PSI - Issue 44

Marco Gaetani d’Aragona et al. / Procedia Structural Integrity 44 (2023) 1760–1767 Marco Gaetani d’Aragona et al./ Structural Integrity Procedia 00 (2022) 000–000

1766

7

4. Application By adopting the procedure outlined in §3, a probabilistic distribution of repair costs is obtained for each building from the database, the distribution accounts for uncertainties in building seismic performances (record-to-record variability), quantification of damageable components, occurrence of given damage states, and unit repair costs. Figure 5 shows the results obtained via the proposed probabilistic framework for a number of simulations equal to 5000. In particular, ARC and the median values of PRC unitary costs obtained considering only IPs; bisector line (continuous) and the 100% error boundary lines (dashed) are also represented. It is interesting to note that despite the dispersion, in about the 66% of cases, the error in prediction is lower than 70%; and it is higher for cases in which the ARC is unexpectedly low. Despite the scatter evidenced in fig.5, for the analyzed building stock the ARC IP was 19.58 M€, while the value predicted with the framework was 18.16 M€. Thus, errors in prediction compensate leading to a loss prediction that underpredicts the total repair cost of about 7% only, which can be considered a remarkable result in a large-scale framework. The probabilistic damage and loss assessment framework in §3.4 required less than 10 minutes to be applied to the entire database.

600

500

400

)

2

300

(euros/m

IP

200

PRC

100

0

0

100

200

300

400

500

600

2

ARC

)

(euros/m

IP

Fig. 5. Probabilistic loss framework proposed (from Gaetani d’Aragona et al., 2022a)

5. Conclusions This paper, resuming main results of a probabilistic loss assessment framework and benchmark study introduced in (Gaetani d’Aragona et al., 2022a), shows the applicability of the methodology for large scale studies purposes. The framework is based on the adoption of a simplified MDOF model, named Stick-IT which allows to rapidly calculate the seismic performances of infilled RC building typologies in terms of engineering demand under specific ground motion records. The proposed probabilistic framework can be easily implemented at the large scale to account for the propagation of uncertainties related to building response, intra-model and record-to record variability, quantification of damageable components, level of damage experienced, and repair costs. The accuracy of the proposed framework has been testes using a database consisting of 120 multistory infilled RC buildings damaged and repaired in the aftermath of the 2009 L’Aquila earthquake. The paper clarifies the steps of the procedure starting from the acquisition of main building features necessary to Stick-IT parameter determination. Then the characterization of the seismic hazard at the site, in terms of ground motion record selection, rotation and scaling according to ground motion Shake-Map is evidenced. The results of Nonlinear Time History Analyses performed by adopting the Stick-IT models is then used to predict damage and repair costs for infills, partitions, and integrated components accounting for the uncertainties in repair cost prediction. It is shown that proposed framework allows predicting a total repair cost for the considered building stock that agrees reasonably well with the actual repair costs, leading to a scatter of about 7% with respect to actual repair costs. Thus, demonstrating that the proposed method is suitable for large scale losses prediction.