IMACS in Worthing did much of the software work at that time.
Weight shift per se apparently was not the cause of the accident. Moving objects supposedly “crippled key hydraulic systems”. See here:
http://www.avweb.com/avwebflash/news/NTSB-Loose-Cargo-Caused-Fatal-747-Crash-224426-1.html
Rediffusion?
Yes; that was one of the intermediate names. Redifon came before or after that.
CFD, "comptational fluid dynamics" is the magic word.
Or Color For Dollars :-) Its accuracy like anything else depends on the skill of the operator/analyst.
I once brushed up against work on cargo barriers for big jets, the innovation of the moment was to use chem milled panels to replace the existing solution, but regardless I have since thought barriers were installed to prevent cargo moving too much even it were to break loose. Apparently not always the case. A sad thing to watch.
I always thought Airbus built their planes like that because they're trying to get rid of the pilots... I "flew" the A330/340 sim once and it was pretty impressive but slightly disconcerting to know that my "job" was merely to suggest to the aircraft computer where I might want to go.
"Computer says No" takes on a whole new meaning... :-)
...flight sim maker - Redifon,
Rediffusion?
They got the same for Concorde for it's sim. When the plane flew for the first time - years later - the pilot reported that it flew just like the sim.
Sometimes it works out, other times it dosen't. Quite a few prototypes were lost because they did not behave like the engineering simulator (e.g. the Canadair Challenger or the German VFW 614 mini-airliner, both crashed during slow flight testing).
I suppose stalling needs a different set of equations but surely they can be determined experimentally in a wind tunnel, or nowadays by CAD?
CFD, "comptational fluid dynamics" is the magic word. But slow flight and especially departure from controlled flight are still very difficult (or almost impossible) to model and predict. Wind tunnel testing helps, but there are still a lot of surprises and lessons to be learned. Because of that, Airbus builds it's aircraft so, that (in theory at least) they protect themselves from entering unexplored regions of the flying envolope!
Many years ago I visited a flight sim maker - Redifon, later called various other names, eventually Thales I think - where the man described how they are supplied with a set of partial differential equations by the aircraft maker, and they solve them approximately 20x per second.
They got the same for Concorde for it's sim. When the plane flew for the first time - years later - the pilot reported that it flew just like the sim.
I suppose stalling needs a different set of equations but surely they can be determined experimentally in a wind tunnel, or nowadays by CAD?
The AF447 recovery would have taken approx. 15000ft.
What? Why would 90 degrees not be in the envelope?
It's not the 90 degrees, but the 80kt (or so) forward speed.
What? Why would 90 degrees not be in the envelope? This was the FFS for the C525 which is Part 23, not 25. I would be surprised if the modeling in the sim stopped at 60 degrees.
I've done that maneuver in the sim, rolling 90 degrees and letting the nose drop. It works ...
... maybe it works, because it is almost certainly outside the specs of the simulator and therefore pure guesswork of the programmers. A full stall is not part of the flight test program of transport category aircraft, therefore no flight data exist to form the basis of the simulation. The same with CofG conditions far outside the limits. There are very few (known) cases of airliners successfully recovering from a stall and all of them lost fife-figure numbers of feet during the recovery. (One of the most famous being TWA 841, a B727 that lost 34.000ft recovering from a stall in cruise flight and even went supersonic during the dive.)
