How to improve efficiency is still a very active research point for pump as turbine. This article comes up with a method for optimal design of pump as turbine impeller meridional plane. It included the parameterized control impeller meridional plane, the computational fluid dynamics technique, the optimized Latin hypercube sampling experimental design, the back propagation neural network optimized by genetic algorithm and genetic algorithm. Concretely, the impeller meridional plane was parameterized by the Pro/E software, the optimized Latin hypercube sampling was used to obtain the test sample points for back propagation neural network optimized by genetic algorithm, and the model corresponding to each sample point was calculated to obtain the performance values by the computational fluid dynamics techniques. Then, back propagation neural network learning and training are carried out by combining sample points and corresponding model performance values. Last but not least, back propagation neural network optimized by genetic algorithm and genetic algorithm were combined to deal with the optimization problem of impeller meridional plane. According to the aforementioned optimization design method, impeller meridional plane of the pump as turbine was optimized. The result manifests that the optimized pump as turbine energy-conversion efficiency was improved by 2.28% at the optimum operating condition, at the same time meet the pressure head constraint, namely the head difference between initial and optimized model is under the set numeric value. This demonstrates that the optimization method proposed in this article to optimize the impeller meridional plane is practicable.
Computational Fluid Dynamics Simulation of Indoor Air Quality and Thermal Stratification of an Underfloor Air Distribution System (UFAD) with Various Vent Layouts
