PSI - Issue 29

Michele Paradiso et al. / Procedia Structural Integrity 29 (2020) 87–94 Michele Paradiso, Sara Garuglieri and Viola Ferrarini / Structural Integrity Procedia 00 (2019) 000 – 000 5 in the conformationof thedifferent soil samples, despite theevident chromatic difference between them. As mentioned above, instead, the large sensitivity of this soil to humidity determined a change in volume as well as the lack of adequate foundations haveadverselyaffected thestructures above bringing them to the current situation. 91

Fig. 3. Garuglieri S., Ferrarini V., (2015), Photo of south-east corner of the arch in south gallery where are present pre-collapse phenomena

Further investiga tions were made based on the fundamental methodologies taken from the Restoration Papers (ICOMOS, 2003); the problem was faced by using several parallel investigation methodologies finally compared in order to grasp different aspects: kinematic analysis with the help of graphic static, use of finite element calculation programs (Strauss 7), experimental approach with construction of a scale model of the arch which recreated thereal load conditions andkinematics. The first model to be processedwith the finite element ca lculationprogramdidnot foreseeanyyieldinga t thebase of the central column, as happened in the rea lity; this made it possible to observe thesituation only for the static load and for its effects on the initial collapse behavior of the two arches. Observing the results in terms of displacements with the application of the entire imposed load, it can be easily no ted theestablishment of hinges on the extrados and on the intrados of the most stressed arch (Figure 3), an evident sing tha t the static loada lone is sufficient to conduct the arch towards collapse (most probably this behavior is duenot only to the dist ributed load, but also to the its irregular geometric conformation of the arch). The scheme of the hinges shows that Couplet's flexural collapse mechanism (1642-1722) is respected, since these are alternated. By reproducing the graphof the deformed weight a t its own weight, it was possible to calculate the opening/closing of each point of contact between the various segments; this was measured as a percentage of the height of theashlar (36 cm). The threehighlightedpoints are the hingesthat visibly open in the model and their presence is in fact evident a lso on the basis of the calculation made, but other less obvious hinges also openwith the nakedeye at this stage, the presence ofwhich is in any case compliant with the collapse described above. Asecondcalculation modelwas started by providing for the insertion of an initia l subsidence a t the base of the central column of the order of 3 cm, a subsequent horizontal displacement of 1 cm and on the deformed part of the latter a further vertical subsidence of 3 cm and1 cm shift. The result once again shows a situation similar to the previous one in which by comparing the two deformedones it can easily to seen that the hinges opened in the deformationwith their ownweight change in thenext one: the distribution of the distance differences that allow the movement is redistributed in a different way so that some hinges close a lmost completely to make other open. If we consider the hinges of theseconddeformedone, it is clear that these no longer follow an a lternating trend, but this is expla inable if we consider the hinges close to each other as unique hinges (5/6 with 6/7 and 9/10 with 10/11) which indicate the distribution on the corners of hinges which are idea lly applied in the middle of the ashlar. In addition to the study of the collapse mechanism, a second phenomenon is evident from reality as well as from an in-depth readingof themodel: thewedge slidingdownwardof thesegment 4 and 5 (Figure 4.a).