Cadkey or Catia? Boeing’s Billion-Dollar 3D CAD Mistake (2017)
tecnetinc.com
tecnetinc.com
Yes, Catia was a PITA to use and support, but it was productive in the right hands. PTC Pro/E was also a PITA to use, but very productive for experts. His outright dismissal of Solidworks is hard to reconcile with my personal experience using it to design components over the last 20+ years. These workflows might be different from his preference, but that doesn't make them invalid.
The referenced article about the failed promise of parametric modelling does have a good point though, although I don't think it's fully expressed: modification of an existing design in parametric CAD feels like it's a lot more difficult (and likely to blow up) than a software project of similar complexity. Perhaps this is inherent complexity? I'm not sure, but I do find that it's easy to get caught in a trap of endless circular revisions on related parts for what feels like a simple design change sometimes.
I don't think so. I use Solid Edge, which is quite good at this. You can push geometry out of the history based parametric scheme into freeform modeling. There's some pretty fascinating YouTube videos from Siemens showing importing of 3d cad models from other programs, and then repairing parts of them. Granted, I've never done anything on a truly massive project like an assembly with ten thousand rivets or something.
The author sells two different CAD products that supposedly do the same (zw3d and ironcad), but I haven't used either of them
In my own personal experience, using Solid Edge outside of the consumed-sketch history mode and in free form mode (which still has both sketches and a history stack) has been much much more pleasant. I started using it long after they introduced that "synchronous technology" as they call it, and didn't even realize that history/consumed-sketch mode was an option for quite a while.
The design intent inferring is pretty slick, some other videos do a good job showing it off (and also how to turn it off if it doesn't get it right). https://www.youtube.com/watch?v=5nPQOalmlrc for one of those.
But I have also had the joy of trying to program IronCAD. The parametric solver is not as good as the other programs (who live and die by it). I suspect, though cannot prove, that the little automation system we used to create a base design would have been a lot more stable running on a different system. But it was too much work to completely re-implement (and convert all the old designs etc.). So I had the joy of fixing changes between versions by e.g. swapping the direction of a constraint, or hard-coding an initial offset so that you approach a solution from a particular direction.
Also, adding 2D annotations to create a production drawing were painfully slow and finicky in IronCAD.
I think I agree with the author that history-based editing caused the whole industry to go down a less-efficient path, but I also think it's too late to change the past now.
It's funny I've been using Solidworks for years and would love to find an alternative that hits the right notes. Closest I've come is OnShape (which I believe was started by Solidworks' x-founder and some top engineers he took with him). Tried a few alternatives like SolidEdge, SpaceClaim etc. but they either didn't offer as intuitive usability (particularly the sketching interface) or weren't stable.
Myself, I clicked with Inventor when I had a free educational license. I later got a solidworks license through EAA (a benefit that no longer exists, thanks Dassault!) and found it frustrating and unstable in so many ways. It felt like so many submodules just jammed together without any concern about consistency of interface or experience. Though it did have more options for putting assemblies together than Inventor did. Comparing and contrasting the two was really enlightening.
Eventually my educational license for Inventor ran out and I got upon Solid Edge. Like you (I presume) I found the sketching to be different from what I was used to. Then I learned that it actually had a completely separate mode that turned on the classic history CAD and suddenly it looked like Inventor again.
These programs all have so much functionality that can be so hidden behind years of changes and updates and best practices. Somehow though, I wouldn't call them UI or discoverability nightmares (though I think solidworks is both, haha). Both Inventor and Solid Edge really did a good job holding my hand as I gradually learn to use more and more of their power. Some of their help systems are really well designed. I'm impressed by how they both can have (mostly) consistent interfaces between their various features so learning a new part feels more natural. The way Solid Edge integrates its help system is really neat too. Maybe the rest of the computer industry can learn a thing or two from how these CAD programs provide gradual discoverability.
I'm not too torn up about it though. I really didn't like Solidworks (as you probably saw in my other posts) and losing the $50/year license for it led me to Siemens Solid Edge, which I'm substantially more productive in. For me the only downside to the free edition of Solid Edge is that it doesn't come with the fastener / standard part library that you get with the commercial license. But hey, guess what, 3DEXPERIENCE Solidworks has the exact same problem! No Solidworks toolbox (when I used it, their version of the standard parts library). Hilarious that that's their offer for the EAA though, you'd think something for designing aircraft parts would at least come with standardized rivets you could use in your assembly!
there are ways to mitigate the side effects (e.g. using reference planes instead of part topology as feature references) but it's a chore sometimes.
software equivalent of topology changes would be perhaps changing from using a certain set of APIs to another for reasons. If only equivalent functions are used it's not an issue. But if there's missing functions that you have no choice but to recompile the library and add logic to manipulate the information, that's a lot of work!
it's the same as if that hole that you parametrically positioned hits a spar and you have to redesign the surrounding structures.
I usually just want to focus on designing a part, and not having to consider if I selected this edge when building the mating part and what will happen if I edit the wrong sketch 20 hours from now.
Edit: I posted because I was curious about your opinion, but now I realize it just looks like I stopped by to insult TurboCAD for no reason.
In 2004, the Dassault Falcon 7X became the first aircraft in industry history to be entirely developed in a virtual environment, from design to manufacturing to maintenance. There was no prototype: the first assembled Falcon 7X was a production model - that was revolutionary and borne by Catia. But maybe the draftsmen were still pissed off...
Sounds like a fiber optics referential database and workflow migration I was part of at a large telco, with the design community growing from 12 to 3000 users: of course the early experts were unhappy with the directions the systems went into, but they were no longer the center of gravity - the whole system was.
Disclaimer: my uncle was part of the Dassault team who sold Catia to Boeing...
"Products as complex as Catia are not separable from their organizational context and only express their values as part of business process transformation."
Just, wow. This should be framed somewhere and enshrined as an example of the utter stupidity and self referential navel gazing that goes into some of these utter nonsense software orgs...
This is completely backwards - software provides features, not business process. When some process gets enshrined in a bit of software, this doesn't make it any good. You also not making business any better - yeah, sure, biz processes change, writing code to enshrine those processes isn't helping anyone.
Unless you create new features, you aren't doing anything new that might mean the business process might change.
Not to mention the viewpoint that software "expresses values" like it is some hipster...
FreeCAD isn't even terrible for any good reasons. If I had to dumb it down it would be "politics" because FreeCAD insists on certain puritanical design choices that have no practical reason to exist.
In FreeCAD, "Body" is a single solid, if you have an operation that fails to create a single solid, you just get an uninspiring error message, you don't even get to see a locked preview that shows you how far off you are from a valid single solid! Ever wondered why the 3D view of FreeCAD and Catia look so similar? Because FreeCAD is using opencascade. Catia has no single solid restriction so this artificial restriction is twice as enraging!
In FreeCAD datum planes must live up to their mathematical counterparts! Officially planes have no orientation i.e. a default direction. Yet when you try to attach a sketch, it will align the sketch based on a random position and you can even use the transform feature to rotate the plane! The first thing I have to do when I make a new sketch is draw an arrow that points up and then rotate the sketch so it actually points up! The demands to let planes show their hidden internal orientation have been shot down because planes have no orientation! Fantastic!
It's very similar to how relational algebra specifies sets of tuples but actual RDBMS allow you to have duplicates. Break the damn eggs, omelette tastes good.
The VP who made the decision decided to go with a DEC VAX solution at about $250,000 per machine. I couldn't believe it. It was the start of my exit of fortune 500 work.
I found out later the DEC rep had taken him on multiple trips to Vegas and the Bahamas. Just they way it was done back then.
That company is no longer anywhere near the fortune 500.
A drop of process in your process? Or perhaps you prefer process. This is unreadable, couldn't make it past the intro :/ (there's also "solution" and "politics").
Corporate cronyism made a really bad piece of software and fucked Boeing's business (translation - your plane tickets in part funded some grossly overpaid people at Dassault)
TL;DR
Boeing had two systems, one was 7g, other was 70g. Feature parity the cheaper solution did more.
Some IT group hated the cheaper solution because they had trouble administering it, so they opted for a multi mill price tag instead of a mil price tag.
Sounds like on top of that, there was a bunch of artificial process which had to do with old artifacts of cad design, which only remained because the IT group built processes around managing those artifacts, not because they made sense in any fashion. The expensive software compounded the issue instead of solving it because of kickbacks and improper interests at Dassault.
Meanwhile the cheap solution was used by pretty much every third party supplying Boeing with parts, to great success. It almost sounds like the internal software at Boeing was used more as a review system and didn't actuall provide any use, they just outsourced the real solutioning to their providers who were not held hostage by an incompetent and corrupt IT dept.
At some point Dassault acquires Solidworks, their software is overengineered and sucks.
Over time things gradually get better - the expensive solution never gains new features, just new headaches.
TBH I'm not sure where the billions lost come from - I presume he is saying doing everything in house would have beem cheaper? But Boeing was just busy hemmoraghing money paying for upgrades to a system that hadn't been doing its job for decades?
I'm not sure I have ever encountered someone who would have not picked some other CAD package over Catia flavors if given the choice.
That said, I'm not a mechanical cad professional and only have dabbled as a hobbyist in Fusion 360, Solidworks and FreeCAD. So my opinion about multikilobuck CAD packages is more hearsay then personal experience.
At the OEM I work at, we have one guy that easily spends 80% of his time dealing with legacy V4 models that need to be brought up to CATIA V5. Whenever we get obscure spare parts requests, he has to go exorcise the models out of the V4 mainframe that we still run, massaging them until we can get geometry that's good enough to use for manufacturing purposes.
I will never forgive Dassault for the lack of compatibility between V4 and V5.
That's especially a pain in Aerospace, when 40 year product life cycles meet considwrably shorter software lifecycles. Tgat being said, I'm told there arw still people out yhere maintaibing COBOL, so.
I do see parallels to ERP software (something I'm much familiar with). Release changes are a major PITA, switching vendors close to impossible, and people complain evenly about each and every single option and vendor. Until you take the system they know away, that is. As someone stated above, once you understood the underlyong logoc, things become easy. It does require some major brain accrobatics sometimes, so.
Huge mess. We went from 4 (still have a couple purple Silicon Graphics machines we’re about to toss) and skipped 5 for Solidworks.
SW isn’t by any possible stretch perfect, it’s not even very good, but having a tool that gets the job done and isn’t some rare mythical unicorn has its perks.
https://www.reuters.com/article/us-airbus-a380-history-analy...