Digital Twins, From Jet Engines to Whole Cities
A digital twin is a virtual model of a real object, system or place that is kept up to date with data from its physical counterpart. Unlike a one-off computer model, a twin changes as the real thing changes, so engineers can monitor it, test "what if" scenarios and predict problems before they happen. The term was popularized by Michael Grieves, a product lifecycle researcher, in the early 2000s, and NASA engineers have long described their spacecraft simulators in similar terms.
Jet engines
Aircraft engine makers such as GE Aerospace and Rolls-Royce collect sensor data from engines in service, including temperatures, pressures and vibration. Comparing that data with a model of how each engine should behave helps predict when parts will need maintenance, so airlines can plan repairs instead of reacting to failures.
Factories
Manufacturers build virtual versions of production lines to plan layouts and test changes before moving any machines. BMW, for example, has described using a digital twin of entire factories, built with NVIDIA's Omniverse platform, to plan new production lines. Engineers can check whether robots and workers will have enough space, or how a new car model will flow through the line, without stopping production.
Cities and buildings
Singapore built "Virtual Singapore," a detailed 3D model of the city-state that planners use to study things like how new buildings affect sunlight, wind and crowd flow, and where solar panels would work best. Building managers use smaller twins to track energy use, heating and ventilation across a building and spot waste.
Motorsport
Formula 1 teams rely heavily on simulation, partly because rules limit how much they can test on track and in wind tunnels. Detailed car models, fed with data from real sessions, are used to set up cars for each circuit and to run race strategy scenarios.
Healthcare
Medical digital twins are at an earlier stage. Dassault Systèmes' Living Heart Project created a detailed simulated human heart that researchers and device makers use to study how devices such as heart valves might behave. The longer-term goal is a model of an individual patient that doctors could use to test treatments virtually, but that remains largely in research.
Energy and utilities
Wind farm operators use twins of individual turbines and entire farms to predict output, spot underperforming turbines and plan maintenance during low-wind periods. Water utilities model their pipe networks to find leaks, which can account for a large share of treated water lost before it reaches customers. Electricity grid operators use network models to test how adding solar farms, batteries or electric vehicle charging will affect the grid before connecting them.
Ports and supply chains
The Port of Rotterdam, Europe's largest port, has developed a digital twin that combines data on water depth, tides, weather and ship movements to plan when vessels can safely arrive and berth. Logistics companies use similar models of warehouses and distribution networks to test how changes, such as a new depot or a surge in orders, would affect delivery times before making them.
How a twin differs from a simulation
The words are often used interchangeably, but there is a useful distinction. A simulation models how something could behave under conditions you choose. A digital twin is tied to one specific real-world object, updated with live data, and used across its working life. In practice, twins run many simulations, but the link to real, continuously updated data is what makes them twins.
What it takes to build one
- Sensors on the real object or system to collect data.
- A reliable connection to send that data regularly.
- A model that represents the physics or behavior accurately enough to be useful.
- Software to compare real and expected behavior and flag differences.
Limitations
A twin is only as good as its data and model. Missing sensors, inaccurate models or poor data quality lead to misleading predictions. Building and maintaining detailed twins is expensive, so they're mostly used where failures or downtime are very costly. The term is also used loosely in marketing, sometimes for ordinary 3D models with no live data, so it's worth asking what a particular "digital twin" actually updates and how often.
Why it matters to consumers
Most people won't interact with a digital twin directly, but they benefit indirectly: fewer flight delays from unplanned engine maintenance, buildings that use less energy, and products tested more thoroughly before they're made.