**5. Results and discussion**

Delamination in composites can occur due to tensile stress Mode I. In the present study, AG-50kNG Shimadzu universal testing machine (see in **Figure 5**) has been used to record the load-deflection curves for calculating fracture toughness for DCB specimen and **Figure 6** shows the initial and the progress of fracture in the machine. Specimens have been designed by referring ASTM standards, DIN EN ISO 75-1, 75–3. The woven [0°/90°]16 and woven [±45°]16 specimens made up of woven-fabric-reinforced glass/epoxy composite materials were studied and the mechanical properties and interlaminar fracture toughness *K*IC value were obtained experimentally by using DCB and numerically by using FEM. Four types of tests were performed. Each of them includes five tests corresponding to combinations of different stacking sequence of woven and width. For the DCB specimens, the delamination was generated during manufacturing by placing teflon between the mid-layer therefore an initial delamination or crack was formed. In this experiment, the output data have been recorded and graphs have been plotted as required.

The mean values for every individual test groups can be shown in **Figures 11**–**14**. The SERR, *G*IC can be derived from the recorded load–displacement data by using compliance calibration method and Eq. (2) to determine Mode I interlaminar fracture toughness *K*IC given in **Table 3**. The average *G*IC value of (0°/90°) fiber orientation specimens, for different width with error bars are ±1 standard deviations and the influence of width on average *G*IC for (±45°) fiber orientation are shown with error bars are ±1 standard deviations in **Figure 15a** and **b**, respectively. It has been obtained that woven [0°/90°]16 specimens have higher fracture toughness values than, woven [±45°]16 specimens to show the effect of different lay-up,.

**Figures 7**–**10** show a typical load–displacement curves for woven [0°/90°]16 and woven [±45°]16 specimens, respectively. According to ASTM 5528–1 the loading process consists of two steps. In step1, the loading was continued with crosshead speed 2 mm/min up to the displacement of about 28 mm and until the first crack propagation *Δa* 2–3 mm occurs. Then the crosshead was returned to zero point with 25 mm/min constant displacement speed to form a natural pre-crack. In the step 2, second loading was continued with crosshead speed 2 mm/min up to the displacement of about 80 mm, and after then the crosshead was returned to zero point with 25 mm/min constant displacement speed. It is seen that, some degree of nonlinearity and small permanent deformations were visible in the unloading curves of specimens. The non-linearity can be partly attributed to the specimens undergoing relatively large displacements towards the end of the tests.

As it is seen from the **Table 3**, the fracture toughness values decrease 2.20% in percentage for the woven [0°/90°]16 specimens 25 mm and 40 mm in width, while the specimen width increases. Similarly, for woven [±45°]16, the fracture toughness values decrease 1.27% in percentage, while the specimen width increases. In other words, the energy release rate increases slightly with decreasing width length for both type of woven specimens. Also it is seen that, the obtained fracture toughness value of woven [0°/90°]16 specimen with 25 mm width is greater 3.93% than woven [±45°]16 specimen. Similarly, it has been found that the fracture toughness value of woven [0°/90°]16 specimen 40 mm in width is greater 2.96% than woven [±45°]16 specimens 40 mm in width.

The obtained average maximum load and the critical SERR, *G*IC achieved from Eqs. (1) and (2) for plane strain case, have been given in **Table 2**. In the finite element analysis, the stress intensity factor *K*IC was obtained for DCB specimens numerically. The comparison of experimental and numerical results of *K*IC values are given in **Table 3**. Finite element analysis was conducted to validate the closed

form solution. Results show a good agreement between analytical solutions, numerical simulation shown as in **Table 3**. As it is seen from the table, the differences between numerical and analytical solutions are approximately 2.31–2.69% percentage for 25 mm width specimen and 3.13–3.16% percentage for 40 mm width specimen. It can be said that, when the width range increases the differences between the numerical and analytical solutions may change small amount because of meshing type at delamination through thickness.
