NTSB Marks 40th Anniversary of Flight 90 Accident

Chief of the NTSB Safety Recommendations Division shares a blog that discusses what was learned from this high-profile tragedy.

The only good thing that comes out of an aviation accident is that sometimes we learn from it and use this knowledge to prevent future accidents. This message was driven home by an accident involving Air Florida Flight 90. 

On January 13, 1982, the aircraft took off from Washington National Airport (KDCA) in a snowstorm and failed to climb. The airline collided with the 14th Street Bridge and plunged into the Potomac River. Only four of the 79 persons aboard the jet survived. Four people on the bridge were killed and four more injured.

Marking the 40th anniversary of the event this year, the NTSB released a blog titled “Progress Towards Eliminating Airline Icing Accidents” and written by Jeff Marcus, chief of the NTSB Safety Recommendations Division. In 1982, Marcus was employed by the National Highway Traffic Safety Administration and was working in Washington, D.C. In his blog, he recalls the details of the accident and the attempts to reach those who survived the initial impact before they succumbed to the frigid waters of the Potomac. He also addresses how the deadly accident changed the way airframe icing is addressed on commercial airliners.

Aircraft accidents are rarely caused by a single event or decision. The NTSB has gone through this accident with painstaking detail and identified the issues presented so that other pilots can avoid these mistakes and mitigate risks. 

How it Happened

According to the official NTSB report, Air Florida Flight 90 was scheduled to fly from Washington, D.C. to Fort Lauderdale. The flight’s departure was delayed approximately 1 hour and 45 minutes because of heavy snowfall, which temporarily closed the airport. 

At approximately 2:20 p.m., the captain of Flight 90 requested that the aircraft be deiced because he wanted to be ready for takeoff when the airport reopened. The deicing process began on the left side of the aircraft. However, the captain asked that the process be terminated when it became clear that the airport was going to be shut down longer than anticipated, and that the aircraft was number five or six in the departure line. The deicing process resumed at 2:45 p.m. The deicing fluid consisted of 30-40 percent glycol and 60-70 percent water. No final overspray was applied. The outside air temperature ranged from 24 to 29 degrees Fahrenheit.

The copilot was the pilot flying, and as the throttles were brought up, he asked the captain, “That doesn’t seem right, does it?”

At 3:15 p.m., the deicing was completed, the jet closed up, and the jetway retracted to prepare for departure. The captain asked the station manager, who was standing near the main cabin door, how much snow was on the wings and was told there was a light dusting. It was snowing heavily at this time.

According to the NTSB report, it took several minutes and a combination of airport tugs and reverse thrust on the engines to get the aircraft away from the gate because the ground was slick with snow and ice. When Flight 90 pushed back, there were 16 other aircraft awaiting departure. 

At 3:38 p.m., the cockpit voice recorder captured the conversation between the pilot and copilot during the checklist challenge and response: The captain responded “OFF” when the copilot verbalized “anti-ice.” The engine anti-ice system prevents sensors in the engines from freezing which can result in incorrect power readings.

The first officer and captain repeatedly discussed the need for more deicing, both apparently looking at the wings and commenting on the snow buildup, and how trying to keep the aircraft deiced in the present conditions of a moderate to heavy snowfall was a losing battle.

At approximately 3:49 p.m., the pilots discussed an anomalous power reading on the left engine.

The aircraft was cleared for position and holds on Runway 36 at 3:57 p.m. One minute later, they were cleared for takeoff and advised no delay due to incoming traffic. The copilot was the pilot flying, and as the throttles were brought up, he asked the captain, “That doesn’t seem right, does it?”

They continued with the takeoff roll. The CVR captured their comments questioning the low power reading, the aircraft not climbing properly, and finally the copilot stating, “We’re going down.” The captain’s response: “I know it.”

The NTSB determined that the jet used approximately a half mile more runway than was normal for takeoff and never climbed above 350 feet.

The 737 collided with the bridge less than a mile from the departure end of Runway 36. As it was late afternoon, the bridge was heavily congested. The airplane struck six vehicles and tore up approximately 41 feet of bridge structure, including 97 feet of guard rail.

There were hundreds of witnesses to the accident on the ground. Several noted how low the aircraft was and how right before impact, the nose was pitched up 30 to 40 degrees. One witness claimed that sheets of ice fell off the wings of the airplane as it struck the bridge.

The aircraft hit the water and broke into several pieces. In many cases, the seat restraints failed resulting in fatal impact injuries. One passenger survived the impact, but later drowned after passing floatation devices to other survivors. Those who survived the impact noted that the water was so cold they quickly lost the use of their extremities making it difficult to don the life vests or hold on to the wreckage.

What the Report Said

The NTSB investigation identified several errors on the part of the flight crew related to flying in snow and ice:

• Although the outside temperature was well below freezing and it was snowing, the crew failed to activate the engine anti-ice system. Because of this, the engines were not producing the amount of power that was needed for safe takeoff.

• The pilot’s decision not to return to the gate for the reapplication of deicing fluid, although the airplane had waited in a taxi line for 49 minutes during a snowstorm before reaching the departure runway. During that time more ice and snow accumulated on the wings. The flight had already been significantly delayed and the pilot feared even further delay.

• While waiting in line for takeoff, the pilots maneuvered closely behind a DC-9, mistakenly believing that the heat from the DC-9’s engines would melt the snow and ice that had accumulated on Flight 90’s wings. The NTSB noted this action went specifically against flight-manual recommendations for an icing situation and contributed to the icing on the Air Florida jet, as the exhaust gases from the DC-9 turned the snow into a slush mixture that froze on the wings and the engine of the 737.

• Although the crew was aware of the ice and snow on the wings, they decided to take off anyway instead of returning to the ramp for more deicing.

In short, the probable cause of the accident was the flight crew’s decision to take off with ice and snow contamination on the wings and control surfaces, failure to use the engine anti-ice system, and the captain’s decision not to reject the takeoff when there were anomalous engine instrument readings.

Lessons Learned

Marcus’ blog noted that the number of icing related accidents has decreased dramatically over the years, thanks in part to the lessons learned from Air Florida and other events investigated by the NTSB between 1982 and 1997 where aircraft icing was a factor, killing some 265 people. Similar investigations were conducted by the Canadian Transportation Safety Board, specifically two accidents in 1985 and 1989 that killed a total of 280 passengers and crew.

From the data acquired through these investigations, the agencies took a harder look at how the properties of deicing fluid determine its effectiveness, and how airport congestion and the time needed for air traffic control clearance and delays can affect the deicing process, notably how long a deiced aircraft can wait for takeoff before it must be deiced again.

The investigations into ice-related accidents also reviewed the importance of using the deicing engine instruments to get a correct reading on engine power; how the prolonged use of autopilot in icing conditions can mask developing problems with aircraft control until it is too late; and how the accumulation of ice on swept-wing aircraft can result in an uncontrollable pitch up tendency, resulting in a stall. 

It was noted that even small amounts of ice on an airplane wing, roughly that equivalent to medium weight sandpaper, can disrupt airflow and result in a significant loss of lift.

These findings contributed to a revision of FAA certification standards for airplanes approved to fly in icing conditions and an increased stall speed in icing conditions.

In addition, the accident involving Air Florida Flight 90 is used as a teaching tool in many ground schools. 


New to Flying?


Already have an account?