The first idea of Twin appeared at NASA with its Apollo program. At that time, it was a real copy of the module where it was possible to reproduce the problems encountered in space by the module and find a way to solve them from the earth. Grieves introduced the transformation into a Digital Twin in 2003 as mentioned in “Digital Twin: Manufacturing Excellence through Virtual Factory Replication”. NASA and the US Air Force then started to apply the Digital Twin in the aeronautic domain.
Since then, the concept of Digital Twin continuously evolved, passing from a three-dimensional Framework to a five-dimensional Framework. The 3Dimensional Framework was composed of the Physical Entity, that is the real physical object; the Virtual Entity consisting of a set of models, that are geometry model, physics model, behaviour model, and rules model; the Connections of data and information that bind the physical and Virtual Entity together.
Figure 1: 3Dimensional Framework
5Dimensional Framework integrates two new components: DT Data and Services. DT Data contains data from all entity, fuses these data via different algorithms, and make these data available for all the other entity. Services encapsulate all functions of the Digital Twin into standardized services with user-friendly interfaces for easy and on-demand usage.
Figure 2: 5Dimensional Framework
Digital Twins is classified on a different level, depending on the application, between unit-level Digital Twin, system-level Digital Twin, and system-of-systems level Digital Twin. The unit-level Digital Twin is the minimal independent individual: a person, a machine, or a product. If multiple units-level Digital Twins are connected, they form a system-level Digital Twin and provides more functionalities, like scheduling and optimization services. Finally, the system-of-systems level Digital Twin is a grouping of system-level that gives more functionalities, more precise information and increase the available data due to the growth of Physical and Virtual Entity.
Figure 3: levels of Digital Twin
The Benefits of Digital Twin
During the design phase, the Digital Twin helps by simulating the behaviour of the product as precisely as possible in different environments or by simulating many different designs in a specific environment. These simulations find the most suitable design without building a physical prototype, and that makes the development process faster and cheaper.
During the use phase, the Digital Twin helps to optimize the Physical Entity by continuously making simulations to find the best parameter for the actual environment and task using the real-time data from the Physical Entity.
The Digital Twin allows remote maintenance - the expert investigates the problem directly in the Digital Twin and then teaches remotely to a field agent how to solve that problem. Before going to the field to solve the problem, the field agent practices on the Digital Twin using augmented reality. No physical contact between the expert and the field agent is needed, and it reduces the expense of travelling.
The Digital Twin allows to make predictive maintenance, as the Digital Twin can compare the real-time data with the historical data considering the wear and the environment of the Physical Entity, it can predict the best time to do maintenance. Or if the component is part of a system of system Digital Twin, it can compare the real-time data of the Physical Entity with data of other Digital Twins that have similar wear and environment.
In brief, the benefits of the Digital Twin are:
- Cheapest conception and maintenance
- Fastest conception and problem detection
- Continuous optimization
- Predictive maintenance
Implementation of the Virtual Entity
The Virtual Entity of a Digital Twin is composed of a set of models. The number and type of implemented models depend on the usage of the Digital Twin. The first model is the Geometry model. It represents the model in a 3D space, indicating the physical object like its shape, its material, or assembly of parts. The second model is the physics model representing the physical property such as velocity and force. This model also reflects the physical phenomena such as deformation and corrosion. The third model, the behaviour model, describes the responding mechanism and behaviour of the entity, such as the performance degradation under external disturbance. Finally, the rules model is a set of rules build from historical data of the Physical Entity using a different technique like machine learning. This model gives the Virtual Entity the ability of optimization and prediction over the Physical Entity.
The data of the Physical Entity need to be fed to the Virtual Entity to continuously make the Virtual Entity match as best as possible to the Physical Entity and allow the Virtual Entity to give better optimization and prediction to the Physical Entity.
(2019) Digital Twin Driven Smart Manufacturing Fei Tao, Meng Zhang, A.Y.C. Nee, Elsevier Inc.
(2015) Digital Twin: Manufacturing Excellence through Virtual Factory Replication, Michael Grieves