Cold Realities

January 1, 1999

Cold weather makes aircraft perform differently. The wings and the engines perform better because the air is more dense. But adding moisture to the cold-weather equation makes life tougher for pilots and aircraft alike. Most pilots are aware of icing's dangers (see " Wx Watch: Icing Rules of Thumb," p. 89), but as we shall see, snow can be a hazard as well. Any type of winter moisture can ruin performance and, for those unaware, the pilot's day. While this article is about an air-carrier accident, the lessons apply to all aircraft. If you fly in the winter, cold-weather procedures are essential. Finally, if you have a feeling that something is wrong, you're probably right.

"Palm 90" was the call sign for Air Florida's Boeing 737-200 flight from Washington National Airport to Fort Lauderdale, Florida. Its 74 passengers and five crew were scheduled to depart at 2:14 p.m. EST on January 13, 1982, but moderate snow delayed departure. The airport was closed for snow removal at 1:38 with a scheduled reopening at 2:30. Palm 90's captain requested the deicing crew to spray down the aircraft with deicing fluid in time for the scheduled reopening. Half an inch of wet snow covered the aircraft. The captain stopped deicing upon learning that 11 other aircraft had departure priority and that the airport would not reopen until 2:53.

Around 2:50 deicing was resumed as heavy snow continued to fall. The deicing was completed by 3:10, and at 3:15 the aircraft was closed up and the jetway was retracted. At 3:23, after a six-minute delay, pushback was approved by ground control, but the tug couldn't move the aircraft because of the snow, ice, some glycol deicing fluid on the ramp, and a slight incline. The aircraft engines were started and reverse thrust was applied, but this was unsuccessful. Witnesses stated that snow and slush were blown toward the front of the aircraft; after the engines were shut down and a mechanic inspected the engine inlets, however, he stated that no ice, snow, slush, or water was evident.

A second tug was brought up, and by 3:35 the Boeing was pushed back for engine start. At 3:38, after accomplishing the after-start checklist, the captain responded "off" to the first officer's callout of "anti-ice." Despite this, the crew was thinking about the snow while they sat in the line of departing aircraft. Nine air carrier and seven GA aircraft were awaiting departure ahead of Flight 90.

At 3:40 there was some further discussion about getting deiced again. The first officer (FO) made several comments about the elapsed time since the aircraft was deiced. At 3:46 the FO commented, "Well, all we need is the inside of the wings anyway — the wingtips are gonna speed up on 80 anyway; they'll shuck all that other stuff."

At 3:48 the FO asked, "See the difference in that left engine and the right one?" presumably referring to an engine gauge reading. The captain replied, "Yeah." The FO continued, "I don't know why that's different unless it's hot air going into that right one, that must be it — from his exhaust — it was doing that in the chocks awhile ago...." The FO was referring to the jet exhaust from the preceding aircraft. At 3:53 the FO said, "Boy...this is a losing battle here on trying to deice those things. It gives you a false sense of security — that's all that does."

At 3:57 the crew set the airspeed bug settings to 138 knots for V 1, 140 knots for V R and 144 knots for V 2. The FO asked the captain, "There's slush on the runway — do you want me to do anything special for this or just go for it?" The captain replied, "Unless you've got something special you'd like to do." The FO then said, "Unless just take off the nosewheel early like a soft-field takeoff or something. I'll just take the nosewheel off and then we'll let it fly off."

At 3:59 Palm 90 moved into position on Runway 36 to hold for takeoff and then was cleared for takeoff with no delay because of landing traffic on a two-and one-half mile final. Fifty minutes had elapsed since deicing was completed, and snow had been falling continuously.

The FO began takeoff at 3:59:46. At 3:59:56 the captain commented, "Cold, real cold." The FO replied, "God, look at that thing, that doesn't seem right, does it?" At 4:00:05 the FO stated, "That's not right," to which the captain responded, "Yes it is, there's 80 [knots]." The FO replied, "Naw, I don't think that's right." About nine seconds later the FO said, "Maybe it is," but then two seconds later, after the captain called "one-twenty" (an airspeed callout), the FO said, "I don't know."

Eight seconds after the captain called V 1 and two seconds after calling V 2, the stickshaker (stall warning) activated. At 4:00:45 the captain said, "Forward, forward. We only want 500." At 4:00:50 the captain said, "Come on. Forward, forward — just barely climb." At 4:01:00 the FO said, "Larry, we're going down, Larry," to which the captain responded, "I know it."

Palm 90 struck the northbound span of the Fourteenth Street Bridge and plunged into the ice-covered Potomac River three-quarters of a mile from the departure end of Runway 36. Heavy snow continued to fall, with visibility at the airport varying between one-quarter and five-eighths of a mile. Seventy passengers and four crew members died. Four persons in vehicles on the bridge were also killed.

Ground witnesses agreed that the aircraft was flying at an unusually low altitude, with the wings level and a nose-high attitude of between 30 and 40 degrees nose up.

The findings of the National Transportation Safety Board are extensive and show the complexity of factors involved. This accident illustrates how little information is readily available regarding safe flight operations of light aircraft in similar conditions. While most of us will probably not be operating under these circumstances, the information is instructional. According to the accident report, the aircraft's takeoff roll and acceleration were slower than normal. Under these conditions a 737 should normally accelerate to a liftoff speed of 145 knots in 30 seconds while using about 3,500 feet of runway and then climb at about 2,000 feet per minute. Palm 90 took 45 seconds and used about 5,400 feet of runway to reach 140 knots. There was an initial climb of 1,200 feet per minute, with the aircraft approaching a stall angle of attack shortly after liftoff. The maximum altitude gained was about 200 to 300 feet agl before the Boeing descended into the bridge. Do you know the normal takeoff times and distances for your aircraft? That would help to indicate a developing problem.

Although there was contamination of the wing and slush on the runway that would have retarded the takeoff somewhat, the NTSB felt that the engines were not developing the appropriate power. The poor acceleration and the FO's comments regarding engine power output encouraged them to look further. By doing a sound analysis on the cockpit voice recorder (CVR) it was determined that the engines were not at takeoff power. The appropriate engine pressure ratio (EPR) should have been 2.04; if the pressure probe were blocked by ice, however, the engine gauge would show 2.04, while the actual EPR was around 1.70. The sound level on the CVR matched an engine output at 1.70 EPR.

The crew's failure to turn on engine anti-ice resulted in the freezing of the probe and an erroneous EPR, leading to a reduced thrust lever setting and significantly less thrust. There were several other cases where crews had rejected a takeoff because of low EPR settings when other supporting engine gauges such as fuel flow and EGT showed below-target levels. Upon further research the Board found that, even with engine anti-ice on, there was a possibility that the 737 engine probes could ice at low power settings while sitting on the ground — and made some recommendations to alert crews and air carriers to the problem.

Yet even with the reduced power setting, the 737 — according to test data — should have been able to attain a 1,000-feet-per-minute climb and gradual acceleration. Ground witnesses and surviving passengers, as well as the pilots' conversation on the CVR, indicated that the snow depth on the wing at takeoff was between one-quarter and one-half inch. Every pilot knows that a contaminated wing will not produce maximum lift. Changes in contour shape and roughness of the surface will cause the airflow to begin to separate from the wing at a lower angle of attack than normal and cause a reduction of lift.

The report goes on, literally for pages, analyzing the effects of wing contamination and the Boeing 737's tendency to significant pitch up in conditions where the wing was presumably contaminated by snow, sleet, or rain in near-freezing conditions prior to takeoff. Excerpts from the flight manual concerning proper engine operation and aircraft deicing are extensive compared to what is typically in the pilot operating handbook of light aircraft approved for flight in icing conditions.

The NTSB concluded that neither the low thrust during takeoff nor the presence of snow or ice on the aircraft alone would likely have led to the crash. 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. A concluding paragraph was very quick to point out that aircraft have been lost solely because of airfoil contamination and advised not to bet the farm (my words) that a pilot could expect to power out of a bad situation. Simply put, if the crew had firewalled the thrust levers, chances are very good that Palm 90 would have flown. This was contrary to most crew training practices governed by concern for engine longevity.

The Board also had concerns about the effectiveness of the anti-ice protection afforded by spraying down the aircraft. In light-aircraft operations most airplanes are not equipped for flight into icing conditions, so takeoff into heavy snow and icing is not an issue. For pilots who do fly aircraft with icing certification, a predeparture spraying may or may not buy much protection. The NTSB found that the practices were not uniform. There were no fewer than three recommended procedures: one by Union Carbide, which made deicer fluid; another by Trump, which built the deicing vehicle; and a third from American Airlines, which had the contract to deice Air Florida aircraft at Washington National.

The NTSB reminded pilots that even though there may be residual anti-icing effect after having been properly deiced, the only way to ensure that the aircraft is ready for takeoff is to observe the wings and tail just prior to beginning the takeoff roll. Analysis of the deicer vehicle found that a nonstandard nozzle had been used and was dispensing about half as much deicer fluid as the mixture controls indicated. While other aircraft successfully departed under the heavy snow conditions, the NTSB was unable to reach a conclusion on whether the nonstandard deicing was a factor in the accident.

Flight crew performance, according to the NTSB, was not optimal. At no time did a crewmember leave the cockpit to check the condition of the wings. The use of reverse thrust to back out of the gate was contrary to the operations manual, yet the crew tried it anyway. While the mechanic who inspected the engines said that he could see no ice, slush, or snow, the Board hypothesized that additional moisture could have blown onto the wing leading edge. This could have subsequently frozen, leading to control problems during takeoff.

The failure to use the engine anti-ice is a key factor in this accident. The flight manual is very clear about the need for engine anti-ice under these conditions. The crew failed to appreciate this, although they constantly discussed the snow on the wings. Additionally, a Boeing operations manual bulletin recommends that aircraft should maintain a greater distance behind other aircraft when taxiing in areas of ice or snow. This is to prevent the blow-off from melting and refreezing on the following aircraft. Palm 90 did not follow this advice.

The decision to take off with visible snow on the aircraft was in violation of the FARs (91.527) and good operating practice. The NTSB noted, "The crew's decision to take off with snow or ice on the wings was a direct cause of the accident." They also stated that this was not an isolated incident, but rather a common occurrence, and that pilots failed to understand the risk. Every year in general aviation we have accidents when pilots challenge aerodynamic fact involving prior successful takeoff experience with a contaminated wing.

The least bit of uncertainty during a takeoff is a mandate to abort. The first officer's comments regarding engine performance should have been a signal to both crew members that something was amiss — or at least worth checking on the ground before committing to flight. This is excellent advice for GA pilots. That Palm 90 was advised to make an immediate takeoff because of landing traffic may have predisposed the crew to continue rather than reject and cause another aircraft to go around. Remember that the departing aircraft owns the runway until becoming airborne and rejecting a takeoff for cause or suspected cause is always good form.

Finally, the NTSB discussed flight crew training and noted that the captain had relatively little experience in jet transport operations and very little in winter operations. An investigation showed that he had made only eight arrivals or departures where conditions were conducive to icing. While his flight hours were high compared to those in general aviation (8,300 total, with 2,322 while employed by Air Florida), they were considerably less than the airline industry norm. In the early 1980s trunk carriers were upgrading FOs to captains after an average of 14 years. This pilot upgraded in fewer than three years, so there was little time to acquire winter operations experience. The crew did learn about cold-weather procedures in a classroom and during initial as well as recurrent checks, but the NTSB felt that this was not a substitute for more formal training to emphasize winter's hazards.

For those of us who fly light aircraft in the winter there are many lessons to learn from Palm 90.

  • Confirm proper power setting and use supporting instruments (EGT and fuel flow) as needed.
  • Be sure that the wing and other control surfaces are free of snow, ice, and frost when reaching the end of the runway. If a long delay is anticipated during a snowfall, coordinate with ATC so that you can be deiced and depart immediately.
  • If something seems wrong during the takeoff or the aircraft is not accelerating normally, reject the takeoff and figure out what is wrong.

See also the index of "Safety Pilot" articles, organized by subject. Bruce Landsberg is executive director of the AOPA Air Safety Foundation.