I have been working with converting-machinery controls for almost a quarter century. I've seen controls perform miracles that were undreamed of just a couple of years ago. I've also seen controls misbehave so badly that they kept brand-new machines down for months until they were bypassed or replaced. Because of these extremes of performance, my views on automation, computers and controls have swung from "Techno-Nerd" to skeptic and back again so often that I get dizzy thinking about it.

At first blush it seems that there's no predictable formula for success. Some modern drives can literally sing songs, while others perform more poorly than their mechanical brake, chain and flat-belt predecessors. Some gage-profile controls improve product consistency while others must be shut off because they destabilize controls. Sometimes automation pays off. In other cases, reduction of operating labor is more than offset by the need for increased technical support—at a net loss for the company.

However, upon closer examination of projects, there does seem to be a factor that correlates with success. Those who do their homework fare much better than those who just buy or build stuff. This homework begins with an understanding of the economics that underlie a particular situation. What realistic savings in waste, delay, maintenance or safety are expected for upgrading controls? What are the incremental costs? What additional support is needed? Finally, what are the risks? Some projects fail to live up to expectations.

In this brief treatment, I can share only some generalizations I have found from experience. You must judge for yourself what's best for your particular situation. Each process has an optimum control level beyond which there's no incremental payback. A different response may even be needed for nominally identical side-by-side machines running different grades. A different response may be needed for the second of two nominally identical machines running the same grades, but coming online only a few months later.

AC-vector drives have reduced mechanical maintenance because they do not have the brushes that the venerable DC drives do. They also have better control finesse that can be helpful in some unusual applications. AC-vector drive technology has matured considerably since its introduction to high-speed paper rewinders in the mid-1980s. Obsolescence of early models and difficulty taming the response of these hot engines is long past. ACs are no longer much more expensive than DCs; they are similar, depending on the horsepower rating.

Servo drives can literally sing. Applications, such as printing-press registration and packaging machinery, have been revolutionized by this technology. In-shafts and gears are getting scarcer every year as we trade mechanical complexity for control complexity.

AC drives and servo drives have no practical performance advantage over DC drives in most applications. In fact, I've seen more gruesome drive startups in the last two years than I have in the last two decades. The problem is obviously not with the hardware or software: That is much more capable than when I started in the industry.

The problem is wetware. Very few people understand what a drive is supposed to do (web handling) and know how to do it (drive programming). To make matters worse, the new software is so beguiling that ordinary human beings think they can program drives. This is rarely the case. Only a fraction of those who tune web drives most days of the week, year after year, can tune web drives well for difficult positions like unwinds and centerwinds. The operative word here is WEB drives. You could commission a drive and drive engineer famous for robotic or machine-tool, motion-control finesse, and they could fail miserably on a web application.

Scanners have been around for several decades on paper machinery. Measurement of basis weight, caliper and moisture is standard equipment on thousands of paper machines worldwide. This has revolutionized troubleshooting of process-variation problems.

In the past two decades, measurement has improved to the point where it is (often) trustworthy enough to be given responsibility for profiling controls. Gage controls have made paper ever more uniform and thus more runnable by its downstream customer. Waste and delay have been clearly reduced in most cases, and payback has been achieved in many cases.

Paper scanners cost millions of dollars; the associated gage controls are even more expensive. They are so complex that they often demand the full-time, onsite support service of the equipment manufacturer. However, even with the paper industry's breadth and depth of engineering resources, problems are common. It is not unusual for gauging controls to be shut off or manually overridden because they degrade performance and destabilize control. These challenges are even more severe in other industries that don't have such resources.

It is also more difficult to accurately measure thin materials, like film, than thicker materials like paperboard. Insufficient measurement sensitivity is usually the limiting factor with profile controls. Sensors are not good enough to "see" gauge-related problems that show up in winding. If they can't see them, they obviously can't correct them.

It should go without saying (but I need to anyway) that do-it-yourself gauging projects don't work. Every one I've seen on thin webs has failed so miserably that not a minute of production was accomplished under their direction.

Process controls have made converting more responsive and more flexible while at the same time more consistent. What was a series of grade adjustments made by an operator who may have variable experience and knowledge is now made by the (hopefully) best practices of process-knowledgeable people. At the very least, every shift produces the same as the last because the operator is given less direct discretion on process changes. Grade changes are faster as a new recipe is called up within seconds.

Control-screen ergonomics are universally abysmal. They often look like they were programmed by a computer programmer, electrical engineer or technician who has never operated a machine before. In fact, this is almost certainly the case. It is also almost certain that the control designer never studied ergonomics. In rare cases, the programmers consult operators during the design phase of the project. In even rarer cases, they study the new operator's difficulties after startup to revise their first effort.

Defect detection has been made possible by the increase in computer-processing speeds; 100 percent scanning can be done even at machine speeds of thousands of ft/min. At first I was highly dubious, but I've seen a couple of paper-machine installations that could flag coin-sized defects without too many false alarms. Slower and narrower machines should be easier or less expensive.

Low-contrast defects are more difficult to see with a camera than with your eyes. Also, the correct camera for this application is a 1-D linescan, not a 2-D camera. Again, don't do-it-yourself. Finally, picking up defects does nothing in itself to save money. You must either reject that material (Are you prepared for the costs?) or diagnose and fix the root cause (Do you have the resources to do so?).

Operators at many converting plants never touch the rolls they make. They are wound, discharged, wrapped, labeled and stacked by hard automation. (Robotics is sometimes used, but isn't usually the best choice, because the forte of robotics is reprogrammability, not repeatability). Automation can eliminate one or two operating positions at a machine.

Obviously, reducing operating manpower must be great enough to pay back the difference between fully-automatic and fully-manual roll-handling stations. However, the real issue is reliability. Any component failure not only brings roll handling down, it can sometimes take the producing machine down with it. Older manual systems have fewer parts and thus tend to fail less frequently and are simpler to repair.

Repair, in fact, is not the right way to think of reliability. If a mission-critical component fails in the first year (where recovery takes an hour) or first decade (where recovery takes the better part of a shift), the component should be redesigned. If it failed once, it will likely do so again. The problem is not so much this particular failed part; it's that this part is but one of hundreds of mission-critical pieces. Each and every one must have extraordinary reliability to keep the entire system up and running.

Systems now keep track of nearly every aspect of manufacturing, from order entry and scheduling to production, label printing, inventory and shipping. Waste and delay are tracked and reported. Online diagnostics, training and help are available for operators and maintenance. Thus we can do more with less.

Payback on supervisory controls is even more difficult to calculate because the benefits are not so directly tied to waste, delay and labor reduction. It's not unlike office automation. Will an upgrade in computer hardware or software make a person more productive, or will they eat up any potential gains in equipment costs and the installation and retraining time?

With operating turnover sometimes as short as a month, initial training and recurrent training is imperative. A few high-end systems offer online training or simulator training to assist other more traditional routes.

Practical online training is almost unavailable. Builders are often exhausted by merely making systems work. Instruction manuals, documentation and training are the last parts of the project and thus the first to be dropped if money is short.

Automation can increase safety by having multiple levels of barriers, interlocks and permissions to keep operators away from hazardous situations. Automation can do things that were previously done manually, thus freeing operators from the hazards of repetitive-motion disorders or the risks of close man-machine interaction.

Automatic operations, such as in wound-roll changes and roll handling, can catch people off guard. A robot has such a nasty backhand that it's best to never enter the rink with it unless multiple levels of lockout/tagout have been put in place.

Successful control upgrades return money to the company's stakeholders in any number of ways: Fewer injuries, increased speed, reduced waste or reduced downtime. Quality can be improved so that new orders fill the machines, and price increases can be passed on to those customers who demand the best. Success is not guaranteed. However, chance favors the prepared mind.

It's the supplier's responsibility to make things work smoothly and reliably. Reliability is easily measured and can be specified as a minimum acceptable "mean time between repairs," or "percent of uptime."

It's the converter's responsibility to buy the right things from the right people. They must also have the resources to support the new technology. Everyone must do their homework before the project begins, even if the project must be put on hold to get better answers.

David Roisum can be reached at 920/725-7671, e-mail: DRroisum@aol.com, www.roisum.com