Title:
Use of Fuzzy Neural Network to Evaluate Effect of Composition of Fly Ash in Suppressing AAR
Author(s):
Hu Ming-yu and Tang Ming-shu
Publication:
Materials Journal
Volume:
103
Issue:
3
Appears on pages(s):
161-168
Keywords:
alkali-aggregate reaction; fly ash; test.
DOI:
10.14359/15849
Date:
5/1/2006
Abstract:
Three types of portland cement and nine types of fly ash, collected from different parts of China, were used to carry out a series of experiments from which 196 groups of test data were obtained. The inherent laws on the suppression of alkali-aggregate reaction (AAR) by fly ash were analyzed using different approaches. By the approach of neural networks, it was deduced that the coefficient of K2O, 0.658, in the calculation formula of sodium equivalent is too large. Furthermore, by the approaches of regression and broken-line graph, it was derived that when cement is partially replaced by fly ash, the coefficient of K2O should be 0.38 or 0.2.