Resumen
Spatial variability is unavoidable for soils and it is important to consider such a feature in the design of geotechnical engineering as it may lead to some structure behaviors which cannot be predicted by a calculation assuming homogenous soils. This paper attempts to evaluate the performance of a shallow circular tunnel, in a context of the service limit state, considering the soil spatial variability. The Log-normal distributed random fields, generated by the Karhunen?Loeve expansion method, are used for the spatial modeling. A two-dimensional numerical model, based on the finite difference method, is constructed to deterministically estimate two quantities of interest (i.e., tunnel lining bending moment and surface settlement). The model is combined with the random fields and is implemented into the Monte Carlo simulation to investigate the effects of the soil spatial variability on the tunnel responses. The autocorrelation distance, an important parameter for random fields, is varied within multiple probabilistic analyses. For both of the two tunnel responses, their variabilities are increased with increasing the autocorrelation distance, while a minimum mean value can be observed with this parameter being approximately the tunnel radius. Such finding is very useful for practical designs. A sensitivity analysis is also conducted to show the importance of each random parameter.