Development of gravitational search algorithm model for predicting packing density of cementitious pastes

Abstract

Wet packing approach is recently used to design different kinds of concrete. To achieve an optimal particle packing density, the particles should be chosen in a way to fill the voids between larger particles with smaller ones for obtaining a dense and stiff particle structure. The majority of past research on packing density has focused on the evaluation of the particle size distribution of granular matrix to gain improvements in the packing density of cementitious materials, while limited attention has been paid to the construction of a model to estimate the packing density value. To serve that purpose, in this study, a series of 216 collected samples were used for proposing a model. The dataset was divided into two main sets for construction and verification of the model. As the basis for the packing density modeling, use of the gravitational search algorithm (GSA) was proposed. Design of experiment (DOE) software was used to evaluate the contribution of each variable on the proposed model. The outcomes indicate that among different parameters, water amount has the largest effect on the packing density value of cementitious pastes. Moreover, increases in the amounts of supplementary cementitious materials up to certain levels increase the packing density. The coefficient of variation (CoV) of the proposed model is 6.3% which reflects the accuracy and consistency of the model.