On 2 September 2013, an Airbus A330-200 (PT-MVL) being operated by TAM on a scheduled international passenger flight from Madrid to São Paulo Guarulhos, Brazil, under ETOPS encountered sudden severe turbulence as it transited the Inter Tropical Convergence Zone (ITCZ) at night in Instrument Meteorological Conditions (IMC). As a result of injuries sustained by 12 of the 156 passengers (two being serious) and 3 of the 13 cabin crew (one being serious), a diversion to Fortaleza was made landing there 1 hour and 40 minutes later. Minor cabin trim damage was caused by the impact of unrestrained passengers with the cabin roof. It was subsequently determined that AFM load limits had not been exceeded during the event.
An Accident Investigation was carried out by CENIPA, Brazil's Aeronautical Accidents Investigation and Prevention Centre. FDR data relevant to the event was successfully downloaded and informed the Investigation along with relevant meteorological data.
It was noted that Captain, who had been PF for the accident flight, had a total of 11,861 flying hours of which 1,712 hours were on type and that the First Officer had a total of 4,039 flying hours which included 1,757 hours on type.
The flight crew stated that prior to the sudden onset of turbulence, they had been in VMC with “clear visibility of the stars above” and no weather returns showing on their weather radar displays. FDR data showed that both pilot displays had been set to a range of 80 nm and with the beam tilt at -1° (although the pilots said that the range on the PM display had been 160 nm). They reported that without warning there had been St Elmo's Fire and the sudden onset of a short period of severe turbulence accompanied by the sound of hail. The seat belt signs had not been selected on because of the assessment by the pilots that there had been no risk of significant turbulence despite the forecasts they had seen for the time of their passage through the ITCZ. These forecast data included a SIGMET available before the aircraft had departed Madrid, warning of thunderstorms with tops up to FL450 during the four hour period during which the accident under investigation occurred. The significant weather chart covering the location of the turbulence event is shown below.
The serious injury to a member of the cabin crew occurred as he attempted to secure a service trolley in the rear galley which prevented him from having time to secure himself. The two seriously injured passengers were seated but not secured and both sustained spinal and other injuries after impact with the cabin roof. Once the extent and nature of injuries that had occurred were known, it was decided that the flight should make an en-route diversion to Fortaleza which was achieved without further event.
FDR data showed that at the onset of the severe turbulence, the aircraft had suddenly begun to climb rapidly and the Captain had attempted to prevent this by using side stick inputs. There was an un-commanded disconnection of AP during this period of maximum intensity which FDR data showed had lasted for 15 seconds out of a total for the whole episode of 1 minute and 37 seconds. There was no repetition thereafter.
The SIG WX forecast chart annotated with the positions where the severe turbulence began and ceased (P1 and P2 respectively). [Reproduced from the Official Report]
FDR data (see below) provided more detail on the upset. An initial vertical acceleration of +1.9g was immediately followed by one of -0.3g within 2 seconds and was accompanied by a rapid climb to FL409 at a rate of up to 4600 fpm. This all occurred within 5 seconds during which the aircraft was subjected to a strong vertical airflow which varied from 2000 fpm down to 9300 fpm up. An uncommanded AP disconnection soon after the severe turbulence was attributed to the exceedence of the limits for side stick input and/or angle of attack but it could not be determined which. Whilst it was also possible that an (inappropriate) attempt to oppose the upward movement had made matters worse, it was not possible to say if this intervention had been the cause of the AP disconnect. The AP was quickly re-engaged and did not drop out again.
Variation in vertical speed, load limit and airspeed during the event. [Reproduced from the Official Report]
FDR data also showed that concurrent with the upward airflow, the outside Air Temperature had suddenly risen from -59°C to -52°C (see the first illustration below), which was noted as indicative of the aircraft having been subjected to a rising current of air coming from the convective air movement associated with a cumulonimbus formation “just below it”.
Satellite data close to the time of passage (see the second illustration below) confirmed that the three areas of the most intense convective activity were around but not on the aircraft track which was between build ups exceeding FL 400 but clear of cloud until flying directly over an undetected (relatively) smaller build up directly below and less than 20 nm away from the aircraft track.
FDR recording of a sudden rise in OAT concurrent with the rapid updraught. [Reproduced from the Official Report]
A satellite image of water vapour intensity timed just after the turbulence event had occurred. [Reproduced from the Official Report]
It was concluded that when the sudden severe turbulence occurred, the aircraft had probably been under the indirect influence of two of the three major convective systems in its vicinity and under the direct influence of another (developing) one below it. The latter would have been the source of both the upward air current which disrupted the flight path and the accompanying (dry) hail ejected from its cloud top. Industry guidance seen by the Investigation noted that turbulence is not only associated with flight inside a cumulonimbus-type cloud but also occurs in the clear air around them. This means that it is necessary to take action to avoid such clouds by a significant margin such as keeping at least 5000 feet vertically and 20 nm laterally from active convective cloud formations.
The Investigation also noted that Airbus’ 2007 guidance on the optimum use of weather contained a reminder that tilt determines the range at which a convective cell return will disappear at a given range - e.g. a return disappearing at 40 nm with a -1° tilt has a top 4000 feet below the level of the aircraft - and advocated regular vertical scanning by varying the tilt. This guidance also included the advice that “efficient weather awareness” in the cruise was best achieved with a different range set on each pilot’s weather radar display, as a general “rule of thumb” using 160 nm (and a tilt setting of -1.5°) on the PM display and with 80 nm (and a tilt setting of -3.5°) on the PF display. These settings, in conjunction with the recommended 5000 feet vertical avoidance, would have given the flight crew a much better situational awareness of the actual turbulence risk ahead which it was concluded had been the consequence of flight through a “turbulence dome” containing just dry hail of relatively low radar reflectivity above a convective cloud top approximately 3000 feet below the aircraft.
It was concluded that the flight crew’s inadequate use of flight watch facilities to enhance their weather awareness during their passage through what they were already aware was likely to be an en-route area of heightened turbulence risk was contributory to the accident encounter.
The only formal Contributory Factor definitely determined as a result of the Investigation was:
- Adverse Meteorological Conditions - the prevailing meteorology constituted a critical factor for the event, with characteristics of intense vertical currents of air, great variation of intensity and direction that were provoked directly by the intense and fast convective process of development of a Cumulonimbus (Cb) cell which was less than 20 nm from the aircraft flight path with cloud tops approximately 3,000 feet below it.
Two Potential but Undetermined Contributory Factors were also formally identified:
- Control skills - the limits of side stick command input and angle of attack protection were exceeded causing the autopilot to disengage; the pilot attempted to counteract the effects of the turbulence and made side stick inputs opposite to the aircraft's attitude changes which may have served to increase the disturbance initiated by the turbulence.
- Radar TILT Adjustment - the setting by both pilots of the weather radar range to 80 nm and a ‘TILT’ of -1° in the absence of reflectivity in the upper section of the turbulence dome at the time of the accident may have contributed to the entry of the aircraft into a region of severe turbulence.
As a consequence of the Investigation, two Safety Recommendations were made on completion of the Investigation as follows:
- that the Brazil National Civil Aviation Agency (ANAC) disseminate the lessons learned in this report in order to raise the situational awareness of pilots, especially with regard to the identification of areas with adverse weather conditions, as well as the procedures to be adopted to mitigate the risks caused by meteorological phenomena. [A-158/CENIPA/2013 - 01]
- that the Brazil National Civil Aviation Agency (ANAC) work with the operator to reinforce the appropriate technical procedures for the use of radar and aircraft control in the company's Training Program to avoid areas of severe turbulence. [A-158/CENIPA/2013 - 02]
The Final Report was completed on 10 May 2018 and published online in both the definitive Portuguese version and in an official English language version on 15 May 2018.