IntroductionGeopolymer is a green cementitious material produced via the activation of silica-aluminate-rich precursor materials in an alkaline environment. It has outstanding mechanical qualities and durability. However, the different raw materials and complicated reaction process of geopolymers result in some parameters affecting mechanical qualities that are difficult to effectively manage and forecast. It is thus necessary to examine the variation rule of mechanical properties in geopolymer concrete (GPC) under various conditions via numerical simulation. Though the majority of existing numerical models are applicable to ordinary Portland cement (OPC), they cannot be directly applicable to GPC due to variations in the chemical characteristics as well as isomorphic model of mortar and interface transition zone (ITZ) in GPC and OPC. The existing studies lack clear criteria for material inhomogeneity in each phase of the model, and the majority of them neglect the impact of interior micropores on the model mechanical characteristics. In this work, an innovative method was used to account for non-homogeneity and determine the mechanical parameters and intrinsic model of geopolymer materials through experiment. In addition, the impacts of aggregate content, model size, and porosity on the mechanical properties of GPC were also investigated by the verified model.MethodsThe mechanical parameters and intrinsic model were determined via a geopolymer mortar test, and the model reliability was verified via a geopolymer concrete test. To prepare the specimens, slag and fly ash were used as precursors, and NaOH and water glass were used as alkaline activators. A coarse aggregate was a continuously graded gravel with a particle size range of 5-20 mm. A fine aggregate was a standard sand with a fineness modulus of 2.5. In this proportion, the alkali equivalent (in mass proportion of Na₂O to precursor) was 4%, the water-cement ratio was 0.50, the modulus of alkali solution (SiO₂/Na₂O) was 1.2, and the mortar was kept in the same proportion of mortar phase in the concrete.At the simulation level, the Mesh-Placement-Identification-Assignment (MPIA) procedure was utilized to create a 3D GPC meso-scale model, and the model fundamental conditions were consistent with the experiments. It was assumed that the mechanical characteristics of each phase material inside the concrete could follow the Weibull distribution, and the mortar phase element non-homogeneous parameters could be solved using an equivalent unit approach. Also, initial flaws were introduced into the model, and multi-scale analysis was utilized to assess their impact on mechanical characteristics because the mixing process could form air bubbles in the concrete. After the model was constructed, the material constitutive model was determined via mortar experimental curve fitting, and the mortar test results and test algorithms were used to estimate the material mechanical characteristics. The model was imported into ABAQUS to set the boundary and loading conditions before being compared to the GPC test results obtained after the simulation. The results indicated that the damage patterns and mechanical characteristics of the test and simulation results could be nearly compatible, and the model was regarded as reliable.Results and discussionBased on the validated model, GPC models with different aggregate volume fractions of 30%, 40%, and 50% are simulated under uniaxial compression. The simulation findings reveal that the compressive strength of GPC increases as the aggregate volume percent increases, as does the energy absorbed for damage. This is because the aggregate plays a role in preventing crack development, so as the aggregate content increases in a specific range, the crack morphology inside the model becomes more complex, the energy required for damage increases, thus improving the macroscopic performance of the mechanical properties.Under uniaxial compression, GPC cubes with the side lengths of 70, 100 mm, and 150 mm are simulated, respectively. According to the simulation results, the compressive strength of GPC descends when the specimen volume grows. Compared to OPC, GPC displays the same pronounced size effect phenomena and aligns well with the Bažant theoretical formulation.The GPC models with internal porosities of 1.9%, 4.3%, and 5.9% are simulated under uniaxial compression, having the compressive strengths of 49.10, 47.90 MPa, and 46.19 MPa, respectively. The mechanical characteristics of the geopolymer concrete decrease slightly as the model inside porosity increases. This is since the introduction of pores in this study takes into account both porosity and pore size distribution, and the number of small-sized pores increases as the porosity increases. The small pores have less influence on the mechanical properties, and the results show that the model mechanical properties are insensitive to the change of porosity.ConclusionsA non-homogeneous meso-scale model of GPC was proposed, and the simulation results were similar to the experimental result. Also, an innovative approach to describe the non-homogeneity of GPC was proposed. The equivalent unit method was used to determine the non-homogeneity parameter, and the initial microporous defects were characterized via multiscale modeling and folding coefficients. The energy required to destroy GPC and the compressive strength increased with increasing aggregate content. As the model size increased, the mechanical characteristics of GPC decreased, aligning with Bažant's theory. The mechanical characteristics of geopolymer concrete diminished slightly as initial microporosity increased.