Unity Digital Twin: How to Plan and Build an Industrial Twin

What a Unity digital twin is

A digital twin is a virtual representation of a real asset, such as a machine, a production line or a whole plant, that is connected to data from its physical counterpart. The ISO 23247 series describes a digital twin framework for manufacturing in exactly these terms: observable manufacturing elements, their digital representation and the data exchange between them (NIST).

A Unity digital twin is the interactive, real-time 3D layer of such a system. Unity turns CAD data and live or recorded machine data into something people can actually use: an operator dashboard in 3D, a VR walkthrough of a plant that does not exist yet, a training scenario on the real machine layout or a sales demo that shows the machine running. Unity is rarely the whole twin. It sits on top of your PLCs, MES, historian or IoT platform and makes their data visible and explorable.

Visualization, monitoring or simulation: three levels

Many projects fail because "digital twin" means something different to everyone in the room. It helps to agree on the level first:

• Level 1, visual twin: an accurate, interactive 3D model of the asset or plant, without live data. Useful for planning, layout reviews, sales and training. Fastest to build.
• Level 2, connected twin: the 3D model shows live or historical data, such as machine states, temperatures, throughput and alarms, directly on the affected component. Useful for monitoring, maintenance and remote support.
• Level 3, simulation twin: the model also predicts behaviour, for example material flow, cycle times, robot reach or what happens when a parameter changes. Useful for virtual commissioning and process optimisation.

Most companies get the best return by starting at level 1 or 2 for one clearly defined asset, proving the value and then extending. A level 3 twin of an entire factory as the first step is usually too expensive and too slow to show results.

Typical industrial use cases

• Operations and monitoring: see the state of a line in 3D, find the machine with the alarm and drill down to the sensor, instead of reading tables.
• Maintenance and remote support: show technicians where a component sits, how it is disassembled and what the data said before the failure.
• Training on the real layout: reuse the twin for VR or desktop training, so employees learn on the exact machine before it is installed. My VR safety training guide shows how this works for hazardous procedures.
• Planning and virtual commissioning: review layouts, check reach and collisions and test control logic before hardware is on site.
• Sales and trade fairs: a twin of a machine can become an interactive product presentation. The virtual wind tunnel on this website is a small example of real-time engineering visualization in the browser.

One anonymised engagement presented on this website, a production-facility digital twin, reports 25% lower unplanned downtime, 15% higher efficiency and an eight-month payback. Results like these depend on the baseline and how they are measured, so define your own measures before the project starts. My article on enterprise use cases and ROI explains how.

From CAD data to a real-time 3D model

Engineering CAD is built for manufacturing, not for real-time rendering. A single machine can contain thousands of parts, internal screws nobody will ever see and geometry far too detailed for a tablet or a headset. Preparing that data is often the biggest single task in a digital twin project.

Unity's own tool for this is the Asset Transformer Toolkit, formerly the Pixyz plugin. According to Unity it supports more than 70 CAD, BIM, mesh and point-cloud formats, keeps hierarchy and metadata and can decimate meshes and generate levels of detail. It is included in a Unity Industry subscription (Unity). Keeping the CAD structure and part IDs is important: it is what later lets a sensor value appear on exactly the right component.

The same preparation is reused for training, configurators and sales, which is why a well-prepared CAD pipeline pays off beyond one project. The 3D product configurator guide covers the same step from the sales side.

Connecting live data: OPC UA, MQTT and APIs

A connected twin needs a reliable data path from the shop floor to the 3D application. In industry this usually means one of these:

• OPC UA: the established standard for machine and controller data, with a rich information model and built-in security.
• MQTT: a lightweight publish/subscribe protocol, often used with an IoT platform or a unified namespace.
• REST or WebSocket APIs of your MES, historian, ERP or cloud IoT platform.

In most projects the Unity application does not talk to every PLC directly. A gateway or existing IoT platform collects, filters and secures the data, and Unity subscribes to the values it needs. This keeps OT security in the hands of your automation and IT teams, keeps the 3D app simple and makes it possible to replay historical data, which is very useful for root-cause analysis and training.

Where the twin runs: desktop, web, tablet and VR

One Unity project can be deployed to several platforms: a Windows control-room application, a browser version for wider access, a tablet app for technicians on the shop floor and a VR or mixed reality version for immersive reviews and training. With OpenXR the VR build can target Meta Quest, PICO, HTC VIVE and other headsets; my VR glasses for business guide compares them. Decide early which platform is primary, because it sets the performance budget for the 3D data.

How a project runs

1. Outcome and measures: which decision or problem should the twin improve, for example downtime, commissioning time or training time, and how you will measure it.
2. Data inventory: which CAD data, layouts and data sources exist, who owns them and how they can be accessed securely.
3. Validation prototype: one machine or cell, real CAD, a small set of real or recorded data and the target platform, tested with real users.
4. Production: full model, data integration, user roles, performance optimisation and testing.
5. Rollout and lifecycle: deployment, documentation and a clear process for updating the twin when the real plant changes.

Timeline and investment

A validation sprint, typically one asset or cell with real CAD and a first data connection, starts from EUR 9,500 and usually takes two to four weeks. A production digital twin application typically ranges from EUR 25,000 to 90,000 over eight to twenty-four weeks, depending on the size of the model, the number of data sources, platforms and integrations. Multi-site or enterprise programmes start from EUR 60,000. More detail is in How much does a Unity developer cost in 2026?.

When you compare providers, ask to see a twin they built running on the target device, ask how they handle CAD data and OT security and ask who keeps the model up to date after launch.

Frequently asked questions

What is a Unity digital twin?

A Unity digital twin is the real-time 3D, interactive layer of a digital twin: a model of a machine, line or plant built in Unity from CAD data and connected to live or historical machine data, so people can monitor, plan, train or sell with it on desktop, web, tablet or VR.

Is Unity good for industrial digital twins?

Yes, for the visualization and interaction layer. Unity offers CAD import through the Asset Transformer Toolkit (formerly Pixyz), real-time rendering on many platforms and VR/AR support. Data storage, analytics and control usually remain in existing industrial systems such as PLCs, MES or IoT platforms.

How does a Unity digital twin get live machine data?

Usually through OPC UA, MQTT or the REST or WebSocket APIs of an MES, historian or IoT platform. A gateway collects and secures the data, and the Unity application subscribes to the values it displays.

How much does a digital twin built with Unity cost?

As a guide from unity-freelancer.com: a validation sprint for one asset starts from EUR 9,500, a production digital twin application typically costs EUR 25,000 to 90,000, and multi-site or enterprise programmes start from EUR 60,000.

Can a digital twin be used for VR training?

Yes. The same 3D model and CAD preparation can be reused for VR or desktop training on the exact machine layout, often before the real equipment is installed.