Real-Time Risk Assessment of Tunneling-Induced Building Damage Considering Polymorphic Uncertainty

Abstract

The risk assessment of tunneling-induced damage in buildings is a challenging task in geotechnical and structural engineering. It is important to consider the soil–structure interaction during the tunnel construction process. In this paper, finite-element (FE) simulation models of mechanized tunneling processes are combined with FE models of buildings to predict tunneling-induced damage. The soil–structure interaction is taken into account by considering the building stiffness in the tunneling process simulation model and by applying the computed foundation settlements as boundary conditions of the building model. The building damage risk is assessed by means of strains in the structural members and a corresponding category of damage is determined. Uncertainties of the geotechnical parameters and the structural parameters are quantified as random variables and intervals in the framework of polymorphic uncertainty modeling. For real-time predictions, the FE simulation models are approximated by artificial neural networks. This makes it possible to predict the structural damage risk according to scenarios of the operational tunneling process parameters in order to assist machine drivers during tunnel construction.