Distributed control: another choice for multi-station loading systems.

Initial hardware cost, installation cost, flexibility, and expandability weigh in its favor. But is it too good to be true?

Just three years ago, buyers of multi-station materials conveying and loading systems had a lot less to think about when making controls decisions. Back then, most state-of-the-art systems employed a central controller wired to each individual piece of materials-handling equipment on the shop floor. The only real choices involved the brand of PLC controller and how much input/output (I/O) to specify. Today, however, a growing number of equipment suppliers have come out with another option--"distributed" controls.

Some of these distributed control systems eliminate the hardwired central PLC altogether, replacing it with a collection of "smart" microprocessors controlling each field device--whether it be a pump, hopper, or valve. Other distributed systems retain the PLC but spread out the I/O capabilities among remote modules in the field. In either case, distributed control for plastics materials handling has grown in popularity since AEC reportedly introduced the first distributed I/O system three years ago. "The general trend in industrial control over the past decade, as hardware became better and cheaper, has been to distribute control or I/O. It's clear to us that on large material-handling systems, distributed control makes a lot of sense," says AEC engineering manager Ron Newlun.

To advocates of this approach, distributed I/O and distributed control both mean simplified wiring, low installation costs, and enhanced flexibility when additional loading stations must be added to a system. "We expect distributed control to become standard for materials-handling applications," concludes David Cosner, v.p. of sales and marketing at Universal Dynamics.

Still, distributed control hasn't won over everyone. Many suppliers believe distributed control only confounds troubleshooting efforts and spreads control maintenance tasks throughout the plant. "Our philosophy--and it's one supported by our customers--is to get the controls off the processing machine. The only thing on the machine should be the vacuum receiver," argues Motan president Chuck Thiele.

SPREADING OUT THE SMARTS

The first thing to realize about distributed control systems is that they come in many flavors, with each vendor taking a different approach. Today, the most completely distributed control system for materials-loading tasks is the Echo System from Universal Dynamics. Introduced at NPE '94, Echo has served jobs ranging in size from one-pump systems up to 10-pump JIT systems with purge valves. One of the most complex Echo installations consists of 10 vacuum pumps and 40 chambers. In this plant, many of the vacuum units are portable and just plug into the Echo network as they move around the plant.

The control logic for each device in an Echo system--e.g., vacuum chamber, pump, sequencing valve, or purge valve--runs on a custom microprocessor board sitting right on or next to the device. "All these boards are `smart' microprocessors differentiated only by their software," notes Paul White, software engineering manager. Uniting all these nodes is a licensed network technology more commonly found in "automated building" applications. "We decided it was a good fit for factory automation," says White. For one thing, the network doesn't need any PLCs. And the network is "open" in the sense that it relies on standard computer protocols--such as DDE for exchanging data with Windows application software. As another example of the benefits of "open" systems, Una-Dyn now supplies a PC-based monitoring system based on Wonderware factory-automation software, which provides a gateway to bring data from the Echo system into spread-sheet, scheduling, or alarm programs.

Unlike some other distributed control schemes, Echo has no supervisory controller--only a simple annunciator indicating the status of each loading station as well as diagnostic messages for the other nodes. In place of the supervisory PLC, the microprocessors at the vacuum pumps run all the scheduling logic individually. "Each pump signals its own needs over the network," explains White. To configure the system, users can operate the entire loading system from any node with a suitable operator interface--which could be located on any pump, purge valve, or hopper.

AEC supplies distributed control systems but does not currently choose to disperse the control functions as widely on the plant floor as Una-Dyn does. "We feel that distributed control should go to the conveying cell level, as opposed to the more extreme level of every conveying system component," says Newlun. "I don't see any sense in having the 'brains' at the hopper, but I can see having them at the pump."

One example of AEC's approach to distributed control involves networking together groups of eight loading stations. Each group has its own control logic and is connected by a multi-conductor wire. More often, however, AEC installs distributed I/O systems, in which a central PLC communicates with I/O blocks on the plant floor. Comparing this approach to hardwired, centralized PLC control, Newlun notes that distributed I/O simplifies wiring in the same manner as a truly distributed control system and still allows some of the most important control functions to be brought down to the conveying cell. For example, AEC's distributed I/O system enables adjustments of conveying times and on/off functions at the hopper--without full-fledged operator interfaces or the cost of wiring back to a central PLC. At the same time, Newlun argues, distributed I/O blocks retain some of the benefits of a central PLC system--such as the ability to readily interface with existing PLC networks. "Distributed I/O is a good middle ground," says Newlun.

Other distributed control variants on the market include ones from suppliers such as the following:

* Conair Franklin offers a single-cable distributed control network with its Selectronic Controls. These networks feature Conair's "UTB" design, which pre-wires all electromechanical devices into a single, quick-disconnect box installed directly on hoppers and vacuum pumps. "This means hundreds less electrical terminations to be made in medium-to-large installations," says product marketing manager Amy Reissener.

* Thoreson-McCosh has offered a distributed control system for three years. Called Tech II, these controls still require a central PLC on the vacuum pump--but only to handle the "master on/off" tasks, says v.p. John McLeod. Most of the individual control functions run at the loading stations. Thoreson-McCosh's distributed control installations are mostly in big plants. Very long wiring runs from remote stations to the central controller are difficult and expensive to install, notes McLeod.

* For the past two years, Comet Automation Systems has supplied distributed control systems made up of a field network of small, fixed-I/O PLCs.

* Premier Pneumatics has added distributed systems to its controls repertoire. "Ninety percent of the applications out there are still served by a central PLC, but we do see a growing demand for distributed control on complex jobs," says marketing manager Mark Wedel.

* Walton-Stout also offers a distributed control network, but no details were available at press time.

INSTALLATION COSTS LESS

In complex multi-station applications, one of the most notable advantages of distributed control or distributed I/O systems is simplified wiring. In contrast to hardwired central systems, where individual wires run from each loading station back to the main PLC, distributed systems typically have a single cable connecting all the loading stations in a "daisy-chain" arrangement. For Una-Dyn, that lone cable is a simple 24-v twisted pair, which carries both the power and communication signals between devices. Thoreson-McCosh's Tech II system also relies on just a single twisted pair. Other vendors currently use multi-conductor cable in a daisy-chain arrangement.

Whatever the type of cable, using just one saves some up-front cost. Una-Dyn's Cosner estimates that wiring costs typically represent 25-30% of the total mechanical installation cost. Some vendors cite even higher wiring figures. "Before, wiring was a nightmare. Now we just tie the wire to the vacuum-tubing lines," says Cosner. Una-Dyn describes the cost of Echo's wiring installation as "almost nil" and will often throw in the wiring installation for free when carrying out the mechanical installation.

Another installation-cost factor is the voltage of the system. Twenty-four-volt systems, unlike 110-v, need no conduit. "So you can run cable as easily as phone wire," says Newlun, noting that AEC also supplies 21-v systems.

Even with distributed systems, however, the network choices--and thus, the wiring method--aren't necessarily clear-cut. For instance, although Comet does supply single-cable systems, controls designer Mark Ponder says many automotive customers have response-time and data-acquisition requirements best served by two or even three cables. "Cabling depends on the data and the response times you want and on the money you want to spend," he says. Ponder describes a balancing act in which some commercial networks are "slow but cheap," others "slow but reliable," and still others "fast and reliable but expensive."

Motan's Thiele brings up another side to the wiring issue: Although traditional hardwiring with conduit costs more up front, "We think it's sounder from an engineering perspective," he says. Wires housed in conduit are harder to sever than loose ones. "If someone is interested only in initial cost, loose wiring is probably the right way to go. But if you break a wire, how much will the downtime cost you?"

OLD FAITHFUL?

On the reliability question, two points of view emerge. One holds that distributed systems inherently cause less maintenance trouble; the other, that they cause more.

On the one hand, distributed controls spread the control logic among several physical locations. "If you lose one node of a distributed system, you don't have to lose the whole plant. Put a forklift through the main panel, and you'd still be running," says Comet's Ponder. Even if the main cable is cut, every loader in a truly distributed system could still run independently--though the network itself would not.

On the other hand, Motan's Thiele joins other vendors of central PLC systems--including Process Control Corp. and HydReclaim--in arguing that centralized controls offer a more trouble-free approach because all control maintenance and troubleshooting is reduced to one box that can be quickly replaced in the event of trouble. Ponder concedes that "distributed systems can be tough to diagnose unless you know what you're looking for."

WHEN NETWORKS GROW

Will your plant ever expand? If so, distributed control and distributed I/O systems may offer some benefits over traditional hardwired systems. "If a guy wants to expand, it's easy," says Newlun. "With a distributed system, just find the closest hopper and tap into the cable. When you do that with a hardwired system, you would have to run conduit back to the central PLC."

That doesn't mean that central PLC systems are not expandable. But you need the foresight to order extra I/O capability up front. "Automotive customers typically specify 30% spare I/O and memory," Ponder notes.

Beyond easy expandability, distributed controls foster another kind of flexibility--the ability to add new features easily. "Echo allows us to integrate new, creative features," says Una-Dyn engineering v.p. Bob Crawford. One example is the company's new "line proof" module. Running as a node of the Echo system, it normally prevents operators from making hose connections inconsistent with the instructions entered into the controller. But one customer uses it to do just the opposite: The control system automatically reconfigures itself based on whatever hose hook-ups the operator makes. "In the past, we would have had to integrate mechanical switches; now it's just a software solution," Crawford says.

OTHER HARDWARE CHOICES

Control systems of all types--distributed or not--are also differentiated by their reliance on custom electronics or off-the-shelf components. Una-Dyn builds all its own Echo microprocessor cards in-house. "We're not big PLC fans around here," says White, noting that custom electronics keep hardware costs down. Comet also tends to use its own proprietary electronics, except on the largest factorywide automation jobs, which need more I/O than its custom boxes offer. At AEC, which offers both custom and off-the-shelf electronics, Newlun agrees that custom components can save money in small, simple systems and may be appropriate for "the guy who buys on price alone." But he notes that on the largest systems, the cost of off-the-shelf hardware drops. "The cost per I/O point is becoming very low in large systems," he says.

Newlun adds that off-the-shelf electronics also have the advantage of fitting into existing PLC networks. "People come to us for total plant automation, and they want it quickly, which rules out custom solutions."

For example, an AEC customer needed to add conveying, blending, and drying equipment to its plant. The company already had paint lines and milling machines controlled by Allen-Bradley PLCs connected to a DEC VAX computer via an Allen-Bradley Data Highway network. "The ability to use this existing network was key," Newlun notes. "There are processors who simply don't feel comfortable with custom electronics," Newlun says.

COMPARING THE COST

With hardware costs constantly changing, the cost/benefit balance between distributed and central control is a moving target, but Newlun does offer a rule of thumb: "In general, as the number of hoppers increases, the benefits of distributed I/O increase."

Some suppliers bristle et too much emphasis on installation costs--a mindset that Motan's Thiele calls penny-wise and pound-foolish. He says, "The most important thing is not necessarily the cheapest installation. Many customers can't stand any downtime. What is important to them is troublefree operations and quick troubleshooting."

"Some companies want to empower their operators to make adjustments from any station et any time. Others continue to like central control," observes Conair's Reissener.