Speaker
Description
The thermal treatment of rocks is a frequently used method to initiate micro-cracks. This is done to study experimentally different physical phenomena related to micro-cracks. The effect of micro-cracks on the effective macroscopic properties can be quantified, for instance, by wave propagation measurements or porosity measurements. Micro X-Ray Computed Tomography (µXRCT), as a noninvasive imaging method, offers the possibility to have an insight into the 3D microstructure. With this method, it is possible to improve the understanding of the relation between the micro-scale and the effective properties. Since micro-cracks have a disadvantageous ratio between the crack aperture and the crack length, the imaging, as well as the subsequent segmentation, is a challenging task. In particular, the spatial resolution of µXRCT devices often come to its limitation to resolve the crack aperture reliable. Furthermore, an inherent noise and low contrast of the resulting dataset cause difficulties to achieve an accurate segmentation. Based on an in-house created µXRCT data set of a thermally treated Carrara marble sample, we applied and compared different segmentation methods. By all methods, the full 3D crack network can be successfully segmented. However, an approach based on a 2D Convolutional Neural Network (CNN) model shows the most promising result among the adopted methods. The segmented data is frequently used as direct input, for instance, in digital rock physics. Consequently, all the results depend on the quality of the imaging and the segmentation. Therefore, it is crucial to evaluate the quality of the segmentation. For this, a link back to macroscopic measurement results, for instance, the porosity can be used as an indicator. Besides different segmentation approaches, this contribution shows the importance of combining measurements of different scales, especially to evaluate segmentation methods.
| Time Block Preference | Time Block A (09:00-12:00 CET) |
|---|---|
| Acceptance of Terms and Conditions | Click here to agree |
