Geoscience Reference
In-Depth Information
316
Multiscale Geomechanics
As can be seen in Figure 9.38(a), adjusting the stiffness of rocks yielded a better
dynamic response. This highlights the important role played by rocks on the behavior
of the entire dam and the sensitivity of the dynamic response to a slight modification
of the shear modulus of these materials, especially since this modulus is difficult to
determine
in situ
or in the laboratory. This analysis also shows that several sets of
parameters can lead to the same response in monotonic loading conditions, but only
one can provide the seismic response correctly. The identification of the first set could
be more accurate if the results of cyclic tests were available. However, given the size
of the rock fragments, it is difficult to characterize this material in the laboratory.
Toevaluatethepredictivequalityofthemodel,thedamresponsefromtheearthquake
of September 19, 1985 has been analyzed by this second set of data. The acceleration in
DM was recorded during this earthquake. In Figure 9.38(b), we present the comparison
between the recorded and calculated response spectra at point DM. We can see that
the numerical simulations predicts the structure's response at point DM rather well.
However, we note the acceleration at the crest measured during the earthquake of
May 30, 1990. In Figure 9.38 (c), we can see the response spectra of the dam crest
derived from the accelerations calculated and recorded at that point. There is, once
again, a good prediction obtained using a calibrated set of parameters obtained using
both static and dynamic behavior of the dam. This indicates the effectiveness of the
parameter calibration procedure from a static-dynamic analysis in the absence of a
suitable series of cyclic tests. This procedure may be recommended for estimating
the seismic vulnerability of existing structures because, very often, there is a lack of
experimental data and information concerning the material in place and the construction
procedure.
Having highlighted the importance of stiffness on the model response, we must try
to characterize this parameter as accurately as possible. This might be envisaged by
a vibrational analysis of the structure. Thus, the vibration response and the dominant
modes can be used for indirectly determining the stiffness of rockfills. This analysis
obviously requires the proper equipment.
Through this example, we have presented a comprehensive methodology for
identifying parameters of the constitutive model by using the overall response of the
structure and especially during seismic loading. Being able to obtain both the frequency
content and peak accelerations proved the ability of the model to predict the behavior of
the dam under loadings in the future. This model was then used to analyze the influence
of strain hardening induced during the earthquake on the subsequent behavior of the
dam. Therefore, the dam was submitted again, albeit with different amplitudes, to
the same magnitude of earthquake it had previously experienced. The freeboard loss
resulting from a second application of the earthquake was smaller than that obtained
for the first one because of the hardening induced in the materials. These results are
summarized in Figure 9.39.
Search WWH ::
Custom Search