The article presents the problem of modelling the charging of a constant-phase bed storage in the first hours of the process. The places of errors in the heat transfer calculations for the packed beds were indicated. Granite in the shape of spheres and crushed rocks, with a characteristic dimension of 50 mm, was used for the experimental tests. The material was subjected to tomographic examination and then used as a storage material. The charging process was carried out for three flow rates: 0.006, 0.008 and 0.010 m3/s. After three hours of testing, the temperature of the outlet air for the granite sphere as the storage material was the same as for the granite crushed rock. However, the biggest differences occurred after 1 h of charging. They were equal to: 40.4% for the flow rate of 0.006 m3/s, 22.0% for the flow rate of 0.008 m3/s, and 18.5% for the flow rate of 0.010 m3/s. The differences were greater than the uncertainty of the measurements. As a result, different temperatures of the storage material were obtained. After three hours, they were equal to: 25.2%, 12.3% and 8.6% for the lowest, medium, and highest airflow, respectively. The conducted heat transfer analysis and the relationship Nu = f(Re) was determined. The influence of the calculated and actual surface of the crushed rock on the heat exchange process was explained. For all the tested air flow rates through the bed, higher thermal parameters were obtained for the crushed rock than for the sphere. The maximum differences in the Nu number were: 222.6%, 151.4% and 161.3% for the flow of 0.006, 0.008 and 0.010 m3/s, respectively. This means that the description of the heat exchange process in the piled beds would require a parameter that takes into account the geometry of the storage material.