Beyond buzzwords: digital twin maturity spectrum (Part 2)

The digital twin maturity spectrum has now been published as part of a whitepaper in collaboration with the Institution for Engineering and Technology (IET). Download the whitepaper and flyer here, or visit www.theiet.org/digital-twins for more information.

I presented the spectrum at the CIO Conference in Syon Park, London (2019). The session was kindly recorded, and you can watch it here:

In the first article of this series, I explored the basic concept of digital twins. Fundamentally, they are a digital replica of a physical thing - a ‘twin’. But depending on maturity, this replica can range from a simple 2D or 3D model of a local component, all the way to a fully integrated and highly accurate model of an entire asset, facility or even a country [i], with each component dynamically linked to engineering, construction, and operational data. 

This broad range of what a twin can be has made defining and understanding them extremely difficult, with disagreement on what level of maturity or features represents a ‘true’ twin. Technology and service providers promising more than is currently achievable and inflated market expectations have further complicated things.

In this second blog, I put forward a maturity spectrum in an attempt to offer more clarity and understanding. Undoubtably there will be critics, but it has been tested extensively cross-industry and seems to offer a clear framework for simply articulating what a twin is at each level of maturity. I welcome feedback as we continue working to create a common definition.

Value mapping

The global digital twin market was valued at USD 3.2bn in 2018 and is expected to reach USD 29.1bn by 2025 [ii]. Gartner predict that half of large industrial companies will use some form of one by 2021 - resulting in a 10% improvement in effectiveness [iii]. Irrespective of how various analysts consider value, they all point to one thing – significant growth and importance of the digital twin. The industry and supply chain are all now looking to harness that potential.

It is important to remember that the creation and management of a digital twin is a journey relevant to the entire project life cycle. While a twin can be developed at any point in an asset’s life, it is most effective when deployed at an early stage, so that captured data adds value along the way. 

While a twin can be developed at any point in the asset life, it is most effective when deployed at an early stage of a project, so that captured data adds value along the way

Industry commonly focus on a unicorn conception of what a twin could achieve if fully implemented, despite this currently being cost-prohibitive. Few refer to the milestones along the journey, or incremental value-proving developments. A maturity spectrum helps explain the steps and also provides a framework for developing a twin. 

Maturity spectrum

As a twin develops through time the elements of maturity increase in complexity and connectivity, and subsequently cost. n

It’s important to realise that these elements are not necessarily linear or sequential in development, so a twin might possess early or experimental features of higher-order elements before possessing the lower-order ones. However, their relationship in terms of complexity is best considered as logarithmic, whereby the higher-order elements are significantly more complex that the lower-order foundational elements.

The physical and digital are connected via a constellation of data platforms or aggregators, securely linking the two environments and any external data sources (such as asset management systems (EAMS), document management systems (EDMS), common data environments (CDE), data historians, etc). As the twin matures, this moves from one-way to two-way flow of information.

The ultimate aim is to create a ‘single version of truth’ for an asset. This is distinctly different to a ‘single source of truth’, as a digital twin is about using a constellation, or ecosystem, of technologies that work and connect

As you go through the maturity spectrum, each of the elements further enables removing humans from hazardous processes or tasks, intrinsically improving safety.

Element 0 – Reality capture (for existing physical assets)

The lowest-order element to start a digital twin (relevant only on existing physical assets without existing accurate information) is creation of an accurate, as-built data set of the asset geometry or system design. This is the foundational element, over which data of various types is connected and overlaid.

Data can be captured by a variety of survey and reality capture techniques, such as point cloud scanning, drones, photogrammetry, etc. This can now be captured more accurately, efficiently and cost-effectively than was possible just a few years ago, and significantly so when compared to traditional survey methods.

With Element 0, you can immediately provide value working within these point-cloud data sets through having greater asset certainty, and spatial context and understanding. This is particularly true in sectors where a high proportion of assets are built and ageing, or in high-hazard sectors where it reduces worker exposure to dangerous tasks. Sometimes it is appropriate to work within these point-cloud datasets, but often there is significant value in going to the next level of maturity.

Element 1 – 2D map/system or 3D Model (object-based only)

Element 1 is the typical entry-point for new assets as an outcome of the design process and is often updated from Element 0 post-construction. 

Models are purely object-based (surface, shapes, etc), with no metadata or BIM information attached. Point-clouds from element 0 can be proportionally converted, as and when required, into an object-based 2D map/system or 3D model. The conversion is largely a manual process today but will soon regularly be done through semi-automated methods involving machine learning.  

At this level of maturity, the twin provides significant value through design/asset optimisation and coordination, answering questions, such as: is there space to run a new line through that module? and how would the maintenance team conduct that task? 

Element 2 – Connected to persistent (static) data, metadata and BIM Stage 2

Further benefits are realised when Element 1 is connected to persistent data-sets, such as design information, material specifications, inspection reports, and asset management information; and further enriched with metadata (i.e. BIM). The data is added, tagged and pulled from existing systems, not embedded or stored in the 2D/3D model directly. 

This provides the basis for engineering, project planning, operations, maintenance and decommissioning, creating a single reference point from which all data can be viewed and interrogated, reducing errors, uncertainties and costs, and enabling faster decision making and collaboration; answering questions such as: Are we on target with our schedule and budget? Where are the highest risk items?

Having a data model of this maturity also allows integrated multi-physics, multi-scale, probabilistic simulations to be run against the asset, either directly in the twin or through connected simulation applications; answering ‘what if’ questions such as: If I change x how will it impact y? 

Element 3 – Enrich with real-time (dynamic) data 

Dynamic or operational data can be obtained and displayed in real (or near real) time through one-directional flow from the physical to the digital asset, facilitated by sensors, connected devices and the IoT. This data can be analysed to inform and predict the behaviour of the built asset, and facilitate decision making, with the output or results fed back and updated into the organisation’s existing systems.

This element of maturity is what many technology and service providers would identify as the ‘true’ starting point of a digital twin, though to get to this level of maturity requires several previous steps that are often not detailed.

Developing Element 3 requires sensors and connected devices to actively or passively capture and collect data. This is often the first significant investment.

Element 4 – Two-way integration and interaction

The state and condition of the physical asset can be changed by the twin, with output and results fed back and updated into the twin. For example, an operator could manipulate a physical valve or activate machinery by initiating the action from the twin. This level of integration requires additional sensor and mechanical augmentation of the physical asset.

This integration can also apply between the twin and other digital assets, such as other twins or even engineering systems and applications. For example, a designer using immersive technology modifies the design, the change is pushed to all connected applications, including the engineering design and process simulation package. The connected applications calculate the impact of the change and update the geometry and data accordingly, with these updates and their impact reflected live into the twin for the designer to see.

This full integration demonstrates the two methods of interacting with digital twins, human-to-machine and machine-to-machine.

Element 5 – Autonomous operations and maintenance

In the future it in not hard to imagine that the digital twin learns and evolves as a living repository for institutional knowledge, absorbing enough experience about the behaviour of the physical asset that it could become completely autonomous in its operations, able to react to anomalies and upsets and can take the necessary corrective action with little or no human interaction.

Achieving this level of maturity is purely aspirational at present, with only small facets of it for discrete situations possible now. The full ramifications of what Element 5 maturity means, and the quantifiable benefits it will bring, are yet to be fully understood.

Around the digital twin, wherever it sits on the maturity spectrum, is a data analytics engine. This interrogates the data to surface patterns and relationships, and enables trainable models based on artificial intelligence (AI) and machine learning (ML) methodologies.

There are many consumers of the data within a twin, each of whom will be securely presented a different view depending on their requirements and access permissions. The ultimate aim is to create a “single version of truth” for an asset. This is distinctly different to a “single source of truth”, as a digital twin is about using a constellation, or ecosystem, of technologies that work and connect.

Each digital twin fits into the organisation’s overall digital ecosystem like a node in a network, alongside potentially many other twins for different assets or systems. These twins can be ‘federated’ or connected via securely shared data and will become an embedded part of the enterprise, as intrinsic in management of the organisation as finance or HR.

Where is the industry now?

Although organisations want to achieve the higher-order Elements 3 and 4, the reality is that most are only ready for Elements 0, 1 and 2. This should not be of concern, as each milestone provides incremental value. It is also possible that achieving higher-order elements is not necessary for an organisation’s objectives. 

In my third blog I’ll be exploring this, looking at visualisation, governance and use cases of the digital twins, and their role as economic enablers.

Author: Simon Evans, Director of Digital Engineering, SNC-Lavalin

Article series also published:

• www.snclavalin.com/en/beyond-engineering/beyond-buzzwords-the-true-meaning-and-value-of-digital-twins
• www.snclavalin.com/en/beyond-engineering/beyond-buzzwords-digital-twin-maturity-spectrum

[i] Centre of Digital Build Britain, The Gemini Principles

[ii] www.zionmarketresearch.com/report/digital-twin-market

[iii] www.gartner.com/smarterwithgartner/prepare-for-the-impact-of-digital-twins; Top 10 Strategic Technology Trends for 2019: www.gartner.com/document/3891569