Performance, sustainability, and machine learning of UHPC with coarse aggregates: a critical review

Abstract

Ultra-High-Performance Concrete (UHPC) is an advanced cementitious material characterized by its exceptional mechanical strength, durability, and resistance to extreme loading conditions. However, the incorporation of coarse aggregates in UHPC remains an active area of research because of the trade-offs among cost-effectiveness, sustainability, and mechanical performance. This systematic review critically evaluates the comparative performance of UHPC with and without coarse aggregates, while examining the application of emerging machine learning (ML) techniques for performance prediction and mix design optimization. The review synthesizes findings from 63 peer-reviewed studies selected through a PRISMA-based screening process from an initial pool of approximately 150 publications published between 2000 and 2026. The reviewed studies collectively indicate that the incorporation of coarse aggregates enhances the economic feasibility and environmental sustainability through reduced cement consumption, lower autogenous shrinkage, and enhanced dimensional stability, while conventional aggregate-free UHPC generally provides superior tensile strength, crack resistance, and matrix homogeneity owing to improved fibre dispersion and a stronger interfacial microstructure. Machine learning methods, particularly Artificial Neural Networks (ANN) and Support Vector Machines (SVM), have demonstrated considerable potential for predicting mechanical properties and supporting durability- and sustainability-oriented performance optimization. Machine learning applications are emerging in UHPC research, particularly for performance prediction and mix-design optimization; however, their broader adoption remains limited by small and heterogeneous datasets, inconsistent validation procedures, and insufficient integration of life-cycle assessment parameters. The major research gaps include physics-informed ML methods, optimization of aggregate gradation to enhance interfacial transition zone (ITZ) behaviour, and systematic long-term durability assessment under diverse environmental conditions. Based on these findings, a future research roadmap is proposed to support the development of sustainable, cost-effective, and intelligent UHPC systems. Unlike previous UHPC review articles that primarily focus on constituent materials, mechanical performance, or durability, this review integrates coarse aggregate incorporation, sustainability assessment, interfacial transition zone (ITZ) behaviour, and machine learning-based optimization within a single systematic framework, thereby providing an integrated roadmap for the development of sustainable and intelligent UHPC systems.