Design and Analysis of The Control and Stability of A Blended Wing Body Aircraft (PDFDrive)

Design and Analysis of the Control
and Stability of a Blended Wing

Body Aircraft
Roberto Merino Martínez

Supervisor: Arthur Rizzi
Royal Institute of Technology (KTH)

Stockholm, Sweden
May 2014
Abstract
Future aircraft generations are required to have higher performance and capacities.
This achievement should be fulfilled with the minimum cost and environmental
impact. This calls for the design of new unconventional configurations, such as the
Blended Wing Body (BWB), a tailless aircraft which integrates the wing and the
fuselage into a single lifting surface. It has been proven in previously published
works that this concept is feasible, has an efficient economical performance and
is a promising candidate for solving the current air traffic problems, despite its
challenging control and stability features. Moreover, the size of the vertical sur-
faces, such as the winglets, condition the radar detectability of the BWB model,
especially for military missions. The goal of the department of Aeronautical and
Vehicle Engineering at the Royal Institute of Technology (KTH) and of the de-
partment of Air Transport Systems of the German Aerospace Centre (DLR) is to
investigate new ways to improve the conceptual design process of the aircrafts in
a multidisciplinary environment. In order to design future unconventional aircraft
configurations (such as the Blended Wing Body), the CEASIOM (Computerised
Environment for Aircraft Synthesis and Integrated Optimisation Methods) geom-
etry module, AcBuilder, is replaced and enhanced by implementing the Common
Parametric Aircraft Configuration Scheme (CPACS), developed by the DLR as
a basis technology. CPACS is meant to become a unified software framework to
allow the sharing of the work and information, making it accessible for every per-
son. It requires an implementation of the software modules in a framework using
a common language for all the tools, in order to make later alterations of this
framework easier. A detailed research of the latest developments and advances
in the BWB concept was performed in order to identify the main principles and
best design options. Afterwards, by using the implemented improved tool CPAC-
SCreator (CC) based on CPACS, instead of Acbuilder, a BWB aircraft baseline
was designed. The aerodynamic behaviour and performance of this model were
v
then analyzed with the aid of the improved CEASIOM platform, with an special
emphasis on its control and stability features. This analysis enables to improve
the baseline design and the allocation and size of the control surfaces was studied
and optimized. The minimum winglet required for a target flight performance was
identified, due to its importance for reducing the drag and the radar detectability
of the aircraft.
vi
Preface
This Master Thesis was accomplished during my Erasmus course at the Royal
Institute of Technology (KTH) in Stockholm. This experience was definetely
enriching and helped me to complement my studies at the Technical University
of Madrid (UPM), in the “Escuela Técnica Superior de Ingenieros Aeronáuticos”.
First of all I would like to address special thanks to my supervisor Professor
Arthur Rizzi for giving me the opportunity to carry out my Master Thesis in his
department and for his patience and all his support and helpful discussions. I
would also like to kindly thank Professor Jesper Oppelstrup for his helpful advice
and interesting conversations. Moreover, I am very thankful to Mengmeng, Mio
and Cong, also from the department of Aeronautical and Vehicle Engineering at
KTH, for their help and useful discussions.
I would like to express my sincere gratitude to my family, especially to my
parents and to my brother, who where always there to encourage me and making
everything easier to me. Without them this work would not have been possible.
Last but not least, I would like to acknowledge the presence of all the wonder-
ful people I have met this year in Sweden, specially my girlfriend Lenka, whose
understanding and motivation were very important for me.
Email of the author: robertomerinomartinez@gmail.com
vii
viii
Contents
1 Introduction 1
1.1 Motivation for this work . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Crucial and distinguishing aspects of the BWB configuration . . . 3
1.3 Aircraft design process . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 A new design solution . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Objectives of this work . . . . . . . . . . . . . . . . . . . . . . . . 5
1.6 Thesis breakdown . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Aircraft conceptual design 9

2.1 Traditional methods . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2 Computational tools for conceptual design: CEASIOM . . . . . . 16
2.2.1 Coordinate systems and control sign conventions . . . . . . 16
2.2.2 AcBuilder (Geometry) . . . . . . . . . . . . . . . . . . . . 18
2.2.3 SUMO (Geometry mesher) . . . . . . . . . . . . . . . . . . 21
2.2.4 Weight & Balance . . . . . . . . . . . . . . . . . . . . . . . 21
2.2.5 AMB-CFD (Aerodynamics) . . . . . . . . . . . . . . . . . 23
2.2.6 Propulsion . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.2.7 SDSA (Stability and Control) . . . . . . . . . . . . . . . . 27
2.2.8 NeoCASS (Aeroelasticity) . . . . . . . . . . . . . . . . . . 33
2.3 CPACS, a new solution . . . . . . . . . . . . . . . . . . . . . . . . 35
2.3.1 CPACS XML file structure . . . . . . . . . . . . . . . . . . 38
2.3.2 CPACSCreator . . . . . . . . . . . . . . . . . . . . . . . . 44
3 The Blended Wing Body concept 51

3.1 Main features of the BWB . . . . . . . . . . . . . . . . . . . . . . 53
3.2 Previous models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.2.1 Horten Ho 229 . . . . . . . . . . . . . . . . . . . . . . . . 57
ix
CONTENTS
3.2.2 Northrop YB-49 . . . . . . . . . . . . . . . . . . . . . . . . 58

3.2.3 MOB BWB . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.2.4 SAX-40 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.2.5 Cranfield University BW11 Eagle Ray . . . . . . . . . . . 62
3.2.6 NASA-Boeing X48-B . . . . . . . . . . . . . . . . . . . . . 63
3.2.7 Main features in common . . . . . . . . . . . . . . . . . . 64
3.3 ELSA BWB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4 Definition of the baseline BWB geometry 67

4.1 Definition in CPACSCreator . . . . . . . . . . . . . . . . . . . . . 67
4.1.1 Wing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.1.2 Fuselage . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.1.3 Control surfaces . . . . . . . . . . . . . . . . . . . . . . . . 79
4.1.4 Engines . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.2 Geometry refinement using CATIA and SUMO . . . . . . . . . . 83
4.3 Final model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5 Generation of aerodynamic data 91

5.1 TORNADO fundamentals . . . . . . . . . . . . . . . . . . . . . . 92
5.2 B-747 example . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.2.1 Description of the B-747 model . . . . . . . . . . . . . . . 95
5.2.2 Validation of aerodynamic sources . . . . . . . . . . . . . . 97
5.2.2.1 Low speed aerodynamics (M=0.15) . . . . . . . . 98
5.2.2.2 High speed aerodynamics . . . . . . . . . . . . . 100
5.2.2.3 Stability analysis . . . . . . . . . . . . . . . . . . 103
5.2.2.4 Linear mode analysis . . . . . . . . . . . . . . . . 104
5.2.3 Conclussions . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.3 TORNADO results . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.4 XFLR5 results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5.5 Comparison of the two tools . . . . . . . . . . . . . . . . . . . . . 115
5.6 Parasite drag estimation . . . . . . . . . . . . . . . . . . . . . . . 116
6 Analysis of the stability and control of the BWB model 125

6.1 Theoretical introduction . . . . . . . . . . . . . . . . . . . . . . . 125
6.1.1 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6.1.1.1 Static stability . . . . . . . . . . . . . . . . . . . 125
x
CONTENTS
6.1.1.2 Dynamic stability . . . . . . . . . . . . . . . . . . 128

6.1.2 Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
6.2 Stability analysis of the BWB baseline . . . . . . . . . . . . . . . 136
6.2.1 Static stability of the BWB baseline . . . . . . . . . . . . 137
6.2.2 Dynamic stability of the BWB baseline . . . . . . . . . . . 138
6.2.2.1 Stability for the longitudinal motion . . . . . . . 138
6.2.2.2 Stability for the lateral motion . . . . . . . . . . 141
6.3 Sizing of the control surfaces and consequences . . . . . . . . . . . 146
6.4 Sizing of the winglets and rudders . . . . . . . . . . . . . . . . . 148
6.4.1 Configuration with no winglets and no rudders . . . . . . . 153
6.5 Final model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
6.6 Animations of the motions . . . . . . . . . . . . . . . . . . . . . . 155
7 Conclusions 157
7.1 Main achievements . . . . . . . . . . . . . . . . . . . . . . . . . . 158
7.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
xi
CONTENTS
xii
List of Figures
1.1 Worldwide air traffic evolution prediction from 2012 to 2032. [1] . 1
1.2 Brent Crude Oil and US Gulf Coast Jet Fuel average prize between
2000 and 2013. [1] . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Boeing X-48B Blended Wing Body developed by NASA. . . . . . 3
2.1 The different stages in the creation process of an aircraft. . . . . . 10

2.2 The new virtual aircraft design method. . . . . . . . . . . . . . . 11
2.3 Table of analysis of disciplines and tools depending on the level of
fidelity required for aircraft design, according to Nickol [7]. . . . . 12
2.4 The scheme of the traditional conceptual design process.[10] . . . 14
2.5 Comparison between the classical design methods (a) and the use of
CEASIOM (b). As illustrated, CEASIOM provides higher fidelity
earlier in the design process, saving time and cost. [46] . . . . . . 15
2.6 CEASIOM software with core modules: Geo-sumo, AMB-CFD,
NeoCASS, S&C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.7 Different coordinate systems used in CEASIOM. [12] . . . . . . . 18
2.8 Positive control surfaces deflections convention for CEASIOM.[12] 19
2.9 AcBuilder data flow including the user interface. . . . . . . . . . . 20
2.10 Sketch of the Transonic Cruiser TCR geometry in the Weights and
Balance section of the old AcBuilder module. . . . . . . . . . . . . 20
2.11 Example of mesh generated using SUMO. . . . . . . . . . . . . . 22
2.12 Volume mesh of the Transonic Cruiser TCR geometry using TetGen
and SUMO. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.13 Graphical view of the different centres of gravity of the Transonic
Cruiser TCR. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.14 Interface between the AMB module and a generic CFD solver. . . 24
xiii
LIST OF FIGURES
2.15 Adaptable fidelity options within the AMB-CFD module in CEA-

SIOM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.16 Adaptative fidelity geometry modelling requirements in CEASIOM. 27
2.17 Different levels of complexity geometry requirements for different
CFD tools fidelity levels. [15] . . . . . . . . . . . . . . . . . . . . 28
2.18 SDSA Coordinate systems to describe the position and the attitude
of the aircraft. [16] . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.19 SDSA Velocity coordinate system. [16] . . . . . . . . . . . . . . . 29
2.20 SDSA Stability analysis scheme. . . . . . . . . . . . . . . . . . . 31
2.21 SDSA flight simulation scheme. . . . . . . . . . . . . . . . . . . . 32
2.22 SDSA structure and functionality. . . . . . . . . . . . . . . . . . . 33
2.23 NeoCASS structure and data flow. [18] . . . . . . . . . . . . . . . 35
2.24 A design environment includes a common language, several analysis
modules and an integration framework. . . . . . . . . . . . . . . 36
2.25 Advantages of unified data management, such as CPACS, in a
multi-module framework. On the left (a), a framework without
unified data and N to N interfaces = N(N-1) interfaces. On the
right (b), a framework with unified data and a hub with N to 1
interfaces = 2N interfaces. . . . . . . . . . . . . . . . . . . . . . . 37
2.26 CPACS XML file hierarchical structure. . . . . . . . . . . . . . . . 39
2.27 CPACS subelement structure. . . . . . . . . . . . . . . . . . . . . 41
2.28 CPACS wing elements definition. [19] . . . . . . . . . . . . . . . . 42
2.29 CPACS wing sections definition. [19] . . . . . . . . . . . . . . . . 43
2.30 CPACS wing segments definition. [19] . . . . . . . . . . . . . . . . 44
2.31 CPACS wing component segment definition. [19] . . . . . . . . . . 45
2.32 CPACS global geometry coordinate system. [19] . . . . . . . . . . 45
2.33 CPACS fuselage elements definition. . . . . . . . . . . . . . . . . . 46
2.34 CPACS fuselage sections definition. . . . . . . . . . . . . . . . . . 47
2.35 CPACS fuselage segments definition. . . . . . . . . . . . . . . . . 47
2.36 CPACSCreator user interface for an example fuselage definition. . 48
2.37 CPACSCreator data flow scheme. . . . . . . . . . . . . . . . . . . 48
2.38 Scheme of modules using CC instead of AcBuilder. . . . . . . . . 49
2.39 Examples of unconventional aircraft configurations that can be cre-
ated using CPACSCreator: Box Wing concept (left) and Strut
Braced Wing concept (right) . . . . . . . . . . . . . . . . . . . . . 49
xiv
LIST OF FIGURES
3.1 Aircraft design evolution. From left to right: the Writght Flyer,
the Boeing B-47, the Airbus A330)[20] . . . . . . . . . . . . . . . 52
3.2 Cannonical forms and the effect of body type on surface area.[20] 52
3.3 Surface difference between the conventional configuration and the
BWB configuration.[20] . . . . . . . . . . . . . . . . . . . . . . . . 53
3.4 Comparison of aerodynamic, inertial and cabin pressure loads for
a conventional configuration (left) and a BWB (right)[20] . . . . . 53
3.5 Potential benefits of the BWB configuration compared to conven-
tional aircrafts in 2015. [23] . . . . . . . . . . . . . . . . . . . . . 55
3.6 Horten Ho 229 illustration (a) and sketch (b) . . . . . . . . . . . . 58
3.7 Northrop YB-49 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.8 Northrop Grumman B-2 Spirit . . . . . . . . . . . . . . . . . . . . 60
3.9 CAD illustration of the MOB BWB configuration [22] . . . . . . . 60
3.10 MOB BWB pressure distribution simulation. [58] . . . . . . . . . 61
3.11 Cambridge University and MIT SAX-40 BWB model. [60] . . . . 62
3.12 Cranfield University BW11 Eagle Ray . . . . . . . . . . . . . . . . 63
3.13 Cranfield University BW11 Eagle Ray passenger distribution sketch. 63
3.14 X48-B BWB from NASA . . . . . . . . . . . . . . . . . . . . . . . 64
3.15 The ELSA BWB aeroelastic wind tunnel model. [5] . . . . . . . . 65
4.1 Martin Carlsson’s ELSA BWB sketch for wind tunnel testing in
reference [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.2 NACA 0017 airfoil. . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.3 Polar of the NACA 0017 airfoil . . . . . . . . . . . . . . . . . . . 69
4.4 NACA 0017 graphics showing from left to right: CL vs α, CL /CD
vs α and CL /CD vs CL . . . . . . . . . . . . . . . . . . . . . . . . 70
4.5 NACA 0012 airfoil . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.6 Polar of the NACA 0012 airfoil . . . . . . . . . . . . . . . . . . . 71
4.7 NACA 0012 graphics showing from left to right: CL vs α, CL /CD
vs α and CL /CD vs CL . . . . . . . . . . . . . . . . . . . . . . . . 71
4.8 N2Asta072 airfoil . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.9 Polar of the N2Asta072 airfoil . . . . . . . . . . . . . . . . . . . . 72
4.10 N2Asta072 graphics showing from left to right: CL vs α, CL /CD vs
α and CL /CD vs CL . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.11 Comparison between N2Asta072 (blue) and NACA 0017 (red) . . 73
xv
LIST OF FIGURES
4.12 Wortmann fx 60-126 airfoil . . . . . . . . . . . . . . . . . . . . . . 74

4.13 Polar of the Wortmann fx 60-126 airfoil . . . . . . . . . . . . . . . 74
4.14 Wortmann fx 60-126 graphics showing from left to right: CL vs α,
CL /CD vs α and CL /CD vs CL . . . . . . . . . . . . . . . . . . . 75
4.15 Comparison between Wortmann fx 60-126 (blue) and NACA 0017
(red) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.16 Airfoils placement defining the BWB wing in CPACSCreator. . . 76
4.17 Isometric view of the BWB geometry defined as a wing. . . . . . . 76
4.18 Fuselage sections of the BWB geometry visualized in CATIA. . . 78
4.19 Fuselage sections of the BWB geometry defined in CC. . . . . . . 78
4.20 BWB geometry visualized in CC with the separated fuselage and
wing parts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.21 Trailing edge control surface definition in CPACSCreator. [19] . . 80
4.22 Trailing edge control surface deflection in CPACSCreator. [19] . . 81
4.23 Illustration in MATLAB of the first and last airfoil of each of the
4 TEDs defined in the BWB right semiwing. . . . . . . . . . . . . 82
4.24 Top view of the BWB geometry visualized in CC with the control
surfaces defined. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.25 MATLAB visualization of the 4CM036 engine selected for the BWB
model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.26 BWB geometry with the incorporated engines . . . . . . . . . . . 84
4.27 Engine performance with respect to the altitude and Mach number
for the BWB baseline. . . . . . . . . . . . . . . . . . . . . . . . . 85
4.29 Illustration of the CATIA Shape Morphing function used to get rid
of the ringe in the central part of the BWB baseline . . . . . . . 85
4.28 BWB geometry visualized in SUMO showing some irregularities in
the fuselage geometry. . . . . . . . . . . . . . . . . . . . . . . . . 86
4.30 BWB geometry rendered in CATIA . . . . . . . . . . . . . . . . . 86
4.31 Refined mesh of the BWB geometry using SUMO automatic mesh
generator. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.32 Sketches views of the BWB geometry. . . . . . . . . . . . . . . . . 88
4.33 Cross section area distribution in the X direction for the BWB
baseline, obtained with SUMO. . . . . . . . . . . . . . . . . . . . 88
4.34 Final BWB geometry represented in the Weight and Balance tool. 89
xvi
LIST OF FIGURES
5.1 Example of the geometry panelling in TORNADO for the TCR

cruise example. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.2 SUMO geometry (a) and volume mesh (b) for the Boeing 747-100
model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.3 TORNADO geometry for the lifting surfaces of the B747-100 model.[6] 96
5.4 TORNADO geometry with fuselage for the lifting surfaces of the
B747-100 model.[6] . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.5 Views of the comparison between the TORNADO geometry and
the B747 model.[6] . . . . . . . . . . . . . . . . . . . . . . . . . . 98
5.6 Surface mesh of the Boeing 747-100 model. (a): Control surfaces
on the main wing. (b): Empennage and its control surfaces. . . . 99
5.7 Lift coefficient vs. angle of attack of the B-747 model compared
with previous data from [6] . . . . . . . . . . . . . . . . . . . . . . 99
5.8 Pitching moment coefficient vs. angle of attack of the B-747 model
compared with previous data from [6] . . . . . . . . . . . . . . . . 100
5.9 Lift coefficient vs. outboard elevator deflection of the B-747 model
5.10 Pitching moment coefficient vs. outboard elevator deflection of the
B-747 model compared with previous data from [6] . . . . . . . . 102
5.11 Lift coefficient vs. angle of attack of the B-747 model for M = 0.75
5.12 Pitching moment coefficient vs. angle of attack of the B-747 model
for M = 0.75 compared with previous data from [6] . . . . . . . . . 104
5.13 Drag polars of the B-747 model for M = 0.75 compared with previ-
ous data from [6] . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.14 Short period mode prediction analysis for the Boeing 747-100 model.
(a): Results presented by Da Ronch et at. in [6]. (b): Results ob-
tained in this experience using the new TORNADO code. . . . . . 106
5.15 Phugoid mode prediction analysis for the Boeing 747-100 model.
(a): Results presented by Da Ronch et at. in [6]. (b): Results
obtained in this experience using the new TORNADO code. . . . 106
5.16 Dutch roll mode prediction analysis for the Boeing 747-100 model.
(a): Results presented by Da Ronch et at. in [6]. (b): Results
obtained in this experience using the new TORNADO code. . . . 107
xvii
LIST OF FIGURES
5.19 2D model for the BWB geometry in TORNADO, including the

mesh grid, the Mean Aerodynamic Chord (MAC) and the centre of
gravity position. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
5.17 Trimmed angle of attack of the B-747 model for different Mach
numbers compared with previous data from [6] . . . . . . . . . . . 108
5.18 Trimmed elevator deflection of the B-747 model for different Mach
numbers compared with previous data from [6] . . . . . . . . . . . 109
5.20 Lift coefficient vs. angle of attack of the BWB using TORNADO
for air speed, u=40 m/s . . . . . . . . . . . . . . . . . . . . . . . 110
5.21 Pitching moment coefficient vs. angle of attack of the BWB using
TORNADO for air speed, u=40 m/s . . . . . . . . . . . . . . . . 110
5.22 Drag polar of the BWB using TORNADO for air speed, u=40 m/s 111
5.23 Lift coefficient vs. deflection of the outer elevator of the BWB using
TORNADO for air speed, u=40 m/s . . . . . . . . . . . . . . . . 111
5.24 Pitching moment coefficient vs. deflection of the outer elevator of
the BWB using TORNADO for air speed, u=40 m/s . . . . . . . 112
5.25 3D model for the BWB geometry in XFLR5, including the panel
discretization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
5.26 Lift coefficient vs. angle of attack of the BWB for both methods
for air speed, u=40 m/s . . . . . . . . . . . . . . . . . . . . . . . 115
5.27 Pitch Moment coefficient vs. angle of attack of the BWB for both
methods for air speed, u=40 m/s . . . . . . . . . . . . . . . . . . 115
5.28 Drag polar of the BWB for both methods for air speed, u=40 m/s 116
5.29 Aerodynamic efficiency vs. angle of attack of the BWB for both
5.30 Aerodynamic efficiency vs. lift coefficient of the BWB for both
5.31 Pressure coefficient, Cp, distribution for different angles of attack
in the BWB model. . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.32 Local lift force spanwise for the BWB model for air speed, u=40 m/s119
5.33 Local lift coefficient spanwise for the BWB model for air speed,
u=40 m/s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.34 Lift coefficient vs. angle of attack of the BWB using TORNADO
and XFLR5 for air speed, u=40 m/s . . . . . . . . . . . . . . . . 120
xviii
LIST OF FIGURES
5.35 Pitching moment coefficient vs. angle of attack of the BWB using
TORNADO and XFLR5 for air speed, u=40 m/s . . . . . . . . . 120
5.36 Drag polar of the BWB using TORNADO and XFLR5 for air speed,
u=40 m/s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.37 Aerodynamic Surface Form Factor estimations for different thick-
ness ratios and sweep angles. [37] . . . . . . . . . . . . . . . . . . 122
5.38 Skin friction coefficient vs. Reynolds number for different turbulent
transition locations. [37] . . . . . . . . . . . . . . . . . . . . . . . 123
5.39 Drag polar of the BWB using TORNADO, XFLR5 and the empir-
ical estimation for air speed, u=40 m/s . . . . . . . . . . . . . . . 124
6.1 Statically stable behaviour. . . . . . . . . . . . . . . . . . . . . . . 126

6.2 Statically unstable behaviour. . . . . . . . . . . . . . . . . . . . . 126
6.3 Neutrally stable behaviour. . . . . . . . . . . . . . . . . . . . . . 127
6.4 Course of motions of the dutch roll mode. [39] . . . . . . . . . . . 134
6.5 Course of motions of the roll mode. [39] . . . . . . . . . . . . . . 135
6.6 Course of motions of the spiral mode. [39] . . . . . . . . . . . . . 135
6.7 Static margin vs. angle of attack of the BWB for air speed u=40
m/s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
6.8 S-Plane diagram showing the eigenvalues for the longitudinal mo-
tion of the BWB baseline . . . . . . . . . . . . . . . . . . . . . . . 139
6.9 Simulation of the short period stability mode for the BWB baseline
for a initial pitch perturbation of 10º of amplitude . . . . . . . . . 140
6.10 SDSA results for the short period mode of the BWB baseline com-
pared to ICAO recommendations found in [40]. . . . . . . . . . . 140
6.11 Simulation of the phugoid stability mode for the BWB baseline for
a initial pitch perturbation of 10º of amplitude . . . . . . . . . . . 142
6.12 SDSA results for the phugoid mode of the BWB baseline compared
to ICAO recommendations found in [40]. . . . . . . . . . . . . . . 142
6.13 S-Plane diagram showing the eigenvalues for the lateral motion of
the BWB baseline . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
6.14 SDSA results for the Dutch roll mode of the BWB baseline com-
pared to EASA CS-23 recommendations found in [41]. . . . . . . . 144
6.15 SDSA results for the roll mode of the BWB baseline compared to
Cooper-Harper pilot assessment ratings explained in [40]. . . . . . 145
xix
LIST OF FIGURES
6.16 SDSA results for the spiral mode of the BWB baseline compared
to MIL-F-8785C recommendations. . . . . . . . . . . . . . . . . . 146
6.17 SDSA results for the short period mode of the BWB Model 2 com-
6.18 SDSA results for the phugoid mode of the BWB Model 2 compared
6.19 SDSA results for the Dutch Roll mode of the BWB Model 2 com-
6.20 SDSA results for the roll mode of the BWB Model 2 compared to
6.21 SDSA results for the spiral mode of the BWB Model 2 compared
6.22 Different winglet and rudder configurations considered: (a) Base-
line. (b) Model 3: Reduced winglet (and rudder) span and in-
creased rudder chord. . . . . . . . . . . . . . . . . . . . . . . . . . 150
6.23 SDSA results for the short period mode of the BWB Model 3 com-
6.24 SDSA results for the phugoid mode of the BWB Model 3 compared
6.25 SDSA results for the Dutch Roll mode of the BWB Model 3 com-
6.26 SDSA results for the roll mode of the BWB Model 3 compared to
6.27 SDSA results for the spiral mode of the BWB Model 3 compared
6.28 BWB model with no winglets or rudders. . . . . . . . . . . . . . . 153
6.29 Illustration of the MOB BWB with no winglets and two vertical
fins where the rudders are placed. [26] . . . . . . . . . . . . . . . 154
6.30 Comparison between the drag polars of the optimized Model 3 with
reduced control surfaces areas and the baseline . . . . . . . . . . . 155
6.31 Detail of the comparison between the drag polars of the optimized
Model 3 with reduced control surfaces areas and the baseline . . . 155
6.32 MATLAB simulation for the BWB geometry motion in the roll
direction. (a): Initial conditions. (b): Final state. . . . . . . . . . 156
xx
LIST OF FIGURES
6.33 MATLAB simulation for the BWB geometry motion in the yaw
direction. (a): Initial conditions. (b): Final state. . . . . . . . . . 156
xxi
LIST OF FIGURES
xxii
List of Tables
2.1 Aerodynamic database, output of the AMB-CFD module. . . . . . 33
3.1 Horten Ho 229 characteristics. . . . . . . . . . . . . . . . . . . . . 58

3.2 Northrop YB-49 characteristics . . . . . . . . . . . . . . . . . . . 59
3.3 MOB BWB characteristics . . . . . . . . . . . . . . . . . . . . . . 61
3.4 SAX-40 characteristics . . . . . . . . . . . . . . . . . . . . . . . . 62
3.5 Cranfield University BW11 Eagle Ray characteristics . . . . . . . 63
3.6 X48-B characteristics . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.1 NACA 0012 data . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

4.2 Wortmann fx 60-126 data . . . . . . . . . . . . . . . . . . . . . . 74
4.3 Control surfaces geometry for the BWB baseline model. . . . . . . 81
4.4 Location and technical information of the two 4CM036 engines of
the BWB baseline. . . . . . . . . . . . . . . . . . . . . . . . . . . 83
4.5 Geometric characteristics of the BWB baseline model obtained. . 89
4.6 Geometric characteristics of the baseline BWB wing. . . . . . . . 89
4.7 Center of gravity coordinates and inertia moments of the BWB
model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
5.1 Aerodynamic coefficients for the BWB baseline and α = 2º and

u = 40 m/s obtained in TORNADO. . . . . . . . . . . . . . . . . 108
5.2 Angle of attack derivatives for the BWB baseline and u = 40 m/s
obtained in TORNADO. . . . . . . . . . . . . . . . . . . . . . . . 109
5.3 Sideslip angle derivatives (side force, rolling moment and yawing
moment coefficients) for the BWB baseline and u = 40 m/s ob-
tained in TORNADO. . . . . . . . . . . . . . . . . . . . . . . . . 109
xxiii
LIST OF TABLES
5.4 Roll rate angle derivatives (side force, rolling moment and yaw-
ing moment coefficients) for the BWB baseline and u = 40 m/s
5.5 Pitch rate angle derivative (pitching moment coefficients) for the
BWB baseline and u = 40 m/s obtained in TORNADO. . . . . . . 110
5.6 Yaw rate angle derivatives (rolling moment and yawing moment
coefficients) for the BWB baseline and u = 40 m/s obtained in
TORNADO. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
5.7 Inner elevator deflection derivatives for the BWB baseline and u =
40 m/s obtained in TORNADO. . . . . . . . . . . . . . . . . . . . 112
5.8 Outer elevator deflection derivatives for the BWB baseline and u =
40 m/s obtained in TORNADO. . . . . . . . . . . . . . . . . . . . 112
5.9 Elevon deflection derivatives for the BWB baseline and u = 40 m/s
5.10 Rudder deflection derivatives for the BWB baseline and u = 40 m/s
5.11 Comparison between the TORNADO predictions for the BWB
baseline and the available experimental data from [5] for u = 40 m/s.
113
6.1 Stability derivatives for the different motions of the BWB baseline
and the static stability criterion. . . . . . . . . . . . . . . . . . . . 137
6.2 Control surfaces geometry for the BWB baseline model. (Model 1) 146
6.3 Control surfaces geometry for the reduced control surfaces BWB
configuration, including the comparison with the baseline surfaces.
(Model 2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
6.4 Control surfaces geometry for the reduced control surfaces and
winglets BWB configuration, including the comparison with the
baseline surfaces. (Model 3) . . . . . . . . . . . . . . . . . . . . . 150
xxiv
LIST OF TABLES
Latin symbols
Symbol Unit Description

AR [-] Aspect ratio
b [m] Wing span
c [m] Chord
c1 [m] Inner chord
c2 [m] Outer chord
CD [-] Drag coefficient
CD0 [-] Parasite drag coefficient
CDi [-] Induced drag coefficient
Cf [-] Skin friction coefficient
CL [-] Lift coefficient
Cl [-] Roll moment coefficient
Cm [-] Pitch moment coefficient
Cm 0 [-] Zero-angle of attack pitch moment coefficient
Cn [-] Yaw moment coefficient
Cp [-] Pressure coefficient
CY [-] Side force coefficient
D [kg m/s2 ] Drag force
e [-] Oswald efficiency number
f [-] Slenderness (ratio between length and diameter)
F [kg m/s2 ] Force
fa [kg m/s2 ] Aerodynamic forces
FF [-] Form factor
hac [m] Aircraft height
H [m] Altitude
I [kg m2 ] Inertia moments
k [-] Induced drag factor
k̄ [-] Reduced frequency
Kn [-] Static margin
L [kg m/s2 ] Lift force
L [kg m2 /s2 ] Roll moment
LBW B [m] BWB length
xxv
LIST OF TABLES

Lf [m] Fuselage length
M [kg m2 /s2 ] Pitch moment
M [-] Mach number
MAC [m] Mean Aerodynamic Chord
N [kg m2 /s2 ] Yaw moment
p [1/s] Roll rate
PAX [-] Number of passengers
q [1/s] Pitch rate
q [kg/(m·s2 )] Dynamic pressure
r [1/s] Yaw rate
Re [-] Reynolds number
S [m2 ] Planform area
SD0 [m2 ] Zero-lift drag area
Sref erence [m2 ] Reference area
Swet [m2 ] Wetted area
Swing [m2 ] Wing area
t [m] Thickness
t [s] Time
T1/2 [s] Time to damp half the amplitude
TAS [m/s] True Air Speed
u [m/s] Air speed
U [m/s] Velocity in the X body axis
V [m/s] Velocity in the Y body axis
VT [m/s] True Air Speed
W [m/s] Velocity in the Z body axis
Greek symbols

α [º] or rad Angle of attack
β [º] or rad Sideslip angle
Γ [º] or rad Dihedral angle
δ [º] or rad Deflection angle of a control surface
xxvi
LIST OF TABLES

η [-] Real part of the Dutch roll eigenvaule
θ [º] or rad Pitch angle
λ [-] Eigenvalue
λ [-] Taper ratio (Ratio between tip chord and root chord)
Λ [º] or rad Sweep angle
µ [-] Imaginary part of the Dutch roll eigenvaule
µ [kg/(m·s)] Air dynamic viscosity
ξ [-] Damping ratio
ρ [kg/m3 ] Air density
σ [-] Real part of the phugoid and short period eigenvalues
φ [º] or rad Roll angle
ψ [º] or rad Yaw angle
ω [1/s] Oscillation frequency
ωd [1/s] Damped frequency
ωn [1/s] Natural undamped frequency
Indices
Index Description
ac Relative to the neutral point
BW B Relative to the Blended Wing Body
cg Relative to the centre of gravity
component Relative to a component of the aircraft
DR Relative to the Dutch roll mode
engine Relative to the engines
f Relative to the fuselage
LE Relative to the leading edge
p Derivative with respect to roll rate
ph Relative to the phugoid mode
q Derivative with respect to the pitch rate
r Derivative with respect to the yaw rate
sp Relative to the spiral mode
wing Relative to the wing
xxvii
LIST OF TABLES
Index Description
x Component in the X direction
y Component in the Y direction
z Component in the Z direction
α Derivative with respect to the angle of attack
β Derivative with respect to the sideslip angle
δInnerElev Derivative with respect to the deflection angle of the inner elevator
δOuterElev Derivative with respect to the deflection angle of the outer elevator
δElevon Derivative with respect to the deflection angle of the elevon
δRudder Derivative with respect to the deflection angle of the rudder
φ Derivative with respect to the roll angle
∞ Relative to the conditions in the farfield
xxviii
Nomenclature
AcBuilder Aircraft Builder (Module of CEASIOM)
AIC Aerodynamic Influence Coefficients
AMB Aerodynamic Model Builder (Module of CEASIOM)
ASCII American Standard Code for Information Interchange
BWB Blended Wing Body
CAD Computer-Aided Design
CATIA Computer Aided Three-dimensional Interactive Application
CC CPACS Creator
CDE Computational Design Engine
CEASIOM Computerised Environment for Aircraft Synthesis and Integrated Optimisation

Methods
CFD Computational Fluid Dynamics
CORDIS Community Research and Development Information Service
CPACS Common Parametric Aircraft Configuration Scheme
DATCOM Data Compendium
DLM Doublet Lattice Method
DLR Deutsches Zentrum für Luft- und Raumfahrt (German Aerospace Center)
xxix
LIST OF TABLES LIST OF TABLES
DOC Direct Operational Cost
EULA End-User License Agreement
FAR Federal Aviation Regulations
FCS Flight Control System (see FCSDT module from CEASIOM)
FOI Totalförsvarets forskningsinstitut (Swedish Defence Research Agency)
GPL GNU General Public License
GUI Graphical User Interface
ICAO International Civil Aviation Organization
IGES Initial Graphics Exchange Specification
ISO International Organization for Standardization
JAR Joint Aviation Requirements
KTH Kungliga Tekniska Högskolan (Royal Institute of Technology)
LLT Lifting Line Theory
LQR Linear-Quadratic Regulator
MAC Mean Aerodynamic Chord
MATLAB Matrix Laboratory
MEW Manufacturer’s Empty Weight
MIT Massachusetts Institute of Technology
MLS Moving Least Squares
MOB Multidisciplinary Design and Optimisation for Blended Wing Body
MTOW Maximum Takeoff Weight
NACA National Advisory Committee for Aeronautics
NASA National Aeronautics and Space Administration
xxx
NASA National Aeronautics and Space Administration
NeoCASS Next Generation Conceptual Aero-Structural Sizing Suit
NLR Nationaal Lucht- en Ruimtevaartlaboratorium (Dutch National Aerospace Labo-

ratory)
NOx Mono-Nitrogen Oxides
Pkm Passenger · Kilometer
RANS Reynolds Averaged Navier-Stokes
RBF Radial Basis Functions
SAS Stability Augmentation System
SAX Silent Aircraft eXperimental
SAX Silent Aircraft eXperimental
SDSA Simulation and Dynamic Stability Analyzer
SimSAC Simulating Stability And Control Characteristics for Use in Conceptual Design
SMX Statistica Matrix Spreadsheet
SUMO Surface Modeler
TCR Transonic Cruiser by SAAB
TCR Transonic Cruiser by SAAB
TED Trailing Edge Devices
TetGen Tetrahedral Mesh Generator
UAV Unmanned Aerial Vehicle
uID Unique Identifier
UPM Universidad Politécnica de Madrid (Technical University of Madrid)
USAF United States Air Force
xxi
VLM Vortex Lattice Method
VSP NASA open source parametric geometry
XML Extensible Markup Language
XSD XML Schema Definition
xxxii
Chapter 1
Introduction
1.1 Motivation for this work

World air traffic has been increasing in a constant way over the last decades, and
all forecasts predict that this pace will continue to accelerate in the near future
due to the great economic development of China, India, Brazil and other areas.
Some authors [1, 2, 3, 4] estimate that the overall revenue passenger·kilometre
figure will grow at a pace above 5%, well over the world economic growth. Short
range transport is expected to increase in developing countries with their own
manufacturing industry and long range transport will experiment a remarkable
growth between the Western World, Middle East and the developing countries.
Boeing made a prediction of the worldwide air traffic evolution compared to 2012
which is presented in figure 1.1, with an average growth of 5.1%.
Figure 1.1: Worldwide air traffic evolution prediction from 2012 to 2032. [1]
In addition, several main hubs already exhibit congestion troubles and some
1
CHAPTER 1. INTRODUCTION
strategic areas like Western Europe or the Northeast coast of USA show a sat-
urated airspace. Freight traffic is expected to increase at even higher rates than
passenger traffic. This process will require more than 35000 new jet aircrafts and
the reconversion of a large number of ageing airliners in the next years (85% of
the world fleet will be new by 2032, [1]). However, this enormous demand will
occur in a moment with high pressure to reduce both direct operating cost and
environmental impact, in part also to the high oil price (see figure 1.2).
Figure 1.2: Brent Crude Oil and US Gulf Coast Jet Fuel average prize between
2000 and 2013. [1]
The conventional aircraft configuration has remained essentially unchanged

for the last six decades and is approaching a productivity and capacity asymptote
around the size of Airbus A380. The continuously changing market and technol-
ogy scenario is leading to new revolutionary unconventional designs and concepts
to address this increasing air traffic demand. One of the most promising concepts
is the Blended Wing Body (BWB) aircraft (see figure 1.3). It is a tailless aircraft
design that integrates the wing and the fuselage into a single lifting surface, by
the thickening of the wing in the central part of the airplane. Because of its eco-
nomical performance and its higher capacity1 , it is a promising candidate for the
future large airliner. The feasibility, efficient performance and airport compati-
bility of the BWB concept have been assessed in several previously publications,
with encouraging results. Consequently, there is a great research activity on it,
performed by the aircraft industry, academia and research centres.
1
Some designs may carry between 800 and 1500 passengers in routes of about 13000 km.
These very large aircrafts benefit even more about the BWB potential.
2
1.2. CRUCIAL AND DISTINGUISHING ASPECTS OF THE BWB
CONFIGURATION
Figure 1.3: Boeing X-48B Blended Wing Body developed by NASA.
1.2 Crucial and distinguishing aspects of the BWB

configuration
The main advantages of the BWB concept are its reduced wetted area to volume
ratio, which leads to a drag reduction and, therefore, to lower fuel burn and
environmental impact, significant payload advantages and noise reduction. All in
all, this leads to the achievement of a lower operational cost.
One of the critical aspects for the BWB configuration is the control and sta-
bility, due to the lack of a tail and its unconventional shape. Therefore, the design
and allocation of the control surfaces becomes crucial. The design of the fuselage
is a more complex problem than in usual aircrafts, due to its aerodynamic and
lifting performance. The fuselage cross sections have to be carefully designed in
order to fulfill both the structural and aerodynamic requirements. Finally, the
size of the vertical surfaces, such as the winglets, condition the radar detectability
of an aircraft. For this reason, the minimum size for these surfaces is a goal to
achieve, also for drag reduction.
1.3 Aircraft design process

Traditional commercial aircraft conceptual design methods often employ hand-
book methods based on semi-empirical theory and data, which are not reliable
enough for treating novel and unconventional designs and can easily lead to re-
markable sizing errors.
3
It is necessary to improve the aircraft design process and make it more effi-
cient. One possibility to achieve this is to implement fast, high fidelity software
frameworks earlier in the design process for making physics based predictions of
the performance and stability available. Obtaining highly accurate information
earlier helps to avoid expensive redesign of the aircraft in the future and might
allow being “first-time-right”. This frameworks should be able to perform multi-
disciplinary aircraft analysis and design.
To cope with missing link in the chain, CEASIOM was introduced. CEASIOM
(Computerised Environment for Aircraft Synthesis and Integrated Optimisation
Methods) is a multidisciplinary support tool for the conceptual aircraft design. It
can help to obtain an improved, faster prediction of stability and control calcula-
tions with high-fidelity methods, facilitating the aircraft design process, compared
to contemporary aircraft design tools. CEASIOM integrates into an application
the main aircraft design disciplines simulation (geometry, structures, aerodynam-
ics, aeroelasticity, stability, control...) in different interconnected modules. This
way, the definition of a virtual aero-servo-elastic aircraft model is possible earlier
in the design process, reducing technical and financial risks.
1.4 A new design solution

However, the geometry module of CEASIOM, AcBuilder, is very limited and can-
not handle unconventional aircraft configurations and has several constraints, such
as the necessity of having only one tubular fuselage, reduced number of kinks in the
wing, and, in general, usual restrictions for conventional aircraft configurations.
These restrictions render the current state of the CEASIOM platform largely,
making it unapplicable for the study of unconventional configurations, such as
the BWB. To overcome these and other limitations, the Common Parametric Air-
craft Configuration Scheme (CPACS), developed by the DLR, are implemented
for describing the BWB geometry. Using the tool CPACSCreator (CC), instead
of Acbuilder, a BWB aircraft is designed and created. However, CPACS uses a
different XML format Extensible Markup Language) than CEASIOM, so a con-
version between them is required.
4
1.5. OBJECTIVES OF THIS WORK
The initial BWB baseline incorporates concepts from previous BWB designs;
in particular, the one developed by Martin Carlsson[5], which presents a brief de-
scription and experimental data of the ELSA BWB, a low speed UAV (Unmanned
Aerial Vehicle) model which was created and tested in flight in KTH Department
of Aeronautics. The available information on this model is not very extense, but
experimental data is a very useful source and the fact that it actually flew is very
encouraging and a good starting point for an improved version of BWB. Once the
baseline geometry is defined, the model is analyzed in the rest of the modules of
CEASIOM, after modifying the CPACS XML file via wrappers in order to be-
come loadable by CEASIOM. Before analyzing the BWB aerodynamic features,
an example test is studied to validate the modified TORNADO code which deals
with multiple control surfaces. The selected example is the Boeing 747-100 con-
figuration, making use of the experimental and numerical data presented in [6]
as a reference for comparison. In order to get more results to compare with, the
BWB geometry was also implemented in the XFLR52 software [45], for studying
its aerodynamical features. After obtaining the information for weight, balance,
aerodynamics and propulsion, the control and stability features of the BWB was
analyzed. Once this data is obtained, the sizing and allocation of the control sur-
faces was studied and optimized. The minimum size of the winglets which offer a
satisfactory flight performance was determined due to its importance in the radar
detectability and drag, as stated before.
In case of having larger computational resoruces, higher-fidelity analysis tools,

such as Edge or RANS (Reynolds Averaged Navier-Stokes), could also be em-
ployed for studying the aerodynamic behaviour in order to get more precise re-
sults. Moreover, a more convenient user interface should be implemented when
further implementing CPACS in the rest of the design tools. Finally, wind tunnel
and flight tests could show the real performance of the model studied, validating
or not the results obtained numerically.
1.5 Objectives of this work

The main objectives of this thesis are listed in the following points:
2
XLFR5 is an open source software based on a improved version of the XFOIL[63] code
developed by Mark Drela at MIT
5
• Implement the use of CC instead of AcBuilder and make it compatible with

the rest of CEASIOM modules.
• Accomplish an in-depth literature research of the BWB concept and exist-

ing BWB models in order to identify the key principles and current design
trends.
• Design a new BWB geometry in CC, sustained in the existing know-how

and lessons learnt from previous designs.
• Enhance the numerical tool CESIOM and use it to study the performance
of the model (with an special emphasis on its stability and control features)
and achieve the optimal design. Compare with the available results in [5].
• Find optimal design solutions for the control surfaces sizing and allocation
that provide a target performance while ensuring a low radar detectability
(minimum winglet size).
1.6 Thesis breakdown

A brief overview of the main chapters of the thesis are presented in the following
lines.
Chapter 2 focus on the numerical tools to be used in this project. New design
and analysis methods such as CEASIOM are introduced, offering a higher
fidelity and detail in earlier stages of the design process compared to tra-
ditional methods, especially in the stability and control performance of the
aircraft. The main modules of CEASIOM are explained in this section,
providing a brief presentation of their fundamentals. A new alternative to
CEASIOM’s module AcBuilder is presented afterwards with the so-called
CPACS. A brief overview of the data handling and structure of this tool is
provided, with a highlight of the software used in this project, CPACSCre-
ator.
Chapter 3 presents in detail the innovative Blended Wing Body aircraft concept,
after showing the current need of a revolutionary alternative to conventional
6
1.6. THESIS BREAKDOWN
aircraft. The main advantages and drawbacks of this configuration are then
listed, confirming the great potential of this concept. When designing an
aircraft, it is strongly recommended to research similar models in order to
learn from previous experience and not making the same mistakes. Unfor-
tunately there are not so many BWB models, due to the innovative nature
of this concept, but some of the most important aircrafts designed with this
configuration are presented. A detailed description is provided to the ELSA
BWB by Martin Carlsson [5], in which this project is inspired and compared
with.
Chapter 4 provides an overview of the design process of the BWB model em-
ployed in this research. The design process started with the definition of the
whole aircraft as a wing for defining later a fuselage with the same shape
from it, using the softwares CC, SUMO and CATIA. The implementation of
the control surfaces and engines was performed inmediatly afterwards. Fi-
nally, an iterative geometry refinement process was used using SUMO and
CATIA in order to smooth the outer shape of the aircraft. After that, the
final model used for later testing is illustrated.
Chapter 5 is focused on the aerodynamic analysis of the BWB design. First

of all, a more detailed introduction about TORNADO’s fundamentals is
provided. Then, an example test is performed in a Boeing 747-100 model,
in order to test the fidelity of the results of the new TORNADO code which
handles with a higher number of control surfaces. Afterwards, the results
obtained for the BWB model using TORNADO and XFLR5 are presented
and compared. Finally, an empirical parasite drag estimation for the model
is performed.
Chapter 6 analyzes the stability and control characteristics of the BWB model
after providing a brief introduction of the stability and control fundamen-
tals. Several analysis cases are performed in order to define the optimal
control surface size that still allows a satisfactory aerodynamic and flight
performance. The winglets are then designed to the minimum size that pro-
vides satisfactory flight performance. In addition, a MATLAB animation of
the motion for the roll and yaw directions is presented.
Chapter 7 gathers the main conclusions and comments of the whole thesis. A
7
review of the fulfillment of the objectives previously listed in the introduction

is presented. In addition, an outlook about the future work that can be
performed in this field is suggested.
8
Chapter 2
Aircraft conceptual design
Aircraft design is a very long and complex process that can be divided into three
major stages, as shown in figure 2.1, with an increasing level of detail and fidelity:
1. The conceptual design: It starts with a set of requirements, such as size,

weights and performance, for the aircraft that is to be designed, including
desired range, payload, takeoff and landing distances and several speed re-
quirements, such as landing sink speed or stall speed. A first sketch is used
to estimate the necessary geometry of the aircraft to meet the mentioned
requirements. This process can be repeated with an increasing level of detail
to get a more accurate insight of the basic aircraft configuration, such as
fuselage shape, wing and tail geometry, engine location, etc. Hence, concep-
tual design is a fluid process where new ideas and problems can occur. At
the end of this stage, all the design layouts have been analyzed and the best
options (one or two usually) will pass to the preliminary design phase.
2. The preliminary design: After taking the major decisions in the conceptual
design stage, the layout passes to the preliminary design stage. During
this stage, the subsystems of the aircraft, such as the landing gear, control
systems or the structure, are further analyzed by experts using higher fidelity
methods. At this step the configuration is frozen and first tests will be
initiated. Another subject in this step is the “lofting” of the outer skin
to ensure the right interaction of the single components. Despite being
supposed to be a one-way process, the obtained results from the analysis
might make it necessary to go back to the conceptual design stage to redefine
certain parameters, with the consequent expense involved. Usually, a lot
9
CHAPTER 2. AIRCRAFT CONCEPTUAL DESIGN
of re-work and repair of the aircraft model developed are required. The
preliminary design has to establish confidence to build the aircraft on time
with the estimated costs, meeting the customer’s requirements.
3. The detail design: In the detail design phase the precise geometry of the com-
ponents are designed for fabrication considering producibility. This phase is
divided into the detailed design and analysis of the actual components and
the production design. The last one handles the planning and arranging of
the aircraft fabrication, including the fabrication of every component and
every step till the final assembly of the aircraft. The last step is the actual
testing of the manufactured structure components and a leadoff flight sim-
ulator flown by test pilots to test the control systems. The end of the detail
design stage is achieved when the entire aircraft has been assembled.
Figure 2.1: The different stages in the creation process of an aircraft.
During the three phases, several tools are used to design the different variables
in order to define the aircraft that meets the given requirements. A multidisci-
plinary study should be done, including geometry, weight estimation, structural
analysis, aerodynamics, stability and control, flight dynamics, noise and emissions.
The data obtained in one stage is introduced into the next one where the fidelity
and detail level will increase. After successful completion of the three phases in
aircraft design the aircraft will be manufactured.
Nowadays, the aeronautical industry demands the development of more com-

plex products with a compressed time-span and more cost-effectiveness while using
innovative business models involving multiple partners. This means that devel-
oping a new aircraft cannot be addressed only by improving existing practices.
Thus, there is a variation from the different and compartmentalized three-stages
sequency (shown in figure 2.1) to a more fluid and homogeneous method where
10
the design loops and project studies are now compacted into a single multicollab-
orative, multidisciplinary and multifidelity process illustrated in 2.2. This process
is used in both the aircraft level (classical conceptual design) and the component
level (preliminary design). All in all, the process based on design is being trans-
formed into a process based on a modelling and simulation approach throughout
the entire aircraft development program. New efficient tools, such as CEASIOM
and CPACS, are easing the transition from the traditional and rigidly structured
sequential design process to the new continuous and fluid design process, allow-
ing bidirectional work flow between the conceptual and preliminary phases. This
software frameworks improve the way of designing and analyzing conventional
aircraft configurations during the conceptual design phase and provide an inte-
grated design and decision making environment, where all the required predictive
computations can be performed, providing a fidelity selected by the user.
Figure 2.2: The new virtual aircraft design method.
Nickol[7] defined a table including six columns of disciplines: aerodynamics,

structures and weights, noise, emissions, stability and control and geometry gen-
eration. Each of these disciplines counted with three different levels of fidelity,
presented in the table as three rows, low, medium and high. These three levels of
fidelity correspond to the three stages of the aircraft design explained above. As
the design proceeds, data must travel across the different disciplines and upwards
to the increasing fidelity levels [8]. Some of the former gaps in this table have
been filled by other authors, such as Rizzi et al. [8] who describe a custom CAD
11
solution with automatic meshing for medium fidelity geometry analysis. All in
all, the current table adapted from Nickol [7], is presented in figure 2.3.
Figure 2.3: Table of analysis of disciplines and tools depending on the level of
fidelity required for aircraft design, according to Nickol [7].
Due to the enormous resources required for the preliminary and detail design,
the scope of this thesis is just focused on the conceptual design stage and the start
of the preliminary definition, but it is enough to create the first view of the BWB
aircraft and make the first analysis.
In the following paragraphs, the traditional methods for the conceptual design
and the new tools CEASIOM and CPACS are described.
2.1 Traditional methods

Traditional commercial aircraft conceptual design methods make extensive use
of handbook methods based on semi-empirical theory and data. More sophisti-
cated and powerful in-house developed industrial software systems usually alter
the widely recognised handbook methods by way of recalibrating the said algo-
rithms with data gleaned from experience and previous design projects. A growing
consensus of opinion, however, finds even such recalibrated handbook methods are
12
2.1. TRADITIONAL METHODS
not reliable enough for treating novel and unconventional designs and are easily
error-prone [46].
These processes is based on mathematical modelling (usually empirically based)

and the main utilities required to conduct the design process are [9]:
• Design specification: Several minimum requirements that any aircraft design

candidate should fulfill.
• Design parameters: Independent variables that define the airplane in a tan-

gible sense.
• Prediction variables: A set of quantities which indicates the degree of ful-

fillment of all the requirements.
• Prediction methods: Explicit relationships between the design parameters

and prediction variables. The accuracy of these methods depends on the
quantity and fidelity of the input information.
• Design space: the amount of aircraft candidates that fulfill the design spec-
ifications completely or partially.
Before 1960 the computer was rarely integrated into this design process, so per-
forming this tasks manually usually required a lot of time and led to remarkable
errors. Up to now, the aerodynamic data considered in the conceptual design stage
were mostly based on tabulated data, previous experience and/or semi-empirical
approaches. Such simplified methods do not provide sufficient fidelity and usually
can lead to errors in the sizing process. Some of the errors can be due to Reynolds
number effects, configuration sensitivities, dynamic motion effects and related is-
sues. Such errors generally can be detected only with a significant increase in the
fidelity of the aerodynamic data base, for instance with wind-tunnel data or even
flight test data, which become available in the late design process. The later in
the design process the error is identified, the higher the cost of its correction.
Figure 2.4 illustrates the two design loops in the traditional conceptual design
stage that follow the first-guess sizing (usually done by a spread-sheet) and provide
the baseline layout of the aircraft configuration:
13
Figure 2.4: The scheme of the traditional conceptual design process.[10]
1. The first one is the pre-design loop, which defines very quick technically
consistent sized configuration with a predicted performance.
2. The second loop is the concept-design and is an extended intensive effort

that includes more advanced trade studies to refine the definition of the
minimum goals that a candidate project should fulfill.
In the scheme represented in figure 2.4, the crucial disciplines for the aero-servo-
elastic aircraft desing are shown:
• Aerodynamics
• Staibility and Control
• Structures
• Performance Optimization.
Stability and control design is a crucial discipline that should be considered as

soon as possible in the design process, especially in tailless aircrafts such as the
14
2.1. TRADITIONAL METHODS
BWB. Previous experience [11] has proven that inadequate design of the stability
and control fundamentals can cause the demise of any project. Nearly every
aircraft project has experienced flying qualities problems at some moment during
the flight-testing process. Difficulties can also appear in extreme conditions, such
as engine-out or actuator failure cases. The current tendencies in aircraft design
towards augmented stability and expanded flight envelopes demands stability and
control information as early as possible in order to be “First-time-right” and reduce
the process cost.
Figure 2.5: Comparison between the classical design methods (a) and the use of
CEASIOM (b). As illustrated, CEASIOM provides higher fidelity earlier in the
design process, saving time and cost. [46]
Therefore, nowadays there is a common consensus towards replacing the tra-

ditional methods with new computational procedures that calculate the required
information from first principles. Figure 2.5 illustrates the difference in the design
process between the classical methods and the new computational method using
CEASIOM. This software provides higher fidelity earlier in the process, during
the product definition, were approximately up to 80% of the life cycle budget is
expended. The information about the aircraft is more uniform in the different
disciplines than the former case, with a remarkable higher focus on the manufac-
turing features only at the detailed stage of the project. The stability and control
characteristics are well defined since the beginning, avoiding expensive situations
where the whole subsystem has to be redesigned thanks to a design mistake. This
new process allows verification in wind tunnel and flight tests much earlier, with
the consequent saving of resources and time.
15
2.2 Computational tools for conceptual design:

CEASIOM
CEASIOM (Computerised Environment for Aircraft Synthesis and Integrated Op-
timisation Methods) is a support tool for engineers for the conceptual aircraft
design process. CEASIOM was developed within the SimSAC project (Simulat-
ing Stability And Control Characteristics for Use in Conceptual Design), funded
by the European Commission 6th Framework Program on Research, Technological
Development and Demonstration. The software development began in November
2006 and continues to date. It can help to obtain an improved prediction of sta-
bility and control calculations with high-fidelity methods, facilitating the aircraft
design process, compared to contemporary aircraft design tools.
CEASIOM gathers into an application the main aircraft design disciplines sim-
ulation (geometry, structures, aerodynamics, aeroelasticity, stability, control...) in
different interconnected modules that share unified geometrical description. This
way, the definition of a virtual aero-servo-elastic aircraft model is possible earlier
in the design process, reducing technical and financial risks. However, this soft-
ware does not perform the whole conceptual design process, requiring an initial
layout as input for the baseline configuration to modify and an engineer in-the-
loop who controls if the results achieved are correct and make sense. By using a
set of parameters, the user can define the aircraft for various analysis cases. CEA-
SIOM models can be handed over to later desing stages as IGES files required in
higher-fidelity CFD analysis. In figure 2.6 the main modules of CEASIOM and
their connections are represented.
CEASIOM is offered as a freeware, according to the EULA (End-User License
Agreement) conditions and it is downloadable free of charge from the official
website [46].
2.2.1 Coordinate systems and control sign conventions

Before starting to analyze the different modules included in CEASIOM, it is nec-
essary to define the different coordinate systems that they handle [12]. Figure 2.7
presents the main coordinate systems used by CEASIOM:
• In the XML file, the origin is the nose of the aircraft. The X-axis goes from
16
2.2. COMPUTATIONAL TOOLS FOR CONCEPTUAL DESIGN: CEASIOM
Figura 2.6: CEASIOM software with core modules: Geo-sumo, AMB-CFD, Neo-
CASS, S&C.
the nose to the tail, the Y-axis goes to the right wing, and the Z-axis points
up.
• Tornado uses a Cartesian coordinate system with the X-axis along the air-
craft body, increasing aft. The Y-axis is aligned positive out through the
starboard wing when no dihedral is present. The Z-axis is right-hand per-
pendicular to the X- and Y-axis, thus pointed up. The origin can be fixed by
the user, and is normally fixed at the aerodynamic center. The coordinate
system is the same then in the XML file, just with a different origin.
• SDSA uses a body axis system. The origin is placed in a point usually
defined by 1/4 MAC (Mean Aerodynamic Chord). The axis OX is parallel
to the MAC and pointed forward to the nose of the aircraft, axis OZ is
pointed down, and axis OY is pointed to the right wing. This is coherent
with the international normed coordinates system ISO 1151-1.
• In AMB, the origin is the reference point. The reference point can be spec-
ified by the user, it can be the center of gravity, the 1/4 MAC or a user
defined point. AMB uses the same coordinate system then SDSA.
The convention for the control surfaces deflections should also be specified
[12]. The SimSAC convention will be as defined by Cook [11] which is: a positive
17
Figure 2.7: Different coordinate systems used in CEASIOM. [12]
control action by the pilot (from the pilot’s point of view) gives rise to a positive
aeroplane response, whereas a positive control surface displacement gives rise to
a negative aeroplane response. Thus:
• Roll: positive right push force on the stick ⇒ positive starboard stick dis-
placement ⇒ right aileron up and left aileron down (negative) ⇒ starboard
wing down roll response (positive)
• Pitch: positive pull force on the stick ⇒ positive aft stick displacement ⇒
elevator trailing edge up (negative) ⇒ nose up pitch response (positive)
• Yaw: positive push force on the right rudder pedal⇒ positive rudder bar
displacement ⇒ rudder trailing edge displaced to the right (negative) ⇒
nose to the right yaw response (positive)
The positive control surfaces deflection are shown on the figure 2.8:
CEASIOM integrates different modules:
2.2.2 AcBuilder (Geometry)

This module can be used for creating a parametrized geometry of the aircraft,
defining the components and their dimensions. The user can also change the fuel
and weight parameters (see figure 2.10) and visualize the sketch of the aircraft
while changing the geometry parameters. This module is composed of different
sections:
18
Figure 2.8: Positive control surfaces deflections convention for CEASIOM.[12]
• Geometry: Defines the components of the aircraft and describes the fuel
tanks, cabin, luggage definition and the wingboxes.
• Weights and Balances: Estimates weights, centers of gravity and the inertia
moments of the aircraft and its components using the information from the
previous section.
• Technology: Visualizes the input parameters for the following module, Neo-
CASS, and the beam model, aero panels and spar locations can be defined.
The user defines the distribution of the aero mesh and the structural mesh.
Material characteristics can be set and the structural concepts of the single
components is selected.
AcBuilder is a mixture of MATLAB and Java (for the graphic interface), which
makes it difficult for engineers with no experience in Java to make alterations to
the code. In figure 2.9 the data flow and structure of this module is illustrated.
In this module it is allowed to import and export parameters and files in XML
format.
An airfoil library is available and can be extended with DAT files with the
point coordinates written in 2 columns. The format for this files consists in num-
bers with 4 fractional digits scaled from 0 to 1 and ordering them with the upper
surface first and then the lower surface.
AcBuilder enables the visualization of conventional aircraft configurations, but

the display of unconventional aircrafts is restricted due to a limited number of el-
ements (1 cylindrical fuselage, 1 main wing, 1 horizontal tail, etc.). In addition,
19
Figure 2.9: AcBuilder data flow including the user interface.
these mentioned elements must be defined in a particular order, because the coor-
dinate systems employed for defining the parameters of the individual elements are
dependent on each other. Therefore, the geometry definition process in AcBuilder
is very unflexible. This led to the development of CPACSCreator (CC) (see sub-
subsection 2.3.2), wherein the user is allowed to add an arbitrary number of lifting
surfaces and aircraft components. AcBuilder employs an old XML scheme for its
files, but CC uses a more extended XML scheme for its files which increases the
detail in the supported data.
Figure 2.10: Sketch of the Transonic Cruiser TCR geometry in the Weights and
Balance section of the old AcBuilder module.
20
2.2.3 SUMO (Geometry mesher)

The graphical surface modelling tool SUMO1 (SUrface MOdeler) can be used
to quickly define an accurate aircraft geometry based on a moderate amount of
b-spline surfaces. SUMO also has a graphical user interface for generating or
modifying the geometry and a large library for creating or changing wing airfoils
or fuselage sections. The b-spline surfaces are generated using control points and
spline curves. SUMO works with basically two different types of surfaces: body
surfaces (for fuselages and pylons) and wing surfaces (for wings, horizontal tails
and lifting surfaces). Body surfaces are defined by their top and side views and
their cross sections definition. SUMO also performs automatic CAD repair by
closing wingtips and fuselage noses and tails, if necessary.
SUMO is able to automatically generate surface and volume unstructured

meshes, for using them in 3D panel methods and CFD solutions based on the
Euler equations via TetGen (Tetrahedral Mesh Generator)2 . In a first step a tri-
angular mesh is generated on the aircraft surface. The triangular elements are
based on a three-dimensional in-sphere criterion. In comparison to the Delaunay
methods they yield a better mesh quality for skewed surfaces such as thin, swept
delta wings. Moreover, triangular elements generates surface meshes easier, at
least when a sufficient quantity of elements are used. Based on surface meshes,
unstructured volume meshes are created by TetGen to fill the space between the
aircraft and the farfield.
Results are exportable as an IGES file. An example of a generated surface

mesh can be visualized in figure 2.11 and a volume mesh in figure 2.12.
2.2.4 Weight & Balance

The weight and balance parameters of non-structural masses and their location
can be estimated using semi-empirical methods (Howe, Torenbeek, Raymer, USAF
and Cessna) mainly based on statistical handbooks. The values can also be settled
1
SUMO was developed by Larosterna (http://www.larosterna.com), a software development
business started in June 2009 by Dr. David Eller from the Flight Dynamics Lab at KTH in
Stockholm.
2
TetGen (http://www.tetgen.org) was developed by Hug Si at the Weierstrass Institute for
Applied Analysis and Stochastics (WIAS) in Berlin
21
Figure 2.11: Example of mesh generated using SUMO.
Figure 2.12: Volume mesh of the Transonic Cruiser TCR geometry using TetGen
and SUMO.
by the user, as well as which method to use. This way the inertia moments of
the aircraft can be calculated. The different centers of gravity can be visualized
(fig.2.13) as well as pie charts with the repartition of the weights for each available
method. FInally, the module calculates the inertia moments and the coordinates
of the center of gravity of the aircraft.
To put this process into practice the XML file is read and introduced in 4
MATLAB scripts:
1. The script wb-weight computes the weight by empirical estimation and sta-
tistical data.
2. The script rcogs calculates the centres of gravity of each component.
3. The script riner is responsible for the calculation of the inertia moments.
4. The script rweig obtains the global centre of gravity with regard to the
MTOW (Maximum Takeoff Weight) and MEW (Manufacturer’s Empty Weight)
22
Figure 2.13: Graphical view of the different centres of gravity of the Transonic
Cruiser TCR.
2.2.5 AMB-CFD (Aerodynamics)

Up to now, the available data considered in this design stage consisted mostly in
tabulated data from previous experiences or in linear fluid mechanics assumptions,
leading to errors when sizing the aircraft components. Wind-tunnel measurements
are usually available only late in the design process so, for obtaining reliable data
for expanded flight envelopes and augmented stability early in the process, Com-
putational Fluid Dynamics (CFD) should be used. CFD methods are able to
describe non-linear flight dynamic behaviour of the aircraft. In the AMB-CFD
module (Aerodynamic Model Builder), tabular data with several parameters val-
ues can be obtained and then generate an aerodynamic database with forces and
moments for flight dynamic analysis.
A MATLAB graphical interface was developed to enhance the available aero-

dynamic sources included in this module. This interface preprocesses the data flow
generated in the AMB module and generates suitable input files to the particular
CFD solver used. When the CFD work is finished, the output is postprocessed
and converted in a format compatible with the Stability and Control tools in-
cluded in CEASIOM (see SDSA subsection later, 2.2.7). An illustration of the
CFD interface included in this module is presented in figure 2.14.
Within this module in CEASIOM, there are four different available methods
23
Figure 2.14: Interface between the AMB module and a generic CFD solver.
for calculating the results, depending on the fidelity required :
1. DATCOM (Data Compendium) : Handbook methods which have shortcom-

ings in the transonic speed envelope and is only available for the types of
aircraft that exist in the handbook. It was developed by the United States
Air Force (USAF) [49] as a Stability and Control Data Compendium. In
the 1960s it started as a compilation of more than 1500 pages of analytic
and semi-empirical formulas collected in notebooks. In the 1970s the devel-
opment of the digital DATCOM started based on Fortran. It estimates the
aerodynamic derivatives based on geometric details and flight conditions.
DATCOM calculates the forces and moments acting on the aircraft’s indi-
viual components after adding the flight conditions to the data file. This
method interprets the input file from the AcBuilder module and formats
the geometric descriptions. It should be only used for conventional aircraft
configurations, therefore it was not used for this project.
2. TORNADO3 : Steady or unsteady and based in the potential Vortex Lattice

Method (VLM) corrected by the strip theory, it can be used for low-speed
aerodynamic cases and aero-elasticity calculations. It is implemented as an
open source in MATLAB and uses a modified horse shoe vortex singularity
method for computing steady and low reduced frequency at time-harmonic
3
TORNADO was developed as a cooperation between the Royal Institute of Technology
(KTH), University of Bristol and Redhammer Consulting Ldt. The code is intended for linear
aerodynamic wing design applications, in conceptual aircraft design or in aeronautical education
24
flows over wings. Curved surfaces (such as airfoil camber and leading edge
control surfaces) are generated by surface normal rotation. Trailing edge
devices (TED) deflections are generated by mesh deformations during the
modification of the horseshoe vortices. Zero-lift drag is modelled according
to Eckerts flat plate analogy. Viscosity can be approximately taken into
account by an empirical extension and incompressible fluid conditions can
be calculated using the Prandtl-Glauert correction. More information about
this software is presented in section 5.1.
3. Edge : Inviscid parallelised 2D/3D CFD Euler code used for high-speed
aerodynamic cases and aero-elasticity calculations, including compressibility
effects. Edge is based on Navier-Stokes equations for viscous and inviscid
compressible flow problems in unstructured grids. Edge is an edge-based
formulation which uses a node-centred finite volume technique. Different
turbulence models are available for steady-state and time accurate calcu-
lations including maneuvers and aeroelastic simulations. Edge applies the
Euler mode with preset values for the maximum and minimum deflection of
each control surface and creates aerodynamic tables with the possible com-
binations of control surfaces deflection angles. The aerodynamic of control
surface deflections is computed by using the transpiration boundary con-
ditions instead of deforming the grid (with a limitation on the amount of
maximum and minimum deflections). The input files for Edge are volume
meshes generated by SUMO and TetGen. Edge is available as a complete
source package, subject to the FOI license agreement, in FOI website [51].
More information about Edge can be found in [14].
4. RANS (Reynolds Averaged Navier-Stokes) : high-fidelity flow simulator for

extreme flight conditions analysis including viscous effects. The input files
are the volume mesh and an IGES file generable by SUMO. The compu-
tational cost of this method is too high for this research, so it will not be
further explained.
The choice of the most suitable method is a compromise between fidelity and
computational cost (time). Depending on the flight envelope of the aircraft, the
applicable method has to be selected by the user. In preliminary design, for low-
speed aerodynamics and within the limits of the fligh envelope, Tier 1 module
(TORNADO and DATCOM) provide acceptable aerodata. This module generates
25
satisfying results for cases with low angles of attack. Tier 1+ module (Edge)
is suitable for computing aerodata in the transonic flight case and provides a
improved design of the flight control system. Finally, Tier 2 module uses RANS
methods for higher fidelity computations and for nonlinear aerodynamic cases. An
scheme about this three options is shown in figure 2.15. Regarding the adaptative
fidelity geometry modelling requirements, figures 2.16 and 2.17 are presented.
Figure 2.15: Adaptable fidelity options within the AMB-CFD module in CEA-
SIOM.
In this project, the VLM code TORNADO was used, due to the low-speed
and low-angle of attack conditions of the analysis. Therefore, the mathematical
models that consider linear aerodynamics are suitable for this model. This tool
achieves reasonable accurate results with low computing time, compared to the
much larger time required by higher fidelity tools.
2.2.6 Propulsion
In the propulsion module, thrust is calculated for several Mach Numbers and
different altitudes selected by the user. It generates a database of the engines
which is required in the SDSA module for the stability and control calculations.
This module comes after the AMB-CFD module and the information generated
is added to the XML data file.
26
Figure 2.16: Adaptative fidelity geometry modelling requirements in CEASIOM.
2.2.7 SDSA (Stability and Control)

The SDSA (Simulation and Dynamic Stability Analyzer) module allows a six-
degree-of-freedom flight simulation about stability and control, and assess about
the flying quality and performance. It requires the aerodynamic coefficients for
the total flight envelope obtained in the AMB-CFD module as well as the center
of gravity coordinates and inertia moments calculated in the Weights & Balance
module. The user can introduce the initial conditions and weather conditions
through the system Graphical User Interface (GUI). Disturbances and single or
double step controls also have to be considered for the flight test settings.
The SDSA module considers the following physical model [16]:
• The aircraft is a rigid body with 6 degrees of freedom: 3 translations along

the axis (x, y, z) and 3 rotations (pitch, roll and yaw).
• Control surfaces are movable but cannot perform free vibrations
• Aerodynamics are seen as quasi steady.
• A standard, undisturbed, windless atmospheric model is considered.
The coordinates systems used to describe the motion of the aircraft are the fol-
lowing ones:
27
Figure 2.17: Different levels of complexity geometry requirements for different

CFD tools fidelity levels. [15]
• A fixed coordinate system of the earth: O1 − X1 Y1 Z1
• A movable coordinate system parallel to the gravity coordinate system, with

origin at 1/4 of the MAC (Mean Aerodynamic Chord): Og − Xg Yg Zg
• The body coordinate system, with OX parallel to the MAC and pointed
forward, OZ oriented down and OY pointed to the right wing and the same
origin as the movable coordinate system (O ≡ Og ): O − XY Z
• A velocity coordinate system, created by the rotation of the body coordinate

system around the OY axis (angle of attack α) and around the OZ axis
(sideslip angle β) and with OXa parallel to the free stream. The origin is
setted by the user, but usually it is placed at 1/4 of the MAC because a
lot of aerodynamic characterisctics are referred to this point: Oa − Xa Ya Za .
This coordinate system is presented in figure 2.19.
To perform the transformation from the Og − xg yg zg coordinate system to the

body coordinate system, the following rotations in this order should be followed:
1. Rotation around the OZg axis: Yaw angle Ψ
2. Rotation around the new OY axis (after yaw rotation): Pitch angle Θ
3. Rotation around the new OX axis (after yaw and pitch rotation): Roll angle
Φ
28
All these coordinate systems and angles and the components of the main velocity
vector Vo (U, V, W ) and the main angular velocity vector Ω(P, Q, R) are shown in
figures 2.18 and 2.19.
Figure 2.18: SDSA Coordinate systems to describe the position and the attitude
of the aircraft. [16]
Figure 2.19: SDSA Velocity coordinate system. [16]
All these coordinate systems are used to obtain the kinematic relations and
create propper equations in an easier way. The relation between the coordinates
of the gravity inertial system and the linear velocities defined in the body axis is:
   

ẋ1  
U 
ẏ1 = Av ·  V (2.1)
   
  
   
ż1 W
29
 

cos Θ cos Ψ cos Ψ sin Θ sin Φ − cos Φ sin Ψ sin Φ sin Ψ + sin Θ cos Φ cos Ψ 
AV =  cos Θ sin Ψ sin Ψ sin Θ sin Φ + cos Φ cos Ψ sin Θ cos Φ sin Ψ − cos Ψ sin Φ
 

 
− sin Θ sin Φ cos Θ cos Φ cos Θ
(2.2)
The relation between the quasi Euler angles and the aunglar velocity is also
presented:
   

Φ̇  
P 

 Θ̇

 = AΩ · 
 Q

 (2.3)
   
Ψ̇ R
 

1 tan Θ sin Φ − cos Φ sin Ψ cos Φ tan Θ 
AΩ = 
 0 cos Φ − sin Φ

 (2.4)
 
sin Φ cos Φ
0 cos Θ cos Θ
The equations of motion, based on the balance of forces and moments are
detailed in [16] and a user guide is found in [17]. The calculated forces are summed
up in Taylor series.
The structure of the SDSA module is based on the stability analysis. The
scheme for this analysis is presented in figure 2.20. To get the results, the aircraft
mathematical model is transformed into a matrix form. The non-linear equations
of motions are formulated and linearized computin the Jacobian matrix of state
derivative around the trimmed condition (state of equilibrium). The eigenvalues
and eigenvectors for the equilibrium state are then computed. Finally, the modes
of motion and the stability characteristics (damping and frequency coefficients,
damping ratio, period, etc.) are calculated.
The module Equilibrium state computation included in SDSA, defines the
eigenvalue problem states and computes the initial conditions in the flight sim-
ulating module, setting the time derivative to zero. Therefore, the non-linear
equation from the equations of motion can be solved. Using the flight simulating
module, the flight parameters can be computed in real time. Stability character-
istics can be verified by using a full non-linear model. The scheme for the fligh
simulation is presented in figure 2.21. It is noticeable that this scheme shares
several modules with the previous computation.
30
Figure 2.20: SDSA Stability analysis scheme.
The input data provided by the AMB-CFD module is presented in a table

similar as the one in table 2.1 and includes nine flight condition parameters as
independent variables (angle of attack,α, Mach number, M, sideslip angle, β, the
three rotation parameters in pitch, Q, yaw, P, and roll, R, and the deflections
of the elevator,δe , rudder, δr , and aileron,δa ). The variables α, Mach and an
arbitrary variable define a set of 7 three-dimensional tables for obtaining the six
aerodynamic coefficients (CL , CD , Cm , CY , Cl , Cn ). Each input variable should
at least consist of two different values, so the SDSA table contains at least 60
calculated cases. For more accurate results, more entries (and more computational
cost) is required.
This module includes the following functionalities4 (shown in figure 2.22):
• Stability analysis: with eigenvalues analysis of linearized model in open and

closed loop case. For the eigenvalues analysis, the model is linearized by
computing the Jacobian matrix of the state derivatives around the equi-
librium (trim) point numerically. It also allows to obtain the time history
4
A more detailed description of each tool can be found in [16]
31
Figure 2.21: SDSA flight simulation scheme.
identification using the nonlinear model.
• Six Degree of Freedom flight simulation: test flights including trim response
that can record flight parameters in real time and turbulence.
• Flight Control System (FCS) based on a Linear Quadratic Regulator (LQR)

approach: calculates control matrices of the total flight envelope and provide
results for simulation or stability analysis. Other options are the human pilot
model or the Stability Augmentation System (SAS).
• Performance prediction (flight envelope, maneuvers, range, endurance) and

other issues (data review, results review, cross plots).
With the recorded data, the SDSA module can provide the phugoid, period,
damping coefficient, phase sift and short period characteristics. It can also provide
the trimmed angle, the trimmed angle of deflection and the drag coefficient. The
stability analysis results are presented as ’figures of merits’ based on JAR/FAR,
ICAO and MIL regulations. A more extended explanation of the Stability and
Control fundamentals is presented in
32
α M β Q P R δe δr δa CL CD Cm CY Cl Cn
x x x - - - - - - x x x x x x
x x - x - - - - - x x x x x x
x x - - x - - - - x x x x x x
x x - - - x - - - x x x x x x
x x - - - - x - - x x x x x x
x x - - - - - x - x x x x x x
x x - - - - - - x x x x x x x
Table 2.1: Aerodynamic database, output of the AMB-CFD module.
Figure 2.22: SDSA structure and functionality.
2.2.8 NeoCASS (Aeroelasticity)

The NeoCASS (Next Generation Conceptual Aero-Structural Sizing Suit) mod-
ule helps to dimension the structure of the aircraft under development processing
quasi-analytical structural analysis and taking into account its static and dynamic
aeroelastic behaviour. NeoCASS provides more accurate details of the bearing air-
frame (such as stiffness and mass distribution, global weight) in the first design
steps, which are usually poorly represented by the structural weight data from
empirical statistical formulas. Thus, the code can be used to design innovative
and non-conventional aircraft configurations.
The aero-structural analysis is based on a combination of computer related,

analytical and semi-empirical methods. NeoCASS also includes [46]:
33
• Flutter solver to efficiently assess flutter (and divergence) instabilities in the

whole flight envelope static aeroelastic.
• Vortex Lattice Method (VLM) for steady subsonic aeroelastic analyses de-
veloped starting from Tornado code.
• Doublet Lattice Method (DLM) for oscillatory subsonic stability derivatives

and Aerodynamic Influence Coefficients (AIC) matrix calculation required
for flutter assessment.
• Identification of the beating of the wing and analyses for deformed flight
shape calculation.
• Analytical structural sizing at different maneuvering flights and ground tax-

ing coming from NASA-PDCYL code; this first solution is merely used as
guess and starting point for the different numerical aeroelastic solvers.
• Structural optimization loop to satisfy user defined criteria.
• Linear and non-linear beam elements with distributed and lumped masses
for efficient structural analyses of high-aspect ratio airframes.
• Moving Least Squares (MLS) and Radial Basis Functions (RBF) spatial
coupling methods for data transfer between non-matching structural and
aerodyanamic meshes.
• Vibration modes solver (results are exported in FFA-format files for Edge
CFD solver when high-fidelity aeroelasticity is required)
Regarding the structure and the data flow inside it, figure 2.23 is presented. Neo-
CASS includes two different programs: GUESS and SMARTCAD:
GUESS The data from AcBuilder module (geometrical, aircraft.xml and geom-
etry.xml, and technical information, techno.xml) and from the Weight and Balance
module (parameter.xml) is introduced in GUESS for the structural model and a
first estimation of the weight, respectively. In addition, a file with the aircraft’s
conditions (states.xml) is also provided to GUESS. This program balances empir-
ical and finite element methods and its main aspects are loads and sizing. In the
end, the corrected mass and stiffnes distributions, an analytical mesh and a stick
model are outputted as an ASCII file.
34
2.3. CPACS, A NEW SOLUTION
SMARTCAD The output of GUESS is introduced in SMARTCAD, as shown

in figure 2.23, which contains numerical aero-structural analysis based on beam
models and VLM/DLM aerodynamics. This program also includes stabilizer static
analysis, vibration mode calculations, linear buckling and linearized flutter anal-
ysis. The user can choose between rigid or deformable aircraft model and steady
or unsteady aerodynamic analysis.
Figure 2.23: NeoCASS structure and data flow. [18]
A more detailed description of NeoCASS and more information can be found

in reference [18] and in its website [52], were it can be downloaded and used in
MATLAB under GNU GPL conditions.
The major dataflow between the CEASIOM modules is based on an XML

format. Including necessary information about the description of the aircraft, the
data volume extends within the conceptual design process. In the next section
CPACS, a new type of XML file, more general and including more details is
explained.
2.3 CPACS, a new solution

Aircraft design process often requires collaboration between disciplinary experts
independently of geographical barriers. There are already several tools to address
35
each different design discipline. However, these tools are usually isolated and do
not share information between them, making the design process slower and less
practical. Therefore, dataflow and workflow should be shared as efficiently as
possible. For this purpose, an engineering framework software is required to link
all the disciplines shown in the table of figure 2.3 and launch them from an efficient
system-level interface. A design environment enables engineers and experts to
cooperate using structured and mostly autonomous exchange of information. A
more detailed view of this issue is well explained in by Rizzi et al. [8].
Figure 2.24: A design environment includes a common language, several analysis

modules and an integration framework.
Since 2005 the DLR has been developing the Common Parametric Aircraft
Configuration Scheme (CPACS), a standard syntax definition for the exchange of
data within conceptual and preliminary aircraft design stages, including all the
parameters necessary to describe the aircraft configuration and tool specific data
for the connected analysis modules. This system is in operational use at all aero-
nautical institutes of DLR and has been extended for civil and military aircraft,
rotorcraft, jet engines and entire air transportation systems [8]. In figure 2.24,
CPACS would play the role of the common language with the modules of CEA-
SIOM. Design teams in various institutions benefit not only from standardized
product descriptions but also from identical coupling of analysis modules. It is
therefore a driver for multi-disciplinary and multi-fidelity design in distributed
environments.
In this unified data-centric setup, the number of interfaces required decreases
36
to a minimum, establishing a more effective and flexible communication between

the modules. Figure 2.25 shows the main advantage of using a unified data man-
agement, the decrease of the number of interfaces required from N(N-1) to 2N.
The inputs and outputs of every module are linked to the common namespace
(CPACS) via wrapper scripts that translate the information. CPACS offers bene-
fits in efficiency and agility and provides a more unified geometry management in
CEASIOM than AcBuilder module. Hence, it provides a more unified version of
CEASIOM, eases the extension and adaptation of the toolset and mitigates the
problems encountered in handing over the model to later design phases. CPACS
is designed so that it is prepared for all eventualities which might occur in future
aircraft congurations.
(a) (b)
Figure 2.25: Advantages of unified data management, such as CPACS, in a multi-
module framework. On the left (a), a framework without unified data and N to N
interfaces = N(N-1) interfaces. On the right (b), a framework with unified data
and a hub with N to 1 interfaces = 2N interfaces.
CPACS describes the characteristics of aircraft, rotorcraft, engines, climate

impact, airports, flight plans, fleets and mission in a structured, hierarchical man-
ner (see figure 2.26). Geometric elements such as aircraft, wings, sections, ele-
ments, profiles, points, and transformations are defined covering the bandwidth
of the named projects. Not only product but also process information is stored in
CPACS, helping to set up workflows for analysis modules.
CPACS offers the following advantages compared to the old CEASIOM XML
files:
37
• More specific details included
• More versatile geometry definition
• Various coupled tools
• Task specific
• Supporting libraries
• High accessibility
• Comparably small
2.3.1 CPACS XML file structure

CPACS is based on XML schema files (XML Schema Definition XSD[53]), mak-
ing it human readable and computer processable. The XML/XSD technology also
provides validation of data files, mitigating the risk of creating inconsistent data
sets. Therefore, no manual editing is foreseen. It holds an object-oriented struc-
ture to increase the modularity and to distinguish aspects such as file handling
and convergence control from design aspects, such as parameter definitions (in-
cluding empirical and analytical knowledge) and calculation methods (for process
control). CPACS XML files contain detailed type definitions of the air transporta-
tion system and other data of all the disciplines mentioned in Nickol’s table. The
main levels of the structure are components and parameters containing general
routines. CPACS could provide the unified data format for export to complete
“Product Life-Cycle Management” software.
Components are structured in a hierarchical way and linkage between the ele-
ments is feasible using the Unique IDentifier (uID) of the elements. Consequently,
the analysis modules are able to interact with each other in a progressional way
and the framework allows splitting and merging of CPACS data sets. Each com-
ponent includes several parameters and attributes grouped by disciplines. The
complexity of the CPACS structure is variable and dependent on the level of de-
tail considered in the specific approach. Hence, CPACS is able to define aircrafts
with several levels of fidelity and complexity. Figure 2.26 illustrates the structure
of a CPACS XML file.
38
Figure 2.26: CPACS XML file hierarchical structure.
For defining the parametric geometry, the most important components are
aircraft, engines and profiles, defined inside the 2nd level component vehicles.
The 3rd level component engines includes information about the engine type,
comprising geometric data, thurst and bupass ratio. The shapes of the components
are defined in profiles. This component is subdivided in fuselageProfiles, described
by fuselage sections defined by a point list, and wingAirfoils, which are described
by a point list normalized from 0 to 1. Each profile has its own profile uID. Some
of the elements and attributes included in the aircraft component are described
in the following list:
uID: element Unique IDentifier that allow linkage between elements, such as
positioning or parenting. Each element has its own different uID.
39
Name: element name
Description: description of the element with characteristics and comments.
Reference: reference length, area and reference point of the aircraft.
Fuselages: fuselages of the aircraft defined using the fuselage sections defined in
profiles. Each section is defined with a positioning, orientation and scaling
in the three axis.
Wings: wings of the aircraft defined using the wing airfoils defined in profiles.
Each section is defined with a positioning, orientation and scaling in the
three axis. Further information about the control surfaces, fuel tanks and
structural elements (spars and ribs) can be defined as subcomponents.
Engines: engines of the aircraft defined using the engines defined in the 3rd level
component engines. Each engine is defined with positioning, orientation and
scaling in the three axis.
Landing gear: information about the landing gear and its position and size.
Systems: description of every important system included in the aircraft.
Global: information about the aerodynamic performance map, passenger seats,

landing weight, and Mach number at cruise point
Analysis: weight breakdown of the aircraft.
Every subelement of the vehicles element can have several attributes and any
number of subelements as defined in the XML fille. The attributes are for ex-
ample the name of the element, a description, the element uID which is used for
crossreferences with other elements within the file and also a symmetry where an
element can be defined to be symmetric to a plane or axis. The parameters of
the subelements are determined within their own subelement coordinate system
which is determined by rotating, translating and scaling it relative to the parent
coordinate system. Therefore, there is an order to follow (much more flexible than
in AcBuilder) so it is not possible to add an element before its parent is deter-
mined. For example it is impossible to create the wing inner structure (spars and
ribs) before the wing is defined. Components on the same level of hierarchy can
40
Figure 2.27: CPACS subelement structure.
be added and sized in any order. Figure 2.27 illustrates an example of the data
structure inside a defined wing.
An additional explanation about the wings definition must be provided. First
of all, an airfoil must be loaded as a point list into a profile. It is very important
that all the profiles have the same amount of points defined in the same order
due to surface continuity reasons required in CC. Profiles are then referenced by
elements, where operations such as translations, scalling and rotations of the
airfoil can be done. An example of elements is shown in figure 2.28. A section
consists of one or several elements and at least two sections must be provided to
define a wing. To define more than one element in a section makes sense in case
of steps in the wing or if the wing splits up into two wings. The sections can be
located either by a transformation (translations, rotations and scalations) with
respect to the wing coordinate system, or by a positioning (lenght, dihedral angle
and sweep angle) relative to another section already defined, as shown in figure
2.29. For parametric geometry input the user must be aware of the coordinate
system. The conguration is a tree structure and child components inherit global
coordinates from parents, and be sized relative to their dimensions. This is a
common practice, for example the control surfaces are children of the wing they
are defined in.
41
Figure 2.28: CPACS wing elements definition. [19]
Finally, the wing segments are defined as the junction of two wing elements
(defined in CPACS as “From” element and “To” element) and the volume between
them. An example of segments definition is illustrated in figure 2.30. If two or
more segments are combined, a component segment is created. The inner
structure of the wing (spars and ribs), the control surfaces and the fuel tanks
are defined in CPACS within the wing component segments. All the profiles,
elements, sections, etc. are referenced by their uID. A very illustrative example
video by D. Böhnke for defining a wing from scratch can be found in the CPACS
website, [54, 55].
In each component segment a new coordinate system is defined, the eta-ksi
coordinate system:
Eta represents the normalized span part. The eta-axis is determined by

the projection of the component segments leading edge onto the global
y-z-plane. The eta values are then set to be zero at the start element
of the component segment and one at the end element. In between
those two values a linear scale is used.
Ksi represents the normalized chord of the component segment. The ksi-
axis is perpendicular to the eta axis and passes through the foremost
and the sternmost point of the start element airfoil. Ksi is zero at
the foremost point of the airfoil and one at the sternmost point. In
between those two values a linear scale is used.
42
Figure 2.29: CPACS wing sections definition. [19]
Additionally to the ksi and eta coordinates a relative height coordinate is intro-
duced. This axis is perpendicular to the other two and always zero at the lower
skin of the wing and one at the upper skin. In between those two values a linear
scale is used.
An example of component segments and the eta-ksi coordinate system is pre-
sented in figure 2.31.
The fuselage definition is very similar than the wing case, and they share the
same structure. The main difference is the coordinate system used. In this case,
it is the same as the global for the whole aircraft with the X axis in the fuselage
longitudinal direction, the Y axis perpendicular to the X axis and pointing to the
right semiwing and the Z axis normal to both of them and pointing upwards, as
illustrated in 2.32. The fuselage cross sections are loaded as point lists into a pro-
file. These profiles are referenced by elements, in a similar way than explained
before (see figure 2.33). A fuselage section also includes one or several elements
and at least two sections must be provided to define a fuselage, as explained before.
Once again, the sections can be located either by a transformation (translations,
rotations and scalations) with respect to the fuselage global coordinate system,
or by a positioning (lenght, dihedral angle and sweep angle) relative to another
section already defined, as shown in figure 2.34. The fuselage geometry is usually
simpler than the wing geometry, being unnecesary to consider sweep angles or
dihedral angles. Finally, the fuselage segments are defined as the junction of
two fuselage elements (defined in CPACS as “From” element and “To” element)
43
Figure 2.30: CPACS wing segments definition. [19]
and the volume between them. An example of segments definition is illustrated

in figure 2.35.
The user interface of CC for a fuselage example definition is presented in figure
2.36, where the geometrical data can be introduced by the user and inmediatly
visualize the graphical changes. Three different modes can be used: Add, remove
or modify sections.
2.3.2 CPACSCreator
As stated previously, an additional problem for designing unconventional configu-
rations is that softwares like CEASIOM are not programmed for unusual geome-
tries, so new software options should be considered. CPACSCreator (CC) is now
introduced, which is being developed nowadays and does not have the geometry
restrictions from AcBuilder from CEASIOM or other softwares used before due
to the advantages of using CPACS explained before.
The main change from the previous software CEASIOM is made in the geom-
etry module (previously AcBuilder) that is subsituted by the CC. The advantages
of compared to AcBuilder are the possibility of having a more general and versa-
tile geometry design without the constraints of usual aircrafts that AcBuilder has
(i.e. the requirement of always having one cylindrical fuselage, reduced number of
wing kinks (2) and control devices, etc). CPACS data format manages all geomet-
44
Figure 2.31: CPACS wing component segment definition. [19]
Figure 2.32: CPACS global geometry coordinate system. [19]
rical congurations, enables multi-fidelity structural analysis modules and loosely

coupled optimization procedures to be applied to problems which require several
analysis modules combined. Other major difference with AcBuilder is that CC
is programmed enterely in MATLAB, including the rendering, making the code
modification and extension easier. The data flow scheme inside CC is presented
in figure 2.37.
The main issue is the interface of CC with the rest of modules from CEASIOM.
In other words, how to make the output/input from/for CC compatible with the
rest of modules of CEASIOM. By now, both of them work with XML files but
with different formats, so it is not possible yet to load an output XML file from
CC directly into CEASIOM. Therefore, wrappers to allow the connexion between
CC and the other modules of CEASIOM are being developed.
45
Figure 2.33: CPACS fuselage elements definition.
Some of them have already been developed before, such as the interface for
exporting CC files as sumo files which enables the use of the CFD program Edge for
the aerodynamic analysis. Other interfaces, such as the interface with TORNADO
software and the Weight and Balance module was developed and improved in this
work. Once the CC file has been processed in TORNADO and some format
changes are made, the file can be used in the rest of the modules in CEASIOM,
such as SDSA.
Therefore, the scheme of the modules and their connexions is approximately
like the one shown in figure 2.38.
Some of the unconventional aircraft configurations modelled using CC are
shown in figure 2.39, but the design possibilities are countless.
46
Figure 2.34: CPACS fuselage sections definition.
Figure 2.35: CPACS fuselage segments definition.
47
Figure 2.36: CPACSCreator user interface for an example fuselage definition.
Figure 2.37: CPACSCreator data flow scheme.
48
Figure 2.38: Scheme of modules using CC instead of AcBuilder.
Figure 2.39: Examples of unconventional aircraft configurations that can be cre-

ated using CPACSCreator: Box Wing concept (left) and Strut Braced Wing con-
cept (right)
49
50
Chapter 3
The Blended Wing Body concept
Taking a look into aircraft history, one can see that after the remarkable improve-
ment and design alterations done for the Boeing B-47 in 1947 , in comparison to
the first aircraft (the Writght Flyer in 1903), no more major changes have been per-
formed in the conventional aircraft configuration. Indeed, the swept-wing Boeing
B-47 includes most of the fundamental design characteristics of a modern subsonic
jet transport aircraft, also known as the Cayley functional decomposition:
• Slender fuselage.
• Swept wing with high aspect ratio.
• Alf-mounted empennage.
• Pod-mounted engines hung on pylons beneath and forward the wing or at-
tached to the rear fuselage.
In figure 3.1 it can be appreciated that the Airbus 330, designed 45 years after the
Boeing B-47, seems essentially similar. Although the design concept has improved
with time, its core design has not changed. There has not been a revolution in its
design for over half a century.
This paradigmatic configuration seems to be approaching its evolutionary po-
tential and an asymptotic design around the size of A380 [21]. New configurations
and designs are being developed in order to search new options and solutions for
air traffic demands. One of them is the Blended Wing Body (BWB) which is
based in another streamlined canonical form, a disk. In figure 3.2 three canoni-
cal forms are presented, each sized to hold the same number of passengers (800
51
CHAPTER 3. THE BLENDED WING BODY CONCEPT
Figure 3.1: Aircraft design evolution. From left to right: the Writght Flyer, the
Boeing B-47, the Airbus A330)[20]
in this example). The sphere has obviously the minimum area of all but is not
streamlined. The two other options, the conventional cylinder and a disk, are
streamlined and have nearly similar surface area (4% lower for the disk).
Figure 3.2: Cannonical forms and the effect of body type on surface area.[20]
After adding the rest of the aircraft elements, this difference becomes larger.
This way, after adding the wing the second option has 20% less surface; after
installing the engines, the difference increases to 25% and finally, after integrating
the control devices, the disk configuration’s surface is approximately 33% lower,
as shown in figure 3.3:
However, the idea of using a flying wing only in order to achieve an efficient
plane is not new. In fact, it has been the subject for much research ever since the
early days of aviation. The BWB concept is meant to combine the advantages
of the flying wing with the loading capabilities of a conventional aircraft, by
52
3.1. MAIN FEATURES OF THE BWB
Figure 3.3: Surface difference between the conventional configuration and the
BWB configuration.[20]
increasing the volume in the center of the wing to allow space for passengers
and cargo. The centerbody must carry the pressure load from the cabin and the
bending load of the wing. An approximate comparison of the loads distribution
in a BWB aircraft and a conventional aircraft is presented in figure 3.4. The peak
wing bending moment and shear force for the BWB is approximately one half of
the conventional configuration.
Figure 3.4: Comparison of aerodynamic, inertial and cabin pressure loads for a
conventional configuration (left) and a BWB (right)[20]
3.1 Main features of the BWB

The BWB is a tailless aircraft design that integrates the wing and the fuselage
into a single lifting surface, by the thickening of the wing in the central part of
53
the airplane. This concept reduces the wetted area to volume ratio and lower
the interference drag compared to conventional aircrafts. The BWB configuration
also increases the volume of the cabin and could provide significant fuel savings
(and hence lower pollution). Noise reduction and safety against failure could be
also be improved if a number of engines are located above the wing.
Authors such as Qin, [22], estimated an increase in the maximum lift to drag
ratio, (L/D)max , of about 20% over the conventional aircraft configurations. This
ratio can be estimated by the following equation:
s s
L πAR π

= =b (3.1)
D max 4kCD0 kSD0
The definitions of each parameter are:
A ≡ aircraft aspect ratio.
CD0 ≡ zero-lift drag.
k ≡ induced drag factor.
b ≡ aircraft span.
SD0 ≡ zero-lift drag area.

With this relation, it can be seen that larger span, reduced wetted area, lower
zero-lift drag and less induced drag could provide substantial improvement in the
aerodynamic performance, represented here by the maximum lift to drag ratio.
Nowadays, considering the current airport capability, the maximum span would
be limitted to 80 m (ICAO F category).
Liebeck et al. [20] concluded that a reduction of about 28% fuel burn pear seat
and a 15% reduction in takeoff weight could be achieved for a BWB configuration
with 800 passengers with 85 m span and a trapezoidal aspect ratio of 10. All in
all, this could result in 10-15% lower direct operational costs.
In 2000, NASA Dryden Flight Research Center made an estimation of the

potential benefits of the BWB configuration compared to conventional aircrafts
for the year 2015. Some of these improvements are presented in figure 3.5 and
were:
54
3.1. MAIN FEATURES OF THE BWB
• 12% lower Direct Operational Cost (DOC).
• 21% higher fuel efficiency.
• 6% lower gross weight.
• 17% lower emissions of NOx.
Figure 3.5: Potential benefits of the BWB configuration compared to conventional

aircrafts in 2015. [23]
The BWB concept also have some drawbacks, due to the stronger coupling
with other disciplines such as structures and flight dynamics. Stability and con-
trol become crucial issues in the design process, because of the tailless quality of
the configuration. Control surface sizing and allocation are cricital issues. Con-
trol surfaces should be multiple, large and rapidly moving and try to minimize
the hinge moments. The winglets provide directional stability and control and in-
crease the effective aspect ratio. Several models, such as the presented by Liebeck
in [20], had a -15% static margin, requiring a fly-by-wire control system.
One option for the position of the engine inlets is to place them right down on
the wing surface, so they ingest the boundary layer and any airflow absorbed into
the engines can be ignored as drag. This could give a great increase in the L/D
55
due to the decrease in drag. Production costs are difficult to estimate, because
this concept has lower number of tight bends but the body design makes mass
manufacturing difficult and expensive, so there is a trade off situation.
All in all, the BWB concept presents the following advantages:
• Reduced wetted area to volume ratio ⇒ Reduced drag ⇒ Reduction of

the fuel burn ⇒ Lower pollution and reduced environmental impact1 (for
example gNOx/Pkm).
• Lower operational cost.
• Structural advantages due to the integration of the wing structure with the
thick centre body.
• Reduced wing loading because the centre body also provides lift, avoiding
wing tip stall and aileron ineffectiveness.
• Reduction of the aircraft total height.
• Lower takeoff weight for the same volume.
• Reduced peak wing bending moment and shear force.
• Significant payload advantages in strategic airlift, air freight and aerial re-
fueling roles.
• Reduced airspace and airport capacity demand per transport perfomance.
• Commonality: allowing both civilian an military applications.
• Engines can be located above the wing and require lower engine speed for
landing ⇒ No need for high lift devices ⇒Noise reduction.2
• High comfort cabin and faster boarding and de-boarding.
1
The IPCC has estimated that aviation is responsible for around 3.5% of anthropogenic
climate change, a figure which includes both CO2 and non-CO2 induced effects. The IPCC has
produced scenarios estimating what this figure could be in 2050 going from 5% to 15% (if no
action is taken) of the total contribution by 2050. [24]
2
NASA audio simulations show a 15.5dB reduction compared to Boeing-777-class aircraft.
(http://www.aviationweek.com)
56
3.2. PREVIOUS MODELS
And the following disadvantages:
• Difficult controlability and stability.
• Not enough previous experience, statistics nor flight tests from previous
BWB commercial programs to take as reference.
• The lack of windows could be an issue for the safety conditions and passenger
acceptance.3
• Potential problem with pressurization of a non-cylindrical fuselage.
• Very thick airfoils in the passenger cabin can be challeging for the transonic
regime.
3.2 Previous models

When trying to define the dimensions and geometry constraints for the cabin
and the whole aircraft, looking for references of previous similar aircrafts is really
helpful to get a first insight about the order of magnitude of the parameters.
However, the Blended Wing Body (BWB) configuration is a recent idea, so there
are not so many references available, and even less experimental data. Some of
the most famous models, like X48-B from NASA, Horten Ho 229 or the MOB
BWB are analyzed and compared in the following sections.
3.2.1 Horten Ho 229

The Horten Ho 229 was a German prototype fighter and bomber built late in the
World War II (1944). It was the first pure flying wing powered by jet engines and
had a 37% reduction in detection range against the radars of the 1940s, but not
anymore with the contemporary radar systems. Only 3 prototypes were built, and
only one survived to the war and is now in American hands. Some characteristics
of this prototype are presented in table 3.1:
3
Al Bowers, Senior Aerodynamicist for NASA at the Dryden Flight Research Center, sug-
gested to place a multi-functional LCD screen on the seat in front of each seat. A selector would
allow the passenger to select from a number of views, including looking to the rear and straight
down. (Blended-Wing-Body: Design Challenges for the 21st Century) [23]
57
(a) (b)
Figure 3.6: Horten Ho 229 illustration (a) and sketch (b)
Parameter Value
Wing span 16.76 m
Wing area 50.20 m2
Length 7.47 m
Height 2.81 m
Maximum speed 977 km/h
Range 1900 km
Service ceiling 52000 ft
Maximum takeoff weight 8100 kg
Empty weight 4600 kg
Thrust to weight ratio at take-off 0.26
Table 3.1: Horten Ho 229 characteristics.
3.2.2 Northrop YB-49
The Northrop YB-49 was an American prototype jet-powered heavy bomber air-
craft developed by Northrop Corporation shortly after World War II. Only 2
YB-49 were built, from converted YB-35 for testing. This aircraft is considered
to be a Flying Wing (not strictly a BWB), but it shares some aspects with the
BWB configuration. Some characteristics of this aircraft are shown in table 3.2:
58
Figure 3.7: Northrop YB-49
Parameter Value
Wing span 52.4 m
Wing area 371.6 m2
Length 16 m
Height 6.2 m
Aspect ratio 4.1
Range 16057 km
Service ceiling 45700 ft
Maximum takeoff weight 87969 kg
Empty weight 40116 kg
Root Airfoil NACA 65-019
Tip Airfoil NACA 65-018
Thrust to weight ratio at take-off 0.23
Table 3.2: Northrop YB-49 characteristics
The design work performed in the development of the YB-49 proved to be

valuable to Northrop decades later (1989) in the eventual development of the B-2
Spirit stealth bomber, which entered service in the early 1990s (see fig 3.8). This
aircraft does not have winglets, due to radar detectability reasons and the stability
in the yaw direction is achieved with split rudders. This aircraft can be classified
as a Hybrid Flying Wing, having similarities with the BWB configuration.
59
Figure 3.8: Northrop Grumman B-2 Spirit
3.2.3 MOB BWB
In 2004, in an European Union project, called Multidisciplinary Design and Op-

timisation for Blended Wing Body (MOB), a multi-disciplinary Computational
Design Engine (CDE) was developed by working groups distributed across Eu-
rope4 . This project was included in the CORDIS Fifth Framework Programme
(FP5) funded by the European Union. The goal was to obtain a full optimal
design from a prototype baseline. The CDE coordinated structural aerodynam-
ics, aeroelasticity and flight mechanics disciplines. More information about this
project can be found in [25], [?], [22], [27] and [28]. Some illustrations of the MOB
BWB are presented in figures 3.9 and 3.10. Additional data is shown in table 3.3.
Figure 3.9: CAD illustration of the MOB BWB configuration [22]
4
The whole member list of the MOB consortium can be found in [25]. Some of them are:
KTH (Sweden), SAAB AB (Sweden), Technische Universitoit Delft (Netherlands), NLR (Nether-
lands), DLR (Germany) and Cranfield University (United Kingdom) .
60
Figure 3.10: MOB BWB pressure distribution simulation. [58]
Parameter Value
Wing span 80 m
Wing area 841.7 m2
Wetted area 3079 m2
Mean Aerodynamic Chord (MAC) 12.31 m
Aircraft mass 371280 kg
Cruise Mach number 0.85
Center of Gravity position 29.3 m
Altitude 11500 m
Table 3.3: MOB BWB characteristics
3.2.4 SAX-40
This aircraft was developed by the Cambridge University and the MIT (Mas-
sachusetts Institute of Technology) Aeronautics and Astronautics Department
within the Silent Aircraft Initiative in 2006. The main project outcome was the
conceptual “Silent Aircraft eXperimental” design SAX-40. The 215 passenger
aircraft has an estimated noise level less than the background noise of a well-
populated area and a predicted 23% fuel burn reduction compared to current civil
aircraft (124 passenger-miles per (US) gallon compared to 101 passenger-miles per
gallon for a Boeing 777). One of the most important aspects of the SAX-40 is
its embedded, boundary layer ingesting propulsion system with ultra-high bypass
ratio (12) engines which reduce the fuel burn and also the noise. The trailing
edge brushes reduce the scattering noise from turbulence near the trailing edges
and, using deployable drooped leading edges, the slat noise is eliminated. These
61
and other devices provide the global noise reduction, making it 25dB quieter than
current aircrafts with a prediction of noise of 63 dBA outside airport perimeter
[59]. Researchers predict that no one beyond the boundaries of a given airport
would be able to hear the SAX-40 taking off or landing. An illustration of this
model is shown in 3.11 and some data is presented in table 3.4.
Figure 3.11: Cambridge University and MIT SAX-40 BWB model. [60]
Parameter Value
Wing span 67.5 m
Range 5000 Nm
Cruise Mach number 0.8
Maximum takeoff weight 151,000 kg
Table 3.4: SAX-40 characteristics
3.2.5 Cranfield University BW11 Eagle Ray
Cranfield University developed a BWB configuration providing a great amount

of details 5 . It presents a distributed propulsion model, a innovative idea that is
getting more important nowadays. In the next images 3.12 and 3.13, the main
overall geometry and the cabin layout are shown. Also the main features provided
are shown in table 3.5.
5
and even prepared a video showing all the process
(http://www.youtube.com/watch?v=UfD0CIAscOI)
62
Figure 3.12: Cranfield University BW11 Eagle Ray
Figure 3.13: Cranfield University BW11 Eagle Ray passenger distribution sketch.
Parameter Value
Wing span 51.1 m
Wing area 552.7 m2
Wing taper ratio 0.163
Aspect ratio 4.72
Cruise Mach number 0.8 - 0.85
Static Margin -3.21
Service ceiling 38,000 ft
Maximum takeoff weight 151,160 kg
Empty weight 97,687 kg
Payload weight 23,760 kg
Table 3.5: Cranfield University BW11 Eagle Ray characteristics
3.2.6 NASA-Boeing X48-B

This model is a 8.5% scale experimental UAV that have been used for flight testing
in NASA facilities, with the help of Boeing and Cranfield University. Nowadays
the new model X48-C is being developed and has already been successfully tested
in April 2013. Some data of this model aircraft is shown in table 3.6:
63
Figure 3.14: X48-B BWB from NASA
Parameter Value
Wing span 6.22 m
Wing area 9.34 m2
Aspect ratio 4.1
Gross Weight 227 kg
Endurance 40 min
Service ceiling 10,000 ft
Powerplant 3 JetCat P200 turbojet with 0.23 kN thrust each
Table 3.6: X48-B characteristics
3.2.7 Main features in common

In general, all these models share some design characteristics. The most important
ones are:
• A smaller aspect ratio compared to conventional aircrafts, around 4.
• A smaller wing taper ratio compared to conventional aircrafts, around 0.15.
• The engines position is generally near the trailing edge of the central body.
• Similar sweep and dihedral angles.
• In general, all of them share the same wing shape, sections division and
geometry.
64
3.3. ELSA BWB
3.3 ELSA BWB

In 2002, this preliminar model was designed by Carlsson in [5] and was intended
to analyze the aeroelastic behaviour of the BWB configuration in wind tunnel
tests. The BWB geometry sketch is presented in figure 4.1. The BWB wind
tunnel model was a half-model approach that presented a segmented structure
with a single beam made of carbon fiber composite, as shown in figure 3.15. It
was equiped with two control surfaces controlled by electrical servos in order to
enable studies involving control surface response. The computer controlled servos
were used to actuate the control surfaces both statically and dynamically.
Figure 3.15: The ELSA BWB aeroelastic wind tunnel model. [5]
Afterwards, and in an amateur way, a remote controlled BWB UAV was built
and tested in flight and named as ELSA BWB. However, data from these flight
tests is not available and only some wind tunnel experimental data was obtained.
65
66
Chapter 4
Definition of the baseline BWB

geometry
In the previous chapter it has been stated the potential benefits and feasibility of
the Blended Wing Body concept. In the following sections, the description of the
BWB model simulated in this work is explained.
4.1 Definition in CPACSCreator

The ELSA BWB geometry was tried to be modelled inspired by some measure-
ments from Martin Carlsson’s papers and files, especially from the reference [5].
In figure 4.1 a sketch from [5] for wind tunnel testing is shown.
Figure 4.1: Martin Carlsson’s ELSA BWB sketch for wind tunnel testing in ref-
erence [5]
67
CHAPTER 4. DEFINITION OF THE BASELINE BWB GEOMETRY
These measurements differ slightly from the ones obtained from the MATLAB
files of the ELSA BWB configuration, but the ones chosen were the ones found in
[5]. However, the available information of the ELSA BWB was by far not enough
to define the geometry in a precise way, so many design decisions were taken, such
as the placement of the control surfaces or the engines. Other aspects, such as the
wing airfoils, were variated in order to improve the aerodynamic performance. In
general, this BWB design would correspond to a 1:30 scale of a real size prototype.
The explanation of the geometry definition process in CC was explained before
in subsection 2.3.1.
4.1.1 Wing
As stated before, some changes were done in order to improve the future aerody-
namic and general performance of the BWB. For example, winglets were added
in order to reduce the induced drag and to allow the placement of rudders in the
wingtips (as the BWB is a tailless aircraft) and the former airfoils (NACA 0017
for the inner fuselage and NACA 0012 for the wing) were substituted for the fol-
lowing ones: N2Asta072 (from the old VSP BWB file [61]) for the inner fuselage
and the Wortmann fx 60-126 (which data was obtained from [62]) for the wing.
The tool employed to study the airfoils aerodynamic performance is called XFLR5
[45], an open source software based on a improved version of the XFOIL[63] code
developed by Mark Drela at MIT. The test conditions for the simulations were a
Mach number of 0.30 and Reynold Numbers: 50000, 75000 and 100000.
The main reason for replacing the airfoils is that the NACA ones are quite
simple symmetrical (no lift with zero angle of attack) and do not offer good aero-
dynamic properties. The new ones (N2Asta072 and Wortmann fx 60-126) offer a
non-symmetrical layout, with camber and slightly lower thickness, with the con-
sequent lower weight, and, in general, better aerodynamical performance. Further
research about the optimum airfoils to be used can be done, but it is out of the
scope of this thesis. Now the mentioned airfoils are going to be described:
• NACA 0017
According to the NACA airfoil specification [64], the NACA 0017 airfoil has no
camber and the maximum thickness is the 17% of the chord value (here normalized
68
4.1. DEFINITION IN CPACSCREATOR
to 1), located at 30,1% of the chord from the leading edge. Figure 4.2 shows the
representation of the point cloud 1 . The point data was obtained from the NACA
4 digit airfoil generator available in reference[64]. In figures 4.3 and 4.4 are shown
the polar and other graphics for this airfoil:
Figure 4.2: NACA 0017 airfoil.
Figure 4.3: Polar of the NACA 0017 airfoil
1
The representation of the airfoils is obtained by using the graphical program Graph 4.4
69
Figure 4.4: NACA 0017 graphics showing from left to right: CL vs α, CL /CD vs
α and CL /CD vs CL
• NACA 0012
NACA 0012 airfoil is very similar to the NACA 0017 one, with the main difference
that the maximum thickness is the 12% of the chord value, which makes it more
suitable for the use in wings than the previous one. In figure 4.5 a sketch of this
airfoil can be observed. Some additional data about this airfoil is also presented
in table 4.1, 4.6 and 4.7:
Figure 4.5: NACA 0012 airfoil
70
Property Value
Maximum thickness 12.0%
Position of maximum thickness 30.10%
Maximum camber 0%
Trailing edge angle 58.6º
Lower flatness 12º
Leading edge radius 1.7%
Maximum CL 0.962
Maximum L/D 36.958
Zero lift angle 0º
Table 4.1: NACA 0012 data
Figure 4.6: Polar of the NACA 0012 airfoil
Figure 4.7: NACA 0012 graphics showing from left to right: CL vs α, CL /CD vs
α and CL /CD vs CL
71
• N2Asta072
This airfoil is not a conventional one and it was obtained from the previous BWB
configuration analyzed from VSP database [61]. The point data was directly
extracted from the XML file used before and transformed into legible data for
representation, as shown in figure 4.8. This airfoil has a lower maximum thickness,
14.28% of the chord, than the NACA 0017 located at 42.60% and a camber of
3.79% located at 7.30%. In figures 4.9 and 4.10 are shown the polar and other
graphics for this airfoil. To compare it in an easier way, both airfoils NACA 0017
and N2Asta072 are represented overlapped in figure 4.11.
Figure 4.8: N2Asta072 airfoil
Figure 4.9: Polar of the N2Asta072 airfoil
72
Figure 4.10: N2Asta072 graphics showing from left to right: CL vs α, CL /CD vs

α and CL /CD vs CL
Figure 4.11: Comparison between N2Asta072 (blue) and NACA 0017 (red)
• Wortmann fx 60-126
This airfoil is mainly used in sailplanes due to its high L/D ratio. The point data
was downloaded from reference [62]. In figure 4.12 the sketch of this airfoil is
shown. Some additional information about this airfoil is shown in table 4.2 and
in figures 4.13 and 4.14. To compare it in an easier way, both airfoils NACA 0012
and Wortmann fx 60-126 are represented overlapped in figure 4.15.
73
Figure 4.12: Wortmann fx 60-126 airfoil
Property Value
Maximum thickness 12.6%
Position of maximum thickness 27.90%
Maximum camber 3.6%
Position of maximum camber 56.50%
Trailing edge angle 2.6º
Lower flatness 52.8º
Leading edge radius 1.7%
Maximum CL 1.491
Maximum L/D 145.519
Zero lift angle -4.5º
Table 4.2: Wortmann fx 60-126 data
Figure 4.13: Polar of the Wortmann fx 60-126 airfoil
74
Figure 4.14: Wortmann fx 60-126 graphics showing from left to right: CL vs α,

CL /CD vs α and CL /CD vs CL
Figure 4.15: Comparison between Wortmann fx 60-126 (blue) and NACA 0017
(red)
After replacing the airfoils from the baseline model, the airfoils are placed and
scaled in a set of 7 airfoils per semi-wing using CC in order to reproduce the BWB
geometry defined. The results of this process is presented in figures 4.16 and 4.17.
75
Figure 4.16: Airfoils placement defining the BWB wing in CPACSCreator.
Figure 4.17: Isometric view of the BWB geometry defined as a wing.
4.1.2 Fuselage
Once defined the BWB wing geometry using the mentioned airfoils, the next step
is to define a proper fuselage utilizing fuselage sections normal to the X direction.
This process is mainly performed in order to obtain a smoother outer shape that
gets rid of the central ridge that appears in figure 4.17 because of the abrupt
surface variation in the plane of simmetry. This fuselage definition also provides
a more accurate weight definition later and to give a first preview of the future
manufacturing process using frames. For obtaining the fuselage sections points
coordinates, it is required to slice the geometry defined in the previous step as a
76
set of airfoils conforming a wing.
Unfortunately, CC or SUMO can not perform this action yet. Therefore, the
CAD program CATIA V5R20 was employed following the next steps:
1. The inner part of the BWB geometry defined in the previous section as
a wing is exported as a SUMO SMX file using the CC wrapper for that
purpose.
2. This SMX file is loaded in SUMO and exported as an IGES file, readable
by most of CAD programs.
3. This IGES file is loaded in CATIA and transformed into a closed surface
joining its several separated surfaces and smoothed using the Surfacic Cur-
vature Analysis, Fill Surface and the Shape Morphing functions from the
Generative Shape Design module (see figure 4.29). This process is performed
iteratively until a satisfactory solution is achieved.
4. Using the Sectioning function in the Mechanical Design module, the geom-
etry is sliced in 86 different sections normal to the X direction. A special
attention was paid to the nose part and to sudden curvature changes. (See
figure 4.18).
5. Later this curves are discretized using the Discretize curves function in the
Digitized Shape Editor module into the same number of points and exported
as ASCII files. As stated before, it is very important that all the profiles
have the same amount of points defined in the same order due to surface
continuity reasons required in CC.
6. These files are post-processed and ordered using Microsoft Office Excel and
finally introduced back in MATLAB as vectors that CC can handle as fuse-
lage sections. (See figure 4.19).
77
Figure 4.18: Fuselage sections of the BWB geometry visualized in CATIA.
Figure 4.19: Fuselage sections of the BWB geometry defined in CC.
All in all, the final visualization of the new geometry is presented in figure
4.20. The main apparent difference between figure 4.17 and 4.20 is that the wing
is no longer the whole outside surface; instead, the inner part is defined as the
fuselage (blue) and the outer part is mantained as the proper wing (green).
78
Figure 4.20: BWB geometry visualized in CC with the separated fuselage and
wing parts.
4.1.3 Control surfaces

As commented before, control surface sizing and allocation are cricital issues for
the BWB concept, so a more detailed analysis on the different control surface
configurations employed will be explained in chapter 6. As stated in subsection
2.3.1, the control surfaces in CC are defined within component segments of the
wing and have to be given in the eta/ksi coordinate system defined before.
The definition of the Trailing Edge Devices (TED) in the CC code was imple-
mented by Reichert in [29]. A TED is created with the same trailing edge shape as
its parent (a wing or another TED), referenced under parentuID. Every trailing
edge device is defined by closer describing its outer shape. The outer shape is
constructed with one airfoil at the inner border and another airfoil at the outer
border, connected assuming a straight leading edge geometry.
After introducing the values of the eta and ksi coordinates and the relative
height for the TED (see figure 4.21), the span, chord and location are defined.
The trailing edge shape of the airfoil will be the same as the parent airfoil till two
points limited by the parameters relChordUpperSkin and relChordLowerSkin as
shown in figure 4.21.
Control surfaces have the ability to rotate and translate relatively to their
79
Figure 4.21: Trailing edge control surface definition in CPACSCreator. [19]
parent to simulate a real deflection. The data required to describe the movement
of a control surface is stored in the subelement path within the CPACS file. Ev-
ery movement is described relative to a hinge line, defined by connecting with a
straight line one point in the inner and one point in the outer border of the trail-
ing edge device. These points are determined with their relative chord and height
coordinates relative to the parent airfoil. The lowering of the control surface is
described within the hinge moving coordinate system, by using:
X axis Equivalent to the global X axis.
Y axis The hinge line.
Z axis Perpendicular to the other two axis.
The deflection process is divided into several steps in CPACS. Every step contains
a translation along every hinge coordinate system axis as well as a rotation around
the translated hinge line, as shown in figure 4.22.
After this introduction in TED in CPACS, eight control surfaces are defined as
an example in the BWB geometry after defining a component segment that span
them. Figure 4.23 presents the wing sections and the first and last airfoil of each
TED and table 4.3 shows the values selected for the baseline’s control surfaces.
In this example an inner elevator, an outer elevator, an elevon (that can operate
as elevator or as aileron when needed) and a winglet rudder were defined for this
80
Figure 4.22: Trailing edge control surface deflection in CPACSCreator. [19]
semiwing, making 8 different control surfaces for the whole BWB. Allocating the
rudder at the winglet reduces the complexity and height of the aircraft, due to
the lack of necessity for a empennage.
Control Surface c1 (m) c2 (m) Span (m) Surface (m2 )

Inner Elevator 0.149 0.162 0.104 0.0162
Outer Elevator 0.108 0.081 0.091 0.0086
Elevon 0.081 0.054 0.427 0.0461
Rudder 0.080 0.080 0.150 0.0120
Table 4.3: Control surfaces geometry for the BWB baseline model.
Some authors like Voskuijl [30], suggest that the control surfaces could be able
to perform multiple functions, especially if a fly-by-wire system is implemented.
The advantage of this system is that the control allocation schedule can be depen-
dent on the flight condition (even for critical conditions, such as engine failure, it
can be reconfigured).
81
Figure 4.23: Illustration in MATLAB of the first and last airfoil of each of the 4
TEDs defined in the BWB right semiwing.
The saturation limits of all the control surfaces are defined as +30º/-30º and
no rate limits were introduced (considered as infinite). Once the control surfaces
are defined with all their parameters, an approxiamte visualization is available
in CC. In figure 4.24, the wing was setted as translucent in order to appraciate
better the control surfaces shape.
4.1.4 Engines
As a first approximation for this project, two engines from the example file D150
from CPACSCreator (4CM036) were imported and geometrically transformed to
fix them to the BWB geometry, see figure 4.25. A slight pitch angle (10º) was
given to both engines to simulate a boundary layer ingestion system, and to fit
in a better way with the BWB tail. The representation of the BWB geometry
with the two engines is presented in figure 4.26. However, the detailed analysis
of the propulsion subsystem is out of the scope of this thesis, so this model of
engine was considered to fulfill the requirements of this study. In this analysis,
the engines are considered as “almost ideal” and they are merely a thrust vector
which instantaneuosly reacts to throttle changes, depending on the flight condition
(Mach number and altitude, see figure 4.27 where the usual range of altitudes and
speeds were taken as conditions). Some technical information and the location
of the engines can be found in table 4.4. A further description of how CPACS
82
4.2. GEOMETRY REFINEMENT USING CATIA AND SUMO
Figure 4.24: Top view of the BWB geometry visualized in CC with the control
surfaces defined.
handles the engines geometry is found in [31].
Parameter Value
Xengine 1.4 m
Yengine ±0.291 m
Zengine 0.07 m
Rotation Y-axis 10º
Maximum diameter 0.22 m
Length 0.461 m
Thrust 45 N each
Table 4.4: Location and technical information of the two 4CM036 engines of the
BWB baseline.
4.2 Geometry refinement using CATIA and SUMO

As commented before, an iterative process was performed to smooth the outer
surface of the BWB baseline using mainly the CATIA surface smoothing tools
and the implicit function implemented in SUMO code that slightly smoothes the
outer shape of an imported IGES file. One of the first iterations obtained is
83
Figure 4.25: MATLAB visualization of the 4CM036 engine selected for the BWB
model.
Figure 4.26: BWB geometry with the incorporated engines
presented in figure 4.28, as the SUMO visualization. It can be better appreciated

that the fuselage geometry still presents some irregularities and bumps. Therefore,
a geometry refinement is required to improve the geometry quality of the BWB
model.
First of all, for solving the bumps problem, an iterative process of employing
the Fill Surface function from CATIA around the bumps and exporting the ge-
ometry as an IGES file from SUMO to CATIA and vice versa was performed till
the outer shape of the BWB is smoothly refined.
The other main objective of this process was to get rid of the ringe that ap-
peared in the plane of symmetry. The function Shape Morphing from the Genera-
84
4.2. GEOMETRY REFINEMENT USING CATIA AND SUMO
Figure 4.27: Engine performance with respect to the altitude and Mach number
for the BWB baseline.
tive Shape Design module permits the deformation of this area in order to smooth
the outer surface and to make it more uniform. Figure 4.29 shows an example of
this function used in the BWB geometry.
Figure 4.29: Illustration of the CATIA Shape Morphing function used to get rid
of the ringe in the central part of the BWB baseline
The final versions in CATIA and SUMO, considered as acceptable, are pre-
85
Figure 4.28: BWB geometry visualized in SUMO showing some irregularities in

the fuselage geometry.
sented in figures 4.30 and 4.31. Once that the geometry has been refined, it was
imported back from SUMO to CC via wrapper in order to redefine the outer shape
of the geometry.
Figure 4.30: BWB geometry rendered in CATIA
4.3 Final model

The final geometry has the following outer shape (without considering the control
surfaces yet), presented in figure 4.32 as the main views of the model. More
detailed skecthes can be found in the Appendix. Figure 4.33 shows the cross
86
4.3. FINAL MODEL
Figure 4.31: Refined mesh of the BWB geometry using SUMO automatic mesh
generator.
section area distribution for the BWB model obtained with SUMO. Table 4.5
gathers some geometric characteristics and table of the BWB model obtained,
table 4.6 presents several geometric parameters of the defined wing2 , and tables
4.3 and 4.4 exhibit the qualities of the control surfaces and engines, respectively,
as commented before. The weight and inertia information of the BWB geometry
is obtained from the adapted Weight and Balance module implemented in the
same interface as CC. Figure 4.34 and table 4.7 show the output achieved.
2
The meaning of these symbols is defined in the Latin and Greek symbols section at the
beginning of this project.
87
Figure 4.32: Sketches views of the BWB geometry.
Figure 4.33: Cross section area distribution in the X direction for the BWB base-
line, obtained with SUMO.
88
4.3. FINAL MODEL
Parameter Value
Wing span, b 3.095 m
Wing area, Swing 1.5164 m2
Mean Aerodynamic Chord, MAC 0.961 m
Mean Geometric Chord, MGC 0.655 m
Total length, LBW B 1.734 m
Fuselage length, Lf 1.609 m
Total heigth, hBW B 0.326 m
Fuselage height, hf 0.253 m
Aspect ratio, AR 3.53
Taper ratio, λ 0.083
Dihedral angle, Γ 5º
MTOW 22.5 kg
T
Thrust to Weight Ratio, W 0.41
Table 4.5: Geometric characteristics of the BWB baseline model obtained.
Wing section c1 c2 Span ΛLE Γ Airfoil

Fuselage 1.609 m 0.993 m 0.581 m 63.51º 5º N2Asta072
Inner wing 0.993 m 0.719m 0.151 m 39.05º 5º Wortmann fx 60-126
Middle wing 0.719 m 0.537 m 0.140 m 39.05º 5º Wortmann fx 60-126
Outer wing 0.537 m 0.179 m 0.639 m 39.05º 5º Wortmann fx 60-126
Winglet 0.179 m 0.179 m 0.150 m 39.05º 80º Wortmann fx 60-126
Table 4.6: Geometric characteristics of the baseline BWB wing.
Figure 4.34: Final BWB geometry represented in the Weight and Balance tool.
89
Parameter Value
Xcg 0.965121 m
Ycg 0m
Zcg -0.03783 m
Ixx 1.97288 kg·m2
Iyy 2.56984 kg·m2
Izz 4.43519 kg·m2
Ixz 0.11429 kg·m2
Ixy 0 kg·m2
Iyz 0 kg·m2
Table 4.7: Center of gravity coordinates and inertia moments of the BWB model.
90
Chapter 5
Generation of aerodynamic data
The software used to obtain the aerodynamic data required for the BWB Stability
and Control analysis is the VLM code TORNADO. In this chapter, the theoreti-
cal fundamentals on which TORNADO is based will be introduced. However, the
conventional TORNADO code implemented in CEASIOM only handles a limited
number of control surfaces. To allow a more complex analysis of the Control
and Stability characteristics of the BWB model, a modified new version of the
TORNADO code that is able to handle more control surfaces was employed. A
wrapper was developed to allow the connection between CC an the new TOR-
NADO code. Before using the new code with the BWB, a study with an example
configuration is performed in order to validate its correctness by comparing to
previous accepted data. The example chosen was the Boeing 747-100 aircraft be-
cause the experimental data and other numerical predictions were gathered in [6]
by Da Ronch et al.
As an additional attempt to compare the obtained results with TORNADO,
the BWB geometry was also modelled in the XFLR5 software mentioned before.
Apart from the 2D airfoil design and analysis used in subsection 4.1.1, this pro-
gram also offers a 3D environment for designing and analysis wings and airplanes
in various conditions, and obtaining its aerodynamic and stability and control
characteristics. The BWB model was modelled as a whole wing, without con-
sidering engines or fuselage. There are three different models available, briefly
introduced in the following list1 :
1. Lifting Line Theory (LLT): Non-linear method for wings with high aspect
1
A more detailed explanation of these methods can be found in [36] and [45].
91
CHAPTER 5. GENERATION OF AERODYNAMIC DATA
ratio and low sweep angle.
2. Vortex Lattice Method (VLM): Linear method considering inviscid induced

drag able to analyze to any usual wing geometry. This method is based in
the same theory as TORNADO.
3. 3D Panel Method: Linear more sophisticated method, taking into account

the thickness of the wing. This method is able to provide the pressure
distributions over the top and bottom surfaces of a wing and model fuselages.
In order to be able to compare the results obtained with Carlsson’s ones in [5],
the same air speed was selected: u = 40 m/s.
5.1 TORNADO fundamentals

As stated before, TORNADO is a software based on the Vortex-Lattice Method
(VLM) theory corrected by the strip theory and integrated in CEASIOM for
conceptual aircraft design. The code is based on a modified horse-shoe vortex sin-
gularity method for computing steady and low reduced frequency time-harmonic
unsteady flows over wings. It can be used on most types of aircrafts, with cranked
and twisted wings with multiple control surfaces at the trailing edge. Tornado is
implemented in MATLAB and published under GNU GPL[50]. Several publica-
tions confirm the satisfactory accuracy of TORNADO code, such as [30, 6]. Even
though calculations carried on Tornado have a strong background on aerodynam-
ics, it is out of the scope of this work to fully explain them, they are included in
Melin’s work [32].
The geometry panelling is automatically produced within AMB from the XML
geometry description. The lifting surfaces are generated as unions of thin quadri-
lateral surface panels (see figure 5.1), each of them has a horseshoe vortex attached
to it. The strength of the vortices is unknown at first and is calculated by making
use of a boundary condition (the flow must be tangent at any point of the surface).
Airfoil camber effects are modelled by the rotation of the surface normal. Leading
edge control surfaces are modelled likewise, and trailing edge devices (TED) are
modelled by actual mesh deformation, instead of using the transpiration approx-
imation as Edge does.
92
5.1. TORNADO FUNDAMENTALS
Figure 5.1: Example of the geometry panelling in TORNADO for the TCR cruise
example.
The fuselage can be considered using several techniques. The simplest ap-
proach is to model it as two flat elements with shapes corresponding to its planar
projections. This model provides remarkably different values of the lifting and
pitching moment coefficients slopes compared to experimental results. Other al-
ternative consists in employing the slender body theory developed by Munk [33],
which assumes that the diameter of the body is far less than its length (as gener-
ally occurs in conventional aircrafts). However, this theory is restricted to body
of revolution shapes, which is a condition rarely met in a real aircrafts. Therefore,
the fuselage surface can be modelled by a number of flat panels approximating its
shape and the value of the sink/source distribution in each panel is the solution
of a system of linear equations. An example of a fuselage modelled this way is
presented in figure 5.4.
The basic flow solver is wrapped by user interfaces to compute forces and
moments and generate tables of aerodynamic coefficients, and stability deriva-
tives, for export to flight simulators and flight control system design software.
To calculate the first order derivatives, TORNADO performs a central difference
calculation calculating small disturbances around the pre-selected state. It can
be used for low-speed aerodynamic cases and general effects of compressibility at
higher Mach numbers are considered by the Prandtl-Glauert similarity rule, which
gives reasonable results up to Mach numbers around 0.6. Zero-lift drag estimates
are based in Eckert’s flat plate analogy. Viscosity can be taken into account by
an approximate empirical extension, combining the results with the 2D viscous
93
airfoil code XFOIL.
For this analysis the aircraft is considered to have six degrees of freedom in
the global axis system, as in classical mechanics:
1. Pitch moment (around the Y body axis): M
2. Yaw moment (around the Z body axis): N
3. Roll moment (around the X body axis): L
4. Velocity in the X body axis: U
5. Velocity in the Y body axis: V
6. Velocity in the Z body axis: W
Among these parameters, the pitch and the roll moment are the critical aspects
for the stability of a trimmed aircraft.
The aerodynamic coefficients represent the forces and moments in an adimen-

sional way in order to facilitate their analysis. In the following list, q∞ represents
the dynamic pressure, S is the planform area and c is the lenght of the chord:
Fz
• Lift coefficient: CL = q∞ ·S
Fx
• Drag coefficient: CD = q∞ ·S

Fy
• Side force coefficient: CY = q∞ ·S
= CYβ β+ VbT CYr r + CYp p +CYδRudder δRudder
Mx
• Pitch moment coefficient: Cm = q∞ ·S·c

Mz b
• Yaw moment coefficient: Cn = q∞ ·S·c
= Cnβ β + VT
Cn r r + Cn p p +
CnδRudder δRudder + CnδAileron δAileron

My
• Roll moment coefficient: Cl = q∞ ·S·c
= Clβ β+ VbT Clr r + Clp p +ClδRudder δRudder +
ClδAileron δAileron
The aerodynamic derivatives of this coefficients can be calculated with respect to:
• Angle of attack, α.
94
5.2. B-747 EXAMPLE
• Angle of sideslip, β.
• Roll-pitch-yaw rotations, p, q, r.
• Control surface deflections, δsurf ace .
TORNADO was selected for the approach followed in this study because it was
considered as a reasonable approximation for linear and quasi-linear aerodynam-
ics, the main objective of this thesis. Other reasons were the low-speed condition
of the case and the unavailable enormous computation cost required by the higher
fidelity level options.
5.2 B-747 example

In order to validate the correctness of the new TORNADO code and guarantee
its use for the BWB model, an example with known results is tested. The model
selected was the Boeing 747-100 and Da Ronch et al. present their results in the
paper [6]. The reason for choosing this aircraft is that it employs more control
surfaces beyond the conventional elevator, ailerons and rudder, so it is a good
choice to test the new TORNADO code and observe how it handles with more
control surfaces than the previous version. In addition, the aircraft geometry and
fuselage cross sections information was available from the SMX file provided by
the authors of [6], and was imported to CC in order to perform the aerodynamic
analysis with TORNADO.
5.2.1 Description of the B-747 model

The Boeing 747 is a very large four-engined turbofan transport aircraft designed
to transport more than 350 passengers on medium–long haul flights. The aircraft
was designed in the 1960s to meet growth demands for mass long haul transport
and has suffered several derivatives with varying operational requirements. As
stated before, the 747-100 version uses a large number of control surfaces:
• Triple-sloted trailing flaps and Krueger leading edge flaps allow the necessary
low speed characteristics.
• Four elevator segments and a movable stabilizer provide longitudinal control.
95
• Five spoiler panels, an inboard aileron and an outboard aileron provide

lateral control.
• Two rudder segments provide directional control.
Figure 5.2 illustrates the SUMO geometry and tetrahedral volume mesh generated
for the Boeing 747-100 model, including fuselage, wing, empennage and fairing.
(a) (b)
Figure 5.2: SUMO geometry (a) and volume mesh (b) for the Boeing 747-100
model.
The geometry of the aircraft is approximated for different fidelity levels ap-
proaches:
1. The simplest and lowest fidelity approach consists of the lifting surfaces
only (see figure 5.3), suitable for TORNADO VLM-based calculations and
defined in [6] to closely match the aircraft’s geometry and control surfaces.
Figure 5.5 presents the comparison between the TORNADO geometry and
the Boeing 747-100 aircraft, showing a quite accurate reproduction of the
main features of the model.
Figure 5.3: TORNADO geometry for the lifting surfaces of the B747-100 model.[6]
96
5.2. B-747 EXAMPLE
2. The second fidelity level corresponds to the addition of the fuselage influ-
ence, modelled as a slender body in TORNADO calculations. In the case
presented in figure 5.4, the lifting surfaces were not modified to accomodate
the fuselage.
Figure 5.4: TORNADO geometry with fuselage for the lifting surfaces of the
B747-100 model.[6]
3. Finally, the highest fidelity is achieved with a watertight geometry used in

CFD calculations. An example of this case was shown in figure 5.2, where
the engine nacelles were removed and the fairings were represented in a
simple way for simplifying the surface geometry. These simplifications are
reasonable considering that the interest is on the flow development on the
upper lifting surfaces.
All the cases considered in this approach in TORNADO consist of inner and
outer ailerons, an all-movable stabilizer, a two-segment elevator and a two-segment
rudder, sized from the aircraft model. These control surfaces are highlighted in
other colour in figure 5.6. The modelled geometry has sharp trailing edges which
are adequate for inviscid flow models.
5.2.2 Validation of aerodynamic sources

The wind tunnel experimental data is provided by the manufacturer [34]. In this
analysis, TORNADO was used for obtaining the aerodynamic data with inviscid
conditions, but in [6] CFD-based results from Edge and DATCOM are also pre-
sented, so a further comparison can be achieved. In order to distinguish between
the data from TORNADO presented in [6] and the data from this work, the pre-
dictions from this research are labelled as “New TORNADO”. In the following
pages, the results obtained in this analysis are compared with the ones shown in
the paper.
97
Figure 5.5: Views of the comparison between the TORNADO geometry and the
B747 model.[6]
5.2.2.1 Low speed aerodynamics (M=0.15)
First of all, the prediction of the lift coefficient, CL , is compared with the one
presented by Da Ronch et al. The graphs labelled as "TORNADO Baseline Paper"
and “New TORNADO” are based on the inviscid formulation of the VLM, which
predicts a linear increase in the lifting coefficient if the angle of attack is increased.
The “TORNADO Viscous Paper” data has the viscous approximation correction
included. Between the angles of attack of 10º and 15º, a noticeable slope change
is observed, indicating the start of the stall. In figure 5.7 it can be observed that
the viscous prediction from TORNADO underpredicts the stall for lower angles
of attack than in the reality. The resemblance between the two inviscid results
98
5.2. B-747 EXAMPLE
(a) (b)
Figure 5.6: Surface mesh of the Boeing 747-100 model. (a): Control surfaces on
the main wing. (b): Empennage and its control surfaces.
using TORNADO is remarkable and the new one is even slightly closer to the
experimental data.
Figure 5.7: Lift coefficient vs. angle of attack of the B-747 model compared with
previous data from [6]
The pitching moment coefficient, Cm , prediction is analyzed now in figure 5.8,

presenting a non-linear behaviour above the static stall condition. The predictions
obtained from the paper include two configurations for the TORNADO calcula-
tions: one without considering the fuselage, just the Wing-Vertical-Horizontal
tail “WVH” as shown in figure 5.3, and the other one that models the fuselage
with sink/source singularities presented in figure 5.4, labelled as Fuselage-Wing-
99
Vertical-Horizontal tail “FWVH”. The only difference between the two options is
a positive offset of about 0.2 in the Cm direction for the “FWVH” with respect
to the “WVH” option. Considering the fuselage, improves the overall agreement
to the experimental data but provides worse results for low angles of attack. The
new TORNADO data did not consider the fuselage and shows an intermediate
behaviour between the two predictions mentioned before.
Figure 5.8: Pitching moment coefficient vs. angle of attack of the B-747 model
compared with previous data from [6]
In order to analyze the behaviour and influence of the control surfaces deflec-
tion, a study about the deflection of the outboard elevator was performed using
TORNADO without considering the viscosity corrections. A clean configuration
for the lifting surfaces was selected. Figures 5.9 and 5.10 show a close agree-
ment between both predictions and the experimental data. The force coefficient
is slightly overpredicted by the old TORNADO and the moment coefficient is
slightly underpredicted by the new TORNADO code.
5.2.2.2 High speed aerodynamics
For these cases, Da Ronch et at. also used Edge and DATCOM in order to show
the differences between the variable fidelity methods. In [6], the study was per-
formed for three Mach numbers: 0.75, 0.80 and 0.85. To avoid becoming tedious,
100
5.2. B-747 EXAMPLE
Figure 5.9: Lift coefficient vs. outboard elevator deflection of the B-747 model
only the case for M = 0.75 is considered in this analysis.
First of all, the lifting coefficient is analyzed again and compared in figure
5.11. As it could be expected, the Euler-based outcomes from Edge show the
closest agreement to the experimental data of the B747-100. DATCOM results
show a similar curve slope, with slightly lower values than the experimental data.
TORNADO results without compressibility correction, labelled as “TORNADO
Paper”, do not vary when the Mach number is increased, but the error with
respect to the experimental data increases due to the bigger influence of the
compressibility effects at higher speeds. TORNADO outcomes considering the
Prandtl-Glauert similarity rule, labelled as “TORNADO Comp Paper” overpre-
dict the curve slope, diverging at higher Mach numbers. This poor agreement
with the experimental data might be due to the simple compressibility correction
which begins to fail at the higher end of the transonic regime.
The next feature to analyze is the pitching moment coefficient, which is a fair
indicator of the aircraft static stability. At this point, the experimental data seems
to have a non-linear behaviour because of the migration of the static stability to-
ward neutral stability. This might be caused by non-linear effects due to viscous
phenomena. The numerical values show in figure 5.12 differ from the experimental
101
Figure 5.10: Pitching moment coefficient vs. outboard elevator deflection of the
B-747 model compared with previous data from [6]
data a constant offset, which suggests discrepancies in the value of the pitching
moment for zero angle of attack, Cm0 . This might be caused by the high depen-
dency of this term on the airfoil section and fuselage geometry used. Once again,
the Euler-based results from Edge present the best correlation with the experi-
mental data. TORNADO including the compressibility effects predicts a higher
slope, amplifying the compressibility effect on the static stability. DATCOM and
the two TORNADO cases without considering the compressibility effects predict
a constant slope similar to the linear part of the experimental data, but with a
higher offset than the Edge predictions. The new TORNADO code even provides
slightly better results than the older version.
Finally, the drag polar is considered in figure 5.13. In this case, DATCOM
outcomes presents the closest correlation with the experimental data, because
it was developed using semi-empirical methods and conventional configurations,
such as the Boeing 747. The Euler results from Edge underpredicts the drag
compared to the experimental data due to the absence of any estimation of the
friction drag. TORNADO does not achieve a satisfactory agreement this time,
and the case considering compressibility effects is even slightly worse than the
ones that do not consider them. However, the similarity between the results of
the two TORNADO codes without compressibility effects is again noticeable.
102
5.2. B-747 EXAMPLE
Figure 5.11: Lift coefficient vs. angle of attack of the B-747 model for M = 0.75
The results included in [6] indicated that a better correlation is obtained for
lower Mach numbers, for all the tools. For the high speed aerodynamics cases, the
Euler-based results of Edge are the ones who present the best correlation with the
experimental data, confirming the requirement of using higher fidelity CFD tools
in the transonic regime, where the VLM methods, such as TORNADO, start to
fail.
5.2.2.3 Stability analysis
The basics of stability and the concepts of short period, phugoid and dutch roll
modes will be explained in detail in section 6.1. In the following figures the
analysis results for these modes from the Stability and Control module, SDSA,
obtained in this study will be compared with the ones presented in [6]. In figure
5.14, the short period mode analysis is represented according to ICAO stability
criteria, showing almost identical results in both cases. The phugoid mode is
analyzed in 5.15 according also to ICAO stability criteria. This time the results
differ slightly from each other in slope, especially for the 12 km height case, but
overall the agreement is satisfactory. Finally, figure 5.16 presents the results for
the dutch roll mode in accordance with CS stability criteria. Besides a slightly
higher slope in the new experience, the results are very similar.
103
Figure 5.12: Pitching moment coefficient vs. angle of attack of the B-747 model
for M = 0.75 compared with previous data from [6]
5.2.2.4 Linear mode analysis
The final aspect to validate is the linear mode analysis. The trimmed values for
the angle of attack and the elevator deflection for different Mach numbers are
presented in figures 5.17 and 5.18. In general, as the Mach number increases, the
aircraft requires a lower angle of attack and the required elevator deflection to
achieve the trimmed state increases, as could be expected.
For the trimmed values of the angle of attack, Edge and the TORNADO con-
sidering compressibility effects present the best correlation with the experimental
data. TORNADO cases that do not consider compressibility effects have a similar
slope but are separated a higher offset from the experimental data. The outcomes
from DATCOM do not really match correctly with the experimental data, possibly
because of the compressibility effects.
For the trimmed values of the elevator deflection, Edge is the closest represen-
tation to the experimental data, followed by the two TORNADO codes without
compressibility effects, which have a higher offset. TORNADO code including
compressibility effects are separated a higher distance from the experimental data.
DATCOM results suggest that the influence of the elevators is significant on the
vehicular lift, leading to a smaller efficiency value or CL /Cm .
104
5.3. TORNADO RESULTS
Figure 5.13: Drag polars of the B-747 model for M = 0.75 compared with previous
data from [6]
5.2.3 Conclussions
In general, the new TORNADO code presents a similar or slightly better behaviour
than the one used in the previous work by Da Ronch et al. The few discrepancies
found could be due to the different parameter values used in the tabular data
to obtain the stability derivatives when using TORNADO. All in all, this study
validated that this code provides a satisfactory approximation and is a powerful
tool for low speed aerodynamics analysis. As it could be expected, when the Mach
number increases, the results start to diverge from the experimental data, but this
case will not be considered for the BWB configuration.
5.3 TORNADO results

The procedure for obtaining the aerodynamic data of the BWB model with the
new TORNADO code is identical as the one used in the B-747 example before.
Therefore, the results obtained are shown directly. The conditions for this analysis
were setted as an air density of ρ = 1.225 kg/m3 and an air speed of u = 40 m/s.
The panel model employed in this analysis is illustrated in figure 5.19. A total
of 368 panels in the mesh were employed: 10 chordwise and 8 spanwise for the
105
(a) (b)
Figure 5.14: Short period mode prediction analysis for the Boeing 747-100 model.
(a): Results presented by Da Ronch et at. in [6]. (b): Results obtained in this
experience using the new TORNADO code.
(a) (b)
Figure 5.15: Phugoid mode prediction analysis for the Boeing 747-100 model.
fuselage, 10 chordwise and 4 spanwise for the inner wing and 8 chordwise and 8
spanwise for the outer wing. This number was considered as a good approxima-
tion, because other publications, such as [35], confirm that the convergence of the
results is achieved using approximately 300 panels with a similar distribution as
the one considered in this project.
106
(a) (b)
Figure 5.16: Dutch roll mode prediction analysis for the Boeing 747-100 model.
Figure 5.19: 2D model for the BWB geometry in TORNADO, including the mesh
grid, the Mean Aerodynamic Chord (MAC) and the centre of gravity position.
First of all, the prediction of the lift coefficient, CL , with respect to the angle of
attack is presented in figure 5.20. The pitching moment coefficient, Cm , variation
with the angle of attack is prediction is presented in figure 5.21. The drag polar
for this model is exhibited in figure 5.22.
In order to analyze the stability characteristics later, a study about the deflec-
tion of the control surfaces of the BWB model was performed using TORNADO.
Due to the low-airspeed nature of the approach, the aerodynamic derivatives al-
most do not depend on the airspeed because there are no compressibility effects
for those airspeeds. As an illustrative example, the graphs for the lift coefficient
107
Figure 5.17: Trimmed angle of attack of the B-747 model for different Mach
numbers compared with previous data from [6]
and pitching moment coefficient variation with respect to the deflection of the
outer elevator are represented in figures 5.23 and 5.24, respectively.
All in all, the most important aerodynamic derivatives obtained for the baseline
BWB model for the mentioned conditions are shown in the following tables. The
selected condition is for angle of attack α = 2º and air speed of u = 40 m/s
These results will be used in the SDSA module to obtain the stability analysis, as
explained later in chapter 6.
Coefficient Value
CL 0.1989
CD 0.0037
Cm -0.0169
Table 5.1: Aerodynamic coefficients for the BWB baseline and α = 2º and u =
40 m/s obtained in TORNADO.
108
Figure 5.18: Trimmed elevator deflection of the B-747 model for different Mach
numbers compared with previous data from [6]
Coefficient Value
CLα 2.7312 rad−1
CDα 0.6017 rad−1
Cmα -0.1139 rad−1
Table 5.2: Angle of attack derivatives for the BWB baseline and u = 40 m/s
obtained in TORNADO.
Coefficient Value
C Yβ -0.2712 rad−1
Clβ -0.1089 rad−1
Cnβ 0.0379 rad−1
Table 5.3: Sideslip angle derivatives (side force, rolling moment and yawing mo-
ment coefficients) for the BWB baseline and u = 40 m/s obtained in TORNADO.
Coefficient Value
C Yp -0.2122 s/rad
Clp -0.2089 s/rad
Cn p -0.0349 s/rad
Table 5.4: Roll rate angle derivatives (side force, rolling moment and yawing mo-
ment coefficients) for the BWB baseline and u = 40 m/s obtained in TORNADO.
109
Figure 5.20: Lift coefficient vs. angle of attack of the BWB using TORNADO for
air speed, u=40 m/s
Figure 5.21: Pitching moment coefficient vs. angle of attack of the BWB using
TORNADO for air speed, u=40 m/s
Coefficient Value
Cmq −24.2s/rad
Table 5.5: Pitch rate angle derivative (pitching moment coefficients) for the BWB
baseline and u = 40 m/s obtained in TORNADO.
110
Figure 5.22: Drag polar of the BWB using TORNADO for air speed, u=40 m/s
Figure 5.23: Lift coefficient vs. deflection of the outer elevator of the BWB using
TORNADO for air speed, u=40 m/s
Coefficient Value
C Yr -0.0953 s/rad
Cl r -0.0781 s/rad
Cnr -0.1098 s/rad
Table 5.6: Yaw rate angle derivatives (rolling moment and yawing moment coef-
ficients) for the BWB baseline and u = 40 m/s obtained in TORNADO.
As it will be analyzed in chapter 6, the fact that Cmα and Clβ are negative
and Cnβ is positive imply that the BWB baseline is statically stable in all the
directions. Furthermore, the damping derivatives, Clp , Cmq and Cnr are negative,
indicating that the aircraft damps its rotations by itself in all directions.
111
Figure 5.24: Pitching moment coefficient vs. deflection of the outer elevator of
the BWB using TORNADO for air speed, u=40 m/s
A special emphasis should be made in the control surfaces deflection deriva-

tives, tables 5.7 to 5.10 show the influence of the deflection of the different control
surfaces of the BWB baseline in the aerodynamic coefficients.
Coefficient Value
CLδInnerElev 0.5124 rad−1
CDδInnerElev 0.0217 rad−1
CmδInnerElev −0.2622 rad−1
Table 5.7: Inner elevator deflection derivatives for the BWB baseline and u =
Coefficient Value
CLδOuterElev 0.3897 rad−1
CDδOuterElev 0.0038 rad−1
CmδOuterElev −0.1873 rad−1
Table 5.8: Outer elevator deflection derivatives for the BWB baseline and u =
112
5.4. XFLR5 RESULTS
Coefficient Value
CYδElevon 0.0194 rad−1
ClδElevon −0.3281 rad−1
CnδElevon −0.0061 rad−1
Table 5.9: Elevon deflection derivatives for the BWB baseline and u = 40 m/s
Coefficient Value
CYδRudder 0.0812 rad−1
ClδRudder 0.0384 rad−1
CnδRudder −0.0295 rad−1
Table 5.10: Rudder deflection derivatives for the BWB baseline and u = 40 m/s
Finally, the available experimental data from Carlsson’s wind tunnel experi-
ment extracted from [5] are compared and analyzed in table 5.11, showing a very
satisfactory agreement, with approximately 5% error with respect to the experi-
mental data, which could be caused for the changes implemented on the former
geometry. Even if the amount of available experimental data is not too large, it is
representative enough and the fact that they correlate with the predictions from
TORNADO is very important to validate the numerical results in this project.
Coefficient Prediction Experimental data from [5] Error (%)

−1
CFzα 4.4392 rad 4.23 rad−1 -4,95
−1 −1
CMyα -1.6813 rad -1.74 rad 3,37
C M xα 2.6478 rad−1 2.79 rad−1 5,10
−1 −1
CMxδ 0.0443 rad 0.046 rad 3,70
Mxδ 19.8733 Nm/rad 20.25 Nm/rad 1,86
Table 5.11: Comparison between the TORNADO predictions for the BWB base-
line and the available experimental data from [5] for u = 40 m/s.
5.4 XFLR5 results

As stated before, the BWB geometry was implemented in the XFLR5 software in
a similar way than in CC, as shown in figure 5.25. A similar mesh was used in
this analysis to the one used for TORNADO. Two different methods were used to
113
analyze the aerodynamic performance of the model, the VLM and the 3D panel
method, in order to compare their results and observe possible differences. Both
of them are designed to offer best results for low angles of attack.
Figure 5.25: 3D model for the BWB geometry in XFLR5, including the panel
discretization.
In the following figures, the results obtained for the BWB model with a air
speed of u = 40 m/s are presented. Both methods provide very similar results,
showing a close data agreement. A special comment must be done about the
aerodynamic efficiency analysis, found in figures 5.29 and 5.30. For negative values
of angle of attack, the results exhibit some irregularities and should be considered
carefully. Therefore, the average real aerodynamic efficiency should be between
the values of 20 and 35 (depending on the opearting point), still remarkably higher
than conventional aircrafts, such as Boeing 747, which presents a L/D ratio of
approximately 17.
As an example of the pressure coefficient distribution in the BWB model, for
some different angles of attack, figure 5.31 is presented. As it could be expected,
when increasing the angle of attack, the pressure difference in the leading edge of
the aircraft increases. One of the drawbacks of the VLM method is that it fails to
predict the stall condition, due to its linearized theory. However in figure 5.31, it
can be appreciated that the outer part of the wing starts having bigger pressure
difference, therefore this region is the most suitable to have the stall phenomena
in first place.
This software presents some graphical representation advantages with respect
to TORNADO, such as visualization of the local lift and local lift coefficient
spanwise. Figures 5.32 and 5.33 illustrate this characteristics for several angles
of attack for the BWB model. The spanwise lift forces present an almost-elliptic
114
5.5. COMPARISON OF THE TWO TOOLS
Figure 5.26: Lift coefficient vs. angle of attack of the BWB for both methods for
air speed, u=40 m/s
Figure 5.27: Pitch Moment coefficient vs. angle of attack of the BWB for both
methods for air speed, u=40 m/s
distribution, which is considered as the ideal distribution for minimum induced

drag possible. This fact indicates that the design process is oriented in a satis-
factory way. As commented before, other advantage of the BWB configuration is
that the outer wing presents a smaller loading, due to the lifting characteristics
of the inboard fuselage, reducing the bending moment.
5.5 Comparison of the two tools

In order to ensure the reliability of the obtained results, a representative sample
of the results obtained using both numerical tools are shown and compared in the
115
Figure 5.28: Drag polar of the BWB for both methods for air speed, u=40 m/s
Figure 5.29: Aerodynamic efficiency vs. angle of attack of the BWB for both
following figures. Figure 5.34 present the lift coefficient variation with the angle
of attack for both methods, showing a remarkable agreement with a slight under-
estimation of the lift coefficient in the TORNADO code compared to the XFLR5
software. In figure 5.35 the pitching moment coefficient is slightly overestimated
in the TORNADO case, presenting a more pronounced slope.
5.6 Parasite drag estimation

Finally, the drag polar is compared in figure 5.36, displaying a satisfactory agree-
ment between all the methods, but a divergent underestimation of the drag co-
efficient for high angles of attack in the case of the TORNADO code. However,
116
5.6. PARASITE DRAG ESTIMATION
Figure 5.30: Aerodynamic efficiency vs. lift coefficient of the BWB for both
it must be noticed that TORNADO (and in general the VLM method) fails to
predict the parasite drag, CD0 , whose values are extremely low (CD0 = 0.0011 for
TORNADO, CD0 = 0.0016 for the VLM method from XFLR5 and CD0 = 0.0017
for the 3D Panel method from XFLR5) compared to usual values of around 0.02.
This is due to the the fact that TORNADO does not consider properly the viscos-
ity, which is one of the main causes of the parasite drags. This fact was mentioned
in [6], but no further estimations were done.
Therefore, an alternative procedure for estimating the parasite drag, the so-
called flat plate method, was performed. The total drag coefficient can be approx-
imated using the usual formula, based on a parabolic model, as shown in equation
5.1:
CL2
CD = CD0 + CDi = CD0 + (5.1)
π · AR·e
,where CDi is the induced drag coefficient, AR is the aspect ratio and e is the
Oswald efficiency number.
The parasite drag of each component can be estimated with the following formula,
extracted from [37]:
Cfcomponent Swetcomponent F Fcomponent

CD0component = (5.2)
Sref erence
where Cfcomponent represents the skin friction coefficient of the component,
Swetcomponent refers to the wetted area of the component, F Fcomponent represents
117
Figure 5.31: Pressure coefficient, Cp, distribution for different angles of attack in
the BWB model.
the form factor of the component and Sref erence is the reference area (selected to
be the wing area).
Wing parasite drag estimation
First of all the wing was considered. However, the wing is not really a flat plate,
therefore the total wetted area for the BWB baseline (considered as a whole wing)
was calculated with CATIA and found to be 3.3672 m2 and also estimated by an
empirical method from [37]:
t

Swetwing ≈ 2Swing 1 + 0.2 = 2·1.5164 (1 + 0.2·0.13) = 3.1117m2 ≈ 3.3672 m2
c
Both values are quite close, but the numerical value obtained with CATIA was
used in this approach.
The form factor for the wing was estimated from an empirical graphic from [37],
presented in figure 5.37, and determined to have a value of F Fwing ≈ 1.217.
The skin friction coefficient depends on the Reynolds number and whether the
airflow is laminar or turbulent. Therefore, the Reynolds number is calculated for
the usual flight conditions (altitude 4000 m and M=0.1):
118
Figure 5.32: Local lift force spanwise for the BWB model for air speed, u=40 m/s
Figure 5.33: Local lift coefficient spanwise for the BWB model for air speed, u=40
m/s
µ·c·V0 0.8194·0.961·324.6·0.1
Re = = −5
= 1538856 ≈ 1.5·106
µ 1.661·10
This value indicates that the airflow is turbulent, if the usual transition Reynolds
number is considered to be 5·105 . With this information, the skin friction co-
efficient can also be estimated from an empirical graphic from [37], presented in
figure 5.38 or calculated from equation 5.3:
119
Figure 5.34: Lift coefficient vs. angle of attack of the BWB using TORNADO
and XFLR5 for air speed, u=40 m/s
Figure 5.35: Pitching moment coefficient vs. angle of attack of the BWB using
TORNADO and XFLR5 for air speed, u=40 m/s
0.45 0.45
Cfwing ≈ 2.58 0.65 = = 0.00412
(log Re) (1 + 0.144M 2 ) (log 1538856) (1 + 0.144·0.12 )0.65
2.58
(5.3)
With all these data, the parasite drag of the whole wing airframe (and fuselage)
can be estimated:
Cfwing Swetwing F Fwing 0.00412·3.3672·1.217

CD0wing = = ≈ 0.01113
Sref erence 1.5164
120
Figure 5.36: Drag polar of the BWB using TORNADO and XFLR5 for air speed,
u=40 m/s
Engine nacelles parasite drag estimation
Following the previous procedure, the wetted area of each engine nacelle was
estimated using CATIA and determined to be Swetengine = 0, 1486 m2 . The form
factor for an engine nacelle is approximated in a better way with equation 5.4,
learned from [37]:
0.35 0.35
F Fengine = 1 + =1+ = 1.167 (5.4)
f 0.461
0.22
where f is the slenderness of the body (the ratio between length and diameter).
In a similar way as previously done for the wing calculations, the skin friction
coefficient can be approximated using figure 5.38 and was considered to be the
same value as before: Cfengine ≈ 0.00412.
Thereby, the parasite drag of both engine nacelles can be now predicted:
Cfengine Swetengine F Fengine 0.00412·0.1486·1.167

CD0engine = 2· = 2· ≈ 0.00092
Sref erence 1.5164
Therefore, the total estimated parasite drag coefficient will be the combination
of the effects of both the wing and the engine nacelles:
CD0 = CD0wing + CD0engine = 0.01113 + 0.00092 = 0.01205
121
Figure 5.37: Aerodynamic Surface Form Factor estimations for different thickness
ratios and sweep angles. [37]
Induced drag coefficient estimation
In order to follow with the empirical estimation, the induced drag coefficient was
also calculated. First, the Oswald efficiency number must be approximated with
the equation 5.5, extracted from [10]:
e = 4.61 1 − 0.045·AR0.68 (cos Γ)0.15 − 3.1

(5.5)
where Γ represents the dihedral angle of the wing.

According equation 5.5 and with the geometrical data of the BWB baseline, the
Oswald efficiency number is approximately:
e = 4.61 1 − 0.045·3.530.68 (cos 5)0.15 − 3.2 = 0.9185

With this value and using equation 5.1, the induced drag coefficient can be
estimated:
CL2 CL2
CDi = = = 0.09817·CL2
π·AR·e π·3.53·0.9185
To conclude, the total drag coefficient is the combination of both parasite drag
and induced drag coefficients, as shown in equation 5.6:
CD = CD0 + CDi = 0.01205 + 0.09817·CL2 (5.6)
122
Figure 5.38: Skin friction coefficient vs. Reynolds number for different turbulent
transition locations. [37]
In order to compare this estimation with the numerical approaches shown

previously, they are all represented in figure 5.39, as well as the TORNADO
calculations with an additional offset of CD0 . It can be observed that the zero
lift drag coefficient value, CD0 , for the TORNADO calculations combinated with
the empirical estimations is 0.0112, around 10 times higher than the consider
before, which is probably much closer to the actual performance. The enhanced
TORNADO calculations with the drag estiamations (in light green in figure 5.39)
has a close agreement with the empirical estimations for low lift coefficient values,
but for CL values higher than 1, it correlates better with the VLM computations of
XFLR5. This results are satisfactory, and therefore, the new TORNADO values,
considering the parasite drag, will be taken into account from now on.
All in all, the majority of the aerodynamic results and tendencies obtained
using both methods correlate in a satisfactory way, enhancing the accuracy of this
analysis. With all this obtained information, the stability and control qualities of
the BWB model can be studied in the next section.
123
Figure 5.39: Drag polar of the BWB using TORNADO, XFLR5 and the empirical
estimation for air speed, u=40 m/s
124
Chapter 6
Analysis of the stability and

control of the BWB model
6.1 Theoretical introduction

As mentioned before, stability and control design allows sustained authority over
the aircraft at any point included in the flight envelope. The control system should
make the aircraft response to the pilot’s commands easy and effective. Stability
and Control are crucial factors in aircraft design, especially in tailless aircrafts
such as the BWB configuration. It is, therefore, essential to understand the fun-
damentals and physics behind this concepts.
6.1.1 Stability
Stability is a property of an equilibrium state. An aircraft in steady uniform
flight, also known as trimmed flight, must fulfill that its resultant forces and its
resultant moments referred to the center of gravity must be equalized to 0. There
are two different types of stability: static and dynamic.
6.1.1.1 Static stability
The main concept of static stability consists in that the aircraft tends to recover by
itself to its original state whenever a disturbance in the flight occurs, such a wind
125
CHAPTER 6. ANALYSIS OF THE STABILITY AND CONTROL OF THE
BWB MODEL
gust. This idea can be explained with an example of a ball on a surface, initially in
equilibrium, that experiences a perturbation from the equilibrium position, caused
by forces or moments. Three different possible behaviours can be distinguished:
1. Statically stable: The body tends to return inmediately to its equilibrium

position, as in the example in figure 6.1.
Figure 6.1: Statically stable behaviour.
2. Statically unstable: After the acting of forces and moments due to a dis-
turbance, the body continues moving away from its equilibrium position, as
would happen in the example in figure 6.2.
Figure 6.2: Statically unstable behaviour.
3. Neutrally stable: The body is disturbed but the moments remain equal to
zero, and the body stays in equilibrium. This option rarelly occurs in flight
dynamics. An example of this behaviour is shown in figure 6.3.
126
6.1. THEORETICAL INTRODUCTION
Figure 6.3: Neutrally stable behaviour.
Within the concept of static stability, three different aspects can be considered:
Longitudinal static stability: It refers to the aircraft’s stability in the pitching

plane. In case the aircraft is longitudinal statically stable, a small increase (or
decrease) of the angle of attack will cause a variation of the pitching moment
which decreases (or increases) the angle of attack, causing that it comes back
to the equilibrium condition. An approximate definition of the pitching moment
coefficient is given in the following equation:
Cm = Cm0 − CL (xcg − xac ) (6.1)
where Cm is the pitching moment coefficient, Cm0 is the pitching moment

caused by the wing coefficient, CL is the lift coefficient and (xcg − xac ) is the
distance between the centre of gravity, xcg , and the aerodynamic centre, xac .
The following equation shows the condition necessary of an aircraft for being
longitudinal statically stable, the derivative of the pitching moment coefficient
with respect to the angle of attack must be negative:
∂Cm
Cm α = <0 (6.2)
∂α
Lateral static stability: It deals with the aircraft’s ability to maintain the
wing level equilibrium with respect to the roll. The most important parameter
for setting this aspect in an acceptable level is the dihedral of the wing. Another
condition for an aircraft to be laterally stable is that the rolling moment resulting
from a positive disturbance in the roll direction, φ, must be negative as presented
in the following equations:
127
BWB MODEL
∂Cl
Cl β = <0 (6.3)
∂β
∂Cl
Cl φ = <0 (6.4)
∂φ
where Cl is the rolling moment coefficient, β is the sideslip angle and φ is the
roll angle.
Directional static stability: It is concerned with the aircraft’s stability with

respect to the Z axis. Thereby, a directional statically stable aircraft tends to
return to the equilibrium condition when it experiences a yawing disturbance, ψ,
by generating a yawing moment that restores it to the trimmed state. Therefore,
the derivative of the yaw moment with respect to the sideslip angle, β, needs to
be positive as shown in the following equation:
∂Cn
Cnβ = >0 (6.5)
∂β
where Cn is the yaw moment coefficient.
The results presented in section 5.1 show that the baseline BWB model studied
in TORNADO fulfills all these conditions, so it can be defined as statically
stable.
6.1.1.2 Dynamic stability
Dynamic stability refers to the vehicle’s characteristics as a result of a disturbance

from an equilibrium condition, taking into account the time history of the mo-
tion of the aircraft. This indicates that there is a resistance to the motion and,
therefore, an energy dissipation (also known as positive damping). It is important
to notice that static stability does not guarantee dynamic stability. Within this
concept, two different categories of response exist:
1. Dynamically stable: Implying positive damping, is once again subdivided

into two situations:
(a) One with a gentle recovery back to original condition with damped
128
oscillations.
(b) One with a gentle recovery back to original condition without damped
oscillations.
2. Dynamically unstable: Implying negative damping, is also subdivided into

three cases:
(a) Undamped oscillations of constant amplitude, usually known as neutral

dynamic stability.
(b) Forced oscillations, where the amplitude increases over time.
(c) A statically unstable situation consisting in a divergence from the equi-
librium condition after a disturbance. The aircraft should not possess
any statically unstable properties in any dynamic mode, or at the very
least, the divergence that results from such instability should not be
fast, so the pilot can reasonably apply a correction. In case the aircraft
shows any dynamically unstable oscillatory behaviour, the oscillations
need to be totally eliminated or damped sufficiently to reduce the fre-
quency through passive or active means.
The aeroelastic aircraft motion is governed by a non-linear set of equations, from

where the ones describing the motion of the structure and the ones describing
the aerodynamic forces have to be solved simultaneously. These equations are
usually linearized for small displacements with respect to an equilibrium point,
also known as trim condition. If a linear and discretized structural model is used,
such as the Finite Elements Methods (FEM), the equations of motion for an
aeroelastic system can be expressed as a set of differential equations according to:
Mẍ + Cẋ + Kx = fa (t) (6.6)
Here, M is the mass matrix, C the structural damping matrix, K the stiff-
ness matrix, x the nodal displacement vector and fa is the vector or aerodynamic
forces. A dot represents differentiation with respect to the time, t.
In the most general case, the vector of aerodynamic forces depends arbitrarily
on t. Using a linear incompressible aerodynamic model, such as the Doublet
129
BWB MODEL
Lattice Method (DLM), the aerodynamic forces can be expressed as a function of

the nodal displacements of the structure and the free-stream dynamic pressure,
q∞ , according to:
fa = q∞ A (k) x (6.7)
Here A is the complex aerodynamic influence matrix, and k̄ is the reduced

frequency and is defined as:
ωc
k̄ = (6.8)
u∞
where ω is the oscillation frequency, u∞ is the free-steam velocity and c is a
reference length, typically the Mean Aerodynamic Chord (MAC) of the wing.
For static stability analysis for rigid aircrafts, a classic eigenvalue problem is
obtained, expressed in the classical form as (A−λI) x. Solving this linear prob-
lem provides the eigenvalues, λi , and the corresponding eigenvectors, xi , which
define the modes of motion. However, the modes identification might fail when
the eigenvalues are not conjugate for some modes.
The SDSA module is used to determine the trim characteristics (such as the
static margin, the angle of attack, the deflection angle), aerodynamic data, such as
the drag coefficient and the dynamic modes of motion, that indicate the handling
quality characteristics. These modes and other cases are presented now and are
further explained by Rizzi in [38]. They are divided into those dealing with longi-
tudinal motion (short period and phugoid modes) and the lateral motion (dutch
roll, roll and spiral modes).
Dynamic stability modes for the longitudinal motion:
The dynamic stability of the longitudinal motion can be defined with the help
of the eigenvalues of the matrix A, which can be found by solving the following
equation:
det |λI−A| = 0 (6.9)
130
,where I is a 4x4 identity matrix. Solving the determinant of this equation

leads to a fouth degree characteristical polynomial expressed as follows, where ai
are the coefficients:
λ 4 + a1 λ 3 + a2 λ 2 + a3 λ + a4 = 0 (6.10)
The aircraft is considered dynamically stable if the real part1 of all its eigen-
values, λi , for the lateral and longitudinal motions is negative. In the rest of the
cases (were the real part of any eigenvalue is zero or positive), the aircraft will be
dynamically unstable. The different eigenvalues obtained are usually represented
in pole plots in order to observe better the stability characteristics.
In case the polynomial 6.10 is factorized, the short period and the phugoid modes
can be identified, as shown in equation 6.11:
s2 + 2 ξsp ωsp s + ωsp

2
s2 + 2 ξph ωph s + ωph
2

=0 (6.11)
,where the coefficients ξsp and ωsp are, respectively, the damping ratio and the
natural undamped frequency of the short period mode and ξph and ωph are, re-
spectively, the damping ratio and the natural undamped frequency of the phugoid
mode.
The solution of the quadratic equation for each factor is given by equation 6.12,
showing pair of conjugated solutions:
s1,2 = σ ± i ωd (6.12)
where the real, σ, and imaginary, ωd , (also known as the damped frequency2 )
parts are:
σ = −ξ ωn (6.13)
q
ωd = ωn 1 − ξ 2 (6.14)
The damping ratio for both modes of the longitudinal motion is a function of
the real part and the damped frequency as seen in equation 6.15.
1
considering the eigenvalues as complex numbers: λ = x + y i
2
In these expressions ωn referrs to the natural undamped frequency introduced before.
131
BWB MODEL
s 2
ωd 1

ξ= 1− =r 2 (6.15)
ωn
1 + ωσd
The period for both modes can be easily calculated with the equation 6.16
2π 2π
T = √ = (6.16)
ωn 1 − ξ 2 ωd
With this information, the short period and the phugoid modes can be identi-
fied and the flight qualities of an aircraft defined. A qualitative definition of these
two modes is provided in the following paragraphs.
The short period mode generates fast oscillations of the angle of attack while
the flight path is almost constant. It can be excited by rapidly deflecting
the elevator and it is usually fast and well damped in few seconds. In case
of simulated flight, it can be excited by an initial disturbance in pitch rate.
Figure 5.14 shows the time to halve the motion amplitude is plotted against
the period of oscillation for the B747-100, including the ICAO recommen-
dations in the plot.
The phugoid mode has very slow oscillations in the pitch angle and velocity,
with an almost constant angle of attack. It is excited during a climb, starting
from a trimmed flight condition, where the speed and lift decrease causing
the aircraft to pitch downward and enter a dive. During the dive, the air-
craft recovers speed and lift, restarting another climb. This case is not as
heavily damped as the short period one. This mode is mainly generated
by lift, which induce oscillation, and drag forces, which is a damper. An
example of the phugoid period for the B747-100 is presented in figure 5.15.
Dynamic stability modes for the directional and lateral motion:
Proceeding similarly as in the longitudinal motion case commented before, a

new characteristical polynomial when solving equation 6.9 for the lateral motion,
where di are the coefficients:
λ5 + d1 λ4 + d2 λ3 + d3 λ2 + d4 λ = 0 (6.17)
132
The polynomial in 6.17 can be factorized as shown in 6.18:
λ (λ + e) (λ + f ) λ2 + 2ξDR ωDR + ωDR

2

=0 (6.18)
The first factor, λ, represents the directional mode. For the root λ = 0, the
aircraft is neutrally stable for the directional mode. The term (λ + e) corresponds
to the spiral mode and the term (λ + f ) to the roll mode. Finally, the last factor
of 6.18, (λ2 + 2ξDR ωDR + ωDR 2
), agrees with the Dutch roll mode, where ξDR is
the damping ratio (usually small) and ωDR is the natural undamped frequency
for this mode. Thereby the Dutch roll eigenvalues will be λDR = µ ± i η.
Operating likewise as for the longitudinal motion, the damping ratio and the
natural undamped frequency for the Dutch roll mode can be calculated with the
Dutch roll eigenvalue real and imaginary parts, as shown in equation 6.19:
1
ξDR = r 2 (6.19)
η
1+ µ
−µDR
ωDR = (6.20)
ξDR
For the roll mode, the roll damping time, TξRoll , is defined in equation 6.21,
where λRoll is the eigenvalue for the roll mode.
−1
TξRoll = (6.21)
λRoll
The last required value for the stability analysis is the time to damp to half
amplitude, T1/2Spiral , of the spiral mode, defined in equation 6.22, where λSpiral is
the eigenvalue for the spiral mode.
ln 2
T1/2Spiral = (6.22)
λSpiral
A qualitative definition of these three modes is provided in the following para-
graphs.
The Dutch roll mode involves fast oscillatory swinging of the aircraft. It con-
sists in low frequency oscillations in bank, yaw, and sideslip angles. In a
flight-state display, a dutch roll will be noticed by the velocity vector circle
oscillating from side to side. This mode can be excited by rapidly deflecting
the rudder, and should be fast and well damped. In simulated flight it can
133
BWB MODEL
be excited by an initial disturbance in the sideslip angle. The most suscep-

tible aircrafts for suffering this motion usually have large aft swept wings
and with a dihedral wing form. Numerical results for the dutch roll for the
B747-100 can be observed in figure 5.16. An illustration of the dutch roll
mode is can be visualized in figure 6.4.
Figure 6.4: Course of motions of the dutch roll mode. [39]
The roll mode is a fast convergent and heavily damped motion. This mode
follows a lateral directional disturbance causing the aircraft to experience a
motion mainly with respect to the longitudinal body axis. The rest of the
parameters vary very slowly, allowing the neglection of their time derivatives.
An scheme of this mode is presented in figure 6.5.
134
Figure 6.5: Course of motions of the roll mode. [39]
The spiral mode is a slowly convergent or divergent motion characterized by

variations of the roll, φ, and yaw, ψ, angles. Despite its small value, the
sideslip angle, β, cannot be neglected because the aerodynamic moments
depend on it. A graphic explanation of this motion mode is presented in
figure 6.6.
Figure 6.6: Course of motions of the spiral mode. [39]
Information about the flight quality specifications published by the FAA and
further explanations about stability can be found in Nelson [39].
135
BWB MODEL
6.1.2 Control
Aircraft control deals with the vehicle’s response after the intentional manipula-
tion of the control devices in order to alter from a given equilibrium condition.
Aerodynamic control is mainly achieved with the employment of three sets of
dedicated control surfaces, as shown before in figure 2.8:
1. Elevators in pitch.
2. Ailerons in primary roll and secondary yaw.
3. Rudder in primary yaw and secondary roll.
Control is a distinct problem from that of stability, although they are both re-
lated. The control system of an aircraft should provide a characteristic response to
a control device action, such as a deflection, so that the desired action is achieved
and control reversal3 is avoided entirely. The sensitivity of the control devices
displacements should also be calibrated in a manner that a statically stable air-
craft is not too stable, in order to avoid trim drag. In addition, the magnitude
of the force required to manipulate the control should increase steadily with the
control device displacement and be correspondingly reflected in the magnitude of
the aircraft response. Finally, a little or negligible time lag should prevail in the
aircraft response to a control input, so that the global performance improve.
Further information about the Stability and Control theoretical basics and
how the SDSA module from CEASIOM adapts them can be found in the doc-
umentation in the documentation included in the CEASIOM files [46] and [16],
prepared within the SimSAC project.
6.2 Stability analysis of the BWB baseline

In the following subsections the static and dynamic stability qualities of the BWB
baseline will be analyzed, in order to illustrate the calculations. As commented
before, the control surfaces allocation and sizing for the BWB baseline are the
3
In this situation, the stick input from the pilot will have no effect on the aircraft, and after
the control reversal, the desired effect will be reversed, causing a very confusing situation for
the pilot flying the aircraft.
136
6.2. STABILITY ANALYSIS OF THE BWB BASELINE
ones presented in table 4.3. The procedures explained for obtaining the aerody-
namic data and the stability analysis are repeated for every design iteration with
different control surfaces configuration. The output from TORNADO has to be
manipulated and enhanced with the weight and balance data before it can be
loaded in the SDSA module.
6.2.1 Static stability of the BWB baseline
This feature can be defined utilizing the stability derivatives calculated in section
5.3 and regarding the criteria explained before in this chapter. Table 6.1 shows
that the conditions for static stability are fulfilled in all the directions: longitudi-
nal, lateral and directional. Therefore, it can be concluded that the BWB baseline
is statically stable. Moreover, the beta derivative of the side force, CYβ , is also
negative so the aircraft is also static stable in the sideslip motion.
Motion Parameter Condition to be statically stable Value

Longitudinal Cmα <0 −0.1139 rad−1
Lateral Cl β <0 −0.1089 rad−1
Directional Cnβ >0 0.0379 rad−1
Table 6.1: Stability derivatives for the different motions of the BWB baseline and
the static stability criterion.
The static margin, Kn , is a concept used to characterize the static stability,

defined as the longitudinal distance between the center of gravity and the neutral
−xcg
point of the aircraft normalized to the MAC: Kn = xac M AC
. The static margin
variation with the angle of attack is represented in figure 6.7. The usual require-
ment states that the static margin for airplanes should be approximately 5%, in
the BWB case, it can be observed that for higher angles of attack, the static
margin tends to a value between 5 and 6%, therefore the satisfactory stability
performance is confirmed.
137
BWB MODEL
Figure 6.7: Static margin vs. angle of attack of the BWB for air speed u=40 m/s
6.2.2 Dynamic stability of the BWB baseline

The analysis in the SDSA module were performed for altitudes from H = 0 m
to H = 5000 mwith values every 1000 m, and airspeeds from u = 20 m/s to u =
120 m/s with values every 20m/s.
6.2.2.1 Stability for the longitudinal motion
First of all, the longitudinal dynamic stability will be studied. As explained

before, the main motion modes for this direction are the short period and the
phugoid modes. The required information about them can be obtained from the
eigenvalues of the matrix A, by solving the characteristic polynomial for this
configuration, shown in equation 6.23:
λ4 + 5,4268λ3 + 15,0188λ2 + 8,0712λ + 15,1368 = 0 (6.23)
By solving this polynomial, the eigenvalues can be easily obtained:
λ1 = −2.63843 + 2.47146 i
λ2 = −2.63843 − 2.47146 i
λ3 = −0.07497 + 1.07358 i
λ4 = −0.07497 − 1.07358 i
138
These eigenvalues can be visualized in the S-plane representation in figure 6.8,

and it can be observed that both longitudinal dynamic modes are stable because
all the values are located on the left side of the S-plane, i.e., they all have negative
real part.
Figure 6.8: S-Plane diagram showing the eigenvalues for the longitudinal motion
of the BWB baseline
Short period mode
Conjugate eigenvalues λ1 and λ2 correspond to the short period mode due to

their higher frequency (and therefore shorter period) compared to the other two
eigenvalues. The factorization of the polynomial 6.23 with the mentioned eigen-
values leads to the characteristic polynomial of the short period mode, as shown
in equation 6.24:
s2 + 5,2769s + 13,0694 = 0 (6.24)
With the explanation previously provided in equations 6.11 and 6.16, the nat-
ural undamped frequency, damping ratio and period of the short period mode can
be defined easily:
√
ωsp = 13.0694 = 3.6152 rad/s
139
BWB MODEL
5.2769
ξsp = = 0.7298
2·3.6152
2π
Tsp = √ = 2.5422 s
3.6152 1 − 0.72982
The positive damping ratio indicates that the aircraft is able to damp out the
short period mode by itself, bringing it back to a stable longitudinal position. A
simulation of this stability mode for the obtained values is shown in figure 6.9
and the results from the SDSA module for several airspeeds and altitudes are
presented in figure 6.10. Both figures confirm the stability of the aircraft for this
mode.
Figure 6.9: Simulation of the short period stability mode for the BWB baseline
for a initial pitch perturbation of 10º of amplitude
Figure 6.10: SDSA results for the short period mode of the BWB baseline com-
pared to ICAO recommendations found in [40].
140
Phugoid mode
As stated before, conjugate eigenvalues λ3 and λ4 correspond to the phugoid mode.

In a similar way as perfomed with the short period mode, the factorization of the
polynomial 6.23 leads to the characteristic polynomial of the phugoid mode, as
shown in equation
s2 + 0,1499s + 1,1582 = 0 (6.25)
With the same procedure as explained in equation 6.11 and 6.16, the natu-
ral undamped frequency, damping ratio and period of the phugoid mode can be
computed:
√
ωph = 1.1582 = 1.0762 rad/s
0.1499
ξph = = 0.0696
2·1, 0762
2π
Tph = √ = 5.8525 s
1.0762 1 − 0.06962
The positive damping ratio indicates that the aircraft is able to damp out
the phugoid mode by itself, bringing it back to a stable longitudinal position. A
simulation of this stability mode for the obtained values is shown in figure 6.11.
A simulation of this stability mode for the obtained values is shown in figure 6.11
and the results from the SDSA module for several airspeeds and altitudes are
presented in figure 6.12. Both figures confirm the stability of the aircraft for this
mode.
6.2.2.2 Stability for the lateral motion
The main stability derivatives for this motion were presented previously in tables
5.3, 5.4 and 5.6. As explained before, the main motion modes for this direction
are the Dutch roll, the roll and the spiral modes. In a similar way as performed
for the longitudinal motion, the required information about them can be obtained
141
BWB MODEL
Figure 6.11: Simulation of the phugoid stability mode for the BWB baseline for
a initial pitch perturbation of 10º of amplitude
Figure 6.12: SDSA results for the phugoid mode of the BWB baseline compared
to ICAO recommendations found in [40].
from the eigenvalues of the matrix A, by solving the characteristic polynomial for
this configuration, shown in equation 6.26:
λ4 + 5,8910λ3 + 1,5982λ2 + 4,5163λ + 0,1035 = 0 (6.26)
By solving this polynomial, the eigenvalues can be easily obtained:
λ1 = −0.0594 + 0.8810 i
λ2 = −0.0594 − 0.8810 i
λ3 = −5.7491
λ4 = −0.0231
142
A remarkable difference of this polynomial compared to the one presented in

6.18 is that the characteristic polynomial is a fourth degree polynomial without
the root λ = 0. Therefore, it can be concluded that the BWB baseline is stable
for the directional motion.
These eigenvalues are represented in the S-plane representation in figure 6.13,

and it can be noticed that all the lateral dynamic modes are stable because,
once again, all the values of the eigenvalues have negative real part. However, in
this case, the Dutch Roll and Spiral modes eigenvalues are notably closer to the
imaginary axis. Therefore, a special attention should be paid to these two modes,
because they are in the limit to be unstable.
Figure 6.13: S-Plane diagram showing the eigenvalues for the lateral motion of
the BWB baseline
Dutch roll mode
According to the definition given in equation 6.18, conjugate eigenvalues λ1 and

λ2 belong to the Dutch roll mode. The factorization of the polynomial 6.26 with
the mentioned eigenvalues provides the characteristic polynomial of the Dutch roll
mode, as observed in equation 6.27:
s2 + 0,1188s + 0,7797 = 0 (6.27)
With the help of the Equations 6.20 and 6.21 the naturally frequency !D and
the damping ratio D for the Dutch Roll mode can be computed with the following
143
BWB MODEL
results: If the investigated aircraft experiences the Dutch Roll, it can damp it out
by itself, because the D value is positive.
With the explanation previously provided in equations 6.19 and 6.20, damping
ratio of the Dutch roll mode can be defined easily:
1
ξDR = r 2 = 0.0673
1 + 0.8810

0.0594
0.0594
ωDR = = 0.8830 rad/s
0.0673
The positive damping ratio indicates that the aircraft is able to damp out the
Dutch roll mode by itself. However, it is a quite small value, so this mode is not
so strongly damped as the modes studied before in the longitudinal motion. The
results from the SDSA module for this mode for several airspeeds and altitudes
are presented in figure 6.14, confirming its stability, with smaller margin than the
previous two studied modes.
Figure 6.14: SDSA results for the Dutch roll mode of the BWB baseline compared
to EASA CS-23 recommendations found in [41].
Roll mode
The next eigenvalue, λ3 , is associated to the roll mode, which is negative, indi-
cating that the BWB baseline is stable for the roll mode, as commented before.
144
This case is simpler than the previous ones considered before, therefore no factor-
ization is required. Using the equation 6.21, the roll damping time can be easily
calculated:
1
TξRoll = = 0.1739 s
5.7491
This value and the eigenvalue indicate that the aircraft is very stable for the
roll mode, being able to damp it out by itself. The results from the SDSA module
for this mode for several airspeeds and altitudes are presented in figure 6.15,
confirming its stability, with smaller margin than the previous two studied modes.
Figure 6.15: SDSA results for the roll mode of the BWB baseline compared to
Cooper-Harper pilot assessment ratings explained in [40].
Spiral mode
Finally, the last eigenvalue, λ4 , is related to the spiral mode, which is also negative,
indicating that the BWB baseline is stable for the spiral mode, as commented
before. Similarly than before, using the equation 6.22, the time to damp to half
amplitude can be easily calculated:
ln 2
T1/2Spiral = = 30.0064 s
0.0231
This value indicates that the aircraft is slightly stable for the spiral mode,
and a special attention should be paid to this mode. The results from the SDSA
module for this mode for several airspeeds and altitudes are presented in figure
6.16, confirming its stability, with smaller margin than the previous two studied
modes.
145
BWB MODEL
Figure 6.16: SDSA results for the spiral mode of the BWB baseline compared to
MIL-F-8785C recommendations.
6.3 Sizing of the control surfaces and consequences

As mentioned in subsection 4.1.3 about the control surfaces, the allocation and
sizes of the control surfaces of the BWB baseline (Model 1) are the ones presented
in table 6.2. In the previous section it was demonstrated that the stability and
control qualities of the BWB baseline are satisfactory. However, this is probably
due to the conservative design of its control surfaces, which were, apparently,
larger than necessary. Therefore, an optimization study was performed in order
to obtain the minimum control surfaces size that still provide satisfactory flying
qualities, within a compromise solution. Due to its importance, the study of the
winglets will be commented separately in section 6.4. In addition, the effect of the
rudders is relatively quite uncoupled in other directions besides the yaw (where
the most critical stability modes, Dutch roll and spiral, are included). Therefore,
only the elevators and the elevons will be sized in this section.
Control Surface c1 (m) c2 (m) Span (m) Surface (m2 )

Inner Elevator 0.149 0.162 0.151 0.0235
Outer Elevator 0.108 0.081 0.140 0.0132
Elevon 0.081 0.054 0.639 0.0431
Rudder 0.080 0.080 0.150 0.0120
Table 6.2: Control surfaces geometry for the BWB baseline model. (Model 1)
One of the main issues for the BWB concept is that the control power is low
in pitch and yaw due to relatively small moment arms, compared to conventional
tailed aircrafts. In addition, due to the lack of empennage and smaller aspect ratio
146
6.3. SIZING OF THE CONTROL SURFACES AND CONSEQUENCES
of this aircraft, the available space to locate control surfaces is limited. There-
fore, the main design variable for the control surfaces used in this optimization
was their relative chord, keeping the spans constant (except for the winglet case).
Smaller spans would mean larger deflections to obtain the same control effect than
before, involving more drag and more adverse conditions for the boundary layer
or even saturation of the control devices.
Several configurations were tested in order to achieve the minimum control

surface with satisfactory stability. The optimum one which was obtained (Model
2), is presented in table 6.3. The inner elevators surface was reduced around 20%,
the outer elevators around 15% and the elevons approximately 9%. As commented
before, the rudders remained constant for this optimization process.
Control Surface c1 (m) c2 (m) Span (m) Surface (m2 ) BL Comparison

Inner Elevator 0.119 0.129 0.151 0.0187 -20.43%
Outer Elevator 0.092 0.069 0.140 0.0113 -14.59%
Elevons 0.074 0.049 0.639 0.0393 -8.82%
Rudder 0.080 0.080 0.150 0.0120 —
Table 6.3: Control surfaces geometry for the reduced control surfaces BWB con-
figuration, including the comparison with the baseline surfaces. (Model 2)
The flying qualities of Model 2 with the reduced control surfaces are presented
as its stability modes from figures 6.17 to 6.21. In general, both longitudinal
stability modes, short period (fig. 6.17) and phugoid (fig. 6.18), were markedly
displaced to less stable positions if compared with the baseline’s results. This
can be explained due to the high dependency of the longitudinal stability modes
with the geometry and deflections of the elevators, which were notably reduced
for Model 2. The roll mode (shown in figure 6.20) also became less stable than
the previous model, due to the reduced value of the elevons surface that are the
main responsibles to control this stability mode.
As it could be expected, the Dutch roll (fig. 6.19) and spiral modes (fig. 6.21)
remained almost exactly the same as for the baseline, they only turned slightly
less stable after the control surface reduction. This fact is caused because they
are mainly motion modes in the yaw direction, which is mostly controlled with
the rudders, which were not modified in this optimization step.
147
BWB MODEL
Figure 6.17: SDSA results for the short period mode of the BWB Model 2 com-
Figure 6.18: SDSA results for the phugoid mode of the BWB Model 2 compared
However, all the applied modifications still kept Model 2 inside the stability
margins recommended by the international flying criteria and left some safety
margins in order to be able to reduce the winglets size in the next section.
6.4 Sizing of the winglets and rudders

Following a similar procedure as in the previous section, different winglet config-
urations were tested in order to achieve the minimum winglet and rudder surface
with satisfactory stability. However, in this case the span of the winglet was mod-
ified due to the importance of vertical surfaces for the radar detectability.
For this optimization process, the operating margin is not as wide as for the
rest of the control surfaces due to the smaller stability qualities for the Dutch
148
6.4. SIZING OF THE WINGLETS AND RUDDERS
Figure 6.19: SDSA results for the Dutch Roll mode of the BWB Model 2 compared
Figure 6.20: SDSA results for the roll mode of the BWB Model 2 compared to
roll and spiral modes, which are mainly handled by the rudder, as stated before.
Smaller control surface areas obviously generate smaller control forces.
The rudder chord was increased around 20% and the winglet (and rudder)
span was reduced around 20% with respect to the baseline, meaning a surface
reduction of the 4% for the rudder. The differences between both configurations
is shown in figure 6.22 . The optimum case obtained (Model 3) is presented in
table 6.4.
149
BWB MODEL
Figure 6.21: SDSA results for the spiral mode of the BWB Model 2 compared to
(a) (b)
Figure 6.22: Different winglet and rudder configurations considered: (a) Baseline.
(b) Model 3: Reduced winglet (and rudder) span and increased rudder chord.
Control Surface c1 (m) c2 (m) Span (m) Surface (m2 ) BL Comparison

Inner Elevator 0.119 0.129 0.151 0.0187 -20.43%
Outer Elevator 0.092 0.069 0.140 0.0113 -14.59%
Elevons 0.074 0.049 0.639 0.0393 -8.82%
Rudder 0.096 0.096 0.120 0.0115 -4%
Table 6.4: Control surfaces geometry for the reduced control surfaces and winglets
BWB configuration, including the comparison with the baseline surfaces. (Model
3)
The flying qualities of Model 3 with the reduced rudder are presented as its
stability modes from figures 6.23 to 6.27. In general, both longitudinal stability
modes, short period (fig. 6.23) and phugoid (fig. 6.24), and the roll mode (fig.
6.26) remained almost constant compared to the previous model, due to the weak
influence of the rudder effect in these three modes.
150
However, the Dutch roll (fig. 6.25) and spiral modes (fig. 6.27) reduced their
stability noticeably. As commented previously, this can be explained because
they are mainly motion modes in the yaw direction, which strongly depend on the
rudders. Both modes are almost unstable for some flight conditions (especially
at higher flight altitudes) in this model, what caused the end of the optimization
process, in order not to make the BWB unstable.
Figure 6.23: SDSA results for the short period mode of the BWB Model 3 com-
Figure 6.24: SDSA results for the phugoid mode of the BWB Model 3 compared
151
BWB MODEL
Figure 6.25: SDSA results for the Dutch Roll mode of the BWB Model 3 compared
Figure 6.26: SDSA results for the roll mode of the BWB Model 3 compared to
Figure 6.27: SDSA results for the spiral mode of the BWB Model 3 compared to
152
6.4.1 Configuration with no winglets and no rudders

The performance of the BWB model with no winglets (and, therefore, no rudders)
at all was studied in order to investigate wheter it could be a feasible option. This
concept involves a reduced radar detectability and reduced parasite drag but, on
the other hand, the induced drag increases due to the lack of winglets, therefore
it is a compromise solution.
As commented previously in section 3.2, actual aircrafts like the stealth bomber
B-2 Spirit do not have any winglets and achieve the stability in the yaw direction
with split rudders and differential thrust. However, the B-2 bomber required a
complex quadruplex computer-controlled fly-by-wire system to mantain its stabil-
ity. In this project, the fly-by-wire system was not considered as an option due to
its complexity.
Figure 6.28: BWB model with no winglets or rudders.
As it could be expected, the no-winglet model (figure 6.28) becomes too un-
stable for the lateral motion (with the dutch roll and spiral eigenvalues becoming
unstable, with positive real part) and there is no further interest in investigating
it if no augmented stability systems are to be considered, such as the fly-by-wire
system.
Other BWB design option that does not involve winglets was presented by
Rahman and Whidborne in [26] and consists in placing the rudders in two twin
153
BWB MODEL
vertical fins near the trailing edge of the center body, as shown in figure 6.29.
The main reason for modifying the MOB BWB thereby was mainly due to the
poor qualities of the previous design (with rudders in the winglets, as shown in
figure 3.10) for Dutch roll (noticeably poorer than the presented in this project).
Therefore, bigger rudders were placed in this new configuration, improving the
general directional stability of the aircraft. However, the no-rudder configuration
was not addressed, confirming the importance of these devices.
Figure 6.29: Illustration of the MOB BWB with no winglets and two vertical fins
where the rudders are placed. [26]
6.5 Final model

After observing the influence of the size of the control surfaces in the stability of
the BWB aircraft, the model considered as definitive was Model 3, which com-
bines the reduced control surfaces and the improved rudder and reduced winglet
with satisfactory flight performance and slight drag reduction, as shown in figures
6.30 and 6.31.
All in all, a global reduction of 11.6% of the control surfaces area was achieved,
reducing the weight, complexity and drag of the design compared to the baseline
(Model 1). In addition, the winglets span was reduced 20%, which would notably
improve the radar detectability features of the BWB.
Finally, an overall drag reduction of 3.2% was achieved for the whole model
compared to the baseline design. As an example, the zero-lift drag coefficient, CD0 ,
decreased from 0.0112 to 0.0108. The sketches for the final model are gathered in
154
6.6. ANIMATIONS OF THE MOTIONS
Figure 6.30: Comparison between the drag polars of the optimized Model 3 with
reduced control surfaces areas and the baseline
Figure 6.31: Detail of the comparison between the drag polars of the optimized
Model 3 with reduced control surfaces areas and the baseline
the Appendix.
6.6 Animations of the motions

As a final study for this project, the actual motions of the BWB final model were
simulated as MATLAB animations. The motions chosen were the two more in-
teresting and characteristic directions: yawing and rolling. The code employed
for these animations was a modified version from a code developed by Thomas
Perrot in [42].
155
BWB MODEL
This study is very interesting and illustrative for visualizing the real movement
of the BWB for different perturbations and initial conditions introduced by the
user. Unfortunately, it is very difficult to represent in paper a video animation4 .
As an example, few screenshots of the MATLAB output are presented in figures
6.32 for the pure roll motion and 6.33 for the pure yaw motion.
(a) (b)
Figure 6.32: MATLAB simulation for the BWB geometry motion in the roll
direction. (a): Initial conditions. (b): Final state.
(a) (b)
Figure 6.33: MATLAB simulation for the BWB geometry motion in the yaw
direction. (a): Initial conditions. (b): Final state.
4
For the videos and simulations, please contact the author: robertomerino-
martinez@gmail.com
156
Chapter 7
Conclusions
In general, this project has been a satisfactory and interesting experience for
analyzing an unconventional aircraft like the Blended Wing Body with new com-
putational tools like CEASIOM and CPACScreator. The design procedure of a
new aircraft is surely of multidisciplinary nature and it usually requires several
specialists for each field. However, most numerical tools used for aircraft design
till now have been developed and verified in order to analyze conventional aircraft
configurations. The research conducted in this thesis is probably one of the first
attempts to design and analyze an unconventional aircraft using these tools.
However, the available resources were limited and, at this stage of the design
process, higher fidelity numerical analysis, with tools such as Edge or RANS, could
be performed (in case of having larger computational power and time) in order
to get a more realistic idea of the actual aerodynamic and flight performance. In
addition, experimental validation with a 3D model in a wind tunnel becomes very
useful and important in the next design steps.
All in all, this has been a very interesting opportunity to give an insight of
the BWB concept and its potential advantages and to introduce promising and
powerful tools for aircraft design, such as CEASIOM and CPACS. These two ideas:
new aircraft configurations and new computational tools could mean a revolution
in the traditional design process, making it more accessible and comfortable to
the user.
157
CHAPTER 7. CONCLUSIONS
7.1 Main achievements

Looking back to the list of objectives proposed in the introduction chapter, it can
be stated that all of them have been fulfilled in a successful way:
First of all, the new CC tool is now able to work with the rest of the CEA-
SIOM modules and can even handle unconventional configurations like the BWB
aircraft. In fact, this was the first time (at least documented) that an unconven-
tional aircraft has been analyzed in CEASIOM.
The literature research of the BWB concept was useful but, unfortunately,
the amount of BWB aircrafts and available information about them was very lim-
ited. However, nowadays many researchs are taking place about this new concept,
therefore in the near future it may be easier to have more data to compare with.
The BWB design procedure was satisfactory and challenging but, at the end,
provided satisfactory results, perhaps better than was expected. The agreement of
the experimental results by Carlsson in [5] with the obtained results and the cor-
relation between the outputs of the different tools used (TORNADO and XFLR5)
indicates that the design procedure was accurate.
In general, the most important drawback of the BWB configuration, the dif-
ficulties in stability and control, was solved in a successful manner indicating an
encouraging future for this aircraft concept and confirming that computational
tools such as CEASIOM are very powerful and useful in the first stages of the
aircraft design process.
7.2 Future work

To conclude, some outlook about what could be done in the future is presented
in the following list:
• It is desirable to connect further design tools to CPACS to exploit a wider

data set and seize its format advantages. In the future, CPACSCreator
could be transformed into the hub of the design framework in order to make
158
7.2. FUTURE WORK
the information transfer between different modules easier and more com-
fortable for the user1 . Furthermore, the drag estimations that TORNADO
performs are not accurate enough, as shown in section 5.6, and should in-
clude the fuselage effect and the parasite drag with predictions similar than
the presented here.
• A more complex 3D geometry optimisation could be done in order to improve

the aerodynamic performance of the BWB aircraft. Other planforms and
airfoils may also be employed. In addition, the winglet shape could also
be transformed into a sharklet, like in the new Airbus A3202 , in order to
improve the global aerodynamic performance.
• In case of having enough resources and computational power, higher-fidelity

analysis tools, such as Edge or RANS, could be employed for studying the
aerodynamic behaviour in order to get more precise results and, perhaps,
study the compressibility phenomena involved in the transonic regime. Fur-
ther research could also include the airframe flexibility analysis, also known
as aeroelasticity, and engine dynamics study. In case a flexible aircraft is
considered, the fuel transfer from one part to another of the BWB aircraft
could also be simulated to balance the moment changes without elevator
deflections (and therefore with less drag).
• More complex wind tunnel and flight tests would show the real performance
of the BWB model, leading to a more accurate comparison of the obtained
values. An UAV radio-controlled model would be a nice opportunity to
verify all the results obtained here. In addition, a full investigation into the
controllability of the BWB aircraft should include a wider range of flight
conditions and geometry configurations.
• In case of implementing augmented stability systems to the model, such as

the fly-by-wire system, more radical modifications could be performed in the
baseline, and it can even turn into a naturally unstable aircraft and perform
correctly.
1
Right now the transition between some modules of CEASIOM using CPACS is still quite
arduous and requires to undergo manually
2
Sharklets are expected to result in a reduced fuel burn of up to 4 percent, which may
correspond to an annual CO2 reduction of around 700 tonnes per aircraft. [65]
159
CHAPTER 7. CONCLUSIONS
160
Bibliography
[1] Boeing Commercial Airplanes. Current Market Outlook 2013-2032. USA.

2013.
[2] Eurocontrol. Long-term forecast of air traffic 2004-2025. Brussels. 2005.
[3] Airbus. Global Market Forecast 2007-2026. Blagnac, France. 2008.
[4] Boeing Commercial Airplanes. Current Market Outlook 2007. Seattle, USA.
2007.
[5] M. Carlsson. Design and Testing of a Blended Wing Body Aeroelastic Wind-
Tunnel Model. J. AIRCRAFT, VOL. 40, NO. 1.
[6] A. Da Ronch, C. McFarlane, C. Beaverstock, J. Oppelstrup, M. Zhang, A.

Rizzi. Benchmarking Ceasiom Software To Predict Flight Control And Fly-
ing Qualities Of The B-747. ICAS 2010, 27th International Congress of the
Aeronautical Sciences.
[7] C. Nickol. Conceptual Design Shop. Presentation to Conceptual Aircraft De-

sign Working Group (CADWG21). September 2004.
[8] A.Rizzi, M.Zhang, B.Nagel, D.Boehnke, P.Saquet. Towards a Unified Frame-

work using CPACS for Geometry Management in Aircraft Design. AIAA
Paper 2012-0549.
[9] M. Zhang. Application and Development of the CEASIOM-SUMO-EDGE

Suite for Rapid AeroData Assessment of Aircraft Flying Qualities. KTH
Stockholm. 2011.
[10] D. Raymer. Aircraft Design: A Conceptual Approach. Third Edition, AIAA

Education Series.
161
BIBLIOGRAPHY
[11] B.H. Cook. Flight Dynamics Principles, Butterworth Heinemann. Cranfield.

1997.
[12] P. Molitor. CEASIOM Tier I and Tier I+ modules for simulation of aircraft
stability and control characteristics. KTH Stockholm. July 2009.
[13] D.Eller. SUMO User Guide. KTH Royal Institute of Technology, Stockholm,
Sweden.
[14] P. Eliasson. EDGE, a Navier-Stokes Solver for Unstructured Grids. FOI Re-
port, FOI–R– 0298–SE, 2005.
[15] A. Rizzi, J. Oppelstrup, M. Zhang, and Tomac M. Coupling parametric air-

craft lofting to CFD & CSM grid generation for conceptual design. 49th AIAA
Aerospace Sciences Meeting, Orlando, USA. 2011.
[16] SimSAC. SDSA - Theoretical basics. Stockholm. 2009.
[17] SimSAC. SDSA - User manual. Stockholm. 2009.
[18] L. Cavagna, A. Da Ronch, S. Ricci. NeoCASS - Next generation Conceptual

Aero Structural Sizing. Milano, Dipartimento di Ingegneria Aerospaziale -
Politecnico di Milano. February 2009.
[19] D. Boehnke. CPACS Documentation. German Aerospace Center (DLR),

2011.
[20] R.H. Liebeck , M.A. Page , B.K. Rawdon. Blended wing body subsonic com-
mercial transport. AIAA Paper 98-0438. The Boeing Company Long Beach,
CA. 1998.
[21] D.Schmitt. The Airbus A380 - Towards a New Future for Air Transport.
EWADE 2007 Conference. Samara, Russia. June 2007.
[22] N. Qin, A. Vavalle, A. Le Moigne, M.Laban, K.Hackett, P.Weinerfelt. Aero-

dynamic considerations of blended wing body aircraft. Progress in Aerospace
Sciences. 2004.
[23] A. Bowers. Blended-Wing-Body: Design Challenges for the 21st Century.

TWITT (The Wing Is The Thing) meeting. NASA Dryden Flight Research
Center. September 2000.
162
BIBLIOGRAPHY
[24] IPCC. Aviation and the Global Atmosphere: A Special Report of the Inter-
governmental Panel on Climate Change. Cambridge University Press. 1999.
[25] A.J. Morris FREng. MOB A European Distributed Multi-Disciplinary Design

and Optimisation Project. Department of Engineering Mechanics and Struc-
tures. Cranfield University. Bedford, United Kingdom. AIAA Paper 2002-
5444.
[26] N.U. Rahman, J.F. Whidborne. A Lateral Directional Flight Control Sys-
tem for the MOB Blended Wing Body Planform. Department of Aerospace
Sciences, Cranfield University.
[27] G. La Rocca, L. Krakers, M.J.L. van Tooren. Development of an ICAD Gen-

erative Model for Blended Wing Body Aircraft Design. Delft University of
Technology. The Netherlands. AIAA Paper 2002-5447.
[28] M. Strettner, R. Voss. Aeroelastics, Flight Mechanics and Handling Qualities

of the MOB BWB Configuration. AIAA Paper 2002-5449.
[29] L. Reichert. Further Developments of the AcBuilder Tool for constructing

Geometrical Models of Aircrafts: Trailing Edge Devices. Royal Institute of
Technology (KTH), German Aerospace Center (DLR). Stockholm, January
2012.
[30] M. Voskuijl, G. La Rocca, F.Dircken. Controllability of Blended Wing Body

Aircraft. Delft University of Technology. 26th International Congress of the
Aeronautical Sciences, 2008.
[31] T. Hiebel. Further developments of the AcBuilder tool for constructing ge-
ometrical models of aircrafts: Engines, Spars, Ribs and Fuel Tanks. KTH
Royal Institute of Technology, Stockholm, Sweden. January 2012.
[32] T. Melin. A Vortex Lattice MATLAB Implementation for Linear Aerody-

namic Wing Applications. KTH Royal Institute of Technology, Stockholm,
Sweden. December 2000.
[33] M. Munk. The Aerodynamic Forces on Airship Hull. NACA Report No. 184,
1924.
163
BIBLIOGRAPHY
[34] C. H. Rodney, D. R. Nordwall. The Simulation of a Jumbo Jet Transport

Aircraft. Volume II. modelling Data (B747). The Boeing Company, Wichita
Division. 1970.
[35] S. Ramirez. Aerodynamic analysis of a sustainable-energy sailplane using

CEASIOM. KTH Royal Institute of Technology, Stockholm, Sweden. 2013.
[36] Guidelines for XFLR5. Analysis of foils and wings operating at low Reynolds
numbers. February 2011.
[37] J. Anderson. Fundamentals of Aerodynamics. 4th Edition. McGraw-Hill.

ISBN 0-07-125408-0.
[38] A. Rizzi. Aerodynamic Design, a Computational Approach. Course SD2610:

Computational Aerodynamics in Aircraft Design. KTH Royal Institute of
Technology, Stockholm, Sweden. Spring Term 2011.
[39] R. Nelson. Flight Stability and Automatic Control. WCB McGraw-Hill, New
York, USA. 1998.
[40] G. Cooper, R. Harper. The Use of Pilot Rating in the Evaluation of Aircraft
Handling Qualities. NASA TN D–5153, 1969.
[41] EASA. Certification Specifications for Normal, Utility, Aerobatic, and Com-
muter Category Aeroplanes CS-23. Amendment 2 (Corrigendum) Annex to
ED Decision 2010/008/R. September 2010.
[42] T. Perrot. Study of the stability of an unmanned combat aircraft. KTH Royal
Institute of Technology, Stockholm, Sweden. January 2014
[43] M. Zhang , A. Rizzi , P. Meng , R. Nangia , R. Amiree and O. Amoignon.

Aerodynamic Design Considerations and Shape Optimization of Flying Wings
in Transonic Flight. AIAA Paper 2012-5402.
[44] L. Cöllen. Standardized Geometry Formats for Aircraft Conceptual Design

and Physics-based Aerodynamics and Structural Analyses. Royal Institute of
Technology (KTH), German Aerospace Center (DLR). September 2011.
164
BIBLIOGRAPHY
Websites consulted
[45] XFLR5 website: http://www.xflr5.com
[46] CEASIOM Homepage: Computerised Environment for Aircraft Synthesis

and Integrated Optimisation Methods. Lausanne, Switzerland: CFS Engi-
neering 2010. URL: http://www.ceasiom.com/
[47] SUMO website: http://www.larosterna.com
[48] TetGen website: http://www.tetgen.org
[49] DATCOM website: http://www.pdas.com/datcom.html
[50] TORNADO website: http://www.redhammer.se/tornado/
[51] Edge website: http://www.foi.se/edge
[52] NeoCASS website: https://www.neocass.org/
[53] XML Schema: http://www.w3.org/TR/xmlschema-1/
[54] CPACS website: https://code.google.com/p/cpacs
[55] CPACS website: http://software.dlr.de/p/cpacs/home
[56] Aviation week magazine: http://www.aviationweek.com
[57] Wikipedia website: http://en.wikipedia.org (accessed January 2014)
[58] TU Delft Colloquium D.R. Dielbandhoesing.

http://www.lr.tudelft.nl/nl/actueel/laatste-nieuws/artikel/detail/colloquium-
dr-dielbandhoesing/
[59] Aero-Astro Magazine Highlight website: http://web.mit.edu/aeroastro/news/magazine/aeroast

no4/silentaircraft.html.
MIT Department of Aeronautics and Astronautics.
[60] The ’Silent’ Aircraft Initiative website: http://silentaircraft.org
[61] NASA open source parametric geometry VSP: http://www.openvsp.org/
[62] Airfoil library website: http://www.airfoildb.com/
165
BIBLIOGRAPHY
[63] XFOIL website: http://web.mit.edu/drela/Public/web/xfoil/
[64] NACA 4 digit airfoil generator website: http://airfoiltools.com/
[65] Airbus website: http://www.airbus.com/
166
Appendix: Final Sketches of the
BWB
167
D
A
4
4
1547.76
1511.44
291.4
872.83
732.68
581.22
220
3
3
5 1
6 3.
993.4
719.38
537.26
2
2
05
39.
179.29
179.29
_
1
SIZE
1
A4 RM23: BWB TOP VIEW D 05/2014

SCALE WEIGHT (kg) DESIGNED BY: SHEET C _
1:15 22.5 ROBERTO MERINO MARTÍNEZ 1/4 B _
This drawing is our property; it can't be reproduced or communicated without our written agreement. A _
D A
D
A
4
4
1733.92
3
326.24
3
460.7 88
7
10
253.32
1609.03
2
2
1
SIZE
1
A4 RM23: BWB SIDE VIEW D 05/2014

SCALE WEIGHT (kg) DESIGNED BY SHEET C _
D A
D
A
4
4
43
3
3
150.
80
2
2
1
SIZE
1
A4 RM23: BWB FRONT VIEW D _
SCALE WEIGHT (kg) DESIGNED BY SHEET C _
1:15 149,71 ROBERTO MERINO MARTÍNEZ 3/4 B _
D A
D
A
4
4
129
119
151
3
3
140
92
69
74 2
2
639
49
_
1
SIZE
1
A4 RM23: BWB CONTROL SURFACES D 05/2014

SCALE WEIGHT (kg) DESIGNED BY: SHEET C _
D A

Design and Analysis of The Control and Stability of A Blended Wing Body Aircraft (PDFDrive)

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Design and Analysis of The Control and Stability of A Blended Wing Body Aircraft (PDFDrive)

Uploaded by

Copyright:

Available Formats

Design and Analysis of the Control

and Stability of a Blended Wing

Roberto Merino Martínez

Royal Institute of Technology (KTH)

Email of the author: robertomerinomartinez@gmail.com

2 Aircraft conceptual design 9

3 The Blended Wing Body concept 51

3.2.2 Northrop YB-49 . . . . . . . . . . . . . . . . . . . . . . . . 58

4 Definition of the baseline BWB geometry 67

5 Generation of aerodynamic data 91

6 Analysis of the stability and control of the BWB model 125

6.1.1.2 Dynamic stability . . . . . . . . . . . . . . . . . . 128

2.1 The different stages in the creation process of an aircraft. . . . . . 10

2.15 Adaptable fidelity options within the AMB-CFD module in CEA-

4.12 Wortmann fx 60-126 airfoil . . . . . . . . . . . . . . . . . . . . . . 74

5.1 Example of the geometry panelling in TORNADO for the TCR

5.19 2D model for the BWB geometry in TORNADO, including the

6.1 Statically stable behaviour. . . . . . . . . . . . . . . . . . . . . . . 126

2.1 Aerodynamic database, output of the AMB-CFD module. . . . . . 33

3.1 Horten Ho 229 characteristics. . . . . . . . . . . . . . . . . . . . . 58

4.1 NACA 0012 data . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

5.1 Aerodynamic coefficients for the BWB baseline and α = 2º and

Symbol Unit Description

Symbol Unit Description

Symbol Unit Description

Symbol Unit Description

AcBuilder Aircraft Builder (Module of CEASIOM)

AIC Aerodynamic Influence Coefficients

AMB Aerodynamic Model Builder (Module of CEASIOM)

ASCII American Standard Code for Information Interchange

BWB Blended Wing Body

CAD Computer-Aided Design

CATIA Computer Aided Three-dimensional Interactive Application

CDE Computational Design Engine

CEASIOM Computerised Environment for Aircraft Synthesis and Integrated Optimisation

CFD Computational Fluid Dynamics

CORDIS Community Research and Development Information Service

CPACS Common Parametric Aircraft Configuration Scheme

DATCOM Data Compendium

DLM Doublet Lattice Method

DOC Direct Operational Cost

EULA End-User License Agreement

FAR Federal Aviation Regulations

FCS Flight Control System (see FCSDT module from CEASIOM)

FOI Totalförsvarets forskningsinstitut (Swedish Defence Research Agency)

GPL GNU General Public License

GUI Graphical User Interface

ICAO International Civil Aviation Organization

IGES Initial Graphics Exchange Specification

ISO International Organization for Standardization

JAR Joint Aviation Requirements

KTH Kungliga Tekniska Högskolan (Royal Institute of Technology)

LLT Lifting Line Theory

LQR Linear-Quadratic Regulator

MAC Mean Aerodynamic Chord

MATLAB Matrix Laboratory

MEW Manufacturer’s Empty Weight

MIT Massachusetts Institute of Technology

MLS Moving Least Squares

MOB Multidisciplinary Design and Optimisation for Blended Wing Body

MTOW Maximum Takeoff Weight

NACA National Advisory Committee for Aeronautics