PSI - Issue 37

Rogério Lopes et al. / Procedia Structural Integrity 37 (2022) 123–130 R. F. Lopes et al./ Structural Integrity Procedia 00 (2019) 000 – 000

125

3

= 7.8 / 3 = 355( )

= 7.8 / 3 = 280( )

= 2.5 / 3 = 80 ( )

Density

Yield stress Ultimate tensile stress 18 % In this paper, the numerical analysis contemplates two distinct guidelines: (i) an analysis using the Abaqus software, where the total displacement obtained experimentally will be applied as a boundary condition on the rigid body part (impactor), and (ii) the analysis using PamCrash , also imposing the same displacement. Both approaches (i) and (ii) use different software tools, however, with identical formulations. In this way, we will have a basis for comparison between the two tools for the same problem and a direct comparison between the experimental model and the numerical model. 2.1. Numerical modelling using Abaqus Initially, a midsurface was firstly assigned to the solid parts, in order to generate the shell elements. A centerline was also created for the parts modelled with beam elements. Table 2 shows the number of elements, including its formulations, and the number of and nodes, for the entire mesh. = 680 ( ) = 410 ( ) = 250 ( ) Total elongation 20 % 28 %

Table 2. Definition of the mesh used in the model via Abaqus

Element type

Nº

Total number of nodes

S4R (4-node doubly curved thin or thick shell, reduced integration, hourglass control, finite membrane strains) S3 (3-node triangular general-purpose shell, finite membrane strains)

40948

1

S8R (8-node doubly curved thick shell, reduced integration)

2548 189

B31 (2-node linear beam in space)

50493

a)

b)

Fig. 2. Door discretization via Abaqus; a) back view and b) front view

Subsequently, all interactions that existed in this system were defined. Mostly of the connections between the components were welded connections, with the exception of the front panel which was bonded to the door by adhesive. This essay also serves to verify if it is possible to treat welded connections as tied connections, in which all degrees of freedom would be connected. Regarding the boundary conditions, this system is supported by 3 supports, while the impactor is imposed a displacement that will be the same as that verified experimentally. In the Fig. 3 the boundary conditions are represented. Respecting the simulation parameters, due to the high number of nonlinearities involved, a dynamic simulation, at low speed (Abaqus explicit) was conducted. In order to improve the buckling behavior, imperfections were introduced in the mesh using numerical techniques.