Mesoscopic investigation of the dynamic tensile behaviour of concrete from spalling test and implication on interpretation of test data

Abstract

The dynamic tensile strength of concrete in spalling tests is usually evaluated through indirect measurements such as the pullback velocity at the free-end surface of the specimen, instead of directly measuring the critical stress at the fractured plane(s) due to experimental constraints. These indirect measurements rely on several strong assumptions which could affect the validity of the obtained spall data. This study presents numerical simulations of the spalling test using a robust 2D mesoscale concrete modelling framework, which adopts a ratedependant cohesive constitutive description. Beyond a further investigation into the mechanisms underlining the increment of the dynamic strength observed from experiments, special attention is paid to the investigation of the validity of the traditional data processing approach, namely the so called Novikov approximation, for deducing the spalling strength from the experiments. The proposed mesoscale model for the spalling simulation is firstly validated against experimental evidences. Good agreements are achieved between numerical and experimental results in terms of the pull-back velocity of the free end and the cracking process. Based on the simulations, the dynamic tensile strength developed during a spalling test is evaluated directly through the section force, and the results are compared with the indirect measurements based on Novikov approximation. The comparisons show marked discrepancies, which raise questions about the adequacy of inferring the spalling strength from Novikov approximation, and thereby the validity of many existing data from the spalling tests. It is argued that the measured pullback velocity carries response information from the pre-peak damage history which deviates from the basic assumption of this experimental processing approach. Accordingly, the influences of the material heterogeneity, the incident impact amplitude and the strength of the interface between aggregates and mortar on the validity of the results are investigated and discussed. Results demonstrate that the Novikov approximation approach tends to perform better under a higher loading pulse and a higher ITZ strength, but it is negatively affected by the material heterogeneity.