Original Article

Proc IMechE Part G:

J Aerospace Engineering
Conceptual design and analysis 2014, Vol. 228(13) 2452–2474
! IMechE 2014

of blended-wing-body aircraft* Reprints and permissions:

sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0954410013518696
uk.sagepub.com/jaero

Jorrit van Dommelen and Roelof Vos

Abstract
Due to the unconventional nature of the blended wing body (BWB) no off-the-shelf software package exists for its
conceptual design. This study details a first step towards the implementation of traditional and BWB-specific design and
analysis methods into a software tool to enable preliminary sizing of a BWB. The tool is able to generate and analyze
different BWB configurations on a conceptual level. This paper investigates three different BWB configurations. The first
configuration is an aft-swept BWB with aft-mounted engines, the second configuration is an aft-swept BWB with wing-
mounted engines and the third configuration is a forward-swept BWB with wing-mounted engines. These aircraft comply
with the same set of top-level requirements and airworthiness requirements. Each of the designs has been optimized for
maximum harmonic range, while keeping its maximum take-off weight constant and identical. Results show that the
forward-swept configuration with wing-mounted engines has the highest harmonic range. These findings warrant further
investigation in this configuration and other alternative BWB configurations.

Keywords
Blended-wing-body aircraft, conceptual design, Multi-disciplinary design optimization

Date received: 12 November 2012; accepted: 4 December 2013

Introduction Program for Aircraft Synthesis Studies (PASS) by

Since the introduction of the tube-and-wing (TAW) Desktop Aeronautics). However, due to the uncon-
jetliners in the late 1950s, their range parameter, ventional nature of the BWB there exists no such
Mcruise  L=D, has been steadily increasing from 13 tool for their conceptual design based on a given set
in 1960 to 16 in the mid 1990s (see Figure 1).1 of top level requirements. This paper details a first
Because designers of TAW aircraft can rely on realis- step towards the implementation of traditional and
tic solutions from decades of research into this con- BWB-specific design methods into a software tool
figuration, it is unlikely that significant gains can be that aids the designer in the conceptual design of a
achieved unless radically new technologies are BWB aircraft.
employed. There are various alternative concepts pro- In this paper, three designs are presented that are
posed to the TAW configuration. One of those con- generated using this new design tool. The configur-
cepts is the blended-wing-body (BWB) configuration, ation between each of the designs is completely differ-
which consistently promises to increase the aero- ent. However, each of these aircraft has to carry the
dynamic efficiency by another 25%2,3 and is the only same payload of 400 passengers and 22 metric tons of
nonconventional concept that airline manufacturers freight. The cruise Mach number is set to 0.82 and the
have taken an interest in (notably Boeing’s 8.5% landing and take-off runway lengths should be less
scale BWB demonstrator). Although, on a first than 2500 m. Furthermore, Federal Aviation
glance, the 25% increase might not seem as revolu-
tionary, it does when one realizes that it took more
Faculty of Aerospace Engineering, Delft University of Technology, Delft,
than 45 years to achieve this improvement on a con- The Netherlands
ventional TAW jetliner. *This paper has been modified from ‘‘A Conceptual Design and Analysis
For TAW aircraft, there are sufficient airplane Tool for Blended Wing Body Aircraft,’’ 29th International Congress of
design handbook methods available to perform con- the Aeronautical Sciences, Brisbane, Australia, 23–26 September 2012.
ceptual design studies. Several off-the-shelf software
Corresponding author:
packages can be used that have these methods inte- Roelof Vos, Faculty of Aerospace Engineering, Delft University of
grated in a user-friendly computer code (e.g. Technology, Kluyverweg 1, 2629 HS Delft, The Netherlands.
Advanced Aircraft Analysis (AAA) by DARcorp or Email: [email protected]

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

van Dommelen and Vos 2453

Regulations (FAR)/Certification Specification (CS) maximum take-off weight (MTOW). The range
25 certification requirements apply to the climb gra- should be at least 11,000 km.
dients, design load factors, and safety margins. These In the subsequent sections the design, analysis, and
requirements are loosely based on the performance optimization strategy are presented that are used to
characteristics of a Boeing-777 aircraft. The top- generate the conceptual designs. Design and analysis
level requirements that are used for the present methods are presented in detail in the following sec-
study are listed in Table 1 and the mission segments tion. The three optimized configuration designs and
are shown in Figure 2. As a test case, it is investigated their respective characteristics are subsequently pre-
which of these three configurations can achieve the sented in the last sections. The characteristics of
highest range at maximum payload for a given each of the aircraft are discussed and intuitive explan-
ations are presented for the differences between the
three designs.

Structure of the design and

analysis program
All calculations are performed from the Matlab envir-
onment, using customary written calculation modules,
centered around a main module. This paper summar-
izes the design and analysis modules and discusses
exemplary results. The program automatically designs
a single BWB aircraft for a given set of input param-
eters and analyzes several key characteristics. In
Figure 1. Aerodynamic efficiency – (M0  L=D) variation with Figure 3 an overview of the program structure is
date of entry into service.1 presented.
In Figure 4 an N2 chart is presented that shows the
dependencies of the modules within the program struc-
ture. On the diagonal of the N2 chart the modules of
Table 1. Top level requirements.
the program are indicated. The blocks above the diag-
Symbol Requirement Value Unit onal show which variables are fed forward during the
design process, while blocks below the diagonal show
Npax Number of pax 400 –
which variables are fed back during the design process.
Necon Number of pax in economy class 350 – The bold lines in this diagram mark the border
N1st Number of pax in 1st/business 50 – between preliminary sizing (top left) and the design
Wfreight Freight weight 22,000 kg optimization (bottom right). It can be observed that
Rcruise Cruise range 411, 000 km there is no feedback between the output of the opti-
Ralt Alternative destination range 500 km mization and the preliminary sizing. In the present
Mcruise Cruise Mach number 0.82 – incarnation the maximum take-off weight (WTO ) and
hcruise Cruise altitude 11,000 m take-off thrust, (TTO ) are held constant during the
sTO Maximum take-off distance 2500 m optimization process and the range at maximum pay-
sld Maximum landing distance 2500 m load (also known as harmonic range) is maximized.
When the program is started, the configuration of
Elr Loiter time 45 min
the aircraft has to be defined by the user. The input

Figure 2. Mission profile, with flight phase definitions as used in the calculations.

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

2454 Proc IMechE Part G: J Aerospace Engineering 228(13)

consists of four vectors. The first vector (denoted as x0 The third option is the winglet option, which means
in Figure 4) describes the shape of the planform and that the vertical tail planes are fixed at the wing tips of
the vertical tail. In other words, the user defines a the aircraft. The final option is to use no vertical tail
preliminary planform shape and fin height. The at all.
second vector is the airfoil vector, which defines the The first part of the program is the preliminary
airfoils used at various spanwise locations as defined sizing (section ‘‘Preliminary sizing module’’). Here
in streamwise direction. The third vector defines the the MTOW, wing loading, and thrust-to-weight
number of engines (Neng ) and their location. There are ratio are calculated based on the method laid out in
two options for the engine location: body-mounted Roskam.4 After the preliminary sizing the program
engines and wing-mounted engines. The final vector constructs discipline-specific models of the aircraft in
describes the vertical tail configuration. There are four the multi-model generator (MMG) module of the pro-
options for the vertical tail type. The first option is a gram as is presented in section ‘‘Multi-model gener-
single body-mounted vertical tail. The second option ator module’’. The models are subsequently used to
is a twin tail at the trailing edge of the centerbody. analyze the aircraft as is described in section
‘‘Analysis module’’. The ‘‘Optimization module’’ sec-
tion compares the analysis results to a set of con-
straints and calculates the value of the objective
function. As the take-off weight is kept constant
during the optimization process, empty weight is
essentially traded for fuel weight to maximize the
range. This approach means that no iterations in
design weight are required, which reduces the opti-
mization time significantly.

Preliminary sizing module

Preliminary sizing of thrust and wing loading as well
as a Class I weight estimation are carried out accord-
ing to traditional handbook methods.5 Estimates of
BWB aircraft from previous studies2,3,6 are taken into
account along with conventional aircraft data where
necessary. For a given set of range requirements and
assumptions on aerodynamic and engine perform-
ance, this module calculates the take-off weight, fuel
weight, and operational empty weight of the airplane.
With appropriate choices, based on prior research
into BWB aircraft, for the design maximum lift coef-
ficients and wing aspect ratio, the design point in the
Figure 3. Single aircraft calculation structure and blocks. thrust-to-weight ratio versus wing loading diagram is
MMG: multi-model generator; BWB: blended wing body. calculated. The design point determines the take-off

Figure 4. N2 chart of the program structure.

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

van Dommelen and Vos 2455

Figure 5. Flow diagram of preliminary sizing module.

OEW: operational empty weight; MTOW: maximum take-off weight; SFC: specific fuel consumption.

The definition of the flight phases is given by the mis-

Table 2. Input parameters for Class I weight estimation. sion profile, displayed in Figure 2. The fuel fraction
for the three main fuel consuming phases: cruise,
Symbol Requirement Value Unit
alternative destination cruise and loiter, are calculated
Nfd Flight deck crew 2 – using Breguet’s range and endurance equation
Ncrewecon Economy pax per cabin crew 30 –  
Ncrew1st 1st/business pax per cabin crew 18 – V L W4
Rcruise ¼ ln ð3Þ
Wpax Person weight 86 kg cj g D W5
Wlug Luggage weight per person 25 kg  
1 L W6
Elr ¼ ln ð4Þ
cj g D W7
 
thrust and the wing area. Through the aspect ratio, V L W8
the wing span is also fixed. The input vector describ- Rcruise alt ¼ ln ð5Þ
cj g D W9
ing the planform of the aircraft is scaled to match the
wing area and wing span estimations of the prelimin- The specific fuel consumption (SFC), cj is estimated
ary sizing. This approach is used to increase the prob- from already existing engines and is taken equal to
ability to arrive at a feasible aircraft design at the end 17.1  106 N/N/s. This value is based on the values
of the conceptual design phase. The process is sche- provided in Roskam4 and Raymer.7 The aerodynamic
matically shown in Figure 5. efficiency, expressed in terms of lift-to-drag ratio (L/D)
is estimated from literature. ðL=DÞmax of a BWB
Class I weight estimation. In Class I weight estimation, a according to Liebeck2 can be taken equal to 27. The
first estimate of the airplane weight is made based on cruise lift-to-drag ratio is lower. Raymer7 suggests
the requirements of Table 1. For the weight estima- using a cruise lift-to-drag ratio of 0:866ðL=DÞmax . The
tion, additional information is needed to complement fuel fractions for the other phases are small relative to
the requirements, this information is given in Table 2 the cruise and loiter phases and therefore are assumed
and forms part of the input parameters. fixed. Values from Roskam4 are used and are listed in
The payload weight is calculated using Table 3. The overall fuel fraction is found by multiply-
ing all fractions of the individual flight phases
Wpl ¼ Wfreight þ ðWpax þ Wlug ÞNpax ð1Þ
m 
Y 
Wnþ1
The crew weight is given in equation (2). The crew Mff ¼ ð6Þ
n¼1
Wn
weight is assumed to be part of the operational
empty weight (OEW).
The MTOW is determined by the following
Wcrew ¼ ðWpax þ Wlug Þ  ðNfd þ Ncabin Þ ð2Þ equation

The fuel weight is estimated using fuel fractions. For WTO ¼ Wpl þ WOE þ ð1  Mff ÞWTO ð7Þ
each flight phase a fuel fraction is defined.

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

2456 Proc IMechE Part G: J Aerospace Engineering 228(13)

The operational empty weight can only be determined Sizing of wing surface area and engine thrust. Sizing of
by looking at statistical data, since no information wing area and engine thrust is performed by generat-
about the actual aircraft is known. The relation ing a wing loading versus thrust-to-weight ratio dia-
between the operational empty weight and MTOW gram. The drag polar of the aircraft is estimated using
is determined from existing aircraft and other BWB the method described by Roskam.4 The wing loading
research programs versus thrust-to-weight ratio diagram is shown in
Figure 7. The various lines visualize the imposed
WOE ¼ aTO þ b ð8Þ constraints.
Several assumptions have been made. The CLmax
The coefficients a and b are determined from the stat- value in clean and take-off/landing conditions are
istical data extracted from Figure 6. In this graph, set at 1.2 and 1.4, respectively. These values are sup-
several long-range TAW aircraft are plotted along ported by Bolsunovsky et al.10 The aspect ratio has a
with results from the open literature for BWB aircraft. value of 4. Table 4 lists the assumptions made in the
Chauvenet’s criterion is applied to check the reliability preliminary sizing.
of the reference aircraft weight. The mean value of the The optimum point in the W/S–T/W diagram is
operational empty weight over MTOW is 0.5226, with found by moving to the lower right of the diagram,
a standard deviation of 0.061. The dotted lines in without exceeding the constraints imposed by the dif-
Figure 6 indicate the boundaries of the Chauvenet’s ferent requirements. An optimizer searches for the
criterion. The MOB BWB does not meet this criter- optimum point, which might be a dependent on a
ion, and is therefore disqualified. A linear relation is weighing factor between wing loading and thrust-to-
obtained by fitting a curve to the remaining entries. weight ratio. The chosen design point is indicated by a
The MTOW is determined by combining equation (7) star in Figure 7. In this example, the one-engine-
with equation (8). inoperative climb-gradient requirement sizes the
thrust-to-weight ratio and the landing-distance
requirement determines the wing loading.

Table 3. Fixed weight fractions from Roskam.4

Multi-model generator module
Symbol Phase Fuel fraction
An MMG uses the output from the preliminary sizing
M10 Start up 0.990 along with specified user input on the wing geometry,
M21 Taxi 0.995 the selected airfoils, the engine configuration and the
M32 Take-off 0.980 vertical tail configuration to generate the following
M43 Climb 0.980 models:
M65 Descent 0.990
. geometric model of the outer shell including
M87 Landing, taxi, shut-down 0.995
disposition of vertical tail and engines

Figure 6. Take-off weight versus operational empty weight. Conventional aircraft from Roskam4 and Jane’s All the World’s Aircraft,8
SAI data from Diedrich et al.,9 TsAGI data from Bolsunovsky et al.,10 MOB data from Laban et al.,11 and BWB-450 data from Liebeck.2

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

van Dommelen and Vos 2457

Figure 7. Wing loading versus thrust-to-weight ratio diagram.

CS: Certification Specification; MTOW: maximum take-off weight; MLW: maximum landing weight.

Table 4. Assumptions for the preliminary sizing of wing area and engine thrust.

Parameter Symbol Value Unit

10
Maximum lift coefficient clean CLmax ,clean 1.2 –
Maximum lift coefficient take-off10 CLmax ,TO 1.4 –
Maximum lift coefficient landing10 CLmax ,ld 1.4 –
Stall speed clean Vs1 80 m/s
Stall speed take-off configuration Vs0 70 m/s
Stall speed landing configuration Vs0 70 m/s
Friction coefficient Cf 0.0020 –
Zero-lift drag increase take-off slats CDtof 0.010 –
Zero-lift drag increase landing slats CDldf 0.010 –
Zero-lift drag increase landing gear CDgear 0.020 –
Zero-lift drag increase wave drag CDwave 0.006 –
Oswald factor clean eclean 0.85 –
Oswald factor take-off configuration eTO 0.80 –
Oswald factor landing configuration eld 0.80 –
Aspect ratio A 4 –
Engine thrust temperature correction – 0.80 –
Maximum continuous thrust ratio – 0.94 –
Trapped fuel and oil fraction Mtfo 0.002 –
Maximum landing weight MLW 0.84 MTOW kg
Maximum lift-to-drag coefficient2 ðL=DÞmax 27 –
Cruise lift-to-drag coefficient7 ðL=DÞcruise 0:866ðL=DÞmax –
Thrust lapse with altitude – Raymer Fig. 5.1 –
MTOW: maximum take-off weight.

. model of the interior volume: passenger cabin, subsequent paragraphs. Figure 8 schematically
cargo space and fuel tank shows the workings of the MMG.
. structural disposition of the wing box The initial planform shape and surface area of the
. vortex lattice model (VLM) of the wing planform BWB is captured by the input vector, x0 . This vector
is linearly scaled to match the planform required wing
This data is used in the analysis modules of the pro- area (S). This ensures that the wing is sized for the
gram. Each of these is elaborated shortly in the maximum allowed wing loading. The take-off thrust is

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

2458 Proc IMechE Part G: J Aerospace Engineering 228(13)

Figure 8. Graphical overview of the MMG structure.

MMG: multi-model generator; VLM: Vortex lattice model.

Table 5. Wing-body input parameters.

Section 1 2 3 4 5 6 Unit

Chord c1 c2 c3 c4 c5 c6 m
t/c t1 =c1 t2 =c2 t3 =c3 t4 =c4 t5 =c5 t6 =c6 –

Twist 5 5

Sweep 1 2 3 4 5

Dihedral 1 2 3 4 5
Span b1 b2 b3 b4 b5 m
Tail height h m

kept constant during the optimization process and The airfoil at each trunk end is defined in the input
equals the value that is determined during preliminary vector. By interpolating the airfoils, the aerodynamic
sizing. surface of the BWB is formed.
The definition of the vertical tail is similar to the
Geometric model of the outer shell. The geometric model wing-body definition. Two additional input param-
of the outer shell consists of three parts: the aero- eters are needed to fix the location of the vertical
dynamic surface of the wing-body combination, the tail. These are the y-position with respect to the cen-
vertical tail surfaces and the engines. The MMG starts terline and the distance from the leading edge. The tail
by generating the aerodynamic surface of the wing- is automatically fixed to the body in z-direction by the
body combination. This surface is formed by reading MMG. Only a single trunk is needed to describe the
the scaled input vector. The wing-body surface is vertical tail. Not all vertical tail input parameters are
divided in 10 wing trunks resulting in 6 sections per used for certain tail configurations. An example is a
wing half. The planform shape of each wing half and winglet configuration. In that case, the tail root chord
the vertical tail height is described by the 30 param- is set equal to the wing tip chord.
eters displayed in Table 5. The airfoils used for the wing-body are based on
The planform is formed by defining the 2D plan- the Whitcomb supercritical airfoil. The camber is
form and subsequently adding dihedral and twist. The removed, while the thickness distribution is preserved.
aerodynamic surface is formed by a lofting process. The thickness is controlled by the input vector.

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

van Dommelen and Vos 2459

The vertical tail airfoils are symmetric NACA 4-series The fuel tanks are located in the outer trunks of
profiles with a thickness of 14% at the root and 12% the wing. In this study the fuel tanks are assigned to
at the tip. the two outboard wing trunks. The fuel tanks are
The final part of the outer shell is formed by the located inside the torque box and extend to 85%
engines. Two options for the engine positions are con- of the semi-wing span. The fuel tank and cabin
sidered. The engines can be placed at the aft-body or model inside the aerodynamic shell can be seen in
beneath the wings. The position and number of Figure 9(b). The cabin and fuel tanks are not sized
engines is described by the engine input vector. The for the required number of passengers and fuel
MMG finds the exact location of the engine by con- volume, respectively. It is therefore possible that
sidering the aerodynamic surface of the wing-body they are not large enough to accommodate the
and the location of the vertical tail. required number of passengers or fuel, respectively.
However, during the optimization process, bounds
Cabin and fuel volume. The internal model consists of are set on the required floor area per passenger.
the pressure cabin and the fuel tanks. The pressure If these bounds are surpassed, the design variables
cabin is formed by a separate structure, which has to need to be altered such that this constraint is satis-
withstand pressurization loads. Liebeck2 proposed fied. For the fuel tank it is checked if the minimum
two concepts for the pressurized cabin for BWB, one design range (R0) can be flown.
with an integrated skin and pressure shell. This
requires a thick sandwich structure for both the Structural model of the wing-body. The third model cre-
upper and lower wing surfaces, which sustain aero- ated by the MMG is a structural model of the wing-
dynamic and pressurization loads. A second approach body combination. This model forms the basic shape
is a separate pressurization structure, which carries of the wing box, which carries the loads through the
only pressurization loads.12 This pressure vessel is structure. The front and rear spar are assumed to be
thin-walled and is loaded solely in tension. The latter located at fixed chord percentages at 13% and 72%,
approach is used in this study. This structure takes the respectively (see Figure 10(a)). This model is used for
form of a multi-bubble structure. By using a separate the Class II weight estimation. The weight of the
structure, the functions of the aerodynamic structure wing-body structure is determined by considering
and the cabin are separated. The MMG determines the the wing box as the load-carrying structure which is
geometry of the multi-bubble pressure cabin. The statically determined. To accommodate for the cabin,
structure consists of straight cylinders, which are cutouts are made in the spars of wing box as can be
joined together. An example can be seen in Figure 9(a). seen in Figure 10(b).

Figure 9. Internal and external geometric models generated by MMG: (a) multi-bubble cabin structure and (b) internal volume
within external aerodynamic shell.

(a) (b)

Figure 10. Wing box generated by MMG: (a) full wing box and (b) wing box with cutouts for cabin..

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

2460 Proc IMechE Part G: J Aerospace Engineering 228(13)

Figure 11. Lattice model with collocation points as generated by MMG.

Engine model. Based on the required thrust (TTO ), the and strut lengths are calculated based on the CoG
size of the engine is determined. The engine data used range, clearance and loading requirements.
are based on the Trent 900 engine used on the A380- Combining the CoG and aerodynamic information,
800. The dimensions, weight and specific fuel con- the trim deflection in cruise flight is estimated along
sumption are scaled according to the rules given by with estimates for the trim drag. In addition, estimates
Raymer.7 The specific fuel consumption, which is are made for the static longitudinal, directional and
expressed in kg fuel used per N thrust per second, is lateral stability, take-off rotation capability and
assumed constant during cruise flight. The engine one engine inoperative (OEI) controllability.
location is specified by the input vector, containing Subsequently, the performance module calculates the
a description of the engine placement per engine. payload-range diagram and estimates the take-off
The length of the vector determines the number of field length, the landing length, and the attainable
engines. The engines can be placed on the aft-body climb gradients for various engine operating/
or beneath the wings. The exact location of the nonoperating conditions. The analysis module with
engines is determined by the MMG which tries to its submodules is visualized in Figure 12 and discussed
find an optimum location. If the engines are pos- in detail below.
itioned on the aft-fuselage the vertical tail position is
taken into account as a positioning constraint. Payload analysis. The payload module analyzes the
pressure cabin generated by the multi-model genera-
Lattice model. The MMG generates a program-specific tor. The required amount of passenger cabin floor
input for a vortex lattice program (Tornado13). The area and the required cargo volume are determined
aircraft consist of the wing-body and the vertical tail from the top-level requirements. The payload
surfaces and is modeled as a three-dimensional lattice module determines the location of the passenger
with zero thickness. This model is used for the calcu- cabin floor (Figure 13).
lation of the lift and drag polar in cruise flight along Subsequently, the payload module calculates the
with stability derivatives (Figure 11). available floor area and cargo volume. The vertical
dimension of 1.63 m is considered to calculate the
available cargo volume, to give the possibility to
Analysis module
accommodate the most common unit load devices
After the full disciplinary models of the aircraft have (ULDs) and pallets.
been generated, a set of submodules analyzes key
characteristics of the airplane. The first submodule Aerodynamic analysis. The second analysis module is the
is the payload module, which uses the cabin layout aerodynamic module. The aerodynamic module uses
to analyze the volume and area available for the pas- the vortex lattice method to determine the lift polar
senger cabin and the cargo bays. The second submo- for cruise conditions. The determination of the
dule performs an aerodynamic analysis based on a induced drag is done using a Trefftz plane analysis.
vortex lattice method. When the aerodynamic loads For the determination of the zero-lift drag, the
are known, a Class II weight estimation is performed method of Raymer is used.7 This method uses the
and the center of gravity (CoG) travel during loading flat-plate friction drag and form factors to estimate
and flight is determined. The landing gear position the zero-lift drag.

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

van Dommelen and Vos 2461

Figure 12. Analysis block flow diagram.

CoG: center of gravity.

Figure 13. Cross section of the cabin, showing passenger and cargo compartments.

Secondly, the aerodynamic module estimates the The relation between quarter chord sweep, local
maximum lift coefficient. First, the maximum lift coef- thickness and lift coefficient from Kroo15 is used to
ficient in clean configuration is determined. This is estimate the critical Mach number. The drag diver-
done by taking the sections’ maximum lift coefficient gence Mach number is estimated by adding 0.05 to
into account. These values are determined from the critical Mach number.
experimental data.14 The spanwise lift distribution is
interpolated until the sections’ maximum lift coeffi- Weight analysis. With the aerodynamic parameters esti-
cient is reached. The corresponding lift coefficient is mated, a second, more detailed weight estimation is
considered to be the maximum lift coefficient of the performed. This weight estimation is based on the
aircraft at its stall speed, Vs1 . The effect of the add- actual dimensions of the aircraft, whereas the Class I
ition of slats, which are assumed to be the only high- weight estimation was solely based on statistical and
lift devices present, is determined from handbook empirical relations. In this section a so-called Class II
methods. Since only slats are present, the maximum weight estimation is developed to estimate the compo-
lift coefficient in take-off and landing configuration is nent weights of the airframe. The flow chart of the Class
assumed to be identical. Therefore, the stall speed in II weight estimation is shown in Figure 14. Because the
both landing and take-off configuration is the same fuel weight is both input and output for the weight cal-
and indicated by Vs0 . culations a feedback loop is present. This weight iter-
The final part of the aerodynamic module is to ation stops after the operational empty weight does not
estimate the critical Mach number and the drag diver- change more than 1% between two iterations.
gence Mach number (MDD ). These values are used to Engine weight and auxiliary component weight is
conclude if the design is considered feasible. calculated using the empirical methods of Raymer.7

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

2462 Proc IMechE Part G: J Aerospace Engineering 228(13)

Figure 14. Flow chart for Class II weight estimation.

This includes the weight of the vertical tail and the (u , l , mat , mat ) and material density (mat ) is
weight of the aircraft systems. The use of this method required as input. These and other empirical factors
is believed to be appropriate since the vertical tail and are listed in Table 6.
aircraft systems do not differ fundamentally from To size the wing structure, the maximum load con-
those used on conventional aircraft. The weights of dition is determined by investigating two cases: (1)
the components estimated using this method are given maximum maneuver load factor (nlimit ) at MTOW,
in equations (15.25) to (15.44) in Raymer,7 except for (2) maximum gust load factor (ngust ) at maximum
the fuselage weight and the wing weight, which are zero-fuel weight. In both cases a safety factor of 1.5
calculated below. is added. For these two load cases the bending
Traditionally, similar empirical relations between moment at any point in the wing is calculated as
the aircraft dimensions and component weights are follows
used in the Class II weight estimation of structural  
components. However, due to the unconventional MB ¼ n MBL þ MBf þ MBeng ð9Þ
nature of the BWB a physics-based method has been
employed to calculate the weight of the wing and pres- where MBL is the 1  g bending moment due to lift,
sure cabin. For reasons of simplicity, the wing and MBf is the 1  g bending moment due to fuel, and
cabin are assumed to be completely independent MBeng is the 1  g bending moment due to the engines.
structures. They are therefore analyzed separately These bending moments are calculated from the lift
and their weights are simply added. distribution (cl c), the fuel tank location (yft ), and the
The primary structural component of the wing is engine disposition (yeng ). The latter two components
the box structure of Figure 10(a). The weight is esti- contribute to bending moment relieving. In case the
mated using the method of Torenbeek.16 This method gust load is investigated, MBf ¼ 0. Whichever of the
assumes that all loads are ultimately concentrated in two load cases yields the highest bending moments is
the primary wing box structure, consisting of upper the one that sizes the thickness of the wing panels and
and lower stiffened panels, a front and rear spar, and spar webs.
ribs. The method assumes that the wing box is static- To estimate the weight of the secondary wing-body
ally determined, where bending forces are absorbed structure, the empirical relations from Torenbeek16
by the stiffened skin panels and shear forces are trans- are used. The MTOW is input in combination with
ferred by the spar webs. Torsional loads are not taken empirical factors. The complete input is found in
explicitly into account. Maneuvering and gust loads Table 6.
are considered as the determining load cases. The pressure cabin forms a separate structure,
Correction factors are applied to the idealized struc- which is assumed to absorb the load caused by the
ture to account for nontapered skins, splices, joints pressure difference only. A multi-bubble structure is
and access panels. The wing box weight is assumed that is manufactured from composite mater-
determined per wing trunk, using the structural ial. When the material is in a state of uniform equal
model of the wing-body as generated by the MMG biaxial extension, the weight is independent of the
(Figure 10(a)). Input on material allowables actual geometry.17 In this case, the structural weight

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

van Dommelen and Vos 2463

Table 6. Values for weight calculations. Table 7. Assumed location of weight components.

Parameter Symbol Value Unit Component Center of gravity location

Wing primary structure Wing box Defined per trunk, dependent on

Ultimate load factor nult 3.75 – geometry and material
Design vertical gust load Ude 12.5 m/s Wing LEs Defined per trunk, halfway between LE
Material density mat 2855 kg/m3 and front spar
(Al 2024 T3) Wing TEs Defined per trunk, thirdway between
Maximum normal stress u 400 MPa aft spar and TE
upper panels Cabin components Dependent on the volume of the cabin
Maximum normal stress l 350 MPa Vertical tail 40% of MAC of the fin
lower panels Landing gear At calculated positions, halfway
Mean normal stress mat 317 MPa between length
Mean shear stress mat 132 MPa Engine group Halfway between engine position
Rib mass factor kr 0.5  103 – Fuel systems At fuel tank center of gravity, volume
Reference thickness tref 1.0 m weighted
Non-optimality correction NO 1.0  103 m APU 90% chord of first wing section
Instruments 2 m from aircraft’s nose
Wing secondary structure Hydraulics Defined per trunk, at wing’s rear spar
Fixed leading edge factor kfle 1.0 – Electrics Center of the wing box
Flap type correction  0.0 – Avionics 2 m from aircraft’s nose
(single slotted)
Anti-ice systems At wing’s outer three sections in the LE
Trailing edge flap factor ktef 0.0 –
Flight deck crew 3.5 m from aircraft’s nose
(no flap)
Cabin crew At cabin’s center of gravity
Trailing edge flap factor ktef 1.0 –
(plain flap) Trapped fuel/oil At fuel tank’s center of gravity
LE: leading edge; TE: trailing edge; MAC: mean aerodynamic chord;
Pressure cabin APU: auxiliary power unit.
Design pressure difference p 80 kPa
Tensile stress fiber direction 1 1 500 MPa
Tensile stress fiber direction 2 2 250 MPa nonoptimality is used. In addition, a factor for the
Material density mat 1550 kg/m3 presence of doors and windows is applied as well as
Factor for door cut-outs Kdoor 1.25 – a factor for landing gear cutouts. The values of these
Factor for landing KLg 1.12 –
factors are taken from Raymer7 and can be found in
gear cut-out Table 6. The weight of the pressurized cabin is found
according to
mat
Wpress ¼ 6Kdoor KLg pVP qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð11Þ
required for the pressure vessel is independent of the 12 þ 22
shape and can be determined as follows
The weight estimation is completed by including the
mat payload weight and the fuel fraction data from the
Wpressideal ¼ 3pVp qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð10Þ
12 þ 22 Class I weight estimation. The fuel weight is calcu-
lated as follows
Wf ¼ WTO  ðWOE þ Wpl Þ ð12Þ
Here, 1 is the allowable stress is circumferential dir-
ection, 2 the allowable stress in longitudinal direc-
tion, and Vp the pressurized volume. It is assumed Center of gravity analysis. The CoG of the empty aircraft
that the allowable stresses correspond to equal max- does not vary and is dependent on the location of the
imum strain. While this is true for isotropic materials, weight of the components calculated in the previous
it is not necessarily true for anisotropic (composite) section. Using the moment the components create
materials. Equation (10) is, however, always valid about a reference point and the total weight, the
when the maximum strain criterion is used as the fail- CoG can be calculated. The estimated location of
ure criterion for anisotropic materials. The weight each component is given in Table 7.
calculated in equation (10) is the weight of an ideal The cabin components include the pressure vessel
pressure vessel. It is assumed that the tensile stresses weight, cabin furnishings, cabin interior and air con-
between the bubbles are carried by walls. Instead, col- ditioning. The engine group includes the installed
umns in which tensile stresses are concentrated are engines, engine starter, engine controls and nacelles.
present. A factor of 2 to account for this All data is derived from the geometric data produced

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

2464 Proc IMechE Part G: J Aerospace Engineering 228(13)

10

Aircraft z−coordinate (m)

OEW CoG
5 Cabin group Wing group Engine group

Vertical tail
Landing gear group
−5
−5 0 5 10 15 20 25 30 35 40 45
Aircraft x−coordinate (m)

Figure 15. Main groups center of gravity positions.

OEW: operational empty weight; CoG: center of gravity.

be the MMG. The CoG of the main groups are yield the same aft-CoG position, the scenario with the
plotted in Figure 15. The aircraft is assumed to be smallest CoG range is chosen. In the example of
symmetrical around the xz-plane, which means that Figure 16 scenario 1 results in the least aft-position
the y-positions of the centers of gravity are equal of the CoG and also has the smallest CoG range. Also
to zero. note that the position of the aerodynamic center (a.c.)
The CoG at each point in flight depends on the is depicted. This indicates that the aircraft in this par-
amount of payload, the amount of fuel and how the ticular example is longitudinally unstable for certain
fuel is distributed in the tanks during the flight. For a combinations of payload weight and fuel weight.
constant MTOW, various combinations of payload Whether this is acceptable depends on the constraint
weight and fuel weight exist. For each payload that is enforced on the static margin. The latter one is
weight, the fuel weight is distributed over the fuel a user-defined input constraint.
tanks within the two outer most wing trunks. To
investigate the effect of loading scenarios on the Landing gear analysis. Based on the CoG range occur-
CoG travel, four different loading–unloading scen- ring in ground operations, the landing gear position is
arios have been defined: determined. The rules as described by Raymer7 are
used. Based on the clearance margins of the aft-end
1. First fuel the inner tanks completely, then fuel the of the body, the engines and the wing tips, the length
outer tanks until MTOW is reached. During flight, of the landing gear is determined.
first drain the outer tanks, then drain the inner
tanks. Trim and stability module. The trim and stability module
2. First fuel the inner tanks completely, then fuel the starts by determining the trim deflection of the elevator
outer tanks until MTOW is reached. During flight, for each CoG case. The BWB uses control surfaces on
first drain the inner tanks, then drain the outer the trailing edge of the body/wing surface to control
tanks. the pitch. The program assigns control surfaces for
3. First fuel the outer tanks completely, then fuel the pitch control to the four most inboard wing trunks.
inner tanks until MTOW is reached. During flight, The pitch control surfaces are indicated in Figure 17.
first drain the inner tanks, then drain the outer For the control surface sizing, a fixed chord percentage
tanks. of 20% is used. To be trimmed in longitudinally, the
4. First fuel the outer tanks completely, then fuel the lift must equal the drag and the aerodynamic moment
inner tanks until MTOW is reached. During flight, must equal the moment generated by the aircraft’s
first drain the outer tanks, then drain the inner weight. This creates a system of two equations,
tanks. which can be solved to determine the trim deflection
of the pitch control surfaces and the angle of attack.
For a given payload weight, each of these loading The aerodynamic derivatives as determined by the
scenarios is applied while the airplane executes the aerodynamic module are used. Using the angle of
mission of Figure 2. This is repeated for 31 instances attack and the trim deflection, the trim drag is esti-
of the payload weight, ranging from zero payload to mated. The trim drag is found by considering the
maximum payload. The shift in CoG is recorded derivative of CDe at a given . Multiplying this deriva-
during the fueling and the draining of the tanks. An tive by the trim deflection e results in the increase in
exemplary result is plotted in Figure 16. Based on drag coefficient due to the elevator deflection.
these scenarios, bounds are defined on the CoG The static longitudinal stability is determined by
travel. Whichever scenario yields the least aft-CoG calculating the minimum static margin. The static
position is the one that is selected. If two scenarios margin is the difference between the CoG position

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

van Dommelen and Vos 2465

Figure 16. Center of gravity travel during four different loading/unloading scenarios.
OEW: operational empty weight; MTOW: maximum take-off weight; CoG: center of gravity; MZFW: maximum zero-fuel weight.

The directional and lateral stability are taken into

account by considering the weathercock stability and
the rolling moment due to sideslip, respectively. The
directional stability coefficient due to the vertical tail
(Cn ) and the lateral stability coefficient (Cl ) are
obtained from the Vortex lattice (VL) analysis.
Additionally, the directional stability coefficient due
to the body is calculated using the method of
Roskam.14 This method takes into account the pro-
jected side area of the BWB onto the xz-plane as well
as the length from the CoG to the centroid of this
area. Because the BWB often has a larger side area
ahead of the CoG than behind the CoG, the body is
destabilizing. Bounds on Cn and Cl are used as con-
straints in the optimization process.
The one-engine-inoperative trim under the most
unfavorable circumstances is also considered. A
rudder size of 25% of the vertical-tail chord is
Figure 17. Example BWB with pitch control surfaces assumed. The inoperative engine is assumed to cause
assigned. a directional moment (NOEI ) that amounts to 1.25 the
take-off thrust of that engine. This takes into account
the additional drag due to wind milling of the engine.
and the neutral point. The neutral point is found by A minimum control speed of Vmc ¼ 1:2  Vs0 is used
dividing the difference in aerodynamic moment for and the most aft-CoG position is considered. The con-
two different angles of attack, by the difference in trol derivative Cnr about the most aft-CoG position is
lift coefficient at these angles of attack. These obtained from the VL analysis. The required rudder
moments are already known from the aerodynamic deflection is calculated according to
module. The minimum static margin is determined
by considering the distance between the most aft- NOEI
r ¼ ð13Þ
CoG position and the neutral point, expressed in qmc SbðCnr Þcgaft
terms of the mean aerodynamic chord (MAC).

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

2466 Proc IMechE Part G: J Aerospace Engineering 228(13)

The final analysis is the calculation of the minimum included in the fuel weight indicated in the diagram.
rotation speed at the most forward CoG position It should be noted that the diversion range (see the
(xfwd cg ) at the MTOW. Here it is assumed that the mission profile of Figure 2) is not included in the
pitch control surfaces are uniformly deflected evaluation of the range in Figure 18.
upwards over an angle of 15 . The take-off rotation The take-off performance is measured by the
speed is calculated by considering the moment equi- balanced field length (BFL). With the maximum lift
librium about the longitudinal position of the main- coefficient known in clean and take-off configuration,
landing-gear (xmlg ) the BFL can be determined with the method sug-
gested in Raymer.7 This method develops a relation
Vminrot based on the fact that the take-off can be divided in
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi three segments: the ground roll, rotation, and climb to
2WTO xfwd cg  xmlg obstacle height. The BFL can be computed as follows
¼
Sc Cmx¼0 ð ¼ 0,  ¼ 15 Þ  CL ð ¼ 0Þ xmlg c
 
þ 5½m=s ð14Þ 0:863 W=S
BFL ¼ þ hobs
1 þ 2:3G gCLclimb
Here, pitching moment about the aircraft’s nose x ¼ 0     ð15Þ
1 655
is calculated during the VL analysis. The additional  þ 2:7 þ pffiffiffiffiffiffiffiffiffiffiffiffi
Tav =W  U =SL
5 m/s are a safety margin.

Performance module. Using the Breguet range equation, The relevant values can be found in Table 8.
stated earlier in equation (3), the weight decrease over The landing distance is used to measure the landing
a fraction of a flight phase can be determined. The performance. The landing is divided into three sec-
cruise flight phase is divided in a finite amount of tions, for which the distance covered is calculated sep-
segments, for each segment a constant lift-to-drag arately. Since the landing distance depends on many
ratio and a constant altitude are assumed. additional variables, like pilot skills and weather con-
Moreover, the specific fuel consumption is assumed ditions, an additional safety factor of 1.66 is intro-
to be constant during cruise flight. The weight frac- duced, as required by aviation authorities. The
tion, defined as the weight at the end of the flight approach length is calculated as follows
phase divided by the weight of the start, determines
the amount of fuel burnt during cruise flight. hobs  hflare
sapp ¼ ð16Þ
By calculating the lift-to-drag ratio for all flight tan app
missions, the payload range diagram of the aircraft
can be constructed. An example of the payload- To calculate the distance covered during flare, the
range diagram generated is shown in Figure 18. flare radius needs to be approximated
Note that the lines are not linear, due the presence
of trim drag. On the x-axis, the cruise range is indi- hflare
Rflare ¼ ð17Þ
cated. This means that the distance covered during 1  cos app
climb and descent are not included, though they are
The flare distance is subsequently calculated as
follows

sflare ¼ Rflare sin app ð18Þ

To calculate the distance covered during the

ground run, the drag parameter, KA needs to be

Table 8. Values used for take-off runway length calculation.

Parameter Symbol Value Unit


G climb min
1 
Climb gradient climb sin ððT  DÞ=W Þ

Minimum climb min 0.030
gradient
Climb lift coefficient CLclimb CL at 1:2Vmin –
Figure 18. Payload-range diagram as generated by the Obstacle height hobs 10.6 m
program. U 0:01CLmax þ0:02 –
OEW: operational empty weight; MTOW: maximum take-off By-pass ratio BPR 8.7 –
weight; ZFW: zero fuel weight.

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

van Dommelen and Vos 2467

Table 9. Values used for landing runway length calculation. Table 10. Assessed climb gradient conditions.

Parameter Symbol Value Unit CS Nengavail Slats Gear V

Approach speed Vapp 1:3Vstall m 25.111 Nengtot 1 retracted retracted 1:2Vs1


Approach slope app 3 25.121a Nengtot 1 deployed extended 1:1Vs0
Flare height hflare 5 m 25.121b Nengtot 1 deployed retracted 1:2Vs0
Braking friction brake –0.5 – 25.121c Nengtot 1 retracted retracted 1:2Vs1
coefficient 25.121d Nengtot 1 deployed extended 1:5Vs0
Lift coefficient during CLground 0.1 – 25.119 Nengtot deployed extended 1:3Vs0
ground roll
CS: Certification Specification.
Zero-lift drag CD0clean þCDslat þCDgear –
coefficient
Oswald factor e 0.8 –
Touch down speed Vtd 1:15Vstall m/s
chosen configuration an optimizer can be employed
to find a feasible (or even optimal) BWB geometry
Pilots reaction time ttd 1.5 s
that fulfills all the top-level requirements and CS con-
straints. In the present incarnation of the tool, a gra-
dient-based optimization algorithm is employed to
calculated first find the maximum harmonic range (Rharmonic ) at a
  constant MTOW within a predefined set of con-
 CL straints. The harmonic range is defined as the max-
KA ¼  brake  CD0  ground ð19Þ
2WMLW =S Ae imum range that can be flown at the maximum
structural payload weight. The mathematical formu-
The distance covered during the ground run is calcu- lation for the optimization scheme is as follows
lated as follows
R0
minJ ¼ ð23aÞ
1 brake Rharmonic ðxÞ
sground ¼ ln þ Vtd ttd ð20Þ
2g0 KA brake þ KA V2td
subject to:
The total landing distance is found by adding the
three distances and multiplying it with the prescribed y1 ¼ 0 ð23bÞ
safety factor of 1.66.
y2 5 0 ð23cÞ
sld ¼ 1:66ðsapp þ sflare þ sground Þ ð21Þ
where x represents the vector with design variables
Table 9 lists all the values used in the calculations. (Table 5) and y1 and y2 represent the constraints.
To compute the climb gradients in the various con- The constraints that have been used in the design
figurations the drag polar is used that is obtained cases of the subsequent section are presented in
from the aerodynamic analysis module. The climb Table 11. Note that y1 is the vector containing all
gradient is calculated for five one-engine-inoperative equality constraints of Table 11, while y2 is the
situations (CS-111 and CS-25.121a-d) and a single all- vector containing all inequality constraints.
engines-operative situation (CS-25.119). The climb
gradient, is computed as follows
Results and discussion
T Nengavail 1 In this section, the results are presented of three dif-
¼ Fmax cont  ð22Þ
W Nengtot L=D ferent BWB configurations for which the top-level
requirements are stated in Table 1. The results of
The weight used in each of the calculations is the the preliminary sizing process are shown
MTOW. The maximum continuous rating,4 Fmaxcont in Table 12. This is the results for a BWB of arbitrary
is equal to 0.94 and the lift-to-drag ratio is calculated configuration with four engines, and an aspect ratio
from the lift and drag polar for the prescribed condi- of 4, with the assumptions as indicated in Figure 7.
tions. The engine condition, flap configuration, gear These results are valid for all three BWB aircraft that
configuration, and evaluation speed for each climb are discussed in the subsequent sections. These results
requirement can be found in Table 10. form the basis for the generation of the geometry
in the MMG and the detailed analysis modules that
follow.
Optimization module The wing loading is somewhat lower compared to
The input of the geometry module and the output of conventional aircraft in the same category. The A380-
the analysis module are structured such that for a 800, for example, has a wing loading at MTOW of

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

2468 Proc IMechE Part G: J Aerospace Engineering 228(13)

Table 11. Constraints summary for a four-engine aircraft. 6600 N/m2 (almost twice as high) and a thrust-to-
weight ratio of 0.233 N/N, which is comparable to
Parameter Constraints Unit
the value quoted in Table 12.8 The lower wing loading
Wing span b 5 80 m is caused by the lower CLmax values in take-off and
Aircraft overall length l 5 80 m landing conditions, since only slats are applied. The
Cabin floor area Scabin 5Scabinreq m2 lower CLmax values are therefore compensated by the
Cargo volume Vcargo 5Vcargoreq m3 larger wing area. Although not relevant in the prelim-
Take-off distance sTO 4sTOreq m
inary sizing, the larger wing area is also necessary to
accommodate the required payload and the fuel.
Landing distance sld 4sldreq m
By operating the MMG and subsequent analysis
Stall speed take-off VstallTO 4Vstall, TOreq m/s
modules, the designer can almost instantly analyze
Stall speed clean Vstall 4Vstallreq m/s one of the three BWB configurations and compare it
CS-25.111 25:111 4 0:017 – to the set of constraints of Table 11. To construct a
CS-25.121a 25:121a 4 0:005 – BWB, the designer needs to provide a set of 30 input
CS-25.121b 25:121b 4 0:030 – parameters (x). The MMG subsequently generates the
CS-25.121c 25:121c 4 0:017 – geometric module and the analysis modules can be
CS-25.121d 25:121d 4 0:027 – employed to analyze the airplane. Finally, its per-
CS-25.119 25:119 4 0:032 – formance characteristics are evaluated to judge
Maximum trim deflection e 5 12 deg which design is most promising. Alternatively, the
Minimum static margin SM 4  10 % optimizer can be employed to change the design
Directional stability Cn ¼ 0:010 –
vector and find a solution for which the harmonic
range is maximized. This is what has been done for
Dihedral effect Cl 5 0 –
the three examples in this section. They are presented
Take-off rotation speed Vrot 5 VstallTO m/s
in Figure 19. Note that in the optimization process,
Drag divergence Mach MDD 4Mcruise – the wing loading and thrust-to-weight ratio might
Nose wheel load 0:05 5 Fnlg 5 0:20 – change as a result of the altered geometry. All three
Nose landing gear position xnlg 4 xnose þ 0:5 m designs eventually comply with same set of constraints
Main landing gear position xle 5 xmlg 5 xte m (displayed in Table 11).
CS: Certification Specification.
It should be noted that for all three designs the
vertical tail planform shape is identical based on stat-
istical data from Obert.18 It has an aspect ratio of 1.9,
a taper ratio of 0.3, a leading-edge sweep angle of 45 ,
Table 12. Preliminary sizing results for a BWB with four a root thickness of 14%, and a tip thickness of 12%.
engines for Rcruise ¼ 11, 000km, Wpl ¼ 66 metric tons, and If twin tails are selected, the dihedral of each vertical
A ¼ 4. tail is set to 20 .
Symbol Description Value Units

W/S Wing loading 3670 N/m2

Design of a ‘‘conventional’’ BWB
T/W Thrust loading 0.259 N/N The most conventional BWB configuration from pre-
WTO Design take-off weight 406103 kg vious studies has aft-swept wings, two vertical tails at
WOE Operational empty weight 200103 kg the wing tip (doubling as winglets), and the engines
Wf Fuel weight 139103 kg positioned close to the trailing edge of the center sec-
S Wing area 1080 m2 tion (see Figure 20). When trying to find a solution for
b Wing span 65.9 m
this configuration that satisfies all the constraints of
Table 11, various constraints appear to be active:
Teng Take-off thrust per engine 258 kN
landing distance, minimum and maximum nose-
BWB: blended wing body. wheel load, and the minimum static margin,

Figure 19. Three optimized BWB configurations.

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

van Dommelen and Vos 2469

Figure 20. Conventional BWB configuration with aft-mounted engines (all dimensions in meters).

Table 13. Design parameters for optimized ‘‘conventional’’ BWB configuration with aft-mounted engines.

Section 1 2 3 4 5 6 Unit

Chord 39.7 39.1 35.1 23.2 12.0 5.3 m

t/c 0.188 0.176 0.157 0.157 0.142 0.142 -

Twist 1.5 –0.8

Sweep 17.6 57.6 48 48 48

Dihedral 0.6 –0.5 1.0 2.0 4.6
Span 1.2 3.6 8.9 14.7 32.4 m
Tail h 8.67 m
BWB: blended wing body.

to name a few. The most important aspect of this This concept has the highest lift-to-drag ratio. The
configuration is discussed in this section. The opti- maximum lift-to-drag ratio is 27.9, with an average
mized parameters used to construct this concept are lift-to-drag of 27.2 over the harmonic range. The
given in Table 13. cause for this high L / D is the favorable lift distribu-
The wing area of the ‘‘conventional’’ BWB is tion and the presence of the winglets. The fuel con-
1091 m2, which gives a wing loading of 3650 N/m2. sumption is 0.0207 kg/pax/km, which is 10% lower
The empty weight of the three presented aircraft is compared to the other concepts. The harmonic
190  103 kg, which is the highest of the three aircraft. range is 14,400 km and the ferry range is 17,700 km.
The reason for this is the lift distribution, which gives The main influences on the harmonic range are the
a higher bending moment and causes a larger struc- fuel weight and the aerodynamic efficiency.
tural weight of the wing-body structure. Since the The ‘‘conventional’’ BWB has the lowest maximum
payload weight and the MTOW of all aircraft are lift coefficients and consequently the longest take-off
equal, this leaves a smaller amount of fuel weight and landing distances. The lift distribution is unfavor-
compared to the other concepts. able for low speed flight. Tip stall is likely to occur

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

2470 Proc IMechE Part G: J Aerospace Engineering 228(13)

Table 14. Design parameters for optimized BWB configuration with wing-mounted engines.

Section 1 2 3 4 5 6 Unit

Chord 40.1 39.4 34.7 23.2 13.4 3.6 m

t/c 0.178 0.177 0.174 0.130 0.130 0.130 -

Twist 3.0 2.5

Sweep 28.6 64.7 47.7 46.2 46.2

Dihedral 0.3 –0.9 0.9 0.7 5.0
Span 1.0 3.3 8.4 13.3 32.1 m
Tail h 8.03 m
BWB: blended wing body.

and the lift distribution is unfavorable for the max- coefficients in clean and take-off/landing configur-
imum lift coefficient. The aircraft has a minimum ation are 1.47 and 1.63 which is considerably higher
static margin of 8.4%, which occurs when flying than the previous concept. The take-off and landing
without payload and at maximum fuel weight. With distances are smaller and are 1670 m and 2230 m,
these conditions the CoG is shifted far aft, causing the respectively.
aircraft to be unstable. The aircraft is unstable over The most notable aspect of these concepts are the
most parts of its flight envelope. The shift in CoG nose loads. The variation is very large and the nose
expressed in terms of the MAC is the largest of the loads are at the limits of 5% and 20% of the weight,
three concepts (xcg ¼ 14:9% of MAC). This is respectively. This is caused by the short-coupled
mainly caused by the position of the fuel tanks nature of the BWB. The nose landing gear is close
which are far aft relative to the payload and empty to the nose, and cannot be moved forward to increase
weight’s CoG. the margins of the nose loads. The rudder trim deflec-
tion in one-engine-out condition is also large (21 )
compared to the other two concepts, which due to
Design of a BWB with wing-mounted engines the fact that the critical engine is located further out-
The parameters used to construct the planform and board (Figure 21).
vertical tail of the optimized BWB with wing-mounted
engines are given in Table 14. The engine position is
Design of a forward swept BWB
changed from aft-mounted engines, to wing-mounted
engines. The reason for this is to investigate the effect A more unusual BWB configuration that has been
of the engine position on the longitudinal stability of analyzed is a forward-swept concept with engines
the aircraft. The previous concept showed a minimum hanging from pylons under the wing (see Figure 22).
static margin of 8.4%, while this concept has a min- The input parameters used to define this concept are
imum static margin of only 2.5%. Moreover, the given in Table 15. The forward sweep can exhibit con-
concept turns out to have a positive static margin siderable benefits in transonic flow, due to the
throughout most of the flight envelope. Over the har- increased shock-wave sweep, the reduction of tip-
monic range, the aircraft is always stable. The only stall tendencies, and successful implementation of nat-
instability occurs during a ferry range mission, at the ural laminar flow technology.18 Even though none of
very start of the cruise flight, with almost full fuel these effects can be quantified in the conceptual design
tanks. stage, it is of interest to see if such a configuration
Compared to the previous concept, this concept could meet the constraints of Table 11. The engines
has a lower lift-to-drag ratio. The maximum lift-to- are located beneath the wing for longitudinal stability
drag ratio is 25.5 and the average lift-to-drag ratio in reasons.
cruise flight is 23.7. The lower aerodynamic efficiency From the planform the decreased leading edge
is caused by the lift distribution, leading to a 10% sweep over the previous concept can be observed.
higher fuel consumption. The lift distribution tends This confirms the expectations that less leading edge
to unload the wing tips. This, in turn, is favorable sweep is required for forward-swept wing aircraft
for the structural weight and low speed performance. compared to aft-swept wing aircraft, to meet drag-
The empty weight is 183  103 kg, which is smaller divergence Mach number requirements. The other
compared to the previous concept. The impact of most notable aspect is the very low operational
the lower aerodynamic efficiency on the range is lim- empty weight. The operational empty weight is only
ited by the smaller empty weight. The smaller empty 157  103 kg. Since the MTOW of the three concepts
weight means that more fuel is available. The har- is kept equal, the fuel weight is relatively large. This,
monic range is 13,599 km and the ferry range is in turn, means a very large range. The harmonic range
14,300 km. The latter is relatively small, because of is 16,500 km and the ferry range is 21,900 km. The
the limited size of the fuel tanks. The maximum lift aerodynamic efficiency is somewhat lower than

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

van Dommelen and Vos 2471

Figure 21. Conventional BWB configuration with wing-mounted engines (all dimensions in meters).

Figure 22. Forward-swept BWB configuration (all dimensions in meters).

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

2472 Proc IMechE Part G: J Aerospace Engineering 228(13)

Table 15. Design parameters for optimized forward-swept BWB configuration.

Section 1 2 3 4 5 6 Unit

Chord 36.9 36.1 22.1 19.7 15.0 2.6 m

t/c 0.189 0.189 0.180 0.18 0.150 0.15 -

Twist –0.2 4.0

Sweep 29.5 64.6 –28.6 –28.6 –28.4

Dihedral 0.9 4.0 0.5 1.7 5.0
Span 1.1 8.6 11.1 16.7 31.4 m
Tail h 8.34 m
BWB: blended wing body.

the aft-swept wing concepts with the maximum lift-to- conventional aircraft. There exist no empirical data-
drag ratio being 23.7 and the average lift to drag ratio base for BWB aircraft and therefore the results from
in cruise flight being 22.5. The fuel consumption the analysis models should not be interpreted as being
is almost equal to the previous concept and is the absolute truth. The methods for aerodynamic ana-
0.0235 kg/pax/km. lysis, weight estimation, CoG determination, and sta-
The empty weight of the aircraft is small due to the bility and control have not been validated against an
lift distribution. The lift distribution is also favorable experimental database. Further investigation should
at low speed conditions. Stall is likely to occur on the demonstrate the verification of each of the analysis
inboard parts of the wings and the maximum lift coef- methods using, for example, higher order models.
ficient is relatively high. The maximum lift coefficients
in clean and slats extended configuration are 1.44 and
1.58 respectively, which is close to the values of the
Conclusions
previously discussed concept. The take-off and land- A conceptual design tool for the configuration design
ing distances are similar as well. of a BWB has been presented. The capability of this
The minimum static margin lies in between the pre- tool allows the designer to quickly alter the main geo-
vious two concepts and has a value of 5.54%. Since metric components of the airplane and analyze its
the fuel’s CoG, the payload’s CoG and the empty effect on key characteristics such as flight and field
weight’s CoG are located close to each other the performance, static stability, balance, weight, and
CoG travel is small compared to the aft-swept wing passenger accommodation. The tool includes methods
concepts (9.41% of the MAC). This also means that for the sizing of all components (body, wing, fin, land-
the trim variation is less and therefore the maximum ing gear) relevant to this design stage and relies on
trim deflection is less than one degree. The size of the modified handbook methods for BWB-specific com-
control surface may be reduced for this type of ponents, such as the structural lay-out of the cabin,
configuration. the structural weight estimation and aerodynamic
analysis of the aircraft. The methods for aerodynamic
analysis and weight estimation have not been vali-
Interpretation of results
dated. The resulting performance characteristics
In Table 16 the results of the three optimized aircraft should therefore only be used to compare the BWB
are displayed. Values printed in bold highlight the configurations with respect to each other.
best values, while values printed in italic highlight Three configurations have been designed for max-
the worst values. It can be seen from this table that imum range at a payload weight of 66 metric tons and
all of the BWB concepts have better climb perform- a MTOW of 406 metric tons. Each configuration had
ance than what is specified in the regulations. This is four engines, a tricycle landing gear, and a twin-tail
due to the fact that the thrust-to-weight ratio and the configuration. Configuration 1 featured aft-swept
MTOW are held constant during the optimization. wings with aft-mounted engines and winglets dou-
This implies that the airplane thrust equals the bling as vertical tails. This resulted in a harmonic
thrust that is estimated at preliminary sizing, i.e. range of 14,400 km. Configuration 2 also had aft-
Ttake-off ¼ 1030 kN. As the airplane is being optimized, swept wings with winglets but with its engines pos-
it is becomes more aerodynamically efficient. With a itioned under the wing. Its harmonic range was
given thrust-to-weight ratio this leads to improved 13,600 km. The final configuration (3) was a for-
climb performance as per equation (22). ward-swept BWB with engines under the wing and
The overview of characteristics in Table 16 should twin tails on at the back of the fuselage. This aircraft
be used to compare the three BWB configurations had the highest harmonic range of 16,500 km. This
relative to each other. The methods that have been high range was attributed to its small shift in CoG
used to generate these numbers are based on simpli- during its mission, which resulted in very small trim
fied analytical models or empirical data from drag. Further research is required to validate these

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

van Dommelen and Vos 2473

Table 16. Concept comparison table.

Parameter Symbol Unit

Wing area S 1091 1043 1074 m2

Wing span b 64.9 64.4 62.7 m
Aspect ratio A 3.86 3.98 3.67 –
Wing loading MTOW W/S 3650 3818 3706 N/m2
Overall length – 41.4 40.5 38.5 m
MTOW WTO 406 406 406 103kg
OEW WOE 190 183 157 103kg
Payload weight Wpl 66.4 66.4 66.4 103kg
Fuel weight Wf 149 157 183 103kg
Harmonic range Rharmonic 14.4 13.6 16.5 103km
Ferry range Rferry 17.7 14.3 21.9 103km
Fuel consumption – 0.0207 0.0234 0.0235 kg/pax/km
Maximum lift-to-drag ratio ðL=DÞmax 27.9 25.5 23.7 –
Average lift-to-drag ratio ðL=DÞav 27.2 23.7 22.5 –
Maximum lift coefficient CLmax ,clean 1.12 1.47 1.44 –
Maximum lift coeff, slats CLmax ,ld 1.28 1.63 1.58 –
CoG travel – 14.9 13.6 9.41 % MAC
Cabin floor area Scabin 312 276 273 m2
Cargo volume Vcargo 186 216 239 m3
Take-off distance sTO 2.00 1.67 1.65 km
Landing distance sld 2.48 2.23 2.25 km
Stall speed TO land VstallTO 62.6 56.6 56.7 m/s
Stall speed clean Vstall 66.7 59.7 59.3 m/s
CS-25.111 25:111 7.4 5.5 5.0 %
CS-25.121a 25:121a 4.0 2.0 1.4 %
CS-25.121b 25:121b 7.4 5.5 5.0 %
CS-25.121c 25:121b 10.6 8.7 8.3 %
CS-25.121d 25:121d 17.0 16.7 16.4 %
CS-25.119 25:119 16.6 15.6 15.1 %

Maximum trim deflection e, max 4.5 4.8 0.9
Minimum static margin SM 8.43 2.47 5.54 % MAC
Weathercock stability Cn 0.0524 0.0311 0.0747 1/rad
Effective dihedral Cl 0.138 0.100 0.0329 1/rad
Take-off rotation speed Vrot 30.8 31.1 27.1 m/s

OEI rudder deflection r, OEI 6.24 21.2 8.41
Drag divergence outer 5 MDD 0.93 0.90 0.84 –
Drag divergence Mach trunk 4 MDD 0.82 0.82 0.82 –
Nose landing gear x xnlg 1.99 1.90 1.99 m
Main landing gear x xmlg 26.9 26.0 19.2 m
Minimum nose load – 6.3 5.0 8.5 %
Maximum nose load – 18.7 20.0 16.5 %
MTOW: maximum take-off weight; OEW: operational empty weight; CoG: center of gravity; TO: take-off; CS: certification specification.

results and optimize these concepts based on different Conflict of interest

performance metrics such as fuel consumption at con- None declared.
stant range.
Acknowledgements
Funding The authors would like to thank Dr E Torenbeek for shar-
This research received no specific grant from any funding ing his insights on nonconventional airplane configurations
agency in the public, commercial, or not-for-profit sectors. and introduction to relevant literature.

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015

2474 Proc IMechE Part G: J Aerospace Engineering 228(13)

References CD drag coefficient

1. Denning RM, Allen JE and Armstrong FW. The broad Cn yawing moment coefficient
delta airliner. Aeronaut J 2003; 107(1075): 547–558. Cl rolling moment coefficient
2. Liebeck RH. Design of the blended wing body subsonic CL lift coefficient
transport. J Aircraft 2004; 41: 10–25. D drag (N)
3. Hileman JI, Spakovszky ZS, Drela M, et al. Airframe h altitude or height (m)
design for silent fuel-efficient aircraft. J Aircraft 2010; M Mach number, fuel fraction, or moment (Nm)
47: 956–970. l aircraft overall length (m)
4. Roskam J. Airplane design, Part I: Preliminary sizing of
L lift (N)
airplanes. Lawrence, KS: DARcorp, 1990.
N integer number
5. Roskam J. Airplane design, Part I: Preliminary sizing of
aircraft. Lawrence, KS: DARcorp, 2005. R range (km)
6. Morris AJ. MOB A European distributed multi-disciplin- s runway length (m)
ary design and optimisation project. Cranfield S surface area (m2)
University, 2002. AIAA 2002-5444. t thickness (m)
7. Raymer DP. Aircraft design: A conceptual approach. T thrust (N)
Reston, VA: American Institute of Aeronautics and V volume (m3), or speed (m/s)
Astronautics, Inc., 2006. W weight (kg)
8. Anon. Jane’s all the world’s aircraft, http://jawa.janes. x aircraft x-position (m)
com/ (2011) x vector of design variables
9. Diedrich A, Hileman J, Tan D, et al. Multidisciplinary
design and optimization of the silent aircraft.  angle of attack (deg)
Massachusetts Institute of Technology, USA, 2006.  side slip angle (deg)
AIAA 2006-1323.  control surface deflection (deg)
10. Bolsunovsky AL, Buzoverya NP, Gurevich BI, et al. e wing twist (deg)
Flying wing: Problems and decisions. In: Torenbeek E climb gradient (%)
and Deconinck H (eds) Innovative configurations and
 wing dihedral (deg)
advanced concepts for future civil aircraft. 2005.
 wing sweep (deg)
11. Laban M, Arendsen P, Rouwhorst WFJA, et al. A com-
putational design engine for multi-disciplinary optimisa- braking coefficient
tion with application to a blended wing body  density (kg/m3)
configuration. National Aerospace Laboratory, NLR,  normal stress (N/m2)
2002. AIAA 2002-5446.  shear stress (N/m2)
12. Geuskens FJJMM, Koussios S, Bergsma OK, et al.
Non-cylindrical pressure fuselages for future aircraft.
Delft University of Technology, Netherlands, 2008. Subscript
AIAA 2008-1907.
13. Melin T. A vortex lattice MATLAB implementation for 0 zero altitude or initial value
linear aerodynamic wing applications. Sweden: Royal clean clean configuration
Institute of Technology, 2000. DD drag divergence
14. Roskam J. Airplane design, Part VI: Preliminary calcu- e elevator
lation of aerodynamic, thrust and power characteristics. eng engines
Lawrence, KS: The University of Kansas, 1990. f fuel
15. Kroo I. Aircraft design, synthesis and analysis. Palo ff fuel fraction
Alto, CA: Desktop Aeronautics, Inc., 2001.
ld landing
16. Torenbeek E. Development and application of a compre-
hensive, design-sensitive weight prediction method for
le leading edge
wing structures of transport category aircraft. Delft: lug luggage
Delft University of Technology, 1992. mat material
17. Geuskens FJJMM, Bergsma OK and Beukers A. mlg main landing gear
Analysis of conformable pressure vessels: Introducing nlg nose landing gear
the multibubble. Delft University of Technology, 2011. OE operative empty
AIAA 54561-869. pax passengers
18. Obert E. Aerodynamic design of transport aircraft. r rudder
Amsterdam: IOS Press, 2009. rot rotation
s0 stall in landing configuration
s1 stall in clean configuration
Appendix PL payload
TO take-off
Notation
SL sea level
A aspect ratio te trailing edge
b wing span (m) tfo trapped fuel and oil
c chord (m) u ultimate

Downloaded from pig.sagepub.com at SUNY MAIN LIBRARY on January 9, 2015