The continuous expansion of the data set accrued by the DT and the embedding of software capable of data analytics (more and more AI based) on those data is de-facto transforming the DT into a knowledge entity.
The knowledge is about the PhT -what it is, how it is performing, what are the interactions taking place with the environment- and it is rapidly extending to the ambient in which the PhT operates as well as a knowledge derived from the analyses of the knowledge space of other instances of that DT. This latter knowledge is created, usually, outside the DT by an external function (most likely leveraging on AI and ML). Although this knowledge is created outside of the DT it is injected in the DT to expand, and refine, its decision capability, hence it becomes part of the DT itself.
The DT knowledge is accrued to enable decision making and taking in its interaction with the PhT: it is a knowledge “to take actions” not a knowledge about things. This is usually referred to as “executable knowledge”.
The executable knowledge results in interactions among entities (autonomous players). As shown in figure 3 we find this knowledge in the workings of a company, manifesting itself in the ways activities are performed within the company and in the interactions the company has across its value chain.
This creates a knowledge infrastructure that in turns generates an emerging system wide knowledge.
The DTs become knowledge hubs and, as they expand their capabilities, they become independent knowledge entities that can be used in other contexts.
This is an interesting evolution in terms of manufacturing processes and business opportunities since they can be used “independently” of their PhT.
Furthermore, the possibility to share knowledge through interconnection of DTs leads to the creation of DTs cluster.
This is the case in a smart cities. Singapore was the first city to leverage on Digital Twins, independently created to mirror specific resources, by clustering them creating a Singapore City DT. This DT is an abstraction of the city, modelling the interplay of its various infrastructures and components.
Likewise in a manufacturing context. We can cluster the DTs of robots on a shop floor to create the DT of that shop floor. This is not just a static representation of the shop floor. It is a dynamic model of what is going on AND what can go on there. We are seeing more and more application of this DT cluster concept in manufacturing, used both for monitoring and for planning a new production line, determining how to restructure the shop floor and how to change/tune individual components (robots, teams…) up to the redesign of the whole factory (watch the clip).