Scholarlink

A multi-objectiveMatar, C. Cinnella, P.Gloerfelt, X. shape optimization of a high-pressure turbine (HPT) vane for Organic-Rankine Cycles (ORC) is designed by fusing together high-fidelity, scale-resolving simulations, and specifically wall-resolved Large Eddy Simulations (WRLES), and Reynolds-Averaged Navier Stokes (RANS) solutions through a multi-fidelity co-Kriging surrogate model. Given the extremely high cost of WRLES at Reynolds numbers characteristic of ORC turbines, adaptive infill criteria are devised to smartly enrich the surrogate model during optimization on both levels of fidelity. While steady RANS solutions are known to fail in predicting key complex flow features present in turbine vanes, the present surrogate captures elements of the trends from the LES high-fidelity data and provides improved estimations of the better performing designs geometry parameters.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multifidelity Optimization of an ORC Turbine Vane Using Scale-Resolving Simulations

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Camille Matar,
  • Paola Cinnella,
  • Xavier Gloerfelt

摘要

A multi-objectiveMatar, C. Cinnella, P.Gloerfelt, X. shape optimization of a high-pressure turbine (HPT) vane for Organic-Rankine Cycles (ORC) is designed by fusing together high-fidelity, scale-resolving simulations, and specifically wall-resolved Large Eddy Simulations (WRLES), and Reynolds-Averaged Navier Stokes (RANS) solutions through a multi-fidelity co-Kriging surrogate model. Given the extremely high cost of WRLES at Reynolds numbers characteristic of ORC turbines, adaptive infill criteria are devised to smartly enrich the surrogate model during optimization on both levels of fidelity. While steady RANS solutions are known to fail in predicting key complex flow features present in turbine vanes, the present surrogate captures elements of the trends from the LES high-fidelity data and provides improved estimations of the better performing designs geometry parameters.