Location instrumentation optimization for disposal cells' deformation monitoring

Abstract

The Industrial Center for Geological Disposal, Cigéo project, will consist of creating disposal cells, such as tunnels, for high level and intermediate level long-lived radioactive waste. Repository cells type will deform under the rock loading. Although this convergence would be low, its monitoring is mandatory especially in order to know whether the waste packages could be retrievable during secular operating period. The monitoring of the cell shrink shall be made by an optimized instrumentation. The approach consists in determining, by numerical simulation of tunnel sections, the number of sensors to be put in place, their position and their orientation, taking into account their lifetime and precision. A simplified model represents a disposal cell subjected to different loads. This model create allows us to a database of the deformations obtained by the virtual sensors according to the soil stress applied to the repository cell. An inverse model built with a Bayesian approach will allow retrieving the stress of the ground corresponding to a given deformation. The capability of the inverse model to detect the loading condition is the criterion to be optimized. The numerical and inverse models were developed to compare horizontal pressure using a fitness function to classify individual configurations. Genetic Algorithm optimization is then used selection, crossover and mutation to find the best sensors placement for a given number of sensors.