Though the FAA in early November published a final rule requiring U.S. airline pilots to experience and recover from full stalls in the simulator, key details needed to put the training into practice within five years are as yet unfinished and the topic of continuing debate.

Simulator manufacturers and airframers are eagerly awaiting a better definition of the required simulator and training upgrades, expected in early 2014 when the FAA releases an advisory circular and the first draft of proposed changes to its Part 60 flight training device rules. Part 60 defines the nuts and bolts of how simulator manufacturers will have to enhance their products to implement the training rule.

“We don't want to go it alone,” says Lou Nemeth, chief safety officer for global simulator and training provider CAE. “We participated in industry working groups that helped shape these things. We're all for this in the name of cost-effective safety enhancements. Safety is paramount. But [the upgrade] has to be meaningful and cost-effective.”

Chief among the missing ingredients are stall models or guidance for building them for an estimated 50 or more aircraft configurations and variants in use for Part 121 commercial air service in the U.S., and the minimum fidelity necessary for each model to respond closely enough to the real airframe to prevent “negative” training. To have the capabilities in place within five years, the deadline for the new rule, approximately 300 simulators will have to be upgraded at an estimated cost of $30 million, the FAA says. Developing the models could cost another $20 million.

Jeff Schroeder, the FAA's chief scientific and technical adviser for flight simulation systems, says the Part 60 changes will be based on guidance that “several committees of experts comprised of airplane manufacturers, simulator manufacturers, pilot unions, regulators and research organizations” have developed over the past three years.

“Key recommendations, such as 'How good does the simulator have to be to teach full stall recoveries?' have been tested in high-fidelity simulators with test pilots who have stalled the actual airplane and with groups of airline pilots without transport aircraft stall experience,” he says. “We also have several research projects underway on topics such as validating the learning objectives, developing a consistent method for creating stall models, improving the simulator cueing and evaluating potential benefits of extreme maneuvers.”

Implicit in the new rules, mandated by Congress following the 2009 Colgan Air Bombardier Q400 crash near Buffalo, N.Y., is that today's full-motion simulators are not representative of the real aircraft at pitch angles and significant sideslip angles beyond the stall, which is approximately an 18-deg. angle of attack (AOA) for a large single-aisle transport. Nor are the simulators required to be representative in that range.

Under Part 60, the devices must accurately reflect flight-test data up to the stick-shaker stall warning and aerodynamic buffet in a wings-level attitude, which occurs at an AOA of about 12 deg. for a Boeing 757-type aircraft, approximately 6 deg. lower than the stall AOA. Simulators typically use flight-test data from the aircraft manufacturer to reflect aircraft handling up to and slightly beyond the stall AOA, but for a narrow sideslip envelope of 1-2 deg. Pilots are trained to recover from an impending stall when the stick shaker activates.

In the Colgan Air crash, the pilot pulled the control column back rather than reducing the AOA after the stick-shaker and stick-pusher warnings went off, keeping the aircraft in a deep stall state with large sideslip angles until it hit the ground. While pilots will ideally never be in the position of having an AOA higher than the stick-shaker activation, upset conditions or improper inputs can send an aircraft into the stall realm or beyond, a state pilots have never experienced and do not know how to recover from.

But can simulators accurately represent that state? The FAA is confident they can, though not with current stall models that extrapolate flight-test data beyond the certified envelope into the realm of non-linear aerodynamics.

John Foster, senior research engineer at NASA Langley Research Center, Va., says an actual aircraft in the non-linear zone has reduced stability in all three axes, reduced control effectiveness and wing-drop tendencies in random directions due to aerodynamic asymmetries.

NASA and others have been modeling deep stall behavior based on flight and wind tunnel tests since the 1960s, when military fighter pilots were having issues with stalls and spins. “Most fixed-based military simulators have stall and post-stall behavior included, more than anything in the civil industry,” says Jack Ralston, president of Bihrle Applied Research. “It has demonstrably saved lives.”

NASA's most recent campaign to model commercial aircraft stall behavior began in 2000, when the agency teamed with Boeing to develop an enhanced upset recovery (EUR) aircraft simulation for a swept-wing transport with under-wing engines and conventional tail. Along with analyzing Boeing's flight-test data, NASA researched the post-stall regime using computational fluid dynamics (CFD) models as well as scale models of a large single-aisle, twin-engine transport, similar to the Boeing 757, in the wind tunnel and vertical spin tunnel.

Augmenting the wind tunnel data was a free-flight test of a dynamically scaled, remote-controlled, 54-lb., jet-powered model called the T2. Foster says the T2 validated stall and departure data from the wind-tunnel models, with roll-off and nose-drops matching well, but asymmetry characteristics differed in that the T2 tended to drop the same wing in every roll-off.

A program goal for EUR was to demonstrate that simulators could provide positive training for stalls, an objective NASA began testing with pilot evaluations in 2005. Since that time, more than 100 military and civilian pilots, including test pilots with stall experience, have flown the simulator, noting that the extended envelope's stability, roll-off and control responses were representative of stall behavior.

Aviation Week sampled the normal and extended envelope models in the Langley simulator on Nov. 15, noting that the aircraft with the normal model was sluggish but controllable, predictable and simple to recover, even with the yoke fully aft in a full stall. With the extended model, however, the aircraft was unpredictable at post-stall AOAs, with rapid left and right roll-offs that diverged in amplitude, out of phase with roll inputs through the ailerons. Recovering from the stall required a steep nose-down push to what appeared to be an almost vertical attitude.

As little flight data exists in the deep stall regime with large slideslip angles, the FAA is tapping pilots with experience in stalling large transport aircraft to evaluate the extended envelopes.

In February, the agency gathered nine pilots from the International Civil Aviation Organization's International Committee for Aviation Training in Extended Envelopes (Icatee) to fly a Lockheed Martin Boeing 737-800 simulator in Miami, modified for post-stall behavior with a model built by Birhle Applied Research. Icatee has proposed that representative stall models could apply across similar models; for example, a single model could stand in for all large, single-aisle transport aircraft with under-wing engines and conventional tails.

Based on that trial, the agency in July brought eight pilots experienced in stalls and 47 airline pilots with no stall experience to evaluate its 737-800 simulator in Oklahoma City. Each pilot tested four stall models—the original 737-800, a 737-800 with a rolling moment at the stall, a Boeing 737-800 extended model and the Birhle model. The test results have not yet been published.

Ralston says opinions of pilots in the Icatee trials ranged from “so-so to positive,” but he says most of them had not stalled a large transport aircraft before. CAE's Nemeth, who flew the simulator in Miami, says the extended envelope was “very similar to roll models already in [simulators],” with the exception of a more significant roll-off. “When you lower the nose, the effectiveness of the ailerons returns and the aircraft becomes controllable again,” he says.

While CAE has not developed any extended models internally, the company does have an upset-recovery and -prevention training program with partner Aviation Performance Solutions that includes in-aircraft recoveries along with simulator training.

Simulator maker L-3 Link U.K. says it has invested “in the introduction of enhanced stall-modeling techniques.” Like CAE, it provides upset-recovery and unusual-attitude training scenarios that present a “startle” factor to the pilot. Mitesh Patel, L-3 Link U.K.'s product development chief, says using representative stall models could “be of huge importance” due to the costs involved with aircraft-specific models.

Schroeder says the number of math models required for the U.S. fleet may vary depending on how frequently a “model developed for one series could be used for another series within the same airplane model.” He says the models' AOA will likely extend to 10 deg. beyond “stall identification” but that modelers will define the applicable slideslip range.

Foster says NASA expects to partner with the FAA on a variety of studies related to training and generic stall models for various aircraft.