Key takeaways
- Scaling is a governance problem, not a sensor problem.
- Model ROI from avoided cost (downtime hours × cost/hour × avoidable share), not sensor count.
- OPC-UA models the data; MQTT moves it. Use both, bridged at the edge.
- IEC 62443 is zoning, not a rebuild. Segment, inventory, least-privilege, monitor — in maintenance windows.
- One governed control plane turns plant-wide rollout into configuration.
Why pilots stall
The pattern is depressingly consistent. A pilot lights up a dashboard on one machine, everyone is impressed, and then nothing scales. The usual reasons:
- The pilot proved a visualization, but not an operations change people actually adopted.
- Identity and security were hand-wired for one line and don’t replicate.
- Data stayed siloed per machine, so plant-wide insight was impossible.
- No single owner was accountable for the rollout.
- ROI was never modeled, so finance could not approve scale-up.
Scaling IIoT is an architecture and governance problem. Solve those and the sensors are the easy part.
Start from a modeled ROI
Predictive maintenance is the classic entry point — and the classic place to over-promise. Model it from avoided cost, not sensor count:
| Input | How to estimate it |
|---|---|
| Unplanned downtime (hours/yr) | From CMMS history or maintenance logs |
| Cost per downtime hour | Lost output + labor + expedite costs |
| Avoidable share | Conservative %, by failure mode |
| Secondary savings | Avoided collateral damage, spare-parts logistics |
| Cost | Sensors + integration + platform |
Run a conservative and an expected case. For the right machines, a credible payback usually appears within 12–18 months.
Get the architecture right: OPC-UA + MQTT
Don’t treat this as a choice. OPC-UA models machine data and enables interoperability between PLCs, machines and historians on the floor. MQTT (often Sparkplug B) efficiently carries that data north to the cloud. The mature architecture exposes machines over OPC-UA and bridges to MQTT at an edge gateway — and integrates with your existing MES, ERP and historian rather than replacing them.
Secure the OT floor: IEC 62443 without breaking production
OT security fails when it is treated as an IT rebuild. Treat it as zoning instead:
- Segment the network into zones and conduits.
- Inventory every asset.
- Apply least-privilege access.
- Add monitoring before controls that could trip a line.
- Roll out one zone at a time, in maintenance windows.
A governed platform with device identity and audit trails already covers a large part of the IEC 62443 control set.
Design for scale from the first cell
The difference between a pilot and a program is whether the second line costs the same effort as the first. With one governed control plane — device identity, role-based access, governed OTA, audit trail — adding a line becomes configuration, not a fresh integration project. That is what turns a proof-of-concept into a plant-wide deployment.
Where Fundamentum fits
Pilots stall when every new line means re-integrating identity, security and data plumbing by hand. Fundamentum gives an IIoT program a single governed control plane — device identity, role-based access, governed OTA and an audit trail — so scaling from one cell to the whole plant is configuration, not re-engineering. It bridges OT data to your existing cloud (AWS, Azure) when required, inside a SOC 2 Type II perimeter. See the platform →
Frequently asked questions
How do I calculate the ROI of a predictive maintenance project?
Start from avoided cost, not sensor count. Quantify current unplanned downtime hours, the cost per hour, the share you can realistically avoid, plus savings on secondary damage and spare-parts logistics. Subtract sensors, integration and platform cost. Model a conservative and an expected case; a credible payback usually shows within 12–18 months for the right machines.
Why do so many IIoT pilots fail to scale?
Common reasons: the pilot proved a dashboard but not an operations change; identity and security were hand-wired and don’t replicate; data stayed siloed per line; there was no owner for the rollout; and ROI was never modeled, so finance couldn’t approve scale-up. Scaling is an architecture and governance problem, not a sensor problem.
OPC-UA or MQTT on the plant floor?
Use both, for different jobs. OPC-UA models machine data and enables interoperability between PLCs, machines and historians. MQTT (often Sparkplug B) efficiently moves that data north to the cloud. A typical architecture exposes machines over OPC-UA and bridges to MQTT at the edge gateway for ingestion and analytics.
How do I implement IEC 62443 without disrupting production?
Treat OT security as zoning, not a rebuild. Segment the network into zones and conduits, inventory assets, apply least-privilege access, and add monitoring before you add controls that could trip a line. Roll out in maintenance windows, one zone at a time. A governed platform with device identity and audit trails covers a large part of the 62443 control set.
Can you connect to our existing MES/ERP and historian?
Yes. We integrate with existing MES, ERP and historian systems rather than replacing them, exposing machine data over standard protocols (OPC-UA, MQTT, Modbus) and bridging it to your platform of record. Fundamentum can interface with your existing cloud where required while keeping a governed control plane over the devices themselves.
Related reading
Talk to an IoT engineer — free
Book a FREE 30-minute consultation with our team. No slides, no obligation — a working session on your connectivity, platform or compliance questions.
