“We found that a string of events leads to an upset,” Advani says. Included are systems failures, cockpit resources management issues and “not paying attention to signs of a developing threat.” He says simulator training can be improved by teaching pilots to recognize changes in aircraft handling and control as an upset condition is approached in increments, and practicing recoveries along the way.

“For high-altitude stall training, you should not be putting the pilot in the simulator and saying 'recover',” says Advani. “Pilots need to recognize the signs, the buffet, the stall warning, the stickpusher, and learn not to fight those systems. These are basics that we have not adequately or consistently trained for.”

To address the other 25% of its goal, Icatee would like to see extended flight envelopes for full-flight simulators, which today match the actual aircraft in handling qualities to near-stall conditions but not beyond. “There are huge gaps between current simulations and reality [in the stall], says Advani. He notes that a stall in a large transport aircraft simulator today “feels very benign,” marked by “a little buffet” and a stickshaker, followed by “mushy” control characteristics and a “steep glide down to the ground” in a deeper stall.

In the reality, he says the stall experience in an actual transport, as relayed from a Boeing 777 test pilot, is a “physically painful” and “violent” experience in terms of buffet, and much less stable in terms of roll and yaw-damping after the stall compared to the simulator.

“We want to identify which elements of reality are important for a pilot to know to never go there . . . but not making them masters of the proper stall,” says Advani. With that goal in mind, he says, a “representative” model of a transport category aircraft for the simulator will likely be adequate to teach pilots to reduce angle of attack and wing loading, even in a startle condition.

Peter Grant, a human control expert in the vehicle simulation group at the University of Toronto, also believes that a generic model “is plausible” for positive transfer of training between the generic simulator and specific aircraft in terms of upset prevention and recovery training (UPRT), but that more studies are needed. “We have empirical evidence and we believe generic models can solve part of the problem, but we don't have proof,” he says.

Grant's group has developed a new adaptive motion drive algorithm for the university's Boeing 747-100 simulator using NASA wind tunnel data for a generic two-engine large transport. The algorithm maps the limited motion of a full-flight simulator to the large motions that would occur in reality during stall, wind shear and “large attitude deviation” events.

Figuring out how to startle and surprise pilots in a simulated environment without sustained g-levels could prove to be more difficult.

“The problem always remains—how do you create g awareness,” says Advani. “We can't create negative-g or high-acceleration environments in the real transport category aircraft.” Instead, Icatee proposes that newly hired commercial airline pilots be exposed to inflight upset maneuvers with an instructor who will teach them recovery techniques. “They're not learning aerobatics; they're learning upset recoveries,” he says.

Advani says instructors during inflight training can create scenarios, such as the aircraft turning from base leg to final when encountering wake turbulence, and have the pilot recover. “There are two fundamental benefits—it teaches them how to apply the upset recovery skills and how to manage the startle factor,” says Advani, adding that an aerobatic aircraft may not be required.