Category: musings

[Fair warning. I’m writing about the UI design of airplane cockpits in the midst of a mystifying airplane disaster that has consumed the planet for nearly 4 weeks, Malaysia Flight 370. Which made me start reading about previous air disasters, which led me to Air France 447.]

Prior to the MF370 incident I read (on vacation), “The Unthinkable: Who Survives When Disaster Strikes – And Why” by Amanda Ripley. It’s a fascinating look at how our brains are wired (or not wired) to react when disaster strikes, be it natural or man-made. Basically, most people think they’re going to act and behave a certain way, but don’t, which results in living or dying. Ripley examines, via survivor stories and research, that for every super power our brain gives us during a stressful situation (example: super human strength), the brain also takes something away, such as our vision or hearing. Also, seemingly simple motor tasks may become too complex to achieve, like putting on a life jacket, unbuckling a seat belt, opening a door, revealing why muscle memory can be the key to survival. In a nutshell, areas that control certain aspects of our brain under normal situations are minimized while other aspects become more dominant.

In her book Ripley describes an aviation accident in which, during routine landing approach, the landing gear indicator light failed to illuminate. The crew remained airborne while they worked to trouble shoot the incident, unaware that they were still descending, despite the fact that the audio warnings were going off in the cockpit. Ultimately the flight crashed, and it was determined that the crew was so fixated on the indicator light (which turned out to be a burned out bulb), that they didn’t hear the warnings. This is just one example of how our cognitive systems react to extreme situations. Under a normal scenario, the crew would have heard and reacted appropriately to the alarms, but the increased stress and narrowed focus obviously affected what the crew heard and saw.

Air France 447
On May 31, 2009, Air France 447 (an Airbus A330), en route to Paris, France, crashed into the Atlantic Ocean. The BEA’s (Bureau d’Enquetes et d’Analyses – French air accident investigation agency) final report concluded the aircraft crashed after temporary inconsistencies in the airspeed sensors caused the crew to react incorrectly, leading to an aerodynamic stall from which they could not recover in time.

Whether it was pilot inexperience and/or lack of familiarity with the controls and inputs, the PF (pilot flying – total flight crew of 3) kept pulling the stick back, which ultimately caused the Airbus A330 climb to unrecoverable stall.  Without overwhelming you with details regarding angle of attack, lift coefficient, pitch inputs, and etc., the plane entered an electromagnetic thunderstorm which disengaged the autopilot, to which the pilot flying (PF) reacted by pulling the stick back. Despite 75+ repeated audio stall alerts, the flight crew was unable to comprehend what was happening to the aircraft. Even if one pilot was unable to understand what was going wrong, how is it that his two colleagues failed to notice the issue (PF pulling back on the stick)? Possibly contributing to the crew’s confusion and panic were several major factors:

• Some of the primary cockpit display panels were dark (went black, no longer displayed data) because the computer that receives the data was no longer able to make sense of what was coming in (due to lack of minimum airspeed, temperature), and subsequently turned the cockpit displays off
• The other pilot in the cockpit did not likely see the PF pulling his stick back, due to the location of the side stick and the fact that the side sticks to not move in tandem like traditional yoke controls.
• Airbus auto-thrust bypasses the manual levers entirely, they don’t move. Instead, pilots have to check their cockpit display panels.

Side sticks
The A330 is controlled by side sticks that, as the label implies, are located beside the pilots’ seats. They resemble joysticks from video game consoles. The side sticks aren’t connected to the control surfaces by levers and pulleys as in older aircraft, but instead side stick commands are fed to computers that send signals to the engines and hydraulics. Often called “fly-by-wire” technology, it has advantages in reduced weight, therefore saving fuel, they have fewer moving parts, have backup systems, and are programmed to compensate for human error. The sticks also enable pilots to concentrate on other aspects of flying because once a command is given, the pilot can release the stick after a few seconds.

However, because the stick is at the side of the pilots seats and not in front of the pilots like the traditional control yoke, one pilot may not be able to see what the other pilot is doing, unless that pilot makes a big effort to look across the other side of the flight deck, which is not easy or convenient, especially in a tense situation when other inputs require attention. Additionally, side sticks do not reflect what command one stick might be receiving. Recall how a traditional control yoke situated in front of the pilot: when one yoke is pulled forward or backward, the other yoke will move in tandem, indicating what the PF is doing. Another aspect of the side stick is  that if both pilots move their respective control sticks, the inputs are averaged.

Glass Cockpit
The A330 also features a “glass cockpit” (as does current Boeing designs).  A glass cockpit features digital instrument displays, typically large LCD screens instead of traditional analog dials and gauges. While the traditional cockpit (aka “steam cockpit”) relies on mechanical gauges to display information, the glass cockpit displays are driven by flight management systems which can be adjusted to display flight information as necessary. Ultimately this simplifies aircraft navigation and operation. Glass cockpits are popular with airline companies because they tend to eliminate the need for a flight engineer ($$$$ saver).

However, with the dependency on the glass cockpit systems, comes the need to train crews to deal with glass cockpit blackouts and failures. On 25 January 2008 United Airlines Flight 731 experienced a serious glass-cockpit blackout, losing half of the ECAM displays as well as all radios, transponders, TCAS, and attitude indicators. Partially due to good weather and daylight conditions, the pilots were able to land successfully at Newark Airport without radio contact.

In 2010, the NTSB published a study done on 8,000 general aviation light aircraft. The study found that, although aircraft equipped with glass cockpits had a lower overall accident rate, they also had a larger chance of being involved in a fatal accident.  The NTSB Chairman said in response to the study:

Training is clearly one of the key components to reducing the accident rate of light planes equipped with glass cockpits, and this study clearly demonstrates the life and death importance of appropriate training on these complex systems… While the technological innovations and flight management tools that glass cockpit equipped airplanes bring to the general aviation community should reduce the number of fatal accidents, we have not—unfortunately—seen that happen.

Auto-thrust
Airbus computers can automatically adjust the engine thrust to maintain whatever speed is selected by the crew. This means pilots do not need to keep fine-tuning the throttles on the cockpit’s centre console to control the power. But a curious feature of Airbus “auto-thrust” is that it bypasses the manual levers entirely – they simply do not move. This means pilots cannot sense the power setting by touching or glancing at the throttle levers. Instead, they have to check their computer screens. According to Boeing, some airline pilots do not like the absence of motion in the flight deck, which is why Boeing’s system the manual handles move even when the system is in automatic mode.

It has been debated that if AF 447 had been a Boeing aircraft the pilot would have realized  the PF had the stick pulled back, as Boeing cockpits are still designed with center column steering. Boeing has also persisted with some conventional controls in it’s fly-by-wire aircraft, even the 787. Boeing cockpits still require levers that have to be pushed and turned like older mechanical versions, despite the fact that the levers send impulses to computers. The controls still have to be held in place or a maneuver to be accomplished. Some pilots think the Boeing approach overly conservative, and it makes it difficult to transition from one type of flying to another. Despite the fact that hand-flying is not as common because of automation for fatigue reduction and reliability, there is something to be said for visual and haptic feedback, particularly in a scenario when not all senses are functioning normally and inputs are under duress. Not to mention having critical displays visible at all time.

In regards to flight AF447, BEA director Jean-Paul Troadec and his team pointed at human-machine interface issues that made the situation extremely confusing for the crew at the controls of the plane. Ultimately, safety comes from both the pilots’ cognitive abilities and the signals they receive, Troadec concluded. He also stated, in response to the audio stall warnings being ignored, “it could have been a matter of sheer perception…Audio alarms are no longer heard in some situations.”  This has prompted the BEA to recommend the addition of a visual stall warning. The BEA report recommended more training for dealing with unexpected situations as well. 

Article sources/inspiration:
A Startingly Simple Theory About the Missing Malaysia Airlines Jet
Wikipedia: Air France Flight 447
Final AF447 Report Suggests Pilot Slavishly Followed Flight Director Pitch-Up Commands
Air France Flight 447: ‘Damn it, we’re going to crash’