Professional Documents
Culture Documents
1.1. DESCRIPTION
This case is meant to show how perform a fatigue analysis of a seam-welded structure.
The system is show in the following figure.
The pipe (violet) is connected to a heat shield (translucent red) via two brackets (cyan). The
brackets are connected to the pipe with four seam welds (blue). In this rough model the
welds are modelled with shells. Please feel free to modify them as you deem appropriate.
The system is loaded and constrained as shown in the following figure:
2
Figure 2. Case #1. The loads and the constraints.
One pipe’s edge is fully constrained (cyan). The pipe’s temperature is 200 °C (red); the
temperature of the model’s remainder is 0 °C (grey).
As a result, one obtains the following distribution for the Von Mises equivalent stress.
3
Figure 3. Case #1. Von Mises equivalent stress distribution [MPa].
I ask you to please show me how to assess the fatigue safety factor for the pipe near the
welds, assuming an arbitrary temperature time-history.
The seam welds are modelled by rigid regions comprising couples of nodes. This technique
provides the necessary kinematic constraints, while not introducing spurious stiffening of the
structure, which would be present with multi-node rigid regions, for instance.
V.Lab features an automatic detection of welds. This is accomplished by specifying different
materials for the sheets to be welded; by a proximity check the weld models are built up. Of
course, the seam models can be manually generated too.
The fatigue life calculations of the seams themselves can be carried out by means of a library
of load/stress transfer functions for a variety of seam welds. This kind of calculation requires
the sheets to be modelled via 1-st order shells.
V.Lab features a library of welded junctions; for each of them a set of notch factors (on for
each loading condition) and of Wöhler curves is defined. The welded sheets are analyzed
extracting nominal stresses from a shell mesh and multiplying them by the notch factors; the
results are input into the Wöhler curves to obtain the fatigue life or the safety factor. The
output quantities can be graphically analyzed, as shown in the following figure for the safety
factor:
4
1.2.2. FEMFAT
No modelling of the welds themselves is required. The sheets are simply connected by
making them share some nodes. This way nominal stresses are read from the Ansys result
file and then multiplied by notch factors coming from a library of seam weld types. The
results are maps of safety factor, fatigue life, etc.:
5
1.2.3. MSC.Fatigue
The weld are to be modelled with shells which represent the seams’ free surface (typically, at
a 45° angle with the welded plates’ surfaces).
Two special materials must be assigned to the elements adjacent to these shells: one each to
model the membrane and bending behaviour of the welded components. The relative
importance of the two effects is estimated by comparing the top and bottom stresses of the
shells, even if MSC Software has still to ascertain whether this comparison is
automatically done by the code or is to be performed by the user.
The welded sheets must be modelled with 4-noded shells.
The welds themselves can’t be analyzed; i.e., their fatigue life can’t be computed.
6
2. CASE #2 – THE CRANK
2.1. DESCRIPTION
This case is meant to show the gradient effect on a stress distribution with a localized stress
intensification.
The system is show in the following figure.
7
Figure 5. Case #2. The loads and the constraints.
Symmetry constraints (“S”) are applied to both the pin and the crank. One crank’s surface
(blue) is fully constrained. One pin’s surface (green) is fully constrained, except for the z
translational dof, which is imposed a 0.01 mm displacement to.
As a result, one obtains the following distribution for the Von Mises equivalent stress.
8
Figure 6. Case #2. Von Mises equivalent stress distribution [MPa].
I ask you to please show me how the fatigue safety factor changes when the gradient effect
is taken into account.
The gradient effect will be implemented inside the next V.Lab release.
2.2.2. FEMFAT
The gradient effect has been taken into account assuming that the load is pulsating. In the
following figures the safety factor maps area shown, with and without gradient effect.
9
The importance of the gradient effect is apparent.
10
3. CASE #3 – THE BRACKET
3.1. DESCRIPTION
This case is meant to show how to interface the durability code with MSC.ADAMS.
The ADAMS model is show in the following figure.
A flexible bracket (maize) is fixed to a slider (red). The bracket has been imported from
Ansys, whose model is shown in the following figure:
11
Figure 8. Case #3. The FEM model.
One of the bracket’s (cyan) surfaces is stiffened by extra-stiff beams (red), which connect all
the surface’s nodes. Two of them (blue) are used as interface nodes inside ADAMS.
The MBS system is loaded and constrained as shown in the following figure:
12
The bracket is fixed to the slider at its interface nodes (cyan locks). The slider is attached to
the ground by means of a translational joint (cyan). The slider’s unique dof is controlled by a
motion constraint (cyan arrow), whose time (t) expression has been parametrized as follows:
t 10
A 10 sin(2ft) if t
f
uz f (0)
10
Asin(2ft) if t
f
A motion amplitude
f motion frequency
Clamping the interface nodes, the bracket’s first natural frequency is 627 Hz, with the mode
shape shown in the following figure:
Figure 10. Case #3. The clamped bracket’s first mode shape.
So the motion constraint in the MBS model is able to excite the bracket’s first natural
frequency, given an appropriate value for f (see eq. 0).
13
I performed two analyses, named freq100 and freq300 respectively. The chosen parameters
are given in the following table.
Then I monitored the sensor displacement against the slider’s one, identified by the Adams
name slider_XFORM.X. The results of the analyses are shown in the following figure:
.
14
Figure 12. Case #3. The sensor’s displacement time history.
Modal stress are imported via the .mnf file written by ANSYS, while the modal coordinates
time-histories are imported from ADAMS, either in RPC3 or ASCII format (ADAMS/Durability
not needed). The modal coordinates time-histories are associated with the modal stress
distributions, written in the .mnf file, just like the load time-histories are associated with the
stress distributions in the static case.
The dynamic effect is shown in the following figures, depicting the equivalent stress maps
both at 100 and 300 Hz (see paragraph 3.1).
15
16
3.2.2. FEMFAT
The technology implemented in FEMFAT is the same as V.Lab’s (linear combination of the
modal stress vectors, using as coefficients the modal coordinates time-histories).
The dynamic effect is shown in the following figures, depicting the safety factor maps both at
100 and 300 Hz (see paragraph 3.1).
17
18
3.2.3. MSC.Fatigue
The stress tensor time-history is obtained by a linear combination of the modal stress
vectors, coming from the FEM package. The combination’s coefficients are the modal
coordinates, computed by the MBS code.
Within the stress-life method, the equivalent to be used (max princ, signed Von Mises, etc.)
can be chosen.
The safety factor vs. endurance limit can be computed.
Surface conditions and treatments can be taken into account.
Multiple load channels can be considered.
The software is modular (durability, motion, acoustics). Each module features a tree, object-
oriented structure, similar to the one implemented inside ANSYS Workbench. Trees can be
converted into templates, which can be used as guidelines for future analyses. (ANSYS
structural result file).
Just as in LMS/FALANCS, the ANSYS results are processed importing the .rst file (ANSYS
structural result file); however stress experimental data can be used instead.
V.Lab is associative with Excel, whose format can be used to specify the load time-histories,
which can be associated to the stress sets by user-defined name matching. Load time-
histories can be imported from ADAMS; V.Lab can then set up an ANSYS input deck in order
to calculate the stress distribution caused by the loads; these stresses can be imported back
to V.Lab, which finally computes the fatigue life. In the presence of flexible bodies inside the
ADAMS model, the modal coordinates time-histories are imported into V.Lab, instead of the
load time-histories. The modal coordinates information is combined with the modal stresses
V.Lab reads from the .mnf file generated by ANSYS.
First-estimate load cycle counting parameters can be issued to analyze the whole structure;
V.Lab evidences the critical areas, which can be further analyzed with more stringent
counting parameters.
V.Lab features a colour-coded subdivision of the structure into 3 parts: the one where the
stress level goes beyond the static ultimate value, the one with stress level under the
endurance limit and the one with finite fatigue life.
In order to specify non-homogeneous data, hybrid systems of units can be set.
4.3. FEMFAT
This software has been specifically developed to be interfaces with FE packages. It’s based
upon the influence factor method, which consists in building local SN curves at each node of
the mesh (see the relevant German TGL for details).
FEMFAT is modular. To perform the benchmark’s analyses, the following modules are needed:
The analyses are guided by trees, whose elements can be exploded by the uses to input the
necessary data (FE information, load time-histories, material properties, etc.).
19
4.4. MSC.FATIGUE
export all the modal coordinates’ time-histories simply specifying the flexible body’s
name;
visualize the stress distribution inside ADAMS/Post-processor;
coupled with ADAMS/AutoFlex, compute the stresses inside rigid bodies under the load
conditions evaluated with the MBS simulation.
5. PRICES
5.2. FEMFAT
Price
Annual lease, Pay up
Module
maintenance
Software Annual maintenance
included
BASIC 11085 22170 3325.5
PLAST 1820 3640 546
MAX 3945 7890 1183.5
WELD 3870 7730 1159.5
VISUALIZER 1450 2900 435
44330 6650
total 22170
50980
5.3. MSC.FATIGUE
€ 30000
20