PSI - Issue 64

Salvatore Verre et al. / Procedia Structural Integrity 64 (2024) 1508–1515 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

1514

7

that very similar CMLs were obtained when the bilinear, multilinear, or Dai et al. (2006) expressions are considered. Fig. 4a also shows the CMLs reported in (Santandrea et al. 2020), obtained with a calibration procedure similar to that used in this work and using the CML expressed by a tri-linear function with ( ) 0 0   and the CML expressed by Eq. (7). CMLs from (Santandrea et al. 2020) are named as in Fig. 9d of (Santandrea et al. 2020). Single-lap shear tests specimens used in (Santandrea et al. 2020) were made with the same steel textile used in this paper, a lime-based matrix and a clay masonry substrate. Only the results of specimens with a plateau branch in the P - g response were used by Santandrea et al. (2020) to obtain the CMLs shown in Fig. 4a. Since the debonding mechanism observed by Santandrea et al. (2020) mainly involved the textile-matrix interface, it can be argued that the differences between the CMLs obtained by Santandrea et al. (2020) and the CMLs obtained based on the P - g responses presented in this paper (Fig. 2) are due to the different type of matrix. Fig. 4b compares the average experimental load responses with the analytical ( ) bl n P g , ( ) ml n P g , and ( ) Dai n P g functions obtained by introducing in Eq. (1) the CML bl , CML ml , and CML Dai , respectively. It can be observed that the analytical P n (g) functions satisfactory capture the experimental load responses. They slightly overestimate the debonding load for n =1 and underestimate it for n =2. This is because the ratio P plat ,2 / P plat ,1 is equal to 1.48, whereas the analytical load responses always provide 2 1 2 141 = = deb, deb, P P . . Fig. 4b also reports the effective bond lengths associated with the CMLs expressed by Eqs (5), (6), and (7), named bl eff L , ml eff L , and Dai eff L , respectively, for one and two layers of textile. It should be noted that CMLs expressed by Eqs. (5) and (7) don’t allow to determine a finite value of the effective bond length, whereas the multilinear CML allows to obtain a finite effective bond length, since ( ) 0 0   (Focacci et al. 2024). For the bilinear CML (Eq. (5)), the effective bond lengths associated with the descending part of the CML are reported in Fig. 4b, whereas for the Dai et al. (2006) CML (Eq. (7)), the effective bond lengths were determined as the lengths needed to transfer to the substrate a force equal to 96% of the debonding load expressed by Eq. (2) (Ueda and Dai 2005).

P [kN]

 [MPa]

a

b

12

2.5

n =2

10

2.0

8

CAL PPLAT ( g ) – Tri-linear function (Santandrea et al. 2020) CAL PPLAT ( g ) – Dai et al. function (Santandrea et al. 2020)

1.5

6

n =1

Average experimental P - g

1.0

4

0.5

2

2.0 g [mm]

s [mm]

0

0

0 0.2

0.4

0.6

0.8

1.0 1.2

1.4 1.6 1.8

2.0

0

0.5

1.0

1.5

Fig. 4. (a) CMLs obtained with the calibration process; (b) Analytical P n ( g ) functions.

4. Conclusions This paper reports the results of an experimental and analytical investigation on the shear stress transfer phenomenon of an SRG-concrete joints. Results of single-lap shear test were presented. SRG-composites comprised of one layer and two layers of textiles were considered. Independently of the number of textile layers, debonding at the fiber-matrix interface was observed. Based on the experimental results, a cohesive material law of the fiber-matrix interface was presented. The ratio of the bond capacity of specimens with two layers to the bond capacity of specimens with one layer was 1.48. This result is consistent with the dependency of the bond capacity on the number of layers arising from the fracture mechanics Mode II approach, provided that the failure mode does not change in the cases of one layer and two layers of textiles.