PSI - Issue 64

1968 E. Michelini et al. / Procedia Structural Integrity 64 (2024) 1967–1974 2 Elena Michelini, S ł awomir Dudziak, Simone Ravasini, Beatrice Belletti / Structural Integrity Procedia 00 (2019) 000–000

In existing buildings, column settlements can be caused by a variety of factors, such as construction errors, uneven distribution of degradation effects (i.e., rebar corrosion), or local subsidence due to natural hazards or human activities (Negulescu and Foerster, 2010; Brunesi et al., 2015; Fotopoulou et al., 2018; Feng et al., 2021; Martinelli et al., 2022; Michelini et al., 2023; Ding et al., 2023). While in the design of new structures the progressive collapse potential can be reduced through the application of specific regulations and guidelines (i.e., UFC, 2016; GSA, 2013; Ravasini et al. 2021), in the case of existing buildings it becomes instead essential to accurately predict the effects of these “low probability and high consequence” (LPHC) events, so to limit human and financial losses, especially in case of public structures. Within this context, Finite Element (FE) analyses based on pushdown techniques are the most widespread tools for robustness evaluations and progressive damage assessment (Brunesi and Parisi, 2017), also due to their larger flexibility and lower economic costs with respect to experimental tests. The results deriving from numerical analyses can be not only used for the design of retrofit interventions (Scalvenzi et al., 2023), but can be also combined to Structural Health Monitoring (SHM) systems, for a near-real-time performance assessment of the structural behavior. In this work, nonlinear FE analyses are applied to study the evolution of the behavior of an existing building with RC framed structural system subjected to large column settlements. The followed numerical approach is first validated against the outcomes of a well-documented experimental test from the literature (Li et al., 2016), concerning the progressive collapse of a planar RC bare frame. Numerical simulations are carried out for comparison by using two commercial FE software (Seismostruct and ABAQUS), whose main assumptions are discussed in the text. The procedure is subsequently extended to the damage assessment of an existing Italian school building, whose main features are known from previous studies (Lima et al., 2018; Belletti et al., 2022). To this end, an increasing movement triggered by large differential settlements is assumed, and two different column settlement scenarios are applied to a 2D sub-model corresponding to an internal frame. The main goal of these analyses is the identification of significant structural parameters that can be easily monitored through SHM in existing structures of the same typology, and the determination of possible corresponding threshold values, associated to the achievement of well-recognized performance limit states (which are herein defined according to the literature, as in Fotopoulou et al., 2018). 2. Description of the numerical procedure and definition of damage states Numerical analyses are mainly performed with Seismostruct FE code (Seismosoft, 2023), which has efficient fiber beam elements implemented and numerous built-in features that enables users to easily assess, for instance, the achievement of predefined performance-based limit states. Additional analyses are carried out also with ABAQUS code (Dassault Systèmes, 2018) to verify and validate the modelling strategy. In Seismostruct, two- noded infrmDB (inelastic displacement-based) frame elements with two Gauss points are used. Large displacement and rotations are considered through a total co-rotational formulation. The material non-linearity is considered using the fiber approach, in which the beam’s cross-section is divided into 200-300 fibers and sectional moments and normal forces are obtained through integration of the non-linear stress-strain responses of separate fibers. For fibers associated with concrete, the Mander’s material model has been selected, see Figure 1(b). For steel fibers, a bilinear constitutive law with a sudden stress drop at ultimate strain has been assumed, see Figure 1(a). Fairly similar assumptions are made for the analyses carried out with ABAQUS. In this case, two-node shear-flexible beam elements (B31) with one Gauss point are used, whose formulation takes into account large axial strains and large rotations. The sectional stress-strain behavior follows the fiber approach as well. Fibers representing rebar are introduced using the *REBAR option. A simplified no-tension material model with trilinear response in compression is assumed for concrete (Sio et al., 2024) and implemented through a UMAT subroutine, see Figure 1(c), whereas the same bilinear stress-strain relationship used in Seismostruct is assigned to steel fibers. The effect of concrete confinement has been neglected in all the analyses, due to low amount of stirrups in the analyzed case studies. For both the FE codes, mesh size has been selected after initial mesh-dependency studies. Performance-based assessment of RC frames requires the definition of appropriate Limit States (LS) associated with increasing damage levels. The four LS adopted in this work, summarized in Table 1, relate the exceedance of a given structural damage state (slight, moderate, extensive, or complete), with the achievement of predefined limit strains in concrete and steel, according to the literature (Fotopoulou et al., 2018; Negulescu et al., 2010; Crowley et al., 2004; Bird et al., 2005). It is worth noting that the considered LS are related to flexural damage, while the occurrence of shear failures is neglected in this preliminary study. Furthermore, the threshold strain values (for rebar