Distributed Control Systems for Material Handling

Editor's note: Distributed input/output systems for material handling systems and other auxiliaries made a big splash at NPE 2000. Since then, more suppliers have incorporated this networking and control technology into their equipment. Ron Newlun, controls group manager for AEC, outlines here some of the nuts and bolts of distributed I/O.

A recent innovation in material conveying and control systems is a distributed input/output (I/O) design using machine-mounted I/O blocks and device-level networks. Distributed systems use control blocks at each hopper (or other device) attached to two multiconductor cables, one for power and one for communications, that run back to a control panel. It is not necessary to string new wiring from each device back to the control panel.

This alternative to a traditional, chassis-based I/O system makes it possible to reduce the installation time of conveying systems by more than 50 percent. Processors cannot afford any more downtime than is absolutely unavoidable. When molding such commodity items as bottle caps that have a return of about one-tenth of one cent per part, even an hour of downtime could mean the difference between profit and loss in a given week.

Also, all capital expenditures must be tied to the assurance of a significant return on investment. Buying technology on the basis of installation time benefits alone is not always the best idea. The ideal scenario is to find a balance between initial acquisition cost, system quality, and technology level.

Total Cost of Ownership

While reduced installation time is a benefit to processors, the total cost of ownership (TCO) needs to be considered. This includes acquisition cost, installation costs, long-term maintenance costs, and costs associated with future system expansion. While a chassis-based I/O system has a lower initial acquisition cost, a distributed I/O system can have a significantly lower TCO for midsized to large processors, especially those with plans for expansion.

Traditionally, plastics material handling applications have used chassis-based, or local, I/O, which is rack-based I/O modules that require processors to pull wires from each I/O device in the plant back to a centrally located control panel.

This panel can be more than 600 ft away.

For example, consider a local I/O in a medium-sized plant covering 200,000 sq ft: 24 injection molding presses, 36 material distribution hoppers, and three vacuum pumps.

Each hopper in this example has at least five wires running back to a corresponding local I/O module, and each wire needs to be in a special conduit to protect it from environmental hazards. The result is a system of piping or wires ranging from 1 to 4 inches in diameter hanging from a facility's rafters. Whenever there's a system failure involving a wire, a technician must determine which of the many wires is the culprit, and then fix it. This is a technician's nightmare. The processor must also pay the cost of all the wires linked to local I/O modules. At $1/foot per wire in conduit, the wiring costs are at least $60,000.

To reduce wiring costs, plastics industry suppliers began developing distributed I/O systems in the 1990s. Industrially hardened I/O systems were available, but device-level networks, the other vital component of a distributed I/O system, were not. The industrial networks available at the time were susceptible to incorrect wiring, electrical supply spikes, and electrical noise. Device-level networks also did not have the error-checking capabilities processors wanted and were limited by the number of nodes (device addresses) and maximum cable lengths with which they could function.

Networking technology has now matured, and several device-level networks and I/O options are available. AEC uses DeviceNet with Allen-Bradley (Rockwell Automation) KwikLink flat cables and ArmorBlock MaXum I/O for its new vacuum material handling system.

By using DeviceNet open architecture communications protocol technology with Allen-Bradley/Rockwell Automation MaXum self-diagnostic I/O blocks, the system goes beyond providing on/off information. The DeviceNet can connect up to 64 nodes and, using KwikLink flat cable, can extend more than 1300 ft, which is good for for both injection molding and extrusion applications. That distance capability generally eliminates the need to use repeaters (signal boosters).

If a repeater fails, all elements of the network downstream of the repeater are out of commission until the repeater is repaired or replaced.

How It Works

AEC's Vactrac vacuum material handling system, which is capable of controlling up to 50 hoppers and seven pumps, distributes plastic pellets from large storage bins at an injection molding or extrusion facility to hoppers mounted above each machine. Each hopper in the network is connected to a single control by a DeviceNet KwikLink flat network cable and power trunk. ArmorBlock MaXum I/O snaps onto the flat cable and mounts onto a bracket that is fastened directly onto the facility's pre-existing vacuum line. Because of this, the system requires no conduit or mechanical support structure and the I/O does not need any protective enclosure.

The ArmorBlock MaXum is a self-contained package that combines either eight or 16 I/O points accessed through 12-mm micro quick-disconnect connectors, a DeviceNet communications adapter and power supply into a sealed housing.

With this vacuum material handling system, the two DeviceNet flat cables run along the entire length of the vacuum line, and the I/O is placed wherever the end user needs to connect to a vacuum hopper or pump. Patch cords extend from the I/O blocks to the devices on the hopper.

Continental PET Technologies Inc. (Fremont, OH) is one user of the distributed I/O system. "One of the most convenient features of the ArmorBlock MaXum I/O blocks is that they connect to the DeviceNet network by simply piercing the flat cable with their connectors," according to Nick Mercorella, plant engineer at Continental. "We were able to install our entire system without ever screwing a wire down, thus eliminating any potential miswiring and the system debugging costs associated with such problems. It also allowed for the installation to be done by nonelectrical personnel."

Continental's injection molding operation has four pumps and 30 material hoppers, feeding a combination of blenders and dryers. With the pumps on ground level and the dryers and blenders on the mezzanine level, there is up to 700 ft of cable end-to-end.

Since the basic application requirements of a hopper or a pump typically consist of just two inputs and two outputs, there is ample room on the I/O block for many options. Adding a filter cleaner to a hopper, for instance, is done by connecting a 12-mm quick-disconnect cable to the connector on the block, and then enabling the option on the PanelView terminal. One can also add a remote proportioning valve used to integrate unused material with regrind.

By contrast, the same changes on a system using local I/O would require new wiring and potentially a new conduit run from each hopper. In addition, the entire control system would probably have to be shut down in order to download the program changes needed to accommodate the new options. And the whole process begins anew when any options are moved elsewhere in the facility or when any new options are added.

To place a new machine between two existing machines, the processor obtains another I/O block and snaps it onto the flat cable at its nearest point, connects the patch cords, and drops it down to the new hopper. In a local I/O application, the same change would require an electrician to run new wires all the way back to the system controller.

Reducing Maintenance Costs

One way to reduce long-term maintenance costs is to use the diagnostic capabilities of a distributed I/O system. With point-level diagnostic capability, as displayed on an operator interface terminal, maintenance personnel are notified about failing field device components, including their precise location.

Diagnostics let maintenance personnel know what to bring to any given location, whether it be a screwdriver or a replacement part. This simplifies troubleshooting and allows nonelectrical personnel to identify a malfunctioning electrical sensor or solenoid.

Local I/O vs. Distributed I/O

While distributed I/O makes sense for many vacuum material-handling systems, there still are applications in which local I/O makes sense. For example, processors with on-staff electricians may not be as concerned about labor-intensive installation of wiring. Consider the following scenario.

Suppose the initial acquisition cost of the control hardware for a local I/O system is $20,000, with an additional $150,000 in installation costs. And let's say that the hardware cost of a distributed I/O system is $60,000, but the installation costs are only $25,000. Up front, the total costs of the distributed I/O system are half the total cost of the local I/O system. However, if the buyer has ample and capable on-staff electricians, the higher installation costs of the local I/O system could be absorbed by those employee's salaries. In cases such as this, the total costs between the two systems may be much closer.

The future of material-handling technology appears to be in its information-handling capability. By using I/O blocks connected via an intelligent network, instead of individual copper wires, a lot of information is made available to the processor. This information can be used to help improve production scheduling, plan preventive maintenance of mechanical components, and keep a system running at peak capacity.