Hot rolling of bars issued from continuous-casting aims at refining the material structure and guaranteeing the central soundness of the metallurgical product. The rolling route must be designed to achieve the complete closure of the shrinkage porosity inherent in the continuous casting process. To predict the void evolution, many models exist that can be implemented in the finite element simulation of the process. Nevertheless, these models need parameter adjustments to be adapted to the forming process, the formed material, and the real geometry of the void. Real scale tests being very expensive in the long product rolling mill, an improved representativeness experimental configuration was designed to reproduce at the laboratory scale the key characteristics of the thermomechanical path driving the void closure phenomenon. This testing consists of successive forming stages with shaped anvils applied to samples containing a shrinkage cavity. The shaped anvils and the forming conditions are calibrated to reproduce the levels of strain and the stress triaxiality of rolling stands, and the alternation of the forming direction of the industrial process. The geometry of the voids before and after the forming paths are measured by tomography. The simulation of the test with an explicit modelling of the void is developed parallel to the experiments. The simulation/experiment comparison allows the validation of the numerical model. The obtained model will be used in future works to perform a more extended design of experiments to characterise void closure during hot rolling of bars.