High-pressure turbine blades are subject to large thermomechanical loads that may threaten their mechanical integrity. The prediction of the heat transfer on the blade surface, crucial to ensure its durability, thus requires an accurate description of the flow physics around the blade to be reliable. In an effort to better qualify the use of computational fluid dynamics in this design context as well as the need for an improved understanding of the flow physics, this paper investigates a transonic highly loaded linear turbine blade cascade that has been found difficult to predict in the literature using large-eddy simulations. Indeed, the configuration results in shocks and acoustic waves on the suction side of the blade, features that are commonly encountered in high-pressure turbines. Turbulent spots are observed on the suction-side boundary layer with an inlet turbulence intensity of 6%. The turbulent spots are shown to have a complex and highly unsteady effect on the shock/boundary-layer interaction, disrupting flow detachment and creating laminar spots downstream of the shock. To address these transient flow phenomena, conditional averages based on the intermittency level are introduced to show that accurate heat transfer predictions require an accurate prediction of the rate of turbulent-spot production. The analysis then focuses on the effect of intermittency on the turbulent kinetic energy exchanges in the near-wall region, as the turbulent kinetic energy balance must be addressed in Reynolds-averaged Navier-Stokes models.
In the framework of wall-modeled large-eddy simulation (WMLES), the problem of combining sub-grid scale (SGS) models with the standard wall law is commonly acknowledged and expressed through multiple undesired near-wall behaviors. In this work, it is first observed that the static Smagorinsky model predicts efficiently the wall shear stress in a wall-modeled context, while more advanced static models like wall-adapting local eddy (WALE) viscosity or Sigma with proper cubic damping fail. It is, however, known that Smagorinsky is overall too dissipative in the bulk flow and in purely sheared flows, whereas the two other models are better suited for near-wall flows. The observed difficulty comes from the fact that the SGS model relies on the filtered velocity gradient tensor that necessarily comes with large errors in the near-wall region in the context of WMLES. Since the first off-wall node is usually located in the turbulent zone of the boundary layer, the turbulent structures within the first cell are neither resolved by the grid nor represented by the SGS model, which results in a lack of turbulent activity. In order to account for these subgrid turbulent structures, a stochastic forcing method derived from Reynolds-averaged Navier–Stokes (RANS) turbulence models is proposed and applied to the velocity gradients to better estimate the near-wall turbulent viscosity while providing the missing turbulent activity usually resulting from the WMLES approach. Based on such corrections, it is shown that the model significantly improves the wall shear stress prediction when used with the WALE and Sigma models.
As the Reynolds number increases, the large-eddy simulation (LES) of complex flows becomes increasingly intractable because near-wall turbulent structures become increasingly small. Wall modeling reduces the computational requirements of LES by enabling the use of coarser cells at the walls. This paper presents a machine-learning methodology to develop data-driven wall-shear-stress models that can directly operate, a posteriori, on the unstructured grid of the simulation. The model architecture is based on graph neural networks. The model is trained on a database which includes fully developed boundary layers, adverse pressure gradients, separated boundary layers, and laminar–turbulent transition. The relevance of the trained model is verified a posteriori for the simulation of a channel flow, a backward-facing step and a linear blade cascade.
Optimising the design of aviation propulsion systems using computational fluid dynamics is essential to increase their efficiency and reduce pollutant as well as noise emissions. Nowadays, and within this optimisation and design phase, it is possible to perform meaningful unsteady computations of the various components of a gas-turbine engine. However, these simulations are often carried out independently of each other and only share averaged quantities at the interfaces minimising the impact and interactions between components. In contrast to the current state-of-the-art, this work presents a 360 azimuthal degrees large-eddy simulation with over 2100 million cells of the DGEN-380 demonstrator engine enclosing a fully integrated fan, compressor and annular combustion chamber at take-off conditions as a first step towards a high-fidelity simulation of the full engine. In order to carry such a challenging simulation and reduce the computational cost, the initial solution is interpolated from stand-alone sectoral simulations of each component. In terms of approach, the integrated mesh is generated in several steps to solve potential machine dependent memory limitations. It is then observed that the 360 degrees computation converges to an operating point with less than 0.5% difference in zero-dimensional values compared to the stand-alone simulations yielding an overall performance within 1% of the designed thermodynamic cycle. With the presented methodology, convergence and azimuthally decorrelated results are achieved for the integrated simulation after only 6 fan revolutions.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.