Compressive strength optimization and life cycle assessment of geopolymer concrete using machine learning techniques

¹ Department of Civil Engineering, Michael Okpara University of Agriculture, Umudike, Nigeria
² Department of Civil Engineering, School of Engineering, University of the Peloponnese, GR-26334 Patras, Greece
³ School of Science and Technology, Hellenic Open University, GR-26335 Patras, Greece
⁴ Department of Civil Engineering, Ladoke Akintola University of Technology, Ogbomosho, Nigeria
⁵ Department of Chemical Engineering, Michael Okpara University of Agriculture, Umudike, Nigeria
⁶ Faculty of Civil Engineering, Babol Noshirvani University of Technology, Mazandaran, Babol, Iran
⁷ Department of Civil Engineering, Shahid Bahonar University of Kerman, Iran
⁸ Department of Information Technology, Vishnu Institute of Technology, Bhimavaram, Andhra Pradesh 534202, India
⁹ Department of Civil Engineering, University of Birjand, Birjand, Iran
¹⁰ Department of Civil and Mineral Engineering, The University of Toronto, Toronto, Canada

^* Corresponding author: kontoni@uop.gr

Abstract

Fly ash-based geopolymer concrete is studied in this research work for its compressive strength, life cycle and environmental impact assessment contribution to the construction environment. This is in line with the United Nations’ sustainable development goals SDG9 and SDG11. However, the focus of this research paper is on the sustainability of geopolymer concrete and its overall environmental impact. The metaheuristic machine learning approaches have been deployed to predict the compressive strength (CS) of the GPC based on environmental impact considerations of the concrete constituent materials, which included fly ash, sodium silicate, sodium hydroxide, fine and coarse aggregates. The metaheuristic techniques include the k-Nearest Neighbour (kNN), support vector regression (SVR), and random forest regression (RFR), where all are optimized with the particle swarm (PSO). These metaheuristic techniques have been modified for this research work with new codes to enhance innovation in terms of run time and efficiency. The results of the life cycle assessment (LCA) evaluation of the GPC mixes based on the Ecoinvent 3 available in SimaPro and Eco-indicator 99 and CML 2001 modified in the framework of ReCiPe 2016 recent development show reduced potential of environmental acidification due to increased fly ash (FA) in the GPC mixes compared to previous results. The decisive CS and LCA predictive models, RFR-PSO and SVR-PSO respectively performed optimally above 90% and better than previous models from the literature. Overall, they present an innovative metaheuristic smart technology for the prediction of the GPC infrastructure behavior and performance integrity.