New developments for automated multi-physics simulations exploring the global-local Generalized Finite Element Method

Abstract

Turbulent combustion physics in aerospace systems typically features interactions between turbulence and chemical reactions that span a wide range of temporal and spatial scales. Even with the rapid advancement in high performance computing, direct numerical simulations (DNS) remain restricted to small-scale turbulent combustion problems. While large-eddy simulations (LES) are well suited to model large-scale physics, the effects of small-scale physics must be incorporated through closure models, most of which are assumption-based and cannot represent the true physics, especially at conditions of interest to Air Force (AF). Against this landscape, an ongoing revolution in data sciences opens an avenue to incorporate small-scale physics from DNS to LES to enable accurate and efficient turbulent combustion simulations. Therefore, the objective of the present proposal is to formulate a systematic and generally applicable data-driven modeling framework to 1) inform effective reduced-order models (ROMs) from DNS to represent subgrid-scale (SGS) or subfilter-scale (SFS) physics (hereafter referred to as DNSROM); and 2) enhance LES by incorporating DNS-ROM for SGS-SFS physics to enable efficient and accurate turbulent combustion simulations in aerospace systems of direct interest to AF.

Document Details

Document Type
DoD Grant Award
Publication Date
Feb 29, 2024
Source ID
FA95502310212

Entities

People

  • Sergio Persival Proenca

Organizations

  • Air Force Office of Scientific Research
  • United States Air Force
  • University of São Paulo

Tags

Fields of Study

  • Physics

Readers

  • Computational Fluid Dynamics (CFD)

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms
  • Space
  • Space - Hall-Effect Thruster