History of the Flight
The Boeing 777-200B, registration N784UA, was scheduled for a public transport flight from London Heathrow to Los Angeles. While taxiing for Runway 27 Right, the crew completed the briefing and entered payload weights into the Flight Management Computer (FMC). After receiving a weight change, the crew correctly updated the FMC and set V speeds according to company procedures: V1 154 kt, VR 161 kt, and V2 167 kt. The aircraft followed a Gulf Air aircraft that had been cleared to line up, and the surface wind was reported as 180°/15 kt with infrequent gusts of 28 kt.
After the preceding aircraft departed, the Boeing 777 was cleared for takeoff. The commander lined up the aircraft, and the first officer handled the controls for a rolling takeoff. Left bank control input and right rudder were required to maintain the runway centerline; crosswind yaw was noted but not considered significant. Acceleration was normal until near V1, when the airspeed dropped by 8 to 10 kt before steadily recovering. Rotation began at 161 kt. All three crew members believed the rotation rate and angle were normal, and the aircraft accelerated toward V2 of 167 kt. A tail strike message then appeared on the Engine Indication and Crew Alerting System (EICAS).
The aircraft was depressurized in accordance with the emergency checklist, climbed initially to FL 90, then held at Daventry at FL 70, where 130,000 lb of fuel was jettisoned to reduce landing weight. After completing the emergency checklist, the aircraft was given radar vectors for Runway 27L at Heathrow, where an uneventful landing was carried out.
Aircraft Examination
Examination at the operator's maintenance facility at Heathrow revealed signs of runway contact on the underside of the fuselage from approximately Station 2130 to Station 2274, a total length of about 3.65 meters (12 feet). At Station 2150 (pressure bulkhead) and most frames aft to Station 2268, the outer skin was completely abraded along the fuselage centerline. The tail strike sensor at Station 4041 was abraded by about 50%.
Flight Recorder Data
Recorded data from the Flight Data Recorder (FDR) and Quick Access Recorder (QAR) were analyzed. The 30-minute Cockpit Voice Recorder (CVR) had overwritten the event as it continued running after the aircraft parked. The QAR data were sent to the operator's Flight Operations Quality Assurance (FOQA) program and were not made available to the investigation; only FDR data were used.
During the latter part of the takeoff roll, the handling pilot held between 15° and 20° of anti-clockwise control wheel with right rudder. All left side spoiler panels deployed between 4.5° and 7°. Just before rotation, the headwind component fluctuated between 10 kt and 2 kt. At rotation onset (airspeed increasing through 159 kt), the control column moved rearward to 45% of full travel, then slightly forward to 35%, with increasing anti-clockwise control wheel and left spoiler deflection. As the aircraft pitched up through 7.7° at a rate of at least 3.5°/second, maximum anti-clockwise control wheel input of 52° was recorded, and headwind component dropped from 4 kt to 1 kt. A maximum spoiler Panel 6 deflection of 21.7° occurred 0.4 seconds later.
After pitch attitude exceeded 10.7° with the main landing gear still not tilted (i.e., still on the ground), the control column was moved rearward to 22% travel and the control wheel briefly centered. A minimum headwind component of 0.25 kt was derived. The maximum pitch attitude recorded with the main gear not tilted was 12.8°, which is when tail contact likely occurred; roll attitude was between 0.5° and 1.5° left wing down. Pitch rate was sampled every 2 seconds. After liftoff and main gear tilt, pitch attitude temporarily reduced to 11.3° before increasing to 15° as the aircraft climbed.
A comparison with four previous takeoffs on the same FDR showed that the accident rotation had a significantly higher aft control column magnitude and rate, resulting in a pitch rate of at least 3.5°/second versus 2.5°/second for the other takeoffs. The pitch rate was well established before the reduction in headwind component. The aircraft manufacturer judged the effect of asymmetric spoiler deployment on pitching moment as negligible.
An anomaly was discovered in the FDR pitch attitude recording: two data buffers were read alternately every 200 ms, but only one buffer was actually read, causing three out of eight pitch attitude values to be repeated. This design error in the flight data acquisition programming was confirmed, and a correction was developed for the 'Block 2001' update to the Aircraft Integrated Monitoring System (AIMS), scheduled fleet-wide in the fourth quarter of 2001. All pitch attitude references in the report were corrected for this anomaly.
Discussion
The operator's expanded normal operating procedures state: "At VR rotate smoothly using a continuous rate of approximately 2.5° per second to establish an initial pitch attitude of approximately 15°. In gusty or windshear conditions, consider delaying rotation and increasing initial climb out speed. Early, rapid, or excessive rotation may cause aft fuselage contact with the runway. With the main gear on the runway, tail contact occurs at approximately 11°."
While the manufacturer concluded that the effect of asymmetric spoiler deployment on pitching moment was negligible, a drop in headwind component had a more significant effect. This delayed the main landing gear leaving the runway, increasing the risk of a tail strike if the rotation rate was not checked. On this occasion, the recorded pitch rate of 3.5°/second and nose-up attitude of 12.8° with the main gear still on the runway exceeded the normal operating procedure limits.