Physical modeling of settlement in sandy soil due to mechanized tunneling

Prediction of tunneling induced settlement in soft grounds is of particular importance for estimating potential risks to pre-existing structures. Considering soil complex behavior, particularly granular materials, empirical methods are of special significance. In this regard, a study using physical modeling has been conducted in Soil and Rock Laboratory of Department of Civil Engineering at Sahand University of Technology. The observed surface and sub-surface settlements have been recorded and compared with the results of the most common empirical method used for this purpose. Tunnel excavation causes relaxation of in-situ stress, and we expect some deformation at tunnel depth that in soft grounds may reach to ground surface. Prediction of these settlements is of particular importance for estimating potential risks to pre-existing structures. In practice, the prediction of ground displacements is based mainly on empirical relationships. The most common empirical method is Peck’s method. In this method for prediction of settlement, it is generally supposed that ground loss at tunnel depth is thoroughly transferred to surface. This assumption may be correct in clay, however, for drained condition such as one in sands, soil volume does not remain constant and shear deformation causes dilatation or contraction of soil leading to ground loss at surface, which is not exactly the same as that at the tunnel depth. Therefore, settlement in sand needs more consideration. In this paper, we present a physical model to monitor soil displacement due to mechanized tunneling. The obtained results from tests on sandy soil are then interpreted. Methodology and Approaches: Ground loss at tunnel depth has been used to model tunneling effect on surrounding soil. Front wall of the physical model is transparent and soil layers movements have been measured by image processing technique. Sand sample movement during tests has been recorded by digital camera. Then, frames have been extracted and processed by means of particle image velocimetry (PIV) and observed surface and sub-surface settlement curves have been compared with the Peck curve. Since sand behavior varies in respect to its relative density, two different states have been considered. Results and The obtained results show that surface and subsurface settlement curves in the dry sand sample follow Gaussian curve with acceptable accuracy. During excavation in drained sand, ground loss at depth may not be the same as that at the surface, and therefore, the use of ground loss at tunnel depth may lead to incorrect prediction of surface settlement. Moreover, the results indicate that the magnitudes of ground movements in looser sand increase and, the shape of the settlement trough becomes narrower.