Host: The Japan Society of Mechanical Engineers
Name : [in Japanese]
Date : November 12, 2022 - November 13, 2022
Centrifugal pumps, which are widely used in industry for transporting liquids in plants and in water purification systems, and it is necessary to improve the instability characteristics at extremely low flow rate conditions. The negative slope of the characteristic curve is the evidence of instability of the flow, and it must be based on the local reverse flow zone and pre-whirl at impeller inlet. It is possible to predict the negative slope of the characteristics by unsteady numerical analysis at low flow rate, however, optimization design search requires many numerical analysis of individuals, it is not realistic solution to use the unsteady numerical analysis for the optimization.
In this study, a meta-model assisted genetic algorithms are used for design search. A neural networks is applied as the meta-model for accelerate the ranking solution in the genetic algorithms. Moreover, a characteristic of the flow field at low flow rate that can be obtained by steady-state analysis, and incorporated the characteristic of the flow field as a constraints for the optimization system. After conducting an optimization geometry search using steady analysis to improve the characteristic flow field, unsteady numerical analysis was conducted on the optimal geometry to investigate the improvement of the negative slope characteristics and the design parameters that have a significant impact on the improvement of the instability characteristics.
