Influence of surface roughness on computed tomography dimensional measurements
Introduction
X-ray computed tomography (CT) has emerged over the last years as an advanced dimensional measuring technique [1]. CT offers several important advantages, such as the possibility of performing non-contact and non-destructive analyses on difficult to access geometries, including internal features. However, the accuracy of CT dimensional measurements is affected by multiple error sources. A relevant influence factor is surface roughness of scanned parts [2], [3]. In particular, parts with highly rough surfaces – e.g. castings and additive manufactured (AM) parts – can cause significant deviations between CT and tactile dimensional measurements, due to the different surface filtering characteristics of the two measuring methods [4], [5]. Tactile coordinate measuring machines (CMMs) produce a mechanical low-pass filtering which depends on the probe size: the larger is the probe diameter, the higher is the probability that the acquired points will lie only on the peaks of the measured surface [6]. CT also generates a low-pass filtering, which has a different nature: it filters both peaks and valleys depending mainly on the unsharpness of CT images and on spatial discretization of the reconstructed volume using voxels of finite size [1].
Previous works have dealt with the influence of surface roughness on CT dimensional measurements. Schmitt and Niggemann [2] included the surface roughness as one of the uncertainty contributions related to the dimensional measurement of a sandblasted aluminium part with mean Rz of about 7 μm; a rectangular distribution was assumed and an uncertainty contribution of 0.6/2 times the mean Rz was calculated. Bartscher et al. [7] stated that an uncertainty contribution in the order of Rz/2 can be considered as an upper limit; in particular, effects of less than a quarter of Rz were determined for a cast aluminium workpiece with Rz up to 134 μm. Aloisi and Carmignato [4] showed that correcting for systematic effects due to roughness enables a significant decrease of measurement uncertainty when dealing with AM parts characterized by high surface roughness. Boeckmans et al. [8] observed an offset of Rp (maximum peak height of the profile in the sampling length) between CT and tactile CMM measurements in the case of turned surfaces. Although the abovementioned works document that an influence of surface roughness on CT dimensional measurements actually exist, the dependence of measurement deviations from surface morphology has not yet been studied in detail. In particular, no information is available so far about influences provided by different surface profiles (i.e. surfaces with different material distribution) and by CT surface filtering characteristics.
In this work, the influence of surface roughness on CT dimensional measurements is investigated considering the combined effect of surface morphology and surface filtering characteristics of CT measurements. In particular, the work aims at determining systematic errors of CT measurements with respect to reference tactile measurements. Since the effects of roughness for tactile measurements have already been studied [6], this work focuses on the effects for CT measurements, considering the deviations from tactile measured points that ideally lie only on the peaks of the measured surface. Determining the systematic effects is relevant especially for measurements of surfaces with high roughness, as they produce larger bias, which should be corrected for (rather than considered as uncertainty).
Section snippets
Materials and methods
Experimental analyses were performed using various CT spatial resolution settings on samples with periodic roughness profiles having different bearing properties, i.e. different Abbott curves [9] (Sections 2.1 and 3.1). In addition, to extend the study, simulation analyses were implemented, considering a wider set of roughness profiles, imaged with a range of different voxel sizes (Sections 2.2 and 3.2). Further influence factors such as focal spot size, surface determination method and fitting
Results
In this section, results obtained from experimental investigations and numerical simulations are presented and compared.
Conclusions
In this paper, the influence of surface roughness on CT dimensional measurements was investigated by experimental and simulation analyses on periodic roughness profiles. CT scans of FDM and turned cylinders featuring various surface topographies were performed using different voxel sizes up to 125 μm. From the experimental results, diameters measured by CT using least-squares fitting were found to be smaller than reference measurements of approximately 2Rp on average. This systematic difference
Acknowledgements
This work received funding from University of Padova, projects Nr. CPDA151522/15 and BIRD167853/16.
References (13)
Computed Tomography for Dimensional Metrology
CIRP Annals
(2011)- et al.
Influence of Surface Roughness on X-ray Computed Tomography Dimensional Measurements of Additive Manufactured Parts
Case Studies in Nondestructive Testing and Evaluation
(2016) - et al.
Probing Systems in Dimensional Metrology
CIRP Annals
(2004) Accuracy of Industrial Computed Tomography Measurements: Experimental Results from an International Comparison
CIRP Annals
(2012)Accurate Subpixel Edge Location Based on Partial Area Effect
Image and Vision Computing
(2013)- et al.
Uncertainty in Measurement for X-ray-computed Tomography Using Calibrated Work Pieces
Measurement Science and Technology
(2010)
Cited by (42)
Comparison of X-ray computed tomography and coordinate-measuring system dimensional measurement for additive manufacturing parts using physical and simulation methods
2024, Measurement: Journal of the International Measurement ConfederationTwo different experimental approaches for the uncertainty determination of X-ray computed tomography dimensional measurements on lattice structures
2023, CIRP Journal of Manufacturing Science and TechnologyRoughness evaluation of turned composite surfaces by analysis of the shape of autocorrelation function
2023, Measurement: Journal of the International Measurement ConfederationAn automated surface determination approach for computed tomography
2022, NDT and E InternationalToward traceable XCT measurement of AM lattice structures: Uncertainty in calibrated reference object measurement
2022, Precision EngineeringCitation Excerpt :Thus, it is important to understand the uncertainty in XCT measurement of lattices prior to the inclusion of these factors. The additional contribution by roughness in the measurement of AM lattices can then be accounted for at a later time using previously proposed methods [32]. Thus, further study is needed to evaluate the uncertainty in XCT measurement of a lattice component under conditions (XCT settings, measured features, workpiece construction) that are similar to the evaluation of AM lattice structures.