Fatigue characterisation of additively manufactured aluminium alloy

Additive manufacturing (AM) is becoming increasingly popular in producing various engineered structures, with many applications in aerospace, automotive engines and marine vehicles, both for novel designs and for repairing existing parts. With AM, complex lightweight designs can be achieved, reducing assembly and machining time and are, therefore, more fuel-efficient and environmentally friendly than conventional manufacturing. Despite this, AM parts are still in their infancy within the manufacturing industry since structural reliability is an essential prerequisite when it comes to applications in the real world. Recent years have seen some attention given to aluminium parts produced by laser powder bed fusion (L-PBF). Nevertheless, one of the critical challenges is the presence of defects that compromise the mechanical properties of the parts, especially when it comes to fatigue life. Considering these aspects, this study investigates the effects of defects, such as surface roughness and internal porosity, on the fatigue performance and failure mechanisms of L-PBF AlSi10Mg alloys to predict FAIL-SAFE as well as damage tolerant design criteria. This objective is pursued using both experimental and numerical approaches. To achieve the project objective, the thesis work starts with an explanation of the experimental results. For this purpose, the fatigue properties of L-PBF AlSi10Mg samples are investigated in the Z-direction to determine the effect of different processing conditions, such as as-built (AB), machined (M), machined & polished (M&P) and heat-treated (HT). The same AM system, powder and machine parameters are used to produce all samples. The samples are tested under tension-tension (R= 0.1), tension-compression (R=-1) and strain loading (strain amplitude = 0.25% and 0.40%) conditions. Surface roughness and porosity are two of the main defects considered in this study. For L-PBF AlSi10Mg, the as-built samples exhibit higher surface roughness (1.5-2 µm) than the machined (0.8-1.0 µm) or polished (0.3-0.75 µm) ones. Under similar loading conditions, the M or M&P samples have a longer fatigue life than the AB samples. T6 heat treatment (5000C for 4 hrs) of AlSi10Mg results in further improvement in fatigue life for both AB and the M samples; however, the improvement is significantly higher for the M samples (15 times higher than the non-heat-treated). For the AB samples, surface roughness was found to be the most important factor affecting fatigue life. However, subsurface porosity becomes the dominant factor affecting fatigue failure for M or M&P samples with relatively low surface roughness. Pore size and location effects are analysed for M samples using X-ray computed tomography (CT) and fracture surface data, adapting linear elastic fracture mechanics theory. The findings indicate that the critical stress intensity factor (K), a measure of both pore location and size, are closely related to fatigue life. An empirical formula is developed to predict fatigue life using X-ray CT porosity data. Furthermore, crack growth patterns, stable crack growth regions (initiation and propagation of cracks) and unstable crack growth regions (final fracture) are observed on the fracture surfaces of different samples (non-heat-treated, NHT, and heat-treated, HT). Ductile failure is evidenced by uniform and larger pits in the final fracture region of the HT samples. Under stress loading, all NHT samples exhibit brittle fracture regions. However, at similar loading, the fracture surfaces of R=0.1 show a smaller stable crack growth region (10%) than those of R=-1 (23%), indicating faster crack propagation and shorter fatigue life. Since this alloy does not exhibit strain hardening, strain-loaded samples show brittle fractures like stress-loaded fracture surfaces. The fracture surface can be used to identify the critical reasons for the failure of L-PBF AlSi10Mg and the differences in fatigue life. In conjunction with the experimental tests, a finite element analysis (FE) is also performed to determine the stresses around the pores and the resulting effects on the fatigue life of the L-PBF AlSi10Mg alloy. The stress field is calculated for individual pore models to evaluate the stress concentration as a function of pore location and size. Later, a multi-scale model FE is developed using representative volume elements (RVE) and based on porosity data from computed tomography (CT), which predicts fatigue life with 90% accuracy with experimental results. The predicted fatigue cycles are calculated using the Fe-Safe software's 'Rainflow' cycle algorithm, using stress-strain data obtained from the proposed FE model. In the design phase of AM components, the proposed model can be used as an alternative to experimental analysis to generate S-N curves for arbitrary loading conditions based on porosity conditions. Another essential parameter for load-carrying members is the fatigue crack growth rate (FCGR). The FCGR of any alloy varies with the load ratio (R) and generating individual FCG curves is tedious. Therefore, it is imperative to define the FCG curves (da/dN versus ΔK) with a generic mathematical representation for AM alloys using FCG similitude laws. The similitude study is performed for the AM alloys Ti-6Al-4V, AlSi10Mg and 316L stainless steel, using Hartman and Schijve's FCC equation and some modified Paris’ law equations (Walker, Kujawski, and Huang-Moan). Interestingly, the similitude equations obtained here predict the FCGR of AM alloys regardless of the process, build, and test conditions. The results of this work can help optimise the materials for future mechanical or structural designs to achieve structural integrity and durability.