вход по аккаунту



код для вставкиСкачать
David J. Moorhouse*
David B. Leggett**
Kenneth A. Feeser***
AF Wright Research and Development Center
Wright-Patterson AFB
Downloaded by UNIVERSITY OF NEW SOUTH WALES (UNSW) on October 27, 2017 | | DOI: 10.2514/6.1989-3390
The STOL and Maneuver Technology Demonstrator
(S/MTD) program was defined to develop technologies
that would enable a supersonic fighter to land on a 1500
x 50 ft usable strip of runway. A short ground roll is
achieved using thrust reversing plus control laws
to minimize touchdown dispersion. The
final /h4TD configuration includes a speed hold
system (i.e. augmented airspeed stability) and pitch rate
command from the stick. Pilot ratings in a piloted
simulation were Level 1 with the speed hold and Level
2 without it.
An additional piloted simulation
experiment was conducted to investigate the trade-off of
pitch axis bandwidth and speed stability.
The 03jectives of this paper are to, first, document
the rationale and SIMTD approach to precision landing
flying qualities and second, to discuss more general
design criteria developed from the piloted simulation.
The S/MTD program was structured to investigate,
develop and validate through analysis, experiment and
flight test, four specific technologies related to
providing current and future high performance fighters
with both STOL capability and enhanced combat
mission performance. These technologies are:
Two-dimensional thrust vectoring and reversing
exhaust nozzle
Integrated flight/propulsion control (IFPC)system
Pilot Vehicle Interface (PVI)
Roughlsoft field landing gear
These technologies have been incorporated into a
two-place F-15B together with all-moving canard
surfaces (see Figure 1).
The contract was written in terms of performance
requirements, as much as possible, and those pertaining
to landing are:
Minimum operating strip of 1500 ft x 50 ft
Crosswinds up to 30 kts (disturbances as defined
by MIL-F-8785C, MODERATE levels)
Wethcy runway surface conditions
No ground-based landing guidance aids
*Chief Engineer, S/MTD ADPO, Senior Member
**Flying Qualities Engineer
***Simulation Test Manager, Member AIAA
This paper is declared a work of the U.S. Government and
is not subject to copyright protection in the United States.
Integrated Fllghff
ProDuislon Control-
Crow Statlon
Landing Gear
Figure 1. S/MTD Configuration
Field length allowances for flare and touchdown
dispersion can be on the order of 1,000 ft which is
obviously not compatible with the above requirements.
The specified field length is an implicit requirement that
landing flying qualities in pitch be designed for precise
touchdown and not allowed to "fall out". Similarly,
the 50 ft width and 30 kt crosswind drive the
lateral/directional flying qualities development.
Another factor in the solution to these requirements is
that the approach speed of 119 kts is not significantly
less than the production F-15's. The short landing
distance is achieved by reverse thrust rather than
reduced touchdown speed.
The ovemding requirement of the IFPC system was
stated to be "capable of functionally integrating all
aspects of flight, engine, and nozzle control including
aerodynamic control surfaces, engine thrust, thrust
vectoring, thrust reversing and differential efflux
modulation, control and stability augmentation, high lift
system, steering and braking". The intent was to
convey the understanding that integration was an
objective of the demonstration program, not just a
means to .achieve mission requirements. The IEPC
system was required to provide "good inner-loop
stability and positive manual control in all axes of the
air vehicle throughout its intended operating envelope
both in flight and on the ground (satisfying the intent of
ME-F-8785C)". This subjective requirement was
intended to convey that we were seeking good flying
qualities over the whole envelope guided more by the
intent than the letter of the specification. This
recognizes that, while the intent is to provide flying
qualities clearly adequate for the mission, the letter of
the specification is no guarantee. In addition, the
requirement for "positive manual contrql" was
intended to preclude consideration of automatic landing
systems, for instance.
Downloaded by UNIVERSITY OF NEW SOUTH WALES (UNSW) on October 27, 2017 | | DOI: 10.2514/6.1989-3390
The objective of this paper is, first, to document the
control law approach that allows touchdown in a
"box" that is 60 ft long by 20 ft wide. The second
objective is to document results of a piloted simulation
experiment to develop pitch axis design criteria.
At the start of the S/MTD program there were
questions as to whether Navy experience could be used.
The Navy has stringent requirements for touchdown
accuracy in the case of landings on aircraft carriers.
Four arresting wires are spaced within a distance of 150
ft, so that touchdown must be made to this accuracy or
a go-around has to be executed. The requirements are
met with carrier-based guidance, and either a manual or
coupled approach using a flight control system known
as an Approach Power Compensator System (APCS).
Reference 1 analyzes and documents development
problems with APCS: "Trial and error is the essence
of the design procedure now employed in APCS
development. While such a procedure is usually to be
avoided and certainly not condoned, it has proven
necessary and reasonably effective under the
circumstance". Each new aircraft is the subject of
intense development to meet the operational
requirements. The current APCS implementations are
angle-of-attack based which implies coupled airspeed
and flight path responses. Quoting from Reference 1:
"The basic manual control technique, involving
operation of both stick and throttle, emphasizes the
necessity of using thrust plus attitude as the combined
means of maintaining reference airspeed and angle of
attack on the glide slope. The (old) notion of using
throttle to control altitude and nose attitude or stick
input to control airspeed is pointed out as not being
wholly valid in all situations along the approach".
Conversely, in discussing requirements for a coupled
approach, "He must depend on the APCS to provide
the proper thrust inputs; and quite often this means he
must change to a more suitable piloting technique to
obtain reasonably acceptable APCS action". Thus,
Navy pilots typically must learn (and train) to fly two
different landing techniques, manual and coupled, and
possibly different from one aircraft to another. The
purpose of this section is not to criticize the Navy
system, the Air Force typically puts little effort into
landing flying qualities, but rather to provide
background for the selection of the S / M T D design
There are also significant differences between the
landing tasks of the two services. In normal operation,
the carrier steams into the wind producing a turbulence
environment which tends to be deterministic and
repeatable, although definitely severe. The S/MTD
configuration was designed to land in winds gusting to
30 knots from either direction, turbulence as a function
of altitude and both magnitude and vector wind shears.
Further, in a carrier landing the pilot either catches the
wire or goes around. Conversely, the S/MTD has to
touch down with heading under control to remain
within the 50 ft width during the rollout in 30 kt
crosswinds, on wet or dry surfaces and using reverse
thrust. Lastly, low pilot workload was required with
only conventional (Air Force) pilot techniques. The
S/MTD landing requirements are believed to be unique
and not achievable by any other aircraft. They led to
the development of a special SLAND control mode.
The Landing Control Laws
An overall requirement of the S/MTD development
was to integrate all the control effectors to satisfy the
requirements with conventional piloting techniques.
The short landing distances are facilitated by providing
maximum reverse thrust immediately after touching
down. To achieve this, the final approach is made with
the engine spooled up to 100% RPM and the exhaust
passing through vanes which are controllable from 45
degrees aft to 45 degrees forward of normal. These
fast-acting vanes also provide for high-bandwidth
control of airspeed because there is no influence of
engine spool-up time.
Design of the SLAND longitudinal control laws
(pitch and thrust axes) was accomplished using
multivariable control techniques. The complete design
requirements are presented in Figure 2, showing the
required form of the pitch and airspeed responses.
There are also requirements to decouple the two axes,
so that there is neither airspeed response to a stick input
nor pitch rate response to throttle input. The final form
of the control laws is given in Figure 3. Two features
are to be noted -- decoupling is achieved by sending
both pitch and thrust commands to the upper and lower
reverser vanes, and the form of the response in each
axis is defined by a command prefilter. Now, in terms
of pilot action, the feedback of airspeed to vane angle
holds airspeed constant until the throttle is moved. The
pilot retains control because throttle movement
commands a new airspeed and consistent flightpath
change. The stick commands pitch rate directly but,
with speed held constant, it is effectively a flightpath
rate command. On approach, the pilot sets airspeed
with throttle and, once the correct airspeed is acquired,
flightpath is controlled solely with stick input.
The lateral/directional control laws are conventional
with the addition of direct sideforce control to satisfy
the crosswind requirement. This is achieved with
differential canard plus rudder deflections commanded
by the rudder pedals during the approach and by lateral
stick after touchdown. This integration of direct
sideforce is totally natural in terms of pilot technique,
giving the appearance of a crosswind half the actual
magnitude. In the piloted simulations, the lateral
touchdown requirement was easier to meet than the
longitudinal (pitch) requirement. The development
effort (and this paper), therefore, concentrated on the
pitch and thrust axes. The final S/MTD configuration
does meet the requirements and is rated Level 1 in the
piloted simulation.
Existing Criteria
The S/MTD contract was written to the "intent" of
MIL-F-8785C in general, but a change was directed for
landing characteristics. Precise landing was designated
a Category A tracking task rather than a Category C
Downloaded by UNIVERSITY OF NEW SOUTH WALES (UNSW) on October 27, 2017 | | DOI: 10.2514/6.1989-3390
Figure 2 .
SLAND Mode Design Requirements
Figure 3 .
SLAND Mode A r c h i t e c t u r e
Downloaded by UNIVERSITY OF NEW SOUTH WALES (UNSW) on October 27, 2017 | | DOI: 10.2514/6.1989-3390
terminal area task. The requirement and the effect of
the change are shown in Figure 4. The minimum
allowable short period frequency is increased, which
will tend to increase pitch attitude bandwidth, all other
things being equal.
Explicit requirements for
bandwidth form an alternate criterion suggested in
MIL-STD-1797 (which has replaced MIL-F-8785C as
the Air Force flying qualities specification), as shown in
Figure 5. It is worth noting that the difference between
the two categories is far more pronounced in Figure 5
than in Figure 4. The various requirements for
precision landing flying qualities were discussed in
Reference 2 before the SIMTD contract award. A
minimum pitch attitude bandwidth of 3.5 radsec was
suggested because the proposed MIL-Standard did not
differentiate between "conventional" and STOL or
precision landings.
Figure 4. Modified MIL-F-8785C Requirements
Designing to a given bandwidth poses a problem of
exactly how to increase the value of a deficient
response. In designing the fix to an up-and-away
tracking problem during the S/MTD development
(Reference 3), both the numerator time constant and
short period frequency were increased equally to
increase pitch bandwidth without introducing other
problems. Conversely, the required change in Figure 4
would imply increasing short period frequency while
keeping a fixed (natural airframe) value of numerator
time constant. This raises the question of whether the
optimum increase in bandwidth depends on the relation
between numerator time constant and short period
frequency. Another question was raised by the final
configuration of the SLAND mode. The SIMTD pitch
attitude bandwidth is approximately 2.6 radsec with a
phase delay of 0.075 seconds. The Category A
requirements of Figure 5 indicate that Level 1 cannot be
achieved because of the magnitude of the phase delay.
The Category C requirements suggest a bandwidth
between 2.5 and 4.2 rad/sec is required for Level 1.
The S/MTD simulation pilot ratings were Level 1 with
the speed hold and Level 2 without it. A piloted
Figure 5. Bandwidth Requirements in MIL-STD-1797
simulation experiment to address the above questions
was therefore set-up in the Flight Dynamics Laboratory.
Simulation Experiment
Downloaded by UNIVERSITY OF NEW SOUTH WALES (UNSW) on October 27, 2017 | | DOI: 10.2514/6.1989-3390
Simulator Descri~tion
The study was conducted in the Large Amplitude
Multi-mode Aerospace Research Simulator (LAMARS)
at the Flight Dynamics Laboratory, Wright Patterson
Air Force Base. The LAMARS consists of a single-seat
cockpit installed in a 20 ft diameter dome attached to
the end of a 30 ft beam. The combined beamldome
movements result in a five degree-of-freedom motion
system which can generate angular velocities of up to
60 degrees per second, linear velocities up to 13 feet per
second, and instantaneous accelerations of up to 3 g's.
The LAMARS cockpit represented to the extent
possible the forward crew station of the S/MTD aircraft
(Figure 6). An F-16 LANTIRN Wide Field of View
(WF,OV) Head-Up Display (HUD) and two 6 inch
square Multi-Purpose Displays (MPD's) were the
pilot's primary flying instruments. The S/MTD center
stick and rudder characteristics were modeled by a
McFadden three axis feel system. Toe brakes with the
standard F-15 force-feel characteristics and S/MTD
throttle quadrant with a spring loaded thrust reversing
region were also provided.
Task Descri~tion
The experiment consisted of an approach task to a
STOL landing on a 50 x 1500 ft Minimum Operating
S e p . The simulation was initialized at 150 kts
calibrated airspeed (KCAS), 1000 ft above ground level
(AGL), approximately two nautical miles from
touchdown, with flaps and gear down in the SLAND
control mode. The task was for the pilot to follow the
command steering guidance on the HUD (Figure 7) and
achieve a landing speed of 119 KCAS to the
touchdown. Atmospheric disturbances consisting of
light turbulence and a head-to-tail wind shear were
given to keep the pilot in the loop at all times during the
task. Also, the command steering cues were initially
offset 200 ft from the touchdown area both in the
longitudinal and lateral direction. At approximately 200
ft AGL, these cues were then corrected to the
touchdown area on the landing strip. This offset
provided the pilot with a gross acquistion task which
was a valuable aid in his evaluation of the various test
conditions. The pilot gave Cooper-Harper (CH) ratings
only for the approach task which was from initialization
to approximately 50 ft AGL. The CH ratings were
based on adequate and desired criteria that had been
determined from earlier evaluations.
performance was defined as keeping the velocity vector
"boxed" by the command steering bars (i.e., keeping
the flight path marker within the length of the steering
bars though not necessarily on them) and hold the
airspeed within 10 kts of the 119 kt landing speed.
Desired performance was defined as keeping the
velocity vector within half the length of the command
steering bars and hold the airspeed within 3 kts of the
119 kt landing speed.
Test Mamx
At the appropriate value of phase delay, Reference
4 indicates that pitch attitude bandwidth for a landing
task should be between 2.5 and 4.5 rad/sec (see Figure
5). Neal-Smith (Reference 5) indicates the Level 1
range to be 1.6 to 3.1 rad/sec. The S/MTD pitch
attitude response to stick input was nominally designed
for a bandwidth of 2.75 rad/sec, i.e. apparently Level 1.
Because of the mixed results noted previously, an
experimental text matrix was set up to investigate
bandwidths from 1.5 to 4.5 rad/sec. These changes
were achieved by varying appropriate parameters in the
stick command prefilter (Figure 3) which defines the
response characteristics. The bandwidth changes were
produced two different ways consistent with the
previous discussion, by varying short period frequency,
o only (the A configurations) and also by varying
bc% as and the numerator time constant, L,, (the B
config&tions) as shown in Figure 8.
Figure 6. S/MTD Cockpit Displays on LAMARS
The influence of speed stability was investigated by
changing the gain on airspeed feedback to the reverser
vane angle. With the gain at its nominal value ("with
speed hold") the speed response would be independent
of pitch inputs, as in the S/MTD. With this gain zeroed
out ("without speed hold") speed response would be
more like that of a normal F-15, which is stable, but
allows the speed and angle of attack to change with
pitch changes. The complete experimental mamx was
Downloaded by UNIVERSITY OF NEW SOUTH WALES (UNSW) on October 27, 2017 | | DOI: 10.2514/6.1989-3390
Barornetrlc Altlfude
Thrust Reversing
Vane Advisory
Operable Strip
'E Bracket
Radar Altitude
Vertical Velocity
Velocity Vector
Autl S t e e r l n ~
Flare Cue 1
Figure 7. SLAND Mode HUD Display
With SDeed Hold
Pilot ratings as a function of bandwidth are shown in
Figure 9 for the configurations with speed hold. The
height of the bars indicates the range of ratings received
for each bandwidth. The horizontal bars show the
average ratings. The numbers in parentheses at the top
of the chart indicate the number of ratings given for
each bandwidth.
Figure 8. Test Configurations
to fly the landing task at each value of both series of
pitch bandwidths both with and without speed hold.
Simulation Results
A Confinurations vs. B Confinurations
It is difficult to draw any conclusions when only one
pilot has flown the cases where we changed bandwidth
by changing both L, and wsp (the B configurations).
However, comparison of his results with the results
from changing only w indicate the same comments as
those given by the othg pilots for each bandwidth. All
of his ratings for the B configurations fell within the
ranges of those given for the A configurations and they
followed the same trends. At this point, over this range
of bandwidths (1.5 to 4.5 rad/sec), we have seen no
difference between changing bandwidth by changing
only w and changing bandwidth by changing both
L, andPws
Because of the similarity between A
configuratiogi and B configurations,we have combined
the results for both cases in the discussion that follows.
Comments for the 1.5 rad/sec case describe it as
sluggish and slow to respond and the stick forces felt
heavy. One pilot said "it felt more like a bomber than a
fighter". The pilots felt like they had to overdrive it to
achieve desired performance and, consequently, there
was a tendency to overshoot. All of the pilots felt that
they could achieve desired performance but it took
moderate workload to compensate for the slow
response. Pilot ratings were borderline Level 1/Level2,
mostly Level 2.
Comments for the 2.0 radlsec case were mixed. One
of the pilots had comments similar to those for the 1.5
rad/sec case. The other pilots felt that it was quite
good, though sometimes they said it was a little
sluggish, but not too bad. The pilot who did not like it
gave CH 4's because of the sluggish response. The
other pilots' ratings were CH 2's and 3's. For them the
turbulence and t!e slightly sluggish response, when
they did perceive it, were minor annoyances. All of the
pilots agreed that they could achieve desired
performance, the difference of opinion was how much
workload was involved.
There was no such difference of opinion for the next
two cases. AU of the pilots liked both the 2.6 and the
3.0 rad/sec cases and the comments and ratings were
very similar for both. Speed of response, precision, and
stick forces were all satisfactory. The pilots felt that
they could put the flight path marker exactly where they
wanted to and hold it, but occasionally they sensed a
slight sensitivity when trying to hold it on. The
Downloaded by UNIVERSITY OF NEW SOUTH WALES (UNSW) on October 27, 2017 | | DOI: 10.2514/6.1989-3390
BANDWlDTH (radlsec)
Figure 9. Pilot Ratings With Speed Hold
sensitivity was not regarded as annoying, they
mentioned it only as an observation. Workload was a
factor only because of the turbulence, which forced
them to stay in the loop. Pilot ratings for both of these
cases were all Level 1.
Comments for the 3.5 and 4.0 rad/sec cases were
also very similar. Response was described as quick and
crisp, but there was a slight tendency for pilot-induced
oscillations (PIO). For the most part the pilots felt that
the PI0 tendency was no more than mildly unpleasant.
They could track within desired criteria, they just had a
little more trouble holding the flight path marker
precisely on target. Ratings were slightly more diverse,
ranging from CH 2 to 4. The 3.5 rad/sec case received
only one CH 4, all of the other ratings were Level 1.
The 4.0 radlsec case received all Level 1 ratings.
In the 4.5 rad/sec case the PI0 tendency became
more pronounced. Though the pilots still described it as
a "slight" PI0 tendency, it graduated from mildly
unpleasant to annoying. For the most part the pilots
were still able to achieve desired performance, but they
had to compensate more to avoid overcorrecting. In
some cases they felt that they could do no better than
adequate. One pilot felt that the PI0 tendency was
insignificant and rated this case CH 3. The other pilot's
found it more annoying and their ratings ranged from
CH 3 to 5, mostly Level 2.
From these results it would seem that for a phase
delay of .075 the Level 1 boundaries for bandwidth
would go from about 2 rad/sec to around 4 rad/sec,
which corresponds very closely with the criteria in
MIL-STD-1797. Below 2 rad/sec the aircraft response
is too sluggish. Above 4 rad/sec the response is too
sensitive, at least for the stick gradient we used.
Without S ~ e e dHold
Unfortunately, at this point we only have one pilot's
evaluation of the configurations without speed hold.
But his comments and ratings show significantly
different trends from those configurations with speed
hold. The pilot ratings for the cases without speed hold
are shown in Figure 10.
Comments and ratings for 1.5, 2.0, and 2.5 rad/sec
were all similar. The pilot said that the pitch response
was too slow and also somewhat unpredictable. He
described it as a tendency to "balloon". He said this
tendency made it very difficult to put the flight path
marker precisely where he wanted to. It felt like a
low-frequency PIO. He also mentioned that he had to
devote considerably more attention to speed control and
that this aggravated his problems in pitch and flight path
It was impossible to achieve desired
performance and it was difficult to keep it in the
adequate range in both speed and flight path control.
He rated the 1.5 and 2.0 rad/sec cases CH 6, and the 2.5
rad/sec case CH 5.
He had similar comments for the 3.0 rad/sec case
except that it was much more predictable. It was still a
little sluggish and, consequently, there was a tendency
to overshoot when he med to force it, but he did not get
into the low-frequency PIO. He still had to devote
considerable attention to airspeed, and this kept his
workload high, but he was able to achieve desired
performance. He rated this case a CH 4.
Comments and ratings for the 3.5, 4.0, and 4.5
Downloaded by UNIVERSITY OF NEW SOUTH WALES (UNSW) on October 27, 2017 | | DOI: 10.2514/6.1989-3390
BANDWIDTH (radlsec)
Figure 10. Pilot Ratings Without Speed Hold
rad/sec cases were all similar to each other as well. For
these cases, speed of response and predictability were
good. There was no tendency to overshoot or PIO.
Speed control was easier and, as a result, interfered less
with his attention to flight path control. The pilot
achieved desired performance with minimal workload.
He rated all three of these cases CH 3. The minor
deficiency was the workload involved in controlling
both speed and flight path in the turbulence. He
commented that he had definitely preferred the speed
hold cases, because it left him free to concentrate solely
on flight path control.
With only one pilot evaluation so far it would be a
mistake to try to draw any conclusions about boundaries
for the case without speed hold, but we do feel that
these results indicate that the pilot prefers a higher
bandwidth without speed hold than he does with speed
hold. His comments also indicate a definite preference
for speed hold, regardless of bandwidth.
Control laws have been designed to enable an F-15B
to achieve very precise touchdowns in stringent
environmental conditions. These control laws feature
augmented airspeed stability (speed hold) and
decoupled airspeed and pitch rate responses. Piloted
simulation verified the predicted Level 1 pilot ratings,
but only with the speed hold feature.
A piloted simulation was conducted to define the
required pitch attitude bandwidth with and without
speed hold. The Level 1 requirements on pitch attitude
bandwidth in MIL-STD-1797 are verified with the
speed hold, i.e. with a high level of airspeed stability
and with minimal coupling between the speed and pitch
responses. Preliminary results indicate that higher
values of pitch bandwidth are required if there is
coupling between the axes or airspeed stability is low.
1. Craig, S. J.; Ringland, R. F.; and Ashkenas, I. L.,
"An Analysis of Navy Approach Power Compensator
Problems and Requirements," Systems Technology,
Inc., ST1 Technical Report No. 197-1, March 1971.
2. Moorhouse, D. J., "STOL Flying Qualities and the
Impact of Control Integration," NAECON 83, May
3. Bland, M. P.; Citurs, K. D.; Shirk, F. J.; and
Moorhouse, D. J., "Alternative Design Guidelines for
Pitch Tracking," AIAA Paper 87-2289, August 1987.
MIL-STD-1797(USAF), "Flying
Piloted Vehicles' ', March 1987.
Qualities of
5. Neal, T. P., and R. E. Smith, "An In-flight
Investigation to Develop Control System Design
Criteria for Fighter Airplanes", AFFDL-TR-70-74,
December 1970.
Без категории
Размер файла
710 Кб
3390, 1989
Пожаловаться на содержимое документа