Julien Bodart

Solving the compressible navier-stokes equations on up to 1.97 million cores and 4.1 trillion grid points

We present weak and strong scaling studies as well as performance analyses of the Hybrid code, a finite-difference solver of the compressible Navier-Stokes equations on structured grids used for the direct numerical simulation of isotropic turbulence and its interaction with shock waves. Parallelization is achieved through MPI, emphasizing the use of nonblocking communication with concurrent computation. The simulations, scaling and performance studies were done on the Sequoia, Vulcan and Vesta Blue Gene/Q systems, the first two accounting for a total of 1,966,080 cores when used in combination. The maximum number of grid points simulated was 4.12 trillion, with a memory usage of approximately 1.6 PB. We discuss the use of hyperthreading, which significantly improves the parallel performance of the code on this architecture.

Large eddy simulation of high-lift devices

Accurate predictions of flow around high lift devices is of interest for airframe noise predictions as well as wing design with or without active flow control. In particular, prediction of the maximum lift coefficient remains challenging using RANS solvers. We discuss how large eddy simulation can be used in this context, together with wall-modeling to reach flight Reynolds numbers. The study is focused on the flow around the McDonnellDouglas 30P/30N airfoil at several angles of attack and at Reynolds number (based on the stowed chord) of Rec = 9 · 10.

Direct numerical simulation of unsheared turbulence diffusing towards a free-slip or no-slip surface

The physics involved in the interaction between statistically steady, shearless turbulence and a blocking surface is investigated with the aid of direct numerical simulation. The original configuration introduced by Campagne et al. (G. Campagne, J.B. Cazalbou, L. Joly, and P. Chassaing, Direct numerical simulation of the interaction between unsheared turbulence and a free-slip surface, in ECCOMAS CFD 2006, P. Wesseling, E. Onate, and J. Periaux eds., TU Delft, The Netherlands, 2006) serves as the basis for comparing cases in which the blocking surface can be either a free-slip surface or a no-slip wall. It is shown that in both the cases, the evolutions of the anisotropy state are the same throughout the surface-influenced layer (down to the surface), despite the essentially different natures of the inner layers. The extent of the blocking effect can thereby be measured through a local (surface) quantity identically defined in the two cases. Examination of the evolution and content of the pressure–strain ...

Assessment of Wall-modeled Large Eddy Simulation For Supersonic Compression Ramp Flows

An assessment of a wall-model technique for large-eddy simulation (LES) is conducted in the context of a supersonic turbulent boundary layer flowing over a compression ramp. Both wall-resolved LES (WR-LES) and wall-modeled LES (WM-LES) calculations are performed of this canonical shock-turbulent boundary layer interaction (STBLI). The free stream Mach number is M∞ = 2.9 and ramp angle β = 24 ◦. Two Reynolds numbers are considered. The first is Reθ = 2400 which is accessible to WR-LES and is used for a direct comparison between WR-LES and WM-LES at matching conditions, while the second Reynolds number, Reθ = 120, 000, is inaccessible to WR-LES and therefore requires a wall modeled approach. Results at both Reynolds numbers are compared against experimental data at matching conditions. The use of WM-LES enables the study of high Reynolds number wall bounded flows that are inaccessible to traditional WR-LES. Analysis of wall-model performance, wall model deficiencies, and possible improvements are discussed.

ANALYSIS OF TURBULENT SCALAR FLUX MODELS FOR A DISCRETE HOLE FILM COOLING FLOW.

Algebraic closures for the turbulent scalar fluxes were evaluated for a discrete hole film cooling geometry using the results from a high-fidelity Large Eddy Simulation (LES). Several models for the turbulent scalar fluxes exist, including the widely used Gradient Diffusion Hypothesis, the Generalized Gradient Diffusion Hypothesis, and the Higher Order Generalized Gradient Diffusion Hypothesis. By analyzing the results from the LES, it was possible to isolate the error due to these turbulent mixing models. Distributions of the turbulent diffusivity, turbulent viscosity, and turbulent Prandtl number were extracted from the LES results. It was shown that the turbulent Prandtl number varies significantly spatially, undermining the applicability of the Reynolds analogy for this flow. The LES velocity field and Reynolds stresses were fed into a RANS solver to calculate the fluid temperature distribution. This analysis revealed in which regions of the flow various modeling assumptions were invalid and what effect those assumptions had on the predicted temperature distribution.

A parametrized non-equilibrium wall-model for large-eddy simulations

Wall-models are essential for enabling large-eddy simulations (LESs) of realistic problems at high Reynolds numbers. The present study is focused on approaches that directly model the wall shear stress, specifically on filling the gap between models based on wall-normal ordinary differential equations (ODEs) that assume equilibrium and models based on full partial differential equations (PDEs) that do not. We develop ideas for how to incorporate non-equilibrium effects (most importantly, strong pressure-gradient effects) in the wall-model while still solving only wall-normal ODEs. We test these ideas using two reference databases: an adverse pressure-gradient turbulent boundary-layer and a shock/boundary-layer interaction problem, both of which lead to separation and re-attachment of the turbulent boundary layer.

A Machine Learning Approach for Determining the Turbulent Diffusivity in Film Cooling Flows

In film cooling flows, it is important to know the temperature distribution resulting from the interaction between a hot main flow and a cooler jet. However, current Reynolds-averaged Navier-Stokes (RANS) models yield poor temperature predictions. A novel approach for RANS modeling of the turbulent heat flux is proposed, in which the simple gradient diffusion hypothesis (GDH) is assumed and a machine learning algorithm is used to infer an improved turbulent diffusivity field. This approach is implemented using three distinct data sets: two are used to train the model and the third is used for validation. The results show that the proposed method produces significant improvement compared to the common RANS closure, especially in the prediction of film cooling effectiveness.

Wall-modeled large eddy simulation of the McDonnell-Douglas 30P/30N high-lift airfoil in near-stall conditions

Accurate predictions of the ow around high lift devices is of particular interest for airframe noise predictions and wing design with or without active ow control. Whereas RANS modeling is often su ciently accurate for external aerodynamics at cruise conditions, it is less trustworthy at high-lift and near-stall conditions. We show that combining large eddy simulation with a wall-model is an attractive way to compute such ows at realistic Reynolds numbers, in conditions relevant to take-o and landing. At these Reynolds numbers, the grid requirements to resolve the viscous layers near the walls prevent the use of standard large eddy simulation without additional near-wall modeling. A wall-modeling procedure that is applicable to complex geometries is developed and incorporated in a massively parallel, unstructured grid compressible Navier-Stokes solver. A boundary-layer state sensor is developed in order to switch o the wall-model in laminar regions. The method is applied to the McDonnell-Douglas 30P/30N airfoil at a Reynolds number of Rec = 9:6 10 and a near-stall angle-of-attack = 19 .

Wall-Modeling in Complex Turbulent Flows

Resolution of wall layer turbulent structures in large eddy simulation of high Reynolds number flows of aeronautical interest requires inordinate computational resources. LES with wall models is therefore employed in engineering applications. We report on recent advances at the Center for Turbulence Research (CTR) in the development of wall boundary conditions for complex turbulent flows computed on unstructured grids. We begin by describing a novel application of wall modeled LES to a high lift airfoil system. This flow field is very complex involving boundary layers, free shear flows, separation and laminar/turbulence transition. We then describe a non-equilibrium model that requires the solution of the full 3D RANS equations in the near wall region. This model is successfully applied to a spatially evolving transitional and a high Reynolds number flat plate boundary layer. Finally we describe a new approach to LES using differential filters. An important byproduct of this approach is the derivation of slip velocity boundary conditions for wall modeled LES. This methodology is successfully applied to flow over NACA4412 airfoil at near stall conditions.

Local Large Scale Forcing of Unsheared Turbulence

With the objective of studying the interaction between turbulence and a solid wall, a new way to generate a statistically steady turbulent state in a DNS setup is presented.