Stabilize quality with process data analytics

Traditionally a quality leader, Japan is being challenged by China and Taiwan. “They’re catching up, increasing their quality, pushing them to their limits,” said Bradley Ford, global business development manager, Yokogawa, at the 2016 Yokogawa Users Group Conference this week in Orlando. Meanwhile, in Japan, “Quality problems are happening and they don’t know why. It’s true that equipment is aging, and they’re running it harder to meet higher demands. But there’s also a problem we call the ‘silent change’”

The Japanese manufacturing industry is traditionally close-knit, with long-established vendor/supplier relationships and highly evolved processes. “Then the earthquake happened, and everything was broken,” Ford said. “Manufacturers had to find new vendors, to go out on the market to other Japanese and foreign suppliers.”

The vendors are meeting the incoming quality specifications, but problems are showing up downstream and in products. “They don’t know why. Sometimes the product is on spec, sometimes it’s not. On working with them, we often find the feedstock was changing. It met the specification, but gave different results. A supplier three levels out was changing something, and it was having a significant impact.”

Manufacturers can’t just change suppliers until a problem stays away, and waiting until the product is tested is too late. “How can we see the problem before it affects the process?” Ford asked. “We did this as a consulting service, and found the existing technology wasn’t good enough. We had false detections and events not occurring as they should. And it took a long time to do the analysis, typically days. Production runs were finished and lost.

“Now we’ve established a comprehensive methodology—advanced analytic software that allows operators to see differences early and take charge, using the knowledge of skilled data analysts and engineers from around the world.”

Solution for stabilization

Yokogawa began offering a process data analytical service to its customers in 2008. More than 100 service contracts have been concluded with Japanese chemical and other industry companies, which have come to rely on this service. Based on the insights that Yokogawa engineers gained by providing this service to their customers, the company developed an analytical tool to improve its efficacy and thereby help its customers maintain and improve product quality. This software makes use of the Mahalanobis Taguchi (MT) method, a pattern-recognition technique employed in multivariate analysis.

[sidebar id =1]Called Process Data Analytics, the application can detect a decline in quality or productivity at an early stage of the manufacturing process by analyzing process data, facility status information, operation history, and other data. Used in combination with the company’s analytical services, the software can help manufacturers stabilize and continuously improve the quality of their products.

Process Data Analytics will run on Windows PCs and analyze production operations using temperature, pressure, flow rate, liquid level and other process data, as well as data on facility operations and equipment maintenance collected by a plant information management system (PIMS), DCS or PLC. While data from such systems must normally be converted to CSV format for use in another program, data from Yokogawa's Exaquantum plant information management system can be used as-is, without file conversion.

The software uses statistical methods to compare the collected data and accurately detect deviations from normal conditions. Any deviation will trigger a warning that quality may have deteriorated. By using the "four M" criteria of material, method, machine and manpower to analyze process data, it can visualize changes in production processes and thereby improve operations at manufacturing sites. Key benefits of the software cited include:

1. Early detection of abnormalities in production processes - By detecting changes in production process data, the software can spot quality and productivity issues at an early stage of the manufacturing process. Based on this information, measures can then be taken to bring production operations back to a normal condition and recover quality.
2. Fail-proof quality inspection - By detecting changes in the data from production processes, the software can detect any sign of deteriorating quality and thereby catch any fault that might be overlooked in a conventional pre-shipment inspection. This can help quality assurance departments improve their quality inspection process.
3. Extensible via integration with MATLAB - The software supports MATLAB, the widely used numerical analysis tool from MathWorks. Custom MATLAB calculations can be integrated within Process Data Analytics to leverage unique business and domain knowledge.
4. High speed and accuracy through use of AngleTry Associates' proprietary technology - By using a pattern-recognition technology licensed from AngleTry Associates, the software delivers quick and accurate analyses. This technology is particularly useful with consulting and systems construction.

The software is used for production quality control in the oil, petrochemical, chemical, pulp and paper, iron and steel, pharmaceutical, food, automobile, glass, rubber, electrical equipment/electronics and other industries, and will be released in March, 2017.

Click here for more information on the Process Data Analytics offering.

[sidebar id =2]