Monitoring muon flux variations besides and on a tunnel-boring machine

International audienceTunnel-boring projects require a considerable amount of planning, well ahead of drilling operations. Despite these precautionary measures, unexpected ground geological features (local density variations, cavities, unstable superficial ground, instabilities induced by the driller) impose real-time adaptations of the drilling operations.The direct relation of muon flux absorption with the density of a given medium, makes muography a promising solution to provide a real-time density analysis of geological objects in front of the tunnel-boring machine (TBM). To test its applicability during the drilling of the "Grand Paris Express" subway network, a muon telescope was used in two different experiments. First it was placed besides the TBM, directed towards the drilled cylinder from a lateral perspective. It was then moved directly inside the TBM for the rest of the drilling operation. These experiments provide an unique dataset to optimize the methodology.The data interpretation remains technically challenging due to the poissonian nature of muon events. The muon flux is estimated through the averaging of point events during a limited time window, leading to possible misinterpretations of pure statistically natural fluctuations of the muons events. Numerical routines have been developed to compute the theoretical muon flux variations for different drilling trajectories, depths, and speeds, in a medium with heteregenous density.For the first experiment, good matches are obtained with experimental values. A blind analysis of the reconstructed TBM's time dependant position puts strong constraints on the time and spatial resolution of the method. For the second data set, because the telescope is moving forward and at an irregular pace, the muon flux crossing a particular geological object measured by the telescope has different directions with respect to time, allowing 3D density estimates. First attempts to reconstruct 3D density distribution of the ground, using inhomogeneous poisson likelihood, are presented