Abstract
The post-crack performance of fiber-reinforced concrete (FRC) is widely reported to exhibit a high degree of variability. Several reasons exist to explain this problem, most notably the small area of crack surface involved in most beam tests. Of less widely recognized significance to variability is the relationship between the sample standard deviation and the population standard deviation for this material. Standard statistical theory can be used to estimate the relationship between the variance in a sample of observations and the variance in the population from which the observations are drawn, but this critically depends on the observations being drawn from a population displaying a normal distribution. Moreover, the observations must also exhibit covariant stationary characteristics. In this investigation, it is shown that neither of these assumptions are necessarily true for FRC specimens produced using standard mixing procedures. Instead, the distribution of post-crack performance parameters obtained from standard beam and panel-based tests exhibit inconsistent characteristics that possibly require an empirical approach to effective estimation of the population standard deviation.
