Taehwan Kim

As part of the concerted effort to reduce the carbon footprint of the concrete industry, limestone-calcined clay cement (LC3) has garnered attention as a highly sustainable alternative to OPC due to a high clinker replacement. This study explores the possibility of combining LC3 with a crystalline admixture to further improve the durability of LC3 systems. LC3 possesses the capability of self-healing, and the mechanisms are similar to those of regular cement mixes. Thus, it is expected that Crystalline Admixture (CA) addition will also improve the durability of LC3. Three identical mortar mixes using LC3 with varying CA dosages (0.5%, 1.0%, 1.5% of cement wt.) were designed for a durability study, alongside a reference mix that does not contain CA. Their performance was also compared with that of OPC concrete. A General Purpose (GP) cement mix without CA and a GP cement mix with 1.0% CA (cement wt.) were selected for comparison. In this study, artificial chloride exposure was employed to simulate real-life applications, and rebar corrosion was evaluated using electrochemical experiments and image analysis. The effect of CA on the compressive strength, volume of permeable voids (VPV), and electrical resistivity of LC3 mortar was found to be very similar to that of OPC. A faster self-healing rate, resulting in mitigation of corrosion activities, was noted in LC3 samples with 0.5%-1.0% CA. However, in this study, LC3 mortar did not outperform OPC in protecting steel rebar from corrosion and was completely outclassed in evaluations with 500 µm cracks. The findings suggest that self-healing can be utilised in LC3-based systems; however, the mixtures need further optimisation to match the performance of OPC specimens, especially for larger cracks above 300 µm width.

Cracks significantly affect the safety and durability of concrete structures, while existing image-based crack analysis methods primarily focus on crack detection from single images, with limited consideration of temporal crack evolution and physical prop agation behavior. Therefore, this study proposes a Physics-Informed Convolutional Long Short-Term Memory (PI-ConvLSTM) framework using time-series crack images and external load information. The proposed model combines a ConvLSTM-based spa tiotemporal prediction model with Partial Differential Equation (PDE)-based physical constraints and irreversibility constraints to represent physically consistent crack propagation behavior. Time-series crack datasets were constructed under 3-point and 4-point bending conditions for single-crack and complex-crack cases. Experimental results showed that the proposed model achieved an IoU of 0.912 and Dice coefficient of 0.954 for the single-crack condition and an IoU of 0.802 and Dice coefficient of 0.888 for the complex-crack condition while preserving crack continuity and propagation morphology better than conventional models during sequential prediction under progressive concrete loading.

Unlike blast furnace slag, steel furnace slags (SFS) have not been widely adopted as supplementary cementitious materials (SCMs), owing to their limited reactivity and free lime expansion. While single-stage aqueous CO2 mineralisation of SFS consumes free lime and improves reactivity, the reactivity of the carbonation products remains inferior to conventional SCMs such as fly ash. This study employed a multi-stage CO2 mineralisation approach – comprising separate acid-leaching and carbonation stages – on basic oxygen furnace slag (BOFS), electric arc furnace slag (EAFS), and ladle furnace slag (LFS). The acid-leaching stage was optimised by investigating the effect of acid concentration on elemental dissolution; a concentration of 1 M was identified as optimal, promoting selective dissolution of Ca while limiting the loss of Fe and Si, thereby supporting the efficient CaCO3 precipitation and solid-liquid separation. The acid-leached residues were characterised for chemical and mineralogical composition, textural properties, and pozzolanic behaviour. The characterisation revealed that acid leaching removed the low- and non-hydraulic phases (β-C2S, γ-C2S) and significantly increased the specific surface area of the residues, thereby enhancing their reactivity. The filtered leachate was subsequently carbonated via the addition of CO2 and NaOH, and its carbonation behaviour was examined through pH monitoring and thermogravimetric analysis. Multi-stage CO2 mineralisation significantly improved carbonation performance, achieving CO2 uptake and CaCO3 precipitation of 200–238 g/kg and 456–583 g/kg, respectively, compared with 120 g/kg and 265 g/kg for single-stage carbonation.