The sheer lack of airline accident data in the first six months of 2023 makes this year – to its mid-point at least – exceptional, in that there were so few fatal or serious incidents globally.

The number of commercial airline fatal accidents – worldwide – was only one, and the number of deaths 72, and this single fatal crash did not involve a mainline carrier or a jet; rather a regional turboprop aircraft.

Yeti Airlines ATR wreckage

Source: Skanda Gautam/SOPA Images/Shutterstock

Yeti Airlines crash took place in good weather and with experienced pilots

To compare like with like in safety terms, the last decade’s best pre-pandemic half-year had been 2015, when there were five fatal accidents and 65 fatalities. In the first six months of 2022, which witnessed the beginnings of air travel’s commercial recovery, the respective figures were six and 186.


This year’s performance to 30 June ought, therefore, be a reason to celebrate. Celebrating good safety statistics, however, goes against the instincts of safety analysts, especially when considering results from such a short period.

The rationale for that scepticism is that these good results may prove to be a brief anomaly. So FlightGlobal will retain its sang froid for the time being, and will look again at the industry’s performance at the end of this year.

Among all the accidents in this year’s first half – including non-fatal mishaps – only the Yeti Airlines ATR 72-500 crash in upcountry Nepal merits serious discussion. This loss aside, our listing of non-fatal events contains a rather mundane litany of familiar mishaps: mishandled approaches and landings, runway excursions/overruns, and routine (but consequential) technical problems.

Is this mundanity something to celebrate in its own right? Only in the sense that there were few serious injuries – but on the other hand, if the industry is succeeding in stamping out major accidents, perhaps it could work on eliminating the carelessness or complacency probably involved in many of these “routine” incidents.

The Yeti Airlines crash on 15 January took place in good weather and with two experienced pilots at the controls. The more senior pilot was in the right-hand seat, operating as pilot monitoring (PM) and checking out the pilot flying (PF) in the left-hand seat to clear him to fly approaches at Pokhara airport. Like many upcountry airports in Nepal, the surrounding terrain and changeable weather can make approaches and departures challenging.

On this fair-weather day, however, the technical challenges of the visual approach the pilots chose to fly do not look as if they were the cause of the accident, which killed all 72 people on board.

According to the Nepalese investigators’ preliminary factual report, the twin-turboprop was established on the visual downwind leg of a circling approach which was intended to lead to a landing on runway 12. The flaps were set to 15° and the gear was down.


When the PF called for flaps to be lowered to 30°, the PM responded “flaps 30 and descending”. But according to the flight-data recorder (FDR) the flaps were not descending. This raises the question which lever did the PM actually move, because at that moment the propeller speed reduced to 25% and the engine torque to zero.

There are six levers across the throttle quadrant on the ATR 72’s centre pedestal. From left to right these are: the parking brake; the left engine power lever; the right engine power lever; the left engine condition lever; the right engine condition lever; and finally, the flap lever.

ATR 72 throttles

Source: ATR

ATR crew may have moved wrong levers before Nepal accident

The levers are each designed to look and feel different according to their purpose. For example, the parking brake is a simple lever angled toward the captain’s seat by about 30°, the power levers’ tops are rounded, the condition lever handles have a rectangular feel, and all these are coloured black – except the flap lever, which is white. The flap lever also features a distinctive top that is supposed to represent a cross-section of the aerodynamic shape of the flaps.

The controls that can order the propellers to feather are the two condition levers: just to the left of the flap lever. When fully retarded, the condition levers shut off the fuel supply to the engines. When the pilots want the engines to run normally, they set the condition levers to AUTO: the forward position. There is an intermediate setting position marked FTR (feather) between shut-off and AUTO.

Propellers on an ATR can auto-feather in the event of a power failure, but the system is designed to prevent this from happening on both sides at once.

Nepalese investigators are still working on the inquiry and have not spelled out exactly what they believe happened. However, when – according to the FDR – the propellers stopped delivering thrust, the crew did not remark on the power loss, but carried out the before-landing checklist and began the left turn toward final approach.

After a few seconds the PM suggested that the PF apply a little more power, and just after that, the flaps were set to 30°, without any request from the PF or status report from the PM.

Pokhara air traffic control cleared the ATR 72 to land, and in response the PF stated twice that there was no power from the engines. A few seconds later the stick-shaker – which indicates the risk of imminent stalling – activated twice, with the second coinciding with the dramatic left wing-drop that sealed the aircraft’s fate. By this time the PM had taken control, but to no avail.

Global safety performance to the mid-point of this year was definitely exceptional, but one thing seems still to be true: years of global airline accident data shows that almost all serious crashes over the last decade or more involved small or medium-sized propeller-driven aircraft operated by commuter, regional or freight operators.

It is not the propellers that are the problem. Other indicators provide clues as to what that might be. Year after year, most such accidents take place in nations – like Nepal – that have, statistically, a below-average safety score in terms of global standards.


So, there is an industry cultural problem here. Not national culture, but safety culture within the national industry. That culture relates to how seriously safety – and human life – is taken within the government’s transport department, the national aviation authority, and the individual airlines, right down to the training of individual pilots and engineers, the quality of which influences their attitudes to their job.

Among countries with below-average aviation safety performance, Nepal and its aviators face particularly serious challenges, given the country’s extraordinary terrain and the fickle weather that goes with it. Having such challenges to face, however, should not degrade safety. Nepal has a duty to its air travellers to become the world’s expert in navigating its own local terrain, and flying safely – or deciding when not to fly – in its extreme conditions.

All countries whose aviators routinely face extreme or unusual conditions have a duty to become experts in the challenges unique to their environment, and to be proud of that expertise.

Separately, China, whose good safety record for nearly two decades was broken by a China Eastern Airlines Boeing 737-800 crash on 21 March 2022 that killed all 132 on board, has still not published its report on that disaster.

On the first anniversary of the accident, however, the Civil Aviation Authority of China issued a statement – in Mandarin only – saying simply that it is continuing a “meticulous and rigorous technical investigation”.

The aircraft – registered B-1791 – plunged from its cruising altitude of 29,000ft to the ground without an emergency call. ICAO standards call for a completed report within one year, or alternatively a release of available facts at about the time of the anniversary.

China Eastern 737

Source: AirTeamImages

Report on crash of China Eastern Airlines 737 in 2022 has still not been published

IATA has highlighted the importance of timely accident data and called for ICAO signatory states to meet their obligations to investigate thoroughly and report in a timely manner. IATA director general Willie Walsh said in June that fewer than half of all accident reports meet the standards for thoroughness and timeliness.

In March this year the US Federal Aviation Administration (FAA) called a Safety Summit in Washington DC mainly because of concerns about a series of near collisions at US airports caused by some form of pilot or controller mistake or omission.

US National Transportation Safety Board (NTSB) chairman Jennifer Homendy said that six of these events – none of which led to an actual collision – had one thing in common: there was no cockpit-voice recorder (CVR) material of the events, because the information had been overwritten.

The NTSB has made a formal recommendation that the present mandated recording capacity of CVRs – which is 2h – should be increased to 25h – a requirement already implemented in Europe. The FAA has responded that it is “committed to addressing the NTSB recommendations” and is initiating rulemaking on the subject.

The FAA has also promised to set up an Aviation Rulemaking Advisory Committee to investigate how to make the best use of recorded information, including increased use of airline flight data monitoring (FDM).

The FAA proposal to capture more FDR/CVR data – and to take advantage of FDM where appropriate – harmonises perfectly with Boeing’s continuing efforts to improve safety feedback at all levels, both within the company and from its customer airlines. This stems from the manufacturer’s commitment in 2020 to overhaul its corporate safety management system (SMS), which had been shown by the two infamous fatal 737 Max crashes to be seriously inadequate – a verdict in which the FAA’s poor oversight of Boeing was also implicated.

On 23 May, Boeing’s chief aerospace safety officer Michael Delaney released the company’s second annual report on SMS progress. A February safety conference in California saw participation by 90 airlines, 200 delegates, the FAA, and four pilot associations. If the first year’s conference had seen a concentration on internal transparency and improved employee feedback, 2023’s version focused on setting up a much-improved system of feedback from customer airlines and national aviation authorities (NAAs).

Boeing now embeds its own pilots (operations experts, not engineering test pilots) with airlines – more than 60 so far – to improve its operations interface with the carriers. It calls this the Global Aerospace Safety Initiative, and it is intended to improve its own awareness of local concerns and generate better feedback on training and technical issues from the line-pilot level.


Delaney says the airlines so far have been enthusiastic about this initiative, but adds that if any reject the approach Boeing will have to consider whether delivery of its aircraft to a non-participating carrier should go ahead.

Boeing says it is similarly improving its participation with NAAs, and has replaced its task-based approach to pilot training with competency-based training and assessment (CBTA). The company admits it has been way behind the airlines and training industry in doing this.

As an essential part of the process of tailoring training according to demonstrable needs, Boeing is cooperating with customer airlines to use pilot performance data from FDM. So far it is delivering CBTA training to four airlines with the 737NG, 737 Max, and 787, and this will expand. If a system like this had been in place with Lion Air and Ethiopian Airlines back in 2018, the two Max 8 disasters would probably not have happened.

Delaney says Boeing is “not there yet”, referring to the SMS upgrading process as a journey that continues. And he says he is still “on the fence” about the part that artificial intelligence (AI) will play in future safety strategies.

Meanwhile, across the Atlantic, the 2023-2025 European Plan for Aviation Safety notes that “the European Union aviation system is emerging from the pandemic without having lowered its high safety standard”, and its emphasis for the future is on maintaining a safety structure that acts on data, and which is flexible and resilient rather than formulaic and compliance-based.

Europe anticipates that AI – particularly machine learning (ML) – will advance further automation, and warns this “will be likely to modify the paradigm of interaction between the human and AI-based systems”. In straightforward language, that means Europe is predicting ML will open the path to reduced crew operations and, in some sectors, autonomous operations, but it does not attempt to forecast when such things could happen.

Time will tell whether 2023 remains an exceptionally safe year in practice. Meanwhile, technical and operational change, as always, is in the pipeline, and change always presents safety challenges of its own.

Supporting documents

Click link to download and view these files