[syhprum]

One day the Americans will lose a city due to their Imperial/SI units confusion while testing nuclear warheads then maybe they will adopt the metric system.

[graham.d]

Whilst I have great sympathy with the idea that a pilot should have a final override because there are circumstances where a good pilot will always make better decisions than computer software, it is worth looking at these stats:

http://www.planecrashinfo.com/cause.htm

They show pilot error to be the main cause of serious accidents. One danger is that introduction of sophisticated systems will further de-skill the pilot's job, reduce the training he may get and reduce the need for as good a qualification to get the job. In the example above I wonder whether if a pilot who had not been a glider enthusiast would have coped as well. Airbus Industrie's argument is that they reduce the chance of getting into situations where the pilot has to make such decisions. I don't think there are enough comparable statistics to judge at the moment.

As a matter of interest I worked on some of the early fly-by-wire systems (chip designs). Fly-by-Wire systems really just replaced direct mechanical linkages with electronic systems. There were quite a few people who did not think this wise too as there were people who did not like control lines replacing with hydraulics. Of course these systems do not remove decision making from the pilot, but they did remove a degree of physical feedback - though actually many later aircraft give artificial physical feedback.

I think the concept in a modern aircraft is to allow the pilot to fly the plane simply and not have to try to manage too many functions simultaneously. The pilot who landed on the Hudson River did really well - if a wing had dipped a bit it could have flipped the plane over. By the way, it was an Airbus A320 with the computer doing a lot to manage the attitude of the craft.

[Geezer]

Graham - I agree. Finding the right balance between experience and automation is the challenge. BTW, did you see in the news recently that a flight bound for Minneapolis overshot its destination by a about a hundred and fifty miles? The air traffic controllers could not get the crew to respond on radio. They were about to scramble fighter jets to intercept it when they finally managed to contact the crew! Precisely what they were up to is still a matter of conjecture.

I don't know much about avionics. Do you happen to know if there is a trend to consolidate all the fly-by-wire controls into a single computerized system, or are they deliberately implemented as discrete control systems? These days I would think it's entirely possible to control an entire aircraft from a single computer so that it's no longer quite "fly-by-wire", more "fly-by-software". This would look like a weight, power and complexity advantage, but I also think it would be extremely dangerous.

[graham.d]

I am no expert on avionics either even though I worked on a small part in the past. All avionics systems have multiple redundancy and a lot of effort goes into ensuring there are as few as possible common components whose failure would be critical. This extends to software, so not only are there multiple computers with "voting rights", but also the critical software is written by teams operating independently. In passenger aircraft, safety will trump weight, and electronic systems are generally not heavy. The fly-by-wire technology reduced the weight of copper wire considerably as all the data goes via a serial bus (multiple and separately routed for redundancy).

I didn't know about the Minneapolis flight. It's hard to know how they could do that because everyone has GPS now even if flying manually. 150 miles is not far though - they may have mistakenly thought they had much further to go and had their attention drawn to an in-flight snack :-) I guess that they should have been monitoring ATC though.

[Geezer]

Thanks for the reassuring information!

[syhprum]

When I used to come home from Hamburg in the eighties the computer used to bring the plane to about 10M above the long LHR runway then what seemed to be an interminable time later the pilot would wake up "oh yes we are supposed to be stopping here" and switch it off and we would come down with a bump.

[Geezer]

When I used to come home from Hamburg in the eighties the computer used to bring the plane to about 10M above the long LHR runway then what seemed to be an intermingle time later the pilot would wake up "oh yes we are supposed to be stopping here" and switch it off and we would come down with a bump.

Yes. I've noticed something like that quite often. I think it might be something to do with "ground effect". A plane tends to float on a cushion of air the closer it gets to the ground. Sometimes it seems as if they really have to force them on to the runway. Other times they seem to be able to land so gently you don't even notice they have landed. Any pilots out there?

[graham.d]

Syphrum, are you saying that you think the pilot decided to disengage auto landing because the hold-off period was too long? I would suppose he could do this but the only reason I can think of would be because he felt that it would delay his getting the aircraft to its parking bay because it would involve a longer period of taxiing back. LHR runways are quite long. You get two effects approaching the ground, ground effect and wind shear (plus residual turbulence from previous aircraft). Normally landings are into wind and, because the wind drops as you get nearer the ground, the aircraft's airspeed falls as you descend, which helps to counter ground effect and lets the aircraft get down gently. I related a personal incident in another post where I had to land downwind and where it can be hard to get on the ground in the usual way. Flying the aircraft on to the ground can be more bumpy and it can be hard to stop the craft wanting to get airborne again :-). It is a technique used though and is especially recommended if damage to an aircraft has rendered knowledge of the stalling speed unreliable.

[LeeE]

When I used to come home from Hamburg in the eighties the computer used to bring the plane to about 10M above the long LHR runway then what seemed to be an intermingle time later the pilot would wake up "oh yes we are supposed to be stopping here" and switch it off and we would come down with a bump.

Yes. I've noticed something like that quite often. I think it might be something to do with "ground effect". A plane tends to float on a cushion of air the closer it gets to the ground. Sometimes it seems as if they really have to force them on to the runway. Other times they seem to be able to land so gently you don't even notice they have landed. Any pilots out there?

When a tricycle landing gear aircraft lands (as opposed to a tail-dragger, where the third wheel is at the tail of the aircraft), the aircraft has to be stalled to get it on the ground and the skill in making a smooth landing is to get it to stall just as the gear touches the ground, so that it doesn't simply drop on to the ground.

The aircraft has to come down the glideslope at a speed safely above the stall speed, so that there's no risk of it crashing short of the runway, but this means that once over the runway it has to shed the excess speed.  To do this, the nose-up attitude of the aircraft is increased (called the flare), which increases its effective cross section, which in turn increases its induced drag and slows it down.  However, increasing the nose-up pitch also increases the Angle of Attack (AoA) over the wings, which increases lift, which tends to keep the aircraft in the air.

The actual stall speed depends upon both the speed of the aircraft and its weight and although the landing weight of the aircraft can be calculated quite accurately, based upon fuel level sensors and loading records, getting it perfectly right, so that the stall occurs just at the touchdown point, requires a very high sensing rate, whether it be by an autopilot/Flight Control System or the pilot's backside.  Another factor that complicates the issue is that runways are not perfectly flat or level and there may be quite an appreciable slope to the runway.  While it's relatively easy to land 'uphill', landing 'downhill' is more problematic because the runway is dropping away beneath you, so getting the stall just right is even harder.

The effect of a tailwind, if is relatively steady, is mostly just to increase the ground speed, but this then means you'll be further down the runway when you touch down and have got to slow down from a higher speed, so it's likely that a pilot will be more inclined to just get the aircraft on the ground and stopped before he runs out of runway than making the smoothest of landings.

Most land based aircraft cannot be flown into the ground without sustaining some degree of damage - have a look at this MD-80 that failed to flare:

http://www.youtube.com/watch?v=x9deMKE-iek

Not only does the tail fail off (because it carries the engines, which are very heavy) but the gear nearly collapses and it's also possible to see the fuselage flex, and nearly buckle.

However, aircraft that land on aircraft carriers are just flown onto the 'deck' as there's no room for a flare.  These aircraft have to be specifically strengthened to do so though, both in regard to their landing gear and their structure, as not only do they have to cope with the high vertical speed at touchdown but also the rapid deceleration once they've caught the wire.

[Bored chemist]

Just a thought.
All aircraft are flown by computers; in practically all cases the computer concerened is between the pilots' ears.
The difference is that in one case the software has been in development for rather longer than the other. The problems highlighted earlier are not due to the use of computers- just that the computers were not programed properly.

[graham.d]

Lee, there is flying into the ground and flying into the ground. Good grief, in the video the pilot made no attempt to reduce is descent slope at all. Perhaps he was an ex-navy pilot thinking he was hitting the deck of a carrier. There is a compromise that can be reached without such disasterous results. It can be done quite safely. Carrier pilots do it because they have to hit the deck in a particular region to catch the arrester system and, also, the deck is likely to be going up and down just to make it harder.

In fact this site suggests that (gently) flying the aircraft into the ground is commonly used. You level out close to the ground and cut the engine, but the implication is there is no need to wait for a stall to drop you down. 

http://www.auf.asn.au/groundschool/umodule12.html

[Geezer]

Thanks guys! These are very informative posts.

BTW, here's some of Captain Sullenberger's take on it

http://cityroom.blogs.nytimes.com/2009/11/15/sullenberger-takes-issue-with-new-book/

[Eric A. Taylor (OP)]

One day the Americans will lose a city due to their Imperial/SI units confusion while testing nuclear warheads then maybe they will adopt the metric system.

There is currently a ban on nuclear weapons testing so it's not even being done anymore so you can relax. I have a friend who's a cop. He pulled over a guy for speeding and he told her he got confused between MPH and KPH. Unfortunately for him, If this were true he was speeding thought he was going much faster than he was. A kilometer is much shorter than a mile (.61 mile to 1 kilometer) so 61 MPH is 100 KPH.

I'm used to miles and feet but I'd prefer using metric. It's just a mater of getting used to the new system. They are now teaching metric to kids in school so maybe the switch will take place in the next generation.

[LeeE]

Lee, there is flying into the ground and flying into the ground. Good grief, in the video the pilot made no attempt to reduce is descent slope at all. Perhaps he was an ex-navy pilot thinking he was hitting the deck of a carrier. There is a compromise that can be reached without such disasterous results. It can be done quite safely. Carrier pilots do it because they have to hit the deck in a particular region to catch the arrester system and, also, the deck is likely to be going up and down just to make it harder.

In fact this site suggests that (gently) flying the aircraft into the ground is commonly used. You level out close to the ground and cut the engine, but the implication is there is no need to wait for a stall to drop you down. 

http://www.auf.asn.au/groundschool/umodule12.html

When you level out (from a nose-high attitude) close to the ground you reduce the lift produced by the wings so that they can no longer support the weight of the aircraft.

The most important thing though, is that once you've got the aircraft on the ground you want to ensure that it has insufficient speed to take-off again i.e. the wings have insufficient lift and the aircraft is effectively stalled.  So yes, just as you can drop an aircraft onto the runway, albeit from a very low height, you can put a tricycle aircraft down while its wings are still generating enough lift to fly as long as you can lose that lift by putting the nose down (which is one of the greatest benefits of tricycle landing gear over tail draggers; if you bring a tail landing gear aircraft down on its mains, and then drop the tail, you increase the AoA over the wings, which increases lift and which may cause the aircraft to take-off again.  This will only be temporary though, as increasing the AoA by bringing the tail down also increases the drag, which means that once the aircraft is back in the air it rapidly loses speed and finally stalls, but probably from an unsafe height).

[graham.d]

Exactly right Lee. I remember a few times when, despite a tricycle undercarriage, the aircraft I flew, err, bounced a bit! When this happens you can get quite a change to the AoA in a rather uncontrolled manner. Fortunately it tended to eventually settle down but tends to worry onlookers a bit :-) I got better though. It takes a while to be able to judge how high you are off the ground, though not quite as bad as in the video you posted; he must have thought the ground was another 100 feet below him.

I gave up flying long ago although my wife has bought me (for my last birthday) an hour flying a Phantom - regrettably only in a simulator though. I have yet to take it up. I am told by an ex-airforce pilot that I probably won't get far without a lot of help. Still, it should be fun. It was either that of a 737, so no choice really.

[LeeE]

I've no idea what happened with that MD-80 and can only assume some sort of failure prevented prevented the flare.  It did look a bit like a carrier landing though.

Re your landing 'moment'; well, they say that any landing you can walk away from is good, but be thankful that you weren't flying a B-52.  Controlling lift, via speed, in these motorised swept-wing gliders at touch-down is crucial because of their quadracycle landing gear.  If you try to touch down on the front gear first, and then bring the tail down, the increase of AoA across the wings, which are set at +six degrees incidence at the roots, generates a lot more lift with the result that the nose of the aicraft balloons up, perhaps to 50-60 degrees, and with the tail dragging along the ground.  Pretty quickly though, the extra induced drag stops and stalls it, at which point it comes crashing down again, usually resulting in a complete write-off, even assuming no fire.

[techmind]

When I used to come home from Hamburg in the eighties the computer used to bring the plane to about 10(metres?) above the long LHR runway then what seemed to be an interminable time later the pilot would wake up "oh yes we are supposed to be stopping here" and switch it off and we would come down with a bump.

When talking to some commercial airline pilots in the past, I was told that the plane's manufacturers recommend pilots drop the plane down fairly hard on the runway - although no doubt plenty of experienced pilots pride themselves on smooth landings. From what I understand, the hard landing is safer particularly if there's crosswinds or other such complications.

[techmind]

This sort of question comes up increasingly with technology taking over traditional expertise.

You could ask the similar: "Would you trust a machine over a surgeon to perform major surgery on you?"

"Would you want to travel in a fully computer-controlled car on the public highway?"


The debate also rages on in various aspects of medical diagnostics where machines with sophisticated software perform blood tests etc in the GPs surgery instead of sending samples away to a lab. The results come back far quicker, and in the main are probably more reliable - but over time you end up with fewer and fewer people who understand the test and its limitations... and an organisation which abruptly becomes paralysed when the wonder-machine breaks down.


I suspect that in many cases where computer-control is deployed, the machine is at least as good as the average professional it replaces (in terms of narrow well-defined conditions like straight-and-level flight, it probably outperforms all humans). Of course that means it's worse than the 'best' professionals. But then the 'best' professionals probably started out as average professionals, who then learned from experience as their career progressed. Computers don't normally learn from experience.
If we fully replace professionals with computers, although we may make short-term gains, there is a grave danger we will cease learning and advancing the skill.

[Geezer]

Latest on the flight that flew past its destination

http://news.yahoo.com/s/ap/us_flight_overflown