On 5 September 2014, an Embraer 145LR (N16954) being operated by ExpressJet Airlines on a scheduled domestic passenger flight from Grand Rapids MI to Newark NJ as flight 4538 and en-route over Canada in day VMC entered a large and active thunderstorm at FL370 and the flight crew lost control in IMC and severe turbulence for around 3 minutes during which 4,000 feet of altitude were lost and Mmo was exceeded. None of the 29 occupants were injured and the aircraft was subsequently found not to have been damaged.
The event was not reported to the Canadian TSB by the aircraft operator or the pilot in command of the aircraft but it was so reported the following day by the Toronto ACC, which had been working the aircraft at the time. Neither the CVR nor FDR were quarantined after the flight and relevant data on the CVR were therefore overwritten. The Investigation was subsequently provided with a download FDR data which included the flight under investigation.
It was found that the Captain, who had been PF for the investigated flight, had accumulated approximately 10,000 flying hours which included 7,000 hours on the incident type of which 5,000 had been in command. He had been employed by ExpressJet for 10 years and was a Company Check Pilot. The First Officer had accumulated approximately 3,400 flying hours which included 2,200 hours on the incident type during his 3 years employment with ExpressJet.
It was established that the departure of the flight from Grand Rapids had been interrupted during taxi out by a "ground stop" due to thunderstorm activity in the vicinity of the aircraft. After a 45 minute delay during which the flight crew considered if and how to deviate from their flight-planned route to avoid severe weather, the aircraft took off and flew to a position around 50 nm north of Grand Rapids before taking up a generally easterly track which allowed it to fly parallel to a line of thunderstorms in VMC. The aircraft was initially climbed to FL330 and then to FL370, the maximum operating altitude for the aircraft type, and was working with Toronto ACC. Clearance to deviate from track for weather avoidance was obtained.
Opportunities to turn south through gaps in the line of storms to the right were observed to be rapidly diminishing and, already experiencing moderate turbulence, 23 minutes after take-off, the crew contacted Company Dispatch by ACARS to request a suitable route through the storms. Based on flight-following weather software designed to inform proactive weather avoidance rather than as a means to select tracks through areas of adverse weather, the Dispatcher suggested turning south on track towards Sarnia which the Captain then initiated into IMC. Thrust was increased and the result was a significant and sustained exceedance of the AFM Recommended Turbulence Penetration Speed of M 0.63 for unexplained reasons.
Almost immediately, as turbulence increased rapidly from moderate to severe, there was a transient minor exceedance of Mmo, which the Investigation considered was "consistent with the aircraft encountering horizontal gusts associated with the mature cumulonimbus clouds into which it was flying" (see the Composite Weather Radar picture reproduced below). The AP disconnected as Mmo was exceeded and the aircraft then climbed above FL370 without ATC clearance. The flight crew responded by re-engaging the AP but it almost immediately disconnected as the range of vertical acceleration increased to 0.2g - 1.7g. As the speed decreased through M 0.68, the aircraft continued to climb reaching FL 376 with an un-commanded right bank of 45°. During continued severe turbulence, the AoA was recorded as having "fluctuated rapidly within the range of −11.9° and 3.8°".
Shortly after reaching the maximum upward deviation from FL370, FDR data showed that the aircraft had pitched to 11° nose down and rolled 30° to the right. In response, the crew had, for no discernible reason, increased both thrust and the bank to the right with a resultant 63° right bank angle. Severe turbulence continued, and pitch attitude, airspeed and rate of descent all continued to increase, the latter eventually peaking at 9,300 fpm. Right bank increased to 77° due to crew action which, for the second time, was contrary to widely-promoted aircraft upset recovery techniques. As the aircraft descended through FL364, the rate of descent had increased to 7,500 fpm and the pitch angle to 24° nose down with severe turbulence persisting.
At this point, the crew applied left roll inputs and the right angle of bank decreased to 26°. Passing FL 355, the level of turbulence began to lessen from severe to moderate and, following a crew-commanded reduction in thrust, aircraft pitch attitude and rate of descent both began to reduce. Airspeed continued to increase to a maximum of M 0.80 until just before the minimum altitude of FL336 was reached some 23 seconds later. With the un-commanded descent finally arrested, the climb back to FL370 began, which took about 3 minutes and during which the level of turbulence reduced to light and icing conditions ceased.
During the loss of control, which had lasted for several minutes, ATC had made a number of attempts to re-establish radio contact with the aircraft but had only received a 'stand by' response from the aircraft in a "tone and volume that suggested that something of an urgent nature was occurring". Crew action to increase an initially un-commanded right bank had resulted in a 200°/minute average rate of turn which had changed the aircraft heading from the intended 180° to 240° and on this heading, and still IMC but now back in control of the aircraft, the crew had requested a heading to "exit the weather system". ATC suggested a southbound heading which was followed and "soon afterwards" the aircraft emerged into VMC and the aircraft turned eastward to proceed to its intended destination without further event.
The track flown superimposed on the NOAA ground-based composite weather radar picture up to the point of loss of control. [Reproduced from the Official Report]
The Investigation compared the actual severe weather encountered and the forecast weather for that time and concluded that the former had been essentially as forecast, with cloud tops in the area which penetrated at up to 52,000 feet. It was found that the flight crew had been fully aware of the likely conditions prior to departure including the content of a SIGMET which warned of severe turbulence between FL 330 and FL 390 associated with a cold front lying east-west across the intended route. It also found that the weather radar picture available to the Dispatcher in the ExpressJet Flight Following Unit was generated using just the lowest reflectivity angle and displayed a base reflectivity product aimed at proactive flight planning such as re-routing around an entire area of storms rather than tactical guidance within such an area which would be better informed by the onboard weather radar. It was also noted that FAA regulations on the subject of the aircraft commander's responsibility for the safe operation of an aircraft were unequivocal, whatever advice may be provided by others such as, in this case, flight-following Dispatchers.
Whether the Flight Crew had optimised their use of the onboard weather radar could not be assessed since the FDR did not record parameters from the weather radar unit. No evidence was found to suggest that the weather radar was faulty in any relevant respect but it was noted that the equipment required manual rather than automatic control of tilt.
It was noted that "the roll inputs by the flight crew during the descent phase of the loss of control were contrary to known aircraft upset-recovery techniques" and by increasing rather than reducing the angle of bank had increased the altitude loss and recovery time. It was also noted that although both pilots had received upset-recovery training, the investigated event was their first exposure to an upset attributable to such severe turbulence. It was considered that both the fidelity limitations inherent in current full flight simulators and the inevitable element of expectation inherent in simulator training scenarios would preclude a realistic replication of a convective weather penetration episode like that encountered.
Finally, it was noted that the unintentional exceedance during the upset of both the AFM Maximum Operating Altitude FL 370 by 600 feet and the AFM maximum operating Mach Number M 0.78 by M 0.02 had introduced the risk that flight safety would be compromised "resulting in injury to the occupants or damage to the aircraft".
The formally stated Findings as to Causes and Contributing Factors were as follows:
- The flight crew were aware of the weather conditions and departed, anticipating that they would be able to navigate through an extensive line of thunderstorms. However, as the flight progressed, the line of thunderstorms intensified and, as a result, the flight crew’s intended route became obstructed.
- The flight crew operated the aircraft through a mature thunderstorm and, as a result, an aircraft upset and loss of control occurred.
- During the loss of control, the flight crew momentarily applied flight control inputs that exacerbated the roll attitude of the aircraft. As a result, altitude loss and recovery time were increased.
The formally stated Findings as to Risk were as follows:
- If flight crews operate aircraft outside of manufacturer recommendations, the risk of encountering an adverse consequence is increased.
- If aircraft are operated outside of manufacturer limitations, there is a risk of compromising flight safety, resulting in injury to the occupants or damage to the aircraft.
- If cockpit voice recordings are not available to an investigation, the identification and communication of safety deficiencies to advance transportation safety may be precluded.
Safety Action taken by ExpressJet Airlines as a result of the event was noted as having included the following:
- Improved the use of flight-following software as a source of weather advice to fight crews.
- Enhanced training on various operational matters including the policy and procedures related to adverse weather phenomena and the suspension of flights, the duties and responsibilities of Dispatchers, the interpretation of weather radar displays and the function and use of AIRMETs, SIGMETs and PIREPs.
- Additional training for pilots on the effective use of weather radar when seeking to identify developing storm activity and to avoid severe weather.
The Final Report of the Investigation was authorised for release on 8 June 2016 and it was officially released on 6 July 2016. No Safety Recommendations were made.