A novel evolutionary learning to prepare sustainable concrete mixtures with supplementary cementitious materials

In this study, sustainable mixture designs of three concrete types, including fly ash concrete, silica fume concrete, and ground granulated blast furnace slag concrete, were investigated. To this end, the compressive strength formulas of each concrete type made with supplementary cementitious materials were obtained by introducing a new machine learning algorithm, called coyote optimization programming. The accuracy of this algorithm proved to be greater than that of conventional and recently developed machine learning methods. An optimization problem is modeled, in which the compressive strengths, price, and environmental impact of the sustainable concrete mixture designs were estimated using global warming potential, energy consumption, and material consumption as the sustainability parameters. Results reveal that increasing the compressive strength reduces the sustainability of concrete, and thus, manufacturing concrete with a higher compressive strength than the one obtained from the design process contradicts the concrete’s performance. Moreover, the 30-MPa sustainable fly ash concrete was proven to be the most sustainable mix with a gray relational grade of 1. This optimal mixture designed in this study can decrease the unit cost, global warming potential, energy consumption, and material consumption by 36.6%, 51%, 43%, and 11%, respectively. Research highlights A novel machine learning method called coyote optimization programming was introduced in this study. Applying optimization techniques to design concrete mixture proportions can reduce the unit cost by 36.6%. The introduced approach can decrease the global warming potential, energy consumption, and material consumption by 51%, 43%, and 11%, respectively. The application of supplementary cementitious materials in the concrete mixtures significantly enhances sustainability.