Abstract:
The movement at the surface or near the surface and the potential damage in built heritage is an essential aspect in the safety of underground works with proximity interferences. The current methodologies for estimating subsidence require, firstly, the calculation of volume lost in the excavation, and secondly, the use of parametric curves, e.g. the Gaussian curve. Often, the selected parameters of these curves are the volume of subsidence and the maximum surface settlement. Numerical geotechnical models are currently used to determine the latter. To keep the models as simple as possible, the layers are represented by homogeneous sub-domains with fixed parameters, usually defined by statistical criteria.
Within the classical approach of design, the analysis of a conservative scenario leads to the definition of limits for the response of ground and support, which are then compared to its monitoring counterparts in order to validate the design assumptions. When applying the observational method, on the other hand, two ground model scenarios must be considered to design underground works: in the first one, which is used for preliminary design, the “most likely" values are assigned to the parameters; in the second one, which is related with collapse safety, the most unfavourable values are assumed for the parameters.
In this paper the influence of ground variability and its spatial correlation properties on the estimates of ground settlements caused by tunnelling is analyzed through probabilistic finite element modelling. Numerical results, concerning the excavation of a near surface tunnel in clayey overconsolidated ground are presented. Following the definition of a simplified geomechanical model, random samples of geotechnical parameters are generated, using the fast Fourier transform method. The characterization and statistical analysis of the surface settlement, the vertical displacement in a profile linking the tunnel crown to the surface and the displacements in a profile at 45 degrees with the horizontal axis of the model, is presented and discussed.
