Scholarlink

The increasing demand for engineers’ design capabilities motivates the enhancement of numerical tools to leverage the advancements in computing capabilities. In this context, advanced tools capable of interactively evaluating designs by adjusting component shapes can make a significant difference. The growing adoption of physics-informed artificial intelligence (AI) appears to be a promising avenue. In this paper, we present an interactive dashboard for designing a NACA (National Advisory Committee for Aeronautics) intake, applied to an aeronautical device. The objective was to identify the geometry that maximizes outlet pressure while optimizing pressure recovery. The workflow began with defining 8 design parameters and 1 physical parameter (inlet velocity), applied directly to a high-fidelity CFD model. Shape parameterization was achieved using Radial Basis Functions (RBF) mesh morphing. Subsequently, 80 snapshots were computed using high-performance computing (HPC) and then compressed using Proper Orthogonal Decomposition (POD). Machine learning algorithms facilitated linking the design parameters to mode coefficients. The resulting Reduced Order Model (ROM) enabled real-time prediction of intake performance and visualization of pressure distribution and flow field. An advanced design dashboard was generated, allowing for the selection of a geometry capable of increasing outlet pressure by 160% compared to a baseline intake designed using preliminary design tools.

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

An Interactive Design Tool for NACA Intakes Based on High Fidelity CFD Simulations and Reduced Order Models

  • 章节

  • 发表日期：

    2026年

查看全文
  • Andrea Lopez,
  • Viola Rossano,
  • Ubaldo Cella,
  • Corrado Groth,
  • Marco E. Biancolini

摘要

The increasing demand for engineers’ design capabilities motivates the enhancement of numerical tools to leverage the advancements in computing capabilities. In this context, advanced tools capable of interactively evaluating designs by adjusting component shapes can make a significant difference. The growing adoption of physics-informed artificial intelligence (AI) appears to be a promising avenue. In this paper, we present an interactive dashboard for designing a NACA (National Advisory Committee for Aeronautics) intake, applied to an aeronautical device. The objective was to identify the geometry that maximizes outlet pressure while optimizing pressure recovery. The workflow began with defining 8 design parameters and 1 physical parameter (inlet velocity), applied directly to a high-fidelity CFD model. Shape parameterization was achieved using Radial Basis Functions (RBF) mesh morphing. Subsequently, 80 snapshots were computed using high-performance computing (HPC) and then compressed using Proper Orthogonal Decomposition (POD). Machine learning algorithms facilitated linking the design parameters to mode coefficients. The resulting Reduced Order Model (ROM) enabled real-time prediction of intake performance and visualization of pressure distribution and flow field. An advanced design dashboard was generated, allowing for the selection of a geometry capable of increasing outlet pressure by 160% compared to a baseline intake designed using preliminary design tools.