Influence of Geological Conditions on Measured TBM Vibration Frequency

This paper examines TBM vibration as a source of information about geological conditions. An EPB TBM was outfitted with accelerometers to monitor vibration during excavation of the University Link light rail tunnel project (U230) in Seattle Washington. Impact-response testing of the TBM indicated that significant signal over a wide range of frequencies transfers from the cutterhead where vibration due to ground interactions emanate, to the bulkhead where sensors can be installed. Analysis of the vibration data collected during excavation indicates that both amplitude and frequency content appear to be influenced by TBM operating parameters and by geological conditions. INTRODUCTION The vibration characteristics of a system when subjected to external stimulus have long been used as a way to monitor the system itself. For example, the field of vibration based condition or health monitoring uses the measured vibration response and its changes to identify wear and damage of rotating shafts, wind turbines, hydroelectric turbines, bridges and buildings. The vibration characteristics of earth construction equipment have been used in the same manner and also to interrogate the condition of the ground with which the equipment is interacting. Examples include intelligent vibratory soil compactors and smart drilling (Mooney & Rinehart 2009, Richard et al. 2002). In the former case, the elastic stiffness and compacted state of the soil is estimated based on measured drum vibration. In the latter case, the rock hardness and strength is estimated based on measured drill bit vibration. The goal of the study described in this paper is to develop a similar approach where TBM vibration can be used to assess geological conditions. The underlying premise is that the measurable vibration characteristics of equipment interacting with the ground will be influenced by the ground properties under certain operating conditions. The ground conditions, therefore, can be estimated by back-analysis that uses either statistical or physical models of the ground/machine interaction. For example, a vibratory drum operating on soil can be physically modeled with lumped masses, springs and dashpots (see figure 1). The model predicts the contact force vs. deflection response that is a function of the roller parameters, operating frequency and amplitude, as well as the assumed ground stiffness and damping. Through a process of matching experimentally measured vibration response to modeled response, the ground stiffness and damping are estimated. This estimation of soil stiffness and compacted state is provided continuously and in real time, and is now routinely performed in earthwork construction practice. If a physical model is difficult to develop, statistical approaches can also be used. For example, in smart drilling, empirical relationships between measured vibration amplitudes and frequencies with rock types and stress conditions have been developed through statistical analysis. The rock type can then be estimated based on these empirical models.