tas code results for the sixth drag prediction …...2016/06/23 · aiaa cfd drag prediction...
TRANSCRIPT
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)
Tiny Coarse Medium Fine Extra Fine
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.
Tiny Coarse Medium Fine Extra Fine
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
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: 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
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: 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
NASA_TNASA_CNASA_MNASA_FNASA_XNASA_UBoeing_TBoeing_CBoeing_MBoeing_FBoeing_XBoeing_U
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
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
NASA_TNASA_CNASA_MNASA_FNASA_XNASA_UBoeing_TBoeing_CBoeing_MBoeing_FBoeing_XBoeing_U
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
WB-NASA-qcr_k1WB-NASA-qcr_k5WB-NASA-qcr_k10
0.5
0.55
0.6
0.65
0.025 0.03 0.035 0.04 0.045
CL
CD
WB-NASA-qcr_k1WB-NASA-qcr_k5WB-NASA-qcr_k10
0.5
0.55
0.6
0.65
2.5 3 3.5 4
CL
α
WB-NASA-qcr_k1WB-NASA-qcr_k5WB-NASA-qcr_k10
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.