In the following article, Valerio Alessandroni, Technical Director at EFCC, Professor at Tallinn Technical University and member of the faculty board of the Zigurat’s Global MBA in Digital Business, takes an in-depth look at Digital Twins and delves into how these virtual replicas of devices, machines or plants can be used to perform in-depth simulations prior to physical implementation.The Digital Twin is a real-time digital replica of a product, process or system that can be used for testing, diagnostics and analysis even before physically creating the object represented. Digital twins therefore combine techniques of data analysis, artificial intelligence, machine learning and simulation and are commonly used for modeling Internet of Things (IoT) systems and product lifecycle management (PLM).On the other hand, the amount of data collected by the sensors present in a machine or an industrial plant is huge - so much so that we talk about Big Data, but if this data is not aggregated and organized in such a way as to facilitate the decision-making process (becoming therefore information), their usefulness is null.The combination of physical and virtual worlds is very useful, allowing you to solve design or operating problems through simulations, prevent machine or plant downtime and better plan the product life cycle.The concept of digital twin is not new but, under the umbrella of Industry 4.0, it is emerging with new vigor. For the record, the concept of Digital Twins was first introduced and clearly defined by Michael Grieves in 2003, during a conference at the University of Michigan. However, at the time it was difficult and expensive to implement the concept and make it available for general use. The situation has changed for about ten years, with the development of the IoT, artificial intelligence, Big Data and cloud computing.To obtain a Digital Twin, first of all it is necessary to integrate sensors in the affected objects to collect in real time data on their status, operating conditions, physical position and so on. This, in fact, is the connection with the Internet of Things. The 'intelligent' objects are then connected to a cloud-based system that receives and processes all data, allowing you to perform analyzes based on specific needs or on the basis of other data, for example historical data.In this way, in the virtual environment it is possible to draw conclusions or discover and analyze opportunities that can subsequently be applied to the physical 'twin'. For example, it will be possible to modify a project to avoid the criticalities identified on the virtual 'twin', or it will be possible to optimize certain production activities by having simulated the results. The digital twin also allows you to create an ideal maintenance situation, where an on-site technician and a remote specialist, connected via the internet, can access the same data and discuss what to do.It is interesting to observe that the virtual prototype of an object is 'alive' and dynamic, which means that it is updated every time its physical twin undergoes changes. It is also able to learn, absorbing the knowledge of people, machines and the environment in which it exists.More specifically, Grieves' model of Digital Twins consists of three main parts: physical products in real space, virtual products in virtual space, and connected data that bind physical and virtual products together.Finally, digital twins must meet three requirements: they must look identical to the original object, including all minor details; behave the same way as the original object during testing; be able to analyze information about the original object, predict possible problems and suggest solutions.The so-called Product Twins are models of specific products, used before setting up a production line to analyze their behavior under different conditions and the problems that could occur. As a result, digital product twins help reduce production expenses and time-to-market while improving quality. Thereafter, the product twins can be used to check the performance of the product in the physical world.The Process Twins models instead simulate the production processes. A virtual production process allows you to create different scenarios and show what will happen in different situations, allowing you to develop the most efficient production methodology.The process can be further optimized with the help of product twins for each piece of equipment involved, allowing for preventive maintenance and avoiding costly plant downtime. Production operations will be safer, faster and more efficient.Finally, System Twins are virtual models of the entire system, such as a plant or a factory. They collect huge amounts of operational data produced within the system, acquire information and create new business opportunities to optimize all processes.But let's see some real examples.With the goal of using, maintaining or repairing physically inaccessible systems, NASA was among the first to try simulation technology similar to what we now call Digital Twins since the early days of space exploration. And when the disaster struck Apollo 13, it was Digital Twin technology that made it possible to save the mission, using virtual systems on Earth.Today, digital twins can be used not only in space, but also in manufacturing, energy, transportation and construction. Complex objects such as aircraft engines, trains, offshore platforms and turbines can be digitally designed and tested before being physically produced. A very significant example is what, in Milan, can be considered a 'watershed' in the field of design, maintenance and management of buildings and / or civil works: the dynamic digital copy of the Milan Central station that was created using about 6000 photographs. Thanks to Digital Twin technology it becomes easier to interact with the building and in the building for the most different purposes: to test new architectural solutions or verify safety plans, plan and carry out plant inspections, or archive the information necessary for the management and maintenance of individual components.Also in the field of woodworking machines, the digital twins of Cloud-based machines and tools are proving very useful for more efficient machining processes, quick order preparation and documentation of the machine life cycle.A final example: a world leader in the design and manufacture of pumps and systems, is expanding the use of ANSYS simulation software to harness the power of the Internet of Things and create a digital version of its products. The company will use Digital Twins to offer its customers enhanced product quality and performance, increased development capacity, optimized maintenance and reduced costs and risks associated with unplanned downtime. It will also be possible to analyze the performance of the products in real operating conditions, and on the basis of this data predict future performance.In conclusion, the way of doing business and above all the ways of designing, producing and managing the products made are changing substantially thanks to the support of technologies that refer to digitization and Industry 4.0. The physical world is transposed into a world made up of digital twins which, through modeling and big data, allow us to understand which areas (product, process, supply chain, business model) can be used to improve functional and operational efficiency.Industry 4.0 and Digital Twin are therefore two expressions destined to go hand in hand, to establish continuity between design and production.