Experimental verification of the boundary conditions in the success of the Brazilian test with loading arcs. An uncertainty approach using concrete disks

The present work analyses the reliability of the Brazilian test with loading arcs. A new testing set up has allowed to determine in an effective way the real load of the failure initiation as this moment was not always or correctly detected by the universal testing machine. The instrumentation used is a simple and low-cost method that allows to know the possible pressure distribution in the contact zone as well as the final contact angle. It has been observed that the success of the test depends mainly on the surface finish of the parts involved, their geometric tolerances and the symmetry of the applied load. These boundary conditions have a direct effect in the contact pressure distribution. The possible failure modes observed experimentally have been simulated with the finite element methods. For this, the contact boundary condition has been changed and the possible stress distribution in term of Griffith equivalent stress has been obtained. The numerical analysis allows to study the influence of the initial contact condition on the success of the test and agrees with the experimental results. Furthermore, an uncertainty analysis in the expression of the tensile strength confirms that, when the test is valid, a crack appears suddenly in the central area of the disk, as observed experimentally, so there is no need to determine if the starting point is in the centre. Additionally, it has been observed that the initial crack length depends on the type of pressure distribution in the contact zone. Finally, a series of recommendations are given in order to minimize both the variability of the final contact angle and the risk of premature failure of the Brazilian disk.