Time-resolved X-ray micro-CT: facilitating data visualization and analysis
Marijn Boone, Jan Dewanckele, Wesley De Boever, Arno Merkle
Time-resolved X-ray micro-CT imaging is rapidly emerging as an essential technique to gain a better understanding of fluid flow in porous materials. The push toward time-resolved dynamic 3D imaging has been spearheaded by synchrotron radiation facilities, with temporal resolutions going below one second. Recent advances in laboratory-based X-ray micro-CT imaging are pushing achievable temporal resolutions from hours down to seconds, enabling real-time imaging of dynamic processes.
Time-resolved X-ray micro-CT imaging generates a vast amount of data, both in acquisition (projection data) and after reconstruction (3D reconstructed volumes). Devising the right workflow strategy prior to reconstruction is essential to quickly identify the interesting moments in time and space and helps to reduce the amount of data that is generated. We present two methods for facilitating the analysis of dynamic micro-CT data and we apply this to a multiphase flow experiment and a reactive fluid flow experiment.
The first experiment is an in situ drainage experiment in a porous sintered glass. The drainage process is imaged using a dynamic imaging approach, where an uninterrupted series of projections is acquired while the sample is continuously rotating. This enables significantly more flexibility in reconstruction and allows one to tune the reconstruction windows to pinpoint the movement and reduce motion artefacts in a given reconstruction. This ‘sliding window’ approach is illustrated by pinpointing pore filling events in time and analyzing the time steps just before and after the pore filling.
Further analysis on reconstructed datasets using flip point detection or automated detection of voxel changes through time is also presented. Instead of analyzing the reconstructed 3D images individually, we directly incorporate the time dimension within the 3D analysis flow, by analyzing grey value changes of individual and clusters of voxels though time. This enables automatic pinpointing of changes inside the volume in time and space, with substantially less calculation overhead and time investment and greater flexibility compared to conventional post-3D reconstruction evaluation approaches (e.g. comparison of segmented volumes). We illustrate this by analyzing wormhole formation in a limestone sample in a reactive flow experiment.
