The pseudo-dynamic test method combines well-established analytical techniques in structural dynamics with experimental testing. It is especially efficient for testing structures that are large, heavy or strong to be tested on the shaking tables. The objective of this study is to improve the numerical algorithm used in the Pseudo-dynamic test and investigate the major sourses of experimental errors.
Since experimental errors are introduced into numerical computations through displacement control and restoring force feedback, significant cumulative errors can be introduced in pseudo-dynamic test results. Particularly Coulomb damping due to friction between a test structure and hydraulic actuators can influence the restoring forces developed by structural deformations. As a results, energy dissipation phenomenon is presented and test results are deteriorated. In this study, efficient procedure to mitigate these error effects, which is divided into three major parts, is proposed. In the first part, the stability, accuracy and error propagation characteristics of numerical integration algorithm are investigated and appropriate integration time steps to suppress the growth of experimental errors are selected. In the second part, dry-friction and other kinds of errors are detected from linear elastic free-vibration tests, and these errors are removed with improvement of test apparatus and with caution. In the last part, the proposed numerical algorithm, which includes equivalent energy compensation procedure to correct restoring forces distorted by experimental errors, is used to get structural behaviors. For verifing the reliability and capacity of the pseudo-dynamic test, test results obtained in this study are correlated with analytical results and shaking table test results.
The results of this study indicate that some systematic experimental errors are detrimental to pseudo-dynamic test and induce significant energy effects. However, it is shown that reliable test results can be obtained if adequate experimental equipment and technique, and appropriate numerical method for mitigating error propagation effects are used.
