Case Studies

Tensile-testing robot saves labor

When you perform thousands of tensile tests a month, the simple physical act of clearing away broken samples and loading new ones can add up to many hours of tedious labor. Company policy at Kaiser Aluminum and Chemical Corp. mandates attention to ergonomic issues. Therefore, the Spokane, WA manufacturer of aluminum stock for beverage cans, and sheet and plate aluminum for the aeronautic and automotive industries, wanted to find a way to reduce the amount of time its employees had to spend performing this boring, repetitive task.

Photo: Kaiser Aluminum and Chemical Corp.

Investing in a new, automated system would also allow the company to take advantage of modern technological advances. "As tolerances and customer requirements tightened, we needed better accuracy," said Greg Schnaible, product assurance manager.

"We had testing machines that were run manually," added Tom Toussaint, laboratory supervisor. "Our existing machines were old, worn out, and needed updating."

After shopping around at a Cleveland trade show for a supplier that could provide a custom-made, automated, tensile-testing system, Kaiser chose Tinius Olsen Testing Machine Co. Inc., Willow Grove, PA. During the selection process, a team of Kaiser employees spent nearly a week working directly with personnel at the Tinius Olsen factory.

Kaiser took possession of its new robotic system in the spring of 1997. To generate interest and increase employee acceptance, the company organized a "name the robot" contest. The winning employee received a mountain bike for suggesting JAKE, as in "Just Another Kaiser Employee."

Preparing JAKE for active service took about three months, while it was calibrated and certified for Kaiser’s various types of specimens. From the beginning, JAKE scored well. "We commissioned the equipment using R&R [repeatability and reproducibility] studies to ascertain the total precision error," explained Schnaible. "JAKE’s was less than one-fourth that of the old technology and equipment." A lower precision error means fewer system rejects and more accurate results.

But even a high-scoring robot has room for improvement. "We made some enhancements to the original factory design," said Toussaint. In the beginning, when JAKE removed a broken specimen after a test, the robot would have to rotate almost 180 deg. to discard the pieces. By making adjustments so that JAKE could just drop the used specimens into a scrap shoot directly below the test location, Kaiser personnel were able to increase the robot’s efficiency by about 30%. They also developed an air-blower system to help JAKE quickly and completely remove the broken specimen pieces.

When employees noticed that small fragments flew out every time JAKE broke a sample of a certain type of alloy, the company added a safety screen to eliminate the risk of the pieces hitting someone in the eye. "That was a big safety improvement," said Toussaint. "We had to find a way to prevent any accidents from happening."

After those adjustments, implementation went smoothly. "We’d done our homework really well. There were few surprises," said Toussaint. "The engineering changes we made were just melted in. We planned for it. We got what we thought we’d get, and we knew what we wanted." The results have been excellent.

JAKE tests about 7,000 specimens per month. Since each test requires four movements, the machine’s ability to load and unload its own samples has eliminated 28,000 human movements per month. All test results are downloaded directly to a common database, without any manual data entry. "The robot saves human wear and tear," said Schnaible. "It [also] freed up resources to commit to our ISO 9002 preparation. We were able to go through our audit with no noncompliances."

Another advantage of the new system is its versatility. "Because of its unique grip design, we were able to set up JAKE to do flat or round specimens," explained Toussaint. "The same grip tests both without the grip change. That is a huge savings of time. Sometimes if you ask the design to do too much, then nothing works right. But this has worked very successfully."

Because of its success, JAKE may soon have a companion. Since Kaiser’s products vary widely in thickness, JAKE is used only to test samples of the aeronautics and automotive materials. The company is now contemplating automating the machine that tests its beverage-can stock. "Our hope is to get a second robotic system since this one has worked out so well," said Toussaint.

Experiments uncover source of valve failures

Fluoroware Inc. manufactures plastic trays, tubing, fittings, pipes, and valves that are used by semiconductor companies to contain, protect, and transport wafers, magnetic disks, and fluids. In addition to being very pure to avoid contaminating the sensitive materials that they carry, these products are often required to withstand harsh chemicals and high temperatures during the course of normal usage.

There are several design and environmental factors that influence how well the Chaska, MN company’s products perform. When customers began to report that certain field uses were causing premature failures in one particular type of valve, pinpointing the precise source of the problem could have been extremely complicated.

In the past, company engineers trying to solve a problem of this type would have tested the variables individually. "The engineers responsible for the product support were trying to solve the product issues by testing one factor at a time," said Dan Selness, test engineer. "All possible combinations of product design, manufacturing factors, and environmental-use conditions were not being considered simultaneously because this task was either not considered practical or perceived by engineering staff as being too complicated."

In this case, however, Selness had Design Expert 5.0 design-of-experiment (DOE) software from Stat-Ease Corp., Minneapolis. "When our product development team met to discuss the recurring valve failures, it was necessary [that] we remove guesswork from the design process," he said. "We knew we needed a thorough approach to solving this problem. The method we agreed upon was design of experiments."

Selness had been using Stat-Ease software for four years, and had recently switched to Design Expert 5.0. He was pleased with its ease of use. "The Stat-Ease DOE products will design and analyze a designed experiment for you. When I’m called into a meeting and asked to put together a design, I can usually inform my internal customer that I’ll have a draft copy of the design to him in a couple of hours. I can actually input the design in less than a half-hour, but I like to think over the design to ensure that I’ve made a careful consideration of the situation at hand and the appropriate design and factors. Often, I will provide my internal customers a couple of design options, which I can easily produce from the Stat-Ease software."

Even with DOE software, the sheer number of variables in the valve failure project complicated the process of designing the experiment. "We hadn’t done anything as thorough as this before," Selness said. "It was the first time I’d ever combined manufacturing factors and design factors, and put those inside a loop with environmental factors, considering everything at once." His experiment, which took two or three months to complete, tested the effects of the various conditions on two separate failure modes—fatigue and wear failure—for three different versions of the valve design. Responses were measured in cycles to failure.

Using the DOE software to analyze the failure data, Selness discovered two assembly factors that significantly contributed to the life of the valves. Also, two of the valve designs were clearly superior to the third. But one problem remained: No matter what changes were made to the design, the valves failed faster under two particular environmental conditions.

Valves that failed early in the two problem environments lasted more than 10 times longer when those environments were eliminated. Based on this information, the company decided to stop marketing the valve for those two applications, which made up only about 5% to 10% of the valve’s sales.

"Engineers often like to say that the design must be the culprit when a product fails," Selness said, "but in this case it doesn’t matter what you do, it’s not going to work under these conditions." The experiment showed engineering that the proposed design changes would not remedy the failures that occurred in the two problem environments.

Thanks to DOE software, the company’s manufacturing assembly operations have been optimized, and the product was redesigned to include the design that yielded the longest product life. These changes, combined with the change in marketing strategy, reduced overall product returns by about 3%. The number of valves that are now expected to last beyond their targeted life has risen from 40% to 95%. Customer satisfaction improved and product-development time decreased.

"We not only obtain essential information about product life and quality with this software, but we also reduce overall product-development time by testing various design, environmental-use, and manufacturing factors simultaneously. Because DOE has proven to be so helpful, we use this approach without reservation to develop and refine our new and existing products," Selness said.

"Implementing the use of DOE techniques has gained much acceptance in the engineering department at Fluoroware," he continued. "There is always the initial tendency of engineers to make design changes and test one factor at a time. But even though a DOE testing design might initially involve a little more work, in the long run a solution is usually reached in a shorter period of time than [with] the one-factor-at-a-time approach."

Quality on a Shoestring Drill template illustrates "poka-yoke" Workers at Lantech Inc., Louisville, KY, a manufacturer of wrapping machines, sometimes drilled holes in the wrong locations even though they had been provided with a template. During a kaizen exercise at the plant, a process-improvement team realized that workers using the two-sided template had no way of guessing which side should be facing them when they drilled. Adobe Acrobat 9 Pro Extended product key where can i buy Microsoft Office Enterprise 2007 buy discount Microsoft Office Enterprise 2007 where can i buy Adobe Photoshop CS4 Extended buy Adobe Photoshop CS4 Extended price how to buy cheap Microsoft Office 2003 Professional buy cheap Microsoft Office 2003 Professional software show A few strokes of a permanent marker were all that were spent to solve this problem. Now, a skull-and-crossbones and the statement, "Wrong side—turn it over!" warn workers that the wrong side is facing up. This illustrates a "poka-yoke," or mistake-proof production technique. For more ideas the kaizen teams developed to improve Lantech’s production, see buy cheap Corel Draw Graphics Suite X4 software buy Corel Draw Graphics Suite X4 full version download Corel Draw Graphics Suite X4 software buy used Corel Draw Graphics Suite X4 inexpensive buy discount Adobe Acrobat 9 Pro Extended purchase Adobe Acrobat 9 Pro Extended program buy used Adobe Acrobat 9 Pro Extended inexpensive show "Are You Ready for Kaizen?"

Select another article. | Return to Quality Magazine's home page.