Everyone knows flight simulators are used to train pilots. That is the core function of these $14- to $20-million machines. Less well known is that simulators play other safety-enhancing roles. United Airlines, for example, which operates 22 full flight simulators–covering its fleet of Boeing 737-300s, 747-400s, 757s, 767s, 777s and Airbus 320s–has employed the equipment for a wide variety of purposes, including:
Researching proposed procedures,
Developing training for recovery from windshear conditions,
Testing avionics installations, and
Refreshing and educating non-pilots, such as emergency responders and air traffic controllers.
United Airlines’ emergency responders periodically use the carrier’s Denver training facility to simulate an aircraft emergency, such as an unruly passenger or critical aircraft failure, in order to practice skills and validate procedures for handling an actual case. These teams, specific to different aircraft fleets, include engineers, pilots, flight operations personnel and other specialists. A 777 simulator aircrew, for example, would work through an emergency scenario. Special equipment and programming are used to insert the simulator into the company’s operational communications systems during the exercise, which tests preparedness and identifies any weak points in the emergency response system. United has supported these sessions with B737, 757, 767, 777 and A320 aircraft simulators in recent years.
Simulator exercises also can help air traffic controllers. In one case United’s Simulator Engineering Group enabled the Oakland, Calif., air traffic control (ATC) center to iron out some issues with a future air navigation system (FANS) upgrade. Oakland Center was having problems with transfer-of-control procedures, as aircraft crossed flight information region (FIR) boundaries. United engineers at the carrier’s flight center were able to duplicate the problem in a recently updated B747-400 simulator and demonstrated to Federal Aviation Administration (FAA) technical personnel what was happening and how to fix it, says Steve Ferro, the airline’s manager of B737 and B747 training devices.
Turbulence Models and SOIA
United has worked with the U.S. government for years to help develop procedures, some of which have come into use. An ongoing project with NASA Langley involves the automatic transmission of turbulence warnings through the airborne communications addressing and reporting system (ACARS). Starting with the B737, the carrier is helping NASA to develop mathematical turbulence models, using the simulator "to measure aircraft response to sudden longitudinal accelerations," explains Bob Ireland, director of training devices and facilities. The idea is for a predetermined pattern of accelerations to trigger an automatic warning report. "We’re showing what the airplane response is to specific gusts of wind at magnitudes considered to be representative of turbulence," he says.
United also has collected data for FAA on the use of simultaneous offset instrument approach (SOIA) procedures at San Francisco. Today, without SOIA, in clear weather, two planes are allowed to land simultaneously on parallel runways, providing the pilots maintain separation visually. In low-visibility conditions, airplanes are not allowed to perform such parallel approaches.
SOIA, however, will allow these capacity-enhancing parallel landings to take place down to lower weather minimums. In the procedure, both airplanes have a localizer and a glideslope; one pilot flies a laterally displaced approach (LDA) procedure, while the other flies a straight-in approach. The pilot flying LDA has the responsibility to see the other aircraft visually in order to continue the approach, says David Jones, United’s manager of capacity enhancement. (LDA involves a localizer displaced angularly, such that it converges with the runway heading at a location/altitude where visual separation can be established.)
In the simulator, United tested SOIA human factors issues. The carrier gathered data in areas such as go-around, localizer overshoot and the closest point of approach to the other aircraft. Researchers, for example, looked at "how much can you offset the localizer and not make too heavy a workload for the pilot?" Ferro says.
For its SOIA studies, United also brought in air traffic controllers from San Francisco to give them a sense of how the procedure would work in an airplane. "The simulator gets all the players together," he says.
Items, such as the exact point at which the pilot makes the turn to the final runway heading and the angle of approach, also are important in SOIA procedures. There is concern about overshooting the runway centerline or coming in at too steep an angle, making it harder to turn, Ireland adds. "We want to be sure it’s an absolutely safe procedure."
United and the Air Line Pilots Association (ALPA) used the carrier’s simulator to prove the feasibility of the concept, Jones says. The FAA Flight Standards organization also used the simulator data as input to its study that determined the flyability of SOIA, considering wake turbulence and collision risk, he adds. Flight Standards approved a "national SOIA order" in August 2002, laying down the requirements for SOIA, and the air traffic side of the house has begun developing national air traffic control (ATC) procedures. United hopes to be able to fly a SOIA approach down to a 2,100-foot ceiling at San Francisco by May of 2003.
United and other airlines also gathered simulator data for FAA on the use of land and hold short operations (LAHSO) at Chicago O’Hare. LAHSO allows two landings to occur on crossing runways or a landing on one runway and a takeoff on the other, increasing airport capacity. The simulator work was necessitated by an FAA rule in 2000, which restated LAHSO requirements for certain runway configurations. As a result of the order, LAHSO operations on the 14R/27L configuration at O’Hare were effectively suspended, pending the development and modeling of rejected landing procedures there. A major concern was the possibility of a rejected, or "balked," landing on one intersecting runway while a takeoff was proceeding on the other runway. (A rejected landing occurs when a landing is aborted between the runway threshold and the first 3,000 feet of landing.)
A pilot using the LAHSO procedure agrees to stop before reaching the intersecting runway. Without LAHSO, an aircraft is not allowed to approach until all crossing traffic is out of the way. United collected LAHSO data using its B737, B777 and A320 simulators in the 2000 timeframe. Pilot reaction times and techniques in executing landings and go-arounds were fed into a database for analysis. From the data, United determined a range of reaction times and pilot performance data, and calculated airplane performance factors, such as climb speed. Engineers, for example, looked at the time from deciding to go around to initiating the climb and the time to execute the turn to the new heading.
FAA employed an offline simulation, meanwhile, to test all of the different combinations and permutations of reaction times and performance data in order to determine whether the LAHSO procedure would be safe. The government’s offline simulation "gives [FAA] a lot more simulator runs to check for the one odd case that might be unsafe," Ferro explains. The rejected landing procedure that had been developed for the 14R/27L configuration at O’Hare, however, is not usable at this time. According to FAA, the procedure has not been shown to meet the agency’s safety requirements. United has asked FAA to review the criteria the agency is using to model the approach, and "FAA is reviewing every criteria affecting LAHSO," an airline official says. (FAA uses both the airline’s simulator results and its own criteria as inputs.)
Perhaps the biggest, airline-wide simulation success was the development of windshear training. Before 1983 simulators were programmed to replicate what happened in nature. Depending on the pilot’s reaction to the "windsheer" episode, the simulated effects would differ. Thus no two pilots’ experiences in the simulator were the same and procedures could not be standardized. Eventually, that approach was changed: the simulated weather condition’s effects–such as airspeed loss–were the same, no matter what the pilots did. The revised training approach allowed standard procedures to be verified and reinforced with success.
United first joined forces with Boeing, the then-Douglas Aircraft, Lockheed and the government’s National Center for Atmospheric Research (NCAR). "We looked at what was causing accidents," Ireland says. "We determined that pilots had been trained instinctively to maintain airspeed, so when a windshear robbed them of some, the first thing they’d do is push the nose down to catch some airspeed. You don’t want to do that close to the ground."
United had pilots in the simulator pull up, close to the stall warning limits of the airplane, without stalling the airplane. "We put together wind models that would be the same for every pilot," Ireland says.
At the same time, the training stressed the need to avoid windshear in the first place. "We went to great lengths to set up one occasionally used situation in the aircraft simulator with enough clues that the pilots should divert from their intended approach. If they didn’t avoid [the simulated event], it was not going to be recoverable."
The training was put in place in the summer of 1985 and it’s still in use. From 1984 to 1986 FAA had Boeing under contract to produce a training program that other airlines could adopt. United, NCAR and the original equipment manufacturers were on the Boeing team. Thanks to the training, the domestic windshear accident rate, industry-wide, went from about one per 16 months, before 1985, to hardly any now. (The most recent domestic incident occurred about 10 years ago.)
Another early project with FAA involved "in-trail climb." Aircraft flying over large bodies of water used to be restricted in altitude by the position of the airplanes in front of them. The concern arose, however, that if the aircraft in front developed a problem–say, an engine failure–and was forced to slow down, it could create an unsafe situation.
United programmed a traffic alert collision avoidance system (TCAS) target in the simulator to see how close the in-trail airplane would come to the airplane in front in several in-trail scenarios. United was able to determine that, at a certain distance, pilots would be able to execute in-trail climb. This testing process helped FAA to determine to put in place the procedure that is used today, says Ferro.
A relatively recent application for flight simulators is new avionics installation testing. United performed a test installation of Honeywell’s Pegasus flight management system (FMS) in one of the carrier’s B767 simulators, Ferro recalls. This was not to see how it "flew," but to test the installation before taking an airplane out of service for FMS updating. Engineers, for example, checked whether the FMS communicated correctly with other avionics boxes over the ARINC data buses.
Recently, the carrier also has used a simulator to help fine-tune the placement of two cabin surveillance monitors in an actual B747-400 cockpit, in line with an FAA grant to evaluate security-enhancing technology. The airline selected Rockwell Collins to install a four-camera version of Collins’ Video Intelligence System. As of August, Collins expected the installation to be complete by the end of the month, supplemental type certification approval to occur shortly afterward and system activation in early October. The six-month trial will look at both wired and wireless networks for video image transmission.
United used the simulator to identify the following human factors issues with the surveillance monitors:
Did the proposed location of the monitors block the pilot’s view?
Did it cause excessive head movement or distract from the pilot’s scan?
Did it interfere with arm movement?
Could the monitors be clearly seen during turbulence?
United expects to conduct further simulator studies on issues such as how easy it is to make selections on the monitors.
At the request of the National Transportation Safety Board (NTSB), United Airlines also has provided simulator time to help investigators better understand aircraft incident scenarios. The carrier, for example, replicated the conditions leading to a 1989 accident, based on information from the flight data recorder. An engine explosion on a DC-10 resulted in loss of all hydraulics and moved the aircraft into an "asymmetric" flying condition. In a rare display of airmanship, the crew, controlling the aircraft with throttles only, crash landed at Sioux City, Iowa. The hull broke into several pieces but many passengers were able to walk away.
United was able to estimate the aerodynamic effects of having the air ducted through the hole in the side of the No. 2 engine and programmed the simulator with these changes. None of the pilots who tried to fly the scenario in the simulator was successful at landing the simulator on the runway, says Steve Ferro, United’s manager of B737 and B747 training devices.
One pilot thought of unbalancing the fuel, which reduced the asymmetric condition but only slightly increased the controllability of the aircraft. McDonnell Douglas designed a modification to the hydraulics that would prevent the loss of all flight controls in a similar situation. United was able to simulate this modification in a flight simulator and demonstrate that, under similar conditions, the change probably would have enabled the aircraft to be landed safely.
Looking at Upsets
United was one of the first airlines to standardize training for recovery from unusual attitudes, says Steve Ferro, manager of B737 and B747 training devices. Developed in 1995, the carrier’s Advanced Maneuvers training program recognized the need to fill in the gaps for the growing number of pilots with no previous military service and associated recovery training. Although rare in the real world, these events do happen. Roll upset can be caused by rudder problems or wake turbulence, for example; and excessive pitch-up attitude can be triggered by problems with the autopilot, stabilizer or other control surfaces.
In the carrier’s roll upset maneuver, the simulated aircraft is rolled to 135 degrees of bank and then released. (Although the simulator can’t reproduce the exact aircraft position or the physical sensation of an upset, all indications, both visual and through instrumentation, give the pilot a compelling impression of a roll upset.)
The pilot’s objective is to roll the aircraft back to wings level and gain control of altitude, without exceeding the operating load factor limit. The load factor also is displayed on the DME indicator to inform the pilot of how many Gs the aircraft would be experiencing in real life. If the simulator exceeds the operational G limit for the airplane–typically 2.5 Gs flaps up and 2 Gs, flaps down–a chime is rung in the cockpit. The maneuver is run several times to demonstrate the airplane’s roll capabilities, using the control wheel only and then using a combination of wheel and smooth rudder input.