Studies in general aviation aerodynamics

University Maryland, College Park

Department of Aerospace Engineering

1121., -';';- :'-" 7 "/._'7"--+

f,l ,F_

..?-0

Final Technical Report for NASA Grant NAG-I-681

Studies in General Aviation Aerodynamics

Allen E. Winkelmann

Department of Aerospace Engineering

University of Maryland

College Park, MD 20742

N_)2-28511

(HAga-CR-190_3I) STUDI_S IN GENEP, AI.

AVIATION A_ODYNAM[£S Final Technical Report

(Maryl,Jn_ Univ.) 20 p

G31oz

Final Technical Report for NASA Grant NAG-I-681

Studies in General Aviation Aerodynamics

Introduction

The Department of Aerospace Engineering at the University of Maryland

has completed a research study for NASA Langley on the application of drooped

leading edges to high aspect wings. The study was supported under NASA

Grant NAG-I-681 with Mr. Daniel DiCarlo as grant monitor. Additional support

for a graduate student (Hugo A. Gonzalez) was obtained from NASA

Headquarters (Underrepresented Minority Focus Program) Grant NTG-70090.

The experimental study conducted for this grant was a natural extension

of work previously conducted at NASA Ames, the University of Michigan,

NASA Langley and the University of Maryland. Previous research had shown

that wing planform modifications (commonly referred to as drooped leading

edge roods) could have a significant effect on reducing or eliminating the

stall/spin characteristics of General Aviation (GA) aircraft. All aircraft studied in

the earlier work had relatively low aspect ratio wings (AR = 6). Since future GA

aircraft will feature higher aspect ratio wings, the obvious question was - "how

well will the dropped leading edge work on higher aspect ratio wings"? The

focus of the current study was to examine the effectiveness of the dropped

leading edge modifications to higher aspect ratio wings with AR = 9 to 12.

Research l_Iighlights

The principal results of this study were presented in the following reports:

Gonzalez, H. and Winkelmann, A.E., "Design of a Three-Component Wall-

Mounted Balance", AIAA Paper No. 90-1397, Seattle, WA, June 18-20, 1990.

Gonzalez, H., "An Experimental Study of Drooped Leading Edge

Modifications on High Aspect Ratio Wings," Master Thesis, Department of

Aerospace Engineering, University of Maryland, December, 1991.

Both reports listed above are currently in preparation for submission to

the AIAA Journal and the Journal of Aircraft for publication. The first report was

presented at the AIAA 16th Aerodynamic Ground Testing Conference held at

Seattle, Washington on June 18-20,1990. A copy of this report is attached. The

principle results of this study are summarized in the following (edited) chapter

of Mr. Gonzalez's thesis:

Chapter 4: Conclusions and Recommendations

A series of force, moment, and surface flow visualization tests were

conducted on reflection plane rectangular wings with NACA 642-415 (Modified)

and NLF (1) - 0414 airfoil sections and with effective aspect ratios of 6, 9, and 12.

The tests were conducted on unmodified (baseline) and modified wings. The

modified wings consisted of leading edge gloves which drooped (leading edge

droops) below the baseline airfoil leading edge. The leading edge droop span

was varied in length. The leading edge extension length was measured from the

wing tip to the discontinuity between the leading edge glove and the baseline

The leading edge droop improved the stall characteristics of the NACA

642-415 (Modified) wings with greater success than the NLF (1) - 0414 wings. The

better performance of the NACA 642-415 (Modified) wings may have been

attributed to the larger droop leading edge radius over the NLF droop radius.

The larger radius allowed for a gentler stall and a large leading edge favorable

pressure gradient.

The use of 3/4 span leading edge droop glove lead to the best

improvement in stall characteristics. A 3/4 glove span reduced the primary stall

of the baseline wing. The loss of lift coefficient at stall was approximately the

same for all aspect ratios. Both the NACA 642-415 (Modified) and NLF (1) - 0414

wings began to recover lift immediately after stall. However, the amount of lift

onaspect ratio. The lower lift recovered by the NLF wings shows the dependence

a droop leading edge has on airfoil shape.

The small loss of lift associated with a 3/4 span leading edge droop

generated a small -dCR/d0_ over a small angle of attack range when compared to

a baseline wing. The small negative change in dCR/d0¢ associated with a 3/4

span droop potentially reduces the divergence of a longitudinal flight path into a

spin. Flight path departure is associated with (large) unsymmetrical stall of an

aircraft's wing. An unsymmetrical stall causes the stalled wing to roll and yaw

due to the lower lift and higher drag - this results in undesired favorable spin

conditions. The small change and short duration in - dCR/dO_ decreases the lift

and drag difference between the stalled and the unstalled wing. The droop

leading edge also causes the port and starboard wing panels to stall at the same

time. Spin characteristics of the different wing configurations could not be

assessed since the force moment data was taken by a static balance.

The rise in lift coefficient after primary stall of a 3/4 span leading edge

droop provides a safety margin in which control of an aircraft can be established.

After control has been obtained, the aircraft's angle of attack can be reduced.

The lift recovered after initial stall can be attributed to the large leading

edge radius of the drooped glove, the discontinuity vortex between the baseline

wing and droop, and the large percent of attached flow on the dropped portion.

Since 75% of the wing leading edge is covered by the glove, only the inboard 25%

of the wing stalls at primary stall while the outer 3/4's of the wing maintained

attached flow except for a small region in the trailing edge. The discontinuity

between the glove and the baseline wing generates a vortex which keeps the

attached and separated flows apart. This vortex generated a "fence" which lets

the outer wing panel act as an independent wing without any large

contamination of the inboard separated flow. The "fence" vortex also acts as a tip

vortex for the outer wing panel which increases its downwash (decreases

effective angle of attack). The downwash generated by the discontinuity vortex

increases in strength with increasing angle of attack since the pressure gradient

between the lower and upper surface increases. Thus, the outer wing panel

behaves as a wing with a large leading edge radius which gradually stalls from

the trailing edge with increasing angle of attack.

If the physical length between two tip vortices of a plane rectangular wing

is increased, the central portion of the wing sees a larger, effective angle of attack

than a smaller wing (due to longer distance between the wing center line and the

tip vortex). Hence, a large aspect ratio plane rectangular wing stalls earlier than a

low aspect ratio wing. The outer droop wing panel experiences the same effect;

hence, the secondary stall of a given droop span occurs earlier for higher aspect

ratio wings.

Minimum drag for a drooped wing increased a small amount over

baseline wing minimum drag. But drag coefficients associated with lift

coefficients less than the CL corresponding to CD,rnin were substantially larger

than baseline wing data. This increase in drag coefficient would provide a

smaller range of cruise CL's.

To extend the secondary stall angle of attack of a high aspect ratio wing

with a 3/4 span droop, a second droop which spans 1/2 - 3/4's of the wing

droop could be added to the first droop. The second droop would have a larger

leading edge radius than the first droop glove. This configuration would

generate two vortex "fences" and three independent stall cell regions which in

turn would generate a triple hump lift curve. Based on the flow visualization

tests, a suggested flow field and wind loading model of a double droop wing at

difference angles of attack was proposed.

An interesting observation was made between the NASA results and the

date of this study. The best droop results obtained by NASA were with a droop.

which covered about half of the aircraft span. This configuration also

corresponded to a droop glove which spanned 3/4's of the wetted wing surface.

This was the same configuration which generated the best results in this thesis.

The above results leads one to believe that the droop span should be sized by

wetted wing surface span rather than absolute wing span.

The studies carried out in this thesis suggest that a droop configuration

does exist which may alleviate stall and longitudinal flight departure for aircraft

with high aspect ratio wings. Further studies to determine the effectiveness of a

double droop wing and whether a low, middle, or high wing has any effect on

drooped wings should be conducted.

-,- -.----- _ --__ i al_lllllmmUmlllmm

am_lmnl_llB_Z u !

AIAA 90-1397

Designof a Three-Component Wall-Mounted

Balance

_ .Gonzalez and A. Winklemann

nlv. of Maryland

College Park, MD

AIAA 16th AerodynamicGroundTesting

Conference

June 18-20, 1990 / Seattle, WA

II III For permissionto copyor republish, contact the AmericanInstitute of Aeronautics andAstronautics

370 L'Enfant Promenade, S.W., Washington, D.C.20024

AIAA-90-1397

DESIGN OF A THREE-COMPONENT WALL-MOUNTED BALANCE Hugo A. Gonzalez* and Allen E. Winkelmann t Department of Aerospace Engineering University of Maryland College Park, MD 20742 INTRODUCTION Abs_Fact The design and evaluation of a three- This paper will discuss the design, component, wall-mounted pyramidal balance fabrication, and testing of a new three- for a small wind tunnel is discussed. The component balance for use in the Aerospace balance was designed to measure lift, Boundary Layer Tunnel at the University of drag, pitching moment, and angle of Maryland. Considering the importance of direct force and moment measurements in attack. The specific design of each component and mathematical models used to wind tunnel testing, it is surprising how design the balance are covered. Balance little information exists on the design of external balances. The literature evaluation consisted of calibration, tare, available on external balances is small and interaction analysis.

when compared to the amount of information available on sting balances. One explanation may be that many balances are Nomenclature typically custom made by companies which keep their design techniques proprietary.

b table width Since little information on external b 2 flexure height balance design is available, the emphasis c restraint coefficient (c = 4 for of this paper will be on design aspects fixed ends) the authors think are crucial to a three- D drag component, pyramidal wall-mounted balance.

E elastic modulus F applied force The balance which is described in I area moment of inertia this paper was designed to be used in K' sensitivity coefficient in units of semi-infinitewing tests with effective load cell signal per interaction load wing aspect ratios up to 12 (Fig. i).

K sensitivity coefficient in units of These tests required a balance with the load cell signal per unit load following minimum requirements: 1 flexure height Le effective flexure link height Test Reuuirements: L load table depth (length) L lift • Maximum Lift Force of 458.17 N M pitching moment (103 Ibs) P applied load • Maximum Drag Force of 240.20 N (54 ibs) t flexure width • Maximum Pitching Moment of 915.18 N. cm x distance along a load table member (81 in-lb) X I unknown normal force • Maximum Angle of Attack Range from X 2 unknown shear force -30" to 60" X3 unknown bending moment • Motorized Angle of Attack Positioning E load cell signal • Capability of Withstanding a 26483.6 u angle of attack N. cm (2344 in'lb) Combined Rolling o stress Moment and Yawing Moment Poisson's ratio v coefficient depending on (b2/(L/_) The new balance was based on an existing (Ref. 7) three-component balance in the Aerospace p radius of gyration (t/4_) Laboratories at the University of Maryland which could not meet the minimum test Subscripts requirements. This existing balance was D drag similar in design to a balance used by the i i th balance component term staff of the High Speed Laboratory, L lift National Aeronautical Establishment of M pitching moment Canada. I Additional references used in designing the new balance were Refs. 2 through 5.

* AIAA Student Member, Graduate Student, phone (301) 454-2922 t AIAA Member, Associate Professor, phone (301) 454-2414 Copyright © 1990 American Institute of Aeronautics and Astronautics, Inc. All rights reserved.

Deflection analysis based on the unit C D S GN load method and nomenclature of Fig. 7 resulted in Eqs. 1 and 2. A series of Figure 2 shows the basic operating calculations to determine load table principle of a wall-mounted pyramidal deflection (q), using Eqs. 1 and 2, were balance. Lift and drag are measured completed for a series of flexures 19.05 directly from two perpendicular load cm long with varying thicknesses and tables which are aligned parallel to the heights, a flow. Pitching moment is read about the balance resolving center. The lift, drag, and pitching moment resolving forces are transmitted to load cells using flexure links.

Based on the resolving and operating

2z2 2z 0

system of a pyramidal balance, the design of the balance was divided into five X2 segments:

II

• Load Table- designed to deflect in the -nTl j t-i-; -iTj t I:, lift and drag planes.

• Cylindrical Core- designed to hold the

"i'TJ tW-{J t{ -ITJ

model, set the angle of attack, and transmit pitching moment.

(i)

• Motor and Angle Measuring Base- designed to set and measure angle of attack.

• Flexure Links- designed to transmit ab2+ alb+ 12b] deflections to load cells.

• Load Cells- measure lift, drag, and pitching moment.

-/ai__bb +ai__ 2 + i__ 3

Figures 3 and 4 show a cross-sectional

( 12 2I I 6I I

view of the balance and a photograph of the assembled balance.

ab+al+ 12 ] Load Table Design A load table responds to a force by deflecting in the direction of the force (as shown in Fig. 5), but resists movement in other directions. As shown in Fig. 6, with the load tables lined up with the

i-#I X 3 + - + + F (2)

wind axis, the lift load table moves

[_2i 2 i' i'b- {_ _} J

q = _ __x2 -_x, vertically, while the drag load table moves horizontally. To accomplish this, the load tables need to be supported on very thin flexures. The drag flexure assembly is mounted directly to the base of the balance, while the lift table Flexure buckling was of major concern assembly rides on top of the drag table. since the resulting rolling and yawing The lift load cell is attached to the test moments put the flexures under compressive section wall and essentially helps support and tensile loads. Figure 8 depicts flex- the weight of the balance (along with the ure loading due to a rolling moment. To determine the critical buckling loads, the drag flexures). Using the lift load cell flexures were modeled as fixed end plate to help support the balance allows one to use very thin lift and drag flexures for columns. Using Eq. 3, several calcula- tests at low speed and low angle of tions were carried out by assuming a attack. Any balance interactions caused column length of 19.05 cm and varying the by this arrangement can readily be flexure height and width. Tension was not accounted for in computer processing of of great concern since buckling occurs the data.

before yield.

With a thin flexure, one must consider the danger of buckling under compressive loads created by rolling and yawing moments produced by lift and drag.

To determine flexure size, two mathema-

(3)

tical models were used: the Unit Load Method to predict deflection 6 and a Plate Column Model to predict buckling 7.

The resulting deflection and buckling lift load cell and motor base, and calculations suggested that a flexure with clearance between the motor and wind a 0.051 cm web (plate column thickness) tunnel wall.

and a height of 4.445 cm was required.

As a safety factor against buckling, the Although Ref. 4 cautions against the web was reinforced at the center, as shown use of bearings in balances, the in Fig. 9, and elongated to 21.59 cm. A authors believe that the use of bearings web reinforcement reduces the effective is a viable option provided that the height of the plate column making it more radial force on the bearings is kept well stable against buckling. Two reinforced below the manufacturer's specified maximum flexures were manufactured and tested for radial force. Moreover, hysteresis deflection and buckling under expected effects due to the bearings may be reduced load conditions. Although the flexures by wind tunnel shaking and wing flutter.

had web reinforcements, they closely The use of ball bearings (to provide the followed the calculated deflection of very small rotational deflection needed unreinforced flexures as shown in Fig. I0.

for the pitching moment load cell to Buckling calculations were validated by respond) leads to a relatively simple applying a 444.82 N load with a moment arm mechanical design when compared with a of 65.405 cm. The flexures showed no sign design using flexures. Ball bearings have of buckling, even when the load swayed been used previously in a number of slightly from side to side.

different balance designs, as noted in Refs. i, 2, and 3.

As shown in Fig. 3, the sleeve Cylindrical Core Assembly housing holds the sting sleeve and turn- table base. The sting is held in place For design purposes, as depicted in with two collars which are attached to the Fig. 3, the cylindrical core assembly was sting sleeve. A maximum sting diameter of divided into six parts: i) sting sleeve, 3.16 cm was incorporated into the design 2) sleeve housing, 3) turntable base, 4) to allow for pressure lines for boundary turntable, 5) pitching moment arm, and 6) layer control, circulation control, and collars. A model is mounted in the surface pressure measurement tests. As balance through the cylindrical core shown in Fig. 12, the pitching moment arm assembly. As a result, the cylindrical on the sting sleeve is connected to a load core assembly must provide good model cell on the turntable via a flexure link.

alignment and be able to withstand all This allows the turntable to set the model aerodynamic forces which are transmitted angle of attack. To change or hold angle through the sting (mounting shaft). The of attack, the turntable is connected to a cylindrical core assembly must also be stepper motor through a plastic cable able to change the model angle of attack chain, as shown in Fig. 13.

either manually or mechanically.

The turntable and its base were The main concern in designing the designed such that the turntable would cylindrical assembly was the sting sleeve rotate freely. This was accomplished by bearing spacing required for model align- ment. An additional concern was the housing a pin roller bearing in the base plate, which mates with the turntable hub.

ability of the bearings to withstand large The turntable diameter was based on the rolling and yawing moments encountered ability of a stepper motor, with a holding when testing high aspect ratio wings with torque of 105.92 N-cm, to support a 915.18 flaps. The importance of model alignment N'cm pitching moment. To prevent the is shown in Fig. ii and explained below.

turntable from wobbling, a raceway for Consider a bearing misalignment of 0.005 0.3175 cm steel balls was cut into the cm (which would be the typical machining turntable and baseplate at a radius of tolerance for this piece) and bearing spacings of 2.54, 15.24, and 20.32 cm. 8.255 cm. The depth of the raceway For point A, which is 111.76 cm from the provides a 0.079 cm spacing between the turntable and baseplate. The turntable is left bearing (distance to the tip of a held in place by teflon covered ball 91.44 cm wing), the corresponding bearings, as shown in Fig. 3. The teflon deflections due to a .005 cm bearing tires are used to prevent the steel misalignment are 0.223, 0.037, and 0.028 bearing from cutting into the aluminum.

cm. This indicates that a separation of 15.24 to 20.32 cm would produce a relatively small induced dihedral or yaw angle compared to the dihedral and yaw Motor and Angle Measurlng Base angle caused by model deflection during tests. The bearings mounted in the sleeve The motor and angle measuring base housing (MPB-3TKCR29-36) have a maximum was designed to serve as a mounting allowable radial load of 4049 N. This platform for a stepper motor, a IK ohm 10- indicated a minimum spacing of 13.08 cm turn potentlometer, and a sprocket for the bearings to withstand the expected ratloing system as shown in Fig. 14.

maximum rolling moment of 26483.6 N'cm.

ASuperior Electric MO93-FCII stepper motor The final bearing spacing was 17.15 cm with a holding torque of 31.777 N'cm was after considering other design aspects attached to the turntable and sprocket such as providing clearance between the ratioing system through a William Berg Co.

Flex-E-Pitch 25CCF plastic cable chain.

To increase the effective holding torque 1-50 m-75 1-150 of the motor, a large turntable to motor 222 333 666 Rsted Capacity sprocket ratio was used (10.05:1). A Newtons sprocket ratioing system which links the 0.03 0.03 0.03 ACCUrBCy turntable and potentlometer via the Flex- X Rsted Output E-Pitch and a Min-E-Pitch 3CCF plastic cable chain was needed to use the full Oef|ect ion st 0.010 ca 0.010 m 0.013 ca Rated Cap.of ty range of the 10-turn potentiometer. The effective rotation ratio of potentiometer -15"to 65" -1S'to 65" -15"tQ 65" Temp. Remge Celsius Cetsius Cetslus to turntable is 38.6 potentiometer C __persated turns/turntable turn. This gives 9.65 Temp. Effect on potentiometer turns for a 90" angle sweep.

0.08 o.o_ o.o6 Rated Output- X of Readfng/55.6"C Table i Load Cell Specifications Flexure Links As shown in Figs. 3 and 4, flexure links are used to transmit load table and BP, IJtN¢8 F_BRZCATION pitching moment arm deflections to the The entire balance was machined out load cells. Flexure links are designed to be strong in tension but weak in bending of 6061-T6 aluminum, except for the sting - if a flexure link transmits a bending sleeve and collars which were made out of moment to the load cell, an erroneous carbon steel. The sting sleeve was made measurement will result. A review of of steel to minimize balance deformation.

flexure links used in Refs. I, 2, and in The lift and drag flexures were machined the Glenn L. Martin Wind Tunnel at the out of a single piece of aluminum to University of Maryland suggested a flexure prevent mechanical slippage under com- column height to width ratio of 6 to I. A pressive and tensile loads.

6:1 (0.952 cm to 0.158 cm) ratio proved to be adequate for lift and drag but not for pitching moment, where the resolving force was considerably less. The pitching BALANCE ELECTRONICS moment link was modified to a 32:1 height to width ratio (2.54 cm to 0.0794 cm) with Figure 15 shows a schematic of the a reinforced center. To prevent flexure electronic equipment used with the link buckling and deformation, the flexure balance. The stepper motor was powered by links were placed in tension when loaded a Superior Electric SPI53B preset indexer.

in positive lift, drag, and negative Measurements Group 2310 amplifiers were pitching moment. Critical buckling load used to power and amplify load cell calculations were based on Eq. 4. 9 The signals. The angle of attack potentio- flexure links were machined from a single meter was powered by a 9.5 volt power piece of 8-32 stainless steel threaded supply. Voltmeters were used to monitor rod.

the amplifiers, potentlometer, and power supply output. Load and angle of attack readings were processed by a DSP A/D converter and an HP-1000, A900 computer.

w2EI Pcr- (4 ) BALANCE EVALUATION Le To evaluate the balance, a series of angle of attack, lift, drag, pitching moment, and tare calibrations were conducted. Tare calibrations consisted of sweeping the balance with and without a model through an angle of attack range Load Cells (with wind off) to determine gravitational effects.

The load cells used in the balance were Interface MB-150, MB-75, and MB-50 The angle of attack, lift, drag, and strain gage, cantilever beam load cells.

pitching moment calibrations were linear, The specified accuracy of the three load repeatable, and showed no hysteresis (see cells is 0.03% of the rated output. Load Figs. 16, 17, 18, and 19). When in use, cell accuracy was based on the maximum the overall balance/data acquisition width of the error band of data scatter system is calibrated before and after each from a load cell callbratlon curve. The test to assure that any drift in the data scatter band includes nonlinearity, overall system can be taken into account.

hysteresis, and nonrepeatability. TM Load Although the long term calibration cell specifications are summarized in stability of the balance has not been Table 1.

assessed, the calibration curves obtained each day during tests lasting several Figure 21 shows the balance pitching weeks were very similar. The small variations that were noted were attributed moment tare with a wing. The pitching moment data follows a cosine curve, which to drift in the data acquisition system.

The basic calibration results showed that corresponds to the wing center of gravity there is no mechanical slippage in the following a circular path. A small hysteresis loop in Fig. 21 is apparent load tables, cylindrical core assembly, or between 0 N-cm and 6.8 N. cm (0 to the angle of attack positioning/measuring .6 in. lb).

system. However, a change in angle of attack due to a negative pitching moment was encountered. The change in angle of attack is attributed to the stretching of Balance Interactions the Flex-E-Pitch chain. Equation 5, based on calibration tests, relates the change As noted in Ref. 4, no balance is in angle of attack to the pitching moment.

free of interactions. Balance interac- tions are both linear and nonlinear.

Linear interactions (first order) are =-1.457xI0 _M 2 ÷ 6.263xI0 qM caused by machining tolerances and

(5) component mlsalignment. Second order

nonlinear interactions are due to the u in degress elastic deformations that modify the M in N'cm geometry of a balance under load. Plastic deformation of balance parts produce nonlinear third order interactions. If plastic deformation is encountered, then a The sensitivity of the balance under balance has been improperly designed. 4 the maximum loading conditions are listed in Table 2.

In the case of a three-component balance, the output signal E i of each load Component Maximum Load Sensitivity cell is a function of all three components (e.g. L, D, and M). An expression for E i Lift 458.2 N 0.0549 N which includes all first and second order interactions is given as Eq. 6. The Drag 240.2 N 0.0186 N equation for the drag load cell signal Pitching 915.2 N. cm 0.112 N'cm (i = D) is given as Eq. 7. The subsequent Moment discussion will be limited to the drag load cell signal. The other component signals are evaluated similarly.

Table 2 Balance Sensitivity Tare calibrations were conducted with and without a wing to determine the offset corrections to be made to the pitching moment data. Tare offsets are due to small pitching moments produced by the weight of the wing. The center of gravity Ei = K_,LL + K_,DD + K_,#M of the wing moves relative to the balance resolving center during an angle of attack sweep.

(6)

+K_,LL 2 ÷ K},o.D 2 ÷ K_,_M2

Figure 20 shows the balance pitching moment tare without a wing. The slight curvature in the plot is due to the moment +K_,0LLD + K_,0, DM + K_.LNLM arm center of gravity following a circular path. The hysteresis loop may be attributed to mechanical run-out between the turntable and turntable base and to hysteresis in the bearings used to support the sting sleeve housing. If the balance had not been used for several days, the initial tare curve was shifted slightly on

E 0 = K_,LL + K_,oD ÷ K_,.M

the plot. However, after an initial angle of attack sweep, a repeat of the calibration showed the same tare curve as

(v)

+K;,LLL 2 + K;._D 2 + K;,..M 2

obtained in previous tests. This effect was apparently due to a slight "sticking" of the bearings that set in after a number

÷K_,oLLD + K_._DM ÷ K_.L.LM

of days. Preliminary tests to evaluate hysteresis in the bearings when a large lift force is placed on the balance indi- cates a similar tare curve with the hyste- resis loop opening up by a factor of two or three. However, since this effect is repeatable a correction for it can be made in data processing.

The coefficients of this polynomial The sensitivity coefficients K' in correspond to K'D, L and K'0,LL.

Eqs. 6 and 7 have units of load cell output signal per unit of interactive The cross product coefficients such load. The principal sensitivity coeffi- as K'D.Q, were also obtained by plotting the cient for the drag load cell is _D; the drag slgnal data against the quantity that other coefficients represent interactions.

was varied in the calibration. The data This creates three first order and six were fitted to a straight line and the second order sensitivity coefficients per cross product coefficient K' was obtained component, as listed in Table 3.

from the slope of the line (i 't order coef- ficient). For example, in the case of the nonlinear linear drag and pitching moment cross product first order second order (DM), the slope is EoM/Mmx (where ED. is the small voltage contribution to E D due to K I ](ID,LL D,L the DM interaction). EDm/M_x was divided by Dmx to obtain K'D.D, as glven by Eq. 8.

K I K' K' D,D D,DD D,OL K' K' K' KID, s D,MM D,LN D,DI4 EDS K_,D. - Dmax Mm------- _ Table 3 Linear and Nonlinear Sensitivity

(8)

Coefficients For the Drag Signal i st Order Coefficient * Mm. x Dmax Mmax An interaction calibration was conducted to obtain the K' coefficients.

The calibration consisted of the loading configurations listed in Table 4 and To obtain D in terms of engineering explained below. This loading procedure units, Eq. 7 was divided by the principle closely followed the technique described sensitivity coefficient K'00 _ which is in in Ref. ii. The balance was first loaded units of load cell signal per unit load.

in pure lift or drag in five equal The final result for D is given in Eq. 9.

intervals to the maximum expected load.

The corresponding expressions for L and M When calibrating for pitching moment, a are given in Eqs. i0 and ii. Eqs. 9, I0, small secondary load (L_x/10) was moved and II cannot be solved directly because along a moment arm to produce a range of the loads appear on both sides of the pitching moments from 0 to Mmx. This equations. Instead, an iterative secondary load resulted in loading the technique is required where the initial or balance in a small, but constant negative raw data for L, D. and M are used to start lift. For combined loads such as LD, the the calculations. _ The sensitivity coef- secondary load Lm_ is held constant while ficients for the balance in this paper are the primary load is varied from 0 to Dmx summarized in Table 5.

in 5 equal increments.

Secondary Term Be|rig Primry load added Evatuated E 0 (constant) D- Ko.LL - Ko..M L,L 2 L None K6..

D,D 2 D None

(9)

-Ko. LL L2 - KO,OD D2 - K0,_M 2 M,N 2 N lmax/10 -Ko,0LLD - Ko,D. DM K0,LN is LD D Line x LM M Lmax/lO, Lma x M DN Lmx/10, Dma x E L L=-- - KL,oD - KL,.M Table 4 Loading Configurations KL,L The drag signal data from each of the _KL,LLL2 _ KL,00D 2 _ F_,_M 2 (i0) three primary loading tests (L, D, and M) were used to get the K' coefficientg generally referred to as the first order -KL,DL LD - KL,DM DM - KL,L. LM and quadratic coefficients. For example, the drag signal data obtained in the pure lift calibration were plotted against lift and fitted to a second order polynomial.

helping complete this project. This work was sponsored by NASA Langley Research Center. grant No. NAG-I-681 and NASA Head-

Ew

M=-- - KN.LL - KN,DD quarters (Underrepresented Minority Focus Program) grant No. NTG-70090.

K_,.

(11)

-KW,LL Lz - K.,00 Dz - KW,NNM z

References 1.

"Development of Half-Model Wind Tunnel -I._,DLLD - I'_,DMDM - Y,_,LMI.,.M Balance", AGARD report i0, February 1956.

2. Lambourne, N. C.; "A Note On a Half- Model Strain-Gage Balance", AGARD L D M report ii, February 1956.

Ki, t 1.78 x 10"2 1.85 X 10.2 3_ Huber, Arthur F., If; "Airfoil Moment Coefficients at Low Reynolds Numbers: Ki, D -5.07 x 10.3 -3.56 x 10.2 The Design, Fabrication, and Testing of a Simple Torquemeter", AIAA Midwest Region Student Conference, March 1985.

gi, M -1.39 x 10 .4 -2.48 x 10.3 4. Rae, William H., Jr. and Pope, Alan; Ki,LL 1.95 x 10.5 -3.23 x 10"6 °4.83 x 10.5 Low-Speed Wind Tunnel Testinu, John Wiley & Sons, New York, 1984, ICi,DO -8.42 x 10.6 1.52 x 10.4 3.70 x 10.5 pp. 152-161.

gi, _ 4.82 x 10.7 1.14 x 10.5 -1.78 x 10.5 5.

Bardowicks, H.; "A New Six-Component Balance and Applications on Wind Tun- Ki,Ot 8.94 x 10.5 3.88 x 10.4 -7.58 x 10.4 nel Models of Slender Structures", Journal of Wind Engineering and Indus- trial Aerodynamics, Vol. 16, No. 2, Ki,LM 4.07 x 10.6 -2.88 x 10.5 5.02 x 10.4 April 1984.

K_,DN 3.28 x 10-5 1.35 x 10.5 1.96 x 10.4 6.

Allen, David H. and Haisler, Walter E,; Introduction to Aerospace Table 5 Interaction Coefficients Structural Analysis, John Wiley & Sons, New York, 1989, pp. 307-311.

7, Houbolt, John C. and Stowell, CONCLUSION Elbridge Z.; "Critical Stress of Plate Columns", NACA TN-2163, August 1950.

This paper has discussed the design of a new three-component wall mounted 8. Gonzalez, Hugo A.; "An Experimental pyramidal balance. The balance was speci- Study of Droop Leading Modifications fically designed for testing high aspect on High Aspect Ratio Wings", Masters ratio wings with flaps. In addition, the Thesis, University of Maryland balance has been configured to allow (pending).

future work with high C , boundary layer , . t control, and circulation control models. 9. Beer', Ferdinand P. and Johnston, E. Russell; Mechanics of Materials, The balance may be modified for low angle of attack and low Reynolds number tests by McGraw Hill Book Company, New York, changing the load cells and reducing 1981.

flexure thickness. The specifications for the balance described in this paper are: i0. Interface Pamphlet 12-10G, 1985.

• Maximum Lift of 667.23 N (150 ibs) 11. Hausen, Raymond M.; "Evaluation and • Maximum Drag of 333.62 N (75 ibs) Calibration of Wire-Strain-Gage Wind • Maximum Pitching Moment of 3186.1 N'cm Tunnel Balances Under Load", (282 in-lb) AGARD report 13, February 1956.

• Angle of Attack Range from -30 ° to 60 ° • Capability to Withstand a 26483.6 N'cm 12. Dubois, M.; "Six-component Straln-Gage (2344 in'ib) Rolling and Yawing Moment Balance for Large Wind Tunnels", Proceedings of the Fourth SESA International Congress on Experimental Mechanics, Boston, May 25-30, 1980.

ACKNOWLEDGEMENTS 13. Dubois, M.; "Calibration of Aero- The authors wish to extend a special dynamic Dynamometers and Balances at note of thanks to: Kevin Bruestle, Victor the Modane-Avrieux Test Center", Pre- Hwang, Ian Matlick, Paul Vieira, Taylor sented at the Mesucora Congress, Hale, Dan Skane, and the University of Paris, April 17-21, 1967, (N68-32016).

Maryland Physics shop for the hard work in

3-Component_ 7

Balance /

Boundary Layer

Plate

.j

! \ /_,°, l i " m _

I

115.61 _.cm _-- i 91.44 cm

/

_- 15.24 cm

\

Pitching Moment ' Resolving Force Flexures Lift -- Load Tables

44.0 cm

Fig. 2 Pyramidal Balance Resolving System Fig. 1 Wind Tunnel Set-Up Flexure Hold Down with Bearing Raceway" Pitching Moment Load Cell

/

Orag Flexure / 14kxle1 Sting I Test_ Section _-L gall Sting Sleeve j e Sleeve Housing __I Base

Stepper

Motor Load Cell :xure llnk Potenttometer Sprocket

Fig. 3 Cross Sectional View of Balance

Fig. 6 Load Tables Fig. 4 3-Component, Wall-Mounted, Pyramidal Balance

b "1

,i

_, F,q

®

a_!.27cm

LJ

)F

h

1 @

1_

NN"_ ,,N'_ Fig. 7 Unit Load Method Nomenclature Fig. 5 Load Table Deflection

ORIGINA£ PAGE

BLACK AND. WHITE PHOTOG_R.APH. -_ --_i

Lift Unreinforced Reinforced Fig. 9 Unreinforced and Reinforced Flexures .04 , Trl I_I rT I T--I_IrT[rrT I--|! l-rT] T_rl_r |I -i--l'-l-Flrrl-T _r T_rl"_ '"O'" THEORY (Eq$. I a 2) 1 o" / d/x EXPERIMENT :_ 0 .- ._ e- .."" "l r_..03 C 0 ."" :J v 0"""°"" 1 z ..,'"

_o .02

0..'" -_ I-- c .4 U IM

..9," 1

--I L=.

.0""' _ IM .01 Flexure Tension Compression ..

i'i i * [! JL_ _I I LLJ] [dl IJ [* LJ I l_.' I Jll I ,IJ. I llZ I I. [J. l._J.

Fig. 8 Flexure Loading Due To Rolling 0 1 2 ,3 4 5 6 7 8 9 Moment

APPLIED LOAD (Ibs.)

Fig. i0 Load Table Deflection I0

111.76 cm ._ I < Bearing Spacin 9 )I

Displacement

Displace- Bearing ment Spacing 2.54 cm 0.223 cm 15.24 cm 0.037 cm O. 028 cm 20.32 cm Fig. ii Model Alignment Example Fig. 12 Balance Cylindrical Core with Load Cell and Flexure Link 13 Turntable/Motor and Angle Fig.

Measuring Base Set-Up

ORIGINA[ PAGE

II

BLACK AND WHITE PHOTOGRAPH

ORIGINALPAGE

BLACK AND ,,,_ruiTr,.-,- PHOTOGRAPH

g llt tl 1111|tI rlrT vT T|TTy;T_ r I rl? i w y.llvT| 11 Tr;Tr,1 [IrrT_Tr_ "6 I P, ,,,,, +,, | +,,, ,, ,,, I ,,,, +,*,, I,,, 1,, +,+ h + + LU,,+I,, _, ,IU_J -10 0 10 20 30 40 50 ANGLE OF ATTACK (DEG) Fig. 16 Angle calibration 9 I_'_T*-T_ 1 ' T _ I I 1--1"-1 f-I-1-"1 rl T_rT-r--r-1_r--r-_r-_ Fig. 14 Motor and Angle Measuring Base

//i

>

i i.i i,iLii

0 10 20 30 40 50 60 70 80 90 100110120130140150 LIFT LOAD (lbs.)

Fig. 17 Lift calibration

(

:Iz / . I°NI°*° i

01___ 1 • I ,_ I _ I L I_ L 1 J [ L_--I 0 10 20 30 40 50 60 70 80 DRAG LOAD (Ibs.)

Fig. 15 Balance Electronics Set-Up Fig. 18 Drag Calibration i E" ' "1 tl 'l -lr -T---T "1-"1---1 r----] -- I T--T-] >.

_, _ I , l _ J j I , I , • • I ._'1RI11 --80 --70 --60 --50 --40 --30 --20 --10 0

PITCHING MOMENT LOAD (in-lb)

Fig. 19 Pitching Moment Calibration , , , ' I , . . ' | ' ' ' ' I ' ' ' ' | ' ' ' ' I ' ' ' " I E m v 2nd TARE i.i _ 1st TARE 1 rw 3rd TARE Z 0 f-_--,m.-_j_. _ - I,I o o z u I,-- m 1--1 --I -I--L--I,I]_ jJII_[_U±IJL*.*IJIIJ]*JIIJJttL -20 - 10 0 10 20 30 40 50 60

ANGLE OF ATTACK (DEG)

Fig. 20 Pitching Moment Tare Without a Wing (Magnified Scale) m I I- '"0"" COSINE CURVE TARE WITH WING ,., 0 tw <[ I.- I-- Z --1

L,.I "%

_E

%

"0 _E

o

(.9 -2 o.

Z

b

"1- I-- _. -3 -20 -10 0 10 20 30 40 50 60

ANGLE OF ATTACK (DEG)

Fig. 21 Pitching Moment Tare with a Wing

NOTES