A full-scale water physical model of a degassing unit is built and used to evaluate the performance of several impeller designs. Four impeller designs are tested: a) one smooth not commercial impeller for reference purposes, b) a commercial design by FOSECO®, called standard impeller in this work, c) a commercial design by FOSECO® with notches, and d) a new design proposed in this work. Since the physical model is easy and safe to operate, a full experimental design is performed to evaluate the effect of the most important process variables, such as impeller rotating speed, gas flow rate, impeller design and the point of gas injection (a conventional gas injection through the shaft and a novel method of injecting gas through the bottom of the ladle) on the kinetics of oxygen desorption of water which is similar to dehydrogenation of liquid aluminum. The new design of impeller proposed in this work shows the best performance in degassing of all impellers tested in this study. It is found that the rotor speed and its design are the most significant variables affecting degassing kinetics, and therefore the analysis of the existing commercial impeller designs may be useful to optimize the fluid dynamics of the process, which in turn would increase efficiency and productivity of the process. Finally, the novel gas injection method through the bottom, proposed by our own group, presents slightly faster degassing kinetics than the conventional injection of purge gas in the conventional way through the impeller.