PSI - Issue 44

Stefano Bracchi et al. / Procedia Structural Integrity 44 (2023) 394–401

399

6

Stefano Bracchi et al. / Structural Integrity Procedia 00 (2022) 000 – 000

place in some models, causing some results to be shifted with respect to the largest part of the data. Dispersion is lower than the case of masonry with average quality.

KL2-A

KL2-B

30

10 15 20 25 30 35 40

25

20

15

10 Frequency

Frequency

5

0 5

0

0

0.2

0.4

0

0.2

0.4

0.6

0.8

1

0.6

0.8

1

a g [g]

a g [g]

Fig. 3. Distributions of acceleration capacity corresponding to ULS obtained from analyses at KL2.

Fig. 5 reports the calibrated values of partial safety factors. The comparison among the various knowledge levels also takes into account the quality of masonry tested by in-situ tests (weak, average and good, in Fig. 5a, b and c, respectively). It is evident how, independently from the mortar quality, a reduction of uncertainties is always present at KL2, although it is more significant in case of good quality mortar (KL2-B). At KL3 with masonry of bad quality, both considering good or weak mortar, no reduction of uncertainties is present, since coefficients are larger than KL2, even performing more than a single in-situ test. This can be caused by the bad quality of the masonry element. In case of good quality mortar, performing more tests can lead to a slight reduction of uncertainties with respect to performing few tests, even though this is not taking place in case of weak mortar. Even in this case, the variation of factors among the KLs, especially in case of weak mortar, is very limited. At KL3-UPP, the behavior is very similar to KL3-CEN, with a reduction in uncertainties with the increase of knowledge more evident than KL3-CEN and with factors lower than KL3-CEN. Partial safety factors in case of good mortar are significantly lower with respect to weak mortar. In this case, for a building with good quality masonry and with good mortar, no difference is evident varying the number of tests performed, since partial safety factors at KL3-UPP-1 and KL3-UPP-2 are identical. Comparing all cases, it can be concluded that, for bad quality masonry, a reduction of uncertainties moving from KL2 to KL3 is not present, in both the cases of weak and good mortar and even increasing the number of tests performed. On the contrary, for masonry with average and good quality, a reduction of uncertainties at KL3 is present in all cases and it is larger in presence of a good quality mortar. Performing more in-situ tests leads to a slight reduction of uncertainties only in case of good quality mortar, whereas it seems more convenient to perform less tests on an element with weak mortar. Indeed, with the adopted methodology, the interval from which results of experimental tests at KL3 are sampled is significantly reduced with respect to KL1. In this way, the coefficient k’’ tends to be very small with respect to k’ and the effect of the number of tests n at the denominator of k’’ is very limited. This partially justifies the cases where, even performing more tests, no reduction of uncertainties is apparent. Moreover, it is evident how values obtained from sampling lead in various cases to models with different failure mechanisms, influencing the dispersion of displacement capacity. The considerations made at KL3-LOW can be extended also to the case of masonry of average quality. Uncertainties hence decrease increasing the knowledge more significantly with respect to KL3-LOW. The difference between weak and good mortar is very slight. At KL3-UPP the situation is similar: uncertainties reduce with the increase of knowledge in a slightly larger way with respect to the case of average quality masonry.