PSI - Issue 13

Adam Smith et al. / Procedia Structural Integrity 13 (2018) 566–570 Smith / Structural Integrity Procedia 00 (2018) 000 – 000

568

3

2. Method

2.1. Finite Element Model

The specimen geometry, loading, and boundary conditions used in the present study were based on experiments reported by Kilpatrick and Cargil (1981) and Kilpatrick (1997), which involved cyclically loaded tee-butt welded specimens subjected to three-point bending using an R-ratio of 0 (Fig. 2(a)). The geometry of the test specimen was simplified for finite-element modelling so that the load could be applied to the central region at the top of the specimen (50 mm either side of the centre-line). The load applied to the model cycled from a maximum of 219 kN to a minimum of 0 kN (Fig. 2(b)). The maximum load (P max ) was calculated using simple bending theory (equation 1) to ensure that the required maximum nominal stress (S max ) was achieved and that the weld toe was subjected to tensile loading (assuming no residual stresses). This maximum stress combined with an R-ratio of 0 led to a stress range ( ∆ S) of 518 MPa, which was equal to the stress range applied in the work reported by Kilpatrick (1997). In equation (1), W is the plate width (200 mm), t is the plate thickness (38 mm), and L is the distance from the weld toe to the vertical loading support (228 mm).

(1)

Fig. 2. (a) Diagram showing the geometry (dimensions in mm) and loading configuration of fatigue test specimens used by Kilpatrick and Cargil (1981). (b) Diagram showing the geometry, loading, and boundary conditions used in finite-element and fracture-mechanics modelling in the present study. The geometry was based on the specimen shown in (a), except the web was altered to simplify the loading and meshing. The shaded area in (b) shows the volume that was re-meshed by FRANC3D to embed a crack at location A (at the weld toe). The values of t web and W w were 25 mm and 45 mm, respectively, and the weld angle assumed for the model (not shown on the diagrams) was 45°. The finite-element model was generated in Abaqus FEA (2016) and first-order brick elements were used to mesh the model (8 elements through the plate thickness and 20 elements across the width of the plate). The volume of the model in which the crack was located was re-meshed using second-order elements by the fracture-mechanics modelling software (FRANC3D, Version 7.0.6). A linear-elastic material model was applied with a Po isson’s ratio of 0.3 and a Young’s modulus of 207 GPa.