We developed a method of self-configurable multipopulational parallel genetic algorithm (SGA) and successfully applied it to a standardless automatic full-profile X-ray phase analysis of aluminum bath samples. The method spares a user-analyst from working on the empirical choice of the genetic algorithm settings and shows the efficiency of the automatic QPA by the Rietveld method under the control of the SGA, as well as the possibility of using it for express monitoring of the bath composition in aluminum production.