Reverse Engineering and Aerodynamic Analysis of a Flying Wing UAV

Navabalachandran s/o Jayabalan1 , Low Jun Horng2, G. Leng3

Aeronautical Engineering Group

Department of Mechanical Engineering

National University of Singapore

Abstract This study is centered primarily about two main objectives with the first one being the

complete structural reconstruction and aerodynamic data generation of a pre-existent

Unmanned Flying Wing Air Vehicle with inadequate contractors’ aerodynamic and

stability data, construction specifications and knowledge of materials used. With this

accomplished we then focus on further aerodynamic analysis and scientific modification

to the original design and power plant to enable the platform to carry additional payloads

of an autonomous navigation system and a real time operating camera to meet various

practical mission requirements. The paper describes in detail the systematic reverse

engineering procedure adopted to analyze and synthesize the entire model. Some of the

techniques adopted are 3D Laser profile scanning of the reflex airfoil and fuselage,

material research and selection and cost effective reconstruction of the non-conventional

airfoil. We also present weight and balance matching techniques , the usage of

commercially available CFD programmes to generate aero coefficients and forces and

estimate the aircraft’s aerodynamic center, results of the extensive verification flight tests

conducted and performance matching procedures used in general of the reverse

engineered craft to the existing UAV flying wing model.

1

Nomenclature CL Coefficient of Lift d Horizontal distance CD Coefficient of Drag h Height Cm Coefficient of Moment S Wing Area c Chord Length M Mass of craft Greek α Angle of Attack Ω Glide Angle θ Angle of Pitch µ Relative Density Introduction

Recent technological advances in the areas of propulsion, guidance systems, and

microelectronics, have made commercially viable miniature autonomous flying vehicles,

Micro Air Vehicles (MAV), possible. Most of the current concepts and prototypes

attempt to scale down traditional aircraft design to meet defense specifications. However,

classical aerodynamic concepts for fixed wing aircraft become impractical at the reduced

scale of MAV’s. Hence the special attention of this paper to the reverse engineering and

aerodynamics of a miniature Flying Wing the Golden Eagle.

The usual method of developing an aircraft is to decide what the mission requirements of

the new aircraft are, finding an aerofoil shape specific to it by testing, do a sizing and

performance optimization and integrate it together with the other parts of the aircraft, i.e.

controls, propulsion systems, payloads etc. However, only a physical model of the UAV

was given without adequate contractor’s aerodynamic propulsion and stability data. This

breaks the chain of development and it is required therefore to do a fair amount of reverse

engineering to determine a good estimate of these required data. Moreover, conventional

methods of testing and analysis may not apply to this UAV as it is much smaller and

2

slower than normal aircraft. New methods may have to be developed by trial and

evaluation.

Description of the UAV

The UAV given is basically a flying wing but with a central fuselage that follows the

reflex airfoil shape longitudinally and adapts to the curved ‘M’ shaped, tip to tip wing

layout when viewed from the back. (Figure 1) The entire aircraft (modular wings and

fuselage) is constructed using ultra-light weight composite Kevlar fibre. Its fuselage is

specifically designed to house 4 Lithium batteries, a speed controller and a rear pusher

propeller unit. The craft is estimated to be able to carry a payload of 1.5 kgs and fly at

speeds up to 20 m/s. Effectively, there are only two control surfaces on the UAV. These

are the left and right elevons found at the ends of the wings of the aircraft. These control

the pitching and rolling on this UAV.

Figure.1. Shape of the UAV (rear view) Figure..2. The given UAV

The wing tips are angled upwards at about 30 degrees to the horizontal to compensate for

the lack of the rudder surfaces, acting as a pair of winglets to provide lateral stability to

the aircraft. Neither exactly a Sweptback wing or a Delta wing, its unconventional airfoil

structure was carefully analyzed and pre-existent aerodynamic theories have been

adapted to suit it where possible.

3

3D Mesh Generation

Unable to match this particular wing with any of the standard NACA airfoils present, we

had to generate a full 3 Dimensional CAD model of the craft from scratch. With the

simplistic construction drawings provided we could not accurately determine the wing

curvature at the concave leading edge and at the convex tail. Hence, using the Minolta,

VIVID 900, Non-Contact-3D Digitizer Image Laser scanner, we photographed the entire

wing profile and fuselage with a tolerance of ±1.5 mm, which we then assembled and

merged using the commercial scan programme RapidFormTM 2002- Reverse Modeler

Version. Working with the photographed scattered points, we had to systematically

connect each coordinate to attain the complex curves on the wing. Plot linearization and

CAD editing was needed to marginalize the inaccuracy inherent in scanning.

4Figure.3. 3-D Laser scanning and Reverse aerofoil CAD modeling procedure

The model was then sectioned and sliced at critical intervals to obtain the exact structural

coordinates to be used to design and construct the wings. The entire CAD model was also

imported into GAMBITTM, a mesh preprocessing programme, and modified to avoid any

skewed edges before generating FLUENT compatible 3D surface and volumetric meshes.

Figure.4. 3-D Volumetric Mesh

Structural Construction

Because design development was heavily dep

and precision of manufacturing and repair was

development could begin. All components we

break away during impact. This ensures minim

and time. Various materials such as low

cardboards, balsa, paper march’es and la

manufacturing processors were experimented w

bi-directionally laid tissue carbon fiber (CFRP

body because of its high rigidity, superior stre

availability. CFRP also displays excellent me

consideration for a UAV without landing gear m

5

Figure .5. CAD model of the UAV

endent on flight-testing, the ease, speed,

a fundamental process we had to before

re determined to be modular and are to

al damage and hence reducing repair costs

and high density foam, stiff ¼-1/2 in.

minate resins together with different

ith and finally we singled out single ply

) as the desired material for the wing and

ngth-to-weight ratio, low cost and ease of

chanical properties upon impact- crucial

echanisms.

Reusable male and female clay molds were created and checked for consistency against

the acquired wing curvature dimensions. The carbon fiber framework was then laid on

the molds and covered with a thin layer of synthetic polymer (Ethylene Glycol, wt. %

99.9 - Polyester). Specifically measured quantities of resin were applied equally on each

of the two wings, maintaining symmetry in weight. The viscous resin was poured down

on the wing, with the mold propped vertically up. This ensures an even distribution of

resin throughout the cast. It was then allowed to drip and air dry in an enclosed area. This

procedure we discovered, gave a smoother and more even exterior finish compared to the

conventional method of brushing on the polyester. The entire manufacturing process is

highly repeatable with the usage of durable and reusable molds and cost effective readily

available materials.

Figure.6. Pouring of the resin on the fiber. Figure.7. Fabricated CFRP right wing’s top shell Estimation of Aerodynamic Coefficients and Forces To derive the aerodynamic derivatives, we use the CAD model of the UAV we reverse

engineered. It is first converted to a STEP file and a volumetric mesh is generated using

GAMBIT™ to be compatible with FLUENT™ a commercially available Computational

Fluid Dynamic (CFD) programme. The Flying wing is subsonic UAV operating at low

6

Reynold’s Numbers hence we ignored compressibility effects for the lift and drag models

and modeled laminar flow conditions sighting the fact our craft operates close to the

transition region; making it simpler to assume a laminar case rather then a turbulent

scenario. In FLUENT™, we set up numerous models with different boundary conditions

to find how the UAV reacted to changes in speed, angle of attack and sideslip.

Figure.8. CFD Static Pressure Profile Plot

The coefficients attained were put into the equations of motions of the aircraft, and the

transfer functions of the UAV were derived. To derive the PID gains for the UAV, an

optimization was done to find the optimal gains for the UAV. MATLAB™ was used to

find the gains, using the transfer functions that were derived.

CL vs Angle of Attack

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 0.1 0.2 0.3 0.4 0.5 0.6

AOA in rad

CL

7

The aerodynamic plots obtained are reflective of a flying wing UAV aircraft. The general

shapes of the graphs are very similar to conventional airfoils and the aerodynamic forces

obtained are logical too. These will be experimentally verified in glide tests.

Estimation of CG and Inertias of Mass

Since the entire Flying wing model is fabricated using different materials from that used

in the original prototype, we need to do a comprehensive weight and balance analysis. In

general the term “weight and balance” refers to the mass properties of an aircraft and the

resulting stability or lack thereof as a consequence of its mass properties. The term “mass

properties” usually includes the following values: volume (or mass or weight), center of

mass (or center of gravity), and the moments and products of inertia. CG is the point (or

centroid) of the craft about which moments summed, due to the mass of the object, equal

zero. Therefore this point represents a balancing point for the whole craft, and the total

weight or gravity force can be represented as acting at this point. We weighed each

component individually and marked out their CG respectively. Using simple geometric

summation and parallel axis theorem, the combined CG position of the craft was found.

8

To experimentally verify our calculations, the conventional method of CG determination

was employed - the entire assembled model was mounted on a pivot and shifted

accordingly to attain the mass centre of the craft.

Pusher Propeller Unit 100g

Radio control electronics (two servo motors, servo card, RC receiver)

65g

Video electronics (camera, transmitter)

55g

Batteries (9-volt, 50 mAh NiCd)

350g

Micro Pilot Card & Cables

29g

Structure 750g

Total 1469g

Table 1: Equipment and weights Figure 9. Conventional CG balancing

The remaining of the payload was then strategically positioned within the fuselage to

shift the CG to the desired position before the aerodynamic centre. Fine tuning of this

exact location is to be done during the trimming routine to attain longitudinal stability

after glide tests. The equations for moment of inertia, are also referred to as “second

moment” equations. This is due to the squared moment arm that multiplies each

infinitesimal volume during the integration. In the case of the Ixx

, the distance from the x-

axis is the moment arm to be squared, and due to the Pythagorean Theorem, this squared

distance is y2

+ z2. The same method is used for the other moments of inertia. We can

approximate the Inertias with the geometric summation of the various components of

different masses in the structure, as per equations (1)-(6).2 We must assume that each

9

component has a constant density and mass distribution throughout. Thus, we obtain the

Inertia Tensor.

( ) ( )

( ) (

( ) ( )

( ) ( )

( ) ( )

( ) ( ) ∑

∑

∑

∑

∑

∑

=

=

=

=

=

=

=

=

=

=

=

=

−−=

=−−=

=−−=

−+−=

−+−=

++−=

ni

i

ni

i

ni

i

ni

i

ni

i

ni

i

XcgXiZcgZimiIzx

ZcgZiYcgYimiIyz

YcgYiXcgXimiIxy

YcgYiXcgXimiIzz

XcgXiZcgZimiIyy

ZcgZiYcgYimiIxx

1

22

1

22

1

22

1

22

1

22

1

22

0

0

)

(1) (2) (3) (4) (5) (6)

Symmetrical Aircraft

Longitudinal Stability-Balancing of Pitching Moments

Stability is a very important criterion in the design of aircraft. For aircraft, two conditions

must be met for longitudinal stability.

0&0 0 >< mm CC (7) α

0434.00985.20911.0

−=−

=αα

CLCm

(8)

As shown in our Cm vs α curve, the gradient is negative and the graph intersects the x-

axis on the positive end. The Cmα / CLα calculation tells us where our aerodynamic

centre lies, the point where the moment acting on the body is independent of the angle of

attack, and since this is a flying wing with a comparatively small central fuselage which

also rides the wing profile, we conclude that the neutral point too lies at the a.c location

calculated. The negative value (8) tells us that the ac actually lies behind the CG location.

10

Figure.10. Location of AC with respect to the CG

Experimental Verification

As flight-testing is an imperative step in the development any MAV design, the ability to

evaluate test-flights was critical. The primary flight characteristics we aimed to validate

via glide tests were stability, lift and drag. The glide tests must be conducted with the

engine installed and the propeller removed. As removing the engine would have created

an unrealistic mass distribution and a non-feathered propeller would have created an

uncharacteristically large drag.

Figure.11. Lift vs Drag Ratio verification The glider's flight path is a simple straight line, shown as the inclined black line in

Figure11. The flight path intersects the ground at an angle a called the glide angle, Ω.

Ω

h

d

11

Horizontal Force Equation: )cos()sin( Ω=Ω dL (9)

Ratio: hd

DL

=Ω

==tan

1DRAGLIFT (10)

(11) SD

SL

VV

××××=

××××=2

D

2L

C5.0

C5.0

ρ

ρ

D

L

CC

DL= (12)

We now would have verified our CFD simulated results of Drag and Lift forces and their

respective coefficients. The MAV prototype will also be fitted with the dummy camera

system and glide trials will continue to assess the trim and stability condition of the craft

with this additional payload. This will then allow further refinement of the Centre of

Gravity position to achieve acceptable flying qualities. The goal in trimming a flying

wing is to get the center of gravity as far aft as possible and still maintain stable control

over pitch. Since the flying wing has very little tail moment there is a tendency for the

wing to be very pitch sensitive. As this craft also does not originally have a rudder, its

yaw and lateral characteristics will also be closely assessed with easily interchangeable

rudders of different configurations fitted on hand during testing.

Concluding Remarks

The project satisfies the two primary goals. For the first loop of the design iteration, a

considerable achievement has been made. A working prototype has been designed, built

and ready for flight testing on time and within the allocated budget. Inline with the

second goal, Computational Simulations have given the required aero coefficients, forces

and moments which will be verified by flight tests. Problems with the CFD and mold

12

building highlighted the larger timescales involved with the preparation of a model mesh,

and the airfoil structure which had not been anticipated. Attempts have since been made

to shorten the time needed to accomplish this task, so future students investigating

MAV’s may easily replicate any wing form even when faced with a hard date-line.

Apart from the vast commercial viability of the reverse engineering procedures

introduced in this paper, they can also serve a crucial role in military and defense

applications, where one side may be able to replicate a captured enemy drone (obviously

with no supporting data) and reconfigure it to carry a micro-camera, flying it into the

enemy ground, without them even engaging it.

Acknowledgements I would like to sincerely thank my supervisor and mentor A/P Gerad Leng of the

Department of Dynamics, National University of Singapore for granting me this rare

opportunity and for his great encouragement and guidance all along the way.

References 1) Karl Nickel and Michael Wohlahrt,, Translated by Capt. E. Brown RN,, Second

Edition-1996 “ Tailess Aircraft in Theory and Practice”, AIAA, Education Series.

2) Dr.Jan Roskam, Ackers Distinguished Professor of Aerospace Engineering.

University of Kansas Lawerence, Kansas. “ Airplane Design, Part Five: Component

Weight Estimation” First Edition-1985

3) Daniel P Raymer, ‘Aircraft Design: A conceptual approach’, AIAA Education Series,

ISBN 1-5634/-281-0

13