You are on page 1of 7
TCAS-04-2.2.3 [EXPERIMENTAL VALIOATION BY FLIGHT MEASUREMENT OF THE PRESSURE DISTRISUTION ‘COMPUTED ON PILATUS PC-7 WING USING A THR ‘DIMNENSTONAL AERODYNAMIC PANEL PROGRAM. 8. Dulio A. turk PILATUS AIRCRAFT LD. STANS 7 SWITZERLAND Apsteact Wind tunnel tests add a considerable amount to’ the development cost of an aircraft. Tt 1s then edvicable, particularly for small companies not equipped with thelr own feci- Lities, to replace, as far es possible, ex periments with numerical calculations. “This Toads to tine and money saving especially during those phases of the design requiring Lengthy optimization precesses. To do #0, the three dimensional eerodynanle penel method of Hunt and Semple haz been introduced at PILATUS. ‘The code validation hat been conducted first using the available Literature presenting Pressure distributions (generally from wind tunnel experiments) and Lift and drag data fon several combinations of wings end bodies Subsequently it has been decided to undertake a Flight test campaign to obtain reference Gata for a further checking of the minerical code. The test resulted in pressure measure ents at two spen stations on the wing of the Company PC-7 demonstrator. Pressure vas nea. sured and recorded in flight using PCH (Pulse Code Modulation) date acquisition system. Chordwise pressure distributions have then been obtained for different flight speeds. Tho numerical results have been com- puted approximating PILATUS PC-7 wing with « podel sing? spanwise stripe of 30 pancls ‘each. The pressure computed et the sume span Wise location as used in flight measurements, has been corrected with # three dimensional boundary layer code to account for viscous effects ‘An important feature of the numerical code, anely the possibility of evaluating off-body Flow field, hes boon exploited to account for the upwash’angle affecting the reading of the langle of attack vane Once this correction was done comparisons have been made st the same incidence for the unericel model and the wing in flight. The apreenent achieved is good for the severe] fangles of attack considered. Therefore the test campaign has proven that three dimensional panel codes are able to provide within the Limite of their validity SMresults which are affordable for a small ‘aircraft company and which can be confidently Used during an aicceatt development. phas Copyriht © 1984 by ICAS and AIAA. Albis reset Introd; At the bepinning of the development phase of ite latert project, the Pc-9, the Engineering Office of PILATUS Aircraft decided to acquire numerical code which could enable. the Group to conduct, ia a quick and reliable manner, e whole series of computational act— ivities. This would have reduced, on one side, the necessity for long and exponsi: wind’ tunnel tests and, on the other, would Ihave provided all the different development offices with the cequiced aerodynanic data, ‘Among all computer prograne evailable on the arket, it war decided to select the Hunt & Semple’ panel program developed by British Aerospace in Warton, for its high Flexibility of use and the well proven numerical approach Although the code nas been succesfully used in BAe for sone yoers, PILATUS decided to conduct ite own validation campaign, mainly to "tein" the method with the kind of conti~ turations and the range of applications typi Eel of PILATUS products, and slightly dif— Ferent from those considered up to then by BAe. Tho program was in the meantime conver: ted from its original Iam code to the VAK code and this further activity led to the Fequicement For a final check. Since the possibility existed to conduct a orice of flight tests on the company testbed Sircraft (a Po-7 with two instrumented wing Sections: ee Figure 1), it was thought thet © comparison between pressure distributions ‘obtained in flight and those computed by the Progran on the elreraft wing, would Yon’ good dare for the validation of both Uhe method and the model adopted to represent the aieeraft Such a comparigon wes Intended to prove the bility of the code to reproduce local flow Gotails (agreement between measured and com pated pressure distributions) and confirm See capability of accurately predicting the global loads ‘acting on the aircraft major Components. Thus enabling the aerodynamic group to confidently use the program as an Optimization tool during the development Phase of « project and as a relisble source For the computation of the aerodynamic loads quired throughout the whole detailed stressing of the aircraft. ‘The_computer model and the numerical results Tho Hunt & Semple MKIA panel program is a first order three dimensional panel propran; it cen treet any given combine- tion of Lifting and non lifting three Gimensional bodies and It incorporates 4 boundary layer routine to evaluate Viscous effects on lifting surfer ‘The progran requires a subdivision of the body under consideration into « set of planer panels which approximate as much as possible to the original shape (see Figure 2). In the present test can— paign only the wing of ‘the aiccraft hes been represented (see Figure 3), since ure distribution in that part of the wing where the instrumented Sections are located. The ocixinsl wing and its model The Wing 1s based on two different pro- files GUA 64 series) of 15 per cent thickness at the root and 12 per cent. at the tip. These profiles have been no~ ified by PILATUS to reduce the pitching, moment coefficient The wing is charecterized by a 7 degree dihedral and a 2 degree twist; wing taper ratio is 0.56 and aspect ratio is es ‘To model the surface 7 spanwise sections have been used. These sections wore 10- cated along the span according to # cor ine law to give « better representation of the Lift distribution towards the wing tip. On each section the profile was’ approximated by 30 panels, whose edges wore fixed along the chord by Using again a cosine distribution, to accurately ropresent the profile shape in the leading edge and trailing edge The location of the spanwise stripe was chosen so that two of the control see~ tions wore in exactly the sane position fas the two instrumented sections avail- able on the test-bed wing, corresponding, te rib 14 and rib 27 of the wing. struc~ Mo detailed modeling of the wing tip regione was performed ax only symmetri— cal flight conditions at Tow angles of attack would have been considered in the test phase. In this situation the cal- culated fFlowflelds ate insensitive to the modelizetion used for the wingtips (closed wingtips or open wingtips) 197 Le The Numerics] Results During the sunerical evaluation pha two sete of computations have been per- formed. In the first set a preliminary "eeimming” of the configuration was done by compering measured and conputed pres— sure distributions at the came integr: ted Lift coefficient. This suggested the existence of « dis. cepancy ‘between the angle of attack of the alecraft, es read by & wing mounted incidence boom, and the nominal free stream value used in the calculations of the wing flowfield. A cow of external off-body points was then added in the numerical model at the boom location to work out an angle of attack correction. The sane configur ton was then recomputed at the correc ted values of the angle of attack. This improved the quality of the results by giving the possibility to compare pres Sure distributions at the same inciden— In thie second pact of the computational phase meny cases were run including vis- cous effects (boundary layer iteration) to investigate the influence of the Rey- holds nunber in this sort of comput tion. Special care was taken in defining the combination of Mach number and Reynolds unber to be used in the numerical a lysis. Each single measurement corre- sponds to « well defined combination of angle of attack, Mach number and Rey- olds number. In order to save computa tional time it was decided to evaluate the pressure distribution over the whole range of angles of attack obtained in Flight, only for two combinations of Mach aad Reynolds nunbers The complete numerical evaluation over ‘ehe full range of data points (combine Eins of incidences, ‘Mach numbers and Reynolds nusbers) took something like a5 hrs of computationel time one VAK 11-750 computer. Tho_flight test setup (PILATUS Data ‘Aeguisition System ‘The date acquisition system (DAQUS) used by PILATUS is based on the Pulse Code Modulation (POH principle and is i1lu- strated In fg 4. The PILATUS systen, that Is to say the encoder, decoder and all the perigheral devices, can be. bro- Ken down into the "Date storage unit” fand the "Data processing unit” aa 22 Data Storage Unit SENSORS - For this test the main sensors were pressure transducers, in conjunction with « scannivelve, and a potentio- eter: SIGNAL CONDITIONING - This unit converts the transducer output signal to the required + 5 ¥ input signal specified for the kn coding unit. ANCODING UNIT - This Is the true core of tho Pow system. It accepts all outputs from the signal con @itioning unit in paraliel, fand the results output to a Single data bus and stored on fone track of a tape recorder. ‘TAPE RECORDER - The PCH system allows for the use of conventional sterco tape recorder. At only fone track is used for data re~ cording the other is available for simltaneous vole record Data Processing unit ‘TAPE RECORDER - An Identical tape recor der ac described above 1s used For date replay, DECODER - The decoder identifies the digital information from the tape and sorts it into ite respective channel identifi- cation and signal size. The output mode we fan analog, trace. OUTPUT DEVICE - The output device used was © U-V paper trace recor der, which allows the output channel to be plotted against time In an quesi-analog form Comparison between theoretical end perimental results As a Elrst step In the validation cam peige, reparate comparisons have. been Performed between computed and measured Pressure distributions relative to the two. Instrunented sections on the alr= craft wing. Figure 6 shove the first of these com perisone at rib 14 (the indoerd one) at fan angle of attack of 6.1 degrees. Fig- lire 7 shows & comparison at rib 20 (the outboard one) for an angle of attack of 5.6 degrees As it can be seen « satisfactory agro nent is obtained in both cases. Tn fig- te 8 the discrepancy at the trailing edge region betwoon the calculated pros Sure (continuous Tine) and measured pressure (symbols) could be due to the presence of the split Flap on the last 20. per cent of the chord, modifying the profile shepe. The pressure peck near the leading edge is slightly overosti— mated by the program. Ageln a fairly good agreement is ob- Esined on rib 20, despite sone inregu- Lecities Introduced on the wing shape by the presence of the sileron hinge and gp As mentioned before, the greatest dif- flevity encountered during this preli— inary tect phase was obtaining © g00d Agreement with flight test date when Celeulations were performed at the sane fangle of attack ae mearured by the Flight tost boom. The good accuracy of figures 9 and 10 iz achieved by showing the comparison at the sano integrated Lift cosfficient. In both calculations no boundary layer effects have beon ac counted for. Tt was then decided before proceding in the validation campaign, to obtain « correction which could relate the angle of attack measured by the doom with the free stream value. The correction proved that at the boom location (approxinately one wing chord in front of the leading cdge and at 99% of the wing span) & significant upwash from the wing was Present. The calibration of the probe Shows that the boom always read an angle of attack greater than the value at in Finity by a quantity increasing With the wing 1Let. Calculations were repeated at the cor rected angle of attack corresponding to the free stream value. Viscous effects were also taken into account. Caluiations have boon performed only For ‘two combinetione of Mach and Reynolds unbers corresponding to the extremes of thelr ranges as experienced in flight. Table 1 summarizes the results for the First combinstion: Re = 7.18 x 108 and M = 0.306. Tt. shows the Freesteeam Angle of attack ( &% gs), the calculated ‘and the measured angles ‘at the boom 1o- cation "COC g,. Ogu respectively), the Life coefficient (c) calculated With and without viscous correction (Chew + Ching Fespectively) and ex- potientan vibe of €y Chace? Figures # to 11 show the corresponding. comparison between nunerical and experi~ mental resulte The pressure distribution agroonent is Acceptable over the whole range of cidences. The shape irregularity of the peasured distribution for the lowest tangle of attack, derived from atmosphe- Fie turbulence, is always present during any of the teat Flights ‘The crossing of the computed pressure disteibution at the teailing edge is due to the particular formulation of the Kutta condition adopted to enhance the Life losses due to viscous effects. ‘As a comparison, figure 12 shows t sane pressure distribution as figure 11 without viscour correction. By eccount— ing for boundary layer effects there was a better representation of the leading edge suction peak. Table 2 summarizes the results for second Re and M combinetion (Re = 3.318 x 10e = 0.142), Figure 13 presenting the corresponding pressure distribu- tions. The graphs show that there is ot a dramatic difference when compered With the previous set of results, from Which it 12 possible to conclude that H 1d Re did not play an important effect on pressuce distributions over the ange considered. Sonclusions During the described test campaign, it hes to acquire a wide experience sd interpretation of the Hunt Semple program, and a confidence in its cape bilities. has in fact been proven capable of predicting, within an acceptable approxima tion, the flow behaviour on the PILATUS test— bed wing. Therefore the designer has the pos- sibility to use the code to investigate the effects of design changes on the flowricld Sound the aircraft or to evaluate configure: tions during. the development phare, in = quick and rellable manner, ‘The accuracy in the prediction of loctl loads (pressure coefficients) is in fairly accept- able Limite, and conficms that the program ‘can be confidently used as a means of evalu Sting aerodynamic Loads on structural compon- fente, with the result of bringing to an Gnigue source the deteraination of the sero- Gynamic leeds for the structucal sizing or checking phase ‘me boundary layer simulation improved the Flowfield prediction only at relatively high isles of attack, thus allowing (For low langle of attack) the use of the code to simi ate potential Flow with e considerable sav— ing in computationel tine. Finally, it has deen found that the accuracy Inherent to this kind of computer code (thees dimensional panel progeans) is adequate for their application a2 optimization tools dur— Ing the development phare of « new project. 199 Pc-7 Testbed Figure 1 Hunt & Semple Mode? Figure 2 i ca] 4 [estan ores] este) Piste ; a 3 PILATUS DAQUS Figure § 200 « -07 | -0.8 oa 7 ors | 0.46 0.190 185 | 2.08 0.323 aa | 6.00 0.635 600 | 20 0.77 Re = 7.14 108 je = 0,306 able 1 5g £2 foe ge OD | 1 0.14 0.190 0.322 0.685 0.77 Re = 3.318 16 jn R= ote | able 2 202

You might also like