Abstract
The incorporation of supplementary cementitious materials, nano additives and sustainable fillers in Ultra-High Performance Concrete (UHPC) has been recognized as a valuable solution to reduce cement usage while enhancing the mechanical and durability performance. However, replacing cement with sustainable alternatives presents a significant challenge, as optimizing UHPC mix designs involves multiple parameters and extensive hit-and-trail experimentation. However, Artificial Intelligence (AI) has gained attention as a powerful approach for predicting the performance of advanced cementitious composites, enabling more efficient mix design optimization and reducing extensive laboratory testing. Despite these advancements, most available UHPC datasets are limited in scope, focusing on specific regions, narrow ranges of properties, or select mix parameters which restricts their effectiveness for global research and comprehensive data-driven modelling. To address this gap, a comprehensive global dataset of UHPC mix designs featuring 2188 mix designs from 168 publications has been compiled across a diverse range of countries, cement types, supplementary cementitious materials (SCMs), nano particles, fillers, fibre, and curing regimes. The dataset is formatted and standardized for use in machine learning (ML) model development and inverse design strategies for developing sustainable and ecofriendly UHPC. By offering a globally diverse dataset, this work provides a valuable resource for advancing the design of sustainable UHPC mixtures.