A digital twin is a virtual replica of a physical object, machine, or system that is kept in sync with its real counterpart through a stream of sensor and operational data. Rather than a static 3D model, a digital twin is a living simulation: as the physical asset runs, wears, and ages, the twin updates so it can mirror the asset’s current state and predict how it will behave under different conditions.
The idea was formalised by Michael Grieves around 2002 in product-lifecycle management and later named “digital twin”; the definition most cited in engineering comes from NASA’s 2012 paper by Glaessgen and Stargel, which described a twin that integrates high-fidelity simulation with onboard health monitoring, maintenance history, and fleet data to mirror the life of a physical vehicle. In practice a twin combines physics-based models, real-time IoT sensor feeds, and machine learning to answer questions like when a part will fail, how a change to operating conditions will affect output, or how to schedule maintenance.
Digital twins now appear across manufacturing, energy, aerospace, construction, and even city planning. For a business reader, the value is the ability to test decisions and anticipate failures on a virtual copy before acting on the expensive, hard-to-reverse physical system.