Comparison of Numerical Predictions and Wind Tunnel

Results for a Pitching Uninhabited Combat Air Vehicle

Russell M. Cummings, Scott A. Morton, and Stefan G. Siegel

Department of Aeronautics

United States Air Force Academy

USAF Academy, CO

Outline

Introduction Dynamic Stall/Lift UCAV Configuration Experimental Results Numerical Method Static Results Pitching Results Conclusions

Introduction

UCAV’s are playing an important role in current military tactics

Predator and Global Hawk are becoming essential elements of current operations

X-45A represents future configurations

Introduction

Some issues to explore in order to take advantage of a UCAV’s uninhabited state:

– High g maneuvering– Compact configurations– Novel control actuation– Morphing wings– MEMS-based control systems– Semi-autonomous flight– Increased use of composites– Novel propulsion systems– Dynamic stall/lift

Dynamic Stall/Lift

Utilizes rapid pitch-up and hysteresis to produce increased lift

A great deal of work has been done on airfoils and simple wings

Very little work has been done on UCAVs

From http://hodgson.pi.tu-berlin.de/~schatz/PIZIALI/osc.html

a—separation beginsb—leading-edge vortex formsc—full stalld—reattachment and return to static state

UCAV Configuration

Straight, swept leading edge with 50o sweep

Aspect ratio of 3.1 Round leading edges Blended wing & body Top/front engine inlet B-2-like wing planform Low observable shaping

Boeing 1301 UCAV

33.27

62.43 276.15 46.19 ft

78.78

33.27

32.58 ft

MAC 241.93BL 92.4

MAC/4FS 170.63

24.1950°

30°

30°

30°

27.38 sq ft

21.02 sq ft

Experimental Results

1:46.2 scale model Academy 3 ft × 3 ft open

return low-speed wind tunnel

Less than 0.05% freestream turbulence levels at all speeds

Freestream velocity of 20 m/s (65.4 ft/s)

Chord-based Reynolds number of 1.42 × 105

Static Testing

Linear lift characteristics up to 10o to 12o

Stall occurring at about 20o

Lift re-established up to 32o, where an abrupt loss of lift takes place

Effect of leading-edge vortices and vortex breakdown? Angle of Attack,

Fo

rce

Co

eff

icie

nt

0 10 20 30 40 50 60 700

0.2

0.4

0.6

0.8

1

1.2

1.4

Lift Coefficient

Drag Coefficient

Dynamic Testing

The configuration was pitched at 0.5, 1.0, and 2.0 Hz (k = 0.01, 0.02, and 0.04)

Center of rotation at the nose, 35% MAC, and the tail

The pitch cycles were completed for three ranges of angle of attack:– 0o < α < 20o

– 16o < α < 35o

– 25o < α < 45o

))cos(1()( tmt

Pitching About 35% MAC @ 2 Hz

Dynamic lift is greater than static lift during pitchup

Pitchup lift is also greater post stall

Dynamic lift is less than static lift during pitchdown

Little impact on dragAngle of Attack, (deg)

Fo

rce

Co

eff

icie

nt

0 5 10 15 20 25 30 35 40 45-0.5

-0.25

0

0.25

0.5

0.75

1

1.25

1.5

(t) = 16o to 35o (t) = 25o to 45o

(t) = 0o to 20o

Static Drag

Static Lift

Pitching About Nose @ 2 Hz

Similar results to pitching about 35% MAC

Slightly less effective at producing lift during pitchup

Essentially identical results in post-stall region

Angle of Attack, (deg)

Fo

rce

Co

eff

icie

nt

0 5 10 15 20 25 30 35 40 45-0.5

-0.25

0

0.25

0.5

0.75

1

1.25

1.5

(t) = 25o to 45o

(t) = 0o to 20o

(t) = 16o to 35o

Static Drag

Static Lift

Pitching About Tail @ 2 Hz

Drastically different than previous results

Much more lift at higher angles of attack in pitching cycle

Reduced lift at lower angles

Significant impact on dragAngle of Attack, (deg)

Fo

rce

Co

eff

icie

nt

0 5 10 15 20 25 30 35 40 45-0.5

-0.25

0

0.25

0.5

0.75

1

1.25

1.5

(t) = 25o to 40o

(t) = 0o to 20o

(t) = 16o to 35o

Static Drag

Static Lift

Numerical Method

Cobalt Navier-Stokes solver– Unstructured mesh– Finite volume formulation– Implicit– Parallelized– Second-order spatial accuracy– Second-order time accuracy with Newton sub-iterations

Run on Academy 64 processor Beowulf cluster, Origin 2000, and USAF HPC computers

Laminar flow with freestream conditions set to match Reynolds number of wind tunnel experiment

Mesh and Boundary Conditions

Three unstructured meshes:– Coarse (1.3 million cells)– Medium (2 million cells)– Fine (4 million cells)

Half-plane model No-slip on surface Symmetry plane Freestream inflow Inlet covered to match model No sting modeled

Mesh Convergence

Steady results from all three meshes yield identical forces

High angle of attack flowfields will be unsteady

Use 2 million cell mesh for following calculations

Detailed time step study to follow for unsteady flow

Number of Iterations

No

rma

lFo

rce

250 500 750 1000-2

-1

0

1

2

3

4

Fine Mesh (4 million cells)Medium Mesh (2 million cells)Coarse Mesh (1.3 million cells)

20

Steady-State Static Results

Good results in linear angle of attack range

Qualitatively similar results in post-stall region

Lift and drag are significantly over-predicted in post-stall regionAngle of Attack,

Fo

rce

Co

eff

icie

nt

0 10 20 30 40 50 60 700

0.2

0.4

0.6

0.8

1

1.2

1.4

Exp. Lift Coefficient

Exp. Drag Coefficient

CFD Lift Coefficient

CFD Drag Coefficient

Steady-State Static Results

Wide shallow vortices Vortex breakdown fairly

far back on configuration Vortical structures

behind breakdown maintain lift on aft of vehicle

Rounded leading edge creates weaker vortices that breakdown sooner

20

Time-Accurate Static Results

Flowfields in post-stall region are unsteady

Time-accurate results match experiment much more closely

Fairly good modeling of flowfield, including drag, up to α=45o

Differences in lift from α=20o to α=30o (sting, surface roughness, transition ?)

Angle of Attack, (deg)

Fo

rce

Co

eff

icie

nt

0 10 20 30 40 50 60 700

0.2

0.4

0.6

0.8

1

1.2

1.4

Exp. Lift Coefficient

Exp. Drag Coefficient

CFD Lift Coefficient

CFD Drag Coefficient

Time-Accurate Static Results

5

15

10

20

Time-Accurate Static Results

25

35

30

40

Dynamic Pitching Results

Pitchup doesn’t capture full lift increase at low α

Overprediction of lift also seen in pitchup case

Drag is fairly well modeled during pitchup

More cycles requiredAngle of Attack, (deg)

Fo

rce

Co

eff

icie

nt

0 5 10 15 20 25-0.25

0

0.25

0.5

0.75

1

1.25

Exp. Lift Coefficient UpExp. Drag Coefficient UpExp. Lift Coefficient DownExp. Drag Coefficient DownExp. Static Lift CoefficientExp. Static Drag CoefficientCFD Lift CoefficientCFD Drag Coefficient

Δt*=0.0075, nsub=5

Dynamic Pitching Results

Static Pressure Pitching Pressure

15

Conclusions

A generic UCAV configuration has been wind tunnel tested both statically and pitching

The configuration generates increased dynamic lift during pitchup maneuver

Numerical simulation helps to understand causes of wind tunnel results

– Stronger leading-edge vortex during pitchup– Leading-edge vortex persists to very high angles of attack– Vortex breakdown causes the non-linearities in lift

Collaboration between experimentalists and computationalists leads to greater understanding of aerodynamics

Questions?