EMI Connects the Dots of Industrial Information

Enterprise Manufacturing Intelligence (EMI) refers to fast, meaningful data sharing between shop floor and business office.

Aw 1863 1005 Team
It refers to the “actionable intelligence” necessary to achieve both continuous process improvement and demand‑driven production. While EMI is implemented as one or more software applications, it is best considered as a system of capabilities that will assist decision making across the enterprise.

“The best EMI systems support decisions both up and down—in executive offices and at shop-floor workstations,” says Dan Miklovic, research vice president for Gartner Research Inc. (www.gartner.com), in Stamford, Conn. By making the shop floor visible to management, EMI informs business strategy; by bringing business context straight to the operator, EMI puts production decisions in line with business objectives.

EMI architectures

EMI performs five core functions: it aggregates, contextualizes, analyzes, visualizes and propagates. All EMI functions act on a mix of data concerning manufacturing processes, the status of individual products and business factors. After applying its analytics to an aggregated data set, EMI output typically consists of representative data samples for graphing and summary metrics called KPIs (Key Performance Indicators). EMI output is contextualized, with each piece of output answering a question for one of three primary targets: operators, operations supervisors or business managers. Output propagates only to its target and is then displayed for quick visual evaluation, often in the form of dashboards and charts.

Architecting an EMI system begins by defining what you want out of it. “To get a high return on EMI investment,” says Miklovic, “the enterprise must have some maturity to map its manufacturing processes to its products and then onto its business strategies.” Choosing the right metrics for each target audience is critical. Aligning EMI with existing process improvement initiatives is almost as important.

With respect to performance, the two primary selection criteria for an EMI product are the data sources from which it can aggregate and the time it takes to deliver output—its latency. Manual entry may be an important EMI data source, and some environments require EMI to read directly from devices. Most EMI data, however, is aggregated from databases: specialized “data historian” databases containing physical process measurements, relational databases that store scheduling and product-tracking information, and relational databases that hold enterprise business data.

Geographically distant operations will add a degree of difficulty. “When the enterprise manufactures in Asia, assembles in Eastern Europe, and distributes in North America, data aggregation with current products is not likely to be fully satisfactory,” says Raj Rao, vice president for enterprise application services for information technology services supplier Keane Inc. (www.keane.com), in Boston. “This is an area in which EMI is still growing.”

EMI output should be timely relative to the decision that the output supports. EMI system latency varies enormously, from less than seconds to weeks. Case-by-case assessment is unavoidable, but an overall EMI system effectiveness threshold seems to be output available within one shift or less of the triggering event.

Perhaps the biggest problem with EMI is the EMI marketplace. The ready availability of high-capacity, commercial, off‑the‑shelf network components and the wide adoption by industry of standard communications protocols have removed the principal obstacles to acquiring a capability that has long been desired. Vendors are responding by offering solutions from all points of the compass. It’s a fluid market in which pure-play EMI products are being absorbed into larger control system and enterprise resource planning (ERP) packages quite rapidly.

Done right, guided by some homework up front, and mindful that every system selection involves trade‑offs, EMI can be implemented with little disruption to current operations while adding significant value at modest cost.

Marty Weil, martyweil@charter.net, is an Automation World Contributing Writer.

Gartner Research Inc.
www.gartner.com

Keane Inc.
www.keane.com

Subscribe to Automation World's RSS Feeds for Columns & Departments

More in Control