6th AIAA CFD Drag Prediction Workshop · 6th AIAA CFD Drag Prediction Workshop Summary of results...

6th AIAA CFD Drag Prediction WorkshopSummary of results from the CFD++ software suite

Brian A. Edge

Metacomp Technologies, Inc.

June 2016

slide 1 of 28

Corresponding Participant

Name: Brian EdgeEmail: [email protected]: (818) 735-4880

Address: Metacomp Technologies, Inc.28632-B Roadside Drive, Suite 255Agoura Hills, CA 91301

slide 2 of 28

Metacomp Participation

I Tasks Performed:1, 2, 3, and 5

I Software used:CFD++ Software Suite: CFD++, CSM++, MetaFSI, MIME

I CFD++ Basic Algorithms:

I Unified unstructured higher-order TVD interpolationconvection scheme

I Cell- and vertex-based polynomial reconstructionI Positivity-preserving Riemann solver-based flux computationI Advanced algebraic multi-grid agglomeration linear solver

slide 3 of 28

Task 1: Verification Study

Conditions:Ma=0.15, Re=6 million, α=10 degrees, farfield νt/ν=0.2104

104 105 106 107

1

1.02

1.04

1.06

1.08

1.1

N

CL

Coefficient of lift versus number of cells (N)

104 105 106 107

0.0125

0.0150

0.0175

0.0200

0.0225

0.0250

0.0275

N

CD

Coefficient of drag versus number of cells (N)

SA

SST

Reported Values

104 105 106 107

0.007

0.008

0.009

0.010

0.011

N

Cm

Pitching moment (Cm) versus number of cells (N)

SA

SST

Reported Values

0 0.2 0.4 0.6 0.8 1

−6.0

−5.0

−4.0

−3.0

−2.0

−1.0

0.0

1.0

x/c

Cp

Coefficient of Pressure(Cp) curves (SA vs SST)

Mesh 1 - SA

Mesh 1 - SST

Gregory and O’Reilly

The dashed lines represents the infinitely-refined results obtained from 3 codes (FUN3D, CFL3D, and TAU).

slide 4 of 28

Task 2: Drag Increment Study

slide 5 of 28

Task 2: Drag Increment

I Two grids:I “unstructured NASA GeoLab.REV00”I “Boeing Babcock Unstructured CC.REV00” grid families.

I Four turbulence models:I Linear Eddy Viscosity

I Spalart-AllmarasI SST

I Non-Linear Eddy ViscosityI HellstenI SA-RC-QCR

slide 6 of 28

Task 2: Grid Comparison

Number of cells

unstructured NASA GeoLab.REV00 Boeing Babcock Unstructured CC.REV00Grid Grid Level WB WBNP WB WBNP

TINY 1 83,578,942 120,909,566 20,657,615 27,015,892COARSE 2 122,816,245 178,924,829 26,271,819 35,271,269MEDIUM 3 181,953,555 266,818,466 33,683,206 45,687,005FINE 4 271,262,930 399,877,018 43,126,748 60,174,840XTRAFINE 5 404,235,547 597,491,792 56,413,328 79,548,552ULTRAFINE 6 606,531,721 901,459,751 71,169,688 101,639,992

slide 7 of 28

Task 2: Grid Comparison (Symmetry Plane)

Boeing Babcock Unstructured CC.REV00Grid Level 1 (TINY)

unstructured NASA GeoLab.REV00Grid Level 1 (TINY)

slide 8 of 28

Task 2: Grid Comparison (Surface and Prism Layers)



slide 9 of 28


Cp contours on WB GeometryGrid Level 6 (ULTRAFINE)



slide 10 of 28


Angle-of-Attack at CL = 0.5± 0.0001. Results were obtained using theSpalart-Allmaras turbulence model.

TINY

COARSE

MEDIUM

FINE

XTRAFINE

ULT

RAFINE

2.4

2.6

2.8

3

WB Data

WBNP Data

α

Angle-of-attack versus grid refinement level

TINY

COARSE

MEDIUM

FINE

XTRAFINE

ULT

RAFINE

0.0240

0.0250

0.0260

0.0270

0.0280

0.0290

WB Data

WBNP Data

CD

CD versus grid refinement level

NASA-WBBoeing-WB

NASA-WBNPBoeing-WBNP

slide 11 of 28

Task 2: Grid Computational Resource Needs

System Specs:192 Intel(R) Xeon(R) E5-2620 v3 CPUs running at 2.40GHz,

InfiniBand interconnect 4X FDR 56GB/sec.

WB Geometry“Boeing Babcock Unstructured CC.REV00”

Grid Grid Level Run Time [h] RAM [GB]

TINY 1 1.5 87COARSE 2 1.8 108MEDIUM 3 2.0 135FINE 4 2.7 170XTRAFINE 5 3.2 218ULTRAFINE 6 3.8 272

For Comparison:unstructured NASA GeoLab.REV00 (WB) - MEDIUM: 22.4 [h]

slide 12 of 28


Our decision: use the “Boeing Babcock Unstructured CC.REV00”grid family for remaining studies.

I It provides “similar” results to the larger unstructured NASAgrid

I It has fewer cells and requires less CPU resource

slide 13 of 28

Task 2: Turbulence Model Comparison

Angle-of-Attack (α) versus grid refinement level.TINY

COARSE

MEDIUM

FINE

XTRAFINE

ULT

RAFINE

2.4

2.6

2.8

3

WB Data

α

Wing-Body (WB) geometry

TINY

COARSE

MEDIUM

FINE

XTRAFINE

ULT

RAFINE

2.4

2.6

2.8

3

WBNP Data

α

Wing-Body-Nacelle-Pylon (WBNP) geometry

SASST

HellstenSA-RC-QCR

slide 14 of 28

Task 2: Turbulence Model Comparison

CD versus grid refinement level.

TINY

COARSE

MEDIUM

FINE

XTRAFINE

ULT

RAFINE

0.0240

0.0260

0.0280

0.0300

WB Data

CD

Wing-Body (WB) geometry

TINY

COARSE

MEDIUM

FINE

XTRAFINE

ULT

RAFINE

0.0240

0.0260

0.0280

0.0300

WBNP DataCD

Wing-Body-Nacelle-Pylon (WBNP) geometry

SASST

HellstenSA-RC-QCR

slide 15 of 28

Task 2: Turbulence Model Comparison: Cp Curves

0 0.2 0.4 0.6 0.8 1

−1

−0.5

0

0.5

1

x/c

Cp

Cp at Wing Section 14

0 0.2 0.4 0.6 0.8 1

−1

−0.5

0

0.5

1

x/c

Cp


0 0.2 0.4 0.6 0.8 1

−1

−0.5

0

0.5

1

x/c

Cp


SA (α = 2.44)

SST (α = 2.53)

Hellsten (α = 2.67)

SA-RC-QCR (α = 2.50)

Run44 Data (α = 2.50)

Run44 Data (α = 2.70)

slide 16 of 28

Task 3: Static Aero-Elastic Effect

slide 17 of 28

Task 3: Results versus angle-of-attack (α)

2.5 3 3.5 4

0.45

0.50

0.55

0.60

0.65

α

CL

CL versus Angle-of-Attack (α)

2.5 3 3.5 4

0.025

0.030

0.035

0.040

0.045

α

CD

CD versus Angle-of-Attack (α)

2.5 3 3.5 4

−0.100

−0.080

−0.060

−0.040

−0.020

0.000

α

Cm

Cm versus Angle-of-Attack (α)

SASST

HellstenSA-RC-QCR

Data: NTF Run 44

slide 18 of 28

Task 3: Isosurface of Separated Flow with differentTurbulence Models (α = 3.75)

SA SST

SA-RC-QCR Hellsten

slide 19 of 28

Task 5: Coupled Aero-Structural Simulation

slide 20 of 28

Task 5: Summary of effort

The CFD++ Software suite was used to predict aero-elasticdeformation of the test model.

I FE model of the NTF wind tunnel geometry was obtained from CRM website

I Reduced the FE model to the wing only

I Simulated ETW run 182 test conditions (Ma=0.85, Re=5 million)

Full Model

4.0E6 Degrees of Freedom

Reduced Model

1.7E6 Degrees of Freedom

slide 21 of 28

Task 5: Summary of effort

Differences from previous tasks:

I Simulation at model scale

I Mesh created with MIME

I Wall-distance-free Realizable k − ε model

I CL-driver combined with coupled aero-elastic analysis

slide 22 of 28

Task 5: Software Suite for Aero-elastic analysis

Aero-elastic calculations used four software components:

1. MIMEunstructured mesh generation

2. CFD++general unstructured finite volume-based flow solver

3. CSM++finite element-based structural solver that can be used to perform static,

transient, and eigen-mode analyses.

4. MetaFSIefficiently transfers loads and morphs the CFD++ grid to follow the CSM++

deformations.

slide 23 of 28

Task 5: Aero-elastic analysis process

CFD++ normal and

shear stresses

CSM++ forcesMorph CFD++ gridIs displacement less than

tolerance?

CSM++ displacements

Start Analysis

Exit the analysis or

proceed to next time step

Yes

No

slide 24 of 28

Task 5: Resulting deformation

Computed Deformation

Undeformed

slide 25 of 28

Task 5: Resulting deformation closely matches AE2.75CAD model

Translucent Surface is CAD

Cyan Surface is Computed Result

Computed bending is within5% of the AE2.75 value

The offset at full scale is 1.18 inch

slide 26 of 28

Closing Summary

I Task 1: ValidationI CFD++ results show excellent agreement.

I Tasks 2 and 3:I CFD++ effectively handled all of the grids from the

“unstructured NASA GeoLab.REV00” and the“Boeing Babcock Unstructured CC.REV00” grid families.

I Results were shown for a sample of the turbulence modelsavailable within CFD++.

I Task 5: Aero-elastic deformationI Demonstrated coupled aero-elastic analysis with CFD++ in

co-simulation with CSM++ and MetaFSI.I Computed deformations closely matched experiment.

slide 27 of 28

Thank You!

Name: Brian EdgeEmail: [email protected]: (818) 735-4880

Address: Metacomp Technologies, Inc.28632-B Roadside Drive, Suite 255Agoura Hills, CA 91301

slide 28 of 28

6th AIAA CFD Drag Prediction Workshop · 6th AIAA CFD Drag Prediction Workshop Summary of results...

Documents

Transcript of 6th AIAA CFD Drag Prediction Workshop · 6th AIAA CFD Drag Prediction Workshop Summary of results...