APY PREMIUM GROUPThe job plant owners most often underestimate is replacing network hardware inside a process control system. It sounds like "swapping a few switches" — but those switches are the arteries that let the entire control system talk to itself. Pull the wrong one and the line stops immediately. This article describes the approach our team used on a project at a large beverage plant. We describe the engineering method only: no client name, no equipment brands or models, and no commercial details, in line with our confidentiality obligations.
The brief
The plant runs a process control system installed many years ago. The system itself still performed well, but the industrial network hardware at its core had entered end-of-life status — no longer available new, thinning spares in the market, and declining vendor support.
The plant's engineering team set three objectives, which are the same three we encounter on nearly every project of this type:
- Improve the performance and stability of the control system network
- Reduce the risk of downtime originating from network hardware
- Create a foundation that supports future upgrades to other machine control systems
The third objective matters most in budget terms, because it reframes the work from "maintenance expense" into "infrastructure investment" that several later projects can build on.
Why a plant network is not an office network
This is precisely why this type of work should not be handed to a general IT team alone. The key differences:
| Aspect | Office network | Control system network |
|---|---|---|
| Effect of a brief dropout | A file loads slowly; the user hits refresh | The control system sees a device as missing and may trip the process or halt the line |
| Timing requirements | Flexible, measured on averages | Must be deterministic; acceptable latency measured in milliseconds |
| Environment | Air-conditioned room, clean rack | High temperature, dust, humidity, vibration, electrical noise from motors and drives |
| Path redundancy | Standard spanning tree is usually fine | Requires a fast-recovering ring, and recovery time must be measured in practice — not taken from a datasheet |
| Work window | Friday evening works | Must wait for a planned production shutdown, which may occur only a few times a year |
So the deliverable is not "new hardware that links up." It is new hardware that reproduces every function of the old system, within an agreed shutdown window, with a rollback path if things do not go to plan.
The 6-step method we used
The guiding principle is to move as much work as possible outside the shutdown window, leaving only what genuinely cannot be done while the line runs.
- Survey the site and document reality — trace every cable run, record which ports are actually in use, photograph every cabinet, and compare against existing drawings. This step gets the most time, and it usually reveals that drawings and reality have diverged.
- Match replacement models — map capabilities one by one: port count and type, fibre type, ring protocol, supply voltage, temperature range, and mounting method. The item most often forgotten is physical size — a replacement larger than the original means cabinet rework in the middle of the night.
- Configure and test off site — configure all new hardware in the office, build the ring on the bench, cut links to measure real recovery time, and save a labelled configuration file for every device. The goal is to reduce on-site work to "remove, mount, plug in."
- Prepare the site in advance — install rails, pull additional cable, prepare power feeds, and label every cable before the shutdown day, all while the line is still running normally.
- Cut over during the planned shutdown — work section by section in the planned order, testing communications before moving to the next. The old hardware stays in place until the new system is confirmed working — that is a rollback plan that genuinely works within minutes.
- Verify original functions, train, and hand over — confirm the control system behaves exactly as it did before the change, then train the plant team to configure and replace the hardware themselves, and hand over manuals, backup configuration files for every device, and an updated network diagram.
A 9-point checklist before starting a control network upgrade
- Do you have a network diagram that matches the plant as it is today — not as it was originally commissioned?
- Do you know the vendor lifecycle status of each device?
- Are configuration files for every existing device stored somewhere other than the devices themselves?
- Do you know which part of the process each device is responsible for communicating?
- What ring recovery time can the system actually tolerate, and has it ever been measured?
- How many hours does a usable shutdown window give you, and how often does one occur?
- If the cutover fails, how many minutes does it take to revert to the original system?
- Will the new hardware physically fit the existing cabinets — size, mounting, and power?
- After handover, can your own team replace and configure the hardware, or must a contractor be called every time?
Points 7 and 9 are the ones usually missing from unusually cheap quotations, and they are the two that most affect long-term total cost.
Lessons that transfer to other plants
- Hardware working well today is not the same as hardware without risk — the risk lies in your ability to source a replacement, not in the condition of the unit.
- Infrastructure should be replaced while you still control the timing — a planned replacement always costs less than an emergency one mid-production-week, in both price and quality of work.
- The survey budget is not the one to cut — the accuracy of the survey determines whether cutover night is smooth or chaotic.
- Always have the contractor train your team — a good outcome leaves the plant more self-sufficient after handover, not more dependent on the contractor.
- Design for the next project from the start — leaving spare ports and bandwidth means the next control system upgrade can begin without touching the network again.
If your plant faces a similar situation, see our article on obsolete PLC and HMI migration, or read more about our PLC control system services.
Summary
A control network upgrade is the kind of project that, done well, nobody notices — because the best possible outcome is that everything behaves exactly as before, only now on hardware that will have spare parts and vendor support for the next decade. Success here has little to do with the technology chosen. It comes down to the thoroughness of the survey, the readiness of the rollback plan, and the discipline to test one step at a time.