CIM team integrates multiple plants and processes.

A CIM implementation group at Reynolds Metals' Flexible Packaging Div. finds 'open' solution that extend beyond plant walls.

If establishing a computer-integrated manufacturing system at just one plant sounds daunting, try coming up with systems for six far-flung plants carrying out several distinct processes on dozens of lines--everything from blown film and extrusion coating to laminating and printing, and even from injection molding to thermoforming. To make the CIM problem even tougher, these varied processes do not really end at their own plant walls but instead share a constant flow of materials and information.

Reynolds Metals' Flexible Packaging Div., headquartered in Richmond, Va., tackles this challenge every day with the help of a team of integration pros dedicated to meeting the company's wide-ranging and never-ending CIM requirements. More than just picking the right computer products or "black boxes" to install at a given plant, this CIM group helps plant personnel build Reynolds' growing network of information highways--ones that bring together the company's business and processing worlds electronically.

Many smaller companies will never have the breadth of applications of giant Reynolds Metals, but the team's methods offer important lessons in good implementation practices. "Our biggest challenge, and maybe our biggest accomplishment, has been getting people in the plant to buy into CIM," says division engineering manager and CIM team member Floyd Schrimsher. And since many of CIM's challenges and solutions extend beyond the nuts and bolts of a single process, more generalized insights from the company's experts can be valuable for anyone wanting to boost their CIM capabilities.

COMBINING KNOWLEDGE

After five years of working with the various Reynolds packaging plants in Virginia, Kentucky, Pennsylvania, Nevada, and Arkansas, the CIM group's experience underscores the synergistic benefits of computer and processing specialists working closely together. Dr. Keith Vetter, a division CIM manager, holds up computer networks as an example. "Networking itself is not a strong point for most plant personnel," he says. "But knowing what to do with the process data--that's where plant people have the knowledge and desire." The company's organizational culture influences CIM implementation in other important ways. For one thing, CIM group members target information exchange across processes and locations as a prerequisite to CIM success--especially when that exchange means interaction between business and technical staffs. "Before it was like two armed camps," says Bobbie Strum, a process engineering supervisor from corporate R&D. To foster this interdepartmental communication, the company has established both corporate and divisional "manufacturing automation committees" to bring together staff members involved in disparate processes but sharing an interest in CIM. "The committees are forums for the for exchange of information" says Strum. "We get together to share successes and horror stories." The committees are the forerunners of the division's full-time CIM team efforts.

While not all CIM solutions apply across processes, a surprising amount of know-how does translate broadly--even techniques borrowed from the division's aluminum processes. "Drives are a common thread throughout our division," says Vetter. "So the plastics processors have been able to learn from the aluminum guys and vice versa." For Reynolds' many extrusion lines, Shackleton Drive Systems--a Eurotherm International company in Reston, Va.--provides real-time distributed control over a fiber-optic network.

Yet another implementation key has been corporate financial support for the team, despite the fact that CIM is tough to justify through conventional cost analyses. "It's difficult to show return-of-investment on a single CIM product because a lot of the benefits are intangible," explains Strum. "But these benefits ultimately speak to incremental improvements in quality and performance." By recognizing this fact, "management has been satisfied with the longer-term CIM mission rather than demanding immediate results," notes Schrimsher.

STAYING FLEXIBLE

Long-term results, however, do translate to increased costs up front. "You have to spend more money to get the flexibility you need," says Vetter, who estimates that flexible, expandable systems can carry as much as a 25% cost premium. The systems recommended by the group got their flexibility from an "open-systems" approach.

On the positive side of the cost equation, "open" architectures impart a number of benefits that may eventually offset any initial price premium. Perhaps the most compelling reason for favoring openness is the frustrating tendency of computer products to become obsolete so quickly. "Unlike mechanical equipment that lasts a lifetime, a CIM product may last 6-10 months," says CIM manager Wiley Via. "You need to make systems flexible or you're stuck with short-term projects producing short-term results."

Openness also fights the big-company penchant for centralizing technology choices, alleviating threats to individualized solutions. To free each plant's technology options, some of the CIM group's work involves research into communications standards. Consider programmable logic controllers (PLCs), for example. In the past, the division might have imposed a particular brand of PLC on all plants for an initial cost savings, even if that choice might not have made the most sense from flexibility or servicing standpoints at some plants. Today the plants use as many as five different PLC brands. "We suggested a certain communication standard rather than dictating a certain brand of PLC," Schrimsher explains. And to further boost flexibility, the company's PLCs are now specified with more expansion slots than the plants need up front--15 instead of five, for example.

On the software side, the CIM group tests and ultimately recommends products that foster independence from a specific hardware platform. Vetter cites FactoryLink from U.S. Data Corp., Richardson, Texas, as one of these software products, and the Packaging Division uses it in four of six plants. "We chose it for its fairly open architecture," Vetter says. This supervisory-control and data-acquisition software (SCADA) gives the team a head start in creating its own control applications--in this case through a graphical programming method that lets users create applications by linking "objects" depicted on the monitor. Without such a package, the group would need to develop a system from scratch, complete with a graphic user interface and database. "With FactoryLink, a lot of structure is already there for us," notes Strum. And because FactoryLink runs on any open platform, it insulates the applications from hardware changes.

FRUITS OF LABOR

One important offshoot of the group's work with FactoryLink has been a program to continuously track emissions of volatile organic compounds (VOCs) at its printing, laminating, and extrusion coating operations in two different plants. Called Reynolds' Emissions Monitoring and Reporting System (REMARS), this application uses a proprietary mathematical model to project emissions at a given production rate for any point in time. Before REMARS, it would take from two days to two weeks to do the materials-balance calculations that reveal VOC emissions.

But with real-time emissions information in hand, production supervisors now run through "what-if games" and make production decisions in order to stay within the plant's EPA-dictated VOC limits. "REMARS helps identify the most effective way to stay in compliance," says Strum. And it helps maintain the complex equilibrium between emissions requirements and production goals. "Sometimes the highest-emitting machine isn't the one that should be shut down," notes Strum. In keeping with the group's open-systems approach, the REMARS calculations are programmed as a FactoryLink module, thereby preserving the application if the hardware changes down the road.

One more application under development will better integrate the division's quality-control functions into existing CIM systems--especially for printing, which has about 500 quality-related parameters to track. "The goal is to put process control and quality control at the machine level, to perform quality control as the process runs," says Schrimsher.

For now, quality control is often one step removed from the process. Reporting systems, for example, currently reside on plant-level mainframe computers that store product reports known as "history cards." These reports list important parameters, setpoints and run lengths for each job. During post-job reviews, the history cards are compared with actual production conditions to see if the set-up needs modification.

Eventually, process setpoints will come through the PLCs at each machine. And at each operator display station, every important process parameter will correspond to a color-coded block. If the process strays from control, the system will generate a defect report and automatically flag the product.

Other processes besides printing will receive similar sorts of automation upgrades in the future as the group brings in inputs from thickness gauges, temperature sensors, and drive-control subsystems into its SCADA system. Some processes, however, already have a jump on CIM. Gauge control in extrusion coating applications is already handled by the CIMLine system from Eurotherm's Gauging Group in Billerica, Mass. And there's also a CIM system in place at Reynolds' PVC film plant in Grottoes, Va. This system has been continuously evolving; it was first based on an Allen-Bradley Data Highway and now has an Ethernet backbone.

WIDER NETWORKS

While it's tempting to think of the division's six plants as stand-alone operations requiring separate CIM solutions that stop at their own walls, Vetter points out that physical materials are not the only thing that flow between foil, plastics processing, printing and converting operations. "Information needs to move too," he says. So, much of the CIM group's work today focuses on establishing links between plants. The goal here is to get production and materials-property data from one plant to another. The second plant will then base its setup on data from the first. To cultivate this kind of exchange, Reynolds' is in the process of implementing a wide-area network--a more far-reaching version of the local-area networks now in effect at individual plant levels. "Our goal is to collect information only once, at the most appropriate place, and make it available where needed without requiring specialized knowledge," says Strum.