Despite recommendations from one of its own safety specialists, the Federal Aviation Administration failed to require changes to a popular commuter plane a full year before one of those planes crashed near Detroit, killing all 29 people on board.
The recommendations in a January 1996 internal FAA memo called for measures that would allow pilots flying the 30-seat Embraer Brasilia 120 to better detect ice building up on the wings of their planes, known as EMB-120s.
In-flight icing is the suspected cause in the Jan. 9, 1997, crash of Comair-Delta Connection Flight 3272 near Detroit.
Several aviation experts and the Airline Pilots Association, the nation’s largest pilot union, contend that the fixes would have prevented the crash of Flight 3272.
A Tribune examination shows the FAA did not adopt the recommendations until last January, two years after the memo and a year after the Comair crash.
The National Transportation Safety Board on Thursday will release the results of its investigation into the Flight 3272 accident. The board is expected to focus on the EMB-120’s problems when flying in icy weather and on the FAA’s failure to address it.
John Dow, an FAA icing expert, wrote the memo after discovering a trend. There had been six previous cases involving the EMB-120 since 1989, all of them similar to the Flight 3272 accident. None resulted in a loss of life, but each occurred in icy rain and drizzle, which can adhere to the wings and disrupt airflow.
FAA and Embraer officials refused to discuss their actions regarding the EMB-120. Dow also would not comment. Six U.S. airlines fly a total of 220 EMB-120s. United Express and Comair fly the plane in service to O’Hare International and Midway Airports.
In addition to Dow’s recommendations, the NTSB also is expected to address wind tunnel tests conducted by NASA after the Flight 3272 crash. Those tests showed that ice can form on the underside of the EMB-120’s wing, behind de-icing devices.
According to the pilots union, this causes drag on the wings, slowing the airplane so much that it cannot fly.
Embraer knew about ice forming behind the de-icers as early as 1980, the pilots union argues, when it was predicted in a study by BFGoodrich Aerospace, which makes the de-icing equipment.
“ALPA believes that this accident was avoidable and was caused by the actions (or inactions) of many organizations,” the union report contends. “There were several significant warnings during the history of EMB-120 operations that should have resulted in proactive actions to preclude an accident.”
The crash of Flight 3272 came as the plane was descending at the end of a 76-minute hop from Cincinnati to Detroit. Pilots Dann Carlsen and Ken Reece had no indication that something was going wrong.
Trouble began just after Detroit air controllers ordered the pilots to slow the plane and then to turn left. As the plane began to turn, it dramatically lost speed.
“Looks like your speed indicator,” Carlsen told his first officer with professional calm. The engine power was increased, but the EMB-120’s turn kept getting steeper, its airspeed lower. “Power,” Carlsen commanded. The plane’s bank angle reached 45 degrees, and then the autopilot disconnected. Suddenly, Flight 3272 flipped onto its back.
The plane swung back and forth as the pilots tried to regain control, but just 29 seconds after the first hint of trouble, Flight 3272 hurtled into the wet ground near Monroe, Mich. In less than a half-minute, 29 lives were ended.
Immediately afterward, pilot error was suspected. The plane had been flying at 150 knots, which an Embraer executive said was too slow given the icy conditions. At the time, Embraer had not issued minimum speeds for flying the plane in ice.
Dow and others at the FAA, however, already had flagged the EMB-120 as a plane that could have problems in icy weather.
Their suspicions were linked to another deadly aviation accident, that of an American Eagle ATR-72 at Roselawn, Ind. The plane, bound for O’Hare, crashed in October 1994, killing 68 people, many of them from the Chicago area.
As in the Comair crash, aviation analysts said immediately afterward that the ATR pilots were at fault. The NTSB, though, investigated and found something unexpected.
The ATR had a de-icing device, or “boot,” that wasn’t big enough. Boots are strips of rubber that run along the front edges of the wings and the tail’s horizontal stabilizers.
When ice adheres to those surfaces, the pilot flips a switch that inflates the boots with air. As they inflate, the boots expand and the ice breaks away. The technology is found on most turboprop planes. (Jets, with more powerful engines, usually have heated metal leading edges instead of boots and rarely have icing difficulties.).
The size of the boots is also an issue in the Comair Flight 3272 crash. Icing experts and the pilots union claim that the boots on the EMB-120 are not big enough to protect the wing.
In the case of the ATR, ice was forming in ridges on the top of the wing behind the boot, disrupting airflow. Though FAA engineers knew as early as 1986 that the ATR had an icing problem, the agency for years did not address the core issue, which was that the de-icing boot did not cover enough of the wing.
Until the Roselawn crash, the FAA also failed to notice a pattern in 12 other “roll upsets,” cases in which ATR planes had flown into ice and lost control.
After Roselawn, the FAA examined the ATR’s vulnerabilities by flying one of the planes behind an airborne tanker that sprayed water to simulate in-flight icing. As a result, the FAA and ATR decided that the plane needed a bigger de-icing boot.
As part of the ATR tests, the FAA also reviewed incidents involving other turboprop planes. That’s when Dow, working out of the FAA’s Kansas City office, discovered the pattern of previous EMB-120 incidents–five roll upsets and one loss of speed, all in icing conditions.
On Nov. 7, 1995, the FAA sat down with the airlines that flew the plane to discuss the potential dangers.
“Some action was taken at that meeting, the issuing of an operations bulletin and so on,” said one participant. “But there was no examination of specific things” related to the six EMB-120 incidents.
Two months after the meeting, Dow issued his memo. It noted that the EMB-120 seemed susceptible to ice. He recommended a close examination of the plane to determine if pilots get an early enough indication that ice is building up on the wings.
Dow also called for the FAA to “develop appropriate corrective means to prevent ice formation or remove ice,” on wings. That would suggest, icing experts say, putting larger boots on the EMB-120, just as the FAA did in the ATR case.
Finally, Dow recommended that if pilots were not getting enough warning about ice, the EMB-120 should have “a reliable means” to assess the danger of ice and to fly out of it. Icing experts say that would mean installation of an ice detector, which is one of the fixes the FAA finally adopted, at a cost of about $16,000 per plane. The improvement came after the Flight 3272 crash.
Embraer, meanwhile, was conducting its flight tests in late 1995. The plane exhibited control problems, says the pilots union, which reviewed the flight test results.
In April 1996, Embraer issued a bulletin stating that the plane had performed well in icing tests. The bulletin recommended increasing the holding speed from 160 to 170 knots and adding five knots to the landing approach speed in icing conditions. Embraer also suggested that pilots turn on the de-icing boots at the first hint of ice. At the time, pilots were following an FAA recommendation to wait until one-quarter to one-half inch of ice had built up on the wing.
Comair, based near Cincinnati, chose not to adopt Embraer’s changes. Wayne Wolke, a Comair captain and the company’s program manager for the EMB-120, told NTSB investigators that Embraer did not provide data to support the changes.
A few days after the Embraer bulletin, the FAA issued a mandatory airworthiness directive affecting 16 commuter airplanes.
It gave pilots “visual cues” to look for in icing conditions and required pilots to request priority handling from air-traffic controllers if their planes were flying in ice.
The directive did not, however, take up the EMB-120 specifically. Nor did it take up Dow’s recommendations regarding the plane.
No further changes were made in the months leading up to Flight 3272’s departure for Detroit, but the crew members on that flight were no strangers to ice.
Carlsen had 5,329 hours of flying time as a pilot, including 2,302 hours in the EMB-120. He also had worked as an instructor on the new jets that Comair was acquiring, and he had conducted performance analysis reviews on the new jet for the airline.
“He was well-respected and regarded as an instructor,” one NTSB interview states. “Very detail-oriented, professional, and serious about his job. He never received negative comments from the FOs (first officers) or other pilots.”
In interviews with colleagues, NTSB examiners heard similar raves about Reece. He had 2,582 hours of flying time, including 1,494 hours in the EMB-120. Both pilots had received training in recovering planes from “unusual attitudes” such as those encountered by Flight 3272.
The crew’s apparent high skill level proved especially vexing to investigators after the crash. No one could figure out how two talented pilots could have allowed it to happen.
Picking over the rubble and flight-data recorders, investigators focused almost immediately on the plane’s airspeed. The EMB-120 can fly safely at 150 knots (about 172 m.p.h.), and airline pilots generally comply with air-traffic controllers’ requests for certain speeds to keep a safe distance between landing airplanes.
Many pilots who fly the EMB-120 said they extend the wing flaps at that speed, according to interviews conducted by NTSB investigators. Flaps, when extended, essentially make the wing bigger, creating more lift and allowing a plane to fly at a slower speed while descending. Embraer, though, had not issued minimum flying speeds for the EMB-120 in icing conditions.
The pilots of Flight 3272 had not extended the plane’s flaps. They also apparently had not turned on the de-icing boots. That might mean that they either did not think they were flying in icing conditions or that the plane had not accumulated enough ice to warrant activating the boots.
Weather information and reports from planes flying nearby showed ice at the altitudes that Flight 3272 passed through as it approached Detroit. The pilots union argues that weather radar information and the path of Flight 3272 show that the plane passed though icing conditions for just 40 seconds. Still, the union maintains that was probably enough exposure for the plane to accumulate ice under its wings, where the pilots could not see it.
When the pilots applied full power to recover from their stall, the plane didn’t respond as usual. The union says that condition corresponds to the results of wind tunnel tests conducted at the NASA Lewis Research Center in Cleveland, which showed that “drag” on the wing increased dramatically when ice formed.
In earlier statements, Embraer said the pilots and not ice-related drag were at fault. A company executive said the Comair pilots simply flew the plane too slowly, causing it to stall. Others, however, say that the six previous incidents experienced by EMB-120s suggest that something more than pilot failure was involved.
“Embraer’s approach to those things was primarily to blame it on the crews,” said Pete Hellsten, an icing expert, and in the lawsuit that has resulted from the crash Flight 3272 crash, an expert witness for Comair.
“They would say, `Your speeds got slow, and if you’d done it right, those things wouldn’t have happened.’ But it’s inherent in a part of the design of the plane.”
NTSB officials refused to discuss their conclusions until Thursday’s hearing. But the urgency of their work is underscored by a case last March, in which a Westair EMB-120 flying in California experienced a roll upset, the eighth case since 1989. It occurred after the FAA’s mandated changes. The pilots regained control. Icing is the suspected cause.
