Virtual replicas of devices, machines or plants may be used to carry out accurate simulations before the physical realization. This useful match allows to solve design and functioning issues, and more,
by Valerio Alessandroni
The Digital Twin is a real-time digital replica of a product, process or system, which may be used for testing, diagnostics and analysis operations even before the object represented is physically created. Digital twins therefore combine data analysis, artificial intelligence, automatic learning and simulation techniques, and they are normally used for the modeling of Internet of Things (IoT) and of product life-cycle management (PLM) systems.
On the other hand, the amount of data collected by the sensors present on a machine or industrial plant is enormous, so much so that they are considered Big Data, but if these data are not aggregated ad organized so as to favour the decision process (thereby becoming information) they are useless. The combination of the physical and virtual world is very useful, allowing to solve by means of simulations design or functioning issues, preventing machine or plant downtimes and planning the product life-cycle better.
The concept of digital twins is not new but, under the patronage of Industry 4.0, it is emerging with new energy. For the record, the concept of Digital Twin has been introduced for the first time 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 changed about ten years ago, with the development of IoT, artificial intelligence, Big Data and cloud computing.
Integrating sensors to collect all the different data in real time
To obtain a Digital Twin, first of all, sensors should be integrated in the objects concerned to collect in real time data on their state, the conditions of their functioning, their physical position and so on. This is the connection with the Internet of Things. “Smart” objects are then connected to a cloud-based system which receives and processes all the data, allowing to carry out analyses bases on specific requirements or based on other data, such as historical data. In this way, in the virtual environment it is possible to draw conclusions or to discover and analyze opportunities which it will then be possible to apply to the physical “twin”. For instance, it will be possible to modify a project in order to avoid critical aspects detected on the virtual “twin”, or to optimize some particular production activity having simulated the results. The digital twin also allows to create an ideal maintenance situation, whereby a local technician and a remote specialist, connected via the Internet, may access the same data and discuss what should be done.
The virtual prototype is updated to reflect every change in its twin
It is interesting to note that the virtual prototype of an object is “alive” and dynamic, which means, it is updated every time its physical twin undergoes any changes. It is also capable of learning, by absorbing the knowledge of persons, machines and the environment where it is found.
In greater detail, the model of the Digital Twin as defined by Grieves consists of three main parts: physical products in a real space, virtual products in a virtual space and connected data which link the physical and virtual products to one another. Lastly, digital twins must satisfy three requisites: they must look exactly like the original object, including all minor details; they must behave in the same way as the original object during tests; they must be capable of analyzing information regarding the original object, foresee possible problems and recommend solutions.
Three types of Digital Twin
So-called Product Twins are models of specific products, used before setting up a production line to analyze their behaviour in different conditions and the problems which might arise. As a consequence, digital product twins help in reducing production costs and time-to-market, improving quality. Afterwards, product twins may be used to check the product’s performances in the physical world.
Process Twins on the other hand simulate production processes. A virtual production process allows to create different scenarios and to show what will happen in different situations, enabling the development of the most efficient product methodology. The process may be further optimized with the use of product twins for every machine involved, thereby allowing to carry out predictive maintenance and avoid costly plant downtimes. Production operations will be safer, faster and more efficient. System Twins, lastly, are virtual models of the entire system, such as, a plant or factory. They collect enormous amounts of operative data produced within the system, obtain information and create new business opportunities to optimize all the processes. But let us see some real examples.
From space to constructions
With the aim of using, maintaining or repairing systems which are physically not accessible, the NASA was one of the first to try out a simulation technology similar to what we now call Digital Twins, in the early days of space exploration. And when disaster struck on Apollo 13, it was this technology which saved the mission, using virtual systems on Earth.
Nowadays, digital twins may be used not only in the space sector, but also in production, in the energy industry, in transportation and constructions.
Complex objects such as aircraft engines, trains, offshore platforms and turbines may be designed and tested digitally before being physically produced.
A very significant example is the one which, in Milan, may be considered as a divide in the field of design, maintenance and management of civil works and buildings: the dynamic digital copy of of Milan’s Central Station was created using about 6,000 photographs.
The Digital Twin technology makes it easier to interact with the building for the most diverse purposes: testing new architectural solutions or verifying safety plans, planning and carrying out inspections to the plants, or storing information necessary for the management and maintenance of the single components.
Other interesting application examples
Even in the wood industry, Cloud-based Digital Twins of machines and tools are proving very useful for more efficient machining processes, a faster preparation of orders and the documentation of the life-cycle of machines.
One last example: a world leader in the design and production of pumps and systems is broadening the use of the ANSYS simulation software to exploit the power of the Internet of Things, creating a digital version of its products. The company will use Digital Twins to provide its clients with improved product quality and performances, greater development capacity, optimized maintenance and reduction of the costs and risks associated with unplanned inactivity periods. It will also be possible to analyze the performances of products in real operating conditions, and, based on these data, to foresee future performances.
Marching in step with Industry 4.0
To conclude, the way of managing an enterprise and above all of designing, producing and managing products is changing substantially, thanks to the support of technologies which fall within digitization and Industry 4.0. The physical world is transposed into a world made of Digital Twins, which, by means of modeling and big data, allow us to understand which are the areas (product, process, supply chain, business model) where action should be taken to improve functional and operational efficiency. Industry 4.0 and Digital Twins are therefore two expressions destined to march in step, so as to establish a continuity between design and production.