An operational digital twin is a 1:1 model of a running plant or line that stays connected to live data from PLCs and SCADA, with business context from ERP, so you can monitor performance, spot anomalies and test changes on the model before touching the floor. iUNI, the Wartens group's digital twin and immersive technology company, builds these twins from CAD data with SCADA, PLC and ERP integration and a web-based 3D viewer that needs no software install (iUNI Digital Twin Development). The hard part is not the 3D. It is getting the right data into the model safely.
Operational twin vs commissioning twin
A virtual commissioning twin is built to test control software before startup: the model reacts to the PLC program and nothing is connected to a live plant. An operational twin comes after startup. It reads the plant as it runs, and it is used by operations, maintenance and engineering rather than by the commissioning team. The best projects build one model that serves both, which is how iUNI lists it: planning, training, monitoring and virtual commissioning from the same twin.
Wartens' angle is that we are automation engineers first. The twin is built around the control system: the same tag structure, the same states and the same logic the plant runs, so what the model shows matches what the PLC is doing.
Where the data comes from
ISA-95 is the usual map for this. It defines the interface between enterprise systems and control systems, the levels from field devices up through supervisory control, operations management and business planning (ISA-95). An operational twin draws from several of those levels at once:
| Source | Typical data | Used in the twin for |
|---|---|---|
| PLC | States, interlocks, counts, fault codes | Live machine status on the 3D model |
| SCADA and historian | Trends, alarms, events | Anomaly detection, replay of incidents |
| IoT sensors | Vibration, temperature, energy | Predictive maintenance |
| MES or ERP | Orders, SKUs, schedules | Context: which product, which batch |
For the PLC and SCADA layer, OPC UA is the common choice, because it is platform-independent and carries both data and its structure (OPC Foundation). Most modern Rockwell and Siemens systems can publish data this way, directly or through the SCADA layer. ISO 23247 adds a reference framework for digital twins in manufacturing, which is useful when you write the specification (ISO 23247-1).
What is a Digital Twin?
IBM Technology
This video is hosted on YouTube. Playing it loads content from Google, which may set cookies.
Watch on YouTubeSecurity: read the plant, do not write to it
Our default design rule is simple: an operational twin reads plant data and does not write to controllers. If somebody wants to push a setpoint from the twin, that is a SCADA change and goes through SCADA's change control, not a feature of the 3D viewer. Data flows out through a DMZ or a historian replica, never through a new inbound connection to the control network.
NIST's report on security and trust considerations for digital twin technology, published in February 2025, sets out both the familiar cybersecurity issues and the new trust questions a twin raises, such as whether the model still reflects the real asset (NIST IR 8356). Use it alongside your plant's OT security architecture when you scope the twin.
What you get and how long it takes
- Digital twin platform with a 3D viewer in the browser, at 1:1 scale from your CAD.
- IoT integration layer connecting PLC, SCADA, sensor and ERP data.
- Analytics and reporting dashboard with what-if scenario simulation and anomaly detection.
- VR/AR visualization module for walkthroughs and training.
iUNI quotes a typical timeline of 12 to 24 weeks for a digital twin project, and publishes price ranges on its pricing page. The range depends mostly on how many data sources are integrated and how much of the plant is modeled, not on the 3D detail.
Questions to answer before you scope a twin
- What decision will it change? A twin that answers a named question, such as where micro-stops come from, gets used. One that shows everything gets admired once.
- Which tags already exist? Most of the data is usually in the PLC or historian already. Missing sensors cost more than modeling.
- Who owns tag names? If the PLC program renames a tag, the twin breaks. Agree a naming standard and a change process up front.
- Where does the data cross into IT? Decide on the historian, DMZ or broker before anyone builds a dashboard.
- Who keeps the model current? Layout changes, new machines and software updates must reach the twin, or it slowly stops matching the plant.
If those five answers are clear, the modeling work is the predictable part of the project. If they are not, no amount of 3D detail will rescue it.
How to start without boiling the ocean
A beverage plant in Colorado, for example, might start by mirroring the filler, labeler and case packer with their state and fault data. Within weeks the team can see micro-stops on the 3D line that never showed up in the daily report, then use the what-if model to test a different accumulation length before moving any conveyor.
Next step
Read more about our digital twin and virtual commissioning service, or start with four practical uses of digital twins in manufacturing. For the controls work underneath, see our controls engineering service. If you have a line and a question you want answered, contact us.
