V53PR0906133 Exp
Uploaded by
Calin ComanV53PR0906133 Exp
Uploaded by
Calin ComanDetailed FEM Modelling guidelines
REFERENCE DATE
for Linear and Non Linear Analysis V53PR0906133_exp Issue3.0 November, 2011
– EXPORTABLE VERSION
ATA CHAPTER 00 GENERAL - TECHNICAL INFORMATION WEIGHT AND CG
A/C SECTION 00 COMPLETE AIRCRAFT
A/C SUB ASSEMBLY
A/C PART
CONFIDNETIALITY Company Use Only Exportable
ANALYSIS TYPE FINITE ELEMENTS
A/C APPLICABILITY* XW-A350|SA|WB|LR|DD-A380|MT-A400M
*Indicate the last A/C-Type-Series WV, which had an impact on this report. For a Post Mod, indicate it in the title (Post Mod XXXXX)
SUMMARY
• This document intends to provide guidance to Airbus COE stress teams for Linear and
Non Linear Detailed Finite Element Static and Fatigue Analyses.
• The scope of applicability in terms of software and versions vs. analysis type is the fol-
lowing:
- MSC-Nastran 2005 for Linear Static (Sol 101) and Buckling Analysis (Sol105)
- Abaqus Standard 6.8 / 6.9 / 6.10 for Non Linear Static Analysis
- SAMCEF 13.1-x for F&DT Analysis
© AIRBUS FRANCE S.A.S. 2011. ALL RIGHTS RESERVED. CONFIDENTIAL AND PROPRIETARY DOCUMENT.
PAGE B-1
Detailed FEM Modelling guidelines
REFERENCE DATE
for Linear and Non Linear Analysis V53PR0906133_exp Issue3.0 November, 2011
– EXPORTABLE VERSION
LIST OF DISTRIBUTION
DEPARTMENT/ NAME P.O. BOX COVER NOTE NOTE
COMPANY PAGE WITHOUT WITH
ONLY ATTACH- ATTACH-
MENT MENT
Diderot M0199/6
Agreement for
Airbus distribution
© AIRBUS FRANCE S.A.S. 2011. ALL RIGHTS RESERVED. CONFIDENTIAL AND PROPRIETARY DOCUMENT.
PAGE B-2
Detailed FEM Modelling guidelines
REFERENCE DATE
for Linear and Non Linear Analysis V53PR0906133_exp Issue3.0 November, 2011
– EXPORTABLE VERSION
RECORD OF REVISIONS
EFFECT ON
ISSUE DATE REASONS FOR REVISION
PAGE PARA
1.0 4 June 2009
2.0 11 Feb 2010 Addition: F&DT section, Specific Applica-
tions
Further information: Materials, Fasteners
3.0 Nov 2011 3-4 Scope Abaqus versions supported
Release of AM documents
6 Analysis Definition of model levels
objectives
15 to 18 Nodes Use of Nastran offsets in linear analyses
reference
for Shell
FEM
25 Mesh den- Mesh density for omega stringers
sity
29 to 39 Element Sign convention of bending stress flows in
types shell elements vs. ISAMI
Fastener modelling and properties for sizing
purpose
44 Element Fastener modelling with Solid Meshes (Na-
types stran)
52 Material RF calculation with Plasticity
modelling
53 Material Composite material
modelling Reduction Factors for laminates
54 to 56 Material Composite material modeling with Nastran
modelling
62 to 64 Material Typical fuselage parts orientations
modelling
73 to 75 DFEM DFEM Connection to the Global FEM (Ab-
Connec- aqus)
tion to the
Global
FEM
78 Loading Mechanical load case
79 Loading Thermal loading
105 Model Va- Element quality checks
to107 lidation
101 Model Va- Reference to DFEM Validation Check List
lidation
131 to Post- Reword
134 processing Reference to technical notes
© AIRBUS FRANCE S.A.S. 2011. ALL RIGHTS RESERVED. CONFIDENTIAL AND PROPRIETARY DOCUMENT.
PAGE B-3
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
performance inspired
Centre of Competence Structure
General Detailed FEM Modelling
for Linear and Non Linear Analysis
Prepared by
ESAZN (Stéphane Le Drogo, Marion Touboul, Olivier Le Roux)
ESKA (Sébastien Vidal)
ESAS (Bruno Rosemberg)
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Content
• Scope
• Analysis objectives
• CAD Environment
• Nodes reference for shell FEM
• Mesh density
• Element types
• Material Modelling
• DFEM Connection to the Global FEM
• Loading
• Model Reduction Techniques
• Model Validation
• Specific Applications
• Post-Processing
• Fatigue and Damage Tolerance Analysis
Page 2
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Scope
• This document has been developed to provide guidance to Airbus COE
stress teams for Linear and Non Linear Detailed Finite Element Static
and Fatigue Analyses
• It was initially dedicated to A350 but can be extended to other Airbus
programs
• Finite Element codes referenced in this document are
• MSC-Nastran 2005 for Linear Static (Sol 101) and Buckling Analysis
(Sol105)
• Abaqus Standard 6.8 / 6.9 / 6.10 for Non Linear Analysis
• Some general recommendations are given, that can be applied also to
SAMCEF software
Page 3
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Scope
• Transnational Airbus Methods documents released by Airbus Non Linear
Technical Centre (ESAZN) for Abaqus usage supersede some sections of the
previous issue of this document (V53PR0906133 Issue 2):
• Element types
• General
• “Airbus Recommendations for the Selection of Structural Elements in Abaqus NL FE
Analyses“ , Ref. M5325.1
• Fastener Modelling
• “Airbus Recommendations for Mesh Independent Fasteners in Abaqus NL FE Analyses”
Ref. M5325.4
• Material Modelling
• Plasticity Modelling
• “Airbus Recommendations for Modelling Plasticity in Metallic Materials”, Ref. M5325.2
• Other official documents have to be consulted for the use of Abaqus in
Airbus
• “Airbus Recommendations for Use of Abaqus for Non-Linear Finite Element Analysis”,
Ref. M5325.0
• ”Airbus Recommendations for Modelling Mechanical Interactions in Abaqus NL FE Analyses”,
Ref. M5325.3
Page 4
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Analysis objectives
• The objectives of an analysis are a very important point to consider
before going through the modelling process. These objectives
determine the analysis methods and strategies to be used.
• It is important to fully understand the analysis expectations / needs /
requirements, as this will help to anticipate the potential non-linear
behaviour and failure modes / sites of the structure
• At this stage, it must be decided whether and where linear
assumptions can be made. Linear assumptions always lead to a
compromise between accuracy (↓) and performance (↑) and
therefore must be used with care.
• An analysis strategy must be built in order to support the design of
‘fit for purpose’ model or multi-level models (see next slide) with
respect to the objectives aimed.
Page 5
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Analysis objectives - Modelling Levels
Screening,
Level0 Model Driving,
Predicting
Screening,
Driving,
Predicting
Level1 Model
Sub-model, no refinement,
L1 modelling methodology
Shell / Cont. Shells
Level2 Model Analysis “Wishbone”
An Sub-model, refinement,
a ly L1modelling methodology
s is Level3 Model Shell / Cont. Shells
“W
is
h bo Sub-model, refinement
ne re-mesh from CAD, significant
” change in modelling methodology
U Level4 Model
pp Shell / Cont. Shells / Solids
er
Ar
m
Level5 Model Sub-model, refinement
re-mesh from CAD, significant
change in modelling methodology
Shell / Cont. Shells / Solids
Page 6
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Analysis objectives
• The document will now detail some recommendations for the
following typical steps of the modelling phase:
• CAD-CAE
• Type of elements
• Meshing rules
• Modelling of connections between parts
• Material modelling
• Boundary conditions
• Loads
• Model reduction techniques (sub-modelling, condensation)
• Post-processing methods
Page 7
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
CAD Environment
mailto:[email protected]
Page 8
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
CAD Environment
• The simplification of a CAD part is the first step of the modelling
phase
• It is linked to the idealization principles and assumptions that have
been decided for the problem to be solved
• The level of details to be kept is linked to the analysis objectives and
the size of the structure to be analyzed
• The simplifications operated in the CAD must not have a significant
influence on the stiffness of the parts considered (internal load
distribution) and on the behaviour of the area of interest (boundary
condition effect).
• Useful information linked to composite stacking can be directly taken
from the CAD if Catia CPD files exist. Some tools allow the automatic
mapping of the properties to the elements of a model.
• However, be aware that in Catia V5r18 (current Airbus version) the
CPD is linked to a part and is not repositioned in a CatProduct
Page 9
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
CAD Environment
• Examples:
• the fillets in omega stringers may have a clear influence on the stiffness
and the local buckling behaviour
• For clips unfolding problems, or for the simulation of a frame 4-point
bending test, the fillets of all parts must be kept in order to obtain a more
realistic behaviour
• For fuselage stiffened panel stability problems, the modelling of the
frames and the clips is not the key point. Simplifications can be applied on
the surface fillets.
• In composite panels, the ply-drop-off areas are generally simplified to
avoid creating small elements in areas not to be analyzed (lack of
knowledge of the ply-drop-off real structure)
• In bolted joint areas where bolts are modelled in a simplified way (spring-
type elements), it is not necessary to keep bolts holes in the CAD as this
type of modelling does not allow to analyze the stresses in the vicinity of
the holes. Keeping the holes in the mesh generates mesh refinements
constraints, penalizes CPU time and incites to process wrong local
stresses.
Page 10
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
CAD Environment
• Impact of element choice (shell vs. solid elements)
• The use of solid elements is recommended when the objective of
the analysis is to capture stresses in local areas (ex: fillets) where
the shell theory is not applicable (plane state of stress). In this
case, the initial CAD part should be kept intact with the only
objective to prepare the mesh.
Page 11
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
CAD Environment
• Impact of element choice (shell vs. solid elements)
• For all cases where the shell theory is adapted (thin-wall
structures), some simplifications of the 3D CAD parts have to be
made in order to ease the generation of the shell mesh.
• The user has to keep in mind that the solution will not be accurate
in areas where the initial CAD has been significantly simplified
Deletion of fillet Strains close to the fillet will not be fully accurate
Page 12
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Nodes reference for Shell FEM
(mailto:[email protected])
Page 13
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Nodes reference for Shell FEM
• Two options
• Mid-plane
• Can be used in all areas
Mid-plane reference
• Reference surface + offset
• Not to be used in bolted joints areas, where transfer of momentum between
plates may not be well estimated
Tie surface reference (example)
Page 14
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Nodes reference for Shell FEM (Nastran)
• Use of offsets vs. Linear Static analysis
• Use ZOFFS, especially if bending stress flows have to be extracted in shell elements
(REF plane always at mid-surface).
• Using Z0 leads to wrong results in homogeneous stresses and bending stress flows
(see table below)
Page 15
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Nodes reference for Shell FEM (Nastran)
• Use of offsets vs. Linear Buckling Analysis
• General recommendation = AVOID using implicit offsets (ZOFFS)
• If used where differential stiffness effects are important (buckling),
the results will be wrong because the offset is represented
internally by an internal displacement constraint which does not
generate differential stiffness.
• However, if the implicit offset is small, the error to due omission of
the differential stiffness will be small, so the results can be
plausible.
• Other cases where the offset does not play an important role in the
results also give plausible answers (far from analysis area)
Page 16
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Nodes reference for Shell FEM (Nastran)
• Use of offsets vs. Linear Buckling Analysis
• In buckling analysis, use PCOMP Z0 field with care
• This assumes that the neutral axis is positioned at the mid-surface
of the lay-up, which is not the case when the layup is unsymmetric.
If the geometric error is small, the error on the results will be small,
but this needs to be checked case-by-case.
• In case of unsymmetric lay-ups, try to EVALUATE the possible
error made by using Z0 on a simple case before running complex
models
• The question of mesh density vs. the order of buckling modes has to
be considered here, as a coarse mesh and high modes orders will
imply larger nodal rotations and thus higher sensitivity to differential
stiffness effects.
• Mesh density recommendations for buckling problems is addressed in
the next section
Page 17
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Nodes reference for Shell FEM (Nastran)
• Use of offsets in Linear Buckling Analysis
• Example to EVALUATE the possible error made by using Z0
Error acceptable but growing with order or buckling mode for a given mesh density
Page 18
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Mesh Density
(mailto:[email protected])
Page 19
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Mesh density
• The accuracy of some FEM output data (stress/strain fields, buckling loads) is
linked to the mesh density used
• A mesh is said to be converged when further mesh refinement produces a
negligible change in the solution
• For example, the level of stress or strain accuracy, which can be
characterized by the discontinuity between adjacent elements (d),
depends on the objectives of the analysis
• Very accurate : d<5%
• Accurate : 5%<d<10%
• Medium : 10%<d<20%
• Coarse : 20%<d
• A model is not only a mesh, but a full set of hypotheses
• idealization principles, material and modelling data, boundary conditions,
loads, …
• The mesh refinement must not make the user forget about the model
assumptions
Page 20
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Mesh density
• Buckling problems
• Skin mesh density
• The mesh density must be sufficient to capture the highest potential number of
buckling waves in the skin pockets
• The buckling pattern depends on the pocket geometry and the loading
• n ≥ 5 elements per semi-wave
Page 21
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Mesh density
• Buckling problems
• Skin mesh density
• Examples
• Shear
> 10 elements
> 25 elements
Page 22
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Mesh density
• Buckling problems
• Skin mesh density
• Examples
• Compression
> 5 elements
> 25 elements
Page 23
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Mesh density
• Stability problems
• Stiffeners : L, I, J, T profiles
• n ≥ 3 elements per half attached flange in order to provide realistic
stiffness boundary conditions for skin pocket buckling
• n ≥ 3 elements per half free flange in order to capture in-plane bending
behavior and accurate local stresses, especially for post-buckling and
collapse analysis
• n ≥ 5 elements in the web if any local buckling mode is expected (n ≥ 3 if
not)
Page 24
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Mesh density
• Stability problems
• Stiffeners : omega profile
In order to provide realistic stiffness boundary conditions for skin pocket
buckling for Level 1 models following recommendations should be followed:
• n ≥ 2 elements per attached flange
• n ≥ 2 elements in the head of stringer
• n ≥ 2 elements in the web
Note : if any local buckling mode is expected on the stiffener or if detailed local analysis is needed, the
mesh density should be increased in a sub model of superior level (see previous slide Modelling levels).
Increase of number of elements in
Typical meshing of an omega head and web of an omega Meshing of an omega stringer
stringer in a Level 1 model stringer in a Level 1 model in a sub-model (example)
3 elements
4 elements
Page 25
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Mesh density
• Joint areas (plates shell FE mesh)
• Mesh constraints in the plates
• No mesh coincidence constraints at fastener locations
• Use of mesh independent fasteners
• Regular meshing of plates
• Respect FE codes element shape criteria
• Avoid triangles
Page 26
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Mesh density
• Joint areas (plates shell FE mesh)
• Mesh density
• At sub-component and higher scale
• 3 elements between 2 fasteners can be sufficient to have a good accuracy
for bolted joint analysis
• Example
• Same fastener diameter, pitch = 4.5 φ, element size ~1.12 φ
• Same fastener diameter, pitch = 5 φ, element size ~1.25 φ
Page 27
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Element types
(mailto:[email protected])
Page 28
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Shell and solid element types (Abaqus)
• Element type selection in Abaqus models must be done following
recommendations of
• M5325.1 document – “Airbus Recommendations for the
Selection of Structural Elements in Abaqus NL FE Analyses”
• General comment for 3D meshes using 2nd order Tetra elements
• C3D10M have to be used without projection of edge nodes on
the geometry, see the picture below
Page 29
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Shell and solid element types (Nastran)
• Linear Static analysis (sol101)
• Shell:
• CQUAD4
• Solid
• CHEXA8 (min 4 elements through thickness)
• TET10 (for complex geometry , to be used without projection of edge nodes
on the geometry, see picture of previous slide)
• Linear Buckling analysis (sol105)
• Shell
• CQUAD4
• Solid
• CHEXA8 (min 4 elements through thickness)
• TET10 (for complex geometry, to be used without projection of edge nodes
on the geometry, see picture of previous slide)
Page 30
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Shell element types (general)
• Sign convention of bending stress flows in shell elements vs.
ISAMI
• Abaqus is using the same sign convention as ISAMI
• Nastran and Samcef (used outside ISAMI) are using opposite sign
conventions
Page 31
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Fastener modelling (general)
• General recommendations
• Element type
• Spring type elements
• Mesh-independent (no direct node-to-node connection)
• Axial elastic properties
Method 1 Method 2
(not recommended) (Preferred, refer to M5325.4)
1 single element
Plate 1 Plate 1
for axial
Both Traction stiffness Plate 2
Plate 2
and Shear
Elements for
Stiffness Plate 3
Plate 3 shear stiffness
- series of elements - 1 single element for axial among the
- with homogeneous axial stiffness plates stacking
(106N/mm) - in parallel with elements for shear
- Shear stiffness (see next slide) behaviour (see next slide)
- no detailed analysis should be - with arbitrary axial stiffness (106N/mm…)
done concerning traction behaviour or ES/L (L= Fastener grip length)
- contact to be defined between the plates
Page 32
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Fastener modelling (general)
Shear elastic properties
Measured properties if available, or Huth formulation
If Huth is used
simple shear condition has to be taken between 2 consecutive plates (k, k+1)
in an assembly even if composed of more than 3 plates
a
k k +1 b k b k +1 bk b k +1
k t +t
Ci = ⋅ + + +
2.d f t k .E k t k +1.E k +1 2.t k .E f 2.t k +1.E f
i i
Cik Flexibility of the joint composed of plates k and k + 1 in direction I
tk Thickness of plate k
t k +1 Thickness of plate k + 1
Eik Young modulus of plate k in direction i
Eik +1 Young modulus of plate k + 1 in direction i
Ef Young modulus of fastener
b k , b k +1 Coefficients depending on the joint plates material
a Coefficient depending on the fastener type
Page 33
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Fastener modelling (general)
Notes:
k k
For composites, Ei & E j represent the homogenized moduli of the stacking of plate k in i
and j directions.
The choice of the direction in which the homogenized properties are computed must be
consistent all over the part stack.
Huth rigidity computation : illustrative example of a joint composed of 2 composite plates
with mismatched 0°fibre directions
0°fibre direction of part 1
Local-1
Local-2
0°fibre E11 E12
direction of
part 2
E12 E22
Page 34
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Fastener element type (Abaqus)
• Please refer to «Airbus Recommendations for Mesh Independent Fasteners
in Abaqus NL FE Analyses», Ref. M5325.4
• The fastening method of Abaqus, called “*FASTENER”, allows to define mesh
independent fasteners
• 2 methods are possible depending on the user’s needs:
• Fully automatic based on reference nodes (see next slide), which does
not allow to control the numbering of the connector elements created in
the *FASTENER procedure
• Semi-automatic, based on existing connector elements, which allow to
control the connector element numbering (use ELSET in *FASTENER
definition)
Page 35
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Fastener element type (Abaqus)
• The *FASTENER procedure automatically does the following operations (use of
reference node):
1- Projection of the reference node on the closest surface : creation of a first attachment
point
2- Coupling between attachment point and surface mesh nodes (structural coupling)
3- Repeat action (1 and 2) on the following surfaces
4- Associate a connector element between the two attachment points
Reference
point multiple
surfaces
attachment
points
Connector
element
radius of
influence
Page 36
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Fastener element type (Abaqus)
• Fastener orientation
• The local coordinate system defined on the connector section, Tconnector, operates
on the local coordinate system for the fastener,Tfastener, to determine the final local
coordinate system of the connector element,Tconnector_final
• General case where the directions 1 and 2 do not have to be controlled
• Keep the default orientation for the fastener (see next slide)
• Define the connector section without the use of *ORIENTATION.
Then, Tconnector=Identity
• With this convention, the previous relationship becomes
1 2
and the axial direction is the direction 3, the shear (in-plane) ones correspond to the 1 and 2
directions (see next slide)
Page 37
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Fastener element type (Abaqus)
• Fastener orientation
• General convention in Abaqus to define Tfastener orientation
• The global X-axis is projected onto the closest surface at the location of the
fastening point to determine the local x-axis. The local z-axis is computed to be
normal to the closest surface. The remaining local y-axis is computed to form a
right-handed set.
• Particular case
• If the global X-axis is (quasi-)normal to the surface, the global Z-axis is
projected instead to form the local x-axis. The local z-axis is computed to be
normal to the surface. The remaining local y-axis is computed to form a right-
handed set.
Page 38
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Fastener element type (Abaqus)
• Fastener orientation
• Particular case where the directions 1 and 2 have to be controlled (ex on
composite joints)
• Associate appropriate user defined system (*ORIENTATION) to have correct
local-1 and local-2 directions (anisotropic in-plane properties)
*ORIENTATION, name=user_defined_system
...
*FASTENER, ORIENTATION=user_defined_system, ...
Page 39
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Fastener element type (Nastran)
• Preferred way of modelling fasteners
• CBUSH element
• RBE3 couplings as defined in Patran Airbus « Stuck Node » PCL
Page 40
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Fastener element type (Nastran)
• CBUSH orientation
• The CBUSH Coordinate System is generally defined as follows :
X axis is given by the 2 Nodes defining the element
Y axis is determined by a user defined Orientation Vector lying in the
(X,Y) plane
Alternatively, the CBUSH can have a user defined Coordinate System. In this case the element X, Y
and Z axis are along direction 1, 2 and 3 respectively of this coordinate system
For instance, the Y axis can be oriented parallel to one of the 0°material axis of the connected plate s if
composite materials are involved (for example, positive X aircraft axis for fuselage skin and stringer)
In any case, it is recommended to have the same bush orientation for all fasteners
that are perpendicular to the same plane (i.e. Fasteners belonging to the same
structure)
Example of bush
orientations in a corner
fitting
Page 41
September 2011
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Fastener element type (Nastran)
• CBUSH orientation
• Remarks concerning the X axis
A special care has to be taken when defining the CBUSH node connectivity, as this
generates the X axis of the Element Coordinate System (ECS).
To avoid eventual confusion while post-processing (same Y-axis, but opposite X axis, will
generates opposite Z-axis, see below) it is recommended to have the Element Coordinate
System oriented in the same way for all CBUSH elements defining one fastener in multi-
plate joint.
Therefore, nodes 1 and 2 of CBUSH elements have to be defined in the same
order/direction through the bolted joint.
CBUSH A
Node 1
Node 2
Same orientation
CBUSH B Same Y direction
Node 1 Same order for
Node 2 Nodes definition
Page 42
September 2011
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Fastener element type (Nastran)
• Rotational Stiffness
• A minimum value for rotational stiffness is recommended, especially when contact
is not modelled
• K5min and K6min have to be computed as a function of the highest shear stiffness
and the CBUSH element length LCBUSH
1 L2CBUSH
K 5− 6 = × Max(K 2 , K 3 )×
ε 4
with 1/ε = 100, and
LCBUSH is the CBUSH length, (distance between the modeling
reference planes of the joint plates
• To avoid to calculate rotational stiffness values per fastener, the order of
magnitude of 109 is recommended for K5 and K6
• Rotational Stiffness around the fastener axis K4 can be fixed arbitrary to 100.
Page 43
September 2011
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Fastener modelling with Solid Meshes (Nastran)
• Element type :
- Parabolic Tetra elements (TET10) or Hexa elements (CHEXA8) recommended
- 2 “Internal” CBUSH elements in 3D solid + 1 CBUSH element for the fastener
RBE3
x • K1,2,3 = 107
CBUSH • K4 = 100
y • K5,6 = 1010
1
CBUSH 0-length fasteners
Coincident Nodes
(1 & 2)
K1,2,3 defined as shown on slide 33
K4 = 100
2 K5,6 = 1010
Same Y direction
Same order for • K1,2,3 = 107
Nodes definition CBUSH • K
4 = 100
• K5,6 = 1010
• Complete RBE3 coupling :
- dof 123456 for the Slave node
- dof 123 with coefficients for the Master nodes
Remark: the last term with dof 456 (to improve the stability of the relation) is removed.
Page 44
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling
(mailto:[email protected])
Page 45
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling
• Metallic material
• typical response of uni-axial test
stress
A A′
Elasto-Plastic
domain
Typical uniaxial stress-
strain curve for a metal
Elastic domain
B strain
Page 46
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling
• Metallic material
• Elastic domain
• Tension/Compression E-modulus
• Max Difference observed in conventional metals
• 5% for Aluminium
• Reminder of GFEM assumption: Tension modulus is implemented
• If comparisons with GFEM internal loads are to be made
• Tension modulus are to be implemented in the DFEM
• The average between tension and compression moduli can also be
chosen
Page 47
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling
• Metallic material
• Elasto-Plastic domain
• Data generally available
• Yield stresses
• For tension (Fty) and compression (Fcy)
• In L, Lt and St directions
• Failure
• Max stress in tension (Ftu)
• Max elongation in tension A
• Selection of relevant data for the problem to be studied
• Analyze the difference between yield stress values
• Select the material data that fit the more with the problem (structure, load
case) or build an average behaviour
Page 48
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Abaqus)
• Metallic material
• Yield data should always be given in Abaqus as true stress versus
logarithmic plastic strain.
• Logarithmic strain, ε, is defined as
l
ε = 1n = 1n 1+ε nom ,
l0
( )
where l is the current gauge length, l0 is the original gauge length, and
ε nom is the engineering or nominal strain : ε nom = (l- l0) / l0
• True stress, σ , is defined as
where σ nom is the engineering or nominal stress : σ nom = F/A0
(
σ = σ nom 1 + ε nom , )
Page 49
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Abaqus)
• Metallic material
• Defining Mises plasticity in Abaqus : σvm = f(εp)
1400
Typical material coupon tension test curve
1200
1000
Stress (MPa)
800
600
Eng Stress vs. Nom Strain True Stress vs. Log Strain Abaqus Plasticity Curve
400
Material test data are expressed in
200 engineering strain/stress. Conversion to
true stress/strain is mandatory
0
0 0,02 0,04 0,06 0,08 0,1 0,12 0,14 0,16
Page 50
Strain
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Abaqus)
• Metallic material
• Defining Mises plasticity in Abaqus : σvm = f(εp)
• Example
500
450
*MATERIAL, NAME=ALU
400
*ELASTIC
350
71000., 0.33
300
*PLASTIC
Stress (MPa)
250
306., 0.
200
385., 0.001936
150
423., 0.006757
100
443., 0.012021
50
450., 0.02913
0
0 0,005 0,01 0,015 0,02 0,025 0,03 0,035 0,04
Total Strain
Page 51
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Abaqus)
• Metallic material
• For more details about how to build plastic curves, consult document
M5325.02 – “Modelling Plasticity in Metallic Materials”
• The post-processing of NLFE plasticity analyses must be done with care,
especially with respect to Reserve Factors calculation
• RF must be based on the following formula
where the Allowable Load is the load for which the criterion in allowable
plastic strain is reached.
If this criterion is not reached for a given load level above Ultimate Load
(ex 1.1 UL), then RF can be reported as RF>1.1
Page 52
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling
• Composite material
• 2 modelling options are possible
• Macroscopic modeling
• Homogeneous material
• Elastic properties
• Ply-by-ply modelling
• Real composite ply stack
• Homogeneous material in the centre ply + 4 strain gauges plies on top and bottom
(Global FEM principle)
• The average between UD Tension/Compression(*) E-modulii has to be used in the FEM
(internal load distribution purpose), either for the calculation of homogeneous properties or
directly in the ply material properties
(* Use the B-basis value of compression Young’s modulus to account for the variation in this
property for composite materials)
• For stability analysis, an additional B-value coefficient has to be considered on the
RF(selected as the maximum between the relevant parts), leading to the lowest RF
• More generally, consult the document «Reduction Factors for laminates outside the
limitations for percentage of plies in one direction (min. 7.5%)»,
Ref.V029PR1003358 issue 3.0
Page 53
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Nastran)
• Composite material
• Macroscopic modeling (GFEM modelling rules)
• Plates complete behaviour (membrane + bending + transverse shear behaviour)
• Use PCOMP card entry with MAT2 (laminate homogenous properties) and MAT8
0°
(gauge plies) 4 “gauge” plies - MAT8
45°
90°
thickness: 0,001 mm 135°
Middle ply MAT2
135°
Total laminate thickness 90°
Monolithic 9 plies 45°
4 “gauge” plies - MAT8
0°
thickness: 0,001 mm
0°
45°
Sandwich 11 plies 4 “gauge” plies - MAT8
90°
thickness: 0,001 mm 135°
Outer skin
135°
Core 90°
Inner skin 45°
0°
4 “gauge” plies - MAT8
thickness: 0,001 mm
σ 1 G11 G12 G13 ε1 A1
σ 2 = G12 G 22 G 23 ε 2 − (T − T0 ) A2
τ G13 G 23 G33 γ A3
12 12
Page 54
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Nastran)
• Example
Page 55
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Nastran)
• Composite material
• Ply-by-ply modelling
The PCOMP property entry may be used when the element is a composite consisting of
layers of unidirectional fibers
The information on the PCOMP entry includes the thickness, orientation, and material
identification for each layer
This information is used to compute the entries of a equivalent PSHELL entry
Special layer-by-layer output is provided when PCOMP option is used
The ply material properties can be modelled with an orthotropic material (MAT8)
Page 56
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Abaqus)
• Composite material
• Each layer in a composite can have its own material and
orientation, if necessary.
• Defining Orthotropic Elasticity with TYPE=LAMINA
• Under plane stress conditions, such as in a shell element, only the
values of E1, E2, v12, G12, G13, and G23 are required to define an
orthotropic material
• The shear moduli G13 and G23 are included because they may be
required for modelling transverse shear deformation in a shell
• Defining Fully Anisotropic Elasticity
• For fully anisotropic elasticity 21 independent elastic stiffness
parameters are needed : (σ) = [D] (ε)
D1111, D1122, D2222, D1133, D2233, D3333, D1112, D2212
D3312, D1212, D1113, D2213, D3313, D1213, D1313, D1123
D2223, D3323, D1223, D1323, D2323
Page 57
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Abaqus)
Page 58
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Abaqus)
Page 59
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Abaqus)
• Composite part
• Example
*SHELL SECTION, ELSET=SKIN_S13, OFFSET= 0.5, ORIENTATION = SKIN, COMPOSITE
** AN OFFSET OF 0.5 IS USED HERE TO TAKE ACCOUNT FOR THE POSITION OF SKIN NODES AT THE OUTER
SURFACE OF THE BARREL
0.125,3,UNIAXIAL,-45
0.125,3,UNIAXIAL,45
0.125,3,UNIAXIAL,90
0.125,3,UNIAXIAL,0
0.125,3,UNIAXIAL,-45
0.125,3,UNIAXIAL,45
0.125,3,UNIAXIAL,0
0.125,3,UNIAXIAL,45
0.125,3,UNIAXIAL,-45
0.125,3,UNIAXIAL,0
0.125,3,UNIAXIAL,90
0.125,3,UNIAXIAL,45
0.125,3,UNIAXIAL,-45
** PLY THICKNESS, NUMBER OF SECTION POINTS THROUGH THE PLY, PLY MATERIAL NAME, PLY ORIENTATION
ANGLE WRT. THE MATERIAL LOCAL COORDINATE SYSTEM
*ORIENTATION, NAME = SKIN, DEFINITION = COORDINATES, SYSTEM = CYLINDRICAL
22961.6 ,0.0 ,0.0 ,23061.6 ,0.0 ,0.0
** DEFINITION OF A CYLINDRICAL LOCAL COORDINATE SYSTEM WITH THE COORDINATES OPTION
1,90.0
** ADDITIONAL ROTATION OF 90° AROUND LOCAL 1-AXIS (RADIAL)
** FOLLOWED BY A CYCLIC PERMUTATION OF THE AXES AND A PROJECTION OF THE AXIS FOLLOWING
** THE AXIS FOR THE ADDITIONAL ROTATION ONTO THE SURFACE OF THE ELEMENT TO FORM THE LOCAL
** MATERIAL 1-DIRECTION
Page 60
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Abaqus)
• Composite part typical orientation problem
• Example : Skin orientation and stringer orientation
1
Initial local coordinate system
3 (cylindrical)
2
3
2 After additional rotation of 90°around
1
1-axis
1 After cyclic permutation of axes leading
to the projection of 2-axis to form the
3
material 1-direction
Page 61
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Abaqus)
• Skin/stringer orientation
• a cylindrical coordinate system can be used to define fuselage skin
and stringers orientation
• Using the 2-axis (orbital) of this system as the additional rotation axis will
imply the projection of 3-axis (axial) to act as material system 1-axis (0°
ply fiber direction)
• Validation of 0°fiber direction
2-axis
3-axis
• Validation of 45°fiber direction
2-axis
3-axis
Page 62
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Abaqus)
• Frame/clip orientation
• The same cylindrical coordinate system can still be used to define
frames and clips orientation.
• Using the 1-axis (radial) system as the additional rotation axis will imply
the projection of 2-axis (orbital) to act as material system 1-axis (0°ply
fiber direction)
• Validation of 0°fiber direction
Page 63
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Material modelling (Abaqus)
• Stabilizer orientation
• One coordinate system is not enough to define all the stabilizer
required orientations due to the 3 perpendicular planes
• The user has to define two systems to fully define these orientations
Page 64
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
DFEM Connection to the Global FEM
(mailto:[email protected])
Page 65
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
DFEM Connection to the Global FEM (Nastran)
• For loading and Boundary conditions reasons, it may be required to
integrate a DFEM in the Global Fem (especially for static linear analysis
where a reduction of the model is not needed).
• Actually, whatever the technique used :
• the integration of a Nastran DFEM is suitable at least for checking
(behaviour at the boundary for example)
• a clear definition of the set of boundary nodes is mandatory.
Global FEM
DFEM to be studied
Page 66
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
DFEM Connection to the Global FEM (Nastran)
• Set of boundary nodes
Take care of the following points:
1. Choose the DFEM model boundary far from the studied area :
The size of this area is of the order of magnitude of the GFEM element
width at the boundary, which leads to
~ 3 stringer pitches in the circumferential direction
~1 frame pitch in the longitudinal direction
2. The boundary nodes should not be constrained
• No constraint node
• Boundary nodes should not be dependent in kinematic or distributing
couplings
s M M
s M s
M s
s s M M : master node
s : slave node
s
s M s M M : dof external relation
M s : dof internal relation
s s : dof forbidden relations
Page 67
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
DFEM Connection to the Global FEM (Nastran)
• Shell mesh sewing technique
For DFEM/GFEM connection purpose, the mesh sewing method is recommended.
DFEM GFEM
Slave Master
nodes nodes
Connection area
Disturbed zone
Page 68
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
DFEM Connection to the Global FEM (Nastran)
• Shell mesh sewing technique
Use of RBE3/MPC distributing coupling along the 6 DOFs (barycentric equations,
the coefficients of the kinematic relations take into account the geometrical positions
of the slave nodes with respect to the master nodes)
Each mesh sewing node must tend to be perfectly aligned with GFEM nodes
If the tolerance is too high, the quality of the rigid modes will be altered
These relations are explicit relations: if some nodes have to be moved, the
relations must be updated
Check that the slave nodes of the kinematic relation are not involved in another
relation (RBE2 or MPC for example) and, if this is not the case, check the
sequencing of relations
Page 69
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
DFEM Connection to the Global FEM (General)
• Shell mesh sewing recommendations
• An area of altered stress distribution in the DFEM can be present at the mesh
sewing level
• The size of this area is ~ the same order of magnitude of the GFEM element width
• Put the mesh sewing border away from the area of interest, at least at a distance
equal to a GFEM element width
• Split the coarse elements into smaller ones to ensure a smooth transition between
GFEM and DFEM and a better stress distribution in the GFEM in the mesh sewing
area.
• Classical mesh density of level 1 Model leads to ~ 10 ratio between DFEM and GFEM
• This ratio can be considered as the maximum order of magnitude
Page 70
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
DFEM Connection to the Global FEM (Nastran)
• DFEM/GFEM stiffeners transition
• Generally the stiffeners are modelled with beam elements in the GFEM
• In a DFEM they are very often modelled with shell elements
• At the DFEM/GFEM boundary a transition technique has to performed
between shell and beam elements in order to ensure a correct load transfer
• The recommendation is to add rigid body element to connect the GFEM node
to the corresponding DFEM nodes
Example of rigid element that
connect 1D gfem V stringer to
the 2D dfem V stringer
Page 71
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
DFEM Connection to the Global FEM (Nastran)
• DFEM/GFEM stiffeners transition
• In some particular cases, as for Frame connection, the rigid body elements can
be replaced by so called “stiff beam” elements that aim at transferring correctly
bending and shear loads
1D Beam element with high
stiffness values for inertia in 1st
bending plane and shear
Usual values (may be case-by-
case dependent)
109 mm4 for bending inertia
If S=1mm², Ks=106
• Note that rigid body elements have a bad behaviour for thermo elastic load cases and
add some induced torque.
• Overstress caused by these elements should be dissipated rapidly along the refined
mesh (see section about ‘0-stress’ validation)
Page 72
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
DFEM Connection to the Global FEM (Abaqus)
A GFEM and DFEM at the same reference (no gap)
Creation of Truss Elements (T3D2) between GFEM boundary nodes.
Use of *TIE between the DFEM boundary (Slave Surface) and Truss Elements (Master Surface) without
adjusting the slave surface (ADJUST=NO).
Use of *Kinematic Coupling to connect the GFEM 1D stiffener to the DFEM 2D stiffener.
Slave (DFEM)
Master (GFEM Truss )
*KINEMATIC
COUPLING
*TIE
T3D2 (Master Surface of TIE) used to connect
DFEM to GFEM
T3D2
Page 73
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
DFEM Connection to the Global FEM (Abaqus)
B GFEM and DFEM NOT at the same reference (gap)
Projection of GFEM boundary nodes on the DFEM skin
Creation of Truss Elements (T3D2) between these nodes.
For each node, use of equations (*EQUATION) between GFEM nodes and projection on DFEM plane,
on each DOF.
Use of *TIE between the DFEM boundary (Slave Surface) and Truss Elements (Master Surface)
without adjusting the slave surface (ADJUST=NO).
Use of *Kinematic Coupling to connect the GFEM 1D stiffener to the DFEM 2D stiffener.
If the gap between GFEM and DFEM stiffeners become too inconsistent, create an area of transition
on the GFEM beam/rod properties
Page 74
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
DFEM Connection to the Global FEM (Abaqus)
Slave (DFEM)
Master (GFEM Truss) *KINEMATIC
COUPLING
*TIE
T3D2
2
*EQUATION
ui1 = ui2
i = dof 1,…,6
Trusses T3D2 (Master Surface of TIE)
used to connect
DFEM to GFEM 1
On Dof 1 1 *Equation for each node pair and each DoF
(5 nodes × 6 DoF =30 *Equation cards for the
example below)
Page 75
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Loading
(mailto:[email protected])
Page 76
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Loading
• Loading a DFEM needs particular care when the objective is to apply
loads coming from GFEM load cases
• The 3 main load case types the have to be considered are:
• Mechanical load case
• Thermal load case
• Pressure load (cabin pressurization, fuel pressure …)
Page 77
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Loading
• Mechanical load case
• When applying a GFEM mechanical load case to a DFEM, it is
advised to distribute the loading initially defined at the GFEM nodes
on the corresponding areas in the DFEM
• Example :
Fr. i-1 Fr. i Fr. i+1 Fr. i-1 Fr. i Fr. i+1
- Use the GFEM node as the slave node of the
Loading defined at each frame
RBE3/Distributing
and 2 stringer stations (ex)
- Distribute the loading on the nodes in the area
of influence of the GFEM loading
- Do not move or renumber GFEM nodes
Page 78
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Loading
• Thermal loading
• Prior to start the process of mapping the temperature field from the
GFEM grid to a more refined grid (DFEM), the user has to analyse the
temperature gradient observed in the GFEM in the area of the DFEM.
• If this gradient is important, it indicates that the assumptions made by
the thermal mapping from GFEM to DFEM – linear interpolation - will
possibly be wrong.
• In this case, it is fully recommended to get in contact with ESAZT
(Thermal Technical Centre) for advice on how to load the structure in
this area.
• The next slides show the steps to follow in order to map a
temperature field from a GFEM grid to a DFEM grid.
• A Patran PCL tool has been developed to automate
this process (Airbus Menu)
Page 79
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Loading
• Thermal Mapping (GFEM->DFEM)
• Thermal load cases have to be rebuilt for the detailed mesh loading
• The method consists in plotting in Patran the original thermal mapping on the
GFEM. Then a scalar field is created with the thermal values and applied on the
detailed model.
Thermal field on GFEM Thermal field applied on DFEM
• It is necessary to have in the Patran data base of both Global and Detailed meshes
• For complex assembly, it is mandatory to apply the method part by part in order to
avoid extrapolation issues.
• Next slides present the steps for the DFEM thermal mapping
Page 80
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Loading
• Thermal Mapping (GFEM->DFEM)
• First step : plot the temperature field on the GFEM mesh
• Second step : create a spatial field from the fringe contours obtained
1
2
GFEM
Page 81
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Loading
• Thermal Mapping (GFEM->DFEM)
• Third step : post the DFEM group, and create the new temperature load case
from the previous spatial field
DFEM
Page 82
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Loading
• Thermal Mapping (GFEM->DFEM)
• Last step : plot the temperature load case created on the DFEM and check
the consistency with with the GEM original load case4
Page 83
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Loading
• Pressure Load Case
• Pressure load cases have to be rebuilt for the detailed mesh loading
purpose
• First remove in the DFEM area the forces coming from the GFEM
• Apply the pressure load on the DFEM
• Be careful at the boundary with the GFEM as the loads of the boundary
nodes have to be updated
• Load balance (closed pressurized envelope) needs to be checked
• Total applied load = 0
Page 84
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Loading
• Pressure Load Case
Initially the GFEM Pressure force calculated at the border
point O is :
1 1 1 1
FO = PAG + PBG + PCG + PDG
4 4 4 4
External area
AG BG GFEM: black elements
O
FEM Border
EDF FDF
Internal CG DG DFEM :red elements
Studied area
When the model is refined, then the new pressure
force at border point O is:
1 1 1 1
FO = PAG + PBG + PEDF + PFDF
4 4 4 4
Then It is necessary to take into account compensatory forces
1 1
FCF = −( PCG + PDG )
4 4
Page 85
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Loading
• Pressure Load Case
• As an example, the final pressure load on the detailed mesh corresponds to:
Pressure forces applied on
Pressures or Forces applied on the - Pressure forces applied on the + the global mesh at the
detailed mesh global mesh at boundary nodes external side of the studied
area
Page 86
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Reduction Techniques
(mailto:[email protected])
Page 87
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Reduction Techniques
• Sub-structuring approach (often called “Condensation” in Nastran)
• Appropriate when the assumption of a linear static behaviour of
the surrounding Global model is valid (constant stiffness matrix)
• Appropriate for Fuselage applications (see below)
• Advantages
• Provides linear aircraft-like boundary and loading conditions to a detailed
model
• Computational efficiency (reduction of the number of d.o.f of the
surrounding model) if the number of boundary nodes is limited
• Can be fully validated in the Nastran GFEM environment
• Avoids the translation of the GFEM from Nastran to Abaqus, which is more
problematic for the fuselage GFEM than for the wing GFEM
• Limitations
• Linear static assumption for the surrounding Global model (geometry,
material)
• The non-linearity must not occur too close from the boundaries
Page 88
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Reduction Techniques
• Nastran GFEM Condensation
• The condensation of Nastran GFEM (stiffness + external load), when
appropriate, is a convenient way of decreasing the size of a problem
by reducing the size of the model for Buckling and NL analysis
• It can be fully validated within the Nastran GFEM environment
• It can be used with Abaqus, by converting the stiffness matrix and the
external load vector to Abaqus format
• This technique requires a particular care, especially when:
• some boundary nodes in the GFEM are some degrees of freedom
involved in couplings
• some external loads are applied to some boundary nodes
Page 89
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Reduction Techniques
• Nastran GFEM Condensation
Condensation method
This aim of this method is to reduce stiffness and loads to a boundary set of nodes.
This method is divided into 3 steps :
1. Stiffness and loads condensation
2. Condensed loads transformation
3. Used of condensed stiffness and loads on reduced model
The following picture shows the reduced model we want to extract from a global FEM
Reduced local FEM to be studied
Page 90
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Reduction Techniques
• Nastran GFEM Condensation
First step: stiffness and loads condensation
First, we remove the reduced model we want to analyse , we obtain the following FEM model :
Spc
123456
Spc
123456
Spc
123456
Spc
123456
Spc
123456
ASET Nodes
This model will be the model to condense ( GFEM model without the reduced local model )
ASET nodes are the condensation nodes, the condensed model will include the local model frontier and in this example,
It will include boundary conditions (constraints + loading )
Page 91
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Reduction Techniques
• Nastran GFEM Condensation
SOL 101 A typical nastran file for condensation
TIME 100
CEND
ECHO = NONE
$ ----------------------------------------------------- DMIG creation activation
PARAM,EXTOUT,DMIGPCH reduced stiffness matrix and reduced loading are output on the punch file in DMIG
$ -----------------------------------------------------
SUBCASE 1
format. Punch Format mandatory.
$ Subcase name : Default
SUBTITLE=Default
SPC = 2
LOAD = 2 Condensation nodes ( frontier nodes)
BEGIN BULK definition of the condensation grid points (interface nodes) between local
$ ----------------------------------------------------- model and the rest of complete model.
ASET1,123456, 6,17,28,39,50
ASET1,123456,61,72,83,94,105 The reduced stiffness matrix and loading of the rest of complete model will
$ ----------------------------------------------------- be performed at theses interface nodes.
PARAM,POST,-1
PARAM,GRDPNT,0
PARAM,WTMASS,0.001
Note: This condensation must be perfomed only with the rest of the
PARAM,AUTOSPC,NO complete model, without local model !
PARAM,SNORM,20.
PARAM,K6ROT,100.
PARAM,PRTMAXIM,YES
$
$
Page 92
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Reduction Techniques
• Nastran GFEM Condensation
Second step : condensed loads transformation
•Reduced loads at boundary nodes are stored in DMIG punch file in PAX format.
•PAX format can be used as it is for Nastran usage. But this format is not very convenient when the user
has to manage several load cases, when he/she wants to keep the link with the initial LOAD card load
factors, or when he/she wants to combine different load cases (thermal, pressure, mech), furthermore at
different load factors (LL, UL, fatigue loads, etc …)
•Some tools allow to transform PAX loads to usual FORCE and MOMENT Nastran Cards (ex: Airbus
France PSN41 tool)
Condensed load in
Condensed load in DMIG Dedicated Nastran format files
file on PAX format tool (condensed-load.dat)
Page 93
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Reduction Techniques
• Nastran GFEM Condensation
Third step : use of condensed stiffness
and loads
Boundary nodes
GFEM Boundary nodes have to be imported in used for
condensation
the DFEM (see picture beside)
Some couplings have to be defined between
DFEM nodes and GFEM nodes, in order to
ensure a satisfying continuity in displacements
and internal loads across the boundary DFEM Nodes
In Nastran, the techniques used to define these Couplings
couplings have been described in the section
DFEM Integration in the Global FEM
For Abaqus usage, similar techniques can be
used
Kinematic couplings for stiffeners
connections
Equations for shell mesh connections
Page 94
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Reduction Techniques
• Nastran GFEM Condensation
A typical nastran file for condensation re-use
In this Nastran Data file
•Interface Nodes-aset nodes- must have same coordinate system than
global FEM
•Reduced stiffness matrix is activated by K2GG command - Reduced
loading is activated by LOAD=1,
• Reduced stiffness matrix is given by include ‘dnastran_condens.pch’
and loading data by include ‘forcepax.dat’
•No constraint data is given. The local model is constrained by reduced
stiffness matrix .
If constraints and/or external forces are applied on local model, they have
to be added in the local model nastran file(SPC1 data, FORCE data,
MOMENT data, etc..)
Page 95
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Reduction Techniques
• Nastran Condensation ABAQUS
• The condensation method can also be used for Abaqus usage
• The Nastran condensation (stiffness matrix + load vector) is easier
to convert from Nastran to Abaqus than the GFEM itself at the
fuselage level.
• This method is not adapted to wing structure which exhibits a more
pronounced non-linear geometrical behaviour than the fuselage. In
the case of the wing, the best approach is to modify and refine the
GFEM principles to enable the model to calculate more accurately
the NL deflection of the wing, and to make use of the sub-modelling
approach to analyse in more details lower levels.
Page 96
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Reduction Techniques
Definition of the critical area to be
ANALYSIS OBJECTIVES
Nastran Condensation ABAQUS modelled, class of problem to be
solved, type of NL expected, …
Boundary with Creation of Abaqus
Global FEM detailed FEM
Creation of the Global FEM part Nastran input file Abaqus input file
Abaqus input file
to be condensed Model_ext.dat Model.inp
Boundary.inp
ASET cards
Definition of the boundary (set *NSET
of nodes)
Nastran execution
Static Condensation PARAM, EXTOUT, DMIGPCH
Execution Nastran punch file
Stiffness matrix (KAAX)
Applied loads vector (PAX) Connection with Global
Analysis of the Nastran input file FEM boundary nodes
residual Global Static Condensation K2GG=KAAX
P2G=PAX Abaqus input file
FEM part + Verification Include ‘Model_ext.pch’ Boundary.inp
condensation
Processing of Nastran results
*EQUATION
Comparison of analysis results with full
NOK Global FEM (must be identical)
*COUPLING
Abaqus input file (*KINEMATIC
Condens.inp or *DISTRIBUTING)
OK Nastran to Abaqus
Translation *MATRIX
Abaqus fromnastran *CLOAD
KAAX
PAX to be first converted to
Nastran FORCE/MOMENT
format
Creation of Abaqus
final input deck
Page 97
ANALYSIS
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Reduction Techniques
• Sub-modelling (ABAQUS)
• 2 separate models are required
• a detailed representation of a region of a structure (submodel)
• a relatively less detailed representation of the structure considered
(global model)
• The global model is run first. It can be linear or non-linear, but
has to be representative of the stiffness response of the global
structure and of the region to be analysed with the submodel
• In a second phase, the submodel is run, using at each
increment the displacements of the global model.
• At the boundary, there is no need for nodes coincidence
between the submodel and the global model (spatial
interpolation)
Page 98
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Reduction Techniques
• Sub-modelling (ABAQUS)
• Advantages
• Computational efficiency : often more efficient than running a single analysis
containing sufficient detail to capture all behaviour of interest.
• Easier to perform, use the same requirement as for the DFEM integration
boundary node selection, i.e. nodes without constraints
• Limitations
• The region of interest must have the same stiffness in the global and the
submodel (prescribed displacements condition), which means that the
boundary must be chosen far enough from the area where the response is
expected to be refined
• If high differences in the geometric stiffness of the submodel compared to
the global model an investigation needs to be performed at the sub-model
boundary in order to validate the technique.
• Ex : buckling occurring in the submodel which does not occur in the global model
• Changes in the contact behaviour in the submodel compared to the global
model that may impact the global stiffness of the submodel
• Changes in the material response in the submodel compared to the global
model (occurrence of plastic strains for example)
Page 99
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation
(mailto:[email protected])
Page 100
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation
• The aim of this chapter is to describe the main steps necessary to check a model
• The main steps are :
• Parts verification
• Parts assembly
• Integration DFEM/GFEM
• Boundary and loading conditions
• Coupling elements (kinematic, distributing)
• Analysis checks
• Linear analysis with few load cases (mechanical, pressure, thermal).
• Modal analysis to detect mainly BC’s , kinetic / rigid elements, stiffness ratio , connection issues.
• Post-processing
• Use « DFEM Validation Check List » , Ref. X51FM1001022 Issue 2.0,
in order to check A350 Detailed FE models
Page 101
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation
• Parts verification
• Mesh
• Consistency
• Free edges
• Duplicate elements
• Normal direction
• Quality
• following AM2036 criteria
• aspect ratio, warping, skew, min and max angles
• Material
• Orientation
• Properties
• Shell properties
• Thickness/offset values
• Modal analysis
• Check of rigid and first flexible modes
• Deviation with geometry
• Mass calculation
Page 102
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation - Mesh checks (general)
• Mesh checks
• Free edge/face : to confirm there are no voids in the model
• Shell elements
• Normal direction : to confirm consistency of element orientations and
connectivity
• Beam : bending planes and offsets : for element consistency
• Identification of duplicate grids and elements
• Identification of all coordinate systems
• Dependant/Independent node conflicts for rigid elements
Page 103
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation - 2D element quality checks
(Nastran)
g Shapes recommendations :
Aspect ratio ≤ 4
– PATRAN: Finite element, verify, quad/tria, aspect
CQUAD Nastran Skew ≥ 30°, CTRIA Nastran Skew ≥ 10°
or CQUAD Patran Skew ≤ 60°, CTRIA Patran Skew ≤ 76°
– PATRAN: Finite element, verify, quad/tria, skew
CQUAD Patran Taper ≤ 0.5
– PATRAN: Finite element, verify, quad/tria, taper
CQUAD Patran Warping ≤ 5.7°
– PATRAN: Finite element, verify, quad/tria, warping
l
S3
h2 h
α S2
h1 S4
S1
4. min( Surfacetria ) h
Aspect =
h1 Taper = Warp = A sin
h2 Skew = 90 − α α ∑ Surfacetria
h2 l
4 tria
h1
Page 104
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation - 3D element quality checks
(Nastran)
g Shapes Criteria :
Aspect Edge Angle Face Skew Face Taper Twist
α
S3 S2
b θ
α S4 S1
a
Aspect =
a Edge angle = α − 70.53° Face Skew = 90° − α
b
θ S3 S2
S4 S1
αi θ
c
Twist = θ
α 4. min( Surfacetria )
b
αi Taper =
a ∑ Surfacetria Warp
Edge angle = Face Skew = Max (90° − α i )
a a 4 tria
Aspect = ou
b c Max (60° − θ , 90° − α )
Collapse
hi θ
α h
l
b li
a On each quad face of
α Min(hi ) WEDGE / HEXA:
a Collapse =
Aspect = Edge angle = Max (90° − α ) Face Skew = 90° − α Max(li ) h
b Face Warp θ = sin −1
l
Page 105
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation - 3D element quality checks
(Nastran)
g Shapes recommendations :
Aspect ratio ≤ 100 (TETRA, HEXA and WEDGE)
Patran/Nastran Edge Angle for TETRA, WEDGE & HEXA ≥ 30°
Patran/Nastran Face Skew Angle for TETRA, WEDGE & HEXA ≥ 30°
Patran Collapse Ratio for TETRA ≥ 0.01,
Patran Face Warp Cosine for WEDGE, HEXA ≥ 0.71
Patran Twist Angle for WEDGE, HEXA ≥ 45°
Patran/ Nastran Face Taper for WEDGE, HEXA ≤ 0.5
Page 106
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation - 2D & 3D element quality checks
(Abaqus)
g 2D-Element Shapes recommendations :
ratio ≤ 5.
Aspect
CQUAD Skew ≥ 40°
CQUAD Max interior angle = 140°
CQUAD Min interior angle = 40°
g 3D-Element Shapes recommendations :
TETRA, WEDGE, HEXA Aspect ratio ≤ 10.
5° ≤ TETRA Interior angle ≤ 170°
10° ≤ HEXA, WEDGE Interior angle ≤ 160°
Page 107
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation - analysis checks (general)
• Compulsory analysis checks
• Static checks
• Analysis of significant static cases to determine deformations, path loads
and local stresses
• A particular care is required for the load equilibrium check
• applied load = - reaction load.
• 0 Stress with a thermo-elastic analysis
• To determine that negligible stress are induced by a uniform DT applied to
the model
• with unique isotropic expansion coefficient
• Rigid body checks analysis / Normal modes analysis
• Only rigid body motion will be acceptable, there should be no local
deformation, no constraint.
• using density properties on MAT cards
Page 108
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation - analysis parameters (Nastran)
• Compulsory
• PARAM AUTOSPC NO
• PARAM BAILOUT 0
• PARAM GRDPNT 0
• PARAM WTMASS 1.E-3
• if model is in N, mm, kg
• PARAM K6ROT 100.
• Use with care and with agreements
• PARAM MAXRATIO
Page 109
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation - global stiffness matrix check
(Nastran)
For all verification static runs it is important to check the following terms:
MAX RATIO :
When Nastran assembles the matrix, it calculates for each column the ratio between the smallest term
and the biggest term, eg the MAXRATIO.
A degree of freedom that has no stiffness (mechanism), will lead to a big value of MAXRATIO.
Nastran gives the list of MAXRATIO bigger than 107 and the corresponding nodes.
MAXRATIO’S are due to:
• bad model constraint.
• Rigid body mechanism
• Connected elements that have very different stiffness
EPSILON :
After each calculation, Nastran determines an error criteria, EPSILON. It enables to detect modelling
mistakes
Nastran is solving the following equation
It calculates the solution U, and then gets the calculated KU .
ε represents the gap between F and the numerical KU.
This gap is due numerical discrepancies and matrix mistakes,singularity.
The result is surely wrong if ε > 10Ε−7
F = K ×U
Page 110
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation - zero stress thermal load case
(Nastran)
g Purpose of the check
• To verify that no modeling errors will generates stress in the model
• All thermal expansion coefficient should have the same value (use PSN1 option
DILAT at Airbus-F)
g Input cards :
• EXECUTIVE DECK
SOL 101
• CASE CONTROL
DISPL=ALL
GPFORCE=ALL
TEMP(INIT)=20
TEMP(LOAD)=120
• BULK DATA
TEMPD,20,20.
TEMPD,120,120.
all MAT thermal expansion coefficient with the same value
g Output results
• No significant stress (<1 Mpa) should appear in the area of interest
Page 111
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation - modal analysis (general)
g Modal analysis check
• Compulsory Normal mode check :
• The highest frequencies for rigid body modes, checks 1 and 2, should be as close to
0.0 as possible, but should not be greater than 1/100 of the magnitude of the lowest
predicted flight mechanics frequency for first flexible mode.
• AM2026 ⇒ Values below 0.01 Hz are desirable (1.0E-04 Hz preferred for DFEM);
any values above this level must be discussed between all parties.
g Output results
• The model should obtain a minimum of six rigid body modes. These modes
should have frequencies less than or equal to 1.0E-04 Hz.
• The following modes (7 to 21) should not be local modes.
Page 112
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation - modal analysis (Nastran)
g Purpose of the check
• To verify that the model will act as a rigid body when it is unconstrained.
g INPUT REQUIRED
• EXECUTIVE DECK
SOL 103
• CASE CONTROL
METHOD=10
(without SPC except special cases as 2D normal rotation dof)
• BULK DATA
PARAM WTMASS 0.001 (or agreed value)
EIGRL 10 21
Page 113
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation - synoptic (Nastran)
• A simplified synoptic of the validation process would be :
Beginning
initial checks performed
(Patran, …) Nastran Y Y DFEM
Only 123456 SPC DOF
SOL101 run -> punch MAXRATIO<107
in PCH file + K6ROT,100. SNORM,20.
+ K6ROT,100. SNORM,20. -> listing + AUTOSPC,NO
+ EPZERO,0. SPCGEN 1 N
N
+ AUTOSPC,YES
Verify BC, Y MAXRATIO=1015 - Complete SPC on reference node
Modifying for translation dofs - Suppress R dofs with no stiffness
- Mechanism : Model modifying N
according to punch dmig
Are constrained rotation dof
Y
- Complete SPC on reference node
with no stiffness acceptable
- Suppress R dofs with no stiffness if acceptable
- Possible help of GROUNDCHECK
N
Model modifying
O stress Modal analysis Gravity
+ K6ROT,1. SNORM,20, - 6 rigid modes only
Y SOL103 Y SOL101 - Realistic shapes Y
AUTOSPC,NO with f<0.01 Hz
stress<1MPa + GROUNDCHECK() + K6ROT,1. - Symmetries
same thermal expansion - No local mode
+ K6ROT,1. + SNORM,20. remaining ?
(with PSN1, DILAT) from 7 to 21 modes
N + SNORM,20. 3 Loadcases
TEMP(INIT)=20.
EIGRL,ND=21, NORM=MAX (1 GRAV for each direction)
TEMP(LOAD)=120. N
N DFEM
- Verify RBEi Modifying… validated
Modifying…
- Modifying…
Page 114
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation - Loading and BC’s
• When the DFEM is loaded with load cases coming from GFEM , the following
steps are mandatory :
•The DFEMhas to be integrated into the gfem (specific to Nastran linear dfem) to verify
the behaviour of the boundary between DFEM and GFEM.
• the following checks have to be performed :
•1 Mechanical load case.
• 1 thermal load case.
•1 pressure load case
⇒The results have to be compared to the GFEM results :
⇒displacements gaps must be reported .The recommended maximum gap value for
displacement magnitude is around 1%
⇒There must be a displacements continuity across the boundary between GFEM and
DFEM.
⇒Comparison of Stress flows must be reported. They must be similar in the running area
of the model.
⇒At the boundary , a particular check is requested for a continuous load path between
DFEM and GFEM ( 1D/2D transition) .
⇒The static equilibrium must be still respected.
- Modal analysis (GFEM+DFEM) mandatory for the mesh sewing validation (specific to Nastran
linear dfem)
The same checks must be performed when using condensation method
Page 115
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Model Validation
•The previous checks will ensure good conditioning of the stiffness matrix ie that the
model response will not be affected by analysis anomalies(singularities, mechanism)
•Engineering judgment is necessary to evaluate if the model meets the initial objectives.
•If some modelling techniques used are not documented in these guidelines or AM ,
please consult EDSAZ to check if some tests have been perform.
•If not, tests results could be required ( see test simulation chapter )
Page 116
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Specific Applications
(mailto:)
Page 117
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Specific Applications
• A350XWB Sizing Method for CFRP Circumferential Joint
• Follow methods and processes recommended in Ref. V53RP0917013 issue 2.0
(EDSASM)
Page 118
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Specific Applications
• A350XWB Sizing Process for Fuselage Window Area
• Follow methods and processes recommended in Ref. V53RP0709584 issue 5.0
(EDSASM)
Page 119
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Specific Applications
• Target Sizing process for A350 PAX Door Surrounding Structure
• Follow methods and processes recommended in Ref. V53RP0723483 issue 4.0
(EDSASM)
Page 120
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Post-Processing
(mailto:[email protected])
Page 121
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Post-Processing
• Objectives
• This section aims to provide some guideline in order to extract from
detailed models specific output data in order to be able to analyse the
following typical problems:
• Onset of buckling
• Internal Load redistribution
• Contact interactions
• Tie interface loads
• Bolted joints load extraction
• Open holes
• Solid mesh
• Failure modes that can be addressed and criteria to be used
• Global failure scenario assessment
Page 122
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Post-Processing
• Onset of Buckling – 1/2
The buckling onset of panels is a key parameter for stiffened panel design. This value is obtain or from a linear buckling
analysis (Nastran sol105 or Abaqus *Buckle) or from a non linear analysis (non linear = yes in Abaqus).
Linear analysis
The eigenvalue obtained, multiplied by the load applied, is the buckling onset if the buckling is concentrated in a local
area (inter-frame/inter stringer). If stringer/ribs are concerned by the buckling shape, a general buckling is detected.
Local buckling Global buckling
Non-linear analysis – Local buckling
In that case, no direct value is proposed by the calculation, the idea is to exhibit the non-linearity on the panel skin, close
to the buckling waves. For this purpose, strain on inner and outer skin ply may be extracted from the model, directly
from elements or through additional elements as the strain gauges described in part 2.4.
1st non linearity on
the skin membrane
(mean) B.O level
Gauges
implementation
Page 123
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Post-Processing
• Onset of Buckling – 2/2
Non-linear analysis – Global buckling
In that case, no direct value is proposed by the calculation, the idea is to exhibit the non-linearity on the stringers. For this
purpose, strain on inner and outer may be extracted from the model, directly from elements or through additional
elements as the strain gauges described in part 2.4.
The level of the global
buckling onset is detected
when the stringer membrane
( average at the stiffness
centerline of the stringer) is
facing a slope change
Page 124
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Post-Processing
• Load distribution – 1/2
• The load distribution may be important to understand and improve the local behavior
of an assembly. 2 methods are proposed:
• Use of flows on a path which can be defined on existing or non existing nodes
• Use of “free body” diagram
Path Freebody diagram
Node path
defined in
abaqus
viewer
200
Panel_SF1_0
Panel_SF1_1
One curve
Panel_SF1_2
0
0 100 200 300 400 500 600 700 800 900 1000
Panel_SF1_3
Panel_SF1_4
Panel_SF1_5
by load step
Panel_SF1_6
Panel_SF1_7
-200
Panel_SF1_8
NFORC output is required in the Abaqus
Panel_SF1_9
Panel_SF1_10
-400 Panel_SF1_11
Panel_SF1_12
Panel_SF1_13
-600
Panel_SF1_14
Panel_SF1_15
Panel_SF1_16
Panel_SF1_17
calculation.
-800 Panel_SF1_18
Panel_SF1_19
A section is defined on existing nodes in the
Flow
Panel_SF1_20
Panel_SF1_21
Abaqus viewer + a local section coordinate
-1000 Panel_SF1_22
Panel_SF1_23
Panel_SF1_24
Panel_SF1_25
-1200
Panel_SF1_26
-1400
Panel_SF1_27
Panel_SF1_28
Panel_SF1_29
Panel_SF1_30
system.
-1600
Panel_SF1_31
Panel_SF1_32
Panel_SF1_33
Normal and shear loads + momentum are
calculated at the section location
Chord abscissa
In that case we can see the flow distribution at each
load step and its evolution wrt the buckling onset
Page 125
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Post-Processing
• Load distribution – 2/2
• Abaqus Viewer usage
Path Freebody diagram
Page 126
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Post-Processing
• Contact interactions
Contact interactions are defined in Abaqus non-linear analysis and may be post-processed in order to evaluate
pressure concentration due to the local contact. This feature is directly available in abaqus Viewer by the plot of the
CPRESS parameter (to be declared as an output in the input file). In addition, for a comprehensive understanding of
the contact interaction behavior, the “contact distance“, CDISP, can be plotted. It allows to evaluate the clearance and
really in contact areas along the non-linear calculation.
Page 127
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Post-Processing
• Tie interface loads
• The tie link is an easy way to connect parts without creating any fastener. In some cases, it may
be interested to extract the load at the tie interface in order to detect potential debonding or
fastener failure.
• The idea here is to extract the local nodal forces and to sum them for a tie contact condition. The
use of freebody in Abaqus may be the good way to do it.
Page 128
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Post-Processing
• Bolted joints load extraction
• FASTPPH2 tool can be used on Nastran FE models that fulfill bolted joint
mesh density and element types requirements
• For Abaqus models, a Plug-in has been developed to extract loads and
stress flow data for the ISAMI Bolted Joint Module. It can be used in
Abaqus/Viewer.
• Consult the following documents for more information about Filled Holes
criteria calculation:
• “A350 XWB Method for Sizing of Bolted Joints”, Ref. V020RP0818061 Issue 5.2
• “Preliminary bolted joint method for LSP fasteners applications”
Ref. V029RP0909190 Issue 1.1
Page 129
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Post-Processing
• Open holes
• The post processing method for open holes is based on a Yamada+Puck
plain strength evaluation performed at a d0 distance from the hole. This d0
distance is depending on the material, the lay-up and the diameter of the
hole.
• Consult document «Composite harmonized open holes calculation
method», Ref. RP0405620 Issue 4.0
Page 130
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Post-Processing
• Solid mesh
• The specificity of solid model post-processing is the way to extract
strain/stress on the part. Indeed, by default, Abaqus is extrapolating the
strain/stress calculated at the element integration point using the shape
functions. However, when the mesh is not regular or when plasticity occurs
this extrapolation may not be accurate.
• The proposal is to map a 2D membrane mesh on the free faces of the solid
mesh (see next slide), with a very small thickness and the material of the
part) in order to post-process strain/stress
Page 131
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Post-Processing
• Solid mesh
• Stresses in solid mesh have to be post-processed with 2D marker elements mapped
on the solid mesh free faces
• Skin membrane elements with thickness of 0.001mm
• 1st order element is sufficient to capture skin strain/stresses
• Membrane material only
• Same material as 3D solid mesh.
Example with parabolic Tetra elements :
Skin membrane elements (e=0.001 mm and same material
as the solid mesh)
2nd order TETRA
mesh
Page 132
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Post-Processing
• Failure modes that can be addressed and criteria to be used
• Plain strength criteria (Yamada Sun + Puck)
• Nastran : manual extraction of stress flows and use of ISAMI standalone module.
• Abaqus : a python plug-in is available.
• Unfolding (CORIN method)
• Nastran : manual load extraction for use in ISAMI Corin module
• Abaqus : a python plug-in is available to fill the Excel template to be used by
ISAMI Corin module
• Consult the following documents about Corin Criteria
• “Composite Unfolding Method: Analytical Approach for Stress Analysis and Failure
Criteria”, Ref. RP0907477 issue 1.0
• ISAMI: “CORIN Analysis - Theoretical Method”, Ref. RP0719493 issue 3.0
• Toldom criteria
• Skin
• Nastran : Manual flow extraction and use of ISAMI standalone module.
• Abaqus: Manual flow extraction and use of ISAMI standalone module
• Stiffeners
• Consult document “Edge Impact Damage Tolerance Method for A350” ,
Ref.V57RP1039013 issue 1.0
Page 133
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Post-Processing
• Global failure scenario assessment
• The failure assessment of a part is linked to the comprehensive behaviour
of the area analyzed.
• For a typical study like a stiffened panel, the following “failure criteria”
history can be proposed, even if not necessarily comprehensive (used in
A350 Fuselage Barrel Test Campaign)
xx LL
1 LL
0
Local buckle Global
onset Stringer
buckle
disbonding
Rivet first
failure
Plane strength criteria
reached
Fastener / hole RF=1
Page 134
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
Fatigue and Damage Tolerance Analysis
(mailto:[email protected])
Page 135
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
October 2011
F&DT Analysis aspects
• Scope of the document with respect to F&DT aspects
• Provide guidelines for Detailed FE modelling at component and sub-
component scale in order to determine more accurately stress
concentration factors (Kt) involved in Fatigue analyses.
• Fatigue and Damage Tolerance analysis objectives that can be
achieved at these scales
• Fatigue life assessment of complex parts
• Identification of critical areas in bolted joints areas
• Out of scope :
• F&DT analysis that follows the stress assessment (done with ISAMI or
legacy tools)
• Calculation of Stress Intensity Factor (SIF) based on FE modelling;
associated FE elements and post-processing aspects
Page 136
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
October 2011
F&DT Analysis aspects
• Fatigue life assessment of complex parts
• Ex : complex machined parts where stress concentration factors
are difficult to assess without detailed modelling
• Due to the complexity of such parts and the type of information
investigated, a 3D modelling is required
• Element types
• 2nd order Tetra elements in geometrical complex areas
• No projection of the edge nodes on the curvature of the
geometry
• Sufficient mesh refinement to capture stress gradients
• Use of modified formulation for contact analysis to avoid
unphysical distribution of contact forces
• 1st order Quad or Hexa elements in regular shapes (ex: open holes)
• Instrumentation of 3D elements free faces with marker 2D elements
(membrane) to extract stresses
Page 137
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
F&DT Analysis aspects
• Fatigue life assessment of complex parts
• 2nd order Tetra elements in geometrical complex areas such as Frame
Foot, Cruciform, fitting, etc…
Page 138
Frame radius analysed
Horizontal flange radius analysed
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
F&DT Analysis aspects
• Fatigue life assessment of complex parts
• 1st order Quad or Hexa elements in regular shapes
• Example of open hole
Panel Mesh
Buffer zone
Quad ring for post-treatment
Page 139
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
F&DT Analysis aspects
• Fatigue life assessment of complex parts
• Modelling strategies
• In a first step a coarse mesh can be performed to identify high stress
areas that will have to be further investigated in more details
• If the model size is not too large, it can be envisaged to modify the
model in these areas with adequate element type selection and mesh
refinement.
• The connection with the coarse mesh has to be done sufficiently far from
the areas of interest.
• It can be done either by mesh transition or by tied contact technique.
• The mesh transition technique can be difficult or even impossible in complex
geometrical parts and when different element types are considered
• If the model size is too large, sub-structuring (condensation) or sub-
modelling techniques can be used when appropriate to reduce the size
of the problem to be solved (see the chapter about Model reduction
techniques)
Page 140
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
F&DT Analysis aspects
• Fatigue life assessment of complex parts
• Modelling strategies
• (1) Areas that have to be addressed at global scale
• Contact zones
• (2) Areas that can be addressed at lower scale with a sub-modelling
technique
• Areas apart from contact zones
(2)
(2) (1)
(1) (2)
Page 141
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
F&DT Analysis aspects
• Fatigue life assessment of complex parts
• Modelling strategies
• Example 1: introduction of detailed fastener models in a large-scale model
• FEM level 1: initial model with relatively coarse meshing, rigid or 1D
elements for fasteners
• Objective : model performed to identify and select areas with potential
risks of fatigue damages
• Minimum of 3 elements between 2 consecutive fasteners
• Sufficient mesh refinement to capture stress gradients for complex area
such as radius, system hole, etc…
• Exploitation : loads in rigid fastener elements, overall displacements,
overall stress values in fastener areas, with potential over-stress effect
due to rigid elements. Local stress in fastener areas not exploitable
Page 142
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
F&DT Analysis aspects
• Fatigue life assessment of complex parts
• Modelling strategies
• Example 1 : introduction of detailed fastener models in a large-scale model
FEM1 model FEM2 model FSFT test damages
('blind' numerical analysis with) ('blind' numerical analysis with) (numerical results compared to)
Panel: D1281
Panel: D659
Panel: D662
Panel: D667
Panel: D660
Doubler: D1270
Panel: D661
Panel: D669
Panel: D650 Doubler: D1271, D1272
Panel: D651
Panel: D654
Panel: D649 Panel: D400
Doubler: D1276
• FEM level 2 : detailed model with refined meshing, 3D fasteners with solid
elements (≅2Φ area meshed with solid, between 16 and 64 elements along
perimeter of bores, number of elements in depth being thickness
dependant), and contact elements where needed + bush elements for
fasteners of the surrounding (need to correct the stiffness of 1D elements to
avoid modifications of load transfers due to 1D/3D mixed modelling)
• Objective : FEM2 model performed to process local fatigue calculations
• Exploitation : overall displacement, overall and local stress values in
fastener areas. Loads in fasteners not exploitable
Page 143
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
F&DT Analysis aspects
• Fatigue life assessment of complex parts
• Modelling strategies
• Example 2 : use of the sub-modelling technique to refine local stress
solutions
• The limited size of a sub-model allows the user to refine the mesh
iteratively in order to reach a convergence of the stress gradients with
respect to an objective of accuracy
Global scale
Sub-models
Page 144
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
F&DT Analysis aspects
• Identification of critical areas in bolted joints areas
• In bolted joint areas involving thin-shell like structures, it is possible to
make use of shell element modelling and spring-type fasteners to
identify the most critical areas (bearing loads, by-pass stress flows)
• The mesh recommendation made in Chapter “Mesh density” in bolted
joint areas can be applied.
• The use of mesh-independent fasteners is preferred as the load
transfer is not punctual and the mesh quality is better (see Chapter
“Element type, Fasteners”) Multi criteria DFEM
MNT Static (composite, metallic) and Fatigue
Y Recommendation for shell mesh in the plates and
X For fastener element connection technique : 4 rigid links,
Z
Page 145 set Krz = 0 or negligible
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
F&DT Analysis aspects
• Identification of critical areas in bolted joints areas
• With bolts modelled with 1D elements, at the global scale, it will
rarely be possible to determine sufficiently accurate local data for
F&DT analysis
• For this specific need, a more detailed FE approach is often needed
(full 3D for fasteners at least)
• 2 cases :
• 1) a 3D approach is decided for a number of critical bolts inside a joint
(global scale)
• 2) a 3D approach is decided for a all bolts inside a joint (detail scale)
Page 146
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
F&DT Analysis aspects
• Identification of critical areas in bolted joints areas
– Case 1 : 3D approach for a number of critical bolts inside a joint (global scale)
• Ensure consistent stiffness between 1D and 3D fasteners
K1D = SAFE-cfrp * %CORREC
1D 3D
Calculate and apply corrective factors on 1D fastener
stiffness law to avoid modification of load transfers
due to 1D/3D mixed modelling
3D 1D
Page 147
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
F&DT Analysis aspects
• Identification of critical areas in bolted joints areas
• Case 2 : 3D approach for a all bolts inside a joint (detail scale)
• The modelling strategy to address this lower scale can be based on
• Application of M,N,T loads (GFEM or intermediate DFEM)
• Substructuring (condensation)
• Sub-modelling conditions
• Ex: investigation of local stresses at bores, load transfer not directly available
(use free-body diagram or contact analysis)
Page 148
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
F&DT Analysis aspects
Validation: the DFEM should be validated by stress expert :
• robustness to boundary conditions (ref: DFEM integration GFEM)
• robustness to Meshing refinement (ref: mesh density)
Page 149
General Detailed FEM Modelling for Linear and Non Linear Analysis October 2011
© AIRBUS S.A.S. All rights reserved. Confidential and proprietary document. This document and all information contained herein is the sole property of AIRBUS S.A.S. No intellectual property rights are granted by the
delivery of this document or the disclosure of its content. This document shall not be reproduced or disclosed to a third party without the express written consent of AIRBUS S.A.S. This document and its content shall not be
used for any purpose other than that for which it is supplied. The statements made herein do not constitute an offer. They are based on the mentioned assumptions and are expressed in good faith. Where the supporting
grounds for these statements are not shown, AIRBUS S.A.S. will be pleased to explain the basis thereof.
AIRBUS, its logo, A300, A310, A318, A319, A320, A321, A330, A340, A350, A380, A400M are registered trademarks.
Page 150
REFERENCE V53PR0906133_exp
Distribution List
ISSUE 3.0 DATE 23 Jan 2012
Detailed FEM Modelling guidelines for Linear and Non Linear Analysis
Distribution List
ELECTRONIC DISTRIBUTION LIST
PARREAUX-EY FRANCOIS [KBEX32]
MOREAU DOMINIQUE [ESA]
MAIXANT GUY [KBENA]
LE-DROGO STEPHANE [ESAZN]
LE-MORLEC ALAIN [KBEXA]
JUILLEN LIONEL [KBEXA]
LE-ROUX OLIVIER [ESAZN]
VIDAL SEBASTIEN [ESKAT]
GALLAIS BRUNO [ESAZT]
DELORME PHILIPPE [ESAB]
MASVESY CHRISTEL [ESAB]
GRILLON ROLAND [KBEX42]
ODRY LAURENT [KBEX54]
ORBAN FREDERIC [KBEX22]
FRIARD VINCENT [KBEX52]
TANGHE FREDERIQUE [KBEX12]
CARRONNIER DELPHINE [ETEC]
ESCAUT JACQUES [KBEXA]
CUTULI JEAN-LUC [KBEA]
Lauster Johannes [KBEA]
CAVAILLES BENOIT [KBEXA]
JACQUES EMMANUEL [KBENA]
GUILLET EMMANUEL [TWEYK5]
LEROUX STEPHANE [KBEN11]
BEAUFILS JEAN-YVES [ESAS]
Ostergaard Morten [ESAZN]
Rollema Gerrit-Jan [ESAZAU]
MAHE MICHEL [ESKA]
ROSEMBERG BRUNO [ESASIT]
DE LA GANDARA-VERANO JUAN LUIS [ESAZAM]
Ohrloff Nikolaus [ESAS]
Dean Michael [ESAW]
Lane Ian [ESAS]
Beumler Thomas [ESKA]
LE-DIVENAH LAURENCE [ESASI]
SANTGERMA ALAIN [ESASA]
Wohlers Wolfgang [ESASM]
RISSE LAURENT [ESKAT]
Schieber Wolfgang [ESKAG]
CAMPASSENS DENIS [ESASMT]
MAHDI STEPHANE [ESKN]
Pellenkoft Frederik [ESASOG]
Liu Jun [ESASOU]
Vultaggio Biagio-Claudio [ESAC]
FUALDES CHANTAL [ESKN]
DIDA STEPHANE [TWEYA2]
PETERS PHILIPPE [TWEK9TJ]
Hunter Alastair [TWEK9V]
BRAS PHILIPPE [TBENY3]
© AIRBUS S.A.S. 2012. ALL RIGHTS RESERVED. CONFIDENTIAL AND PROPRIETARY DOCUMENT.
REFERENCE V53PR0906133_exp
Distribution List
ISSUE 3.0 DATE 23 Jan 2012
MANUAL DISTRIBUTION LIST
-------------------------------------------
Winkler Erhard [XRD2]
-------------------------------------------
GOSSET STEPHANE [EYCDC]
-------------------------------------------
CBURLEIGH
© AIRBUS S.A.S. 2012. ALL RIGHTS RESERVED. CONFIDENTIAL AND PROPRIETARY DOCUMENT.
You might also like
- AM 2036 (10 May 2004) - Normes - Fem - Global100% (5)AM 2036 (10 May 2004) - Normes - Fem - Global68 pages
- ANSYS Mechanical APDL Rotordynamic Analysis GuideNo ratings yetANSYS Mechanical APDL Rotordynamic Analysis Guide154 pages
- Impact Analysis of Aircraft Using ANSYS Explicit Dynamics100% (1)Impact Analysis of Aircraft Using ANSYS Explicit Dynamics13 pages
- Dynamic Analysis of Shaft Structure Based On ANSYSNo ratings yetDynamic Analysis of Shaft Structure Based On ANSYS6 pages
- SIMSOLID ThermalAnalysisConceptsGuide 1No ratings yetSIMSOLID ThermalAnalysisConceptsGuide 114 pages
- A Tire Model For Air Vehicle Landing Gear Dynamics100% (1)A Tire Model For Air Vehicle Landing Gear Dynamics12 pages
- Unit 3 - Aircraft Materias and Structures - CompressedNo ratings yetUnit 3 - Aircraft Materias and Structures - Compressed87 pages
- Introduction To The Computer Simulations - BorovinsekNo ratings yetIntroduction To The Computer Simulations - Borovinsek90 pages
- MSC Training Catalogue 2014: Hängpilsgatan 6, SE-426 77 Västra Frölunda, Sweden Tel: +46 (0) 31 7485990No ratings yetMSC Training Catalogue 2014: Hängpilsgatan 6, SE-426 77 Västra Frölunda, Sweden Tel: +46 (0) 31 748599044 pages
- Modal and Linear Static Analyses in Inspire With SimSolid Solver PDFNo ratings yetModal and Linear Static Analyses in Inspire With SimSolid Solver PDF10 pages
- Numerical Investigation of Structural Design of Torsion Links in A Landing Gear Retraction MechanismNo ratings yetNumerical Investigation of Structural Design of Torsion Links in A Landing Gear Retraction Mechanism7 pages
- Finite Element Analysis: GE393 Computer-Aided Design, Analysis and PrototypingNo ratings yetFinite Element Analysis: GE393 Computer-Aided Design, Analysis and Prototyping14 pages
- 3 SimSolid Cantilever Modal Analysis-1 PDFNo ratings yet3 SimSolid Cantilever Modal Analysis-1 PDF8 pages
- Predictive Engineering White Paper On Small Connection Elements-Mpc and Cbush Rev-1No ratings yetPredictive Engineering White Paper On Small Connection Elements-Mpc and Cbush Rev-140 pages
- MD Nastran Demonstration Problems 2010No ratings yetMD Nastran Demonstration Problems 20101,347 pages
- MSC - Nastran 2003 Linear Static Analysis User's GuideNo ratings yetMSC - Nastran 2003 Linear Static Analysis User's Guide792 pages
- Chained Analysis: Fan Blade Out With Rotor Dynamics100% (1)Chained Analysis: Fan Blade Out With Rotor Dynamics21 pages
- Design and Analysis of A Ladder Frame ChassisNo ratings yetDesign and Analysis of A Ladder Frame Chassis10 pages
- Siemens SW Analysis and Simulation in Aircraft Structure Certification WP - tcm27-65939No ratings yetSiemens SW Analysis and Simulation in Aircraft Structure Certification WP - tcm27-659397 pages
- Tensile Strength of Monolithic Advanced Ceramics at Ambient TemperaturesNo ratings yetTensile Strength of Monolithic Advanced Ceramics at Ambient Temperatures18 pages
- THAIOIL - DOUBLE SHEAR CLEAT ANGLE - BOLTED - 005 - Rev ANo ratings yetTHAIOIL - DOUBLE SHEAR CLEAT ANGLE - BOLTED - 005 - Rev A11 pages
- Geotechnical Characteristics of Copper Mine TailingsNo ratings yetGeotechnical Characteristics of Copper Mine Tailings12 pages
- 8.SC8-CVR-SHEAR (H200x100x5.5x8) - PL12-M20-W8-L75X75X6No ratings yet8.SC8-CVR-SHEAR (H200x100x5.5x8) - PL12-M20-W8-L75X75X618 pages
- Chapter 2 - Physical & Mechanical PropertiesNo ratings yetChapter 2 - Physical & Mechanical Properties42 pages
- Lecture 6-Utility of Energy Release Rate-NewNo ratings yetLecture 6-Utility of Energy Release Rate-New23 pages
- W1 - Introduction Design of Steel Structure Pak DaniNo ratings yetW1 - Introduction Design of Steel Structure Pak Dani38 pages
- Mechanical Metallurgy 3 TMP31-1: National DiplomaNo ratings yetMechanical Metallurgy 3 TMP31-1: National Diploma6 pages
- Páginas DesdeB.H.G.brady, E.T.brown - Rock Mechanics-8No ratings yetPáginas DesdeB.H.G.brady, E.T.brown - Rock Mechanics-81 page
