The cooling structure of the air-cooled turbine first-stage inlet guide vane is complex, requiring a refined multi-factor design. This study focuses on the impingement insert in the forward cavity of turbine vanes. A multi-objective optimization method is developed to improve the internal cooling structure, considering the coupled effects of jet hole diameter, streamwise spacing, and spanwise spacing. Latin Hypercube Sampling (LHS) method is employed for the design of experiments, and a Backpropagation Neural Network (BPNN) is developed to model the relationship between the geometric parameters of the impingement insert and the area-averaged Nusselt number (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">${\overline{Nu}}$</EquationSource> <EquationSource Format="MATHML"><math display="block"> <mrow> <mover> <mrow> <mi>N</mi> <mi>u</mi> </mrow> <mo accent="false">¯</mo> </mover> </mrow> </math></EquationSource> </InlineEquation>), uniformity index (UI), and overall pressure drop (ΔP). Furthermore, multi-objective optimization of <InlineEquation ID="IEq2"> <EquationSource Format="TEX">${\overline{Nu}}$</EquationSource> <EquationSource Format="MATHML"><math display="block"> <mrow> <mover> <mrow> <mi>N</mi> <mi>u</mi> </mrow> <mo accent="false">¯</mo> </mover> </mrow> </math></EquationSource> </InlineEquation>, UI and ΔP is performed using the NSGA-II algorithm. The results show that the Pareto-optimal solutions achieve an approximately 20% reduction in ΔP compared to the original design, along with improvements of 1.48%–17.14% in <InlineEquation ID="IEq3"> <EquationSource Format="TEX">${\overline{Nu}}$</EquationSource> <EquationSource Format="MATHML"><math display="block"> <mrow> <mover> <mrow> <mi>N</mi> <mi>u</mi> </mrow> <mo accent="false">¯</mo> </mover> </mrow> </math></EquationSource> </InlineEquation> and 5.46%–8.17% in UI.

Multi-Objective Optimization of Turbine First-Stage Inlet Guide Vane Internal Impingement Cooling Structure Based on NSGA-II Algorithm

期刊论文
发表日期：

2026年4月15日

查看全文

Yi Liu,
Yiran Li,
Qiuping Ma,
Xueying Li,
Jing Ren

摘要

The cooling structure of the air-cooled turbine first-stage inlet guide vane is complex, requiring a refined multi-factor design. This study focuses on the impingement insert in the forward cavity of turbine vanes. A multi-objective optimization method is developed to improve the internal cooling structure, considering the coupled effects of jet hole diameter, streamwise spacing, and spanwise spacing. Latin Hypercube Sampling (LHS) method is employed for the design of experiments, and a Backpropagation Neural Network (BPNN) is developed to model the relationship between the geometric parameters of the impingement insert and the area-averaged Nusselt number ( ${\overline{Nu}}$ $\bar{N u}$ ), uniformity index (UI), and overall pressure drop (ΔP). Furthermore, multi-objective optimization of ${\overline{Nu}}$ $\bar{N u}$ , UI and ΔP is performed using the NSGA-II algorithm. The results show that the Pareto-optimal solutions achieve an approximately 20% reduction in ΔP compared to the original design, along with improvements of 1.48%–17.14% in ${\overline{Nu}}$ $\bar{N u}$ and 5.46%–8.17% in UI.