The pilot flying a Cessna Caravan that crashed after takeoff on Oct. 6, 2005, in Winnipeg, Manitoba, violated operational requirements, according to the Canadian Transportation Safety Board’s final report. Among the violations were taking off at a weight greater than the legal maximum takeoff weight and exceeding the time allowed between wing contamination inspection and takeoff. Shortly after takeoff, the Caravan pilot told controllers that she wanted to return to land at Winnipeg International Airport. The airplane stalled and crashed on railway tracks near the airport and caught fire. The pilot died in the crash.
On Jan. 31, 2006, the TSB recommended that Caravan pilots not take off in conditions worse than light icing. The accident occurred in an area of forecast moderate and severe icing conditions.
This accident was one of the most recent of a series of Caravan icing accidents. It has spawned a lawsuit filed by the pilot’s family; TSB and NTSB recommendations; an FAA Airworthiness Directive; changes in Cessna’s Caravan operating parameters; and changes in the way that Morningstar Air Express, operator of the crashed Caravan, conducts flights in icing conditions and freight operations.
The pilot began working for Morningstar Air Express in 2000 and had logged 4,750 total hours and 1,500 in the Caravan. Her most recent training included simulator work in January 2005 and a pilot proficiency check–which included “a satisfactory rating for icing encounter”–the following month. In April, she completed ground training that included ground and in-flight icing, emergency procedures, surface contamination and other subjects.
The day before the accident, the Caravan flew to Winnipeg from Thunder Bay, Ontario, and it was stored overnight in a heated hangar. At 0410 on October 6 the Caravan was pulled out of the hangar, refueled, then moved to another section of the airport for loading for the flight to Thunder Bay.
The TSB determined that the Caravan carried 488 more pounds than permitted for operations in icing conditions, 288 pounds more than the normal maximum takeoff weight.
The TSB investigators believe that the pilot was aware that the Caravan was loaded beyond maximum takeoff weight. Under the Morningstar Air Express self-dispatch system, the pilot was supposed to perform weight-and-balance computations and leave a copy at the point of departure, “where practical,” according to the TSB.
Investigators did not find the form at Winnipeg, but they did find one in the wreckage, with numbers written in the blocks for the Caravan’s basic weight, takeoff fuel and pilot weight. The remainder of the form was not filled out, nor were computations done for takeoff weight or center of gravity, but there were three numbers written in the remarks section: 7,030, 4,842 and 2,188.
The cargo manifest weight was 2,288 pounds. The TSB estimated that the actual cargo weight was 2,474 pounds; mistakes had been made in calculating the tare weight of cargo containers. The 2,188 figure written in the remarks section likely was the weight of cargo that could be carried after subtracting the aircraft’s weight, pilot’s weight and takeoff fuel from the maximum takeoff weight of 8,750 pounds, the TSB explained.
“The investigation did not determine why the pilot accepted the cargo manifest weight of 2,288 pounds when the calculations showed a maximum weight of 2,188 pounds to remain within the 8,750-pound maximum takeoff weight.”
In fact, given the forecast icing conditions, the Caravan should have been limited to a maximum takeoff weight of 8,550 pounds.
Not knowing exactly how the Caravan was loaded, the TSB couldn’t be sure whether the cg was within limits.
Normally, the FBO used by Morningstar pilots received a faxed package of weather information from the Winnipeg Flight Information Centre (FIC), including textual weather and graphic area forecast charts. The agreement between Morningstar and Nav Canada had been amended before the accident so that only textual data, not the graphic charts, was in the faxed package.
The textual weather included METARs, TAFs, notams and upper winds information. The pilot telephoned Winnipeg FIC at 0410, according to the TSB report, “and
indicated that the charts had not been faxed.” The briefer provided information from turbulence and icing charts valid at 0100 and 0700, including icing forecasts for Winnipeg and eastern Manitoba.
The weather was worse than forecast. A post-accident “aftercast” found that “at 0500, there was a large air mass of icing conditions over Winnipeg and eastward with more than a 70-percent chance of producing airframe icing.” The report also noted that “the weather system over the Winnipeg area was moving somewhat more slowly than forecast. As a result of the slower weather system movement, icing conditions prevailed in the Winnipeg area longer than anticipated.”
At 0504, a TAF was issued showing wind 360 at 15 gusting to 25 knots, visibility greater than 6 sm, light snow, scattered clouds at 800 feet agl, broken clouds 3,000 feet, overcast 7,000 feet and temporary conditions of visibility three miles with light snow and ceiling broken at 600 feet.
The charts that the pilot didn’t see predicted, starting at 0700, “extensive areas of ceilings from 400 to 1,000 feet agl [and] local visibility of three-quarters [of a statute mile] in light and blowing snow. The chart depicted moderate mixed icing in cloud from the freezing level to 16,000 feet over southeastern Manitoba, with the western edge of the forecast icing just to the east of Winnipeg.”
At 0531, a special weather observation at Winnipeg indicated visibility four miles in snow and mist, broken clouds 1,000 feet and overcast at 5,300 feet. Airplanes operating at Winnipeg around the time of the accident (0543) were experiencing icing, with pilot reports showing “nil to light” in airplanes ranging from a Piper Navajo to a Boeing 727. Nav Canada received the reports an hour after the accident occurred.
One airplane that left Winnipeg two minutes before the accident provided flight data recorder information to the TSB showing that temperatures aloft were -2 degrees C at 1,800 feet and -3 degrees C at 2,400 feet. The TSB report stated that the “critical temperature range just below 0 degrees C…prevailed in the Winnipeg area at the time of the accident” and that the Caravan “operates at relatively lower airspeeds than most other turboprop aircraft, such that the [ice-mitigating] effects of friction and compression are reduced.” Forecasters don’t take this phenomenon into account, the TSB wrote.
Because the pilot didn’t see the weather charts, “it is possible that the pilot’s understanding of the icing situation was less than complete,” the report noted. “However, the telephone briefing described the forecast moderate icing conditions over Winnipeg and eastern Manitoba, and those conditions prevailed throughout the flight route.”
Cessna recommends the use of ice-adhesion depressants such as Icex II, and the Caravan’s boots were treated with Icex II the day before the accident flight.
Although the Caravan sat outside for more than an hour in the light snow, TSB investigators don’t believe that any ice or snow accumulated on the airplane’s critical surfaces. They found that the pilot did perform a tactile check of the Caravan’s wing 12 minutes before takeoff, seven minutes earlier than required, but could not verify that she checked other critical surfaces such as the horizontal stabilizer, pitot-static system, stall warning heater or propeller blades “as required by the AFM.” The AFM requires takeoff to occur within five minutes of the tactile check.
Even if there was no contamination when the pilot felt the wing, the TSB believes that “it is unlikely that critical surface contamination would have accumulated between the time of the tactile inspection and takeoff. Because de-icing was readily available and pilots were encouraged to use it when necessary, it is unlikely that the aircraft would have departed with contaminated critical surfaces had the pilot noted such contamination during the tactile check. The investigation found nothing that indicated that the aircraft departed with ice or snow adhering to its critical surfaces.”
Regarding the pilot’s assumed choice to use the normal maximum takeoff weight of 8,750 pounds instead of the icing weight of 8,550, the TSB speculated that the pilot might have believed that she would be able to take off in non-icing conditions and avoid icing until the Caravan burned enough fuel to reduce the weight to 8,550 pounds.
After takeoff, the Caravan accelerated to an airspeed below the zero-flap best-rate-of-climb speed, but this was also higher than the 20-degree flap speed. The TSB couldn’t determine whether the flaps were retracted, although Morningstar’s normal procedure was to depart with 20 degrees of flap and then retract them at 400 feet agl. The flaps were found in the retracted position in the wreckage.
“If the aircraft entered icing conditions in cloud with 0 degree flap, the airspeed would have been below the minimum for those conditions. However, if the flaps had been extended during the climb, the 105-knot minimum airspeed would not have applied.”
During the flight, the Caravan’s calibrated airspeed remained at less than 100 knots except for a brief period just after the pilot requested to return to Winnipeg, according to the TSB. “The reduction in aircraft performance began soon after the aircraft entered cloud and was likely the result of the icing conditions.”
The TSB explained that residual ice on areas outside boot coverage can increase the Caravan’s stall speed and that stall speed also increases in a turn (such as a return to the airport) and when de-ice boots are cycled. The Caravan supplement adds that stall speed can increase by 20 knots or more and climb rate decrease 500 fpm or more with airframe ice accumulations.
“These increases in stall speed could approach or exceed the airspeeds maintained by the accident aircraft,” the TSB noted, “and the aircraft would depart controlled flight. The stall could have occurred without warning; airframe icing decreases the stall warning margin and the stall warning may be concurrent with the stall.”
The wreckage indicated that the Caravan hit the ground inverted in a nose-low, left-wing-low attitude. The engine was operating at full power at the time of impact.
Early in the investigation, TSB investigators performed tests in a level-D simulator to try to replicate the Winnipeg accident flight. This was before they had determined that the Caravan was loaded 488 pounds over maximum takeoff weight, so the simulations were done at a maximum weight of 8,809 pounds instead of 9,038 pounds. Even at the reduced weight, in moderate (simulated) icing conditions, no successful return to Winnipeg could be accomplished.
The TSB noted that it could find no evidence that Morningstar pressured its pilots to fly into bad weather or over weight.
In its findings, the TSB highlighted the excess weight of the Caravan, the icing conditions, the pilot’s loss of control at low altitude, weather forecasts that don’t predict the effect of icing on different aircraft types and an incomplete weather package delivered to the pilot.
As a result of the accident, the TSB recommended that Caravans not be allowed to dispatch into conditions worse than light icing. It also recommended that Canadian Caravan operators be required to maintain a minimum of 120 knots in icing conditions and exit those conditions if that speed can’t be maintained.
The U.S. NTSB mirrored those recommendations and added one suggesting that Caravan pilots be required to disengage the autopilot in icing conditions.
After the FAA issued Airworthiness Directive 2006-06-06, Transport Canada made that AD mandatory in Canada. The AD prohibits Caravan operators from continued flight into moderate or worse icing conditions; specifies minimum speeds for icing operations; requires disconnection of the autopilot at the first sign of ice accumulation; and mandates the addition of more icing information in the AFM.
Morningstar Operational Changes
Morningstar Air Express altered many procedures as a result of the Caravan accident, including:
- improved weight-and-balance practices;
- Rosemount Aerospace ice detectors installed on Caravans;
- flights prohibited in any intensity of freezing rain, freezing drizzle, ice pellets, snow pellets,
- mixed conditions and any known or forecast moderate or severe icing conditions;
- pilots required to obtain a graphic area forecast before every flight;
- gross weight reduced to 8,550 pounds or less in icing conditions;
- pilots required to maintain a minimum of 120 knots in icing conditions;
- minimum ceilings and visibilities established for circling approaches in icing conditions.