Hybrid LES Approach for Practical Turbomachinery Flows
Transcript of Hybrid LES Approach for Practical Turbomachinery Flows
![Page 1: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/1.jpg)
Hybrid LES Approach for Practical Turbomachinery Flows
Paul Tucker, Simon Eastwood, Christian KlostermeierHao Xia, Prasun Ray, James Tyacke and William Dawes
Whittle Laboratory, University of Cambridge
Summary
• LES computational cost discussed
• LES hierarchy is proposed → LES model last !!
• Practical hybrid LES approach is proposed
• Applied to predict canonical flows: HDT, free shear flow, wall shear flows (+TS), ribbed passage; convex surface impingement
• Consider turbine and compressor endwall flows; fan blade section; jets; cutback trailing edge; idealized high pressure compressor drum cavity
• Encouraging results. Challenges remain for complex BL physics …..
• Need for best practices, better validation data discussed
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 2: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/2.jpg)
LES = Resolve all large eddies10% modelling
RANS = Resolve time average of flow,all modelling
x
y
Δ
Resolved/solved forModelled < 2Δ
l
What is LES, MILES, NLES …?
•Practical LES (DeBonis)
•Rigorous LES
KEY LES PROBLEM
Hinze (1975)
•Resolving streaks Δz+ ≈ 100
•Trent 1000 fan at cruise 107
•LES Cost α Re2.5*
•Hybrid LES-RANS Cost α Re0.5
y+=90
*Piomelli, AIAA-2008-396
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 3: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/3.jpg)
LES Resolution Requirements
Adapted from Leschziner (2009), Piomelli and Balaras (2002)
δ+
y+ ≈ 100
Grid Requirements
Re
Hybrid
LES
y+<≈100
δ+>y+>≈100
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 4: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/4.jpg)
NUMERICS/SGS SENSITIVITY
Liu, Y., Tucker, P. and Kerr, R. Linear and nonlinear model large-eddy simulations of a plane jet. Computers and Fluids, 37(4): 439-449, 2008
2-6 3-7th order
Homogenous decaying turbulence
LES model Numerical parameters
ScatterScatter
1/l
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 5: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/5.jpg)
Shear Layers – significant numericalinfluence
• Average predicted shear stress error for five LES models, five numerical schemes and 10% Ti delta at inflow
4 % -Inflow (Ti = 10%)
14%19 %Numerical scheme
12 %4 %LES model
Ave. errorStd. Dev.
U
For NLES N/Re0.4 - 3.5 time higher i.e. better resolved
Shear Flow - Significant Numerical Influence
Chow and Moin JCP (2003) – numerical error
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 6: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/6.jpg)
TS Wave in Shear Flow - Grid-Solver Compatibility/Sensitivity
θ
OMIT LES MODEL TO LESSEN DISSIPATION?
LES Model Validity? Omit?
• Smagorinsky model erroneously drains energy from all scales (AIAA ban)
• It provides no backscatter
• Eddy viscosity alters the effective Reynolds number and so alters the fine scales (non-linear LES models OK?)
• Existence of Kolmogorov -5/3 region in turbomachinery? If not new LES model needed
0.0 35.5 71.0 106.5 142.0 177.6 213.1 248.6 284.1
-5/3 law
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 7: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/7.jpg)
‘Law of the wall’ for under-resolvedLES grids
U+
log y+
Linear LESNonlinear LES
Hybrid NLES-RANSx3 - blendings
Δz+ ≈ 100
Industrial Turbomachinery LESHierarchy
WALL MODELLING: separation; transition
PROBLEM DEFINITION:Boundary conditions; real
geometry influences
GRID-SOLVER COMPATABILITY
SGS MODEL: LES, MILES
BEST PRACTICES
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 8: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/8.jpg)
ENGINEERING MODELLING APPROACH – RANS-NLES Blending
| ∇φ|=1+ f(φ)∇2φ + g(φ)
f(φ) = ε0φ, g(φ)= ε1(φ /L)n
NLES
RANS
RANS
NLES
Mixed zone
y
‘d’
l = κ d, d → 0
0 5 10 15 20x/h
0
50
100
150
200
250
300
Nu
Ribbed channel (Re = 14,000)
LES removes scatter
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 9: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/9.jpg)
Impinging Jet Flow – Re = 23000
3 x 106 cells
Compressor Blade – End Wall
Interface
Re ≈0.25 million, N = 5 million
Ptot loss50% after blade
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 10: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/10.jpg)
Streamlines at Endwall - Compressor
ExperimentNLES
Hybrid NLES-RANS
Averaged Total Pressure Loss Coefficient
Interface
Re ≈ 0.6 million, N > 5 million
10% DS
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 11: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/11.jpg)
Streamlines at Endwall - Turbine
Experiment NLES
Hybrid NLES-RANS Excessive separation
Coaxial Nozzle
Re = 300,000, 5<N< 50 (million)
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 12: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/12.jpg)
HJ Equation Distance Function
Vorticity Contours
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 13: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/13.jpg)
Statistics
Time averagedstreamwise velocity.
Normal stress, u’u’
Shear stress, u’v’
Grid Sensitivity
6M
12M
50M
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 14: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/14.jpg)
SENSITIVITY TO REAL INFLOW/ GEOMETRY
Geometry andBlocking Structure
Axialvelocity
Vorticity Contours
Turbulence profiles for pylon and nozzle with internal geometry
Pylon – circumferentially averaged Internal geometry – on blade trailing edge
Baseline geometry
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 15: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/15.jpg)
Pylon Geometry
Instantaneous Streamwise Velocity Time Averaged Streamwise Velocity
Deflection
Chevron Nozzle
12x106 < N < 60x106, Re = 1x106
θ
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 16: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/16.jpg)
‘Numerical Schlieren’
Flow Visualisation
Vorticity iso-surface coloured by streamwise
velocity magnitude. Chevron K-H rollers
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 17: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/17.jpg)
Statistics
Tip
Root
9/30/2008 CAA Overview
RANS-LES + FWH
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 18: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/18.jpg)
Chevron Far-field Sound (SPL)
5o 17o
Θ, deg
ΔSP
L,d
B
20 40 60 80 100 120-6
-4
-2
0
2
4
6
Far-field sound PSD of SMC006 at 90 degree.
blue dash: data of NASA Glenn.
St
PS
D,d
B/H
z
10-1 100 101
20
40
60
80
St ~ 2
Incremental OASPL from SMC001 to SMC006 (left),
We get the delta!!
Power spectral density
Stmax < a∞ /4Δ
Assume 4 nodes can resolve a wave accurately
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 19: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/19.jpg)
Fan Blade Section (Re = 3 x 106)
Time averaged MInstantaneous vorticity magnitude
Blade surface pressure spectra
M=1 isosurface
Goody (2004) AIAA J.
N = 8 million
Cut back trailing edge
Hex rich mesh
Instantaneous flow
‘Martini & Schultz’ (2004) type case, Kacker and Whitelaw (1969)
Instantaneous flow – vacillates– periodicity issues
Temperature
x
x
x
x
x
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 20: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/20.jpg)
High pressure compressor drum
PROBLEM AEROSPACE FLOWS
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 21: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/21.jpg)
Open rotor engines
Y0 PLANE
Total Pressure Contours
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany
![Page 22: Hybrid LES Approach for Practical Turbomachinery Flows](https://reader034.fdocuments.us/reader034/viewer/2022042609/6263371563725b1fd202d86d/html5/thumbnails/22.jpg)
Experimental data & Best Practices
• To correctly validate/exploit detailed comparisons with u’u’ and E(k) data is desirable
• Lacking in many turbomachinery studies
• More importantly LES can need inflow (U, l & Ti) and careful definition of general BCs
• Frequently missing from experimental studies
• To advance LES, both numerical practitioners and experimentalists need to move forward together
• Formal framework for recording simulations and what to record
Conclusions
• High cost of testing and low fidelity of RANS makes the development of LES attractive
• Considered field of ‘practical LES’, where dissipative RANS based solvers are frequently used
• LES model omission is an attractive option with near wall RANS model – hybrid RANS-NLES
• Combined with suitable grid topologies (hexahedral meshes seem highly preferable) method gives useful results
• For industrial LES hierarchy is: problem definition, wall modeling and grid-solver compatibility with, last of all, the much debated LES model
• Better defined and more detailed validation data and best practice
• Caution must be exercised – still needs user expertise and best practices should be developed
Second Symposium "Simulation of Wing and Nacelle Stall", June 22nd - 23rd, Braunschweig, Germany