13th International Conference on Fracture June 16–21, 2013, Beijing, China -8- 0 5 10 15 20 25 30 500 1000 1500 2000 2500 3000 3500 4000 4500 Δa (mm) P(N) : Test : δ 0 =0.10mm, α c =4deg : δ 0 =0.10mm, α c =3deg : δ 0 =0.10mm, α c =5deg Fig.10 Crack extension against load of C(T) specimen (Δa-P curve)[10] 3.2 Experimental and predicted stable crack growth With the critical δ0 and αc values determined, the main task for stable crack growth analysis of sheets with MSD is to solve the strip yield model to obtain the crack opening displacement for cracked panels. As a result, the efficiency and accuracy of the stable crack growth prediction are much dependent on the method for solving the strip yield model. The unified method was applied to determine the COD for stable crack growth analysis. Here, the results for sheets (600mm×1140mm) with five different cracks shown in Fig.11a are presented as examples. Figure 11b and c shows the predicted and experimental crack growth behaviors for two sheets with five different cracks. In both cases, the length of the side cracks is 15mm. The length of the center lead crack and ligaments between cracks are different. In Fig.11b, the experimental and predicted maximum residual strengths occurred after the fracture of all the ligaments. In Fig.11c, the sheet failed immediately at the fracture of ligament l2 shown. Due to the quick fracture of this specimen, the crack growth information for the outside crack tip was not recorded. However, the predicted result was able to describe the entire crack growth process, see the solid lines in Fig.11. When the applied load reached at the maximum residual strength 99.0KN, all the crack tips started to grow, resulting in the fracture of the whole sheet. More detailed information about the experiment and prediction on various MSD configurations was given in Ref.[10]. The elastic-plastic FEM is widely used to predict the residual strength for cracked structures. Using the “plane strain core” model [16] and CTOD criterion embedded in ABAQUS software, the ligament fracture loads and residual strengths for some of the MSD configurations were given in Ref.[10]. It is found that the accuracy of both methods is comparable. However, the computational and modeling demands are quite different. For a given MSD configuration, it takes at least two hours (a computer with a Pentium® Dual-Core CPU E5300@2.60GHz and 3.00GB RAM) to complete a residual strength prediction by using FEM. Yet, it does not include the time for creating the finite element model. In order to model the crack growth, the “debond” technique in ABAQUS was used. The FEM involves material and contact non-linear analysis [10]. Rich experiences on finite element modeling and analysis are required. However, for most of the MSD configuration
RkJQdWJsaXNoZXIy MjM0NDE=