The synthesis of N-A-S-H gel with high Si/Al ratios (>2) has been rarely reported in the literature, leaving the establishment of a reliable synthesis route as an open challenge. This paper aims to synthesize N-A-S-H gels with Si/Al ratios ranging from 1 to 3 and establish their thermodynamic database. The effects of reaction temperature, reaction time, initial Si/Al, concentration of reactants and pH of the matrix on the Si/Al ratios of the synthesized N-A-S-H gel were investigated. Results showed that N-A-S-H gels with target Si/Al ratios can be synthesized by controlling the concentration of reactants, pH and initial Si/Al ratios. The solubility products of the obtained N-A-S-H gels were determined via dissolution tests at different temperatures, to determine thermodynamic data. The development of this experimentally derived thermodynamic database of N-A-S-H gels constitutes a crucial step in the advancement of thermodynamic modeling of geopolymer, providing valuable insight into geopolymer reactions and phase assemblages.

In this paper, the atomic structures of sodium aluminosilicate hydrate (N–A–S–H) gels with different Si/Al ratios are studied by molecular dynamics simulation. An N–A–S–H gel model was obtained from the polymerization of Si(OH)₄ and Al(OH)₃ monomers with the use of a reactive force field (ReaxFF). The simulated atomic structural features, such as the bond length, bond angle, and simulated X-ray diffraction pattern of the gel structure are in good accordance with the experimental results in the literature. Si–O–Al is found to be preferred over Si–O–Si in the N–A–S–H gel structure according to the amount of T–O–T bond angles and distribution of Si⁴(mAl). Pentacoordinate Al is identified in all simulated N–A–S–H models. It provides strong support to current knowledge that pentacoordinate Al in geopolymer does not only come from raw material. Furthermore, the structural analysis results also show that N–A–S–H gel with lower Si/Al ratios has a more cross-linked and compacted structure.

A structure-based modelling framework was established to simulate the three-dimensional autogenous shrinkage of cement paste. A cement hydration model, HYMOSTRUC3D-E, was used to obtain the microstructures and ionic concentrations of the cement paste. A lattice fracture model based on the effective stress and effective modulus was used to consider the elastic and creep parts of autogenous shrinkage. For Portland cement pastes with water-to-cement ratios of 0.3 and 0.4 (where the time zero of autogenous shrinkage was set as the time of the drop in internal relative humidity), the simulated linear elastic autogenous shrinkage was respectively −188 and −79 μm/m at 160 h. The obtained linear total autogenous shrinkage including elastic and creep deformations was respectively −501 and −236 μm/m at 160 h. These values of the elastic autogenous shrinkage and total autogenous shrinkage are close to the predications of poromechanical models and experimental data obtained using a corrugated tube.

Non-uniform stresses, strains and microcracking of the concretes with three coarse aggregate sizes (5–10 mm, 10–16 mm, 16–20 mm) dried under 40% relative humidity (RH) for 60 days were quantified using digital image correlation and lattice fracture modelling. The influencing mechanism of coarse aggregate size on the drying-induced microcracking of concrete was clarified: (1) As the coarse aggregate size decreases, propagation paths of microcracking are increased, which increase the number of small microcracks and release the drying shrinkage force from mortar phase. (2) Tensile stress shells surrounding the coarse aggregates become thinner, thereby decreasing the area of large microcracks. As the coarse aggregate size decreased from 16-20 mm to 5–10 mm, the average thickness of tensile stress shells decreased from 2.13 mm to 1.09 mm at the beginning of drying, and the area of the microcracks >5 μm in width decreased from 796.6 mm²/m² to 340.2 mm²/m² at 60 days since drying.

New self-healing agents that can chemically bind seawater ions invading cracked cementitious materials were proposed. The potential of self-healing and binding of seawater ions were investigated by thermodynamic modeling. It was found that CaO-NaAlO₂ and CaO-metakaolin agents can have Cl⁻, SO₄²⁻ and Mg²⁺ chemically bound by reacting with sea water to form Friedel's salt, Kuzel's salt, ettringite and hydrotalcite. The removal of Cl⁻ from seawater firstly increased and then decreased with the increase of Ca/Al molar ratio in both agents, while the removal of Mg²⁺ and SO₄²⁻ were hardly influenced and approximated 100%. Because NaAlO₂ dissolves and releases Al(OH)₄⁻ rapidly, precipitates binding Cl⁻, SO₄²⁻ and Mg²⁺ were formed fast. In comparison, the reaction of metakaolin binding aggressive ions occurred after 3 days. Because of the faster reaction and the capacity to make [Cl⁻]/[OH⁻] lower in the solution, CaO-NaAlO₂ would be more efficient for self-healing and mitigating reinforcement corrossion than CaO-metakaolin.

To assess the effect of relative humidity (RH) on drying-induced damage in concrete, the non-uniform strains and microcracks in concrete under different RH conditions were obtained using the digital image correlation (DIC) technique and lattice fracture model. The simulated non-uniform displacements were consistent with those captured using DIC. A new damage index was proposed by considering all the subsets with equivalent strain larger than the threshold tensile strength. The calculated damage index showed good correlation with the microcracks' total area and indicated that RH equal or lower than 55% could cause relatively high cracking risk. This work provides an attractive method for quantifying drying-induced damage in concrete using the DIC technique.