Once the redundancy of the thrust-vectoring system was assured and ground clearance of the vanes on takeoff and landing was determined, the vanes were installed and then flown on the 10th flight of Ship 1 on February 14, 1991, with Ken Dyson at the controls. The vanes were not used to vector thrust on that flight. The initial test with the thrust-vectoring vanes installed consisted of flying with the vanes in a commanded fixed position out of the exhaust plume to validate that vibration, acoustic levels, and temperatures were within thrust-vectoring vane specifications. Following this checkout and postflight evaluation, the X-31 flight-test team would be better positioned to plan the first use of vanes in the exhaust to enhance agility.
With Ships 1 and 2 both active, the program moved quickly along with new international and American pilots. On March 15, 1991, Karl Lang of the Wehrtechnische Dienststelle-61 (WTD-61) at Manching flew the X-31 for the first time.6
Following the start of flight envelope clearance, the X-31 test team under-took initial Government Performance Evaluation (GPE) flights even though it was still relatively early in the test program. The concept of a Government Performance Evaluation began as a Navy process whereby a Government pilot would fly an aircraft that was involved in a contractor flight-test program to
Into the Air: Initial Flight Testing
X-31 Initial test pilots at Plant 42: (left to right) Fred Knox, Dietrich Seeck, Ken Dyson, and Karl Lang. (Rockwell)
gain an early evaluation of the performance, flying qualities, and systems opera-tion of the aircraft, but only within the flight envelope already cleared by the contractor pilots. In the case of the X-31, U.S. Marine Corps test pilot Maj.
Bob Trombadore undertook the first GPE flight on April 24, 1991. German Air Force Maj. Karl-Heinz Mai followed this with a second flight on April 30, 1991, and the two pilots undertook two more flights on May 2 and 3, 1991, to complete the first series of GPE tests.
Initial use of the thrust-vectoring paddles occurred on a test flight on May 31, 1991, with Dietrich Seeck at the controls. These missions were known as “plume line” flights, during which Seeck moved the vanes up to 15° at alti-tudes from 10,000 to 40,000 feet MSL. The vanes were moved to map the jet plume line as a function of nozzle area and nozzle pressure area. This sort of data was required to produce a table for the flight control computers so that nozzle effects could be included in the flight control computer calculation, a requirement for an integrated flight and propulsion control system. Flights with the vanes automatically tracking the plume line showed that there was no effect on aircraft handling and no uncommanded aircraft movements.
Fred Knox subsequently reported that these early evaluations during the first year of flight testing indicated that the X-31 had excellent handling qualities,
Flying Beyond the Stall
particularly in the power-approach configuration; its measured flight-test per-formance was well matched to preflight ground simulations; it was shown to be a reliable aircraft, easy to fly and operate; and its GE F404 engine was flawless in its operation.7
In contrast to Rockwell, however, the German view of the X-31’s reliability was less positive. German test team members believed that, despite the use of many “off-the-shelf” components (all of which were flight-certified), flight testing had uncovered many weaknesses and problem areas that needed to be resolved. The majority of flight maintenance squawks or discrepancies in the first 63 flights (representing nearly a year of flying) related to the environmental control system, flight control system hardware and software, and flight-test instrumentation.8 Since the flight control system was pushing the state of the art in combining flight and thrust control, it is perhaps not surprising that this system should have some difficulty in attaining maturity. Also, since flight-test instrumentation for a given aircraft type is usually a “one-off” design for just that specific aircraft flight test, it is not unusual for there to be flight-test instrumentation problems early in a flight-test program. Maintenance prob-lems did require parts from one airplane to be “borrowed”—or “cannibalized,”
as the military terms it— more often than was desired. Despite this occasional cannibalization, Rockwell and the Navy consistently provided parts quickly and efficiently through their normal supply channels.
Initial flight clearance limits for the aircraft were set by Naval Air Systems Command, the Navy being DARPA’s agent for the program. These initially were an AOA limit of 30°, an altitude limit of 30,000 feet MSL, a Mach limit of 0.67 (365 knots true airspeed), and a structural load limit of just 4 g’s. These were gradually expanded to approach the desired conventional envelope of the aircraft, which was an altitude limit of 40,000 ft MSL, Mach limit of 0.9, a structural limit of 7.2 g’s, and an AOA limit of 30°.9 Throughout the clearance of the conventional flight envelope, pilots reported excellent handling quali-ties; Level 1 handling qualities were reported up to 30° angle of attack with and without use of thrust vectoring. (Flying qualities are reported in varying levels as defined in a military specification, with Level 1 being the best and indicating “[f]lying qualities clearly adequate for the mission flight phase.”10) While some flying quality anomalies were found in this first year of flight test-ing, they were not so significant as to impede the ever-so-important envelope expansion. Some of these anomalies are discussed below.
While conducting initial flutter testing, pilots discovered a roll-response asymmetry: there was more roll rate and acceleration to the right than to the left, and roll sensitivity to the right was unacceptable. This sensitivity reflected an almost full-right roll-trim requirement above 300 knots calibrated airspeed that was caused by the flight control law’s lateral trim mechanization. This
Into the Air: Initial Flight Testing
problem was fixed by adjusting the trailing-edge flap rigging to minimize lat-eral trim requirements and roll-response asymmetry. This provided acceptable roll response and allowed flutter testing up to 500 knots calibrated airspeed.
A subsequent software change that modified trim inputs from rate commands to direct position bias was tested in simulation and applied as a permanent fix to the roll-response asymmetry problem.
Overall, project test pilots felt that the pitch and roll response of the aircraft was, in their terms, “snappy” above 300 knots calibrated airspeed. After exam-ining damping, frequency, and bandwidth in the pitch axis, project engineers made adjustments to the stick mechanization. Also, roll-time constants seemed normal, but after a permanent fix to the roll-trim issue, this was evaluated further to determine if a software change would be required to reduce roll-rate onset at high dynamic pressure.
The X-31 had three reversionary modes built into its flight control system because there was limited redundancy in three critical measurement channels.
There were only two means of sensing angle of attack and angle of sideslip.
Additionally, there was only one inertial navigation unit. Therefore, it was important not only that the reversionary modes operate properly with good handling qualities, but also that there were essentially no transient responses when transitioning from a normal mode to a reversionary mode. The “R1”
reversionary mode handled inertial navigation unit failures, the “R2” reversion-ary mode handled failures of angle-of-attack and angle-of-sideslip sensing, and the “R3” reversionary mode handled air data failures. Intentional flight-test events plus unintentional anomalies allowed assessment of these modes. (There was one failure of the angle-of-attack/angle-of-sideslip sensing that required a landing in the R2 reversionary mode.) On July 12, 1991, Fred Knox was flying a flutter-data-focused flight when an unresetable sideslip data failure caused an R2 request. Fred slowed the airplane and selected R2 with no apparent transient responses. He configured for landing and performed a brief flying qualities evaluation that was satisfactory. Due to relatively high crosswinds on the only available runway at Palmdale and the fact that the X-31 had a fairly high weight due to the high remaining fuel load, the decision was made to land at Edwards AFB. The landing was made uneventfully and the airplane was ferried back to Palmdale 2 days later. These reversionary modes all exhibited Level 1 flying qualities.
During a level deceleration in the R1 reversionary mode (which is the INU failure mode), a 13-Hz surface oscillation occurred as the X-31 passed through 200 knots calibrated airspeed. Postflight analysis revealed a coupling of the noseboom (which furnished secondary AOA sensing) structural mode to the flight control system in the R1 mode. This did not appear in the normal (i.e., INU operating normally) mode because the INU acts as a low-pass filter for
Flying Beyond the Stall
the boom signal. This was initially filtered out with a low-pass filter to allow continued expansion of the high-AOA envelope, with the final fix being a notch filter at 13 Hz.
Flutter testing was conducted out to the limiting dynamic pressure of 800 pounds per square foot. The 800-pounds-per-square-foot dynamic pressure line is along the 485 knots equivalent airspeed line, which is equal to 485 knots calibrated airspeed at sea level and Mach 0.9 at approximately 10,100 feet MSL. The flutter excitation was provided by direct electrical commands to the actuators of individual flight control surfaces through a flutter-test box that was a part of the flight-test instrumentation system. This is a normal mechanization on an aircraft with an electronic flight control system. Flutter margins were as predicted and pilots observed that the aircraft’s ride quality in turbulence was excellent, allowing for continuation of flutter testing even when atmospheric conditions were somewhat turbulent, as is characteristic of the hot desert afternoons around Edwards AFB and the R-2508 airspace complex.11
Flight at high angles of attack while using thrust vectoring was the heart of the X-31’s reason for being. As discussed in the first chapter, the program’s object was to demonstrate departure-free operations with thrust vectoring on or off (initially at 30° angle of attack or below). Having demonstrated this, the same departure-free characteristics were then to be demonstrated between 30°
and 70° with thrust vectoring on. Additionally, the thrust-vectoring system had to exhibit “fail safe” operation above 30° angle of attack. Initially, AOA expansion was conducted to 30° with thrust vectoring off, then with thrust vectoring on. Testing of the thrust-vectoring system started at Mach 0.6 and proceeded to both lower and higher speeds. Since the low-speed side of the flight envelope was as important as the high-speed side, this was a true “build-up” technique to expand the thrust-vectoring envelope. The integrated flight and propulsion control system provided identical flying qualities with thrust vectoring both on and off below an angle of attack of 30°. No differences in flying qualities were expected and, in fact, none were noted once flight testing explored this environment. Instead, the pilots reported that the airplane felt the same whether thrust vectoring was engaged or not. Testers evaluated the X-31’s high-AOA handling qualities using the standard flight-test maneuvers of doublets, rolls, steady-heading sideslips, and windup turns. Before flight, engineers had predicted that the X-31 would demonstrate Level 1 handling qualities, and they were validated in full-flight, again confirming the basic fidelity of the preflight modeling and simulation to actual flight-test results.
At elevated angles of attack above 13°, the airplane showed light buffet. Rapid pitch step inputs to 20° angle of attack or above produced small, rapid wing drops. Control of angle of attack was reported as precise, and angle of sideslip remained at 4° or less during maximum deflection rolls at 20° angle of attack.
Into the Air: Initial Flight Testing
Roll performance at 1 g and elevated g was termed outstanding. As the airplane decelerated to slower flight speeds, the thrust vectoring provided increasing amounts of control power in comparison to the control power generated by its conventional aerodynamic controls. Test pilots characterized the X-31 as
“comfortable and solid” during this phase of testing, all the way up to 30°
angle of attack.12 The final test of the integrated flight and propulsion control system was to complete a 360° roll around the velocity vector, which the aircraft executed with extreme accuracy.13
Having cleared the conventional envelope, it was time to penetrate the stall barrier and enter the post-stall envelope. On November 21, 1991, during Ship 2’s 36th flight, Fred Knox flew the airplane to 40° angle of attack. This was near the maximum lift coefficient for the aircraft and is most critical because sudden airflow detachment or a vortex “burst” at that point could cause real surprises.
Unfortunately, a computer failure triggered the automatic recovery mode, termi-nating post-stall flight before Knox could explore this regime in detail.14
The day before Knox’s foray to high AOA, a new series of GPE flights began.
U.S. Navy Cmdr. Al Groves joined the test team, and between November 20, 1991, and December 13, 1991, Groves and Lang completed seven GPE flights. At this stage, there were seven pilots that had flown in the two-airplane test program and, though this was an unusually high number of pilots for a program involv-ing only two aircraft, even more pilots were soon to be added. A large number of pilots on a program of this size has its plusses and minuses. On the plus side, pilots representing all of the stakeholders (contractor and government [NASA, Navy, USAF, German Air Force]) have an opportunity to fly and comment on the airplane. This is ultimately good for the test team, which is trying to obtain data that often takes the form of pilot commentary. On the minus side, it is difficult for the pilots to maintain “currency” of recent experience in flying the test airplane.
This can be mitigated by having the pilots also fly often in similar-type aircraft.
It has been the author’s experience that often there are too few pilots rather than too many, and if one gets sick or is reassigned, a scheduling crisis must be averted.