A high-resolution DNS study of compressible flow past an LPT blade in a cascade

TL;DR

This study employs a high-resolution DNS approach with a novel code to accurately simulate compressible flow past a low pressure turbine blade, achieving detailed flow features and resolving previous discrepancies with experimental data.

Contribution

Development and validation of the ANUROOP code for high-resolution DNS of compressible flow on hybrid grids, enabling more accurate turbine blade flow simulations.

Findings

Discrepancies in pressure distribution are resolved.

Detailed characterization of the trailing edge separation bubble.

Surface curvature effects are analyzed in detail.

Abstract

Flow past a low pressure turbine blade in a cascade at $Re \approx 52000$ and angle of incidence $\alpha = 45.5^{0}$ is solved using a code developed in-house for solving 3D compressible Navier-Stokes equations. This code, named ANUROOP, has been developed in the finite volume framework using kinetic energy preserving second order central differencing scheme for calculating fluxes, and is compatible with hybrid grids. ANUROOP was verified and validated against several test cases with Mach numbers ranging from 0.1 (Taylor-Green vortex) to 1.5 (compressible turbulent channel flow). The code was found to be robust and stable, and the kinetic energy decay obeys the compressible Navier-Stokes equations. A hybrid grid, with a high resolution hexahedral orthogonal mesh in the boundary layer and unstructured (also hexahedral) elements in the rest of the domain, is used for the turbine blade simulation. Total grid size (160 million) is approximately an order of magnitude higher than in previous simulations for the same flow conditions and using similar numerical methods. The discrepancy in the pressure distribution in earlier studies compared to experimental data has been removed in this simulation. The trailing edge separation bubble has been characterized and a detailed discussion on the effect of surface curvature is presented.