Toward sustainability in optimizing the fly ash concrete mixture ingredients by introducing a new prediction algorithm

Concrete is the most utilized human-made material that has been used in the construction industry. The global concrete demand is rapidly increasing, and accordingly, designing green and sustainable concretes is of great concern. This study aims to optimize the mixture proportion of green and sustainable concretes. In this regard, concrete functional characteristics (i.e., compressive strength, slump, carbonation, and rapid chloride permeability test), unit cost, and environmental impacts (i.e., global warming potential, hazardous waste disposed, non-hazardous waste disposed, and radioactive waste disposed) are considered in the model to be optimized. Accordingly, a new prediction algorithm called Marine predator programming is introduced to model and predict concrete functional characteristics. Three prediction techniques, including artificial neural network, support vector machine, and 2nd polynomial regression, are employed to assess the performance of the introduced machine learning model. Consequently, a novel sustainability modeling is developed, and the mixture proportions of sustainable and green concretes are designed for different compressive strength classes. An optimization process is performed to find the optimal solutions to the mentioned sustainability model. The results indicate that Marine predator programming is highly qualified to estimate different concrete features. Green and sustainable mixtures can reduce the environmental index by 74.37% and 67.83%. The sustainability index of sustainable mixture proportions is up to 80.03% lower than the mixture proportion of mixtures designed by other conventional experimental.

Automated summary

This study aims to optimize the mixture proportion of green and sustainable concretes by considering functional characteristics, unit cost, and environmental impacts. A new prediction algorithm called Marine predator programming is introduced, and three prediction techniques are employed to assess the performance of the machine learning model. The results show that the introduced model is highly qualified to estimate different concrete features and can significantly reduce environmental impacts compared to conventional experimental mixtures.