Digital Twins for Resilient Geo-Infrastructure

Digital Twins are an emerging technology enabled by recent innovations in areas such as the Internet of Things (IoT). A Digital Twin is a virtual representation of a physical asset, process or system, used to understand and predict its physical counterpart's performance characteristics. A Digital Twin is coupled to its physical counterpart by establishing a data stream of real-time information obtained through monitoring the physical asset, or 'Physical Twin'. The Digital Twin provides a live simulation of the Physical Twin, using the data stream for validation of the simulation. Throughout the life-cycle of the Physical Twin, its Digital Twin may be calibrated by identifying discrepancies between predicted simulation data and real monitored data. Through calibration, the fidelity of the virtual model and the accuracy with which the Digital Twin represents the Physical Twin continuously increases. Machine learning techniques may be applied to automate some facets of Digital Twin calibration.

Digital twins could be created to couple the thousands of kilometres of roads, railways and flood defences in the United Kingdom, allowing for calibrated simulations of earthworks to produce reliable real-time predictions about maintenance needs and possible failures.

Failures in geotechnical assets are common (twice weekly) and costly. The cost of failures dwarf that of maintenance, but the timing of these failures is often unpredictable, making reliable maintenance planning exceedingly difficult using current techniques. Currently, separate computer models are used to produce weather projections, predict water flows within a geographical region and simulate the effects of water pressure on individual assets. While these models help forecast potential failures, they are unaware of real data and are disconnected from one another, limiting the accuracy of their predictions.

This project will investigate tools and techniques for engineering Digital Twins. As a primary case study, a Digital Twin for Resilient Geo-Infrastructure will be created and coupled to an existing geotechnical asset. The Digital Twin will provide a near-live simulation of the physical geotechnical asset, allowing for numerous potential benefits:
* Observation of the geotechnical asset from an arbitrary location.
* A Digital Twin provides an accurate representation of its coupled Physical Twin. Observations of a calibrated Digital Twin are therefore interchangeable with observations of the Physical Twin. Due to the Digital Twin existing in a virtual environment, the location of observations of the Digital Twin may be arbitrary.
* Non-destructive investigation into predicted or planned changes of the geotechnical asset.
* Changes to geotechnical assets are common, either through predicted natural events (flooding, earthquakes, landslides etc.) or planned activity such as construction. Complete testing of changes is costly and often unreasonable, as it may involve the destruction of the asset. Potential destructive testing may instead be performed on the Digital Twin.
* Autonomous failure detection of the asset.
* Through either observed deviation from a normal operational pattern, or discrepancies between the Physical Twin and its digital counterpart, anomalies indicating potential asset failure may be automatically detected.
* Accurate failure projection of the asset, allowing for predictive maintenance.
* Once a Digital Twin is calibrated to the point which it forms an acceptable representation of its Physical Twin, the Digital Twin may be used to simulate future performance of the asset. By forecasting when and how an asset may fail, predictive maintenance plans can be developed to prevent failure. The techniques used to prevent these failures can be tested against the Digital Twin to ensure they provide the intended outcome.