Fatal crash shines light on in-flight icing danger

 - April 1, 2009, 6:16 AM

The crash of Colgan Air/Continental Connection Flight 3407 (a Bombardier Dash 8 Q400) on February 2 has again raised the same issues about in-flight icing that came to light after the 1994 icing-related crash of American Eagle Flight 4184 in Roselawn, Ind., and other icing accidents. The issues have to do with autopilot use in icing conditions, training that pilots receive for flight-in-icing conditions, especially regional airline pilots flying airplanes at altitudes more conducive to icing, and why air traffic controllers and pilots don’t work together to avoid icing conditions instead of keeping airplanes in icing for long periods of time during terminal operations. Although the manufacturer of the ATR-72 that crashed in Roselawn was criticized for design features that some believed contributed to the ice accumulation on the American Eagle turboprop, the significant amount of time that the pilots spent in what turned out to be severe icing conditions was a major factor in that accident.

Since the Roselawn accident, two widely publicized accidents, in which pilots kept their airplanes on autopilot in icing conditions to the point that the airplanes stalled and then crashed, have added to the debate about pilot technique in icing. That the Colgan pilots did exactly the same has generated concern about whether key information is being delivered to pilots. The other two accidents in which autopilot use was a factor were Comair Flight 3272, an Embraer EMB-120 Brasilia on Jan. 9, 1997, in Monroe, Mich., and the crash of a Circuit City Citation 560 in Pueblo, Colo., on Feb. 16, 2005. For unexplained reasons, the NTSB’s conclusions on the causes of these two accidents skipped over the autopilot use, although it was covered in detail inside the accident reports.

Pilot Training Questioned
In Safety Alert SA-014 published in December 2008, the Safety Board recommended that pilots “Turn off or limit the use of the autopilot in order to better ‘feel’ changes in the handling qualities of the airplane.” It is clear that the NTSB feels strongly about the use of autopilots in icing conditions.

In these accidents, including the Colgan accident, airspeed decreased to a point that the autopilot automatically kicked off. The NTSB doesn’t address this point specifically, but this common thread in these accidents raises the question: Was there something distracting the pilots so that they failed to notice the decrease in airspeed, especially while carrying a load of ice?

In the recent Colgan accident, it is too early to say. The NTSB has not yet released the cockpit voice recorder transcript. AIN has submitted a Freedom of Information Act request to the NTSB for release of the CVR transcript but had not received a response by press time.

Colgan Air has issued a series of questions and answers that address the training its pilots received, but these do not get into detail about what pilots are taught with regard to flying in icing conditions. “Our training programs specifically and thoroughly address emergency situations such as icing, stalls and other in-flight scenarios. When our crews fly our aircraft, we believe, and the FAA has certified, that our crews are prepared to handle emergency situations they might face. In November 2008, Capt. Renslow and first officer Shaw [the crew of the accident airplane], along with all other Q400 pilots at Colgan, reviewed the Winter Ice Bulletin in preparation for the upcoming winter season. The bulletin provided the latest information on flying the Q400 in icing conditions, and it supplemented the training and experience the captain and first officer already had with the Q400. Again, our training programs address emergency situations such as icing and other possible emergency situations,” said the company. Colgan has not released the contents of this bulletin.

During the arrival into the Buffalo area on the night of the accident, the crew of Flight 3407 reported “significant” icing to ATC as they descended from 11,000 feet. What does ATC do when pilots report icing? Do controllers reroute airplanes around where the icing conditions are reported or assign them to a higher altitude for a longer time so that they can approach the airport above the icing conditions and then quickly descend through icing before landing?

ATC does not as a normal procedure change inbound altitudes just because
of icing conditions, according to responses to AIN’s questions from the FAA, Eurocontrol and Nav Canada. “It is always incumbent on the pilot to cry uncle when they see the situation worsening,” said one former FAA controller.

Another controller said, “There are no specific policies I’m aware of in regard to icing. Airspace design around airports is kind of like airline schedules. It’s designed for the best of conditions, and you can’t change it when the weather gets bad.”

One pilot for a major airline flying a 757 complained about this set of circumstances, after a recent flight into a northern U.S. airport: “Everyone was getting ice while being vectored all around at low altitudes today as [the airport] struggled to keep runways plowed, opening and closing them very frequently. ATC doesn’t seem to understand the concept of keeping everyone high and above it. The current system functions the same whether [there is] icing or not. Pilots don’t have a lot of info about how long they’ll get vectored and maybe just a pirep or two on icing. I’m sure it’s a mess for ATC to keep us high–with departures to intermix–but it would be safer.”

However, controllers will help if asked. According to an FAA spokesman, “If aircraft encounter icing and need ATC to accommodate different altitudes or routing, we do what we can to allow for that. When an aircraft needs priority, it’s given priority. If it’s not to that level, we accommodate routing and altitude requests.”

ATC Procedures for In-flight Icing
In Europe, the same is true regarding pilot requests concerning icing. “Pilots forward their requests to ATC asking for different altitude or rerouting to avoid or mitigate icing conditions,” explained a Eurocontrol spokeswoman. “When ATC is aware of icing conditions, [the controller]  will also issue information to pilots on any known icing conditions and possible flight levels affected and propose or instruct alternative levels or routings to them.”

In Canada, according to Nav Canada, “Upon pilot request or when deemed appropriate, controllers instruct pilots to climb or descend away from altitudes where known icing conditions are occurring, after ensuring there is no conflict with other aircraft. This is normally initiated by a pilot requesting an altitude change after observing icing conditions, or by a controller when he or she is aware of potential icing conditions that could directly affect aircraft performance.”

The problem with trying to anticipate and observe icing conditions is that, unlike thunderstorms, which can be detected by radar, there is no such device for icing. Weather reports might indicate ripe conditions for icing and some new forecasting and reporting tools might help, but the only way to know that icing actually exists in a specific area and altitude is for a pilot to report the condition.

Finally, there is the question of de-icer-boot-equipped airplanes and those with heated leading edges. Is one system better than the other, and why don’t turboprops use heated leading edges?

According to icing research engineer Tom Ratvasky at NASA’s Glenn Research Center in Cleveland, turboprops don’t provide enough excess power–and hence have adequate bleed air–to run heated leading edge systems, although bleed air is used to inflate and deflate de-icer boots and pressurize the cabin. A large enough engine on a turboprop with heated leading edges would probably negate the turboprop’s efficiency compared with a turbofan-powered airplane.

Ratvasky disputes the hypothesis that the higher speed of jets is enough to prevent ice from forming or melt it. “Some schools of thought say that heat due to stagnation is sufficient to boil ice off,” he said. “There’s not a speed [at which] ice will go away; that’s a false assumption.”

There is the perception that turboprops are more prone to ice than jets, but that can be explained by the fact that jets spend less time at the lower altitudes, where icing occurs most often and where turboprops fly more often. Therefore, jets simply are not exposed to icing as frequently as turboprops are. 

NASA Glenn is working with other research organizations to help the FAA define new certification standards for dangerous supercooled, large-droplet (SLD) icing, which was implicated in the Rose-lawn accident. Ratvasky expects the FAA to propose a new SLD icing certification envelope soon.