On the dispersion of data collected by in situ diagnostic of the existing concrete

The compressive strength of in-place concrete is a crucial mechanical parameter which influences the performance level and the safety assessment of existing RC structures both under seismic loads and under dead loads. Italian NTC and Eurocode 8 [1,2] have established that the fundamental basis of the safety assessment is the "Knowledge Level (KL)", which shall be defined according to the procedures adopted for in situ inspection and materials testing (destructive and non-destructive). Within this framework - in the case of RC constructions - a major requirement is the explicit identification of the reference values for the strength of in situ concrete. The afore mentioned building codes, however, do not include any consideration about the uncertainty level affecting the results of in situ tests, which can indeed invalidate the reliability of the mechanical parameters. This problem is instead well addressed by FEMA 356 [3], which prescribes a limit value (14%) to the statistical dispersion of the measures performed on a set of concrete specimens. In this paper, after presenting and discussing a procedure applied for processing experimental data provided by in situ tests in a number of real case studies (school buildings in the Province of Foggia, Italy), the attention will be focused on the problem of the data scattering in the case of in situ concrete strength. It is shown that the variance is a crucial parameter, which should always be considered when performing the safety assessment, even if FEMA requirements, in this respect, are too much restrictive.