A unified RVE-based mesoscale framework for predicting mechanical performance of conventional cement concrete and recycled aggregate slag-dolomite geopolymer concrete.

Concrete is inherently heterogeneous but is often modelled as homogeneous at the macroscale for simplicity. Accurately representing this heterogeneity using mesoscopic modelling minimises reliance on extensive laboratory testing, enables efficient parametric studies, and thereby reduces both experimental effort and overall costs in developing sustainable cementitious composites. Mesoscale modelling, which treats concrete as a heterogeneous three-phase system comprising mortar, aggregate, and the interfacial transition zone (ITZ), offers a practical balance between accuracy and computational demand. It can simulate both the mechanical properties and the nonlinear behaviour of concrete, while capturing size effects that significantly influence these properties. It simplifies parametric investigations, deepens understanding of crack initiation and propagation, and facilitates the efficient generation of comprehensive datasets. Despite these advantages, mesoscale models can be computationally intensive due to numerous contact regions and thin ITZ layers. To address this challenge, the present study develops a random-aggregate Representative Volume Element (RVE) framework for simulating uniaxial compressive and tensile strength, as well as biaxial strength, of conventional ordinary Portland cement (OPC) concrete. Parametric studies are conducted to evaluate the effects of RVE size, aggregate shape, and ITZ thickness on the simulation of uniaxial compressive and tensile strength of OPC concrete. The optimal configuration, a 5 cm RVE with circular aggregates and ITZ thickness of 0.1 times the aggregate diameter, reproduced uniaxial compressive and tensile strengths while reducing the number of finite elements by approximately 90% relative to full-scale (15 cm) modelling, which significantly reduces computational time. The framework is validated under biaxial loading, demonstrating its capability to capture complex multiaxial behaviour. It is further extended to simulate the uniaxial compressive response of recycled aggregate geopolymer concrete (RAGPC), advancing low-carbon, circular-economy materials. The results demonstrate that the proposed mesoscopic RVE framework offers an effective, extensible, and computationally efficient approach for simulating the mechanical response of different types of concrete, highlighting its potential as a robust tool for sustainable cementitious composites.