Document
AIAA-92-4109-CP
FLIGHT TESTING AND SIMULATION OF AN F-15 AIRPLANE
USING THROTTLES FOR FLIGHT CONTROL
Frank W. Burcham, Jr.+ Trindel Maine"'* Thomas Wolft NASA Dryden Flight Research Facility Edwards, California Abstract VC airspeed, kts a angle of attack, deg Flight tests and simulation studies using the throt- tles of an F-15 airplane for emergency flight control have been conducted at the NASA Dryden Flight Re- Introduction search Facility. The airplane and the simulation are ca- A multi-engine aircraft with a major flight-control pable of extended up-and-away flight, using only throt- system failure (such as loss of hydraulic pressure) may tles for flightpath control. Initial simulation results use throttle manipulation for emergency flightpath con- showed that runway landings using manual throttles- trol. Differential throttle control generates yaw, which only control were difficult, but possible with practice.
through dihedral effect, results in roll. Collective throt- Manual approaches flown in the airplane were much tle inputs may be used to control pitch. The DC-10, more difficult, indicating a significant discrepancy be- B-74 7, and L-1011 aircraft have had to use throttles tween flight and simulation. Analysis of flight data for emergency flight control.
and development of improved simulation models that resolve the discrepancy are discussed. An augmented To study the use of the propulsion system for throttles-only control system that controls bank angle emergency flight control, the NASA Dryden Flight and flightpath with appropriate feedback parameters Research Facility at Edwards, California, conducted has also been developed, evaluated in simulations, and flight, ground simulator, and analytical studies. The is planned for flight in the F-15.
study had three objectives. The first objective was to determine the degree of control power available for Nomenclature various classes of airplanes. Results from this objective have shown a surprising amount of control capability CG center of gravity for most multi-engine airplanes. The second objec- CAS control augmentation system tive was to investigate control modes that could be DEEC digital electronic engine control developed for future airplanes. An augmented control system that uses pilot flightpath inputs and feedback EMD engine model derivative control to provide throttle commands for emergency HUD heads-up display landings has been developed. This augmented system PCA propulsion controlled aircraft has been evaluated on a transport airplane simulation,2 and an F-15 simulation. A flight evaluation on an PLA power lever angle, deg F-15 is planned. The third objective was to provide PLF power for level flight, deg awareness of throttles-only control capability and sug- gested manual throttles-only control techniques for pi- lots. Reference 1 presents Dryden results of simulation •chief, Propulsion Branch. Associate Fellow, AIAA. and flight studies of several airplanes, including the •• Aerospace engineer.
B-720, Lear 24, F-15, B-727, C-402, and B-747.
t Simulation engineer.
Copyright @1992 by the American Institute of Aeronau- More recently, additional flight tests have been flown tics and Astronautics, Inc. No copyright is asserted in the to investigate the details of throttles-only control for United States under Title 17, U.S. Code. The U.S. Govern- the F-15 airplane, and to develop data to compare ment has a royalty-free license to exercise all rights under with the F-15 simulation. Significant discrepancies the copyright claimed herein for Governmental purposes. All other rights are reserved by the copyright owner. were found when the flight data were compared with
l
engines. These engines ( company designation F-15 simulation data. Additional flights and a series of improvements to the simulation have been made to PW-1128) include a redesigned fan and other improve- resolve the flight-to-simulation discrepancies. ments. The FlO0 EMD engines are controlled by a dig- ital electronic engine control (DEEC). Interim control This paper reviews the principles of throttles-only system software was incorporated in these EMD en- control, recent results of propulsion-only flight con- gines. This software produces slower, nonproduction trol for the F-15, comparisons of flight to simulation engine response characteristics at low power settings data, and simulation upgrades. Although the Fl00 en- that make it more representative of higher bypass tur- gines are equipped with afterburners, all tests discussed bofan engines.
in this paper were limited to nonafterburning power.
The inlets are mounted on the sides of the forward Plans for implementation of an augmented system for fuselage, and are external compression horizontal ramp flight on the NASA F-15 are also discussed.
inlets with variable geometry. A variable capture-area Description of F-15 Airplane and capability exists in which the inlet cowl rotates about Instrumentation a point near the lower cowl lip. At subsonic speeds, the inlet cowl angle is normally positioned by a control The F-15 airplane (Fig. 1) is a high-performance system as a function of angle of attack. The cowl may fighter airplane with a maximum Mach capability of be moved to the full-up inlets emergency position by 2.5. The F-15 (McDonnell Aircraft (McAir) Division the pilot.
of the McDonnell Douglas Corp., St. Louis, MO) has The NASA F-15 flight-control system has the stan- a high wing with 45° of leading-edge sweep and twin dard mechanical flight-control system and a digital con- vertical tails. It is powered by two Pratt & Whit- trol augmentation system (CAS). For throttles-only ney (West Palm Beach, FL) FlO0 afterburning tur- control research, the CAS can be turned off and the bofan engines mounted close to the centerline in the mechanical system can be operated in an emergency aft fuselage. The thrust-to-weight ratio is very high, mode. This eliminates any flight-control system mo- approaching 1 at low altitudes with maximum after- tion except that caused by pilot inputs.
burning power. The NASA F-15 is the number 8 pre- production F-15A, has no weapons systems installed, The F-15 is equipped with a heads-up display (HUD) and has additional extensive instrumentation. The which provides flight information such as airspeed and zero-fuel weight is 29,450 lb. Fuel capacity is 11,600 lb.
altitude. A velocity vector symbol is available for de- termining the precise flightpath relative to the ground.
The engines installed in the NASA F-15 are the The F-15 airplane was instrumented to measure the developmental FlO0 engine model derivative (EMD) ( ( Figure 1. Three-view drawing of the F-15 airplane.
r parameters required for the throttles-only flights. All The simulation may be run in a batch (non-real time) typical engine and airplane parameters were measured. mode or inay be flown from a simulated cockpit shown Data from individual sensors and from the digital con- in Fig. 2(a). The cockpit simulates the key instruments trol system data buses (each engine and the digital in the NASA F-15 airplane. An actual F-15 stick and flight-control system) were recorded on an onboard throttle quadrant are provided. The control panel on pulse code modulation system and also telemetered to the left allows the operator to select special modes as the ground. Data were presented in a ground control required.
room for real-time monitoring and analysis. An HUD The visual display provides a limited out-the-window camera was also provided and the signal was teleme- color view of the world with an optional IIUD overlay.
tered to the ground for real-time display. Data were The HUD information is similar to that available in the also recorded for post-flight analysis.
F-15 airplane, and includes the velocity vector symbol.
The lakebed, main runway, and Edwards area are mod- F-15 Simulation eled with adequate realism for the approach-landing Two F-15 simulations (Fig. 2) were used in this task of this study. Upgrades to the Dryden simulation study, one at NASA Dryden and the other at the McAir that have evolved over the course of this project will be discussed later in the Results and Discussion section.
Simulation Facility in St. Louis, MO. The NASA Dryden F-15 simulation is a fixed-base, full-envelope, Similar tests were conducted at the McAir simula- six-degree-of-freedom aircraft simulation. This model tion (Fig. 2(b)). This fixed-base simulation features an contains nonlinear aerodynamics, a nonlinear flight- actual F-15 cockpit and high-fidelity visual equipment control system, and originally, a first-order engine which projects scenery onto a 40-ft dome. The aerody- response model. It is written in FORTH.AN and namic, control system, and propulsion system models is modular in construction. The integration in- were similar to those at Dryden.
terval is 25 msec. Because it is an engineering simulation, only those elements necessary to sup- port the flight research programs are implemented.
EC 90-227-1 (a) Dryden F-15 simulation cockpit.
Figure 2. F-15 simulation cockpits.
i
.. ·····.·.· .
. ... ~=««---<-;~:;::.:::;:;::;·.
(b) McAir F-15 simulation cockpit.
Figure 2. Concluded.
Over a short period of time (approx 15 sec), added Principles of Throttles-Only Control thrust causes a speed increase, which increases The principles of throttles-only flight control 1 will be lift, causing a pitch rate increase, and a climb (if reviewed here, using examples for the F-15 airplane.
allowed to continue for a longer period of time, this effect will be oscillatory, sec Phugoid, page 5.
Roll: Differential thrust generates sideslip, which, The degree of change to the flightpath angle is through dihedral effect, results in roll. Roll is con- proportional to the difference between the initial trolled to establish a bank angle, which results in a turn trim airspeed and the current airspeed, hence, the and change in aircraft heading. Figure 3 shows a typ- change in flightpath angle tends to increase as ical roll response to differential throttle. Once the dif- speed increases.
ferential throttle is applied, the differential thrust be- gins to increase, inducing sideslip and roll. As sideslip 2. Pitching moment due to thrust line offset. If the increases, the airplane directional stability generates a engine thrust line does not pass through the center moment equal to the moment from differential thrust, of gravity (CG), there will be a pitching moment and equilibrium is reached (in this case for the F-15) introduced by thrust change. For many transport with approximately 12 deg/sec of roll rate.
aircraft, the thrust line is below the CG, and in- Pitch: Pitch control due to throttle changes is more creasing thrust results in a nose-up pitching mo- complex. There are several effects that may be present, ment, the magnitude being a linear function of depending on the aircraft characteristics. These effects the thrust change. This is the desirable geom- are shown in concept in Fig. 4(a).
etry for throttles-only control, because a thrust change immediately starts the nose in the same 1. Flightpath angle change due to speed stability. direction as will be needed for the long-term flight- Most airplanes exhibit positive speed stability. path angle change. The effect is more a function
r
of change in thrust than change in speed, and oc- which kinetic and potential energy (speed and al- titude) are traded. The degree of oscillation in curs near the time of the thrust increase, as seen in Fig. 4(a). High mounted engines result in a pitch speed and altitude is related to the speed stabil- down, which counters the effects of speed stability. ity. The phugoid oscillation is excited by a pitch, Pitching moment due to thrust will cause a change or velocity change, and will have a period of ap- in angle of attack, and hence, lift. For the F-15, proximately 1 min., and may or may not damp the thrust line passes within ± 1 in. of the verti- naturally. Figure 5 is an example of the phugoid cal CG, depending on fuel quantity, and this effect response from the F-15 simulation in its initial is small. configuration as excited by a 10°-step increase in PLA. The flightpath angle increase results in a 3. Flightpath angle change due to the vertical compo- steepening climb and speed peaks, and begins to nent of thrust. If the thrust line is inclined to the decrease after about 15 sec, oscillating about the flightpath, as is commonly the case, an increase in initial trim speed. In the oscillatory phugoid mo- thrust will cause a direct increase in vertical ve- tion, pitch rate is in phase with velocity, while locity, i.e., rate of climb, and a resulting increase flightpath angle (and rate of climb) lags by 90°, in flightpath angle. For a given aircraft configu- and altitude lags by 180°. Although a very small ration, this effect will increase as angle of attack amplitude phugoid is nearly a constant angle-of- increases (i.e., as speed decreases).
attack motion, for the size phugoid oscillations typically seen in throttles-only control, pitch rates Figure 4(b) is an actual time history of pitch rate for are significant, as shown. This results in a varia- the F-15 for a throttle increase to intermediate power. tion in angle of attack, in this case varying over It shows the overall result of the effects previously men- a 2- to 3°-range. Properly sized and timed throt- tioned, with a maximum pitch rate of 2 deg/sec. tle inputs can be used to damp unwanted phugoid oscillations.
4. Phugoid. The phugoid is the longitudinal long pe- riod oscillation of an airplane. It is a motion in Left 12 X 103
-
I _,.-
-----------
,,,, ✓ ✓ ~ 8
Throttle -
Net ✓✓-..
angle, thrust per ✓ ,,,, _,.
deg engine, ...
......
lb
......... __
Right
\
-
---------
I I I I I I ~ ,_..
Sideslip, Roll rate,
'
-1 deg deg/sec
'
_,. _,. .... --------
~', ---- .,,,.
.,, ' ...
'
-2 -1 0 2 3 4 5 Time, sec 910687 Figure 3. Roll control resulting from differential thrust.
Commanded Change Phugoid in thrust .r~sponse
..... __ _
Change In ..
..
..
..
airspeed Change In Thrust offset flightpath Vertical angle component due to: of thrust Overall 2 4 10 12 14 16 0 6 8 Time, sec 920290 (a) Schematic pitch effects of thrust increase.
Measured pitch rate, deg/sec
~ ~
,?
~I 4 10 12 14 16 0 2 6 8 Time, sec 920317 (b) Flight data, NASA F-15, VC = 170 kts.
Figure 4. Pitch effects of a step increase in thrust on both engines.
16000 12000 Altitude, ft Pitch rate, deg/sec
-~
Rate of climb, ft/min.
-500 a, deg Flightpath angle, deg -10 vc, kts 40 120 160 200 0 80 Time, sec 920291 Figure 5. Phugoid oscillation from the F-15 simulation, VC = 170 kts, 10°-increase in PLA at O sec.
Speed Control maximum negative pitch rate. Tests were repeated over a range of speeds, and in some cases, for a suitable Once the flight-control surfaces of an airplane are range of fuel quantities (with resulting CG positions).
locked at a given position, the trim airspeed of most airplanes is only slightly affected by engine thrust. Re-- Another test was the full-differential throttle test, trimming to a different speed may be achieved by other used to determine the maximum roll rate. The airplane techniques, such as variable stabilizer control, CG con- was gently rolled to 30°-bank, then full-differential trol, lowering of flaps, landing gear, etc. In general, throttle was applied, and the airplane rolled back through level and to at least 30° in the other direction.
the speed will need to be reduced to an acceptable landing speed; this implies developing nose-up pitch- This test was also conducted over a range of speeds.
ing moments. Methods for doing this include moving The small throttle movement test was also performed the CG aft, lowering the flaps, and extending the land- on the F-15 airplane. In this test, beginning at PLF, ing gear. For the F-15, moving the inlets to the full-up the throttles were advanced-retarded by 1 in., and the emergency position reduces the trim speed by 20 kts.
resulting pitch rates were measured. For roll rate tests, Thrust Response the throttles were split by 1 in. These results are more like the types of throttle movements that are commonly Thrust response of turbofan engines may be slow rel- used in engines-only flight control.
ative to piston or turbojet engines. The Fl00 EMD Typical pilot-in-the-loop maneuvers were also used engine controllers in the NASA F-15 have interim soft- to evaluate throttles-only control capability of the ware, and respond quickly at higher thrust levels, but F-15. With the flight-control surfaces fixed, the pilot at low thrust levels, respond more slowly. Idle to in- was asked to fly tests which included (1) achieve and termediate power throttle snaps take approximately 2.5 sec. Reductions to idle power exhibit a rapid re- maintain level flight, (2) turn to and hold a given head- sponse until low thrust is reached, but a very slow ing, (3) initiate and attempt to maintain a constant spooldown taking up to 10 sec occurs before idle thrust rate of descent, (4) use various techniques to damp a is reached. phugoid oscillation, and (5) make approaches to a run- way. In the simulator, the pilot was also asked to make Effects of Speed on Propulsive Control Power landings on a runway and make go-arounds from a low- For turbine-powered airplanes, engine thrust is not a approach situation.
strong function of airspeed, however, the stabilizing ef- Results and Discussion fects of vertical and horizontal stabilizers are a function of dynamic pressure, and are inversely proportional to This section discusses the development of the sim- the square of airspeed. The result of these characteris- ulation and flights of the NASA F-15 airplane for tics is that the relative propulsion system control power throttles-only control in chronological order. All data increases as airspeed decreases.
presented are with the landing gear down. Also in- cluded are the plans to implement the augmented Test Techniques throttles-only control system on the NASA F-15 Test techniques were developed to assess the airplane.
throttles-only control capability of the F-15 airplane The initial throttles-only control tests were con- and simulation. To avoid flight-control system inputs, ducted on the NASA Dryden F-15 simulation. It was the CAS was turned off, and the emergency mode was found that the F-15 had pitch capability at speeds be- selected for the mechanical system. In this mode, the low 300 kts, and roll capability at all speeds. The air- flight-control surfaces would not move as long as the plane was quite stable in the initial simulation configu- pilot did not move the stick or rudder pedals. One test ration. Flightpath control with throttles worked well; if used was the full-throttle (maximum nonafterburning) the HUD velocity vector was below the desired flight- range test. Although full throttles are rarely used dur- path, the pilot simply added thrust until it reached ing throttles-only flight, this test provides an assess- the desired position. If the flightpath was higher than ment of the maximum capability, and an easily repeat- desired, the pilot reduced the thrust until the desired able metric with which to make comparisons between flightpath was reached. With some practice, the F-15 flight and simulation.
simulation could be landed repeatedly on a runway.
From power for level flight (PLF) conditions, both Some initial throttle step tests were also conducted. At throttles were advanced to intermediate power (max- this point, initial flight tests were flown on the NASA F-15 airplane. Open-loop tests, including full-throttle imum nonafterburning) to determine the maximum pitch rate capability. The same test was then repeated steps, were flown and control capability appeared like by going from PLF to idle power to determine the the simulation.
Full-Throttle Steps more PLA increase is available than decrease, and that more nose-up than nose-down control is available. The Typical results from the full-throttle step tests on flight data for thrust decreases are less than predicted the F-15 airplane and simulation are shown in Figs. 6 by the simulation and will be discussed later.
and 7. The flight and simulation maximum-minimum pitch rates are shown in Fig. 6, and exhibit a response Full-differential thrust test results are shown in inversely proportional to the square of the speed. For Fig. 7. Again, the inverse square relationship to speed the F-15 and most other turbine-powered airplanes, is evident. The flight data show somewhat less roll rate PLF in the approach to landing phase is rather low, than the simulation results.
typically 25 to 35 percent of thrust. This means that Flight data Simulation ◊ PLF to Intermediate With separate gross 0 PLF to Idle thrust and ram drag With gross thrust and ram drag assumed coincident Pitch rate, deg/sec PLA from PLF to Intermediate
i
I PLA from
' PLF to idle -1 -2 100 150 200 250 300 VC, kts 920318 Figure 6. Effect of calibrated airspeed on pitch rate for the F-15 flight and simulation.
0 Flight
Simulation With separate gross thrust and ram drag With gross thrust and ram drag assumed Maximum coincident roll rate, deg/sec 0 _____ __._ _____ ..._ _____________ ..._ ____ __, 150 200 250 300 350 VC, kts 920319 Figure 7. Effect of calibrated airspeed on maximum roll rate for F-15 flight and simulation, full differential thrust.
The engine gross thrust and ram drag terms Pilot-in-the-Loop Tests needed to be separated since the inlet and noz- In the next flight phase, the manual throttles-only zle axes were significantly displaced. This was flight tests (with the pilot actively controlling flight- done on the Dryden simulation, and resulted in ap- path in a closed-loop fashion) were flown. These tests proximately IO-percent less roll due to differential showed that the F-15 airplane was much more difficult thrust, slightly less pitch up due to increased thrust, to fly than the simulation. Figure 8 shows a compari- and significantly more pitch down due to decreased son of approaches to a runway for the F-15 airplane thrust. The same change was also made to the and simulation. The simulation is relatively stable, McAir simulation. The effect of the changes to and only small PLA changes were required. The ac- separate the gross thrust and ram drag effects on tual F-15 airplane was never stabilized, large throttle pitch rate and roll rate is shown in Figs. 6 and 7.
Flight Simulation Pitch attitude, 4 deg -4 Bank angle, 0 deg -10 Average power ,, setting, 40 "✓ deg vc, kts 180 0 20 40 80 100 Time, sec 920293 Figure 8. Comparison of flight and simulation results for a landing approach, landing gear down, VC = 170 kts.
r
excursions were evident, and the flightpath control was some of the small throttle step maneuvers were used to compare the simulated response to that of the actual much poorer. The pilot reported strong coupling be- tween the pitch and roll axes, large thrust lags, and aircraft. The following describes the method used to mismatches between engines. Even maintaining level make the comparisons. The simulator was set to at- flight was difficult; it was not possible to attain a tain a straight and level trim that matched the flight hands-off trim condition for more than a few seconds, Mach, altitude, and fuel weight with the CAS off and even in perfectly smooth air. The flightpath control flight control in emergency, the inlets in the emergency technique in which thrust was modulated relative to position, the gear down, and the speed brake in the the velocity vector position resulted in a large ampli- proper position. The pilot had been asked to re-trim tude oscillation. the aircraft before each maneuver, and for this study an effort was made to select time segments that started The McAir F-15 simulation was flown by the same with the aircraft more or less in trim.
pilot who had flown the NASA F-15 airplane. The McAir simulation flew much like the Dryden simula- To avoid step jumps caused by any mismatch be- tion, and also did not predict the great difficulty found tween the simulation trim and the flight trim, the initial in the flights.
values of the left and right PLA from flight were sub- tracted from the respective time histories to create in- Since the F-15 simulation model was being used to cremental PLA time histories. These incremental PLA design and evaluate the augmented mode, it was criti- time histories were then added to the simulation trim cal to resolve the major differences between the flight values to drive the simulation. The flight time histories and simulation pilot-in-the-loop results. First, the en- were plotted with the time histories generated by the gine model in both simulations was improved to in- simulation for a variety of variables characterizing the corporate the nonlinear response characteristics of the response of the aircraft. There were several problems FlO0 EMO engines present at low throttle settings.
with this analysis. Since this is an open-loop compar- This made the F-15 simulation more difficult to fly, but ison between the flight data and the simulation, even with practice, it was still possible to make repeatable small differences between the model and flight tend to runway landings in the Dryden and McAir simulations.
accumulate and become large with time. Thus, these There was an additional destabilizing effect in the air- comparisons are only potentially useful for short-term plane not being modeled in the simulation that made responses. Second, there is no record of the random the airplane much more difficult to control.
external forces acting on the aircraft available to drive Additional effects were modeled, including engine the simulation. The pilot reported still air during these gyroscopic moments, which were found to be insignif- maneuvers so it can be assumed that the effects of un- icant. Vertical CG effects were also investigated. Ex- modeled atmospheric disturbances are at a minimum.
treme values (thrust line 6 in. above the vertical CG) Third, reflecting the overall difficulty of flying the air- could destabilize the simulation to the degree seen in craft engines-only, the pilot had considerable difficulty flight, but the actual range of vertical CG travel is only establishing a trim condition prior to the step inputs.
± 1 in.
There were three cases where both throttles were in- Fuel slosh was investigated. It was thought that in- creased about 1 in. In all three cases the simulation creasing power would move the fuel aft, adding more properly predicted the direction of the response, but nose-up pitching moment, and adding to the pitch re- somewhat underpredicted the pitch rate. The throt- sponse. In the roll axis, differential thrust could move tle step also excited roll rate oscillations in all three fuel in the wing tanks in a direction to reduce the cases. A typical case is shown in Fig. 9. Fan RPM is rolling moment. shown responding to the throttle increase, along with the corresponding pitch rate, roll rate, and angle of An additional flight was flown and small (approxi- attack. These small roll oscillations resulting from en- mately 1-in.) throttle steps were tested. In addition, gine mismatches were adequately modeled in the simu- tests were flown at high, medium, and low fuel levels to lation, the primary difference was that the oscillations investigate the effects of fuel quantity. The amount of in the simulation damped out more quickly than those fuel affects fuel slosh and horizontal and vertical CG, in the airplane. These differences are in accord with pi- but only small effects of fuel quantity were seen.
lot comments on the differences observed between the A batch version (non-real time) of the Dryden F-15 flight and simulation. Note that only a very small de- simulation was modified to permit throttle positions crease in angle of attack occurred, whereas the simula- measured inflight to drive the simulation. This way, tion showed a larger decrease.
t 10°-step Increase
In both throttles 1.5 Pitch 1.0 rate, deg/sec .5 -.5 1.0 .5 Roll rate, deg/sec -.5 -1.0 ...... ...... _.,_ .. _ a, ---·------------ *-- .........
........ _
deg Simulation _/
Left_/
Fan RPM 10 15 0 5 Time, sec 920294 Figure 9. Comparison of flight and simulation data for a 10°-step increase in throttle setting, VC = 175 kts.
Figure 10 shows results for a typical PLA reduc- conditions also showed the same initial pitch up and tion. The pitch rate comparisons of flight and simula- angle-of-attack increase. These results showed a serious tion data are shown where both throttles were reduced discrepancy between the simulation and flight. Fan from PLF to idle. While the long-term response of the RPM and thrust take almost 9 sec to stabilize because flight data was the expected pitch down, there was a of the slow responding engine control logic. Fan RPM significant initial pitch up. There was also a signifi- and angle of attack show a direct inverse relationship.
cant increase in angle of attack. Data at other flight Figure 11 shows a cross plot of fan RPM and angle of attack for the data of Fig. 10 and also for several and inertias. The right scale of Fig. 11 is the approxi- other cases, including another step throttle reduction mate pitching moment that is required to obtain such a and phugoid damping tests. These data represent a change in angle of attack. Although there is some vari- range of airplane weights and therefore, CG positions ability in the data, the trend with fan RPM is clear.
t Throttles PLF to Idle
.5 Pitch Simulation \ rate, -.5 deg/sec -1.0 -1.5
,,,---
,,
\r Right
' Net ' ' Leftj ' thrust, ' ' ' .
' lb -1000 ____ ""'-Right Left~ Fan RPM a, deg
---------------- --------
.... - -
.. -- -
8 --------------- -----~:~~t-i~~-~
--- -- .......
0 5 10 15 20 25 30 Time, sec 920295 Figure 10. Comparison of flight and simulation data for a step throttle decrease to idle, VC = 175 kts. (Simulation without inlet airflow effect modeled).
.008 2.0 Step throttle reduction __ ..,._, Phugold damping test \
'
\ .006 1.5 \ \ \ \ \ .004 1.0 Change Change in in angle pitching of attack, moment deg coefficient .002 0.5 -0.5 4 10x10 5 6 7 8 9 Fan RPM 920296 Figure 11. Effect of fan RPM on change in angle of attack and pitching moment coefficients, landing gear down, VC = 175 kts, angle-of-attack range 7. 7 to 11 °.
drag, and also increases the overall airplane lift, drag, Effects of Inlet Airflow and pitching moment (this would be expected with the Since the fan RPM is proportional to engine airflow, overhanging ramp configuration of the F-15 inlet). The possible airflow effects of the inlet on airplane pitching wind-tunnel pitching moment coeflicicnt data is shown moment were investigated. There had been extensive in Fig. 12 for the inlcl ramp-full-up emergency posi- wind-tunnel tests previously conducted on the effects tion. The fairing extrapolates, based on other data, of inlet airflow on F-15 inlet and overall airplane drag, to higher values of mass fl.ow ratio that occur at lower lift, and pitching moment. These data show that speeds. This pitching moment effect would produce an reducing the inlet airflow increases the inlet lift and -.018 ■ Test data Assumed fairing -.020 -.022 Pitching moment coefficient __ -.026 -.028 .20 .30 .40 .50 .60 .70 .80 Inlet mass flow ratio 920297 Figure 12. Pitching moment due to inlet mass flow ratio, F-15 7.5-percent wind-tunnel model test results, a= 8°, Mach= 0.6.
effect in accordance with the flight data, i.e., a throttle well. Although only one case is shown, similar results
r
reduction would result in a pitch up and an increase in were observed for all other tests. This airflow effect has also been incorporated in the piloted simulation.
angle of attack, which would eventually be overcome The pilot commented that with the inlet airflow effects by the speed stability effects as the velocity is reduced.
modeled, the simulator flies much more like the air- The lowest Mach number in the wind-tunnel study plane. Attempts are continuing to refine this pitching was 0.6. It is not clear how to extrapolate the results to moment effect to better match the flight data.
Mach 0.3 where the flight studies are being conducted, particularly since the mass flow ratio would have been The inlet airflow effect is small, and would often be higher at the lower Mach number. Two things are neglected in an airplane simulation. However, when the only moments being used for control are the small mo- noted from the flight data. First, the significant change in angle of attack as a function of engine RPM seems to ments from the propulsion system, normally neglected be limited to an intermediate range of fan speeds. Sec- effects may become significant. This is particularly ond, the cases where the engine was stepped up instead true for airplanes with highly integrated propulsion sys- of down did not have a comparable initial pitch down tems such as fighters where inlet-airframe interactions or significant angle-of-attack decrease as seen in Fig. 9. are strong. It would likely be less true for subsonic air- These effects are consistent with the wind-tunnel inlet planes with podded engines where the inlets tend to be airflow effects shown in Fig. 12. simple pitot inlets normal to the flow.
Based on these observations, the data from Figs. Differential Throttle Tests 11 and 12 were used to develop a piecewise linear There were four cases with primarily differential increment to the pitching moment as a function of throttle input. In all cases, the simulator responded inlet airflow with no increment being added at the with somewhat more roll rate in response to the dif- higher airflow. With this airflow effect, it has been ferential throttle input than the aircraft did. A typ- possible to substantially improve the simulator's abil- ical case is shown in Fig. 14 where the pilot initially ity to match the flight data. The results of this air- split the throttles approximately 2 in. and held that for flow effect are shown in Fig. 13, the flight data of 3 sec, then split the throttles 2 in. in the opposite di- Fig. 10 are shown with the original and updated simu- rection. The yaw rate match is very good. The result- lation. The changes in pitch rate are properly mod- ing roll rate oscillations were comparable in frequency eled, and the trend for angle of attack is predicted 'Throttles PLF to Idle .5 I ' .,,,,-- Flight
r Simulation with
/ airflow effect Pitch \ Simulatio ....
rate, -.5 deg -1.0 -1.5
r Simulation with
/ airflow effect
--
--
a, deg \ Simulation -------------------------- --------------- ........
0 5 10 15 20 25 30 Time, sec 920298 Figure 13. Comparison of flight and simulation results for a throttle step from PLF to idle, VC 175 kts (simulation with and without inlet airflow effect modeled).
and damping in the flight and the simulator response, simulation roll rates shown in Fig. 6. The inlet airflow although the roll rates were higher in the simula- effects that are important in pitch have only a minor tion than in the flight data. These roll rates agree effect on the yawing and rolling moments due to differ- with the previously collected data comparing flight and ential throttle.
:·--·---·--·· .. - \_ Right .
-- .. - ..... - ..............
' ' I ,_ PLA, ' ' I I '
deg ,_ ................. -·
r Left
Pitch rate, deg/sec Simulation Roll rate, deg/sec -5 -10 Yaw rate, -2 deg/sec -4 Simulation ___,;·-- -6 0 5 10 15 20 25 Time, sec 920299 Figure 14. Comparison of flight and simulation response for a differential throttle input, VC = 175 kts.
Augmented Throttles-Only Control system will provide the feedback signals and digital en- gine control systems on each engine will be used to System move the throttles to the commanded position. The Manual throttles-only control is difficult for up-and- PCA control logic will reside in the digital flight-control away flight and a successful landing on a runway would computer.
be extremely unlikely for the NASA F-15, based on pi- lot comments. However, an augmented propulsion con- Concluding Remarks trolled aircraft (PCA) concept shows promise of being A flight and simulation evaluation of the throttles- able to make repeatable runway landings. Figure 15 only control capability of the F-15 airplane has been shows an augmented PCA system designed for the conducted. Principles of throttles-only control have F-15. Appropriate feedbacks are used to stabilize been shown. Initial flight-to-simulation comparisons the pitch and roll axes. Thumbwheel controllers re- were good for differential throttle and increasing throt- mind the pilot that the system is a slow-response, low- tle, but were poor for decreasing throttle. Detailed authority system. Initial simulation results based on comparisons of flight and simulation data have revealed the first NASA Dryden and McAir simulation showed an unmodeled pitching moment effect thought to be that the system worked well. More recently, the up- caused primarily by inlet airflow. The inlet airflow ef- dated simulation model, which flies much like the air- fect is small. However, when the only moments being plane and incorporates inlet airflow effects, has been used for control are from the propulsion system, nor- used to evaluate the PCA system. Although phugoid mally neglected effects may become significant. This damping is reduced, PCA system performance is still is true for airplanes with highly integrated propulsion adequate at the lower speeds. At higher speeds, gain systems such as fighters where inlet and engine interac- changes and the addition of airspeed feedback make tions are strong, but less true for airplanes with pod- the performance of the PCA system satisfactory. The ded engines. Incorporating this effect into the simu- flight-test control laws have the capability for changing lations has greatly improved the simulation-to-flight gains, which will help with solving problems that oc- comparisons. Based on simulation results, an aug- cur during the flight evaluation. Based on the simula- mented throttles-only feedback control system shows tion, repeatable runway landings with this PCA system promise of making repeatable runway landings of the should be practical.
F-15 airplane practical.
A flight demonstration of this PCA system on the NASA F-15 is planned. The digital flight-control Left Throttle Command Computer Right Throttle Command Yaw Angle Bank angle ._'4-_R_o_ll_R_a_te-'--_ Bank angle thumbwheel Bank angle software Bank Angle lllllll I 1111 I !llllli--c.;:..;:o=m=m=a=n=d-H~ Pitch Rate Pitch attitude Pitch Pitch Attitude thumbwheel attitude Velocity software Pitch attitude command 920300 Figure 15. Schematic view of the augmented propulsion controlled aircraft system for the F-15.
References Burcham, F., Jr., C. Fullerton, G. Gilyard, T. Wolf, and J. Stewart, "A Preliminary Investigation of the Use Burcham, Frank W. Jr. and C. Gordon Fullerton, of Throttles for Emergency Flight Control," AIAA-91- Controlling Crippled Aircraft - With Throttles, NASA 2222, June 1991.
TM-104238, 1991.
Kamrnan, J.H. and H.W. Wallace, "Assessment of Gilyard, Glenn B., Joseph L. Conley, Jeanette L.
Installed Inlet Forces and Inlet/ Airframe Interactions," Le, and Frank W. Burcham, Jr., "A Simulation Eval- AFFDL-TR-76-62, July 1976.
uation of a Four-Engine Jet Transport Using Engine Thrust Modulation for Flightpath Control," AIAA-91- 2223, June 1991.