Thanks for the post!

So, my first (albeit, tongue-in-cheek) piece of advice is to not make large assemblies. Of course, this may not even matter if their whole way of modeling is completely jacked.

Nevertheless, it's a place to start and helps people visualize assembly hierarchy which helps with the engineering to production.

Thanks for the comment!

It's all in the detail.

I think the key here is the second step, simplify parts. HOWEVER, I would add that you do it with configurations, because there WILL be a time when you need those crazy details that everyone thinks you're nuts for putting into the models.

The biggest one I always get criticized for are threads, helical springs, wiring, decals, logos, and purchased parts. And each time I got scorned, laughed at, or yelled at for “wasting time” putting in those details, I had something happen, where having those details in the models paid off, big time. Or the flip side, not having those details cost the company big money.

One crazy example: Worked at a manufacturing plant, architectural goods. Many screws used. Some fully threaded, some partially, some were rivets. First complaint I got from the shop floor was that the workers were confused because they couldn't tell in the exploded views whether the fastener was fully or partially threaded. Ok, so they can read the parts list and find out, right? yeah, but that since people are more visual, it would greatly improve the manufacturing process by having them being able to tell simply from the diagram. So YES, sometimes YOU DO need schematic threads, people…

Another crazy example: worked at a place where the product used trilobular screws. Yes, I actually swept a cut on a trilobular path, although I didn't have it follow a helical path. Well, manufacturing was having an issue with the threads stripping out of the extrusions that they were used in. Since these were tapered at the ends, they needed to know how many untapered threads were actually engaging in the material, at least by design. Well, presto, I already had that level of detail available to me, and was able to quickly pull up the models, and see that by design, we didn't have enough thread engagement because of the tapered part, and the revision to the extrusion began. To add to that, I later ended up creating a Simplified configuration for that screw, so that at the higher top level assemblies, where the threads were too small to be noticed or needed, the assemblies could load much faster. At this same company, the issue always came up of whether or not an manufacturing, or even an end user, could install a screw/bolt using a cordless drill (or the tools on the shop floor). Well, guess what, I went to ye ol' 3D Content Central, found a drill, a bit, and slapped that mutha in a test assembly to check for that. Many times, it was apparent that you couldn't do it, and a redesign was able to be done before any tooling had been made. Screws, zip ties, tags, stickers, labels, packaging? Yep, all important things to consider.

An example that let to mass chaos and mayhem: Worked for a company, large industrial equipment. Projects ranging in the $3-5 million range. Part of the product included conveyors that were designed and outsourced elsewhere, but used in the plant. Boss thought it only necessarily to represent these conveyors as “simple” blocks that represented the basic shape and the footing locations of the conveyors. After all, our assemblies were 5000+ components, we don't need to be bogging down our assemblies with “pretty parts”. When I was sent on site to oversee the installation, the first person I met was a welder with a cutting torch asking me, “What do you want me to cut off?” meaning what part of the conveyor did he want me to cut off because it wasn't fitting… by like 3 feet!!! Nothing fit, in fact. Almost every single conveyor and support needed to be cut/modified/jury-rigged…. all costing thousands of dollars a day in delays, and an eventual lawsuit after I had left the company… Why? Because details like motor guards, safety shields, catwalks, etc, weren't taken into account, and weren't in the models.

As designers, are we not supposed to design for form, fit, and function… and manufacturing, and packaging, and shipping, and consumer assembly, and usage? How can we do any of this if we don't have enough information to really make an informed, educated, and “engineered” decision?

And doesn't this CAD data get used for more than just design? One thing mentioned on this and other blogs is new tools like 3DVia for technical documentation, and PhotoWorks/HyperShot for rendering. Anyone else who's done technical documentation, or photo-rendering/animation, please pipe up! They'd all be cursing those of you who pass on 3D data to them, that doesn't have enough detail for them to complete THEIR task efficiently. Assembly instructions, replacement parts lists, etc. all require highly detailed illustrations in order to be effective and provide manufacturing and the end consumer with a good experience. Don't even get me started on photorendering/animation. I can't tell you the countless hours of model prep that needs to go into some models just to get it ready for the materials/rendering process. You can spend that time putting in detail into the model, or you can spend just about the same amount of time trying to apply funky textures, bumpmaps, and decals in the rendering program, hoping that it will look right in the end.

So yes… DETAILS MATTER people! Sorry, I don't mean to fly off the handle or offend anyone, but this discussion constantly comes up wherever I go, and companies are constantly scratching their heads wondering why they can't seem to reduce high costs associated with errors/inefficiencies in their manufacturing and design process. It's because they keep ignoring the need to have enough details in their design data, be it 2D or 3D, to truly be able to analyze a design and check for these types of issues. The promise of 3D was that you could almost completely flesh out the design, work out kinks, and reduce costs, before even making a prototype. Well, if you don't actually put the effort into making that first “digital prototype” as kick-butt as possible, you'll never realize those benefits.

Is is realistic for every single company to have models/assemblies detailed to the hilt? No. Will the tips Josh has provided help you deal with the situation and alleviate some of the pain? ABSOLUTELY. But is there great value in going the extra mile to get your models as detailed as possible? You betta-believe-it.

As far as large assemblies, I don't really know how you get away from it, especially in industries like automobiles, ships, industrial plants, airplanes.

One other thing. The other term I keep hearing from managers and what not are “pretty pictures” referring to renderings and technical illustrations. They say it with such disdain and ridicule sometimes. It's so funny to see the change in their demeanor when they realize, that “pretty picture” can actually help the company MAKE MONEY. Oh, you realize that these pretty pictures can help seal the deal in selling a project or a product in a struggling economy? You finally realize that these pretty pictures can be used for all SORTS of marketing/advertising material, for tradeshows and conventions, and for meetings with upper management and even clients? Oh, you like pretty pictures now, do ya? Well, better late than never I guess…

Wow, I think I need therapy…

There's a lot of things that play in to how assemblies function right, and basic principles that can make creating them a lot easier. The first point should read, “define your hierarchy.” Know up front how you're gonna build the assembly, what is feeding into it to aid manufacturer and make design changes easier for engineering and manufacturing. Use the heck out of Top level assemblies, just do it smart… use simplified configuration, etc.

Performance in Large Assemblies that you refer to is what I'm trying mostly to address. I actually avoid using lightweight mode and run assemblies with 100-1000 unique parts. It gets doggy on the high end, but my computer is substandard (2GB XP 32bit DualCore with FX1500) – sidenote: what's annoying is that set-up could push a pretty good size assembly a few years back. ahy.

Anyway, I see where you're coming from and it's what I'm trying to help out with. I'm hoping assemblies get faster without having to employ any speedpaks of modes to help. That would be the most ideal. Then ya wouldn't have to put up with posts like this thanks for the comment Chris.

And on points 2 and 3, you should always strive for simple part design, and even if you are only making a single part you should fully define the sketch. And of course following these rules will improve the performance of the software.

But the performance of SW with even a small number of parts is pretty sad. I've had assemblies where all the parts are fairly simple (prismatic parts, simplified hardware, no complex surfaces) but it still seems like more than 50 – 100 part assemblies is enough to bring the software to it's knees. Using lightweight parts helps. I haven't had a chance to use any of the new large assembly functions in 2009 yet but it sounds similar to what ProE offers with simplified reps. We'll see how well it works.

My advice is this: Don't use freakin' SolidWorks for large assemblies.

If I'm working on a product with less than 50 or so unique parts, say a toy or a consumer product, I'll stick it out with SW because I like the interface. But if the product has more parts than that, or requires lots of complex surfaces, or simulation or FEA (we have Mechanica), then I'll opt for ProE.

I am constantly amazed at work at the tiniest little details that people use to make pretty parts and then wonder and gripe about their computer or SolidWorks slowing down! I've sent emails, had discussions and meetings, and it still happens…