Performance prediction and optimization strategy for LNG multistage centrifugal pump based on PSO-LSSVR surrogate model

• DOI: 10.1016/j.cryogenics.2024.103856
• Publication Date: 2024-05-12T22:06:05Z
• Authors: Bo Liu; Wei Zhang; Feng Chen; Jie Cai; XunMing Wang; Yi Liu; JinLing Zhang; Qian Wang
• Keywords: Surrogate model

The Multistage LNG cryogenic submersible pump is one of the crucial power equipment in transport process. However, high hydraulic losses and low efficiency this make it necessary to optimize structure its impeller order reduce energy losses. In study, an optimization strategy based on particle swarm (PSO) least squares support vector regression (LSSVR) machine proposed impeller. with PSO-LSSVR method increased head by 0.57% 2.72%, respectively. Comparing calculation results CFD results, we found that relative error between them don't exceed 3%, which verified accuracy method. Then, enstrophy dissipation theory introduced quantitatively analyze loss. impellers before after revealed maximum reduction loss was 18%.. distribution mechanism vortex generation inside were analyzed Q-criterion equation. equation horseshoe at leading edge optimized blade suppressed, stretching term Coriolis force term, dominate evolution structure, reduced. This study will provide related reference for optimal design pumps.