By Dr. David McKee, Fractional CTO, Digital Twin Thought Leader
In July 2023, the Digital Twin Consortium launched the Digital Twin Platform Stack Architectural Framework (PSAF) as a high-level guide to the critical components of a digital twin system. In this blog, we review some of the essential elements of the paper and framework, as well as address some of the common questions. We start with a quick look at the primary aspects of the framework and then revisit some of the companion frameworks you can use to accelerate the design and development of digital twin systems. Following that, we will look at how digital twin systems can be composed of multiple digital twins and some of the benefits this brings, particularly in deployment and synchronization. Finally, we will examine various use cases and consider the balance between predictive and prescriptive insights.
In the guide, we define a digital twin system as a system of systems that uses virtual representations of physical or abstract entities to provide insights, optimize processes, and enable decision-making. The guide outlines the main components of a digital twin system, such as the IT/OT infrastructure, the virtual representation, the service layer, and the applications. It also discussed the importance of security, trustworthiness, and governance for digital twin systems. We provided five use cases of different maturity levels to illustrate how the framework works in practice:
- Buildings as batteries: A digital twin system that models the energy consumption and generation of buildings and optimizes their energy management.
- Emergency communication services: A digital twin system that simulates the communication network of emergency responders and identifies potential failures and bottlenecks.
- Manufacturing quality control via remote operator: A digital twin system that enables a remote operator to monitor and control the quality of a manufacturing process using sensors and cameras.
- Scope 3 carbon reporting emissions: A digital twin system that tracks and reports the carbon emissions of an organization’s supply chain and operations.
- Infectious disease management: A digital twin system that models the spread of infectious diseases and evaluates the effectiveness of interventions.
Digital Twin Platform Stack Architectural Framework
The Digital Twin Platform Stack Architectural Framework identifies the critical building blocks for a digital twin system. These include the primary layers of the IT/OT platform, virtual representation, integration, and synchronization services, followed by an application layer for insights.
Figure 1: Digital twin platform stack architectural framework
The IT/OT infrastructure is the foundation of a digital twin system. It comprises the hardware and software components that enable data collection, processing, storage, and communication between the physical and virtual worlds. The IT/OT infrastructure includes sensors, devices, gateways, networks, cloud services, databases, and analytics platforms. The IT/OT infrastructure needs to be scalable, reliable, secure, and interoperable to support the complex and dynamic nature of digital twin systems.
The virtual representation is the core of a digital twin system. It is a dynamic, virtual model that reflects the properties, conditions, and behavior of a physical or abstract entity. You can create a virtual representation using various methods, such as 3D modeling, simulation, machine learning, or reasoning. You can use the virtual representation to perform multiple tasks, such as testing scenarios, predicting outcomes, optimizing performance, or generating insights. You must update the virtual representation from real-time data collected from sensors or other sources to ensure accuracy and relevance.
The service layer is the interface between the virtual representation and the applications. It provides various services to access and manipulate the data and information stored in the virtual representation. The service layer is vital for synchronizing systems within the digital twin and the real world. Ensuring this synchronization between the real and virtual domains is one of the critical challenges in implementing digital twin systems.
Synchronization means keeping the states of digital twins consistent with those of their real-world counterparts. You can achieve synchronization by transferring data from sensors or other sources in real-time or near real-time to update the virtual models. Synchronization can also involve sending commands or feedback from the virtual models to control or influence the physical systems. In some particular circumstances, it could also apply human-in-the-loop operators.
There are different types of synchronization depending on the direction and frequency of data transfer:
- One-way synchronization: Data is transferred only from the physical domain to the virtual domain or vice versa.
- Two-way synchronization: Data is transferred bidirectionally between the physical domain and the virtual domain.
- Continuous synchronization: Data is transferred at regular intervals or whenever there is a change in state.
- Discrete synchronization: Data is transferred only at specific points in time or upon request.
In the guide, we also discuss the maturity of digital twin systems and suggest that if the synchronization is only one-way, it can not be considered a digital twin system, which leaves us with either continuous or discrete two-way synchronization.
Synchronization can also vary depending on the level of granularity or abstraction:
- Full synchronization: All aspects of state are transferred between domains.
- Partial synchronization: Only selected aspects of state are transferred between domains.
- Aggregated synchronization: Data is aggregated or summarized before being transferred between domains.
- Detailed synchronization: Data is transferred without any aggregation or summarization between domains.
As shown in the figure above, these can each be expanded into further stack elements as detailed in the published guide.
System of Systems Composition
You can combine this framework with the Digital Twin Capabilities Periodic table to help define the capabilities required for any given set of use case scenarios in a digital twin project. Using both tools in combination provides a powerful tool for rapidly and iteratively architecting digital twin systems. You can realize each capability individually by using a subset of elements from the platform stack, typically combining IT/OT elements with another element from the stack. This way, we can quickly compose from a range of capabilities a digital twin system specification for a given use case.
By building digital twin systems in a composable manner, we can iteratively build digital twins for individual scenarios, which provides numerous benefits in managing complexity, interoperability, and more. Less commonly discussed, though, is the advantage that composing digital twins provides from a cost and expertise as well as deployment perspective. By composing a digital twin system from multiple smaller systems, we can utilize dedicated expertise or specialist tooling for the given problem rather than building in-house expertise across every aspect of the system. For example, this is particularly useful from a modeling language angle where the most appropriate languages and tooling can be used to represent the specific subset of the problem.
Ongoing Work and Next Steps
Within the Capabilities & Technologies working group at the Digital Twin Consortium, we continue to develop the platform stack architectural framework in combination with other tools, including the capabilities periodic table, as well as for security and trustworthiness. Over the next few months, we will be publishing a range of more detailed worked examples and expanding the open-source examples.
The views expressed in this blog are from Dr. David McKee, Fractional CTO, Digital Twin Thought Leader.
About the Author
Dr. David McKee
Dr. David McKee, Fractional CTO, Digital Twin Thought Leader