Specify Your Digital Twin :
The Data Synthesis Model
Key points to remember
Three key points must be tackled to specify a digital twin:
The modelling perimeter. A "digital twin" only has meaning when compared to its "physical twin". What is your digital twin the representative of? A building system, of a building in its entirety, of a campus?
The data representation structure. The digital twin must be able to represent the rich, heterogeneous and scalable data. A graph type representation is the only adequate structure to model a building in its entirety (the BIM uses this data structure)
The data dynamic. As a synthesis model, the digital twin must be able to integrate static data (BIM, documentation, description...), transaction data (tickets, calendars...) and state and history dynamic data (time series, measure point, alarms, instructions...)
Definition and modeling parameter
Globally, today the building is badly described in all systems because none have as a goal to bring a global and transverse vision of the building's operation.
The BMS systems bring a partial modelling of the building centred on the managed networks (HVAC, stores, lights)
The CMMS systems bring a partial modelling of the building centred on the rooms and equipment to maintain (the details, walls, dynamic data.... are not represented)
The room reservation systems bring a partial representation of the building centred on the available booking spaces.
The BIM brings, from this point of view, a huge novelty as it is the first initiative that has as a goal to describe in a fine and precise manner the whole of the elements composing a building. It constitutes an essential element for the construction of the digital twin. However, it stays static and does not bring any information on the state, history or behaviour data of the building.
Modeling scale of the building's digital twin
The building's digital twin is the data model representative of the building in its entirety, and only the building. Its perimeter is therefore limited to the building's scale.
It describes its structure, components, sub-systems and imposes a unique username base to each element of the model.
It describes its state and history
It allows the simulation of its behaviour
NB : A campus' digital twin will therefore be represented by an assembly of digital twins
Definition of the structure of the representation model
The information pyramid presents in a figured manner the data typologies according to the richness of the information processed:
Layer 1, raw data: that is the data that we find directly from the sensor. For example, a temperature sensor will supply, for a temperature of 23 °C, a raw piece of data: 23.
Layer 2, information: the information corresponding to this raw data enriched with meta-data, attributes or more generally semantic that enables their specification.
Layer 3, context: a context is a data graph in which each piece of information is a knot and the knots may be linked to each other by connections. This layer makes it possible to represent the data in their context (and contextualize the data) to be able to extract more sense and values. The data organizations represented as such are called ontologies.
Layer 4, wisdom: in an IS, the applications, analysis systems.... exploit this layer. The applications use contextualized data in order to bring value (decision-making, action, consulting...)
The digital twin (also called dynamic building graph) is a graph of rich data operating on layer 3.
Definition of the data dynamic
The digital twin must make the synthesis of data from varied natures presenting different dynamics:
BIM: static data graph
BMS: dynamic time series
Ticket: transaction data
The digital twin is a graph of data enabling the representation of static data (BIM, objects...), transaction data (tickets, events...) and dynamic data (time series, alarms...).