The 2009 crash of Colgan Air Flight 3407 near Buffalo, New York, U.S., reverberated in April as training and safety specialists debated its effects on initial pilot qualifications, the adequacy of airline pilots’ hand-flying skills and adding hours to recurrent flight simulator training. Some predicted during sessions of the World Aviation Training Conference and Tradeshow (WATS 2012) in Orlando, Florida, U.S., that derivative regulatory changes will have unintended consequences. Others credited public pressure on legislators in the United States with breakthrough decisions on air transport safety issues.
“We are focused on fostering the kinds of behaviors that lead to professional conduct,” said Michael Huerta, acting administrator of the U.S. Federal Aviation Administration (FAA). He reminded attendees that the final rule addressing pilot fatigue, completed after years of disagreement, requires every pilot to exercise personal responsibility to arrive for work fit for duty, and every passenger airline to accommodate more rest in instances where flight delays or other reasonable circumstances take away the pilot’s opportunity for eight hours of uninterrupted sleep.
Some of the latest industry upheaval has surrounded updates to certification and qualification for airline pilots through a rule proposed in February (see, “Safety News March 2012”), which Huerta termed “the most significant overhaul in crew training in the last 20 years.” Essentially, the FAA wants all pilots to benefit from initial and continuing qualification comparable to that of the airlines that voluntarily participate in the agency’s advanced qualification program (AQP). Advanced flight simulators will be a key enabler.
“Not only do we want to require [that first officers hold an airline transport pilot (ATP)] certificate, but we propose to greatly increase the training to achieve it,” Huerta said. “For example, we believe it is necessary to have both academic and flight training in critical operating skills. [The rules also] would require pilots to demonstrate their skills in real scenarios … rather than have the pilot executing a recovery in a highly choreographed event. … Scenario-based training will enhance safety for the kinds of emergencies that we all acknowledge are extremely rare. We want pilots to have sufficient knowledge, experience and confidence so that they can appropriately handle any situation that life throws at them.”
Corridors of Power
Understanding the continued influence of the Colgan accident on aviation rulemaking requires familiarity with contemporary political dynamics of Washington, said John Allen, director, FAA Flight Standards Service. Government leaders, many in new positions when the accident occurred, and families of the passengers and crewmembers pressed for substantive changes to mitigate risks they perceived in airline industry practices.
The FAA was directed to gather data, identify and analyze airline pilots with documented problems in training and proficiency, and to explain the nation’s overall state of training and mentoring. This included reassessing the value of an ATP to the performance of first officers. Before the accident investigation was completed, senior transportation officials asked Allen for his perspective of the human factors.
“My perspective [then was] that a pilot needs to be mentally at least 10 nm [19 km] ahead of the airplane, has to know what’s coming up, has to be prepared — on the edge of the seat, anticipating the unusual,” he recalled. “That did not happen, so therefore [the Bombardier Q400 Colgan accident airplane] departed controlled flight. … [The crew] missed something, and then they were unable to recover because it was all a surprise.”
The FAA in short order was directed to complete 22 studies and task force reports to Congress, the National Transportation Safety Board and other entities; initiate eight rulemaking processes and create two databases. Again this year, a February congressional reauthorization of funding for the agency included further Colgan-derived rulemaking and study requirements, he said.
“There are some very positive things out of this, even though we are very cautious on legislating safety,” Allen said. “We are [first asking ourselves,] ‘What is the intent of these legislative requirements?’ … We will build upon them what we believe will be a better system. But there are unintended consequences that we are trying to work through, and they are [requiring] quite a bit of effort.”
Among responses to the changes, the FAA has narrowed its scope of rulemaking for mandating safety management systems (SMS) at major airlines. “If operators have a robust SMS, that means that they have programs for data collection tools, statistics and transparency to show the regulator how well they are managing their safety,” Allen said. “They are capturing the risks and the hazards, they are mitigating them, they’re getting a positive response, and they’re being forthright about it. … FAA inspectors are more efficient and more effective [if] they don’t have to waste time [in low-risk areas or] trying to ferret out whether there are unseen risks and hazards.”
Adoption of SMS has prompted reconsideration of some deeply ingrained FAA policies. “I’m not sure that our enforcement posture is serving safety very well right now,” he said. “We have over 4,000 enforcement actions in the pipeline, over 1,000 of them [more than] three years old. [Having] too many enforcement actions inhibits our attention to the significant ones. We also have a culture [of] inspectors who reflexively — because they don’t have guidance to say otherwise — initiate enforcement action [whenever they see a violation of regulations]. We are amending our guidance to provide a mechanism for our inspectors to work in a collaborative fashion to do the right thing for safety … to be judicious in our enforcement [yet apply penalties] when it is appropriate.”
Regarding first officer qualifications, the FAA’s senior leaders agree with the strong industry and academic view that quality of experience — not just flight hours — establishes “the quality of the pilot,” Allen said. “The legislation requiring the ATP is self-enacting, which means that the requirements become effective no later than July 31, 2013, regardless of any rulemaking action by the FAA. … [If the rule becomes] self-enacting … we are going to be scrambling to make sure that the pilot in the right seat in Part 121 [air carrier] operations has an ATP and 1,500 hours. It’s going to be a challenge. We are already working with industry to put together a posture as to how we are all going to manage … the ramifications.”
Much of the focus has been on methods of crediting specific academic study or experiences toward the hours. AQP advanced flight simulation for scenario-based training and checking also is a factor in the FAA’s thinking about how the qualification of airline pilots needs to evolve. The context of loss of control–in flight (LOC–I) warrants reconsideration of airline policies on maintaining manual flying skills and systematically evaluating “how much proficiency crews have on manually flying the aircraft,” Allen said. “[Airlines should] really think outside the box about extending the realm of training beyond just in a simulator and in initial qualification[to] look at line operations as well.”
Scheduled completion at the end of July of a notice of proposed rulemaking on professional pilot development, concerning mentoring programs, has been delayed.
Valuing Manual Flight Skills
Aircraft automation has been instrumental to air transport safety gains, said Jacques Drappier, a captain and senior adviser training, Airbus. Therefore, caution should be exercised in drawing conclusions about the causal role of automation in accidents, he said.
“Without automation … reduced vertical separation minimum and required navigation performance approaches would just be impossible,” Drappier said. “Continued efforts from all aviation manufacturers to further enhance the safety and economy of flight will bring more automation. But nothing is perfect, [and] maybe there have also been some side effects. One could be the loss of manual flying skill, and one may be an overreliance on automation.”
Some avenues for further scientific research into the question include measuring the effects of practice and the causes of skill erosion, poor quality ab initio training and level of experience before promotion of first officers to captains. “Hand-flying, in most cases nowadays and especially in the long hauls, is limited to one minute after takeoff and about two or three minutes in approach,” Drappier said. “But in some respects, automated aircraft may require a higher standard of basic stick-and-rudder skills [because they are not practiced often]. These skills are still necessary today when, at certain moments, there are abnormal situations or extreme weather conditions. The transition between smooth autopilot [flight] and a hair-raising situation can be very abrupt in modern cockpits.”
LOC–I and runway excursion events cannot be assumed to be attributable to flight crews’ use of automation. “Are we really looking at … erosion of our manual flying skills, or are we looking at an issue of airmanship?” he said. “When [Airbus] looked at cases where flying skill was blamed, often the real cause of the accident was a lack of situational awareness, lack of airmanship or disregard of rules. … It is too easy to blame automation.”
Careful consideration should be given to whether the skills-erosion hypothesis is correct, whether that issue is important even if correct, and if so, what changes would make a measurable difference in risk or outcomes, Drappier said.
So-called “missing or eroded manual flying skills” can become a catchall in LOC-I discussions and in discussions of training. “A lot of [LOC–I] accidents show a lack of recognition and a lack of anticipation,” he said. Yet, other than a 1985 study of automation side effects, a literature review for Airbus found a few studies from the mid-1990s specifically discussing erosion of airline pilot skills. “So the problem is probably not new, and it may be getting worse, but there are no [current] scientific data,” he said.
Anecdotal evidence at Airbus, however, does not support the assertion that significant numbers of active airline pilots have “lost” these skills. “What we see in our training centers is a few pilots who are a little bit rough on the edges, but the majority are still very capable and are doing a fine job in hand flying,” Drappier said.
Fly-by-wire automated airplanes let pilots fly manually while protecting them against gross errors, and significantly reduce the difficulty of handling. Compared with some earlier systems, “handling skills needed in daily life are extremely limited,” he said. “The automation has moved the emphasis away from flying the airplane, and has allowed the commander to free [his or her] brain from lower-level tasks … to manage the flight.”
The conventions of aviation have held that maintaining proficient manual flight skills to maneuver an airplane per regulatory standards only can be achieved through regular practice. “Clearly we have a problem here,” Drappier said. “On one hand, company policies are being set up to encourage, foster or even mandate the maximum use of automation for the benefit of safety and comfort. There is a strong pressure to avoid any hand flying. On the other hand, without practice, the motor skills will diminish. … Operational circumstances such as fatigue, weather and traffic will reduce further the opportunity to practice.”
Flight training adhering to U.S. or European evidence-based training principles does not necessarily address manual handling proficiency for abnormal or difficult situations such as upset prevention and recovery or crosswind landings. “We need dedicated sessions,” he said, citing a decision by Emirates to introduce four hours of additional simulator sessions per pilot in 2012 dedicated to manual flying proficiency. “Every three months or every six months, their pilots are back in the simulator to do flight director-off, autothrust-off [sessions such as] manual flying of patterns,” Drappier said. “I am sure that, if we take this problem seriously, [other airlines] will come to the same conclusion: [Pilots] need more [of this flying] time.”
Airbus suggests that even pilots with a solid foundation of hand-flying proficiency from earlier training should have manual handling skills developed or refreshed during type rating training. “We believe that at least two sessions’ worth of manual flying are needed during a type conversion,” he said. “We must use time in the full flight simulator to do handling, and push the automation exercises into flight simulation training devices. We also need to put more effort into the recurrent [hand-flying experience], where in recent years we’ve seen a reduction in overall time spent in the simulator.”
Memorization Overdose
Mike Carriker, a captain, aeronautical engineer and chief pilot, new airplane product development, Boeing Commercial Airplanes, told attendees that time spent designing airplanes has made him wary of persistent-but-obsolete pilot training practices.
Memorizing operations manual details — such as a 40-gal-per-minute flow rate of fluid through a hydraulic pump — makes little sense for flight crews in contemporary airline flying, he said. “We, as engineers, sized the hydraulic pump to make it work,” Carriker said. “Fear not, between [the manufacturers], FAA, European Aviation Safety Agency and [other authorities], we sized the hydraulic system [correctly].”
The first reform should be to stop requiring rote memorization from books, Carriker said. “No place — in 50,000 hours of analysis of failures in the 787 — was there anything [to suggest, for example,] a better outcome if a crewmember had recalled that the airplane has a 15-kVA electrical system.” Similarly, oral exam questions for a recent, non-Boeing, airplane type rating included one about the number of loops and detectors in the fire-suppression system, he said, adding, “There is no safety analysis anywhere that says that it is important for the crew to know that.”
Far more important than conserving a tradition of memorization is accelerating advances in airline pilot training and adapting to the learning strengths/preferences of multiple generations, he said. This includes “turning the airplane loose,” that is, taking full advantage of the latest technology for precise flight paths.
“This is my plea to the regulators and principal operations inspectors,” Carriker said. “When I was in [the recent type rating] training … the airplane was fully certified to do RNAV [area navigation providing] glide path and lateral direction.” When he asked to practice RNAV approaches, however, the request was denied because of objections from the principal operations inspector for the airline. “I told them, ‘How can one guy stop the whole industry?’” Instead, Carriker and the instructor conducted rarely used nondirectional beacon approaches.
“[A current Boeing] airplane possesses the capability to [utilize] billions of dollars worth of satellites and a multimillion-dollar, multisensor, integrated FMS [flight management system that provides] up/down, left/right guidance to the end of every runway in the world — with indication of deviation from the path and warning for excessive deviation,” he said. “The airline industry has to turn that [technology] on.”
Simulator Operations Quality Assurance
Flight data–driven flight training has been demonstrated recently in feasibility research for the U.S. Department of Defense (DoD) involving a Boeing 737 full flight simulator and military versions of this aircraft type, said Lou Németh, chief safety officer, CAE. The inspiration is flight operational quality assurance (FOQA) programs, typically collecting 1,500 variables (also called parameters) 11 times a second during routine flights for subsequent analysis.
“We use computer algorithms to see if the pilots are performing as they were trained, and to see if the aircraft is performing the way it was engineered and maintained to fly,” Németh said. “The simulator operations quality assurance [(SOQA) research now addresses the questions] ‘Is there value in SOQA data to look the training system as a whole, to see if it’s performing as it should?’; ‘Do SOQA data match the realities that we are seeing from FOQA data?’ and ‘Is there a correlation between FOQA and SOQA data?’”
SOQA basically comprises a full flight simulator, a data capture station, automated reports, analyses transmitted to the training manager, and data visualization/animation capability. “We’re monitoring the system, not necessarily the individual pilot performance,” he said.
Nevertheless, one simulator session during the SOQA feasibility research underscored the system’s ability to clarify risks when a pilot and/or an instructor is ambivalent about the seriousness of errors — or possibly even denies that errors in the simulator would have had serious safety consequences during a real flight.
A simulator-flown approach northbound into Colorado Springs, Colorado, U.S., specified a right turn in the published missed approach procedure to avoid the Rocky Mountains. In one observed event, however, with the simulator’s crash-inhibit function selected for unrelated reasons, miscommunication between the pilots led the pilot flying to turn left toward charted terrain, and the error was not detected until audible terrain alerts activated.
Shortly after this session, the crew and instructor told Németh that their error had been resolved. “The crew said, ‘We got pretty close, but we saved the day at the last moment, and we did not [strike] the terrain,’” he recalled. “I said, ‘Oh, really?’ In actuality, as seen from the animation, they ‘flew through’ a mountain.” The SOQA data replay with animation and data visualization showed controlled flight into terrain. “The visualization tool made it very clear to the instructor, and the students came away with entirely new behaviors because they could [relive] the problem from the outside looking in,” Németh added.
The DoD feasibility study concluded that applications of FOQA to these flight operations “brought a lot of value” and that SOQA was technically feasible and similarly valuable for flight training.
Correlations of FOQA and SOQA data were documented, he added, and automatic detection of deviations from standard operating practices or procedures was a significant, objective method of performance enhancement.
One common deviation, for example, involved violations of the procedure for setting approach flaps. Most frequent was late extension of landing gear during approach and landing. “We also wanted to see pilots land the airplane [with touchdown] just above 1 g [that is, one times standard gravitational acceleration], but the data showed a number of landings at 2 g to 3 g and one at about 5 g to 6 g,” he said.
As in an actual hard landing, the touchdown may not be perceived by pilots or instructors as a significant exceedance. “We don’t know [which hard landings] should require retraining unless we have the analytical basis to decide. [The SOQA] system will give us that information. The student and client benefit from a more precise indication of performance — and know they’re going to be treated impartially.”
Diminished Craftsmanship
As airplanes increasingly interface with their pilots through computers — with less, or no, direct connection with the control surfaces — hands-on “craftsmanship” in flying has diminished, said Christof Kemény, a captain and manager, cockpit crew training, Lufthansa CityLine. “However, if things go wrong … we immediately require [reverting to] these old craftsmanship skills for the pilot to recover from the situation,” he said.
As part of the expansion of its Embraer fleet, his airline has been developing a holistic approach to aircraft training and control to help mitigate the risk of LOC-I. “We have to take the human and machine as a system, and look at how we can optimize both,” Kemény said. “We might have lost the pilots a little bit along the way in not making them equal on the flight with … the purely technical systems.” Research on human cognitive performance — such as evidence for faster mental processing of subconscious versus conscious stimuli — suggests frontiers for developing habits and effective reactions during flight training, he said.
When taken by surprise, as often occurs in a stall or upset condition, “it [may take] six times longer than the subconscious perception, up to 600 ms, for a conscious perception … before pilots can start to analyze what they are seeing and then form a reaction,” Kemény said. “It’s in our best interest that flight training be geared toward habits [that] connect policies and procedures with the actual machine and the human being.”
The need to give flight crews every advantage to quickly perceive their flight path and energy state was a factor in the company’s decision to equip one fleet of new airplanes with dual head-up display systems (see, “Wait and See”) and to train pilots to use them during every flight, rather than equipping the fleet with autoland systems, he said.
The overarching Lufthansa CityLine philosophy includes attention to optimizing the pilot-machine interface. “We are talking about redefining pilot monitoring as a skill, not as tasks and duties … and, with this, comes the question ‘Do our present aircraft have the correct indications or annunciations to support the skill of monitoring?’” he said.
Faster Global Harmony
The International Civil Aviation Organization (ICAO) called for the earliest possible certification of new air transportation systems in the United States, Europe and elsewhere and global synchronization of related investment decisions. “We expect, in the next 10 years, more than $120 billion to be spent worldwide on new aviation systems,” said Nancy Graham, director, ICAO Air Navigation Bureau. “New technologies and procedures — complete with the right number of well-trained people — can serve to keep our excellent record [that is, the improved 2011 rate of commercial jet accidents] intact. In fact, we can improve the record if it’s done in a harmonized way. Harmonization is a nice word for negotiation … very easy to talk about but very hard to do.”
Work underway by individual states — now at different stages of launching initiatives to handle traffic growth — could proceed more rationally under a proposed worldwide timeframe of block upgrades, she said. The strategy, for example, supports decisions about which elements need to be interoperable and which systems need to be harmonized to achieve seamless flight operations.
Meetings by an informal “challenge team,” including technical representatives for states with the most advanced efforts, have propelled ICAO down this road. “The challenge team is going to look at a couple of modules that have to be mandates,” Graham said. “Elements that are backbones of the ‘system of systems’ — datalink might be one, systemwide information management might be another —ultimately will have to be mandated [by ICAO] for safety reasons. … as opposed to leaving [them] to regions.”
Each block upgrade would encompass “a group of elements that are operationally driven to provide a cost-effective upgrade with … a clearly defined operational improvement that is measurable … a positive business case over a defined period of time, and an operational or certification approval path,” Graham said. Current systems have been designated as block 0, and four block upgrades lie ahead. Block 1, containing about 15 modules including new ICAO standards, would be available for operators to use and achieve their intended benefits from 2018, while block 2 modules would be available from 2023, for example.
“[This] allows flexibility for states that don’t have high-density traffic to come along as they need to, based on their operational requirements,” she said. “At the highest level, [ICAO’s] Global Aviation Safety Plan and Global Air Navigation Plan are intrinsically linked … but driven by separate processes. … Every three years, [these plans] will be updated to be reflective of new priorities and activities.” The benefits are grouped as “green” airports; globally interoperable systems and data; optimal capacity and flexible flights; and efficient flight paths. An ICAO conference in November will provide further details, including efforts to build a global business case.
Airlines, the International Air Transport Association, air navigation service providers and airports, meanwhile, are working with ICAO to aggregate their safety data into one global aviation safety index designed to “give us a much better indication of the overall situation of safety in any particular state,” she added.
General Aviation Risks
The quality and effectiveness of initial pilot training logically would affect the safety of the National Airspace System, said Gerald Dillingham, director of civil aviation issues, U.S. Government Accountability Office (GAO). After the Colgan accident, “the integrity of all levels of training came under scrutiny” by the Congress, he said. Data showing the effects have been difficult to obtain, however.
“One of the issues that became part of the public debate after the Colgan accident and during the development of the 2010 Airline Safety Act was effectiveness of training on pilot performance,” Dillingham said. “Stakeholders told the GAO that they were concerned about … [students’ use of] rote memorization to pass knowledge tests … rather than understanding aeronautical issues. They were also concerned that practical [flight] tests focus on mastery of skills rather than ability to respond [safely] to various scenarios. … They pointed out that although airline operations have evolved operationally and technologically, training requirements for commercial pilot certificates largely have not been revised for about 15 years.”
GAO analysis of general aviation accidents from 1991 through 2011 looked — with difficulty —for specific links to training. “[The year] 2011 marked the fifth consecutive year of declines in accidents and fatal accidents,” he said. “We are currently at a 20-year low [in general aviation fatalities but they] still account for the vast majority of U.S. aviation fatalities. … Because of the lack of reliable flight activity data, we cannot know exactly what [changes in rates occur]. … [Pilots holding a private pilot certificate] were disproportionately represented in accidents. … General aviation traffic was involved in about 60 percent of [all runway] incursions but represented less than 33 percent of operations at towered airports. About 30 percent of these incursions involved two general aviation aircraft, and 77 percent of incursions [classified as] pilot deviations involved general aviation pilots.”