Airliner chat

**clean32** · 11th August 2009, 07:22 PM

Originally Posted by JDNSW

The problem of errors in software is, I suspect a sleeping monster in a lot of areas. I ran across one of these in the 1990s, in a piece of navigational software that had been used by one of our contractors for twenty years (with the code lifted from one set of software to new systems without rewriting). Increasing precision of work showed a worrying systematic degradation of data that eventually was traced to a navigational problem. This in turn was eventually found to be a software error that only operated in the southern hemisphere in changing from grid north to true north. The error was that the programmer explicitly changed the sign of a cosine function for negative angles, forgetting that the cosine is negative for negative angles.

More recently, my brother, who is a US resident, advised me of a New York court case where a drink driving case has successfully managed to get the code for a breathalyser to be opened for the court, and has found that there was a simple error in the method of determining an average of readings - instead of summing figures and dividing by the number of figures, the code added each additional figure to the previous sum and divided by two. Which does not give the same answer!

A well documented case was an unmanned space mission, one of the Mars landers, I think it was, where a mission failure was tracked down to an incorrectly handled change of units from imperial to metric.

What I am talking about is errors which do not affect whether the software runs reliably, but gives incorrect results. In most software, testing ensures that no serious errors exist for normal ranges of input, but most software today is so complex that it is impossible to test all possible routes through the software, and there is a real risk that results can be either slightly wrong a lot of the time, or wildly wrong on rare occasions, or both. And they can continue a long time like that.

In very critical systems, such as Airbus fly by wire software, they use triplicated systems, with different hardware, different software teams (not allowed to talk to each other) and different operating systems.

Any reasonable view of the software that is pervasive today in everything we do must come to the conclusion that there are a lot more of this sort of problem lurking to bite us.

John

intel P60 P 75 and P90

**flagg** · 11th August 2009, 07:39 PM

Originally Posted by JDNSW

Not up on gliders - only ever flown them a couple of times, many years ago. Just to give it a try. Most of my flying has been on a Cessna 180 and Beech 36, but also on a variety of others - learnt to fly in Victas, first plane I owned was an Auster, then the 180 when I moved to PNG, then various rental aircraft. Haven't flown for ten years.

John

My first flight was in a PA28.. then a C152.. then a Robin.. and most recently an Extra (only an hour though

). There isn't much I wouldn't do to go flying in PNG.. One day I will. One day

**JDNSW** · 12th August 2009, 03:29 PM

Originally Posted by flagg

My first flight was in a PA28.. then a C152.. then a Robin.. and most recently an Extra (only an hour though

). There isn't much I wouldn't do to go flying in PNG.. One day I will. One day

All my flying in PNG as PIC was in the Cessna180, except for a brief flight in a 185, but I did a lot of time there as a passenger in light aircraft, mainly BN2a, C402, C206, C185, Pilatus Porter, Bell 206b & 47, Hiller UH1100, Beech B55, Do26.

Flying in PNG is (or was when I was there, and I doubt much has changed except we now have GPS - but the maps are probably still not too good) very different from Australia. For a start, you really have to know about Density Altitude and its effect on performance - I have flown out of an airstrip with a density altitude of 12,000ft, in a normally aspirated plane (C185). Then there are the sloping airstrips. And the changeable weather. When I was there, there were still large areas on the WAC charts labelled "unexplored", where there was almost perpetual cloud cover and hence no airphotos and no maps.

Quite a few airstrips are one way, which means that you need to be up on downwind landings, and some have surroundings that ensure that by the time you cross the threshold, you are committed to a landing, and a few, you are committed to a landing before you come in sight of the airstrip.

One airstrip, Nuku, comes to mind where on final you approach to cross the ridgeline at an altitude of about thirty feet, with landing flaps already selected, and as soon as you cross the ridgeline, close throttles and maximum allowed sideslip to just short of the threshold on the opposite slope of the valley, then straighten up, and touchdown - once stable on the ground, keep it moving, otherwise the station tractor will have to tow you to the top of the hill! On takeoff, turn left almost as soon as airborne and fly down the valley until you have enough speed to climb.

John

**JDNSW** · 12th August 2009, 03:31 PM

The wreckage from the PNG flight has been found. No survivors. It appears to have flown into a near vertical cliff while executing a missed approach in bad weather.

John

**Captain_Rightfoot** · 12th August 2009, 03:57 PM

Originally Posted by JDNSW

The wreckage from the PNG flight has been found. No survivors. It appears to have flown into a near vertical cliff while executing a missed approach in bad weather.

John

I heard someone from up there talking on the radio. Apparently their destination was clouded in so they moved on looking for somewhere to land. He said they would have been light on fuel too.

**JDNSW** · 12th August 2009, 04:26 PM

Originally Posted by Captain_Rightfoot

I heard someone from up there talking on the radio. Apparently their destination was clouded in so they moved on looking for somewhere to land. He said they would have been light on fuel too.

I can't see why they would have been short on fuel - thirteen is a reasonable load for a twin Otter, but it is only a short flight by Australian standards.

I wonder if the use of GPS has resulted in pilots flying into cloud they would not have risked previously? Problem is, the GPS might be right about where the plane is, but the mapping is mostly nowhere near good enough for that sort of stunt. And worse, you don't know which bits of the mapping are good. Far to many pilots lost in PNG flying in cumulo-granitus!

John

**rmp** · 12th August 2009, 08:02 PM

Originally Posted by JDNSW

Don't know that one, but I remember years ago doing a circuit and successful landing in my Auster with no airspeed indication - and then cleaned the mudwasp out of the pitot tube. It had an automatic cover, which I had assumed protected it - but it was parked head to wind and this lifted the cover!

John

Did that in a glider after not removing the pitot cover before takeoff ;-)

Worked it out during the takeoff roll, who needs ASI anyway!

**JDNSW** · 12th August 2009, 08:38 PM

Originally Posted by rmp

Did that in a glider after not removing the pitot cover before takeoff ;-)

Worked it out during the takeoff roll, who needs ASI anyway!

You can do without most instruments at a pinch - like the time all the fluid leaked out of my compass on a flight from Roma to Archerfield. I knew the route well enough to fly it by simple pilotage. Also worth noting that the only stall warning on the Auster was the controls got sloppy, wind noise got quiet, and just before the break it started to shudder a bit. And no aileron control in the stall, had to pick the wing up with rudder or you would find yourself in a spin - "don't do this near the ground!"

John

**VladTepes** · 12th August 2009, 08:53 PM

That U2 video was spectacular....

Pontiac GTO my bum, that's a Monaro ! Damn yankees.

**Captain_Rightfoot** · 13th August 2009, 07:19 AM

Originally Posted by JDNSW

I can't see why they would have been short on fuel - thirteen is a reasonable load for a twin Otter, but it is only a short flight by Australian standards.

John

The guy said they had been trying to land for some time and had made a few attempts and had given up. Maybe diversions and go-arounds had used their margin.

Thread: Airliner chat

Thread Tools

Bookmarks

Bookmarks

Posting Permissions