Session: 32-01 High Fidelity CFD

Paper Number: 121022

121022 - Towards New Insights in Gas Turbine Aerothermodynamics With Wall-Modeled LES and Immersed Boundary Method 

High-Reynolds gas turbine aerodynamics represent a salient and contemporary engineering challenge. One of the most notable challenges associated with the operation of gas turbine vanes at high speed is the potential for the flow to reach transonic regimes, thereby introducing many intricate aerodynamic phenomena, including boundary layer separation and shock-wave/boundary layer interactions.
In addressing these issues, Direct Numerical Simulations (DNS) are recognized as the most accurate approach; however, they often prove unfeasible in practical scenarios due to their computational demands. As a promising alternative, Large Eddy Simulation (LES) has emerged, surpassing conventional Reynolds Averaged Navier-Stokes (RANS) strategies by directly resolving the dynamics of energy-dominant, flow-dependent large eddies on the computational grid, as opposed to modeling them (Bose et al., 2018).
Nonetheless, the computational requirements of Wall-Resolved LES (WRLES), primarily driven by the necessity for high-resolution near solid boundaries, still render it unsuitable for real-world operating conditions. This is because the number of grid points required for a WRLES arrangement scales with the second power of the Reynolds number, making computation infeasible on standard architectures. Consequently, various approaches seeking to merge the LES framework with RANS methods have emerged in recent years. Among these, the Wall-Modelled LES (WMLES) approach has garnered significant attention. The method combines conventional LES techniques for resolving the most significant flow structures with a wall-stress/heat-flux model to address near-wall dynamics, demonstrating superior accuracy compared to other hybrid/zonal numerical discretizations (Kawai et al., 2012).

In addition to these challenges, efficiently handling the complex geometries inherent to gas turbine technologies and combining complex geometries with highly efficient and massive parallel solvers presents another significant computational obstacle. The Immersed Boundary Method (IBM) has emerged as a promising strategy in this path. IBM, in fact, allows the body surface to intersect computational cells, enabling the use of a Cartesian mesh, regardless of geometric complexity, thus making the solver structure prone to scale over thousands of computational units. However, IBM faces difficulty accurately resolving near-wall regions due to local mismatches between the mesh and the physical body. 

Thus, our receipt is to provide a robust combination of WMLES, a method designed to set the first off-the-wall point as far as possible from the body surface, with the IBM strategy. This integrated approach provides an efficient computational framework for tackling the challenging conditions of high Reynolds/high-Mach number flows. Moreover, it seamlessly integrates with contemporary Graphics Processing Units (GPUs) architectures, offering simulation cost advantages and enhanced system physics description.

In light of these challenges, the present study introduces an innovative technique that combines IBM with a WMLES approach to analyze the aerothermodynamics of a gas turbine stator cascade. To the best of our knowledge, no prior research in this direction has been documented in the literature, marking this as a pioneering effort to elucidate the intricate physics of these systems.
In particular, the present research aims to rigorously validate the numerical approach over the well-experimentally documented transonic turbine vane by Arts et al., 1990, "Aero-thermal investigation of a highly loaded transonic linear turbine guide vane cascade". Following a detailed comparison between the experimental and numerical arrangements, the full potential of the proposed method will be harnessed by examining the time-dependent behavior of the system. This critical phase of the study will shed light on the dynamic aspects of gas turbine aerothermodynamics, providing insights that are challenging to explore with standard CFD approaches, thus offering an alternative to overcome the limitations often associated with conventional RANS in capturing the intricate transient behaviors inherent to gas turbine systems. 

Presenting Author: Francesco De Vanna Università degli Studi di Padova

Presenting Author Biography: Francesco De Vanna is an Assistant Professor in Machines Fluid Dynamics and Computational Fluid Dynamics at the University of Padova, Italy. His research primarily focuses on Numerical Gas Dynamics and flow modeling in both fundamental and engineering applications. Francesco De Vanna is the main developer of URANOS, an open-source, fully compressible Navier-Stokes solver tailored for complex geometries internal and external aerodynamics simulations.

Authors:

Francesco De Vanna Università degli Studi di Padova
Ernesto Benini Università degli studi di Padova