Deutscher Luft- und Raumfahrtkongress 2013

DocumentID: 301281

WING WEIGHT ESTIMATION METHODOLOGY FOR HIGHLY NON-

PLANAR LIFTING SYSTEMS DURING CONCEPTUAL DESIGN
A. Spohr
J. Schirra, J. Hoefling, A. Schedl
FH Aachen University of Applied Sciences
Flight Systems, Flight Guidance and Control
52064 Aachen, Germany

ABSTRACT
Unconventional and especially highly non-planar aircraft configurations are supposed to contribute to a significant
increase in efficiency of civil air transport vehicles. So called multidisciplinary design optimization tools are used to find
an optimum concerning aerodynamic, flight mechanic and structural performance but traditional methods to estimate the
wing weight of the lifting system are not always applicable to highly non-planar configurations. For this reason a wing
weight estimation tool is developed that combines the advantages of existing methods to achieve high physical accuracy
while reducing the computational effort simultaneously. The approach proposed in this paper unites classical analytical
methods with simple FEM models to exclude the dependency on statistical sets of data. A simplified beam model of the
structural wing box is used to determine the inner forces and moments caused by outer airloads. Based on these inner
loads the cross sections are designed by the use of analytical equations. During this process the cross-sectional area is
minimized by an automated algorithm while limiting the occurring stresses to a given maximum stress level. Two
validations are performed subsequently including one conventional and one unconventional highly non-planar lifting
system.

NOMENCLATURE 1. INTRODUCTION
A Cross sectional area [m²] Unconventional aircraft configurations like the Box Wing,
a st
Width of 1 reinforcement [m] Joined Wing aircraft or C-Wings are supposed to
nd contribute to a significant improvement of efficiency in civil
b Width of 2 reinforcement [m] air transport. Provided with aerodynamic benefits due to
st
c Height of 1 reinforcement [m] low induced drag these configurations might have
nd
d Height of 2 reinforcement [m] drawbacks concerning flight mechanics and structural
FoS Factor of safety [-] weight in comparison to conventional cantilever wings. So
𝐹𝑥 Shear force in x-direction [N] called multidisciplinary design optimization (MDO) tools
are often used for weighing up the advantages and
𝐹𝑦 Normal force in y-direction [N]
disadvantages of promising configurations. With these
𝐹𝑧 Shear force in z-direction [N] tools several disciplines like aerodynamics, structural
h Height of the structural wing box [m] mechanics and flight mechanics are combined and
4
𝐼𝑥 Second moment of inertia along x-axis [m ] coupled by performance requirements to find the optimal
𝐼𝑧 Second moment of inertia along z-axis [m ]
4 solution. Since this procedure is done in the conceptual
𝐼𝑥𝑧 Mixed moment of inertia [m ]
4 design phase, the used models are kept as simple as
𝑀𝑥 Bending moment along x-axis [Nm] possible and as accurate as necessary to minimize the
computational effort.
𝑀𝑦 Torsional moment along y-axis [Nm]
𝑀𝑧 Bending moment along z-axis [Nm] The student hosted research project Advanced Aircraft
𝑆𝑥 First moment of inertia along x-axis [m³] Configuration is dealing with the investigation and
𝑆𝑧 First moment of inertia along z-axis [m³] development of unconventional aircraft configurations and
has set itself the task to develop such a MDO tool. For the
t Thickness [m] structural module high-fidelity FEM approaches do not
w Width of the structural wing box [m] make sense during this early design stage but simple
𝜎 Normal stress [MPa] methods for estimating the structural mass of the lifting
𝜏 Shear stress [MPa] system are appropriate. A suitable approach to do so is
proposed in this paper.
Subscript
all allowable 1.1. Empirical methods based on statistics
b bending
During the last decades several well-engineered methods
encl enclosed were developed for conventional cantilever wings which
n normal are common use in aircraft industry during the conceptual
sk skin design phase. Since these methods mostly rely on
sp spar statistical data of existing aircraft, they are not applicable
vM von Mises to unconventional configurations or highly non-planar

1
 Deutscher Luft- und Raumfahrtkongress 2013

lifting systems. Over-determined systems like Box Wing or actual wing structure is replaced by equivalent beam
Joined Wing aircraft may moreover induce specific inner models with more accurate cross-sections than in the
load distributions which are not taken into account by double plate model (see Figure 1.2). In contrast to the
traditional formulas developed for cantilever wings. A very approach presented in chapter 1.2 the shape of the cross-
famous approach for example has been presented by section is more detailed and can account for more
Torenbeek [1]. His formula for estimating the wing weight complex load cases. The overall mass of the structural
during early design stages unites several factors of wing box is estimated by designing all sections according
influence like the aspect ratio, relative thickness and the to the inner loads with a given maximum stress and
wing area with constant factors based on statistics from summing up the masses of the needed material.
existing aircraft.

1.2. Double plate model

More suitable methods like the double plate model are
capable of taking load distributions of over-determined
lifting systems into consideration, but the design is
restricted to pure bending along one axis while shear
forces and torsion are neglected completely (see
Leoviriyakit [2]). The basis for this method is the
calculation of the inner bending moments according to the
outer loads. Analytical or simple computer codes are
commonly used for that. In a second step the cross
section of the structural wing box is assumed to consist of
two plates, one at the upper and the other one at the lower
side of the box (see Figure 1.1 (b)).

Figure 1.1: Double plate model acc. to Leoviriyakit [2] Figure 1.2: Cross section according to Hajela et al. [3]

The vertical distance 𝑡 between the plates and the width 𝑐𝑠 1.4. Synthesis of advantages
are given by the dimensions of the airfoil and thus the skin
In this paper several approaches of existing methods are
thickness 𝑡𝑠 , which is same for both plates is the only
combined to exploit their advantages for developing a
design variable remaining. The inner bending moment
wing weight estimation tool especially for highly non-
resulting from the lift distribution is split into a pair of forces
planar lifting systems. In contrast to some existing
which are acting on the two plates causing normal
methods (see Kroo et al. [4]) the proposed approach
stresses. By rearranging equation (1) and defining a
assumes that the cross-sectional faces have to carry all
maximum allowable stress level the required skin
occurring loads simultaneously which is close to reality.
thickness 𝑡𝑠 can be evaluated for every cross section.
Hence, a rectangular cross section with reinforcements in
the corners is preferred and the cross section is
𝑀 considered with its real shape instead of smearing and
σ= (1)
𝑡 ∙ 𝑡𝑠 ∙ 𝑐𝑠 simplifying the material distribution like mentioned for
example in Bindolino et al. [5]. Although no full size FEM
The associated cross-sectional areas as well as the total analyses of the wing box is done, a simple FEM approach
weight of the simplified structural wing box can be for a beam model is needed to derive the inner forces and
estimated subsequently. This method is very simple but moments based on the outer airloads. Thereby it is only
the actual load situation is simplified significantly and necessary to divide simple line models into a finite number
many effects are neglected. of elements instead of a time consuming meshing process
for complex geometries. Basically the main task of the
1.3. Equivalent beam model proposed tool is the design of the structural wing box
which has to carry the loads. Estimating the masses of
To account for oblique bending, shear forces and torsional secondary structures like the control systems, fuel
moments more accurate methods for example by Hajela systems and additional aerodynamic shapes is not in the
et al. [3] were developed, still being suitable for focus of the paper, but proven approaches are existing as
unconventional highly non-planar configurations. The for example stated in Gallman et al. [6] and Samuels [7].

2
 Deutscher Luft- und Raumfahrtkongress 2013

2. METHODOLOGY 2.2. General structure of the proposed tool

Since the multidisciplinary design optimization tool
2.1. Summary of existing methods
including the structural module is part of the conceptual
The shared objective of all existing methods presented in design phase, a compromise between accuracy and
chapter 1 or proposed by Bindolino et al. [5] or Seywald [8] simplicity has to be found. Consequently the following
is to reduce the cross-sectional area while limiting all requirements were set for the development of the wing
occurring stresses to the maximum allowed stress; the so weight estimation tool:
called fully stressed design (see Hajela et al. [3] and
 Applicability for highly non-planar lifting systems
Gallman et al. [6]). Additional constraints concerning the
maximum displacement, aeroelastic effects and minimum  Suitability for over-determined systems
gauge thickness are usually part of the optimization. The
suitability for unconventional configurations and  Consideration of up to two planforms
associated effects like oblique bending is given by using
 Low computational effort
asymmetric reinforcements in the corners of the wing box
(see Figure 2.1) and appropriate models for calculating the  No classical FEM analyses of the whole wing box
inner loads.
 Waiver of iterative processes as far as possible
 Consideration of a minimum gauge thickness
Additionally to these requirements two versions of the tool
will be established. These are one stand-alone version
with graphical user interface plus graphical output and
another release in a black box manner which can be
integrated into the MDO tool. To guarantee flexibility,
TXT-files are used as input and output files.
The general structure of the tool is depicted in Figure 2.2
and explained in the following.

Figure 2.1: Asymmetric material distribution of a Joined

Wing according to Hajela et al. [3]

In the case of over-determined configurations like Joined

Wings or Box Wing configurations, the inner load
distribution depends on the stiffness distribution as
mentioned in Kroo et al. [4] and Gallman et al. [6]. The
stiffness distribution in turn depends on the cross-sectional
area and its distribution. Thus a straight forward procedure
is not possible and an iterative approach becomes
unavoidable. An initial stiffness value has to be estimated
somehow to start the iteration with a first set of inner
loads. Based on that, a more reliable stiffness distribution
can be evaluated due to the cross-sectional design.
All methods developed during the last decades differ
concerning their level of accuracy, individual assumptions,
design variables and cross-sectional shape. In Kroo et
al. [4] for example, bending and shear effects are
decoupled in a way that the skins carry shear and
torsional loads only. The bending loads have to be
absorbed by the spars and the rectangular reinforcements
in the corners. The representation of the cross-sectional
shape varies from simple rectangular plates to more
realistic versions with stiffeners and lumped stringers in Figure 2.2: General structure and process of the tool
the corners (see Bindolino et al. [5]).

3
 Deutscher Luft- und Raumfahrtkongress 2013

The basis for the wing weight estimation consists of the The next step of the tool is the Cross section design
outer geometry of the lifting system in terms of planform module on which this paper is focused. Here the inner
data, general parameters like material properties and the forces and moments are used to calculate the necessary
aerodynamic data set. The latter, especially the lift cross-sectional area and its distribution to keep all
distribution is obtained from a vortex lattice method called occurring stresses below a given limit. Basically a two
LamDes, which has been improved for the demands of dimensional design is performed for every cross section,
highly non-planar lifting systems (see Spohr [9]). Due to which is completely independent of the type of static
the flexibility of the TXT-file as input format other codes specification of the system. Since the inner forces and
can be used instead easily after adapting the output moments of over-determined systems depend on their
format. stiffness distribution (see Kroo et al. [4] and Gallman et al.
[6]) an iteration loop becomes necessary. At the beginning
The resulting set of data is read into the Preprocessor
a stiffness distribution is estimated based on the minimum
where the airloads are calculated according to the given gauge thickness. The resulting inner forces and moments
lift distribution. Drag forces are comparably low and are lead to more realistic cross-sectional areas and therefore
thus neglected at the moment but can be considered stiffnesses. This process including the FEM line model
easily in future versions. The flight state including things and the cross section design module is repeated until
like the load factor and the flight velocity is also taken into convergence is reached.
account in this module. The planform coordinates together
with positioning data for front and rear spar lead to the After computing the needed cross-sectional areas a
location of the actual structural wing box and its centerline. Postprocessor module is used to calculate the actual wing
This centerline is divided into a finite number of elements weight. Therefore the average cross-sectional area within
according to a suitable element size which can be chosen one section is multiplied by the according element length
by the user. Due to the fact that the resulting aerodynamic and at the end the sum is multiplied by the density of the
loads are acting approximately at the quarter chord line, a used material to obtain the mass. For the stand-alone
lever arm to the centerline of the structural wing box version graphical outputs in terms of diagrams and tables
arises. The resulting moment along the global y-axis is showing the needed areas and the shape of the
also computed in this module. reinforcements are generated. In addition it is possible to
activate an interface to CATIA V5 that consists of an Excel
After transferring the data to the FEM line model the inner sheet with embedded VBA macro. The macro is capable
forces and moments are calculated based on a classical of exporting the vertices to the CAD environment and to
FEM beam approach. Since the FEM model does not connect them by polylines automatically. Finally the user
have to handle a complex geometry but only a line which can decide if the macro is supposed to create a solid body
is located in the three dimensional space, the of the whole structural wing box based on the polylines.
computational effort is desirable low. For statically
determined systems the influence of the cross-sectional
2.3. Cross section design module
area and thus of the stiffness cancels out and a straight
forward calculation becomes possible. In contrast to that The task of the cross section design module in particular is
an iteration loop with initial stiffness estimation is to lay out the material distribution in every cross section
necessary for any over-determined lifting system. according to the inner forces and moments. Figure 2.3
shows the general shape of each cross section and how it
is located within the airfoil.

Figure 2.3: Structural wing box within a Joined Wing configuration

4
 Deutscher Luft- und Raumfahrtkongress 2013

Figure 2.4: Cross section with rectangular reinforcements Figure 2.5: Cross section with triangular reinforcements

While the outer dimensions width and height are given by An integrated sub module of the algorithm calculates all
the preprocessor according to chord length and airfoil occurring stresses according to the given analytical
thickness, all other parameters are determined by the formulas. Based on that, the combined von Mises stress is
design module. The thickness of front and rear spar, calculated along the spar and the skin which has to be
sometimes also called webs and the size of the upper and lower than the maximum allowed tensile stress of the
lower skin (also: caps) is the same at front and rear material divided by a factor of safety. This iterative
respectively at the upper and lower side. Similar to that the process is repeated until the needed cross-sectional area
rectangular reinforcement in the upper right corner has the reaches a minimum while fulfilling the maximum stress
same size as the one in the lower left corner. Same holds criteria. The individual formulas and calculation steps are
for lower right and upper left corner (see Figure 2.4). explained in the following.
Thereby point symmetry around the center is achieved
and asymmetric material distributions are still possible. As First of all the inertias are evaluated according to the
coordinate system the aircraft fixed system is used and equations (2), (3) and (4).
the cuts through the wing are oriented parallel to the
global x-axis.
𝐼𝑥 (𝑧) = 𝑧 2 𝑑𝐴 (2)
The shape described above is not ideal for a practical 𝐴
implementation because of its sharp corners and the
involved notch effects. Despite the simple mathematical
description of the rectangles a more appropriate design 𝐼𝑧 (𝑥) = 𝑥 2 𝑑𝐴 (3)
with triangular reinforcements in the corners as depicted in 𝐴
Figure 2.5 is established in addition. The proposed tool
performs an optimization with rectangular and triangular
reinforcements and the comparison of the resulting wing 𝐼𝑥𝑧 = − x ∙ z 𝑑𝐴 (4)
weights leads to the lighter solution. It is ensured that 𝐴
either all reinforcements are rectangular or triangular but
changes from one cross section to the next one are Then the normal stresses due to bending moments along
prevented. the x- and z-axes are determined by using the formula for
oblique bending:
For determining suitable values for all six parameters
(𝑡𝑠𝑝𝑎𝑟 , 𝑡𝑠𝑘𝑖𝑛 , 𝑎, 𝑏, 𝑐, 𝑑) they have to be chosen in a way that 1
all occurring stresses are below the given limit. Since 𝜎𝑦𝑏 = 𝑀𝑥 𝐼𝑧 − 𝑀𝑧 𝐼𝑥𝑧 ∙ 𝑧 − 𝑀𝑧 𝐼𝑥 − 𝑀𝑥 𝐼𝑥𝑧 ∙ 𝑥 (5)
∆
there is an endless number of possible designs, which
would ensure that, additional constraints are necessary to
with
achieve the most effective material distribution. Thus an
optimization algorithm is chosen to minimize the cross- 2
sectional area while constraining the maximum occurring ∆ = 𝐼𝑥 𝐼𝑧 − 𝐼𝑥𝑧 (6)
stresses. Unfortunately it is not possible to avoid the
iterative process here, so it has to be ensured that the The resulting stresses for both a moment along x and z
optimization results are all valid and that the process is show a linear distribution with its climax in two opposite
working as fast and effectively as possible. Therefore corners and its minimum in the other two corners.
simplified formulas are used to estimate appropriate
starting values for spar and skin thickness and for the four The influence of normal tensile or compressive loads is
reinforcement variables. For this estimation it is assumed evaluated by dividing the acting force by the cross-
that all reinforcements are quadratic and have the same sectional area of the respective section (see equation (7)).
size which leads to an axisymmetric structure. Another Superposition of all stresses acting normal to the section
simplification hypothesizes that the skins and spars have in y-direction leads to the resulting normal stresses.
to carry shear stresses only and the bending moment is
absorbed by the reinforcements. Without decoupling both 𝐹𝑦
effects the dependency inbetween would allow for more 𝜎𝑦𝑛 = (7)
than one possible solution and more complex formulas 𝐴
would become necessary.

5
 Deutscher Luft- und Raumfahrtkongress 2013

The occurring shear stresses are also caused by two

different effects. First of all, shear forces in x- and
z-direction are leading to shear stresses in the skins and
spars. The equations (8) and (9) are used to calculate the
distribution of the first moment of area, which looks similar
to that depicted in Figure 2.6 for a force in only one
direction (Note: the positive y-direction in the figure
corresponds to the negative x-direction used in the tool).
The actual distribution for the shape used in this tool
shows additional kinks near the locations where the
reinforcements begin or end respectively (see Figure 2.7).

𝑆𝑥 (𝑧) = 𝑧 𝑑𝐴 (8)
𝐴

𝑆𝑧 (𝑥) = 𝑥 𝑑𝐴 (9)
𝐴

Figure 2.6: Principle Shear flow in thin-walled rectangle

(extracted from Lenz [10])

Figure 2.7: Qualitative depiction of the shear flow in one half of the cross section

Based on that, the resulting shear stresses can be (Equation (12)) for thin walled cross sections leads to a
determined according to equation (10) and equation (11). constant shear flow along the perimeter and in
combination with the local thickness, the actual shear
𝐹𝑥 ∙ 𝑆𝑧 stress at every location can be determined.
𝜏𝑦 = (10)
𝐼𝑧 ∙ 𝑡
𝑀𝑦
𝜏𝑡 (𝑠) = (12)
𝐹𝑧 ∙ 𝑆𝑥 2 ∙ 𝐴𝑒𝑛𝑐𝑙 ∙ 𝑡(𝑠)
𝜏𝑦 = (11)
𝐼𝑥 ∙ 𝑡
Since the location of the maximum normal stress does not
Secondly shear stresses are caused by the torsional match necessarily with the location of maximum shear
moment along the y-axis. Applying Bredt’s first formula stress, the perimeter of the cross section is divided into a

6
 Deutscher Luft- und Raumfahrtkongress 2013

finite number of points. The combined von Mises stress is

then derived at every single point to find the critical spot. This set of equations prevents the algorithm from choosing
values for the design variables that would cause an
For the spars the normal stress in y-direction and the overlapping of reinforcements or reinforcements being too
shear stress in z-direction are decisive to calculate the large to fit into the wing box.
according combined von Mises stress (see equation (13)).
The optimization process is stopped automatically if the
relative change in cross-sectional area falls below a given
𝜎𝑣𝑀 𝑠𝑝𝑎𝑟 = 𝜎𝑦2 + 3 ∙ 𝜏𝑦𝑧
2 (13) threshold. At the same time the constraint violation and
thus the stress deviation has to be lower than an accuracy
parameter preset by the user.
For the skins the normal stress in y-direction and the
shear component in x-direction are relevant (see 3. VALIDATION
equation (14)).
3.1. Planar lifting system: Boeing 727
𝜎𝑣𝑀 𝑠𝑘𝑖𝑛 = 𝜎𝑦2 + 3 ∙ 𝜏𝑦𝑥
2 (14) First of all a planar lifting system according to the Boeing
727 in the paper of Samuels [7] is used to validate the
cross section design module. Therefore it is combined with
Taking advantage of the point symmetry the maximum of the FEM line model only, since not all parts of the tool are
the von Mises stress distribution is evaluated in only four completely finished at the current stage of the
different sections to ensure that all other occurring development. The used outer shape and the dimensions
stresses are below the given limit. Considering the right of the structural wing box are depicted in Figure 3.1.
half of the cross-sectional shape in Figure 2.7, drawn in
solid black, the four sections are:
 Upper skin, right half
 Right spar, upper half
 Right spar, lower half
 Lower skin, right half
In all of these sections the maximum von Mises stress is
computed and set equal to the limit:

𝜎𝑎𝑙𝑙
𝜎𝑣𝑀 𝑚𝑎𝑥 = (15)
𝐹𝑜𝑆

Especially near the wing tip the inner forces and moments
can be that low that even a cross section with minimum
gauge thickness would resist the loads, with the maximum
occurring stress being significantly below the limit. Figure 3.1: Structural wing box according to Samuels [7]
Therefore an upstream process stage verifies if the
minimum material application is sufficient. If this is not the The acting outer airloads are calculated with the vortex
case, the normal algorithm is used to scale the lattice tool and applied to the centerline of the structural
reinforcements and spar and skin thicknesses. wing box. As mentioned in the paper the actual chordwise
lift distribution is simplified to point loads and the moment
In addition to the maximum stress constraint, further induced by the lever arm between quarter chord line and
conditions limit the range of values for the six design center line of the structural wing box is neglected. Thus
variables (𝑡𝑠𝑝𝑎𝑟 , 𝑡𝑠𝑘𝑖𝑛 , 𝑎, 𝑏, 𝑐, 𝑑). The lower boundaries for the airloads consist exclusively of lift forces in z-direction.
skin and spar thickness are set to minimum gauge
All additional parameters needed for the cross section
thickness and the parameters of the reinforcements have
design were chosen consistent to the paper and can be
to be greater or equal to zero. The upper boundaries are
taken from Table 3.1.
set quite loose, just excluding unrealistic values. More
important is the definition of the linear inequalities shown Parameter Unit Value
in equation (16).
Load factor: [-] 2.5
1 1 0 0 2 0 𝑎 𝑤
Factor of safety: [-] 1.5
0 0 1 1 0 2 𝑏 ℎ
2 0 0 0 2 0 𝑐 𝑤 Allowable stress: [MPa] 193
∙ 𝑑 ≤ 𝑤 (16)
0 2 0 0 2 0
𝑡𝑠𝑝 ℎ Density: [kg/m³] 2768
0 0 2 0 0 2
0 0 0 2 0 2 𝑡𝑠𝑘 ℎ Minimum gauge thickness: [mm] 3.175

Table 3.1: Constant parameters according to Samuels [7]

7
 Deutscher Luft- und Raumfahrtkongress 2013

The execution of the FEM line model leads to shear forces The distribution of the cross-sectional area needed to
in z-direction, a bending moment distribution along the resist the loads, is depicted in Figure 3.5. The trend line is
global x-axis and a torsional moment along the global descending towards the wing tip which is in accordance
y-axis as shown in Figure 3.2 and Figure 3.3. with the loads but the distribution is not as smooth as
expected. Additionally some irregularities occur
Relative spanwise position [-] concerning the use of reinforcements. In some sections
0 0.2 0.4 0.6 0.8 1 large reinforcements are present, while other cross
0.0E+00 sections are designed nearly without any reinforcements.
Possibly more than one solution can be found satisfying
-1.0E+05 the maximum stress constraint but the reason for that is
not clarified completely and the effect will be investigated
Inner Force [N]

-2.0E+05 in future work.

-3.0E+05 1.6E+05

-4.0E+05 1.4E+05

Cross-sectional area [mm²]

Fx
-5.0E+05 Fy 1.2E+05
Fz
1.0E+05
-6.0E+05
8.0E+04
Figure 3.2: Inner force distribution
6.0E+04
4.0E+04
4.0E+06
Mx
2.0E+04
3.0E+06 My
Mz 0.0E+00
2.0E+06
Inner Moment [Nm]

0 0.2 0.4 0.6 0.8 1

1.0E+06 Relative spanwise position [-]

0.0E+00
Figure 3.5: Distribution of cross-sectional area
-1.0E+06
3.2. Non-planar lifting system: Joined Wing
-2.0E+06
Since there is no Joined Wing or Box Wing configuration
-3.0E+06 ever built in the size of a transport aircraft, no data sets
0 0.2 0.4 0.6 0.8 1 are available for validating against existing aircraft. Thus a
Relative spanwise position [-] Joined Wing according to the paper of Miura et al. [11] is
chosen to prove the capabilities of the proposed wing
weight estimation tool in terms of highly non-planar
Figure 3.3: Inner moment distribution configurations. The used outer geometry and its
dimensions are depicted in Figure 3.6 and Figure 3.7.
Based on the inner forces and moments the cross-
sectional areas with its reinforcements are designed
accordingly which leads to a total mass of the structural
wing box of 2280 𝑘𝑔 for one wing. Compared to the
2303 𝑘𝑔 evaluated by Samuels this value is slightly lower
but a relative error of −1 % is more than acceptable.
The resulting structural wing box with its cross sections is
depicted in Figure 3.4 showing the outer contours in black
while representing the inner cut-outs in grey. All sections
are axisymmetric which is in great accordance to the fact
that only bending moments along the x-axis are acting and
thus no oblique bending effects are present.

Figure 3.4: Representation of the structural wing box Figure 3.6: Topview of the Joined Wing configuration

8
 Deutscher Luft- und Raumfahrtkongress 2013

For the near future all modules of the tool will be merged
together and the iteration loop for over-determined lifting
systems will be established. The integration into the MDO
environment is also planned for future work. For the cross
section design module in particular the starting values will
be improved by more accurate formulas or correction
factors based on the acting loads. Thereby the
convergence behavior can be increased furthermore.
Additional constraints concerning aeroelastic effects,
maximum deflection of the wing tip or buckling problems
Figure 3.7: Frontview of the Joined Wing configuration can be integrated easily into upcoming versions of the
tool.
The shape is reconstructed from the tabular values and The completed tool will be used afterwards to perform
sketches given in the paper and the position of the rear several studies about highly non-planar lifting systems
wing is adapted in a way that both centerlines of the wing concerning the influence of joint fixativity, the position of
box meet at the joint location to guarantee compatibility to joint locations, staggering dimensions and variations in
the tool. The dotted lines symbolize the centerline of the spanloading.
structural wing box and the solid dots show the locations
where the cross section design is performed. 5. REFERENCES
Unfortunately the applied loads are not completely defined [1] Torenbeek, Egbert. 1982. Synthesis of Subsonic
in the paper. Only the inner moment distributions along the Airplane Design. Dordrecht, Netherlands. Kluwer
x- and z-axes are given in two diagrams. Since the other Academic Publishers. ISBN: 90-247-2724-3.
moments and forces are unknown the calculation is
performed without them. For simplification reasons it is [2] Leoviriyakit, Kasidit. 2005. Wing planform
assumed that the given inner moments are the final result optimization via an adjoint method. Stanford
of an iterative process, because this feature is not University. Stanford . Dissertation.
implemented in the code at this stage of the development.
[3] Hajela, Prabhat and Chen, Jahau Lewis. 1988.
As expected the evaluated structural wing weight is lower Preliminary weight estimation of conventional and
than predicted by literature (see Table 3.2). joined wings using equivalent beam models. Journal
of Aircraft Vol. 25.
Structural Wing Box mass [4] Kroo, Ilan, Smith, Stephen and Gallman, John.
1991. Aerodynamic and structural studies of joined-
Calculated Value Reference Value Relative Error wing aircraft. Journal of Aircraft Vol. 28.

3864 kg 4544 kg -15,0 % [5] Bindolino, G., Ghiringhelli, G.; Ricci, S.; and
Terraneo, M. 2010. Multilevel Structural Optimization
Table 3.2: Structural wing box mass for the Joined Wing for Preliminary Wing-Box Weight Estimation. Journal
of Aircraft Vol. 47.
This effect is caused by the fact that the assumed loads
[6] Gallman, John W., Smith, Stephen C. and Kroo,
are most likely lower than those applied to the actual
Ilan M. 1993. Optimization of joined-wing aircraft.
model from the paper. Nevertheless the result shows that
the order of magnitude is correct and that the tool is also Journal of Aircraft. Vol 30.
applicable to highly non-planar lifting systems. [7] Samuels, Mary Fairchild. 1982. Structural Weight
Comparison of a Joined Wing and a Conventional
4. CONCLUSION AND OUTLOOK Wing. Journal of Aircraft Vol. 19.
The approach presented in this paper is characterized by [8] Seywald, Klaus. 2011. Wingbox Mass Prediction
its balance between physical accuracy and computational considering Quasi-Static Nonlinear Aeroelasticity.
effort. Although the suitability for conventional cantilever Stockholm, Sweden. Königlich Technische
wings and highly non-planar lifting systems has been Hochschule. Diploma Thesis.
proven, additional validations are necessary. High fidelity
FEM codes will be used to guarantee that the analytical [9] Spohr, Alexander. 2012 Konzeptioneller
model of the code matches with the stress distributions Flügelentwurf für eine Box Wing Konfiguration unter
evaluated by higher order methods. Verwendung eines Wirbelleiterverfahrens. Berlin,
Germany. Deutscher Luft- und Raumfahrtkongress.
As desired the computational effort for computing one
lifting system is very low and iterative processes have [10] Lenz, W. 2005. Vertiefung zur technischen
been avoided as far as possible. Only for optimizing the Mechanik III. Magdeburg, Germany. Otto-von-
cross-sectional shape and in the case of over-determined Guericke-Universität.
lifting systems iteration loops are needed. The fast [11] Miura, Hirokazu, Shyu, Albert T. and Wolkovitch,
convergence behavior of the cross section design makes it Julian. 1988. Parametric weight evaluation of joined
possible to calculate the structural wing box mass of a wings by structural optimization. Journal of Aircraft
cantilever wing with a resolution of 20 sections in under a Vol. 25.
minute.