Probabilistic fracture of Ti–6Al–4V made through additive layer manufacturing

Abstract

The large deformation response of Ti–6Al–4V parts made through additive layer manufacturing is investigated. A wire-feed process is chosen instead of a powder process in an attempt to reduce the oxide contaminations of the final part. The experimental program includes uniaxial tension experiments along different part directions and fracture experiments on flat specimens with cut-outs covering stress states ranging from pure shear to equi-biaxial tension. More than 100 experiments are performed in total to characterize the randomness in the material's fracture response. It is found that the stress-strain response of the ALM material is comparable to that of Ti–6Al–4V sheet stock, while its average ductility is substantially lower. For example, for pure shear loading, the average strain to fracture for the ALM material is 0.47, while the mill product of the same alloy failed at a strain of 0.65. A probabilistic extension of the stress state dependent Hosford–Coulomb fracture initiation model is proposed to account for the significant standard deviation in the identified strains to fracture. Microscopic and surface strain field analysis demonstrate that the initiation and propagation of ductile fracture in the ALM material is strongly affected by the presence of prior-beta grains.

Introduction

The Ti–6Al–4V alloy is the most widely used titanium alloy with applications in jet engines, airframes and biomedical implants. Consequently, its mechanical behavior has been studied extensively. For example, Zhang et al. (2007) made use of a crystal plasticity model to describe the mechanical response of a Ti–6Al–4V alloy to cyclic loading. Przybyla and McDowell (2011) introduced a microstructure-sensitive extreme value probabilistic framework to compare the fatigue failure of four different Ti–6Al–4V microstructures. Khan et al. (2012) formulated an anisotropic criterion with tension-compression asymmetry to describe the yield behavior of Ti–6Al–4V. The tension/compression asymmetry, anisotropic yielding and anisotropic strain-hardening in Ti–6Al–4V ingots has also been characterized experimentally and modeled at the macroscopic level by Tuninetti et al. (2015). A theoretical model predicting the spacing of periodic adiabatic shear bands during high speed machining of Ti–6Al–4V has been prosed by Ye et al. (2013). Li et al. (2014) observed an increase in the ductility of Ti–6Al–4V during ring expansion experiments at strain rates above about 7 × 10³/s.

In aerospace engineering, Ti–6Al–4V components are traditionally manufactured through intense milling of bulk parts, the hot-forming of sheets and assembly welding (Tersing et al., 2012). Additive Layer Manufacturing (ALM) provides a promising cost-effective alternative to traditional machining. Historically, ALM has been intensively used for rapid prototyping, where shape is more important than the mechanical properties of the manufactured parts. Examples are the selective laser sintering with metal powders (Agarwala et al., 1995, Kruth et al., 2003, Levy et al., 2003) or the development of 3D printing with polymers (Levy et al., 2003, Wendel et al., 2008). The mechanical properties of components made from metal powders are often affected by contaminations that are associated with the high surface-to-volume ratio of powders. To the best of the authors' knowledge, ALM components built from powder stock are not yet used in the safety critical load carrying structures of modern jet engines.

Wire-feed processes feature a lower surface-to-volume ratio and thus a lower risk of contamination (Brandl et al., 2008). Other advantages of wire over powder include material availability, cost and quality. Brandl et al., 2008, Brandl et al., 2009 presented an argon flooded open ALM system composed of a Nd:YAG laser beam and a wire-feeder mounted on a 6-axis robot. Baufeld et al., 2009, Baufeld and Van der Biest, 2009, Baufeld et al., 2010 proposed the so-called Shaped Metal Deposition (SMD) process composed of a tungsten inert gas welding torch mounted on a 6-axis robot. The ALM part is built from wire stock on a 2-axis table inside a closed chamber with argon atmosphere. As compared to laser made parts, the SMD parts feature a lower nitrogen contamination (Baufeld et al., 2011).

The basic mechanical performance of ALM materials are typically characterized through uniaxial tension experiments (e.g. Baufeld et al., 2011). In view of using ALM parts in load carrying structures, the multi-axial material response needs to be known. A first objective of the present paper is to characterize experimentally and model numerically the average large deformation response of SMD made Ti–6Al–4V. Given the stochastic nature of the fracture response of ALM materials, a second objective of this work is to formulate and calibrate a probabilistic stress-state dependent fracture initiation model for SMD made Ti–6Al–4V. In Section 2, the macro- and microstructure of an SMD produced Ti–6Al–4V box structure is characterized. Subsequently, a comprehensive plasticity and fracture testing program is executed which includes the tensile testing of smooth, notched and central hole specimens as well as selected shear, bending and punch experiments. In Section 3, a non-associated plasticity model is presented along with a probabilistic formulation of the Hosford–Coulomb fracture initiation model. Finite element simulations are performed for all experiments to determine the loading paths to fracture in terms of the stress triaxiality, the Lode parameter and the equivalent plastic strain. Based on the hybrid experimental-numerical results, the four material parameters of the probabilistic fracture initiation model are identified. The final discussion is primarily concerned with the third objective of this work, which is the comparison of the observed ALM material fracture response with that of conventional Ti–6Al–4V sheet stock, and the identification of the physical origin of the observed randomness in the ALM-made Ti–6Al–4V fracture response.