Advance Exploration by Stochastic Inversion of Tunnel Seismic Waves - A Numerical Study

"With the aim to mitigate excavation risks and reduce the costs caused by tunnel boring machine (TBM) stoppage when tunneling under uncertain soil conditions, prediction ahead of the tunnel face has become an increasing interest in the tunneling community. Our research topic is concerned with developing effective concepts for nondestructive identification of geological anomalies ahead of the tunnel face by waveform inversion of reflected tunnel seismic waves. The authors demonstrate that this new technique is promising in providing tunnel engineers with an accurate image of the geological situation that the TBM is going to encounter thanks to the waveforms’ rich information about the geological structure that affects propagation, reflection/refraction and mode conversion behavior of the elastic waves. By setting up a 2-dimensional numerical model for simulation of tunnel seismic wave propagation influenced by a dip layer change ahead of the tunnel face, the authors show that the waveform misfit to be minimized by the inversion process is highly multimodal and therefore propose a hybrid stochastic optimization method that can quickly and robustly converge to the global minimum. DEVELOPMENTS IN TUNNEL ADVANCE EXPLORATION Conventional site investigation by drilling boreholes helps to draw a general image of geological structures along the borehole axes but that image is likely to misrepresent the detailed geological structure due to limited sampled locations. Due to high costs and restricted boring space which deep tunneling and tunneling under urban areas encounter, many industrial and academic research groups have had great interest in developing nondestructive ahead of the tunnel face prediction systems. That means only excitation sources and receiving sensors, which are usually placed on the ground surface or on the side walls or on the cutter head of the TBM, are used for acquisition of the measurement data. The recorded signals in the form of seismic or electromagnetic/ electrical waves, which travel through the geological structure or are reflected back from it, hold the signature of the geological structure of concern, such as geological layer changes, fault zones, erratic rocks, boulders, construction remnants, etc., and therefore can be used to reconstruct the spatial and material properties of those hidden geological structures. Most common nondestructive ahead-of-tunnel investigation systems to date are based on interpretation of reflected elastic/ acoustic waves (seismic waves) and analysis of electromagnetic/ electrical resistivity signals (Mooney et al. 2012). However, due to insufficient penetration depth and spread of signals generated by radar and geoelectric sources, the use of seismic waves for ahead of the tunnel face prediction (tunnel reconnaissance) is more preferable (Kneib et al. 2000; Bohlen et al. 2007)."