Originally Posted by
compressor stall
Extra crew training is just keeping a link in the chain.
A modern civil airliner should be designed not to rely on the flight crew having to adopt non standard (unique?) piloting techniques to counter a faulty bandaid put there to mask inherent aerodynamic flaws.
So, there was Qantas with an A380, where the assumed failure modes of certification didn't bear much resemblance with the real world. There was also Qantas' A330 FBW going nuts, and the B744 where the O2 bottle went cross country out the side of the plane. There was UAL 811, which didn't bear much resemblance with simulator sessions, Alaskan 261, UAL232, BA037, and the hudson ferry....
There are more things in heaven and earth, Horatio,
Than are dreamt of in your philosophy.
- Hamlet (1.5.167-8), Hamlet to Horatio
On the other hand, we have AF447... Adam Air, and a panoply of headlines of crew being outside the loop. Every day, crew deal with oddities with the aircraft, that come to light where reality meets assumptions and a gap exists. The majority of occasions do not end up in headlines, the crews just get on with the task, write up the events and go home and take crew rest. Occasionally, that doesn't work out, and we have bad days. The direction of air transport has added distance between the operator and the aircraft, through design and perceived safety benefits of reliance on automation. That is a high stakes assumption and it is routinely shown to be invalid. Consider AZ214 and similar bad days. Automation is great, but it is only great if it is working as expected, and where the monitoring is not degraded by the out of loop condition, or the degradation and or inertia that develops to the operator intervening with a system. All RPT aircraft fly like Cessnas and Pipers at the basic level, yet we have flight crew that are reticent to intervene and take control of the aircraft when wacky stuff starts. The reticence is a paradoxical response to the focus on compliance, reporting and rigid management directives to reliance on automation to reduce errors.
We can't design a reliable toaster, yet the assumption is made that an engineer 30 years ago can predict the conditions of weather, fuel state, fuel policy, system malfunctions, traffic, ATC capability, and crew training etc, such that errors cannot occur. We can't make windows or iPhones work without error, the system couldn't imagine foam bringing down a spacecraft, or cold temperatures mussing with sealing of an O ring... Passengers hop on an aircraft in country A, and fly for a day and night to country B, and land in weather that you won't drive your car in. The crews do that every day, and thousands of times, in ice storms, weather fronts, fog and sometimes sunny calm days. They do it with controlled aircraft flying in proximity to other aircraft that are not controlled, and hopefully the 15 hour student doesn't mess up, and enter the airspace that the punters are spending time in a tube complaining about the peanuts, coke and pretzels that are part and parcel of unregulated air travel. When we can build reliable toasters, then the necessity for human competency to be maintained may be mitigated, maybe.
The balance of reinforcement training is hard to get right, back in the late 90's there was a push to make crew more aware of handling of aircraft, and that led in passing to AA587. A little knowledge can be a dangerous thing. Getting the mix right is not easy, but we certainly do not have it at this time, and have not for many years. This is not generational, it is institutional. Last time I looked, the world was not flat, supported by turtles all the way down, it is a weird oblate spheroid and strange stuff happens that takes care to avoid having bad days, that care takes SA, and knowledge and skills to respond to the information that is provided to the crew. We have a problem in getting the loop right all of the time. 10^-9 reliability is fantastic, except for the poor pax that are the statistic. [yeah, -6 is the basic level, but the events we see that get into the headlines are much more remote than that]
PS: links in chains presume a simple model of the world; a chain is a linear analogue, and the world flat or oblate spheroid, is much more entertaining. When you place a band aid onto a cut, you have introduced new failure modes and vectors for badnesses, directly and indirectly. Have a fuel fright on one flight, add more, and have an aircraft roll off the end of the runway, Too many unstabilised approaches? demand AP/FD use form minimum engagement, and then wonder why the crew ding pods and tails, or ham fist AP faults. The world is stochastic, non linear, and every barrier that gets put up acts as much as a new surface to bounce issues off in new directions as it acts as a barrier. This is not a pessimistic view of the world, it is one that SMS programs (ICAO DOC 9859 etc) and audit processes need to be comprehended in their constraints and limitations. Functional resonance is a reality and that suggests the key to safe management is comprehensive awareness of the interaction of policy, procedures and practices with the real world.