NASA has been investigating propulsion-controlled landings: Flight International tries the system.


NASA HAS COMPLETED its latest Propulsion Controlled Aircraft (PCA) programme, using a McDonnell Douglas (MDC) MD-11 test-bed with modified flight-control software. Supported by the US Federal Aviation Administration, the Department of Defense, industry and academia, the project culminated with a week of demonstration flights. Guest pilots from airlines, manufacturers, the US Navy and Air Force, the National Transportation Safety Board (NTSB) and the FAA were given the opportunity to evaluate the proof-of-concept system.

During the 30h flight-test programme, PCA performance exceeded expectations, according to NASA. PCA manual and auto-pilot-coupled instrument-landing system (ILS) approaches have been flown to touchdown, including hands-off PCA landings. Cruise-phase investigations included extended flight with flight-control hydraulic pumps off, automatic PCA recoveries following intentional upsets, two- and three-engine PCA operations, and PCA flight with an aft centre of gravity.

The PCA programme represents the culmination of research, which began with a concept-development study at NASA Dryden, California, only months after the MDC DC-10 crash at Sioux City, Iowa, in July 1989.

United Airlines Flight 232 (a DC-10-10) experienced a catastrophic engine failure while cruising over the mid-west USA. According to the NTSB, the "...separation, fragmentation, and forceful discharge of stage 1 fan rotor-assembly parts from the number-two engine led to the loss of the three hydraulic systems that powered the aircraft's flight controls". While the crew was immediately aware of the loss of the tail-mounted engine, it did not realise that it had also lost the capability to move any of the flight-control surfaces.

Shortly after the explosion, despite full-left control-column and rudder-pedal deflection, the aircraft continued its slow roll to the right. The quick-thinking crew shoved the right throttle forward, pulled the left one back and the DC-10 slowly rolled towards wings level.

Keeping the aircraft's attitude excursions small, the crew experimented with the only control it had. The crew continued to use this differential throttle technique to control the crippled aircraft in yaw and pitch, and this enabled degraded heading and vertical flight path control. The pilots also struggled with high control-column forces, believing that a small amount of normal control was available.

Gaining skill as the flight continued, the crew made its way to the Sioux Gateway Airport, to attempt what it would later learn was a thrust-control-only landing. A dramatic video showed the DC-10 in an apparent good position to land, just a few metres above the runway, when it again began to roll to the right.

Later investigation revealed that the number-two engine's disintegration had created a large hole on the right side of the aircraft's vertical tail. The effect was the same as that of right-rudder deflection - a yawing moment to the right. Because of the aircraft's positive dihedral effect, the resulting left sideslip caused the DC-10 to roll right. When both throttles were retarded for landing, the asymmetric-thrust compensating for the yawing moment was removed. The resultant right roll caused the right wing to contact the ground, triggering a crash, which destroyed the aircraft. Of the 296 people on board, 185 survived.

NASA cites nine other incidents involving the loss or partial loss of primary flight-control systems resulting in the loss of more than 1,000 lives. The aircraft, ranging from single-seat fighters to wide-body airliners, all had most or all of their engines operating normally after the flight-control failures.

Following the Sioux City mishap, the NTSB recommended research into back-up flight-control systems, using an alternative source of motive power separate from that used for the conventional controls.

NASA's propulsion-control investigations have been under way for years using a variety of aircraft, ranging from light twins to a Boeing 747, in simulation, flight test, or both. The agency has concluded that every aircraft evaluated has adequate propulsion-control capability for extended cruise flight, and that every aircraft is very difficult to land using manual manipulation of the throttles for control. It was determined that automatic thrust-control using feedback from the aircraft is necessary for repeatable, successful, landings.

NASA/MDC F-15 PCA test flights were perhaps the most visible previous propulsion-control efforts. During that programme, NASA Dryden research test-pilot Gordon Fullerton found control using manual throttle movements to be very difficult: safely controlling the F-15 in cruise flight required practice; safely landing the F-15 using manual throttles was deemed unlikely. In April 1993, however, the aircraft was landed safely with its PCA system engaged. The F-15 PCA project proved that the concept was viable as a back-up means of control.


The MD-11 PCA test-bed has standard hardware and the only software modifications are to the number-one flight-control computer and the engine controls. Honeywell provided the modified software. The standard auto flight-system heading-track/knob and vertical speed/flight path-angle thumb-wheel are used for PCA control.

Other functions accessed through the flight-control panel, located on the glare shield, remains available, but were not used in the demonstration. The MD-11's normal longitudinal stability-augmentation and yaw-damper systems were turned off.

Operation of the PCA system is straightforward and intuitive. Depressing the auto-flight button engages the system, and both pilots are notified on their primary-flight displays (PFD). The aircraft is then flown, on whatever track and flight path angle the pilot has selected. The knob and thumb-wheel remain active, allowing changes as fine as 1° in track and 0.1° in vertical flight path angle.

Although the track set is displayed as the knob is turned, the PCA system ignores the setting until the knob is pulled, enabling pre-setting of the next track. Pushing the track knob at any time commands the aircraft to a zero bank angle. Bank angle, is software limited to 20° although, a smaller bank angle limit, can be selected by the pilot. Optionally, vertical speed can be commanded in lieu of flight-path angle. Bank angle can also be selected in lieu of track by toggling the heading/track knob to the heading position.

With selection of the approach/land mode, an "armed" notification is displayed on the PFDs. Automatic localiser and glide-slope coupling occurs when the aircraft receives reliable ILS signals. In this mode, bank angle is limited to 10°. Once coupled, the PCA system intercepts the inbound course and automatically flies the approach. Current programming commands a 1-dot-low glide-slope (0.25° flatter than that of published glide-slope).

This results in a higher thrust approach, reducing engine spool-up time, when additional thrust is required and reduces, the flight-path-angle change needed for the landing. Both the track knob and flight path-angle thumb-wheel remains active, but any adjustment disengages the approach/land mode and causes reversion to the manual PCA system. Re-engagement is accomplished by merely pressing the approach/land button again and allowing the system to self-couple. It is possible for the system to couple to the localiser or glide-slope signal alone if a reliable remaining signal is not received. The pilot is kept aware of the system's automatic approach status by the PFDs.

The PCA system performs a two-stage landing flare when in the approach/land mode. At 125ft (40m) above ground level (AGL), the flight-path angle is automatically commanded to 1.5°, at 30ft AGL, another adjustment is commanded, to 0.75°, which is then maintained until touchdown.

With a 165kt (300km/h) landing-approach speed, this flight path angle should yield a touchdown rate of descent of about 215ft/min (1.1m/s). The pilot is expected to take control at this point, but if no action is taken, the PCA system disengages upon nose wheel spin-up.

An automatic go-around can be initiated by depressing a button on the number-two throttle stalk. This action commands a 3° climb straight ahead. Several PCA system-disengagement options are available. The yoke-mounted auto-pilot disengage switch, the number-one throttle-mounted disengage button, or a manual number-one or number-three throttle-angle change of more than 2° completely disengages the PCA system.


The modified software in the number-one flight-control computer uses pilot input, aircraft track, bank angle, flight path, roll rate and pitch rate, to determine its output to the engines. Controlling each Pratt & Whitney PW4462 engine is a full-authority digital engine-control unit. These are production units using modified software.

The control principle in the PCA system is the same as that used by Capt Haynes' crew on the ill-fated UA232 flight. The number-one and -three engines are located outboard and below the MD-11's centre of gravity. Symmetric changes in thrust, cause a pitching moment, asymmetric changes cause a yawing moment, which leads to roll. What makes the PCA system so user-friendly is that the pilot commands the result, and the computer handles the details. Within the limits of the control authority provided by the moment generated by thrust, this arrangement should automatically compensate for failures, which result in deflected control surfaces.

For the approach and landing flight-tests, the number-two engine was kept at a nominal setting near idle. Since the same two engines are used for pitch and roll (through yaw and the dihedral effect) control, precautions have been incorporated in the software, to ensure that adequate thrust is available in both axes regardless of the demands in an individual axis. A priority logic slightly favours roll over pitch control.

Extensive PCA simulation was performed in the MD-11 simulator at Douglas Aircraft in Long Beach, California. The simulation uses the same PCA-modified software as the aircraft - only some of the control gains are different. These differences arise from the continual refinement in the aircraft, and are essentially transparent to the new pilot.

The fixed-base, limited-visuals, simulator provides a good appreciation of the PCA system before experiencing it in the aircraft. Controls and displays are identical, and the simulation fidelity is laudable.

Wind and turbulence models allowed operation of the PCA system to be experienced under conditions unavailable, and undesirable, in the aircraft. The simulated system did a marvellous job of eliminating gust upsets in moderate turbulence while tracking the approach. The simulator was also graphically successful in demonstrating the difficulty of making an approach using manual-throttle control.


To keep the approach speed to a reasonable 165kt, the leading-edge slats were extended and the flaps were set at 28¡. While this would not be possible following a total hydraulic failure (unless the surfaces were already deployed at the time of the failure), the programme personnel thought it reasonable for this proof-of-concept demonstration. The landing gear was lowered and raised with the PCA system engaged.

Engaging and disengaging the system resulted in no transient aircraft motions. Pitch and roll rates experienced with the PCA system engaged were comparable to those routinely used by the airlines. This is true during pilot selection of track and flight path angle, as well as during automatic approaches. A small lateral acceleration was felt as the system commanded differential thrust to yaw the aircraft to establish a roll rate.

Noticeable, but not uncomfortable, this sideways pulse was only evident with roll initiation, not with termination. The only other telltale sign of PCA system activity might have been, the increased engine activity audible in the sparse, test-system-equipped, cabin, but not heard in the cockpit.

Any guest pilots who accepted MDC safety pilot John Miller's invitation to fly the MD-11 using manual throttles readily acknowledged the inflation of basic airmanship to a full-concentration task. No approaches were flown with manual throttles - and no pilot asked to fly one, either.

Pilot-selected changes to aircraft track (or bank angle) and flight path angle using the PCA system were all smooth and comfortable. Inbound course intercepts were lead-computed and, for reasonable offsets, resulted in no overshoots. Glide-slope captures were similarly smooth and effective.

For safety considerations, approaches were made to a simulated runway 100ft above the actual runway at Edwards AFB. Each coupled approach was terminated at the simulated-runway elevation. With no pilot action, the PCA flight-path became shallower at 225ft AGL, then dipped again a few seconds later at 130ft AGL. Both flight-path changes were easily recognised, providing immediate reassurance to the pilot that the PCA system was functioning during the most critical landing phase. Go-arounds were initiated with the single number-two throttle button and were performed automatically as advertised.

Before an impressive PCA automatic landing, Fullerton demonstrated his manual-throttle technique. As the project pilot for both the F-15 and MD-11 PCA programmes, he was skilled. Despite his constant, small, throttle adjustments, excursions were evident, again verifying the need for a feedback-augmented control system.

With a 5kt crosswind from the right, Fullerton flew the 2.5° ILS glide slope, to Edwards runway 22 using the PCA system manually. Glide-slope was established with the thumb wheel after two corrections of less than one dot (0.25°), and no further flight-path-angle inputs were required. He continued to refine the PCA system's track all the way to the automatic flare (implemented on the PCA even in manual mode) in 1° increments with the track knob every 10s or so. Using this technique, he was able to keep the aircraft's bank angle to 3° or less. By the time Fullerton selected the automatic go-around, the MD-11 was in a good position for landing.

For the automatic PCA landing, the artificial 100ft elevation buffer, was removed by changing the two flare altitudes stored in the number-one flight-control computer. A coupled ILS approach was flown. Just as in the go-around cases, the two-stage approach was flare executed, as it should have been. The modified software does not account for ground effect, so the PCA-controlled aircraft floated to a touchdown further down the runway than might have been expected, but essentially on the centreline. The landing, only the fourth using automatic propulsive control in the MD-11, was more firm than had been anticipated. As project personnel point out, however, a flight-control-crippled aircraft should be landed safely and quickly. Data traces indicated a touchdown vertical speed of approximately 300-360ft/min (1.5-1.8m/s).


Fullerton disengaged the PCA system on touchdown and manually deployed the thrust reversers. The reversers are hydraulically driven on the P&W PW4462 engines, but the MD-11 can also be equipped with General Electric CF6-80C2D1F power plants. The GE engines have pneumatic reversers, which should be available regardless of the aircraft's hydraulic-system status. Runway controllability issues resulting from hydraulic failures, including loss of reversers, spoilers, wheel brakes, nose wheel steering, and were beyond the scope of this programme.

The MD-11 PCA programme ends with these demonstration flights. It appears that the concept of a propulsion-based back-up flight-control system is viable. Although this research effort was aimed at demonstrating sufficient propulsion control of a damaged aircraft's flight path to land safely, there are implications of other propulsion-control benefits. Considering the success of the MD-11 PCA tests with software changes only, the potential for back-up propulsion-based flight-controls in new designs looks promising.


Source: Flight International