A parametric study of segmentation thresholds for X-ray CT porosity characterisation in composite materials


Porosity in composite structures is one of the most critical manufacturing defects because it leads to a reduction of mechanical performance, particularly matrix-dominated properties. Most published research work has focused on correlating the average through-thickness void volume fraction to the mechanical properties. However, individual void features, such as morphology, size and location, have now been shown to be more important in predicting failure.

Non-destructive testing (NDT) techniques are used to inspect composite structures after manufacture in order to identify the extent of any defects and to help assess their potential impact on mechanical performance. Micro X-ray computed tomography is a promising NDT technique which can give information about pore location, size and morphology in three dimensions. To identify and characterise voids in the specimen, an image ‘segmentation’ must be applied to the full CT dataset, which is simply a rule-based decision about whether a given voxel is inside a pore or not.

This work analyses the effectiveness of previously accepted threshold methods for segmentation across a range of values of several different material and porosity parameters. Then a new, simple, rigorous, reproducible and accurate CT-segmentation thresholding method is proposed to characterise voids in a wider range of composites and with reduced errors than the accepted segmentation methods. The sources of uncertainties in the results are investigated using modelling and simulation, and recommendations are made to minimise these uncertainties.

The authors would like to acknowledge Rolls-Royce plc for their support of this research through the Composites University Technology Centre (UTC) at the University of Bristol, UK.