Geology Reference
In-Depth Information
advantage of the gradient search in the case of a
small number of dof has disappeared.
improvement of a given design. On the other
hand there is no guarantee that we will find the
absolute optimum in the case of challenges that
are not very trivial. The choice of the optimization
strategy therefore has to be made very carefully.
Evolutionary procedures are able to cover large
regions of parameter spaces even in the case of
many local maxima and very irregular landscapes
(Figure 1d). But they need very large numbers of
trials or individuals. They do not converge very
well if a local maximum has to be approached
closely, so the option to switch to locally efficient
gradient searches should be included.
The use of reduced models always inhibits
the loss of information due to reduction. The
discrepancy between fast proposals and qualified
statements about compensators has to be dealt
with. Experience and a critical check of the data
produced by the optimization systems are indis-
pensable to interpret the results. Nevertheless, the
approach is fast and easy to handle. It may serve
as a starting point in compensator design. The idea
of introducing extra dof for the compensators may
be extended to any other dynamic problem and
the evolutionary optimization as well. Motivated
engineers will profit from this way of dealing
with questions and will surely add new ideas,
modifications and extensions as result of their
growing experience.
The expansion of the approach to more difficult
problems e.g. to non-linear or transient studies may
be done without difficulties. But like for all other
optimization the large number of variants to be
checked is less or more prohibitive as long there
is no efficient management of the jobs clusters of
computers or any other parallelization.
Discussion, Strengths, and
Limitations of the Method
Passive compensator systems are used as earth-
quake prevention installations with satisfactory
success. Like any other anti-earthquake strategy,
their efficiency is limited by the way they are
adapted to the loading history of the expected
seismic events. On the other hand, they are useful
for many other anti-oscillation applications such
as bridges under traffic load or chimneys under
wind exposure.
The main advantage of the evolutionary opti-
misation approach lies in the capability to handle
systems with many compensators, especially
distributed dampers (Fu and Johnson 2009). In
such cases, conventional methods are difficult to
apply and tend to find only local maxima of the
objective function.
The method outlined shows high potential
to improve the design of passive compensator
systems to enhance the resistance of buildings.
It allows a fast check of many variants and finds
even unexpected proposals, as it does not stop
at local optima. It is a powerful tool to quantify
the effects of the design. Furthermore, due to its
drastic simplification of the building, the total
system only has a small number of dof. It has to
be linear in geometry and stiffness. Such reduced
systems always inhibit the danger of ignoring es-
sential properties of the original part. The results
should therefore be looked at with sufficient care.
It may be helpful as a tool to propose initial designs
that may be qualified by more detailed studies. It
reduces the numerical effort of the optimization
process by a great amount.
FUTURE RESEARCH DIRECTIONS
Solutions and Recommendations
The studies done show that the method proposed
may help to accelerate the design of passive com-
pensator systems. The application to real structures
and the comparison with more detailed approaches
Optimization is one of the central tasks not only
in engineering. Most of our work deals with the
Search WWH ::
Custom Search