Practical intelligence
Manufacturing intelligence boils down to greater agility for Celanese Acetate, other manufacturers
By Malcolm Wheatley, contributing editor -- Manufacturing Business Technology, 6/1/2002 6:00:00 AM
Back in April 1999, when Schmalbach-Lubeca opened its manufacturing facility in Blythwood, S.C., "the plant was conceived pretty much as a lights-out operation," says systems engineer Thomas Iwancio. The latest of 62 such plants operated by the company in 22 countries around the world—16 of them in the U.S.—the heavily automated factory contains 11 injection lines producing pre-forms, which are then fed into nine blow-molding lines producing plastic bottles and other food & beverage containers.
According to Iwancio, with such a high degree of automation, it can be hard for production managers and shift supervisors in the front office to know exactly what is going on. Consequently, in November 2001, the plant leveraged its existing investment in industrial automation software, from Lake Forest, Calif.-based Wonderware Corp., by implementing Wonderware's SuiteVoyager software, a manufacturing information portal that abstracts information from the plant's Wonderware-based supervisory control nodes, using Wonderware's IndustrialSQL Server as a data repository for the roles-based portal.
"All you need is Internet Explorer," says Iwancio. "From a single portal, users can tell which lines are up, which are down, which are on planned maintenance, and what the current cycle counts are."
Iwancio is enthusiastic about the portal's ability to convey what might be called manufacturing intelligence information, but he also says it's hard to pin down a hard return-on-investment for the portal. "While there's a certain feeling of comfort from knowing what the production status is, and what the efficiencies are, putting a dollars-and-cents figure to the value of the information is tough," he says.
Which, in a sense, reflects the emerging state of manufacturing intelligence solutions. Like Ghandi, when asked of his opinion of Western civilization ("I think it would be a very good idea," he's reported to have responded), manufacturing intelligence sounds like a good idea, but is so broad it loses relevancy unless viewed as specific measures, each with practical benefit.
"The real problem is that the term means different things to different people," says Simon Bragg, European research director with ARC Advisory Group, Dedham, Mass., "Interpretations range from optimization, on through to various forms of visibility."
Yet, as Schmalbach-Lubeca found, the starting point is clear enough. There's a wealth of data at the supervisory control level that can provide insights into the operation of production machinery around the plant. And, correctly interpreted, that data can provide meaningful information about manufacturing glitches, and where opportunities exist to make improvements. So to some vendors—and many users—manufacturing intelligence means doing precisely that: taking raw data and displaying it in a way that facilitates making such interpretations.
Range of approaches
Currently, plant portal solutions are central to the notion of manufacturing intelligence, but they aren't the only approach.
For example, Intellution, a Foxborough, Mass.-based industrial automation software vendor, is carefully positioning its historian products as manufacturing intelligence applications. Plant-floor historians can capture information from specific machines, generally at the supervisory-control level. Plant-wide historians, on the other hand, can capture data from a variety of sources, combine it, store it, and enable analysts to perceive links between an event in one part of the plant, and an associated event in another. A filling line may halt, for example, when a palletizer in the warehouse jams up.
Intellution's new infoAgent is a Web-based tool that in conjunction with iHistorian, Intellution's plantwide data historian, targets the need for the analysis and visualization of production data. The product supports Web-based access to key performance indicators and roles-based information access.
Wonderware also offers appropriate analytic software, even though Schmalbach-Lubeca was not using it. Its ActiveFactory application, says Janie West, director of corporate marketing for the Production Management division of Herndon, Va.-based Invensys, of which Wonderware is a major brand, is a tool for analyzing plant and process data, and delivering real-time or historical trend data in a graphical format. Using the built-in query language, even more sophisticated analyses are possible.
But not everyone in the manufacturing function has the need—or the inclination—for such detail. It's inevitable, for example, that plant managers and vice presidents of manufacturing are more likely to want a quick "heads-up" display of what's going on, rather than having to burrow into a mass of detail. Indeed, that's partly the reason for the launch of Wonderware's easier-to-understand SuiteVoyager portal product. "People tell us that ActiveFactory is a great tool, but complain that it's for a power user, says West. "They say, 'We have three power users—but 100 users who need pieces of that capability.'"
Another vendor in the industrial automation software space—GE Fanuc Automation, Charlottesville, Va.—also has developed portal solutions. Its recently launched CIMPLICITY Digital Cockpit, says Kevin Roach, a GE Fanuc vice president, aims to collect, aggregate, and disseminate data from multiple sources around the enterprise, in a graphically rich format. "If you're a plant manager, you'll have one view of the cockpit—and if you're a CFO, you'll have another," he says.
But is starting with a collection of supervisory control-level interfaces and adding analytical and display capability the best way of getting that heads-up information? Or is it better to redesign the whole information-gathering process so as to work backwards from the information needs of users? Some experts say the key is that solution providers and end-users define the ultimate business case behind a plant portal solution.
"Sometimes technology allows us to do things, but we don't yet have a game plan for what we're going to do with that capability," says Ken Brant, a research director with Gartner, a Stamford, Conn.-based analyst firm. "People need to define what information is required, and what it will be used for—and not just grab information from across the plant and stuff it on a desktop simply because it's possible to do it."
Tackling improvement
It's a given that plant managers might appreciate a ready insight into how well the plant is performing in near-real time, but they're more likely to open their wallets for a tool that produces bottom-line benefits.
ABB, Wyckliffe, Ohio, offers plant management and industrial automation applications that address improved asset performance. "Plants can lose up to 40 percent of their capacity through stops and interruptions," says Tom Mueller, global director of asset optimization products, ABB.
Unleashing this "hidden plant" is the task of the vendor's Production Intelligence system, which—like many of the company's other products these days—benefits from the adoption of ABB's "Industrial IT" integration architecture that simplifies connectivity to plant-floor data.
And such diversity is necessary, points out Mueller, because first impressions of a problem often are deceptive. "Changes in the way that people interact with machinery can have a significant impact on performance," he notes. "Shift coverage, how breaks are handled, and the details of how people physically work a machine—all these have a bearing on the issue, and often are overlooked."
Harnessing three analysis measures—Overall Equipment Effectiveness (OEE), Root Cause Analysis (RCA) and Total Productive Maintenance (TPM)—explains Mueller, the Production Intelligence software aims to pin problems down to their real cause, rather than their apparent cause. "Often, you think you've got one problem, such as maintenance, but actually you've got another: waiting for material, or waiting for operators," he says.
But what about more broadly based applications of manufacturing intelligence? Lighthammer, based in Exton, Pa., and EMT, based in London, Ontario, are two vendors that both claim to provide a variety of benefits by going beyond portals as mere tools. Both offer digital cockpit-style plant portals, but see those as almost incidental to the real task of making sense of what a portal is saying. However, there is some divergence between their approaches.
EMT's approach, according to CEO Dennis Cocco, is to start from the construction of a database. Its VisualPlant solution, built from the ground up on a Microsoft .NET architecture, aims to automate both the collection of data and its analysis. "Where we differ from the supervisory control-based approaches," he says, "is that we aim to make it easy to get the data, and then provide a number of off-the-shelf tools for analyzing that data."
This affords considerable flexibility, Cocco believes. "To us, downtime is an event," he says. "One person might want to look at it by shift, another by operator, someone else by product code, someone else by lot number, and someone else by machine. Competitors' products don't always make that possible because they don't necessarily store the data in the right way."
It's certainly a degree of flexibility appreciated by Cindy Wilson, continuous improvement manager at Presstran Industries, St. Thomas, Ontario. The company manufactures heavy metal stampings, and has implemented VisualPlant to direct its continuous initiatives, she explains. Hooked up to each stamping machine's PLCs, VisualPlant captures the time spent both running, and not running, faulted, and the cycle count—in conjunction with which die is in the machine at that time.
While this is useful, there's one other slightly out-of-the-ordinary benefit, according to Wilson. "We can use it to verify that the improvements that we were expecting to get actually occurred," she says. "In other words, when we projected a 10-percent improvement in cycle time from a particular improvement activity, did we actually get it?"
For its part, Lighthammer believes that the emphasis should be placed more on the links that collect the data, and less on the database that stores it. To chief executive Russell Fadel, the database question is almost a non-issue: storing data twice is poor practice—better by far to integrate it at the point of use.
Take OEE calculations, for example. A seemingly simple sum to do, it's deceptively difficult to actually get all the data together. "You need three pieces of information from three different manufacturing systems, and no single vendor produces a system that includes all three," says Fadel. It's preferable, he adds, to concentrate on the question of creating the link than storing the information.
Enterprise integration
One thing, that the two vendors agree on: in the manufacturing intelligence space, the most promising future belongs to those who can offer enterprise system integration into the portal, to provide even more meaningful manufacturing intelligence. The logic is simple. "As supply chains intrude ever more tangibly onto the factory floor, plant managers need to be able to monitor not just machines inside the plant, but the supply chain itself," says Nigel Montgomery, a research director with analyst firm AMR Research, Boston.
Lighthammer already offers enterprise system connectivity, while Cocco says EMT's upcoming Enterprise Edition will add new capabilities in this area. Real-time process management system vendor Verano, Sunnyvale, Calif., also offers it—albeit with a couple of caveats.
How significant these caveats are depends on the user. The first is simple: Verano, which in 2000 acquired the automation integration software division of Agilent Technologies of Palo Alto, Calif., presently focuses on the process industries, where it has more than 200 blue chip customers. The second caveat is that Verano provides the middleware-connectivity part of the manufacturing intelligence problem—including the enterprise links—but not the full display and rich analytical tools that other vendors offer.
Nevertheless, users such as David Chesson, information technology director at Celanese Acetate, Charlotte, N.C., are more than happy with Verano's capability. The company uses the Verano solution to integrate data captured from shop-floor systems (e.g., materials movements) with enterprise system and supply chain management systems from SAP, Walldorf, Germany. "[Verano's software] maps our shop-floor systems onto our business systems with guaranteed delivery, and no loss of data," says Chesson. "It's made us much more agile. We're not trying to guess where we are now, based on yesterday's data."
Which, at a time when "manufacturing intelligence" most definitely means different things to different people, is as good a working definition of it as any.
| For More Info: | ||
| ABB: www.abb.com | ||
| EMT: www.visualplant.com | ||
| GE Fanuc Automation: www.geindustrial.com | ||
| Intellution: www.intellution.com | ||
| Lighthammer: www.lighthammer.com | ||
| SAP: www.sap.com | ||
| SAS Institute: www.sas.com | ||
| Verano: www.verano.com | ||
| Wonderware Corp.: www.wonderware.com | ||
