Blade profile optimization of pump as turbine
Sen-chun Miao,Jun-hu Yang,Guang-tai Shi,Ting-ting Wang
DOI: https://doi.org/10.1177/1687814015605748
IF: 2.1
2015-09-01
Advances in Mechanical Engineering
Abstract:One major focus on the performance researches of pump as turbine is how to enhance the efficiency of energy recovery. While the key point of increasing the efficiency is to improve the performance of the blade profile which is structural basis of the blade geometry. This article presents an optimization method for the blade profile. It contained the parameterization of blade profile, the Latin Hypercube experimental design, the computational fluid dynamics techniques, the back propagation neural network, and genetic algorithm. Specifically, the nonuniform cubic B-spline curve was used to parameterize the blade profile, the Latin Hypercube experimental design method for the acquirement of the sample points of back propagation neural network. The performance analysis of each sample point was accomplished by the computational fluid dynamics techniques. Then, the learning and training of the back propagation neural network was carried out. Finally, the optimization techniques of combining the back propagation neural network and genetic algorithm were used to solve the optimization problems of the blade profile. Based on the above method, the blade profile of a pump as turbine was optimized and improved. The result shows that the efficiency of the optimized pump as turbine under the optimum operating condition was increased by 2.91%, with the constraint condition to ensure that the difference between the head and the initial head of the pump as turbine is less than the specified value. This proves that using the above method to optimize the blade profile is feasible.
engineering, mechanical,thermodynamics