October 2004 Volume 47 / Number 10
Landmark Accidents: Much More Than a Bad Day
Basic skills could have saved a senator's life
Pilots have to demonstrate mastery of basic airmanship and as they move up the ranks in certificates and ratings, flight checks provide an opportunity to evaluate those skills. Professional pilots flying for charter or the airlines are under closer scrutiny than those who fly for personal reasons. This accident review is a reminder that no matter how long one has been flying or the certificate level, the basics of flight are always with us. Since this accident involved a U.S. senator, it attracted considerable attention and provides detailed insight into the pilots' backgrounds and some breakdowns in flight operation management. The lessons apply to all pilots regardless of whether they fly for hire or not.
The flight was a routine charter on October 25, 2002, in a King Air A100, equipped for flight into known icing conditions, from St. Paul Downtown Holman Field to Eveleth-Virginia Municipal in Minnesota about 150 miles north. The purpose was to transport Sen. Paul Wellstone, a rising star in the Democratic Party, and some of his family and staff to a funeral. The pilot called the flight service station (FSS) for an abbreviated weather briefing shortly after 7 a.m. The specialist advised of airmets for IFR and icing over the entire route. IFR conditions were reported throughout central and northern Minnesota and ceilings were running between 800 and 1,000 feet broken and 1,300 to 1,600 feet overcast. At Eveleth there was a 1,000-foot scattered layer, overcast at 2,000 feet with visibility as four statute miles in light snow. Nearby stations had ceilings between 300 and 600 feet with visibilities between one and four miles in light snow and mist. Duluth International, the alternate, indicated an overcast of 500 feet and visibility of two statute miles in light snow and mist. The terminal forecast called for ceilings of 500 to 900 feet and visibility of one to three statute miles in rain or snow. The pilot stated, "You know what, I don't think I'm going to take this flight."
It was certainly the pilot's prerogative to cancel, for whatever reason, but this type of weather is flown routinely in turbine equipment. After some uncertainty, a call to the senator's scheduler to plan for a delay, and a second call to the FSS, the pilot decided to fly. On the second briefing the controller provided some pilot reports of icing in the cloud tops but nothing unusual for late-fall operations. The pilot also spoke to the charter company's dispatcher and attempted to reach the chief pilot and director of operations to get their thoughts on the flight but was unable to contact them. An IFR flight plan was filed and the flight departed St. Paul at 9:37 a.m. and was cleared direct to Eveleth at 13,000 feet.
Upon leaving the Minneapolis Center frequency the controller noted that the last icing report "was from a Saab [Saab 340 airliner] that descended into Duluth, had light rime ice.... Just about an hour ago, a DC-9 had moderate rime between niner thousand and one-one thousand." The copilot acknowledged and at 10:02 a.m., the flight was instructed to descend to 4,000 feet at pilot's discretion.
At 10:04 a.m., the copilot contacted the Duluth Approach controller, indicating that he had the current Eveleth weather and would like the VOR Runway 27 approach. The controller advised the crew to expect vectors. At about 10:06 a.m., the controller asked what they intended to do in the event of a missed approach. The pilot responded, "Well, let's hope we don't have that. If we do have a missed approach, we'll go up and circle and figure this out. I'll hold at the VOR."
At 10:09 a.m., the flight reported leaving 13,000 feet for 4,000 feet. Radar indicated that the King Air was approximately 34 miles south of Eveleth and the flight was cleared down to 3,500 at pilot's discretion. At 10:15:17, the radar showed a descent through 6,200 feet at 194 knots when the controller instructed the crew to turn right to a heading of 050 degrees. At 10:17:21, a left turn to 360 degrees was issued. By 10:17:35, the flight had leveled off at 3,500 feet and was heading northward with the airspeed decreasing through 190 knots. So far, everything had been routine but it was about to unravel quickly.
At 10:18:13, when the airplane was less than about one-half mile south of the VOR Runway 27 approach course, the controller advised, "One-zero miles from the VOR, turn left heading three-zero-zero, maintain three thousand, five hundred until established on the final approach course, cleared for the VOR Runway Two-Seven approach, Eveleth." The copilot acknowledged the instruction at 10:18:31 and shortly thereafter, the airplane began turning left while maintaining 3,500 feet and slowing to 164 knots. It overshot the approach course, based on a late turn by ATC and possibly by a slow response from the crew. It traveled almost one mile north while turning toward the course and then established a ground track of 262 degrees.
At 10:19:12, the controller cleared the King Air to the advisory frequency and that was the last transmission. The airplane began to descend from 3,500 feet and the airspeed increased from 155 knots to 170 knots with vertical speed increasing to 1,400 feet per minute. At 10:20:06 the airplane passed through (south of) the approach course five miles east of the Runway 27 threshold and made a slight right turn. The airspeed and vertical speed began decreasing and the ground track settled on 269 degrees until the end of the radar data at 10:21:42. The final two radar returns indicate the airspeed had dropped to 76 knots. The King Air stalled and collided with trees about 1.8 nautical miles southeast of the approach end of Runway 27. Both pilots and all six passengers were fatally injured.
The charter company had hired the 55-year-old ATP-rated pilot about 18 months earlier. In April 1989 the pilot failed a flight check for his airline transport pilot certificate because of unsatisfactory performance in area arrivals, ILS approaches, normal/abnormal procedures, and judgment. He passed the test on August 4, 1989. The pilot flew for a regional airline for several months in 1990 and then resigned to serve 15 months in prison for mail and wire fraud.
According to the charter company employment application, the pilot worked as a registered nurse for four different employers between March 1992 and March 2001. The pilot had approximately 5,116 total flight hours, 598 hours of which he flew with the company, 200 hours of which were as pilot in command (PIC) in King Airs. He had flown approximately 101, 53, and 36 hours in the 90, 60, and 30 days, respectively, before the accident. His last recurrent ground training occurred on April 23, 2002. Since completion of his last simulator training on August 21, 2002, the pilot had flown 32 flights as PIC in King Airs and 11 as copilot in Cessna Citations. Six of the 21 King Air flights were conducted in the A100, including one flight the day before the accident in the accident airplane. His last proficiency and line checks were conducted on October 23, 2002. FAA and company records showed no accident or incident history and no enforcement or disciplinary actions.
The charter company's lead ground instructor felt that the pilot was average on systems knowledge but several company pilots indicated that his flying skills were below average. He was described as "very meticulous, by the book, calm, laid back, friendly, and cheerful." Several copilots thought the accident pilot was well liked because of his reputation for letting them fly, but they were uncertain of his skills since they handled the controls consistently.
One pilot expressed concerns about the pilot's abilities, monitoring capabilities, and tendency to become distracted. Several felt that the accident pilot was not particularly assertive but others disagreed. One described him as "too timid to be a pilot." Another King Air pilot reported that he had taken the controls away from the accident pilot during an instrument approach because of his inability to maintain altitude. A copilot also assumed command when during level flight in instrument meteorological conditions (IMC) the pilot allowed the airplane to enter a 45-degree bank and a 1,000-fpm descent.
Another copilot indicated that two months before the accident the pilot did not have his navigational radio tuned to the proper VOR during an approach, which caused the pilot's course deviation indicator (CDI) to provide erroneous indications during the entire approach. The copilot, who was flying, tuned to the right frequency and completed the approach without incident. The copilot later explained to the accident pilot why his CDI was not indicating properly.
According to the charter company's director of operations and the company pilots who witnessed these incidents, none of them were ever reported to company management. Charter company records showed that the pilot and copilot had flown together four times in King Airs and that the pilot had landed at the Eveleth airport on four previous occasions, twice in King Airs and twice in Citations.
On October 22, 2002, three days before the accident, the pilot flew Wellstone from St. Paul to Greater Rochester International Airport, Rochester, New York, in a King Air 90. The copilot, who was the flying pilot, stated that during takeoff, instead of activating the yaw damper switch, the pilot activated the adjacent autopilot switch, which caused the airplane to pitch down. The copilot stated that he (the copilot) immediately applied back-pressure and then disconnected the autopilot, which caused the airplane to pitch up erratically before it returned to normal climb. After everything was stabilized, the copilot explained to the pilot that he had engaged the autopilot instead of the yaw damper and that the pilot had replied, "Oh, that could have been pretty bad."
The pilot had successfully completed a six-month proficiency check with the chief pilot two days prior to the accident and, according to his wife, had adequate rest the night before the accident. While the pilot did take an early morning flight the day before the accident, when he let the copilot fly all the legs, his actions seemed normal. Working a second job as a nurse, he went to work in the afternoon for several hours before returning home and retiring at about 9:30 p.m.
The 30-year-old copilot was hired by the charter company in February 2001 and held a commercial pilot certificate with single- and multiengine land and instrument airplane ratings. While unknown to the charter company, logbook entries and other records indicated that the copilot worked for a skydive operator from October 1998 through March 1999. He was terminated when he could not meet standards for flying a Cessna 182. In February 1999, an airline hired the copilot to provide instruction on Airbus A320 systems and cockpit procedures during ground school. The copilot was not able to successfully complete his initial training despite special assistance from the company and resigned.
The copilot had flown approximately 701 total flight hours, 304 hours of which were with the company, 107 hours of which were in King Airs, including 36 hours in the 30 days preceding the accident. His last recurrent ground and flight training occurred on August 2, 2002, and was satisfactorily completed. However, the lead ground instructor characterized the copilot's performance in ground school as below average. He had problems with memory items, calculating weight and balance, and applying formulas.
Pilots who had flown with the copilot described him as "friendly, happy, organized, motivated." Several described him as not assertive and expressed concern about his flying skills, especially his inability to land the airplane without assistance. Two pilots stated that the copilot had difficulties with power management when flying an approach and that he had to be reminded to keep one hand on the throttles and to monitor his power gauges. Another pilot, who had been mentoring the copilot and flew with him often, stated that this was a consistent problem. The power management issue had manifested itself earlier when the pilot was in A320 training.
The weather at 10:14 a.m., according to the AWOS, was three statute miles in light snow with scattered clouds at 400 feet and overcast at 700 feet.
Two aircraft that landed prior to the accident and one that departed shortly afterward did not find icing to be a problem.
A standard procedure after an accident involving IFR is to check all involved navaids. The Eveleth VOR showed minor anomalies but nothing serious enough to cause an accident. Several tests were flown by hand and with an autopilot-equipped aircraft and there was some discrepancy. The NTSB asked for a special flight demonstration and while the VOR was out of tolerance, pilots were able to make normal transitions to landing and the VOR was ruled out as a factor.
As noted earlier, the last two radar returns showed the airplane flying about 76 knots at 1,800 feet. The wings-level, gear-down stall speed, assuming an estimated landing weight of about 10,500 pounds with the flaps in the approach setting, would have been about 77 knots with the power at flight idle. The NTSB then duplicated the approach scenario using a King Air C90B simulator to recreate the condition and to assess the crew's workload.
The tests conducted without simulated airframe icing showed that the flight path could be matched with minor pitch and power inputs. Stall warnings occurred between 81 and 84 knots but prompt power application allowed a quick recovery. With light to moderate simulated airframe icing the flight path could also be duplicated, although more power was needed for recovery.
Considerable simulation time was devoted to duplicating the approach scenario and how the aircraft was configured. The King Air passed through maximum gear extension speed of 156 knots while descending through 2,200 feet. Particular attention was focused on cockpit setup, duties for the flying pilot and the nonflying pilot, callouts, and checklist usage.
The charter company's procedures guide states that upon passing the final approach fix (FAF), the pilot should begin the descent and extend the landing gear. Landing flap extension can be delayed until the landing is assured. The recommended approach speed for precision approaches was 120 knots, but none was specified for nonprecision approaches. The chief pilot recommended all approaches be flown at 130 knots until the runway was in sight. The chief pilot further indicated that with the gear down, about 400 foot-pounds of torque, and 130 knots of airspeed, the descent rate was less than 1,000 fpm.
The NTSB reached the following conclusions: The flight crew failed to maintain an appropriate course and speed for the approach and did not properly configure the airplane at the start of the approach, making the later stages of the approach more difficult. During the final portion of the approach, the flight crew failed to monitor the airspeed and allowed it to decrease to more than 50 knots below the company's recommended approach airspeed.
The flight crew failed to recognize and recover from a stall and instead of going around with a full CDI deflection and an inadequate airspeed, they chose to continue. If the flight crew had been adhering to procedures and applying crew resource management (CRM) techniques, at least one crewmember would have been monitoring the instruments and would have noted airspeed and CDI deviations.
Icing was not determined to be a factor and while the clouds were just above or near the minimums for the approach, which might have precluded the pilots from seeing the airport, that should not have resulted in a stall.
The FAA came in for some criticism for failing to exercise appropriate oversight of the charter company, for not having an active CRM program, and for not conducting enough inspections to detect various operator/pilot shortcomings. In fairness, it should be noted that the FAA district office responsible for oversight was significantly understaffed and a hiring freeze was in effect at the time preceding the accident and for several months afterward.
The NTSB found that both pilots had previously demonstrated serious performance deficiencies with below-average flight proficiency. The charter company was not operating in accordance with its procedures, and it did not have adequate stall recovery guidance, consistent deicer boot operational guidance, an effective CRM program, or an in-range checklist, which helps crews plan for the arrival.
In the most basic terms, a professional flight crew allowed the aircraft to get ahead of them during an approach and went from being too fast and too high into a stall following an unstabilized approach. They were clearly distracted and, based on the detailed crew profiles that the NTSB uncovered, were not well suited for the positions they held. Even the most rudimentary airmanship of just noting that the aircraft was getting way too slow likely would have prevented this loss. The charter company operations management and the other pilots who knew the problem were remiss, perhaps in a misguided hope that by not confronting the problem it would somehow go away. A high price was paid for ignoring numerous warning signs.
Our system provides checks and balances, including FAA oversight, operator oversight, and crewmember and fellow pilot oversight. Most of us have seen other pilots who've had a bad day, maybe even had one ourselves, and the system worked as designed — nobody got hurt. We're not talking about an isolated incident, a fluffed landing, or some other minor transgression. However, when a pilot is consistently weak, is as uneasy about flying as this captain clearly was, and no one takes positive steps to intervene, it can turn ugly very quickly.
Bruce Landsberg is the executive director of the AOPA Air Safety Foundation.