Neural network prediction of unconfined compressive strength of coal fly ash–cement mixtures

• Published in: Cement and concrete research Vol. 33; no. 8; pp. 1137 - 1146
• Main Authors: Sebastiá, Marta, Fernández Olmo, Iñaki, Irabien, Angel
• Format: Journal Article
• Language: English
• Published: New York, NY Elsevier Ltd 01.08.2003; Elsevier Science
• Subjects: Applied sciences; Buildings. Public works; Cement; Cement concrete constituents; Cements; Compressive strength; Computation methods. Tables. Charts; Concrete; Concretes. Mortars. Grouts; Exact sciences and technology; Fly ash; General (composition, classification, performance, standards, patents, etc.); Materials; Neural network analysis; Properties and test methods; Properties of anhydrous and hydrated cement, test methods; Structural analysis. Stresses; Fly ash; Concrete; Compressive strength; Neural network analysis; Cement; Sensitivity analysis; Pozzolanic cement; Coal; Numerical simulation; Euro standard; Neural network; Forecast model
• ISSN: 0008-8846; 1873-3948
• DOI: 10.1016/S0008-8846(03)00019-X

Addition of coal fly ash is a common practice in cement and concrete; an important amount of information can be found in the literature of the unconfined compressive strength (UCS) of these products. Prediction of mechanical properties such as UCS of cement-based pastes, mortars and concrete containing coal fly ash has been done using neural network analysis (NNA) based on the Trajan Neural Network Simulator. The application of NNA has been able to identify the main variables showing an influence on UCS, and the best model to describe UCS with a root mean squared error of 6 MPa for all formulations and 5.5 MPa when formulations are restricted to the maximum addition of coal fly ash established in the European Standards (35% for cement and 55% for concrete). These results allow a good description of the experimental data for the European limits based on cement and concrete, where UCS ranges between 32.5–52.5 MPa and 12–60 MPa, respectively.