Statistical quality control is coming out of the laboratory and assuming a bold position on the production line. Not content to simply identify defects in finished products, quality departments are moving to the front lines of production to effect continuous improvement.

Instead of inspecting and throwing out defective products, companies find it more productive and cost-effective to control the process. Software such as Quality Analyst from Northwest Analytical Inc. (NWA), Portland, Ore, automates the extensive number-crunching used to ensure quality. It speeds up the calculation time for statistical analyses, improves accuracy, and rapidly creates usable charts and graphic displays.

Statistical process control software allows operators to avoid undesirable variations in online production. Statistical quality control software, on the other hand, interacts with various databases off line to plot long-term production trends. It helps determine if production is stable and predictable, if yields can be increased economically, if production costs could be reduced, or if maintenance is excessive.

"We make two or three different blends of lubricant additives daily," says Jim Moore, senior technical services advisor, Ethyl Corp., Houston. "Each batch is a different formulation that is shipped to manufacturers of passenger car motor oils such as Texaco, Pennzoil, and Quaker State. They mix our heavy concentrates with other motor oils at their plant. Our products constitute 10-12% of the finished product. Because the ingredients and volumes vary widely (5,000-120,000 gallons each), we rely on Quality Analyst statistical quality control charts from NWA. We use Shewhart charts, run charts, histograms, regression, and correlation to determine such things as cost per pound and provide a timely, high-quality product for our base of more than 50 customers."

At the Tosco Refinery in Arroyo Grande, Calif., Quality Analyst is used in a continuous process application, according to Gary Davis, advising engineer. "Everything we produce travels by pipeline to our plants in Rodeo and Martinez in the Bay area," says Davis. "We continuously make about a dozen different products. This means our systems must be highly balanced and accurate to meet refinery accounting and operations requirements. Data is gathered in intervals of anywhere from five seconds to five minutes. This is needed to check meter accuracy, running averages, product quality, and yields. In five years of use, Quality Analyst has helped us maintain our high standards with almost no need for technical support."

"We don't even make a product," says Mike Connolly, quality assurance manager, Casey Products Co., Lisle, Ill. "Casey is a fastener distributor, but we depend on Quality Analyst software to check variable-gauge parts, dimensional control, and metallurgy of the fasteners we ship to our major customer, Caterpillar Inc. As a small-volume distributor, we can use the software to shop the world markets and compare not only suppliers, but individual lots. Instead of just running ring gauges, we can now conduct tensile tests, Rockwell hardness tests, plating thickness tests, and instrument calibrations. As a result, our customer reject rate has dropped from about 5% to less than 0.1% in just several years. And profits have risen substantially."

As a manufacturer of pharmaceutical and industrial grade-chemicals, ISP Technologies, Texas City, Texas, depends heavily on statistical quality control software to chart the operation of its three reactors. "We also use Quality Analyst to test the quality of common incoming substances such as methanol and formaldehyde," says Simon Lin, control chemist. "With more than 100 customers, each with different requirements, we take great care in plotting standard deviations, range charts, and process capability. Using information from our Laboratory Information Management (LIM) database, a detailed certificate of analysis is prepared and given to each customer to certify the quality of product received."

For Mark Giaudrone, total quality improvement manager, Huntsman Packaging, Macalester, Okla., the ability to build in software limitations and self-checks is one of the most important aspects of QA. "We make more than 200 different thin-gauge plastic products for food and medical applications. They vary from 0.5-2.5 mil component thickness, different shades, different compositions and strengths. QA enables us to filter through the information and select specific data, and then permits us to build in self-checking routines to question possible number transpositions, latest chemical formulations, and quality parameters. This allows us to prepare certificates of analysis for our customers with complete confidence, and at considerably less cost than other comparable software."

back to top