Automated image analysis techniques for assessing the corrosion in historical cementitious composites: An experimental analysis of ferrocement samples

For the ACE industry, the preservation of capacity and aesthetics in historical structures raises issues. Existing structures and heritage buildings often encounter degradation processes driven by extreme environmental conditions. One of the consequences of inadequate protection against these actions is corrosion. Such effects are difficult to assess and quantify, especially for structures constructed with thin cross-sections and highly dense reinforcement materials like “Ferrocement”. These data are essential for evaluating the capacity and structural integrity of existing structures. An extensive experimental campaign was conducted on replicated samples to overcome and analyse the corrosion degrading phenomenon for historical materials such as ferrocement. To replicate those, the real ferrocement materials that “Ing. Pier Luigi Nervi” utilized at the “Torino Exposition” (To-Expo) centre were reproduced in the lab. This study builds upon two prior experimental campaigns, utilizing the mechanical testing results in a total of 48 samples and photographic datasets originally obtained weekly during the accelerated ageing process by previous researchers. In this paper, two types of studies are presented: experimental work (Microscopic analysis) and computational work (Image Processing). The initial part of the newly performed investigations in this paper involved cutting the samples in cross-section and obtaining microscopic images for the purpose of observing patterns and quantifying corrosion. To see how corrosion was distributed, several portions were cut close to the cracks and other critical regions. Correlations with fracture behaviours were investigated by measuring the corroded or lost area in each sample. A non-destructive technique, such as image processing, tracked the surface corrosion manifestation, while microscopic investigation described the interior corrosion. As a supplement to the experimental work, the second part of the study used sophisticated digital image processing techniques, like image colour segmentation, to measure the rate of corrosion over time and the severity of localized corrosion based on the surface rust stain. To meet this goal, RGB thresholds have been applied to the segmenting colour space, and a rules-based algorithm has been implemented. Based on the results, the impact of corrosion on mechanical characteristics and corrosion distribution on reinforcement was investigated. Furthermore, the relationship between section corrosion and surface corrosion with ultimate load, ultimate deflection, and covermeter data was examined.