THE BUSINESS OF VISCOSITY BLOG



    Five Reasons SPC is Nearly Impossible for Liner Operation

    Posted by Mike Bonner

    May 11, 2016 1:00:00 PM

    Statistical process control is difficult in liner operations.

    In our recent blogs, Can Liner Weight Variations Impact Quality, Delivery, and Cost Structure, and Variation in Liner Placement Impacts Delivery, Quality, and Cost, too!, we identified weight and placement as the two primary criteria for liner effectiveness.

    Though a high-throughput, repetitive process, which would seem to make it an ideal candidate for Statistical Process Control (SPC), the reality of these criteria makes it almost impossible to manage using these methods.

    Placement is Easy

    With the advent of high-speed vision systems, liner placement can be evaluated on every part in real time. These are so fast that when a defect is detected, there is still time to eject the end from the sequence – even at production speeds in excess of five ends per second! This has virtually eliminated placement as a source of QC holds.

    Not So Weight…

    For all of its importance, liner weight is an extremely difficult metric to manage – and it turns out that SPC is no help whatsoever! 

    The reason? 

    The measurement process.

    Unlike machined parts, for which critical features can be identified and hard gauging created to allow inspection right at the machine, producing immediate results, liner weight is much more difficult to measure.

    But, how can weight be hard to measure?

    Weigh – Add – Weigh

    Like most coating weight specifications, liner weight is based on two measurements:  the weight without compound, and the weight with compound. But unlike coatings, the measurements are separated.

    Coating weights are measured using a process called weigh – strip – weigh. A sample of known size (often a 4 cm2 circle) is cut from the metal being coated. That sample is weighed, then the coating is stripped from the metal substrate, which is then weighed again. The difference is the weight of the coating – usually expressed in mg/cm2.

    Liner, on the other hand, cannot be evaluated in this way.

    Why?

    Because liner is specified in mg/end. 

    The weighing process is exactly the opposite. The bare end is marked so it can be identified, and then weighed.  Then it is inserted into the process and the compound applied. The markings cause the ends to be rejected by the vision system and they are collected and taken to the lab to be cured and weighed again. As with coatings, the difference between the two is the compound weight.

    So what’s the difference?

    Time is the Enemy

    In between measurements, there are a host of things that can influence the outcome of this test.  First, it turns out that the sensitive balances required are influenced by magnetism. To make matters worse, steel ends are transported through the process on magnetic conveyors.  As a result, the ends have to be demagnetized before each weighing step – an imprecise process. That’s one.

    Next is the curing process. The weight is influenced by the amount of moisture in the compound. The cure time and temperature is extremely important, but is also influenced by the mass of the end, so different ends cure at different rates – this introduces variations between ends, operators, etc. That’s two.

    Generally, the weighing process is performed in a remote location – like a QC lab – due to equipment required. Those sensitive balances require a controlled ambient environment – not an option on the factory floor. That’s three.

    Too Late to be Effective

    In the end, there is too much handling, and too much variation, and too much time for this to be a repeatable process. It is not uncommon for these test to take more than an hour to complete – and by the time they are done, literally thousands of parts have been run – all of which may be bad – if you trust the measurements. That’s four.

    In addition to potentially creating large volumes of scrap, this long measurement cycle also means that the sample size is too small to be statistically significant. So, instead of being “control," it is more like a “spot check." The resulting Cpk’s are both inaccurate and irrelevant. QC Holds are inevitable. And that’s five!

    So why do we bother?

    The answer is a real-time, in-process control strategy that can make these sparse data points relevant, but that’s the subject for another day…

    Download our free Process Excellence Diagnostic to more closely evaluate the operation costs, maintenance, overall performance, and product quality of your liner process.

    CEC_process_excellence_diagnostic

    Topics: Manufacturing

    Saint Clair Systems Blog

    We post our thoughts on temperature control, technology, viscosity and industry regularly. Subscribe to get updates on our business and our philosophy.

    Subscribe to Email Updates

    Recent Posts