Some review of what statistics there are may be in order.
The A320 family was introduced in 1988 and until October 2006 there were 2,880 aircraft of this family delivered (Flight International, 24-30 October 2006). That amounts, assuming a constant rate, to 160 aircraft per year, assuming no attrition (there has been some, of course, namely 18 hull losses, but I am ignoring it). Taking a service year to have 333.33 days and taking each aircraft to fly an average of 6 legs per working day, that is 2,000 landings per aircraft per year. At 160 aircraft per year from 0 (at this point in 1988) to October 2006,
that is a total of 18 years at 160 aircraft per year, which amounts to 27,360 aircraft years (take the arithmetic mean of 18 and multiply by 160). At 2,000 landings per aircraft per year, that is a total of between 54 million and 55 million landings.
Of those (let us take it as) 55 million landings, there have been 3 in which some issue related to thrust control has resulted in an accident, with a total of 3 deaths (ground deaths, no passengers and no crew). These incidents are
* 22 March 1998, at Bacolod
* 28 Aug 2002, at Phoenix
* 18 Oct 2004, at Taipei-Sung Shan
Even if we suppose that TAM Congonhas on 17 July was an addition to that list, that is still less than one serious problem every 10 million landings.
Just to put that in perspective, the U.K. Health and Safety Executive considers a risk to be "broadly tolerable" if it is less than one in a million per year. British roads are the second-safest in Europe, and one has one chance in 20,000 per year of dying on the roads in the U.K. The "background level of risk" from all sources is taken to be one in a hundred per year.
So, at most one thrust-control screw-up every ten million landings. If we take the average flight in an A320-series airplane to be 1.5 hours (including the customary 15 minutes for ground ops), that amounts to at most one thrust-control screw-up every 15 million flight hours. How many total career hours on A320-family airplanes are people likely to have? Say 10,000? That is, just one in 1,500 career A320-series pilots may have a thrust-control problem on landing just once in hisher career.
I don't see much of an argument there for querying the logic of the system design. With the best available digital-system design principles we can only guarantee freedom from error to within about one in 100,000 logic actuations, and we are talking about this system design to be about 100 times more reliable than that.
I was taught, from day 1 of my flight training:
* stabilised approach
* approach speed about 1.3 times stall speed
* gear down (and welded
)
* flare when within appropriate distance from ground
* throttle to idle
As far as I know, this is standard procedure on all airplanes.
When designing automation for flying, then one principle enunciated by Charles Billings and others which has gained currency is that of "human-centered automation". That is, in a device with human operators, the automation should be designed to support the human-operator tasks. Since this list above is more or less axiomatic, the question becomes how automation could best support these actions. Since the pilot must fly the airplane, heshe must ensure a stabilised approach and not the airplane - there are times when one
wants to deviate (to avoid hitting intruders, for example). V.Ref is ensured through autothrottle. Gear down is aided through warnings and inhibitions, as is throttle retardation.
The question whether the thrust should automatically be set to idle at flare is a question of subtleties and extraordinary situations which would occur, if at all, then very rarely. As has been pointed out (subsequent to this post) by others, the human mind is limited in its ability to conceive of these rare extraordinary situations, so one must be very careful about reaching too far.
For all but a handful of cases in 55 million landings, relying on the pilot to retard the throttle upon flaring has proven to be appropriate, as one might expect. Because all pilots are trained to do it from Day 1. Before the phenomenon, of some rare pilots in some very rare landings in the A320 not retarding thrust levers, is well understood, it seems to me premature to suggest building in prophylaxis in the landing logic, unless one can be assured that it introduces no new anomalies into other rare situations.
This is an edited post, the original being shortened due to my apparent inability to handle the editing SW.