It's called "GDS." It's General Motors' term for the rigorous final evaluation every car built in its assembly plants gets from independent teams of quality assurance people. The Global Delivery Survey found that the GM plant in Lansing, Michigan, the corporation's largest North American assembly plant (over 400,000 units per year) was doing an outstanding job in the production of the Pontiac Grand Am and Oldsmobile Alero: in their first year of build, they rated #1 and #3 overall, as compared with more than five dozen car and truck models produced by all of GM's plants in North America.
"It has never happened before for any car to rank #1 in its first production year," says plant manager Jim Zubkus. He goes on to point out that the Alero had reached the #3 spot after just six weeks in production.
How was this achieved? "There are a huge number of factors behind this success, beginning with the individual efforts of thousands of hourly and salaried people," Zubkus notes. "A lot of new methods, processes and technologies" employed by these people also had a role.
"We attribute a portion of the GDS audit success to the extremely good body fits on the Grand Am and the Alero," Zubkus says. One of the tools that facilitated this is a portable coordinate measuring machine (CMM) built by Romer Inc. (Carlsbad, CA). Zubkus explains, "The portable CMM allows us to qualify the tooling in the body weld-assembly systems better than ever before."
He adds, "For the first time, the Romer lets us measure the tools directly now rather than indirectly. Before, we verified the tools by carrying sample panels or subassemblies to the CMM room and measuring the product. Now we measure the tools right where they sit in the production system." That is, instead of bringing the parts to the measuring machine, the measuring machine is brought to the parts right on the line.
"It doesn't matter if the CMM room is just three feet away from the line," Zubkus continues. "By the time you get the measurements back it may be two days. And even if it's only two hours, dozens of cars have been built which may have a dimensional discrepancy. All of them have to be chased down, inspected, and fixed."
New Technology's Benefits. "Some of the auto industry's tool verification methods that are now being supplemented by portable CMMs have been in use 20 years or longer," says John Sturgis, a GM tooling expert at Lansing. These older methods include hard-checking fixtures and gages, and witness panels. Sturgis is a member of a process validation engineering group based at the GM Technical Center (Warren, MI). This group, and others like it, follow GM's new car programs, helping implement new technology. In effect, Sturgis's group is a traveling best-practices team for dimensional integrity.
In the case of hard-checking, the fixtures and gages are certified then used to verify the accuracy of the tools' settings. This tends to involve a bit of guesswork, which is unacceptable if every dimension is to be related directly to the math model in the CAD database. "The Romer takes us into the new millennium," Sturgis says. "Because we can measure the weld tool right on the floor, we know it is dimensionally correct. There is no more guesswork."
He adds, "Witness panels were even less reliable." Witness panels, if you're not familiar with them, are sets of sheet metal parts that are assembled into a check fixture and held in place while alignment holes are drilled. The holes are used to visually verify tool alignment and positioning when clamped up in the weld tool. "This is time consuming and error-prone," Sturgis says, adding, "And the amount of care taken with aligning and drilling could significantly affect the process. Witness panels are also very susceptible to differences in visual interpretation among operators and users. Often, several sets of witness panels had to be made, drilled and checked before we could be sure the tooling was correct."
It's not that witness panels are no longer used. According to Sturgis, "They are still used as aids to understanding how a tool operates, in identifying root causes, and in troubleshooting."
|Here Joe Croyle, a panel checker at GM Lansing Assembly, is measuring a body-in-white.|
But the implementation of the portable CMM is accelerating throughout the Lansing plant. There have been over 350 tools measured during a three-year period, a number that Sturgis says is "several times more than could have been done with the stationary CMM and check fixtures or witness panels."
There are six portable CMMs at Lansing. Some are Romer model 1028, which have 9-ft arms. The others are model 2500, which have 6-ft arms. The six machines have eight operators; more people are being trained. One reason why the plant needs lots of operators is because the CMMs can generally be used only when production is not running.
Getting CAD data into and out of the CMMs' control is simplified because the interface is a laptop PC running Microsoft Windows 95. CAD surfaces and vectored points are used for checking tooling. For checking the body-in-white, surface data is used. Either way, a straight-forward download provides the data from the math model.
Asked about the selection of the equipment, Sturgis responded that they looked at a variety of devices, including other portable CMMs, as well as computerized theodolites, digital photogrammetry, and laser tracking. Ticking off the reasons why the options weren't selected, Sturgis notes:
In Action. The Dimensional Integrity teams at GM Lansing use the portable CMMs primarily for location, position and orientation of the tooling to defined nominal dimensions. "They are also used to determine root causes of dimensional build problems by measuring the welded subassembly and the tooling, and developing dimensional correlations between them," Sturgis says.
Now, weld assembly tools are measured as soon as they arrive on the plant floor, before they are set in place, wired and plumbed. The CMMs are not only used to measure tooling and parts, but even car body interiors.
As Sturgis concludes, "Now there is little or no guesswork about tooling problems and build variations. There are fewer of those frustrating exchanges of opinion which can happen when no one really has good data."