Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Efficient Wall-Modelled Large Eddy Simulation of Rotors using Homogenized Lattice Boltzmann Methods

Adrian Kummerländer, Shota Ito, Maximilian Schecher, Davide Dapelo, Stephan Simonis, Mathias J. Krause, Fedor Bukreev

TL;DR

The work introduces a blade-resolved wall-modeled LES framework for rotor flows using a homogenized lattice Boltzmann method that targets the filtered Brinkman–Navier–Stokes equations. By combining a D3Q19 HL-BM discretization with a hybrid regularized collision and a Smagorinsky SGS model, and coupling it to a turbulent wall model and moving-geometry FSI, the approach achieves accurate blade-scale predictions while maintaining computational efficiency on GPUs. Validation against experimental and blade-resolved references demonstrates good agreement for thrust and wake profiles, supported by grid-convergence evidence (EOC ≈ 1.38 and GCI ≈ 0.92%). Performance analysis shows scalable weak performance up to 384 GPUs on 41 billion cells, confirming the method’s potential for simulating entire wind-farm layouts with blade-resolved WMLES in HPC environments.

Abstract

Accurately capturing the dynamic forces acting on rotors as well as their wake effects presents a significant challenge for computational fluid dynamics (CFD) due to high Reynolds numbers and a large range of spatio-temporal scales. The present work proposes a novel blade-resolved wall-modeled large eddy simulation (WMLES) approach based on the lattice Boltzmann method (LBM). A homogenized hybrid regularized recursive collision scheme targeting the filtered Brinkman--Navier--Stokes equations is combined with a novel wall-model. This is implemented in the context of a platform-transparent framework for fluid-structure interaction in the open source LBM framework OpenLB. Convergence order and accuracy are validated against both experimental and numerical data for a model wind turbine, demonstrating excellent agreement for integral forces and wake velocity profiles. Computational efficiency and parallel scalability was investigated by roofline analysis and weak scaling studies for up to 384 rotors resolved by 41 billion lattice cells on the Karolina supercomputer. The proposed framework enables efficient blade-resolved WMLES of entire wind farms and offers a new methodology for other complex wall-modeled fluid-structure interaction applications.

Efficient Wall-Modelled Large Eddy Simulation of Rotors using Homogenized Lattice Boltzmann Methods

TL;DR

The work introduces a blade-resolved wall-modeled LES framework for rotor flows using a homogenized lattice Boltzmann method that targets the filtered Brinkman–Navier–Stokes equations. By combining a D3Q19 HL-BM discretization with a hybrid regularized collision and a Smagorinsky SGS model, and coupling it to a turbulent wall model and moving-geometry FSI, the approach achieves accurate blade-scale predictions while maintaining computational efficiency on GPUs. Validation against experimental and blade-resolved references demonstrates good agreement for thrust and wake profiles, supported by grid-convergence evidence (EOC ≈ 1.38 and GCI ≈ 0.92%). Performance analysis shows scalable weak performance up to 384 GPUs on 41 billion cells, confirming the method’s potential for simulating entire wind-farm layouts with blade-resolved WMLES in HPC environments.

Abstract

Accurately capturing the dynamic forces acting on rotors as well as their wake effects presents a significant challenge for computational fluid dynamics (CFD) due to high Reynolds numbers and a large range of spatio-temporal scales. The present work proposes a novel blade-resolved wall-modeled large eddy simulation (WMLES) approach based on the lattice Boltzmann method (LBM). A homogenized hybrid regularized recursive collision scheme targeting the filtered Brinkman--Navier--Stokes equations is combined with a novel wall-model. This is implemented in the context of a platform-transparent framework for fluid-structure interaction in the open source LBM framework OpenLB. Convergence order and accuracy are validated against both experimental and numerical data for a model wind turbine, demonstrating excellent agreement for integral forces and wake velocity profiles. Computational efficiency and parallel scalability was investigated by roofline analysis and weak scaling studies for up to 384 rotors resolved by 41 billion lattice cells on the Karolina supercomputer. The proposed framework enables efficient blade-resolved WMLES of entire wind farms and offers a new methodology for other complex wall-modeled fluid-structure interaction applications.
Paper Structure (12 sections, 26 equations, 13 figures, 1 table)

This paper contains 12 sections, 26 equations, 13 figures, 1 table.

Table of Contents

  1. Introduction
  2. Methodology
  3. Filtered Brinkmann--Navier--Stokes Equations
  4. Homogenized Lattice Boltzmann Method
  5. Turbulent Wall Model
  6. Fluid-Structure Interaction
  7. Implementation
  8. Representation of Moving Geometries
  9. Moving Wall Model
  10. Validation
  11. Performance
  12. Conclusion

Figures (13)

  • Figure 1: Illustrative volumetric rendering of the vorticity generated by a four-turbine offshore wind farm modeled as a two-way coupled blade-resolved LES in OpenLB.
  • Figure 2: Schematic of the discrete velocity set $D3Q19$.
  • Figure 3: Schematic of the wall-modeled region. Wall-modeled cell marked in green.
  • Figure 4: Efficient local update of porosities from signed distance geometry
  • Figure 5: Representations of rotor geometry for wall-modeling
  • ...and 8 more figures

Theorems & Definitions (4)

  • Definition 1: Lattice Porosity
  • Definition 2: Tip chord based Reynolds number
  • Definition 3: Thrust Coefficient
  • Definition 4: Grid Convergence Index