Aaron wrote:
Back in the day when CATIA was used to design Mirage fighter jets, I remember reading that what made the software so powerful was that it could define the location of an infinite number unlike its competitors that could only approximate points. I thought 'defect' this due to polygons ultimately being a representation of a surface. CATIA sounds like it deals absolutely with a virtual surface.
Is that true? True still? Is that why it costs trillions of dollars to use?
Yes, that's how CATIA used to work back in the day, but that was not necessarily nor primarily the reason for the its power or the cost. And it has since undergone (at least one) a major or massive overhaul from that particular engine: - I believe when they upgraded from version 4 to version 5.
Incidentally, an interesting anecdote about this software engine's overhaul has to do with the direct effect it had on one of Dassault's (the developers of CATIA) biggest clients. When the Airbus was being designed, since different parts of the plane are contracted to different manufacturing plants across Europe (partly to do with the fact that the Airbus is a multinational or at least multi-country effort), this was occurring just as Dassault were upgrading the software, and certain manufacturing parts upgraded their CATIA versions to the new engine in version 5 while others kept working in version 4 and the result was one of the most expensive and infamous blunders in the construction industry when the separate manufactured parts couldn't fit together due to the different ways in which CATIA v4 and CATIA v5 calculated or geometry defining the holes that the rivets go through in connecting the various parts. It was fractional error and really very very minor discrepancy, but nonetheless one that added up to billions of dollars (or Euros) in re-designs, re-simulations, etc, and an embarrassing set of delays and overall cost overruns for the Airbus corporation. Needless to say, they have since instituted a common coordinated software upgrade policy for all their various plants across Europe. Boeing have been looking to avoid suffering the same design pitfalls with their own 787 Dreamliner design, which is also being manufactured all over the world as they have sub-contracted the various parts of the plane to different manufacturing plants - and indeed completely independent firms and companies, outside their N. American plants.
But yes, Dassault Systemes have since upgraded the software since version 4 and it actually calculates geometry differently (more accurately? more efficiently?) now that before. I also heard that it underwent another overhaul for the recently released version 6 as well.
Regardless, CATIA's power and cost, aside from the 3D engine which is powerful as it is, really lies in the modular software structure that they have adopted which allows them to develop several separate modules for separate functions which then run fluidly atop the main base 3D Engine. This allows their customers to easily customize the software to their individual needs while also streamlining how it handles and manages information across the board. The modular approach is something that GS could greatly learn from in terms of how they dispense their limited R&D resources for developing various tools for ArchiCAD - to some extent I see that they seem to be adopting this approach as well when you consider the MEP module, and to a lesser extent the VBE explorer. I have always long hoped that they would adopt such an approach to developing other features that are often clamored for by users, but which may not necessarily be considered to be nor fall under the core functions of the design and modeling engine - such as a rendering module possibly developed by a ASGVIS/Chaosgroup, but licensed by GS users directly ala Vray in Rhino and Sketchup; or a Mass modeling/ conceptual design module or a Energy calc. module, a Take-off/Quantization module, etc etc etc.
All of which would keep the main engine light and streamlined while giving the customers the option of purchasing (and directly paying for) additional modules as they see fit for their individual needs and their firms.
Dassault systems is not that large a firm neither.
Aaron wrote:
Clarence: do you know anything about cost overruns and improperly manufactired assemblies at the Art Gallery of Ontario. Does anyone know what that was about? The inside poop, as it were?
I may be wrong, but my understanding of that situation was that it was more of a bureaucratic kaffafel on the bidding/construction end of the project (i.e. Ontario provincial govt. rules and issues vs. the private sub-contractors) more than it had anything to do with Gehry's design or project delivery. It's not exactly that complex of a design anyway (morphologically speaking) aside from the curvy curtain wall/facade and the spiral stair feature that he designed.
But then again, he is currently involved in litigation over MIT's Stata science research building that he designed (they're suing him over some leaks of something of that nature.) - although again therein lies the issue of the lack of control of the design once it goes into construction, and the contractors decide to cut shortcuts, use inferior materials or methods (as he contends happened in this case under the client's(MIT) direction, to cut down construction costs).
