Relationship between Computed Rate Constants and the Variability in Maturity-Based Strength Predictions

Abstract ASTM C1074-11, Standard Practice for Estimating Concrete Strength by the Maturity Method, provides instructions for applying the Nurse-Saul and Freiesleben-Hansen Pedersen maturity methods, including ways to obtain the key parameters of datum temperature (T0) or the Arrhenius activation energy (Ea). Default values are provided along with three alternative computational methods for determining a mixture-specific value of T0 or Ea from experimental data, as detailed in C1074-11 Annex A1. This paper demonstrates via examples that the accuracy of the resulting strength estimates can vary considerably depending on which of the three alternatives is selected, potentially masking the more intrinsic variability of these parameters as influenced by composition or temperature. One source of this observed variability is the statistical bias known to characterize the “double reciprocal linearization” regression technique, which is utilized in two of the three ASTM methods. While recommending the deletion of these statistically biased estimators, the authors support the more rigorous nonlinear regression method (A1.1.8.1) as well as simpler functional forms; they affirm the reasonableness of the ASTM default values for T0 and Ea.