Predicting the Influence of Aggregate Size and Distribution on Cementitious Concrete Properties: A Review

Abstract

Cement concrete has been in use for centuries as one of the primary construction materials. Its demand in the construction industry is expected to continue for several centuries before the full development of alternative products. However, one of the main areas of research interest is understanding how its constituents can be tailored to make its properties predictable to reduce risks associated with structural failures, reconstruction and reduced durability. These hindrances associated with cementitious concrete result from several attributes, including constituent material characteristics, mixing ratios and workmanship. Understanding the predictability of cement concrete properties requires computer modelling tools to provide reliable information for the mix design, construction, management and operation of cement concrete, and cement concrete structures. This paper reviews progresses in machine learning models for predicting cement concrete properties. Several algorithms have been reviewed, highlighting their applications, knowledge gaps and suggestions for future research. The paper provides a basis for selecting appropriate algorithms for predicting different concrete properties.