Sie sind auf Seite 1von 5

2004-32-0085 / 20044372

Combining FEM-Optimization and Durability Analysis


to Reach Lower Levels of Component Weight
Klaus Puchner, Christian Gaier and Helmut Dannbauer
MAGNA STEYR, Engineering Center Steyr, Austria

Copyright © 2004 SAE International and Copyright © 2004 Society of Automotive Engineers of Japan, Inc.

ABSTRACT INTRODUCTION

Utilizing the Finite Element Method (FEM), two The reduction of vehicle weight is, at least
evolving CAE methods have recently reached a since the drastic raise of fuel prices in the
high level of efficiency and accuracy to optimize seventies, a basic requirement for the
the properties of components in respect of development of any new component or whole
stiffness, stress level or fatigue resistance. vehicle. Nevertheless there is a general
tendency in the opposite direction. Reasons
The first type of CAE-methods is a family of can be found in additional features for more
optimization methods known as (parameter-free) comfort and safety as well as increased engine
shape and topology optimization. Very torque.
remarkable and useful results can be generated
with these methods to reach considerable In order to safe weight, the use of high strength
stiffness increases or stress and weight materials is a solution to realize lightweight
reductions. structures. For components undergoing cyclic
loading, one has to consider that the effect on
The second CAE-method is fatigue life fatigue life by stress concentrations in notches
prediction which gives reasonable accurate generally increases with the use of high
outputs for component life if input data like stress strength materials. Vehicles have to withstand
history and material properties are well known. different types of loading requirements
State of the art algorithms and software can including effects from notches, weld seams and
handle a combination of complex load histories, spot joints. How can engineers handle these
detailed material description and large FEM- multiple goals during the development
data to give reliable results in short time. process?

It will be demonstrated, using test cases and real Generally there are two methods available for
world components, that combining these two this optimization process: “test” and “numerical
CAE-methods can significantly increase the simulation”.
optimization level in respect to lifetime or weight. By performing tests for complete vehicles or
In the case of combining topology optimization components, many influences can be
with fatigue life prediction, which is a rather new considered without any mathematical model.
idea, it can be detected that both material type Unfortunately, experimental tests give only poor
(brittle / ductile) and loading history often have a information about over-designed areas and are
decisive influence on the resulting design of expensive and time consuming.
components. Compared to common usage of
FEM-based optimization, higher savings of Numerical simulation and optimization can help
weight can be obtained by including fatigue life to achieve a quicker engineering process and
prediction. is the only way to get higher levels of
performance at a lower weight. The following
investigations focus mainly on optimization of up” by the stress history in a location.
components in respect to fatigue, but multi- For even higher cycle numbers and low load
disciplinary optimization is about to be used levels accumulated damage gets extremely
soon. These other disciplines can be NVH and small for many materials so it is practical again
crash-safety for instance. to work with a safety factor. It defines if a crack
is initiated for a infinite number of load cycles.
FATIGUE LIFE PREDICTION In this region notches and the fatigue limit of the
The structural stress distribution from a finite material are most important, ductility is nearly
element analysis, the loading of the part, e.g. irrelevant.
from a Multi Body Simulation or measurements From these simple considerations it clear that a
of a test drive as well as the material properties component will have different critical locations
are the basic input data for fatigue life for different load levels or – vice versa – a
prediction. component has a different optimal shape for a
different load level and material type.
During the analysis, many things have a
significant effect on the predicted fatigue life
and have to be considered including, for STRUCTURAL OPTIMIZATION
example:
In a traditional development process the design
• Dynamic stress
is the result of an iterative procedure. Based on
• Mean stress an initial design from experience, analyses for
• Notch sharpness durability, stiffness, buckling, energy absorption
etc. are performed and the results are used to
• Temperature of operation
improve the design in the next development
• Surface conditions step. This loop has to be repeated until the
• Joints (e.g. welds, rivets) analysis results meet the requirements.
Depending on the team experience of the
• Plastic effects
engineers, this process may require a variety of
• Material properties loops and, therefore development time.
Durability analysis can be done in different
regimes. This means basically defining a
criterion of failure for a component. In ground
vehicle industry the failure criterion is always
the appearance of cracks with a “technical” size
(detectable by visual inspection and a few mm
length). In aerospace industry, due to much
higher importance of low weight and adapted
maintenance systems, the criterion is also often
related to a safe crack length.
For predicting crack initiation, different regions
of load level and cycle number can
distinguished. At very high load levels and
consequently very low cycle numbers (<10)
until crack formation a static safety factor can
be defined. It defines if a spontaneous crack is
formed at a specific load level. In this region the
ductility of the material is of the utmost
importance. Fig. 1: Traditional development process
In the range between more than 100 and about
1 million load cycles, damage values are used
to quantify which part of the life has been “used
In this analysis process, the input data is the density. By including durability analysis after
geometry of the design as well as the boundary finite element analysis (see figure 4) and using
conditions. The result of the analysis shows the results like damage values or safety factors, the
design performances like stresses, fatigue life, optimizer can directly and automatically
deformations, strain energy density or natural optimize a component in respect to fatigue life.
frequencies. Weak areas can be identified by
checking the peak value locations as well as
the result’s distribution over a component. This
enables the engineer to make a proposal about
the required changes for the next design.

Fig. 2: Traditional analysis process

Due to the available computer power today, Fig. 4: Optimization process including fatigue
there is another way to come up with a life prediction
lightweight design by reversing this analysis
process. Rather than defining the design and The following examples show a topology
boundary conditions and analyze the optimization and a shape optimization
components performance, the usage of including durability analysis.
structural optimization tools enables the
engineer to define the performance and the EXAMPLE 1
boundary conditions first and let the software
come up with a design which meets the The first example deals with a topology
prescribed requirements. Again, these optimization. It’s result is a generated new
requirements can be the maximum deflection, global design for a defined design space.
the lowest natural frequency etc. Therefore, This is performed for a rectangular design
new designs can actually be created and a space with a pulsating load on one side and 3
large optimization potential can be utilized. frozen areas (the software is not allowed to
perform changes in this areas) for the boundary
condition on the other side.

Fig. 3: Analysis process utilizing structural


optimization

Therefore, time savings in the development


process can be achieved by using this
technique.
A coupling of fatigue life prediction with shape
and topology optimization has been
investigated. Fig. 5: Example 1 – Topology optimization of
Traditional structural optimization software generic beam
uses direct results from finite element analysis
like displacements, stresses of strain energy
The analysis was performed for 2 different history for optimization. Instead it would be
materials; steel and gray cast iron respectively, necessary to select a few stress distributions at
see figures 6 and 7. discrete instances. The introduction of multi-
axial fatigue life prediction eliminates this
difficult task and computes one single damage
value per node. Therefore damage can be
easily used for the automatic optimization
process.

Fig. 6: Shape and damage values for steel

Fig. 8: Example 2 – Shape optimization of


suspension component

Fig. 7: Shape and damage values for gray cast Contrary to the topology optimization in the
example 1 where a “global design” was
It can be observed that the usage of fatigue created, the area for which the shape
results rather than using direct finite element optimization is performed is restricted to a
analysis results generates two completely known critical area as shown in figure 9.
different designs for the 2 material each. For
obvious reasons, this could not be achieved by
only changing the modulus of elasticity and
Poisson’s ratio which is necessary to run the
finite element analysis. The tiny changes in
stress would create no effect during topology
optimization. Furthermore, a very good material
utilization in both designs can be observed,
which is expressed by very uniform damage
distributions.

EXAMPLE 2
Fig. 9: Critical area for damage
In the second example, a shape optimization
shall be performed for the critical area of a A comparison of the damage distributions for
wheel carrier. Three forces are acting on this both, the original and the modified design is
component with three different and shown in figure 10 and 11. It can be seen that
independent time histories. Since the loading the max. damage was reduced from D = 0.0383
condition is changing in time, it would be in the original design to D = 0.00224 in the
impossible to use the resulting stress or strain optimized design – which is a reduction of
about 96%.
CONCLUSION

Today, the use of numerical analysis methods


plays an important role in the developing
process in automotive industry and there is a
wide range of possible applications. Fatigue
life prediction and automatic optimization are
powerful software tools which help the
engineers to improve products to a higher level
more quickly.
While shape and dimension optimization tools
are in use for years already, the increased
computer power recently enables the
integration of topology and shape optimization
at the very beginning of a design process.
Fig. 10: Critical area, damage original design
Therefore, new designs can be created and a
large optimization potential can be utilized.
The effectiveness of these approaches was
demonstrated with examples.

CONTACT

Klaus Puchner, Dr. Christian Gaier and Helmut


Dannbauer are members of the Engineering
Center Steyr, located at Steyrer Strasse 32, St.
Valentin, Austria, Europe. The Engineering
Center Steyr is part of the MAGNA STEYR
group.

REFERENCES

1. FKM-Richtlinie, “Festigkeitsnachweis”,
Forschungsheft 183-2, Vorhaben Nr. 154,
Frankfurt. 1994.
2. Eichlseder, W. “Rechnerische Lebens-
Fig. 14: Critical area, damage new design daueranalyse von Nutzfahrzeugkompo-
nenten mit der FE-Methode”, Dissertation,
As it turnes out, the integration of fatigue life TU Graz. 1989.
prediction i.e. the usage of damage or safety 3. Köttgen, V.B., Oliver, R., Seeger T.,
factors as the design optimization criteria “Schwingfestigkeitsnachweise für
improves the quality of structural optimization Schweissverbindungen auf Grundlage
results. The reasons can be found in: örtlicher Beanspruchungen”, Forschungs-
hefte, Heft 143, FKM, Frankfurt. 1989.
• Adequate interpretation of static and 4. Radaj, D., “Gestaltung und Berechnung von
dynamic loads Schweisskonstruktionen, Ermüdungs-
• Consideration of load histories festigkeit”, DVS, Düsseldorf, 1985.
• Consideration of material properties 5. Grün F., “Form- und Topologieoptimierung
• Other fatigue influence factors can be unter Berücksichtigung der Betriebsfestig-
considered keit “, DA MU Leoben, 2002.

Das könnte Ihnen auch gefallen