A digital twin is a virtual model of a physical asset, process, or system that is kept in sync with its real counterpart through live data. It is used to monitor, diagnose, predict, and test changes before making them on the real equipment.
A digital twin combines three things: a model of the physical asset (geometry, physics, or a data-driven model), a live data connection (sensors, PLC tags, historian data), and software that uses the model to answer questions. Typical questions: what is the current state, why did it drift, what happens if we change a setpoint.
Twins exist at several scales. A single pump or spindle. A production line. A whole plant or supply chain. The model can be physics-based, statistical, or a mix of both.
The term is used loosely. A 3D CAD model or a one-off simulation is not a twin unless it is updated from the real asset. Some definitions separate a digital shadow (data flows one way, from asset to model) from a full twin (the model can also send changes back). A twin is only as good as its data and its model. Keeping both current as equipment is modified is often the hardest part.
Questions about Digital twin
Is a 3D CAD model or a simulation a digital twin?
No, not unless it is updated from the real asset. What makes a twin is that live data keeps it in sync with its physical counterpart.
What is the difference between a digital twin and a digital shadow?
Some definitions separate them by the direction of the data. In a digital shadow, data flows one way, from the asset to the model. In a full twin, the model can also send changes back.
What do you need to build a digital twin?
Three things: a model of the asset, which can be geometry, physics or data-driven; a live data connection such as sensors, PLC tags or historian data; and software that uses the model to answer questions. Keeping the model and data current as equipment is modified is often the hardest part.
Digital twin in real plants
- Kansai Electric Power and Cognite: Kansai Electric Power adopted Cognite as the integrated data platform for its hydro power business under its Next Generation Hydro Reborn Project.
- Dangote Petroleum Refinery and Honeywell: Honeywell Performance+ Services, delivered through Honeywell Forge, will be deployed on core units of the Dangote refinery in Lekki, with operator training simulators built on digital twins of the facilities.
- Xiang Piao Piao and ABB: Xiang Piao Piao runs ABB FlexPicker IRB 360 robots with AI-based vision and PickMaster Twin digital twin software to pick and place milk tea ingredients at its Huzhou, Chengdu and Tianjin factories.
- PepsiCo and Siemens: PepsiCo is building digital twins of U.S. manufacturing and warehouse facilities with Siemens Digital Twin Composer to test and validate new line configurations before changing the plant.
- BPMI and Gecko Robotics: BPMI and its suppliers are using Gecko robotic and AI platforms to inspect forged components for U.S. Navy nuclear propulsion and to build digital twins of critical components.
Vendors linked to Digital twin
Mujin (2) · ABB (1) · AVEVA (1) · Cognite (1) · Gecko Robotics (1) · Honeywell (1) · KUKA (1) · SICK (1) · Siemens (1) · Sight Machine (1) · Yokogawa Electric (1)
Digital twin in the digest
Related terms
Digital thread
A digital thread is the connected flow of data that links a product's information across its lifecycle, from design and engineering through manufacturing, inspection and service, so each stage can use and feed back into the others.
Process historian
A process historian is a time-series database that collects, compresses and stores data from plant control systems, such as sensor readings, setpoints and alarms, so it can be trended, analyzed and retrieved later.
Predictive maintenance
Predictive maintenance is maintenance triggered by a prediction of failure, based on measured equipment condition such as vibration, temperature, or noise.
Synthetic data
Synthetic data is artificially generated data that mimics some statistical properties of real data.
Industry 4.0
Industry 4.0 is the umbrella term for connecting and computerizing manufacturing through cyber-physical systems, networked sensors, and data analysis.