Master in Aerospace Engineering

Structure design project: Final Report 2024-2025

A350 Central Wing Box (CWB) preliminary

desing study

Authors:
Supervisors:
Aleksander Pryzyblkowski
Jérôme Durand
Andrea Talamonti
Claire Maffre
Swapnil Tole
Eric Bouchet
Shruti Puntambekar
Jérémie Cabot
Teetat Techasrivorakul

June 12, 2025

Contents
1 Introduction 2
1.1 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Wing Box geometry 2

2.1 CAD modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3 Applied loads 4
3.1 Running loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.2 Safety factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

4 Super-stringers design and assessment 5

4.1 Material selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
4.2 Stiffness ratio and preliminary geometry definition . . . . . . . . . . . . . . . . . . . . 6
4.3 Design of composite layups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.4 Assessment against buckling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.4.1 Skin buckling approach and computation . . . . . . . . . . . . . . . . . . . . . 10
4.4.2 Blade buckling approach and computation . . . . . . . . . . . . . . . . . . . . 11
4.4.3 Super stringer buckling approach and computation . . . . . . . . . . . . . . . 12

5 Repairability 13

6 Root joint assessment 15

6.1 Bolting pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
6.2 Composite skin assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
6.3 Crown fitting sizing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
6.4 Cruciform sizing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
6.5 Splice sizing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
6.6 Triform sizing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
6.7 Fatigue life . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
6.8 Thermal induced loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

7 Conclusions 20
7.1 Manufacturability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
7.2 Critical review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
7.3 Further developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

A Technical data sheet 23

A.1 Bolts shear strength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
A.2 Metallic materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
A.3 Fatigue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
A.4 2D Drawing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
A.4.1 Wing Box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
A.4.2 Crown Fitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
A.4.3 Stringer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
A.4.4 Cruciform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
A.4.5 Splice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
A.4.6 Triform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

B Useful formulas 26
B.1 Equivalent bending stiffness for composite multi-regions inhomogeneous sections . . . 26
B.2 Huygens-Steiner theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

1
1 Introduction
This article deals with the preliminary study, analysis, and assessment of the central wing box (CWB)
of the Airbus A350. In a large commercial aircraft, the central wing box (CWB) serves as the pri-
mary load-bearing structure that unites the left and right wing panels with the fuselage. Acting as
the structural backbone of the wing assembly, the CWB is responsible for transmitting aerodynamic
loads, fuel weight, and landing forces between the wings and the main fuselage. Its design must
accommodate bending moments, shear forces, and torsional stresses encountered during flight ma-
neuvers, gust encounters, and ground operations. Furthermore, the central wing box often houses
integral fuel tanks and provides attachment points for critical systems such as landing gear, engines
(in some configurations), and control surfaces.
The importance of the CWB lies in its dual role as both a structural element and a system
integration platform. By efficiently distributing loads, it ensures wing stiffness and strength while
minimizing overall airframe weight, thus contributing directly to the aircraft’s performance, fuel
efficiency, and safety. Advances in materials, such as the adoption of composite laminates and high-
strength aluminium alloys, have further optimized the CWB’s strength-to-weight ratio. As such, the
central wing box remains a focal point of structural innovation in contemporary commercial aircraft
design.

1.1 Objectives
The objective of this work is to present ready conceptual model of the central wing box. This means
all of the parts should be assessed against static failure modes, buckling, fatigue as well as thermal
loads. The key points to be respected are easiness of comprehension in the description of the steps
followed during design, manufacturability of the solutions proposed and respect of the usual safety
criteria imposed to a big commercial aircraft.

2 Wing Box geometry

A fundamental starting point for any structural design process is a clear visualization of the geometry.
Accordingly, a basic CAD model was provided to us in advance, based on which we were required
to further develop a detailed model of the central wing box. The design was to be carried out in
adherence to the following preliminary layout guidelines:
• The front and rear spar should be located at around 17% of the chord starting from the leading
edge and at around 34% of the chord starting from the trailing edge respectively, which is
essential for optimal load distribution and structural integrity .

• Frame pitch which is the longitudinal distance between two fuselage frames should be 685.8
mm. This will ensure consistent support and rigidity.

• For internal structuring, at the upper panel, there should be 5 inter-stringers between two
fuselage frames and at the lower panel, the distance between the two inter-stringers should be
152.4 mm. The upper panel is usually subjected to compression loading and hence is required to
have higher number of stringers to resist buckling loads. As the lower panel experiences tension
loading, it can accommodate a lower number of inter-stringers. This helps in maintaining a
good balance between strength and weight.

• The number of internal ribs in the wing box should be 8. This aids in maintaining the aero-
dynamic profile of the wing as well as supporting the skin. It also acts as a baffle as the wing
box also carries fuel.
To comply with the assumptions, the front and rear spar was adjusted to be located at 17% from
the leading edge and 33% from the trailing edge.

2
2.1 CAD modelling
The CAD modelling process was carried out using the CATIA as well as the Autodesk Fusion 360
software. The CAD began with extracting the basic aerodynamic profile of the air-foil and modelling
the wing box according to the dimensions and the basic wing box was created which is represented
in Fig. 1 and Fig. 2.

Figure 1: Wing box 3D view Figure 2: Basic Geometry with the wing box

The inter-stringers were incorporated by first interpolating the skin thickness at 20% and 80% of
the chord, based on values obtained from self-stiffened panel calculations, as presented in Table 3.
Then the T- section was added on the upper and lower panels which can be seen in Fig. 3 and 4.
The stringer placement of the upper and the lower panel is shown in the Fig. 5 and 6.

Figure 4: Stiffener view: only half of the

Figure 3: Upper and Lower Panel of the U is visible on both sides. These would
central wing box extend to both adjacents stiffeners

Figure 5: Upper Stringer Placement: Figure 6: Lower Stringer Placement:

stringer pitch 137.16mm stringer pitch 152.4mm

It was necessary to maintain the angle between the fuselage and the wing box below 12 degrees
considering a good design practice as shown in Fig 8 and 7.

3
 Figure 7: Front view: 7.9° Figure 8: Rear view: 5.4°

3 Applied loads
Before starting the actual static assessments, one has to know the actual loads applied to the structure
itself and the safety factors to be taken into account. Safety factors are prescribed by the existing
regulations and are dictated by experience; the actual loads, in a preliminary phase, are derived from
a master file on which a finite element analysis is performed. In such an analysis all aerodynamic
forces and forces due to the structure’s characteristics are taken into account with uncertainty factors
that depend on the lack of knowledge of the actual final model.

3.1 Running loads

The running loads are the results obtained in first approximation by the master FEM performed on
the structure. A more accurate model can be later retrieved with more specific knowledge of the
behaviour of the structure. The data provided is the following. The running loads’ unit of measure
is N/mm.

Figure 9: Running loads per flight. On the abscissa: chord percentage starting from the front edge
of the CWB towards the end of the CWB. On the ordinate: running loads in N/mm for a flight
envelope at 1g.

4
 For a positive gust, the upper panel of the central wing box is in compression whilst the lower
panel in tension. The situation is flipped for a negative gust. Because of characteristics attributable
to the shape of both the CWB and the wing that tapers, the loads are higher as we move along the
cord downwards. We here already state that the sizing will be performed solely, for simplicity, at
20% and 80% of the chord, being these meaningful locations for the loading.

3.2 Safety factors

As already mentioned, the running loads are obtained for a flight envelope at 1g. We must take into
account the presence of safety factors. The first one is the limit load safety factor, the limit load
being defined as the maximum load that an aircraft is expected to see at any point in the its service
life. It is such that
LL
= 2.5.
n(= 1)
The second safety factor to be taken into account is the ultimate load, which is defined as the Limit
Load multiplied by a prescribed Safety Factor of 1.5. Any part of the structure of an aircraft must
be able to support the Ultimate Load and, with certain exceptions, be able to do so without failure
for at least 3 seconds (Strength and deformation - CS 25.305 and Section 25.305).

UL
= 1.5.
LL
In the case of negative gusts, not all the running load has to be taken into account, just the 45%
of the ultimate running load: thus in the case of compression of the lower panel and tension of the
upper panel.

Figure 10: Schematic representation of the loads applied to the CWB due to gusts

4 Super-stringers design and assessment

The super-stringer is the crucial reinforcement of both the upper and lower panel of the central wing
box. It is essential for the prevention of buckling of the panels. The Euler approach for composite
materials will be used. The characteristic length of the stiffening beam is the distance between ribs’
pitch. In the preliminary study of the central wing box it constitutes one of the two fundamental
phase of sizing.

5
4.1 Material selection
It is obvious that it is desired that the central wing box is as light as possible while maintaining
acceptable cost of manufacturing. The material of our interest are presented below. By comparing
the maximum compressive stresses in the materials, after having applied a safety factor due to the
dispersion in the behaviour of the composite parts in terms of elastic modulus(KDF ), the IMA
is superior to HTS although it is twice as expensive. However, the increased strength allows for
reduction in weight, in first approximation, by 23%, hence the total cost increase is reduced to 77%.
Considering that the operating cost of the aircraft are much higher than manufacturing, the team is
willing to design the structure with IMA. However, using plies of areal weight of 268 g/m2 this shall
provide a reasonable compromise between the costs and reduction in weight. Moreover, the selected
material has better bearing resistance both in single lap shear and double lap shear.

Elastic modulus E Critical ϵ σ = Eϵcrt KDF Ecompression

IMA 91374 [M pa] 3000 274 [M pa] 1.17 78094 [M pa]
HTS 76054 [M pa] 3500 266 [M pa] 1,20 63378 [M pa]
E
Table 1: Comparison between IMA and HTS materials in compression, the sizing criteria: Ec,IM
c,HT S
A
=
0.23. The IMA shows the chance to reduce the weight of the structure for the sizing criteria of
buckling

Figure 11: Materials available for the design

4.2 Stiffness ratio and preliminary geometry definition

The geometry of interest of super stringer is presented in Fig.12. For the analysis we are going to
take the repeating part of the skin which is one stringer pitch. The stiffness ratio is defined as
E2 × S2
λ= (1)
E1 × S1 + E2 × S2
where the lower script 1 is for the skin and 2 for the blade of the stinger. Limits are imposed for
lambda, in particular λmax,80% = 0.3 and λmax,20% = 0.2. In a first approach, not knowing the values
for the elastic moduli of the two parts of the super-stringer, an exclusively geometrical approach will
be used.

6
Figure 12: Super-stringer visualization: bonding between stringer and skin. Stringer pitch b = 137.16
mm

The ratio λ between the load taken by the stiffener and skin needs to be selected. The initial
assumption in the calculation is that λ assumes its maximum allowable values which are 0.3 at 80%
of the chord and 0.2 at 20% of the chord. Initially, the section is assumed to have a homogeneous
Young’s modulus, equal to the one characterising the IM A material selected, and to work together
as one. This in return allows us to treat it as a rectangular section of length equal to the stringers
pitch. By applying the maximum strain at ultimate load the equivalent skin thickness is obtained.
Knowing the total area of the section necessary to sustain the load, it can be divided into the portion
dedicated to the skin and the one dedicated to the stiffener part, according to λ. However, the section
is subject to certain constraints that must be respected as in Fig.15. It is interesting to notice that
the U section in composite changes the stacking orientation between one stiffener and the adjacent
one: if for one a ply is at 45 degrees, the same ply for the adjacent stiffener is at −45 degrees.

Figure 13: Super-stringer close-up look and main

dimensions definition Figure 14: Splitting of surfaces into blade (S2 )
and equivalent skin (S1 )

Figure 15: Constraints for the geometry of the super-stringer. These relations come from experience,
they are basically thumb-rules

7
 Table 2: Geometric Parameters for first approach in the sizing of the super-stringer

Upper Panel 80% Upper Panel 20% Lower Panel 80% Lower Panel 20%
e 9.58 mm 5.75 mm 10.90 mm 6.54 mm
e1 2.88 mm 2.22 mm 2.22 mm 1.33 mm
e2 1.37 mm 1.73 mm 2.2 mm 2.13 mm
e′ 8.50 mm 6.90 mm 8.9 mm 8.50 mm
h 66.28 mm 48.99 mm 70.39 mm 30.05 mm

Table 2 carries the values obtained fulfilling all the geometrical constraints. Apart from satisfying
geometrical constraints, the thickness must be divisible by 0.25 mm (the thickness of one ply of
IM A) in order to make it manufacturable. This is way some values will be finally rounded in the
last iteration comprising the layup of the parts themselves. As can be seen always from table 2, the
computation was performed only for location at 80% and 20% of the cord both for upper and lower
panel, as already stated previously.

4.3 Design of composite layups

The assumed composition of the stacking was 50% for plies at 0 degrees, 20% for plies at +45
degrees, 20% for plies at −45 degrees and 10% for plies at 90 degrees. The obtained thicknesses
could not always be divided into this composition while keeping the stacking symmetric that is
why the contents were slightly modified which resulted in Young’s Modulus of the skin and stiffener
slightly different from the one assumed. The final values defining the geometry are listed below
(table 3). The stacking sequences are also displayed. From these, using an ad hoc tool the A, B, D
matrices(membrane, coupling, bending behaviours) of the layup were computed and the resulting
elastic moduli were used to recompute the λ coefficients, which still respect the boundary conditions
imposed.

80 % upper panel 20 % upper panel 80 % lower panel 20% lower panel

e 10 mm 6 mm 11 mm 6.50 mm
e1 3 mm 1.75 mm 2.25 mm 2.25 mm
e2 3 mm 1.75 2.25 2.25 mm
e’ 12 mm 7.25 mm 9 mm 8.5 mm
h 42 47.5 mm 48.4 mm 30 mm

Table 3: Geometric Parameters for last approach in the sizing of the super-stringer. These values
are obtained taking into account the actual

Figure 16: Stacking sequence of the skin 80 % upper panel

Figure 17: Stacking sequence of the stiffener 80% upper panel

8
Figure 18: Stacking sequence of the skin 20% upper panel(the only non-symmetric full stacking)

Figure 19: Stacking sequence of the stiffener 20% upper panel

Figure 20: Stacking sequence of the skin 80% lower panel

Figure 21: Stacking sequence of the stiffener 80% lower panel

Figure 22: Stacking sequence of the skin 20% lower panel

9
 Figure 23: Stacking sequence of the stiffener 20% lower panel

Having the all of the stacking sequences the Young modulus of skins and stiffeners can be recal-
culated to check if the condition of λ ≤ 30% for upper panel and λ ≤ 20% for lower one is satisfied
in all of the domain(eq. 1). At the interface between e1 and e − e1 and between e2 and e1 two 0
degrees plies come into contact to guarantee the co-curing compatibility.

80 % upper panel 20 % upper panel 80 % lower panel 20% lower panel

λ 26% 28% 20% 20%

4.4 Assessment against buckling

After the dimensions and properties of the super-stringer have been defined, buckling analysis is per-
formed. In accordance with the Composite Margin Policy, the Buckling Analysis must be performed
with the B-value of the modulus in compression. The process is separated into three sections: skin,
blade, and globally.

4.4.1 Skin buckling approach and computation

Since the dimensions of the skin differ depending on the location on the wing box. The calculations
are split into 4 cases: at the 80% and 20% chord of the upper and lower panel. For the skin, the
four edges are simply supported in pure compression in which the buckling flow (Ncrit ) of the skin is
computed by the following equation:
1 π p
Ncrit = × kc × ( )2 × D11 × D12 (2)
KDF b
where b is the stringer pitch and KDF is the knockdown factor. D11 and D22 are constants of the
bending matrix D. Other variables used in the previous equation can be computed by the following:

kc = h(ᾱ) + qβ (3)
r
a 4 D22
ᾱ = (4)
b D11
D12 + 2D33
β= √ (5)
D D
(11 22
( ᾱ1 )2 + ᾱ2 if ᾱ ≤ 1
h(ᾱ) = (6)
2 else
q=2

where a is the rib pitch, kc is the buckling coefficient, ᾱ is the dimensionless aspect ratio factor, D22
and D33 are terms in bending matrix, and q is a constant multiplier.

From the plies layup present in section 4.3, matrix D constants can be derived for the skin at the
four locations which are presented in Table 4 below.

10
 80 % upper 20 % upper 80 % lower 20% lower
D11 (kN mm) 8844.1 2109.5 13166.0 3027.7
D22 (kN mm) 2700.9 469.2 2961 699.8
D12 (kN mm) 1395.9 259.0 1478.0 340.9
D33 (kN mm) 1538.5 289.8 1668.0 384.7

Table 4: Stiffness matrix D of the skin at four locations

Knockdown factor for the composite material (IMA/M21E) is taken as 1.17. For the upper panel,
the stringer pitch is 137.16 mm and rib pitch is 750 mm. The results are shown in Table 5 below.

80 % upper 20 % upper 80 % lower 20% lower

b (mm) 137.16 137.16 152.4 152.4
a (mm) 750 750 750 750
kc 3.83 3.69 3.54 3.53
ᾱ 4.065 3.755 3.389 3.412
β 0.9152 0.8430 0.7710 0.7628
h 2 2 2 2
Ncrit (kN/mm) 8.394 1.644 8.032 1.864
Fcrit (kN ) 1151.3 225.5 1224.1 284.1
Applied Load (kN ) 363.4 221.8 459.6 273.3
RF 3.168 1.017 2.664 1.039

Table 5: Input parameters, buckling flow and reserve factor of the skin

Applied loads used for the reserve factor are at the ultimate load. It can be seen that the reserve
factors for all four skin configurations are greater than one. This indicates that, when the wing box
is subjected to the ultimate load, the skin section will not experience buckling.

4.4.2 Blade buckling approach and computation

Similar to the skin, the critical buckling load is computed for the blade in four different cases based
on location. However, the blade has a different support where three edges are simply supported with
one free edge under pure compression. The buckling flow is computed with the following equation:
1 π p
Ncrit = × kc × ( )2 × D11 × D12 (7)
KDF b
where b is blade height. Different support configuration changes the equations for computing the
parameters used in the previous equation which are the following:

kc = h(ᾱ) + qβ + rη (8)
r
a 4 D22
ᾱ = (9)
b D11
D12 + 2D33
β= √ (10)
D11 D22
2D33
η=√ (11)
D11 D22
1
h(ᾱ) = ( )2 (12)
ᾱ
q=0 (13)
r = 0.608

11
where a is the blade height, and η is the shear interaction term, and r is a constant multiplier equal
to 0.608. KDF is taken as 1.17, identical to the skin.

D matrix constants can be calculated from the proposed plies layup in section 4.3 The values are
presented in Table 6 below.

80 % upper 20 % upper 80 % lower 20% lower

D11 (kN mm) 15689.0 3618.0 7774.0 5066.0
D22 (kN mm) 4535.1 557.0 1172.0 976.0
D12 (kN mm) 2274.9 371.0 753.0 649.0
D33 (kN mm) 2521.4 420.0 857.0 722.0

Table 6: Stiffness matrix D of the blade at four locations

The input parameters and results are as shown in Table 7 below.

80 % upper 20 % upper 80 % lower 20% lower

b (mm) 42 47.5 48.4 30
a (mm) 750 750 750 750
kc 0.369 0.370 0.356 0.398
ᾱ 13.09 9.90 9.66 16.54
h 0.01021 0.00583 0.01071 0.00366
η 0.5978 0.5917 0.5678 0.6494
Ncrit (kN/mm) 14.9 1.97 3.87 8.30
Fcrit (kN ) 625.7 93.28 187.3 249.1
Applied Load (kN ) 150.9 86.86 111.9 69.61
RF 4.146 1.075 1.673 3.579

Table 7: Input parameters, buckling flow and reserve factor of the blade

The applied loads in the table are derived from the ultimate load on the wing box. The reserve
factors of the blades at each location meet the required threshold of one, indicating that even under
ultimate load, the blade sections of the super-stringer will not buckle.

4.4.3 Super stringer buckling approach and computation

The global super-stringer was assessed against buckling as a composite beam of equivalent bending
stiffness (see equation in appendix B). The critical loading, assuming that the effective skin width
that works in buckling is the total stringer pitch, becomes:

1 π 2 × (EI)eq
Fcrt = × (14)
KDF (k × Lcharacteristic )2

where k = 0.9 is due to the boundary condition at ribs and allows to define an equivalent length
k × Lcharacterisitc and Lcharacteristic = 750 mm is the length of the super-stringer, hence the rib pitch.
These critical values of buckling are compared to the total loads applied to the super-stringer due
to the running loads multiplied, by the safety factors. The lower panel stringer pitch is blow = 152, 4
mm, for upper panel bup = 137, 16 mm.

12
 Table 8: Buckling parameters for upper/lower panels at 20% and 80% chords

80% chord, upper panel 20% chord, upper panel

Symbol Value Symbol Value

I1 (skin I) 1.1430 × 104 mm4 I1 2.4689 × 103 mm4
I2 (stiffener I) 1.5300 × 105 mm4 I2 3.7803 × 104 mm4
E1 (skin E) 9.1374 × 104 MPa E1 9.1124 × 104 MPa
E2 (stiffener E) 8.5193 × 104 MPa E2 8.9108 × 104 MPa
S1 1371.6 mm2 S1 822.96 mm2
S2 506.6 mm2 S2 290.20 mm2
e 10.00 mm e 6.00 mm
L 60.00 mm L 39.70 mm
zG 14 mm zG 8.8594 mm
(EI)eq 9.3207 × 109 N mm2 (EI)eq 2.3447 × 1010 N mm2
k 0.90 k 0.90
Fcrit 1.7256 × 106 N Fcrit 4.3411 × 105 N
Fappl 5.1435 × 105 N Fappl 3.0861 × 105 N
SF = Fcrit /Fappl 3.36 SF = Fcrit /Fappl 1.41

80% chord, lower panel 20% chord lower panel

Symbol Value Symbol Value

I1 1.6904 × 104 mm4 I1 3488 mm4
I2 8.4824 × 104 mm4 I2 19 221 mm4
E1 9.7526 × 104 MPa E1 92 992 MPa
E2 9.1498 × 104 MPa E2 91 498 MPa
S1 1676.4 mm2 S1 1676.4 mm2
S2 435.24 mm2 S2 270.45 mm2
e 11.00 mm e 11 mm
L 48.4 mm L 30 mm
zG 11.3134 mm zG 8.3117 mm
(EI)eq 6.4222 × 1010 N mm2 (EI)eq 2.0076 × 1010 N mm2
k 0.90 k 0.90
Fcrit 1.1890 × 106 N Fcrit 3.7169 × 106 N
Fappl 5.7150 × 105 N Fappl 2.7329 × 105 N
SF = Fcrit /Fappl 2.08 SF = Fcrit /Fappl 1.36

5 Repairability
Knowing how a repair should be performed is crucial in case of an accident. The materials used to
repair a part are the same ones used in the original part. The loads to consider are the running loads
applied to the structure to be repaired multiplied by the safety factors. The bolting pattern has to
be specified(number of columns and rows of bolts). In the case of composite, the sizing criteria are
the bearing of the bolts on the structure and the failure of the plies(Yamada Sun criteria of failure).
The localized damage area must be able to be restored using mechanically fastened patch. The
load on the fastener is calculated to determine whether the fasteners would fail or not. The material
of the fastener is chosen to be titanium since it provides the highest toughness to weight ratio
compared to steel alloy and inconel. The diameter of the fasteners are later chosen based on the
buckling criteria. It is assumed that the skin-panel configuration is modelled as a single lap shear
joint. Failure modes are split into three criteria: bearing, fastener shear and fastener hole.

13
 First, the fastener load is computed with:

Fx/y = CLT P × CLT N × max|Nx/y | × k × d (15)

where CLT P is the percentage of load in the repair and CLT N is the percentage of load at the first
fastener row. k is the ratio between fastener pitch to diameter and d is the fastener diameter. CLT P
is taken as 0.5 while CLT N is taken as 0.3. There is a requirement for k to be between 4 and 5.

The safety factor of the bearing is calculated with:

σallbear
RFbearing = (16)
σbea
Fx/y
σbea = (17)
d×t
where d is the fastener hole diameter, t is the thickness of the skin, σallbea is the allowable bearing
stress which is determined from the composite material property, and σbea is the bearing stress.

The safety factor of the fastener shear is computed by:

Psu
RFshear = (18)
Fx/y
where Psu is the ultimate shear strength of the fastener which depends on the material and diameter
of the fastener. The table of Psu can be seen in Appendix B.

Lastly, the safety factor of the fastener hole is computed with:

1
RFY S = q (19)
στxymax 2
( σxmax
R1
)2 + ( S12
)
1 Nk
σmax = ( + Km σbea ) (20)
Kt t(k − 1)
Several configurations of k and d are experimented to achieve the reserve factor of one for three
buckling criteria. Fastener diameter code 3A and fastener pitch-to-diameter of 5 are chosen for all
location. The analysis results of the selected configuration are shown in Table 9 below.
Constant parameters:

k=5
d = 5.556 mm [diameter code 3A]
CLT P = 0.5
CLT N = 0.3
σallbea = 500 MPa
Psu = 16000 N

80 % upper 20 % upper 80 % lower 20% lower

Nx (N/mm) 3750 2250 3750 2250
Fx/y (N ) 15627.0 9376.2 15627.0 9376.2
σbea (N/mm2 ) 281.25 281.25 255.68 259.62
RFbea 1.778 1.778 1.956 1.926
RFshear 1.024 1.707 1.024 1.706
RFY S 1.00 1.01 1.00 1.22

Table 9: Fastener loading criteria results

14
 The safety factors at every location are either greater than or equal to one for all three repairability
criteria, suggesting that the panel would not buckle in case of a fastening patch repair. Since the
repairability requirements are satisfied with reserve factors close to one, an SFY S value near one is
not a concern, and the current configuration is deemed acceptable.

6 Root joint assessment

The root joint in the central wing box (CWB) is the primary structural interface where each wing
panel attaches to the fuselage. It typically comprises a series of bolted or riveted fittings—often rein-
forced by doubler plates or splice beams—that transfer bending moments, shear loads and torsional
forces from the wing into the fuselage structure. Because the root joint must carry the full spectrum
of aerodynamic and inertial loads encountered in flight, its design balances stiffness, strength and
fatigue resistance, while minimizing added weight. Proper detailing of the root joint is therefore
critical to ensure that the wing-to-body load path remains secure throughout the aircraft’s service
life. Here, a picture of how the load are redistributed between skin, cruciform and crown fitting at
upper panel and between skin, triform and splice at lower panel. The analysis is computed at 80%
and 20%. The factor of 45% for the negative gust at ultimate load for tension of upper panel and
compression of the lower panel are taken into account in the following analysis.

Figure 24: Root joint redistribution of loads for both upper and lower panel

σb σyc σyt
670 MPa 450 MPa 275 MPa

Table 10: Aluminium T2024 T351 properties

The strength for the aluminium is taken at ultimate load.

6.1 Bolting pattern

An initial assumption was made to start the assessment of all the structures in the following subsec-
tions. In particular, it was verified that within the crown fitting internal surface a solution of 3 rows

15
and 5 columns was viable(the space is limited as the crown fitting fits between two stiffeners’ pitch:
from this dimension, a play of 3 mm has to be removed from both sides as well as the thickness
of the edges of the crown fitting, which again is 3 mm). The bolting pattern in the crown fitting
determines the number and the size of the fasteners, as well as the repartitioning of loads and the
bearing load that has to be sustained by the composite skin. The first candidate for the material
for the crown fitting, splice, cruciform and triform is T2024 T351, because the alloys from this series
perform the best against fatigue while maintaining good mechanical properties.

6.2 Composite skin assessment

The composite skin at the root joint connects with two metallic parts to form a double shear lap
joint. If the percentage rule is respected also for the root joint, the skin layup will not affect the
elastic modulus of the skin itself. Here, the connection has to be carefully designed, as the skin has
to support both the stresses applied to the central wing box and the loads coming from the bearing
stresses of the fasteners. The safety factors for the shearing of the bolts and the maximum bearing
stresses that the skin can withstand are therefore computed. For the skin, the critical sizing criteria
is the bearing stress, as the original skin thickness iteration is the one coming from the equivalent
section, which by definition sustains the running load at ultimate load. For the composite material
of choice, in double shear lap joints:

σbearing = 600M pa.

The fasteners’ pitch and edge distances are computed to later assess cruciform, crown fitting, splice
and triform fro assembly. This is because the space in between stringers is limited as already men-
tioned.

Table 11: Structural assessment of composite skin at root joint

Parameter 80% Upper 20% Upper 80% Lower 20% Lower

Geometric Parameters
e2 [mm] 23.17 23.17 23.17 23.17
e1 [mm] 28.7 28.7 28.7 28.7
Columns 5 5 5 5
Rows 3 3 3 3
at [mm] 5 5 5 5
Bolt pitch 33.41 35.80 42.39 42.78
[mm]
Thickness 14 7 12 7
[mm]
Stringer pitch 137.16 137.16 152.4 152.4
[mm]
Fastener 7.9 7.9 7.9 7.9
diameter
[mm]
α 0.25 0.25 0.25 0.25
Load Analysis
Load (total) 514,350 308,610 571,500 342,900
[N]
Load per row 171,450 102,870 190,500 114,300
[N]
Load per bolt 4,286 3,429 6,350 4,771
[N]
σbolt [MPa] 655.8 333.5 728.7 437.2
Bearing Stress Analysis
Bearing 387.6 465.1 592.4 516.7
stress [MPa]
Safety Factors
Bolt Safety 1.51 2.52 1.36 2.27
Factor
Bearing 1.55 1.23 1.18 1.16
Safety Factor

16
6.3 Crown fitting sizing
The bearing force was taken as αF where α is the fraction of the total load per row F , taken by the
most loaded fastener. Knowing the mechanical properties of the alloy, as well as the width of the
fitting (with the clearances taken into calculation), they were used to size the thickness of the crown
fitting. To our surprise it was not the bearing that was decisive but compression. The outcome of
the calculation is the following geometry. The 2D drawing can found in the appendix 34.

Position Rows Columns Width [mm] thickness of the base [mm]

80% Upper Panel 3 5 119.2 6.4
20% Upper Panel 3 5 123.7 3.7

Table 12: Crown fitting dimensions

Figure 25: Crown Fitting CAD

6.4 Cruciform sizing

The same approach was used as in the case of crown fitting with the exception of clearances as here
they do not apply. The cruciform is a unique part which hosts all the internal and external crown
fittings. The name carries the shape of the part itself, as its section is similar to a cross. Strict
tolerances on curvature and geometry must be respected. The results are presented in the table
below.
Position rows columns Width thickness of the base [mm]
80% Upper Panel 3 5 137.16 8,3
20% Upper Panel 3 5 137.16 5,5

Table 13: Cruciform dimensions

Figure 26: CAD Model of the Cruciform

17
Figure 27: Assembly of cruciform(in yellow), external, internal crown fittings and composite skin:
the cruciform is visible in the middle as main connection element

6.5 Splice sizing

The driving sizing criteria for the splice at 20% of the chord is tension, compression at 80%. The
negative gust load reduction was considered for the compression loads giving following results.

position rows columns Width[mm] thickness of the base[mm]

80% Lower Panel 3 5 152.4 3.3
20% Lower Panel 3 5 152.4 3.3

Table 14: Splice dimensions

Figure 28: CAD Model of the Splice

6.6 Triform sizing

Both at 20% and 80% the driving sizing criteria is again tension, despite the compression modulus
being lower (the -45% due to negative gusts compensates).

position rows columns Width[mm] thickness of the base [mm]

80% Lower Panel 3 5 137,4 5.6
20% Lower Panel 3 5 137,9 3.4

Table 15: Triform dimensions

18
 Figure 29: CAD Model of the Triform

6.7 Fatigue life

In this section, the fatigue life of the structure is assessed only for the lower junction, as the upper
one mainly works in compression, whilst the lower one mainly works in tension. The standard design
life goal value of 28800 should be multiplied by a safety factor of 5 giving design life goal. The ratio
between the cycles of useful life and DLG constitutes the safety factor, which should result greater
than 1). Another approach is to simply compute the ratio Ncycles /DLG and aim for a safety factor
of SF = 5. To compute such statistics, the running load at 1g flight are taken at the 20% and 80%
of the chord. The equivalent load in Newtons is obtained by multiplying by the width of the part.
Dividing by the area the stress is computed. Upon this value, another factor is taken into account,
the Cs = 1.5 or fatigue spectrum coefficient, which allows to convert a complex fatigue spectrum in
an equivalent one for which only one equivalent load is taken into account.
σequiv = cs σn=1g .
To later compute the number of allowable cycles Ncycles the following formula is used
AF I 4.5
Ncycles = 105 [ ] ,
σequiv
where AF I(Airbus fatigue index) reads
Fn · M · E · C
AFI =
Kt
where C = 510M pa is a reference load for bolted joints, Kt depends on the geometry(see appendix
A, Fn on the technologies(faster mounted with interference, shot-peening, surface treatments, cold
working of the whole et cetera), M depends on the material of the fastener (M = 1.2 for T i fasteners)
and E = (1/r)0.08 the scale factor with r = hole radius. It was computed AF I = 116 M pa.

Table 16: Chord Analysis Comparison: Triform and splice fatigue life assessment

Part Thickness [mm] Load [N ] Area [mm2 ] Eq. Stress[M pa] Ncycles SFold
Triform 80% 8.9 137160 1534.30 89 146734 5.1
Triform 20% 5,4 82296 320.96 89 146372 5.4
Splice 80% 6 91440 1032.40 89 153017 5.3
Splice 20% 3,6 54864 618.77 88 152210 5.4

In the end, fatigue is actually the sizing criteria for some of the parts. Compared to subsection
6.1, indeed for triform and splice their new dimensions are highlighted in yellow. These new values
now actually satisfy also the static requirements. Given that the thermal loads are only assessed for
in the upper junction, the values for the bottom section will not change any further.

19
6.8 Thermal induced loads
Fittings and skin are made from different materials, which can lead to thermal loads when temper-
atures rise due to their differing thermal expansions. In this section, temperatures can range from
Tmin = −50°C to Tmax =70°C, so these thermal effects must be considered during the preliminary
sizing phase. While the case involving two plates is straightforward, the analysis becomes more com-
plex when three plates are assembled. Therefore, for the purposes of preliminary sizing, the top and
bottom fitting sections are treated as a single metallic component. Then the force due to thermal
loading is given by:
E1 S1 · E2 S2
F = ∆α∆T (21)
E1 S1 + E2 S2
where ∆T = T − Tref , Tref = 23°C and ∆α is the difference between the expansion coefficients of the
materials. Additionally, it is assumed that the the thermal load is not evenly distributed between
the plates. Thus, each plate feels different percentage of the total thermal load. From its two cases
T = −50°C, T = 70°Cthe more severe is used in assessing the fittings against bearing by adding its
absolute value to the running load which ignores the possibility of opposite senses of the thermal
and running loads yet assures that the current iteration is suitable to withstand worst case scenario.
The results of computation are summarized below:

Table 17: Thermal load fractions, forces and recomputed safety factors at root-joint
locations
Location Load fraction F @ –50 ◦ C [N] F @ 70 ◦ C [N] SFold SFnew
80% Upper panel crown fitting 0.40 −29 046 45 673 1.92 1.24
20% Upper panel crown fitting 0.40 −16 618 25 810 1.84 1.23
80% Upper panel cruciform 0.60 −44 109 68 510 2.96 1.52
20% Upper panel cruciform 0.60 −24 926 38 715 1.46 1.05

In all of the cases, the added thermal load does not require another iteration to recompute the
thicknesses of the fittings, as all of the recomputed safety factors are over 1. Finally, it is necessary
to check if the bearing loads induced from the thermal loading will change the size of the skin or
bolts or both. The results are presented in the table below.

Table 18: Comparison of safety factors for the upper panel skin and bolts before and
after addition of thermal loading
skin skin bolts bolts
Location SFold SFnew SFold SFnew
80% Upper panel skin 1.55 1.23 1.51 1.24
20% Upper panel skin 1.27 1.07 2.52 2.1

Since all the safety factors are above 1 there is no need to resize neither the skin nor bolts . Thus,
concluding the initial sizing of the CWB.

7 Conclusions
The design of the Composite Wing Box (CWB) presented in this work has aimed to balance structural
efficiency, manufacturability, and repairability within the operational and regulatory constraints of a
regional aircraft platform. A progressive design logic has been followed, from the preliminary struc-
tural layout to detailed considerations of materials, joints and damage tolerance. Nonetheless, the
development process has highlighted areas where additional refinements or extensions are both nec-
essary and feasible, either to improve performance or to better support certification and operational
reliability. These aspects are discussed in the following critical review and further developments.

20
7.1 Manufacturability
The solutions proposed for the sizing are readily implementable. The thicknesses for composite parts
are multiple of the thickness of a single ply and for the lamination of the super-stringer a co-curing
approach can be used where the U sections and the vertical body of the blades are added to the
curing process towards the end of the process of cure of the actual skin. This requires an autoclave
big enough to host such a structure. The metallic parts bolting pattern satisfies the edge margin
policy, especially in the crown fitting region where space is limited. The geometry of the whole
structure has been well clearly defined in terms of main dimensions and sizes and distances between
parts(section 2). However, the merging of plies between different regions is not tackled, nor the space
to actually mount the fasteners or use the adequate equipment is validated.

7.2 Critical review

The analytical framework laid out in this report provides a clear, conservative baseline for the CWB’s
initial sizing, but several limitations and pending tasks should be noted:

• Design loop: As it was clear from the report, some properties and values changed along
the way due to different driving sizing criteria. Loops in the design process are unavoidable,
but at least for fatigue a first computation could have been made to later assess the bolting
pattern for the reference parts. Thermal loads could have been taken into account right from
the beginning, but that’s just a matter of approach that in the end does not change the sizing
that much, mostly the readability of the report itself.

• Analytical simplifications: Reliance on Euler buckling formulas, uniform stiffness assump-

tions and first-order laminate theory neglects three, dimensional stress gradients, local stress
concentrations (e.g., bolt holes), geometric imperfections and non-linear skin-stringer interac-
tion. These effects must be captured in a high-fidelity finite-element model—with composite,
capable elements, detailed contact definitions at joints and global wing assembly boundary con-
ditions—prior to final sizing. The same applies for thermal loading case as it used simplification
that might turn too far to render true results, hence it is neccessary to verify it numerically.

• Expanded load spectrum: Only symmetric limit and ultimate gust loads have been ad-
dressed at the 20% and 80% chord stations. Asymmetric maneuvers (roll, yaw), landing-gear
drop and taxi loads, panel rip-off scenarios, and aeroelastic (flutter) or impact loads may govern
local stiffening or root-joint reinforcement. Each case must be quantified and incorporated in
the load matrix.

• Fastener bearing and joint detailing: The chosen bolting patterns respect generic edge
margin rules, but detailed bearing, shear out and clamp up analyses remain to be performed
for every fitting (crown, cruciform, splice, triform). Subsequent phases will compute bearing
capacities, washer effects, preload dispersion and potential need for local doublers or filler
materials. Moreover, the direction of the thermal load was ignored which if taken into account
could decrease the net load thus allowing for optimization of the thickness. Also, other bolting
solutions could be analysed to see how they could affect the sizing criteria, hence the thicknesses
of the parts and the weight of the structure. Moreover, the enhanced stacking sequence at the
root joint should be computed to determine the resistance to failure according to the Yamada
Sun criteria even in the presence of thermal loads.

• Environmental durability and lightning protection: The CWB’s composite panels and
metallic fittings must withstand combined effects of moisture absorption, fuel exposure, UV
ageing and galvanic corrosion at composite-to-metal interfaces. Moreover, modern airliners
mandate integrated lightning-strike protection (e.g. embedded metal meshes or conductive foils)

21
 to safely dissipate strike currents without damaging the laminate. The impact of these conduc-
tive layers on local stiffness, fatigue life and joint detailing has not yet been considered, and
should be incorporated into both the material specification and joint designs.

In conclusion, this study defines a robust starting point for geometry, lay-up and preliminary load
checks, but must be augmented by detailed FEM, joint level testing (bearing, tear-out, preload),
probabilistic manufacturing assessments, and a full load spectrum fatigue / damage tolerance analysis
before the CWB concept can be finalized for production.

7.3 Further developments

Some key design points are yet missing. The computation of the ribs’ loads and the sizing of ribs,
as well as the computation for every super-stringer in between the locations at 20% and 80%, before
and after, and the contribution to the root joint sizing from the cabine’s and fuel-tank’s pressure.
Moreover, optimization is not performed to obtain the best possible solution in terms of the ratio
of rigidity to weight. The front and rear spars are not analysed in this preliminary study, and no
detailed numerical simulation is conducted. A section dedicated to in-service life should be added
to periodically and systematically assess the operating state of the structure, especially considering
the difficulty of certification of composite parts and their continued airworthiness(damage tolerance).
In addition, the assembly process should be clearly defined in terms of the phases and steps to be
followed in order to obtain the best possible result at the end of the assembly chain process. These
are some possible further adjustments that would deepen the level of the preliminary study.

22
A Technical data sheet
A.1 Bolts shear strength

Figure 30: Bolts shear strength for double lap shear. In case on single lap joint, the values are simply
half

A.2 Metallic materials

Figure 31: Properties of some metallic materials

23
A.3 Fatigue

Figure 32: Kt computation

A.4 2D Drawing
A.4.1 Wing Box

Figure 33: Wing box Drawing

24
A.4.2 Crown Fitting

Figure 34: Crown Fitting Drawing

A.4.3 Stringer

Figure 35: Stringer Drawing

A.4.4 Cruciform

Figure 36: Cruciform Drawing

25
A.4.5 Splice

Figure 37: Splice Drawing

A.4.6 Triform

Figure 38: Triform Drawing

B Useful formulas
B.1 Equivalent bending stiffness for composite multi-regions inhomoge-
neous sections
X X
(EI)eq = Ei Ii + (zi − zG )2 Ei Si .
i i
In this case i = 1, 2 for skin and blade respectively. The zi values are obtained as geometrical
barycentres of each of the regions defining the global geometry, Si are the respective areas and Ei
the respective moduli. The zg value is the global barycentre of the geometry computed as a weighed
average on moduli and areas: P
zi Ei Si
zG = Pi .
i Ei Si

B.2 Huygens-Steiner theorem

The moment of inertia about an axis a, parallel to another axis c passing through the centre of mass,
is obtained by adding to the moment of inertia about axis c the product of the region’s area and the
square of the distance between axes c and a.

Iz = IG + Sd2 (22)
where:

26
• Iz is the moment of inertia about axis z (or axis a)

• IG is the moment of inertia about the axis passing through the geometrical barycentre

• S is the area of the section of the geometrical region

• d is the distance between the two parallel axes

27