Structural Behaviour of Masonry Arches on Moving Supports: from On-site Observation to Experimental and Numerical Analysis

Since ancient times, master builders have used arches to cover large spans in masonry structures. As a consequence, nowadays the safety assessment of these structural elements plays a fundamental role in the conservation of built cultural heritage. Due to their frequent occurrence, support displacements are one of the primary sources of damage for masonry arches. Among the potential causes of support displacements, slow-moving landslides have received very little attention from the scientific community. The present thesis is motivated by the observation of extensive and severe damage in the arches of historic masonry churches exposed to slow-moving landslides. These phenomena produce a combination of vertical and horizontal supports displacements, whose effect on the arch structural behaviour has never been thoroughly investigated in the literature, especially in the framework of large displacements. In view of the above, this thesis aims at providing a full understanding of the mechanics of masonry arches subjected to large support displacements, with special attention to inclined displacements. The methodology used to accomplish this goal included both experimental tests and numerical analyses on a segmental scaled dry-joint masonry arch subjected to different combinations of horizontal and vertical displacements at one support. The numerical simulations were carried out in the framework of large displacements using two different numerical approaches based on finite element (FE) and rigid block (RB) modelling. A micro-modelling strategy was adopted, where the arch was modelled as an assemblage of voussoirs, very stiff and infinitely resistant in compression in the FE model and rigid in the RB model, interacting at no-tension friction interfaces. Preliminary numerical simulations, aimed at designing the experimental set-up and gaining a first insight in the arch response, were carried out considering the arch as a rigid-no tension structure. To this aim, a very large value of interface normal stiffness was adopted in the FE model. A large experimental campaign was performed on a 1:10 small-scale model built as a dry-joint assemblage of voussoirs made of a bicomponent composite material. The results of the tests allowed, for the first time in the literature, to accurately assess the effect of the direction of the imposed support displacements on the arch response in the framework of large displacements. The comparison between numerical and experimental results showed that the numerical models were not able to accurately predict the experimental response, especially in terms of ultimate displacement capacity. To investigate this discrepancy, a sensitivity analysis on the effect of the interface normal stiffness on the FE predictions was performed. The results demonstrated that the difference between numerical and experimental results could be attributed due to the imperfections, and resulting deformability, of the joints of the physical model. A strategy to include imperfections in the numerical modelling, consisting in calibrating the interface normal stiffness based on the experimental results, was thus proposed and validated by performing further FE simulations, whose results were in very good agreement with the experimental evidence. Finally, to investigate the effect of geometrical imperfections on the arch response, a further experimental test was performed on a physical model made of bicomponent composite voussoirs exhibiting more imperfections. The test was simulated using a FE calibrated model to further validate the strategy proposed to model imperfections. The comparison between the experimental results for the two tested physical models showed that imperfections play a fundamental role in the response of small-scale arches to large support displacements. Furthermore, reducing the interface normal stiffness with respect to the large value adopted to model rigid interfaces proved to be an effective strategy to simulate the amount of imperfections of the experimental models.