PSI - Issue 28

Thomas Bergmayr et al. / Procedia Structural Integrity 28 (2020) 1473–1480 C.V. Thomas Bergmayr et al. / Structural Integrity Procedia 00 (2020) 000–000

1476

4

Table 1. Material properties of the sandwich structure. Section t (mm) E 11 (GPa)

E 22 (GPa)

E 33 (GPa)

G 12 (GPa)

G 13 (GPa)

G 23 (GPa)

GFRP fabric

0.125

22.55 0.001

20.90 0.001

-

4.5

3.5

3.5

Nomex honeycomb Aluminum fittings

15

0.5

0.001

0.066

0.034

-

70

70

70

26.3

26.3

26.3

Table 2. Debonding states and its corresponding sizes. Debonding states Debonding area (mm 2 )

Debonding diameter (mm)

Percentage of one inch (%)

DS0 DS1 DS2 DS3 DS4

142.51 253.35 395.87 570.05 775.90

19.05

75

25.4

100 125 150 200

31.75

38.1 50.8

z

edge tting (EF)

test rig attachment

center hinge tting (CHF)

sandwich panel

ball bearing

load introduction (torque free)

x

y

whifetree

load cell

hydraulic cylinder

Fig. 3. The experimental set-up of the reduced spoiler model test rig.

the di ff erence of deformation between the reference FE model of the real spoiler loaded by the defined load case and the FE model of the idealized spoiler loaded by four local forces. However, details on the used test rig and numerical optimization of the load introduction will be published soon (Winklberger, 2020). The whole set-up is shown in Fig. 3. For recording the strains and deformations at the top face layer, a digital image correlation (DIC) system was used.

4. Finite Element Model

The test set up is modeled with the Finite Element Method (FEM) using the FEM Software package Abaqus. In order to to reduce computation time and in the same time increase the accuracy of the DIC system, by means of reducing the monitored area the simulation model is designed with the submodel technique, depicted in Fig. 4. The implemented face layer debonding at the sandwich structure was modeled with removing the tie constraint between sandwich core and skin in the damaged area. After performing a buckling analysis, the resulting buckle deformation is scaled and implemented as imperfection at the load analysis step of a successive analysis. Whereas, the global FE-model is geometrically linear, the submodel with the implemented imperfection is geometrically non-linear. The loading of the idealized spoiler Fe-model was performed via concentrated forces acting on the undamaged skin, imitating the load introduction of the whi ffl etree.