In the low-cost mode, the flight-test planning targeted a fly rate somewhere between the then-recent YF-16 and X-29 flight-test programs. Initially, the YF-16 program achieved an average fly rate of about 0.8 flights per day, with 100 flights in the first 125 days after first flight. The X-29 program achieved about 0.075 flights per day, with 30 flights accomplished in the first 400 days after first flight. It is important to note that the YF-16 was a prototype-aircraft test program, whereas the X-29 was an experimental, or research, flight-test program. It would be expected that a prototype effort in which the airplane is closer to a production flight article would have a higher flight rate than an experimental aircraft program. At this stage, the planned fly rate for the X-31 was about 0.30 flights per day, or 100 flights in the first 325 days after first flight. Needless to say, this was a very aggressive flight-test schedule for an entirely new experimental research aircraft design. Flight-test support was to be provided through the use of a Rockwell flight-test control room at Palmdale and the Rockwell resources at Palmdale and El Segundo for data reduction. As mentioned in the previous chapter, a flight simulator using a 24-foot dome was
Into the Air: Initial Flight Testing
available at the Rockwell Downey facility to support flight testing. The two X-31 aircraft were equipped with identical flight-test instrumentation suites that could provide real-time and near-real-time data to the control room.2
At this juncture, the test team planned to clear the aircraft’s conventional flight envelope and then to move on to do an incremental expansion of the post-stall envelope. The post-stall envelope would include flight up to 70° angle of attack, with rolling maneuvers at 70° angle of attack and “dynamic entries”
into the high-AOA regime. Dynamic entries would be aggressive, mimicking the urgency of air combat, with rapid-g-onset maneuvers to the 70°AOA limit.
The flight-test organization was a combined test team composed of a Rockwell team leader and an MBB deputy team leader. Members of the flight-test team came from both contractors as well as from the U.S. and German governments.
Following the establishment of an adequate post-stall flight envelope, the test program was to move to NAS Patuxent River for the portion of flight test that would involve the evaluation of tactical effectiveness during close-in air-to-air combat. The overall objectives of the flight tests following envelope clearance were to demonstrate and measure Enhanced Fighter Maneuverability perfor-mance and to accomplish an evaluation of the X-31 during close-in air-to-air combat. At this time, first flight was scheduled for February 1990. There were to be about 600 flights from both locations (Palmdale and Patuxent River), and the entire program was to be completed in 2 years.
Initial ground testing illuminated a number of technical challenges that needed to be solved prior to the start of taxi testing. These challenges were by no means unusual for the initial checkout of a brand-new aircraft design;
however, they did require troubleshooting and the development of appropriate fixes before first flight could occur. The first issue was encountered when the flight control system was engaged for the first time. The flight control surfaces entered a 5° deflection at 4 hertz (Hz). Ground vibration test data showed a 4 Hz “on-gear” pitching mode, and analysis indicated that this mode was coupling with the flight control system to cause the oscillation. The fix to this problem was to stiffen the rate-gyro mounting platform (perhaps surprisingly, given the “high-tech” nature of the program, simple ½-inch marine plywood was used). The oscillation disappeared.
During initial ground testing of the flight control system, the flight control surface motion was jerky, especially with large control surface movements. An interesting solution was devised for this problem. A smoothing filter was used to quiet the control surface motion, but this produced a bandwidth issue with the flight control surfaces, so a digital lead filter was installed upstream of a analog converter, followed by an analog lag filter after the digital-to-analog converter. This combined filter fix had no effect on the frequency response of the flight control system, but the jerky surface motion was at least “smoothed.”
Flying Beyond the Stall
A potential flight safety issue arose regarding flight control redundancy. The primary source of angle of attack and angle of sideslip for the flight control system was a single-ring laser gyro. The flight control system’s redundancy-management system depended upon an internal flag in the inertial navigation unit (INU) to cause the system to switch to the alternate sensing of angle of attack and angle of sideslip that came from the aircraft noseboom. The prob-lem was that this “flag setting” took as long as 200 milliseconds after an INU failure, whereas flight-hardware-in-the-loop simulation testing had shown that the aircraft had a likelihood of controlled-flight departure after only 20 mil-liseconds of bad data were input to the flight control computers. The fix here was to compare INU body rates and accelerations to flight control system sensors and to reject the INU data in the first instance that it was out of toler-ance. With this fix to the INU’s redundancy management, the airplane was ready to start its taxi tests.
Taxi testing commenced on June 20, 1990, with a low-speed test. During the test, some flight control computer anomalies arose, as well as a rudder problem. Two medium-speed taxi tests to 80 knots and 85 knots followed, accompanied with steering evaluations. These also uncovered some flight con-trol computer issues and a problem with the “weight-on-wheels” logic—an important input to the aircraft’s computer systems. There was objectionable directional sensitivity both with and without nosewheel steering. At about 70 knots, the aircraft started a zigzag oscillation that was deemed unacceptable by the pilot. Both nosewheel steering and rudder pedal response were felt to be too sensitive. Since the X-31 was using an F-16 nosewheel steering control box, it was modified to produce only 10° of deflection, and its sensitivity was also reduced. This became the new “normal” mode. Pilots could still select the original F-16 nosewheel steering mode (±30°) for hard turns. Rudder sensitivity was fixed by lowering the effective rudder pedal to rudder-surface deflection gain at speeds below 110 knots. This seemed to be adequate during subsequent medium- and high-speed taxi tests. The taxi tests progressed to higher speeds with two medium-speed taxi-steering evaluations to 105 knots, including a taxi in the R3 flight control reversion mode to 70 knots. These also identified an R1 reversion-mode problem because the X-axis acceleration-sensing threshold was set too tightly and there was a leading-edge flap failure.
Often in the development of a new aircraft, the test team will avail itself of the services of one of Calspan Corporation’s variable-stability Learjets. These are experimental aircraft in which the flight control system has been modi-fied to be programmable to simulate the flight dynamics of the aircraft under development—in this case, the X-31. The test pilots can then fly the variable-stability Learjet and actually observe the flight dynamics of what they can expect to see in the test aircraft. This is a particularly useful tool in preparation
Into the Air: Initial Flight Testing
for a first flight. Rockwell Chief Test Pilot Norman K. “Ken” Dyson strongly advocated for this resource, but it was not made available due to funding.
Dyson did spend many hours in the Rockwell flight simulation dome in prepa-ration for the first flight.3 Finally, a high-speed taxi test was accomplished on October 3, 1990, that included a drag chute deployment. There were some minor problems with the flight-test instrumentation’s data link. This was the final flight readiness test in preparation for first flight.4
The first flight of Ship 1 (U.S. Navy [USN] Bureau Number 164584) took place on October 11, 1990, with Ken Dyson at the controls. Dyson was a retired USAF test pilot who had participated in the flight testing of the Lockheed XST Have Blue project, an early stealth demonstrator aircraft and progenitor of the F-117 stealth fighter and many other noteworthy airplanes, including his “low-g fighter,” the B-1B. The first flight lasted 38 minutes and attained a little over 300 knots calibrated airspeed and reached an altitude of 10,000 feet mean sea level (MSL). Dyson reported that the aircraft’s flying qualities were excellent and matched ground simulation predictions. He cycled the landing gear, and flying qualities were evaluated in both power-approach and cruise configurations. Subsystem performance matched preflight expec-tations, and after landing, he had only minor maintenance discrepancies to report.5 The X-31 program was now a flight-testing reality. It should be noted that the X-31 was flown without the thrust-vectoring paddles on this first flight and for several of the initial flights. This was done because an inadvertent hard-over of a thrust-vectoring vane on takeoff could cause loss of aircraft control
First landing of the X-31, followed by a T-38 chase airplane. (Rockwell)
Flying Beyond the Stall
and because thrust-vectoring-vane ground clearance during takeoff rotation was critical. Once the thrust-vectoring redundancy management was validated and takeoff and landing tail clearance was observed, the thrust-vectoring vanes were installed.
On November 6, 1990, MBB Chief Test Pilot Dietrich Seeck flew Ship 1 for the first time, followed on November 15, 1990, by Fred Knox, a former Navy fighter pilot and test pilot who had joined Rockwell as an engineering test pilot.
In the meantime, Ship 2 (USN Bureau Number 164585) was progressing toward first flight with its series of ground tests and taxi tests. A disturbing event occurred during the first taxi test of Ship 2. As the aircraft accelerated to a relatively high speed, a divergent oscillation occurred in pitch. Two of the flight control system’s rate gyros had been wired backward. It was a surprising anomaly, and one that should not have occurred. Previously, other high-performance aircraft had been lost because of installation errors involving control gyros, including a Lockheed A-12 Blackbird and, years later, the first production F-117A stealth fighter. The aircraft was slowed to a stop with no untoward effects. On January 19, 1991, Ship 2 took to the air with Dietrich Seeck at the controls. The very first international X-plane was now an airborne accomplishment!