News  Briefs  Features  Calendar  Opinions  References  Products  Subscriptions  Editorial FAQ  Media Kit

06 April 2000
The Ethernet Decision
by John McGilvreay

Straight answers on Ethernet's suitability for industrial control networks

Ethernet links office automation and industrial control networks. This is the most significant advantage of Ethernet-based industrial control networks (see Figure 1). It promises to revolutionize the way management interacts with manufacturing operations. Open systems unchain live manufacturing data, enabling it to be distributed freely across enterprise networks to help a company run more effectively. Even at this early stage, this capability taps a range of tools too extensive to list comprehensively, but at the very least, these tools would include the following:

• manufacturing execution systems
• computer-based information and command for managing production resources, processes, costs, labor, data collection, documentation, work in progress, and so on
• new technologies for traditional functions such as time and attendance recording, labor reporting, and materials tracking
• computerized maintenance management systems
• graphical views of process equipment and processes accessed by multiple users to pinpoint faults and failures, order design fixes, and accelerate repairs and changes
• closed-circuit television for monitoring
• data warehouses
• quality information systems for tracking compliance with ISO 9000 and industry-specific benchmarks
• decision-support systems
• electronic forms processing
• data support for control and automation processes

Effectively using these and other yet-to-be-developed tools hinges on the seamless integration of office automation and industrial control networks.

The factors discussed previously provide an almost overwhelming rationale for using Ethernet in your next control network. Nonetheless, significant challenges exist in adapting Ethernet technology to mission-critical manufacturing system requirements. Fortunately, most of these challenges are being met. The most important questions about Ethernet implementation include the following:

Can my existing legacy control system be migrated to an Ethernet-based network? Your company has spent a lot of time and money developing its existing industrial control networks. Although legacy fieldbus systems, with an upper bandwidth limit of 2 megabits per second, will be an early casualty as more management tools emerge, many companies will want to keep using parts of their legacy fieldbus control network where practical. Integrating legacy devices, instrumentation, and I/O is achieved using gateways that support existing control and device networks. These gateways provide a smooth migration of installed control networks to Ethernet (see Figure 2).

Can Ethernet-based networks be made sufficiently secure? Using Ethernet-based networks eliminates the need for costly gateways and translational devices that access and distribute data across all levels of the industrial control network. This, however, opens your industrial control network to a potential Ethernet weakness: unauthorized users tapping into the network's control, automation, and/or business level. One way to stop this is to use VLANs, whose inherent security can restrict traffic. Also, using well-thought-out access policies provides stringent levels of security. This lets different operators use the system 24 hours a day, 7 days a week, without compromising mission-critical operations. Organizational policies can help restrict access to control rooms and remote-access locations. Firewalls, which let only authorized users access and distribute data, are another useful security tool. Finally, encryption/decryption and user authentication allow virtual private networks to offer secure local and remote communications through a nonsecure Internet channel.

Can Ethernet-based industrial control networks withstand harsh environments? System robustness refers to how a product's design is suited to a particular environment. Networks installed in industrial environments may need to contend with potentially explosive situations, extremely high and low operating temperatures, or excessive electromagnetic noise caused by large motors or conductors affecting the communication path's transmission characteristics. Carefully considering the challenges of a given environment and the various products available can ensure a high mean time between failure. Selecting Ethernet products specifically designed for industrial applications (e.g., industrial switches, interfaces, and hubs) can eliminate problems down the line.

Are Ethernet-based industrial control networks resilient enough for mission-critical applications? The cost of downtime in a mission-critical industrial application extends well beyond losing manufacturing capability to consequential losses (e.g., raw materials, equipment damage, personnel injuries, and environmental pollution). Research has shown that about 20% of downtime is due to network-related errors; thus, it is crucial to consider how robust a network is against failures. Although downtime can't be eliminated completely, two basic techniques can reduce the risk and minimize the impact of unscheduled outages: fault avoidance and rapid recovery.

Fault avoidance prevents faults from occurring. It encompasses the right processes and properly trained people, as well as the right technology. An example of fault avoidance is using redundant power supplies, which can prevent network failures due to power outages.

Rapid recovery minimizes downtime when outages or system failures occur. Redundant components can maintain operations in the event of a failure, while online repair of faulty components prevents further downtime. An example of rapid recovery is network path redundancy, as well as protocols that govern the switch to a redundant network path.

The IEEE 802.1d "spanning tree" standard is the traditional method for Ethernet recovery when a link or repeater fails in the network path. Communication reliability/resiliency is manifest in path redundancy. With redundancy implemented, the crucial question becomes "At what speed can the attached system detect and bypass a network failure?" You can design an Ethernet network to detect when a particular path cannot pass data and then automatically switch to a backup path within 1 second.

For industrial control networks, however, the time to recover from a path failure is critical. If a real-time application, for example, requires a deterministic response within 500 milliseconds and your spanning tree recovery time (e.g., 1 second) is longer than the application requires, the system will fail. Industrial rail switches and redundancy managers can enable detection and switching within 300 ms—well within the necessary time constraints for most applications (see Figure 3).

The evolution of Ethernet technology has alleviated past concerns about its ability to deliver the level of deterministic performance that industrial applications and processes require. In turn, this has enabled Ethernet's incorporation into scalable, highly resilient fieldbus networks that are simple to design, build, implement, and manage.

Most current control and automation networks can't meet the data-intensive demands of the process industry into the next millennium. Ethernet provides a practical high-performance solution, with the ability to operate distributed high-bandwidth networks that deliver unmatched performance, determinism, and resilience through sophisticated robust design. We believe Ethernet can meet or exceed the current requirements of more than 98% of industrial control networks, despite the broad range of applications and processes they support.

Additional Information

Figures and Graphics

• Figure 1: Linking computers from the office to the factory is key in accessing and exchanging data.
  
• Figure 2: It's important to migrate legacy control systems into the new industrial control network.
  
• Figure 3: In this example of a network, a redundancy manager and rail switches are used to ease redundancy.
  

Author Information

John McGilvreay is distribution manager for industrial automation with Hirschmann Inc.'s Automation & Networking Solutions division in Pine Brook, N.J. Formerly, he was an engineer on IBM's mainframe, midrange, and token-ring network product lines. Hirschmann is a division of Rheinmetall Elektronik AG of Germany. The company produces networking products and systems, as well as mobile communication antenna systems for automotive, AM, FM, cellular, GPS, and wireless LAN and local loop applications.