A model for concrete mixture design using fuzzy logic

Abstract

Concrete, a mixture of Portland cement, water and aggregates, is commonly used in the construction industry with high percentage of in-situ fabrication. The proportions of each mixture component is very important to ensure its quality (i.e., compressive strength). Nowadays, several empirical mixture design procedures are utilized; however, the majority of them are based on equations, tables and/or correlations, without considering past experiences and/or experimental data. The present study illustrates the application of fuzzy logic theory for developing a model for estimating concrete mixture proportions by weight, without admixtures. The selected independent variables (input data) were those that are commonly used for concrete mixture design. Historical experimental data and concrete technician’s experience were utilized for creating membership functions (FMs) and fuzzy rules. Mamdani fuzzy inference system (SIL) was used for developing the model since this allows to have several outputs (e.g., water – cement ratio and aggregates). The results indicate that the SIL is able to estimate concrete mixture proportions very well (R²=95.1%); however, the model can be improved as long as new knowledge of the system is available. Furthermore, the model is able to use stored data and technical personal experience in order to develop particular models for each project.