Recent developments towards digital concrete

Abstract

Concrete formulation, despite recent progress, remains challenging due to the increasing number of constituents and performance requirements. This study explores a parallel approach to physical and semi-empirical experimental proportioning methods, utilizing machine learning (ML) or deep learning (DL). By harnessing big data from concrete production and high throughput experimentation, ML/DL algorithms can optimize mixes, minimizing waste and maximizing performance. This approach, combined with digital fabrication, addresses environmental and productivity concerns in construction. However, challenges such as diverse non-digital experimental methods, lack of basic data on waste streams, and limited open repositories on cement-based materials hinder the widespread use of data-driven techniques. This paper examines the background and potential of data-driven techniques in the field of concrete technology.