A digital twin in manufacturing is a virtual model of a machine, line or plant that is kept in step with the real one through data, so you can test, plan, monitor or train on the model instead of the equipment. NIST describes digital twins as synchronized virtual models (NIST). The useful question is not whether you have a digital twin. It is which decision the twin helps you make faster or more safely.
Why the interest is real
Grand View Research valued the global digital twin market, across all industries, at about $35.8 billion in 2025, and forecasts growth of 31.1% a year from 2026 to 2033 (Grand View Research via State Journal, June 2026). Behind the market numbers is a simple cost: NIST, citing its own research, puts the cost of unplanned downtime in U.S. discrete manufacturing at about $245 billion (NIST digital twins). Tools that find problems before they reach the floor have an obvious target.
Four twins that pay back
| Twin | Question it answers | Typical buyer |
|---|---|---|
| Virtual commissioning twin | Does the control code work before startup? | Machine builders, integrators |
| Planning and layout twin | Will this line fit, flow and be safe to work on? | Plant engineering |
| Operational twin | What is happening now, and what if we change it? | Operations, maintenance, energy |
| Training twin | Can people practice without risk to equipment? | Training, safety, HR |
1. Virtual commissioning
The most proven use. The PLC program runs against the model and errors are found before site. We cover it in detail in virtual commissioning.
2. Planning and layout
A 1:1 model of a new line inside the existing building, walked through in VR, shows clashes, access problems and awkward reaches before money is spent on steel. For brownfield expansions it is often the cheapest twin with the fastest payback. See brownfield line integration.
3. Operations
An operational twin connects the model to live data from PLCs, SCADA and sensors, with a 3D view, analytics, anomaly detection and what-if simulation. It helps when many people need to understand a complex system, for example a large warehouse or a multi-line packaging hall, and when changes can be tested on the model first.
4. Training
The same model, put in VR, lets new staff practice startups, changeovers and fault finding. Read more in VR training for manufacturing.
The digital twin: Transforming factories into adaptable systems
Siemens Software
This video is hosted on YouTube. Playing it loads content from Google, which may set cookies.
Watch on YouTubeBuild once, use four times
The strongest business case is reuse. Our iUNI team builds twins from CAD at 1:1 scale, connected to live data, for planning, training, monitoring and virtual commissioning (iUNI digital twins). A model built to commission a line becomes the training simulator before startup and the operational twin afterwards. Each step adds data connections and views, not a new model.
Scale matters too. ISA's InTech reported that Unilever built twins for dozens of its roughly 300 plants, at three to four weeks per twin (ISA InTech, August 2022). Speed like that comes from templates and standard data models, not from building every twin from scratch.
What a twin needs from your plant
Most twins stall on data, not on 3D. Before starting, check what you have. CAD for the equipment, ideally native or STEP files rather than PDFs. Up-to-date PLC programs with readable tag names. A SCADA system or historian that can share data over OPC UA or a database. And someone in the plant who owns the twin once the project team leaves.
Budget for the data work honestly. Mapping a few hundred PLC tags to the right objects in a model, cleaning up inconsistent names and agreeing units takes longer than most teams expect, and it is the part that decides whether anyone trusts the twin. Missing pieces are not a reason to wait. Layout twins can start from point clouds and simple models. Operational twins can start with a few dozen tags from one line. The point is to know the gaps at the start, so the scope and the budget reflect them.
Standards and openness
ISO 23247, published in 2021, sets out a digital twin framework for manufacturing, and NIST is proposing a verification, validation and uncertainty quantification guideline as a new part of it (NIST). For buyers, the practical points are simpler: insist that the model and its data connections use open formats and interfaces such as OPC UA, and that you own the model at the end of the project.
How to start small
- Pick one question with a cost attached: a startup you can't afford to delay, a layout you aren't sure of, a recurring fault nobody can explain.
- Pick one asset: a cell or a line section, not the whole plant.
- Use what you have: CAD, PLC tags, existing SCADA data.
- Define success in hours, dollars or incidents before you start.
- Plan for reuse: keep the model structured so it can grow into the next use.
We build virtual commissioning, planning, operational and training twins for U.S. plants in pharma, semiconductors, packaging and warehousing. See our digital twin service or contact us with the question you want a twin to answer.
