CMStatistics 2023: Start Registration
View Submission - CMStatistics
B0396
Title: Disentangling long term latent dynamics of the Brunelleschi's Dome cracks Authors:  Silvia Bacci - University of Florence (Italy) [presenting]
Bruno Bertaccini - University of Florence (Italy)
Fabrizio Cipollini - University of Florence (Italy)
Abstract: The Brunelleschi's Dome in Florence, an iconic symbol of the Renaissance, has been equipped with a monitoring system consisting of 166 electronic sensors since 1988. A model is proposed that captures the joint dynamics of the crack sizes on the Dome, measured daily using 57 deformometers. The empirical data reveals significant similarities in the crack evolution, including non-stationarity, a dominant yearly seasonal pattern resulting from the alternating weather seasons, and the presence of a slow-moving non-seasonal pattern indicating a gradual enlargement trend. Driven by this evidence, the model incorporates five components: 1) a non-seasonal, non-stochastic slow-moving pattern; 2) a direct seasonal effect of the masonry's temperature; 3) an indirect seasonal, non-stochastic effect of the masonry's temperature resulting from the movement of other cracks; 4) a stochastic dynamic latent component due to the joint movement of the cracks; 5) a noise term. Components 1) to 3) are structured non-parametrically using spline functions. The separation of components with different characteristics is valuable both descriptively and interpretatively, enabling a better understanding of the long-term evolution of the Dome cracks, as well as facilitating predictions.