LAX99LA291

The flight had just leveled off at flight level 240 with the autopilot still engaged. The seat belt sign had been off for about 5 minutes, and the cabin crew was in the process of preparing a beverage service. The airplane was crossing a VORTAC when the flight encountered turbulence the crew categorized as 'severe,' with a rapid right roll and pitch excursion. The captain disconnected the autopilot and attempted to counter the excursions with control inputs. As a result of the aircraft motions, 1 flight attendant sustained serious injuries, and another flight attendant and 13 passengers sustained minor injuries. The airplane sustained minor damage, limited primarily to ceiling tiles and seats. A review of the meteorological data showed no evidence of any atmospheric phenomena in the area. A simulation study comparing the known airplane response to control inputs versus the motion of the airplane as recorded on the DFDR data showed that the flight controls alone did not cause the recorded motion of the airplane. Radar data established that the flight was in trail of a MD-11 by 11.5 nmi and 97 seconds when the upset occurred. The MD-11 had climbed through the flight's cruise altitude and was 600 feet above the flight at the time of the encounter. Wake vortex studies by NASA have documented vortex descent rates between 270 and 440 feet per minute.

History of flight

On September 2, 1999, at 1406 hours Pacific daylight time, United Airlines Flight 2036, a Boeing 737-322, N371UA, encountered turbulence at flight level (FL) 240 near Santa Barbara, California. The airplane was operated by United Airlines, Inc., under 14 CFR Part 121 as a regularly scheduled domestic passenger flight, en route from Los Angeles to San Francisco, California. The airplane sustained minor interior damage to some ceiling tiles and seats, and the flight continued to San Francisco, where an uneventful landing was accomplished. The 2 flight crew, 1 flight attendant, and 95 passengers were not injured. One flight attendant suffered serious injuries, while another flight attendant and 13 passengers sustained minor injuries. An IFR flight plan was filed for the flight, which departed Los Angeles at 1326. Visual meteorological conditions prevailed at the altitude, time, and location of the accident.

According to information provided by United Airlines Flight Safety Department, the flight had just leveled at FL240 and was crossing the San Marcus VOR when the turbulence was encountered. The flight crew characterized the event as an encounter with severe turbulence. The seat belt sign had been off for about 5 minutes. The seriously injured flight attendant was standing at row 22, just outside the aft galley, and the flight attendant with minor injuries was inside the galley area. One injured passenger was in the aft lavatory and the others were in their seats.

Radar data obtained from the Federal Aviation Administration (FAA) Oakland Air Route Traffic Control Center established that the aircraft was 11.5 nautical miles and 97 seconds in trail of a preceding MD-11 aircraft on the flight route when the event occurred. The data also showed that the MD-11 had climbed through FL240 and had leveled at FL270. No significant weather was forecast on the high altitude prognostic charts. The winds at FL240 were westerly at 25 to 34 knots. Safety Board specialists prepared factual reports for meteorology, radar data, and aircraft performance/flight recorders. These reports are appended to this file.

According to the Captain's written statement, they were at cruise flight at FL240 with a smooth and clear ride. The autopilot was +ACI-on+ACI- and they encountered a left roll with a corresponding change in pitch. He stated he disconnected the autopilot and rolled the aircraft level to regain a 200-foot altitude loss. The Captain then checked with the Flight Attendants in the back to see if any injuries had occurred. He confirmed that there were injuries. He asked the air traffic controller if there were any airplanes near them and found out that a MD-11 had climbed through their altitude ahead of them.

The First Officer stated that he had just turned off the fasten seat belt sign and made an announcement to the passengers when they encountered light buffeting, followed immediately by a left roll and a loss of 200-300 feet altitude. He said that this occurred near the San Marcus (RZS) VORTAC.

The uninjured flight attendant was interviewed after the accident and stated that she was at row 5 taking drink orders after the Captain had turned off the fasten seat belt sign. She said she felt her feet come off the floor and then found herself on the floor. She stated that it felt like the airplane dropped for approximately 8-10 seconds.

The seriously injured flight attendant said that about 20 minutes into the flight, they hit severe turbulence. She said she had been standing in the aisle at row 19, and then she suddenly found herself on the floor. She stated that due to her injuries she remained on the floor for the rest of the flight. She received injuries to her face, right elbow, thigh, and ankle.

The third flight attendant said that about 15 minutes into the flight she was working in the aft galley. She recalled that the airplane began to shake and she grabbed the galley counter. As she tried to get to her jump seat, the airplane suddenly rolled to the left and dropped about 200 feet. She said that she hit her head on the ceiling and passed out. She ended up sitting down in row 24B, where she remained for the rest of the flight.

Two doctors, two nurses, and a cabin crewmember onboard as passengers assisted in treating the injured.

Flight recorders and airplane performance

Radar showed that United Flight 2036 was about 11.5 nautical miles in trail of China Eastern 584, an MD-11, at the time of the upset. The Digital Flight Data Recorder was read out and used as data input to a computer-modeling program, which simulates the Boeing 737-300 series. The simulation study was run using control inputs from the Flight Data Recorder (FDR) to help separate motion in response to the controls from motion due to any external turbulence. DFDR control (wheel, rudder pedal, column, and engine rotor speed N1) time histories from the input data file drove the simulation. The takeoff weight and cg were obtained from United as 108,290 pounds at 17.0 plus ACU minus Mean Aerodynamic Chord. This was used for the simulation since the aircraft was not far from takeoff at the time of the upset.

As shown in the attached simulation study, in the case of the control column, the FDR shows that the initial nose down column input occurred between the time 1098.69 and 1099.69. The simulation pitch response to recorded controls occurs sooner and has greater magnitude than the recorded aircraft response. The simulated load factors generated also led the load factor recorded by the DFDR.

In the lateral axis, the simulator roll response to DFDR recorded control inputs did not match the recorded roll. The recorded roll did not respond to the recorded left wheel at 1098 seconds. About 1101 seconds, the recorded bank angle became significantly left of the bank angle that would exist purely from following the controls.

The rudder trace did not reflect the pedal input since the rudder is hinge moment limited at this flight condition. The initial simulated heading response is close to the recorded heading. However, the simulated heading departs from the recorded heading after the initial response due to the higher bank angle of the simulation.

The DFDR recorded airspeed represented in figure 6 of the attachment, shows a 6-knot jump at 1100 seconds. The 6-knot airspeed jump corresponds to a change in pressure differential between total and static pressure of 0.031 psi. The simulation showed that this jump is not a response to control input. DFDR recorded altitude shows a 30-foot jump in altitude at this time. The simulation altitude did not match the recorded altitude because of the higher bank of the simulation run. According to the report, airspeed is measured based on the difference between total and static pressure, and an airspeed jump represents some combination of an increase in total pressure or a decrease in static pressure.

The report states that the simulation showed that flight controls alone did not cause the recorded motion of the aircraft. Specifically, an external nose up pitching moment is required in addition to the control inputs at 1100 seconds to explain the aircraft motion. An eternal right rolling moment is also required at 1098 seconds, followed by a larger external left rolling moment at 1101 seconds. A complete copy of the simulation study is appended to this report.

United Airlines also conducted a data analysis of the upset event. Comments in their report were made relative to the FDR printouts provided by the Safety Board dated September 10, 1999. Their report summarized that based upon the data presented it appeared that the airplane did not experience an upset resulting from a rudder, or other flight control system malfunction. The upset appeared to have been caused by an atmospheric disturbance of unknown origin.

Radar information

Preliminary review