Modeling Digital Twins
A key element for valuable virtual representations
Digital twins are virtual representations of entities or processes within the real world. This article will explain the importance of modeling Digital Twins, the key element to achieving a valuable virtual representation of your entities/processes. A high overview of three alternatives to implement Digital-Twins-based solutions and their modeling strategies is made.
Modeling Digital Twins
The modeling of the available resources and data sources achieves the virtual representation within the digital world of an entity/process. Usually, we can divide those resources into two categories: Virtual entity/process and Real entity/process.
- The Virtual entity/process comprises design data, usually 2D or 3D models of the physical Thing. Simulation data on how the entity/process should ideally perform and process/application data, basically the behavior of your entity through an application flow, is very common, for instance, in business applications such as an ERP.
- The Real entity/process mainly comprises real-time data provided by sensors, control systems, and the entity’s user interactions. Historical data is typically represented by time-series data, a common way to store telemetry data from entities/processes. And, finally, context data, for which a very illustrative example is weather data.
The starting point of a digital twin implementation will be determined by the maturity of the integration and aggregation of all data sources and resources needed to feed the virtual representation of a given entity/process.
These representations enable their integration into the digital world to develop data visualization applications, XR experiences (virtual/augmented), simulations, applications services integration to aggregate context, and data analytics.
A digital twin-based application aims to orchestrate integration through digital solutions and collaboration from areas of a given business value chain, enabling data-driven decisions to generate business impact.
Alternatives to model Digital Twins
There are different perspectives on developing a Digital Twin based solution. This article will focus on the IoT perspective and, in particular, on what is happening in the vendor’s ecosystem for Industrial IoT space based on the latest Industrial IoT Platforms Magic Quadrant from Gartner. Let us review what is trending with the three key vendors: Azure Digital Twins, AWS TwinMaker, and PTC, focusing on ThingWorx.
Azure Digital Twins
Azure was named in 2021 as the leader of the IIoT Platforms Magic Quadrant, and they include a dedicated service for Digital Twins within the Azure IoT offering.
As we can see in the architecture reference diagram below, they placed the Digital Twin in the middle of all the services to ingest and aggregate data sources and resources to model the virtual representations. Then they enable alternative pipelines of data/services to obtain the value of that modeling in a wide variety of applications.
Within Azure Digital Twins, a model is similar to a class in an object-oriented programming language, defining a data shape for one particular concept in your actual work environment. Models have names and contain elements such as properties, telemetry/events, and commands that describe what this type of entity in your environment can do. Models for Azure Digital Twins are defined using the Digital Twins Definition Language (DTDL) based on JSON-LD. A DTDL model may contain zero, one, or any of the following fields: properties, telemetry, relationships, and components.
In the diagram below, we can see an example of the Azure Digital Twins service; it enables the creation of twin graphs based on digital models of entire environments, such as buildings, factories, farms, energy networks, etc.
AWS TwinMaker
The second vendor that we will review is AWS, named the challenger of the IIoT Platforms Magic Quadrant. During the latest reInvent, they dedicate attractive spaces to talk about the new approach of AWS TwinMaker, their Digital Twins service.
The main concepts of the service connect, model, and compose, and as an outcome, the enablement of applications with specific use case examples and the integration of components for data visualization and data analytics. The diagram below shows the high-level architecture on how the service fits in. Similar to the Azure approach, they illustrate different types of data sources.
IoT TwinMaker provides tools to model your system using an entity-component-based knowledge graph. You can use the entity-component architecture to create a representation of your physical system. This entity-component model consists of entities, components, and relationships. This approach is based on a software architectural pattern used in video game development.
Once the entities are modeled, the idea is to connect these models to data sources such as sensor data. And then create visualizations that help users understand the data and insights. This visualization component is achieved through the Scene composer, a tool for creating scenes in 3D, based on 3D/CAD models and optimized for web display to create visual representations of a given operation. Apart from the scenes, dashboards can be developed using the plug-in for Grafana enabling end-user applications.
PTC
As a third key player, it is crucial to consider PTC, the former leader of the IIoT Platforms Magic Quadrant. It is also important to remember that they have different ways to achieve a Digital Twin across their solution ecosystem. For example, Creo models and Windchill BOMs represent the physical configuration of a physical asset rather than its current properties and functionality. Vuforia experiences represent the physical object digitally in the real world using XR. Finally, in Thingworx, their IoT platform, our focus in this article, a Thing is a Digital Twin, used to represent a physical asset in the digital world.
The following diagram presents a high-level architecture for Thingworx. On the bottom are two types of storage, the one dedicated to the models of Things and one specific for time series data. Both users and Things interact with the Connection Server that optimizes the communication with the platform.
In terms of modeling in Thingworx, a Thing is like an object or instance in a programming language; an instantiated entity that encapsulates its functionality into the following attributes types: general information, properties, services, or functions that can be executed against the Thing, events and configuration tables that are like properties not intent to change frequently. Most of the time, Things are used to represent connections to physical assets; however, in specific scenarios, the connection can also support the integration of enterprise systems and remote data stores, among others, with the purpose of, as mentioned within the intro, to achieve a valuable virtual representation. The following diagram shows an example of the entity’s relationship that consolidates a Thing within Thingworx.
To finalize this section from Azure Digital Twins is important to highlight the integration of this service with Azure’s services ecosystem, especially in IoT, and with a strong focus on IIoT and OPC integration. From AWS, TwinMaker is relevant to stand out the focus on the 3D model integration as part of the service and the support/integration to Grafana to ease the data visualization component. And finally, from ThingWorx is remarkable in the capability to easily integrate with other tools from PTC’s ecosystem extending the capabilities of your digital twin, for example, combining with Vuforia to develop augmented/virtual reality experiences quickly.
Conclusion
This article covered the value of modeling Digital Twins as the key to achieving a valuable virtual representation. As mentioned before, there are different perspectives to implementing Digital Twins based solutions. We reviewed the offering of three key players in the IIoT space, where Digital Twins are an essential element in making business decisions more efficiently to drive business impact.
The invitation is to review other key players’ offerings to support Digital Twins implementations in scenarios that could extend from the IIoT space. Key players on the radar are Nvidia with its Omniverse Platform, Digital Twins by Unreal Engine, and PlantSight from Bentley.
