Digital Twins, ready for adoption

This underhyped industry approaches its inflection point

The Digital Twin concept 

As the world undergoes an ever-accelerating digitalization process, even physical assets (like a building or a plant) are increasingly intertwined with and, more importantly, managed through digital controls and processes. It becomes crucial, therefore, to integrate the two worlds across the entire lifespan of the asset, from its design to its decommissioning.

And here is where the concept of a “Digital Twin” takes all its relevance. Indeed, by creating an exact virtual replica of an asset, it becomes possible to better design, test, and control all relevant aspects across the various stages of its life.

On one side, this translates into avoiding decision-making based on incomplete (at best) or incorrect (at worst) data throughout the asset’s lifecycle, which eventually compounds negatively over time. On the other side, this also allows for experimentation and testing of multiple scenarios and options, enabling the best choices to be made in a consistent way.

The full benefits of digital twin technology, such as operational gains, predictive maintenance, and improved product design all boil down, in the end, to lower total life costs for any given asset, and significantly improved ROIs.

Reaching an inflection point

The concept is not entirely novel and was touted a few years ago, only for hyped expectations to face the reality of slower adoption. Nevertheless, as technologies have improved and digitalization pervasiveness has increased, the industry looks on the verge of exiting the current trough of disillusionment.

The current economic juncture is particularly favorable for the broader adoption of the Digital Twin concept. A dire global business outlook is forcing companies to significantly increase efficiency, an unescapable path that will rely on digitalization and automation. After the last two years, where most CapEx was put on hold or deprioritized versus infrastructure and work-from-home spending, there is an impellent need to catch up. As a cherry on the cake, the de-globalization process and its consequent production capacity re-shoring are likely to drive a renewed investment cycle. Also, the investment necessary to put in place Digital Twins build on each other, making the upfront cost more manageable: most companies have already started, and are looking to ramp up spending on the technology.  

All in all, most indicators are flashing green for a technology that is maturing into its commercialization stage.

Despite encompassing multiple segments and actors, the overall Digital Twins industry is estimated by CIMdata (a specialized industry consultancy) as having a ~$70bn 2026e TAM with a 10% 5Y CAGR. The key subsegments bound to show the strongest growth are software-related, notably architecture, engineering, construction (or AEC), simulation, and 3D visualization.

Simplifying complexity

Complexity is a key driver

The greater the level of complexity the more likely decision-makers will lose visibility about the implications of their decisions down the road. Uncertainty breeds more safeguards and ultimately higher costs. 

This general truth is particularly relevant for complex physical “assets” like a production plant or even an office building. In this context, a Digital Twin allows for experimentation and testing at full-scale different scenarios, with the consequences of a given choice playing out across the whole system and not only within the domain of concern of the decision maker.

Consider an office building: it relies on various systems to ensure the business runs smoothly. The technical system takes care of heating, ventilation, elevators, power supply, etc.; the logistic system takes care of supplies, floor space organization, etc.; the commercial system takes care of occupancy rates, activity distribution, etc. – all relying on individual datasets to plan and execute at best their tasks. Similarly, in a production plant, different systems take care of various organized and intertwined tasks to allow operations to run smoothly. Modifying one single system requires adjustments in all others. And none of those systems can individually provide a comprehensive view of what type of change would be optimal to achieve a specific goal. They rely on their historical dataset and processes to optimize their output. They can provide indications about the impact of an eventual change only in what concerns their output. Still, they cannot consider the subsequent effects on other systems or if a better outcome could be achieved by modifying a different (albeit related) system.

For example, the goal of making a building more eco-friendly could be achieved in different ways (e.g., retrofitting its ventilation system with more modern equipment or reconfiguring the floor plans to accommodate fewer people). The technical system would be able to provide the impact of retrofitting the ventilation system, while the logistic system would be concerned with reconfiguring the floor plans for reduced staff occupancy. But one system cannot consider the impact a change may have on the other. Nor which change would be optimal to reach the initial goal.

A Digital Twin, on the contrary, would provide the full impact of a change in one system across all the other systems that are connected and intertwined, directly or indirectly. As a result, an optimal decision can be taken at every step. Also, as the data generated to run the different systems continues to grow with the increased deployment of sensors and IoT devices, Digital Twins allow for Machine Learning and AI deployment to improve and optimize processes and asset usage.

Enter the Digital Twin System

To use and exploit a Digital Twin, a Digital Twin System is needed. And here is where commercial vendors come into play, as we move from the conceptual idea to the actual hands-on, usable product.

A Digital Twin System consists of the overarching digital infrastructure that allows the Digital Twin to “exist” by feeding it with relevant data and connecting it both up- and down-stream in the value chain.

The main components of a Digital Twin System can be provided by different suppliers and be integrated by further third-party vendors. Still, they usually consist of the following:

• Digital models (e.g., legacy business systems, company databases, engineering models, IoT data feeds, imagery, videos, etc.);
• Synchronization mechanisms to keep the digital models in sync with the real world;
• A digital thread implementation to weave the digital models together and allow a holistic view across multiple dimensions;
• Subsystems for visualization, analysis, automation, data integration, data security, etc.;

On top of this, simulation capabilities lay as the cornerstone of a Digital Twin System. More details on all these later. 

Multiple Digital Twins, multiple options

Within a company, setting up a Digital Twin System is the prerequisite to creating and deploying the Digital Twin of a given asset. But the investment offers the best returns when multiple Digital Twins are created, covering each a different asset and having them interact with each other. In a virtuous loop, once a company adopts the Digital Twin concept, it will naturally “grow” into it by expanding it to incorporate more and more assets, processes, and systems.

Also, Digital Twins may be “expanded” to include features that allow them to control the corresponding asset. More and more physical assets incorporate more or less sophisticated sensors and controls that are digitally linked to control and operational systems. In such cases, Digital Twins can first simulate changes and, once validated, actually deploy them to the real-world asset.

Creating a sustainable competitive advantage

In building a Digital Twin, several tools, mostly software-based, are used. Also, Digital Twins tend to develop incrementally, starting from the simplest models and combining them in increasingly complex systems.

Collecting and linking the data

It all starts from incorporating existing data: CAD design data is often a good starting point, and CAD software providers such as Dassault Systemes SE, PTC Inc, or Siemens AG were among the pioneers of the Digital Twin concept in industrial settings. Autodesk Inc is blazing a similar trail for buildings.

Today, more and more data is available thanks to IoT and LiDar-based scanning technologies that recognize and tag the different components of a given asset. Players like Open Text Corp or Matterport Inc are critical enablers on this front. This information is often needed to complement the data derived from initial CAD designs, especially when the Digital Twin concerns an already existing real-world asset. Digitizing the data collection process is key in facilitating the adoption of Digital Twins. Today it is possible to walk (or even fly a drone) through a plant or a building and laser-scan it to obtain a fully detailed list of its components, reducing the time needed to collect the data by orders of magnitude versus a manual approach. 

Simulating possible scenarios

Once all the relevant data have been sourced, the next step is to add simulation capabilities. In this domain, ANSYS Inc is an undisputed leader. The role of simulation is to enable the play out of various scenarios in virtual conditions to test them very safely and very inexpensively compared to real-life tests. The simulation software is used to investigate the model and perform various design analyses for functional verification.

The more a Digital Twin replicates a real-world asset, including its dependencies (up and downstream in the value chain), the more useful the simulation will be. 

Moreover, AI can be used as a real-time accelerator for simulations. For example, accurate mathematical simulation of fluid dynamics takes months of offline computation, but AI can be trained to accurately mimic the simulation results and replicate them for real-time purposes.

Providing a real-time 3D view

The final piece of the Digital Twin puzzle is visualization. Rendering a real-time 3D view of the Digital Twin (and thus the asset it represents) with all the links, references, and tags related to it is not easy. The amount of data to be handled is exponential to the complexity of the Digital Twin. CAD software's classical 3D rendering capabilities are easily overwhelmed, and real-time adjustments (necessary during simulations) become impossible.

This is an opportunity for the 3D capabilities of game engines, like those developed by Unity Software Inc or Unreal (owned by privately-owned Epic Games). These software (known as game rendering engines) were developed to provide video game makers with tools to render 3D images and universes quickly, allowing for fluid animations in video games. Now, these capabilities are being integrated into the Digital Twin concept, offering a win-win scenario: gaming 3D engines get a horizontal market expansion, while proponents of Digital Twins get an attractive and easy-to-interact-with interface. The deal between Autodesk and Epic we recently wrote about falls exactly into this context.

Conclusion

Digital Twins have the power to transform an organization through the insights and use cases that they deliver. By bringing together the three critical components of data models, analytics, and knowledge base, they enable a framework that allows companies to fully leverage their existing digitalization investments in the most powerful way possible. The byproduct of leveraging Digital Twins across a whole company is an optimized operational model that translates into a sustainable competitive advantage.

A confluence of factors, both on the technical and economic side, is setting the stage for a significant inflection in the adoption of Digital Twins. We are at the onset of a significant investment cycle and stand ready to capitalize on it through our strategies, notably AI & Robotics and Sustainable Future. 

Catalysts

• A confluence of efforts. Leaders in different software end-market niches are pushing for Digital Twin adoption. As more verticals get involved, the snowball effect comes into play.

• Standardization and interoperability. Vendor models are converging and making different Digital Twins easier to interface with each other. 

• 3D visualization. Having proper and timely 3D rendering is key in making Digital Twins more user-friendly, and thus spurring adoption.

Risks

• Acceleration may take longer than anticipated. The exact shape of the adoption curve is difficult to predict, as the recent story shows. 

• Time to ROI. Digital Twins investments are compounding, but without a significant initial increase, they may take much longer to reach significant ROI.

• Constrained budgets. CapEx spending was reined in during the pandemic, and companies may still be reluctant to invest, facing a dire economic outlook.