Professional Documents
Culture Documents
ABSTRACT
The hybrid simulation test method is a versatile technique for evaluating and analyzing
the seismic performance of structures. Its ability to seamlessly integrate both a physical
and numerical simulation of a structure into a single model allows for a cost-effective test
method that can be chosen as an alternative to other tests, such as shake-table tests and
quasi-static tests.
A difficulty in the application of the hybrid simulation test method is that the
delay and dynamic response of the servo-hydraulic actuators can introduce errors into the
results. These actuator delays tend to add more energy to the structural system, and as
these errors propagate, the system may become unstable. The compensation of these
delays is critical to the stability and the accuracy of hybrid simulation. Various
procedures have been introduced to help compensate for these delays in order to improve
the effectiveness of hybrid simulations. In this study, four different delay compensation
methods were studied: the polynomial extrapolation method for both force and
displacement, kinematic equations used in explicit Newmarks Beta Method, and the
Force Correction method. These tests were evaluated analytically, numerically, and
experimentally. The results of these tests showed that all of these methods greatly
improved the stability and accuracy of the simulation through modifications of the
command and measurement signals. The polynomial extrapolation procedure for
displacement and Newmarks Beta method proved to be the most effective in
compensating for the actuator delays that occur during hybrid simulations.
Introduction
In the midst of catastrophic earthquakes affecting the Middle East and Southeast Asia,
and recognition of these risks closer to home, more and more Americans are shifting
focus to reducing the existing vulnerabilities in our nations infrastructure. Earthquakes
are known to cause devastating damage, and as engineers, we seek to lessen this damage.
More research is needed to improve our current knowledge of building behavior during
such severe earthquake shaking. The most effective method to understand the behavior
of structures and components is through experimental evaluation.
Three main experimental methods exist to evaluate the performance of structures
during seismic excitation. This includes shake table tests, quasi-static tests, and hybrid
simulations. Shake table tests allow for models to be subjected to the most realistic
seismic loadings. The disadvantage of using shake tables is that they are very expensive
to build and payload is limited to the size of the table and its payload capacity. Expenses
increase greatly when building larger shake tables with multi-degree of freedom
capabilities, so the reduction in scale required for models remains a major drawback for
using shake tables for seismic testing. An alternative is quasi-static testing which is much
(1)
where M is the mass, C is the damping coefficient, R is the restoring force, f is the applied
load, and a(t), v(t), and d(t) are the time-dependent acceleration, velocity, and
displacement, respectively.
One difficulty regarding Equation (1) is the difficulty in
determining the restoring force R. The problem can be simplified by doing a linear
elastic analysis and letting R(d) = Kd, where K is the stiffness factor. This allows the
problem to be reduced to a linear second order equation. However, this limits the scope
of the problem to linear elastic analysis. Structures are designed to respond in the
nonlinear range during strong earthquake shaking and this should be considered in the
analysis.
The main task of the numerical model is to integrate the equation of motion using
the restoring force data R(d) measured from the experimental setup. The solution to
getting this restoring force is based on a time-stepping integration procedure and enforces
the discretized equation of motion
mai cv i ri f i
(2)
at time intervals t i it for i=1 to N. The time-dependant variables are noted by the
subscript i and t is the integration time step and N is the number of integration steps.
Therefore, the displacementresponse time at time step t i 1 is calculated based on the
response from time step ti (Mosqueda et al. 2005; Stojadinovic 2005).
Tn
(3)
di1 di1
d 2di di1
, ai i1
2t
t 2
(4)
By combining these two expressions with the descretized equation of motion (2), the
target displacement at the end of step i+1 can be seen as
2m
m
1
c
di1
di 2
di1 ri f i
2
m
c t
(5)
t
2t
2
t
2t
Now that the displacement at time ti+1 can be solved, the numerical side of the hybrid
simulation is functional and only needs to follow a simple stepwise procedure to advance
adds more energy to the system, causing the structure to become unstable (Mosqueda et
al. 2005).
One of the most widely used method to compensate for this actuator delay was
introduced by Horiuchi et al. in 1999 (Ahmadizadeh et al. 2006). Horiuchi devised a
procedure using a polynomial extrapolation of the command displacement to help offset
the delay and to lessen the errors that may have been produced. Following Horiuchis
method, the last four data points are fitted with a third order polynomial and is used to
predict the command displacement ahead of the actual simulation time.
1 2
x1 x o t x
t x
o
o
2
(6)
x x o (t jt) x
(t jt) 2 x
o
o
2
(7)
Plugging in the known quantities into Equation (6), the desire displacement can be found.
Similar to Equation (6), taking into account the delay jt and plugging in the other known
Delay Compensation
The effects of actuator delays have been studied for many years in the areas of both
hybrid simulations and active structural control. The delay in a hybrid simulation in
effect can act as negative damping, which causes more energy to be added to the system.
In some cases, this negative damping can eventually cause the system to become
unstable. Compensation of delay is therefore very critical to the accuracy of both the
experimental and numerical simulations of hybrid experiments.
In this paper, established methods of actuator delay compensations have been
tested. The accuracy and reliability of these tests were studied analytically, numerically,
and experimentally. Based on these results, the effectiveness of each procedure was
evaluated.
Procedures
The behavior of the structure is modeled numerically during the hybrid testing by the
equation of motion (1) in discrete parameter form. The equation of motion (1) is solved
using numerical integration algorithms. These algorithms can also incorporate errors into
the model such as actuator delays and random noise from the load cells to better study the
actual experiment.
Proper care should be taken to effectively tune the actuator controllers and to
calibrate the measurement instrumentation. Hybrid simulations are very sensitive to
these errors since experimental measurements are used to compute future commands in
the time-stepping algorithms.
Errors caused by actuator delays and faulty
instrumentation may cause for errors to propagate as the simulation progresses, possibly
causing for the structure to become unstable. Methods have already been discussed to
help compensate for actuator delays in order to minimize the effects of experimental
errors. These methods will be explored further into detail later on in this report.
Simulink, a computer software often used for simulations and model-based
designs, was used to help model the system (Mathworks 2003). Shown below is the
Simulink model used to replicate the equation of motion.
be seen in Figure 10 where it is being compared to the exact simulation. This method can
be seen to efficiently compensate for the delays that are present during the simulation.
Figure 11. Simulation with compensation using Polynomial Extrapolation for force
The third method tested was Newmarks Beta method. This method utilizes
Equation 7 which uses a basic kinematic equation assuming constant acceleration and
evaluates the equation at time t + jt to find the command displacement. This is a very
simple method that is very easy to use. The results shown in Figure 12 also show that
this method is efficient and very accurate in compensating for the delay. The graphs for
the experimental test and the exact test are very similar and indeed do show that this
method is accurate despite the noise that was introduced during the experimentation.
Much appreciation is also given to Dr. Reinhorn, Tom Albrechcinski, Carmella Gosden,
Sofia Tangalos, and Melanie Brown of NEESinc for all their help and support. Lastly, I
would like to thank NEESinc and NSF for giving me the opportunity to participate in this
NEESreu program.
References
Ahmadizadeh, M., Mosqueda, G., and Reinhorn, A. M. (2006). "Compensation of
Actuator Delay and Dynamics for Real-Time Hybrid Structural Simulation."
Chopra, A. K. (2000). Dynamics of Structures, Prentice Hall.
Horiuchi, T., Inoue, M., Konno, T., and Namita, Y. (1999). "Real-time Hybrid
Experimental System with Actuator Delay Compensation and its Application to a
Piping System with Energy Absorber." Earthquake Engineering and Structural
Dynamics, 1121-11141.
Magonette, G. (2001). "Development and Application of Large-Scale Continuous
Pseudodynamic Testing Techniques." Dynamic Testing of Structures: Papers of a
Theme, Philosophical Transactions of the Royal Society: Mathematical, Physical
and Engineering Sciences.
Mahin, S. A., and Shing, P.-s. B. (1984). "Pseudoynamic Test Method for Seismic
Performance Evaluation: Theory and Implementation." Earthquake Engineering
Research Center, 84/01, 162.
Mathworks. (2003). "MATLAB, The Language of Technical Computing."
Mosqueda, G., Stojadinovic, B., and Mahin, S. (2005). "Implementation and Accuracy of
Continuous Hybrid Simulation with Geographically Distributed Substructures,"
University of California, Berkeley, Berkeley, California.
Nakashima, M. (1984). "Stability and Accuracy of Integration Techniques in
Pseudodynamic Testing." Building Research Institute (No. 105), 114.
Nakashima, M. (1999). "Real-Time On-Line Test for MDOF Systems." Earthquake
Engineering and Structural Dynamics, 393-420.
Nakashima, M., and Takai, H. (1985). "Use of Substructure Techniques in
Pseudodynamic Testing." Building Research Institute (No. 111), 115.
Sivaselvan, M. V. (2005). "A Unified View of Hybrid Seismic Simulation Algorithms."
Stojadinovic, B. "Hybrid Simulation: Basics." NEES Hybrid Simulation Workshop,
Berkeley, California.