Figure: Accident aircraft fuel system diagram, from NTSB Aircraft Accident Brief, figure 1.
This mishap was caused by a pilot's over-aggressive rotation, the so-called "snatching the controls" problem. I had just checked out that year in this airplane and as soon as I heard Transport Canada's explanation, I knew it was wrong. They blamed fuel migration for a shifting center of gravity. Most CL604 pilots suspected the pilots almost immediately.
Canada stuck to the fuel migration story for almost four years, placing restrictions on the airplane but not really addressing fuel migration. I flew with a Canadair test pilot in 2001 and took the airplane away from him during takeoff rotation. "Take it easy," I told him, "your manual says 3 degree per second, maybe you should try that for a change." He dismissed my warnings as being overly cautious.
Four years after the Air Transport Canada report, the FAA took the unusual step of reexamining the data. They found the pilot had rotated the nose at 9.6° per second, three times the rate specified in the aircraft operating manual. Looking at other Bombardier test pilots it appeared everyone was rotating at least double the prescribed rate. Had the pilot used the correct rate, the aircraft would have flown under perfect control.
So why does an entire cadre of "test pilots" ignore the flight manuals they are responsible for writing? I think it can be blamed almost entirely on complacency. The best paying jobs at this end of corporate aviation tend to be flying for private owners, and to a lesser degree, charter companies. The aircraft manufacturer offers less pay but the psychological benefit of a business card that says "test pilot." Being among a cadre of test pilots with little or no external oversight leads many to start ignoring established procedure. More about this: Pilot Psychology / Complacency.
The official accident report, while citing the rapid rotation, still mentions fuel migration. The telling fact, however, is that nothing has been done to address this supposed problem. That tells me there was no problem with fuel migration. There are two lessons here:
- Don't do anything with the flight controls at a rate faster than the flight controls can provide you feedback. Part of the reason you are sitting in row one is you can provide finesse the computers cannot.
- If you are in an environment where there is little oversight and your peers tend to ignore procedure, find another job or blaze a new trail and set an example for them to follow.
What follows are quotes from the relevant regulatory documents, listed below, as well as my comments in blue.
- Date: 10 October 2000
- Time: 1452
- Type: Canadair CL-600-2B16 Challenger 604
- Operator: Bombardier Aerospace
- Registration: C-FTBZ
- Fatalities: 3 of 3 crew, 0 of 0 passengers
- Aircraft Fate: Destroyed
- Phase: Takeoff
- Airports: Wichita-Mid-Continent Airport, KS (ICT), USA
[NTSB Aircraft Accident Brief, AAB-04/01]
- At 1448:45, the tower issued a takeoff clearance and instructed the flight crew to fly a heading of 230°.10 At 1449:21, the pilot stated, "okay, here we go," and a sound similar to an increase in engine RPM was recorded 2 seconds later. At 1449:29, the pilot stated, "set thrust," and the copilot responded, "thrust set" 6 seconds later. At 1449:37, the copilot called out "airspeed's alive eighty knots." At 1449:48 the copilot called out "V one" (takeoff decision speed) and "rotate". The pilot responded, "okay, we're flying," followed by the copilot calling out "V two" (takeoff safety speed).
- Main gear liftoff occurred about 1449:50, as the airspeed reached 143 knots. FDR-derived data indicated that the PF used about 10° of nose-up elevator to initiate rotation, and main gear liftoff occurred about 1.2 seconds later, with a pitch angle of between 2.8° and 3.8°. The 10° nose-up elevator input was maintained for 0.8 second after liftoff until the pitch attitude reached 12°, according to FDR data. Pitch attitude continued to increase over the next 1.4 seconds, peaking at 20°, while nose-up elevator input decreased from 9° to 1° nose up. According to the FDR, the vane AOA reached 23° about 3.4 seconds after start of rotation. According to Bombardier, the airplane enters the stall warning region after reaching an AOA of 19°.
- The airplane's ADAS [Airborne Data Acquisition System] indicated a pitch rate of 9.6° per second.
- FDR data indicated that the airplane began an uncommanded right roll just before reaching peak pitch attitude. The CVR recorded the sound of the stick shaker at 1449:51, and the stick shaker sound continued for 2.2 seconds. During this time, a nose-down elevator input of about 14° was recorded, followed by a 5.5° nose-up elevator control input, consistent with pilot control inputs to correct the airplane's pitch and roll oscillations.
- At 1449:51, the CVR recorded a sound similar to stick shaker for 2.2 seconds, during which time the pilot stated "whew," and the flight test engineer stated "what are you doing?".
- The pitch attitude decreased to 4.3° nose up and the bank angle increased to about 80° right-wing down. During the next 3 seconds, the airplane rolled left to about wings level as the pitch attitude increased to 18° nose up. The vane AOA on the second pitch up was 26.4°.
- The CVR then recorded the mechanical voice warning "bank angle" and a sound similar to stall aural warning for 1.1 seconds at 1449:53.. "Bank angle" was recorded at 1449:54 and again at 1449:55. A sound similar to stick shaker was recorded for 0.15 seconds beginning at 1449:57, followed by "bank angle" again at 1449:57.36.
- The second pitch up oscillation was followed by a second pitch down to -2°, and a right-wing down roll to 61°. This pitch down was followed immediately by a pitch up and roll back to wings level, reaching nearly level pitch attitude and 40° right-wing down at impact, according to FDR data. Peak nose-up elevator input at this time (1449:55) was about 16°.
- At 1449:58, and for the next 2 seconds, a sound similar to stick shaker was recorded for 0.22 second, and the pilot stated, "hang on." A sound similar to stick shaker" was recorded again for 0.3 seconds, the flight test engineer repeated "what are you doing?," followed by a sound similar to stall aural warning for 0.82 seconds, and "bank angle" again. At 1449:59.59, the pilot stated, "hang on." The recording ended at 1450:00.
- Witnesses reported seeing the airplane bank to the right after takeoff. They stated that the airplane's right wing rolled and impacted the ground first and that the airplane exploded on impact. The airplane crashed through an airport perimeter fence and came to rest adjacent to a two-lane, north-south road.
The pilot flying
[NTSB Aircraft Accident Brief, AAB-04/01]
- From August 1989 to October 1990, he performed avionics certification testing as a flight test engineer for an avionics manufacturer. He was employed as a captain on an Aero Commander 500 for 14 CFR Part 135 cargo operations from October 1990 to September 1993.
- According to company records, he had logged 6,159.3 hours flying time, including 1,187 hours at Tucson Production Flight Test; 359.3 hours engineering flight test flying time at BFTC; 557.2 hours of production flight test PIC time at Tucson; and 126.4 hours of engineering flight test as PIC at BFTC.
- He had logged 189 flying hours in the Challenger 604, of which 94.6 hours were as PIC. He received his initial type rating in the Challenger 604 on October 15, 1998. His last proficiency check was accomplished on March 24, 2000.
- According to BFTC's manager of flight test operations and safety, there was no record that the pilot flying had received formal test pilot training.
The pilot was not, by any definition, an experienced jet pilot ready to take on even a routine test flight of this airplane. He was probably a very good pilot placed into a situation he was ill prepared to cope.
- Safety Board staff reviewed flight data to determine the peak pitch (rotation) rates per second during previous takeoffs performed by the PF. [Results:] 7.2° per second, 6.5° per second, 6.0° per second, 8.3° per second, 6.8° per second, and 5.8° per second.
The pilot's rotation rate during the accident flight was 9.6° per second, triple the rate required by the airplane flight manual.
Normal takeoff procedures
[NTSB Aircraft Accident Brief, AAB-04/01] According to the Bombardier Aerospace Challenger 604 Operations Reference Manual, the PF rotates to 14° at 3o per second after the "rotate" call from the PNF. The same rotation rate is used for an abnormal takeoff (engine failure after V1) but with a reduced pitch attitude of 10°. The rotation rate value listed in the Challenger 604 Operations Reference Manual is based on an industry average for transport-category aircraft takeoff profiles.
Aircraft pitch sensitivity
[NTSB Aircraft Accident Brief, AAB-04/01]
- In a second c.g. study, an FAA test pilot and a Transport Canada test pilot, who were rated in the CL-604, performed takeoffs in the Bombardier engineering flight simulator to determine the effects of c.g. location on rotation rate (and the ability to capture the prescribed takeoff pitch attitude) and to examine whether there were perceptible differences between the handling characteristics of the modified PFS and the production PFS installed on in-service CL-604 airplanes. The pilots performed takeoffs with c.g. locations ranging from 35.0 percent MAC to 42.0 percent MAC. The pilots reported that aft c.g. positions caused them to rotate at a somewhat higher rate. The pilots noted that these effects were more noticeable when they used increased rotation rates (about 6° instead of the normal 3° rotation rate). When increased rotation rates were used, the pilots noted that the stick shaker frequently activated but only briefly. The pilots also indicated that the simulator was controllable at all c.g. locations using both normal and increased rotation rates.
The importance of rotation rate
[Davies, page 183]
- There is a 'natural' rotation rate appropriate to each aircraft type; a few take-offs will soon show you just what this is. One can say that this rate then becomes the datum rate of a mid-weight I.S.A. sea-level take-off.
- The execution of a good take-off demands a smooth rotation of the aeroplane from VR, through the lift-off speed, through the screen at V3 (approximately V2 + 10 knots) to a setting V4 by, say, 300 feet, in the all engines case.
- Finally, a word of warning. Don't indulge in 'snatch' rotations; apart from the risk of getting near the stall under 'g' conditions after lift-off it frightens those dead-heading pilots riding in the back.
Of course this is precisely the point. Even a minor increase in G-loading decreases stall speed enough to cause one wing or the other to drop off.
[NTSB Aircraft Accident Brief, AAB-04/01] The National Transportation Safety Board determines that the probable cause of this accident was the pilot's excessive takeoff rotation, during an aft center of gravity (c.g.) takeoff, a rearward migration of fuel during acceleration and takeoff and consequent shift in the airplane's aft c.g. to aft of the aft c.g. limit, which caused the airplane to stall at an altitude too low for recovery. Contributing to the accident were Bombardier's inadequate flight planning procedures for the Challenger flight test program and the lack of direct, on-site operational oversight by Transport Canada and the Federal Aviation Administration.
Davies, D. P., Handling the Big Jets, Civil Aviation Authority, Kingsway, London, 1985.
NTSB Aircraft Accident Brief, AAB-04/01, Bombardier CL-600-2B16 (CL-604), C-FTBZ, Mid-Continent Airport, Wichita, Kansas, April 14, 2004