Control networks and Ethernet
Anyone reading Control Engineering realizes the growing impact of Ethernet networks in control systems, and much of its impact is based on sound economic considerations. The significantly lower cost of network cable, interface cards, network analysis tools, and network management software is letting Ethernet become a major component of most control systems.
Anyone reading Control Engineering realizes the growing impact of Ethernet networks in control systems, and much of its impact is based on sound economic considerations. The significantly lower cost of network cable, interface cards, network analysis tools, and network management software is letting Ethernet become a major component of most control systems. Use of Ethernet as a control network is changing the way we control, manage, and support our networks.
Control networks had always been point-to-point, relying on 4-20 mA loops or direct dc signals. One of the big advancements in digital networks was the move to multi-drop lines, where more than one device could sit on the same wire. Unfortunately, with standard Ethernet we have taken a step backwards; 10 Base-T and 100 Base-T Ethernet networks are not multi-drop networks but are cascaded star networks. Each device communicates to a hub or switch, and hubs and switches communicate to other switches. One of the advantages of Ethernet is that switches can be inexpensive and most switches can mix 10MB and 100MB devices. Because of this topology, the typical industrial network contains many hubs and switches. There are single wire runs for medium and long distances, and small star networks at each end for devices located close together.
This differs from most IT Ethernet installations wired like a phone network, with each device wired to a large central cabinet. IT departments use this topology to simplify their network management, provide easy access to switches, and allow quick network reconfiguration.
An industrial Ethernet network with distributed switches will put more requirements on the network support. Most hubs and switches must be managed, which means that they can be remotely queried for their status and remotely controlled. Unfortunately, this means we cannot use the $29.95 8-port 10-100MB switch purchased from the local computer superstore. Instead we need managed hubs, switches, and network management software.
This network topology requires a different approach to support and leads to one important question. Who owns the control network—the IT department or the control department? The owner of the network has responsibility for ensuring that the network devices are operating correctly and that the cabling is configured correctly, yet many networked devices are not standard servers or workstations.
In many companies the initial answer is that the IT department owns the network and network devices. However, this can be a recipe for disaster when the control department owns the control applications. Many system failures are not the result of failure of any single element but are because elements simply do not work together. In addition, changes in one element, while correct, can cause failures in apparently unrelated areas. When standard IT networks are used in control applications, they need a different organization than the typical IT support organization. Control applications are highly connected and often tightly coupled to the operating system and network services. The typical IT department has separate support groups for servers, networks, and applications.
Supporting Ethernet-based control systems requires a team of application specialists, network specialists, and server specialists. This team will have to work closely together on any significant problem and needs to be under central management to be effective. The conventional approach would be to have the support team part of the IT department, but a better approach is to have the team part of the control department. If correct and safe operation of the plant is based on the reliability of the network and network devices, then the people responsible must have the resources to ensure correct and safe operation. Separating network support from control application support is not an effective support strategy.
Incorporation of Ethernet networks in control systems requires that support organizations accept responsibility for the entire control system, including the applications, servers, networks, hubs, switches, and other network connected devices.
Dennis Brandl is the president of BR&L Consulting, a consulting firm focusing on manufacturing IT solutions, based in Cary, N.C.