Analyzing the compressive performance of lightweight foamcrete and parameter interdependencies using machine intelligence strategies

Abstract As an alternate to regular concrete, foam concrete, also called foamcrete, has several useful applications. It saves money on transportation and production costs as well as dead weight on buildings and foundations, which helps with energy efficiency. Nevertheless, there is still a lack of practical applications, which calls for more research, especially in strength studies, to increase its use in the actual world. For this purpose, the compressive strength (C-S) of foamcrete was assessed using two machine learning algorithms: gene expression programming (GEP) and multi-expression programming (MEP). A sensitivity analysis was conducted to determine how important certain aspects were. For predicting foamcrete’s compressive strength, MEP was better than GEP. By comparison, the MEP model had an R 2 value of 0.970, while the GEP models only managed 0.94. This is further supported by the findings of the statistical analysis and the ML models’ cross-validation using Taylor’s diagram. The sensitivity analysis results indicated that density (28.0%), cement content (11.0%), and age (8.5%) were the three most significant criteria influencing overall strength. The generated models can determine the compressive strength of foamcrete for different input parameter values, hence enhancing its practical uses and saving time and financial resources compared to laboratory testing.