Source Term-Based Synthetic Turbulence Generator Applied to Compressible DNS of the T106A Low-Pressure Turbine

João Anderson Isler, Guglielmo Vivarelli, Chris D. Cantwell, Francesco Montomoli, Spencer J. Sherwin, Yuri Frey, Marcus Meyer, Raúl VázquezInternational Journal of Turbomachinery Propulsion and Power 2025 · DOI

TL;DR

This paper introduces a novel source-term based synthetic turbulence generator (SEM) for high-fidelity simulations of low-pressure turbines. By applying this method to the T106A turbine, the researchers demonstrate its ability to accurately capture complex flow phenomena like separation and transition, revealing that turbulence length scale plays a critical role, comparable to turbulence intensity, in dictating flow behavior and losses.

What Problem Does This Paper Address?

Accurately simulating the complex aerodynamic flows within turbomachinery, particularly low-pressure turbines, is crucial for improving engine efficiency and reducing emissions. A key challenge in these simulations, especially for high-fidelity methods like Direct Numerical Simulation (DNS) and Large Eddy Simulation (LES), is generating realistic incoming turbulence at the inlet. Traditional methods often struggle with stability, computational cost, or fail to capture the anisotropic nature of turbulence relevant to turbine flows. This paper aims to address this by developing and validating a robust and efficient synthetic turbulence generation method that can be integrated into advanced computational frameworks.

Methods

The core of this work is the implementation of a source-term formulation of the Synthetic Eddy Method (SEM) within the Nektar++ spectral/hp element framework. Unlike traditional SEM that injects turbulence only at the inlet, this source-term approach injects eddies directly into a specified region of the computational domain. This offers greater flexibility and can reduce the need for extensive mesh refinement near the inlet.

Key methodological aspects include:

  • Source-Term SEM: Eddies are introduced as a source term in the compressible Navier-Stokes equations, driving the flow towards a desired turbulent state. This method was validated using a turbulent plane channel flow simulation.
  • Nektar++ Framework: This open-source, high-order spectral/hp element solver was utilized for its scalability and flexibility.
  • Sponge Layer: A sponge layer was implemented at both the inflow and outflow boundaries to absorb spurious acoustic waves and prevent reflections, enhancing numerical stability.
  • Variable Polynomial Order: Nektar++'s capability to use different polynomial orders for adjacent elements was leveraged. A lower order was used in the outflow region to diffuse vortices and turbulence, further aiding stability and reducing computational cost.
  • Anisotropic Turbulence: The SEM was configured to generate anisotropic turbulence, which is more representative of real-world flow conditions in turbines.

Datasets

The study utilizes experimental data and existing DNS data for validation. Specifically, the authors compare their results with:

  • Experimental data: From Stadtmuller [40] and other sources for the T106A turbine.
  • DNS data: From previous studies on similar turbine configurations (e.g., Michelassi et al. [3], Garai et al. [5], Sandberg and Michelassi [7], Kim et al. [34], Moser et al. [35]).

No new datasets were introduced in this paper.

Key Insights and Ideas

  • Turbulence Length Scale Matters (A Lot!): The most significant finding is that the turbulence length scale has a profound impact on flow behavior, comparable to turbulence intensity. This challenges the common perception that intensity is the primary driver.
  • SEM Source-Term is Robust: The source-term formulation of SEM proves to be a reliable and efficient method for generating realistic turbulence in complex turbomachinery flows, even for compressible simulations.
  • Sponge Layer + Variable Order = Stability: The combination of a sponge layer and variable polynomial order in Nektar++ significantly enhances numerical stability, crucial for high-fidelity simulations.
  • Delayed Separation and Transition: Increased turbulence intensity and specific length scales can delay flow separation and transition, leading to improved aerodynamic performance.
  • Wake Behavior is Sensitive: The mixing and characteristics of the turbine wake are highly sensitive to both turbulence intensity and length scale.

Key Results

  • Turbulence Intensity Effects:
    • Increasing turbulence intensity delays separation and reduces the separation bubble size.
  • For intensities > 2%, open separation bubbles form near the trailing edge.
  • Peak wake loss decreases with increasing turbulence intensity.
  • Turbulence Length Scale Effects (at 3.2% Tu):
    • Smaller length scales (2% chord) lead to higher turbulence and mixing downstream, but don't significantly influence transition or wake profiles.
  • A length scale around the boundary-layer height (5% chord) was most effective in energizing the boundary layer, delaying separation and transition, and creating an open bubble.
  • The 8% chord length scale showed less mixing and a well-behaved vortex.
  • The 5% chord length scale had the most significant effect on wake losses.
  • Validation: The simulation results show good agreement with experimental and existing DNS data for pressure distribution, skin friction, and wake loss profiles.

Limitations and Future Work

  • Limitations:
    • The study acknowledges that obtaining more precise turbulence intensity statistics would require longer time-averaging periods.
  • While the method is validated, the paper focuses on a specific turbine geometry (T106A).
  • The paper notes that spurious pressure fluctuations can still occur if the synthetic velocity field is not perfectly divergence-free.
  • Future Work:
    • The authors suggest that this source-term SEM approach can be extended to other turbomachinery applications.
  • Further investigation into the interplay between turbulence length scales and boundary-layer height is implied.
  • The paper's contribution to developing more robust and efficient models for high-fidelity turbine simulations is highlighted, suggesting continued development in this area.

Summary updated

Abstract

Direct numerical simulations (DNSs) of the T106A low-pressure turbine were conducted for various turbulence intensities and length scales to investigate their effects on flow behaviour and transition. A source-term formulation of the synthetic eddy method (SEM) was implemented in the Nektar++ spectral/hp element framework to introduce anisotropic turbulence into the flow field. A single sponge layer was imposed, which covers the inflow and outflow regions just downstream and upstream of the inflow and outflow boundaries, respectively, to avoid acoustic wave reflections on the boundary conditions. Additionally, in the T106A model, mixed polynomial orders were utilized, as Nektar++ allows different polynomial orders for adjacent elements. A lower polynomial order was employed in the outflow region to further assist the sponge layer by coarsening the mesh and diffusing the turbulence near the outflow boundary. Thus, this study contributes to the development of a more robust and efficient model for high-fidelity simulations of turbine blades by enhancing stability and producing a more accurate flow field. The main findings are compared with experimental and DNS data, showing good agreement and providing new insights into the influence of turbulence length scales on flow separation, transition, wake behaviour, and loss profiles.
BibTeX
@article{isler2025source,
  author = {João Anderson Isler and Guglielmo Vivarelli and Chris D. Cantwell and Francesco Montomoli and Spencer J. Sherwin and Yuri Frey and Marcus Meyer and Raúl Vázquez},
  title = {{Source Term-Based Synthetic Turbulence Generator Applied to Compressible DNS of the T106A Low-Pressure Turbine}},
  journal = {International Journal of Turbomachinery Propulsion and Power},
  year = {2025},
  doi = {10.3390/ijtpp10030013},
}
Figure