What is redundancy anyway?
We all want more for less. And nowhere is that truer than in industry’s constant quest for energy efficiency. But in the big energy-intensive process and infrastructure industries, electric motors account for 70% of their electricity usage – mostly in processes where continuous availability is what counts. So companies are waging competitiveness war on the continuity of service front. Their weapon? Redundancy.
But just what is “redundancy”? I sometimes think it is whatever you want it to be: duplication, fault tolerance, producing the same result in alternative ways, etc. Businesses need to get it straight in their minds what they mean by “redundancy”. Otherwise they impact their process with no improved reliability to show for it.
Ethernet: tops for topology configuring
What I’m interested in here is redundancy in distribution and protection switchboards. There’s huge potential there in Ethernet-based networks. They bring flexibility and scalability and high-speed communications. And you can configure them into different topologies and build in levels of redundancy. That’s the second big question you should ask yourself: What level of redundancy yields the dependability that meets your needs at the right cost?
Meaning and level of redundancy
So what do I mean by redundancy here? Applied to switchboards, I mean addressing a specific motor or motor starter subsystem via two different paths with no manual intervention.
What level of redundancy do I want? Let’s take two:
- Redundancy at the level of a network’s backbone so it can address sub-networks of motor starters via two different paths.
- Redundancy at the sub-network level, so that sub-networks can reroute when they talk to individual motor starters (or other devices).
Ring topology for network backbone
Traditionally, big companies and industrial facilities use Ethernet tree topology for their electricity distribution. More and more, however, are switching to fault-tolerant ring topologies as their network backbone. Personally, I think they’re right. With their twisted pair cables they’re heavily wired – with all the wiring costs that entails. You can’t connect very many devices to them, either. Anything over seven and the computation and test packets required balloon out of all proportion.
Ring topology is much simpler, which makes calculation simple. And on industrial plant’s shop floor, there is no practical limit to the number of devices that can be connected.
Redundancy in the ring
Ring topology lends itself to redundancy. In many networks you’ll find an additional counter-rotating ring. In the event of a switch failure, data automatically flows across to the other ring. And failover recovery times are extremely fast.
The only disadvantage? No international standard. So failover methods and times depend on the switch manufacturer. You should, therefore, try to use switches produced by the same manufacturer to be sure of homogeneous, continuous service.
To sum up, then, the ring constitutes a backbone of managed switches that all sub-systems connect to. The ring tolerates any single point of failure and automatically reconfigures to reroute subsystem traffic.
Failover is automatic, transparent – and very fast. The typical recovery time is 500ms but can be as fast as 3ms.
Let’s take a look at redundancy in the subsystem level.