LOS ANGELES -- In the wake of the American Airlines Flight 587 accident, manufacturers of large transports are scrutinizing the systems that affect pilots' feel of the rudder and limit its travel. Large rudder motions are the primary suspect for why the vertical tail came off the airplane last Nov. 12, but the reason for the motions is still a mystery.

Several flight control experts and pilots contacted by Aviation Week & Space Technology are critical of certain rudder systems, one saying they are "tailor-made for overcontrol." But Airbus and Boeing say existing systems are acceptable, noting that they have an excellent safety record.

The manufacturers' rudder reviews are part of a Feb. 8 NTSB safety recommendation that described how some airplanes can give full available rudder deflection with forces and deflections that are much smaller at high speed than at low speed, and noted that some pilots may not be aware of the sensitivity at high speed (AW&ST Feb. 18, p. 44). At least one major airline agrees. "Our training department's jaws dropped when they learned about that," an official said.

Factors cited by critics include:

• The rudder on jet transports is rarely used, except for engine failure and
  countering crosswinds in takeoff and landing. Many pilots have never used
  the rudder above 200 kt. and are unfamiliar with its characteristics at higher
  speeds.
• Feel systems treat the rudder pedals more as a secondary than as a primary
  control. Designers put in large breakout forces, typically in excess of 20
  lb., to prevent the pedals from moving if bumped when pilots take their feet
  on and off. Breakout is the amount of force that must be applied before the
  control moves; if it is large compared with the extra force for full deflection,
  precise control may be difficult.Rudder feel systems are an acknowledgment
  that the rudder is rarely used and pilots often have their feet off the pedals.
• Some rudder systems become more sensitive with speed. "Sensitive"
  means instantaneous yaw acceleration per incremental pound of pedal force.
  That is, the pilot will get more of a sideways kick in the cockpit when he
  steps on the pedal. Systems with this characteristic have been made by Boeing,
  Airbus and Douglas.
• Upset recovery courses teach pilots to use the rudder more than they would
  in the past. Simulators do not monitor structural loads and may not properly
  convey the g-forces in the cockpit from recovery actions. It is possible that
  pilots have done maneuvers in simulators that would have torn the fin off
  and have no idea they did so.

[[Rudder limiters prevent structural overload and are implemented in several ways. Modern Boeing aircraft use the ratio changer, which varies the pedal-to-rudder gearing, and modern Airbus aircraft use the variable stop, which has fixed gearing. The stops are actually in the tail at the input to the rudder actuator. When the ratio changer starts operating, it keeps the pedal sensitivity constant with airspeed, but the variable stop continues to become more sensitive (see graph below).]]

If pitch and roll controls were designed like some rudder systems, the airplanes would be almost unflyable, several handling qualities experts said. One cited the example of a McDonnell Douglas C-17 transport where the roll breakout forces became high because of a maintenance error. The pilot had trouble controlling the airplane, and it wobbled around the pattern to make an immediate landing.

In this case, the breakout force was about the same as the incremental force for full control, or a 1:1 ratio. On the A300-600 rudder system at 250 kt., the approximate speed of the Flight 587 accident, there is 22 lb. of breakout compared with 10 lb. of incremental force, or a 2.2:1 ratio. And this is controlled
by the feet, which are less practiced than the hands.

These designs are historically acceptable because they are almost exclusively used at low speed. Pilots have experience there and the pedal forces and deflections are high, making the breakout force less important. But at high speed, which is where the breakout force dominates and the response gets more sensitive on some systems, the rudder is essentially never used and thus has not been much
of an issue -- until the tail came off Flight 587.

PART 25 OF THE Federal Aviation Regulations governing large transport design doesn't have much to say about pedal feel, other than ". . . rudder control movements and forces must be substantially proportional to the angle of sideslip in a stable sense . . ." (section 25.177), and that the maximum temporary force required can be no more than 150 lb. (section 25.143). There is no guidance on sensitivity or breakout force. "Rudder is not a big player in design, it's more wheel and stick,"
said a test pilot who had a long career at Douglas. "They don't fine-tune rudder feel. The critical case is engine failure at liftoff and V2, and they don't expect much use after that. But you can't ignore it, you have to look at the sensitivity. Maybe they never thought the rudder would be used at high speed. That's a dangerous assumption." The A300-600 pedal breakout force is not unique; for example, most Boeing transports have 18-22 lb. pedal breakout force, and this has proven acceptable, a company official said. But high breakout followed by sensitive response can be a problem. Handling experts say this tends to give bang-bang control--full deflection one way, possibly followed by full deflection the other way as the pilot responds to the unexpectedly large airplane reaction to the first input. This prompted the "tailor-made for overcontrol" comment by one expert. The rudder motions from the Flight 587 flight data recorder are a series of full-deflection reversals before the fin came off.

"If there is an abrupt transition from breakout to control, coupled with a ponderous aircraft, the pilot could easily bang controls back and forth and wonder what's going on," a senior research pilot said. "In a low-frequency pilot-induced oscillation, the pilot feels detached from the response of the aircraft; he doesn't feel like the response is connected to his inputs. And yaw is an unfamiliar axis to control."

The effect of control deflection and force on susceptibility to pilot-induced oscillations (PIOs) is complex, and it may be that a pilot anxious to exit a situation will initially apply full control no matter how the pedal feels. However, the ensuing control reversals that make up a PIO can be reduced or encouraged by control characteristics, several experts said.

TESTS IN VARIABLE stability aircraft like Calspan's B-26 showed that large breakout forces in a responsive aircraft would be very difficult to fly. Big aircraft with slow response can tolerate more breakout.

However, a Boeing official noted that pilots achieve precise pedal control every day during the high-speed part of the takeoff roll. An FAA test pilot agreed, but did say that his first few takeoffs in a 737 had jerky pedal control until he got used to it.

Maximum pedal deflection and force shrink as the variable stop limiter system on the A300-600 takes effect above 165 kt., making breakout force become larger than control force. A ratio changer system would have the same characteristics at all speeds.

But unlike takeoffs, inflight operation of the rudder at high speed is rare and first use may be clumsy, as the FAA pilot commented. Also, the pilot knows beforehand he must use the pedal during a takeoff roll, whereas use in an upset is less premeditated.

A line check airman with a major airline noted that some pilots, when practicing roll upset recovery in the simulator, will "fight initially with aileron, but at the panic point they tend to stomp rudder, while some are better with coordinated controls. Many have their feet flat on the floor on climbout--it's very disturbing to see this."

Strong rudder use is seen in service as well as in the simulator. In 1997, an American Airlines crew used large rudder motions in recovering from a stall in an A300-600 (AW&ST Mar. 18, p. 47). Recent calculations showed that ultimate load was exceeded on the fin.

Manufacturers are concerned that large rudder use in upset recovery can result in overcontrol. "If you're using rudder to get out of a roll, by the time the wings go level you've built up such a large yaw that you go shooting out the other side," one official said.

Investigators reportedly think the rudder motions started in Flight 587 after it encountered the wake of the preceding 747. At their speed, just the yaw acceleration from full rudder would create roughly 0.5g lateral force in the cockpit, a level that might be higher than expected and cause some confusion.

It must be noted that the NTSB has not determined that the pilot moved the rudder, and is actively considering a system malfunction. In one case investigated by the NTSB, an autopilot flaw caused large rudder motions and the board has not ruled out this possibility (see p. 47).

Rudder controls have evolved from manual actuation. Aerodynamicists view rudder properties nondimensionally -- a rudder deflection of, say, 10 deg. will have a certain hinge moment coefficient that resists deflection, and will create a lift coefficient on the tail. These coefficients are multiplied by the dynamic pressure, or the square of the equivalent airspeed, to calculate the actual moments and forces.

MOST MANUAL systems have constant gearing between the pedal and rudder, and the pedal becomes stiffer as the square of the airspeed. The pedal force to hold 10 deg. of rudder may be 20 lb. at 125 kt. but grow to 80 lb. at 250 kt.

The resulting fin yaw force also grows as the square of the airspeed. To the first order, the pedal force is a measure of the fin loads over a wide range of airspeeds -- what you feel is what you get. "Forces are a clue to limit loads," the research pilot said.

Another nice feature of manual systems is that pedal sensitivity remains constant with varying airspeed. This is because yaw acceleration and pedal force grow equally with airspeed, though rate damping becomes an important effect at higher speeds.

Higher weights and speeds forced hydraulic systems upon manufacturers. Designers realized that an irreversible hydraulic actuator that did not feed hinge moments back to the pilot would result in overcontrol and structural damage. No longer does "what you feel is what you get" hold true. They devised artificial feel systems to give a semblance of manual feel, including in the yaw axis.

Boeing's 1947-vintage B-47 bomber had a "Q bellows" that mechanically sensed dynamic pressure and made the pedals stiffer with speed, like a manual rudder. Most 707 versions have feel provided by an aerodynamic tab on the rudder and a Q bellows, and have a similar manual feel.

But while newer aircraft continued to have pitch sensitivity that was roughly equivalent to manual, rudder control systems took a different evolution that strayed from the manual roots. Clever design meant the pedals were no longer needed to coordinate every turn, and they became an almost vestigial control
for occasional use at low speed.

Newer systems fall into two broad groups with a few common features. One similarity is that pedal deflection works against a fixed spring. Unlike manual feel, it does not become stiffer with speed. Another common feature is that both groups limit rudder travel at high speed to prevent overloading the fin. However, the limiters may not protect against rapid rudder motions in a sideslip (AW&ST
Jan. 21, p. 24).

One group has fixed pedal-to-rudder gearing with rudder deflection limited at high speed either by blowdown (actuators overcome by airloads) or by variable mechanical stops. As the airplane goes faster there is less pedal travel available. This is the "fixed ratio" group.

The other group has variable pedal-to-rudder gearing so that full pedal travel is always available but results in less and less rudder motion. This is the "ratio changer" group.

THE FIXED RATIO systems become increasingly sensitive because a given rudder position causes more acceleration at high speed with no rise in pedal deflection or force. "Having sensitivity get larger with speed seems opposite to what you want," the research pilot said. "Why on Earth would you do that?" Also, the limiter chops pedal travel to shorter amounts, making the breakout force an increasingly dominant player just when the pedal is becoming more sensitive.

A Boeing official noted that roll also becomes more sensitive with airspeed, but pilots have more practice in this axis than in yaw, and aeroelastic and rate damping effects mitigate aileron sensitivity more strongly than rudder.

The ratio changer group typically shifts the pedal-to-rudder gearing in proportion with dynamic pressure, making the sensitivity approximately constant with speed, like manual feel. And because full pedal throw is always needed to reach the rudder limit, the breakout force remains relatively small.

Manufacturers have had different design evolutions. Douglas has used fixed-ratio systems on all its jets, limiting motion with variable stops or blowdown. Boeing used fixed ratio with blowdown limiting on the 727 and 737, then switched to the ratio changer for the 747 and subsequent aircraft, finally enshrining it in software on the 777 fly-by-wire rudder. Airbus started with the ratio changer (called Variable Lever Arm) on the A300B2/B4, then switched to a fixed-ratio variable-stop system (called Rudder Travel Limiter) on the A310 and subsequent aircraft, including the A300-600. Similarly, the variable-stop system is now implemented in software on Airbus' first fly-by-wire rudder on the A340-600.

Airbus says the Rudder Travel Limiter (RTL) has been thoroughly flight tested "and does not lead to overcontrol." "At high speeds, the rudder deflections required to counter major lateral asymmetries are small and so are the rudder pedal inputs," the company said, noting that this is similar to pitch and roll controls where deflections also reduce with speed. "From a technical standpoint, the benefit of the RTL is that it is simpler than the Variable Lever Arm, and hence overall more reliable."