Standards Development

Scope

This document provides a generic digital twin maturity model, definition of assessment indicators, and guidance for a maturity assessment.

Purpose

A digital twin technology can be widely used in a number of application domains such as manufacturing, energy, environments, agriculture and smart city. During the practical discussion about a digital twin, it is occasionally observed that people explain technical features of a digital twin from different levels of perspective, and some miscommunication happened occasionally caused by their different expertise such as GIS, autonomous deli very vehicles, industrial manufacturing, and transportation.

This may cause disorder in integration of the digital twins, i.e., interoperable digital twins, in different application domains.

A digital twin maturity model is used for self-assessment to find improvement points towards a higher level, and can be used to set the analysis boundary and the same level of consideration issues even in different digital twin application domains.

There are several digital twin maturity models are being introduced by m arket researches, papers and articles. Atkins/IET introduced digital twin maturity spectrum with defining principles and outline usage.

Maturity level 0 − Defining principles: Reality capture (e.g., point cloud, drones, photogrammetry) − Outline usage: Brownfield (existing) as-built survey Maturity level 1 − Defining principles: 2D map/system or 3D model (e.g., object -based, with no metadata or BIM (Building Information Modeling)) photogrammetry) − Outline usage: Design/asset optimization and coordination Maturity level 2 − Defining principles: Connect model to persistent (static) data, metadata, and BIM Stage 2 (e.g., documents, drawings, asset management systems) − Outline usage: 4D / 5D simulation (i.e., time and cost additionally to 3D), D esign / asset management, BIM Stage 2 Maturity level 3 − Defining principles: Enrich with real-time data (e.g., from IoT sensors) − Outline usage: Operational efficiency Maturity level 4 − Defining principles:

Two-way data integration and interaction − Outline usage: Remote & immersive operations, Control the physical from the digital Maturity level 5 − Defining principles:

Autonomous operations and maintenance − Outline usage: Complete autonomous operations & maintenance Gartner also defined three-level maturity model. 3 JTC1-SC41/298/NP Maturity level 1: 3D visualization and simulation Maturity level 2: Real-time monitoring

Maturity level 3:

Analysis, prediction, and optimization Rainer Stark and Thomas Damerau introduced 8 -dimensional model in terms of integration breadth, connectivity mode, update frequency, CPS intelligence, simulation capabilities, digital model richness, human interaction and product lifecycle.

See document SC41WG6N076 for more detail. At PWI stage, Korean expert has introduced 5 -level maturity level in terms of capabilities as below. See document SC41WG6N076 also for tabular-form view.

Maturity level 1:

Mirroring − Physical entity modelled to have a similar visual appearance and rendered in 2D or 3D Maturity level

2: Monitoring − Persistent, static, and initial data connection

− No models of behaviours and dynamics but process control logics applied

− Realtime monitoring through twinning interface

− Partial automatic control, but mainly through human intervention for action Maturity level 3:

Modelling and Simulation

− Behaviours and dynamics modelled for operation and simulation

− What-if simulation provided

− Cause analysis by reproductive simulation

− Synchronization through twinning and threading interfaces throughout the lif e-cycle Maturity level 4:

Federation − Federated, synchronized, and interactive operations among Digital Twins, but through human intervention for action − Application/service-specific data exchanges through federation interface among cross - domains Maturity level 5:

Autonomy − Autonomous operations by live synchronization and orchestration without any human intervention 4 JTC1-SC41/298/NP Irish expert has also three-dimensional digital twin maturity model for discussion in terms of convergence, capabilities and integrated view.

See document SC41WG6N077 also for tabular-form view.

Maturity level 1 − Convergence (Disconnected):

Digital twin exist for individual systems of interest based on single data collections.

Digital twin is in no means directly connected to the syst ems of interest.

− Capabilities (Descriptive):

Digital twin can describe the behaviour system of interest and allows to monitor specific states in its life cycle.

This can be done through visualizations, data analytics, etc. or any combination of those.

− Integrated view (Task-specific):

Each digital twin provides an isolated view on a specific task in the life cycle of a system of interest. Users utilize the digital twin for this task and rarely interact with users in other life-cycle stages and have not full insights and understanding in the impact of decisions.

Maturity level 2 − Convergence (Synchronized):

The states of the digital twin are synchronized (read and write) with the systems of interest in an appropriate rate through common interop erable interfaces.

Multiple digital twins for different systems of interests, but do not exchange data.

− Capabilities (Diagnostic):

Digital twin uses AI to enable identification, diagnosis and tracing of root causes of issues in the systems of interest life cycle.

− Integrated view (Connected):

Digital twin of the same system of interest for different tasks are connected and allow users to analyse and diagnose problems across the life -cycle, understand the impact of decisions across the life-cycle and underlay their own decisions with data from the digital twin.

Maturity level 3 − Convergence (Federated):

Multiple digital twins of the same or different interacting systems of interests are connected to each other and exchange information through common int eroperable interfaces − Capabilities (Predictive):

Digital twin uses appropriate AI methods to predict digital twin behaviour allowing what-if analysis and time-ahead predictions for optimization of production, operation, maintenance, or risk and impact assessment and improved design processes.

− Integrated view (Preventive Views):

Digital twin enables the users to run what -if experiments across the systems of interests’ life cycle for design, manufacturing, and operation.

This allows them to predict the impact of decisions taken and enable mitigation of potential risks and provide user with confidence in operational changes.

Maturity level 4 − Convergence (Collaborative): Multiple digital twins collaborate in decision processes and can negotiate strategies to optimize the performance of a single system of interest.

− Capabilities (Preventive):

Fully integrate tools and data pipe line to automate the AI to enable the digital twin to continuously learn from new data. AI models are continuously updated and 5 JTC1-SC41/298/NP form an synchronize-learn-optimize cycle.

This allows to predict and prevent issues ahead of time.

− Integrated view (Augmented Views):

The automated and collaborative digital twin enable users to automatically analyse data and decisions in the life cycle an d highlight identified issues and provide the users with decision variants. Optimisation steps are being introduced as a result of in-depth analyses.

Maturity level 5 − Convergence (Unified): Digital twin operates autonomously with other digital twins in the context of the system of interest and connected systems to optimize systems of systems performance.

− Capabilities (Autonomous): Digital twin is capable of automatically learning from new data and run predictions and optimization on this new data.

This enables autonomous decision making in the digital twin with limited manual intervention and can communicate with external systems, including other digital twins. In addition, the digital twin is able to quantify its own accuracy in approximating real-life systems.

− Integrated view (Supervising Views):

The users are able to delegate re -occurring tasks to the autonomous digital twin, such as generating design variants, running standard maintenance tasks, etc. They monitor operation, investigate issues iden tified by the digital twin and set strategic guidelines for the digital twin.

A standardized maturity model of a digital twin can guide an architect and a developer to understand a digital twin in a consistent way from the point of capabilities, interopera tion, convergence and so on. In addition, it resolves disorder in integration of the digital twins in different application domains.