Why History-Based Modeling Still Matters
Imagine the horror. The untimely disintegration of a particularly brittle parametric model explodes right in the face of an unsuspecting designer, an unwary victim of a thousand unresolved dependencies. What manner of creature developed such an intrinsically complex system? After all, direct modeling is the new hotness. History-based modeling is old and busted. We're going to push and pull our way to a better tomorrow. Hold on there just a minute, slick.
Complaints about history-based modeling are well known: models that fail if you look at them funny, the chore of defining explicit constraints, or the inability to easily discover model controls and dependencies. However, blaming parametric modeling for these things is kind of like blaming rollover accidents on trucks. Because you can rollover a truck pretty easily we should all really be driving sedans, right? Rollover accidents are chiefly caused by inattentive driving or driving like a lunatic. Turns out you can model like an inattentive lunatic too.
That's not to say that direct modeling doesn't have its place in modern design. Of course it does. After all, direct modeling was old before it was new again. But much like trucks and sedans it's all about the right tool for the right job. That's where CAD systems that support both or offer hybrid solutions really show their strength. Direct modeling is your quick fix; history based is for organizing with the intent to re-use. But where does each technique fit best? For direct modeling, it's front and back end. For history-based, it's everything in between. Let's explore each of those phases with my own recollection of a particularly complex fairing design:
Initial design concepts are fast and loose. Possibilities abound, and trying to account for such wide variability in a model is counterproductive. Direct modeling shines here. But this phase doesn't last very long, because very few designs are a truly clean sheet. We know something of what they are going to look like, based on prior experience and stable solution options emerge quickly. Unless it's furniture or assorted decorative plastic. But in those cases I'd argue that's not engineering but rather industrial design.
During this stage in the fairing design I was lofting rough cuts of the Outer Mold Line (OML) myself, developing load paths and notional substructure with the stress analysts and creating all kinds of shapes to approximate equipment and structure that didn't exist yet and constantly rearranging them all. Each successive concept was dramatically different, so the immediacy of direct modeling made the difference here.
In this stage, design automation, and hence history-based modeling is critical. The design is much better understood. The iterations don't stop here, they only change in character. Wild concept upheavals are replaced with progressive refinement, driven by refined analysis from a wide variety of disciplines ranging from the expected (aerodynamics) to the esoteric (electromagnetic effects and lightning strike analysis). Parametric design makes these iterations as cost effective as possible by maximizing re-use at this stage. Additionally, they support creation of engineering documentation at this stage, be it MBD or your favorite drawing format.
During this stage of the fairing design, I was receiving regular surface updates from CFD runs with newly tweaked molded surfaces or changes in structure size and/or position from stress. At this point the assembly was relatively mature, consisting of hundreds of interrelated parts. They were driven by a steering file which contained all the relevant datum references, surfacing, and parameter descriptions in one place. I would take updated surfaces, replace it in my steering file, adjust datums and update all the structure simultaneously.
Something like that is just not possible with direct modeling. I didn't have a hundred engineers to go push on a hundred parts every time a new surface or stress change trickled down. I had looming deadlines, and a resilient parametric model was the only way to get there.
Direct modeling once again becomes relevant at this stage. Chances are production changes occur well after initial design release and involve different people, sometimes on different CAD systems on perhaps on different continents. That's where interpretation and update issues are most likely to crop up with parametric modeling, but again properly resilient models can hold true, provided editing at this stage is on the same CAD system. In the fairing design, the latter was definitely the case and direct modeling was unnecessary, but depending on how the company was structured or what partners were involved this could have been very different.
In short, both parametric and direct modeling are not mutually exclusive, they are each tools that should be present in any designers toolbox. It's all about re-use versus accessibility to make the design change. If each design iteration is dramatically different from the last, accessibility rules and so does direct edit. If each design iteration is a variation on a theme (i.e. an evolution), history-based modeling provides a clear cost advantage by maximizing re-use of work already completed. No designer these days should go without both.
The Next Generation of PDM
More teams are using Cloud, Analytics, Mobile, and Social tools to speed up product development. Independent analyst firm, Consilia Vektor, explains how this changes Product Data Management (PDM) as you know it and how this can help your team work smarter.
About the author: (Ed Lopategui)
Technology evangelist, entrepreneur, and aerospace engineer specialized in the software tools and technology which enable engineering, design, and product development - PDM, PLM, CAD, CAE, CAM. Any views, opinions, prophecies, and sarcastic remarks are my own and are in no way associated with any current or past employers.
All posts by Ed Lopategui — Follow on Twitter