Sei sulla pagina 1di 34

1

INLET FINITE ELEMENT MODELING STANDARD WORK


6/25/04

TABLE OF CONTENTS

1) INTRODUCTION.......................................................................................................................... 3
1.1 MATERIAL PROPERTIES AND ALLOWABLES ............................................................ 3
1.2 LOADS APPLICATION....................................................................................................... 4
1.2.1 Aerodynamic Pressure .................................................................................................... 4
1.2.2 Cavity Pressure ............................................................................................................... 4
1.2.3 Temperature .................................................................................................................... 4
1.2.4 Inertial ............................................................................................................................. 4
1.2.5 Burst Duct ....................................................................................................................... 4
1.2.6 Blade Out ........................................................................................................................ 4
1.3 BEAM SECTION PROPERTIES............................................................................................. 4
1.4 CATIA TO PATRAN GEOMETRY TRANSFER................................................................... 4
1.5 FILE NAMING CONVENTION.............................................................................................. 4
2) LOADS MODELS ......................................................................................................................... 5
2.1 NOSE LIP.................................................................................................................................. 5
2.1.1 General Mesh Refinement Requirements ........................................................................... 5
2.1.2 Geometry and Interface Modeling ...................................................................................... 9
2.2 FORWARD BULKHEAD...................................................................................................... 10
2.2.1 General Mesh Refinement Requirements ......................................................................... 10
2.2.2 Modeling of Stiffeners ...................................................................................................... 12
2.2.3 Geometry and Interface Modeling .................................................................................... 12
2.3 INNER BARREL.................................................................................................................... 13
2.3.1 General Mesh Refinement Requirements ......................................................................... 13
2.3.2 Honeycomb Panel Modeling............................................................................................. 14
2.3.3 Geometry and Interface Modeling .................................................................................... 15
2.4 OUTER BARREL................................................................................................................... 16
2.4.1 General Mesh Refinement Requirements ......................................................................... 16
2.4.2 Outer Barrel Frame and Stiffener Modeling ..................................................................... 17
2.4.3 Geometry and Interface Modeling .................................................................................... 18
2.5 AFT BULKHEAD................................................................................................................... 19
2.5.1 General Mesh Refinement Requirements ......................................................................... 19
2.5.2 Stiffener Modeling ............................................................................................................ 21
2.5.3 Geometry and Interface Modeling .................................................................................... 21
2.6 ATTACH RING AND ENGINE MODEL ............................................................................. 22
2.6.1 General Mesh Refinement Requirements ......................................................................... 22
2.6.2 Geometry and Interface Modeling .................................................................................... 22
2.6.3 Engine Interface ................................................................................................................ 24
2.7 ACCESS PANELS.................................................................................................................. 24
2.7.1 General Mesh Refinement Requirements ......................................................................... 24
2.7.2 Modeling of Stiffeners ...................................................................................................... 24
2.7.3 Geometry and Interface Modeling .................................................................................... 25
2.7.4 Bolted Joint Interface Stiffness ......................................................................................... 25
2
2.8 PRESSURE RELIEF DOOR................................................................................................... 28
2.7.1 General Mesh Refinement Requirements ......................................................................... 29
2.7.2 Modeling of Stiffeners ...................................................................................................... 29
2.7.3 Geometry and Interface Modeling .................................................................................... 29
2.9 ANTI ICE TUBE AND EXIT DUCT..................................................................................... 29
2.9.1 Anti Ice Tube Modeling.................................................................................................... 29
2.9.2 Anti Ice Exit Duct Modeling............................................................................................. 30
2.10 ECU COOLING AND FAN COMPARTMENT VENTILATION INLETS ....................... 31
2.11 NOTES ON INCOMPATIBLE FINITE ELEMENTS ......................................................... 31
3) STRESS AND FATIGUE MODELS........................................................................................... 32
4) LINEAR BUCKLING MODELS ................................................................................................ 32
5) ACOUSTIC RESPONSE MODELS............................................................................................ 33
6) VIBRATION RESPONSE MODELS.......................................................................................... 33
7) REFERENCES............................................................................................................................. 34

3
INLET FINITE ELEMENT MODELING STANDARD WORK
12/**/00 DRAFT FOR REVIEW


1) INTRODUCTION

The purpose of this document is to provide guidelines for the inlet FE modeling of inlet structures.
The guideline is separated into the following five sections:
1. Loads Models
2. Stress and Fatigue Models
3. Linear Buckling
4. Acoustic Response Analysis Models
5. Vibration Response Analysis Models
Each of these subject areas will provide the preferred mix of finite element modeling and the hand
calculations available though the standard method of analysis and reporting tools.

Documentation will address the following finite modeling issues:
Mesh refinement/hand calculation issues regarding:
Cylindrical shell edge bending affects captured at interfaces and stiffeners
Stress concentration
Punch loads
Mode shape
Preferred modeling and required output for:
Connections between nose lip, inner barrel and fwd bulkhead
Connections between nose lip, outer barrel and fwd bulkhead
Connections between aft bulkhead and inner barrel
Inner Barrel models through:
Solid elements for core, membrane for skins and shells for lands
Shell elements with PCOMPS
Shell elements with PSHELL
Stiffeners modeled
Beams with offsets
Beams without offsets (what inaccuracy is induced)
Beams with warping
Explicit as shells


1.1 MATERIAL PROPERTIES AND ALLOWABLES

Materials data is obtained from the CENTOR data base. The data base is accessed through the
BFG intranet under the point of use tools.


4
1.2 LOADS APPLICATION

Load case naming convention shall follow:
1.2.1 Aerodynamic Pressure

1.2.2 Cavity Pressure

1.2.3 Temperature

1.2.4 Inertial

1.2.5 Burst Duct

1.2.6 Blade Out


1.3 BEAM SECTION PROPERTIES

Section properties for beams are calculated with the Standard Method of Analysis and Report Tools
available on the BFG intranet. If these tools are not yet available, the properties are to be
calculated in PATRAN using the section property calculation utility.


1.4 CATIA TO PATRAN GEOMETRY TRANSFER

Surfaces imported to PATRAN should be CATIA faces that are translated through the
CATEXPRESS program or via the CATIA Direct interface. The CATIA Direct interface may
require a database from an older release of CATIA. It is currently working with CATIA version
4.19.


1.5 FILE NAMING CONVENTION

File naming convention will follow corporate business practice, CBP-D330.03.01 Engineering
Reports, Preparation And Control Of.
5
2) LOADS MODELS

The finite element load model approach suggested here was developed for small deformation linear
elastic analyses. Some of the shell and beam element offset options that are suggested in this
section will produce errors if run in a bucking, modal, dynamic or non-linear solution sequence in
NASTRAN. In general, shell and beam offsets are not included in the differential stiffness
algorithm.

2.1 NOSE LIP

The nose lip is a doubly curved shell structure. The nose lip construction is shown in the BFG
Intranet under the lessons learned index and the best practices sub-bullet.

2.1.1 General Mesh Refinement Requirements

Finite element modeling of nose lip requires a minimum of elements to achieve a reasonable
stiffness for the component. One of the more commonly overlooked mesh refinement issues is the
membrane-bending stiffness coupling that occurs in curved shells. The most basic example of this
type of coupling is that of an axis-symmetric edge loading on a cylinder. These edge loads
generally arise from interfaces between the shell and a bulkhead, ring stiffener or connection to an
elliptical end cap. In general, stresses that are induced in the shell decay over a length that is small
in comparison to the radius of the shell. In fact, the axial decay length is small compared to the
element maximum edge length sizing rule of one element every ten degrees of arc around the
circumference of the shell, Reference 1. The recommended edge length around the nose lips
circumference corresponds to an arc that is between 5 to 7.5 degrees.

In order to illustrate the membrane-bending decay, closed form solutions from reference 2 are
examined here. The three cases provide the analytical stress as a function of axial distance from an
axis-symmetric load application point on a long cylinder with radius R, thickness t, elastic modulus
E and Poissons ratio v.

V
Cylinder Skin

M





X
Cylinder Centerline X
6





The three cases and their functional forms are:
I. An axis-symmetric line load in the middle of a long cylinder resulting bending stress decay
formula: )] sin( ) [cos( /
max
x x e
x
b

. This equation best represents the affect of a


hoop stiffener or splice on a cylindrical shell.
II. An edge prying moment which has the following bending stress decay formula:
)] sin( ) [cos( /
max
x x e
x
b

+ =


III. An edge shear load that results in a hoop stress decay formula of: ) cos( /
max
x e
x
h


=
Where x is the distance along the axis of the shell from the load application point and
4 2
4 / DR Et = is the membrane-bending stiffness parameter. For a uniform thickness isotropic
shell, this equation reduces to
4
2 2 2
) /( ) 1 ( 3 t R v = where )] 1 ( 12 /[
2 3
v Et D = . The form of
the above equations allows the stress decay to be plotted as a function of a non dimensional
parameter x/, as shown in the following figure.


CYLINDRICAL SHELL AXIS SYMMETRIC LOADING
STRESS DECAY PLOT FOR DIFFERENT LOADS
-0.25
0
0.25
0.5
0.75
1
1.25
0.000 1.000 2.000 3.000 4.000
Axial Length x Lamda
S
t
r
e
s
s


/


M
a
x
i
m
u
m

S
t
r
e
s
s
Line Load
Edge Moment
Edge Shear



7
Based on the above figure, at the very least a single element would be required to account for this
local affect. The length of the edge of the single element would approximately correspond to x=1
resulting in x=1/.

Finite element size and the number of elements used to model the edge decay was studied using the
MSC/MARC finite element code. The internal study looked at the resulting shear and moment at
the interface of an internally pressurized cylindrical shell with radial and circumferential rotation
end constraints. The study used an aluminum cylinder with the following dimensions: a radius of
50 inches, a length of 50 inches and a thickness of 0.060 inches. Element size used in the model
was about 8.3 inches axially and 8.7 inches circumferentially, which is consistent with a typical
coarse loads model. Three FE configurations corresponding to one, two and three elements used to
model the edge decay were analyzed. Each of these three configurations was analyzed with the
following three different axial element edge lengths: 0.5/, 1.0/ and 2.0/, where 1/=1.35 inches.
Results of the study present the finite element edge shear load divided by the analytical result as a
function of element edge length. The results are illustrated as three curves, one for each of the
configurations reflecting the number of elements used to model the edge affect, in the following
figure:

PRESSURIZED CYLINDRICAL SHELL WITH FIXED ENDS
EDGE SHEAR REACTION LOAD AS A FUNCTION OF
MESH REFINEMENT AND ELEMENT SIZE
0
0.5
1
1.5
2
2.5
3
0 1 2 3 4 5 6 7
Axial Element Edge Length x Lamda
F
E

S
h
e
a
r

R
e
a
c
t
i
o
n

/

A
n
a
l
y
t
i
c
a
l
One Element
Two Element
Three Element



Note that there is a substantial reduction in shear load with the addition of even one element over
the typical large elements used in a loads model. Near analytical shear loads can be obtained by
8
using just two elements with an axial edge length equal to 1/ adjacent to interfaces like a hoop
stiffener or splice joint. If this edge affect is ignored, the interface shear loads can be higher and
the membrane stresses in the skin will be lower than they actually are. In addition, for composite
materials with poor inter-laminar shear strength, ignoring this affect could lead to not being able to
validate a good design.

With this membrane coupling affect in mind, three checks must be made in the development of the
mesh density. These checks are:
A) The nose lip element at the interface to the forward bulkhead/nose lip splice needs to reflect the
affect of the membrane-bending decay length. Modeling this affect will better model the load
transfer between the two structural components in the fore-aft direction. The edge length of the
first two interface elements, L
x
, in the fore-aft direction should be
4
2 2 2
) 1 ( 3 / v t R L
x
= , where
R is the nose lips smallest radius in the plane of the interface to the forward bulkhead.
B) The nose lip element edge length along the inlets circumferential direction is taken as the
smaller of the following two numbers: a circumferential arc length subtending an angle of less
than 10 degrees, and eleven (maximum aspect ratio) times the fore-aft edge length, L
x
, at the
bulkhead interface as defined in item A.
C) Element edge length in the fore-aft direction, other than at the interface identified in item A,
shall have an included angle of less than 10 degrees or less, Reference 1. This corresponds to a
difference between the actual shell surface and the finite element edge of 2.2% of the element
edge length. Because the radius of curvature changes dramatically in this direction, the latter
definition can be used in PATRAN when defining the element density around the parabolic
fore-aft cross-section of the nose lip.
Check B above reflects the overall global stiffness requirement along with the local edge affects
coupled with the element edge aspect ratio (long edge divided by the short edge) requirements for
element accuracy. For rectangular elements with stress gradients primarily in one direction, the
maximum aspect ratio should be limited to eleven, Reference 1. In general, the aspect ratio should
be less than four, and even less when in the presence of large stress gradients, with the best aspect
ratio being equal to one, Reference 3. Element skewness and taper have a more pronounced affect
on element accuracy than does the element aspect ratio, Reference 3.

One additional investigation was performed on the mesh refinement on the nose lip. Fore-aft
stiffness of the bulkheads, outer barrel and nose lip is generally governed by the fore-aft stiffness of
the nose lip. As a result, a mesh size sensitivity study was performed on the finite element model
of the BR715 nose lip. The study addressed the mesh refinement impact on the fore-aft stiffness of
the nose lip due to a unit fore-aft load applied at the interface to the outer-barrel. The baseline
mesh refinement was compared to the results of a mesh that was twice as dense in the fore-aft
direction while the circumferential distribution was kept at its original 72 element uniform-density.
The stiffness was evaluated by applying a unit deflection in the for-aft direction of the outer edge
and comparing the reactions between models on the inner edge. The inner edge was constrained in
all degrees of freedom, while the outer edge additional constraints were fixed in all rotational
DOFs. Two surfaces were fit to the baseline mesh in PATRAN. These surfaces were then meshed
to a density that was twice the baseline density and analyzed with the same loads and boundary
conditions as were applied to the baseline model. This increased mesh in density was found to
change the axial reaction forces by less than 3%. Sizing of the recommended element edge length,
9
L, is based on keeping the error, h, between the actual surface and the straight element edge to:
h<0.06 L .

The nose lip skin will be modeled with using the following MSC/NASTRAN inputs:
CQUAD4 without offsets
PSHELL with membrane, bending and transverse shear stiffness
MAT1 material definition referencing MATT1 for temperature dependence
MATT1 entries for modeling the temperature dependence of the material stiffness
TEMP entries provide the grid temperatures for the element
The PSHELL cards have a stiffness entry for membrane-bending coupling. This coupling models
an asymmetric composite material where there is an offset between the lay-up neutral axis and the
mid-plane of the section. The asymmetric material will result in out of plane deflections due to in-
plane loads. This is different from the shell membrane coupling in that the source of the coupling
is due solely to the geometry of the shell structure.

2.1.2 Geometry and Interface Modeling

This section addresses modeling details for the following three areas:
I. Geometry used for nose lip
II. Interface to the forward bulkhead
III. Interface to the splice plates
These details are called out in a separate section to highlight their requirements.

The geometry required for the nose lip corresponds to the mid plane surface of the nose lip. The
CATIA surface should be broken at the fastener lines of the forward bulkhead and hoop splice plate
locations. A mid-plane surface for the splice plates should also be obtained for reference.

The nose lip interface to the forward bulkhead requires the same circumferential element spacing
on both components to allow for a clean mesh interface between them. Having the same
circumferential element spacing and the surfaces broken along the fastener lines allows for a clean
line of near coincident nodes between the two components. For the purpose of the loads model, the
forward bulkhead nodes along the fastener line are equivalenced to the nose lip nodes as shown
below.


Nose Lip Outer Barrel










10
Forward Bulkhead







Nose Lip Inner Barrel


It is recommended to equivalence these nodes rather than use MPC equations or RBE2 elements
between the interface nodes. This is recommended for model simplicity and because substantial
care must be taken to insure that the nodes that are tied together are directly in line with their
surface normal. Otherwise, there will be an artificial moment applied to the nodes when the nose
lip and bulkhead bear against each other.

Historically, the material overhang nose-lip between the fastener line and the edge of the part has
been modeled with beam elements. If the distance between the fastener line and the edge of the
part is small compared to the shell bending-membrane decay length, beam elements along the
fastener line are appropriate. Otherwise, shell elements representing the nose-lip between the
fastener line and the edge of the part should be added to the model. Likewise, the same thing can
be done for the inner barrel and outer barrel components.

Splice plate interfaces between the three or so segments of the nose lip are generally not explicitly
modeled. Additional bending stiffness and in the fore-aft direction and mass, due to the splice/skin
over-lap, can be added by beam elements such as the CBAR or CBEAM element. Hand analysis of
the joint needs to consider that it is a single lap shear joint, which has an eccentricity that induces
bending in the joint. These bending loads can affect the joints strength and fatigue life.

2.2 FORWARD BULKHEAD

The forward bulkhead is a doubly curved shell structure. Its construction is shown in the BFG
intranet under the lessons learned index and the best practices sub-bullet.

2.2.1 General Mesh Refinement Requirements

Mesh refinement suggested for use in the forward bulkhead was determined from a mesh sensitivity
study using two load cases, on the BR-715 forward bulkhead geometry. The two load cases are:
I. Relative fore-aft stiffness between the outer and inner edges of the bulkhead
II. Relative radial stiffness between the outer and inner edges of the bulkhead
The mesh only included the typical forward curved section of this bulkhead. The inner radius of
the model was 27 inches. The shell thickness was 0.060 inch. A uniform mesh density was used
around the circumference model with total of 72 elements. The radial mesh density was non-
uniform in two directions, with the center element being 2.5 times larger than the center element.
11
The outer and inner flanges were not considered since they are effectively short cylinders that
already have mesh size criteria. The models outer edge was constrained in all degrees of freedom,
while the inner edge additional constraints were fixed in all rotational degrees of freedom. The
fore-aft stiffness was evaluated by applying a unit deflection in the for-aft direction of the inner
edge and comparing the reactions between models. Radial stiffness was evaluated by applying a
unit deflection in the radial direction of the inner edge and comparing the reactions between
models.

Three models were developed each with a different number of elements from the inner to the outer
edge of the forward bulkhead. Models were developed with 7, 12 and 20 elements from the inner
to the outer edge (radial direction) of the bulkhead. The corresponding minimum radial element
edge length for the three cases are: 0.766 inch, 0.423 inch and 0.249 inch, respectively. The 7
element model already existed. The 12 and 20 element models were generated from two surfaces
that were fit to the left and right sides of the 7 element mesh using the PATRAN create-surface-
mesh option. The results of the study are tabulated in the following tables.

RADIAL UNIT DISPLACEMENT ON INNER RIM
Number of
Elements
Edge Elements
Radial Length
Inner
Reaction
Force
Outer
Reaction
Force
Inner
Reaction
Moment
Outer
Reaction
Moment
Bilinear
VM Stress
(ksi)
7 0.766 4630 582 1420 9.44 944
12 0.423 4850 514 1510 2.39 1140
20 0.249 4860 521 1520 5.01 1190

AXIAL UNIT DISPLACEMENT ON INNER RIM
Number of
Elements
Edge Elements
Radial Length
Inner
Reaction
Force
Outer
Reaction
Force
Inner
Reaction
Moment
Outer
Reaction
Moment
Bilinear
VM Stress
(ksi)
7 0.766 156 1930 67.7 406 151
12 0.423 182 1860 86.8 414 189
20 0.249 169 1820 83.0 414 195

Results indicate that 9 radial elements should be used in this case with a 0.57 inch radial element
edge length on the inner and outer edges of the bulkhead. Non-uniform radial mesh density is to be
used with the center element 2.5 times the size of the innermost and outermost elements.

In general, the forward bulkhead mesh size will use the non-uniform mesh distribution with a
maximum ratio of 2.5 between the middle and edge elements. The maximum outer and inner
element radial edge length, L
R
, shall be:
4 / Rt L
R
=
where R is the inner radius of the inlet and t is the thickness of the bulkhead.

12
Additional elements are added to the mesh to account for the fore-aft return flange between the
forward curved section and the two fastener line interfaces to the nose lip. The minimum fore-aft
element length, L
x
, is calculated by the following formula:
4
2 2 2
) 1 ( 3 / v t R L
x
=
where R is the inner radius of the inlet and t is the thickness of the bulkhead.

The forward bulkhead skin will be modeled with using the following MSC/NASTRAN inputs:
CQUAD4 without offsets
PSHELL with membrane, bending and transverse shear stiffness
MAT1 material definition referencing MATT1 for temperature dependence
MATT1 entries for modeling the temperature dependence of the material stiffness
TEMP entries provide the grid temperatures for the element

2.2.2 Modeling of Stiffeners

The forward bulkhead stiffeners are generally asymmetric cross-section curved beam-columns.
These beam-columns are typically stopped before the interfaces to the inner and outer barrel. As a
result, beam offsets along with beam warping coefficients are required to model the stiffness of this
element. Beam offsets are modeled by placing nodes that are offset from the shell element nodes to
the beams shear center. The offset node is then connected to the adjacent shell element node by a
RBE2 element with all six dofs assigned. The offsets model the load transfer from in-plane skin
loads into the stiffeners. And the warping coefficient models the roll over of the asymmetric beam
cross-section under axial loading.

Beam shear loads normal to the skin of the bulkhead induce bending loads in the stiffener flange
between the stiffener shear web and the fastener line, in addition to the rivet pull through loads.
Rivet shear and pull through interface loads can be extracted from the finite element model through
the grid point force balance option in NASTRAN. Additional shell element refinement along the
stiffener length may be required to extract the interface loads between the stiffener and skin.


Stiffeners will be modeled with using the following MSC/NASTRAN inputs:
PBEAM beam properties including the warping coefficient for stiffeners
CBEAM
MAT1 material definition referencing MATT1 for temperature dependence
MATT1 entries for modeling the temperature dependence of the material stiffness
TEMPRB entries provide the temperature for the element

2.2.3 Geometry and Interface Modeling

This section addresses modeling details for the following three areas:
13
I. Geometry used for bulkhead
II. Interface to the nose lip
III. Interface to the splice plates
These details are called out in a separate section to highlight their requirements.

The geometry required for the forward bulkhead corresponds to the mid plane surface of the
bulkhead. The CATIA surface should be broken at the fastener lines of the nose lip, outer barrel,
inner barrel, and hoop splice plate locations. A mid-plane surface for the splice plates should also
be obtained for reference.

The forward bulkhead interface to nose lip, outer barrel and inner barrel requires the same
circumferential element spacing on all components to allow for a clean mesh interface between
them. Having the same circumferential element spacing and the surfaces broken along the fastener
lines allows for a clean line of near coincident nodes between the two components. For the purpose
of the loads model, the forward bulkhead nodes along the fastener line are equivalenced to the nose
lip nodes. See noselip interface section for illustration.

Splice plate interfaces between the generally three forward bulkhead segments are generally not
explicitly modeled. Additional bending stiffness and in the fore-aft direction and mass, due to the
splice/skin over-lap, can be added by beam elements such as the CBAR or CBEAM element. Hand
analysis of the joint needs to consider that it is a single lap shear joint, which has an eccentricity
that induces bending in the joint which can affect the joints strength and fatigue life.

2.3 INNER BARREL

The inner barrel is a doubly curved shell structure that is of honeycomb construction. Its
construction is shown in the BFG intranet under the lessons learned index and the best practices
sub-bullet.

2.3.1 General Mesh Refinement Requirements

Honeycomb sandwich construction is generally used in inner barrel structure. The structure is
usually made in three circumferential segments, which are spliced together along the fore-aft
direction split in the honeycomb panel. These splices interrupt the hoop bending capability of the
inner barrel and induce local bending moments due to the offset between the honeycomb section
neutral axis and the splice, which connects one of the two honeycomb skins together.

As a result, the inner barrel splices must be considered explicitly in the model. As a minimum, one
element will be used to model the honeycomb panel land between the panel and the fastener line at
the splice/land interface. And at least one element representing the splice between the splice/land
interfaces. This will result in a local mesh refinement in the circumferential direction at this
interface. This refinement will need to be considered in how the mesh is continued through the
other inlet components.

14
Mesh refinement required for membrane-bending coupling in a honeycomb panel is not as
significant an affect as in thin solid shells. The membrane-bending decay parameter is:
4 2
4 / DR k = . For a panel with core height c and skin thickness t, the bending stiffness, D, is:
)] 1 ( 2 /[ ) (
2 2
v t t c E D + = and, k, is the membrane stiffness k=E(2t). The decay parameter equation
reduces to
4
2 2 2
] ) ( /[ ) 1 ( t c R v + = . For a cylindrical honeycomb panel with a radius of 50
inches, a core thickness of 0.5 inch, a skin thickness of 0.020 and a Poissons ratio of 0.3, the
resulting decay length, 1/, is 5.2 inches. This length is easily four times greater than that of a
typical nose lip. The minimum mesh refinement for the honeycomb panel will follow the same
basic rule as discussed in the nose lip section.

The axial element width at the forward and aft interfaces should consist of at least two elements
that each should have an axial length, L
x
, of: Rc L
x
= , where R is the minimum radius in the inner
barrel.

Inner barrel lands will be modeled with the following NASTRAN input options:
CQUAD4 without offsets
PSHELL with membrane, bending and transverse shear stiffness
MAT1 material definition referencing MATT1 for temperature dependence
MATT1 entries for modeling the temperature dependence of the material stiffness
TEMP entries provide the grid temperatures for the element

2.3.2 Honeycomb Panel Modeling

Shell element representation of the inner barrel in the initial stages of a program can provide the
capability to respond to panel core height and skin thickness changes without having to make
substantial changes to the geometry of the model. As a result, it is recommended that the inner
barrel will be modeled with using the following MSC/NASTRAN options:
CQUAD4 for the honeycomb panel, panel lands and splice plates.
Honeycomb panel element will be modeled with PCOMP entries with offset representing
the distance from the element location and the center of the section.
Isotropic skins can use MAT1 or MAT8 material card. MAT1 definition uses E and NU for
membrane and bending stiffness and G for transverse shear stiffness (quick reference guide,
version 68, page 633)
Core transverse shear modulii should be entered on a MAT8 material card with the in-plane
modulii set to a default value of 100 psi, the in-plane shear modulus set to 50 psi and a
Poissons ratio of 0.
TEMPP1 cards may be necessary if there is a temperature difference between the inner and
outer skins of the honeycomb panel.
The usage of the PCOMP option is fairly straight-forward in the absence of thermal loads. In
general, when PCOMP property option is used, the following issues to be addressed:
i. PCOMP entries do not allow for temperature dependant material properties (uses values
based on TREF only)
15
ii. Element loads will be relative to the element plane not the neutral axis of the PCOMP
section.
iii. Use composite stress output (param,nocomps,+1)
iv. Homogeneous stress will be incorrect if asked for. This is because of the Z0 offset and
because the membrane and bending thickness are equal to the total honeycomb panel
thickness which are then used to calculate stress by: S = N/h +6M/h
2
, where h is the
thickness of the core plus both skins ( Should be: S= N/(2t
s
) + M/[(c+t
s
)t
s
] , where c is
the core thickness and t
s
is the skin thickness).
v. Core flatwise compression and tension stresses can be approximated by the following
equation: S
fw
= M/(Rc) , where M is the applied bending moment on the section, R is the
shell radius of curvature in the direction of M, and c is the thickness of the core.
vi. Thermal loads can be modeled by defining unique material numbers for each ply in the
PCOMP. Each material will have the modulii and the thermal expansion coefficient that
corresponds to the temperature of the ply. These values must be recalculated for every
different temperature load case. This approach can lead to a separate PCOMP along with
its unique material cards for each QUAD4 element in the model.
Provided these limitations of the PCOMP entry are carefully considered, its use in a loads-model is
a reasonable way to go. The PCOMP has the advantage of being able to mix different materials
along with different material thicknesses.

2.3.3 Geometry and Interface Modeling

This section addresses modeling details for the following three areas:
I. Geometry used for the inner barrel
II. Hoop Splice Modeling
III. Interface to the forward bulkhead
IV. Interface to the aft attach ring
These details are called out in a separate section to highlight their requirements.

The geometry required for the inner barrel corresponds to an offset from the wetted surface of the
inner barrel. The offset will be equal to half the thickness of the panel land areas. The CATIA
surface should be broken at the fastener lines of the forward bulkhead, the aft bulkhead (if separate
from the aft attach ring), the aft attach ring, and the hoop splice plate locations. A mid-plane
surface for the splice plates should also be obtained for reference.

The inner barrel mesh requires the same circumferential element spacing on all components to
allow for a clean mesh interface between them. Having the same circumferential element spacing
and the surfaces broken along the fastener lines allows for a clean line of near coincident nodes
between the two components.

Inner barrel forward land to forward bulkhead attachment modeling requires at least one element
between honeycomb panel and the fastener line at the splice/land interface. And at least one
element representing the splice between the splice/land interfaces. For the purpose of the loads-
model, the forward bulkhead nodes along the fastener line are equivalenced to the inner barrel
nodes, as shown below.

16

Forward Bulkhead






Nose Lip Inner Barrel



Inner barrel honeycomb panel segment connections are modeled using at least one element between
honeycomb panel and the fastener line at the splice/land interface. And at least one element
representing the splice between the splice/land interfaces. This is illustrated in the following figure.









Inner Barrel Segment 1 Splice Plate Inner Barrel Segment 2


Hand analysis of the latter two joints needs to consider that they are single lap shear joints, which
have an eccentricity that induces bending in the joint which can affect the joints strength and
fatigue life.

2.4 OUTER BARREL

The outer barrel is a doubly curved shell structure that is a frame stiffened skin structure. Its
construction is shown in the BFG Intranet under the lessons learned index and the best practices
sub-bullet.

2.4.1 General Mesh Refinement Requirements

Outer barrel consists of a skin that may be stiffened with up to two hoop frames. These hoop
frames also provide supports for local punch loads from the pressure relief door. The outer barrel is
usually made in three circumferential segments due to fail-safe and maintenance issues. These
segments are spliced together along a land in the fore-aft direction of the inlet. These splices
interrupt the hoop bending capability of the inner barrel and induce local bending moments due to
the offset between the frame neutral axis and the splice.

17
As a result, the outer barrel splices must be considered explicitly in the model. As a minimum, one
element will be used to model the honeycomb panel land between the end of the frame and the
fastener line at the splice/land interface. In addition, at least one element should be used to model
the splice between the splice/land interfaces. This will result in a local mesh refinement in the
circumferential direction at this interface. This refinement will need to be considered in how the
mesh is continued through the other inlet components.

Membrane-bending coupling affects discussed in the nose lip section, is also applicable in the outer
barrel. Two checks must be made in the development of the mesh density due to the coupling
affect. These checks are:
A. The outer barrel element size at the interface to the forward and aft bulkheads and the hoop
frame splices need to reflect the affect of the membrane-bending decay length. Modeling this
affect will better model the load transfer between the structural components in the fore-aft
direction. The edge length of the first two elements at the interface, L
x
, in the fore-aft direction
should be
4
2 2 2
) 1 ( 3 / v t R L
x
= , where R is the nose lips smallest radius in the plane of the
fore-aft interface.
B. The outer barrel element edge length along the inlets circumferential direction is taken as the
smaller of the following two numbers: a circumferential arc length subtending an angle of less
than 10 degrees, and eleven (maximum aspect ratio) times the fore-aft edge length, L
x
, at the
axial interfaces as defined in item A.
Check B above reflects the overall global stiffness requirement along with the local edge affects
coupled with the element edge aspect ratio (long edge divided by the short edge) requirements for
element accuracy. For rectangular elements with stress gradients primarily in one direction, the
maximum aspect ratio should be limited to eleven, Reference 1. In general, the aspect ratio should
be less than four, and even less when in the presence of large stress gradients, with the best aspect
ratio being equal to one, Reference 3. Element skewness and taper have a more pronounced affect
on element accuracy than does the element aspect ratio, Reference 3.

The outer barrel skin will be modeled with using the following MSC/NASTRAN inputs:
CQUAD4 without offsets
PSHELL with membrane, bending and transverse shear stiffness
MAT1 material definition referencing MATT1 for temperature dependence
MATT1 entries for modeling the temperature dependence of the material stiffness
TEMP entries provide the grid temperatures for the element

2.4.2 Outer Barrel Frame and Stiffener Modeling

The outer barrel frames are generally asymmetric cross-section curved beam-columns. These hoop
beam-columns are typically stopped at splice locations between the segments of the outer barrel.
As a result, beam offsets along with beam warping coefficients are required to model the stiffness
of this element. Beam offsets are modeled by placing nodes that are offset from the shell element
nodes to the beams shear center. The offset node is then connected to the adjacent shell element
node by a RBE2 element with all six DOFs assigned. Beam offsets are also acceptable, but are not
supported in some solution sequences. The offsets model the affect of stopping and starting the
18
hoop frames. And the warping coefficient models the roll over of the asymmetric beam cross-
section under axial loading.

Beam shear loads normal to the skin of the bulkhead induce bending loads in the stiffener flange
between the stiffener shear web and fastener line, in addition to the rivet pull through loads. Rivet
shear and pull through interface loads can be extracted from the finite element model through the
grid point force balance option in NASTRAN.

Stiffeners will be modeled with using the following MSC/NASTRAN inputs:
PBEAM beam properties including the warping coefficient for stiffeners
CBEAM
MAT1 material definition referencing MATT1 for temperature dependence
MATT1 entries for modeling the temperature dependence of the material stiffness
TEMPRB entries provide the temperature for the element

2.4.3 Geometry and Interface Modeling

This section addresses modeling details for the following three areas:
I. Geometry used for the outer barrel
II. Interface to the forward and aft bulkheads
III. Interface to the splice plates
These details are called out in a separate section to highlight their requirements.

The geometry required for the outer barrel corresponds to the mid plane surface of the outer barrel.
The CATIA surface should be broken at the fastener lines of the forward and aft bulkheads, hoop
frames and hoop splice plate locations. A mid-plane surface for the splice plates should also be
obtained for reference.

The outer barrel interface to the bulkheads and frames requires the same circumferential element
spacing on all components to allow for a clean mesh interface between them. Having the same
circumferential element spacing and the surfaces broken along the fastener lines allows for a clean
line of near coincident nodes between the two components. For the purpose of the loads model, the
forward bulkhead nodes along the fastener line are equivalenced to the outer barrel nodes, as shown
below.

Nose Lip Outer Barrel





Forward Bulkhead


Splice plate interfaces between the segments in a unstiffened outer barrel are generally not
explicitly modeled. Additional bending stiffness and in the fore-aft direction and mass, due to the
19
splice/skin over-lap, can be added by beam elements such as the CBAR or CBEAM element. Hand
analysis of the joint needs to consider that it is a single lap shear joint, which has an eccentricity
that induces bending in the joint which can affect the joints strength and fatigue life. For outer
barrels with stiffeners that are not continuous, at least one row of elements is required to model the
splice joint stiffness.

2.5 AFT BULKHEAD

The aft bulkhead is a flat stiffened plate structure. Its construction is shown in the BFG intranet
under the lessons learned index and the best practices sub-bullet.
2.5.1 General Mesh Refinement Requirements

The aft bulkhead mesh needs to interface with the outer barrel mesh and the inner barrel or engine-
attach ring meshes.

Mesh refinement for the aft bulkhead was determined from a mesh sensitivity study. The study
was performed on a 0.060 thick flat square plate with ten inch long edges that were fully fixed.
NASTRAN finite element models with 4, 6, 10 and 20 uniformly spaced quad4 elements along
each edge of the plate were analyzed. FE results were compared to the following theoretical
values:

2 2
2 2
4
/ ) 0513 . 0 ( 6
/ ) 0231 . 0 ( 6
/ 00126 . 0
t pa S
t pa S
D pa w
edge
x
cent
x
=
=
=


obtained from reference 6. The study determined that a 6x6 FE element mesh density was capable
of predicting the center deflection to within 6.9% of the theoretical value. It was also capable of
determining the maximum FE edge moment (grid point reaction moment divided by element edge
length) to within 4.4% of its theoretical value. As a result, the aft bulkhead mesh refinement
should provide at least six elements across its width (radial direction).

Convergence of the FE analysis is shown in the following figure.

20
NASTRAN MESH SENSITIVITY STUDY FOR A SQUARE PLATE
WITH FIXED EDGES ( t=0.060", a=10.0", E=10.7msi, v=0.3)
BILIN AND CUBIC STRESS NODAL OUTPUT
0
0.2
0.4
0.6
0.8
1
1.2
0 5 10 15 20 25
Number of Elements Along An Edge
F
E

R
e
s
u
l
t

/

A
n
a
l
y
t
i
c
a
l

R
e
s
u
l
t
w (Center)
Sxx (Center-Bilin-Cnr)
Sxx (Edge-Bilin-Cnr)
Mx (Edge-SPC)
Sxx (Edge-Cubic)

Stress recovery at the edge of the panel did not converge was quickly as the deformation or edge
moment reactions. As can be seen in the figure, the 6x6 FE model with the cubic stress option was
only within 20% of the theoretical value, while the Bilinear and Corner options were 50% of the
theoretical value. The element centroidal stress was 3% lower than the bilinear and corner options.
This trend carries over to the element force Bilinear, Corner and Cubic output options. Additional
convergence could be obtained by applying a non-uniform mesh density approaching the edge of
the panel. Aspect ratio for these non-uniform elements should be less than four. Note that the edge
stresses are twice the center stress, so the more critical stress has the lowest convergence rate.

A curious result of the FE analysis (with the NASTRAN QUAD4 element) was that the
displacements started on the higher than the analytical value and then converged to the correct
value. This is curious in that H type finite elements tend to be over stiff. As a result, one would
expect to have low displacements with a only a few elements. As one then adds elements the
displacements would increase, converging to the theoretical value. The MSC/MARC FE code is an
example of the latter convergence trend.

The aft bulkhead skin will be modeled with using the following MSC/NASTRAN inputs:
CQUAD4 without offsets
PSHELL with membrane, bending and transverse shear stiffness
MAT1 material definition referencing MATT1 for temperature dependence
MATT1 entries for modeling the temperature dependence of the material stiffness
21
TEMP entries provide the grid temperatures for the element

2.5.2 Stiffener Modeling

The aft bulkhead stiffeners are generally asymmetric cross-section straight beam-columns. As a
result, beam offsets along with beam warping coefficients are required to model the stiffness of this
element. Beam offsets are required to model the affect of the free rotation condition at each end of
the beam. And the warping coefficient models the roll over of the asymmetric (non axis-
symmetric) beam cross-section. Beam offsets are modeled by placing nodes that are offset from the
shell element nodes to the beams shear center. The offset node is then connected to the adjacent
shell element node by a RBE2 element with all six DOFs assigned. Beam offsets are also
acceptable, but are not supported in some solution sequences.

Beam shear loads normal to the skin of the bulkhead induce bending loads in the stiffener flange
between the stiffener shear web and fastener line, in addition to the rivet pull through loads. Rivet
shear and pull through interface loads can be extracted from the finite element model through grid
point force balance option in NASTRAN.

Stiffeners will be modeled with using the following MSC/NASTRAN inputs:
PBEAM beam properties including the warping coefficient for stiffeners
CBEAM
MAT1 material definition referencing MATT1 for temperature dependence
MATT1 entries for modeling the temperature dependence of the material stiffness
TEMPRB entries provide the temperature for the element

2.5.3 Geometry and Interface Modeling

This section addresses modeling details for the following three areas:
I. Geometry used for bulkhead
II. Interface to the outer barrel
III. Interface to the inner barrel or engine attach flange
These details are called out in a separate section to highlight their requirements.

The geometry required for the aft bulkhead corresponds to the mid plane surface of the bulkhead.
The CATIA surface should be broken at the fastener lines of the stiffeners, outer barrel, inner barrel
or engine attach flange, and hoop splice plate locations. A mid-plane surface for the splice plates
should also be obtained for reference.

The aft bulkhead interface to outer barrel and inner barrel/attach ring requires the same
circumferential element spacing on all components to allow for a clean mesh interface between
them. Having the same circumferential element spacing and the surfaces broken along the fastener
lines allows for a clean line of near coincident nodes between the two components. Nodes along
the fastener line are equivalenced to the interfacing component nodes.

22
Note that, if the bulkhead is attached to the inner or outer barrel through a bending element, the
stiffness of that bending element needs to be included in the FE model.

Splice plates between the segments of the aft-bulkhead are generally not explicitly modeled.
Additional bending stiffness and in the fore-aft direction and mass, due to the splice/skin over-lap,
can be added by beam elements such as the CBAR or CBEAM element. Hand analysis of the joint
needs to consider that it is a single lap shear joint, which has an eccentricity that induces bending in
the joint which can affect the joints strength and fatigue life.

2.6 ATTACH RING AND ENGINE MODEL

The engine attach ring provides the structural connection between the aft end of the inlet and the
engine A1 flange. The attach ring construction is shown in the BFG Intranet under the lessons
learned index and the best practices sub-bullet.

2.6.1 General Mesh Refinement Requirements

The attach ring interface to the inner barrel and aft bulkhead requires the same circumferential
element spacing on all components to allow for a clean mesh interface between them. Having the
same circumferential element spacing and the surfaces broken along the fastener lines allows for a
clean line of near coincident nodes between the two components. The maximum fore-aft element
edge length should be less than Rt L
x
39 . 0 = , where R is the inner radius of the attach ring and t
is its thickness.

The inlet interface to the fan case occurs on the upstanding aft flange of the attach ring. At least
two elements are required between the fastener line and inner edge of the flange to capture the
bending stiffness of the flange. Similar radial mesh refinement should continue outboard of the
fastener line.

If scallaps are present in the attach ring flange, they need to be modeled.

The attach ring will be modeled with using the following MSC/NASTRAN inputs:
CQUAD4 without offsets
PSHELL with membrane, bending and transverse shear stiffness
MAT1 material definition referencing MATT1 for temperature dependence
MATT1 entries for modeling the temperature dependence of the material stiffness
TEMP entries provide the grid temperatures for the element

2.6.2 Geometry and Interface Modeling

This section addresses modeling details for the following three areas:
23
I. Geometry used for the Attach Ring
II. Interface to the inner barrel
III. Interface to the aft bulkhead
These details are called out in a separate section to highlight their requirements.

The geometry required to develop the finite element model is the mid-plane surface of the engine
attach ring. The CATIA surface should be broken at the inner barrel and aft bulkhead fastener lines
and hoop splice plate locations. A mid-plane surface for the splice plates should also be obtained
for reference. Fastener locations for the interface between the attach ring and the engine fan case
flange will also be provided.

Attach ring model nodes are connected to the inner barrel nodes by RBE2 rigid elements at the
fastener locations. Nodes on the attach ring will be the RBE2 dependent nodes with all six DOFs
assigned. However one must keep in mind that the shell normal stiffness is zero, so any miss
alignment will result in a zero in-plane stiffness in a direction. This is overcome by creating a local
coordinate system between the nodes and constraining the rotation in the panel normal direction.
Connection of the attach ring to the aft bulkhead needs to reflect the short cylinder bending
stiffness between the bulkhead and the adjacent row of fasteners to the inner barrel. This bending
stiffness directly affects the stiffening influence the aft bulkhead has on the fan case.


Aft Bulkhead















RBE2 RBE2
Inner Barrel Fan Case


Hoop splice plate interfaces in the attach ring are generally not explicitly modeled. Joint loads are
extracted from the continuos ring model at the splice joint locations for use in hand sizing analysis.
Hand analysis of the joint needs to consider if there is a single lap shear joint, which has an
eccentricity that induces bending in the joint which can affect the joints strength and fatigue life.

24
2.6.3 Engine Interface

The flexibility of the engine fan case significantly affect the reactions exerted on the attach flange.
As a result, it is best to determine the fan case stiffness as soon as possible.

The engine fan case model will interface with the attach ring at the fastener locations. The
connection will be with RBE2 connection between the attach-ring grids and the engine model grids
at the fastener locations. The RBE2 independent grid will be associated with the attach ring grid
and the fan case grid set as the dependant grid. All of the translation degrees of freedom and the
rotation about the hoop direction will be tied via the RBE2 element.

When an engine model is not available, the fan case will be assumed to be rigid. The attach ring
flange will be constrained at grids that correspond to the fastener locations. All of the translation
degrees of freedom and the rotation about the hoop direction will be rigidly constrained at the
fastener locations.


2.7 ACCESS PANELS

The access panel is shown as part of the outer barrel in the BFG Intranet under the lessons learned
index and the best practices sub-bullet.

2.7.1 General Mesh Refinement Requirements

Finite element modeling will follow the same guidelines used to model the outer barrel. Finite
element modeling of the access doors also needs to have a mesh density that interfaces cleanly with
the surrounding component element edge lengths.

The access door will be modeled with using the following MSC/NASTRAN inputs:
CQUAD4 without offsets
PSHELL with membrane, bending and transverse shear stiffness
MAT1 material definition referencing MATT1 for temperature dependence
MATT1 entries for modeling the temperature dependence of the material stiffness
TEMP entries provide the grid temperatures for the element

2.7.2 Modeling of Stiffeners

The access panel stiffeners are generally asymmetric cross-section curved beam-columns. These
beam-columns are typically stopped before the interfaces to the inner and outer barrel. As a result,
beam offsets along with beam warping coefficients are required to model the stiffness of this
element. The offsets model the affect of stopping and starting the hoop frames. Beam offsets are
modeled by placing nodes that are offset from the shell element nodes to the beams shear center.
The offset node is then connected to the adjacent shell element node by a RBE2 element with all
25
six DOFs assigned. Beam offsets are also acceptable, but are not supported in some solution
sequences. And the warping coefficient models the roll over of the asymmetric beam cross-section
under axial loading.

Beam shear loads normal to the skin of the panel induce bending loads in the stiffener flange
between the stiffener shear web and fastener line, in addition to the rivet pull through loads. Rivet
shear and pull through interface loads can be extracted from the finite element model through the
grid point force balance option in NASTRAN.

The access door stiffeners will be modeled with using the following MSC/NASTRAN inputs:
PBEAM beam properties including the warping coefficient for stiffeners
CBEAM
MAT1 material definition referencing MATT1 for temperature dependence
MATT1 entries for modeling the temperature dependence of the material stiffness
TEMPRB entries provide the temperature for the element

2.7.3 Geometry and Interface Modeling

This section addresses modeling details for the following three areas:
I. Geometry used for the Attach Ring
II. Interface to the inner barrel
III. Interface to the aft bulkhead
These details are called out in a separate section to highlight their requirements.

The geometry required to develop the finite element model is the mid-plane surface of the skin on
the door. The CATIA surface should be broken at the interface lines, frame and stringer fastener
line locations.

2.7.4 Bolted Joint Interface Stiffness

Bolted joint modeling generally affects the membrane load transfer across the joint. The rotational
and stiffness normal to the surface of the shell are considered a rigid connection.

The membrane load transfer is a function of:
i. Distance the fastener moves to bottom out in the over sized hole
ii. Contact stiffness between the bolt and skin when the bolt bottoms out in its hole
iii. Splice joint rotation from offset load
iv. Friction coefficient and pre-load in the bolt
The first two considerations are applicable to ultimate load calculations. Consideration three may
be of concern in some lightly loaded areas. However, friction in joints is ignored unless it will
make the situation worse.

Joint stiffness must, at a minimum, reflect the hole/bolt tolerances and contact stiffness issues.
First, the issue of contact stiffness will be addressed followed by the tolerance issue. The contact
26
stiffness for a cylinder in contact with a cylindrical groove was obtained from Reference 4. The
deflection, , is given as a the following function of load p in load per inch of cylinder length:
2
2
2 1
2
1
2 1 2 1
/ ) 1 ( / ) 1 (
where
)]} /( 64 . 1 ln[ 1 {
E v E v C
D D D pCD C p
c
+ =
=

and the subscript one corresponds to the property of the hole and subscript two corresponds to the
property of the fastener. Note that the deflection,
c
, is a nonlinear function of the applied load. If
one takes the load that corresponds to the bearing strength of the material, a linear estimate can be
made for the contact deflection. There are two contact surfaces in the joint, one on the skin and one
on the splice plate. As a result, the joint contact displacement is twice
c
. The contact load per
inch of skin thickness is then p=F
bu
D
2
, where F
bu
is the bearing ultimate strength of the fastener
and D
2
is the diameter of the bolt. The contact load is then: P=pt. The tolerance between the hole
and the bolt,
tol
=D
1
-D
2
, also must be considered in the calculation. If this tolerance is considered
as deflection under ultimate loads, the resulting in-plane stiffness is:
)]} /( ) ( 64 . 1 ln[ 1 { 2
) (
2
2 1 2 1 2 2
2
D D D CD D F C D F
t D F P
K
bu bu tol
bu
c tol
tc
+
=
+
=


As an example, consider a number 10 (3/16D) A286 close tolerance fastener through a class III
hole in 2024-T62 aluminum sheet. The minimum fastener diameter is 0.1885 inches and the
maximum hole diameter is 0.202 inches. Material properties for A286 are: E
2
=29.1 msi and
v
2
=.31 . Material properties for 2024-T62 are: E
1
=10.7 msi, v
1
=.33 and F
bu
=103 ksi. The
resulting stiffness for a 0.060 thick sheet is then:
lb/in 4 06 . 3
0123 . 0 2 0135 . 0
1164
)]} 1885 . 202 . 0 /( 1885 . 202 . 0 7 14 . 1 4 94 . 1 64 . 1 ln[ 1 { 7 14 . 1 4 94 . 1 2 ) 1885 . 202 (.
060 . 0 ) 4 94 . 1 (
=
+
=
+

=
e K
e e e e
e
K
tc
tc
Note that the displacement due to contact deformation is on the same order as the maximum
difference between the hole and fastener. The offset displacement for bearing yield is 2% of the
fastener diameter, which for this case is a displacement of 0.0038 inches. Introducing the bearing
yield deflection into the joint equation, reduces the stiffness by 10%. Displacement offset for
ultimate loads would further reduce the stiffness.

Stiffness of the fastener in shear is K
f
=GA/t=1.87e6 lb/in, which is forty times the stiffness
calculated above.

Additional axial displacements occur from the splice joint rotation from the offset load. The
rotation may be calculated from the equations for a tensile loaded beam-column with end moments
obtained from reference 5. The end moments are due to the axial load through the joint times the
offset between the skin and splice plate. The equation allows the rotation of the joint to be
calculated. The additional axial displacement due to the rotation is the rotation times the distance
between the neutral axis of the skin and splice plate. Assuming the skin and splice plate have the
same thickness and there are no shims present, the additional displacement from joint rotation is
then:
27
inch 00925 . 0 9073 . 0 170 . 0
3
tanh
12
2
) 12 / /( ) (
tanh
12 / ) (
) (
2
/
tanh
2
3
3
= =

= =
t L
Et E
t
L Et t
Et t
t t
t
L EI P
PEI
M
t t
m m
sp
m
m
m
sp



where E=10.7 msi, t=0.06 inches, L=0.754 inches (4D distance between fastener lines on the splice
plate) and ksi 75 . 25 4 / ksi 103 ) 4 /( = = = D F D
bu m
for a single row of number ten fasteners
spaced at 4D. The stiffness of the splice joint is then:
in) 1.26e5(lb/ inch 00925 . 0 / inch) 060 . 0 psi 25750 )( inch 1885 . 0 4 ( / ) 4 ( = = =
sp sp
P D K
which is more than four times the contact/tolerance stiffness discussed above. If one ignores the
stabilizing influence of the axial load, the stiffness becomes:
lb/in) ( 5 07 . 1
inch 754 . 0 6
inch 060 . 0 psi 6 7 . 10
) inch 1885 . 0 4 (
6
) 4 (
12 / 2
) 4 (
2
) 4 ( / ) 4 (
2
) (
2
2
3
2
e
e
K
L
Et
D
L t
Et
D
L t
EI
D P D K
EI
L Pt
t
EI
ML
t t
sp
sp sp
sp
=

=
= = = =
= = =



This stiffness is only 15% lower than when one included the affect of the joint in tension. If the
joint was in compression, the stiffness would be significantly lower than the shown here.

This offset between the splice plate and the skin not only reduces the stiffness of the joint, it also
significantly increases the stress level above the membrane stress level. The moment in the joint is
just the membrane load (membrane stress times the skin thickness) times the offset between the
neutral axis of the skin and splice plate. The total moment is split between the skin and splice plate.
The bending stress level for a splice plate thickness equal to the skin thickness is then:
[ ]
m b m total
m m b
b
m
t t t
t M
t t M

4
3 / 2 / ) ( 6
/ 6
2 / ) (
2
2
= + =
= =
=
=

The total stress is four times higher than the membrane stress level, assuming that there were only
membrane stresses present in the first place. These bending stresses are more of a concern for
fatigue conditions than for ultimate load capability. This is due to plastic deformation in the skins
and the large deflection and rotations that absorb/reduce the bending loads in the joint that occur
when subjected to ultimate loads.

The combined joint stiffness under tensile loading for a 4D spacing is then:
K
j
=1/[1/1.07e5+1/1.87e6+1/3.06e4]=2.35e4 (lb/in)
Which is the combined joint rotation, bolt shear, contact/tolerance stiffness. This combined
stiffness is dominated by the low contact/tolerance stiffness.

28
If one considers an end loaded 10 inch long strip of 0.060 thick aluminum that has a width equal
to four times the diameter of a number ten fastener. The resulting strip stiffness, K
s
, assuming it is
fastened at both ends (5=10/2), is:
lb/in 4 68 . 9 5 / 1885 . 0 4 060 . 0 6 7 . 10 / ) 4 (
2 1 1
= = = e e L D t E K
s

For this situation, the joint stiffness is 25% of the sheet stiffness. The effective stiffness of the
sheet and joint is 1.89e4 lb/in, which is only 20% of the sheet stiffness alone.

If this approach is used for a class II 1/32 oversized hole, the effective tolerance/contact stiffness
will be substantially reduced. The class II minimum fastener diameter is 0.187 inches with a
maximum hole diameter of 0.229 inches, which results in a class II joint stiffness of:
lb/in 4 64 . 1
0146 . 0 2 042 . 0
1164
)]} 187 . 229 . 0 /( 187 . 229 . 0 7 14 . 1 4 94 . 1 64 . 1 ln[ 1 { 7 14 . 1 4 94 . 1 2 ) 187 . 229 (.
060 . 0 ) 4 94 . 1 (
=
+
=
+

=
e K
e e e e
e
K
tc
tc

The resulting total joint stiffness begins to be dominated by the contact/tolerance stiffness. The
combined joint stiffness is:
K
j
=1/[1/1.07e5+1/1.87e6+1/1.64e4]=1.41e4 (lb/in)
Which is 15% of the sheet stiffness. The effective stiffness of the sheet and joint is 1.23e4 lb/in
which is a mere 13% of the sheet stiffness alone.

These results indicate that a Class III hole and a single lap joint reduce the membrane stiffness of
an access panel by about 87%. Class II holes in a single lap joint reduces the membrane stiffness
by 80%. Stiffness reduction in the joint is dominated by the contact stiffness and bolt/hole
clearance. Even a class VI joint only has a stiffness of 3.87e4 lb/in, which yields a 71% reduction
in joint stiffness for this 10 access panel. Joint stiffness reduction in the access panel membrane
stiffness decreases as the panel size increases. A class VI joint would yield only a 50% reduction
in combined stiffness for a 25 long access panel. Bolt stiffness has only second order affect on the
stiffness of the joint.

SUMMARY
Check bolted panels and the surrounding structure for two cases. The first case assumes no
membrane load transfer through the joint. However, the proper load transfer from the access
panel to the inner barrel must be accounted for. The second case assumes that the joint is rigid.
This approach avoids the assumption of a joint stiffness.
Or, use the joint stiffness as calculated from the above equations. Then connect the door to the
outer barrel through with these springs in the in-plane (membrane) direction of the shell. The
remaining degrees of freedom, all rotational and surface normal degrees of freedom, may be
rigidly connected.

2.8 PRESSURE RELIEF DOOR

The access panel is shown as part of the outer barrel in the BFG Intranet under the lessons learned
index and the best practices sub-bullet.
29
2.7.1 General Mesh Refinement Requirements

Finite element modeling will follow the same guidelines used to model the outer barrel and aft
bulkhead. Finite element modeling of the access doors also needs to have a mesh density that
interfaces cleanly with the surrounding component interface points. The interface/connection of the
pressure relief door to the outer barrel is at the hinge and latch locations. The stiffness of the any
door land interface around the perimeter of the door will be ignored.

The access door will be modeled with using the following MSC/NASTRAN inputs:
CQUAD4 without offsets
PSHELL with membrane, bending and transverse shear stiffness
MAT1 material definition referencing MATT1 for temperature dependence
MATT1 entries for modeling the temperature dependence of the material stiffness
TEMP entries provide the grid temperatures for the element

2.7.2 Modeling of Stiffeners

Finite element modeling of the door stiffeners will follow the same guidelines used to model the
outer barrel.

2.7.3 Geometry and Interface Modeling

The geometry required to develop the finite element model is the mid-plane surface of the skin on
the door. The CATIA surface should be broken at the interface lines, frame and stringer fastener
line locations, as well as, at the hinge and latch locations.

Latch stiffness may be modeled as CELAS2 elements using the appropriate stiffness in each of the
translational degrees of freedom. If the latch is capable of transmitting moment by heal toe affects,
the equivalent rotational stiffness also needs to be added to the model. Offset of the latch from the
outer shell surface can be taken into account using a node at the latch interface location. Care must
be taken to insure that the latch coordinate system is applied to the latch interface grids to obtain
the correct latch stiffnesses.

Hinges will be modeled with ten CBEAM elements following the same guidelines as the outer
barrel stiffeners. The hinge pivot point can be modeled through coincident grids (one from the
hinge and one from the outer barrel hinge fitting) at the pivot point that are tied via CELAS2
elements in the degrees of freedom linked between the hinge and outer barrel.

2.9 ANTI ICE TUBE AND EXIT DUCT
2.9.1 Anti Ice Tube Modeling

30
The anti-ice tube transfers hot air from the engine to the nose lip to provide heat for anti-ice
devices. The tube transfers this hot air through the inlet cavity between the forward and aft
bulkheads. It is connected to the aft bulkhead and the anti-ice nozzle assembly on the forward
bulkhead.

The anti-ice tube is modeled as a bar element due to the circular/axis-symmetric cross-section.
NASTRAN options used to model the tube are:
PBAR property card for bending and torsion
MAT1 isotropic material card referencing MATT1 temperature dependency
MATT1 temperature dependence cards
CBAR simple beam elements
TEMPRB cards to specify element thermal loads on the CBARS (not nodal
temperatures since the bars can be at a different temperature than the bulkheads to
which they attach)
The tube will be modeled with no less than ten elements between the forward and aft bulkhead
attachments. Note, that the connection on both ends of the beam, need to reflect the torsional
connections. If the CBAR runs normal to a shell element it is connected to, the normal shell
stiffness is zero so there is no torsion coupling between the beam and the shell elements.

Local stiffness of the bulkheads at the tube interface needs to be considered.

2.9.2 Anti Ice Exit Duct Modeling

The anti-ice exit duct transfers hot air from the nose lip cavity to the outside air. It is connected to
the forward bulkhead and then to the outer barrel through a tight bend in the duct.

The exit duct is modeled as a bar element due to the circular/axis-symmetric cross-section.
NASTRAN options used to model the straight portions of the duct are:
PBAR property card for bending and torsion
MAT1 isotropic material card referencing MATT1 temperature dependency
MATT1 temperature dependence cards
CBAR simple beam elements
TEMPRB cards to specify element thermal loads on the CBARS (not nodal
temperatures since the bars can be at a different temperature than the bulkheads to
which they attach)
For the curved portion of the duct, the following NASTRAN options for a curved beam are used:
PBEND property card for curved beams and curved pressurized tubes
MAT1 isotropic material card referencing MATT1 temperature dependency
MATT1 temperature dependence cards
CBEND curved beam and curved pressurized tube elements
TEMPRB cards to specify element thermal loads on the CBARS (not nodal
temperatures since the bars can be at a different temperature than the bulkheads to
which they attach)
31
Note, that the connection on both ends of the beam, need to reflect the torsional connections. If the
CBAR runs normal to a shell element it is connected to, the normal shell stiffness is zero so there is
no torsion coupling between the beam and the shell elements.

Duct interface stiffnesses at the forward bulkhead and outer barrel need to consider the stiffness
and size of the interface relative to the diameter of the duct and the size of the finite elements the
duct is attached to.

2.10 ECU COOLING AND FAN COMPARTMENT VENTILATION INLETS


2.11 NOTES ON INCOMPATIBLE FINITE ELEMENTS

Finite element compatibility can become a significant issue when mixing different element types.
These incompatibility issues show up routinely when modeling honeycomb panels with solid and
shell elements, as well as, modeling skin-stiffened structures with shell and beam elements.
Element compatibility is when elements have matching displacement fields between nodes. This is
a fundamental requirement for element selection when using the finite element method.

When honeycomb panels are modeled with solid elements for core and shell elements for the thin
skins, it must be recognized that the shell rotations are not compatible with the solid elements.
Shell element rotations implicitly result in a non-linear displacement field between nodes resulting
in an incompatibility with the linear displacement field of the solid element. This can lead to
phantom FEM buckling modes because the thin shell elements are essentially pinned at each node
and large in size. This is similar to intra-cell skin buckling (dimpling) where the finite element
edge length is equivalent to the core cell size. Buckling modes of this type are principally
dominated by rotations with little or no displacement, resulting in a random looking displacement
mode shape in PATRAN. Please note that these (large) rotations are not read into PATRAN with
the default setting. In general, the skin should be modeled as a membrane element to maintain
compatibility with the solid elements. This is achieved in NASTRAN by only entering a material in
the membrane stiffness location in the PSHELL card. When PCOMPS are used, NASTRAN
creates a PSHELL card with a bending stiffness making it incompatible with solid elements.

In honeycomb panel to land transition areas, where the solids elements transition to lands modeled
with shell elements, there is a need to couple the rotations of the solid element edge to the
rotational degree of freedom used by the shell element. In simple cases, this can be achieved with
RBE2 elements to couple the solid element edge rotation to the rotational DOFs. Care must be
taken to make sure that the nodal coordinate systems are consistent and to exclude the core though
thickness displacement in the RBE2 element. The resulting interface moments and shear loads can
be used with a beam on an elastic foundation hand calculation to determine the skin/doubler to core
peel stresses. Small element sizes are required to account for the coupling between the core flat-
wise stiffness and skin bending stiffness. In order to get the correct coupling, the shell elements
must be smaller than the decay length for a beam on an elastic foundation. This decay length may
be calculated from the following formula:
32
4
2
3
) 1 ( 3
d c
d d
d
E
ht E
L

=
Where E
d
is the skin modulus, E
c
is the core flat-wise compression modulus, h is the core height
and t
d
is the combined thickness of the skin and doubler. For a 3 lb/cu-ft aluminum core, an
aluminum skin and douber combined thickness of 0.060 and a 1.0 core-height, the largest
element edge length would be 0.13 (L
d
/2). This is very small compared to typical element sizes
used to model nacelle honeycomb structures.

Complications can also arise when modeling skin-stiffener type structures. These structures are
often modeled with shell and beam elements that have incompatible rotational stiffness. The shell
element only has only two rotational stiffness while the beam element has all three. This can
result in a similar buckling issue as found when mixing shell and solid elements. The shell normal
rotational stiffness is zero. Any rotation about the shell normal is modeled by the in-plane linear
displacement field along the element edge and not by this normal nodal rotation. Beam elements
are cylindrical bending elements that have a curved displacement field between nodes (defined by
the nodal rotations) along with a linear field defined by the nodal displacements. In the plane of the
shell element, the shell element edge displacements are linear while the beam elements are non-
linear. As a result, the displacement fields do not match up making the elements incompatible.

The third class of incompatibility is found when shell elements are attached at a right angle to one
another. This is commonly found when stiffeners are explicitly modeled with a right angle between
the caps and shear web or shear web and skin. This attachment scheme has a compatibility issue
similar to the latter beam/shell connection. Once again, the shell element edge displacement field
is linear in the shells plane (defined only by in-plane displacements) while the other element
rotated by 90 degrees has a cylindrical/nonlinear field defined by the rotations and in-plane
displacements. This can be rectified by coupling the rotations on this linear interface with the
rotations of the element edges that are perpendicular to the common cap/web element edge. If the
FEM model is required to capture cap or shear web buckling, element refinement on the order of
flat plates would be required. Generally speaking, cap and shear web buckling is best left to hand
calculations due to the many elements required to accurately model the stiffener buckling modes
coupled with the fact that stiffeners are generally small compared to the overall size of the
structure. As a reminder, the interaction equations should be used to assess the effect of stiffener
crippling with the global collapse loads for panels.




3) STRESS AND FATIGUE MODELS

4) LINEAR BUCKLING MODELS

33
5) ACOUSTIC RESPONSE MODELS

6) VIBRATION RESPONSE MODELS



34
7) REFERENCES

1) Harry H. Schaeffer, MSC/NASTRAN Primer, Sections 9.9.1.3-9.9.1.6, Fifth Printing,
Wallace Press, Inc., Milford, New Hampshire, December 1988.
2) Raymond J. Roark, Formulas for Stress And Strain, Third Edition, TABLE XIII, McGraw-
Hill Book Company, Inc., New York, 1954.
3) Getting Started With MSC NASTRAN, Users Guide (Second Edition), Section 11.3,
MacNeal Schwendler Corp., 1996.
4) Raymond J. Roark, Formulas for Stress And Strain, Third Edition, TABLE XIV, McGraw-
Hill Book Company, Inc., New York, 1954.
5) Raymond J. Roark, Formulas for Stress And Strain, Third Edition, TABLE VI Case 7,
McGraw-Hill Book Company, Inc., New York, 1954.
6) S. Timoshenko, Theory of Plates and Shells, Second Edition, Table 35, page 202, McGraw-
Hill Book Company, Inc., New York, 1959.
7)

Potrebbero piacerti anche