Air Florida Accident Second of Two in Sequence
The accident report describes “the Boeing 737’s tendency to significant(ly) pitch up in conditions where the wing was presumably contaminated by snow, sleet, or rain in near-freezing conditions prior to takeoff.” This phenomenon directly relates to asymmetrical lift. Asymmetrical or differential lift is not new, and is, in fact, used every day by every airplane that uses roll control spoilers. Asymmetrical or differential lift is used to assist not only in rolling the airplane, but in assisting in maintaining coordinated flight by compensating for adverse yaw. Here, however, in the B-737’s tendency to pitch up, the asymmetry comes not laterally but longitudinally. With the outboard portion of the wings sufficiently deformed by leading edge ice accretion, they provide less, or no lift, leaving the inboard portions of the wings (also farther forward) producing an asymmetrically large lift differential. As the airplane accelerated during the takeoff roll, the pilot, to ensure nose gear contact with the runway surface, normally would hold the controls in a forward, or nose-down, position. However, as the rotation speed is approached, the pilot is instructed to move the control column to a neutral or slightly aft of neutral position in preparation to initiate the rotation to the takeoff attitude. My opinion is that when this action occurred, or immediately when beginning the rotation, as the wing was rotated from an essentially negative angle of attack to a positive angle of attack, the inboard portion of the wing, producing virtually all of the lift being generated and that forward of the center of gravity, rotated the airplane, without the assistance of the pilot, and continued that rotation to a point where the wing was aerodynamically stalled.
An “aside” here, I think, may be appropriate. I know for a fact that the Air Force KC-135, Strato-Tanker airplane (essentially a hybrid of the original B-707 and the B-720)was equipped with roll control spoilers. However, additionally, the KC-135 had spoiler control valve switches (both of which were guarded switches) located just under the center of the glare shield. The left switch activated the inboard spoiler control valve and the right switch activated the outboard spoiler control valve. These valves were checked prior to each and every flight. What was their use? Well, that’s a questionable call about which many would not be willing to offer an opinion. Personally, I believe it was a safety feature that was made available in case loading of fuel in the massive fuel tanks located both forward and aft of the center of gravity created a longitudinal shift in that center of gravity that might needed to be managed. In fact, all instructors were trained on their use, as during takeoff, a pilot could reach up under the glare shield, open the left guard, depress the switch (thereby cutting off hydraulic fluid to the inboard spoilers) and without doing anything with the control column, reach down an raise the speed brakes (now, because of deactivating the inboard spoilers, only the outboard spoilers would respond – raising only the outboard spoilers) to create a differential lift on the swept wings of the KC-135, where the inboard portion of the wing is farther forward than the outboard portion of the wing – resulting in the outboard portion of both winds having the lift “spoiled” and the lift on the inboard portion remaining unchanged, the airplane would rotate smartly – so smartly, in fact, the speed brake extension had to be moderated to avoid driving the tail of the airplane onto the runway prior to becoming airborne. Asymmetrical lift is quite powerful.
In the accident B-737, if this condition existed, it is very likely that continued forward control column movement and continued forward pressure on the control column would be insufficient to bring the nose down. We see on the CVR transcript that the Captain is urging “Forward, forward.” “Easy.” “We only want Five Hundred.” “Come on forward.” “Forward.” “Just barely climb.” What do you think was going on in that cockpit? I think that the flight crew had the control column fully forward and the airplane was not responding the way they had expected it to respond. Speculation has been raised that trimming the stabilizer to its maximum nose down position may have been able to provide some additional nose down elevator authority. However, this action would have been completely non-intuitive to a pilot never having experienced this phenomenon previously, and being airborne for only 22 seconds would probably not have allowed sufficient time to allow the trim to achieve sufficient movement to make enough of a difference.
Interestingly, the accident report also states “there were other incidents similar to this one reported, in which the crew was able to overcome the contamination, but they needed the proper thrust level of 2.04 EPR.” I believe this statement inaccurately and unfairly characterizes the recoverability of any B-737 that has aerodynamically stalled due to a significant pitch up resulting from contamination of the wing by freezing precipitation. In support of this belief I call your attention to an occurrence on the same day as the Air Florida accident, with another B-737 operator’s experience out of Oslo, Norway. While perhaps one may be able to conclude that flight and ground personnel in Washington may misunderstand snow and ice conditions, there is no way that anyone is going to believe that pilots or ground crew in Oslo, Norway, are going to misunderstand the necessities or the procedures for deicing an airplane. However, on this day, the B-737 departing from Oslo was exposed to a snowstorm during taxi out and takeoff, and the problems experienced immediately after takeoff were chillingly identical to the Air Florida accident.
Because of the storm in Oslo that day, ice was built up on the wing leading edge. But, because of the nature of the storm, the build-up was asymmetrical. As in the Air Florida airplane, at takeoff there was a resulting pitch-up, but in Oslo, it was a rolling pitch (due to the asymmetry of the deformation). Even though the crew slammed the throttles to the firewall immediately, used full opposite aileron and full opposite rudder, they were unable to control the pitch/roll of the aircraft. Fortunately, because of the radical bank angle (approaching 90 degrees), the nose dropped back down to the horizon, and with the throttles fully forward, the airplane began to accelerate. As the airplane accelerated, lift was produced over the outboard portions of the wing, including the ailerons, and that allowed the crew to roll back to level flight. The crew did recover the airplane; but the recovery was below 100 feet above the ground. Had they not had this bit of inconsistency from Mother Nature, resulting in the asymmetrical lift on the wings, providing the roll, and allowing the nose to fall, the world would have seen B-737s on opposite ends of the world crash on the same day from the same problem.
I contend that this problem is a very slight accumulation of ice on the leading edge of the wing causes asymmetrical lift, longitudinally, resulting in a nose up pitching moment. The relationship is that the greater the accumulation of ice, the more asymmetrical lift, and the greater the pitch tendency. Perhaps there were some incidents where the crew was able to recover the airplane by, among other things, advancing the power. But in Washington and in Oslo, the ice build-up was sufficient to result in uncontrollable pitch. Even with both throttles full forward throughout the event, the Oslo aircraft was not recovered by assuring the engines were producing at least 2.04 EPR. In fact, had the differential lift been symmetrical, and the aircraft not rolled over on its own, full thrust on the under-slung engines on the B737 may have exacerbated the pitch tendency.
Is this what happened to the Air Florida airplane? I believe it is. Look at the CVR and the Flight Data Recorder (FDR) readouts. The CVR has the Captain calling “V2,” and scarcely 2 seconds later the transcript describes “sound of stickshaker starts.” When listening to the actual recording of the CVR there is little doubt that there are sounds of severe buffeting as well as the stick shaker operating. The airplane had gone well past the stall warning (the stick shaker) and had gone well into the stall buffet. Opinions differ as to what pitch angle would have been required to get into the stall buffet – but they range from 22 to 24 degrees of pitch. Think about that for a minute. In just over 2 seconds the airplane goes from level attitude to at least somewhere between 22 and 24 degrees of pitch. That is a rate of rotation in the neighborhood of 12 degrees per second. Recall that the normal rotation rate is 3 degrees per second; and the F/O, who was at the controls, had indicated he was going to “takeoff the nose gear and then just let the airplane fly off by itself.” The F/O wanted to deliberately limit the rotation to just getting the nose gear into the air, and the airplane was rotated at 12 degrees per second or more. Does that indicate anything inconsistent to you?
The bottom line is this. I have no explanation as to why one of the crew did not look more closely at the wings prior to departure. I don’t know why the crew was lead into accepting a clearance that would have been “challenging” on a clear day with unlimited visibility. I regret that we all didn’t know more about deicing effects and “hold-over” times. I don’t know why we were not made aware of the tendency of the B737 to have pitch-up and roll-off problems in winter weather conditions. What I do know is that I am grateful that the errors made that day, by all who made them, have been talked about and hopefully will be used to educate others and save lives. I also know that two competent aviators, former colleagues of mine, and extremely nice young men, lost their lives, as did most of the rest of those on board the airplane along with four motorists who died on the bridge that fateful afternoon.
My only goal in providing this information is to provide at least an alternative to the understanding that many have had for a number of years. I believe that the real cause of this tragic accident wasn’t because these young men forgot to use, or chose not to use, engine anti-ice. The acceleration rate was immaterial. It was not because they attempted to power-back out of the gate. They did not do that. It was not because they chose to “deice” their airplane by deliberately taxiing too close behind other aircraft “jet wash.” I don’t believe they did that. It was not because they disregarded the snow that accumulated on their aircraft while waiting to takeoff. I believe they placed their confidence in the deicing / anti-icing process. It was not because these professional pilots lacked critical knowledge of cold weather and de-icing operations. (I probably should point out that prior to being hired at the airline, the F/O who was at the controls of the accident airplane, had been an F-15 pilot stationed at Minot, North Dakota – where, as everyone probably knows, the residents do have something more than a nodding acquaintance with ice and snow.) I am firmly convinced that the accident was caused by the fact that the airplane was highly susceptible to very minor amounts of ice deformation of the leading edge of the wing; that it was caused by the fact that they were deiced with an inadequate system that coated their aircraft and engine intakes with water – and only a small token of the glycol solution on which they depended.
I believe that once this crew pushed the throttles forward with the intent to takeoff, they were doomed. The only way that an accident could have been avoided at that point, was to have kept the airplane on the ground until a sufficiently higher airspeed was reached prior to initiating the rotation. Unfortunately, not only did the crew not know that this would be necessary, they wouldn’t have known the “magic” airspeed number. Even if they had pushed both throttles all the way to the firewall from brake release, and then rotated at the computed rotation airspeed (as they did) the airplane would have performed in exactly the same way. It would have uncontrollably pitched up to at least the 22 – 24 degree attitude; likely more given witness statements … where some said they saw “the aircraft was flying at an unusually low altitude with the wings level at a nose-high attitude of 30 degrees to 40 degrees before it hit the bridge.” It would have entered the same deep aerodynamic stall. The flight crew would have been unable to bring the nose down aerodynamically. Unfortunately, in that condition, full thrust on both engines would have been insufficient to maintain flight. All who were affected by this tragedy were victims of longitudinal differential lift. The flight crew did not have the luxury of time to analyze, consider, and choose accordingly.