On 14 September 2021 a Cranfield University-operated and recently acquired Saab S340B (G-NFLB) configured as a flying classroom was being used to facilitate type training of two of the new operator’s senior pilots by Loganair Training Captains current on type. After engine-out handling familiarisation, the left engine could not be restarted because the right engine starter-generator had failed. The resulting priority to achieve a landing before battery-source electrical power was exhausted was met by a successful diversion to Campbeltown led by the Training Captain in command at the time.
A Serious Incident Investigation was carried out by the UK Air Accident Investigation Branch (AAIB). It was noted that the 59 year-old Training Captain-in-command at the time had a total of 14,894 hours flying experience which included 7,208 hours on type and was occupying the right hand seat. Another similarly qualified Training Captain from the same airline was seated in the passenger cabin after completing his planned training duties in command with another trainee from the operator earlier in the same flight. The pilot in the left hand seat when the investigated problem arose was a Captain and Nominated Post Holder with the aircraft operator undergoing type training. A second trainee who had been under training earlier, was occupying the supernumerary crew seat and observing. He was the operator’s programme manager responsible for overseeing the introduction of the type into service and had previously held training qualifications for the aircraft type.
The training provided was intended to cover aircraft handling characteristics including stalling as well as an intentional in-flight engine shutdown and restart, the latter to enable the trainee to experience the handling characteristics of the aircraft with one engine inoperative noting that the available full flight simulator was considered not to replicate the engine out handling characteristics very accurately. Briefing prior to single engine operation included inability to re-light an engine.
In VMC at FL120, a scenario involving a loss of oil pressure in the left engine was followed which led to the shutdown of that engine. After a demonstration of single engine handling, briefing for the intended engine re-light was given including for a start malfunction and it was then attempted. It was not successful - a hung start occurred with the engine speed settling at only 40% Ng before dropping. As the crew were discussing the failed start, the two EFIS screens used to display a composite EADI/EHSI went blank and FMC stopped working. The Captain handed control to the trainee and established that the right engine starter-generator, essential for a cross start, had failed. The reset procedure was not successful which meant that the electrical power was now being provided only by the batteries with obvious implications for endurance.
The Training Captain in the cabin noted the time and began to assist his colleague in command. Completion of the emergency procedure for the Loss of Both Generators (Both Engines Running) and load-shedding of non-essential electrical equipment was delegated to the Observing trainee in the supernumerary crew seat and the Captain declared a ‘MAYDAY’ and requested the weather for Glasgow, Prestwick, Islay and Campbeltown. The trainee in the left hand seat continued as PF with only the standby mechanical flight instruments available and descended to FL100 whilst remaining in VMC.
The aircraft was over the sea between the Mull of Kintyre and Islay and whilst a layer of cloud could be seen below, it was also possible to see that cloud was more widespread over land - although the southern end of the Mull of Kintyre was visible. A T-DODAR review considered attempting a re-light of the left engine using battery power but since even after load shedding, and from a full charge, the two batteries would only provide 60 minutes power. It was not known how much battery power would be consumed by such a start attempt and given that there was also no guarantee that such an attempt would be successful, it was decided not to do this. It was decided to connect the two battery bus bars together even though this was not a stated action in the applicable emergency procedure.
It was clear that flight on instruments would be necessary to descend safely below MSA and possibly to conduct an approach. In the latter case, it would only be possible given the remaining aircraft systems to fly a VOR or an ILS approach set on NAV 1 since the loss of the FMC had removed the option to make a GNSS approach. Also, only the Standby OBS was available - the RMI (Radio. Magnetic Indicator) and ADF were no longer functioning. This ruled out Prestwick which only had an NDB approach and whilst both VOR and ILS approaches were available at Glasgow, the weather there was deteriorating to the extent that only an ILS was likely to be possible and the distance to get there was not compatible with likely battery power endurance.
The weather at Islay and Campbeltown was more favourable and these airports were visible through the cloud layers, but they only offered GNSS approaches. Fortunately however, the Captain in command had his Loganair EFB with him and this included a VOR approach to runway 11 at Campbeltown which was approved for company use so this was selected as the only practicable option. Radar vectors towards Campbeltown were requested and received and on reaching the MSA, the sea surface was visible through breaks in the cloud. The aircraft was established on the VOR procedure FAT (final approach track) and the runway was visually acquired at about 6 nm out and a landing was completed approximately 40 minutes after the right engine starter-generator had failed.
Why it happened
It was found that the aircraft had been acquired by the operator after previously having been in long term storage and had only flown approximately 31 hours since the change of ownership.
An examination of the right engine starter/generator found that it had, unknown to the aircraft operator, failed which meant that there was no power to enable a cross start of the left engine. This failure was considered by the operator’s MRO to have been the result of multiple starts at intervals insufficient to allow the unit to cool. Since the current operator reported that no such circumstances had occurred since they had acquired the aircraft, it was concluded that this must have occurred whilst the aircraft was in service at a previous operator. The failed starter-generator had accrued 261 hours since its last overhaul and was found on examination that its generator armature had two raised bars and damage to the brushes (see the illustration below).
The removed right starter generator armature and its lifted and damaged commutator bar. [Reproduced from the Official Report]
It was noted that initial engine starts in normal operations use a GPU power supply which the design of the aircraft only allows to be connected to the left side electrics. This means that whilst the left starter-generator regularly works under a high electrical load, the right starter-generator only does so in the event of a cross start such as the one attempted and for almost all its installed time it is only functioning at a much lower electrical load. This would potentially mean a longer life but one which may eventually make it vulnerable to a very occasional high power demand during a cross start.
The critical importance of load shedding in the event that electrical power is reduced to batteries only as in this event was noted. Without it, the endurance of the main batteries is less than 15 minutes whereas it is approximately 45 minutes if immediate load reduction is carried out and “a minimum of 60 minutes” if only essential equipment is connected to the hot battery, essential and emergency bus bars.
It was found that the AFM has no emergency checklist for the loss of both generators when relying on single engine operation but seeking to restore a serviceable second engine.
The Conclusion of the Investigation was formally documented as follows:
"The engine was shutdown intentionally as part of a training exercise. The right starter/generator had latent damage probably caused by insufficient time between starts in service with a previous operator. The right starter/generator then failed under loads higher than those it was normally subjected to, which had not occurred previously because it was not normally used for engine starts. The crew of four, who all had relevant experience, coordinated their activities to produce a successful outcome, and were assisted by the availability of a non-precision approach procedure approved for use by another operator for whom one of them also flew.
The Final Report was published on 13 October 2022. No Safety Recommendations were made.