Welcome to the HyperFlux Project

Contents:

Introduction

Computational Fluid Dynamics (CFD) is used in a wide range of engineering sectors for the analysis and design of products – from aircraft to racing cars and jet turbines. Current CFD technology is lacking in accuracy and efficiency for unsteady flow problems (including acoustics) and for resolving shed vortices and wakes. This is a real problem for industry: expensive physical prototypes are still required since computational tools are unreliable for these unsteady flow phenomena. High-order methods (HOMs) provide a potential solution, providing higher fidelity solutions than currently achievable with low-order schemes on unstructured grids. Dr. Peter Vincent and his group at Imperial College are experts in this field - underpinned by an EPSRC Early Career Fellowship and DTA PhD Studentships. The group will work with technical specialists at the Centre for Modelling and Simulation (CFMS) and cloud high performance computing (HPC) specialist Zenotech to create a prototype software base for industrial evaluation. HOMs has the potential to speed up the design cycle, reduce costs and improve products.

Airbus, ARA, BAE Systems, Airbus Group Innovation, Rolls-Royce, DSTL, Arup, Williams F1 and the newly formed UK Aerospace Technology Institute will be non-financial partners / provide evaluation. Each of the industrial primes will contribute a test case to evaluate the prototype software – validating and verifying it against existing data and processes, and providing feedback on the impact on the need for physical prototyping. The KTN will be directly involved to provide additional dissemination. CFMS will lead the integration and dissemination activities from its supercomputer facility at the Bristol and Bath Science Park. The software prototype will embody the latest in high-performance computing, particularly heterogeneous configurations of conventional and many-core processors for speed and energy efficiency. NVIDIA will support the project via its technical team in Bristol. Via remote (cloud) access to its virtual engineering hub, CFMS will make the prototype software available to other sectors (civil engineering, automotive and renewable energy) and support its uptake with local specialists.

Hyper Flux will be a shared UK-based software tool, underpinned by expertise within the UK. This is in line with government strategies for HVM and ICT, and forms a cornerstone for the new UK aerodynamics ATI. This will further establish a center of expertise in the application of the new models to on-ramp new users – particularly SMEs.

Algorithm

Test Cases

Unit Test Cases

Name Status Comment
Taylor Green vortex decay    
Decaying Isotropic Turbulence    
3D Cylinder static    
2D vortex across sliding interface    
3D Cylinder rotating    

Open Test Cases

Name Status Comment
NASA High Lift Prediction Workshop    
NASA Drag Prediction Workshop    
NATO RTO AVT VFE 2    
SMC006 Serrated Nozzle    
DARPA SUBOFF    
DARPA HIREP    
NASA Rotor 37    
Civil Aircraft Landing Gear    

Closed Test Cases

From ARUP, ARA, Willams F1

High Lift Prediction Workshop - Validation

Authors: A. Cimpoeru (CFMS), J. Appa (Zenotech) and D. Standingford (Zenotech)

November 2014

Abstract

This document summarizes some initial results obtained within the second order benchmarking of a new CFD software. The zCFD code was used for the DLR F11 high lift configuration in order to compute the flow field using the k-ω SST turbulence model. The results consist of high Reynolds number computations for the simplified (Case 1) and complex (Case 2B) configurations at \(7^\circ\) angle of attack. The results were validated against the numerical solutions obtained using CFD++ and ANSYS FLUENT and showed a good agreement with the experimental data.

Introduction

The zCFD code is a GPU accelerated high performance computational fluid dynamic software. The zCFD solver is an unstructured cell centered finite volume code which solves explicitly the compressible Navier-Stokes equations. The solution is preconditioned and the convergence is accelerated using the Multigrid technique along with a dual time stepping method.

The results consist in high Reynolds number computations using k-ω SST turbulence model for the simplified (Case 1) and complex (Case 2B) configurations at \(7^\circ\) angle of attack. The zCFD solutions were compared with wind tunnel data and other CFD codes such as Metacomp CFD++ and Ansys FLUENT. These codes were selected since they use the same cell-centered approach in combination with k-ω SST model.

Results

Case 1, Coarse mesh (9.5M cells), \(Mach=0.175\), \(Re=15e6\), \(\alpha=7^\circ\)

Case 2b, Solar mesh (50M cells), \(Mach=0.175\), \(Re=15e6\), \(\alpha=7^\circ\)

_images/case2b_oil_flow.png

DLR F11 High Lift Configuration; Locations of Pressure Sections. For the full definition of cutting planes refer to workshop website.

NASA Drag Prediction Workshop - NASA Common Research Model - Validation

Authors: A. Cimpoeru (CFMS), J. Appa (Zenotech) and D. Standingford (Zenotech)

December 2014

Configuration ( Wing-Body no Tail ) , L3 mesh unstructured hexahedral - 5.1 x 106 cells - Match CL = 0.5

alternate text

NASA Common Research Model

Conditions

Reynolds number = 5.0 x e06 based on MAC

Mach = 0.85

Angle of attack = 2.217

Reference static pressure = 101325 Pa

Reference static temperature = 310.928 K

RANS steady state

k-W SST model

Preconditioned

MUSCL scheme for

Results

\(\circ\) Isosurfaces of Mach number

alternate text

\(\circ\) CL match for L3 mesh unstructured hexahedral mesh

zCFD Experiment Error
CL = 0.4884 CL = 0.508 -0.46 %

\(\circ\) Code to code validation

alternate text
alternate text
alternate text

Numerical investigation of the flow field around \(65^\circ\) delta wing using the unstructured code zCFD

Authors: A. Cimpoeru (CFMS), J. Appa (Zenotech) and D. Standingford (Zenotech)

December 2014

Abstract

In the present study the flow field around \(65^\circ\) delta wing is investigated using the unstructured density based solver zCFD. The simulations were carried out by solving the steady state RANS equations using the k-ω SST turbulence model at :math:‘13.3^circ’ flow angle for :math: ‘Re=6.0e06’ (NASA) and :math:’Re=3.0e06’ based on MAC. Within this analysis the numerical solutions are compared against wind tunnel results performed at NASA and DLR.

Geometry Specification

The geometry has been generated using the analytical definition available in the reference (Chu and Luckring). For the present study two sets of simulations were performed: (1) Comparison against the experiment performed at NASA and (2) Comparison against the experiment performed at DLR as part of the VFE 2 project.

alternate text

Delta wing geometry

Results

Steady state RANS, SOLAR - octree mesh (30M cells), \(Mach=0.4\), \(Re=6.0e06\) (based on MAC = 0.3268 m)

\(\circ\) Large Radius - Validation against NASA Wind Tunnel Experiment

alternate text

\(\circ\) Sharp Radius - Validation against NASA Wind Tunnel Experiment

alternate text

\(\circ\) Large (right) and Sharp radius (left)

alternate text

See also

Experimental data and geometry specification (4 volumes available on NASA Technical Reports Server)

DLR Project VFE-2

J Chu and J M Luckring. Experimental Surface Pressure Data Obtained on 65 degrees Delta Wing Across Reynolds Number and Mach Number Ranges. NASA TM 4645, 1996

R M Cummings and A Schute. Detached-Eddy Simulation od the vortical flow field about the VFE-2 delta wing. Aerospace Science and Technology 24 (2013) 66-76

R Konrath, C Klein and A Schroder. PSP and PIV investigations on the VFE-2 configuration in sub- and transonic flow. Aerospace Science and Technology 24 (2013) 22-31

Delta Wing Sharp Radius Notebook

Validation of the flow field around DARPA SUBOFF model

Authors: A. Cimpoeru (CFMS), J. Appa (Zenotech) and D. Standingford (Zenotech)

December 2014

Abstract

In this section the flow field around DARPA SUBOFF geometry is computed using the unstructured cell centered finite volume density based solver zCFD. For this study the AFF8 configuration was employed due to complex flow features such as boundary layer - vortex and vortex-vortex interactions. The Reynolds number for this test case is \(Re=1.2e07\) and the employed turbulence model is k-ω SST. The steady state solutions are validated against wind tunnel data (see Reference) and have shown agreement.

Configuration (Axisymmetric hull + Fairwater + Stern appendages )

alternate text

SUBOFF Configuration

Results

Steady state RANS, SOLAR - octree mesh (14 and 30 M cells), \(Mach=0.0179\), \(Re=1.2e7\) (model length based)

alternate text

Coefficient of pressure in the symmetry plane

alternate text

Coefficient of pressure contours

Wake development ( Left – Coarse Mesh ; Right – Fine Mesh )

alternate text

x/L = 0.3

alternate text

x/L = 0.4

alternate text

x/L = 0.5

alternate text

x/L = 0.978 (Propeller Plane)

Comparison against zCFD, experimental data and Bhushan et al (2013) (FLUENT)

alternate text

x/L = 0.978 (Propeller Plane)

See also

Summary of DARPA Suboff Experimental program data

Geometric Characteristics of DARPA SUBOFF models

‘S. Bhushan, M. F. Alam and D. K. Walters. Evaluation of hybrid RANS/LES models for prediction of flow around surface combatant and Suboff geometries. Computer and Fluids 88 (2013) 834-849’

DARPA SUBOFF Notebook

RANS simulations of cold jet flows from a serrated nozzle

Authors: A. Cimpoeru (CFMS), J. Appa (Zenotech) and D. Standingford (Zenotech)

February 2015

Abstract

Geometry and Mesh Generation (SMC-006 Nozzle)

alternate text

SMC006 chevron type nozzle. Left (Model) and Right (Domain Topology ) (Xia et. al. (2009))

alternate text

SMC006 mesh on x-y plane. Multiblock Stuctured Mesh (19M cells) (Xia H. and Tucker P. (2009))

Initial Conditions (Steady-State RANS)

Ambient Conditions

Variable Value Unit
P_amb 97000 Pa
T_amb 280.2 K
Rho_amb 1.225 Kg/\(m^3\)
mu 1.79e-5 Pa.s
speed_of_sound 335.549 m/s
gas_constant 287.0 KJ/KgK

Jet Conditions

Variable Value  
NPR 1.83 Nozzle Pressure Ratio
TPR 1.022 Nozzle Temperature Ratio
Mjet 0.9 Jet Mach number
Ujet 300 Reference Velocity [m/s]
Reynolds 1.03e06 Reynolds Number (See Note)

Results

alternate text

Jet velocity profiles in the Tip-to-Tip plane at different stations

alternate text

Jet velocity profiles in the Notch plane at different stations

alternate text

Comparison between LES study of Xia et. al. (2009) (top) and present RANS study (bottom) at different stations

alternate text

Isosurfaces of Q criterion (present study)

  • Note

The Reynolds number is based on the jet diameter and reference velocity

See also

‘Hao Xia, Paul G. Tucker and Simon Eastwood (2009). Large-eddy simulations of chevron jet flows with noise predictions. International Journal of Heat and Fluid Flow 30 (2009) 1067-1079.’_

‘Hao Xia and Paul G. Tucker (2011). Numerical Simulation of a Single-Stream Jets from a Serrated Nozzle . Flow Turbulence Combust 2011.’_

Serrated Nozzle Notebook

NASA Rotor 37 zCFD Code Validation - Flow Field in a Transonic Axial Compressor

Authors: A. Cimpoeru (CFMS), J. Appa (Zenotech) and D. Standingford (Zenotech)

May 2015

Abstract

Geometry and Mesh Generation (NASA ROTOR37)

alternate text
alternate text

Parameters

alternate text

Initial Conditions (Steady-State RANS)

Ambient Conditions

Variable Value Unit
P_amb 101523 Pa
T_amb 288.15 K
Rho_amb 1.225 Kg/\(m^3\)
mu 1.79e-5 Pa.s
gas_constant 287.0 KJ/KgK

Inflow Total Conditions.

Variable Ratio
Ptotal/P_amb 1.0
T_total/T_amb 1.0

Results

Note that the map has been initially generated by varying the static pressure ratio on the outflow from 0.7 to 1.8 in order to determine the chocked and stalled conditions.

alternate text

Cp distribution on the rotor blades

alternate text

Mach number in the flow field and pressure contours.

alternate text

Turbulent eddy viscosity and pressure contours

More results will be publised in the following weeks.

References