Resonant ultrasound spectroscopy for quality control of geometrically complex additively manufactured components

One of the barriers to the adoption of metal additively manufacturing by some industries is the absence of reliable part qualification procedures especially for components with complex geometries. Nondestructive evaluation (NDE) methods such as x-ray computed tomography (CT) and conventional ultrasonic testing (UT) have limitations in their abilities. X-ray CT is costly, hazardous, and offers limited resolution for larger components while many UT methods have limited applicability for inspection of parts with complex geometries or rough surfaces. Here, we conduct an integrated numerical and experimental study to investigate the feasibility of resonance ultrasound spectroscopy (RUS) as an alternative NDE method to inspect complex AM lattice structures with a varying number of missing struts. The results of numerical simulations including eigenfrequency and frequency domain analyses are promising, indicating that the pristine and defective lattice samples should theoretically be distinguishable. In addition, given a reference intact sample, characterizing the extent of the defect in terms of the number of missing struts appears feasible. We introduce a similarity metric to compare the spectra after being locally normalized. However, the experimental results are not as conclusive. Although pristine and defective lattices may be distinguished for some cases, the number of missing struts cannot be inferred. The discrepancies between the numerical and experimental results are likely due to our simplified assumptions about material properties in numerical simulations and/or the presence of other unaccounted defects and heterogeneities in test samples.

Automated summary

The study investigates the feasibility of using resonance ultrasound spectroscopy (RUS) as an alternative nondestructive evaluation (NDE) method to inspect complex AM lattice structures with various numbers of missing struts. Numerical simulations suggest that pristine and defective lattice samples can theoretically be distinguished, and a similarity metric can be used to compare spectra after normalization. However, experimental results are less clear, possibly due to simplified material assumptions in numerical simulations and/or the presence of unaccounted defects and heterogeneities in test samples.