Title:
Flow and Compressive Strength of Alkali-Activated Mortars
Author(s):
Keun-Hyeok Yang, Jin-Kyu Song, Kang-Seok Lee, and Ashraf F. Ashour
Publication:
Materials Journal
Volume:
106
Issue:
1
Appears on pages(s):
50-58
Keywords:
alkali-activated mortar; fly ash; slag; neural network; regression analysis
DOI:
10.14359/56316
Date:
1/1/2009
Abstract:
Test results of 36 slag-based mortars and 18 fly ash (FA)-based mortars activated by sodium silicate and/or sodium hydroxide powders are presented. The main variables investigated were the mixing ratio of sodium oxide (Na2O) of the activators to source materials, water-binder ratio (w/b), and fine aggregate-binder ratio (s/b). Test results showed that slag-based alkali-activated (AA) mortars exhibited much higher compressive strength but slightly less flow than FA-based AA mortars for the same mixing condition. Feed-forward neural networks and simplified equations developed from nonlinear multiple regression analysis were proposed to evaluate the initial flow and 28-day compressive strength of AA mortars. The training and testing of neural networks and the calibration of the simplified equations were achieved using a comprehensive database of 82 test results of mortars activated by sodium silicate and sodium hydroxide powders. Compressive strength development of slag-based AA mortars was also estimated using the formula specified in ACI 209R calibrated against the collected database. Predictions obtained from the trained neural network or developed simplified equations were in good agreement with test results, although early strength of slag-based AA mortars was slightly overestimated by the proposed simplified equations.