Dynamic Parameter Calibration of an Analytical Model to Predict Transient Groundwater Inflow into a Tunnel

Abstract

The use of either analytical or numerical models of groundwater inflow into a tunnel, ignoring the excavation process, will result in inaccurate prediction. More researchers have begun to study prediction methods with temporal and spatial variables, in accompaniment with the tunnel excavation progress. The hydrogeological parameters significantly affect tunnel inflow prediction; thus, it is always worthy to carry out an inversion analysis when the inflow rate observation is available for the excavated part of a tunnel. Based on a transient analytical model and the Trust Region Reflection (TRR) algorithm, this paper proposes a Dynamic Parameter Calibration (DPC) method to sequentially optimize parameters for a tunnel that is divided into several sectors. The results of two case studies indicate that the fitting effects by DPC are significantly improved compared with the empirical trial-and-error method, indicating good optimizations. The proposed scheme can conduct calibration simultaneously for multiple tunnel sectors and also several optimizations for the same sector. The parameter optimization results reflect the lithologic heterogeneity of different strata. Parameter sensitivity analysis proves that the hydraulic conductivity K has a greater influence on water inflow calculation than the specific storage coefficient S.