Abstract

The aim of this study is to investigate the estimation ability of the effects of utilizing different amount of the class C fly ash on the mechanical properties of cement using artificial neural network and regression methods. For this reason, 0, 5, 10, 15 and 20% amount of the class C fly ash were substituted with cement and 40 x 40 x 160 mm dimension specimens were prepared. On the prepared specimens unit weight, flexural tensile strength and compressive strength tests were performed after the 2, 7 and the 28^th days. 2 different estimation models regression techniques (RT) and the artificial neural network (ANN) methods were used for determining the flexural tensile strength and the compressive strength of the cement specimens. Experimental results were used in the estimation methods. Fly ash content (%), age of specimen (day) and unit weight (g/cm³) were used as input parameters and flexural tensile and compressive strengths (N/mm²) were used as output parameters. The developed models and the experimental results were compared in the testing data set. As a result, compressive and flexural tensile strength values of mortars containing various amounts class C fly ash can be predicted in a quite short period of time with tiny error rates by using the multilayer feed-forward neural network models than regression techniques.

Key words: Fly ash, cement, flexural tensile strength, compressive strength, regression, ANN.