CUNIM,
I thought EDDUS was relatively successful for an ATC project. The system was originally proposed as single computer to replace NAS and the Locus 16s and to act as a common database for military and civil users. Echoes of the huge Linesman/Mediator project from the 1960s. The usual specification problems arose and a fairly major reorganisation left NAS in place and the minimalist EDDUS serving the new MASOR (military area services). NAS and EDDUS never spoke to each other but EDDUS was paperless from its introduction. Getting Tandem machines to talk to DEC machines was a nightmare. Does this fit your recollection? And as for translucent windows – absolutely! I just haven’t bothered at this stage for the demonstrator.
Scott Voigt,
Having worked on both sides of simulators I agree that the ‘perfect’ quality of the representation can give misleading results. In an earlier post I noted that a crystal clear voice channel could enable a 33 per cent increase in capacity. The New Model Demonstrator uses two weather and performance models. One is used for trajectory prediction and the other to drive the simulated aircraft. Tests have been conducted with large differences between the parameters in each model. The results match what one would expect for a stormy day; the overall traffic solution changes quite frequently but the next RT advisory is always a good one. I have studied noise and stochastic processes for military systems but I don’t think anyone will want to know more about the maths involved!
I can’t actually see any human factors links – did you intent to include some?
Vector4fun,
I haven’t forgotten the unusual traffic that makes up most of the difficulties in ATC. I was actually in TC observing when the aerial photographs were being taken for the multimap
www.multimap.com database, mostly from 6,000 feet. The frenetic shouting across the room was quite something. Because all aircraft in controlled airspace must have a flightplan we can start with the assumption that the system is able to allocate airspace resources as necessary. Surveillance data and controller inputs will assist in making the allocation as efficient as possible. After that, the problem is simply one of testing possibilities and evaluating the results. And computers are now very, very good at doing that. The New Model Demonstrator completely recalculates a solution for the London TMA SWE sector with one new arrival per minute for the next 25 minutes in about one second. I can post again (or PM me) if you’d like a detailed explanation for your examples but for the moment could I just say that the calculations would be very similar to those performed at CFMU for slot allocation. For the approach sequencing part of the problem FAST here and in the USA is getting better all the time.
Scot and Vector,
OACC FDPS I and the new ATOP handle traffic automatically. The ocean has quite a lot going on, air-to-air refuelling, submarine hunting and air-sea rescue amongst other things. FDPS I operated both organised traffic and so-called random traffic. Even moving blocks of airspace could be reserved. It changed the tracks twice a day, handled supersonics as a matter of course and acquitted itself pretty well on the black day of 9/11. None of this is intended to take anything away from the hard working controllers. The New Model an honest attempt to offer something useful that builds on previous experience.
Flight International 10 May 2005:
ATC safety over Africa set to worsen as traffic increases
ALPA-SA President Captain Gawie van Rooyen
Anybody from Africa reading this thread and looking for a solution?