Archive

Monthly Archives: March 2013

On March 13th 2012, an incident took place with an Air France Airbus A340-300 on approach to Paris Charles de Gaulle (France).

A summary of the report is given here: http://avherald.com/h?article=45f1317a

The French BEA have offered recommendations in French: Final report available at http://www.bea.aero/docspa/2012/f-zu120313/pdf/f-zu120313.pdf

F-GLZU

As is often said, there are many aspects to an incident and no single cause can be solely attributed to an unintended end result. However, this case reminds me of a message drilled home by one of my instructors during training on the Airbus A320. We all know that Aviate is our number one priority: to fly the aeroplane in a safe manner.

This means we have to make the aeroplane do what we want. In this incident the crew wanted the Airbus on the proper approach slope and the correct speed and configuration, but was unable to achieve that result.

But how did they know that they were so far outside the normal approach parameters that a go-around was inevitable? Reading the report, it certainly seems that they did not in fact know that this was the case. How do any pilots find themselves saying “This isn’t going to work.”.

We do it by asking the question “Is this going to work?”. During a normal approach we have distances or heights mandated (either legally or by airline Standard Operating Procedure, SOP) at which we are required to be stable within the normal parameters in order to be allowed to continue the approach. We ask the question at these points, and if the answer is “no” we perform a go-around or discontinued approach. In this case, having crossed the final approach point higher than they should have been, they either

  1. Specifically chose to continue
  2. Were simply unaware of their speed/height/distance – they didn’t ask the question “Is this going to work?”.

In the first case, choosing to continue regardless, what motivating factors would have made them continue? Not wanting to have to accomplish a go-around and second approach, followed by paperwork and explanation? Not having a sufficiently strong SOP culture, either due to there being no requirement to abandon the approach, or working in an environment where the crew felt they could get away with ignoring the SOP?

In the second case, simply being unaware, could fatigue or tiredness have contributed to slower processing of a rapidly changing environment? Did they brief themselves that they may be kept high and, if so, what modes would be selected on the autopilot control unit in order to still make a safe and stable approach? If not… did the element of surprise cause such tunnel-vision that they simply forgot to ask the question… “Is this going to work?”

In any case: not having been there, it is sensible to assume that we are all capable of making the same mistakes. So, we put aside blame on CRM, SOP adherence, ATC or fatigue and look at how to handle the situation. That situation is that we are now so far above the normal approach profile that the aircraft starts to pick up spurious signals from the runway transmitters, causing it to start climbing instead of descending. No bad thing to be climbing away from terrain, except that in this case the spurious signals commanded such a sharp climb by the autopilot that the speed of the aircraft washed off, bringing the aircraft close to a stall. For another example of why this is bad, see the Dutch Safety Board final report in to Turkish Airlines 1951 at Amsterdam Schiphol, here: http://www.onderzoeksraad.nl/en/index.php/onderzoeken/Neergestort-tijdens-nadering/#rapporten

Turkish 1951 Amsterdam

Thankfully, the crew of Air France 3093 noticed the sharp climb, reacted by disconnecting the autopilot and shoved the nose of the aircraft forwards in order to accelerate. The autothrust disconnected, adding further distraction to the task of ensuring the Airbus A340 remains airborne. Having stabilised, the crew FINALLY made the decision to go-around.

But that is not where the story ends. Having been put in to a near-stall by incorrect autopilot commands 30 seconds previously, the crew decided to make an automatic go-around. In order to do this, they re-engage the autopilot. When the autopilot re-engages, it picks up a false runway approach slope again and now enters a nose-dive back towards runway 08R. Again disconnecting the autopilot, a manual go-around is finally flown.

It is easy to say that the go-around on 08R is a simple straight-ahead climb to 4000ft and that the autopilot was not to be trusted. So why was it re-engaged again for the first go-around attempt? Reversion to an automatic motor-response in a time of stress? Reluctance to make a manual go-around when tired and surprised by what has just happened? What about focus on manual flying skills brought about by the investigation in to Air France 447 which crashed in the South Atlantic ocean following a …stall situation?

(Incidentally, the A340 involved in this incident was also previously involved in an incident over the South Atlantic bearing eerily similar hallmarks to that of the fateful A330 of AF447 – for further information see http://avherald.com/h?article=44280b2a)

We can be grateful for two things: the aircraft and occupants are now safe; and the incident has reminded us of the need to always fly the aircraft in a safe manner regardless of the external factors.

Once a month (normally), commercial pilots pick up their new roster to see what they will be doing, or at least expect to do for the next month.

Seeing a CRM – Crew Resource Management – training session on the roster often results in a small inward sigh, not least because they are often combined with ‘business awareness’ training, ‘customer expectation’ training and other things which, for many pilots, are clear from previous careers in corporate business and customer service.

Sometimes, however, sad accidents remind us in commercial aviation of the need for a flight deck operation which is coordinated and well-planned. The case of Afriqiyah 771 (Airbus A330-200, registration 5A-ONG) brings to mind the harsh reality that, in a scrappily-managed and poorly-briefed approach, the often-present fatigue can combine with some handling mistakes to result in the total loss of an aircraft and almost all occupants.

How can we always be expected to properly brief our intentions when the other pilot is clearly tired, bored, irritable or just looking at something interesting out of their window?

Good CRM is about more than just keeping their interest for long enough to share a mental model. It serves as a backup and safety net for situations where SOPs are ignored (such as flying below the minimum altitude to get visual with the runway before performing a go-around), when holes in training or currency lead to poorly-flown go-arounds, or perhaps where fatigue leads us to succumb to tricks of the mind, such as somatogravic illusions (accelerating at a normal nose-up angle leads to mistakenly believing a climb is steeper than it really is, maybe leading to an unintentional push forwards on the controls).

Briefing these risks in advance may for many pilots be a secondary, less-interesting aspect to flying than, well, the actual flying. The truth is that we manage risk for a living: this assessment is our primary job. Our CRM (non-technical) tools are there to catch us when the other holes in the cheese line up. Indeed, they will probably keep us away from the go-around in the first place.

Image