The cement and concrete industry is responsible for about 5% of global anthropogenic CO2 emissions, and the largest source of CO2 emissions from carbonate decompositions. Fly ash can be used as a partial replacement of cement in concrete. However, the heterogeneity of fly ash poses a challenge to its optimal utilization in concrete. Traditionally, the classification of fly ash for use in concrete is based on bulk chemical composition. This classification does not account for the inherent variability in the fly ash glass, which is the main reactive component of fly ash and a major determinant of its performance in concrete.

This research utilizes simulated fly ash modeled after a real fly ash to investigate the effect of the variation of the fly ash glass chemical composition on the performance of the fly ash, while other major properties are held constant. The chemical compositions of several thousands of individual particles from a Coal Creek, IA source fly ash was determined by computer-controlled SEM (CCSEM). The particle data was then grouped into four clusters having relatively narrow ranges of composition. A set of four synthetic fly ash glasses were prepared based on the mean chemical composition of each cluster. Three simulated whole fly ashes were made by combining varying fractions of the synthetic fly ash glasses over a range of 10% weight difference.

The microstructure and reactivities of the simulated fly ashes in cement paste at a 20% replacement factor were analyzed using an array of analytical techniques including low-temperature differential scanning calorimetry, isothermal calorimetry, SEM/EDX with multispectral image analysis, thermogravimetric analysis, electron microprobe analysis and compressive strength tests.

The simulated fly ash with the highest contents of Si, Al and Ca appears to be the most reactive across the various reactivity measurements, indicating the significance of these phases in the pozzolanic reaction of fly ash with cement. The reactivities of the individual synthetic glasses were similar within the bounds of experimental error. This could be due the smaller variations in the glass compositions as well as the 20% replacement factor.