Adaptive Aeroelastic Wing Shape Optimization for High … 06, 2014 · agenda 2 ①’’motivation( ②’’introduction( ③’’framework( ④’’method( ⑤((results( ⑥’’future(work(

Applied Modeling & Simula1on Seminar Series, August 6, 2014 NASA Ames Research Center

Adap%ve Aeroelas%c Wing Shape Op%miza%on for High-‐Li; Configura%ons

Mentor: Nhan T. Nguyen, PhD NASA Ames -‐ Intelligent Systems Division

Gustavo E. C. Fujiwara M.S. Aerospace Engineering -‐ University of Illinois at Urbana-‐Champaign PhD Candidate Aeronau1cs and Astronau1cs -‐ University of Washington

2014 Summer Intern at NASA Ames

Funded by NASA ARMD – Fixed Wing Project

AGENDA

2

① MOTIVATION

② INTRODUCTION

③ FRAMEWORK

④ METHOD

⑤ RESULTS

⑥ FUTURE WORK

⑦ AKNOWLEDGEMENTS

AGENDA

3

① MOTIVATION

② INTRODUCTION

③ FRAMEWORK

④ METHOD

⑤ RESULTS

⑥ FUTURE WORK

⑦ AKNOWLEDGEMENTS

4

•  Global demands for more sustainable technologies ð more energy-‐efficient airframes

•  New genera1on aircra] concepts feature: ✈ Be_er engine performance (quieter, cleaner) ✈ Higher aerodynamic efficiency ✈ Lighter materials (advanced composites)

① MOTIVATION

THE CHALLENGES •  The use of composite materials increase airframe flexibility for same load capacity

•  More flexible structures ð can change aircra] op1mal shape during flight

ð can degrade the aerodynamic efficiency

•  Reduced rigidity can also ð affect flight dynamic characteris1cs

ð reduce structural safety margins (flu_er, etc)

•  Boeing 787 wing at 0-‐g load

5

① MOTIVATION

Example of flexibility of current composite aircra]

Photo credits: YK , Kenneth Low (Airliners.net)

6

① MOTIVATION

12-‐] wing 1p deflec1on à ~12% of wing semispan

Example of flexibility of current composite aircra] •  Boeing 787 wing at 1-‐g load

Photo credits: YK , Kenneth Low (Airliners.net)

AGENDA

7

① MOTIVATION

③ FRAMEWORK

④ METHOD

⑤ RESULTS

⑥ FUTURE WORK

⑦ AKNOWLEDGEMENTS

② INTRODUCTION

8

•  A current research effort to address the aerodynamic efficiency/flight control issues of flexible wings:

ð Variable Camber Con1nuous Trailing-‐Edge Flap (VCCTEF)

② INTRODUCTION

VCCTEF CONCEPT IDEA ð  Tailor the wing’s spanwise load distribu1on

ð  Aeroelas1cally re-‐adapt it back t o t he op1ma l s hape , a s condi1ons change during flight

Courtesy of The Boeing Co.

VCCTEF Anima1on

Courtesy of Michael A]osmis and David Rodriguez

② INTRODUCTION

VCCTEF Anima1on

② INTRODUCTION

CRUISE Configura1on HIGH-‐LIFT Configura1on

Courtesy of Michael A]osmis and David Rodriguez

•  Cruise Configura1on Model

Computa%onal Work in Drag Reduc%on Op%miza%on ¥ -‐  Sonia Lebofsky, Eric Ting, Nhan Nguyen (Vorview) -‐  Michael A]osmis, David Rodriguez (Cart3D)

Experimental Work (Wind Tunnel Tests) * -‐  University of Washington (UW): Prof. Eli Livne, Nathan Precup -‐  Boeing: James Umes, Sr., Chester Nelson -‐  Model based on the Generic Transport Model (GTM) – B757 devised -‐  Designed to match flexibility of about 10% wing 1p deflec1on

11

② INTRODUCTION

¥Lebofsky, S., Ting, E., Nguyen, N., “Aeroelas1c Modeling and Drag Op1miza1on of Aircra] Wing with Variable Camber Con1nuous Trailing Edge Flap”, AIAA Avia1on 2014, 32nd AIAA Applied Aerodynamics Conference

*Nguyen, N., Precup, N., Umes, J., Nelson, C., Lebofsky, S., Ting, E., Livne, E., “Experimental Inves1ga1on of a Flexible Wing with a Variable Camber Con1nuous Trailing Edge Flap Design”, AIAA Avia1on 2014, 32nd AIAA Applied Aerodynamics Conference

Courtesy of University of Washington

•  High-‐Li] Configura1on Model

12

② INTRODUCTION

Computa%onal Op%miza%on Framework -‐  Currently under development at NASA Ames -‐  Michael A]osmis, David Rodriguez – OVERFLOW -‐  Challenges: Grid deforma1on tool (surface/volume meshes) Computa1onal cost ro run RANS solvers

Experimental Work (Wind Tunnel Tests) -‐ UW: Prof. Eli Livne, Nathan Precup -‐  Currently being tested -‐  Includes a variable camber Krueger (VCK) Flap -‐  Single element slo_ed inboard flap -‐  3-‐chordwise-‐segment VCCTEF elsewhere


•  High-‐Li] Configura1on Model

13

② INTRODUCTION


•  Objec1ve

14

Develop an aeroelas1c framework ð quick turnaround mul1disciplinary op1miza1on of a flexible wing with variable camber con1nuous trailing edge flap (VCCTEF) for high-‐li] configura1ons (takeoff and landing)

② INTRODUCTION

§ Develop a low-‐fidelity 3D aerodynamic code capable of handling stall characteris1cs (Cl, Cm) §  Couple structural code to calculate sta1c wing deflec1on §  Allow fast computa1on for op1miza1on purposes

•  Requirements

• Poten1al Modeling Complica1ons

Ø Flap deflec1ons increase not only the loads but also the nose-‐down pitching moment Ø  Nose-‐down moment loads tend to twist the wing downwards, thereby decreasing angle of a_ack and li] (similar to control reversal)

Ø  Op1mizing flexible wings for high-‐li] can be counterintui1ve

AGENDA

15

① MOTIVATION

② INTRODUCTION

④ METHOD

⑤ RESULTS

⑥ FUTURE WORK

⑦ AKNOWLEDGEMENTS

③ FRAMEWORK

16

③ FRAMEWORK Mul%disciplinary Op%miza%on Framework Strategy: •  Coupled aerodynamics/structures •  Sta1c aeroelas1c calcula1ons •  Op1mizes flap deflec1on schedule for:

Ø  CLmax, or Ø  (CL/CD)max

OPTIMIZER

GEOMETRY GENERATOR

AERODYNAMICS (Nonlinear 3D code) STRUCTURES

CONVERGENCE ?

CLmax (CL/CD)max

YES

NO

TASK 1

TASK 2 OPTIMIZER

GEOMETRY GENERATOR AERODYNAMICS STRUCTURES

CONVERGENCE ?

CLmax

(CL/CD)max

YES

NO

17

Low-‐fidelity aerodynamic code structure:

AERODYNAMICS 2D Viscous Data

MSES

2D RANS

XFOIL

Wind Tunnel

3D Code VLM

Panel Method

Weissinger

③ FRAMEWORK

AGENDA

18

① MOTIVATION

② INTRODUCTION

③ FRAMEWORK

⑤ RESULTS

⑥ FUTURE WORK

⑦ AKNOWLEDGEMENTS

④ METHOD

19

•  Baseline Airfoil example: ‘bacj.dat’ (supercri1cal) publicly available at UIUC – Airfoil Database

④ METHOD

2D solver to obtain sec%onal viscous data (XFOIL)

TASK 1 2D Viscous Data

MSES

2D RANS

Wind Tunnel

XFOIL

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

x

z

VCCTEF

Supercritical Airfoil δ1 = 5°, δ2 = 5°, δ3 = 5°

δ1 =10°, δ2 =10°, δ3 =10°

δ1 =15°, δ2 =15°, δ3 =15°

δ1 =20°, δ2 =20°, δ3 =20°

•  Build a 2D Viscous Aerodynamic Data Bank

20

α = -‐5˚ α = -‐4˚ α = -‐3˚ ...

Re = 100k Cl, Cd, Cm

Re = 200k ...

Re = 300k ...

... ...

α

Re

④ METHOD

Offline Calcula1ons -‐ XFOIL •  Re = 100k -‐ 1.1M (UWAL model, chords) •  AoA = -‐5˚ – 25˚ (Cl, Cd, Cm)

TASK 1 2D Viscous Data XFOIL

21

TASK 2 3D Code VLM

Panel Method

Weissinger

Vortex La^ce Method code chosen •  AVL – Athena Vortex La�ce* •  Publicly available (open source) •  Validated

*Mark Drela, Harold Youngreen – MIT Based on the classic work of Lamar (NASA codes), E. Lan and L. Miranda (VORLAX)

④ METHOD

22

Nonlinear Vortex La^ce Method •  Lower computa1onal cost and lower fidelity compared to a Navier-‐Stokes solver •  Itera1vely changes wing’s local incidence/camber to match Cl/Cm from 2D viscous data

④ METHOD TASK 2 3D Code VLM

Decambering Method Inspira%on •  XFOIL – Boundary Layer Thickness Growth

23

α = 4˚ α = 0˚

α = 12˚

α = 18˚


24 2003 – Funded by NASA Langley – North Carolina State University *Mukherjee, R., Gopalarathnam, A., Kim. S., “AN ITERATIVE DECAMBERING APPROACH FOR POST-‐STALL PREDICTION OF WING CHARACTERISTICS USING KNOWN SECTION DATA”, AIAA 2003, 41st AIAA Aerospace Sciences Mee1ng


DETAILS: THE DECAMBERING APPROACH

2)  Get Cl, Cm and downwash angles (ω) along span

6) Change incidence δ1 and camber δ2 in each sta1on and rerun STOP CRITERION

| Cl (sec1on) -‐ Cl 2D Visc (αeff) | < 0.001 | Cm(sec1on) -‐ Cm 2D Visc(αeff) | < 0.001

3) For each sta1on (strip) calculate local αeff

4) Calculate gradients/sensi1vi1es (Jacobian): -‐  ∂ Cl in each sta1on i due to incidence (δ1) perturba1on p at sta1on j -‐  ∂ Cm in each sta1on i due to incidence (δ1) perturba1on p at sta1on j -‐  ∂ Cl in each sta1on i due to camber (δ2) perturba1on p at sta1on j -‐  ∂ Cm in each sta1on i due to camber (δ2) perturba1on p at sta1on j

1) Input Aircra] Geometry and Run AVL (Linear VLM code)

Incidence

δ1 δ2

Camber

5) Mul1variable Newton-‐Raphson Itera1on to compute δ1 and δ2 ~ 2π

α

Cl

2D viscous data

ΔCl Residual

α effec1ve

Linear VLM sec1on

α

Cm

α effec1ve

ΔCm Residual

25 *Mukherjee, R., Gopalarathnam, A., Kim. S., “AN ITERATIVE DECAMBERING APPROACH FOR POST-‐STALL PREDICTION OF WING CHARACTERISTICS USING KNOWN SECTION DATA”, AIAA 2003, 41st AIAA Aerospace Sciences Mee1ng


DETAILS: THE DECAMBERING APPROACH

Jacobian Matrix (Gradients/Sensi1vi1es)

Residuals (ERROR)

Incidence/ Camber

Correc1ons

Aircra; Geometry Input

Run VLM Get Cl, Cm, ω along span

At each sta%on:

Cl2D VLM (i) = Cl2D visc (αeff) ? Cm2D VLM (i) = Cm2D visc (αeff) ?

YES

NO Change Incidence/Camber

Mul%variable Newton-‐Raphson

END

START

αeff

~ 2π

α

Cl

α effec1ve

Linear VLM sec1on

2D viscous data

α

Cm

α effec1ve

26

For Improved Efficiency…

•  All decambering variables can be dealt with using the RHS only •  Avoid matrix inversion every itera1on

[AIC][Γ] = [V∞.n+Vind⊥]

④ METHOD

Incidence

δ1 δ2

Camber

TASK 2 3D Code VLM

AGENDA

27

① MOTIVATION

② INTRODUCTION

③ FRAMEWORK

④ METHOD

⑥ FUTURE WORK

⑦ AKNOWLEDGEMENTS

⑤ RESULTS

28

⑤ RESULTS TASK 1 2D Viscous Data XFOIL

Matlab automated tool to run XFOIL in Batch

29

TASK 1 2D Viscous Data XFOIL

Effect of Boundary Layer Transi1on – Ncri1cal Method

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

-0.05

0

0.05

x

z

VCCTEF

Supercritical Airfoil δ1 =5°, δ2 =5°, δ3 =5°,

-5 0 5 10 15 20

0.6

0.8

1

1.2

1.4

1.6

1.8

α [°]

Cl

0.6 0.8 1 1.2 1.4 1.6

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Cl

Cd

-5 0 5 10 15 20 25-0.28

-0.26

-0.24

-0.22

-0.2

-0.18

-0.16

-0.14

-0.12

-0.1

α [°]

Cm

Re=0.662M - Ncrit=0.5

Re=0.662M - Ncrit=9

Re=0.662M - Ncrit=12

⑤ RESULTS

30

⑤ RESULTS TASK 2 3D Code VLM

Matlab automated tool to run AVL

0 2 4 6 8 10 12 14 16 18 200

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

α [deg]

CL

Wing-BodyWing-Body-Botton-Top-WallsWing-Body-All-Walls

Streamwise Cuts Conversion

31 2 3 4 5 6 7 8

1

1.5

2

2.5

3

3.5

4

4.5

5

5.5

6UWAL - VCCTEF Model

X [ft]

Y [ft

]


Half Wing-‐Body Symmetry Condi1on

32


Half Wing-‐Body Symmetry Condi1on + Top/Bo_om Walls Only

33


Half Wing-‐Body Symmetry Condi1on + All Walls

34


0 2 4 6 8 10 12 14 16 18 200

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

α [deg]

CL

Wing-BodyWing-Body-Botton-Top-WallsWing-Body-All-Walls

35

Agrees with UWAL Rigid Wing Data from Nhan’s Paper (see next slide)

CLα = 4.33-‐4.53

CLα = 4.47-‐4.68

CLα = 4.64 – 4.83

⑤ RESULTS

LINEAR AVL TASK 2 3D Code VLM

36

UWAL RIGID -‐ Paper CLα = 4.49

•  Reconstructed data for a rigid wing based on flexible wing tunnel data •  Data corrected for wind tunnel walls

⑤ RESULTS

*Nguyen, N., Precup, N., Umes, J., Nelson, C., Lebofsky, S., Ting, E., Livne, E., “Experimental Inves1ga1on of a Flexible Wing with a Variable Camber Con1nuous Trailing Edge Flap Design”, AIAA Avia1on 2014, 32nd AIAA Applied Aerodynamics Conference

UWAL RIGID -‐ Paper CLα = 4.31

TASK 2 3D Code VLM

NON-‐LINEAR AVL

37

⑤ RESULTS

•  Code Modifica1ons to Include: – Maximum 2D Cl and Non-‐lineari1es in li] (near stall)

– Viscous drag (profile + skin fric1on)

TASK 2 3D Code VLM

Valida1on with Experimental Data

38

⑤ RESULTS

*Mukherjee, R., Gopalarathnam, A., Kim. S., “AN ITERATIVE DECAMBERING APPROACH FOR POST-‐STALL PREDICTION OF WING CHARACTERISTICS USING KNOWN SECTION DATA”, AIAA 2003, 41st AIAA Aerospace Sciences Mee1ng

TASK 2 3D Code VLM

39

Aircra; Geometry Input

Run VLM Get Cl, Cm, ω along span

At each sta%on:

Cl2D VLM (i) = Cl2D visc (αeff) ? Cm2D VLM (i) = Cm2D visc (αeff) ?

YES

NO Change Incidence/Camber

Mul%variable Newton-‐Raphson

END

START

αeff



40

-10 0 10 20 30 40 50 60-0.5

0

0.5

1

1.5

2

2.5

α [°]

CL

2D Viscous Data3D Exp. - AR=12 - λ=1.0AVL (3D linear)AVL-Modified (3D nonlinear)



41 0 0.5 1 1.5 2 2.5-0.12

-0.1

-0.08

-0.06

-0.04

-0.02

0

0.02

0.04

2y / b

Cm

Lin-α=0Nonlin-α=0Lin-α=5Nonlin-α=5Lin-α=10Nonlin-α=10Lin-α=15Nonlin-α=15Lin-α=19Nonlin-α=19Lin-α=27Nonlin-α=27

0 0.5 1 1.5 2 2.50

0.5

1

1.5

2

2.5

3

2y / b

Cl

Cl Max 2DLin-α=0Nonlin-α=0Lin-α=5Nonlin-α=5Lin-α=10Nonlin-α=10Lin-α=15Nonlin-α=15Lin-α=19Nonlin-α=19Lin-α=27Nonlin-α=27


42

0 1 2 3 4 5 60

1

2

3

c.C

l

0 1 2 3 4 5 6

0

0.5

1

b/2

Cl

α =0α =2.5α =5α =7.5α =10α =12.5α =15α =17.5α =20

2D viscous sec1onal stall limit

•  UWAL Model Geometry with NACA0012 sec1onal data •  The developed Nonlinear Vortex La�ce code results:

•  CLα: slope agreement with the linear prediciton for the finite wing •  Cl x span: good agreement with theory. As angle of a_ack increases, the sta1ons increase local li] coefficient un1l they reach the maximum 2D Cl from the viscous data, yielding the 3D stall behavior and propaga1on characteris1cs of the CLxα curve

0 5 10 15 20 25

0

0.2

0.4

0.6

0.8

1

1.2

AoA [deg]

CL

NACA0012 - 2D Viscous Data - XFOILAVL - 3D linearAVL-Modified - 3D nonlinear


43

•  Established a mul1disciplinary op1miza1on framework strategy

•  Developed a low-‐fidelity nonlinear 3D aerodynamic tool Ø Using known sec1onal data Ø  Capable of capturing CLmax and handling stall Ø Modified an exis1ng Vortex La�ce Method code (AVL)

•  Results showed successful predic1on of 3D nonlinear aerodynamic characteris1cs, while maintaining low computa1onal cost

This summer so far …

⑤ RESULTS

AGENDA

44

① MOTIVATION

② INTRODUCTION

③ FRAMEWORK

④ METHOD

⑤ RESULTS

⑦ AKNOWLEDGEMENTS

⑥ FUTURE WORK

45

⑥ FUTURE WORK

Program the structural code

FEM

STRUCTURES Galerkin Method

46

AERODYNAMICS 2D Viscous Data 2D RANS (OVERFLOW)

XFOIL

3D Code VLM (AVL)

Panel (Panair)

VLM (Vorview)

Modify the aerodynamic tool

⑥ FUTURE WORK

47

Integrate aero/structural code with the op%mizer

OPTIMIZER

GEOMETRY GENERATOR

AERODYNAMICS (Nonlinear 3D code) STRUCTURES

⑥ FUTURE WORK

48

⑦ AKNOWLEDGEMENTS

• NASA ARM – Fixed Wing Project • Adap1ve Aeroelas1c Shape Control (AASC) under Aerodynamic Efficiency sub-‐project

• Dr. Ce1n Kiris • NASA AMS Division: Michael A]osmis, David Rodriguez, others

• Dr. Nhan Nguyen • NASA Intelligent Systems Division: Ezra Tal, Sonia Lebofsky, Eric Ting, others

• Dr. Michael Bragg, Dr. Eli Livne at the University of Washington

THANK YOU

BACKUP SLIDES

49

Nonlinear Weissinger •  Swept Wing 2D and 3D CFD data

50

The results are from CFD computed at a Re = 3.0 M for NACA 4415 airfoil Wing: AR = 12, Taper =1 (constant-‐chord wing)

Source: Paul, R. C., Gopalarathnam, A., "Itera1on schemes for rapid post-‐stall aerodynamic predic1on of wings using a decambering approach", INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN FLUIDS, Int. J. Numer. Meth. Fluids (2014), Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/fld.3931

CFD Data

51 Source: Paul, R. C., Gopalarathnam, A., "Itera1on schemes for rapid post-‐stall aerodynamic predic1on of wings using a decambering approach", INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN FLUIDS, Int. J. Numer. Meth. Fluids (2014), Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/fld.3931

52 Source: “Itera1on schemes for rapid post-‐stall aerodynamic predic1on of wings using a decambering approach” -‐ 2014 -‐ Interna1onal Journal for Numerical Methods in Fluids -‐ Wiley Online Library

Documents

Adaptive Aeroelastic Wing Shape Optimization for High … 06, 2014 · agenda 2 ①’’motivation( ②’’introduction( ③’’framework( ④’’method( ⑤((results( ⑥’’future(work(