the thrust. Therefore negative CD -values are found. The effet of the roeller rotation diretion on the drag oeffiient onfirms the (small) benefiial effets of the inboard u rotating roeller. At a tyial ruise lift oeffiient of C = the differene between the outboard u and the inboard u rotating roeller is 4 drag ounts in favor of the latter. The differene between the two roeller rotation ases deends on the lift oeffiient. One may exet that with inreasing lift ating on the wing, the relative effets of the oosite swirl distribution introdue larger hanges in the drag. Fig. 10 shows that the ositive influene 6

7 REVIEW OF PROPEER-WING AERODYNAMIC INTERFERENCE Re=1 million T =0.127 Clean of the inboard u rotation indeed inreases for the higher C values. C D =4 ounts The effet of the rotation diretion was investigated also with the enhaned VM-ode. The alulations were based on a Fokker 50 like onfiguration. The most imortant results are summarized in Table 1. The data are analyzed based on the roulsive effiieny that is defined as: CD η tot = (2) P Here C D tot is the total drag oeffiient orreted for the thrust of the roeller and P is the ower oeffiient. C D inb-u outb-u Fig. 9 ift-drag olar for the F27 at T = in the lean onfiguration. (C D ) inb-u -(C D ) outb-u (ounts) Re=1 million T=0.127 Clean Fig. 10 Differene in drag oeffiient for the inboard u and the outboard u rotation ase ; F27 in lean onfiguration. Table 1 Influene of roeller rotation diretion on the lift oeffiient and the effetive roulsive effiieny; Fokker F50 onfiguration. High seed ase Rotation diretion C ( η ) orr inboard u o-right outboard u ow seed ase Rotation diretion C ( η ) orr inboard u o-right outboard u The IU ase indeed rodues the highest roulsive effiieny for both a tyial low seed and a high seed flight ondition but the magnitude of the effets is rather small. The roulsive effiieny for the IU ase is only 3% higher than for the (onventional) CRase for the high seed (ruise) ondition. For the low seed ase this even redues to 0.14% in favor of the IU-onfiguration. These values are onsiderably lower than exeted from earlier exeriments and alulations on otimized wing geometries [12]. Nevertheless, for wing onfigurations with a muh lower aset ratio or with a wider roeller (i.e. higher value of D/ b) the differenes between the IU and the OU ase are likely to inrease. 5.2 Streamwise roeller osition By hanging the roeller streamwise osition referened to the wing loation the aerodynami ouling between the two elements hange. The swirl veloity maintains its value obtained diretly behind the roeller lane while the axial veloity at the wing inreases onsiderably with inreasing distane, x, between the roeller and the wing. Ref. [13] resents results of windtunnel test that were on a full 3D-airraft model with the roeller loated at 5 and 0 ahead of the wing leading edge. It was found that the installations with the roeller lose to the wing were more effiient than the onfiguration. To 7

8 ..M. VEDHUIS η unorreted onstant lift, C=1.204 result is a hange in the wing lift and in the drag distribution that diretly influenes the effiieny of the roeller-wing onfiguration. It should be remembered here that in these alulations the full interation between the roeller and the wing was modeled (FIM). 000 ow seed ase (HSC) 1.02 verify these findings the effet of different streamwise ositions was analyzed for the Fokker 50 like onfiguration, further denoted as Model-1,using the VM-method. Both a tyial high seed ase ( J = 1.63, T = 46 ) and a low seed ase ( J = 1.00, η x /R High seed ase (HSC) unorreted onstant lift, C= x /R Fig. 11 Proulsive effiieny versus the roeller streamwise osition of Model-1 for a tyial low seed and high seed ase taking the full interation of the roeller and the wing into aount. T = 51) were used for this analysis. Evidently the streamwise osition of the roeller affets to some extend the roulsive effiieny,η, as indiated by Fig. 11. The oen symbols indiate the trend in ase the roeller osition is hanged without taking notie of hanges in the wing lift, while the line with the losed symbols show the data oints that were orreted to a onstant lift oeffiient. When the roeller is brought loser to the leading edge of the wing, the uflow in the roeller lane hanges. This in turn leads to a distortion of the veloity distributions in the slistream. The /( ) x/r=2 C D /(C D ) x/r= C D /(C D ) x/r=2 HSC SC x /R (a) HSC, indued drag HSC, rofile drag SC, indued drag SC, rofile drag x /R HSC SC x /R () Fig. 12 Effet of roeller streamwise osition on the wing of model 1 for the low seed ase (SC) and the high seed ase (HSC) ; (a) the lift oeffiient ;, the drag oeffiient omonents ; () total drag oeffiient 8

9 REVIEW OF PROPEER-WING AERODYNAMIC INTERFERENCE The soure for the slightly higher roulsive effiieny with inreasing distane to the wing E an be found in the higher lift oeffiient due to the augmented dynami ressure at greater distanes from the roeller disk. This effet is deited in Fig. 12a where the ratio between the lift oeffiient for a given roeller osition and the one found at x / R = 2.0 is given. Although the slistream veloity distributions are quite different for the low seed ase (SC) and the high seed ase (HSC) the ratio C /( C ) x / R= 2.0 hanges almost identially with x. The hange in the drag oeffiient with x (Fig. 12b) is slit into the two main ontributors: rofile drag and indued drag. The rofile drag omonent exhibits the same behavior for both thrust ases in that inreasing x results in higher CD values due to dynami ressure effets (inreased "srubbing drag"). However the indued drag runs in a different way when the roeller aroahes the wing. In the low seed ase the thrust fore is relatively high and the roeller indued veloity omonents omared to the undisturbed veloity are higher than in the high seed ase. This aarently leads to stronger roeller indued angle of attak effets for the roeller that is in lose roximity of the wing. From x / R = 1.0 on both C D i urves show a negative gradient whih ould be antiiated sine due to the higher loal lift oeffiient, in the slistream washed area of the wing, stronger swirl reovery due to the resene of the wing ours. When the total drag oeffiient is alulated versus x a small hange is found, however. /CD /C D z/r= z/r=0 z/r= z/r= y /b/ z/r= z/r=0 z/r=1.017 (a) α = 1.05 z/r= z/r= y /b/2 α = 4.2 Fig. 13 Effet of the roeller sanwise loation, y, on the lift/drag ratio of the APROPOS wing for several vertial roeller loations, z ; J = 0.92 ; T = ; α = 0. The range over whih the roeller was translated was hosen rather wide to be able to identify the streamwise effet as omlete as ossible. For ratial reasons however the hoie for the roeller streamwise osition is onstrained by the sae needed for the engine in relation to the wing strutural layout. Therefore the variation of x is to be seen only as an "aerodynami test ase" without onerning the roblems related to the naellewing struture. Still, in a ratial range of x / R = a small effet on the roulsive effiieny is found. One may onlude that from the fuel onsumtion oint of view a roeller osition not to lose to the wing leading edge is benefiial. 5.3 Sanwise roeller osition The sanwise gradient lift distribution at the osition where the slistream washes the wing lays an imortant role with reset to the ossible erformane benefits introdued by the roeller. This was already shown by in earlier investigations [1,9,11]. The influene of the sanwise roeller osition was investigated using the APROPOS test setu. In Fig. 13 and Fig. 14 the effets of the sanwise osition is resented for a rather low thrust ondition. 9

10 ..M. VEDHUIS z /R=0 C CD y /b/ Fig. 14 Effet of the roeller sanwise loation, y, on the lift and the drag oeffiient of the APROPOS wing ; J = 0.92 ; T = ; α = 0 ; α = 4.2. The different urves in these figures orresond to different vertial ositions of the roeller, whih will be disussed in a subsequent setion. As exeted the erformane of the wing imroves when the roeller is moved in the diretion of the wing ti indiated by the inreasing wing lift / drag ratio. The reason for this hange in the lift / drag ratio an be found in a inrease in the lift oeffiient ombined with a onurrent redution of the drag oeffiient towards the ti, as indiated by the urves in Fig. 14. Aarently the vortiity field (swirl) indued by the inboard u rotating roeller attenuates the wing ti vortex influene. As a result, the effetive aset ratio of the wing is inreased whih leads to the lift inrease and drag derease. Tests at several angles an attak and vertial roeller ositions have shown that the effet at the wing ti is maximum when the slistream enterline is exatly in line with the wing ti vortex. Although the sanwise roeller osition strongly affets the erformane for ositions lose to the wing ti, it should be remarked that small hanges in y for realisti ositions ( 5 < y / b / 2 < 0.30 ) shows negligible hanges. The effet of the sanwise roeller osition was analyzed with the VM-ode based on Model-1. To kee the onfiguration more C D C l / av y /b/ y/b/2 Fig. 15 Sanwise loading distribution for three lateral roeller ositions; Model-1; high seed ase. realisti than onsidered in the exerimental amaign for APROPOS, the sanwise roeller osition was hanged over a small range of y/( b / 2) = 0 8 only. The sanwise loading distributions are resented in Fig. 15 while the harateristi oeffiients are given in Table 2. The small shift in the roeller osition is learly visible in the sanwise loading given as C l / C versus y / b / 2. Although the most outboard osition rodues a somewhat higher roulsive effiieny, as exeted, the differenes between the 3 ositions are too small to be of signifiane for the design roess. This result an be attributed to the fat that the lift distribution is rather flat for the given roeller osition. Table 2 Influene of sanwise roeller osition on the lift oeffiient and the effetive roulsive effiieny of Model-1. High seed ase y /( b / 2) C ( η ) orr ow seed ase y /( b / 2) C ( η ) orr

11 REVIEW OF PROPEER-WING AERODYNAMIC INTERFERENCE z 1 4 Fig. 16 Definition of the vertial roeller oordinate, z. 5.4 Vertial roeller osition Sine twin-engined turboro airraft show quite different vertial roeller ositions with referene to the wing, the z oordinate (Fig. 16) was hanged in the APROPOS tests to investigate the erformane effets C CD z /R (a) C D The tests showed that the vertial roeller osition has a remarkable effet on the erformane of the wing. For realisti sanwise loations of the roeller a high osition is benefiial with reset to the wing lift/drag ratio, mainly due to a lift enhanement indued by the ombination of dynami ressure inrease at the wing's uer surfae and ontration of the slistream whih leads to inreased flow angles of attak. In Fig. 17 the effet of the vertial roeller osition on the APROPOS lift and drag oeffiient is deited. Whereas only moderate hanges in C and C D our for the lower thrust oeffiient a more ronouned effet of the roeller slistream an be notied for the higher thrust ase. The loal drag minimum for the mid osition of the roeller is likely to be aused by the fat that for z values lose to zero a smaller art of the immersed art of the wing is washed by the slistream annulus that ontains inreased dynami ressure ( doughnut effet). This rinile is illustrated in Fig. 18. The inrease in the lift oeffiient that is found when the roeller is moved from negative to ositive Z values, is attributed to the effet of the ontration of the slistream. For high z / Rvalues this results in a loal wing angle of wing A area with max. dynami ressure inrease Area with max. dy C D Proeller disk Z =0 =0 C CD z /R 1 Fig. 17 Effet of roeller vertial osition on the lift and drag oeffiient of the APROPOS wing atα = 4 ; (a) low thrust ( T = ) ; high thrust ( T = ) Fig. 18 Inrease in average dynami ressure inrease over the wing due to off-entre osition of the roeller ( z 0 ). Z <0 <0 11

12 ..M. VEDHUIS C/(C)z/R= HSC SC naelles. This in turn might then lead to exessive high naelle-wing interferene drag. The effets of the vertial roeller osition were examined by seleting 5 ositions saed 5R aart; assuming that no signifiant hanges would our for intermediate ositions. C D /(C D ) z/r= z /R attak inrease and a lift inrement; for the lower z / R values an oosite effet ours. Considering these results one an state that the wing may benefit from the resene of the roeller sine C / C D rises for the higher roeller ositions. The trends that were found exerimentally seem to onfirm the (limited) observations by other researhers that the rojetion of the roeller lane onto the wing strongly influenes the loal wing lift. To study the effets of the vertial osition further VM-simulations were onduted on Model-1 for a range of z / R = This range is somewhat beyond that found for tyial twin owered roeller airraft. It is obvious that values z / R > 5 lead to very low/high (a) HSC, rofile drag HSC, indued drag SC, rofile drag SC, indued drag z /R Fig. 19 Calulated effet of vertial roeller osition on the lift oeffiient (a) and the drag oeffiient of Model-1 alulated for the high seed and the low seed ase. Fig. 19 gives an imression of the effet of z on the lift and the drag oeffiient. The results are exressed in the form of a ratio between the values found at the seifi loation and z / R = 0. The first thing that an be notied is the minimum in the lift oeffiient found at z / R = 0. Aarently the overall dynami ressure inrease over the wing art immersed in the slistream inreases somewhat when the roeller is ut at an off-zero osition. The drag oeffiient is mostly affeted by a hange in the rofile drag omonent showing variations u to 5% while the indued drag alters not more than 1% for the given range of roeller ositions. Comaring the data of Fig. 19 with the earlier found data for PROWIM the VM-ode overestimates the lift oeffiient for the low roeller ositions. A ossible ause for this is twofold. Firstly the inflow into the slistream, whih determines the loal wing angle of attak, may be not modeled orretly. A seond ause for this disreany is very likely the result of the ode's inability to take the slistream deformation into aount. From the exerimental investigations on PROWIM a onsiderable deformation was found. Negleting this henomenon introdues different slistream veloities to the wing aarently ausing some asymmetry in the lift and drag distributions with referene to the z / R = 0 osition. Sine the effetive roulsive effiieny is diretly orrelated with the total lift and drag values, a more aetable value of η an be obtained by alying a orretion suh that the lift oeffiient resembles that of Fig. 17 i.e. lower values for negative z. The original and the (lift) orreted values of η are resented in Fig. 20. It is remarkable to notie that, in 12

13 REVIEW OF PROPEER-WING AERODYNAMIC INTERFERENCE η η High seed ase (HSC) unorreted orreted lift z/r ow seed ase (HSC) (a) unorreted orreted lift z /R Fig. 20 Unorreted and lift orreted values of the effetive roulsive effiieny versus the vertial roeller osition; high (a) and low seed ase for Model-1; VM. ontrast with the lift and drag distributions for both the unorreted and the lift-orreted ase no symmetry with reset to z / R = 0 is found. Eseially the high seed ase, whih is reresentative for the ruise ondition of the airraft, shows a onsiderable higher roulsive effiieny for the low roeller ositions. Again it should be made lear that the statements resented with reset to the vertial roeller osition are onsidered only from the roellerwing aerodynamis oint view. No onlusive remarks an be made regarding the best osition when both strutural and stability and ontrol asets are taken into aount as well. Searate investigations, inororating interferene between aerodynami and strutural loads are needed for this enhaned design analysis, whih fall outside the soe of this aer. 5.5 Proeller angle of attak effet Signifiant flow non-uniformity might be introdued in the roeller lane due to wing indued uwash as well the resene of the naelle(ositioned at a non-zero angle of attak). The fat that this has a negative effet on the roeller is another reason to exlore the ossible advantageous effets of a Proeller- Tilt-Down onfiguration, further denoted as PTD. Normally the angle of inidene given to a roeller is limited to say 2 whih effetively redues the yli loading due to the wing uwash generated skewed flow field. To exerimentally verify the ossible erformane effet of PTD onfigurations muh bigger angles were analyzed. Although all the tests were erformed for different y and z values ombined with several roeller inidene angles, α, only the exerimental results for α=4.2 deg C CD α (deg) (a) α=8.4 deg α (deg) C CD Fig. 21 Effet of roeller angle of attak on the wing lift and drag oeffiient of APROPOS;(a) α = 4.2 ; α = 8.4 ; J = C D C D 13

14 ..M. VEDHUIS alha=4.2 deg alha=8.4 deg 0.6 / C D 20 HSC wing lift roeller lift total α (deg) Fig. 22 APROPOS wing lift/drag ratio versus the roeller angle of attak relative to the wing hord referene line; J = y / b / 2 = 69 and z / R = 0 will be resented here. In Fig. 21 the lift and the drag oeffiient of the APROPOS model are given versus the roeller angle of attak, α. The tilting-down of the roeller evidently leads to imroved erformane of the wing through an inrease in the lift and a signifiant derease in the drag. This auses in a notable rise of the lift / drag ratio, as indiated in Fig. 22. Note that the underlying ause is on rinial different from the effet of vertial dislaement of the roeller sine now the veloity distribution in the slistream has undergone a major hange [9]. Furthermore the (small) negative normal fore ating in the roeller lane and the (small) redution in effetive thrust hardly redues the ositive effets of the PTD. The remarkable strong effets ofα were also analyzed with the VM-ode using the more realisti Model-1. Fig. 23 shows the build-u of the lift oeffiient. For dereasing roeller angle of attak the wing lift inreases slightly due to the uwash enountered by the wing immersed in the slistream α ro (deg) SC wing lift roeller lift total α ro (deg) Fig. 23 Comonents of the lift oeffiient for Model-1 versus the roeller angle of attak; (a) high seed ase; low seed ase. However, the diret lift fore that ats on the roeller (denoted 'roeller lift') lowers the total lift as a result of tilting down the thrust vetor and the negative roeller normal fore that is assoiated with negative α. The trend for both the high seed ase and the low seed ase is similar though the hanges due to α are somewhat stronger for the latter. To omare the α -effets on the lift and the drag, ratios were defined taking the values at α = 0 as a referene. Fig. 24 shows the results for C and C D, where the latter again is slit in the two searate ontributors: rofile drag and indued drag. 14

15 REVIEW OF PROPEER-WING AERODYNAMIC INTERFERENCE /( ) αro= HSC SC η C D /(C D ) αro= α ro (deg) HSC, rofile drag HSC, indued drag SC, rofile drag SC, indued drag α ro (deg) Fig. 24 Effet of roeller angle of attak on the lift (a) and drag ratios for Model-1. The values at α = 0 were taken as a referene. Small inrements in the lift oeffiients are found in the low seed ase with dereasing α while more signifiant effets are redited for the high seed ase. In the latter ase the most rominent ontribution is due to the hange in the wing lift. By lowering the advane ratio, as is done in the low seed ase, the relative ontribution of the roeller normal fore inreases leading to a smaller hange of with dereasingα. The iture for the drag oeffiient beomes more omlex sine the total drag fore is onstituted of 3 omonents: rofile drag, indued drag and roeller drag (in fat thrust) HSC SC α ro (deg) Fig. 25 Proulsive effiieny of Model-1 versus the roeller angle of attak for the high seed and the low seed ase. In Fig. 24b the drag ratio CD /( C D) α = 0 is shown for both flight ases. One more the effet of lowering the roeller angle of attak is more ronouned for the high seed ase. That is: the relative hange in the drag oeffiient is higher whih is of ourse to be artially ontributed to the lower absolute values of the drag oeffiient in this ase. The hange in the rofile drag oeffiient is negligible, whih ould be exeted from the fat that the average dynami ressure in the slistream is almost not affeted by hanges in the roeller angle of attak. The indued drag, on the other hand, diminishes exressively as a onsequene of the inreased uflow in front of the wing. As disussed earlier, the inreased uwash tilts the fore vetor ating on the wing forward whih moderates the indued drag. The ombined effets on the hanged lift and drag ontributions leads to favorable roulsive effiienies for low values ofα, as indiated by Fig. 25. In the high seed (ruise) ase η rises aroximately 9.5% by hanging α from 0 to 10. This erformane imrovement is interesting enough to be onsidered further in design studies of trator roeller wing onfigurations. 5.6 Numerial modeling omlexity 15

16 ..M. VEDHUIS The roblem of roeller wing interferene needs redition odes whih inororate all imortant asets as disussed above. To be able to determine to what level of omlexity the redition odes should be develoed to aquire aetable alulation results, the two numerial methods will be disussed briefly VM method The signifiane of omleting the full interation sheme, e.g. take into aount the roeller effets on the wing and vie versa, was assessed by omaring the final results of ste 1 (no wing effet on the roeller) and ste 2 (full interation) in the VM aroah. The rimary indiation whether the full interation is needed is found from a omarison between the lift distributions found for both situations. In Fig. 26 this is exressed in the form of a differene between the loal lift oeffiient found for ste 1 and ste 2, C l. Again the alulations were erformed for both the high seed and the low seed ase. Aarently a signifiant differene exists between the results of the alulation without interation and the full interation ase. The differenes are eseially found in the viinity of the roeller were the effet of the slistream is the most rominent. The fat that larger values of C l are found for the low seed ase is trivial sine for this ondition a higher lift is rodued by the wing whih auses a stronger deterioration of the inflow field of the roeller HSC SC From the data found it may be onluded that all roeller-wing alulations should be erformed in the full interation mode (FIM). The roulsive effiieny hanges 0.3% 1.0% for the high seed while a 0.7% 2.5% hange is found for the low seed ase. The suggestion, offered by some authors, that the wing effet on the roeller, in ase of the trator roeller loated well ahead of the wing, an be negleted should therefore be questioned. In Fig. 27 the lift oeffiients of two exeriments are omared with the VM-data. As an be seen an exellent agreement between the three methods is found. Aarently the VM-method redits the roeller effets on the overall lift oeffiient of the onfigurations balane wake survey alulated α (deg) ro on (a) α (deg) ro off balane wake survey alulated balane wake survey alulated Cl 12 C D ' y/b/2 Fig. 26 Differene between the loal lift oeffiient distribution of ste 1 (no interation) and ste 2 (full interation) alulation on Model-1 for the high (HSC) and the low seed ase (SC). 1 0 balane wake survey alulated α (deg) α (deg) Fig. 27 Comarison of exerimental and redited lift (a) and drag oeffiient (VM) of PROWIM. 16

17 REVIEW OF PROPEER-WING AERODYNAMIC INTERFERENCE aurately. In general an aurate value of the drag oeffiient is very diffiult to ature due to diffiulties in the estimation of the rofile drag. Nevertheless the VM-ode gives a reasonable redition of the total drag oeffiient, as an be seen in Fig. 27 The differenes omared to the exerimental values are likely to be aused by inauraies in the rofile drag data read by the VM-ode. The fat that a reasonable agreement is found suorts the onlusion that the tyial roeller effet; a drag rise in the roeller washed area due to dynami ressure inrease, is effetively envisaged by the ode. The aabilities of the VM-ode an be further be aknowledged by omaring the alulated sanwise loading with exeriments and other odes. In Fig. 28 some results for the PROWIM model are omared. The agreement is very aetable for both angles of attak and a lear deformation of the elliti-like shae as a result of the slistream swirl veloity omonent is found. The shar hanges in the exerimental lift distribution are softened to some extend by the ode, ossibly as a result of an inaurate alulation of the true slistream veloities (BEM analysis). It should be remarked that the VMalulations were erformed with an swirl reovery fator of about SRF = 0.5. As is found with many odes that do not inororate an SRF, ignoring the effet of swirl reovery leads to overestimated roeller slistream effets on the wing C l Cl y/b/2 (%) (a) exeriment vlm4d exeriment vlm4d y/b/2 (%) Fig. 28 Comarison of the exerimental and the alulated lift distribution from the PROWIM exeriment; roeller rotating inboard u; J = 0.85 ; T = ; (a) α = 0 ; α = 4. Additional verifiation of the reditive aabilities of the VM-ode were erformed by omaring results obtained from flight tests [14]. During these test the lift distribution over the wing was obtained by erforming ressure measurements in hordwise diretion using ressure belts. Two tyial flight onditions were onsidered to wit: a low thrust ase (TC) withα =, T ' = 0.63, J = 1.0 and a high thrust ase (HTC) whereα =, T ' = Fig. 29 shows the omarison between the exerimental values, the VM-aroah and the results obtained with an dediated Euler ode [15]. The VM ode shows a remarkable good agreement with the other data, inluding the loal effet of the naelles and the fuselage. Unmistakably the aliation of the SRF revents unrealisti swirl veloity effets in the wing art washed by the roeller slistream. The main disadvantage of the VM-method is unmistakably the limited level of detail that is obtained. Sine only integral harateristis are obtained the ossibilities to verify the flow henomena in detail with available data is imossible. 17

18 ..M. VEDHUIS Cl Euler alulation, ro on exeriment, ro on vlm4d, ro on y/b/2 (a) layer thikening and small areas of flow searation that were found in the exeriment, led to the onlusion that the intermediate ste in alulation omlexity (anel odes) is not interesting for an quik or aurate redition of roeller-wing interation effets. Hene the Navier-Stokes alulations were set u for the PROWIM model. As exeted, a muh better agreement with exerimental data were found. Some results are resented in Fig. 30 to Fig C l Euler alulation, ro on exeriment, ro on vlm4d, ro on y/b/2 Fig. 29 Sanwise lift distributions found with an advaned Euler ode [15] and the VMmethod omared with exerimental results [14] ; Fokker-50 in the low thrust ase ; (a) HTC ; TC. (a) NS-aroah The advantage of higher order alulations odes is to be found in the ossibility to determine the hordwise ressure distribution in more detail. However, some alulations on the PROWIM model that were erformed with a anel ode [16] revealed the roblem with orret definition of the relation between the roeller loading and the slistream indued flow roerties at the loation of the washed wing anels. The main ause for the disreany between the exerimental and the numerial results of anel odes is the omission of the visous effet (no deambering effet) and the lak of a swirl reovery fator (that must be set by the user). The anel ode's inability to redit the seondary flow henomena, like flow boundary Fig. 30 Calulated surfae ressure distribution of PROWIM; α = 0 ; (a) ro off; ro on. The ressure distributions in Fig. 30 learly show the effet of the roeller slistream that washes the wing. Eseially the imat of the swirl veloity omonent is very ronouned. The aability of the NS-ode to inororate the 18

19 REVIEW OF PROPEER-WING AERODYNAMIC INTERFERENCE Naelle wake Wing wake Ti vortex Table 3 Comarison of numerial and exerimental (balane) data; RKE=Realizable k-ε model; RSM=Reynolds Stress model.. roeller off α = 0 α = 4 α = 4, ro off C D Predited (RKE) Predited (RSM) Fig. 31 Combined flow ath lines and ontours of total ressure loss behind showing the wake and the ti vortex struture; α = 4 ; ro off. deformation of the wing wake and the slistream, as skethed in Fig. 31 and Fig. 32 resetively, is essential for a detailed analysis of the roeller-wing interative flow. The strongest oint of the NS-ode in the analysis of the roeller-wing interation roblem is its intrinsi modeling of the swirl reovery effets. Furthermore, no user intervention is needed to resribe the slistream osition within the omutational domain. Integration of the surfae ressure and frition fores leads to a very aetable agreement with exerimental values found earlier (Table 3). C C D C Balane data Predited (RKE) Predited (RSM) Balane data roeller on α = 0 α = 4 Predited (RKE) Predited (RSM) Balane data Predited (RKE) Predited (RSM) Balane data z/b/ Predited z/b/ Exeriment roeller disk y/b/2 Fig. 32 Examle of redited and exerimental ontour lines of onstant total ressure oeffiient; ro on; α = 4. 6 Conlusions Exeriments on roeller-wing onfigurations reveal a very omlex flow with high levels of vortiity and onsiderable shearing fores in the wing area that is washed by the roeller slistream. Surfae ressure measurements learly showed the effet of the swirl veloity and the inreased total ressure due to the roeller whose loal influene is diretly ouled to the roeller rotational diretion. The fore measurements and the surfae ressure measurements demonstrated a erformane benefit when the roeller rotational diretion is inboard u. This finding indiates the ossibilities to design otimum wing shaes whose exat shaing and rofiling deends on the struture of the inoming slistream. 19

20 ..M. VEDHUIS With reset to the ouling between thrust and drag it seems that in many investigations the auray as well as the hanges in the roeller thrust fores due to interferene reeived limited attention. This aroah is mostly refleted in ignoring the wing effet on the roeller (SIM). The full interation mode (FIM) that used in the VM alulations showed that even for a normal layout, with the trator roeller well in front of the naelle, small effets on the thrust oeffiient are found in the range of several drag ounts. From the arametri study of the roeller osition effets the following onlusions may be drawn. roeller inboard u rotation is benefiial as higher lift/drag ratios are obtained at onstant ower settings The effet of the roeller sanwise osition is negligible for loations well within the range that is normally used for modern turboro airraft. Nevertheless, an inboard u rotating roeller ositioned lose to or at the wing ti delivers a onsiderable erformane inrease. When the vertial osition of the roeller is hanged, high ositioned roellers obtain the highest lift oeffiients. The drag oeffiient, on the other hand, rises for both high and low roeller ositions and shows a relative derease for z 0 due to the "donut" effet. It is imortant to notie that the alulations rove a low roeller osition to be benefiial regarding the roulsive effiieny. The reason for this effet is attributed to the redued roeller inflow distortion due to the wing. The streamwise variation of the roeller osition shows small variations in the roulsive effiieny of the onfiguration. For ositions further away from the wing ( x < 0 ) the roulsive effiieny inreases slightly due to the rise in the axial flow veloity inside the slistream and the diminished wing effet on the roeller. Negative roeller inlination angle with reset to the wing are in general benefiial for values of α < 15 The main reason for this ositive effet is the hange in the slistream veloity distribution where inreased asymmetry is introdued and the fat that the omlete slistream enveloe is rotated to rodue an average inrease in the loal wing flow angle of attak. The alulations erformed with the different odes desribed herein in general onform to the trends found in the windtunnel exeriments. The VM-aroah delivered quite aurate results omared to the anel methods and NS-ode. This lose agreement is found only after alying a swirl reovery fator (SRF). In the in the anel methods based on a slistream enveloe model that is solved simultaneously with the other arts of the numerial model, the absene of the SRF always leads to an overestimation the slistream swirl veloity omonent and hene the loal lift effet. In the inlusion method where the slistream generated veloities and ressure are resribed on a er-anel basis, SRF an be inororated without diffiulty. Hene this anel method aroah is referred over the slistream enveloe model. One aurate results are needed for the roeller-wing interferene roblem and details of the flow are needed to determine the seondary flow effets that influene the lift and drag erformane of the model, the NS-ode beomes indisensable. The NS-aroah failitated the identifiation of tyial flow henomena, like the deformation of the slistream when assing the wing. As indiated before, the sanwise distributions of lift and drag fores are sensitive to the form the veloity distribution in the slistream as well as the way the slistream deforms when assing the wing. Hene, a alulation model based on the NS-equation yields a more realisti estimation of the roeller wing interative flow sine the slistream is allowed to develo and 20