Feature-Based Interoperability Goes Global

By David Prawel

The global economy’s effect on manufacturing is well documented, and it’s no secret that competitive pressures have decentralized production facilities. But more and more design work is being distributed globally as well and, as designers, engineers, and manufacturers become further separated by time zones and cultures, effective collaboration has emerged as the foremost key to success.

Few single issues present a more formidable impediment to accelerating design cycles and time to market than sharing critical data. Studies have suggested that poor MCAD interoperability costs the automobile industry $20 billion per year. To collaborate and communicate effectively, designers must be able to easily and accurately share their MCAD data and related information.

Effective MCAD interoperability depends on many factors, such as intended use of the resulting model, destination application, and the business relationship of those involved. While industry standards like STEP, IGES, or STL might suffice for the analysis team needing only the geometry, topology, and some physical properties of the model, a more robust tool is needed when a close partner needs all features, history, and other critical design elements of the MCAD model to modify its features.

While there are many excellent tools for sharing geometry and topology, interoperability of full-featured models has proven much more difficult. Until recently, because of the limitations of industry standards like STEP when it comes to feature-based translations and a lack of motivation on the part of many MCAD vendors, robust MCAD models were absolutely required, and the only choice was to recreate the model.

New technology has emerged that is proving successful at sharing fully featured models. Most of the major vendors of data-exchange software provide tools for feature-based translation. Geometry-based translation comes up short by “sharing and recalculating mathematical representations of part geometry and topology,” Elysium Director of Product Marketing Peter Heath explains, but “feature-based translation uses the features and history of a model as its recipe, and tells the destination CAD system what the recipe is to recreate the model.”

Peter Roy, senior CATIA designer at Pollack, a manufacturer of sensors and switches for instrument panels, has become a real believer in feature-based translation. Like most suppliers, Pollack uses many of the major design systems, primarily Pro/Engineer, and delivers designs to customers in the customer’s preferred MCAD format. The company’s approach to translating features had been to remaster wire-frame models into STEP files before sharing—an expensive and error-prone process. After an exhaustive cost analysis of person hours spent remastering data using this approach compared to a feature-based approach, they decided Proficiency’s Collaboration Gateway would be a wiser choice. The company anticipates saving hundreds of person hours annually.

However, as with all technologies, there are trade-offs to consider.

It can be difficult to set up the often complex feature mappings that are required to make feature-based translation work dependably. Each MCAD design tool has its own representation for each feature. Some features are simple and can easily be mapped to other MCAD systems, but some are complex. To implement feature-based translation, one must set up a procedure to identify the features in the source MCAD system that should be mapped to the features in the destination MCAD system. This can take time and requires a lot of experience. As a result, feature-based translation tools can be relatively expensive to implement.

For this reason, feature-based translation is most valuable when you’re dealing with large MCAD databases, rather than smaller sets of MCAD files. By spreading the initial investment over many translations, the solution increases in efficiency and becomes more cost effective. It is particularly well suited to migrating MCAD databases between MCAD systems, either when upgrading to a new major version or when changing MCAD systems. Feature-based translation can also provide good value in cooperative design situations, as indicated above, and when starting a new project where the design will be sufficiently different to necessitate near complete remastering of the model anyway.

But for translations of smaller sets of MCAD models and when the applications don’t require features, as in most analysis and machining applications, geometry-based translation is more suitable. Stuart Thurlby, president of Theorem Solutions Inc., reports seeing much less demand from customers for feature-based translation than for translating geometry.

Commercial feature-based translation tools use two basic strategies. The more common approach is to deliver as many of the features as possible in the destination MCAD file using feature-based translation technology, and then to use geometry translation as a worst-case for those features that can’t be translated correctly. This is the approach used by Elysium with its CAD•feature translator.

Another company takes the other approach. Ben Parrish, technical support manager at Translation Technologies feels that “most MCAD processes and applications don’t require features, but when they do, feature-based translation is the only answer. Our philosophy is that when you need features, you need all the features. You need the complete model. So we provide added-value services to make sure the translation is complete and accurate.”

Accuracy and quality is very difficult to assess in feature-based translation. In many cases, it may not be possible to map features between MCAD systems; for example, a lofted surface in one MCAD system may be geometrically quite different in another. Or, the mapping, when completed, produces an inaccurate MCAD model. According to Elysium’s Peter Heath, “it’s very difficult to assess the quality of a model after feature-based translation. With geometry translation, the model geometries must match. But with feature-based translation, you’re comparing the recipes for making the model, not the geometry of the model.” For some types of models, feature-based translation is relatively dependable. For example, prismatic 2D parts with relatively simple features translate dependably. But surfaces, blends, and drafts can get tricky.

Two types of accuracy need to be considered. First, the geometric accuracy of the result must match the required precision. Software tools are available that will check this, usually by measuring mass properties and other analytical metrics that compare two MCAD models for “exactness” at a given precision. A second type of accuracy, which is much more difficult to analyze and resolve, is an assessment of the feature mapping itself. For example, two different MCAD systems can place a fillet on an edge with the same radius and produce different results. MCAD systems define swept curves differently, such that one might allow a sweep to “float” more than another as the curve rounds a corner. The result will be two very different models. In the final analysis, successful feature-based translation has less to do with the quantity of features, and more with the type and complexity of the features. This makes accuracy difficult to assess and success with feature-based translation difficult to predict.

Another potential disadvantage of feature-based translation is confidentiality: Sharing feature definitions with the wrong people could spell disaster. Proficiency addresses this risk in the new release of Collaboration Gateway by providing security management to help you control who sees what. For example, some partners might be allowed to receive the features, while another only the geometry, and someone else might only receive the assembly structure. This powerful new capability should go a long way to help protect intellectual capital, removing another barrier to broader deployment of feature-based translation tools and methodology.

Excellence in MCAD interoperability is a fundamental element of collaboration and efficient communication. Feature-based translation is a good alternative for migrating large MCAD databases or for switching systems, so compare the cost of software and implementation to the time and cost of recreating MCAD models, and don’t forget to factor in the opportunity gained or lost in the equation.

David Prawel is the founder and president of Longview Advisors Inc., a consulting firm serving manufacturers. His career spans 25 years in high-tech businesses and he has published many articles covering 3D MCAD, product design, and related subjects. Send David your comments about this article via e-mail by clicking here. Please reference Interoperability Goes Global September 2005 in your message.

Product Information

Elysium Inc.
Southfield, MI

Proficiency
Marlborough, MA

Theorem Solutions Inc.
Loveland, OH

Translation Technologies Inc.
Spokane, WA