Coalescing Networks

Keith Jones isn't the type who'd use a hammer to turn a screw. He believes in using the right tool for the job. For Jones, president of automation systems integrator Prism Systems Inc. of Mobile, AL, that means that the right network for some parts of particular automation and control installations isn't one with more bandwidth and greater capabilities, like Profibus or ControlNet.

By Hank Hogan for Control Engineering August 1, 2006

AT A GLANCE

Right-sized network

Easier configuration

Greater transparency

Network families

Keith Jones isn’t the type who’d use a hammer to turn a screw. He believes in using the right tool for the job. For Jones, president of automation systems integrator Prism Systems Inc. of Mobile, AL, that means that the right network for some parts of particular automation and control installations isn’t one with more bandwidth and greater capabilities, like Profibus or ControlNet. Instead it may be something much simpler like ASi.

‘You get more bandwidth on some of these bigger networks,’ says Jones. ‘Sometimes you need that bandwidth. But the simplicity of some of these little low-end networks is great. So a lot of times we’ll end up using multiple networks.’

However, he notes, there’s always a main network, one that binds everything together. Consequently, it must handle disparate communications in a secure way that reaches from the plant floor and beyond. The governing bodies of the various networking standards, along with automation vendors, are reacting to that need. As a result, installations are being implemented faster, significant sums are saved, and capabilities are increasing every day.

Standards groups respond

Not surprisingly, advocates for the various standards believe their brand of networking is well suited to be the plant backbone. Carl Henning, deputy director of the Profibus Trade Organization (PTO), notes, for example, that Profinet employs software proxies to help integrate dissimilar networks and devices. These proxies act as translators, allowing different brands of hardware to look the same for applications and end-users trying to get one to talk to the next.

‘You don’t really care what brand controller, what fieldbus, what anything is on the next one. It’s all transparent, handled through the proxy for you. So that’s the beauty of this from a user standpoint,’ says Henning.

Equipment manufacturers create and maintain the proxies, which are XML files that define the input, output, and other needed information. Using testing labs, PTO certifies that each proxy conforms to the standard. Devices, therefore, appear as components, which can be linked together graphically. By avoiding labor-intensive programming, Henning says, automation projects can cost significantly less.

FOUNDATION fieldbus isn’t to be outdone when it comes to its ability to handle different networks, notes David Glanzer, director of technology development at the Fieldbus Foundation. Communication is done over Ethernet, with intelligent switches ensuring that messages go where they’re supposed to and aren’t broadcast over the entire network. The architecture puts control level communication on the 31.25 kbit/s network, dubbed H1, and on 100 Mbit/s Ethernet—HSE (for high-speed Ethernet). Management communication is done via OPC, with servers talking to one another.

As for integrating different devices and networks, that’s done through a flexible application-specific block model found in both H1 and HSE. This approach provides a way to map foreign nodes into something similar to all the other devices. Glanzer notes that this concept is being extended. ‘We’re trying to get a common interface for the applications software,’ he says.

At ODVA, executive director Katherine Voss points to a media-independent platform, the Common Industrial Protocol (CIP). This approach, she says, isolates the object model from the lower network layers. Thus CIP extends across both Ethernet-based and other networks, and offers extensions for safety, motion control, and other applications. ODVA jointly maintains CIP with other interested parties, with users reaping the benefits, she explains.

‘CIP allows companies to integrate I/O control, device configuration, and data collection across multiple networks,’ says Voss.

At the OPC Foundation, president Thomas Burke points out that facilitating collaboration and interoperability among all network standards has always been a goal of the group. Work on the Enhanced DDL (EDDL) standard, an effort in which OPC is taking part, should allow end-users to diagnose, configure, and extract real-time and asset management information from more than 15 million devices and instruments from a host of manufacturers. OPC is also pushing for and actively helping develop the emerging OPC Unified Architecture (OPC-UA) standard. (See related ‘Technology Update’ in this issue.)

The goal is to have OPC servers communicate to a physical device that exposes its information and provides seamless standardized access to those data. Vendors would have to implement this for their respective devices. When that’s done, OPC-UA would allow for the complex information in a device to be generically discovered and browsed. Corresponding data would be exchanged between the device and higher-level software applications.

Burke notes that users don’t want to spend time writing code to move such information. ‘Transparency is the key to interoperability. End-users buy products from multiple vendors and they expect that the products are able to exchange information without writing any software,’ he says.

Doing one thing well

As for vendors, they’re responding in various ways to the need to integrate networks. Tim Sweet, manager of product marketing for Experion control and I/O at Honeywell, can appreciate why this effort is necessary. ‘Each of these networks is optimized on some parameter that is mutually exclusive with the other parameters,’ he says.

An example is Profibus, which he notes is best at very high-speed control that quickly moves fairly simple data structures. In contrast, F OUNDATION fieldbus has more complex data and control structures and a lower transmission rate. To Sweet, this different focus isn’t bad. He’d like to have various networks that do one thing very well, rather than a single, all-encompassing network that excels at nothing.

Vendors like Honeywell that provide host systems for the various networks face two challenges. The first is how to ensure reliable and secure communications. The second is how to make end-user configuration easy and foolproof. Whatever networking standard is used has to provide the basis for both.

Sweet reports that this job is getting easier. For one thing, there are more tools for bridging between networks. Third-party communication stacks are now available that allow for more reliable and secure connections. He also notes that collaborations like EDDL can provide more commonality in the way a host system communicates with various devices on different networks. In addition, work by the FDT group is also increasing harmony among networks.

This increased cooperation between standards is fairly new, with EDDL, for example, only a few years old. It’s driven in part by the need to hold down cost and thereby spur adoption of new networking technology.

Alex Johnson, director of system architecture for process systems at Invensys, agrees that different networks are designed for a particular application domain and that each has its strengths and weaknesses. On top of that, many of these advanced networks aren’t deployed in brand new locations. ‘There’s a lot of installed equipment that no one can afford to replace even if the perfect solution was found,’ he says.

In the case of Invensys, Johnson says the approach is to develop a large set of standard interfaces to traditional and intelligent I/O, as well as to SCADA products and intelligent equipment protocols and buses. Standardizing data transfer, alarms and messages, system diagnostics and configuration is done by normalizing the physical and logical communication layers so that each application gets data in a standard format. There’s also software normalization involved, through the use of FDT technology for configuration and system diagnostics. Invensys also has been heavily involved in development of OPC-UA, which implements Web services in process control.

Other vendors are also involved in and support this effort. Emerson Process Management and Siemens Automation and Drives, for example, recently announced a technology exchange that by mid-2007 will result in Siemens products supporting F OUNDATION fieldbus as well as Profinet, while Emerson products will support Profinet as well as F OUNDATION fieldbus. The two product groups will achieve this using EDDL and OPU-UA. In a statement released when the joint effort was announced, Anton Huber, a Siemens A&D board member, noted this integration will prove particularly useful in mixed networking environments. He said, ‘Users can choose from a variety of process instruments that have the interfaces to suit the control system concerned.’

Rolling tobacco, rotating discs

For examples of how this network coalescence and integration works in practice, consider a water utility, a compact disc manufacturer, and a tobacco plant. Each presented its own challenges and produced a unique, yet generally applicable, solution.

The water utility, located in the southeastern U.S., provides water and wastewater treatment to more than 700,000 people over a 1,500 square mile area though more than 200,000 service points. In addition to increasing demand, the utility had to test for an increasing number of contaminants, with more than 200 currently monitored. To handle control and reporting needs, the utility implemented an Invensys DCS and installed a wireless SCADA network. The 23 application workstations communicate via TCP/IP over a T1 line, while eight hubs receive 900 MHz spread spectrum wireless signals from Modbus-configured PLCs at 155 remote sites.

Invensys’ Johnson noted that the utility needed reliable, but not necessarily redundant, real-time control. So critical processes, like adjusting pH or chlorination, were isolated to local treatment or pumping stations while configuration, historical, and maintenance data were shared among the 23 workstations. Software sharing of data took place by mapping values from the DCS standard application to values from remote sites and vice versa. One benefit of this setup is that the cost in time and money of the production of operations and compliance reports has been cut significantly.

PTO’s Henning points to a compact disc maker that had a mix of PLC controllers from Omron, Siemens, Sharp, Rockwell Automation, and Mitsubishi. These were running Profibus and DeviceNet as their fieldbus. Through use of proxies, the controllers talked to one another and an OPC server, enabling one machine to reliably tell the next in the production flow when it was running and when it was stopped.

33% less engineering time

As a result, Henning says, engineering time was reduced 33%, trouble-shooting improved, and commissioning time was cut by four weeks. Taken together, substantial savings resulted.

Finally, there’s a project Jones of Prism Systems knows well. Called in at the last minute to save an automation project at a tobacco rolling and automated cigar production plant, Prism Systems faced a tangle of dozens of PLCs and not much time to connect them. Prism went with a Profinet-based Siemens solution, except for a few spots where ASi and similar low-level networks were used. The main data network choice largely turned on the availability of software that would make the job easier. ‘It moves the object level up and it allows you to incorporate communications into it,’ explains Jones.

Freed from writing code, Prism’s staff shaved 20% off the development time and was therefore able to implement the project faster and on deadline. The cigar maker saw decreased production time and labor, increased productivity, and better work-in-progress tracking thanks in part to an RFID tagging system.

That melding of networks and products from different vendors is something Jones sees as a natural outgrowth of the development of standards-based networks. ‘That’s a part of the reason you’re going to these open networks. It lends itself to multi-vendor projects,’ he says.

Related Control Engineering articles include:

Product Research: Industrial Networks

Legacy Moves to Ethernet

Fellowship of the Fieldbuses

Also, see more on Enhanced DDL (EDDL) from HART Foundation and Fieldbus Foundation