Accessible and reproducible measurement of carbonation depth in cementitious materials using an open-source image analysis framework

Cement-based materials are among the most widely used materials worldwide, and carbonation is one of the main mechanisms affecting their long-term durability. Carbonation depth is a key measurand for assessing the durability of cementitious materials. However, its determination typically relies on manual interpretation of phenolphthalein treated fracture surfaces, leading to operator-dependent variability and limited reproducibility. This study presents Carbon Depth, an image-based measurement framework for automated quantification of carbonation depth from digital images. The workflow integrates image acquisition, preprocessing, color-based segmentation, carbonation front detection, and automated extraction of depth metrics. The method was applied to three cementitious systems (latex-modified, silica-modified, and reference Class G) exposed to accelerated carbonation for 30, 60, and 90 days after initial curing (CO 2 /H 2 S, 3000 psi, 60 °C). A total of 54 fractured surfaces were analyzed and compared with manual measurements obtained by two independent operators. The results showed a strong agreement, with a mean bias of 0.10 mm, RMSE of 0.22 mm, and R 2 of 0.973. Bland–Altman analysis indicated consistent performance across the measurement range. A semi-synthetic dataset (14 images) was further used to assess robustness under varying optical conditions, yielding R 2 = 0.9991 and RMSE = 0.22 mm. The proposed approach preserves experimental trends and provides a reproducible, low-cost measurement solution with reduced operator dependency. The full processing pipeline is made available as an open-source implementation, enabling transparency, reproducibility, and adaptation across different laboratory and field conditions