Digital Twin-Based Real Time Back Analysis of System Behaviour in Supported Excavations

Digital twins are receiving increasing attention for geotechnical projects (e.g., for storing monitoring data, ground models) but are often limited to either digital models (digital representation of the physical asset) or digital shadows (real-time data from monitoring the physical asset). We use the term digital twin, as we present a case study of a project where a digital shadow's functionality is enhanced with real time system behaviour simulation capabilities. So far, digital twins are not commonly used for improving the design and execution of geotechnical activities (e.g., deep excavations), although commercial software exists for specific excavation problems. The case study addresses the construction of the new office for the Norwegian Geotechnical Institute in Oslo, featuring an up to 7 m deep excavation pit in marine clay ground conditions in a densely populated urban environment. The construction pit is equipped with several inclinometers, load cells and fibre-optic sensors to monitor soil and structure response at different stages of excavation. Most sensors were installed in the first half of 2023, providing continuous monitoring data since then. This paper presents a workflow for using monitoring data to continuously update the digital twin (numerical 2D/3D model), thereby enabling more realistic prediction of the system behaviour and optimizations in the design (e.g., adaption of strut forces during construction, calibrate soil parameters). The project's design phase provides the starting point of the simulation. The digital twin case study highlights the challenge of integrating multiple different geotechnical data sources within one model, while also demonstrating the potential of real time back analysis to improve system behaviour predictions during construction. Given a suitable contractual framework, digital twin based real time back analysis improves geotechnical work and is the next step in the evolution of the observational method.

Keywords: Digital twin, Sensor data.