Digital twin tech is beginning to see more use cases across a variety of industries today. A digital twin, which is basically a virtual replica of a physical system, process, or product, provides a real-time look at how a physical asset is performing.

Initially designed for manufacturing, the technology is now being applied in more industries worldwide. In the automotive and construction industry, a digital twin is used to design and test concept models, be they vehicles or architecture designs.

In the oil and gas industry, digital twinning enables fuel companies to have a virtual replica of drills and other components on oil rigs. Sensors installed on the drills and rigs transmit data to the digital twin, which in turn indicates the specific problem areas. The process has enabled the industry to speed up repair and maintenance work on oil pipelines, rigs, and deep-sea drills.

The digital twin market size is expected to reach US$ 86.09 billion by 2028, according to a study by Grand View Research. Apart from the pandemic expediting the adoption of the technology, the ability for it to be managed by a smaller workforce is also a contributing factor as to why digital twins are growing in demand.

While Europe accounted for more than 30% of the overall digital twin market last year, other regions around the world are also showing more interest in the tech. With AI, machine learning, and IoT enabling more automation use cases, more organizations are looking to implement digital twins to enhance their productivity and reduce time to market as well.

More recently, end-use industries are increasingly demanding digital twin portfolios capable of addressing specific requirements. Leading market players are responding to this growing ask by offering different digital twin solutions tailored for specific end-use industries such as in healthcare.

According to Asha Poulose Johnson, VP and Global CIO Data Analytics at GE Healthcare, digital twins are transforming industrial operations and are now starting to find applications in domains like personalized medicine. Infusing AI-based learning models is going to be critical for digital twins to provide scalable results with consistent accuracy, requiring several advanced technology components to come together.

“Digital twins are very personalized. It learns continuously from other twins and the physical factors around it. Highly scalable, it needs to be able to adapt to the environment it’s in. For example, in healthcare, the digital twin needs to adapt to the use cases of that specific hospital requirements. Digital twins can be created to mimic the physical asset, providing users visibility without needing to be present on-site,” noted Johnson in her presentation at the AI Accelerator Summit.

Digital twin use cases 

The three main use cases for a digital twin tech are for early warning, continuous prediction, and dynamic optimization. With these three use cases, the digital twin can be customized to provide the best performance on more detailed industry-specific use cases.

As such, applying digital twin to use cases would mean infusing it with AI and machine learning algorithms. Machine learning tests new features as well as self-learning to understand processes, enabling the digital twin to obtain more data. Machine learning forms a subset of AI, which is a real part of how machine learning and digital twins feed into each other.

When it comes to AI, digital twins learn from the AI models, which includes learning from peers, learning from solutions, learning from self, and learning from humans. The digital twin becomes a comprehensive learning system across the board.

A continuous learning system 

Johnson pointed out that digital twin tech is basically a learning system that produces a specific outcome for a specific use case, based on the three aforementioned main use cases. For early warning use cases, digital twin in healthcare is used on a CT machine for example. The digital twin looks at the operational data and provides warning alerts, up to 2 weeks in advance, on any problems in the machine such as an impending CT tube issue. This can improve healthcare center performance, especially in maintaining hospital machines.

For optimization use cases, digital twins can be harnessed on locomotives for example. The digital twin can track locomotive topological data and give input on how fuel saving can be optimized, as well as how carbon emissions can be reduced.

Digital twins for prediction use cases can be utilized within industries that use steam turbines. The digital twin boiler applies erosion physics and machine learning along with design, operational and inspection data to understand boiler tube thickness, and how to conduct maintenance.

With more use cases being developed for digital twins, Asha believes that organizations need to be certain of how they can apply digital twin tech in their workloads, ensuring they have the right infrastructure in place to support it.

For Asha, the important thing from a technology perspective for businesses is to invest in the architecture, be it in a very distributed on cloud or on-premise, to having strong foundational ways to improvise data and creating templates that are industrialize and domain specific to partner with AI and ML models to generate the desired outcome with the digital twin.