Application of the Improved JAYA Algorithm in Searching the Worst Mistuning Patterns for the Splittered Impeller

To investigate the worst mistuning pattern of the impeller with splitter blades, the parallel multipopulation-based elitist JAYA (PME-JAYA) algorithm is proposed based on the original JAYA algorithm, which is a new metaheuristic algorithm and has a very simple structure. Compared with the original JAYA, the PME-JAYA increases the diversity of the search process and improves the population quality by introducing the multipopulation search mechanism and the elite selection strategy, and which accelerates the convergence of the algorithm. At the same time, the parallel computing capability of the algorithm during the process of updating subpopulations is developed, which improves the computing efficiency. Using the PME-JAYA, the worst mistuning patterns of the impeller are searched for the following typical cases of mistuning distribution: (i) the mistuning only distributes on main blades. (ii) the mistuning only distributes on splitter blades. (iii) the mistuning distributes on both main and splitter blades with larger amplitude for the main blades. The forced response characteristics are compared and the influence of mistuning amplitude on the vibration localization characteristics in the worst mistuning cases are analyzed. The results show the main blade dominated modes are sensitive to the main blade mistuning, but not to the splitter blade mistuning. And the splitter blades have the similar property. When mistuning follows uniform distribution, the worst mistuning pattern search results indicate multiple blades have reached the set mistuning amplitude and the worst mistuning patterns under different mistuning amplitudes are not the same. Besides, the position where the maximum response occurs is not fixed for some specific blade.