Documente Academic
Documente Profesional
Documente Cultură
5
E being Youngs modulus of the material and n
Poissons ratio. The general form of the solution of
differential equation (4) is
wz w
4
i1
A
i
exp
z
l
i
6
where 1=l
i
are the roots [ C of the characteristic
equation associated with equation (4) and w is the
particular solution that can be derived by setting
derivatives to zero in equation (4) according to
w 1
n
2
DpR
2
Et
1
n
2
w
m
7
For the simple case of a single frame, solution (6)
must satisfy the boundary conditions
wj
z0
0, ? rigid frame,
w
f
, flexible frame,
lim
z??
w ww
dw
dz
z0
0, lim
z??
dw
dz
0
8
In the case of four real roots
*
, the solution assumes the
form
w
w
m
1
n
2
1
1 w
f
=w
l
2
l
1
l
2
exp
z
l
2
l
1
exp
z
l
1
9
where 1=l
1
and 1=l
2
are given by the relationships
1
l
1
R Dp
4D
1
1
4E
2
t
4
31 n
2
Dp
2
R
4
1
l
2
R Dp
4D
1
1
4E
2
t
4
31 n
2
Dp
2
R
4
10
For the case of multiple equally spaced frames,
equation (4) yields similar results by adopting suitable
boundary conditions; such results have been extended to
the more realistic case of a stiffened shell by using a
modied equation proposed in reference [4], which is
based on the concept of spreading the stringer section
area along the fuselage perimeter, resulting in an
unstiffened shell of equivalent thickness t
eq
.
4 FINITE ELEMENT MODELS OF THE
FUSELAGE BARREL
The geometrical model of the fuselage barrel has been
built by using the parametric modelling tool included in
the commercial software ANSYS 5.7; the same software
has been used for the meshing, solution, and post-
processing phases.
Since the real structure is mainly constituted by thin-
walled components, shell elements have been used
extensively in the meshing phase together with beam
elements to model stiffening members (low-resolution
meshes of frames and stringers). Shell element number
63 has been selected [13], which is a four-node plane
element having six degrees of freedom per node and
capable of simulating bending and membrane beha-
viour.
As far as meshing strategies are concerned, an
analysis region has been selected that is enclosed
between two boundary zones where border effects due
to loads and constraints disappear. For such main zones
a structured mesh is adopted with a ner grid for the
analysis region and a coarser grid for the boundary
zones; between these two, transition regions exist for
which it is convenient to use a free meshing approach
(an example of such meshing strategy is given in Fig. 5).
4.1 Model for mass load analyses
For these kinds of analysis, the fuselage is assumed to
have a clamped section, corresponding to the wing rear
spar, and to be subjected to constant mass load per unit
length on both the upper-deck and the lower-deck oor
beams [10, 14].
The analysis barrel has a length of seven bays and is
contained between two boundary zones: a rst zone,
composed of 12 bays, is on the constraint side, while the
second zone extends for 19 bays towards the tail cone
(Fig. 5). The presence of the stringers is accounted for by
means of an equivalent shell thickness. As far as the
modelling of the frames and of the oor beams is
concerned, in the boundary zones they are represented
by means of beam elements, while in the analysis region
they are modelled according to the example shown in
Fig. 6. Moreover, in the analysis region the frames and
the skin are connected by using node-to-node rigid
elements (CERIG) [13].
The main data used for the analyses carried out to
study mass loads effects are summarized in Table 1.
They refer to the outputs of the optimization code
relevant to a one-g load index of 200 N/mm, for the
design of stiffened panels with hat stringers at a spacing
*
The case of real roots is relevant to fuselages characterized by small
values of the ratio t=R
4
for typical pressurization levels (5060 kPa).
L BONI AND D FANTERIA 320
Proc. Instn Mech. Engrs Vol. 218 Part G: J. Aerospace Engineering G05603 # IMechE 2004
of 200 mm, and to mass loads equivalent to resultant
section shear loads of 10 000 and 5000 N for upper-deck
and lower-deck oor beams respectively. The dimen-
sions refer to a Z geometry for the frame cross-section
and an I geometry for the cross-section of both oor
beams.
All components are made of aluminium alloy 2024
with Youngs modulus equal to 72 000 MPa and
Poissons ratio equal to 0.32.
4.2 Models for pressure load analyses
The effects of the pressure load have been studied by
means of models of increasing geometrical complexity,
in order to evaluate the contributions to the stressstrain
eld of the different fuselage components [10, 15].
The models have been developed using the hypothesis
of cylindrical symmetry, so that only a cylindrical sector
has been modelled of nite extension along the long-
itudinal axis of the fuselage (Fig. 7). To respect the
symmetry and equilibrium conditions, the cylindrical
sector must be properly constrained and loaded at its
edges: those parallel to the fuselage axis (side edges) are
constrained against circumferential displacement while
on the edges orthogonal to the axis (normal edges)
different conditions are applied; one is restrained against
axial displacement and to the other a tensile load per
Fig. 5 Meshing strategies
Fig. 6 Frame meshing example
Table 1 Main dimensions used for the FEAs
Value (mm)
Component
Mass load
analyses
Pressure
load
analyses
Fuselage diameter 5640
Spacing of frames (bay length) 500
Equivalent skin thickness 3.6
Skin thickness 2
Spacing of stringers 200
Web height of hat stringers 56
Free ange width of hat stringers 16.8
Connection ange width of hat stringers 56.6
Thickness of hat stringers 1.97
Tear-strap width 47
Tear-strap thickness 0.64
Frame web height 165
Frame ange width 36.5
Frame thickness 1.5
Web height of upper-deck oor beam 240
Flange width of upper-deck oor beam 156
Thickness of upper-deck oor beam 2.5
Web height of lower-deck oor beam 180
Flange width of lower-deck oor beam 60
Thickness of lower-deck oor beam 1.5
Fig. 7 Fuselage sector meshing example
ANALYTICAL METHOD FOR FUSELAGE DESIGN BY MEANS OF FE ANALYSES 321
G05603 # IMechE 2004 Proc. Instn Mech. Engrs Vol. 218 Part G: J. Aerospace Engineering
unit length is applied in order to establish the
equilibrium of the structure when subjected to a pressure
load. The value of such an axial load corresponds to that
given by the analytical model for the case of a
circumferentially non-stiffened shell (i.e. in absence of
frames).
The sector extension in the circumferential direction
has been determined so that the effect of the constraints
on the side edges is almost suppressed in the central
portion of the arc, where the analysis section is located.
As far as the extension in the longitudinal direction is
concerned, two cases have been analysed: for the models
characterized by a single central frame, the axial length
of the sector has been selected in such a way as to allow
for the effect of the frame to be suppressed at the normal
edges; for the models with multiple frames, a number of
bays in the longitudinal direction have been modelled so
that the stressstrain distribution in the central bay is
not affected by the boundary effects at normal edges.
In the model used to study the effects of the
longitudinal stringers, the real geometry of the frame
has been accounted for, which includes the holes
necessary for the crossing of the stringers and the
presence of the tear strap between the skin and the
frame.
Finally, a model has been developed that takes into
account the presence of a structural element that
interrupts the axial symmetry of the fuselage section
(Fig. 8). The upper-deck oor beam has been added,
lying on the bisector to the fuselage cylindrical portion;
symmetry constraints are applied to the oor beam
at the longitudinal plane of symmetry of the fuselage
(Fig. 9).
In all the models the cross-section of the frames has a
Z shape and the tear strap is modelled as a simple strip;
hat stringers have been used and the oor beam has an I
cross-section. Main dimensions of the components are
summarized in Table 1 and correspond to the output of
the optimization code relevant to a pressure load of
68 000 Pa.
Also for this model all components are made of
aluminium alloy 2024, except for the tear strap, for
which titanium alloy is used.
5 RESULTS OF THE FINITE ELEMENT
ANALYSES
In this section the results relevant to the models and
load conditions described in the previous section are
discussed. In particular, as far as the mass load analyses
are concerned, a comparison of analytical versus
numerical results of the shear ow transfer between
the skin and the frame is presented. Numerical results
have been obtained by means of a linear FEA.
The results concerning the pressure load analyses are
subdivided into two parts: the rst considers the effects
of the stringers while the second shows the modications
of the stress eld due to the presence of a structural
element which causes the loss of axial symmetry. The
results of the two subsets have been computed by using a
geometrically non-linear nite element method (FEM)
which had been previously tested against well-known
theoretical results relevant to a constant thickness shell
stiffened by means of equally spaced frames. For such
cases the agreement between the numerical results and
the theoretical data was found to be excellent. Fig. 8 Meshing example of a fuselage sector with oor beams
Fig. 9 Circumferential control stations
L BONI AND D FANTERIA 322
Proc. Instn Mech. Engrs Vol. 218 Part G: J. Aerospace Engineering G05603 # IMechE 2004
5.1 Skin-frame shear ow transfer: comparison of
analytical and numerical results
In Fig. 10, results relevant to the case of a single loaded
frame are shown. It can be observed that the skin shear
ow, induced by the frame, predicted by FEA is in fairly
good agreement with the theoretical distribution along
the section perimeter. Both results highlight that the
region above the upper-deck oor beam is relatively
unloaded, while the major part of the overall shear load
is carried by the regions between the two oor beams.
Results relevant to the case of multiple loaded frames
are compared with the results of a single-loaded frame in
Fig. 11. It is evident that the presence of loaded frames
in the vicinity of the section under analysis redistributes
the skin shear ow, so that the peak value is shifted
towards the upper-deck oor beam; as a consequence,
the zone right above the oor beam, where the shear
ow is close to zero in the previous case, results in a
signicant amount of shear load.
As a nal remark, it can be observed that the shear
ow distribution in the case of multiple loaded frames is
%
Fig. 10 Single-loaded frame: comparison between theoretical and FEM results
Fig. 11 FEM results: comparison of a single loaded frame and multiple loaded frames
ANALYTICAL METHOD FOR FUSELAGE DESIGN BY MEANS OF FE ANALYSES 323
G05603 # IMechE 2004 Proc. Instn Mech. Engrs Vol. 218 Part G: J. Aerospace Engineering
quite similar to the distribution predicted by the
elementary theory.
5.2 Effects of the stringers on the stressstrain eld due
to pressurization
In Fig. 12, results relevant to the case of a single and
innitely stiff frame inserted in a shell stiffened by
longitudinal stringers are shown in terms of non-
dimensional radial displacement versus longitudinal
non-dimensional coordinate
*
, while in Fig. 13 the
variation in the radial displacement along the section
Fig. 12 Comparison of FEM and theory for a single, innitely stiff frame and stiffened shell
Fig. 13 FEM results for a single, innitely stiff frame and stiffened shell
*
The origin of the longitudinal coordinate is at the location of one of
the frames; the reference length used to make the longitudinal
coordinate non-dimensional is l
2
&68:3 mm; the radial displacement
is made non-dimensional by dividing it by w
m
Dp R
2
=Et&3:75 mm.
L BONI AND D FANTERIA 324
Proc. Instn Mech. Engrs Vol. 218 Part G: J. Aerospace Engineering G05603 # IMechE 2004
perimeter, for different sections distributed in long-
itudinal direction, is presented.
About the radial displacement distribution along the
longitudinal axis, the followingobservations canbe made.
1. Far from the frame, the presence of stringers
diminishes the membrane axial stress in the skin,
causing a larger radial expansion of the stiffened shell
with respect to the non-stiffened shell.
2. Close to the frame, halfway between two stringers,
the radial displacement approaches the asymptotical
value of the theoretical distribution relevant to the
solution with equivalent thickness t
eq
; as the distance
from the frame increases, the numerical solution
tends to the theoretical value relevant to the non-
stiffened shell of thickness t.
3. Close to the frame, at the stringers location, the
radial displacement is signicantly lower than the one
between the stringers; such difference disappears as
the distance from the frame increases.
Fig. 14 FEM results for multiple frames and stiffened shell: longitudinal variation in radial displacement
Fig. 15 FEM results for multiple frames and stiffened shell: circumferential variation in radial displacement
ANALYTICAL METHOD FOR FUSELAGE DESIGN BY MEANS OF FE ANALYSES 325
G05603 # IMechE 2004 Proc. Instn Mech. Engrs Vol. 218 Part G: J. Aerospace Engineering
In Figs 14 and 15, results relevant to the case of
multiple frames, of nite stiffness, in a longitudinally
stiffened shell are shown in terms of variation in the
radial displacement along the longitudinal axis and
along the section perimeter respectively.
Close to the frames, the radial displacement halfway
between two stringers is lower than that at the stringers
location (Fig. 14); at mid-bay between the frames the
opposite is true; i.e., the stiffened shell expands more
halfway between two stringers than in correspondence
to the stringers themselves. The behaviour mentioned
above can be observed also in Fig. 15 by comparing the
circumferential distribution of radial displacement at
stations 1 and 4. An explanation of such behaviour can
be given by realizing that the stiffness of the frame is not
uniform but decreases signicantly at stringers locations
due to the presence of the holes to allow for the crossing
of the stringers themselves.
5.3 Perturbation of the stressstrain eld due to
pressurization caused by the oor beams
Figure 16 shows the radial displacement along the
longitudinal axis, at several stations distributed along
the circumferential direction (see Fig. 9), relevant to the
case of a stiffened shell with multiple frames and taking
into account the presence of the oor beam. It is evident
that the high longitudinal stiffness of the oor beam
strongly modies the distribution of the radial displace-
ment in the vicinity of the oor beam itself (station 1);
nevertheless, such effect decreases rapidly as the distance
of the control stations increases, so that at station 6 the
value of the displacement at mid-bay between the frames
reaches the value calculated in absence of the oor beam
(see also Fig. 13 for comparison).
6 CONCLUSIONS
The results of some specic FEAs, carried out to
support the development of an integrated procedure
based on analytical and semi-empirical methods, for the
design of aircraft fuselage components, have been
presented and discussed with the aim of achieving a
deeper understanding of the advantages and the limita-
tions of the analytical approach.
FEM results demonstrate the effectiveness of the load
coefcient method in estimating the shear ow that a
single loaded exible frame introduces into the skin and
highlight the importance of the interactions between
stressstrain elds induced by a set of equally spaced
frames subjected to typical mass load distributions on
the oor beams. From this point of view, the elementary
theory provides anyway a solution that is a good
starting point for further high-delity nite-element-
based design activities.
Results of nite element models for pressure load
analyses highlight that, as far as a rst-approach design
is sought, the theoretical methods can be regarded as
fully satisfactory; nevertheless, more advanced tools are
needed when dealing with detailed design of the regions
in the vicinity of the stringers and of those structural
elements (such as the oor beams) that invalidate the
hypothesis of axial symmetry on which theoretical
methods are based.
Although the results presented cannot be regarded as
complete and exhaustive, they are a rst step towards the
validation of the analytical design procedure; further
Fig. 16 FEM results for multiple frames and stiffened shell with oor beams
L BONI AND D FANTERIA 326
Proc. Instn Mech. Engrs Vol. 218 Part G: J. Aerospace Engineering G05603 # IMechE 2004
investigations are planned for which the use of more
rened FEAs are being considered. From this point of
view, the combined use of a three-dimensional parametric
modeller, which can manage assemblies of many compo-
nents, and a FEM solver, with enhanced submodelling
and interface capabilities, would be very advantageous.
ACKNOWLEDGMENTS
The authors would like to express their sincere
appreciation to Professor Attilio Salvetti and Professor
Luigi Lazzeri for the encouragement, the many helpful
suggestions and fruitful discussions.
The help given by Massimiliano Cartolano and
Francesco Calvetti in carrying out the FEAs is gratefully
acknowledged.
REFERENCES
1 Wignot, J. E., Combe, H., and Ensrud, A. F. Analysis of
circular shell-supported frames. Technical Note NACA TN
929, National Advisory Committee for Aeronautics, 1944.
2 Khun, P., Peterson, J. P., and Levin, L. R. A. Summary of
diagonal tension. Technical Note NACA TN 2661,
National Advisory Committee for Aeronautics, 1952.
3 Gerard, G. The crippling strength of compression elements.
J. Aeronaut. Sci., 1958, 25, 3752.
4 Williams, D. An Introduction to the Theory of Aircraft
Structures, 1960 (Arnold, London).
5 Rivello, R. M. Theory and Analysis of Flight Structures,
1969 (McGraw-Hill, New York).
6 Bruhn, E. F. Analysis and Design of Flight Vehicle
Structure, 1973 (Jacobs Publishing, Indianapolis).
7 Niu, M. C. Airframe Structural Design, 1988 (Conmilit
Press, Hong Kong).
8 Engineering Sciences Data Units, Structures Sub-Series
(ESDU International Limited, London).
9 Schmidt, A., Lapple, M., and Kelm, R. Advanced fuselage
weight estimation for the new generation of transport
aircraft. SAWE paper 2406, 1997.
10 Boni, L. Methodologies for the optimum design of fuselage
structures of transport aircraft. PhD thesis, University of
Pisa, Pisa, Italy, April 2004.
11 Poe Jr, C. C. Stress intensity factor for cracked sheet with
riveted and uniformly spaced stringers. Technical Report
NASA TR-R-358, 1971 (National Aeronautics and Space
Administration).
12 Swift, T. Damage Tolerance in Pressurized Fuselages. In
Proceedings of the 14th International Committee on
Aeronautical Fatigue Symposium, Ottawa, Canada, 812
June 1987, pp. 177.
13 ANSYS Users Manual, Version 5.7, 1997 (ANSYS Inc.,
Canonsburg, USA).
14 Cartolano, M. Analisi numeriche di strutture di fusoliera
soggette a sollecitazioni di essione e taglio. Final degree
thesis, University of Pisa, Pisa, Italy, 2002.
15 Calvetti, F. Analisi numeriche di strutture di fusoliera
soggette a sollecitazioni di pressurizzazione. Final degree
thesis, University of Pisa, Pisa, Italy, 2002.
ANALYTICAL METHOD FOR FUSELAGE DESIGN BY MEANS OF FE ANALYSES 327
G05603 # IMechE 2004 Proc. Instn Mech. Engrs Vol. 218 Part G: J. Aerospace Engineering