Wednesday, January 28, 2009

The Push & Pull of Direct Modeling

The CAD industry can be a tight, competitive industry. Yet, we are still seeing some serious innovative technology coming out of the various players. Every year there seems to be the "new thing". The buzz over the last 12-18 months has been history-based vs. direct modeling.

A good majority of mechanical engineers are familiar with history based solid modeling MCAD systems. Everyone knows the major players - SolidWorks, Inventor, Pro/Engineer, NX, Solid Edge, CATIA etc.. There are numerous advantages to a history based system. I won't bother pitching it as I think most will agree with some of the advantages.

But let's talk about some of the disadvantages. Probably the biggest disadvantage is when it comes to collaboration. This is a very broad category that involves collaborating with other CAD designers, the CAM guys, the CAE folks as well as other CAD systems. Many of the CAD systems can fail quickly if you attempt to break a parent/child relationship. In essence, if you are trying to edit a feature that depends on another, which may depend on yet another. If you are not exactly clear how this model was built or how it all ties together, it can be very frustrating to try and edit a simple feature. It is somewhat the nature of these systems, but a good majority of the issues come back to POOR cad modeling practices. One can easily argue that the system allows the poor practices, therefore, its the systems fault. I'm not sure we will all ever agree on this one.

So, what is the solution? Many, me included, think that Direct Modeling can be the answer. Almost every vendor out there has some form of direct modeling initiative. Siemens has probably been the most aggressive in marketing their Synchronous Technology. So, imagine that you can have a powerful CAD tool like NX or Solid Edge and you can simply then just grab, pull, and drag the geometry around. From a simulation guy's perspective, this is an amazing capability. So, let's say I have a valve assembly and I run a quick CFD simulation to measure pressure drop and attempt to identify the inefficient flow locations. I can then simply grab and shape the internal flow pattern and "morph-it" as if it were clay to "smooth" out the flow patterns? This quite bluntly will change the way engineers design if implemented properly.

Recently, Autodesk announced their version - Inventor Fusion. Their technology will be embedded inside Inventor (to be released soon). Autodesk may be at an advantage as they are still (others as well) converting companies from 2D to 3D. So, feasibly, you will have a sector of engineers that are not familiar with history based systems and will see the best of both worlds. Not only will they have parametric capabilities, but they will also have the power of interacting with their models in a unique way. But most importantly, this will seem natural to them as they won't know any different. It will be very interesting to see if this methodology takes off.

Gotta mention our friends at SpaceClaim. They may not have invented the push/pull concept, but they certainly are taking it and running with it. They are one of the easiest/coolest modelers out there, especially for those of us in the simulation world. Not only can I model the way I expect and want, but I can also bring in geometry from essentially anywhere and edit, delete, push/pull features at will. I have praised SpaceClaim many times here. But its only fair to say that these direct modelers have some really powerful grab and pull capabilities. BUT -- they must allow some for of parametrization, some sort of dimensioning, some way to tie my model together.

WHOA, sounds like I am suggesting the dreaded parametric, history-based modeling? Or am I suggesting some sort of hybrid modeler. Call it what you want, fight it out in your strategy meetings. But here is what I (the market) wants for simulation, at least..

I want a very powerful (capable of complex shapes), SUPER modern and fun to use, ability to edit other CAD systems, ability to create constraints/dimensions/parameters (ideally with the option to drive from a table/excel) that can create multiple "configurations" on the fly and have unique simulation tools (gap/interference detection/repair, small feature removal, part simplification (dumb down part by part).

Is that too much to ask? The CAD system/modeling system that can dial this in will have a leg up on everyone else.


Jeff Waters said...

Great write-up Coop! Business is changing right now (and I'm not talking about the current economic conditions- which is all you seem to hear about on the news). It's changing in ways that will see "companies" look more like a collection of suppliers and temporarily hired guns working on bits of products "virtually" from all corners of the world. This idea that you can open and edit models in any CAD format will be required to support that new reality.

As for today, there are a ton of areas where Direct Modeling offers value to folks who could be using 3D- but who can't, won't, or shouldn't climb the learning curve of a traditional CAD tool. These are untapped areas of productivity hiding in the existing workforce.

Course, I work for SpaceClaim and am a bit biased. I do want to point out one wrong assumption a lot of people are making as they learn about Direct Modeling: Many people think Direct Modeling means there are no parameters, because they are seeing it being positioned in opposition to traditional "Parametric" CAD tools. This is just an error in symantics causing some confusion. You absolutely can setup parametric relationships within Direct Modeling tools like SpaceClaim. You just get the freedom to set them up on-the-fly. And, you only need to define the parameters that you really care about, where traditional CAD requires you to manhandle every feature and parameter. And yes, it is possible to control these parameters from outside the DM tool... at least, I know that's true for SpaceClaim.

Solid DNA said...

Jeff you talk about symantics, I agree with you.

I place a lexis on my blog in order to make some distinction.

Solid DNA

Big Joe said...

Nice write-up Derrek. I've been using NX for almost 3 years now and I rarely use any of the usual parametric modeling tools. This is after spending the previous 7 years prosthletizing design-intent-driven parametric modeling.

I wrote a quick description of parametric vs. direct-modeling workflow on Product Design Forums:

Anyone Using "direct Modelers" (e.g. Spaceclaim, Keycreator, Ironcad, Cocreate, Aso)?

One of the biggest benefits of direct modeling over parametric modeling is that it brings the world one step closer to a standard 3D file format. When making changes to dumb-solids the direct modeling tools make more sense than parametric tools.