Unrestrained and Restrained Shrinkage Behavior of Sustainable Lightweight Concrete Using Air Foam and Bottom Ash Aggregates

The objective of this study is to propose reliable models to assess the unrestrained and restrained shrinkage strains of lightweight concrete prepared using bottom ash aggregates and air foam (LWC-BF), on the basis of the nonlinear regression analysis using test data measured in the present specimens. Ten mixtures for LWC-BF were prepared with air foam volume fraction varying between 0% and 40% and water-to-binder ratios ($W/B$) at 25% and 30%. To grant sustainability with LWC-BF, the binder was composed of 30% ordinary portland cement, 50% ground-granulated blast-furnace slag, and 30% fly ash. All the specimens under the plastic shrinkage condition exhibited no crack development. The equivalent porosity ($P_{eq}$) and $W/B$ could be regarded as a critical factor to assess the unrestrained (${\epsilon }_{fsh}$) and restrained (${\epsilon }_{rsh}$) shrinkage strains of LWC-BF. The net time to cracking and average stress rate at cracking recorded in the ring specimens under restrained condition ranged between 1.33 and 1.75 days and 1.12 and $1.31\mathrm{MPa}/\mathrm{day}$, respectively. Thus, the cracking potential of LWC-BF attributable to shrinkage was classified as high level. Based on these analyses and the concept of time function, ${\epsilon }_{fsh}$ and ${\epsilon }_{rsh}$ can be simply formulated as functions of $P_{eq}$ and $W/B$. The statistic comparisons revealed that the proposed models predict accurately the shrinkage behavior of LWC-BF. The mean and standard deviation of normalized root-mean-square error values determined from the measured and predicted shrinkage curves are 0.083 and 0.014, respectively, for unrestrained specimens, and 0.083 and 0.005, respectively, for restrained specimens. Meanwhile, code models result in unreasonable predictions for unrestrained shrinkage curves.