Experimental Methods in Earthquake Engineering - Dynamic Behavior of Concrete and Seismic Engineering - page 179

insert the finite element method into the control program instead of the ramp
generator.

Figure 5.4. Conventional and continuous pseudo-dynamic method

We have seen that both PSD conventional and continuous methods use the same
integration schemes. However, for each b acceleration value at the base (read in the
recorded acceleration file with a sampling period

' ), the continuous method will

introduce a series of

n

values with

b

( 0

) accelerations obtained by kn

ss ikn / ss

ss

interpolation between b and

b . For each

' large step , the step by step

1

integration process will be carried out

' operating period of the
control system (typically 1 or 2 ms), as if O sub-steps were introduced into the
conventional method (see Figure 5.4).

n

times, at the

ss st

The O time dilatation factor between a PSD test and a real time test is given by:
/ O '' . For a test on a several-storey building, we will typically obtain:

500

nt T

ss st

' , i.e. O = 200. One second of the real earthquake
will develop in 200 seconds in the test.

s n

,

t ' , 2ms 5ms

st

The results of a series of tests carried out on a 3-storey, 10.4 m high steel
structure have shown that the PSD method could be implemented 10 to 20 times as
fast as the conventional method, depending on the importance of the masses present
in the experimental device. Let us actually note that besides the static stresses linked
to the displacements, the force cells of the jacks measure inertia forces about O
times lower than the real inertia forces, and therefore they can often be neglected.
Nevertheless, when the mass of the tested structure is large and O is smaller than
1| , the inertial effects have to be taken into account when we try to resolve the
equations of motion.