| About this Abstract |
| Meeting |
2024 TMS Annual Meeting & Exhibition
|
| Symposium
|
Characterization of Minerals, Metals and Materials 2024: Process-Structure-Property Relations and New Technologies
|
| Presentation Title |
Characterization of Cements and Concretes Using 3D Automated Quantitative Mineralogy and Enhanced Deep-learning Reconstruction via X-ray Microscopy |
| Author(s) |
Ria L. Mitchell, John Provis, Dan Geddes, Giacomo Torelli, Antonia Yorkshire, Richard Taylor, Andy Holwell |
| On-Site Speaker (Planned) |
Ria L. Mitchell |
| Abstract Scope |
There is a growing need to quantify and characterize the micro-to-nano scale processes within cements and concretes in multiple dimensions and scales. Evolving areas of research include recycled concretes, self-healing concretes, nuclear waste containment, and lowering the CO2 footprint in cement/concrete production. They are prone to degradation and change over time, making processes such as hydration, cracking, corrosion, and mineralogical/phase transformations under differing conditions (temperature, stress/strain, humidity) important to characterize for these fundamental building materials.
Here, we have used non-destructive 3D imaging via X-ray Microscopy (XRM), combined with a novel 3D automated quantitative mineralogy technique, to spatially characterise the mineralogical phases in a variety of cements. Additionally, we have applied a deep learning-based reconstruction method to improve the quantification of pores and cracks. This combined workflow applies advanced denoising and vastly improves artefact detection in 10 mm+ diameter cores to enhance quantification, as well as accelerating throughput in tomographic investigations. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Characterization, Machine Learning, Phase Transformations |