Network Resiliency on the Manufacturing Floor

Unplanned downtime is a significant detriment to any business. In order to make sure that processes remain uninterrupted, consider implementing these resiliency protocols when designing your industrial network infrastructure.

Ricardo Romero is a systems analyst at Interstates Control Systems
Ricardo Romero is a systems analyst at Interstates Control Systems

On the plant floor, communication uptime in the manufacturing process is a top priority. And—for any business—ensuring the supply-chain process remains uninterrupted is also a top priority. Corporate and industry standards, and many other factors, play into the level of robustness and system availability on a network. A resilient network reduces the risk of unplanned downtime on the plant floor and application communication downtimes.

Resilient network technologies include layer 2 protocols, EtherChannel, and others, which can be used to construct a loop-free logical topology, and are designed specifically for Ethernet networks. The primary function of a spanning tree protocol (STP)—a layer 2 protocol—is to prevent loops and broadcasts but can also be used to provide redundant links if the active links fail. Consider a ring topology—at the switch- and device-level—for automation applications that require high-speed convergence and single fault recovery for perpetual manufacturing. Many automation devices have embedded switch technology which allows them to participate in ring topologies. Select a topology that meets the performance, cost, and spatial requirements of your industrial applications.

Convergence times are defined as the time it takes a switch port to go from forwarding to blocking on a ring port and blocking to forwarding on another. The IEEE and industrial network hardware manufacturers have also introduced significantly faster protocols with quicker convergence times in the low milliseconds. Some examples of convergence times of the resiliency protocols are: 1-3 ms for a Device Level Ring which is supported on Allen-Bradley hardware, 30 ms for N-Ring which is supported on N-Tron hardware, and 250 ms for Rapid PVST+ which is supported on most common switching hardware.

It is important that the application requirements are understood when selecting a protocol, such as the type of traffic (motion control, time sync, and I/O and safety control), Requested Packet Intervals (RPIs), and bandwidth. Most major vendors have tools or charts that can assist with determining which protocol will work best.

Convergence is controlled by a manager or supervisor and is typically applied to a primary switch which monitors the health of the ring topology via ring packets. When the role holder stops receiving health check packets, it converts the ring topology to a linear bus topology by blocking one of the ring ports in a matter of milliseconds depending on the protocol used. Convergence times in an industrial network architecture must be considered to avoid application and device timeouts due to connectivity failure. Human-machine interface (HMI), message instructions, I/O, and produce tags are a few examples of network connections that must go uninterrupted.

To support further resiliency and real-time communications, any latency and jitter can be minimized with the use of Internet Group Management Protocol (IGMP) to control the delivery of multicast traffic and the use of Quality of Service (QoS) to achieve real-time requirement of multiple types of traffic flows.

Remember to include these resiliency protocols as a topic of discussion in your next design meeting and you will be on the way to a 100% production uptime status.

 

Ricardo Romero is a systems analyst at Interstates Control Systems, a certified member of the Control System Integrators Association (CSIA). For more information about Interstates Control Systems, visit its profile on the CSIA Industrial Automation Exchange.

More in OEE