tas code results for the sixth drag prediction …...2016/06/23 · aiaa cfd drag prediction...

TAS Code Results for the Sixth Drag Prediction Workshop

Y. Ito, M. Murayama & K. Yamamoto (JAXA) K. Tanaka (Ryoyu Systems Co., Ltd.)

6th AIAA CFD Drag Prediction Workshop, Washington DC June 16, 2016

Outline

Objective Flow Solver: TAS Code Case 1: NACA0012 Grids Cases 2 & 3: CRM Grids

Computational results Comments on grids

Concluding Remarks

Objective

Evaluate our unstructured grid solver, TAS Code, with committee-provided grids.

Case 1: Verification Study (NACA0012 Airfoil) Grid Family II on TMR

Case 2: CRM Nacelle-Pylon Drag Increment unstructured_NASA_GeoLab.REV00 Boeing_Babcock_Unstructured_CC.REV00 (as reference)

Case 3: CRM WB Static Aero-Elastic Effect unstructured_NASA_GeoLab.REV00

No optional test cases

Flow Solver: TAS Code

TAS (Tohoku Univ. Aerodynamic Simulation) code Originally developed by Nakahashi et al.

Venkatakrishnan’s limiter (1995. JCP, 118, 120-130.)

Constant K = 10, 5, 1 Less Limiter effect More

5 is recommended in the paper. TAS

Grid type Unstructured hybrid grids

Discretization Cell-vertex finite volume

Convection flux HLLEW 2nd-order with Venkatakrishnan’s limiter

Time integration LU-Symmetric Gauss-Seidel

Turbulence model SA-noft2 (Case 1)

SA-noft2-R(Crot=1)-QCR2000 (Cases 2 & 3) Yamamoto, et al. AIAA 2012-2895

Case 1: NACA 0012 Grids

Grid Family II http://turbmodels.larc.nasa.gov/naca0012numerics_grids.html

7 structured grids: 1 (Finest) through 7 (Coarsest) Converted to unstructured one-layer hexahedral grids

Flow condition M∞ = 0.15, Re = 6M, AoA = 10° Farfield: Dirichlet BC (not Riemann BC)

Case 1: Cfx

Cfx K=1 K=5 K=10

Cfx was sensitive to K in Venkatakrishnan’s limiter. Cp was not.

Case 1 Grid Convergence Study

Compared with results from other solvers, TAS code predicted similar converged coefficients. K in Venkatakrishnan’s limiter created variations when the grids were coarse.

Case 1 Grid Convergence Study: CL & CD

Compared with no limiter case, CL and CD predicted with K = 5 converged similarly.

CL

CD CDf CDp

Cases 2 & 3: CRM Grids

Two unstructured grid families were used for WB & WBNP configurations

unstructured_NASA_GeoLab.REV00 Except WBNP Ultra-Fine due to limitation in grid partitioning

Boeing_Babcock_Unstructured_CC.REV00 (as reference)

# nodes (million)

NASA GeoLab Boeing Babcock WB WBNP WB WBNP

Tiny 20 28 8.0

Not used

Coarse 30 41 10 Medium 44 61 13 Fine 66 91 17 eXtra fine 101 138 22 Ultra fine 151 209 28

Correction – Case 2A CMS

The method to calculate CMS in our submitted data (sectional lift and moment) was incorrect.

Medium Grid Case

NASA GeoLab WB Grid Family (1)

Cross-sections around LE through the kink Dense surface grids Relatively a small # of prismatic layers Dents on medium grids for several angles of attack

Tiny Coarse Medium Fine Extra Fine

Ultra Fine

2.50° 3.00° 3.25° 3.50° 3.75° 4.00°

2.75°

NASA GeoLab WB Grid Family (2)


Ultra Fine

2.50° 3.00° 3.25° 3.50° 3.75° 4.00°

2.75°

Cross-sections around TE through the kink Small # of extreme slivers close to TE in several grids were fixed to properly run TAS code.

Boeing Babcock Grid Family

Cross-sections around LE & TE through the kink A large # of prismatic layers The # of prismatic layers is almost constant

Only grid convergence study with this Boeing WB grid family.


Ultra Fine

Case 2 Grid Convergence Study: Cp

K in Venkatakrishnan’s limiter affected the shock location especially when the grids were coarse. K = 5 appeared to provide the most consistent result.

K=10 K=5 K=1

Tiny Coarse Medium

Fine Extra Fine

Ultra Fine

η = 0.8456

0.0255

0.0260

0.0265

0.0270

0.0275

0.0280

0.0285

0.E+00 5.E-06 1.E-05 2.E-05 2.E-05 3.E-05 3.E-05

CD

N-2/3

WB-NASA-qcr_k10WB-NASA-qcr_k5WB-NASA-qcr_k1WB-Boeing-qcr_k5WBNP-NASA-qcr_k10WBNP-NASA-qcr_k5WBNP-NASA-qcr_k1

Case 2 Grid Convergence Study: CD

K in Venkatakrishnan’s limiter affected CD by 1-2 counts. The two grid families provided different results for the WB case.

WBNP

WB Difference in grids

Difference in “K”

Difference in “K”

0.014

0.0142

0.0144

0.0146

0.0148

0.015

0.0152

0.0154

0.0156

0.E+00 5.E-06 1.E-05 2.E-05 2.E-05 3.E-05 3.E-05

CDp

N-2/3


Case 2 Grid Convergence Study: CDp

Boeing Ultra-Fine grid showed a different trend. Boeing Extra-Fine & NASA Tiny grids were similar in size, but produced a difference in drag count.

Due to relatively coarse tetrahedra around CRM in the Boeing grid family.

WBNP

WB 1.6 counts

Boeing Grids on Symmetry Plane

To check the size of tetrahedra, surface grids on the symmetry plane were visualized. Grid density was controlled on the CRM and the farfield boundary, but was not well controlled in the middle.

The Ultra-Fine grid still had relatively coarse tetrahedra.

Tiny Coarse Medium Fine Extra Fine Ultra Fine

0.0110

0.0115

0.0120

0.0125

0.0130

0.0135

0.E+00 5.E-06 1.E-05 2.E-05 2.E-05 3.E-05 3.E-05

CDf

N-2/3


Case 2 Grid Convergence Study: CDf

CDf estimated by TAS code with SA was usually not sensitive to grid density, but a different trend was observed with the NASA grids.

Due to unexpected growth rates for the near-field grids.

WBNP

WB

NASA Grids

NASA Grids

Boeing Grids

0.4

0.45

0.5

0.55

0.6

0.65

0.7

0.75

0.8

0 0.002 0.004 0.006 0.008

Ux

Wall Distance/MAC

NASA_TNASA_CNASA_MNASA_FNASA_XNASA_UBoeing_TBoeing_CBoeing_MBoeing_FBoeing_XBoeing_U

1.0

1.1

1.2

1.3

1.4

1.5

1.6

1.7

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07

Gro

wth

Rat

e

Wall Distance/MAC


Prism Growth Rates & Ux

To check boundary layer profile in the near-field grids, nodes with the same coordinates on the junction of the fuselage and the symmetry plane were selected.

According to the Gridding Guidelines, “Growth Rates < 1.2X Normal to Viscous Walls” The large growth rates of the NASA grids made the friction drag by TAS code grid dependent.

NASA: Large growth rates

Boeing: constant growth rates

BL thickness in the location

0.5

0.52

0.54

0.56

0.58

0.6

0.62

0.64

0.66

0 0.0005 0.001 0.0015 0.002 0.0025

Ux

Wall Distance/MAC


1.0

1.1

1.2

1.3

1.4

1.5

1.6

1.7

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07

Gro

wth

Rat

e

Wall Distance/MAC


Prism Growth Rates & Ux

To check boundary layer profile in the near-field grids, nodes with the same coordinates on the junction of the fuselage and the symmetry plane were selected.

According to the Gridding Guidelines, “Growth Rates < 1.2X Normal to Viscous Walls” The large growth rates of the NASA grids made the friction drag by TAS code grid dependent.

NASA: Large growth rates

Boeing: constant growth rates

BL thickness in the location

Case 3: α Sweep

To perform α sweep, there is a set of grids that do not have the same element connectivity.

Should we restart a CFD simulation based on a solution at a lower α (= 2.75°) even for this case? Can we use an impulsive start for each grid?

For NASA WB grids, the two approaches gave almost the same result.

Impulsive starts were selected for other cases.

-0.11

-0.1

-0.09

2.5 3 3.5 4

CM

α

WB-NASA-qcr_k1WB-NASA-qcr_k5WB-NASA-qcr_k10

0.025

0.03

0.035

0.04

0.045

2.5 3 3.5 4

CD

α


0.5

0.55

0.6

0.65

0.025 0.03 0.035 0.04 0.045

CL

CD


0.5

0.55

0.6

0.65

2.5 3 3.5 4

CL

α


Case 3: Result of α Sweep

Surface Stream Lines

No significant side-of-body separation found on the wing upper surface.

2.50° 3.00° 3.25°

3.50° 3.75° 4.00°

2.75°

Concluding Remarks

In Case 1, K in Venkatakrishnan’s limiter was evaluated by using three constants, 10, 5 and 1.

K = 5 recommended in the original paper was the best. In Cases 2 & 3, grids had a great impact on TAS code in terms of prism growth rates & farfield grid density.

We will generate our own grids and run TAS code to see if a difference is found in the grid convergence study.

Impulsive starts and restarts based on a solution at α = 2.75° gave almost no difference.

tas code results for the sixth drag prediction …...2016/06/23 · aiaa cfd drag prediction...

Documents