Materials Data Science for Microstructural Characterization of Archaeological Concrete

Ancient Roman concrete presents exceptional durability, low-carbon footprint, and interlocking minerals that add cohesion to the final composition. Understanding of the structural characteristics of these materials using X-ray tomography (XRT) is of paramount importance in the process of designing future materials with similar complex heterogeneous structures. We introduce Materials Data Science algorithms centered on image analysis of XRT that support inspection and quantification of microstructure from ancient Roman concrete samples. By using XRT imaging, we access properties of two concrete samples in terms of three different material phases as well as estimation of materials fraction, visualization of the porous network and density gradients. These samples present remarkable durability in comparison with the concrete using Portland cement and nonreactive aggregates. Internal structures and respective organization might be the key to construction durability as these samples come from ocean-submersed archeological findings dated from about two thousand years ago. These are preliminary results that highlight the advantages of using non-destructive 3D XRT combined with computer vision and machine learning methods for systematic characterization of complex and irreproducible materials such as archeological samples. One significant impact of this work is the ability to reduce the amount of data for several computations to be held at minimalistic computational infrastructure, near real-time, and potentially during beamtime while materials scientists are still at the imaging facilities.

[1]  Emmanuelle Gouillart,et al.  scikit-image: image processing in Python , 2014, PeerJ.

[2]  Marie D. Jackson,et al.  Unlocking the secrets of Al-tobermorite in Roman seawater concrete , 2013 .

[3]  Marie D. Jackson,et al.  Phillipsite and Al-tobermorite mineral cements produced through low-temperature water-rock reactions in Roman marine concrete , 2017 .

[4]  Pascal Fua,et al.  SLIC Superpixels Compared to State-of-the-Art Superpixel Methods , 2012, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[5]  R. Ritchie,et al.  Real-time Quantitative Imaging of Failure Events in Materials under Load at Temperatures above 1,600 , 2012 .

[6]  Singanallur Venkatakrishnan,et al.  Convolutional neural networks for grazing incidence x-ray scattering patterns: thin film structure identification , 2019, MRS Communications.

[7]  P. Monteiro,et al.  Multiscale X-ray tomography of cementitious materials: A review , 2020, Cement and Concrete Research.

[8]  Sabita Khatri,et al.  Implementing WEKA for medical data classification and early disease prediction , 2017, 2017 3rd International Conference on Computational Intelligence & Communication Technology (CICT).

[9]  P. Noël,et al.  The evolution of image reconstruction for CT—from filtered back projection to artificial intelligence , 2018, European radiology.

[10]  Marie D. Jackson,et al.  Building for Eternity: The History and Technology of Roman Concrete Engineering in the Sea , 2014 .

[11]  Ian H. Witten,et al.  Data mining: practical machine learning tools and techniques, 3rd Edition , 1999 .

[12]  Matti Pietikäinen,et al.  Performance evaluation of texture measures with classification based on Kullback discrimination of distributions , 1994, Proceedings of 12th International Conference on Pattern Recognition.

[13]  Francesco De Carlo,et al.  TomoPy: a framework for the analysis of synchrotron tomographic data , 2014, Journal of synchrotron radiation.

[14]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

[15]  Robbie M. Andrew,et al.  Supplementary material to "Global CO2 Emissions from Cement Production" , 2017 .

[16]  Dilworth Y. Parkinson,et al.  Evaluating Fiber Detection Models Using Neural Networks , 2019, ISVC.

[17]  Guido Gerig,et al.  3D Multi-scale line filter for segmentation and visualization of curvilinear structures in medical images , 1997, CVRMed.

[18]  N. Kelez,et al.  Beam Measurements and Upgrade at BL 7.2, The Second Diagnostics Beamline of the Advanced Light Source , 2005, Proceedings of the 2005 Particle Accelerator Conference.

[19]  Chao Yang,et al.  IDEAL: Images Across Domains, Experiments, Algorithms and Learning , 2016 .

[20]  Geoffrey E. Hinton,et al.  Deep Learning , 2015, Nature.