Experimental protocols to study the transmissivity (to fluids) of localized cracks in concrete samples are often based on the use of a Brazilian test. According to the protocol proposed by Rastiello et al. (2014, 2018) for water flows, and extended to air flows by Tailhan et al. (2018), such a test is coupled with a specific measurement setup to assess fluid flow properties under loading. The splitting load, which generates the localized macrocrack, is indirectly controlled by the measurement of the mean diametrical sample expansion. The plane faces of the sample stand thus available for positioning the upstream and downstream flow chambers, in which pressures are controlled. The induced fluid flow is then measured and put into relation with imposed pressures and crack geometrical properties. However, fluid mechanics arguments suggest that considering the pressures in the downstream/upstream chambers could lead to a wrong characterization of the crack transfer properties. For that reason, the present contribution proposes a more in-depth investigation of the characteristics of the experimental setup itself. The development of an “idealized” version of the air permeability test allows for replacing the real crack by a “plane crack” equipped with additional internal pressure sensors. These additional measurements show that pressures acting at the inlet/outlet crack sections can be very different from those controlled in the chambers due to localized head losses. Moreover, knowing the real pressure drop within the crack leads to interpret experimental results more accurately.