Stephen Wolfe, P.E.
Publisher, Computer Aided Design Report
In January 1997, we wrote about work by the International TechneGroup Incorporated (ITI) to analyze the geometry of three-dimensional CAD models. ITI has found that many models contain geometric elements that cause problems for programs and processes that employ CAD geometry. The errors may include cracks and sliver faces, poor accuracy, and loss of the relationships between edges and faces. Sometimes solid models enclose parts of themselves. Sometimes edges become disassociated from model faces or contain loops or knots.
These errors occur even in data created with some of the best solid-modeling systems, such as Pro/Engineer or systems based on the Parasolid kernel developed by Unigraphics. Sometimes they are the results of operator error. But poor operator training or sloppiness is not the cause of all bad data. Even skilled workers make mistakes. In the course of developing a design, it is common to add or delete features or to move them around. Occasionally, a feature may be squeezed until it is nearly invisible. Or two features may overlap, leaving a sliver or crack between them. In the rush to finish a project, such errors can easily go unnoticed.
Setting model tolerances too loose can create unacceptably large gaps between model faces and edges. But modelers with very tight tolerances can cause errors by generating unrealistically small faces and edges needed to fill in the gaps between surfaces.
Even if the only purpose of CAD models is to validate fit and appearance or to make drawings or colorful images, the types of errors identified by ITI may cause problems. Poor model quality may prevent additional features from being added to a model. It can also prevent models from regenerating after a change is made. Such stubbornness on the part of solid-modeling software is often blamed on inadequacies of the program when in fact it may be caused by a bad set of data.
And if CAD geometry will be used for finite element analysis or translated to another CAD system via IGES or STEP, then seemingly insignificant errors in geometry can cause big problems. Finite-element mesh generators -- even those made for a specific CAD software program -- can stop working for no obvious reason. And data sent through IGES or STEP may fail to "knit" or "sew" into a valid solid in the receiving system. Even straightforward processes, such as programs that produce the stereolithography (STL) format used by rapid prototyping systems, may fail because of cracks or slivers in the geometry.
Diagnostic Aid
On February 27, ITI began shipping a program for checking solid-model geometry called CADIQ. The program reads CAD models from Pro/Engineer or CATIA. It also reads Parasolid geometry employed by Unigraphics, SolidWorks, DesignWave, ICAD, and soon-to-be-released versions of Bentley Microstation and Solid Edge. ITI is currently developing CADIQ versions for SDRC's I-DEAS and for the standard formats of IGES and STEP.
CADIQ analyzes geometry for the types of defects listed above. It produces tabular listings and a statistical analysis of errors, and individually identifies errors in the CAD model. With this information, CAD operators can rework models in regions with known problems to eliminate faults that can cause other processes to fail. CADIQ sells for $3,600 per license and runs on Windows computers as well as Unix systems from Sun, HP, SGI, and IBM.
Do You Need It?
If your company translates models from one system to another or employs finite-element analysis, numerically controlled tools, or additive rapid prototyping systems, CADIQ belongs in your toolkit. Identifying geometry problems can help CAD operators fix them more quickly and save both time and frustration.
CADIQ also can be a good investment if your engineers are frustrated by mysterious behaviors in their solid models. Analysis of the data may reveal that difficulties are caused by geometry defects that can be repaired instead of worked around.
The use of CADIQ can be blended into your company's design procedures in one of several ways. The first is to employ it only on models that cause trouble. This approach works if all your processes are under the control of a small group of people who communicate well. In larger organizations, however, the people who create CAD models may not know how others will use them. In this situation, it is wise to run every model through CADIQ before it is released for production. This extra step takes time and costs money, but it will save money in the long run by making models less trouble-prone for others.
It is also a good idea to run CADIQ periodically (perhaps each night) as complex designs are being developed. This procedure enables errors to be corrected before they are compounded. It is better, for example, to eliminate a crack in a feature before it is replicated in an array.
The reports produced by CADIQ are laden with terms used by the folks who program solid modeling software. Many designers will find them intimidating. (We certainly do.) It is therefore wise to have a few people who are trained to interpret the reports assisting other workers in correcting problems. ITI offers such training and consulting services in addition to the software.
CADIQ is not a substitute for Prescient's Design QA software (described in the January 1998 CAD Report). Design QA checks feature definitions and other modeling practices. CADIQ checks geometry. Both tools should be part of an engineering release process that assures that CAD models can be used as digital masters for manufacturing and analysis.
It's unfortunate that CAD users must employ sophisticated data-checking software to avoid seemingly erratic behavior with CAD applications. CAD salespeople rarely advertise this fact when they are pitching senior managers on the concept of digital product definition. But it's a fact of life that even the best CAD software isn't very smart.
Some day, CAD systems may perform internal checks to validate geometry as it is created. But we believe that day is a decade or two away. Meanwhile, tools such as CADIQ deserve serious consideration.