EFT_Stage3_Results

The base plate and its anchorage in EFT study are different from those in the shake table study; hence, the structural stiffness during the EFT tests was 1% greater than that in the shake table study due to a slight change in the column boundary condition.

   Free vibration tests were conducted, and parametric simulations were made to determine a combination of viscous damping and friction force which would minimize the error between measured displacements and simulation displacements based on a least square technique. The structural mass increased by 2% during the EFT tests, which was in part due to the addition of a thick plate for connecting the actuator. The damping during the EFT tests was decreased by 10%, which was attributed to an unknown change in the dampers and a change in test environment.

With the above structure properties, the natural frequency of the structure in the two studies changed approximately by 1% (from 2.89 Hz in the shake table study to 2.87 Hz in the EFT study). With a reduced damping (from 9.6% in the shake table study to 8.2% in the EFT study), it was anticipated that the displacement and velocity of the structure in the EFT study would be slightly greater than those in the shake table study.

   The results with a 0.55 g Northridge earthquake input are presented above.  Only 11 seconds of response (from 6 sec to 16 sec) are shown to make the graphs more readable.  The force comparison shows that the effective force command was followed by the actuator closely in the EFT test.  The Fourier amplitude of the force applied to the structure by the actuator was slightly greater than the force command in the frequency domain, indicating a slight overcompensation of the natural velocity feedback due to uncertainties in the estimation of the servovalve flow property.  Both the global responses (displacement, velocity, and acceleration) and local response (column base shear) of the EFT test match well with those of the shake table test.  The structural responses in the EFT test are slightly greater than the shake table results, which was attributed to the slight overcompensation of the natural velocity feedback and the decrease in structural damping.  In addition, the after-shock free vibration, which began at 15 sec, was accurately captured.

In addition, after-test inspection indicated that the column ends were partially yielded during the test. The maximum spool opening in the test was about 25%, which is beyond the linear range (10%) of the servovalve performance (refer to servovalve flow curve). These observations indicate that with nonlinear velocity feedback compensation, the EFT method can be used to test nonlinear structures with large hydraulic demands.

   The EFT tests without dampers were problematic as shown by the first 12 seconds of the test with 0.3g El Centro earthquake. Although the actuator seemed able to follow the force command in the time domain, large force overshoots are evident in the frequency domain. The system was slightly driven into the unstable region after 4s, and the structural responses were much larger than those in the shake table study.

The unstable system was attributed to the fact that the current velocity feedback compensation was based on a predetermined flow curve, which did not consider the uncertainties in the servovalve flow property. The system uncertainties may cause instantaneous overcompensation, which in turn may cause instability of the test system. The test structure without the dampers had little damping (0.25%), such that the test system was not able to tolerate the instantaneous instability.

   To further explore the system stability in testing the structure with small damping, two car struts were used to replace the viscous dampers towards the end of the study. As shown above, the strut provided three times damping force in one direction than in the other though an equivalent damping of 2.3% critical damping was obtained based on a free vibration test. The test system was stable, and the force comparison indicates that the actuator was able to apply forces at all frequencies within 10 Hz. A small amplitude drop around 3 Hz (the natural frequency of the test structure) is evident, which was attributed to the system uncertainty not included in the velocity feedback compensation.

Therefore, it was concluded that a structure to be tested using EFT with the current velocity feedback compensation scheme should have at least 2% damping to prevent instantaneous instability.  Tests with more advanced compensation schemes, which consider the system uncertainties, might release the damping requirement to the test structure.