Automotive production has embraced metrology for more than a century because of its demand for high production rates and low tolerance for defects. Every major assembly or component plant has an expansive quality lab with coordinate measurement machines (CMMs), laser scanners, and material gages. Virtually all of that equipment connects to computer networks to feed data into tracking and management software. However, there are some spots in the quality inspection process where techniques are still old-school and subjective, words that make data-focused managers cringe.
“There are a lot of things being done today that are being done manually that could be done with highly automated systems,” says Robert Wasilesky, global key account manager for automated inspection at Zeiss Industrial Metrology.
Many of those manual processes show up in vehicle bodies – stamped panels used to create fenders, hoods, and other components and in welded car bodies heading to the paint shop.
Wasilesky says getting the human element out of inspecting vehicle bodies and assemblies can speed the process, eliminate differences in opinion, and feed critical data into systems that can monitor processes and improve production throughout time.
Minor, barely perceptible flaws in a bare-metal door panel can be problematic later in the manufacturing process. A 1.0mm drift from tolerance spec can make it difficult to weld the inner door supports to the exterior metal. A minor scratch that looks fine in steel can create a noticeable dimple after employees prime and paint the vehicle, leading to expensive rework late in the build process.
To combat this, many auto plants turn to white-glove treatment. While that terminology may bring up images of meticulous place settings at five-star restaurants, in stamping plants, dozens of employees wearing thin cotton gloves feel the surfaces of body panels with their fingertips, looking for flaws that their eyes can’t see. At Nissan’s Smyrna, Tennessee, car plant, quality engineers once challenged automotive journalists to don the gloves and feel for flaws on a test panel, marking each inconsistency with a magnet. Most of the journalists noted two or three flaws. Trained quality employees spotted 25 to 30.
Wasilesky says Nissan’s demonstration shows two big problems with white-glove inspection – it’s highly subjective, and it’s a very tough skill to teach to employees. Often, more-sensitive inspectors will find more flaws than their colleagues, so data are inconsistent and harder to track for process flaws.
“A lot of people with white gloves are effectively giving a pat-down of the automobile, a lot like the TSA at the airport,” Wasilesky says. “The Zeiss ABIS II sensor takes that process and automates it 100%. We take extremely quick images of surface. With two sensors on robots, we can do a full auto body in less than 60 seconds.”
The ABIS II optical sensor system can spot dents, dings, and cracks, identifying visual defects up to 0.5m wide and dents as small as 7µm deep. Many of those dings are too small to see or feel, but can mar paint jobs, creating ugly ripples in surfaces.
“We have cells that replace, on a high-volume line, 36 people on the line – 12 per shift. We have one operator monitoring equipment instead of 12 per shift,” Wasilesky says. “Data collected by the automated scanners are consistent, making it easier to track flaws through time.”
Flaws in body panels can come from dies going out of spec on the stamping press or workers grabbing the panels improperly to put them into inspection jigs. By digitizing and automating inspection, quality managers can track flaw trends – separating one-time handling flaws from progressively worsening die performance. The automated, optical inspection system has become popular in Europe with BMW and Mercedes-Benz showing off their uses of the system, and Wasilesky says interest is growing in North America, adding that one manufacturer is testing an automated cell that can identify flaws and repair them before sending them on to the next stage of the build process.
Body in white
In a perfect world, every panel sent to the body shop from stamping would be perfect and identical to every other panel. In the real world, 100% repeatability isn’t practical. Dies used in stamping presses, much like cutting tools or weld heads, break down with repeated use. Engineers set a tolerance spec, and stamping plants work from one end of the acceptable range to the other before replacing or repairing dies. So, in-spec parts can vary in size and shape, making alignment and fit difficult when welding parts into a car body.
Wasilesky says planning for that variability starts with good metrology from the stamping process, but it continues in body-in-white assembly. Scanning equipment can measure alignment issues to create offsets for slightly misaligned parts – allowing a great weld for slightly mismatched parts.
“We can move a robotic sensor to anywhere you want on an automobile, take a shot, do analysis, and move on to the next position in one second or less,” Wasilesky says. “Holes, threaded holes, fasteners, self-piercing rivets – these have all become standard in the toolbox, so you need to measure alignment for many different joining systems.”
Zeiss’ AIMax cloud digital, optical, 3D-sensing system can take 70mm x 70mm scans of flat or curved surfaces in less than a second. Placed on a robot, it can inspect hem edges, rivets, surfaces points, and T pins – 3D shapes that have traditionally been tough for automated scanners to capture.
Wasilesky says increasing computer power and advanced sensor systems have dramatically improved the speed of data collection – allowing AIMax cloud to create a point cloud of 3D data quickly enough to be used in-line at auto plants. Traditionally, manufacturers have sent samples of pre- and post-fastened components to the quality lab for detailed CMM inspection and destructive testing. With automated systems, AIMax cloud can test all products 100% instead of samples, potentially improving overall quality.
“We can see everything about a closure’s edge. We can quantify all of the characteristics – radius, form, location, whether there’s bleed-through (ripple) – things that were very hard to do with sensors in the past,” Wasilesky says. “With a 60-second measure time, that’s enough to measure most critical features of a car body. A four-robot cell can take hundreds of location scans, and each location could include three or four features that we’re analyzing. So, it can measure hundreds of features per vehicle in standard cycle times.”
ABIS II sensors and AIMax cloud systems feed data into manufacturing executive systems (MES), data collection and analysis software such as Zeiss’ PiWeb Suite, and product lifecycle management (PLM) software, allowing engineers to track quality performance and optimize manufacturing. Automakers are leading users of such systems, so any technology that improves the quality of data entering the network is welcome.
“If you know where the problem is, you can fix it, so it’s important not just to produce data but to make it usable,” Wasilesky says. “When components start to drift within tightly held specs, you can adjust tool and robotic offsets to adjust manufacturing without slowing the line or sacrificing quality. We have capabilities that didn’t exist five years ago. We can sense robot movement down to an accuracy of 50µm. With this feedback, we can control fine robot moves and automation processes. So, you can use robot guidance, based on the metrology results.”
While such metrology systems can be retrofit into existing assembly lines, typically big investments in in-line metrology come with new vehicles when automakers completely retool their factories. So, full adoption industry adoption of in-line inspection and enhanced body-panel scanning could take years, but Wasilesky says interest is growing rapidly as automakers look to boost quality and productivity.
“There are still lots of gains in efficiency to be had by automating more processes in manufacturing, let’s close that process loop,” Wasilesky says.