1 AIAA 2000-3978 A Study of Airline Passenger Susceptibility to Atmospheric Turbulence Hazard Eric C. Stewart NASA Langley Research Center Hampton, VA Atmospheric Flight Mechanics Conference 14-17 August 2000 Denver, CO For permission to copy or to republish, contact the American Institute of Aeronautics and Astronautics, 1801 Alexander Bell Drive, Suite 500, Reston, VA, 20191-4344. AIAA 2000-3978 A STUDY OF AIRLINE PASSENGER SUSCEPTIBILITY TO ATMOSPHERIC TURBULENCE HAZARDS Eric C. Stewart* NASA Langley Research Center Hampton, VA 23681-2199 Abstract dfall equivalent fall height in a I-g gravity field (see eqn. 28), ft A simple, generic, simulation math model of a f frequency of flexible fuselage mode, Hz commercial airliner has been developed to study the susceptibility of unrestrained passengers to large, force due to flexible fuselage spring, lb discrete gust encounters. The math model simulates component of force along X body axis (i=w, t, the longitudinal motion to vertical gusts and includes T), lb (1) motion of an unrestrained passenger in the rear cabin, (2) fuselage flexibility, (3) the lag in the component of force along Z body axis (i=w, t, downwash from the wing to the tail, and (4)unsteady ta, Y), lb lift effects. Airplane and passenger response contours 2 are calculated for a matrix of gust amplitudes and gust acceleration of gravity, 32.2 ft/sec lengths of a simulated mountain rotor. A comparison position of unrestrained mass above aft cabin of the model -predicted responses to data from three floor, ft accidents indicates that the accelerations in actual collision velocity with the floor or ceiling, fps accidents are sometimes much larger than the simulated gust encounters. moment of inertia about y(pitch) axis, slug- IY Nomenclature ft2 alt altitude, ft k flexible fuselage spring constant, lb/ft az acceleration of cg vertically downward, g's k, efficiency factor of the tail, non-dimensional acceleration at aft passenger cabin, g's kq pitch rate gain in autopilot, CD,i total drag coefficient of ith surface, non- radians/(radians/sec) dimensional k0 pitch angle gain in autopilot, radians/radian CL,i lift coefficient of ith surface, non-dimensional kh altitude hold feedback gain, radians/ft lift curve slope of airplane, per radian kge elevator effectiveness, non-dimensional lift curve slope of tail, per radian gust length (see equation 29), ft C Lot ,I zero lift drag coefficient of the tail, non- m mass of wing-body, slugs C i)o,t dimensional me equivalent mass of aeroelastic structure, slugs mean aerodynamic chord of wing, ft Anc_ delta normal acceleration of center of gravity, C w *Aerospace Engineer, Senior Member AIAA g's An_jl delta normal acceleration of aft passenger Copyright © 2000 by the American Institute of cabin, g's Aeronautics and Astronautics, Inc. No copyright is asserted in the United States under Title 17,U.S. Code. q pitch rate, radians/second The U.S. Government has a royalty-free license to dynamic pressure, psf exercise all rights under the copyright claimed herein r radial distance from center of mountain rotor for Government Purposes. All other rights are reserved (see equation 29), ft by the copyright owner. R ratio ofx position of aft passenger cabin to tail length, non-dimensional 1 American Institute of Aeronautics and Astronautics b body (axis system) S W wing area, 1994 ft2 i ith aerodynamic component t time, seconds ic initial value At lag in wing downwash reaching the tail, aft in the aft passenger cabin seconds plunge motion time constant Operators TO A change first derivative with respect to time ( m//pVS,rCL, a ), seconds second derivative with respect to time component of inertial velocity along X body f,o non-linear lift function U o axis, fps f,0 non-linear drag function V true airspeed, fps Knax maximum tangential velocity of wind rotor Introduction (see equation 28), fps tan tangential velocity of wind rotor (see equation The NASA has initiated an aviation safety 28), fps program aimed at reducing the accident rate of W weight of airplane, lb transport airplanes. One element of this program is concerned with the fatalities and injuries caused by Wo component of inertial velocity along Z body encounters with turbulence. Although the literature is axis, fps full of studies on airplane gust loads, these studies have Xi x location of point of application of the ith primarily been concerned with the structural integrity of the airplanes and certification issues. However, in force component, ft most accident scenarios the airplane structure is not vertical displacement of horizontal tail due to Ylrim damaged, but flight attendants and unrestrained aeroelastic effects at trim flight condition, ft passenger are injured or killed 1. Often the airplane Ay change in length of spring from trim length, ft unexpectedly encounters a discrete gust of only a few z location of point of application of ith force seconds duration while flying in otherwise relatively 2i calm air. A study that specifically addresses the component, ft interaction of the gust/airplane/autopilot/pilot is needed O_ angle of attack, radians or degrees help identify critical factors in a typical accident. For Dc example, the parameter(s) that best describe the gust downwash parameter, non-dimensional 0_ characteristic(s) that induce accidents need to be identified. Once these parameters are identified, 8e elevator deflection, radians sensors such a radar and lidair can be designed to component of elevator deflection due to detect and avoid dangerous gusts. In addition, the (_ec simulated autopilot, radians responses of the airplane, autopilot, and pilot that contribute to the problem need to be identified. Once O pitch angle, radians or degrees these responses are identified, improvements to the P density of air, slug/ft 3 autopilot and piloting procedures may be possible. A three step procedure was used to study these A7 change in flight path angle, radians factors. First, a simulation math model was developed g inclination of thrust axis with respect to x to explore the response of unrestrained passengers to body axis (positive up), radians gusts of various shapes and sizes. This model was natural frequency of aeroelastic mode, generic with a minimum of airplane characteristics. (/')71 radians/sec However, the model included representations of such things as non-linear aerodynamics, an autopilot, Subscripts: fuselage bending, the lag inthe downwash from the w wing wing to the tail, unsteady lift effects on the wing, and t (horizontal) tail inelastic collisions of unrestrained objects with the ta (horizontal) tail aerodynamic interior of the rear cabin. T thrust The second step was to analyze the flight data g gust recorder (FDR) measurements from three recent e earth (axis system) turbulence accidents involving domestic commercial 2 American Institute of Aeronautics and Astronautics airlinersA.pproximagteustprofilewsereproduced fromestimateadirplanaeerodynamaincdmass with the non-linear, table-lookup aerodynamic characteristpiclussthemeasurepdarameteornsthe characteristics of the wing: flightdatarecordesrsuchasverticaalcceleratiaonnd pitchattitude. CL,w _- L. (O_w) and Crow = L, (a _) (6) and the estimated aerodynamic characteristics of the The third and final step was to compare the tail accident gust profiles and airplane responses to those predicted by the math model. Ct.., = Cl_,,a ' and CI,, = C1_,,' + k,C[, (7) This paper presents a short description of the The force components for the thrust are simply: simulation math model and various gust wave shapes Fx.r = Thrust *cos(e) (8) that were investigated. The simulation response calculations were then summarized into constant F::r = -Thrust *sin(e) (9) response contours as a function of gust length and The aerodynamic Z forces for the wing and the elastic amplitude. Next, the equations used to analyze the tail are: FDR data are presented along with the extracted gust velocities. Finally, the simulation calculations and the F._ = _S..[-C_._ cos(a_ )- C_)j sin(a_)] (10) FDR data are compared, and some observations are where i=w,ta made. The Z force from the tail (Fzj ) that is transmitted to the wing-body through the simulated Simulation Studies spring is not the same as the aerodynamic tail force Math Model: (f:.l.) given inequation (10). It is, instead equal and The equations of motion are written in a body axis system that is free to translate and pitch ina opposite Fsin equation (4) above: vertical plane. The airplane is modeled as a wing- fz. I = -F, = -k(y,ro, , + Ay) (1I) body mass connected by a spring to the effective The spring constant, k, in equation (10) and horizontal tail mass. The airplane mass is acted on by the wing body aerodynamic forces, gravity forces, the the effective tail mass, m e , in equation (4) were thrust forces, the X aerodynamic force of the horizontal estimated as follows. First, a natural frequency of the tail, and a Z tail force produced by the connecting elastic tail system was assumed, i.e. f = 3 Hz. Then a spring. The effective tail mass is acted on by the Z structural stiffness was calculated by assuming that the aerodynamic tail force, the connecting spring force, tail would deflect a given ratio, R=.02, of the tail length and gravity. A diagram of model ispresented in figure if loaded with the entire weight of the airplane. That is, I. The equations of motion for this system are given k = (mg)/(-x, *R) (12) below: Finally, the effective mass iscalculated from the spring Airplane wing-body: constant and the assumed natural frequency _" F_i - mg sin(0) = m(it, + qw,) (1) k i m e = - (13) _" F:a + mg cos(0) = m(fv,, - qu,, ) (2) (2rff) 2 i The angles of attack on the wing and tail are, (3) of course, critical to the proper calculations of the aerodynamic forces and moments. The angle of attack i J where i= w,t,T at the wing was calculated from the following equation: Flexible tail: a., = tan-' (w.,) (14) Ay = F,"."+---Fg:a,.,-.g+x,(1 (4) U w where the velocity components for the wing are a m e 1Tle straightforward sum of the inertial components and the The X aerodynamic forces for the wing and tail in gust components: equations (1) and (3) are derived from the lift and drag u,. = u,, + u_ (15) (5) w,. = w, + w_, (I6) Fr., = _S,, [-C _),icos(ai )+ C_._sin(aj )] where i=w,t 3 American Institute of Aeronautics and Astronautics Theprimedparametearrsetheunsteadliyft The commanded elevator was passed through a 14 radians/sec first-order actuator model and added to the effectrsepresentbeydtheapproximafotermulas2: trimmed elevator position. t t _i, = w,(I-.36e _,,) and w'_= w_(I-e _) (17) 6,. = 5.,. + (_,,,,ro, (23) with In order to simulate the collisions of an unrestrained passenger with the ceiling and floor, an 3cw 3cw = -- and rg (18) auxiliary calculation was made. The first step was to r,, 2V =--4--V- calculate the normal acceleration at an estimated The velocity components for tail are more complicated extreme aft cabin position: even though unsteady lift effects are ignored. For example, the gust hits the tail slightly after ithits the a4, =-a: +0.8x,0- (0.8)2A j) (24) wing. Also, the velocity at the tail ismodified by the were the 0.8 factor was the estimated ratio of the aft downwash from the wing. Of course, the pitching cabin location to the tail location, xt .The factor is motion of the airplane induces a vertical velocity at the squared because the deflection curve of the rear tail that isprimarily responsible for the pitch fuselage was assumed to be quadratic. Whenever the damping 3. Adding the flexible tail degree of freedom unrestrained body (passenger) was not in contact with adds more terms to these basic effects. The vertical either the floor or the ceiling (0 < h < 4 ),its motion of elastic tail induces a vertical velocity acceleration relative to the (accelerating) cabin floor or component at the tail that is responsible for damping of ceiling was this structural mode. In addition, the deflection of the flexible tail surface changes the incidence angle of the = -a,r, (25) tail and reduces the effective angle of attack of the tail. This acceleration was then integrated to produce the All these effects are modeled inthe following relative velocity and position. The inelastic collisions equations for the velocity components and angle of with the floor and ceiling were simulated by setting attack at the tail: (1= 0 whenever ut =Uo+Ug(t-At)+z t*q (19) h-<0 and((l<0 or [1<0) (26) 0o_ w,= w,, w,,(t- At)... or h > 4 and ( la> 0 or _i>0) (27) This simulated a purely inelastic collision (no rebound) of the passenger with the interior of the cabin. c3c (t At) x,q + A_ (20) +O-}-d)w - _ The 4 (ft) constant in equation (27) was a compromise between the distance between apassenger's head while where the (t - At) notation indicates the parameter is standing inthe aisle and the ceiling and the much evaluated at an earlier time [with longer distance from the ceiling to the floor. The At = (x w - xt) /V ]to account for the time collision velocity with the floor or ceiling, 1:1c ,was difference ineffects at the tail and the wing. All the transformed into an equivalent height of a fall in a other parameters are evaluated at the current time t. normal l-g gravity field using The angle of attack at the tail was calculated from the above velocity components plus two additional terms: d_,, - (28) 2g a, = atan -l(w,) + k,_6e -_2(AT +Y,ri,,,) (21 ) where the 4a,i ,term is added to account for the fact Itt -- Xt The second term on the right hand side of equation (21) that the floor (or ceiling) is either accelerating away is the effect of the elevator (used from trimming the from or toward the passenger when he impacts the simulated airplane and as the autopilot effector), while floor (or ceiling). The constant of 4 is based on the the third term is the effect of the flexible tail degree of assumption that the body of the passenger deforms 3 freedom assuming a quadratic structural deflection inches upon impact. curve. Gust inputs: The autopilot control law contained feedback Several different gust shapes were terms for the pitch rate, pitch attitude, and altitude: investigated including "l-cosine," simple ramps, 6, =k_q+ko(O-O,r,,,)+kh(alt-alt,r,,) (22) sawtooth, sine waves, and a mountain rotor, figure 2. However, only results from the mountain rotor are presented herein. These results are fairly typical of the other wave shapes if they are compared on the basis of 4 American Institute of Aeronautics and Astronautics aconsistesnettofgusltengthasndamplitudeTs.he airspeed of approximately 800 fps. The structural mountairnotorwasrepresentbeydasimplealgebraic deflection of the horizontal tail, Ay, was relatively relationshtihpatwasafunctionoftwoparametegrsu:st small (less than 0.2 feet) indicating that the simulated amplitudeVm,ax,and gust length, l_. The general airplane was relatively stiff. Since the change in the acceleration inthe aft cabin dropped below -lg, the equation was simulated passenger came off the floor and impacted the 4-foot ceiling at approximately 2 seconds into the 6 r time history, see the time history of h. The relative velocity at impact was approximately 20 fps, see ]_. However, after the acceleration reversed, the simulated passenger fell back down to the floor and impacted with an even greater velocity than before. where r is the radial distance of the airplane from the A matrix of gust lengths and gust amplitudes were run through the simulation program to produce center of rotation of the rotor. The tangential velocity, time histories. A second program then extracted the gtan, was transformed into a vertical and horizontal maximum values of the accelerations and impact velocity before being applied to the airplane math velocities. These data were then transformed into model. However, since the airplane was initially set to contour plots such as those shown in figure 5 for the go through the center of the rotor and because the normal acceleration at the c.g. As can be seen in the simulated airplane did not translate vertically any figure, the maximum accelerations are a function of significant distance, the gust velocity was almost both the gust length and the gust amplitude. For entirely vertical as will be seen later. A sketch of the example, at a gust amplitude of-40 fps, the vertical velocity is presented in figure 3 for the acceleration varies from 1.5g's at a gust length of assumed condition of no vertical translation of the about 200-300 ft to only 0.5 g's at agust length of airplane. about 2200 ft. However, at very short gust lengths the Simulated Responses: accelerations drop off again because of the unsteady The above equations were translated into a lift effects used in the simulation. Matlab V4.2 Simulink block diagram for solution. The The contour plots for the normal acceleration model was trimmed to a cruise flight condition (Mach at the aft cabin location, figure 6, are practically =0.8 at 30,000 ft) by an iterative routine that identical to those at the c.g. at gust lengths greater than determined the trim thrust, angle of attack, elevator 1000 ft. However, at the shorter wavelengths, the aft position, and spring deflection to produce zero cabin accelerations are significantly higher. Part of this accelerations in level flight. The simulated airplane increase is due to the pitching motion of the simulated characteristics are presented intable I. airplane, but a significant part isdue to the excitation A time history of a typical gust penetration is of the structural mode of the tail. The combined effect shown in figure 4. All the gust responses shown are of the pitching and structural mode is enough to offset for gusts that have a negative peak followed by a most of the reduction due to the unsteady lift effects. positive peak. As can seen in the figure, the gust input, The contour plots of the equivalent fall heights are presented in figure 7. The maximum Aw_, has peaks of almost exactly +/-100 fps contours occur for gust lengths less than 500 ft. indicating that there was very little vertical translation However, they drop off sharply for the smallest gust away from the center of the rotor by the fast moving lengths despite the fact that the aft cabin accelerations airplane. The maximum change in angle of attack was were very high at the same gust lengths, figure 6. The about 7degrees which did not stall the wing since its reason for this, of course, is that for the shorter gust initial trim angle of attach was less than 1degree. The lengths the accelerations reversed so quickly that the normal acceleration at the c.g, An,g, and the resulting velocities and displacements were relatively small. In fact, for the shorter gusts the simulated acceleration at the aft passenger location, Anqfi, are passengers did not impact the ceiling at all, but only practically equal because the illustrated gust length of came offthe floor a small distance before impacting 500 ft is so long that the pitching motion, A0, is with floor again. relatively slow. This was not true for shorter gust lengths. The change in true airspeed, A V, was much less than the maximum gust velocity of 100 fps because the latter was nearly perpendicular to the true 5 American Institute of Aeronautics and Astronautics Data from Accidents The recovered gust velocities for three different accidents are presented in figures 8to 10. Analysis equations: Each accident is different, but there are some common The first step inthe calculations was to features. The gust amplitudes are very larger--up to calculate the change (delta) in the vertical acceleration 100 fps, and they attain their maximum values in a few and pitch attitude from the initial, undisturbed flight seconds. Another interesting characteristic is that the largest magnitude accelerations are inthe negative values, ni,,:and Oic: direction and they occur when the initial upward gusts reverse direction. The fact that all 3 accidents had their Ancg =n-nio and A0=0-0i_ (30) largest magnitude accelerations inthe negative Next, the dynamic pressure was calculated from the direction may be because the NTSB data collection true airspeed: process would naturally eliminate positive 1 (31) accelerations. That is, large positive accelerations will not generally throw passengers around the cabin so that these incidents do not result in accidents and would be where /9 and V were derived from the pressure altitude omitted from the available data base. However, the and indicated airspeed assuming a standard atmosphere correlation between gust reversal and the peak and zero position error corrections. The third step was accelerations may indicate that there may be a control to calculate the change in angle of attack from the change in vertical acceleration ignoring the effect of input phasing problem between the airplane and the the elevator and other second-order parameters on the gust that contributes to the accidents. This speculation acceleration: is supported by accident #2, figure 9. From about I0 seconds to 15seconds there is a strong oscillatory acceleration response that is not correlated with the Aa = _S_,,C_,,_ gust velocity. This oscillatory response was produced by control inputs to the elevator. However, it is The weight of the airplane was estimated from the impossible to tell from the recordings whether the accident records of the fuel and passenger load. The control inputs were produced by the autopilot or the wing area, S,,, was determined from publicly pilot. There is a strong tendency to overcontrol the . 6,7,8 available information on the aircraft type. The lift pitch attitude insevere gust encounters . The curve slope, C1,,<_,was estimated using aerodynamic autopilot usually has a relatively small effect 9. In any case, the present sample of only 3accidents istoo data charts 4 and publicly available information on the small to generalize to the larger world• wing and tail for each accident airplane. The next step was to calculate the change in Comparison of Simulation to Accidents flight path angle from the vertical acceleration and true airspeed: Although the regular gust shape used in the simulation was not the same as the irregular gust shape g IAn_,gdt (33) encounter in the real accidents, a comparison was Ay = attempted. That is, the time intervals in the accident It should be mentioned that the vertical acceleration, data where the gust gradients and amplitudes were the An<.g, in equations (32) and (33) included a small greatest were analyzed. The gust amplitudes were non- correction determined separately to make the second dimensionalized by dividing by gTo,"or integral of An<_ more consistent with the changes in mg / pVS,.Ct.,, _. This factor compensates for measured pressure altitude. W The final step was to calculate the change in different air speeds V, wing loadings _-,,., altitudes the vertical gust velocity from the previously calculated parameters: (through /9 ), and aerodynamic characteristics CI_,,_. Aw_ = V(Aty - A0 + AT") (34) Likewise the gust length was normalized by dividing Equation (34) is the NASA aircraft turbulence by the true airspeed to account for differences in the • 5 measurement equation with the acceleration-derived effective gust length due to different airspeeds. The angle of attack substituted for the vane-measured angle results are presented in figure II for the cg of attack. acceleration. The accident data symbols are plotted Recovered gust velocities: according to the estimated maximum gust amplitudes 6 American Institute of Aeronautics and Astronautics andgustlengthsN.exttoeachsymboislanumerical accelerations during the actual accidents were valueinparenthesinisdicatintghemaximumrecorded sometimes twice as large as the model predictions. acceleratioBny.comparinthgemaximumrecorded This large difference may be due to the transients acceleratitoonthenearessitmulateadcceleration during the autopilot disconnect sequence or pilot contouar,measuoreftheagreemebnettweethnetwo control inputs. However, the accident data does not datacanbemadeI.tisapparetnhtattheagreemeinst contain enough parameters to draw any firm relativelpyoorforacciden#t1and#2.Infactt,he conclusions. acceleratiofnosrthosetwoaccidenatsreovertwice thatpredictebdythesimulationIn.othewr ordsit, appeatrhsattheaccidenatcceleratiownesremuch largetrhantheywouldhavebeeniftheairplanehsad respondetodthegivengustsinthesamemannearsthe References simulateadirplanesA.moredetaileadnalysoisfthe 1. Anon: In-Flight Safety of Passengers and Flight accidenitnsdicatetdhattheelevatocrontroml otions Attendants Aboard Air Carrier Aircraft; NTSB- andpitchingresponsweerelargetrhanthoseofthe AAS-73-1, March, 1973. simulationI.twasimpossibtloedetermintheereason 2. Jones, Robert T.: The Unsteady Lift of a Finite forthelargecrontroml otionasndpitchingresponse Wing. NACA TN 682, 1939. fromthelimitednumbeorfparametererscordeodnthe 3. Phillips, William H., and Kraft, Christopher C.: flightrecordersT.hatis,itcouldnotbedetermined Theoretical Study of Some Methods for Increasing whethethreautopilootrthepilotwasmakintghe the Smoothness of Flight through Rough Air. controilnputsO. nlythedataforacciden#t3(good NACA Technical Note 2416, July 1951. agreemewnitthsimulationin)dicatethdestatuosfthe 4. Etkin, Bernard: Dynamics of Flight-Stability and autopilosthowintghatitwasdisengagaefdterthe Control. John Wiley and Sons, 1959. (positivep)eakintheacceleration. 5. Houbolt, John C.; Steiner, Roy; and Pratt, Kermit G.: Dynamic Response of Airplanes to Concluding Remarks Atmospheric Turbulence Including Flight Data on A simple mathematical model for studying the Input Response. NASA TR R-199, 1964. longitudinal effects of gusts has been described. The 6. Anon. Civil Aeronautics Board: Aircraft Accident model includes the effects of fuselage flexibility and Report SA-372; Northwest Airlines, Inc. Boeing the motion of an unrestrained mass (simulated 720B, near Miami Florida, February 12, 1963; passenger) in the aft passenger cabin. The model has Report Released June 4, 1965. been used to calculate the response to different gust 7. Andrews, William H.; Butchart, Stanley P., Sisk, shapes and amplitudes and flight conditions. However, Thomas R.; and Hughes, Donald L.: Flight Tests only calculations for amountain rotor gust profile are Related to Jet-Transport Upset and Turbulent-Air presented because of the similarity of the results for Penetration, NASA SP-83, May, 1965. different wave shapes. Gust length was shown to be 8. Bray, Richard S., and Larsen, William E.: nearly as important as gust amplitude in predicting Simulation Investigation of the Problem of Flying airplane response. The simulated passenger inthe aft Swept-Wing Transport Aircraft in Heavy passenger cabin had the most severe collisions with the Turbulence; NASA SP-83, May, 1965. cabin interior for gust lengths of approximately 300- 9. Goldberg, Joseph H.: Gust Response of 400 ft. However, aeroelastic effects on simulated Commercial Jet Aircraft Including Effects of passenger collisions with the cabin interior were Autopilot Operations. NASA CR-165919, June minimal. 1982 The model predictions have been compared to data from three airline turbulent accidents. The 7 American Institute of Aeronautics and Astronautics Table1.SimulateAdirplanCeharacteristics Characteristic Value Weight, W 140,000 lb Wing area, Sw 1994 ft 2 .01 Zero lift drag coefficient of tail, Cj_,,.r 1.19/rad Tail lift curve slope, C Let,t .55 Oz Wing downwash on tail, -- &z Pitch moment of inertia, ly 525,000 slug-ft _ .1058 Efficiency factor of tail, kt Elevator effectiveness, kse 0.5 x location of wing aerodynamic center, xw -0.59 ft 0.0 ft z location of wing aerodynamic center, z w -76.6 ft xlocation of tail aerodynamic center, x t z location of tail aerodynamic center, z t 0.0 ft 0.0 ft x location of thrust, xT 0.0 ft z location of thrust, z T pitch gain in automatic control system, k0 2.0 rad/rad 0.3 rad/(rad/sec) pitch rate gain in automatic control system, kq .0001 rad/ft altitude gain in automatic control system, kh flexible tail spring constant, k 91,0001b/ft 257 slugs flexible tail effective mass, m e Thrust 16074 ib trim angle of attack, O:mm 0.67 degrees trim elevator angle, _ejri,,, -0.89 degrees trim flexible tail deflection, Ytrim 0.0118 ft flexible tail natural frequency, f 3 Hz flexible tail stiffness factor, R .02 8 American Institute of Aeronautics and Astronautics Earth Axis T_h._rust /r -..-. -. -.. Airplane AxXise _ 1 _,, -.. Xb ,oo.o / ----... -z,w Zb """-.._.._.:""--. ,i_ _ "-.. .v,BB "'X / / '\ k i / " . h \ Fx,t....._ AA . BB i----_ / Fz,t=-Fs "-_./ ,, .- -_. / / x Aft / x,t Passenger / __j-Tailac Cabin I ___"me AA \\ / x .... J Fz.ta / Flexible Tail Figure 1. Schematic of simulated airplane math model showing the main components: wing, horizontal tail, elastic fuselage, and aft passenger cabin with an unrestrained mass. step ramp peak _S Y_ block cos-ramp cos-ramp-cos sine 1-cosine rotor Figure 2. Sketches of some of the different gust wave shapes that were studied. 9 American Institute of Aeronautics and Astronautics