Conceptual Design of a Hybrid Hydrogen Fuel Cell/Battery Blended-Wing-Body Unmanned Aerial Vehicle—An Overview

Abstract

The manuscript presents the conceptual design phase of an unmanned aerial vehicle, with the objective of a systems approach towards the integration of a hydrogen fuel-cell system and Li-ion batteries into an aerodynamically efficient platform representative of future aircraft configurations. Using a classical approach to aircraft design and a combination of low- and high-resolution computational simulations, a final blended wing body UAV was designed with a maximum take-off weight of 25 kg and 4 m wingspan. Preliminary aerodynamic and propulsion sizing demonstrated that the aircraft is capable of completing a 2 h long mission powered by a 650 W fuel cell, hybridized with a 100 Wh battery pack, and with a fuel quantity of 80 g of compressed hydrogen.

Full text

Citation: Suewatanakul, S.; Porcarelli,
A.; Olsson, A.; Grimler, H.; Chiche,
A.; Mariani, R.; Lindbergh, G.
Conceptual Design of a Hybrid
Hydrogen Fuel Cell/Battery
Blended-Wing-Body Unmanned
Aerial Vehicle—An Overview.
Aerospace 2022,9, 275. https://
doi.org/10.3390/aerospace9050275
Academic Editor: Konstantinos
Kontis
Received: 15 October 2021
Accepted: 10 May 2022
Published: 19 May 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
aerospace
Article
Conceptual Design of a Hybrid Hydrogen Fuel Cell/Battery
Blended-Wing-Body Unmanned Aerial Vehicle—An Overview
Siwat Suewatanakul 1,†, Alessandro Porcarelli 1,† , Adam Olsson 1,†, Henrik Grimler 2,† , Ariel Chiche 2,† ,
Raffaello Mariani 1,* and Göran Lindbergh 2
1Engineering Mechanics, Unit of Aeronautical and Vehicle Engineering, KTH Royal Institute of Technology,
114 28 Stockholm, Sweden; siwat@kth.se (S.S.); apor@kth.se (A.P.); adolsso@kth.se (A.O.)
2Applied Electrochemistry, KTH Royal Institute of Technology, 114 28 Stockholm, Sweden;
hgrimler@kth.se (H.G.); chiche@kth.se (A.C.); gnli@kth.se (G.L.)
*Correspondence: rmariani@kth.se
† These authors contributed equally to this work.
Abstract:
The manuscript presents the conceptual design phase of an unmanned aerial vehicle, with
the objective of a systems approach towards the integration of a hydrogen fuel-cell system and Li-ion
batteries into an aerodynamically efficient platform representative of future aircraft configurations.
Using a classical approach to aircraft design and a combination of low- and high-resolution computa-
tional simulations, a final blended wing body UAV was designed with a maximum take-off weight
of 25 kg and 4 m wingspan. Preliminary aerodynamic and propulsion sizing demonstrated that the
aircraft is capable of completing a 2 h long mission powered by a 650 W fuel cell, hybridized with a
100 Wh battery pack, and with a fuel quantity of 80 g of compressed hydrogen.
Keywords:
conceptual design; blended wing body; unmanned aerial vehicle; hydrogen-electric
propulsion; fuel cells
1. Introduction
For more than 60 years, since the design of the Boeing 707, civil aviation has shown
constant progress towards more efficient aircraft. Initially driven by pure economics for
fuel efficiency, recent years have seen efficiency improve also as an effect of concerns for
environmental issues such as greenhouse gas emissions—aviation contributes for 2% to 3%
of global carbon dioxide emissions, and this number is projected to grow [
1
,
2
]—and noise
pollution. As a result, there has been a significant push towards “more electrical aircraft,”
with a focus on battery as well as hydrogen power. Both hydrogen fuel cells and batteries
are electrochemical devices that convert chemical energy into electricity. The difference
between the two systems resides in the energy storage method: as hydrogen (in gaseous or
liquid for) in an external storage for use in a fuel cell; and inside the electrode material for
a battery [
3
]. Hydrogen fuel cell systems offer high specific energy and are therefore an
interesting candidate for replacing the APUs in commercial aircraft [
1
,
2
,
4
–
6
]. In addition,
feasibility studies and test flights have been carried out in light aircraft by fitting fuel cells
to existing two-seater [6–8] and six-seater airframes [9].
Prior to these larger scale efforts in civil aviation, the use of hydrogen fuel cells has
been evaluated on small to medium sized unmanned aerial vehicles (UAV) [
10
], as fuel
cells in combination with batteries. The resulting hybrid system exploits the high energy
density of hydrogen and high power density of batteries, and from an operation standpoint
they combine low environmental impact, low thermal signal, and quiet performance. Since
the energy is stored as hydrogen, it is also possible to achieve high endurance without a
significant impact on the aircraft weight, keys design factors for successful UAV operations.
The most prominent among these demonstrators is the IonTiger designed by the US Naval
Research Laboratory, which competed a record-breaking flight of 48 h [11,12].
Aerospace 2022,9, 275. https://doi.org/10.3390/aerospace9050275 https://www.mdpi.com/journal/aerospace

Aerospace 2022,9, 275 2 of 23
Several other large-scale projects also tested the concept of fuel-cells for use in fixed-
wing UAVs. Within the United States, long endurance small UAVs powered by hydrogen
fuel cells have been designed and successfully flown by Georgia Tech [
13
] and Colorado
State University [
14
], and California State University and Oklahoma State University [
15
]
collaborated with Horizon Fuel Cell Technologies in their design. Medium to large scale
efforts were also conducted by Gong et al. [
16
] in Australia, through ground testing
simulation of fuel-cell-based power system for small UAV missions, and in Korea by
Gang et al. [17]
and Kim et al. [
18
]. A thorough study, both analytical and flight tested, has
also been conducted by Lapeña-Rey et al. at Boeing Europe [
19
], who analysed the in-flight
performance of the fuel cells stack in hot and dry conditions.
In addition to these major efforts, several small UAVs have been designed, primar-
ily as university research efforts, all of which share the same general configuration and
mission: high-aspect ratio wings for long endurance, and hydrogen fuel cells enhanced
in a hybrid system with LiPo batteries. Some studies are primarily analytical to evaluate
the operational aspects of fuel cell stacks, such as Gadalla et Zafar [
20
], Renau et al. [
21
],
Depick et al. [22]
, and the design by Verstraete et al. [
23
], which is a development of the Hy-
perion blended wing body project, and which was tested in flight for aircraft performance,
but not with the hybrid/electric system. Few examples exist, to the best knowledge of the
authors, of flight-tested aircraft, such as Özbeck et al. [
24
] and Savvaris et al. [
25
]. Most
recently, ISAE-Supaero [
26
] in France has launched a project developing a fixed-wing UAV
for ultra-long endurance (the Mermoz Challenge), powered by a hybrid-electric system
incorporating a fuel-cell with liquid hydrogen as fuel, which features thin, high aspect ratio
wing configuration for dynamic soaring, and TU Delft NederDrone [
27
], a unique tilt-wing
fixed wing drone for maritime operations, which has been flight-tested.
Setting itself apart is the Phoenix Project by AeroDelft, which is of added significance
as it is planned to transition from an unmanned version of a specifically designed aircraft,
powered by fuel cells utilising gaseous hydrogen, to a crewed aircraft, powered by fuel
cells utilising liquid hydrogen [28].
From a commercial perspective, French company Delair [
29
] is making the case for
the transition from battery-powered to hybrid-hydrogen powered fixed-wing drones with
the Hydrone project, as it has been recognized that the use of hydrogen fuel cells provide
significant advantages in terms of range for industrial applications, in particular with the
need of carrying a given payload.
Two aspects have been highlighted by this brief review of known past and current
projects: on one hand research in aircraft propulsion is clearly moving towards new, sus-
tainable, and environmentally conscious technologies, while on the other hand little or
no emphasis appears to having been given to push the boundaries in terms of aircraft
configurations towards high aerodynamic efficiency, for optimized use of these new green
technologies. An exception to this approach is the work by Guynn et al. [
30
], in which
the application of hydrogen fuel cells, in combination with the aerodynamically efficient
blended wing body (BWB) aircraft configuration was evaluated. Most recently, this philos-
ophy of combining zero-net emissions and high aerodynamic efficiency was adopted by
Airbus in its ZEROe program [
31
]. Literature has indicated that BWB configurations may
may result in improvements in fuel consumption of up to 10% for a nominal mission of
6000 nmi
and
300 passenger
, compared to a classical tube-and-wing configuration in the
same class [32,33].
It is therefore the objective of this multidisciplinary engineering work to develop
a UAV test-platform centred on the integration of a hybrid/electric hydrogen fuel cell
system in an aerodynamically efficient blended wing body configuration representative of
a potential configuration for future transport aircraft in the wide-body class.
The structure of the manuscript follows the steps of a traditional aircraft conceptual
design process. Section 1presents the background review of the existing work in the topic
of hybrid electric-fuel cell aircraft, and the goals and objectives of the current work are
stated. Section 2summarizes the design process and mission specifications and provides

Aerospace 2022,9, 275 3 of 23
the background information for the definition of the design constraints, as well as pre-
senting the educational aspect of the project. In Section 3, the initial sizing is discussed,
covering the weight and aerodynamic sizing as independent parameters to the final aircraft
configuration. Starting from a design maximum take-off mass of 25 kg, the structural and
mass energy fractions are estimated as 0.565 and 0.160 of the full mass of the aircraft. Reflex
airfoils were selected to retain the natural pitching stability of the aircraft. The conceptual
design phase is discussed in Section 4describing the design constraints for the body-section
of the aircraft, and the design choice of a 4 m span, 30
◦
swept wing with positive 4
◦
dihe-
dral and spanwise varying twist angle for enhanced performance. A stall angle of 9
◦
at
a maximum lift coefficient of 0.8 is shown. The preliminary power analysis and fuel cell
and hybridization are discussed in Section 5, where a 650 W fuel cell in connection with a
100 Wh
battery is shown to be sufficient to complete a flight of the duration in excess of
60 min in ideal conditions. Conclusions are presented in Section 7.
2. Pre-Conceptual Phase
2.1. Educational Aspects
An accompanying aspect of the project presented in the manuscript is the introduction
of aspects of net-zero carbon offset aviation research within the academic curricula. Broad
scoped research and education oriented projects with a strong emphasis on multidisci-
plinarity outside of the controlled environment of a scheduled course, as well as a strong
emphasis on practical work are not commonly available within school curricula, making
the current work an added value to education at KTH, The Royal Institute of Technology.
The combination of two aspects—research and education—are addressed at the core of
the project.
The work presented in the manuscript is built on the teaching concept of “Conceive-
Design-Implement-Operate” (CDIO) [
34
] as a series of complementary projects and master
thesis work, with topics aligned with the existing curricula within KTH, enhanced through
a multidisciplinary perspective so that no single project is “stand-alone” but is built on
previous work and becomes the basis for future work. This aspect mimics in part work
conducted in a research-focused environment as well as industry, emphasizing the need
for collaborative work. Student involved in the project have the opportunity of improving
technical skills within their “field of expertise” whilst at the same time exposing them to
learning soft skills necessary in industry.
Within each phase of the work, students are faced with challenges found in research
and industry. Throughout the pre-conceptual phase, students must assess the needs for
future aviation and highlight a “gap” in technology. In this particular case, this process
resulted in the choice of a hydrogen fuel cell and blended-wing-body aircraft as a good
balance between existing technology and knowledge to be developed. At every step of the
design phase, students need to select the appropriate tools and assess their suitability and
limitations and at times adapt well-tried methodologies to novel topics.
Finally, students involved in the project are developing soft skills comparable to those
necessary to those required in industry dynamically and in a synergistic approach with the
need of being aware of the project as a whole, in an approach rarely found in course work.
2.2. Design Methodology
The design methodology for the current conceptual work followed that presented by
Raymer [
35
], and shown in Figure 1, from Mission Requirements to the Final Configuration,
or design freeze. One aspects differs from the conceptual design process described by
Raymer [
35
], and that is the Technology Available here is replaced with Novel Technology and
placed at an equal level with Mission Requirements as this is a driver for the design and
mission specification. In addition, only introductory studies of the stability of the aircraft
are discussed.

Aerospace 2022,9, 275 4 of 23
Figure 1. Design Methodology [35].
2.3. Mission Specifications and Design Parameters
The first step in the design process is to determine the mission specifications that drive
the design and which ultimately define the configuration.
The purpose of the aircraft is to act as a flying test platform for compressed hydrogen
fuel-cell/battery electrical hybrid system. The aircraft is required to have a target endurance
of 1 h, at a cruise altitude of 500 m, and is expected to operate from paved runways. The
mission profile is shown in Figure 2. The long-duration mission requires beyond-line-of-
sight (BVLOS) and autonomous flight capabilities, which is beyond the scope of the work
presented herein and will be explored at later stages of the project.
Figure 2. Mission Profile.
A comparison of fuel-cell powered aircraft that completed flight testing was conducted
to provide a design frame for the current project, in particular the relation between fuel-cell
power and maximum take off weight, as presented in Table 1.
Table 1. Existing Demonstrators.
Aircraft Mass
[kg]
Wing Span
[m]
Cruise
Velocity
[m/s]
Aspect
Ratio
Fuel Cell
Power [W]
Bradley et al. [13] 16.4 6.58 14.5 23 500
Rhoads et al. [14] 13.4 5.54 - 22.76 600
Swider-Lyons et al. [11] 16 5.18 13.89 17 550
Renau et al. [22] 16 4 - - 700
Savvaris et al. [25] 13.04 - - 8.5 500
Lapena-Rey et al. [19] 11 4.7 - 22 500
Özbek et al. [24]6.5 3 16 9.76 250
The data in Table 1show that heavier UAVs require fuel-cells with power exceeding
500 W, and that in general all models evaluated have high aspect ratios representative of
glider-type aircraft as well as relatively low cruise velocities. Based on these values and
on the emphasis of designing a UAV with a configuration applicable to future commercial
aircraft, target values have been defined of a maximum take-off mass (MTOM) of 25 kg
(maximum take-off weight MTOW 245.3 N), which corresponds to the EU class C3C4 for
UAV certification, a maximum wingspan of 4 m, a design cruise velocity of 20 m/s, and a
stall speed of 12 m/s [
10
,
13
,
23
,
25
]. A fuel cell in the range of 500 W to 800 W is used for
the initial sizing. The initial parameters and sought aircraft configuration set the current
design beyond the boundaries of existing work.

Aerospace 2022,9, 275 5 of 23
3. Initial Sizing
For the conceptual phase of the design, a progressive approach was implemented,
starting from a classical method, which applies regression models based on available
data [
35
,
36
]. The size of the fuselage section of the aircraft was determined by the constraints
imposed by the dimensions of the fuel cell, which was preliminary defined by averaging the
size of commercially available models in the range of 500–800 W and 2 L to 3 L compressed
hydrogen tanks, pressurized at 300 bar [37,38].
3.1. Weight Estimation
The conceptual sizing was conducted initially using Raymer’s approach [
35
,
36
], subse-
quently transitioning to the preliminary weight estimation method by Gundlach [
39
]. The
approach by Gundlach was preferred when considering two aspects of a small to medium
size, unmanned aerial vehicle:
•
The approach takes the aspects of electric or hybrid-electric fuel cell-based propulsion
into consideration;
•
The breakdown of the weight components is more detailed, as in lighter aircraft it is
more critical to differentiate the weight of subsystems and avionics from the generic
empty weight, as they might have more significant impact in terms of weight fractions.
The maximum take-off weight WTO is then estimated using Equation (1) [39]:
WTO =WStruct +WSubs +WPro p +WAvion +WOther +WPL +WEnergy , (1)
where the subscripts Struct, Subs, Prop, Avion, PL, and Energy correspond to the structural,
subsystems, propulsion system, avionics, payload, and energy storage weights. The term
WOther
refers to the weight of items in the UAV that do not belong to any of the standard
categories, and will be neglected in the current design. This is because the mission of
the UAV under development is to test the hybrid/electric propulsion system, and any
instrumentation required for in-flight measurements is included in the payload category.
Preliminary weight estimations of fixed items is shown in Table 2, where a total of
six motors
+propeller-esc have been estimated, and a total of 15 servos have been estimated
to take into consideration multi control-surface designs for aircraft stability.
Table 2. Fixed items weight.
Item Mass [g]
Energy and Storage
Fuel Cell 810
Hybrid Battery 230
Hydrogen Mass 80
Hydrogen Storage 1400
Regulator 250
Extra Battery 2000
Avionics
Pixhawk 4 20
Holybro GPS Module 30
Holybro Telemetry 30
Propulsion
Emax Gran Turbo Motor 140
APC Propeller 25
Brushless ESC 95
Servos 50
Subsystems
Tricycle Landing Gear 420

Aerospace 2022,9, 275 6 of 23
Equation (1) can then be rewritten in terms of component weight fraction WF with
respect to the take-off weight, and careful attention must be paid to the weight fraction of
the propulsion
WFProp
and of the energy storage
WFEnergy
, as the estimation needs to be
broken down into sub-components for a fuel-cell powered aircraft. The mass fraction of the
propulsion is estimated, taking the weight of the fuel-cells stack into consideration, along
with the associated propulsion chain controller-motor-propeller. For a fuel-cell powered
aircraft, the propulsion mass fraction is:
WFProp =finstall
(P/WTO)Ai rcra f t
P/(WFuelCel l +WMotor +WController +WPropell er ). (2)
As can be seen from Equation (2), the mass fraction of the propulsion is directly related
to the power output Pof the fuel-cell, so it was important to have an initial estimation of
the power output required. In addition, the scaling factor
finstall
in Equation (2) is based on
the assumption that all items in the propulsion chain which are down-the-line from the
powerplant are scaled directly with the engine power. In the current initial evaluation, the
value for
finstall
is assumed to be equal to 1, as the term (
P/(WFuelCel l +WMotor +WController
)
include all installation items and propeller, meaning that the term
WPropeller
is set to zero. It
is acknowledged that this assumption underpredicts the mass fraction of the propulsion.
The weight of the compressed hydrogen, of the tank, and of any additional batteries
for the UAV are included in the category of energy weight
WEnergy
, and the energy mass
fraction
MFEnergy
is therefore estimated by calculating the ratio between the mass fraction
of the weight of the hydrogen and the storage, and by adding the mass fraction of the
estimated additional batteries required for high energy maneuvers and/or reservoirenergy:
WFEnergy =W FH2
WFStorage
+WFBattery. (3)
The mass fraction of the storage
WFStorage
of the compressed hydrogen must be
estimated taking the weight of the compressed hydrogen itself into consideration:
WFStorage =WH2
WH2+WStorage
. (4)
The remainder of the mass fractions, namely avionics and subsystems, were calculated
by adding up weights of the expected components, which have been listed in Table 2in
Section 3.1. Finally, the weight fraction of the structure was initially estimated following
Gundlach’s recommendation of a starting value of 0.6 [
39
], as there is no existing available
data for this class of aircraft, and Raymer’s fudge factor of 0.95 for composite material was
applied [
35
], resulting in a wight fraction of the structure of 0.565. The mass fractions and
corresponding absolute weights are presented in Table 3, including the values for payload,
which are a function of the fixed MTOW.
Table 3. Aircraft Initial Weight Sizing.
Symbol WF W [N]
WTO 1.000 245.3
WStruct 0.565 138.65
WSub 0.075 18.4
WProp 0.095 23.3
WAvion 0.050 12.3
WPL 0.055 13.5
WEnergy 0.160 39.2

Aerospace 2022,9, 275 7 of 23
3.2. Aerodynamic Sizing
The definition of the aerodynamic shapes of the aircraft are a compromise between
the required volume and dimensions imposed by the pre-selected fuel-cell system, and the
aerodynamic and stability constraints of a tailless BWB configuration, with the need of reflex
cambered airfoils to naturally restore the pitching moment stability of the
aircraft [40–42]
.
These constraints resulted in the selection of the Martin Hepperle MH 104 airfoil for the
body section with 15% thickness for additional combined favourable aerodynamic and
geometric properties, and the MH61 for the wing sections with 10% thickness. Specifications
for the two airfoils are presented in Table 4. Airfoil profiles are shown in Figure 3.
Table 4. Airfoil Profiles [43].
Aircraft Airfoil (t/c)max Max Camber
Section Section % %c % %c
Fuselage MH104 15.2 26.4 1.9 31.1
Wings MH61 10.2 27.6 1.4 37.3
Figure 3. Profiles of the MH61 and MH104 reflex airfoils.
The performance of these two airfoils was analyzed using the open-source software
XFOIL [
44
], obtaining lift and drag polars for both airfoils at a range of Reynolds number
between 250,000 and 5,000,000, providing preliminary estimations of
Clmax
and
Cl/Cd
.
The initial sizing of the lift coefficient for a wing of finite span was obtained by classical
approach [35,45]
, where the indicative value of
CLmax
is calculated by simplifying the
geometry of the wing and treating it as a “constant airfoil section wing” based on averaging
the maximum lift coefficient of the wing root and wig tip, corrected for 3-dimensionality of
the wing flow and quarter chord sweep at 30◦as shown in Equation (5).
CLmax =0.9Clm ax cosΛ0.25. (5)
The value of the maximum lift coefficient of this simplified wing was then used to
define the initial planform area required for an initially defined stall speed of 12 m/s.
Subsequently, for the design maximum take-off weight of 245 N, the initial value of the
planform area was estimated [
45
,
46
]. Finally, the sizing of the drag coefficient can be
obtained, with the simplified assumption that the minimum drag coefficient is equivalent
to the zero-lift drag coefficient [36]:
CD=CD0+KC2
L=CD0+C2
L
πeA R . (6)

Aerospace 2022,9, 275 8 of 23
Wing sweep was set at 30
◦
to be representative of a larger scale transport aircraft, and
the Oswald efficiency factor was calculated both for the unswept (or
Λ
< 30
◦
) (Equation (7))
and swept wings (or Λ> 30◦) (Equation (8)) configurations, and the values averaged [35].
e=1.78(1−0.045AR0.68 )−0.64 (7)
e=4.62(1−0.045AR0.68 )cos Λ0.25 −3.1. (8)
The initial value for zero-lift drag was taken from references [
36
,
47
,
48
] for small, fixed
wing UAVs, and it must be noted that it is a statistical value obtained for a class of UAVs,
rather than based on a specific configuration. It is expected that, for an aerodynamically
efficient BWB, the zero-lift drag of the final design will be lower than this initial estimation.
The value for the aerodynamic sizing, and the related geometric parameters are summarized
in Table 5.
Table 5. Initial Aerodynamic Sizing Parameters.
Parameter Symbol Unit
Aspect Ratio AR 5.47 [-]
Planform Area S 2.92 [m2]
Wing Sweep Λ30 [◦]
Wing Loading W/S 83.9 [N/m2]
Stall Speed VStall 12 [m/s]
Max. Sectonal Lift
Coefficient Clmax 1.22 [-]
Max. Lift Coefficient CLmax 0.95 [-]
Drag Coefficient CD0.078 [-]
Zero-Lift Drag
Coefficient CD00.015 [-]
Oswald Efficiency
Factor e0.83 [-]
4. Conceptual Design Phase
The next phase in the design was to complete the iterative process of defining the
final aircraft configuration, based on the initial design specifications of 25 kg MTOM and
4 m maximum wingspan presented in Section 3, the initial geometric and aerodynamic
sizing discussed in Section 3.2, and the requirement of fitting the fuel cell system in the
main section of the body. Three additional aerodynamic parameters have been considered
in the iterative process of the design: spanwise wing lift distribution, stall characteristics,
and natural static stability of the aircraft. Optimization of the wing design focused on
approximating the lift distribution towards an elliptical shape for increased efficiency [
49
].
Wing sweep promotes both longitudinal and directional stability, although an undesirable
effect of wing sweep is the first onset of stall occurring at the wing tip. To counteract this
phenomenon, and delay and shift the onset of stall towards mid-wing to retain aileron
effectiveness and roll control at high angles of attack, geometric twist was evaluated.
Dihedral angle, which provides inherent lateral stability, was evaluated as well.
4.1. Configuration Layout
With the BWB configuration being a lifting body, the first step in the design process
is to define the three sections of the aircraft: fuselage, fuselage-to-wing blending section,
and wing. As previously discusses, the design of the fuselage was driven by the geometry
and size of the fuel cell system for which the thicker MH104 airfoil was used. The sizing
parameters of the fuel cell are presented in Table 6.

Aerospace 2022,9, 275 9 of 23
Table 6. Fuel Cell System Initial Sizing.
Width Length Height Weight
[mm] [mm] [mm] [kg]
Fuel Cell Stack 170 200 110 1.0
3 L Composite
Tank 122 400 122 2.3
The result of the design process is a fuselage with a centreline chord (
ccl
) length of
1.7 m
and an outboard chord of 1.52 m. The main section of the fuselage has a width of
0.4 m
, a maximum thickness of 0.259 m at 0.449 m from the nose of the aircraft, and a
sectional planform area of 0.644 m2.
Keeping the geometry of the fuselage section fixed, the design was subsequently
parametrized using OpenVSP [
50
], an open-source software provided by NASA for the
conceptual design phase of an aircraft, for fast geometry optimization process of the
blending section and the wing. The wing shape was then aerodynamically optimized in
xflr5 [
51
] using classical methods such as vortex lattice, panel, and the non-linear lifting line
theory methods discussed in Section 4.2.1. Final geometric parameters and aerodynamic
analysis data are discussed in Section 4.3. The final configuration is shown in Figure 4. The
aircraft geometric parameters are presented in Table 7.
Table 7. Aircraft Geometry.
Parameter Symbol Unit
Wing Span 4 4 [m]
Centre Chord ccl 1.7 [m]
Root Chord cr1.29 [m]
Tip Chord ct0.25 [m]
Aspect Ratio AR 5.57 [-]
Planform Area S 2.87 [m2]
Wing Sweep (avg) Λavg 30 [◦]
Wing Loading W/S 85.5 [N/m2]
Oswald Efficiency
Facort e0.88 [-]
The root chord
cr
corresponds to the distance from the leading edge at the centreline
to the projected trailing edge of the wing, as shown in Figure 4.
The final step concerned the geometric definition of the centre of gravity based on the
components which, apart from the structure of the aircraft, impact the location, namely
those belonging to the “Energy and Storage” section as presented in Table 2. Each com-
ponent was modelled as a point mass centroid. The structure of the aircraft was initially
estimated as a carbon fiber shell and the centroid estimated using OpenVSP.
As previously mentioned, the volume and shape of the fuel cell and the hydrogen
storage constrained the size of the fuselage section of the aircraft, resulting in the hydrogen
tank being positioned in the thickest part of the fuselage and the fuel cell and hybrid
battery ahead of it towards the nose. These geometric constraints limited the movement of
the components along the longitudinal axis leaving the extra battery as the primary mass
for shifting the centre of gravity, resulting in the battery being placed at the nose of the
aircraft. The preliminary centre of gravity is
cg = (
0.718; 0.000; 0.038
)m
from the nose of the
aircraft. The location of these components and the centre of gravity are shown graphically
in Figure 4.

Aerospace 2022,9, 275 10 of 23
Figure 4. Aircraft Dimensions. All dimensions are in [mm].
4.2. Aerodynamic Analysis Methodology
4.2.1. Low-Fidelity Numerical Methods
The choice of a low-fidelity analysis approach was implemented as it allowed to
quickly and efficiently iterate the design at a low computational cost for both aerodynamic
and stability parameters [
52
]. Three viscous solutions were considered: the vortex lattice
method (VLM2) based on ring vortex analysis, and the panel method, which are both
linear solutions, and the non-linear lifting line theory (LLT) approach as it is able to
indicatively predict stall conditions as the solution includes non-linear approximations [
51
].
These methods were employed in parallel to obtain an overall initial understanding of
the aerodynamic performance, since all of them present limitations in the analysis with
respect to the current design [
44
,
51
]. The VLM2 method considers only the airfoil mean
camber line for the analysis and assumes the small angle approach in the aerodynamic
analysis, limiting the accuracy of the coefficient of lift at extreme (outside the linear region
of the lift curve, therefore near stall) low and high values of angles of attack, resulting in
the trailing vortices being misaligned with respect to the free-stream velocity. In addition,
viscous parameters for the coefficient of lift are obtained from a preceding airfoil analysis
using XFOIL [44], therefore adding an additional degree of approximation in the solution.
Oppositely, the non-linear LLT method has been primarily developed for high-aspect ratio
and low-sweep wings [
51
], such as those for sailplanes, and therefore of limited application
to the current design. The panel method is the only approach that takes the wing thickness
into consideration and is expected to lead to more accurate lift and drag initial estimations,
at least within the linear region of the lift curve. These results are discussed in Section 4.3.
4.2.2. CFD Methodology
Numerical Set-Up: A higher resolution analysis has been conducted using computa-
tional fluid dynamics (CFD) by means of Star-CCM+ commercial software by Siemens

Aerospace 2022,9, 275 11 of 23
(Release 16.02.008). A Reynolds-Averaged Navier–Stokes (RANS) model has been applied
(Equations (9) and (10)):
∂U1
∂xi
=0 (9)
∂Ui
∂t+Uj
Ui
xj
=−1
ρ
∂p
∂xi
+∂
∂xj v∂Ui
xj
−u0
iu0
j!. (10)
Two approaches have been used, with the RANS coupled with: the k−ωturbulence
model equations in the SST formulation; and with the Spalart–Allmaras model. Menter [
53
]
has shown that the RANS
k−ω
SST model provides excellent approximation of pressure-
induced separation and the resulting viscous-inviscid interaction in comparison to other
closures. Specifically designed for aerospace applications involving wall-bounded flows,
the Spalart–Allmaras model has been shown to give similarly good results for boundary
layers subjected to adverse pressure gradients [54].
As most of the RANS turbulent closures, both
k−ω
and Spalart–Allmaras are based on
the Boussinesq Hypothesis, where the traceless part of the Reynolds stress tensor is assumed
to be proportional to the Strain Rate Tensor
Sij
with the coefficient for eddy viscosity
vT[55]:
−u0
iu0
j=2vTSij −2
3kδij. (11)
By dimensional analysis, it holds that
vT
is proportional to the product of a turbulence
velocity and length scale. The
k−ω
models consist of solving a transport equation for the
turbulent kinetic energy kand the turbulent frequency
ω
, which allows the derivation of
the turbulence scales and closes the system.
The RANS and turbulence model equations have been discretized through a second
order upwind scheme. A steady-state solving approach has been employed to reduce
the computational expense. The inlet turbulence intensity has been set to 1% and the
Turbulent Viscosity Ratio
vT/v
to 10. While the intensity is in accordance with Spalart and
Rumsey [
56
] for external aeronautics-applied flows, the default value of viscosity ratio
in Star-CCM+ has been kept for simplicity and is a few units above the one suggested in
Spalart and Rumsey [
56
]. Further analysis performed within the project showed that this
boundary parameter has nearly no influence in the output aerodynamic coefficients for this
case-study and thus proved the reliability of the condition.
Mesh Validation: Following best practice cases for domain definition, a box-shaped
domain (4c–6c box) has been constructed, with dimensions shown in Figure 5. With the goal
of preventing the introduction of an unknown error related to the presence of the domain’s
external walls, a “slip” boundary condition has been introduced in their correspondence.
Furthermore, in order to optimize computational time, only half of the geometry has been
simulated, with a symmetry plane imposed on the xz-plane along the axis of symmetry. In
order to resolve the boundary layer, an inflation of high aspect-ratio cells has been designed
in the near-wall region of the aircraft geometry, as shown in Figure 5. Preliminary boundary
layer calculations have been performed with the turbulent flat plate formulas, yielding
a total thickness of
δ=
3.5 mm, which is prescribed as the maximum thickness of layers
inflation. Moreover, to ensure
y+<
1 and fully resolve the viscous sublayer region, a first
cell thickness of 1 ×10−5has been set.
The automated mesh algorithm of STAR-CCM+ allows to build a high-quality poly-
hedral mesh provided some general input parameters are given. For each of the surfaces
to be meshed, the user can prescribe a “minimum target size” (MSS) and a “target surface
size” (TSS). The former represents the minimum allowed cell dimension in the most curved
surface region, i.e., where the algorithm tends to refine the mesh to capture the curved
geometrical features. For instance, this parameter comes into play in the wing leading
edge limiting the refinement in its correspondence (Figure 6a). The latter sets a maximum
cell size in the regions where the algorithm tends to build a coarser mesh, i.e., in the least
curved or flat surfaces. For instance, this parameter is used to limit the mesh coarsening at

Aerospace 2022,9, 275 12 of 23
the wing upper and lower surfaces (Figure 6b,c) and at the far-field of the external domain
boundaries (Figure 5a). Furthermore, Star-CCM+ allows the design of simple geometrical
shapes to delimit domain regions where a mesh refinement can be introduced. Hence, in
order to better capture the wingtip vortices and thus better predict the induced drag, a
truncated-cone in the wingtip proximity has been designed.
Figure 5. Volume Mesh.
The mesh validity has been assessed through a grid-convergence study, at an angle of
attack of
α=
10
◦
, and the dependence of the aerodynamic coefficients on the following
mesh parameters has been explored:
•
The wing minimum surface size, which in practice defines the wing leading edge
smallest surface cell dimension;
•
The wing target surface size, i.e., the upper and lower wing surface cell maximum
dimension;
•
The external domain boundaries target surface size, i.e., the far-field maximum cell
surface dimension;
• The characteristic cell dimension within the cone-delimited wingtip mesh region.
The initial steps in the validation process consists in evaluating the wing minimum
surface size, for which a cell-size range between 5 mm to 0.5 mm was selected. Results
indicated that a cell dimension of 1 mm is the equilibrium between numerical accuracy
and computation time. Similarly, the wing target surface cell size has been evaluated for
a range between 30 mm to 7.5 mm, with mesh independence reached at a cell value of
7.5 mm
. Finally, the external domain boundaries target surface size and the wingtip cone
cell size were evaluated, the first for a range between 600 mm to 75 mm, and the second for
a range between 50 mm and 10 mm. Results show that mesh independence is reached at a
size of 150 mm for the simulation domain target surface size, and at a cell size of 30 mm for
the wingtip cone mesh.
The far-field domain shape and extension should be chosen to allow a safe boundary
condition setup without introducing non-physical impositions to the problem, and without
affecting the resulting aerodynamic coefficients. Further literature research showed that for
low Mach number flows, it is recommended to employ a bullet-shaped domain rather than
a box-shaped one [
57
]. Goetten et al. [
57
] also provide an overview of desirable and safe
domain extensions, which turn to be much larger than the ones employed in this study. For
instance, they report Versteeg’s and Malalasekera’s [
58
] suggestion to position the inlet and
the outlet significantly more than ten body lengths behind the geometry. These findings
highlight the necessity to further prove the validity of the results obtained using the initial
box-shaped domain shown in Figure 5. Hence, two additional extended domains, one
box-shaped and one bullet-shaped as shown in Figure 7, have been investigated. The cases
simulated are summarized in Table 8, with dimensions expressed, in this case, in terms of
wingspans.

Aerospace 2022,9, 275 13 of 23
Figure 6. Surface Mesh.
Table 8. Domain Evaluation for Spalart–Allmaras Turbulence Model.
Domain αDomain Domain Domain CLCD
Shape [◦]Length Width Height [-] [-]
Box 2 80 40 30 0.354 0.0193
Bullet 2 40 radius [-] [-] 0.354 0.0193
4c–6c Box 2 [-] [-] [-] 0.362 0.0190
Box 10 80 40 30 0.709 0.0935
Bullet 10 40 radius [-] [-] 0.727 0.0937
4c–6c Box 10 [-] [-] [-] 0.725 0.0958
Results showed that in both flow regimes, i.e.,
α=
2
◦
attached flow and
α=
10
◦
stalled flow, the lift and drag coefficient vary in a range below 2.5%. In particular, the
lift and drag coefficients show respectively a 2.3% and 1.6% discrepancy at low angles of
attack, and a 0.3% and 2.2% at stalled angles of attack. The lower lift coefficient variation
in the latter configuration is probably due to the higher numerical random errors which
characterize this flow regime. This very limited error in the CFD computations can be
considered acceptable for the conceptual design purposes of the present work, and thus
proves the reliability of the limited-extension domain initially designed.
Final Set-Up: After completing the mesh studies, initial aerodynamic evaluations
have shown a more stable convergence of the solution provided by the Spalart–Allmaras
turbulence model in comparison to the one of
k−ω
SST at high angles of attack, where
the flow is partially stalled, as it will be discussed in Section 4.3. In contrast, the results
of the two models are almost perfectly matching at low angles of attack, where the flow
is attached. Hence, in order to derive more reliable results at high angles of attack, and
ultimately predict a reasonable
CLmax
, Spalart–Allmaras was selected for the remainder of
the study. A similar choice has been performed by Panagiotou and Yakinthos [
47
]. They
compared the models for the prediction of aerodynamic forces on UAVs, and based on their
studies opted for the use of the Spalart-Allmaras turbulence model in their subsequent
work involving a BWB configuration [36,48].

Aerospace 2022,9, 275 14 of 23
Figure 7. Box (left) vs. Bullet (right) Domains.
Domain 4c–6c Box in Table 8has been used for the remainder of the work, mesh
parameters are summarized in Table 9and shown graphically in Figure 6.
Table 9. Final Mesh with 4c inlet and 6c outlet (4c–6c Box).
Simulation Cell Wing Wing Wing Tip External Domain
Reference Count MSS TSS Cone Cell DBoundaries TSS
4(PL2) 4.7 ×1061 mm 7.5 mm 30 mm 150 mm
4.3. Aerodynamic Performance
The preliminary configuration was tested using OpenVSP and xflr5 for fast iteration of
the effects of geometry variation on the aerodynamic performance of the aircraft. Nominally,
twist angle effects were iterated with the objective of refining aerodynamic efficiency and
obtaining a nearly elliptical lift distribution at cruise flight conditions, and dihedral angle
to provide natural lateral stability. The final configuration had a wing dihedral of 5
◦
(not including the blending and fuselage sections), and a twist distribution, from the
longitudinal axis y = 0.0 m, of 2.5
◦
up to y = 0.2 m, of 3
◦
at 0.2 m
≤y≤
1.5 m, and twist of
0
◦
at the wing tip. Tip stall was negatively affected, but this twist variation improved the
overall lift span distribution.
Results are presented in Table 10. These results are referenced to the initial aerody-
namic parameters listed in Table 5where available. All data were obtained at the cruise
velocity of 20 m/s, at a range of angles of attack between −4◦and 12◦.
Table 10. Comparison of aerodynamic parameters.
Parameter Unit VLM2 Non-Linear Panel CFD Initial
LLT Method - Sizing
a[◦−1]0.068 0.079 0.071 0.068 -
CLmax [-] - 0.90 - 0.789 0.95
αmax [◦] - 9 - 8.5 -
CD0[-] 0.008 0.009 0.009 0.011 0.015
Cm0[-] 0.004 0.000 0.009 0.002 -
Cmα[◦−1]−0.0031 −0.0026 −0.0039 −0.0024 -
Here, ais the lift curve slope,
CLmax
is the maximum lift coefficient,
αmax
the maximum
angle of attack,
CD0
the zero-lift drag, and
Cmα
the change of pitching moment coefficient
with respect to angle of attack α. Lift, drag, and moment coefficients with respect to angle
of attack are shown in Figure 8.

Aerospace 2022,9, 275 15 of 23
Figure 8.
Aerodynamic results comparison.
CL
vs.
α
and
CD
vs.
α
(
top left
),
CD
vs.
CL
(
top right
),
Cmvs. α(bottom right), and L/Dvs. α(bottom left).
The comparison of the lift curves obtained with the three methods indicate good
agreement between the results of the panel method and the CFD simulations in terms of
lift curve slope, with an error of 3.4%. The large difference in lift curve slope with respect
to data obtained using non-linear LLT method has already been witnessed.
The comparison of the drag coefficients shows a consistently larger drag coefficient
value in the CFD results, and this is expected as it is known that the analytical models
used in xflr5, due to the way friction and form drag are applied, underpredict drag. At low
angles of attack, where viscosity has a smaller effect as the flow is fully attached, this results
in a discrepancy of only 2–3 counts of drag, and this difference progressively increases as
the angle of attack is increased and viscous effects become more significant.
This aspect of modelling viscosity also affects the prediction of lift in the non-linear
LLT model in particular, for which the curve slope is steeper, and lift is over predicted
with respect to the CFD result. On the other hand, the lift curve slope obtained with the
CFD model agrees well with the VLM2 and panel method predictions, to a value of
±
3%.
Results obtained with the panel method agree particularly well, as shown in Figure 8, as
the values of the lift coefficients in the linear section of the curve are within less than a
1% difference. As a consequence of the larger drag predicted by the CFD simulations, the
aerodynamic efficiency is severely affected and decreases from a value of approximately 26
to a value of approximately 19, impacting the performance of the aircraft, as can be seen in
Figure 8.
As previously mentioned, the non-linear LLT is the only analytical method capable of
estimating the maximum lift coefficient and stall angle, providing an initial prediction and
a first level of validation comparing to the classical initial aerodynamic sizing described in
Section 3.2. As shown in Table 10, a 6% overprediction of maximum lift coefficient occurs
between the initially estimated value and that obtained with the non-linear LLT, a difference
that becomes more significant, in excess of 15%, if compared with the value predicted by
the CFD solution. On the other hand, the non-linear LLT and CFD results appear to be quite
consistent when predicting the maximum angle of attack, with only a 0.5
◦
difference, which
could be improved by decreasing the angular steps between
two simulations
currently

Aerospace 2022,9, 275 16 of 23
set at 0.25
◦
.The stall angle is in agreement with aircraft of similar size and comparable
configuration [48].
The prediction of the longitudinal static stability parameters appears to be highly
dependent of the solution methodology applied, although all models indicate a stable
aircraft, as
Cmα
is negative. Of the three analytical methods, the non-linear LLT and panel
methods predict a
Cm0=
0 and a stable aircraft up to
α
= 6
◦
. The pitching moment
coefficient obtained from CFD simulation shows a similar trend extending the limit of the
longitudinal stability to
α
= 8
◦
at the onset of stall and confirming a positive value of
Cm0
.
Post-stall, with a departure of the aerodynamic centre from the quarter chord due to flow
separation, the prediction of the longitudinal stability becomes more complex and therefore
the current values should be considered indicative
To fully assess the stall conditions of the aircraft, the phenomenon was evaluated in
more in detail by looking at the surface pressure fields on the top surface of the aircraft.
As shown in Figure 9, at an angle of attack of 6
◦
the flow is fully attached to the surface of
the wing, while the surface pressure distribution at 8
◦
indicates that the desired onset of
stall at mid-wing for keeping control surface effectiveness is achieved prior to separation
extending towards the wingtip. Finally at a 10
◦
angle of attack, the flow is fully stalled
over the outer section of the wing, but it remains locally attached on the fuselage of the
aircraft with clean flow allowing more design options for the integration of the propulsion
system at trailing section of the fuselage. Qualitatively, the stall behaviour is also shown
via plotted streamlines, in Figure 10.
Figure 9. Development of stall pattern.
Figure 10. Stall Pattern at α=8.5◦.
5. Power Requirements and Battery Definition
5.1. Power Required Analysis
Literature has shown [
10
–
25
] that existing UAVs have implemented propeller driven
propulsion as it is normally less energy demanding and therefore preferred for long en-
durance designs. Oppositely, work on BWB at full-aircraft scale [
30
–
33
] has favoured
configurations with air-breathing engines. As the propulsion system of the current design
is yet to be defined, both power and thrust required have been estimated. The use of

Aerospace 2022,9, 275 17 of 23
power in the calculations for the sizing of the fuel cell and extra battery do not indicate a
pre-selection of a propulsion system. The values expressed in Watts are used for consistency
in the calculations in Section 5.2
Preliminary estimations were completed using aerodynamic data obtained through
CFD simulations, at a velocity range between 5 m/s and 30 m/s [
46
] where the power
required curve is calculated through the components of the aircraft drag (Equation (12)):
PR=1
2ρ∞V3
∞SCD0+W2
1
2ρ∞V
πeAR . (12)
The thrust required [46] is, for the aircraft at zero angle of attack (Equation (13)):
TR=PR
V∞
. (13)
Preliminary power and thrust required curves are presented in Figure 11, where data
for velocities below 10 m/s are not available, as the aircraft is not capable of sustained flight
below that speed in clean wing configuration. Results indicate that, at the target velocity of
20 m/s, the aircraft requires 262 W of power, or 13 N of thrust, at a trim angle of 1.7
◦
to 2
◦
,
demonstrating that, for target flight conditions, sufficient excess power is available, even
when taking into consideration that these values are normally an underestimation as they
do not take the efficiency losses inherent to the propulsive system into consideration.
Figure 11. Power require (left) and thrust required (right) vs. Velocity.
5.2. Hybrid Hydrogen Fuel Cell/Battery System
Based on the power demand of the UAV discussed in Section 5.1, a typical 2 h mission
is designed and presented in Figure 12. It corresponds to a 10 min take-off at 1.2 kW,
100 min cruise at 600 W and 10 min landing procedure requiring 1 kW. A sizing strategy
using the power demand as an input for designing hybrid fuel cell/battery systems was
previously developed by the authors for underwater applications [
59
], and was adapted
for the UAV application. It finds the best, i.e., most compact or lightest, combination of
fuel cell stack and battery pack in order to fulfill the given mission. Results showed that a
hybrid system combining a 650 W Proton Exchange Membrane Fuel Cell (PEMFC), with a
tank containing 80 g pressured hydrogen, and a 100 Wh battery pack is the most suitable
(lightest) for the hybrid electric propulsion system. Only certain fuel cell power sizes were
considered, based on what is available commercially. As a comparison, this described
hybrid energy system weights around 5 kg, whereas, based on the assumptions from [
59
],
a full secondary battery system with the same amount of energy weights around 9 kg.
Parameters used for simulating a mission are shown in Table 11. The amount of
hydrogen left in the tank during the mission can be calculated using Equation (14), with the
fuel cell output power
PFC
, the Low Heating Value of hydrogen
LHVH2
at 25
°
C and the
fuel cell efficiency
ηFC
. As a result, the fuel cell delivers the base power while the battery
is used to reach high-power demands and is being recharged by the fuel cell during times

Aerospace 2022,9, 275 18 of 23
with low power demand. This configuration means that the flexibility of the UAV, in terms
of power and therefore speed, mostly depends on the battery pack.
mH2(t) = mH2(0)−Zt
0
PFC(t)dt
LHVH2×ηFC(t). (14)
Table 11. Electrochemical Parameters for the hybrid fuel cell and battery system.
Parameter Symbol Unit Source
Reaction enthalpy LHVH2119.96 MJ
kg
Thermodynamics
Battery efficiency ηbatt 0.90 - Assumed, [60]
Fuel cell efficiency ηFC 0.50 - Assumed, [49]
SoC lower limit SoClow 0.20 - Set value
SoC upper limit SoChigh 0.95 - Set value
The battery State-of-Charge (SoC) corresponds to the ratio between the energy content
of the battery pack at a given time, and its energy content when fully charged. It can be
estimated as shown in Equation (15), where
Pbatt
is the battery power (positive during
discharge and negative during charge),
Ebatt
is the energy content of the battery when fully
charged and
ηbatt
is the battery efficiency for charging or discharging. Going to very high
or very low SoC can accelerate the degradation of the battery [
61
], so to avoid degradation
the acceptable SoC-window was therefore set to 20%
≤
SoC
≤
95% in the simulations. To
compare different charge and discharge rates the charging rate,
Crate
, is often used. A
Crate
of 1C means that it takes one hour to fully charge the battery from a SoC of 0%, and 2C
means it takes 30 min. The maximum
Crate
reached during the mission is estimated using
Equation (16).
SoC(t) = SoC(0)−ηbatt Rt
0Pbatt(t)dt
Ebatt (15)
Crate =max(|Pbatt|)
Ebatt . (16)
A procedure for checking if the mission can be fulfilled with the chosen fuel cell,
hydrogen storage and battery size was developed by the Chiche et al. [
60
]. If the hydrogen
tank is depleted or the battery SoC goes outside its chosen window at any time then
the mission is considered unsuccessful. It also considers that when the battery SoC is at
95%, charging stops. Results, in Figure 12, show that the mission is fulfilled. In addition,
the battery SoC goes from being at 95% SoC to 55% during the take-off phase and is then
recharged up to 90% during the cruise period. The landing phase also deplete battery power,
when the SoC decreases to 50% at the end of the mission. Furthermore, the maximum
Crate
reached during the mission is 5.5, meaning that the battery will have to be power-
optimized. There is no clear boundary for a high
Crate
, however, a
Crate >
1 is usually
considered high. Most batteries can handle higher discharge currents than charge currents,
so it is almost always the charge step that is limiting for battery design and selection [
62
].
A power-optimized battery can handle relatively high
Crate
, but generally has a reduced
energy density. It is interesting to note that the fuel cell power remains constant at 650 W,
in other words, the fuel cell continuously works at steady-state at a high efficiency point.
At the end of the mission, there is 8 g of hydrogen left in the tank.

Aerospace 2022,9, 275 19 of 23
0 20 40 60 80 100 120
Time [min]
600
800
1000
1200
Power [W]
0
0.2
0.4
0.6
0.8
1
Battery SoC
Power Profile
FC nominal power
Battery SoC ==>
Figure 12.
Electric power profile and estimated battery SoC of the typical mission consisting of three
phases, 10 min take off, 100 min cruising and 10 min landing. The simulation considers a fuel cell
stack with 650 W nominal power, 80 g hydrogen, and a 100 Wh battery with an initial SoC at 95%.
6. Future Work
Work is currently ongoing for the continuous development of the aircraft, as well as
targeting specific issues encountered in the conceptual phase of the design.The work is a
combination of applied research within the frame of the educational and research emphasis
of the project itself.
Further aerodynamic work is being conducted focusing on the improvement of the stall
characteristics of the aircraft—namely to delay it to higher angles of attack and therefore
increase the maximum lift coefficient—through the use of passive flow control. This work
also focuses on a more detailed comparison of low-fidelity and CFD simulations towards a
more robust design process when non-conventional configurations are concerned.
As noticeable in the literature [
10
–
25
], the overwhelming majority of UAVs powered
by fuel cells are propeller aircraft as they are less energy demanding, while future aircraft
of the BWB design are expected to be powered by air-breathing engines [
31
]. Consequently,
detailed work is being carried out comparing the use of electrical ducted fans (EDF) and
brushless-motors with propellers, to determine the number of motors needed for each
configuration and the propeller size. Characterization of the performance of an acquired
fuel cell from those evaluated in the literature and that are commercially available [
37
,
38
] is
being carried out to improve the power analysis, and will be followed by bench tests of the
full propulsion architecture with additional focus on energy management.
The design process is also transitioning to ground testing first through the planning of
a preliminary wind tunnel campaign to validate the aerodynamic loads obtained through
CFD simulations and improve the understanding of the stability of the aircraft by testing for
both angle of attack and sideslip/yaw angle. These tests will enable the full understanding
of the aerodynamic and stability characteristics of the aircraft and provide information for
future development of the aircraft.
Data obtained experimentally will be imported in commercially available flight simu-
lation software to study further the behaviour of the aircraft, coupled with the design and
manufacturing of a geometrically representative sub-scale version for flight testing.
The results from simulations and experimental campaigns will provide the neces-
sary foundation knowledge for any redesign prior to commencing the final phase of the
development involving the structural design and manufacturing of the full scale aircraft.

Aerospace 2022,9, 275 20 of 23
7. Conclusions
The present work focuses on the conceptual design of a hybrid-electric, fuel-cell
powered blended wing body UAV representative of a future aircraft configuration. A
simple mission was defined for the aircraft to act as a test platform for data acquisition,
with a target endurance of 60 min. Future work covering all aspects of aircraft design
has been discussed showing the broad scope of the project. The platform is to act as the
first stepping stone for the development of a new research topic at KTH, Royal Institute
of Technology.
Design constraints were based on existing, successfully flown, fuel-cell powered
UAVs, leading to sizing constraints of a MTOM of 25 kg, a wing span of 4 m, and a cruise
velocity of 20 m/s. The technical specifications of the UAV and of the desired mission,
the multidisciplinarity aspects of the work, and the long-term emphasis and perspective
on novel configurations for future aviation in combination to the efforts in embedding
the project within the education curricula at KTH sets the current project apart from the
majority of existing work reported in the literature.
A classical design approach was initially implemented supported by recognized
analytical tools for aircraft design such as XFOIL, OpenVSP, and xflr5 for rapid design
iterations. The aerodynamic analysis was enhanced using CFD methods for improved
understanding of the performance of the aircraft. These results indicate that the classical
design approach has limitations when departing from a traditional aircraft configuration
and emphasize the need to implement more advanced methods and tools already at early
stages of the design process, as also previously shown by
Panagiotou et al. [36]
. At the
same time, this classical approach can still provide a quick “rule of thumb” approximation
and starting point for the design process.
The current design features a fuselage and wing section with reflex airfoils for natural
longitudinal stability, and a wing sweep of 30
◦
. Dihedral and sweep angles were set to
optimise the aerodynamic efficiency in cruise and the natural lateral stability of the aircraft.
An aerodynamic efficiency of 19 was achieved, and a maximum coefficient of lift of 0.789 at
8.5
◦
. Initial longitudinal stability analysis shows a longitudinally stable aircraft between
−
4
◦
and 8
◦
, according to results from CFD simulations. Post-stall pitching moment is more
difficult to predict as a result of the departure of the aerodynamic centre from quarter chord
due to flow separation, and therefore data should be taken as indicative.
Power calculations demonstrated that a 650 W fuel-cell, complemented by a tank of
80 g of compressed hydrogen and a 100 Wh lithium-ion battery is capable of completing
the mission required by the specifications.
Author Contributions:
Data curation, S.S., A.P., A.O., H.G., A.C. and R.M.; Formal analysis, S.S.,
A.P., A.O. and R.M.; Investigation, S.S.; Methodology, H.G. and A.C.; Project administration, R.M.;
Supervision, R.M. and G.L.; Writing—original draft, A.C. and R.M.; Writing—review & editing, H.G.
and G.L. All authors have read and agreed to the published version of the manuscript.
Funding:
The project has been funded by the KTH Integrated Transport Research Lab, KTH In-
dustry Transformation Platform, the KTH Energy Platform, KTH XPRES, and KTH Excellenta
Utnildningsmiljöer (KTH Excellence in Education.
Institutional Review Board Statement: Not Applicable.
Informed Consent Statement: Not Applicable.
Data Availability Statement: Data is contained within the article.
Acknowledgments:
Malin Åkermo for financial support through XPRES, Selma Rahman and Mattias
Olausson for providing high-quality CAD models of the aircraft, which were used in the CFD
simulations, and finally Alessio Galfione and Enrico Trevisan for the engineering drawing.
Conflicts of Interest:
The authors confirm that there are no potential or existing conflict of interests
associated with their work or the funding.

Aerospace 2022,9, 275 21 of 23
References
1.
Centracchio, F.; Rossetti, M.; Iemma, U. Approach to the weight estimation in the conceptual design of hybrid-electric-powered
unconventional regional aircraft. J. Adv. Transp. 2018,2018, 1–15. [CrossRef]
2.
Hordé, T.; Achard, P.; Metkemeijer, R. PEMFC Application for aviation: Experimental and numerical study of sensitivity to
altitude. Int. J. Hydrogen Energy 2012,37, 10818–10829. [CrossRef]
3.
Winter, M.; Brodd, R.J. What Are Batteries, Fuel Cells, and Supercapacitors? J. Chem. Rev.
2003
,104, 4245–4269. [CrossRef]
[PubMed]
4.
Baroutaji, A.; Wilberforce, T.; Ramadan, M.; Ghani Olabi, A. Comprehensive investigation on hydrogen and fuel cell technology
in the aviation and aerospace sectors. Renew. Sustain. Energy Rev. 2019,106, 31–40. [CrossRef]
5.
Peters, R.; Samsun, R.C. Evaluation of multifunctional fuel cell systems in aviation using a multistep process analysis methodology.
Appl. Energy 2013,111, 46–63. [CrossRef]
6.
Guida, D.; Muntillo, M. Design methodology for a PEM fuel cell power system in a more electrical aircraft. Appl. Energy
2017
,192,
446–456. [CrossRef]
7.
Romeo, G.; Borello, F.; Correa, G. Setup and test flights of all-electric two-seater aeroplane powered by fuel cells. J. Aircr.
2011
,48,
1331–1341. [CrossRef]
8.
Romeo, G.; Borello, F.; Correa, G.; Cestino, E. ENFICA-FC: Design of transport aircraft powered by fuel cell & flight test of zero
emission 2-seater aircraft powered by fuel cells fueled by hydrogen. Int. J. Hydrogen Energy 2013,28, 469–479.
9. Barrett, S. ZeroAvia completes first hydrogen-electric passenger plane flight. Fuel Cells Bull. 2020, 6.
10.
Gong, A.; Verstraete, D. Fuel cell propulsion in small fixed-wing unmanned aerial vehicles: Current status and research needs.
Int. J. Hydrogen Energy 2017,42, 21311–21333. [CrossRef]
11.
Swider-Lyons, K.E.; MacKrell, J.A.; Rodgers, J.A.; Page, G.S.; Schuette, M.; Stroman, R.O. Hydrogen fuel cell propulsion for long
endurance small UAVs. In Proceedings of the Name of the Conference,AIAA Centennial of Naval Aviation Forum “100 Years of
Achievement and Progress”, Virginia Beach, VA, USA, 21–22 September 2011.
12.
Stroman, R.O.; Schuette, M.W.; Swider-Lyons, K.; Rodgers, J.A.; Edwards, D.J. Liquid hydrogen fuel system design and
demonstration in a small long endurance air vehicle. Int. J. Hydrogen Energy 2014,39, 11279–11290. [CrossRef]
13.
Bradley, T.H.; Moffitt, B.A.; Mavris, D.M.; Parekh, D.E. Development and experimental characterization of a fuel cell powered
aircraft. J. Power Sources 2007,39, 11279–11290. [CrossRef]
14.
Rhoads, G.D.; Wagner, N.A.; Taylor, B.J.; Keen, D.B.; Bradley, T.H. Design and flight test results for a 24 h fuel cell unmanned air
vehicle. In Proceedings of the 8th Annual International Energy Conversion Engineering Conference, Nashville, TN, USA, 25–28
July 2010.
15.
Herwerth, C.; Chiang, C.; Ko, A.; Matsuyama, S.; Choi, S.B.; Mirmirani, M.; Gamble, D.; Paul, R.; Sanchez, V.; Arena, A.; et al.
Development of a Small Long Endurance Hybrid PEM Fuel Cell Powered UAV; Aerospace Technology Conference and Exposition, SAE
Technical Paper; SAE International: Warrendale, PE, USA, 2007.
16.
Gong, A.; Palmer, J.L.; Brian, G.; Harvey, J.R.; Verstraete, D. Performance of a hybrid, fuel-cell-based power system during
simulated small unmanned aircraft missions. Int. J. Hydrogen Energy 2016,41, 11418–11426. [CrossRef]
17.
Gang, B.G.; Kim, H.; Kwon, S. Ground simulation of a hybrid power strategy using fuel cells and solar cells for high-endurance
unmanned aerial vehicles. Energy 2017,141, 1547–1554. [CrossRef]
18.
Kim, K.; Kim, T.; Kwon, S. Fuel cell systems with sodium borohydride as hydrogen source for unmanned aerial vehicles. J. Power
Sources 2011,196, 9069–9075. [CrossRef]
19.
Lapena-Rey, N.; Blanco, J.A.; Ferreyra, E.; Lemus, J.L.; Pereira, S.; Serrot, E. A fuel cell powered unmanned aerial vehicle for low
altitude surveillance missions. Int. J. Hydrogen Power 2017,42, 6926–6940. [CrossRef]
20.
Gadalla, M.; Zafar, S. Analysis of a hydrogen fuel cell-PV power system for small UAV. Int. J. Hydrogen Energy
2016
,41, 6422–6432.
[CrossRef]
21.
Renau, J.; Lozano, A.; Barroso, J.; Miralles, J.; Martín, J.; Sánchez, F.; Barreras, F. Use of fuel cell stacks to achieve high altitudes in
light unmanned aerial vehicles. Int. J. Hydrogen Energy 2015,40, 14573–14583. [CrossRef]
22.
Depcik, C.; Cassady, T.; Collicott, B.; Burugupally, S.P.; Li, X.; Alam, S.S.; Arandia, J.R.; Hobeck, J. Comparison of lithium ion
batteries, hydrogen fueled combustion engines, and a hydrogen fuel cell in powering a small unmanned aerial vehicle. Energy
Convers. Manag. 2020,207, 112514. [CrossRef]
23.
Verstraete, D.; Lehmkuehler, K.; Wong, K.C. Design of a Fuell Cell Powered Blended Wing Body UAV. In Proceedings of the
ASME 2012 International Mechanical Engineering Conference & Exposition, IMECE 2012, Houston, TX, USA, 9–15 November
2012; IMECE2012-88871.
24.
Özbek, E.; Yalin, G.; Ekici, S.; Karakoc Hikmet, T. Evaluation of design methodology, limitations, and iterations of a hydrogen
fuelled hybrid fuel cell mini UAV. Energy 2020,213, 118757. [CrossRef]
25.
Savvaris, A.; Xie, Y.; Malandrakis, K.; Lopez, M.; Tsourdos, A. Development of a fuel cell hybrid-powered unmanned aerial vehicle.
In Proceedings of the 24th Mediterranean Conference on Control and Automation (MED), Athens, Greece, 21–24 June 2016.
26.
Project for an Unmanned Airplane with Zero Emissions to cross the South Atlantic. Available online: https://www.isae-supaero.
fr/en/news/project-for-an-unmanned-airplane-with-zero-emissions-to-cross-the-south/ (accessed on 2 March 2021).
27.
De Wagter, C.; Remes, B.; Smeur, E.; van Tienen, F.; Ruijsink, R.; van Hecke, K.; van der Horst, E. The NederDrone: A hybrid lift,
hybrid energy hydrogen UAV. Int. J. Hydrogen Energy 2021,46, 16003–16018. [CrossRef]

Aerospace 2022,9, 275 22 of 23
28. Project Phoenix. Available online: https://aerodelft.nl/project-phoenix/ (accessed on 12 May 2022).
29. DELAIR. Available online: https://delair.aero/delair-commercial-drones/hydrone/ (accessed on 7 September 2021).
30.
Guynn, M.D.; Freeh, J.E.; Olson, E.D. Evaluation of a Hydrogen Fuel Cell Powered Blended-Wing-Body Aircraft Concept for
Reduced Noise and Emissions. In NASA/TM-2004-212989; National Aeronautics and Space Administration: Washington, DC,
USA, 2004.
31.
ZEROe, towards the World’s First Zero-Emission Commercial Aircraft. Available online: https://www.airbus.com/innovation/
zero-emission/hydrogen/zeroe.html (accessed on 7 September 2021).
32.
Reist, T.A.; Zingg, D.W. Aerodynamic design of blended wing body and lifting-fuselage aircraft. In Proceedings of the 34th AIAA
Applied Aerodynamics Conference, Session: Innovative Aerodynamic Concepts & Designs, Washington, DC, USA, 13–17 June
2016; AIAA 2016-3874.
33.
Bradley, K.R. A sizing methodology for the conceptual design of blended-wing-body transports. In NASA/CR-2004-213016;
National Aeronautics and Space Administration: Washington, DC, USA, 2004.
34.
Crawley, E.F.; Malmqvist, J.; Östlund, S.; Brodeur, D.R.; Edström, K. Rethinking Engineering Education. The CDIO Approach, 2nd ed.;
Springer: Berlin/Heidelberg, Germany, 2014.
35.
Raymer, D.P. Aircraft Design: A Conceptual Approach, 6th ed.; AIAA Educational Series; AIAA: Reston, VA, USA, 2018; pp. 27–94;
discussion 834–838.
36.
Panagiotou, P.; Fotiadis-Karras, S.; Yakinthos, K. Conceptual design of a Blended Wing Body MALE UAV. Aerosp. Sci. Technol.
2018,73, 32–47. [CrossRef]
37. HES Energy Systems. Available online: https://www.hes.sg/ (accessed on 7 September 2021).
38. Intelligent Energy. Available online: https://www.intelligent-energy.com/ (accessed on 7 September 2021).
39.
Gundlach, J. Designing Unmanned Aircraft Systems: A Comprehensive Approach, 3rd ed.; AIAA Educational Series; AIAA: Reston,
VA, USA, 2012.
40. Liebeck, R.H. Design of the blended wing body subsonic transport. J. Aircr. 2004,41, 10–25. [CrossRef]
41.
Qin, N.; Vavalle, A.; Le Moigne, A.; Laban, M.; Hackett, K.; Weinerfelt, P. Aerodynamic considerations of blended wing body
aircraft. Prog. Aerosp. Sci. 2004,40, 321–343. [CrossRef]
42.
Qin, N.; Vavalle, A.; Le Moigne, A. Spanwise lift distribution for blended wing body aircraft. J. Aircraft
2005
,42, 356–365.
[CrossRef]
43. Airfoil Tools. Available online: http://airfoiltools.com/search/index (accessed on 7 September 2021).
44.
Drela, M. XFOIL: An Analysis and Design System for Low Reynolds Number Airfoils. In Lectures Noted in Engineering, Proceedings
of the Conference in Low Reynolds Number Aerodynamics, Notre Dame, IN, USA, 5–7 June 1989; Mueller, T.J., Ed.; Springer: Berlin,
Germany, 1989.
45. Anderson, J.D. Aircraft Performance and Design; McGraw-Hill Int. Editions: Singapore, 1999.
46. Pamadi, B.N. Performance, Stability, Dynamics and Control of Airplanes; AIAA Educational Series; AIAA: Reston, VA, USA, 1998.
47.
Panagiotou, P.; Yakinthos, K. Parametric aerodynamic study of Blended-Wing-Body platforms at low subsonic speeds for UAV
applications. In Proceedings of the 35th AIAA Applied Aerodynamics Conference, AIAA AVIATION Forum, Denver, CO, USA,
5–9 June 2017; AIAA 2017-3737.
48.
Panagiotou, P.; Kaparos, P.; Salpingidou, C.; Yakinthos, K. Aerodynamic design of a MALE UAV. Aerosp. Sci. Technol.
2016
,50,
127–138. [CrossRef]
49.
Koster, J.; Balaban, S.; Hillery, D.; Hmbargar, C.; Nasso, D.; Serani, E.; Velazco, A. Design of a Blended Wing Body UAS with
Hybrid Propulsion. In Proceedings of the ASME 2011 International Mechanical Engineering Congress & Exposition, IMECE2011,
Denver, CO, USA, 11–17 November 2011; IMECE2011-62126.
50.
McDonald, R.A.; Gloudemans, J.R. Open Vehicle Sketch Pad: An Open Source Parametric Geometry and Analysis Tool for
Conceptual Aircraft Design. In Proceedings of the AIAA SCITECH 2022 Forum, San Diego, CA, USA, 3–7 January 2022.
51. xflr5. Available online: http://www.xflr5.tech/xflr5.htm (accessed on 7 September 2021).
52. Tomac, M.; stenfelt, G. Predictions of stability and control for a flying wing. Aerosp. Sci. Technol. 2014,39, 179–186. [CrossRef]
53.
Menter, F.R. Two-equation eddy-viscosity turbulence models for engineering applications. AIAA J.
1994
,32, 1598–1605. [CrossRef]
54.
Douvi, E.C.; Tsavalos, I.A.; Margaris, P.D. Evaluation of the turbulence models for the simulation of the flow over a National
Advisory Committee for Aeronautics (NACA) 0012 airfoil. J. Mech. Eng. Res. 2012,4, 100–111.
55.
Menter, F.R.; Kuntz, M.; Langtry, R. Ten Years of Industrial Experience with the SST Turbulence Model. Turbul. Heat Mass Transf.
2003,4, 625–632.
56.
Spalart, P.R.; Rumsey, C.L. Effective inflow conditions for Turbulence Models in Aerodynamic calculations by Spalart and Rumsey.
AIAA J. 2007,45, 2544–2553. [CrossRef]
57.
Goetten, F.; Finger, D.F.; Marino, M.; Bil, C. A revire of guidelines and best practices for subsonic aerodynamic simulations unsing
RANS CFD. In Proceedings of the Asia Pacific International Symposium on Aerospace Technology, Gold Coast, Australia, 4–6
December 2019; ISBN 978-1-925627-40-4.
58.
Versteeg, H.; Malalasekera, W. An Introduction to Computational Fluid Dynamics: The Finite Volume Method, 2nd ed.; Pearson:
London, UK, 2007.
59.
Chiche, A.; Lindbergh, G.; Stenius, I.; Lagergren, C. A Strategy for Sizing and Optimizing the Energy System on Long-Range
AUVs. IEEE J. Ocean. Eng. 2021, 1132–1143. [CrossRef]

Aerospace 2022,9, 275 23 of 23
60.
Chiche, A.; Andruetto, C.; Lagergren, C.; Lindbergh, G.; Stenius, I.; Peretti, L. Feasibility and impact of a Swedish fuel cell-powered
rescue boat. Ocean. Eng. 2021,234, 109259. [CrossRef]
61. Wikner, E.; Thiringer, T. Extending Battery Lifetime by Avoiding High SOC. Appl. Sci. 2018,8, 1–16. [CrossRef]
62.
Sturm, J.; Frank, A.; Rheinfeld, A.; Erhard, S.V.; Jossen, A. Impact of electrode and cell design on fast charging capabilities of
cylindrical lithium-ion batteries. J. Electrochem. Soc. 2020,167, 130505. [CrossRef]