Data Stamps Out Defects
The stamping presses stand in a row, 10 of them, each two or more stories tall and a half a football field long. They pump out parts, massive amounts of them every hour, 23 hours a day. On an LED, the number of parts produced click by and the raw, dull-silver components, roll out from the presses’ jaws and are offloaded by robot onto conveyor belts where operators load them up onto holding fixtures.
Inspectors, four of them in all, pull the first and last of every run of every part that is produced at ThyssenKrupp Budd Co.’s Detroit plant. In between, they randomly inspect parts at least every hour, and sometimes more, depending on the part, and collect reams of data, mountains of information and tons of clues that can help solve existing or potential problems.
But until recently, each notation on the inspectors’ paper concern form was a sheet unto itself. This island of data reveals nothing more than the status of that particular part at that particular time. Each form was kept, properly stowed away in a cabinet, but the data contained within those sheets for the most part was never seen again. But, things have changed.
Today, inspectors at this facility, considered one of the world’s largest stamping operations, use portable data collectors that they carry from part to part, press line to press line. Three handheld, portable pen tablet units are in place on the factory for use as parts come off the line. ThyssenKrupp Budd integrated the lightweight, portable units because of their versatility; they did not want to put a PC on the shop floor that would have required the inspector to go to it to enter each new set of data. This is important considering the amount of production that goes on at this facility. ThyssenKrupp Budd makes automotive components including full-body underframes, car roofs, doors and sidepanels. More than 400 different parts are produced at this plant alone.
“Our inspectors carry the portable computer around because they have more than one area to watch and it makes it a lot easier for the inspector to do his checks,” says Joseph Locricchio, the continuous improvement supervisor who set up the data collection program at ThyssenKrupp Budd as part of his Six Sigma Black Belt project. “Instead of doing his checks and going to a personal computer and inputting the data, now he can walk with the unit.”
The unit is a completely portable pen table computer that weighs less than 4 pounds, has an 8-hour battery life and is constructed to withstand drops and environmental factors. It operates on the Windows CE operating system and collects real-time data using an image-based, touch-screen unit. The digital assistant is an image-based computer that uses customized screens. At ThyssenKrupp Budd, the paper concern sheets have been digitized, and by using the pen to touch the screen, the location of the defects is automatically entered. The portable units also have radio frequency (RF) capability and the data is sent real time through RF to the database. “We are able to know everything that goes into a database,” says Locricchio. “We set the program up so that it basically is the same as the concern sheet.”
Inspectors such as Bill Paluda use it. Paluda’s years of experience helps him pick out a defect that is almost “unseeable” to the untrained eye, but, down the line, would create painting or assembly problems. He pulls a part and “highlights” it by rubbing the component with a specially treated rag that will give the silver-colored part a shine when it is exposed to the bright lights of the inspection area and make defects visible—visible, at least, to the trained inspectors.
“We highlight it to bring out defects, to see if we can find them,” says Paluda. “With the angle of the light and the training of the eyes, you can pick it up. The average person wouldn’t pick it up, but as you become more familiar with highlighting and look at it from a different angle, your eyes become successful at picking the defects up.”
On one sidepanel, Paluda discovers a spot of dirt, for the most part unseen, and marks it with something akin to a grease pencil. He then uses the pen tablet to mark up the defect, its location, and the severity of the defect and from what shift and press the product was stamped.
Some typical problems that Paluda and the other inspectors see include burrs, slivers and dirt. The inspectors will also look for missing cutouts, hole locations and other parameters.
Depending on the severity of the defect, the part will either go through a touch-up process or be discarded. The inspectors rank the defects by severity giving them a range from 0.1 to 1.0 with 1.0 being a major defect that would require the part to be scrapped. Defects classified as less than a 0.5 are considered minor and anything between 0.5 and 0.9 requires some sort of repair. “We have it set up so that if an inspector rates a defect at 0.5 or above, the software will send out e-mails,” says Locricchio. The e-mails will go to the quality manager, press-shop manager and assembly manager.
The digital sheets feature a line drawing of the part, and using a pen, the inspector will touch the screen in the areas of the defect and that act will leave a mark at the point it was touched. The severity of the ranking will cause the dot to change colors, from a green for a minor repair to a hot-red major repair.
The data collected helped ThyssenKrupp Budd improve several processes. Defects can be sorted by severity, parts, dates, shifts, type of problems and other parameters. On one particular Friday, for instance, three sliver defects were discovered, on another day dirt was found on some parts. An often-used sort is by severity of defect, which allows the company to look at the most important problems. “We will sort by repair and look at defects that are 0.5 and higher. Even though the ‘incidentals’ (minor problems) are still important to us, we want to focus on the top problems,” Locricchio says. “We want to work on those problems first, and after we get them solved, we will work on the others.”
Every piece of data that the inspector enters into the portable unit will show up on concern reports or can be used for statistical analysis. At his desk computer, Locricchio can click on a problem area in a report and be able to analyze that data. He can then run statistical analysis such as Pareto charts, trend charts, X-Bar, R-charts or U-charts, to track down problem areas.
The analysis is then used during daily and weekly meetings in what Locricchio calls the war room. By printing out the color-coded concern sheets, the quality and manufacturing representatives at the meeting can visually see the concern areas and data can be compiled to determine if problem areas are improving.
“What we are able to do is use the data to lead us to the problem and from there we can start to improve the process,” says Locricchio. “Some of our roofs had problems with slickers, and by using the software, we were able to find out the areas that they were occurring, and we were able to make adjustments to the dies. We also know the rack number, so that if we ever find something severe, we are able to contain it.”
Slivers, or small metal chips on the parts, were an area of concern for roofs for a popular automobile. The slivers were in the rear and along the edge of the roof, and because the part was so important, the company had a repair area dedicated to these roofs. By determining the location of the problem, they were able to determine its root cause. Two months ago, the workers called the area the mountain because of its stacks of defect products. Today, that mountain is more of a molehill.
“What we found out was that the first press was actually grinding part of the blank and some of the scrap would fall out and the chips would get in there and cause a sliver,” says Locricchio. “We knew we were getting the slivers, but we didn’t know why. Using the software we were able to say, ‘we are getting slivers in this area,’ and tooling was able to go out to the press and make adjustments to the press to solve the problem.”
Dirt, which can be anything from actual dirt and dust to oil drips and metal shavings, is a common problem at an older plant such as ThyssenKrupp Budd’s Detroit facility and one problem that has been the focus of much statistical analysis. “One of the things that we found was that on one of the lines we were getting a lot of dirt at the front part of the roof. Having the data to back it up, we were able to get a washer for that line,” says Locricchio.
The washer uses two rollers and a filter to remove any slivers and any contaminants from the blank before it goes through the process. “Since we installed the washer, we have seen a significant drop in dirt,” adds Locricchio.
While dirt is typically the number one problem at the plant, for the past six months, waviness had taken over the top spot. “I know where the waviness is occurring, and now we have to determine why it is occurring,” he says.
This new problem is an example of how the system can allow the quality group to tackle new problems. With the success of the program, ThyssenKrupp Budd is looking to expand into other areas of the Detroit plant. This expandability is one of the strengths of the system. If a new part is designed and is to be manufactured, it is simply a matter of generating a new form. If another area of the plant needs the benefits of data collection, one of the portable units can be walked over to that area.
The next deployment area for the portable units may be in the subassembly area where inspectors look for surface problems and also fit- and function-type problems. In the past, a portable unit was diverted from other areas for use in that area, but Locricchio wants to have a dedicated unit for that department. “We want to roll out another portable unit in the body sides area,” he says. “When I was doing a project regarding weld pin holes, we found out that the way we were programming the steppers to keep the weld tips from wearing wasn’t optimized. I hope that we can get a unit back there so that we can start quantifying where a lot of our assembly problems are and start to narrow some of those problems down. As of right now, we are taking the data but not doing anything with it.
The use of the portable data collection devices is starting to get some momentum in terms of showcasing process improvements. Others in the company are beginning to realize what can be done with the portable data collection devices and the software.
“Instead of intuitively knowing where problems are, we have data. Now we know,” says Locricchio. “Before we didn’t know, we thought, ‘I think this is what it is, probably what it is.’ I think this is what it is, probably what it is. Now we can say for certain. ‘This is where we are seeing our waviness, our buckling, our dirt.’ It is very powerful.”