Download as pdf
Download as pdf
You are on page 1of 9
Journat or Amcnart Voll 32, No. 3. May—fune 1995 Analysis of Some Interference Effects in a Transonic Wind Tunnel Giovanni Lombardi” University of Pisa, Pisa 56126, Maly and Mauro Morellit Council for Scientific and Indusirial Research, Pretoria 0001, South Africa Ie effets of the walls ofa test seston on a mode! in transanie flow were investigated by using the AGARD Catineatlon Model, Tests were caried ut i 2 closed-circuit pressurized tunnel, with a confined square test section of 1S m width, with tapered slots giving a 5% porosity. Two modes with different dimensions were tse, with 0.95 and 0.056% blockage rates. Longitudinal aerodynamic characteristics were analyzed by means ‘of measurements performed at varying angles of attack (ap to 24 deg) and Mach numbers from 6.3 to 1.2. In me flow conditions wall itererence eects were prebably present. However, the forces amd moments de- pendent on the pressure distribution were likely to be related to the same factors, and therefore, the above lMfects tended to disappear when longitudinal stability and Hi-dependent drag were analyzed asa function of| lit characteristics. The drag rise Mach number evaluation seems be fully free from blockage effets. The dimensions of the tested larger model can be considered tobe the largest reasonable ons for industrial app ‘ations, bt, probably, not suiciendly small when high accuracy is required. Nomenclature = mode! span, Fig. 1, mm forebody drag coefficient minimum value for Cy for a given Mach number zorovlift drag coefficient for a given Mach number sources of error affecting the accuracy of experimental data ‘The classical cortection criteria are mot satisfactory for an accurate data correction, being based on insufficiently rep- resentative theoretical linear models,’ whose validity is lim. ited to low velocities and angles of aitack; however, even in these conditions, the accuracy of these criteria is not high Cy = forebody lit coefficient since they do not account for physical tunnel characteristics Ci, = forebody pitching moment coefficient (Ge. filets, re-entry doors, ete). With the introduction of 6. = chord at the body-wing junction, Fig. 1, mm the ventilated test section for high-speed subsonic and tran- ©, = mean aerodynamic chord, mm sonic testing, new procedures have been devised to extend = ee cords Figs Lem the classical wall interference methods for correcting model = model fuselage diameter, Fig. 1 mm test data, Beeause ofthe complex nature of the interference, a Giameter of the support sting, rim 2 satisactory general analytical solution to this problem for K = coefficient inthe telation Cy ~ Dpy + KCE Nentiated walls is far from being achieved. More recently, 1, = base to wing triling-edge distance, Fig. 1, mm new correction methods were introduced, based on more co EL) = oaie length, Fig. 1, mm plex procedures, coupling measurements—typeally pressure 1, = ‘total length, Fig. 1./mm Endior velocity on the wall or in the field—with mumerical Mo = Mach number Caleulations.°* Anyway, these procedures experience dif Sf = static pipe average Mach number Culties because of the small quantities to be measured, the Me prondo-Aach number calolated from the uncertainty of the devices used to make the measurements Pyualibrium plenum pressure find the interference trom other sources.* as well as because P= confidence level ‘of the complexity of flow calculation. Furthermore, in thre 5, = gross wing area, Fig. 1, mm? dimensional problems, the number of field measurements fet scetion wih, fom necessary for a proper wall correction evaluation is 0 large streamwise coordinate in the test section, Fig. 2, tat it yields an experimental procedure very difficult to pe form, at least regarding industrial applications. This is par- ticularly true for transonic conditions or for high angles of attack, because linear models are not vatid and itis almost impossible to use a reliable numerical code The above considerations explain why limiting model di mensions as much as possible remains the best way to avoid unacceptable errors. On the other hand. the importance of testing the largest possible mode! is evident, not only to max imize the Reynolds number but, especially, ro improve the accuracy of force measurement and of the model itself. Thus, itis important to have reliable criteria for choosing modsi ‘advancing knowledge on this matter isthe main purpose Uistance of the moment reference point from the model base, Fig. 1, mm angle of attack, deg Introduction HIF interference om the flowfield around a modet ceused by wind-tannel walls is known to be one of the main ‘Received March 26. 199; revision received Aug. 31, 1994 a epics toe publication Aug, 31.1994, Copyright © 1984 by the Aimer ican Lsurute of Aromatics sind Astroautics, Tee, Al Aight te ‘sistant Profesor, Department of Aerospace Engin Dotti, 2, Member MAA. Rescatch Engineer. Mediam Speed Wind Tunnel, P.O. Box 395 gs Via sol of the present research activity The effects of the walls of the wind tunnel on a mode) in transonie flow were investigated by using the AGARD Cal- ‘bration Model B specified in Rel. 7 (see Fig. 1), [tis an sgive-etlinder boul with 1 wing having a symmettical eurcular are airfoil section. alzesdy tested in many uiflerent wae tannel.* “Two models with vifeent dimensions were vised. the Tist_ one (ranted rmwdel) had s nomi sliamter Hig. 1} 13 mm, with corresponding blockage at zero angle of atiack of 88% Gneluding the wingsceticmal area}. and a mudel-tn-(est seetion width ratio hit of 0.400 The other one (named “small” model} had & nominal di ameter D ~ 38.5 mm, with a corresponding blockage ott 365% and a fr ratio of 02103, Because of the very low bloc af the small model. the latter could be considered free of wall interferenee, The dimensions of 1 chosen as indicative of the largest possible for the test n geometry To avoid differences caused dy different flow characterise ies, which may significantly affect forces (particulary. the Iurbulenee scale and intensity") the tes eandhitions were kept the sme for boil models: in this way the two models were Fested at different Reynolils numbers, but this influcnes can he considered to be (atheist yualitatively) known. particularly tor the tested model, whieb hil heen estensiveiy stadied in many diferent conditions ae lg Nowa icenaons Ly = 85D Tet be sap % = 1402 & Fig 1 AGARD U. Experimental Setup Tests were carried out in the Medium Speed Wind Tunnel of the CSIR Laboratories, in South Afvica, ‘This isi elosed circuit variable deasity transonie wind tunnel. Its operational spoed ranges from AP = 11.25 to 1.5, with presse vay from 20 to 250 kPa: Reynolds number eat! be changed by ‘modifying the pressure. The test section his w LS % 1.5 m: ‘ross section, 4.5 m in lengths. All four walls are eqully fon situdinally slotted (“coke bottle” type) fora total porosity a 3%, following the scheme in Fig. 2; the lateral walls sve par ich, while the upper and lower ones are slightly di 00.7 deg cue, starting from the test section inet ‘The main flow characteristics in the test section are sum marized in Table 1. derived Isom the complete calibration «i ‘the wind tunnel.” The models were assembled on the main mode support ‘Fix. 3). which consists of a puch sector with w tol Head that transmits the attitude movements to the model. ‘hey were vonnected through internal Balances onto & sting, a base imressure probes were fitted onto bath models, On the large model to pressure orifices wore fitted in diametrically Posite positions on the sting at the base of the model, whe: ‘on the small mode! @ single enifice was used 10 messure the hase pressure. The hase axial force coefficient was evaluated by calculating the pressure acting on the model hase rea this value was subtracted from the measured axial foree cv ficient to determine the forebody axial force coellcient.! which was used to evaluate the forebedy drag and lift vou ficients C,, and C;, ‘The mounting scheme is identical for the two mvs, in particular the sting diameter to base diameter ratio «l/D Tor the large model is 0.500, and for the small mode it is 0.504, ‘Two internal, six component strain gauge balances were used 4 measure the forces on the models; the balianees were cal ‘nated with a second-order matvix technique. ‘The two models were positioned in the same zone ot the ‘wind-tunnel test section; the large model between the stations v= 2213-MS8 (Fig. 2). the small model hetween the stations = 2620-2948. Both models were stationed in the “Standard {est section™ as defined by the primary twnnel calibration, = namely from x = 2100 t0 3600. Consequently. the tunne! Mach number determination is obtained from the offset pa Fameter (Mf — M,) caleulated during the statie pipe corte: lation where the value of 47 is calculated across the standard Both models were tested ‘conditions and attitudes, samely, from MF nder the same environmental 0.320119 in U4 4500 tow direction —> || Fig. Dafinitian a the walls of the wind-tunnel fest section, Te sors on the foue I are pnstioned in the sare way (all inensions ate in Sia umber with tine Stability af sagnation pessine wt time Stahity of stagnotion temperatune with Hie Hui fev sewn Spatial variation of Mach umber Spatial variation of ow angutacity ‘Aooulic pressure coeticent fhvctuation Acoustic Hutuation Frequency euntent Turhalcae tee! D 130. 1 as, > a ¢ 388.) L 23 Wetted wing area, mn = Model volume, m Blockage factor at = Udeg — Blockage factor at @ = 24d = wD Phishing Foe em ainetion Fig. 3. Scheme of the support and reference system steps. from MF = 0.95 t0 1.1 in 0.05 steps, and at M12 Aupiliary suction was used at Mach numbers between 0.9— 1.2. The mass flow evacuated from the test section was 27% cof fae plenumn evacuation system (PES) tor M >= 0.91.1 and 2.5% of the PES at M = 1.2 Under the texted conditions the Reynolds number, referred to the model total length £,, varied trom 9.2% 10" te 1.7 > UP tor the large model, and from 2.4 > 10" 0 4.5.10 For the small model ihe models were tested at piteh angles that ranged from 4 to 24 dee for Much numbers up to 0.7, and from —410 17.5 dea for the remaining Mach numbers. in these eanitions ho model oscillation was detected during the angle-of-attack sweep for either model. The measured body axis forces were converted into drag and lft values, and the effects ofthe base Pressure were eliminated, Le, only the forees acting on the forenody were consider The balance data were acquired through a NEFF system 820 series 500 provided with a 16-bit A/D converter. The pressure data were acquired through a Seanivalve 20°C system with a 12-bit A/D converter. The adopted reference system is shown in Fig. 3, The force coef cients were nondimensionalized with the dynamic prey Oy = BATE On AUNT KPa mys 2K 2 dey below tunnel static temperature faa = OM (subsonic) 174 OM (Supersonic) 6, 5 0.18 dog (sonic) 0, = 0.30 deg ‘in, = Out infor Superson) =00n 8.001 (low Mach 10! = 0002 (high Mach) Real on 70 me 8.68 Isis 3.90 rina 20 Sus 19.0 133.36 Mot soo 422 B12 155.5371 10.4657 81,3516 5,185.0 air vani298 U8 ss0% sae 406% hue oo ‘ius oi 0.36 sure and the gross wing area (S,. including the wing carry through the fuselage, see Fig. 1}. the moment eoeflcients with the dynamic pressure, the gross wing area and the mean aerodynamic chord (c,,,) and Were referred to the quarter. chord point of the meat aeradynamie chord The quality control inspection carried out on the motels is fully described in Ref. 14; it was accomplished on a modet 22 DEA [OTA coordinate messuring machine with com puterized measuring and recording capability, In Table 2 the hhontinal and real dimensions of the two moxlels are reported the uncertainty in the linear measurements being less than 8.01 mm ‘An automatic Mach controller maintained a constant Mach ‘number, compensating for Mach number variations such as, those mduced by the model piteh eyele, The uncertainty in Mach number vas 0.002%; that in the angle-of-attack pos Homing was fess than OL deg, and dita were corrected lor sting deflections, In Table 3. for the large model, the coral uncertainties in the dat that can be attributed to insiaumentation errors (forces ‘dynamic pressure, and ase pressure measurements), refer~ ence dimension evaluation (Surfaces, lengths, and moment reduction points), und data acquisition procedure are show and expressed as Coefficient uncertainty: the small mod! ‘Mach Fig. 6 Pict angles of attach. ratios ¥s Mach number, for different and b) 17-8 deg ‘18 models also characterized by an almost linear correlation bewoen the Ci and the C), values, In Fig. 8 the least-square best-fit interpolated linear coetficient K between CC; avd Cy ~ Cia is reported (because the symmetry of the body is Cohan = Cra) a8 a function of the Mach number; differences ave almost negligible, and there are very high values of the correlation factor, in any case higher than 0.99: Exentin this setting, blockage effects are likely to oveur when changing the angle of attack, but again, they have the seme origins as those related to lift, which is why they disappear ‘when drag (as s usually the cise) is studied as a function of Wi The zero-ft dragis dificult 10 measure because ofthe small values of the forces involved, with a consequently high esror Feel (Table 4); an accurate analysis of the zero-lfe drag chat= acteristics would require a different test campaign. with a ‘completely different choice of test procedures, In Fig, 9 the Cy. asa funetion of the Mach number, is shown; differences are present, but they may be due 10 a Reynolds number effect (in this ease, a greater extent of laminar flow on the small nodel_ alfecting results in the way previously shown): indeed, this parameter is markedly affected by viscosity. In any ease, dilferences are of the (high, in this case) uncertainty order, ‘except for the higher Mach number where differences ure ‘more important, being significantly higher than the uncer- tulnty level. Thu. it Tikely that. in supersonic flow condi- tions. « blockage effect affecting zero drag mneasurements was present, while no definite conclusions are possible with re- spect {0 the ether flow conditions, However, it is of interest that the evaluation of the drag rise Mach number does not appear to be affected by model scale (Fig. 9). J Mesa : 02°04 06 08 1 ia is 2 a ey — | ao AE 3 » cy OS creeper e-- Lance | ae SMALL j 0 02 04 06 O8 1 12 14 c2 a 08 pepe or Lance 4 SMALL o4 oe 0406 8 1 12 14 * L Drag, vert drag squared it for diferent Stach mum Ms a) 08, Dp OR. €) Lend d) 13 Hig. 7 bers. 040 pr — a 038, ey 0.30 arse 2 SMALL 035 0 02 u4 06 08 1 12 Mach Fig. 8 Linear & eetfcient, relating increment rag 0 vs Mach wed iN, a4 7 + LARGE SMALL, aga oan oon oo , 0 02 04 06 08 1 12 Mach big. 9 Zerit drag 98 Mach number. Conclusions The use of ditfereat scale mols, operating in given wind Iumel under identical flow conditions, appears to be an ape propriate procedure to gain information on wall interference teflecs: in fact, this approach abolishes any differences related to different flow conditions, and the uncertainty in measur imcnt comparisons is considerably reduced, being limited to the randont component (which can be reduced, theorstically, 0 anv desired vatlues)_ of the measurement procedute as well sto the biss uncertainty related to balances, The present research activity acklressed wall interference celfoets on longitudinal acrodynamie characteristics, especially as rehited to pressure-tependent forces, An accurate analysis fof the zero-if drag would actually requite a specialized fest Ccunpaign, with an appropri choice of instraments and test ing techniques, The test section geometry is very elose to chat generally Sugested as the design producing a wall interference-free ceweironment for moderately sized models (such as the large imcdel used here}. The results presented in this article would ‘confirm that this solution is able to significantly reduce wall Interference elfeets, particularly @ the very critical eoncitions ‘of Mach auanners close te However, in some How conditions, despite the moderate ions of the Jurge model. wall interference effects are pprohubly present. In particular, the pressure-dependent forces Luppear io be atfected by blockage effects at high angles of stack i the fw transonte regime (Af = 0.70.95). However Suse Wockaye effects on pressure actions were likely to be ccatsed by the same factors, they tended! to disappear when Fongitudinal stability and lift-dependent drag were studied as Inte RPERENCE eETEC TS « function of lift characteristics. This is for the taille. eon figurations like the tested one: the extension to a complete uration is not necessarily immediate Another parameter that may he eonsidered affected by blockage effects is the zero-lift drag in the supersonic regime although no definite conclusions ean be reached in the sul sonic regime, sinee the measurement procedure adopted is not sufficiently accurare for exalunting the zero-it dr However. itshould he observed that the drag rive Mich nur ber evaluation appears to be free of blockage effets, even, though the lage error in the measurement of the 2er0-lit drag does completely grant this ‘These conclusions cannot be immedistely extended 16 dif- ferent configuration iypes.it being dificult to generalize wall interference results from any one configuration, However. it 's possible to indicate that when n model with dimensions comparable with the large one is utilized in atest section like the one used here (which is typical Tor the latest transonie wind tunnels), ican probably be eoasidered not afected by significant blockage effects (with respect to lift characteristics and not to the angles of attack}. Te appears that the above dlimensions shoul! be considered the largest aecepsable for ‘corrected inuiuscrial applications, even though they ae prosh- ably not sufficiently small if high wccuracy is required It must be noted that the principle objective of this work was to analyze the globat blockage effeets, therefore, no cor reelion procedure was taken into consideration, and the re sults thus achieved are not the “best” obtainable in the tunnel, By applying the clussic corrections «significant inesease i the accuracy of the results can be expected: improvement in the accuracy ean alsa be achieved by testing the models in a redefined” test space, where the offset calibration parameter (Mf — M,) is re-evaluated for each applicable test space with every different model installation, thereby granting minimal bias References "Vavssite, 1 C., “Survey of Methods for Corseting Wall Con. ints in Transoni Wind Tunes.” Problems a Wiad Tunnel est ohnigues. AGARD Rept. 61H. April 1973 ‘Bengeink, R. 1. ind Zinsrling, N. 1. “Wall Interlerence Mew surements for ‘Three-Dimersional Models in Transonie Wind rls: Expesimental Difficulties.” NASA CP 2319, Jan. 1S, ‘Asti, PR, "Boundary Flow Measurements Meth for Wail Inverforence Assesment and Correction —Cissifcation arnt Re view." Fel AGARD Fla! Dynamics Pane! Mk Symposiany fon Wall Interference. Support laterderence and’ Flow Fil Mes Stentens. Brusiels, Belen, Oct. 1995 “Ferri, A and Baron P., "A Method for Transonie Wind nel Corrections." AUAA doural. Vol. 11. No. 1 1078, pp, 61-66, ‘Mont, “Wall Corretions for Airplanes wis Ltn Tramsonie Wind Tanne! Tess.” AGARD Ad Hoe Commitice on Engine — Aisplane tnerterence ard Wall Coreeetions in Transonie Wink Tu hel Tests, AGARD AR 36, Aug. 1971 Naucheret, X,. and Vayssate. J €.. °Correetions de Pais en Exovlement Trdimensionnet Transsomgue dans des Veines a Paris Ventiées." al AGARD Fluid Dynnnes Panel Mieting sid Sgn psiuin on Wal: lnterorenee, Support Intererense and Foy Pcl Meassremenis, Brussels, Belgium, Jet, 1993 Speciiaiton forthe AGARD Wind Tune Calibration Max AGARD Memorandsim AG-AIMS, Aug. 55, “Mls, R. (ed). "A. Review af Meastrements on AGARD Cal bration Models,” AGARDogniph 64, Nav, 196) ‘Ongiratn. 1. "WindTunnel Wal ntetensnae Statics at Hi Subsonic Speeds.” nal of Aireraf. Wo. 4, No. 2.49. 19, "Odo, 1. “Sysemaieal Investigations ofthe Iflsense nf Wine! Tunnel Turhutenee oa the Results of Mocel Force Mesnurenenis Wind une! Design anu Testing Techies, AGARD OP 71. Oct "Anderson. ©. &.. “An Investigation of the Acrodhtinie Char acteristics of the AGARD Model B for Mach numisers Irom 12 to lc" Arnold Engineering Development Center, AEDCTR-70- iI May 1970 ing | | LOMARDI AND MOREL INTERFERENCE EIT ers ° Hurlin, RS. aid Davis. MLW. “Prolninary Calbeation Re Davis. M. W., “Analysis uf Pressure Menuet Eras sls or the CSIR Mesum Speed Win! Tunnch” DASE Rept. 80. plicl the Siac Pipe Cabiratin of MSW Ie ith Som SOU, CSIR Pretaria, Sov Alvi, Now. 10 ‘lvnamic Testing Aswociaion, Tori, Tals, ay 0, Martindale, W. Rl, "The CSIR Medium Spcd Wind ‘Tunnel "Lombardi G.-and Morell, M..-" Wind Vue! Blockage Foes Automation for High Productivity wi Excellent Flaw Quilty.” on the AGARD Bi Model sn Transomic Flow." oth Cones of the Svezdrap Penjets 9339, CSIR, Pretoria, Somth Alea, Dee. 1950 International Couns ofthe Aeronauticl Scioases, Amatinn CA ‘Christin, “Dimensional Inspection Report Projects Exe Sept 1994 peed (PLY L1D Caihraiton Fyuipmont, Hage | CSIR. Preto, ‘Coleman. H, W.. and Steck. W. G.. "Experimentation nd Sot Asc. 198, certainty alysis For Engines” Wile). New York. PS f i al > WV JUCEOOOOCOOOO AIAA Aerospace Design Engineers Guide | Third Edition | This thin, revised and enlarged edition provides a condensed collection of ccominonly used engineering reference data specifically related 10 aerospace design. I's an essential tool for every design engineer! TABLE OF CONTENTS: Mathematics + Section properties + Conwersion factors # Structural elements * Mechanical design Elecrical/electronic + Aircraft design + Earth, sea ard solar system + Materlals and specslications Spacecrat design * Geometric dimensioning and tolerancing 1993, 294 pp, ilus, 9x 316" leather-tone wite binding, ISBN 1-56347-045-4 AIAA Members § 29.95, Nonmembers $49.95, Order #: 45-4(945) ace your ode taday! Call 160/682 ALAA GAIA Sasicecatin Sonu arenes Vanes aene Se

You might also like