The process industries were ahead of the curve in realizing the value of computerized data analytics, which is a subject of general business interest today. After all, statistical process control (SPC) applications were among the first on the production engineer’s desktop.
The much-heralded work of Edward Deming and the Quality Management Institute in industry was, among other things, the application of statistical analytics in production environments just beginning to make wide use of computers.
Today, it is big news that "big data" is working its way down from the enterprise-business level, hoping to selectively draw on production data in support of business decisions. It is a streamlined version of the "board room to plant floor" integration about which there was much talk. But that talk ended just before the Great Recession bit.
Now, most known big-data industrial applications seem to pertain more to the marketing side than the production side.
Not fade away
Nevertheless, SPC is alive and well. Desktop applications are available, and more significantly, the SPC footprint has evolved into an execution-layer application also known as "enterprise manufacturing intelligence" (EMI).
Across the process industries, a long history of corporate mergers and acquisitions means facilities that today are part of the same company have different legacy IT and automation systems in place, said Peter Guilfoyle, vice president of solutions, Northwest Analytics, at November’s Chem Show in New York.
Data collection and analysis is complex in itself, but comparing production facilities and product quality, identifying constraints, and developing best practices across multiple facilities can seem scarcely possible.
Dow does it
At Dow Chemical of Midland, Michigan, one of the world’s largest chemical companies, Guilfoyle points out that the integration challenge is addressed by enabling more general data access using EMI, but it is done without moving the data to some redundant, parallel data space. Databases remain in place, but it is easier to access data where it already resides.
This is because EMI allows direct data access from existing data sources. Then, by passing it through an analytics layer, critical process information becomes visual in real time.
"Dashboards are shared across departments," says Mary-Beth Seasholtz, Ph.D, a senior scientist at Dow, in an article published last year in Processing. "When behaviors are contrary to agreed-upon metrics, the decision becomes data-driven."
