The Modeling Pendulum Swings

Vendors share their perspectives on the evolution from explicit interaction with geometric objects to indirect manipulation of parametric formulas and features.

A few weeks ago I was invited to attend an event hosted by Collaborative Product Development Associates (CPDA) titled “Product Lifecycle Management Road Map 2007.” This event was the 14th annual PLM Road Map conference that, from the beginning, has focused on the evolving elements that continue to influence product development. It’s an interesting format where CPDA's analysts and key industry players share their experiences to address strategic PLM topics.

One of the sessions that piqued my interest was a panel discussion titled “Leveraging Design Models and Reuse Downstream: A Shift in the CAD Modeling Pendulum.” Moderating the panel was Ken Versprille, a CPDA partner and PLM research director. Participants in the panel included Robert W. Bean, P.E., executive vice-president of Kubotek USA; Richard Bush, director of marketing for Siemens PLM Software; Aaron Kelly, product manager with SolidWorks; and Michael Payne, CEO of SpaceClaim.

As you might imagine with this mix of panel participants, things started out with a civil tone, but it quickly got heated as both the moderator and audience members raised various interoperability issues.

The discussion boiled down to the fact that MCAD geometric modeling technology has experienced major leaps in features and capabilities over the past four decades from 2D to 3D wireframe, solid, and surface modeling. Methods for interacting with construction geometry have also evolved from explicit interaction with geometric objects to indirect manipulation of parametric formulas and features that determine model shape.

As this evolution has progressed, however, users have been forced to deal with changes in how they approach MCAD design. Core geometry structure and architecture advances in MCAD packages have led to a movement to direct geometry-editing techniques that will continue to swing the modeling pendulum from history-based to non-history-based approaches. This panel of vendors brought together representatives who are among those leading this technology shift, and they each had insights and advice on balancing the two different approaches to MCAD.

It’s clear that parametric, history-based MCAD has dominated the market since the late 1980s, although CoCreate and Kubotek USA are notable exceptions, with both promoting non-history-based (also known as direct) modeling for some time. However, several other vendors in the past year or so have also gotten on the non-history-based bandwagon, and as you might expect, they are all using their own terminology (marketing jargon) to describe roughly the same thing.

Today, the majority of mechanical CAD products are parametric, history-based applications. However, with a history-free architecture, design changes to 3D models can be easier to make without needing to know how the design was originally created. At first some users may not like the absence of the history, especially those who have experience and a comfort level with parametric, history-based systems. However, I think that many users will appreciate the shorter learning curve and the faster changes that can be made to a design without having to deal with its history.

Parameters also can be a hindrance where design change speed is concerned. In parametric models, each entity, such as a line or an arc, has parameters associated with it. These parameters control the various geometric properties of the entity, such as the width and height of a rectangle or the radius of a fillet. They also control the locations of these entities within a part model. Parameters can usually be changed to create a desired part for design optimization or maintaining original design intent (although that’s another potential point of contention).

History-based parametric modelers create and maintain a record of how a part model is built. When parameters in a model are changed and it is regenerated, the CAD application repeats the operations from its history, using the new parameters to create a new part model.

While parametric history-based modeling can be a powerful design method, it can also complicate things and make models created using this method difficult or impossible to work with, especially late in the design process, based on inherent dependencies. This is exactly what makes non-history-based packages shine for collaborative design teams who need to perform design changes or reuse design data without having to concern themselves with constraints, parametrics, and history-based feature trees.

This is what struck CPDA’s Versprille last year, beginning with Kubotek USA and its direct dimension-driven editing technique, then with the non-history-based approaches that began to proliferate shortly thereafter from other vendors. Let’s take a brief look at each of the technologies offered by the various panel members’ companies.

Kubotek USA unveiled direct modeling capabilities that redefined the ability to design, edit, and reuse 3D solid models with its ability to change solid model geometry by selecting and editing dimensions in real time without predefined constraints. Kubotek’s direct dimension-driven editing requires no fixed relationships built into the dimensions or any special data or entity types. This flexibility lets you decide during an edit how the change should take place as well as what features or faces should be included in the change. Underlying face selection technology, called Face Logic, allows you to add driving dimensions to a 3D solid at any time, regardless of how the model was originally defined. Face Logic technology never requires constraints on the model, nor is the model constrained by rigid design intent or subsequent model edits. Instead, the technology lets you make edits whenever you want to make them.

A somewhat similar process is enabled with a new feature in SolidWorks 2008 called Instant 3D that lets you modify a model based on the relations you have in a sketch. For example, on parts created with SolidWorks 2008, you can change a boss to a cut and back, although you cannot perform this on models created in previous versions. You can also create sections of parts and still change features. Although heavily reliant on sketches, this tool has real potential for SolidWorks users.

Recognizing an ongoing conflict between the rules of parametric design and the constraints placed on designers arising out of intent history-based systems, UGS launched what it calls “Design Freedom” in NX 5. This capability lets you modify geometry without having to understand the history of, or undo, the design tree. NX 5’s various design tools eliminate the constraints imposed by parametric systems. For example, Advanced Selection Intent automatically selects geometry and infers the correct relationship, which allows you to quickly make design changes. The ability to work on geometry without defined features or history adds a new level of flexibility, enabling changes to be made relatively quickly.

SpaceClaim Professional 2007’s Flexible Modeling technology provides an adaptive design environment that supports unanticipated design directions, making SpaceClaim well suited for conceptual design, detailed design, and modification. Its geometric inferencing works unobtrusively and in real time to highlight design similarities, such as equal radius holes or coplanar surfaces, to assist you during geometry creation and modification. SpaceClaim’s flexible modeling environment provides the freedom to explore design changes without needing to know how the model was built. Helping the cause is SpaceClaim’s Open XML data format that makes all design data accessible for PDM and PLM.

So, will history-based, parametric modeling go away? Admittedly, a history-free design approach, while not for everybody, can provide some surprising benefits. According to Kubotek USA’s Bean, “The history-based, parametric paradigm will not go away, but the fact remains the paradigm just does work well for dealing with pure geometry. Maybe a better way to go is ‘parametrics on demand.’” I tend to agree with this opinion; while many of us have had history-based methods ingrained in us, they aren’t always the best way to go. It seems that a growing number of vendors are thinking along the same lines.