Pro-tips exist to make you a better automation professional. Often that means helping reduce human workloads, sometimes it means you need to plan for eventualities that you hope never happen. In this context redundancy is a means to continue your employment.
The key equation is M = xN + y
M is how many devices you actually install
N is the minimum number of devices required
x and y are the variables for your backup
With pumps or motors, this is fairly straightforward. It becomes less straightforward with other subjects.
Consider a "critical" bar screen system, for the influent of a waste-water plant. The plant needed one screen running, so they put two screens (x=1, y=1) in a single cabinet (x=1, y=0) and with a single power feed in (x=1, y=0).
Another good one is fiber. Many older sites may have a network layout that has a single cabinet with all of your lines connecting. Some fiber rings are "virtual," where a single fiber is used but four strands are picked up inside to make it appear to be a ring. Lifting a single point will not drop anything but putting a drill through the fiber will take everything down.
What I would want is a ring going outside your campus, dedicated to OT and managed by redundantly fed switches at each node, then run a mesh in the middle. Have servers in multiple buildings at either end of your campus. However, that can get quite expensive quite easily. How much uptime do you need?
Budgetary restraints are simple enough to understand. It becomes less clear when you consider complexity. Take redundant CPUs: two processors sharing I/O. I found one system that had such a set up taken out by a single switch, which failed in such a way to cut communications from the I/O off to both CPUs.
If you set this up, think before you ask. Make a plan, document it and make sure to keep your hands on a copy of your proposal. If something goes bad and you recommended the fix beforehand, you should be (partially) covered.
Comments