Quantitative analysis of the correlation between dual-deposition parameters and porosity in wire arc additive manufactured Ti-22V-4Al alloys

The highly interdependent nature of deposition parameters in wire arc additive manufacturing (WAAM) complicates process optimisation for effective defect mitigation, particularly in controlling porosity formation. This study quantitatively investigates the influence of concurrently varied wire feed speed (WFS) and travel speed (TS) on porosity characteristics in Ti-22V-4Al titanium alloys fabricated by WAAM, with a focus on understanding the interplay of deposition parameters. Quantitative examination of porosity volume, morphology, sphericity, and spatial distribution was conducted leveraging machine learning (ML)-based segmentation, as Mask R-CNN provided more accurate and reliable results than traditional Fiji thresholding, particularly for connected and irregular pores. Synchrotron micro-computed tomography (micro-CT) was employed for its high resolution and efficient processing capabilities. A critical TS threshold of 105.14 mm/min was identified from the developed empirical model, where the relationship between WFS and porosity shifts, whilst the correlation between TS and porosity is influenced by both TS and WFS, highlighting the non-monotonic effect of the dual deposition parameters on porosity. This threshold provides a process-specific reference point for optimising parameter selection in industrial WAAM applications aimed at porosity mitigation. The study also found that porosity exhibits a periodic layer-by-layer distribution pattern, with irregular pores predominantly concentrated at the component's centre, while small gas pores dominate the porosity across all examined samples. The superiority of ML models over traditional methods in porosity analysis was demonstrated in terms of accuracy and consistency. This work provides guidance on the synergistic optimisation of deposition parameters in titanium alloys fabricated via WAAM for porosity mitigation. • Machine learning-based quantitative porosity examination. • Interplay and impact of multivariate printing parameters on porosity. • Periodic porosity pattern with central concentration of irregular pores. • Small gas pores dominate the porosity across all samples.