The following information provides an awareness of problems that might be avoided in the future. The information is based on final reports by official investigative authorities on aircraft accidents and incidents.
‘Aggressive Schedule’ Cited
Gulfstream G650. Destroyed. Four fatalities.
The manufacturer’s “persistent and increasingly aggressive attempts to achieve V2 [takeoff safety] speeds that were erroneously low” and its “inadequate investigation” of previous uncommanded rolls during takeoff performance tests were among the probable causes of the accident that killed all four crewmembers during certification flight testing of the Gulfstream G650, according to the U.S. National Transportation Safety Board (NTSB).
The accident occurred at Roswell (New Mexico, U.S.) International Air Center the morning of April 2, 2011. The flight crew was conducting a simulated one-engine-inoperative (OEI) takeoff and was not able to correct an uncommanded roll that occurred when the right wing stalled on liftoff. The wing tip struck the runway, and the experimental ultra-long-range, fly-by-wire business airplane veered right, struck a concrete platform housing electrical equipment and was consumed by a fuel-fed fire. The pilots and both flight test engineers succumbed to smoke inhalation and thermal injuries.
Flight testing that day, as well as during several previous tests, had focused on achieving the manufacturer’s target for V2, basically the minimum speed that a transport category airplane must attain at a height of 35 ft to meet the required OEI climb gradient to 400 ft. The V2 speeds achieved during the tests had consistently been too high to meet the manufacturer’s goal of providing a balanced field length of 6,000 ft (1,829 m).
Various rotation techniques and angle-of-attack (AOA) targets had been tried by Gulfstream Aerospace’s flight-test crews, but none had succeeded in meeting the target V2 without exceeding the 20-degree pitch angle that had been set to assure passenger comfort. The most recently developed technique included an abrupt and rapid rotation, using the maximum allowed 75 lb (34 kg) of pull force on the control column, to an initial 9-degree pitch attitude, then a further increase in pitch attitude to achieve V2. This technique had produced the best results, exceeding the target by only 3 kt.
However, in the course of about a dozen test flights earlier the morning of April 2, the pilot-in-command (PIC) had decided that a smooth rotation, pausing only briefly at 9 degrees before increasing pitch to about 16 degrees, might be a better and more repeatable technique. “I’m not doing that jerk stuff,” he told a flight test engineer. “It doesn’t work … and I don’t think the FAA [U.S. Federal Aviation Administration] is going to like it, either. It’s such a great-flying airplane, you shouldn’t have to abuse it to get [it] flying.”
The NTSB report noted that both flight crewmembers had extensive experience as test pilots. The PIC had 11,237 flight hours, including 263 hours in G650 certification testing. The second-in-command (SIC) had 3,940 flight hours, including 140 hours in type.
During some of the earlier test flights that morning, the use of a smooth rotation with a brief pause at 9 degrees produced V2 speeds within 4 kt of the target (135 kt). The pilots agreed to try an even briefer pause at 9 degrees during the next takeoff, which was conducted on Roswell’s Runway 21 with flaps extended 10 degrees. As planned, the SIC moved the right thrust lever to idle at 105 kt, simulating an engine failure. At 127 kt, the PIC began rotating the airplane for takeoff. Recorded flight data indicated that there was no pause when the pitch attitude reached 9 degrees. AOA quickly exceeded 11 degrees, the outboard section of the right wing stalled, and the airplane rolled right. The PIC attempted to level the wings, but the bank angle increased beyond 16 degrees.
The pilots had received no warning of the asymmetric stall. Because of a miscalculation of the G650’s stall AOA in ground effect, the threshold for activation of the stick shaker and the primary flight display pitch-limit indicators had been set too high. “Ground effect refers to changes in the airflow over the airplane resulting from the proximity of the airplane to the ground,” the report explained. “Ground effect results in increased lift and reduced drag at a given [AOA], as well as a reduction in the stall AOA.”
The stick shaker activated and the indicated pitch attitude reached the limit shown on the primary flight displays only after the stall occurred. The PIC pushed the control column forward, applied full left control wheel and rudder, and called for “power.” The SIC already had moved the right thrust lever full forward. Despite these inputs, the airplane remained in a stalled condition. The sound of an automatic warning when the right bank angle exceeded 30 degrees was captured by the cockpit voice recorder shortly before the recording ceased about 24 seconds after the takeoff was initiated.
The report noted that uncommanded rolls resulting from right outboard wing stalls had been encountered during two previous test flights. In both cases, the pilots recovered by reducing AOA. The events subsequently were attributed to stalls induced by over-rotation. “If Gulfstream had performed an in-depth aerodynamic analysis of these events shortly after they occurred, the company could have recognized before the accident that the actual in-ground-effect stall AOA was lower than predicted,” the report said.
Investigators found that the stall precipitating the accident had occurred at an AOA of 11.2 degrees, or about 2 degrees lower than the predicted stall AOA in ground effect.
The report said that contributing to the accident was the manufacturer’s “aggressive” flight-test-program schedule, which was designed to achieve certification of the G650 by the third quarter of 2011. “The schedule pressure … led to a strong focus on keeping the program moving and a reluctance to challenge key assumptions.”
The report noted that after the accident, Gulfstream suspended performance flight testing and implemented several corrective actions. The target V2 was increased by 15 kt, while maximum takeoff thrust was increased by 5 percent to meet the takeoff performance goals. Certification of the new airplane eventually was achieved in September 2012.
The accident investigation generated 10 safety recommendations, including the NTSB’s call for the FAA to work with the independent Flight Test Safety Committee to develop detailed guidance for aircraft manufacturers on flight test operations (ASW, 11/12).
Illness Prompts Diversion
Boeing 777-200. No damage. No injuries.
More than an hour after the airplane departed from Paris for a flight to New York the morning of Jan. 17, 2011, the captain became ill. A physician among the passengers diagnosed gastroenteritis and applied basic antispasmodic treatment, after which the captain decided to continue the flight, said the report by the French Bureau d’Enquêtes et d’Analyses.
About 90 minutes later, the captain felt abdominal pain. “The doctor observed that the captain was very pale, with stiffness, shaking and severe pains in the abdominal region,” the report said. The copilot declared an emergency and diverted the flight to Keflavik, Iceland. “During the descent, an improvement in the captain’s condition allowed him to assume the duties of PNF [pilot not flying],” the report said. The airplane was landed without further incident.
The captain was taken to a hospital, where he was observed and released the same day. “The investigation could not determine the exact nature of the captain’s pains,” the report said.
Microburst on Short Final
Airbus A340. Substantial damage. No injuries.
Visual meteorological conditions (VMC), with winds from 360 degrees at 5 kt, had been reported at Darwin (Northern Territory, Australia) Airport the night of Feb. 28, 2012, but, while completing the instrument landing system (ILS) approach to Runway 29, the flight crew saw heavy rain close to the threshold.
The crew asked the airport traffic controller for an update on the weather conditions. The controller replied that there was a storm extending to the east from the runway threshold but that the reported wind was still from 360 degrees at 5 kt. The Australian Transport Safety Bureau (ATSB) report noted that the wind data were derived from sensors located about 2.3 km (1.2 nm) from the threshold.
“The crew briefed the possibility of a missed approach if the conditions deteriorated,” the report said.
Rainfall increased as the aircraft neared the runway. The crew set maximum continuous thrust to arrest an increased sink rate encountered at 55 ft above ground level (AGL) but then reduced thrust to idle shortly thereafter. “As the aircraft entered the flare, the rain intensified, significantly reducing visibility,” the report said.
The A340 touched down hard, with a recorded vertical acceleration of 2.71 g. None of the 116 passengers and eight crewmembers was hurt, but subsequent engineering inspections disclosed a broken engine mount and damage requiring replacement of several main landing gear components.
The report said that analyses of recorded flight data indicated that the aircraft might have encountered a downburst, an intense localized downdraft. The data showed that just before touchdown, the wind had changed from a 9-kt headwind to a 6-kt tailwind. “At touchdown, the tailwind was recorded at 18 kt, and the rate of descent was 783 fpm,” the report said.
Fuel Leak Causes Fire
Boeing 767-300. No damage. No injuries.
The 767 was climbing through 9,000 ft after departing from New York’s John F. Kennedy International Airport (JFK) for a flight to Haiti with 201 passengers and 12 crewmembers the morning of Feb. 8, 2012, when the flight crew heard a bang that was immediately followed by warnings of a fire in the right engine.
The crew shut down the engine and discharged a fire bottle into the right nacelle. The fire warning persisted until the crew discharged the second fire bottle, the NTSB report said. They declared an emergency and turned back to JFK. The first officer flew the airplane while the captain and standby first officer completed the associated checklists and communicated with the flight attendants. The crew then landed the airplane without further incident.
Subsequent examination of the 767 revealed no damage from the fire or the overweight landing. Investigators found that during maintenance the night before the incident, a bracket and spray shield for the integrated drive generator’s fuel-oil heat exchanger had been reassembled incorrectly. “A seal under the fuel tube flange that is held in place by the bracket had the O-ring partially missing, which was the source of the fuel leak,” the report said. “Contributing to the cause of the fire was the 767 Aircraft Maintenance Manual’s lack of any graphical or pictorial displays of the correct assembly of the two-piece bracket and spray shield.”
Head-On Over the Atlantic
Airbus A319, Boeing 737-800. No damage. No injuries.
A controller’s loss of awareness of the airplanes’ flight paths resulted in the issuance of a climb clearance that placed the A319 and the 737 on a head-on collision course off the eastern coast of the United States the evening of Nov. 11, 2010, according to an NTSB incident report.
The flight crews of both airplanes received, and followed, traffic-alert and collision avoidance system (TCAS) resolution advisories that resulted in the A319 and the 737 passing about 1,800 ft vertically and 2.8 nm (5.2 km) laterally of each other about 66 nm (122 km) east of Hobe Sound, Florida.
The A319 had been southeast-bound at Flight Level (FL) 360 (approximately 36,000 ft), en route from Washington to Bogotá, Colombia. The 737 was northwest-bound at FL 370, en route from Oranjestad, Aruba, to Atlanta.
The report said that the 737 was still “well within” Miami Air Route Traffic Control Center (ARTCC) Sector 21 when the sector controller handed off the flight to the controller of an adjacent sector, Sector 2, which the 737 eventually would enter — and which the A319 was transiting. About a minute later, the Sector 2 controller handed off the A319 to the Sector 21 controller.
The Sector 21 controller had the 737’s data tag from his radar display and “did not maintain awareness” of the airplane’s position after handing off the 737 to the Sector 2 controller, the report said. Unaware of the conflict he was creating, the Sector 21 controller cleared the A319 to climb to FL 370, “which placed the flight in direct conflict with the 737,” the report said.
Shortly thereafter, the ARTCC’s radar data processing system generated a conflict alert, and the sector controllers radioed traffic advisories and instructions to resolve the conflict. However, the flight crews of both airplanes replied that they were following TCAS resolution advisories. The 737 crew also reported that they had the A319 in sight.
Tug Slides on Slippery Ramp
Bombardier CRJ200. Substantial damage. No injuries.
The airplane had been dispatched with its auxiliary power unit (APU) inoperative per the minimum equipment list and was two hours behind schedule for departure from Salt Lake City the night of Nov. 23, 2010. “The captain stated that he started both engines [using an external power cart] due to a concern that by starting one engine only he would encounter control problems taxiing in the slippery conditions,” said the NTSB report.
Snow was falling, and the ramp area was covered with snow and ice. “During pushback, the tug was unable to gain enough traction to move the airplane and was subsequently replaced with a larger tug,” which initially was able to move the CRJ, the report said. However, the airplane, with both engines at idle power, began to overpower the tug.
“The captain … experienced a sensation of unusual movement [and] asked the tug driver if the driver still had control of the airplane,” the report said. “The tug driver confirmed that he had control; however, the airplane subsequently moved forward while still attached to the tug, which rotated to the right, striking the airplane’s fuselage.” The collision damaged the CRJ’s lower fuselage skin and several stringers.
The report said that a factor contributing to the accident was that “no guidance existed for either the flight or ground crew regarding pushback procedures in low-traction ramp conditions with an inoperative APU.”
Altitude Deviation Unnoticed
Bombardier DHC-8-100. No damage. No injuries.
Inadequate monitoring of flight instruments resulted in the continuation of a gradual descent that placed the Dash 8 on a head-on collision course with another aircraft off the east coast of the Hudson Bay the afternoon of Feb. 7, 2011, said a report by the Transportation Safety Board of Canada (TSB). The TSB also faulted the absence of simulator training on TCAS maneuvers for the initial incorrect reactions by the pilots of both aircraft to TCAS resolution advisories.
The aircraft, both Dash 8s operated by the same airline, were flying in opposite directions between Puvirnituq and La Grande-Rivière, Quebec, in VMC but in airspace not covered by air traffic control (ATC) radar. The DHC-8-100, with 28 passengers and three crewmembers, was northbound to Puvirnituq at FL 230; the other aircraft, a DHC-8-300, was southbound at FL 220 with three crewmembers.
The -100 crew was using the autopilot’s vertical-speed mode to maintain altitude because of pitch oscillations that often occurred when the altitude-hold mode was selected. “The use of [vertical-speed] mode is neither intended for nor evaluated for this function, but nothing prohibited the flight crew from using it to maintain altitude,” the report said.
The captain was alone in the cockpit, after the first officer left for “physiological reasons,” and did not notice when the -100 began to descend at about 50 fpm, the report said. “Since the rate of descent was very slow, the speed and attitude of the aircraft were very similar to those of cruise flight. It was therefore impossible to note the descent without reference to the flight instruments.”
During the next 14 minutes, the -100 descended about 700 ft. The captain apparently did not notice the altitude-warning light. The first officer was re-entering the cockpit when the TCAS generated a traffic advisory, then a resolution advisory to climb. The captain disengaged the autopilot and began a 38-degree-banked right turn. “During this turn, the aircraft lost just over 50 ft in altitude before beginning to climb,” the report said, noting that the right turn might have been an automatic reaction to opposite-direction traffic.
Meanwhile, the pilot flying the -300, the first officer, had begun a shallow left climbing turn after misinterpreting the TCAS “descend” resolution advisory displayed on the vertical speed indicator. “It is possible that the appearance of the [-100 target symbol] in the upper right-hand corner of the display may have caused the [pilot] to turn left,” the report said.
During these maneuvers, the aircraft passed within 1,500 ft vertically and 0.8 nm (1.5 km) laterally. Both aircraft then continued to their destinations without further incident.
Lesions Induce Disorientation
CASA 212-100. Destroyed. Five fatalities.
ATC radar and radio contact with the aircraft were lost about 26 minutes after it departed from Batam, Indonesia, with two pilots and three company engineers for a functional check flight the afternoon of Feb. 12, 2011. The wreckage was found on Bintan Island, and subsequent examination revealed that the propeller on the left engine was not rotating on impact, said the report by the Indonesian National Transportation Safety Committee.
Investigators found signs that the left engine had erroneously been placed in reverse before the pilots lost control of the aircraft. They found no check flight plan or authorization for the PIC to conduct the check flight following replacement of the left engine. The PIC, 61, had 13,027 flight hours, including 3,311 hours in type. The first officer, 50, had 2,577 flight hours, including 152 hours in type.
A postmortem examination of the PIC revealed lesions that had caused paralysis of the vestibular organs in his right inner ear. The report said this condition meant that the pilot “could not [respond] normally to three-dimensional motion or movement” and may have induced spatial disorientation.
Beech C90 King Air. Minor damage. No injuries.
The airplane had departed from Champaign, Illinois, U.S., and was climbing through 23,400 ft the afternoon of Feb. 9, 2011, when the flight instructor and commercial pilot receiving instruction heard a loud bang as the inner ply of the copilot’s windshield shattered.
“The [instructor] noted that the flight was in clouds and that the temperature aloft … was minus 23 degrees C [minus 9 degrees F],” the NTSB report said. “Additionally, he stated that there was no visible structural icing present and the electric window heat was on.”
The instructor declared an emergency and requested and received vectors from ATC to the nearest suitable airport, Evansville, Indiana, where the airplane was landed without further incident.
Examination of the windshield revealed a peel-chip fracture of the inner ply. This type of fracture “has historically been an issue on the King Air” and prompted the windshield manufacturer in 2001 to incorporate a layer of urethane “that relieves stresses on the inner glass ply and prevents peel chip-type fractures,” the report said. “This airplane did not have the improved windshield.”
Task Overload Suspected
Beech B55 Baron. Substantial damage. Two serious injuries.
The landing gear warning horn sounded shortly after the airplane departed from San Luis Obispo, California, U.S., the morning of Feb. 7, 2011. “The pilot diagnosed the problem and determined that the landing gear had retracted successfully and that the indication system was in error,” the NTSB report said. “He continued the flight with the horn intermittently sounding.”
Nearing the destination — San Bernardino, California — the pilot received an unexpected clearance from ATC to navigate directly to the airport. “As a result, he rushed through the descent checklist items,” the report said.
The pilot then requested and received clearance from ATC to make a low pass over the runway so that the airport traffic controllers could perform a visual check of the landing gear. During the low pass, the pilot began to have difficulty controlling the Baron and did not realize that the right engine had lost power due to fuel starvation, as confirmed by the airplane’s on-board engine-monitoring system.
The pilot subsequently lost control of the airplane, which crashed in a nose-down, inverted attitude in a recreational vehicle storage facility.
Examination of the Baron revealed no airframe or engine anomalies that would have precluded normal operation. Investigators found that the pilot had not switched from the auxiliary fuel tanks to the main tanks during descent or approach, and that the right engine had lost power after the fuel in the right auxiliary tank was exhausted.
The report said that the unexpected approach clearance and the low pass over the runway might have caused the pilot to experience “task overload” that contributed to his improper fuel system management.
Descent Beneath Low Clouds
Cessna 340. Destroyed. Two fatalities.
The pilot was en route under visual flight rules from Henderson, Nevada, U.S., to Compton, California, the afternoon of Jan. 18, 2010. He held a private pilot certificate with a multiengine rating but did not have an instrument rating. His total flight time was 474 hours.
The airplane was cruising at 10,500 ft when it encountered instrument meteorological conditions, and the pilot initiated a descent. Recorded ATC radar data showed an “erratic and circling flight path” before radar contact was lost at 4,800 ft. “It is likely that the pilot was having difficulty determining his location and desired flight track,” the NTSB report said.
A glider pilot, who was driving on a highway in the area, saw an airplane matching the description of the 340 flying at 200 ft AGL just below the clouds and in “bad” visibility. About eight minutes after the witness lost sight of the airplane, the 340 struck the slope of a gully at about 2,490 ft near Lytle Creek, California.
Engine Fails, Control Lost
Piper Twin Comanche. Substantial damage. One serious injury.
The airplane recently had undergone maintenance that included overhaul of both engines. The pilot had called the maintenance facility after landing in Big Bear City, California, U.S., on Jan. 29, 2011, to report that the right engine was running rough. “A mechanic was not available to help him, and he was told that he should not fly the airplane,” the NTSB report said.
The pilot apparently had no further contact with the maintenance facility before attempting to depart from Big Bear City for a flight to Pacoima, California, the next morning. A witness heard the engines popping and backfiring before the airplane began the takeoff roll.
The pilot told investigators that he was turning onto a left crosswind when the right engine lost power. “He continued the left turn to downwind and made sure to keep the airspeed above the single-engine control speed of 90 mph,” the report said. “The pilot’s last recollection was turning to final approach and seeing the runway.”
The report said that the pilot allowed airspeed to decrease below the single-engine control speed; information recovered from the Twin Comanche’s global positioning system indicated that the groundspeed was 76 mph about 1,400 ft (427 m) from the runway threshold. “The airplane subsequently impacted the roof of a private residence located about 900 ft [274 m] from the runway threshold,” the report said. “The airplane came to rest inverted in the front yard.”
Examination of the wreckage revealed nothing to explain the loss of power, but the report said that a contributing factor in the accident was “the pilot’s decision to fly with a known deficiency in one engine.”
Control Fastener Detaches
Robinson R44 Astro. Destroyed. Two fatalities, one serious injury.
As part of a biennial flight review at Cessnock Aerodrome in New South Wales, Australia, the morning of Feb. 4, 2011, the instructor simulated a failure of the flight control hydraulic boost system. After landing the helicopter, the pilot told the instructor that the hydraulic system would not re-engage.
The passenger, who normally flew the helicopter, told the instructor that the system had been leaking and that he had replenished the reservoir that morning. “The instructor announced that he would reposition the helicopter to the apron to facilitate examination of the hydraulic system,” the ATSB report said.
The instructor lost control of the R44 shortly after becoming airborne. The helicopter was in a steep left bank when it struck the runway and came to rest on its left side. A fire erupted and rapidly engulfed the helicopter. The pilot was able to escape, but the instructor and the passenger succumbed to thermal injuries.
Investigators found that the bolt securing the lower flight control push-pull tube to the left-front hydraulic servo had detached while the helicopter was on the ground following the simulated hydraulic failure. “The ‘feel’ of the flight control fault [would have] mimicked a hydraulic system failure,” the report said.
Although the precise cause of the bolt detachment was not determined, investigators found that “a number of self-locking nuts from other aircraft, of the same specification as that used to secure safety-critical fasteners in [the accident helicopter] were identified to have cracked due to hydrogen embrittlement.”
The report noted that the aluminum fuel tanks in the helicopter had not been replaced with bladder tanks, as recommended by a service bulletin issued by the manufacturer in December 2010 to “improve the R44 fuel system’s resistance to a post-accident fuel leak.”
Control Lost After Tail Strike
McDonnell Douglas 369FF. Substantial damage. One fatality, two serious injuries, one minor injury.
The police helicopter was being used to scout a 3,600-ft mountaintop near Marana, Arizona, U.S., for installation of emergency communications equipment the morning of Jan. 31, 2011. The NTSB report said that the weather conditions, which included 10- to 15-kt winds, were “within the helicopter’s and the pilot’s performance capabilities.”
The pilot circled the peak before attempting a pinnacle landing. “The passengers reported that during the landing attempt, they felt a bump, the helicopter rose a few feet, then the nose pitched down and the helicopter began to spin to the right,” the report said. “The helicopter tumbled and slid about 120 ft [37 m] down a shallow canyon … before it was halted by rocks and scrub vegetation.”
Investigators determined that the pilot, who was killed in the crash, had lost control of the helicopter after the tail rotor struck the ground during the attempted pinnacle landing.