Make the Network(s) Fit the Application

For all the discussion of industrial networking and all the installations around the world, they aren’t as common as one might expect. In fact, many more plants aren’t networked than those that are. Michael Bryant, director of the Profibus Trade Organization, estimates that only 25% of the potential market for industrial networking has been penetrated so far, in spite of all the ind...

By Peter Welander November 1, 2006
  • Should you network?

  • Networking benefits

  • Design basics

  • Choosing a platform

Two real-world examples

For all the discussion of industrial networking and all the installations around the world, they aren’t as common as one might expect. In fact, many more plants aren’t networked than those that are. Michael Bryant, director of the Profibus Trade Organization, estimates that only 25% of the potential market for industrial networking has been penetrated so far, in spite of all the industry’s efforts. Most companies still hardwire individual instruments and control devices to PLCs or less sophisticated systems. When a pressure sensor and readout box exist in isolation, the “network” is people who walk around carrying clipboards.

If you have not yet made the leap, keep in mind that the products available now offer the widest options ever. Industrial networking has seemingly endless possibilities from a host of vendors. For example, there are at least 50 different fieldbus platforms alone. When approaching the process for the first time, the enormity of the subject can boggle the mind. Moreover, a network of any size will virtually always involve at least two technologies, depending on how much distance you want to span from individual field devices to ERP or IT networks.

In addition, a networking project can be driven with very different objectives in mind. On one hand, you may simply want to save cabling while you connect field devices to a PLC on a specific piece of machinery. On the other, your president may want to be able to see current output levels for all manufacturing processes in real-time from the front office in another state. These are vastly different projects, but fall under the same industrial networking umbrella.

Why network?

Given this often confusing technology, why would someone leave the hard-wired world? What’s wrong with a clipboard anyway?

The first and oldest answer is that networking saves on wiring costs and cabling. Networks started originally to reduce the need to run individual cables from each device to an I/O card on the controller. Companies that made controllers designed proprietary fieldbus platforms, which is one reason why there are so many. Depending on your situation, saving on wiring and installation costs could still be a compelling reason.

As networks become more sophisticated, reasons for implementing them also increase:

  • Plant safety systems now operate on the same networks as control functions;

  • Controllers offer ways to diagnose system problems and reconfigure equipment without having to rewire;

  • New instruments can communicate more information, if there is a means;

  • Production asset management (PAM) programs operate with control networks; and,

  • As IT networks move down from the enterprise level, the ability to communicate with the plant floor provides motivation to tie it together.

“Networking makes sense when the cost is right, but might not as a retrofit. ‘If it ain’t broke, don’t fix it,’” advises Tom Phinney, senior fellow with Honeywell Process Solutions. “If you don’t have an economic incentive to change, then don’t. That doesn’t mean there aren’t potential areas for improvement. There are tons of applications that cry for a way to reduce costs, reduce emissions, and reduce maintenance, but there has to be a will to do things. It all comes down to building awareness, gaining focus, and making the business case. Pushing smarter instrumentation farther into the plant using the network is one way do that. You can use ancillary information from smart devices to predict maintenance problems. Tell your flowmeter, ‘Give me your noise signature.’ If you do that over time, you can look for changes and detect sludge buildup. You can’t do that with 4-20 mA.”

Adds Katherine Voss, executive director of ODVA, “Today, the mandate in corporations around the world, both large and small is ‘Get connected! We don’t want just control of our manufacturing processes, we want information about our processes and we want it anywhere, anytime!’ The network wave for manufacturing is coming. Copper wire on the plant floor is out; seamless, multi-hop routable networking is in.”

Starting at the bottom

Given the complexity of industrial networking, any attempt to divide it into categories will be messy since so many areas overlap. The suggestions here are general and designed to present topics on the conceptual, rather than specific level.

Networks can cover a wide span from individual field devices to IT and ERP networks. So far, no single networking platform has proven practical to cover the entire distance from a single limit switch to the top. Source: Control Engineering

If we consider an individual discrete field device to be the lowest point on the spectrum, the PLC is the first major level working up from the bottom. There are networking technologies that exist below the PLC and others above, but few cross the line and function on both sides. The same network that links PLCs to each other and to devices at a higher level usually does not extend lower. Conversely, the networks below PLCs are not normally used to link PLCs to each other, although some of the more sophisticated fieldbus platforms can to varying degrees. However, these rarely extend into higher levels.

The simplest network is probably a group of individual devices tied to one PLC for a single process or machine. These devices can be hard wired to I/O cards one at a time with no networking at all. However, when the number of devices multiplies and distances increase, this becomes expensive and unwieldy. The simplest type of network ties these together using a fieldbus, reducing the amount of wiring and makes configuration of the PLC easier.

Even when building a relatively small network, it is critical to know what types of devices you are tying together. This answer determines the amount of data your network must process:

  • Small data items, e.g., limit switches, photo cells, proximity sensors, etc., are one-bit devices and require simple bit data fieldbus networks;

  • Medium data items, e.g., intelligent devices, smart instruments, variable frequency drives, small loop controllers, etc., will need more sophisticated fieldbus technology; and,

  • Large data items, e.g., robots, HMIs, larger loop controllers, etc., will need sophisticated fieldbus technology or Ethernet.

Multiplying the amount of data per device times the number of devices determines the overall capacity and speed a network will require. If you have enough one-bit devices, the capacity need can reach that of more sophisticated instrumentation.

“What type and volume of I/O will the network be supporting?” asks system integrator Mike Darnell of Prism Systems. “A paper mill, for example, may move 5,000 data points around as a normal part of their operating process. Ethernet is sometimes overkill for low I/O count and availability of Ethernet-enabled devices is limited at the low-end market, but it is a great solution for large applications. Customers at the high-end generally have a better understanding of Ethernet technology and often view it as less proprietary than fieldbus platforms.”

Those starting out need to think about how their networking needs are likely to change over the next three to five years, suggests Larry Komarek, automation product manager, Phoenix Contact. “Will you be using the same kinds of devices but more of them, or are you seeing the need for the amount of data per device to increase? When you add more intelligent devices or those with diagnostics or status information, the data per device goes up. If that’s the case, the decision will balance on the point of paying now or paying later. If you decide to save money now and stay with a fieldbus, think about how fast things change in your factory. Will you have time to replace a less expandable system because you’ve outgrown it? If you make a management decision to save cost now and don’t put in an Ethernet platform, be sure to factor in the future cost and changeover time to add it later.”

Moving above the PLC level

Once a network passes the individual PLC level, project scope grows significantly and becomes more open ended. PLCs need to talk to each other, communicate with HMIs, deliver data to historians, and pass information up to engineering, enterprise level users, IT network, etc. These applications change the nature of networking dramatically, and what goes on above the PLC level has little resemblance to what goes on below. While some sophisticated fieldbus platforms support PLCs talking to each other peer-to-peer, extending communication higher toward enterprise level requires Ethernet technology.

Use of Ethernet networking has been growing rapidly with deliveries of industrial Ethernet devices climbing at double-digit rates. There are at least 10 competing platforms, each with its own strengths and weaknesses. Some have specialties designed to complement specific applications. Most are associated with one or more sizeable equipment manufacturers, although they all generally claim to be open architectures.

Currently, Ethernet networks link PLCs, smart sensors, remote I/O blocks, HMIs, etc, with each other and higher level systems, such as IT and ERP networks “Use the right tool for the right job,” suggests Carl Henning, deputy director, Profibus Trade Organization. “It is likely that a number of networks are needed to complete a project. Look for the right combination of platforms. Don’t use an IT network to connect limit switches! If I’m the engineer responsible for the project, I’ll look for the number of devices and their complexity. I’ll consider the requirements for integrating to a production system. There’s no substitute for good engineering judgment.”

Eddie Lee, national industrial Ethernet product sales manager for Moxa Technologies, sees Ethernet technology moving to span the entire range. “Today the Ethernet is powering barcode scanners at Wal-Mart checkouts,” he notes, “so it’s happening at a device level in the commercial world. That same technology is moving more extensively onto the manufacturing floor. Maybe I’m looking farther ahead, but it’s coming.”

While a small fieldbus installation can exist in relative isolation, Ethernet solutions provide potential means for outsiders to hack into your networks and do serious damage. While no such network is completely secure, there are levels of protection to keep out unauthorized users. This is another topic that you must consider carefully, and there are many resources available.

Making sense of the choices

Before spending too much time reviewing the options, it’s best to find out if your company has specific directives on network platforms, preferred vendors, established training, or parts inventory. Such policies are difficult to ignore when they involve investments of time and money. Likewise, it’s best to determine if you are building strictly for yourself and your own plant, or if you need to follow practices established at other locations. For many reasons, there may be influences at work on your choices beyond finding the ideal solution for this specific situation.

Once you have established your working constraints, the process needs to commence with an assessment of your situation and needs:

  • What is the network supposed to do? Is this strictly a means to save cable connecting I/O points to a single PLC, or will it carry real-time data to the president’s office and beyond?

  • Is there existing equipment or networking with which you need to communicate or interface? Unless you’re beginning with a greenfield plant, machine, or process, legacy equipment and systems likely will need to be integrated.

  • Make sure you enquire up the ladder as well as down. Your IT department may want input since they will have to connect to the system.

  • How will you integrate network security?

  • Will this be a home-grown project, or will you use a system integrator?

  • Do you have know-how going forward to use and maintain the system once it’s in place?

Look at the larger system or process that you are networking to establish the scope of the project. Begin by cataloging the actual devices you have and what they do. Individual devices need to be separated according to their sophistication and amount of data that they transmit.

One group of choices you will make as part of the project will be specific software for networking, security, and control functions. A fieldbus or Ethernet network does not help run your process, manage plant assets, schedule maintenance, or generate production reports by itself. Those operations come from applications that run on the network. The lists of choices available for networking and control software covers many pages and their discussion has filled many volumes. The Control Engineering Website is an excellent place to continue your research.

Two real-world examples

Glad they changed

A tobacco processing plant uses a combination of machines and skilled operators to manufacture a variety of cigars and other products. While the facility was historically un-automated, the management understood the potential savings and increased production efficiencies of an effective networking system. Prism Systems of Mobile, Alabama took on the task to integrate a plant full of old, legacy equipment that wasn’t automation friendly. Ultimately, the system tied together 21 PLCs, 61 RFID readers, 223 conveyor drives, 7 HMI consoles and countless odd devices using two fieldbus systems and an Ethernet network to tie it all together.

“With this modern network architecture in place, our customer now has instant access to real-time production data,” says Keith Jones, president of Prism Systems Inc. “The results include dramatically improved production efficiencies for the plant and a fresher, higher-quality product for the consumer.”

Still holding out

Hytrol is one of the largest suppliers of standard and custom conveyors in North America. Their products range from simple stick conveyors to high speed sliding shoe sorters used by major shipping companies and big box retailers in their distribution centers. These sorters demand networking performance that pushes the envelope. A complete unit can be 500 to 600 feet long, running at 700 fpm with up to 1,000 I/O points. These are generally one-bit discrete I/O devices including proximity switches, photo eyes, encoders, barcode scanners, ID readers, and others. The amount of information coming back to the processor at any given time is staggering, with all I/O points updated every 3-4 ms.

This puts huge demands on the speed and determinism of the PLC and supporting network. “Lots of I/O over a long distance functioning at high speed is the worst case situation. Most field buses and controllers can’t handle it,” says Wendell Rogers, controls specialist. “They were either not fast enough or not deterministic enough.” Currently, Hytrol has to depend primarily on hard-wired I/O blocks using quick-connectors to simplify the massive cabling task, but they are always experimenting with new networking technologies in search of exceptional performance.