Challenging the conventional wisdom for assemblies

So much of what SolidWorks users assume about assembly modeling appears to come from the official SolidWorks training. The official SW training has been around for a good long time, and it receives high marks for what it does. But face it. We’ve all learned as much at the school of hard knocks. Maybe its time for the traditional basic training mantra to become a little more sophisticated. We have already seen that the conventional SolidWorks wisdom on part building is that you should just build relationships to what ever geometry is conveniently available, and we know that this works for fast modeling, but it’s not a great idea for production work.

Anyway, the notion that I’m interested in challenging is that you should mate each part to the one next to it. I know there is this mental goal of trying to make the computer model the real world, but in the real world, that doesn’t really work. In the real world, you can’t change the width of a part, and if you could fillets wouldn’t disappear, features wouldn’t fling out into space, and assemblies wouldn’t turn inside out.

A recent comment on one of those wish list discussions I read very astutely pointed out that the real problem with modern CAD products is not that they don’t model reality very well, it’s that they don’t model theory very well (very heavily paraphrased). By that I mean that CAD software across the board doesn’t react to change very well. If I think about how I spend most of my time in SolidWorks, it is in cleaning up the consequences of changes. It’s not in figuring out design, it’s not making geometry for the first time. The biggest obstacle is the aftermath of change, which happens all the time. Do I need another way to put meta data into my geometry data? No. I need better ways to make changes without creating more work for myself. Do I need to run SolidWorks from my cell phone? No. Cloud answers problems that don’t exist. I need software that reacts better to change – whether its flipping mates, flipping trims, flipping sketch planes for SolidWorks or limitations placed on changes by fillets and technical brep issues for ST3.

So what are your top problems or recommendations with assemblies? We’ve talked about master model concepts, “Horizontal Modeling” and layout/control sketches before. These are alternative methods, but should they be the default? Should the convenience things like sketching on model faces and referencing model edges become things of the past? Are you willing to put in a little more work in return for a model that reacts more predictably through changes? Or this this sort of problem supposed to be solved at the software level? Would you expect this “best practice” stuff to be hard-coded into the software? Do you mate face to face? plane to plane? Or do you use layout sketches? Is mate-based dynamic assembly motion a valuable part of your process? How much time do you spend cleaning up after changes? Why do you think SolidWorks thinks that “the cloud” is a more compelling center piece for a fresh start than something related to solving your most obvious CAD-related pain?