Air Florida Accident First of Two in Sequence
Please understand that my motive in saying what I’m saying (and what I’ve said previously), is in no way intending to impugn the fine reputation or the integrity of the NTSB or of the dedicated and professional employees at the Safety Board. My only motive has been a continuing effort to describe the actions of the flight crew from a slightly different perspective while providing what I believe to be clarifying information regarding the cause of the accident and, to the best I am able, set the record straight– as much as one can this long after the fact.
I should also say, up front, that I, too, continue to have questions for which I still do not have answers. There is not a winter flying season that comes and goes that I do not wish that the flight crew of that Air Florida flight had been more inquisitive about the condition of the wings. Perhaps if one of them had taken a few moments to observe the wing surfaces on a walk back through the cabin at some point prior to the takeoff, he would have noticed something … anything … that would have triggered a different course of action. But that didn’t happen. So, what I’m left with is what happened … the results of the investigation … and a long-time consideration of what that investigation revealed to me. For simplicity, I would like to address some issues I believe are important, and I’ll do that in chronological order, beginning with the airplane de-icing.
The De-icing. The NTSB Aircraft Accident Report described an “analysis of the deicer vehicle found that a nonstandard nozzle had been used” and was dispensing deicer fluid with a mixture different from what was expected. The accident report states “the mixture dispensed differed substantially from the mixture selected…” such that the “deicing fluid in solution was about 18 percent rather than 30 percent.” The report further states that the right side of the aircraft was “deiced with 100 percent water and a final overspray applied with a 20 to 30 percent deicer to water solution selected.” The “20 to 30 percent” mixture for the overspray was the selected value, not the mixture that was actually applied. Standard deicing procedures called for deicing and overspray to be conducted with 30 to 40 percent solution of glycol in heated water. What the report does not contain is that the nonstandard nozzle dispensing fluid at 18 percent glycol mixture was a reasonably accurate measurement for the minimal flow of fluid from the deicing wand at a nominal, or minimal, flow rate. However, when the operator increased the flow rate, all of the additional volume flow came from the heated water tank, and none from the glycol tank – effectively lowering the solution percentage by an unknown but substantial proportion. From my perspective, deicing over half of the airplane and the wing surfaces with 100 percent hot water and over-spraying the entire aircraft with a maximum concentration of 18 percent glycol (and very probably significantly less depending on the volume flow at any given point in the overspray process) cannot simply be catalogued and filed away. I believe this to be the pivotally significant action in the series of actions leading up to the accident and was, singularly, the accident cause.
The Takeoff Clearance and the Absence of an Aborted Takeoff. The accident report goes into some detail about the lack of assertiveness on the part of the F/O and the lack of receptiveness on the part of the Captain during the initial stages of the takeoff roll when the F/O apparently noted some engine indication anomalies. Many have quoted from the CVR transcript indicating that during the takeoff roll there were a number of times the F/O questioned the accuracy of the engine instruments. My question is how do the readers of the transcript know that the F/O was referencing the engine instruments? If it was an “anomaly” with the engine instrument readings, the logical question that the casual observer would make is “why didn’t they simply abort the takeoff?” In fact, the accident report cites 3 factors as the probable cause of the accident, with the third one being “the captain’s failure to reject the takeoff during the early stages when his attention was called to anomalous engine instrument readings.” The report also states that the investigation “considered the possibility that the captain was aware of and concerned about the decreasing separation between his aircraft and the aircraft landing behind him.” However, the report concludes that this decreasing separation “would likely have become a factor only after the landing aircraft reported ‘over the lights.’” I’d like to look at that conclusion for just a moment.
Pilots sit at the arrival ends of runways multiple times each day – day in and day out. They watch arrival after arrival – big airplanes, little airplanes, fast airplanes, and airplanes not so fast. What they do is get a “feel” for the timing involved. I know of no pilot who has ever taxied into position on an active runway, knowing there is an airplane on final, who has not had at least a slight “tingle” down the back of his or her neck. That tingle grows until takeoff clearance is received, the throttles are pushed forward, the airplane accelerates, and the crew rotates into the air. Then, and only then, does the tingle subside. The investigators concluded that the Captain would not have considered the approaching airplane to be worrisome until that crew reported “over the lights.” I respectfully disagree with that conclusion; and I know of few, if any pilots who would accept that conclusion. Would you?
The traffic that was referred to was Eastern flight 1451, a B-727. I know that the final approach speed is dependent on several factors (weight, flap setting, etc.), but because the tower operator asked the Eastern crew to fly “reduced speed,” I’ll use a speed of 130 knots. With a head wind of approximately 10 knots, the ground speed would have been approximately 120 knots, or 138 miles per hour. At that speed it would take the airplane approximately 64 seconds to cover the “two and a half miles,” or the 13,200 feet, reported by the tower operator. From that time of that report (59:28) until the F/O took control of the throttles (59:46), 18 seconds had passed. It was 16 seconds later (00:02) when the F/O first indicated that there was something that he questioned. This was now 34 seconds after the tower’s notification that landing traffic was “…two and a half out for the runway.” When the Captain called “80-knots” (00:09), the airplane had probably reached a position 1000 to 1500 feet down the runway. The investigators believed that the F/O was concerned about engine “anomalies,” but I believe he was confused about physical throttle position being different from what he had become used to recognizing. He was used to having his arm and his hand at a position that “felt familiar” (i.e., muscle memory), but when the instruments told him that the proper position had been reached, he recognized “something” different but didn’t recognize that his arm and hand were in a slightly different position – and this is what was confusing for him. I believe that it is quite likely that both crewmembers were attempting to identify anything out of the ordinary – but I contend that the engine instruments were reading what the crew expected them to read – and it is only after-the-fact that we’ve learned that the PT2 probes being blocked would have allowed the EPR gauges to indicate the desired setting for takeoff with the engine thrust actually set to a lower value. The other engine instruments were likely close to what should have been “normal” but, and I think significantly, those indications were steady and symmetrical. The flaps were properly set. The speed brakes were stowed. Everything that either crew member could see seemed normal.
The investigators believe that once the F/O voiced those concerns the Captain should have aborted the takeoff. Before agreeing or disagreeing with that belief, I think the significant question should be … where was the B-727 at this time? Assuming the airspeed of the B-727 was 120 knots and that this speed remained constant, and that its position was accurately relayed by the controller to the departing flight crew, at this 34 second mark, mathematics tells us that the B-727 would have been approximately 6300 feet from the threshold. That’s well over a mile out … but, that is ONLY if the assumptions are correct. To some – perhaps to many – that distance may seem to be adequate spacing for the departing airplane to abort with little or no problem. However, the fact is that a scant 2 seconds after the B-737 reached 80 knots (which happened at 00:09) the B-727 crew reported “cleared to land, over the lights” (which happened at 00:11) The approach lighting system extends 2,400 feet from the threshold. If the B-727 was indeed “over the lights” as they, themselves reported, they were something less than 2,400 feet from the threshold – still maintaining that constant 120 knots. Either they were flying a lot faster than my assumption or the tower operator provided inaccurate information to the departing crew. At that point, the B-737 was probably less than 2000 feet down the runway. Again, estimates vary, but there are some who believe that the B-727 actually landed prior to the B-737 lifting off … meaning that both aircraft were on the runway at the same time! Had the B-737 crew aborted, I believe there is at least a reasonable possibility that two aircraft would have been involved in an accident on the runway, and the probability of that accident occurring goes up dramatically, second by second.
I probably should point out that the both the aircraft immediately preceding the accident B-737, “Apple 58” (a New York Air DC-9) and the aircraft immediately following the accident B-737, “Six Eight Golf” were both in the takeoff que for approximately the same amount of time and both aircraft took off from the same runway and departed without incident.
The Airplane Performance During the Takeoff. It would come as no surprise to anyone that a lower engine power setting would result in a longer than normal takeoff roll given that all other parameters were the same. In this case, as I’ve said previously, I have little doubt that the accident airplane PT2 probes were blocked – most likely blocked with ice. I also believe that this resulted in erroneous EPR readings in the cockpit. I believe that the flight crew set the takeoff EPR at 2.04 and that, because of the blocked PT2 probes, the actual engine thrust was set to approximately 75% of maximum. Undoubtedly, this increased the ground roll. But the point that goes unmentioned in this discussion is that the airplane was certificated to be able to accelerate to decision speed, or V1, (and I recognize that airplane certification is based on all engines operating at full power up to that V1 speed), experience the complete failure of one engine, and be able to continue the takeoff safely. What happened in this case was that the airplane accelerated down the runway with both engines producing approximately 75% power. Again, I acknowledge that the takeoff roll would be longer to get to this point. However, once the computed V1 speed was reached, regardless of how long it took to get there (downhill roll, rubber-band, whatever…), from that, “V1” point forward, the airplane should have been able to fly on one engine at 100% power. From the CVR transcripts we know that the V1, Vr, and V2 speeds were noted to be “thirty eight, forty, forty four” – meaning 138, 140, and 144 knots, respectively. In fact, after reaching “V1” (and we know this speed was reached, as the Captain’s “Vee One” call-out is noted on the transcript) this airplane had 2 engines operating at 75% power, which is clearly 50 percent more power than should have been required – and as an added benefit the power that was being produced was produced symmetrically. We also see on the CVR transcript that in addition to the “V1” speed being reached, we know that “V2” speed was also reached, and was also confirmed by the Captain’s callouts. We should also recall that V2 is generally 20% above the stall speed for that aircraft weight and configuration. Yet after having reached a speed of more than V2 (actually 150 knots), as confirmed by the Flight Data Recorder (FDR) tracings, with symmetrical power of two engines, each producing 75 of maximum power, the airplane failed to fly. Why?
Many people instantly respond with the assumption that the airplane had added weight due to the accretion of ice. Today we know that if an airplane the size of a B-727 is completely covered in ice, the total weight of that ice is something on the order 180 to 350 pounds – somewhere between 1 and 2 additional passengers. So, if there was additional weight of ice, that small increase should not have had such significant detriment to the performance of the airplane. However, it would be interesting to note where it was that this ice came from. Recall that the aircraft that was immediately in front, and the aircraft that was immediately behind, the Air Florida B737 in the takeoff line, took off with no problems. All three airplanes were deiced at approximately the same time. All three airplanes were exposed to the elements for approximately the same time. Yet two of the three flew successfully, and one did not. If it wasn’t weight, what prevented the airplane from flying? In my opinion, it aerodynamically stalled.
Pilots fly airplanes according to airspeeds, not power settings. Airplanes fly or fail to fly because of the wing moving through the air at sufficient or insufficient speed. Did this accident airplane have insufficient airspeed? No. The airplane reached a speed slightly higher than V2, computed to be 144 knots. At 20 percent above the stalling speed, this meant that the airplane would have stalled at 120 knots. In that any additional weight due to ice accretion seems not to be the answer, why did the airplane stall at 24 – 30 knots above stalling speed? In my opinion it was because of the pitch attitude of the airplane.