Experimental Methods in Earthquake Engineering - Dynamic Behavior of Concrete and Seismic Engineering - page 173

modified, improved and can take new types of sensors into account. Therefore, it is
easy to adjust the system by introducing hybrid solutions to implement the PSD
method, especially as far as the sub-structuration technique is concerned (see section
5.2.3).

The control system operates in the following way: each degree of freedom,
corresponding to the displacement of a jack, is driven by a PC equipped with the
interface needed to receive data from the experiment and send orders to the jack
([MAG 91], [MAG 93]). In the system, a digital displacement sensor measures the
output signal. The real state of the system is then described by means of a variable
y , which is updated at each step of the sampling. This variable is compared with
the displacement orders

k

d . In fact, the control algorithm operates from the

error (and other information) to produce correction orders u .
Finally, these numerical orders are transformed into a u (t) step-wise continuous
signal, which is sent to the jack's servo-valve. From the displacement

k k k

edy

n

d

at the

d at the end of the step, the control unit generates a
function in the shape of a ramp followed by a plateau. This function represents the
displacement path the jack will have to follow to gradually pass from

1

beginning of the step to

n n . 1

d to d

k

The ramp is generated in the shape of a discrete sequence

d

that is used as orders.

It is presented on the left in Figure 5.5.

In such cases, the control algorithm itself is a numerical implementation of a
PID (Proportional Integral Derivative) analog method.

Note that the displacement rate for making the ramp is generally specified. The
length of a loading step is therefore proportional to the displacement increment, thus
it is unknown beforehand. On the other hand, the force sent back to the central
processing unit is an average of values supplied by the acquisition of the last points
of the displacements' plateau. The control units are geographically distributed
according to the position of the jacks. Each control unit is connected to the central
processing unit via a high bitrate optical fiber. Each unit works independently, in
real time, for each of the checked dofs in the structure, and it is controlled by a
central processing unit that integrates the motion equations.

The structural displacement is measured by an optical data line counter,
engraved with a 2 P m resolution [MAG 91]. Units allowing a 0.5 or 1 m
displacement are currently used. These digital sensors are directly connected to the
control units, without resorting to any digital/analog conversion and without any
analog pre-conditioning. The sensors therefore require no calibration and deliver a
signal that gives a correct measure of the speed, by counting on a given time period