Experimental Methods in Earthquake Engineering - Dynamic Behavior of Concrete and Seismic Engineering - page 174

(the sampling period, for example). The sharp measure of the sensors allows us to
reach an accuracy of the displacement control around 50 P m.
It should be noted that the displacement sensors are mounted on a supporting
structure independent of the movement of the jacks or those of the reaction wall on
which they rest (see Figure 5.1). In fact, as the stiffness of the wall is not infinite, it
bends with the motion of the tested structure. This means the lengthening of the jack
is not characteristic of the motion of the tested structure alone, especially if we
consider the accuracy of the displacement. Therefore, an independent reference
system has to be introduced, and we should ensure its neutrality (for example, its
mass should be such that no resonance system can be induced by the vibrations
either of the hydraulic system or of the control system).
The hydraulic system which powers the jacks operates under a 210 bar pressure
and delivers a 25-liter per second maximum flow. The jacks are two different types:
maximum force: 0.5 and 1.0 MN; displacement: r 250 mm and r 500 mm.
5.2.3. The sub-structuration method
The PSD method is a hybrid method that combines the digital integration of the
motion equations of a complex structure (condensed on a reduced number of dofs)
with the measure of the reaction forces that results from the imposed motion.
Despite its potential, direct experimentation on very large civil engineering
structures such as bridges would be difficult: apart from the size problem, the
simultaneous control of a large number of dofs would be quite laborious. Yet in
order to deal with such cases, it is possible to expand the application field of the
PSD method, at least whenever part of the structure can be modeled; this is called
the sub-structuration procedure [DER 85].
This procedure takes advantage of the hybrid nature of the PSD method by
combining the modeling of part of the structure (numerical structure) with the real
testing of the remaining structure (tested structure). The procedure naturally applies
to a bridge, as its biggest part, the deck can often be considered as elastic linear, and
can therefore be modeled with finite elements. Only the piers subjected to damage
will be tested in the laboratory. Another advantage, which is particularly welcome in
the case of bridges, lies in the fact that we can impose asynchronous seismic
loadings or loadings that have different amplitudes along the foundations.
The numerical part of the structure has to be modeled suitably, and then the
resulting motion equation discrete system must be integrated in time. The numerical
model can reveal a number of dofs that are far more important than those checked in
the lab. Therefore, it is not straightforward to modify the software developed by the