Enterprise Architecture – the details

The process we follow when architecting a digital twin within an existing EA is two way. It always starts with what we expect the digital twin to do. We call this the business layer.

In the following information systems layer, we look at the data coming from existing assets and determine the services the digital twin can deliver based on this data. Next to that, we also look at extending those services by adding intelligence or even new measurement data (including measuring devices) to the physical environment. We then reach the technology layer.

This process is based on the Architecture Development Method (ADM), as described in The Open Group Architecture Framework (TOGAF).

Although this model is used to define large enterprise architectures, it can also be applied to relatively simple solutions.

If we look at this framework in greater depth, we see that it starts with a preliminary phase. This phase is about defining the “why” (Motivation Stack) and determining the scope of EA work. In our case, the scope is usually defined as integrating digital twin functionality in the existing enterprise architecture. The result of this preliminary phase is a description of the work to be carried out by the person or team architecting the digital twin for this specific situation.

The next phase involves creating an architecture vision. 

Phase A. Architecture Vision

Here we create a high-level overview of what the digital twin will add to its physical counterpart. This could be anything from advanced analytics with its results shown to operators or a manager, to suggesting new setpoints to an operator or even to a Supervisory Control and Data Acquisition (SCADA) system. The result of this phase might contain “artist impressions” of the solution rather than detailed models.

When developing an architecture vision for a digital twin in terms of content, we use our digital twin reference architecture.

The next sections show examples based on this reference architecture.

Phase B. Business Architecture

Once a vision has been agreed, the business architecture can be developed. Detailing this vision may mean asking questions like “what services will the digital twin deliver to its environment?”

The result of this phase may look like this:

The diagram shows the process carried out by an operator – either a traditional operator or a more modern, highly skilled operator with advanced knowledge of process optimisation. The right part of the model shows business processes like simulate and analyse, traditionally performed by people like process engineers but increasingly supported by IT.

Phase C. Information Systems Architecture

In the information systems architecture phase, we add data and functionality from IT systems to the model. This diagram shows how the phase might look.

The blue blocks in the model show:

  1. What services the IT systems deliver to enable – or even fulfil – the necessary business processes identified in the business architecture.

  2. The functions needed to deliver these services.

  3. Which application components deliver the functions?

Although not explicitly show in this example, this part of the architecture largely deals with data flowing between functions.

Phase D. Technology Architecture

In the technology architecture phase, the layer describing the generic components with a role in the digital twin functions are added. This covers hardware like internet of things (IoT) measuring devices, facilities, more traditional IT hardware, and even network and security functions. The full picture in this example is shown below.

The green boxes show the industrial control systems with the connected actuators and sensors, security, and integration systems. Increasingly, we see digital twins being implemented on the edge (i.e., as near to the operational technology as possible), and in the cloud. Elements like machine learning model training and advanced analytics require large data storage capacity and processing power, meaning they’re often positioned in the cloud. ML process models can then be applied to the edge where they can support operational systems.

Phase E-H. Planning and implementation

In phases E to H, digital twin architecture is planned and guided. More on these phases, including the experienced good practices, will follow in future versions of this report.