Quality drives productivity and is now, more than ever, seen as an enabler for advanced manufacturing, rather than a cost that has to be endured,” stated Ken Woodbine, president of the Metrology Software Division at Hexagon Manufacturing Intelligence (North Kingstown, RI). He also stressed that a metrology process that simply reports measurements accurately may not be enough. Quality engineers must create measurement programs that give confidence that the part is meeting the measurement intent of the design and designer as well as the needs of the assembly the part is going into. The art of creating measurement programs is still commonly done as in the past, through highlighted mark-ups of 2D paper drawings or even informal discussions with on-site designers. But these are proving insufficient and even archaic in an increasingly global and complex supply chain.
A third solution is electronically attaching quality information to a 3D CAD model, making it more intelligent. This is part of the growing trend towards Model-Based Definition, or MBD. “Embodying quality information such as GD&T electronically with the master 3D CAD model is the most efficient way of communication,” stated Woodbine. He sees many of his key customers moving in this direction. PC-DMIS has routines to help automate creation of measurement programs using such embedded GD&T as input. It supports creating these automated programs for CMMs, laser trackers, and portable CMMs, as well as specialized routines for gears and blades.
One problem Woodbine was quick to point out is that GD&T alone does not specify the methods or machines for measuring—it only states the tolerances to be met. “But if it can be consumed by measurement software, as we can with PC-DMIS, we can convert into a measurement program,” he said. “It saves a lot of time and creates confidence in the result, as long as that GD&T is created correctly. PC-DMIS has routines for checking the validity of attached GD&T.”
Besides accuracy and design intent, another important problem MBD is solving is speeding up data flow, according to Phil Hewitt, manufacturing product manager at Autodesk Inc. (San Rafael, CA). The flow of data in today’s complex supply chains is an important bottleneck. “A 2D representation on a drawing requires human interpretation,” he said, which, besides being prone to error, is also slow. MBD offers the possibility of automating the data flow. “Machine-readable semantic data can link the real requirements electronically to the relevant features in the CAD file,” he said.
He also sees this in the context of a broader vision. Attaching GD&T information to CAD models is essential in meeting the vision of Industry 4.0 or Smart Manufacturing. “Think of it as enabling the full digital twin,” Hewitt explained, especially linking it to in-process control, such as inside CNC cutting machines. “Realizing in final Inspection that there is a problem is too late. We see measurements becoming much more integrated with manufacturing processes—a common example is on-machine verification—and this can be greatly enabled by explicit machine-readable data.”
However, he also noted the process of embedding and using GD&T is not fully mature. It is just starting to become useful. While in many instances it does a better job than previous methods of transferring what the design intent was, according to Hewitt, it is still part of an ecosystem that needs to grow in unison. The data must be comprehensive and all systems needing to use it must have sufficient capabilities to be able to exploit it. For example, PowerInspect can read in the typically high-end CAD formats that include PMI (part manufacturing information) in them, including CATIA, NX and SolidWorks. PMI appears as GD&T callouts that the user selects to create an inspection plan.
A sign that the ecosystem is indeed growing is the ability of mid-priced CAD systems to output models with MBD, noted David Olson, director of global sales and marketing for Verisurf Software (Anaheim, CA). According to the company, Verisurf features an open 3D measurement platform strategy to create inspection plans for many stationary or portable CMMs and analyzing point cloud data from scanners. “Verisurf has always been able to read in attached GD&T with high-end CAD models in their native format, but mid-priced CAD systems like SolidWorks, SolidEdge and Autodesk Inventor can now support model-associated GD&T and PMI data,” he said.
He also believes that whatever barriers now exist in model-associated GD&T and intelligent tolerance information will be relieved by the adoption of the STEP AP242 standard. This is an ISO standard for managing model-based engineering and its development is managed by NIST, according to Olson. “This format includes the model topology as well as PMI data that includes the intelligent GD&T data,” he said. The word “intelligent” carries specific meaning for Olson—features such as surfaces or holes will have tolerance annotation in the part history definition tree in the CAD model. Intelligence is important because then software can more easily automate creating measurement programs. “A standard opens the doors for entrepreneurship by unlocking the embedded design-intent information,” he said.
“All my customers are looking for more intelligent CAD models,” he added, “and looking for metrology software to support the import of those intelligent models to create measurement programs. It promises cost and time savings by not forcing redundant effort to recreate that information, or worse, introduce errors through misinterpretation.”
All of this points to cost savings through universally read data formats and even universal software. The goal of the president of InnovMetric Software (Québec City, Québec), Marc Soucy, is to provide a single software platform that can read in any CAD data and program any machine. “People want to reduce the number of metrology software packages they are using,” he said. “We believe there is a huge cost reduction in reduced training and personnel costs and it makes it easier to share data.” InnovMetric’s PolyWorks creates programs for portable metrology devices and recently expanded into CNC-driven CMMs as well.
In the spirit of universality, Soucy also recognizes that a software that can use GD&T in an intelligent CAD system is important. His company’s PolyWorks Inspector has a Smart GD&T toolset module that comes standard for reading in attached GD&T and helps automate creating inspection programs. “Creating an inspection program from a CAD model with attached GD&T is very involved, a tough job. A lot of mathematics need to be developed to achieve that,” Soucy said. “There are 20 or 30 man years of effort to make a GD&T engine that really works.” The company’s Smart GD&T toolset was created with a partner in the field, Multi Metrics Inc. (Menlo Park, CA).
He offers further cautions. “In theory, the concept of MBD is great, but I don’t think it works all that well in actual implementation,” Soucy said. “The only area where MBD really works is with GD&T. But even then, there is more to measurement than GD&T. The reality is that GD&T only represents a fraction of what measurement specialists need to measure when they analyze geometry of manufactured products.” Specific industries such as automotive sheetmetal use unique measurements, such as a feature line profile radius or gap-and-flush. These are not covered in standard GD&T specifications. He sees customers creating some of their measurement programs from attached GD&T and then creating more specific measurements as well.
The key point here is that metrology data is used in a decision loop in a specific way. Measurements in and of themselves provide no value to the customer. “We strive to provide critical dimensional characteristics and analysis back to manufacturing or engineering in real time,” agreed Bertrand Gili, president of the Metrologic Group (Meylan, France and Wixom, MI), suppliers of the Metrolog X4 inspection package. “Measurement is good, but having meaningful results directly from 3D metrology devices with clear instructions and clear feedback on how to, say, adjust tooling—that is the true value of measurements,” he stated. While faster inspections today might mean robotic data collections and automation, it also means understanding exactly what those critical characteristics are, as rapidly as possible.
Again, this is where embedded GD&T in an intelligent CAD model is so valuable. Metrologic has been in the game a long time, according to Gili. “We started working with embedded GD&T on a project with Boeing in 1999 on the 7X7 program,” he stated. “Many customers are realizing that GD&T is a very serious topic.”
He stresses that embedded GD&T needs to be encoded properly, using strict semantic protocols, to ensure a proper decoding of the data to create valid measurement programs. Over-constrained datums and other GD&T that is annotated as text, but meaningless in a mathematical sense, needs to be detected and dealt with. That is why that strict semantic coding sequence of the embedded GD&T is important. “Semantic tolerances are meaningful GD&T, encoding it in a particular way so if three different individuals are creating an inspection program from that semantically correct GD&T, it will systematically decode it the same way. If it has not been encoded properly, it cannot be decoded,” he said. In those cases, Metrolog X4 will ignore it. The software also boasts of a new GD&T processing engine and expert system for handling complex cases.
Other software programs offer routines for checking validity. Look for future releases to help correct invalid GD&T the programs detect. While some correction of GD&T to finalize measurement programs is probably inevitable, it is best if the tolerancing is verified and validated by the design engineer in the CAD system itself.
A good example of this comes from Siemens PLM Software (Plano, TX). Producing valid tolerancing to begin with was stressed as part of a lecture from John Zhang, solution architect and business consultant, at the Siemens PLM World conference on May 8, 2017, in Indianapolis, IN. He described a process he called model-based quality that answered the challenges of ensuring valid GD&T is attached to CAD in the first place.
“To avoid unnecessary cost, you need to avoid over engineering a part and over inspecting it,” he said, while maintaining quality and robustness. That is where variation analysis techniques are useful, such as the Siemens PLM Variation Analysis (VSA) software tool, which is offered within its Teamcenter Lifecycle Visualization toolset. The VSA tool simulates manufacturing and assembly processes and predicts the amounts and causes of variation, the reason anyone attaches GD&T specifications. The model-based quality process he describes is iterative, starting with the 3D CAD and an initial GD&T specification. This is then tested virtually in VSA to analyze the tolerances, adjusting and validating it to ensure it results in acceptable variation at minimum cost and acceptable quality.
Once validated, the GD&T is then annotated and put into the PMI that would be attached to the CAD—ready for use in downstream applications and in creating inspection plans and programs. “The VSA program will clean up incomplete GD&T, or an engineer can start to create it within the VSA program,” explained Zhang. It will flag over constrained conditions or invalid feature combinations. “VSA is CAD neutral, utilizing the lightweight JT CAD format, and can inherit geometry and PMI from NX, CATIA, and ProE. It can simulate large assemblies and support multi-CAD environments,” he said. Siemens also offers inspection programming of CMMs through its NX CMM software as well as Dimensional Planning and Validation (DPV), a closed-loop system for the real-time collection of measured quality data, according to the company.
While large companies with well-organized information systems—think of OEMs or Tier One suppliers—seem to be embracing MBD and intelligent CAD, Scott Green, director of software product management for 3D Systems (Rock Hill, SC), thinks future growth will come from smaller firms and job shops, who comprise the bulk of manufacturing activity. “MBD hasn’t quite caught on yet at the job-shop level; in auto and aero manufacturing you have this potentially huge economy of Tier Two and Tier Three suppliers,” he said. Geomagic Control X, his company’s software for metrology planning, includes many native CAD importers, including those which contain model-based GD&T definition.
One of the reasons he thinks this is true is that these smaller Tier Two and Tier Three shops are forced to deal with many different flavors of CAD. “More historically, the kind of predominant versions of MBD were attached to the major CAD packages, such as NX, CATIA, and CREO, each of which have their own nuances and downstream usability, because of their format as well as magnitude of seats in the field. And recently SolidWorks, with a huge installed base, enters the scene with MBD that’s available through third-party translators,” he said, noting that Control X recently added a SolidWorks translator. It is very hard for a smaller job shop to only accept work in a single native CAD format of their choosing, and because of that neutral CAD definition becomes the popular choice for design transfer, thereby historically losing all intelligence and nativity.
So, he sees the adoption of future neutral formats, such as STEP 3.0 in conjunction with AP242 for GD&T, as the next step. “Though GD&T itself is a standard, each CAD system has its own native file format description and MBD is written in that format. Because of this the success of MBD downstream then becomes the problem of the CAD translator. The cool thing for the future is having an open, neutral standard for MBD, that is a good step. That means more people will accept it,” he said.
Connect With Us
