Fatigue Assessment of AZ31 Magnesium Alloy Weldments - Application of a Local Energy-Based Method

Abstract

Traditional stress-based fatigue criterions are highly dependent of geometry and loading mode and are often based on idealized geometries and loading conditions. These fatigue criteria are seldom applicable in general engineering assessments of structural components.

The scope of the thesis is therefore to apply the energy-based fatigue criterion of average local strain energy density (ASED) to welded joints of AZ31 magnesium alloy. A fatigue criterion which is both geometry and loading mode independent. The application of ASED as a new criterion was based on experimentally gathered fatigue data for three different weld geometries, namely: fully penetrated butt welded plate; non-penetrated butt welded plate; plate with a welded transverse stiffener. These three geometries were subjected to three different loading ratios of R=-1, R=0, and R=0.5.

Application of ASED criterion was performed with the aid of a numerical investigation using the Computer Aided Engineering (CAE) software Abaqus 6-14. Finite element models were constructed to best represent the actual occurring constraints, which were applied to the experimental specimens during testing. The models were then subjected to the stress range (∆σ) used experimentally.

The control radius (Rc) which defines the in-closed volume used in the fatigue evaluation, was found with the aid of a "best-fit" approach. Fatigue data from the three geometries under R=0 loading were tested with incrementally changing Rc. The Rc that yielded the best regression fit was select to be the appropriate one, resulting in the selection of Rc = 0.15 mm.

In addition to the numerical investigation, an analytical evaluation of ASED was preformed to compare the results with the FEM results. Through this evaluation it was found that the discrepancy between calculating ASED from the stress intensity factor (SIF, which is found through fine mesh FEM) and the extracted ASED was at 11.8%. It is important to mention that SIF was extracted by using ∼1500 elements, and ASED with 40, which gives a ratio of 1:37.5.