On January 15, 2009, US Airways Flight 1549 experienced a bird strike shortly after take-off that crippled both its engines, leading to the famous water ditch landing we’ve come to know as the “Miracle on the Hudson.”
After the incident, the aircraft’s pilot, Dr. Chesley “Sully” Sullenberger, became the center of international acclaim and spotlight, as his talent and skill was credited with saving the lives of all 155 passengers and crew aboard. The recently released movie “Sully” revisits Flight 1549 and the subsequent 15-month National Transportation Safety Board (NTSB) investigation to explore a critical question:
Was the Hudson River landing truly the best option after dual-engine thrust loss, or would it have been possible for Captain Sullenberger to land the plane safely back at La Guardia?
As reflected in the movie, the key factor to consider when answering this question is human delay: in recognizing and diagnosing the problem after the bird strike incident, in weighing possible remedial options, and in ultimately selecting a plan of action.
In a series of emergency simulations run after the incident from Airbus’s headquarters in Toulouse, France, test pilots were able to safely land the aircraft at La Guardia airport if they initiated their approach immediately. The NTSB ultimately rejected these simulations as reflections on Captain Sullenberger’s performance by noting that they failed to “account for real-world considerations such as time delay [between the human crew’s perception and reaction].”
In her AIAA paper, “Emergency Landing Automation Aids: An Evaluation Inspired by US Airways Flight 1549,” AE Professor Ella Atkins proposes an adaptive or emergency flight planning architecture that would erase this time delay and enable flights such as Airways 1549 to land safely in crisis situations. If Atkins’ architecture, which contains landing site search, post-failure trajectory planning, and datalink capabilities, had been installed aboard Flight 1549, the event may have played out quite differently:
In under a second after the bird strike, the emergency flight planning software could have diagnosed the dual-engine failure by detecting throttle levers at nominal position with indications of no thrust produced. Immediately afterwards, the software would have initiated a search for desirable and feasible landing sites based on the aircraft’s post-failure state. Using data from the NTSB accident docket report to simulate Flight 1549 conditions, Dr. Atkins concluded that her software would have determined four possible runways at La Guardia on which the airplane could have safely landed. The software would have then calculated a safe trajectory to the most preferred runway, and communicated the flight plan to both the pilot and air traffic control system. Once the pilot provided flight plan approval, or the certified system autonomously activated the new emergency landing plan, the aircraft could be safely flown to a landing with much lower risk to people and property than was present for the Hudson River landing.
Even when automated in this fashion, timing is still of the absolute essence when it comes to a safe landing. In the moment following the bird strike, four La Guardia runways, Runways 31, 13, 4, 22, were feasible landing sites. However, as time progressed and the aircraft continued to turn away from La Guardia, these runways eventually became unreachable. Four seconds after the bird strike, with the aircraft having descended less than one hundred additional feet, Runway 31 became unviable. After a mere 16 seconds, none of the four were feasible options.
In the absence of emergency flight planning software on his aircraft that could inform and expedite his decision-making, Captain Sullenberger was forced to execute the water ditch landing as his most promising option. Indeed, attempting to land at a runway without automated trajectory planning could have resulted in not only disastrous consequences for his passengers and crew, but for civilians on the ground. As Captain Sullenberger explained in his book Highest Duty: My Search for What Really Matters:
“I had to be certain we could make it [back to La Guardia] because it would rule out every other option [and could kill] who knows how many people on the ground.”
A necessary ingredient to successfully executing an emergency flight trajectory is clear and efficient communication of the plan to air traffic control (ATC). Traditionally, such communication is conducted verbally, with the pilot and ATC attempting to conduct a complete discussion of the situation in the height and confusion of a crisis. Precious seconds can be lost in this exchange; for example, during Flight 1549, the ATC recommended runways long after they were unreachable due to incomplete understanding of the situation. Atkins’ architecture would enable near instantaneous transmission of status and possible trajectory options to ATC that would enable clearance of the appropriate flight path as efficiently as possible.
Overall, Dr. Atkins does not consider her software to be a substitution for skilled and experienced pilots. Rather, she hopes that her analysis can “link emergency flight planning algorithms, designed to be intuitive, provably-correct, and computationally-tractable, with the US Airways [Flight 1549] incident, to show how cockpit automation can improve safety through its proliferation into NextGen cockpits and onto NextGen datalinks.”