PSI - Issue 21

Hande Yavuz / Procedia Structural Integrity 21 (2019) 112–119 H. Yavuz/Structural Integrity Procedia 00 (2019) 000 – 000

116

5

The first group of preprocessed finite element models were free from holes, slots, notches and cracks. The second group of preprocessed finite element models were included 1 identical circular bolt hole with 6 mm in diameter for quasi-isotropic (QI, [0/+45/-45/90]), symmetric balanced (SYMB, [+45/-45/0/90] s ), and symmetric (SYM, [+45/+30/0/90] s ) laminates. Concentrated force was defined as a boundary condition for QI, SYMB, and SYM laminates in both models (Figure 3). To evaluate the damage characteristics of unnotched laminates, both sides of the parts were clamped and length and width were set to 250 mm and 25 mm, respectively as per ASTM D3039 (ASTM 2014). All the edges of elements have the same dimensions with element seed size of 6.0 mm with total number of elements of 164. To compare the damage response of various notched laminates as in unnotched ones, both sides of the notched laminates were clamped and their length and width were set to 200 mm and 36 mm, respectively as per ASTM D5766 standard (ASTM 2018). The edges of elements have the same dimensions with element seed size of 3.0 mm with total number of elements of 888. All plies were meshed with general purpose S4R shell elements; quadrilateral stress/displacement with 4-nodes shell element having reduced integration formulation. This element has 6 degrees of freedom per node (three degrees of freedom are node linear displacements; the others are node angular displacements) and computationally less expensive than S4 elements (Laulusa 2006). Based on the first order shear deformation theory, transverse shear stress was assumed constant through thickness of the shell. In the formulation of this element, thickness change as a function of in-plane deformation is allowed, thus it does not suffer from transverse shear locking (Abaqus 2012). No elements of different orders of shape functions were utilized; same type of elements was used throughout the whole study. The mesh convergence study was performed in both analyses by gradually varying mesh density from coarse to fine to count on damage distribution as well. Aspect ratio of elements were kept controlled to avoid inaccuracy of numerical solution especially around hole in notched laminates.

(a)

(b)

Fig. 3. Illustration of boundary conditions and loads of (a) unnotched laminate, (b) notched laminate.

3. Results and Discussion 3.1. Computational damage analysis for QI, SYMB, and SYM laminates

Damage analysis of AS4/3502 composite laminates have been widely studied since 1980s (Witcomb 1988, Martin 1995). Damage initiation and propagation requires intralaminar and interlaminar failure analysis along with properly defined elastic and strength properties. However, in many of these studies, material properties of unidirectional lay ups were not completely matched with each-other regardless of manufacturing details. This may cause an accumulation of multiple sequential variations through the computational structural design and analysis. In this study,