Digital twin technology for industrial transformationFollow article
Digital twins: the birth of a new era in manufacturing, but what is a digital twin?
A digital twin can create a highly accurate digital replica of physical assets and deliver enormous advantage to manufacturers, right across the value chain.
Modern manufacturing facilities have undergone a digital revolution in recent years. The convergence of Industry 4.0-enabling technologies such as automation, wireless connectivity, and artificial intelligence offers an ability to continuously extract valuable data, providing real-time insights into production activities. This information is being used by engineers in increasingly imaginative ways, enabling them to deliver operational performance improvements right across the shop floor.
These underlying technologies having driven incredibly rapid advances with digital twins – the concept of creating computerised representations of physical assets. These virtual replicas of objects or processes can be updated in real time using data from sensors, enabling engineers to perform highly accurate simulations of even the smallest variances. This kind of action can allow finely tuned calibration across the manufacturing value-chain – deliver extremely valuable outcomes in areas such as products and production, through to predictive maintenance.
Indeed, as manufacturers become more confident with applying Industry 4.0 methodologies, and the collection and streaming of data continue to get cheaper, so the demand for digital twins has increased. The market was worth around $3.8 billion in 2019 but is expected to reach $35.8 billion by 2025, according to a report published by Research and Markets. This represents a compound annual growth rate of 37.8%. Digital twins are clearly here to stay and will find widespread adoption across various manufacturing sectors including automotive, aerospace and electronics.
Deriving value from multiple data sources
So, let us now look a little deeper at the concept of digital twins to understand how they have evolved, before outlining some specific ways in which they enhance the manufacturing process in areas such as overall equipment effectiveness and production efficiency optimisation. In terms of historical development, engineers have used software to simulate the performance of products and processes for a long time. But the creation of a digital twin goes far beyond that, using data gathered from sensors incorporated into the physical twin to develop an identical computerised replica. The data is used to build an extremely faithful representation of each asset – whether that is an individual product or an entire production line. Therefore, the model is always based on factual information, rather than being a simulation of behaviour based on the perceived operation.
Indeed, the digital twin can be built from multiple data sources, including historical performance statistics, real-time sensors and manufacturing outputs, and future data provided by machine learning. It is all about creating the most accurate representation possible, with continuous refinement giving engineers a real understanding of the physical counterpart’s characteristics and even its performance over time.
Application across the value chain
But how can that information be applied to derive meaningful value at different stages of the production cycle, and beyond? At production planning, the digital twin can be a hugely valuable means of optimising assembly line layouts and how components flow around the factory floor. And having established a single model of truth for production, the digital twin can be used to monitor the performance, effectiveness, and quality of the manufacturing lines, flagging up any intolerable variances in output by comparing the digital asset with the finished product.
Maintenance is another area of potential value. With information related to the operation of key machinery built into the digital twin, maintenance engineers can capture data from sensors on a wide range of parameters including speed, vibration, temperature, and humidity to predict when equipment might be set to fail. This enables maintenance teams to carry out any rectifications or repairs before unplanned downtime occurs.
The way that maintenance is performed could also be impacted, bringing advanced technologies such as augmented reality (AR) to the fore. For instance, a maintenance team could attend a customer site, with headsets providing an AR view of components and systems incorporated within a wider piece of equipment. The maintenance operative can access a mine of relevant information – from usage data to maintenance history – in real-time, with drawings and specs delivered in their line-of-sight.
But digital twins are not restricted to the factory floor. It is a concept that lends itself to the demand for customised products, helping manufacturers loop customer input back into the design process. Data from existing equipment used out in the field can be relayed back to design teams, enabling them to prototype potential improvements and new ideas. Any custom configurations can be assessed against the digital twin, with usage data allowing engineers to measure the impact on product performance.
Deploying digital twins in retrofit environments
These are some of the advantages derived from digital twins. But how do manufacturers go about creating such multi-layered models? Deployment scenarios are relatively straightforward when starting with a blank sheet of paper at a greenfield site. However, existing production facilities with legacy systems, which are often siloed, require a different approach based upon retrofit and effective assimilation of the collected data.
Here, sensors and actuators provide the first touchpoints to the physical asset, enabling the collection and streaming of data. But that is just the start. Digital twins are often built from multiple streams of information, including enterprise systems such as CAD and from other divisions such as logistics and supply. These are brought together and fed into the modelling software to create the first iteration. The key is to start small – maybe with an individual part within an asset - and build things up over time. This is often done by bringing many smaller twins together to produce a representation of an entire machine or a production process. Security is also a vital consideration, both in terms of device identification and who within an organisation has access to the twin, and at what level.
Widespread adoption, from aerospace to automotive
When done correctly, digital twins can become a mainstay of design, production, and maintenance activities, as recognised by some of the world's biggest manufacturers. For instance, Boeing uses digital twins to replicate a broad range of physical parts and systems and is extending the use throughout its design, engineering, and development processes.
In the automotive sector, meanwhile, digital twins are being employed to drive the trend towards connectivity and electrification, as engineers design new architectures and production layouts within their plants. This will allow them to speed up the development of new powertrains and deliver more efficient manufacturing processes as new platforms come on stream.
And in the electronics sector, digital twins are being used to provide a digital replica of semiconductor circuit performance, as well as manufacturability and integration into components and hardware.
Delivering more optimised operations
To conclude, digital twins are emerging as a central tenet of the Industry 4.0 revolution. Ultimately, they allow companies to ‘sweat the assets’ – whether that be in design, production, operation, and maintenance, or all of these areas combined. They give engineers a clarity of insight that simply is not possible with the disparate data flows usually associated with manufacturing environments, providing a constantly updated ‘source of truth’ from which to make informed decisions.
In short, digital twins can provide a solid bedrock for more streamlined operations, ushering in an exciting new era for manufacturing optimisation.