RULES FOR MODELLING CONTENTS
STRUCTURES
Introduction Selection of Geometry Appliccation to model .................••... Idealisation and Interpretation ...•........... Fuselage Frames Machined frames Pressure bulkheads ...........•....•........ Note on the use of CSHEAR elements Understanding nonrectangular panels Use of PARAH NOELOF for edge loads •........ Wings  ribs, spars, skins •...•......•..... Fuselage skins ...........•••.•.......•..... Trailing edge id8alisation ....•............ Composite panel idealisation .......•..•..•. Single anisotropic panel .••...•........• Multiple QUAD4 elements ...•............. The Nastran PCOMP facility .•..•..•••...• Through thickness CFC modelling .•....... Honeycomb idealisation ......••.••.•..•••... Joints and hinges ....•......•••••..••.•.•...
Mechanisms
General
1 2 3 5 8 10 10 12 13 15
16 5
Floors
17 19 21 22 22 24 25 26 27 29
30
Holes and reinforcements
Symmetry
.............•••.•.
32
34
Solid modelling ...•..........•...••.••.•••. 36 Strain Gauge elements .•...•.........•.••.•. 36 Rigid load paths ....••......•.••.........••.•• 37 Removal of potential singularities .•....... 37 Enforcing local displacements .•......•.•... 39 Very stiff items of structure .....•.•..•... 40 Merging a change in mesh size ....•.....•... 41 Obtaining average motions ..........•.•..•.. 42 Distributing loads to a structure ......•... 43 Changing the global D.O.F ...••.........•.• '. 44 Note on dependant and independant freedoms . 45 Exchanging freedoms, in RBE1' s .........•.•.. 45 Constraints and supports .... :.........•...•... 46 Statically determinate supports .....•..•... 46 Boundary constraints .......••.....•........ 47 Imposition of deflected shapes 48 Imposi tion of plane sections ..........•.... 49 Multipoint constraints using MPC's 50 Errors in Nastran elements 51 Straight cantilever beam tests .....•....... 53 Rectangular plate test .......••...•....•... 54 Methods of avoiding errors in modelling .... 55
Application of loading . Description of common cards used Definition 6f loading Checking the loading Symmetric and antisymmetric components Un it cas es Rigid body movement cases Point loads Distributed loading ........•............... Pressure loads Fu e1 pre ssur es Inertia loads Balanced cases Loading balance sheet .................•..•.
56 56 58 58 59 60 60 63 63 64 66 67 68 69
: 
.
i
section 3 b RULES FOR MODELLING Introduction The idealisation of the real structure into an acceptable geometrical grid, set of elements, sizes, constraints, materials, loading etc. has but one purpose, to enable the REAL structure loads, stresses and deflections to be calculated. The idealisation should thus represent all the load paths and realistic loading distributions set up. At the stage of formulating the size and scope of the analysis, decisions will be made which force the idealisation of the structure along fairly recognisable paths. For example in the modelling of a fuselage a coGrse mesh with skin panels bounded by major frames would imply the use of shear and rod elements for the skins. Unless the idealisation is carried out keeping in mind the full requirements, pitfalls are numerous. Overkill of the mesh size may create a dinausaur with no future; over simplification can leave out load paths which may be critical. The use of preprocessors such as Patran gives the engineer the ability to generate complex structures with ease, but this does not necessarily mean that they will be more accurate. . Intrinsically the solution of the stiffness matrix, application of loading and constraints, the generation of displacements, and the back substitution to obtain loads and stresses is an accurate MATHEMATICAL process. This does not imply that the results from any model are correct. Accuracy should not be confused with correctness. The correctness of the result is in the hands of the user. Overall accuracy, no doubt, remains linked to the source and quality of the information available. It is pointless setting up a complex analysis based on preliminary scheme drawings, approximate loading and uncertain materials. The question "How accurate is the available,data?" is rarely asked, and if qualified would generally lead to suprising tollerances on the results. The process of idealisation requires geometry to be defined and element sizes to be calculated which satisfy area or stiffness requirements, and only rarely both. Thus in most instances the model is a compromise, and the subsequent interpretation to the real structure is treated differently. The stress engineer should not use analysis results blindly. Modelling, loading and constraint errors should be eliminated in the checking process, but may still b~ present. Thus be on guard for unexpected results. The user should only launch into the idealisation process having first determined for himself a clear method of approach and a good understanding of the total requirements. STRUCTURES
I
t,
r
!
i
~
• r f~
,
,l
or if available on Anvil by conversion into a mesh via Patran. Both of these stages have several options.• Catia items. Source of geometry:The source can playa vital part in the scope of an analysis. c) Key diagrams . using a digitiser. The mesh may be created using Anvil and converted to a Nastran set of points or passed to Patran for further refinement. these may be interpreted several ways. d) Surfaces . and its application to the model. for example. . to
')
... The process of conversion an analysis grid is currently being studied and remains a speCialist activity. available on Anvil or Catia.SELECTING
THE GEOMETRY
There are two stages in determining the model geometry. A wing analysis.• giving the intersect points of major items. b) Scale drawings . a) Defined explicitely . e) Solid geometry .hand measurement.. locating the source. drawings gi~ing all the required geometry. would require a greatly extended time scale if the surface geometry was not available. not only in the accuracy but in the elapsed time to definition. These are used to form up the basic model using intersects as patch boundaries which are then put to Patran for further refinement. If key diagrams for a structure are not available a complex analysis should not be undertaken since by implication the results would be approximate.. This may be input directly into Nastran or Patran. ribs/spars etc. or to generate profile geometry by project~ ing a previously defined 2D mesh of points onto the surface.
Some comments are given below. modelled as bar elements. or z direction. geometry as 2 above and use offset vectors to section N/A. a) Flying surfaces. b) Fuselage shells . but in other cases should be modified to represent the best interests of accuracy or expediancy to the user. are relatively thin structures and thus in order to keep the skin membrane loading and stiffness correct. the geometry of the hinge points must be set up colinear in order to avoid 'locking up'. y. Geometry errors which create kinks both reduce the effective stiffness and impose unreal loads on .Application
to model :
The geometry may be used direct~y in some cases. is
e) Ring frames •. ensure the geometry planar. wings etc . at the outer CPt. It is better to put the joint coincident with a skin/frame/longeron junction. d) Fuselage floors. The joint loads can be recalculated and offsets taken into account in the subsequent detailed stressing work. h) Hinge lines . Because of the relative size errors are small whichever approach.. PPS offers an automatic conversion of coordinate systems in option 6. c) Fuselage frames •. modelled as shear webs and rods. g) Offset joints ..
. or the mid skin point. f) tongerons/spars .. ensure that these follow the correct (straight) lines. shear webs •. the mid skin thickness geometry should be used. the inner skin point.• beware of setting up joiqts involving offsets from the main load paths unless the offset has been modelled correctly and the local structure is capable of reacting the kink loads. See the section on the modelling of Hinges. a decisi6n should be made on the geometry profile which should be consistant. convert to local axes if nesessary. Rules for calculating frame flange areas etc can thus be fixed..the support points. sloping frames should be converted to local axes so that the in plane stiffness is not lost due to small kinks. Use a coordinate system with the hinge line axis nearest to the basic x. ie. geometry at skin points and internal points (observe 3 above) • •. There is a proceedure within the Optimisation routines which will carry out this task automatically.
.J Q
V)
en ~
:%
~
CJ..
\
.:c: 0
"
en
cC
') ..
tn
a
e
..


...~ .£J
c.. <A~ UJ\. 0
..J
.J
I
4.
VI
"%'
U..I
~ .:x.
'.. _

_ .lJ
"
0
Z
\
f.I....J
::..
a
<.
I
r
't~
.
Ql .
~ ~
uJ
uJ
4:fJ
:z::
w
.
Z
~
'r
fl
!. . ~
cd
c:t::
e '" 'Z
r ~ ~ ~ J..
Z
..
fA
~
<
~
w
'3
U.J
''A "....I..>I(j
.')
uJ
.I
3
z 8
.
:i
2:
1:
Q
...._.
t.:c: =
CI..
. _. toth
Vl
tJ
. ~'I.
tn
~I
~
.b
0
._..
\
uJ U.:>tI)
a.
F <l._~
__
.
(J
r
z"1
(1)2
A<: :z:: ~A
AUJ
>~
Y.b
0 LJJ
~ u.".
c:t
a. I:
oJ.:J
tn
~
r
.J.J:
~
u...J
\'Z
CJ7
Z .!:l Z
t.
_
.J
:z a
d: c:>I!
..
c..J 01..
'"
f:
tI)
I to
~
Z
Q
:z::~
t~
t.
'"
:::l
\.<t.b
UJ
..".I"...
_
_

<41
::>
(I
~ :..I
t:)
a..J......:CI)
~
~
t:
A
0 1:
..ij
A
k
I'
I
:::A
01..J....1 <C 4: w<.. _ _ ./I .

IDEALISATION
. as previously stated. Nastran element forces and stresses balance the applied loading. What should be clear is :The size of the grid The position of all major load paths The interface points to other componants or between frames and shell etc. Additional points may have to be created. only be undertaken when the full purpose and scope of the analysis has been determined.RULES FOR MODELLING
STRUCTURES
GENERAL The idealisation of the real structure into the Nastran model should. The loading and support requirements. The user must ensure the element and real structure forces are in balance when correcting for geometry and size idealisation methods (or changes to the structure). some general idealisation
h
. If a single element represents several items with different areas or materials. but in many instances these are modified due to a compromise in the selection of the geometry. Thus the general aim in the idealisation process is to model the CORRECT STIFFNESS. however STRESSES are only accurate for the real structure in the instances where the GEOMETRY is at the centroid of the item and the element SIZE is correct.it solves for displacements at grid points. This normally requires the specification of the correct thickness and area of the items. Thus in most instances the element FORCES are used in the subsequent interpretation the real structure. and the subsequent interpretation of the results will have to use the same relationships to obtain the real structure loads. the calculation of the idealised size will be linked to 'the selected idealised material stiffness. Nastran is a displacement method program . The following page illustrates and interpretation methods.
Co
4: %
....
uJ
.
~
w
~ ~
3
::c
1
fA
a..
z o
4..C
W1
...
UI'
(n'
l: LtJ
VI

4':
Z
.
z:
Q
Q.J
W nJ
(f)
u. ..
t)..
uJ
~ ~
Q
~
UoI
~ F "" t.i
(/\
~
Jf/)
!:
\U
<C.
I
f
f
i
al
a:
r
.._
f.. ~
oct
'U
u....
fII

ell
Z
)w
"

.J W
~.
t:
0
.
:$
...
... " P.I: <JJ
w
~ ~ ~
(. 2
.
t
uJ
tz
..
4:
..
~ 1:
w
Q 0
W d&
A
C4
2
2
. .J
..\
W 1:
W
.. )w
W
<
a.::....J
....U u."
::s
~
= .I
~ \U ~ ~ ~
¢
A 0
u. .l.t.Y
A
% W
\II
.
uJ
cJ
o·
V\.....c
~
Co)
r
.J 0 en
. ....l
~
:1
~
:.
4f)
..
~
"""
d:
W ~ LL1
r
fJJ.L
...l
s .J
~
l:
c::)
V)
cI\
~
'Z
$
l.
w
' cz:
Co»
...1 (4
q)
<t:
~ ~
Vl V'
c¥
:s
:s:
'% ""' t»
<:)
~
V\ tn
~
.J Vl
.b
\ll
~
~
al
&$I
uJ
<ll ~ ~
en
':::t:
A ~
w
.f
~
()
<C VI
u.. ~.J
~
i
>~
\.&.U
A
~
':It < ::
()
2
\U
¢
q!
~
tn ~
.
~
~ ~
z
w c.
%
:
w
c ' ao VI
CJ
~
u. S
4
:..."
'Z
~
:: "" .... 
(/I
«.
:s
0 4 c.. .J
%
.
Q2
....

c
tit
<t
F1

w
V)
CIt
~ ~
¢
..J.
C
Pf
'%
C
'"
:>
0
u.J....
o
'»
Z
eft:
cS:'
2:
~:
0:
..
"

a.b
W
Q
U..U
..:)
UJ
%
....
.:.
1:
~ ~
¢
x:
O
Q..
4
::
lJ
~
= ~
Qj
'Z
4:
c.J'
"':).
w
\LJ .
.
'X
'J
4:
<..J IoU'
z:
.
Z
~
Q
.
ell
1:
\6..:...
t
% cU
tw
uJ
q: cJ

~
Q
t. ~
<:)

'Z
<t
w e:t
~ ~
t=
CI)
0 \=
Ul
4IJ
~
~
(fJ
%:
:l
.J
.......
fl '"
tU't
4:
~
GJ
~
<:)
....
("l
r<>(
a: 3
0
~l~•
4:
Q...J
~
"'"
A
1&1
u.
~
ul
CII
:e e._
...::.
:I
~
a:
~
In <41 "'l£
.J PI
:0::
0
'
'Z
i
en
c.l
f'l
~
?'
•
..
..j
•.
0.
~
tit
U.... 4:
%
to
.iI
? +
~
+
.LV
... ..u
I &oil
~ ~
'
0
u.
0
c:
::z:
3
Q
A
4:
~
~l~
QC..
_
..
'%
_______
•
w· .!
z:
. ~
C
~
G.
Q
UI
CII <C 0
A
a:
..::.
j

~ :z 3 ~ 4:
"t
II
Us
'"~ "" ~ ~
VI
(S ~
}e.
Iq:'
0 01.Ia
1..
::
Q. A
Q1
c:::
\100 In
rr
r
:l
A
(/'j
"
ee
A
Ul
"Z
tn
._
~ ~ co..~
A
c:i I
w '"
III
\A
c.
...li
\)
.1
"
III <C
:r: 2 II.J
UJ

~
t
~ ~
"4
:%
Z
~
til %
I0

Q..
Q
•
tQt
r:
Q!
<J...
.
.
.!
..
s:
:r.
ell
<C uJ
(J)
'
~
IJ)
t
0
.
~
~
+
Z
'Z
Q
" :
...9
~
..J .10
P
~
7: 'Z
0
~
z: 0
w
(oJ
~
...... __
··. "":c
CIl
0
:1 Ul
l.:=..<C
~
..J
0
.
Q
to
+I~
.s
z
C
.
"" ~
til
d
Q.~
~
W
\J')
"Z
~ ~ uJ ~ ~
}
V'"
uJ
:: ...:..J
Q
Q
u...
u.
__
. ' ..U
"I
•
A
0
r:l
c:
'"
.
II
...."Ie»
~
I
.
j
W
en
~
..
J.
0.i
W
A
~ z ~
I/>
ca::: u.:.~
UJ
" t1
fQ
t
4:
Z
~ ~ en
~
~
~
s
I
d:
u....
II
I.:=..Q
c::
...
1L
..Q
~ ~.. ~.lol
A
<:I
& .... '..
.)
...
ul
):
~ ~ I<
i' c
4: A
"Z
~
U1
3~ c~
f8 
ul
<b
ci.
w
J:
"%
I
<
"l Ql
'..
q
Q...
~
<C
I'
•
'" ~ '"
Ii>
"" ..
.... ::~ ::"'_:"'.
"
c
"Z
b
'"
uJ
IU
q:.
s:
Cot <.. ::! CJ 4:
Q
_.I
" l
ce:
H
u..J
a:
A
(/'j
.b W .:::_ ::.
!J
~
tI
0'"'1:
'0
::
Ul
til
4.
..
":.J
l
Q
t:
u
~
01
.!! ....
VI
a:
OJ
A
::a
s
~
..
.
A "'7
...
.4:
(J
0 .. : .!:I
~I
IU
A
c.iI
\01:1 VI
A
a: ~
J:
.
~
A
a:. ~
<.J
4)
..
t= ~
IU
\J..J
IQ
a
u.
UJ
{i.l
c...

::> 'Z
x:
c¥
.
w
I4.J
0
"Z
~
< :::1 ~
1M
~
~
0
ct.
•
'z
A o ~
I
f
f
~..u
~
l
...
B.F. Fuselage Frames These can be in various forms. The stresses in the real structure can then be calculated using E. Stressing method : Some interpolation is required since the skin shears are applied at the grid points and thus steps appear in the BM at the grid points . It is assumed that the fuselage skin effective area is modelled as a separate CROD and thus the frame loading can be readily found. The element sizing is linked to the selection of the geometry and the types of element chosen. and the analysis stress levels.T. The moments at the centre of the elements will be correct. The geometry should be set at the average depth between flanges and flange efficiency factors should be used in determining the CROD areas.use the average. A method is shown below :
. Heavy Ring Frames :
The idealisation is calculated to maintain the correct moment of inertia and N/A position.
a) Shear web and boom.I2. Stressing method : use the bending moment calculated from the boom end loads and shear values to carry out detailed stressing round the frame.O. Light Ring frames :Two methods of idealisation are possible. CBAR elements are specified using the outer geometry and offset vectors to the N/A of the section. ring frames. Note that additional D. machined part/full frames or pressure bulkheads. b) Bar representation. Make sure that the bars are coplanar (eg Xconst) and if not supply all properties A. are required for this method. If the outer CROD includes the effective skin. some load sharing is nesessary before the outer flange load can be determined and thus the frame bending moment. thus in the examples that follow the sizing method is shown explicitely. the geometry being dictated by the fuselage skin and internal skin if present. The frame is represented as a shear web and boom using CSHEAR and CROD elements and sized to combine the effective web and flange areas. Remember the default for shear flexibility is ZERO (infinitely stiff).J etc. The frame bending moment and shear values are output directly.Conveniently the various A/C componants can be covered seperately and the recomended modelling techniques are shown below.Il. INTERPRETATION of the results is linked to the method of idealisation and thus guidelines are included where convenient.
::
a::
r:
~ <It <IJ ~~ ~
CIt
uJ
~
Z

r:~:.
....l
lob Z
CI.
>
:I:
\ %.
(j1
W
¢
at
A
'"
Q
uJ
:z:
::r::
I
.
+
III
I
I I
J
VJ
'%
at
es
H
II
(I).I
z:
= 1:
4::

q
ClI...
" c:
" .
1
<r.... A 2
<lC
c::
II)
e
A "Z
%
<t
~ r
tA
4.::n
c:::
.c
>c::C !l.
A
uJ
a.
j
c.
:?
:::l
(II <JJ
s~ :::
Z
VJA...)
uJ
U7
t:
.
'5 c)
..
Q
o
.
:::rt
s»
w
I
2:
0
~~ ~
<r.J
~
.
W)
u ~
:::z fU :)
'l:
tI)<
~
4:
(J)
~ ~
"='

\i
J
7:
..)
'

.
Vl
j::
•
at l()
.
... Cit
'::
..JJ
0
"1
<:I(
'"
~
I
..1
'
c.....
'j )(
>..
N.::
. ..
:n
Z
c
!
s
<C'
.)
IJ
Q
Z
::i
A
til % o
F <.l
..q
:j
() Q
~
tJ
IJ
:il
~ ~
'4J
c'
.I
I
c .l
Z cC
s
t
C"4
cr
)(
r
u.
:z.f4
...w
fA
x:
l0
..w_
eo..
using CRODS for the stiffeners and CSHEAR's for thin (non end load effective) panels and CQUAD4's for thick end load carrying panels. Idealisations without offsets would require recalculation of the bar loads to take into account the offset shear loads.remember to resolve the loads into the axis of the edge concerned . Provided the areas and thicknesses are accurate no corrections to the real structure are nesessary. This enables the model to be kept simple and does not involve any additional rotational D.in which case the bending moments will be correct at the centre of the elements and need averaging at the ends.F. However.GPFB'S are produced in GLOBAL axes.
. The CBAR's should be used with offsets to their correct N/A position . modelled and pressure loading supplied as part of the loading case definition. the pressure loading effects being taken into account in the detailed stressing. Pressure Bulkheads:For an initial simple idealisation of a fuselage model the pressure bulkheads are usually modelled for in plane loading only.O. CSHEAR's will give the panel edge shears directly whilst QUAD4 elements quote the centre of panel values (in element axes). The stiffeners should then be modelled using CBAR' s. for more detailed work the bending effects should b. Thus all stiffeners have the correct area but are not subject to bending loading. Using the GPFB for a QUAD4 panel will enable the edge shears and loads to be calculated .Machined frames:The idealisation should follow the stiffener pattern as close as possible.
.a
'_)
\
c/. u.s
A u.~
CI' 'Z: <C
....)
"Z
<..J uJ
AI
..::I
Q
01.
Qo
'Z 4:
. C) u..
f'I
~cn
~~
~4 ex:
I
c:a:..
z¢
..)
cr:W
IU . 0 <It at 0
~
CI\
~
II.lit¥.:) It
ol...I)
:z
~
'Z
uJ
1
:..
U..
U.
...
..
4::
...4._ Z
s
w
...~
.
<l!c
Wz
e
I/)
\
II')
~
j ==
II)
A ~"Z
:d
<t
t¥
IA
..
dl4'll
'"
...
l
t.
~
0
w ...
uJ In
...
uI~ .
l
.
....J '
~w
%
z
. ''
:: IofA c
qJ
j
z "& ~ A
Clot
(j
'>1..i1N
crl.l
".".
f..! 'Z
.
~ If!M
01..r~
w I..c.....x ~ .
.u.t...
:>
10.J
fl..
ac:
4:
._
.:
:..
tt
5
~
c:e:
' "t:
0
.J W Q In C&:..g
<:l 0 QI e.J
<.
VI ...Q
til <..
5=
0
:i
Ao oj.s..
t ~...
.
o
x
. A
e
V
cr:
~x A
j
...
!J
<oJ
.J
.
..
.....I
Z
lZ
'" e
_ z
Q.
(.II
.~ ~ .l

..
::l I
::l
<.s:
...
.~.
~
~'l ...&
z:
\&J
.
j.J
'%
::s:
<...I
4
'::>
~
'2
:::::
%:
W % tJ
U .
(J
~
c:::
~ o (J
'"
c::: uJ
VI
..!)
~
IP
w
CI)
V\
~
0 ...I
)(
.
Z
:.a
I
I .
1
!
.
V\ III 0
CI)
W
C.N I
Alel
dJaI
VI
.1>\..7":'"':..~
~al
Col ~w7
~ !:!
::t
'Z
I
....
".J
%....II (..r~ VI u. r=1
I
~
<
VI Q( <.1
""
I.....
~
Z
Z
'Z
<
\&J l:
..
~
~
<J.oJ
0.
..J
J
!
Q.tJI
G.lJ
u
uI
4:
c:
~
=
\
..c o
1:
~
CI)
.
""
~
. ... t
..
d
.u.
()
c
c" UA I.~
~
I )0(
..!I
Q(

"''
~
I(.
':% ~ 2
c..
lit
.a.u. ~ .g
0
f1
\.. '%

''
.a...oJ A
en
'W
VI
III
1::.
e
"Z
I
uJ ... 0
oj ~
U ulw
u..
01.. If this extensional stiffness is not present the panel will be singular and would probably lead to a FATAL error or excessive displacements.t ..
In calculating flange areas the web thickness componant Aw/6 can be included using the F factors by setting the required F factor to 1/3. a) r < 1.t.. a function of the panel THICKNESS.. WARNING .
2
1.4 F2 . two methods of approach are possible.Wl or 1/2.!'
Note on the use of CSHEAR elements The CSHEAR element normally carries shear loads only and thus should be surrounded by ROD elements to carry end loads in the equivalent areas.
(
". USE These factors can be used to remove the singularity ring frames across the frame depth.W2 If the panel is less than fully effective. Without F factors the corner forces are of equal magnitide and a~ount to the shear loading on the edges.F2. l" '. With F factors the corner forces will be different and the shear contribution should be removed to determine the end loads in the panel.
f
i
!
.t.Fl. 2.factor for areas on sides 1.. the areas are set to 1/2. b) F > 1.01.. A facility to give the shear element extensional stiffness in either direction exists using the rl and/or F2 factors. For rectangular panels orthogonal to the global axes the corner forces are identical to the GPFB values .
.
I
I
:~
I
r2 rods
r1 .• 3 If F1 or F2 are set to 1.0 the equivalent rods would be set to ie.
•
AND
fl'ange
RODs
e.4 and 2. .factor for areas on sides 1.0 would lead to an over stiff
DIRECT LOADS IN PANELS .• The only method of determining the end loads carried in the SHEAR panel when an F factor is used is to look at the corner forces FFROM4 FFROM2 etc. (See PSHEAR description)..W1 or 1/2.F2.t..see next page.F1. for
use F1 or F2 to create equivalent rods across frame.
:
I

wa
r
Wi
J.2 and 3.t.tor 1/2... fully effective for end loa'': areas of 1/2. the areas are set to 1/2. Using F1 or F2 bending section.3
I
(
4.W2 a function of the panel WIDTH.
Z{.:: 3~=::: . 113 '4
=
'12.
(!4.140
.sctS"·3+
'ObS'01 .
.
2fo·4

.
CoNT£.'t7
. For nonrectangular panels the GPFB output can be used to calculate the edge loads provided the output is transformed into the edge directions and components from adjacent edges removed.AR.see pages 6 and 10.t~\)TI~
..... COlfPOHEHT
of DGe S"if.
2S~·30
I__:
b
2! o 40
L
~.)NT~. ..
lo~h
It..b3·lo
434' ~4
3o·h3
t

'S
.. Consider the example below....f~tJTIOfl/
• __....z.H
SHEAR.

='V=__
A~:~·· ~
SS~'II
S~c.... and these will be identical to the panel FORCE output for both CSHEAR and CQUAD4 panels .2.UNDERSTANDING
NONRECTANGULAR
PANELS
For rectangular panels the GPFB output can be easily used to calculate the shear and end load components of the panel loading.b~·l
'h
SoH
.. Su..
T
tv E..
I ...
HS"
1
of ?AtJEL
s~cJtR.
S25'·7
!
~ S"~~
~
3Ib'Z
\113·7
CONT£I~UTtON
~ :...
ouc. 'T1H:.
cr:.. ~NEL
nib
IS TKEN ANb LE.. ..
~/
lo...ol'lbS
L...1. ~+P'l..b l>/f:Fc~cNe.. bib E.... HA)(IHUH 3e.!. 'lo·/lj.
'f'l.Nl:> lot\3:)S
fAWE.CVLAT£.S
CAN .
\lib
L...l...ve.CtJ"ATU
Po ( N TS
..
eAcH
TM\S Ew.GTIONS
~£
/
.lS e.·g
t
\t
~ ..
...
CM. FR.b \.AtI
eFf:£eT!uf.~
AtIb
HALF AT
~1':.l.oH : 5·l7 1'1/ .cl\~ i"'HE..lUl::
ttlH IHvrt
S rRESs
s Il:>~
CJH. 1'1·7y:2 '" 3· <?S NJ ..'i1f
11+uS
a ~ :f!:_
'~Pt.
(XuA:b4
0
urPUi
/
cAleUL.ot./lJ •
'/'1. .:
LE.l:>GE.AtioN
E.Z1S'5'
ItS9·Q
s'""I.
i\~OUr...
~E.
fANE.
J:>fRE..£
oN
() ~/N ~
'i"HE GR.
...
1.R fC.EDt\e. Flo
2
e F•+ rt SHEAR
E. Fz
b
\U~f ~E
WI
w
t \~ \H(~N s s s
bTH
PAN £.
CONTRI P>VTloN
IS
'/2..
INlo
EDGt.£Nr
E...I~uTloNS
FRoM AbTA~£NT
~b~S
.FfR0Jc..ONTR.... L.TA N GULAR./ I
• It:
c.~OLVe.
I.
ANY
eDMPONANTS
A"J>J'Ae.]..tF2
~
P~NE:.N GTH
El::>G£
STRE.
~F.rr£
.
f't ~·~Ih
2.E.1~:l(ONS
LD
Glo'BAL
A2<E. S S
cr::
'1zw
F.
..Fz.
'12
J:)(f..___.....
___
_.
SIGN.l
IS.L.
z)
NON .P.v c..
__..
LoAbS
FRDT1
INto
f_J)G'£
l>IR£CTroNS
• •
REHouE.....
ro~
PAN (L S
EAC!\
Sl>~
Rf:.E.
D~
'A..L NoRHAL TD
1$
1
z..
SUH
Z
CoN~t~uT(oJo.
LO~l:>S
bIRE..JJGfS
~Q. rt.S
R.
PtNl:> '>

PAR~M
NOe.t=.CTLY
s or
of
DPPOS...2 LE. 2
.X I
ke
v.~DLVE.LO~
WILL
DUrPUI
'f.i
'NOTE
:
USE.).DcSF_b F\S ~ 'S DU E
.. LoA ~ S
RE:So t.
2E. :brReC TIONS
1t£t10VE.
I
1..
C..bGE....
I Nro
\ xt
E..E
S 11EA~
E...\~'1
rs
NeT
oRLttOGDNAL
J>11<..L
hl:b
F./"
L
(i._ f.'l..
ei
t=".l:> ~
_.
0 0.:1(+02 6. The output is requested by using a PARAM card in the BULK DATA PARAM NOELOF
1
The panel/element loads in the edge axes are output in a similar manner to the GPFB.0937Z.geeHoE..{l.654%lftOZ 0.0 0.
IA NfL
'1.4115:25E+02 0.0
Z .D JIOlHTIO
]
fRCtI
0 " 0 .0 0.)4.5
o.~
0.0 4.344448£:+02 2.lI S
~~oVE.
1
9 6
Z
O.0 1.0 0.6612iOEt02 2.~ o.0
o.
THoSe..9111s.0 1.0 0.0 1..
~R~'
DN
l>£N TIc.
H:J It
Qu1"PUr.0 0.Ol 6.'::EtOZ 0..0~7<l:'!EfOl IS.V
0. E N T S ) to LOADPOINT
'ORCE':'Z 1..0 0..OZ 2.
t. Only if the panel edges are orthogonal to the global directions c~n the GPFB terms be used directly to calculate the edge shears and direct loads.Z65'755E+02 3.(j
O.b O."<.5[+02 5..lS LO. FRoH
ChI.1el .O"l0eCE+02 It.0 0.004949Et02 5..t!
o. 70 PltE:V
e.0 3.0 2.."
0...8EOl
<.0
5 5 5 5
1
5
..Ol 0.0 0.0 0. .ti
ROO ROO ROO
2 It 6 1 3
0.766S43E.~
Use of PARAM NOELOF to obtain panel edge loads GPFB output is presented at the corner grids of panels with forces given in the GLOBAL directions for the grids concerned...6 6
.0450Z7f+02 2. .0 1.A'T£. If the panel is skewed a complex process such as on the previous pages is necessary to deduce the edge loads.0 (
.590H3E +02 3.! .AL.0
E.
Tltf..e
e.45'.ti
0.771zec.0 0.01 3..t.a
0.9815.u I...16Z0b3E+0? 1..7
.la.953369Et02 0.0 0.0 0.136974Eo02 5. POINTZ
REFERENCE 6 It 0 0 0 5 0 0 6 12 0 0 0'7
1
] ] ]
12
10
.1.~
O. of output for the previous
'OR CE5 ItEFEREHCEPOIH'TS tIOttENTl 0.c A.6Z0Z5eE...0 0. oEOZ " 5.l>G1:.. 6]5554[t02 0.0 0.0
O.:E+02 3..453421E+02 4.0 t.. ACTICf. 5E .0 1.
Le'
..O 1.l>
leu ~ PAGE.0 0. Loads are positive in the direction of the reference point to the load point.80.0 0. and from these the shears and direct loads are easily calculated using the 1/2 sum and difference method.&
6 11 12 11 Z1
5
QUAD 4 QUA04 QUAD 4 QUAD 4
l 5
7
ROO
ROO ROO ROO
11 5 l 7 11
11
5 0 0 0 0
3
4.0 0.. N
example is shown
IHT~IlH4l El [ " EHT I IH'TUH.362942E+OZ 1. This is because the ElF forces are ~ement Internal Forces and thus balance the GPFB loads.
6
10 11 U Z 1
QUAD4 QUADlt
ROO ROO ROO
12
4 6
It 6
o.0 0. vector wise these ar~ opposite to the GPFB loads.0 O.t..:<.E+OZ 1.0
lLD1fHT
ID
Z 1 IS
fLEt1EHT TYPE
QUAD 4 QUAD 4
REFERENCE: POIH'Tl
6
fCJ1CEl Z. There exists within Nastran a facility to output edge loads directly.0 0. A typical example below.883735E+02 1.S" f.0
4 4 It
ROO ROO
QUAD 4
5
0.583323E+02 3.lel''40E+02 6. . .0 .
The normal idealisation is by using ROD and SHEAR elements.~... "
_ r' ~
u. GEOMETRY .
..'" /". BARs should be used with offset vectors. l. . i
tzz:
(L
.FLOORS usually these can.O.~
I
j

.t. leaving the RODS only for the panel end load paths. put into a local axis system if nesessary._'L.
t_.. are needed.. A slight out of plane kink in any of the grids will destroy the load carrying capability...
'.
. and may be required to carry pressure loading. t: /1.
. make sure the floors are planar.. ' For initial project work the pressure effects can be added in the detail stressing. If the panel has any significant stiffener offsets and/or is subject to pressure loading.0 on the PSHEAR card.
The panel effective end load areas can be set up automatically by setting F1 and F2 . the model reduced to in plane effects only and thus no rotational D. the ROD elements representing any stiffener areas plus the effective panel areas.G~ \. be regarded as being fully effective in end load and shear.. in which case each effective rod area will be 1/2 the panel width x thickness.1.F.
Thus the idealisation should aim at providing the correct stiffness for these items... The geometry should be set up at the mid depth of the skins.~
t ! . the'spar/ rib webs as SHEARs and the flanges as RODs. Since the CSHEAR element only has ROD elements along the skin edges~ some stiffness must be given in the through thickness direction. If one of these methods is not used excessive displacements will be generated since the webs will be near singular.
e. thus use the appropriate F factor on the PSHEAR card. Alternatively use RRODS.RIBS. linking only the through thickness D. fins etc.O. SPARS.F. The skins should be idealised as QUAD4 elements. between top and bottom skin grid points. "
.
I'
i
I
t
[
I
if
~
I
~. SKINS Wings. thus the loading and stiffness will be correct.
l
'~.WINGS . The spar/ rib idealisation should aim at obtaining the correct moment of inertia .in a similar way to that for heavy ring frames.. are thin structures domonated by bending loading for the skin design and shear loading for the internal spar/rib construction.
.
Cut. F2.SE..SS
AI
IN
AN'!)
tJPPeR.ORRE.:
TO SPAR..:
IN
\"'OWER.:..
<c::
l.
SPAR/R..RibS
S rtOUL.
F.
AI
SJJF\J>4
t..J:> \
R..
f12..TR.ATE!> PR.CT
. ~
N _~ ~
cROb
C Roo:%) A ReAS ARE C3~\. IS
MOMENT
F :r:NE R.
01..NT S E.:.
G.
MOI1Et.vctf
()t...\YIN GS
~TC..CTloN STI~FN
r"J:)E.PTI1
.
..E.~uL.1.I~
~AcTcR.~.b
fHJ. ~oRReCT' CAt..S
~s S
GuA'b"+
.NCE... SPA~I R.C5
FHtE.:.
PLANs£.iJ
c::::
:z::
c.
A2.ol>
PlRCAS
P'I '3..ott:0
1:~E...
SE. + A~ ~~
AI\).QUA~'t
\'U.. IS
Et=FEC'TWE:.l<
0~
F
GEDME.
~
':1'2.ATG..
PSt\fAR.. \1.'f F\CcuRArE.l>
1$ IF
t.
'By
~..':'
oArl~~)
AN»'
NOTE...
ENl:> loA'b
(0
GltJ£
TrlROVGH
THICKNess
StiFFNESS..
crlvE
0
tw
~z
~HCAlt TltICl<NE....SS
QUAl>4
~
(J')
1:.Nc~""A
'.Tl r. 
=
I '..:...
.:..
us E. o s..ISATcort
REPR.AL..T
IJ
P
AT
SKIN
rUl>
l>e..:D
SPARS
E.P" .1l>S ftRE
Sl~E.Ij'Z
H E...
STRE.
'BY
OA'2.
')
FLANGE.'2.
CQUAD4s. which should have been modelled separately. if QUAD4s are used for skin elements globally the structure could/will be significantly overstiff.O.PSHEAR card may be used to provide this. The panels should be bending QUAD4s and any longerons or stringers should be CBARs.
f
t
WARNING .
.).
I
f
f
r
i
f• t l
i
\
!
~
The design condition will dictate the material ie. and this increases the running time considerably. Alternatively if the F factor is not used the CROD areas should represent the sum of the effective end load and longeron/stringer areas. For thick ducts etc. It is not possible to represent any skin bending (pressure loading will load the frame grids only) or hoop tension effects . This forces the model to have rotational D.
Coarse Mesh Fuselages :In the coarse mesh idealisation the skin between the modelled frames is represented by single elements. The mesh should be sufficiently fine (minimum 4 panels) so that the skin effects can be seen in the output. In this case CRODs would only for the longerons etc.
into two groups. which would lead to incorrect load paths and deflections.
buckled/unbuckled
etc
However it is possible to divide the methods coarse mesh and fine mesh.tudinal ·CRODs will represent any·longeron or stringer areas.these must be taken into account in the detail stressing work.•. a reduced E for a buckled structure. Pressure loading should now load the midbay poirits. ie be required skins the effective areas may approach 100% (intake and in this case the CSHEARs may be replaced by fully end load effective.FUSELAGE
SKINS
The idealisation of fuselage skins is probably the most difficult to model correctly in that the modelling is dependant on several factors :a) b) c) d) the the the the size of the mesh frame support skin thickness design condition. The CRODs along the panel edges bounded by the frames will be sized depending upon the effective area of the skins in the circumfrential direction.
/
f.F. The skins are thus represented by SHEAR elements and the effective end load areas by ROD elements. Long~tudinally the skin will be fully effective and thus the F factor on the. in which case long. Fine Mesh Fuselages :In a fine mesh situation the skins between frames are modelled in sufficient detail to enable skin bending and hoop tension effects to be predicted. stiffness used. ESDU 71004 gives formulae for the effective areas for single and multiple frame attachment lines. The CRODs will be independant of any frame flange areas.
f'.'
TRAILING
EDGE IDEALISATION
The idealisation of trailing edges has in the past been of minor concern since for purely stressing purposes local pressure design has been critical. With the advent of optimisation methods using aeroelastic data the trailing edge displacements are of interest, and for flutter the T/E stiffness and mass can be critical. In reality most trailing edges do not meet at a point and it is vital to represent the real finite depth. The usual method is to include a slender QUAD4 closing element with a thickness representing the closing material. The skin elements are thus separated at the T/E and now represent the correct stiffness.
1Rf11l\NG
E.bGE.
Il>E.AL1SA,i!oN
REAL
STftv~TuR
E
')
4.
'.
•••

_.

..:~~.
(
'.'"~~ ,;;';'_~~....;..   '~=='.:.:..~':'~' . =~\
~.
=====
\
at. tP
"" w
II)
'Z
.
'_',
~~
<It.
')I: (I)
~ •
u..
tv
... 0
.~
.,..._
_j
.n
"
~
J
"
", ,
_;
.J

1
I,
'}
uJ
A
...
<
:z
"
r
.
.!.
COMPOSITE
a) b) c) d)
PANEL IDEALISATION
<
Composite panels form a large part of present day structures and several methods of idealisation are possible. single QUAD4 anisitropic panel mulitple QUAD4 panels Nastran PCOMP facility Through the thickness modelling
using 2D elements
Correct bending stiffnesses can be obtained with methods a) and c), method b) is used for membrane effects only (usually in optimisation work). a) Single Anisitropic Panel :ror membrane only panels the QUAD4 material is specified by a singleMAT2 card, the stiffness matrix being calculated using the ADS cr02 program. The fibre angles are specified relative to the 0 deg direction, the orientation of this layer can then be input on the QUAD4 card as either an angle or a coordinate system, as shown below.
200
201
o·
C:J." \. __.,..
HATc.2IAL
~~Clt\ \ ~
'l:ll
R.E.C..TloN
ot..
sr~I~I~1>
A~ AN'l..E
E.:l>G £.
eo~
Sysn H 1
101
Nastran QUAD 4 element CQUAD4 or CQUAD4 1 1 1 1
would be specified 100 100 .101 101
as 200 200 17.5 1
201 201
where the 1 in field 8 would point to a CORD caret. WARNING •• the material direction is the PROJECTION of the X axis onto the panel. Beware of panels out·of plane with the CORD specification. Consider a layup of total thickness Layer +45 45 0
90
4.0mm
:
Thi"ckness 0.5 0.5 2.0 1 •. 0
% Thickness 12.5 12.5 50.0 25.0
1/2 unit t .0625 .0625 .25 .125
CF02 should be run with layer thicknesses totalling UNITY, option 0 giving the stiffness matrix as shown below.
PROGRAMME CFB2 ISSUE 2  ANISOTROPIC PLATE STIFFNESS ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••
SYMMETRIC MATERIAL
M
PLANE El
1 36BBB.JIgaB
NO TEMPERATURE 8:,) NO. OF MATERIALS· E2 6JIgB.BagB 1
OR MOISTURE
C;lZ
USED B.BBS MAT. NO. NU12 11.3BBS
NO. OF PLIES.
AHGl E SF • DEG. DATUM • 3BBJI.JIggg
66 0.IIBSSS 45.IISggB 9H.BBHHHBBBE .B625B B. r
~
g.B625B 'S.BBHHS g.SSgSB II.BBfUg 8.SSgBSBSS£+SB S.79856396E+B4 S.1I62SS S.SBBB8 S.B625g B.35851282E+S3
8.223S536SE+S3
MATRIX
S. '~
I
f
INPLANE
g.6BIIS'g ~5.35851285E+83 1I.35B51279E+0'3
.ISSSS
.1B93H83E+85
S.S'62SB S. 244 14863ES2
I
RIGIDITY
B.r
nl!(t:::.79663344E+HS H.SSgSg 4S.22868916E+84 S.J
. 8B
STIFFNESS
MATRIX
B.974166B2E+H4 B.16796389E+g4 S..BgSSH s.0
If BENDING effects are to be modelled.B625S S.Sgggg (S.3
F4 FS 9741.SSSSS 9B.0 and a separate MAT2 card specified for the bending material.BBBBB (S.8
I
~"
F3 79663.LE 'S.6ggBS' 4S.
i
!
'r
I
f\ (_ I_.BBHHB 45.2SSgg S.
PLY NO.1779B41IE+84
8.
1
1
1 1 1 1 1 1 1
IB
11 12 13 14
1
1 1 1 1
15 16
1
1
II.Fcc._IN.79663336E+SS 8.97416582E+S4 S.S62SS S.SSggg 9i1.BIISS8 45 .SBBBB II.47H337BIE+H5 S.PSHELL PID
1
4.16796389£+84 8.IIfJSgS 911.ZUIU63ES2 8.B625B B.'<'::
.6535S332E+84 S.BSSIIS (5.1.
".BBgSB 9S.ggggg 45. A typical stacking sequence is shown below together with the CF02 output.
S:ll\C ~
\..BBBSB II.S625S S.
1 1 1 1 1
1 1 1
5 6 7 8
ANG:lE 'WITH DATUM 4S.I
U~~C
UI.T n(l(_~('J
I
f
1
I
\ .0
F8 10934.35B51285£+B3
MATRIX
8.SSgSS 45..S625S g.HBgSB II.974166S2E+84 S.gSBBS
WITH DATUM
ANGLE
1
l
\.gBBBS 8.lBBBSSgE+1I11 S.SIISIII1 II.862SS' 1I.gllgllS' •• gSIlIlS 9..12SSg S. BSBBB 45.B625S g.
i
::"0::..gBBBB fI. 47S33777£+85 S.SSggg II.B625B S.11217456E+S( II.SSSSSSBSE+SS
STIFFNESS
MATRIX
S.IISgS8 9B.12211477£+8(
)
S.PLY NO.gSSSS 9B.1S934483E+B5
f.S625B
ANGLE
II.gSSSS lI.
2 3 4 5 6 7 8 9
THICKNESS
S.BBgBS 9S._) < ~ ~:::
!
I
• .195257B4£"'0'4 8.gSgSS II.SSSSS II.5
The PSHELL card would thus be .SSSSS
MATER IAL HO.IIBSBB 9S.S625S S.rl'::':'. T
I·l~.SSSSIl 45.223B536BE+g3
The MAT2 material data is specified terms from the stiffness matrix :F1 MAT2 F2
1
using r6 47033.244141163E82 B.BSBBB S.IZSgg S.B62SS S.2SSSS S.SBBBH
MATERIAL
1
NO.B625B B.B62SB B. 244 14863ES2
RIGIDITY
S. CF02 is used with the full laminate specification in order to obtain a rigidity (bending) matrix for a UNIT thickness of laminate.BS9IIg 9B.97416582E+S4 8.SSSSS II.11217456£+84 S.
INPLANE
S.0
F7 0.223SS36BE+S3 S.S62SS
AH(.S.223SS371E+S3 S.gHggB 4S. The rigidity matrix is then multiplied by 12.
1 Z
3
(
THICKNESS 8.BIIIIIIII 911.g625S B.
one for each layer direction.0
6023.136S.3
b) Multiple QUAD4 elements The laminate is split up into four QUAD4 elements.0 1 PSHELL PID where the transverse MATI
3
3
0. say.IBB717SBE+AJ' AJ. <
.0
.. : ::'/~tI'~.
Multiplying the matrix by 12. The main use for this type of idealisation is in optimisation work where the laye~ thicknesses are a variable.0
3000..6B23918SE+B.. It is not possible to represent bending effects correctly and this work should be carried out in the detail stressing stage using the cr programes in ADS or the postprocessor. and referencing a single MAT2 card which giv~s the unidirectional stiffness matrix for the material.0
If the MID3 field is blank the TRANSVERSE SHEAR flexibility is zero (infinitely stiff).lSAJ717SBE+S'. or by inserting angles from edges 12.7
13461. JBBAJBBBBE+B4
and the MAT2 card is thus MAT2 1
.
C ".BAJAJBAJBBS'E+BB AJ. and an effective thickness of between 0. c
.
t ··2·"
t.0 4206.0
2
1..v r F~ 1_. ijormally this should be specified and for crc laminates the matrix (glue) material should be used with E6000. the full thickness being used. .2
0. The final PSHELL card would be :.1
23430.8
4206. for buckling etc.AJSAJIISAJAJAJE+AJAJ AJ.".4 and 1.1·0
For UD material
AJ. each having the individual fibre direction total thick~ess.
As can be seen the rigidity matrix is quite different for the fully specified layup.8
and the PSHELL would now be PSHELL PID
1
4.0 on the PSHELL card.C 2 1.''"'' . .. The ratio 121fT' should be set at 1.BBBS'BBBBE+BB
B.8
shear material
is given on a MATI card :
6000.cZ16E+AJ6 AJ.0
0. The Bending MAT2 card would be :MAT2
2
95827. O'S'
Material angles may be specified using coordinate systems in the r8 of the CQUAD4.9
0.)
136542.0 gives an equivalent material for BENDING effects. the exact stacking 5equen~e should be used.0 4.c
AJ.o·S' t ':.
1807.1
14653.0.c
the CF02 output would be
INPLANE STIFFNESS MATRIX
'.BBBBBBBBE+AJB
AJ. and if bending effects are critical.
7414E+04 0.3110[+03 0.0
PCOt'lP UUC DATA c:. It is possible to put in the full stacking sequence and to recover stress output for individual plies using an RFALTER see the Nastran users handbook.6500£~6 0.0
1.43032
\
+eU30zes +eU3030S
O. orientations and materials are specified and the equivalent PSHELL and MAT2 cards are generated as part of the solution.
The material The PCOMP data is shown below
PCOMP 1
ID is 5
:lBlA10.0
o
400000001 0.0
.uIDS ARE REPUCED B 1. 7690E+04 0.\.0 0 5. The various layer thicknesses.0
o o
2. The PCOMP card is referenced by the usual CQUAD4 property ID.0 0.5000£.6866Ei04 0.5
O.077&E+04 0.S IBlA1030 IB1.0127E+03 0.
USER IHFORtu. I· 0
t~ oS"
The angles on the layers are relative to the angle or Cid given on the QUAD4 card ••••• +ve into the panel from edge
~
t .6500£0' 0.0
6.
**.0000E+00 Z.0 0.0 0. bending and transverse shear.0 3.55Z4E+04 1.0 200000001 0. the three MAT2 cards being for membrane.0127E+03 0.TIClN I'I£SSAG£ 4379.6825E+04 1.6500EO~ 0. and complexi ty of t he job and thus the PCOMP data is seldome used within the Stress Office. Thi s increases the size.3303E+01 2.c) The Nastran
PCOMP facility
Nastran has available a layered composite element property card PCOMP.0 6.
YES YES
5 5
1. THE USER SUPPLIED 1 1.0
45. for a laminate
t~ o·S"
t. A typical set of data would be as shown below with total thicknes~ 3..0 0.0 1. Bending and transverse shear effects are automatically included and thus a membrane only model cannot be set up using thi s type of data.
90.Z647E+04 0.2647£+042.0 0. .00 6.0 1.5000E+00 10000000l 0.0 0.077&£+04 0.0 8.0000E+00 400000001 2.s
0.0
BY THE I'OLlOWIN6 0.0 100000001 1.1217£+03 2.1585E03 0.0 1.3237E+03 4.
YES YES
t
The computed PSHELL and MAT2 data is shown below.7836E03 0.0mm.0000E+00 6..0
PSHElL ·KATZ KATZ KATZ
Z00000001 2.903ZEi04 0.0
1.0
12.0
8.0 1. 135.
~ .i
. t.b
L.H(S
For a particular QUAD4 element the stiffness in the X direction will be dependant upon the layup stack which it represents. . All these aspects should be discussed with the materials group prior to the creation of the model.'
.. Having determined Ex and Ey the usual anisotropic MAT2 data can be created by inverting the compliace (flexibility) matrix. whilst the Y stiffness will be the glue only. '.
.
x 1. In more detail the method would be :a) determine the layups to be represented through the thickness element.
This form of idealisation is common for investigating details of test specimens or portions of structure on a very detaile~ level.
.
..' .E..d)
Through
thickness
CFC modelling
using 2~. Through the thickness modelling using 2D elements requires the determination of the material stiffess matrix where the 'through the thickness' material is the matrix (glue). c) the known material details are now Ex deduced X direction stiffness Ey glue E G glue G 0.
.
c.HE.
..
1
f
e
The above method is the general application for through the thickness modelling.
::
.IC
e
. e) create MAT2 data from stiffness matrix. Boundary constraints and loading should be known and quite often symmetry can be used to reduce the size of the model. Slices at a free edge would again be treated differently.3 poisons ratio d) calculate the element stiffness matrix by either inverting the compliance matrix formed from the data at c).
.
.
...' .
".AHINI\Te. however other aspects may require the modification of the material data before use.
. It is vital to know the full specification of the problem to be solved and the expected information required from the model. . ~oissons ratio constraints need to be accounted for if the slice is from the centre of a model as this would imply an increase in effective stiffness ~n the X direction. elements.
i'L. or use the above data in CF02 and output the stiffness matrix for unit thickness. by each QUAD4
b) for each element use CF02 to produce the compliance matrix (option 0) of the group of plies (for unit thickness obviously) and deduce the value of E in the QUAD4 x direction..
.
. A typical honeycomb PSOLID...
c.
.. make sure that the material direction follows the manufacturing drawings. ' ")
I~' /'.. )( .~·whic"h··' planes are the Land W directions..
ID is 15 and direction
\....
.
"'
1
_..HONEYCOMB
IDEALISATION
It is nesessary to model honeycomb where it is ued as the internal structure for flaps.
/"1
c . local or element axes.. whilst for the CPENTA element the it is dependant on the conectivity list."... WARNING . t_f..
. Thus it is advisable to specify the material direction in all cases via field 4 of the PSOLID card. It is possible to specify additional grid points at the centre of the edges..
PSOLID ID
::
S
~'
15
12~3 :c
5
Where the material system 5. the output being in the MATERIAL direction. Take care with the data to ensure th&t the matrix ties up with the material' axis' system. fins etc.(..··±e·.. would be as shown below. The CHEXA and CPENTA solid elements are used for this purpose and are simply defined by their corner grid points.
'_
.. 'I. foreplanes.... If a single direction is used for a thin curved panel the out of plane stiffness will be lost as the elements curve away. which holds the 6x6 stiffness matrix. tr1z
~·s
ss z
L. ..)
.. but this facility is not nesessary for honeycomb and is mainly used for solid modelling.since honeycomb is extreemly anisotropic.". stresses may be recovered at the centroid of the elements and at the corners.. The material can be specified using a KAT9 card.(~lnow)
2300
X'f
I·"
IIt~
c: J.
. This can be in the global._
is in coordinate
~ r l~'I[
r::
'1'. stiffness matrix and MAT9 card
Honeycomb
Itiffne .. ror the CHEXA the element axis depends upon the SHAPE of the element (X running between the longest direction of the mid faces).IlL
.
0 0.MIDI and K into the PSHELL fields 5 and 6 The transverse shear material MID3 must be specified based on the shear stiffness of the core and the ratio. the ratio K .0 0.:.0 1293.
Set the PSHELL membrane thickness to tl+t2 (if fully effective) with the associated material MIDI.0
The Land W stiffnesses are obtainable from the usual data sheets. ~ .0
0. 0.0 \' 0.~
SPE:CrFlc.L~TIIJE.12I/t~
I
where t .>
.0 1.0
This would show that the sandwich is 250 times stiffer in bending than the 3. . or implicitly where a single bending element represents the assembly of the three componant~.~
'i
!
.0
F4
F5 0.l I L III I 1. and may be inserted as 1.
i
: i
Ii
I
I
! .0 0.y and xy values are small due to the concertina effect.
F'6
F7
Fa
0.I> El£H£Hi
I ~N. . SANDWICH PANELS
These can be modelled explicitly with elements for the two skins and internal core.0 1149.
5.
Ht>sr"!>E.0
1
250.1 HII~ ! t i t++t+
.0mm.Fl
MAT9 + +
F2
F3 6.0 if unknown.5 0.0 2300..0
2
10.
PSHELL card may look like this
1
3. on LocAL Ax\S
E. the crushing (z) stiffness possibly.
t. The full PSHELL card is thus :PSHELL PID MIDI t MIDI bending MID3 ts/t shear
membrane The Nastran are specified A typical PSHELL 5
transver~e
output gives the correct skin loads if Zl and Z2 correctly on the PSHELL continuation card.5 0.0 0.tl+t2
Assume the bending is taken as skin loads.r'
No'ff.
"
I
=
i
"""~
RE:. The x.
. The method for the latter case is shoWn below.0 0.
'lrl
i:
. Calculate Calculate the section moment of inertia.0
0.0mm skin and that the core thickness is 30. where ts is the transverse shear thickness.)
~ 'Z. put MID2 .0 0. ts/t.0
F9
15 0.
Early work at BAe Warton produced reports on the stiffness of bolted and rivetted joints (SOR(P) 75 etc.) and these have now been incorporated. or z direction. y. the local. ie. It is better to move a bolt position to a local 'hard' point on the model and ensure the overall stiffness is maintained.stressing. If spring elements are used the fastener loads are known and the loads in the panel (real thickness) too.F. ie. Care should be taken to ensure that any load offset effects can be carried by the local structure and that unreal flexibilities are not created.
Method b) is preferrable in that the effects of sliding fittings can be modelled by only connecting the pairs in certain directions. ie.fuselage fitting involving many bolts. One of the problems for bolted or rivetted joints is the derivation of the individual or group bolt stiffness. lug etc.. and fastener loads will be calculated from the edge loads on the panel. or complex. the derived loads will be applied to the real structure sizes.F. do not significantly affect the stiffness of the structure as a whole. (CELASi or MPCs).~ bolt positions and the offset effects. Once thecq have been determined it is a simple task to create the spring / element values. The modelling of large lug and pin assemblies in some detail _ is part of the design process and if carried out correctly can predict the overall jOint stiffness. HINGES
(
Hinges can be modelled by
a) using common grids between componants with only translational D. If a structure is connected by several hinges it is important to make these colinear.~· . the best method being to use a coordinate system for the geometry of the hinge line points.
')
.2'.taking into account··the'real. The remainder of the joints can be modelled explicitly .into ESDU85Q. If the element property is modified.
JOINTS The manufacture of aircraft componants involves all kind of joints.. rotations are possible..34_ ncL 85035. or implicitly by modifying the local element properties or grouping the effects of several joints/fasteners to pairs of coincident grids and using spring elements to represent the equivalent stiffness.._ a . although subject to local stressing problems .shear build up etc.O.. which can then be used as a simple spring element in the overall coarse mesh model.by representing each bolt. In the idealisation process the majority of these can be ignored in that they are the continuous bolted/rivetted/ bonded joints which. a wing .O. Explicit modelling of joints can be simple. for transport type joints a single bolt is used at each position. in some detail. or b) using coincident pairs of grids and connecting these only in certain translational D. Try and make the hinge line direction follow the nearest basic x.
WARNING . the pa i rs of coincident grids~..F.'s. but in general the above comment holds.
. but the loads are calculated with th~ geometry at the undeflected position (unles large displacement solution is used). if the hinge system is redundant make sure that the geometry of the hinges is colinear.'MUSThave identical global D. If this is not so the hinges will lock up and present a hinge moment load path which is unreal. MECHANISMS These can be modelled using ROO or RROO elements to represent the links.WARNING . Nastran will calculate the displacements under load. jacks etc. Thus beware of exessive deflections.. If this is not so the spring elements will transform the displacements (and loads) across the pair of grids and the resulting structure will be out of balance.O. This implied transformation is occasionally used to advantage for setting flaps at angles etc. using a hinge line coordinate system if necessary.
\":>
\~
'" z "" .
\c:t
I
III
\?
u.J
~
• •
(11
t:
~ )<..l
Vi
ct
u:
Q:
_' '
\Q \'!"".j
A
CI
e
A
... ~
v :::t
4:w
..J
\
i!
\
)
\
.Q
.~
.
b a••octatad Wih fiance 'aDell with low .ili..d .laLa ..a2 )t1on..
"ra.
Feature
Type
'{.. Features such as repeated stiffeners on a skin or a series of lightening holes in a web .
.HOLES AND REINFORCEMENTS
The treatment of
these on a coarse mesh idealisation is to modify the element size to give an equivalent stiffness...
l"...a..Chr1ch. are. and he catorigises the guide shown below.d
Iffacti .7 c~ ..
For weakened shear webs the equivalent thickness and end load area is shown in· the table·below··(fromAGARD series14·7+.a. Web With .. dL.halaDt SHffts. r of Uta or bt/6
bt
T
bt
0.t • .. of Web
~_tt~~~_~~~ . lbear IUff •• n (aquhalellt Gt)
. •• !lotch"idth ..) to lint ) of 3 ) e•• tallltion.)
T
%ero aor...l
.tallated ".. nt <10' ot element flexibility under relevant unifora loading
AHD
Theae rough rulea are intended to enaure that the strain energy in the el__ent with it.
Joints
Aggregate volu.~ Web...
a. 11 cltalion.C...i:.. feature. D hoi.'. Corrq.trel.
.hallow "..e of perforations <s...
%ero Iloraal) to
La •• ar of 15tl or bt
T !l aor..
ce
UD/. b. D ...aro "loa. The criteria to be used is whether or not the local features will have any noticable effect beyond the boundaries of the element in which they occur.items which occur regularly.. 4 of . AGARD lecture series 147 included papers given by I..rldieat
ryp.. IffactL . of WealtaMeI Sb. tallatLon. of basic element volume Aggreqate flexibility over periphery of el.Uu
T
Le . b
IheIrbucU. cannot be ignored..
! c:t
for .. 1a flaxure . ~'3• aotch depth pttch plate deptb
• 41 •
loped lauath pllte dapth Y"' Dotch pitch
cI. not continuous AND Aggregate volume of reinforcements <lOt of basic element volume No IIOre than 20t aection loat in any continuous loadpath aero•• the element
Weakening features
'.l to corna·ctODa
bt
{!:.
2
se (ala 41)
!! parallal)corn.___
~. b..'..6
c:t
Web with lLahtaaiua bol.ba
Q 8 1 + ('i"':Q)
+
taro alona liae of c&. ter
c:t
ca. b.~~
_
...
Can
be
i9'tlor~ if:
Reinforcing features
Reinforcement i. is within! 10' of the basic element under any relevant loading.Taig dealing with the practical application of finite element analysis.
C$HtAit
e fI.
AbJlITIONAL.N act
AR E:AS.Ro.
.»S
ItE. ONl.E.NTIN~
STI F fE. Wherever an element is changed in stiffness. Alternatively the element material E value can be adjusted to reflect the stiffness change.
PLATE.PQ.(
•
c.. in the case of stiffeners in a single direction.Reinforcements can be modelled by increasing the overall thickness or.
c.Se. by adding CRODson the panel edges in the stiffener direction. s\) 'l> "t~
i='J
e. the interpretation of the results should originate from the element LOADS which are then applied to the real structure for detail stress calculations.
..TR. A table of the constraints required for the various planes of symmetry is shown below..ON~TRPdNTS
STRuCTuRt:.shOl:t·ld··be··· . For the antisymmetric case freedoms 1.
s"n1 HETRI e
SYt1I1ET~le
AN'll S'1Hn E. e.le. . vertical and pitch movements. a given time. ·· constrained in the 2.3 and 5 are constrained.

SYMMETRY
Symmetry can be reduce the runing time. preventing lateral.4 and 6 freedoms. If symmetry is to be used. Symmetric constraints X Fix constraints X Fix > Fix :Y Z Fix Fix :Y Fix Fix Z Fix Fix
MX MX
Plane of symmetry XY YZ ZX Antisymmetric
I
MY Fix Fix
MZ Fix Fix
Fix Fix
Plane of symmetry XY YZ Z~
MY
MZ Fix
Fix
Fix
.
co NSTRAUln
For symmetric loading the grids·· on the 'ZX p Le ne. only a portion of the structure is modelled. ie.
a) is the structure geometry symmetric b) how many planes of symmetry are present c ):does the material behave symmetrically . roll and yaw movements. SPCs etc. preventing aft.. CFC ? d) are the constraints symmetric e) is the loading symmetric . and the grid points on the plane or planes of symmetry have imposed on them constraints. Two kinds of symmetry are used._
:. the most common use being for fuselages as shown below. Some users would prefer not to do this for small models. These type of constraints can be applied to any structure to FORCE symmetric or anti symmetric BEHAVIOUR.. symmetric and antisymmetric. There are taken into account when
used to simplifY'the model and thus or enable a finer mesh to be set up in various considerations which should be determining its use. which can then be applied to the model as seperate cases with different constraints and added. it is possible to break down any loading into symmetric and antisymmetric parts.
A"lE. and the model can be reduced to 1/4 size.Two further examples are shown below :Simple test speci~en with hole
(:::'+..J
ARrc..
2:) \Ix IN
H~
PL..TRle
(tQuAL
*
\.
'd) rbc
Sill E.lbS
H.A ~E. '1 uP ere f:.
To ~E. The centre section of the jOint is to be modelled.c
H.r. Bonded test joint
conditins. XY and ZX. I
I
f
')(
I
I
*"
N££M
kP.
.
q S'tnti E.R
f=.
GRll:l5
oN
nHS IN
Fi\(:£. CFC and axial loading. I)
A Nb
I
.
•
f
r
f
.x'f
CoN SiR AI N E..' =
f 

.
l1Y
GR 11>5
oN
UN bE.A 1"0 ~~
'r\Ol>HL£t>
There are two planes of symmetry for SYMMETRIC conditions.E:~)
. and the model can be reduced to 1/4 size..2:.__. .
MU)
G(}. tfSo i=1~R..A1NE.ex
I\Nb
PLANE
c:oNSrR. ZX and
Symmetric lap joint using symmetric layupsin.
r
t
~////~//~41 
'f ~ t
.io
There are two planes of symmetry for SYMMETRIC YZ.
cD~~~
c'ON~TR~IN IN ~
I
TlftSE.
Thus more information than is usually available from solids can be recovered (mid face stress levels) and plotting is enhanced. It is only possible to recover element STRESSES at the centriod and at the corner grid points. Thus the plotting or calculation of stresses by hand is difficult.foreplane spigot frame is an example where the Catia model is being converted to an equivalent Nastran model. are present at the grid points since by definition bending loads are carried through the elements depth. CQUAD4. A plate previously modelled using bending CQUAD4s is now by solid elements with grid points on the both surfaces plate. CROD etc. Patran performes this averaging in the contour plotting routines. GPFB's will obviously give the corner forces acting on the grids.O. STRAIN GAUGE ELEMENTS These are normal Nastran elements..SOLID MODELLING By definition this is the representation of the using threedimensional elements. see below. the geometry being obtained from Anvil or Catia. c) as core elements for flaps.F. CHEXA. carbon fibre layers/interlaminar features etc. One of the drawbacks with solid elements is that the stress levels for the adjacent elements at the same grid point have to be averaged to find the correct corner stresses. This is specialised analysis work and is rarely undertaken. Thus the geometry mirrors the exact geometry of structure thick defined of the the item. b)' macro/micro investigations of detailed areas. A method of finding the correct surface stresses is to use strain gauge elements. The EtA. Solid elements can be used in a variety of situations :_
a) large machined items.
No rotational D. holes. CPENTA etc. test specimens. and this is the usual method of gaining an overall picture of the surface stress levels. fins etc. which have insignificant thicknesses or areas and are "smeared" over the surfaces and edges of a solid element model in order to register surface stress levels.
.
the CRODs along the flanges cover the end load singularity in these directions. large local displacements will be seen (several metres). these can be seen in the GPFB for the grids co~cerned as the out of balence loads. and the local skin membrane stiffness will be reduced. CSHEAR panel singularities. at a grid point ill
Rigid elements are covered in the Handbook for Linear Analysis section 2. a) to remove near singularities modelling. these are not necessarily only the independent freedoms. due to the method of on singular points.
the global D.RIGID LOAD PATHS Rigid load paths or elements do not occur in real life..4 and several applications are described the Applications Manual section 2. Examples of the above types are described singularities.10.
')
2. only is 'beamed' to adjacent hard points. However there are many instances in finite element modelling where it is necessary or essential to use these. d) change in mesh size. Where the mesh is finer than the modelled internal structure and the skins do not have bending stiffness. joining different elements. or by the use of the GRDSET card which applies constraints to ALL the grids in specified global D.. f) to distribute g) changing loads to a structure.F.5. The resulting loads on the three points should be insignificant. but through the wing depth the end load capability is usually neglected. the unsupported grid points will have a very low stiffness out of plane. If not constrained.O.
.F. Unsupported skin panels.O. SPCs. However near singularities cannot be automatically dealt with and the user must remove thes with rigid elements. below
ti
r
a) Removal of potential
AUTOSPC will remove EXACT and VERY NEAR singularities. Care should be taken to ensure that the dependant freedoms (UM set) do not conflict. but itis better to include these on the GRID cards in field a. In modelling shear webs for wing spars or ribs etc. Using an RBEl the unsupported D. Some corr~on situations are described below. 1.F. all applications of rigidelementsrequiretheuse of independant and DEPENDANT freedoms.O. with each rigid element or with other modelling requirements.
e) to obtain average motions of a set of points.
WARNING .F. boundary points etc.
b) to enforce local displacements
c) to represent very stiff items of structure. The singularity can be removed by linking the top and bottom grids with a RROD in the through the thickness D. Freedoms from which stiffness is to be derived should not appear in the S set (Singular) .O.
IN
P1.nEUr
3.. and usually for flat surface models using bending elements thes~ are removed automatically.Uti
A
.. E...s.»
Qo:
J
' '.
3
+
.L.l>
.I
A\l>
Itt
.
J~'
'f .
e
Tit!
e2HTltE NIl>
u
. CS"EA~
"
3~ =:. in common wi th many fini t e element packages."flLI. Nastran plate elements..S
•
f~t: \
.14J1~
iao'b
£'L.&y
Q.'F..__ .S
\.r:..

3.
•
ST.I"~
'To
.. . __t_ t
fl!E.
tI
Q.1
• ~ll~ :l 1.. . do not support the in plane stiffness at the grid points. However if the surface is slightly curved the in plane stiffness will be made up of small components of the plate bending stiffnesses. Thus there are implicit singularities for these elements.au..M
~1~tCTIO""
II
"A~
~ef'NlIA~~
.ANE.. i. These rotations may be removed using RBEl elements linking the offending freedoms to local in plane freedoms..
elll> hNl> FUUoM _ ~2ol>
PANEL
..'
8
A.Hp\'AH£
s» I. !f... AUTOSPC will not remove these and thus high rotations may be generated.
iOTATIONS
.
e
3
""
2.
The removal of high in plane rotati~ns....i~
10
A
A
.
_ __
. J.
III
rr
. .E..""T
SIHfoaLAltITle.1.1 I
')
. ''''loAN£'
STIFFNesS NO .ING
~Ne..1
AWl> HI'"
4 1.NUS
4
J)
I
IZ3
S
13
C.
TIOMS S'RlNGctt\..eE. "f.OeAl.oc~L ~ISPL."1l>
U.. Thus if two structures are attached via brackets with potential singularities removed using RIGID elements.0
a
t
1
~IJ)
. MOM )
r·
'
"
I
~£2>UNl>~NT
i.
FUuon
NoTE. WARNING ••..AeEl1EHTS
\:..HbENT
.
!
of
l.EJ>Ut<l>AHT
~1.
It is sometimes necessary to force a LOCAL displacement onto a singular point in order to set up a load path not modell~d in adequate detail.. I~Pf.4 x ~IR£Cnc"
FICT/CIOuS.lf.b
IH
~l~T""LE &l
olol(
:..llol1
2.S£
t1oll£..t:N flXE.blfEeTIOIJ . F 1bJ>$ ~_
I.
~ISPI.
b) E"'~o~C~h
\. but make sure that only ONE load path is present.MTSo
IN
x
..."0 IN
ItO TO 'to. Hf\t1b'lO~ R:~O~?
: ... 0(
So
.L.
ro~ttA.N . and more than one load path is present THROUGH the rigid elements. Several brackets may be treated in this way..
•
\.f:~TloN
~
·•·
flCTlalo\lS )/~'LAt!f.
)E
Sf'P. and RRODs may be used to transfer out if plane loads to the main structure.\Ll.e. lot.
nAI.eTE.eTIDN'
2.l1E.
sPRINGWIl./f) U.
L..:
a..ttfNT
OF
.oAbS
•
1).Hr ~o£ To CA«TI~G.
1)
WILl.£n~NTS
1t4
x
J
.:)1R."I.T
Z.MS ~:rlllf. but beware of setting up redundant load paths.DeAL
.b) Enforcing
local displacements..
. The RROD will maintain a FIXED distance between the two grid points used.. An RROD links the bracket point to a grid on the main structure with the specification using one of the global components in that direction.. Brackets modelled in 2D membrane only form usually are singular out of their plane.
IRol»
lQOl>
l:))
:'2>
aol
.:1 CC»CNE.~C~t\~NTS·
I!~s
To
..IONS
e.01 .TA..bi'f. In this case the rigid element will carry load if any load exists in its direction.
~ ~
. . see below.
:..bIAe.
~"NfCT"HL 2. the resulting relative displacements may give rise to unreal loads in the structure. a.01
FUf. The application of enforced displacements will le~d to erroneous results if a closed load path is formed.
:.)
R.b HI:I:IS blSTAwcE
Stllt"'"
To
eoNNEe.01 AN~
Wolb
.
\
•..
(00
I. heavy.:. (Usually noticed by high values of Epsilon and diagonal ratio terms).
2.ItS
t ~
a. One of these is the inclusion of very stiff elements. '':'"
'
.
'
c) Very stiff items of structure. U/C legs.'
t:O~"AR.. brackets etc..f"'.3
~oo
3
~o Ul1 ioo
1
113~S'"
. any effects in the modelling which spoil the aatrix conditioning should be avoided.
. no~\..l"IHI\Te.Tc.
nSTE.\..>:. which make the conditioning poor because of the increase in the numerical range of the stiffness matrix terms. Thus an engine connected to the structure in a statically determinate manner can simply be represented by a RBEl linking the CG point to the attachment points.00
. Very stiff items such as engines.lo \.
+
.
. 'Tit.l>I~I\"t'e.IM~ !I.>
...I
~OOQ.\. It is better to represent these items using rigid elements. ~U4\:S.
..
"~11
... (STAT~L.T u..:'.\"'t ~e.)
.
•
~
.~IN E.. boo.0
«.. and can lead to quite implausable results... . Because Nastran has to form and~invert the stiffness matrix. will caus~this illconditioning if modelled.AL.IHl($
S£."
..DC..
RPt(..+
LIU
"To
..+ ONTo ~.
..S.
APr t'lel
I.£t1$
Pu1'" Ll..
I.
'~~e.."
I
I
US£
tJ. 70
V
~~E.
it
f
r . and a rigid element should be used to enforce an average displacement on the gr"id concerned.E
1S ACCfPTI\SI.
Merginqa change of mesh size can be accomplished using the usual modelling techniques.
Lx
. A more complex merge of a plate to the centre of the CHEXA could transfer the shear as the average of the corners using MPCs. .1
70
'1
.d) Merging
a change of mesh and joining different
elements.I 'L:b + un ~o
ItO
12. AN1:I
2. Instances occur where dissimilar elements have to be joined and continuity of load paths created using rigid elements..
2.1. If the plate locates to one face of a CHEXA a single RBEl can transfer the missing rotational load paths as differential loads at the CHEXA corners. S'
Il>
~t~£NT . Do not use this technique in areas of rapidly changing loads.IIA1:.
teuAl:>+
.0
22.. A common example is a bending plate joining a solid element..r ~o 15" 11!jj..1
lS"
11
3~
.p
III I. and pairs of RBARs to carry the bending loads to the corners.
~"".. It is not recommended to use these higher order elements. RIl>
USIN f.
:rOIN/NG1)tSSI" ILAR.£. 2.
(.0
U'
•
eJlE:XA
3.."0
1
. AN 1:> 2.
bl btH"f
.l>M~S
To
'jLx
MPC
1"
11
31
•
~E~l>IN6
"TE. but sometimes this is not convenient and additional points have to be added directly as mid points along an existing element side.
E l€)1eNT
U'
S
VHblN .tlb USE:. I
&.u~l:.
t1PC
To
To
AVt.
PoINT 40
10
IS
INTlt~ee.~..l>5
22..0
12..
I:
t~1
+
CHE:xA
"T
~oTA"TIONS
20 1$
floM I\l.
1·0
as'
t
a'S'
o'~ 11 13
~EJ.
1$
~
~o
iT.5"
el\UJ\
R.[~]
1.RlbS
2:2.S$
.lI
TO
ttU~II!!.s
lit
10
e4vl'l~ .ffHtSS
ST/FFNe$S
IN
IN
r
x
~IUCTloli4
~ltce"(/OH
7'0
~c
AvttMel>
f~Otf
~tll>S
:to 1\"'1:.
2. 20 2. I.l>
un
20
2."E
Y
AT
20
a.
.l>
To
11'"
rw
..l1>
.c. Another RBEl use is in the definition of interface points between structures where the points are not part of the local model.
THt't"
".TU
~~~NT. I"TE. The RBE3 element then links the motion of this new grid point to the average of all the 'hard' points.1
r.R.Po.~OL.e £
of
I'oIN1" It
0
"\I E: I A ~E. \.
c·
fOINTS
FRl\nE..e) Obtaining
average motions.
\
. I ".I I.<
J4_
t'?>E.00.flCE
~\SPLACEl1~I>ITS
ON itt£. and secondly creating a new grid at the A/C C/L.
USE. b ISP\.~ 'r \
i"~T.J(T 'T~t.1
so
2 IF
rioen!rOIN""
nit\.
I1I:.f\(:.E.N'1'e.oL.
p::r\'
.A\ . One of its uses is in the extraction of the motion of aeroelastic points not present in the model. A scalor can be used for for the displacements of any of the selected points prior to the averaging. ~
)0
A
31 10
I)~
12.~El1uu·
oF
A...b
UM
to
~
III \2..
Fo/<
.
IItr
A.. IS
SE.
t'\oTloHS
110\10.
..J.eT
1oISrt..
XF
TloN
rOu·~TS
.tHfA~
PoINT'
1"00
AElloE.UT'S. 'l
+
1.£. by first selecting a subset of 'hard' pOints.
~
I~l
I~ ~E.f..ll.I\~'TIC..RnIHA'TE..\
10
13
12.AGE. '" TS AlE.i.ll> ".l\eM
.Uoc.
E.
~I.3tH
$'0
~\ Sl
II
10
1\
ul1
12..
e) o~TAININ Gtt'llA.'1 >ETE. thus soft points can be made less effective.
)t
"Of t.E.
No __
100.F£.l)
»ISf'l.
~
.A
.l•.t.
oF
b
"
$1
" !/
.
The RBEl element is capable of defining the motion of a free point to be dependent upon a statically determinate set of motions of selected grids and freedoms.'l»E.
.tHeA 1.:::::1
.T
MAY
TO
:lKI. The RBE3 element allows the average motions of any number of points to be enforced on a specified point.
100 I "
sFoR." Po IN'T ~
~.
A"".. Take care that over flexible points are not used in the definition of the element. .0
'1
+
PoiNT
A
~IIS
ASSICHL.
i! t'S
.I'M 'b'
fOINTS
AAE.$
STAilC. The normal use of this rigid element is to predict the average motion of a fuselage frame.)1S'PlAeErtt.I'ISIJR."..\L.a... eN
loo
Ih'IS
c. . otherwise incorrect interface loads may be generated..LOW£.
100
e:
300
fxrs
'""~
F.
. / .AT'Nw.
Avf.
R.
'OINT'
.01\ l:I AT'
TO f.F
\.
".aTU
~~t.£ ioltfT' '\!AnIN.f) Distributing
loads to a structure.3H'".PI.
"(0
A STR.E. An RBE3 element can be used to distribute load from a fuselage generalised point to a selected set of loads on a frame..ISEl)
')
.oa
LQ~b~
1'1
A Pl"1.
~'J.3 200
AS
J(
\.IE. .
W.
SO
..oA~S
.~M.
I1ETl101i.. .
K.l:>
TO THt:
I~T
TO
!'oINT
A
"m.
"UTICAI.AJo "H~
HO'lcoHTftL. ')(
l.l. .'b
SOO.PI.
AfflLIe./'"
I'
l. 'To
.1 I!>
(0
"
~
IF
111
U:r'
II
SO
'"
Ati>I.: 1\.
~~E.G. "nit.
A
Xb
UH
IIf.
11" '1.."'" E. TA':'£N
0.I
. This method is used to load fuselages in coarse loading situations.0~IU~$
U)Ab
Ft
....IEl>
.
12.b TO
.IC>
100 .I.1E.1..l."£
cr~I'l>S IN
ftPPl.'.
~e.AT£.Lo w£.
Loading defined at a point not coincident with any grid point may be applied by setting up an additional grid at the load point and using rigid elements to beam the ·load to the required grid points.
Ut'\
~t.
~ . A simple beaming to 3 points is carried out using an RBE! with the loads being applied to the model in the way the statically determinate set is defined..NT.4.:b ~o~f>'
II
_\.
HOt1£WT
100
I
TAteE.IE.N
I.
se.uc TuR.1'l> 100.lt.O~~
aoo....
ho •
".
u)"
:t
'l>
F~
IIItTK
wil. the distribution being in the classic bolt group type of calculation.
t'&e.00
It !tE.H AS
°'00
:t. POIttT "
~NEU".:::~:::rl4
\
1/
\
\
I_
'T'N E..'.l>
I
~£'l.:'CC.lo
TO ~E.
'0
sa
MAY
'·0
~e.~ :..2.ol\l>S U.
.
lXuAt. oN
l1ot'\~""T
tOO.!>
+
to
Z~ ~o
uH
A
31
12.
IS
"I..
Fl..l 12.c..~IN~
.
'TO C~'I>
loA"':> 100
I
FlC
..0.11..t"t'\o\
W"I.c.
WITH l1A/N STRuc. . and the latter cannot be changed for other reasons. carries out the transformation between the pairs of coincident grids. and connecting them to the original grids using RBARs tL.
t\"e'~
X.TCJ~f.. but the displacements relative to the main structure are of interest. at a grid point.
{/o/
(. It is sometimes required that the output displacements need to be known in a different displacement system to global.
X
y
~
GLo~RL ~ ErtN1TlON
~l> t
PS
.ING101
He.£. The MPCs can then be applied to the new grids in a consistant manner.l. By setting up coincident grids with the required GLOBAL D. Sloping frames etc..
e'Ll>
craot1e.w
l:.b
1lE:.O.b of 101
l:l>
100
10 I
...F.R ~s..O.
rf.h
GRl'bS
I'r v E. This may occur at hinges where the hinge line connecting springs dictate local global D.g) Changing the global D.A
1\" vf..
I .F.FINrTlo'"
. Tft'f
~E. New grids are created coincident with the existing grids and connected byRBARs.l e.l>l>E:. Another use would be in the speci0fication of MPC data which for geometrical reasons needs to be in a consistant coordinate system.
IN ~A!i. ~bfl'lT'. of
\ l1 POS~l>
c:rR.b
)00
.N'1>E
t
I
R~~R..CE.w
t
P01NT~
\tPc.E. .E.
101
(I
t
~
~o
/'"
10\
•.F.
10/ ) 1.01
HOTloNS _ IN
).
IU4S..F.
l f t I
s.o
1·0
.E..O. J3.
'To
~\SPLP."....: new grids will output the displacements required.
~E.1
I ••
:
• eREPl1"i:
Ne.o
12.
~
~.o.O.01 IO~
.
'_)
•
L\N~
~Hn"
R~AAS
101 ).
..00
}. The global D. Thus the RB}L.
\
\ .
~.1..~ E. defined in a local coordinate system.00
t\ AvE.
Pf.:b Z"b
sre.1DS
ON
fH'~ l)
2.
j
10\
.
2.
"}.. R'BA~
Pt."ol
l.l1 cNTS
cohPf\A.F..
101
II
PAl ft.F.W
GRlb
wrn\
eAN
GLc~A\. may have the geometry and the global D.l>::'1.
:S>...
\
~~AR. This is a common problem involving spring elements also.O. I e.
. is defined incfield 7 of the grid data and thus all displacement related data will be in whatever coordinate system is contained in this field.l).
2.!o<
e:t.~E). the user must ensure that the pairs of grids are in the same global system.F.~4Sb
•
WfUTc.
A fatal error will occur if this is not so..
. An error will result if any of of these freedoms are singular and alternative freedoms are not to be found in the element .
1 "
t
j.F.O. The indpendant D.
UM
ID
301 161
123 123
171
i
. an RBEl with freedoms swapped. are put in the M set and must be exclusive._ .I . f
(
t
RBEl to give point 301 stiffness in freedoms 1.F. are used in general to calculate the stiffness of the dependant points and usually comprise of a set of statically determinate freedoms.F.1.O..3. and thus a singular freedom may be used unwittingly in error.
t
f
~ ~ ~ ~
*
~o\
RBE1 +
UM
ID
161 301
123 123
171
13
261
3
RBEl with freedoms RBE1 +
reversed 13 261 3
.2.
~~the use of rigid elements always or implication of dependant and
and independant
freedoms
All dependant D. a fatal error will occur if the point is left dependant.O. Thus ensure that these are referenced only once within all rigid data and constraints. Exchanging freedoms in RBEl elements
If a dependant grid in a RBEl equation is required to be on the boundary of a superelement. Obscure errors will result if singular freedoms are referenced in RBE3 elements.~
Note on the use of dependant As mentioned previously requires the specification independant D. However the boundary point may be exchanged with one of the independant points and the solution will be unaffected. Freedoms in the S set (auto SPC set) will be removed from the set as Nastran thinks stiffness is to be obtained within the element. for example.
tLon.t~ Xl>
I
. ~eST2Aft(
o~
IN q.
~
Q 1<.
C i.L ~E.00
10 10
~l>
23
1'3
~oo
2.
of:
k::' "
'tIII.
4)
~ATIC.CONSTRAINTS
AND SUPPORTS
. generated and can be checked by statics.
.AH CEo. single and multipoint constraints. atthese points wi·ll be .ALL'1
J>ETE. and will simply serve as a base from which the displacements are calculated.~. ..
These two items are interlinked~ since what would be regarded as a simple set of constraints to remove rigid body motion on a componant would also generate support loads if the loading was unbalanced.TS
Su/l'oa.
SPC:I
SP~I
SPe I
".
)( )
oJ."\l>
\J
MovE.
c:oH~T'lAIHTS
eAPA~e. .t"'NATE
SuPPoR.:l . Other implied forces from MPCs or RBEs etc. Imposition of plane sections. Other rigid body elements are used in certain applications. a) b) c) d) e) Statically determinate supports.TS tNST ~e. Boundary constraints..~E))
IF
LoAl>I~G
NoT'
IN
SAL.s SPC forces.. Imposition of deflected shapes. If any loading is applied to the model the reac. Constraint equations using MPCs. in which case the choice of the constraint points will have no effect en the internal load distribution. ie. The only forces output simply are SPC forces using the case control deck SPCi and SPCrORCEALL statements.00
1000
SPC
to~ES
Rtcovf. neWTS. are only seen from the GPFB output as out of balance fOrces at the grids involved. Several uses are listed below.. determinate supports. If the loading is in balance the SPC forces will be zero or insignificant. The elements used for constraints and supports are SPCs and MPCs.
'"
o~
flu.
~
a) Statically
. statically determinate supports or constraints. The minimum requirement for all analysis models is that they are supported against rigid body motion.
. or mounted on a plug of centre fuselage structure and all grids at the end of the plug held in freedoms 16.1 ..b) Boundary
constraints.
A front fuselage model.1>
1:..~!)/HG
S'REAR
"
SKEA"
I~
~.t."0
1:'1
Ilo
.•
SPCI
spel S'CI
T.l.
Usually the support or constraint system is redundant....!"b
~T
J"CINT
P~ITIO~S
0.3
$fCI
l:b
.Ll> IN I..J)S
S'CT"I<:»I
w(
IU.. . for example.. fb4T/tui.I..
lilT
AI..7>
1
I
2. 1
12.~~.. A wing may be constrained at the Fuselage in the freedoms capable of taking load.
l'\OlIE. PjU lug points
Similarly a test specimen model may be regarded as fully built in at a section away from the area of interest..
~AG
WUfG
fOOT
eONSTiAIHTS
~I
0
. may be held in freedoms 13 at the transport joint bolt positions..
"I~S
AT A
IU . Ii
~E.
11\
100
110
1.
I 3
.. ~"7
Ak» .
2..HeNr~
A.~
1... and it is better to impose boundary LOADS.$'
I. 0
0
"
1'l°S
1
'J)I~'LAC E:. A more common use is in the analysis of pipework systems.:1
.. for example a wing fuel pipe model would have maximum wing deflections imposed at the attachment points.S.
•
&. The pipework model has imposed upon it the deflections at the attachment points obtained from a maximum deflection case for the structure to which it is attached.2.Ll>
10
1~00
.eont'l~NlIU>~SE lloOUHbAU ~ _
II
~ot1 '""' ~NAL.£CT'li
I"foc£.{)OH~AR'f "oI'.
!
Su'e$IGE.
•~
•
I
/" _j
/' .E.M~HiS t:.. ~
eAR..:b
10
"0
3
1
foo·
2.F.ASE..0.c:~
Srell
~~
· ·
.
100
1.Pt.
.l> l.01'::'
I\LTE.b PIPE ATTAeHn~'i
flOII.L. N£NTS ST~"CTuRAV woAST
ON PIPEWo~" t1.. . though.
»/S'LAe.
Uf'OSE... One could use this facility to impose a boundary shape onto a detailed model of a component.
The SPC card has the provision for imposing a set of displacements at any grid points and in~any of the global D.S
i
t
i
I
c
_
.Y~IS
bISPI.ACE..ii'
.~.$ 0
0
10
2. IMPosE. 0
0
sPe
"0
¥o '10
( I
1
noe
vC.JtNf\TIIlf.vE!.bS ".bIH.c) Imposition
of deflected
shapes.:r>E.TS
SPe
sPe
l:l>
l:.oA~
SE:r
o
u. Thus several sets of deflections can be requested._
.
I
}.
~."."'D'S IS !:!!! le.lI
F~
.L
ellS! AeE.
I
'" .. This.oR'). The SpeD card enables enforced displacements to be requested at the Subccase level along with other loading if present..
T
I
. the shape being determined from a previous analysis.O. y
spe l> ". \..:a~~
111\'1 3£
"N~
T~
l>fFl. is a dangerous technique...f
eASe.
_.ll
I "0
L.0
280
••
10 10
l
10
0
.i..
E)Jl>
AB
FoR.T.
R.QuATIOtHi
\. ~ 1\Nl> e.·...L.
I
3
uN
20
1
I
!
2.C£ 1> 10 I. ee.0
10
I
FRAME SECTION ~II. Since the technique FORCES a plane section. .'lt
.ANE TO itUitE.INeAR. tR~A·Ll..uuc.)
A
~
12. .~1.b
1'10tlE. and E.B..Jole£
f1.L. For 2D models this approach will give a linear deformed shape in the plane of the model.~ 1 13 e.
THE. ooToP.d) Imposition
of plane sections.£1
•
I.. To THE.
~c.~tl
t. and the condition can be obtained by using sets of RBEls relating the section grid displacements to three reference grids. and these can..Y
Itl>EI
•
1:. conditions are considered to apply at the boundaries. '1
MOvE.l>oHS oF 6al~$ \1GINTS 1..a
yUE. This implies that sections remain PLANE. a set of self balancing forces will be generated at the grids.TC!
I
. Pl.
This sort of requirement occurs when portions of a structure are modelled in isolation.
l
I
I
i 1
!
.
c
un
1\
113
"
20
e
30
i..PLAwA£ • ~ PLANE. SECT/ot</.I==iE.
Nce~SARII
. be seen from the GPFB at the section.
1 1' 1·0 1.0..01 301 'lol lol
3 3
I
~b
1. the sum of all the
An example of their use is to force the movement of a set of grids to be equal. Again the equations would be simply Uxa .Ae£nENjS 6R.•.:z:.>
'00
1'0 1'0
1·0
2.1·0 1 1·0
.lol wou!.11
·9333
1'0 10
+
+0
So
I
I .·.
100. The out~r two grids can be forced to have identical displacements.AI11b·
USINW:
tHC'S
.'~('1
+
Mile %.:!> I
1'0
)C.E.0.b
300
a
MPe
~oo
l
"0
2.
MPCs specify the behaviour of a group of points by solving a set of equations relating the DISPLACEMENTS of the points to each other.
IbS
'TC .lbS.0
:2 0
20
A t(b
40 I
30
1'0
eAloJ
10
'!.7
.Uxb . MPCs can also be used to transfer loads in specified directions between coincident grids.e) Multipoint
constraints
using MPCs.100 2.0 displacements • 0. and the inner two grids can have linear equations written to link the displacements to the outer two.
oNTo
Mi'C Il)
POINTS npc.0 I· 0
0
t1Pe
1.1.0.2.uAL l>1~pt.. b
'L
I
1·0
hPe 1.Uxb.
'"P>e.. The general equation for an MPC is as follows :
Uxa + Uxb + Uxc .
e)
MULTIPOINT
~oNSTI{P\INTS
USING
nrc_'s
FO£CING5cTS of
eG. etc.b
10
!
1'0
+0
I
_1.b.t • ~
x r
I
:
l.
HPe
E~uATIONS
I) 100 100
PAIRS
100.Uxb. .. at hinges etc. 100 1 .1.~ 1.
101
10 .00 . In a si~ple 4 grid section the grids are to be forced to move with the same X displacement. otherwise the model will be forced out of balance and the resulting loading will be affected. or Uxa . Normally these equations MUST be in balance..oAbS
MPe
11Pc: MPe
1oE. Uxa .10 I
of
eOINCIl>C:WT
G R.0
ie.
CARiY
30O.01
To
.
loading'om:iit·ions'.e to be expec.. The derivation of the stiffness of any of the elements is simple provided the elements are regular and unwarped.me."For··''~ . ie. Increasing the number of elements through the depth.tecL. For a rectangular plate subject to uniform or point loads the results are shown on the second sheet.
•
r
. The results for a cantilever beam are shown on the following page..
. into regular and irregular elements. The comparison of the displacements for simple loading cases then gives an indication of the element stiffness accuracy for the various shapes and for various element types. would improve the accuracy as expected.'l'he_.thod_ofrs _ detecting the errors is to subdivide a regular structure. for critical items such as flaps in particular. The conclusions reached in the note are that whilst all elements produced failures due to locking or mechanisms in certain instances and none met all the tests.o ar. For rectangular element geometry all elements (the QUAD2 is now obsolete). Recent developments in structural optimisation using aeroelastic criteria have brought to light the need to predict deflected shapes accurately.. Elements which have an incorrect stiffness will degrade the solution and thus the final stress distribution in the structure. the general set of elements if used with these failures in mind can be used with confidence. All elements are accurate for pure tension/compression loads regardless of their shape. trapezoidal and parallelogram elements some shear terms are as low as 5% of the theoretical values.ERRORS IN NASTRAN ELEMENTS Nastran is a displacement method finite element program and as such derives all its information from the solution of the stiffness matrix and the displacements of the grid points. and the QUAD4 being poor for low mesh size and high aspect ratio eleme?ts. c . A Nastran application note of March 1984 deals with the problem of element accuracy in some detail and highlights particular situations where large err. The QUAD4 is poor only for in plane shear whilst the HEXA is poor for both in plane and out of plane shear. for which the theoretical stiffness is known. Unfortunately Nastran has to deal with all conceivable geometry variations and in general the more obscure the shape the more difficult it is to arrive at an acceptable stiffness matrix.behave well unde r all·. rectangular and planar. The centre deflection errors can be seen to be small in all cases with the exception of the HEX20 being poor in most cases. Thus the behaviour stiffnes wise of each element under a variety of loading conditions should be as accurate as possible.
~
1. use low aspect ratio rectangular elements. a) Avoid obscure element shapes where high loads are input . uniform end loads are dealt with correctly. use a mesh size of at least 4 across the chord...
{
~ ~ ~ ~
I I
~ !II
1 J
r'
"
~'>
1 I
. use QUAD4 bending elements.. b) For deflection critical items such as flaps... A mesh size of 2 will give rise to an over stiffness of 100% in torsion. d) Avoid using non uniform solid elements where out of plane displacements are critical .. c) Refine the mesh at high load input points and coarsen up at least two elements away from the area.. It is the high strain gradients which cause the stiffness problems . and evidence from actual model comparisons have led to the following recommendations.At Warton a series of single element tests have over a period of time produced similar conclusions.
.sis
.999 ..
.HZ
.416
..
of
slnqut.014 !
.
Shut"
..
aet...loqr" (~)
E'~ts
t
't
l.%' .032
..984
.
{
.886 .918
< ~
SlIt.986
.". .
'.961 ..999 .
~lfzld Tfp D1s0J.985 .
.)
'UI..069 ..ssa
....977 ...991
....ll.
I
[nefts 1an
i .
£XTE.943
.tl1..080
..904
c
.989 .sss
.90'
~
f...I'"
.9ZO
.1
7
1
:
U'
f
I
~~~
SIIf.t.Ktlll9!!llt'
EXUIfI 1'" s
E'~u
.971
.sso
(b) TnC)«%oidll
.566
.9l1
.ftfO .945
..
.911
.1 1.055
.tIIs 1'"
1"...995
.
(
(.1st
Sl'Ie.IOM
S
~
b)
w6
\
~
/\ .906
..913
•ceo
..941
.tl'lAllit Twist
.J94 .
t ~
Ift~lIM
..J46 .a 1GIl lit I..981
•910
Outofl'lane T.999
..904 .951
5
.HZ
SM..1 09 Mill S1Iape E1.
'>
ut!'1x.969
.910
.Ml
.&\R
Sh&pe E1 tlllitfTtS
e LAN€...~
..lrHy
In
..995 .99' .
.999
..
l'
!
The
thit good to uc...99" .016
(Jut.972
.
£lHlltnts
E.11M
.~nt 1ft 1H.071
..) .051 .970 ..ll"'t'd
1n spH.998 .0I41l'19 DfNtCt1C1'
~
~
~
IlEU(!}
~
KEX20(R) 
(.11
~
z
c:)
z
lnults
A·sz z
fOt' StJ"lfvtrt CMtl1.
r lr .. 7
O.
SM.967
....%5
I ..964 .904
l."u
Tlbl .~Its
: ..999 .a
.961 .N<.cl
/rTW(~r
S
~1d.·Plane SMIt'
(JutofI' 11M T.p.HZ
..Ml
.981 .996
. 's Itl ""4'5S
snown Mrt
for
)£120(11)
.994 ..J88
.osd
z
..
.nt ~ ..996 .
Tfp 1..1"
.987 ..
0
2
..Z23 1..
of ."u
:t
CllAD4
!!JAn8
~
.995 1.867
..001
1.972
.121
.t40
~ .975
.ni
.1
Z
~
.004 1.998
1.002
.m .693
!EX20( R)
1.ction
I£U(8)
2
(b) A~

WAD2
Rat10 • 5.
.71 ~
JfztG
Literal
Defl.
a
rtf NodI SPlasEdg.~
tM ~
1.000
.K7
.91t
.010 1.421
.000' 1.
a
.0
2
.972
.519 .at10 • 1.
I
1.nt .997
.163 . of Mod.991 ..c:t1C11t It c. ItUIIICef'
at
wfth afds1d11t 1l00H.012 1.l. 1f Zed Lnerll
Defl.011 1.052 .995
1..M9
f£XA( 8)
~ .Jst
1.028 .754
HEXA(8)
~
_ !EX20(R~
'\
" ."8
It
.072
H020(R)
(I) AsPO<% ..~ .992 .. I Defl.
at NodI SI'IC" PI.993
1.nts
pet'
1<191 rtf .nter
!EU(S)
~
(Il
4~ct
.~t10 • 1.983 .994 1.006
"ffJl'
tIt..731
.955
1.957
.988
. Edge of 1b'J..994
.363
•!So
.000
.ction
It
c..m
.1
Nar.1
1_
1
I
i~
~ ~
.003 1.010
.02'
.UZ
.433 .
~
rtf ~ SplenItt!" Ed9! of 1iIod.885
..054 1.002
" ... !ElZO{ltt
1.S42 .139 ....0
" .978
.. ~ .990
.l
.991
Cant.~3
.l.995
\
...0lS
.003 1..~
\
.m
.'1:1
.m
I£X20
...997
.9l'!
•
.813
" .001
.073
.1
fs equal to onenaIf
NwIOer'
Per
NodI SI'len Edge (Sf fib1tt I
at
NanIIltztd
LUI... IIOde SpICIS.942
"'.075
.
~ .121
..
I
I
1 I S)'a
.. Per
1Icr.
.000
1.552
..
H
b') Do
NOt
tJ S £..N T c.rINE..) ~ \lOll:>
OuT
SOL! 1) £.(£$
eORN€~
'biS
COULl::> 13~
1"'00
uP
"10
5"
'')
TIMES
'"'il P
S:TIFi= IN :Bftl.
Pt•.rrteAl
M..c·
II
S'Mf\ll
I
E"'~Ro~S
ITE..S
roR. QUA1><f
sa.
~PtRSE.J\NE.rs s
'!.LE.ECT
J)/
.:>ING\IJ\TI\
PLA
.obG..tS.£Nl.E H e.
1'1 f.
l1E.t1SNlS
wiTH
~ON
PARA lL~L
FAe£.R.SH
At'
Ht~H
L.hIt{G
S !LAC.$ JI
0N
TUIN
S
c)
RE. oR.
e~.oAb
PoINTS
/
II
of
\
~
d.
0+6 1000000.0 Nl
.N3 coordinate system CID.
M
. A moment of 5. N3
FORCE
ID
G
Nl
The FORCE card applies direct forces at grid point G.2. At the initiation of the model it is vital to know the type of loading to be applied.
MOMENT
The MOMENT card magnitude M and relative to the appiied to grid MOMENT or MOMENT ID ID 200 200
applies a moment load to grid point G.
1.. 0
0.8Kn applied in the z direction of the basic system to grid 200 can be defined by FORCE or FORCE 2 ID ID ID 200 200
G
2800 .0 0..0
r.APPLICATION
OF LOADING
The application of the loading is an important part of structural analysis..
. Methods of combining the types of loading include the use of ·theLOAD· ca rd s. 5.
! r
l'
.ase in c control I deck the subcases can be combined in any linear manner.0 0. The most commonly used being FORCE. MOMENT and PLOAD cards.. . as this can affect the modelling process.
10
6zcL.N2.
5.0KnM 200 in the MX direction (basic) .
of the common cards ~sed CID
F
follows. 2 ..
00 0 .N3 relative to the coordinatesystem 'CID. A description
1.the .0
t. of in the direction given by vector Nl. a) b) c) d) e) f) g) The loading can be broken down into several distinct Symmetric and Antisymmetric unit cases point loads distributed loads pressure loads inertia loads fully balanced cases components
Each of the types can load the model using any of the available Nastran loading cards. One could say that without loading all models are unusable.N2.and. of magnitude F 'and in the direction given by the vector Nl.. types.FO"t'"examp'lecrrorCe" O""f. 1000 CID
00 0 . A wrong assumption can render the model useless or produce meaningless results.0
0..0 N2
.8
N3
.
C. and z as defined by the right nand rule. 1'1 refers to G1. ID3*53 etc.etc
~
This is a simpler pressure loading card in that a constant pressure l' is applied to panel elements E1.3. with an overall ultimate factor of 1. The pressure loading acts as forces in the Z direction. Y into the panel. Load case 10 is formed by adding together the case I01 multiplied by the scalar 51.0. and care should be taken to check that the loading is as expected.
6.E..
G\\JE.0. 102*52. the DIRECTION of which is related to the OROER of the grids on the card.'.
eL. X is in direction Gl to G2.0
30
. but this can be corrected by changing the sign of the pressures on the offending panels. with an overall scalar of 5.
This pressure loading card is used if the pressures are varying across the panel in that pressures at the corners of the elementcan be specified.0
20
1. POSITIVE pressure gives rise to loads in the element local axis Z direction. Fot triangular panels only 1'1 to 1'3 are defined..E3 etc.
~
f
l t
r
f
PLOA02
IO
P
E1
E2
E3
••. 5.'
PLOAD
ID
P
Gl
G2
G3
G4
The PLOAD card' applies a pressure P to the panel defined by grids Gl to G4. see above.E. 20
.0.ttexr
E4
10 ~A L. Thus in the following example :Aerodynamic Inertia U/C load 10
I
i
i
!'t'
I
=
10 . 1>1sae TloN .E2. 1'2 to G2 et~ where G1 . +Ve.0
10
9. 4. or Gl to G3 in the case of triangular panels. The resulting LOAD card is :LOAD 10 1.
tt
~ ~.~SUR.
3
PRE.S
koAl>
IN
THE.5.'
i'
~ ~ ~
~
LOAD
10
5
51
I01
52
ID2
53
103
The LOAD card combines any of the previously set up types or componants of loading to form a single load case for use in the requested analysis run.G4 are the grids defining the panel connections. Thus defining the grids in different orders results in different P05ITIVE pressure directions.5
')
2. The pressure loading results in direct loads at the corner grids.. The positive direction of the loading applies as above. and the U/C loads ~ scalar of 1. the inertia a scalar of 9. PLOA04 1·0· EID 1'11'3' 1'4·
~. 30
The aero loads have to have a scalar of 2.
X aft. All accelerations also follow this sign convention. if they are available. but it is important to check each componant of the loading before it is used in the analysis.. as it can be run online on the Vax. Condition .same '"::omponant. All loading should be checked against the documentation before proceeding with the analysis runs. Type of loading . The balance information is vital.g. However this has yet to be implimented fully and a mixture of sign conventions'.. c. If the summation of the loading about a reference point is known from the loading document then checking is eas.y. totals about a reference point
There is a general requ i remerrtsccn.
._Both overall resultants and shear. 4. forall loading to besupplied in a consistant manner. case identifier etc 2. Name . Items should include :1. if on a file 7. aerodynamic.. Configuration . flight parameters. moment and torque about an axis (which may be swept)can be generated. Data set name and where to find the data. other than simple cases to check the model. Z up. Sign Convention 5."even. net etc.~appear' inthe . The PPS check loading facility should be used for this. position.. Loading details if not on file 8.
3. where the sign conventions and balance information are capable of a simple transformation to the basic A/C axes. Thus be care~~ll to check the sign conventions used. EFA..r •• _
.. It is one thing to apply a pressure distribution or set of loads. Balance information .DEFINITION

OF THE LOADING
All loading.for _the . fuel state etc. It is quicker to use PPS than wait for Nastran runs on the mainframe. if it is not known then ASK for it to be supplied.. CHECKING THE LOADING The Nastran runs will output the overall applied loading resultants. it another thing to be in a position to check that the loading is correct. inertia. mass.)
6. and the rotations following the right hand rule._ loading documentation. and convert these to the basic A/C system (or your model system) for checking. with/without stores + details. should have an adequate description/definition supplied with the data. No left hand sign conventions should be used... Y to stbd. Proof/Ultimate
(UF. ie.
Thus symmetric loading will be applied to the model constrained symmetrically. Symmetric component . which will be Asymmetric.a) Symmetric
and Antisymmetric
components
Loading applied in particular to fuselages and wings can normally be subdivided into symmetric and antisymmetric components.Port) Antisymmetric
100
component • 1/2 Difference ISo 'if(
Kr/
t
~
I
SO'/.. The method of splitting up the loading is shown below.200 Kn The true Stbd loading is thus Symm+A/S The true Port loading is thus SymmA/S
This process can be carried out on any fuselage/wing assembly. The two components of loading are thus applied to the half model with the model constrained in the relevent manner. and on any structure modelled with planes of symmetry. I
PoRr
!
i
!
t
t
I
Full model Symmetric Antisymmetric The full model loading can be split up into the components using the folowing formulae. It is assumed tha Starboard is +ve Y. The two resulting sets of subcases can be combined to give the correct Port or Starboard output.H
I
Total loading Symmetric Antisymmetric ISO50 . This is vital if only a half fuselage has been modelled. and antisymmetric loading applied to the model constrained antisymmetrically.100 Kn 150+50 .1/2 Sum (Stbd + Port) (Stbd .
.
flaps etc. unit
2
cases . Rigid body movement cases •• The normal checks of applying uni t cases to a . unit cases conSisting of distributed (perhaps coarse) loads with unit resultants about the origin. b) if the model is part of a large assembly of superelements. sho..d..O..ON/mm internal store attachment cases.:ree.ul. Types of unit cases are :1.
. unit shear. Balancing cases . 4. Checking cases . Flexibilities are normally produced by inverting the KAA matrix.. movement cases where the structure has imposed upon it displacements corresponding to the 6 D. 19 inertia. The reasons for the use of these cases are as follows. This is usefull for full A/C work or in setting up noncritical cases where the resultants only are known.st ruc. 5. torque etc. l. .1 Radians. 2. the structure should be'·st·ress <f. However if this is not possible unit load cases have to be set up for each freedom required and the resulting displacements at the points used.._
b) un it cas es < By definition these ~re cases set up in order to be used either alone for checking purposes or by scalar combinations to form up actual cases. There should be no significant SPC forces recovered.
3. U/C cases etc. Rigid body checks' may show these to be significant and thus some fix must be applied to the model. moment. c) freefree stiffness matrices (KAA's) should have rigid body checks carried out on them if they are to be passed outside the department or issued officially and archived.. The structure is supported in a statically determinate manner and the 6 rigid body motions applied using SPCD data applied to the support D. Unit cases for flexibilities etc.inc~ude. body. present.fOT" oody:·motitm:. However the portion of structure may be subject to large overall movements when assembled (a fin for example moves bodily on the rear fuselage) and these small SPC forces may be significant.ri. These can be used to trim existing cases to meet the exact balance resultants. ie.tu re. loads on tips of wings."rig:._..O.d .~··These checks are of particular importance if models are to be passed outside the department in that they represent a more severe test than is usually carried out.F.
a) insignificant SPC forcces produced from a loading case may occur at areas of low displacement.F.i:d ·.
.also. Unit actual pressure. Thus there will be 6 subcases.. checks on local rigid supports may not reveal significant SPC forces. typical translations being 100mm and rotations 0.gi.
dummy load data for each subcase.
. where n • 1 to 6. 3.F. 2.O. The following page illustrates the method.d} similarly freefree matrices recieved should always be checked for rigid body behaviour before use. The displacements for each case should have visible the enforced case and the SPC forces should be insignificant. 4. given in 1. The Nastran deck for carrying out the checks should contain the following items. and having ID's 16. 6 subcases calling up enforced displacements using LOADn. 1. The run should request all displacements and SPC forces. SPC data to hold the structure satically determinate and referenced above the subcase level. ·SPCD data setting up the 6 cases using the D. ID • 1 to 6.
Hf:N'r
sue...L.t=ottef..0
o· 0
SP~ 1:>
Sfe...
"f'fPIC AL Sf e
r:...s
~PPll£S
.:b
+
+
.00
r. X
222
2.0
101
200
lOG .l:>
2.£: WITH ~IG'l:> NO 30b't )1o\JE.COuE.0
.
sf'~
l> :.
Sfe.Jt FoR.eE:. SuPPd~. 2
t'oc ~~
y
I.
3
100
100
101 2...
J)lspu~eE..>HQ U E H
. . 100 SPCI t"b
src
~pel
I
I:b
2....
E}o( i
e.S.
2
100 ..E.l)
E..
I
x .•
0.0 101
's 2..t\R
~Obl\""'f
fvSc.00
2.
..
00
2~
200
100 101
S?e 1>
SPCb
SPCb
100 100
100 .~
bATA I
2.t\ Ll
A PPII
IOO.Nr $(~USC.
2 2..
oR..
NASTRAN
1:.:/
:bf\TA
SPc: !:>
FoRce
2
100
'.
/00 200
20·0
200
+
iQ
/0 I
2
2..o At> ::.
. 100 ._.
'Nr~NAL._J 1
I
SPc..0
.
_ 

(
I
t=IN
"evES
R.He.A
oN
~T~ueTt)R.:t>
SPe....0'0
1 0'
fI4J. II.S:
GE..TS . ALL ::.suacAS~
L...eK..
STATl e.
Sl4oUL..0
100'':)
:butt tl Y FoRCE
~
'c..1: _
"
~..1>
ItE.
_}:
lC&000
FRAME.·
o
Ii. test specimens etc may have uniform loading which can be easily calculated using the geometry of the grids on the section.
These can be from many sources
:
1.
014 APPLIEl)

TO
DNLY
~EN E ItALISE. Calculated
simple loading •. Applied
directly
d) Distributed loading Often the aerodynamic and inertia loading is seperated and supplied as shear. from the loading documents.. The analysis user must spread this loading onto the available grids in whatever detail is required. The final distributed loading must be checked against the known resultants and trimmed if not correct. Ole or engine loads. A LoN G.roo
1.AG€
.L
G=1t. applying the resultant to all the grids or to a generalised point linked to a selected set of 'hard' points.
2.oAb(N
6.~S
APPLtf. local calculated
loading .Ae.F TO AL.F uSE \. . from attached items . moment and tor. For fuselages this can be carried out by dividing the diagrams into blocks and applying the block loading to grids at the available frame stations..h
• .0Al>ING
fOtNT'
AERc)b'1NAH1C Sl+e.que diagrams along a fuselage...b
&... The frame· loading carr t·husbe as·· detailed as necessary.
\. for example. given in the loading
3.. foreplane loading can be applied directly to a reference point if it is connected via a RBEl to the bearing and jack points. .• due to equipment inertia not modelled.c) Point loads.
fuel etc
(}
canopy side loads. This adjustment in loading should be noted since for balanced cases the anlysis loading may now be incorrect. . Application by hand is frought with the problems of panel local axes.. one face at at time until a balance is obtained. Adjustments made to the loading planform can then be either ammended back to the analysis condition and the loading regenerated. PPS applies these pressures to a set of surface elements and generates either a set of PLOAD cards (taking into account the loacl axes) or a set of FORCE cards. if acceptable.
Using the PPS pressure loading facility. Internal pressures 3. the basic data 'is normally obtained from a file on the mainframe. Extrenal pressures 4. It may be usefull to run a constant pressure case for checking out. either the panels should be adjusted in shape or additional grids and panels added to give EXACTLY the document planform. radome lops etc. followed by a table of pressures.
Pressure loading has to be applIed folbwing instances :
1. If in doubt use the Case Contro~f' card OLOADALL. The PPS pressure loading faclity.e) Pressure
loads to a model in the
lift. Blanket pressures cabin. with dummy elements. or by using the PPS pressure loading facilities . Refer back to the document source if excessive discrepencies are present. and should be applied in stages.for the case. or the loads on the additional grids disarded. This will print out the forces on the grids from the P~OAD data and any sign reversal should easily be seen. Flying surface pressures which generate 2. This consists of a set of spanwise locations and chordwise percentages at which pressures are defined.
The pressure loading must be applied by hand using PLOAD data. If the analysis planform does not match or cover the loading document planform. for designing local structure.
. intake. Applying the pressure cases should then give the required resultants as given in the loading douments. The user should set up a set ·of surfacepanels lorthe ·loading·· and fill in any gaps between controll surfaces etc.for pressure loading and for internal pressure generation due to fuel inertia.
.ottl:>INGWIl. f\N J) J:) IR...lbS .. UP.: . \.NERA't£l:> SETTINGGE:ol1£ TR Y..
\'c: .
.ut1E.loss IN !.xTRA GR..
.V£F!Y
7.0 P\ 'b IN l1C.AP~
\N
l10:bf.ATE:.»
E..
c
=0'0
foR.. H.L 'Be GE. uP Wlnt . tH et~INGPuR ?OS~ Lo t\ l> I N G.
SuRFACE.l
r ~ r! r
t
loA
t
1>1 MG
t
Fl.'f
c.o n
PtSSUt1e.oAl> Htt'f "BE
~AS
"bl\LF\NC. Y
I
~
"
\
• FoR. '1
_
j •
•
J. MAl>e.L.
DC () M E.ATE L.
SPAN WISt:
STF\TloH5
l>ISTR..l:::t>.'5HAPE....I1.l> STArto N ..
AN:t:>
"to HAKE.. PU~NFDRt1
~F
• PRt.AT
i
ii
E.TR'f
•
•
1..\J)S uP PL~~ FoR.CT
TH E.rll.NTAT10N.t
PANe. ell.1!oN 0 F 10 Al> 1N G
» oc. S.
II
A'DJ) EXTRA GR.Nl' P~AN ~D RM .:hoc.
J)
GEDl1E.
001
Ot=
Pl!\Nt.. l>Ei=INE.
LOt\1)ING
1=01<.D~b
POINT
AT
fVER.AT Pt.
CHECK
J)lS~AItl>
of
LoA'l>It4GkOAl>INGINT E REST
USINGDN
IN
PPS .oRRE:..N TA IfoN ASS U H ES A FL.uT\ONS
I
[ PtRE.SlioOL:b fiE:.Nr
f\CTUAL wING. E.EC.s
Pt.
.SSVRE:5 SP!\NVJtS E.
PRESSuRE.L
BDUN))~
R.)OCUH E.f:. E.PAN E. S Sri00Ll::> 'BE.L IN AN. PANElS
ANAl'f$l$
HO'J)E.ro lOPrl>INGJ.
The PPS Fuel Tank Pressure program calculates the internal fuel pressures and inertia effects of a fuel tank subjected to any form of linear/angular accelerations and angular velocities." .generalthe acdee. h
BY
L.t1
'J)E.~
IC \... at a constant Z or on a flat plane using a local coordinate system. and create the loading panel geometry likewise...
l>E. FIN I:..'
SUHnp. . The loading may be output as FORCE.. the latter being used for optimisation purposes.FIHlifON
Po fNTS
~ UtL
• S'fSTE..r fA
..Geometry of Loading Panels . Thus the user should find out what assumptions have been made in the loading.inwill not correspond to the real t.. a unit pressure case should be run in order to find the volume and mass.. fuel density and system pressure are also input tOgether with the controll data to define the inertia cases required. and that the loading is normal to this. The load summations are printed out together with the fuel inertia properties relative to thetank···cg.
t
i
. Bt implication drag or thrust loading due to the curvature of the real surface is not generated from the pressure distributions..
rAN K
} NER. \".E.O~IT1£S
hE.. Fuel Pressures .. If the mass is different to the required mass the fuel density should be adjusted before actual cases are =un.. • : ~ (:. A ({oN
l>ATA
S
0 UTPuT
eAN 'E=>E..
•
lNTE..I NeA~
J ANGULfti!?
Ac.RNPlL.b DUrPI..'.ank due to internal structure volumes not being accounted for.J'i
rt. The aerodynamic theory which produces the flying surface loading assbmes that the surface is a flat plate.rfONS

~ . The centre of rotation. The program is easey to use and has been used extensively on the EFA wing to produce fuel tank loading..SSURE
• C AS ~ '5
~N'b •
V e t.~\T 'f PR... In addition if the surface has a small dihedral the loading may still be specified as being in the A/C 'Z' direction.C
e.. PLOAD2 or OPTPRES cards. The tank is defined by a set of panel elements and geometry on the boundary together with additional internal points to indicate which side of the panels are internal to the tank. '"""[C .\~~
L
.Since.
I)\. these effects must be added later. L£ R..
j ~ \.
)
CALC rJ LA Tf..
Fuel Inertia may be generated from the fuel mass allocated to the tank panel grids. Lve Ly.. ·This method. the inertia cases must be formed from the mass definition.. 1. However supplied. Alternatively the fuel pressure loading program in PPS may be used in which case the correct fuel inertia FORCE data will be generated together with the pressure effects.The weights department carry out this task..
.ens en. structural.I
f) INERTIA LOADING Inertia loading forms a vital part of many loading cases. If several concentrated masses are present it is better to deal with them as a whole and use the PPS Inertia Program to calculate the loading. concentrated (non structural) mass etc. where the mass and geometry information is used to generate equivalent FORCE data for any combination of inertia . or as unit FORCE data for linear accelerations only.using rigid or elastic elements . Concentrated items of mass can be treated in several ways. and their data is converted to CONM2 card data giving the mass at the grids. CONM2 data can be used directly by the PPS Inertia Loading program.be uaed. either as CONM2 data for use with the PPS inertia program (in which case no fuel PRESSURES will be generated). A change in mass specification may give rise to discrepanccies which need to be resolved before acceptance of the inertia loading.. 3.the grid then having inertia forces calculat by hand. Structural inertia loading can be supplied in many ways.._. Several types may be encountered.ext. Recently the trend has been to define the mass distribution directly at a set of interface grids . some times by mass per unit area in preliminary calculations. In this case the CONM2 data will include the inertia properties of the section.. by mass distributions.. Fuselages tend to be defined by running mass diagrams. on EFA.. Wings etc. Large masses may have self inertia which requires aplyuing as rorational forces.c.parameters. CHECKS Always check that the resulting inertia loading is as expected in the loading document... In fuselages the mass of two half bays may be lumped to a generalised point at the associated frame position. It may be lumped to a grid at the cg of the mass and this grid connected to the structure . The mass may be allocated to a set of grids and the inertia effects calcuated by hand. fuel. has .
. 2.
The agreed set of loading cases have to be applied to all components and the resultants on each componant should agree to a predetermined total.fairing loads 4.
In order to ensure that cases are in balance. A full AIC model may consist of seperate componants assembled in a reduced form using Superelements.
The organisation necessary to impliment balanced loading the items to be agreed include :
1. eg. However for full AIC cases an EXACT balance is necessary. A typical type of full A/C balance sheet is shown on the next page. Small resulting SPC forces may flnally be seen. Critical loading. Mass ~tandard 3. cases 2.. The full AIC model should be supported on a set of statically determinate freedoms at retlatively 'hard' points.•
2.
is extensive. In the case of flying surfice loading the planform of the model usually does not correspond exactly with the loading document.. Componant boundaries for loading . but these should be insignificant compared to the local stress levels«< I.
g) Balanced Cases All loading cases should have balance information in order to check the case. and all componants should be trimmed to give EXACTLY the correct loads. ego front fuse load cases would not require detailed Fin loading. Once this is obtained the models can then be interacted to generate the interface loads and post interaction output produced. and some sligh~~out of balance may be acceptable.
A matrix type of table should be drawn up to define the type of loading and responsibility for all balanced cases. Standard of loading . not all compnants need to be loaded in detail for all cases. each case should have a full AIC balance sheet.1.
. load sharing may have to be carried out between componants.OKn). and thus a compromise loading may be accepted.
WING
Po&rr
FIN
ST~l>..
y ••
Con fig Z.
I
ST13~.Asle
S/lSN
f /.
r
x
t: I
I
!trrA~E e /FuSE ~/FtJSf.
./t.£ FusfLA c.
sr.Nr'ON
loAb
CD~\JtNTIO"'"
')
RE):"ER£NCE.iA
N(S
At
Ale
oRIGIN
'0
r
C!E.A (..£.~Itl
I
~. FuSG.
q
.A~ PO~I WIN(.
POI~'TS '(
COI1Pf
(1.f.LA<rE REAR.E.
Mass State
x.AGE INTA r::.
~ k"
C Ot1 PoNE..S£L.£ .f {: L I~ E.
~». ~T
F:RONI
Fr). FoR£Pt...v..NiR.
GFoR£..
I
POII\f'rS
"<. sus« LAGf. W /101 a:PORT
LoA'bING)(
US
'(
U L.AGf 'l:NTA K.. WING~r.
LOA "J)/NCF(o~rr
FUSO.AtJ£..rAN'7"S
=
AT
CDitfONENT'
fUf..

I
I
ST13~. PLA
we.
POINTS
~
H~
~
.
FIN
WI'"
foR"7" .Ale
lO~/N
G
S!GN
~/llNVt.
S'ISTCH
SHOwlV
l.
FoREP!.
5..
"'
.
"Z:N
Ale. kft Mass
q
61
61
6.&.
. l.f_ P. az
q'
r p r/sJ r/sJ r/s2 r/s
P
..
r
Mach Alt .
/USl. CeNTRE. F /PI.A "'c fOR£PUtN£
rorAL S
Inertia Parameters "Flight Parameters
)
q
ax
q
a.
q
r/s
C "'lL
rls
t.