Experimental Methods in Earthquake Engineering - Dynamic Behavior of Concrete and Seismic Engineering - page 186

In big servo-valves, there is usually a pilot stage (outflow from 1 to 60
liters/min), which is used only to drive an enslaved stage allowing an important
outflow (up to 700 liters/min capacity). A displacement sensor on the enslaved stage
provides a signal proportional to the position of the enslaved stage and then roughly
proportional to the output flow of the servo-valve. There also exists an inner loop
identical to the previous one but which uses the LVDT feedback of the pilot stage.
In spite of the settings of the various loop gains, this kind of control does not
allow us to obtain a flat transfer function within a wide frequency range. This is the
reason why three variable controls are used on some systems.
The system combines the information from each of the following signals:
- displacement of the low frequency;
- speed in the middle frequency range;
- acceleration for high frequencies.
The servo-controller subtracts the feedback for each of the three variables of the
corresponding reference signal, which thus form three error signals. The three errors
are then amplified and added in order to make up a global error signal making it
possible to drive the system. In order to stabilize the control system, pressure-
measuring feedback is injected at the level of the error signal. The technique is
generally used on the hydraulic circuit to improve the stability of the system. The
function stabilizes the high frequencies produced during the seismic tests to provide
more stable acceleration and above all it allows us to reduce the influence of the
resonance frequency of the oil column of the jacks.
For multi-axial tables where several jacks enable the table to be driven in a given
direction, the control systems are split into degrees of freedom, that is, that the
reference signals do not correspond to the signals sent to each jack but to the control
in one degree of freedom. From the reference signal and the geometrical positions of
the jack, the control system generates a control signal per jack.
In spite of the control loops, it is difficult to obtain a flat transfer function for the
whole frequency range. Therefore, a computer is used. From a pre-test (impulse or
random-loading type) it measures the transfer functions for each degree of freedom.
Then, from these transfer functions, the reference signal is corrected and a control
signal is determined, which makes it possible to obtain the desired accelerogram on
the table. After each test, the actual response is measured on the table and compared
with the reference signal. An iterative calculation system allows the control signal to
be corrected again, in order to converge to the desired response.