Computational analysis of unmanned aerial vehicle (UAV)

Sakhr Abudarag, Rashid Yagoub, Hassan Elfatih, and Zoran Filipovic

Citation: 1798, 020001 (2017); doi: 10.1063/1.4972593

View online: http://dx.doi.org/10.1063/1.4972593
View Table of Contents: http://aip.scitation.org/toc/apc/1798/1
Published by the American Institute of Physics

Articles you may be interested in

World Congress: ICNPAA 2016 has been Organized with the Support of: International Conference on
Mathematical Problems in Engineering, Aerospace and Sciences
1798, 010002010002 (2017); 10.1063/1.4972592

Preface: International Conference on Mathematical Problems in Engineering, Aerospace and Sciences

1798, 010001010001 (2017); 10.1063/1.4972591

Shock waves as generalized solutions of thermoelastodynamics equations. On the uniqueness of boundary

value problems solutions
1798, 020003020003 (2017); 10.1063/1.4972595

Numerical study of a finite volume scheme for incompressible Navier-Stokes equations based on SIMPLE-family
algorithms
1798, 020004020004 (2017); 10.1063/1.4972596
Computational Analysis of Unmanned Aerial Vehicle (UAV)

Sakhr Abudarag1,a), Rashid Yagoub1,b), Hassan Elfatih1,c) and Zoran Filipovic 2,d)

1
Sudan University of Science and Technology, Aeronautical Engineering Dept., P.O. Box: 407, Khartoum, Sudan
2
Institute GOŠA, Milana Rakica 35, Belgrade, Serbia.
a)
[email protected]
b)
[email protected]
c)
[email protected]
d)
[email protected]

Abstract. A computational analysis has been performed to verify the aerodynamics properties of Unmanned Aerial
Vehicle (UAV). The UAV-SUST has been designed and fabricated at the Department of Aeronautical Engineering at
Sudan University of Science and Technology in order to meet the specifications required for surveillance and
reconnaissance mission. It is classified as a medium range and medium endurance UAV. A commercial CFD solver is
used to simulate steady and unsteady aerodynamics characteristics of the entire UAV. In addition to Lift Coefficient (C L),
Drag Coefficient (CD), Pitching Moment Coefficient (CM) and Yawing Moment Coefficient (CN), the pressure and
velocity contours are illustrated. The aerodynamics parameters are represented a very good agreement with the design
consideration at angle of attack ranging from zero to 26 degrees. Moreover, the visualization of the velocity field and
static pressure contours is indicated a satisfactory agreement with the proposed design. The turbulence is predicted by
enhancing K-ω SST turbulence model within the computational fluid dynamics code.

INTRODUCTION
In last two decades, extensive researches were carried out to develop the UAVs field. This attention reveals from
the fact that the UAV can be utilized for many different missions such as surveillance and reconnaissance. Coincide
with the global growing trend towards it is applications; the SUST-UAV has been designed by the Department of
Aeronautical Engineering at Sudan University of Science and Technology {SUST}. The UAV derivatives were
found by DATCOM but there has been no detailed verification of its aerodynamic properties [1]. The application of
computational fluids dynamics has revolutionized the process of aircraft design. With increasingly rapid advances in
the development of powerful computers, the numerical simulations of various approximations to the Navier-Stokes
equations were performed. CFD has joined the wind tunnel and flight test as primary tools because of the need of
Boeing aerodynamicists to a tool that could accurately predict and confirm vehicle flight characteristics. This tool is
consisted almost of analytic approximation methods [2]. The main challenge faced by many computational
simulations was the drag and pitching moment predictions which have been experienced by several researchers.
Therefore in 2001, the CFD Drag prediction workshop series (DPW) were initiated. Since 2001 four workshops
have been held addressing different transonic speeds and aircraft configuration. The Fourth AIAA Drag Prediction
Workshop (DPW-IV) was focused on the prediction of both absolute and differential drag levels for wing-body and
wing-body-horizontal-tail configuration. The Drag, lift and pitching moment predictions from numerous Reynolds-
Averaged Navier-Stokes computational fluid dynamics methods were presented. Solutions were performed on
structured, unstructured and hybrid grid systems. The CFD simulations of the previous workshops have been
compared with both, the existing experimental data from previous wind tunnel tests and tested data from new wind
tunnel experiments using refurbished and new aircraft models [3]. The University of Technology, Malaysia was

ICNPAA 2016 World Congress

AIP Conf. Proc. 1798, 020001-1–020001-10; doi: 10.1063/1.4972593
Published by AIP Publishing. 978-0-7354-1464-8/$30.00

020001-1
developed UTM Elang-1 UAV and carried out an analysis for drag and lift characteristics. The aerodynamic lift and
drag obtained from the CFD simulation and analytical based on DATCOM calculations had been compared. The
results proved they were agreeable especially at a low angle of attack [4]. The National University of Singapore was
validated the aerodynamic and stability data by using different CFD programs and compared between them and the
numerical data without using wind tunnel. They were reached to the same results in less time [5]. The mentioned
studies highlights the need for utilizing the current computational fluid dynamics method for investigating the
complex flow fields around the SUST-UAV and verifying the computed results with the analytical data.

This paper aims to verify the conceptual design of UAV, in particular the aerodynamic characteristics by
performing a computational Fluid Dynamic (CFD) investigation. The results from this analysis in addition to the
wind tunnel experimental data were used to validate the existed design. This work is part of a project reveals in
increasing knowledge in design an efficient UAV in order to be able to evaluate the performance and stability during
flight test of a full-model of the UAV.

MATHEMATICAL FORMULATION
For the SUST-UAV analysis, the flow was assumed to be incompressible due to the maximum speed of 47 m/s.
Therefore, only Navier-Stokes equations with the K-Omega SST and continuity equation were solved
simultaneously.

Governing Equations
Continuity Equation

wu i
0 (1)
wxi

Navier-Stokes Equations

These equations were employed in the following form [6]:

* * * * * & &
wq wE wF wG wR wS wT
     (2)
wt wx wy wz wx wy wz
where,

ª1 º ªu º ªv º ªw º ª0 º ª0 º ª0 º
«u » & « 2 » «uv » & «uw » & «W » & «W » «W »
& u  P / U» & », G », R « xx », S « yx », T& « zx »
q U « », E U« ,F U« 2 U« «W yy »
«v » «uv » «v  P / U » «vw » «W xy » «W zy »
« » « » « » « 2 » « » « » « »
¬ w¼ ¬«uw ¼» ¬«vw ¼» ¬« w  P / U ¼» «¬W xz »¼ «¬W yz »¼ «¬W zz »¼

Turbulence Models Equations

Selection of turbulence model depends on the type of grid i.e. structured or unstructured grid and for the present
simulation The SST model has been selected. Historically standard two-equation models miss the separation and
predict attached flow even for strong pressure gradient flows. But SST model is one of the most accurate two-
equation models for separation prediction it provide very good calculations of wall bounded flows even with highly
separated regions. This model is more accurate than k-epsilon especially near wall layers and for flows with
moderate adverse pressure gradients. It has been developed for aerospace applications as follows.[7, 8].

020001-2
 wUk wUu j k w §¨ wk ·¸
 Pk  E * UZk  P  R k1 P t  (3)
wt wxi wx j ¨© wx j ¸¹

wUZ wUu j Z w §
¨ P  RZ1 Pt  wZ ¸
·
 J 1 PZ  E1 UZ 2  ¨
(4)
wt wxi wx j © wx j ¸¹
the eddy-viscosity:
k
Qt (5)
Z
turbulent stress tensorɒ୧ǡ୨ is given by:
§ wu wu j · 2
W ij vt ¨ i  ¸  kG ij (6)
¨ wx wx ¸ 3
© j i ¹

the shear stress is calculated as follows:

W Ua1 k (7)

with the constant ƒଵ ൌ ͲǤ͵, the shear-stress in two-equation models is computed from.

W Pt : (8)

The turbulence intensity for the present study is assumed 0.1%, whereas the turbulent viscosity ratio, Pt / P  is
directly proportional to the turbulent Reynolds number Re t
2

k / HQ  . Re t is large in high-Reynolds number
boundary layers, shear layers and fully developed duct flows. However at the free-stream boundaries of most
external flows, Pt / P  is fairly small [9]. Therefore, the turbulence parameter is set to 1  P t / P  10 .
COMPUTATIONAL EXERCISE

Case Study

The Table 1 below shows the specifications of the designed UAV.

TABLE 1. UAV specifications.
Parameter Magnitude/Type
Altitude 3 km
Cruise Speed 47 m/s
Gross Weight 43 kg
Endurance 9.6 hrs
Range 250 km
Wing Span 4m
Reynolds Number 1.126.106
Total Length 3m
Stall Speed 25 m/s
Wing Airfoil NACA 4415
Tail Airfoil NACA 0012

020001-3
 Geometry and Mesh
The aerodynamic characteristics of the designed UAV were obtained by performing a computational approach.
The program used for the investigation, was ANSYS 14 FLUENT and the result has been processed subsequently
within Microsoft Excel environment.

Computational Domain
The computational domain was selected after extensive evaluation for different types. For Drag estimations on
experimental aircraft using CFD, Staffan Hardie et al. [10] were compared between spherical domain which
characterized by the model setting at the same distance from all directions to provide minimum volume and the
cubical domain in which the modeling and applying boundary conditions is more easier to set up. The disadvantage
of the cubical domain is the creation of the wasted space at the corners of the domain that affects the result and
increases the computational time. The technique adopted for the present investigation to create a large computational
domain, is the integration of both, hemispherical and cylindrical domains as shown in Figure 2. In this case the
domain effect is minimized by setting the far field values of velocity and pressure, without missing details of wing
downwash and wake behind wing and fuselage.

FIGURE 1. UAV Geometry FIGURE 2. Computational Domain

Mesh Size
Four million elements were created on the surfaces and volume of the computational domain of the UAV by
sizing the minimum spacing (minimum element size) to 4 mm. Unstructured grid with triangles and tetrahedral in
the surface and volume meshes was implemented.

Boundary layer Consideration

Considerable attention was taken during the creation of the mesh at the proximity area to the UAV body where
the boundary layer was performed. Therefore the Inflation was applied in order to take into account the effect of the
boundary layer on its aerodynamic behavior. The parameters that control the Inflation were carefully chosen in order
to capture the flow phenomena within the boundary layer such as transition or separation [11]. The boundary layer
thickness was determined based on laminar flow estimation. To precede this value in the commercial software, the
Total Thickness option in the inflation menu in ANSYS was selected. This option enabled to create constant
inflation layers by enhancing the value of the Number of Layers and Growth Rate to obtain the total boundary layer
thickness. The Number of Layers control determined the actual number of inflation layers in the mesh except in
places where layers were removed locally for reason of improving mesh quality they had a valid values from 1 to
1000. The default value was 5 for solid bodies and 2 for surface bodies. Moreover, the number of inflation layers
applied must not exceed 100 [6, 11].

020001-4
 FIGURE 3. Inflation layers near wing leading edge FIGURE 4. Inflation layers near wing trailing edge

FIGURE . Inflation layer around the fuselage

The Growth Rate control determined the relative thickness of adjacent inflation layers. Drifting away from face
proximity where the inflation control was applied, each successive layer was approximately one growth rate factor
thicker than the previous one [9].

RESULTS AND DISCUSSION

For computational analysis, ANSYS FLUENT 14 was used to compute the aerodynamics forces and moments at
cruise flight conditions. The results for two solutions, turbulent and laminar models were validated against the
theoretical design of the UAV. The turbulence model k-ω was used to capture the wake behind the aircraft. Figure 4
below shows the variation of the lift coefficient verses angle of attack at RE=1.126.106 for the designed UAV
according to the type of model used. The maximum lift coefficient for the theoretical calculation is 1.7 occurs at
angle of attack 16 degrees, whereas for both computational models is occur approximately at angle of attack 22
degrees, but with different maximum lift coefficient. The maximum lift coefficient for turbulent model is 2.3,
whereas for the laminar model is 2.2. It is observed that the results obtained from the computational analysis of
SUST-UAV for the two models and the analytical data based on DATCOM were agreeable at a low angle of attack
(0 < α < 10) but by increasing angle of attack there is a significant difference in maximum lift coefficient value. The
difference or the delay in stall angle for computational analysis might reveal to steady state assumption. Figure 5
represents the curve of drag coefficient verses angle of attack. It is observed from the results that the computational
analysis drag curve slope is higher than the drag curve slope for DATCOM method. This difference might caused by
inaccuracy and imperfections of the 3D UAV model or by the assumption of one vertical tail in the analytical
analysis because the DATCOM does not permit twin boom configuration. Besides that, the analytical analysis based
on DATCOM is well known more suitable for supersonic aircraft such as rocket. For subsonic, it is more suitable for
aircraft which has Mach number above 0.3, whereas for lower values, some assumptions are considered in order to
be able to calculate using the figures provided in DATCOM method. However, the graphical curve for drag
coefficient is almost indicating the similar trend. Therefore, the good agreement between the two methods used for
determining the drag coefficient is definitely able to validate the accuracy of the analysis work in the present study.

020001-5
 cL 2.5 cD 0.35
0.3 cd k-ω
2 cd theoretical
0.25 cd-laminar
1.5 0.2
1 0.15
cl k-ω 0.1
0.5 cl theoretical 0.05
cl-laminar
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Alpha Alpha
FIGURE . Lift coefficient vs. angle of attack FIGURE . Drag coefficient vs. angle of attack

0.6 0.1

Pitching moment
Yawing moment

0.4 cm k-ω
0 cm theoretical

coefficient
coefficent

0.2 cm laminar
-0.1
0
-0.2 -0.2
cn
-0.4 -0.3
Cn theoretical
-0.6 -0.4
-10 -8 -6 -4 -2 0 2 4 6 8 10 0 2 4 6 8 10 12 14 16 18 20
Side slip angle Alpha
FIGURE . Yawing moment coefficient vs. side slip angle FIGURE . Pitching moment coefficient vs. angle of attack

Figure 6 shows the correlation between yawing moment coefficient and side slip angle (β). The airplane is said
to be directionally stable if it has an inherent capability to realign itself into the resultant wind whenever disturbed
from steady level flight [12]. Mathematically, the requirement for directional stability criterion is that CNβ > 0, and
from this figure, it is obvious that this criterion is satisfied. The variation of moment coefficient verses angle of
attack is indicated by Figure 7. The moment coefficient values start positive and then decrease gradually until the
trim point where they become negative. The positive moment coefficient at zero angle of attack (‫ )ܱܯܥ‬and the
negative moment coefficient variation with angle of attack are satisfied the requirement of longitudinal stability
criterion. The value of negative (‫ )ߙܯܥ‬and positive (‫)ߚܰܥ‬, demonstrate that the UAV is statically stable.

Figures 8 and 9 below show the change of pressure contour on the upper, lower, front and side surfaces of the
aircraft before and after the stall, respectively. Figure 8 illustrates the pressure contours before stall at angle 16. It is
clear that the pressure increases gradually from the leading edge till the trailing edge over the wing surface. The
pressure is almost constant at the lower surface of the wing but with high magnitude than the upper surface. The
pressure at the wing tips still higher than the pressure at the leading edge, but less than the pressure side. The same
scenario of the pressure distribution is also observed for the horizontal tail whereas it is almost constant for the
vertical tail. This result for vertical stabilizer may reveals from utilizing a symmetrical airfoil section. The maximum
pressure on the fuselage is concentrated at the stagnation point that represented by the nose of the fuselage and then
degreases to a certain value for the rest of the fuselage for both side surfaces. The pressure has the same magnitude
of the stagnation point on the fuselage precisely in the region under the wing. Figure 9 represents the pressure
difference at angle of attack 24 after the stall. The pressure on the upper surface of the wing start increases earlier
than the pressure before the stall, therefore the entire upper surface is at higher value of pressure compared to one
before the stall. The lower surface of the wing is faced with highest constant pressure. The pressure over the
horizontal tail increases very rapidly to the maximum value, whereas it is keep constant and at higher value than
pressure before the stall for the vertical stabilizer. The pressure on the fuselage sides under the wing is increased
after the stall. It obvious that the magnitude of pressure on the upper and lower surfaces of the wing of the UAV in
figure 9 is more higher than in figure 8, which indicates the increment in pressure due to the stall phenomena.

020001-6
FIGURE . Pressure contours at angle 16 before stall

The effect of the wing drag on the horizontal tail is illustrated in both Figures 8 and 9. In Figure 8 at angle of
attack 16 degrees before the stall, the developing of the boundary layer over the upper surface of the tail indicated
clearly that the horizontal stabilizer is not influenced by turbulent flow behind the wing.

FIGURE . Pressure contours at angle 24 after stall

Figures 10, 11, 12 and 13 indicate the change of pressure contour at the Mean Aerodynamic Chord (MAC) of the
wing and the horizontal stabilizer before and after the stall.

020001-7
FIGURE 1. Pressure contour around MAC for wing FIGURE 1. Pressure contour around MAC for
before stall at angle of attack 16 degree horizontal tail before stall at angle of attack 16 degree

Figure 10 represents the pressure contour at MAC around the wing at angle of attack 16 degrees before the stall.
Constant pressure is observed from the stagnation point on the leading edge till the maximum thickness of the
airfoil. Then the pressure gradient increases upward and backward over the upper surface. The pressure on the lower
surface is greater than the pressure on the upper surface. The wing surface is considered as a Dirichlet boundary
condition, therefore the pressure distribution is a function of the wing camber. On the other hand, Figure 11
represents the pressure contour at MAC for the horizontal tail before the stall at angle of attack 16 degrees. The
horizontal tail is symmetric airfoil, but according to angle of attack the distribution of pressure is cleared. Constant
pressure is captured from the leading edge to about 25% of the wing chord. Then the effect of the Dirichlet boundary
condition is verified. The pressure on the pressure side is greater than the pressure on the suction side. The pressure
distribution on the upper surface of the tail is predicted.

FIGURE 1. Pressure contour around MAC for wing FIGURE 1. Pressure contour around MAC for horizontal
after stall at angle 24 degree tail after stall at angle 24 degree

Figure 12 represents the pressure contour at angle of attack 24 degrees at mean aerodynamic chord (MAC) for the
UAV wing after the stall. The pressure gradient increases rapidly and greater than pressure before the stall in the
axial and vertical axes. The propagation of pressure at stagnation point is less compare with the pressure at angle of
attack 16 degree. Whereas, Figure 13 represents the pressure contour at MAC for the horizontal tail after the stall at
angle of attack 24 degrees. The pressure increases over the whole tail reveals in complex vertices. Figures 14 and 15
show the change of velocity contour for the wing and the tail before the stall at the mean aerodynamic chord (MAC),
respectively. The velocity wake is predicted around the wing and tail surfaces before stall. The wake at the trailing
edge of the wing and horizontal tail is observed but with minimum velocity. Moreover, the velocity is decreased
over the pressure surface from leading to trailing edges. Figures 16 and 17 represent the change of velocity contour
for the wing and the tail after the stall at the MAC. The wake increases gradually as angle of attack increased and
wake begin to rise downstream when it is stall.

020001-8
 FIGURE 1. Velocity contour around MAC for wing FIGURE 1. Velocity contour around MAC for horizontal
before stall at angle 16 degree tail before stall at angle 16 degree

FIGURE 1. Velocity contour around MAC for wing FIGURE 1. Velocity contour around MAC for horizontal
after stall at angle 24 degree tail after stall at angle 24 degree

Figure 18 illustrates the flow pathlines over the upper surface of wing before the stall at angle of attack 16
degrees. It is obvious there are no vortex existence and the flow moves linearly. Figure 19 shows the flow pathlines
after the stall at angle of attack equal to 24 degree. In this figure, the existence of vortex on upper surface and tip
vortices are indicated. The results reveal that, the separation occurs near wing root first and starts to extend spanwise
by increasing angle of attack. The high kinetic energy of the wing tip vortices pursues the swirl and rotation of the
vortices to increase the induced drag.

FIGURE 1. Velocity path line for wing before stall at FIGURE . Velocity path line for wing after stall at angle 24
angle 16 degree degree

020001-9
FIGURE 2. The pressure coefficient at angle 16 degree FIGURE 2. The pressure coefficient at angle 22 degree

Figures 20 and 21 show the calculated pressure coefficient distribution on the pressure side and suction side of
the wing at the Mean Aerodynamic Chord (MAC) for angle of attack 16 degree and 22 degree respectively. The
pressure coefficient is plotted versus the axial chord length. The calculations show that at angle 16 degree, the
maximum pressure coefficient at the leading edge on the lower surface is approximately at 0.55m and then decreases
curvedly till the trailing edge at 0.92m. On the other hand, for angle of attack 22 degree the maximum pressure
coefficient at 0.52m from the leading edge and drop slightly with constant slope till the trailing edge at 0.95m. for
the negative pressure on the upper surface of the wing, the pressure coefficient is convex upward till approximately
0.8m and then concave upward till the trailing edge for angle 16 degree whereas, for angle 22 degree the pressure is
convex upward till the trailing edge.
CONCLUSION
The design of SUST-UAV has been computationally validated. The computational analyses were carried out for
different angles of attack enhancing both, viscous laminar model and K-omega SST turbulence model. The
aerodynamics parameters were represented agreeable results with the design consideration at low angle of attack but
at high angle of attack there was a considerable overprediction in maximum lift coefficient. The designed UAV was
found to be stable longitudinally and directionally.

REFERENCES
1. A. H. Madni, A. H. Al-Noor, A. A. Ehamer, M. A. Ismael, M. M. Osman and M. M. Ali, “ Design and Build of
UAV with Co-Axial Propeller,” B.Sc. Thesis, Sudan University of Science and Technology, 2014.
2. E. N. Tinoco, “Thirty Years of Development and Application of CFD at Boeing Commercial Airplanes,
Seattle,” 2003.
3. E. N. T. John C. Vassberg, Mori Mani, Ben Rider and others, “Summary of the Fourth AIAA CFD Drag
Prediction Workshop,” AIAA, 2010.
4. M. N. B. D. Haris Ahmad Bin Israr Ahmad, “Estimation of Lift and Drag Characteristics of UTM Elang-1
UAV,” in 2nd Regional Conference on Vehicle Engineering and Technology., Kuala Lumpur, 2008.
5. L. J. H. a. G. L. Navabalachandran Jayabalan “Reverse Engineering and Aerodynamic Analysis of a Flying
Wing UAV.”
6. M. M. Abdulla, “Computational Aerodynamic Analysis of Light Aircraft,” Military Technical College, Kobry
Elkobbah, Cairo, Egypt, 2009.
7. F. R. Menter, M. Kuntz and R. Langtry, “Ten Years of Industrial Experience with the SST Turbulence Model,”
Turbulence, Heat and Mass Transfer 4, Begell House, Inc, 2003.
8. F. R. Menter, “Improved Two-Equation k-ω Turbulence Models for Aerodynamic Flows,” NASA Technical
Memorandum 103975
9. ANSYS Fluent 12.0 User's Guide, ANAYS, Inc, 2009.
10. S. Hardie, “Drag Estimations on Experimental Aircraft Using CFD,” Mälardalen University.
11. A. T. Piperas, “Investigation of Boundary Layer Suction on a Wind Turbine Airfoil using CFD,” Technical
University of Denmark, 2010.
12. H.Schlichting and K. Gersten, Boundary Layer Theory: McGraw Hill, 2000

020001-10