The four things that must be known

Based on a survey of its members, the National Association of Manufacturers says U.S. production output will increase measurably in 2007. Research groups see increasing investment in information technology by manufacturers in 2007 as well.

For these and other reasons, a range of technology providers—from major automation vendors to small, privately owned software developers—hope to collect on bets they've placed that manufacturers will respond soon to the challenges of globalization by augmenting their capabilities for manufacturing intelligence.

A number of manufacturers—including Ballard Power Systems, Nucor, and Chemtura Corp.—say they're already seeing benefits from applying more real-time intelligence to their operations.

For anyone not immersed in the ongoing dialogue concerning the relationship between globalization and information technology, the logical first questions are: what is manufacturing intelligence, and what does it have to do with today's most pressing production and supply chain challenges?

In recent years, several small software vendors—including the former Lighthammer, the former IndX, and Activplant—introduced solutions that used Internet technologies to gather data from disparate plant-floor systems. Aggregation and analysis of the resulting information, delivered in near-real time as key performance indicators (KPI), allows for quick reactions in dynamic environments, including changes in demand coming down from ERP systems.

In recent years, automation and ERP vendors have taken notice of this emerging solution category. Lighthammer was bought by SAP, and IndX by Siemens, for example. Through internal development or similar acquisitions, other vendors soon offered their own manufacturing intelligence solutions—many focused on a particular mode of manufacturing or industry.

A more expansive view of manufacturing intelligence includes other means to the same end, with data integration achieved by plant operations software suites that integrate with third-party systems in a manner similar to solutions from more narrowly focused manufacturing intelligence vendors.

Manufacturing intelligence could be defined as the application of analytics to data integrated from diverse sources in support of collaborative, real-time decision-making. It also implies integration of the plant floor with the enterprise, and the blending of financial and operational metrics to ensure the enterprise is focused on a common goal rather than optimization of one operational silo or another.

The tendency of manufacturing intelligence to make production more responsive to the true needs of the supply chain leads those most bullish on IT use in manufacturing to see it as a tool for optimized, demand-based manufacturing, potentially lowering inventories, increasing customer responsiveness, and fostering competitiveness in global markets.

"Our 2006 IT spending survey reveals investments in manufacturing operations capturing the top spot—ahead of ERP—in the coming year," says Colin Masson, a director with Boston-based AMR Research.

Regarding manufacturing intelligence, "Step one is to understand baseline performance so as to prioritize investments to fill gaps in your enterprise system," says Masson. "You also need dashboards and metrics that enable you to drive changes in behavior."

Changing behavior was important for Ballard Power Systems, a Vancouver-based maker of proton-exchange fuel cells. "As software spread throughout the company, we got more requests from various divisions to access the data," says William Blakeman, senior engineer and team leader for Ballard's Integrated Plant Data Systems (IPDS). "It was a nightmare with everybody trying to hit all the different databases on different servers with their own homegrown queries. It became uncontrollable."

As a supplier to automotive, heating, and other commercial and industrial applications, "We're primarily a research company, so we handle enormous amounts of data," adds Blakeman.

To resolve the tangle of multiple servers and databases, and proliferating ad hoc queries, Ballard implemented Incuity EMI manufacturing intelligence software to wrapper and virtualize all data sources and stores, making them accessible through a single portal using service-oriented architecture (SOA). Manufacturing intelligence vendors, besides tending to have strengths suited to particular vertical industries, also take different technology approaches (see Different approaches taken by intelligence vendors).

"The huge advantage of Incuity is the ability to combine data from different sources in a single report," Blakeman says. In one use, "The researchers see test results when a fuel cell stack was assembled and everything passed. They then compare that against data coming back from a field trial to see the differences."

The company is adding enterprise data to the manufacturing intelligence mix to aid financial reporting.

"Incuity is emerging as the place to find data in the entire company—no matter what it is," says Blakeman. Eventually, he adds, controlled access via Web services and the Incuity portal may be extended to customers as well.

The need for better manufacturing intelligence is a key driver in the emergence of production and performance management suites that have their own common SOA-based infrastructure, giving manufacturers opportunity to manage data comprehensively while rolling out applications incrementally. Further, this approach is well-suited to multiplant engagements common in an age of globalization.

Rockwell Software, Invensys Wonderware, and Invensys Process Systems are among the major automation vendors to introduce production and performance suites.

Announcements at the recent Rockwell Automation Fair for the FactoryTalk production and performance management suite included a freshly inked partnership with data historian vendor OSIsoft; the availability of ProductionCentre, which leverages Rockwell's Datasweep MES acquisition for visibility, traceability, and quality; and availability of FactoryTalk Integrator for enterprise connectivity, and FactoryTalk Portal for a single enterprise portal adapted to manufacturing—both incorporating IBM WebSphere middleware technology. FactoryTalk AssetCentre was on display on the show floor at the Fair, but won't be available until next spring.

Invensys Wonderware and Invensys Process Systems both made use of the Invensys ArchestrA architecture in the development of the Wonderware Production and Performance Suite, and the Invensys InFusion Enterprise Control System. In general, InFusion is better suited for continuous- and batch-process environments involving DCS-based control, while Wonderware Software Solutions are best suited for discrete manufacturing and batch operations that use PLCs and supervisory control systems.

Wonderware says 14 brand-name manufacturers in the consumer goods, aerospace, electronics, and other industries are engaged in global rollouts of Wonderware Software Solutions—involving more than 300 production plants, with individual company rollouts encompassing as many as 50 plants.

At the recent Invensys Process Systems user conference, Peter Martin, VP of strategic ventures, said the company has more than $240 million in InFusion-related business in the sales pipeline following its introduction last April.

"Almost every plant has different vendors' systems and equipment—even different vintages of the same systems," says Martin. "In the past, none tended to work together."

This situation must be addressed, says Martin, because of the need to maximize the economic value of industrial assets and the volatile nature of supply and demand in the global economy. InFusion provides a "common space for collaborative decision-making based on integrated systems."

John Snodgrass is a process control manager with Chemtura Corp., a maker of specialty chemicals for the rubber, plastics, and other industries. Its Morgantown, West Va.-based facility makes phosphate-based stabilizers used in plastics; and runs batch, continuous, and packaging production processes controlled by a sitewide Foxboro DCS, batch management systems, PLCs, and manual controls.

Snodgrass says the problem with the previous process at the plant was that humans were at the center of the system architecture.

"People were hoarding information, using private systems to do their jobs and then entering the results into the SAP ERP system. Operators entered production data manually. Data flowed out of the DCS, but very little information came back."

Data integration achieved by means of InFusion, "takes people out of the center of the architecture, resulting in improved customer service and budgeting and forecasting data at the plant—all leading to better supply chain management," says Snodgrass.

Quantitative results at Chemtura include a 5-percent to 10-percent manpower reduction based on better process order execution; a 1-percent reduction in operating costs; and an 80-percent reduction in ERP account requirements since operators no longer log into SAP.

Martin says the ultimate goal of an enterprise control system is the application of something much like closed-loop control to an entire manufacturing enterprise. "Finance works in monthly increments, but operations are managed in real time, on a daily basis, and in longer horizons as well," he says.

"Closed-loop control concepts can be applied to financial information," says Martin, "and Invensys Process Systems has performed this exercise in its work with more than 50 companies. But having a balance sheet for each production unit in real time isn't necessarily helpful. We've noticed that it's in the most measured plants that dashboards and portals seem to be ignored the most. For people to make use of the results, you have to get it down to about four visual fields."

This idea provides a true picture of a plant's operations, and programs can launch on that basis, says Martin. "Why shouldn't plants today be run on the same sound statistical basis that is today found in the sports world?"