SUMMARYWhen subjected to long-period ground motions, high-rise buildings' upper floors undergo large responses. Furniture and nonstructural components are susceptible to significant damage in such events. This paper proposes a full-scale substructure shaking table test to reproduce large floor responses of high-rise buildings. The response at the top floor of a virtual 30-story building model subjected to a synthesized long-period ground motion is taken as a target wave for reproduction. Since a shaking table has difficulties in directly reproducing such large responses due to various capacity limitations, a rubber-and-mass system is proposed to amplify the table motion. To achieve an accurate reproduction of the floor responses, a control algorithm called the open-loop inverse dynamics compensation via simulation (IDCS) algorithm is used to generate a special input wave for the shaking table. To implement the IDCS algorithm, the model matching method and the H ∞ method are adopted to construct the controller. A numerical example is presented to illustrate the open-loop IDCS algorithm and compare the performance of different methods of controller design. A series of full-scale substructure shaking table tests are conducted in E-Defense to verify the effectiveness of the proposed method and examine the seismic behavior of furniture. The test results demonstrate that the rubber-and-mass system is capable of amplifying the table motion by a factor of about 3.5 for the maximum velocity and displacement, and the substructure shaking table test can reproduce the large floor responses for a few minutes.

E‐Defense, which is operated by the Japanese National Research Institute for Earth Science and Disaster Resilience, currently contains the world's largest 3‐dimensional shaking table (size: 20 × 15 m, maximum loading capacity: 12 000 kN). This facility has already implemented over 80 full‐scale or large‐scale experiments since its operation commenced in 2005. During these research activities, E‐Defense has encountered many challenges in the successful execution of various types of experiments within the designed experimental capacity and under the guidance of safe operational management. The experience gained in the past years using E‐Defense has led to the accumulation of knowledge on loading/shaking techniques, measurements for large‐scale specimens, table maintenance and operation, safety management, and specimen construction. This article highlights the major technical issues that have been encountered during the operation of E‐Defense and introduces the solutions that have been achieved. This is followed by brief summaries of some of the major experiments conducted since E‐Defense was commissioned.

Summary Nonlinear signal‐based control (NSBC) is very powerful for controlling structural systems with parameter variations and has the advantage that the controllers can be designed using classical control theory and expressed by transfer functions. This report describes the first application of NSBC to shake table tests with nonlinear specimens and compares the performance of NSBC with that of a basic control approach that relies on the inverse transfer function of the controlled system. NSBC and the basic approach were numerically applied to shake table tests to excite a nonlinear single‐degree‐of‐freedom system with earthquake acceleration motion, considering the nonlinear specimen characteristics and estimation errors associated with the table dynamics. NSBC achieved excellent control with near 100% accuracy, whereas the basic approach provided insufficient control. Although inaccurate estimation of the pure time delay in the controlled system causes instability at the practice of NSBC, proper design of the nonlinear signal feedback controller prevented instability. In experimental examinations, controllers for the two approaches were designed on the basis of the table dynamics with/without a specimen, which were preliminarily identified by performing tests using a random wave with small amplitude excitations. With no specimen present, both approaches yielded the expected acceleration motion on the table with high accuracy. However, with the specimen present, only NSBC successfully achieved excellent control of the shake table with near 100% accuracy, whereas the basic approach did not because of the specimen nonlinearity. These results numerically and experimentally demonstrate the efficiency and practicality of NSBC for shake tables supporting nonlinear structures.

No abstract