Utilizing Detector Filters for Noise Reduction in X-Ray Computer Tomography Scanning for the Inspection of the Structural Integrity of Additive Manufactured Metal Parts

Abstract

Abstract The recent development of industrial computer tomography (CT) has enabled inspection of the integrity of mechanical parts without physical sectioning. Using X-ray CT (XCT) presents many challenges prohibiting industry from widely implementing the technology. The existence of several variables such as filament current, filter material, and filter thickness is directly related to the presence of noise and influences the accuracy of the inspection process. Noise in the resultant reconstructed image is the result of low-energy X rays being absorbed by the detector causing large variations of brightness on the computer image; other noise sources such as detector variations and electronic noise are further sources of error. Additionally, the reconstruction method, which also has a response to the noise scatter, can contribute to the resulting image noise. The presence of noise can skew the resulting image and create the illusion of pores/defects that are not actually present, thus vastly compromising the results of the analysis. Noise reduction is vital in improving the reliability of CT imaging of additive manufactured components. This article investigates the possibility of reducing noise by using detector filters. The study will investigate the impact of source and detector filters (shown in fig. 1) on image quality. Two filter types of 100-μm-thick aluminum and 100-um-thick copper filters were used, and the results were compared to a conventional tube filter. The workpiece used in this study consisted of a 6-mm Ti6Al4V round bar with designed internal features ranging from 50 to 1,400 μm containing a mixture of voids, two filled with unfused powder and two unfilled. The diameter and depth of defects were characterized using focus variation microscopy and then scanned with a Nikon XTH225 industrial CT to compare the measured internal features. The analysis was carried out using VGStudio Max 3.0 (Volume Graphics, Germany) software package to evaluate surface determination and defects/porosity. Preliminary results indicate that using the aluminum detector filter can reduce the noise by 20 % when scanning the part under certain conditions. Also, for a low-magnification scan using the aluminum detector filter, no difference in noise was evident, but it was noted that the filament current could be reduced, potentially further reducing noise.