A digital twin may be characterized by the following three keywords:
- Virtual Twin
- Predictive Twin
- Twin Projection
At SINTEF we have a strong tradition to develop high-fidelity numerical tools for accurate prediction of an physical asset or a physical process. In order to enable "Predictive Twin" we are addressing two key challenges in the creation and update of digital twins:
- Transfer of the physical realism inherent in the analysis tools used in the design phase to analysis tools used in operation and management phase by means of reduced order modelling (ROM).
- Combining physical models and data-driven models, i.e., hybrid analysis and modelling (HAM), to open new vistas for run-time simulations that enable more realistic coupling of the real and virtual world.
A level of physical realism of analysis tools used in the operational phase that can enables a digital twin to be predicative (with acceptable accuracy) of the product in the real world will certaintly radically change our ability to understand and control the scientific or engineering situation at hand. New paradigms for product lifecycle management could then be established e.g., to facilitate significant reduction of cost of energy produced by offshore wind parks.