PSI - Issue 13

Fuminori Yanagimoto et al. / Procedia Structural Integrity 13 (2018) 2095–2100 Author name / Structural Integrity Procedia 00 (2018) 000–000

2097

3

80

unit:mm

13

8

8

30

Upper Web

R0.5 w

w

59.5

Bonded surface

Upper flange surface

127

Flange

t f

a1

50

67.5- t

Lower Web

Lower flange surface

Fig. 1 Specimen geometry

3. Experiment 3.1. Experimental results

Three experiments were carried out using three types of flange thickness. The experiments are summarized in Table 2. In Mark A, the crack was arrested in the flange and did not enter the lower web. Although the cracks penetrated the flange and entered the Lower Web in both Mark B and C, the crack was arrested just after it entered the Lower Web in Mark B. In Mark C, the crack penetrated the Web. Table 2 Experimental results Mark Notch radius [mm] Flange thickness [mm] Fracture load [kN] Arrest position A 0.6 13 2.90 Flange B 1.0 5 3.45 Lower Web C 1.0 10 3.64 Propagate The crack behaviors observed by high speed camera are shown in Fig. 2, 3, and 4. Zero second is the time when the crack entered the flanges. In Mark A as shown in Fig. 2, the crack front kept to be shape similar to surface crack. In Mark B, although the crack front shape was initially like surface crack just after the cracks entered the flange, the crack became a through crack immediately. On the other hand, although the crack eventually a through crack, it required time for the crack become a through crack shape in Mark C. Therefore, the crack tip on the upper flange surface preceded on the lower flange surface and the crack front is not same as that in Mark B. 3.2. Mechanics of crack arrest in structural crack arrest design The differences among the above experimental results are attributed to the crack front shape. According to past analyses, SIF at the crack tip decreases when the depth of crack front from the surface is larger (Handa et al., 2014). The high speed observation clearly showed that such crack front shape effect is critical for structural crack arrest design. When the flange is thin such as in case of Mark B, the crack front contacts with the lower surface of the flange and becomes through crack. On the other hand, the crack front remained surface crack front shape until SIF decreased enough to stop crack propagation in Mark A because the flange thickness was enough. In Mark C, although the crack front contact the lower surface of the flange before the crack was arrested, the ligament of the flange was larger to reduce SIF of the crack in the web because the crack velocity on the lower surface was lower than that on the upper surface. This trend is also supported by the crack velocity history shown in Fig. 5. In Mark B, the crack velocity kept about 300 m⁄s even after the crack tip entered the flange. Furthermore, the crack velocities in the right and left in flange were much higher compared to those in other experiments. This higher velocity was caused by the crack front shape similar to through crack, which did not reduce SIF.