PSI - Issue 44

Corrado Chisari et al. / Procedia Structural Integrity 44 (2023) 1108–1115 Corrado Chisari et al./ Structural Integrity Procedia 00 (2022) 000 – 000

1112

5

(a)

(b)

(d)

(c)

Fig. 3. (a) Experimental force-displacement plots of the tested arches, (b) hinge distribution in UR and RI1, (c) sliding and detachment at the base blocks (hinge D) in specimen RI2, and (d) mechanism observed in RI2. 3. Numerical simulations 3.1. Description of the models The tests on the arches previously described were modelled with Abaqus software (Dassault Systemes, 2013) to further investigate the mechanical behaviour of the proposed retrofitting strategy. The modelling strategy selected for the arch is based on simplified micromodelling; within this approach, units, discretised as C3D8R elastic solid elements, are connected at interfaces representing potential fracture surfaces. Mortar joints are thus lumped into zero thickness interfaces, and units are expanded by half joint thickness at each side. The interaction between two adjacent blocks is characterised by friction-cohesive behaviour, in which the tangential behaviour is governed by Mohr Coulomb friction coefficient; the normal behaviour is defined as Hard Contact with very low tensile strength. Specific normal and tangential stiffnesses of the surface were defined by appropriate uncoupled values K nn and K ss =K tt . Similar contact interfaces connected the arch to base blocks, assumed made of stiffer material. In the reinforced arches, the mortar layer was modelled with C3D8R solid elements, having Concrete Damage Plasticity constitutive relationship (Lee & Fenves, 1998). The connection between the blocks of the arch and the layer of mortar was represented by a cohesive interaction in the case of FRLBM, while a rigid Tie interaction was selected for ordinary mortar. The motivation for this difference lies in the macroscopic mechanical behaviour observed in the experimental tests, in which detachment was only observed for FRLBM. The analysis was divided into two phases: (1) application of self-weight and additional vertical loading on the arch, and (2) application of a horizontal displacement to the base. Similarly to the experimental test, a horizontal restraint was applied to a point located at the level of the load cell. A quasi-static implicit dynamic solver was employed in order to increase the robustness of the analysis thanks to the stabilising effect of the mass matrix in the solution algorithm.