3D GD&T provides higher-quality parts and greater productivity, but speed bumps slow its full acceptance into the future.

Authored by:
Bryan R. Fischer

President
Advanced Dimensional Management LLC
Sherwood, Oreg.

Edited by Leslie Gordon

Resources:
Advanced Dimensional Management LLC
, (503) 625-2480, www.advdm.com

In Chapter 26 of the 11th Edition of the IHS Global Drawing Requirements Manual, Fischer details the different design deliverables, offers advice, and explains the extra steps needed to put the techniques into practice.

Acknowledgements:
Thanks to Rick Zuray, Tom Thurman, and Doug Cheney for their support in preparing this article. Thanks to Lothar Klein, Ed Paff, and Jochen Boy, who along with Rick and Tom, are comembers of the STEP AP242 GD&T and PMI team.

Today, industry stands at the crossroads of the old world of 2D drawings, the world of annotated 3D models, and the future where 3D engineering data can be consumed and understood directly by software. Today, the two international geometric and tolerancing (GD&T) standards that cover annotation largely define engineering specifications in terms of how they are graphically presented on a 2D drawing or annotated 3D model. This “presentation data” is only useful for human consumption. For computer systems to consume data without human intervention, the data must be explicitly modeled in a manner that captures its underlying meaning. This underlying meaning is called “representation (semantic) data.”

Fortunately, several groups are working to develop semantic models of GD&T and other annotation in STEP and other standards that will help move industry away from the world of 2D drawings to that of 3D model-based product definition (MBD). MBD uses product definition data (datasets) to provide complete specifications for components and assemblies. Both ISO and the Long Term Archival and Retrieval of product definition data (LOTAR) group are currently working on new standards.

No doubt, modern 3D CAD software lets engineers model, visualize, and test accurate representations of complex geometries much more quickly than in the past. But CAD geometry alone cannot define acceptable as-produced parts and assemblies.

At best, a CAD model depicts a theoretically perfect part or assembly. Mechanical components work in the physical world, and perfect parts have never been made. So, allowable variation (tolerances) must be part of the product definition.

However, because a CAD model represents a theoretically perfect entity, simply applying plus and minus tolerances to it in a general note or tolerance block proves insufficient. A CAD model provided as all or part of the design deliverable makes it legally binding, similar to an engineering drawing. The only way to clearly and unambiguously define the acceptable geometry of as-produced parts — through the use of GD&T.

Moving into the future, that there are two international GD&T standards poses problems. ASME Y14.5-2009 is defined by the American Society of Mechanical Engineers (ASME). ISO 1101 is defined by the International Standards Organization (ISO). At first glance, the two standards look almost identical, but there are significant differences. This makes it difficult for multinational corporations whose staff must understand both standards.

New standards for product definition
In the past, when 2D CAD started moving to 3D, it became possible to make a 3D model and link it to a 2D engineering drawing. This associativity meant that a change to the 3D model changed the views on the associated 2D drawing. In this era, engineers created designs in 3D, but documented products with 2D drawings. The 2D drawing was, thus, the engineering design deliverable.

In the last decade, several CAD packages have included tools for documenting products, allowing engineers to apply GD&T, notes, surface texture symbols, welding symbols, and other annotation to 3D models. This advancement was welcome to many, and made others apprehensive because it represented a change in the status quo.

In 2003, ASME released Y14.41-2003, which was used as the basis for the ISO 16792:2006 standard. It establishes the rules for using 2D and 3D digital data as the design deliverable and, importantly, sets the groundwork for using annotated or attributed 3D CAD models as the sole design deliverable. The aerospace and automotive industries play a key role in the move to eliminate or reduce reliance upon 2D drawings, but the committees developing these standards include company representatives from many other industries.

Reasons to reduce or eliminate 2D drawings include data interoperability and engineering-data reuse. 3D CAD data exports to other programs for use in analysis, manufacturing, and inspection, eliminating the need to manually reenter product-definition data. Manually reentering data means redundancy and greater chance for errors, while associativity supports data interoperability and fewer errors.

Also, personnel can easily change dimensional data on 2D drawings, which often causes problems downstream in manufacturing. And, less annotation is required in axonometric views to communicate the same information as in traditional 2D orthographic drawing views. In fact, once individuals are familiar with the technique, they more easily understand properly formatted annotation in 3D. Again, time equals money, so the elimination of drawings can reduce engineering costs.

However, the migration to 3D GD&T and MBD does carry risk. For example, workers who rely on drawings will no longer be able to do their jobs without new tools and training. Proper forethought should avoid this problem. Another risk: Personnel often resist change. Many workers in drawing-based settings are comfortable with the way they do their jobs and ill-disposed to change. A shift in cultural mind-set must spread throughout an organization, from top to bottom.

Design deliverables
Overall, the CAD model includes the geometry that represents the product, and the supporting geometry and annotation (also called the product and manufacturing information or PMI) needed to complete the product definition.

Most mechanical products today are designed in 3D CAD. High-end software such as Creo Elements/Pro (formerly Pro/Engineer), Catia, and NX give engineers options in documenting designs. They can create a design in 3D, but document it on a 2D drawing; partially document a design on a 2D drawing and complete the product definition with the 3D CAD model data; or completely define the design in 3D, documentation and all.

Generally, engineers choose an option depending on the needs of downstream operations such as manufacturing, inspection, and assembly. However, most downstream activities already use 3D data, either in native CAD format or translated into STEP or other neutral formats. It will take a long time for MBD and annotated 3D models to become the primary design deliverable throughout industry. However, in some industries, the transition is already underway, and it will happen quickly. That said, there are many additional considerations and best practices needed to implement MBD that are not found in the ASME and ISO standards.

Presentation versus representation
Data works in engineering product documentation in different ways. One way involves presentation data — for example, format, layout, and symbols. Another way involves representation data, the underlying mathematical and legal meaning.

Presentation data targets human consumption. Users understand the meaning of the data by reading a drawing or annotated model and correlating it with the specifications and standards referenced in the product documentation. With presentation data, layout and format are critical. Moving a note to another location, moving a tolerance to a different view, or rotating a dimension 90° generally changes the meaning of the data.

Representation data, on the other hand, reveals underlying meaning in mathematically accurate terms. Representation data remains independent of presentation — it does not rely upon the graphical depiction of the data (the presentation data). Other software programs and automated manufacturing processes can use representation data directly without human intervention.

Use cases
The many uses for design data generally fall into the broad categories of data-rich and visualization.

Data-rich use cases require all the information needed to perform some activity, such as manufacturing or inspection. If annotation or PMI is included, downstream activities need to get the PMI information in its proper form. In the future, these activities will require full semantic PMI data to consume information without human intervention.

High-definition models contain a complete semantic definition of the product (different from the high definition of videos, which merely have a higher resolution). Native CAD data and other data formats today are almost complete enough for industry-wide adoption of 3D MBD. The next five years should bring exciting developments.

In visualization-use cases, humans must either merely understand the visually presented data or also enter it into another system. Industry and standards communities are expanding the role of visualization formats, such as the JT lightweight data format, to make them more data-rich and, thus, increase their use in automated life-cycle processes.

How the standards relate
To develop fully semantic data models, the hierarchy of standards needs improvement. Standards currently reside in a three-tiered structure:

Tier 1 standards define the techniques for applying PMI and the meaning of annotation and PMI symbols. Mostly drawing based, the standards were developed during the era of 2D drawings. In the ISO system, standards related to drawing annotation include ISO 128, ISO 129, ISO 1101, ISO 1302, ISO 2553, and others. The ASME Y14 standards cover engineering drawing practices, including ASME Y14.5, ASME Y14.36, and others. AWS develops welding standards, and AWS A2.4 covers symbols for welding, brazing, and nondestructive examination.

Tier 2 standards ASME Y14.41-2003 and ISO 16792:2006 define model construction and usage. They establish rules and set the groundwork for developing engineering design data in 3D, managing the data, and using the data as a deliverable instead of drawings. Both in their first revision, the standards slant heavily toward presentation data, presenting engineering data in axonometric views and 3D. They are similar in scope and requirements. Along with setting rules for PMI and the presentation of geometry, the standards provide examples and set guidelines for the logical association of data elements. The goal is to make data usable for manufacturing, inspection, and assembly, and revise data, manage revisions, and data reuse. ASME Y14.100 and MIL-STD-31000 also play a role.

Tier 3 standards define data models for presenting and representing product geometry data and PMI data in native CAD data formats, neutral formats such as STEP and IGES, and, to a lesser extent, visualization data formats. High-end CAD packages have implemented semantic representational models of GD&T and other PMI, but these models are still incomplete. That’s because the models can only be as complete as the definitions in the Tier 1 PMI-defining standards. High-end CAD software vendors are working to include STEP’s new MBD capabilities; this work is still in progress and not complete. That there are many customized, industry and company-specific symbols and annotation techniques poses additional challenges.

The future should bring a full semantic representation of all relevant PMI in CAD, STEP, and other formats. More downstream applications will be capable of directly importing this data without human intervention.

What does “semantic” mean?
In the context of data modeling for engineering product-definition data, semantic data and representation data are synonymous.

Semantic data captures the meaning of a character, word, phrase, sentence, paragraph, specification, or symbol without using visual characters or constructs, such as the letters, graphical symbols, lines and arrows used on engineering drawings. Semantic data includes the meaning of a specification, but it does not include its graphical depiction that is needed for a human to understand it. For example, a semantic model of a linear dimension includes all of the information needed to understand the specification without the aid of any of its graphic components, such as dimension lines and extension lines, their direction, arrowheads, and the dimension value.

© 2011 Penton Media, Inc.