/'/'9-3&q

NASA Technical Memorandum 105977AIAA-93-0398

Icing Effects on Aircraft Stabilityand Control Determined FromFlight Data

Preliminary Results

T.P. Ratvasky and R.J. RanaudoLewis Research CenterCleveland, Ohio

Prepared for the31st Aerospace Sciences Meeting and Exhibitsponsored by the American Institute of Aeronautics and AstronauticsReno, Nevada, January 11-14, 1993

NASA

https://ntrs.nasa.gov/search.jsp?R=19930005642 2018-02-12T05:49:48+00:00Z

ICING EFFECTS ON AIRCRAFT STABILITY AND CONTROLDETERMINED FROM FLIGHT DATA. Preliminary Results.

T.P. Ratvasky, and R.J. RanaudoNational Aeronautics and Space Administration

Lewis Research CenterCleveland, Ohio 44135

Abstract:

The effects of airframe icing on the stability and control characteristics of the NASA DH-6 Twin Ottericing research aircraft were investigated by flight test. The flight program was developed to obtain the stabilityand control parameters of the DH-6 in a baseline ("united") configuration and an "artificially iced" configuration

for specified thrust conditions. Stability and control parameter identification maneuvers were performed overa wide range of angles of attack for wing flaps retracted (0°) and wing flaps partially deflected (10°). Enginepower was adjusted to hold thrust constant at one of three thrust coefficients ( Cr =0.14, Cr =0.07, C, =0.00).

This paper presents only the pitching- and yawing-moment results from the flight test program. Stabilityand control parameters were estimated for the uniced and artificially iced configurations using a modifiedstepwise regression algorithm. Comparisons of the uniced and iced stability and control parameters arepresented for the majority of the flight envelope. The artificial ice reduced the elevator and rudder controleffectiveness by 12% and 8% respectively for the 0° flap setting. The longitudinal static stability was alsodecreased substantially (approximately 10%) because of the tail ice. Further discussion is provided to explainsome of the effects of ice on the stability and control parameters.

List of Symbols and Abbreviations:

A,,A,,,AZ longitudinal, lateral, and vertical p,q,r roll rate, pitch rate, and yaw rate,accelerations, respectively, g units respectively, rad/s or deg/s

b wing span, m R'- squared multiple correlation coefficientC. Iq pitching-, yawing-moment coefficients, S wing area, of

respectively V total airspeed, m/s or knotsC,,C,,, initial conditions of pitching-, yawing- (t) measured aircraft response or control

moment coefficients, respectively surface inputCr thrust coefficient y(t) computed aerodynamic coefficientsT mean aerodynamic chord, m a angle of attack, rad or degcg aircraft center of gravity angle of sideslip, rad or degg acceleration due to gravity, m/s 4 change in parameter due to ice shapeklk l lz moments of inertia about roll, pitch, 6 ,6,,5, elevator, aileron, and rudder

and yaw axes, respectively, kg-mZ deflection, respectively, rad or degT„Z product of inertia, kg-mz 6, flap deflection, rad or deg

JMSR cost function for modified stepwiseregression algorithm

aerodynamic stability and control

parametersconstant offset term

derivative with respect to jth response orcontrol surface inputpitch and roll angles, respectively, rad ordeg

standard deviation estimate of measure-ment noise

AMI Analytical Methods, Inc.LeRC NASA Lewis Research CenterMSR modified stepwise regressionPID parameter identificationWFF NASA Wallops Flight Facility

A

80

6i

0,0

C

Pitching moment derivatives:

c = ac,^ c = ac. c =.

as ' qc ' aa^a-2V

Yawing moment derivatives:

c =

ac n' c _ ac^ c _

ac"

^o a p ' a pb ' a rb

2V 2V

acC =nyaar

2

Introduction:

A principal objective of the NASA LewisResearch Center's (LeRC) icing research programis to develop and validate computational methodsfor predicting the effects of airframe icing onaircraft flight characteristics. Recently a multi-yearcontract was awarded to Analytical Methods, Inc.(AMI) to develop a three -dimensional flow codeto predict the stability effects and performancelosses of a complete aircraft with known icecontamination. The code will be validated withwind tunnel data supplied by a large commercialaircraft company and full-scale flight data fromNASA LeRC's icing research aircraft.

NASA's DeHavilland DH-6 Twin Ottericing research aircraft was utilized in an extensiveflight test program to acquire the needed flight database. Although the DH-6 does not represent thegeometry of a modern swept wing transport, it doesprovide numerous advantages in terms of it's lowoperating cost, simplicity, adaptability for theintended experiment, and known flightcharacteristics from previous stability and controlflight experiments'. Also, a digitized panel modelof the DH-6 that was previously developed', wasavailable for performing the AMI code validation.

Relying on extensive past experience inmeasuring icing effects on aircraft performance,stability and control ', NASA developed a com-prehensive flight program to generate the requiredaerodynamic data base. The key elements of theflight program include:

1. A high-fidelity, onboard instrumentation systemwith complete system characteristics knownfrom comprehensive ground and flightcalibration tests.

2. Use of measured aircraft weights, center ofgravity, and moments of inertia for datareduction and analysis programs.

3. Flight testing at specific thrust settings (C r =0. 14, Cr = 0.07, Cr = 0.00) to distinguish power

effects from icing effects on the derivativeestimates. Previous experience indicated thatpower had a measurable effect on certainstability and control parameters °.

This report is organized in sections whichdescribe the research aircraft, instrumentationsystems, flight test procedures, data analysismethods, a discussion of the derivative estimates,and conclusions drawn from the work. Because ofthe volume of data acquired in the flight program,and the need for brevity in this particular reportformat, only the pitching- and yawing-momentresults are presented. A complete data report willbe written at a later time.

Research Aircraft:

The NASA Lewis icing research aircraft isa modified DeHavilland DH-6 Twin Otter (figure1). It is powered by two 550 SHP Pratt andWhitney PT6A-20A turbine engines drivingthree-bladed Hartzell constant speed propellers.The flight controls are mechanically operatedthrough a system of cables and pulleys. Controlsurfaces consist of elevator, ailerons, rudder, andwing flaps. Physical characteristics of the aircraftare in Table 1.

The DH-6 was tested in a baselineconfiguration (clean airfoils), and an artificially icedconfiguration (figure 2). Artificial glaze ice shapeswere attached to the leading edges of the horizontaland vertical stabilizers only. No other surfaces werecontaminated. The ice shapes were determined bycombining the geometry from actual tail icephotographs with a well-known ice area calculationprocedure '. The ice shapes were cut fromStyrofoam blocks, and attached to the leading edgesof the tail with double sided tape. The ice shapesdid not incorporate surface roughness or 3D effects(scalloping).

Extensive ground tests were conducted onthe DH-6 to obtain the center of gravity (cg) andmoments of inertia along the longitudinal, lateral

3

and vertical axes. These characteristics varied with

fuel and crew loading and any modifications thatwere made. Aircraft configuration was closelymonitored to account for any changes in center ofgravity and moments of inertia.

For that phase of testing when the DH-6engines were shut down while in flight, an auxiliarypower unit (APU) was installed to supply theresearch equipment with electrical power. Theaircraft was modified structurally and electrically toaccommodate the APU.

Instrumentation System:

The stability and control data system flownon the Twin Otter incorporated the followingcomponents: inertial data, air data, control surfacedeflection data, signal conditioning, data acquisitionand recording systems. The inertial sensorsconsisted of three orthogonally mounted linearaccelerometers, three orthogonally mounted angularrate gyros, and a vertical gyro to provide pitch androll angle data. These sensors were near theaircraft center of gravity. Yaw angle data wasprovided by the ship's directional gyro. Air dataconsisted of airspeed, angle of attack, angle ofsideslip, pressure altitude, and outside airtemperature. All air data parameters (except OAT)were sensed by a Rosemount 858 probe headextended from the aircraft on a 9 foot noseboom.The control surface deflections, (6., 6,, 6,), weremeasured using linear control position transducers(CPT's) located near the control horns whicheliminated cable stretching errors. Transducersignals were amplified and filtered by a PrecisionFilters System 6000 unit and then digitized with aKeithley series 500 data acquisition system. A totalof 26 channels of data were digitized at anacquisition rate of 100 samples/second and a 12 bitresolution. A ruggedized AT-class microcomputerwas used to control the data acquisition system, anda removable hard drive in the computer provideddata storage. See table II for instrumentationspecifications.

Extensive calibrations were conducted on

all components of the data system. Individualcalibrations were conducted for all sensors, andthrough-put calibrations (sensor-filter-dataacquisition-recording) were run where possible.Calibrations were checked periodically during theresearch program.

Flight Test Procedures:

Aircraft weight and balance weredetermined before each flight by weighing the fullyfueled aircraft less crew. In flight, fuel totalizersprovided an accurate measure of fuel burned.These readings were used for cg and moment ofinertia calculations in the post flight data processing.

Flight testing in the two configurations(baseline, artificially iced) was performed with wingflaps retracted (6 F =0°), and with wing flaps partiallyextended (6, =10°). Test point airspeeds wereselected to cover the range of angles of attack ineach configuration from maximum cruise airspeedto near aerodynamic stall. Parameter identification(PID) maneuvers consisting of elevator doublets(figure 3) and rudder-aileron doublets (figure 4)were used to excite the required aircraft response.

To determine the power effects, each PIDmaneuver was flown at three target thrustcoefficients (Cr ): Cr =0.14 (high thrust), q=0.07(low thrust), and C,=0.00 (engines off andpropellers feathered). To attain the target thrustcoefficients in the powered cases, a simple flightprocedure was developed. Initially, an altitudewould be selected where flight conditions weresmooth. This became the reference pressurealtitude for all flight maneuvers. The outside airtemperature was also recorded. Based on thepre-planned indicated test airspeeds, and a constant1800 propeller rpm, the required engine torquepressure settings were calculated from knownrelationships between thrust coefficient, enginepower coefficient, propeller advance ratio, andpropeller efficiency. Normally, the target thrust

4

coefficients did not provide level flight conditions atthe trimmed test airspeeds. Consequently, the PIDmaneuvers were usually performed in shallowclimbs or descents as the aircraft reached thereference pressure altitude. Generally, all PIDmaneuvers were accomplished within ±200 feet ofthe reference altitude.

The tests at C,. =0.00 were performed at theNASA Wallops Flight Facility (WFF), where arestricted test area, tracking support, and adedicated landing site were employed for flightsafety reasons. In performing these tests, the DH-6departed WFF and climbed to a pre-plannedaltitude and position which was within safe glidingdistance to the landing field. While being trackedon radar, the engines were shut down and thepropellers feathered. Test airspeeds were attainedby establishing the proper flight path. PIDmaneuvers were then executed while the onboarddata system, powered by the auxiliary power unit,recorded the flight data. The glide was terminatedand the engines re-started when a specifiedminimum altitude was reached.

Data Analysis:

Measured data were recorded with theonboard data acquisition system in a binary format.Data were post-test processed into engineering unitsand corrected for instrument offsets from theaircraft center of gravity, position errors in airspeedand altitude, and upwash effects in the angle ofattack measurements.

Each flight contained a series of datacompatibility maneuvers designed to verify sensorand data system integrity. These maneuvers wereanalyzed using a maximum likelihood algorithm' toestimate bias and scale factor errors in the datasystem. The analysis indicated a small sidewashcorrection was required in the sideslipmeasurement. This correction was made prior tothe stability and control parameter estimation.

The stability and control derivatives wereestimated using a Modified Stepwise Regression(MSR) technique'. MSR is a version of linearregression which can determine the structure of theaerodynamic model, and estimate the values of themodel parameters. The general form of theaerodynamic model is as follows:

Y(r) = e0 +0 1 MO + 92x2(r) + ... + Oj.(I) (1)

y(t) represents the aerodynamic force ormoment coefficient and is known frommeasurement.eo is a constant corresponding to the initialflight condition.

8 1 to On are constant coefficients known asstability and control derivatives.

x, (t) to Y,(t) terms represent the measuredinput and output variables, or theircombinations (regressors).

MSR determines the model structure oneterm at a time. Each new term enters into theregression equation based on the largest correlationwith the dependent variable, y(t), after adjusting forthe effect of the previously selected terms on y(t).Essentially, the first regressor ) (t) is selected basedon the highest correlation with the dependentvariable y(t). A constant value, 8., is determined tominimize the squared difference in the measuredaerodynamic coefficient y(t), and the modelprediction 80 + 8J- ) (t). It means that at each stepof the MSR, the parameters are obtained byminimizing the following least squares cost function:

N 1 2

.)MSR = ly(1) - e0 - E e,Xi(1)1 (2)=t i-1

N is the number of data points

1+1 is the number of parameters in theregression equation.

5

In addition, at each regression step theinfluence of individual derivative/regressor pairs onthe model is re-evaluated. The estimatedparameters, 6'., may be retained, or removed fromthe model due to their statistical significance. Theprocess of adding and deleting terms to the modelcontinues until no further significant terms can beadmitted to the model, and no further insignificantterms can be removed.

Models based solely on significance ofindividual parameters has proven to contain toomany terms for good predictability'. Criteria forelecting adequate models are the squared multiple

correlation coefficient (W), and the F-statistic value.Che W value indicates the percent of variationexplained by the model. An W close to 100%'<nggests the model perfectly fits the measured data.The F-statistic value is the ratio of regression mean<quare to residual mean square. The model withthe maximum F-value has been recommended asi he 'best' one for a given set of data. 9 Bothcriteria were used in this analysis.

provide an ensemble of data. From the ensemble,a better measure of the variance of the derivativeswas made.

The stability and control derivatives areplotted with respect to trim angle of attack infigures 5-14. For each parameter estimate, errorbars representing 2a variance determined by MSRare included. To clarify trends between the baselineand iced configurations, a third order polynomialregression with respect to angle of attack wasperformed for each ensemble of data. Along withthe regression line, a 95% confidence bound on themean was included to evaluate the statisticalsignificance of configuration change due to icing.

Longitudinal Model:

The pitching moment coefficient wasadequately modeled with the following equation:

C. = Cno + C.. a + C ,' '7c + C. 6e (3)2V

Results:

The analysis performed for this report waslimited to the pitching and yawing momentcoefficients. Ice on the horizontal and verticalstabilizers strongly affects these moment coefficientsbecause the moments are predominantly created bythe lift generated from these tail surfaces. Icecontamination may reduce the maximum lift andlift-curve slopes of the stabilizers, which could resultin lower stability of the aircraft. Separated flowsbehind the ice shape may decrease the effectivenessof control surfaces, which results in reducedcontrollability of the aircraft.

The effect of ice on the pitching and yawingmoment coefficients is evaluated by comparing thevalues of stability and control derivatives for boththe baseline (uniced) and iced cases. Each stabilityand control derivative and its standard error wasestimated using the MSR technique describedabove. Each flight condition was repeated to

The suitability of the model was determinedchiefly by the R= value and the F statistic value fromthe MSR program. For all data analyzed here, theW >_ 90 %, which indicates that over 90%v of thevariation of Cm was described by this model.

The bias term C,,,o ^ 0 for all data runsbecause the elevator doublet inputs were initiatedfrom trimmed conditions.

The derivative, CC„., is known as the staticlongitudinal stability derivative. A statically stableaircraft will have a negative C,. indicating that anose down moment is produced with a positivechange in angle of attack. A more negative C,.implies greater static longitudinal stability.

Figure 5 presents the effect of the tail iceon the static stability derivative with flaps retracted(br =0°) for three thrust coefficients. Static stabilitywas reduced by approximately 10% for each Cr.

6

Because of the exceptional repeatability in thesedata sets, the reduction in static stability due to tailice was determined with high accuracy.

The effect of ice on Cn. with the flapsdeflected to 10 degrees (6 F =10°) is shown in figure6. Similar to the S r =0°case, a reduction in staticstability occurred because of the ice. However, theeffect of ice varied with the thrust setting. Staticstability was reduced by approximately 8% in thethrust cases, and 17% in the zero thrust case.Because of scatter in the derivative estimates, thereduction in Cna for the low thrust case (CF =0.07)was not statistically significant (i.e. the confidencebounds overlapped for most the range tested).Note that the least statically stable conditionappeared at low angles of attack with flaps extendedand a high thrust coefficient.

The derivative, C„ q , is known as the pitchdamping derivative. As the aircraft pitches, amoment is created usually countering the pitchingmotion. The horizontal tailplane is the primarycontributor to the pitch damping.

Figure 7 shows the effects of the ice on C.1,with flaps retracted for three thrust coefficients. Inthe cases with thrust, the trends indicated a slightlylower pitch damping occurred due to the ice(A =5%). This reduction was statistically significantexcept at the low angles of attack where theconfidence bounds intersected. For the zero thrustcase, the pitch damping was virtually unaffected bythe ice except at low angles of attack. Theconfidence bounds overlap for nearly the entirerange of angles of attack tested. This result seemedinconsistent with the Cma results because a loss instatic longitudinal stability should also result in areduction of pitch damping. Further discussion onthis point will follow.

Figure 8 presents the effect of ice on C„,with the flaps deflected to 10 degrees OF =10°). Aswith the 6 F =0° cases, pitch damping was reduced by5% to 9% in the high and low thrust cases.Overlap in the confidence bounds occurred in thehigh thrust case at low angles of attack, but no

overlap occurred in the low thrust case. For thezero thrust case, pitch damping was unaffected bythe tail ice. The confidence bounds overlap for theentire angle of attack range tested, indicating thatno change in Cn, can be attributed to tail ice.

The derivative, C,,,,, is the elevatoreffectiveness control derivative. It is usually anegative value so that a positive elevator deflectionresults in a nose-down (i.e. negative) pitchingmoment. The more negative value for Cn,indicates a more effective elevator.

In figure 9, the effects of tail ice on Cm,are shown for flaps retracted and three thrustcoefficients. Because of the linearity in this dataset, a first order regression was performed toindicate the trends. A clear separation existsbetween the baseline and iced cases for each thrustcoefficient. The tail ice caused an approximate 12%loss in elevator effectiveness over the entire angle ofattack range tested. Also, note that the slope of

C , with angle of attack changes with thrustcondition. Engine power is clearly a factor in theelevator effectiveness with or without ice on the tail.

Figure 10 shows the effect of ice on C,,,,with the flaps deflected to 10 degrees (6, =10°). Aswith the 6 F =0° cases, the ice decreased elevatoreffectiveness for all thrust settings (A­: 16%). Also,the flaps appear to further decrease the dependenceof elevator effectiveness with angle of attack.

Lateral Model:

The yawing moment coefficient wasadequately modeled with the following equation:

C. = C + Cn p + Cn pb + Cn rb + C^ 8r (4)no 0 /2V 2V

The suitability of the model was determinedchiefly by the W value and the F statistic value fromthe MSR program. For all data analyzed here, the

7

RZ >_ 90%, which indicates that over 90% of the

variation of C, was described by this model.

The bias term C, - 0 for all data runsbecause the rudder and aileron doublet inputs wereinitiated from trimmed conditions.

In figure 13, the effect of ice on C,,, areshown for flaps retracted (8, =0°) and three thrustconditions. Although the data are fairly repeatable,the effects of ice on this derivative are negligible ineach thrust case.

The derivative, C,, a , is the directionalstability derivative ('weathercock" stability). Adirectionally stable aircraft will have positive C,eindicating a positive yawing moment is producedwith a positive change in sideslip angle. The yawingmoment will rotate the aircraft so as to decrease thesideslip. A more positive C,, b implies a greaterdirectional stability.

Figure 11 presents the effect of the verticaltail ice on C, a with flaps retracted (6, =0°). Forhigh and low thrust cases, the trends indicate thatice slightly decreased directional stability. However,the decrease is insignificant since the confidencelimits overlap for most of the tested angle of attackrange. For the zero thrust case, the ice decreasedthe directional stability (A=20%) for all but thelowest and highest angles of attack. Also note thatdirectional stability is greater with zero thrust thanwith power on.

The derivative, C„r , is the directionalcross-derivative. It indicates the rate of change inthe yawing moment due to roll rate. For mostaircraft, C^ is usually negative and of low value.

Figure 12 shows the effect of ice on C,Pwith flaps retracted (8,=O') for three thrustconditions. For the zero thrust case, C,, was not asignificant term in the MSR model except at higherangles of attack. Regardless of the thrust condition,the effects of ice are negligible on Cam.

The derivative, C,,,, is the yaw dampingderivative. It indicates the rate of change in theyawing moment due to yaw rate. The verticaltailplane is the primary contributor to the yawdamping. A larger vertical tail will likely increaseyaw damping characteristics.

The derivative, q,,, is the ruddereffectiveness control derivative. It is usually anegative value so that a positive rudder deflectionresults in a negative yawing moment. A morenegative value for q, means a more effectiverudder control.

Figure 14 presents the effect of ice on Crt,,,with flaps retracted at three thrust conditions. Tailice substantially reduced rudder effectiveness(A=8%) in all thrust cases except for the higherangles of attack in the zero thrust case. This maybe a result of low dynamic pressure at the rudder isthe zero thrust condition.

Discussion:

From the results presented, it is clear thatthe ice on the tail surfaces considerably affect someof the aircraft stability and control parameters.Aircraft longitudinal static stability was reduced.Ice contamination reduced the horizontal tail'smaximum-lift and lift-curve slope which resulted ina decreased pitching moment capability. Elevatorand rudder control effectiveness decreased. Flowdisturbances caused by the ice may have resulted inlower dynamic pressure at the control surfaceswhich decreased the effectiveness.

Other parameters indicated no change dueto ice. The directional cross derivative, Cam , anddirectional damping derivative, C,,,, showed virtuallyno difference in the iced configuration. Thisindicates that these derivative were not sensitive toicing on this aircraft.

Two parameters, C,,„q and C,, a , had mixedresults. Ice caused the pitch damping, C", toslightly decrease with engine power on, but wasunaffected by ice in the C, =0 cases. Ice on the

8

horizontal tail decreased the tail lift coefficientwhich should have caused a change in both the pitchdamping and longitudinal stability, C,.. The tail iceclearly affected the C,, in each thrust case, sosimilar changes were expected in C4rq . On further

examination, Cam , was found to be much moresensitive to changes in the tail lift coefficient thanC,,,. As a result, C, may appear unaffected by theice when Cm , was affected by the ice as wasdemonstrated in the Cr = 0 cases. The secondparameter which showed mixed results was thedirectional stability derivative, q,. It showednegligible changes due to tail ice with engine poweron, but as much as 20% change in the zero thrustcase. Also, C, p was greater in the zero thrust casefor both baseline and iced configurations. Onepossible explanation is that the propeller wash hada more destablizing effect than the ice.Consequently, the effect of ice with engine poweron appeared insignificant. Another possibility maybe a shortcoming in the analysis technique. Theeffects of the vertical tail ice may not be apparenton q, until P > 6°, but in these tests each q, wasevaluated about a P^0°. Analysis techniques areavailable to separate the data that was collected intosmall bins of fi so that Cs,p can be evaluatedspecifically at higher sideslip angles. Thesetechniques will be applied for a future report.

Another interesting observation was theeffect of aircraft configuration and flight conditionon static longitudinal stability, C,.. Ice was shownto destablize the aircraft by 10% in the S F =0° casesand as much as 17% in the S F =10° cases. But theaddition of flaps alone destablized the aircraft in thehigh thrust cases at lower angles of attack (seefigures 5 and 6). For the uniced Cr =0.14 cases, atan a =0*, C,, = -1.5 for 6, =0*, and C,. = -0.9 forS F =10°. This example illustrates a 40% reductionin static longitudinal stability with flaps deflectedonly 10°. Tail ice in conjunction with the 10° flapdeflection decreased static longitudinal stability by50% for the same condition. It may be inferredthat if flap deflection was increased beyond 10° andthe aircraft flown at a =0°, the static longitudinalstability would be lowered even more. The effect oftail ice with flaps deflected could drive the aircraft

to a neutrally stable condition (C; . =0). It isimportant to note that a high thrust condition withlarge flap deflections at low angles of attack maycause serious stability problems.

Finally, it also should be noted that theDH-6 is a short takeoff and landing (STOL)airplane. The tail surfaces are designed for lowspeed operations and tend to be oversized for theconfigurations and flight conditions that were tested.The changes in stability and control derivatives dueto tail ice were sometimes small, but measurable.These changes were expected to be small becausethe DH-6 STOL capabilities make it more robust toice. Other airplanes may show greater losses in

stability and control for the conditions tested.

9

Conclusions:

Based on the analysis and results presented, the following conclusions are made:

1. The test techniques and analysis methodsemployed permitted an accurate evaluation ofthe effects of moderate glaze tail ice on aircraftstability and control characteristics.

2. It was shown that ice on the horizontal andvertical tail surfaces significantly affect someaircraft stability and control parameters. Thefollowing parameters were affected:

• elevator and rudder control effectivenesswas reduced by approximately 10% becauseof ice on the horizontal and verticalstabilizers

• static longitudinal stability was reduced byapproximately 10% because of ice onhorizontal stabilizer.

• directional stability was reduced byapproximately 20% because of ice on thevertical tail for the zero thrust case.

3. It was shown that ice on the tail surfaces didnot significantly affect some stability andcontrol parameters. The following parameterswere unaffected:

• yaw damping was not affected by the iceshapes on the tail surfaces.

• Pitch damping was reduced because of tailice for the Cr = (0.14, 0.07) cases, but not atthe Cr =0 case. Further investigation needsto be made to understand this result.

• directional stability was shown to beunaffected by ice for the powered cases,but further investigation needs to be madebefore this is conclusive.

10

References:

1. Ranaudo, R.J., Batterson, J.G., et.al ., Determination of Longitudinal Aerodynamic Derivatives Using FlightData From an Icing Research Aircraft. AIAA Paper 89-0754, NASA TM 101427, 1989.

2. Norment, H.G., Three-Dimensional Trajectory Analyses of Two Drop Sizing Lutrunents: PMS OAP and PMSFSSP. NASA CR 4113, 1988.

3. Jordan, J.L., Platz, S.J., Schinstock, W.C., Flight Test Report of the NASA Icing Research Airplane. KSR86-01, Kohlman Systems Research, Inc., Lawrence, KS, NASA CR-179515, 1986.

4. Ranaudo, R.J., et.al ., The Measurement of Aircraft Performance and Stability and Control After Flight ThroughNatural Icing Conditions. AIAA Paper 86-9758, NASA TM-87265, 1986.

5. Bowden, D.T., Gensemer, A.E., Skeen, C.A., Engineering Summary of Airframe Icing Technical Data.General Dynamics /Convair, San Diego, CA, Technical Report ADS-4, 1964.

6. Klein, V., Morgan, D.R., Estimation of Bias Errors in Measured Airplane Responses Using MavinuunLikelihood Method. NASA TM-89059, 1987.

7. Batterson, J.G., STEP and STEPSPL - Computer Programs for Aerodynamic Model Structure Detenninationand Parameter Estimation. NASA TM-86410, 1986.

8. Klein, V., Batterson, J.G., Murphy, P.C., Detennination of Airplane Model Structure From Flight Data byUsing Modified Stepwise Regression. NASA TP-1916, 1981.

9. Klein, V., Estimation of Aircraft Aerodynamic Parameters Front Flight Data. Prog. Aerospace Sci. Vol 26,pp.1-77, 1989.

Table 1: Physical Characteristics of Research Aircraft

CHARAYrE1xJSYYC LOW ........... HYGN

Mass, kg 4510 4970

INERTIA:

&, kg-nF 26190 26660

1,, kg-m2 33460 34650

LL , kg-m2 47920 51650

k,, kg-n^ 1490 1560

WING:

Area, mZ 39.02

Aspect ratio 10.06

Span, m 19.81

Mean geometric chord, m 1.98

Airfoil section (17% thickness) "Dellavilland High Lift"

HORIZONTAL TAIL:

Area, mZ 9.10

Aspect ratio 4.35

Span, m 6.30

Mean geometric chord, m 1.45

Airfoil section (inverted) NACA 63A213

Table 11: Instrument Specifications

PARAMETERISFNSOR ::» RANGE,. :I:RMb M.0N

A, & A,, Sundstrand QA-700 1g .0002g

A\ , Sundstrand QA-700 +3g, -1g .00098

p, Humphrey RG02-2324-1 ±60°/s 0.0167°/s

q, Humphrey RG02-2324-1 ±60°/s 0.0167°/s

r, Humphrey RG02-2324-1 ±120°/s 0.0138°/s

9, Humphrey VG24-0636-1 ±60° 0.0293°

0, Humphrey VG24-0636-1 ±901 0.0439°

a, Rosemount 858 +15°, -10° 0.003°

8, Rosemount 858 ±15° 0.003°

V, Rosemount 542K 0 to 190 knot 0.076 knot

Alt., Rosemount 542K 0 to 15K ft 8.2 ft

OAT, Rosemount 102AUIP -201 to 30° F 0.041°F

S. . L & 6.. R I SAC series 160 +19°, -16° 0.0091°

8^, SAC series 160 +14°, -26° 0.0128°

d,, SAC series 160 ±16° 0.0080°

12

1.98

(6.5)

--F5.66

18.58)

6.30

(20.67)

^ lO v

DIAM 2.59(8.5)

—.^ 3.81 ^.

(12.5)

19.81(65.0)

4.5

(14.75)

15.77(51.75)

Figure 1: NASA Lewis Research Center Icing Research Aircraft: all dimensions are in meters (ft).

13

Figure 2: Artificial moderate glaze ice attached to horizontal and vertical stabilizers of the icing researchaircraft

14

U.3

0.2A x (g units)

0.1

00O 5 10 15

—0.4

—0.6

A Z (g units) —0.8—1.0

—1.2

—1.4 -- --0 5 10 15

20

10

q (deg/s) o --10

-zoO 5 10 15

90 -

89

V (i<ta S)88

87

86

0 5 10 15

14 --

12

10

—\/V

a (deg) $

6 -

4O 5 10 15

12 -10

8 -

0 (deg) 6 -42O

O 5 10 15

5 -

0

S (deg)-s

e -10

-15

-20--O 5 10 15

Time (s)

Figure 3: Longitudinal Parameter Identification Maneuver

15

0.15

A y (g units) 0.00-0.15

-0 30O 5 10 15

20

10

P ( deg / s ) 0-10

-20O 5 10 15

20

10

r (deg/s) 0

-10

-20O 5 10 15

a

4

(3 (deg) 0-4

-8O 5 10 15

15

10

5

^P (deg) 0

-5

-10O 5 10 15

5.0

2.5

d a (deg) 0.0

-2.5

-5.0O 5 10 15

5.0

2.5

d (deg) 0.0r

-5.0O 5 10 15

Time (s)

Figure 4: Lateral Parameter Identification Maneuver

16

-0.5 • UNICEDCT=0.14 ICED

—1.0

CmCK

—1.5

-2.0 ' 1 '

-2 0 2

4 6 8 10 12

-0.5CT=0.07

-1.0

C ma

—1.5

—2 0

N11,

—2 0 2 4 6 8 10 12

—0.5

—1.0

Cm-1.5

CT=0.00

______________

-------

V.

-2 0-2 0 2 4 6 8 10 12

Angle of Attack (deg)

Figure 5: Static longitudinal stability derivative with 95% confidence limits. S F =0°

17

CT 0.07

-0.6 • UNICEDCT=0.14 ICED

-0.8

-1.0

C ma —1.2

—1.4

—1.6

—18—4 —2 0 2 4 6 8 10 12

—0.6

—0.8

—1.0

Cm —1.2

a —1.4

—1.6

—1 8—4 —2 0 2 4 6 8 10 12

—0.6

—0.8

—1.0

C —1.2m

—1.4

—1.6

—1 8

C T=0.

—4 —2 0 2 4 6 8 10 12

Angle of Attack (deg)

Figure 6: Static longitudinal stability derivative with 95% confidence bounds. S F =10°

18

CT=O. 14

T,

-25

-30

C m -35q

-40

-45

• UNICEDICED

-2 0 2 4 6 8 10 12

-G0

-30

C -35m

q

-40

-45

CT=0.07

—2 0

—25

—30

C —35M q

—40

—45

2 4 6 8 10 12

CT=0.00

-------- ---

—2 0 2 4 6 8 10 12

Angle of Attack (deg)

Figure 7: Pitch damping derivative with 95% confidence bounds. S F =0°

19

CTI 0.07

---------- \^.

—25

—30

Cm —35q

—40

—45

-25 • UNICEDCT=O. 14

ICED

—30

Cm—35

q

—40

—45—4 —2 0 2

—4 —2 0 2 4 6 8 10 12

—25

—30

—35C M

q

—40

—45—4 —2 0 2 4 6 8 10 12

Angle of Attack (deg)

Figure 8: Pitch damping derivative with 95% confidence bounds. S F =10°

20

• UNICEMCT=o.1-t ICV1)

—1.5

C77Ibe

—^ 0 2 4 (7 8 10 12

—1.0

CT 0.O7

CI716e

—2.0

— 2 .5 1 1

—2 0

2 4 6 8 10 11

—1.0

CIn 6e

—2.0

^ I I I I I

CT=0.00

— ---^_____

•--- _______ -^___

—2.5 1 1 1 1 1 1 1 1

—2 0 2 4 6 8 10 12

Angle of Attack (deg)

Figure 9: Elevator effectiveness derivative with 95% confidence bounds. S F. =0°

21

—1.5

Cm de

—2.0------ A_-^-

-1.0 i • UNICEDCT 0. 14 ICED

.-

—25'—4 —2 0 2 4 6 8 10 12

—1.0

CT 0.07

—1.5

Cmde - ------- — — -

—2.0 -------.

—2.5—4 —2 0 2 4 6 8 10 12

—1.0CT=0.00

--------------------_- — -- -------------- : —-----------rte_--_ ^^-----

--------------------

—1.5

Cm 6

—2.0

— 2.5 ' I I I I I I I I

—4 —2 0 2 4 6 8 10 12

Angle of Attack (deg)

Figure 10: Elevator effectiveness derivative with 95% confidence bounds. 6F=100

22

0.15 • UNICED

CT=0.14

ICED

CnR0.10

0.05 ' ' '—2 0 2

4 6 8 10 12

0.15

CT=0.07

C 0.10n

0.05 ' 1'—2 O 2

4 6 8 10 12

0.15 I

CT=0.00

C 0.10

w-"----------------- ---f-----------

-----

- -----°^---- -----------------------------------------------------^ i- — — ---

0 . 05 1 1 1 1 1 1 1 1

—2 0 2 4 6 8 10 12

Angle of Attack (deg)

Figure 11: Directional stability derivative with 95% confidence bounds. SF =0°

23

CT=0.07

o.00 • UNICED

CT=0.14 ICED

—0.03

C —0.06P

—0.09

—0 12—2 0

2 4 6 8 10 12

U.UU

—0.03

C —0.06P

—0.09

—0 12

—2 0

2 4 6 8 10 12

0.00

CT=0.00

—0.03

C n —0.06P

—0.09

—0 12 I I I

—2 0 2 4 6 8 10 12

Angle of Attack (deg)

Figure 12: Directional cross-derivative with 95 % confidence bounds. 6F= 00

24

-0.10 • UNICED

CT=0.14 ICED

-0.15

Cnr

-0.20 --- _

-0 25

i _L - I i

-2 0 2

4 6 8 10 12

-0.10

CT=0.07

-0.15

C.nr

-0.20

-0.25 ' '-2 0

2 4 6 8 10 12

-0.10 1 T

CT=0.00

-0.15

C n

r -0.20

----_ ----^ = T̂_ --------

-0.25 1 1 1 1 1 1 1 1

—2 0 2 4 6 8 10 12

Angle of Attack (deg)

Figure 13: Directional damping derivative with 95% confidence bounds. S F =00

25

-0.10

—0.11

Cndr —0.12

—0.13

—0 14

-T --J^.i

CT=O. 14

• UNICEDICED

—2 0

2 4 6 8 10 12

—0.10

—0.11

C —0.12dr

—0.13

—0 14

CT=0.07

T%--- ------ -----I'^--- ---- __ T

—2 0

—0.1 0

—0.11

Cndr —0.12

—0.13

—0 14

2 4 6 8 10 12

CT=0.00T

------ —---- -- _-------------

—2 0 2 4 6 8 10 12

Angle of Attack (deg)

Figure 14: Rudder effectiveness derivative with 95% confidence bounds. 6,=O'

26

Form ApprovedREPORT DOCUMENTATION PAGE OMB No. 0704-0188

Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources,gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of thiscollection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for information Operations and Reports, 1215 JeffersonDavis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188), Washington, DC 20503.

1. AGENCY USE ONLY (Leave blank) 2. REPORT DATE 3. REPORT TYPE AND DATES COVEREDJanuary 1993 Technical Memorandum

4. TITLE AND SUBTITLE 5. FUNDING NUMBERS

Icing Effects on Aircraft Stability and Control Determined From Flight DataPreliminary Results

WU-505-68-106. AUTHOR(S)

T.P. Ratvasky and R.J. Ranaudo

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATIONREPORT NUMBER

National Aeronautics and Space AdministrationLewis Research Center E-7500Cleveland, Ohio 44135-3191

9. SPONSORING/MONITORING AGENCY NAMES(S) AND ADDRESS(ES) 10. SPONSORING/MONITORINGAGENCY REPORT NUMBER

National Aeronautics and Space AdministrationWashington, D.C. 20546-0001 NASA TM-105977

AIAA-9341398

11. SUPPLEMENTARY NOTESPrepared for the 31st Aerospace Sciences Meeting and Exhibit sponsored by the American Institute of Aeronautics andAstronautics, Reno, Nevada, January 11-14, 1993. T.P. Ratvasky and R.J. Ranaudo, Lewis Research Center. Responsibleperson, T.P. Ratvasky, (216) 433-3905.

12a. DISTRIBUTION/AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE

Unclassified - UnlimitedSubject Categories 08

13. ABSTRACT (Maximum 200 words)

The effects of airframe icing on the stability and control characteristics of the NASA DH-6 Twin Otter icing researchaircraft were investigated by flight test. The flight program was developed to obtain the stability and control param-eters of the DH-6 in a baseline ("uniced") configuration and an "artificially iced" configuration for specified thrustconditions. Stability and control parameter identification maneuvers were performed over a wide range of angles ofattack for wing flaps retracted (0°) and wing flaps partially deflected (10°). Engine power was adjusted to hold thrustconstant at one of three thrust coefficients (C T=0.14, CT=0.07, CT=0.00). This paper presents only the pitching- andyawing-moment results from the flight test program. Stability and control parameters were estimated for the unicedand artificially iced configurations using a modified stepwise regression algorithm. Comparisons of the uniced andiced stability and control parameters are presented for the majority of the flight envelope. The artificial ice reduced theelevator and rudder control effectiveness by 12% and 8% respectively for the 0 0 flap setting. The longitudinal staticstability was also decreased substantially (approximately 10%) because of the tail ice. Further discussion is providedto explain some of the effects of ice on the stability and control parameters.

14. SUBJECT TERMS 15. NUMBER OF PAGES

Aircraft icing; Stability and control; Flight data; Power effects 2816. PRICE CODE

A0317. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACT

OF REPORT OF THIS PAGE OF ABSTRACTUnclassified Unclassified Unclassified

NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89)

Prescribed by ANSI Std. Z39-18298-102

FOURTH CLASS MAIL

ADDRESS CORRECTION REQUESTED 7AU. S.MAIL

Postage and Fees PaidNational Aeronautics anSpace Admin straZnNASA 451

National Aeronautics andSpace Administration

Lewis Research CenterCleveland, Ohio 44135

Official BusinessPenalty for Private Use $300

NASA