Nearly every software vendor whose products are used to create, transport, store, or analyze industrial data now uses “digital twin” to describe its solutions. And while most of these do have elements of a digital twin, few are complete digital twins as characterized by the three elements of a digital twin defined in this ARC Strategy Report. Instead, current examples of true digital twins are created by integrating numerous solutions, oftentimes from many different vendors.
In many ways, digital twins are a culmination of the ongoing convergence of information technology (IT), operational technology (OT), and engineering technology (ET). The result is a digital representation of an asset, product, or process that mirrors the behavior of its real-world counterpart. However, this definition of digital twin, and others like it, are often insufficient to provide a meaningful basis for discussion – especially when comparing digital twins across applications and industries. This Strategy Report attempts to lay a foundation for those discussions by examining key technologies, applications, and maturity measurements. Key findings include:
- Digital twins are built by integrating many different technologies and rely on the cooperation of cross-functional teams to be effective.
- True digital twins share three elements: Context and characteristic data, real-time and operational data, and an information model used to integrate the two.
- There are two fundamental types of digital twins: Project-related digital twins and performance-related digital twins. The former is used to develop and deploy new products, assets, or process; the latter to optimize the performance and improve or enable the broader business functions surrounding a product, asset, or process.
- The methods used to evaluate the maturity of a digital twin are strongly associated with the metrics used to measure a company’s digital and IoT maturity.
Elements of a Digital Twin
Reality as the Foundation
A digital twin reflects one or more physical assets or processes. Every digital twin includes the elements described here. However, the under-lying asset or process will dictate their characteristics. Remember that the physical assets and/or processes are fundamental to defining each unique digital twin.
A digital twin of a production line will appear vastly different than a digital twin of a jet engine. This makes perfect sense since each has different challenges, requirements, and metrics for success. Similarly, the digital twin that represents the construction project of a refinery may share only a few things in common with the digital twin of the same refinery in operation. It is necessary that these unique instances have equally unique digital twins.
The outline of elements hereafter provides a way to generalize digital twins, whether they are of a pump or a powerplant, despite the enormous differences between their counterparts in the physical world. De-constructing the concept of the digital twin to its constituent parts enables comparison between digital twins across industries, creating a foundation for meaningful discussion regardless of the assets or processes they represent.
The Three Core Elements of the Digital Twin
ARC has identified three distinct elements of a digital twin.
First Element: Context or Characteristic Data
Contextual or characteristic data can be thought of as the DNA of the digital twin. This information defines the properties of the real-world asset or process. Many suppliers of digital twin solutions, especially those in the product lifecycle management (PLM) market will assert that this is represented by the three-dimensional model of the asset. In fact, three-dimensional models, though vital for certain activities, are not a requisite of the digital twin definition. More broadly, this information constitutes any type of model or datapoint that can be used to define the identity of the asset or process: its form, functions, and relationships with other assets or processes, including business processes. In-formation that falls within this category includes process diagrams; simulations; schedules; 1D, 2D and 3D models; supply chain models; bills of materials; maintenance procedures; product specifications; and project budgets.
Currently, most of this type of data is created and provides most of its value during the engineering and design phases of a project. For example, a structural analysis of a load bearing component may be used to validate its geometry during new product development. Similarly, the CAD models of a refinery may be used to detect locations where two parts of the structure could clash (interfere) with one another upon construction. After the initial intent of their creation has been satisfied, many such models are filed away, never to be seen again.
However, as part of a digital twin, this data is used to provide context throughout the lifecycle of the underlying asset or process. For instance, the CAD model that was used for clash detection can be referenced throughout the construction process to compare the as-built 3D-scan data to the as designed engineering model. Any discrepancies be-tween the two can be reconciled either physically, or (at much lower cost) by adjusting the CAD model. Rerunning the clash detection with an updated model will help ensure that inevitable divergences from the original design will not cause serious issues as construction progresses. Alternatively, it could confirm that there will be an issue, which can then be addressed and planned around before cascading into more de-lays. The same type of process can be used to validate and reconfigure schedules and budgets. Once construction is completed, the CAD model can be handed over to the refiner owner-operators to continue providing value for maintenance or brownfield projects throughout its lifecycle as part of their digital twin.
Key Challenges Associated with Context and Characteristic Data
- Selecting appropriate modeling and simulation technologies in a growing market of mature and novel solutions
- Identifying and creating useful models and simulations, considering fidelity and complexity requirements and limitations
- Maintaining model accuracy throughout the lifecycle of the underlying asset, product, or process
Second Element: Real-time and Operational Data
The second element of a digital twin is the real-time and operational data created during the lifecycle of the asset or process. The 3D-scanning data mentioned in the previous section would fall into this category. While models establish the identity of the digital twin’s real-world counterpart, real-time and operational data describe its state and behavior. For example, the prototypical digital twin of a manufacturing line tracks equipment health, performance indicators, and statistical process control data. It can be expanded to include work orders, quality inspection records, maintenance records, and other sources of data that describe the behavior of the manufacturing line – current and historical – and the actions taken to influence that behavior.
Just like the models discussed in the previous section, the type of information being tracked is highly influenced by the underlying asset or process. The digital twin of a supply chain, for example, might value shipment location tracking, weather pattern data, material pricing data, or logistic KPIs, such as delivery time, order accuracy, or inventory holding costs.
This element of the digital twin aligns closely with concepts and technologies associated with the Industrial Internet of Things (IIoT). While digital models are conceptual in nature, the real-time and operational data is - for the most part - a digital representation of real physical phenomena. The information isn’t created through a process of abstraction, but instead captured through observation; sensed and digitized. This tethers the digital twin to reality. Combined with the various forms of contextual data, this knowledge provides a foundation for insightful and timely decision-making.
Key Challenges Associated with Real-time and Operational Data
- Identifying what information is needed to achieve digital twin goals and implementing the necessary connectivity and data collection technology
- Determining data fidelity and volume required for accurate analytics and confident decision making
- Ensuring quality of real-time data collected throughout the lifecycle of the underlying asset or process
Third Element: Implementation of a Holistic Information Model
The information model integrates the first two elements of the digital twin: the contextual and operational data of the asset or process. The model formalizes the properties, relationships, and operations that can be performed for each data type that is part of the digital twin. When implemented as a database it federates data from the disparate systems that relate to the underlying asset or process and acts as the single source of truth for any application that requires access to the digital twin. If a modular approach is preferable, an information model can be used to stitch together the data coming from many digital twins such as a fleet of assets across an enterprise or a supplier’s installed base.
Additionally, this element provides a means to update the contextual models to reflect the information gained by collecting operational data. This action is sometimes referred to as tuning or synchronizing the digital twin.
Key Challenges Associated with the Information Model
- Handling unstructured, semi-structured, and structured data within an integrated environment
- Implementing a tuning/synchronization strategy to ensure the digital twin’s predictive capabilities are consistent throughout the lifecycle of the underlying asset or process
- Governing access to the digital twin, as well as managing availability, usability, consistency, data integrity and data security within and outside the enterprise
Types of Digital Twins
ARC Advisory Group has found that digital twins of nearly all applications can be categorized into two fundamental groups: project digital twins and performance digital twins.
The Project Digital Twin
Project-related digital twins facilitate the creation of a new asset, product, or process and rely heavily on the design and simulation elements of the digital twin model. Whether used for new product development or during the design and construction phase of a plant, project digital twins can accelerate program timing, warn against costly errors, in-crease the fitness, and provide a performance baseline of an end product or asset.
Digital Twins for Plant Design and Construction
For example, in the design and construction phase of a greenfield plant, the digital twin, as a single source of truth, provides an up-to-date view for all stakeholders. The same stakeholders can provide their own feedback to ensure the aspects of the digital twin that are relevant to their role are as accurate as possible. These harmonized and easily accessible views of the project enable greater cross-team and cross-domain collaboration. When unforeseen circumstances arise, each participant in the project is alerted and can quickly determine how their individual responsibilities are affected and can act to maintain timelines and budgets.
As mentioned previously, 3D scanning is often a critical tool in this endeavor, providing an as-built digital model that can be used to track and validate progress. Furthermore, simulations of the construction process can be used to optimize logistics, while simulations of the plant operation can be used for operator training. The latter, when combined with the models and simulations from process and detailed design can significantly ease handover and enable virtual commissioning.
Digital Twins for New Product Development
In discrete industries, such as automotive, a project digital twin can be implemented to connect all product data throughout a new product development cycle. One way to envision such a digital twin is as an in-formation model that forms associations between all the tools or methods used by engineers to design, validate, and describe a product. These include customer design requirements, engineering drawings, 3D models, failure mode effects analyses (FMEAs), simulations, validation test data, and pilot production data.
A digital twin creates a chain of logic between all the decisions made in the creation of these data and alerts when decisions or information gained during the course of the program may invalidate that logic. For instance, smarter products require greater cross-domain collaboration. While the design requirements set for an electrical engineer rarely over-lap with those set for a mechanical design engineer, the choices made by either can potentially affect the other’s ability to meet their requirements. A late-stage change in a products geometry could, for example, affect its ability to pass electromagnetic compatibility testing – a test that might never occur to a mechanical engineer, but one that is fundamental to electrical engineering. In such a case, the digital twin will alert all parties involved that the specific change to the 3D model may have in-validated the product’s ability to meet a specific design requirement and could even identify what failure modes may arise and the simulations that must be re-run to verify the new revision.
Table of Contents
- Executive Overview
- Elements of a Digital Twin
- Types of Digital Twins
- Digital Twin Maturity in Brief
ARC Advisory Group clients can view the complete report at ARC Client Portal
If you would like to buy this report or obtain information about how to become a client, please Contact Us